Review For reprint orders, please contact: [email protected] Proximal to distal patterning during limb development and regeneration: a review of converging disciplines

Regeneration of lost structures typically involves distinct events: wound healing at the damaged site, the accumulation of cells that will be used as future building blocks and, finally, the initiation of molecular signaling pathways that dictate the form and pattern of the regenerated structures. Amphibians and urodeles in particular, have long been known to have exceptional regenerative properties. For many years, these animals have been the model of choice for understanding limb regeneration, a complex process that involves reconstructing skin, muscle, bone, connective tissue and nerves into a functional 3D structure. It appears that this process of rebuilding an adult limb has many similarities with how the limb forms in the first place – for example, in the embryo, all the components of the limb need to be formed and this requires signaling mechanisms to specify the final pattern. Thus, both limb formation and limb regeneration are likely to employ the same molecular pathways. Given the available tools of molecular biology and genetics, this is an exciting time for both fields to share findings and make significant progress in understanding more about the events that dictate embryonic limb pattern and control limb regeneration. This article focuses particularly on what is known about the molecular control of patterning along the proximal–distal axis.

KEYWORDS: blastema n digit n limb development n limb regeneration Francesca V Mariani n proximal–distal patterning n stem cell Eli & Edythe Broad Center for Regenerative Medicine & Stem Cell Research, Keck School of Medicine, Limb development & regeneration shell that, in tetrapods (but not fish), forms a University of Southern California, 1450 Biggy Street, NRT-4505, Over 250 years ago, the dedicated Italian sci- thickening at the distal tip called the apical Los Angeles, CA 90033-9601, USA entist, Lazzaro Spallanzani, first described the ectodermal ridge (AER). The AER is a site of Tel.: +1 323 442 7855 ability of urodeles (salamanders) to regenerate expression for a number of growth factors that Fax: +1 323 442 7899 [email protected] appendages [1]. Since that time, scientists have are actively being studied for their role in limb been fascinated by the idea that our organs and bud outgrowth and patterning [4,5]. As cells in appendages may have the capacity to regener- the underlying mesenchyme proliferate, the ate, and hope for a time when regeneration can bud enlarges. Next, some of the cells differen- be enhanced in the clinic. A tremendous effort tiate into chondrocytes that coalesce into con- will be required to understand regenerative densations. Interestingly, these condensations processes in such a way that patients can truly are arranged in a pattern that prefigures the benefit. Interestingly, while it is clear that there different skeletal elements. The condensations are some essential differences between limb elaborate, grow and, subsequently, differenti- regeneration and how they form during embry- ate into cartilage. The cartilage then serves as onic development, recent studies suggest that a ‘template’ for the later differentiation into both processes share a basic molecular ‘toolkit’. bone, a process called endochondral bone for- Indeed, regeneration may involve the reactiva- mation. The limb mesenchyme will differenti- tion of the developmental program to generate ate into chondrocytes and connective tissues missing structures [2,3]. Perhaps if we can learn (tendons and muscle sheath), while muscle, more about this toolkit using modern molecu- blood vessels and nerves develop from cells that lar and genetic techniques, it will be possible, migrate into the limb bud from other locations one day, to stimulate appendage regeneration in the embryo. in humans. „„Limb regeneration „„Limb development In general, vertebrates have a limited ability to The vertebrate limb typically develops dur- regenerate, with urodeles, fish and premeta- ing mid-gestation as a bud of mesenchyme morphosing frogs being the only vertebrates that protrudes from the flank of the embryo known to regenerate a complete appendage. (Figure 1A). This bud is encased in an epithelial Amphibian limb regeneration can be most

10.2217/RME.10.27 © 2010 Future Medicine Ltd Regen. Med. (2010) 5(3), 451–462 ISSN 1746-0751 451 Review Mariani

Limb development

Bud formation AER formation Bud outgrowth Chondrocyte Cartilage formation and patterning condensation, morphogenesis Limb regeneration Urodeles Injury Wound healing and Blastema formation Chondrocyte Cartilage formation AEC formation and growth condensation, morphogenesis Mammals Injury Wound healing Growth and Regenerated bone local proliferation

Limb skeletal morphology

` ximal ` Dista l Pro

Stylopod Zeugopod Autopod

Figure 1. The steps of limb development and regeneration compared. (A) Limb development. (B) Limb regeneration in urodeles and digit-tip regeneration in mammals. After wound healing, the first similarity is the formation of a epithelium at the distal tip. This epithelium is called the AER during development and the AEC during regeneration. Within both the limb-bud mesenchyme and the regenerating blastema, cells proliferate, acquire positional information and respond to inductive signals. The cells differentiate and become organized to form the limb elements. During both limb development and regeneration in salamanders, cartilage prefigures the formation of bone. In mouse digit-tip regeneration, bone forms directly. (C) Final skeletal morphology of a human limb with the three limb segments color coded: stylopod (gray), zeugopod (yellow) and autopod (purple). The proximal and distal axis is indicated. AEC: Apical epithelial cap; AER: Apical ectodermal ridge.

easily described as a series of defined events (out- through the process of patterning, chondrocyte lined in Figure 1B). After an injury occurs, there condensation, cartilage formation and bone is an immediate clotting response and the mobi- deposition [2,6,7]. Interestingly, the site of injury lization of the immune cells involved in wound can occur at any proximal to distal (P–D) loca- healing. Very quickly, epithelial cells cover over tion along the limb and the cells in the blastema the wound area and form a thickened structure faithfully replace the missing parts. called the apical epithelial cap (AEC), a struc- ture that is probably analogous to the AER in the „„What are the questions in the developing limb bud. Underneath the AEC, cells regeneration field? with an undifferentiated or mesenchymal mor- In the mid-1980s, Stocum outlined the central phology accumulate to form a structure called questions in the field of limb regeneration [6] the blastema. Like cells in the developing limb and, although some progress has been made, bud, cells in the blastema proliferate and eventu- these central questions are still unanswered and ally undergo differentiation and morphogenesis have been recast here in modern terms.

