The otic placodes form as a pair of epithelial primordia located on either side of the posterior rhombencephalon. In the avian embryo, the primordia become visible by stage 10 (Hamburger and Hamilton, 1951) as a pair of thickened regions of the head ectoderm. They subsequently invaginate to form the membranous labyrinths of the inner ear. In contrast to other primordia, vesicularization occurs by means of formation of a series of folds in distinct areas of the primordium (Hilfer et al., 1989). Formation of these folds gives the primordium a distinctly box-like appearance during establishment of the otic vesicle. As the folds converge, the otic pit deepens while the vesicle closes and pinches off from the adjacent head ectoderm.
The forces responsible for this folding during invagination may be either intrinsic or extrinsic to the primordium. Intrinsic factors that may be responsible for the folding of the primordium include activation of the cytoskeleton and interkinetic nuclear migration. Interaction with extrinsic agents such as the neural tube, neural crest, and somitomeric mesoderm may also influence this folding.
Invagination of the otic primordium has shown ATP and calcium independence (Hilfer et al., 1989). This finding would indicate that the actin/myosin cytoskeletal motoring model for invagination in other primordia, such as the neural plate (Burnside, 1973; Sadler et al., 1982; Lee et al., 1983), thyroid (Shain et al., 1972; Hilfer and Pakstis, 1977), pancreas (Wessells and Evans, 1968), and optic vesicle (Brady and Hilfer, 1981) is not the primary mechanism responsible for invagination of the otic placode. Rather, it resembles more the elevation of neural folds in the chick embryo which is motivated by extrinsic forces (Schoenwolf et al., 1988). However, changes in internal parameters surely are required in addition to whatever external forces are driving invagination. It is possible that the formation of folds is due to an interaction of the otic primordium with surrounding tissues such as the neural tube along the medial surface or somitomeric mesoderm along the lateral surface of the primordium. This interaction is likely to be mediated through components of the extracellular matrix and cell membrane receptors.
This investigation involved testing the hypothesis that the well-documented attachment of the otic primordium to the lateral wall of the neural tube during early development (Noden and Van de Water, 1986; Hilfer et al., 1989; Alvarez and Navascues, 1990; Mayordomo et al., 1998) is necessary for normal invagination. More specifically, the role of extracellular matrix (ECM) components and cell membrane receptors in the mediation of this association of the otic primordium with the lateral wall of the neural tube was investigated. Antibodies to extracellular matrix components such as laminin, fibronectin, and perlecan (a basement membrane heparan sulfate proteoglycan [HSPG]) were microinjected into the head mesenchyme of Hamburger and Hamilton (HH) stage 10 chick embryos in the region where the basal laminae of the otic primordium and the rhombencephalon normally fuse slightly later in development (Hilfer and Randolph, 1993). Each antibody elicited different degrees of inhibition of attachment of the primordium to the neural tube and subsequent folding. This variability of inhibitory activity coincides with differences in the distribution of these matrix components in this region of the embryo. Antibodies to laminin produced the most marked effect. Microinjection of antibodies to integrin receptors also interfered with association of the primordium with the neural tube and subsequent fold formation. Antibodies were selected based on their immunoreactivity to integrin subunits that serve as receptors solely for laminin as well as receptors for groups of extracellular matrix components, including laminin. Although some of these antibodies exhibited inhibitory effects, they were not effective to the same degree as antibodies to extracellular matrix components.
Injection of Antibodies to Extracellular Matrix Components
The close association of the basal surface of the otic primordium with the basal surface of the posterior rhombencephalon is most likely mediated by components of the extracellular matrix. Perturbation of formation of an organized extracellular matrix could prevent this stabilizing interaction from occurring. See Table 1 for a summary of injection results.
