Birth Defects Research (Part C) 84:265–280 (2008) REVIEW

Mammalian Regeneration and Regenerative Medicine

Ken Muneoka,* Christopher H. Allan, Xiaodong Yang, Jangwoo Lee, and Manjong Han

Mammals are generally considered to be poor regenerators, yet there are is driven by (1) the potential of a handful of mammalian models that display a robust ability to regener- adult stem cells to participate in ate. One such system is the regenerating tips of digits in both humans the formation of various organ and mice. In vitro studies of regenerating fetal human and mouse digit systems when introduced in early tips display both anatomical and molecular similarities, indicating that the embryos (Jiang et al., 2002), (2) mouse digit is a clinically relevant model. At the same time, genetic stud- the feasibility of transforming ies on mouse digit tip regeneration have identified signaling pathways required for the regeneration response that parallel those known to be adult cells into pluripotent stem important for regeneration in lower vertebrates. In addition, recent stud- cells (Yamanaka, 2008), and (3) ies establish that digit tip regeneration involves the formation of a blas- the isolation and characterization tema that shares similarities with the amphibian blastema, thus estab- of adult multipotent stem cells lishing a conceptual bridge between clinical application and basic research from virtually every tissue of the in regeneration. In this review we discuss how the study of endogenous body (Crisan et al., 2008). The regenerating mammalian systems is enhancing our understanding of second approach involves a bioen- regenerative mechanisms and helping to shed light on the development gineering strategy in which a sub- of therapeutic strategies in regenerative medicine. Birth Defects strate or scaffold is introduced Research (Part C) 84:265–280, 2008. VC 2008 Wiley-Liss, Inc. that can either be infiltrated by host cells (Badylak, 2007), or Key words: regeneration; mammal; digit; finger; blastema; ossification seeded with selected cells before implantation (Howard et al., INTRODUCTION and on the body’s endogenous 2008). This approach includes the In this posthuman genome/post- ability to repair wounds following in vitro engineering of specific tis- animal cloning era of modern biol- injury on the other. The goal of re- sues for use in transplantations, ogy, many have turned their generative medicine is to be able and in doing so sidesteps the attention to the prospect of con- to replace adult body parts on problems associated with tissue trolling the regeneration of tissues demand, and to this end we can morphogenesis and patterning or organs that do not regenerate identify three general avenues that are key to the successful in humans. Successes in this new being taken for the development regeneration of injured body field of Regenerative Medicine of novel regeneration therapies. parts. However, it does introduce would have enormous impact on The first is a cell based approach. a secondary problem of integrat- current medical practices and, as This approach has grown largely ing an engineered tissue with the well, on the general quality of from the successes in the use of host that still needs to be human life. Regenerative medicine hematopoietic stem cells in cell addressed (Khan et al., 2008). is strongly influenced by break- replacement therapies for the cure The third approach is to study throughs in our understanding of of blood diseases (Bhattacharya naturally regenerating models for organ and tissue formation during et al., 2008). The potential to comparison with nonregenerating embryogenesis on the one hand, expand into other organ systems injury wounds to discover critical

Ken Muneoka is from Division of Developmental Biology, Department of Cell and Molecular Biology, Tulane University, New Orleans, Louisiana and The Center for Bioenvironmental Research, Tulane University, New Orleans, Louisiana. Christopher H. Allan is from Department of Orthopaedics and Sports Medicine, University of Washington, Seattle, Washington. Xiaodong Yang, Jangwoo Lee, and Manjong Han are from Division of Developmental Biology, Department of Cell and Molecular Biology, Tulane University, New Orleans, Louisiana. Grant sponsor: National Institutes of Health; Grant numbers: R01-HD043277; P01-HD022610 *Correspondence to: Ken Muneoka, Division of Developmental Biology, Department of Cell and Molecular Biology, Tulane University, New Orleans, LA 70118. E-mail: [email protected] Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bdrc.20137

VC 2008 Wiley-Liss, Inc. 266 MUNEOKA ET AL. factors necessary for a regenera- with a focus on limb or tail regen- ofablastemathatmediatesthe tion response, and this is the pri- eration in adult urodeles (Brockes regeneration response. mary topic of this review. This and Kumar, 2005; Tanaka, 2003) approach represents a long history or larval anurans (Slack et al., EARS AND DEERS of experimental inquiry focused 2008). While mammals lack simi- largely on invertebrate models that lar regenerative capabilities, there The ears of some mammals are have enhanced regenerative ability are a handful of model systems in able to undergo scar-free healing (Sanchez Alvarado and Tsonis, which a variation of appendage and regeneration after an exci- 2006) and selected vertebrate regeneration has been described, sional hole punch that removes a groups that possess the ability to and these models provide a cylindrical mass of tissue including regenerate structures such as the glimpse at the limitations and epidermis, dermis, muscle and limb and tail (Brockes and Kumar, potential for regeneration in cartilage. This response in mam- 2005; Gardiner, 2005). Included in humans. These include the closure mals was first characterized in this category are studies on the of excisional tissues in ears rabbit ears and later shown to be developing appendages of mam- following hole punch in rabbits a characteristic not restricted to mals, birds, and frogs which pos- and mice (Metcalfe et al., 2006), lagomorphs (Williams-Boyce and sess regenerative ability that is lost the annual regeneration of antlers Daniel, 1986). In recent years as the animal matures (Muller in deer (Price et al., 2005; Kier- research on ear hole punch regen- et al., 1999). Although the leap dorf et al., 2007), and the regen- eration has been stimulated by the between regenerating systems and eration of amputated digit tips finding that different mouse human therapies may seem large, known to occur in humans and strains display variability in this there is substantial evidence that rodents (Han et al., 2005). regeneration response (Clark signaling pathways important for Although the regenerative capa- et al., 1998; Kench et al., 1999; regeneration have been conserved bility of mammals does not com- Li et al., 2001), raising the possi- through evolution (Sanchez Alvar- pare with that of amphibians, it is bility that the genetic basis of this ado, 2000; Brockes et al., 2001), critical to keep in mind that these variation might be uncovered and there are examples of specific mammalian models provide (Heber-Katz, 1999). In mice, a 2- signaling pathways or genes that insight into how successful regen- mm diameter hole punch under- are important for regeneration in eration can be accomplished goes re-epithelization that in- both traditionally regenerating and within the context of a warm- volves epidermal closure from the nonregenerating organisms (Taylor blooded terrestrial animal with two opposite surfaces of the ear, et al., 1994; Yokoyama et al., similarities to humans. Lessons and centripetal filling in of the hole 2000; Beck et al., 2003; Han et al., learned from such examples are is driven by growth of a blastema- 2003). In addition, there are a likely to provide important insight like structure that forms between handful of mammalian systems into how to effectively modify the the existing ear tissue and the that can successfully regenerate. human wound environment to wound epidermis. The MRL strain, Studies of these systems are pro- elicit an enhanced regenerative also known as the healer strain, ducing important insight into the response, or to establish a func- displays the highest level of regen- feasibility of an enhanced endoge- tional interface with a bioengi- erative ability described (Heber- nous regenerative response in neered or artificial organ or struc- Katz, 1999); however, even in this humans, and also in the design of ture. The mammalian ear punch strain complete regeneration does alternative strategies in regenera- and the deer antler models have not always occur (Rajnoch et al., tive medicine, particularly to recently been reviewed (Price 2003). The regeneration process is address the problem of integration et al., 2005; Metcalfe et al., characterized by a wound healing with host tissues (see Pendegrass 2006; Kierdorf et al., 2007), so a response that involves formation et al., 2006). In this review we very brief introductory overview of a specialized wound epidermis highlight studies on appendage is provided here and the reader is that integrates the epidermal regeneration in mammals, with directed to these excellent re- layers from the inner and outer particular emphasis on the regen- views. We will focus most of our ear surfaces. This wound healing erating digit tip, in the context of attention on the human and response is influenced by the na- how such efforts may impact the mouse digit models that we have ture of the injury and the degree development of successful thera- explored over the past few years. of trauma ear tissues experience pies in regenerative medicine. On the one hand the mouse digit (Rajnoch et al., 2003). This has models share anatomical and mo- led to the suggestion that a high lecular similarities with human degree of trauma leads to a regen- MAMMALIAN MODELS OF fingertip regeneration making it erative response, whereas a low APPENDAGE clinically relevant, and on the degree of trauma results in a rep- other hand, digit tip regeneration is arative response (Metcalfe et al., REGENERATION comparable to other well studied 2006). The existence of a dichoto- Appendage regeneration has been regeneration models in that wound mous switch that triggers an epi- studied primarily in amphibians healing culminates in the formation morphic regeneration response

