Genetics, Vol. 167, 187-202, May 2004, Copyright © 2004
Bipartite Inhibition of Drosophila Epidermal Growth Factor Receptor by the Extracellular and Transmembrane Domains of Kekkon1
Diego Alvarado1,a,
Amy H. Rice1,a, and
Joseph B. Duffya
a Department of Biology, Indiana University, Bloomington, Indiana 47405
Corresponding author:
Joseph B. Duffy, Indiana University, 1001 E. 3rd St., Bloomington, IN 47405., jduffy{at}bio.indiana.edu (E-mail)
Communicating editor: T. SCHÜPBACH
 | ABSTRACT |
|---|
In Drosophila, signaling by the epidermal growth factor receptor (EGFR) is required for a diverse array of developmental decisions. Essential to these decisions is the precise regulation of the receptor's activity by both stimulatory and inhibitory molecules. To better understand the regulation of EGFR activity we investigated inhibition of EGFR by the transmembrane protein Kekkon1 (Kek1). Kek1 encodes a molecule containing leucine-rich repeats (LRR) and an immunoglobulin (Ig) domain and is the founding member of the Drosophila Kekkon family. Here we demonstrate with a series of Kek1-Kek2 chimeras that while the LRRs suffice for EGFR binding, inhibition in vivo requires the Kek1 juxta/transmembrane region. We demonstrate directly, and using a series of Kek1-EGFR chimeras, that Kek1 is not a phosphorylation substrate for the receptor in vivo. In addition, we show that EGFR inhibition is unique to Kek1 among Kek family members and that this function is not ligand or tissue specific. Finally, we have identified a unique class of EGFR alleles that specifically disrupt Kek1 binding and inhibition, but preserve receptor activation. Interestingly, these alleles map to domain V of the Drosophila EGFR, a region absent from the vertebrate receptors. Together, our results support a model in which the LRRs of Kek1 in conjunction with its juxta/transmembrane region direct association and inhibition of the Drosophila EGFR through interactions with receptor domain V.
CELLULAR communication by the epidermal growth factor receptor (EGFR) pathway is widely utilized throughout development to specify cellular fates and behaviors (SCHWEITZER and SHILO 1997 
; DOMINGUEZ et al. 1998 ; NILSON and SCHUPBACH 1999 ; VAN BUSKIRK and SCHUPBACH 1999 ). Work in invertebrates and vertebrates has implicated this pathway in numerous developmental processes including axial patterning in Drosophila, vulval development in Caenorhabditis elegans, and cardiac development in vertebrates (AROIAN et al. 1990 ; LEE et al. 1995 ; NILSON and SCHUPBACH 1999 ). In addition, mutation and misregulation of the EGFR family of receptor tyrosine kinases (RTK) is one of the hallmarks of oncogenic transformation and its associated alterations in cellular behavior: immortalization, proliferation, migration, and invasion (STOSCHECK and KING 1986 ; HARARI and YARDEN 2000 ; BLUME-JENSEN and HUNTER 2001 ; HOLBRO et al. 2003 ). With the exception of the orphan receptor ErbB2 and ErbB3, which encodes a catalytically inactive kinase domain, the vertebrate and invertebrate receptors are type I transmembrane proteins composed of an extracellular ligand-binding domain, a transmembrane region, and a cytoplasmic tyrosine kinase domain (YARDEN and ULLRICH 1988 ; ALROY and YARDEN 1997 ; OLAYIOYE et al. 2000 ). Within the extracellular region the vertebrate ErbBs are composed of four domains. Domains I and III (also known as L1 and L2) mediate ligand binding, whereas the cysteine-rich domains II and IV (also named CR1 and CR2) are involved in dimerization and auto-inhibition, respectively (LAX et al. 1988 ; GARRETT et al. 2002 ; OGISO et al. 2002 ; FERGUSON et al. 2003 ). Interestingly, the extracellular domain of Drosophila EGFR contains an additional cysteine-rich domain distal to CR2, herein referred to as domain V (PRICE et al. 1989 ). Distal to the kinase domain, a noncatalytic carboxy-terminal tail (C-tail) contains tyrosines that become phosphorylated in response to ligand binding (YARDEN and SCHLESSINGER 1987 ; ULLRICH and SCHLESSINGER 1990 ). In contrast to the four vertebrate receptors, Drosophila contains only two isoforms, EGFR1 and EGFR2, derived from a single locus, torpedo/egfr (CLIFFORD and SCHUPBACH 1994 ; LESOKHIN et al. 1999 ). These isoforms differ only at their N termini in the signal sequence and a short stretch of flanking amino acids, but are identical through the ligand-binding, transmembrane, and cytoplasmic domains.
The long-standing model for activation of EGFR signaling involves receptor homo- or heterodimerization upon ligand binding (YARDEN and SCHLESSINGER 1987 ; ULLRICH and SCHLESSINGER 1990 ). This allows for transphosphorylation of a specific subset of tyrosine residues in the C-terminal tail to occur. Transphosphorylation initiates the recruitment of distinct phosphotyrosine binding adaptor/effector molecules and triggers activation of a variety of cytoplasmic kinase cascades [e.g., RAS-RAF-mitogen-activated protein kinase (MAPK) cascade; YARDEN and SLIWKOWSKI 2001 ; SCHLESSINGER and LEMMON 2003 ]. However, a wealth of data obtained in the past few years supports a more complex scenario that distinguishes the EGFR family among RTKs and indicates that long-held notions about the mechanism of EGFR activation need to be reexamined (SCHLESSINGER 2000 , SCHLESSINGER 2002 ; BURGESS et al. 2003 ). For example, the recently solved crystal structures of apo- and ligand-bound soluble forms of EGFR reveal striking differences in the mechanism of ligand binding with respect to other RTKs (GARRETT et al. 2002 , GARRETT et al. 2003 ; OGISO et al. 2002 ; FERGUSON et al. 2003 ). Ligand binding may induce a conformational shift involving rotation about the juxta/transmembrane domain of the receptor, thereby resulting in the access of substrate to the kinase domain (MORIKI et al. 2001 ). Perhaps most surprisingly, structural and functional studies indicate that the EGFR kinase domain is likely catalytically active in a monomeric state (GOTOH et al. 1992 ; STAMOS et al. 2002 ). This is in stark contrast to other members of the RTK family (e.g., insulin receptor), which require tyrosine phosphorylation in the activation loop (A-loop) of the kinase domain for full catalytic activity (HUBBARD and TILL 2000 ). While kinase activity is essential to signaling, receptor dimerization, ligand binding/dissociation, subcellular localization, trafficking, and effector transduction all play important roles in regulating signaling strength (OLAYIOYE et al. 2000 ). The integration of such regulatory mechanisms is therefore essential to specify the appropriate level of EGFR signaling within a given developmental context.
