Classical evolutionary theory maintains that mutation rate variation between genes should be random with respect to fitness 1-4 and evolutionary optimization of genic mutation rates remains controversial 3,5 . However, it has now become known that cytogenetic (DNA sequence + epigenomic) features influence local mutation probabilities 6 , which is predicted by more recent theory to be a prerequisite for beneficial mutation rates between different classes of genes to readily evolve 7 . To test this possibility, we used de novo mutations in Arabidopsis thaliana to create a high resolution predictive model of mutation rates as a function of cytogenetic features across the genome . As expected, mutation rates are significantly predicted by features such as GC content, histone modifications, and chromatin accessibility. Deeper analyses of predicted mutation rates reveal effects of introns and untranslated exon regions in distancing coding sequences from mutational hotspots at the start and end of transcribed regions in A. thaliana . Finally, predicted coding region mutation rates are significantly lower in genes where mutations are more likely to be deleterious, supported by numerous estimates of evolutionary and functional constraint. These findings contradict neutral expectations that mutation probabilities are independent of fitness consequences. Instead they are consistent with the evolution of lower mutation rates in functionally constrained loci due to cytogenetic features, with important implications for evolutionary biology 8 .Monroe et al.
Genome-wide mutation bias in A. thalianaA core maxim of evolutionary biology, codified through experiments completed in the early 1940s 1 , is that the "consequences of a mutation have no influence whatsoever on the probability that this mutation will or will not occur." 2 This assertion has profound implications for understanding organismal evolution and predicting human disease. That genic mutation rates might have been optimized during evolution has been extensively challenged with a strong argument: selection for mutation rates on a gene-by-gene basis cannot overcome the barrier of genetic drift 3 . And while reports of non-random relationships between mutation rates and fitness consequences have been previously made, these have been questioned because they have largely relied on substitution rates in natural populations rather than direct measures of de novo mutations 3,9-12 .More recently though, discoveries in genome biology have inspired a reevaluation of classical theories of mutation rate evolution. It is now recognized that mutation rates across genomes are influenced by DNA sequence composition, epigenomic features, and bias in the targets of DNA repair mechanisms 5,6,[13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32] . It is also known that broad classes of genes (e.g., housekeeping genes) exist in distinct cytogenetic (DNA sequence + epigenomic) states. This provides an opportunity for mutation rates to evolve in beneficial directi...