Formation of regulatory modules by local sequence duplication

Turnover of regulatory sequence and function is an important part of molecular evolution. But what are the modes of sequence evolution leading to rapid formation and loss of regulatory sites? Here, we show that a large fraction of neighboring transcription factor binding sites in the fly genome have formed from a common sequence origin by local duplications. This mode of evolution is found to produce regulatory information: duplications can seed new sites in the neighborhood of existing sites. Duplicate seeds evolve subsequently by point mutations, often towards binding a different factor than their ancestral neighbor sites. These results are based on a statistical analysis of 346 cis-regulatory modules in the Drosophila melanogaster genome, and a comparison set of intergenic regulatory sequence in Saccharomyces cerevisiae. In fly regulatory modules, pairs of binding sites show significantly enhanced sequence similarity up to distances of about 50 bp. We analyze these data in terms of an evolutionary model with two distinct modes of site formation: (i) evolution from independent sequence origin and (ii) divergent evolution following duplication of a common ancestor sequence. Our results suggest that pervasive formation of binding sites by local sequence duplications distinguishes the complex regulatory architecture of higher eukaryotes from the simpler architecture of unicellular organisms.

💡 Research Summary

The paper investigates how regulatory sequences, particularly transcription factor (TF) binding sites, are created and lost during evolution. While point mutations have long been recognized as a source of new sites, the authors propose that local sequence duplications play a dominant role in generating clusters of binding sites—so‑called cis‑regulatory modules (CRMs)—in the fruit fly Drosophila melanogaster.

Using a dataset of 346 experimentally defined CRMs (each >1 kb) from the REDfly database, the authors first compute an autocorrelation function Δ(r), which measures the excess probability that two nucleotides separated by distance r are identical compared with random expectation. They find a clear positive signal up to ~70 bp, decaying roughly exponentially. This signal cannot be explained solely by known sources such as microsatellite repeats, homopolymeric stretches, or background compositional bias.

To link the autocorrelation to functional motifs, the authors develop a similarity‑information framework. A motif of length ℓ is described by a probability distribution Q over ℓ‑mers, and its divergence from background P₀ is quantified by relative entropy H(Q|P₀). For any pair of sites of the same motif, the mutual entropy K(c,ℓ|c₀) captures the extra similarity beyond background, where c is the average nucleotide identity of the aligned positions. By scanning all possible site pairs within each CRM and selecting a non‑overlapping set that maximizes total similarity (via dynamic programming), they obtain distance‑dependent similarity information K_ℓ(r).

The analysis reveals that ℓ = 7 bp yields the maximal total similarity K_tot(ℓ), indicating that the dominant functional elements have a characteristic length of about seven bases—consistent with typical TF binding sites and with the length of many tandem repeats. Moreover, K₇(r) remains positive for distances up to ~50 bp, mirroring the autocorrelation profile. When known TF binding sites are masked, the similarity signal drops by roughly 50 %, showing that functional sites account for a substantial fraction of the observed correlation. In contrast, masking microsatellite repeats reduces the signal by less than 10 %, confirming that repeats are not the primary driver.

To infer the evolutionary origin of correlated site pairs, the authors construct a probabilistic model that evaluates two competing histories for any pair: (i) independent origin followed by point mutations, and (ii) a common ancestor that duplicated locally and then diverged under selection for possibly different TFs. Using a likelihood ratio test that incorporates mutation rates, genetic drift, and selection coefficients derived from biophysical binding models, they find that the duplication pathway is favored for the majority of site pairs separated by ≤50 bp. The probability that a pair arose by duplication declines sharply beyond this distance, matching the decay of K₇(r).

A comparative analysis with intergenic sequences from the unicellular yeast Saccharomyces cerevisiae shows virtually no distance‑dependent duplication signal, underscoring that the duplication‑driven mode of site generation is a hallmark of multicellular eukaryotes with complex regulatory architectures.

From a biophysical perspective, TF‑DNA binding energy is highly nonlinear: binding probability is near‑unity for energies above a threshold and drops sharply below it. Consequently, a “seed” sequence with marginal affinity can be rapidly refined by a few advantageous point mutations, especially if it already resides near an existing functional site. Local duplications provide precisely such seeds in the immediate genomic neighborhood, dramatically increasing the probability that a new functional TF binding site will emerge. The authors argue that this mechanism accelerates regulatory evolution and facilitates adaptive changes in gene expression.

In summary, the study provides (1) statistical evidence that a large fraction of neighboring TF binding sites in Drosophila CRMs share a recent common ancestor created by local duplication, (2) a quantitative model showing that duplication followed by divergent evolution explains the observed distance‑dependent similarity up to ~50 bp, (3) demonstration that this mechanism is largely absent in yeast, indicating it is a distinctive feature of higher eukaryotes, and (4) a mechanistic rationale linking duplication‑generated seeds to rapid adaptive formation of new regulatory elements. These findings reshape our understanding of how complex cis‑regulatory landscapes evolve, highlighting local sequence duplication as a pervasive and functionally important driver of regulatory innovation.