Amputation in most vertebrates induces a fundamental process of wound closure and tissue repair in which fibrosis predominates. Complete regeneration of a missing appendage occurs rarely in mammals but readily in various species of teleost fish, urodele amphibians (newts, salamanders), and larval anurans (frogs, toads). In this type of regeneration, cell proliferation produces a population of undifferentiated cells from which the new structure develops, by a process termed “epimorphic regeneration” by Morgan (1901). In urodele limbs, the best studied model, epimorphosis involves cellular dedifferentiation and reprogramming in the injured tissues, formation from these cells of a mesenchymal blastema, followed by proximodistal, anterioposterior, and dorsoventral patterning of new limb structures from these cells. Blastema cell proliferation sufficient for limb regeneration requires mitogenic signaling from the distal wound epidermis which closes the amputation surface and trophic factors released from regenerating nerve axons that pervade the blastema (Mescher, 1996; Stocum and Cameron, 2011).
Extensive research with new models for mammalian tissue repair has in recent years revealed a complex, dynamic interplay among the resident cells at the injury and the various white blood cells recruited to the damaged site. These interactions underlie the sequential events of inflammation, proliferation of new progenitor cells, and tissue differentiation required for repair (Gurtner et al., 2008; Eming et al., 2009). This work clearly indicates that exaggerated and prolonged local inflammation is a key feature of both chronic, nonhealing wounds and excessive scarring (Eming et al., 2009). Conversely, epimorphic limb regeneration in urodeles includes a brief period of inflammation overtly similar to that of mammalian wounds (Schmidt, 1968), but this is followed by blastema formation and tissue patterning that yields a complete, fully functional new organ, with no scarring whatsoever (Harty et al., 2003). Early evidence suggesting that regenerative capacity declines in both mammals and amphibians as the immune system develops has been reviewed previously (Mescher and Neff, 2005).
Work reported here focuses on limb regeneration in the anuran Xenopus laevis. In addition to its long-standing status as a highly useful model for embryological research, Xenopus has also played a uniquely valuable role in comparative and developmental studies of the immune system, immune tolerance, and tumor immunity (Robert and Cohen, 2011) and for investigations of regenerative mechanisms (Beck et al., 2009). We will first review work on gene expression in amputated Xenopus limbs at regenerating and nonregenerating developmental stages which revealed the inverse correlation between inflammation and regenerative capacity. This will be followed by a brief review of features of the non-mammalian immune system that bear on the role of inflammation in the balance between fibrosis and regeneration. We then present results from initial tests of the hypothesis that factors modulating the inflammatory response to injury are one determinant of the quality of regeneration, comparing the effects of anti-inflammatory agents and the proinflammatory adjuvant beryllium.
GENE EXPRESSION STUDIES IN REGENERATING LARVAL XENOPUS HINDLIMBS SUGGEST A CHANGING BALANCE BETWEEN INFLAMMATION AND BLASTEMA FORMATION
Premetamorphic anuran amphibians regenerate developing hindlimbs well, but in most species lose this ability to various extents during prometamorphosis. In Xenopus laevis, hindlimb amputation at late prometamorphic and postmetamorphic stages gives rise to a spike of skin-covered cartilage that lacks skeletal patterning and most other tissues. The well-defined regeneration-complete and -incomplete stages of hindlimb development (Dent, 1962) and the increasingly more detailed genomic information available for Xenopus (Hellsten et al., 2010) prompted investigations of the genes expressed locally after limb amputation at different developmental stages for insights at the molecular level into the different responses to amputation at different developmental stages.
Our initial studies used subtractive cDNA hybridization to enrich for genes differentially expressed between regeneration-complete (stage 53) limbs and regeneration-incomplete (stage 59) 7 days after amputation (King et al., 2003). Subtractions were done using RNA extracted from stage 53 blastemas and stage 59 “pseudoblastemas” (Komala, 1957) and several hundred clones were identified in which expression was more than twofold higher in the blastemas or the pseudoblastemas (or both) than in the unamputated limbs. These differentially expressed sequences included many genes encoding transcription factors and components of translation and signaling pathways, as well as increased numbers of retrotransposon-related sequences and repetitive elements (King et al., 2003). Comparing gene expression in Xenopus hindlimb blastemas and pseudoblastemas was found to be a useful method to probe important events underlying the capacity for regeneration and its decline during limb development.
