In late adulthood of humans, the intervertebral discs (IVDs) may degenerate, underlying the most common cause of low back and neck pain, which affects two-thirds of the population (Choi et al., 2008; Peng and DePalma, 2018; Raj, 2008). There is no cure and few good prevention strategies for it; current surgical treatments for IVD defects are invasive and unsatisfactory due to low efficacy and high risk for serious complications (Kos et al., 2019). As a result, the healthcare cost of low back and neck pain in the USA is the highest among 154 health conditions, estimated at $134.5 billion in 2016 (Dieleman et al., 2020). This status quo is difficult to change due to an insufficient understanding of the basic biology of IVD development, homeostasis, and degeneration.
Each IVD has four structural components: one central gelatinous nucleus pulposus (NP); one cartilaginous annulus fibrosus (AF), which circles the NP; and two cartilaginous endplates with one positioned anterior and the other posterior to the NP (Lawson and Harfe, 2015) (Fig. 1). In IVD degeneration, the most affected of the four components is the NP, which acts as an elastic cushion to absorb mechanical impacts, thus playing a vital role in the functions of IVDs (Hunter et al., 2003a, 2004b; Lawson and Harfe, 2015; Palacio-Mancheno et al., 2018; Raj, 2008; Trout et al., 1982a). The critical physical property of the NP is largely determined by the extracellular matrix (ECM) secreted by NP cells, which are embedded in the ECM and cluster in 3-dimensional (3D) networks (Hunter et al., 2003a, 2004b; Lawson and Harfe, 2015; Palacio-Mancheno et al., 2018; Raj, 2008; Trout et al., 1982a). Under mechanical stresses, the NP deforms; consequently, NP cells are stressed (Fig. 1). Thus, the organization of NP cells should play an important role in coping with mechanical stresses. The proper organization of NP cells depends on cell adhesions, which can be described and studied from the perspective of orientational cell adhesions (OCAs).
The concept of OCAs was recently proposed to refer to cell–cell or cell–matrix–cell adhesions that define the relative orientations of the intrinsic polarities of neighboring cells; such orientational intercellular relationships underlie various cellular configurations for proper tissue morphogenesis (Zhang and Wei, 2022, 2023a, b). By coupling cell adhesions with cell orientations, the OCA concept offers a new perspective to understanding tissue morphogenesis. OCAs can be categorized according to the types of orientational intercellular relationships they define. Depending on cellular contexts, the same adhesion molecule can act as different types of OCAs to promote the formation of various tissue architectures. For example, in the established neural epithelium, the N-cadherin cell adhesion molecule is concentrated at the adherens junctions and plays an essential role in aligning neighboring cells in parallel configuration for cellular sheet formation, whereas in early neural rod, N-cadherin can also mediate opposing apical OCAs between the apexes of contralateral cells to facilitate their opposing configuration for the formation of the mirror-symmetric neural rod (Guo et al., 2018; Zhang and Wei, 2023a). N-cadherin, also called cadherin 2, belongs to the type I subfamily of the cadherin superfamily (Nollet et al., 2000). N-cadherin mediates versatile cell-cell adhesion via homophilic adhesion between themselves or heterophilic adhesion with other molecules at comparable adhesion strengths (Prakasam et al., 2006). Evidence suggests that N-cadherin also plays important roles in the development of the NP as well as its precursor the notochord, possibly by functioning as different types of OCAs.
The expression and function of N-cadherin in the NP have been an active research topic about IVD degeneration. Evidence suggests that the structural integrity and functionality of the NP are positively correlated with the expression levels of N-cadherin. From the perspective of the OCA concept, this review summarizes the current understanding and outstanding questions about the roles of N-cadherin in NP development and maintenance, focusing on three aspects: First, how the notochord may transform to the NP by adjusting OCAs, and how NP cellular architecture may be involved in NP maintenance. Second, how N-cadherin may promote the developmental transition from the notochord to the nucleus pulposus. Third, how N-cadherin may regulate NP maintenance and homeostasis. In sum, a better understanding of N-cadherin functions in NP physiology helps with developing effective prevention and treatment strategies for low back and neck pain caused by IVD degeneration.