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 267 versus a wound healing response is an important consideration in the development of regeneration therapies and needs to be further investigated. The wound healing process results in the formation of a blastema-like structure that is continuous around the margin of the wound. The blastema is com- posed of proliferating cells and as it grows centripetally, the punch hole eventually closes and the epi- thelial surfaces fuse. Re-differen- tiation of ear cartilage occurs by extension of the existing cartilage sheet or by the differentiation of cartilaginous islands at the base of the blastema (Rajnoch et al., 2003). In some cases ectopic bone formation has been described dur- ing redifferentiation (William- Boyce and Daniel, 1986), suggest- ing that the cells involved in this response are multipotent and re- Figure 1. Fingertip regeneration in humans. (A, B) A fingertip injury of a 7-year old sponsive to the wound environ- girl resulted in an amputation at the base of the nail. The injury was treated conserva- ment. tively with dressing changes and after 8 weeks the fingertip regenerated (From Stocum The annual regeneration of deer DL. Regen Biol Med 2006, 394, copyright 2006, Elsevier, reproduced by permission.). antlers represents another exam- (C–E) A fingertip injury of a 2-year old child resulted in an amputation at a level proximal to the nail. (C) Radiograph at the time of injury indicated that the level of amputation was ple of a naturally occurring regen- through the proximal region of the terminal phalangeal bone. (D) The amputation injury erative response in mammals. was treated conservatively with dressing changes and after 10 months the fingertip Deer, along with many of their rel- healed without significant scarring and a nail rudiment was present. The fingertip had a atives, shed their antlers annually normal contour and sensibility had returned. (E) Radiographic evidence after healing showed that there was no re-growth of the terminal phalangeal bone and indicated that a only to have them undergo a com- regenerative response was not stimulated. (From Han M, Yang X, Lee J, et al. Dev Biol plex regenerative response that 2008, 315:125–135, Copyright 2008, Elsevier, reproduced by permission.). involves outgrowth from the pedi- cle, a bilateral bony protrusion of the frontal bone. Primary antler of the epidermis over the pedicle chondroblasts, hypertrophic chon- development occurs during pu- wound to form a wound epithe- drocytes, and bone, undergoing berty and in response to circulat- lium. Beneath the wound epithe- endochondral ossification similar ing levels of sex steroids. Antler lium is a dermal layer overlaying a to the developing antler. In addi- development involves the initial fibrous perichondral layer that is tion to bone tissue, antler regen- formation of the pedicle from a continuous with the periosteal eration involves the regeneration specialized periosteum associated layer of proximal bone. Just proxi- of epidermis and its derivatives, with the frontal bone. Pedicle out- mal to the fibrous perichondrium dermis, and vasculature. growth and elongation involves at the distal end of the regenerat- many developmental processes ing antler is a mesenchymal FINGERTIP REGENERATION beginning with intramembranous growth zone (also called the IN HUMANS ossification to initiate pedicle for- reserve mesenchyme) where cells mation, and is followed by the for- are actively dividing, and is argu- The prospect of developing strat- mation of a distal endochondral ably the antler blastema (Fau- egies for enhancing regenerative growth zone with proximal ossifi- cheux et al., 2004; Li et al., 2005; ability in humans is encouraged by cation that continues until the ant- Kierdorf et al., 2007). The cells of clinical observations that the ler is fully developed (Price et al., the mesenchymal growth zone human fingertip is capable of a re- 2005). Antler shedding or casting appear to be derived from the generative response (Fig. 1A, B). normally occurs in the spring and pedicle perichondrium and these While the initial descriptions of fin- is mediated by enhanced osteo- cells display characteristics of gertip regeneration were made in clast activity at the distal region of mesenchymal stem cells (Rolf children (Douglas, 1972; Illing- the pedicle, leaving an open pedi- et al., 2008). Proximal to the mes- worth, 1974), they were followed cle wound that forms a scab (Goss enchymal growth zone are regions by descriptions of fingertip regen- et al., 1992). The regeneration containing, in distal to proximal eration in adults as well (Lee process is initiated by the closure order, chondroprogenitor cells, et al., 1995). The key for human

Birth Defects Research (Part C) 84:265–280, (2008) 268 MUNEOKA ET AL. fingertip regeneration is to treat first step, we provide evidence by connective tissue and encased the amputation wound in a con- from a case report that begins to within a nail (Fig. 2A). Like human servative manner, e.g., clean and define the proximal extent of re- fingertips, regenerative ability is dress the wound so as to allow it generative capabilities (Han et al., level specific within the terminal to heal by secondary intention 2008). This case report involved phalanx. In adult mice amputa- (i.e., without assisted wound clo- an amputation injury in the proxi- tions midway through the terminal sure). It is thought that such con- mal region of the terminal phalan- phalanx result in a robust regener- servative treatment in humans geal bone that was conservatively ative response, whereas amputa- promotes the formation of a treated, and because there was X- tion in which more than 3/4th of wound epidermis that is required ray documentation at the time of the bone is removed fails to mount for the initiation of a regenerative injury and after the healing a response (Neufeld and Zhao, response. Appendage regeneration response was completed, there is 1995; Han et al., 2008). Because in amphibians and in embryonic clear indication that a regenerative the mouse digit is relatively small, limbs of birds has been shown to response that included bone the physical distance between have a similar requirement of a regrowth did not occur (Fig. 1C, D, these two amputation injuries is specialized wound epidermis for a E). Thus, despite the fact that cos- less than 1 mm and the tissue successful regenerative response metic healing and good sensibility composition of both regenerating (see Muller et al., 1999). Never- of the fingertip was restored, this and nonregenerating wounds is theless, the actual closure of the case report begins to identify the similar, making this a good model amputation wound itself in humans proximal boundary of regenerative to investigate the cause of regen- is a very slow process, and much ability in humans. Understanding erative failure. The fact that the of the regenerative growth and the physical boundaries of regen- mouse digit tip is able to regener- remodeling associated with the erative potential in humans is an ate makes it a very unique part of regeneration response occurs important first step toward de- the body, and we have made con- before the completion of wound veloping a protocol that has pre- siderable effort to characterize its healing. Thus, it is safe to conclude dictive value for the treatment of developmental anatomy as well as that continuity of the wound epi- amputation injuries. What is its regenerative potential. dermis is not a prerequisite for needed is a concerted effort to regeneration, but it is unclear now better document the limits of this Developmental Anatomy of the what role the wound epidermis amazing regenerative response in plays in human regeneration. humans that would entail radio- Digit Tip Fingertip regeneration in graphic analysis of amputation The primary structure of the humans is reported to be re- injuries before and after healing to mature digit tip is the terminal stricted to the distalmost, or ter- establish a database that can be phalangeal bone (P3), which is lat- minal phalangeal element and used both for predicting clinical erally flattened and has a triangu- associated with the nail organ outcome and for experimental lar shape with its base articulating (Illingworth, 1974). The clinical studies (see below). Since injuries with the subterminal phalangeal use of the term regeneration is not to the hand alone represent 30% element (P2) and its apex forming strictly defined so the clinical of all reported injuries (Oleske and a sharp point (Fig. 2B, C). The P3 description of amputation injuries Hahn, 1992; Angermann and bone contains a bone marrow often involve cosmetic and neuro- Lohmann, 1993), and there are region localized to the base of the logical assessment of soft tissue 19,000 reported digit amputa- element and associated with a repair, whereas regenerative stud- tions per year in the United States small oblong canal that is contigu- ies in animal models generally alone (Sorock, 1993), it should be ous with the lateral connective tis- focus on the restoration of skeletal possible to establish such a data- sues. There are insertion sites for tissue in addition to soft tissue base in a relatively short time the dorsal extensor tendon and (Han et al., 2008). Using bone frame. This would be the first step ventral flexor tendon in the proxi- regrowth as definitive evidence for toward developing an understand- mal region, and collateral liga- a regenerative response, there is a ing of the limitations of human re- ments join P2 and P3 laterally. The subset of clinical reports that generative ability and, as well, for terminal phalangeal bone is document fingertip regeneration the development of therapies to encased within a nail organ that after conservative management of enhance this amazing response. comes to a sharp distal point amputation wounds in both chil- extending well beyond the tip of dren (Vidal and Dickson, 1993) DIGIT TIP REGENERATION the terminal phalanx. The nail cov- and adults (Lee et al., 1995). ers the dorsal and lateral surfaces IN MICE Thus, it is clear that human finger- of P3 but is not contiguous ven- tips display a true regenerative A valuable experimental model for trally. The nail organ consists of response that establishes the human fingertip regeneration is the nail matrix proximally that foundation upon which we can the digit tip in rodents. The mouse supplies cells to the nail bed which begin to explore ways to enhance digit tip includes the terminal pha- is overlain by the nail plate. Mouse the regenerative response. As a langeal bone which is surrounded nail growth is continuous through-

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 269

Figure 2. Developmental anatomy of the mouse digit tip. (A–C) Mature mouse digit tips. (A) A digit tip showing that the terminal phalange is encased within a nail dorsally and laterally. The terminal phalangeal bone is visible through the nail (arrow). (B) Whole mount terminal phalangeal (P3) bone stained with Alizarin Red S showing a sharp point at the apex. Proximal end of the P3 bone articulates with the subterminal phalangeal (P2) bone. (C) Sagittal sectioned sample stained with Mallory’s triple stain. The loose connective tissue between the nail bed and the bone contains cells that appear fibroblastic. (D–H) Gene expression in the developing digit tips of E14.5 embryos. (D) Msx1 is expressed in the apical mesenchymal cells surrounding the forming terminal phalanx. (E) Msx2 is expressed in the apical epidermis and in mesenchymal cells subjacent to the epidermis. (F) Bmp4 is expressed in apical mes- enchymal cells in a domain similar to that of Msx1.(G)Ihh is expressed in digit tip cells, initiating endochondral ossification of the terminal phalanx. (H) The nail organ marker, Hoxc13, is expressed in the distal epidermis associated with presumptive nail tissue. (I–M) Gene expression in the developing digit tips at birth. (I–K) Expression of cartilage-specific genes. (I) Type II collagen (Col II), a marker for proliferating chondrocytes; (J) Indian hedgehog (Ihh), a marker for prehypertrophic chondrocytes; (K) Type X Collagen (Col X), a marker for hypertrophic chondrocytes. (L–M) Expression of osteoblast-specific genes. (L) Osteocalcin transcripts are first expressed at the apex of the terminal phalangeal bone then extend along the dorsal surface; (M) Type I collagen (Col I) is first expressed in a similar pattern. (N–P) Whole-mount skeletal staining of postnatal digit tips stained with Alizarin Red S and Alcian Blue. Chondrogenic tissue stains blue and osteogenic tissue stains red. Ossification begins from the distal tip at birth (N) and pro- gresses in a proximal direction. (O) At postnatal day 7 (PN7), the distal 3=4 of the terminal phalangeal bone has initiated ossification and by PN14 (P) ossification has commenced along the entire proximal-distal length of the bone. [B,C, I–P (From Han M, Yang X, Lee J, et al. Dev Biol 2008, 315:125–135, Copyright 2008, Elsevier, reproduced by permission.); D–H (From Han M, Yang X, Farrington JE, et al. Development 2003, 130:5123–5132, Copyright 2003, Company of Biologist, reproduced by permission.)]

