Abstract
Regulation of epidermal growth factor receptor (EGFR) signaling requires the concerted action of both positive and negative factors. While the existence of numerous molecules that stimulate EGFR activity has been well documented, direct biological inhibitors appear to be more limited in number and phylogenetic distribution. Kekkon1 (Kek1) represents one such inhibitor. Kek1 was initially identified in Drosophila melanogaster and appears to be absent from vertebrates and the invertebrate Caenorhabditis. To further investigate Kek1's function and evolution, we identified kek1 orthologs within dipterans. In D. melanogaster, kek1 is a transcriptional target of EGFR signaling during oogenesis, where it acts to attenuate receptor activity through an inhibitory feedback loop. The extracellular and transmembrane portion of Kek1 is sufficient for its inhibitory activity in D. melanogaster. Consistent with conservation of its role in EGFR signaling, interspecies comparisons indicate a high degree of identity throughout these regions. During formation of the dorsal-ventral axis Kek1 is expressed in dorsal follicle cells in a pattern that reflects the profile of receptor activation. D. virilis Kek1 (DvKek1) is also expressed dynamically in the dorsal follicle cells, supporting a conserved role in EGFR signaling. Confirming this, biochemical and transgenic assays indicate that DvKek1 is functionally interchangeable with DmKek1. Strikingly, we find that the cytoplasmic domain contains a region with the highest degree of conservation, which we have implicated in EGFR inhibition and dubbed the Kek tail (KT) box.
SIGNAL transduction via receptor tyrosine kinases (RTKs) constitutes one of the major modes of cellular communication in metazoans (Schlessinger 2000). One of the first RTKs to be identified was the epidermal growth factor receptor (EGFR) and it represents, arguably, the most extensively studied RTK (Schlessinger 2000). The EGFR family is highly conserved and representatives have been identified in invertebrates, such as Drosophila melanogaster and Caenorhabditis elegans, as well as in vertebrates (Wadsworthet al. 1985; Schejteret al. 1986; Aroianet al. 1990). Invertebrates appear to have a single EGF receptor, while four receptors, EGFR (or ErbB1), ErbB2, ErbB3, and ErbB4, exist within vertebrates (Stein and Staros 2000). Modulation of receptor activity is essential to normal development and aberrant regulation has been linked to oncogenic situations (Blume-Jensen and Hunter 2001). The classical notion of EGFR signaling invokes receptor dimerization and activation in a ligand-dependent fashion (Schlessinger 2000). However, numerous mechanisms, both stimulatory and inhibitory, exist in vivo to govern the levels of receptor activity. For example, dimerization, coupled with the expansion of family members in vertebrates, allows for the formation of distinct heterodimeric complexes, thereby diversifying signaling outputs (Alroy and Yarden 1997). Diversity also exists at the ligand level, as numerous ligands have been identified—consistent with the concept of combinatorial signaling activities. With one exception, all of these ligands function in a positive fashion. The exception, Argos (Aos), acts in an inhibitory fashion and has not been described outside the dipteran lineage (Schweitzeret al. 1995; Howeset al. 1998). In addition to Aos, another inhibitor of EGFR, Kekkon1 (Kek1) was identified in D. melanogaster (Musacchio and Perrimon 1996; Ghiglioneet al. 1999). One possibility is that in lieu of the genomic duplications in vertebrates, which expanded the receptor family, inhibitors have arisen within other lineages to provide additional signaling diversity (Perrimon and Duffy 1998).
Such diversity in output is crucial to the developmental role of the EGFR family, which functions in numerous cellular events including determination, proliferation, migration, survival, and differentiation. In D. melanogaster, EGFR's role in axis patterning has been well characterized and provides a useful system in which to investigate the regulation of receptor activity (Riechmann and Ephrussi 2001). During oogenesis, communication between the germline-derived oocyte and its overlying somatically derived epithelium or follicle cells orients axial polarity (Schüpbach 1987; Gonzalez-Reyeset al. 1995; Rothet al. 1995; Wasserman and Freeman 1998; Riechmann and Ephrussi 2001). Central to this process is the establishment of follicle cell fates by EGFR in response to a germline signal from the ligand Gurken (Grk; Neuman-Silberberg and Schüpbach 1993). Subsequently, the follicle cells are responsible for the generation and morphology of the chorion, which manifests the underlying polarity of the oocyte (Spradling 1993). During the latter stages of oogenesis, EGFR is stimulated in the follicle cells by Grk secretion from the oocyte (Neuman-Silberberg and Schüpbach 1993). This initiates a suite of transcriptional responses and autoregulatory loops responsible for directing dorsal patterning. Interestingly, two of the transcriptional targets during this process, kek1 and aos, function as inhibitors of EGFR signaling (Schweitzeret al. 1995; Wasserman and Freeman 1998; Ghiglioneet al. 1999). kek1 is expressed in a dorsal anterior gradient within the follicle cells, while aos is expressed along the dorsal midline (Musacchio and Perrimon 1996; Wasserman and Freeman 1998). In contrast, other transcriptional targets (e.g., rhomboid) function in a stimulatory manner, revealing the existence of tiers of feedback regulation (Ruohola-Bakeret al. 1993). Through this complex network of positive and negative feedback mechanisms, EGFR signaling is refined to establish dorsal follicular fates.
