Directed cell migration is crucial for development, but most of our current knowledge is derived from in vitro studies. We analyzed how neural crest (NC)cells migrate in the direction of their target during embryonic development. We show that the proteoglycan Syndecan-4 (Syn4) is expressed in the migrating neural crest of Xenopus and zebrafish embryos. Loss-of-function studies using an antisense morpholino against syn4 show that this molecule is required for NC migration, but not for NC induction. Inhibition of Syn4 does not affect the velocity of cell migration, but significantly reduces the directional migration of NC cells. Furthermore, we show that Syn4 and PCP signaling control the directional migration of NC cells by regulating the direction in which the cell protrusions are generated during migration. Finally, we perform FRET analysis of Cdc42, Rac and RhoA in vitro and in vivo after interfering with Syn4 and PCP signaling. This is the first time that FRET analysis of small GTPases has been performed in vivo. Our results show that Syn4 inhibits Rac activity, whereas PCP signaling promotes RhoA activity. In addition, we show that RhoA inhibits Rac in NC cells. We present a model in which Syn4 and PCP control directional NC migration by, at least in part,regulating membrane protrusions through the regulation of small GTPase activities.
Considerable progress has been made recently in understanding cell migration, which is crucial to the comprehension of normal and disease-related processes, such as morphogenesis and metastasis in cancer. However, most of these studies have been performed in vitro, and some discrepancies have been found between cell behaviour in vivo and in vitro(Even-Ram and Yamada, 2005). Here, we use the migration of an embryonic tissue, the neural crest (NC), as an in vivo model to study cell migration.
The NC has been called `the explorer of the embryo' because of its inherent migratory abilities. NC cells migrate from the dorsal neural tube, covering extremely long distances and colonizing almost all the tissues of the embryo. Upon reaching their destination, they differentiate into a wide range of cell types, including neurons, glial cells, skeletal and connective tissue, and adrenergic and pigment cells (LeDouarin and Kalcheim, 1999).
The migration of the NC is a highly ordered process; individual NC cells migrate with high persistence towards the direction of their targets(Teddy and Kulesa, 2004), but it is not known how this directionality is controlled. A number of molecules have been identified as being key players in neural crest migration (for a review, see Kuriyama and Mayor,2008). However most of these molecules function as inhibitory signals, which are required to prevent the migration of NC cells into prohibited areas. Although chemoattraction has been one of the proposed mechanisms to explain this directional migration, no chemoattractant has thus far been found in the NC. This prompted us to look for alternative mechanisms that might generate directional migration. Interestingly, researchers studying cell migration in vitro have observed that cultured cells can migrate with a high directionality even in the absence of external chemoattractants. It is known that in vitro cell migration requires the formation of membrane protrusions at the leading edge of the cell, membrane adhesive interactions with the substrata and the coordinated dynamics of the cytoskeleton(Lauffenburger and Horwitz,1996; Pollard and Borisy,2003; Ridley et al.,2003; Sheetz et al.,1999). Small GTPases (Rac, Rho and Cdc42) are well-known modulators of several of these activities (for reviews, see Ridley et al., 2003; Jaffe and Hall, 2005). Moreover, it has been shown that directional migration in vitro in the absence of extrinsic chemoattractants is controlled by the level of Rac activity(Pankov et al., 2005). Rac promotes the formation of peripheral lamella during random migration, while slightly lower levels of Rac suppress peripheral lamella and favour the formation of a polarized cell with lamella just at the leading edge(Pankov et al., 2005).
Syndecan-4 (Syn4) is a proteoglycan that is involved in the migration of cells cultured in vitro, and it has been proposed as a key regulator of RhoA and Rac activities, focal adhesion formation and planar cell polarity (PCP)signaling (for a review, see Alexopoulou et al., 2007). In this study, we examined the role of Syn4 in neural crest migration in Xenopus and zebrafish embryos. We show that syn4 is expressed specifically in the migrating neural crest and that it is essential for its migration. In addition, we show that Syn4 controls directional migration by regulating the polarized formation of cell protrusions, in a manner similar to non-canonical Wnt signaling. In order to understand the molecular mechanism by which Syn4 and planar cell polarity(PCP) signaling control the orientation of cell protrusions, we performed fluorescence resonance energy transfer (FRET) analysis to measure the activity of the small GTPases, Cdc42, RhoA and Rac. This is the first time that this kind of FRET analysis has been carried out in vivo. Our results indicate that whereas Syn4 inhibits Rac activity, PCP signaling activates RhoA. In addition,we show that RhoA, through Rock, is an inhibitor of Rac activity in the neural crest. Thus, the convergence of Syn4 and PCP signaling through the regulation of small GTPases contributes to the directional migration of neural crest cells.
