Stem cell system in tissue regeneration in fish


*Author to whom all correspondence should be addressed.


During evolution from single-cell to multi-cellular organisms, organisms developed the needed machinery by which a vast number of functionally different types of cells could be unified into an individual. To attain this goal, organisms evolved the developmental strategies that produced different cell types and unified them into complex body architecture. However, a more intriguing feature of multi-cellular organisms is that they can maintain their bodies throughout long life. For tissue maintenance, stem and/or progenitor cells in many tissues and organs are thought to play an important role; however, we know little about their control and the process of tissue reconstitution. As cells are fragile, all animals have the ability, more or less, to replace damaged or dead cells; however, there are large variations in such abilities, depending on the type of organs and the species. Though vertebrates cannot reconstitute their bodies from a small piece as do planarians, some lower vertebrates, unlike mammals, have the ability to regenerate body appendages and many internal organs. If we unveil the nature of stem cells in striking examples of such regeneration, this information can be applied to mammals and greatly benefit us. The focus in the present review is on the recent advances in our knowledge about the regeneration mechanism in fish, including the stem cells that are involved.


During the evolution from single-cell to multi-cellular organisms, animals attained longevity by developing the machinery by which a vast number of functionally different types of cells could be unified into an individual organism. To maintain the integrity of the multi-cellular body, animals needed to have specific sizes and morphologies, and to maintain their properties by a tissue maintenance mechanism.

The word regeneration is defined as the “reproduction or reconstitution of a lost or injured part” or as “a form of asexual reproduction.” With such definitions, regeneration encompasses a broad spectrum of natural phenomena that operate through seemingly different mechanisms. Regeneration may include phenomena such as physiological regeneration (renewal of blood and epithelial cells and seasonal replacement of deer antlers), morphallaxis (regeneration of hydra and some annelids), hypertrophy (compensatory or regenerative increase in the mass of internal organs), and reparative regeneration (cellular regeneration, tissue regeneration), and epimorphic regeneration mediated by the formation of a blastema (Fig. 1; Carlson 2007). However, in any of these tissue-maintenance phenomena, how organisms sense the loss or damage of tissues and control the proper tissue restoration has not been elucidated yet.

Figure 1.

 Major types of regenerative phenomena in multi-cellular organisms. Physiological regeneration includes seasonal or hormonal cycles of tissues such as deer antlers and replacement of blood and epithelial cells. Tissue regeneration is defined as the replacement of damaged tissues without the mediation of a blastema, whereas epimorphic regeneration refers to a type of regeneration mediated by the blastema. Hypertrophy is thought to be a type of regeneration, in which there is a compensatory increase in the size of a paired organ such as kidneys and lungs after its pair has been lost or damaged. The regenerative type refers to restoration of the mass of damaged internal organs such as liver and pancreas. Morphallaxis refers to reconstruction of form after severe damage by remodeling the body.

One of the difficulties in revealing the regeneration mechanism is due to our ignorance about the entity of positional information. When we look at the animal kingdom, there are many varieties of sizes even between close species. For example, the largest modern rodent, the capybara, is over 1 m in body length and 45 kg in weight (there is an even larger rodent in the fossil record) in contrast to the smallest one, the pygmy mouse, whose body length is only 5–6 cm. Similarly in primates, the smallest pygmy mouse lemur is about 6 cm in body length; whereas the largest gorilla is nearly 2 m tall and weighs over 200 kg. Considering the rapid and continuous turnover of cells within the body, it is surprising that animals are able to maintain their specific body sizes and morphologies during their long lives, which implies that the positional information exists in adult organisms and it keeps directing the size and shape according to a genetic program.

