Abstract
In the course of a large-scale program of ENU mutagenesis, we isolated a dominant mutation, called Velvet. The mutation was found to be uniformly lethal to homozygotes, which do not survive E13.5. Mice heterozygous for the Velvet mutation are born with eyelids open and demonstrate a wavy coat and curly vibrissae. The mutation was mapped to the proximal end of chromosome 11 by genome-wide linkage analysis. On 249 meioses, the locus was confined to a 2.7-Mb region, which included the epidermal growth factor receptor gene (Egfr). An A → G transition in the Egfr coding region of Velvet mice was identified, causing the amino acid substitution D833G. This substitution alters an essential triad of amino acids (DFG → GFG) that is normally required for coordination of the ATP substrate. As such, kinase activity is at least mostly abolished, but quaternary structure of the receptor is presumably maintained, accounting for the dominant effect. Velvet is the first known dominant representative of the Egfr allelic series that is fully viable, a fact that makes it particularly useful for developmental studies.
THE mammalian genome is believed to encompass ∼30,000–40,000 genes (Landeret al. 2001; Venteret al. 2001), but the number of phenotypes cataloged to date is much smaller than this, and, hence, it may be said that the essential function of most genes remains undetermined (Brown and Peters 1996). To close the “phenotype gap,” the chemical mutagen N-ethyl-N-nitrosourea (ENU) has been widely used to produce random germline mutations in mice (Ehlinget al. 1985; Hrabe de Angelis and Balling 1998; Justiceet al. 1999; Hrabe de Angeliset al. 2000; Nolanet al. 2000; Brown and Balling 2001), so that novel phenotypes can be identified, and the mutations responsible for them can be resolved by positional cloning. In this report, we focus on the development of the eyelids in mice, which undergo specific developmental changes both in utero and during early postnatal life.
Under normal circumstances, the eyelids begin to grow across the surface of the developing eye at E12.5. Between days 14 and 16 of gestation, the top and bottom eyelids continue to flatten, come to lie in close approximation to one another, and finally fuse tightly with each other at E16.5. They remain fused until ∼12 days after birth (Findlateret al. 1993). Failure of fusion leads to a readily apparent “open-eyelids-at-birth” defect. Eyelid closure is a process involving the migration of epithelial cells. Mutations that disrupt the signaling interactions between epithelium and the underlying mesenchyme can cause eyelid closure defects. Some examples include recessive mutations in the Tgfα, Egfr, MEKK-1, and Fgfr2 loci (Luetteke et al. 1993, 1994; Mannet al. 1993; Yujiriet al. 2000; Liet al. 2001). A number of unidentified loci, such as eob, lgGa, oe, and gp (Steinet al. 1967; Juriloffet al. 2000), can also produce an open-eyelid defect. In some cases, a wavy coat and curly vibrissae accompany open eyelids. This complex of traits is exemplified by the spontaneous mutations wa-1 and wa-2, which represent recessive lesions in the Tgfα and Egfr genes, respectively.
In the course of a large-scale ENU mutagenesis screening effort carried out in our laboratory, a dominant mutation (Velvet) was identified among a total of 16,606 F1 mice born to mutagenized C57BL/6 males and normal C57BL/6 females. The designation “Velvet” refers to the velvet-like texture of the coat of affected animals. However, Velvet mice are born with eyelids open and with curled vibrissae as well. They present a phenotype very similar to that of waved-1 and waved-2 mice. Although two similar phenotypes (so far unmapped) have been observed among 23,221 F3 animals generated and weaned to date, Velvet was the sole dominant mutation of its type. It quickly became evident that the mutation could not be maintained in the homozygous state (i.e., that it was homozygous lethal). We expanded the phenotype, creating a heterozygous stock, and cloned the mutation positionally.
MATERIALS AND METHODS
Animals: The C57BL/6J animals used in ENU mutagenesis were purchased from Jackson Laboratories, and C3H/HeN mice used for outcrossing and mapping were purchased from Taconic Farm. All studies involving animals were carried out in accordance with institutional regulations.