452 Regen. Med. (2010) 5(3) future science group Proximal to distal patterning during limb development & regeneration Review

The first question concerns the origin of P–D patterning: three models the blastema cells. Where do these cells come to consider from? Do they originate from differentiated „„The Progress Zone Model skeletal, connective tissue and muscle cells in Currently, there are three models for how P–D the remaining stump? If so, do they dediffer- patterning information might be acquired and entiate to revert to a stem cell or embryonic interpreted either during limb development or state? Alternatively, are the blastema cells regeneration: the Progress Zone Model, the derived from resident stem cell populations in Early-specification Model and the Two-signal the adult stump that then move into the blas- Model (Figure 2). For many years, the Progress tema and build new structures? Related to these Zone Model, proposed by Lewis Wolpert [21], questions is the issue of whether or not cells in has been the prevailing framework for explain- the blastema are multipotent cells or are already ing how the limb develops. This model proposes restricted in potential. These are active research that a special region of defined size, called the questions [8] and the answers will be of great progress zone, exists in the distal mesenchyme. interest to the field. Wolpert and colleagues postulated that cells in The second central question, and the focus of this zone are under the influence of the AER, this article, concerns the patterning of the blas- which plays a ‘permissive’ rather than ‘instruc- tema, a structure that appears to have similari- tive’ role, meaning the AER does not impart pat- ties to the embryonic limb bud. What signals terning information to the cells but keeps the instruct the blastema cells to rebuild the missing cells ‘labile’ and able to change positional infor- parts of the limb so that the regenerated struc- mation. The cells are also proposed to be able ture has the correct patterning, morphology and to ‘mark time’ as they are within the progress polarity? Where do these signals come from? zone and become more ‘distalized’ the longer Do cells at the cut edge retain some informa- they reside in the zone. Therefore, as the limb tion that can be conveyed to the new structures bud expands, the first cells to leave the progress as they form? For convenience, the vertebrate zone are the least distalized and postulated to limb can be described as having three axes: contribute to proximal structures. Cells that an anterior–posterior axis (thumb to pinky), leave the progress zone last contribute to distal a dorsal–ventral axis (back of the hand to the structures as they have resided the longest in the palm) and a P–D axis (base of the limb to tip progress zone and are, therefore, the most distal- of the finger). The P–D axis can be further ized. Thus, the Progress Zone Model proposes divided into three different segments: a most that the P–D axis of the limb becomes specified proximal element (stylopod or upper arm/leg), a in a P–D direction through time. middle element (zeugopod or forearm/leg) and While no one has been able to identify a a distal element (autopod or wrist/ankle and mole­cular mechanism whereby progress zone hand/foot) (Figure 1C). So far, clues for how the cells mark time, for many years the results of an axes are molecularly defined largely come from important experiment carried out on the limb studies on how limbs form during development. of developing chickens provided support for the Although there are some differences in expres- Progress Zone Model [22]. In this experiment, sion patterns, genes important for limb develop- the AER was removed from the limb at different ment, such as the growth factors Fgf8 [9,10] and stages of embryo development (chicken embryos Shh [11], and the Hox [12–14], Tbx [15,16] and Dlx are accessible via a window through the shell) [17] family of transcription factor genes, are all and the resulting limb was analyzed. Early AER expressed again during regeneration and, thus, removals resulted in an abrupt proximal trunca- the same molecular pathways used in develop- tion while later AER removals resulted in a distal mental patterning may be recapitulated during truncation. This result was then interpreted in the regeneration. context of the Progress Zone Model. The removal Of the three axes, the most progress has come of the AER was thought to halt the counting of from studies investigating how the P–D axis is time by cells in the progress zone. Thus, after established in both developing and regenerating early AER removal, all the cells contribute to the limbs. And, although P–D patterning has been most proximal structures and resulted in limbs studied for many years and there are now some with proximal truncations. While in older limb clues for how patterning occurs [14,18–20], how buds, proximal and middle specification had cells specifically acquire the necessary informa- already occurred at the time of AER removal. tion to become different proximal and distal AER removal would then halt any further distal structures is still an active area of investigation. specification resulting in distal truncations.

future science group www.futuremedicine.com 453 Review Mariani

Progress Zone Model

Progress zone

SZ AA

Early-specification Model

SZ AA

Two-signal Model with intercalation

SZ AA

Figure 2. Models for proximal–distal patterning. Each model is depicted with three different domains that will prefigure the three segments of the limb. The gray box represents stylopod precursors (S), yellow represents the zeugopod precursors (Z) and purple represents the most distal autopod precursors (A). The apical ectodermal ridge (AER)/apical epithelial cap is represented by a purple line at the distal tip. (A) The Progress Zone Model postulates that cells underneath the AER (depicted in green) are able to mark time and that the AER keeps them in a labile state. As growth occurs, cells leaving the progess zone early will form proximal structures. The longer cells reside in the progess zone, the more distalized they become, thus, the last ones to leave the zone form progressively more distal structures [21]. (B) The Early-specification Model postulates that all the domains are specified early and expand with growth [23]; where the patterning information comes from is not clear. (C) The Two-signal Model with intercalation is a modification of the Early- specification Model. Two opposing signals are proposed to arise from the proximal end and from the distal end. These signals establish proximal and distal domains. The third, middle domain is proposed to form as a consequence of these two signals or two domains interacting. Retinoic acid has properties of the proposed proximalizing signal but so far there is no genetic evidence to support retinoic acid as a proximalizing factor [38,40], and FGFs from the AER likely act as the distalizing signal. This model is supported by experiments in both limb development and limb regeneration [19,20]. A: Autopod; S: Stylopod; Z: Zeugopod. This figure is drawn based on concepts presented in [4,19].

454 Regen. Med. (2010) 5(3) future science group Proximal to distal patterning during limb development & regeneration Review

