Crystal Structure of a Conserved Ribosomal Protein-RNA Complex

See allHide authors and affiliations

Science  14 May 1999:
Vol. 284, Issue 5417, pp. 1171-1174
DOI: 10.1126/science.284.5417.1171


The structure of a highly conserved complex between a 58-nucleotide domain of large subunit ribosomal RNA and the RNA-binding domain of ribosomal protein L11 has been solved at 2.8 angstrom resolution. It reveals a precisely folded RNA structure that is stabilized by extensive tertiary contacts and contains an unusually large core of stacked bases. A bulge loop base from one hairpin of the RNA is intercalated into the distorted major groove of another helix; the protein locks this tertiary interaction into place by binding to the intercalated base from the minor groove side. This direct interaction with a key ribosomal RNA tertiary interaction suggests that part of the role of L11 is to stabilize an unusual RNA fold within the ribosome.

The most important functional sites of ribosomes are associated with highly conserved domains of ribosomal RNAs (rRNAs); among these are the mRNA decoding site, the peptidyl transferase, and the site at which two elongation factors (EF-Tu and EF-G) alternately bind to position tRNA in the A site or to catalyze translocation of the mRNA (1). The latter site is associated with two extraordinarily well conserved regions of the large-subunit rRNA: the 17-nucleotide (nt) sarcin/ricin loop, and a 58-nt domain associated with ribosomal protein L11. Both of these RNA fragments are targeted by antibiotics, attesting to their functional importance. The 58-nt domain is particularly interesting: The sequence derived fromEscherichia coli 23S rRNA adopts a specific, though marginally stable, tertiary structure that is further stabilized by L11 protein, itself highly conserved, or thiazole antibiotics (2, 3). We report here the crystal structure of a complex between this 58-nt rRNA fragment and the 76-residue RNA binding domain of L11, at 2.8 Å resolution. The RNA is tightly folded by an extensive set of tertiary interactions that is locked into place by direct interaction of the protein with a key tertiary feature. This creates an unusually large interior of stacked bases and exposes a surface of conserved bases available for interaction with EF-G or antibiotic.

The RNA binding domain of Bacillus stearothermophilus L11 (L11-C76) was crystallized with an E. coli rRNA fragment containing a mutation that stabilizes tertiary structure (Fig. 1, A and B) (4), and its structure was determined by multiple wavelength anomalous diffraction (MAD) of a selenomethionyl derivative (5) (Table 1). Two complexes are present in the asymmetric unit and differ in only very minor ways. This suggests that crystal packing has not significantly perturbed the structure of the complex.

Figure 1

Components of the L11-C76–rRNA complex. (A) Sequence of the B. stearothermophilus L11-C76 protein fragment (19, 25). Secondary structure determined by NMR is marked, and colored residues are >80% conserved among 34 bacterial and plastid sequences (green) or additionally conserved in 22 archaeal and eukaryotic homologs (red). (B) Secondary structure of the 1051 to 1108 rRNA fragment of E. coli 23S rRNA as determined by comparative sequence analysis. Position 1061 has been mutated from U to A to stabilize tertiary structure. Bases in open typeface are conserved in greater than 97% of bacterial, archaeal, and eukaryotic rRNAs in a database of ∼480 aligned sequences (8). The base triple C1072(C1092/G1099) was suggested by comparative sequence analysis and supported by compensatory mutations (26). (C) Secondary and tertiary structure of the same RNA. Color coding of bases is as in (B). Thin black lines with arrows indicate 5′→3′ direction of the backbone, black horizontal bars are Watson-Crick base pairs, dots are non–Watson-Crick base pairs, open horizontal bars are tertiary bonds between bases, and thin red lines indicate uridine N3 to phosphate hydrogen bonds characteristic of the U-turn motif. (D) Stereo view of the folded RNA and protein backbone. Mg2+ (green) and Os(NH3)6 3+(magenta) are shown. The RNA bases in (B), (C), and (D) are color-coded for clarity. Abbreviations for the amino acid residues are as follows: A, Ala; D, Asp; E, Glu; F, Phe; G, Gly; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; R, Arg; S, Ser; T, Thr; and V, Val.

Table 1

: Summary of crystallographic statistics.

