An egg-stremely useful interaction
In 1941, when Esmond Snell and co-workers were investigating the cause of “egg-white injury” -a vitamin H (biotin) deficiency in hens fed raw egg white- they coined the term ‘avidin’ for a protein that they had isolated from egg white which sequestered biotin ( ). Because of avidin’s hungry interaction with biotin (hence the name: avid + biotin), they suggested “its possible use as a tool in the purification of biotin.” In fact, the interaction is so strong that even though there was plenty of biotin in the hens’ diet, it was all bound by the avidin in their egg white food. Seventy years on, the interaction between avidin (or its prokaryotic homologue streptavidin) and biotin has been extensively studied. With a of 10-14 M, it is one of the tightest non-covalent interactions known, and its essentially irreversible nature makes the system a very useful tool in biotechnology and medicine. Snell could hardly have imagined how useful the protein would become.
Inside the barrelThe structure of biotin bound to streptavidin (from the bacterium Streptomyces avidinii) was first solved in 1989 (PDB entry 1stp) and that of the avidin-biotin complex followed a few years later (PDB entry 2avi). While only moderately conserved at the sequence level (30% identity), the two proteins share a common secondary and tertiary structure. Streptavidin and avidin fold into an eight-stranded beta barrel (view-1) typical of the lipocalin family and both exist as tetramers. Like in all lipocalins, the ligand-binding site is located inside the barrel. There is a high degree of shape complementarity between the binding site and biotin. The mostly hydrophobic ligand biotin is bound through hydrogen bonds and hydrophobic interactions within the monomer (view-2). One face of the biotin is exposed but in the tetrameric assembly residues from another monomer complete the biotin-binding site, almost completely burying the biotin within the protein. In total, the interaction between the monomers buries some 1600 Å2 of protein surface and most notably, a conserved tryptophan located in a loop between the seventh and eighth strands of the barrel (called loop7-8) stacks against the biotin in an interaction which is reciprocated between the two subunits (view-2). Mutating this tryptophan to a phenylalanine (PDB entry 1swp), reduces the binding constant for biotin by two orders of magnitude. Because each biotin-binding site is made up of residues from two monomers, (strept)avidin can be thought of as a functional dimer.
Putting a lid on itHaving a biotin binding site buried within the protein would make it difficult for the ligand to enter, and the network of interactions alone doesn’t account for the tight binding. Structures solved in the absence of biotin (e.g., PDB entries 1ave and 1swb) show that the loop between the 3rd and 4th strands of the beta barrel is flexible, only becoming ordered once biotin is bound in the cavity and residues in this loop are able to interact with it. The ordered loop effectively forms a ‘lid’ on the binding site (view-3) such that the biotin is almost completely buried within the interior of the protein.
While (strept)avidin can be considered to be a functional dimer, two of these form the in vivo tetramer (view-1). This quaternary structure is stabilised through a network of hydrogen bonds and hydrophobic interactions between the molecules. The dimer-dimer interface buries some 4000 Å2 of protein surface, rendering the tetramer stable and resistant to proteolysis. As (strept)avidin is secreted from the cell, stability is a prerequisite for it to retain its biotin binding function in the harsh external environment, and also of course useful for its in vitro applications.
Avidin engineeringAvidin has evolved as an antibiotic; it sequesters an essential vitamin and so helps prevent bacterial growth within the egg. Knowledge of the structures has not only elucidated the nature of the remarkable affinity of (strept)avidin for biotin, but also inspired the engineering of mutations that tune the properties of the protein to make it an even more useful tool. Some mutations weaken the affinity of the protein for biotin. In PDB entry 4ekv, for example, the tryptophan from the second monomer which packs against the biotin has been mutated to an aspartate, and other residues in loop7-8 mutated to make this part of the protein more flexible. As a result, loop7-8 does not pack against the biotin which changes the binding site into a cleft on the protein surface (view-4). The dissociation constant for this interaction is 10-8 M, which means that the binding is reversible and the protein can be used as an affinity agent to purify biotinylated molecules.
Currently the PDB archive contains over 180 structures of avidin, streptavidin or the equivalent proteins from other organisms; you can explore all of these structures using the PDBe Pfam browser.
Further explorationMany molecules of the lipocalin fold bind ligands within their barrel. You can explore structures which have this fold using the PDBeXplore Fold browser. This mini-tutorial will demonstrate how.