Although the relevance of positive effectors of EGFR signaling has long been appreciated, only in the past few years has the importance of inhibitory molecules in regulating signaling strength and duration come to the forefront (FREEMAN 1998 ). Molecules such as Kekkon1 (Kek1), Argos (Aos), D-Cbl, and Sprouty (Spry) inhibit EGFR signaling, resulting in a refinement of signaling output (SCHWEITZER et al. 1995 ; HIME et al. 1997 ; WASSERMAN and FREEMAN 1998 ; CASCI et al. 1999 ; GHIGLIONE et al. 1999 , GHIGLIONE et al. 2003 ; KRAMER et al. 1999 ; PAI et al. 2000 ). Furthermore, all these inhibitors exert their effects via different mechanisms, resulting in distinct effects on receptor signaling. Characterizing the mechanisms by which all EGFR regulators work will be essential to understanding the balanced interplay between these molecules and their contribution to EGFR-mediated developmental decisions.
The specification of dorsal-ventral (DV) polarity in Drosophila provides one well-characterized example of the interplay between positive and negative effectors of EGFR signaling. During the latter stages of oogenesis, receptor activity is modulated by positive and negative feedback loops to pattern the DV axis (STEVENS 1998 ; VAN BUSKIRK and SCHUPBACH 1999 ). Throughout dorsal fate specification, expression of the transmembrane molecule Kek1 is induced in a graded fashion by receptor activity (MUSACCHIO and PERRIMON 1996 ; GHIGLIONE et al. 1999 ). kek1 was initially identified in an enhancer trap screen for genes involved in the development of the Drosophila nervous system. Subsequent molecular and genomic analyses indicated that Kek1 is the founding member of a family of six related transmembrane proteins in Drosophila that contain leucine-rich repeats (LRRs) and a single C2-type immunoglobulin (Ig) domain in their extracellular regions (MUSACCHIO and PERRIMON 1996 ; ADAMS et al. 2000 ). Three lines of evidence support a role for Kek1 in attenuating EGFR activity during DV patterning (GHIGLIONE et al. 1999 ). First, loss of Kek1 activity results in wider spacing between the appendages. This phenotype is dramatically different from that observed in aos mutants, which produce a single, wide appendage, suggesting that Kek1 and Aos utilize distinct mechanisms to inhibit EGFR. Second, reduced EGFR activity can be suppressed by the simultaneous elimination of Kek1 activity. Last, misexpression of Kek1 in follicle cells results in inhibition of EGFR signaling, observed phenotypically as ventralization of the chorion. More recently, loss-of-function studies in the eye, as well as misexpression studies in both the eye and the wing, indicate that Kek1 is likely to function as a general inhibitor of EGFR throughout development (GHIGLIONE et al. 2003 ; ALVARADO et al. 2004 ). In addition, deletion analyses and mutagenesis experiments have demonstrated that the LRRs of Kek1 play an essential role in receptor inhibition (GHIGLIONE et al. 2003 ; ALVARADO et al. 2004 ).
Despite recent progress in elucidating the mechanism of Kek1 function, a number of questions remain unanswered. Currently, it is unclear what elements of Kek1 suffice for inhibition, whether other Kek molecules function redundantly with Kek1, and if Kek1 is a substrate for EGFR-mediated phosphorylation. Here we present evidence addressing these questions. Importantly, we show that while the Kek1 LRRs promote binding, the juxta/transmembrane region of Kek1 actively contributes to receptor inhibition. We demonstrate that EGFR inhibition is unique to Kek1 among the Kek family members, that Kek1 is not a substrate for EGFR phosphorylation, and confirm that Kek1 inhibits EGFR in multiple developmental contexts. Finally, we have isolated a unique class of EGFR alleles that specifically disrupt its association with Kek1.
| MATERIALS AND METHODS |
|---|
Drosophila genetics:
Flies were raised at 27° on standard media. The following stocks were used: follicle cell drivers P{GawB}CY2 (QUEENAN et al. 1997 ) and P{GawB}T155 (FREEMAN 1996 ); eye driver P{GAL4-ninaE.GMR} (FREEMAN 1996 ); P{w + mC = Act5C-GAL4}; P{UAS-Egfr.DN.B}29-77-1(FREEMAN 1996 ); P{UAS-grk
TC}, grkHK36 (NEUMAN-SILBERBERG and SCHUPBACH 1993 ); egfr2X51 (WIESCHAUS et al. 1984 ); egfrQY1 (CLIFFORD and SCHUPBACH 1994 ); egfrCO (CLIFFORD and SCHUPBACH 1989 ); Df(2R)egfr3F18; kek1RA5, and kek1RM2 (MUSACCHIO and PERRIMON 1996 ). Transgenic lines were generated by coinjecting each pUAST construct with transposase (pUChs 
2-3) at a 4:1 ratio into w1118 embryos.
All pUAST constructs were misexpressed in follicle cells using CY2-GAL4 or T155-GAL4 and in the developing eye with GMR-GAL4. For chorion preparations, eggs were collected, washed, and cleared in lacto-Hoyer's solution for 48 hr at 60°. Images were captured under darkfield on a Zeiss Axiophot microscope. Ovaries from females expressing green fluorescent protein (GFP) constructs used in egglays were dissected in PBS and fixed for 10 min in 3.7% formaldehyde/PBS, washed three times in PBT (0.1% Tween-20), and brought to volume in 70% glycerol/PBS with SlowFade (Molecular Probes, Eugene, OR). GFP fluorescence images were captured with a Leica TCS SP confocal microscope. Wings were dehydrated in 100% ethanol and mounted in polyvinyl lactophenol. Scanning electron microscopy was performed as in KIMMEL et al. 1990 .
To identify suppressors of Kek1 misexpression in the eye, P{GAL4-ninaE.GMR}, P{UAS-kek1}/CyO recombinants were generated by standard methods from stocks containing individual P elements. w;iso2;iso3 males were mutagenized with 25 mM EMS according to previously described methods (ASHBURNER 1989 ). Mutagenized males were crossed to P{GAL4-ninaE.GMR}, P{UAS-kek1}/CyO females and maintained at 27°. A total of 110,000 chromosomes from straight-winged F1 males and females were screened for suppression of the P{GAL4-ninaE.GMR}, P{UAS-kek1}/CyO phenotype. Five straight-winged strong suppressors were isolated independently and crossed to w1118; the segregation of the suppressor away from the P{GAL4-ninaE.GMR}, P{UAS-kek1} chromosome in the F2's demonstrated that all five suppressors were on the second chromosome. Stocks were generated by crossing F1 w; SOK/+ males to w; P{GAL4-ninaE.GMR}, P{UAS-kek1}/CyO females; w; SOK/P{GAL4-ninaE.GMR}, P{UAS-kek1} males were then crossed to yw; Sco/Cyo and balanced stocks established. The suppressors were then mapped using standard recombination mapping techniques to genetic position 97–99 on 2R with the following stocks: al1 dpov1 b1 pr1 c1 px1 sp1 and crossveinless-2. The egfr alleles egfrQY1, egfrCO, and Df(2R)egfr3F18 were used for complementation tests.
