While single-cell transcriptomics has just become popular in the field of IVD biology to deconvolute the cell types within the heterogeneous populations of the NP, another way to determine NP phenotypic markers is to follow the cells as they differentiate. In mice, fate mapping strategies together with whole transcriptome analysis have been key to exploring differential marker expression associated with different stages of cell transition within the NP. NP of young and adult mice (<1 year of age) mainly consist of NCs (Winkler et al., 2014; Mohanty and Dahia, 2019; Mohanty et al., 2019). Notochord-specific lineage-tracing studies often make use of tissue-specific-inducible Cre-based reporters to trace the fate of eNCs (Chan et al., 2014). Previous work based on the SHH-Cre (Choi et al., 2008; Peck et al., 2017), NOTO-Cre (McCann et al., 2012), and the tamoxifen-inducible SHH-CreERT2 (Choi et al., 2008) and KRT19-CreERT (Mohanty et al., 2020) mouse models indicate that all cell types within the NP of mice are derived from the embryonic notochord. Studies on aged mice are rare and scarce work indicate that the NC to NPC transition occurs between 9 months 1 and 2 years of age (Winkler et al., 2014; Mohanty and Dahia, 2019; Mohanty et al., 2019).

Another attractive strategy is the use of reporters to select specific cells at different stages of differentiation. These fate-mapping studies can only be done in mice, but are instrumental as exploratory studies to find markers that can later be verified in human IVD tissue. For instance, cells isolated from SHH-Cre; ROSA:YFP mice at E12.5 and postnatal day 0 can be employed to investigate specific cells populations that represent eNC to NC transformation (Peck et al., 2017). Significant SHH pathway expression was demonstrated in eNCs, while the transforming growth factor-β (TGF-β)/insulin-like growth factor pathway, and ECM expression was significantly higher in NCs (Peck et al., 2017).

Such strategies can be further enriched by using mouse models to study IVD degeneration (Table 2), which can provide insight into the role of genetics and specific genes in NCs towards NPC transition, as well as the use of live-cell imaging technology to monitor this transition (Joo Han Kim et al., 2009). These fundamental and mechanistic studies involving mice will not only shed light on NC (patho)physiology but will also assist in further identifying and confirming cell markers for the specific life-stages of NCs; its journey from the notochord to the disc and its role during maturation and ageing. With the identification of the different NP cell phenotypic profiles and the ability to isolate the targeted cell populations becoming more achievable, the following sections highlight the advantages of using NCs for the treatment of IVD degeneration-related LBP.

NCCM contains the NC-secreted factors, either soluble or ECM-bound, as well as ECM components that are released due to tissue remodeling during culture. To date, five studies focused on the identification of (bioactive) proteins present in NCCM (Erwin and Inman, 2006; Purmessur et al., 2011; Gantenbein et al., 2014; Bach et al., 2016b; Matta et al., 2017) (Figure 7). Based on these studies, it appears that NCCM contains many proteins primarily associated with ECM, anti-catabolism, and growth factors (Figure 7), while upstream analysis identified possible transcription factors involved. The detected matricellular proteins reflect the matrisome produced and remodeled by the NCs. Several of those (including clusterin, tenascin, and α2-macroglobulin) have been suggested to have protective effects for NP cells (Tortorella et al., 2004; Purmessur et al., 2011). In line with this thought, decellularized ECM derived from passaged porcine NCs induced proliferation and chondrogenic ECM production in porcine synovium derived MSCs, indicating that this microenvironment favors differentiation into an NP-like phenotype (Pei et al., 2012).

The NC secretome, rich in ECM molecules, is deposited in the form of a GAG-rich matrix in the NP tissue and brings up challenges in detecting growth factors. Bach et al. (2016) by employing NCCM in proteomic analysis did not detect growth factors. Growth factors secreted by NCs may be overshadowed by large quantities of ECM proteins released by the NP tissue during culture. Furthermore, growth factors may sequestrate within the robust GAG content. The study of Matta et al. (2017) fractionated non-chondrodystrophic canine NCCM by size-exclusion and identifoied that upto 31% of the identified proteins were ECM-related. Several growth factors were also detected, and these include connective tissue growth factor (CTGF, also known as CCN2), TGF-β, Wnt-induced soluble protein 2, insulin-like growth factor binding protein-7, angiopoietin-like 7, and their modulators (chordin, sclerostin and cartilage intermediate layer protein) (Matta et al., 2017).

