Clathrin-mediated endocytic signals are required for the regeneration of,as well as homeostasis in, the planarian CNS

Planarians are free-living Platyhelminths, and are among the most primitive animals possessing a CNS, which is composed of an inverted U-shaped brain in the anterior region of the animal and two longitudinal ventral nerve cords along the body (Agata et al.,1998). The planarian CNS is relatively simple but well organized,consisting of several functional and structural domains, as defined by the discrete expression of three otd/Otx-related homeobox genes and a complex set of other regulatory genes(Umesono et al., 1997; Umesono et al., 1999; Cebrià et al., 2002b; Cebrià et al., 2002c; Mineta et al., 2003; Nakazawa et al., 2003; Koinuma et al., 2003; Inoue et al., 2004). Furthermore, planarians possess a remarkable regenerative ability that derives from pluripotent somatic stem cells (Agata and Watanabe, 1999; Saló and Baguñà,2002; Agata et al.,2003; Reddien and Sánchez Alvarado, 2004), whereby even a tiny fragment cut from most parts of the body can regenerate the entire body with a functional nervous system within 5 days (Inoue et al., 2004). To identify genes involved in functional brain regeneration, we systematically cloned genes expressed in the regenerating brain (Nakazawa et al., 2003; Mineta et al., 2003) and analyzed their expression timing during brain regeneration(Cebrià et al., 2002c). Those studies revealed that planarian brain regeneration consists of at least five steps: blastema formation, brain induction, patterning, neural circuit formation and functional recovery. Several candidate genes involved in each step have been identified. Djnlg, a planarian noggin homologue, and DjvlgA, a vas-related gene, are the earliest genes found to be involved in blastema formation (Ogawa et al., 2002; Shibata et al.,1999). The second-earliest gene is nou-darake, a dominant-negative-type fibroblast growth factor receptor (FGFR) that is activated in only the anterior blastema within 24 hours after amputation to form a brain rudiment (Cebrià et al., 2002a). DjotxA, DjotxB and Djotp,brain-specific homeobox genes, are activated in the brain rudiment 36 hours after amputation and act to create patterns in the brain(Umesono et al., 1997; Umesono et al., 1999). After patterning, netrin and several immunoglobulin superfamily neural cell adhesion molecule genes (IgCAMs) are activated in the brain approximately 72 hours after amputation to form neural circuits(Cebrià et al., 2002b; Cebrià and Newmark,2005; Fusaoka et al.,2006). Finally, 5 days after amputation, a functional brain is recovered (Inoue et al.,2004). The structure and function of the CNS depend on precisely controlled interactions among the neural cells, but the mechanisms of intercellular interaction required to construct and maintain the functional brain during brain regeneration are unclear. Here, we demonstrate that the clathrin heavy chain gene (DjCHC) of a planarian, Dugesia japonica, was required for proper brain regeneration and investigate at what point it acts during brain regeneration.

Clathrin-mediated membrane trafficking plays an important role in cell signaling and is involved in numerous cell functions, including nutrient uptake, regulation of the number of signaling receptors on the cell surface and the recycling of synaptic vesicles at nerve terminals (Takei and Haucke,2002; Murthy and DeCamilli,2003; Le Roy and Wrana,2005). Functional analyses of clathrin genes has been performed in cultured cells, fungi, trypanosomes and yeast, and have provided insights into the functions of clathrins and their interactions with other proteins in cell signaling (Seeger et al., 1992; Ruscetti et al., 1994; Niswonger and O'Halloran, 1997b; Wettey et al., 2002; Motley et al.,2003; Allen et al.,2003; Hinrichsen et al.,2003). However, much remains unknown about the developmental,molecular and cellular processes that determine the onset and maintenance of intercellular communication and the role of such communication in physiological responses of the nervous system. Here, by in vitro and in vivo analyses, we show that DjCHC-RNA interference (RNAi) inhibits neuronal survival, and neurite outgrowth and maintenance, and that DjCHC is required for proper CNS regeneration and homeostasis.