Within this autoregulatory network, kek1 functions to attenuate receptor activity. Supporting this, loss-of-function mutations in kek1 suppress egfr mutations in a dose-dependent fashion and display phenotypes indicative of an increase in receptor activity (Ghiglioneet al. 1999). In addition, Kek1 misexpression throughout development results in phenotypes similar to those of loss-of-function mutations in egfr (Ghiglioneet al. 1999). While analysis of Kek1 in oogenesis has provided insight into a specific function, its discovery was based on expression in the embryonic nervous system, where it may function in axonal pathfinding (Musacchio and Perrimon 1996; Speicheret al. 1998). Moreover, the related gene kekkon2 (kek2) was recovered on the basis of its similarity to kek1 (Musacchio and Perrimon 1996). Kek2 exhibits the same arrangement of structural domains and a similar expression profile in the embryonic nervous system (Musacchio and Perrimon 1996). The completion of the D. melanogaster genome has revealed that Kek1 is the founding member of a family of six Drosophila proteins sharing a related extracellular structure (Figure 1; Musacchio and Perrimon 1996; Adamset al. 2000). Each family member contains leucine-rich repeats (LRRs), an amino (N) and carboxy (C) cysteinerich region (N and C cysteine caps) flanking the LRRs, and a single C2-type immunoglobulin (Ig) domain. The identification of a Kek family of molecules within D. melanogaster raises a number of questions concerning the evolutionary history of Kek1 and its role as an inhibitor of EGFR signaling. To address these questions and better define the role of Kek1 in EGFR signaling we undertook an evolutionary analysis of Kek1. Here we describe the identification and characterization of Kek1 orthologs from D. virilis, D. pseudoobscura, and Anopheles gambiae. We report that the expression and function of DvKek1 supports conservation of its role as an inhibitor of EGFR signaling. Finally, interspecies comparisons reveal a region of unexpected conservation in the carboxy-terminal tail, which we have termed the Kek tail (KT) box.
MATERIALS AND METHODS
Molecular techniques: Standard molecular techniques were followed throughout the course of this work (Sambrooket al. 1989). To obtain Dvkek1 a D. virilis genomic phage library (kindly provided by J. Tamkun) was screened with a digoxygenin-labeled probe to kek1. Positive phage clones were carried through secondary and tertiary screens and four positive clones (C1-2-2, C2-1-1, C3-4-1, and D1-1-1) representing Dvkek1 were selected for subsequent analysis. Using a combination of restriction mapping, PCR, and sequencing, a general map of the genomic region containing the kek1 gene in D. virilis was generated. The region corresponding to the entire predicted open reading frame (ORF) was then sequenced from phage clone C1-2-2 using cycle sequencing according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). Sequence was extended at the 5′ end for ∼20 bp until the presence of a CA minisatellite region was detected. Generation of a chimeric Dvkek1-green fluorescence protein (GFP) P{UAS-Dvkek1-gfp} construct was undertaken using Gateway cloning technology (Invitrogen, Carlsbad, CA). Briefly, DvKek1 was flanked with AttB sites by PCR and recombined in frame into a P{UAS-GFP} destination vector (W. Wang and J. B. Duffy, unpublished data). To construct P{UAS-kek1ΔIC-gfp} the extracellular and transmembrane portion of Kek1 was amplified and cloned into EGFPN1 using a 5′ kek1 primer flanked with an EcoRI site and a 3′ kek1 primer flanked with a KpnI site. This kek1-GFP fusion was then shuttled into pUAST using EcoRI and XbaI. P{UAS-kek1ΔIC+T-gfp} was cloned using primers encoding the KT box flanked by a 5′ EcoRI and 3′ KpnI sites and fused in frame to gfp flanked by 5′ KpnI and 3′ XbaI sites. The N-terminal portion of Kek1 was amplified by PCR and nondirectionally cloned, taking advantage of an EcoRI site present at the junction with the KT box. Agkek1 was isolated by using BLAST to identify kek1-related sequences from raw sequence reads deposited into GenBank by the A. gambiae genome project (Holtet al. 2002). These sequences were subsequently organized into contigs using BLAST and Sequencher 4.1 to identify overlapping clones and generate a continuous ORF. Dpkek1 was identified in a similar manner using the sequences available from the Human Genome Sequencing Center at Baylor College of Medicine. The programs ProtScale and SignalP were utilized to analyze the presence of signal peptides and transmembrane domains. Hydropathy plots were performed using Kyte-Doolittle values for hydrophobicity. Protein alignments were performed with ClustalW and manually edited. For in situ hybridizations, the templates NB1 (for Dmkek1) and the coding region for Dvkek1 were used to generate digoxygenin-labeled RNA probe and hybridizations were carried out according to Klingler and Gergen (1993), with minor modifications and 65° as the hybridization temperature (Musacchio and Perrimon 1996).
