Limits...
Comparison of strand-specific transcriptomes of enterohemorrhagic Escherichia coli O157:H7 EDL933 (EHEC) under eleven different environmental conditions including radish sprouts and cattle feces.

Landstorfer R, Simon S, Schober S, Keim D, Scherer S, Neuhaus K - BMC Genomics (2014)

Bottom Line: Furthermore, we observed induction of the shiga-toxin carrying phages by antibiotics and confirmed active biofilm related genes on radish sprouts, in cattle feces, and on agar plates.For instance, the environmental function of azoR had been elusive, but this gene is highly active on radish sprouts.Thus, NGS-transcriptomics is an appropriate technique to propose new roles of hypothetical genes and to guide future research.

View Article: PubMed Central - PubMed

Affiliation: Lehrstuhl für Mikrobielle Ökologie, Wissenschaftszentrum Weihenstephan, Technische Universität München, Weihenstephaner Berg 3, D-85350 Freising, Germany. neuhaus@wzw.tum.de.

ABSTRACT

Background: Multiple infection sources for enterohemorrhagic Escherichia coli O157:H7 (EHEC) are known, including animal products, fruit and vegetables. The ecology of this pathogen outside its human host is largely unknown and one third of its annotated genes are still hypothetical. To identify genetic determinants expressed under a variety of environmental factors, we applied strand-specific RNA-sequencing, comparing the SOLiD and Illumina systems.

Results: Transcriptomes of EHEC were sequenced under 11 different biotic and abiotic conditions: LB medium at pH4, pH7, pH9, or at 15°C; LB with nitrite or trimethoprim-sulfamethoxazole; LB-agar surface, M9 minimal medium, spinach leaf juice, surface of living radish sprouts, and cattle feces. Of 5379 annotated genes in strain EDL933 (genome and plasmid), a surprising minority of only 144 had sequencing reads under all conditions. We therefore developed a statistical method to distinguish weakly transcribed genes from background transcription. We find that 96% of all genes and 91.5% of the hypothetical genes exhibit a significant transcriptional signal under at least one condition. Comparing SOLiD and Illumina systems, we find a high correlation between both approaches for fold-changes of the induced or repressed genes. The pathogenicity island LEE showed highest transcriptional activity in LB medium, minimal medium, and after treatment with antibiotics. Unique sets of genes, including many hypothetical genes, are highly up-regulated on radish sprouts, cattle feces, or in the presence of antibiotics. Furthermore, we observed induction of the shiga-toxin carrying phages by antibiotics and confirmed active biofilm related genes on radish sprouts, in cattle feces, and on agar plates.

Conclusions: Since only a minority of genes (2.7%) were not active under any condition tested ( reads), we suggest that the assumption of significant genome over-annotations is wrong. Environmental transcriptomics uncovered hitherto unknown gene functions and unique regulatory patterns in EHEC. For instance, the environmental function of azoR had been elusive, but this gene is highly active on radish sprouts. Thus, NGS-transcriptomics is an appropriate technique to propose new roles of hypothetical genes and to guide future research.

Show MeSH

Related in: MedlinePlus

Comparison of technical and biological replicates. A: The technical replicate was generated from spinach medium by splitting the libraries before PCR of SOLiD sequencing. The mapped counts were normalized and are given in fragments/gene according to Haas et al. (51). The correlation coefficient R2 is virtually 1.0. B: Biological replicates for LB medium are shown. They were sequenced on two different platforms, the SOLiD 4.0 system and the Illumina MiSeq sequencer, to exclude technical artifacts of one platform. The correlation coefficient R2 is 0.72.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4048457&req=5

Fig1: Comparison of technical and biological replicates. A: The technical replicate was generated from spinach medium by splitting the libraries before PCR of SOLiD sequencing. The mapped counts were normalized and are given in fragments/gene according to Haas et al. (51). The correlation coefficient R2 is virtually 1.0. B: Biological replicates for LB medium are shown. They were sequenced on two different platforms, the SOLiD 4.0 system and the Illumina MiSeq sequencer, to exclude technical artifacts of one platform. The correlation coefficient R2 is 0.72.

Mentions: In order to test the reproducibility of the sequencing process, two technical replicates of barcoded libraries of two conditions were generated, spinach medium and LB-nitrite. After cDNA-synthesis the libraries were split and treated independently and the RPKM values of each replicate were compared. The correlation coefficient R2 was analyzed as described in Haas et al.[28] in reads per gene. Since the correlation was excellent (R2 = 1.0, see Figure 1A), as had also been observed for other NGS experiments (e.g., [29]), we combined those technical replicates for further expression analysis. Next, biological reproducibility was tested by sequencing replicates of the LB reference and the radish sprout condition on two different sequencing platforms SOLiD and Illumina. Despite massive differences in library making techniques and in the sequencing strategy of both platforms, we obtained a high correlation of R2 = 0.72 (Figure 1B). This verified that the observed changes in gene regulation were not due to technical or experimental artifacts.Figure 1


Comparison of strand-specific transcriptomes of enterohemorrhagic Escherichia coli O157:H7 EDL933 (EHEC) under eleven different environmental conditions including radish sprouts and cattle feces.

