As the fertilized Xenopus egg undergoes sequential cell divisions to form a blastula, each cell develops a network of cortical actin that provides shape and skeletal support for the whole embryo. Disruption of this network causes loss of shape and rigidity of the embryo, and disrupts gastrulation movements. We previously showed that lysophosphatidic acid (LPA)signaling controls the change in cortical actin density that occurs at different stages of the cell cycle. Here, we use a gain-of-function screen,using an egg cDNA expression library, to identify an orphan G protein-coupled cell-surface receptor (XFlop) that controls the overall amount of cortical F-actin. Overexpression of XFlop increases the amount of cortical actin, as well as embryo rigidity and wound healing, whereas depletion of maternal XFlop mRNA does the reverse. Both overexpression and depletion of XFlop perturb gastrulation movements. Reciprocal rescue experiments, and comparison of the effects of their depletion in early embryos, show that the XLPA and XFlop signaling pathways play independent roles in cortical actin assembly, and thus that multiple signaling pathways control the actin skeleton in the blastula.
Introduction
The actin skeleton controls many aspects of cell behavior (reviewed by Revenu et al., 2004). However,little is known of how cortical actin is assembled in early embryos, or how it is remodeled as cell behavior changes during morphogenesis. The early Xenopus embryo is a useful system in which to analyze this process. The expression of individual cytoskeletal elements can be either up or downregulated during early development by mRNA injection or antisense oligodeoxynucleotide-mediated mRNA depletion, respectively, and the consequences to the cytoskeletal pattern and to normal development assayed in an in vivo system (Quaas and Wylie,2002; Robb et al.,1996; Torpey et al.,1992; Vernos et al.,1995).
We have previously shown that the Arm repeat protein Plakoglobin(γ catenin) is both necessary and sufficient for assembly of the cortical actin skeleton in early Xenopus embryos, and is downstream of the signaling intermediate cdc42(Kofron et al., 2002). In order to identify more fully the pathways leading to cortical actin assembly,an arrayed two- to four-cell cDNA library was divided into mRNA pools, which were expressed in Xenopus embryos by microinjection at the two-cell stage. Pools that caused effects similar to plakoglobin/cdc42 overexpression were split in a matrix fashion (Grammer et al., 2000) to identify the active mRNAs. One of these mRNAs was found to encode a novel G protein-coupled receptor (GPCR), whose sequence is related to a subfamily of GPCRs, the OGR1 (Ovarian cancer G protein-coupled Receptor 1) subfamily. Overexpression of this mRNA dramatically increased the density of the actin skeleton in the early Xenopus embryo, whereas oligodeoxynucleotide-mediated depletion of the endogenous mRNA had the reverse effect, and caused embryos to lose their shape and rigidity. On the basis of the phenotype caused by its depletion, we have given this protein the provisional name of XFlop. In a previous paper, we showed that intercellular signaling is required to maintain the normal density of the cortical actin skeleton in Xenopus, and that the expression of LPA receptors is both necessary and sufficient to maintain the normal density of cortical actin filaments in interphase cells (Lloyd et al., 2005). Here, we compare the roles of these two ligand receptors, and conclude that they each play independent roles in generating the normal pattern of actin filaments in the embryo. Thus, multiple cell signaling pathways are involved in this process.
Materials and methods
Gain-of-function library screen
Approximately 16,000 colonies from a Xenopus two- to four-cell cDNA expression library were arrayed into 38 384-well plates using a QBot(Genetix). Each master plate was split into four 96-well plates (denoted as nA, nB, nC, nD, where n=1-38). After 2-3 days incubation, colonies from each 96-well plate were pooled for plasmid isolation and capped RNA synthesis. 5 ng mRNA from each pool was injected into the animal cytoplasm of two-cell-stage Xenopus eggs. Injected embryos were reared in 0.2×MMR containing 2% Ficoll, and were analysed for cortical actin assembly at the late blastula stage (Stage 9)(Nieuwkoop, 1975). To identify the active colonies from the first-round mRNA pools, active pools from the master plates were amplified in triplicate in 96-well plates (one for the 2nd round screen, one for the 3rd round screen, one for backup). In the 2nd round screen, the eight wells from each column of a 96-well plate were pooled, and 1-2 ng of mRNA from each column was analysed as before. In the 3rd round screen, mRNAs were transcribed from the eight single wells from the active column. Finally, if a mRNA from a single colony mimicked the effect of the original pool, it was sequenced for further analysis. All 384-well and 96-well plates were from Genetix. Each well was filled with 200 μl LB medium containing 8% glycerol, and plates were kept in a humidified chamber to minimize evaporation. The alkali lysis protocol was used for all library plasmid isolation. The plasmids were linearized with NotI and transcribed with SP6 Message Machine kit (Ambion).