MATERIALS AND METHODS
Xenopus and zebrafish embryos, micromanipulation and graft experiments
Xenopus embryos were obtained and dissections carried out as previously described (Mancilla and Mayor,1996). Embryos were staged according to Nieuwkoop and Faber(Nieuwkoop and Faber, 1967). Zebrafish strains were maintained and bred according to standard procedures(Westerfield, 2000). Transgenic sox10:egfp (Carney et al., 2006) embryos were obtained by crossing heterozygous adults. Xenopus NC grafts were carried out as described by De Calisto et al.(De Calisto et al., 2005). Zebrafish cell transplants were performed according to Westerfield(Westerfield, 2000); sox10:egfp donor embryos were injected at the one- or two-cell stage with a mixture of tetramethylrhodamine dextran (RDX) (1 ng/nl; Molecular Probes) and either control or syn4 morpholino oligonucleotides (MOs). At the sphere stage, single or small groups of around 10 donor cells were transplanted into the apical region of unlabelled wild-type host embryos using an oil-filled manual injector (Sutter Instrument Company). Embryos were cultured until 12 somites and observed using time-lapse microscopy (Leica DM5500).
Whole-mount in situ hybridization and cartilage staining
In situ hybridization was carried out according to Harland(Harland, 1991) using digoxigenin-labelled antisense RNA probes (Roche Diagnostics). Probes used were: Xenopus syn4 (Munoz et al.,2006); and zebrafish syn4, snail2(Mayor et al., 1995), fli (Meyer et al.,1995), foxd3 (Kelsh et al., 2000) and crestin(Luo et al., 2001). For cartilage staining, 5-dpf zebrafish were stained according to Barrallo-Gimeno et al. (Barrallo-Gimeno et al.,2004).
RNA synthesis and morpholino microinjection
cDNA was linearized and RNA was synthesized using the mMessage mMachine Kit(Ambion), according to the manufacturer's instructions. mRNA and MOs were co-injected into Xenopus embryos with fluorescein dextran (FDX,Molecular Probes) at the eight- or 32-cell stage(Aybar et al., 2003). For zebrafish microinjection, 4 nl was injected at the one- or two-cell stage. The mRNA constructs used were: Xenopus syn4(Munoz et al., 2006); and zebrafish syn4, DshDEP+ and DshΔN(Tada and Smith, 2000), and mutant syn4. Two translation-blocking MOs against zebrafish syn4 were designed over the 5′UTR region: syn4 MO1,5′CGGACAACTTTATTCACTCGGGCTA3′; and syn4 MO2,5′GAGAAG(ATG)TTGAAAGTTTACCTCA3′. As both MOs produced the same phenotype, we used mainly syn4 MO1 (called syn4 MO), except in some experiments where syn4 MO2 or a mixture of both MOs was used,as indicated in the text and figure legends. A standard control MO was used:5′CCTCTTACCTCAGTTACAATTTATA3′. Injection of this control MO into wild-type zebrafish embryos caused no defective phenotype. Two translation-blocking MOs against Xenopus syn4 were used: syn4 MO1, 5′GCACAAACAGCAGGGTCGGACTCAT3′; and syn4 MO2, 5′CTAAAAGCAGCAGGAGGCGATTCAT3′(Munoz et al., 2006). Throughout this work, a 1:1 mixture of both MOs called syn4 MO was used. A 5-base mismatched MO against Xenopus syn4 was used as a control (5′GCAGAAAGATCAGCGTCCGACTGAT3′). The other MO used was directed against wnt5a (Lele et al., 2001). Unless stated otherwise, 6 ng of MO was used for zebrafish and 8 ng for Xenopus.
For the mutation in the PKCα-binding site of Syn4 (called Syn4*), a mutation was introduced in the PIP2-binding site that enables interaction with PKCα. Amino acid residues Y185KK were changed to LQQ using PCR with mutated primers. According to Horowitz et al., this mutation reduces the affinity of PIP2 binding to Syn4 (Horowitz et al., 1999). We observed that this mutant has the same activity as wild-type Syn4 in a neural plate induction assay (our unpublished results).
