RhoA small GTPase, as a key regulator for actin cytoskeletal rearrangement, plays pivotal roles during morphogenesis, cytokinesis, phagocytosis and cell migration, but little is known about its signaling mechanism that controls cell survival in vivo. Using zebrafish as a model, we show that non-overlapping antisense morpholinos that block either translation or splicing of rhoA lead to extensive apoptosis during embryogenesis, resulting in overall reduction of body size and body length. These defects are associated with reduced activation of growth-promoting Erk and decreased expression of anti-apoptotic bcl-2. Moreover, ectopic expression of rhoA, Mek or BCL-2 mRNA rescues such phenotypes. Consistently, combined suppression of RhoA and Mek/Erk or Bcl-2 pathways by sub-optimal dose of rhoA morpholino and pharmacological inhibitors for either Mek (U0126) or Bcl-2 (HA 14-1) can induce developmental abnormalities and enhanced apoptosis, similar to those caused by effective RhoA knockdown. Furthermore, U0126 abrogates the rescue by RhoA and MEK but not BCL-2. In contrast, HA 14-1 effectively abolishes all functional rescues by RhoA, MEK or BCL-2, supporting that RhoA prevents apoptosis by activation of Mek/Erk pathway and requiring Bcl-2. These findings reveal an important genetic and functional relationship between RhoA with Mek/Erk and Bcl-2 for cell survival control during embryogenesis.
Rho small GTPases are members of the Ras superfamily (Etienne-Manneville and Hall, 2002), which cycle between the GTP-bound active state and GDP-bound inactive state. RhoA is one of the best-characterized members with well-established role in controlling cytoskeletal rearrangement, but little is known on how it regulates cell survival in vivo. Several anti-apoptotic pathways have been implicated in the suppression of apoptosis by RhoA. For example, the expression of constitutively-active RhoA promotes activation of extracellular signal-regulated kinase (ERK) and facilitates glomerular epithelial cells survival (Bijian et al., 2005), while inhibition of Rho induces p53 in human endothelial cells (Li et al., 2002). Also, RhoA upregulates anti-apoptotic Bcl-2 expression in T cells (Gomez et al., 1997), vascular smooth muscle cells (Blanco-Colio et al., 2002) and osteosarcoma cells (Fromigue et al., 2006). On the contrary, it has been reported that activation of RhoA–ROCK pathway also induces apoptosis (Aznar and Lacal, 2001; Coleman et al., 2001). However, all these studies were restricted to in vitro culture system. How RhoA regulates apoptosis in vivo remains to be seen.
Linking cell survival to development, it has been shown that inhibition of RhoA activity by C3 transferase leads to a decrease in the number of thymocyte by increasing apoptosis and reducing proliferation (Henning et al., 1997), while suppression of RhoA–ROCK signaling by conditional expression of dominant negative RhoA or ROCK in transgenic mice enhances apoptosis, leading to reduced number of motor neurons in the spinal cord (Kobayashi et al., 2004). However, many of these findings on the inhibition of RhoA functions in vivo are obtained primarily using inhibitors to block Rho–ROCK signaling, or overexpression of dominant negative form of Rho or ROCK, thus neglecting the possibility of functional interference due to non-specific inhibition on other closely-related members.
The zebrafish, Danio rerio, is an excellent vertebrate model for developmental and genetic studies owing to its external embryonic development, optical clarity and its amenability to specific gene knockdown by morpholino (MO) and functional rescue (Beis and Stainier, 2006). Extending our recent finding that RhoA links non-canonical Wnt signaling to the gastrulation movements in zebrafish (Zhu et al., 2006), we further explore the in vivo signaling mechanism of RhoA during post-gastrulation period of embryogenesis. Here, we report that knockdown of RhoA by its specific MOs causes marked increase in apoptosis during embryogenesis, leading to overall reduction in body size and length, which is closely associated with reduced phosphorylation status of Erk and a lack of expression of bcl-2. Using a combination of gene suppression and functional rescues together with specific pharmacological inhibitors for Mek and Bcl-2, we show that RhoA prevents intrinsic apoptosis during zebrafish embryogenesis by activation of Mek/Erk and Bcl-2 signaling pathways.
