Evolution and regeneration of the planarian central nervous system


*Author to whom all correspondence should be addressed.
Email: umesono@mdb.biophys.kyoto-u.ac.jp


More than 100 years ago, early workers realized that planarians offer an excellent system for regeneration studies. Another unique aspect of planarians is that they occupy an interesting phylogenetic position with respect to the nervous system in that they possess an evolutionarily primitive brain structure and can regenerate a functional brain from almost any tiny body fragment. Recent molecular studies have revisited planarian regeneration and revealed key information about the cellular and molecular mechanisms underlying brain regeneration in planarians. One of our great advances was identification of a gene, nou-darake, which directs the formation of a proper extrinsic environment for pluripotent stem cells to differentiate into brain cells in the planarian Dugesia japonica. Our recent findings have provided mechanistic insights into stem cell biology and also evolutionary biology.


Freshwater planarians are believed to belong to an early group of organisms with defined bilateral symmetry, dorso-ventral polarity, a central nervous system (CNS) and a simple brain structure. One of the most notable characteristics of planarians is their high regenerative ability. They can regenerate whole animals, including a functional brain, from tiny fragments from almost any part of their bodies after amputation (Agata et al. 2007; Cebrià 2007; Agata and Umesono 2008). Thus, planarians were described early on as ‘. . . almost immortal under the edge of the knife’ (Dalyell 1814). The robust regenerative abilities of planarians are based on a population of pluripotent stem cells called neoblasts which are located throughout the body (Baguñàet al. 1989; Agata and Watanabe 1999; Newmark and Sánchez Alvarado 2002; Saló and Baguñà 2002; Agata 2003; Reddien and Sánchez Alvarado 2004; Agata et al. 2006; Saló 2006; Sánchez Alvarado 2006; Sánchez Alvarado 2007). Planarian regeneration involves a host of basic biological events similar to those observed during development, such as cell proliferation, cell differentiation, organogenesis and morphogenesis. In spite of the great interest of biologists in planarian regeneration, until the past decade, studies on the molecular mechanisms underlying planarian regeneration made very slow progress due to the many difficulties in applying molecular approaches. In 1997, we established a whole mount in situ hybridization (ISH) method for the planarian Dugesia japonica, which was a great advance in planarian studies (Umesono et al. 1997). In 1999, Sánchez Alvarado and Newmark developed a method for RNA interference (RNAi) in planarians, which provides a powerful tool to analyze gene function in planarians (Sánchez Alvarado and Newmark 1999). Continuing efforts by planarian workers have succeeded in identifying many genes expressed specifically in brain cells as well as in stem cells by combining EST (expressed sequence tag) projects and DNA chip analyses (Cebriàet al. 2002a,b,c; Mineta et al. 2003; Nakazawa et al. 2003; Rossi et al. 2007; Eisenhoffer et al. 2008).

In the present review, we will describe cellular and molecular dissection of the planarian brain. Our recent progress in studies of planarians has advanced our understanding of the phylogeny of the nervous system and the genetic mechanisms underlying brain regeneration.

Structure of the planarian central nervous system

The freshwater planarian Dugesia japonica is one of the most common planarians in Japan. Its body-length is generally up to 2 cm. The planarian has a variety of organs such as a central nervous system (CNS), a pharynx acting as a mouth and anus in the middle portion of the body and a gut throughout the body (Fig. 1). The planarian CNS is composed of two morphologically distinct structures: the anterior brain and the ventral nerve cords (VNCs; Fig. 1A). The brain is a bilobed structure that consists of a large cluster of neural cells located on the dorsal side relative to the VNCs and is independent of them (Fig. 1D; Agata et al. 1998). The bilobed brain structure is composed of a cortex of nerve cells and a core of axons (Fig. 1D), and can be morphologically divided into a central main lobe, which is a mass of interneurons, and lateral branches (Fig. 1C; Agata et al. 1998; Okamoto et al. 2005). About nine branches in each lobe elongate to the head margin to form sensory organs, such as auricles (Pigon 1974). A pair of eyes is located on the dorsal side relative to the brain. The eye is composed of only two cell types: pigment cells and photoreceptor cells, whose axons form an optic chiasma and project caudally onto the dorso-medial region of the central main lobe, where the photosensory signals are integrated (Fig. 1C; Agata et al. 1998; Sakai et al. 2000; Okamoto et al. 2005).