sox10:egfp was used to analyze NC migration in vivo(Carney et al., 2006). Embryos were processed as described by Westerfield(Westerfield, 2000). Each embryo was staged according to the number of somites and only embryos with equal numbers of somites were compared. The embryos were dechorionated,inserted into a drop of 0.20% agarose in embryo medium(Westerfield, 2000) and mounted in a custom-built chamber. Control and experimental embryos were mounted side-by-side in the same chamber. A compound (Leica DM5500) or a confocal (Leica SP2-DMRE) microscope was used for time-lapse imaging. Digital images were typically collected at 30 to 90 second intervals for a period of between 1 and 14 hours. We performed z-stack in preliminary experiments to establish how deep the NC migrates in the embryo. After 6- to 8-hour time-lapse imaging of 20-somite embryos, we found that cephalic NC cells migrate between 500 and 800 μm in the anteroposterior axis, between 40 and 60 μm in the dorsoventral axis, and between 7 and 9 μm in the periphery-center axis. Consequently, for tracking analysis, we can assume that most of the cell migration is performed in two dimensions; the third dimension(in the z-axis when the embryo has a lateral orientation) can be neglected.
Sequences of images were quantitatively analyzed using the public domain program NIH ImageJ (developed at the US National Institutes of Health) and Matlab (MathWorks). Tracking of individual cells was used to calculate velocity (total distance traveled divided by time), persistence (defined as the ratio between the linear distance from the initial to the final point and the total length of the migratory path) and the angle of migration (with respect to its previous position).
The shape of individual cells was analyzed using NIH ImageJ. Thresholds were fixed at the same value for control and syn4 MO-injected cells. The outline/analyze particles function was used to draw the contour of each cell at different time points, which were overlapped maintaining the original XY positions. Two independent methods were used to analyze cell protrusions. For the first method, we defined Cell extension (CE) as the new positive area between two consecutives frames (separated by the shortest time of 1 minute in the time-lapse analysis). During the course of one minute, the body (and centroid) of the cell does not move a significant distance and most of the new area generated corresponds to lamellipodia extension. Note, fillopodia were not considered in this analysis as they move faster and the intensity of fluorescence is weaker. Using ImageJ, we subtracted two consecutives frames,in a manner that the new growing area was shown in red and the unchanged area in white (Fig. 6C,H,M). The centroid (defined as the average of the x and y coordinates of all the cell pixels) was calculated and a vector between the centroid (x)and the center of the red area was drawn (arrow in Fig. 6N). These vectorial data were used to analyze the distribution of CE orientation under different conditions. As a second method to estimate cell protrusion, we measured the Cell Smoothness (CS), defined as the ratio between the perimeter of an ideal ellipse-shaped cell and the actual perimeter of the cell. The ellipse was the best-fit ellipse and we used the standard built-in ImageJ function. This value gives us an unbiased measure of how folded a cell is (i.e. how many protrusions a cell has). P-values were obtained using a one-way analysis of variance (ANOVA). All statistical analyses and their graphical illustrations were performed on Matlab, using both built-in functions and customized scripts (available from the authors on request).
Fluorescence (Förster) resonance energy transfer (FRET)
Plasmid DNA encoding FRET probes [Raichu-Rac, Raichu-Cdc42(Itoh et al., 2002) and RhoA biosensor (Pertz et al.,2006)] was injected directly into Xenopus embryos at the eight-cell stage and NCs were dissected at stage 15. For in vitro analysis, NC cells were cultured on fibronectin, as described by Alfandari et al.(Alfandari et al., 2003), and fixed with 4% PFA after 5-7 hours migration. For the in vivo analysis,injected neural crests were grafted into wild-type hosts. Embryos were fixed in MEMFA at stage 26, and 12 μm cryostat sections were taken.
Samples for FRET analysis were imaged using a Zeiss lSM 510 META laser scanning confocal microscope and a 63× Plan Apochromat NA 1.4 Ph3 oil objective. The CFP and YFP channels were excited using the 405-nm blue diode laser and the 514-nm argon line, respectively. The two emission channels were split using a 545-nm dichroic mirror, which was followed by a 475-525 nm bandpass filter for CFP and a 530 nm longpass filter for YFP. Pinholes were opened to give a depth of focus of 3 mm for each channel. Scanning was performed on a line-by-line basis with the zoom level set to two. The gain for each channel was set to approximately 75% of dynamic range (12-bit, 4096 gray levels) and offsets set such that backgrounds were zero. The time-lapse mode was used to collect one prebleach image for each channel before bleaching with 50 scans of the 514 nm argon laser line at maximum power (to bleach YFP). A second post-bleach image was then collected for each channel. Pre- and post-bleach CFP and YFP images were imported into Mathematica 5.2 for processing. Briefly, images were smoothed using a 3×3 box mean filter,background subtracted and post-bleach images fade compensated. A FRET efficiency ratio map over the whole cell was calculated using the following formula:(CFPpostbleach-CFPprebleach)/CFPpostbleach. Ratio values were then extracted from pixels falling inside the bleach region,as well as an equal-sized region outside of the bleach region, and the mean ratio was determined for each region and plotted on a histogram. The non-bleach ratio was then subtracted from the bleach region ratio to give a final value for the FRET efficiency ratio. Data from images were used only if YFP bleaching efficiency was greater than 70%.
