SELEX selection and characterization of high-affinity DNA sequences bound by Ff gene 5 protein

The bacteriophage Ff gene 5 protein (g5p) is a model single-stranded DNA (ssDNA) binding protein. It possesses biologically relevant dual features of binding cooperatively to the viral ssDNA genome during replication and of binding sequence-specifically to the viral mRNA to regulate translation. To further explore the binding properties of g5p, I used SELEX to select high-affinity ssDNA sequences. From a synthesized 58-mer ssDNA library with a central random region and two constant tails, two major groups of structure-forming sequences were separately selected at physiologically ionic strength (200 mM NaCl). The first group was isolated from g5p-saturated complexes. Circular dichroism and other studies showed that I-3, the predominant sequence of this group, can fold into an intramolecular G-quadruplex with unstructured constant tails. Folding of the quadruplex occurs through a transiently folded state and is sensitive to the concentration and type of monovalent cation. The G-quadruplex promotes a two-stage g5p assembly, in which three g5p dimers first bind to the quadruplex core of I-3 and slightly distort the structure into a state similar to the transiently folded form. The remainder of the sequence is then saturated by further g5p binding to the tails. The second SELEX-selected group was derived from g5p-bound intermediate complexes. A major family of sequences in this group contained two regions of complete or partial sequence complementarity. Studies of a representative sequence, U-4, showed that the segments with complete complementarity pair to form a hairpin. A dimer of U-4 is also formed by pairing of the segments with partial complementarity at higher oligomer concentrations. The g5p assembles on the hairpin to form an intermediate complex prior to saturation of the oligomer at high g5p concentrations. In an attempt to identify biologically relevant g5p-binding sites, a third SELEX experiment was performed with a well-characterized fd genomic ssDNA library. The preliminary results showed that the selected sequences were unlikely to be g5p binding targets in vivo, because constant tails were involved in base-pairing with the genomic inserts during the selection. A modification of this procedure to eliminate the constant tails is required for future genomic SELEX.