Figure 1.

The gross body structure of the planarian Dugesia japonica. (A) Anti-Synaptotagmin (SYT) antibody staining in green (Tazaki et al. 1999) from the dorsal view visualizes the axonal networks of the planarian central nervous system (CNS). The planarian CNS is composed of an inverted U-shaped brain in the head region and a pair of ventral nerve cords (VNCs) in the trunk region. (B) Double staining in magenta with probe for the muscle-specific DjMHC-A gene (Kobayashi et al. 1998) and anti-Arrestin antibody in the same animal shown in A. The expression of DjMHC-A in the muscle surrounding the intestinal duct visualizes the gross structure of the gut throughout the body. A strong signal of DjMHC-A is also observed in the pharynx in the middle portion of the body and in cells surrounding the pharyngeal opening on the ventral surface. The anti-Arrestin staining visualizes photoreceptor cells and their axons (arrows). (C) Higher magnification of the merged image in the head region (boxed in A). The left and right central main lobes of the brain each have about nine lateral branches (asterisks). The pair of eyes is composed of only two cell types: pigment cells (two black dots) and photoreceptor cells (magenta) associated with the pigment cells, whose axons form an optic chiasma and project caudally onto the dorso-medial region of the central main lobes (arrowheads). (D) Transverse view of the part of the head that includes the brain at the position indicated in A. The bilobed brain, which is composed of a cortex of nerve cells visualized with the strong Hoechst staining in blue and a core of axons visualized with the anti-SYT antibody staining in green, is located on the dorsal side relative to the VNCs (dotted circles). (E) Transverse view of a part outside the head region at the position indicated in A. Only the VNCs are observed (dotted circles). The Hoechst staining visualizes the nuclei of all cells in blue. Anterior is to the left in A and B. Anterior is to the top in C. Dorsal is to the top in D and E.

Molecular genetic analyses in Drosophila and vertebrates revealed that the regional identity of the anterior brain is controlled by the evolutionarily conserved homeobox genes Drosophila orthodenticle (otd) and empty spiracles (ems) and their vertebrate homolog genes Otx1/2 and Emx1/2 (Simeone et al. 1992; Hirth et al. 1995). We succeeded in identifying the three otd/Otx-related homeobox genes in the planarian, whose discrete expression patterns clearly define not only the brain as a structure genetically distinct from the VNCs, but also the structurally distinct domains of the planarian brain (Umesono et al. 1997, 1999). These findings revealed some similarities between the expression patterns of the otd/Otx family genes in planarians, Drosophila and vertebrates despite the great phylogenetic distances among them, strongly supporting the idea that the brains of deuterostomes and protostomes descended from a common ancestor (Reichert and Simeone 2001; Lichtneckert and Reichert 2005).