Table 1. Effects of Microinjection of Antibodies to ECM and Integrins
Development of the otic pit begins with the formation of a series of folds within the otic primordium (Fig. 1, see also Hilfer et al., 1989). The curved surface of the otic placode at stage 10 (Fig. 1a) first forms a crease (the primary longitudinal fold) along the junction of the neural tube with the paraxial mesoderm (Fig. 1b). Slightly later, ventral and then dorsal longitudinal folds appear along the same embryonic axis as the primary longitudinal fold. The rostral and caudal transverse folds rise perpendicularly to the longitudinal folds and mark the rostral and caudal margins of the otic primordium. Collectively, these folds give the otic pit a box-like appearance by stage 13 (Fig. 1c).
Injection of Hanks' balanced salts solution (HBSS) or nonimmune serum did not affect invagination of the otic pit (see Table 1 and below). The primordium contralateral to the site of antibody injection also exhibited normal folding (see Figs. 2–4). Injection of monoclonal antibodies to avian laminin, at stage 10, into the mesenchyme just rostral to the cranial margin of the otic placode inhibited attachment and subsequent folding in a dose-dependent manner. Injection of relatively small amounts of antibody (30–60 pg) inhibited both attachment of the primordium to the neural tube and subsequent folding. In a few cases, a shallow otic pit was formed and the opening on the experimental side was wider in diameter compared with the uninjected side, when viewed from the surface (Fig. 2a). In 90% of cases, however, folding was completely inhibited. The injected side failed to form primary longitudinal, dorsal, and ventral folds (Fig. 2b–d). Higher doses of antibody (150 pg) caused complete detachment of the primordium from the neural tube (Fig. 3). In fact, the ectoderm was so destabilized by these doses that it often fell off during fixation. When viewed from the surface, these primordia did not exhibit a normal folding pattern. Instead, they exhibited numerous, shallow, random folds that were not present in control primordia.
Injection of anti-perlecan (HSPG) antibodies into the same region of stage 10 embryos elicited a different effect. Injections of comparable amounts of antibody (60 pg) did not disturb attachment of the primordium to the neural tube (Fig. 4a). Similar doses of anti-laminin antibody were sufficient to prevent attachment of the otic primordium along the majority of its length to the neural tube. Embryos injected with anti-perlecan were closely attached to the neural tube in a manner similar to control primordia. However, these experimental primordia exhibited a delay in folding (Fig. 4a). The otic primordium appeared curved but did not exhibit the three characteristic primary longitudinal, ventral, and dorsal folds. Another interesting effect of injection of antibodies to perlecan was a distortion of the mesenchyme in the region ventral to the primordium (Fig. 4b). This embryo is an example of more pronounced inhibition of folding on the injected side. The mesenchyme on the injected side appeared compressed, and structures in this region, such as the cardinal vein, had an aberrant morphology (Fig. 4b). Interestingly, embryos injected with comparable doses of anti-perlecan at slightly later stages (stage 10+ and 11) exhibited marked distortion of the neural tube in addition to inhibition of otic morphogenesis and expansion of the ventral mesoderm (Fig. 4c).
Injection of antibodies to avian fibronectin elicited an effect that was different from either anti-laminin or anti-perlecan antibodies. Embryos were injected with comparable amounts (60 pg) of antibody at stage 10. Injection of anti-fibronectin did not elicit detachment of the primordium from the neural tube. Experimental primordia were closely attached to the rhombencephalon and exhibited normal primary longitudinal and dorsal folds. These embryos often exhibited a delay in formation of the ventral fold on the injected side. Concomitant with this event was a notable compression of the mesenchyme ventral to the primordium (Fig. 5). This compression was marked by shrinkage of the mesenchyme away from the basolateral surface of the primordium. At later stages (14–15), the mesenchyme had not begun to migrate between the separating neural tube and otic primordium (not shown).