Molecular cloning and sequence analysis:
keg-based chimeras were subcloned into pUAST in three steps. The kek1 extracellular and transmembrane region was amplified by PCR and subcloned into pUAST using 5' EcoRI and 3' BglII sites (pUAST-k1et). The gfp gene was excised from the pEGFP-N1 vector (CLONTECH, Palo Alto, CA), using 5' KpnI and 3' XbaI restriction sites, and fused in frame to pUAST-k1et (pUAST-k1et-gfp). Next, variants of the cytoplasmic domain of egfr were flanked with 5' BglII and 3' KpnI sites by PCR and incorporated into pUAST-k1et-gfp. Fragments encode the full-length cytoplasmic domain (keg), the kinase region (keCg), and the C-tail (keKg). Point mutations encoding kinase-dead chimeras were generated by PCR-based site-directed mutagenesis utilizing mismatch primers encoding G901R (ke*gg) and K923M (ke*kg) changes in the kinase region. pUAST-mCD8EC-gfp was constructed by substituting the kek1 region in pUAST-keCg with a PCR-based fragment from pUAST-mCD8-gfp (courtesy of Liqun Luo; LEE and LUO 1999 ) encoding extracellular and transmembrane regions of the murine CD8 gene flanked by 5' MfeI and 3' BglII restriction sites.
pUAST-kek1-gfp was made by first subcloning a PCR-based fragment flanked by 5' SpeI and 3' KpnI sites encoding the transmembrane-intracellular regions of kek1 into pUAST-gfp (pUAST-kek1tm-intra-gfp). Next, the extracellular domain of kek1 was amplified by PCR from its cDNA (NB1) and cloned into pUAST-kek1tm-intra-gfp using 5' EcoRI and 3' SpeI sites. pUAST-Kek2-gfp was made by amplifying the full-length gene from the NB7 cDNA and subcloning into a basal pUAST-gfp using 5' EcoRI and 3' KpnI sites. Additional details and maps are available upon request.
Kek1-Kek2 swaps were generated in vitro by overlapping PCR. Purified fragments were cloned into a Gateway-compatible pUAST-GFP destination vector, according to manufacturer's instructions (Invitrogen, Carlsbad, CA).
egfrSOK1–4 were sequenced from genomic DNA isolated from 8–10 hemizygous adult flies [SOK/Df(2R)Pu-D17] with QIAGEN DNeasy columns according to manufacturer's instructions (QIAGEN, Beverly, MA). For egfrSOK5, DNA was isolated from 30–50 homozygous embryos selected from egfrSOK5/CyO, P{Act-GFP} stock. Likewise, genomic DNA was also isolated from w;iso2;iso3 adults. egfr exons were amplified individually by PCR, using primers specific to noncoding sequence flanking each exon, purified using a gel purification kit (QIAGEN), and sequenced using cycle sequencing according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). Primer sequences are available upon request. Sequence alignments were performed with MAP (http://searchlauncher.bcm.tmc.edu/multi-align/multi-align.html), using the extracellular and transmembrane portion of the receptor for each of the indicated species. The start codon corresponds to the first residue with respect to numbering for the alignments.
pUAST-egfrSOK constructs were generated using PCR-based site-directed mutagenesis with mismatch primers encoding changes R738Q (egfrSOK1–3) and E718K (egfrSOK4). PCR fragments encoding point mutations and flanked by 5' FseI and 3' PshAI restriction sites were cloned into the corresponding region of the template pUAST-egfr1 (courtesy of Nick Baker). All pUAST-egfrSOK subclones were sequenced to confirm presence of the point mutations. Primer sequences and construct maps are available upon request.
Cell culture and immunoprecipitations:
Drosophila S3 cells were grown and maintained as described in CHERBAS and CHERBAS 1998 . Cells were grown to a density of 
5 x 106 cells/ml and transfected by electroporation. Cells were cotransfected with 5 µg of metallothionein-GAL4 (mt-GAL4; KLUEG et al. 2002 ), a copper-inducible GAL4 driver, and 5 µg of responder DNA and induced with 1 mM CuSO4 for 22 hr. Cells were collected and gently pelleted by centrifugation at 2000 rpm and subsequently lysed in 1 ml of ice-cold Fehon buffer (FEHON et al. 1990 ) containing 1 mM PMSF, 1 µM leupeptin, 1 µM pepstatin A, and 0.3 µM aprotinin. Lysed cells were cleared by centrifugation at 14,000 rpm for 5 min at 4°. Supernatant was brought up to 5 ml in buffer, and antigen was immunoprecipitated with 0.5 µl of rabbit anti-GFP (CLONTECH). Samples were rotated for 2 hr at 4° and subsequently incubated with 150 µl of a 1:5 slurry of protein A Sepharose beads (Amersham-Pharmacia, Piscataway, NJ) in Fehon buffer for 1 hr at 4°. Beads were collected by gentle centrifugation (3000 rpm for 2 min at 4°) and washed five times in Fehon buffer. The last two washes were performed in Fehon buffer lacking detergent. Samples were resuspended in 30 µl of a 3:2 mix of 5x sample buffer and TBS, boiled for 5 min, and loaded in 8% polyacrylamide gels. Transfer to nitrocellulose membranes (Amersham-Pharmacia) was followed by Ponceau staining and subsequently blocked for 1 hr at room temperature (RT) in 5% nonfat dry milk (NFDM) with TBST (100 mM Tris pH 7.5, 150 mM NaCl, 0.1% Tween-20). Membranes were incubated with primary antisera at the following concentrations: rabbit anti-EGFR (courtesy of Nick Baker; LESOKHIN et al. 1999 ) at 1:5000 (2% NFDM in TBST), monoclonal anti-GFP (CLONTECH) at 1:1000 (5% NFDM in TBST), and guinea pig anti-Delta (courtesy of Marc Muskavitch; HUPPERT et al. 1997 ) at 1:5000 (5% NFDM). Incubations were done overnight at 4°, followed by five washes in TBST. Secondary horseradish peroxidase-conjugated goat-anti-rabbit, mouse, and guinea pig antibody incubations (Jackson Immunoresearch, West Grove, PA) were done at 1:20,000 in 5% NFDM for 1 hr at RT, followed by five washes in TBST. Detection was performed by chemiluminescence (West Pico; Pierce, Rockford, IL), according to manufacturer's instructions, utilizing Kodak Biomax MR-1 autoradiography film. Stripping and reprobing was performed according to manufacturer's instructions.
