Release 4, September 10th, 2010



RIBOSOMAL ACCURACY BLOG
- Or : Where have all the reviewers gone ?

JACQUES NINIO

( included in the web site http://www.lps.ens.fr/~ninio )



Contents:

1-Introduction
2-List of articles commented in this document
3-Comments
4-Notes
5-References


1-INTRODUCTION

It seems that a large number of recent articles on ribosomal accuracy were not handled by competent reviewers. Too many false assertions are circulating about well established experimental results (e.g., what Gorini really showed), and there are severe distorsions about well known theories (e.g., what is really an induced-fit mechanism). There are unsupported claims in the interpretation of otherwise valid kinetic experiments (the conclusions were in fact in the simplifying mathematical assumptions). Articles by high level scientists sometimes contain mathematically erroneous equations, which the authors do not accept to correct, or reactions schemes that are at variance with thermodynamic principles.

In such a situation, a person having faith in science would expect that there would be vigorous debates, with attacks by one camp and refutations or counter-attacks by another camp. Instead, what is happening, article after article and review after review is the emergence of a “negotiated truth”. It is similar in spirit to these official announcements that are issued by heads of states in conclusion to a ritual international meeting, announcements in which every word has been weighted up, announcements that seem to represent a deep consensus and a major progress for the benefit of mankind, and are there in reality to fool honest citizens. This situation is particularly damaging to students who are fed without any warning with false data, self-contradictory theories, or absurd mathematics. It is damaging to people who work on topics like logistic models of protein synthesis or codon usage and unknowingly use absurd kinetic parameters. It is damaging to medical scientists concerned with the pathologies that are due in part to rare events producing protein variants, or due to metabolic defects that impinge on post-transcriptional tRNA modifications.

This is not to deny the fact that there are also excellent scientific articles (for instance, the Kramer and Farabaugh 2007 article commented below) and that there have been fantastic breakthroughs in structural determinations (by crystallography, and by cryo-electron microscopy), in biophysical single molecule studies (using FRET, for instance), and in fast kinetic studies, using fluorescent markers. I understand that scientists who spent so much time in developping the high technology techniques, and spent so much time collecting and interpreting the results are allowed to let their imagination run. They have the right the speculate and give their own views about how the ribosome works. Nevertheless, this is not a reason to ignore basic established facts. Of course, some old « established facts » may have been misleading, and perhaps they were artefacts, or perhaps the old facts will, in the long run, be interpreted differently in the light of the new results. Still, these are not reasons to distort the content of old experimental or theoretical publications. Given the current situation, my first advice to students is « read the quoted original publications, and never trust what the quoting authors make them say. » My second advice is « read also the unquoted articles, they are often more informative than the publicized ones ». As a striking example, I invite students to read the series of articles by Leendert Bosch and co-workers on Elongation Factor Tu mutants. They were left aside in the past, because they did not fit in the conceptual framework of the 1990’s. Fully understanding these mutants (taking into account their subtle, paradoxical properties) is a main challenge for a modern theory of translation mechanisms.



2-LIST OF ARTICLES COMMENTED IN THIS DOCUMENT

Bilgin, N., Claesens, F., Pahwerk, H. and Ehrenberg, M. (1992) Kinetic properties of Escherichia coli ribosomes with altered forms of S12. J. Mol. Biol. 224, 1011-1027.

Döring, V. and Marlière, P. (1998) Reassigning cysteine in the genetic code of Escherichia coli. Genetics 150, 543-551.

Faxen, L.A., Kirsebom, L.A. and Isaksson, L.A. (1988) Is efficiency of suppressor tRNAs controlled at the level of ribosomal proofreading in vivo ? J. Bacteriol. 170, 3756-3760.

Geggier, P., Dave, R., Feldman, M.B., Terryl, D.S., Altman, R.B., Munro, J.B. and Blanchard, S.C. (2010) Conformational sampling of aminoacyl-tRNA during selection on the bacterial ribosome. J. Mol. Biol. 399, 576-595.

Kramer, E.B. and Farabaugh, P.J. (2007) The frequency of translational misreading errors in E. coli is largely determined by tRNA competition. RNA 13, 87-96.

Lavrik, I.N., Avdeeva, O.N., Dontsova, O.A., Froeyen, M. and Herdewijn, P.A. (2001) Translational properties of mHNA, a messenger RNA containing anhydrohexitol nucleotides. Biochemistry 40, 11777-11784.

Lee, T.H., Blanchard, S.C., Kim, H.D., Puglisi, J.D. and Chu, S. (2007) The role of fluctuations in tRNA selection by the ribosome. Proc. Nat. Acad. Sci. USA 104, 13661-13665.

Ninio, J. (2006) Multiple stages in codon-anticodon recognition : double-trigger mechanisms and geometric constraints. Biochimie 88, 963-992.

Sharma, D., Cukras, A.R., Rogers, E.J., Southworth, D.R. and Green, R. (2007) Mutational analysis of S12 protein and implications for the accuracy of decoding by the ribosome. J. Mol. Biol. 374, 1065-1076.

Schmeing, T.M. and Ramakrishnan, V. (2009) What recent ribosome structures have revealed about the mechanism of translation. Nature 461, 1234-1242.

Vaiana, A.C.and Sanbonmatsu, K.Y. (2009) Stochastic gating and drug-ribosome interactions. J. Mol. Biol. 386, 648-661.

Zaher, H.S. and Green, R. (2010) Hyperaccurate and error-prone ribosomes exploit distinct mechanisms during tRNA selection. Molecular Cell 39, 110-120.



XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

3-COMMENTS

Bilgin, N., Claesens, F., Pahwerk, H. and Ehrenberg, M. (1992) Kinetic properties of Escherichia coli ribosomes with altered forms of S12. J. Mol. Biol. 224, 1011-1027.

This is a most important article. It provides an almost complete set of kinetic parameters for various sections of the elongation cycle. All the parameters are biologically plausible.

Contrary to what they had expected, the authors found that « altered ribosomes, which were previously identified as aggressive proofreaders, have the same stoichiometry of GTP hydrolysis to peptide bond formation as do wild-type ribosomes ». They can be applauded for chosing to reveal these findings which contradicted their earlier interpretations.

Furthermore, they found that it is « a ribosome-idling reaction occurring without EF-G, which is drastically amplified in SmD in relation to wild-type ribosomes ». It is striking, they note, that « the ribosome-dependent idling GTPase activity on EF-Tu is reduced to wild-type level for all variants » when streptomycin is added. They were extremely puzzled by these potentially crucial findings, which, to my knowledge, no one attempted to confirm or refute. And, to my knowledge, Ehrenberg never quoted the result in his later publications.

I conjectured, in Ninio 2006, that the idle GTPase reaction was perhaps a sign that the kinetic amplification scheme in Ninio, 1975 (which uses a backward idle cleavage reaction) is effectively used on the ribosome, in conjunction with the classical Hopfield 1974 mechanism.



