Injury-induced activation of the MAPK/CREB pathway triggers apoptosis-induced compensatory proliferation in hydra head regeneration


Author to whom all correspondence should be addressed.


After bisection, Hydra polyps regenerate their head from the lower half thanks to a head-organizer activity that is rapidly established at the tip. Head regeneration is also highly plastic as both the wild-type and the epithelial Hydra (that lack the interstitial cell lineage) can regenerate their head. In the wild-type context, we previously showed that after mid-gastric bisection, a large subset of the interstitial cells undergo apoptosis, inducing compensatory proliferation of the surrounding progenitors. This asymmetric process is necessary and sufficient to launch head regeneration. The apoptotic cells transiently release Wnt3, which promotes the formation of a proliferative zone by activating the beta-catenin pathway in the adjacent cycling cells. However the injury-induced signaling that triggers apoptosis is unknown. We previously reported an asymmetric immediate activation of the mitogen-activated protein kinase/ribosomal S6 kinase/cAMP response element binding protein (MAPK/RSK/CREB) pathway in head-regenerating tips after mid-gastric bisection. We show here that pharmacological inhibition of the MAPK/ERK pathway or RNAi knockdown of the RSK, CREB, CREB binding protein (CBP) genes prevents apoptosis, compensatory proliferation and blocks head regeneration. As the activation of the MAPK pathway upon injury plays an essential role in regenerating bilaterian species, these results suggest that the MAPK-dependent activation of apoptosis-induced compensatory proliferation represents an evolutionary-conserved mechanism to launch a regenerative process.


Hydra, a model system for adult regeneration

The freshwater Hydra polyp is a classical model system to investigate the cellular and molecular basis of regeneration (Galliot & Schmid 2002; Steele 2002; Holstein et al. 2003; Bosch 2007). Hydra, which belongs to Cnidaria, exhibits a tube shape, with a mouth opening surrounded by tentacles at its apical pole, and a basal disc that secretes mucus at the basal one (Fig. 1A). The bilayered cellular organization relies on myoepithelial cells that constitute the endodermal and ectodermal layers, whereas the intermingled interstitial stem cells differentiate into neurons, mechano-sensory cells (nematocytes), gland cells and gametes (Fig. 1B). Therefore three distinct stem cell populations permanently self-renew and differentiate in the adult polyp, where cycling cells, i.e. stem cells and progenitors are found along the gastric region, whereas the extremities (tentacles, basal disc) are made of terminally differentiated cells that are continuously renewed as they get sloughed off. This highly dynamic maintenance of the Hydra polyp likely explains its amazing plasticity in homeostatic conditions as well as its possibilities for adult developmental processes (Galliot & Ghila 2010). Indeed adult animals can stand weeks of starvation without losing their fitness, but simply progressively reducing their size as a result of apoptosis and autophagy (Otto & Campbell 1977b; Bosch & David 1984; Chera et al. 2009a). By contrast when regularly fed, Hydra polyps make use of the excess of cycling cells by reproducing through budding, a form of asexual reproduction (Otto & Campbell 1977a). Moreover, as discovered 270 years ago, when bisected at any level along the body column, the animal regenerates all missing parts in a few days (Trembley 1744). This regeneration potential, which is not dependent on the feeding diet, actually goes to the point where animals can regenerate from dissociated tissues (Noda 1971; Gierer et al. 1972; Technau et al. 2000).

Figure 1.

 Apoptosis-induced compensatory proliferation in head-regenerating Hydra. (A) Anatomy of the Hydra polyp: the head region is formed of a dome (the hypostome) terminated by the mouth/anus opening and a circle of tentacles. At the aboral extremity the basal disc secretes a mucus that fixes the animal to substrates. The double arrow indicates the 50% body length bisection. (B) Hydra differentiated cells derive from three distinct stem cell populations (written red). The endodermal and ectodermal myoepithelial cells form the inside and outside layers, respectively. The interstitial stem cells are located in the ectodermal layer; they provide sensory neurons and nematocytes that remain in the ectoderm, ganglionic neurons that can migrate across the layers and gland cells that are all found in the endodermal layer. Most neuronal precursors produced in the body column migrate towards the extremities where they differentiate. (C, D) Apoptotic interstitial cells detected at 1 h post-amputation (hpa) in head-regenerating tip (region-1) with Hoechst staining combined to ribosomal S6 kinase (RSK) (green) and cAMP response element binding (CREB) (red) immunostainings (see also Fig. S3). Panel C shows three successive stages of the typical nuclear fragmentation: at the early stage (upper panel) the ring-shaped nucleus is formed of small islands of condensed chromatin, CREB staining is predominantly nuclear and homogenous, whereas RSK staining is cytoplasmic. At the advanced stage (middle panel) the nucleus is clearly fragmented and the CREB signal colocalizes with the large islands of condensed chromatin, while an intense RSK staining surrounds the nuclear area. At the late stage (lower panel) the CREB and RSK proteins colocalize at the periphery of the apoptotic bodies detected in the cytoplasm. The modulations of CREB subcellular localization during early head-regeneration suggest that pairs of i-cells and single i-cells display different sensitivity to pro-apoptotic signals (see Table S1). (D) Engulfment of apoptotic bodies (arrows) by an endodermal epithelial cell. Note the strong RSK peripheric expression. Scale bars: 8 μm. (E) Scheme depicting the structure of a head-regenerating stump at 4 h after mid-gastric bisection. In the tip (region-1), most interstitial cell derivatives underwent apoptosis after bisection and have released Wnt3. The cycling interstitial cells located in the sub-jacent region-2 have responded to Wnt3 exposure by activating beta-catenin signaling as evidenced by the nuclear translocation of beta-catenin in cycling cells between 60 and 90 min pa. These cells then rapidly undergo mitosis. Concomitantly interstitial progenitors from the basal regions (region-3 and below) migrate towards the wound.

