Article. Phosphorylation of Groucho Mediates RTK Feedback Inhibition and Prolonged Pathway Target Gene Expression

Transcription

1 Current Biology 21, , July 12, 2011 ª2011 Elsevier Ltd All rights reserved DOI /j.cub Phosphorylation of Groucho Mediates RTK Feedback Inhibition and Prolonged Pathway Target Gene Expression Article Aharon Helman, 1 Einat Cinnamon, 1 Sharon Mezuman, 1 Zvi Hayouka, 2 Tonia Von Ohlen, 3 Amir Orian, 4 Gerardo Jiménez, 5 and Ze ev Paroush 1, * 1 Department of Developmental Biology and Cancer Research 2 Department of Biochemistry Institute of Medical Research Israel-Canada, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem 91120, Israel 3 Division of Biology, Kansas State University, Manhattan, KS 66506, USA 4 Cancer and Vascular Biology Research Center, Faculty of Medicine, Technion-Israel Institute of Technology, Haifa 31096, Israel 5 Institució Catalana de Recerca i Estudis Avançats and Institut de Biologia Molecular de Barcelona-CSIC, Parc Científic de Barcelona, Barcelona, Spain Summary Background: Signaling by receptor tyrosine kinase (RTK) pathways plays fundamental roles in processes of cell-fate determination, often through the induction of specific transcriptional responses. Yet it is not fully understood how continuous target gene expression, required for irreversible cell-fate specification, is preserved after RTK signaling has ended. Here we address this question using the Drosophila embryo, a model system that has been instrumental in elucidating the developmental functions of RTK signal transduction. Results: The Groucho corepressor is phosphorylated and downregulated in response to RTK signaling. Here we show that RTK pathways use Groucho phosphorylation as a general mechanism for inducing expression of pathway target genes encoding cell-fate determinants as well as feedback antagonists, indicating that relief of Groucho-dependent repression is an integral element of RTK signaling networks. We further demonstrate that after mitogen-activated protein kinase (MAPK) has been deactivated, sustained phosphorylation of Groucho is essential for persistent RTK-induced target gene expression and cell-fate determination in several developmental contexts. Conclusions: Phosphorylation of Groucho by MAPK plays a dual role in the regulation of RTK responses: (1) it mediates rapid feedback inhibition, and (2) it provides a stable memory mechanism of past MAPK activity. We propose that, in this manner, phosphorylation of Groucho enables transiently active RTK pathways to fix the spatiotemporal expression profiles of downstream targets over time. Introduction Receptor tyrosine kinase (RTK) signaling pathways control diverse cellular processes such as proliferation, migration, differentiation, and fate determination. RTK-transduced *Correspondence: zparoush@cc.huji.ac.il signals are often relayed via the canonical Ras/Raf/MEK cascade, culminating in the phosphorylation and activation of mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/Erk). Once activated, MAPK/Erk enters the nucleus, where it phosphorylates and modulates the activity of various transcriptional regulators, leading to specific changes in pathway target gene expression [1]. In development, RTK signaling activity is tightly restricted in time and space by various negative regulators. Accordingly, prominent among the downstream targets induced by RTK pathways are feedback antagonists that restrain RTK-dependent signaling. In Drosophila, these inhibitors include MAPK phosphatases (e.g., Mkp-3) and Argos (Aos) [2, 3]. As a result of negative feedback regulation, MAPK/Erk is quickly deactivated [4, 5], yet remarkably, expression of pathway targets is often continuous. How transient RTK signaling is converted into prolonged transcriptional regulation is not well understood. Here we address this question using the Drosophila embryo as a model system [6]. In early embryogenesis, patterning along the dorsoventral (DV) axis is controlled by the transcriptional activity of the Dorsal (Dl) morphogen. Graded nuclear distribution of Dl specifies three presumptive embryonic tissues: mesoderm, neuroectoderm, and dorsal ectoderm [7]. One of the many genes directly activated by Dl in the neuroectoderm is rhomboid (rho), which encodes a serine protease required for stimulating the epidermal growth factor receptor (EGFR) RTK pathway [8, 9]. Accordingly, expression of rho fully coincides with the pattern of activated MAPK in the neuroectoderm and other contexts [10]. rho is initially expressed at stage 5/6 (st5/6), about 2:40 3:00 hr after egg laying (hr AEL), in a wide lateral stripe in the neuroectodermal region [8, 11]. Later on, rho expression responds to dynamic changes in the Dl gradient and gradually narrows down to the ventral midline [12]. Once activated by Rho, the EGFR pathway, together with Dl, induces transcription of intermediate neuroblasts defective (ind), a gene required for the formation of neuroblasts (NBs) comprising the intermediate of three neuroectodermal NB columns [13]. Although ind is clearly a target of the EGFR pathway at st5/6 [14], the mechanisms regulating its expression are not well understood. Furthermore, ind transcription is maintained in the neuroectoderm from st7 (3:00 hr AEL) onward, long after EGFR pathway activity retracts toward the ventral midline following the changes in rho expression. Thus, ind is an EGFR target whose expression in the neuroectoderm persists long after RTK signaling has terminated in this region. Here we show that downregulation of the Groucho (Gro) developmental corepressor, via MAPK phosphorylation [15, 16], plays a dual role in the spatiotemporal regulation of RTK responses. First, it provides a mechanism required for rapid induction of RTK target genes encoding cell-fate determinants as well as negative feedback regulators. Second, it enables extended expression of targets long after pathway activity has terminated. Specifically, we demonstrate that the induction of ind and aos expression by the EGFR pathway occurs by relief of Gro-mediated repression. Additionally, we show that phosphorylation of Gro is sustained for several hours after