A clonal strain (GI) of the planarian Dugesia japonica maintained in autoclaved tap water at 22-24°C was used in this study. In all experiments, 8-10 mm-long planarians that had been starved for 2 weeks were used. For regeneration studies, animals were cut posterior to the auricle and pharynx on ice. To prepare X-ray-irradiated planarians, worms were irradiated with 12R X-rays using an X-ray generator B-4 (Softex).

cDNA clones (Dj_aH_000_01290HH, Dj_aH_507_P09, Dj_aH_402_I20, Dj_aH_302_F09 and Dj_aH_316_I10) encoding the respective proteins (DjCHC, Djmu-1, Djmu-2,Djunc13A and DjsbpA) were first identified in a previously constructed database of expressed sequence tags (ESTs)(Mineta et al., 2003). The longest cDNA of DjCHC was obtained by the stepwise dilution method(Watanabe et al., 1997) from a cDNA library constructed from poly A(+) RNA of whole planarians in the ZAPII vector (Stratagene). The sense primer 5′-TGATCATCGAAACTTACAAAACC-3′ and antisense primer 5′-ATACCCATATGTGCTCGTTCTAA-3′ corresponded to the sequence of Dj_aH_000_01290HH. The positive cDNA clone with the longest insert was re-cloned into pBluescript SK(-) (Stratagene).

Whole-mount in situ hybridization was carried out using 20 ng/ml of digoxygenin (DIG)-labeled riboprobes (Roche), as previously described(Umesono et al., 1997; Agata et al., 1998).

A fragment of Dj_aH_000_01290HH (amino acids 1190-1462) was inserted into pQE30 vector (Qiagen) and the construct was introduced into E. colistrain JM109. The fusion protein with a histidine tag was induced with isopropyl-β-D-thiogalactopyranoside and purified according to the supplier's protocol (Qiagen). For further purification, the gel slice containing the target protein was cut out after SDS-PAGE, and the purified protein was collected using an electroeluter (Amicon). The purified recombinant DjCHC protein was injected with Titer Max Gold adjuvant (CytRx)into mice (Balb/c) three times at intervals of 1 month. The antiserum against the fusion protein was prepared according to a standard procedure(Orii et al., 2002).

Whole-mount immunostaining was performed as described previously(Cebrià et al., 2002c). Planarians were stained using mouse antibodies: 1/2000 anti-planarian synaptotagmin (anti-DjSYT) (Tazaki et al.,1999); 1/2000 anti-DjCHC; 1/2000 anti-planarian G-protein βsubunit (anti-DjGβ) (AB245430); 1/5000 VC-1 monoclonal antibody(Sakai et al., 2000); and 1/200 anti-phospho histone H3 antibody (anti-H3P)(Hendzel et al., 1997)(Upstate Biotechnology), all of which were diluted in 10% goat serum in 0.1%Triton X-100-containing PBS (TPBS).

Double-stranded RNA (dsRNA) was synthesized essentially as previously described (Sánchez Alvarado and Newmark, 1999). Control animals were injected with dsRNA for green fluorescence protein (GFP), a protein not found in planarians. At 4 hours after the third injection, planarians were amputated immediately posterior to the auricles and pharynxes, and the resulting pieces were used for various assays.

BrdU labeling was performed by microinjection basically as described previously (Newmark and Sánchez Alvarado, 2000). RNAi-treated planarians (RNAi planarians) were prepared as described above and then cut into three pieces. At 3 hours after amputation, 10 mg/ml of BrdU was injected once into the intestinal tract of the trunk-piece. At either 4 or 48 hours after the injection, planarians were fixed and BrdU was detected immunohistochemically(Agata et al., 1998).

Fluorescence-activated cell sorting (FACS) and the culturing of planarian neural cells were performed basically as previously described(Asami et al., 2002). In total,50 RNAi planarians were cut into head, trunk and tail pieces. After 4 days,the regenerated heads of the trunk and tail pieces (total of 100 pieces) were collected and used for FACS. The collected cells were maintained statically in a humidified incubator at 22°C for 3 days in vitro (DIV).

Cultured cells were fixed in 4% PFA, 5/8 Holtfreter's solution for 60 minutes at 4°C, washed three times with TPBS and blocked in 10% goat serum in TPBS for 30 minutes at 4°C. The cells were then incubated with 1/2000 diluted mouse anti-DjCHC, mouse anti-DjSYT, mouse anti-DjGβ or hamster anti-planarian 14-3-3ϵ (whose signal was distributed throughout the whole cell; AB245429; used as counterstaining) for 60 minutes at 4°C. The samples were washed with 10% goat serum in TPBS for 5 minutes three times and signals were detected with 1/400 Alexa Fluor 488-conjugated goat anti-mouse IgG(H+L) (Invitrogen) or 1/400 Alexa Fluor 546-conjugated goat anti-hamster IgG(H+L) (Invitrogen) diluted 1/400 in 10% goat serum in TPBS for 60 minutes in the dark. After the samples were washed three times with TPBS, cell nuclei were labeled with Hoechst 33342 (Invitrogen). Fluorescence was detected with an LSM 510 confocal microscope (Carl Zeiss).

