|Home | About | Journals | Submit | Contact Us | Français|
Parietal endoderm (PE) contributes to the yolk sac and is the first migratory cell type in the mammalian embryo. We can visualize PE migration in vitro using the F9 teratocarcinoma derived embryoid body outgrowth system and, show here that PE migration is directed by the non-canonical Wnt Planar Cell Polarity (PCP) pathway via Rho/ROCK. Based on golgi apparatus localization and microtubule orientation, 68.6% of cells in control outgrowths are oriented in the direction of migration. Perturbation of Wnt signaling via sFRP treatment results in a loss of orientation coupled with an increase in cell migration. Inhibition of the PCP pathway at the level of Daam1 also results in a loss of cell orientation along with an increase in cell migration, as seen with sFRP treatment. Constitutively active Daam can inhibit the loss of orientation that occurs with sFRP treatment. We previously demonstrated that ROCK inhibition leads to an increase in cell migration, and we now show that these cells also lack oriented migration. Canonical Wnt signaling or the Rac arm of the PCP pathway do not appear to play a role in PE oriented migration. These data suggest the PCP pathway via Rho/ROCK modulates migration of PE.
Directed cell movements are a critical component of embryogenesis and also play a role in many inflammatory diseases, as well as in cancer metastasis. The non-canonical Wnt signaling pathway, also known as the Planar Cell Polarity (PCP) pathway, regulates specific types of directed cell movements. The PCP pathway was first identified in Drosophila, where mutations in pathway components lead to defects in adult cuticular structures, ommatidia organization, and bristle orientation (Klein and Mlodzik, 2005). In vertebrates, the PCP pathway is responsible for orienting hair cells in the inner ear, fur pattern of rodents, cell movements during neural tube closure, wound healing, cell orientation during division in many tissue types, and convergent extension (De Calisto et al., 2005; Guo et al., 2004; Keller, 2002; Klein and Mlodzik, 2005; Magdalena et al., 2003; Nobes and Hall, 1999; Torban et al., 2004; Ybot-Gonzalez et al., 2007).
In PCP signaling, Wnt binds to its receptor, Frizzled, activating Disheveled (DVL). The PDZ and DEP domains of DVL activate the PCP pathway exclusively, and not the β-catenin dependent canonical pathway (Boutros et al., 1998; Habas et al., 2001; Sato et al., 2006). Signaling downstream of DVL can be mediated through the small GTPases Rho or Rac, which are inactive when bound to GDP and active when bound to GTP (Arthur and Burridge, 2001; Etienne-Manneville and Hall, 2002; Habas et al., 2001; Keller, 2002; Sato et al., 2006). In the Rho dependent pathway, the Formin homology adaptor protein Dvl associated activator of morphogenesis 1 (Daam1), associates with the DEP domain of DVL (Habas et al., 2001; Sato et al., 2006). Daam1 interacts with a Rho guanine nucleotide exchange factor (GEF) leading to Rho GTP formation, which in turn activates Rho Kinase (ROCK). ROCK action is mediated in part via non-muscle myosin II (Ishizaki et al., 1996). ROCK directly activates myosin II regulatory light chain (MLC) through phosphorylation at serine 19 and threonine 18 (Amano et al., 1996; Feng et al., 1999; Kawano et al., 1999; Kureishi et al., 1997; Totsukawa et al., 2000). ROCK also indirectly activates MLC by phosphorylating and inactivating myosin light chain phosphatase (MLCP) at its myosin binding subunit. Phosphorylated MLC promotes migration through the formation of actin stress fibers and focal adhesion maturation (Amano et al., 1998; Chihara et al., 1997).
Alternatively to activating Rho/ROCK, Dvl can initiate Rac signaling and its downstream effector c-Jun N-terminal kinase (JNK) (Boutros et al., 1998; De Calisto et al., 2005; Kishida et al., 2004; Klein and Mlodzik, 2005; Krylova et al., 2000; Rosso et al., 2005). Rac directs lamellipodia protrusion at the leading edge and focal complex formation and turnover just behind it (Hall, 2005; Ridley, 2001; Ridley et al., 2003; Webb et al., 2004). Rac GTP promotes lamellipodial extension by activating WASP family proteins (Ridley, 2006). WASP proteins interact with the Arp 2/3 complex initiating complex actin branching by promoting sites of actin polymerization thereby modulating lamellipodia extension (Ridley, 2006). Activation of Rac can also phosphorylate JNK, which localizes to focal adhesions and phosphorylates the focal adhesion protein paxillin, promoting focal adhesion turnover (Almeida et al., 2000; Huang et al., 2003; Rosso et al., 2005; Yamanaka et al., 2002). JNK also co-localizes with the kinesin motor protein KIF3 along microtubules, and this association may be involved in directed migration (Huang et al., 2003; Nagata et al., 1998).
