Regulation of the Rac GTPase pathway by the multifunctional Rho GEF Pebble is essential for mesoderm migration in the Drosophilagastrula

Gastrulation represents the first major morphogenetic event in the development of most multicellular animals. One key aspect of gastrulation is the specification of the presumptive mesoderm cells and their morphogenetic rearrangements within the embryo by dramatic cell movements. In Drosophila, these mesoderm movements can be divided into two major stages: internalization and migration(Costa et al., 1993). Mesoderm cells are derived from a population of ventral cells of the blastoderm epithelium. Internalization involves actin-myosin-mediated apical constriction that promotes formation of a ventral furrow and internalization of the mesoderm as an epithelial tube-like structure(Leptin and Grunewald, 1990; Sweeton et al., 1991). Once internalized, the cells undergo epithelial to mesenchymal transition and mitotic divisions. During the following migration stage, the multilayered cell aggregate spreads out to form a monolayer. At this time, mesoderm cells begin to differentially express transcription factors that identify distinct fates along the dorsal/ventral axis of the embryo(Jagla et al., 2001; Furlong, 2004; Stathopoulos and Levine,2004).

The genetic control of gastrulation has been attributed to the function of a limited number of genes. Internalization is controlled by targets of the zygotically active transcription factors Twist (Twi) and Snail (Sna)(Leptin and Roth, 1994; Leptin, 1999; Seher et al., 2007). Cell signalling through the secreted glycoprotein Folded Gastrulation and the transmembrane protein T48 are both implicated in local activation of Rho1 at the apical cell cortex of invaginating mesoderm cells(Costa et al., 1994; Leptin and Roth, 1994; Barrett et al., 1997; Kolsch et al., 2007). Migration of the mesoderm depends on signalling via the FGF receptor Heartless(Htl) and its two FGF8-like ligands, Thisbe (Ths; FGF8-like1) and Pyramus(Pyr; FGF8-like2) (Shishido et al.,1993; Beiman et al.,1996; Gisselbrecht et al.,1996; Shishido et al.,1997; Gryzik and Müller,2004; Stathopoulos et al.,2004). In most developmental contexts, Htl acts through the adaptor protein Stumps (Sms) via the conserved Ras/Raf/MAP kinase pathway(Michelson et al., 1998a; Vincent et al., 1998; Imam et al., 1999). However,targets of MAPK with a role in mesoderm migration remain elusive, and genetic evidence suggests that activation of MAPK by Htl is neither required nor sufficient for the early morphogenetic events occurring during early mesoderm spreading (Schumacher et al.,2004; Wilson et al.,2005).

A major unresolved issue is how signalling from the FGF receptor is transduced to trigger changes in cell behaviour, which eventually results in the collective cell movements to form a monolayer. Guanine nucleotide exchange factors (GEF) activate Rho GTPases and provide entry points for the regulation of Rho activity in different signalling contexts(Rossman et al., 2005). RhoGEF2 and Rho1 promote the recruitment and assembly of cytoplasmic myosin that drives apical constriction during ventral furrow formation(Barrett et al., 1997; Hacker and Perrimon, 1998; Nikolaidou and Barrett, 2004; Dawes-Hoang et al., 2005). Another GEF called Pebble (Pbl) is indispensable for Htl-triggered cell shape changes and thus represents an excellent candidate that links FGF signalling to the modulation of cell shape(Schumacher et al., 2004; Smallhorn et al., 2004).

Pbl is the single fly orthologue of the human proto-oncogene ect2and plays an evolutionarily conserved role in cytokinesis. Pbl localizes to the cell cortex and activates Rho1, which acts through its effector Diaphanous to promote formation of the contractile actin-myosin ring(Piekny et al., 2005). The two functions of Pbl, cytokinesis and cell migration, can be separated genetically: Pbl function is still required for cell migration in a genetic background in which no mitosis occurs, indicating that Pbl plays independent roles in cytokinesis and cell migration(Schumacher et al., 2004). Whereas protein interactions of Pbl during cytokinesis appear to be highly conserved, to date nothing is known about the mechanisms of Pbl function in mesoderm migration. Pbl belongs to a large family of GEFs that contain a Dbl-homology (DH) domain, which harbours catalytic activity(Whitehead et al., 1997). The function of Pbl in cell migration involves activation of Rho GTPases, as a point mutation in the highly conserved CR3 region within the DH domain compromises its catalytic activity and exhibits equally severe defects as pbl null alleles (Whitehead et al., 1997; Liu et al.,1998; Schumacher et al.,2004; Smallhorn et al.,2004). The only currently known Pbl substrate, Rho1, is unlikely to be involved in mesoderm migration, because Rho1 dominant-negative constructs fail to block mesoderm spreading while efficiently inhibiting cytokinesis (Schumacher et al.,2004).

