This series of three short reviews is an attempt to summarize our current knowledge of the in vivo tests of hypotheses of laminin functions. The structures of the laminins have been thoroughly reviewed recently (P. Ekblom and R. Timpl, in press), and I will not attempt to repeat this information here. Instead, I will focus on the recent evidence gathered from gene knock out experiments in mice and from naturally occurring human and mouse gene mutations. The most obvious lesson from the above studies--other than demonstrating the importance of laminins in general--is that the structural diversity of the laminin family members makes highly specialized functions possible. While all laminins may share many functional properties, the individual chains are involved in interactions which cannot be substituted for by other laminins or by other basement membrane components. While this concept is not new, it is very satisfying to see its validity so dramatically confirmed. It is therefore predictable that additional gene ablation experiments using other known and yet undescribed laminin genes will be equally interesting and informative. To me, one of the most striking lessons from these studies is how strongly the induced mouse mutations mimic human disease. With all the concerns with genetic background differences and species specific effects, manipulation of the laminin genes appears to be a particularly good first approach to identifying the causes of human disease. There is an abundant literature accumulated from biochemical and, more recently, molecular structural analyses, and from in vitro systems, suggesting a role of laminins contributing directly to the stability of the basement membrane. There is an equally vast literature supporting an indirect role in mediating cellular behavior, through interactions with various receptors. It is interesting that the in vivo studies summarized above support both activities. In the case of laminin 5 mutations, the phenotypic consequence appears to be due primarily to the loss of an important structural link between the epithelial cytokeratins and the dermal anchoring fibrils. The ultrastructure of the epithelium appears normal, as does the architecture of the papillary dermis. Only the anchoring complex itself is aberrant. The absence of laminin 5 appears not to compromise the development or viability of the epidermis. The basement membrane appears normal-other than the anchoring complex itself. The pathology observed in the newborn is believed to be due to the frictional trauma of birth, with the expectation that the function of the fetal skin is normal in utero. The Herlitz epidermolysis bullosa phenotype is obvious immediately at birth, and it does not progress postnatally beyond the extent to which the affected individual experiences additional frictional trauma or secondary consequences such as infection or fluid loss. Since laminin 5 is only one of a series of structural links within the anchoring complex, one would predict that a loss of any of these links would result in the same phenotype. Current evidence supports this view, as the absence of integrin alpha 6 beta 4 (Vidal et al., 1995; Dowling et al., 1996; Georges-Labouesse et al., 1996; van der Neut et al., 1996) or of collagen VII (A. M. Christiano and J. Uitto, in press) also results in dramatic neonatal dermal-epidermal fragility. The differences in phenotype, such as the pyloric atresia in the case of loss of integrin alpha 6 beta 4, are presumably due to additional functions of the integrin in other tissues or in other developmental processes. Therefore, the laminin 5 mutations may be unique, in that the in vivo studies suggest that the primary role of the molecule is in the elaboration and stability of the anchoring complex, but not in the basement membrane itself. Of course, since the in vivo phenotype reflects only losses that cannot be compensated, this interpretation may be much too narrow. (ABSTRACT TRUNCATED)