Ptenb Mediates Gastrulation Cell Movements via Cdc42/AKT1 in Zebrafish

Phosphatidylinositol 3-kinase (PI3 kinase) mediates gastrulation cell migration in zebrafish via its regulation of PIP2/PIP3 balance. Although PI3 kinase counter enzyme PTEN has also been reported to be essential for gastrulation, its role in zebrafish gastrulation has been controversial due to the lack of gastrulation defects in pten-null mutants. To clarify this issue, we knocked down a pten isoform, ptenb by using anti-sense morpholino oligos (MOs) in zebrafish embryos and found that ptenb MOs inhibit convergent extension by affecting cell motility and protrusion during gastrulation. The ptenb MO-induced convergence defect could be rescued by a PI3-kinase inhibitor, LY294002 and by overexpressing dominant negative Cdc42. Overexpression of human constitutively active akt1 showed similar convergent extension defects in zebrafish embryos. We also observed a clear enhancement of actin polymerization in ptenb morphants under cofocal microscopy and in actin polymerization assay. These results suggest that Ptenb by antagonizing PI3 kinase and its downstream Akt1 and Cdc42 to regulate actin polymerization that is critical for proper cell motility and migration control during gastrulation in zebrafish.


Introduction
Gastrulation is a morphogenetic process involving cell migration and rearrangements to establish three germ layers: ectoderm, mesoderm, and endoderm [1]. In zebrafish, three distinct morphogenetic cell movements occur during gastrulation, including epiboly, involution, convergence and extension [2]. Gastrulation starts after the blastula stage when embryo proper appears as a mass of cells situated on top of yolk cells. The yolk sphere is then forming a dome cap that pushes the mass of blastomeres to become thinner and start to spread over the yolk sphere in a process called epiboly. After 50% of the yolk sphere is enclosed by the blastoderm, the front runner cells at the putative dorsal side begin to involute retrogradely toward the future anterior part to form mesoderm and endoderm progenitor cells. At about midgastrulation, convergence and extension movements occur to narrow medio-lateral and elongate anterior-posterior of body axis, respectively, that is essential to set up the dorsal-ventral and anterior-posterior axes [3,4,5]. These gastrulation cell movements are well demonstrated to be mediated by cell adhesion and cytoskeleton rearrangement [6,7,8].
Cell adhesion and cytoskeleton rearrangement can be associated to the metabolism of membrane lipids. One of the key enzymes for metabolizing membrane lipid is phosphoinositide 3-kinases (PI3 kinase). PI3 kinase can phosphorylate the D3 position hydroxyl group of the inositol ring of phosphatidylinositol-4,5-diphosphate (PIP 2 ).
In mouse, fruit fly and chicken, PTEN is known to regulate cell migration, cell cycle length, and cell survival during early embryogenesis [24,25,26]. Pten2/2 knockout mice die at around embryonic day 7.5 [24,27] and altering pten expression before midblatula transition causes gastrulation delay in Xenopus embryos [28]. It appears that Pten is an indispensable gene during embryogenesis. However, the effects of Pten on the dynamic gastrulating cell movements have not been examined because of experimental constraints. Zebrafish is a well-established model to study the dynamic processes of gastrulation cell movements [29]. There are two zebrafish pten isoforms, ptena and ptenb. Each isoform has two alternatively splicing forms. Using the morpholino (MO) knockdown approach, Croushore et al. [30] showed that ptena or ptenb MO-injected embryos exhibit distinct morphological defects in 24-48 hours post fertilization (hpf), but the effects of those MOs on gastrulation were not described. Comparing genomic synteny, it reveals that zebrafish ptenb and the human PTEN have the conserved locus. It suggests that ptenb is probably the ortholog of human PTEN [31]. Thus, we set out to examine the effects of Ptenb on gastrulation cell movements and found that knockdown of ptenb by MO cell non-autonomously disturbs epiboly and convergent-extension cell movements during gastrulation in zebrafish.

