Developmentally regulated expression of α₆ integrin in avian embryos

During development, interactions between cells and extracellular matrix molecules are thought to play an important role in many morphogenetic processes. Numerous extracellular matrix molecules (ECM) including fibronectin, laminin, collagens, and proteoglycans are present in regions of active cell movement and neurite outgrowth. The best-studied ECM molecules are the glycoproteins, fibronectin and laminin, which are widely distributed and appear to be developmentally regulated. Some of these molecules have estabfished functions in development; for example, fibronectin and/or laminin are required for gastrulation (Boucaut et al., 1984), cranial neural crest cell migration (Poole and Thiery, 1986; Bronner-Fraser and Lallier, 1988) and neurite outgrowth (Lander et al., 1985).

The integrins are a family of cell surface receptors that mediate adhesion of cells to a variety of extracellular matrix molecules including fibronectin, laminin, tenascin and various collagens (Horwitz et al., 1985; Buck et al., 1986; Tomaselli et al., 1988; Bourdon and Ruoslahti, 1989). Integrin receptors are heterodimers of two non-covalently linked transmembrane glycoproteins, one α and one β subunit. Within the integrin family, there are at least fourteen α subunits, each with significant sequence identity, and at least eight β subunits. Some integrin heterodimers recognize a single ligand specifically whereas others bind to several different extracellular matrix molecules. For example, α₆β₁ ft only binds to laminin (Sonnenberg et al., 1988; 1990) whereas α₃β₁ recognizes fibronectin, laminin and collagens (Wayner et al., 1988; Gehlsen et al., 1988; 1989; Tomaselli et al., 1990).

Although the structural and biochemical characteristics of various integrin heterodimers have been well-studied, little is known about their tissue-specific distribution or function during development. The β₁ subunit is present on most embryonic cell types at early stages of development (Duband et al., 1986; Krotoski et al., 1986) and antibodies against the β₁ subunit interfere with cranial neural crest cell migration (Bronner-Fraser, 1985; 1986a) and myoblast migration (Jaffredo et al., 1988). In avian embryos, this suggests a functional role for integrins during cell migration. However, because the β₁ subunit can pair with multiple α subunits, it remains unclear if single or multiple integrins are involved in these morphogenetic events. To determine which of the possible integrin hetero-dimers are present in the early embryo, a-specific probes are required. Unfortunately, the numerous antibodies that recognize human α subunits do not cross-react well with other species. Furthermore, few α- specific probes are available for avian embryos (de Curtis et al., 1991; Muschler and Horwitz, 1991), in which perturbations experiments have indicated a functional significance for integrins. In the present study, we utilize a monoclonal antibody that recognizes the avian α₆ subunit of integrin to describe its distribution in the developing embryo. Our results show that it appears in neuroepithelial cells, subsets of neurons in the developing nervous system, muscle cells and their precursors in a developmentally regulated fashion.

A monoclonal antibody against the chick α₆ subunit of integrin was prepared by immunizing nine week old female Balb/c mice at 21 day intervals with approximately 50 μg of integrin α subunits purified from adult chicken brain tissue. The chicken integrins were purified using a CSAT antibody (against the β₁ subunit) affinity column, as described by Muschler and Horwitz (1991). The dimeric receptors were denatured in 6 M guanidine isothiocyanate, dialyzed against 200 volumes of extraction buffer and passed several times over a W1B10 (anti-β₁) column to deplete the β subunit. For immunizations, the sample was mixed (1:1) with Freunds adjuvant (Gibco Laboratories; Life Technologies Inc, Grand Island, NY) to a final volume of 0 5 ml/mouse and injected into the peritoneal cavity. The fusion was performed by the methods of Kennett et al. (1982). One hybridoma was identified as specific for a 130 × 10³M_r integrin α subunit and was subsequently designated α₆C6. Having been generated against denatured protein, this antibody immunoblots well and immunoprecipitates the monomeric α subunit after the dimeric receptor has been denatured, but does not precipitate the native receptor Verification that the α₆C6 recognizes the chick subunit of integnn was demonstrated by transfecting NIH3T3 cells with chicken α₆ cDNA (kindly provided by Dr. L. Reichardt; de Curtis et al., 1991) followed by immunoblots with the α₆C6 antibody The methods describing the ex-pression vector, transfection, cloning and protein extraction from NIH 3T3 cells are as described in Hayashi et al. (1990).