RhoA knockdown results in reduced body size and body length in zebrafish embryos
To study the specific role of RhoA in zebrafish embryogenesis beyond gastrulation, two non-overlapping translation-blocking MOs (rhoA MO1 and rhoA MO2) and one splice-blocking MO (E3I3 MO) were applied. During the whole embryogenesis, the standard control MO- or 5 bp-mismatch MO-injected embryos developed normally (Figures 1a, b, f, g, i, k, l, p and q). In contrast, embryos injected with rhoA MO1 and rhoA MO2 displayed convergence-extension defects at the end of gastrulation (Figures 1c and d) as we reported previously (Zhu et al., 2006) while embryos injected with E3I3 MO appeared normal (Figure 1e) but it still induced the ‘opaque’ phenotype in rhoA morphants after gastrulation (Figures 1o and t). During early somitogenesis, the apparent opaque region was observed in the head of rhoA morphants (Figure 1h), and cell corpuses were detected within or at the boundary of deformed somites (Figure 1j). By mid-segmentation stage, the opaque regions were found throughout the morphants, but mostly in head and tail (‘opaque’ phenotype) (Figures 1m–o). Till pharyngula stage, rhoA morphants exhibited shrinkage in the whole body, with severely reduced size in head and shortened body length including yolk extension and tail posterior to the yolk extension (Figures 1r–t). Most of the severely affected rhoA morphants died within 4 days. Some of mildly-defected embryos could survive longer, but they never reached the similar body size and length as the control morphants (data not shown).
RhoA knockdown induces apoptosis during zebrafish embryogenesis
It is reported that the opaque regions in the developing zebrafish embryos implicate cell death and could be a consequence of apoptosis (Langheinrich et al., 2002). To examine this, we analysed the phenotypic embryos by TUNEL assay. The increased apoptosis in rhoA morphants was first observed at the end of gastrulation (29/34, Figure 2b) while none was detected in control morphants (0/44, Figure 2a). During somitogenesis, only a few scattered apoptotic cells were distributed throughout the control morphants (18/21, Figure 2c), consistent with the observation by Cole and Ross (2001). In contrast, extent of apoptosis in rhoA morphants became more prominent during early somitogenesis (29/31, Figure 2d). By mid-segmentation stage, TUNEL-positive cells distributed throughout the whole embryo especially in head and tail (52/52, 33/35, 28/29, Figures 2f–h). And such increased apoptosis was maintained throughout somitogenesis (47/48, 18/18, 23/26, Figures 2j–l). By 2 dpf, the TUNEL-positive cells were detected in forebrain, midbrain, hindbrain (23/24, Figures 2n, p, r and t) and particularly concentrated in tissues known to be highly proliferative during embryogenesis, such as the posterior tectum (Figure 2p) and neural retina (Figure 2r) (Wullimann and Knipp, 2000). This could result in much smaller head and eyes than those in control morphants. In summary, the enhanced apoptosis correlated with the cell death observed in rhoA morphants, suggesting that the developmental defects in reduced body size and shortened body length was a consequence of increased apoptosis.
To verify the specificity of RhoA MOs, western blotting, semi-quantitative real-time polymerase chain reaction and rescue experiments were applied. As shown in Supplementary Figure S1a, the level of endogenous RhoA was dramatically reduced in either rhoA MO1- or rhoA MO2-injected embryos, indicating that RhoA translation can be successfully blocked by rhoA MO1 and 2. Moreover, the sequencing analysis (data not shown) and semi-quantitative RT–PCR showed that E3I3 MO resulted in a frame shift and a premature stop-codon in the transcript, leading to premature termination of RhoA synthesis (Supplementary Figure S1b–c). Furthermore, forced expression of rhoA mRNA which had no complementary sequence to rhoA MO1 can correct both developmental defects and enhanced apoptosis in rhoA morphants (see later in Figure 4). In addition, the pattern of apoptotic cells in rhoA morphants partially overlaps with that of rhoA expression pattern that we reported previously (Zhu et al., 2006), and the abnormal developmental defects in rhoA morphants were distinct from those induced by Ras knockdown (Liu and Low, unpublished data). All these data support that the increased apoptosis in rhoA morphants was due to loss of RhoA function and not due to unspecific toxicity from the MOs used. Taken together, all these data suggest that RhoA knockdown could induce apoptosis during zebrafish embryogenesis, resulting in reduction of body size and shortening in body length. Since all the specific rhoA MOs induced very similar effects, rhoA MO1 was used for the subsequent studies.