Injection of Antibodies to Integrin Subunits
Two possibilities were considered to explain how microinjection of antibodies to extracellular matrix components interrupted attachment of the otic primordium to the posterior rhombencephalon. First, antibody binding to cognate ligand may prevent the formation of a stable and organized matrix between the primordium and the neural tube. The other possibility is that antibody binding prevents interaction of the matrix component with its cell surface receptor. This second hypothesis was investigated by using antibodies against integrin subunits. The first approach was to inject antibodies that would abrogate the binding of β1 integrins to substrate. Two antibodies that block β1 binding were used. (See Experimental Procedures section for details.) The first antibody, the CSAT antibody, is a monoclonal antibody to β1 integrins whose effects have been demonstrated to be time- and dose-dependent (Horwitz, 1983; Buck et al., 1986; Drake et al., 1991). It was expected that interruption of β1-containing integrin binding would have a pronounced effect, because the majority of epithelial integrin receptors for laminin, fibronectin, and collagen type IV contain a β1 subunit (Albeda and Buck, 1990; Ginsberg et al., 1990; Roman et al., 1991; Ruoslahti et al., 1994). However, injections of CSAT did not have as severe an effect as did injection of antibodies to laminin. A minimum dose of 320 ng of antibody was required to prevent attachment of the otic primordium to the neural tube. This dosage was sufficient to inhibit subsequent fold formation (Fig. 6a). It is noteworthy that this is an appreciably higher dose (104-fold greater) than was required to elicit the same effect with anti-laminin. Considerably higher doses were required to cause complete detachment of the ectoderm from the neural tube. Lower doses did not give reproducible results.
Injection of another antibody that blocks binding of the β1 integrin subunit, JG22 (Greve and Gottleib, 1982), had an effect similar to CSAT. Although this antibody could effect a complete detachment of the otic primordium from the neural tube, the more common effect was formation of a normal primary longitudinal fold followed by an apparent collapse of the primordium along this hinge point (Fig. 6b). To show that the effects of these two antibodies were due to blocking of β1 binding activity, Horwitz's V2E9 (Hayashi et al., 1990), an antibody that binds to the cytoplasmic domain of β1 integrins without perturbing function, was microinjected into the same region of similarly staged embryos. Embryos injected with this antibody appeared normal (Fig. 6c); there was no distinguishable difference between injected and control primordia.
Finally, antibodies to α6-containing integrins were microinjected. The α6β1 integrin is the predominant epithelial laminin receptor (Akiyama et al., 1990), and the α6β4 is the major basement membrane receptor (Quaranta, 1990; Ruoslahti et al., 1994). Injection of comparable amounts of this antibody had no effect. Attachment was not inhibited and formation of folds occurred in a normal manner (Fig. 7a). Injection of nonspecific mouse IgG (the same general isotype as the experimental antibodies) or HBSS elicited no effect (Fig. 7b,c).
Localization of Injected Antibodies
Injected primary antibodies were immunolocalized with fluorescein-conjugated goat anti-mouse immunoglobulin G (IgG). Fixed, whole mounted injected embryos were viewed with a 20× objective lens (0.5 NA) on the Nikon Optiphot microscope with fluorescence optics. These embryos exhibited bright fluorescence at the injection site (Fig. 8). The basal surfaces of the neural tube and the otic primordium were brightly fluorescent. There was some cranial and caudal spreading of antibody along the neural tube on the injected side of the embryo. Laser scanning confocal microscopic analysis supported the localization observed in embryos studied in whole-mount, i.e., there was axial and lateral spread of antibody on the injected side of the embryo without diffusion either into the deeper ventral tissue or to the contralateral side of the embryo (not shown). Control embryos exhibited no fluorescence above background levels (not shown).
This study provides experimental evidence that extracellular matrix components, particularly laminin, play a key role in the morphogenesis of the otic primordium. The data suggest that extracellular matrix in the region between the otic primordium and the rhombencephalon has an important role in stabilization of the primordium as it is shaped by extrinsic developmental events. Microinjection of antibodies to laminin and integrin cell surface receptors interrupted binding of the otic primordium to the lateral wall of the neural tube. Perturbation of the association of these two epithelial structures inhibited subsequent folding of the otic primordium. Injection of antibodies against laminin was most effective in inhibiting the association of the otic primordium with the neural tube and subsequent invagination. This inhibition was dose-dependent. Low doses partially detached the otic epithelium from the neural tube and interfered with subsequent folding of the primordium. Higher doses elicited a complete detachment of the primordium from the neural tube and spreading along the surface of the embryo. These results are supported by previous studies in our laboratory which revealed that laminin is one of the most ubiquitously located basal lamina components in this region of the embryo (Hilfer and Randolph, 1993).
