Table 1. Families of orthologous genes regulated by transcription attenuation¹

¹On the basis of existing genome annotation, the 400 nt immediately upstream of every gene in the 180 fully-sequenced bacterial genomes available in GenBank (ftp://ftp.ncbi.nih.gov/genbank/) were identified and examined. We selected only those genes predicted to be immediately preceded by a site of intrinsic transcription termination. These sequences were considered the analysis window in our compilation. In this window, we searched for a “run of T's” (rts) that would correspond to a run of U's in RNA, an essential characteristic of an intrinsic terminator. The length of the rts selected was 6 nt, at least five of which must be T (“ts”). The arbitrarily selected maximum acceptable spacing between the rts and the first base of the downstream gene or coding region, was 110 nt. For every rts, our program searched for the most stable secondary structure that could be formed within the preceding 60 nt. Initially, the secondary structures that could be formed from this 60 nt was predicted using the RNAfold program. The only secondary structure that was considered to be part of a terminator (T), was a stem-and-loop structure (SLS). If the sequence analyzed contained more than one SLS, only the one closest to the rts was considered to be the terminator. In the event that folding predictions yielded structures different from a SLS (e.g., cloverleaf-like), the first nt of the 60 nt long sequence was removed and a search for the SLS repeated recursively until the program located a sequence, which when folded, corresponded to the most stable SLS. This SLS was considered part of a T only if the base at the bottom of the stem was no more than 4 nt from the rts, the maximum number of loops in the structure was 3, and the energy of the structure was at least -0.2 kcal/mol per nucleotide. For every T found, the program then searched for the presence of an associated anti-terminator (AT). To do this the program initially scanned a 60-nt sequence immediately upstream of the middle of the T loop. The program then located the most stable SLS using the procedure described for the T analysis. A SLS was considered an AT on the basis of the following somewhat arbitrary preferences: If at least 3 bases of its stem overlapped with the T sequence, the maximum number of loops in the structure was 4, and if it had a free energy that was at least one fourth that of its corresponding T. The previous analysis was also repeated with each AT, searching for its corresponding anti-anti-terminator structure (AAT). To consider the AAT and AT as mutually exclusive structures, they must share at least 3 bases. The free energy of the AAT must be at least one fourth that of it’s associated AT. The predicted terminator/antiterminator attenuators as well as those attenuators that do not depend on an antiterminator structure can be found on our web page (http://cmgm.stanford.edu/~merino/).

²According to the Cluster of Orthologous Groups of proteins database (COG). Where there is a predominant phylum regulated by transcription attenuation within each COG, this is indicated according to GenBank listing (http://www.ncbi.nih.gov/Genbank/). Fi stand for Firmicutes and Pr stands for Proteobacteria. COGs associated with riboswitches or conserved motif sequences are indicated in bold face.

³p-value of the over-representation of genes regulated by transcription attenuation in a specific COG assuming a hypergeometric distribution. COGs with p-values lower than 1x10^-4 or relevant COGs previously reported to have a clear tendency to be regulated by transcription attenuation are shown in the table. Only non-redundant strains were considered in our statistical analyses. We considered strains as redundant if they share more than 85% of their genes as orthologous genes with another organism previously included in this analysis.

⁴Indicates the type of transcription attenuation mechanism described in Box 1 that is used to regulate the members of this COG.