2 Persistent Phosphorylation of Gro in RTK Signaling 1103 Figure 1. Phosphorylation of Groucho and Downregulation of Its Repressor Activity Are Required for Ind Expression and Function (A and B) Lateral view of st5/6 wild-type embryos stained for dperk (A; green) and for pgro (B; red). (A) Anti-dpERK staining delineates the neuroectodermal region where early EGFR pathway activation takes place. (B) Gro is phosphorylated in the neuroectoderm, as well as in the embryonic termini and in seven transverse stripes [18]. (C) DSor mutant embryo stained for pgro. The lateral stripe of EGFR-dependent Gro phosphorylation is greatly reduced, as is phosphorylation in response to Torso RTK signaling at the poles. (D F) Lateral view of st6 embryos. Expression of Ind (green) is reduced in both DSor mutant embryo (E) and embryo expressing Gro AA (F), whereas Vnd (red) is expressed in both backgrounds. (D) Control embryo expressing LacZ. (D 0 F 0 ) Ind expression alone. (G I) Lateral view of st6 maternal mutant embryos stained for Ind (green). Ind expression is derepressed by 1 3 cell diameters in gro mutant (G; cf. D 0 ), is completely absent in ras mutant (H), and is partially rescued in ras gro double mutant (I). In this and subsequent figures, embryos are oriented with anterior to the left and dorsal side up unless stated otherwise. (J) Simplified model for ind regulation. (K M 0 ) Formation of intermediate neuroblasts in the neuroectoderm depends on phosphorylation and downregulation of Groucho. (K M) Ventral view of st10 DSor mutant embryo (L), embryo maternally expressing Gro AA (M), or control embryo maternally expressing LacZ (K), stained for the pan-neuroblast (NB) marker Hb. (K 0 M 0 ) Higher magnification of embryos similarly stained for Hb. (K and K 0 ) Three Hb-expressing neuroectodermal NB columns form on either side of the ventral midline. (L M 0 ) Note the formation of only two NB columns and the specific loss of intermediate NBs, brought about by blocking either EGFR signaling (L and L 0 ) or Gro phosphorylation (M and M 0 ). In all panels, the ventral midline is demarcated by the dashed line. MAPK has been deactivated and that this modification is required for maintaining continuous ind expression. Similarly, phosphorylation of Gro in response to EGFR, Torso, and fibroblast growth factor receptor (FGFR) RTK signaling correlates with feedback inhibition and maintenance of pathway target expression in the eye imaginal disc, at the embryonic poles, and in mesodermal Eve-positive pericardial cells, respectively. Based on these results, we propose that Gro is a central component of RTK feedback networks and that stable phosphorylation of Gro represents a molecular imprint required for sustained RTK-dependent gene expression in development. Results The EGFR Pathway Promotes ind Expression via Relief of Groucho-Dependent Repression Analysis of a functional ind enhancer uncovered potential binding sites for several Gro-dependent repressors [17], prompting us to investigate the role of Gro in ind regulation. We first stained gro mutants for Ind, finding its expression derepressed by 1 3 cell diameters (Figure 1G; see also Figure S1 available online). Although the effect is not dramatic, it nevertheless implicates Gro as a negative regulator of ind. Furthermore, the observed expansion resembles the effect caused by EGFR overactivation in the embryo [14], raising the possibility that both effects are linked. Thus, we asked whether EGFR signaling modifies Gro in the presumptive neuroectoderm at embryonic st5/6. Embryos were stained for doubly phosphorylated MAPK (dperk) as readout for RTK pathway activity, as well as for phosphorylated Gro (pgro) [18, 19]. As Figures 1A 1C show, Gro phosphorylation overlaps with early EGFR pathway activation. To investigate whether phosphorylation of Gro is relevant to ind expression, we maternally expressed an unphosphorylatable Gro variant in which the phosphoacceptor amino acids residing in two MAPK phosphorylation sites were replaced by alanine residues (Gro AA ; T308A and S510A) [15, 18]. As shown in Figure 1 and Figure S1, Gro AA expression causes a significant reduction in Ind protein (Figures 1F and 1F 0 ) and ind mrna levels (Figure S1; 88%; n = 90; n indicates the number of embryos scored). This result is similar to that caused by mutations in DSor (Drosophila MEK; Figures 1E and 1E 0 ; 100%; n = 25) and ras (Figure 1H; 100%; n = 30). The effect is specific because both Gro AA embryos and DSor mutants still express the medial column homeobox gene ventral nervous system defective (vnd) (Figures 1D 1F) [14]. In comparison, expression of the native form of Gro or a phosphomimetic Gro derivative (Gro DD ; T308D and S510D) [15, 18] exerts a milder effect (Figure S1). Thus, blocking Gro

3 Current Biology Vol 21 No Figure 2. Phosphorylation of Gro Correlates with Mesodermal eve Expression (A B 00 ) Late st10 wild-type embryos stained for Eve (A and B; red), dperk (A 0 ; green), and pgro (B 0 ; blue). (A 00 and B 00 ) Merge. At this stage, MAPK is activated in response to EGFR and FGFR signaling, leading to the phosphorylation of Gro. (C F) Arrowheads demarcate the position of one of the clusters. Lateral view of st12 wild-type (C) and zygotic ras (D), gro (E), and ras gro (F) mutant embryos, stained for Eve (red). In wild-type embryos, Eve expression defines ten mesodermal clusters, composed of heart and muscle progenitors (C). In comparison, the number of progenitor clusters is markedly reduced in ras (D) but not in gro mutants (E). In the latter background, eve is derepressed and many clusters contain additional Eve-expressing cells. (F) Expression of eve and formation of progenitor clusters are partially rescued in ras gro double mutants (cf. D), suggesting that relief of Gro-dependent repression by RTK signaling is required for Eve induction and cluster formation. (G) Average number of clusters per embryo in each background is presented. Note that, on average, less than one cluster is observed in ras mutant embryos, compared to 4.7 clusters in ras gro mutants. Results are the average 6 standard deviation of the rates calculated from ten embryos per each genetic background. phosphorylation mimics a transcriptional defect caused by the loss of EGFR signaling, indicating that