ENU mutagenesis and breeding: Six-week-old male C57BL/6J mice are treated with ENU administered in three weekly doses (90 mg/kg body weight) by intraperitoneal injection. After 12 weeks of recovery from infertility, each mouse is bred to normal C57BL/6J female mice so as to produce a maximum of 20 offspring. These F1 animals are either subjected to phenotypic screens or used to produce F2 mice, which in turn are backcrossed. Two F2 daughters per sire are backcrossed with the F1 male to generate F3 mice. A total of six F3 females and six F3 males are produced for screening. Approximately 50% of the mutations carried by each F1 mouse are transmitted to homozygosity in every panel of six F3 mice produced according to this scheme. To date, 16,606 F1 animals and 23,221 F3 germline mutants have been generated in our laboratory.
Morphological screening: All animals are examined for abnormalities affecting the eyes, the color or quality of the coat, development of the limbs or tail, and neurological or behavioral status at birth and at weaning age. All viable F1 phenodeviants are examined to determine whether the phenotype is heritable. Once transmissibility has been confirmed, each mutation is expanded, bred to homozygosity if possible, and mapped.
Genomic linkage analysis: Heterozygotes (first generation) were mated to wild-type C3H/HeN mice; the affected F1 hybrid mice (second generation) were then crossed with the wild-type C3H/HeN animal again. The offspring of the second generation were phenotyped and tailed for mapping. Genomic DNA was prepared from tail tips with the QIAGEN (Valencia, CA) DNeasy kit and adjusted to a concentration of 50 ng/μl. A total of 59 microsatellite markers were used for genome-wide linkage analysis. After chromosomal linkage was established and low-resolution confinement was achieved, a total of 39 microsatellites were surveyed across the critical region to identify informative markers. A total of 21 novel informative simple sequence length polymorphisms were identified in the Velvet region and are shown in Table 1.
DNA sequencing: Total RNA was isolated from the liver of a Velvet heterozygote on the C57BL/6 background and from a wild-type C57BL/6J mouse, respectively, by the QIAGEN RNeasy mini kit. A total of 1.5 μg total RNA was used in RT-PCR to generate cDNA with the RETROscript kit (Ambion, Austin, TX). The coding region of Egfr was amplified and directly sequenced to detect mutations. Sequencing was carried out using nested primers on an ABI 3100 sequencer and covered both strands of the template. Contigs were assembled by the programs phred and Phrap, and possible mutations were examined with the aid of the consed and polyphred programs.
Embryonic studies: For histological analysis of the eyelids, carriers of the Velvet mutation (C57BL/6J background) were bred with normal C57BL/6J mice. The presence of a vaginal plug was taken at 0.5 day of gestation. At E16.5, female animals were euthanized using CO2 asphyxiation, and embryos were dissected under a dissecting microscope. The heads of affected embryos, which were easily distinguished from those of unaffected embryos, were fixed in 4% paraformaldehyde and embedded in paraffin. Sections were stained with hematoxylin and eosin. The eyelids were examined in transverse section.
To examine the homozygous lethal effect of Velvet, heterozygotes (C57BL/6J background) were mated with each other. At specific embryonic stages (E10.5, E11.5, E12.5, E13.5, E15.5, and E17.5), embryos were dissected free of the uterine wall and subjected to genomic DNA isolation with the QIAGEN DNeasy tissue kit. A total of 100 ng of genomic DNA was used to amplify a 500-bp region that covers the Egfr mutation in question. The primers used for PCR were 5′ GTCATTCATGC CAGATAATTCCAA 3′ and 5′ CCAAATGCCATTCACAAAGTA GAG 3′. Genotyping was accomplished by sequencing the PCR products.
To determine whether midgestational defects of the placenta might result in homozygous lethality, heterozygotes (C57BL/6J background) were mated with each other. At E12.5, embryos and corresponding placentas were dissected from the uterine wall. Placentas were fixed in 10% formalin and embedded in paraffin. Sections were stained with hematoxylin and eosin. The embryos were used for genotyping.
Embryonic fibroblast migration assay: Velvet heterozygotes were mated with wild-type C57BL/6J mice. At E16.5 fetuses were isolated and pooled together by the phenotype. The fetal skin was peeled off and trypsinized at 4° overnight. Cell suspensions were centrifuged at 1200 rpm for 5 min and resuspended in DMEM (Dulbecco's modified eagle medium, GIBCO BRL, Gaithersburg, MD) supplemented with 10% fetal bovine serum. The fifth passage of cells was used for the migration assay.
