All cultures were incubated at 21°C and constantly irradiated with 28 μmol quanta m-2 s-1. Results Transcriptome structure of Prochlorococcus MED4 The Illumina high-throughput sequencing (RNA-Seq) protocols were applied to ten Prochlorococcus MED4 samples cultured in Pro99 and AMP (Table 1; Methods). Altogether, 62.8 million 90-bp pair-end reads were generated, and approximately 51.0 million pair-end reads (81.3%) were perfectly mapped to the genome (Table 1). Collectively, 91.8% of the MED4 genome was transcribed for at least one growth condition,

Wnt inhibitor review and 61.2% of the genome was transcribed in all conditions. The transcribed regions might be larger if more growth conditions are tested. The genome expression cut-off was defined as the coverage of the tenth percentile of the lowest expressed genome regions [23] (Table 1). In contrast, 96.6% of 1965 coding-sequence (CDS) genes were expressed in at least one growth condition, and 80.9% were expressed in all conditions. Gene

expression selleck chemicals cut-off was defined as the mean RPKM (reads per kilobase per million mapped reads [26]) of the ten percentages of the lowest expressed gene regions (Table 1). The RNA-Seq reads mapping allow us to globally identify transcripts’ boundaries and adjacent gene regions [22–24]. To obtain a genome-wide operon map, a putative operon was characterized if it was repeatedly observed in at least three

samples (Methods). Using this criterion, 55.5% of all genes were assigned to 422 primary operons (Additional file 1), representing the first operon map of Prochlorococcus based on experimental data. The operon map completely or partially shared 73.4% of operon genes within predicted operons identified by the Prokaryotic Operon DataBase [27]. The remaining operons comprised many new genes recently predicted by Kettler et al. and Steglich et al.[6, 28] (Figure 1). The majority of the operons (63.0%) identified in this study were composed of two mafosfamide genes. The largest operon identified was a ribosomal protein operon containing 20 genes, and this was consistent with previously published observation made by Steglich et al.[29]. Furthermore, those extensively characterized operons, such as kaiBC circadian clock [30], two-component system phoRB[31], photosystem I core apparatus psaAB[32], and carboxysome shell proteins cso cluster [33], were also included in the operon map (Additional file 1). Figure 1 Operon map comparison. The operon map experimentally generated by this study compared with a bioinformatically predicted operon map generated by the Prokaryotic Operon DataBase (ProOpDB) [27].