|Home | About | Journals | Submit | Contact Us | Français|
Efforts to enhance bone regeneration in orthopedic and dental cases have grown steadily for the past decade, in line with increasingly sophisticated regenerative medicine. To meet the unprecedented demand for novel osteospecific growth factors with fewer adverse effects compared with those of existing adjuncts such as BMPs, our group has identified a craniosynostosis-associated secreted molecule, NELL-1, which is a potent growth factor that is highly specific to the osteochondral lineage, and has demonstrated robust induction of bone in multiple in vivo models from rodents to pre-clinical large animals. NELL-1 is preferentially expressed in osteoblasts under direct transcriptional control of Runx2, and is well-regulated during skeletal development. NELL-1/Nell-1 can promote orthotopic bone regeneration via either intramembranous or endochondral ossification, both within and outside of the craniofacial complex. Unlike BMP-2, Nell-1 cannot initiate ectopic bone formation in muscle, but can induce bone marrow stromal cells (BMSCs) to form bone in a mouse muscle pouch model, exhibiting specificity that BMPs lack. In addition, synergistic osteogenic effects of Nell-1 and BMP combotherapy have been observed, and are likely due to distinct differences in their signaling pathways. NELL-1's unique role as a novel osteoinductive growth factor makes it an attractive alternative with promise for future clinical applications. [Note: NELL-1 and NELL-1 indicate the human gene and protein, respectively; Nell-1 and Nell-1 indicate the mouse gene and protein, respectively.]
Regeneration of bone, as with any other tissue, is a concerted process that requires precise combinations of cells, scaffolds, and relevant growth factors. In particular, osteochondrogenic growth factors have been recognized as crucial components of a favorable microenvironment for bone regeneration in vivo. In addition, controlled delivery at the appropriate window of time and to the correct cells is equally critical to the regeneration process. Existing growth-factor-based therapies have had moderate success, but are also associated with high cost and rare life-threatening adverse effects from various causes (Benglis et al., 2008). Therefore, the search is on for new growth factors for promoting bone and cartilage regeneration. This review focuses on NELL-1, a novel potent osteoinductive factor with relatively high specificity to the osteochondral lineage, which our laboratory has been studying for the past decade. We will describe (1) the discovery of the NELL-1 gene and the characterization of the NELL-1 protein and its role in skeletal development, (2) our ongoing efforts to harness the capacity of NELL-1 protein for regenerating bone in multiple ex vivo and in vivo models, (3) the current understanding of the mechanisms of action of NELL-1 in promoting bone regeneration and in the greater context of bone biology, and (4) future prospects of NELL-1 research in bone and cartilage tissue engineering.
NELL-1 (Nel-like molecule, type 1) was named after its similarity to a gene Nel which was strongly expressed in Neural tissues encoding a protein with Epidermal growth factor (EFG)-Like repeats without high homology to any known molecules (Matsuhashi et al., 1995). The Nel gene was cloned over a decade ago from a phage display screen of a 9-day-old chick embryonic cDNA library with a monoclonal antibody (Pr-28) against a chick DNA-binding protein fraction. Subsequently, the human homologs of Nel were found from a human fetal brain cDNA as part of the Human Genome Project. Two sequences homologous to Nel, named NELL-1 and NELL-2 (NCBI accession nos. D83017 and D83018, respectively), were identified (Watanabe et al., 1996). Our group simultaneously identified an 1800-bp fragment of human NELL-1 (GenBank accession no. U57523) that was not identified from a library, but instead was isolated from, and found to be up-regulated in, surgically resected human cranial bone tissues with unilateral coronal synostosis (UCS), a congenital cranial defect characterized by premature fusion of one of the sutures in the developing calvarium (Ting et al., 1999). This is believed to be the first report linking the NELL-1 gene to a human pathologic condition. With the additional finding of the transient expression of NELL-1 and NELL-2 in developing human B-cells (Luce and Burrows, 1999), it became clear that the expression of the NELL-1 gene is not restricted to brain or neuronal cells.
NELL-1 shares 57% nucleotide and 50% amino acid homology with Nel, much lower than the 75% nucleotide and 80% amino acid homology between Nel and NELL-2. Homology alignment between NELL-2 and a frame-shifted Nel sequence suggests that NELL-2 is very likely the human counterpart of chick Nel (Watanabe et al., 1996). Additionally, rat and mouse Nell-2 are each over 90% homologous to human NELL-2, revealing close interspecies conservation (Kuroda et al., 1999). In humans, the amino acid sequences of NELL-1 and NELL-2 are about 50% homologous, which suggests a close but distinct molecular structure and possible function for these two members of the Nel family (Luce and Burrows, 1999). Human NELL-1 has been mapped to chromosome 11 at 11p15.1-p15.2 and spans about 906 kb with 20 coding exons (http://www.ncbi.nlm.nih.gov) (Fig. 1A).
Nel expression was found in almost all tissues at different stages, starting in 2-day-old chick embryos, with the highest levels in the brain prior to hatching (Matsuhashi et al., 1995). The expression plateau was reached right before hatching, and was maintained at high levels in the brain after hatching. Non-neural tissues such as liver and kidney expressed very low levels of Nel after hatching, except for the retina, where Nel was expressed at levels similar to those prior to hatching. Notably, a transcript of 4.5 kb was detected by Northern blot in all tissues at all stages, indicating that chicken Nel produces only a single mRNA. Similarly, a single 3.6-kb mRNA transcript of human NELL-2 has been found to be highly expressed in fetal and adult brain and weakly expressed in fetal kidney, while a 3.5-kb transcript of human NELL-1 has been found to be expressed only in fetal and adult brain at relatively low levels compared with NELL-2, except for some barely detectable amounts in adult kidney. This suggests that there is a very specific and small population of cells that express the NELL-1 gene (Watanabe et al., 1996). In addition, an immature, unspliced NELL-1 transcript was reported in the primordial B-cells of human normal bone marrow (Luce and Burrows, 1999). Furthermore, our group has reported a splice variant of rat Nell-1 lacking the coding sequence of the fifth EGF repeat by RT-PCR of mRNAs from rat primary calvarial cells (Ting et al., 1999), and subsequently this and possibly other isoforms have been identified in human and/or mouse tissues and cell lines as well (http://uswest. ensembl.org and unpublished observations). Analysis of human, rat, and mouse NELL-1 cDNA revealed 95% nucleotide homology among the species (Watanabe et al., 1996; Kuroda et al., 1999).