Cortical actin-assembly assays and pixel-intensity analysis
Cortical actin assembly assays were carried out exactly as described previously (Kofron et al.,2002; Lloyd et al.,2005). The size of each confocal image was set as 512×512 pixels. Z stacks, that include cortical actin of all the superficial cells of the blastocoelic surface of the cap, are shown in all low power (20×)pictures of entire caps. The intensity of each pixel was calculated by Zeiss 510 software in a linear range of 128-4095 (2-256 for the older version of software). The overall amount of the cortical actin for each cap is expressed as mean pixel intensity across the whole cap. Five to eight caps from each treatment were analyzed in all experiments, and the results presented as mean pixel intensity ±s.d. Treatments were compared for statistical significance by Student's t-test.
mRNA depletion and host transfers
These were carried out as described previously(Kofron et al., 2002). Briefly, stage VI oocytes were obtained by manual defolliculation and cultured in oocyte culture medium (OCM). To deplete the maternal store of XFlop mRNA,8-12 ng of HPLC-purified antisense phosphorothioate-modified oligodeoxynucleotides (18-mer) were injected into each oocyte. After 2-3 days of culture at 18°C, oocytes were matured and transferred using host transfer techniques (Holwill,1987). The sequences of two antisense oligonucleotides used to deplete maternal XFlop mRNA are: oligo 1s,5′-A*A*-G*GGAACACTGTAG*C*C*A-3′;oligo 5s,5′-G*T*T*GTACGTTTTGGC*T*G*G-3′,where * indicates the phosphorothioate diester bond substitution.
mRNA synthesis and injection
Plasmids from the expression library, and those encoding pCS2+XFlop ΔC and FL were linearized with NotI and transcribed in vitro using SP6. XFlop mRNA (10-250 pg) was injected into the animal cytoplasm at the two- or eight-cell stage. mRNA-injected embryos were reared in 0.2×MMR containing 2% Ficoll, and transferred to 0.1×MMR at the mid-blastula stage.
Total RNA isolation and real-time RT-PCR
These were performed exactly as described previously(Kofron et al., 2002).
Immunostaining
Microtubules and cytokeratin filaments in control and XFlop-depleted animal caps were detected as described previously(Kofron et al., 2002).
Results
Gain-of-function screen for maternal genes involved in actin assembly
We constructed a Xenopus two- to four-cell-stage cDNA expression library, using oligo dT-primed polyA+ RNA directionally cloned into the Xenopus expression plasmid pCS2+. The library contains 1×106 independent clones, with an average insert size of>1 kb. The library was plated, and mRNA pools generated, as decribed in the Materials and methods. mRNA pools were injected in 5 ng doses (representing 50 pg of each mRNA) into the animal cytoplasm of two-cell-stage embryos. The embryos were allowed to develop to the late blastula stage(Fig. 1A). Animal caps(Fig. 1B) were excised and cultured until the debris from dead cells could be removed (10 minutes). Caps were then fixed, Alexa-488 phalloidin stained, and examined under the confocal microscope for the presence, pattern and density of cortical actin staining. The remainder of the embryos after excision of the caps, the bases(Fig. 1C), were assayed for rigidity, shape, and the ability to heal the wound left by the excision of the animal cap. Pooled mRNAs from eight of the 96-well plates caused dramatic changes in the actin skeleton. Colonies were pooled from each column of wells from positive 96-well plates. mRNA was transcribed from these, and injected in doses of 1 or 2 ng (representing 125 pg and 250 pg, respectively, of each mRNA) into embryos at the two-cell stage, and assayed as before. Finally, mRNA was transcribed from each well of the positive columns, and doses of 100 pg and 200 pg assayed to identify the active maternal cDNA.