Döring, V. and Marlière, P. (1998) Reassigning cysteine in the genetic code of Escherichia coli. Genetics 150, 543-551.

This work was part of an ambitious project on engineering bacterial strains having an altered genetic code. Although the theme of codon-anticodon recognition was not central to the work, it contains important data concerning this topic, which seem to have escaped the specialist’s attention. The results contained in Table 2 of the article can be condensed as follows :
 


Anticodons


AAU

GAU

UAU

CAU

Codons





AUU

+

++

-

-

AUC

-

++

-

-

AUA

+

-

++

+

AUG

-

-

-

++



These are in vivo results. The anticodons were grafted on a tRNA for cystein. Their post-transcriptional modification status is unknown, and deserves being determined in future studies, particularly in the case of UAU which appears here unable to read AUG. Note that the very rare codon AUA may be read through the third position odd pairs A.A and C.A. This finding suggests that AUA is particularly ambiguous in E. coli, thus explaining perhaps its disuse.

Faxen, L.A., Kirsebom, L.A. and Isaksson, L.A. (1988) Is efficiency of suppressor tRNAs controlled at the level of ribosomal proofreading in vivo ? J. Bacteriol. 170, 3756-3760.

Here is what I wrote about this article in my review on translation mechanisms (Ninio, 2006):

An article by Faxén, Kirsebom and Isaksson [ref.] played an extremely important role in this work. I will stress here the qualities of this article, because it has all the attributes which would make it unpublishable to-day in the so-called "high impact factor journals". First of all, it was deeply honest. Instead of making a narrow selection of their experimental observations, and using such a selection to promote simple-minded theories that would have appealled to a broad audience of researchers from other fields, they laid down all the results of their genetic experiments on the table. Next, part of the data was in line with current theories, others were not, but the authors interpreted their work in an extremely modest way. This attitude would make it impossible to "sell" the work as a revolution in the field. Above all, the work was extremely systematic. It covered, in a consistent way nonsense suppression, codon context effects, streptomycin effects, ribosomal mutations, and anticodon modifications. All the clues were there, but hard to decipher. After all, the ribosome is one of the most complex achievements of molecular evolution. While progressing in the comprehension of the ribosomal puzzle, I often returned to the work of Faxén et al. [ref.] to determine if at this point their data became more intelligible. While the proposals made in this review do not explain in detail all their results, they provide at least sufficient leads to envision tentative explanations.

Geggier, P., Dave, R., Feldman, M.B., Terryl, D.S., Altman, R.B., Munro, J.B. and Blanchard, S.C. (2010) Conformational sampling of aminoacyl-tRNA during selection on the bacterial ribosome. J. Mol. Biol. 399, 576-595.

This work is a great step forward in the characterization of tRNA processing by the ribosome using the FRET technology. The authors are now able to follow 50 ribosomes in parallel, taking pictures at intervals of 10 milliseconds. Each ribosome was followed on a sequence of about two thousand tRNA binding events. As a consequence, not only average times are measured, but complete reaction time distributions can be described for the transitions between different ribosomal states. The results are interpreted with extreme caution. Although the time resolution has been improved by a factor 10 with respect to previous studies, the authors insist on the necessity to further improve their time resolution, as many initial events on the ribosome occur in the 1 ms to 5 ms time scale. The J. Mol. Biol. Editor, D.E. Draper must have faced a difficult situation, because this work needed to be encouraged as a foundation for much of the future work in the field, yet there is no appealing take-home conclusion to the work. He took, I feel, the correct decision.

The FRET technology allows the authors to distinguish 3 states of tRNA binding, in addition to the “zero” state. These states would correspond, roughly, to a an initial codon recognition state, a GTP hydrolysis state, and an accomodation + peptide bond formation state. Fluoerscence transfer was studied in parallel between a dye attached to the elbow position of the A site tRNA, and dyes attached to the P site tRNA, and the large subunit ribosomal protein L11. The signals obtained with respect to the two anchoring points are mutually consistent.

I am particularly satisfied with the fact that the mathematical treatments (in the Supplementary Materials) are correct.

There is a very curious convergence between this article, and the Zaher and Green 2010 article commented below. In both papers, the results are presented in the light of a kinetic scheme which involves an “initial codon-independent binding step“, and in both articles this step is in practice dismissed in the calculations, as though the authors did not really believe in this “initial binding“.

Geggier et al. describe “initial binding” as follows: “Initial binding of ternary complex to the ribosome is mediated by interactions between EF-Tu and the C-terminal domain of ribosomal protein L12 located on the 50s subunit. This codon-independent contact localizes ternary complex to the leading edge of the ribosome to facilitate entry of the tRNA anticodon into the A-site decoding site”. This step is captured in their working model (Fig. 6 of Supplementary Material, scheme C) by a reversible reaction with apparent k_on and k_off kinetic constants. I have nothing to say against the description of ternary complex admission to the ribosome described above. The crucial question is how the tRNA leaves the ribosome? Is it through the reversal of the admission sequence, as proposed by Gromadski and Rodnina? If it were the case, the ribosome would behave in an extremely stupid manner, because after rejecting a tRNA (through the k_off) it would process it again and again (through the k_on).

In their mathematical analyzes Geggier et al. (see the Supplementary Material, Table 1), compute everything as though k_on and k_off did not exist, and the real tRNA exit was given by the codon-dependent k_2. This is reasonable, but clearly incompatible with the Gromadski and Rodnina scheme. Actually, the authors have under-exploited their data bearing on this issue. After what they consider as a tRNA departure event - as a matter of fact, a return to the zero FRET state, either the tRNA did leave the ribosome, as it is implicit in their treatments, or it came back to the initial binding stage, ready for a next incorporation attempt. Since they are able to follow the succession of 2000 incorporation attempts by a single ribosome, the distribution of time-intervals between apparent exits and apparent admissions can be determined, and this can perhaps throw light on the initial binding issue. Or perhaps, they would need to construct an experiment addressing specifically this question, playing with the ternary complex concentrations of cognate and near-cognate tRNAs.

Symmetrically, they include - to my great satisfaction - a tRNA rejection branchpoint before GTP cleavage, assign a kinetic constant k7 to it and ... take it equal to zero in their subsequent treatments. Do they believe in the existence of this branchpoint? In their Discussion, page 590, bottom of 1st column, they write that “the existence of premature aa-tRNA dissociation pathways in the selection process” raises serious theoretical problems, which they solve by postulating “passive, thermally accessible relaxation processes that rapidly return the system to ground-state configuration(s) following aa-tRNA dissociation”. Once again, how the ribosome returns to the ground state after a tRNA rejection appears to me to be a key question in translation accuracy.

Looking very closely at their initial kinetics, I am somewhat surprised by the relative positioning of the accomodation and GTP hydrolysis curves. Improved time resolution is required here. The authors also note a discrepancy but only in the case of the near-cognate case.