Initiation of head regeneration and the activation of the MAPK/CREB pathway

Four successive phases characterize head regeneration, the immediate wound healing response, the early setting-up of the head-organizer, the early-late patterning phase that precedes the emergence of the tentacle rudiments, and the late growth of the head once the tentacle rudiments have emerged (Galliot et al. 2006). Within the first 10 h following mi-gastric bisection, beside the immediate wound healing process, transplantation experiments have evidenced a steadily rising head-organizer activity in head-regenerated tips. Indeed grafting experiments have shown that head-regenerating tips grafted 10 h after amputation onto an intact host polyp induce ectopic head formation (Browne 1909; MacWilliams 1983). During this time period the cells located in the head-regenerating-tips upregulate numerous “early” genes, which are supposed to deploy this head-organizer activity (Galliot et al. 1995; Martinez et al. 1997; Gauchat et al. 1998; Technau & Bode 1999; Hobmayer et al. 2000; Shimizu et al. 2002; Chera et al. 2007; Rentzsch et al. 2007; Lengfeld et al. 2009). Among those, the cAMP response element binding protein (CREB) transcription factor is submitted in the minutes that follow amputation to an asymmetric ribosomal S6 kinase (RSK)-dependent phosphorylation event, which occurs exclusively in the head-regenerating tips (Kaloulis et al. 2004). In fact when the MAP Kinase (MEK) is inhibited, RSK that is no longer phosphorylated is inactive and CREB remains unphosphorylated, then head-, but not foot-regeneration is blocked. Systematic pharmacological approaches also showed the importance of the MAPK pathway in the formation of the head organizer activity (Arvizu et al. 2006; Manuel et al. 2006). All together these studies lent weight to a key function for the MAPK/RSK/CREB pathway in the initiation phase of head-regeneration.

Immediate apoptosis-induced compensatory proliferation in head-regenerating tips after mid-gastric bisection

Recently a detailed analysis of the immediate and early cellular remodeling that takes place in regenerating halves identified a massive wave of apoptosis in early head-regenerating tips (Chera et al. 2009b). Beside typical apoptotic nuclei (Fig. 1C), one can detect apoptotic bodies engulfed by the endodermal epithelial cells (Fig. 1D) that transiently lose their epithelial polarity (Fig. S3). Interestingly this wave of apoptosis affects 50% of the cells in the head-regenerating tips (region-1) at 1 h post-amputation (hpa) whereas in the sub-jacent regions (region-2 and region-3) less than 1% of the cells undergo apoptosis (Fig. 1E). By contrast the level of apoptosis does not exceed 7% in the foot-regenerating tips. Hence this injury-induced wave of cell death is asymmetrically activated and remains localized in the vicinity of the wound. Importantly these apoptotic cells were shown to transiently overexpress the Wnt3 protein that subsequently activates the beta-catenin pathway in the adjacent cycling cells, which in turn synchronously divide by 4 hpa (Chera et al. 2009b).

Four types of functional evidences currently support the importance of apoptosis-induced compensatory proliferation in head regeneration in Hydra: First, treatment of head-regenerating Hydra with the pan-caspase inhibitor Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (ZVAD) inhibits apoptosis and head regeneration while, as expected, foot regeneration is not affected. Second, it is possible to force foot-regenerating tips to regenerate a head by briefly heating them locally. In this context the level of apoptosis is increased, Wnt3 is produced and the beta-catenin pathway is activated in the cycling cells. Third upon RNAi knockdown of the Wnt3 or beta-catenin genes, cell proliferation is abolished and the animals no longer regenerate their head. Fourth head-regeneration in ZVAD-treated animals can be fully rescued (and even speed up) by adding some Wnt3 protein in the medium at the time of amputation. All together these data indicate that apoptosis-induced compensatory proliferation is necessary and sufficient to launch head regeneration in Hydra (Chera et al. 2009b).

One key question regarding apoptosis-induced head regeneration concerns the characterization of the injury-dependent signaling that activates it. Given the immediate asymmetric activation of the MAPK/CREB pathway in head-regenerating tips (Galliot et al. 1995; Kaloulis et al. 2004; Arvizu et al. 2006; Manuel et al. 2006), this pathway appeared as an obvious candidate. In this paper we have investigated with pharmacological and genetic tools the role of the MAPK/CREB pathway in apoptosis-induced compensatory proliferation during head regeneration after mid-gatric bisection. The results indicate that the activation of this pathway plays an essential role for launching injury-induced apoptosis.

Materials and methods

Culture of animals, chemical treatments of live Hydra and regeneration experiments

Hydra vulgaris (Hv, Basel strain) and Hydra magnipapillata (Hm, 105 strain) were cultured as in (Gauchat et al. 2004). Regeneration experiments were performed at 18°C on mature budding Hv polyps, gently detached from their parents, bisected at mid-gastric position after 2–5 days of starvation and kept in Hydra Medium (HM, 1 mL/polyp). For pharmacological treatments, triplicates were pretreated for 90 min in U0126 (Cell Signaling, 20 μmol/L) or ZVAD (OMe)-FMK (Alexis, 20 μmol/L), bisected and left to regenerate in the same medium up to 90 min. These regenerating halves were either exposed to BrdU and fixed as indicated, or macerated for Hoechst and immunostaining to evaluate the level of apoptosis.

Hydra maceration and immunofluorescence

At indicated time-points, 30 regenerating halves were sliced off to provide each region as described in Figure 1E and gastric regions of 10 intact Hydra were used as control. Tissues maceration and immunodetection were performed as in (Chera et al. 2007) with the anti-RSK antibody (BD Bioscience N°610226, 1:1000) mixed either with the anti-hyCREB (Galliot et al. 1995; Kaloulis et al. 2004) (1:4000) or the anti-CBP (CREB binding protein) (Santa Cruz, sc-1211, 1:1000) antibodies.