Microinjection of antibodies against other components of the extracellular matrix did not affect the attachment of otic primordium to the rhombencephalon. In embryos injected with anti-perlecan (a basement membrane HSPG), otic primordia were closely attached to the neural tube but exhibited a delay in folding accompanied by a compression of the mesenchyme in the region of the injection. The rather limited effects of anti-perlecan injection on otic morphogenesis may also be explained by the limited distribution of HSPGs along the exterior of the basal lamina. These results are interesting as HSPGs serve as binding sites for growth factors (Vlodavsky et al., 1996; Larrain et al., 1997; Schulz et al., 1997; Zhou et al., 1997). Interruption of HSPG in the basement membrane may affect the activity or bioavailability of cytokines or other morphogenetic cues in the injected region. This affect may prevent transmission of mitogenic signals across the cell membrane or might interfere with proliferation or expansion of the mesenchyme.
Injection of antibodies to fibronectin elicited a different set of abnormalities, including delayed formation of the ventral fold, a marked compression of mesenchyme in the region of the injection, and a noticeable gap between the basolateral surface of the primordium and the somitomeric mesoderm. Subsequently, the mesenchyme failed to migrate between the basal surfaces of the otic vesicle and the neural tube. This result is consistent with the distribution of fibronectin between the otic pit and the rhombencephalon when they become separated during later stages of otic morphogenesis (Hilfer and Randolph, 1993). At this point in development, invading mesenchymal cells become surrounded by a fibronectin lattice. Disruption of the fibronectin component of the ECM could prevent or delay expansion of the head mesenchyme ventral to the primordium and elicit a delay in lateral fold formation. Fibronectin, in addition to laminin, hyaluronate, collagens, and chondroitin sulfate proteoglycans have been implicated in neural crest migration. Perturbation of fibronectin in this region of the embryo may also interrupt neural crest migration in this region of the embryo. This interruption may, in turn, affect shaping of the otic primordium; streaming of migrating neural crest in this region of the embryo could contribute to the shaping of the otic vesicle (Bronner-Fraser, 1990; Bronner-Fraser et al., 1991; Birgbauer et al., 1995).
Laminin binds to cell surfaces through specific integrin receptors. The antibodies to laminin could prevent attachment of the otic primordium to the posterior rhombencephalon by two possible mechanisms. First, antibody binding could prevent formation or maintenance of an ordered ECM and prevent the association of the basal laminae of the otic epithelium and the neural tube. The other possibility is that the epitope which the antibodies bind is necessary for the interaction of the matrix component with its cell surface receptor, producing the same result. Because the ECM antibodies used in this study were not epitope mapped, it was expected that the results of microinjection of anti-integrin antibodies would facilitate distinction between these two possibilities. Antibodies to the β1 integrin subunit were chosen because the majority of epithelial integrin receptors for laminin, fibronectin, and collagen IV contain the beta 1 subunit (Quaranta, 1990). Antibodies reactive to the α6 subunit were injected because the major epithelial integrin laminin receptors contain an α6 subunit (α6β1, α6β4). Microinjection of these antibodies did not elicit the expected results. Anti-β1 antibodies prevented attachment of the primordium to the rhombencephalon and inhibited subsequent folding. However, these antibodies did not cause complete detachment of the ectoderm in the same manner as anti-laminin antibodies. The α6 antibodies exhibited no effect on attachment of the otic anlage to the neural tube and subsequent folding. There are several interpretations for these results. First, there may not have been adequate penetration of the antibodies between the basal laminae of the otic primordium and the neural tube. This is an unlikely explanation as high magnification laser scanning confocal microscope images reveal bright staining along the basal laminae in this region (not shown). Alternatively, it is possible that other epithelial laminin receptors are expressed in this region of the embryo. Other reported integrin receptors for laminin are α3β1, α1β1, and α7β1 (Ruoslahti et al., 1994). This redundancy of function across the integrin family may be the best explanation for failure of antibodies to α6 integrin subunits to perturb the morphogenetic events involved in otic pit formation. It is reasonable to speculate that there are multiple laminin receptor integrins expressed, in a developmentally regulated manner, in the region where the basal surface of the otic primordium abuts the basal surface of the neural tube. Another possibility is that integrin subunits that are bound to their ligand are inaccessible to binding by immunoglobulins. This could prevent abrogation of integrin function using our experimental protocol. This argument is supported by our observations which suggest that antibodies to integrins often do not recognize integrin subunits that are bound to their ligand(s) at the cell surface (Visconti, unpublished observations).
