David S. Lawrie; Philipp W. Messer; Ruth Hershberg; Dmitri A. Petrov

“ Linking polymorphism with divergence data, we further find that the fraction of synonymous sites exposed to strong purifying selection is higher for those positions that show slower evolution on the Drosophila phylogeny. The function underlying the inferred strong constraint appears to be separate from splicing enhancers, nucleosome positioning, and the translational optimization generating canonical codon bias. The fraction of synonymous sites under strong constraint within a gene correlates well with gene expression, particularly in the mid-late embryo, pupae, and adult developmental stages. ”

“ We tested a number of controls to rule out the possibility that our observation of strong purifying selection results from other forces with possibly similar signals: A lower mutation rate, for example, can cause a signal indistinguishable from strong selection in polymorphism if the sample depth of the population is too shallow. To account for this, and at the same time account for any variation in the amount of linked selection between 4D sites and short intron sites, we used a bootstrap to control for GC content and distance between the 4D and short intron sites. ”

“ More evidence for weak selection acting at synonymous sites comes from the study of polymorphism within species. Purifying selection reduces the frequency of deleterious alleles in the population. To measure its effect, the site frequency spectrum (SFS) tabulates the fraction of observed SNPs in all frequency classes across the sites of interest. The overabundance of low frequency SNPs relative to the neutral expectation is the signal of purifying selection operating on the test sites. ”

“ If synonymous sites harbor highly deleterious variants under strong purifying selection, then that must change our view of the functional importance of synonymous sites and their potential role in genetic disease, as a possible source for adaptation, and as the neutral foil in tests for selection. ”

“ Even so, strong selection, rather than a mutational difference, would still underlie our signal, as genetic draft cannot alter the frequency of synonymous mutations that are simply absent from the population. On the other hand, sweeps of weakly advantageous alleles in coding sequences could eliminate polymorphism in 4D sites more so than in short introns. Narrow selective sweeps in coding sequences reducing variation at otherwise neutral 4D sites is, however, an unlikely explanation for our observations. ”

“ Directed mutagenesis studies targeting synonymous sites as well as studies of natural polymorphism have found consequential changes in protein levels and functionality due to natural synonymous variation and induced mutations [87] – [85] , [88] – [89] . In an experiment done on the Alcohol dehydrogenase (Adh) gene in D. melanogaster, changing 10 wild-type preferred Leucine alleles to unpreferred alleles in the 5′ region of the gene lowers the enzymatic activity of collected Adh by 25% [90] . ”

“ Mutations at strongly constrained synonymous sites should never rise above low frequency in the population and certainly will never fix, barring tight linkage to a very advantageous allele or a shift in the functional properties of the site. ”

“ Nucleosome-bound regions are associated with both the presence of purifying selection and lower rates of mutation [51] – [52] . Canvassing all 4D sites in the 146 bp regions around known bulk nucleosomal binding sites [53] , we find a small increase in the fraction of missing polymorphism in these 4D sites bound by nucleosomes (Table 3) . However, we have reason to believe that this slight increase above 22% is due to weak selection acting on nucleosome-bound sites and is likely not related to the strong constraint we measure in this paper (see Text S3) . ”

“ Strong constraint can be difficult to capture as strong selection has a greater effect on the amplitude of the SFS, the total number of observed mutations, than on its shape, the frequency distribution of observed mutations. ”

“ Human gene Multidrug Resistance 1 (MDR1) contributes to the drug resistance of cancer cells [92] . Both naturally occurring alleles as well as induced novel mutations at synonymous sites in MDR1 affect the resulting protein's conformation, altering its substrate specificity in human cell lines [92] . In the E. coli gene ompA, exchanging eight frequently-used codons for synonymous infrequently-used codons near the gene start results in a 3-fold reduction in mRNA levels and a 10-fold reduction in synthesis of protein OmpA [81] . ”

“ Short intron sites are defined as those sites falling in introns of less than length 86 bp, 16 bp away from the intron start and 6 bp away from the intron end in order to eliminate any functional sequences at the edges of the introns [30] . Eliminating 16 bp from each side did not change SNP density (not shown) . Any remaining purifying selection, especially strong purifying selection, in short introns makes our results more conservative. Four-fold (4D) sites are the collection of 3rd codon positions for the following amino acids: Proline, Alanine, Threonine, Glycine, and Valine. ”

