Recognition of the Codon-Anticodon Helix by Ribosomal RNA

See allHide authors and affiliations

Science  10 Sep 1999:
Vol. 285, Issue 5434, pp. 1722-1725
DOI: 10.1126/science.285.5434.1722


Translational fidelity is established by ribosomal recognition of the codon-anticodon interaction within the aminoacyl–transfer RNA (tRNA) site (A site) of the ribosome. Experiments are presented that reveal possible contacts between 16S ribosomal RNA and the codon-anticodon complex. N1 methylation of adenine at position 1492 (A1492) and A1493 interfered with A-site tRNA binding. Mutation of A1492 and A1493 to guanine or cytosine also impaired A-site tRNA binding. The deleterious effects of A1492G or A1493G (or both) mutations were compensated by 2′fluorine substitutions in the mRNA codon. The results suggest that the ribosome recognizes the codon-anticodon complex by adenine contacts to the messenger RNA backbone and provide a mechanism for molecular discrimination of correct versus incorrect codon-anticodon pairs.

The fidelity of protein synthesis is determined by the interaction of an mRNA codon with the anticodon of the correct (cognate) transfer RNA (tRNA) within the aminoacyl-tRNA site (A site) of the ribosome. The ribosome distinguishes the correct codon-anticodon pair from all noncognate pairs. Despite the relatively low specificity of the codon-anticodon interaction, the measured fidelity of translation is about one error per 104 amino acids (1). The ribosome achieves high fidelity through a kinetic discrimination mechanism that couples codon-anticodon recognition on the ribosome with hydrolysis of guanosine triphosphate by elongation factor Tu (2). Rate constants for tRNA binding to the ribosomal A site are tuned by formation of a cognate or noncognate codon-anticodon pair (3). These processes suggest ribosomal recognition of the codon-anticodon pair within an active site for decoding (4). The molecular basis of this recognition is not known.

Highly conserved regions of ribosomal RNA (rRNA) form the tRNA-binding sites (5). The codon-anticodon interactions in the peptidyl-tRNA site (P site) and A site occur on the small (30S) ribosomal subunit (6). Fewer nucleotides, located primarily within the 1400 to 1500 region of rRNA, have been implicated in A-site tRNA binding than in P-site tRNA binding (6). This agrees with the affinity of a cognate tRNA for the A site being one-fiftieth that for the P site. RNA mutations in 16S rRNA affect the fidelity of translation (5), and aminoglycoside antibiotics that decrease the fidelity of translation (7) bind to 16S rRNA and perturb rRNA structure (8). These data suggest that the ribosome recognizes the codon-anticodon complex within a defined region of 16S rRNA in the 30S subunit.

Here, we describe experiments that map possible molecular contacts between 16S rRNA and the mRNA-tRNA complex in the A site. Chemical modification interference identified A1492 and A1493 in 16S rRNA as required for mRNA-dependent tRNA binding in the A site. Mutations of these two universally conserved nucleotides are lethal in Escherichia coli and decrease A-site binding affinity. The deleterious effects of A to G changes at these positions are compensated by 2′F modifications in the mRNA codon. The results support a model for ribosomal decoding in which A1492 and A1493 recognize the helical structure of cognate codon-anticodon complexes in the A site.

A selection scheme was developed to identify bases in 16SrRNA whose chemical modification disrupts A-site tRNA binding (Fig. 1A) (9). 3′-biotin-tRNAPhe (biotin- tRNAPhe) was directed to the A site by saturating the P site with E. colitRNAfMet at the first two codons of phage T4 gene32 mRNA (10). In the absence of tRNAfMet, biotin-tRNAPhe binds exclusively to the ribosomal P site in a mRNA-dependent fashion (11). A-site or P-site complexes were isolated by means of capture of biotin-tRNAPhe by streptavidin beads (10). Specific selection of biotin-tRNAPhe was revealed by toeprint experiments (12) on captured 30S subunits (Fig. 1, A and B). With saturating concentrations of tRNAfMet, only A-site complexes of tRNAPhe were captured; in the absence of tRNAfMet, P-site complexes were captured.