The transition from the notochord to the nucleus pulposus
Evidence suggests that the vertebrate NP is derived from a morphologically distinct tissue—the notochord. The notochord is the defining midline axial organ of the phylum chordata, which traditionally includes three subphyla: tunicata or urochordata, cephalochordata, and vertebrates (Satoh et al., 2014). The anteroposteriorly elongated, rigid, and yet flexible notochord is composed of cells cohered in a tandem fashion as rod-shaped tissue; it supports embryos’ structural integrity and locomotion (Corallo et al., 2015; Stemple, 2005). In addition, the notochord is an organizer that secretes signals to orchestrate the development of many surrounding tissues (Placzek, 1995; Saude et al., 2000).
Notochordal morphogenesis can be divided into four stages according to the cytoarchitectural features unfolding along the morphogenesis: the notochordal plate, the two-cell-wide notochordal plate, the immature notochord of a file of cells of single-cell width, and the mature notochord which is encircled with an extracellular matrix sheath (Fig. 2). These four stages can be organized into two phases: a convergent extension phase, followed by a differentiation phase. During the convergent extension phase, the wide notochordal plate reorganizes to form the immature notochord via mediolateral cell intercalation. This process is governed by many apicobasal and planar cell polarity proteins (Balmer et al., 2016; Heisenberg and Tada, 2002; Jiang and Smith, 2007; Keller et al., 2000; Kida et al., 2007; Kourakis et al., 2014; Munro et al., 2006; Munro and Odell, 2002; Ninomiya et al., 2004; Shindo, 2018; Wallingford et al., 2002; Zallen, 2007). During the subsequent differentiation phase, immature notochordal cells differentiate and reorganize into the mature notochord via three modes that differ among chordates (Fig. 2): (1) In ascidians, notochordal cells reshape and reorganize to form a hollow tube, which contains an extracellular osmotically-pressured lumen to maintain mechanical rigidity (Dong et al., 2009; Jiang and Smith, 2007; Munro et al., 2006). (2) In frogs, rodents, and humans, notochordal cells take on a wedge shape in the transverse section to assemble into a solid rod-shaped tissue (Balmer et al., 2016; de Bree et al., 2018; Keller et al., 2000; Murakami et al., 1985; Shinohara and Tanaka, 1988). (3) In zebrafish, notochordal cells differentiate into two types of cells, organized in a center-surround configuration, with the inner vacuolated cells surrounded by non-vacuolated endothelial-like outer sheath cells (Glickman et al., 2003; Norman et al., 2018; Stemple, 2005). Regardless, the outcome of notochordal morphogenesis is the production of a rod-shaped tissue architecture, with its constituent cells organizing largely in a tandem configuration along the anteroposterior axis.
The vertebrate embryonic notochord is short-lived; in adults, the notochord has regressed and transformed from a single tissue entity to the segregated NPs. This transformation was demonstrated by two landmark cell lineage tracing studies in mice (Choi et al., 2008; McCann et al., 2012). In young NP, cells no longer align in tandem; rather, they cluster into 3D cellular networks (Fig. 1) (Hunter et al., 2003a, 2004b; Lawson and Harfe, 2015; Palacio-Mancheno et al., 2018; Raj, 2008; Trout et al., 1982a). Like notochordal cells, young NP cells have vacuoles. Besides the formation of the NP, the notochord also contributes to other aspects of spine development. Recent studies in zebrafish showed that the notochord sheath cells are segmented along the anterior-posterior axis into alternating domains: the cartilaginous domains, which align with the future IVDs, and the mineralizing domains, which recruit osteoblasts to form the vertebral bodies (Wopat et al., 2018). Despite these findings, one outstanding question remains to be answered: How does the tandem and continuous notochordal cellular configuration transform into the segregated and randomly clustered 3D networks in the NP?