The F9 embryoid body outgrowth system, in which multiple embryonic and extraembryonic lineages emerge, provides an excellent model to study early development (Gardner, 1983; Grabel and Casanova, 1986; Rossant et al., 1997). Yolk sac formation is critical to mammalian embryogenesis because it provides nutrients and gas exchange to the embryo prior to allantois fusion and establishment of the placenta (Rossant and Tam, 2002). Without proper yolk sac formation, the embryo either dies or its growth is stunted. In the mouse pre-implantation embryo, primitive endoderm (PrE) ‘sorts out’ from the inner cell mass (ICM) to cover the surface of the ICM facing the blastocoel (Chazaud et al., 2006). After implantation, the PrE covering the ICM differentiates into visceral endoderm (VE). VE, or PrE precursors of parietal endoderm (PE), contacts the trophectoderm at the lateral margins of the ICM and differentiates into PE (Rossant and Tam, 2002). PE then migrates along the inner surface of the trophectoderm, contributing to the parietal yolk sac. This is the first migratory event in mammalian embryogenesis, but is difficult to study in vivo in the post implantation embryo (Rossant and Tam, 2002). F9 teratoarcinoma cells cultured in suspension in the presence of retinoic acid (RA) form embryoid bodies that contain an inner, undifferentiated core of stem cells surrounded by an outer layer of VE (Grabel et al., 1998; Hogan et al., 1981). When embryoid bodies are plated on an extracellular matrix-coated substrate, VE transdifferentiates into PE and migrates away from the embryoid body, mimicking early mouse embryogenesis (Casanova and Grabel, 1988; Grabel and Casanova, 1986).
Based on time lapse imaging, PE migrates in a manner reminiscent of convergent extension, with cells in the outgrowth changing their relative positions by intercalation. Outgrowth cells maintain close contacts with their neighbors and migrate as a cohesive sheet. These observations lead us to hypothesize that the PCP pathway directs PE migration. We show here that under control conditions, PE cells are polarized preferentially in the direction of migration based upon both the position of the golgi apparatus relative to the nucleus and the alignment of microtubules. Inhibition of the Wnt pathway using the secreted Frizzled Related Protein (sFRP) leads to a loss of orientation in the outgrowth, and an increase in migration distance, suggesting that the Wnt pathway controls oriented migration in PE cells. Transfection of outgrowths with a dominant negative Daam1 construct (N-Daam), leads to a loss in oriented migration and an increase in migration distance, whereas transfection of PE outgrowth with a constitutively active Daam1 construct (C-Daam) prevents the loss of oriented migration observed upon sFRP treatment. Previously, we demonstrated that inhibition of ROCK, the downstream effector of Rho, leads to an increase in cell migration and a decrease in focal adhesions and actin stress fibers (Mills et al., 2005). We now show that ROCK inhibited outgrowth also lacks golgi orientation and microtubule organization in the direction of migration, consistent with a role for Rho in mediating PCP pathway activity. Inhibition of JNK, a downstream effector of Rac, does not affect cell orientation during migration, suggesting this GTPase is not involved. In addition, promoting β-catenin stability by treatment with the GSK-3β inhibitor BIO does not rescue the sFRP mediated loss of cell orientation, suggesting no role for the canonical Wnt pathway. Taken together, these data suggest the PCP pathway, acting via Rho/ROCK, regulates oriented cell migration of PE.
F9 teratocarcinoma cells were cultured on gelatin coated tissue culture dishes (Corning) in DMEM (Gibco) supplemented with 10% bovine serum (Hyclone), Penicillin-streptomycin, and L-glutamine (Gibco). Embryoid bodies were formed by plating F9 stem cells in suspension and treating daily with 7.5 × 10−7 M RA. After 6 days or once VE formed on the outer surface of the embryoid bodies, they were plated on fibronectin (Sigma) coated coverslips (coated overnight with 30 μg/ml fibronectin at 37° C).