The head-abundant cell fraction on day 4 of regeneration (R4HAC) was cultured and collected at 0, 1, 2 or 3 DIV, and TUNEL reactions were performed(Hwang et al., 2004). The signal was enhanced using 1/400 Alexa Fluor 488-conjugated anti-fluorescein(Invitrogen). Immunostaining was performed using 1/400 hamster anti-planarian 14-3-3ϵ for counterstaining and 1/400 Alexa Fluor 546-conjugated anti-hamster IgG(H+L) (Invitrogen) as a counterstain.

Quantitative data were analyzed by one-way analysis of variance (ANOVA) and the statistical significance of differences between test results was determined by Student's t-test; P values greater than 0.05 were taken as not significant (NS). For in vitro analysis, data from at least eight images were averaged.

The full-length cDNA for DjCHC containing a 5381-bp insert with an open reading frame (ORF) encoding 1683 amino acids was screened by stepwise dilution screening (Watanabe et al.,1997) and named DjCHC. Sequencing analysis revealed that DjCHC has five highly conserved functionally distinct domains - the globular N-terminal domain, knee region, distal domain, proximal domain and C-terminal end (Ybe et al.,1999; Fotin et al.,2004; Wilbur et al.,2005) - and seven clathrin heavy-chain repeat (designated CHCR1-CHCR7) motifs (residues 546-1583) (see Fig. S1A in the supplementary material) (Ybe et al., 1999; Wilbur et al., 2005). Phylogenetic analysis using the full-length DjCHC showed high homology of this protein to the clathrin heavy chains (CHCs) of other eukaryotes (see Fig. S1B in the supplementary material).

Intact planarians were examined for the expression pattern of DjCHC. In agreement with previous reports, whole-mount in situ hybridization showed strong and ubiquitous expression of DjCHCtranscripts (Fig. 1A). In X-ray-irradiated planarians, weaker DjCHC signals were detected than in intact planarians, suggesting that DjCHC was expressed in various tissues, including in stem cells (Fig. 1B). To analyze the expression pattern of DjCHC protein,anti-DjCHC antibody was produced. DjCHC protein was also detected ubiquitously, but was detected especially strongly in the spongy region composed of the neurites of the neural cells(Fig. 1D). Confocal microscopy of the brain revealed much higher expression of DjCHC protein in the CNS compared with in other organs (Fig. 1E). Thus, the DjCHC gene might play a major role in the planarian CNS. Expression of the DjCHC gene within the newly formed blastema was first detected at day 2 of regeneration as a weak signal in two clusters of cells, which probably corresponded to the primordium of the inverted U-shaped brain region (Fig. 1G).

To identify genes involved in regeneration of the planarian brain, we performed expression analysis of the genes expressed in the regenerating CNS and a functional screen of these genes by RNAi, and thereby found that the DjCHC gene was essential for regeneration. Fig. 2A shows western blot analysis of equal protein loadings of the newly regenerated head region at 7 days after amputation of control and DjCHC-RNAi-treated planarians. In the controls, the DjCHC gene was expressed at a high level,whereas only a weak signal was detected in DjCHC-RNAi-planarians,indicating that RNAi-treatment downregulated DjCHC protein expression. The DjCHC-RNAi-treated planarians (139/139) seemed to be unable to regenerate their head or tail (Fig. 2B). When we cut planarians at levels posterior to the auricle and the pharynx into three pieces (head, trunk and tail pieces), extensive degeneration of the head pieces was observed (see Fig. S2A in the supplementary material). By contrast, almost all of the trunk pieces and tail pieces survived beyond 4 weeks, although they could not regenerate their heads. This suggests that DjCHC gene function is specifically required for regeneration and maintenance of the head region. Immunostaining of DjCHC-RNAi-treated planarians with anti-DjSYT as a pan-neural marker (Tazaki et al., 1999)and anti-DjGβ, which stains the lateral branches of the brain, revealed that DjCHC- RNAi-treated planarians could not regenerate their brain normally (Fig. 2C), which was confirmed by the fact that they could not recognize the direction of light and they hardly moved (see Fig. S2C in the supplementary material). To determine whether DjCHC gene function is specific to the brain, we investigated the effects of DjCHC-RNAi in other tissues and organs by using molecular markers, including a tail-specific homeobox gene, DjAbd-Ba(Nogi and Watanabe, 2001). Interestingly, the homeobox gene was expressed with the normal spatial pattern, but the regenerating ventral nerve cords (VNCs) in the posterior blastema were strongly disorganized (Fig. 3A). No disorganization was detected in the newly formed pharynx(Fig. 3B) or in the regenerating intestinal duct or body-wall musculature(Nogi and Levin, 2005; Orii et al., 2002)(Fig. 3C,D).

