The single magnesium ion on the surface of the I domain is thought to bind an anionic surface residue on the ligand, for example Glu34 in ICAM-1.
Huang & Springer (JBC'95) found four residues to be crucial to the ability of human CD11a to distinguish human ICAM-1 from mouse ICAM-1. These residues were in the I domain and are highlighted in yellow in the image to the right (top to bottom: Glu146, Met140, Thr243, and Ser245). They appear to define a binding face surrounding the magnesium ion.

A crystal structure of the extracellular domains of ICAM-2 (1ZXQ.PDB) has been deposited at the PDB and is on hold until September, 1997. No structures have been reported for the complex of the I domain and an ICAM fragment.

These RasMol images were made from 1ZOP.PDB published by Qu & Leahy (S'96). The 1ZOP crystal actually contained manganese ion in place of magnesium. However, the structure was very similar to that of the magnesium form of the crystal, 1ZOO.PDB. The ion is referred to as magnesium above because that is the physiological ion. 1ZOP was used because a larger number of waters are resolved (however, the waters are not shown in the above images). Not shown in the images above are a chloride ion which was bound to and directly in front of the mangnesium ion, and numerious water molecules, two of whose oxygens are bound directly to the magnesium ion.