However, it may also be the case that much of the earlier kinetic studies, which made use of simple kinetic models have reached their limit. What is represented with a single arrow in a kinetic model may be an oversimplified representation of a situation in which tRNA moves between two states according to a very large number of trajectories. This was studied in computer simulation for the early binding events (see Vaiana and Sanbonmatsu, 2009, commented below) and for the CCA end migration just before peptide bond formation (Whitford et al., 2010). Both articles offer, in my opinion, a much more physically realistic picture of ribosomal processing than current linear models. These complexities are not necessarily reflected in the reaction time distributions (see Bel, Munsky and Nemenman, 2010).

On the other hand, it is my feeling that much of the current high technology work is putting the cart before the horses. More time should be spent in learning from very elementary but highly systematic experiments, in order to characterize various types of errors, determine which drug is effective against which class of errors, which class of error is sensitive or not to EF-Tu concentrations, etc. For technical limitations that are mysterious to me, the protein synthesis systems used in the FRET experiments have low speeds and very low accuracy. So, something essential is missing or going wrong in these systems. Understanding why should be a top priority before engaging into new ambitious projects.

Kramer, E.B. and Farabaugh, P.J. (2007) The frequency of translational misreading errors in E. coli is largely determined by tRNA competition. RNA 13, 87-96.

This is an outstanding contribution to codon-anticodon recognition, not reflected in the title, and hardly reflected in the summary, which are both too modest. Using the very elegant luciferase assay, the authors determined « the whole range of possible misreading errors by a single tRNA species at a single site ». A luciferase gene was inactivated by substituting an essential lysine codon at position 529 by a missense or a nonsense codon. Some luciferase activity could be measured due to misreading of the substituted codon, by a tRNA for lysine, with anticodon UUU. Kramer and Farabaugh determined the misreading levels relative to 24 different codons. It is the systematic character of this experiment which gives it its value. There are so little data on the decoding rules in vivo ! There are so many extrapolations from experiments in which at most 2 or 3 interactions were studied! With 24 results side by side, you are compelled to think differently.

Furthermore, the authors show how the reading patterns are modified in the presence of streptomycin or paromomycin, and the effect of high accuracy ribosomal context (rpsL). Such systematic work is reminding of the little known but seminal work of Faxen, Kirsebom and Isaksson (1988).

The data provided by Kramer and Farabaugh (see in particular their Table 1) are exceptionally complex, and challenging for theoreticians. So, what are we waiting for ? Perhaps we woudl like to have more data of the same kind : Misreading by one or two other tRNAs at 20 or 30 different codons. Having these data would be fantastic. The problem is that the publication system is becoming more and more perverse. The criteria of selection by some of the so-called high impact factor journals have nothing to do with the scientific quality of the artcles, they discourage serious systematic work, and are an almost open incitation to fraud.

I will single out here two results in the Kramer and Farabaugh article.

One is a side result shown in their figure 4. They show that the misreading levels at two codons is reduced by 50% when a competing cognate tRNA is overexpressed. The result goes in the expected direction. But is this twofold effect a mere reflection of the change in the in vivo concentration of the cognate tRNA ? If not, and the in vivo concentration had been in fact multiplied by three or more, this would be a clue for hidden kinetic complexities on the ribosome.

Another curious result is the way misreading levels vary in a ribosomal mutant rpsD context : « The only misreading error greatly affected by the rpsD mutant was wobble misreading of AAU, which increased almost ninefold, while the effect on the other codons was no more than twofold. »

In my recent discussion of the ribosomal accuracy mutants (Ninio, 2006), I suggested that rpsD favoured « shortcut events », that is misincorporations occurring when a tRNA being processed by the ribosome is subject to premature dissociation, and is replaced by another tRNA which therefore skips the earlier codon-anticodon recognition stage (See Note [2] in this document). I pointed out, in the same review that « late » codon-anticodon recognition could be almost blind to third position misreading. So, the rpsD enhanced misreading of AAU might tell us something important about the late stages of codon-anticodon recognition.

Note in this respect that « only those near-cognate codons frequently misread during normal translation showed increased misreading in rpsD mutants ». This observation reinforces the link between rpsD misreading and shortcut events. Actually, Kramer and Farabaugh showed that normal misreading correlated with cognate tRNA scarcity. When the cognate tRNA is not abundant, there will be a relatively large number of near-cognate or non-cognate binding events, leading to a large number of abortive incorporation events, which in turn will create favourable conditions for shortcut incorporation events.

List of codons used in the Kramer-Farabaugh study :

AUU, AUC, AUA (Ile), AUG (Met)
ACU, ACC, ACA, ACG (Thr)
UAA, UAG (Term)
CAA, CAG (Gln)
AAU, AAC (Asn)
GAA, GAG (Gln)
UGA (term)
CGU, CGC, CGA, CGG (Arg)
AGA, AGG (Arg)

Farabaugh and co-workers have now produced very informative data, in the same spirit, this time in yeast (Kramer et al., 2010).

Lavrik, I.N., Avdeeva, O.N., Dontsova, O.A., Froeyen, M. and Herdewijn, P.A. (2001) Translational properties of mHNA, a messenger RNA containing anhydrohexitol nucleotides. Biochemistry 40, 11777-11784.

Here Lavrik and co-workers show that they can translate codons having a very unusal sugar-phosphate backbone. The ribose is replaced by hexitol, a six-membered ring. This article takes on importance in the context of the current debate on the role of the sugar-phosphate backbone in codon-anticodon recognition.

The authors compared the coding properties of three nucleic acids :

M0, a 31 nucleotide long synthetic mRNA which contains a Shine-Dalgarno sequence, and an internal AUG initiating codon followed by a UUC phenylalanine codon.

M1, a modified form of M0 in which the riboses of the UUC codon are replaced by hexitol molecules.

M2, a modified form of M0 in which both the AUG and the UUC codons have their riboses replaced with hexitol.

Their studies allow them to conclude that « the 2’-OH function in mRNA is not necessary for binding and movement through the ribosome. Groove shape recognition of the codon-anticodon complex, more than hydrogen-bond interactions of ribose residues in mRNA, is an important factor for correct translation ».

This work appears to me to be methodologically excellent and intellectually rigorous. It should be taken seriously into account in forthcoming discussions of small groove interactions in codon-anticodon recognition on the ribosome.

Earlier work relevant to the same theme can be reminded :

McCarthy and Holland (1965) which demonstrates DNA translation by the ribosome.

Potapov, Triana-Alonso and Nierhaus (1995) on the importance of the 2'OH group in translation by the ribosome.



Lee, T.H., Blanchard, S.C., Kim, H.D., Puglisi, J.D. and Chu, S. (2007) The role of fluctuations in tRNA selection by the ribosome. Proc. Nat. Acad. Sci. USA 104, 13661-13665.

