cWhile the model organism Escherichia coli has been the subject of intense study for decades, the full complement of its RNAs is only now being examined. Here we describe a survey of the E. coli transcriptome carried out using a differential RNA sequencing (dRNA-seq) approach, which can distinguish between primary and processed transcripts, and an automated prediction algorithm for transcriptional start sites (TSS). With the criterion of expression under at least one of three growth conditions examined, we predicted 14,868 TSS candidates, including 5,574 internal to annotated genes (iTSS) and 5,495 TSS corresponding to potential antisense RNAs (asRNAs). We examined expression of 14 candidate asRNAs by Northern analysis using RNA from wild-type E. coli and from strains defective for RNases III and E, two RNases reported to be involved in asRNA processing. Interestingly, nine asRNAs detected as distinct bands by Northern analysis were differentially affected by the rnc and rne mutations. We also compared our asRNA candidates with previously published asRNA annotations from RNA-seq data and discuss the challenges associated with these cross-comparisons. Our global transcriptional start site map represents a valuable resource for identification of transcription start sites, promoters, and novel transcripts in E. coli and is easily accessible, together with the cDNA coverage plots, in an online genome browser.A fter many years of study, we are only now beginning to understand and appreciate the complexity of bacterial transcriptomes. With the recent advances in deep-sequencing technology, transcriptome sequencing (RNA-seq) now allows for the detection of transcripts that are present at low levels or were previously missed by other methods of detection, the generation of global transcript maps, and improved genome annotation (reviewed in references 1 and 2). While these studies provide vast amounts of information about bacterial transcriptomes and regulatory elements, they also raise challenges regarding comparisons between studies and functions of the newly identified transcripts.One group of underappreciated transcripts being uncovered by these genome-wide analyses are RNAs that map opposite annotated coding regions, termed antisense RNAs (asRNAs). The abundance of pervasive antisense transcription start sites (asTSS) was first highlighted in an RNA-seq survey of the human pathogen Helicobacter pylori, where asTSS were identified opposite ϳ46% of the genes (3). Subsequent RNA-seq studies in cyanobacteria (4) and Gram-negative (5, 6) and Gram-positive (7-9) bacteria identified asRNAs expressed opposite 2 to 30% of annotated genes. This wide range in numbers of asRNAs reported may reflect differences in bacterial lifestyle or differences in the experimental setup or analyses of the RNA-seq data sets.Even for the transcriptome analyses of the well-studied model organism Escherichia coli (10-22), the numbers of asRNAs reported range from hundreds to thousands. This significant variation is due, in part, to differences i...