MAPK-mediated downregulation of Gro repressor activity is essential for the expression of ind in response to EGFR activity (Figure 1J). If EGFR signaling induces ind by relieving Gro-dependent repression, could the removal of Gro bypass the requirement for EGFR activity? Indeed, whereas Ind is never detected in ras single mutants (Figure 1H), it is partially restored in all developed ras gro double-mutant embryos (Figure 1I; Figure S1; 100%; n = 11). Given that Ind expression is not rescued to its normal extent in ras gro mutant embryos, we also conclude that additional Gro-independent inputs by EGFR signaling are necessary for full ind expression in the neuroectoderm. To explore the functional consequences of Gro phosphorylation, we examined the formation of the EGFR-dependent intermediate column NBs in its absence. Mutations in genes encoding EGFR or other elements of the pathway, such as DSor, lead to loss of the intermediate NBs column (Figures 1L and 1L 0 ; 100%; n = 30) [20 22]. Likewise, we find that intermediate NBs are significantly lost in Gro AA -expressing embryos (Figures 1M and 1M 0 ; 86%; n = 50), whereas the expression of Gro and Gro DD causes a more moderate decrease of NBs (56%; n = 32 and 51%; n = 33, respectively; data not shown). These data suggest that downregulation of Gro repressor activity is essential for EGFR-mediated formation of intermediate NBs. RTK-Induced Mesodermal even-skipped Expression Requires Downregulation of Groucho-Mediated Repression The attenuation of Gro repression by RTK signaling is also required for the induction of pathway targets in other contexts. For example, FGFR signaling contributes to heart formation by activating the mesodermal expression of even-skipped (eve), thus specifying Eve-positive pericardial progenitor cells (EPCs) [23, 24]. Correspondingly, MAPK is transiently active in the ten mesodermal Eve-expressing clusters that form at st10 (5 hr AEL) (Figures 2A 2A 00 )[25, 26]. In the absence of RTK signaling at this stage, eve expression is lost and these cells are misspecified (Figure 2D) [26]. Gro is phosphorylated in st10 Eve-positive clusters (Figures 2B 2B 00 ), raising the possibility that RTK signals induce eve by downregulating Gro-mediated repression. Consistent with this idea, Eve-positive clusters are significantly rescued in a zygotic ras gro double-mutant background (Figures 2F and 2G). These results further support the idea that attenuation of Gro repressor function by phosphorylation is a prevalent mechanism employed by multiple RTK pathways in turning on their targets. Phosphorylation of Groucho Is Required for Negative Feedback Regulation of RTK Signaling Besides ind and eve, phosphorylation of Gro mediates derepression of other key RTK pathway targets. Significantly, several genes encoding feedback antagonists of RTK signaling, generally viewed as the universal targets of all RTK pathways, are also under this mode of regulation. For instance, the negative inhibitor argos (aos) is normally transcribed in the ventral ectoderm at st7 and onward, and in the termini at st4 (1:20 2:10 hr AEL), in response to EGFR and Torso RTK signaling, respectively (Figures 3A and 3G) [2, 27, 28]. We find that aos is derepressed in gro mutant embryos, indicating that Gro is normally a repressor of aos expression (Figure 3H). By contrast, aos is repressed in both DSor mutants (Figure 3B; 100%; n = 25 and Figure 3I; 100%; n = 25) and Gro AA embryos (Figure 3C; 96%; n = 29 and Figure 3J; 74%; n = 23). Additionally, expression of the MAPK phosphatase mkp-3, detected at the poles of st4 wild-type embryos (Figure 3K) [3, 29], is significantly derepressed in gro mutants (Figure 3L) and, conversely, is reduced (particularly at the anterior) in embryos with decreased Torso signaling (Figure 3M; 100%; n = 25), as well as in Gro AA -expressing embryos (Figure 3N; 79%; n = 54). Consistent with the above findings, we note that dperk staining in the neuroectoderm of st9 (3:40-4:20 hr AEL)

4 Persistent Phosphorylation of Gro in RTK Signaling 1105 Figure 3. Relief of Groucho-Mediated Repression by MAPK Phosphorylation Mediates Feedback Inhibition of RTK Signaling (A C and G N) In situ hybridization using digoxigenin-labeled RNA probes. (D F and O R) Embryos stained for dperk (green). (A C) Ventral view of st9 embryos stained for aos, an EGFR target gene encoding a central feedback antagonist of the pathway. (A) Control embryo showing the normal aos expression pattern at this stage. (B and C) aos is never expressed in the neuroectoderm of Dsor mutants (B) and is repressed in Gro AA -expressing embryos (C). Note the two unaffected aos head patches in all three backgrounds. (D F) Ventrolateral view of st9 embryos. MAPK activation is stronger in the neuroectoderm of Gro AA -expressing embryo (F) than in wild-type control (D) and Gro DD -expressing embryo (E). (G J) St4 embryos stained for aos. aos expression is derepressed in maternal gro mutant (H). By contrast, terminal aos expression domains are reduced in DSor mutant (I) and embryo maternally expressing Gro AA (J), compared to control embryo maternally expressing LacZ (G). (K N) St4 embryos stained for mkp-3. (K) Control embryo maternally expressing LacZ. (L N) Expression of mkp-3 expands throughout maternal gro mutant (L), and is blocked at the termini (particularly at the anterior pole) in torsolike mutant in which the Torso pathway is inactive (tsl 691 ; M) and in embryo maternally expressing Gro AA (N). (O R) Lateral view of st5/6 embryos. MAPK activation is significantly reduced in gro mutant (P) compared to wild-type control (O), in accordance with the ectopic expression of feedback inhibitors in this background (H and L). In comparison, prolonged MAPK activation is evident at the termini of Gro AA -expressing embryo (R, arrows; cf. O), in the domain of Torso RTK signaling, corresponding to the reduction in aos and mkp-3 expression in the pole regions (J and N). dperk staining is undetectable in DSor mutant (Q) as a result of impaired RTK signaling. Gro AA -, but not Gro DD -, expressing embryos is stronger than in wild-type (Figures 3D 3F). Notably, this effect is not observed at st5/6 when EGFR activation is initiated (Figures 3O and 3R), suggesting that