Using fluorescent probes and FRET measurements, Lee and co-workers studied tRNA selection by ribosomes. Their conclusion, in favour of an induced-fit mechanism, rests almost entirely on the low value of a forward kinetic constant (k3) which they proposed for the nearcognate interaction at 5mM magnesium concentration. However, they warn the reader that this constant has “large uncertainties” because of the limited data set.

Their kinetic scheme (Fig. 3) involves both a reversible tRNA exit from the “low FRET” state, through the reversal of the incorporation path (straight arrow and kinetic constant k-2) and an irreversible tRNA exit from the same state (curved arrow and kinetic constant k-2’ ). Therefore, this state has no definable energy level, and the scheme cannot work, unless an energy-coupling reaction is associated to the “irreversible” exit or to all premature tRNA dissociation events (Hopfield, 1974; Ninio, 1975).

The two exit pathways should have qualitatively different effects on accuracy, yet they are lumped as a single constant (k-2all or k-2,nall) in the theoretical equation of Fig. 4 which is, unfortunately, incorrect. The probability P that a tRNA reaches the “mid FRET” stage is given, according to published methods (Ninio, 1986, 1987), by :

P = [k2/(k2+k-1)]x[(k3/(k3+k-2all) + k-2 P/(k3+k-2all)] (1).

Discrimination between cognate and nearcognate tRNAs at equal binding rates is the ratio of the P’s for the two competitors. The equation shown in Lee et al.’s Fig. 4 could have been obtained by replacing inadvertently k-2 in Eq(1) here, by k-2all.

This paper should be given to analyze to students in molecular biology and biophysics, asking them, in particular, to derive the equation. It will be very interesting to have statistics on their reactions. How many students will be proud of supplying their own mathematical derivations of the erroneous equation ? How many will derive instead the correct equation ? And how many will see why, obviously, the published equation cannot be correct, without going into the derivations ? Or better: the students should ask their university professors to derive the equation. (It can be done in two lines, using probabilities, as explained in Ninio, 1986).

It is gratifying to note that in a subsequent article (Geggier et al., 2010, analyzed in this blog) Blanchard and co-workers use correct mathematics, and discuss the problem created by the early irreversible tRNA exit.



Ninio, J. (2006) Multiple stages in codon-anticodon recognition : double-trigger mechanisms and geometric constraints. Biochimie 88, 963-992.

In this extensive review (30 pages, 213 references), I attempted, among other things to see how the kinetic proofreading/amplification mechanisms could be implemented on the ribosome, in the light of recent progress. I conjectured that a substantial fraction of errors in protein synthesis occurred through “shortcut events’ - events in which a tRNA skips the early steps of the incorporation pathway (see Note [2]).Subsequent reviews on translation accuracy rarely mention the existence of my 2006 review. The most recent reviews merely repete current catechism, without ever mentionning the existence of divergent views. (The authors of a review may of course promote vigorously their ideas, but not even mentioning the existence of alternative ideas is unfair).

Here is the summary of my 2006 review :

Thirty years of kinetic studies on tRNA selection in the elongation cycle are reviewed, and confronted with results derived from various sources, including structural studies on the ribosome, genetic observations on ribosome and EF-Tu accuracy mutants, and codon-specific elongation rates. A coherent framework is proposed, which gives meaning to many puzzling effects.

Ribosomal accuracy would be governed by a "double-trigger" principle, according to which the ribosome uses energy in the forward direction to create new configurations for tRNA selection, and energy in the backward direction to regain its initial configuration, in particular after a premature dissociation event. The conformation energy would come in part, in Hopfield's mode, from GTP cleavage on the ternary complex (TC). The reset energy would be provided in part, in the author's mode, from GTP cleavage on a binary EF-Tu.GTP complex (BC). There would be several paths for amino acid incorporation. The path of highest accuracy would involve TC binding followed by BC binding, followed either by GTP hydrolysis on the TC, or by TC dissociation and GTP hydrolysis on the BC.

Codon-anticodon recognition would occur in at least three kinetically and geometrically distinct stages. In a first stage, there would be a very rapid sorting of the TCs with unstrained anticodons contacting a loosely held mRNA. This stage ends with the anchoring of the codon-anticodon complex by a cluster of 3 nucleotides of 16S RNA. The second stage would be the most discriminative one. It would operate on the 5 msec time scale and terminate with GTP cleavage on the TC. The third stage would provide a last, crude selection involving "naked" aa-tRNA, partially held back by steric hindrance. Streptomycin and most EF-Tu mutants as well as high accuracy ribosomal mutants would produce specific alterations at stage 2, which are mapped on the stage 2 kinetic mechanism. The ram ribosomal ambiguity mutants, and anticodon position 37 modifications could be markers of stages 1 and 3 selection.

Dissociation events at stage 2 or stage 3, when they are not immediately followed by reset events create a leaky state favourable to shortcut incorporation events. These events are equivalent to an "error-prone codon-anticodon mismatch repair". From the recent evidence on ribosome structure, it is conjectured that the L7/L12 flexible stalk of the large ribosome subunit acts as a proofreading gate, and that the alternation of its GTPase activation center between "TCase" competence and "BCase" competence is a main factor in the control of accuracy.



Sharma, D., Cukras, A.R., Rogers, E.J., Southworth, D.R. and Green, R. (2007) Mutational analysis of S12 protein and implications for the accuracy of decoding by the ribosome. J. Mol. Biol. 374, 1065-1076.

Can we explain the properties of ribosomal accuracy mutants in terms of the recent ideas on ribosomal mechanisms (induced-fits, domain closure, etc.) ? This article attempts to do it, but leaves me unconvinced.

The article by Sharma et al. studies a number of ribosomal high fidelity mutants, mutated in the gene for ribosomal protein S12. They identify precisely the amino acid substitution, and locate it on a 3d structure of the ribosome. So far, there is no problem. They also give a scenario for the role of S12 in ribosomal accuracy – S12 would have a role in transmitting information on the status of the current codon-anticodon association on the 30s subunit to the GTPase center on the 50s subunit (OK, I do not necessarily agree, but I do encourage authors to speculate). They also provide precise kinetic parameters that characterize wild type and mutant ribosomes. More precisely, they provide, for each ribosome a typical GTPase rate constant, an accomodation rate constant, and a « saturation » value. Very good, we need precise data of this kind. I merely regret the absence of data on GTP cleavage stoichiometries.

The article also contains data on the effects of paromomcin and streptomycin. The way these two antibiotics affect the decoding properties was examined, and there are detailed results on how the two drugs bind to the ribosome in various concentration ranges. Excellent, this is solid work, combining structural and kinetic experiments.

The authors decided to interpret their results in terms of the induced-fit model of codon-anticodon recognition. Oops, this is becoming tough, because induced-fit models are definitely incompatible with Gorini’s data on high accuracy ribosomal mutants. (See Note [1] in this document).

