Weak Evolutionary Conservation of RNA-Protein Dispersion across Human and Arabidopsis Orthologs

Gene-specific relative RNA-protein dispersion, the difference between RNA and protein variation across tissues, may be conserved over evolutionary distance if particular genes experience consistent post-transcriptional regulatory constraints. We tested this in human and Arabidopsis thaliana by comparing a scale-invariant dispersion phenotype, B_disp, across 855 prespecified one-to-one orthologs, using tissue-resolved transcriptomic and proteomic atlases. Human and Arabidopsis B_disp showed a weak positive association (Spearman's ρ = 0.173, P < 0.001). Adjusting for protein detectability, RNA and protein abundance, and sequence conservation reduced it to ρ = 0.068 (bootstrap 95% confidence interval [CI] - 0.0003 to 0.136), although its direction was stable across four prespecified cohorts (ρ = 0.063-0.088). Masking 89 orthologs in four prespecified ancient systems (ribosome, proteasome, chaperone complexes and respiratory chain) lowered the adjusted association (Δρ = - 0.019, 95% CI -0.038 to -0.001) but did not change two alternative RNA-protein metrics. Exploratory analyses added after the primary analysis showed that this reduction was no larger than that from removing 89 random orthologs (two-sided P = 0.11) and was attributable to the respiratory-chain complex alone (24 pairs; Δρ = - 0.022; P = 0.00025). Among 447 sufficiently represented Gene Ontology (GO) terms, 119 unadjusted correlations, but no covariate-adjusted correlations, passed Benjamini-Hochberg correction. Genome-wide conservation of RNA-protein dispersion between human and Arabidopsis is therefore weak once abundance and detectability are considered, the residual signal is not localized to ancient core systems as a class, and the data do not identify a mode of natural selection.