RNA secondary structures are conserved but random

Abstract

Noncoding RNAs perform a wide range of essential biological functions, and their secondary structures are often conserved by purifying selection. However, such conservation does not necessarily imply that positive selection shaped their evolutionary origins. Here, we test for global signs of positive selection by studying the distribution of secondary structures in naturally occurring noncoding RNA. We find that, to first order, these structures are statistically indistinguishable from those produced by a relatively small set of randomly generated sequences. The distributions are, however, profoundly shaped by a strong bias in the arrival of phenotypic variation, such that only an exponentially small subset of all possible structures is likely to occur in nature. In other words, the secondary structure repertoire of natural noncoding RNAs largely reflects this developmental bias rather than further adaptive fine-tuning. Detecting genuine signatures of selection, beyond randomness, in the distribution of secondary structures, therefore requires careful calibration against appropriate null models that account for the underlying bias. We perform a large-scale and detailed analysis of four extensive datasets covering a wide spectrum of functional RNA classes. We describe one potential signature of adaptation on structure: archaeal ribosomal RNA structures are simpler and more robust than predictions of the sequence null model, and hyperthermophiles are less complex than archaea in other niches, but the effects are relatively small. This example illustrates the difficulty of inferring a creative role for natural selection in shaping evolutionary outcomes for RNA secondary structure.