subsequent MAPK overactivation reflects a defect at the level of feedback inhibition. Similarly, there are no significant differences in Torso-dependent dperk levels between wild-type and Gro AA embryos at st4/5 (2:00 2:20 hr AEL; data not shown); however, by st5/6 we observe persistent dperk staining only in Gro AA -expressing embryos (Figure 3R, arrows). Conversely, in gro mutant embryos, a significant reduction in dperk staining is seen (Figure 3P; Figure S2). Remarkably, the prolonged and increased levels of dperk staining in Gro AA -expressing embryos (Figures 3R and 3F) are comparable to those evident at the termini of st5/6 mkp-3 and in the neuroectoderm of st9 aos mutant embryos, respectively (Figure S2). Taken together, our results strongly indicate that activation of multiple feedback antagonists of RTK signaling requires downregulation of Gro-dependent repression, suggesting that Gro is a central component of the feedback networks controlling the magnitude and duration of MAPK signals. Persistent Phosphorylation of Groucho Enables Sustained Expression of RTK Pathway Target Genes As mentioned above, ind is continuously expressed in the lateral neuroectoderm, even while EGFR pathway activity gradually retracts toward the ventral midline. In fact, at st7 and later, the Ind expression domain does not overlap any longer with that of dperk (Figure 4E). What could be the mechanism that maintains expression of ind and other RTK pathway target genes after MAPK has been deactivated? The Capicua (Cic) repressor restricts the early (st5/6) dorsal limit of ind expression, and this repression is alleviated by EGFR-dependent downregulation of Cic protein levels (Figure S1)[30]. We initially explored a possible role for Cic in maintaining accurate limits of ind expression as development proceeds, by analyzing late (st8) Cic protein accumulation relative to the Vnd expression domain. We find that as EGFR pathway activity progressively retracts toward the ventral midline from st7, newly synthesized Cic accumulates in the same nuclei that continuously express Ind on both sides of the Vnd domain (Figures 4A 4A 00 ). Thus, the pattern of Cic degradation accurately reflects dynamic RTK signaling activity, and when MAPK activation terminates, new Cic protein is rapidly produced. Accordingly, downregulation of Cic likely plays a role only in the induction, but not in the maintenance, of ind expression. In contrast, Gro remains constantly phosphorylated in the neuroectoderm, and, from st7 onward, the pattern of pgro is broader than that of dperk (Figure 4C). Could prolonged phosphorylation of Gro serve as memory of previous MAPK activity, to maintain ind expression in the neuroectoderm even in the presence of newly synthesized Cic protein (Figures 4B 4B 00 )? First, we unequivocally established that

5 Current Biology Vol 21 No Figure 4. Stable Phosphorylation of Groucho Maintains Continuous Ind Expression in the Absence of EGFR Activity (A and A 00 ) Ventrolateral view of st8 wild-type embryo double stained for Vnd (A; green) and Cic (A 0 ; red). (A 00 ) Merge. By this stage, EGFR pathway activity has retracted from the region of Ind expression abutting the Vnd domain (arrowhead), where newly synthesized Cic protein now accumulates. (B and B 00 ) Ventrolateral view of st8 wild-type embryo double stained for unphosphorylated Gro (B; green) and Cic (B 0 ; red). (B) There is no anti-gro staining in the neuroectoderm, in the region of earlier EGFR activity. The lack of anti-gro signal in the Ind domain (arrowhead) indicates that Gro is in its phosphorylated state [18]. (B 0 ) Cic is freshly produced in the Ind domain, following the cessation of EGFR signaling in this region. Thus, phosphorylated Gro and Cic overlap where ind is continuously expressed. (B 00 ) Merge. (C D) Ventral view of st10 wild-type (C) and sim mutant (D) embryos, stained for dperk (green) and pgro (red). (C) The pgro domain is broader than that of dperk. (C 0 ) pgro alone. (D) dperk staining is insignificant in sim mutants, yet Gro is still markedly phosphorylated in the neuroectoderm. (E) Ventral view of st10 wild-type embryo stained for dperk (green) and Ind (red). EGFR activation has narrowed down by this time to a region adjacent to the ventral midline, whereas Ind expression in the neuroectoderm extends beyond the dperk domain and overlaps with that of pgro (cf. [C]). (F G) Ventral view of st10 embryos ectopically expressing Gro (F) or Ras DN (G) under Kr-Gal4 regulation, stained for unphosphorylated Gro (blue) and Ind (red). Gro represses Ind expression in the Kr domain (F), whereas Ras DN has no effect (G). In st6 Kr>Gro embryos, Gro is undetectable and the Ind pattern is indistinguishable from the wild-type (not shown), ruling out the possibility that the Kr-Gal4 driver is leaky at earlier stages. (H and H 00 ) Ventral view of st10 embryo, ectopically expressing unphosphorylated Gro (H; red) under Kr-Gal4 regulation and stained for the panneuronal marker Hb (H 0 ; green). (H 00 ) Merge. Misexpressed Gro compromises the formation of intermediate column NBs cell-autonomously. neuroectodermal phosphorylation of Gro is indeed stable with time by analyzing pgro in single-minded (sim) mutants in which only early (st5/6) EGFR activation takes place. Although sim is not an inherent component of the EGFR pathway, the ventral midline does not form in its absence and, consequently, late EGFR activation at st10 (4:20 5:20 hr AEL) does not occur [31]. Nonetheless, Gro remains phosphorylated in the neuroectoderm of st10 sim mutants, including in Ind-expressing cells (Figure 4D; Figure S3), indicating that pgro persists at least throughout the 2 hr interval separating the two waves of EGFR activation. To test whether pgro plays a role in maintaining neuroectodermal ind expression independently of EGFR activity during st7 10 (3:00 5:20 hr AEL), we expressed the native form of Gro under the regulation of Kr-Gal4 (Kr-Gro). Kr-Gal4 drives expression starting from st8 (3:10 3:40 hr AEL), after the early wave of EGFR signaling has terminated [32]. An antibody specific for the unphosphorylated form of Gro [18] confirmed that Kr-Gal4 drives Gro expression from st8 only and that Gro is unmodified by MAPK