Detection of apoptosis on whole mounts

The in situ cell death detection kit fluorescein (Roche) was used for terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assays. Hydra were fixed at 1 hpa in 4% paraformaldehyde (PFA) for 1 day, washed 3× 10 min in phosphate-buffered saline (PBS) and permeabilized for 10 min in 0.1% sodium citrate, 0.5% triton X100 at room temperature (RT). The samples were then heated for 10 min at 70°C, washed in PBS for 5 min, incubated in the TUNEL mixture at 37°C for 90 min, washed 2× 10 min in PBS, stained in Hoechst 33258.

BrdU-labeling, immunofluorescence on whole mounts and cell density

Hydra were incubated in 5 mmol/L BrdU/HM (Sigma) for 2 h as indicated, then relaxed in 2% urethane and fixed in 4% PFA overnight at 4°C. Samples were dehydrated in 100% methanol for at least 1 day, stepwise rehydrated in 75%, 50%, 25% methanol/PBS (10 min each step), then washed 2× 10 min in PBS, treated 20 min at RT in 2N HCl, washed in PBS 3× 5 min and blocked 2 h in bovine serum albumin (BSA) 2%. After overnight incubation in the mixed anti-BrdU (1:20, Roche Kit III) and anti-hyCREB (1:4000 clone 81 524) antibodies at 4°C, samples were washed in PBS 3× 15 min and exposed to the mixed AlexaFluor 488 and AlexaFluor 555 (Molecular Probes 1:400) secondary antibodies. After 3× 15 min PBS washes, samples were stained in Hoechst for 5 min, briefly washed, mounted in Mowiol and analyzed with the Leica SP2 confocal. Nuclei were counted on squared confocal snapshots with l = 190 μm. Cell density was established as the ratio between the number of CREB+ nuclei counted in a given region of the regenerating half over the number of CREB+ nuclei counted in the body column of intact Hydra. The BrdU-labeling index (BLI) was measured as the percentage of BrdU+ nuclei over the CREB+ nuclei.

Cloning the RSK and CBP Hydra genes

The Hv_RSK-609 cDNA was isolated with the RSK-for1 (gattgaaatcatgccacttgc), RSK-rev1 (ctacatctaacagaatattctctg) primers designed from the Hm_RSK EST-clone tac30b10.y1 (Genbank, Hydra EST sequencing project). This cDNA that encodes the N-terminus domain (200 AAs), is highly conserved, 66% identical (75% similar) to the rat RSK2 N-domain expressed to raise the anti-RSK antibody used here (see Fig. S1). The TAZ-KIX-332 Hm cDNA that encodes the full KIX domain was obtained with the CBP-TAZ-for2 (YtWYtWcatgcWcataaRtgY) and CBP-KIX-rev3 (tgggggttaatcatatttgaactg) primers derived from the Podocoryne CBP sequence (kindly provided by V. Schmid). The Hydra KIX domain is 56% identical (72% similar) to the mouse KIX domain that was expressed as immunogen to raise the anti-CBP antibody used here (see Fig. S2). The CBP-620 Hv cDNA was obtained with the CBP-for1 (ctaccgcattcaacacaagttaca) and CBP-rev1 (cataaagtttggactgttgaagag) primers designed from the Hm_CBP EST-clone tab 48d04.x1 (Genbank, Hydra EST sequencing project). Accession numbers: FR796474, FR796475.

RNA interference experiments

The Hv_CREBα-628, Hv_RSK-609, Hv_CBP-620 cDNAs were inserted into the pPD129.36 (L4440) double T7 vector. Hydra were exposed to dsRNAs as in (Chera et al. 2006; Buzgariu et al. 2008). Control animals were similarly exposed to the 183 bp dsRNAs produced from the L4440 vector. Briefly, 2× 2 mL bacteria suspension were pelleted after isopropylthio-β-d-galactopyranoside (IPTG) induction, resuspended in 200 μL 0.5% low melting point (LMP) agarose in HM, 5 mmol/L Tris pH 7.5 (HMT), solidified on ice, ground and given to 50 to 70 Hydra previously transferred in 0.5 mL HMT and exposed to glutathione (50 μmol/L). Bacteria feeding was conducted every other day. For regeneration experiments, Hydra were bisected 1 day after the last dsRNA feeding (Fig. 2A) and gene silencing was confirmed by reverse transcription–polymerase chain reaction (RT–PCR) analysis on 10 regenerating halves per condition.

Figure 2.

 Silencing of ribosomal S6 kinase protein (RSK), cAMP response element binding protein (CREB) and CREB binding protein (CBP) by RNA interference in Hydra. (A) Scheme depicting the regeneration assay in RNAi knocked-down Hydra. After one or several feedings with the bacteria-dsRNAs mixture, Hydra are bisected at the mid-gastric position the day following the last feeding (day +1). Animals are then either left to regenerate, or used for RNA extraction, or fixed for immunofluorescence on whole-mounts, or macerated for immunostaining. FR, foot regeneration (upper half); HR, head regeneration (lower half). (B) RSK, CREB and CBP expression detected in head-regenerating Hydra vulgaris (Hv) at 4 h post-amputation (hpa) by reverse transcription–polymerase chain reaction (RT–PCR) (24 cycles) after either 1× or 3× dsRNAs exposures. Beside the expected RSK, CREB and CBP knocked-down expressions in RSK(RNAi), CREB(RNAi) and CBP(RNAi) animals, respectively, note also the lower level of RSK expression in CBP(RNAi) Hydra compared with that detected in L4440 control animals. (C) RSK, CREB and CBP expression in regenerating Hydra magnipapillata (Hm) previously exposed 5× to dsRNAs detected at 4 hpa by whole-mount in situ hybridization. As previously reported (Chera et al. 2006), Hm requires more numerous dsRNAs feedings than Hv to obtain RNA interference.