The 130 × 10³M^r chicken α subunit was purified from a CSAT affinity column by denaturing in 6 M guanidine isothiocyanate, heating to 60°C for 15 minutes, dialyzing, and then incubating with α₆C6 monoclonal antibody conjugated to Sepharose beads The beads were washed in extraction buffer and eluted with SDS-PAGE sample buffer. CSAT, W1B10 and α₆C6 matrices were constructed by coupling 5 mg/ml of pure monoclonal antibody to CNBr-activated Sepharose 4B (Piscataway, NJ).

SDS-PAGE was performed by the method of Laemmh (1970). Separating gels were 7% acrylamide and 0.12% bisacrylamide. Samples were prepared in sample buffer (62.5 mM Tris-HCl, pH 6 8, 2.0% SDS, 10% glycerol). Proteins were visualized by silver staining using Bio-Rad Silver Stain kit (Bio-Rad, Richmond, CA). Immunoblots were performed as described in Muschler and Horwitz (1991). All antibodies were used at an approximate IgG concentration of 25 μg/ml.

The CSAT antibody was prepared as described by Neff et al. (1982). The W1B10 antibody was prepared as described by Hayashi et al. (1990). Rabbit anti-human α₆ and α₅ cytoplasmic domain serum were a gift of Dr. L. Reichardt (UCSF; Tomaselli et al., 1988; de Curtis et al., 1991). The JG22 antibody against the chick β₁ subunit of integrin was purchased from Developmental Studies Hybridoma Bank. An antibody against chick laminin (LM-33) which recognizes the B chains (Krotoski et al, 1986) was kindly provided by Dr. Douglas Fambrough. A polyclonal antibody against mouse laminin was purchased from E-Y laboratories. An antibody against a nonphosphorylated epitope of rat neurofilament was kindly provided by Dr. Virginia Lee (Lee et al., 1987).

White Leghorn chick embryos ranging from Hamburger and Hamilton (1951) stage 8 (embryo day 1 5) to stage 35 (embryonic day 9) were obtained by incubating eggs in a forced air incubator at 37°C. Some tissue also were obtained from adult chickens

Embryos were fixed in methanol at 4°C overnight, washed in PBS, transferred to 5% sucrose in PBS with azide for 4-24 hours, and placed in 15% sucrose in PBS with azide overnight at 4°C Embryos were placed in 7% gelatin (Sigma; 300 Bloom) m 15% sucrose/PBS for 3 hours at 37°C, and embedded in fresh gelatin Embryos were stored for up to two weeks in the refrigerator until the time of sectioning when they were rapidly frozen in liquid nitrogen. 14 μm sections were cut on a Zeiss Microm cryostat and mounted on gelatin subbed slides.

Primary neural crest cultures were prepared from the neural tubes of Japanese quail embryos (Coturnix coturnix japonica). Embryos were incubated for 48 hours, at which time their developmental age was comparable to that of chick stages 13 - 15. The region of the trunk neural tube adjacent to the six to nine most posterior somites as well as the unsegmented mesenchyme was dissected away from the embryo. The neural tubes were isolated from adjacent tissue by treatment with 160 units/ml of collagenase (Worthington Biochemical, Freehold, NJ), followed by trituration. After stopping the enzymatic reaction with MEM containing 15% horse serum and 10% chick embryo extract, neural tubes were plated onto fibronectin- or laminin-coated culture dishes. After several hours, the neural crest cells migrated away from the explant and the neural tube tissue was removed Neural crest cells were grown for several days prior to fixation in cold methanol for 15 minutes.

Cryostat sections of embryos were warmed to 25°C and adjacent sections were incubated with either α₆. neurofilament, β₁ or laminin antibodies. For staining neural crest cultures, dishes were incubated with anti-α₆ antibody as described for sectioned tissue. All antibody dilutions were made in 0.1% bovine serum albumen in PBS Anti-α₆ was used at a concentration of 35 μg/ml. Anti-neurofilament was diluted 1:400 from hybridoma supernatant. JG22 and LM-33 were used as straight hybridoma supernatant. The rabbit polyclonal antibody against laminin was diluted 1:50. Following an overnight incubation, sections were rinsed in PBS, and incubated with a highly fluorescent FITC-goat antibody against mouse Igs (Antibodies Inc., Davis, Ca.; diluted 1:300) for detection of monoclonal antibodies for 1 hour at 25°C. Slides then were rinsed in PBS, coverslipped with glycerol (90% glycerol/10% 05 M sodium carbonate, pH 7.8) and viewed with an Olympus Vanox epifluorescence microscope. Data were recorded photographically. For control slides or cultures, the primary antibody was omitted and the sections or dishes were stained with fluorescinated second antibody alone.