Syn4 is expressed in the NC and is required for NC development
The expression of zebrafish and Xenopus syn4 was compared with that of specific neural crest markers. In both species, syn4 is expressed in the NC as soon as the cells start to migrate(Fig. 1A,G). In zebrafish, the migrating NC can be recognized at 20 hpf as disperse cells in the head and as streams of cells migrating through the somites in the trunk(Fig. 1B,C). This pattern of expression is indistinguishable from that of the NC marker crestin(Fig. 1E,F). Visualization of trunk NC is more difficult in Xenopus, but a similar pattern of expression of syn4 was observed in the migrating cephalic NC(Fig. 1G-I), where NC cells were identified with the migratory NC markers twist and fli(Fig. 1J-L). The early expression of syn4 in Xenopus does not delineate the NC as precisely as do the NC markers (Fig. 1G,J), which suggests that syn4 is also expressed in cells adjacent to the migrating NC.
This highly localized expression of syn4 prompted us to analyze its function in NC development. Two morpholino antisense oligonucleotides(MO1, MO2) were developed to inhibit zebrafish syn4. We tested the efficiency of MO1 by analyzing its ability to reduce the fluorescence of a Syn4-GFP fusion protein (Fig. 2A-C). Injection of syn4 MO1 or syn4 MO2 into zebrafish produces a dramatic reduction of neural crest derivatives, including cartilage (Fig. 2D,E,G,H; see also Fig. S1 in the supplementary material) and melanocytes(Fig. 2F,I; Fig. S1 in the supplementary material). We ruled out a possible role for Syn4 in NC specification by analyzing the expression of early NC markers. No effect on the expression of NC markers was seen in Xenopus(Fig. 2J) or zebrafish(Fig. 2K,L) embryos injected with syn4 MOs. In addition, when either control ectoderm, or ectoderm injected with syn4 MO, was combined with dorsolateral mesoderm to test for induction of the NC, no difference was observed(Fig. 2M-O).
Syn4 is required for NC migration
Next, we analyzed whether Syn4 is involved in NC migration. Migration of cephalic neural crest can be recognized in Xenopus as three streams of cells (Fig. 3A), whereas the migrating trunk NC can be visualized in zebrafish as streams of cells in the somites (Fig. 3F). Injection of syn4 MO or mRNA leads to a strong inhibition of cephalic and trunk NC migration (Fig. 3B,C,G,H). Co-injection of syn4 mRNA, which does not contain the MO-binding site, together with the MO can rescue NC migration, demonstrating that the MO specifically targets syn4 (Fig. 3D,E,I,J). Importantly, however, overexpression of syn4carrying a mutation in the PKCα-binding site does not rescue NC migration in syn4-MO-injected embryos(Fig. 3E, two last columns),which is different to what has been reported for the effect of this mutation in cells cultured in vitro (Bass et al.,2007).
Our results show that syn4 MO affects NC migration; however, it is also possible that the MO interferes with NC migration in a non-cell-autonomous manner, as it is known, for example, that syn4 MO affects convergent extension of mesoderm(Munoz et al., 2006) (L.M. and R.M., unpublished). Two experiments were performed to analyze this possibility. First, we injected zebrafish embryos with a MO against has2, the synthesizing enzyme of Hyaluronan, which has been described to perturb convergent extension (Bakkers et al., 2004). Injection of the has2 MO produced a strong phenotype in convergent extension, as previously described, but no difference in the migration of NC between control and has2 MO-injected embryos was observed (see Fig. S2A,B in the supplementary material). A second set of experiments were performed to show that syn4 MO inhibits cell migration in a cell-autonomous manner. Grafts of fluorescein dextran (FDX)- or syn4 5-base mismatched MO (syn4 5mm MO)-injected Xenopus NC into control embryos(Fig. 4A) show normal migration(Fig. 4B,C; uninjected, 87%grafts migrated, n=15; syn4 5mm MO, 75% grafts migrated, n=12); however, grafts of NC injected with syn4 MO exhibit a complete inhibition of migration (Fig. 4D; 0% grafts migrated, n=15). This result suggests that Syn4 is required in a cell-autonomous manner for NC migration. However, the NC graft contains a large number of cells that could affect each other in a non-cell-autonomous manner; therefore, true cell-autonomy can only be examined by grafting single cells. As it is technically difficult to do this in Xenopus, we used zebrafish embryos for this experiment, using a transgenic line that expresses GFP only in NC cells (sox10:egfp)(Carney et al., 2006). We grafted either NC cells injected with the syn4 MO into wild-type embryos (Fig. 4E) or wild-type NC cells into embryos injected with the MO(Fig. 4F). Fig. 4G-N shows the GFP-positive cells after 4 hours of migration. The average distance traveled in 4 hours by several grafts is shown in Fig. 4O,P. The control MO did not inhibit NC migration (Fig. 4O,P; compare position at time 0, indicated by the black arrow in Fig. 4G-J with the position 4 hours later, white arrow in Fig. 4H,J). However, syn4 MO had a significant effect on NC migration, whether present in the grafted NC or in the host(Fig. 4K-P). These results suggest that Syn4 is required autonomously in the NC to control its migration,but that an interaction with other NC cells expressing syn4 is also required.