One gene found to be upregulated in blastemas by the subtractive screens is the Xenopus homolog of Sall4, a member of the evolutionarily conserved Spalt/sal gene family encoding zinc finger transcription factors. This gene was chosen for further study because various Sal-like genes in other vertebrates have important roles in limb development and are involved in clinically significant congenital limb anomalies. Localized expression of Sall4 in developing Xenopus hindlimbs was found throughout the mesenchyme of the early limb bud, with smaller mesenchymal fields of expression that were temporally and regionally dynamic during limb patterning and digit formation, in addition to expression in regeneration blastemas and delayed expression in pseudoblastemas (Neff et al., 2005). Sall4 is now known to be a positive regulator of differentiated cell reprogramming toward an embryonic stem cell-like state and is likely to play a central role in the cellular dedifferentiation and reprogramming in the initial phase of epimorphic limb regeneration (Neff et al., 2011).
We also analyzed gene expression in regenerating Xenopus hindlimbs using gene array screening with the commercially available Xenopus Genechip (Grow et al., 2006). This more comprehensive screen of gene expression during the early post-amputation period at regeneration-complete and -incomplete limb stages used tissues collected at both 1 and 5 days after amputation. Results of this analysis included not only expected differences in various gene expression profiles, such as those for limb patterning but also revealed differential expression of newly identified immune-related genes which allowed us to more completely characterize molecular and cellular events of tissue/organ regeneration. Another report of a Genechip® analysis of genes up-regulated early in regeneration-competent Xenopus limbs identified other such immune-related genes and emphasized activation of the stress response during regeneration (Pearl et al., 2008). The striking observation that numerous immune response genes are differentially expressed between regeneration-complete and -incomplete tissues led us to look further at how local inflammatory activity may affect hindlimb regenerative capacity in developing anurans (Harty et al., 2003; Mescher and Neff, 2005, 2006; Mescher et al., 2007).
Although the influence of immune cells in epimorphic regeneration has been considered previously (Schmidt, 1968; Sicard, 1985), recent high-throughput analyses of gene expression in many regenerating systems suggest a central role for local immunity and inflammation in determining the balance between repair of an injury and regeneration of a histologically complete, fully functional organ. Many immune functions have been implicated, from innate immune components such as the complement system to the regulatory T cells of adaptive immunity, but detailed examinations of immunoregulatory gene expression and leukocyte activities are still lacking in the context of regeneration. The array data of Grow et al. (2006) clearly showed that many genes previously identified as important for the immune response were indeed differentially expressed during regeneration-complete and -incomplete stages of limb development. Together with other data from urodele limb (Del Rio-Tsonis et al., 1998) and Xenopus tail regeneration (Tazaki et al., 2005), the results of Grow et al. (2006) suggested that regulation of innate and adaptive immunity may be important in establishing a local tissue environment allowing for epimorphic regeneration (Mescher and Neff, 2005, 2006).
Grow et al. (2006) also found differential expression of several genes involved in patterning and growth-related responses, including Shh, Tbx3, FGF8, and Msx2. Our array data demonstrated that each of these genes is expressed in regeneration-complete limbs and blastemas, with little or no expression in regeneration-incomplete limbs or pseudoblastemas. Given that amputation of a stage 57 Xenopus hindlimb produces only a pattern-deficient spike of cartilage (Dent, 1962), the observed pattern of expression of these genes is not surprising.
PROTEOMIC ANALYSIS OF XENOPUS LARVAL LIMBS FURTHER SUGGEST THE IMPORTANCE OF REGULATING INFLAMMATION DURING REGENERATION
Whereas the analysis of differences in gene expression profiles can be useful in identifying important pathways involved in tissue regeneration, it is also important to determine if there are equivalent changes in the levels of proteins synthesized from the expressed mRNAs. A global proteomic analysis of unfractionated tissue from stage 53 Xenopus hindlimbs at the time of amputation and 3 days later allowed us to identify and quantify over 1,500 mostly abundant peptides (King et al., 2009). Of significance to the concept of immune modulation of regenerative capacity was the abundant and differential expression of proteins involved with innate immunity or inflammation in the regenerating limbs, a finding consistent with the global analysis of gene expression (Grow et al., 2006). Similar proteomic analyses of axolotl limb blastemas (Rao et al., 2009), coupled with large-scale gene expression studies (Monaghan et al., 2009) and systems biology approaches to examine the networks of transcription factor and other protein interactions (Jhamb et al., 2011) in the same system, are expected to provide a firm basis for eventually understanding epimorphic regeneration at the molecular level. For example, the activation of transcription factors by transforming growth factor-β (TGF-β) during the wound healing phase of axolotl limb regeneration found by bioinformatic analyses (Jhamb et al., 2011) is shown by specific inhibitors of TGF-β signaling to be required for cell proliferation and limb blastema formation (Levesque et al., 2007).
