Selectional versus mutational mechanism underlying genomic features of bacterial strand asymmetry: a case study in Clostridium acetobutylicum.

Strand biases are widespread in bacterial genomes. In this review, we discuss 5 types of bias, including gene orientation, the number of open reading frames, nucleotide composition, substitution rate, and gene length, between leading and lagging strands during replication. For each type of strand bias, related studies were summarized and Clostridium acetobutylicum ATCC 824 was used as a representative example to illustrate bias. Our results in C. acetobutylicum indicate that there is little asymmetry between 2 replication strands on open reading frame number and gene length, whereas the other 3 features presented significant strand bias. The underlying mechanisms of mutation and/or selection are discussed. It is hoped that this review will improve the understanding of the extent and reasons for various types of strand bias in bacterial genomes.

Statistical analyses of conserved features of genomic islands in bacteria.

We performed statistical analyses of five conserved features of genomic islands of bacteria. Analyses were made based on 104 known genomic islands, which were identified by comparative methods. Four of these features include sequence size, abnormal G+C content, flanking tRNA gene, and embedded mobility gene, which are frequently investigated. One relatively new feature, G+C homogeneity, was also investigated. Among the 104 known genomic islands, 88.5% were found to fall in the typical length of 10-200 kb and 80.8% had G+C deviations with absolute values larger than 2%. For the 88 genomic islands whose hosts have been sequenced and annotated, 52.3% of them were found to have flanking tRNA genes and 64.7% had embedded mobility genes. For the homogeneity feature, 85% had an h homogeneity index less than 0.1, indicating that their G+C content is relatively uniform. Taking all the five features into account, 87.5% of 88 genomic islands had three of them. Only one genomic island had only one conserved feature and none of the genomic islands had zero features. These statistical results should help to understand the general structure of known genomic islands. We found that larger genomic islands tend to have relatively small G+C deviations relative to absolute values. For example, the absolute G+C deviations of 9 genomic islands longer than 100,000 bp were all less than 5%. This is a novel but reasonable result given that larger genomic islands should have greater restrictions in their G+C contents, in order to maintain the stable G+C content of the recipient genome.