|Home | About | Journals | Submit | Contact Us | Français|
The ability to regenerate bone across a critical size defect would be a marked clinical advance over current methods for dealing with such structural gaps. Here, we briefly review the development of limb bones and the mandible, the regeneration of urodele limbs after amputation, and present evidence that urodele and anuran amphibians represent a valuable research model for the study of segment defect regeneration in both limb bones and mandible.
Injuries and diseases of the musculoskeletal system, including elements that support dental tissues, are among the most prevalent and disabling problems of the global human population. These include amputations of appendages, crush injuries, segment defects created by removal of shattered or malignant tissues, arthritis, bone and tooth loss due to periodontal disease, and temporal mandibular joint disease. Although mammals (including humans) can regenerate digit tips distal to the terminal phalangeal joint, they cannot regenerate digits amputated proximal to this joint, nor can they regenerate limbs or jaws, whereas urodele amphibians are able to regenerate these complex structures throughout life.1–3 Small defects in bones and associated muscles of both mammals and amphibians are regenerated by osteoprogenitor cells and satellite cells, but regeneration does not occur across defects larger than critical size (critical size defect, CSD).4–6 Several methods are used clinically to repair non-union fractures and CSDs in bones, including bone grafts, Ilizarov distraction and prosthetics.1,7 Alveolar bone damaged by periodontal disease can be induced to regenerate via stem cells of the periodontal ligament using a “guided regeneration” technique.8
Urodele and anuran amphibians are the only research models available to explore the mechanisms of appendage and jaw regeneration in tetrapod vertebrates. They also represent new models for research on the regenerative biology and medicine of CSDs. Amphibians have several advantages over rodents in this regard: they are easy and inexpensive to maintain and breed in the laboratory, microsurgery to remove bone segments and install scaffolds is simple, wound healing is rapid and requires no sutures, minimal post-operative care is required, morbidity and mortality are low, results are obtained relatively rapidly, and tissues are easy to harvest and process for histological or immunocytochemical staining or molecular analysis. Here, we review the use of amphibians as research tools for appendage and jaw regeneration, and for regeneration across CSDs.
The structure and development of bone has been well described9 and will be only briefly repeated here. Intramembranous bones (primarily the flat craniofacial bones) develop primarily by the direct differentiation into bone from neural crest-derived condensations of ectomesenchyme sandwiched between two periostea. In some cases, intramembranous bones also use endochondral bone formation. For example, in the mandible, Meckel's cartilage and secondary accessory mandibular cartilages ossify to form the major part of the mandible, condyle head and mental protuberance.
The long bones of the limbs develop by a combination of intramembranous and endochondral mechanisms. Condensations of mesenchymal cells first differentiate into a cartilage template surrounded by a periosteum. Progenitor cells of the periosteum then differentiate directly into osteoblasts that form a shell of intramembranous bone around the cartilage template. Next, the cartilage template is replaced by bone in endochondral fashion. The chondrocytes hypertrophy and die, their matrix calcifies, and is degraded by osteoclasts to create a marrow cavity. Periosteal capillaries accompanied by perivascular mesenchymal stem cells (MSCs) invade the cartilage template through the periosteal bone shell. The MSCs differentiate into osteoblasts, marrow stromal cells and hematopoietic stem cells. Each successive process, from mesenchymal condensation to ossification, begins in the center of the bone and progresses toward the ends. The shaft, or diaphysis, of the bone occupies most of its length. The ends of the bone ossify in their centers to become the epiphyses, leaving growth zones of cartilage around the ossification centers and between them and the diaphysis. The epiphyses become completely ossified after bone growth is completed.
Regardless of its mode of development, adult bone takes two histological forms. The first is compact bone, in which the osteocytes and bone matrix are arranged in Haversian systems around central canals. Compact bone lies directly under the periosteum. The second is trabecular (spongy, woven) bone, which extends inward from the compact bone. In endochondral bones, trabecular bone faces the marrow and is lined with an endosteum. Intramembranous bones have no marrow cavity and their interior is composed entirely of trabecular bone. The periosteum, endosteum and linings of the Haversian canals contain fibroblasts, MSCs, pre-osteoblasts and osteoblasts that can add bone by intramembranous ossification and repair fractures by either intramembranous or endochondral ossification.