In the present paper, we define domains of Pbl involved in regulating mesoderm migration. We provide evidence that the catalytic tandem DH-PH domain is essential for mesoderm migration and interacts with Rho1, Rac1 and Rac2. Mis-expression of the tandem DH-PH domain interferes with normal mesoderm migration. Biochemical assays suggest that the interaction between Pbl and Rac is direct. We further show that Pbl localizes to the cell cortex of migrating cells and that the conserved C-terminal tail and the PH domain are important for this cortical localization. These data suggest that Pbl acts through the Rac pathway during mesoderm migration in Drosophila.

Flies were kept under standard conditions. The following stocks were obtained from the Bloomington stock centre: w¹¹¹⁸, twi::Gal4(2x); Dmef2::Gal4,GMR::Gal4, pbl^11D/TM3[ftz::lacZ], pbl³/TM3[ftz::lacZ],rho1^l(2)k07236/CyO, Cdc42⁴/FM6, yw;Rac1^J10,Rac2^Δ,FRT2A,Mtl^Δ/TM3[ftz::lacZ],Df(2R)ED2238/CyO[ftz::lacZ], yw,hs::Flp;cx^D/TM3,w;P[ovo^D1-18]3L,FRT2A/βTub^85D/TM3,UAS::pbl^ΔBRCT_myc/TM3[ftz::lacZ],UAS::RhoL^N25/CyO, UAS::RhoL^V20, UAS::Rac1^V12,UAS::Rac1^N17, UAS::Rac1.L, UAS::Rho1.Sph and EP(3)3118/TM3.

All rescue assays were performed using virgins from a twi::Gal4;pbl³/TM3[ftz::lacZ] stock. Genetic interactions of Pbl with Rac1 and Rho1 were examined using a UAS::pbl^ΔBRCT,pbl³recombinant chromosome crossed in trans to pbl³ with UAS::Rac1.L on the second chromosome or in trans to a recombinant UAS::Rho1.Sph,pbl³ chromosome, respectively. These experiments required distinct crosses to control for the genetic background:for the Rac1 experiment, twi::Gal4;pbl³ crossed to UAS::pbl^ΔBRCT,pbl³was used as control; for the Rho1 control experiment, twi::Gal4;UAS::pbl^ΔBRCT,pbl³was crossed to pbl³.

The pbl cDNA constructs were generated through PCR amplification using the pbl-RA cDNA as a template. Fragments were inserted in frame into the pUAST-HA vector to create C-terminal fusions of the HA epitope. The Pbl-GFP and GFP-Pbl^PH constructs were generated using the Gateway system (Invitrogen) and cloned into the pTGW or pTWG expression vectors (DGRC, Bloomington). The Pbl constructs encode the following amino acids of the Pbl-A protein: Pbl-A 1-853, Pbl^ΔN-term 386-853,Pbl^DH-PH 386-775, Pbl^DH 386-581, Pbl^PH595-719, Pbl^C-term 716-844 and Pbl^ΔC-term 1-720. The Pbl^DH-PH_V531D and Pbl^{ΔN-term_V531D} constructs were generated using the QuikChange II Site-Directed Mutagenesis Kit(Stratagene) to generate a single amino acid exchange (Pbl-A Val⁵³¹to Asp) of the respective construct.