Within the context of the mode of action of NCs, CTGF is a multifunctional growth factor that interacts with other growth factors such as TGF-β and bone morphogenetic proteins (BMPs), thereby mediating ECM interactions (Figure 7). Furthermore, CTGF can independently induce ECM synthesis (Nakanishi et al., 2000) and plays a role in apoptosis (Gygi et al., 2003; Croci et al., 2004). Pathway analysis demonstrated that the signaling pathways TGFβ, Wnt/β-catenin, axonal guidance, nitric oxide synthase, leucocyte extravasation signaling, and inflammation associated matrix degradation were significantly covered (Matta et al., 2017). Complementary upstream analysis of detected proteins may allow for the identification of potential binding sites and transcription factors involved in NC physiology, including activator protein 1 (AP-1) and paired box gene 4 (PAX4) (Bach et al., 2016b). Furthermore, linking to the possible anti-angiogenic and anti-neurogenic effects (Cornejo et al., 2015; Purmessur et al., 2015), class 3 semaphorins (SEMA) were identified in proteomic analysis of canine (Matta et al., 2017) and porcine (Bach et al., 2016b) NCCM. SEMA3C plays a role in the innervation and vascularization of degenerated human IVDs and has been linked to LBP (Binch et al., 2015). The finding of class 3 semaphorins in canine and porcine NCCM, supports the inhibitory role of semaphorins in nerve and blood vessel growth (Bagnard et al., 1998).

What is the ideal combination of bioactive (growth) factors that can be employed to harness the therapeutic potential of NCs? Anecdotally, only specific growth factors (CTGF and TGF-β) were selected for further analysis given their pivotal role in regulating the expression of several ECM proteins and interaction with other growth factors (Matta et al., 2017). Combinatory treatment with TGF-β₁ and CTGF restored a healthy, NC-rich NP in a preclinical rat model with induced IVD degeneration (Matta et al., 2017). Also, a single intra-discal injection into degenerated canine IVDs resulted in increased disc height, expression of healthy ECM proteins, and reduced expression of inflammatory mediators (Matta et al., 2018). However, the study did not look into the possible enhancement of tumor development or tissue fibrosis (Bach et al., 2017b; Chen et al., 2020; Zhao et al., 2020). As for an ideal combination for bioactive factors, the key consideration will be to maximize the positive effects of detected bioactive factors in the NCCM, while minimizing their detrimental effects.

The NCCM-based proteomic studies observed a plethora of possible (bioactive) components and signaling pathways involved in disc physiology. These possible candidates were recently confirmed in proteomic analysis of human fetal NP tissue at 24 weeks of gestation (Rajasekaran et al., 2020a; Rajasekaran et al., 2020b), bovine fetal NC-rich matrisome at 7 months of gestation (Caldeira et al., 2017) and young murine matrisome (Kudelko et al., 2021) where mainly ECM-related proteins were found (Figure 7). In contrast, the matrisome of NC- and NPC-containing adolescent human NP tissue was distinct from the NC rich matrisome; while these NC low matrisomes contained many ECM-related proteins, they contained even more non-matrisome proteins (Tam et al., 2020). Taken together, these studies shed light on factors important for the composition of healthy NC-containing NPs, but since the fetal tissues of these species, even at these early stages, may have contained NPCs besides NCs (Alini et al., 2008; Bach et al., 2015), the results cannot be confirmed as NC-specific.

Challenged by the complexity of pinpointing the hub-bioactive factors secreted by NCs that exert essential disease-modifying effects, the ECM of NC-rich tissue, the so called NC-derived matrix (NCM), has recently become a focus of attention. It is being considered as a potential candidate, or at least a model system, because of both the bioactive factors it contains, secreted by NCs, as well as replenishing the NP ECM constituents. NCM may, when applied as clinical therapy, comparable to demineralized bone matrix that is widely used in bone regeneration (Gruskin et al., 2012), serve as an ‘instructive matrix’ by locally releasing growth factors and in this way facilitate tissue repair (Bach et al., 2018). Therefore, intradiscal injection of NCM could be a promising regenerative treatment for IVD disease, circumventing the cumbersome identification of bioactive NC-secreted substances or the ethical and regulatory issues raised by the use of NCs in cell-therapy approach (see, Ethics).