Semi-quantitative RT–PCR experiments

At indicated time points, the mRNA of 10 head-regenerating halves was prepared after removing the basal disc with the QuickPrep micro-mRNA Purification Kit (GE Healthcare) and resuspended in 10 μL H2O. 50 ng mRNA was used for Sensiscript Reverse Transcription (Qiagen) and specific cDNAs were PCR amplified over 20 and 24 cycles with the pairs of primers RSK-for1/RSK-rev1 (609 bp), CREB-upfish2 (agcatccaagaaagcgtaaca)/CREB-rev564 (tgcgagttgatggggcgatgct) (628 bp), CBP-for1/CBP-rev1 (620 bp), hyBra1-f1 (gacattgatggagttgcgcatc)/hyBra1-r1 (gatttaaggcatctctttcgc) (680 pb) and Act-4HF (aaggattcctacgtcggtgacgaa)/ Act-4HR (ggataccahctgattccataccaa) (683 bp).


Silencing of the RSK, CREB and CBP genes in Hydra

To test whether RSK, CREB and CBP are involved in injury-induced apoptosis in head-regenerating tips, we knocked-down their expression by exposing the animals up to three times prior to amputation to dsRNAs produced from the empty L4440 vector or the RSK-, CREB-, and CBP-L4440 constructs (Fig. 2A). A single exposure to dsRNAs was sufficient to knockdown RSK, CREB and CBP expression in Hydra vulgaris as detected by RT–PCR (Fig. 2B) while several exposures were required in Hydra magnipapillata (Fig. 2C). Indeed whole mount in situ hybridization confirmed that the early upregulation of RSK, CREB and CBP observed after mid-gastric bisection, was no longer detected after several exposures to dsRNAs (Fig. 2C). Therefore all subsequent RNAi experiments were conducted in Hydra vulgaris. As the anti-CREB, anti-RSK and anti-CBP antibodies specifically cross-react with the cognate proteins in most cell types from intact animals (Fig. 3A and not shown), we also verified the level of protein silencing after one or three exposures to dsRNAs in Hydra vulgaris (Fig. 3B, C). This analysis showed that the RSK, CREB and CBP proteins were actually present after a single exposure to dsRNAs (Fig. 3B). In the absence of detectable transcripts (Fig. 2B), we concluded that these proteins were produced prior to dsRNA feeding. By contrast after three dsRNAs feedings, RSK, CREB and CBP were undetectable (Fig. 3C) while the levels of RSK, CREB and CBP were unaltered in control cells (from L4440(RNAi)3× animals –Fig. 3C, first row). This result was confirmed by western blot analysis (Fig. S4).

Figure 3.

 Progressive suppression of the ribosomal S6 kinase protein (RSK), cAMP response element binding protein (CREB) and CREB binding protein (CBP) after RNA interference in Hydra (Hv, Basel strain). (A) Immunodetection of RSK (green) and CREB (red) in cells from the body column of non-regenerating Hydra. Bars: 8 μm, except for the endodermal epithelial cells (endo. epith.): 16 μm. (B, C) Immunodetection of either RSK (green) and CREB (red) or RSK (green) and CBP (red) in non-apoptotic cells isolated from region1 at 1 h post-amputation (hpa) after a single (B) or three (C) exposures to dsRNAs as indicated on the left. Note that three feedings are necessary to fully silence RSK, CREB and CBP protein expression. Note also the epistatic interactions: CREB and CBP are no longer expressed when RSK is knocked-down (C, 2nd row); CBP is undetectable when CREB is knocked-down (C, 3rd row) but CREB expression is not altered when CBP is knocked-down (C, 4th row, white arrows). See also Fig. S4. Bars: 8 μm.

These cellular and biochemical analyses also showed that these three proteins likely interact in an epistatic fashion as in RSK(RNAi)3x cells RSK but also CREB and CBP were fully depleted, whereas in CREB(RNAi)3x cells both CREB and CBP were depleted (Fig. 3C). Therefore a sufficient level of RSK protein appears necessary to maintain detectable levels of CREB and CBP, whereas CREB seemingly sustains the CBP protein level. At the functional level these three proteins are essential to maintain homeostasis, as their full depletion was lethal in few days (not shown).

The pro-apoptotic function of the MAPK/CREB pathway in head-regenerating Hydra

To test whether RSK, CREB and CBP need to be expressed at the time of amputation to observe injury-induced apoptosis in head-regenerating tips, we knocked-down their expression by feeding once or repeatedly the animals with unspecific (L4440) or specific (RSK, CREB, CBP) dsRNAs prior to amputation. As a result neither apoptosis nor engulfment was detected in cells from head-regenerating tips (Fig. 4). TUNEL assays performed on whole mounts showed that apoptotic nuclei were indeed very few in head-regenerating tips of RSK(RNAi), CREB(RNAi) and CBP(RNAi) Hydra, although still present in foot-regenerating tips (Fig. 4A). The quantification of these apoptotic cells was performed on macerated tissues and indeed proved that <8% of the cells from region-1 were apoptotic at 1 hpa (Fig. 4B). The pharmacological approach confirmed this pro-apoptotic function of the MAPK/CREB pathway as Hydra exposed to the MEK-inhibitor U0126 that prevents RSK and CREB phosphorylation, (Kaloulis et al. 2004) also exhibited a drastic reduction in the number of apoptotic cells after bisection (Fig. 4C). Both assays thus indicate that the MAPK/CREB pathway is required for immediate apoptosis in head-regenerating tips. However a single exposure to dsRNAs was sufficient to prevent injury-induced apoptosis, indicating that the RSK, CREB and CBP proteins produced in homeostatic conditions prior to RNAi silencing do not suffice to induce apoptosis. The regulation of this pathway might actually be even more complex as RT–PCR analyses detected the transient presence of an injury-induced truncated CREB isoform (Kaloulis 2000, L. Ghila unpubl. data, 2009). This suggests that a stress-induced alternative splicing event is involved in this regulation as previously reported in vertebrates (Hua et al. 2006).

Figure 4.