„„Challenges to the Progress Zone Progress Zone Model, despite Sauder’s original Model & the Early-specification Model efforts to examine these parameters [22] (see [4] In 2002, Dudley, Ros and Tabin revisited this for a more detailed discussion). key experiment using precise lineage tracing and A further challenge to the Progress Zone modern techniques for detecting cell death and Model came from experiments in which proliferation in chicken embryos. Their experi- growth-factor signaling in the AER was modi- ments suggested that the limb truncations upon fied during limb bud outgrowth. By employing AER removal could have a very different inter- genetic techniques during mouse development, pretation than that originally envisioned by sup- the signaling of proteins in the FGF family porters of the Progress Zone Model [23]. This (four of which are expressed in the AER) was prompted a rethinking of the Progress Zone eliminated or reduced, and the resulting skeletal Model and the authors postulated a very dif- phenotypes were analyzed [19,24–26]. Some of the ferent view that P–D specification occurs early phenotypes were very interesting because distal on in limb development rather than progres- structures could form even when more proximal sively through time. This model, called the ones were reduced or missing [19]. This observa- Early-specification Model, was envisioned as the tion is not easily explained by the Progress Zone establishment of three zones that might prefig- Model, where specification is proposed to hap- ure the three segments of the limb (Figure 2B). pen in a P–D sequence. However, a model like To support this model, they first demonstrated the Early-specification Model, where proximal that cells in the very early limb bud (just as the and distal specification occurs early does fit with bud begins to protrude) contribute to proximal, these observations. middle and distal structures without mixing. For example, cells labeled in a proximal location „„Two-signal Model: a synthesis (300 µm under the AER) gave rise to proximal between regeneration & development structures, while those labeled in a more distal During limb regeneration, P–D specification location (100 µm under the AER) gave rise to of the different proximal, middle or distal seg- distal structures. Thus, they constructed a well- ments could also happen early on in the blas- defined fate map for the very early limb bud. tema. However, similar to developing limb This fate map is consistent with the idea that buds, the question remains, how does this early specification happens early, although, in itself, specification occur? Where does the ‘positional does not rule out the Progress Zone Model. information’ come from to establish these three However, importantly, this lineage tracing could domains, especially during limb regeneration, then be used to re-examine the AER removal when only the missing structures need to be experiments that had seemed to fit the Progress regenerated? While it is entirely possible that Zone Model so nicely. Cells at different P–D patterning during development occurs by a dif- locations were labeled and then the AER was ferent mechanism than patterning during regen- removed at early and late stages. Remarkably, eration, a modification of the Early-specification in limbs with proximal truncations, distally Model that includes two opposing signals can be marked cells were not converted to proximal applied in both contexts. This Two-signal Model structures as predicted by the Progress Zone (Figure 2C) suggests that opposing signals emanate Model but instead, were removed by apoptosis, from the distal or proximal end to either distal- which occurs in the mesenchyme after AER ize or proximalize the limb mesenchyme. These removal (also see [24]). Similarly, distally marked signals are proposed to then specify a proximal cells in limbs with distal truncations failed to and distal domain, with the size of each domain proliferate and contribute to the distal limb, being proportional to the strength of the signal. while proximally labeled cells (which at this late A third middle domain might subsequently stage are far from the AER) expanded normally form as a result of an interaction between the and contributed to proximal structures. They two signals, or between the two domains at the concluded that distal progenitor cells did not border. This mechanism for acquiring positional change their fate specification as suggested by information was put forward to explain the phe- the Progress Zone Model but that the popula- notypes resulting from a loss or reduction in FGF tion fated to become the missing structures was signaling in the AER [19]. In addition, studies simply depleted. Thus, the AER was demon- using these FGF mutants provided evidence that strated to provide an important survival and FGFs emanating from the AER act as an instruc- proliferative function for the underlying cells, tive signal to specify distal cell fates and, there- which was not appreciated by advocates of the fore, may be the distalizing signal in this model.

future science group www.futuremedicine.com 455 Review Mariani

Two-signal Model for P–D fact, as we shall see in the next section, there patterning: origin & possible appears to be a conserved ‘toolkit’ for pattern- molecular mechanisms ing of appendages despite the lack of homology „„Studies on invertebrates appendage between arthropods and vertebrates [27,33]. regeneration & development The idea that proximal and distal domains might „„P–D patterning during vertebrate be specified first and that a middle domain might limb development be specified later via P–D signal interactions has Potential proximalizing factors actually been around for a long time (since the During early mouse limb development, reti- 1970s). Early studies in insect appendage develop- noic acid (RA) signaling can be detected in the ment and regeneration proposed that patterning lateral-plate mesoderm and is later enriched in occurs via an interaction between proximal and the proximal portion of the limb [34]. When distal domains (reviewed in [27]). For example, the RA-soaked beads are applied to developing ablation and juxtaposition of leg parts from dif- chicken limbs, distal cells can be converted ferent P–D locations in larval cockroach append- into proximal ones [20,35], suggesting that RA ages suggested that proximal and distal domains signaling can overcome distal specification and, communicate with each other to regenerate therefore, might act as the proximalizing signal the missing intermediate part, a process called during limb development. Interestingly, the intercalation (reviewed in [28]). For example, if a homeobox genes, Meis1 and Meis2 (related to distal part is placed on a proximal one, interme- the proximal factor Hth in Drosophila) can be diate structures form with normal P–D polar- upregulated by RA [20]. In addition, analogous ity. Conversely, if a proximal part is placed on a to what occurs when Hth is ectopically expressed distal amputation, intermediate structures form during Drosophila leg development, overexpres- with reverse polarity. Similarly, during the for- son of Meis genes inhibits distal development mation of the Drosophila leg, Schubiger observed [36] and can cause distal cells to express more that proximal and distal parts are the first to dif- proximal markers [20,37]. While these results are ferentiate when metamorphosis was stimulated intriguing, solid genetic evidence that either RA precociously, while intermediate structures were signaling or Meis1/2 is required for specifying only identifiable in older animals[29] . proximal identities is still needed. Genetic stud- During limb development, P–D specification ies removing Raldh2 and Raldh3 (genes encod- in the Drosophila leg involves coordination of ing enzymes essential for RA production) have Wnt and TGF‑b signaling (Dpp) to set up proxi- demonstrated that RA is required for limb bud mal and distal domains [30]. Expression of the initiation (reviewed in [38,39]). Further studies, transcription factor Dll marks the prospective in which RA is exogenously provided to these distal cells in the leg as the imaginal disc [27], enzyme-deficient animals in order to rescue while the Meis family gene (Hth) and Esg are limb bud development, resulted not only in expressed in the prospective proximal domain. the formation of normally patterned limbs, These genes establish proximal and distal but also in the lack of any detectable expres- domains, which then become separated by an sion of a RA-response element reporter in the intermediate domain marked by the expression proximal limb bud [40]. This suggests that RA of dac, a Ski/Sno-related gene that is required in signaling is not instructive for proximal limb Drosophila for the specification of intermediate identities. Instead, the authors suggest that fates [31]. Ectopic expression of the proximally RA may be important for providing a permis- expressed Meis family gene Hth in the distal sive environment in which P–D patterning domain induces dac, the intermediate marker. can occur. Likewise, RA does not appear to be In addition, cells expressing Hth aggregate and required for zebrafish P–D patterning using a assort from neighboring cells [27]. By contrast, similar approach [41]. Further ana­lysis will be ectopic expression of Dll in the proximal domain needed utilizing other RA-responsive reporters turns on distal markers [32]. Thus, during inver- to ­confirm this observation. tebrate appendage development, proximalizing and distalizing patterning factors appear to have Potential distalizing factors important opposing roles that define the early Signals that specify distal cell types could come proximal and distal domains. from the distal mesenchyme and/or from the Based on these observations, one possibility specialized distal ectoderm – the AER. Although is that insect and vertebrate appendages share signaling from the AER was only thought to be a similar P–D patterning mechanism and, in permissive by keeping underlying mesenchymal