Laminin has been shown to play an important role in the formation of other embryonic organs, particularly lung (Klein et al., 1990; Schuger et al., 1990; Thomas and Dziadek, 1994), salivary gland (Kadoya et al., 1995), and kidney (Sorokin et al., 1992; Vanden-Heuvel and Abrahamson, 1993). In all of these studies, secretion of the α chain is the critical factor for branching morphogenesis. These studies point to the role of laminin in development not only as a permissive adhesion substrate, but also as a signaling molecule for events such as branching morphogenesis and cell division (Schuger et al., 1997; Stahl et al., 1997). The role of integrin receptors has also been considered (de Curtis and Reichardt, 1993; Kadoya et al., 1995; Wu and Santoro, 1996; Stahl et al., 1997). These studies, as well as the current work, raise the question whether laminin is needed as a support for the developing epithelium and acts mechanically or whether it is involved in a signaling pathway. Otic morphogenesis, in contrast to other epithelially derived organs, seems to be independent of intracellular events (see Hilfer et al., 1989). ECM interactions in this region may act to stabilize the primordium as it is shaped by events that occur in surrounding tissues. A working model for the invagination of the otic primordium would include the following. Step 1: Palisading of the presumptive otic epithelium. Step 2: Fusion of the basal laminae of the otic primordium and the neural tube. This step would seem to be absolutely required for subsequent morphogenesis of the otic pit. This attachment of the otic primordium to the lateral wall of the rhombencephalon serves to stabilize the primordium as it is shaped by other developmental events. Step 3: Folding and shaping of the otic primordium in response to extrinsic forces. These events include (1) ventral and caudal migration of neural crest cells that may contribute to rostral and caudal fold formation; (2) epiboly of the head ectoderm, which may contribute to dorsal fold formation; and (3) expansion of the head mesenchyme, which seems to be another important effector of otic fold formation. This study and others (Gerchman et al., 1995) suggest that inhibition of mesenchymal expansion in the region ventral to the otic primordium prevents or delays ventral fold formation as well as deepening of the primary longitudinal fold. In addition, head mesenchyme has been shown to migrate into the region between the primordium and the neural tube as they begin to separate. This migration may cause the dorsal reorientation of the otic pit and likely contributes to shaping of the neural tube at these stages. Although intracellular events do not seem to be the major influences of otic morphogenesis, they are likely responsible for maintenance of shape changes in the otic primordium effected by morphogenetic events in the surrounding tissue space.
This study provides experimental proof that the close association of the basomedial surface of the otic primordium with the lateral surface of the neural tube is an essential event in the early morphogenesis of the otic vesicle. This investigation further shows that this interaction is mediated through specific components of the extracellular matrix, particularly laminin. When the extracellular matrix in this region is perturbed, association of these epithelial layers is interrupted and subsequent fold formation in the primordium is inhibited. This result is most likely due to loss of stability in the region of the primordium where contact with the neural tube has been inhibited. As a result, external pressures and internal forces do not have a fixed region against which to exert pressure. Although the role of the cytoskeleton may seem to be less important in otocyst formation, it should not be dismissed. The extracellular matrix may function in a number of ways in this region: it may be responsible for mediating the association of the otic primordium with the neural tube so that intracellular events or changes in surrounding tissues that alter the morphology of the primordium, are stabilized or, conversely, elicit cytoskeletal changes in response to signals triggered by events in surrounding tissues. Events on both sides of the cell membrane are integrated by transmembrane receptors. In vivo experimental evidence indicates that an intact and well-organized extracellular matrix between the basal surfaces of these epithelia is an important factor in the close association of the medial surface of the otic primordium with the lateral wall of the neural tube during early development. It is likely that this association stabilizes the primordium in the region of contact. It is reasonable to speculate that this stabilization influences the manner in which pushing forces from surrounding tissues effect changes in the shape of the primordium.