The cellular organization in healthy, aging, and degenerated NP
In the NP of young animals of many vertebrate species, cells form 3D network clusters and are embedded in the extracellular matrix that is composed of largely water, collagen II, and proteoglycan aggrecan (Hunter et al., 2003a). In aged animals, NP cells are more dispersed and resemble chondrocytes morphologically, as they are small and do not contain the vacuoles observed in young NP cells (Hunter et al., 2003b, 2004a; Sakai et al., 2009). This change in cell morphologies is observed in many vertebrate species (Hunter et al., 2004a), suggesting that it is a common phenotype of NP aging. Although NP cells have lost the original 3D clustering architecture in aged NP, some long processes were observed to connect these cells (Errington et al., 1998). These processes can allow some intercellular communications between NP cells, but the level of mechanical coupling by intercellular adhesion is unlikely comparable to that of 3D cellular clustering in young and healthy NP. In addition to cell dispersion, the cellularity is also reduced in degenerated NP, further disrupting the 3D clustering architecture (Gruber and Hanley, 2002; Hunter et al., 2003a; Trout et al., 1982b).
The 3D clustering of NP cells is unique because such organization is not observed in the IVD AF and endplates. The function of this 3D cellular organization is not known. But it is tempting to speculate that it enables NP cells to sense mechanical stresses, as reasoned below. Under mechanical loads and multidirectional movements of the spine, the NP deforms in 3D and passes the mechanical stresses to NP cells. Thus, if NP cells need to and are responsible for sensing mechanical stresses in normal physiology, they may sense the stresses through mechanisms that are based on the stretching, compressing, shearing, or twisting of NP cells (Fig. 1). This demand may be fulfilled by the 3D clustering organization of NP cells because this organization can ensure that there are always some NP cells connected and aligned in the right orientation to sense a particular type of stress.
Thus, we hypothesize that NP cells sense mechanical stresses by monitoring conformational changes of intercellular adhesions, which can be dynamically affected by tensile, compressive, shear, and torsional mechanical stresses; this mechanosensation leads to changes in gene expression through mechanotransduction to adjust NP physical properties and handle mechanical stresses (Fig. 3). Such stress sensing capability is likely compromised or lost in aged and degenerated IVDs because the NP cells are dispersed in these NP (Gruber and Hanley, 2002; Hunter et al., 2003a, 2004a, b; Trout et al., 1982b) and lack the required intercellular adhesions needed for mechanosensation. This reduction in stress-sensing capability may make NP cells unable to adjust their gene expression under stress, thus compromising the maintenance of NP cells and ECM. This chain of events can become a vicious cycle, leading to the eventual failure of the NP.
According to the Yin-Yang philosophy, mechanical stresses may be either beneficial or detrimental to NP health, depending on the interactions between stresses and NP cells. For example, extended, static, and high mechanical loads can cause a reduction in the expression of NP cell genes, including N-cadherin, aggrecan, collagen II, Brachyury, laminin, keratin 19, glypican-3 (Li et al., 2017b; Zhou et al., 2018). Thus, knowledge about how NP cells sense and respond to stresses at the molecular level is central to understanding IVD health and degeneration.
N-cadherin in the development of the notochord
N-cadherin is expressed in the notochord in all vertebrate species examined thus far by in situ hybridization and/or immunostaining, including frogs (Simonneau et al., 1992), zebrafish (Warga and Kane, 2007), and chickens (Hatta et al., 1987; Inuzuka et al., 1991; Lin et al., 2014), rats (Hatta and Takeichi, 1986; Sakamoto et al., 2008), and mice (Radice et al., 1997), suggesting that N-cadherin has conserved functions in the vertebrate notochord. Loss of N-cadherin results in deformation of the notochord (Warga and Kane, 2007); this suggests that N-cadherin is required for notochordal cells to organize into the rod-shaped architecture, possibly by mediating tandem OCAs. To verify this OCA function of N-cadherin, we need to know where N-cadherin localizes at the cellular and subcellular levels. Such information is yet to be obtained because the membrane localization of N-cadherin makes it challenging to determine which of the cohered cells expresses N-cadherin and whether it mediates homophilic or heterophilic adhesion (Prakasam et al., 2006).