Ptenb expresses maternally and exists throughout early embryogenesis
Previous studies have reported that ptenb is ubiquitously expressed in zebrafish early embryos at a few selected stages [30,32], but its expression patterns at most cleavage and gastrulation stages are still lacking. RT-PCR analysis showed that ptenb mRNA was indeed expressed in every early embryonic stage examined, reduced at 30% epiboly then gradually recovered afterward (Fig. 1A). Whole-mount in situ hybridization (WISH) analysis revealed ubiquitous expression patterns of ptenb in embryos up to 18-somite stage (Fig. 1B). The maternal and ubiquitous expression patterns of ptenb during early embryonic stages suggested that it may play a pivotal role during early embryogenesis. Specificity and potency examination of translation-blocking ptenb MO Two published antisense translational blocking MOs [30] were adopted to study the role of Ptenb during gasrulation. These two MOs bind to ptenb mRNA 59 untranslated region (59 UTR) at nonoverlapping sites as indicated in Fig. 2A. To further confirm the specificity and translational blocking efficiency of the first ptenb MO (tMO 1 ), we co-injected plasmids of a PCS2+ construct containing the tMO 1 binding site (PCS2+_ptenb 59 UTR) to check the tMO 1 translation blocking ability. Embryos (n = 148) injected with the PCS2+_ptenb 59 UTR plasmids alone expressed green fluorescent protein (GFP) in a mosaic pattern with a high ratio of 87.260.23% (N = 3) at 40% epiboly stage (Fig. 2C). By contrast, co-injection of 5 ng tMO 1 (n = 160) completely blocked the expression of GFP (Fig. 2E). These results indicated that ptenb tMO 1 specifically and potently blocks zebrafish ptenb translation.

Knockdown of ptenb impairs gastrulation
To understand the role of Ptenb during early embryogenesis, we injected embryos with tMO 1 and examined its effect on embryonic development. A standard control MO (StdMO), which has a minimum effect on embryonic development, was used as a MO control. The MO-injected embryos will be called morphants hereafter. The epibolic progression in ptenb morphants was notably slower and some morphants displayed a cylinder-like shape. Most of the StdMO morphants (98.760.9%) reached over 95% epiboly stage at 10 hpf (Fig. 3A). However, the epiboly progression was notably lagged in tMO 1 morphants in a dose-dependent manner (N"4). To test if the observed defects were specific to the loss of ptenb, we further tested the effects of a second published ptenb MO, tMO 2 [30] and a 5-bp mismatched tMO 1 (mis-tMO 1 ). At a dosage of 15 ng, only 2.761.1% tMO 1 , 20.9613.8% tMO 2 and 16.6611.7% 5-bp mis-tMO 1 morphants reached over 95% epiboly stage at 10 hpf. Eventually (,1-3 h lag) most of them could complete epiboly at MO dosages tested. This delay in epiboly-induced by ptenb MOs could be partially rescued by coinjecting ptenb mRNAs (Fig. 3B). We then examined the effect of ptenb MOs at 14 hpf and found the shortening of embryonic axis. The tail rudiment of tMO 1 morphants was either less extended (intermediate phenotype, Fig. 4B) or stalled at vegetal pole (severe phenotype, Fig. 4C). This tail defect was tMO 1 -dose dependent (Fig. 4D). mis-tMO 1 ( Fig. 4E). Taken together, the ptenb MO-induced defects appeared to be caused by the specific loss of ptenb. With the higher potency of tMO 1 observed we used it at the rest of experiments unless otherwise stated.
Since the observed shortening of embryonic axis is a typical convergent extension defect [33], we examined the effect of ptenb MO on convergence. Embryos were fixed after checking early morphological defects and subjected to WISH against myod to reveal somites. The width of the widest somite of the last three posterior somites was measured to estimate the extent of convergence (Fig. 5A). The average somite width was 173.763.8 mm in StdMO (15 ng) morphants, which was comparable to the untreated embryos (data not shown). Embryos injected with tMO 1 showed expanded somite width in a dose-dependent manner (Fig. 5B). The somite width was 189.962.0 mm, 196.361.4 mm and 215.966.3 mm for embryos injected with 5 ng, 10 ng and 15 ng tMO 1 , respectively. tMO 2 (15 ng) morphants also exhibited severe convergence defect with 236.466.0 mm in somite width. This convergence defect could be rescued by co-injecting 15 ng tMO 1 and 50 pg ptenb mRNA with a somite width at 170.461.8 mm, which was similar to that of StdMOinjected embryos.
To examine the effect of tMO 1 on dorsal axis extension, we measured the angle between the anterior end of prechordal plate (as revealed by WISH against ctsl1b) and the posterior end of myod. The tMO 1 morphants showed notable increase in angle (i.e. the inhibition of dorsal axis extension) compared to that of StdMO morphants (Fig. 6A). The ptenb MO-induced extension defect could also be partially rescued by ptenb mRNA co-injection as shown in Fig. 6B. Since the ctsl1b probe-labeled prechordal plates were observed in both control embryos and morphants (Fig. 6A), it appeared that the mesendodermal involution was not affected by the loss of Ptenb.

