Why did translation develop a specific codon for initiation?

Question

The translation of mRNA is initiated by a specific methionine-accepting tRNA at a specific initiation codon, usually AUG (complementary to the tRNA anticodon). However translation at suitable (albeit unphysiological) conditions in vitro does not require a specific initiation codon or a specific amino acid, so it is not difficult to envisage an earlier stage in the evolution of the translation apparatus in which this specific initiation system was absent. Instead there might have been random initiation or initiation from the 5′ end of the mRNA.

A specific initiation codon puts constraints on the sequence of the mRNA and adds an unnecessary amino acid to the N-terminus — one that is sometimes unwanted and needs to be removed.

What advantages can one envisage for initiation at a specific internal site, rather than at, say, the 5′ end of the mRNA?

I appreciate it is impossible to prove which factor was actually responsible during evolution, but by soliciting concrete explanations I trust that this question is as valid as the original from which it is derived (see Footnote).

FOOTNOTE

This question has been expanded from one part of a three-part question posted some time ago on “Why is AUG the Initiation Codon?”. This recently resurfaced, and as nobody had answered this particular part, and as it is site policy to ask one question at a time, I decided to remove it from the original and repost it here so it could be considered on its own merits, without the distraction of other questions such as the choice of codon or of methionine over other amino acids. I shall contribute an answer with some thoughts of my own, but, of course, anyone can join in.

Answer 1

It is not clear to me where initiation took place in the translation of artificial polynucleotides such as (UC)_n which Khorana used in experiments that, following up the work of Nirenberg and Matthaii, helped decipher the genetic code. It may well have occurred at random positions rather than the 5′ end. Such a mode of translation would obviously have been wasteful if it occurred during early evolution as a mixture of products would be formed, many of which would be inactive and might even compete with the active species. If this were the original situation, it is easy to see how a change to initiation at a specific point would be advantageous.

But why not start at the 5′ end? Contemporary eukaryotic initiation complexes recognize the 5′ end of the mRNA, even though they then scan further to find a specific initiation codon that will respond to the initiator tRNA. As there is no physical separation of genes on a chromosome, it is obvious why transcription needs to start at a specific point within the DNA, but the need not be true for the translation of individual mRNA transcripts.

I can see two ways of approaching this question.

The first is from the standpoint of the contemporary ‘DNA World’. The arguments one can make include:

• In a general sense 5′-end initiation may be regarded as restrictive in that it would not allow the 5’ end of mRNA to provide other functions such as forming a secondary structure to regulate the ease of translation.
• 5′-end initiation would preclude the polycistronic mRNAs we see in contemporary bacteria.
• In 5′-end initiation it might have been difficult to distinguish the ends of mRNAs from other RNAs before the modification of bases that we see in contemporary eukaryotic mRNAs had developed.
• Likewise, as suggested by @John in a comment on the question, 5′-end initiation might have caused translation of fragments of RNA as they degraded.
• The selection of a specific codon was secondary to the need for a single initiator tRNA, which has been discussed in relation to another SE question, although the arguments there relate to more recent stages in evolution.

However there is another approach which considers the situation in a putative ‘RNA World’.

The difference here is that — like some contemporary RNA viruses and bacteriophages — the genome might also have been the mRNA but without a system to produce many copies of the ‘plus’ strand for translation. Translation may have been on the genomic RNA.

• To translate several proteins it would have been advantageous to have several initiation sites on what would have been a polycistronic messenger (rather synthesize a polyprotein like poliovirus).
• The 5′-end of the RNA may not, in any case, have been available for encoding protein as it might have another (e.g. replicative) function, and it is also possible that part of what in contemporary systems is a separate part of the translation apparatus was physically incorporated into the genome.