Survey of the differences between regenerative and non-regenerative animals


Author to whom all correspondence should be addressed.


To investigate the boundaries between regenerative and non-regenerative animals, we first survey regenerative ability across animal phyla from sponges to chordates (including mammals). There are both regenerative and non-regenerative animals in each phylum. The cells participating in regeneration also vary among different species. Thus, it is hard to find clear rules concerning regeneration ability across the animal kingdom, suggesting that it is not useful to compare the difference of regenerative ability across phyla to seek the boundary between regenerative and non-regenerative animals. Instead, if we carefully compare the differences of regenerative ability between closely related species within each phylum and accumulate these differences at the cellular molecular levels, we may be able to clarify the boundary between regenerative and non-regenerative animals. Here we introduce our comparative analysis of cellular events after amputation of lower jaws between frogs and newts. Then we propose that such comparative analyses using closely related species within the same phylum should be accumulated to understand the boundary between regenerative and non-regenerative animals in order to apply this understanding for realizing regenerative medicine in the future.


Regeneration phenomena have fascinated many people for a long time. A lot of biologists have investigated the regeneration capacity of a variety of animals from simple multicellular organisms to complex mammals. However, it is very hard to find a boundary between regenerative and non-regenerative organisms (Fig. 1). Although both planarians and Caenorhabditis elegans have simple body structures, planarians show high regenerative ability, but C. elegans cannot regenerate their body parts at all (Sulston & Horvitz 1977). Among vertebrates, amphibians can regenerate limbs, but other vertebrates cannot (Tanaka 2003; Tamura et al. 2010). Even among amphibians, only some urodels can regenerate limbs: frogs and adult salamanders cannot. We have not been able to find a clear rule concerning regeneration ability. This is the one of the reasons why regenerative biology has kept a mysterious atmosphere even in the 21st century, when we have succeeded in the production of induced pluripotent stem (iPS) cells and regenerative medicine has become a rapidly advancing science. Here we attempt to investigate the boundaries between regenerative animals and non-regenerative animals across animal phyla, and then to consider how to induce regeneration in non-regenerative animals.

Figure 1.

 Phylogenic tree of regenerative animals and non-regenerative animals across the metazoans. Open circles indicate regenerative animals that can regenerate whole body or structure such as limb, tentacles in adult. Black circles indicate non-regenerative animals. Gray circles indicate the partially regenerative animals that can regenerate some internal organs such as lens, or tissues but not structures in adults, or structures only in cretin period such as embryo, or larval stage. Even in the same phylum, class, or order, no common rules in regenerative ability are found.

Phylogenetic implications about regenerative ability

Sponges and cnidarians

It is well known that sponges can reconstitute their bodies after dissociation. Wilson clearly demonstrated the power of multicellular systems in which the cells of various organisms have self-organizing capacity (Wilson 1910). Although Wilson’s experiments showed an important aspect of regeneration at the cellular level, it is somewhat controversial whether such reconstitution of the body can be categorized as regeneration or not. In his experiments we do not know to what extent stem cells were involved in the reconstitution process. Recent studies using desmosponges clearly indicated the existence of pluripotent stem cells in sponges (Funayama 2010; Funayama et al. 2010). Can sponges reconstitute their bodies from only differentiated cells without stem cells? Or are stem cells required for reconstitution of the body with differentiated cells, just like in regeneration? We also do not know whether hexactinellid sponges can regenerate their bodies.

