Microevolution of nematode miRNAs reveals diverse modes of selection

Abstract

Micro-RNA (miRNA) genes encode abundant small regulatory RNAs that play key roles during development and in homeostasis by fine tuning and buffering gene expression. This layer of regulatory control over transcriptional networks is preserved by selection across deep evolutionary time, yet selection pressures on individual miRNA genes in contemporary populations remain poorly characterized in any organism. Here, we quantify nucleotide variability for 129 miRNAs in the genome of the nematode Caenorhabditis remanei to understand the microevolution of this important class of regulatory genes. Our analysis of three population samples and C. remanei's sister species revealed ongoing natural selection that constrains evolution of all sequence domains within miRNA hairpins. We also show that new miRNAs evolve faster than older miRNAs but that selection nevertheless favors their persistence. Despite the ongoing importance of purging of new mutations, we discover a trove of >400 natural miRNA sequence variants that include single nucleotide polymorphisms in seed motifs, indels that ablate miRNA functional domains, and origination of new miRNAs by duplication. Moreover, we demonstrate substantial nucleotide divergence of pre-miRNA hairpin alleles between populations and sister species. These findings from the first global survey of miRNA microevolution in Caenorhabditis support the idea that changes in gene expression, mediated through divergence in miRNA regulation, can contribute to phenotypic novelty and adaptation to specific environments in the present day as well as the distant past.

Keywords: Caenorhabditis; gene expression; miRNA; nucleotide variation; regulatory networks.

Fig. 1.—

The level of selective constraints varies among different regions of the miRNA hairpin. (A) Diagram of a stereotypical miRNA hairpin with a stem-loop structure. The number of mutations segregating in the population from Ohio and the number of substitutions between Caenorhabditis remanei and Caenorhabditis latens are respectively shown in parentheses for each region of the miRNA (i.e., polymorphisms: substitutions). (B) Nucleotide differences in miRNA regions are lower than nucleotide differences at synonymous sites of protein-coding genes. Light blue, within species variation; dark blue, between species divergence; miR, mature miRNA; miR*, star sequence. Means are represented ± 1 standard error of the mean.

Fig. 2.—

miRNA loci experience strong purifying selection to maintain regulatory interactions and to preserve the integrity of the hairpin structure. Average nucleotide differences within population (A) and between species (B), calculated using a sliding window, are lower in the region corresponding to the miRNA hairpin, represented by a black box. Black lines indicate mean nucleotide diversity and divergence and the blue areas indicate the 95% confidence interval. Nucleotide differences in the mature miR and in the backbone (hairpin − miR) are lower than nucleotide differences at synonymous sites of protein-coding genes (C, D). Similarly, both paired and unpaired sites of the miRNA hairpin show signatures of purifying selection when compared with synonymous changes (C, D). Median polymorphism and divergence are shown by a horizontal line. The box represents the IQR between the first and third quartile. The whiskers extend to the furthest data point within 1.5 times the IQR from the box. Means with different letters are significantly different with Wilcoxon two-sample tests. IQR, interquartile range.

Fig. 3.—

(A) SNP frequencies in mature miRs among populations of C. remanei. Most variants are found in a single population and at low frequency, with noticeable exceptions for instance for SNPs in mir-64a, mir-248, mir-787, block2892, and block3297. Columns represent separate populations and rows represent distinct SNPs. Each circle represents the frequencies of the ancestral or major allele (in purple) and the derived or minor allele (in blue). The different alleles and their position relative to the start of the mature miR are indicated in the right panels. Ancestral alleles identified by comparison with C. latens are marked with a thick line. SNPs located in a same miRNA are joined by a horizontal bar. (B) A 14-bp long deletion present in 22% of the population from Ontario removes the seed motif of the mature miR in miRNA block2890 and also alters the hairpin structure.

Fig. 4.—

Nonneutral pattern of sequence variation in miRNA hairpins. miRNAs with SFS deviating from neutral expectations with Tajima’s D (A) and Fay and Wu’s H (B) are labeled in orange, and miRNAs with SFS compatible with neutrality are labeled in ivory. The distributions of D and H for protein-coding genes are shown for comparison with dashed lines.

Fig. 5.—

Signatures of purifying selection in miRNAs belonging to novel families unique to C. remanei and C. latens. (A) Selective constraints are stronger for mature sequences and paired sites. (B) Distribution of nucleotide variation in different region of miRNA hairpins.

Fig. 6.—

miRNAs from families unique to C. remanei are expressed at lower levels than miRNAs from families conserved in other Caenorhabditis species. Expression read counts from de Wit et al. (2009).

Fig. 7.—

Rapid evolution of a miRNA cluster. (A) The mir-64 cluster expanded by ancient and recent tandem duplications (supplementary fig. S8, Supplementary Material online). The number of mir-64 paralogs differs between C. remanei and C. latens because of the possible loss of mir-64c-1 in C. remanei. (B) Genetic differentiation of mir-64 members among three populations of C. remanei. mir-64c is highly differentiated between samples from Ohio and Ontario, suggesting that mir-64c may be the target of adaptive evolution. The dash line represents the 95 percentile of F_ST values in a set of protein-coding genes. (C) Distance matrices between paralogs and orthologs in C. remanei and C. latens for the entire hairpin sequence (below diagonal) and for the mature sequence (above diagonal). Crem, C. remanei; Cla, C. latens.