RhoA knockdown inhibits Mek/Erk activation
Ras–MAPK pathway regulates multitude of cellular processes, including cell survival, gene expression and cell proliferation (Giehl, 2005). To investigate whether the regulation of cell survival by RhoA during early embryonic development is through the cross-talk with Ras–MAPK pathway, western analysis was first performed. As shown in Figure 3a, the level of phospho-Erk (an indicator of Mek/Erk activation) was significantly reduced in rhoA morphants, while the total Erk level remained unchanged. Consistently, the phosphorylation of Mek was also decreased in rhoA morphants (data not shown). In contrast, neither phospho-p38 nor total p38 showed detectable changes in rhoA morphants. These results suggest that RhoA could be important for the activation of Mek/Erk pathway during zebrafish embryogenesis.
To determine whether the inactivation of Mek/Erk pathway was responsible for the developmental defects and enhanced apoptosis in rhoA morphants, a series of rescue experiments were performed. In rhoA MO1-injected group, 88.3% of embryos (n=149) showed ‘opaque’ phenotype with reduced body size and shortened body length during somitogenesis (around 15–18 somites stage; Figures 4b and k). In contrast, such developmental defects were corrected in 37.6% of embryos (Figures 4c and k, n=156, at P<0.05) and the induced apoptosis was dramatically decreased in those phenotypically rescued embryos (42/44, Figure 4h) after the forced expression of rhoA mRNA. Although the difference between RhoA knockdown and rhoA mRNA rescue was statistically significant at P<0.05, the percentage of rescue was relatively low. This could be due to the sensitivity of embryos to the co-injection of rhoA mRNA with rhoA MO1, or the early onset of translation of the rescue mRNA during the zygote that could derail the proper developmental process. Alternatively, RhoA was ubiquitously expressed during early zebrafish embryogenesis (Zhu et al., 2006) where the mosaic nature of the mRNA and MO injections as reported by others (McClintock et al., 2002; McWhorter et al., 2003) may also affect the efficiency of rescue. Furthermore, to avoid the overexpression phenotype or possible toxicity of high expression of rhoA, sub-optimal dose of mRNA was applied, and this level might not be sufficient to fully restore normal development to all the rhoA morphants. In addition to rhoA mRNA rescue, ectopic expression of Mek mRNA not only restored normal development in 28.9% of the embryos (Figures 4d and k, n=161, at P<0.05), but also prevented apoptosis in 93.6% phenotypically corrected embryos (n=47, Figure 4i). To further confirm that Mek/Erk acts as downstream of RhoA to promote cell survival, the Mek inhibitor (U0126; Hong et al., 2006) was used to block the activation of Mek/Erk in the embryos which were injected with rhoA MO1 alone, or co-injected with rhoA MO1 and mRNA encoding either Mek or rhoA. As shown in Supplementary Figure S2a, rhoA mRNA rescue could be inhibited by U0126 in a dose-dependent manner, while no inhibition of such rescue was seen in embryos treated with JNK inhibitor (SP600125; Supplementary Figure S2b). And compared to the group injected with rhoA MO1 alone (10.6% normal, n=251), U0126 (at the non-toxic concentration) abrogated both rhoA mRNA (from 40.5% normal, n=219, to 15.2% normal, n=240) and Mek mRNA rescue (Figure 4l, from 34.9% normal, n=250, to 15.1% normal, n=248, at P<0.05). Furthermore, combined application of rhoA MO1 (at low concentration) that would only cause mild developmental defects in few injected embryos with U0126 (at the non-toxic concentration) could dramatically induce ‘opaque’ phenotype and increased apoptosis similar to RhoA effective knockdown (Supplementary Figures S3i and k). Taken together, these results demonstrate that RhoA could prevent apoptosis during zebrafish embryogenesis at least through the Mek/Erk pathway.