Among cnidarians, many hydra and jellyfishes have strong regenerative ability. Hydra is one of the model animals for studying regeneration (Wolpert et al. 1971; Gierer & Meinhardt 1972; Bode et al. 1976; Fujisawa 2003; Bosch 2007). Hydra regeneration was found by a French scientist, Trembley, in the 18th century (Trembley 1744). The body of hydra is composed of three different cell lineages: endodermal and ectodermal cells, and interstitial cell-derived cells. Each lineage has its own stem cells (Bode et al. 1976; David & Murphy 1977; Fujisawa et al. 1986; Bosch & David 1987). Cell type conversion into different lineages has not yet been observed (Galliot et al. 2006; Bosch 2008). Interstitial cells are known to include pluripotent stem cells, differentiating neurons, nematocytes and germ cells. However, a mutant hydra lacking interstitial cells survives, and is called epithelial hydra (Bode et al. 1976; Achermann & Sugiyama 1985). Hydra is known to show morphallactic regeneration (Holstein et al. 2003; Bosch 2007; Chera et al. 2009). Hydra regeneration has provided the idea of the positional information and reorganization of positional information during regeneration (Wolpert 1971). Recently, Nematostella has become a new model animal for evodevo research in cnidarians (Darling et al. 2005; Technau et al. 2005). In Nematostella, the foot (or physa) part of the body shows strong regenerative ability (Franka & Bleakney 1976; Hand & Uhlinger 1992, 1995; Darling et al. 2005; Reitzel et al. 2007), but other parts do not. Thus, it is difficult to say that evolutionarily primitive animals show especially high regeneration capability. Some jellyfishes show strong regenerative ability. Turritopsis nutricula regenerates the entire body structure from the part of stolon after transformation to a medusa (Schmidt and David 1986). It is known that the jellyfish Podocoryne carnea can regenerate from a part of the bell in which muscle cells can transdifferentiate into gametes and neuronal cells (Schmid & Alder 1984). However, the most common jellyfish, Aurelia aurita, shows lower regenerative ability, although jellyfishes generally have similar body structures. Thus, the regeneration abilities vary among different species even in cnidarians.


Generally, ecdysozoa, including arthropods and nematodes, are known as weakly regenerative animals, although some of them regenerate appendages. However, studies of their regeneration have contributed a lot to understanding the mechanisms of regeneration. Leg regeneration of cockroaches provided the idea of “intercalary regeneration”, which is a landmark model in the field of biology (French et al. 1976; French 1978; Bryant and Iten 1976). Insect legs have positional values along the distal-proximal axis or according to a polar coordinate system, and when a distal part is joined to a proximal part, they start to regenerate the intermediate region to fill up the gaps of the missing positional values between these two parts. This model was proposed around the 1970s and fascinated many biologists (Sánchez-Alvarado 2000; Agata et al. 2003; Nye et al. 2003; Brockes & Kumar 2008; Bando et al. 2009; Bely & Nyberg 2010), but lost prominence because of the lack of molecular and cellular knowledge.

Drosophila do not regenerate legs or wings at all, but their imaginal discs can regenerate according to the polar coordinate model (which has now been revised to a boundary model) (Bryant 1975; Strub 1979; Meinhardt 1982; Campbell & Tomlinson 1995; Gibson & Schubiger 1999). Genetic studies to explain regeneration of the imaginal discs revealed molecules involved in intercalary regeneration. Interaction of dorsally expressed dpp and ventrally expressed wg, whose expression is induced by posteriorly expressed hh, can induce generation of the most distal part of the leg (called “distalization”) and then interaction of the newly induced distal parts and remaining proximal parts induces intercalary regeneration to restore the intermediate region (Campbell & Tomlinson 1995; Marsh & Theisen 1999; Gibson & Schubiger 1999; Agata et al. 2007; Smith-Bolton et al. 2009).

Arthropods sometimes show “heteromorphic regeneration” during appendage regeneration. For example, stick insects regenerate a leg after amputation of an antenna. Crustaceans sometimes regenerate antennae after their arms are removed (Maginnis 2006). These phenomena were a mystery for a long time, but now they can be explained as a kind of homeotic transformation of segments of arthropods. However, most regeneration studies using arthropods are lacking a cellular level approach. Thus, we do not know the extent to which differentiated and stem cells contribute to regeneration of these animals.