out life with nail loss occurring api- the ventral epidermis is contigu- tissue is more prominent along the cally. The ventral surface of the ous with the thickened interdigi- dorsal and lateral surfaces by digit tip is covered with a thin tating epidermis of the ventral fat comparison to the ventral, and layer of keratinized epidermis that pad. Between the nail bed and the there are regions where the con- extends to the distal tip where it is terminal phalanx is a layer of loose nective tissue cells are organized thickened and contiguous with the connective tissue that surrounds perpendicular to the bone surface dorsal nail epidermis. Proximally, the bone. This layer of connective and appear to attach the nail bed

Birth Defects Research (Part C) 84:265–280, (2008) 270 MUNEOKA ET AL. to the bone surface. The loose connective tissue consists primar- ily of fibroblasts and cells associ- ated with the vascular system. Developmentally, the digit tip is first identified as an autonomous structure at embryonic day 14.5 (E14.5) when the P2-P3 interpha- langeal joint becomes visible. At this stage, there are several genes that are specifically expressed by cells of the digit tip that distin- guishes this region from the proxi- mal phalangeal elements (Fig. 2D– H). For example, the expression of Hoxc13 in cells of the dorsal epi- dermis marks the forming nail organ (Godwin and Capecchi, 1998), Msx2 is expressed in the apical epidermis and also in cells of the most distal mesenchyme (Reginelli et al, 1995), and Msx1 and Bmp4 are expressed in a larger domain of the distal mesenchyme (Reginelli et al., 1995; Han et al., 2003). Other genes known to be expressed at the developing mouse digit tip include Dlx5 (Acampora et al., 1999), Bambi (Grotewold et al., 2000), and Dachshund (Hammond et al., 1998; Davis et al., 1999). The distal specific expression of these genes reinfor- ces the conclusion that the digit tip is a unique structure both in devel- opment and in regeneration. The uniqueness of the digit tip in birds and mammals has recently been reviewed by Casanova and Sanz- Ezquerro (2007) in the context of Figure 3. Ossification and growth of the terminal phalangeal (P3) bone. (A) Growth digit evolution. curve of P3 bone. The terminal phalangeal bone continues to elongate during the post- Ossification of the mouse digit natal period and reaches its mature length at 8 weeks. The lower curves show that fol- tip initiates just before birth (Han lowing amputation at PN3, the terminal phalangeal bone never catches up with unam- et al., 2008). Gene expression putated controls. (B–G) Calcein incorporation into the P3 bone. (B) Calcein labeling for one day identifies two ossification centers at 3 weeks of age, one associated with the studies indicate that hypertrophic proximal growth plate and one at the distal tip. (C) By 5 weeks of age only the distal chondrocytes begin to mature ossification center is observed. (D) Calcein incorporation is used as a vital marker to between E17.5 and E18.5 (birth), identify existing bone by long-term labeling studies. New bone deposition distal to the with the onset of ColX expression calcein label identifies bone deposition that has occurred since the initial labeling (white arrows). Calcein was injected at PN1 and analyzed at 4 weeks (D) and 7 weeks at the distal tip. At this stage pro- (E) of age. (F) By measuring the length of the terminal phalangeal bone and the proxi- liferating chondroblasts identified mal-distal length of new bone deposition, we show that both proximal and distal ossifi- by ColII expression are localized cation centers contribute equally to bone elongation from birth until 4 weeks of age. to the proximal half of P3 and (G) Similar analysis of the period between 4 and 9 weeks of age indicate that length- there is a band of prehypertrophic ening of the terminal phalanx results solely from the distal ossification center. [A–E (From Han M, Yang X, Lee J, et al. Dev Biol 2008, 315:125–135, Copyright 2008, cells identified by Ihh expression Elsevier, reproduced by permission.)] between the ColX and ColII expression domains (Fig. 2I–K). During this same period ossifica- genes for ossification, Osteocalcin In mice, elongation of the termi- tion is initiated at the digit tip in and ColI (Fig. 2L, M). As the digit tip nal phalanx continues until it association with the ColX domain matures, ossification progresses reaches a mature length at 8 as evidenced by histological stain- in a distal to proximal direction weeks of age (Fig. 3A). Elongation ing and expression of marker (Fig. 2N–P). rate is rapid during the first 3–4

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 271 weeks following birth, after which the rate of elongation levels off until the mature length is reached (Han et al., 2008). Calcein is a marker that incorporates into newly forming bone and pulse labeling is used to identify regions of the bone that are undergoing active ossification (Suzuki and Mathews, 1966). Calcein pulse labeling of the terminal phalanx indicates that the early stages of bone growth (before 4 weeks) are associated with two ossification centers, one proximal and one dis- tal (Fig. 3B), whereas by 5 weeks of age only the distal ossification center remains (Fig. 3C). The proximal ossification center is linked to the proximal endochon- dral growth plate which closes at 3–4 weeks of age, and the distal ossification center results in appo- sitional bone growth similar to that occurring during diametrical growth of the bone collar. Because ossification by appositional growth is typically linked to growth of the bone collar in long bones, it seems Figure 4. Regeneration response in the fetal digit tips of mice and humans. (A): The central digits (digits 2, 3 and 4) that were amputated at E14.5 exhibit normal regener- reasonable to consider the termi- ation response. Note that the regenerated digit tips (asterisk) are little bit shorter than nal phalangeal bone as equivalent a non-amputated control digit tip (arrowhead). (B–D) In situ hybridization of frontal to the proximal half of a subtermi- sections at 2 days after amputation showing expression of Msx1 (B), Msx2 (C), and 2/2 nal phalanx with its central collar Bmp4 (D). (E) Digits of Msx1 mutant mice display a regeneration defect. (F–H)In situ hybridization of frontal sections at 2 days after amputation. (F) A nonfunctional region constricted to form the api- transcript of Msx1 is expressed at the amputation wound. (G, H) Msx2 and Bmp4 cal tip of P3. Calcein can also be transcripts are not upregulated in the failed regeneration response. (I) Treatment of introduced as a vital marker of wildtype and Msx12/2 mutant digits with the BMP antagonist, NOGGIN, inhibits the existing bone to quantitate ossifi- regeneration response. Despite the absence of a regeneration response by NOGGIN cation that takes place after its treatment, stump tissues maintain expression of Msx1 (J), Msx2 (K) and Bmp4 (L), suggesting that all three genes are upstream of the NOGGIN inhibitory effect in digit introduction; in this case newly regeneration. (M–P) Embryonic human digits initiate a regeneration response in vitro. formed bone is unlabeled (Fig. 3D, (M) A fingertip from a human embryo with an estimated gestational age (EGA) of 57 E). Measurements of distal ossifi- days displays MSX1 immunostaining that is localized to the nail forming region. (N) cation by comparison to the elon- Amputated fingertips cultured for 7 days postamputation (DPA) initiate a regeneration response, forming a blastema at the wound site (cytokeratin-19 immunostaining). (O) gation of the entire terminal pha- Amputated fingertips cultured for 4 DPA and immunostained for MSX1 show upregula- lanx show that during the period tion in dorsal mesenchymal tissue. (P) At 7 DPA, blastemal cells stain positive for from birth to 4 weeks of age about MSX1. [A–H, J–L (From Han M, Yang X, Farrington JE, et al. Development 2003, half the length of the terminal 130:5123–5132, Copyright 2003, Company of Biologist.); M–P (From Allan CH, Fleck- phalanx results from distal ossifi- man P, Fernandes RJ, et al. Wound Repair Regen 2006, 14:398–404, Copyright 2006, Wound Healing Society.)] cation (Fig. 3F). After 4 weeks of age the distal ossification center is responsible for 100% of terminal regenerative decline associated 1995) and in vitro (Han et al., phalanx elongation (Fig. 3G). with maturation (Muneoka and 2003). In mice, the proximal limit Sassoon, 1992; Muller et al., of regenerative capability is asso- 1999). For example, the limb buds ciated with the proximal extent of Embryonic Digit Tip as a Model of rats and mice have been shown the Msx1 expression domain dur- for Mammalian Regeneration to partially regenerate following ing digit development (Reginelli The developing limb bud and amputation in vitro (Deuchar, et al., 1995). During embryonic digits of higher vertebrates pos- 1976; Lee et al., 1991) and in vivo digit tip regeneration a number of sess enhanced regenerative capa- (Wanek et al., 1989), and the em- genes specifically expressed at the bilities by comparison to adults, bryonic mouse digit tip undergoes apex of the developing digit are and represent models for investi- a rapid and complete regenerative upregulated, suggesting that gating regenerative responses and response in utero (Reginelli et al., these genes, Msx1, Msx2, Bmp4,

Birth Defects Research (Part C) 84:265–280, (2008) 272 MUNEOKA ET AL.