The cell migration assay was performed using modified Boyden chambers (Transwell; Costar, Cambridge, MA) containing polycarbonate membranes as previously described (Klemkeet al. 1997). The cells were allowed to migrate for 5 hr; the migratory cells attached to the bottom surface of the membrane were stained with 0.1% crystal violet for 20 min at room temperature. The stain was eluted with 10% acetic acid, the absorbance was determined at 590 nm, and the percentage of migratory cells of wild-type or Velvet origin was compared.
In vivo phosphorylation: Wild-type and Velvet heterozygotes (1- to 2-day-old pups) were injected subcutaneously in the neck with phosphate-buffered saline (PBS) or with epidermal growth factor (EGF) dissolved in PBS, at a dose of 10 μg/g body weight. Ten minutes later, mice were killed, and their livers were harvested and homogenized in lysis buffer (20 mm Tris, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1% Triton X-100, 2.5 mm sodium pyrophosphate, 1 mm β-glycerol-phosphate, 1 mm Na3VO4, 1 μg/ml leupeptin, and 1 mm phenylmethylsulfonyl fluoride) as described previously (Donaldson and Cohen 1992). Homogenates were centrifuged at 14,000 × g for 5 min, and the protein concentrations in the supernatant were determined with Coomassie plus protein assay (Pierce, Rockford, IL). Equivalent amounts of protein were immunoprecipitated with anti-EGF receptor (EGFR) antibody (Cell Signaling Technology) and protein A-agarose (Sigma, St. Louis) prior to separation by SDS-PAGE using a 10% gel and subsequent immunoblotting. The immunoblot was incubated with anti-EGFR antibody and horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling Technology) to examine the expression level of EGFR. The protein was detected using the Phototope-HRP chemiluminescent detection system (Cell Signaling Technology). The blot was subsequently stripped and reprobed with anti-phosphotyrosine antibody (Cell Signaling Technology) to evaluate the autophosphorylation level of EGFR. The signals were measured using a personal densitometer (Molecular Dynamics, Sunnyvale, CA).
EgfrVelvet expression construct and transfection of HEK293 cells: The cDNA of EgfrVelvet with internal signal sequence removed was amplified using primers 5′ CCCAAGCTTTTGGAG GAAAAGAAAGTCTGC 3′ and 5′ GGAAGATCTTCATGCTCC AATAAACTCGCTGCT 3′. The cDNA was then cloned into pFLAG-CMV-3 expression vector (Sigma) between HindIII and BglII sites. The construct was verified by DNA sequencing.
Informative microsatellite markers identified in the Velvet region on chromosome 11
HEK293 cells were purchased from the American Type Culture Collection and grown in DMEM (GIBCO BRL) supplemented with 10% FBS, 2% penicillin/streptomycin. Cells were transfected with either pFLAG-CMV-3 vector or EgfrVelvet expression construct (amount indicated) using Fugene 6 (Roche). The cellular response to EGF stimulation was tested 48 hr posttransfection. At that time, transfected cells were serum starved for 24 hr and then stimulated with EGF (1 nm final concentration) for 10 min at 37°. Then whole cell lysates (12 μg/lane of protein) were subjected to Western blotting using antibodies against phospho-mitogen-activated protein kinase (phospho-MAPK), MAPK, and a Flag epitope.
Immunoblotting of activated MAPK: Transfected 293 cells were serum starved for 24 hr before being stimulated with 1 nm EGF for 10 min at 37°. Cells were rinsed twice with PBS, lysed in lysis buffer, scraped from the plates, transferred into a microcentrifuge tube, and kept on ice. Samples were sonicated for 5 sec four times to shear genomic DNA and reduce viscosity. The lysate was clarified by centrifugation at 14,000 rpm for 10 min, and the amount of total protein was determined with Coomassie plus protein assay (Pierce). Equivalent amounts of protein were separated by a 10% SDS-PAGE gel and transferred to nitrocellulose membrane. The activation of MAPK was monitored by Western blotting using antibodies against the phosphorylated form of MAPK (Cell Signaling Technology). The amount of total endogenous MAPK loaded for each sample was verified by probing against MAPK with a nonphosphospecific antibody.
