Global transcriptional landscape and promoter mapping of the gut commensal Bifidobacterium breve UCC2003

Abstract

Background: Bifidobacterium breve represents a common member of the infant gut microbiota and its presence in the gut has been associated with host well being. For this reason it is relevant to investigate and understand the molecular mechanisms underlying the establishment, persistence and activities of this gut commensal in the host environment.

Results: The assessment of vegetative promoters in the bifidobacterial prototype Bifidobacterium breve UCC2003 was performed employing a combination of RNA tiling array analysis and cDNA sequencing. Canonical -10 (TATAAT) and -35 (TTGACA) sequences were identified upstream of transcribed genes or operons, where deviations from this consensus correspond to transcription level variations. A Random Forest analysis assigned the -10 region of B. breve promoters as the element most impacting on the level of transcription, followed by the spacer length and the 5'-UTR length of transcripts. Furthermore, our transcriptome study also identified rho-independent termination as the most common and effective termination signal of highly and moderately transcribed operons in B. breve.

Conclusion: The present study allowed us to identify genes and operons that are actively transcribed in this organism during logarithmic growth, and link promoter elements with levels of transcription of essential genes in this organism. As homologs of many of our identified genes are present across the whole genus Bifidobacterium, our dataset constitutes a transcriptomic reference to be used for future investigations of gene expression in members of this genus.

Keywords: Bifidobacteria; Gene expression; Probiotics; Transcription.

Fig. 1

Tiling arrays and RNA-Seq alignment. Genome-wide transcription of the B. breve UCC2003 chromosome obtained using a combination of tiling array and RNA-Seq technologies. a Alignment of transcription patterns of B. breve genes (ORFs) as obtained from tiling array (blue) or RNA-Seq (red) data. Expression levels are indicated as RPKM (Reads Per Kilobase of transcript per Million mapped reads) in RNA-Seq dataset and FC (Fold-Change expressed as RNA signal strength vs gDNA baseline) in tiling arrays. b Genome-wide transcription of the B. breve UCC2003 chromosome as determined by a strand-specific tiling array (red for the forward and blue for the reverse strand) or RNA-Seq (grey)

Fig. 2

B. breve transcriptional unit organization. Characteristic elements of a typical, highly expressed TU in B. breve. a Comparative transcription patterns of highly transcribed TU encoding the transaldolase/transketolase proteins. Directional tiling arrays signals are expressed in red (forward strand) and blue (reverse strand), while RNA-Seq expression levels are indicated in grey. From this representation it is possible to appreciate the non-even distribution of RNA-Seq reads across the transcript as compared to the array signals. b Schematic representation of a bifidobacterial TU with relative main features (e.g. the promoter elements −10 and −35 located upstream a TSS as well as the TTS (rho-independent transcriptional termination)

Fig. 3

Tiling arrays and RNA-Seq accuracy. Evaluation of the obtained accuracy in determining the exact position of TSSs using tiling array and RNA-Seq data. a Barplot showing the discrepancy between TSSs predicted by tiling array or RNA-Seq data as based on all genes. A green line indicates the ranges of detection with higher frequency of hits, while an orange line indicates the expected discrepancy value. From this chart it is possible to assess how the discrepancy in predicting a TSS ranges between 25 and 50 bp. b Discrepancy between TSS predicted by tiling array and RNA-Seq approaches when based only on high and medium transcribed genes. A green line indicates the ranges of detection with higher frequency of hits, while an orange line indicates the expected discrepancy value. As from the chart is possble to observe a decrease of discrepant predictions located at the extremities of the distribution when the lowly transcribed genes are removed

Fig. 4

B. breve transcriptional unit elements and termination. a Barplot showing the different strategies in transcriptional termination identified in B. breve UCC2003, grouped by level of transcription (high, medium and low). From the Barplot it is possible to observe the high occurrence of rho-independent termination in this organism. c Barblot comparing the length of 5’-UTR and 3’-UTR to genes with transcription level high, medium and low. From the chart it is possible to observe a higher 5’-UTR length in highly transcribed genes as compared to gene with a medium and low level of transcription

Fig. 5

Promoter strength. Results of the Random Forest (RF) computation and analysis of the variation of promoter consensus across various transcription levels in B. breve UCC2003. a Barplot showing the results of the RF analysis conducted on 62 bp regions upstream of predicted TSSs. The plot shows the importance of each base position across the 62 bp region where the promoter elements (−10 and −35) can be detected (red blocks). Importance values of a number of additional classifiers is also indicated (green block). b Weblogo showing the consensus motif identified for the −10 and −35 sequences of the canonical promoter in B. breve across the levels of expression high, medium and low. Also in this case the promoter elements (−10 and −35) can be detected

Fig. 6

Essential expressed genes. Analysis showing the overlap between transcription level in core and essential genes. a Barplot showing the distribution of lowly, medium and highly transcribed genes of the essential genes of B. breve UCC2003 and the core genome of Bifidobacterium compared to the whole transcriptome (total of genes). The chart also illustrates the overlap between the core and essential transcribed genes in this organism. b Stacked-column plot showing the percentage of essential (blue) and total (red) expressed genes organized by COG categories. The plot displays how housekeeping COG categories (e.g. translation, ribosomal structure and biogenesis as well as RNA processing and modification) are the most represented in our dataset