Search

The heterophilic CD2-CD58 adhesion interface contains interdigitating residues that impart high specificity and rapid binding kinetics. To define the hot spot of this counter-receptor interaction, we characterized CD2 adhesion domain variants harboring a single mutation of the central Tyr86 or of each amino acid residue forming a salt link/hydrogen bond. Alanine mutations at D31, D32 and K34 on the C strand and K43 and R48 on the C' strand reduce affinity for CD58 by 47-127-fold as measured by isothermal titration calorimetry. The Y86A mutant reduces affinity by approximately 1000-fold, whereas Y86F is virtually without effect, underscoring the importance of the phenyl ring rather than the hydroxyl moiety. The CD2-CD58 crystal structure offers a detailed view of this key functional epitope: CD2 D31 and D32 orient the side-chain of CD58 K34 such that CD2 Y86 makes hydrophobic contact with the extended aliphatic component of CD58 K34 between CD2 Y86 and CD58 F46. The elucidation of this hot spot provides a new target for rational design of immunosuppressive compounds and suggests a general approach for other receptors., (Copyright 2001 Academic Press.)

Stopped-flow fluorescence studies on the N-terminal domain of rat CD2 (CD2.d1) have demonstrated that folding from the fully denatured state (U) proceeds via the transient accumulation of an apparent intermediate (I) in a so-called burst phase that precedes the rate-limiting transition leading to the native state (N). A previous pH-dependent equilibrium hydrogen exchange (HX) study identified a subset of amides in CD2.d1 which, under EX2 conditions, exchange from N with free energies greater than or equal to the free energy difference between the N and I states calculated from the stopped-flow data. Under EX1 conditions the rates of HX for these amides tend towards an asymptote that matches the global unfolding rate calculated from the stopped-flow data, suggesting that exchange for these amides requires traversing the N-to-I transition state barrier. Exchange for these amides presumably occurs from exchange-competent forms comprising the kinetic burst phase therefore. To explore this idea further, native state HX (NHX) data have been collected for CD2.d1 under EX2 conditions using denaturant concentrations which span either side of the denaturant concentration where, according to the stopped-flow data, the apparent U and I states are iso-energetic. The data fit to a two-component, sub-global (sg)/global (g) NHX mechanism, yielding Delta G and m value parameters (where the m value is a measure of hydrocarbon solvation). Regression analysis demonstrates that the (m(sg), Delta G(sg)) and (m(g), Delta G(g)) values calculated for this subset of amides correspond with those describing the kinetic burst phase transition. This result confirms the ability of the NHX technique to explore the structural and energetic properties of kinetic folding intermediates., (Copyright 2001 Academic Press.)

Domain 1 of CD2 (CD2.D1) forms a conventional Ig fold stabilised by non-covalent antiparallel contacts between beta-strands. Removing two residues from the middle of the protein sequence, where the polypeptide chain normally folds back upon itself, stabilises an open conformation. In this modified molecule, the optimum evolved contacts between side-chains can only be satisfied through the antiparallel association of two chains to create a symmetrical pair of pseudo-domains. Here, we describe the dynamics of the switch between monomeric and dimeric states and demonstrate the extension of this novel underlying principle to trimer and tetramer formation. The ability of a protein molecule to form higher-order antiparallel structures is reminiscent of the behaviour of hairpins, duplexes, three-way and Holliday junctions in DNA., (Copyright 1999 Academic Press.)

To determine the general relation between the sequence and structure of variable domains of immunoglobulins we have carried out an analysis of their atomic structures, some 5300 different expressed sequences and the human germline gene segments. Variable domains are formed by two beta-sheets, packed face to face, and the inter-strand turns. Comparison of the different known structures shows that they have a core of 76 residues which has the same main-chain conformation in all structures. This common core contains almost all of the beta-sheet structure and three inter-strand turns. The regions that differ in conformation are the three hypervariable regions, three other inter-strand turns and a few adjacent residues. The 5300 expressed sequences currently known for variable domains were examined to determine the residues that occur at the 76 sites. Ignoring site conservations that occur for functional reasons, there are eight sites that have the same residue in almost all sequences; 12 that have one of a small group of very similar residues, and 52 where the chemical character of the residues is strongly conserved but not their volume. The role of residues at each site in the core was determined from the examination of their accessible surface areas, contacts, packing and buried side-chain hydrogen bonds. The most strongly conserved sites form the "deep" structure of the domain at the centre of the interface between the beta-sheets. It includes eight invariant sites and 11 sites that have one of a set of very similar residues. Around the deep structure there are buried hydrophobic residues that, in different variable domains, can differ greatly in volume. These differences in volume are accommodated by conformational changes in turn regions that are outside the common core. On the surface nearly all residues not involved in function or turn conformations strongly conserve hydrophilic or neutral residues. The implications of these results for the general relations between the sequence and structure of proteins are discussed., (Copyright 1998 Academic Press Limited.)