DNA sequencing of a 1.7-kilobase cloned cDNA allowed determination of the complete COOH-terminal noncollagenous region (NC1) of the human alpha 2(IV) collagen chain. This 227-residue domain is composed of two equal-sized sequential "repeats" as had been observed for the corresponding 229-residue domain of the alpha 1(IV) chain. Alignment of the alpha 2(IV) repeats with each other and with those in alpha 1(IV) suggests that the type IV NC1 regions evolved via an intra- to intergenic duplication. In addition, smaller internal units having a consensus sequence GYSSCLFLWYLAF are present multiple times in the 110-115-amino acid halves. Consonant with the predominance of Tyr, Leu, and Phe in the heptads, hydrophilicity profiles of the homologous alpha 1(IV) and alpha 2(IV) regions revealed extended hydrophobic stretches which are similar to those in the noncollagenous COOH terminus of the chicken alpha 1(X) chain (Ninomiya, Y., Gordon, M., van der Rest, M., Schmid, T., Linsenmayer, T., and Olsen, B. R. (1986) J. Biol. Chem. 261, 5041-5050). Isolation of an alpha 2(IV) clone also enabled us to investigate if the alpha 2(IV) and alpha 1(IV) collagen mRNAs were coordinately transcribed and if one or both were consistently associated with either type I (alpha 1 and alpha 2), III, or V (alpha 2) transcripts as a function of cell type. Northern blot hybridization of collagen cDNA probes to poly(A) RNA extracted from human and bovine cell cultures showed that only the alpha 1(III) and alpha 2(V) genes were expressed in all cells examined. Unexpectedly, neither type IV mRNAs were found in bovine endothelial, smooth muscle, or fibroblast cells, whereas both type IV species were present in human fibroblasts and to a much greater extent in human umbilical vein endothelial cells.