Syn4 controls the directionality of NC migration and the orientation of cell protrusions
As the migration of NC cells is a complex process involving an early delamination step followed by active cell migration, we explored which step is affected by syn4 MO. We used the sox10:egfp zebrafish transgenic line to visualize migrating NC cells(Carney et al., 2006).
The movement of individual cephalic NC cells was followed(Fig. 5A-D; similar results were seen with trunk NC, data not shown) and their migration path was tracked(Fig. 5E). A strong inhibition of the migration of NC cells in the syn4 MO-injected embryos was observed (Fig. 5F-I; see also Fig. S3 in the supplementary material); the cells were motile, but their overall migration was reduced, as shown by the cell tracks(Fig. 4J). We confirmed that these effects were not due to a delay in cell migration by analyzing embryos at later stages of development (see Movie 1 in the supplementary material). Moreover, these data suggest that delamination and cell motility are not inhibited, as cells can migrate as individuals. Tracks of individual cells(Fig. 5K) showed no significant difference in the velocity of migration between control and syn4 MO cells (Fig. 5L; P=0.2276, n=15). However, the directionality of migration[measured as persistence (the linear displacement of the cell divided by the total distance traveled)] was significantly affected by the syn4 MO(Fig. 5M; P=0.0049, n=15). The distribution of angles of migration for each individual cell at each time point also demonstrated a significant difference between control MO (Fig. 5N,O) and syn4 MO (Fig. 5N,P; P=0.0044, n=665) cells. Taken together, these results indicate that the syn4 MO affects NC migration by interfering with the directionality of migration. As far as we know, this is the first time that a specific effect on the directional migration of NC cells has been described during development.
As persistence of migration usually depends on the orientation of cell protrusions (Ridley et al.,2003), we analyzed whether cell protrusions are affected by the syn4 MO. In Fig. 6A,B,the NC cells migrating around the optic vesicle (mandibular stream) are shown at two time points. Strikingly, the cell morphologies observed in control MO-(Fig. 6A,B,D; Movie S2 in the supplementary material) and syn4 MO-(Fig. 6F,G,I; Movie S3 in the supplementary material) injected cells are very different; control cells are elongated along the axis of migration, whereas syn4 MO-injected cells are more rounded. In order to quantify these differences, we measured two different parameters. First, we measured cell smoothness (CS; defined as the ratio between the perimeter of an ideal ellipse-shaped cell and the actual perimeter of the cell, which gives us an unbiased measure of how folded a cell is, i.e. how many protrusions a cell has). A significant difference in CS was observed between the control and syn4 MO(Fig. 6K,L; P<0.0005, n=59). A second parameter related to cell protrusions that we call cell extension (CE) was also measured(Fig. 6M). We define CE as the new positive area between two consecutives frames (separated by 1 minute). During the course of a minute, the body (and centroid) of the cell does not move a significant distance and careful comparison of the morphology of the cell between consecutive frames strongly suggests that CE corresponds to cell protrusions, such as lamellipodia (Fig. 6C,M). However, fillopodia cannot be observed, as they move faster and the intensity of fluorescence is weaker. Comparison with control and syn4 MO-injected NC cells cultured in vitro, where the lamellipodia can be easily identified, strongly suggest that our CE corresponds to lamellipodia (Fig. 6C-E). Most CEs are formed at the anterior edge of control cells(Fig. 6C), whereas equivalent CEs are found all over the cell in embryos injected with the syn4 MO(Fig. 6H). We measured the orientation of CE by calculating the direction of a vector drawn from the centroid of the cell to the centroid of the CE(Fig. 6N). A significant difference in the orientation of the CE was found between control and syn4 MO cells (Fig. 6O,P; P<0.005, n=180). In conclusion, syn4 MO is required for the polarized formation of CEs.