—Failure of eyelid closure in Velvet mice. (A and C) Wild-type mice; (B and D) V/+ heterozygotes. Neonatal Velvet mice are born with eyelids open (B), whereas the eyelids of wild-type littermates remain closed until 12 days after birth (A). The upper and lower eyelids of a normal embryo are fused in the midline at E16.5 (arrow), whereas the eyelids of a Velvet embryo remain apart (C and D). Embryos were fixed in 4% paraformaldehyde and embedded in paraffin, cut in transverse section, and stained with hematoxylin and eosin.
RESULTS
Velvet was identified in F1 mice born to an ENU-treated sire and was assumed to be a dominant mutation. The phenotype proved to be readily transmissible. Analysis of the transmission ratio, determined by counting the number of normal and Velvet pups on a hybrid background (C57BL/6J Velvet × C3H/HeN), was consistent with the conclusion that Velvet heterozygotes are fully viable to term. Among 360 animals produced in such crosses, 184 showed the Velvet phenotype. Hence, it also appears that the mutation is fully penetrant.
—Embryonic fibroblast migration assay. Cells from Velvet embryos showed defective migration ability. The result was based on two individual experiments. Error bars indicate standard deviation.
Heterozygous Velvet mice are born with eyelids open and with curly vibrissae. Their first coat is wavy; later, the pelage hair loses its wavy appearance and becomes disoriented, with individual shafts projecting in various directions relative to neighboring shafts. The eyes of adult Velvet mice are often small, with corneal opacity, and excessive secretions are always seen at the edges of eyelids. Aside from the eyelid and fur abnormalities, heterozygous animals appear healthy and show normal fertility. Compared to their littermates, Velvet mice have a normal body size and exhibit normal cage activities.
The eyelid defect as it appears in neonatal Velvet mice is depicted in Figure 1. As seen in histological sections (Figure 1, C and D), the upper and lower eyelids of normal mice become approximated and then fuse to a state of closure by E16.5, whereas the eyelids of Velvet embryos remain far apart, never coming into contact with one another. The protruding ridges of Velvet eyelids were well formed, and growth across the cornea seemed to be initiated in the presence of the mutation as well, but migration of epithelia cells was defective compared to that in normal mice.
Migration of eyelid epithelial cells is difficult to quantitate in vivo. To provide a numerical index of the migration defect, and to demonstrate that the mutation imparted a cellular phenotype in vitro, we analyzed the migration of embryonic fibroblasts isolated from wild-type and Velvet embryos. The fibroblasts from Velvet embryos demonstrate ∼53% of the migratory ability of wild-type embryos (Figure 2).
—Genetic mapping of the Velvet locus. (A) Whole-genome linkage analysis. On 51 meioses, the Velvet locus is confined to the proximal end of chromosome 11 with a peak LOD score of 9.8. A total of 59 microsatellite markers were used. (B) Fine genetic mapping of the Velvet locus. On 249 meioses, the mutation is confined within the region flanked by markers Vel_6 and D11Mit227. There are three crossovers on the proximal side and five crossovers on the distal end. The position of markers is presented with reference to the Celera database.
Aside from the eyelid defect, other structures of the eye seem normal. Since the open eyelids constantly leave the cornea exposed, the corneal opacity and excess secretion displayed by adult Velvet mice might not be due to developmental defects, but rather to infection, drying, and possibly trauma.
By analyzing 249 meioses, we were able to confine the Velvet mutation to an interval 3.2 cM in length, demarcated by the microsatellite markers Vel_6 and D11Mit227. The critical regions lie near the proximal end of chromosome 11, occupying a 2.7-Mb interval (11.9–14.6 Mb from the centromere; Celera distances; Figure 3). Within the critical region, 38 annotated genes are listed in the Celera database. Ten of these are pseudogenes and 28 are authentic genes. Among the authentic genes, Egfr was considered the most promising candidate, given the fact that the wa-2 phenotype (very similar to the Velvet phenotype) was caused by a spontaneous recessive mutation in Egfr (Luettekeet al. 1994). The EGF receptor is expressed in a wide range of adult tissues and cell types (Paria and Dey 1990), and it is believed that activation of the EGF receptor signaling pathway contributes to the regulation of numerous cellular processes, both during embryonic development and in the adult (Carpenter and Cohen 1990; Wells 1999).