Drosophila genetics: D. melanogaster transgenics were generated by co-injecting the P{UAS-Dvkek1-gfp} construct with the transposase-encoding plasmid pUChsΠΔ2-3 at a 4:1 ratio into w1118 embryos. Transformant lines were mapped with w-; Sp/ CyO; Sb/Tm6, Hu and subsequently balanced. Misexpression analysis was performed by mating transgenic lines to the follicle cell driver P{GawB}CY2 (CY2-GAL4)(Queenanet al. 1997).
Phenotypic analysis: Eggs from adult females raised at 27° were collected, mounted in Lacto-Hoyers, and cleared at 60° for 48 hr. Images were captured under dark-field illumination on a Zeiss Axiophot microscope. For GFP fluorescence imaging, ovaries from the same females were dissected in PBS, fixed in 3.7% formaldehyde for 10 min, washed four times in PBT, and mounted in 70% glycerol/PBS with SlowFade (Molecular Probes, Eugene, OR). Images were obtained with a Leica TCS SP confocal microscope.
Cell culture and co-immunoprecipitations: Drosophila S3 cells were grown, maintained, and transfected by electroporation as described in Cherbas and Cherbas (1998). Co-immunoprecipitations were performed as follows. Briefly, S3 cells were transiently cotransfected with GFP-tagged constructs, P{UAS-egfr1} or P{UAS-egfr2}, and mt-GAL4 and induced with 1 mm CuSO4 (Klueget al. 2002). GFP-tagged proteins were immunoprecipitated from lysed cells with anti-GFP (CLONTECH, Palo Alto, CA), coupled to protein A Sepharose beads (Amersham-Pharmacia, Piscataway, NJ), and washed thoroughly. Samples were subjected to SDS-PAGE and transferred to nitrocellulose membranes (Amersham-Pharmacia). Blots were probed with mouse anti-GFP (CLONTECH) at 1:1000 and rabbit anti-EGFR (kindly provided by Nick Baker) at 1:5000 (Lesokhinet al. 1999).
RESULTS
Identification of Kek1 orthologs within the Diptera: In D. melanogaster, kek1 is a member of a multigene family and represents a nonvital function (Figure 1; Musacchio and Perrimon 1996). The best-characterized contribution of Kek1 to development is its ability to attenuate EGFR signaling (Ghiglioneet al. 1999). Current evidence suggests that this function is likely to be unique to Kek1 among the six Kek family members identified in D. melanogaster (Alvaradoet al. 2004b). Thus, one possibility is that Kek1's role in EGFR signaling is not representative of a conserved or ancestral function. To gain insight into this possibility we undertook the identification of kek1 orthologs. We searched for kek1 in a related drosophilid, D. virilis, with an estimated divergence time from D. melanogaster of ∼40–65 million years ago (MYA; Beverly and Wilson 1984; Russoet al. 1995). To identify Dvkek1, a D. virilis genomic phage library was screened under moderately stringent conditions with a probe to the Dmkek1 coding region. From this screen four kek1-positive phage clones were selected for further characterization. Analysis of the D. virilis genomic kek1 sequence predicts an uninterrupted ORF of 910 amino acids (aa) for the DvKek1 protein, as compared to 880 aa for DmKek1. Subsequently, we were able to identify kek1 orthologs in D. pseudoobscura and A. gambiae (Holtet al. 2002). Analysis of the kek1 genomic sequences in these species predicts uninterrupted ORFs of 886 aa and 769 aa for DpKek1 and AgKek1, respectively. Interspecific comparisons indicate that these molecules represent orthologs of kek1 and its overall structure as a single-pass transmembrane molecule has been well conserved (Figure 2). In its extracellular region DmKek1 contains N and C cysteine caps (N and C flanks) flanking a set of LRRs, followed by a single Ig domain (Figures 2,3,4). These features and their relative orientation are all conserved in DvKek1, DpKek1, and AgKek1 (Figures 2,3,4). In agreement with their identification as Kek1 orthologs, these molecules display >60% identity with DmKek1 throughout this region, but <40% identity when compared with other Kek family members. In contrast, the N-terminal insert, a feature distinguishing Kek1 from other members of the Kek family in D. melanogaster, is not well conserved (Musacchio and Perrimon 1996). In DmKek1, the insert lies between the signal peptide and the N cysteine cap, but is absent from Ag-Kek1 and displays only two short stretches of identity with DvKek1 and DpKek1 (Figures 3 and 4).