The Nel gene product was predicted to have a histidine-rich domain, a cysteine-rich domain similar to the C-terminal domain of von Willebrand factor (vWF), five EGF-like repeats, and two more cysteine-rich domains. It was postulated to be a membrane-integrated protein, since it possesses a hydrophobic transmembrane domain close to the C-terminal of its last EGF-like repeat (Matsuhashi et al., 1995). Alternatively, a frame shift may have generated an extra EGF-like repeat and resulted in a secreted protein instead (Watanabe et al., 1996). Thus, the possibility that Nel exists as two different forms—membrane-integrated and soluble secreted—remains unclear. Recently, a new cytosolic isoform of Nell-2 in rat has been identified and confirmed resulting from differential splicing of exon 3 of Nell-2 (Hwang et al., 2007).
NELL-1 encodes a secreted protein containing an 810-amino-acid open reading frame with molecular weight of about 90 kDa before N-glycosylation and oligomerization. NELL-1 contains several highly conserved structural motifs, including a secretory signal peptide, an NH2-terminal thrombospondin-1 (TSP-1)-like module (overlapping with a laminin G domain), 4 chordin-like cysteine-rich (CR) domains (or vWF domains), and 6 EGF-like domains (Fig. 1B). It is suggested that the secreted rat Nell-1 is a phosphorylated homotrimeric oligomer with molecular weight over 400 kDa (Kuroda et al., 1999). Recombinant human NELL-1 is also an 810-amino-acid protein that shares 92.6% protein sequence homology with rat Nell-1 (Table 1). However, recombinant human NELL-1 expressed from CHO cells has a molecular weight of about 140 kDa under reducing conditions and over 700 kDa under non-reduced conditions by SDS-PAGE, suggesting that NELL-1 may be secreted as a pentamer (unpublished observations). Presumably, the coiled-coil structure at the 5′ end of the first vWF domain in NELL-1 may confer oligomerization similar to that observed in cartilage oligomeric matrix protein (COMP) (Kajava, 1996).
Interestingly, Nell-1 was recognized as a TSP-1-like molecule because of its TSP-1 heparin-binding domain at the N-terminus (Kuroda et al., 1999). However, it lacks some major TSP-1 motifs, including type I and type III TSP repeats, RGD-binding domains, and the C-terminal domain. Heparin binding is a characteristic of Nell-1 biochemistry. The EGF repeats in Nell-1 were identified as a key component for interaction with the PKC subunit, which was recognized as a novel mode of interaction between proteins containing EGF repeats and their corresponding factors (Kuroda et al., 1999). Thus, the splice variant of Nell-1 (from our previous study) which lacks the fifth EGF repeat may possibly influence this interaction by modulating Ca2+ binding to distinguish itself functionally from other isoforms. The vWF domains are also assumed to be involved in Nell-1 oligomerization and to mediate cell adhesion (Kajava, 1996). Notably, TSP-1 can bind and activate the latent form of transforming growth factor (TGF)-β1 (Bornstein, 1995). NELL-1/Nell-1 has not been proven to possess binding capacity to TGF- β1 (personal communication with Kuroda); however, the CR domains in Nell-1 suggest potential binding with BMP members of the TGF-â superfamily (Garcia Abreu et al., 2002).
Northern blot analysis revealed a barely detectable 4.0-kb transcript in E11 embryos, the levels of which gradually increased from E14 to E17 during mouse embryogenesis with a rat Nell-1 full-length cDNA probe (Kuroda et al., 1999). Recently, a single 3.5-kb transcript was identified in mouse embryos, which was expressed faintly beginning at E10 and increased from E14 to E18. This transcript was detected by Northern blot with a PCR-generated 1.92-kb cDNA probe from mouse ETS sequence matching human NELL-1 (Desai et al., 2006). From the middle to the late stage of mouse embryogenesis, Nell-1 expression is clearly higher in the head compared with the body; however, the brain is the only organ with Northern-blot-detectable levels of Nell-1 in the adult mouse. Analysis of these data suggests that Nell-1 plays critical roles in mouse development, particularly in neural tissues. Indeed, the massive apoptosis of neural cells in E15.5 transgenic mouse embryos overexpressing Nell-1 has been identified as a possible cause of acrania, a syndromic condition involving severe defects in both brain tissue and neural-crest-cell-derived calvarial bone plates (Zhang et al., 2006). However, the penetrance of this defect was relatively low compared with the skeletal abnormalities in the same transgenic mouse model of Nell-1 overexpression (Zhang et al., 2002). Although Nell-1 was ubiquitously overexpressed under CMV promoter control, the major pathological changes were limited to the skeletal system, with the craniofacial bones and cartilage particularly affected. The recapitulation of human craniosynostosis (CS) in this transgenic model strongly suggests that Nell-1 plays a pivotal role in craniofacial bone and cartilage development, and that its overexpression is a causative factor for CS. This finding is significant not only because it links Nell-1 to human CS, but also because it confirms Nell-1 function outside of neural tissues.
The generation of a gene deficiency model is a common, yet critical, step for functional study of a specific gene. The ENU-induced Nell-1-deficient mutant mouse has been an invaluable tool for the analysis of the in vivo effects of Nell-1 protein deficiency (Desai et al., 2006). The mutant allele is a nonsense point mutation truncating the Nell-1 protein from 810 to 502 amino acids and induces a rapid nonsense-mediated decay of its transcript. Homozygous mutants die perinatally, with no detectable level of Nell-1 mRNA. Post mortem characterization of the homozygous mutants revealed enlarged cranial vaults, short body length, and anomalous curvature in the cervical spine, without detectable abnormalities in other organs compared with their wild-type and heterozygous littermates. Skeletal staining revealed enlarged, but less mineralized, calvarial bones, compressed cervical intervertebral spaces, and severe deformity of the ribcage, which is believed to cause the death of Nell-1 mutants due to respiratory failure. Histological analysis also revealed that the amount of extracellular matrix in the intervertebral spaces of Nell-1 mutant mice was reduced compared with that in age-matched (P0) wild-type mice. Quantitative, high-resolution microCT analysis with emphasis on mineralized bone deformities revealed distinct differences in the calvarial, vertebral, and long bones between Nell-1-deficient newborns and their wild-type littermates (Zhang et al., 2008). The parietal bones were much thinner and the sagittal sutures were much wider in Nell-1-deficient mice compared with their wild-type littermates, and the prominent defects in these bones strongly resembled cleidocranial dysplasia mice with Runx2 haploinsufficiency (Komori et al., 1997). This work is the first demonstration that Nell-1 is indeed expressed in mouse calvarial bones during development, and RT-PCR revealed that the pattern was more specific than what had been described with mRNA from the whole head or body (Kuroda et al., 1999; Desai et al., 2006). There is a drastic decline of Nell-1 expression from E14.5 to E16.5 and E18.5 in the calvarial bone, in contrast to the gradual increase from E14 to E18 with whole-head RNA. Obviously, this discrepancy may be related to high Nell-1 expression levels in brain tissue, but it is worth reiterating that E14 is a critical time-point for osteogenesis of the calvarial bone plates (Ogle et al., 2004; Morriss-Kay and Wilkie, 2005), and thus Nell-1 may be required at higher levels than it would be at later stages. Analysis of these data clearly indicates that Nell-1 is a critical factor in the development and formation of intramembranous calvarial bones in mice (Zhang et al., 2008). Furthermore, detailed examination of vertebral and long bones by multiple approaches also revealed significant differences between Nell-1-deficient newborns and their wild-type littermates. Skeletal staining revealed that thoracic and lumbar vertebral bodies are smaller, less mineralized, with compressed intervertebral spaces in Nell-1-deficient mice. High-resolution microCT revealed less-mineralized trabeculae in the central zone of vertebral bodies in Nell-1-deficient newborns compared with their wild-type littermates. Histological analysis showed that Nell-1-deficient mice exhibit delayed ossification at the center of vertebral bodies and form fewer and thinner trabeculae compared with their wild-type littermates. In addition, the long bones from Nell-1-deficient mice exhibited cortical bone malformations. Thus, analysis of our data strongly indicates that Nell-1 is involved in the development of mouse bones formed through both intramembranous and endochondral ossification (Siu et al., 2009). Collectively, the data from the CMV-driven Nell-1 overexpressing and ENU-induced Nell-1-deficient models strongly suggests a pivotal role for Nell-1 in the development of the mouse skeletal system, and that Nell-1 is a novel osteochondrogenic molecule.