Figure 1

(A) Isolation of chemically modified or mutant 30S ribosomal subunits that retain their A-site tRNA-binding activity. Shine-Dalgarno sequence (SD) and codons for methionine (AUG) and phenylalanine (UUU) of T4 phage gene32 mRNA are indicated. Subunits whose binding affinity is not affected are captured with magnetic streptavidin beads, which bind to biotin-tRNAPhe. A-site complexes are formed by addition of tRNAfMet and biotin-tRNAPhe, whereas P-site complexes are formed by addition of only biotin-tRNAPhe. Stops of reverse transcription (RT) caused by 30S subunit in the toeprint experiment are indicated by a small and a large arrow. (B) Autoradiograph showing the toeprint stops for A-site and P-site complexes after capture with streptavidin beads. With excess tRNAfMet over 30S subunits, tRNAPhe binds exclusively to the A site, whereas in the absence of tRNAfMet, it binds to the P site. A and G are dideoxy sequencing lanes. Numbers at right indicate positions on the gene32 mRNA. (C) (Left) Base modifications of 16S rRNA that interfere with A-site tRNA binding. The autoradiograph shows a region of 16S rRNA around position A1483 (control), A1492, and A1493. 30S subunits were modified with DMS. Lanes 1 and 4 show primer extension of unmodified 16S rRNA, lanes 2 and 5 show the total population of modified 16S rRNA, and lanes 3 and 6 show the captured subpopulation of modified 16S rRNA. A-site and P-site labels indicate A-site and P-site selection, respectively. The lanes labeled A, C, G, and U are dideoxy sequencing lanes. (Right) The graph shows quantification of interference of A-site tRNA binding by modification of N1 positions of A1492 and A1493. Error bars are derived from at least three independent experiments. (D) Secondary structure of 16S rRNA showing critical nucleotides for A-site tRNA binding to 30S subunits (•). Methylation of G1494 (N7) only interfered with tRNA binding in the presence of 100 μM paromomycin (▴).

The chemical groups in 16S rRNA required for high-affinity A-site tRNA binding to 30S subunits were identified with this selection scheme. 30S subunits that were competent for A-site or P-site biotin-tRNAPhe binding after chemical modification were captured with streptavidin beads (10,13). Comparison of the modification levels of all bases in 16S rRNA in the total population and in the bound fraction of 30S subunits revealed critical bases for A-site tRNA binding. Only methylation of A1492 and A1493 N1 positions interfered with tRNA binding at the A site (Fig. 1, C and D). A1492 and A1493 are a subset of the bases protected from chemical probes by A-site tRNA (6). A1408 forms an internal loop structure with A1492 and A1493, but N1 modification of A1408 did not interfere with A-site tRNA binding. In the presence of the aminoglycoside antibiotic paromomycin, methylation of G1494 (N7) also interfered with A-site tRNA binding (Fig. 1D); G1494 (N7) is required for paromomycin binding (8), indicating coupling of drug and tRNA affinity in the A site. No interference was observed in the 530 loop after modification with different reagents. This conserved loop is likely involved in A-site tRNA binding indirectly (14).

To test the functional importance of A1492 and A1493, we expressed rRNA containing mutations at these two positions in E. coli(15). These changes included A1492G, A1492C, A1493G, A1493C, A1492G-A1493G, and A1492C-A1493C. All mutations at 1492 and 1493 confer lethal phenotypes when expressed in E. coli. Change of A1408, which is not universally conserved, to G produces functional ribosomes (16). Thus, A1492 and A1493 are essential for ribosome function in vivo.

