Signal transduction pathways are key elements in the regulation of cell behavior and tissue organization during the development of multicellular organisms. The members of each pathway are linked by molecular interactions, and, in general, they control the activity of transcription factors that contribute to the temporal and spatial regulation of gene expression. The Ras/Mitogen activated protein kinase pathway (Ras/MAPK) is conserved from yeasts to humans and participates in a variety of developmental processes, including mating in yeast, vulval development in Caenorhabditis elegans and appendage formation in both vertebrates and invertebrates (Herskowitz, 1995; Simcox, 1997; Martin, 1998; Shilo, 2003; Wang et al., 2000; Zecca and Struhl, 2002a). The Ras/MAPK pathway is also implicated in the formation and progression of multitude of human cancers (Edwards, 1999; Farassati et al., 2001; Porter and Vaillancourt, 1998). The core elements of the pathway include the small GTPase Ras, several serine-threonine protein kinases and several nuclear effectors such as the ETS proteins Pointed and Yan (Baonza et al., 2002; Perrimon, 1994; Rebay and Rubin, 1995; Rebay, 2002; Shilo, 2003). Growth factors, hormones, cytokines, and cellular stress, acting through receptor complexes belonging to different superfamilies, activate specific MAPK proteins. MAPK belong to three large families exemplified by p38, JNK, and extracellular-regulated kinase (ERK). ERK proteins are activated by receptors with tyrosine kinase activity, such as epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR), and fibroblast growth factor receptor (FGFR), and also by receptors coupled to a heterotrimeric guanine nucleotide binding protein (G-proteins; Rebay, 2002). In general, these receptors are activated upon binding of extracellular signals and initiate a phosphorylation cascade culminating in the activation of ERK by phosphorylation of threonine and tyrosine residues in a conserved domain. Activated ERK proteins, in turn, phosphorylate serine and threonine residues in a variety of target proteins, including cytoskeletal components and transcription factors involved in the response to growth factors (Shilo, 2003).
The regulation of the places where the Ras/ERK signaling pathway is active and the control of appropriate levels of signaling, are two key aspects in the implementation of the pathway biological activities (Casci and Freeman, 1999). This outcome is achieved through several mechanisms, including the regulation of both ligands and receptors expression and the activity of antagonists that can turn down the pathway at multiple levels (Perrimon and McMahon, 1999; Rebay, 2002). Of interest, many of these antagonists are expressed in response to signal activation, and constitute negative feedback loops that modulate the levels of signaling and contribute to signal termination (Perrimon and McMahon, 1999; Rebay, 2002). Examples of antagonist are the secreted protein Argos, which plays a determining role in photoreceptor formation in the Drosophila eye, and several Drosophila and vertebrate proteins such as the transmembrane protein Kek-1 and the cytoplasmic components Sprouty, Spred, and Sef (Golembo et al., 1996; Ghidlione et al., 1999; Vinos and Freeman, 2000; Wakioka et al., 2001; Yusoff et al., 2002; Ghiglione et al., 2003; Kovalenko et al., 2003). These proteins act at different levels of the FGF and EGF signaling pathways, antagonizing the receptor itself (Kek-1, Sef) or the serine-threonine kinase Raf (Sprouty and Spred).
A key point of regulation of Ras/MAPK signaling involves reversible changes in the phosphorylation of ERK. The kinase activity of ERK depends on the phosphorylation of particular threonine and tyrosine residues, and the opposing effects of ERK kinases (MEK) and ERK phosphatases (MKP) determine the cellular levels of activated ERK. The places of ERK activation during development can be visualized in vivo using antibodies against the phosphorylated protein (dp-ERK; Yung et al., 1997; Bier, 1998). The restricted expression of dp-ERK suggests that many mechanisms operate to generate a dynamic ERK activation/deactivation balance in different tissues and developmental stages. The down-regulation and eventual termination of ERK signaling is brought about by dephosphorylation of either the threonine or tyrosine residues within the ERK activation loop motif (Camps et al., 2000). The elimination of phosphate from these residues depends on the activity of specific phosphatases such as tyrosine phosphatases, threonine phosphatases and dual-specificity phosphatases (DSPs; Camps et al., 2000). The proteins of the DSP family have a characteristic structure of two domains, a N-terminal domain involved in ERK interactions and a catalytic C-terminal domain (Camps et al., 2000). The specificity of binding is determined by the N-terminal part of the protein, which shows two domains of homology found in the yeast phosphatase CDC25 (CH2; Muda et al., 1998). Binding of MKP to its specific MAP kinase triggers the activation of the phosphatase catalytic domain (Muda et al., 1998). Of interest, the expression of several DSPs depends in part on MAP kinase activation, and many of them are transcriptionally up-regulated in response to growth factors, both in cell culture experiments (Boschert et al., 1998; Camps et al., 1998, 2000) and in developing embryos (Dickinson et al., 2002; Eblaghie et al., 2003; Kawakami et al., 2003).