456 Regen. Med. (2010) 5(3) future science group Proximal to distal patterning during limb development & regeneration Review cells in a labile state or by providing survival or Thus, it is hard to know how this segment might proliferative signals, the idea that FGFs found be specified. One possibility is that RA and FGF in the AER might have distal patterning activi- signals travel over several hundred micrometers ties was examined by Mercader et al. [20]. Using to cooperate at the center and induce the third the chick system, beads soaked in FGF‑8 pro- segment. Another possibility would involve a tein were found to restrict the expression of the new signal that arises at the juxtaposition of the proximally expressed Meis1, while the applica- two domains. tion of an FGF inhibitor resulted in the expan- sion of Meis1 expression into the distal tip of the „„P–D patterning during regeneration limb bud. Furthermore, FGF‑8 bead applica- In urodeles, after the blastema forms, the first tion inhibited the expression of the RA receptor, structures and genes to become apparent are Rarb, suggesting that FGF signaling from the the distal ones [14]. Later, intermediate struc- AER not only has a distal patterning function, tures located between the distal and remaining but may also act to oppose the proposed RA proximal regions are regenerated. Although this proximal patterning signal [20]. phenomenon has been observed for some time, Genetic evidence that FGFs from the AER it is still unclear how patterning information is have an instructive patterning function came conveyed to recreate the missing structures. For from the ana­lysis of mouse limbs that still had now, we can look to clues from studies of limb an AER, but lacked specific AER FGFs [19]. development to identify the potential molecular Limb buds that lacked Fgf8 and Fgf9 (moder- players involved. ate decrease in overall FGF signaling) were com- pared with limb buds that lacked Fgf8, Fgf4 and Specification of proximal identity one copy of Fgf9 (more severe decrease in FGF For proximal specification, RA has been shown, signaling). Importantly, gene expression was both during development (see previously) and compared at a stage when the limb buds were regeneration, to at least be capable of stimulating the same size so that the extent of gene expression proximal cell fates. RA application can stimu- could be measured. When FGF signaling was late proximal development in urodele limbs with moderately decreased, Meis1 expression encom- distal amputations in a dose-dependent fashion passed the first two thirds of the proximal limb [43–45]. For example, if the limb is amputated at bud. However, when FGF signaling was severely the level of the wrist, the blastema would nor- decreased, Meis1 expression extended almost to mally reform the rest of the autopod. If exposed the distal tip. This observation provided genetic to RA, however, this blastema can form a whole evidence that a decrease in FGF signaling could new limb distal to the cut site [43]. RA can also influence gene expression patterns independently downregulate distal markers in the blastema, of any effect FGFs might have on proliferation or such as Hoxa13 [12], and can stimulate proximal cell survival. In addition, these results provided markers, such as Hoxd10 [15]. support for the idea that FGF signaling from the In terms of what might lie downstream, a AER has a distalizing function that may establish glycosyl­phosphatidylinositol-anchored cell sur- a distal domain within the early limb bud. face protein called CD59 (also known as Prod1) Together, both these studies support the is more strongly expressed in proximal limbs and model that during development, limb pattern can be upregulated by RA application [46]. When is specified as two proximal and distal domains overexpressed in 4‑day‑old blastemas, CD59- with a signal from the flank (still to be geneti- expressing cells integrate into the proximal cally verified as RA) acting as the proximalizing region of the limb instead of the distal region of signal and FGFs from the AER acting as the dis- the regenerate. Thus, CD59 appears to promote talizing signals (Figure 2C). Evidence that at least proximal development in cells found in the distal two distinct populations exist in the early limb blastema (although this interpretation is some- bud comes from in vitro studies of early chick what confounded by the observation that the limb mesenchyme, where proximal and distal cells were compromised in their ability to divide cells were shown to not only behave differently via BrDU incorporation). In addition, regener- but to independently assort [42]. ated limbs in these experiments had reduced The lack of definitive markers for different distal development [47]. segments of the limb makes further investiga- Similar to Drosophila and vertebrate limbs, tion difficult. In particular, little is known about Meis genes are expressed in the proximal por- what molecular code might mark early progeni- tion of developing limbs and in the proximal tors of the intermediate segment (the zeugopod). portion of regenerating vertebrate blastemas, and