Mouse anti-vinculin and mouse anti–α-tubulin were from Sigma and mouse anti-GM130 (golgi marker) from BD Biosciences. Mouse anti-phospho-histone H3 was from Chemicon, mouse anti-active β-catenin from Millipore, and rabbit anti-β-catenin from Cell Signaling. Mouse anti-α-fodrin (spectrin) was from MP Biomedicals, rabbit anti-phospho myosin light chain phosphatase from Upstate, and rabbit anti-myosin light chain phosphatase obtained from Covance. Both rabbit anti-c-jun and mouse anti-phospho c-jun antibodies were from Santa Cruz. For immunofluorescence, Alexa 488 conjugated rabbit anti-GFP, goat anti-mouse Alexa 568, goat-anti mouse Alexa 488, rhodamine conjugated phalloidin, Alexa 488 conjugated phalloidin, Alexa 634 conjugated phalloidin, and the nuclear stain Hoechst 33342 are from Molecular Probes. For western blotting, goat anti-mouse HRP and goat anti-rabbit HRP were from Sigma. The ROCK inhibitor Y27632 was from Calbiochem Inc. Soluble Frizzled Protein (sFRP) was from R&D Systems. For transient transfection, Lipofectamine 2000 was from Invitrogen. The JNK inhibitor SP600125 and the GSK-3β inhibitor 6-bromoindirubin-3′-oxime (BIO) were from EMD Biosciences/Calbiochem. The Daam 1, C-Daam, N-Daam, and GFP plasmids were the generous gift of Raymond Habas (Habas et al., 2001; Sato et al., 2006). The C3 Transferase vector was the generous gift of Keith Burridge (Worthylake et al., 2001).
After day 6, embryoid bodies were plated on fibronectin coated coverslips, PE was allowed to migrate for 48–72 hours before cells were fixed in 3.7% formaldehyde and washed twice with PBS. Cells were then permeabilized in 0.5% Triton-X and then blocked in 1% BSA-PBS for 45 minutes at 37° C. Cells were then incubated overnight at 4° C in the presence of primary antibody in 1% BSA-PBS. GM130 antibody was used at a dilution of 1:50, α-tubulin antibody at 1:500, and the vinculin antibody at 1:400. The following day, cells were washed five times in 1% BSA-PBS, then incubated in either goat anti mouse Alexa 568 or Alexa 488 (1:1000) in 1% BSA-PBS for one hour at room temperature. Cells were then washed five more times in 1% BSA-PBS before incubation with phalloidin (1:100) for 30 minutes at room temperature. Cells were then washed five times in 1% BSA-PBS prior to 10 minute incubation in Hoechst (1:10,000). Cells were washed five times in PBS before mounting in gelvatol/NPG. Slides were analyzed using a Nikon fluorescent microscope equipped with a Photometrics Cool Snap EZ digital camera controlled by NIS Elements.
Day 6 embryoid bodies were plated on fibronectin coated coverslips in F9 media without antibiotics, and outgrowth was transfected after 24 hours as previously described (Mills et al., 2005). Briefly, 1.6 μg of plasmid DNA was incubated with 4 μl of Lipofectamine 2000 in 200 μl antibiotic free media for 20 minutes at room temperature. 800μl of antibiotic free media was added to the plasmid DNA mix. Old media was aspirated off the cells and was replaced with the 1 ml transfection mixture. 24 hours post transfection, cells were allowed to recover in fresh F9 media for 24 hours before being fixed for immunofluorescence. The rate of transient transfection was relatively high based on quantification of GFP positive cells (GFP is linked to the Daam1 plasmids) as previously described (Hong and Grabel, 2006; Mills et al., 2005). Overall rate of transfection was 79% with 27.8% of cells expressing GFP at a high level.
For C-Daam plus sFRP experiments, cells were transfected with C-Daam as described above. Cells were treated with 2.5 μg/ml sFRP 24 hours post transfection. Cells were then fixed for immunofluorescence a total of 72 hours after plating.
After 6 days in suspension culture in the presence of RA, embryoid bodies were plated on fibronectin coated coverslips (30 μg/ml) in 1 ml of media. After 24 hours, half the dishes were either treated with pharmacological inhibitors or transiently transfected with Daam 1 plasmids as described above. sFRP was used at 2.5 μg/ml. 2.5 μg/ml sFRP was used after we tested 2.5 μg/ml and 5 μg/ml concentrations for both golgi orientation and outgrowth distance and there was no significant difference between 2.5 μg/ml and 5 μg/ml. The BIO inhibitor was used at 3 μM after initial experiments with the following concentrations: 1μM, 2μM, and 3 μM (Krylova et al., 2000; Rosso et al., 2005; Sato et al., 2004). The JNK inhibitor was used at 10 μM (Rosso et al., 2005). The ROCK inhibitor Y27632 was used at 10 μM (Mills et al., 2005). 24 hours later phase images of five embryoid bodies and their associated outgrowth were acquired and outgrowth measurements were performed as previously described (Mills et al., 2005).