The involvement of N-cadherin in the notochord formation may also depend on the surrounding tissues because interactions between the notochord and surrounding tissues influence their development (Placzek, 1995; Saude et al., 2000). For example, the neural tube and the notochord are known to influence each other’s development (Corallo et al., 2015; Stemple, 2005). N-cadherin is expressed in the neural tissue and required for neurulation (Guo et al., 2018; Harrington et al., 2007; Hong and Brewster, 2006; Malicki et al., 2003; Radice et al., 1997; Warga and Kane, 2007). Thus, to what extent N-cadherin regulates notochordal development in a tissue-autonomous or tissue-non-autonomous fashion remains to be determined. Regardless, the tandem cellular configuration of the notochord is consistent with our hypothesis that N-cadherin may be part of the tandem OCAs that are responsible for adhering to notochordal cells head-to-tail along the anterior-posterior axis. Testing this hypothesis will require high-resolution imaging of N-cadherin during notochordal development.
N-Cadherin expression in the nucleus pulposus:
The transition from the notochord to the NP does not eliminate the expression of N-cadherin. Rather, N-cadherin has been identified in the NP of many vertebrate species, including humans (Zhang et al., 2020), cows (Minogue et al., 2010), pigs (Hwang et al., 2016), rats (Tang et al., 2012), and mice (Zhang et al., 2019). This conservation suggests N-cadherin plays an essential role in the NP. In fact, besides keratin 19, N-cadherin is uniquely expressed in the NP cells but not in AF cells and articular chondrocytes in many vertebrate species, even though these three cell types express some genes in common (Minogue et al., 2010; Sakai et al., 2009). By contrast, the expression profile of N-cadherin in the NP is similar to that of notochordal cells (Minogue et al., 2010). These expression profiles suggest that N-cadherin is a unique marker for NP cells in the IVDs and notochordal cells.
Theoretically speaking, one would expect that N-cadherin is not necessary for the NP’s cushion function, which depends mainly on the ECM. This is because the IVD endplates and AF can handle the same mechanical loads and stresses experienced by the NP, but they use N-cadherin-negative chondrocytes to maintain their ECM (Sakai et al., 2009). Thus, the conserved expression of N-cadherin in the NP suggests that NP cells use N-cadherin to mediate unique functions that cannot be performed by chondrocytes.
Roles of N-cadherin in the 3D clustering of NP cells
Given that N-cadherin can mediate cell adhesions, N-cadherin may contribute to random OCAs by localizing to NP cell membranes randomly, thus promoting the 3D clustering of NP cells. This hypothesis is consistent with the observations that suppression of N-cadherin by either antibodies or CRISPR mutations inhibits the formation of NP cell clusters in cell and tissue culture analyses (Hwang et al., 2016; Hwang et al., 2015; Niu et al., 2018; Palacio-Mancheno et al., 2018). Furthermore, in aged NP where expression of N-cadherin is significantly lower than that of juvenile NP, NP cells become dispersed (Hunter et al., 2003a, b, 2004b; Hwang et al., 2016; Hwang et al., 2015). Nevertheless, as far as we know, this function of N-cadherin in the NP has not been verified by in vivo approaches. Because the 3D clustering of NP cells is drastically different from the tandem configuration of notochordal cells, it is tempting to hypothesize that N-cadherin contributes to the architectural transition from the notochord to the NP by switching from mediating tandem OCAs to random OCAs. This functional switch may depend on changes in N-cadherin’s subcellular localizations, which we know very little about. Thus, understanding how N-cadherin mediates OCAs in the NP is expected to provide important insight into the development, maintenance, and degeneration of the NP.