Amphibian bones develop in the same way as mammalian bones. In tadpoles and larvae, periosteal bone forms around the cartilage templates of endochondral bones, but significant ossification does not take place until metamorphosis.5 Ambystoma mexicanum (axolotl), a favorite animal for limb regeneration studies, is neotenous, meaning that it matures to adulthood while maintaining a larval form externally, and its bones exhibit less ossification than urodeles and anurans that undergo metamorphosis to adult form.
Fractures of intramembranous bone are repaired by direct differentiation of MSCs in the periosteum into osteoblasts that re-form compact and trabecular bone. Regeneration of endochondral bones is accomplished primarily by MSCs in the periosteum, with lesser contributions from the endosteum and marrow stroma.10,11
Following fracture, blood vessels within and without the bone are torn, resulting in the formation of a fibrin clot (hematoma) in and around the break. Hypoxia results in osteocyte death for a limited distance on either side of the fracture. Platelets in the clot release PDGF and TGFβ, initiating an inflammatory phase in which the hematoma is invaded by neutrophils and macrophages. 11 Some of the macrophages in the bone marrow become osteoclasts that degrade the matrix of the dead bone.
Structural repair of the fracture largely duplicates embryonic development in both intramambranous and endochondral bones. In endochondral bones, periosteal MSCs differentiate on both sides of the fracture to osteoblasts (hard callus) in a process of intramembranous ossification.12 The osteoblasts secrete a bone matrix rich in Type I collagen, and containing osteocalcin, the mineralization-associated glycoproteins osteonectin, osteopontin and bone sialoprotein II (BSP-II), and numerous proteoglycans.13
Within the fracture space itself, regeneration is by the endochondral process. MSCs in the periosteum, endosteum and bone marrow proliferate to form a “soft callus.” These MSCs condense and differentiate into chondrocytes that secrete cartilagespecific matrix composed of Type II and XI collagens, aggrecan, hyaluronic acid and fibronectin. The chondrocytes undergo hypertrophy that is characterized by a switch to the production of Type X collagen and downregulation of the other collagen types. Subsequently, the cartilage matrix becomes calcified and the chondrocytes undergo apoptosis.11 Osteoclasts excavate the matrix of this calcified matrix template, and periosteal capillaries, induced by angiogenic factors produced by the hypertrophic chondrocytes, invade the matrix.12 The invading blood vessels are accompanied by perivascular MSCs that differentiate into osteoblasts, which replace the cartilage matrix with bone matrix.