GST fusion proteins were expressed from pGEX plasmids in BL21DE E. coli cells. After lysis in 50 mM Tris-HCl (pH 8), 100 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 1 mM PMSF, the fusion proteins were purified by affinity chromatography (wash buffer, 50 mM Tris-HCl pH 8, 500 mM NaCl, 10 mM MgCl₂, 1 mM DTT; elution buffer, 50 mM Tris-HCl pH 8, 50 mM NaCl,20 mM glutathione, 1 mM DTT). The GEF assay was performed as described previously (Grosshans et al.,2005). Briefly, 0.2 μM GST-GTPases were loaded with[8-³H]GDP (Amersham). The ³H-GDP loaded GTPases were incubated as duplicates with 0.1 μM of the corresponding GEF in the presence of GTP at 25°C for 20 minutes. After nitrocellulose filtration,the radioactivity bound on the filter was determined by liquid scintillation counting.

Embryos were obtained, fixed, stained and sectioned as described previously(Müller, 2008). Microscopy was performed on a Zeiss Axiophot, an Olympus BX61 as well as Zeiss 510 Meta and Leica-SP2 confocal microscopes. Images were processed using Adobe Photoshop and Volocity (Improvision). Heads of adult flies were prepared for scanning electron microscopy as described(Meyer et al., 2006). The following antibodies were used: mouse-anti-Eve, mouse-anti-βGal (both at 1:100, DSHB), rabbit-anti-βGal (1:5000, Cappel), mouse-anti-HA (1:1000,Roche), mouse-anti-GFP (1:800, ABCAM), rabbit-anti-Myc (1:35, Santa Cruz),mouse-anti-CD2 (Serotec), rabbit-anti-Twi (1:1000) and rat-anti-Pbl (1:350). Pbl antiserum was generated against a GST-Pbl-A fusion protein. A 1.6 kb fragment of pbl-RA cDNA (encoding amino acids 1-532 of Pbl-A) was cloned into pGEX-4T-2. The corresponding GST fusion protein was used to immunize rats (Eurogentec, Belgium).

Pbl is a modular multi-domain protein(Fig. 1). The amino(N)-terminal part contains two BRCT domains, which act as protein-protein interaction domains and are required to localize Pbl to the cleavage furrow during cytokinesis (Somers and Saint,2003). The central region of Pbl contains a PEST sequence and a nuclear localization signal, whereas the C-terminal half harbors the catalytically essential DH domain associated with a PH domain, also called tandem DH-PH domain.

The full-length pbl cDNA, when expressed in the mesoderm using the UAS-Gal4 system, rescues both the migration and cytokinesis defect of pbl-null mutants and thus provides an excellent assay for identifying domains of the protein required for Pbl function(Schumacher et al., 2004)(Fig. 2A-C). As a quantitative measure of mesoderm migration, we scored the segmental expression of even-skipped (eve) in a cluster of dorsal mesoderm cells(Frasch et al., 1987). Expression of eve in the dorsal mesoderm represents a reliable marker for proper dorsal mesoderm migration in pbl mutants because, unlike Htl, Pbl is not directly involved in the activation of eve expression in those cells (Carmena et al.,1998; Michelson et al.,1998b; Schumacher et al.,2004; Smallhorn et al.,2004).

A point mutation (V531D) in the DH domain that is known to compromise its catalytic activity abolished the activity of Pbl in cytokinesis and cell migration (Liu et al., 1998; Schumacher et al., 2004; Smallhorn et al., 2004). To identify other functionally important protein domains of Pbl, we tested the rescue potential of a range of tagged deletion constructs(Fig. 1). Expression of Pbl^ΔBRCT, which has both N-terminal BRCT domains deleted,with twi::Gal4 was unable to rescue cytokinesis, but still rescued migration at ∼55% compared with wild type(Fig. 2D,J; Table 1)(Smallhorn et al., 2004). Similarly, a construct lacking the conserved C-terminal tail,Pbl^ΔC-term, did not rescue cytokinesis, but was still able to partially rescue the migration defect to a similar extent as Pbl^ΔBRCT (Fig. 2I,J; Table 1; see below). These data indicate that neither domain alone plays an essential role,because in the absence of either domain there is still a partial rescue. However, as the rescue is not complete, both the BRCT domains and the C-terminal tail must be important for Pbl function in mesoderm migration.