Outgrowths that were transiently transfected were fixed and stained with Hoechst and measurements were obtained as described above. Hoechst-stained nuclei were used because embryoid bodies flatten by day 3 and the edge of the embryoid body can be discerned with greater precision using Hoechst staining. All experiments were performed in triplicate and analyzed for statistical significance using a two-tailed T-test assuming equal variance in Microsoft Excel.
Cell orientation was measured using techniques described by Nobes and Hall (Nobes and Hall, 1995). Briefly, immunofluorescence was performed for GM130 (golgi), phalloidin, and Hoechst 24 hours post inhibitor treatment and 48 hours after transient transfection. Multi-channel fluorescent images at 20 X were acquired for four different regions of PE outgrowth for each treatment (a minimum of 200 cells per experimental group). Cells are scored as oriented if greater than 50% of the golgi was localized in the front third of the cell in the direction of migration. Cells were scored as not oriented if the golgi was either localized in either third of the cell not oriented in the direction of migration or if diffusely distributed. Outgrowth migration is considered to be directed if over 34% of the cells are oriented, as 33% is random. All experiments were performed in triplicate and arcsine transformations were performed to allow for statistical analysis. A two-tailed T-test assuming equal variance was performed for each set of conditions.
Day 6 embryoid bodies were plated in 10 cm tissue culture dishes for 3 days. Outgrowths were rinsed in ice cold PBS then scraped in ice cold PBS in the presence of protease and phosphatase inhibitors (1 μg/ml leupeptin, 1 μg/ml pepstatin, 1 μg/ml benzamide, 1 μg/ml aprotinin, 2 mM PMSF, and 2 mM sodium orthovanadate). Cell suspensions were pelleted at 8,000 × g for 5 minutes then lysed in buffer composed of 50 mM Tris pH 7.4, 150 mM NaCl, 1 mM CaCl2, 0.1% SDS, 0.05% Sodium Deoxycholate, 0.1% NP-40, with the addition of protease and phosphatase inhibitors as described above, for 30 minutes on ice. Cell lysates were precleared by centrifugation at 13,000 × g for 5 minutes and then a Bradford Protein Assay was performed to determine protein concentration. After normalizing, lysates were solubilized in Laemlli’s buffer and boiled for 5 minutes. Proteins were resolved on 4–20% gradient gels using SDS-PAGE. Proteins were then transferred to Immobilon-P PVDF Membranes electrophoretically via fast transfer and were then blocked for 30 minutes at room temperature in Blotto (0.3% TBS-Tween [TBST]), 3% non-fat powdered milk, and 0.5% BSA). Blots were incubated in primary antibodies overnight rocking at 4° C at the following concentrations and conditions: β-catenin 1:5000 in 5% non-fat powdered milk; active β-catenin 1:1000 in Blotto; c-jun 1:500 in Blotto; phospho c-jun 1:200 in Blotto; α-actinin 1:5000 in Blotto; phospho-myosin light chain phosphatase, myosin light chain phosphatase, and α-fodrin (spectrin) 1:1000 in Blotto. The next day blots were washed 5 times for 5 minutes in TBST. Blots were incubated in either goat anti-mouse HRP or goat anti-rabbit HRP at 1:10,000 in Blotto for 1 hour at room temperature. Blots were then washed as described and visualized using Western Lightning Chemiluminescence Reagent (Perkin Elmer).
Based on time lapse imaging studies, cells in PE outgrowth migrate as a sheet of cells in a manner somewhat reminiscent of convergent extension (Supplemental Mov1). PE outgrowth cells appear to jostle for position with neighboring cells while migrating away from the embryoid body. To determine if migrating PE cells in outgrowth cultures (Figure 1A) are polarized relative to the direction of migration, we used a method based on golgi localization adopted from Nobes and Hall (Nobes and Hall, 1995). This approach divides the cell into thirds (Figure 1B), and a cell is polarized in the direction of migration if greater than 50 percent of the golgi apparatus, as visualized by the golgi specific antibody GM130, is localized in front of the nucleus in the third of the cell oriented closest to the outgrowth leading edge. We define cells as oriented if they are polarized in the direction of migration. If cells are not oriented during outgrowth, then only 33 percent of PE will have their golgi localized in front of the nucleus in the direction of migration. We found that 68.6 percent of outgrowth cells have their golgi positioned in front of the nucleus in the direction of migration (Figure 1C, D). Cells lacking orientation are randomly distributed throughout the outgrowth. Many cells in these outgrowths also align their microtubules in the direction of migration as shown by α-tubulin immunofluoresence (Figure 1E, E′, E″). These data suggest PE outgrowth cells are oriented.