Soft callus formation and chondrocyte differentiation are controlled by several transcription factors.14 Commitment to chondrogenesis of the soft callus requires expression by MSCs of the transcription factor Sox9, which induces the expression of genes for cartilage markers such as type II, X, IX and XI collagens and aggrecan.15 As the chondrocyte callus matures, Indian hedgehog (Ihh) transcripts are detected in chondrocytes and Gli1 transcripts are expressed in a population of cells on the periphery of the callus that will re-form the periosteum.16 In the embryonic development of endochondral bone, the products of these genes and of the genes for PTH and PTHrP, are part of a feedback loop that controls the rate at which chondrocytes mature.17–19 During the replacement of the cartilage template with bone, transcripts of genes encoding transcription factors active in osteoblast differentiation, such as Runx2 and osteocalcin, are detected.16
Growth factors essential for fracture repair are released from degrading bone matrix and are synthesized by soft callus cells of the regenerqating bone. Members of the TGFβ family are particularly important for chondrogenesis and osteogenesis in fractured bone.5,20,21 BMPs, which induce MSCs to commit to the chondro/osteogenic lineage, are released from degrading bone matrix and they and their receptors are also strongly expressed by soft callus cells.22,23 BMPs are detectable by antibody staining in cells of the soft callus as early as three days after fracture in a rabbit mandible model.24,25 Antibody staining increased as cartilage differentiated, and by two weeks after fracture osteoblasts were stained. BMP-4 transcripts are expressed in mesenchymal cells of the periosteum and marrow during the early phase of rib fracture repair, then disappear when chondrogenesis begins.26 Monoclonal antibodies to BMP-2, 4 and 7 exhibit increasing intensity of staining in periosteal mesenchymal cells in the region of hard callus formation, in the proliferating mesenchymal cells of the early soft callus and in the chondroblasts differentiated from these cells. Less intense staining is seen in maturing and hypertrophic chondrocytes, but staining is again intense in osteoblasts replacing the cartilage with bone.22 A mutation in the mouse short ear gene, which encodes BMP-5, results in congenital bone defects and a reduced capacity to repair fractures, suggesting that this BMP plays an important role in both bone embryogenesis and regeneration.27 BMPs are not expressed in uninjured bone, but immunolocalization studies show that their receptors, BMPR-IA and IB, are expressed in periosteal cells and are upregulated after fracture, parallel with the upregulation of BMPs.22
Northern blotting and immunolocalization studies indicate that high levels of TGFβ and FGF-1 and 2 are expressed during chondrogenesis of the soft callus, but not in the region of hard callus formation.28 FGF-2 and other members of the FGF family regulate chondrogenesis.29 TGFβ is present earlier in the hematoma and periosteum, but its source appears to be release from platelets and degrading bone matrix rather than synthesis by periosteal cells. PDGF and IGF-I are expressed in the soft callus, 30,31 suggesting that these growth factors are also involved in fracture healing.
Transcriptional profiling of intact vs. fractured rat femur by subtractive hybridization and microarray analysis has revealed that gene expression patterns change dramatically during fracture repair.32 Sixty-six percent of the genes expressed in fractures had homology to known genes. These genes were part of multiple families involved in the cell cycle, cell adhesion, extracellular matrix (ECM), cytoskeleton, inflammation, general metabolism, molecular processing, transcriptional activation and cell signaling. Consistent with studies on other regenerating tissues, members of the Wnt signaling pathway (Wnt 5a, Frizzled and β-catenin) were also identified.
It seems unlikely that molecules such as the Hoxa, Hoxd and T-box transcription factors, sonic hedgehog (Shh), FGF-4 and FGF-8, and Lmx1, which are involved in axial patterning of the skeletal condensations of the limb bud,33 play a role in endochondral fracture repair, or if they do, it is not a patterning role. This is because chondrocytic differentiation of the soft callus is taking place within an already delineated space and therefore probably does not require patterning.
Urodele amphibians such as the axolotl and newt are unique in being able to regenerate their limbs throughout life, although the process is slower with advancing age of the animal.34 Regeneration takes place in two major phases, the formation of a blastema that resembles the early limb bud, and the growth and differentiation of the blastema into the missing limb parts (Fig. 1). The ability of urodele amphibians to regenerate limbs and other complex structures has relevance for regenerative medicine, since were we able to understand the mechanism by which it occurs, we might be able to confer regenerative capacity on tissues and complex structures where it does not now exist.
