Fibronectin, a fibroblast surface protein, was purified from human and chicken plasma and extracts of cultured chicken fibroblasts with affinity chromatography on gelatin coupled to Sepharose particles. A fibronectin-like protein was also isolated from the plasma of Torpedo fish. The collagen binding properties of fibronectin were studied with several genetically distinct collagens. Heat denatured types I, II, and III collagens were equal in their binding capacity and more active than the native collagens or A and B chains. Native type III collagen was more active than the other native collagens. Human and chicken fibronectins showed approximately the same pattern of specificity. Identical specificities were shown by the plasma and fibroblast forms of chicken fibronectin. Two cyanogen bromide peptides of the collagen alpha1 (II) chain, CB8 and CB12, derived from different parts of the chain, were active in fibronectin binding. A polymer of the tripeptide pro-gly-pro, and polyproline were inactive. Fibronectin also binds to fibrinogen and fibrin. Comparison of this binding to collagen binding showed that fibrinogen inhibited binding of fibronectin to collagen, but was less active than native collagen. Two other fibrous proteins, tropoelastin and keratin, did not bind fibronectin. The binding of fibronectin to fibrinogen was inhibited by collagen and incorporation of fibronectin into blood clot in the cold was inhibited by gelatin. These results suggest that the binding of fibronectin to collagen and fibrinogen depends on the same binding site in the fibronectin molecule. It is proposed that cell surface fibronectin mediates attachment of cells to the collagenous extracellular matrix and to a temporary fibrin matrix in a wound.
A nuclear antigen was detected in the mouse liver nonhistone protein fraction by using antibodies to whole liver cells. The antigen was purified to homogeneity from perchloric acid extracts of liver tissue. It gave a single band corresponding to tool wt 21,000 in sodium dodecyl sulfate gel electrophoresis. Amino acid and carbohydrate analysis showed predominance of the acidic amino acids, lack of proline, and absence of carbohydrate. Immunofluorescence staining of liver sections confirmed the nuclear localization of the antigen. Its tissue distribution was studied by using radioimmunoassay. Of the various tissues extracted for analysis, the liver contained the highest amounts of the antigen, about 1 μg/mg of solubilized liver protein. Other tissues examined showed 2-4 percent of the amount of antigen present in the liver. Two transplantable hepatomas in C3H/HeJ and C57L/J mice, respectively, and three spontaneous C3H hepatomas showed greatly decreased levels of the antigen compared to normal liver. The amount of antigen in hepatomas varied from nondetectable to 2 percent of the amount of antigen found in the livers of the mice. The antigen was also found in the blood. The antigen was found in high concentrations (up to 13 mg/ml) in the urine of normal mice. This suggests identity with the previously known mouse urinary protein (MUP). In addition to the extremely high urinary output, the properties found to be shared by MUP and the nuclear antigen included similar serum concentrations (2-60 μg/ml), a sex difference with lower values in females, same molecular size as determined by gel filtration, and immunological identity. The nuclear localization of MUP and its disappearance from hepatomas suggest that it may have an important regulatory function.