Phosphorylation assays:
Ten ovaries per genotype were dissected in PBS and homogenized in 500 µl of RIPA buffer with protease and phosphatase inhibitors (150 mM NaCl, 100 mM Tris pH 7.4, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM PMSF, 1 µM leupeptin, 1 µM pepstatin A, 0.3 µM aprotinin, 5 mg/ml ß-glycerolphosphate, 1 mM sodium orthovanadate). Ovaries were homogenized and cleared by centrifugation for 10 min. Supernatant was brought up to 1 ml and immunoprecipitations and Western blots were performed as described above. Anti-phosphotyrosine (pY99; Santa Cruz Biotechnologies, Santa Cruz, CA) was used at 1:1000 in 4% BSA. Rabbit anti-EGFR and mouse anti-
tubulin (Accurate Chemical, Westbury, NY) antibodies were used at 1:5000.
| RESULTS |
|---|
Within the Kek family, EGFR inhibition is unique to Kek1:
In its extracellular region Kek1 is composed of an N-terminal insert, a set of seven LRRs flanked by cysteine-rich caps, and a single Ig domain, which together with the transmembrane domain suffice to inhibit EGFR signaling. Within the Drosophila proteome, significant sequence similarity and a similar arrangement of motifs are found in five additional transmembrane proteins that together with Kek1 constitute the Kekkon (Kek) family (MUSACCHIO and PERRIMON 1996 ; DERHEIMER et al. 2004 ). The identification of these structurally related molecules coupled with the subtlety of kek1 null phenotype raised the possibility of functional redundancy among Kek family members. To address this we tested four of the five additional members, Kek2, Kek4, Kek5, and Kek6, for effects on EGFR signaling in vitro and in vivo. Initially, we confirmed that both EGFR isoforms (1 and 2) associate with Kek1 and each other by co-immunoprecipitation (co-IP) experiments from Drosophila S3 cells (Fig 1A and Fig B). Kek1, in addition to binding the receptor, is also able to associate with itself (ALVARADO et al. 2004 ). Indicating specificity to binding, Kek1 does not associate with the transmembrane molecule Delta (Dl; Fig 1B). In contrast to the strong interaction detected between Kek1 and EGFR, Kek2, Kek4, and Kek5 interact either weakly or not at all with EGFR in co-IPs (Fig 1B).
View larger version (57K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1.
EGFR inhibition is not a common feature of Kek family members. (A) Graphical representation of Kek family members and Kek1 variants fused to GFP (circle). Kek family members are single transmembrane proteins consisting of an extracellular domain composed of seven LRRs and a single Ig domain. Their intracellular regions share little similarity. (B) Co-immunoprecipitation experiments from Drosophila S3 cells. GFP-tagged constructs were immunoprecipitated with a polyclonal anti-GFP antibody (IP) and immunoblotted with anti-EGFR (IB; top left), anti-Delta (top right), and a monoclonal anti-GFP antibody as a control (middle). Whole-cell lysates were directly immunoblotted with anti-EGFR and anti-Delta as a loading control (bottom). Kek1GFP associates with both isoforms of EGFR (iso1 and iso2) whereas Kek2GFP, Kek4GFP, and Kek5GFP display minimal or no affinity for EGFR. (C) Chorion phenotypes and subcellular distribution of GFP-tagged constructs misexpressed in follicle cells with CY2-GAL4. Kek1GFP exhibits strongly ventralized chorions, whereas Kek2GFP, Kek4GFP, Kek5GFP, and Kek6GFP do not display ventralized chorions when misexpressed with CY2-GAL4, demonstrating that the extracellular and transmembrane domains of Kek1 are unique in their ability to inhibit EGFR among Kek family members.
|
|
Consistent with the ability of Kek1 to physically associate with the receptor, its misexpression during oogenesis with CY2-GAL4 results in inhibition of EGFR signaling and ventralization phenotypes similar to those observed with loss-of-function alleles of the receptor. To determine if other Kek family members displayed inhibitory effects similar to Kek1 in vivo, we generated UAS lines for Kek2, Kek4, Kek5, and Kek6 and misexpressed them during oogenesis using CY2-GAL4. Misexpression of these Keks in follicle cells has no effect on DV patterning, indicating Kek2, Kek4, Kek5, and Kek6 are not functionally analogous to Kek1 with respect to EGFR inhibition (Fig 1C). Thus, the ability to associate with and inhibit the EGFR is not a common feature shared by Kek family members and is unlikely to provide an explanation for subtlety of the kek1 null phenotype.
Kek1 attenuates receptor signaling in multiple tissues:
Initially, the subtlety of the loss-of-function phenotype for kek1 hindered identification of its role in EGFR signaling. Subsequently, however, a role for kek1 in attenuating EGFR signaling during oogenesis was uncovered in DV patterning, as determined by misexpression analysis and confirmed through loss-of-function studies (GHIGLIONE et al. 1999 ). In addition to its role in DV patterning, the EGFR participates in a multitude of cellular decisions throughout development, where distinct ligands regulate receptor activity in a tissue-specific fashion. It has recently been shown through misexpression experiments that Kek1 can inhibit EGFR signaling in the wing and eye (GHIGLIONE et al. 2003 ; ALVARADO et al. 2004 ). To determine if endogenous kek1 functions more generally throughout development to attenuate receptor activity, we generated flies with reduced egfr activity and asked whether the resulting phenotypes could be rescued by simultaneous removal of kek1. Prior to DV patterning, EGFR activity in the posterior follicle cells results in the establishment of the anterior-posterior (AP) axis. Hypomorphic combinations of egfr result in abnormal AP axis specification and mislocalization of the posterior determinant oskar (osk; ROTH et al. 1995 ; Fig 2D). Simultaneous removal of kek1 activity suppresses this phenotype, allowing the proper establishment of the AP axis (Fig 2F). Likewise, a similar role for kek1 is observed during wing vein specification, which also requires EGFR signaling (DIAZ-BENJUMEA and GARCIA-BELLIDO 1990 ; STURTEVANT et al. 1993 ). Hypomorphic combinations of egfr result in vein loss (Fig 2C) and simultaneous removal of kek1 also suppresses this phenotype (Fig 2E). Consistent with an increase in EGFR activity in a kek1– background, patches of ectopic vein are also present in kek1– wings (Fig 2F). Strong dose-dependent inhibitory effects for endogenous kek1 on EGFR signaling have also been observed in the developing eye (ALVARADO et al. 2004 ). Since EGFR activation within each tissue is initiated by distinct ligands, our results suggest that Kek1's inhibitory activity is neither ligand nor tissue specific, consistent with misexpression analyses (NEUMAN-SILBERBERG and SCHUPBACH 1993 ; ROTH et al. 1995 ; SIMCOX 1997 ; TIO and MOSES 1997 ; GHIGLIONE et al. 1999 , GHIGLIONE et al. 2003 ; GUICHARD et al. 1999 ; ALVARADO et al. 2004 ). However, it is important to note that these results do not exclude the possibility that Kek1 binds all EGFR ligands and might function as a ligand sink. Given that Kek1 can also associate with itself, one possibility is that Kek1 acts as a homodimer and competes with the receptor for ligand binding (ALVARADO et al. 2004 ). Work with Kek1 and the vertebrate receptor has recently provided support against such a model by indicating that Kek1 does not appear to act simply by binding to vertebrate EGF (GHIGLIONE et al. 2003 ). Below we also rule out the ligand sink model in Drosophila.
View larger version (103K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2.