RhoA knockdown suppresses bcl-2 expression
In addition to activation of Mek/Erk pathway, RhoA may also promote cell survival through antagonizing the intrinsic apoptotic pathway. To test this, the protein level of Bcl-2, a critical anti-apoptotic gatekeeper for the intrinsic mitochondria pathway, was examined in rhoA morphants. Compared to control morphants, Bcl-2 level was significantly reduced in rhoA morphants (Figure 3b). Consistently, ectopic expression of mRNA encoding human BCL-2 could correct both developmental defects in 29.1% embryos (Figures 4e and k, n=186, at P<0.05) and increased apoptosis in 97.7% phenotypically rescued embryos (n=43, Figure 4j). Besides, antagonizing the anti-apoptotic function of Bcl-2 by its pharmacological inhibitor, HA 14-1 (Wang et al., 2000; at non-toxic concentration), effectively abolished all functional rescues elicited by rhoA (from 41.5% normal, n=149, to 21.9% normal, n=133, at P<0.05), Mek (from 35.7% normal, n=135, to 16.1% normal, n=106, at P<0.05) and BCL-2 mRNA (Figure 4m, from 31.6% normal, n=214, to 17.5% normal, n=168, at P<0.05). However, Mek inhibitor, U0126, had no effect on the BCL-2 mRNA rescue (Figure 4l, from 28.2% normal, n=144, to 27.6% normal, n=95). Moreover, combined treatment of HA 14-1 with injection of rhoA MO1 at low concentration resulted in ‘opaque’ phenotype, overall reduction of body size and body length, and enhanced apoptosis in 51.1% embryos (Supplementary Figures S3j, l and m, n=141, P<0.05). Supporting this, high dose of HA 14-1 could induce very similar developmental defects as RhoA knockdown (Supplementary Figure S5c). Taken together, these results suggest that Bcl-2, albeit mechanism unknown, act as downstream response to anti-apoptotic signal from RhoA-Mek/Erk during zebrafish embryogenesis.
RhoA controls cell survival via Mek/Erk activation during embryogenesis
Rho GTPases are key regulators for cytoskeletal dynamics, but their involvement in executing survival control during animal development is less well understood. Our present study shows that knockdown of RhoA in zebrafish causes increased apoptosis, which is at least resulted from the reduced activation of Mek/Erk pathway. Consistently, Erk2 knockout mice also show enhanced apoptosis (Yao et al., 2003). However, knockout of Erk leads to the loss of mesoderm, whereas suppression of RhoA signalling does not affect the cell fate determination in both zebrafish and Xenopus (Tahinci and Symes, 2003; Zhu et al., 2006). Similarly, high dose of U0126 can induce embryonic lethality in the majority of treated zebrafish embryos and severe apoptosis in few survivors (Supplementary Figure S5b). These suggest that strong inhibition on Mek/Erk signaling could cause more severe embryonic defects than RhoA knockdown, reflecting broader biological roles of Mek/Erk in cell fate determination, gene transcription and cell proliferation (Giehl, 2005). In addition, the genetic link between RhoA and Mek/Erk pathway can be further supported by their reminiscent expression domain during embryonic development, for example, rhoA (Zhu et al., 2006) and erk (Krens et al., 2006) are both highly expressed in forebrain, midbrain–hindbrain boundary, hindbrain, eyes and tail region during zebrafish embryonic development. And, the similar expression profile of RhoA and the phosphor-Erk are very conserved from fly to vertebrate (Gabay et al., 1997; Corson et al., 2003), suggesting that functions of these proteins in embryogenesis could be genetically linked during evolution.