Interestingly, we have previously shown that the inhibition of PCP signaling in vitro in Xenopus NC cells leads to the formation of long cell protrusions all over the cells (De Calisto et al., 2005), reminiscent of the phenotype observed here for syn4 MO. In order to confirm this finding in vivo, we carried out a similar analysis of CE in zebrafish. First, we had to test that non-canonical Wnt signaling is also required for NC migration in zebrafish embryos, as this has only been shown in Xenopus. Three different treatments that specifically affect PCP signaling were used and migration of trunk neural crest was analyzed (Fig. 6S). Embryos injected with a dominant-negative form of Dishevelled(Dsh) specific for the PCP pathway (DshDep+), or with a MO that has been shown to specifically inhibit wnt5(Kilian et al., 2003; Lele et al., 2001)(Fig. 6S-V), and the PCP mutant tri (trilobite, also known as strabismus; Fig. 7C) showed defects in NC migration, although the tri mutant exhibited a milder NC phenotype than those caused by the other two treatments. These results show that in zebrafish, as in Xenopus (De Calisto et al., 2005), PCP signaling is required for NC migration. Next, we used the sox10:egfp transgenic line to analyze the orientation of CE after Dsh inhibition by injection of DshDep+. A significant difference in the orientation of CE was observed compared with control NC cells (Fig. 6Q,R; P<0.0005, n=101). Taken together, our results show that both the activity of Syn4 and PCP signaling are required to restrict the formation of CEs to the anterior edge of the cell in vivo.
As the inhibition of Syn4 and PCP signaling produce a similar NC phenotype,we investigated whether there was an interaction between these two signals by performing a genetic interaction study in zebrafish embryos. syn4 MO was co-injected with a MO specific for wnt5(Kilian et al., 2003; Lele et al., 2001). When 1 ng of syn4 or wnt5 MO was injected into wild-type embryos only a mild inhibition of NC migration was observed(Fig. 7B,F); however,co-injection of both MOs together produced a much stronger NC phenotype(Fig. 7G,H). An equivalent experiment, with similar results, was performed with the tri mutant(Fig. 7A,C,D). These results indicate that Syn4 interacts genetically with the PCP signaling pathway. This genetic interaction could be compatible with either a hierarchical relationship between Syn4 and Dsh, or an interaction between two parallel pathways.
Syn4 and PCP signaling control the localized activity of Rac and RhoA
It has been observed in many in vitro studies of cell migration that cell polarity and the formation of cell protrusions are dependent on the activity of members of the small Rho GTPase family(Jaffe and Hall, 2005). To analyze the activity of GTPases in the NC in vitro and in vivo, we used fluorescence resonance energy transfer (FRET) biosensors for Cdc42, Rac and RhoA (Itoh et al., 2002; Pertz et al., 2006). The inhibition of Syn4 leads to a fivefold increase in Rac activity(Fig. 8A-C). No significant effect on the activity or localization of Cdc42 or RhoA was seen between control and syn4 MO-injected cells(Fig. 8A). Therefore, Syn4 inhibits Rac activity in the NC cell migrating in vitro.
Next, we investigated whether Dsh could also play a role in GTPase signaling. The activation of Cdc42, Rac and RhoA was compared among control NC cells and cells from embryos injected with DshΔn or DshDep+(Fig. 8D-F). The activation or inhibition of Dsh had no significant effect on the activity of Cdc42 or Rac(Fig. 8D). However, the activation of Dsh led to a significant increase in RhoA activity, whereas the inhibition of Dsh produced a significant decrease in RhoA activity(Fig. 8D-F). Thus, Dsh-PCP promotes RhoA activity in NC migration in vitro.
Crosstalk between Rac and RhoA has been described in a number of different cell types. We wished to discover whether a similar feedback loop was present in the NC cells. Cells were treated with Y27632 to specifically inhibit the RhoA effector Rock (Uehata et al.,1997), or with NSC23766 to inhibit Rac activity(Gao et al., 2004), and the activity of Rac and RhoA was analyzed by FRET. The inhibition of Rock led to a significant increase in Rac activation(Fig. 8G-I), suggesting that RhoA (Rock) can act as an inhibitor of Rac activity in the NC cells. In addition, the inhibition of Rock produced an increase in cell protrusions in different directions, whereas the inhibition of Rac reduced the number of cell protrusions (not shown).