N-cadherin as a mechanosensor in the NP
Besides cell-cell adhesion, what other biological functions N-cadherin may carry in the 3D NP cell networks remains unclear. As reasoned earlier, 3D cellular clustering may underlie the capability of mechanosensation. Given that N-cadherin may mediate random OCAs for such 3D cellular clustering, we hypothesize that N-cadherin at the random OCAs may carry out such mechanosensation (Fig. 3). This mechanosensation may be coupled to mechanotransduction via N-cadherin-mediated signaling (Yulis et al., 2018). Although it is unclear what signaling pathways are involved, evidence suggests that β-catenin-regulated signaling (Hwang et al., 2016) and PI3K/Akt-GSK-3β signaling (Li et al., 2017a) may be involved. The challenge of verifying whether these pathways are employed in vivo is the lack of animal models (Peng et al., 2006) because IVD degeneration takes time, thus making experimentation inhibitive in terms of time and financial costs. In addition, systemic loss of N-cadherin is embryonic lethal, hindering analyzing N-cadherin functions in the NP.
The gene expression regulation controlled by the supposed N-cadherin-based mechanosensation and mechanotransduction may be the molecular bases for the differential gene expression between NP cells and chondrocytes, which determine the different physical properties of these tissues. For example, in the human NP, the proteoglycan to collagen ratio (in the form of glycosaminoglycans to hydroxyproline ratio) is as high as 27:1, whereas it is about 2:1 in the cartilaginous endplates (Mwale et al., 2004). This makes the NP much more hydrated and less fibrous to better buffer the mechanical loads and movements imposed on the IVDs.
N-cadherin in NP degeneration:
The expression levels of N-cadherin are correlated to the NP health status. During aging, NP gene expression profiles change (Chen et al., 2006). N-cadherin and some other genes, such as SNAP25, KRT8, and KRT18, are expressed at higher levels in young and healthy NP cells (large and vacuolated) than in aged NP cells (small and non-vacuolated) (Chen et al., 2015; Lv et al., 2014; Minogue et al., 2010; Tang et al., 2012) (Hwang et al., 2016; Lv et al., 2014). This correlation suggests that N-cadherin may play an essential function in NP maintenance and homeostasis, particularly under mechanical stresses. Loss of N-cadherin may reduce or eliminate the intercellular adhesions between NP cells and thus the ability to sense mechanical stresses. This defect may render NP cells unable to adjust their gene expression to modulate cell organization and ECM properties to cope with the stresses. Supporting this notion, loss of N-cadherin reduces the expression of aggrecan, collagen II, and Brachyury (Hwang et al., 2016; Hwang et al., 2015). Injury to mouse tail IVDs can also reduce the expression of N-cadherin in the NP and cause NP degeneration (Zhang et al., 2019). However, this injury model may not represent the natural condition of injury, and it is difficult to evaluate to what extent it reflects the natural degeneration condition. Interestingly, N-cadherin is also reduced in the IVDs of human patients with hypertension (Chen et al., 2015). By contrast, overexpression of N-cadherin promotes NP cells’ resilience to senescence caused by compressive stresses, as manifested by increased cell proliferation, reduced expression of senescence markers p16 and p53, increased telomerase activity, and upregulation of matrix proteins collagen II and aggrecan (Hou et al., 2018; MING NIU 2018; Niu et al., 2018; Wang et al., 2017).
The extracellular domain of N-cadherin has the potential to interact with the ECM, which is known to interact with NP cells and promote the maintenance of the NP phenotype and the expression profiling of NP cells. For example, it has been shown that laminin-mimetic peptides that interact with integrins can stimulate in vitro cultured NP cells to increase the production of N-cadherin, collagen I and II, and aggrecan and the maintenance of NP cell phenotype, suggesting the interaction between the extracellular matrix and integrins play an important role in NP cells maintenance (Bridgen et al., 2017). But whether NP cell–matrix interactions also concern N-cadherin is an open question.