Technical Comments

Mechanism of Ribosomal Peptide Bond Formation

See allHide authors and affiliations

Science  12 Jan 2001:
Vol. 291, Issue 5502, pp. 203
DOI: 10.1126/science.291.5502.203a

The mechanism of peptide bond synthesis constitutes a fundamental and long-debated question in molecular biology. For many years, ribosomologists championed a protein-based mechanism, similar to the charge relay system that has been proposed for peptide hydrolysis by serine proteinases [references in (1)]. As it has become apparent that the peptidyl transferase center is composed mainly of RNA, however, two likely mechanisms for catalysis have emerged that are compatible with the available biochemical data: divalent metal ion catalysis (1), or acid-base catalysis mediated by a cytosine (N3) or adenosine (N1, N3). The environment of the catalytic nucleotide would create the unusal higher pK a (whereK a is the acid dissociation constant) necessary for it to behave analogously to the histidine of the serine proteinases. This pK a shift has been shown for a catalytic cytosine in the active site of the hepatitis delta virus ribosome (2).

In support of the acid-base catalysis hypothesis, Muth et al. (3) have demonstrated that the highly conserved nucleotide A2451 (Escherichia coli numbering) in the active site has a pK a shifted to a value of around 7. In addition, in a 2.4 Å x-ray map of the large ribosomal subunit, Nissen et al. (4) show that the N3 of this nucleotide is in close contact with a transition state analog of peptide bond formation, CCdA-p-Puro. These results were combined into a model in which A2451 functions both in general acid-base catalysis and in transition state stabilization (3, 4).

Although this evidence is seductive, a definitive assignment of A2451 as the catalytic nucleotide must be treated with caution, for three reasons. (i) The structure observed with the transition analog CCdA-p-Puro might not be an active one, because small analogs of P site tRNAs require different reaction conditions (5). Furthermore, the dA substituted for the terminal A in the analog should interact differently with the active site and has indeed been shown to be inactive as a P-site substrate (6). (ii) A2451 was the main cross-link site of a P-site bound Phe-tRNA whose amino group was acylated by a benzophenone derivative (7). Significantly, the cross-linked tRNA was still active in peptide bond formation (8). (iii) A chloramphenicol-resistant mutant exists that harbors an A2451-to-U transversion (9).

Taken together, these data argue against A2451 being the sole catalytic nucleotide, which in turn leaves open the question whether the catalysis of peptide bond formation is mediated by catalytic nucleotides, divalent metal ions, or both. The answer to this question will come from biochemical experiments, coupled with high-resolution x-ray analysis of active ribosomal 50S subunits containing maps of the relevant metal ions and water molecules. We eagerly await the results.


The generation of a model for the molecular structure of the large ribosomal subunit, from x-ray crystallographic studies at 2.4 Å resolution, is one of the most exciting biological advances in recent years (1). Of particular interest is the identification of the active site for peptidyl transfer based on the determination of the structure of a complex with a transition state analog (2). A key feature of this active site is a completely conserved adenosine that is proposed to act as a general base catalyst. The elegant study of Muth et al. (3), which revealed that this adenosine has a markedly shifted pK a near 7.6, supports the importance of this base.

Based on these results, Nissen et al. (2) have proposed a mechanism that involves the deprotonation of the amino group of aminoacyl-tRNA (aa-tRNA), leading to the nucleophilic attack of the deprotonated amine on the carbonyl group of the ester linkage holding the growing polypeptide chain. Such a mechanism, however, has three significant difficulties: (i) The amino group would be expected to be largely protonated (that is, present as a primary ammonium group, −NH3 +) at neutral pH. The pK a value for a terminal amino group is expected to be approximately 8.0. (ii) Even with its shifted pK a, the adenosine is not nearly a strong enough base to deprotonate an amino group, −NH2, that is expected to have a pK a near 35. Based on the difference of more than 25 pK a units, the rate of deprotonation of the amino group would be much too slow to support catalysis. (iii) The amino group, −NH2, would presumably be a sufficiently strong nucleophile to attack the ester linkage without the need for assistance from a base.