Cellular aspects of the planarian brain

Our extensive EST analysis indicated that the genome of D. japonica possesses hundreds of nervous system-related genes that are highly conserved with those of other animals, including human and mouse (Mineta et al. 2003). Interestingly, almost all of them are expressed with discrete patterns in the brain, and some of them are brain-specific (Cebriàet al. 2002a,b,c; Nakazawa et al. 2003). When we looked at the planarian brain at the level of single neurons, we found that the brain is composed of many distinct neuronal populations, such as dopaminergic, serotonergic, octopaminergic and GABAnergic neurons, which form distinct neuronal networks (Nishimura et al. 2007a,b, 2008a,b; Fig. 2). These findings strongly suggest that the basic machineries of the brain might have been acquired early in the course of the evolution of the CNS. In fact, planarians possess a variety of genes encoding evolutionarily conserved axon guidance molecules, such as Netrin, Slit and Robo, which are required for the formation of neural networks during regeneration (Cebrià and Newmark 2005; Okamoto et al. 2005; Cebrià and Newmark 2007; Cebriàet al. 2007). In contrast, little is known about the molecular aspects of glial cells in the planarian nervous system (Cebrià 2007). Glial cells are the other main type of neural cells generated from neural precursors that give rise to both glial cells and neurons in vertebrates and invertebrates (Turner & Cepko, 1987; Luskin et al. 1988; Udolph et al. 1993; Condron & Zinn 1994). Electron microscopic studies revealed glial-like cells in planarians (Morita and Best 1966). In Drosophila, glial cells missing (gcm) encodes an evolutionarily highly conserved transcription factor that functions as a binary switch to promote glial cell fate and simultaneously repress neuronal cell fate in the embryonic CNS as well as peripheral nervous system (Hosoya et al. 1995; Jones et al. 1995; Akiyama et al. 1996), and thus seems to be a good candidate marker for glial cells. We identified Djgcm, a gcm-related gene in D. japonica showing a high degree of conservation with Drosophila gcm. However, expression of Djgcm was not detected in the neural lineage in either intact or regenerating animals (Fig. 3). We also tried to identify neural committed stem cells in planarians by focusing on the expression of the planarian musashi gene family. Musashi encodes an evolutionarily well-conserved RNA binding protein known to be expressed in the neural lineage, including neural stem cells and progenitor cells, in the animal kingdom (Okano et al. 2002). Three musashi family genes (DjmlgA, DjmlgB and DjmlgC) were obtained in our planarian EST database (Higuchi et al. 2008). Contrary to our expectations, all of them were predominantly expressed in subsets of X-ray-resistant terminally differentiated neurons, but not in X-ray-sensitive proliferating cells (i.e. stem cells) (Higuchi et al. 2008b). Furthermore, neither single nor combinatorial knockdown experiments of these genes by RNAi affected the gross number of neurons during brain regeneration (Higuchi et al. 2008).

Figure 2.

An example of the cellular diversity in the planarian brain. (A) Distribution of octopaminergic neurons revealed by detecting DjTBH (Dugesia japonica tyramine β-hydroxylase)-immunopositive neurons (green). (B) Distribution of dopaminergic neurons revealed by detecting DjTH (D. japonica tyrosine hydroxylase)-immunopositive neurons (magenta). (C) Merged image. DjTBH and DjTH were not coexpressed in any brain neurons. All images are horizontal views of the whole brain (Nishimura et al. 2008a). Anterior is to the top. (Reprinted from Neurochemistry International, 53/6-8, Nishimura, K., Kitamura, Y., Inoue, T., Umesono, Y., Yoshimoto,K., Taniguchi, T. & Agata, K., Identification and Distribution of tryptophan hydroxylase (TPH)-positive neurons in theplanarian Dugesia japonica/Comparison of distribution of DjTBH-immunopositive neurons and DjTH-immunopositiveneurons in theplanarian CNS, Pages 190, Copyright (2008), with permission from Elsevier.)

Figure 3.

Expression pattern of a planarian gcm homolog gene (Djgcm). (A) Djgcm expression in a tail-regenerating head fragment. (B) Neural-specific expression of a planarian prohormone convertase 2 gene (DjPC2; Agata et al. 1998) in a tail-regenerating head fragment. (C) Djgcm expression in a head-regenerating tail fragment. (D) Neural-specific expression of DjPC2 in a head-regenerating tail fragment. Djgcm was expressed in subsets of mesenchymal cells outside the head region, and was not detected in either differentiated or differentiating brain cells (arrows), or in cells forming the VNCs. All samples were analyzed at 36 h after amputation. Anterior is to the left.

In conclusion, we do not yet have any information about glial cells or neural stem-like cells in planarians.