Bone and cartilage loss caused by traumatic injury, congenital defects, periodontal disease, surgical resection, and craniofacial reconstruction represents a significant biomedical burden. Defects that exceed the size at which complete healing occurs naturally are termed “critical” and require adjunctive therapies such as tissue grafts to heal properly. Autologous bone grafts from the iliac crest and calvarium are considered the most optimal source of bone graft material (Canady et al., 1993; Frodel et al., 1993). This, of course, results in an additional surgical site for the patient and causes donor site morbidity, including pain, neurologic defects, infection, hematomas, and gait disturbance and thigh paresthesia for iliac crest donor sites (Laurie et al., 1984; Kurz et al., 1989; Kline and Wolfe, 1995; Vail and Urbaniak, 1996; Sawin et al., 1998; Ahlmann et al., 2002; Silber et al., 2003), all of which contribute to increased surgical time and length of hospital stay. In addition, the amount of harvestable bone is finite, limited by the amount required for normal function at the donor site. To overcome these obstacles in current therapies for bone regeneration, tissue engineering is a rapidly evolving strategy in which the combination of a recombinant growth factor with the appropriate carriers and/or target cells offers the greatest promise of creating an autologous graft alternative that is both readily available and minimally infectious or antigenic. In fact, the autologous bone graft gold standard is already being challenged by commercially available orthobiologic molecules such as bone morphogenetic proteins (BMPs).
BMPs are potent inducers of bone (Urist, 1965), and approval of the INFUSE Bone Graft (based on BMP-2) and OP-1 (based on BMP-7) has allowed them to be used to substitute for autogenous bone while achieving equal or superior bone formation without donor site morbidity (Grauer et al., 2001; McKay and Sandhu, 2002). However, BMPs are involved in multiple developmental processes during embryogenesis (Ducy and Karsenty, 2000; Rivera-Feliciano and Tabin, 2006; Vogt et al., 2006), and this functional heterogeneity of BMPs can contribute to clinical complications. Adjunctive use of growth factors requires supraphysiologic doses to overcome diffusion away from the implantation site (Walker and Wright, 2002), further exacerbating the potential for adverse effects. Indeed, native bone resorption, implant fracture, soft-tissue swelling, osseous overgrowth, inflammation, and other complications at and away from implant sites have been linked to BMP therapy (Riew et al., 1998; Poynton and Lane, 2002; Haid et al., 2004; Shields et al., 2006; Smucker et al., 2006; Perri et al., 2007; Vaidya et al., 2007; Tumialan et al., 2008). Treatments with other growth factors have been explored, but most have limited osteoinductive capacity (Laurencin et al., 1999). Thus, a crucial component of bone regeneration research is to find a molecule that can specifically induce bone formation while avoiding excessive or ectopic bone formation.
As discussed earlier, the discovery of NELL-1 up-regulation in fused or prematurely fusing coronal sutures by our group was originally motivated by the search for local factors responsible for premature suture closure in human CS patients (Ting et al., 1999). Indeed, the recapitulation of human CS—without detectable defects in other tissues and organs—in CMV-driven Nell-1-overexpressing transgenic mice has further demonstrated that Nell-1 is a critical factor with relative specificity to promoting differentiation of neural-crest-cell-derived osteoblasts (Zhang et al., 2002), and that Nell-1 may potentially be an osteogenic growth factor with specificity higher than that of those currently used for bone tissue engineering. We tested this hypothesis in a rat calvarial defect model, a location closely matching the predicted tissue origin and site of Nell-1 function. Following the success of Nell-1-mediated bone regeneration in this animal model (Aghaloo et al., 2006), we have applied this technology to multiple animal models for the past decade and have since developed many successful bone tissue engineering systems. The pre-clinical animal models that have been used to assess the ability of Nell-1 and BMPs to form bone are summarized in Table 2, and their results are compared.
The calvarial defect model is a well-defined, widely used in vivo model for bone regeneration studies, especially for bone healing via intramembranous ossification processes (Aalami et al., 2004; Cowan et al., 2004). This model was ideal for testing the osteoinductive capacity of Nell-1, a growth factor involved in an anomaly of intramembranous cranial bones and capable of promoting osteoblastic differentiation of calvarial cells in vitro. Two 3-mm full-thickness craniotomy defects were created in each parietal bone of 3-month-old male Sprague-Dawley rats. Each defect was implanted with a control PLGA scaffold (containing only PBS) on one side and a PLGA scaffold coated with 200 ng of either Nell-1 or BMP-2 protein on the other side (Aghaloo et al., 2006). BMP-2 was selected as a positive control, since it is a benchmark bone growth factor that has induced successful healing in this model (Cowan et al., 2005). Live microCT imaging allowed us to monitor bone regeneration dynamically, and clearly demonstrated equivalently increasing bone formation for both Nell-1- and BMP-2-treated defects throughout a 12-week period. The healing patterns were similar as well, with defects starting to be filled with new bone growing from the uninjured edge of the parietal bone plates, rather than from the center of the defect. This result implies that both Nell-1 and BMP-2 stimulate existing osteoblasts or other potential target cells residing close to the defect edges to begin the process of intramembranous bone regeneration. Interestingly, Nell-1-treated defects also showed consistently greater mineralization within the defects at early weeks compared with BMP-2, although both were equivalent by week 12. This may be explained, at least partially, by our Osterix (Osx) expression data, which showed that BMP-2 induced much more Osx-positive cells than did Nell-1 in 2-week samples, possibly indicating BMP-mediated cell recruitment and proliferation rather than Nell-1-mediated osteogenic differentiation and mineralization. In addition, Nell-1 and BMP-2 may repair calvarial defects with comparable potency but through different mechanisms, which has been verified in our recent studies to be discussed later. In fact, a much lower molar dose of Nell-1 was applied in this model, due to the difference in molecular weight between BMP-2 (dimer MW: 26 kDa) and Nell-1 (oligomer MW: 400 kDa).