RhoA prevents apoptosis via activation of Mek/Erk and Bcl-2 pathways
Bcl-2 plays a central role in inactivating the intrinsic apoptotic pathway by sequestering pro-apoptotic Bcl-2 family proteins (such as Bax) and blocking their targeting to mitochondria for cytochrome c release. Increasing evidence has shown that RhoA and Mek/Erk signalings are essential for cell survival by upregulating the expression of Bcl-2 in variety of cell lines (Navarro et al., 1999; Rios-Munoz et al., 2005). Also ERK1/2 has been implicated to phosphorylate Bcl-2, leading to its full potency in anti-apoptotic function in murine IL-3-dependent myeloid cell lines (Deng et al., 2001). Our study provides the first evidence in vivo that the inactivation of Mek/Erk pathway and reduced Bcl-2 protein level caused by RhoA knockdown are closely associated with increased apoptosis in zebrafish rhoA morphants. However, involvement of other anti-apoptotic or pro-apoptotic Bcl-2 family members in RhoA-mediated cell survival during zebrafish embryogenesis cannot be completely ruled out. This is because the pharmacological inhibitor, HA 14-1, may also inhibit Bcl-2 close homolog, such as Bcl-xL (Doshi et al., 2006). Besides, RhoA and Mek/Erk have also been implicated to control the expression of other anti-apoptotic Bcl-2 family members, such as Bcl-xL, myeloid cell leukemia-1 (Mcl-1) and pro-apoptotic Bcl-2 family proteins including Bim, Bad and Bax, in various cell lines (Rios-Munoz et al., 2005; Del Re et al., 2007). Nonetheless, our study supports the notion that RhoA-Mek/Erk signaling prevents intrinsic apoptosis in vivo at least through upregulating the expression of bcl-2 (Figure 5).
Actin dynamics control by RhoA as a possible link to apoptosis
In addition to regulating the ratio of anti-apoptotic and/or pro-apoptotic Bcl-2 family proteins, the alteration of actin dynamics induced by RhoA knockdown may also contribute to the apoptosis. It has been reported that an intact cytoskeleton mediated by Rho/ROCK is necessary for overall ERK1/2 activation and their nuclear translocation, as well as activation of transcription in SDF-1-stimulated cells (Zhao et al., 2006), while disruption of actin cytoskeleton can induce apoptosis mediated by Bcl-2 in MEK-transformed EpH4 and MCF10A mammary epithelial cells (Martin and Leder, 2001; Pinkas et al., 2004). On the contrary, stabilization of actin cytoskeleton by downregulation of the actin severing protein, gelsolin (Harms et al., 2004) or by addition of jasplakinolide that induces large F-actin aggregates can also lead to increased apoptosis (Posey and Bierer, 1999). These data suggest that alterations in actin dynamics, either disruption or stabilization of actin cytoskeleton, could result in apoptosis. As such, this may help to explain the paradox that both downregulation and overexpression of RhoA could cause apoptosis in different cellular contexts.
Cell survival is uncoupled from gastrulation control by RhoA
Besides the prevention of cell death, cytoskeleton rearrangement regulated by RhoA during cell movements may also be mediated by Mek/Erk and Bcl-2 pathways. It has been shown that Mek induces phosphorylation of myosin light chains during cell motility (Klemke et al., 1997), while overexpression of Bcl-2 inhibits actin depolymerization to promote cell migration in myelocytic cell lines (Korichneva and Hammerling, 1999). Moreover, overexpression of Mek or Bcl-2 enhances the migration of endothelial cells (Rikitake et al., 2000), bladder cancer cells (Miyake et al., 1999) and glioma cells (Wick et al., 1998). In comparison, our study shows that forced expression of either Mek or BCL-2 can correct gastrulation defects in rhoA morphants (Supplementary Figure S4), suggesting Mek and Bcl-2 could act as downstream of RhoA to control cell movement during zebrafish gastrulation. However, results from our splicing MO experiments revealed that knockdown of zygotic RhoA exhibited only reduced body size/length with increased apoptosis without any gastrulation movement defects. Therefore, these results imply that although these three proteins are important in both processes, the mechanisms of their actions during convergence-extension movement and cell survival are likely to be uncoupled.
In summary, we have uncovered the first genetic link in vivo between RhoA and Mek/Erk signaling as well as Bcl-2 where RhoA prevents Bcl2-dependent intrinsic apoptosis via activation of Mek/Erk pathway. This could pave the way to our better understanding of regulation for apoptosis by these key proteins and their related members during normal development and pathophysiological conditions.
Materials and methods
Maintenance and breeding of zebrafish
Fish were purchased from local supplier and raised under standard laboratory conditions (Westerfield, 2000). Embryos were collected and fixed at different stages based on the standard morphological criteria (Kimmel et al., 1995).