Hoxc13, and Ihh (Fig. 4A–D), are 2003). In other studies we have The Msx genes are implicated in important for the regeneration investigated the role of Dlx5 in em- other models of regeneration and/ response. These genes are not bryonic digit regeneration (J. Lee, or cell renewal. In the regenerat- expressed at a proximal digit unpublished data). Dlx5 is ing urodele limb, Msx genes are amputation wound that is normally expressed in the apical ectoderm down-regulated in the mature nonregenerating (Reginelli et al., and mesenchyme in the E14.5 limb, and following limb ampu- 1995; Han et al., 2003). We have digit tip in a domain that overlaps tation both Msx1 and Msx2 are analyzed the Msx1 mutant (Sato- with Msx2 (Acampora et al., 1999; re-expressed during regeneration kata and Maas, 1994) and the Han et al., 2003). In amputation then down-regulated after redif- Msx2 mutant (Satokata et al., studies we do not find a regenera- ferentiation (Crews et al., 1995; 2000) to determine whether either tion phenotype in amputated Simon et al., 1995). Msx gene of these genes are playing a func- homozygous Dlx5 mutant embryo expression during regeneration is tional role in the regeneration digit tips. In addition, we have largely similar to developmental response (Han et al., 2003). Our tested mutant embryos lacking expression: Msx1 is expressed by studies show that the Msx1 mu- both the Dlx5 and Msx2 genes, mesenchymal cells, whereas Msx2 tant, but not the Msx2 mutant, dis- and in response to digit tip ampu- is expressed by both mesenchy- plays a regeneration phenotype tation, these embryos can also mal and apical epidermal cells (Fig. 4E), suggesting that Msx1 is successfully regenerate. Thus, we (Carlson et al., 1998; Koshiba playing a critical role in the regen- can conclude that both Msx2 et al., 1998). Msx1 is also eration response (Han et al., and Dlx5 are not essential for expressed in association with limb 2003). The distal tip of the Msx1 embryonic digit tip regenera- regeneration in developing Xeno- mutant digit following amputation tion even though both are promi- pus limbs (Endo et al., 2000) and behaved like a proximal level nently expressed in the forming in association with FGF-induced amputation, resulting in a trun- digit tip. regeneration of the amputated cated digit phenotype and a failure In separate studies on embry- chick wing bud (Taylor et al., to upregulate the distal digit onic human digits, Allan et al. 1994; Kostakopoulou et al., marker genes (Fig. 4F–H). Since (2006) established that cultures of 1996). In these cases the re- the Msx1 mutant digit does not embryonic human digits under se- expression of Msx1 is initiated at have a limb phenotype in develop- rum-free conditions were able to the wound surface and establishes ment, these results suggest that go through early stages of a re- an expression domain in the Msx1 is functioning in a regenera- generative response. Digits tested regenerate that is similar to the tion-specific manner. We also were from embryos with an esti- developing limb. Msx1 has been found that Bmp4 expression in the mated gestational age (EGA) of shown to be required for tail digit tip was controlled by the com- 53–117 days and were maintained regeneration in Xenopus (Beck bined action of Msx1 and Msx2, in culture from 4 to 28 days. MSX1 et al., 2003) and fin regeneration and we therefore carried out stud- expression was analyzed immuno- in Zebrafish (Thummel et al., ies to determine whether exoge- histochemically and was found in 2006). Developmental studies nous BMP4 could rescue the Msx1 control digits to be expressed in show that Msx1 acts to inhibit dif- regeneration phenotype. Indeed, the connective tissue between the ferentiation of a variety of cell we found that BMP4 rescued digit nail bed and the terminal phalan- types (Hu et al., 2001 ) and there tip regeneration in a dose-depend- geal bone in digits up to 70 days is evidence that over-expression ent manner and expression of all EGA (Fig. 4M). An analogous of Msx1 can induce myotube dedif- distal digit marker genes was expression domain of Msx1 is ferentiation in vitro (Odelberg upregulated in this induced regen- observed in the late mouse et al., 2000). Transcription studies erative response. To corroborate embryo and early neonatal digit show that the Msx1 protein func- this finding, we used the BMP an- tips (Reginelli et al., 1995). Ampu- tions as a transcriptional repressor tagonist, Noggin, to determine if tated 57 day EGA digits initiated a acting with a TATA binding protein BMP signaling was required for regenerative response with the (Catron et al., 1995; Zhang et al., digit regeneration in wildtype formation of a blastema like struc- 1996, 1997) and the linker his- digits, and found that treating ture (Fig. 4N) and MSX1 expres- tone, H1b, to inhibit differentia- amputated digits with exogenous sion was found to be associated tion-specific gene expression (Lee Noggin inhibited the regeneration with the regenerating apical cells et al., 2004). Since we find that response (Fig. 4I). Gene expres- (Fig. 4O, P; Allan et al., 2006). Msx1 expression is required for sion studies demonstrated that These results provide evidence digit regeneration, is acting in a Noggin treatment did not affect that human digit tissues share regeneration-specific manner, and Msx1, Msx2,orBmp4 expression gene expression and injury is upregulated during the regener- (Fig. 4J–L), suggesting a linear responses with that of the mouse ation process, we speculate that it pathway in which Msx1/Msx2 digit, and point to the MSX1 gene is functioning to repress an activ- regulated Bmp4 and that BMP4 as a candidate regulator in the ity that is normally inhibitory for a played a key role in controlling the control of a human regeneration regenerative response. Our BMP4 regeneration response (Han et al., response. rescue data suggests that the

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 273

amputated bone. The blastema, on the other hand, appears to integrate cells derived from the connective tissue surrounding the bone with cells derived from the marrow to form a proliferation center associated with the regen- eration response (Fig. 6C, D; Han et al., 2008). The cell contribution to the blastema has not yet been documented, and this is an area of research that is critically important for our understanding of mamma- lian regeneration. The third phase Figure 5. Digit tip regeneration in neonatal mice. Amputations were carried out at a of regeneration involves the differ- distal level through bone (A) and at a proximal level through cartilage (D) at postnatal entiation of regenerated struc- day 3 (PN3). After 6 weeks, digits were analyzed using whole-mount bone stain with Alizarin Red S (B, E) and histological analysis with Mallory’s triple stain (C, F). Distal tures, i.e., bone, connective tis- amputations regenerate anatomically normal digit tips (B, C), however, proximal sue, and nail. The regrowth of amputations show no signs of regeneration (E, F). bone is most critical because it structurally defines the regener- ate. For example, nail regrowth Bmp4 is a target gene for such an sion of the amputated stump bone continues in proximally amputated anti-regeneration activity. resulting from enhanced osteo- digits but the regrown nail lacks clast activity (Revardel and Che- anatomical structure and simply bouki, 1987), which extends the covers the truncated stump (Fig. Postnatal Digit Tip Regeneration actual limit of regeneration to a 5F). In the neonate, bone re- The regeneration of neonatal more proximal level so as to growth following amputation occurs and adult digit tips is thought to include the eroded bony tissues. A between 7 and 14 days postampu- be largely equivalent (Neufeld and similar region of tissue remodeling tation and proceeds by direct ossi- Zhao, 1995), although regenera- during amphibian limb regenera- fication, i.e., without expression of tion of neonatal digit tips occurs tion is associated with the upregu- any endochondral marker genes in during a time of rapid skeletal lation of MMPs that are thought to the distal digit region where ossifi- elongation. Amputation midway mediate this effect (Yang et al., cation is commencing (Fig. 6E). through the terminal phalanx 1999). Similarly, microarray stud- Osteoblasts present at the inter- results in a regeneration response ies of digit tip regeneration in mice face between the blastema and (Fig. 5A–C), whereas amputation identify several upregulated histo- the bone stump at 7 days postam- through the proximal region fails lytic genes, including MMPs, asso- putation suggest that the blas- to mount a regenerative response ciated with this phase of regenera- tema is organizing and perhaps (Fig. 5D–F). These two amputation tion (Chadwick et al., 2007). contributing to the regenerated planes transect similar tissues yet The second phase involves blas- new bone (Fig. 6F). There is a the repair response differs dra- tema formation. The question of burst of ossification that occurs matically. Digit tip regeneration whether or not a ‘‘blastema’’ forms between 7 and 14 days postampu- can be described in terms of in mammals is addressed below, tation that restores the character- distinct phases that are character- and for purposes of discussion the istic pattern of the P3 bone and istic of other regeneration models blastema is simply defined as an completes the regenerative such as the urodele amphibian aggregation of proliferating cells response (Fig. 6G). limb (Gardiner et al., 2002). After involved in the regeneration proc- distal amputation the regeneration ess. Based strictly on histological THE MAMMALIAN response is characterized by an observations and cell proliferation BLASTEMA initial wound closure response that studies the blastema appears to requires multiple days, despite the form from two sources: (1) the One of the hallmarks of limb relatively small surface area of the migration of connective tissue cells regeneration in urodele amphib- amputation injury (Fig. 6A, B). across the amputation wound ians is the formation of a blastema The slow rate of epidermal closure (Neufeld et al., 2004), and (2) of undifferentiated cells that prolif- following amputation is similar to cells arising from the marrow cav- erate, go through morphogenesis, the human response and unlike ity of the skeletal stump. In nonre- and differentiate to replace struc- the rapid re-epithelialization that generating proximal digit amputa- tures lost by amputation (Bryant occurs during amphibian limb tion, the skeletal stump forms a et al., 2002; Brockes and Kumar, regeneration (Carlson et al., periosteum across the injury site 2005). The blastema is a transient 1998). During the wound healing and undergoes ossification result- phase in regeneration that has phase there is considerable ero- ing in a truncated cap of the been described in terms of the

Birth Defects Research (Part C) 84:265–280, (2008) 274 MUNEOKA ET AL.