Functional properties of the planarian brain

Planarians can sense a variety of signals coming from the outside, such as light and chemicals, and display distinct behavioral traits depending on the type of signals. One of the typical behaviors is a specific light avoidance behavior, known as negative phototaxis, mediated via the eyes acting as a light-sensing organ (MacRae 1964; Carpenter et al. 1974; Asano et al. 1998). This experimentally tractable behavioral trait can be quantitatively measured in detail for parameters such as movement speed, distance and direction, and therefore is suitable for behavioral analysis to elucidate brain functions (Inoue et al. 2004). However, it still remains unknown how the neuronal activity is regulated in the planarian brain during negative phototaxis, in spite of the remarkable recent increase of available genetic tools. To elucidate the roles of genes in the brain circuitry, we established a negative phototaxis assay system (Inoue et al. 2004) and demonstrated that a gene (Djsnap-25) encoding a planarian synaptosome-associated protein of 25 kDa is indispensable for the brain function, by combined behavioral assays and loss-of-function (RNAi) analyses (Takano et al. 2007). Djsnap-25 is expressed exclusively in the nervous system, brain, and VNCs, as well as in putative sensory cells along the body periphery, but not in the photoreceptor cells (Fig. 4B). Within the brain, Djsnap-25 is expressed in the main lobes (Fig. 4B). To distinguish the specific role of Djsnap-25 in the brain circuitry from those in other regions such as the VNCs, we developed a unique RNAi technique named Readyknock, which is an easy and powerful approach for silencing gene activity in a body-position-dependent manner by taking advantage of the high regenerative ability of planarians (Takano et al. 2007). The principle of Readyknock is as follows: when the head region of Djsnap-25 RNAi-treated animals once regenerates, the effect of RNAi, especially at the protein level, seems to be stronger in cells newly differentiating from stem cells (which do not express any differentiation marker genes), than in the already-existing Djsnap-25-positive differentiated cells (Fig. 5). The protein stability of DjSNAP-25 explains the difference in the expression level of this protein between the newly differentiating cells and the terminally differentiated cells. After the resultant selective elimination of the DjSNAP-25 activity in the head region while leaving it intact in the trunk region, the Readyknock knockdown planarians behave like headless animals, showing uncoordinated and more random movement under the light condition, although they show no effect on head or brain morphology (Takano et al. 2007). These findings provided the first strong molecular evidence of a particular functional property of the planarian brain as an information-processing center.

Figure 4.

Whole-mount in situ hybridization views of neural-specific genes from the ventral side. (A) Pan-neural expression of the planarian synaptotagmin gene (Tazaki et al. 1999). (B) Neural-specific expression of Djsnap-25 (Takano et al. 2007). (C) Neural-specific expression of DjGAD (Nishimura et al. 2008b). Djnsnap-25 and DjGAD were expressed in a limited number of neurons in the brain. The expression of all of these genes in the head region was required for displaying negative phototaxis (Inoue et al. unpubl. data in A; Takano et al. 2007; Nishimura et al. 2008b). Anterior is to the left. (C is reprinted from Neuroscience, 153/4, Nishimura, K., Kitamura, Y., Umesono, Y., Takeuchi, K., Takata, K., Taniguchi, T. & Agata, K., Identification of glutamic acid decarboxylase gene and distribution of GABAergic nervous system in the planarian Dugesia Japonica/Expression pattern of DjGAD mRNA and protein, Pages 1108, Copyright (2008), with permission from Elsevier.)

Figure 5.

Readyknock experiments in planarian brain cells. (A) Immunohistochemical detection of DjSNAP-25 protein in brain cells of normal head-regenerated animals. (B) DjSNAP-25 protein in the ventral nerve cords (VNCs) of the same animal shown in A. (C) Effect of silencing of the DjSNAP-25 protein in brain cells of the Djsnap-25 knockdown head-regenerated animals. (D) DjSNAP-25 protein in the VNCs of the same animal shown in C. Severe reduction of DjSNAP-25 protein was observed in the newly regenerated brain cells. However, prominent expression of DjSNAP-25 was still detected in the already existing VNCs of the same animal. Yellow in A and C is a pseudo-color of the Hoechst staining for nuclei to visualize clusters of brain cells. The signals of DjSNAP-25 protein were detected in axons of the brain as well as the VNCs (Takano et al. 2007). Animals were injected with Djsnap-25 dsRNA and subsequently their head region was removed, and then 7 days were allowed for head regeneration. Anterior is to the top.

Recent studies on a gene encoding glutamic acid decarboxylase (GAD), a key enzyme that converts gulutamic acid into GABA, a major inhibitory neurotransmitter, provided further information about a particular cell type in the planarian brain that is indispensable for eliciting negative phototaxis (Nishimura et al. 2008b). A small number of the planarian GAD (DjGAD)-positive, GABA synthetic (GABAergic) neurons are observed only in the brain, and none are observed in cells forming the VNCs or in the photoreceptor cells (Fig. 4C). The DjGAD-knockdown animals showed a significant reduction of the amount of GABA and loss of negative phototaxis. Interestingly, we found a fine difference of behavior between the Djsnap-25 knockdown and the DjGAD knockdown animals. Loss of the Djsnap-25 activity in the brain results in defects of both locomotive and directional movement during negative phototaxis, which is regulated by two distinct types of primary sensory information coming from the head margin (probably mechanosensory organs) and eyes, respectively (Takano et al. 2007). In contrast, loss of the DjGAD activity in the brain results in only a defect of directional movement during negative phototaxis (Nishimura et al. 2008b). The data strongly suggest that the mechanosensory and photosensory circuits are mutually independent pathways in the planarian brain.