White Leghorn eggs (Truslow Farms, Chesterton, MD) were incubated in a commercial egg incubator to provide embryos from stage 10 to 15 (Hamburger and Hamilton, 1951).
Microinjection of Antibodies
Monoclonal antibodies to various avian extracellular matrix components and integrin subunits were microinjected according to procedures described previously (Gerchman et al., 1995). Briefly, chicken embryos were microinjected at Hamburger and Hamilton stage 10 by using a General Valve Picospritzer (Fairfield, NJ). Embryos were visualized by the method of Sandor (1968): a suspension of India ink and acetone-washed agar in Hanks' balanced salts solution (HBSS) was injected with a 1-cc syringe and 22.5-gauge needle under the embryo to provide contrast. This method facilitated easy localization of the desired injection site. Unbeveled microneedles with a tip diameter of 5.0 to 7.0 μm were pulled from 1.0-mm electrode glass. Antibody solution was delivered by using compressed air at 10 psi. The antibodies were microinjected under the head ectoderm into the mesenchyme just anterior to the cranial margin of the otic primordium and immediately lateral to the junction of rhombomeres 3 and 4. The microneedle was inserted into the tissue at a 45 degree angle, orientated along the longitudinal axis of the embryo. This orientation ensured that the direction of flow of the injected antibody would be in a posterior direction along the basal surfaces of the otic primordium and the neural tube. Volume of injection was estimated by collecting 10 ejections in a calibrated 1.0-μl Microcaps pipette (Drummond Scientific, Broomall, PA), determining this volume and averaging. To quantitate the amount of antibody injected, this volume was multiplied by the concentration of IgG in solution as determined by enzyme-linked immunosorbent assay. Volume of injection was controlled by regulation of injection duration. Previous studies in our laboratory (Gerchman et al., 1995) demonstrated that injection volume affects the distance to which macromolecules diffuse along the anterior-posterior axis. In the case of immunoglobulin injection, we considered that the IgG would be titrated out of the solution phase by the presence of a high local concentration of cognate ligand. Therefore, we chose to deliver larger amounts of antibody by increasing the injection volume rather than adjusting the concentration of the injection solution. HBSS (Sigma Chemical Company, St. Louis, MO) and nonspecific mouse IgG (Sigma Immunochemicals, St. Louis, MO) were used in control injections. Coordinate volumes of control reagents were microinjected for each experiment.
Three monoclonal antibodies that are immunoreactive with the β1 integrins were chosen: the CSAT (Buck et al., 1986) and JG22 antibodies (Greve and Gottlieb, 1982) are immunoreactive with the extracellular portion of the β1 integrin complex. Although not epitope mapped, they have been shown not to recognize the same peptide sequence. Both antibodies have been demonstrated to perturb the binding of cells to substrate in vitro (Greve and Gottlieb, 1982). The third anti-integrin β1 antibody used in this study, V2E9, has been shown to bind to the cytoplasmic region of the β1 integrin (Hayashi et al., 1990). This antibody served as a control; the presence of intact cell membranes in living, cultured embryos would render the cognate epitope unavailable for binding by the antibody. Therefore, it was considered to be a reasonable control reagent to demonstrate the specific effects of microinjection of CSAT and JG22 antibodies on otic primordium morphogenesis. Monoclonal antibodies to chick laminin (31-2: Bayne et al., 1984), fibronectin (B3/D6: Gardner and Fambrough, 1983), integrin β1 subunit, extracellular domain (JG22: Greve and Gottlieb, 1982), integrin β1 subunit, cytoplasmic domain (V2E9: Hayashi et al., 1990), and integrin α6 subunit (P2C62C4: Bronner-Fraser et al., 1992) were obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, and the Department of Biology, University of Iowa, Iowa City, IA, under contract N01-HD-2-3144 from the NICHD. Murine anti-avian perlecan (a basement membrane heparan sulfate proteoglycan) was the gift of Dr. John Hassell, University of Pittsburgh, Pittsburgh, PA. CSAT (Buck et al., 1986) was the gift of Dr. Clayton Buck, Wistar Institute, Philadelphia, PA. Antibodies to collagen IV were not injected due to the inaccessibility of collagen IV to antibody binding as a result of masking by laminin, fibronectin, entactin, and heparan sulfate proteoglycan (Inoue, 1989; Hilfer and Randolph, 1993). Primary antibodies to avian laminin were localized by using fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). The embryos were cultured on the yolk in 100 mm × 20 mm plastic culture dishes in a humidified, 5% CO2 incubator for 22 hr after injection.