Deletions of N-terminal regulatory domains extending beyond the NLS and PEST sequences create variants of Pbl that are characterized as oncogenic forms of Ect2 as they promote transformation in mammalian cells(Rossman et al., 2005; Saito et al., 2004)(Fig. 1). Expression of Pbl^ΔN-term in the mesoderm of pbl³homozygotes did not rescue the mesoderm differentiation defects(Fig. 2E,J; Table 1). Moreover, even in heterozygous embryos expressing Pbl^ΔN-term the mesoderm cells failed to internalize (see below). By contrast, Pbl^DH-PH lacking the conserved C-terminal tail was able to suppress the pbl mesoderm defect (Fig. 2F,J; Table 1). The V531D point mutation completely abolished the rescuing activity of Pbl^DH-PH;both constructs were expressed at very similar levels(Fig. 2G,J; Fig. 5G,H,K,L). Importantly,the DH domain alone did not exhibit any rescue activity(Fig. 2H,J). Thus, the activity of the tandem DH-PH domain of Pbl requires both a functional DH domain and the presence of the PH domain. Moreover, the rescue capability of the tandem DH-PH domain was dependent on the absence of the C-terminal tail, suggesting that this domain might impinge on the activity of the DH-PH domain.

In addition to the different rescue potentials of Pbl^ΔN-term and Pbl^DH-PH, we noticed that these constructs also exhibited distinct dominant phenotypes. Expression of Pbl^ΔN-term in a wild-type background blocked invagination and the cells failed to undergo cytokinesis(Fig. 3C,D,O,P; Fig. 5F). As null mutants of pbl do not exhibit any defects in mesoderm invagination (S.S. and H.A.J.M., unpublished), Pbl^ΔN-term exhibits an abnormal activity interfering with that process. By contrast, expression of Pbl^DH-PH exhibited defects in mesoderm spreading, whereas cytokinesis was unimpaired (Fig. 3I,J,Q,R). The expression levels of the constructs were in a similar range and even when the level of Pbl^DH-PH was increased using multiple copies of transgenes, the occurrence of phenotypic classes did not change (Fig. 5) (A.v.I. and H.A.J.M., unpublished). Introducing the V531D mutation into either Pbl^DH-PH or Pbl^ΔN-term abolished the dominant activity (Fig. 3K,L and data not shown). Expression of Pbl^ΔBRCT,Pbl^ΔC-term or of the DH domain alone in the mesoderm of wild-type embryos had no adverse effects on development(Fig. 3E-H). Similarly,expression of the C terminus alone did not have any effect on development(data not shown). In summary, the distinct dominant mis-expression phenotypes of Pbl^ΔNterm and Pbl^DH-PH support the idea that the C-terminal tail plays an important role in modulating the activity of the tandem DH-PH domain.

Activation of Rho GTPases is thought to occur by recruiting GEFs to specific subcellular locations. We therefore reasoned that one possible means by which the C-terminal tail might promote Pbl activity would be by controlling its localization. Thus far Pbl has been reported to accumulate at the cleavage furrow during cytokinesis and in the nucleus during interphase(Prokopenko et al., 2000). When endogenous Pbl function was complemented by expression of HA-tagged Pbl-A, we found prominent localization of HA-Pbl to the cytokinesis furrow and the nucleus (Fig. 4G-O). Importantly, HA-Pbl was also associated with the cell cortex and cell protrusions of migrating mesoderm cells(Fig. 4A-F; see Movie 1 in the supplementary material). By contrast, our Pbl antiserum revealed prominent staining of the nuclei, but only very weak staining of cell borders in wild-type embryos, suggesting that the fraction of total Pbl protein at the cell cortex might be low (Fig. 4P,Q). To examine the dynamics of Pbl distribution in vivo, we generated eGFP-tagged Pbl. In mesoderm cells, eGFP-Pbl was present in the nuclei but expression was too low to detect cortical Pbl. However, in migrating haemocytes, levels of eGFP-Pbl were much higher and the protein was localized to the cell periphery and actin-rich microspikes as well as the nucleus (Fig. 4R; see Movie 2 in the supplementary material). Taken together, these data demonstrate that in addition to its prominent nuclear localization, a subpopulation of Pbl localizes to the cell cortex and actin-rich structures.

