Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.


Important genomic regions mutate less often than do other regions

Salvador Luria and Max Delbrück made a profound discovery in 1943 that won them a Nobel prize, shared with Alfred Hershey, 26 years later. What they found was that bacterial mutations that confer resistance to a virus arise at the same rate, regardless of whether the virus is present1. That the generation of mutations (a process called mutagenesis) is blind to its consequence has since become an established principle of genetics. Writing in Nature, Monroe et al.2 report that, in stark contrast to this tenet, the rate of mutation in the model plant Arabidopsis thaliana is lower in genomic regions that are functionally more important, and in regions where mutations are more frequently harmful.

By analysing thousands of mutations collected in mutation-accumulation experiments, the authors find that the mutation rate is 58% lower inside genes than in regions immediately outside them, and 37% lower in essential genes (those indispensable for viability or fertility) than in non-essential genes. Furthermore, the authors observe a negative correlation between the proportion of mutations in a gene that are deleterious and the mutation rate of the gene.

Monroe et al. are not the first to describe such apparently advantageous patterns of variation in the rate of mutation across a genome. For example, a previous study3 reported that highly expressed genes in the bacterium Escherichia coli have relatively low mutation rates. This trend has been suggested to be an evolutionary ‘risk-management’ strategy3, because the detriment imposed by a mutation tends to increase with the expression level of the mutated gene4. Similarly, another study5 proposed that gene expression in the human testes is regulated to optimize gene-specific rates of mutations that are transmitted to the next generation. However, the results of both of these studies have been contested, owing to confounding factors in mutation-rate estimation, and a lack of viable mechanisms68.

What mechanisms cause crucial genomic regions to mutate less in A. thaliana? Monroe et al. noticed that the mutation rate of a given genomic region (in the study, a stretch of 1,000 nucleotide bases) is correlated with several genomic features. Among these are the percentage of nucleotides in the region that are guanine or cytosine, and epigenetic features of the region — molecular modifications that affect gene activity without changing the DNA sequence. These include various modifications to histone proteins that bind to DNA and affect gene regulation, DNA replication and DNA packaging. Monroe et al. propose that these genomic features and (especially) epigenetic features together form part of the machinery that is shaped by natural selection to reduce mutagenesis of important genomic regions.

The evolutionary selective pressure for mutagenesis-reducing machinery should be weak, because the machinery does not directly affect the fitness of the organisms that carry it. Rather, it affects the fitness of their offspring, owing to differences in their numbers of newly generated mutations9. In organisms such as A. thaliana that reproduce by selfing (the union of male and female sex cells from the same organism), the strength of selection for this machinery approximates the number of deleterious mutations per individual per generation that the machinery prevents9,10. Monroe et al. estimate that, in the face of genetic drift (random fluctuation of frequencies of genetic variants in a population), a machinery that lowers the mutation rate of essential genes by 30% must influence at least one-third of all coding sequences of all essential genes in A. thaliana for it to be established by natural selection. Hence, a mutagenesis-reducing machinery is unlikely to have emerged through adaptive evolution unless it has large and broad effects.

The suppression of mutagenesis in important genomic regions could, in theory, originate in two ways. First, because epigenetic modifications regulate gene expression, epigenetic features probably differ between genomic regions within and outside genes, and also between genes that show drastically different expression levels or regulation (for example, those that are continuously expressed and those that are expressed only in certain tissues or in response to certain environmental factors). The relationship between the expression or regulation of a genomic region and the functional importance of the region might thus create a correlation between the epigenetic feature of a region and the probability that a mutation in the region would be deleterious. Consequently, selection might lead to the evolution of machinery that lowers mutagenesis in regions that exhibit an epigenetic feature that correlates with high probability of a mutation being deleterious (Fig. 1a).

figure 1

Figure 1 | Routes to lower mutation rates in more-important genomic regions. Monroe et al.2 analysed mutations in the model plant Arabidopsis thaliana, and found that genomic regions important for plant viability and reproduction mutate less often (show reduced mutagenesis) than do other regions. This variation in mutation rate could originate in one of two ways. a, If a genomic feature or an epigenetic feature (a modification that affects gene activity without changing the DNA sequence) is present at important genomic regions, but not at non-important regions, natural selection could drive the emergence of a mutation-rate-reducing machinery that is associated with the feature. This would reduce the number of mutations that occur in the important regions over multiple generations. b, Alternatively, the association between the feature and reduction in mutation rate could be intrinsic or a by-product of some other biological processes, without selection.