The effects of 1492 and 1493 mutations on A-site tRNA binding were determined with the selection scheme described above (17). Mutant 30S subunits were purified as a mixture with wild-type subunits. Changes in P- or A-site tRNA binding affinity for the mutant will change the mutant/wild-type ratio after selection. G1492-G1493 and C1492-C1493 mutations were deleterious to A-site tRNA binding (Fig. 2A) as were single mutations at 1492 and 1493. Mutations at position 1492 decreased only A-site tRNA binding affinity, whereas those at 1493 or double mutations at 1492 and 1493 decreased both A- and P-site binding affinity. The effects on P-site binding were relieved by higher concentrations of P-site tRNA. 30S subunits containing a G1492 or G1493 (or both) mutation showed decreased protections at the N1 position of A1408 by A-site tRNA compared with that observed on the wild-type 30S subunits (18). The results demonstrate decreased affinity for A-site tRNA binding in these mutants.

Figure 2

The A-site selection was applied to mixtures of wild-type and two mutant ribosomes, G1492-G1493 or C1492-C1493 bound to gene32 mRNA containing either 2′OH, 2′F (A), or 2′H uridines (B). (A) Lanes 1 and 4 show the ratio of chromosomal and plasmid encoded 16S rRNA in the wild-type/mutant 30Ssubunit mixture before A-site selection. Lanes 2 and 5 and 3 and 6 show the 16S rRNA ratio observed with capture of 30S subunits with 2′OH or 2′F mRNA, respectively. (B) Lanes 1 and 3 show the ratio of the wild-type/mutant 30S subunits mixture before A-site selection. Lanes 2 and 4 show the 16S rRNA ratio observed with capture of 30Ssubunits with 2′H mRNA. (C) Bar graph representing the variations in the mutant/wild-type ratio after selection for A-site binding with mRNA containing either 2′OH, 2′F, or 2′H uridines. Error bars are derived from at least three independent experiments. (D) Autoradiograph showing kethoxal probing experiments with 30S subunits containing G1492-G1493 mutation. N1 position of G1492 was protected from kethoxal modification by binding of tRNAfMet, yeast tRNAPhe, and 2′F mRNA. Lane 1, unmodified 30S subunits; lane 2, 30Ssubunits + 2′F mRNA; lane 3, 30S subunits + 2′F mRNA + tRNAfMet ; lane 4, 30S subunits + 2′F mRNA + tRNAfMet + tRNAPhe (200 pmol); lane 5, 30S subunits + 2′F mRNA + tRNAfMet + tRNAPhe (500 pmol); lane 6, 30S subunits + 2′OH mRNA + tRNAfMet; lane 7, 30S subunits + 2′OH mRNA + tRNAfMet + tRNAPhe (200 pmol); and lane 8, 30S subunits + 2′OH mRNA + tRNAfMet + tRNAPhe (500 pmol). In all experiments, 10 pmol of 30S subunits was bound to 30 pmol of 2′F or 2′OH gene32 mRNA and with or without 50 pmol of tRNAfMet in the total volume of 12.5 μl. Modifications were monitored by primer extension from a DNA oligonuclotide primer complementary to priming site V (16).

The modification-interference and mutational data suggest rRNA chemical groups for interaction with the tRNA-mRNA complex in the A site. In the structure of a model RNA oligonucleotide, the N1 positions of A1492 and A1493 point into the minor groove of the RNA (8). Methylation of the N1 positions of A1492 and A1493 and mutations to guanine remove a hydrogen bond acceptor, the adenine N1 position. The hydrogen bond donor likely resides in the mRNA. A-site tRNA binding is strictly mRNA dependent (19), and the 2′OH groups of the mRNA are important for translation (20). Substitution of 2′OH groups with deoxynucleotides in the A-site codon decreases the binding affinity of the cognate tRNA but does not affect P-site tRNA binding (21). These data suggest that the 2′OH groups within the codon may interact with rRNA in the A site, possibly through hydrogen bonds to the N1 positions of A1492 and A1493.