Kek1 effects are not ligand or tissue specific. Loss-of-function combinations of egfr result in loss of the L4 wing vein and oskar mRNA mislocalization. In each situation, reduction or elimination of kek1 rescues the associated egfr mutant phenotype. (A, C, E, and G) Dark-field images of wings from (A) wild-type, (C) egfrQY1/egfrCO, and (E) kek1RA5 egfrQY1/kek1RM2 egfrCO males. As observed in C, loss of the anterior cross-vein (100%, n = 49) and partial loss of the L4 wing vein (69%, n = 49) are evident. In E, restoration of L4 in egfr mutants is observed in response to elimination of kek1 as no gaps were observed in the double mutants (0%, n = 47). (G) kek1 mutants (kek1RM2/kek1RA5) display ectopic wing vein patches. This phenotype is 72.3% penetrant in kek1RM2/kek1RA5 flies (n = 176), whereas in kek1RM2/+ (n = 105) and kek1RA5/+ (n = 69), this phenotype is not observed. (B, D, and F) Nomarski images of oskar mRNA localization in egg chambers from (B) wild-type, (D) egfrQY1/egfrCO, and (F) kek1RA5 egfrQY1/kek1RM2 egfrCO females. As seen in D, mislocalization of osk to the middle of the oocyte is evident. In F, osk localizes properly in egfr mutants in response to elimination of kek1.
|
|
Kek1 acts in association with EGFR in vivo and not as a ligand sink:
To address the potential for Kek1 to function as a competitive inhibitor of EGFR through ligand binding in Drosophila, we took advantage of the knowledge that chimeric proteins have been widely used in Drosophila to study the function of transmembrane receptors (DICKSON et al. 1992 ; REICHMAN-FRIED et al. 1994 ; MURPHY et al. 1995 ; QUEENAN et al. 1997 ; SCHNEPP et al. 1998 ; BOUTROS et al. 2000 ; KELEMAN and DICKSON 2001 ; KELEMAN et al. 2002 ). Since the extracellular and transmembrane portions of Kek1 mediate inhibitory activity, we were able to test the ligand sink model, using a chimeric Kek1-EGFR molecule. The extracellular and transmembrane portions of Kek1 were fused to a portion of the cytoplasmic domain of the EGFR including the kinase domain, but lacking all residues following the kinase domain—the C-tail. In Drosophila, as with the vertebrate receptors, the EGFR C-tail contains a series of tyrosines that recruit adaptor proteins upon phosphorylation and is essential for signaling activity (RAZ et al. 1991 ; CLIFFORD and SCHUPBACH 1994 ). The absence of the C-tail from this chimera prevents it from undergoing autophosphorylation as a homodimer. This Kek1-EGFR chimera was tagged with GFP and termed KECG (Fig 3A).
If Kek1 functions as a competitive inhibitor through homodimerization and ligand binding, the KECG chimera would still function as an inhibitor, since it could bind ligand but would fail to activate EGFR signaling due to the absence of the C-tail. Alternatively, if Kek1 and EGFR function as a heterodimer in vivo, KECG would bind and cross-phosphorylate the endogenous receptor (Fig 3B). In this scenario, misexpression of KECG would result in EGFR activation, rather than inhibition, confirming the presence of a Kek1/EGFR complex in vivo and providing evidence against the ligand sink model. Misexpression of KECG in follicle cells resulted in dorsalized chorions, an EGFR gain-of-function phenotype, strongly suggesting that it is able to activate signaling via a direct interaction with the receptor (Fig 3C). Consistent with this, KECG associates strongly with EGFR1 by co-immunoprecipitation (Fig 3D). Similar stimulatory effects were also observed in the wing and eye (data not shown). If this chimera requires endogenous EGFR for its activity, then reducing access to EGFR should result in suppression of its activity. Consistent with this, hemizygosity for egfr, heterozygosity with receptor alleles lacking the C-tail, or misexpression of a dominant-negative allele of egfr (DNegfr) resulted in partial or complete suppression of the KECG phenotype, respectively, indicating that KECG requires endogenous EGFR to activate signaling (Table 1). In contrast, reduced activity of the EGFR ligand Gurken (Grk) did not suppress the KECG phenotype, indicating that association of KECG with EGFR is ligand independent (Table 1). To demonstrate that the interaction between KECG and EGFR is not directed by the presence of the kinase domain in the chimera, a control chimera encoding the murine Ig-containing transmembrane molecule CD8 was similarly tagged with the EGFR intracellular domain lacking its C-tail and GFP (mCD8ECGFP; Fig 3A). Misexpression of either this molecule or a form of EGFR lacking the extracellular domain (TOP; QUEENAN et al. 1997 ) did not activate signaling in our assay (Fig 3C and data not shown), indicating that the kinase domain of EGFR alone does not trigger signaling. Therefore, the extracellular and transmembrane portions of Kek1 are sufficient to direct association with EGFR in vivo in a ligand-independent fashion. A similar conclusion was also made using a Kek1-EGFR chimera that includes the C-tail of EGFR (GHIGLIONE et al. 2003 ).
Kek1 forms an inhibitory complex with EGFR:
The physical association of Kek1 with the receptor noted in vivo is consistent with a model for inhibition in which receptor dimerization and autophosphorylation are precluded by association with Kek1 (GHIGLIONE et al. 2003 ). Although this provides a simple explanation for the lack of receptor signaling, the status of receptor kinase activity when complexed with Kek1 remains an important and unresolved issue (BURGESS et al. 2003 ). Current data indicate that, unlike most RTKs, the kinase domain of a monomeric EGFR is catalytically active, which is further supported by the observation that the EGFR A-loop does not require tyrosine phosphorylation for catalytic activity (GOTOH et al. 1992 ; STAMOS et al. 2002 ; BURGESS et al. 2003 ). In contrast, most RTKs require tyrosine phosphorylation in the A-loop of the kinase domain for full catalytic activity (GOTOH et al. 1992 ; STAMOS et al. 2002 ). Thus, we were prompted to investigate two possible models for Kek1 inhibition of EGFR. In one model, the receptor kinase domain has access to and phosphorylates Kek1, which then acts as a "dead end" substrate by failing to bind the appropriate downstream adaptors. Alternatively, Kek1 may not be phosphorylated when complexed with EGFR, suggesting that the receptor kinase domain is unable to gain access to Kek1 for trans-phosphorylation. Such a scenario would be consistent with evidence suggesting that rotation of the receptor transmembrane domain is necessary for activation (MORIKI et al. 2001 ). To address these two models we utilized both a biochemical and a genetic approach to examine the status of Kek1 as a phosphorylation substrate in vivo.
