A recent paper in Nature from the lab of Pavel Baranov and collaborators suggests a new type of mechanism of translation regulation. However, After discussing this paper in a journal club today – I’m not convinced about their model.
They looked at the ribosome profiling of the AMD1 mRNA (encoding S-adenosyl-methionine dcarboxylase 1) from public data and noticed something strange. This mRNA encodes a known upstream open reading frame (uORF) which is known to regulate translation initiation of the AdoMetDC coding sequence. Indeed, they found ribosomes accululating at the uORF region. What they also noticed that there is ribosome accumulation at the 3′ end of the 3′ untranslated region – which should not have any ribosomes.
Their model claims that ribosomes readthrough the native stop codon and continue to the next stop codon, where they stall. To that region of the 3’UTR they call “tail”. Ribosomes pile up as if in a trafic jam, until the next ribosome can’t pass through the native stop codon. This halts the translation altogether.
However, already in fig 1 there’s a problem: we only see on peak at the end of the “tail”, but not ribosomes piling up along the “tail.
They do a number of very nice experiment, the key finding is that if the replace the stop codon at the end of the coding sequence to a sense codon (so there is 100% readthrough, compared to ~1.6% in the native mRNA) then there is very little protein product compare to the native situation (65-fold reduction, which exactly the ration of the readthrough efficiency).
Why would that happen? As I mentioned, they claim that ribosomes pile up from the tail’s stop codon backwards. But they don’t show it. The KEY POINT in their model, and they don’t provide any evidence that ribosomes actually pile up (i.e. performing ribo-seq on their constructs to show accumulation of ribosomes). They claim that this is a molecular counting mechanism. Then show me the numbers – show me that in the native situation, 1:65 proteins have tails of differnt lengths, representing ribosomes stuck along the tail (e.g. by inserting FLAG after the CDS stop codon and enriching the the tail peptide by IP with anti-FLAG). They claim there are no proteins with tails but only show western blot at one exposure. Also, according to their model, at a rate of 1.6% readthrough, translation will stop only after 65 (readthrough events) * 13 (ribosomes piling along the the tail, at 30nt/ribosome) = 845 proteins /mRNA. is that truely the case? How many mRNA and protein moleculaes are there? Can a single mRNA really translate that many? At a translation rate of 5aa/sec, AdoMetDC1 (334aa) should be translated in ~1min. Which means that for ~850 proteins, with an average of ~1ribosome/200nt, should take roughly 3 hours of non-stop translation. What is the situation in the cell? How long will it take an mRNA to translate 850 proteins in the cells in reality?
What happens when you replace the tail’s stop codon with a sense codon? Will you get more os less protein if you change the distance between the two stop codons?
I can think of an alternative explanation to their model: the first ribosome that gets to the tail’s end is stalled, and then conveys information to the 5′ region to stop translating (a similar situation like NMD or no-go decay). In this case, translation will cease already after 65 proteins on average.
Another unexplored question is why would the ribosome stall at the tail’s stop codon? Is the seqence there full of rare codons? is there a unique structure that captures ribosomes?They show that it is probably the 21 codons upstrem of the stop codon, but don’t go beyond that claimor using that knowledge later. Also, why doesn’t it activate no-go decay? could you imitate the same outcome with a RNA stracture that halts the ribosome?
Finaly – the authors did not examine the function of this mechanism in vivo – what happens to AdoMet levels in case you eliminate this mechanism (e.g. by deleting the tail)? Does it have any physiological effect?
Their abstract promised a molecular counting mechanism that limits translation. But there are no numbers here and not really supporting data for this mechanism.
Yordanova et al (2018) AMD1 mRNA employs ribosome stalling as a mechanism for molecular memory formation. Nature (epub Jan 3 2018) doi: 10.1038/nature25174 .