Among the proteins that showed a substantial increase 3 days after amputation were members of the annexin family, particularly ANXA1 and ANXA2 and the annexin binding partner S100A10 (King et al., 2009). Annexins bind phospholipids in a Ca2+-dependent manner and thereby help regulate many cellular activities. The 37 kDa protein annexin 1, also known as lipocortin-1, blocks production of the lipid mediators of inflammation, leukotrienes and prostaglandins, by inhibiting activity of phospholipase A2, an enzyme early in the synthetic pathway. In this way, ANXA1 exerts a potent anti-inflammatory effect locally that helps ensure the transient nature of the inflammatory reaction and prevent chronic inflammation with its potential for tissue damage (Perretti and Flower, 2004). The important roles of ANXA1 and ANXA2 as highly localized anti-inflammatory factors make these proteins prime candidates not only in regulating local activities in the amputated limb but also in determining whether inflammation is short-lived or prolonged.
ANXA2 expression is up-regulated during both tail (Tazaki et al., 2005) and limb (Grow et al., 2006) regeneration in larval Xenopus. Quantitative PCR analysis showed that ANXA1, ANXA2, and S100A10 are all increased following amputation in limbs of both stages 53 and 57 larvae. However, whereas the expression of all three remains high in the stage 57 pseudoblastemas up to five days post-amputation, in the regeneration-complete stage limbs their expression returns to baseline levels by five days after amputation (King et al., 2009). These results suggest that inflammation is prolonged in the pseudoblastemas of regeneration-incomplete limbs, but is resolved more quickly in the limb stumps of stage 53 larvae which regenerate almost completely.
Our proteomics work, in conjunction with the analyses of gene expression during early regeneration, reveals that both proinflammatory and anti-inflammatory activities are upregulated in amputated amphibian limb stumps and suggest that the balance between such signals may be important in determining the quality of limb regeneration at different larval stages. Continuing work on the role of inflammation during the early phase of regeneration will yield additional insights into the molecular basis of dedifferentiation and other aspects of epimorphic regeneration.
STUDIES OF REGENERATION IN OTHER NON-MAMMALIAN VERTEBRATES REINFORCE THE RELATIONSHIP BETWEEN REGENERATIVE CAPACITY AND IMMUNITY
Comparative vertebrate immunology has long been centered on the Xenopus model and a broad overview of recent immunological studies using Xenopus has been published (Robert and Cohen, 2011). Aspects of amphibian immunity relevant to their capacity for epimorphic regeneration have also been discussed previously (Harty et al., 2003; Mescher and Neff, 2005; Godwin and Brockes, 2006). Of key relevance are the profound immunological changes that occur as tadpoles metamorphose into adults, which corresponds to the period in which hindlimb regenerative capacity is gradually lost (Flajnik et al., 1987). In effect, two developmentally distinct immune systems exist in Xenopus. The larval immunity reflects characteristics of an ancestral immune system that functions without classical MHC class I antigen presentation or efficient effector mechanisms (Robert and Cohen, 1998). After metamorphic climax, however, the immune surveillance system appears more highly evolved in that it is remarkably similar to that of mammals (Robert and Cohen, 1998).
The changes in the larval immune system during prometamorphosis greatly affect immune tolerance in anurans, producing histoincompatibility between larval and adult organs. Using a highly inbred strain of Xenopus, Izutsu (2009) found that if skin is grafted from the tail or flank of stage 54 tadpoles (shortly after the stage when hindlimb regeneration begins to decline) to syngeneic adults the grafts are quickly rejected. The larval immune system easily tolerates new antigens, but the adult frog generates strong immune responses comparable to those of the mammalian immune system against antigens of larval skin. During metamorphosis, anuran skin undergoes a complete morphological transformation from a larval type to the adult type. Only tail tissue behaves differently; it remains a larval-type tissue until it disappears at the end of metamorphosis (Izutsu, 2009). These studies provide strong evidence that the emerging adult immune system contributes not only to defense but also to tissue remodeling processes and cell elimination during metamorphosis (Mukaigasa et al., 2009).