First, we assayed for tyrosine phosphorylation of Kek1 under conditions in which its misexpression results in strong inhibition of EGFR signaling. In ovaries from CY2-GAL4, UAS-Kek1-GFP females, no detectable tyrosine phosphorylation of Kek1 is evident, supporting the latter model (Fig 4). To provide additional support for this, we tested a series of chimeric Kek1/EGFR molecules designed to functionally assay the activity of the receptor when complexed with Kek1. If Kek1 is phosphorylated by EGFR and acts as a dead end substrate, then exchanging the cytoplasmic domain of Kek1 with a phosphorylation and signaling-competent substrate (EGFR cytoplasmic domain) would enable recruitment of the appropriate adaptor proteins, thereby converting Kek1 to an activator (Fig 5A and Fig B). In contrast, if Kek1 is not a substrate for phosphorylation, no trans-phosphorylation of this Kek1/EGFR chimera would occur, thereby maintaining Kek1's function as an inhibitor (Fig 5B). These chimeras contain the extracellular and transmembrane domains of Kek1 fused to kinase-deficient forms of the intracellular domain of EGFR tagged with GFP (KEG; Fig 5A). Three distinct kinase-deficient forms of KEG (KEgG, KEkG, and KEKG; HANKS et al. 1988 ) were engineered to prevent the chimeras from phosphorylating the endogenous receptor. Previous reports have demonstrated that a receptor with an intact kinase can trigger signaling by trans-phosphorylation of a kinase-deficient EGFR, indicating that a kinase-deficient receptor is capable of acting as a functional signaling substrate in vivo (RAZ et al. 1991 ; CLIFFORD and SCHUPBACH 1994 ; GUICHARD et al. 2002 ). Likewise, the Kek1/EGFR chimeras could act in a manner analogous to the vertebrate kinase-deficient receptor ErbB3, which signals by serving as a phosphorylation substrate for other ErbB family members (SLIWKOWSKI et al. 1994 ; KIM et al. 1998 ). The first two chimeras, KEgG and KEkG, contain point mutations that disrupt ATP binding and phospho-transfer, respectively, whereas the third chimera KEKG contains an internal deletion encompassing the entire kinase domain (QIAN et al. 1995 ; KLINGBEIL and GILL 1999 ).
View larger version (45K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4.
Kek1 is not phosphorylated in vivo. GFP-tagged constructs were misexpressed and immunoprecipitated (IP) from ovaries and immunoblotted (IB) with anti-phosphotyrosine (top). The blot was stripped and reprobed with anti-GFP (middle). A whole-lysate sample was probed with anti- tubulin as a loading control (bottom). With the exception of the control chimera KEG, none of the molecules tested were tyrosine phosphorylated.
|
|
As with Kek1, all three chimeras associate with the receptor by co-IPs, but are not phosphorylation substrates in vivo (Fig 4 and Fig 5A and Fig C; data not shown). Demonstrating that these chimeras are functional inhibitors, their misexpression in follicle cells resulted in strongly ventralized chorions, a phenotype comparable to that observed with Kek1 misexpression (Fig 5D). Similar inhibitory effects were also observed in other tissues (data not shown). Similar results were obtained with a form of KEgG that lacks the GFP tag (KEg), indicating that GFP does not interfere with the function or structure of the chimeras (Fig 5D). As an additional control, we confirmed that the cytoplasmic domain of EGFR is able to act as a functional phosphorylation substrate in a chimeric situation. We took advantage of the fact that Kek1 is able to homodimerize and created a control chimera, KEG, with a functional EGFR kinase domain. KEG is heavily phosphorylated in vivo and its misexpression results in dorsalization of the chorion in the presence or absence of receptor activity, consistent with the formation of homodimers and EGFR-independent activation of signaling (Fig 4 and Fig 5A, Fig C, and Fig D). This demonstrates that the cytoplasmic domain of EGFR in a Kek1 chimera is able to act as a functional signaling substrate in vivo and that the kinase activity of the inhibitory chimeras was effectively abolished by the kinase domain point mutations. Together our results argue that Kek1, as well as the chimeras, is not a phosphorylation substrate for EGFR. This supports the latter model in which the receptor kinase domain is unable to access Kek1 for phosphorylation and argues against a model in which Kek1 acts as a pseudo-substrate.
The juxta/transmembrane portion of Kek1 is essential for inhibition:
One potential explanation for the inability of the receptor to phosphorylate Kek1 is that rotation through the receptor's transmembrane region is required to appropriately position the kinase domain relative to its substrate. In this scenario, the structure of the Kek1 juxta/transmembrane (jt/tm) region might act to sterically hinder rotation. Consistent with an active role for the Kek1 jt/tm region, analysis of a series of domain swaps between Kek1 and Kek2 indicates that while the LRRs suffice for binding, the jt/tm region of Kek1 is necessary for full inhibitory activity in vivo (Fig 6). Three swaps, each containing progressively less of the extracellular/transmembrane portion of Kek1 fused to Kek2, were constructed and assayed for binding and inhibition of EGFR activity in vivo (Fig 6). All three swaps bind to the receptor, but display different degrees of inhibitory activity when assayed in vivo (Fig 6B and Fig C). The LRR (L), Ig (I), and jt/tm (T) swap (LIT), which includes the entire extracellular and transmembrane portion of Kek1 in place of the corresponding portion of Kek2, displays inhibition equivalent to full-length Kek1 (refer to Fig 1 and Fig 6C). In contrast, the next two swaps, which include only the LRRs and Ig (LI) or the LRRs (L) of Kek1, respectively, but lack the jt/tm portion of Kek1, have minimal inhibitory activity in the ovary, eye, and embryo (Fig 6C). Thus, while the LRRs of Kek1 suffice for receptor binding, the jt/tm region of Kek1 is specifically required, in addition to the LRRs, for full inhibition of the receptor in vivo. Supporting this, secreted forms of Kek1 (sKek1, sKek1GFP), display no inhibitory activity when assayed with CY2-GAL4 (data not shown; GHIGLIONE et al. 2003 ).
View larger version (41K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6.
The juxta/transmembrane region of Kek1 actively contributes toward EGFR inhibition. (A) Three constructs were generated where increasingly smaller portions of Kek1 were swapped with the corresponding regions of Kek2. LITGFP encodes Kek1's LRRs, Ig, and juxta/tm regions and Kek2's cytoplasmic domain. LIGFP encodes the LRRs and Ig domain of Kek1 and the juxta/tm and cytoplasmic regions of Kek2. LGFP contains only the Kek1 LRRs (including the cysteine-rich flanks). (B) Like full-length Kek1, these chimeras bind EGFR (top), demonstrating that the Kek1 LRRs are sufficient for associating with EGFR. In contrast, Kek2 binds EGFR weakly. (C) Misexpression of LITGFP in follicle cells with CY2-GAL4 results in strongly ventralized chorions, a phenotype similar to that of full-length Kek1. LIGFP and LGFP inhibit weakly, demonstrating that the juxta/tm region of Kek1 is necessary for full inhibition. Similar activities were also observed with GMR-GAL4 and Act5C-GAL4. V1–V3 denotes increasing degrees of ventralization, while R1–R3 denotes increasingly stronger rough eye phenotypes. Parentheses represent the number of independent transgenic lines tested.
|
|
Novel alleles of EGFR disrupt binding and inhibition by Kek1:
One simple interpretation of the current data is that the LRRs direct association with the receptor, while the jt/tm region of Kek1 facilitates inhibition. In light of this possibility and the knowledge that Kek1 and EGFR interact directly, we reasoned that identification of suppressors of Kek1 misexpression phenotypes would likely identify mutations in EGFR that disrupt this association, providing further insight to the mechanism underlying their association. To identify such suppressors, an F1 mutagenesis screen was employed to detect dominant mutations that suppress the effects of Kek1 misexpression in the eye (FREEMAN 1996 ; Fig 7A). From this screen five dominant suppressors of Kek1, or SOKs, that suppress the effects of Kek1 misexpression both in the eye (GMR-GAL4; UAS-kek1) and in the ovary (CY2-GAL4; UAS-kek1), were recovered (Fig 7, B–K). To further address the specificity of the SOKs, we examined their ability to suppress phenotypes associated with misexpression of DN-EGFR, Aos, Kek family members, and the Kek1/EGFR chimeras. Strikingly, the SOK mutations suppress only the effects of molecules containing the extracellular and transmembrane domain of Kek1, including both the inhibitory chimera KEgG and the activating chimera KECG (Fig 8 and data not shown). The SOKs do not interact indiscriminately with other molecules, but rather act in a dominant fashion to suppress Kek1-mediated increases or decreases in EGFR signaling.