To determine if the PCP pathway regulates PE cell migration in outgrowth cultures, the Wnt inhibitor sFRP was used. sFRP, which lacks the transmembrane domain of frizzled, binds Wnt ligand in the culture media, preventing it from associating with the membrane bound frizzled receptor (Rosso et al., 2005). Reverse Transcriptase Polymerase Chain Reaction (RT PCR) data show that Wnts 3a, 5a,7a, 7b, and Wnt 11 mRNAs are expressed in F9 embryoid bodies (Figure S1.) Wnt 7a and 7b expression increases over time in a manner similar to the VE marker alphafetoprotein (AFP), suggesting that these Wnts are localized to the VE. sFRP treated cultures displayed ample outgrowth but the PE cells are not oriented, based on random golgi localization (27%), compared to control cultures (66.8%) (Figure 2A, A′, B, B′, E). sFRP treated cells also lack microtubule orientation, with microtubules aligned randomly, including perpendicular to the axis of migration in many cells (Figure 2F, G). sFRP treated cells migrate 2.3 fold further than control cells by 48 hours after embryoid body plating (Figure 2H, I, J). This increase in the extent of migration is not due to increased proliferation as levels of phospho-histone H3 expression, a marker for cells in M phase, were similar for control and sFRP treated outgrowths (Figure S2). Outgrowths recover oriented migration when incubated for an additional 24 hours in sFRP free medium (Figure S3). Interestingly, the actin cytoskeleton in sFRP treated cells was not arranged in thick stress fiber bundles as observed in control cells, leading us to hypothesize that the PCP pathway acts through Rho to promote oriented PE migration (Figure 2A, A′, B, B′). Similar to our previous findings using Rho/Rock inhibitors, we see an apparent decrease in strong focal adhesions in sFRP treated cells based on localization of the focal adhesion protein vinculin (Figure 2C, D)(Mills et al., 2005). Control outgrowths have numerous vinculin rich adhesions localizing to the tips of actin stress fibers in the leading edge along with an abundance of immature focal complexes observed throughout the cell (Figure 2C). sFRP treated cells contain very few focal adhesions, and those that remain are localized to filopodial protrusions (Figure 2D). The filopodia in sFRP treated cells are randomly distributed around the cell and not preferentially localized to the front of the cell, as observed for lamellipodia of control cells (Figure 2B, B′, D, G, G′).
sFRP treated outgrowth cells do not appear to migrate as an adherent sheet. Control outgrowth cells remain adherent to one another throughout their cell-cell boundaries (Figure 2A, A′, C, F, F′, H and Figure S4). In contrast, sFRP-treated outgrowth cells appear to exhibit less extensive cell- cell contact, and associate mainly via filopodia (Figure 2B, B′, D, G, G′, H). These data suggest that sFRP treated outgrowths may be migrating further due to a decrease in focal adhesions co-localizing with actin stress fibers and that the PCP pathway promotes focal adhesion and stress fiber formation, regulating oriented migration.
To determine if the PCP pathway works via Rho in orienting PE migration, we focused on the role of the adaptor protein Daam1, that acts downstream of Wnt/Fz/Dvl exclusively through Rho. The N-Daam mutation contains only the amino terminus of Daam 1, which contains the Rho binding domain and lacks the DVL binding domain, preventing Rho, Daam, and DVL from forming a complex, and so acts as a dominant negative regulator of Rho (Habas et al., 2001; Sato et al., 2006). C-Daam lacks the amino terminus of Daam 1, which is proposed to regulate auto-inhibition of Daam 1, and so acts as a constitutively active regulator of Rho (Habas et al., 2001; Sato et al., 2006). Outgrowth cells expressing N-Daam are not oriented based upon golgi localization (Figure 3A), and migrate further after 72 hours than control cells or C-Daam expressing cells, consistent with our observations that a loss of orientation also results in an overall increase in migration (Figure 3B). Both PE outgrowth cells and F9 stem cells express Daam 1 mRNA, based on RT PCR analysis, and knockdown of Daam1 utilizing siRNA results in a loss of orientation (Figure S5).
