Post-transcriptionally altered nucleotides are common in ribosomal RNA (rRNA) in all organisms; however, prokaryotes and eukaryotes differ significantly in the numbers and types of these modifications (1–6
). For example, the rRNA of Escherichia coli
contains a total of 35 modifications. Most (69%) are methylated, with CH3
groups added to heterocyclic bases (mN; 60%), and in a few cases to the ribose moiety (Nm; 9%). Another subset is comprised of uridines converted to pseudouridines (Ψ; 31%) (2
). In eukaryotes the number of modifications in rRNA is significantly higher, with over 100 in yeast and more than 200 in vertebrates (3
). The most abundant modifications in eukaryotic rRNAs are pseudouridines and ribose methylations, which typically account for ~95% of the alterations. Less numerous are nucleotides with a base methylation(s).
Many protein and RNA factors that participate in rRNA modification have been defined and many remain to be identified (6
). Thus, far eubacteria appear to use protein-only enzymes to catalyse the various modifications. However, while eukaryotes and archaea also have such enzymes, most 2′-O
-methylations and pseudouridines are created (respectively) by C/D and H/ACA families of small ribonucleoprotein complexes. The RNP complexes in each family consist of one unique, site-specific guide RNA and a small set of family-specific proteins (3 or 4). The guide RNA component identifies the nucleotide to be modified based on complementarity between the guide element(s) in the small RNA and the rRNA substrate (11–14
The discovery that Saccharomyces cerevisiae
C/D and H/ACA small nucleolar RNAs (snoRNAs) guide Nm and Ψ formation in rRNA (15–17
) opened the way for identifying nearly all of the modifying snoRNPs and sites of Nm and Ψ in the first eukaryotic organism. Knowledge about the components of the yeast C/D and H/ACA snoRNPs spawned novel computational and experimental approaches for discovering guide snoRNAs and archaeal homologs (sRNAs) in many organisms (18–30
). Exploiting the complementarities between the guide sequence(s) and rRNA by computational approaches has identified both novel guide RNAs and sites of modification (including non-rRNA substrates). Indeed, databases for plant snoRNAs and human snoRNAs (and the functionally related scaRNAs) contain large amounts of data about predicted and confirmed snoRNAs and potential targets (9
). Knowledge about snoRNAs from other species is also developing quickly (31–37
). Thus, often the number and positions of many rRNA modifications can be predicted for several organisms before they have been established experimentally.
Little is yet known about the significance of the modifications in ribosomal RNA, although targeted depletion studies have shown impaired behavior in a few cases (e.g. (38
)). Strikingly, modeling of modification data of E. coli
, archaea and S. cerevisiae
into ribosome crystal structures revealed that many are concentrated in functionally important regions, for example, the peptidyl transferase center (PTC) in the large subunit (LSU), the decoding center in the small subunit (SSU) and on interacting surfaces of the two subunits (40–42
). Although atomic-resolution structures of ribosomes from higher eukaryotes are not yet available, 3D modification maps can be developed from such organisms, by modeling modification data onto high-resolution structures that do exist. The merit of this approach is supported by the very highly conserved structure of the ribosome and its functions (43–46
). While we eagerly await the availability of 3D maps that have been experimentally derived, development of deduced maps of the type described here will allow structure and function studies of modification effects to proceed at a high level.