Processing for Scanning Electron Microscopic Viewing
Control and experimental embryos were removed from the yolk on filter paper rings and transferred to HBSS before fixing. The embryos were fixed for 20 min at room temperature in freshly prepared phosphate-buffered 2.0% glutaraldehyde, pH 7.4, washed in phosphate buffer, and postfixed in 1.0% osmium tetroxide in phosphate buffer. Embryos were transected through the transverse midline of the otic primordia and removed from the membranes with a sharpened tungsten needle, dehydrated in a graded series of acetone, critical point dried under liquid CO2, and sputter coated with gold (Seevac, Pittsburgh, PA). Whole-mount and transected embryos were viewed in a Philips (Mahwah, NJ) 501 scanning electron microscope.
Whole-mount embryos were processed for immunofluorescence microscopy to localize the site of primary antibody binding on a Nikon (Melville, NY) Optiphot microscope equipped for epifluorescence and for laser scanning confocal microscopic (LSCM) image analysis on a Leica (Deerfield, IL) TCS 4D confocal microscope. Two dozen stage 10 embryos were microinjected with undiluted hybridoma culture supernatant with an antibody concentration of 2 μg/ml and incubated on the yolk for 30 min. Embryos were removed from the yolk on filter paper rings and fixed for 20 min in fresh phosphate-buffered 3.0% paraformaldehyde. The embryos were washed in 0.1 M phosphate buffered saline (PBS) for 20 min, removed from the membranes with sharpened tungsten wires, and then permeabilized with two 30-min washes in ice cold (−20°C) absolute methanol. The embryos were rehydrated at 4°C in a graded series of ethanol/PBS solutions. The embryos were subsequently washed in 0.1 M PBS containing 3.0% goat serum albumen (GSA) for 6–8 hr at 4°C to block nonspecific binding. Next, the embryos were incubated in a solution of FITC-conjugated goat anti-mouse IgG in 0.1 M PBS containing 0.2% GSA overnight at 4°C. The embryos were washed for 1 hr in 0.1 M PBS before mounting under 22 mm × 60 mm glass coverslips (Thomas Scientific, Philadelphia, PA). The embryos were mounted in glycerol-based anti-photobleach mounting medium. The edges were sealed with clear fingernail polish before viewing. Photomicrographs were produced on Kodak TRI-X Pan film with the Nikon microscope. Optical sections were stored to the confocal workstation, and projection plates were produced on Kodak Ektachrome Elite 100 film.
Light photomicrographs of postfixed embryos immersed in 0.1 M PBS were produced on Kodak Technical Pan film by using a Wild (Heerbrugg, Switzerland) dissecting microscope.
Analysis of Defects
Scanning electron photomicrographs of embryos were printed at the same magnification. They were examined in a blind manner by the senior author, and each side was scored for the presence or absence of the various morphologic features characteristic of normal otic primordia. The degree of abnormality was divided into four classes as described in the Results section and compiled in Table 1.