future science group www.futuremedicine.com 457 Review Mariani

its expression can be enhanced by RA applica- between the distal tip and proximal stump [56]. tion [18]. In addition, distal cells overexpress- Interestingly, this manipulation resulted in the ing Meis proteins in chicken embryos (with or formation of intermediate structures, suggesting without the cofactor Pbx1) tend to contribute to that either FGF signaling stimulates the expan- more proximal structures whereas cells express- sion of intermediate progenitors or may have a ing control lineage markers remained distally direct role in specifying intermediate structures located. Furthermore, overexpression of Meis1 at the P–D boundary. Clearly, further studies in mice using a transgenic approach results in will be needed to determine if FGFs serve a dis- altered expression of P–D patterning genes in tal patterning role during limb regeneration and a Pbx1-independent manner [48]. While these what role, if any, FGFs may play in specifying results are intriguing, a requirement for Meis intermediate or intercalated structures. genes has not been established since morpholino During limb bud development, the homeobox knockdown of Meis1 and Meis2 in regenerating (Hox) genes are expressed in interesting dynamic limbs has no effect. The knockdown of Meis1 patterns with HoxA cluster genes marking P–D and Meis2 did affect the ability of RA to have segments in accordance with their 5´–3´ chro- its proximalizing effect, suggesting that these mosomal locations. For example, Hoxa‑13 marks transcription factors are necessary to mediate an the distal tip of the developing limb bud while RA signal [18]. However, without genetic tests, the more 5´ gene Hoxa‑9 (which does extend the possibility that RA acts as a proximalizing expression into the more distal areas) marks the factor via CD59 and/or Meis1 and Meis2 can be more proximal end [57,58]. In regenerating uro- called into question. dele blastemas, as soon as the blastema begins to form, Hoxa‑9 and Hoxa‑13 are expressed and Specification of distal identity mark the cells with a distal code [12,14]. Over FGFs may be involved in several aspects of limb the next 10 days, as the blastema enlarges via regeneration from stimulating cell proliferation the recruitment of cells in the stump and by to providing the distalizing signal. Analogous to proliferation, Hoxa‑13 becomes restricted to the developing limb bud, Fgf8 expression can be the distal end leaving Hoxa‑9 to mark the more found in the blastema at the distal tip. In partic- intercalating structures. Thus, although it is ular, Fgf8 is expressed in a layer of mesenchyme not clear what signals (RA or FGFs) might be underlying the AEC, as well as within the AEC controlling the spatial or temporal expression basal layer [10,49]. Exogenous expression of FGFs pattern of these genes, it appears that the regen- in nonregenerative (denervated) adult urodele erating tip is distalized early on and that this is limbs (FGF2) and in frog (Xenopus laevis) lar- followed by the expression of genes that mark val hindlimbs (FGF10) can stimulate the regen- the intermediate/intercalating structures. It will erative process, suggesting that FGF signaling be interesting to see if this trend holds as more could be involved in distal patterning but may markers are examined. also be involved in blastema initiation and/or the Thus, there appear to be compelling pieces proliferation of blastema cells [50,51]. Ideally, the of evidence coming together that P–D pat- experiments applying FGF10 on larval Xenopus terning occurs by the same mechanism in hindlimbs need to be repeated since at least one both developing and regenerating limbs. To group was unable to repeat the result that FGF10 summarize, this mechanism could involve the application enhanced regeneration [52]. action of proximalizing and distalizing signals X. laevis is only able to regenerate the hindlimb that establish two domains. It is intriguing at larval stages. In urodeles, intercalary regenera- to think that these signals involve RA and tion of intermediate structures occurs when a FGFs; however, there is much to be done to distal blastema is juxtaposed with a proximal verify this and to determine what additional stump [53–55]. However, in the Xenopus hindlimb, signaling pathways might be involved. In addi- although excellent regeneration occurs when the tion, since limb development has been studied distal limb bud tip is removed, the juxtaposition extensively in chicken and mouse embryos, and of a distal limb bud tip with a proximal stump limb regeneration has been largely studied in fails to result in the formation/intercalation of urodeles (Ambystoma mexicanum) and frogs intermediate structures [56]. Shimizu-Nishikawa (X. laevis), ideally future work could compare et al. observed that Fgf8 expression is not as these events within the same species. This may extensive in the larval limb bud (compared be possible in X. laevis or Xenopus tropicalis, with the urodele blastema [49]) and implanted a where limb development could be compared bead soaked in FGF8 protein at the boundary with tadpole hindlimb or froglet forelimb

458 Regen. Med. (2010) 5(3) future science group Proximal to distal patterning during limb development & regeneration Review

regeneration [59]. Furthermore, since zebrafish or more completely. In amphibians, younger can regenerate fins[60] , genetic ana­lysis of both urodeles and frogs regenerate more quickly and fin bud development and regeneration could be completely than adult animals [55,59,66] and this investigated. Finally, we understand least about trend might also be evident during human digit- how an intermediate domain might be formed tip regeneration. In one study, the oldest child and if it forms subsequent to the establishment to display some regeneration was 11 years old of the proximal and distal domains. As devel- [64], but it is unclear what the human age limit opmental and regeneration biologists move might be. forward, increased progress in these areas can only be enhanced by increased communication „„Rodent digit tips as a model for and appreciation of the common questions still regeneration studies to be addressed. Hopefully, these findings can In mice, healing and regeneration of the digit also lead the way toward understanding more tip is also restricted to the most distal phalanx about regeneration and patterning of append- [67–69] and can occur in neonates as well as ages in a wider variety of species, and possibly adults. A recent study by Han et al. examined even humans. digit-tip regeneration of postnatal day‑3 mice and found that although the new tip looks very Mammalian digit-tip regeneration good, the regenerated structures are not reliably „„Human digit tips complete, that is, full length is not reached [62]. Although humans are not able to regenerate Over 10 years ago, studies indicated that a true complete limbs, human children are docu- blastema forms during mouse digit-tip regenera- mented to at least regenerate distal fingertips. tion [70]. Recently, Han et al. defined the region If the tip is available to reattach, healing occurs more carefully by analyzing proliferation at the nicely. However, in the absence of tissue for reat- distal tip with a BrDU incorporation assay. Just tachment, regeneration can occur with excel- 4 days following amputation, mesenchymal lent cosmetic and functional outcomes (sum- cells can be found underneath the wound and, marized in [61]; see also [62–65]). Interestingly, by 7 days, local proliferation is observed at the if the surgeon sutures the ends of the digit base of the blastema-like region. Thus, as dur- closed or uses a skin flap to repair the stump, ing amphibian regeneration, a blastema appears regeneration fails to occur and the digit never to form in regenerating digit tips. However, in attains normal length or morphology. Instead, contrast to amphibian regeneration and to limb a simple method of treatment allows digit-tip development, regeneration of the new skeletal regeneration to occur. This method starts with structure occurs by direct ossification (i.e., ossi- cleansing of the finger and debridement. Then, fication without a cartilage template) rather than no sutures or antibiotics are used, and a simple via the endochondral bone pathway (Figure 2C). bandage is applied. Even if the bone protrudes, No change is evident in various cartilage mark- substantial restoration of the tip can occur in ers but osteocalcin (indicating the presence of 11–12 weeks [64]. osteoblasts) is upregulated. Unfortunately, to date, a comprehensive study Very little is known about what molecules of these digit injuries in children has not been might be involved in the various steps of digit- carried out to determine the precise extent of tip regeneration and identifying the key players regeneration. Ideally an x‑ray image or other (RA or FGFs?) is an important area of future imaging techniques need to be performed at the inquiry. Certainly, bone morphogenetic pro- time of injury and compared with a follow-up tein (BMP) signaling is involved as mice null x‑ray after healing so that soft tissue and/or new for Msx1 (a downstream effector of the BMP bone regeneration can be quantified. In particu- pathway) fail to regenerate properly [71]. BMP4 lar, it would be interesting to determine how expression is also reduced in these animals and much can be removed from the tip to still result the regeneration defect can be rescued with in regeneration. So far, it appears that if the the application of BMP4 as well as BMP2 and injury occurs as far back as the second phalanx BMP7 [72]. Future studies will be needed to then regeneration does not occur and in one case characterize other players required for digit-tip no regeneration was observed when 70% of the regeneration as well as to identify key molecu- terminal digit was lost [62]. In addition, the qual- lar pathways that can be stimulated to enhance ity of regeneration needs to be correlated with the process. Given the trend, clues as to what the age of the patient to determine if younger pathways are involved are likely to come from patients might heal and regenerate more quickly studies on limb development.