We next sought to determine the domains that are required for cortical localization of Pbl in mesoderm cells. Pbl^ΔBRCT was localized similarly to wild-type Pbl, whereas the two BRCT domains alone localized to the cytoplasm (Fig. 5A,B)(A.v.I. and H.A.J.M., unpublished). Thus, the BRCT domains appear not to be involved in the association of Pbl with the cell cortex in interphase cells. Pbl^ΔCterm was present at high levels in the nucleus, but low amounts in the cytoplasm and cell cortex(Fig. 5C,D). The importance of the C-terminal tail for the cortical localization was even more evident in constructs lacking N-terminal PEST and NLS sequences, in which cytoplasmic levels are accumulating. Pbl^ΔNterm exhibited a strong accumulation at the cell cortex (Fig. 5E,F). Even when the C-terminal tail alone was expressed it was enriched at the cell cortex of mesoderm cells, suggesting that this domain is to some extent sufficient for cortical localization(Fig. 5O,P).

Despite the importance of the C-terminal domain, constructs lacking this domain still exhibit some cortical localization. Pbl^DH-PH, which lacks the C-terminal tail, was also localized at the cell periphery in a conspicuous punctate fashion - similar to that described for the tandem DH-PH domain of Ect2 in mammalian cells (Fig. 5G,H) (Solski et al.,2004). This result suggested that the PH domain might contribute to membrane association of Pbl. Indeed, the DH domain alone was localized in the cytoplasm, indicating that the PH domain is required for the punctate cortical localization of Pbl^DH-PH(Fig. 5M,N). Moreover, a Pbl PH-GFP fusion protein was enriched at the cell cortex, suggesting that the PH domain was to some extent sufficient to mediate cortical localization(Fig. 5Q,R). In summary, these localization studies indicate that both the C-terminal tail and the PH domain are involved in the localization of Pbl to the cortical cytoplasm.

As Pbl^ΔCterm can still partially rescue mesoderm defects in pbl mutants, cortical localization through the C-terminal tail appears to be important but not essential for the activity of Pbl in cell migration. By contrast, Pbl^ΔCterm was unable to rescue cytokinesis in pbl mutants (Fig. 6G-J). The failure of Pbl^ΔCterm in rescuing cytokinesis was not due to a requirement for subcellular localization. Pbl^ΔCterm was localized to the cleavage furrow of dividing cells as in the wild type (Fig. 6A-F). These data indicate that the C-terminal tail is required for the activation of Rho1 during cytokinesis and suggest that the C-terminal domain might play a more direct role in regulating the activity of the DH domain. Thus, the Pbl^ΔCterm construct uncouples the dual functions of Pbl, in cytokinesis and cell migration, and supports the previous model that Pbl activates a different Rho pathway during mesoderm migration(Schumacher et al., 2004).

We sought to determine the Rho GTPase specificity of Pbl in vivo by testing genetic interactions in the developing eye using GMR::Gal4. The dominant activities of Pbl^DH-PH and Pbl^ΔNterm were both dependent on a functional DH domain. Thus, the overexpression phenotypes are most probably consequences of over-activating the respective Rho GTPase pathway downstream of Pbl. As Pbl^DH-PH is able to partially rescue the mesoderm defect in pbl mutants, it represents an excellent tool with which to identify the substrate of Pbl in cell migration through testing genetic interactions with Rho GTPases. Expression of Pbl^DH-PHresults in a rough eye phenotype that is characterized by a reduction of the size of the eye and highly abnormal ommatidial structures(Fig. 7A,B). Expression of Pbl^DH-PH_V531D did not produce any phenotype, indicating that the Pbl^DH-PH rough eye phenotype is a result of overactivation of downstream Rho GTPase pathways (Fig. 7C). Moreover, expression of Pbl^DH-PH in a pbl³ heterozygous background mildly suppressed the rough eye phenotype (Fig. 7D). Therefore, Pbl^DH-PH probably acts in the normal Pbl pathway, but is hyperactive. Hence, it should be possible to suppress the eye phenotypes similarly by reducing the expression level of the target GTPases of Pbl.