In this paper, we report on the role in actin assembly of one clone identified in this screen, 12D10D, which encodes a novel G protein-coupled receptor. Fig. 1D shows that embryos injected with mRNA from pool 12D were more rigid and spherical than controls; the wound made by excision of the animal cap is arrowed. Fig. 1E shows the dramatic enhancement in wound healing, 45 minutes after animal cap removal, in embryos injected with mRNA from the single column 12D10 of colonies. Fig. 1F shows that animal caps dissected from late blastulae injected at the two-cell stage with 100 pg of mRNA from the single colony 12D10D healed faster than control caps, and contain dramatically increased cortical actin. The same results were seen in embryos injected with the original pool, the active column, and the single active colony. In each case, batches of 5-10 embryos were examined, and five to six caps were analyzed. In each case the results were identical to the representative samples shown in Fig. 1D-F.
Based upon the effect of its depletion in the early embryo (see below), we temporarily named this protein XFlop, until a more systematic name, based upon function and structure, can be allocated.
Characterization and expression pattern of XFlop
Three independent colonies from 12D10D were sequenced in both strands. The sequence has been deposited in GenBank (AY766161). The XFlop cDNA sequence is 1.5 kb long. The predicted open reading frame encodes a protein of 348 amino acids (Fig. 2A),which contains a seven-pass transmembrane domain, suggesting that XFlop is a G protein-coupled receptor (GPCR). Furthermore, XFlop contains a putative N-linked glycosylation site (N*QSC, aa4-7) in the N-terminal extracellular domain (aa1-aa18), and a conserved acidic-arginine-aromatic triplet (D-R-F) in the N-terminal extremity of the second cytoplasmic loop(aa111-aa113), each characteristic of GPCRs(Bause, 1983; Wheatley and Hawtin, 1999). The acidic-R-aromatic triplet has been implicated in G-protein coupling(Oliveira et al., 1994). XFlop shows 46-53% sequence homology with members of a recently proposed subfamily including mammalian ovarian cancer G protein-coupled receptor 1 (OGR1), G protein-coupled receptor 4 (GPR4), G2 accumulation (G2A) and T cell death-associated gene 8 (TDAG8) (Fig. 2B) (Heiber et al.,1995; Xu,2002).
Recently, another member of the GPR4 family was identified in Xenopus (Xgpcr4) (Chung et al.,2004). Fig. 2Ashows that the putative amino acid sequences are very divergent in the C termini of Xgpcr4 and XFlop, which suggests that they are encoded by different genes that are not pseudoalleles of the same gene. We also found that Xgpcr4 and XFlop are encoded by different genomic sequences even though they reside in the same chromosome (Fig. 2C). Comparison with the genomic scaffold sequence (Scaffold 3407,JGI) shows that the ORF of XFlop is encoded by a single exon, a feature shared by mammalian GPR4 proteins (Heiber et al.,1995).
Comparison of the sequence of the original 12D10D clone with the contig sequence from the NIBB database suggested that there might be a single adenosine insertion at position 947 in the open reading frame of the 12D10D clone (Fig. 2D, highlighted in green), which would result in a premature translation stop, affecting the C-terminal 30 amino acids (Fig. 2A). To establish which coding sequence is expressed in the embryo, we used the contig-coding sequence to PCR-amplify the putative full-length coding sequence from a late blastula cDNA preparation. Sequencing of the amplified cDNA showed that the original 12D10D clone did indeed contain a single adenosine insertion at nucleotide 947. We further compared this cDNA sequence with the Xenopus tropicalis genomic sequence (scaffold 3407)available from the JGI database, which further supported the notion that the adenosine insertion in the 12D10D clone is an artifact of library construction. Fig. 2D compares the coding sequences of the 12D10D clone and the full-length cDNA isolated from late blastula mRNA, and shows that the 12D10D clone has a stop codon earlier than in the full-length sequence. We therefore denoted the original 12D10D sequence as XFlop ΔC. Interestingly, these two forms of the protein had exactly the same activities in regulating actin assembly (see below), suggesting that the C-terminal 30 amino acids are not required for the action of XFlop on the actin skeleton.