Identification of nou-darake gene

Regeneration is the sophisticated process by which animals reconstitute certain missing body parts after disease or injury. How can planarians regenerate a functional brain? The differentiation of brain cells from stem cells definitely depends on the body position along the anterior-posterior body axis during regeneration. Stem cells outside the presumptive head region never differentiate into brain cells, even though they clearly have the ability to differentiate into brain cells during regeneration. This observation indicates that stem cells receive proper positional information from their surrounding environment that regulates the brain regeneration in planarians. How is such an environment encoded in planarians? We identified the gene nou-darake (ndk means ‘brains everywhere’ in Japanese) and demonstrated that ndk inhibits stem cells from differentiating into brain cells outside the head region of the planarian Dugesia japonica (Cebriàet al. 2002a).

ndk encodes a putative transmembrane protein with two extracellular immunoglobulin (Ig)-like domains related to those of fibroblast growth factor (FGF) receptors, but lacks the cytoplasmic kinase domains characteristic of this receptor family. When ndk mRNA was injected into Xenopus embryos, it inhibited FGF signaling (Cebriàet al. 2002a). This inhibitory effect of NDK was still observed when both its intracellular and transmembrane regions were deleted (Cebriàet al. 2002a), indicating that the extracellular domain of NDK is responsible for its ability to inhibit FGF signaling. These findings strongly suggest that NDK functions as a dominant negative form of FGF receptor in Xenopus. We do not know yet whether NDK has the ability to bind to FGF ligand molecules or FGF receptors to inhibit FGF signaling in Xenopus.

In planarians, ndk is specifically expressed in both brain cells and non-brain cells in the head region. During regeneration, ndk is highly activated in the anterior stump early in the course of regeneration (Fig. 7B). Loss of function of ndk in planarian by RNAi caused gradual brain expansion to more posterior regions beyond the head region. This observation together with the fact that ndk is specifically expressed in the head region indicates that ndk functions in a non-cell-autonomous manner. The ectopic brain formation is suppressed by combined inhibition of two planarian FGF receptor genes (DjFGFR1 and DjFGFR2), which are expressed in pluripotent stem cells and also in brain cells (Ogawa et al. 1998, 2002). We found that ndk RNAi-induced ectopic brain structures are composed of a variety of neurons, and their formation is correlated with uncoordinated behavior of the knockdown animals. These data strongly suggest that FGF signaling promotes the formation of the brain structure as a functional unit in planarians.

Figure 7.

Predicted protein structures of ndk and vertebrate FGFRL1 genes, and their expression patterns. (A) Predicted protein structures. Both NDK and FGFRL1 lack the cytoplasmic kinase domains characteristic of the fibroblast growth factor receptor (FGFR) family. Ig, immunoglobulin-like domain. (B) Anterior-specific ndk expression (green) during regeneration. A middle fragment containing a pharynx (out of focus) allowed both head and tail regeneration. Immunostaining with anti-DjSYT visualized the ladder-like structure of the ventral nerve cords (VNCs) (magenta). (C) Anterior-specific Xenopus FGFRL1 (XFGFRL1) expression during embryogenesis (Hayashi et al. 2004). Anterior is to the left.

Based on the non-cell-autonomous function of ndk in planarians, we formed the following working model to explain the role of ndk (Fig. 6).

Figure 6.

Interpretation of ndk function and its RNA interference (RNAi) phenotype. The left panel shows an example of head regeneration from a tail fragment based on a working model proposed in this review (details in text). Green indicates the ndk-positive presumptive head region during regeneration. Dark and light gray indicate regenerating brain and regenerating pharynx, respectively. The upper right panel shows a control animal stained with a planarian brain-specific glutamate receptor gene as a probe (Cebriàet al. 2002a). The lower right panel shows ndk RNAi-induced ectopic brain formation. In many ndk RNAi planarians, ectopic brain formation occurs predominantly in the prepharyngeal region rather than in more posterior regions, suggesting that a low level of the brain-inducing activity may be retained in the prepharyngeal region (black line in left graph), but does not promote brain formation there under normal conditions. This may result in an increase in the level of the brain-inducing activity predominantly in the prepharyngeal region in the ndk RNAi planarians (red broken line in left graph). Anterior is to the left.