In this work, we have studied the role of the DSP member MKP3 during Drosophila wing development and in Xenopus anteroposterior (A/P) neural patterning. We have characterized the phenotype of both Drosophila and Xenopus MKP3 loss- and gain-of-function genetic conditions, studied their patterns of expression during normal development and analyzed the functional and regulatory relationships between MKP3 and the Ras/MAPK signaling pathway. Our results indicate that MKP3 expression is induced in response to EGFR and FGFR receptor activation in Drosophila and Xenopus, respectively. Upon induction, MKP3 and phosphorylated ERK (dp-ERK) are expressed in the same territories, suggesting a role of MKP3 in sustaining appropriate levels of ERK phosphorylation. These results implicate a conserved negative feedback loop mediated by MKP3 on Ras/MAPK signaling and suggests a key role of MKP3 in processes where the level of Ras/MAPK signaling is critical for normal developmental, such as in the formation of the Drosophila wing veins and during A/P patterning of the Xenopus neuroectoderm.
Isolation and Characterization of Drosophila MKP3 Gain of Function Alleles
Three independent insertions of a P-GS element in the 5′ untranslated region (UTR) region of the MKP3 gene were obtained in a genetic screen designed to identify genes that, when overexpressed, affect the differentiation of veins. All three insertions (MKP3M76, MKP3S, and MKP3160; Fig. 1) cause very similar phenotypes in combination with a variety of Gal4 lines (see Fig. 2). The phenotypes depend on the temporal and spatial expression of the Gal4 drivers used. Thus, in combinations with the Gal4-shv3kpn driver, which is expressed only in the developing pupal veins L2, L3, L4, and distal L5 (see Experimental Procedures section), the wings are of normal size and these longitudinal veins fail to differentiate (Fig. 2B). When the expression of MKP3 is driven in the imaginal disc, such as in combinations with Gal4-sal (expressed in a central domain in the wing disc), the resulting phenotypes are milder and affect the veins L2 and L4 (Fig. 2C). Increased levels of MKP3 in its normal domain of expression (in the combination Gal4-NP3640/MKP3M76) also results in a mutant phenotype of loss of veins (Fig. 2D), indicating that both the pattern and level of MKP3 expression are critical for normal development. When the driver Gal4-638 (expressed in the wing blade from early stages, not shown) is used, the resulting phenotypes are the complete absence of the wing and its substitution for proximal hinge tissue (Fig. 2E, arrow). Combinations with Gal4 drivers expressed in the notum region of the wing imaginal disc affect the development of macrochaetae (Fig. 2F) or the formation of the thorax (Fig. 2G). In summary, MKP3 overexpression in the wing imaginal disc affects developmental processes, such as vein differentiation (see also Rintelen et al., 2003), macrochaetae formation and wing–thorax subdivision, that require EGFR signaling (Diaz-Benjumea and Hafen, 1994; Guichard et al., 1999; Culi et al., 2001; Wang et al., 2000; Cavodeassi et al., 2002; Zecca and Struhl, 2002a, b). Furthermore, the resulting phenotypes of MKP3 overexpression correspond to strong reductions of EGFR activity, as they are very similar, albeit stronger, than those resulting from ectopic expression of dominant negative forms of the EGFR (data not shown) and loss-of-function alleles of genes of the EGFR partway (Diaz-Benjumea and Hafen, 1994; Guichard et al., 1999; Wang et al., 2000; Culi et al., 2001; Cavodeassi et al., 2002; Zecca and Struhl, 2002a, b). The phenotypic analysis of MKP3 overexpression suggests that this protein affects EGFR signaling during imaginal development, and indicates that the spatial–temporal pattern and levels of MKP3 expression are critical for normal EGFR activity.
The Ras/MAPK pathway plays a key role in the formation of veins, and for this reason, the relationships between MKP3 and other members of the EGFR signaling pathway were further analyzed in the developing pupal veins. We find that the loss of veins characteristic of MKP3 ectopic expression (Fig. 2B) is retained when MKP3 is coexpressed with rhomboid (Fig. 2H), EGFR (not shown), activated-Ras (Fig. 2I), and activated-Raf (not shown). Furthermore, the ectopic and thicker veins differentiating when Rho, EGFR, activated-Ras, and activated-Raf are expressed using the same Gal4 driver (Sotillos and de Celis, 2005) are completely suppressed by coexpression of MKP3 (Fig. 2H,I and data not shown). The only member of the EGFR pathway able to suppress the loss of vein caused by ectopic MKP3 is the mutant form sevenmaker of the gene rolled (rolsem; Fig. 2J). Rolled encodes the Drosophila ERK protein, and the mutant form Rolsem affects in vitro its binding to MKP3, behaving as a constitutively activated form (Brunner et al., 1994; Kim et al., 2002). The specific suppression of the MKP3 overexpression phenotype by rolsem coexpression was also observed by Rintelen et al. (2003) and is compatible with Rolled being the in vivo target of MKP3 activity. Taken together, these results indicate that, during vein differentiation, MKP3 antagonizes the activity of the MAPK protein Rolled.
