Nucleolin is a sequence-specific RNA-binding protein: characterization of targets on pre-ribosomal RNA

Nucleolin is an abundant nucleolar protein, which plays an essential, but largely unknown role in ribosome biogenesis. Nucleolin contains four consensus RNA-binding domains (CS-RBD), the presence of which suggests that the molecular function of this protein is likely reflected by its RNA-binding properties. Indeed, by immunocytological analysis performed on ribosomal transcription units, we have found several nucleolin molecules associated with nascent pre-rRNA. In mouse, two high-affinity binding sites with an apparent dissociation constant (Kd) of 50 to 100 nM have been mapped in the 5' ETS upstream from the early pre-rRNA processing site. Interestingly, nucleolin of mouse origin has recognized analogous sequences in the 5' ETS of human pre-rRNA. In parallel, selection-amplification (SELEX) experiments have identified an 18-nucleotide long RNA sequence that binds nucleolin with high affinity (Kd 5 to 20 nM) and shares a common UCCCGA motif with the characterized pre-rRNA binding sites. By mutagenesis and a structural analysis, we have characterized the nucleolin RNA binding site and found that it is constituted by a minimal 18-nucleotide long stem-loop structure. The sequence UCCCGA that is found within the hairpin loop is necessary for the specific interaction. Mutation of any of the C or G residues within this motif abolishes nucleolin interaction. Furthermore, point mutation in the stem that completely disrupt the hairpin structure also prevents nucleolin binding. By determining the minimal 5' and 3' ends of the RNA that is bound to the protein we concluded that nucleolin binding site is constituted by a short four to five-base-pair stem and an eight-nucleotide loop. This structural motif is very similar to hairpins recognized by two other CS-RBD-containing proteins (U1 snRNP A and U2 snRNP B"). Possible functional implications of our findings are discussed.