If we could figure out the entity of such information and elucidate how the regeneration mechanism surveys the tissue integrity, it will impact many fields of biology including regenerative medicine. Though trials that aim to reconstruct tissues by using the induced and/or natural stem cells are already ongoing, a mechanistic understanding of the regeneration process is necessary for making organs with specified size, morphology, and function. Particularly, elucidation of the mechanism of epimorphic regeneration, a natural process that completely restores the size, morphology, and function, will greatly benefit regenerative medicine and make it feasible. This review focuses on the stem and/or progenitor cell system of tissue regeneration, especially the epimorphic regeneration, in fish, and overviews the recent progress in our understanding.

Structure of the tail fin in fish

Technically, epimorphic regeneration has been defined as a form of regeneration that is initiated by the formation of a blastema, which is a mass of proliferating undifferentiated progenitor cells, followed by their differentiation to attain restoration of the tissue. This phenomenon has attracted the attention of biologists since the first description of limb regeneration of crayfish in the early 18th century by René-Antoine Ferchault de Réaumur (1712). Since that era, the fish fin has also been used for regeneration research (Broussonet 1786), which has now spanned more than 200 years.

The fish tail fin is an easily accessible tissue for examining the regeneration process due to its simple and radial symmetrical structure, rapid and robust regeneration, and accessory function for animal survival. Actually, fish are the champions of regeneration among vertebrates, since they possess a striking potential to regenerate not only fins, but also scales, retina, spinal cord, and many internal organs including the heart and pancreas. Regarding the tissue origin, the fish tail fin is an organ analogous to the tail of terrestrial vertebrates; but unlike the tail, it does not contain the extension of the notochord and muscles. Common to all types of fish fins, the main constituents of the tail fin are the fin rays, which are composed of an array of fin-ray bones that have a shape like a longitudinally split piece of bamboo (Fig. 2). These bones are radially arranged to form a fan-like frame and are covered with an epithelium that has a basal cell layer at its bottom. Within the concave fin-ray bones, the hollow spaces contain the actinotrichia, which is the bundles of collagen fibers, arteries and nerve axons that run down to the caudal end. The rest of the space within the fin is filled with non-characteristic mesenchymal or connective tissues.

Figure 2.

 Structure of adult fish fin. (a) Appearance of a caudal fin. (b) Magnification of boxed area in (a). The segmented fin-ray bones (arrowheads) are surrounded by mesenchymal cells and covered with thin epithelial cells. Irregularly distributed black spots are the chromatophores. (c) Schematic illustration of fin rays. A segment of fin-ray bone is composed of two hemi-rays. The pink line depicts the blood vessel.

Epimorphic regeneration in fish species

A number of classical studies have afforded an extensive description of the process of regeneration. According to those studies and our own observations, the process goes through the following steps: (i) wound closure; (ii) epithelial wound healing and formation of a wound epidermis; (iii) the formation of a blastema at the distal end of the mesenchyme; and (iv) proliferation of blastema cells, their differentiation, and tissue reconstruction (Fig. 3).

Figure 3.

 Epimorphic regeneration in adult fish. Process of fin regeneration (a) and actual tissue appearances (b). After the amputation of an adult caudal fin, the amputated plane is covered with new epithelial cells by 3 h. However, a tight epithelial sheet is not formed at 3 h. The wound epidermis is formed by 1 day. In the following stages, the blastema (red ovals) is formed, and its active cell proliferation and growth recover the original fin morphology in approximately 10 days. Irregularly distributed black and golden spots are the chromatophores.

After the amputation, the opening of the epidermis is in many cases quickly sealed by prompt contraction of epithelial tissue surrounding the amputation site, which is known to be mediated by the formation of the F-actin purse string around the wound and its rapid contraction (Martin & Lewis 1992). Such a rapid reaction of wounded tissue is commonly observed in flies and vertebrates (Martin et al. 1994; Wood et al. 2002), and it appears to be a mechanism conserved through evolution as a first tissue reaction against tissue trauma. Though the process of early wound closure is accompanied by neither cell proliferation nor migration, in the following step when the actin purse string disappears, the epithelial cells around the stump actively migrate to form the wound epidermis, a specialized epithelial thickening formed over the stump, which has a molecular identity different from the surrounding epidermal regions (Poss et al. 2000b; Kawakami et al. 2004).

