Genome reduction has more than one cause

A comparative genomics study suggests that genome size in bacteria and archaea cannot be explained by the same evolutionary rules alone. In the study published in PNAS, selective constraint generally strengthened as bacterial genomes grew larger, whereas the analyzed archaeal clusters showed the opposite relationship. The researchers interpreted these patterns as consistent with functional diversification and genome streamlining, respectively.

Several forces shape prokaryotic genome evolution. Selection favoring streamlining can maintain smaller genomes, while selection favoring functional diversification can support the expansion of genes and functions. Horizontal gene transfer provides a route for acquiring genes needed for functional expansion. But a small genome is not always the product of strong selection for streamlining. In host-associated microbes with small effective population sizes, selection can become less efficient, and genetic drift and gene loss can combine to shrink genomes.

Earlier studies relied mainly on comparisons between genome pairs or on averages across multiple lineages. These approaches made it difficult to adequately distinguish differences among lineages from constraints acting on individual genes.

Differences among lineages across 791 genome clusters

The researchers began with ATGC, a database of closely related genome clusters containing more than 22,000 genomes. After applying quality criteria, including criteria concerning the number of variants, the final analysis included 791 clusters: 772 bacterial and 19 archaeal. These figures count clusters of closely related genomes, not species.

Selective constraint was estimated using dN/dS, the ratio of the nonsynonymous substitution rate to the synonymous substitution rate. Nonsynonymous substitutions are nucleotide substitutions that change an amino acid in a protein; synonymous substitutions do not. The researchers calculated this ratio while accounting for phylogenetic relationships, then decomposed it into lineage-specific and gene-specific components. The lineage-level measure of selective constraint used in the analysis varies in the opposite direction to dN/dS, so a higher value indicates stronger constraint.

Across bacteria, the Spearman correlation coefficient between genome size and selective constraint was 0.28. This is consistent with the interpretation that larger genomes experience stronger selection to maintain a broader range of functions. By contrast, the correlation across the 19 archaeal clusters was −0.82, showing a pattern consistent with streamlining, in which smaller genomes had stronger selective constraint. In 10,000 resampling analyses of bacterial clusters, drawing 19 clusters each time, no negative correlation as strong as the one in archaea was obtained.

Bacteria were not uniform, either. Some taxonomic groups showed no significant relationship, while the order Bacillales showed a negative correlation consistent with streamlining. The overall bacterial pattern therefore cannot be applied to every bacterial lineage.

The causes of lineage differences remain unresolved

The findings provide evidence that the relationship between genome size and selective constraint varies across lineages. However, the study did not experimentally establish causal links with environmental conditions. Functional diversification may contribute to the positive correlation in bacteria, but so may a process in which weak selection and genetic drift in populations with small effective sizes promote genome reduction.

The researchers proposed differences between microbes that exploit diverse ecological opportunities and those specialized for stable or resource-limited environments as a possible explanation. This is the authors’ hypothesis; the cause of the contrast between bacteria and archaea has not yet been established. The resampling analysis provides evidence that sample-size differences alone are unlikely to explain the result, but the archaeal analysis was limited to 19 clusters, which must also be considered.

Future work should test whether these relationships hold across a wider range of lineages and ecological conditions. The methods, results, and basis for the interpretation are available in the open full text of the PNAS paper.