Figure 6. Histological and gene expression analyses of regenerating digit tips. (A–C) Histological sections of regenerating digit tips stained with Mallory’s triple stain at the time of amputation (A), 4 days postamputation (DPA) (B) and 6 DPA (C). Note the forma- tion of a blastema by 6 DPA. (D) BrdU incorporation at 7 DPA shows robust proliferation in the connective tissue and the bone stump. (E, F) In situ hybridization analyses documenting the expression patterns of a hypertrophic chondrocyte marker (E), Type X Collagen (Col X), and an osteoblast maker (F), Osteocalcin. Note the absence of chondrogenic marker gene transcripts associated with the blastema indicating that the regeneration response involves direct ossification. (G) Calcein was used as a vital label to identify ossification in the stump 1 week after amputation and Alizarin Red S was used to stain bone after 2 weeks. Differentiation of the regenerate is largely completed by 2 weeks after amputation. (H) Msx2 expression is induced in the dorsal connective tissue during wound healing stages but is absent in the blastema (not shown). (I) Bmp4 transcripts are present at the distal tip of the blastema and also in the dorsal connective tissue. (J) Dlx5 is expressed in cells at the base of the blastema and also in the marrow region of the stump. (K, L) Pedf is specifically expressed in the bone marrow and distal apex of the blastema underneath the wound epidermis during regeneration.

characteristics of cells with respect and important to identify parallels healing); however, in limb regener- to both their tissue of origin as with, as well as deviations from, ation, the term has been redefined well as their ultimate fate in the best characterized regenerat- to include characteristics of urodele regeneration. Thus, for example, ing systems. blastema cells, and/or their inter- we know that the amphibian blas- Many models for mammalian actions with the overlying epider- temal cells (1) arise from either appendage regeneration involve mal layer. For example, the regen- dedifferentiation of, and/or stem the formation of a proliferating ag- eration blastema has been defined cells present in, mature tissues, gregate of undifferentiated cells as (1) a structure derived from (2) appear undifferentiated and that undergoes differentiation to the dedifferentiation of cells at the express developmental genes dur- form a regenerated structure. This amputation wound, (2) arising ing the blastema phase, (3) prolif- cell aggregate is often described as through epithelial-mesenchymal erate, and (4) differentiate in ei- ‘‘blastema-like’’ because it does interactions, and (3) contains in- ther a homotypic or heterotypic not have all of the characteristics of trinsic morphogenetic information (metaplastic) manner (Brockes the classical amphibian blastema (see Carlson, 2005, 2007). Some and Kumar, 2002; Han et al., that mediates appendage regener- have even proposed that regenera- 2005; Morrison et al., 2006). ation. The term blastema is used tion itself is defined as a response There is currently no mammalian generally to describe a cell aggre- that must involve the dedifferentia- counterpart to the urodele blas- gate involved in development tion of cells at the amputation tema, and because there exists a (e.g., blastema condensations dur- wound (Kostakopoulou et al., growing interest in developing ing skeletal formation, metaneph- 1996). With respect to mammalian strategies to induce regenerative rogenic blastema during kidney de- regeneration, this strict definition responses in mammals, particu- velopment) or regeneration (e.g., of a regeneration blastema impairs larly humans, it is both necessary osteoblastic blastema in fracture the utility of the blastema concept

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 275 simply because it presumes that invaded by osteocytes following an integration of the stump tissues mammalian regeneration must osteogenic process that recapitu- with the regenerate is a research proceed via mechanisms similar to lates development (Faucheux topic that has not received much amphibians. Whether or not this is et al., 2004). Cultures of antler attention, yet this interface is true remains to be seen. Indeed, blastemal cells provide for a way to clearly important for success in we argue below that there are likely characterize the control of blas- functional tissue engineering and multiple ways to regenerate a sin- tema growth and differentiation regenerative medicine. In regen- gle structure, and if a blastema is (Price et al., 1994; Sadighi et al., erating systems, this interface involved, it follows that the charac- 1994). Factors shown to enhance represents a site where terminally teristics of the blastema must not cell proliferation in vitro include differentiated cells of the stump be constant. Thus, there is value in IGF-I, IGF-II, FGF2, and PTHrR, must functionally interact with the broadly defining the blastema as and in vivo support for this prolifer- undifferentiated cells of the regen- an anatomical structure so as to ative effect comes from immuno- erate. Further studies on this topic bring regeneration studies under a histochemical studies showing that will prove to be important for the common umbrella where we can the corresponding receptors (IGFR, development of therapeutic strat- discover similarities and differen- FGFR, and PPR) are expressed in egies critical for functional tissue ces when comparing regeneration the blastema (Price et al., 1994; engineering. events of distinct animal groups Sadighi et al., 1994; Barling et al., Recently, we characterized the (e.g., amphibians versus mam- 2004; Faucheux et al., 2004; expression of digit development mals). We therefore favor the sim- Lai et al., 2007). Other studies pro- markers during neonatal digit tip ple definition of a regeneration vide evidence that key develop- regeneration using in situ hybrid- blastema as an aggregate of prolif- mental signaling pathways, such ization (Han et al., 2008). Msx1 erating undifferentiated cells as canonical Wnt signaling (Mount and Msx2 are expressed in the dis- involved in the regeneration of a et al., 2006), BMP signaling tal digit tip during development lost body part. (Barling et al., 2005) and retinoic and are prominently expressed All three mammalian regenera- acid signaling (Allen et al., 2002), during embryonic digit tip regen- tion models described earlier (ear are also playing a critical role in eration (Fig. 4). In the neonatal punch, antler, and digit tip) regen- antler regeneration. digit Msx1 is expressed in the dor- erate via a blastema. In ear punch The regenerating mouse digit tip sal connective tissue, whereas regeneration, the blastema forms forms a blastema of proliferating Msx2 is expressed in the nail epi- around the circumference of the cells that later undergoes direct dermis. Following digit tip amputa- hole punch and it grows inward ossification to restore the distal tion we see Msx1 expression upre- progressively filling in the hole region of the terminal phalanx gulated in association with the with tissue. Several genes and (Revardel and Chebouki, 1987; healing dorsal connective tissue, proteins are expressed in the blas- Neufeld, 1992). While the cellular and in this same region we find tema, including BMP2 (Urist et al., origins of the blastema remain to Msx2 expression induced (Fig. 1997), the EGF family member be explored, studies of the neona- 6H). This upregulation of Msx Pref-1 (Samulewicz et al., 2002), tal blastema identify the loose genes is transiently associated and the matrix metalloproteinases connective tissue surrounding the with the wound healing response (MMP), MMP2 and MMP9 (Goure- terminal phalangeal bone and the since we do not observe expres- vitch et al., 2003). In addition, an marrow-forming region as areas sion of either gene in the blas- angiopoietin-related growth factor where enhanced cell proliferation tema. We note that the early (AGF) expressed in injured skin is occurring in association with the induction of Msx2 has also been promotes regeneration in the regenerative response (Han et al., observed following amputation transgenic mouse model (Oike 2008; see Fig. 6D). The digit blas- during amphibian limb regenera- et al., 2003). The blastema of the tema is characterized by its conti- tion (Carlson et al., 1998), thus regenerating deer antler is the nuity both with the connective tis- perhaps Msx2 may be serving a proliferating mesenchymal region sue surrounding the stump bone parallel function. Once the blas- called the mesenchymal growth and with the stump bone itself. tema has formed, we find Bmp4 zone or the reserve mesenchyme Unlike digit amputation at a proxi- transcripts expressed in the dorsal (Price et al., 2005; Kierdorf et al., mal (nonregenerating) level where connective tissue, in the forming 2007). These cells express the a periosteum forms a distal cap bone, and in the distal blastemal mesenchymal stem cell marker associated with bone truncation mesenchyme (Fig. 6I). In the STRO-1, and have been shown to (Neufeld, 1985), the smooth inte- unamputated digit, Bmp4 expres- be multipotent when challenged in gration of the regenerated bone sion is restricted to the forming in vitro differentiation assays; tissue with the stump is a likely bone, so it appears to be upregu- thus, it is proposed that the blas- outcome of the stump-blastema lated specifically in the blastema tema is stem cell derived (Rolf continuity, and the extensive and dorsal connective tissue. We et al., 2008). In vivo, these cells remodeling of the stump that also find the homeodomain con- give rise to chondrocytes that occurs during the wound healing taining gene Dlx5 expressed in the undergo hypertrophy and are phase appears to play a role. The blastema. Dlx5 is expressed in