View larger version (78K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7.
SOK mutations suppress misexpression of kek1 in multiple tissues. (A) A screen was aimed at identifying F1 suppressors of a Kek1 misexpression. (B–K) Adult eye SEMs and chorion images. (B and G) Wild type. Kek1 misexpression in the eye with GMR-GAL4 (C) and in follicle cells with CY2-GAL4 (H) is rescued by SOK3 (D and I), SOK4 (E and J), and to a lesser extent by SOK5 (F and K). GMR-kek1 refers to the [GMR-GAL4], [UAS-kek1] chromosome.
|
|
The first hint that the SOKs were not standard loss-of-function (LOF) mutations was their ability to suppress increases and decreases in EGFR signaling. Further supporting the notion that the SOKs represent unique mutations, the SOKs all map to the same region and deficiencies for this region fail to suppress the GMR-kek1 eye phenotype. Since the SOKs mapped to the vicinity of the receptor, we hypothesized that they represent unique alleles of the receptor, defective primarily in their ability to associate with Kek1. By eliminating the receptor's ability to associate with Kek1, these alleles would effectively prevent Kek1 or the Kek1/EGFR chimeras from affecting receptor signaling. Confirming this, SOK5, the only homozygous lethal SOK mutation, fails to complement alleles of egfr and contains a single missense mutation altering codon 750 (TAC to TGC), converting tyrosine to a cysteine (Y750C; Fig 9A). Thus, SOK5 represents a novel allele of the receptor that acts in a dominant fashion to suppress Kek1-dependent effects on EGFR signaling. Likewise, molecular analysis indicated that the remaining four SOKs, SOK1, SOK2, SOK3 and SOK4, also contain missense mutations in the receptor. SOK4 alters codon 718 (GAG to AAG), converting glutamic acid to a lysine (E718K), while SOK1, SOK2, and SOK3 alter codon 738 (CGA to CAA), converting arginine to a glutamine (R738Q; Fig 9A). Together the five SOKs represent changes to three residues spanning only 32 amino acids in the extracellular portion of the receptor. Moreover, they all lie within domain V, a region absent from all vertebrate orthologs of the receptor (Fig 9C). Finally, all of the SOKs are able to promote eye development (in the presence of Kek1 misexpression) and therefore must retain the ability to dimerize and initiate downstream signaling events. Consistent with this, the egfrSOK1–4 alleles are viable mutations and hence do not affect any vital functions of the receptor. In contrast, the lethality of egfrSOK5 indicates that it disrupts functions of the receptor essential for viability, in addition to its effects on Kek1.
View larger version (55K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9.
The SOKs are EGFR alleles that impair association with Kek1's. (A) Schematic representation of the Drosophila and human EGFR. Functional domains are labeled as I–V. Domains I and III are the ligand-binding domains while II and IV regulate dimerization. The SOK mutations all map to domain V within a 32-amino-acid stretch. (B) Mutations corresponding to SOK1–3 and SOK4 disrupt binding to Kek1 and KECG, but not to a GFP-tagged form of EGFR1 (top). GFP and EGFR loading controls are displayed in the middle and bottom, respectively. (C) Sequence alignment of Drosophila EGFR with orthologs from A. gambiae, C. elegans, and Homo sapiens. Domain V is conserved in the invertebrate receptors, but is absent in the vertebrate receptor. The SOK residues are highly conserved between Drosophila and Anopheles. In contrast, LET-23 contains one of the three SOK residues. Residues altered in the SOK mutants are indicated with arrows.
|
|
On the basis of our model, we predict that the SOK alleles retain the receptor's ability to homodimerize, while concomitantly disrupting its ability to associate with Kek1. To test this directly, we assayed the ability of the EGFRSOK molecules to associate with full-length Kek1, the activating chimera KECG, or wild-type EGFR in co-immunoprecipitations. Supporting our model, EGFRSOK has reduced affinity for both full-length Kek1 and KECG, but is able to associate efficiently with the wild-type receptor (Fig 9B). Consequently, domain V is crucial in mediating the receptor's interaction with and subsequent inhibition by Kek1.
| DISCUSSION |
|---|
Throughout development, EGFR activity specifies distinct cellular responses. Essential to this ability is the existence of an integrated network of regulatory molecules that direct receptor activity. Kek1, a member of a family of LRR- and Ig-containing molecules, represents a component of this network through its role as a feedback inhibitor of receptor activity. Deletion and mutagenesis studies have now demonstrated that the LRRs of Kek1, specifically LRR2 and G160, are essential for its association with, and consequently inhibition of, the receptor (GHIGLIONE et al. 2003 ; ALVARADO et al. 2004 ; Fig 10). The Kek1 cytoplasmic domain and associated Kek1 tail (KT) box have also been implicated in the inhibitory process, possibly through effects on subcellular trafficking (GHIGLIONE et al. 2003 ; DERHEIMER et al. 2004 ; Fig 10).
View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10.
A model of Kek1-mediated inhibition of EGFR. Full inhibition of EGFR activity requires the LRRs, transmembrane portion, and cytoplasmic domain of Kek1, as well as EGFR's domain V. The LRRs mediate binding to the receptor, the transmembrane portion promotes inhibition, while the cytoplasmic domain and associated KT box have been implicated in trafficking of Kek1.