Recent studies suggested that partially digested rabbit NC-rich NP tissue co-cultured with human NPCs increased ECM production, proliferation, T and KRT18 expression (Bai et al., 2017), indicating that NC-derived NP tissue indeed has regenerative properties (Supplementary Material S2). Furthermore, porcine NCM increased the DNA and GAG content of bovine NPCs in vitro (de Vries et al., 2018). In this study, the NCM concentration in the medium was adjusted to obtain a similar protein concentration as NCCM (∼0.4 mg/ml) to allow for comparison. Interestingly, NCM’s anabolic effect on adult human NPCs was stronger compared with NCCM derived from the same porcine spines. In another in vitro study, however, 2 mg/ml NCM did not exert any anti-angiogenic and anti-neurogenic effects (de Vries et al., 2018b). Based on the overall promising anabolic in vitro benefits, the effect of 10 mg/ml NCM (concentration based on the dose-finding in-vitro pilot on canine NPCs) was tested in vivo on mildly (spontaneously) and moderately (induced) degenerated chondrodystrophic canine IVDs (Bach et al., 2018). NCM injected in moderately degenerated canine IVDs exerted beneficial effects at a macroscopic level and on MRI, induced type 2 collagen-rich ECM production and improved the disc height. Furthermore, NCM exerted anti-inflammatory effects in bovine NPCs in vitro (de Vries et al., 2018) and ameliorated local inflammation in the canine IVD degeneration model in vivo (Bach et al., 2018). Although these results look very promising, when translated towards the clinic, one of the bottlenecks that need to be addressed is the decellularization of the NCM to avoid host versus xenogeneic graft responses.

In parallel, promising studies have been conducted with decellularized porcine NCM to support cell-based translational studies (Supplementary Material S2). With this decellularized NCM, human NPC viability was maintained and GAG synthesis in vitro was enhanced (Wachs et al., 2017). Furthermore, human MSCs cultured on decellularized porcine NCM expressed NP-cell markers and deposited GAGs and collagens (Mercuri et al., 2011; Mercuri et al., 2013; Zhou et al., 2018; Xu et al., 2019). In the latter study, the cultured constructs of NCM that were seeded with MSCs counteracted IVD degeneration in a rabbit model (Xu et al., 2019). Similar findings were also reported for decellularized porcine NCM combined with rabbit MSCs that partly restored the ECM content of degenerated rabbit NPs in vivo (Zhou et al., 2018). Xu et al., 2019 (Xu et al., 2019) also analyzed the MSC-NCM tissue constructs by mass spectrometry and detected the presence of abundant important signaling molecules, e.g. TGF-β_1, which is in line with the observations from the upstream analysis of proteins detected in NCCM, indicating the involvement of transcription factors mediating TGF signaling (Bach et al., 2016b) Affirmatively, the in vitro results of this study were (partially) mediated by increased TGF-β Smad2/3 signaling (Xu et al., 2019)_, albeit functional inhibitory studies to confirm this observation is still lacking.

Decellularization is necessary to allow the use of porcine NCM for clinical application. Different decellularization methods have been employed to remove the NCs from the tissue and maintain the ECM. (Supplementary Material S2) Nonetheless, the GAG and collagen content of the decellularized constructs was negatively affected by the decellularization processes in all studies. Since cellular components have numerous ligand-binding domains important for cell-ECM interactions, specific associations with cytokines, chemokines, morphogens, and growth factors that regulate cell differentiation and proliferation (Caterson, 2012), decellularization could affect these bioactive factors; impacting the instructive role of the ECM. Therefore, optimal decellularization agents and protocols remain to be established for this specific application, balancing the safety and maximum instructive capacity of the NCM. Within this context, while fundamental research focuses on the mode of action of NC-based decellularized scaffolds, and follow-up translational work should focus on injectability, dose-finding, and subsequent translation from bench to bedside. Although porcine tissue-derived products have already been clinically applied (Stone et al., 2007; Konertz et al., 2011), safety aspects including methods to remove α-Gal (Park et al., 2009; Wong and Griffiths, 2014) and genetic material [endogenous retroviruses (van der Laan et al., 2000)] without affecting its biologic activity would need to be optimized before the decellularized tissue can be applied clinically in patients. Despite the reduced instructive capacity, decellularized NCM could serve as a cell-free approach or be employed in combination with cell-based therapies. The decellularized NCM could be used as a carrier or even as a primer during the production of the cells to enhance the effects of cell therapy alone.