Head atrophy was also seen in X-ray-irradiated or RNAi-treated planarians with targeting of stem cell-related genes(Reddien et al., 2005a; Reddien et al., 2005b; Salvetti et al., 2005). Therefore, we speculated at first that the phenotype of the DjCHC-RNAi-treated planarians was caused by a stem cell defect,because DjCHC is expressed ubiquitously (including in stem cells). To test this possibility, we compared the phenotypes of planarians in which the stem cells were ablated by X-ray irradiation(Wolff and Dubois, 1948; Baguñà et al.,1989), DjCHC-RNAi-treated planarians and planarians that had undergone both treatments. It is known that the cell number in planarians is determined by the balance between cell proliferation and cell death, and even starved planarians can maintain cell proliferation while their body is shrinking in size via `tissue homeostasis'(Baguñà and Romero,1981; Sánchez Alvarado,2006). We investigated the phenotype of DjCHC-RNAi-treated planarians whose stem cells were destroyed by X-ray irradiation. Although, in control and X-ray-irradiated planarians, we could not find any defects through 14 days after irradiation or RNAi-treatment, CNS atrophy could be detected in DjCHC-RNAi-treated planarians from 5 day after RNAi-treatment, and the number of CNS-regressed planarians progressively increased thereafter until almost all planarians showed head atrophy by 14 days after RNAi-treatment(Fig. 4A,B). Furthermore,ablation of the stem cells by X-irradiation enhanced the CNS atrophy of the DjCHC-RNAi-treated planarians, perhaps because it caused a decrease in the cell number in the CNS(Baguñà and Romero,1981; Hwang et al.,2004). These results support the notion that the CNS atrophy caused by DjCHC-RNAi treatment was due to effects on differentiated neuronal cells rather than on stem cells.

To investigate the stem cells of DjCHC-RNAi-treated planarians in greater detail, the cell division and migration of stem cells were assessed using BrdU (Newmark and Sánchez Alvarado, 2000). The number and distribution of BrdU-positive stem cells (which might migrate to the blastema) did not differ between control and DjCHC-RNAi-treated planarians at 4 or 48 hours after BrdU injection(Fig. 4C). Furthermore,analysis of the number of mitotic stem cells using anti-H3P(Newmark and Sánchez Alvarado,2000) in uncut DjCHC-RNAi-treated planarians 14 days after RNAi treatment showed that H3P-positive cells were present in DjCHC-RNAi-treated planarians and survived for over 3 weeks even though the CNS was atrophied (Fig. 4D, and see Fig. S2B in the supplementary material). There were around 130 H3P-positive cells in an 8-mm-long planarian in both control(130.3±6.5 cells) and DjCHC-RNAi-treated (128.2±7.0 cells) planarians (not significantly different)(Fig. 4D,E). On the other hand,in planarians at 14 days after X-ray irradiation, when half of the treated animals had died (see Fig. S2B in the supplementary material), there were no H3P-positive cells and the CNS was normal(Fig. 4D,E). However, Reddien et al. (Reddien et al., 2005b)reported that a planarian homolog of the piwi gene was involved in supporting the generation of cells that promote regeneration and homeostasis,but not the division or migration of stem cells; we therefore performed further histological analyses using several molecular markers under DjCHC-knockdown conditions to identify the stage at which DjCHC is required for proper brain formation during head regeneration.

To test the ability of DjCHC-RNAi-treated planarians to form the blastema and brain rudiment, expression analysis was performed for Djnlg [a planarian noggin homologue rapidly induced by dorsoventral interaction after amputation to provide a positional cue to stem cells (Ogawa et al., 2002)],for DjvlgA expressed in the early blastema and stem cells(Shibata et al., 1999), and for nou-darake (ndk) expressed in the predicted brain region in the anterior blastema (Cebrià et al., 2002a). Djnlg, DjvlgA and ndk were detected in DjCHC-RNAi-treated planarians as in control planarians(Fig. 5A-C), indicating that DjCHC was not required for the formation of the blastema or brain rudiment early in regeneration.

