The multiple flavors of GoU pairs in RNA

Abstract Wobble GU pairs (or GoU) occur frequently within double‐stranded RNA helices interspersed within the standard G═C and A─U Watson‐Crick pairs. However, other types of GoU pairs interacting on their Watson‐Crick edges have been observed. The structural and functional roles of such alternative GoU pairs are surprisingly diverse and reflect the various pairings G and U can form by exploiting all the subtleties of their electronic configurations. Here, the structural characteristics of the GoU pairs are updated following the recent crystallographic structures of functional ribosomal complexes and the development in our understanding of ribosomal translation.


| INTRODUCTION
GoU pairs have been previously analyzed and reviewed (see, for example, Masquida and Westhof, 1 Varani and McClain, 2 and Ananth et al 3 ).
GoU pairs, first suggested by Crick 4 for decoding the third codon base and then called "wobble," are since regularly observed and predicted in RNA secondary structures, and folding programs include measured energy parameters 5 for GoU pairs. The importance of their functionality is emphasized by the high conservation of GoU pairs in critical positions in sequence alignments or in RNA structures or complexes. In selfsplicing catalytic RNAs, GoU pairs are at the cleavage site, for example, in group I introns. 6 In RNA complexed with proteins, GoU pairs can be determinant. 2,7,8 GoU pairs are also key recognition elements for small ligands (see, for example, Burgstaller et al 9 ). GoU pairs are critical for long-range packing interactions between RNA helices in crystals 10 and in the ribosome. 11,12 An extensive analysis of GoU pairs in ribosomal structures is presented in Mokdad et al. 12 This quick overview gives a glimpse of the multiple structural roles played by GoU pairs in RNA folding and recognition. Here, the focus is on the various arrangements between G and U that are observed in GoU pairs.

| THE USUAL GOU PAIRS
In all the examples above, the GoU pairs are canonical with the U protruding in the deep major groove of the RNA helix ( Figure 1). This minor movement is at the core of the structural and functional properties of the GoU pairs. The following characteristics follow: • GoU pairs are easily accommodated within regular RNA helices, with minimal distortions in the sugar-phosphate backbone.
• A GoU pair is not isosteric to a UoG pair, unlike the standard G═C and A─U pairs 4,13 ( Figure 1).
• The angle between the 5′GoU3′ and the following pair is undertwisted, and that between the 5′UoG3′ and the following pair is overtwisted 1 ; the stacking of a 5′GoU3′ pair with the following pair in the 3′ direction is therefore more pronounced than that of a 5′UoG3′ pair. 14 • For entering a helical stem, a 5′GoU3′ is thus more frequently • The slippage of the U into the major groove leaves a cavity on the minor groove side frequently occupied by a water molecule that links the O2′(U), O2(U), and the N2(G) (Figure 2). [15][16][17] • The displacement of that water molecule allows for a tighter packing with another base pair 11 or to the insertion of a protein atom in a protein complex. 18 • The displacement of the U also creates a binding site frequently occupied by a hydrated potassium ion with binding via hydration water molecules to O4(U), O6(G), and N7(G) in the major groove ( Figure 2). 19,20 With those usual GoU pairs, it is the departure from standard In nonhelical regions (internal or hairpin loops), GoU base pairs occur in unusual configurations called "bifurcated" where either of the uracil O4 or O2 carbonyl groups points directly to both the N1 amino and N2 amino of the guanine, for example, in the sarcin module, 22 the UNCG tetraloop, 23   Tautomeric forms of nucleic acid bases have been discussed from the early days of structural biology 28,29 and were recently reviewed. 30 In the past years, tautomeric GoU pairs have been implicated 31,32 but FIGURE 2 A wobble GoU pair (from PDB 4PCO 73 ) with a water molecule in the minor groove (green cross) and a hydrated ion (here a cobalt hexamine) in the major groove. A water molecule linking a phosphate oxygen, the N7(G), and an ammine group is also shown. All distances are in angstrom and between the heavy atoms, except for the two base-base H bonds Yokoyama. 56 More recently, NMR experiments support an anionic state of the modified 5-oxyacetic acid-uracil base. 59 It was therefore FIGURE 5 At the top is shown the usual GoU pair with the displacement of the U in the major groove, and below is shown the novel GoU pair with the U displaced into the minor groove. In the latter case, the electronic structure of the U could not be determined by X-ray crystallography, but the shape of the base pair was clearly indicated by the electron density. 35 In the drawing, we follow the choice made by Sochacka et al 58 where the negative charged is shown delocalized. In the crystal structure, 35 X = S and R = methylaminomethyl FIGURE 4 Two theoretical forms of tautomeric GoU pairs. At the top, the G adopts the enol form (and not the usual keto form), at the bottom, the U adopts the enol form (and the usual keto form). Crystallography cannot distinguish between these two possibilities. Physically, the two states are equivalent with the protons oscillating between the O6 and O4 oxygen atoms and between the N1 and N3 nitrogen atoms. Such tautomeric GoU pairs are isosteric between themselves like Watson-Crick G═C pairs. Such tautomeric base pairs are undistinguishable from G═C pairs through interactions in the minor groove side. 30 Such tautomeric Watson-Crick-like pairs have been observed by crystallography with natural bases 33,36,38,39 or modified Uridines 52,53 suggested that, like mnm 5 s 2 U34, cmo 5 U34 forms a minor groovedisplaced UoG pair and not a Watson-Crick-like tautomeric UoG pair. 52 Further studies are required to assess which modified U*34 adopts the minor groove-shifted UoG pair or whether depending on the type of U modifications some of the pairs adopt instead a Watson-Crick-like tautomeric geometry. Structurally, either of these two types of pairs is accommodated by the ribosomal grip and would allow for proper translation.