During prometamorphosis (stages 53–57 in Xenopus) the larval skin becomes heavily populated with cells having features of monocyte-derived dermal dendritic cells (DCs) and epidermal Langerhans cells (LCs; Du Pasquier and Flajnik, 1990; Mescher et al., 2007). Cutaneous antigen-presenting cells (DCs and LCs) are abundant after metamorphosis in Rana (Carrillo-Farga et al., 1990; Castell-Rodriguez et al., 1999). Putative dendritic epidermal T cells were also reported (Mescher et al., 2007). As all these immune cells have been implicated as regulators of mammalian tissue repair, it is of interest to examine anuran skin wound healing before and after metamorphosis. Full-thickness skin excisions from the hindlimbs or backs of postmetamorphic Xenopus froglets are reported to heal completely without scarring (Yokoyama et al., 2011). A study of healing in similar dorsal skin wounds in larval and adult Rana found that during the prometamorphic period the repair process shifts from scarless, regenerative healing to wound closure by contraction and formation of dense, fibrous subepidermal tissue similar to mammalian scar tissue (Yannas et al., 1996). This change resembles that occurring in mammalian skin during the transition from embryonic to fetal development. Yokoyama et al. (2011) suggest that the Xenopus froglet's capacity for scarless wound healing is related to their ability to regenerate partial and non-patterned limbs, an ability lacking after metamorphosis in Rana species.
Manipulation of immune cell formation has become an important tool in studies of the molecular events of inflammation and wound healing. The PU.1 gene in mammals codes for a transcription factor that promotes myelopoiesis and differentiation of monocytes into macrophages, DCs, and LCs (Heinz et al., 2006). PU.1 null mice lack monocyte-derived cells but as neonates heal excisional skin wounds on the same schedule as wild-type wounds (Martin et al., 2003). Moreover, compared to controls PU.1 null mice heal such wounds with minimal inflammation and in a scar-free manner resembling that of fetal wound healing (Martin et al., 2003). Not only are the debris-removal and cytokine functions of macrophages and related cells superfluous for efficient wound healing under these experimental conditions, but the absence of these cells attenuates inflammation and reduces scarring. This report and more recent studies using other genetically immunodeficient mouse models have produced a paradigm shift in understanding the role of leukocytes and resident immune cells in tissue repair and regeneration.
IMMUNOMODULATORY GLUCOCORTICOID SIGNALING INHIBITS TAIL REGENERATION IN ZEBRAFISH
Down-regulation of PU.1 has also provided important insights into the relationship between inflammation and epimorphic regeneration in non-mammalian systems. Mathew et al. (2007) pioneered a high-throughput method for screening small molecule probes that specifically modulate pathways required for epimorphic regeneration by assessing caudal fin regeneration in larval zebrafish treated and maintained for 3 days in 96-well plates. Of 2,000 small bioactive molecules tested, none were reported to stimulate tail regeneration but 17 (∼0.8% of the library) specifically inhibited the process. These represented different chemical classes but the largest cluster of positive hits (five compounds) were glucocorticoids, which consistently and specifically blocked regeneration without inhibiting normal larval growth. The effect of these compounds occurred at nanomolar concentrations and was found to activate expression of primary glucocorticoid receptor (GR) target genes. Transient knockdown of GR with antisense morpholinos produced no obvious developmental defects, did not inhibit tail regeneration, and restored regenerative capacity in the presence of beclomethasone, a prototype GR agonist (Mathew et al., 2007). Thus, the GR seems not to be required for regeneration of larval fish caudal fins, but inappropriate activation of the GR blocks that process.