Sequence conservation among Kek1 orthologs: Conservation of the extracellular cysteine caps and LRRs: Although DmKek1 was first characterized as a nonvital locus with no overt phenotypes, subsequent studies have ascribed two functional roles. In oogenesis Kek1 functions to attenuate EGFR signaling during axial patterning, while in neuronal development it appears to function in axonal guidance (Speicheret al. 1998; Ghiglioneet al. 1999). Kek1's function during the former process requires the extracellular and transmembrane portion of the molecule and is mediated by a direct interaction with the EGFR (Ghiglioneet al. 1999). This portion of Kek1 contains seven LRRs and it has recently been revealed that they are essential to Kek1's interaction with the receptor, with the second LRR appearing to play a particularly crucial role (Ghiglioneet al. 2003; Alvaradoet al. 2004a, accompanying article). LRRs have been described as a tandem array of ∼24 amino acid repeats, which is often flanked by cysteine-rich caps. LRRs represent the second most prevalent repeat within the Drosophila proteome, where they are present in a diverse set of secreted, membrane-bound, and cytoplasmic proteins (Pruesset al. 2003). Structurally, LRRs are believed to mediate protein-protein interactions through a characteristic horseshoeshaped structure (Kobe and Deisenhofer 1994, 1995; Kobe and Kajava 2001). Throughout the N and C flanks and LRRs, a high degree of conservation is observed among all four Kek1 orthologs, exhibiting 66% amino acid identity overall (Figure 3; Table 1). Conservation in Kek1 orthologs is apparent in the organization of these motifs, as well as of residues known to impart key structural information. For example, the sequence Lx2LxLx2N/C is believed to represent the smallest element of a full LRR that may be essential to impart the characteristic horseshoe structure of LRR proteins (Kobe and Kajava 2001). In all four species the first LRR of Kek1 is missing the initial leucine of this element, suggesting the first LRR represents only a partial repeat. The subsequent six repeats contain this element in full, although the last (seventh) repeat also appears to represent an abbreviated version of a complete LRR.
—Diagrammatic representation of Kek family members in D. melanogaster. Kek family members are single transmembrane pass proteins that share sequence and structural identity in their extracellular domains. This family is defined by the presence in the extracellular domain of seven leucine-rich repeats (LRRs), flanked by N- and C-terminal cysteine-rich caps (N and C flanks), and followed by a single immunoglobulin (Ig) domain. In contrast, their cytoplasmic domains share little identity. Two features distinguish Kek1 from other Kek proteins. The first is the presence of a small insert between the signal peptide and the N flank. This insert is partially conserved in other Kek1 orthologs. The second is a highly conserved region in the carboxy tail of all Kek1 orthologs.
Conservation of the Ig domain: A single C2-type Ig domain of ∼100 amino acids is present in all Kek1 orthologs identified, as well as in all Kek family members (Figures 1,2,3,4). Ig domains represent the second most common domain in D. melanogaster and mediate a wide array of protein-protein interactions (Pruesset al. 2003). These modules are structurally defined by seven to nine antiparallel β-sheets, which fold to form a β-sandwich. Although a strict consensus sequence does not exist, a number of sequence elements aid in defining an Ig fold. Depending on the Ig domain subtype, the presence of specific disulfide bonds, aromatic residues, and turns are thought to help nucleate and stabilize the fold into an energetically stable conformation. Thus, specific cysteine, proline, and tryptophan residues are often indicative of the Ig signature (Borket al. 1994; Barclay 1999; Stewardet al. 2002). Within the Ig domain Kek1 orthologs share 63% identity, where cysteine, proline, and tryptophan residues are highly conserved (Figure 3; Table 1). Conservation of the transmembrane domain of Kek1: The transmembrane (tm) region spans 21 residues positioned approximately in the middle of the coding sequence and exhibits 62% identity among all four orthologs (Figure 3; Table 1). Structurally, tm domains are typically defined as α-helical stretches composed of hydrophobic amino acids. Given such a general definition, the high degree of conservation manifested by the Kek1 tm regions is striking. In contrast, the signal sequence, similarly constrained to a hydrophobic nature, is only 14% identical (Figure 3; Table 1). Thus, the tm region is likely to contribute more specifically to Kek1 function, in addition to its role in anchoring Kek1 within the lipid bilayer.