Formation of the limb regeneration blastema is a reverse developmental process that takes place by histolysis of the tissues local to the amputation plane, resulting in tissue disorganization and liberation of individual cells. Histolysis is achieved by proteolytic degradation of the tissue ECM. The liberated cells undergo dedifferentiation to an earlier mesenchymatous state. Along with satellite cells from muscle (and probably MSCs from periosteum and endosteum) the dedifferentiated cells accumulate under the wound epidermis as a blastema. Simultaneously, the regenerating nerves induce the wound epidermis to thicken into the apical epidermal cap (AEC), a primary signaling center for epithelialmesenchymal interactions in the proximodistal axis of the regenerating limb.3,35–38
During histolysis and dedifferentiation, a high percentage of blastema cells enter the cell cycle and synthesize DNA, but only a tiny fraction undergoes mitosis. Signals from the AEC are thought to maintain the blastema cells in an undifferentiated state that allows them to synthesize DNA, while signals from the nerves promote their mitosis.39 Depriving the amputated limb of either epidermal or neural signals does not prevent histolysis and dedifferentiation, but a blastema does not form because the dedifferentiated cells do not accumulate and probably undergo premature redifferentiation.40,41 At the accumulation blastema stage, the epidermal and neural signals stimulate the proliferation of blastema cells and the blastema grows rapidly. The major nerve signal is the anterior gradient protein (AGP), which is first induced in Schwann cells by regenerating axons during histolysis and dedifferentiation.42 As the accumulation blastema forms and the axons enter the wound epidermis, they induce the expression of AGP by the gland cells of the AEC. Though not proven, the AEC appears to take over AGP expression from the Schwann cells and promote the proliferation of blastema cells throughout growth and differentiation of the blastema. The AEC is continually nerve-dependent for this function, for if the blastema is denervated at any time during its growth proliferation ceases.43 Consistent with this scenario is the fact that the gene encoding AGP supports the growth of the blastema to fingerbud stages when electroporated into the tip of a denervated limb during histolysis and dedifferentiation.42 Other important signaling molecules for blastema formation and growth are Fgf-2, 8, 10, BMPs and Wnt,3,44 but how they are integrated with AGP is not known.
The pattern of the replacement parts is specified during the formation and growth of the blastema, and is manifested in the spatial organization of the new musculoskeletal and connective tissues. Limb cells know their positional identity or address in three-dimensional space. The regeneration of missing structures in the circumference and proximodistal (PD) axis of the limb is known to require the confrontation of non-neighboring dedifferentiated cells.45,46 Thus proximal and distal boundaries must be established and confronted so that the gap in PD structure is recognized, leading to proliferation and intercalation of missing positional identities until a normal neighbor map is re-established.47
Retinoic acid (RA) is an important signaling molecule that regulates transcription of multiple target genes crucial to patterning during limb development and regeneration. When administered to regenerating axolotl limbs during histolysis, RA enhances the extent of histolysis and also induces dose-dependent proximalization of blastema cell PD positional identity via activation of the nuclear RA receptor RARδ2.48,49 This results in the serial duplication of limb segments (Fig. 2). The positional identity of blastema cells is manifested as a gradient of cell surface adhesion from distal to proximal, as shown by both in vitro and in vivo assays,50–52 and RA coordinately proximalizes blastema cell adhesion and positional identity in regenerating axolotl limbs.51–53
The adhesive gradient of blastema cells, the recognition of structural gaps by confrontation of non-neighboring cells that represent PD boundaries, and the boundary cell interaction resulting in the proliferation and intercalation of intermediate positional identities, all appear to converge on AGP and its cell surface receptor, a GPI-anchored protein called Prod1.3,42,54 Blastema cell adhesion and proximalization is abolished by antibodies to Prod1, or by its removal from the cell surface. Furthermore, there is a distal to proximal gradient of Prod1 in the limb, and RA treatment of blastema cells proximalizes the level of Prod1 expression.55 These observations imply that Prod1 plays an important role in recognizing neighbor vs. non-neighbor positional identity, and that when non-neighboring cells are confronted, this recognition allows Prod1 to interact with its ligand AGP to stimulate intercalary regeneration.
Several transcription factors, including Meis, HoxA and HoxD have also been identified as important PD patterning regulators during axolotl limb regeneration.1,2 RA proximalizes the level of expression of these genes, which probably act upstream of the Prod1 gene in the pathways that encode positional identity. RA undoubtedly affects genes involved in ECM degradation as well, and may affect the expression of genes involved in dedifferentiation and redifferentiation.