 Inhibition of the mitogen-activated protein kinase/cAMP response element binding protein (MAPK/CREB) pathway upon ribosomal S6 kinase protein (RSK),, CREB and CREB binding protein (CBP) RNAi or U0126 exposure prevents the immediate wave of apoptosis in head-regenerating tips. (A) Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining of regenerating halves at 1 h post-amputation (hpa) from Hydra vulgaris (Basel strain) exposed once to RSK(RNAi), CREB(RNAi) or CBP(RNAi). The head-regenerating tip contains numerous apoptotic cells in L4440(RNAi) Hydra but none when RSK, CREB or CBP are knocked-down (arrows). In contrast the level of apoptosis is similar in the foot-regenerating tip (arrowhead). Scale bars: 100 μm. (B, C) Percentage of apoptotic cells in region 1 after 1× or 3× exposures to dsRNAs (B) or after U0126 treatment (20 μM, C). Apoptotic cells were detected after maceration as in Figure 1C at 1 hpa (B, C) up to 16 hpa (C). For each condition, triplicates were macerated and 600 to 700 cells were counted. Mean values and standard deviations are represented (see Table S2). DMSO, dimethylsulfoxide.

Inhibition of MAPK/CREB pathway alters the formation of the proliferative zone

To assess the role of the MAPK/CREB pathway in apoptosis-induced compensatory proliferation, the abundance of cycling cells was evaluated in Hydra knocked-down for CREB or pharmacologically inhibited for MEK signaling. In Hydra exposed once to CREB(RNAi), head-regenerating tips contained few apoptotic cells but exhibited a surprisingly high density of CREB+/BrdU+ cells (Figs 5A,B and 6A–C). When BrdU labeling was performed for 2 h before bisection, <40% of the cells were BrdU labeled at 4 hpa, showing similar BrdU-labeling index to that measured in untreated animals (Figs 5C, 6G,H). By contrast, at 8 hpa CREB+/BrdU+ cells were twice more numerous in region-1 (76%), significantly higher than that measured in region-2 in untreated animals (34%, Figs 5C, 6G,H). This high density of cycling cells in region-1 at 8 hpa was likely obtained at the expenses of lower regions that appeared strongly depleted in BrdU+ cells (9% in region-3), indicating a massive migration of cycling cells from the lower regions to the tip. When BrdU labeling was performed after bisection, the number of BrdU+ cells in head-regenerating tips at 4 hpa was extremely high (83% and 91%, Fig. 6G, H). If we assume that the period of BrdU exposure does not influence the speed of migration towards the wound after amputation, then the striking difference observed between untreated and CREB(RNAi)1× cells suggest that CREB(RNAi)1× cells massively re-enter the cell cycle between 0 and 4 hpa. Hence despite the absence of apoptosis, cell migration and cell cycle re-entry were not inhibited in CREB(RNAi)1x leading to the formation a proliferative zone in region-1 instead of region-2.

Figure 5.

 Alterations of apoptosis-induced compensatory proliferation in head-regenerating Hydra (Hv, Basel strain) exposed to cAMP response element binding protein (CREB)(RNAi), Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (ZVAD) or U0126. Confocal views of BrdU+ (green) CREB+ (red) cells along head-regenerating halves of Hydra either exposed either 1× (A–C) or 3× (D) to CREB dsRNAs, or treated with ZVAD (E) or U0126 (F). BrdU labeling was performed for 2 h before (C) or immediately after mid-gastric section (A, D, E left, F), or from 2 to 4 h post-amputation (hpa) (B, E right). Head-regenerating halves were then fixed at 4 or 8 hpa. Scale bars: 300 μm. A–C) After a single exposure to CREB dsRNAs (CREB(RNAi)1x) the CREB protein is still present even though transcripts are no longer detected (see Fig. 3B). This partial silencing leads to a mislocalization of the proliferative zone as observed here at 4 and 8 hpa with a strong increase in CREB+ (arrowhead) and BrdU+ (asterisk) cell density in region-1. All animals displayed a wound closure defect (C, D, dashed line) still visible at 8 hpa. (D) In CREB(RNAi)3×head-regenerating Hydra, CREB expression is either residual (left, arrow) or undetectable (right). Note that the few cycling cells are all CREB+. E) In ZVAD-treated animals, the cycling activity is dramatically decreased and the region-1 often collapsed (arrowhead). (F) In U0126-treated animals, CREB expression is hardly detectable and cycling cells are rare.

Figure 6.

 Inhibition of the mitogen-activated protein kinase/cAMP response element binding protein (MAPK/CREB) signaling pathway upon CREB(RNAi) or U0126 prevents cell proliferation in head-regenerating halves (Hv, Basel strain). (A) Schematic view of the three regions where the density of BrdU+ cells was analyzed. (B–F) Confocal views of BrdU+ (green) and CREB+ (red) cells at 4 h post-amputation (hpa) in animals exposed to BrdU over the first 2 h (0 to >2) after amputation. Scale bars: 50 μm. (G–J) Graphs showing the values of cell densities (upper panels) and BrdU-labeling index (BLIs, lower panels) along head-regenerating halves in untreated, CREB(RNAi)1x, Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (ZVAD) and U0126 treated hydra. BrdU-labeling was performed for 2 h either before (–2 to >0), or immediately after (0 to >2) amputation, or starting at 2 hpa (2 to >4). Animals were fixed at 4 or 8 hpa and BLIs were counted on confocal views. Note the 3× increase in cell density measured at 8 hpa in region-2 of untreated animals when compared to cell density measured in the body column of intact animal (1×). This increase was never observed in animals exposed to CREB(RNAi), ZVAD or U0126. See corresponding values in Table S3.

However, the proliferative regions in CREB(RNAi)1× and untreated animals are likely not equivalent: despite an extremely high density of BrdU+ cells in CREB(RNAi)Hydra the observed increase in cell density at 8 hpa in head-regenerating tips never exceeded 2× when this increase reached 3× in the wild-type context (Fig. 6G, H). This indicates that the synchronous mitotic activity observed between 2 and 4 hpa in untreated Hydra (Fig. 7A) was not observed in CREB(RNAi)Hydra. The analysis of the size of the nuclei at 4 hpa indeed confirmed this lack of synchronous cell division (Fig. 7B, C). When RNAi was continued for three feedings, CREB+/BrdU+ cells were either very rare and clustered, or completely absent (Figs 5D and 6D). Therefore when CREB silencing was partial, the amputation-driven apoptosis was abolished but the CREB+ cells were still able to enter S-phase and gather close to the wound although unable to synchronously divide (Fig. 7B, C). When CREB was completely silenced, apoptosis and cell cycling were no longer observed. Moreover, a defect in the wound healing process was noted in both contexts, still present at 8 hpa (Fig. 5C, D).