It has been known for more than 100 years that the presence of the wound epidermis is an absolute requirement for the initiation of regeneration. In a well-designed study, Goss (1956) amputated the forelimbs of newts and inserted the distal epidermis-free ends into the body cavity; a procedure that keeps the apical surface epidermis free. These limbs did not regenerate. Thus, the apical wound epidermis plays a necessary role in regeneration; however, it remains unclear how the wound epidermis affects regeneration.

In the successive stages, the blastema proliferation provides an adequate number of cells to fulfill restoration of the lost tissue part; and further cell differentiation and tissue remodeling reconstitutes the original tissue architecture.

Origin of the blastema – Are blastema cells stem cells?

It is widely accepted that the presence of the wound epidermis and blastema is absolutely required for the initiation of regeneration and that the formation of the blastema and its active cell proliferation are the crucial and necessary driving forces for the progression of regeneration. Although the blastema cells seem to have the characteristics of undifferentiated cells, are blastema cells really stem cells? According to the definition, stem cells are cells that have the capacity to self-renew for an extended period of time and the ability to differentiate into a mature cell. In a broad sense, the blastema in fish regeneration nearly fulfills the criteria of stem cells, although little is known about the exact origin and fate of the blastema.

In terms of the origin of the blastema, there are two opposing hypotheses. One is that the blastema is a collection of activated cells derived from a population of dormant and widely distributed stem cells. The tissue regeneration by such a stem cell system can be seen in the regeneration of planarians, annelids and other invertebrates (see other reviews in this issue); however, an apparent example of stem cells in vertebrate tissue regeneration has not been proved. If this is true, the blastema cell is precisely a stem cell or a progeny derived from a stem cell.

Alternatively, another prevailing idea is that the blastema originates from cells around the wound site by de-differentiation. Cell lineage studies in urodele amphibians have suggested that this actually occurs in amputated tissues (Tanaka & Brockes 1998; Echeverri et al. 2001; Echeverri & Tanaka 2002). In such a case, the blastema cells should be called precursor cells as a transient cell population that is exhausted at the end of regeneration. As well as in urodeles, it has been observed that the mesenchymal cells in the amputated fish fin undergo disorganization and migration toward the amputation plane, where the blastema is formed (Poleo et al. 2001). In addition, the blastema cells express msx molecules, a family of transcription factors that may have a role for maintaining cells in an undifferentiated state (Akimenko et al. 1995; Odelberg et al. 2000). From these and other observations, it has been thought that the “de-differentiation” is a more favorable hypothesis, although a direct demonstration of the origin of the blastema in fish has not been made yet. Thus, we cannot completely rule out either of these possibilities before carrying out a detailed tracing of cell lineage. In the remainder of this review, I will tentatively use the term “stem cells” to refer to the blastema cells responsible for fish tissue regeneration.

Heterogeneity of cells in regenerating tissues

As well as the origin of cells for regeneration, the differentiation potential of the blastema has also not been clarified. A cell lineage analysis was recently carried out during limb regeneration in the axolotl, and it surprisingly showed that the de-differentiated cells derived from various types of cells such as dermis, muscle, epidermis, and Schwann cells produced the same respective type of cells after regeneration (Kragl et al. 2009). Though it has been postulated that the blastema is a sort of stem cell that can produce several cell types, the result in the axolotl suggests that the blastema itself is a group of heterogeneous cells that are “weakly” de-differentiated and re-enter the cell cycle.