More precisely, the authors attempt to interpret their results within the Gromadski and Rodnina 2004 framework. The Gromadski-Rodnina scheme is displayed in their Fig. 2a, with names for the kinetic constants, but no numerical value attached to them.

At this stage, you would think, the authors have all the cards in their hands for the Great Synthesis : They have precise structural information, they have kinetic parameters for all their mutants, and they have a kinetic scheme in which they believe. So, they have just one more thing to do now, feed the kinetic scheme with their parameters, and show that everything fits !

As a reviewer, I would have asked the authors to do the calculations and show either that everything fits or that things do not fit, and say it clearly to the reader. I do hope that the situation will be cleared up in the next publication on ribosomal accuracy mutants.

My own analysis of high accuracy ribosomes (Ninio, 2006) indicates that what is crucial with these ribosomes, is what they do after a tRNA falls off the ribosome, and the A site becomes vacant (See Note [2] in this document). By design, the Sharma et al. experiments do not probe these situations, so they are blind to the kinetic parameters which are, in my opinion, critical for accuracy. The Sharma et al. work, by design, could not investigate properties 1, 2 and 4 of Gorini's mutants as defined in Note [1] of this document. Its results are in line with property 3. So, much work lies ahead. 

Schmeing, T.M. and Ramakrishnan, V. (2009) What recent ribosome structures have revealed about the mechanism of translation. Nature 461, 1234-1242.

When my friend Henri Grosjean sent me this article, I feared the worse. I thought that the article would merely repeat the usual catechism. It took me an entire week before I found the courage to read the article. Actually, I had the agreable surprise to find that the doctrine of the Church is evolving.

Francis Crick used to say "I am not a wobblist myself". After publishing the wobble hypothesis, he never returned to the subject again, except modestly and tangentially in articles on the origin of the code, and he never wrote a review article on the topic.

He was a reviewer for my article submitted in 1970 to the Journal of Molecular Biology "The missing triplet hypothesis" in which I proposed that "the situation is less unsymmetrical with respect to the three positions of the codon-anticodon association than assumed by the wobble hypothesis" and the article was published thanks to his positive review - which of course does not imply adhesion to my views, but implies to the very least, that there was in these times a climate of tolerance towards heretical ideas.

A number of events convince me that the climate has changed in recent years, that there is much censorship and self-censorship going on, and this motivated me to initiate this ribosomal accuracy blog.

The part of the review by Schmeing and Ramakrishnan over which I have competence is that concerned with the mechanisme of elongation, from page 1235 to the first paragraph of page 1237, references 26 to 44.

Since much of what I have to say on the subject can be found either in Ninio, 2006 or elsewhere in this blog, I will make here only a few brief comments.

Schmeing and Ramakrishnan use at several places arguments about how energy is being used by the ribosome. Perhaps someone should, at this point, transcribe their ideas in terms of a reaction diagram, and annotate the diagram with the postulated energy levels along the various reaction pathways. In the absence of such a reaction scheme with its associated energy values, I cannot say whether the proposals make sense or not. For instance, my intuition about "domain closure" is that it is damaging to accuracy.

The part of the review which was for me an agreable surprise was at the end of the elongation secion, where Schmeing and Ramakrishnann write:

"In the recent experiments (ref 43 to Ehrenberg and co-workers), the accomodation of tRNA into the PTC is apparently too fast to allow significant discrimination by proofreading after GTP hydrolysis. If so, the structural basis of how one could have such a high accuracy with little or no proofreading is not clear, nor why measured in vivo rates of misincorporation are so much higher (reviewed in ref. 29 -to Ogle and Ramakrishnan 2005). Further experiments with other reporters and in varying conditions are required to clarify these differences".

There is an important turning point, here. For the first time, I think, Ramakrishnan acknowledges the existence of controversial points.

About proofreading, it is clear that the original ideas of the 1970's needed to be updated and adapted to the recent developments on ribosome mechanisms. I attempted to do this in Ninio, 2006. It would be fair to discuss kinetic proofreading in terms of the updated analyses. In the 2006 review I demonstrated that so far, the evidence in favour of kinetic proofreading mechanisms had been rather indirect. A much more direct way of testing the kinetic proofreading ideas would be to study shortcut events (Note 2 at the end). These are tRNA binding events which would occur right after a tRNA rejection, and before a return of the ribosome to its initial state. A susbtantial fraction of the amino acid incorporation errors occur, I propose, through shortcut events. All the tools to study shortcut events are available, or should be easy to construct. Please feel free to consult me if you would like to design a system for studying these events. There are plenty of possibilities. And, I believe, a lot to discover about codon-anticodon recognition ahead.



Vaiana, A.C.and Sanbonmatsu, K.Y. (2009) Stochastic gating and drug-ribosome interactions. J. Mol. Biol. 386, 648-661.

Usually, I pay little attention to molecular simulation studies. Although the predictive power of these studies is constantly improving, how can they compete with high-resolution crystallographic data, when such data are available? Here, this work is a happy exception, because it focuses on a "blind spot" of the crystal structures, and it frames the so-called induced-fit models of codon-anticodon recognition into a physically sound perspective. On the surface, this work seems to be about the mechanism of binding of the error-inducing drug gentamicin to the decoding site of the ribosome. But its real value is in what it says about the decoding site in general.

We learnt from crystallographic and other structural studies that there is on the ribosome a kind of "reading head", and that two residues of the 16S rRNA subunit, A1492 and A1493 are, at one or more stages of the decoding process, flipped out from their helix, and the adenines are making contact with the small groove of the codon-anticodon mini-helix.

Both the flipped-in and the flipped out states of the two adenines have been observed. From this people have deduced a linear sequence of events for codon-anticodon recognition. The ternary complex would bind to the codon while the two adenines are in the flipped-in state, then the adenines would switch to the flipped out state, then depending on the quality of the interaction, the ternary complex would remain on the ribosome or be ejected. The alternative model, in which, when the ternary complex binds to the ribosome, the two adenines are either flipped in or flipped out has very few advocates.

The "then-then" linear engine model of the ribosome is now coated with a doctrine according to which the then-then sequence of events is in fact an "induced-fit". Furthermore, this induced-fit would be, in large measure, responsible for ribosomal specificity. This cannot be correct, because induced-fits, as defined by enzymologists (see, e.g., Koshland, 1958 and all subsequent enzymological papers on the subject), are either-or mechanisms: Two alternative states of the enzyme are accessible to the competing substrates. And all enzymologists know that induced-fits do not create specificity. Whatever the specificity achieved by an induced-fit enzyme, it is the specificity related to the binding energies of the substrates to the enzyme in the productive complexes. There is no way to avoid these chemical kinetic constraints, and as far as I know nobody dared to produce an induced-fit model of the ribosome, with an associated energy diagram. One thing I learnt from Leslie Orgel when I was working in his laboratory in 1972-1974 is that although "either-or" mechanisms (e.g., the random binding of two substrates) look more complicated on paper than the "then-then" ones (e.g., sequential binding) they are in fact less constraining and far more natural, physically.