Ptenb knockdown effects can be alleviated by inhibiting PI3 kinase
To understand whether ptenb MOs impair gastrulation cell movements through reinforcing PI3 kinase-mediated responses, we examined whether ptenb MO-induced gastrulation defects could be mitigated by inhibiting PI3 kinase. By analyzing the ratio of somite width to embryo diameter ( Fig. 7A-C), we observed that in the presence of 5 mM of PI3 kinase inhibitor LY294002 (n = 72), the somite width ro embryo diameter ratio was not significantly affected (0.30860.013 mm v.s. 0.29560.003 mm, n = 85) in 10 ng StdMOinjected embryos (N = 3). By contrast, LY294002 almost fully rescued the impaired somite-convergent defect in tMO 1 morphants (Fig. 7G). However, it appeared to be less potent to rescue the extension defect induced by tMO 1 at the dosage tested (Fig. 7F,G). We attempted to use higher dosages up to 30 mm of LY294002, but the decorinated embryos were too fragile to be examined under those conditions. These results indicated that ptenb may regulate zebrafish convergent movement by antagonizing PI3 kinase.

Ptenb regulates cell movements via modulating cell protrusive activity
To examine how ptenb regulates convergence and extension cellular movements, we analyzed cell migration in different regions of gastrulating zebrafish embryos. As shown in movie S1 and S2, the leading edge hypoblast cells of a StdMO morphant formed different types of protrusion but mainly the long and thick The behaviors of prechordal plate leading edge cells at 90% epiboly were also analyzed for extension abnormality. In StdMO morphants, prechordal plate cells formed multiple layers and migrated as a sheet with the leading edge cells formed protrusions at front and adhered cells in the back. The leading edge cells (n = 17) were quite active, they formed protrusions, including filapodia, lamellapodia, and blebbing, and migrated fast over the yolk sphere at a velocity of 2.360.2 mm/min (N = 5, movie S3). The leading-edge cells of tMO 1 morphants (n = 18) also had the same protrusions as those in StdMO morphants but their velocity of leading-edge cell migration was significantly slower at1.860.2 mm/min (N = 5, p,0.01, movie S4).

Ptenb functions non cell-autonomously for gastrulation cell migration
To assess cell-autonomy of ptenb function in mediating migration of lateral hypoblast cells, we performed cell transplantation experiments by using rhodamine dextran as a cell tracer. Blastomeres from StdMO or tMO 1 morphants were transplanted to host embryos treated without or with tMO 1 , and examined under confocal microscopy or epifluorescent microscopy. Under confocal microscopy, cells from a StdMO morphant showed highly protrusive activities with well formed protrusions when transplanted into an untreated host embryo (''STD.UT'', arrow,  Knockdown of ptenb enhanced actin polymerization PTEN is known to control protrusion formation and adhesion during cell migration [34] via modulating actin dynamics [35]. To determine whether actin polymerization during gastrulation in tMO 1 morphants was affected, we used rhodamine phalloidin staining to visualize filamentous actin (F-actin) in gastrulating embryos under confocal microscopy. The vegetal pole view showed that, F-actins were mainly located at cell boundaries corresponding to the cortical actin structure underneath cell membrane in StdMO-treated embryos (Fig. 10A). In contrast, the overall intensity of F-actin was enhanced without disturbing actin distribution patterns (Fig. 10B). To further confirm the increase of F-actin in tMO 1 morphants, we measured the F-actin contents of these embryos by using an actin polymerization assay [36]. In 3 experiments, tMO 1 morphants exhibited ,1.8 folds of fluorescence compared to that of control embryos (Fig. 10C). These results demonstrated that knockdown of ptenb results in an increase in actin polymerization.