During head regeneration, yet-unidentified secreted brain-producing factors are first expressed in the anterior stump after amputation. These unknown signals, which are likely to be FGF-like ligand molecules, activate the FGFRs in pluripotent stem cells (the functions of DjFGFR1 and DjFGFR2 still remain unclear, because DjFGFR1 and 2 double knockdown animals did not show any defects in regeneration. Additional FGFRs appear to be required for brain formation during normal regeneration; Ogawa et al. 2002) to promote brain regeneration. One of the crucial downstream target genes of the FGF signaling is ndk, which may regulate the diffusion range of the brain-producing factors from the anterior end through a direct interaction with these factors, thereby trapping a proper amount of the active brain-producing factors within the presumptive head region (we speculate that the affinity of NDK for these factors should be lower than that of the FGFRs to keep NDK from interfering with the FGFRs within the brain-forming region). This ndk-based feedback loop ensures the formation of the proper extrinsic environment, in which stem cells are allowed to differentiate into brain cells during regeneration.

In the future, we must identify the secreted brain-producing factors in order to verify this model. However, no FGF-like ligand molecules have been identified yet in D. japonica or in the accumulating EST and genome sequence of the planarian Schmidtea mediterranea (Sánchez Alvarado et al. 2002; Robb et al. 2008). This suggests that it will be difficult to identify such molecules simply by glancing at their deduced protein sequences, and therefore we will need to use some other strategy, such as functional screening, in planarians to identify such factors.

Evolutionary implications of CNS development

The studies on ndk provide strong molecular evidence for the existence of a brain-inducing circuit based on the FGF signaling pathway in planarians. In chordates, FGF signaling also acts as a key regulator for neural induction during embryogenesis (Launay et al. 1996; Streit et al. 2000; Wilson et al. 2000; Wilson and Edlund 2001; Akai and Storey 2003; Bertrand et al. 2003). Furthermore, FGF signaling plays a crucial role in adult neurogenesis in vertebrates. In mice, FGF2 stimulates the proliferation and differentiation of neuroprogenitor cells in the adult hippocampus after brain injury (Yoshimura et al. 2001). The evolutionarily conserved requirement for FGF signaling for the establishment of the basic framework of the CNS leads us to imagine that FGF signaling may be related to commitment to a CNS lineage during evolution. Regarding the course of evolution of the nervous system, cnidarians occupy a unique phylogenetic position, in that they possess well-characterized nerve cells, but not the greater complexity of a CNS. Recently, the draft genome of one of the model cnidarians, the starlet sea anemone Nematostella vectensis, was reported (Putnam et al. 2007). Interestingly, N. vectensis possesses 15 FGF genes and two FGFR genes in its genome (Rentzsch et al. 2008). Furthermore, morpholino-mediated knockdown experiments indicated that some of these genes control the formation of the apical sensory organ during embryogenesis (Rentzsch et al. 2008). These observations indicate that the presence of genes encoding FGF signaling components in an animal's genome does not simply have an evolutionarily novel function inducing commitment to a CNS lineage.

ndk also provides another interesting standpoint from which to understand CNS evolution. In vertebrates, a transmembrane receptor, FGFRL1, containing three extracellular Ig-like domains related to FGF receptors but lacking the cytoplasmic kinase domain, exhibits striking similarity of all of these domain structures to those of NDK (Wiedemann and Trueb 2000, 2001; Trueb et al. 2005). Interestingly, the Xenopus ortholog of FGFRL1 (XFGFRL1) is first expressed specifically in the anterior region of embryos and is coexpressed with XFGF8 in many regions throughout embryogenesis (Hayashi et al. 2004), suggesting that ndk and FGFRL1 may play a common role in anterior specification by modulating FGF signaling. In contrast, N. vectensis does not have any ndk-related genes in its genome (Osamu Nishimura, pers. comm., 2008). These findings suggest that ndk/FGFRL1 might serve as one of the novel genes promoting cephalization that appeared during evolution (Fig. 7). Elucidating the role of ndk will help us to evaluate this idea.