View this table:

The RNA fold is stabilized in three locations by tertiary contacts between bases (Fig. 1, C and D). The junction loop is closed by helix B, which contains five canonical pairs and is terminated by a side-by-side G·A pair. From this point the RNA diverges into two helices. On one side, the reverse Watson-Crick U1082/A1086 base pair, previously deduced by comparative sequence analysis and compensatory base mutations (6), stacks onto the neighboring A helix. The three-base loop closed by U1082/A1086 forms a U-turn, a common motif for reversing chain direction (7). This turn places A1085 in position to hydrogen bond to G1055 in the minor groove of helix B. Both these bases are 99% conserved among sequences from the three phylogenetic domains (Bacteria, Archaea, and Eucarya) (8), suggesting their importance for stabilizing the structure. The other strand of helix B extends into helix C, which is interrupted by a two-base bulge loop. The first of these bases, A1088, is in the syn conformation and intercalates into helix A, where it forms a Hoogsteen pair with U1060; both of these bases are nearly invariant among the three phylogenetic domains. This pair is a key feature for protein recognition, discussed below. The second base of the bulge loop, A1089, lies in the same plane as A1090 with A1089 N3 hydrogen-bonded to A1090 N6 in what might be called an “A-A sidestep.” Two more base triple interactions, involving bases from the helix A hairpin loop, stack on top of the A1089-A1090-U1101 triple (Fig. 2). An identical “A-A sidestep” has been seen in the “A platform” motif (9), where it also supports base triples but in a different way than seen here.

Figure 2

Stereo view of the stack of three base triples. Shown is a 2F o F celectron density map contoured at 1.0σ, calculated with the nine nucleotides in both RNA molecules omitted from the model (11% of the atoms in the asymmetric unit).

The structure of the helix C hairpin was previously solved by nuclear magnetic resonance (NMR) methods in the context of a 12- or 14-nt hairpin (10). It has essentially the same structure when part of the larger RNA domain, including a characteristic U-turn (Fig. 1, C and D). A third U-turn is found within the A helix hairpin. It places A1067 at the top of the turn, with G1068 stacked underneath and the rest of the hairpin loop extending deep within the structure to make contact with helix C. This is an unprecedented configuration for a hairpin loop, which depends on nonhelical torsion angles about the A1070 5′O-P, C4′-C5′, and 3′O-P bonds that unstack A1070 and reverse the chain direction. As a result, the highly conserved A1070 appears in a crevice between the A and C helices, where it stacks with A1061. It was previously thought that the U1061→A mutation stabilized RNA structure indirectly by preventing incorrect Watson-Crick pairing with A1077 (11). It now appears that this explanation is only partly correct, and that stacking with A1070 directly stabilizes the tertiary structure as well.

A pronounced feature of the overall RNA fold is the large interior core formed by stacked planes of bases. U1060, A1088, A1089, A1090, and U1101 all lie in a plane across the middle of the molecule, and additional planes of three or four bases are stacked above or below (Fig. 1D). This extensive stacking may account for the large heat capacity change of unfolding this RNA, which is larger than seen for any other nucleic acid and similar in magnitude to that seen for unfolding of protein hydrophobic cores (2). Two Mg2+ ions have been located within the interior core (Fig. 1D) and may correspond to two ions that strongly stabilize tertiary structure in solution (12).

The RNA contact surface of L11-C76 has previously been defined by NMR experiments to comprise the entire surface of helix 3, much of loop 2, and a central portion of loop 1 (13). Footprinting (14) and a compensatory mutation study (15) had suggested that the RNA contact surface would be centered about the conserved U1060 and A1077 nucleotides within helix A. These expectations are borne out by the crystal structure, which shows α-helix 3 against the minor groove surface of helix A (Fig. 3B). In the two known cases of α helices binding in an RNA major groove, the α helix is deep within the concave surface of the groove (16). In contrast, the helix A minor groove is relatively flat, in part due to distortions in the middle of the helix, and α-helix 3 is lying on its surface.

Figure 3

Interactions of the RNA fragment. (A) Schematic of protein-RNA contacts. Bases lettered in red are ≥97% conserved among rRNA sequences. Protein residue contacts with RNA are coded by color: green, hydrogen bonds; red, electrostatic contact of basic residue with nonbridging phosphate oxygen; blue, nonpolar contact. Contacts are with amino acid side chain, unless noted as O (backbone carbonyl) or NH (backbone amide). Note that the backbone reverses direction at U1060, and that A1088 derives from another part of the RNA. (B) Backbone of L11-C76 on RNA surface. Many of the side chains indicated in (A) are shown. (C) Nucleotides potentially interacting with thiostrepton or EF-G. The RNA is shown as a space-filling model in two approximately orthogonal views. The sites of chemical protection by EF-G (N1; blue) and a 2′ OH where methylation confers thiostrepton resistance (red) are highlighted.

