Evidence from a variety of sources indicates that the extracellular matrix forms an important part of a feedback loop governing the migration, proliferation, and differentiation of the cells that produce it. In keeping with this, we showed previously that the extracellular matrix of a multipotential mesenchymal clonal cell line (ROB-C26) induced to differentiate into a more osteoblastic cell type by the addition of exogenous retinoic acid produces an extracellular matrix capable of osteoinductive activity in vivo and of stimulating alkaline phosphatase activity in vitro. Since type I collagen is the major structural component of this extracellular matrix, we sought to determine whether and to what extent this protein is responsible for the previously observed inductive/stimulatory activity. To this end, C26 cells are cultured on plastic, in the presence of retinoic acid, on a type I collagen film, or on an extracellular matrix from retinoic acid-treated C26 cells, and cell differentiation is assessed by measuring changes in the abundance of a number of osteoblast-related mRNAs. These determinations are made by RNAse protection assay after 3 or 6 days of incubation and include measurements of the RNAs for type I collagen, alkaline phosphatase, osteopontin, transforming growth factor a1 and p2, and Vgr-l/BMP-6. In addition, C26 cells are incubated in the presence of retinoic acid and several established inhibitors of the synthesis or assembly of extracellular matrix components and the effects on induced alkaline phosphatase activity determined. Our data show that while the collagen substrate mimics some of the effects of retinoic acid and the extracellular matrix, it cannot reproduce all of them. Specifically, while the latter two culture conditions increase the abundance of all six mRNAs, type I collagen film increases the levels of only three of the six (collagen I, alkaline phosphatase, and osteopontin). Moreover, while type I collagen film produces an increase in alkaline phosphatase message, it fails to produce a similar change in alkaline phosphatase activity, an effect seen with both retinoic acid and extracellular matrix. However, interruption of collagen I synthesis by cis-4-hydroxy-~-proline blocks the increase in alkaline phosphatase activity associated with retinoic acid treatment. Thus, it appears likely that type I collagen is a necessary but, by itself, insufficient factor to elicit the comprehensive expression of the osteoblastic phenotype in immature mesenchymal cells. (