Identification of a novel motif, the intracellular KT box: Previous data suggested that the cytoplasmic domain of Kek1 was not essential for its role in EGFR signaling (Ghiglioneet al. 1999). Thus, it was somewhat surprising to discover that the cytoplasmic tail represented the most highly conserved region of Kek1 (Figure 4). Spanning the last 48 residues of the cytoplasmic domain, the Kek tail (KT box) exhibits 92% identity across the four Kek1 orthologs (Figure 4; Table 1). Within this novel motif, two potential elements exist. The first encompasses the last three amino acids (TDV) of Kek1 and represents a putative type I binding site (S/T-X-I/V/G) for PDZ domains (Figure 4). PDZ domains most commonly function in subcellular trafficking and signaling (Hung and Sheng 2002). The second proposed element within the KT box is the sequence SPDEGY previously noted in Kek1 and Kek2 in D. melanogaster (Musacchio and Perrimon 1996). The SPDEGY element is conserved in all four Kek1 orthologs, as well as Kek2 in both D. melanogaster and A. gambiae (Figure 4; data not shown). Functional conservation of the KT box is supported by recent work, which has uncovered a role for the cytoplasmic domain of Kek1 in subcellular trafficking. Deletion of the cytoplasmic domain results in aberrant Kek1 localization and less efficient inhibition of EGFR signaling in oogenesis (Ghiglioneet al. 2003; Figure 5). Strikingly, we find that addition of a portion of the KT box, including the SPDEGY element, restores full inhibitory capability, thus providing one possible explanation for its high degree of conservation (Figure 5).
—Diagrammatic representation of Kek1 orthologs in D. melanogaster (Dm), D. pseudoobscura (Dp), D. virilis (Dv), and A. gambiae (Ag). Kyte-Doolitle hydropathy plots are depicted on the right (positive numbers depict hydrophobicity). All four Kek1 proteins share a similar distribution of hydrophobic regions, centered about the signal peptide (SP) and a transmembrane domain (TM). AgKek1 is the shortest and most divergent protein, displaying a truncated N-terminal insert and cytoplasmic domain. Remarkably, the KT box is present in AgKek1 and highly conserved with respect to the other Kek1 molecules.
—Protein sequence alignment of the extracellular and transmembrane domains of Kek1 in Diptera. Sequence alignment of Kek1 orthologs in D. melanogaster, D. pseudoobscura, D. virilis, and A. gambiae reveals a high degree of conservation at the amino acid level. This conservation is manifested predominantly throughout the LRRs and the Ig domain. In addition, the transmembrane and juxtamembrane regions display remarkable conservation. In contrast, the N terminus is highly divergent, due in part to the presence of a variable signal sequence. Vertical arrows denote putative cleavage sites for the signal sequence. Solid circles represent consensus residues in the LRRs. + indicates conserved cysteine residues in the Ig domain.
Besides the KT box, conservation within the cytoplasmic domain is limited to small pockets for which no functional attributes have been identified. However, these pockets encompass most of the tyrosine residues within the cytoplasmic domain, as 9 of 11 tyrosines are conserved and may reflect functional significance (Figure 4).
—Protein sequence alignment of the cytoplasmic domains of Kek1 in four dipterans. The cytoplasmic domains of Kek1 orthologs are more divergent overall than the extracellular domains. However, of particular interest is the terminal ∼50 amino acids of all proteins, which exhibit remarkable conservation. In addition, short stretches of identity tend to be centered on tyrosine residues.