If Daam1 acts downstream of Wnt, expression of C-Daam will prevent the loss of cell orientation observed with sFRP treatment. Outgrowths that were transfected with C-Daam were treated with sFRP 24 hours post transfection. C-Daam expressing cells treated with sFRP maintain their orientation relative to the direction of migration, as observed for C-Daam transfected cells, unlike cells treated with sFRP alone (p<0.05)(Figure 3C). Cell migration extent is also similar to control and C-Daam transfected outgrowth in C-Daam transfected plus sFRP treated cells (Figure 3D). These data provide evidence that cell orientation in PE is directed by the PCP pathway through Daam1 since cells treated with N-Daam lack oriented migration, and C-Daam expression can protect PE cells from the loss of orientation mediated by sFRP treatment.
To determine whether Daam 1 acts downstream of Wnt and upstream of Rho in the PCP pathway and to begin to establish the molecular mechanism of Wnt action, we examined the effect of N-Daam expression on the ROCK substrate myosin light chain phosphatase (MLCP). ROCK phosphorylates, and thereby inactivates MLCP, leading to higher levels of MLC activity (Amano et al., 1998; Amano et al., 1996; Chihara et al., 1997). When cells are transfected with N-Daam, a decrease in phosphorylation of MLCP relative to both MLCP (Figure 4A, B) and α-fodrin (Figure 4A, C) is observed, consistent with inhibition of Rho signaling and ROCK activity. C3 transferase is a direct inhibitor of Rho as it ADP ribosylates Rho, inhibiting GTP binding that is necessary for activation (Worthylake et al., 2001). As expected, C3 transferase transfection also leads to decreased phosphorylation of MLCP relative to both MLCP (Figure 4A, B) and α-fodrin (Figure 4A, C). Interestingly, total MLCP levels are decreased in both N-Daam and C3 transferase transfected outgrowths suggesting transcriptional or translational regulation of MLCP expression in addition to regulation via phosphorylation (Figure 4A). Levels of phosphorylated MLCP were similar in both control outgrowths and C-Daam transfected outgrowths along with total levels of MLCP (Figure 4A). Interestingly, there is an increase in fluorescence intensity when comparing phosphorylated MLCP and MLCP for C-Daam (Figure 4B), but not when comparing phosphorylated MLCP and α-fodrin (Figure 4C). These data suggest both that Daam 1 is upstream of Rho and that it plays a role in the modulation of PE migration via MLCP.
To further examine whether Wnt acts via Rho, Rho’s downstream effector ROCK was inhibited using Y27632, a small, permeable pharmacological inhibitor highly specific for ROCK (Ishizaki et al., 2000). Previously, we showed that ROCK inhibition results in increased migration distance, a decrease in focal adhesions in favor of focal complexes, and disruption of actin stress fibers (Mills et al., 2005), properties that also characterize sFRP treated outgrowths. We therefore examined the orientation of Y27632 treated outgrowths and determined that only 32% of the cells have their golgi localized in front of the nucleus, consistent with an absence of orientation (Figure 5A, A′, B, B′, C). In these cultures, the microtubules are not aligned with the direction of migration (Figure 5E, E′, F, F′), and there is also a 2.5 fold increase in migration distance, as previously observed(Mills et al., 2005). On average, control cells migrate 3.4 microns per hour and ROCK inhibited cells migrate 4.4 microns per hour (p<0.05) (Figure 5D). Similar to what we observe with sFRP treated outgrowths, ROCK inhibited outgrowths display more filopodial protrusions than control outgrowths (Figure 5B, B′, F, F′)(Mills et al., 2005). Along with an increase in filopodial protrusions, cells in ROCK inhibited outgrowths are not in close association with one another, though they appear to maintain contact via elongated filopodial projections (Figure 5B, B′, F, F′ and Figure S4). These results are consistent with what we observed with sFRP inhibition, suggesting that cell orientation in F9 PE outgrowths is directed by the PCP pathway acting via Rho.
The PCP pathway can also act via Rac to modulate cell migration (Boutros et al., 1998; Kishida et al., 2004; Rosso et al., 2005; Yamanaka et al., 2002). To determine if Rac plays a role in oriented migration of PE cells, we inhibited JNK, a known downstream effector of Rac, using the pharmacological inhibitor SP600125 (Rosso et al., 2005). Cells remained oriented, relative to the direction of migration, based on golgi localization and microtubule distribution (Figure 5G and data not shown). The extent of migration in SP600125 treated cultures was similar to control outgrowths (Figure 5H). SP600125 treatment inhibited JNK activity at 10 μM, based upon reduced levels of phospo-JNK in treated versus control outgrowths, while total levels of JNK remained unchanged (Figure S6). These data suggest that the PCP pathway in the F9 outgrowth system is modulated through Rho, and not Rac, in promoting oriented PE migration.