In line with such a heterogeneity of regenerating axolotl tissue, a number of observations in fish have shown that the expression of sonic hedgehog (shh) and bone morphogenetic protein 2b (bmp2b) is localized in lateral domains of the basal layer of the wound epidermis (Laforest et al. 1998) and that lef1 has a similar overlapping expression in the basal wound epidermis (Poss et al. 2000a). Further, a study by Nechiporuk and Keating (2002) demonstrated that the distal most part of the blastema contains a group of slow cycling cells that are apparently segregated from the proximal blastema cells. Moreover, we recently demonstrated from the expression analysis of a number of regeneration-induced genes that the wound epidermis and blastema are composed of several cellular compartments with distinct molecular identities (Yoshinari et al. 2009; Fig. 4). According to our results, the wound epidermis contains at least four different compartments including the apical cells and several types of basal epidermal cells. The blastema is also divided into at least three types, the distal, proximal, and lateral populations, which appear to be the early osteoblasts (Yoshinari et al. 2009; Kawakami in press). Thus, the emerging data suggest that the regenerating tissue is a much more complex tissue than previously thought.

Figure 4.

 Cellular compartments in the blastema of a regenerating adult fin. The illustration schematically depicts a regenerating fin ray (approximately 2 days post amputation [dpa]). Gene expression analysis (Yoshinari et al. 2009) revealed that the regenerating tissue contains many cell types with different molecular identities. Respective cellular compartments are expressed with different colors.

However, there is no concrete study on fish that addresses the fate of each blastema cell. So, the fate mapping study will be one of the important ones to carry out. Though the cell lineage tracing in the axolotl was carried out with transplanted cells (Kragl et al. 2009), it can ideally be carried out by using genetic labeling. Recently, it was reported that inducible or tissue-specific in vivo recombination using the Cre-ER and lox system was successfully used for a long-term marking of specific cells in zebrafish (Hans et al. 2009). By using such genetic labeling, we will be able to trace the origin and fate of the cells participating in regeneration.

Stem cells in heart regeneration

Despite the histologically distinguishable morphology of the blastema in limb and fin regeneration, the blastema in other tissues has been poorly characterized. However, regeneration of other tissues in fish is thought to be mediated by the blastema or blastema-like cells. One such tissue is the heart, as the partially resected organ in fish can be completely restored without scar formation (Poss et al. 2002; Poss 2007). In mammals, however, the lesions in heart muscle caused by an infarction are never recovered to normal, but form fibrous scar tissues that may cause severe contractile dysfunction and lead to heart failure.

During the course of regeneration of the zebrafish heart, cardiac progenitor cells are successfully driven to regenerate by interaction with the epicardium, the thin epithelial layer enveloping the chambers. The epicardial cells close to the site of injury invade the regenerating tissue through a process reminiscent of the epithelial-to-mesenchymal transition (EMT) that occurs in the developing heart, and provide new vasculature to the regenerating muscle (Lepilina et al. 2006). In this process, epicardial and myocardial cross-talk is mediated by FGF signaling, and inhibiting the FGF receptor blocks cardiac regeneration (Lepilina et al. 2006). Not only in regeneration but also during homeostatic cardiac growth the epicardium regulates the addition of new myocardial and epicardial cells (Wills et al. 2008). Although dedifferentiation of pre-existing cardiomyocytes has also been postulated for the zebrafish heart, this possibility has not been supported by experimental evidence (Lepilina et al. 2006). Thus, it remains unclear whether newly formed cardiomyocytes originate from epicardium-derived cells or from de-differentiated progenitor cells after epicardium-mediated activation. However, in any case, a group of cells that undergo active mitosis and produce the mature cardiomyocytes, which can be called blastema-like cells, is formed around the site of injury.