For further exploration of the possible applications of Nell-1 within the craniofacial complex, the ex vivo model of a rapidly distracted rat intermaxillary palatal suture was adopted for a comparative study with BMP-7 as positive control (Cowan et al., 2006). Rat intermaxillary sutures were mechanically expanded and cultured ex vivo in a chemically defined serum-free medium with 100 ng/mL Nell-1 protein, 200 ng/mL BMP-7, or PBS vehicle for a total of 8 days. MicroCT analysis of mineralized bone within the expanded suture area revealed significantly more new bone formation induced by both Nell-1 and BMP-7 compared with the PBS negative control. Secondary cartilage formation, assessed by Alcian Blue staining, was also remarkably accelerated along the osteogenic fronts of palatal bones, with great expansion within the sutures exposed to Nell-1 and BMP-7. In contrast, the carrot-shaped secondary cartilage in untreated controls remained well apart, similar to previous reports (Mizoguchi et al., 1992; Takahashi et al., 1996). Detailed examination of H&E-stained histological sections of both Nell-1- and BMP-7-treated samples revealed subtle differences in matrix deposition and mineralization and chondrocyte maturation within the sutures. However, more type X collagen-positive hypertrophic chondrocytes and BSP-positive dense connective matrix was observed in Nell-1-treated sutures compared with BMP-7-treated sutures. Of note, some areas within Nell-1-treated sutures appeared as chondroid bone, showing the involvement of intramembranous bone formation within existing cartilaginous tissues (Dhem, 2001). Significantly, this finding was the first demonstration that Nell-1 is capable of promoting new bone formation through an endochondral mechanism, in addition to the previously defined intramembranous mechanism (Zhang et al., 2002; Aghaloo et al., 2006). Although complete fusion of the palatal sutures was not observed, due to the limited long-term viability of this ex vivo model, the BMP-7-like osteoinductive capacity and potency of Nell-1 were highly encouraging, and the model demonstrated the versatility of Nell-1 in osteochondral differentiation and bone regeneration.
Another biotechnological approach to augment bone regeneration is ex vivo gene therapy, which involves transduction of target cells with osteoinductive genes to enhance bone repair, a strategy previously investigated for BMP (Lieberman et al., 1999; Chang et al., 2003). Bone marrow stromal/stem cells (BMSCs) are among the most frequently used target cells in gene therapies, due to their multipotency, and hold great promise for the future of regenerative medicine (Cancedda et al., 2003; Derubeis and Cancedda, 2004). However, the induced osteogenic differentiation capacity of BMSCs in humans and other higher mammals tends to decline drastically compared with BMSCs from rodents, which are the most frequently used in vivo models in published reports of bone regeneration (Derubeis and Cancedda, 2004). To stringently challenge the osteoinductive capacity of Nell-1 and test the feasibility of ex vivo gene therapy using an adenoviral Nell-1 delivery system in a higher mammalian model, we used an adenovirus vector (AdNell-1) to transduce the Nell-1 gene into goat BMSCs, both in vitro and in vivo, with transductions of BMP-2 adenovirus (AdBMP-2) and LacZ adenovirus (AdLacZ) as positive and negative controls, respectively (Aghaloo et al., 2007). Investigators implanted 5 × 106 BMSCs transduced with AdNell-1, AdBMP-2, or AdLacZ directly into the thigh muscle of nude mice, while the untransduced cells were also implanted as an additional negative control. As expected, all 7 mice implanted with AdBMP-2-transduced BMSCs formed large masses of ectopic bone after 4 wks, while untransduced and AdLacZ-transduced BMSCs formed no mature bone tissue except for some cartilaginous and fibrous tissues at the implantation site. Of the 6 sites implanted with AdNell-1-transduced BMSCs, 5 exhibited mature bone formation, but with much less volume after 4 wks compared with the AdBMP-2 group. Interestingly, microCT analysis of the AdNell-1-induced bone showed masses that were more solid and localized to the injection site compared with the AdBMP-2-induced masses. Although AdBMP-2-treated mice showed a larger average bone mass, bisected microCT images of the mass showed a hollow cavity with only a shell of calcified tissue. Also, while histological analysis of AdNell-1-treated samples showed areas of both mature and immature bone formation, AdBMP-2-treated samples had extensive amounts of fatty marrow tissue in addition to bone, which is consistent with other findings involving BMP-2 ex vivo gene therapy (Sugiyama et al., 2003, 2005). This study showed that Nell-1 is capable of promoting mature bone formation using BMSCs as target cells through an endochondral ossification mechanism in vivo, in addition to its effects of osteoblastic differentiation in vitro.
Given the structural differences in the new bone formed in this study, Nell-1 has the potential advantage over BMP-2 of achieving precision and efficacy in robust bone regeneration and may be able to overcome therapeutic challenges in some clinical conditions, such as restoration of maxillofacial bone defects, where more delicate bone regeneration strategies are necessary (Schilephake, 2002). This study also validated ex vivo gene therapy by adenoviral delivery of Nell-1 as a viable alternative to the application of recombinant Nell-1 protein. With respect to the differences in bone volume and structure of the regenerated bone induced by BMP-2 and Nell-1 in this model, it may be due to the availability and different responsiveness of target cells in the host to BMP-2 over Nell-1. BMP pleiotropy (Ducy and Karsenty, 2000) and relative specificity of Nell-1 to osteoblasts may account for these different outcomes. Thus, we became interested in whether Nell-1 and BMP-2 had any additive, synergistic, or complementary effects if they were applied together.
A significant drawback to most growth factors studied for bone regeneration is their lack of specificity to osteoblasts—even potent osteoinductive factors such as BMPs are not specific to osteoblastic cells. BMPs are involved in multiple physiologic processes and can have profound and specific effects on organogenesis (Hogan, 1996; Ducy and Karsenty, 2000). Specifically, BMP-2-deficient mice manifest an open pre-amniotic canal and/or cardiac malformation and die between days 7.0 and 10.5 of gestation (Zhang and Bradley, 1996), while BMP-7-deficient mice exhibit failed renal morphogenesis, potentially absent lens induction and eye formation, and perinatal lethality (Dudley et al., 1995; Luo et al., 1995). In contrast, Nell-1 was primarily associated with CS—a bone anomaly—and its known functions thus far have been specific to the skeletal system and supported by in vitro studies and in vivo overexpression and deficiency mouse models (Ting et al., 1999; Zhang et al., 2002, 2003; Desai et al., 2006). As described earlier, adjunctive use of BMP is susceptible to non-bone-related adverse effects, while Nell-1 may hold great potential to be a novel bone growth factor with high potency and specificity. Nell-1 may also be applied together with a relatively lower dose of BMP-2, to meet specific needs in bone regeneration while simultaneously reducing the risk of undesired effects.