The distribution of the chick α₆, subunit of integrin was examined in embryos ranging from 1.5 to 14 days of incubation, as well as some adult tissues. Its expression pattern was compared with that of the β₁ subunit of integrin, neurofilament proteins and laminin, the extracellular ligand of α₆β₁ integrin.

A monoclonal antibody specific for a 130 × 10³M_rchicken integrin α subunit was generated by immunizing mice with purified integrins from chicken brain tissue. Immunofluorescent staining of adult tissue sections using this monoclonal antibody, α₆C6, demon-strated that the antigen was localized on neural tissue and on epithelial cells, where it was concentrated on the basal surface contacting the basal lamina (not shown). This localization is similar to that of the human α₆ subunit. To further test homology of the α₆C6 antigen with human α₆, we utilized an antiserum that recognizes a 35 amino acid polypeptide corresponding to the cytoplasmic domain of the human subunit (de Curtis et al., 1991). This anti-α₆ antiserum exclusively recognized the 130 × 10³M_ra band purified by the α₆C6 antibody (Fig. 1). The 160 × 10³M_r band in lane 2 of Fig. 1 is monoclonal antibody released from the beads used for immunoprecipitation.

To demonstrate that the α₆C6 monoclonal specifically recognizes the chick subunit of integrin, cDNA encoding for chick (% (de Curtis et al., 1991) was expressed in NIH 3T3 cell lines (Hayashi et al., 1990). Extracts of the transfected and untransfected cells were immunoblotted with the α₆C6 antibody. The antibody clearly recognizes a 140 × 10³M_r protein product in the α₆ transfected cells which is absent in the untransfected cells (Fig. 2). The gene product in transfected cells migrates slightly larger than α₆, purified from chick tissue (140 compared with 130 × 10³M_r). This difference is likely to be due to differential glycosylation. These results suggest that by immunological relatedness, similar tissue distribution and recognition of the gene product o0066 chick α₆ cDNA, the 130 × 10³M_rα subunit recognized by α₆C6 is the chicken homolog of human α₆. Therefore, this antigen will be referred to as the chicken α₆ subunit.

In the developing head, the α₆ subunit of integrin was observed on neuroepithelial cells of the neural plate in stage 8 (1.5 day) embryos. Its expression continued during neural tube closure and was also observed on premigratory neural crest cells (Fig. 3). Interestingly, the immunoreactivity appeared to drop off gradually caudal to the hindbrain and was not observed in trunk neuroepithelial tissue until after neural tube closure. α₆ expression was not observed outside of the forming nervous system in neurulating embryos.

In the trunk region of a 2 day chick embryo (stage 15), the onset of α₆ immunoreactivity appeared to follow an anterior to posterior sequence, mirroring the developmental progression of the embryo. Because neurulation begins in the head and moves progressively tailward, several stages of α₆ expression could be viewed simultaneously in a single embryo. Within the somites, the posterior extent of its expression was detected approximately five somites above the last somite, at a level where the epithelial somites undergo a transition to form the dermomyotome and sclerotome. In transverse sections, expression can first be seen on a few presumptive myotomal cells in the portion of the somite closest to the neural tube (Fig. 4A) within the newly formed dermomyotome. Approximately 5 - 6 somites above the last-formed somite, a subpopulation of immunoreactive cells is observed in the anterior portion of the somite just medial to the dorsal neural tube (Fig. 4E). The number of α₆-positive cells increases and these cells appear to progressively fill in the space between the dermatome and sclerotome, in an anterior to posterior sequence (Fig. 4E), such that only 1 - 2 somite lengths contained ‘partial’ α₆immunoreactive myotomes. In slightly more mature somites located in more anterior (older) regions of the same embryos, α₆ reactivity within the somite appears restricted to myotomal cells (Fig. 4C,F).

Concomitant with its appearance in the myotome, α₆ becomes apparent in the trunk neural tube and the intensity of immunoreactivity increases in progressively more anterior regions of the embryo. Some α₆ reactivity is detectable in the surface ectoderm and in ventral regions of the embryo where the mesonephros begins to form.

When the neural tube is a single cell layer thick, α₆ appears to outline the columnar neuroepithelial cells of the tube. Its expression pattern in the neural tube overlaps with that of β₁integrin (Fig. 4B). Though neurons are morphologically distinguishable by neurofilament staining at this stage (Fig. 4D), levels of α₆ are not elevated on neurons compared with undifferentiated neuroepithelial cells until a slightly later stage.

In anterior (pharyngeal) regions of the stage 15 - 16 embryo or in the trunk of stage 18 - 19 embryos, the pattern of immunoreactivity changes within the neural tube. Although low levels of reactivity remain detectable on neuroepithelial cells, some subpopulations of cells within the neural tube express particularly high levels of α₆. The most notable are the cells with the morphological appearance of commissural neurons which extend their axons toward the floorplate (Tessier-Lavigne et al., 1988). The cell soma as well as axonal processes are outlined by α₆ immunoreactivity (Fig. 5A,C). Some of these processes appear to project toward the laminin-rich basement membrane (data not shown) along the lateral margin of the neural tube. Some motor neurons with axons extending into the sclerotome also appear to express α₆ (Figs 4F, 5A, 7A). This is particularly clear when viewed in longitudinal section (Fig. 4F), where its expression can been seen on processes and/or Schwann cells emerging from the ventral neural tube into the anterior portion of the sclerotome. Laminin is present surrounding the neural tube and in the sclerotome through which the axons project (data not shown; and Krotoski et al., 1986).

The pattern of α₆ expression within the neural tube changes with progressive development. High levels of reactivity continue to be observed on neuroepithelial cells in the germinal region of the neural tube, toward its lumen (Figs 6A, 7A, 8A). Immunoreactivity becomes reduced in the region of the floorplate (Figs 8A, 9A), but remains high in regions that are neurofilament-positive as well is near the lumen (Fig. 8D). By stage 25, immunoreactivity is apparent on cell bodies and processes within both the neural tube and dorsal root ganglia (Fig. 8A.B). In contrast to α₆ expression, β₁ immunoreactivity is only present at low levels in the ventrolateral regions of the neural tube containing neuronal cell bodies and processes, but remains high on neuroepithelial cells. In the developing brain, the pattern of α₆ expression is similar to that observed in the developing spinal cord. It also is observed in the optic cup of the 2 day old embryo, and later on the ganglion cell layer of the neural retina during embryogenesis and into adulthood.

Throughout the nervous system, levels of α₆ immunoreactivity remained relatively unchanged in distribution and intensity through stage 28 (day 6). By stage 30 (day 7), however, the intensity of staining appeared to diminish. In embryos examined at stage 33, the levels of reactivity were markedly lower than those observed at stage 30 (Fig. 9). This trend continued such that by stage 37, the latest stage examined, the neural tube contained little α₆ immunoreactivity with the exception of the germinal region adjacent to the lumen (Fig. 10A). Interestingly, the ventral roots continued to express high levels at these late stages, suggesting that this subunit of integrin is expressed on axons and/or Schwann cells, though only low levels of immunoreactivity are observed on motor neuron cell bodies (Fig. 10A), α₆ is also prevalent on the sciatic nerve from day 11 to adulthood. Most of the regions containing immunoreactivity are adjacent to regions containing high levels of laminin (Fig. 9C), though the distributions of the two molecules do not always overlap.

Faint immunoreactivity for the subunit of integrin is visible on early migrating neural crest cells within the cell-free space between the neural tube, ectoderm and somites in the trunk region (Fig. 4C). Slightly later in development, trunk neural crest cells enter the somitic sclerotome, where they migrate through the anterior but not the posterior half of this tissue (Rickmann et al., 1985; Bronner-Fraser, 1986b). After entering the anterior half of the sclerotome, α₆ immunoreactivity is no longer detectable on neural crest cells (Fig. 5A), identified on the basis of their HNK-1 immunoreactivity though these cells express abundant β₁ integrin immunoreactivity (Fig. 5B). With further development, neural crest cells reach the level of the dorsal aorta, where they will differentiate into catecholamine-containing cells of the sympathetic ganglia, adrenal medulla and aortic plexuses. α₆ immunoreactivity is again detectable on the subpopulation of neural crest cells at the level of the dorsal aorta (Fig. 5A), before these cells express catecholamines.

The dorsal root ganglia begin to aggregate at stage 19 - 20 in the chick embryo (Lallier and Bronner-Fraser, 1988). At the level of the forelimb bud, low levels of immunoreactivity are visible on the forming dorsal root ganglion cells (Fig. 6A), though the levels are far less prevalent that those of the β₁ subunit (Fig. 6B). In contrast, immunoreactivity for both the α₆ and β₁ subunits are high around the dorsal aorta by stage 19 (Fig. 5A,B), though the sympathetic ganglia do not aggregate until stage 20 - 21 (Lallier and Bronner-Fraser, 1988). By this stage, prominent α₆ and β₁ immunoreactivities are observed in the ventral roots (Fig. 6A,B). This immunoreactivity could be on axons and/or Schwann cells (Fig. 6A).

The relative levels of ay expression remain constant on forming sympathetic ganglia from stage 19 to stage 28. However, the levels of expression of dorsal root ganglia increase dramatically after ganglion formation. While α₆ immunoreactivity is faint at stage 21 (Fig. 7A), it increases to become comparable in intensity to that observed on the sympathetic ganglia by stage 24 to 25 (Fig. 8A). The expression of α₆ appears to be associated with both cell bodies and processes within the ganglia (Fig. 8D). At these times, there is prominent laminin immunoreactivity surrounding the ganglia and their processes (data not shown).

In embryos observed between stages 30 and 37, α₆ expression remains detectable on dorsal root and sympathetic ganglion cells. However, the levels of immunoreactivity decline relative to those observed at earlier stages. For example, the relative level of immunoreactivity on dorsal root ganglion cells at stage 33 (Fig. 9A,B) is less than that observed at stage 25 (Fig. 8A). In the stage 37 embryo, α₆ immunoreactivity is present on cells of the dorsal root ganglion (Fig. 10A), but of lower intensity than that observed either in the ventral roots or on dorsal root ganglion cells of stage 25 embryos.

Neural crest cells were grown in tissue culture on fibronectin- or laminin-coated dishes for 1-7 days prior to fixation and staining. Similar results were found for both substrates and the results will be presented collectively. In 1 to 2 day old cultures, little or no α₆ immunoreactivity above background was observed on neural crest cells. By 3 days in vitro, expression was observed in a subpopulation of neural crest cells. The immunoreactivity appeared to be punctate in the cytoplasm, as well as outlining the cell periphery, suggesting that the integrin is present in cytoplasmic pools as well as on the cell surface. A similar pattern was maintained throughout the culture period examined. The α₆ immunoreactive cells represented a significant subpopulation of cells in these neural crest cultures, comprising approximately one quarter of the total cell number.

After the differentiation of the somite into dermatome, myotome and sclerotome, α₆ is only detectable on myotomal cells (Fig. 4C). In the trunk, immunoreactivity remains on myotomal cells as long as the myotome is visible (Figs 5A,7A,8A). In the limb of a stage 21 embryo, no ay reactivity is detectable (Fig. 7B) with the exception of a few positive cells beginning to enter the limb from the base of the dermomyotome (Fig. 7A). The numbers of immunoreactive cells in the limb increase with time. By stage 25, the limb is populated by numerous α₆-reactive cells in regions where myogenesis will occur (Fig. 7C). One interpretation of these results is that the presumptive α₆-bearing myoblast cells derived from the myotome invade the limb and proliferate, while continuing to express the antigen.

As the myotubes differentiate, α₆ is expressed at high levels on early myotubes. Its is expressed in skeletal muscle of the limb and body wall and in intervertebral muscle (Fig. 10B) throughout the embryonic stages examined, but does not appear to be concentrated at the neuromuscular junction. In contrast, the presence of α₆ on adult muscle is ambiguous. Laminin immuno-reactivity is prevalent in regions containing myotomal cells, myoblasts and differentiated muscle at all times examined (data not shown).

Although low levels of immunoreactivity can be detected on ectodermal cells as early as stage 15-16, staining becomes more intense at approximately stage 21 compared with earlier times. Particularly bright reactivity is observed on the apical ectodermal ridge (Fig. 7B) and in the developing mesonephric rudiment. By stage 21, intense reactivity is present in the mesonephric region (Fig. 7A). The developing gonads have high levels of expression, most prominent in the cortical region. However, the germ ceils themselves are α₆ negative. β₁ and laminin immunoreactivities are observed in the medulla of the gonads but not the cortex (data not shown). β₆ immunoreactivity remains associated with the developing kidney tubules through development and into adulthood. It is also observed at low levels on the lung and some other endodermal organs. For example, it is found on intestinal epithelium from embryonic day 14 through adulthood, where it is concentrated on the basal cell surface.

We have examined the distribution of the α₆ subunit of chick integrin in the early embryo. Because the expression of this subunit overlaps with that of the β₁ subunit in most tissues examined, it is likely that α₆β₁ heterodimers are formed in the majority of the tissues expressing α₆. α₆β₁ heterodimers function as laminin receptors (Sonnenberg et al., 1988; 1990; Hall et al., 1990) and laminin is prominent in the basal lamina and in some interstitial matrices such as in the sclerotome at the stages examined in this study (Krotoski et al., 1986). Thus, it appears likely that the α₆β₁ receptor complex is expressed in numerous sites in the vicinity of its natural ligand. However, α₆ can pair with the β₄ subunit of integrin (Sonnenberg et al., 1988). The ligand for α₆β₄ has not been defined, nor is the distribution of this β subunit known in the embryo. Therefore, we cannot rule out the possibility that α₆ also forms heterodimers with the β₄ subunit. In the neural tube, laminin is absent, levels of β₁ are low and levels of the α₆ subunit are high; therefore, it is tempting to speculate that α₆ may pair with a different β subunit there. At the stages examined in our study, immunoreactivity for the β₁ subunit of integrin is present in most parts of the embryo. An exception is in the germinal ridge of the gonads. In the cortical region, α₆ immunoreactivity is high whereas β₁ is absent. In the medullary region of the gonads, both subunits are present. Interestingly, laminin is present in the medullary but not cortical portion of the gonad. Thus, it is conceivable that α₆ pairs with a different β subunit, such as β₄, in this organ.

As expected, β₁ immunoreactivity is present in numerous regions which lack α₆ reactivity. These include dermatomal cells, sclerotomal cells, migrating neural crest cells and notochordal cells. Interestingly, all of these cells are present in regions containing high concentrations of laminin, fibronectin, collagens and proteoglycan (Krotoski et al., 1986; Duband and Thiery, 1987; Perris et al., 1991a,b). For example, lamiriin is prominent around the neural tube, dermo-myotome, under the ectoderm and throughout the somitic sclerotome (Krotoski et al. 1986). Therefore, these tissues are likely to express other β₁ integrin heterodimers for recognition of extracellular ligands. Numerous integrins that interact with laminin have been described. In addition to α₆β₁ laminin is recognized by α₁β₁ (Ignatius and Reichardt, 1988), α₂β₁ (Languino et al., 1989) and α₆β₁ ft (Wayner et al., 1988). Multiple integrins that bind to fibronectin and collagens have been described as well (Pytela et al., 1985; Turner et al., 1987; Wayner and Carter, 1987; Edwards et al., 1987; 1988; Gailit and Ruoslahti, 1988; Kirchhofer et al., 1990; Elices et al., 1991; Guan and Hynes, 1990; Mould et al., 1990; Dedhar and Gray, 1990; Vogel et al., 1990; Charo et al., 1990; Gullberg et al., 1990; Staatz et al., 1989; 1990).

Due to the paucity of available markers for α subunits, little is known about the distribution of other integrin heterodimers in the early embryo. In a recent study examining the distribution of the 05 subunit of integrin, Muschler and Horwitz (1991) found that α₅ has a widespread distribution in the early embryo, being associated with endothelial and mesenchymal cells at stages when reactivity is highest. α₅ immunoreactivity is persistent at later developmental stages (E10 and beyond), with a widespread distribution on a myriad of cell types. In contrast, it appears to be absent from most cells in the adult. Taken together with the present study, these results suggest that integrins are developmentally regulated with different α subunits being expressed at different times and on selected populations. In contrast to the α subunits, expression of β₁ appears to be more uniform both spatially and temporally. It is likely that regulation of heterodimers occurs by altering a expression with the β expression remaining constitutive.

α₆ expression on presumptive myotomal cells begins just as the epithelial somite undergoes an epithelial/mesenchymal transition to form the dermomyotome and sclerotome. The appearance of α₆ is coincident with that of earliest myotomal markers, such as acetylcholin-esterase (Layer et al., 1988; Kaehn et al., 1988). By examining its expression as a function of time, one can infer the pattern of myotome formation. A subpopulation of immunoreactive cells first appears in the anterior and medial portion of the somite. The number of immunoreactive cells increases to rapidly fill in the space between the dermatome and sclerotome, in an anterior to posterior sequence. Partially formed α₆-reactive myotomes only were observed along 1-2 somite lengths and each somite takes approximately 110 minutes to form. This suggests that myotomal precursors arise and/or migrate between the dermatome and myotome during a 4 hour time interval. This migration pattern is consistent with that inferred from observations of acetylcholinesterase staining of the forming myotome (Layer et al., 1988; Kaehn et al., 1988). Interesting, antibodies against the β₁ subunit of integrin have been shown to disrupt trunk myoblast migration into the body walls and limbs (Jaffredo et al., 1988). Because α₆ codistributes with β₁ in the myotome, the receptor disrupted by the injected antibody is likely to be the α₆β₁ heterodimer, α₆ expression persists on myoblasts as they populate the limb. It is present on myotubes and differentiated muscle throughout the embryonic stages examined. In turn, the muscles are surrounded by a laminin-nch matrix.

In the neural tube, α₆ is expressed on subpopulations of neurons, including commissural neurons and motor neurons. These may extend their axons along the laminin-rich basal lamina on the lateral margin of the neural tube and, in the case of motor neurons, out into the periphery containing interstitial laminin. This raises the possibility that laminin may serve as a permissive substrate for commissural and motor axons. Although α₆ is expressed prominently on numerous neurons and their cell bodies, it is not clear that the expression correlates with initial axon outgrowth. For example, neurofilament-positive cells are observed within cell bodies in the neural tube and on some processes extending out the’ forming ventral root prior to the expression of high levels of α₆ on their processes.

The α₆ subunit is observed on neural crest cells during the initial stages of their emigration from the neural tube. At this time, neural crest cells enter a cell-free space rich in laminin, fibronectin and other matrix molecules (Krotoski et al. 1986; Perris et al., 1991a,b). After migrating further ventrally and entering the cellular environment of the sclerotome, which contains lower levels of ECM molecules, α₆ immunoreactivity is no longer detectable on neural crest cells. It reappears on a subpopulation of neural crest cells at the level of the dorsal aorta. These cells belong to the sympathoadrenal sublineage (Doupe et al., 1985) and will form sympathetic neurons, chromaffin cells and cells of the aortic plexuses. α₆ expression apparently preceeds that of catecholamines which are first observed about a half day later (unpublished observation) and expression is maintained on both primary and secondary sympathetic chains. In contrast, α₆ is expressed on dorsal root ganglion cells only after gangliogenesis.

In a recent report by de Curtis et al. (1991), the expression, regulation and sequence of the avian α₆ subunit of integrin was described. The antibody used in our study recognizes this same subunit. Our immuno-cytochemical study suggests that α₆ integrins are high in the retina as early as the optic cup stage. Furthermore, we observed α₆ expression in the retina through adulthood.

In conclusion, our results show that the α₆ subunit of integrin has a developmentally regulated pattern of expression in the embryo. Notably, it appears in the developing nervous system and in presumptive myo-blasts. In many places, its distribution overlaps with that of the β₁ subunit of integrin as well as with laminin, the natural ligand for α₆β₁ heterodimers. Its expression pattern is dynamic and is down-regulated in the neural tube, but not on muscle, by embryonic day 7. The distribution pattern of this subunit is consistent with a possible role for α₆β₁ heterodimers in numerous morphogenetic processes such as axon guidance and myoblast migration.

We thank Mary Flowers and Kristin Artinger for excellent technical assistance and Dr. Scott Fraser for helpful comments on the manuscript. Anti-human α₅ and α₆ antiserum and cDN A encoding chicken α₆were the generous gifts of Dr. L. F. Reichardt. This work was supported by USPHS HD-15527 to MBF and GM-23244 to AFH.