The unusual structure of helix A is a crucial factor in protein recognition. Insertion of A1088 into helix A requires that U1078 be turned out of the helix. U1078 then forms a stack with A1077 and G1062 across the width of the helix, sharply changing the direction of the helix axis (Fig. 3, A and B). On the opposite strand of the helix, A1061 is bulged and the backbone twists U1060 into a reverse polarity, placing the major groove edge of U1060 in the helix minor groove. A similar reversal of backbone polarity was seen in the Rev-RRE complex (16). The nearly invariant bases U1060, A1088, and G1059 are hydrogen-bonded to invariant parts of α-helix 3: The hydroxyl of Thr66, which only varies to Ser, is hydrogen-bonded to O4 of U1060; the carbonyl of the invariant Gly65 bonds with N6 of A1088; and the Met62carbonyl is bonded to N2 of G1059. The α carbon of Gly65is packed against G1059 N2 and C1079 O2 and 2′O, which explains why no other amino acids are found at this position. Mutation of either Gly65 or Thr66 to alanine drastically reduces L11 binding affinity, demonstrating the importance of this set of hydrogen bonds (17). The invariance of these base–amino acid contacts accounts for the interchangability of L11 homologs between different organisms (18).

L11 dramatically stabilizes the entire tertiary structure of the 58-nt rRNA domain (19). The reason for this stabilization is now clear: By binding A1088 directly and stabilizing the helix A conformation that accommodates A1088, L11 locks a key tertiary interaction into place. Direct protein contact with a critical tertiary interaction has not been seen in other RNA-protein complexes and suggests a role for ribosomal proteins in stabilizing correct rRNA tertiary folds.

Besides Gly65 and Met62, a third carbonyl-base hydrogen bond takes place between Ser69 and G1062; it is conserved in nearly all bacteria. But in Synechocystis and plastid rRNAs, U1062/A1076 or C1062/U1076 substitutions have pyrimidine O2 in place of guanine N2. L11 sequences from these organisms have Asn in place of Ser69, and this change has been shown to partially compensate for the deleterious effects of a U1062/A1076 substitution (15). On that basis, a specific contact between Ser69 OH and G1062 N2 was proposed. Instead, Ser69 OH bonds to both Thr66 carbonyl and G1062 2′OH, and it is the Ser69 carbonyl that interacts with G1062 N2. Whether an Asn69 side chain bonds to U1062 or some other accommodation takes place in plastid ribosomes remains to be seen.

At either end of α-helix 3, loop 1 or loop 2 of L11-C76 closely approaches the helix A surface and makes a number of contacts, all but one of which are with the backbone (Fig. 3, A and B). Additional backbone contacts are made by α-helices 1 and 2 with the 5′ strand of helix A. Thus, L11 uses α-helix 3 to recognize conserved bases, and other parts of the protein to recognize the distorted shape of the helix A backbone.

The adjacent U1066 and U1094 U-turns expose the bases of A1067 and A1095 to solvent and bring them into proximity at one end of the RNA; the invariant base A1070 protrudes in a cleft between the two loops (Fig. 3C). A1095 and A1070 are conserved in more than 97% of sequences from the three phylogenetic domains, and position 1067 is always a purine (8). The conservation of these exposed bases is difficult to rationalize in terms of the RNA structure, which suggests that they are required for interactions with other components of the ribosome. Considerable biochemical evidence in fact points to these two loops as interaction sites for both thiostrepton and EF-G. A1067 and A1095 are associated with thiostrepton resistance in vivo or altered thiostrepton binding in vitro (20). Simultaneous interaction of thiostrepton with both loops—for example, by binding in the cleft between the loops—would account for its ability to stabilize the same (or very similar) rRNA tertiary structure as recognized by L11. The NH2-terminal domain of L11 stimulates thiostrepton binding to the rRNA (19) and would be positioned nearby this cleft.

Thiostrepton inhibits the activity of EF-G (21), and this protein factor has also been localized to the U1066 and U1094 U-turns by various chemical methods (Fig. 3) (22). In direct binding assays, EF-G has been seen to interact with an 84-nt rRNA fragment containing the 58-nt sequence described here (23). Thus, direct contact of EF-G with either or both of the U1066 and U1094 U-turns seems likely. The bases protected by EF-G from chemical reaction are on the opposite face of the RNA from the proposed thiostrepton binding site (Fig. 3), suggesting that the factor and antibiotic bind at adjacent sites within this RNA.

As more ribosome functions are attributed to rRNA, it seems likely that a main function of ribosomal proteins is to stabilize required rRNA conformations. The fact that the E. coli rRNA fragment has a marginally stable tertiary structure in the absence of L11 supports this idea (2) and parallels observations that specific proteins trap active RNA conformations of some ribozymes (24). This look at a conserved protein-RNA complex from the ribosome will serve as a basis for more precise functional and structural analysis of the mechanisms by which L11, EF-G, and thiostrepton promote or inhibit ribosome activity.

  • * To whom correspondence should be addressed. E-mail: draper{at} and lattman{at}


View Abstract

Navigate This Article