Recognition of the codon-anticodon complex through adenine N1-2′OH interactions was tested experimentally by exchange for guanine NH1-2′F interactions (22). The energetics of C-F–H-N hydrogen bonds are not known, but crystal structures have demonstrated their formation (23). The A-site selection was applied to mixtures of wild-type and two double mutant ribosomes (G1492-G1493 and C1492-C1493) bound to gene32 mRNA containing either 2′OH, 2′H, or 2′F uridines (24) (Fig. 2, A and B). The mutant/wild-type ratio reveals the relative effects of rRNA or mRNA changes on A-site binding affinity. As shown above, decreases in the mutant/wild-type ratio were observed for the G1492-G1493 and C1492-C1493 mutants with 2′OH mRNA. An increase in the mutant/wild-type ratio was observed only for the G1492-G1493 mutant with 2′F mRNAs (Fig. 2, A and C). Unlike with 2′OH mRNA, the C1492-C1493 mutant with 2′F mRNA and the G1492-G1493 and C1492-C1493 mutants with 2′H mRNA showed no change in the mutant/wild-type ratio after selection. Because these mutations disrupt A-site tRNA binding with the 2′OH mRNA, the results suggest equally deleterious effects of rRNA mutations and mRNA backbone changes on A-site tRNA-binding. 30S subunits containing a single G1492 or G1493 mutation gave results similar to those of the G1492-G1493 mutant.

The selection results with mutant ribosomes and 2′F mRNA were confirmed by footprinting. Addition of tRNAfMet and tRNAPhe with 2′F mRNA to G1492 and G1493 mutant ribosomes protected G1492 N1 and N2 from kethoxal, whereas no protection was observed with 2′OH mRNA. A similar protection at G1493 N1 and N2 was obtained with G1493 mutant ribosomes (18). These results strongly suggest that the N1 position of adenines 1492 and 1493 contacts two 2′ OH groups of an A-site bound mRNA codon.

To ensure translation of all codon-anticodon pairs, the ribosome should contact the codon-anticodon complex in a sequence-independent manner (8). A1492 and A1493 contact 2′OH groups in the mRNA codon, which forms an A-form duplex with a cognate tRNA anticodon; similar recognition of base-pair shape occurs within the active site of DNA polymerase (25). The small energetic effects of rRNA and mRNA modifications on A-site tRNA affinity are consistent with the dominant contribution of the codon-anticodon interaction to A-site tRNA binding (3). Aminoglycoside antibiotics, which distort the structure of A1492 and A1493 and the surrounding helix (8), increase the affinity of A-site tRNAs by favoring contacts between the codon-anticodon helix and these nucleotides (7). The interactions between rRNA and the codon-anticodon complex are probably more intricate than a pair of hydrogen bonds between A1492 and A1493 and the mRNA backbone (26) and may involve a network of interactions between rRNA and the codon-anticodon duplex, as observed in group I introns (27).

The backbone contacts described here suggest a molecular mechanism for ribosomal discrimination between correct (cognate) and incorrect (near or noncognate) codon-anticodon pairs. The correct codon-anticodon duplex is a two–or three–base pair stretch of A-form helix. Molecular contacts between A1492 and A1493 and the codon-anticodon duplex (Fig. 3A) decrease the dissociation of cognate tRNA and may trigger additional conformational changes in the 530 loop (14) and 900 regions (28) of 16S rRNA, which are transmitted to the 50Ssubunit (3). Incorrect codon-anticodon duplexes contain mispairs within the first two positions that distort their A-form geometry and the positions of critical hydrogen bonding groups for recognition by A1492 and A1493. Disruption of rRNA contacts (Fig. 3B) increases the dissociation rate of incorrect tRNAs and decreases conformational signaling rates (3). The ribosome thus distinguishes between different helical structures formed by correct and incorrect codon-anticodon complexes during decoding.

Figure 3

Schematic model for ribosome mRNA-tRNA interaction in the A site. (A) The N1 positions of A1492 and A1493 in 16S rRNA can contact two 2′OH in the mRNA codon. The interaction between the cognate tRNA and mRNA places the codon in the proper orientation to interact with the ribosomal A site and leads to efficient conformational signaling. (B) The interaction between near-cognate tRNA and mRNA leads to a mispair in the first two positions of the codon-anticodon helix, which distorts its structure, disrupting N1 contacts and decreasing conformational signaling.

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


View Abstract

Navigate This Article