To assay the spatial and temporal expression pattern of XFlop mRNA during early development, we made cDNA from embryos frozen at different stages, and assayed the level of XFlop mRNA by quantitative real-time PCR. XFlop was expressed maternally, and ubiquitous expression was observed in the blastula throughout early development(Fig. 2E).
Overexpression of XFlop accelerated wound healing and enhanced actin assembly
We transcribed both XFlop ΔC and full-length XFlop mRNA and microinjected them, in doses of between 10 and 250 pg per embryo, into the animal cytoplasm at the two-cell stage. After excision of animal caps at the late blastula stage, the remainder of the embryos (the bases) were found to heal significantly faster than controls(Fig. 3A,E, compare blue and yellow bars), in a dose-dependent fashion. The blastocoel roofs of XFlop-injected embryos were also thicker than those of controls(Fig. 3B). The mean(±s.d.) thickness of the blastocoel roofs (the distance from the middle of the roof to the animal pole) was measured in ten embryos from photographs of fixed, bisected embryos. The mean roof thickness (±s.d.) of XFlop-overexpressing embryos was 195±37 μm, which compared with 73±25 μm for control embryos.
The effect of XFlop overexpression on cortical actin is shown in Fig. 3C,D. At low magnification, the level of staining in the whole cap is dramatically increased, as shown by the pixel intensity charts(Fig. 3C). The mean pixel intensity of five to eight caps (±s.d.) is shown in Fig. 3D, together with the P-value from the comparison by Student's t-test of the control, and mean and standard deviation of a 250 pg dose. Increasing doses of XFlop mRNA caused a corresponding increase of phalloidin binding, and therefore of F-actin concentration in the cell cortices. Examination of individual caps showed that not all cells have increased actin staining (arrow in Fig. 3C). This could be due to one of two possibilities. First, the injected mRNA may not have filled this area of the injected embryo. This is unlikely, as cells surrounding the injected one have increased actin staining. Second, and more likely, we have shown previously that cells undergoing cytokinesis change their cortical actin to a less-dense network (Lloyd et al.,2005). As cells divide normally in XFlop-overexpressing embryos,it is likely that they go through the same cycles of actin assembly and disassembly during the cell cycle. However, we have not experimentally discriminated between these two possibilities.
Higher magnification images of control and XFlop-overexpressing animal caps are shown in Fig. 4. Cortical actin is increased in both inner (Fig. 4A,B) and outer (Fig. 4C,D) surfaces of the caps. Interestingly, all actin-containing structures show an increased intensity of staining. The cortical network is denser, regions where cells overlap show increased staining, and there is an increased density of small projections from the cell surfaces. The fact that no actin-containing structures are enhanced at the expense of others, and that there is not any change in the type of network or processes, suggests that XFlop may control the overall amount of F-actin polymerization, rather than controlling any specific actin-containing structure in the embryonic cells.
XFlop and XLPA receptor overexpression both cause increased cortical actin assembly. We sought to exclude the possibility that this is a general property of G protein-coupled receptor-mediated signaling by testing the effects of overexpression of two other G protein-coupled receptors: TDAG8 (which is in the same sub-family in mammals) and the distantly related GPR160. In each case, 200 pg of full-length mRNA was injected into two-cell-stage embryos, and the animal caps assayed as above at the late blastula stage. In both cases,there was no upregulation of cortical actin (data not shown), showing that these are specific properties of XFlop and LPA signaling.
XFlop is required for actin assembly in the early Xenopusembryos
We then asked whether XFlop is necessary, as well as sufficient, for actin assembly in early Xenopus embryos. Twelve antisense deoxyoligonucleotides were tested for their ability to deplete maternal XFlop mRNA. Two of these (oligo 1s and 5s) reduced the maternal store of XFlop mRNA to 10%, or less, of the levels in control oocytes, when doses of 8-12 ng per ooctye were used (Fig. 5A). These oligos were synthesized with the three 5′ and three 3′ phosphorodiester bonds replaced by phosphorothioate linkages. These chimeric oligos were injected, either individually or together, into cultured oocytes, which were then fertilized using the host transfer method(Holwill, 1987). XFlopmRNA levels remained at low levels until the gastrula stage, after which zygotic XFlop mRNA began to be synthesized(Fig. 5A). Therefore, the embryos can develop at least until the mid-gastrula stage with a significant depletion of XFlop mRNA.