In mammals, it has long been believed that the cardiac muscles never renew in their postnatal lives (Leu et al. 2001). However, a recent study has shown that the adult heart achieves a modest, but nonetheless convincing, self-renewal after injury that was attributed to a pool of resident stem cells (Hsieh et al. 2007). In addition to this, it has been suggested that the epicardium also has a crucial stimulatory role essential for proper development and regeneration, as in the zebrafish. One of the factors that influences this instructive role is thymosine β4 (Tβ4), a G-actin monomer-binding protein implicated in cytoskeleton reorganization. Tβ4 secreted from the developing myocardium stimulates the proliferation, differentiation, and inward migration of epicardial cells (Smart et al. 2007). Another potential source of cardiac stem and progenitor cells may be the vasculature. Vessel-associated progenitor cells, called mesoangioblasts, have recently been identified in the juvenile mouse heart (Galvez et al. 2008). Although it is controversial as to which stem cell source is responsible for the natural process of self-renewal in the heart (Ausoni & Sartore 2009), it is intriguing to compare the heart stem cell systems in different species with different regeneration potentials.

Stem cell system in scale regeneration

The fish scale is also known as one of the regeneration-competent tissues (Sire 1989). Though the term “scale” often covers a variety of different structures, the elasmoid scale is the commonest type of scale within teleost species (Sire & Akimenko 2004). In medaka and zebrafish, the body is covered with several hundreds of large elasmoid scales, arranged in longitudinal and vertical rows, forming a regular pattern. The elasmoid scale, like the other elements of the dermal skeleton including the membrane bones, forms in the dermis without the presence of a cartilaginous initium (Zylberberg & Nicolas 1982).

Though hair and fish scales are distributed over the body surface in an orderly pattern, it has been thought that they are morphologically and evolutionally different (Sharpe 2001). However, recent studies have shown that the formation of these skin appendages is governed by the same signaling pathway including a transmembrane protein, ectodysplasin or EDA, and a TNF-like receptor, EDAR, that interacts with EDA (Kondo et al. 2001), indicating the existence of a common developmental and molecular mechanism for appendage formation involving epidermal–dermal interactions. In addition, the expression of several signaling molecules such as shh is shared in mammalian hair, bird feather, and fish scale (Iseki et al. 1996; Nohno et al. 1995; Sire & Akimenko 2004), supporting a molecular commonality among these skin appendages.

During the past decade, it has been shown that the hair bulge, an upper permanent region of the hair follicle below the sebaceous glands, is the residence of stem cells that can contribute to epidermal repair when a wound cannot spontaneously repair itself through the migration of epidermal cells from the neighboring unwounded epidermis (Ito et al. 2005). On the other hand, little is known about the stem cells within the epidermis and/or dermis in fish skin. It would be intriguing to characterize such cells in the fish skin and compare the underlying mechanism with that operating in mammals.

Control of cellular supply during regeneration

Regardless of the type of regeneration, either epimorphic regeneration, simple tissue regeneration or other types of tissue restoration, it is essential for organisms to supply the cells that replenish the damaged parts. Although we are not sure of the nature of blastema cells, cells that are not terminally differentiated and retain their proliferation activity are involved in the process and are the source of new cells that participate in the repair of tissues. The control of morphogenesis and tissue architecture is also an important issue in regeneration, but it acts later on. Thus, in a sense, regeneration can be defined as a process in which new cells are supplied in response to tissue loss or incompleteness of positional information.

The strategies of cell supply are different depending on the tissue types and/or extent of the damage. Small wounds can completely be repaired by cell division of surrounding epithelial cells and fibroblast cells, whereas larger tissue loss requires the consecutive occurrence of several distinct processes including epithelial wound healing, induced proliferation of stem cells such as the blastema, and production of new cells to form new body parts. Seemingly, re-epithelialization occurs in a similar way in all animals, and appears to be possibly universal machinery that serves as a primitive response of organisms to their body damage (Gurtner et al. 2008). In regeneration-competent species or tissues, the first step for complete regeneration is the formation or activation of stem cells such as the blastema. Otherwise, in many mammalian tissues, damaged tissues form a scar, a fibrous tissue that contains plenty of collagen matrices by way of the transient formation of granulation tissue. A leading hypothesis is that the immune system is involved in the switch between regeneration and fibrotic healing, because human fetuses, which heal without scarring, have immature immune systems (Mescher & Neff 2005). Although the significance of scarring and its relation to regeneration have not been elucidated, the mechanism of cell supply, in other words, the activation of the cell cycle and formation of proliferating cells, is one of the critical issues to be addressed in order to understand the process of tissue repair either in fish or in mammals.