Synergistically increased osteoblast differentiation and bone formation in AdRunx2 and AdBMP-2 co-transduced C3H10T1/2 cells have been described (Yang et al., 2003). Since Nell-1 is a downstream mediator of Runx2 activity (discussed later), we expected similar augmentation of BMP activity following Nell-1/BMP combotherapy. To examine this possibility, investigators transduced C2C12 myoblasts with AdLacZ, AdNell-1, AdBMP-2, or a combination of AdNell-1 and AdBMP-2, and assayed for osteoblastic differentiation in vitro. To evaluate bone formation in vivo, they injected the adenoviruses directly into the thigh muscles of nude mice (Cowan et al., 2007). As expected, AdLacZ did not affect alkaline phosphatase (Alp) activity, osteopontin (OPN) secretion, or mineralization of C2C12 cells in vitro, while AdBMP-2 up-regulated both Alp and Opn and promoted mineralization in C2C12 cells. Transduction of C2C12 cells with AdNell-1, in contrast, showed no sign of osteoblastic differentiation at any given dosage. However, a significant increase of Alp induction after 9 days and OPN secretion after 12 days was observed in the combined AdNell-1+AdBMP-2 group compared with the singly transduced AdBMP-2 C2C12 myoblasts. Furthermore, both AdNell-1+AdBMP-2 transduction and the conditioned medium enhanced C2C12 mineralization significantly after 4 wks. Remarkably, direct intramuscular injection of up to 1 to 4 × 109 pfu AdNell-1 failed to induce bone formation similar to that in AdLacZ controls after 8 wks, even though Nell-1 was successfully expressed in muscle and some stromal cells at the site of injection. In contrast, the combined AdNell-1+AdBMP-2 (5 × 108 pfu each) treatment group exhibited a rapid and significantly enhanced process of bone regeneration compared with the AdBMP-2 (1 × 109 pfu) treatment group. The mineralized bone volume in the combined group plateaued within 4 wks—at which time the bone in the AdBMP-2 group reached only half of its final volume. Analysis of these data suggests that, like Runx2, Nell-1 can enhance the responsiveness of C2C12 cells to BMP stimulation and synergistically promote bone regeneration, but cannot initiate osteoblastic differentiation of muscle cells to form ectopic bone by itself. Histology and immunohistochemistry have shown subtle differences in protein expression of the bone-related transcription factors Cbfa-1/Runx2 and Sox9 in newly regenerated bone tissue, which may suggest several possible mechanisms involving Nell-1 and BMP-2 in this bone regeneration model, to be described later.
Interestingly, BMP-2 and -7 and BMP-4 and -7 can act synergistically, but BMP-2 and -4 cannot (Zhao et al., 2005). Co-transduction of either AdBMP-2 or AdBMP-4 with AdBMP-7 synergistically increased Alp activity two- to four-fold in both C3H10T1/2 multipotential cells and the marrow stromal cell line ST2 (Franceschi et al., 2004). Furthermore, stable cell lines co-expressing BMP-2 and BMP-7 yielded BMP-2/7 heterodimers that exhibited 20-fold higher specific activity than BMP-2 or BMP-7 homodimers when added to an in vitro Alp induction assay (Israel et al., 1996). Thus, Nell-1 combotherapy with either BMP-2 or -7 may yield disparate results, possibly due to mechanistic differences when either BMP is used independently. Hypothetically, Nell-1/BMP-2 synergy, like BMP-2/Runx2 cooperativity, may be more effective than Nell-1/BMP-7 combotherapy in promoting bone regeneration. Thus, Nell-1, as a novel osteogenic factor different from BMPs in initiating bone regeneration, may mechanistically satisfy current clinical needs for a soluble factor that can potentially replace or augment BMP-based therapies in various clinical conditions. Our most recent study using a combination of BMP2 and NELL-1 on a clinically relevant calvarial defect model provided further support for this notion (Aghaloo et al., 2010).
Spinal fusion is a common procedure in the orthopedic clinical setting. As discussed earlier, the greatest challenges to spinal fusion are the low rate of solid arthrodesis achievable by standard autogenous bone grafting and donor site morbidity associated with collection of the autograft bone. The development of growth-factor-based therapies has improved spinal fusion strategies significantly. However, the rate of successful fusion has been less than optimal, and adverse effects have been major safety concerns.
We asked if Nell-1 would be a good adjunct or substitute for BMP-2 or BMP-7 in spinal fusion procedures. As proof of principle, we directly delivered AdNell-1 from a demineralized bone matrix (DBM) carrier in a rat posterolateral intertransverse process spinal fusion model (Lu et al., 2007b). In this study, two groups of 20 athymic rats received DBM carrier containing 1 × 109 pfu of either AdNell-1 or AdLacZ. The rats were killed at 6 wks and evaluated for spinal fusion by radiography, manual spine palpation, and high-resolution microCT. Results showed significantly higher rates of spinal fusion in AdNell-1-treated groups compared with controls, at 60% and 20%, respectively. This result is comparable with those of other studies using BMP-2 and BMP-7 gene transfer in a spinal fusion model evaluated after 8 wks (Zhu et al., 2004), demonstrating that Nell-1 is able to induce rat spinal fusion successfully as effectively as can BMP, when administered via adenoviral gene transfer.
Our success in achieving spinal fusion in rats with AdNell-1 and the availability of purified recombinant NELL-1 protein led us to move forward to a sheep interbody spinal fusion model as a large animal model more biomechanically similar to human spines (Wilke et al., 1997; Han et al., 2005). In the first set of experiments, 8 sheep divided into 4 groups (2 animals/group and 4 sites/group) underwent fusion at L3-L4 and at L5-L6 with a radiolucent vertebral spacer. The control group was filled with DBX, a sheep demineralized bone matrix product, alone, and the other 3 groups received sheep DBX mixed with recombinant NELL-1 at final concentrations of 0.3, 0.6, and 1.5 mg/mL within and around each vertebral spacer. At 3 mos after surgery, CT images demonstrated successful spinal fusion in 100% of samples implanted with the 0.6 mg/mL NELL-1 dose, while sheep implanted with DBX alone achieved only a 30% fusion rate. Notably, the fusion rate was reduced to ~80% for doses lower or higher than the “optimal” dose (0.6 mg/mL), presumably because 0.3 mg/mL NELL-1 was insufficient for achieving a higher fusion rate, or 1.5 mg/mL Nell-1 contained excessive impurities with perturbing effects. Compared with controls, the central bone within the spacers exhibited increased bone density and bone volume in NELL-1-treated specimens, as measured by microCT. Histological examination of new bone formation bridging the two transverse processes in the rat fusion model and the two vertebral bodies in the sheep fusion model gave definitive evidence of true bone fusion (Lu et al., 2007a). Overall, NELL-1 delivered from DBX may induce comparable, if not better, sheep interbody fusion than BMP-2 and -7 delivered from absorbable collagen sponges, as published BMP studies concluded at 6 mos rather than 3 mos (Magin and Delling, 2001; Sandhu et al., 2002).
Appropriate biomaterials are required both to form a physical scaffold on which cells can attach and grow and to provide a platform from which to deliver growth factors to facilitate cell growth. To achieve controlled and sustained in vivo delivery of recombinant NELL-1, we have developed a biomimetic apatite-coated alginate/chitosan microparticle system and demonstrated its efficacy using a rat spinal fusion model (Lee et al., 2009). The in vitro release profile of NELL-1 protein is significantly different between apatite-coated and apatite-uncoated microparticles—the cumulative release of protein bound to apatite-coated microparticles over 30 days was 15%, while 40% of bound protein was released in a burst from uncoated particles on the first day. We believe that an innovative delivery system specific to NELL-1 protein will be critical for its biological actions and for optimizing its potential application in the clinic; thus, more work needs to be done in this regard.
From the positive results in our various animal models, it became clear to us that NELL-1/Nell-1 is a novel osteoinductive molecule with great potential for future clinical applications. But how exactly does it work? Based on our laboratory's work, the current understanding of the mechanisms of NELL-1/Nell-1 in promoting bone regeneration can be summarized as follows:
Within the context of promoting osteochondral cell differentiation in vitro and bone formation in vivo, Nell-1 has exhibited a relatively high specificity to target cells, with several lines of experimental evidence from previous studies.
First, AdNell-1 transduction or recombinant Nell-1 protein stimulation can induce osteoblastic differentiation, apoptosis, and mineralization only in the pre-osteoblastic cell line MC3T3 and primary calvarial cells, but not in the fibroblastic cell line NIH3T3 and primary fibroblast cells in an appropriate differentiation medium in vitro (Zhang et al., 2002, 2003). This may be partially due to the preferential binding of Nell-1 to its putative cell-surface receptors expressed only on osteoblastic cells but not on fibroblasts (Fig. 2). Both modes of Nell-1 stimulation targeting cells of the osteogenic lineage induced significantly high expression of the late-stage osteoblastic markers OPN, osteocalcin (OCN), and BMP-7 gene and protein, as well as mineralization, the hallmark of complete osteoblastic differentiation. The expression levels of Alp and osterix (Osx), relative early marker genes of osteoblastic differentiation, were inhibited, while Nell-1 was overexpressed in MC3T3 cells (Aghaloo et al., 2006), implying that Nell-1 may act to promote bone differentiation at a relatively late stage and not earlier in the differentiation process. This is also true in our rat calvarial defect in vivo model, where Nell-1 promoted more bone formation after 4 wks (μCT bone area, 97 ± 3%; bone volume, 49 ± 11%), while BMP-2 promoted faster bone formation after 2 wks (BMP-2/Nell-1: μCT bone area, 78 ± 15%/60 ± 11%; bone volume, 36 ± 23%/20 ± 13%), but with slightly less bone formed after 4 wks (μCT bone area, 93 ± 8%; bone volume, 45 ± 12%). Concomitantly, the in situ level of Osx protein in Nell-1-treated samples was much lower than that in the BMP-2 group at an early (1 wk) time-point, but the levels of OCN, bone sialoprotein (BSP), and BMP-7 were not significantly different between the two treatment groups.
In addition, Nell-1 also induced pre-osteoblasts and chondrocytes to undergo apoptosis when the cells were transduced with AdNell-1 at a high multiplicity of infection (MOI) or given high doses of recombinant Nell-1 protein in vitro, (Zhang et al., 2003, 2006). Apoptosis is believed to be an important process during normal osteogenesis in vivo, as well as in craniosynostosis (Rice et al., 1999). In vitro work has shown that Nell-1-induced apoptosis of primary calvarial cells was dependent on dose and differentiation status. At an MOI of 50 pfu/cell, AdNell-1 induced 32.52 ± 7.83% cell death at 12 days post-transduction, while only 15.6 ± 5.63% at 20 pfu/cell and 13.6 ± 3.37% at 10 pfu/cell were observed. Nevertheless, the removal of ascorbic acid and β-glycerophosphate, the critical components necessary for in vitro osteoblastic differentiation, from the culture medium almost completely abrogated Nell-1-induced apoptosis in primary calvarial cells. Additionally, primary chondrocytes were also susceptible to Nell-1-induced apoptosis when a higher MOI of 200 pfu/cell was used for transduction. Significantly, the apoptosis of osteoblasts near the cranial suture osteogenic front and of the chondrocytes in the basal chondrocranium induced by exaggerated Nell-1 overexpression has been recognized as an important mechanism of the development of mouse CS in our transgenic model (Zhang et al., 2003, 2006).
Second, Nell-1 can drive the differentiation of BMSCs toward the osteochondral lineage in vitro and endochondral bone formation in vivo, but cannot transdifferentiate myoblastic C2C12 cells into osteoblasts in vitro or induce ectopic bone formation in muscle in vivo (Aghaloo et al., 2007; Cowan et al., 2007). BMSCs, also referred to as mesenchymal stem cells (MSCs), are multipotent under appropriate conditions (Owen and Friedenstein, 1988; Bianco et al., 2001). AdNell-1 transduction significantly promotes osteoblastic differentiation of BMSCs, as evidenced by elevated Alp activity and formation of mineralized nodules in differentiation medium. In contrast, AdNell-1 transduction of C2C12, a cell line with a predetermined myogenic lineage, cannot convert the elongated, tubular myoblast cells to spindle-shaped or cuboid osteoblast cells or promote Alp and OPN up-regulation or mineralization in differentiation medium. More importantly, intramuscular implantation of AdNell-1-transduced BMSCs resulted in the formation of cartilage and new mineralized bone starting at 2 wks post-implantation, while direct intramuscular injection of AdNell-1 did not induce ectopic bone formation even after 8 wks. These findings clearly demonstrate that Nell-1 selectively promotes cells in the osteochondrogenic lineage or multipotent cells to undergo osteochondral differentiation.
Third, Nell-1 has profound, specific effects on cells in the osteochondrogenic lineage compared with cells of other lineages during development. This notion was backed with the finding that, in the CMV-Nell-1 transgenic mice and ENU-induced Nell-1-deficient mice, defects were limited to the skeletal system (Zhang et al., 2002; Desai et al., 2006). Thus, it is highly likely that Nell-1 functions specifically and selectively on target cells in the osteochondrogenic lineage to promote osteochondrogenic differentiation and bone formation.
A bone-specific cDNA microarray analysis revealed that significant differential expression of osteochondrogenic genes occurred concomitantly with the beginning of mineralization of the AdNell-1-transduced MC3T3 cells at 12 days post-transduction (Zhang et al., 2002). Genes up-regulated more than two-fold included OPN, OCN, and BMP-7, all late-stage markers of osteoblastic differentiation. However, no differences in RNA expression were observed in Runx2; TGFβ-1, -2, -3; TGFβ receptors I, II, III; or FGF receptors I and II. This suggested that Nell-1 may act downstream of these critical molecules of bone biology (Zhang et al., 2002), or may operate in different pathways to promote osteoblastic differentiation at a relatively late stage. In fact, both TGFβ-1 and FGF-2 can significantly stimulate Nell-1 expression in primary calvarial cells in vitro (Aghaloo et al., 2006), which demonstrated that Nell-1 is transcriptionally downstream of these two important osteogenic growth factors and further implied that Nell-1 may act as a local factor critical to the development of human CS.
To study the regulation of NELL-1 gene expression, we analyzed the NELL-1 promoter and found 3 functional osteoblast specific-binding elements 2 (OSE2) sites in a 2.2-kb human NELL-1 promoter upstream of its translational start site (Truong et al., 2007). We thus concluded that NELL-1 was directly regulated by Runx2 through direct binding to the OSE2 sites in the promoter. In addition, in silico analysis and in vitro transfection further verified that the rat and mouse Nell-1 promoters also contain functional OSE2 sites in this study. The identification of NELL-1/Nell-1 as a direct downstream target of Runx2, a master gene controlling osteoblastogenesis and bone formation, is a key finding that warranted further investigation of the role of NELL-1/Nell-1 in skeletal development as well as bone tissue engineering.
The formation and function of osteoblasts are two of the most central concepts of bone biology and regeneration (Aubin, 2001). Osteoblastogenesis begins with mesenchymal stem cell (MSC) commitment to an osteoprogenitor lineage, and culminates with sequential differentiation to osteoblasts and osteocytes. Osteoblast function, in turn, depends on matrix deposition and bone formation by already-differentiated osteoblasts. Commitment of undifferentiated MSCs to an osteochondroprogenitor lineage is first marked by expression of the transcription factor Runt homology domain transcription factor-2 [Runx2, also known as corebinding factor-1 (Cbfa1)]. The Runx2 gene is essential for both osteoblast formation and function (Ducy et al., 1997; Komori et al., 1997) and has also been recognized to be critical for the maturation of chondrocytes (Adams et al., 2007). Runx2 null mice (Runx2-/-) exhibit arrest of osteoblast differentiation, resulting in complete lack of mineralized bone formation and perinatal lethality due to respiratory failure secondary to a cartilaginous ribcage (Komori et al., 1997). Notably, the levels of Nell-1 mRNA and protein detected by immunohistochemistry and real-time PCR were significantly reduced in craniofacial bones and primary calvarial cells in Runx2 null mice (Runx2-/-) (Zhang et al., 2008). Furthermore, ENU-induced Nell-1-deficient mice died of respiratory insufficiency secondary to a malformed ribcage (Desai et al., 2006) similar to Runx2 homozygous (Runx2-/-) mice. In contrast to Runx2-/- mice, the heterozygotes (Runx2+/-) manifest clavicular hypoplasia, delayed development of membranous bones, and delayed ossification of cranial bones, causing open anterior and posterior fontanelles, smaller parietal and interparietal cranial bones, and multiple Wormian bones (small bones in the sutures), a phenotype similar to cleidocranial dysplasia (CCD) in humans (Komori et al., 1997). Again, the calvarial bone defects in ENU-induced Nell-1-deficient mice are highly reminiscent of the Runx2+/- phenotype in CCD. By crossing Runx2+/- mice with CMV-Nell-1-overexpressing transgenic mice, we obtained Runx2+/-/CMV-Nell-1, which exhibited partial rescue of the calvarial defects associated with Runx2 haploinsufficiency (Zhang et al., 2008). This finding is strong proof that Runx2 exerts some of its functions through Nell-1 as a downstream mediator. In contrast, most described osteoinductive factors appear to function upstream of Runx2. For instance, BMP-2, BMP-7, IGF-I, and TGF-β1 are known to up-regulate Runx2 transcription (Lee et al., 2000; Tou et al., 2003; Koch et al., 2005; Hassan et al., 2006), which may cause more undesired effects than Nell-1 by affecting different downstream targets of Runx2 in the context of bone formation. Thus, these models have elucidated the direct regulatory relationship between Runx2 and Nell-1 within cells of the osteochondral lineage. More importantly, these models strongly suggest that Nell-1 may work as a key functional mediator downstream of Runx2 in control of osteochondrogenesis and maturation during skeletal development (Fig. 3). This functional relationship may also account for the high specificity and effectiveness of Nell-1 in promoting bone formation.
Structural analyses have revealed that the NELL-1 protein contains a well-defined signal sequence, making it a secreted factor and possibly a signaling molecule. Of the multiple motifs contained in NELL-1 (Fig. 1B), the N-terminal TSP-1 domain, the EGF-like repeats, and the cysteine-rich domains have the best potential of binding to their putative receptors or extracellular matrix molecules to initiate signaling cascades (Kuroda and Tanizawa, 1999). Most soluble factors involved in osteoblastogenesis and bone regeneration, such as BMP-2 and FGF, act on osteoblasts by binding to their cell-surface receptors to activate various intracellular signaling pathways (Hughes et al., 2006). The ability to produce highly pure recombinant NELL-1 protein from mammalian cells greatly facilitated the search for signaling pathways that NELL-1 may use to promote osteoblast differentiation (Cowan et al., 2007; Bokui et al., 2008). With our limited knowledge of proteins that bind to NELL-1 or its possible receptors, work on the NELL-1 signaling pathway has focused on MAPK, since it is the most active pathway elicited by various growth factors (Lee et al., 2002). A 100 ng/mL quantity of NELL-1 protein can significantly activate the ERK1/2 and JNK1/2/3 kinases in primary rat fetal calvarial (RFC) cells after only 10 min of stimulation, and this activation can happen transiently only within 30 min of stimulation in these normal osteoblasts. Nevertheless, the activation of ERK and JNK was verified to be responsible for the up-regulation of OPN by NELL-1 stimulation of RFC cells. siRNA-mediated degradation of Ras in RFC cells prior to NELL-1 stimulation can severely reduce OPN production, probably through the effective blockade of ERK activation. Interestingly, the selective activation of different MAPK pathways by NELL-1 depended somewhat on cell type. NELL-1 stimulation can activate all three major MAPKs transiently in RFC cells with different intensities, but activated both ERK and JUN transiently in multipotential C3H10T1/2 cells and only JNK1/2/3 in myoblastic C2C12 cells (Cowan et al., 2007; Bokui et al., 2008). In addition, NELL-1 stimulation of Saos-2 human osteosarcoma cells prolonged activation of ERK1/2 from 60 min to several hours.
BMP-2 primarily promotes osteoblast differentiation and mineralization by binding to its specific receptors BMPRII and BMPRI to activate Smad1/5/8 and Smad4 to further transactivate its downstream target genes (Heldin et al., 1997). Aside from the Smad pathway, BMP-2 can also stimulate osteoblast differentiation by activating the MAPK pathway (Gallea et al., 2001; Lee et al., 2002; Derynck and Zhang, 2003). However, NELL-1 does not appear to participate in the Smad1/5/8 pathway to initiate its intracellular signaling, regardless of cell type (Bokui et al., 2008), which may partially explain the functional and mechanistic differences between NELL-1 and BMP-2, despite their similar osteoinductive effects. It has been reported that FGF-2 may increase Runx2 activity through the MAPK pathway (Xiao et al., 2000, 2002; Ge et al., 2009), and activating mutations in FGF receptor-1 (FGFR1) dramatically increases Runx2 expression, causing premature cranial suture fusion in mice (Zhou et al., 2000). Similarly, stimulation of RFC cells with NELL-1 for 1 hr can drastically increase Runx2 phosphorylation, and the introduction of Ras siRNA can effectively decrease Runx2 phosphorylation (Bokui et al., 2008). Recently, we have assayed primary calvarial cells with a 6xOSE2 Runx2 reporter construct and confirmed that Runx2 phosphorylation induced by NELL-1 can significantly enhance its transactivation capacity (unpublished observations). Clearly, NELL-1 indeed modifies Runx2 bioactivity by enhancing its phosphorylation, which is thought to act upstream of the Runx2 signaling cascade (Fig. 4). However, this does not conflict with the fact that NELL-1/Nell-1 is a direct downstream target of Runx2, and that NELL-1 expression is tightly regulated by Runx2 through binding to the OSE2 sites in the NELL-1 promoter. Most likely, Runx2 phosphorylation by NELL-1 is a feedback loop for more precise regulation on cellular gene expression and function. It is also suggestive of how actively and closely NELL-1 participates in regulatory networks with Runx2 as a hub controlling osteochondral differentiation and bone formation (Lian et al., 2006) (Fig. 3).
As a novel and potent osteoinductive factor more specific to target cells in the osteochondrogenic lineage, NELL-1 has a bright future in the field of regenerative medicine, and in particular for bone and cartilage regeneration. The translational studies in mice, rats, and sheep have demonstrated that NELL-1/Nell-1 has the unique ability to promote adequate bone formation where it is needed, and in addition produces more robust bone. Recently, an in vitro study has shown that NELL-1 is capable of prolonging and maintaining the chondrogenic phenotype of isolated rabbit chondrocytes during in vitro expansion in a 3D culture (Lee et al., 2010). This is a step toward uncovering the role of NELL-1 in chondrocyte differentiation and finding the ideal delivery system of NELL-1 protein for cartilage regeneration. BMP-2 and BMP-7 are the only two bone growth factors currently approved by the FDA for use in specific orthopedic and dental cases, although severe adverse effects have been reported. Therefore, there is an urgent need for well-defined alternatives to better serve the unmet demands of the clinic. NELL-1, as a CS-associated protein having been found to be a novel osteoinductive growth factor, has many exciting applications in dental/craniofacial tissue regeneration applications beyond what has already been achieved in our pre-clinical models. With respect to safety, NELL-1 applications have not elicited exaggerated inflammation, soft tissue swelling, or severe immune responses with cGMP-ready purified NELL-1 (unpublished observations), and, due to its tumor suppression properties (Agaram et al., 2008), is not expected to induce malignancy.
Although NELL-1 has been found to use the ERK and JNK MAPK pathways during osteoblastic stimulation, the exact underlying mechanisms remain elusive. The identification of NELL-1 specific receptors or binding proteins in different tissues and cells is a very challenging, but important, goal for delineating the mechanistic reason for NELL-1 selectivity and specificity to cells of the osteochondrogenic lineage. The binding capacity to TGFβ superfamily members including BMPs, and subsequent effects on bone and cartilage regeneration, is a high priority of future research, because it may elucidate the mechanisms of synergistic or complementary effects between NELL-1 and BMPs. The direct downstream targets of NELL-1, including NELL-1 primary response genes, are also active topics of investigation. Finally, further research on signaling pathways activated by NELL-1 will be a critical piece of the puzzle that may help ‘connect the dots’ of identified actions of NELL-1 and allow us to draw a clearer picture of the NELL-1 network.
Growth-factor-based therapies for bone regeneration have enjoyed some success, but are not yet perfect. The most challenging obstacle is to find novel, potent osteoinductive molecules specific to osteochondral targets, to promote bone regeneration more precisely while minimizing adverse effects.
NELL-1, a protein originally associated with CS, is a promising candidate to fill this void. The identification, characterization, and application of NELL-1/Nell-1 in various in vitro and in vivo systems over the last decade have been our contribution to the search for alternative or complementary bone growth factors to fulfill clinical needs. Nell-1 is a developmentally regulated gene and plays major roles in the skeletal system, as shown by abnormalities observed in ENU-induced Nell-1-deficient and CMV-driven Nell-1-overexpressing mouse models. The osteoinductive potency and osteochondral specificity of NELL-1/Nell-1 in promoting bone regeneration have been well-demonstrated in animal models ranging from rodents to sheep. Although the detailed molecular mechanisms underlying the osteoinductive actions of NELL-1 are still a mystery, the Smad-independent Ras-ERK/JNK signaling pathway is believed to be primarily involved in NELL-1-mediated signal transduction. Identifying possible specific receptors for NELL-1, factors interacting with it, and direct downstream targets of it, are active projects in our laboratory. Finally, we are continuing to investigate other human conditions for which NELL-1-based therapies may be efficacious. A more complete understanding of the mechanism of NELL-1's action as a unique bone growth factor, either different from or complementary to known mechanisms, will open new doors for future applications of bone tissue engineering.
Work from the authors' laboratory has been supported by NIH grant R01 DE016107–01, and by UC Discovery Grants Bi005–10489 and Bi007–10677. Bone Biologics, Inc. licensed NELL-1-related patents from UCLA. X.Z., K.T., and C.S. are founders of Bone Biologics, Inc. and inventors of the related patents.