at this stage (Figure 4F). Strikingly, double staining revealed that Ind expression is clearly reduced in cells that express ectopic Gro protein. Some embryos showed a severe reduction in Ind expression (Figure 4F), whereas others showed a milder effect (15% and 43%, respectively; n = 35). Notably, even the latter embryos lack Ind expression in cells that strongly express Gro (Figure S3). As control, Ind expression was only mildly affected by Gro DD (10%; n = 30) and was not at all affected in embryos similarly expressing a dominant-negative form of Ras (Ras DN ) (Figure 4G; Figure S3) [33]. Taken together, our results indicate that phosphorylation of Gro is crucial to permit the continuing expression of ind starting from st7, when Cic protein is produced anew. Thus, initiation of ind expression depends on EGFR pathway-mediated downregulation of both Cic and Gro; however, its maintenance responds solely to persistent phosphorylation of Gro. Furthermore, long-term ind expression is functionally important, given that misexpression of Gro blocks formation of intermediate column NBs in the Kr domain (Figures 4H 4H 00 ). Continuous RTK Pathway Target Gene Expression Correlates with Sustained Groucho Phosphorylation in Multiple Developmental Processes Persistent phosphorylation of Gro also correlates with prolonged pathway target gene expression at other stages, downstream of multiple RTK pathways, suggesting that this memory mechanism is a general one. For instance, in thirdinstar eye imaginal discs, dperk staining is most prominent in the row of photoreceptor clusters immediately posterior to the morphogenetic furrow, whereas it is significantly less so in more posterior ommatidial rows (Figures 5A 5A 00 ) [19]. Expression of the pathway target aos is detected only posteriorly to the first row, where it presumably acts to dampen MAPK activation (Figures 5A 5A 00 ). Notably, phosphorylation of Gro is discernable both in the first row and in more posterior cells in which dperk levels have declined yet aos expression is robust

6 Persistent Phosphorylation of Gro in RTK Signaling 1107 Figure 5. Stable Groucho Phosphorylation Correlates with, and Is Required for, Continuous RTK Pathway Target Gene Expression (A and B) Third-instar eye discs derived from aos-lacz flies, stained for dperk (A; green), pgro (B; blue), and LacZ (aos-lacz; A and B; red). (A 0,A 00,B 0, and B 00 ) Enlargements. Stable phosphorylation of Gro correlates with sustained aos expression even in cells with little or no EGFR pathway activity. (C E) Stable phosphorylation of Gro correlates with sustained hkb expression. (C) Lateral view of st7 wild-type embryo stained for dperk (green) and pgro (red). Early activation of MAPK at the poles in response to Torso RTK signaling has diminished by this stage; however, phosphorylation of Gro persists (arrows). (D) At the same time, the Torso pathway target hkb is still expressed in the domain of phosphorylated Gro (arrows). (E) Ventrolateral view of st7 wild-type embryo stained for Cic (green). Cic, previously phosphorylated and degraded by Torso signaling at st4, is freshly produced at the poles (arrows). Note the degradation of Cic in the neuroectoderm at this stage. (F I) Stable Gro phosphorylation is essential for prolonged Eve expression in pericardial cells. (F and G) St14 wild-type embryos stained for dperk (F; green), pgro (G; blue), and Eve (F and G; red). By this time, RTK activity has diminished in Eve-positive cells, yet Gro remains phosphorylated. (H and I) St14 wild-type (H) and Gro-expressing (I) embryos stained for Gro (green) and Eve (red). (I) Nonphosphoryated Gro, driven by 24B-Gal4, interferes with continuous Eve expression cell autonomously (cf. [H]). Eve-positive pericardial progenitor cells (EPCs) that do not express Gro still stain for Eve. Arrows in all panels demarcate pairs of EPCs. (Figures 5B 5B 00 ). Thus, in the eye, phosphorylation of Gro follows the pattern of EGFR signaling, and pgro remains stable even in cells in which MAPK has been deactivated, correlating with sustained pathway target gene expression. Phosphorylation of Gro in response to Torso-mediated signaling also illustrates this trend. Torso is briefly active at the embryonic poles at st4, where it phosphorylates and downregulates Cic- and Gro-dependent repression allowing huckebein (hkb) transcription [18, 34]. We find that both phosphorylated Gro as well as hkb expression persist long after the Torso pathway has been turned off (Figures 5C and 5D). Again, stable Gro phosphorylation correlates with sustained RTKinduced gene expression at the time when newly synthesized Cic is accumulating (Figure 5E). Finally, stable phosphorylation of Gro also plays a role in maintaining prolonged Eve expression in EPCs at st12 (7:20 9:20 hr AEL) and later, when dperk is no longer detectable [25, 26]. Induction of eve expression in cardiac cell clusters is initiated by transient FGFR activity at st10 (Figure 2), but eve is continually expressed in mesodermal EPCs long after termination of the pathway [25, 26]. Double staining for both Eve and pgro reveals that pgro persists in these cells independently of Ras/MAPK activity at least until st14 (10:20 11:20 hr AEL) (Figures 5F and 5G). Strikingly, expression of native Gro under the regulation of 24B-Gal4, which drives expression in cardiac and somatic mesoderm from st11 onward [26], reduces or abolishes Eve expression cell autonomously (Figures 5H and 5I). These results, together with those presented above, indicate that phosphorylation of Gro is a primary mechanism by which expression of many RTK pathway targets is induced and then maintained after signaling activity has ended. Discussion In this work, we have uncovered central roles for Gro as a key effector of RTK signaling networks. Our results indicate that, prior to signal transduction, Gro suppresses the expression of multiple pathway target genes. Indeed, DNA adenine methyltransferase identification (DamID) experiments in resting Drosophila Kc cells show that Gro is tethered to DNA in proximity to many known RTK downstream genes, including ind and aos, suggesting that these represent direct targets for Gro-mediated repression [22, 35] (A.O., unpublished data).

