The tertiary structure of the integrin heterodimer is currently unknown, although several predictive models have been generated. Detailed structural studies of integrins have been consistently hampered for several reasons, including the small amounts of purified protein available, the large size and conformational flexibility of integrins, and the presence of transmembrane domains and N-linked glycosylation sites in both receptor subunits. As a first step toward obtaining crystals of an integrin receptor, we have expressed a minimized dimer. By using the Fc dimerization and mammalian cell expression system designed and optimized by Stephens et al. Integrins are ␣,␤ heterodimeric transmembrane receptors that play central roles in cell adhesion, migration, differentiation, and survival (1). Several lines of evidence indicate that integrins also contribute to the progression of a wide variety of diseases, including inflammatory, thrombotic, and neoplastic conditions (2-4), and that the integrin families are valid therapeutic targets. The rational design of integrin antagonists based on ligand peptide motifs such as RGD and LDV is currently well advanced. Although the tertiary structure of the integrin heterodimer is unknown, this information would aid the process of drug development, and it represents one of the most important outstanding questions in the field.The overall shape and dimensions of the ␣IIb␤ 3 and ␣ 5 ␤ 1 integrin heterodimers have been revealed by rotary shadowing electron microscopy (5-7). Both receptors consisted of an Nterminal globular head of 8 -12 nm with two extended tails of 18 -20 nm that corresponded to the C termini (7). Similarly, a soluble ␣IIb␤ 3 integrin, generated by removal of the ␣IIb and ␤ 3 transmembrane and cytoplasmic domains, and the ␤ 3 cysteine-rich repeats also contained a globular head, but its tails were 4 -6 nm shorter (8).In the absence of a tertiary structure for the integrin heterodimer, several predictive models have been generated (9 -12), and these have subsequently been supported by biochemical analyses (13)(14)(15)(16). Whereas the ␣-and ␤-subunits are unrelated in primary sequence, they share common structural features including an N-terminal globular ligand-binding domain, C-terminal stalk regions, transmembrane domains, and short cytoplasmic domains (17,18). The N-terminal portion of ␣-subunits contains seven homologous repeats, each 60 -70 amino acid residues in length. These repeats are quite similar in sequence, and repeat four in some integrins and repeats five to seven in all receptors contain EF-hand-like divalent cationbinding motifs. The seven repeats have been predicted to fold cooperatively, forming an all-␤ structure known as a ␤-propeller fold (9) (Fig.