Expression of Dvkek1 supports a conserved role in oogenesis: During the latter stages of oogenesis, EGFR functions to pattern the dorsal epithelium. Its role in this process has been well studied and represents an excellent example of the role of feedback loops in regulating signaling outputs and directing pattern formation (Perrimon and Duffy 1998; Wasserman and Freeman 1998; Freeman and Gurdon 2002). Briefly, a gradient of Grk emanating from the oocyte initiates EGFR activation within the follicle cells. EGFR activation then initiates a transcriptional response that leads to an amplification of EGFR activity within the dorsal region through the production of two additional ligands, Spitz and Vein, in the follicle cells. Amplification results in a peak of EGFR activity along the anterior-dorsal midline, which then directs expression of the inhibitory ligand Aos. Aos expression effectively abolishes receptor activity along the midline, thereby splitting the single broad peak of EGFR signaling into two domains, representing the bilateral pattern of dorsal appendages on the chorion (Wasserman and Freeman 1998). Throughout this process, expression of kek1 is under the control of EGFR signaling and provides a measure of feedback inhibition (Ghiglioneet al. 1999). Loss of Kek1 activity results in an overall increase in EGFR signaling leading to an increase in the spacing between the appendages, presumably as a result of an expansion in the aos expression domain (Ghiglioneet al. 1999).
While species in the subgenus Sophophora, including D. melanogaster, have only two chorion appendages, a wide range of patterns is displayed among the drosophilids. For example, D. virilis chorions display four appendages and initial results suggest that EGFR signaling modulates dorsal patterning in this species (Periet al. 1999). We asked if kek1's expression profile and, thus, regulation by EGFR signaling was conserved during oogenesis in D. virilis. During the early stages of dorsalventral (DV) axis formation, kek1 is expressed in a dorsal anterior gradient in both species (Figure 6, A, B, F, and G). During the ensuing stages, specific elements of the pattern diverge, although expression remains dorsally restricted in both species (Figure 6, C–E and H–J). Dmkek1 expression remains primarily in a dorsal anterior gradient, until stage 11 when it begins to resolve into two bilateral stripes (Figure 6E). In contrast, from its initial dorsal anterior gradient, Dvkek1 develops a broad and distinctive domain of expression during later stages. Two small regions of bilateral repression disrupt the broad domain of Dvkek1 expression (Figure 6, H and I). This results in the appearance of two triangular patches along the outer edges of the expression domain and a central dorsal patch of expression flanked on either side by the regions of repression (Figure 6, H and I). This pattern is then resolved into two bilateral domains of expression that appear to extend laterally around the egg chamber (Figure 6J). The dorsally restricted expression of Dvkek1 supports the conservation of its role as a transcriptional target of EGFR signaling.
Sequence conservation among Kek1 orthologs
—A portion of the conserved KT box rescues defects associated with loss of the Kek1 cytoplasmic domain. (A) Graphical representation of full-length Kek1 tagged to GFP (Kek1-GFP), a form of Kek1 lacking the entire cytoplasmic domain (Kek1ΔIC-GFP), and a similar truncation where the last 15 amino acids of the KT box have been added back (Kek1ΔIC+T-GFP; see also C). The KT box is depicted in gray. (B) Misexpression of these constructs in follicle cells with the CY2-GAL4 driver generates ventralized chorions. V1 represents weak ventralization as indicated by appendages that are closer together or fused at the base. V2 represents moderate ventralization indicated by a single thin appendage, whereas V3 represents severely ventralized chorions with only a small patch of appendage material. Deletion of the cytoplasmic domain results in a phenotype (V1-V2) weaker than full-length Kek1-GFP (V3). Addition of the terminal 15 amino acids of Kek1 displays a phenotype comparable to full-length Kek1 (V3). The number of independent transgenic lines tested is in parentheses. (C) Alignment of the KT box in the four Kek1 orthologs. The bar indicates the portion of the KT box added back in Kek1ΔIC+T-GFP.
Functional conservation of DvKek1 as an inhibitor of EGFR: Dorsal expression of kek1 in D. virilis, together with the extensive sequence conservation noted among all Kek1 orthologs, supports but does not directly demonstrate conservation of its feedback role in EGFR signaling. To address this we tested the ability of the DvKek1 to both associate with and inhibit the DmEGFR. In D. melanogaster the utilization of distinct 5′ exons results in the production of two isoforms of the EGFR. These isoforms differ solely at the amino terminus preceding the ligand-binding domain and are capable of binding DmKek1 (Figure 7). Similarly, association of DvKek1 with either isoform of DmEGFR can be detected in S3 cells (Figure 7). To directly test the ability of DvKek1 to inhibit EGFR activity we took advantage of the well-documented role of Kek1 in DV patterning (Ghiglioneet al. 1999). During oogenesis, DmKek1 acts to inhibit EGFR signaling in the dorsal follicle cells (Ghiglioneet al. 1999). This inhibitory activity is easily observed through the effects of Kek1 misexpression in the follicle cells, which results in a loss of dorsal fates and ventralization of the chorion (Ghiglioneet al. 1999). Using the same misexpression system, we assayed the ability of DvKek1 to inhibit DmEGFR signaling in vivo. Misexpression of UAS-Dvkek1 in the follicle cells of D. melanogaster results in ventralization, similar to that observed with DmKek1 (Figure 8, C and D). Consistent with this and its predicted transmembrane structure, DvKek1 exhibits a subcellular localization pattern identical to DmKek1 (Figure 8, A and B). Likewise, misexpression of DvKek1 in other tissues also results in inhibition of EGFR signaling, causing phenotypic effects analogous to those observed with DmKek1 (data not shown).