In contrast to the Eczysozoa, Lophotrochozoa includes many highly regenerative animals, such as flatworms (planarians in the phylum Platyhelminthes) (Agata & Watanabe 1999; Newmark & Sánchez-Alvarado 2002; Saló & Baguñà 2002) and earthworms (oligochaeta in the phylum Annelida) (Christensen 1964; Yoshida-Noro & Tochinai 2010). Stem cells participating in the regeneration of these regenerative animals have been identified.

Planarians show outstanding regenerative ability. Amazingly, most small fragments can regenerate the entire body structure within one or a few weeks. Such high regenerative ability is supported by adult pluripotent stem cells, called neoblasts, which comprise 20% of the cell population (Hayashi et al. 2010; Shibata et al. 2010a). Thomas Hunt Morgan reported complete regeneration from a 1/279th piece of the body (Morgan 1898). He also proposed several new concepts about 100 years ago, such as “body polarity” and “molecular gradient”, from the studies of planarian regeneration (Morgan 1904). These concepts have now become very important for understanding the morphogenesis of animal bodies, including vertebrate embryogenesis. Also, recent cellular level studies using planarians have provided a lot of unique information about regulation of the pluripotent stem cells (Reddien et al. 2005; Yoshida-Kashikawa et al. 2007; Higuchi et al. 2007; Rossi et al. 2008; Rouhana et al. 2010; Wagner et al. 2011). However, some planarians cannot regenerate a head from the tail pieces. Among planarians there is a deep gap between regenerative and non-regenerative animals (Kawakatsu & Iwaki 1967; Teshirogi & Ishida 1979).

It is well known that octopus (a cephalopod in the phylum Mollusca) can regenerate arms (Norman & Finn 2001), sometimes accompanied by bifurcation. When under attack, some octopuses can perform arm autotomy, in a similar manner to how lizards detach their tails. The default number of their arms is eight. However, in Japan a 96-arm octopus was reported (Okada, 1935; pers. comm. from Shima Marineland Aquarium).

Recently, small earthworms, Enchytraeidae, have become new model animals for regenerative research (Yoshida-Noro & Tochinai 2010). They have segmental structures like arthropods, but can regenerate the entire body, including a brain, from small segmental pieces (Yoshida-Noro & Tochinai, 2010). In contrast, the regenerative ability of ordinary earthworms is limited. In general, the head fragment can regenerate the posterior part of the body, but the head part is not regenerated from the posterior fragment. It is also known that ribbon worms have regenerative capability (Newth 1958), but this ability has not been extensively studied.


Among deuterostomes, regeneration of starfishes and feather star is well known. Their arms can be regenerated after amputation or autotomy (Achituv & Sher 1991; Alves et al. 2002; Rubilar et al. 2005; Shibata et al. 2010b). However, the molecular and cellular mechanisms have not yet been well studied.

Sea cucumbers possess unique regeneration ability (Garcia-Arraras et al. 1998). They can regenerate intestinal ducts after their complete removal. We do not know the precise biological meaning of this phenomenon. However, sea cucumbers may show unique features of regeneration.


In comparison with the above invertebrate animals, most vertebrates show lower regenerative ability. However, newts show exceptionally high regenerative ability among the vertebrates. They can regenerate not only limbs, tail and jaw (Maden 2008; Kurosaka et al. 2008), but also a part of their heart, eyes and brain (Reyer 1954; Oberpriller & Oberpriller 1974; Mitashov 1996; Okamoto et al. 2007). The larvae of salamanders and neotenous axolotl show high regenerative ability similar to that of adult newts (Kragl et al. 2009). We do not know why they lose regenerative ability after metamorphosis. In the case of frog, regeneration of the limbs of Xenopus has been extensively studied. The larva of Xenopus can regenerate the complete structures of limbs (Dent 1962; Muneoka et al. 1986), but after metamorphosis froglets regenerate only a spike-like structure without muscle regeneration or limb pattern formation (Dent 1962; Endo et al. 2000; Satoh et al. 2005; Yakushiji et al. 2007).