Isolation and Characterization of MKP3 Loss-of-Function Alleles
The genetic analysis of gain-of-function conditions in the MKP3 gene is fully compatible with a role in regulating the activity of ERK by direct dephosphorylation. To identify the requirements of MKP3 during development, we induced loss-of-function alleles in the gene. Two ethyl methane sulphonate (EMS) -induced revertants (MKP3M76-M and MKP3M76-R2b; see Experimental Procedures section) of the P-GS insertion MKP3M76 cause in the heteroallelic combination MKP3M76-M/MKP3M76-R2b the differentiation of ectopic vein stretches in several regions of the wing (Fig. 2K, arrows). The allele MKP3M76-R2b is homozygous viable with a stronger phenotype of ectopic vein formation (Fig. 2L, arrows). Mitotic recombination clones of the lethal chromosome MKP3M76-M have normal viability in all wing regions, and only caused the formation of ectopic veins when they appear in the proximal region between the veins L4 and L5 (data not shown). This phenotype is weaker than that of MKP3M76-R2b homozygous flies, suggesting that the lethality of the MKP3M76-M chromosome is not related to the MKP3 gene.
To confirm the presence of point mutations in the MKP3M76-M and MKP3M76-R2b chromosomes, we sequenced all the exons and the UTR regions of the MKP3 transcript. In the MKP3M76-M mutation, we only identified a T to A transition in position 19018080 (exon 5) that causes the substitution of Tyr316 for Asn (Fig. 1). This Tyr is situated in the catalytic core of the protein, in a region highly conserved in the MKP3 gene of several species (Fig. 1). The allele MKP3M76-R2b has only a nucleotide substitution in the first ATG of the coding region, changing the ATG consensus for an AAG codon. The next ATG in frame is located 111 amino acids toward the C-terminal end of the protein, and, if produced, the putative mutant MKP3M76-R2b protein would lack all of the MKP3 N-terminal domain, which is required for interaction with ERK and phosphatase activation (Fig. 1). Thus, the MKP3M76-R2b allele most likely corresponds to the null condition of the gene.
The phenotype of homozygous MKP3M76-R2b flies is very similar to that of genetic combinations in which there is a moderate increase in EGFR signaling, such as in rlsem mutant animals, or when rlsem or EGFR are ectopically expressed in the pupal interveins (Fig. 2M, compare with K and L). Taken together, these data suggest that one role of MKP3 is to maintain generalized low levels of EGFR signaling. In addition, the expression pattern of the gene (see below), and the effect of increasing MKP3 expression in its normal domain, suggest that MKP3 expression is also critical to maintain appropriate signaling levels in places where the EGFR pathway is more active.
MKP3 Affects ERK Phosphorylation During Vein Differentiation
It has been shown that purified Drosophila MKP3 produced in Escherichia coli has intrinsic phosphatase activity in vitro and that it can inhibit both the phosphorylation of ERK (encoded in Drosophila by rolled) and its activity in Schneider cells (Kim et al., 2002). Furthermore, MKP3 interacts with ERK through its N-terminal domain, and this interaction is needed for its activity (Kim et al., 2002). No effect of MKP3 on other MAPK proteins such as p38 and JNK were observed in cell culture experiments (Kim et al., 2002). To visualize in vivo the level of phosphorylated ERK (dp-ERK) in different experimental situations where EGFR signaling is affected, we expressed the wild-type receptor (UAS-EGFR), a dominant negative version of EGFR (UAS-EGFRDN), and MKP3 (MKP3M76) in the central domain of the wing using Gal4-sal as a driver (Fig. 3). In wild-type wing discs, the expression of dp-ERK is maximal in the developing longitudinal wing veins L3 and L4 and in the marginal veins, both in the dorsal and ventral compartments (Gabay et al., 1997; Fig. 3A). This pattern indicates where the activity of the EGFR pathway is maximal, and it is very similar to that of rhomboid (rho) expression (Sturtevant et al., 1993; Gabay et al., 1997). When the level of EGFR is increased in the spalt domain of expression (Gal4-sal/UAS-EGFR wing discs) the presence of dp-ERK is highly augmented in this territory (Fig. 3B). Conversely, when either EGFRDN or MKP3 are misexpressed in the sal domain, we observe a reduction in the level of dp-ERK (Gal4-sal/UAS-EGFRDN wing discs; Fig. 3C) or its absence in the presumptive veins L3 and L4 (Gal4-sal/MKP3M76 wing discs; Fig. 3D). As expected, the expression of dp-ERK in homozygous MKP3M76-R2b wing discs is increased (Fig. 3E), indicating that MKP3 plays a significant role in maintaining appropriate levels of dp-ERK and EGFR signaling in vivo.
Role of MKP3 During Vein Pattern Formation
To further characterize the role of MKP3 during vein patterning, we studied the expression of blistered (bs) and Delta (Dl), which are known to be regulated by EGFR signaling during imaginal development (de Celis et al., 1997; Huppert et al., 1997; Roch et al., 1998). In these experiments, we compare the effects of ectopic expression of MKP3 with those of misexpression of EGFRDN (Fig. 4). In wild-type discs, Bs protein expression is restricted to the developing interveins by EGFR-mediated repression (Fig. 4A), and the expression of Dl is maximally localized in the veins L3, L4, and L5 and in the marginal veins due in part to EGFR-mediated activation (Fig. 4D; Kooh et al., 1993; Sturtevant et al., 1993; Fristrom et al., 1994; Montagne et al., 1996; Huppert et al., 1997; Roch et al., 1998). The characteristic down-regulation of Bs in the developing veins L3 and L4 is lost when MKP3 in expressed using Gal4-sal, and only the L5 gap in Bs localization was observed (Fig. 4B). The L5 vein lies outside the domain of Gal4-sal expression (see Fig. 3F). Ectopic expression of EGFRDN has similar consequences, although the effects on Bs protein localization are weaker (Fig. 4C). Ectopic expression of MKP3 and dominant-negative EGFR cause a similar reduction of Dl expression in the veins L3 and L4 (Fig. 4E,F).