At the blastula stage, XFlop-depleted embryos looked morphologically normal, but when the vitelline membranes were removed, they became less spherical than control embryos (Fig. 5B,C), which made them look a little larger when viewed from above(Fig. 5B, compare the lengths of the lines of control and depleted embryos). When fixed and bisected along the animal/vegetal axis, and viewed from the side, the less spherical architecture becomes more obvious (Fig. 5C). After excision of the animal caps, the bases healed more slowly than control bases did. After 45 minutes, the wound margins of the control bases were smooth, and the wound diameter had contracted. By contrast,the wound margins of XFlop-depleted embryos were still open(Fig. 5D). These data are quantitated and compared with the effect of XFlop overexpression in Fig. 3E. Animal caps from these embryos were cultured for 10 minutes, fixed, stained with Alexa 488-coupled phalloidin, and their blastocoelic surfaces examined by confocal microscopy. Fig. 5E shows low magnification images of whole caps (upper panels), and high magnification images (lower panels) of control and XFlop-depleted animal caps. The overall level of cortical actin staining was significantly reduced by XFlop depletion. Quantitation was performed by comparing the mean pixel intensity(±s.d.) of six caps from each treatment(Fig. 5F). In addition, none of the XFlop-depleted caps had actin-rich purse strings around their circumferences (arrow in Fig. 5E). High magnification images(Fig. 5E) show the reverse of XFlop overexpression; staining of all actin-containing structures (except for the contractile rings of dividing cells) was reduced. There were fewer cell processes, and those that were present had less actin staining than did those of controls. The cortical actin network was also reduced. As is the case for overexpression, this suggests that XFlop signaling is required to maintain the normal amount of F-actin in all actin-containing structures, except those required for cell division.
We next examined the cortical actin network in dissociated cells from control and XFlop-depleted embryos (Fig. 5G). Animal caps were dissociated and cultured in calcium/magnesium-free saline for 10 minutes before fixation and staining. This culture time is not sufficient for the cortical network to transform from the dense to the coarse network type, a process that takes at least 30 minutes(Lloyd et al., 2005). Actin staining was significantly reduced in individual dissociated cells that were XFlop depleted. This was quantitated by the comparison of mean pixel intensity(±s.d.) from 20 control and 28 XFlop-depleted cells, which showed that the decrease is statistically significant (P<0.01, Fig. 5H). Fig. 5G (middle panels) shows the images in rainbow scale, in which the Zeiss LSM510 software assigns different spectral colors to different pixel intensities. This illustrates the fact that, even though staining levels are reduced in XFlop-depleted embryos,there is still an actin network present, but it contains less actin. In some areas (see arrow, Fig. 5G),areas denuded of actin filaments can be seen, through which underlying yolk platelets can be seen.
In order to check whether microtubule and intermediate filament assemblies were also affected by XFlop depletion, we stained animal caps from XFlop-depleted and control late blastulae with antibodies against tubulin or cytokeratin. In the absence of maternal XFlop, the microtubule and cytokeratin networks in the animal caps were both intact, although the F-actin network was significantly reduced (Fig. 6A,B), suggesting that the cytoarchitectural defects seen at the blastula stage are primarily due to the loss of the cortical actin skeleton.
In order to show that these defects are specific for the loss of the maternal store of XFlop mRNA, and were not caused by non-specific degradation of other mRNAs by the oligo, or by random toxicity, we injected XFlop mRNA-depleted embryos at the two-cell stage with 50 pg of full-length XFlop mRNA. The injected mRNA rescued the rigidity and spherical architecture of the late blastula(Fig. 6C), as well as their ability to heal wounds (Fig. 6C) and the density of the cortical actin skeleton(Fig. 6D,E). This rescue was seen in four repeat experiments. As an additional control for oligo specificity, we found that two antisense oligos targeting different regions of XFlop mRNA gave identical phenotypes, and that a 1:1 mixture of these two oligos had additive effects (data not shown). The XFlop-depletion phenotypes have been reproduced in five independent experiments. In wound healing assays, we analyzed five to six embryos from each treatment at late blastula stage. Taken together, these data show that XFlop is necessary, as well as sufficient, for maintenance of the correct levels of cortical actin and the maintenance of the spherical architecture of the embryo.
