The LXXLL nuclear receptor motif is among the most conserved interaction modules in metazoan transcription. Across a 48-species sarcopterygian sample, the NR1 region was invariant as LVQLL in all non-monotreme species. Yet in both extant monotremes—the platypus and the short-beaked echidna—the motif was fixed as PGQLP, a change so radical that it demanded explanation.

Several independent lines of evidence converged on gBGC as the mechanism. In both monotreme genomes, the NR1 region formed a local GC peak. The monotreme branch carried a weak-to-strong substitution bias across the entire gene, not merely at NR1, exactly as expected under gBGC. Branch-level tests reported positive selection on the monotreme lineage, but no site-level test detected adaptive change at NR1 itself.

When the analysis was restricted to GC-conservative substitutions—which gBGC cannot drive—the monotreme-branch dN/dS remained below 1. This finding is critical: it demonstrates that once the confounding influence of gBGC is removed, there is no evidence of adaptive protein evolution at this locus.

The derived PGQLP motif was identical between platypus and echidna across their approximately 55-million-year divergence and was fixed at the DNA level across 52 platypus genomes. This uniformity indicates that the replacement occurred before the two lineages split and has been stably maintained since.

Structural modeling added a functional dimension to the story. FoldX predictions indicated weaker receptor binding for PGQLP than for the ancestral LVQLL, with a ΔΔG of approximately +6.9 kcal/mol. This suggests the replacement was not beneficial—it was, if anything, mildly deleterious, yet it fixed anyway because gBGC overrides selection at weakly selected sites.

The same process appears to have acted independently in ray-finned fishes. There, NR1 was the least constrained NR box, and the direction of its substitutions tracked the local GC background with striking precision (Spearman r = 0.93, p = 0.008). This parallel finding suggests that gBGC-driven collateral replacement of conserved motifs may be more widespread than previously appreciated.

The implications extend beyond monotremes. gBGC is a known cause of false positive-selection signals in molecular-evolution analyses, yet documented examples of gBGC fixing a protein-altering change at a specific, conserved coding element remain rare. This locus provides a stringent test case, given the extreme conservation of the LXXLL motif across metazoan transcription.

The combination of tests used in this study—examining local GC content, substitution bias direction, GC-conservative dN/dS, site-level selection tests, and structural modeling—offers a framework for distinguishing gBGC from adaptive evolution when conserved coding motifs are altered in lineages with unusual recombination patterns or GC content. As genomic datasets grow, such approaches will become increasingly essential for accurately reading the evolutionary record.

To appreciate why this finding is so striking, it helps to understand what the LXXLL motif actually does. The motif takes its name from a five-residue pattern in which leucines occupy the first, second, and fifth positions, flanked by two variable residues. This short stretch of amino acids serves as the primary docking interface by which coactivator proteins such as NCOA1 attach themselves to the ligand-binding domains of nuclear receptors, the family of transcription factors that respond to steroid hormones, retinoids, thyroid hormone, and vitamin D. The hydrophobic leucine side chains slot into a complementary hydrophobic groove on the receptor surface, an interaction so central to hormone signaling that its disruption would be expected to ripple through endocrine physiology. The fact that the first of NCOA1’s several NR boxes could be swapped for a proline-rich sequence without obvious catastrophic consequences underscores how much redundancy and buffering exists in these signaling networks, and how weakly selected individual interaction motifs can be.

The mechanism implicated here, GC-biased gene conversion, arises from the mechanics of meiotic recombination. When homologous chromosomes exchange information during meiosis, the heteroduplex intermediates formed at recombination sites contain mismatches that must be resolved by DNA repair machinery. This repair is not perfectly unbiased: in many vertebrates, the machinery preferentially restores G and C nucleotides over A and T. The result is a non-Mendelian segregation distortion that behaves, from the perspective of population genetics, like selection favoring GC alleles, even when those alleles carry no benefit or even a modest cost to the organism. Because the effective strength of this bias scales with recombination rate and local GC content, genomic regions experiencing high recombination can accumulate weak-to-strong substitutions at rates that mimic adaptive evolution in standard comparative tests.