Regulation of MKP3 Expression by EGFR Activity
An important characteristic of the EGFR pathway is that the levels and pattern of signaling are highly regulated during development (Bier, 1998). Correct modulation of the place and intensity of signaling relies on the restricted expression of several elements of the pathway, such as the ligands Vein and Spitz, as well as of several components involved in the proteolytic processing and activation of the ligands (Rho and Star) (Rutledge et al., 1992; Sturtevant et al., 1993; Simcox et al., 1996; Golembo et al., 1999; Guichard et al., 1999). To study whether MKP3 expression is related to the activity of the pathway, we analyzed its expression pattern by in situ hybridization in different experimental situations. The MKP3 transcript is preferentially detected in wild-type wing imaginal discs in two stripes of cells interrupted at the dorsoventral boundary (Fig. 5A). These regions correspond to the developing veins L3 and L4, as revealed by the expression of the MKP3 Gal4 line NP3640 (Fig. 5B,C). In pupal wings (24 hr after puparium formation [APF]), the maximal levels of expression are present in all longitudinal veins, with low levels detected in most intervein cells (Fig. 5K,L). The specificity of our mRNA probe was confirmed in overexpression experiments using the Gal4 lines sal (Fig. 5D), nub (Fig. 5G), and shv3kpn (Fig. 5J) in combination with the P-GS line MKP3M76. We also studied the expression of MKP3 in experimental situations where the activity of the EGFR signaling pathway was altered in the wing blade. In the combinations Gal4-sal/UAS-EGFRDN and Gal4-sal/UAS-EGFR, we obtained opposite results. Thus, when the activity of the pathway is lowered (Gal4-sal/UAS-EGFRDN; Fig. 5E), the levels of MKP3 expression are very reduced, and, conversely, when the level of signaling is augmented (Gal4-sal/UAS-EGFR; Fig. 5F), the expression of MKP3 is increased in the central domain of the wing. Similar changes in MKP3 expression were observed in the combinations between Gal4-nub and the UAS lines UAS-EGFRDN and UAS-EGFR (Fig. 5H,I). The correlated changes of EGFR signaling and MKP3 expression indicate that the transcription of this gene in the developing wing blade is positively regulated by the EGFR signaling pathway.
We also studied the expression of MKP3 during embryonic development, as the domains of MAPK signaling have been defined accurately in different tissues and stages (Bier, 1998; Gabay et al., 1997). MKP3 mRNA is uniformly distributed in the syncytial blastoderm (stage 4), indicating an important maternal contribution (Fig. 6A). Later on, as cellularization proceeds, there is a drastic change in mRNA distribution. At early stage 5, mRNA accumulates at the embryonic poles and is absent from the central region (Fig. 6B). This pattern evolves very rapidly with the formation of a third central domain of expression from 85% to 40% egg length (Fig. 6C). In older embryos, the main places where high levels of MKP3 transcript accumulate correspond to the invaginating mesoderm (Fig. 6D, arrow), the tracheal pits at stage 11 (Fig. 6E, arrow), the visceral mesoderm at stage 13 (Fig. 6F, arrow), the heart (Fig. 6G, arrow), the midline and the apodema (Fig. 6G,H, arrows and arrowheads, respectively). These territories correspond to places where the activity of several receptor tyrosine kinases (RTK), such as EGFR, Breathless and Torso are required, indicating that there is a good correlation between places of RTK activity and the expression of MKP3. Thus, the regulatory relationship between RTK activation and MKP3 expression observed in the wing disc could also be operative during embryonic development. We have not attempted to analyze the requirements of MKP3 during embryonic development. To study the functional conservation of MKP3 regulation and function in other organisms, we characterized the XenopusMKP3 homologous gene (XMKP3).
Xenopus MKP3 Expression Depends on Ras-Mediated FGF Signaling
The restricted expression of DrosophilaMKP3, which appears to be controlled by the RAS/MAPK pathway, prompted us to re-examine the expression profile of Xenopus MKP3. Previous studies showed that MKP3 mRNA is localized in restricted domains only at neurula stages (Mason et al., 1996). In contrast, we found that MKP3 is expressed with a highly restricted pattern all through early development. Thus, at gastrula stages, MKP3 is found in the mesoderm marginal zone and in the dorsal ectoderm (Fig. 7A–D). During neurulation, MKP3 mRNA is localized to the posterior neuroectoderm, at the midbrain–hindbrain boundary, and in two horseshoe-shaped domains within the anterior neural plate (Fig. 7E–I). At late neurula and tail bud stages, MKP3 is also expressed at the branchial arch region, somites, and tail tip (Fig. 7I,J). MKP3 expression domain is very similar to that of Fgf8 (Fig. 7K,L), which can signal through the MAPK pathway in vertebrates and invertebrates (Umbhauer et al., 1995; Martin, 1998; Michelson et al., 1998; Borland et al., 2001). Thus, we examined whether the FGF/MAPK pathway controls MKP3 expression. Interfering with MAPK signaling by overexpressing a dominant negative Ras mRNA (DN-ras; 1 ng) strongly reduces MKP3 expression (Fig. 7M, black arrowhead; 58%, n = 41). Conversely, injection of 1 ng of constitutively activated Ras mRNA (CA-ras) promotes ectopic MKP3 expression (Fig. 7N, black arrowhead; 63%, n = 43). We then examined whether modulation of FGF signaling at the receptor level also affects MKP3 expression. Interfering with FGF signaling by overexpression of a dominant negative FGFR1 mRNA (DN-FgfR1; 200 pg) reduces MKP3 expression in the posterior neuroectoderm (Fig. 7O, black arrowhead; 45%, n = 49). Conversely, overexpression of constitutively activated FGF receptor mRNA (CA-FgfR1; 500 pg) caused ectopic expression of MKP3 (Fig. 7P, black arrowhead; 66%, n = 38). Taken together, our data indicate that FGF, signaling through Ras, regulates the expression of MKP3. These results are consistent with recent reports showing that in other vertebrates MKP3 is expressed in regions in which FGF signaling is active (Klock and Herrmann, 2002; Dickinson et al., 2002; Eblaghie et al., 2003; Kawakami et al., 2003).