In mammals, neonatal and adult mice can only regenerate the distal part of the finger when amputated from the first joint (Borgens 1982; Muneoka et al. 2008). However, they cannot regenerate the distal part beyond the first joint. Although they retain regeneration capability of major tissues composing limbs, such as skin, muscle and bone (Borgens 1982; Neufeld & Zhao 1995). Thus, the regeneration capacity of mammalian organs is strongly limited in spite of maintaining tissue repair ability.

There are two unique examples of regeneration in mammals. It is known that deer antlers can rapidly regenerate after amputation (Li et al. 2009). The process of antler regeneration has been well studied and the findings suggest that the perichondriummay have an important role in regeneration. The other example is very unique. An immunologically defective mutant mouse showed regeneration of the ear after hole-punching. Usually we distinguish each individual mouse used for experiments by punching patterns in their ears. However, amazingly, a certain kind of immunologically defective mutant mouse regenerated the hole in the ears after punching (Heber-Katz & Gourevitch 2009). This mutant mouse shows defects of scar formation, suggesting that scarring may interfere with regeneration by blocking initiation signal(s) for regeneration.

In conclusion, it seems that the ability of tissue repair using stem cell systems has been maintained during evolution. However, to regenerate organs and appendages, not only stem cells but also the systems for morphogenesis may be required. We discuss this point later.

Recently, for elucidating molecular mechanisms underlying vertebrate regeneration, medaka and zebrafish have become new model animals for studying regeneration, although they do not have strong regenerative ability. They can regenerate fins. Thus, fin regeneration provides information for elucidating molecular mechanisms underlying limb regeneration, since the limbs of four-footed animals evolved from fins of fish (Akimenko et al. 2003; Ishida et al. 2010). Now studies of heart regeneration using zebrafish have become a hot topic. It was shown using transgenic zebrafish that cardiomyocytes dedifferentiate and proliferate after resection injury without scar formation and regenerate electrically coupled cardiac muscle (Kikuchi et al. 2010; Jopling et al. 2010). These findings have important implications for promoting regeneration of the injured human heart in the future.

Regeneration of eyes and brains

Eye regeneration

Some jellyfish have eyes in the stump of each tentacle. One Swiss scientist reported regeneration of these eyes (Stierwald et al. 2004). Common precursor cells derived from ectodermal cells may differentiate both pigment cells and visual neurons in the process of regeneration (Nordström et al. 2003). The planarian Dugesia japonica can regenerate functional eyes within 5 days. After head amputation, two spots of miniature eyes are detected 2 days after amputation. At day 3, they connect with each other, forming a chiasma. Visual axons start to project to the brain at day 4 and negative phototactic behavior is completely restored at day 5 (Inoue et al. 2004; Yamamoto & Agata 2011). It has already been shown that sine oculis (Six) families of genes are required for the eye regeneration process (Pineda & Saló 2002).

Among vertebrates, newts show strong regenerative ability of eyes. Newts can regenerate not only the lens but also the neural retina from pigmented epithelial cells (PECs) of the iris and retina, respectively (Eguchi 1988; Araki 2007). Neural tube-derived PECs retain multipotent differentiating potentiality. Thus, when we remove the anterior half of the eyes, the remaining part of the eyes containing PECs can regenerate the lost tissues and structures. Interestingly, such regeneration of the neural retina from PECs can be induced in chick embryos when fibroblast growth factor (FGF)-containing beads are placed in the ocular space after removing the neural retina (Park & Hollenberg 1989). Regeneration of a lens-like structure from PECs was also reported in the early mouse embryo of a transgenic mouse in which the lens was destroyed using a toxic gene driven by a lens-specific promoter (Breitman et al. 1989). Also, under in vitro cell culture conditions including FGFs and other growth factors, PECs of many vertebrates, including adult humans, show highly proliferative activity and multipotent differentiating ability (Okada 1980; Tsonis et al. 2001), although lens regeneration has never been observed in humans after lens removal for cataract therapy. In some frogs and the fish Misgurnus anguillicandatus (Cobitidae), lens regeneration is observed from the cornea (Sato 1961; Freeman 1963; Bosco 1988; Mizuno et al. 1999; Henry 2003).