These measurements of small Rho GTPase activity were performed in NC cells cultured on fibronectin in vitro. A growing number of reports suggest that there are important differences in the migration of cells cultured in two-dimensional (2D) versus 3D matrices(Even-Ram and Yamada, 2005),and that those differences could be even bigger when cell migration is analyzed in vivo, so we decided to perform FRET analysis of NC cells migrating in the embryo. To our knowledge, this is the first time that small GTPase activity has been observed in vivo using FRET. DNA coding the FRET probes for RhoA and Rac was injected in blastomeres fated to become NC cells. At the early neurula stages, the NCs were dissected from the injected Xenopus embryos and grafted into control hosts. NC cells could be identified by the fluorescence of membrane-RFP, which was co-injected with each probe. During migration the embryos were fixed and sectioned, and FRET analysis was performed. Fig. 9A-D shows the migrating NC, and in Fig. 9A′-D′ cells expressing the biosensors can be seen. Examples of FRET efficiency for individual cells migrating in vivo are shown in Fig. 9E-H. A clear increase in Rac activity is observed in syn4 MO cells(Fig. 9I), whereas an inhibition of RhoA is observed after Dsh inhibition(Fig. 9J). This confirms our in vitro observations, indicating that, in terms of Rho GTPase regulation, there is no apparent difference between the 2D in vitro and in vivo migration of NC cells. Taken together, our results indicate that Syn4 acts as an inhibitor of Rac and that the Dsh-PCP pathway promotes the activity of RhoA. In addition,activation of RhoA by PCP signaling may also result in Rac inhibition via the inhibitory activity of Rock upon Rac. These data all support a model whereby the Syn4 and PCP activities converge to polarize the formation of cell protrusions, restricting them to the front of the cell. More specifically,they control the levels of Rac, by both a Rac-Syn4 and a RhoA-PCP dependent pathway.
Here, we reveal a crucial role for Syn4 in the migration of NC cells in vivo during embryo development, and its interaction with PCP signaling. Our main conclusions are that: (1) syn4 is expressed almost exclusively in the NC; (2) Syn4 and PCP signaling control the directional migration of NC cells by regulating the polarized formation of cell protrusions; and (3) Syn4 inhibits Rac, whereas Dsh promotes RhoA activity in the NC cell. Thus, Syn4 and PCP signaling work in a coordinated manner to control the directionality of NC migration both in vitro and in vivo, by regulating cell polarity and the cytoskeletal machinery that controls the formation of cell protrusions.
Cdc42 has been shown to be active at the front of migrating cells and it has been suggested that this may be required for cell polarity(Etienne-Manneville and Hall,2002). However, more recent studies show no significant effects of loss of Cdc42 upon cell polarity [either by genetic ablation or siRNA targeting (Czuchra et al.,2005; Pankov et al.,2005)]. Our data demonstrate that the inhibition of Syn4 or PCP has no effect on the activation status of Cdc42, despite having a profound effect upon cell polarity and directed migration. This would suggest that in NC migration, Cdc42 is not the primary GTPase regulating polarization and protrusion formation.
Rac activation at the front of a cell has also been shown to be a key event for directional migration (Grande-Garcia et al., 2005; Nishiya et al.,2005). Several factors control the localized activity of Rac, such as specific guanine nucleotide-exchange factors (GEFs) that are delivered to the front of the cell in a PI3K-dependent manner(Welch et al., 2003), and the formation of lipid rafts (del Pozo et al.,2004). Once Rac is active, numerous feedback loops help to maintain directional protrusions (Ridley et al., 2003). We have shown here that Syn4 contributes to the inhibition of Rac, as an antisense MO against syn4 increases the levels of Rac in the entire cell, leading to loss of cell polarity similar to that seen in other cells types in vitro(Bass et al., 2007; Saoncella et al., 2004). Bass and colleagues showed that the regulation of Rac levels by Syn4 contributes to a persistent migration in vitro, with Syn4-null fibroblasts showing an activation of Rac around the cell periphery(Bass et al., 2007), similar to what we have observed in the NC. However, there are some notable differences between their results and those shown here, primarily that they demonstrate that Syn4 stimulates a wave of Rac activation upon initial cell spreading,whereas we only observed a negative regulation of Rac by Syn4. Additionally,Bass et al. showed that Syn4 containing a mutation in the PKCα-binding site was able to rescue persistent migration in vitro, whereas we were unable to rescue the embryonic phenotype of the syn4 MO with this same mutant. Thus, our results support the previous notion that the PKCα-binding site is essential for Syn4 function(Alexopoulou et al., 2007). The differences between our results and those of Bass et al.(Bass et al., 2007) could be due to the added complexity of migration in an in vivo environment, or to an intrinsic difference between NC cells and fibroblasts.