MKP3 Negatively Modulates FGF Function in Anterior–Posterior Patterning in Xenopus
The expression pattern of MKP3 and its regulation by the FGF/MAPK pathway suggests that MKP3 is acting as a negative feedback loop of the pathway during Xenopus neural development, as it has been reported during mesoderm formation (Umbhauer et al., 1995). During neural development, gain- and lost-of-function analysis of FGF/MAPK indicates its requirement in the generation of posterior neural identities (Mariani and Harland, 1998; Salzberg et al., 1999; Umbhauer et al., 2000; Ribisi et al., 2000). This function involves reprogramming of neural cells with an initial anterior character (reviewed in Stern, 2001). To examine MKP3 function during neural development, we overexpressed the wild-type form of MKP3 mRNA (4 ng) (Umbhauer et al., 1995) or the mRNA (4 ng) encoding a phosphatase dead mutant (Mason et al., 1996). This molecule acts as a partial dominant negative form of MKP3 by sequestering MAPK in the cytoplasm (Brunet et al., 1999; Tsang et al., 2004). As previously reported (Umbhauer et al., 1995), most of the embryos injected with MKP3 mRNA failed to gastrulate due to malformation of mesoderm structures (Fig. 8A, lower embryo; 71%, n = 70) and will not be considered here. Among those embryos that proceeded normally through gastrulation (29%), most of them (12 embryos from 20, 60%) developed shorter A/P axis associated with enlarged anterior structures (Fig. 8A, two upper embryos). Analyses of A/P neural patterning in MKP3-injected embryos showed that, in the injected side, the neuroectoderm is anteriorized, as shown by the caudal shift of different neural A/P markers (Fig. 8B, 63% of 19 embryos that proceed normally through gastrulation). Similar effects have been reported in embryos in which FGF signaling is impaired (Amaya et al., 1991; Holowacz and Sokol, 1999). These results suggest that overexpression of MKP3 interferes with FGF signaling. The anteriorization observed upon injection of MKP3 mRNA was not observed upon misexpression of XMKP3-mut mRNA. Indeed, most of these injected embryos developed normally (Fig. 8C, upper embryo; 85%, n = 65). However, a small fraction of these embryos showed truncation of anterior structures (Fig. 8C, two lower embryos; 15%). This anterior truncation is also observed at neurula stages by monitoring different A/P neural markers, with anterior markers reduced and posterior genes displaced anteriorly (compare Fig. 8D with a control embryo in Fig. 8F). Similar anterior truncation is observed upon overexpression of several Fgf molecules (Lombardo et al., 1998). That the anteriorization observed upon XMKP3-mut mRNA injection occurs only in a small fraction of embryos and that even in these embryos not all markers are similarly affected (see in Fig. 8D expression of HoxB9, which is not up-regulated anteriorly), suggests that XMKP3-mut behave only as a partial dominant negative molecule. This suggestion is consistent with previous data showing that the phosphatase dead mutant form of MKP3 binds to active MAPK, preventing its translocation to the nucleus. This process impairs the activation of nuclear MAPK targets but does not affect that of the cytoplasmic ones (Brunet et al., 1999). Consistent with MKP3-mut behaving, at least in part, as a dominant negative molecule, in embryos coinjected with MKP3 and MKP3-mut mRNAs an increased number of embryos developed normally (Fig. 8E, 43%, n = 60) compared with those injected with MKP3 alone.
The control of appropriate levels of Ras/MAPK signaling is of central importance to determine cell behavior and fate during development. The pathway is regulated at various hierarchical levels, and MAPK proteins are key targets in signal regulation (Garrington and Johnson, 1999; Rebay, 2002). Several mechanisms determine MAPK activity, including scaffolding of MAPK modules in macromolecular signaling complexes, and the balance between the kinases and phosphatases that modify the activity and subcellular localization of MAPK (Garrington and Johnson, 1999; Rebay, 2002). The inactivation of MAPK depends on dephosphorylation of either threonine or tyrosine residues in the MAPK activation loop. The DSP dephosphorylate both serine-threonine and tyrosine leading to reversible inactivation of MAPK. Many aspects of DSP proteins activity and mechanism of action are known by biochemical and cell culture analysis. Thus, individual DSP proteins have specific subcellular localization, they can display high specificity toward particular MAPK targets, and their transcription is generally induced by growth factors, both in cell culture and in vivo (reviewed in Camps et al., 2000). In this work, we have studied the role of the DSP member MKP3, focusing in the biological aspects that determine its function in vivo. To this end, we have used Drosophila wing vein formation and Xenopus A/P neural patterning as model systems to study the functional requirements of MKP3 and its relationships with other members of the Ras/MAPK signaling pathway.