Tail regeneration in zebrafish larvae is complete within 72 hr after amputation and glucocorticoids affect events during the initial wound healing/blastema phase of the process. Beclomethasone (1 μm) inhibited regeneration if present during only the first 4 hr after amputation; if such exposure was delayed until 4 hr post-amputation, larval tails nevertheless regenerated completely. Regeneration is slower in tails of adult zebrafish, but cell dedifferentiation, migration, and proliferation in the injured tissues begin within 24 hr of amputation (Sousa et al., 2011). Mathew et al. (2007) found that beclomethasone exposure throughout this period was required to inhibit tail regeneration in adults, indicating developmental differences in the window of glucocorticoid sensitivity during regeneration. To analyze further the effect of GR signaling on regeneration, expression of dlx5a in the wound epidermis and msxe and junbl in distal mesenchymal cells were examined by in situ hybridization. Expression of all three genes failed completely in the presence of beclomethasone (Mathew et al., 2007).
Reasoning that the inhibitory effect of glucocorticoids on caudal fin regeneration was likely due to their well-studied, potent inhibitory effects on the acute inflammatory response, Mathew et al. (2007) examined whether beclomethasone affected immigration of leukocytes to the site of injury. Only slight reductions in the numbers of macrophages and neutrophils arriving at the amputation site were found after such treatment. Moreover, larvae developing without most myeloid cells after embryonic injection of antisense morpholinos targeting PU.1 nevertheless regenerated tails indistinguishably from control larvae. Collectively, these results illustrate clearly that although inflammatory events mediated by myeloid-derived cells are not required to initiate tail regeneration in zebrafish, other events inhibited by glucocorticoid signaling are needed for epimorphic regeneration in this system (Mathew et al., 2007).
Commentary by regeneration biologists on this report has focused exclusively on the PU.1 knockdown experiment showing that larvae without normal numbers of monocyte-derived cells nevertheless regenerated tails normally (Brockes and Kumar, 2008; Eming et al., 2009; Neff et al., 2011). The finding that the absence of myeloid cells has no effect on tail regeneration in zebrafish is important, but not unexpected in light of the earlier report that wound healing in neonatal PU.1 defective mice was actually enhanced to scar-free skin regeneration (Martin et al., 2003). The major finding of Mathew et al. (2007) is that GR signaling does inhibit critical events that are required for blastema formation, such as expression of the developmentally important genes dlx5a, msxe, and junbl. The study did not address the possibility that excessive activities triggered by cytokines from macrophages or DCs may attenuate or eliminate patterning events underlying regeneration.
INFLAMMATION AND DEVELOPMENT OF THE IMMUNE SYSTEM ARE DIRECTLY IMPLICATED IN THE CAPACITY FOR XENOPUS TAIL REGENERATION
In Xenopus, the larval tail normally fails to regenerate for a relatively brief period (stages 45–47) at the beginning of premetamorphosis, when tadpoles begin to feed and hindlimb buds appear (Beck et al., 2003). Although the cause of this “refractory period” in anuran tail regeneration remains unknown, it has been shown to correlate with a drop in bone morphogenetic protein (BMP) levels (Beck et al., 2006), a higher rate of apoptosis (Tseng et al., 2007), and a failure of cells in the wound epithelium to become polarized normally (Adams et al., 2007). Fukazawa et al. (2009) compared gene expression profiles in tail stumps during the refractory and post-refractory periods and found significant differences with immune and inflammation-related genes. Post-amputation expression of three such genes by leukocytes, RIN3, cathelicidin-like, and CXCR2, failed to occur during the refractory period, while genes for MHCIIa, one CXC chemokine, and three CC chemokines were upregulated by amputation during the refractory period but not at earlier or later periods (Fukazawa et al., 2009). These results suggest that differences in the inflammatory responses in tail tissues amputated during the refractory period versus the pre- or post-refractory period may produce the temporary failure to regenerate.
Fukazawa et al. (2009) tested the hypothesis that conditions similar to those of chronic inflammation were responsible for the impaired tail regeneration. Groups of tadpoles with tails amputated during the refractory period were exposed to four different immunosuppressants: celastrol, an inhibitor-κB kinase (IKK) inhibitor; an unrelated IKK; and cyclosporine A and FK506, two different and unrelated immunosuppressants. Through multiple experiments, all four agents were found to restore normal tail regeneration. Similarly, embryos injected with antisense morpholinos against PU.1, which significantly reduced expression of the pan-leukocyte marker CD45, showed higher regenerative abilities during the refractory period than did control larvae (Fukazawa et al., 2009). These results clearly raise the possibility that activities of myeloid cells whose formation requires PU.1 interfere with and impair the regenerative capacity of larval Xenopus tails during the refractory period.