Vesicle related proteins have been identified in porcine, canine, and human NCCM (Bach et al., 2016b) and NC-rich fetal bovine NP matrisome (Caldeira et al., 2017), indicating a role of extracellular vesicles (EVs) in NC biology. EVs are small lipid bilayer-enclosed particles released by cells, that have attracted attention as potential targets for the development of regenerative therapies in the cartilage/IVD field (Malda et al., 2016; Piazza et al., 2020). EVs serve in intercellular signaling, since they are protective carriers for biologically active molecules (e.g., mRNA, miRNA, DNA, protein, lipid) and have the ability to interact with cells in a context-dependent manner (Malda et al., 2016).

Within the context of the IVD, SHH, a key factor in notochord and IVD development, was found to be the cargo of two specific EV fractions from primary chick cells derived from the notochord (Vyas et al., 2014). These two EV fractions contained distinct sets of vesicular miRNAs and proteins with differential effects on the SHH target genes (Vyas et al., 2014), indicating that notochordal EVs play a complex role during embryonic development. A recent study demonstrated that EVs are abundantly secreted by porcine NCs and that these EVs exert anabolic effects upon canine and human NPCs derived from degenerated IVDs (Bach et al., 2017a). Thus, the involvement of EVs within the effects observed by NCCM warrants further investigation. It remains to be determined whether EVs also hold promise for symptom modification, e.g., by inhibiting inflammation, angiogenesis, neurogenesis, or catabolism. Further identifying the bioactive EV-associated cargo, how the micro-environment influences it, the subsets of EVs and their possible distinct biologic effects will provide a better understanding of the NC’s mode of action. This fundamental knowledge will instruct future cell-free regenerative therapies that employ either the native EVs or synthetic nanovesicles delivering specified EV-associated biomolecules (Malda et al., 2016).

The regenerative potential of NCs is evidenced by the increasing body of literature. NC-based approaches are considered promising for regenerative and/or symptom modifying therapies for LBP due to IVD degeneration. Further efforts to bring NC-based therapies from the bench towards the bedside have to involve alternatives to employing primary NCs. These strategies are hampered by the lack of consensus terminology in defining the cells residing within the NP from development to adulthood (see Box 2), and by the lack of more precise phenotypic descriptions that single-cell technologies are likely to enable in the near future. Such terminology does not only allow for improved communication within the scientific community, most importantly it will assist in defining the specific set of NP cell phenotypic markers. These markers are essential in defining the outcomes of fundamental and translational research. Even more so, these markers will be used to benchmark NC-based therapies being developed.

In the developing notochord, NC vacuoles are considered to be the driving force of axis elongation (Ellis et al., 2013). The NC vacuoles are presumably mechanoprotected by the caveolae, flask-shaped structures of the plasma membrane that express caveolins (Nixon et al., 2007; Sinha et al., 2011; Bach et al., 2016a) and by the ECM of the notochordal sheath, rich in collagens, proteoglycans (mostly chondroitin sulfate and heparan sulfate), laminins, and fibronectins (Yasuoka, 2020). Since the vacuoles play an important biomechanical role in the developing embryo, they might also have a biomechanical role in the postnatal IVD by maintaining intracellular pressure to resist mechanical loading (Hong et al., 2018). Since NC vacuoles are important organelles in the embryonic notochord and postnatal NP, they might also be involved in the NC-mediated regenerative effects. For instance, they may contain bioactive substances in their membrane or vacuole content. During IVD maturation, vacuolated NCs are replaced by non-vacuolated NPCs (Hunter et al., 2003). It is interesting to determine if and how the cargo changes during maturation of the NCs and transition into non-vacuolated NPCs. In addition to the cargo content, also the self-renewal ability of the NCs is a topic of debate. There is evidence that the presence of large intracytoplamic vacuoles is a limiting factor for cell proliferation (Joo Han Kim et al., 2009; Hong et al., 2018), which is important to consider when aiming for NC-based regenerative therapies. Taken together, studying the NC vacuoles may provide novel information, possibly enabling guidance for NC-based IVD regeneration. For this reason, future studies should elucidate the content, composition, and specific role(s) of the NC vacuoles. Lastly, it would be worthwhile studying the exact relationship between the EVs and vacuoles of NCs.