Landstorfer R, Simon S, Schober S, Keim D, Scherer S, Neuhaus K - BMC Genomics (2014)

Comparison of technical and biological replicates. A: The technical replicate was generated from spinach medium by splitting the libraries before PCR of SOLiD sequencing. The mapped counts were normalized and are given in fragments/gene according to Haas et al. (51). The correlation coefficient R2 is virtually 1.0. B: Biological replicates for LB medium are shown. They were sequenced on two different platforms, the SOLiD 4.0 system and the Illumina MiSeq sequencer, to exclude technical artifacts of one platform. The correlation coefficient R2 is 0.72.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4048457&req=5

Fig1: Comparison of technical and biological replicates. A: The technical replicate was generated from spinach medium by splitting the libraries before PCR of SOLiD sequencing. The mapped counts were normalized and are given in fragments/gene according to Haas et al. (51). The correlation coefficient R2 is virtually 1.0. B: Biological replicates for LB medium are shown. They were sequenced on two different platforms, the SOLiD 4.0 system and the Illumina MiSeq sequencer, to exclude technical artifacts of one platform. The correlation coefficient R2 is 0.72.
Mentions: In order to test the reproducibility of the sequencing process, two technical replicates of barcoded libraries of two conditions were generated, spinach medium and LB-nitrite. After cDNA-synthesis the libraries were split and treated independently and the RPKM values of each replicate were compared. The correlation coefficient R2 was analyzed as described in Haas et al.[28] in reads per gene. Since the correlation was excellent (R2 = 1.0, see Figure 1A), as had also been observed for other NGS experiments (e.g., [29]), we combined those technical replicates for further expression analysis. Next, biological reproducibility was tested by sequencing replicates of the LB reference and the radish sprout condition on two different sequencing platforms SOLiD and Illumina. Despite massive differences in library making techniques and in the sequencing strategy of both platforms, we obtained a high correlation of R2 = 0.72 (Figure 1B). This verified that the observed changes in gene regulation were not due to technical or experimental artifacts.Figure 1

Bottom Line: Furthermore, we observed induction of the shiga-toxin carrying phages by antibiotics and confirmed active biofilm related genes on radish sprouts, in cattle feces, and on agar plates.For instance, the environmental function of azoR had been elusive, but this gene is highly active on radish sprouts.Thus, NGS-transcriptomics is an appropriate technique to propose new roles of hypothetical genes and to guide future research.

View Article: PubMed Central - PubMed

Affiliation: Lehrstuhl für Mikrobielle Ökologie, Wissenschaftszentrum Weihenstephan, Technische Universität München, Weihenstephaner Berg 3, D-85350 Freising, Germany. neuhaus@wzw.tum.de.

ABSTRACT

Background: Multiple infection sources for enterohemorrhagic Escherichia coli O157:H7 (EHEC) are known, including animal products, fruit and vegetables. The ecology of this pathogen outside its human host is largely unknown and one third of its annotated genes are still hypothetical. To identify genetic determinants expressed under a variety of environmental factors, we applied strand-specific RNA-sequencing, comparing the SOLiD and Illumina systems.

Results: Transcriptomes of EHEC were sequenced under 11 different biotic and abiotic conditions: LB medium at pH4, pH7, pH9, or at 15°C; LB with nitrite or trimethoprim-sulfamethoxazole; LB-agar surface, M9 minimal medium, spinach leaf juice, surface of living radish sprouts, and cattle feces. Of 5379 annotated genes in strain EDL933 (genome and plasmid), a surprising minority of only 144 had sequencing reads under all conditions. We therefore developed a statistical method to distinguish weakly transcribed genes from background transcription. We find that 96% of all genes and 91.5% of the hypothetical genes exhibit a significant transcriptional signal under at least one condition. Comparing SOLiD and Illumina systems, we find a high correlation between both approaches for fold-changes of the induced or repressed genes. The pathogenicity island LEE showed highest transcriptional activity in LB medium, minimal medium, and after treatment with antibiotics. Unique sets of genes, including many hypothetical genes, are highly up-regulated on radish sprouts, cattle feces, or in the presence of antibiotics. Furthermore, we observed induction of the shiga-toxin carrying phages by antibiotics and confirmed active biofilm related genes on radish sprouts, in cattle feces, and on agar plates.

Conclusions: Since only a minority of genes (2.7%) were not active under any condition tested ( reads), we suggest that the assumption of significant genome over-annotations is wrong. Environmental transcriptomics uncovered hitherto unknown gene functions and unique regulatory patterns in EHEC. For instance, the environmental function of azoR had been elusive, but this gene is highly active on radish sprouts. Thus, NGS-transcriptomics is an appropriate technique to propose new roles of hypothetical genes and to guide future research.

Show MeSH
Related in: MedlinePlus