The authors suggest that tail regenerative capacity is lost as effector cells of the immune surveillance system appear peripherally around stage 45, shifting the balance of activities in the wound environment from regeneration to autoimmunity, and then returns with the later development of regulatory T cells, which suppress local immune activation and induce local tolerance of new antigens. Support for this hypothesis includes the observation that markers for cytotoxic T cells were first expressed at the onset of the regeneration refractory period and that strong (10-fold), transient induction of FOXP3, which identifies regulatory T cells, was seen in amputated tails during the post-refractory period (stage 53) but not during the refractory period (Fukazawa et al., 2009). As the authors note, this work not only provides direct evidence for a link between immune responses and regenerative ability in Xenopus tails but also provides insight into the molecular mechanisms underlying the early events of epimorphic regeneration.
Regenerated tails of urodeles and larval Xenopus lack certain anatomical features of the original structure (Slack et al., 2008). Recent work indicates that, like the loss of regenerative competence in developing hindlimbs, tail regeneration in Xenopus becomes increasing imperfect during prometamorphosis (Franchini and Bertolotti, in press). Compared to tails regenerating at stage 50, regeneration at stage 55/56 produced malformed structures and included delayed re-epithelialization of the amputation surface with formation of an atypical wound epithelium, more rapid angiogenesis of the wound, and a more intense and prolonged cellular inflammatory response (Franchini and Bertolotti, in press).
IMMUNE INTERACTIONS ARE ALSO IMPORTANT FOR LENS REGENERATION IN NEWTS
Finally, the immune system has been linked to the onset of lens regeneration from cells of the dorsal pupillary margin in newts (Godwin et al., 2010). Lentectomy induces these cells to express tissue factor, a transmembrane cell surface receptor for the serine protease factor VIIa, which upon binding activates thrombin to produce fibrin and triggers cytoplasmic signaling. Tissue factor activation of thrombin at the dorsal iris was shown to produce local fibrin deposition and to be required for lens regeneration. Godwin et al. (2010) suggest that leukocytes normally found associated with the fibrin on the dorsal iris act as a paracrine source of FGF2, the key driver of dorsal iris epithelial cell proliferation and Wnt signaling that produce the new crystalline lens (Hayashi et al., 2008).
Lens regeneration in newts can also be induced by injuring the lens or by injecting DCs from an injured eye into the anterior chamber of a control eye, both of which produce lens histolysis and then formation of a new lens (Kanao and Miyachi, 2006). Interestingly, this phenomenon was found not to occur in splenectomized animals, in eyes injected with immature DCs, or if immune signaling by nitric oxide was inhibited systemically. In line with the phenomenon of immune privilege in the mammalian eye (Streilein, 2003), trafficking of sensitized DCs to the spleen presumably allows the signaling required for amplification of cells that return to the eye, produce local immune tolerance, and allow subsequent lens regeneration (Godwin and Brockes, 2006; Kanao and Miyachi, 2006). The importance of DCs for regeneration may be unique to the formation of a new lens, since splenectomy does not inhibit limb regeneration in newts (Fini and Sicard, 1980).
MODULATION OF AMPUTATION-RELATED INFLAMMATORY EVENTS IS IMPORTANT FOR PATTERNING AND REGENERATION IN LARVAL XENOPUS HINDLIMBS
Work reviewed above indicates that local activities triggered by the injury–events suppressed by glucocorticoids–are required to successfully initiate regeneration. Moreover, at least some repair processes such as skin wound healing which typically involve scarring show improved organ regeneration if activities or numbers of macrophages and related myeloid leukocytes are reduced. Just as experimental biologists have long recognized the important role of the injury itself in initiating epimorphic regeneration (Tassava and Mescher, 1975), clinicians have appreciated the double-edged nature of inflammation in tissue repair (Henry and Garner, 2003). We have begun studies examining how tissue injury and inflammation affect limb regeneration in larval Xenopus.