Pbl^DH-PH interacted with Rho1, as a reduction of the Rho1 gene dose resulted in suppression of the rough eye phenotype(Fig. 7E). This result was expected, as it has been shown before that Pbl can directly bind Rho1(Prokopenko et al., 1999). Co-expression of dominant versions of RhoL or heterozygosity of a loss-of-function mutation in cdc42 did not modify the rough eye phenotype (Fig. 7F-H). However,in flies heterozygous for a triple mutation in Drosophila Rac GTPases(Rac1^J10, Rac2^Δ and Mtl^Δ), the Pbl^DH-PH rough eye phenotype was strongly suppressed (Fig. 7I). Moreover, co-expression of either Rac1 or Rac2 with Pbl^DH-PH strongly enhanced the rough eye phenotype(Fig. 7J; data not shown). These results suggest that overexpression of Pbl^DH-PH in the eye promotes activation of Rac GTPases. We conclude that Pbl^DH-PHbehaves as a gain-of-function allele and exhibits genetic interactions consistent with activation of Rho1 and Rac pathways.

Expression of Pbl^ΔNterm in the embryo affected two Rho1-dependent processes, cytokinesis and invagination, suggesting that this construct might specifically overactivate the Rho1 pathway in the cell. Unfortunately, expression of Pbl^ΔNterm in the eye results in lethality at pupal stages. However, at a lower temperature (18°C),lethality occurred at the pharate adult stage [0% eclosion (n=43); Fig. 7K]. The lethality is suppressed by removal of one functional copy of Rho1, as those flies eclosed and displayed a strong rough eye phenotype [20% eclosion(n=54); Fig. 7L]. No suppression of the Pbl^ΔNterm lethality was observed in flies heterozygous for Rac1^J10, Rac2^Δ, Mtl^Δ [0% eclosion n=42)]. These results indicate that Pbl^ΔNterm specifically activates the Rho1 pathway and support the idea that the embryonic phenotype produced by Pbl^ΔNterm is caused by overactivation of the Rho1 pathway.

The genetic interactions demonstrated that the tandem DH-PH domain of Pbl activates Rho1 and Rac GTPases. To determine whether Pbl is capable of directly interacting with Rac GTPases, we performed functional guanyl-nucleotide exchange assays using GST fusion proteins of Rho1, Rac1,Rac2, Mtl, RhoL and Cdc42, the DH domain of Pbl, and the first DH domain of Trio as a control. The GTPases were loaded with ³H-GDP and incubated with the respective DH domain or GST as a control in the presence of GTP. The release of ³H-GDP reflects a measure of the exchange activity of a specific DH domain towards a given GTPase. The first DH domain of Trio, an exchange factor for Rac GTPases, exhibited a strong preference for Rac1, Rac2 and Mtl, whereas Trio did not promote nucleotide exchange for Rho1 or Cdc42 and showed a weak activity for RhoL(Fig. 8). GST-Pbl^DHpromotes GDP exchange from Rho1, consistent with our genetic data and previously reported binding studies(Prokopenko et al., 1999). Strikingly, we also detected an activity of GST-Pbl^DH for Rac1 and Rac2 (Fig. 8). The fact that the activity for Rac1 and Rac2 was weaker than for Rho1 might reflect a requirement of the PH domain in promoting full activity or specificity of the DH domain of Pbl. The insolubility of the bacterial GST-Pbl^DH-PHfusion protein prohibited us from directly testing this possibility. Together,these data indicate that the DH domain of Pbl is able to use Rac1 or Rac2 as a substrate and in conjunction with the genetic interactions suggest that Pbl promotes exchange activity towards multiple substrates, including Rac GTPases.

The genetic and biochemical data are consistent with the model that Pbl functions through activation of the Rac pathway to promote mesoderm spreading. As the compound eye represents a heterologous system, we first wanted to investigate whether the genetic interactions between Pbl and Rho1 and Rac also occurred in the embryonic mesoderm. We therefore tested whether Rho1 or Rac1 variants are able to enhance the moderate phenotype produced by the weak loss-of-function allele pbl^11D. Expression of a dominant-negative construct (Rac1^N17) enhanced the mesoderm phenotype of pbl^11D(Table 2). Overexpression of constitutively active Rac1^V12, but not Rho1^V14 enhanced the mesoderm phenotype of pbl^11D mutant embryos, consistent with an adverse effect upon over-activation of the Rac pathway(Table 2).