future science group www.futuremedicine.com 459 Review Mariani

Future perspective skeletal element in the stylopod versus multiple Given the technical advances that are con- small elements in the distal autopod? stantly being made in genetics, genomics, small- ƒƒHow are the other axes, besides the P–D axis, molecule synthesis and cellular imaging, in specified (anterior–posterior, dorsal–ventral)? 5–10 years time we should have a much clearer What specifies left–right symmetry? picture about what molecular pathways and cel- ƒƒ lular changes occur during limb development What characteristics of young animals that are lacking in older animals speed the regenerative and regeneration [73,74]. Some of these findings may allow us to stimulate or enhance regenera- process? Is scarring a major inhibitor? tion in animals that do not usually regenerate ƒƒHow can dedifferentiation and/or the produc- very well or completely. Ultimately, the hope tion of stem cells be stimulated, and once a is that this knowledge will enter the clinic and blastema forms, what further events are make it possible not only to stimulate finger-tip needed to initiate the patterning programs? healing and regeneration but also to encourage regeneration after a more severe injury to the limb or after musculoskeletal injury occuring Acknowledgements elsewhere in the body. The author wishes to thank Gail Martin, Elly Tanaka, David Gardiner, members of her laboratory and several anonymous „„Future questions reviewers for helpful suggestions and discussions. Some of the There are still many areas to be addressed to themes in this review have also been discussed by others, and determine how patterning is orchestrated in both the author regrets that it was not possible to cite all of the developing and regenerating limbs: relevant literature due to space constraints. ƒƒWhile it is intriguing that RA and FGF signal- ing may be the important players in P–D pat- Financial & competing interests disclosure terning, to what extent are these pathways The author has no relevant affiliations or financial involve‑ required? Genetic techniques will be needed ment with any organization or entity with a financial interest to address this issue. in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, con‑ ƒƒHow are the distalizing and proximalizing sig- sultancies, honoraria, stock ownership or options, expert nals interpreted? That is, what gene networks ­testimony, grants or patents received or pending, or royalties. are initiated and what cellular changes occur No writing assistance was utilized in the production of this that result, for example, in a long proximal manuscript.

Executive summary An introduction to limb development & regeneration ƒƒ Similar morphological changes occur during both limb development and limb regeneration. ƒƒ The molecular mechanisms that underlie these morphological changes are likely to be similar. ƒƒ Patterning a developing or regenerating limb requires signaling mechanisms that specify the proximal–distal (P–D), anterior–posterior and dorsal–ventral axes. ƒƒ This article focuses on the recent understanding of P–D patterning during limb development and regeneration. P–D patterning: three models to consider ƒƒ Studies on limb development and regeneration, in both vertebrates and invertebrates, suggest that similar genes might be involved in patterning the P–D axis. ƒƒ A re-examination of the models for P–D axis specification has led to a convergence of views for how P–D domains are established. ƒƒ A model in which two opposing signals, a distalizing signal and a proximalizing signal, establish proximal and distal domains appears to explain both experimental results in developing and regenerating limbs. Intercalary patterning may also be involved. Two-signal Model for P–D patterning: origin & possible molecular mechanisms ƒƒ Early studies on insect limb regeneration set the stage for the Two-signal Model with intercalary patterning. ƒƒ During development, specification of the proximal axis probably involves a ‘proximalizing’ signal. Genetic evidence that this signal is retinoic acid or involves Meis1/2 is still to be obtained. ƒƒ There is strong genetic evidence that FGF signaling may provide the distalizing signal during limb development. ƒƒ Although there is evidence that retinoic acid and FGF signaling can have patterning roles during limb regeneration, further studies will be needed to confirm this and to understand how these signals might be interpreted. Mammalian digit-tip regeneration ƒƒ Human children and rodents are able to regenerate digit tips. ƒƒ Defining how limb development and regeneration occur in various animal models may help us understand how to stimulate the regenerative process in human adults.

460 Regen. Med. (2010) 5(3) future science group Proximal to distal patterning during limb development & regeneration Review