Molecules and signals that are involved in regeneration

Then, how does the epithelial wound healing proceed to regeneration in fish? Further, why does such a regeneration system not operate in large wounds of mammalian tissues? To approach these issues, we need to know the molecular basis of regeneration-competent systems and compare it with that of incompetent ones. During the last decade, studies aimed at the molecular basis of regeneration were conducted and identified a number of molecules and signaling pathways involved in regeneration (reviewed by Poss et al. 2003; Iovine 2007; Nakatani et al. 2007; Stoick-Cooper et al. 2007). A number of signaling pathways such as Hedgehog (Hh), Wnt, Fibroblast growth factor (Fgf) and Activin βA signaling as well as many other molecules have been shown to be involved in regeneration. The details of these respective molecules and signaling pathways are outside the scope of this review. In spite of the increase in our knowledge about the molecular players involved, we still must figure out how these molecules are interconnected to lead to the respective processes of regeneration such as its initiation, blastema formation, cell proliferation, and morphogenesis. As the regenerating tissue has a complex cellular composition, we will need to examine the function of these respective molecules in the cellular compartments as well as the role of each compartment.

Use of larval zebrafish finfold as a model of tissue regeneration

To facilitate the analysis of regeneration at the molecular level, we previously proposed the use of infant tissue regeneration (Kawakami et al. 2004). The newly hatched zebrafish larva at 2 days postfertilization (dpf) has already completed the formation of most of its organs and tissues, and thus it serves as a useful system for analyzing the tissue repair process. According to a previous study, even such infant tissues respond to tissue trauma to form a wound epithelium with a molecular identity distinct from that of the surrounding epithelial cells and to cause apparent activation of the cell cycle in cells of the adjacent mesenchymal region, where the induction of msx-family genes is also observed. These processes are followed by a complete recovery of morphology within 3 days (Fig. 5). Thus, in light of the criteria of epimorphic regeneration such as the formation of specialized epithelial covering, the induction of genes that represent an undifferentiated state, and the activation of cell cycle, the larval finfold regeneration fulfils these criteria of epimorphic regeneration.

Figure 5.

 Regeneration of juvenile tissue. (a) Zebrafish larvae at 2 days post fertilization (dpf). The finfold can be used as a model of regeneration. (b) Temporal process of larval finfold regeneration. Hematoxylin staining. By 6 h, the wound epithelium (arrowhead) composed of tightly packed cells is formed. From 12 to 48 h post amputation (hpa), many cells with condensed chromatin, which seems to represent the actively dividing cells, are observed in areas adjacent to the stump (bracketed areas). (c) Schematic illustration of larval regeneration.

The larval finfold is mostly composed of mesenchymal cells covered with a thin epithelial sheet. At this stage, differentiated cells cannot be detected, except that the growing actinotrichia, bundles of collagen fibers, begin to radially elongate. Though the nerve fibers from the spinal cord have entered and become distributed all over the finfold, there is no detectable presence of blood vessels or their arbors by 5 dpf (not shown). Despite its functional homology with the adult caudal fin, the larval finfold is a transient tissue that is later replaced by the developing adult fin. The precursor cells that form the adult fin rays arise from the ventral side of posterior somites and/or notochord and migrate to form the actively proliferating fin ray progenitors, the process of which was shown to be dependent on Hh signaling (Hadzhiev et al. 2007). These fin rays precursors gradually shift to the dorsal side to form the adult caudal fin as if spreading a fan (Sakaguchi et al. 2006; Hadzhiev et al. 2007; Fig. 6). In view of such differences in origins and structures between the larval finfold and adult fin, the similarity in the repair processes between these tissues is striking and suggests a commonality of basic mechanisms that do not depend on the tissue types.