—Velvet corresponds to a single-base-pair substitution of the Egfr gene. This A → G transition results in Asp replacement by Gly. The consed display shows bidirectional sequencing of the heterozygous Egfr mutation using a template amplified from a mouse with the Velvet phenotype.
The complete coding region of Egfr was amplified from liver cDNA, prepared from normal C57BL/6 mice and Velvet heterozygotes, respectively. A single heterozygous nucleotide substitution was detected in the Velvet template (Figure 4). This, an A → G transition, typical of ENU mutations (Justiceet al. 1999), is predicted to cause the replacement of an aspartic acid residue at position 833 with a glycine. The mutation thereby alters the structure of the EGF receptor cytoplasmic tyrosine kinase domain and falls within a triad of residues—the DFG motif—known to be important for ATP coordination (Stamoset al. 2002).
Previous work has revealed that injection of EGF into a newborn mouse can rapidly induce EGFR tyrosine phosphorylation (Donaldson and Cohen 1992), which occurs through an internal autophosphorylation mechanism (Hubbardet al. 1998). To examine Velvet EGFR activity in vivo, neonatal Velvet mice and their wild-type littermates were injected subcutaneously, either with PBS alone or with EGF dissolved in PBS. As expected, in the wild-type mouse, injection of EGF led to enhanced tyrosine phosphorylation of EGFR. In the Velvet heterozygote, there was a sixfold decrease of EGFR autophosphorylation after EGF challenge as compared with the wild-type control. Moreover, basal tyrosine phosphorylation of EGFR was reduced by threefold (Figure 5).
The dominant effect of EgfrVelvet was observed in vitro as well (Figure 6). Human epithelial cell line 293 cells, known to express EGFR and Erbb2 protein (data not shown), were transiently transfected either with various amounts of an EgfrVelvet expression construct or with empty vector as a control. Upon 1 nm EGF stimulation, cells transfected with empty expression vector showed pronounced induction of the activated form of MAPK (phospho-MAPK, p42/p44); the EgfrVelvet expression construct caused a dose-dependent inhibitory effect on MAPK activation.
Because Velvet homozygotes were never observed at term throughout the course of our studies, we considered the mutation to be homozygous lethal. We wished to determine the stage at which death occurs in utero. We therefore intercrossed Velvet heterozygotes (C57BL/6 background) to determine whether homozygosity for Velvet was compatible with survival to a postembryonic stage of development (Table 2). Dividing gestation into windows of time, we observed that between E10.5 and E11.5, the genotypic distribution (+/+:+/V:V/V) was 5:9:3. Between E12.5 and E13.5, the distribution was 5:13:2. Between E15.5 and E17.5 the distribution was 5:18:0. And at term, the distribution was 11:17:0. These data (along with phenotypic assessments of a much larger number of mice, mentioned earlier) suggest that heterozygotes have no diminution in viability. However, homozygotes show a trend toward diminished viability during embryonic life and die between E13.5 and E15.5, a time that roughly corresponds to the transition between embryonic and postembryonic life (E14.5).
—EGF-induced autophosphorylation of EGFR in vivo. (A) EGFR expression in liver dissected from a neonatal Velvet mouse (V/+ genotype) and a wild-type littermate, respectively. Two-day-old pups were injected with PBS or with EGF in PBS (10 μg/g body weight) subcutaneously in the neck. Ten minutes later, pups were killed and livers were removed and homogenized. Equivalent amounts of protein were subjected to immunoprecipitation and subsequent immunoblotting with anti-EGFR antibody. (B) Tyrosine phosphorylation of EGFR upon EGF stimulation. Immunoblot from A was stripped and reprobed with anti-phosphotyrosine antibody.
—EGF-induced activation of MAPK in vitro. A total of 293 cells were transfected with either 2 μg of empty pFLAG-CMV-3 vector (C) or pFLAG-CMV-3 containing the EgfrVelvet expression construct (amounts indicated). The level of MAPK activation was normalized to take account of the total amount of endogenous MAPK loaded per lane.
Histological analysis of E12.5 placentas revealed that the labyrinth trophoblast layer in homozygous placenta was disorganized. The cell number was greatly reduced when compared to the placenta of wild-type and heterozygous embryos, and a number of dead or dying cells were evident. The organization and cell number of the giant-cell layer of the trophoblast were fairly normal (Figure 7). Since fetal nutrition depends on the placental labyrinthine layer, we consider it likely that homozygosity for EgfrVelvet produces midgestation lethality due to a placental defect.