7 Current Biology Vol 21 No Figure 6. Model; Stable Gro Phosphorylation Provides a Long-Term Mechanism Required to Maintain the Spatiotemporal Extent of RTK Pathway Target Gene Expression after MAPK Activity Has Declined (A) Schematic cross-sections depicting domains of dperk staining (green), phosphorylated Gro (blue), and ind expression (red), in st6 (left) and st9 (right) embryos. Initially rho is expressed, and MAPK is active, in the neuroectoderm. With time, rho expression retracts toward the ventral midline, and RTK feedback inhibitors terminate MAPK activity in the neuroectoderm. Nevertheless, Gro is stably phosphorylated and ind continually expressed in this domain. (B) Lateral view. The Torso RTK pathway is briefly active at st4 (left) at the termini of the embryo (dperk; green), where it phosphorylates Gro (blue) and relieves repression of hkb (red). At st7 (right), both phosphorylation of Gro and hkb expression persist, long after Torso signaling has been inactivated. (C) Lateral view. At st10 (left), FGFR-mediated mesodermal signaling (dperk; green) phosphorylates Gro (blue) and induces expression of eve in ten clusters (only some depicted). At st14 (right), MAPK has been deactivated, yet persistent phosphorylation of Gro enables the continuous expression of eve. Upon phosphorylation by MAPK, Gro-dependent repression is relieved and downstream targets are activated. Products of some of these genes impart stable cellular fates, whereas others mediate rapid negative feedback regulation. Our study thus highlights an intimate connection between phosphorylation of Gro and feedback inhibition of RTK signaling (Figure 3). Significantly, phosphorylation of Gro provides a mechanism of long-term derepression, allowing prolonged transcription of RTK targets after MAPK activity has declined (Figure 6). This mechanism controls the persistent spatiotemporal expression of ind in the neuroectoderm in response to early EGFR activation; prolonged expression of ind is subsequently required for the formation of intermediate NBs and, hence, must be maintained for several hours. Other genes such as rho and short gastrulation are initially expressed in the neuroectoderm but are then progressively lost from this region [12]. These genes are targets of Dl, but not of the EGFR pathway or of Gro (not shown), and are therefore not subject to this mode of regulation. Importantly, long-lasting derepression mediated by Gro phosphorylation is also similarly required for sustained expression of eve in EPCs, as well as that of aos and hkb in eye imaginal discs and at the embryonic termini, respectively. Interference with this mechanism results in the loss of ind and eve expression, consequently leading to depletion of intermediate NBs and misspecification of mesodermal EPCs (Figure 4; Figure 5). The induction of many RTK targets involves degradation of Cic [29, 36 38]. However, Cic is degraded only where RTK signaling actively takes place but is newly synthesized once feedback inhibitors deactivate MAPK and signal transduction terminates (Figure 4; Figure 5). Thus, downregulation of Cic does not appear to play a role in maintaining target expression. We propose that prolonged phosphorylation and attenuation of Gro repressor function is crucial for fixing localized patterns of gene expression while Cic is being produced de novo. If stable pgro maintains transcriptional memory of transient MAPK signals, then unphosphorylated active Gro should not accumulate during this process. Indeed, ectopic expression of unphosphorylated Gro perturbs this regulatory mechanism and leads to the loss of RTK target gene expression. The marked dichotomy between the phosphorylated and unphosphorylated states of Gro [18] suggests that Gro phosphorylation is highly efficient and that the protein is stable in both states. It is also possible that a more active mechanism restricts the synthesis of new unphosphorylated Gro in cells in which it is modified. As shown here, phosphorylation and downregulation of Gro by MAPK provide a mechanism both for initial derepression of pathway target genes as well as for maintenance of their expression over time. However, additional Gro-independent inputs must also be controlling RTK-dependent gene expression, in the form of relief of one or many additional repressors and/or potentiation of activators. For example, ind and eve expression are only partially restored in ras gro double-mutant embryos (Figure 1; Figure 2). Additionally, ind is positively autoregulated, contributing to persistent ind transcription in the developing embryonic nervous system [39]. In conclusion, downregulation of Gro-mediated repression is a basic component of different RTK pathways, playing a dual role in restricting RTK signaling as well as in maintaining spatiotemporal outputs of transient MAPK activity. Future studies will show whether persistent phosphorylation of