As shown in Fig. 1, treatment with the glucocorticoid beclomethasone (1 μM) immediately after amputation at stage 53 inhibited regeneration in this system. Regeneration was blocked completely in over a third of the limbs, while vehicle-treated limbs regenerated as described by Dent (1962). As with the adult zebrafish caudal fin (Mathew et al., 2007), beclomethasone exposure for more than one day following amputation was required to block regeneration. Regeneration was inhibited less effectively if glucocorticoid treatment was delayed by 24 or 48 hr after amputation. Levels of endogenous glucocorticoids increase significantly in Xenopus during metamorphosis and have been implicated in the removal of tadpole lymphocytes at this time (Rollins-Smith et al., 1997). Genes for steroid biosynthesis were found to be up-regulated in amputated stage 52 Xenopus hindlimbs prevented from forming blastemas by over-expression of the BMP inhibitor Noggin (Pearl et al., 2008). The inhibitory effect of added beclomethasone on regeneration and patterning in stage 53 limbs is consistent with that study and results of several early investigations of corticosteroids and amphibian limb regeneration [reviewed by Schmidt (1968)].
Beginning at stage 53 amputation of the developing hindlimbs in Xenopus yields increasingly smaller and more pattern-deficient regenerates, until by stage 57 only a simple skin-covered spike of cartilage typically results (Dent, 1962). To examine the role of inflammation in this decline in regenerative capacity, we tested the effects of nonsteroidal anti-inflammatory agents, including inhibitors of cyclooxygenase-2 (COX-2). This enzyme and its product prostaglandin E2 are critical mediators of inflammation. COX-2 expression is immediately up-regulated in epidermal cells, leukocytes, and fibroblasts after skin injury and represents an important step in the transition from scarless to fibrotic healing in wounded fetal mouse skin (Wilgus et al., 2004). Inhibition of COX-2 in adult skin wounds has been shown to reduce several parameters of local inflammation and significantly improve skin regeneration, with reduced scarring (Wilgus et al., 2003). Consistent with these studies of inflammation in mammals, COX-2 was found by Pearl et al. (2008) to be strongly and specifically up-regulated in stage 52 Xenopus pseudoblastemas with regeneration blocked by BMP inhibition, compared to regenerates of stage-matched wild-type larvae.
The quality of regeneration in limbs amputated at stage 54/55, as determined by the completeness of the new limb, was improved by two inhibitors of COX-2 (Fig. 2a). Typically, only about half of such limbs regenerate three or more digits (Dent, 1962). With celecoxib treatment, 100% of the limbs regenerated to that extent and with diclofenac the percentage regenerating three or more digits was only slightly lower.
Similar regenerative improvement in stage 55 limbs was found with celastrol, a triterpenoid IKK inhibitor which reduces inflammation and auto-immunity and which Fukazawa et al. (2009) found to improve Xenopus tail regeneration during the refractory period. Treatment with this anti-inflammatory agent also enhanced regeneration and patterning in hindlimbs at this stage (Fig. 2b). Neither celastrol nor the COX-2 inhibitors affected the nearly complete failure of late prometamorphic stage 57 limbs to regenerate (data not shown) and one could speculate that in these almost fully developed limbs, in which dendritic antigen presenting cells are more abundant than at earlier stages (Mescher et al., 2007), the inflammatory response triggered by amputation cannot be adequately inhibited by these methods to allow regenerative events to proceed.
These data suggest that in larval Xenopus limbs, anti-inflammatory actions instigated by the GR can block events required to initiate regeneration and that inhibiting other inflammation-related activities can improve growth and patterning during epimorphic regeneration. We have explored this further by examining the effects on regeneration of enhanced inflammation at the amputation site of stage 53/54 hindlimbs. Several different immunological adjuvants were applied topically to the wounds immediately after amputation and the effects on regeneration and expression of proinflammatory genes such as IL-1β during the first week post-amputation were determined.
CHRONIC INFLAMMATION IN AMPUTATED LIMB STUMPS INHIBITS PATTERNING AND BLOCKS REGENERATION
Solutions of lipopolysaccharide, poly-AU, Freund's adjuvant, and nickel salts (10 mM) induced transient expression of IL-1β one day post-amputation, with no effect on subsequent regeneration, but solutions of beryllium salts (10 mM) produced prolonged expression of several proinflammatory genes, triggered intense local inflammation at the wound site, and completely inhibited limb regeneration (in preparation). Unlike the other agents tested, ions of the light metal beryllium are not neutralized or eliminated by phagocytic activity and produce chronic monocyte and lymphocyte infiltration and local immune reactivity, reactions that have been well-studied in the context of occupation-related chronic beryllium disease (Sawyer et al., 2002; Sawyer and Maier, 2011).