We recently proposed that brain evolution may have been accompanied by the evolutionary emergence of the neural stem cell system, which enabled a drastic increase of the variety of neural cell types that can be produced from a single precursor via asymmetric cell division (Agata et al. 2006). This idea also leads us to imagine that distinct master control genes should have been recruited in the neural stem cell system for the generation of cellular diversity during evolution. Here, we are thinking about the gcm gene family during evolution. In fact, the gcm gene family displays quite divergent functions in CNS development among animal species, indicating a lack of strict conservation of the mechanisms of glial fate determination during evolution (Hashemolhosseini and Wegner 2004; Soustelle and Giangrande 2007). In mammals, two Gcm genes (Gcma/Gcm1 and Gcmb/Gcm2) were identified (Akiyama et al. 1996; Schreiber et al. 1997; Kim et al. 1998). However, all reported studies in mammals failed to detect significant levels of Gcm expression in the CNS lineage. A recent study in chicken provided the first strong evidence of Gcm expression in the vertebrate CNS. The chicken Gcm1 (c-Gcm1) is strongly expressed and has a neurogenic role in early neuronal lineages of the developing spinal cord (Soustelle et al. 2007). However, c-Gcm1 is not expressed in glial progenitors, nor did it induce glial cell fates when it was overexpressed (Soustelle et al. 2007). Our data indicated that the planarian gcm is expressed outside the neural lineage (Fig. 3). Furthermore, Caenorhabditis elegans does not have any gcm-related genes in its genome (Osamu Nishimura, pers. Comm., 2008), although it generates a variety of non-neuronal support cells from the same precursors that generate neurons. These findings suggest that dynamic changes of cis-regulatory elements of the gcm gene family may allow the gcm activity to be integrated into the neural stem cell system in animals in a species-dependent manner (Jones et al. 2004), and may thus have generated certain types of cells in the CNS lineage during evolution.

Conclusions and future prospects

The recent rapid advances in human stem cell research have brought the possibility of regenerative medicine much closer to becoming a reality and providing novel clinical applications for curing diseases and injuries by manipulating pluripotent stem cells (Okita and Yamanaka 2006; Takahashi and Yamanaka 2006; Lewitzky and Yamanaka 2007; Takahashi et al. 2007; Wilmut 2007; Yamanaka 2007, 2008a,b), and therefore stem cell biology is now one of the most interesting fields in biological research due to its therapeutic potential. This boom of stem cell biology is also emphasized by the increasing number of papers about the mechanism of regeneration discovered from studies in model-animal vertebrates such as newt, Xenopus and zebrafish (reviewed by Stoick-Cooper et al. 2007). These model animals provide a good opportunity to obtain molecular evidence about how the activity of stem cells is properly regulated in vivo during regeneration. However, it needs to be stressed that stem cells or stem-like cells in these model animals are still poorly characterized. In contrast, planarians have well-characterized somatic stem cells that are their only mitotic cells and that give rise to all cell types during regeneration (Baguñàet al. 1989; Shibata et al. 1999; Newmark and Sánchez Alvarado 2000; Orii et al. 2005; Reddien et al. 2005; Salvetti et al. 2005; Guo et al. 2006; Hayashi et al. 2006; Higuchi et al. 2007; Yoshida-Kashikawa et al. 2007). The well-defined stem cell system of planarians provides a golden opportunity to analyze the mechanisms regulating the activity of animal stem cells in vivo. In the near future, a deeper understanding of the mechanisms underlying brain regeneration in planarians may provide mechanistic insights into not only how to use regenerative medicine to treat neurodegenerative diseases such as Parkinson's disease (Nishimura et al. 2007a), but also the evolution of the brain.


We would like to thank all of our colleagues involved in the planarian brain project, Osamu Nishimura for genome-wide analyses in animal species, and Elizabeth Nakajima for critical reading of the manuscript. This review was written based on works supported by a Grant-in-Aid for Creative Scientific Research to K. A. (17GS0318), the Global COE Program A06 of Kyoto University, and the Naito Foundation.