8 Persistent Phosphorylation of Gro in RTK Signaling 1109 additional MAPK/Erk2 substrates is also important in contexts where brief activation of MAPK/Erk2 signaling controls sustained developmental responses. Experimental Procedures Fly Stocks The following mutant alleles and Gal4 drivers were used: sim H9, EGFR f2, aos I(3) 5959, aos ld7 (Benny Shilo), mkp-3 1 (Yoosik Kim), cic 1, tsl 691, nos-gal4- VP16, UASp-LacZ, UASp-Gro, UASp-Gro AA, UASp-Gro DD [18], UASt-Gro [15], Kr-Gal4, 24B-Gal4, and UASt-Ras N17 (Bloomington Stock Center). yw and histone-gfp stocks (Stas Shvartsman) [40] served as wild-type controls. Embryos lacking maternal or zygotic gro, ras, ras gro, and DSor activities were derived from mosaic gro E48 and gro BX22, ras1 e2f, ras1 e2f gro BX22, and Dsor LH110 (FlyBase) mutant germlines, respectively [41]. The vast majority of ras gro double-mutant embryos are undeveloped. To distinguish those that do mature, we costained embryos together with either Vnd or Wingless. In Situ Hybridization and Antibody Staining Embryos were dechorionated in bleach and fixed in 4% formaldehyde/ phosphate-buffered saline (PBS)/heptane for min. Expression patterns of ind, aos, mkp-3, and hkb were visualized by whole-mount in situ hybridization using digoxigenin-utp-labeled antisense RNA probes and anti-digoxigenin antibodies conjugated to alkaline phosphatase (Roche). Fluorescent immunohistochemical detection of activated MAPK, in freshly fixed embryos (10% formaldehyde/pbs/heptane buffer), was achieved with rabbit adperk (1:100; Cell Signaling) or mouse adperk (1:100; Sigma) [42]. Other antibodies used were as follows: rabbit apgro (1:100) [18], rabbit aind (1:1000), rat avnd (1:1000), rabbit ahb (1:1000), mouse agro (1;1000; Christos Delidakis), mouse alacz (1:1000), rabbit aeve (1:1000; Talila Volk), mouse aeve (1:20; Developmental Studies Hybridoma Bank), and rabbit acic (1:1000). Secondary antibodies were FITC- (1:2000), rhodamine- (1:2000), or Cy5-conjugated (1:800) (Jackson Laboratories). Embryos were mounted using DakoCytomation medium. Supplemental Information Supplemental Information includes three figures and can be found with this article online at doi: /j.cub Acknowledgments We thank Albert Courey, Rutie Finkelstein, Dennis Kuo, Yaron Suissa, Oren Ziv, and members of our laboratory for continued help. We are grateful to David Ish-Horowicz and Benny Shilo for their comments on the manuscript; to Benny Shilo for continuous encouragement; and to Rich Binari, Shari Carmon, Bill Chia, Christos Delidakis, Chris Doe, Greg Fish, Barbara Jennings, Yoosik Kim, Mark Krasnow, Maria Leptin, Norbert Perrimon, Benny Shilo, Stas Shvartsman, Angela Stathopoulos, Lisa Vogelsang, Talila Volk, Nir Yakoby, the Developmental Studies Hybridoma Bank, and the Bloomington Stock Centre for antibodies, fly stocks, and other reagents. This research was supported by grants from the Israel Science Foundation (Center of Excellence; 180/09) and the Król Charitable Foundation to Z.P.; G.J. was supported by the Institució Catalana de Recerca i Estudis Avançats (ICREA), the Ministerio de Ciencia e Innovación (BFU ), and the Generalitat de Catalunya (2009SGR-1075); and A.O. was supported by the Israel Science Foundation. A.H. was the recipient of a PhD fellowship from the Rector of The Hebrew University of Jerusalem. Received: December 25, 2010 Revised: April 5, 2011 Accepted: May 24, 2011 Published online: June 16, 2011 References 1. Shaul, Y.D., and Seger, R. (2007). The MEK/ERK cascade: From signaling specificity to diverse functions. Biochim. Biophys. Acta 1773, Golembo, M., Schweitzer, R., Freeman, M., and Shilo, B.Z. (1996). argos transcription is induced by the Drosophila EGF receptor pathway to form an inhibitory feedback loop. Development 122, Kim, M., Cha, G.H., Kim, S., Lee, J.H., Park, J., Koh, H., Choi, K.Y., and Chung, J. (2004). MKP-3 has essential roles as a negative regulator of the Ras/mitogen-activated protein kinase pathway during Drosophila development. Mol. Cell. Biol. 24, Amit, I., Citri, A., Shay, T., Lu, Y., Katz, M., Zhang, F., Tarcic, G., Siwak, D., Lahad, J., Jacob-Hirsch, J., et al. (2007). A module of negative feedback regulators defines growth factor signaling. Nat. Genet. 39, Santos, S.D., Verveer, P.J., and Bastiaens, P.I. (2007). Growth factorinduced MAPK network topology shapes Erk response determining PC-12 cell fate. Nat. Cell Biol. 9, Shilo, B.Z. (2003). Signaling by the Drosophila epidermal growth factor receptor pathway during development. Exp. Cell Res. 284, Stathopoulos, A., and Levine, M. (2002). Dorsal gradient networks in the Drosophila embryo. Dev. Biol. 246, Ip, Y.T., Park, R.E., Kosman, D., Bier, E., and Levine, M. (1992). The dorsal gradient morphogen regulates stripes of rhomboid expression in the presumptive neuroectoderm of the Drosophila embryo. Genes Dev. 6, Bang, A.G., and Kintner, C. (2000). Rhomboid and Star facilitate presentation and processing of the Drosophila TGF-alpha homolog Spitz. Genes Dev. 14, Gabay, L., Seger, R., and Shilo, B.Z. (1997). MAP kinase in situ activation atlas during Drosophila embryogenesis. Development 124, Rusch, J., and Levine, M. (1996). Threshold responses to the dorsal regulatory gradient and the subdivision of primary tissue territories in the Drosophila embryo. Curr. Opin. Genet. Dev. 6, Bier, E., Jan, L.Y., and Jan, Y.N. (1990). rhomboid, a gene required for dorsoventral axis establishment and peripheral nervous system development in Drosophila melanogaster. Genes Dev. 4, Weiss, J.B., Von Ohlen, T., Mellerick, D.M., Dressler, G., Doe, C.Q., and Scott, M.P. (1998). Dorsoventral patterning in the Drosophila central nervous system: The intermediate neuroblasts defective homeobox gene specifies intermediate column identity. Genes Dev. 12, von Ohlen, T., and Doe, C.Q. (2000). Convergence of dorsal, dpp, and egfr signaling pathways subdivides the drosophila neuroectoderm into three dorsal-ventral columns. Dev. Biol. 224, Hasson, P., Egoz, N., Winkler, C., Volohonsky, G., Jia, S., Dinur, T., Volk, T., Courey, A.J., and Paroush, Z. (2005). EGFR signaling attenuates Groucho-dependent repression to antagonize Notch transcriptional output. Nat. Genet. 