Figure 7.

 Inhibition of injury-induced apoptosis in head-regenerating Hydra (Hv, Basel strain) correlates with a decreased mitotic activity of the interstitial cells. (A–C) Analysis of the BrdU+ nuclei performed on whole mount head-regenerating halves exposed to BrdU for 2 h immediately after bisection. The white dashed line underlines medium size nuclei, the red dashed line underlines their putative progeny at 4 h post-amputation (hpa) (small sized nuclei), the blue dashed line circles the interstitial cells (i-cells) undergoing mitosis. (A) In region-2 of control Hydra, the density of BrdU+ cells is high at 2 and 4 hpa (≥ 50%). Note the dramatic increase in the number of small sized nuclei at 4 hpa that belong to small i-cells arising from large i-cells by mitotic division. (B) In region-1 at 4 hpa only large i-cells can be detected as small i-cells underwent apoptosis. In CREB (RNAi)1×Hydra pairs of i-cells (surrounded by white dashed line) accumulate in region-1 without undergoing mitotic division. These pairs of i-cells are less numerous or even rare in Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (ZVAD)- or U0126-treated Hydra. Scale bars: 10 μm. (C) Sorting of the BrdU+ nuclei at 4 hpa in three distinct classes: the largest nuclei (>12 μm) from the epithelial cells, the middle sized nuclei (8–12 μm) from large i-cells, and the small sized nuclei (< 8 μm) from the small i-cells (see Table S4). Note that most BrdU+ cells are small i-cells in region-2 of untreated hydra while large i-cells are still abundant in the three regions in CREB(RNAi), Z-VAD or U0126 treated animals.

The importance of apoptosis to induce compensatory proliferation is also evidenced in ZVAD-treated Hydra that never form any proliferative zone (Figs 5E and 6E). Indeed when apoptosis is inhibited by a short ZVAD treatment, the cell density remains basically unregulated in head-regenerating halves, except a mild increase to 1.5× noted at the 8 hpa in region-1 (Fig. 6I, upper). Accordingly the density of CREB+/BrdU+ cells stays low, whatever the region, whatever the time-point, except a slight increase at 4 hpa in region-1 (52%) when animals are exposed to BrdU immediately after bisection (Fig. 6I, lower). This suggests that, despite the absence of apoptosis, a limited number of cells are still able to re-enter the cell cycle after amputation, although not dividing at 4 hpa (Fig. 7B). In both ZVAD and CREB(RNAi)1× phenotypes the cycling cells appear shifted to the tip of head-regenerating halves.

Finally the role of the MAPK pathway in apoptosis-induced compensatory proliferation is confirmed by exposing wild-type Hydra to the U0126 MAPK inhibitor (Figs 5F and 6F, J). This treatment causes a dramatic reduction in the number of BrdU+ cells, possibly linked to the strong decrease in the level of CREB expression (Figs 5F and 6F). As expected the regeneration-induced increase in cell density is not observed in U0126-treated Hydra where, in all three regions and at all time points, the cell density remains similar to that measured in intact untreated Hydra (Fig. 6J, upper). Indeed in these animals BrdU-labeling indexes and mitotic activity at 4 hpa are dramatically low (BLI <20%, Fig. 6I lower, Fig. 7B, C), indicating a strongly altered cell cycling response.

Expression of RSK, CREB and CBP is required for head regeneration

To monitor the kinetics of head regeneration in RNAi Hydra, the progression through the four morphologically distinct phases of head regeneration was recorded (Fig. 8A). In RSK(RNAi), CREB(RNAi) and CBP(RNAi) Hydra head-regeneration was dramatically delayed as phase-1 lasted from 2.8× to 4× longer after a single exposure to dsRNAs, over 5× longer after two. After two dsRNAs exposures, the early apical differentiation process (phase-3) was also significantly lengthened. After three feedings, Hydra were no longer able to regenerate and died (not shown). In animals exposed to L4440 dsRNAs, the RSK, CREB and CBP genes were found expressed at each phase of the head-regeneration process, although with some variations (Fig. 8B). Because non-lethal RNAi conditions only delay head-regeneration, we looked at whether gene silencing was merely transient. Indeed in RSK(RNAi), CREB(RNAi) and CBP(RNAi) Hydra, the corresponding genes were re-expressed after several days: after a single dsRNA exposure, RSK was first re-expressed around 72 hpa, CREB 12 h later and finally CBP reached a significant level at 96 hpa (Fig. 8C–E). Hence their respective re-expressions preceded by about 12 h the first morphological modifications appearing in head-regenerating tips. As previously noticed (Fig. 2B), we found RSK downregulated in CBP(RNAi) Hydra, re-expressed only when CBP transcripts were back (Fig. 8E).

Figure 8.

Ribosomal S6 kinase protein (RSK), cAMP response element binding protein (CREB) or CREB binding protein (CBP) silencing significantly delays the early phase of head regeneration. (A) Duration of each head-regeneration phase in Hydra (Hv, Basel strain) bisected after one or two exposures to RSK, CREB and CBP dsRNAs (see Table S5). In wild-type conditions, phase 1 (approximately 25 h) precedes any sign of apical differentiation, phase 2 (approximately 10 h) corresponds to the appearance of the first signs of apical differentiation, i.e. elongated halves with square tips, phase 3 (approximately 5 h) is marked by the emergence of the first tentacle rudiments, phase 4 starts when the length of the newly formed tentacles exceed the head diameter. Note that RNAi silencing of RSK, CREB or CBP predominantly affects the initiation of regeneration: phase 1 is lengthened by 3× to 4× after a single feeding, up to 6× after two feedings. However after two exposures to dsRNAs, phase 3 is also significantly lengthened. (B–E) RSK, CREB, CBP, hyBra1 and actin expression in head-regenerating halves of Hydra (Hv, Basel strain) exposed once to unspecific dsRNAs (empty L4440 vector) or to RSK, CREB, CBP dsRNAs. Note that animals resume the head regeneration process, i.e. get out of phase-1, exactly when expression of the knocked-down gene resumes.