MKP3 Antagonizes Specifically Ras/MAPK Signaling
The biochemical analysis of Drosophila MKP3 indicated that its activity is specific of the ERK protein Rolled, the MAPK central to both EGF and FGF signaling in Drosophila (Kim et al., 2002). This study also showed a cytoplasmic localization of MKP3 in Schneider cells (Kim et al., 2002). The results of our functional analysis of MKP3 in Drosophila are fully compatible with a specific and biologically significant role of the protein in the inactivation of Rolled. Thus, all the phenotypes due to overexpression of MKP3 are indistinguishable from those caused by inactivation of the Ras/MAPK pathway. These phenotypes are very different to those caused by interference with the Notch, Dpp, Hedgehog, JNK or IP(3)K signaling pathways during imaginal development, indicating high specificity in vivo of MKP3 toward the Ras/MAPK pathway. Furthermore, the effects of MKP3 overexpression on the transcription of several targets of the Ras/MAPK pathway during vein formation (bs and Dl) are also similar to those caused by reduction of EGFR activity. The genetic analysis of combinations between MKP3 and other members of the Ras/MAPK pathway indicates that the target of MKP3 antagonism is, as expected, the MAPK protein Rolled. Therefore, we conclude that MKP3 negatively regulates EGFR signaling by inactivating specifically Rolled during Drosophila imaginal development. Rintelen et al. reached independently a similar conclusion in their analysis of MKP3 in the Drosophila eye (Rintelen et al., 2003). During Xenopus neural development, we found a similar negative regulation of the FGF pathway by MKP3. Thus, overexpression of XMKP3 promotes effects consistent with a reduction of FGF signaling such as the reduction of posterior neural structures and the concomitant expansion of anterior ones. Conversely, overexpression of a phosphatase dead mutant form of XMKP3 cause, to some extend, phenotypes similar to those produced by the ectopic activation of the FGF pathway, namely the suppression of anterior neural structures as a consequence of the expansion of posterior ones.
The cytoplasmic localization of MKP3 in Schneider cells is compatible with a mechanism of MAPK inactivation based on cytoplasmic trapping of this protein. However, a MKP3 mutant in the catalytic domain (MKP3M76-M) with an intact ERK-interaction domain affects its activity and causes a mutant phenotype, suggesting that direct dephosphorylation is the more likely mechanism of MKP3 antagonism. The second allele we isolated (MKP3M76R2b) eliminates the first Met codon and would correspond to the elimination of MKP3 activity. This allele displays a phenotype of ectopic veins indistinguishable from that of other null alleles of the gene (Rintelen et al., 2003). Of interest, the pattern and number of ectopic veins is remarkably similar to that caused by a generalized increase in the activity of the Ras/MAPK pathway, further indicating the specific requirement of MKP3 in the down-regulation of Ras/MAPK signaling in vivo. The elimination of MKP3 function is compatible with viability. Furthermore, the phenotype of null alleles is mild when compared with the morphological consequences of other manipulations that activate Ras/MAPK signaling in vivo. Other mechanism involved in signaling down-regulation acting independently of MKP3 might account for the mild phenotypic consequences of MKP3 elimination. For example, genetic analysis identified the tyrosine-phosphatase PTP-ER as another negative regulator of Rolled (Karim and Rubin, 1999). The phenotype of PTP-ER mutant flies is also very mild, and there is a genetic interaction between this gene and MKP3, suggesting some degree of functional redundancy between the two proteins (Karim and Rubin, 1999; Rintelen et al., 2003). Other mechanisms antagonizing ERK activity and Ras signaling might also contribute to regulate negatively the pathway even in the absence of MKP3 and PTP-ER, compensating for the absence of these proteins during wing vein and eye development. In contrast with the moderate effects of MKP3 elimination, the augmented expression of the gene causes very strong phenotypes typical of Ras/MAPK signaling down-regulation, suggesting that high levels of phosphatase are incompatible with MAPK activity during development.