Differences in cell migration in vivo and in vitro have been widely documented. A number of research groups have now begun to switch to 3D models,which provide a better representation of the microenvironment of living tissues (Even-Ram and Yamada,2005). There are some key differences between cell migration in two and three dimensions. For example, αvβ3integrin is not detected in 3D-matrix adhesion, and focal adhesion kinase(FAK) is less phosphorylated at residue Y397 in fibroblasts in a 3D matrix than it is in those on a 2D substrate(Cukierman et al., 2001). It has been suggested that Syn4 plays no role in migration in vivo, as, despite its role in vitro, the disruption of Syn4 in mice causes only a relatively minor and specific defect in wound healing(Echtermeyer et al., 2001). However, here, we have demonstrated a clear role for Syn4 in controlling cell migration in vivo. The minor phenotype in mice could be due to redundancy between syndecans, with Syn1 possibly being able to compensate for the lack of Syn4 activity. Interestingly, Syn1 has not been found in zebrafish (R.M.,unpublished). A double Syn4/Syn1 mouse knock-out would be able to clarify this point.
Unlike with Syn4, no significant effect on Rac activity was observed after modifying PCP signaling in our study, consistent with the role of Rock on convergent extension (Marlow et al.,2002). This is paradoxical, because we show that PCP signaling promotes RhoA activation and that RhoA, via Rock, inhibits the activity of Rac in NC cells. One possible explanation is that a residual amount of RhoA remaining after Dsh inhibition is sufficient to maintain the normal Rac level.
It has been demonstrated that RhoA, Rac1 and Cdc42 all act downstream of PCP signaling (Choi and Han,2002; Habas et al.,2003; Penzo-Mendez et al.,2003). Convergent extension in Xenopus and zebrafish are dependent on RhoA/Rac activity, controlled by PCP signaling(Habas et al., 2003; Habas et al., 2001; Tahinci and Symes, 2003). Our results show that RhoA, but not Rac, is dependent on PCP signaling. Interestingly, a triple deletion of the three Rac genes in Drosophila,Rac1, Rac2 and Mtl, fails to cause PCP defects(Adler, 2002; Hakeda-Suzuki et al., 2002),suggesting that Rac signaling is not essential for the PCP pathway.
One open question is: how exactly does Syn4 interact with PCP signaling to control NC migration? Our results clearly indicate that Syn4 and PCP activate two parallel pathways that lead to the inhibition of Rac activity and the activation of RhoA, respectively (Fig. 10). However, we also show that RhoA inhibits Rac activity in the NC. Thus, both pathways ultimately have the same effect of decreasing the overall levels of Rac, either directly or indirectly, which is necessary for the polarized formation of cell protrusions and for maintaining persistent migration (Pankov et al.,2005). The requirement for precise levels of Rac signaling for persistent migration may explain why both inhibition and overexpression of Syn4 produce an inhibition of migration, as has been previously shown for the PCP signaling pathway (De Calisto et al.,2005; Wallingford et al.,2000). It has recently been shown that Syn4 interacts with PCP signaling during convergent extension of the mesoderm(Munoz et al., 2006). They propose a direct interaction of Syn4 with Dsh and Frz7. Thus, it is possible that, in addition to this interaction between Syn4 and the Wnt receptor, each of these molecules could lead to the activation of parallel pathways that control the activity of small GTPases, as we have shown here.
Chemotaxis has been suggested as one of the mechanisms to explain the directional migration of NC cells; however, there is no sound evidence for this proposal. Moreover, persistent directional migration occurs in vitro in the absence of chemoattractants. Instead, interactions between the extracellular matrix, integrins, and the levels of Rac and Syn4 can control persistent migration (Bass et al.,2007; Choma et al.,2004; Pankov et al.,2005; White et al.,2007). Here, we provide evidence that a similar mechanism regulates the migration of NC cells in vivo. Syn4 and PCP signaling in the NC act on RhoA and Rac to maintain the balance of Rac required for the formation of directional cell protrusions, which results in a persistent, directional migration.
We thank M. Tada for useful comments and reagents during the course of this work. We also thank C. Stern for comments on the manuscript, R. Kelsh for the sox10:egfp transgenic line, M. Matsuda for the Cdc42 and Rac FRET probes, and K. Hahn for the RhoA FRET probe. This investigation was supported by grants from the MRC and BBSRC. H.K.M. and C.C.-F. are MRC and Boehringer Ingelheim Fonds PhD scholarship holders, respectively.