Birth Defects Research (Part C) 84:265–280, (2008) 276 MUNEOKA ET AL. both the epidermis and mesen- addressed largely within the con- that evolved at the expense of re- chyme of the developing digit tip text of animals that possess a high generative ability (Mescher and in a pattern similar to Msx2, and degree of regenerative capabil- Neff, 2005). However, this view genetic studies indicate that nei- ities, i.e., invertebrates and lower leaves no place for endogenous ther Dlx5 nor Msx2 are required vertebrates (fish and amphibians). regeneration models in mammals, for regeneration (see above). Dlx5 For these animals there is indirect because how can we systemically is also known to play a critical role evidence that the ability to regen- forfeit regenerative ability yet in osteogenesis (Holleville et al., erate is an ancient attribute of retain it in select parts of the 2007), thus its expression during metazoans that is linked to devel- body? Regeneration in mammals postnatal digit regeneration is opmental mechanisms, and as is real and how it fits into our probably associated with the onset such, is non selectable (Brockes thinking about the evolution of of ossification by regenerating et al., 2001). It is further pro- regeneration turns out to be criti- cells. In the blastema, Dlx5 ex- posed that regeneration is an evo- cal for understanding how to move pression is restricted to the proxi- lutionary remnant of the process forward in regenerative medicine. mal regions and lies between the of asexual reproduction (Sanchez For example, it is likely that re- Bmp4 domain and the skeletal Alvarado, 2000). It seems likely generative ability evolved second- stump (Fig. 6J). We also find however, that the wound healing arily from a nonregenerative state genes expressed in the blastema component of the regeneration in some mammals (see below), so that are not expressed during digit process that does not involve a we can look upon these regenera- development. For example, pig- reiteration of development is adapt- tion models as examples of how ment epithelium-derived factor ive and has undergone consider- nature got into the regenerative (Pedf), a secreted protein with able evolution among metazoans. medicine business and came up neurotrophic activity (Ramirez- Since wound healing is the initial with a successful product. Simi- Castillejo et al., 2006), is ex- phase of any regeneration larly, if regenerative ability was pressed in the bone marrow, and response, it would seem that an selectively retained in some parts is prominently expressed in the important consideration for our of the body, how is it that these blastema (Fig. 6K, L). To date, understanding of regeneration lies parts of the body can become re- Pedf is the earliest gene that we at the interface between an evolv- fractory to a strong negative have specifically localized to the ing wound healing process and selective force? The answers to forming blastema. Cells express- conserved developmental mecha- these questions may be quite pro- ing Pedf localize to the apex of the nisms. The process of intercalary found and could have a major blastema just underlying the growth (French et al., 1976; Bry- impact on the development of wound epidermis. It is not ex- ant et al., 1981), for example, strategies for how we can success- pressed in proximal amputation may represent a mechanism that fully implement regenerative wounds that fail to regenerate has successfully evolved in therapies. (not shown), suggesting that it amphibians and invertebrates to Deer antler regeneration is per- may be playing a key role in the span this interface (Gardiner haps the best example of an regeneration response. Because et al., 1995). evolved regenerative response as Pedf is prominently expressed in The flip side of the view that re- fossil records clearly indicate that both bone marrow and blastema, generative ability is a conserved antler formation and regeneration one possibility is that Pedf mechanism in animals that can evolved from a nonregenerative expressing cells in the bone mar- regenerate is that negative selec- precondition (Goss, 1969). The row contribute to the establish- tion for regeneration can explain antler develops postnatally as an ment of the blastema. In addition, why many groups of animals, outgrowth of the frontal bone and the neurotrophic activity of PEDF including mammals, lack the exten- its formation and regeneration fol- may be linked to maintaining sive regenerative capacity enjoyed low similar processes. The frontal damaged neurons following ampu- by others. It is here where most dis- bone forms during embryogenesis tation and during the regeneration cussions on the evolution of regen- by intramembranous ossification process. In conclusion, the digit eration ends; the animal kingdom and the initiation of antler out- tip blastema cells display charac- is divided into regenerators and growth involves direct ossification; teristics similar to the developing nonregenerators, and the discus- however, antler elongation itself digit tip, but also characteristics sion moves to addressing the involves an unusual form of endo- that are unique to regeneration. question of what might be the cir- chondral ossification in which the cumstances surrounding the loss growth plate is localized at the of regenerative ability in mam- apex of the outgrowth. These EVOLUTION OF mals. Indeed, a prevalent view is observations clearly demonstrate REGENERATION IN that the loss of regenerative ability that the evolution of antler forma- is linked to a more urgent selec- tion/regeneration is not linked to MAMMALS tive need of developing an adapt- the developmental mechanisms The evolution of regeneration in ive immune system to combat involved in frontal bone formation, the animal kingdom has been pathogens invading injured tissue but to developmental mechanisms

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 277 used to form elongated bony regeneration represents an entiation during regeneration has structures. Thus, the evolution of evolved regenerative response evolved away from the mecha- antlers involves the activation with cells at the amputation injury nisms that guided their initial de- and modification of developmental utilizing a developmental process velopment, i.e., regeneration itself mechanisms utilized in other parts that is novel for the developing is not constrained by existing de- of the body (e.g., long bones) and limb. One possibility is that digit velopment models. This suggests not those involved in frontal bone tip regeneration has been evolu- that in mammals, regeneration is formation itself. One conclusion tionarily conserved because it is not refractory to adaptive selec- that can be drawn from this is that adaptive, but the process itself tion, but that it is a process that when regenerative ability evolved, has evolved as well. We note that continues to evolve. Sorting out it was not constrained by the de- both intramembranous ossification aspects of regeneration that are velopmental history of the tissue and endochondral ossification are evolutionarily conserved versus responding to the injury. involved in bone formation during those that have evolved will prove It is unclear whether the digit tip long bone fracture healing, thus to be important as we begin to in mice represents a model of an both processes have evolved dur- consider human application based evolved regenerative response, ing bone healing and remodeling on discoveries from animal mod- one in which regenerative ability (Schindeler et al., 2008). It is also els. has been maintained, or both. The interesting to note that deer antler The anatomy of the mammalian vertebrate digit has undergone regeneration evolved from bone response suggests that a spatial tremendous anatomical modifica- that undergoes direct ossification system of positional information is tion, including digit elongation, during embryogenesis to form the required to guide regeneration, shortening, and reduction in num- antler by endochondral ossifica- although we have no understand- ber, so it is clear that the digit is tion, so these two regeneration ing of the nature of such a system under strong selective pressure. models appear to have evolved in in mammals. Because the regen- Bmp4 is expressed in developing opposite directions. erative response is limited, we do and regenerating digit tips and not have anatomical assays avail- required for embryonic digit able to explore the role that posi- regeneration (Han et al., 2003), tion plays in regeneration. How- CONCLUSION so its activity at the digit tip could ever, we are encouraged from very well be a target for adaptive By comparison to other well stud- studies of human fibroblasts which selection. It is worth noting that ied regeneration models such as display expression profiles that BMP4 is known to play a critical the amphibian limb, the study of vary with position in the body role in the evolution of tooth for- appendage regeneration in mam- (Chang et al., 2002; Rinn et al., mation in birds (Chen et al., 2000) mals remains at an immature 2006). Since limb fibroblasts play and, as well, is responsible for stage. Regeneration in mammals a critical role in establishing the variation in beak morphologies is very real despite the fact that it urodele blastema and in organiz- (Abzhanov et al., 2004; Wu et al., has clear limitations. In multiple ing the system of position infor- 2004). Thus, the expression of mammalian models, the wound mation required to effect a suc- Bmp4 in both digit development healing response results in the for- cessful regenerative response and regeneration, along with its mation of a blastema of proliferat- (Gardiner et al., 2002), the finding close association with other evolv- ing cells that mediates the regen- that analogous cells in adult ing vertebrate structures, is con- eration response. The source of humans maintain a similar posi- sistent with the idea that digit tip the cells that forms the blastema tional memory is suggestive that regeneration has been maintained remains unknown, although there the most essential component for during vertebrate evolution. On is evidence that these cells display a complex regenerative response the other hand, the regeneration characteristics associated with in humans is largely intact. Finally, of the postnatal digit tip involves mesenchymal stem cells. At the the role of fibroblasts in a mam- intramembranous ossification of same time there is also evidence malian injury response is generally the terminal phalangeal bone (Han that connective tissue fibroblasts linked to fibrosis and the produc- et al., 2008), a process that is not also participate in blastema forma- tion of scar tissue, yet in regener- related to the developmental proc- tion, thus the mammalian blas- ating models these same cells are ess of endochondral ossification tema may be composed of cells viewed as important contributors that originally forms the limb skel- derived from multiple sources. and play an organizing role in the eton. Thus, if digit tip regeneration Although the formation of the regenerated pattern. Thus, it is is an example of a maintained blastema in different mammalian our contention that understanding response, then why would it utilize models appears to share similar- and re-directing the response of a developmental process that is ities, the mechanisms guiding re- these cells, in particular, in a non- completely novel for digit forma- differentiation during regeneration regenerating mammalian wound is tion? Indeed, the intramembra- seem to be quite diverse. We a critical first step in transforming nous ossification of the regener- interpret this observation to sug- a wound healing response to a re- ated bone suggests that digit tip gest that the process of re-differ- generative response.

Birth Defects Research (Part C) 84:265–280, (2008) 278 MUNEOKA ET AL.