Dominant-negative Cdc42 rescues ptenb MO-induced convergence and extension defects
Small GTPases Rac1 and Cdc42 were known downstream of PTEN in in vitro studies [35]. To examine whether Cdc42 and/or Rac1 are involved in the Ptenb-mediated regulation in gastrulation cell movements, dominant negative, T17NCdc42 or T17NRac1 were used to see their effects on the ptenb MO-induced convergent extension defects. Co-injection of T17NCdc42 mRNAs (25 pg) with StdMO (10 ng) did not change the somite width, but increased the extension angle (N = 4, Fig. S1A). In contrast, T17NCdc42 mRNAs could partially restore the somite width and extension angle of tMO 1 morphants (  (Fig. S1B,D, N = 4). T17NRac1 mRNAs (50 pg) and ptenb tMO 1 (10 ng) co-injected embryos only slightly improved the convergent extension defect compared to that of tMO 1 morphants with no statistical difference (p.0.05) between groups.

Overexpression of human constitutively active AKT1 causes convergence and extension defects
Knockdown of ptenb would lead to increase of downstream Akt/ PKB activity, to investigate whether the elevation of AKT/PKB activity may result in similar ptenb MO-induced phenotypes. We overexpressed human constitutively active akt1 (caakt1) in zebrafish embryos to observe its effect on convergent extension. Embryosinjected with 16 buffer and caakt1 mRNAs at 50, 100, or 200 pg were collected and subjected to WISH against myod and ctsl1b. As shown in Table 2, lower doses (50 and 100 pg) of human caakt1 mRNAs caused slight convergent extension defects, but profound inhibition on convergent extension was observed in embryostreated with 200 pg caakt1 mRNAs. These results implied that the ptenb MO-induced convergent extension defects might be due to the elevation of AKT1 activity.

Ptenb MO causes convergence and extension defects in ptenb2/2 embryos
The lack of notable early embryonic defect in ptenb2/2 mutant fish is in contrast to our observations described above that might be due to the presence of maternal pten transcripts [32]. To examine this possibility, we had obtained heterozygous ptenb+/2 embryos, raised to adults to give rise to F 1 offspring. Those F 1 fish were screened by PCR for the presence of mutated ptenb to get ptenb2/2 homozygous fish according to Faucherre et al. [32]. Different dosages of ptenb tMO 1 were injected into ptenb2/2 embryos, fixed and subjected to WISH against myod and ctsl1b. Similar results of ptenb MO-induced convergent extension defects were obtained as that of wild type embryos (Table 3). Extension angle and somite width of ptenb2/2 embryos were significantly increased in the presence of ptenb tMO 1 . These data implied that the presence of maternal ptenb transcripts was sufficient to complete normal gastrulation in ptenb2/2 embryos.

Discussion
The necessity of PTEN in embryogenesis has been demonstrated in several animal species including mice, chickens and fruit flies [24,25,26]. However, the role of Pten in zebrafish gastulation has been controversial. In this study, we demonstrated that ptenb may regulate zebrafish gastrulation cell movements by controlling actin polymerization and directed cell migration via antagonizing PI3 kinase and downstream Cdc42 and Akt activity.

Ptenb regulates convergence and extension during gastrulation
PI3 kinase has been shown to be required for cell polarization of gastrulating mesendodermal cells by overexpressing dominantnegative form of PI3 kinase or using the PI3 kinase inhibitor LY294002 [10]. The authors unequivocally demonstrated the necessity of maintaining PIP 2 /PIP 3 balance in mesendodermal cell polarity of directed migration during gastrulation in zebrafish. It is logical to hypothesize that the negative regulator of PI3 kinase, PTEN, should also be necessary for regulating gastrulation cell movements in zebrafish, but no gastrulation defects was observed in zebrafish pten mutants [37]. Based on the pten mutant study, Pten appears to play no role during gastrulation in zebrafish. However, the presence of maternal pten messages in early mutant embryos cannot be excluded. Secondly, the lack of functional Pten in the zebrafish pten mutants have not be demonstrated due to the lack of antibodies which can recognize zebrafish Pten. Therefore, we took an alternative approach by using pten MOs which have been previously shown to have specifically lowered pten translation and enhanced AKT activity [30] to analyze their effects on gastrulation cell movements in zebrafish. Our results clearly indicated that block of ptenb activity affects convergent extension movements during gastrulation in zebrafish.