We propose an alternative mechanism that avoids these difficulties (Fig. 1). In this mechanism, the aa-tRNA is present initially in its protonated form, which resolves difficulty (i). The ribosomal base (adenosine 2451) then removes a proton from this ammonium group to generate the free amino group. The pK a values for the base and the ammonium group are expected to be nearly matched, so deprotonation should be quite feasible thermodynamically; this resolves difficulty (ii). The reaction products are the aa-tRNA with a free amino group and the protonated adenosine 2451 base. The amino group is a strong enough nucleophile to attack the ester linkage of the peptidyl-tRNA, which resolves difficulty (iii). This reaction generates a tetrahedral intermediate that collapses to release the tRNA previously linked to the polypeptide chain. The protonated adenosine 2451 can act as a general acid to facilitate this reaction, as suggested by Nissen et al. (2). This generates the product with the new peptide bond formed, but in an N-protonated form. The pK a of this product is expected to be very low, so that it would readily give up a proton to generate the final product.

Figure 1

Alternative mechanism for peptide bond formation within the ribosome.

The mechanism proposed by Nissen et al. (2) and the alternative proposed here allow distinct experimental predictions. Most important, if a nucleotide with a dramatically altered pK a is indeed required to deprotonate an amino group, mutation of this base or its surroundings should abolish the ability of the ribosome to catalyze peptide bond formation completely. In contrast, in the alternative mechanism, the role of the base is to deprotonate an ammonium group with a pK a value expected to be near 8. Loss of this base would be expected to reduce the rate of this step by a relatively small (but functionally important) factor of between 5 and 100. Further experimental studies should provide additional data that should help distinguish between these two mechanisms, as well as other possible ones.


Response: Although the mechanistic proposals advanced in our studies (1, 2) are hypotheses that need to be examined critically, we do not believe that the points raised by Barta et al. refute them.

In our crystal structures, both the substrate analog bound to the A-site and the CCdA-p-Puro intermediate analog hydrogen bond with residues identified as crucial for A-site and P-site binding, and the puromycin moieties of both occupy the same position. Hence, there is no compelling reason to believe that the CCdA-p-Puro complex is inactive; indeed, the opposite seems likely, because the one A in 23S rRNA that has an anomalously high pK a is positioned so that it could catalyze peptide bond formation if CCdA-p-Puro were properly oriented. There is no reason to assume that we are studying an inactive 50Ssubunit conformation. We have found small analogs of P-site tRNAs that do not require alcohol to react (2, 3), and preliminary crystallographic experiments suggest that they do indeed react in our crystals to yield products whose binding is consistent with our published complexes.

The suggestion that the absence of a 2′ OH from A76 of the CCdA-p-Puro might cause it to bind abnormally is interesting, but seems unlikely. Our model suggests that a 2′ OH in that position could hydrogen bond with C2104 without much change in the position of the inhibitor. Findings reviewed by Chládek and Sprinzl (4) do not support the conclusion that the 2′ OH at A76 is essential for P-site binding. Finally, when the CCdA in the P site is bonded to the puromycin of the A site, as it is in this case, the binding effect of the missing 2′ OH should be reduced.

Barta et al., citing earlier work (5, 6), also point out that ribosomes carrying Phe-tRNA, whose amino group is cross-linked to the A that we propose is critical for catalysis, remain active. This argument has two important deficiencies. First, it hinges on assumptions about the rate of the chemical step in peptide bond formation. Although the overall rate of ribosomal peptide elongation is reasonably well known, its chemical step is not rate limiting—and there is no convenient way to measure its rate in the ribosome. The assay done to demonstrate the activity of the cross-linked ribosomes required 15 min (5), enough time for a ribosome to make a polypeptide 1.8 × 104 amino acids long. A rate reduction of as much as 102 or 103 due to modifications of the A in question would not have been detected.

Second, in 1983, Barta and Kuechler (5), using 3-(4′-benzoylphenyl)propionyl-Phe-tRNA (BP-Phe-tRNA), did indeed obtain cross-links that targeted 23S RNA exclusively when bound to what they believed was the P site, and they also showed that the cross-linked ribosomes were active in peptide bond formation (5). A year later, the cross-linked residues were identified as U2584 and U2585 (7). Four years after that, however, Steiner et al. (6), using optimized protocols, reported that BP-Phe-tRNA bound to the P-site cross-links to A2451 and C2452, and that the U2584 and U2585 cross-links reported earlier represent A-site binding. The peptidyl transferase activity assay of the crosslinked ribosomes was not repeated as far as we know, and hence it is not known whether ribosomes cross-linked on A2451 possess peptidyl transferase activity. The activity of U2584/U2585 cross-link products, although surprising, does not conflict with our model of catalysis. Indeed, it effectively rules out the possibility that these uridine residues have any other than secondary roles in the activity, which is exactly what we have proposed.