In a second set of experiments, we asked whether Rac1 was able to enhance rescue activity of Pbl^ΔBRCT. Overexpression of Pbl^ΔBRCT provides enough activity to suppress the pbl³ mesoderm phenotype substantially without producing a dominant phenotype, suggesting that this construct is present in the cells at near physiological levels (Fig. 2J; Fig. 3E,F; Table 1). Co-expression of wild-type Rac1 together with Pbl^ΔBRCT leads to a significant enhancement of the rescue of pbl mutants by Pbl^ΔBRCT(Table 3). When wild-type Rho1 is co-expressed with Pbl^ΔBRCT, there was no change in the strength of the rescue of the pbl phenotype by Pbl^ΔBRCT (Table 4). This experiment indicates that Rac1 interacts with Pbl^ΔBRCT and can promote its ability to rescue the pbl³ migration defect. Together, the genetic interactions strongly support a role of Pbl to activate the Rac pathway in mesoderm spreading.

We next asked whether mesoderm spreading depends on Rac GTPases and analyzed maternal-zygotic mutants lacking Rac1 and Rac2 with reduced maternal Mtl expression(Hakeda-Suzuki et al., 2002; Ng et al., 2002). In Rac1 Rac2 Mtl mutant embryos, the mesoderm never migrated dorsally, as assessed by Twi staining (Fig. 9A,B). The phenotype is similar to the mesoderm spreading defects seen in embryos lacking both FGF ligands FGF8-like1 and FGF8-like2 (Fig. 9C)(Gryzik and Müller,2004). These results extend previous findings that embryos with reduced maternal expression of Rac GTPases fail to initiate mesodermal-ectodermal contact after invagination(Wilson et al., 2005). Moreover, when expressed in the mesoderm of wild-type embryos, Rac1^V12 affects mesoderm spreading(Fig. 9D-H). These data indicate that tight spatiotemporal regulation of the Rac pathway plays an important role in mesoderm migration.

The Rho GEF Pbl provides one of the few molecular links between the proximal FGF receptor signalling events and the regulation of cell shape changes. We have previously characterized the loss-of-function phenotype of pbl mutants, showing that Pbl acts in a pathway downstream or in parallel to Htl-dependent MAP kinase activation(Schumacher et al., 2004). Here, we used genetics and biochemistry to determine the regulation of Pbl and its downstream Rho GTPase pathways in migrating cells. Our data demonstrate that Pbl partially localizes to the cell cortex of mesoderm cells and functionally interacts with Rac GTPases in this process.

We show that the tandem DH-PH domain of Pbl is essential for cell migration and employs not only Rho1, but also the Rac pathway. Several lines of evidence strongly suggest that Pbl acts through Rac GTPases during mesoderm migration. The dominant rough eye phenotype induced by Pbl^DH-PH is sensitive to gene doses of Rac GTPases. Expression of constitutively active or dominant-negative Rac1 but not Rho1 enhances the mesoderm phenotype in the hypomorphic pbl^11D allele. Moreover, co-expression of Rac1, but not of Rho1, promotes the suppression of mesoderm migration defects by Pbl^ΔBRCT in pbl-null mutants. In addition, we provide biochemical data that strongly suggest the Rac pathway as a direct target of Pbl.

Pbl has previously been reported to localize to the nucleus in interphase cells. Nuclear localization was interpreted as a means of storing the protein until rapid release at mitosis (O'Keefe et al., 2001). In cultured cells and C. elegans zygotes,homologues of Pbl localize at the cell cortex, e.g. cell junctions or the anterior cortex in the nematode zygote(Liu et al., 2004; Jenkins et al., 2006). We detected functional Pbl-HA in the nucleus and the cytoplasm, including membrane protrusions. These data are consistent with the model that Pbl activates Rac GTPases at the cell cortex during cell migration.