As far as I know, there is not a single piece of experimental evidence demonstrating that ternary complex binding occurs through a then-then mechanism. In fact the then-then mechanism is a tacit assumption of all mathematical treatments of the results, not a result in itself.

The Vaiana and Sanbonmatsu article begins with an excellent introduction. Induced-fits are defined with rigour by explaining the reaction paths, and the associated energy constraints. The introduction also describes a model with the same diagram topology, but different constraints on the reaction paths, the "stochastic gating" model (something like Kurland's "conformation selection"). They write, page 649:
"For the purposes of this discussion, we define stochastic gating as the limit that the binding site switches between these configurations much faster than ligand binding. Likewise we define induced fit as the opposite limit, where the timescale of the binding site conformational change is similar to or slower than ligand binding."

From their simulations Vaiana and Sanbonmatsu derive a picture in which the two adenines, in the absence of gentamicin move extremely rapidly and independently. Futhermore there are multiple paths connecting the flipped in and flipped out states. The energy landscapes in which the adenines move "appear rugged, and with several local minima, connected by higher free-energy pathways. The morphologies of the two landscapes appear to be very different, although the free-energy values at the minima and heights of the barriers connecting them are comparable" (second column, page 650).

Their picture is much closer to the "stochastic gating" model than to the "induced-fit" one. Concerning gentamicin binding, they propose a striking metaphor: "This process is analogous to "jiggling" a key until the correct orientation is found to fit into the keyhole" (first column, page 655).

The authors have now good hopes of being able to simulate tRNA binding to the decoding site. I hope that they will carry out this work with the same independance of mind as that shown in the present work.

The main limitation of their work, I think, is that they are not still able to connect their results with kinetic models: "It is not possible to directly calculate rates by counting events" (first column, page 650). For me there is a very challenging theoretical problem here, beyond the mere practical agreement between experimental and theoretical rates. In the simulations, there are countless states. How to reduce this complexity to a few states, and a simple kinetic diagram which would capture the gist of the scene? With four states, we can already have induced-fits and stochastic gating, but what do we loose? 

Zaher, H.S. and Green, R. (2010) Hyperaccurate and error-prone ribosomes exploit distinct mechanisms during tRNA selection. Molecular Cell 39, 110-120.


This work on ribosome accuracy mutants, published three years after the previous one (Sharma et al., 2007 - see the detailed comments above) was much awaited. The results form a nice, complete set, and offer a timely alternative to Rodnina's work. The conclusions to be derived from the work are not yet clear, due to the unreliability of the methods used to analyse the kinetic data.


The discovery of the ribosome accuracy mutants about 40 years ago by Gorini and co-workers, and Apirion and co-workers played an enormous role in sharpening our ideas on the accuracy of molecular processes. Much work had been performed on the properties of these mutants, culminating, on the genetics side, with the 1988 Faxén et al. paper, and on the biochemical side with the 1992 Bilgin et al. paper. Both papers are commented in this translation accuracy blog. The most conspicuous biochemical correlate of the rpsL mutants according to Bilgin et al. is an increase in the so-called "idle" GTPase reaction. An important, rarely quoted paradoxical result on the ribosome accuracy mutants was published in 1999 by Björkman et al: The combination of a high accuracy mutation in protein S4 with a high accuracy mutation in protein S12 gives a double mutant with a standard level of accuracy! Several insights from the "old" work could be derived (see my 2006 Biochimie review), for instance that ram exerted its effect at an earlier stage than the rpsL mutant, although the effect of ram could also be sustained through late stages. It also appeared that rpsL mutants exerted their effects at a stage in which the presence of EF-Tu was required, so rpsL mutants clearly did not seem to play a role after the dissociation of the ternary complex from the ribosome.


Here, Zaher and Green compare the properties of a low accuracy ribosome (rpsD, or ram) and a high accuracy ribosome (streptomycin-dependent, rpsL) with wild type ribosomes. They do so using Rodnina-type mathematics (fitting kinetic data with one or two parameters when the model counts 6 or more parameters, both of the forward and the backward type).


Zaher and Green start by introducing, in their Fig. 1 a "simplified kinetic scheme of the tRNA selection pathway", inspired from Rodnina (e.g., Gromadski and Rodnina, 2004). The scheme starts with a so-called "non-specific initial binding" of the ternary complex to the ribosome, with associated kinetic parameters k1 and k-1. As I pointed out earlier (Ninio, 2006) this non-specific initial binding step "kills" the discrimination produced by the codon-recognition step. Note that the problem is not with "non-specific binding" in itself, it is with making non-specific binding an obligate step in getting out of the ribosome. This and other problems were also recognized by Ehrenberg and co-workers who wrote, with exquisite politeness: “So, if the model and its parameter values are basically correct, why did evolution not bring it from a suboptimal performance with low accuracy and small elongation rate to an optimal performance with with much higher accuracy and faster protein synthesis?” (Johansson, Lovmar and Ehrenberg, 2008).


Although the initial binding step is conceived as "non-specific" - inasmuch as the kinetic parameters k1 and k-1 are postulated to be independent of the quality of the codon-anticodon interaction, it has a profount effect on ribosomal accuracy. This may sound paradoxical, but this type of effect (of a non-specific kinetic constant on the accuracy of a selection process) was described with appropriate mathematics in Ninio, 1974. Intuitively here, in the limting case in which k-1 = 0, the tRNA cannot leave the ribosome before GTP cleavage, so all the part of the reaction scheme that precedes GTP cleavage contributes nothing to specificity. If on the other hand k-1 is large, the ternary complex may leave the ribosome, and this will occur more often for the near cognate or the non-cognate substrates than for the cognate substrate. All this can be computed very easily.


After being introduced in their Fig. 1, the "non-specific binding" disappears from all the subsequent analyses of Zaher and Green. If k-1 exists, as represented in their Fig. 1, removing it from the kinetic analyses of the data in Fig. 4 is a deep mistake. If it does not exist, then the ternary complex may dissociate directly from the ribosome during the codon-anticodon recognition step, and Rodnina's kinetic scheme collapses. After so many years of brainwashing with "induced-fits", there is some hope to breathe again.


Even when k-1 is taken out of the picture, there are pending interpretation problems concerning the data in the various panels of Fig. 4. For instance, k-2 is not taken into account in Fig. 4A, k2 and k3 are not taken into account in Fig. 4B, and so on. There seems to be, as in the work of Rodnina, a reluctance to incorporate into the kinetic analyses, the multiplicity of binding and departure events. There alo seems to be a reluctance to investigate the idle GTPase reaction - a possible correlate of a kinetic proofreading step occurring prior to the classical GTP hydrolsis on the ternary complex.