Bibliography 14 Gardiner DM, Bryant SV: Homeobox- 24 Sun X, Mariani FV, Martin GR: Functions of containing genes in limb regeneration. FGF signalling from the apical ectodermal Papers of special note have been highlighted as: In: HOX Gene Expression. Papageorgiou S (Ed.). ridge in limb development. Nature 418, n of interest Landes Bioscience and Springer 501–508 (2002). nn of considerable interest Science+Business Media, MA, USA 1–9 (2006). nn Provides evidence to challenge the Progress 1 Spallanzani L: Prodomo di un opera da n imprimersi sopra le riproduzioni animali. Clearly summarizes the data on Hox gene Zone Model. An Essay On Animal Reproductions. T Becket expression during limb regeneration. 25 Lewandoski M, Sun X, Martin GR: Fgf8 and PA de Hondt, London, UK (1769). 15 Simon HG, Tabin CJ: Analysis of Hox‑4.5 signalling from the AER is essential for 2 Bryant SV, Endo T, Gardiner DM: and Hox‑3.6 expression during newt limb normal limb development. Nat. Genet. 26, Vertebrate limb regeneration and the origin regeneration: differential regulation of 460–463 (2000). of limb stem cells. Int. J. Dev. Biol. 46, paralogous Hox genes suggest different roles 26 Boulet AM, Moon AM, Arenkiel BR, 887–896 (2002). for members of different Hox clusters. Capecchi MR: The roles of Fgf4 and Fgf8 in Development 117, 1397–1407 (1993). 3 Gardiner DM, Endo T, Bryant SV: limb bud initiation and outgrowth. Dev. Biol. The molecular basis of amphibian limb 16 Simon HG, Kittappa R, Khan PA, 273, 361–372 (2004). regeneration: integrating the old with the Tsilfidis C, Liversage RA, Oppenheimer S: 27 Goto S, Hayashi S: Proximal to distal cell new. Semin. Cell Dev. Biol. 13, 345–352 A novel family of T‑box genes in urodele communication in the Drosophila leg provides (2002). amphibian limb development and a basis for an intercalary mechanism of limb regeneration: candidate genes involved in 4 Mariani FV, Martin GR: Deciphering skeletal patterning. Development 126, 3407–3413 vertebrate forelimb/hindlimb patterning. patterning: clues from the limb. Nature 423, (1999). Development 124, 1355–1366 (1997). 319–325 (2003). 28 French V, Bryant PJ, Bryant SV: Pattern 17 Beauchemin M, Savard P: Two distal-less 5 Niswander, L: Pattern formation: old models regulation in epimorphic fields. Science 193, related homeobox-containing genes expressed out on a limb. Nat. Rev. Genet. 4, 133–143 969–981 (1976). in regeneration blastemas of the newt. (2003). 29 Schubiger G: Acquisition of differentiative Dev. Biol. 154, 55–65 (1992). 6 Stocum DL: The urodele limb regeneration competence in the imaginal leg of Drosophila. 18 Mercader N, Tanaka EM, Torres M: blastema. Determination and organization of Wilhelm Roux Arch. Entwmech. Org. 174, Proximodistal identity during vertebrate limb the morphogenetic field. Differentiation 27, 303–311 (1974). regeneration is regulated by Meis 13–28 (1984). 30 Lecuit T, Cohen SM: Proximal–distal axis homeodomain proteins. Development 132, formation in the Drosophila leg. Nature 388, n Classic paper that outlines the major 4131–4142 (2005). 139–145 (1997). questions in the field of limb regeneration. 19 Mariani FV, Ahn CP, Martin GR: Genetic 31 Mardon G, Solomon NM, Rubin GM: 7 Brockes JP, Kumar A: Appendage evidence that FGFs have an instructive role in Dachshund encodes a nuclear protein regeneration in adult vertebrates and limb proximal–distal patterning. Nature 453, required for normal eye and leg development implications for regenerative medicine. 401–405 (2008). in Drosophila. Development 120, 3473–3486 Science 310, 1919–1923 (2005). nn Provides genetic evidence that FGFs act as (1994). 8 Kragl M, Knapp D, Nacu E et al.: Cells keep distalizing factors. 32 Gorfinkiel N, Morata G, Guerrero I: a memory of their tissue origin during axolotl 20 Mercader N, Leonardo E, Piedra ME, The homeobox gene Distal-less induces limb regeneration. Nature 460, 60–65 Martinez AC, Ros MA, Torres M: Opposing ventral appendage development in Drosophila. (2009). RA and FGF signals control proximodistal Genes Dev. 11, 2259–2271 (1997). 9 Christensen RN, Weinstein M, Tassava RA: vertebrate limb development through 33 Pueyo JI, Couso JP: Parallels between the Expression of fibroblast growth factors 4, 8, regulation of Meis genes. Development 127, proximal–distal development of vertebrate and 10 in limbs, flanks, and blastemas of 3961–3970 (2000). and arthropod appendages: homology Ambystoma. Dev. Dyn. 223, 193–203 nn Proposes that two signals may be involved in without an ancestor? Curr. Opin. Genet. Dev. (2002). proximal to distal patterning during limb 15, 439–446 (2005). 10 Han MJ, An JY, Kim WS: Expression development. 34 Reynolds K, Mezey E, Zimmer A: Activity of patterns of FGF‑8 during development and 21 Summerbell D, Lewis JH, Wolpert L: the b‑retinoic acid receptor promoter in limb regeneration of the axolotl. Dev. Dyn. Positional information in chick limb transgenic mice. Mech. Dev. 36, 15–29 (1991). 220, 40–48 (2001). morphogenesis. Nature 244, 492–496 (1973). 35 Tamura K, Yokouchi Y, Kuroiwa A, Ide H: 11 Torok MA, Gardiner DM, nn Retinoic acid changes the proximodistal Izpisua-Belmonte JC, Bryant SV: Sonic Original paper that proposes the Progress developmental competence and affinity of hedgehog (shh) expression in developing and Zone Model. distal cells in the developing chick limb bud. regenerating axolotl limbs. J. Exp. Zool. 284, 22 Saunders JW Jr: The proximo–distal sequence Dev. Biol. 188, 224–234 (1997). 197–206 (1999). of origin of the parts of the chick wing and 36 Capdevila J, Tsukui T, Rodriquez Esteban C, 12 Gardiner DM, Blumberg B, Komine Y, the role of the ectoderm. J. Exp. Zool. 108, Zappavigna V, Izpisua Belmonte JC: Control Bryant SV: Regulation of HoxA expression in 363–403 (1948). of vertebrate limb outgrowth by the proximal developing and regenerating axolotl limbs. 23 Dudley AT, Ros MA, Tabin CJ: factor Meis2 and distal antagonism of BMPs Development 121, 1731–1741 (1995). A re-examination of proximodistal patterning by Gremlin. Mol. Cell. 4, 839–849 (1999). 13 Brown R, Brockes JP: Identification and during vertebrate limb development. Nature 37 Mercader, N, Leonardo E, Azpiazu N et al.: expression of a regeneration-specific 418, 539–544 (2002). Conserved regulation of proximodistal limb homeobox gene in the newt limb blastema. nn Provides evidence to challenge the Progress axis development by Meis1/Hth. Nature 402, Development 111, 489–496 (1991). Zone Model. 425–429 (1999).