The events involved in RA-induced proximalization in regenerating axolotl limbs have not been investigated in detail. We have initiated studies using soft X-ray radiography and high-resolution quantitative micro-computed tomography (μCT) to facilitate detailed morphological characterization of the effects of a wide range of RA concentrations on regenerating axolotl limbs amputated through the radius/ulna, starting with doses known to be less than those causing recognizable dose-dependent skeletal duplications (Li B, unpublished results). The RA was administered by a new method, implantation of RA-impregnated poly(lactic-co-glycolic acid) (PLGA) polymer into a tunnel bored into the connective tissue of the dorsal fin.
Figure 3 shows radiograms taken at 45 and 73 days after PLGA implantation in animals treated with 10 or 20 µg of RA per gm body wt whose forelimbs were amputated through the mid-radius and ulna. At 10 µg, two cartilage protrusions developed at angles to the junction of pre-existing and regenerated ulna: one directed anteriorly from the dorsal ulna, and the other directed ventrally from the posterior ulna. At 20 µg, a single cartilage protrusion developed laterally from the junction between old and new ulna. Figure 4 shows quantitative µCT images of the regenerating forelimbs of animals treated with 10 µg RA/g body wt two months post-amputation. Open marrow cavities can be seen in both of the extra skeletal structures protruding from the ulna, suggesting their partial ossification. Masson's trichrome staining (not shown) on other limbs confirmed ossification, which revealed that blood-vessel invasion into the cartilage template took place at the sites of ossification. Further study with progressively higher concentrations of RA at closely spaced intervals after the start of RA administration will be needed to understand proximalization at the cellular level.
Limb regeneration can be studied directly by comparative global analyses of transcripts and proteins in regeneration-competent vs. regeneration-deficient systems. The undifferentiated limb buds of the anuran Xenopus laevis regenerate perfectly after amputation, but as the limb bud differentiates it loses the capacity to regenerate. The regeneration deficiency of amputated Xenopus laevis limbs is correlated with the formation of a fibroblastema instead of a mesenchymatous blastema. Fibroblasts migrate under the wound epithelium and proliferate, but differentiate into a symmetrical cartlage spike lacking muscle.56–58
Transcriptional and proteomic comparisons have been made between amputated regeneration-competent and deficient limb buds in Xenopus.59–61 Since it is not clear whether the events following amputation of an undifferentiated limb bud involve a reverse developmental process like that of a differentiated limb to form a blastema, or are a continuation of the original developmental process, or some combination of these, comparisons of blastema formation have also been made between the fully differentiated amputated limbs of juvenile axolotls and Xenopus froglets. Proteomic analysis in our laboratory revealed high levels of the centrosomal protein EVI5 throughout blastema formation in the axolotl.44 EVI5 functions in mammalian cells to stabilize the Emi1 protein, which prevents cells from entering the mitotic phase of the cell cycle while they replicate DNA.62 Both EVI5 and Emi1 are destroyed at G2, allowing cells to enter mitosis. We have hypothesized that high levels of EVI5 during blastema formation serve to prevent mitosis until the cells liberated by histolysis are fully dedifferentiated and have accumulated under the wound epithelium. In this hypothesis, the function of AGP is to signal for the destruction of EVI5 and Emi1, thus driving mitosis. Importantly, upregulation of EVI5 was not detected during fibroblastema formation in Xenopus froglet limbs (Rao et al., in preparation).
In our proteomic study we also identified the transcription factor Lin 28, the gene for which has been used in combination with the Oct4, Sox2 and Nanog genes to reprogram mammalian fibroblasts to induced pluripotent stem cells (iPSCs).63 In another study, Klf4, Sox2 and c-myc, which have also been used to derive iPSCs, were detected in regenerating newt limbs and lens.64 The ultimate goal of regenerative medicine is to initiate regeneration directly at the site of injury by reprogramming cells to an adult stem cell state that reproduces the original tissue. Since amputated urodele limbs know how to do this, unraveling how they do it will be useful in conferring regenerative capacity on mammalian appendages and other tissues.