When embryos depleted of the maternal stockpile of XFlop mRNA were allowed to develop further, they exhibited defects in gastrulation movements,and post-gastrulation abnormalities (Fig. 7A-C). Defects ranged from mild (distorted axes, truncations and reduced head structures, Fig. 7B, n=10 in a single experiment) to severe (absence of head, unclosed blastopore, severe truncations, Fig. 7C, n=9 in the same experiment). We previously showed that reduction of the cortical actin skeleton by depletion of mRNA for plakoglobin(Kofron et al., 2002), or the two LPA receptors LPA1 and LPA2 (Lloyd et al., 2005), causes defective gastrulation movements. This suggests that XFlop-regulated cortical actin assembly is required for normal gastrulation movements. To test this further, we microinjected XFlopmRNA into single animal cells at the eight-cell stage together with mRNA encoding β-galactosidase. The early ectoderm cells overexpressing XFlop do not undergo epiboly (Fig. 7D,E), the spreading movement by which cells in the animal region,destined to form the ectoderm, spread around the whole surface of the embryo during gastrulation. Thus, both depletion and overexpression of XFlop abrogate the cell movements of gastrulation. The etiology of the post-gastrulation defects is likely to be complex, first because reduced or delayed gastrulation movements will lead to later defects in a pleiotropic fashion, and second,because XFlop may be required for other aspects of early development. To address this latter point, we investigated whether XFlop is required for the cell fate specification events that take place during the blastula stage, by assaying cDNA from XFlop-depleted embryos for molecular markers of the three primary germ layers, and of the dorsal axis. All of these structures are initiated during the mid- to late-blastula stages by intercellular signaling events (Harland and Gerhart,1997; Kofron et al.,1999; Kofron,2003), and by localized maternal transcription factors(Kofron et al., 1999; Sundaram et al., 2003; Xanthos et al., 2002; Xanthos et al., 2001; Zhang et al., 1998). Early zygotic markers of ectoderm (Xlim5, AP2 and E-cadherin)(Houston and Wylie, 2003; Luo et al., 2002), mesoderm(Eomesodermin, Fgf8, Xbra and Xwnt8), endoderm (Xsox17 and Mix.1)(Xanthos et al., 2002), and of the dorsoventral axis (Xnr3, Siamois and Xvent1)(Yokota et al., 2003) were analyzed at late blastula (stage 9) and early gastrula (stage 10.25) stages,using quantitative real-time RT-PCR techniques. Expression of neither the early ectodermal nor the early endodermal genes were changed dramatically(Fig. 7F,G), indicating that XFlop is not necessary for the specification of ectoderm and endoderm. The expression of Fgf3 and Xbra were reduced, but not absent, at the beginning of gastrulation (Fig. 7H). Because Xwnt8 and Eomesdermin were activated on time, this suggests that XFlop may be necessary for the full activation of a subgroup of mesodermal genes. This is also consistent with the gastrulation movement failure, as shown in Fig. 7A-C. The depletion of XFlop does not affect the activation of early dorsal markers(Fig. 7I), although Siamois was reduced, when compared with controls, at the early gastrula stage.
Taken together, we have shown that XFlop is both necessary and sufficient for the cortical actin assembly, and that this actin assembly activity is required for normal gastrulation movements.
XFlop and LPA signaling play distinct roles in cortical actin assembly
We have recently shown that a different signaling pathway, mediated by the phospholipid ligand lysophosphatidic acid (LPA), and its two receptors XLPA1 and XLPA2, also controls the cortical actin skeleton of the early Xenopus embryo, so it was important to establish the relationship between these two signaling pathways. In particular, we wanted to test whether each pathway independently initiates actin assembly, or whether they are in the same pathway.