Conceptually, there are also some confusions in the Zaher-Green article, as in other articles on translation accuracy published in Molecular Cell. Discrimination based on forward kinetic constants is confused with induced fits, discrimination based on off rates is confused with equilibrium discrimination. Actually, both on-rates and off rates discrimination are fully compatible with steady-state Michaelis kinetics (furthermore, steady state and equilibrium should not be confused). Kinetic proofreading may use off-rates (as in the Hopfield or Ninio mechanisms) or on-rates, as in the Yan, Magnasco and Marko (1999) mechanism. Energy levels along the reaction pathway (not to be confused with equilibrium constants) are important in all reaction schemes, including induced fits, and the energy levels are ratios of forward to backward kinetic constants. As a matter of fact it is quite difficult to segregate forward from backward effects in kinetic analyses, and the current emphasis on forward discrimination relies on Rodnina's biased interpretations of her otherwise valuable kinetic results. Last but not least, in both the induced-fits and the kinetic proofreading mechanisms there is - contrary to Michaelis kinetics - more than one exit gate for the substrate, which is also, to some extent, a side-entrance gate. Experimental work probing these mechanisms should explicitely investigate the properties of the side gates.


4-NOTES

[1] Some background on ribosomal mutants :

The discovery of the ribosomal accuracy mutants by Luigi Gorini and co-workers (e.g., Rosset and Gorini, 1969; Gorini, 1971; Biswas and Gorini, 1972) provided a strong impetus to the development of modern ideas on accuracy. There is a rich set of properties associated with these mutants, which is usually forgotten (for a history of the subject and detailed references, see Ninio, 2006).

Property 1 : Mathematical structure. There is a ranking of the ribosomal mutants in terms of their general accuracy, and a definite mathematical structure underlying the variations in tRNA suppression effciencies as a function of the ribosomal context.

Property 2 : The distinction between geometrically correct and geometrically incorrect codon-anticodon association is irrelevant. Actually, what puzzled most Gorini and co-workers, (and Apirion and co-workers too) was the fact that the reading of nonsense codons by tRNAs having complementary anticodons varied in the same direction as misreading ! Most of Gorini’s data was about the reduction of what we call cognate nonsense suppression in the high accuracy mutants. Not reminding it is a lie by omission.

Property 3 : The ribosomal accuracy mutants are not universal. They have very little effect on many misreading errors. In general, they have much less effect on missense than on nonsense suppression. In the case of standard missense errors, they seem to have some effect on the near cognate interactions, and almost no effect on the non-cognate interactions.

Property 2 rules out at once all induce-fit models based upon the geometry of the codon-anticodon association. I could have said « ruled out at once » since the induced-fit models were in vogue in the early 1970’s.

Property 1 led me to interpret the ribosomal accuracy mutants in terms of a kinetic model (Ninio, 1974). I called the effect « kinetic modulation » and proposed an analogous effect in the case of DNA replication, the « next-nucleotide effect » (Ninio, 1975). The model predicted that the probability of forming a peptide bond after a tRNA binding event on the high fidelity ribosomes decreased for ALL interactions, correct or not. I wrote in the summary of Ninio, 1974:

« It is postulated that the tRNA first makes a 'loose' bond with the codon. A second event is required to stabilize binding and to allow transpeptidation. The probability that the second event occurs is related to the time that the tRNA sticks to the codon in the loose binding state. Ribosomal mutations would make the transition from loose to tight binding more probable (ram) or less probable (strA) per collision. (---). In the wild-type cell, when a codon becomes associated with its cognate tRNA or release factor, the probability of ensuing chain elongation or termination is very close to unity. The probability of elongation decreases to about one-half in the strA1 strains. »

The one-half factor is just what is observed experimentally. The fact (property 3) that ribosomal accuracy mutants have little effect on the non-cognate (thus very weak) interactions cannot be explained with my initial model. It becomes far more natural in the context of multi-step discrimination mechanisms, in particular the kinetic proofreading ones (Hopfield, 1974 ; Ninio, 1975).

Kinetic experiments aimed at characterizing the properties of high accuracy ribosomal mutants culminated with the outstanding Bilgin et al. 1992 J. Mol. Biol. article, which showed (property 4) :

Property 4 : High accuracy correlates with high « idle GTPase » activity. It is « a ribosome-idling reaction occurring without EF-G, which is drastically amplified in SmD in relation to wild-type ribosomes ».



[2] On shortcut events.

1- The theoretical rationale for shortcut events.

Everyone agrees that there are several kinetic steps from initial ternary complex binding up to peptide bond formation on the ribosome. Most people agree with the notion that a tRNA may dissociate from the ribosome somewhere in the middle. Most people agree with the notion that some tRNAs are discarded very early in the production line, we may call them non-cognate tRNAs; and most people agree with the notion that some other tRNAs are carried along further, but often discarded somewhat later, at an intermediate stage in the production line, we may call them near-cognate tRNAs. (This could also be the situation for some “weak cognate” tRNAs).

What happens after the dissociation of the near-cognate tRNA? Most people consider that there is no problem. The ribosome returns to its initial state, and a new incorporation cycle begins. Now, suppose for a moment that it does not return to its initial state. Another tRNA or another ternary complex may take the place of the previous one. An enzymologist may call this an “exchange reaction”. If such an exchange reaction is possible, the incoming tRNA would skip the early steps of the incorporation pathway. Errors may arise, because the latecomer tRNA would have been tested on only a subsection of the elongation cycle. I called “shortcut events” the hypothetical events in which a tRNA skips the early steps of the incorporation pathway, presumably by binding to the ribosome right after the dissociation of another tRNA which had brought the ribosome to an intermediate stage in the incorporation pathway.

At this point, I think, most readers will agree with the point that the ribosome needs to go back to the initial state after a premature dissociation event. If the ribosome is not re-initiallized, and if furthermore shortcut events are impossible, then the ribosome is out of business. So, if there are re-initilization difficulties, shortcut events would be beneficial, they would allow the engine to work again, at the cost of making an occasional error.

How the ribosome returns to its initial state after a premature dissociation? Now, you might ask, why should it be difficult to re-initialize the ribosome? It all depends on ribosome energetics. If you believe for instance that the ribosome derives energy from peptide bond formation, and begins the next incorporation cycle loaded like a spring, then goes downhill, the ribosome would be in a low energy state after a premature tRNA dissociation, and will not go spontaneously back to the initial state. It would need an energy coupling activity (e.g., mediated by an ATPase or a GTPase) to return to the initial state. So this ATPase or GTPase activity would control the error-rates by controlling the ratio of shortcut events to re-initialization events after a premature dissociation.

You may on the other hand, believe that the ribosome is going uphill. Since the tRNA binds as a ternary complex, GTP cleavage on EF-Tu may bring the ribosome to a high energy state. After GTP cleavage, if there is a premature dissociation event, the ribosome would spontaneously return to the initial state. Here again a GTPase activity would play a crucial role in accuracy, and accuracy could be controlled by a factor which influences the rate of return from the high energy to the low energy state.