This is precisely why gBGC has long been recognized as a confound in studies that use dN/dS ratios and related codon-based statistics to detect positive selection. Branch-site models, which are among the most widely used tools for identifying episodic adaptation, can interpret an excess of weak-to-strong changes as evidence of diversifying selection, because such changes disproportionately alter encoded amino acids. The monotreme NCOA1 case illustrates the trap in its most vivid form: a motif conserved across hundreds of millions of years of sarcopterygian evolution appears, on a branch-level scan, to have undergone rapid, selection-like change, when in fact the underlying force was recombination-associated repair bias acting on linked sites throughout the gene.

The monotremes are particularly plausible candidates for such effects. As the sole surviving lineage of the mammalian branch that diverged before the emergence of marsupials and placentals, they carry genomes with distinctive GC landscapes and recombination profiles shaped by their long independent history. Their roughly 55-million-year separation, with the motif change fixed identically in both the platypus and the echidna, places the event deep in monotreme ancestry, before the two lineages parted ways. The complete fixation observed across 52 platypus genomes sampled at the DNA level leaves no room for segregating variation at the site, reinforcing the conclusion that whatever drove the change, it acted with the force of a sweep, even though the driving process was conversion bias rather than selection on phenotype.

The structural prediction of weakened binding deserves careful interpretation. A ΔΔG of roughly +6.9 kcal/mol is a substantial penalty for a protein-protein interface, suggesting that the derived PGQLP form interacts less favorably with nuclear receptors than the ancestral sequence did. Proline residues are notoriously disruptive in helical contexts, and the LXXLL motif functions as a short amphipathic helix when docked, so replacing leucines with prolines would be expected to compromise that geometry. Yet the change fixed and persisted. This apparent paradox resolves once one recognizes that gBGC can drive alleles to fixation against weak purifying selection, provided the fitness cost is small relative to the conversion bias. The case therefore offers a concrete example of a molecular change that is neither adaptive nor effectively neutral in the classical sense, but rather the product of a transmission distortion that outruns selection.

The ray-finned fish analysis broadens the significance considerably. If NR1 substitutions in fish track local GC background with a correlation as strong as the reported Spearman coefficient of 0.93, then the same conversion-driven dynamics are shaping this motif across vertebrate phylogeny, not merely in one odd lineage. It also suggests that among the multiple NR boxes in NCOA1, NR1 may be under the weakest functional constraint, making it the box most vulnerable to collateral replacement. Whether other coactivators and other interaction motifs show similar vulnerability remains an open question, but the framework developed here—combining local GC profiling, substitution directionality, GC-conservative dN/dS estimation, site-level tests, and structural modeling—provides a template for investigating them.

For evolutionary genomics more broadly, the lesson is one of interpretive caution. Conserved motifs that suddenly change in a particular lineage have traditionally been read as signatures of adaptation, and often rightly so. But this study demonstrates that the same pattern can emerge from a process with no adaptive content whatsoever, one that fixes changes regardless of their functional consequences. As recombination maps and population genomic data become available for more species, distinguishing these scenarios will become increasingly tractable, and cases like the monotreme NCOA1 motif will serve as valuable benchmarks for validating the diagnostic toolkit.

Subject of Research: GC-biased gene conversion collateral replacement of the conserved NCOA1 NR1 nuclear-receptor coactivator motif in monotremes

Article Title: Collateral replacement of a conserved nuclear-receptor coactivator motif by GC-biased gene conversion in monotremes

Article References: Cho, E. S. (2026). Collateral replacement of a conserved nuclear-receptor coactivator motif by GC-biased gene conversion in monotremes. BMC Genomics. https://doi.org/10.1186/s12864-026-13443-x

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13443-x

Keywords: GC-biased gene conversion, monotremes, platypus, echidna, LXXLL motif, nuclear receptor coactivator, NCOA1, molecular evolution, false positive selection, genome evolution, dN/dS, ray-finned fishes