A major orthopedic goal is to repair bone CSDs by inducing regeneration. The general experimental approach has been to bridge the CSD with a scaffold that is either seeded with osteogenic cells (MSCs, or osteoblast progenitors) or simply encourage the migration of host osteogenic cells into the scaffold. The scaffolds are made of either natural (e.g., collagen hydrogels) or synthetic (e.g., ceramic) materials and can be impregnated with growth factors or vectors with genes encoding these factors to enhance osteoblast differentiation. Most experimental studies on segment defect regeneration have been carried out on rat models. Even after many decades of study in these models, however, we do not have a scaffold/growth factor combination that can induce bone regeneration equivalent to the original.
Like mammals, neither urodele nor anuran amphibians can regenerate bone across a CSD. This has been shown for the cartilaginous radius of juvenile axolotls65,66 and for the tarsus of adult Xenopus laevis (Feng et al. submitted). BMP-2 soaked beads were able to induce cartilage regeneration across a CSD in the axolotl radius.66 Fibroblasts were the source of the regenerated cartilage. The fibroblasts apparently differentiated directly into cartilage without dedifferentiating, since they proliferated without expressing Prrx-1, a marker for the dedifferentiated cells that form the blastema in an amputated axolotl limb.67 Blastema cells derived by limb amputation and grafted into a CSD of the axolotl radius differentiated into cartilage that bridged the defect.66 Furthermore, deviating nerves into the CSD and removing the skin over the defect to create a wound epithelium, also resulted in regeneration. The proliferating fibroblasts that filled the gap and differentiated into chondrocytes expressed Prrx-1, suggesting that the nerve and wound epithelium provided signals that induce histolysis and dedifferentiation, similar to what happens in the amputated limb. These results show that knowledge of the mechanisms of blastema formation in amputated limbs can inform our attempts to regenerate cartilage or bone across a CSD.
Because adult Xenopus bones are more ossified than axolotl bones, a CSD made in these bones more closely approximates the mammalian condition. Recent experiments (Feng et al. submitted) showed that the tarsal CSD in Xenopus laevis was 35% of the length of the tarsus, as opposed to 20% in mammals. Untreated CSDs formed only fibrous scar tissue. Implantation of a biocompatible 1,6 hexanediol diacrylate (HDDA) scaffold loaded with BMP-4 and VEGF into tarsus CSDs resulted in complete regeneration. The scaffold did not act as a bone-forming substrate, but rather to deliver the growth factors over the whole length of the CSD. BMP-4 promoted the regeneration of a cartilage template within the gap, which was subsequently replaced by endochondral ossification promoted by VEGF. The source of the regenerated tissue was not traced, but is likely to be fibroblasts. Importantly, the regenerated cartilage and bone made anatomically correct connections with attached skeletal muscles and interosseus ligaments.
Although it is able to regenerate only a cartilage spike, formation of the fibroblastema is nevertheless dependent on the wound epithelium and nerves.68,69 We might thus expect that nerve deviation to, and creation of a wound epidermis over a CSD in a Xenopus long bone would be capable of supporting fibroblast migration into the defect and differentiation of cartilage to fill the gap. The same might then be true of a defect in a mammalian long bone, since it is known that mouse and human digit tips will regenerate if the amputation surface is left open to be recovered by wound epidermis.70,71 Ultimately, however, we wish to avoid as many surgical manipulations as possible in mammalian CSD regeneration; thus chemical stimulation of regeneration at the site of injury will be preferred.
To this end, axolotl and Xenopus CSDs can be used as platforms for rapid testing of different combinations of growth factors and delivery templates, or templates designed to simultaneously deliver growth factors and act as a physical substrate for cartilage and bone formation. Screening of combinatorial chemical libraries for synthetic small molecules that initiate the cascade of events leading to cartilage and bone regeneration will allow us to bypass the necessity of manufacturing the proteins that normally initiate these events. In this regard, low doses of RA might be useful in stimulating histolysis in the connective tissues bordering CSDs of urodele and anuran, or even mammalian bones to produce a population of cells that might initiate cartilage and bone regeneration in the absence of any other stimulus.