Moreover, expression analysis was performed using the DjotxA,DjotxB and Djotp genes, which are thought to be involved in patterning of the brain 48 hours after decapitation(Umesono et al., 1997; Umesono et al., 1999). The expression patterns of these genes were not grossly different between control and DjCHC-RNAi-treated planarians, suggesting that brain patterning was not affected by DjCHC-RNAi(Fig. 5D,E).

Immunostaining using anti-DjSYT and anti-DjGβ 72 hours after amputation revealed conspicuous abnormalities of the regenerating brain. Positive signals were detected in the newly formed blastema region(Fig. 5F), but the brain architecture was disorganized. Although the regenerated brain was not properly organized, some normal features, such as an inverted U-shaped structure and lateral branches, were observed. Staining of visual cells with specific antibody against visual neurons (VC-1)(Sakai et al., 2000) revealed that eye regeneration was also perturbed. Planarian eye regeneration can be divided into three steps: formation of two visual-cell clusters in the dorsal side of the anterior blastema; projection of the left and right visual neurons to the opposite ones; and connection of the optic nerves onto the brain(Inoue et al., 2004). Although the differentiation of visual neurons was not inhibited by DjCHCsilencing, projection of left and right visual neurons to the opposite ones and connection of the optic nerves onto the brain were inhibited(Fig. 5G), implying that DjCHC might be required for the projection of axons during CNS regeneration.

To address the gene function of DjCHC in CNS regeneration and homeostasis at the cellular level, we used a primary culture system of planarian neural cells (HAC; head abundant cells) purified using fluorescence-activated cell sorting (FACS) and compared the head (with abundant neural cells) with the body (with few neural cells)(Asami et al., 2002). The DjCHC-RNAi-treated HAC contained DjCHC protein (data not shown) due to the fact that RNAi does not affect already expressed proteins. The RNAi technique, combined with the strong regenerative ability of planarians, makes it possible to disrupt the gene expression specifically in newly formed differentiated structures (T. Takano, J. Pulvers, T. Inoue, H. Tarui, H. Sakamoto, K.A. and Y. Umesono, unpublished). We used the newly formed head of DjCHC-RNAi-treated planarians to collect the head-abundant cell fraction on day 4 of regeneration (R4HAC) by FACS. This method is optimal for adjusting the start point of the RNAi effect, and therefore to perform time-course analysis.

Fig. 6A shows a comparison of the FACS profiles of the body and head fractions. The specific cell fraction designated R4HAC (Fig. 6A, black ellipse in middle panel) was detected in cells from the head but not from the body (Fig. 6A, left panel). R4HAC accounted for approximately 10% of the total cells in the head fraction in both control (10.9±2.9%) and DjCHC-RNAi-treated (9.3±3.5%) planarians (not significantly different) (Fig. 6B). Over 95%of the collected control and DjCHC-RNAi-treated R4HAC expressed the neural marker DjSYT (data not shown). Immunostaining of R4HAC at 0 days in vitro (DIV) using anti-DjCHC showed that 90.6±1.4% of control R4HAC were positive for DjCHC protein. DjCHC-RNAi treatment drastically reduced the percentage of DjCHC-protein-positive cells to 23.5±4.6%,and even positive cells expressed lower levels of DjCHC protein(Fig. 6C).

R4HAC began to extend neurites soon after culturing(Fig. 6D). Although the neurites of some cells regressed after 1 DIV, those of other cells showed additional elongation and branching by 3 DIV. In the cultures of DjCHC-RNAi-treated R4HAC, projection of neurites was observed, but almost all of the extended neurites regressed during culturing. The neurites of the DjCHC-RNAi-treated R4HAC (11.0±1.6 μm) were shorter than those of the control R4HAC at 1 DIV, and the length was further shortened to 2.1±0.3 μm at 3 DIV (Fig. 6F). The percentage of neurite-extending cells in DjCHC-RNAi-treated R4HAC was similarly decreased(Fig. 6E). Thus, neuronal cells of DjCHC-RNAi-treated planarians could not progressively extend the growth of or maintain their neurites.