A comparison of XFlop overexpression and depletion (Figs 3, 4, 5, 6) with previously published data on LPA1 and LPA2 depletion (Lloyd et al., 2005) suggests that the two signaling pathways control different aspects of actin assembly. LPA depletion causes the dense actin network characteristic of interphase blastomeres in vivo to become replaced by the coarser network seen in dividing blastomeres in vivo, or in dissociated blastomeres (Lloyd et al.,2005). Overexpression of XFlop, however, caused a general increase in actin in all actin-containing structures (Figs 3, 4), whereas its depletion caused an overall decrease in the actin present in all actin-containing structures (Figs 5, 6) without causing a shift to the looser network characteristic of dividing or dissociated cells. This would suggest that LPA1/2 and XFlop play different roles in actin assembly.
To confirm this, we further investigated the effects of XFlop and LPA1 and LPA2 depletions in dissociated cells from late blastulae. Animal caps were excised at stage 9, dissociated in Ca2+/Mg2+-free 1×MMR, and cultured for 90 minutes. Under these conditions, dissociated blastomeres adopt the coarse configuration of cortical actin that is characteristic of dividing cells in intact animal caps(Lloyd et al., 2005). Phalloidin-stained, single dissociated blastomeres from embryos depleted of LPA1 and LPA2 (Fig. 8A) or Xflop (Fig. 8B) are shown. Quantitation of total cortical actin by pixel intensity measurements of 10 control and 27 LPA1 and LPA2-depleted blastomeres, showed that LPA1 and LPA2 depletion did not significantly reduce the amount of cortical actin in each isolated blastomere (Fig. 8A′) (see also Lloyd et al., 2005). Xflop depletion, however, did significantly reduce the total amount of cortical actin in isolated blastomeres(Fig. 8B,B′). This reduction was quantitated by comparing the mean pixel intensity of 30 individual control and 44 XFlop-depleted blastomeres(Fig. 8B′), and suggests that XFlop controls the overall level of actin assembly.
To test further whether XFlop and LPA lie in the same pathway, we investigated whether depletion of one would block the effect of overexpression of the other. If they lie in the same pathway, then they will be epistatic;one will rescue the other, but the reverse will not be the case. Oocytes depleted of either LPA1 and LPA2, or XFlop, were fertilized, and mRNA encoding either XFlop or LPA1 was injected into the animal cytoplasm at the two-cell stage. Both LPA1 overexpression in XFlop-depleted embryos, and XFlop overexpression in LPA1 and LPA2-depleted embryos, increased the levels of cortical actin (Fig. 8C). This shows that neither receptor requires the other for its action, and strongly suggests that they have separate roles in cortical actin assembly.
Discussion
A cortical network of F-actin has been described at the earliest stages of embryogenesis of a number of species, including Drosophila(Karr and Alberts, 1986; Warn and Robert-Nicoud, 1990),mouse (Albertini et al., 1987),pig (Reima et al., 1993) and C. elegans (Strome,1986). However, little is known about the mechanisms by which each cell of the early embryo synthesizes a specific pattern and density of cortical actin. Forward genetic screens in Drosophila have revealed a number of genes involved in actin dynamics(Postner et al., 1992; Postner and Wieschaus, 1994; Schejter and Wieschaus, 1993; Straub et al., 1996; Sullivan et al., 2000), but most of these studies focused specifically upon cellularization, the process by which a syncytial blastula changes to a cellular blastula. None of the genes identified so far in these studies suggest a role for intercellular signaling. Here, we show that expression screening is a useful tool for identifying proteins involved in cytoskeletal assembly in early embryos. It allows rapid gain-of-function experiments by the injection of mRNA pools or single mRNAs, and rapid confirmation of function by loss-of-function experiments. Most proteins functioning at the blastula stages in Xenopus are encoded by genes expressed in the oocyte, and whose mRNA is stored in the egg. Depletion of this mRNA by antisense oligonucleotide injection into cultured oocytes, followed by fertilization of the oocytes,allows any essential role of the cognate protein to be rapidly identified. In this study, and elsewhere (Lloyd et al.,2005), we have used these assays to show that in a cellular vertebrate embryo, intercellular signaling plays a primary role in the control of F-actin assembly in each cell. Here, we show that the G protein-coupled receptor XFlop is both necessary and sufficient for this process. It is interesting that two G protein-coupled receptors (the LPA receptors are also GPCRs) have now been identified that seem to perform separate, but overlapping roles in the assembly of the correct pattern and density of actin filaments in the embryonic cell cortex.