We searched for a possible delayed apoptotic process at the time genes were re-expressed and screened the cells from head-regenerating tips every 3 h over the 2 days preceding the predicted emergence of tentacle buds. However, we failed to detect any sign of apoptosis and/or metaplasia (not shown), suggesting that these cellular events do not occur in the absence of amputation. If confirmed, we anticipate that head regeneration once RSK, CREB, CBP expression resumes occurs through a different mechanism, possibly by-passing the injury-induced apoptosis and the apoptosis-induced compensatory proliferation phases.


Injury-induced apoptosis requires a functional MAPK/CREB pathway in Hydra

Previous work had shown the importance of apoptosis-induced compensatory proliferation to specifically launch a head regeneration process in Hydra after mid-gastric bisection (Chera et al. 2009b). Beside ZVAD that inhibits injury-induced apoptosis, we have now shown that inhibition of the MAPK/CREB pathway either pharmacological (U0126 treatment) or through RNAi silencing of the RSK, CREB and CBP genes also prevent injury-induced apoptosis. This function of the MAPK pathway in Hydra apoptosis is not surprising as the MAPK/c-Jun N-terminal kinase (JBK) pathway was shown to regulate apoptosis in a variety of contexts (Kuranaga & Miura 2002; Lin 2003) either during development as in C. elegans (Gumienny et al. 1999) and Drosophila larvae (Kuranaga et al. 2002; Moreno et al. 2002), or during tissue repair as in Drosophila larvae (Ryoo et al. 2004; Bosch et al. 2005; Bergantinos et al. 2010a) or in mammalian cells (Wang et al. 2007). Although initially reported as exhibiting an anti-apoptotic effect (Xia et al. 1995), the MAPK/ERK pathway can also promote apoptosis as in the ascidian Ciona larva losing its tail during metamorphosis (Chambon et al. 2007) or in mammals differentiating their thymocytes (Sohn et al. 2008). Despite the fact that the role of JNK during apoptosis in Hydra is currently unknown, these data indicate that the injury-dependent regulation of the MAPK activity was already established in cnidarians, where it can contribute to apoptosis.

Apoptosis-induced compensatory proliferation requires a functional MAPK/CREB pathway in Hydra

Thanks to the three tools that inhibit injury-induced apoptosis (ZVAD, U0126 or CREB[RNAi]), we could compare what is left of the compensatory proliferation process in head-regenerating halves when injury-induced apoptosis is inhibited. Two criteria were considered: (i) the injury-induced increase in the number of cycling interstitial cells close to the wound as deduced from the comparison of the number of BrdU+ cells labeled for 2 h either immediately before or immediately after bisection; (ii) the synchronous division of the cycling interstitial cells as evidenced by the dividing cells observed at 4 hpa, the resulting increase in cell density recorded at 8 hpa and the dramatically low BrdU labeling index when BrdU is given between 2 and 4 hpa.

In all three inhibitory contexts the synchronous mitosis of the interstitial precursors is abolished as evidenced by the lower increase in cell densities recorded at 8 hpa: 2× in CREB(RNAi)1× tips, 1.5× after ZVAD, 1.2× after U0126 versus 3× in untreated polyps. However, the injury-induced increase in the number of cycling interstitial progenitors is fully efficient in CREB(RNAi)Hydra, limited in ZVAD-treated Hydra and absent in U0126-treated Hydra. In fact, a partial CREB silencing (CREB(RNAi)1×) where the CREB transcripts are undetectable but the CREB+/BrdU+ cells still numerous, is sufficient to prevent the activation of apoptosis and to modify the proliferative response: first the proliferative zone is mislocalized at the regenerating tip (where apoptosis normally takes place) and these cells are unable to divide in time and to drive head regeneration until CREB expression resumed. When the CREB and CBP proteins are no longer detectable (CREB(RNAi)3×, Fig. 3C), cell proliferation is totally abolished in most animals that survive amputation for only a few hours.

These data are in agreement with the following scenario (Fig. 9): in the absence of apoptosis, the transient source of Wnt3 signal produced by the apoptotic cells is missing and the subsequent activation of beta-catenin signaling in the surrounding cycling cells cannot take place. As a consequence the cycling cells present at the tip of regenerating halves do not undergo synchronous mitotic division as observed in region-2 of control animals. These results indicate that CREB that is overexpressed in cycling cells in homeostatic conditions (Chera et al. 2007), is actually playing an additional function in the head regenerative context. Beside its positive effect on cell proliferation, CREB that is submitted to RSK-dependent phosphorylation in head-regenerating tips immediately after mid-gastric bisection, exerts a transient pro-apoptotic function.

Figure 9.

 Working model for the immediate activation of the mitogen-activated protein kinase/cAMP response element binding protein (MAPK/CREB) pathway in wild-type Hydra regenerating their head after mid-gastric bisection. The upper graph describes the dynamics of the apoptotic cell population in region-1, assuming a transient induction of the apoptotic event followed by first-order transfers from initially affected cells to subsequent apoptotic stages. According to this model, the highest fraction of early apoptotic cells is expected at 30 min pa; these cells transiently produce Wnt3, which activates beta-catenin in the neighboring cycling cells and promotes their subsequent synchronous mitosis. At the signaling level, the amputation stress leads to the immediate MEK-dependent phosphorylation of ribosomal S6 kinase protein (RSK) in head-regenerating tips, which results in the hyperphosphorylation of CREB isoforms. In foot-regenerating halves RSK activity is low, as a consequence apoptosis and Wnt3 production is very limited, beta-catenin is not activated and compensatory proliferation is not observed. How the asymmetrical activation of the MAPK/CREB pathway promotes apoptosis remains to be established.