To further evaluate the molecular features of DjCHC-RNAi-treated R4HAC, immunocytochemical analysis was performed using antibodies against DjCHC and DjSYT. DjCHC protein was strongly detected in neurites in R4HAC, but was not detected in DjCHC-RNAi-treated R4HAC, at 3 DIV. More than 95%of the control R4HAC at 3 DIV still expressed DjSYT protein, which was accumulated in neurites, consistent with the expression pattern in vivo (Figs 2, 3 and 4)(Tazaki et al., 1999). DjSYT expression was detected at least until 3 DIV(Fig. 6G), suggesting that the DjCHC-RNAi-treated R4HAC might maintain their neural identity.

To investigate how DjCHC mediates neuronal cell survival, the TdT-mediated dUTP nick-end labeling (TUNEL) assay(Hwang et al., 2004) was performed on R4HAC to detect apoptosis. DjCHC-RNAi treatment induced an increase in apoptosis (Fig. 7A-D). The degree of neuronal apoptosis was also quantified by measuring the percentage of TUNEL-positive cells among the total cells, and the data showed that the increase peaked at 3 DIV, when over 80% of the neurons were either dying or dead (Fig. 7A,B,E) and exhibited typical features of apoptosis(Fig. 6G, Fig. 7A-D). To examine whether DjCHC-silencing specifically caused apoptosis in DjCHC-protein-negative cells, we performed double TUNEL staining and immunostaining using anti-DjCHC. In the DjCHC-positive cell fraction, there was no significant difference in the fraction of TUNEL-positive cells between the control and DjCHC-RNAi-treated R4HAC, whereas, in the DjCHC-negative cell fraction, DjCHC-RNAi treatment markedly elevated the fraction of TUNEL-positive cells (Fig. 7C,D,F,G). Furthermore, all TUNEL-positive cells in the DjCHC-RNAi-treated R4HAC lacked neurites(Fig. 7B). Thus, dysfunction of the DjCHC gene might induce cell-autonomous apoptosis in neuronal cells.

Neurons have extensive and specialized requirements for protein trafficking, in order to transduce signals as well as to form the extremely polar shape of neurons. One of the unique functions of endocytosis in neurons is synaptic-vesicle recycling (Murthy and DeCamilli, 2003; Wu,2004). When we knocked-down the planarian unc13 homolog(Djunc13A; AB281585) planarian syntaxin-binding protein homolog(DjsbpA; AB281586) and the Djsyt genes involved in synaptic vesicle transmission, we could not find any abnormalities in the regeneration or maintenance of CNS architecture, although behavioral abnormalities were detected in these gene-knockdown planarians(Fig. 8A). This suggested that clathrin-mediated synaptic-vesicle recycling might not be important for the regeneration and maintenance of the CNS.

Clathrin functions in exocytosis at the endosome and trans-Golgi network(TGN) as well as in endocytosis at the plasma membrane(Hirst and Robinson, 1998). Thus, it was difficult to distinguish whether CNS abnormalities were caused by exocytic or endocytic dysfunction. To address the function of DjCHCnecessary for neuronal survival, we compared the effects of DjCHCwith those of the adaptor protein complex-1 (AP-1) gene,whose protein product is found on the TGN and endosomes, and the AP-2gene, which is specifically involved in endocytosis(Boehm and Bonifacino, 2001; Robinson, 2004). First, we isolated the planarian mu-1 gene, encoding a subunit of AP-1, and the mu-2 gene, encoding a subunit of AP-2, and named them Djmu-1 and Djmu-2,respectively (Fig. 8B). They are expressed ubiquitously (Fig. 8C,D). Then, we performed loss-of-function analysis by RNAi. Like DjCHC-RNAi-treated planarians, Djmu-2-RNAi-treated planarians could not regenerate their brain properly, although they showed a relatively weak phenotype (Fig. 8E). By contrast, no gross effects were detected in the brain regeneration of Djmu-1-RNAi-treated planarians. Moreover, gene knockdown of the γ and σ1 genes, which encode the other subunits of AP-1 (Hirst and Robinson, 1998) had no effect on CNS regeneration (data not shown). No additive phenotype was seen when double knockdown of Djmu-1 and Djmu-2 or of DjCHC and Djmu-2was performed (Fig. 8E,F). These data suggest that AP-2- and clathrin-mediated endocytosis, but not protein trafficking from the TNG via AP-1, are essential for proper CNS regeneration and maintenance in planarians.