—kek1 is expressed in a dynamic pattern during oogenesis in D. melanogaster and D. virilis, as shown by in situ hybridization of species-specific kek1 RNA probes in ovaries of D. melanogaster (A–E) and D. virilis (F–J). In A and F, dorsal is up. In all other images, dorsal is facing outward. Anterior is to the left. (A) In stage 10 egg chambers of D. melanogaster, kek1 RNA is present in the dorsal anterior follicle cells overlying the oocyte nucleus. Expression begins as a large patch that includes lateral follicle cells. (B) Stage 10 egg chamber is rotated slightly to show dorsolateral follicle cells. (C) The area of expression becomes more refined throughout stage 10. (D) Eventually it becomes more focused along the dorsal midline. (E) In stage 11 egg chambers, kek1 RNA is limited to two small patches on either side of the dorsal midline. (F) In D. virilis, kek1 RNA is also expressed in the dorsal anterior follicle cells during stage 10. (G) In early stage 10 it is restricted to a small dorsal triangle. (H) This pattern begins to broaden in the dorsolateral follicle cells during stage 10 and repression starts to appear in two bilateral patches. (I) Repression increases in the two bilateral patches (arrows) leaving a center spot of kek1 expression (arrowhead), as well as two outer dorsolateral patches. (J) By stage 11, kek1 expression resolves into two bilateral domains that extend laterally around the egg chamber.
DISCUSSION
Feedback loops constitute an important mechanism for the regulation of EGFR signaling during development and contribute to its role in pattern formation (Freeman and Gurdon 2002). Many of the underlying molecular components of this pathway are conserved, but in a few instances regulatory components with limited phylogenetic distributions have been identified. The antagonistic ligand Aos represents one such example and has been suggested to be an example of the independent coevolution of a receptor and a ligand (Stein and Staros 2000). On the surface, invertebrates, with a single EGF receptor and fewer than five ligands, appear to have more restrictions on their ability to generate equally diverse sets of signaling output when compared to vertebrates, which have four receptors and at least twice as many ligands. However, novel regulatory components such as Aos, whose phylogenetic distribution is more limited, might represent alternative approaches to signal diversification.
Here we add to comparative studies of EGFR signaling through an analysis of Kek1, a transmembrane inhibitor of the Drosophila EGFR (Ghiglioneet al. 1999). Null mutations in kek1 are viable and display only subtle phenotypic effects, consistent with a minor role in EGFR signaling (Musacchio and Perrimon 1996; Ghiglioneet al. 1999). Thus, Kek1's ability to bind and inhibit the receptor might represent a modern co-option event, whereby Kek1 only recently acquired the potential to attenuate receptor activity. To investigate this possibility and to identify conserved Kek1 sequence elements that would serve to inform functional studies, we characterized kek1 from three dipteran species with an estimated evolutionary divergence time of ∼250 MYA (Gaunt and Miles 2002). Initially, we chose to search for an ortholog of kek1 in D. virilis and, during the course of our studies, the genome sequences for A. gambiae and D. pseudoobscura became available (Holtet al. 2002). We were able to identify Kek1 in all three species, supporting an ancient presence for kek1 within the Diptera. Moreover, the functional interchangeability of DvKek1 and DmKek1 in EGFR signaling also suggests that Kek1's ability to inhibit EGFR was present over 40–65 MYA and is not a recent acquisition.