37, Cinnamon, E., and Paroush, Z. (2008). Context-dependent regulation of Groucho/TLE-mediated repression. Curr. Opin. Genet. Dev. 18, Stathopoulos, A., and Levine, M. (2005). Localized repressors delineate the neurogenic ectoderm in the early Drosophila embryo. Dev. Biol. 280, Cinnamon, E., Helman, A., Ben-Haroush Schyr, R., Orian, A., Jiménez, G., and Paroush, Z. (2008). Multiple RTK pathways downregulate Groucho-mediated repression in Drosophila embryogenesis. Development 135, Gabay, L., Seger, R., and Shilo, B.Z. (1997). In situ activation pattern of Drosophila EGF receptor pathway during development. Science 277, Skeath, J.B. (1998). The Drosophila EGF receptor controls the formation and specification of neuroblasts along the dorsal-ventral axis of the Drosophila embryo. Development 125, Yagi, Y., Suzuki, T., and Hayashi, S. (1998). Interaction between Drosophila EGF receptor and vnd determines three dorsoventral domains of the neuroectoderm. Development 125, Orian, A., Delrow, J.J., Rosales Nieves, A.E., Abed, M., Metzger, D., Paroush, Z., Eisenman, R.N., and Parkhurst, S.M. (2007). A Myc- Groucho complex integrates EGF and Notch signaling to regulate neural development. Proc. Natl. Acad. Sci. USA 104, Buff, E., Carmena, A., Gisselbrecht, S., Jiménez, F., and Michelson, A.M. (1998). Signalling by the Drosophila epidermal growth factor receptor is required for the specification and diversification of embryonic muscle progenitors. Development 125, Carmena, A., Gisselbrecht, S., Harrison, J., Jiménez, F., and Michelson, A.M. (1998). Combinatorial signaling codes for the progressive

9 Current Biology Vol 21 No determination of cell fates in the Drosophila embryonic mesoderm. Genes Dev. 12, Halfon, M.S., Carmena, A., Gisselbrecht, S., Sackerson, C.M., Jiménez, F., Baylies, M.K., and Michelson, A.M. (2000). Ras pathway specificity is determined by the integration of multiple signal-activated and tissuerestricted transcription factors. Cell 103, Liu, J., Qian, L., Wessells, R.J., Bidet, Y., Jagla, K., and Bodmer, R. (2006). Hedgehog and RAS pathways cooperate in the anterior-posterior specification and positioning of cardiac progenitor cells. Dev. Biol. 290, Gabay, L., Scholz, H., Golembo, M., Klaes, A., Shilo, B.Z., and Klämbt, C. (1996). EGF receptor signaling induces pointed P1 transcription and inactivates Yan protein in the Drosophila embryonic ventral ectoderm. Development 122, Schweitzer, R., Howes, R., Smith, R., Shilo, B.Z., and Freeman, M. (1995). Inhibition of Drosophila EGF receptor activation by the secreted protein Argos. Nature 376, Kim, S.H., Kwon, H.B., Kim, Y.S., Ryu, J.H., Kim, K.S., Ahn, Y., Lee, W.J., and Choi, K.Y. (2002). Isolation and characterization of a Drosophila homologue of mitogen-activated protein kinase phosphatase-3 which has a high substrate specificity towards extracellular-signal-regulated kinase. Biochem. J. 361, Ajuria, L., Nieva, C., Winkler, C., Kuo, D., Samper, N., Andreu, M.J., Helman, A., Gonzalez-Crespo, S., Paroush, Z., Courey, A.J., and Jimenez, G. (2011). Capicua DNA-binding sites are general response elements for RTK signaling in Drosophila. Development 138, Kim, I.O., Kim, I.C., Kim, S., Kwon, Y.K., Han, P.L., Jeon, S.H., and Kim, S.H. (2005). CNS midline cells contribute to maintenance of the initial dorsoventral patterning of the Drosophila ventral neuroectoderm. J. Neurobiol. 62, Halfon, M.S., Gisselbrecht, S., Lu, J., Estrada, B., Keshishian, H., and Michelson, A.M. (2002). New fluorescent protein reporters for use with the Drosophila Gal4 expression system and for vital detection of balancer chromosomes. Genesis 34, Prober, D.A., and Edgar, B.A. (2002). Interactions between Ras1, dmyc, and dpi3k signaling in the developing Drosophila wing. Genes Dev. 16, Paroush, Z., Wainwright, S.M., and Ish-Horowicz, D. (1997). Torso signalling regulates terminal patterning in Drosophila by antagonising Groucho-mediated repression. Development 124, Bianchi-Frias, D., Orian, A., Delrow, J.J., Vazquez, J., Rosales-Nieves, A.E., and Parkhurst, S.M. (2004). Hairy transcriptional repression targets and cofactor recruitment in Drosophila. PLoS Biol. 2, E Jiménez, G., Guichet, A., Ephrussi, A., and Casanova, J. (2000). Relief of gene repression by torso RTK signaling: Role of capicua in Drosophila terminal and dorsoventral patterning. Genes Dev. 14, Roch, F., Jiménez, G., and Casanova, J. (2002). EGFR signalling inhibits Capicua-dependent repression during specification of Drosophila wing veins. Development 129, Astigarraga, S., Grossman, R., Díaz-Delfín, J., Caelles, C., Paroush, Z., and Jiménez, G. (2007). A MAPK docking site is critical for downregulation of Capicua by Torso and EGFR RTK signaling. EMBO J. 26, Von Ohlen, T.L., Harvey, C., and Panda, M. (2007). Identification of an upstream regulatory element reveals a novel requirement for Ind activity in maintaining ind expression. Mech. Dev. 124, Coppey, M., Boettiger, A.N., Berezhkovskii, A.M., and Shvartsman, S.Y. (2008). Nuclear trapping shapes the terminal gradient in the Drosophila embryo. Curr. Biol. 18, Chou, T.B., Noll, E., and Perrimon, N. (1993). Autosomal P[ovoD1] dominant female-sterile insertions in Drosophila and their use in generating germ-line chimeras. Development 119, Helman, A., and Paroush, Z. (2010). Detection of RTK pathway activation in Drosophila using anti-dperk immunofluorescence staining. Methods Mol. Biol. 661,