Comparative transcriptomics of pathogenic and non-pathogenic Listeria species

Abstract

Listeria monocytogenes is a human, food-borne pathogen. Genomic comparisons between L. monocytogenes and Listeria innocua, a closely related non-pathogenic species, were pivotal in the identification of protein-coding genes essential for virulence. However, no comprehensive comparison has focused on the non-coding genome. We used strand-specific cDNA sequencing to produce genome-wide transcription start site maps for both organisms, and developed a publicly available integrative browser to visualize and analyze both transcriptomes in different growth conditions and genetic backgrounds. Our data revealed conservation across most transcripts, but significant divergence between the species in a subset of non-coding RNAs. In L. monocytogenes, we identified 113 small RNAs (33 novel) and 70 antisense RNAs (53 novel), significantly increasing the repertoire of ncRNAs in this species. Remarkably, we identified a class of long antisense transcripts (lasRNAs) that overlap one gene while also serving as the 5' UTR of the adjacent divergent gene. Experimental evidence suggests that lasRNAs transcription inhibits expression of one operon while activating the expression of another. Such a lasRNA/operon structure, that we named 'excludon', might represent a novel form of regulation in bacteria.

Figure 1

The Listeria transcriptome browser. (A–C) Represent different windows of the genome in the unified browser. X axis represents the position on the genome; y axis the number of cDNA sequences mapped to the genome (log scale). The browser unifies TSS mapping (vertical lines being 5′ ends of genes, with numbers indicating the amount of supporting 5′ end sequencing reads), total RNA sequencing (black line), tiling-array data (red and blue small triangles represent probes on the forward and reverse strands, respectively), genome annotation (arrows underneath the axis; red and blue representing genes on the forward and reverse strands, respectively), operon annotation (yellow arrow), rho-independent terminators prediction (small black arrows), small RNAs (purple), asRNAs (green), as well as additional information (Materials and methods). The browser is available at http://www.weizmann.ac.il/molgen/Sorek/listeria_browser/.

Figure 2

5′ UTRs in L. monocytogenes and L. innocua. The length distributions of the 5′ UTRs in L. monocytogenes (A) and L. innocua (B) are almost identical, with a median length of 33 nt. The distributions are skewed to the right, representing long 5′ UTRs that include known riboswitches. (C) Divergence of 5′ UTR lengths between L. monocytogenes and L. innocua: the lengths of the 5′ UTRs of genes that have expressed homologs in both organisms were compared. (D) An example for a significant difference in 5′ UTR lengths between homologous genes. The lmo1654 gene is a predicted secreted metalloprotease of L. monocytogenes and has a 179 nt long 5′ UTR. Its homolog in L. innocua, lin1694, has a shorter 5′ UTR of 107 nt.

Figure 3

sRNAs in L. monocytogenes and L. innocua. Small RNAs (yellow box-arrows) encoded in the intergenic regions of L. monocytogenes (Lmo) and L. innocua (Lin) genes (blue box-arrows) are defined by an arrow corresponding to the TSS (red arrows on the plus strand; blue arrows on the minus strand, crossed gray arrow if the transcript is not expressed). Size of the transcript (dashed blue arrow) was estimated from the northern blot (right side of the figure) performed using the RNA extracted from L. monocytogenes and L. innocua in the exponential (exp) and stationary phase (stat) with the labeled probeset generated on either the L. monocytogenes (Lmo probe) or L. innocua genome (Lin probe). (A) Rli146 is a small RNA conserved in both genomes but not detected in L. innocua by either RNA sequencing or northern blot. Probes designed specifically to either L. monocytogenes or L. innocua successfully hybridized only in L. monocytogenes validating that the sRNA is not expressed in L. innocua. Differences in the promoter regions that might explain the differences in expression between the species are shown. (B) SbrA and Rli42 are 70 and 150 nt small RNAs conserved and expressed in both species. The SigB-dependent transcription of SbrA is supported by the presence of SigB box (σ-B) within its promoter region and the inability to detect a transcript in the ΔsigB mutant by northern blot. The conserved SigB box preceding SbrA in both L. monocytogenes and L. innocua is shown. Rli42 is a small RNA transcribed opposite to SbrA. (C) Rli130 is a novel sRNA detected in this study that is conserved and expressed in both L. monocytogenes and L. innocua. Northern blot suggests that Rli130 might exist as part of the lmo1934 mRNA as a band of 600 nt is observed. In L. innocua, the shortest form of Rli130 (230 nt) is missing, possibly due to sequence divergence between the two organisms (11 differences between positions 179 and 230 of that gene).