future science group www.futuremedicine.com 461 Review Mariani

38 Lewandoski M, Mackem S: Limb 50 Yokoyama H, Ide H, Tamura K: FGF‑10 62 Han M, Yang X, Lee J, Allan CH, development: the rise and fall of retinoic acid. stimulates limb regeneration ability in Muneoka K: Development and regeneration Curr. Biol. 19, R558–R561 (2009). Xenopus laevis. Dev. Biol. 233, 72–79 (2001). of the neonatal digit tip in mice. Dev. Biol. 39 Niederreither K, Dolle P: Retinoic acid in 51 Mullen LM, Bryant SV, Torok MA, 315, 125–135 (2008). development: towards an integrated view. Blumberg B, Gardiner DM: Nerve nn Excellent modern introduction to the Nat. Rev. Genet. 9, 541–553 (2008). dependency of regeneration: the role of digit-tip regeneration model in mice. distal-less and FGF signaling in amphibian 40 Zhao X, Sirbu IO, Mic FA et al.: Retinoic acid 63 Douglas BS: Conservative management of limb regeneration. Development 122, promotes limb induction through effects on guillotine amputation of the finger in 3487–3497 (1996). body axis extension but is unnecessary for limb children. Aust. Paediatr. J. 8, 86–89 patterning. Curr. Biol. 19, 1050–1057 (2009). 52 Slack JM, Beck CW, Gargioli C, Christen B: (1972). Cellular and molecular mechanisms of 41 Gibert Y, Gajewski A, Meyer A, 64 Illingworth CM: Trapped fingers and regeneration in Xenopus. Philos. Trans. R. Soc. Begemann G: Induction and prepatterning of amputated finger tips in children. J. Pediatr. Lond. B Biol. Sci. 359, 745–751 (2004). the zebrafish pectoral fin bud requires axial Surg. 9, 853–858 (1974). retinoic acid signaling. Development 133, 53 Iten LE, Bryant SV: The interaction between 65 Vidal P, Dickson MG: Regeneration of the 2649–2659 (2006). the blastema and stump in the establishment distal phalanx. A case report. J. Hand Surg. of the anterior–posterior and proximal–distal 42 Barna M, Niswander L: Visualization of Br. 18, 230–233 (1993). cartilage formation: insight into cellular organization of the limb regenerate. Dev. Biol. 66 Brockes JP: Amphibian limb regeneration: properties of skeletal progenitors and 44, 119–147 (1975). rebuilding a complex structure. Science 276, chondrodysplasia syndromes. Dev. Cell. 12, 54 Stocum DL: Regulation after proximal or 81–87 (1997). 931–941 (2007). distal transposition of limb regeneration 67 Neufeld DA: Bone healing after amputation 43 Maden M: Vitamin A and pattern formation blastemas and determination of the proximal of mouse digits and newt limbs: implications in the regenerating limb. Nature 295, boundary of the regenerate. Dev. Biol. 45, for induced regeneration in mammals. 672–675 (1982). 112–136 (1975). Anat. Rec. 211, 156–165 (1985). 44 Niazi IA, Saxena S: Abnormal hind limb 55 Dent JN: Limb regeneration in larvae and 68 Borgens RB: Mice regrow the tips of their regeneration in tadpoles of the toad, Bufo metamorphosing individuals of the South foretoes. Science 217, 747–750 (1982). andersoni, exposed to excess vitamin A. African clawed toad. J. Morphol. 110, 61–77 Folia Biol. (Krakow) 26, 3–8 (1978). (1962). 69 Reginelli AD, Wang YQ, Sassoon D, Muneoka K: Digit tip regeneration correlates 45 Thoms SD, Stocum DL: Retinoic acid- 56 Shimizu-Nishikawa K, Takahashi J, with regions of Msx1 (Hox 7) expression in induced pattern duplication in regenerating Nishikawa A: Intercalary and supernumerary fetal and newborn mice. Development 121, urodele limbs. Dev. Biol. 103, 319–328 regeneration in the limbs of the frog, Xenopus 1065–1076 (1995). (1984). laevis. Dev. Dyn. 227, 563–572 (2003). 70 Neufeld DA, Zhao W: Bone regrowth after 46 da Silva SM, Gates PB, Brockes JP: The newt 57 Haack H, Gruss P: The establishment of digit tip amputation in mice is equivalent in ortholog of CD59 is implicated in murine Hox‑1 expression domains during adults and neonates. Wound Repair Regen. proximodistal identity during amphibian patterning of the limb. Dev. Biol. 157, 3, 461–466 (1995). limb regeneration. Dev. Cell. 3, 547–555 410–422 (1993). (2002). 58 Yokouchi Y, Sasaki H, Kuroiwa A: Homeobox 71 Han M, Yang X, Farrington JE, Muneoka K: Digit regeneration is regulated by Msx1 and 47 Echeverri K, Tanaka EM: Proximodistal gene expression correlated with the BMP4 in fetal mice. Development 130, patterning during limb regeneration. bifurcation process of limb cartilage 5123–5132 (2003). Dev. Biol. 279, 391–401 (2005). development. Nature 353, 443–445 (1991). 59 Beck CW, Izpisua Belmonte JC, Christen B: 72 Yu L, Han M, Yan M, Lee EC, Lee J, n Includes an excellent discussion of proximal Beyond early development: Xenopus as an Muneoka K: BMP signaling induces digit to distal models during limb regeneration. emerging model for the study of regenerative regeneration in neonatal mice. Development 48 Mercader N, Selleri L, Criado LM et al.: mechanisms. Dev. Dyn. 238, 1226–1248 137, 551–559 (2010). Ectopic Meis1 expression in the mouse limb (2009). 73 Monaghan JR, Epp LG, Putta S et al.: bud alters P–D patterning in a Pbx1- 60 Yin VP, Poss KD: New regulators of Microarray and cDNA sequence ana­lysis of independent manner. Int. J. Dev. Biol. 53, vertebrate appendage regeneration. transcription during nerve-dependent limb 1483–1494 (2009). Curr. Opin. Genet. Dev. 18, 381–386 (2008). regeneration. BMC Biol. 7, 1 (2009). 49 Boilly B, Cavanaugh KP, Thomas D, 61 Muller TL, Ngo-Muller V, Reginelli A, 74 Pearl EJ, Barker D, Day RC, Beck CW: Hondermarck H, Bryant SV, Bradshaw RA: Taylor G, Anderson R, Muneoka K: Identification of genes associated Acidic fibroblast growth factor is present in Regeneration in higher vertebrates: limb buds with regenerative success of Xenopus laevis regenerating limb blastemas of axolotls and and digit tips. Semin. Cell Dev. Biol. 10, hindlimbs. BMC Dev. Biol. 8, 66 binds specifically to blastema tissues. 405–413 (1999). (2008). Dev. Biol. 145, 302–310 (1991).

462 Regen. Med. (2010) 5(3) future science group