REFERENCES velopment, regeneration, and wound erating axolotl limbs. Development healing in axolotl limbs. J Exp Zool 121:1731–1741. Abzhanov A, Protas M, Grant BR, et al. 282:715–723. Gardiner DM, Endo T, Bryant SV. 2004. Bmp4 and morphological vari- Casanova JC, Sanz-Ezquerro JJ. 2007. 2002. The molecular basis of am- ation of beaks in Darwin’s finches. Digit morphogenesis: is the tip differ- phibian limb regeneration: integrat- Science 305:1462–1465. ent? Dev Growth Differ 49:479–491. ing the old with the new. Semin Cell Acampora D, Merlo GR, Paleari L, et al. Catron KM, Zhang H, Marshall SC, Dev Biol 13:345–352. 1999. Craniofacial, vestibular and et al. 1995. Transcriptional repres- Godwin AR, Capecchi MR. 1998. bone defects in mice lacking the Dis- sion by Msx-1 does not require Hoxc13 mutant mice lack external tal-less-related gene Dlx5. Develop- homeodomain DNA-binding sites. Mol hair. Genes Dev 12:11–20. ment 126:3795–3809. Cell Biol 15:861–871. Goss RJ. 1969. Principles of Regenera- Allan CH, Fleckman P, Fernandes RJ, Chadwick RB, Bu L, Yu H, et al. 2007. tion. New York: Academic Press. et al. 2006. Tissue response and Digit tip regrowth and differential Chapter 11, pp.223–255. Msx1 expression after human fetal gene expression in MRL/Mpj, DBA/2, Goss RJ, Van Praagh A, Brewer P. digit tip amputation in vitro. Wound and C57BL/6 mice. Wound Repair 1992. The mechanism of antler cast- Repair Regen 14:398–404. Regen 15:275–284. ing in the fallow deer. J Exp Zool Allen SP, Maden M, Price JS. 2002. A Chang HY, Chi JT, Dudoit S, et al. 264:429–436. role for retinoic acid in regulating the 2002. Diversity, topographic differ- Gourevitch D, Clark L, Chen P, et al. regeneration of deer antlers. Dev entiation, and positional memory in 2003. Matrix metalloproteinase activ- Biol 251:409–423. human fibroblasts. Proc Natl Acad Sci ity correlates with blastema forma- Angermann P, Lohmann M. 1993. Inju- USA 99:12877–12882. tion in the regenerating MRL mouse ries to the hand and wrist. A study of Chen Y, Zhang Y, Jiang TX, et al. 2000. ear hole model. Dev Dyn 226:377– 50,272 injuries. J Hand Surg [Br] Conservation of early odontogenic 387. 18:642–644. signaling pathways in Aves. Proc Natl Grotewold L, Plum M, Dildrop R, et al. Badylak SF. 2007. The extracellular Acad Sci USA 97:10044–10049. 2000. Bambi is coexpressed with matrix as a biologic scaffold material. Clark LD, Clark RK, Heber-Katz E. Bmp-4 during mouse embryogenesis. Biomaterials 28:3587–3593. 1998. A new murine model for mam- Mech Dev 100:327–330. Barling PM, Liu H, Matich J, et al. malian wound repair and regenera- Hammond KL, Hanson IM, Brown AG, 2004. Expression of PTHrP and the tion. Clin Immunol Immunopathol et al. 1998. Mammalian and Dro- PTH/PTHrP receptor in growing red 88:35–45. sophila dachshund genes are related deer antler. Cell Biol Int 28:661–673. Crews L, Gates PB, Brown R, et al. to the Ski proto-oncogene and are Barling PM, Lai AK, Tong AST, et al. 1995. Expression and activity of the expressed in eye and limb. Mech Dev 2005. The distribution of growth fac- newt Msx-1 gene in relation to limb 74:121–131. tors and their receptors in growing regeneration. Proc Biol Sci 259:161– Han M, Yang X, Farrington JE, et al. red deer antlers. In: Suttie JM, edi- 171. 2003. Digit regeneration is regulated tor. Advances in Antler Science and Crisan M, Yap S, Casteilla L, et al. by Msx1 and BMP4 in fetal mice. Product Technology. Taieri Print, 2008. A perivascular origin for mes- Development 130:5123–5132. Dunedin: New Zealand. pp. 37–44. enchymal stem cells in multiple Han M, Yang X, Taylor G, et al. 2005. Beck CW, Christen B, Slack JM. 2003. human organs. Cell Stem Cell Limb regeneration in higher verte- Molecular pathways needed for 3:301–313. brates: developing a roadmap. Anat regeneration of spinal cord and mus- Davis RJ, Shen W, Heanue TA, et al. Rec B New Anat 287:14–24. cle in a vertebrate. Dev Cell 5:429– 1999. Mouse Dach, a homologue of Han M, Yang X, Lee J, et al. 2008. De- 439. Drosophila dachshund, is expressed velopment and regeneration of the Bhattacharya D, Ehrlich LI, Weissman in the developing retina, brain and neonatal digit tip in mice. Dev Biol IL. 2008. Space-time considerations limbs. Dev Genes Evol 209:526–536. 315:125–135. for hematopoietic stem cell trans- Deuchar E. 1976. Regeneration of Heber-Katz E. 1999. The regenerating plantation. Eur J Immunol 38:2060– amputated limb-buds in early rat mouse ear. Semin Cell Dev Biol 10: 2067. embryos. J Embryol Exp Morphol 415–419. Brockes JP, Kumar A. 2002. Plasticity 35:345–354. Holleville N, Mateos S, Bontoux M, and reprogramming of differentiated Douglas BS. 1972. Conservative man- et al. 2007. Dlx5 drives Runx2 cells in amphibian regeneration. Nat agement of guillotine amputation of expression and osteogenic differen- Rev Mol Cell Biol 3:566–574. the finger in children. Aust Paediatr J tiation in developing cranial suture Brockes JP, Kumar A, Velloso CP. 8:86–89. mesenchyme. Dev Biol 304:860– 2001. Regeneration as an evolution- Endo T, Tamura K, Ide H. 2000. Analy- 874. ary variable. J Anat 199:3–11. sis of gene expressions during Xeno- Howard D, Buttery LD, Shakesheff KM, Brockes JP, Kumar A. 2005. Append- pus forelimb regeneration. Dev Biol et al. 2008. Tissue engineering: age regeneration in adult vertebrates 220:296–306. strategies, stem cells and scaffolds. and implications for regenerative Faucheux C, Nicholls BM, Allen S, et al. J Anat 213:66–72. medicine. Science 310:1919–1923. 2004 Recapitulation of the parathy- Hu G, Lee H, Price SM, et al. 2001. Msx Bryant SV, French V, Bryant PJ. 1981. roid hormone-related peptide-Indian homeobox genes inhibit differentia- Distal Regeneration and Symmetry. hedgehog pathway in the regenerat- tion through upregulation of cyclin Science 212:993–1002. ing deer antler. Dev Dyn 231:88–97. D1. Development 128:2373–2384. Bryant SV, Endo T, Gardiner DM. French V. 1976. Leg regeneration in Illingworth CM. 1974. Trapped fingers 2002. Vertebrate limb regeneration the cockroach, Blatella germanica. and amputated finger tips in chil- and the origin of limb stem cells. Int II. Regeneration from a non-congru- dren. J Pediatr Surg 9:853–858. J Dev Biol 46:887–896. ent tibial graft/host junction. J Jiang Y, Jahagirdar BN, Reinhardt RL, Carlson BM. 2005. Some principles of Embryol Exp Morphol 35:267–301. et al. 2002. Pluripotency of mesen- regeneration in mammalian systems. Gardiner DM. 2005. Ontogenetic chymal stem cells derived from adult Anat Rec B New Anat 287:4–13. decline of regenerative ability and marrow. Nature 418:41–49. Carlson BM. 2007. Principles of regen- the stimulation of human regenera- Kench JA, Russell DM, Fadok VA, et al. erative biology. Burlington, MA: Aca- tion. Rejuvenation Res 8:141–153. 1999. Aberrant wound healing and demic Press. Gardiner DM, Blumberg B, Komine Y, TGF-beta production in the auto Carlson MR, Bryant SV, Gardiner DM. et al. 1995. Regulation of HoxA immune-prone MRL/1 mouse. Clin 1998. Expression of Msx-2 during de- expression in developing and regen- Immunol 92:300–310.

