We are occasionally warned not to compare apples with pears (sorry, English speakers, not with oranges this time). Sure, both are pome fruits, and both plants belong to the Rosaceae family, but they differ notably in shape, feel, smell and taste. Geneticists may well need a similar warning when they sometimes use the terms silent mutation and synonymous substitution (or synonymous mutation) interchangeably.
I could easily provide watertight definitions for both terms "silent mutation" and "synonymous substitution." Perhaps illustrated with a nice Venn diagram that makes it clear from which perspective which of the two terms make sense, where their meanings overlap and where they do not. I prefer to go with two examples, though.
A synonymous substitution that does not remain silent but needs an introduction. The work of Betül Kaçar from the Dept. of Bacteriology at the University of Wisconsin–Madison (UW–Madison) and her collaborators has long focused on the intriguing topic of merging ancestral sequence reconstruction ("paleogenomics") with experimental evolution. One of their pet molecules is elongation factor EF-Tu, a core component of the translational apparatus in all bacteria, archaea and eukaryotes (eEF‑1 or EF‑1A in the latter two). EF-Tu shuttles amino acid‑charged tRNAs (aa‑tRNAs) to ribosomes and facilitates their binding to the ribosomal A‑site (see frontispiece), and, given its crucial function, an ancestral EF‑Tu (AnEF) was certainly already part of LUCA's inventory.
To date, there is no known backup mechanism that would allow cells to dispense with the function of EF-Tu. And, despite their highly conserved sequences, it is usually impossible to functionally exchange evolved EF‑Tu alleles between distantly related species as, for example, mesophilic E. coli and thermophilic Thermus thermophilus (Figure 1A). However, De Tarafder et al. (2021) succeeded in getting two sequence-reconstructed 1.3- to 3.3-Gy-old ancestral EF-Tus to "work" with reasonable kinetic parameters in both contemporary E. coli and Thermus.
To round off the introduction and to come closer to the example, Betul Kaçar and Eric Gaucher had found that a sequence-reconstructed 0.5-Gy-old ancestral EF-Tu (AnEF) allows E. coli cells to grow with a doubling time of ~53 min. This corresponds to a two-fold increase of the doubling time compared to the tufA+ tufB+ parent strain (Figure 1B). E. coli has, like most bacteria, two genes encoding EF-Tu, tufA and tufB, with the two proteins being virtually identical. Therefore, one of the alleles had to be deleted to allow a fair comparison between the fitness of the ancestral AnEF and the contemporary EF-Tu (both share ~95% protein sequence identity). Strain constructs tufA+ ΔtufB and ΔtufA tufB+ with a single tuf gene both had doubling times of ~34 min, indicating that the ~53 min doubling time of the ΔtufA ΔtufB::anEF strain is not so bad after all. But could it be improved (=shortened) by experimental evolution in the lab?
This question was addressed in the new study by McGrath et al. (2023), who let the ΔtufA ΔtufB::anEF strain adapt through continuous cultivation over 3,000 generations (with sampling every 500 generations for whole‑genome sequencing, and, in parallel, evolving the ΔtufA tufB+ strain as control).
At generation 2,000, a mutation appeared in the anEF gene coding region that reached fixation by generation 2,500. They found by sequencing that a C→T mutation (transition) at position 45 of anEF changes a valine codon from GUC to GUU (V15V) (Figure 2A). Both the GUC and GUU codons are recognized by the same tRNA species in E. coli, tRNAVal (valV, valW), whose 5'‑G in the GUC anticodon can basepair with both the 3'-C and 3'-U in the codon (a textbook case of a wobble base pair). This is thus a typical "synomymous substitution" as it does not change the encoded amino acid, and AnEFC45T would be expected to be phenotypically "silent." But why is it not "silent," apparently, but beneficial rather as it had swept through the entire population?
Because the ΔtufA ΔtufB::anEF strain had accumulated other mutations in addition to the C45T mutation in anEF during continuous cultivation (see below), the authors had to dissect the effects of the C45T mutation from those of the "evolved genetic background" on fitness. To study this, they engineered four strains, two non‑evolved ('ancestor') and two 'evolved', each pair expressing either the original AnEFC45 or the 'evolved' AnEFC45T (Figure 2B). And this time they did not measure differences in doubling times (see Figure 1B) but differences in fitness via coculture competition assays (using the 'ancestral' and 'evolved' ΔtufA tufB+ strains, respectively, as controls).
Replacement of native E. coli EF-Tu (tufB+) with the 0.5-Gy-old EF-Tu (anEFC45) (orange) causes a 13% decrease in fitness (Figure 2B, orange), which upon 3,000 generations of lab evolution is recovered by AnEFC45T (red) and increased by 60% relative to the ancestor (note that in Figure 2B the fitness of "Ancestor AnEFC45" is set to 1 for an easier visual comparison). Substitution of the synonymous mutation with the native codon in AnEFT45C significantly decreases the fitness of the evolved strain by 30% (grey) whereas whereas introducing the synonymous mutation AnEFC45T in the ancestral background has no relative fitness benefit (cyan). This demonstrates that the fitness effect of the synonymous mutation C45T in anEF depends on the evolved genetic background.
However, as usual, it seems to be a bit more complicated than just the "evolved genetic background" being responsible for the improved fitness of strains carrying the AnEFC45T allele. McGrath et al. (2023) found a moderate but statistically significant increase in mRNA levels (Figure 3A) and in protein levels (Figure 3B) for AnEFC45T compared to AnEFC45 in both the "Ancestor" and "Evolved" strains. They suspect that, similar to what Brandis et al. (2016) found for tufB in Salmonella, a synonymous mutation near the start codon, here anEFC45T, promotes an open conformation of the mRNA that leads to increased AnEF expression. My tentative projection of the anEFC45T mRNA sequence onto the secondary structure proposed by Brandis et al. (2016) supports this.
During its evolution in the lab for 3,000 generations, the 'Ancestor' AnEFC45-strain (ΔtufA ΔtufB ::anEF) strain had accumulated seven non-synonymous mutations − the above mentioned "evolved genetic background" − in addition to the synonymous mutation anEFC45T. None of these mutations did immediately point to a connection to the AnEFC45T mutation, but some had also occurred in the long-term evolution experiment (LTEE) of E. coli B strain REL606, the parent of the "Ancestor" AnEFC45-strain. It is thus reasonable to assume that these mutations reflect the adaptation of the population to the specific growth regime of the experiment, which selects for better growth (=shorter doubling times) over time.
From the Muller plot in Figure 4 that shows the genotype dynamics over time one can almost "read" that by generation 2,000 enough growth-enhancing mutations had accumulated for the improved expression of AnEFC45T to become effective. To put it this way: the synonymous mutation C45T in the anEF gene would remain a silent mutation if it could not whisper in chorus along with other mutations in the host genome.
In the forthcoming part 2, I will present an example of Twin synonymous substitutions that remain silent.