Ras/MAPK Signaling Regulates MKP3 Expression
The aim of our analysis was to identify mechanisms regulating MAPK activity during development in vivo. The main aspects of the MKP3 gene related to its biological activity are its restricted expression and its regulation by Ras/MAPK signaling. Thus, both during embryonic and imaginal development, the expression of MKP3 is localized to places of Ras/MAPK signaling, such as the terminal poles of the embryo or the developing wing veins. Similarly, the expression of the homologous Xenopus gene is also correlated with FGF ligand expression and signaling domains. Moreover, increasing FGF signaling or MAPK activity promotes ectopic MKP3 expression, while interference with FGF/MAPK pathway suppresses MKP3 expression. This data indicate that, during Xenopus early neural development, MKP3 is controlled by FGF/MAKP pathway, as has been reported in chick (Eblaghie et al., 2003) and during Xenopus mesoderm formation (Umbhauer et al., 1995). The presence of MKP3 in places of Ras/MAPK signaling is also observed in other vertebrate embryos, indicating a high degree of conservation in the regulation and in vivo function of MKP3 genes during evolution (Eblaghie et al., 2003). The coincidence between dp-ERK and MKP3 expression is not always manifested. For example, MKP3 is not expressed in the apical ectodermal ridge, the tissue where higher levels of dp-ERK are present during vertebrate limb development (Kawakami et al., 2003). Similarly, the expression of MKP3 in the Drosophila eye disc does not appear to depend on Ras/MAPK activity (Rintelen et al., 2003). These observations indicate that other regulatory inputs besides Ras/MAPK signaling contribute to generate the expression pattern of MKP3.
To what extent the regulation of MKP3 by Ras/MAPK signaling is direct is not known. Recently, it has been proposed that the induction of MKP3 by ERK is mediated by the PI(3)K pathway in chick embryos and cell culture experiments (Kawakami et al., 2003). Another report, however, suggests that the PI(3)K pathway is not involved in inducing MKP3 expression in response to FGF (Eblaghie et al., 2003). Our studies in Xenopus cannot discriminate whether MKP3 is induced by the MAPK and/or PI(3)K pathway, because both of them can activate Ras. However, in Drosophila the phenotypic consequences of changes in IP(3)K and MKP3 activities are entirely different, because IP(3)K affects only cell growth and size without interfering with pattern formation or cellular differentiation (Stocker and Hafen, 2000). This finding suggests that IP(3)K signaling is not related to the regulation of MKP3, at least in Drosophila imaginal discs.
In conclusion, the regulation of MKP3 expression by EGFR and FGFR signaling and the antagonism of MKP3 toward the key pathway transducer ERK, indicates that MKP3 regulates Ras/MAPK signaling acting as a conserved negative feedback loop. In this role, MKP3 plays a comparable function to the specific JNK phosphatase encoded by puckered (Martin-Blanco et al., 1988).
Drosophila melanogaster Strains and Phenotypic Analysis
We used the following Drosophila UAS lines: UAS-EGFR, UAS-EGFRDN (Buff et al., 1998), UAS-rasact (Brand and Perrimon, 1993), UAS-sem (Brunner et al., 1994; Oellers and Hafen, 1996), UAS-rho (de Celis et al., 1997), UAS-NotchDN (Lawrence et al., 2000), UAS-tkvQD (Nellen et al., 1996), UAS-dpp (Staehling-Hampton and Hoffmann, 1994), and UAS-GFP (Ito et al., 1997). We also used the Gal4 lines Gal4-shv3kpn (Sotillos and de Celis, manuscript submitted for publication), Gal4-nub and Gal4-sal (Calleja et al., 1996), Gal4-638, Gal4-1348 (de Celis, 1997), Gal4-253 (de Celis et al., 1999), Gal4-248 (Aldaz et al., 2003), and Gal4-NP3640 (NP consortium at http://flymap.lab.nig.ac.jp/∼dclust/getdb.html). The Gal4-NP3640 line is inserted in the first intron of the MKP3 gene (NP consortium, see Fig. 1). X-ray induced mitotic recombination clones were generated in flies of genotype f36a; M(3)i55P[f+]MKP376-M. Homozygous MKP3M76-M cells were recognized by the presence of the cell marker forked (f). As a source of P transposase, we used Δ2-3 and CyO, hop2. The P-UAS line we used was the double-headed P-GS inserted in a CyO chromosome (CyO, P-GS; Toba et al., 1999). Stocks for male recombination were w; ru e/TM2 and w; CyO, hop2/If; ru e. All phenotypes were analyzed at 25°C, and flies were mounted for microscopic examination in lactic acid–ethanol (1:1). Pictures were taken in an Axiophot microscope with a Spot digital camera and processed in Adobe Photoshop.
Generation of MKP3 Alleles by Male Recombination and by Chemical Mutagenesis
In a mutagenesis screen designed to identify genes affecting vein differentiation (C. Molnar, A. López-Barea and J.F. de Celis, unpublished observations), we isolated three independent P-GS insertions in the 5′ UTR of the MKP3 gene (see Fig. 1). The three insertions, named MKP3M76, MKP3S, and MKP3160 cause a similar loss-of-vein phenotype in combination with Gal4-shv3kpn and pupal lethality in combination with Gal4-638. Loss-of-function alleles were induced in two different ways: (1) by male recombination (Preston et al., 1996) in male flies of genotype w; CyO, hop2/+; ru1e11/MKP3M76 crossed to w; ru1e11 we generated two ru1MKP3M76 and five MKP3M76e11 recombinants. The ru1MKP3M76 chromosomes in combination with any Gal4 line are phenotypically normal, whereas all the MKP3M76e11 recombinants in combination with any Gal4 line retain the MKP3M76 gain-of-function phenotype (data not shown). This observation indicates that the gene responsible for the overexpression phenotype lies to the left end (distal) of the P-GS insertion. (2) By chemical mutagenesis, we treated males of genotype w; MKP3M76/MKP3M76 with 20 mM EMS according to Lewis and Bacher (1968). Treated males were mass-crossed with Gal4-638 females and viable progeny of genotype Gal4-638; MKP3M76/+ were isolated, making the stocks w; MKP3M76-rev/TM2. All flies of Gal4-638; MKP3M76/+ and Gal4-638/+; MKP3M76/+ genotype die in the pupal case. Among 3500 progeny (dead pupae that cannot leave the pupal case), we isolated two full revertants named MKP3M76-M and MKP3M76-R2b. The presence of mutations in both revertants was confirmed by sequencing all exons of the MKP3 gene (see below).