The link with kinetic proofreading/amplification schemes. In these schemes, proposed in the mid 1970's, a substrate binds to an enzyme, product is formed in two steps, the substrate may dissociate at the binding stage or after the first step. Energy coupling is used to prevent a substrate from skipping the first step (now I would say “taking a shortcut”). In Hopfield's scheme, the energy is used to bring the enzyme in a high energy state after the first step. In my scheme, energy is used to bring the enzyme from the intermediate state to the initial state. So, essentially, energy is used in both schemes to prevent the occurrence of shortcut events. The immediate implication is that if energy coupling is not perfect, shortcut events MUST be observed.

2- Biological implications of shortcut events. At present, the field of codon-anticodon recognition is a mess. There are dozens of anticodon post-transcriptional modifications, dozens of strange suppressors, plenty of queer influences by tRNA regions outside the anticodon. Once you begin to think in terms of a multi-stage selection process, with different constraints at each stage, and once you begin to characterise, through the study of shortcuts, which type of error is most likely to arise at which stage, there will be plenty of possibilities to put order in the data, plenty of possibilities to separate early from late events, understand which type of error is sensitive to which antibiotic, or to which ribosomal context, understand what would be the effects of under or over postranscriptional modifications of tRNAs in different cellular types, and their implications in some diseases, etc.



5- REFERENCES.  

Biswas, D.K. and Gorini, L. (1972) Restriction, de-restriction and mistranslation in missense suppression. Ribosomal discrimination of transfer RNA's. J. Mol. Biol. 64, 119-134. 

Bel, G., Munsky, B. and Nemenman, I. (2010) The simplicity of completion time distributions for common complex biochemical processes. Physical Biology 7, 016003.

Björkman, J., Samuelsson, P., Andersson, D.J. and Hughes, D. (1999) Novel ribosomal mutants affecting translational accuracy, antibiotic resistance and virulence of Salmonella typhimurium. Molecular Microbiology 31, 53-58.

Blanchard, S.C., Gonzalez, R.L., Kim, H.D., Chu, S . and Puglisi, J.D. (2004) tRNA selection and kinetic proofreading in translation. Nature Structural and Molecular Biology 11, 1008, 1014.

Gorini, L. (1971) Ribosomal discrimination of tRNAs. Nature New Biol., 234, 261-264.

Gromadski, K.B. and Rodnina, M. (2004) Kinetic determinants of high fidelity tRNA discrimination on the ribosome. Molecular Cell 13, 191-200.

Hopfield, J.J. (1974) Kinetic proofreading: a new mechanism for reducing errors in biosynthetic processes requiring high specificity. Proc. Nat. Acad. Sci. USA, 71, 4135-4139.

Hopfield, J.J. (1980) The energy relay: A proofreading scheme based on dynamic cooperativity and lacking all characteristic symptoms of kinetic proofreading in DNA replication and protein synthesis. Proc. Nat. Acad. Sci. USA, 77, 5248-5252 

Johansson, M., Lovmar, M. and Ehrenberg, M. (2008) Rate and accuracy of bacterial protein synthesis revisited. Current Opinion in in Microbiology 11, 141-147.

Kaul, M., Barbieri, C.M. and Pilch, D.S. (2006) Aminoglycoside-induced reduction in nucleotide mobility at the ribosomal A-site as a potentially key determinant of antibacterial activity. Journal of the American Chemical Society 128, 1261-1271.

Kolitz, S.E., Tacaks, J.E., and Lorsch, J.R. (2009) Kinetic and thermodynamic analysis of the role of start codon/anticodon base pairing during eukaryotic translation initiation. RNA 15, 138-152.

Koshland, D.E. (1958) Application of a theory of enzme speificity to protein synthesis. Proc. Nat. Acad. Sci. USA 44, 98-104.

Kramer, E.B., Vallabhaneni, H., Mayer, L.M. and Farabaugh, P.J. (2010) A comprehensive analysis of translation missense errors in the yeast Saccharomyces cerevisiae. RNA 16, 1797-1808.

Ledoux, S. and Uhlenbeck, O.C. (2008) Different aa-tRNAs are selected uniformly on the ribosome. Molecular Cell 31, 114-123. 

Lemeignan, B., Sonigo, P. and Marlière, P. (1993) Phenotypic suppression by incorporation of an alien amino acid. J. Mol. Biol. 231, 161-166.

McCarthy, B.J. and Holland, J.J. (1965) Denatured DNA as a direct template for in vitro protein synthesis. Proc. Nat. Acad. Sci. USA 54, 880-886.

Ninio, J. (1971) Codon-anticodon recognition: The missing triplet hypothesis. J. Mol. Biol. 56, 63-82.

Ninio, J. (1974) A semi-quantitative treatment of missense and nonsense suppression in the strA and ram ribosomal mutants of Escherichia coli . Evaluation of some molecular parameters of translation in vivo. J. Mol. Biol. 84, 297-313. 

Ninio, J. (1975) Kinetic amplification of enzyme discrimination. Biochimie 57, 587-595

Ninio, J. (1986) Kinetic and probabilistic thinking in accuracy. In Accuracy in Molecular Processes (Kirkwood. T.B.L., Rosenberger, R. & Galas, D.J., eds) Chapman & Hall, London, pp. 291-328.

Ninio, J. (1987) Alternative to the steady-state method : Derivation of reaction rates from first passage times and pathway probabilities. Proc. Nat. Acad. Sci. USA 84, 663-667.

Potapov, A., Triana-Alonso, F. and Nierhaus, K. (1995) Ribosomal decoding processes at codons in the A or P sites depend differently on 2’-OH groups. J. Biol. Chem. 270, 17680-17684. 

Rosset, R. and Gorini, L. (1969) A ribosomal ambiguity mutation. J. Mol. Biol. 39, 95-112.

Gregory, S.T., Carr, J.F., and Dahlberg, A.E. (2009) A signal relay between ribosomal protein S12 and elongation factor EF-Tu during decoding of mRNA. RNA 15, 208-214.

Shoji, S., Abdi, N.M., Bundschuh, R. and Frederick, K. (2009) Contribution of P residues to P-site tRNA binding. Nucleic Acids Research 37, 4033-4042.

Vallabhaneni, H., and Farabaugh, P.J. (2009) Accuracy modulating mutations of the ribosomal protein S4-S5 interface do not necessarily destabilize the rps4-rps5 protein-protein interaction. RNA  15, 1100-1109. 

Whitford, P.C., Geggier, P., Altman, R.B., Blanchard, S.C., Onuhic, J.N. and Sanbonmatsu K.Y. (2010) Accomodation of aminoacyl-tRNA into the ribosome involves reversible excursions along multiple pathways. RNA 16, 1196-1204.

Yan, J., Magnasco, M.O., and Marko, J.F. (1999) A kinetic proofreading mechanism for disentanglement of DNA by topoisomersaes. Nature 401, 932-935.