“ From these alignments we removed the sequences belonging to D. melanogaster and D. willistoni. The D. melanogaster sequences were removed because the polymorphism data was extracted from this species and we wished to avoid a false concordance between the results from polymorphism and divergence. The D. willistoni sequences were removed, because the branch length leading to D. willistoni is long and the codon bias of D. willistoni is significantly different than from the rest of the twelve Drosophila species [46] . ”

“ We find that in genes enriched for strong constraint, we co-enrich for many important functional gene sets. In particular, we co-enrich for genes critical in pupae-to-adult morphogenesis and in late embryogenesis. This finding is consistent with the result that genes expressed highly in late embryos, pupae, and adults have elevated levels of strong constraint at 4D sites. ”

“ This might indicate a role for splicing enhancers in the strong constraint, but Table 3 also shows that multi-exon genes and single-exon genes have similar amounts of strong constraint. The inference on the single-exon genes is particularly noisy, especially so given that our bootstrap method controls for distance to short introns. However, only about one-fifth of our 4D sites fall near splice sites and the modest enrichment of constraint near splice sites is not enough to explain the ubiquitous constraint at 4D sites across the genome or especially in single-exon genes. ”

“ The enrichment score of the overall cluster is then the geometric mean of these p-values. Thus the higher the enrichment score, the lower the p-values are for all terms in the overall annotation cluster and the more significantly enriched the overall cluster is in the test gene set. ”

“ Rosenbloom, KR & Siepel, A, “Detection of nonneutral substitution rates on mammalian phylogenies”, Genome Res 20 (1) : 110–121, [ [pmid:19858363 PollardKSHubiszMJRosenbloomKRSiepelA] ] ↑ Zhou, T; Gu, W & Wilke, CO, “Detecting positive and purifying selection at synonymous sites in yeast and worm”, Mol Biol Evol 27 (8) : 1912–1922, [ [pmid:20231333 ZhouTGuWWilkeCO] ] 1 2 Waldman, YY ”

“ Phylogenetic analysis by maximum likelihood”, Mol Biol Evol 24 (8) : 1586–1591, [ [pmid:17483113 YangZ] ] ↑ Ikemura, T, “Correlation between the abundance of escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes: A proposal for a synonymous codon choice that is optimal for the E. coli translational system”, J Mol Biol 151 (3) : 389–409, [ [pmid:6175758 IkemuraT] ] 1 2 Grantham, R ”

“ Text S1 Demographic correction of SFS. Maximum likelihood results when correcting for demography and other non-mutation-selection balance forces in the SFS. (DOC) pgen.1003527.s007.doc
Text S2 Mutation versus strong constraint. Discusses extra controls for distinguishing the strong constraint from possible confounding mutational effects: S2.A – Tri-nucleotide bootstrap results, S2.B – Power to detect low mutation versus strong purifying selection on 4D sites, S2.C – Slow- versus fast-evolving 4D site bootstrap results. ”

“ Quantification of selection on nonsynonymous and synonymous sites in humans”, Genetics 188 (4) : 931–940, [ [pmid:21596896 KeightleyPDHalliganDL] ] 1 2 Mackay, TFC; Richards, S; Stone, EA; Barbadilla, A & Ayroles, JF, “The drosophila melanogaster genetic reference panel”, Nature 482 (7384) : 173–178, [ [pmid:22318601 MackayTFCRichardsSStoneEABarbadillaAAyrolesJFet al] ] ↑ Clemente, F & Vogl, C, “Unconstrained evolution in short introns?–An analysis of genome-wide polymorphism and divergence data from drosophila”, J Evol Biol 25: 1975–1990, [ [pmid:22901008 ClementeFVoglC] ] 1 2 3 Haddrill, PR ”

“ Codon usage, tRNAs, and ribosomes”, Trends in Genetics 28 (11) : 574–581, [ [pmid:22921354 NovoaEMRibas de PouplanaL] ] ↑ Tuller, T; Veksler-Lublinsky, I; Gazit, N; Kupiec, M & Ruppin, E, “Composite effects of gene determinants on the translation speed and density of ribosomes”, Genome Biol 12 (11) : R110, [ [pmid:22050731 TullerTVeksler-LublinskyIGazitNKupiecMRuppinEet al] ] ↑ Pechmann, S & Frydman, J, “Evolutionary conservation of codon optimality reveals hidden signatures of cotranslational folding”, Nature Structural & Molecular Biology 20: 237–243 PechmannSFrydmanJ 1 2 Qian, W ”