The mandible is an intramembranous bone that develops directly from mesenchymal tissue. The human mandible is derived from an intramembranous ectomesenchymal (neural crest) condensation lateral to Meckel's cartilage of the first pharyngeal arch at 5–6 weeks post-conception.72,73 Ossification begins at each side of the mandible in the region of the bifurcation of the inferior alveolar nerve and artery into mental and incisive branches. The ossification spreads upwards, dorsally and ventrally to form a trough for the developing teeth, the body and the ramus of the mandibles, respectively. Meckel's cartilage becomes surrounded and invaded by bone. Secondary accessory cartilages, which are dissociated from the primary Meckel's cartilage, emerge between the 10th and 14th weeks post conception to form the condyle head and the mental protuberance. In the mental region, on either side of the symphysis, small cartilages appear and ossify during the 7th month post conception to form a variable number of mental ossicles in the fibrous tissue of the symphysis. The ossicles become incorporated into the intramembranous bone when the symphysis menti is converted from a syndermosis into a synostosis. Eventually the mandible fuses into a single dentary bone, which bears the mandible teeth and parallels the maxilla/premaxilla of the upper jaw. Even before the mandible appears, the mandibular division of the trigeminal nerve develops in the region of the lower jaw, and may produce neurotrophic factors essential for inducing osteogenesis.
Ossification of the mandible is much reduced in amphibians. Almost all of the Meckel's cartilage disappears by the 24th week after conception in the human fetus, but is maintained in amphibians. Typical urodeles have three dermal bones, the major dentary, and two bones on the inner surface, which are probably the prearticular and coronoid.
Spallanzani described regeneration of the urodele mandible in 1768 (see Dinsmore74 for a historical perspective on Spallanzani's work). Modern research on regeneration of the salamander mandible can be traced to Goss and Stagg.75,76 Fifty-two years later, research on jaw regeneration in amphibians is scarce compared to limb regeneration, barely ten reports. This is not surprising, because limb structure is much simpler than that of the mandible, and it follows the trend in bone research that the study of endochondral bones has drawn more attention than that of intramembranous bones.
The amputated urodele mandible (and maxilla) regenerates via blastema formation, complete with teeth.75,77–80 Jaw regeneration takes longer than limb regeneration, about 24 weeks compared to only 4–6 weeks for the limb. Only the tooth bearing skeletal element, the dentary of the maxilla and mandible, is ossified 5 months after amputation.78 In both upper and lower jaws, the ability to regenerate the missing part depends on the level of amputation. Regenerative ability of the maxilla is impaired when the amputation site is performed distal to the eyes,81 or closer to the jaw articulation for the mandible.75,76 The blastema cells for mandibular regeneration are derived primarily from dedifferentiating muscle75 and are much like those of the limb, while cartilage also contributes to the maxillary blastema.79 The density of the blastema cells is highest on the cut ends of the mandible, where the alveolar bone, dental lamina, teeth and extra tooth buds are regenerated. The intermandibular blastema cells regenerate the muscle and connective tissue of the mandible, but the tongue and hyoid bone are not regenerated.
Both larval Ambystoma maculatum and adult newt mandibles regenerated alveolar cartilage and bone across the gap after removal of one-quarter or one-half of the mandible,77 suggesting that there is no CSD for regeneration in the urodele mandible. In A. maculatum larvae, the dental lamina containing the toothforming epithelium regenerated in both anterior and posterior directions over the regenerated bone, but the adult newt dental lamina exhibited polarity of regeneration, being regenerated only in a posterior direction.
Intermandibular soft tissue regenerates after removal in the newt.76 Removal of this tissue is followed by epithelial migration to cover the wound and the formation of a blastema on both lateral wound surfaces by dedifferentiation of remaining muscle and connective tissue. The two blastemas grow toward the midline and oral and skin epidermis fuse to re-establish continuity of these tissues. New salivary glands differentiate from the oral epidermis, while the mesenchymatous cells of the blastema enclosed by the oral and skin epithelia regenerate new muscle and connective tissue. In this case, cartilage thought to represent the hyoid bone also regenerates from these cells.