To determine if Wnt acts via the canonical pathway as well as the PCP pathway to promote oriented migration of PE cells, we inhibited the canonical Wnt pathway at the level of GSK-3β. In the canonical pathway, GSK-3β phosphorylates β-catenin, targeting it for degradation. Inhibition of GSK-3β allows β-catenin accumulation and stabilization in the cytosol as well as its translocation to the nucleus, promoting canonical signaling. Treatment with the GSK-3β pharmacological inhibitor BIO at up to 3 μM, which promotes an increase in the level of β-catenin, does not affect cell orientation as measured by golgi apparatus localization (data not shown)(Figure 6A, B, C, D).
If the canonical Wnt pathway promotes cell orientation, then BIO treatment should rescue the sFRP mediated loss of outgrowth orientation. Based on quantification of golgi localization, no significant difference in cell orientation between BIO plus sFRP treated and sFRP treated outgrowths is observed (Figure 6A). These data provide no evidence for rescue of oriented migration in sFRP treated outgrowths by canonical Wnt signaling and therefore suggest the canonical Wnt pathway does not play a role in outgrowth orientation.
Our data suggest that oriented PE migration is promoted by the non-canonical Wnt/PCP pathway. Under control conditions, 68.6% of outgrowth cells are oriented, based upon golgi localization (Figure 1B, C, D). When Wnt signaling is inhibited at the level of Wnt/Frizzled using sFRP, we see a loss of oriented migration (Figure 2A, B, G, H, I). These cells also migrate further over 48 hours and lack strong focal adhesions and a robust actin cytoskeleton (Figure 2C, D). Inhibition of the PCP pathway at the level of Daam 1, which acts exclusively within the Rho arm of the PCP pathway, via transient transfection of N-Daam, leads to a loss in cell orientation coupled with an increase in migration distance after 72 hours (Figure 3A, B). In addition, transient transfection with C-Daam prevents loss of cell orientation with sFRP treatment (Figure 3C, D). When ROCK, the downstream effector of the Rho pathway, is inhibited, cells also lose orientation (Figure 5A, B, C, E, F). Our data also suggest no role for Rac or the canonical Wnt pathway.
In fibroblast cultures in the presence of a chemoattractant, ROCK inhibition leads to increased migration in a non-persistent or random manner, where cells will often double back on their migratory trajectory compared to control cells without Y27632, which move directly towards a chemoattractant (Totsukawa et al., 2004). In the F9 outgrowth system, cells can only migrate away from the embryoid body, with no evidence that cells can migrate over or under each other. Based on our observations, this migration can be oriented or not oriented. In fact, further and faster migration lacks the ties that orient migration. Based on our data, lack of cell orientation is not only coupled with increased cell migration, but PE cells also lack focal adhesions, and the actin cytoskeleton is perturbed. The lack of focal adhesions can account for some of the increase in migration speed and distance, since there may be a loss in cell-ECM interactions. The disrupted actin filaments and microtubules may also contribute to faster and further migration. Under these conditions, neighboring cells do maintain some contacts through filopodia, but not to the same extent observed for control cells. Cells lacking orientation may not be able to respond to signals from the ECM and may also exhibit a decrease in focal adhesion maturation which could disrupt integrin recruitment (Hall, 2005; Webb et al., 2003). A decrease in integrin recruitment could lead to defects in inside-out signaling, which coordinates activation and inactivation of small GTPases as well as for cross talk between signaling pathways (Miranti and Brugge, 2002). Focal Adhesion Kinase (FAK) and Src are both downstream of integrins and can facilitate crosstalk between signaling pathways (Hong and Grabel, 2006; Ishibe et al., 2004; Ishibe et al., 2003). Therefore, inhibition and disruption of the PCP may affect other signaling pathways as well.
Unconstrained migration could result in cells missing cues that provide a stop signal. This suggests that faster migration, due to aberrant or perturbed cell signaling, may not be advantageous to a cell. The PCP pathway may act as the braking system when cells migrate as an adherent sheet to their final destination. Cells may migrate further than their target destination if PCP signaling is disrupted, preventing cells from responding to local cues.