Embryonic lethality in Velvet homozygotes
DISCUSSION
Velvet is a new member of the Egfr allelic series. Although an EGFR construct with at least some dominant negative properties has been expressed in mice as the product of a transgene (Murillaset al. 1995), no in situ modification of the Egfr locus has previously been shown to yield a dominant effect. Hence, Velvet is the sole dominant representative of the Egfr allelic series described to date. It results from a point mutation within exon 21 (of 28 exons in the gene) that causes an amino acid substitution (D833G) within a highly conserved motif of the receptor tyrosine kinase domain.
From the crystal structure of the EGF receptor tyrosine kinase domain, D833 lies within the “DFG motif,” which defines the beginning of the “activation loop” and is important for ATP coordination (Stamoset al. 2002). This residue (and the whole of the DFG motif, as well as several flanking amino acids) is strictly conserved in all known EGF receptor sequences from a broad range of species. Moreover, in PRINTS (a protein fingerprint database; http://www.bioinf.man.ac.uk), it is observed that the DFG motif is invariably found as a part of the tyrosine kinase catalytic domain signature. The inflexibility of the DFG motif suggests its conformational importance: by interacting with other residues, the DFG motif is believed to provide a favorable structural arrangement for the “catalytic loop” to accommodate the ATP binding. At the same time, the DFG motif may help the activation loop attain or stabilize the conformation required for catalytic activity upon phosphorylation (Hubbardet al. 1998; Stamoset al. 2002). The D833G mutation replaces the relatively large, acidic side chain with a hydrogen atom. The steric and/or charge effect of the mutation is evidently sufficient to adversely affect receptor function and to do so in a dominant fashion. In vivo, although EGFR is clearly expressed, inducible autophosphorylation occurred at only one-sixth the level observed in a wild-type littermate. In this respect, the phenotypic effect of the Velvet mutation is distinguishable from that of a null allele, which is strictly recessive.
—Histologic analysis of midgestation placentas. E12.5 placentas were sectioned and stained with hematoxylin and eosin. The genotypes were determined by PCR amplification of genomic DNA from the corresponding embryos. (A) Wild type; (B) V/+; (C) V/V. DC, decidua; GT, trophoblast giant cells; L, labyrnthine placenta.
The outer root sheath of the hair follicle and the epidermal layer of the eyelid probably express EGFR in greater abundance than do most other tissues (Greenet al. 1983; Murillaset al. 1995) and, on this basis, may be markedly dependent upon EGFR, a circumstance that may make them particularly sensitive to perturbations in EGFR signal transduction. Targeted expression of a truncated human EGFR (HERCD-533, which has proved to act as a dominant negative mutant) to the basal layer of epidermis and outer root sheath of hair follicles of transgenic mice severely alters hair follicle development and skin structure. Most of these mice eventually developed alopecia (Murillaset al. 1995). Both the outer root sheath of the hair follicle and the epidermal layer of the eyelid display an immense proliferative capacity and probably depend upon EGFR as a regulator of cell migration through the extracellular matrix, a process that is highly dependent upon interactions between epithelium and underlying mesenchyme. Upon binding its ligands (mainly TGF-α and EGF), EGFR activates multiple signaling pathways to stimulate epithelial cell proliferation and migration (Jhappanet al. 1990; Sandgrenet al. 1990; Wilson and Gibson 1999). EGFR is linked to two pathways of particular importance in the promotion of cell motility: the phospholipase Cγ (PLCγ)-protein kinase C (PKC) and the Ras-MAPK pathways. Once activated, PLCγ catalyzes the hydrolysis of phosphatidylinositol (4,5)-bisphosphate to yield the second messengers diacylglycerol and inositol-1,4,5-triphosphate, which effect calcium release from intracellular stores and activate the PKC-mediated cascade (Wells 1999; Jorissenet al. 2003). At the same time, EGFR activation induces immediate activation of the Ras/MAPK pathway, which can enhance myosin light chain kinase (MLCK) activity, leading to phosphorylation of myosin light chains (MLC; Klemkeet al. 1997). Thus, through the PLCγ and Ras/MAPK signaling cascade, EGFR directly influences the motility machinery of the cell. These two signaling pathways are not only activated simultaneously but also interlinked with each other, presenting a complex but delicate regulatory system. Any mutation that disrupts signal transduction in these pathways will cause dysregulation of cell migration. This effect is likely manifested in defects of eyelid and hair development, as seen in mice with the wa-1, wa-2, MEKK1, and Velvet mutations.