Consistent with the known requirement of the extracellular and transmembrane domains of Kek1 for EGFR binding and inhibition, interspecies comparison indicated a high degree of conservation throughout this region, with the exception of the N-terminal insert. The conservation of the transmembrane domain relative to the signal peptide (62 vs. 14% identity) was particularly striking and supports the notion of an essential function for this region. It will be interesting to determine if this conservation is the result of an EGFR-dependent or -independent function. We have also noted that the transmembrane and juxtamembrane portion of Kek1 displays limited identity with some transmembrane receptor-like kinases from Arabidopsis, also of the LRR superfamily (our unpublished observations). Future functional studies will be required to directly assess the relevance of such conservation and the contribution of this region to the in vivo function of Kek1. Finally, the highest degree of conservation we detected in Kek1 was in the KT box. This striking conservation (92% identity across 48 aa) over 250 MYA argues strongly for an essential role in Kek1 function. Here we demonstrate that this conservation might be due in part to its role in enhancing Kek1's ability to inhibit EGFR signaling. Whether conservation is solely representative of a contribution to Kek1's role in EGFR signaling or of an alternative function, perhaps contributing to Kek1's role in neuronal pathfinding, awaits further analysis.
—DmKek1 and DvKek1 associate with DmEGFR by co-immunoprecipitation. Drosophila S3 cells were cotransfected with EGFR isoforms 1 or 2 in combination with GFP-tagged forms of D. melanogaster Kek1 (DmKek1-GFP) and Kek2 (DmKek2-GFP), as well as D. virilis Kek1 (DvKek1-GFP). GFP-tagged constructs were immunoprecipitated (IP) with anti-GFP and assayed for the presence of EGFR (top) and GFP (middle) by immunoblot (IB). EGFR loading levels were assayed directly from whole-cell lysates (bottom). DmKek1-GFP and DvKek1-GFP associate with DmEGFR1 and -2, whereas DmKek2-GFP does not.
DvKek1 is functionally interchangeable with DmKek1 in assays for EGFR binding and inhibition supporting the notion that Kek1's role in EGFR signaling might contribute to its conservation. However, at least one alternate function for Kek1 has been noted and might also play a role in constraining sequence divergence (Speicheret al. 1998). The expression pattern of Dvkek1 provides additional support for a conserved role for kek1 in EGFR signaling. In D. melanogaster, kek1 expression in the follicle cells is regulated by EGFR signaling and appears in a dorsal anterior gradient. In D. virilis, analysis of grk and activated mitogen-activated protein kinase (MAPK) expression suggest that EGFR signaling specifies dorsal fates in D. virilis (Periet al. 1999; Nakamura and Matsuno 2003). Consistent with this, we observe dorsal expression of kek1 in D. virilis in stage 8 onward, although the spatial and temporal dynamics of the expression pattern within the dorsal epithelium are quite distinct from that observed in D. melanogaster. In part, this likely reflects the differences noted in EGFR signaling using antisera to activated MAPK and further supports the contention that the dynamics of EGFR signaling during dorsal appendage patterning have evolved in different ways in these two species (Periet al. 1999; Nakamura and Matsuno 2003). It will be interesting to identify the molecular and genetic bases for such differences and to determine whether they represent changes in elements of the numerous feedback loops that act on EGFR signaling in D. melanogaster. Alternatively, such diversity might reflect the differential contributions of other signaling pathways involved in appendage patterning, such as the transforming growth factor-β (TGF-β) or decapentaplegic pathway (Deng and Bownes 1997; Peri and Roth 2000; James and Berg 2003). Regardless, it remains to be determined whether such differences in kek1 expression represent changes in cis-or trans-acting elements and whether they contribute to the morphological differences in appendage patterning between the two species. Such answers will require further studies involving the development of functional tools for analysis of EGFR components in other drosophilids, such as D. virilis.
—DvKek1 inhibits DmEGFR in vivo. DmKek1 and DvKek1 were tagged with GFP and misexpressed during D. melanogaster oogenesis with the follicle cell driver CY2-GAL4. (A and B) Both GFP-tagged proteins are present at the apical surface of follicle cells during stage 10. (C and D) Misexpression of DmKek1-GFP and DvKek1-GFP results in strongly ventralized chorions, consistent with Kek1-mediated EGFR inhibition.
Acknowledgments
We thank J. Tamkun, N. Baker, and T. Schüpbach for reagents, W. Wang for construction of the P{UAS-GFP} destination vector, T. Kaufman for advice and equipment, and B. Boswell in whose lab this work was initiated. This work was supported by a National Institutes of Health Genetics training grant fellowship (GM-07757) to D.A. and C.M.M. and by a National Science Foundation grant (IBN-0131707) to J.B.D.
Footnotes
-
Communicating editor: T. Schüpbach
- Received July 1, 2003.
- Accepted October 10, 2003.
- Copyright © 2004 by the Genetics Society of America