One possible mechanism for this injury-induced pro-apoptotic function might be linked to the production of injury-specific isoforms. In vertebrates, multiple CREB isoforms were identified with sometimes opposite activities (Mayr & Montminy 2001). As an example a stress-induced truncated isoform of the CREB-related gene activating transcription factor 3 (ATF3) exhibits a pro-apoptotic function by reducing the CBP-dependent activation of anti-apoptotic genes (Chen et al. 1994; Hua et al. 2006). As in vertebrates, several CREB isoforms were identified in Hydra including one truncated isoform that is transiently produced after mid-gastric bisection (Galliot et al. 1995; Kaloulis 2000). Further experiments will tell us whether this amputation-induced CREB isoform also plays a pro-apoptotic function in Hydra.

Sensitivity to cell death signals seems cell lineage-dependent in Hydra

In Hydra, where the apoptosis machinery is well conserved (Lasi et al. 2010a,b), apoptosis was characterized in a variety of physiological and developmental contexts: (i) in starving Hydra where the proliferative rate is maintained for weeks, the excess of produced cells is eliminated by apoptosis (Bosch & David 1984); (ii) during oogenesis the nurse cells undergo partial apoptosis and engulfment by the oocyte (Honegger et al. 1989; Miller et al. 2000; Technau et al. 2003), whereas during spermatogenesis a subset of sperm precursors become apoptotic (Kuznetsov et al. 2001); (iii) after wound-healing or bisection differentiating nematocytes located along the body column are submitted to cell death (Fujisawa & David 1984); (iv) during head regeneration after mid-gastric bisection, interstitial cell derivatives and progenitors located in head-regenerating tips are rapidly eliminated (Chera et al. 2009b); (v) in the thermosensitive sf-1 mutant, heat shock rapidly induces the apoptosis of the S-phase cells (Terada et al. 1988; Cikala et al. 1999). Apoptosis can also be induced in Hydra polyps either by pharmacological treatments that affect the mitotic spindle as colchicine (Campbell 1976) or inhibit the Pi-3 kinase as Wortmannin (David et al. 2005), or by physical treatments as grafting polyps from different species together (Kuznetsov et al. 2002), or briefly heating the cells of foot-regenerating tips (Chera et al. 2009b). Except mid-gastric head regeneration and heat-induced apoptosis in foot-regenerating tips, it is currently not known whether apoptosis might lead to compensatory proliferation in some of the contexts listed above. However in each of these contexts, the cells that are mostly sensitive to the death signals belong to the interstitial cell lineage and not to the epithelial cell lineages. Indeed those appear extremely resistant to the apoptosis-inducing signals although they quickly respond to the presence of apoptotic bodies by engulfing them. This is in sharp contrast with the Drosophila studies where the epithelial cells of the imaginal discs were shown to easily enter apoptosis upon irradiation, heat-shock or genetic manipulation of the cell death signaling pathway (Bergantinos et al. 2010b).

Injury-induced apoptosis, an evolutionarily-conserved process to launch a regenerative response?

We previously reported that apoptosis-induced compensatory proliferation is necessary and sufficient to drive head regeneration after mid-gastric bisection (Chera et al. 2009b; Galliot & Chera 2010). The data shown here indicate that the asymmetric activation of the MAPK/CREB pathway likely participates in the signaling cascade that leads to apoptosis. Cell death was actually reported as a driving force in several regenerative contexts (Bergmann & Steller 2010): in humans repairing their tissue after irradiation (Kondo 1988), in rodents regenerating their skin or their liver (Li et al. 2010), in Drosophila larvae regenerating their imaginal discs after induction of cell death (Bergantinos et al. 2010b), in Xenopus regenerating its tail (Tseng et al. 2007). Also in newt regenerating its limb (Vlaskalin et al. 2004) and in planarians regenerating their body (Hwang et al. 2004; Pellettieri et al. 2010), cell death was reported but its role in the regenerative process is unknown. More generally these studies point to the non-apoptotic functions of caspases as reported during developmental processes (Kuranaga & Miura 2007).

In Drosophila imaginal discs, “undead” cells (cells that are induced to enter apoptosis but cannot proceed to complete the cell death process) were shown to release Wnt or Dpp signals to induce compensatory proliferation of their neighboring cells and regenerate the missing tissue (Huh et al. 2004; Perez-Garijo et al. 2004; Ryoo et al. 2004; Fan & Bergmann 2008; Smith- Bolton et al. 2009). However, this release of signaling molecules might be linked to the “undead” status of the cells (Perez-Garijo et al. 2009) and the activation of the JNK pathway might suffice to trigger the compensatory growth (Bergantinos et al. 2010a; Warner et al. 2010). In intact Hydra JNK is predominantly expressed in differentiating nematocytes (Philipp et al. 2005) but JNK also appears to be involved in the stress response (Bridge et al. 2010). Therefore JNK might play a role in apoptosis-induced compensatory proliferation in Hydra. If identified, then the balance between CREB and jun would also need to be tested in Hydra as the balance between ATF and jun regulates the sensitivity of mammalian cells to apoptosis (Bhoumik et al. 2002). Comparative investigations in phylogenetically distant contexts where initiation of regeneration relies on apoptosis-induced compensatory proliferation should help identify some shared mechanisms of interest for regenerative medicine.


The authors thank Philippe Jean and Kevin Dobretz for their work on the characterization of the CBP gene. This study was supported by the Canton of Geneva, the Swiss National Foundation (FNS 3100A0-116784), the National Center for Competence in Research (NCCR) “Frontiers in Genetics” Stem Cells & Regeneration Pilot project, the Claraz Donation and the Academic Society of Geneva.