Finally, Barta et al. cite a chloramphenicol-resistant mutant that contains an A2451-to-U transversion (8). This also is not a crucial test of the role of A2486 (2451), however. As already noted, interpretation of these in vivo experiments requires knowing the rate of the chemical step compared with the overall rate of peptide bond synthesis. Moreover, as reported in (2), we have made all possible substitutions of A2451 in a plasmid encoding for 23S rRNA, and find that when expression from plasmids is induced in E. coli, they show a dominant lethal phenotype.

Concerning the alternative model, metal ion catalysis, our electron density maps contain abundant information about the positions of metal ions and water molecules in the large ribsomal subunit—and, although our analysis is still incomplete, we see no evidence for a metal ion at the catalytic site. Thus, we conclude that none of the work cited by Barta et al. casts doubt on the proposal that A2486 (2451) plays a catalytic role, although future experiments may of course do so.

The issue raised by Berg and Lorsch is one that we have also considered. Which α-amino group proton, if any, is the base A2486 (E. coli 2451) removing? Although the pK a of an α-amino group is about 8 in solution, its pK a might be higher or lower when aminoacylated-tRNA is bound to the ribosome. In the case of elongation factor Tu binding of aminoacyl-tRNA, the co-crystal structure establishes that the amino group is in the −NH2 form in this complex (9). The influence of the large ribosomal subunit on the pK a of this α-amino group is unknown, however, and thus the possibility should be considered that A2486 removes a proton from the −NH3 + state of the α-amino group, as Berg and Lorsch suggest. Although it is not likely that the contribution of A2486 to the rate of catalyses using this latter mechanism would be greater than about 10, it is not obvious that the expected contribution to the rate of catalysis of other possible mechanisms (transition state stabilization or removal of a proton from the −NH2 state) would necessarily be much greater.

One reason we have favored the mechanism proposed in (1) is that the protein synthesis reaction is the reverse of the acylation reaction of serine proteases, and thus their mechanisms may be chemically related. In the acylation step of chymotrypsin, for example, His57 is proposed to assist in the removal of a proton from Ser195 as it attacks the carbonyl carbon of the peptide bond being cleaved forming a tetrahedral carbon intermediate (10). In the breakdown of this intermediate, a protonated His57 is proposed to deliver a proton to the substrate peptide −NH, which leaves to form the −NH2product. The third proton needed to form the −NH3 + is presumed to come from water. If A2486 functions in peptide synthesis analogously to His57, one might anticipate that the reverse of the serine protease acylation reaction and peptide synthesis should follow similar mechanistic pathways. The pK a of the α-NH2 is indeed high, but this pK a becomes greatly reduced as the nitrogen attacks the carbonyl carbon. Thus, removal by A2486 would get progressively easier as the C-N bond is being formed. In any event, it appears that this proton would be removed from the −NH2either by water or by A2486.

Furthermore, Fahnestock et al. (11) showed many years ago that ribosomes will catalyze the formation of ester bonds between the formylmethionyl moiety of formylmethionyl tRNA and a puromycin derivative in which the α-amino group is replaced by a hydroxyl group. Although the pK a of the hydroxyl group is much higher than that of the −NH3 + it replaces, the pH/rate profiles for the puromycin and the hydroxypuromycin versions of the reaction are nevertheless the same (11).

The second role for a protonated A2486 proposed in (1) is transition state or intermediate stabilization interaction with the oxyanion of the tetrahedral carbon intermediate. Indeed, this is the interaction observed in the crystal structure of the large subunit complex with CCdA-p-Puro. We can, of course, provide no quantitative estimate of the contribution made by A2486 to general base catalysis (whichever proton is removed), general acid catalysis, or transition state stabilization. Perhaps the largest contribution to the catalysis of peptide bond formation is provided by the positioning of the two substrates by the ribosome in an optimal orientation for attack of the α-amino group of aminoacyl-tRNA on the carbonyl carbon of peptidyl tRNA. Only further mutagenic, kinetic, and structural experiments can address these issues.


Navigate This Article