Zebrafish Ptenb regulates gastrulation in coordination with PI3 kinase
Pten is a known counter enzyme of PI3 kinase by dephosphorylating PIP 3 to PIP 2 [9,18], and also acts via downstream effectors AKT/PKB to regulate early developmental processes [22,25,28]. To examine the coordination of PI3 kinase and Ptenb in regulating zebrafish gastrulation, we showed that LY294002 could rescue convergence defect in tMO 1 morphants. It clearly demonstrated that PI3 kinase and PTEN linkage is required to modulate cell movements during zebrafish gastrulation.

Zebrafish Ptenb regulates protrusive activities of lateral mesendodermal cells but not prechordal plate cells during gastrulation
In the LY294002 rescue experiment, the result suggested that Ptenb coordinates with PI3 kinase to regulate gastrulation cell movements presumably via their regulation on PIP 2 /PIP 3 balance. Activated PI3 kinase can increase PIP 3 concentration, which would localize at the frontal edge of protrusions and induce the directional migration, and Pten acts at the rear edge of cells to dephophorylate PIP 3 to PIP 2 then restrict the spatial expression of PIP 3 in the cell [22,23]. Our time-lapse recordings clearly demonstrated that active protrusive activity is required for cell movements during zebrafish gastrulation. The ptenb MO-induced gastrulation defect was not due to the change in protrusion numbers. In stead, the directionality and persistency of protrusion were more relevant. Lamellapodia were formed at the leading edge of lateral hypoblast cells and these protrusions were mainly pointing anterodorsally to the direction of cell movement. However, tMO 1 morphants' lateral hypoblasts exhibited interference in persistence and directionality of protrusions. In the lateral hypoblasts of tMO 1 morphants, the increase of lamellapodia turnover was evident by the reduction of lamellapodia persistence. At the same time, the directionality was also affected by decreasing the number of protrusions pointed anteriodorsally. The directional defect was also observed at the transplantation experiment. StdMO cells transplanted to untreated embryos revealed a better linearity than the tMO 1 morphant cells transplantated to tMO 1 hosts. The abnormality of protrusion persistence and migration directionality was not observed at prechordal plate cells, this result might be due to the fact that prechordal plate cells migrate as multicellular sheets and cell-cell are highly contact. The rear area of the leading edge cells were in close contact with other cells that might help to maintain cell polarity in the absence of Ptenb [38].