The P-GS lines MKP3M76, MKP3S, and MKP3160 were mapped by inverse PCR following standard procedures. Briefly, genomic DNA from the P-GS strains was digested by using HhaI or MspI and the resulting products circularized by using T4-DNA ligase. For the PCR reaction, we used the P-GS specific primers CTTCTTGGCAGATTTCAGTAGTTGC and ATTGCAAGCATACGTTAAGTGGA. For sequencing, we used the primer CGACGGGACCACCTTATGTTA. The resulting sequences were n-Blast against the Drosophila genomic sequence at NCBI. The two EMS-induced MKP3 alleles were sequenced as follows: Genomic DNA was prepared from homozygous flies (MKP3M76-R2b) or homozygous embryos (MKP3M76-M) and used as template for four different PCRs reactions. The primers used were (1) 5′UTR region, CCCGCGCTGCTTGTGTGTGCTCG (upstream) and GGTCTCGTGCTCCGTTTCTGG (downstream); (2) Exon 1, CCAGAAACGGAGCACGAGAC (upstream) and CTTGTAAAGCAACTTACGCGGC (downstream); (3) Exons 2 and 3, GGCTTCAACAATTTCGCCAGGC (upstream) and CTATAAACTGTATGGCATCCGGG (downstream); (4) Exons 4 and 5, GTTTTGAATGTGACACCAGATTTGC (upstream) and CTTCTTCGGCCATCTCCTGATCCCG (downstream). The resulting PCR products were cloned in pGEMTeasy (Invitrogen) and sequenced using T7 and Sp6 vector primers. At least four independent clones resulting from two different PCR reactions made with the same set of primers were sequenced to confirm the presence of nucleotide substitutions. The only differences in sequence we found were T to A transitions in the positions 19035101 (ATG to AAG in MKP3M76-R2b; Met to Lys) and 19018080 (TAC to AAC; Tyr to Asn in MKP3M76-M). Other sequence polymorphism found in our sequences was in position 19018904 in MKP3M76-R2b (ATC to ATT, both coding Ile).
Drosophila Inmunocytochemistry and In Situ Hybridization
We used rabbit anti-phosphorylated ERK (Cell signaling technologies), mouse monoclonals anti-Dl (Hybridoma bank) and anti-DSRF. Secondary antibodies were from Jackson Immunological Laboratories (used at 1/200 dilution). Pupal wings and third instar imaginal discs were dissected, fixed, and stained as described in de Celis (1997). Confocal images were captured using a Bio-Rad confocal microscopy. In situ hybridization in imaginal discs and embryos were carried out as described in de Celis (1997) and Ruiz-Gomez and Ghysen (1993), respectively. Digoxigenin-labeled RNA probes were prepared from the MKP3 cDNA clone SD06439 (Research genomics).
Xenopus In Situ Hybridization and X-Gal Staining
The Xenopus MKP3 probe was prepared by from the Mochii clone XL034f13 linearized with EcoRI and transcribed with T7 RNA polymerase using digoxigenin (Roche) as labels. Specimens were prepared, hybridized, and stained as described (Harland, 1991). X-Gal staining was carried out according to Coffman et al. (1993).
In Vitro RNA Synthesis
All DNAs were linearized and transcribed as described (Harland and Weintraub, 1985) with a GTP cap analogue (New England Biolabs). After DNAse treatment, RNA was extracted with phenol-chloroform, column purified, and precipitated with ethanol. mRNAs for injection were resuspended in water. DN-Ras encodes a mutant form of Ras with an Asn at position 17 that generate a dominant negative molecule (Whitman and Melton, 1992). Constitutively activated Ras encodes a mutant form of Ras with Val at position 12 that generate a constitutively active molecule (Whitman and Melton, 1992). Dominant negative FGF receptor type1 also denominated XFD is described in Amaya et al. (1991). Constitutively activated FGF type 1 receptor, also named t-R1, is described in Umbhauer et al. (2000).
We thank R. Hernandez for her skillful technical help and E. Amaya, H-F. Kung, R. Old, N. Papalopulu, J-F. Riou, K. Storey, and J. Smith for reagents. We also thank N. Ueno and the NIBB/NIG Xenopus laevis EST project for the Mochii clone XL034f13. We also thank A. García-Bellido and J. Modolell for their continuous support and R. Barrio, S. Campuzano, S. Sotillos, M. Suzzane, L. Baena, J.C. Pastor, and A. Glavic for critical reading of the manuscript. J.F.dC. and J.L.G.-S. were funded by the Dirección General de Investigación Cientófica y Técnica and an institutional grant from Fundación Ramón Areces to the Centro de Biología Molecular “Severo Ochoa” are also acknowledged.