Localization of downstream components of the PCP pathway to specific regions of migrating cells may be important for oriented migration. Sequestering components such as focal adhesions, actin stress fibers, microtubules, and the golgi apparatus to a specific region may allow the cell to migrate efficiently in a specific direction and may be necessary for oriented migration. In migrating fibroblasts, downstream of Wnt and Rho, modulators such as MLC can be phosphorylated by ROCK in the central region of the cell, while MLC Kinase via upstream Map Kinase signaling is responsible for MLC phosphorylation at the cell’s periphery (Totsukawa et al., 2004; Totsukawa et al., 2000). Phosphorylated MLC appears to play distinct roles at different sites in migrating fibroblasts, mediating membrane ruffling at the cell periphery and focal adhesion and stress fiber at the cell’s center (Totsukawa et al., 2004; Totsukawa et al., 2000). While little is known about Wnt and Frizzled subcellular localization in mammalian cells, there is more known about subcellular localization in zebrafish and Drosophila. For example, in the presomitic mesoderm during zebrafish gastrulation, the PCP pathway protein Prickle is localized to the anterior portion of the cell while Dvl is localized to the posterior region of the cell during intercalation (Yin et al., 2008). The subcellular localization of PCP pathway components in the PE outgrowth system may support oriented migration.
Although earlier studies suggest that a subset of Wnts interact with specific Frizzled receptors to promote PCP versus canonical signaling, it has been recently demonstrated that some Wnts and Frizzleds are involved in both the PCP and canonical signaling pathways. Wnt 3a has been described in the literature as a ligand for the canonical Wnt pathway (Gordon and Nusse, 2006; Mikels and Nusse, 2006b), yet recent literature shows that Wnt 3a also plays a role in the PCP pathway promoting neurite retraction (Kishida et al., 2004). Mikels and Nusse also found that whether Wnt5a acted via the canonical or non-canonical pathway was dependent on the presence of the canonical coreceptor LRP5/6 (Mikels and Nusse, 2006a). These data show whether a Wnt ligand promotes canonical versus non-canonical signaling and may be context dependent.
Recent work by Kemp et al. characterizes Wnt expression patterns in the developing mouse embryo (Kemp et al., 2005). They demonstrate that Wnts are expressed in the blastocyst stage embryo (Kemp et al., 2005; Lloyd et al., 2003). Wnt 3a and Wnt 10b localize to the core of the blastocyst, while Wnt 6 and Wnt 7b are found in both the inner cell mass and extraembryonic tissues, including the PE cell layer lining the inner surface of the trophectoderm (Kemp et al., 2005; Kemp et al., 2007; Lloyd et al., 2003). These studies suggest potential candidate Wnts responsible for PE oriented migration.
Our preliminary RT PCR analysis (Figure S1) indicate a number of Wnt ligands are expressed in PE outgrowth cultures, some of which (Wnt 7a and 7b) appear to be enriched in the embryoid body. This observation is consistent with the hypothesis that Wnt ligands can act in a repulsive manner, directing cells away from the embryoid body. PE cells respond to the chemorepulsant cue by organizing focal adhesions and actin stress fibers, while orienting microtubules and the golgi apparatus. Wnts as chemorepulsants have been described in Xenopus, where xWnt 11 is expressed in the dorsal ectoderm, providing a repulsant cue to neural crest cells migrating through this region (De Calisto et al., 2005). When xWnt 11 in the dorsal ectoderm is perturbed, neural crest migration is decreased, creating a build up of cells. Alternatively, Wnt could be acting as a chemoattractant, expressed by cells at the outer margins of the outgrowths. Preliminary data suggest the enrichment of specific Wnt ligands in the PE outgrowth, consistent with this hypothesis. Outer margin cells may be different from inner cells, despite the identity of all outgrowth cells as PE based on markers (Mills et al., 2005), and may secrete Wnt ligands while the cells behind the outer ring respond via oriented migration. In other in vivo systems, such as the developing spinal cord and cerebellum, migrating cells respond to Wnts that are secreted from surrounding tissue (Hall et al., 2000; Krylova et al., 2002).
We show here that the first migratory event in the embryo, using the in vitro F9 PE outgrowth system, contributing to yolk sac formation is guided by the PCP pathway acting via Rho, with no role for Rac. Future studies using gain and loss of function approaches will work to aid our understanding of how Wnts acting via the PCP pathway orient PE migration.
We thank Raymond Habas for generously supplying us with all the Daam1 encoding constructs and Keith Burridge for the C3 transferase encoding construct. This work was supported by NIH grant NIH R15 CA090305-02.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.