Considering the above-mentioned limitation in NC-based approaches, emerging avenues that hold promise involve the generation of NC-like cells from other cell sources, e.g., embryonic (Diaz-Hernandez et al., 2020) or induced pluripotent stem cells (ES and iPS cells) (Liu et al., 2014; Liu K. et al., 2015; Yongxing Liu et al., 2015; Tang et al., 2018; Sheyn et al., 2019; Xia et al., 2019; Colombier et al., 2020; Zhang et al., 2020). Protocols for differentiation into the NC-lineage are highly variable, but generally involve differentiating pluripotent cells towards mesodermal lineages through activation of WNT signaling followed by the transient induction of early embryonic notochordal markers. The published studies do not compare the generated NC-like cells with native NCs in terms of their regenerative capacity or marker expression. While the histologic analysis suggested the presence of vacuolated cells in two studies (Tang et al., 2018; Zhang et al., 2020); their presence needs to be confirmed by markers typically expressed on the membranes of vacuoles. In an in vivo study, iPS cell-generated NC-like cells were tested in a porcine model in which degeneration was induced by annular puncture (Sheyn et al., 2019). Although the iPS-derived NC-like cells were detected 8 weeks after injection by KRT18/19, T, and NOTO expression, histology did not show the typical vacuolated NC morphology. Moreover, the GAG or collagen content was not quantified in the treated IVDs, and therefore, regeneration of the porcine IVDs was not confirmed. Interestingly, in a parallel study, porcine NC-derived ECM promoted the differentiation of human iPS cells into NC-like cells (Liu et al., 2014), indicating that modifying the local niche can improve their differentiation. Lastly, the culture conditions optimal for NC preservation (see section Preservation of the NC phenotype) can also be used to further optimize the differentiation of pluripotent cells into NC-like cells. Taken together, ES/iPS-based technology to generate NC-like cells can be used for fundamental research to better understand NC development, physiology and regenerative therapies. This is an active field of research with exciting possibilities, but also with ES/iPS-specific and general advanced therapy medicinal product-related bottlenecks and challenges. Examples of ES/iPSC-specific challenges include the improvement of differentiation efficiency, further identification of distinct NC-markers, purification of the differentiated cells, and importantly, maturation into the desired NC phenotype that can thrive and exert therapeutic effects within the hostile degenerative disc environment.

Besides ES or iPS cells, the earlier mentioned local tissue-specific NPPCs (TIE2+ cells) can be considered as an alternative for NC-like cell generation. The relationship between NPPCs and NCs is not clear yet, since NCs can be either TIE2+ or TIE2 negative (Sakai et al., 2018). Moreover, NPPCs can be detected in very young, but also (to a lesser extent) in older human individuals (Sakai et al., 2012), an age at which NCs are not present in the disc anymore. Therefore, NPPCs are considered a specific cell population in the NP, and might not be part of the NC or NPC population per se. Also, the fact that NCs have a low (Joo Han Kim et al., 2009), while NPPCs have a high proliferation capacity (Sakai et al., 2012) provides evidence against the hypothesis that NPPCs represent the (entire) NC population. Therefore, TIE2+ NPPCs might be a more suitable tissue-specific cell source for NC-like generation, that might differentiate even more efficiently into the desired cell type than other types of stem cells.

Although this review focuses on LBP due to IVD degeneration, patients with related neck pain (Peng and DePalma, 2018) due to IVD degeneration can probably benefit from NC-based regenerative treatments in the future. Furthermore, during osteoarthritis (OA), the articular cartilage undergoes processes resembling late stages of IVD degeneration, such as hypertrophic differentiation and calcification (Rutges et al., 2010; Rustenburg et al., 2018). Since NPCs closely resemble articular chondrocytes and NCs can induce regenerative effects on NPCs, they may exert positive effects on chondrocytes as well. Confirming this hypothesis, NCCM was demonstrated to exert anabolic and anti-inflammatory effects on human OA chondrocytes (Müller et al., 2016). Moreover, a recent study tested the potency of NC-derived ECM as a bioactive regenerative lubricant for the osteoarthritic joint (de Vries et al., 2018a). This NC-derived matrix exerted anabolic and anti-inflammatory effects on healthy bovine articular chondrocytes and showed promising lubricating properties, comparable to hyaluronic acid (de Vries et al., 2018a). Future studies are warranted to determine the effect of NC-secreted substances on human OA cartilage explants, which may respond differently from (healthy) chondrocytes. Altogether, these two studies show that local application of NC-secreted substances could have potential as a treatment not only for IVD disease but also OA and perhaps even other (orthopedic) diseases.