To investigate the function of clathrin-mediated signaling in neuronal cells leading to proper regeneration of the brain in planarians, we isolated a full-length DjCHC cDNA. The Djmu-1 and Djmu-2genes, as well as the DjCHC gene, showed high structural similarity to the corresponding genes of other eukaryotes(Fig. 8B, and see Fig. S1 in the supplementary material) (Boehm and Bonifacino, 2001; Wilbur et al., 2005).

Clathrin silencing in the planarian prevented proper regeneration and maintenance of the CNS (Fig. 2). The headless or head-atrophied animals resulting from ablation of endocytic components did not die, because planarians have extraordinary regenerative potential. This feature of planarians was advantageous for analyzing the phenotype during the process of CNS formation and maintenance in vivo. Although the DjCHC transcripts were expressed ubiquitously, the DjCHC protein was strongly accumulated in neurites of CNS neurons(Fig. 1). Furthermore, gene knockdown of DjCHC without the regeneration process revealed that the CNS atrophied independently of stem cells(Fig. 4). These data imply that DjCHC might have an important role in neuronal cells in the formation and maintenance of the proper brain architecture. Notably, a tail, pharynx and intestinal tract were regenerated in the head- and tail-pieces of DjCHC-RNAi-treated planarians(Fig. 3A-C), indicating that intercalary regeneration, the principal manner of regeneration, generally occurred normally in DjCHC-depleted planarians(Kato et al., 1999; Agata et al., 2003). We could find no defects caused by silencing of DjCHC except for those in the CNS, despite the ubiquitous expression of this gene.

We could find no deficiency in stem cell division, differentiation or migration, as indicated by BrdU, H3P and FACS analysis. Histological analysis using several genes involved in different steps of CNS formation revealed that blastema formation, brain induction, patterning and cell differentiation occurred normally until 48 hours after decapitation in DjCHC-RNAi-treated planarians(Fig. 5). At 72 hours after decapitation, DjCHC-RNAi-treated planarians could express a postmitotic neural-marker protein and could form a disorganized brain(Fig. 5F). Furthermore,although clathrin dysfunction did not disturb the differentiation or the distribution of visual neurons, projection of left and right visual neurons to the opposite ones and the connection of the optic nerves onto the brain were defective (Fig. 5G), implying that DjCHC is not needed for cell differentiation of the brain and eye, but rather for the projection of the axons and formation of the proper CNS. Thus, clathrin is essential for the construction of the proper architecture of the CNS after patterning and neuronal cell differentiation,and before neural circuit formation during CNS regeneration, as well as for the maintenance of the proper architecture of the CNS after synaptic connection in planarians (Fig. 9).

In vitro analysis using primary cultured planarian neurons revealed that the numbers of control and DjCHC-RNAi-treated R4HAC collected were not significantly different, suggesting that differentiation and mitosis occurred normally in DjCHC-RNAi-treated planarians(Fig. 6B), in contrast to Dictyostelium cells and HeLa cells, in which clathrin inhibition blocks cytokinesis (Niswonger and O'Halloran, 1997a; Motley et al., 2003). The combination of RNAi and the in vitro analysis described here should prove useful for the functional characterization of neural genes at the cellular level during the regeneration and maintenance of the planarian CNS. At 3 DIV, almost none of the DjCHC-RNAi-treated R4HAC could lengthen or maintain their neurites, which seemed to be regressed but not retracted (Fig. 6D-F). These observations are consistent with the features of RNAi-treated planarians in vivo and suggest that DjCHC-silenced neurons could extend some neurites, but could not further extend or maintain these neurites, and seemed to undergo cell death after the completion of differentiation.

Interestingly, although the percentage of neurite-extending cells and the neurite length per cell of the control R4HAC peaked by 1 DIV and were subsequently reduced, the length of the neurite-extending cells tended to be slightly increased (from 30.5±1.9 μm at 1 DIV to 41.5±6.5μm at 3 DIV) (data not shown). It is tempting to speculate that some mechanisms related to appropriate synaptic connections or trophic factors might be important for neurite outgrowth and maintenance in this organism(Vaudry et al., 2002; Howe and Mobley, 2005). In contrast to the findings in control R4HAC, both the percentage of neurite-extending cells and neurite length of DjCHC-RNAi-treated R4HAC progressively decreased during culturing, indicating that dysfunction of clathrin-mediated endocytosis might accelerate neurite regression(Fig. 6E,F). Previous reports suggested that neurotrophic factors such as nerve growth factor are produced and released in target tissues to activate receptors on the presynaptic elements of innervating neurons, thereby signaling to regulate the neuronal survival of these neurons at nerve terminals involved in mediating endocytosis(Casaccia-Bonnefil et al.,1999; Ye et al.,2003; Yano and Chao,2004; Howe and Mobley,2005).