Depending upon their genetic background, Egfr–/– mice die during the peri-implantation stage due to degeneration of the inner cell mass or during midgestation as a result of placental defects or are born with their eyes open but live only up to 3 weeks of postnatal life as a result of respiratory problems, epithelial immaturity, and multiorgan abnormalities (Sibilia and Wagner 1995; Threadgillet al. 1995; Miettinenet al. 1999), while the Velvet heterozygotes show abnormalities in only the eyelids and hair follicles. On the other hand, Velvet homozygotes die in late embryonic life due to the placental insufficiency, whether the mutation is on the C57BL/6 background or on a mixed C3H/HeN × C57BL/6 background. The strong dominant effect of the Velvet allele may have two general explanations, neither of which excludes the other.
First, it is believed that EGFR activation involves the formation of an active homodimer upon ligand binding, and the activation is achieved by autophosphorylation that occurs in a “trans” mode in the receptor complex. Since normal levels of EGFR expression are observed in Velvet heterozygotes, it is likely that the integrity of dimer formation is undisturbed by the mutation; however, EGF signaling clearly is interrupted, consistent with the loss of three-quarters of receptor activity (to be expected with a kinase-dead dominant subunit residing in a dimeric protein) or perhaps even more.
Second, the interaction between EGFR and other cell surface protein kinases of the ErbB family leads to the formation of heterodimers or even hetero-oligomers. For example, ErbB2 is the preferred heteromeric partner for EGFR (Yarden and Sliwkowski 2001; Schlessinger 2002; Jorissenet al. 2003). While the ligand for the ErbB2 heteromer is presently unknown, the biological activity of ErbB2 is at least largely dependent upon the formation of a heterodimer or hetero-oligomer with EGFR. The product of the Velvet allele might potentially disrupt signal transduction through EGFR homodimers and EGFR/ErbB2 heteromers alike, as well as transduction through other receptor complexes. In this respect, a kinase-dead allele would predictably be more deleterious than a null allele.
In vitro, the inhibitory effect of the EgfrVelvet product on the EGF-induced MAPK activation was readily demonstrated, confirming that the mutant allele is capable of disrupting the function of the normal allele. Inhibition was dose dependent and consistent with the poison subunit model proposed.
We further note that homozygosity for the wa2 allele creates a phenotype approximately similar to that of heterozygosity for Velvet. Overall, we suggest that phenotypic severity of different genotypes might be graded as follows:
Not all genotypic components of the series (those that are not underlined) have yet been examined, and, in particular, the phenotypic effect of compound heterozygosity for Velvet and either the wa2 or null allele (not underlined above) remains a matter of speculation.
The fact that a heterozygous animal is morphologically marked and fully viable while homozygosity is embryonic lethal may make the Velvet mutation particularly useful for developmental studies. Moreover, differences between the phenotype of Velvet and the phenotype of other Egfr alleles may permit added insight into the role played by specific components of the EGFR/ErbB receptor family.
Acknowledgments
The authors are grateful for the assistance of Adrian Smith in the histologic analysis of embryonic placentas. This work was supported by funding from National Institutes of Health grants GM067759, GM60031, U54A154523, and 2P01-AI40682 and by a grant from the Defense Advanced Research Projects Agency.
Footnotes
-
Communicating editor: N. A. Jenkins
-
Note added in proof: We have become aware of an independent dominant hypomorphic allele of Egfr, reported by C. Thaung, K. West, B. J. Clark, L. McKie, J. E. Morgan et al. (2002, Novel ENU-induced eye mutations in the mouse: models for human eye disease. Hum. Mol. Genet. 11: 755–767). The molecular defect responsible for this allele, Wa5, has not yet been reported.
- Received July 9, 2003.
- Accepted September 24, 2003.
- Copyright © 2004 by the Genetics Society of America