Zebrafish Ptenb mediates gastrulation cellular movements via controlling actin polymerization and downstream Akt1 and Cdc42
In vitro cell studies have shown that, the increase of PIP 3 concentration in cells activates Rac1 and Cdc42 which would lead to actin polymerization and lamellapodia and filapodia formation [35,39]. Rac1 has further reported to be important in Drosophila mesoderm migration during gastrula [40] and zebrafish convergence by controlling lamellapodia formation [41]. In our results, dominant negative Cdc42 and to a less extent of dominant negative Rac1 also showed to rescue convergent extension defects in ptenb morphants.
AKT/PKB is another known downstream factor of PI3 kinase. The binding of AKT/PKB to PIP 3 can recruit it to membrane for the regulation of actin arrangement and protrusion formation [42]. Knockdown of ptenb should lead to an increase of PIP 3 and subsequent elevation of AKT activity in zebrafish embryos. The elevated AKT activity might disturb the actin cytoskeleton, protrusion formation and cell movements. Over-expression of caakt1 showed convergence and extension defects. These results demonstrated that Ptenb regulates zebrafish convergence and extension during gastrulation via PI3 kinase-Akt pathway that is consistent with the rescue of ptenb tMO 1 morphants by the PI3 kinase inhibitor.
Gastrulation cell movements are tightly regulated by actin cytoskeletons. Here, we showed that actin polymerization was enhanced about two folds in the ptenb tMO 1 morphants than the control embryos. These results suggest that zebrafish Ptenb downregulates small GTPases Rac1 and Cdc42, which further rearrange actin polymerization, then controls the convergence and extension during gastrulation.
ptenb MO induces convergence and extension defects in ptenb2/2 embryos The lack of gastrulation defect in the pten mutants argues a role of PTEN in zebrafish and the authors also questioned that the effects of pten MOs on zebrafish early embryos reported by Croushore et al. [30] may not be due to the specific loss of Pten [32]. However, we demonstrated the specificity of ptenb MO used with the following evidences: (1) Co-injection of ptenb tMO1 and ptenb 59 UTR-GFP plasmid showed no fluorescence signal compared to ptenb 59 UTR-GFP injection alone. (2) Ptenb tMO 1 caused convergence and extension in a dose-dependent manner.
(3) The defects of ptenb morphants could be rescued by exogenous ptenb mRNAs, LY294002 and dominant-negative Cdc42. (4) Since the generation of functional truncated Ptenb protein could not be excluded in the ptenb mutants [32], the ptenb2/2 embryos was treated with ptenb tMO 1 and these mutant embryos showed similar morphological defects with wild type embryos. In addition to our results, Croushore et al., [30] had shown that ptenb tMO 1 inhibited Ptenb protein in vitro translation assay and elevated phospho-Akt (pAkt). Those results supported that the ptenb-induced gastrulation defects observed were specifically due to the loss of ptenb.
In summary, we provide the first in vivo evidence that Ptenb coordinates with PI3 kinase to modulate downstream AKT1 and Cdc42 for rearranging actin polymerization and protrusion formation that can lead to proper cell migration to regulate convergence and extension cell movements during gastrulation in zebrafish as schematically depicted in Fig. 11.

Ethics Statement
All animal handling procedures were approved by the use of laboratory animal committee at National Taiwan University, Taipei, Taiwan (IACUC Approval ID: 97 Animal Use document No. 55).  The definition of embryo stage was according to Kimmel et al. [43], and the stages are indicated as hours post-fertilization (hpf).

Morpholino oligonucleotide microinjections
All antisense morpholino oligonucleotides (MOs) were custom made by Gene Tools, LLC (Philomath, OR). Standard control MO (Std MO, 59-CCTCTTACCTCAGTTACAATTTATA-39) has no homology sequence to any known zebrafish sequence

Whole-mount in situ hybridization (WISH)
Embryos were fixed in desired stages in 4% paraformaldehyde in phosphate-buffered saline overnight, dechorionated manually by fine forceps and stored in 100% methanol at 220uC until use. Antisense digoxigenin (DIG)-labeled RNA riboprobes were synthesized following the manufacture instructions (Roche Applied Science, Penzberg,Germany). The myod construct [45] was linearized by XbaI and transcribed by T7 RNA polymerase (Roche Applied Science, Penzberg, Germany). The ctsl1b construct [46] was linearized by NotI and transcribed by T7 RNA polymerase. In situ hybridization and detection were performed according to Thisse et al. [47] with a phosphatasecoupled anti-DIG antibody. The processed embryos were then transferred to 100% glycerol and photographed by using Nikon CoolPIX 995 digital camera.