|
|
Bipartite inhibition—the LRRs and jt/tm of Kek1 are required for EGFR inhibition:
While it is clear that the Kek1 LRRs are essential for EGFR binding and inhibition, secreted forms of Kek1 are nonfunctional, indicating that membrane anchoring is likely to be an essential element to the inhibitory mechanism (GHIGLIONE et al. 2003 ; this article). Directly testing this, our Kek1-Kek2 swaps demonstrate that while the Kek1 LRRs are sufficient for binding in vitro, they provide only minimal inhibition in vivo. Indeed, full inhibition is restored only when the entire extracellular and transmembane regions of Kek1 are placed in the context of a Kek2 backbone. This result supports an active role for the Kek1 jt/tm domain in inhibition, as a chimera containing the Kek1 LRRs in a Kek2 backbone is membrane tethered, but a weak inhibitor. This indicates that LRR-mediated binding alone is insufficient for receptor inhibition. Rather, our results suggest that Kek1-mediated inhibition of EGFR signaling is a bipartite process, in which the LRRs dictate EGFR binding and the jt/tm region facilitates inhibition (Fig 10). Phylogenetic analysis has indicated that this region is well conserved in Kek1 orthologs, supporting an important functional role for this region (DERHEIMER et al. 2004 ). Given this requirement for the Kek1 jt/tm region in inhibition, it was interesting to note that the SOK alleles identify three amino acids present in domain V of the receptor. Alteration of these three residues renders the receptor refractory to inhibition by Kek1 and activation by KECG, respectively. Moreover, two of the changes, R738Q and E718K, represent viable alleles of the receptor, capable of ligand binding and receptor activation. Together with our binding data, these results assign a role for domain V in mediating regulation by Kek1. It is interesting to note that EGFR domain V represents a third cysteine-rich domain in Drosophila, which is absent in the vertebrate ErbBs. This raises intriguing structural and evolutionary questions, as Kek1 has been reported to associate with all human ErbBs (GHIGLIONE et al. 2003 ). It will be important in the future to define those elements in the receptor that suffice for its inhibition by Kek1 and determine if additional distinctions in the interactions between Kek1 and the different receptor family members exist.
We also provide both direct (absence of phosphorylation) and indirect evidence (chimeras) that Kek1 is not a phosphorylation substrate for the receptor. This was somewhat surprising, as structural work with the vertebrate receptor has indicated that the EGFR kinase domain is in a catalytically open configuration (STAMOS et al. 2002 ). Such a configuration is unique in that receptor tyrosine kinases normally require activation loop phosphorylation to relieve autoinhibitory interactions that prevent substrate binding and phosphorylation (GOTOH et al. 1992 ; BURGESS et al. 2003 ). In light of a distinct mechanism for activation of EGFR, one proposal put forth is the rotation twist model, in which ligand binding induces dimerized receptors to pivot in or near the transmembrane domain, thereby reorienting the kinase domains to their substrates. One potential explanation for the inability of the receptor to phosphorylate Kek1 is that the structure of the Kek1 jt/tm region might act to hinder such a rotation.
Inhibition of EGFR is not a common Kek family function:
Considering that kek1 knockouts exhibit subtle and dose-dependent phenotypes, one important question remaining is, what is the role of Kek1 in a cellular and developmental context? An initial explanation for the subtle LOF phenotype of kek1 with respect to EGFR inhibition was the possibility of functional redundancy between members of the Kek family. However, our data for Kek2, Kek4, Kek5, and Kek6 indicate that EGFR inhibition is not a common feature of the Kek family. Alternatively, Kek1's inhibitory activity might reflect a recently acquired trait and not an ancestral or conserved role. However, analysis of kek1 orthologs in Drosophila virilis, D. pseudoobscura, and Anopheles gambiae argues against such a notion (DERHEIMER et al. 2004 ).
Although it is not a common feature within the Drosophila Kek family, it is unclear whether inhibition of EGFR by Kek1 represents a more widely conserved regulatory mechanism for receptors. For instance, the LRR-containing transmembrane protein Decorin binds to the human EGFR and has been implicated in the regulation of receptor activity (PATEL et al. 1998 ; IOZZO et al. 1999 ; CSORDAS et al. 2000 ; SANTRA et al. 2000 ). It has been reported, however, that the motifs in the Decorin LRRs required for binding EGFR differ from those of Kek1, suggesting these two LRR molecules are unlikely to represent comparable regulatory modes (SANTRA et al. 2002 ; ALVARADO et al. 2004 ).
Finally, a role for Kek1 in the nervous system has also been reported (SPEICHER et al. 1998 ). Expression of most kek family members is observed in the nervous system and recently three molecules that are structurally similar to Kek1, AMIGO1–3, have been implicated in neuronal development in vertebrates (KUJA-PANULA et al. 2003 ). It will be interesting to determine if Kek1 functions in neuronal development in an EGFR-independent manner and if such a role underlies its ancestral function.
| FOOTNOTES |
|---|
1 These authors contributed equally to this work. 
| ACKNOWLEDGMENTS |
|---|
We thank the Bloomington Drosophila Stock Center, N. Baker, L. Luo, A. Michaelson, N. Perrimon, and T. Schüpbach for reagents; T. Kaufman for advice and equipment; K. Klueg and L. Cherbas for advice in cell culture experiments; F. Rudi Turner for assistance with the SEMs; Tim Evans, Brandon Weasner, and Christopher Skipwith for their contributions to this work; and C. Ghiglione and K. Carraway III for sharing results prior to publication. We gratefully acknowledge the help of W. Forrester, A. Prieto, J. Kumar, and K. Cook in critical readings of the manuscript. This work was supported by a National Institutes of Health genetics training grant fellowship (GM07757) to D.A. and A.H.R. and by a National Science Foundation grant (IBN-0131707) to J.B.D.
Manuscript received October 1, 2003; Accepted for publication January 14, 2004.
| LITERATURE CITED |
|---|
ADAMS, M. D., S. E. CELNIKER, R. A. HOLT, C. A. EVANS, and J. D. GOCAYNE et al., 2000 The genome sequence of Drosophila melanogaster. Science 287:2185-2195.[Abstract/Free Full Text]
ALROY, I. and Y. YARDEN, 1997 The ErbB signaling network in embryogenesis and oncogenesis: signal diversification through combinatorial ligand-receptor interactions. FEBS Lett. 410:83-86.[CrossRef][Medline]
ALVARADO, D., A. H. RICE, and J. B. DUFFY, 2004 Knockouts of Kekkon1 define sequence elements essential for Drosophila epidermal growth factor receptor inhibition. Genetics 166:201-211.[Abstract/Free Full Text]
AROIAN, R. V., M. KOGA, J. E. MENDEL, Y. OHSHIMA, and P. W. STERNBERG, 1990 The let-23 gene necessary for C. elegans vulval induction encodes a tyrosine kinase of the EGF receptor sub-family. Nature 348:693-699.[CrossRef][Medline]
ASHBURNER, M., 1989 Drosophila: A Laboratory Handbook. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
BLUME-JENSEN, P. and T. HUNTER, 2001 Oncogenic kinase signalling. Nature 411:355-365.[CrossRef][Medline]
BOUTROS, M., J. MIHALY, T. BOUWMEESTER, and M. MLODZIK, 2000 Signaling specificity by Frizzled receptors in Drosophila.. Science 288:1825-1828.[Abstract/Free Full Text]
BURGESS, A. W., H.-S. CHO, C. EIGENBROT, K. M. FERGUSON, and T. P. J. GARRETT et al., 2003 An open-and-shut case? Recent insights into the activation of the EGF/ErbB receptors. Mol. Cell 12:541-552.[CrossRef][Medline]
CASCI, T., J. VINOS, and M. FREEMAN, 1999 Sprouty, an intracellular inhibitor of
TOP
ABSTRACT
MATERIALS AND METHODS