Figure 4

Small antisense RNAs in L. monocytogenes and L. innocua. Small antisense RNAs (yellow box-arrows) transcribed opposite to L. monocytogenes (Lmo) and L. innocua (Lin) genes (blue box-arrows) are defined by green arrows corresponding to TSSs (crossed gray arrow if the transcript was not detected). The size of each transcript (dashed blue arrow) was estimated from the northern blot. Northern blots (right side of the figure) were performed using the RNA extracted from both organisms in the exponential (exp) and stationary phase (stat) with a probe designed against either the L. monocytogenes (Lmo probe) or L. innocua transcript (Lin probe). Where probes cross-hybridized with transcripts from both species and generated identical hybridization patterns, a blot with only one probe is shown. (A) anti1292 is a conserved and expressed sasRNA in both L. monocytogenes and L. innocua. Its transcription starts within the coding region of lmo1292 and terminates near the TSS of the same gene, generating a 350 nt long product that is presumably processed to 300 nt long product or terminated earlier, as observed by the northern blot. (B) anti0671 is an sasRNA conserved and expressed in both L. monocytogenes and L. innocua. Its promoter region contains a SigB box (σ-B), and SigB regulation was further confirmed by northern blot analysis which did not detect an anti0671 transcript in the ΔsigB mutant. The SigB box is not completely conserved in L. innocua. (C) anti1255 is an antisense RNA conserved in both species. It is transcribed from two promoters, one generating a 1.4-kb transcript, and second promoter generating a 2.6-kb transcript. Both transcripts are absent in the ΔsigB mutant, however, only the first promoter generating the 1.4-kb product contains a well-defined SigB box (shown), whereas the putative SigB box of the second promoter slightly differs in sequence (not shown). (D) anti2092 is a 450 nt long asRNA transcribed opposite to the lmo2092 encoding BetL, an osmolyte transporter mediating bile tolerance in L. monocytogenes. Although a BetL homolog is present in L. innocua (lin2197) no antisense transcript was detected in L. innocua, possibly due to differences between the presumed promoter regions as shown.

Figure 5

Excludon lasRNA in L. monocytogenes and L. innocua. Long antisense RNA (yellow box-arrows) encoded in the opposite strand of L. monocytogenes (Lmo) and L. innocua (Lin) genes (blue box-arrows) are defined by green arrows corresponding to TSSs. The size of the transcripts (dashed blue arrow) was estimated from northern blots. (A) anti0605 is transcribed opposite to lmo0605 in L. monocytogenes and lin0614 in L. innocua. (B) Northern blots were carried out with probes specific to L. monocytogenes (shown) or L. innocua (not shown). As probes cross-hybridized with RNA from both species, and showed identical hybridization patterns, only the blot with L. monocytogenes-specific probe is shown. Transcription appeared to be dependent on SigB in L. monocytogenes as indicated by the failure to detect bands in the ΔsigB mutant. This SigB-dependent regulation is likely conserved in L. innocua due to the conservation of the SigB box (σ-B) within the promoter region (as shown). Several transcripts were detected with the longest being ∼5.8 kb and likely corresponding to the antisense RNA plus the polycistronic mRNA of lmo0606-07-08 operon. lmo0605 codes for MatE (NorM), an Na⁺-driven multidrug efflux pump whereas the downstream operon codes for a predicted transcriptional regulator (lmo0606) and an ABC-type multidrug transport system (lmo0607, lmo0608). (C) Fold-change expression of the antisense transcript, lmo0605 and lmo0606-07-08 operon in the WT and ΔsigB mutant was measured by strand-specific qRT–PCR analysis. In the ΔsigB mutant where the lasRNA is not expressed, expression of lmo0605 is upregulated whereas lmo0606-07-08 is downregulated as shown.

Figure 6

Excludon lasRNAs are a common phenomenon in the L. monocytogenes and L. innocua genomes. LasRNAs (yellow box-arrows) encoded opposite to L. monocytogenes (Lmo) and L. innocua (Lin) genes (blue box-arrows) are defined by green arrows corresponding to TSSs. Transcript sizes (dashed blue arrow) were estimated from the northern blots. Northern blots were carried out with probes specific to L. monocytogenes or L. innocua as indicated, however where probes cross-hybridized with RNA from both species, and showed identical patterns of bands, only a blot with a single probe is shown. (A) anti1846 is a lasRNA transcribed as an antisense to lmo1846, conserved and expressed in both species. Northern blot shows that anti1846 is transcribed as 3800, nt long and two shorter, 3000 and 2400, nt transcripts. Its transcription originates on the opposite strand to the coding sequence of lmo1846 encoding efflux pump from the MatE family and the distal part of the transcript contains the coding sequence of the downstream lmo1845 xanthine-uracil permease and 1843-44 encoding lipoprotein signal peptidase and ribosomal subunit synthase. (B) anti0424 is present in eight different lengths ranging from 500 to 6500 nt. Its transcription starts opposite to the coding sequence of lmo0424, a glucose permease, and continues to the adjacent operon encoding genes involved in fructose metabolism. The first gene of the operon, Imo0425 codes for a fructose-sensor/transcriptional antiterminator that regulates genes involved in fructose metabolism. The following four genes lmo0426-27-28, constitute the PTS lla, llb, llc components of a fructose permease and lmo0429, glycosyl hydrolase. Interestingly, both transcripts were upregulated during exponential phase growth in L. monocytogenes whereas they showed upregulation during stationary phase growth in L. innocua.