Brain regeneration in vertebrates

Brains are the most complex tissues in the mammalian body. Moreover, it was believed for a long time that brain neurons could not regenerate at all after early development. The brain seems to be the most difficult tissue to repair. However, it was shown in 1965 that a fraction of brain neurons maintain proliferative activity and contribute to the renewal of brain neurons (Altman & Das 1965). This finding led to the brain becoming one of the targets of regenerative medicine. Now transplantation of dopaminergic neurons derived from embryonic stem (ES) or iPS cells is being extensively tested for future therapy (Lee et al. 2000; Kawasaki et al. 2000; Wernig et al. 2008).

However, in newts, it has been demonstrated that dopaminergic neurons are spontaneously regenerated from neural stem cells without transplantation of dopaminergic neurons from the outside (Parish et al. 2007; Berg et al. 2011). BrdU incorporation experiments clearly indicated that these cells are derived from neural stem cells after proliferation (Parish et al. 2007). Newts can regenerate a part of the brain using adult neural stem cells after partial brainectomy (Okamoto et al. 2007). It will be very important to understand why newt brain neurons can be regenerated and mammalian neurons cannot.

Brain regeneration in planarians

Planarians regenerate the brain from pluripotent stem cells in the course of regeneration of the anterior body. Thus, planarians may provide information about how we can regenerate a functional brain from pluripotent stem cells. We do not have any idea how to construct a 3D-structured brain from ES or iPS cells. Here we would like to summarize our knowledge obtained from planarian regeneration about how planarians regenerate the 3D brain from pluripotent cells.

The planarian Dugesia japonica has an inverted-U shaped brain with a pair of eyes in the anterior portion of the body (Agata et al. 1998). After amputation, an anterior blastema is formed, and then a brain rudiment is formed from pluripotent stem cells in the anterior blastema within 24 h after amputation. In this process, FGF receptor (FGFR)/Nou-darake-mediated signaling is involved in formation of the brain rudiment (Cebriàet al. 2002; Agata & Umesono 2008; Umesono & Agata 2009). After rudiment formation, a Wnt family gene is activated in the posterior portion of the rudiment to induce anterior-posterior (AP) patterning in the brain (Kobayashi et al. 2007). Three Otx/otd-related homeobox genes are also activated in the rudiment to induce mediolateral patterning of the brain (Umesono et al. 1997, 1999). Both visual neurons and visual center neurons develop in the DjotxA-positive region. Lateral branches, which are composed of chemosensory neurons, are formed from the Djotp-positive region. Inverted-U shaped main lobes are formed in the DjotxB-positive region, where interneurons develop. After pattern formation, neural networks start to be connected about 3 days after amputation. Interestingly, several genes are activated after completion of the neural connections in order to restore the function of the brain (Inoue et al. 2004; Yamamoto & Agata 2011).

These findings clearly indicate the importance of stepwise instruction for brain formation from stem cells. Recently, it has been shown that X-ray-insensitive differentiated cells may have a critical function for such stepwise instruction (Yazawa et al. 2009; Hayashi et al. 2011). Therefore, to investigate the differences of animals’ ability of brain regeneration, we should analyze not only the potential of stem cells but also the capability of the surrounding cells.

Comparative analyses between regenerative and non-regenerative animals among close species within the same phylum

Finally, we would like to briefly introduce our recent study to investigate the difference of the regenerative ability of jaws between newts and frogs, since the difference has many implications (Kurosaka et al. 2008). Newts can regenerate both their upper and lower jaws, but frogs cannot. Therefore, we carefully investigated the difference of cellular and molecular events after amputation of the lower jaw between newts and frogs.