Jaw regeneration after amputation in urodeles exhibits some differences from limb regeneration. Firstly, the cytokeratin NvKII is expressed by the wound epidermis of regenerating newt limbs, but not regenerating jaws.79 Secondly, RA fails to proximalize the blastema cells of the regenerating mandible, but does induce truncation of the regenerating maxilla with a cleft lip and palate morphology.81 Lastly, blastema formation in the amputated mandible is not dependent on innervation.82 The wound epidermis thickens into a structure resembling the AEC of the limb, but whether this epithelium performs the same function as in limb regeneration is unknown. No studies of mitotic index have been done to determine whether the blastema formed by the regenerating urodele mandible is by accumulation or mitosis.
Kurosaka et al.80 have compared the regeneration of the Xenopus tropicalis mandible with that of the newt. BrdU labeling showed that cells of both species entered the cell cycle within a few days after amputation. However, only partial cartilage regeneration occurred in the frog mandible, compared to full restoration of the mandible in the newt. The difference in regenerative capacity was correlated with the expression of myosin heavy chain (MHC) expression and the presence of Pax-7 positive satellite cells in the newt mandible, but not in either the intact or regenerating frog mandible. Interestingly, Pax-7-positive satellite cells are present in the froglet limb, though they do not contribute to the fibroblastema.83 Clearly, our knowledge of how the blastema is formed in jaw regeneration is very limited compared to the limb. What is the origin of the progenitor cells that give rise to the regenerate? How is the pattern of the regenerating jaw determined? Does jaw regeneration share similar mechanisms with limb regeneration?
The “holy grail” of regenerative medicine is to chemically induce the regeneration of damaged tissues or whole complex structures in situ, without resorting to any in vitro culturing and reprogramming steps. Known mammalian regenerative mechanisms in situ include compensatory hyperplasia (liver, pancreas), and activation of resident stem cells, but not dedifferentiation or transdifferentiation (the switch to a cell phenotype other than the one of origin after dedifferentiation).
Amphibian limbs and jaws have a natural mechanism to reprogram differentiated cells to an earlier mesenchymatous state of differentiation after amputation that allows them to re-establish the equivalent of a limb bud or jaw rudiment that replaces exactly the structures that were removed. In addition, the blastema cells derived from dermal fibroblasts by this process can undergo a limited transdifferentiation.53 Thus these organisms are valuable models to help understand on a molecular level how to make mammalian cells undergo a similar limited re-programming directly at the site of an injury, thus eliminating the need to re-program cells to a ESC state, proliferate them in culture, and then direct their differentiation to the desired cell type for transplantation, all of which is time-consuming, expensive and not 100% efficient.84 In addition, identification of the natural molecules involved in this regenerative ability will allow an informed search of combinatorial chemical libraries for synthetic small molecules that can trigger the appropriate molecular cascades that activate dedifferentiation and transdifferentiation.
Despite their ability to regenerate whole limbs and jaws, the fact that urodele and anuran amphibian limb bones are like mammalian limb bones in being unable to regenerate across a CSD, makes them an inexpensive research model to test various combinations of scaffolds, growth factors and synthetic small molecules that will promote regeneration across such a gap. Factors essential for whole limb regeneration may well be useful in devising ways to promote CSD regeneration in both amphibians and mammals. We would also like to know how it is that up to half the urodele mandible can regenerate when removed, and how this knowledge might be applied to regenerating across CSDs in mammals like ourselves.
We thank our colleague Nandini Rao for critique of the manuscript, and the W.M. Keck Foundation and the U.S. Army Research Office (Grant no. W911NF07-10176) for their research support to the laboratory of D.L.S.
Previously published online: www.landesbioscience.com/journals/organogenesis/article/12039