Figure 6.

Transition from larval fin fold to adult fin. (a) Bromodeoxyuridine (BrdU) labeling of zebrafish larva. A population of proliferating cells migrating from the ventral side of the notochord (arrowhead) is the precursor of the adult fin ray. (b,c) Schemes illustrating the change in migration pattern. Cells forming the fin-ray precursors migrating in the ventral direction gradually move to the caudal and then dorsal directions, as the caudal-most few segments of the notochord tilt toward the dorsal side.

Given the commonality with adult epimorphic regeneration, the model of larval finfold regeneration is amenable to the conventional molecular approaches established for the analysis of embryogenesis, such as morpholino antisense oligonucleotide (MO)-mediated gene knockdown. In particular, further merit is that an enormous number of larvae can be used for the assay due to the high fertility of zebrafish. In addition, the small size of newly hatched larvae in combination with their aquatic habitat is useful for carrying out a pharmacological test or screening in a multi-plate dish (Mathew et al. 2007; Yoshinari et al. 2009). Moreover, a useful but a slightly tricky application of this model is that we are able to use plenty of mutant resources for dissecting the regeneration process by a genetic means, because even the lethal mutations can be used for the assay as far as they do not affect finfold formation and survive up to 5 dpf in a good condition (Yoshinari et al. 2009). Though the larval tissue model may not have processes of apparent cell differentiation and complex organ formation, it is certain that it should have a basic regeneration process such as the formation of highly proliferating cells and control of cell supply. We will be able to use this model along with the adult tissue model to further dissect the molecular pathway underlying tissue maintenance and regeneration.

Initiation of regeneration and stem cell proliferation

Whereas the epithelial wound healing appears to be similar in many species and tissues, why does such a common reaction proceed to regeneration only in some species or organs? Still, we cannot completely exclude the possibility that the wound healing process itself, including the early immune reaction, differs between regeneration-competent and -non competent species or organs (Mescher & Neff 2005); however, no apparent early difference has been documented so far, including the immediate induction of Jun family transcription factors (Ishida et al., in preparation), which are suggested to have a crucial role in wound healing (Yates & Rayner 2002). Then, what is the determinant of regeneration? Is there a fish-specific regeneration protein? Furthermore, when does regeneration start?

Previously, we reported that the expression of two members of the junb family genes in zebrafish was induced during regeneration of the larval finfold and adult fin. It appears that the expression of these genes demarcates regeneration from wound healing. The immediate induction of Jun family genes, c-Jun and JunB, is soon attenuated in mammalian tissues; however, only the expression of two zebrafish junb genes, but not that of c-jun, is maintained until the regeneration stage (Ishida et al., in preparation). This maintained expression seems to be the earliest event that is specific to regeneration.

It has been shown in Drosophila that Jun and its upstream kinase, the Jun N-terminal kinase (JNK), are necessary for epithelial cell migration during wound closure (Jacinto et al. 2000). In mammals, the functions c-Jun and JunB for proper epithelial wound healing have also been suggested by studies using skin-specific conditional gene knockout mice (Grose 2003; Zenz et al. 2008). Generally, fish have two junb genes, and their expressions are seen in different cellular compartments during regeneration (Yoshinari et al. 2009). Thus, the original Junb function is thought to be divided into two independent Junb proteins in different cell groups. Intriguingly, we recently observed that two junb genes were immediately induced in the forming wound epidermis, but as soon as the initiation of blastema formation the junb-like (junbl) expression shifted to the blastema. In our recent study, we succeeded in showing that the functions of Junb proteins, particularly the function of Junbl, are indispensable for regeneration, Junbl is thought to be the earliest functional molecule necessary for regeneration (Ishida et al., in preparation).

