Nucleolin is an abundant nucleolar protein involved in several steps of ribosome biogenesis. The protein is highly conserved through evolution and possesses four RNA-binding domains (RBD), which are likely to determine its RNA binding specificity. Previous studies have shown that nucleolin interacts with two different RNA targets. The first is a small stem-loop structure, the nucleolin recognition element (NRE), found all along the pre-ribosomal RNA. The second is a short singlestranded RNA sequence, the evolutionary conserved motif (ECM), located five nucleotides downstream of the first processing site in the pre-ribosomal RNA 5 external transcribed spacer. Biochemical, genetic, and structural studies have shown that the first two RBD of nucleolin are necessary and sufficient for the specific interaction of nucleolin with the NRE motif. In this work, we have studied the interaction of nucleolin with the ECM sequence. Deletion and mutational analyses showed that all four RBDs of hamster nucleolin were required for the interaction with the ECM sequence. This RNA binding specificity is conserved between hamster and Xenopus laevis, whereas the Xenopus protein does not interact with the NRE. Nucleolin is the first example of a protein that requires four RBDs for its interaction with an RNA target, demonstrating that a single protein can use different combinations of RBD to interact specifically with several RNA sequences.Specific RNA-protein interactions play an important role in gene expression. One of the most common protein sequence motifs involved in these interactions is the RNA-binding domain (RBD), 1 also called the RNA recognition motif (RRM) (1, 2). A single or multiple RBD, often in combination with other domains involved in protein-protein interaction, are found in proteins that have diverse functions such as in pre-mRNA maturation and splicing (3, 4), hnRNA packaging (5), and mRNA stability and translation (6, 7).The RBD domain is characterized by two highly conserved motifs called RNP-1 and RNP-2 found within a weakly conserved 80-amino acid sequence (1, 2). The crystal and solution structures of several RBDs have revealed a common structural organization composed of four antiparallel -strands packed against two ␣-helices (Refs. 8 -10; 14). The RNP-1 octapeptide and RNP-2 hexapeptide are located in the 1 and 3 strands, respectively. Structures of RBD found in tandem show that the two independently folded domains do not interact with each other in Sex-lethal (11, 12), HuC (13), and nucleolin proteins (14). In contrast, in hnRNPA1, the two domains interact extensively (15, 16). Despite the highly conserved structure of the RBD, they show a remarkable ability to interact with diverse RNA sequences and structures. The tertiary structures of several RBD⅐RNA complexes are now available. These are the single RBDs from the U1A and U2BЈЈ protein bound to a stem-loop structure (8,17,18), the two RBDs of Sex-lethal, hnRNPA1, and poly(A)-binding protein bound to a singlestranded RNA (12,16,19), and the first two...