Thornton (1951) showed that topical application of Be2+ to larval salamander limbs immediately after amputation inhibits their normal rapid and complete regeneration. Although Be has cytotoxicity produced in part by its direct interactions with G proteins and protein kinases (Bigay et al., 1987), its major deleterious effects on vertebrate tissues are due to immune-mediated reactions. In larval urodeles, local Be treatment completely inhibits the increased inositol triphosphate production by cells of the injured tissue in the first minute after amputation (Tsonis et al., 1991), but the subsequent failure of blastemas to form is more likely due to the local inflammatory response and fibrosis which occur at the treated amputation site (Thornton, 1951). The Be concentration which inhibited Xenopus limb regeneration was 10-fold less than that required for the similar effect in urodele larvae; higher doses induced a systemic inflammatory response in the larvae and high levels of mortality (in preparation). The greater sensitivity of Xenopus larvae to Be exposure likely results from the immunological differences between anurans and urodeles.
The effect of Be exposure on expression of selected genes is shown in Fig. 3. Untreated limbs showed transient expression of the important proinflammatory gene IL-1β 6 hr post-amputation and expression of this gene was increased and prolonged by Be treatment. The treatment had little effect of expression of Sall4, a marker for cell dedifferentiation and reprogramming, but inhibited and delayed expression of the limb patterning gene shh. Similar effects are seen when expression of these and functionally similar genes is studied in limb stumps of regeneration-incompetent stage 57 limbs and compared with that in regeneration-competent stage 53 limbs. Post-amputation expression of inflammation-related genes is prolonged and/or higher in the older limbs, there is little difference in expression of reprogramming genes, and genes for limb patterning fail to be expressed normally in the older limbs (Grow et al., 2006; King et al., 2009; Neff et al., 2011). The failure of shh and possibly other blastema patterning genes to be expressed in regeneration-incomplete pseudoblastemas may involve epigenetic silencing of relevant sequences during prometamorphosis (Yakushiji et al., 2007; Tamura et al., 2010).
The results presented here, together with those of Mathew et al., (2007), Fukazawa et al. (2009), and other groups, indicate that an important determinant of non-mammalian epimorphic regeneration is the balance between factors promoting inflammation on the one hand and blastema formation on the other. As found for normal mammalian wound healing, immediate glucocorticoid-sensitive events that occur in the aftermath of amputation are required for normal regeneration of larval Xenopus hindlimbs. Conversely, chronic local inflammation produced by wound treatment with beryllium also inhibited limb regeneration, suggesting that epimorphic regeneration depends on proper local immunomodulation in the period following amputation. As was found during the regeneration-refractory phase with Xenopus tails (Fukazawa et al., 2009), treatment of larvae with different anti-inflammatory agents during the early phase of regenerative decline can enhance the capacity for regeneration. The data also show that prolonged local inflammatory activities after amputation have little effect on dedifferentiation and expression of genes that control cellular reprogramming, but interfere greatly with gene expression required for patterning the regenerate.
An important current question in the field of epimorphic regeneration is the extent to which dedifferentiating blastema cells resemble mammalian mesenchymal stem cells (MSCs). Amphibian limb blastema cells are being studied extensively with regard to the genes promoting dedifferentiation and reprogramming (Christen et al., 2010; Neff et al., 2011) and epigenetic regulation of important patterning genes such as shh (Tamura et al., 2010). At the same time, the potent multifaceted ability of mammalian MSCs to regulate the local immune response and inflammation provoked by tissue injury, transplantation, or autoimmunity is now recognized as one of these cells' most important properties (Caplan and Correa, 2011; Singer and Caplan, 2011; Weil et al., 2011). Whether similar regulatory interactions occur during epimorphic regeneration between dedifferentiating cells and the local immune cells remains unknown. If mesenchymal cells of the early blastema exert paracrine immunomodulatory, anti-scarring effects like those of MSCs, this would likely be of great importance in determining the success of the patterning events upon which the outcome of regeneration depends. The data from regenerating larval Xenopus tissues reviewed and presented here suggests that studies of the ontogenic immunological changes that occur in this non-mammalian model can be very useful in determining how local inflammation and its resolution affect the capacity for blastema formation.
The authors' research was approved by the Indiana University Animal Care and Use Committee and conducted according to the NIH Guidelines for the care of lab animals.