Birth Defects Research (Part C) 84:265–280, (2008) MAMMALIAN REGENERATION 279

Khan IM, Gilbert SJ, Singhrao SK, Neufeld DA. 1985. Bone healing after Rinn JL, Bondre C, Gladstone HB, et al. et al. 2008. Cartilage integration: amputation of mouse digits and newt 2006. Anatomic demarcation by posi- evaluation of the reasons for failure limbs: implications for induced tional variation in fibroblast gene of integration during cartilage repair. regeneration in mammals. Anat Rec expression programs. PLoS Genet A review. Eur Cell Mater 16:26–39. 211:156–165. 2:e119. Kierdorf U, Kierdorf H, Szuwart T. Neufeld DA. 1992. Digital regeneration in Rolf HJ, Kierdorf U, Kierdorf H, et al. 2007. Deer antler regeneration: mammals. In: Bone, Vol. 5. CRC Press. 2008. Localization and characteriza- cells, concepts, and controversies. Boca Raton, Florida pp. 287–312. tion of STRO-1 cells in the deer pedi- J Morphol 268:726–738. Neufeld DA, Hosman S, Yescas T, et al. cle and regenerating antler. PLoS Koshiba K, Kuroiwa A, Yamamoto H, 2004. Actin fiber patterns detected ONE 3:e2064. et al. 1998. Expression of Msx genes by Alexafluor 488 phalloidin suggest Sadighi M, Haines SR, Skottner A, in regenerating and developing similar cell migration in regenerating et al. 1994. Effects of insulin-like limbs of axolotl. J Exp Zool 282:703– and nonregenerating rodent toes. growth factor-I (IGF-I) and IGF-II on 714. Anat Rec A Discov Mol Cell Evol Biol the growth of antler cells in vitro. Kostakopoulou K, Vogel A, Brickell P, 278:450–453. J Endocrinol 143:461–469. et al. 1996. ‘Regeneration’ of wing Neufeld DA, Zhao W. 1995. Bone Samulewicz SJ, Seitz A, Clark L, et al. bud stumps of chick embryos and regrowth after digit tip amputation in 2002. Expression of preadipocyte reactivation of Msx-1 and Shh mice is equivalent in adults and neo- factor-1(Pref-1), a delta-like protein, expression in response to FGF-4 and nates. Wound Repair Regen 3:461– in healing mouse ears. Wound Repair ridge signals. Mech Dev 55:119–131. 466. Regen 10:215–221. Lai AK, Hou WL, Verdon DJ, et al. Odelberg SJ, Kollhoff A, Keating MT. Sanchez Alvarado A. 2000. Regenera- 2007. The distribution of the growth 2000. Dedifferentiation of mamma- tion in the metazoans: why does it factors FGF-2 and VEGF, and their lian myotubes induced by msx1. Cell happen? Bioessays 22:578–590. receptors, in growing red deer antler. 103:1099–1109. Sanchez Alvarado A, Tsonis PA. 2006. Tissue Cell 39:35–46. Oike Y, Yasunaga K, Ito Y, et al. 2003. Bridging the regeneration gap: Lee H, Habas R, Abate-Shen C. 2004. Angiopoietin-related growth factor genetic insights from diverse animal MSX1 cooperates with histone H1b (AGF) promotes epidermal prolifera- models. Nat Rev Genet 7:873–884. for inhibition of transcription and myo- tion, remodeling, and regeneration. Satokata I, Maas R. 1994. Msx1 defi- genesis. Science 304:1675–1678. Proc Natl Acad Sci USA 100:9494– cient mice exhibit cleft palate and Lee KK, Chan WY. 1991. A study on 9499. abnormalities of craniofacial and the regenerative potential of partially Oleske DM, Hahn JJ. 1992. Work- tooth development. Nat Genet excised mouse embryonic fore-limb related injuries of the hand: data 6:348–356. bud. Anat Embryol (Berl) 184:153– from an occupational injury/illness Satokata I, Ma L, Ohshima H, et al. 157. surveillance system. J Community 2000. Msx2 deficiency in mice causes Lee LP, Lau PY, Chan CW. 1995. A sim- Health 17:205–219. pleiotropic defects in bone growth ple and efficient treatment for finger- Pendegrass CJ, Goodship AE, Price JS, and ectodermal organ formation. Nat tip injuries. J Hand Surg [Br] 20:63– et al. 2006. Nature’s answer to Genet 24:391–395. 71. breaching the skin barrier: an inno- Schindeler A, Morse A, Harry L, et al. Li C, Suttie JM, Clark DE. 2005. Histo- vative development for amputees. 2008. Models of tibial fracture heal- logical examination of antler regen- J Anat 209:59–67. ing in normal and Nf1-deficient mice. eration in red deer (Cervus elaphus). Price JS, Oyajobi BO, Oreffo RO, et al. J Orthop Res 26:1053–1060. Anat Rec A Discov Mol Cell Evol Biol 1994. Cells cultured from the grow- Simon HG, Nelson C, Goff D, et al. 282:163–174. ing tip of red deer antler express 1995. Differential expression of myo- Li X, Gu W, Masinde G, et al. 2001. alkaline phosphatase and proliferate genic regulatory genes and Msx-1 Genetic control of the rate of wound in response to insulin-like growth during dedifferentiation and rediffer- healing in mice. Heredity 86:668– factor-I. J Endocrinol 143:R9–R16. entiation of regenerating amphibian 674. Price JS, Allen S, Faucheux C, et al. limbs. Dev Dyn 202:1–12. Mescher AL, Neff AW. 2005. Regenera- 2005. Deer antlers: a zoological curi- Slack JM, Lin G, Chen Y. 2008. The tive capacity and the developing osity or the key to understanding Xenopus tadpole: a new model for immune system. Adv Biochem Eng organ regeneration in mammals? regeneration research. Cell Mol Life Biotechnol 93:39–66. J Anat 207:603–618. Sci 65:54–63. Metcalfe AD, Willis H, Beare A, et al. Rajnoch C, Ferguson S, Metcalfe AD, Sorock GS, Smith E, Hall N. 1993. Hos- 2006. Characterizing regeneration in et al. 2003. Regeneration of the ear pitalized occupational finger amputa- the vertebrate ear. J Anat 209:439– after wounding in different mouse tions, New Jersey, 1985 and 1986. 446. strains is dependent on the severity Am J Ind Med 23:439–447. Morrison JI, Loof S, He P, et al. 2006. of wound trauma. Dev Dyn 226:388– Suzuki HK, Mathews A. 1966. Two- Salamander limb regeneration 397. color fluorescent labeling of mineral- involves the activation of a multipo- Ramirez-Castillejo C, Sanchez-Sanchez izing tissues with tetracycline and tent skeletal muscle satellite cell F, Andreu-Agullo C, et al. 2006. Pig- 2,4-bis[N,N0-di-(carbomethyl)amino- population. J Cell Biol 172:433–440. ment epithelium-derived factor is a methyl] fluorescein. Stain Technol Mount JG, Muzylak M, Allen S, et al. niche signal for neural stem cell 41:57–60. 2006. Evidence that the canonical renewal. Nat Neurosci 9:331–339. Tanaka EM. 2003. Cell differentiation Wnt signalling pathway regulates Reginelli AD, Wang YQ, Sassoon D, and cell fate during urodele tail and deer antler regeneration. Dev Dyn et al. 1995. Digit tip regeneration limb regeneration. Curr Opin Genet 235:1390–1399. correlates with regions of Msx1 Dev 13:497–501. Muller TL, Ngo-Muller V, Reginelli A, (Hox 7) expression in fetal and new- Taylor GP, Anderson R, Reginelli AD, et al. 1999. Regeneration in higher born mice. Development 121:1065– et al. 1994. FGF-2 induces regenera- vertebrates: limb buds and digit 1076. tion of the chick limb bud. Dev Biol tips. Semin Cell Dev Biol 10:405– Revardel JL, Chebouki F. 1987. Etude 163:282–284. 413. de la reponse a l’amputation des Thummel R, Bai S, Sarras MP, et al. Muneoka K, Sassoon D. 1992. Molecu- phalanges chez la souris: role mor- 2006. Inhibition of zebrafish fin lar aspects of regeneration in devel- phogenetique des epitheliums, stim- regeneration using in vivo electropo- oping vertebrate limbs. Dev Biol ulation de la chondrogenese. Can J ration of morpholinos against fgfr1 152:37–49. Zool 65:3166–3176. and msxb. Dev Dyn 235:336–346.

Birth Defects Research (Part C) 84:265–280, (2008) 280 MUNEOKA ET AL.

Urist MR, Behnam K, Kerendi F, et al. Williams-Boyce PK, Daniel JC Jr. 1986. expression is responsible for regen- 1997. Lipids closely associated with Comparison of ear tissue regenera- erative capacity in Xenopus limb bone morphogenetic protein (BMP)— tion in mammals. J Anat 149:55–63. buds. Dev Biol 219:18–29. and induced heterotopic bone forma- Wu P, Jiang TX, Suksaweang S, et al. Zhang H, Catron KM, Abate-Shen C. tion. With preliminary observations 2004. Molecular shaping of the beak. 1996. A role for the Msx-1 homeodo- of deficiencies in lipid and osteoin- Science 305:1465–1466. main in transcriptional regulation: duction in lathyrism in rats. Connect Yamanaka S. 2008. Pluripotency and residues in the N-terminal arm Tissue Res 36:9–20. nuclear reprogramming. Philos Trans mediate TATA binding protein inter- Vidal P, Dickson MG. 1993. Regenera- R Soc Lond B Biol Sci 363:2079–2087. action and transcriptional repression. tion of the distal phalanx. A case Yang EV, Gardiner DM, Carlson MR, Proc Natl Acad Sci USA 93:1764– report. J Hand Surg [Br] 18:230–233. et al. 1999. Expression of Mmp-9 1769. Wanek N, Muneoka K, Bryant SV. and related matrix metalloproteinase Zhang H, Hu G, Wang H, et al. 1997. 1989. Evidence for regulation follow- genes during axolotl limb regenera- Heterodimerization of Msx and Dlx ing amputation and tissue grafting in tion. Dev Dyn 216:2–9. homeoproteins results in functional the developing mouse limb. J Exp Yokoyama H, Yonei-Tamura S, Endo T, antagonism. Mol Cell Biol 17:2920– Zool 249:55–61. et al. 2000. Mesenchyme with fgf-10 2932.

Birth Defects Research (Part C) 84:265–280, (2008)