| CONCLUSIONS
The adaptable, and almost chameleon-like, behavior of GoU pairs is remarkable. By exploiting tautomerism, they can mimic the geometric and isosteric properties of standard complementary Watson-Crick pairs. Indeed, they can either adopt Watson-Crick-like pairs that appear like standard Watson-Crick base pairs or form GoU pairs that are isosteric between them (see Figure 6). In the latter case, the U must be modified at the C5 position. This adaptable potential of GoU pairs is critically necessary for smooth and efficient ribosomal translation. 48,[60][61][62] The nature of the chemical modification on the U base is, however, very diverse in the various organisms of the phylogeny 60,63 and reflects the historical contingencies of the evolutionary pro- In this respect, it is worthwhile noting that pseudouridines (Ψ) should resist tautomeric changes and solely exist in the diketo form. 66 Thus, a GoΨ pair is expected to occur in a single conformation, the standard wobble pair (the pair stabilizes tautomerism with altered states of both G and U, see Figure 4). Consequently, Ψ at the third position should stabilize a wobble pair (G34oΨ3), but Ψ34 could only promote translation with A(+3) (since Y34oG(+3) requires either a tautomeric form or a change in the electronic configuration of the pyrimidine). This situation occurs in eukaryotes where a second tRNA Ile carrying Ψ34 decodes the (rare) AUA3 codon. 61 Further, Ψ at the first and second positions should prevent miscoding (because tautomeric GoΨ are not expected to occur). In eukaryotes, tRNA Tyr is characterized by the presence of Ψ35. 61 The adaptable potential of GoU pairs is particularly key for a subset of codons. Indeed, for NNY codons, a G34 (modified or not) in tRNAs can decode either C(+3) or U(+3), which is not the case for NNR codons in two-codon boxes (Arg (AGR), Gln, Glu, Leu (UUR), Lys, Trp) that require a modified U34* to decode G(+3) without impairing decoding of A(3). Evolution exploits variations in codon usage and in tRNA species for diversifying the decoding range. 54,60,61,67,68 In order to escape from the dependence on modification enzymes acting on U34, two pathways are possible: One can either impose a strong selection against G-ending codons or duplicate the tRNAs so that the corresponding C34-tRNA is present for decoding G-ending codons. The first solution is seen in fungal mitochondria 69 and insect symbionts, 70 and the second solution is possibly present in a sea cucumber symbiont. 71 Such general evolutionary mechanisms have been observed in various organisms or organelles, but they could also take place within cells of multicellular organisms. Therefore, depending on the developmental stage or the cellular type, the G/C content of tran- The three flavors of GoU pairs discussed: At the top, the usual mode, the GoU wobble has the U displaced into the major groove; below (thick dark lines), the anionic mode, the GoU pair has the U (that must be modified) displaced into the minor groove, and at the right, the tautomeric mode, the GoU pair is isosteric to a G═C pair