Measurement and counting of embryos
For the convergence and extension assay, ptenb MOs or STD MO treated embryos were incubated at 28.5uC until 8 to 11somite stages for morphological observation under stereoscope. After the morphology classification, the embryos were fixed and subjected to WISH against myod to measure the width of myod signal of the widest somite among the last three somites. Furthermore, the extension defects were characterized by measuring the angle between prechordal plate staining using ctsl1b and the vegetal end of myod signals. All experiments were repeated at least three times and the measurements were made using tools in Adobe Photoshop CS4  DIC time-lapse cell migration recording To monitor the migration of lateral hypoblast leading edge cells (marginal deep cells), the embryos at 75% epiboly stage were dechorionated with 0.01 mg/mL protease (Sigma, St. Louis, MO) and mounted in 0.8% low-melting agarose (Amresco, Solon, OH). The migrating cells were recorded using a 406 water immersion objective under a Leica DM5000 B DIC microscope (Leica Microsystems, Wetzlar, Germany). One hour long movies were recorded at 30-sec intervals by a CoolSNAP fx CCD camera (Roper Scientific, Tucson, AZ). The movies were acquired and analyzed by Simple PCI Imagine System software (Compix, Sewickley, PA). The protrusions which merged from cell body with an angle ,135uC and a width .2 mm were defined as a lamellapodium extension. To monitor the migration of prechordal plate cells, the embryos at 90% epiboly stage were dechorionated and mounted as described, and then monitored, filmed (5 min movies at 5 sec intervals) and analyzed using a 636 water immersion as described. To track transplanted cells, 1 hour long time-lapse recordings with 10-sec intervals were taken and analyzed by using the Simple PCI software for migration path, velocity and linearity (the shortest distance between the start and end points of the movement divided by the total distance moved) analyses.

Actin polymerization assay
The filamentous actin (F-actin) of embryo was determined by fluorometric assay which based on the binding affinity of rhodamine phalloidin and F actin [48,49]. Embryos injected with different MOs were fixed at shield stage in 5% formaldehyde in Actin Stabilizing Buffer (ASB) [50] overnight. The embryos were then incubated in 150 mM glycine-ASB for 3 hours, washed by ASB, and then labeled by 165 nM rhodamine phalloidin (Molecular Probes, Inc., Eugene, OR). Control embryos were incubated by 20 mg/mL unlabeled phalloidin for 50 minutes prior to rhodamine phalloidin staining. All samples in a group of 60 embryos were then washed in ASB and extracted in dark with methanol and homogenized by 24G needles, and incubated with rotating for 36 hours at 4uC. The fluorescence signals were detected from the extractant and measured by using a fluorospectrophotometer (Hitachi F-2000, Tokyo, Japan) with an excitation wavelength of 565 nm and an emission wavelength of 580 nm [36].

Cell transplantation
For transplantation preparation, donor embryos were injected with 0.5% rhodamine dextran and Std MO 10 ng or Ptenb tMO 1 10 ng respectively. Donor embryos were transplanted into wild type and Ptenb morphant host separately at sphere stage as described previously [51]. Transplanted cells were further recorded by using Leica DM5000 B microscope and Leica TCS SP5 Confocal Microscope Imaging System.

ptenb2/2 zebrafish screening
Heterozygous ptenb+/2 zebrafish embryos were kindly provided by Jeroen den Hertog (Hubrecht Institute, Utrecht, The Netherlands). Those embryos were raised to adults and mated to produce F 1 fish. Nested PCR was performed to select homozygous ptenb 2/2 zebrafish according to Faucherre et al. [32].

Statistical analysis
All experimental values are presented as mean 6 standard error and were analyzed by unpaired-sample Student's t-test in Microsoft Excel. N indicates the number of experiments repeated; n indicates the total sample number in one experimental condition. Different superscript lettering between values stands for a significant difference at p,0.05.

Supporting Information
Figure S1 Embryos were injected with 10 ng of StdMO with or without 25 pg of T7 NCdc42 mRNAs and treated as described in Table 1. The somite width and extension angle of each embryo were measured and shown (A). Embryos were injected with 10 ng of StdMO without (B) or with 50 pg of T7 NRac1 mRNAs (C), incubated to 10-somite stage and photographed. The percentages of normal and abnormal embryos in each treatment are shown (D). (PDF) Figure 11. Ptenb-mediated signal transduction during gastrulation in zebrafish. By coordinating with PI3 kinase, Ptenb controls the balance between PIP 2 and PIP 3 that may activate downstream Cdc42 and AKT1 for mediating actin dynamics to regulate cell movements during gastrulation in zebrafish. doi:10.1371/journal.pone.0018702.g011 Movie S1 Time-lapse imaging of cellular migration and protrusions in lateral hypoblast cells of a StdMOinjected morphant. Shown here is a 1-h at 30-s intervals DIC time-lapse image sequence of a 75% epiboly stage embryoinjected with 10 ng StdMO. The animal pole is on the top and the dorsal side is to the left.