Our study identified two domains, the conserved C-terminal tail and the PH domain, as candidates to mediate the association of Pbl with the cell cortex in interphase cells. The use of N-terminally deleted constructs facilitated these studies, because the respective proteins were excluded from the nucleus as they lack the NLS. Either domain alone is sufficient to localize to the cell cortex, and deletion studies suggest that both domains are crucial for cortical localization. We propose that the PH domain and the C-terminal tail might cooperate in localizing Pbl to the cell cortex. DH domain associated PH domains are essential for GEF function and are known to promote binding to specific membrane subdomains enriched in phosphoinositides(Lemmon, 2008). An attractive model therefore is that the PH domain provides specificity by targeting Pbl to membrane domains enriched for particular phospholipids, whereas the C-terminal tail functions in anchoring Pbl to the cell cortex. In addition, binding to phospholipids might promote the specific exchange activity of the tandem DH-PH domain, as described for other Dbl family GEFs(Snyder et al., 2001; Rossman et al., 2003).

It is difficult to address the issue of whether cortical localization is important for the function of Pbl in mesoderm migration. The reduced rescuing capability of Pbl^ΔC-term is consistent with a correlation of cortical localization through the C-terminal domain and the function of Pbl in cell migration. A more stringent experiment would involve the generation of a construct that lacks the PH and C-terminal domains for membrane association. However, as PH domains are essential for DH domain function in vivo, deletion of the PH domain will abolish activity in any case, as we have shown for the constitutively active DH-PH construct. Such an analysis would require a way to uncouple the activities of the PH domain that promote the exchange activity and membrane-phospholipid binding. It will therefore remain important to determine whether the function of the PH domain involves its interaction with lipid substrates or directly promotes the activity of the DH domain in migrating cells.

The inhibition of invagination and cytokinesis by Pbl^ΔNterm is probably caused by disruption of the local activation of Rho1 at the cell cortex. During invagination and cytokinesis,the Rho1 pathway is activated locally: in the apical domain of the mesoderm cells to trigger apical constriction or at the cell equator of the dividing cell to promote assembly of the contractile ring. As Pbl^ΔNterm strongly accumulates at the cortex in a non-polarized fashion, it might activate Rho1 ectopically throughout the cell cortex and thereby overriding any polarizing cues for local activation.

The dramatic differences in the overexpression phenotypes of Pbl^DH-PH or Pbl^ΔNterm suggest an important function of the C-terminal tail in controlling the biochemical activities of the tandem DH-PH domain. Strikingly, Pbl^ΔNterm genetically interacts with Rho1, but not with Rac GTPases, supporting the idea that the C-terminus promotes the exchange activity towards Rho1. We propose that in the mesoderm cells this activity of the C-terminal domain is antagonized to activate the Rac rather than to the Rho1 pathway. In the presence of the NLS and PEST motifs, the cytoplasmic levels of Pbl are low and allow for this regulation to occur, whereas the oncogenic forms lacking these motifs are present in the cytoplasm at high levels and might escape regulation. Thus,constructs that lack the C-terminal tail promote interaction with Rac and rescue Rac-dependent mesoderm migration. This model is also supported by the observation that the C-terminal domain is essential for Rho1 activation, but not for Pbl localization in dividing cells. The same construct,Pbl^ΔC-term, is still able to rescue Rac-dependent migration defects. Thus, deletion of the C-terminal tail uncouples activation of Rho1-from Rac-dependent processes and suggests that in the absence of the negative interaction with the C-terminal tail, the tandem DH-PH domain promotes activation of Rac.

Although many receptor tyrosine kinases signal through Rho GTPases, only few FGF receptors have been reported to regulate Rho GEFs(Schiller, 2006). One attractive model is that FGF signalling mediates post-translational modification of the C-terminal tail to trigger the switch in the differential interaction with Rho1 and Rac GTPases. The sequence of the C-terminal tail contains several conserved putative phosphorylation sites that might represent targets for FGF signalling. Interestingly, the exchange factor specificity of oncogenic ect2 for GTPase substrates depends on the C-terminal tail of the protein (Solski et al.,2004). Identification of proteins that interact with the C-terminal domain might shed light on its role in controlling selectivity for distinct GTPase pathways. Such studies will be important to advance our understanding of the mechanism of the transforming potential of Pbl, as well as its mechanism of action in cell polarity and cell migration.