We tested whether clathrin dysfunction interferes with cellular functions such as cell survival by examining apoptosis in R4HAC, and found that DjCHC-RNAi treatment increased apoptosis(Fig. 7). This is consistent with the fact that chicken B cell line DT40 cells deficient in the clathrin gene undergo apoptosis (Wettey et al.,2002). However, clathrin-depleted HeLa cells do not show any apparent increase in apoptosis (Motley et al., 2003; Hinrichsen et al.,2003). These discrepancies will be resolved by further analyses revealing the details of the mechanism of endocytosis in each cell type.

Besides clathrin, one of the most important components of clathrin-mediated endocytosis is AP-2, which binds both clathrin and cytoplasmic tyrosine-based signals on transmembrane proteins (Boehm and Bonifacino, 2001; Robinson, 2004). AP complexes involved in cargo selection for inclusion into coated vesicles in the late secretory and endocytic pathways are widely distributed among eukaryotes(Boehm and Bonifacino, 2001; Boehm and Bonifacino, 2002). Our finding that the disruption of endocytic genes caused defective CNS regeneration, but disruption of an exocytic gene, Djmu-1, did not,suggested that clathrin-mediated endocytosis at the plasma membrane, not protein trafficking at the TNG, is necessary for CNS formation(Fig. 8E). This is consistent with our finding that the DjSYT and DjGβ proteins were detected in neurites in DjCHC-RNAi-treated planarians, indicating that protein trafficking occurred without the DjCHC gene(Fig. 2C, Fig. 5F). However, the possibility that AP-1-independent alternative factors in the clathrin-coated pits at the TGN are involved in proper CNS formation and maintenance could not be completely ruled out.

Interestingly, the gene-knockdown animals for DjCHC and Djmu-2 did not show exactly the same phenotypes: DjCHC-RNAi treatment caused a more-severe phenotype than Djmu-2-RNAi treatment. The distribution of α-adaptin and epsin and the cellular morphologies are different in CHC-siRNA- and mu-2-siRNA-treated HeLa cells(Motley et al., 2003; Hinrichsen et al., 2003). Furthermore, clathrin-mediated endocytosis can still occur in the absence of AP-2, and AP-2 may not be essential for the recruitment and assembly of clathrin at the plasma membrane (Motley et al., 2003; Hinrichsen et al.,2003; Conner and Schmid,2003). Our data, together with these previous findings, lead us to speculate that clathrin, AP-2 and alternative adaptors may co-assemble at the plasma membrane, bringing cargo with different types of internalization signals into the coated pit. In the absence of AP-2, alternative adaptors are still recruited onto the plasma membrane and co-assemble with clathrin(Mishra et al., 2001; Mishra et al., 2002; Motley et al., 2003; Hinrichsen et al., 2003; Sever, 2003). Furthermore,simultaneous silencing of Djmu-1 and Djmu-2 or DjCHC and Djmu-2 did not cause additive effects on CNS formation (Fig. 8E,F),indicating that the more-severe phenotype of DjCHC- RNAi-treated planarians was probably not caused by inhibition of both endocytosis and trafficking at the TNG. In planarians, alternative adaptors in addition to AP-2 in the clathrin-coated pits mediating endocytosis may be necessary for proper CNS regeneration and maintenance.

In summary, our data demonstrate that clathrin-mediated endocytosis is essential for neuronal cell survival, and for neurite outgrowth and maintenance, after the completion of neuronal differentiation to enable the formation as well as the maintenance of the proper functional CNS in planarians in vivo (Fig. 9). Studies of this enormously interesting organism will advance our understanding of the physiological roles, and the cellular and molecular mechanisms of intercellular communication and cellular signaling in maintaining the proper function of the nervous system.

We thank Elizabeth Nakajima and Kosei Takeuchi for careful reading of the manuscript. We also thank Hidefumi Orii (Department of Life Science,University of Hyogo) for his generous gift of anti-DjMHC-B antibody. This work was supported by a research grant from the Special Postdoctoral Researcher Program of RIKEN to T.I., a Grant-in-Aid for Creative Scientific Research to K.A. (17GS0318), a Grant-in-Aid for Scientific Research on Priority Areas to K.A., and the Grant for Biodiversity Research of the 21st Century COE (A14) to K.A.