XFlop is most closely related to mammalian GPR4, a member of a recently proposed sub-family of G protein-coupled receptors, the OGR1 sub-family(Xu, 2002). Comprising OGR1(ovarian cancer G protein-coupled receptor 1), G2A (G2 accumulation), TDAG8 (T cell death-associated gene 8) and GPR4 (G protein-coupled receptor 4), the sub-family is moderately conserved, with 36-51% amino acid identity. Ligands for this group are still being identified, but include related signaling lipids. TDAG8 has been reported to bind psychosine (galactosyl sphingosine)(Im et al., 2001), whilst OGR1 binds sphingosylphosphoryl choline (SPC)(Xu et al., 2000), and GPR4 and 2GA bind both SPC and lysophosphatidyl choline (LPC)(Bektas et al., 2003; Kabarowski et al., 2001; Zhu et al., 2001). More recently, it has been reported that OGR1 and GPR4 respond differentially to changes in extracellular proton concentration(Ludwig et al., 2003). OGR1 increases inositol phosphate formation, whilst GPR4 increases cyclic AMP formation, in response to increased proton concentration. Ligand specificities have generally been identified by overexpressing the receptor in cultured cells, so it is not yet clear which ligands are used in vivo, and whether these change spatially and temporally during tissue differentiation, nor are the in vivo functions of the OGR1 sub-family fully established. G2A null mice have an abnormal expansion of both B and T lymphocytes, and an autoimmune syndrome, suggesting that G2A acts as a repressor in the immune system(Le et al., 2001). GPR4 is thought to mediate the growth stimulatory effects of SPC on Swiss 3T3 cells,whilst OGR1 may mediate the inhibitory growth effects of this same ligand(Xu et al., 2000; Zhu et al., 2001), suggesting complementary functions of these related receptors. We show here that a related receptor, XFlop, in Xenopus, plays an essential in vivo role in the assembly of the correct amount of cortical actin. It will be interesting to study the related mammalian receptors in this respect, to see whether this function is conserved throughout the vertebrates.
Signaling through G protein-coupled receptors has been implicated previously in the control of actin assembly, particularly in cell responses to chemotactic stimuli. Ligands for these can be small molecules such as cAMP(reviewed by Hereld and Devreotes,1992), peptide ligands such as stromal derived factor 1 (SDF1)(Nagasawa et al., 1999), or N-formyl methionine peptides (Panaro and Mitolo, 1999). However, the role of intercellular signaling in establishing the amount, and pattern, of cortical actin that maintains the correct shape and rigidity of the early embryo is less well understood. During the egg to blastula stages, the number of cells in the embryo doubles with each cell cycle. However, after each division cycle, each cell assembles the cortical actin network appropriate to the shape and rigidity of the whole embryo. In general, this could be accomplished by intracellular components of the actin polymerization machinery inherited from the egg, or by intercellular signaling. This work, and a previous paper(Lloyd et al., 2005), shows that at least two intercellular signaling pathways control this process in early Xenopus embryos. The role of the XFlop-mediated pathway appears to be to control the overall amount of F-actin in most, but not all,actin-containing structures, as XFlop depletion causes reduced F-actin in all structures except the contractile rings of cytokinesis, which must be controlled by other signals. By contrast, the role of LPA signaling seems to be to control the switch from dense cortical actin during interphase to a less dense network during cytokinesis (Lloyd et al., 2005). The two pathways can also be distinguished by the fact that each can be initiated in the absence of the other.
The ligand for XFlop has yet to be identified, as have details of its downstream signaling pathways. The multiple effects of XFlop depletion(reduction in the cortical network and cell processes, and absence of wound healing and purse-string formation), and the fact that overexpression increases the amount of F-actin without amplifying only one type of actin-containing structure, suggest that it acts through more than one downstream signaling pathway. It will be important to identify which cytoplasmic signaling intermediates, and which of their effectors, act downstream of LPA and XFlop signaling at the cell surface in this in vivo developing system.
Acknowledgements
The authors would like to thank the NIH for financial support (RO1 HD044764, T32HD07463), and Bertie Puck for excellent technical support.