Integrity of prokaryotic mRNA isolated from complex samples for in vivo bacterial transcriptome analysis.

Even though several in vitro studies have focused on bacterial biology, the extent of such knowledge is not complete when considering an actual infection. As culture-independent microbiology methods such as high-throughput sequencing became available, important aspects of host-bacterium interactions will be elucidated. Based on microbiological relevance, we considered Bacteroides fragilis in a murine experimental infection as a model system to evaluate the in vivo bacterial transcriptome in host exudates. A disproportionate number of reads belonging to the host genome were retrieved in the first round of pyrosequencing, even after depletion of ribosomal RNA; the average number of reads related to the eukaryotic genome was 71.924-67.7%, whereas prokaryotic reads represented 34.338-32.3% in host exudates. Thus, different treatments were used to improve the prokaryotic RNA yield: i) centrifugation; ii) ultrasonic treatment; and iii) ultrasonic treatment followed by centrifugation. The latter treatment was found to be the most efficient in generating bacterial yields, as it resulted in a higher number of Bacteroides cells. However, the RNA extracted after this treatment was not of sufficient quality to be used in cDNA synthesis. Our results suggest that the methodology routinely used for RNA extraction in transcriptional analysis is not appropriate for in vivo studies in complex samples. Furthermore, the most efficient treatment for generating good bacterial cell yields was not suitable to retrieve high-quality RNA. Therefore, as an alternative methodological approach to enable in vivo studies on host-bacterium interactions, we advise increasing the sequencing depth despite the high costs.