Unexpectedly, not only newts but also frogs show active proliferation in the remaining tissues after amputation. We could not detect a significant difference of proliferative ability of cells between these regenerative and non-regenerative animals. However, we found a clear difference of gene activation in the remaining tissues. In newts, strong reactivation of muscle- and cartilage-specific genes was observed in the remaining myotubes and cartilage, respectively, but such reactivation could not be seen in frogs. We also could not detect many Pax7-positive muscle satellite cells in adult frogs’ jaws. These results suggested that both activation of remaining tissues and stem cells are required for proper regeneration (Kurosaka et al. 2008). In newts, after amputation, the remaining tissues may respond to the injury signals and start activation of second signal(s), which not only induce reactivation of tissue-specific genes but also recruit stem cells to the injured tissues (Fig. 2A). In contrast, both reactivation of muscle-specific genes and stem cells are lacking in frogs’ muscle tissues after amputation (Fig. 2B).

Figure 2.

 Schematic drawings of differences between newt and frog. In newts, after amputation, the remaining tissues may respond to the injury signals and start activation of second signal(s). Reactivation of tissue-specific genes and recruitment of stem cells to the injured tissues are required for proper regeneration (A). In frog, both reactivation of muscle-specific genes and stem cells are lacking (B).

Also in the case of limb regeneration, both reactivation of remaining tissues and recruitment of stem cells were observed. A stem cell-specific nucleolar protein, nucleostemin, was rapidly accumulated in injured myotubes in the stump region after amputation (Maki et al. 2007). Recent cell lineage tracing experiments using transgenic axolotl clearly showed participation of stem/precursor cells during limb regeneration (Kragl et al. 2009). In Xenopus froglet limbs, although muscle regeneration was induced by transplantation of hepatocyte growth factor (HGF)-releasing cell aggregates, the induced muscle-like tissue had random locations and disorganized directions, and its connection with bone was not seen (Satoh et al. 2005). These results suggest that activation of remaining tissues may be required for recruiting stem cells into the proper positions to restore original structure and function. Thus, we should consider not only transplantation of stem cells but also activation of remaining tissues for future regenerative therapy.

Concluding remarks

In order to understand the difference between regenerative and non-regenerative animals, here we surveyed regenerative ability across animal phyla from sponges to mammals. In general many people believe that simpler animals may have stronger regenerative ability. However, we can’t find such clear boundaries between regenerative and non-regenerative animals across animal phyla (Fig. 1). Instead, there are both regenerative and non-regenerative animals in each phylum. Even among planarians, we find highly regenerative and non-regenerative planarians. The cells participating in regeneration also vary among different species. Thus, it is hard to find a clear rule concerning regeneration ability across the animal kingdom. However, by carefully comparing regenerative ability between close species within each phylum, we can accumulate information about the differences between regenerative and non-regenerative animals. Here we have introduced our pioneer studies in which we compared both the cellular and the molecular events during jaw regeneration between newts and frogs. The results suggest that we need to consider how to activate the remaining tissues to recruit transplanted stem cells into the remaining tissues to restore the original structures and functions (Kurosaka et al. 2008).

Other groups also reported a difference of epigenetic status in the enhancer region of the Shh gene between larva and froglet limbs in Xenopus, suggesting that epigenetic status might be involved in differences of regenerative ability (Yakushiji et al. 2007). Recently, Brockes’ group identified a taxon-specific gene involved in limb regeneration in salamanders, and hypothesized that regeneration depends on the activity of taxon-specific components in orchestrating cellular machinery that is extensively conserved between regenerating and non-regenerating taxa (Garza-Garcia et al. 2010). We believe that this kind of comparative analysis using closely related species and modernize techniques will be important for realizing the promise of regenerative medicine in the near future.


We thank Dr Elizabeth Nakajima for her careful reading of our manuscript. This review was written with the support of a Grant-in-Aid for Global COE Program A06, a Grant-in-Aid for Creative Scientific Research, and a Grant-in-Aid for Scientific Research on Innovative Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan.