Expression of α1 integrin, a laminin-collagen receptor, during myogenesis and neurogenesis in the avian embryo

Integrins are a group of non-covalently associated heterodimers of α and β subunits that function in cell-to-cell and cell-to-extracellular matrix adhesions (see, for reviews, Humphries, 1990; Hynes, 1992). Integrins have been classified according to their β subunits, each of which may associate with different α subunits. There are at least eight β1 integrins (α1β1, …, α8β1), that are widely distributed on most cell types. β1 integrins act primarily as cell surface adhesion receptors for extracellular matrix molecules including fibronectin, laminin, collagens and vitronectin. Some β1 integrins apparently bind to a single extracellular matrix ligand, while others are more promiscuous in their ligand recognition. However, members of β1 integrins may also function as cell-cell adhesion molecules (Larjava et al., 1990; Elices et al., 1990; Marchisio et al., 1991; Taichman et al., 1991). Moreover, the ligand-binding specificity of β1 integrins may differ in various cell types at least in vitro, thus generating additional functional diversity within the family (Elices and Hemler, 1989; Languino et al., 1989).

The great diversity of integrin receptors and their versatility in ligand binding raises the problem of determining their exact function in the various adhesive processes that take place during embryogenesis and in the adult organism. Are the different integrins that bind common ligands interchangeable in every tissue or do they show tissue and functional specificities and, thus, participate in defined adhesion events? An approach to evaluate the various functions in which each integrin may play a role is to determine their expression pattern during embryogenesis and in adult tissues. β1 integrins have been found in a large variety of cell types cultured in vitro. In the adult organism, many different tissues express β1 integrins (Bozyczko et al., 1989; Korhonen et al., 1990; Sonnenberg et al., 1990a; Virtanen et al., 1990; Terracio et al., 1991; Hertle et al., 1991) and, in the embryo, most cells express the β1 subunit of integrins (Duband et al., 1986). However, little is known about the tissue distribution and possible developmental regulation of the various α chains.

α1β1 integrin is an RGD-independent receptor for both laminin and collagens (Hemler et al., 1985; Kramer and Marks, 1989; Turner et al., 1989; Belkin et al., 1990; Ignatius et al., 1990; Rossino et al., 1990; Tawil et al., 1990; Tomaselli et al., 1990; Syfrig et al., 1991; Vandenberg et al., 1991). The α1 integrin subunit has been demonstrated on a large variety of cell types cultured in vitro, including neuronal cells (Ignatius and Reichardt, 1988; Turner et al., 1989; Rossino et al., 1990; Tawil et al., 1990; Tomaselli et al., 1990), melanoma cells (Kramer and Marks, 1989), microvascular endothelial cells (Korhonen et al., 1990; Defilippi et al., 1991), smooth muscle cells (Kelly et al., 1987; Belkin et al., 1990; Syfrig et al., 1991), cardiomyocytes (Terracio et al., 1991) and hepatocytes (Forsberg et al., 1990). In contrast, in adult human tissues, α1 integrin has been so far detected only in capillary endothelial cells and in smooth muscle cells (Belkin et al., 1990; Virtanen et al., 1990; Defilippi et al., 1991). In cultured cells, α1 integrin expression can be strongly regulated by nerve growth factor (Rossino et al., 1990), transforming growth factor-β1 (Heino et al., 1989), tumor necrosis factor, retinoic acid and phorbol esters (Defilippi et al., 1991; Rossino et al., 1990, 1991). These observations suggest that α1 integrin might be expressed transiently under the control of growth factors in a number of tissues during embryonic development and might play a critical role during cell differentiation.

In this study, we have analyzed the distribution of α1 integrin during avian development. We show that α1 integrin is expressed at high levels transiently in nervous tissues and striated muscle cells, and permanently in smooth muscle cells and capillary endothelial cells. It is practically absent from other tissues, particularly epithelia. This restricted pattern of distribution is indicative of tightly controlled developmental regulation of α1 integrin expression which correlates with defined morphogenetic events occurring during neurogenesis and myogenesis.

Japanese quail (Coturnix coturnix japonica) animals were used throughout the study. Eggs were incubated at 37 ± 1°C and staged according to the number of somite pairs and the duration of incubation. Adult animals were kindly provided by Dr M.-C. Lainé (Centre de Biologie Cellulaire, Ivry-sur-Seine, France). Late embryos and adult animals were killed by decapitation.

Purification of a p175 membrane protein from chicken gizzard and production of rabbit antibodies directed against this molecule were described previously (Koteliansky et al., 1985). The monoclonal antibody TS2/7 against human α1 integrin was from T Cell Sciences (Cambridge, MA, USA). Polyclonal antibodies to human α1 integrin were raised and characterized previously (Belkin et al., 1990). Purification of the α1β1 integrin complex from chicken gizzard and rabbit sera directed to the natively isolated α1 subunit were described previously (Syfrig et al., 1991). A rat monoclonal antibody, ES46, directed against the chicken β1 integrin subunit was described previously (Duband et al., 1988; Mueller et al., 1988) and kindly provided by Dr K. M. Yamada (NIH, Bethesda, MD, USA). A rabbit polyclonal antibody to mouse laminin was described previously (Duband and Thiery, 1987), and a sheep polyclonal antibody also directed against mouse laminin was a generous gift of Dr H. Kleinman (NIH, Bethesda, MD, USA). Rabbit sera to chicken collagen I and to human collagen IV were purchased from Institut Pasteur-Lyon (France). Two mouse monoclonal antibodies, NC-1, which recognizes migrating neural crest cells and a number of neural tissues, and MB-1, which recognizes quail endothelial cells, were characterized previously (Péault et al., 1983; Vincent and Thiery, 1984). Rabbit polyclonal antibodies to α-smooth muscle actin were purchased from Sigma.

Embryos were routinely fixed at room temperature in 3.7% formaldehyde in phosphate-buffered saline (PBS) for 1 –8 hours, depending on the size of the embryos. After several extensive washes in PBS, embryos were embedded in a graduated series of sucrose solutions in PBS (12-18% wt/vol) and subsequently frozen in Tissue Tek (Lab-Tek Products) in liquid nitrogen. Adult tissues were directly immersed in liquid nitrogen after isolation and immediately processed for histological sectioning. Sections were cut at 7 –12 μm on a cryostat (Bright Instrument Co. Ltd., Huntington, England) and mounted on slides coated with a gelatin solution. Immunofluorescent staining of sections was performed essentially as described previously (Duband and Thiery, 1987). Control stainings were performed using preimmune or non-immune immunoglobulins as primary antibodies. No significant staining was obtained with these control antibodies even in compact tissues. Sections were observed with a Leitz Orthoplan epifluorescent microscope and photographed using TMX-400 Kodak film.

Immunoprecipitation of Triton X-100-extracted proteins from human uterus smooth muscle with the monoclonal antibody TS2/7, or with a polyclonal antibody directed against both human and chicken α1 integrin was described previously (Belkin et al., 1990). For immunoblotting, samples of embryonic or adult tissues were briefly homogenized and extracted at 90°C with SDS-sample buffer under reducing conditions. The extracts were clarified by centrifugation and subjected to SDS-PAGE followed by immunoblotting analysis with antibodies to α1 or β1 integrins. PAGE was performed in Laemmli buffer system on slab 7.5% polyacrylamide minigels. The protein bands were electroblotted for 1 hour onto nitrocellulose in 50 mM Tris-glycine, 20% methanol buffer. The nitrocellulose membranes were then incubated with a 4% bovine serum albumin solution in PBS for 1 hour at 37°C, followed by incubation with antibody solution for 1 hour at room temperature. Control stainings were performed using preimmune or non-immune immunoglobulins as primary antibodies. The sheets were then rinsed in PBS supplemented with 1% Tween 20 and incubated with ¹²⁵I-labelled protein A (Amersham) for 1 hour at room temperature. After rinsing, the blot was subjected to autoradiography. For immunodepletion experiments, 14-day-old embryonic gizzards were extracted in 2% Triton X-100 in PBS supplemented with 1 mM Ca²⁺ and Mg²⁺ and 3 mM phenylmethylsulfonyl fluoride. Lysates were clarified by centrifugation at 11,000 g and mixed three times successively with rabbit serum to the avian α1 integrin subunit or with antibodies to p175 protein at the final dilution of 1/100 followed by incubation with protein A-Sepharose (Sigma) and centrifugation. The samples were then subjected to SDS-PAGE followed by immunoblotting analysis with antibodies to p175 protein or to the avian α1 integrin subunit.

In a previous study, a p175 membrane protein was isolated from chicken gizzard smooth muscles (Koteliansky et al., 1985). The tissular origin, surface localization and biochemical properties of this protein raise the possibility that it may be the integrin α1 subunit. The specificity of rabbit antibodies against the p175 protein was analyzed using several different techniques. First, human uterus smooth muscle membrane extracts were immunoprecipitated either with a monoclonal antibody to human α1 integrin (TS2/7) or a previously-characterized polyclonal antibody against human uterus smooth muscle α1 integrin (Belkin et al., 1990), followed by immunoblotting with antibodies to the p175 protein. As shown in Fig. 1A, both antibodies precipitated a dimer of proteins of approximately 110×10³ and 195×10³M_r corresponding to the β1 and α1 subunits of the human VLA-1 integrin, respectively. On the corresponding immunoblots, the polyclonal antibody to p175 protein specifically stained the upper band of VLA-1 (Fig. 1A). Conversely, antibodies to the p175 protein immunoprecipitated a dimer of proteins of 110×10³ and 195×10³M_r from human uterus smooth muscle protein extracts, and the 195×10³M_r band was recognized on immunoblots by the polyclonal antibody to the human α1 integrin subunit (data not shown). Second, immunoblotting of whole extracts of human uterus smooth muscles with the antibody to p175 protein revealed a single band of about 195×10³M_r and a protein of approximately 175×10³M_r on adult chicken gizzard extract (Fig. 1B). In addition, the antibody to p175 protein recognized on immunoblots the α1 subunit of purified avian α1β1 integrin complex (Fig. 1B). Finally, in order to rule out the possibility that the antibody to the p175 protein might also recognize another integrin α subunit comigrating with the α1 integrin, we performed immunodepletions experiments. First, chicken gizzard extracts were immunodepleted with a recently characterized rabbit antiserum to avian α1 integrin (Syfrig et al., 1991), followed by immunoblotting with the anti-p175 antibody. As shown in Fig. 1C, the antibody to p175 did not detect any band at 175×10³M_r after immunodepletion with the antiserum to avian α1 integrin. Conversely, antibodies to α1 integrin subunit did not detect any band after immunodepletion with the antibody to p175 (data not shown). These results indicate that the anti-p175 protein antibody specifically recognized the α1 integrin subunit in both human and avian species. This antibody was used for studying the spatiotemporal distribution of α1 integrin during avian embryogenesis and in the adult animal by immunofluorescence microscopy of frozen sections and by immunoblotting tissue extracts. The distribution pattern of α1 integrin was compared with those of its putative ligands, laminin and collagens.

As shown in Fig. 1D, α1 integrin was present in large amounts in immunoblots of extracts from the gizzard, intestine, aorta and lung. In contrast, a fainter band corresponding to α1 integrin could be detected in extracts from the heart, skeletal muscles, skin and kidney, and no staining was found in extracts from the brain and liver. Using immunofluorescence labeling, it was found that, in the gizzard, lung, aorta and intestine, only smooth muscle cells were intensely stained for α1 integrin and that, in the other tissues examined, only capillary endothelial cells and smooth muscle cells of blood vessels showed significant immunoreactive material (Table 1). Thus, the α1 integrin expression found on immunoblots of extracts from the heart, skeletal muscles and skin could be attributed mostly to capillary endothelial cells and to vascular smooth muscle cells. The distribution of α1 integrin in adult avian tissues is summarized in Table 1. It shows that α1 integrin expression was restricted to smooth muscles and capillary endothelia.

α1 integrin immunoreactivity was detected early, during neurulation in the 6-somite-stage avian embryo. It was found in all tissues deriving from the three primary germ layers (Fig. 2A). However, staining was weak and uniform throughout the embryo. Only a few tissues occasionally were seen to express α1 integrin at higher levels; these were the notochord and the most caudal part of the growing Wolffian duct. The faint staining for α1 integrin persisted in tissues until day 2.5 of development (E2.5). By the 20-to 25-somite stage, α1 integrin distribution gradually changed to become selectively enhanced in the neural tube in cranial regions, in the most rostral myotomes, in the myocardium and in endothelial cells of developing capillaries (see below). By E3, the pattern of expression of α1 integrin became tissue specific, particularly in muscles and in the nervous system.

The skeletal musculature in the trunk derives from the somites, an epithelial structure formed from the segmental plate located in the caudal end of the embryo. Segmental plate cells showed negligible staining for both α1 integrin (Fig. 2A), laminin and collagens I and IV (not shown, but see Duband et al., 1987; Duband and Thiery, 1987). α1 integrin staining was slightly enhanced during condensation of the segmental plate into the somitic epithelium (Fig. 2B). This increase in α1 integrin staining correlated with the appearance of laminin and collagens surrounding the somite (Duband et al., 1987; Duband and Thiery, 1987).

Shortly after their formation, somites undergo compartmentalization into the sclerotome, a mesenchyme from which the cartilage of the vertebrae develop, and the dermamyotome, an epithelial sheet from which the dermis and the musculature arise. This reorganization was accompanied by a relative decrease in α1 integrin staining among sclerotomal cells, but with no noticeable changes in the dermamyotome (not shown). Once formed, the dermamyotome is transformed into a double-layered epithelium comprising the dermatome and the myotome. Cells that constituted the myotome were strongly stained for α1 integrin as soon as they appeared (Fig. 2C). Coincidentally, a basal lamina containing both laminin and collagens I and IV was deposited along the myotome (Fig. 2D). The sequential appearance and increase in α1 integrin staining in the myotome could be detected first in the more rostral segments by approximately stage 20 somites, then appeared gradually in the most caudal segments as the development of the embryo progressed. By E4, all myotomes were formed and expressed high levels of α1 integrin. α1 integrin staining became more pronounced on the myotome as it extended towards the ventral regions of the embryo (Fig. 2E). Laminin also remained abundant along the myotome (Fig. 2F).

At E4 a number of myoblasts start to separate from the most ventral portion of the myotome and migrate within the lateral body wall to form the abdominal muscles. As shown in Fig. 3A, B, these myoblasts were in contact with a laminin matrix and expressed high amounts of α1 integrin. Myoblasts reach their target sites and form the abdominal muscle blastemas by E6. By E9-E10, myotubes have appeared and the abdominal muscles have their defined anatomy (see for details, Jaffredo et al., 1988). At this stage, α1 integrin staining was conspicuous on myotubes (Fig. 3C), and laminin immunoreactive material was also extremely abundant along each myotube (Fig. 3D). After E11, myotubes became fully differentiated and contractile, and α1 integrin disappeared totally from their surfaces, whereas laminin was still retained in the cells’ periphery (Fig. 3E, F). This disappearance of α1 integrin from skeletal muscles could also be evidenced by immunoblotting (Fig. 1E). The other skeletal muscles in the limb and head followed a similar pattern of α1 integrin expression, each with different developmental timing specificities. For example, facial muscles still expressed α1 integrin at stages when abdominal muscles had already lost their staining, probably reflecting differences in the timing of their maturation. After hatching, however, α1 integrin could not be detected in any skeletal muscle throughout the body.

Visceral smooth muscle precursors appear gradually between E3 and E5 after local delamination of the splanchnic epithelium along the developing digestive tract. Even though the splanchnic epithelium was occasionally found to express α1 integrin prior to delamination, smooth muscle precursors were devoid of α1 integrin immunoreactivity at the time that they were released. This absence of α1 integrin expression is illustrated in the gizzard at E4 (Fig. 4A). Differentiation of visceral smooth muscle cells occurs sequentially along the digestive tract. It can be evidenced by the increase in the α-smooth muscle actin content of the cells, first rostrally in the esophagus (Fig. 4C), then progressively more caudally within the intestine, and finally in the cloaca and in the gizzard (Fig. 4E). The appearance of α1 integrin strictly followed the differentiation pattern of smooth muscle cells along the digestive tract. It could be detected first in developing smooth muscle cells in the esophagus by E5.5-E6 (Fig. 4B). α1 integrin staining was intense along the whole intestine by E7.5, and in the gizzard and cloaca by E9.5 (Fig. 4C). The conspicuous expression of α1 integrin in visceral smooth muscle was retained throughout development and was enhanced in adult animals compared to late embryos and newly hatched animals (Fig. 9). Immunoblotting experiments further demonstrated the increase in the α1 integrin content of the gizzard during development (Fig. 1D, E).

α1 integrin was also expressed at high levels by smooth muscles in the pulmonary buds. Due to the relatively late development of the lung, α1 integrin was detectable in pulmonary smooth muscle cells only after E9.5 (Fig. 4F). Similar to what was observed with visceral smooth muscle, α1 integrin expression also increased in pulmonary smooth muscle after hatching (not shown).

By E3 as the aorta becomes entirely surrounded by mesenchymal cells, a number of these cells clearly showed α1 integrin staining (Fig. 5A). Due to their high α-smooth muscle actin content, these cells could be unambiguously designated as early differentiating vascular smooth muscle cells (Fig. 5B). By E4.5, the dorsal aorta in the thoracic region was surrounded by several layers of cells which showed both strong α1 integrin and α-smooth muscle actin stainings (Fig. 5C, D). In contrast, in the most caudal regions, the dorsal aorta was less mature and showed very faint labeling for α1 integrin. Staining for α1 integrin on arterial smooth muscle cells did not change significantly until E14. By this stage, the number of smooth muscle layers around arteries again increased significantly, with a coincident enhancement of α1 integrin labeling (Fig. 9).

α1 integrin staining was not prominent in the neural tube during neurulation (Fig. 2A-C). However, the level of α1 integrin expression in the central nervous system changed at the time when neurons underwent morphological differentiation. The first neurons to exhibit strong α1 integrin staining could be detected in the ventral floor of the forebrain at the 20-to 25-somites stage (Fig. 6A). These cells were also intensely labelled for NC-1 (Fig. 6B) which has been shown previously to appear in early differentiating central neurons (Vincent and Thiery, 1984). High level of α1 integrin expression in the brain occurred between E3 and E5. As stratification of the brain began, staining was predominantly associated with the cellular layers (Fig. 6C). After E6, the α1 integrin content in the brain declined significantly and became almost undetectable by E9 (Fig. 6D).

The spinal cord followed essentially the same pattern of α1 integrin expression as the brain. Motor neuroblasts were the first cells to express high amounts of α1 integrin at the time when they extended neurites out into the sclerotome at E3-E3.5 (Fig. 7A). At this stage, staining was particularly strong on the cell bodies of the motoneurons. Later, as neurites started to regroup and form bundles Schwann cells, which were detected by NC-1 labeling, also exhibited strong α1 integrin staining (Fig. 7C). The high level of α1 integrin expression on Schwann cells could be correlated with the appearance of laminin along the developing nerve shaft (not shown, but see Rogers et al., 1986; Duband and Thiery, 1987). After E3.5-E4, α1 integrin expression was observed in most neurons within the spinal cord (Fig. 7B). The fibrous layers and the ependymal zone were less brightly stained than the neuron cell bodies. At our level of resolution, it was not possible to determine if glial cells were also stained. By E7.5, α1 integrin staining dropped to a relatively low level in the cellular layers (Fig. 7D) and, by E10, it was absent from the entire spinal cord (Fig. 7E). The disappearance of α1 integrin from the spinal cord could also be evidenced by immunoblotting (Fig. 1E).

In the trunk, the peripheral nervous system arises entirely from the neural crest. During migration, trunk neural crest cells were only weakly stained for α1 integrin (Fig. 2B, C). However, at E3 (stage-35 somite), α1 integrin expression was significantly enhanced in neural crest cells regrouping into the dorsal root ganglion anlagen (Figs 2E, 7A). This increase in staining was accompanied by the appearance of laminin and collagen IV within the crest cell population, which were first seen as dots, and subsequently as a thin basal lamina entirely surrounding the mass of the ganglion (not shown, but see Rogers et al., 1986; Duband and Thiery, 1987). Expression of α1 integrin increased in the sensory spinal ganglia by E4-E4.5, as sensory neurons started to extend neurites (Fig. 7B). Spinal sensory ganglia retained strong α1 integrin staining for at least the following five days of development (Fig. 7D, F). During this period of time, laminin and collagen IV could be detected around individual neurons and along neurites (Duband and Thiery, 1987). After E11, α1 integrin staining declined significantly among the mass of the ganglion and was negative before hatching (not shown). α1 integrin expression also increased dramatically in the sympathetic ganglia at the time of their condensation along the aorta (Fig. 8A). However, in contrast to spinal ganglia, staining decreased rapidly to weak levels after E6 (not shown). It should be remembered that sympathetic ganglia express small amounts of laminin and collagens throughout development (Duband and Thiery, 1987).

The cranial peripheral ganglia also exhibited strong labeling for α1 integrin, each with developmental timing specificities. This strong α1 integrin staining could be generally correlated with the presence of laminin and collagen IV within the mass of the ganglion. For example, the trigeminal ganglion expressed high amounts of α1 integrin, laminin and collagen IV almost throughout embryonic development (Fig. 8C). Staining disappeared in the trigeminal ganglion only a few days before hatching. In contrast, the ciliary and the enteric ganglia, which expressed low levels of laminin and collagens, were stained for α1 integrin only transiently between E3.5 and E10 (Fig. 8B).

The ectodermal placodes at the origin of the sensory parts of the inner ear and of the olfactory organ expressed α1 integrin at high levels transiently, first during their formation, then during the extension of neurites from the placode towards the brain (Fig. 8D, E). The placodes lost their prominent α1 integrin staining later during development, possibly as neuronal connections were established. The anterior lobe of the hypophysis, which derives from a local invagination of the ectoderm (the Rathke’s pouch) coming in contact with the floor of the forebrain, also exhibited a transient expression of α1 integrin during its formation (not shown).

Finally, the optic cup, which develops from the neural epithelium in the forebrain, was also found to express α1 integrin during its formation. Later, as the retinal ganglion cells became individualized and extended axons out of the retina into the brain approximately between E5 and E10 (Rager, 1980), they became brightly labelled for α1 integrin (Fig. 8F). Laminin was also found on the internal side of the retina where retinal ganglion cells are found and along the optic nerve (not shown). After E12, however, both α1 integrin and laminin staining decreased among the retinal ganglion cells and along the optic nerve. This corresponded to the stage when virtually all retinal ganglion cells have reached the optic tectum (Rager, 1980).

Beside neural and muscular tissues, the only other cell types that expressed abundant amounts of α1 integrin were capillary endothelial cells and cardiomyocytes. α1 integrin immunoreactivity could be detected on capillary endothelial cells throughout development in a large variety of organs, including skeletal muscles (Fig. 3E), myocardium, kidney and dermis, and along the nervous system (Fig. 5E, F). α1 integrin was only transient in cardiomyocytes; staining was strong at the time of appearance and early differentiation of cardiomyocytes until E6 (Fig. 9). As described previously in the rat (Terracio et al., 1991), adult avian cardiomyocytes did not exhibit α1 integrin staining.

α1 integrin was also occasionally detected in a number of epithelial sheets. The ectoderm and the endoderm expressed varying amounts of α1 integrin only during the first half of the embryonic development. Transient α1 integrin expression in the human epidermis has also been reported recently (Hertle et al., 1991). In contrast, the liver, the pancreas, cartilages and most organs associated with the digestive tract never expressed α1 integrin. In the kidney, all tubules were negative for α1 integrin. In this organ, beside capillary endothelial cells, the only cells that expressed α1 integrin at detectable levels were mesangial cells in the glomeruli, as also described previously in human kidney (Korhonen et al., 1990).

It is commonly accepted that the components of the extracellular matrices and their cell surface integrin receptors play an important role in morphogenesis and cell differentiation. In this report, we have systematically investigated the expression of the α1β1 integrin during avian embryonic development and in the adult quail. We have found that (i) in adult avian tissues, α1 integrin was specifically expressed in visceral and vascular smooth muscles and in capillary endothelial cells, (ii) during development, α1 integrin showed a transient, dynamic distribution in striated muscles and in the nervous system and (iii) the expression of α1 integrin in muscles and nervous system was often correlated with the expression of laminin and collagen IV.

In this study, we used an antibody to a p175 membrane protein isolated from chicken gizzard smooth muscles (Koteliansky et al., 1985). The specificity of this antibody was compared with that of a recently characterized rabbit antiserum to avian α1 integrin (Syfrig et al., 1991), using several different approaches. By immunoblotting of chicken gizzard extracts, both antibodies recognized a band of approximately 175×10³M_r and, by immunofluorescence staining of tissue sections, they gave identical patterns of labelling. In addition, the anti-p175 antibody recognized on immunoblots the α1 subunit of purified avian α1β1 integrin complex and immunoprecipitated the same proteins from human uterus smooth muscle extracts as a monoclonal antibody to human α1 integrin. Finally, when avian gizzard extracts were immunodepleted with the rabbit antiserum to avian α1 integrin or with the anti-p175 antibody, no bands comigrating with the α1 integrin were detected by immunoblotting. All together, these results strongly suggest that the anti-p175 protein antibody specifically recognized the α1 integrin subunit in both human and avian species.

α1β1 integrin has been shown to bind both laminin and collagens I, III and IV (Ignatius and Reichardt, 1988; Kramer and Marks, 1989; Turner et al., 1989; Belkin et al., 1990; Ignatius et al., 1990). More specifically, it interacts with the E1 domain of laminin and with a triple-helical region of collagen IV (Rossino et al., 1990; Sonnenberg et al., 1990b; Tomaselli et al., 1990; Vandenberg et al., 1991). A striking feature of the distribution of α1 integrin is that it is restricted to relatively few cell types both during development and in the adult (Fig. 9). With the exception of the early developmental stages, in which it is occasionally expressed at low levels in the ectoderm and endoderm, α1 integrin was almost never found in epithelial sheets. Moreover, it was found at low levels in tissues that adhere to connective-tissue-type collagens. One would have expected a laminin/collagen receptor to be expressed in a large number of tissues, most particularly in epithelia. It thus seems that α1 integrin may serve preferentially as a receptor for laminin and collagen IV in non-epithelial cells, such as the nervous system, striated and smooth muscle cells, and capillary endothelial cells. Our observation is consistent with a recent report showing that, in avian, collagen IV and to a lesser extent laminin are the preferential ligands of α1 integrin whereas collagen I does not bind this integrin (Syfrig et al., 1991). Therefore, one should assume that other cell types that are known to interact with laminin and collagen would utilize alternative integrin receptors. Other potential laminin and collagen receptors are the α2β1, α3β1, α6β1 and the α6β4 integrins. Connective tissue cells, including fibroblasts, are likely to adhere to collagens I and III through the α2β1 integrin (Elices and Hemler, 1989), whereas endothelial cells of large vessels are known to express α6β1 and α2β1 integrins as laminin receptors (Languino et al., 1989; Sonnenberg et al., 1990a). Among epithelial cells, adult human keratinocytes either in vitro or in vivo express α2β1, α3β1, α6β1 and α6β4 integrins as laminin/collagen receptors but not the α1β1 integrin (De Luca et al., 1990; Stepp et al., 1990; Hertle et al., 1991; Marchisio et al., 1991; Sonnenberg et al., 1991). Likewise, epithelial cells of the human nephric tubules exhibit the α2β1, α3β1 and α6β1 integrins (Korhonen et al., 1990). A complete survey of the distribution of the α6 and β4 integrin subunits in human adult tissues has been published recently (Sonnenberg et al., 1990a). It is of interest to note that most epithelial tissues were found to express α6 and β4 integrins, with the noticeable exception of the liver, whereas most non-epithelial tissues that are known to interact with laminin and collagen IV (i.e. striated and smooth muscles, and nervous system) did not express these two integrins. These results, combined with our observations, would indicate that epithelial tissues would interact with laminin and collagen IV through one or a combination of the α6β1, α6β4, α2β1 and α3β1 integrins, while nonepithelial tissues adhere to the same extracellular matrix components using at least the α1β1 integrin. Such a difference in the expression pattern of laminin/collagen-binding integrins may possibly reflect functional specificities of the different members of the integrin family.

There are a number of possible explanations for the functions and developmentally regulated expression of α1 integrin during development and more specifically during myogenesis and neurogenesis. A possibility is that transient expression of the α1 integrin subunit reflects modifications in the composition of the extracellular matrix and, consequently, may correlate with changes in cellular behavior. This is probably the case for the nervous system. Our study reveals that sensory ganglia express high amounts of the α1 integrin subunit as long as they exhibit laminin and collagen IV. Conversely, α1β1 integrin expression is very transient in autonomic ganglia, correlating with a low laminin/collagen IV content. In the retina, both α1β1 and α6β1 integrins and laminin disappear from the surface of retinal ganglion cells after E11-E12, correlating with the fact that retinal neurons lose their ability to attach and extend neurites on laminin (this study; Cohen et al., 1986; de Curtis et al., 1991; Hall et al., 1987). Finally, retinoic acid and nerve growth factor, two naturally occurring morphogens, were found to increase α1 integrin expression in neuroblastoma cells in correlation with increased neurite response to laminin (Rossino et al., 1990, 1991). Thus, the patterns of expression of integrins and those of the extracellular matrix components are intimately interrelated. It has been found in in vitro studies that matrix components may control the surface distribution of integrins in cell-to-substratum anchorage sites (Dejana et al., 1988; Roman et al., 1989). It is then possible that, in vivo, integrin expression might also be regulated by extracellular matrix molecules during development of the nervous system.

The distribution of α1β1 integrin during neurogenesis at the time of nerve sprouting and on migrating myoblasts is consistent with a functional role of this integrin in cell locomotion over laminin and collagen IV substrata. In vivo, laminin and collagen IV have been detected in the nervous system at the time of neurite extension (Duband and Thiery, 1987; Hagg et al., 1989; Liesi, 1985; Rogers et al., 1986) and alongside migrating myoblasts (Solursh and Jensen, 1988; this study). In vitro, laminin has been shown to exert a key role in neurite sprouting for a large variety of neurons of either the central or peripheral nervous system (for reviews, see Reichardt and Tomaselli, 1991; Sanes, 1989) and to enhance motility of myoblasts (Öcalan et al., 1988). Likewise, collagen IV promotes neuritic outgrowth in various neurons (Lein et al., 1991). Integrin receptors, and more particularly the α1β1 integrin, are actively involved in cell adhesion to laminin and collagen IV during these processes of cell migration, as evidenced by in vitro and in vivo perturbation experiments (Neff et al., 1982; Bozyczko and Horwitz, 1986; Tomaselli et al., 1986; Hall et al., 1987; Jaffredo et al., 1988; Ignatius and Reichardt, 1988; Turner et al., 1989; Rossino et al., 1990, 1991; Tawil et al., 1990; Tomaselli et al., 1990; Goodman et al., 1991; Lein et al., 1991).

Beside its possible role in cell migration during myogenesis and neurogenesis, α1β1 integrin may also be involved together with other β1 integrins in the differentiation program of these cells. Indeed, it has been shown that laminin promotes myogenesis in rat skeletal muscles in vitro (Foster et al., 1987) and that occupation of integrins by their corresponding ligand is a crucial event for several stages of myogenic differentiation (Menko and Boettiger, 1987). In Drosophila, the αPS2βPS integrin is expressed in segmental muscle attachment sites and at Z-bands. This particular integrin is crucial in sarcomere formation, as integrin βPS null myosopheroid mutants lack Z-bands and fail to form mature muscle fibers (Volk et al., 1990). The intense expression of α1β1 integrin in myoblast precursor cells coincides with the pattern of expression of qmf1, the avian homologue of myf-5, a member of the MyoD regulatory gene family (Braun et al., 1989; Charles de la Brousse and Emerson, 1990; Ott et al., 1991). This correlation in expression of a skeletal muscle master gene and of α1β1 integrin suggests that α1β1 integrin expression might be triggered by members of the MyoD family, which are responsible for switching on the differentiation program of myogenic precursors. Alternatively, α1β1 integrin itself may promote the expression of members of MyoD gene family, thus allowing initiation of myogenesis.

We thank Dr M.A. Glukhova for helpful advice. Many thanks to Drs K. Brown and K.M. Yamada for careful reading of the manuscript. This work was supported by the Centre National de la Recherche Scientifique (CNRS). V. E. K. acknowledges the CNRS for supporting fellowship. J.-L. D. acknowledges the CNRS (Programme ATIPE), the Ministère de la Recherche et de la Technologie (Grant 91 T 0011), the Association pour la Recherche contre le Cancer (ARC 6517), the Ligue Nationale contre le Cancer and the Fondation pour la Recherche Médicale for support. J. S. was supported by a grant from the Swiss National Science Foundation.