Second, the association between a genomic or epigenetic feature and mutation rate might not be a result of selection for lower mutagenesis. Instead, it might be intrinsic to the feature (owing to its chemical nature) or a by-product of some other biological processes11 (Fig. 1b). Intriguingly, although selection for lower mutagenesis should be orders of magnitude weaker in the non-selfing forest tree Populus trichocarpa than in the selfing A. thaliana9,10, Monroe and colleagues present evidence suggesting similar mutation-rate profiles between the two species. This finding supports this second approach to explaining the origin of suppressed mutagenesis in important genomic regions.

It is worth emphasizing that, in both scenarios, the enrichment of certain genomic or epigenetic features at important regions occurs not because these regions have a high probability of deleterious mutations, but because of some correlates of that probability, such as gene expression or regulation. Hence, some variations in mutation rate across the genome might merely reflect these correlates. For example, Monroe et al. find that the outermost coding parts of a gene mutate more than other coding parts do. Moreover, genes that lack untranslated regions in their messenger RNAs have higher coding mutation rates than do other genes. And genes with few non-coding segments (introns) have higher coding mutation rates than do genes with more introns. Whether these mutational patterns are beneficial to the plant is unclear.

Even when mutagenesis-reducing machinery recognizes a particular genomic or epigenetic feature, selection for lower mutagenesis cannot drive the acquisition of the feature at an important genomic region. This is because the feature’s beneficial effect on mutation rate in that one region is too small to overcome the effect of genetic drift6,8.

Monroe et al. propose that, because A. thaliana’s mutation-rate profile reduces the overall chance that a new mutation is deleterious, the profile increases the chance that a mutation is beneficial. This statement, however, need not be true, because lowering mutagenesis in crucial genomic regions could reduce the proportion of mutations that are deleterious as well as the proportion of those that are beneficial — provided that these types are concentrated, and neutral mutations under-represented, in important regions.

Mutation and selection are generally considered to be distinct evolutionary forces. But if mutation rate is shaped by selection to different extents in genomic regions of different importance, as Monroe et al. suggest, this distinction would be blurred, and many evolutionary phenomena would require reinterpretation. Most notably, differences between genomic regions in DNA-sequence variation within a species (known as polymorphism) and between species have been commonly explained by a variation in selection — but they might also be caused by a variation in mutation rate. Indeed, the authors observe a striking similarity between mutation-rate variation and polymorphism variation among genomic regions in A. thaliana, suggesting that the latter is largely attributable to the former.

Although I am not ready to throw out the fundamental tenet of Luria and Delbrück, the intriguing mutation-rate pattern of A. thaliana makes me wonder whether the same pattern exists in many other species — and, if so, what the underlying mechanism is, and how it originated in evolution.

Nature 602, 38-39 (2022)



  1. Luria, S. E. & Delbruck, M. Genetics 28, 491–511 (1943).

    PubMed  Article  Google Scholar 

  2. Monroe, J. G. et al. Nature 602, 101–105 (2022).

    Article  Google Scholar 

  3. Martincorena, I., Seshasayee, A. S. N. & Luscombe, N. M. Nature 485, 95–98 (2012).

    PubMed  Article  Google Scholar 

  4. Zhang, J. & Yang, J. R. Nature Rev. Genet. 16, 409–420 (2015).

    PubMed  Article  Google Scholar 

  5. Xia, B. et al. Cell 180, 248–262 (2020).

    PubMed  Article  Google Scholar 

  6. Chen, X. & Zhang, J. Mol. Biol. Evol. 30, 1559–1562 (2013).

    PubMed  Article  Google Scholar 

  7. Maddamsetti, R. et al. Mol. Biol. Evol. 32, 2897–2904 (2015).

    PubMed  Article  Google Scholar 

  8. Liu, H. & Zhang, J. Mol. Biol. Evol. 37, 3225–3231 (2020).

    PubMed  Article  Google Scholar 

  9. Kimura, M. Genet. Res. 9, 23–34 (1967).

    Article  Google Scholar 

  10. Lynch, M. Genome Biol. Evol. 3, 1107–1118 (2011).

    PubMed  Article  Google Scholar 

  11. Chen, X., Yang, J.-R. & Zhang, J. Genome Res. 26, 50–59 (2016).

    PubMed  Article  Google Scholar 

Download references

Competing Interests

The author declares no competing interests.


Nature Careers


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing


Quick links