During the process of translation, molecular machines in the cell called ribosomes use sequences encoded by messenger RNAs as templates for protein synthesis. On page 356, Yordanova et al.1 propose an intriguing mechanism that might limit the number of protein molecules that can be synthesized from a single mRNA. It involves the formation of a queue of ribosomes on the mRNA, downstream of the main protein-coding region.
The conventional view of translation in eukaryotes — organisms such as fungi, plants and animals — is that each mRNA consists of a stretch of nucleotides that contains an open reading frame (ORF), which encodes a single protein containing more than 100 amino-acid residues. But over the past decade, the advent of technologies such as ribosome profiling2 has revealed that a more-diverse range of ORF sequences can, in fact, be translated. For example, numerous small upstream ORFs (uORFs) have been identified whose translation might regulate expression of the main ORF.
Ribosome profiling has also revealed a wealth of events in which translation is initiated at alternative start codons3 (triplets of nucleotides other than the triplets at which translation is normally assumed to initiate), and read-through events4 in which translation continues beyond the stop codon (the nucleotide triplet at the end of the ORF). Not only do these two types of event increase the overall diversity of proteoforms (molecular forms of proteins produced from genes)5, but they have also emerged as regulatory mechanisms for hundreds of genes in eukaryotic genomes. Other regulatory mechanisms for translation are also known, including ribosome stalling, in which obstacles impede ribosome movement along mRNAs.
Yordanova et al. now propose another evolutionarily conserved mechanism for translational control. They suggest that sporadic stop-codon read-throughs can lead to the formation of ribosome queues at downstream stalling sites, such that the queue length is proportional to the number of protein molecules that have been synthesized. The authors define the region between the end of the main ORF and the next in-frame stop codon (that is, the next nucleotide triplet that would be recognized as a stop codon by a ribosome translating beyond the main ORF’s stop codon) as the tail ORF. They suggest that translation is halted when queuing ribosomes in the tail ORF extend into the main ORF (Fig. 1).
The authors were inspired to propose this mechanism after inspecting publicly available ribosome-translation profiles for a protein called adenosylmethionine decarboxylase 1 (encoded by the AMD1 gene), the translation of which is tightly controlled. The profiles revealed translation of a uORF in the AMD1 mRNA, as previously reported6, but also a high density of ribosomes in a region known as the 3ʹ trailer of the mRNA, downstream of the AMD1 stop codon. This suggested that a stop-codon read-through had occurred, allowing ribosomes to accumulate in the tail ORF of AMD1.
Yordanova and colleagues performed experiments showing that stable peptidyl–transfer RNA complexes (which form between tRNA and the nascent protein chain during translation) are generated when tail-ORF sequences are translated, and that complex formation occurs before translation reaches the stop codon at the end of the tail ORF. This confirmed that the proposed read-through could occur, and that translation could stall in the tail ORF. The authors also constructed a mutant mRNA in which the AMD1 stop codon was replaced by a sense codon (a nucleotide triplet that encodes an amino acid), in the expectation that translation would occur uninterrupted through the mutated sequence. However, almost no AMD1 translation occurred with this mutant — the expected extended proteoform was produced in nearly undetectable levels.
To explore the mechanisms that affect the levels of expressed protein, Yordanova et al. used a strategy7 known as StopGo, which allows the cleavage and release of nascent protein chains at chosen positions during translation, but then allows ribosomes to resume translation of the downstream sequence. The authors used StopGo to cleave nascent proteins before translation of the AMD1 stop codon in the wild-type mRNA, and before translation of the sense codon in the mutant mRNA. They observed that the amount of AMD1 protein subsequently produced from the mutant mRNA was lower than the amount produced from the wild-type mRNA — even though the amino-acid sequences of the proteins were identical.
This result suggests that the tail ORF must lower protein expression by influencing translation, rather than by reducing the stability of the produced protein. The finding is at odds with previous work8 showing that proteins are generally destabilized when their sequences are extended by a stop-codon read-through. Yordanova and colleagues’ experimental data collectively show that the effects of translation of the AMD1 tail ORF are independent of the main coding sequence, mRNA stability, common protein-degradation and cleavage pathways, or whether the expressed protein is secreted by cells.
The findings are therefore consistent with the idea that translation is halted when a ribosome queue in the tail ORF extends into the main AMD1 coding region. Note that such regulation could work only if AMD1 translation is dysregulated and high, and would not apply under standard conditions. The authors also observed that translational output was more strongly reduced in experiments that increased the efficiency of read-throughs, thereby accelerating queue formation — in agreement with the model.
A note of caution is warranted, because Yordanova and co-workers have not directly observed long ribosome queues. The proposed ribosome stalling might also occur only transiently, thereby increasing the time required to attain full ribosome coverage of the tail ORF and so decreasing the overall impact of ribosome-queue formation on translation. Furthermore, besides the experimentally validated traffic-jam model9 (in which ribosomes collide and form queues, blocking translation initiation), other models for how ribosome stalling interferes with translation have been proposed. For example, it has been suggested that stalled ribosomes fall off mRNA following collision with a trailing ribosome; this model conflicts with the idea that long ribosome queues could form10. Peculiarities of the expression systems used by Yordanova et al. might also underlie some of the authors’ observations. Finally, their data do not rule out the possible involvement of other factors that could cause the downregulation of protein expression, such as protein-degradation pathways that occur at the same time as translation.
Nevertheless, the authors’ findings will surely inspire future endeavours to obtain concrete proof of the proposed mechanism, and to assess how widely it is used to limit protein synthesis. Single-molecule imaging of translation on individual mRNA molecules, in real time and in live cells, might eventually allow simultaneous observation of mRNAs and their protein products11.
Nature 553, 289-290 (2017)