Conformational diseases and the protein folding problem: role of amino-acid propensity to be embedded in context of specific usage of synonymous codons. Genetic code degeneracy and folding

Abstract

Translation was long thought to be a smooth process, producing invariably native proteins. However, it appeared recently that this may not always be the case. First, a family of apparently unconnected diseases, including important neurodegenerative conditions, like Alzheimer, Parkinson, Huntington, and prion disorders, was shown to be linked to misfolded proteins ('conformational diseases'). Secondly, up to 70% of nascent proteins are recognized as misfolded and tagged for proteolysis whilst still attached to the ribosome. Misfolding seems therefore to be a quite common event, which is however most often corrected by efficient proteolysis of the misconformers produced. Misfolding therefore usually remains unnoticed, and safe when proteolysis is unable to remove the misconformers, which is the case in conformational diseases. Since the latter usually onset in the second half of human life, the mechanism of misfolding is likely to be linked to defects in the cellular translation and proteolysis machinery. This speculation is further substantiated by the observation that the amount of nascent proteins, recognized as misconformers whilst still attached to the ribosome, depends on the cellular state. On the other hand, the large amount of work devoted to prion proteins has highlighted the fact that certain proteins are prone to become easily misfolded. This implies that particular motifs in the primary structure of proteins enter metastable or 'soft' conformations upon synthesis, which can easily flip from one state to another. The present study investigates a possible link between particular primary structures and soft conformations. The study is based on experimental evidence, showing that silent codon exchange can affect protein conformation. Synonymous codons correspond to cognate tRNAs that may be present in the cell in rather different concentrations. Since tRNA availability determines the cognate codon translation rate, it follows that synonymous codons may have rather different translation rates. Taken together, part at least of the factors determining the native protein conformation may be encoded in the protein translation rate, i.e. in the selection of synonymous codons. Finally, the codon translation rate can be approximately estimated from the relative codon usage. Indeed, since the cellular concentration of charged tRNA on one hand controls the translation rate of the cognate codon, and on the other correlates closely with the usage of that codon in the cell, a reasonable approximation is to take as relative rate of codon translation the inverse of the usage of this codon. These various assumptions allow us to compute the local translation rate at any codon of a given messenger RNA. We were struck by the observation that 'fast' (frequent) or 'slow' (rare) codons tend to cluster in most mRNAs, and that the clusters tend to harbour preferentially certain amino acids. We therefore set out to analyse all the genes of E. coli, identify the 'fastest' and 'slowest' codon window in each and check if certain amino-acids are favored, or disfavored, within these windows, with respect to the general amino-acid usage in the genome of the species under investigation. We find that indeed, most aminoacids are constrained within these windows, and more so as the size of the window is reduced from 21 to 5 codons. Significant correlations are found between the type (excess or depletion) of constraint and the physical characteristics of the amino acids (hydrophobic, aromatic, charged or polar) or their propensity to enter particular secondary structure. It is therefore likely that synonymous codon selection in the mRNA locally sets kinetic factors contributing to the native conformation of the protein. Modifications of these kinetic factors, whether by silent codon exchange or by changes of the translation rate due to modifications of the cellular concentrations of components of the translation machinery (charged tRNA, elongation factor, foldases, etc.) would then possibly result in misconformation of the protein. We surmise that conformational diseases may result from spontaneous protein misfolding resulting from the incompatibility of the translation machinery components with the folding-specific translation rate. The misfolded proteins would either adopt a conformation that cannot be proteolyzed, or aggregate so as to be unable to enter the proteasome.