Morpholino injections and mRNA rescues
Two non-overlapping anti-sense MOs against 5′UTR (rhoA MO1: 5′-TCCGTCGCCTCTCTTATGTCCGATA-3′) or translation start-site of zebrafish rhoA gene (rhoA MO2: 5′-CTTCTTGCGAATTGCTGCCATTTTG-3′), one splicing MO targeting the rhoA splice donor site of exon 3 (E3I3 MO: 5′-ACACCAAAGAGCATTCTTACTAAAC-3′) and one standard control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′) as well as one 5-bp mismatch MO of rhoA MO1 (5′-TCgGTCcCCaCTCTaATGTCgGATA-3′; mismatches in lower case) were synthesized (Gene Tools). All the MOs were resuspended in 1 × Danieau buffer (58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO4, 0.6 mM Ca(NO3)2 and 5.0 mM HEPES pH 7.6), and injected into embryos at one-cell stage. For effective functional knockdown, the dose of MOs injected per embryo are as follows, rhoA MO1 (2.4 ng), rhoA MO2 (5.7 ng), E3I3 MO (5.7 ng), 5-bp mismatch MO (5.7 ng) and standard control MO (5.7 ng). In the combination of gene suppression with inhibitors treatment, 1.15 ng of rhoA MO1 or control MO was applied per embryo. For the rescue experiments, capped full-length mRNA encoding zebrafish rhoA (PCS2+-rhoA), mouse prototype Mek1 (pXJ40-mMek1) and human BCL-2 (pRCCMV-hBCL-2) were synthesized in vitro using the mMessage mMachine Kit (Ambion, Foster City, CA, USA). Each of these mRNAs was titrated and co-injected with rhoA MO1 (2.4 ng) at one-cell stage. The sub-optimal dose for rhoA, Mek1 or BCL-2 mRNA which could effectively prevent RhoA knockdown defects was 11.25, 60 and 5 pg, respectively.
Pharmacological inhibitors treatments and statistical analysis
Mek inhibitor (U0126, Promega, Madison, WI, USA), Jnk (SP600125, Calbiochem, San Diego, CA, USA) and Bcl-2 inhibitors (HA 14-1, Calbiochem) were dissolved in dimethyl sulfoxide as 10 mM stock solution. They were titrated on wild-type embryos before applied on morphants, for example, U0126 (5–40 μM), SP600125 (0.3–1.2 μM) and HA 14-1 (0.5–2.5 μM). The optimal concentrations were chosen whereby the corresponding pathways could be effectively inhibited without generating excessive and non-specific global defects (suggesting no toxicity). Embryos at 2.5 hpf were emerged into the egg water with inhibitor and continuously incubated till 15–18 somites stage, then they were either collected for western analysis and microscopy or fixed for apoptosis assay. For statistical analyses, data are presented as means±s.d. Each experiment was repeated at least three times. The statistically significant differences in mean values were assessed with the two populations (paired) t-test (Origin Pro 6.1).
Cryostat section and apoptosis assay
Embryos injected with control MO or rhoA MOs were fixed in 4% paraformaldehyde at different developmental stage (from 7 hpf to 2 dpf) overnight at 4 °C. For cryostat section, embryos at 2 dpf were embedded in OCT as described in the zebrafish book (Westerfield, 2000) and 6-nm cryosections were made on a CM 1900 cryostat (Leica, Germany). Both embryos and cryosections were incubated in acetone at −20° C for 10 min, followed by two 5–10 min rinses in PBS. Then they were subjected to TUNEL assay (terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling; Clontech, Mountain View, CA, USA) according to the manufacturer's instructions.
Pooled embryos injected with control MO or rhoA MOs were dechorionated and deyolked manually at 18-somite stage, and lysed in RIPA buffer (1% NP-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate (SDS) in PBS) containing protease inhibitor cocktail (Roche, Switzerland). After determination of protein concentration (Bradford's reagent), 100 μg of protein extract was loaded onto 15% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Western analysis was performed with the primary antibodies (1/1000 diluted in TBST/BSA) against RhoA, BCL-2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), phospho-ERK1/2 (Sigma, St Louis, MO, USA), total ERK, phospho-p38, total p38, β-Tubulin (Cell Signaling Technology, Danvers, MA, USA).
This work was supported by a Graduate Research Scholarship awarded to SZ and grant from the Biomedical Research Council of Singapore.