To our further surprise, we also found that Junb family proteins were phosphorylated in response to the wounds (Ishida et al., in preparation). It has been shown in mammals that the consensus target sequence for JNK is not conserved in JunB and that this sequence in humans and mice is never phosphorylated (Kallunki et al. 1996). However, the fish Junb proteins partially retain the consensus sites and indeed are phosphorylated in vivo (Ishida et al., in preparation). However, curiously, in spite of the early role of JNK in the Drosophila epithelial healing, our analysis in zebrafish suggested that JNK signaling was not necessary for the epithelial wound healing, but was required for regeneration some time after wounding (Ishida et al., in preparation). From these observations, we propose that Junb family molecules and their JNK-dependent phosphorylation are an important determinant that specifies the initiation of the regeneration process (Fig. 7). Our data imply that the initiation of regeneration is determined some time after injury, where the wound healing itself has a minor role for determination, although it is still possible that the early phosphorylated Junb proteins might have some effect on the wound healing. In any case, it appears that regeneration proceeds in a stepwise way.

Figure 7.

 Function of Junb family molecules in fish regeneration and comparison with that of mammalian c-JUN and JUNB. (a) Function of c-JUN and JUNB in mammals. The active phosphorylated form of c-JUN has a strong positive role in cell-cycle progression, whereas JUNB lacks the phosphorylation site and acts antagonistically toward c-JUN. (b) Function of fish Junb proteins. The fish Junb and Junbl retain a phosphorylation site and are activated by the wound. These proteins are expressed in different cells and have distinct functions in the blastema and wound epidermis.

Maintenance of blastema cells

The induced blastema increases in size by cell division and at the same time its cells differentiate to produce a number of mature cells. Thus, the blastema stem cells possibly keep self-renewing at least during regeneration. Our recent study has suggested that an additional factor is required for the maintenance of blastema cells. We have identified a zebrafish mutation in which the larval blastema cells once formed undergo apoptosis instead of active cell proliferation (Fig. 8). The molecule responsible for this mutation has not been identified, but it is thought to be a key molecule for blastema maintenance. Another zebrafish mutant, pinball eye (piy), has the phenotype of retinal apoptosis (Yamaguchi et al. 2008), in which a missense mutation occurred in the small subunit of DNA primase (Prim1), a molecule essential for DNA replication; however, this mutation does not affect cell proliferation but rather induces neuronal apoptosis. The cellular maintenance mechanism may be different between the stem cells in regeneration and that of other precursor cells during retinal development, which is intriguing considering the mechanism for stem cell maintenance in regeneration.

Figure 8.

 A zebrafish mutant that cannot maintain the blastema. (a) Mutant phenotype during regeneration (left) and abnormal apoptosis in the blastema (right). Arrowhead indicates a characteristic abnormal tissue appearance, which is probably due to the blastema apoptosis; and the bracket indicates the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL)-positive apoptotic cells. (b) Wild-type sibling.


During the 1980s, little was known about the molecular background of regeneration, although there were some trials for identifying molecular players (Tabin 1989). Nowadays, many advances have been made in our knowledge about the molecules involved in regeneration as well as the nature of the blastema and its induction. However, many questions still remain to be answered. The following are some of them:

  • What is the first trigger for wound healing?

  • What regulates the regeneration-specific gene expression?

  • What is the origin and fate of cells participating in regeneration?

  • What are the roles of the cells in the respective cellular compartments?

  • What regulates the JNK activity?

  • How is the blastema maintained during regeneration?

  • How is the positional information integrated into regeneration?

By elucidating the answers to these questions along with clarification of the molecules involved and their networks at the cellular level, we should be able to obtain a complete picture of the epigenetic landscape of tissue maintenance.


I would like to thank lab members T. Nakajima and N. Yoshinari for their help in preparing the illustrations, providing pictures, and valuable discussions.