I’ve seen a few criticisms/caveats floating around in response to this paper already.
First, some people are highly skeptical that the wild-type controls in the study are sufficient, as they weren’t edited in the same way as the mutants so don’t control for any off-target effects etc. As a result, the fitness of all mutants may reduced relative to the WT regardless of their identity, potentially exaggerating fitness effects of synonymous mutations. Indeed, if you set a stricter threshold of fitness reduction to be counted as truly deleterious, more non-synonymous mutations pass it than synonymous ones. In other words, non-synonymous mutations tend to cause larger fitness declines than synonymous mutations.
Second, with regards to generalisability of these results (assuming they’re all perfectly accurate), it seems likely that this kind of noticeable fitness decline from synonymous mutations affecting the transcripts is going to be much more relevant to a unicellular haploid organism growing in rich media than other organisms, such as multicellular diploid organisms in a natural environment.
We know that having a large number of cells buffers organisms against perturbations in transcript abundance in some contexts, for example. Similarly, in a heterozygous setting, non-synonymous mutations are more likely to result in a substantial fitness decline than synonymous mutations acting only through transcript effects because again there would be a buffer in the form of the WT transcript. The rich media environment of the experiment because, to borrow from a commenter on twitter, a little rattle in a car is more likely to cause problems when driving flat out than it is in everyday use. If you push the yeast to operate at maximum, minor differences that usually go unnoticed might suddenly be limiting.
All this to say, even if their results are flawless, the experiment is still very likely to have limited generalisability and therefore relevance to evolutionary biology.
Finally, as this paper was partially brought up as a response what I said about the substitution rates of non-synonymous/synonymous mutations in another thread ("I'm treating the mutation rate as a substitution rate" - Dr. Nathaniel Jeanson - #68 by evograd), I will point out that this paper, even assuming it’s accurate and that it can be generalised to all contexts in all organisms, doesn’t actually change my overall point.
The explanation the authors of this new article propose for why we see more synonymous mutations than non-synonymous mutations is still natural selection, just operating on a larger scale (timeline) than the traditional model. To quote the paper:
We explored whether the low d N/d S could also be caused by a difference between synonymous and nonsynonymous mutants in their fitness variation among environments40,41. Considering this variation is relevant, because the fixation of a neutral mutation takes 4N e generations42 on average, a period during which the environment is highly likely to have changed many times. In addition to influencing the mRNA level and/or mRNA folding strength that can exert a fitness effect, nonsynonymous mutations also alter the protein sequence and potentially alter function, which synonymous mutations do not. Because each of the molecular phenotypic effects could be environment-dependent, nonsynonymous mutants may naturally have a larger across-environment fitness variance than synonymous mutants, especially given recent reports that amino acid substitutions often show environment-specific fitness effects43,44,45. Under the most extreme scenario, the fraction of deleterious mutations is identical between synonymous and nonsynonymous mutations in each environment, but the specific deleterious mutations vary across environments for nonsynonymous but not synonymous mutations. Consequently, when the environment of a population fluctuates within the typical fixation time, some synonymous mutations are never deleterious so may be fixed, whereas nearly every nonsynonymous mutation is deleterious under some environments so cannot be fixed, resulting in d N/d S much lower than 1.
Selection would still be the cause of dN/dS<1, as was proposed before.