Horizontally acquired AT-rich genes in Escherichia coli cause toxicity by sequestering RNA polymerase

Abstract

Horizontal gene transfer permits rapid dissemination of genetic elements between individuals in bacterial populations. Transmitted DNA sequences may encode favourable traits. However, if the acquired DNA has an atypical base composition, it can reduce host fitness. Consequently, bacteria have evolved strategies to minimize the harmful effects of foreign genes. Most notably, xenogeneic silencing proteins bind incoming DNA that has a higher AT content than the host genome. An enduring question has been why such sequences are deleterious. Here, we showed that the toxicity of AT-rich DNA in Escherichia coli frequently results from constitutive transcription initiation within the coding regions of genes. Left unchecked, this causes titration of RNA polymerase and a global downshift in host gene expression. Accordingly, a mutation in RNA polymerase that diminished the impact of AT-rich DNA on host fitness reduced transcription from constitutive, but not activator-dependent, promoters.

Figure 1. Characterisation of the yccE locus.

a) Genomic context of yccE and its promoter. The panel shows yccE, and surrounding genes, alongside data describing H-NS binding (ChIP-seq) and RNA abundance (standard RNA-seq; this work, done in duplicate). Data are representative. b) Sequence of the yccEΔ200 DNA fragment containing PyccE. The PyccE -10 and -35 elements are in bold and underlined. A consensus σ³² promoter sequence is in grey for comparison. Mutations made to disrupt sequence elements are in red. Transcription can initiate at adjacent nucleotides. These are labelled (+1) and are highlighted by a bent arrow. c) Analysis of transcripts generated from PyccE in vitro. The gel image shows transcripts generated by RNA polymerase, associated with either σ⁷⁰ or σ³², from pSR plasmid DNA templates. The schematic diagrams above the gel image represent the native cloning site of pSR (left hand side) or derivatives containing a PyccE insert (right hand side). The different PyccE containing DNA fragments inserted are indicated below the gel image in parenthesis. PyccE derived transcripts manifest as a 107/106 nucleotide (nt) doublet. The 108/107 nt RNA-I transcript is derived from the plasmid replication origin. The experiment was done three times. Data are representative. d) Effect of PyccE mutation in vivo. The graph shows LacZ activity data obtained from E. coli JCB387 cells carrying different yccEΔ200 derivatives cloned in pRW50. Assays were done in triplicate and error bars show standard deviation from the mean. e) Induction of yccE transcription in the absence of H-NS does not require PyccE in vivo. The panel illustrates a series of yccE::lacZ fusions labelled i-iv. Genes are shown as block arrows and PyccE is shown as a bent line arrow. The ß-galactosidase activity was measured in lysates of M182 (grey bars) and M182hns::kan cells (open bars). Assays were done in triplicate and error bars show standard deviation from the mean.

Figure 2. H-NS represses intragenic yccE transcription and associated fitness costs

a) Identification of intragenic yccE promoters. The data are β-galactosidase activities driven by short intragenic yccE DNA fragments in strain JCB387. The bars align with the location of the DNA fragment relative to yccE and show sense (upper) and antisense (lower) transcription. Bars labelled “a”-“k” have at least 2-fold over background activity (empty pRW50; shown by dashed line). The black bar represents the canonical promoter PyccE. Note that, two DNA fragments resisted cloning. Hence, 19 of the 21 potential promoters were tested. Assays were done in triplicate. Error bars show standard deviation from the mean. b) Transcription can initiate at multiple sites within yccE in vitro. Gel image showing RNA generated in vitro separated by denaturing gel electrophoresis. DNA templates, with the yccE gene cloned upstream of the λoop terminator in plasmid pSR, are illustrated above the gel. Transcripts generated by RNA polymerase (400 nM) with empty pSR plasmid (lane 1) are highlighted by a black dashed line. Transcripts initiating within yccE in the forward (lanes 2-5) or reverse (lanes 6-9) orientation are highlighted by a blue dashed line. The control RNA-I transcripts are highlighted by a grey dashed line. H-NS was added at concentrations of 0.8, 1.5, or 3.0 μM. The experiment, done three times, is representative. c) Mutation of promoters within yccE prevents induction in Δhns cells. The lower illustrations show yccE (blue arrows) cloned upstream of lacZ (red arrow). A solid blue arrow represents wild type yccE whereas open arrows indicate yccE with mutated intragenic promoter -10 elements. For each lacZ fusion, β-galactosidase activity was measured in lysates of M182 (grey bars) and M182hns::kan cells (open bars). Assays were done in triplicate and error bars show standard deviation from the mean. d,e) The fitness cost of yccE is reduced when intragenic promoters are mutated. The figure illustrates changes in culture OD650 following inoculation of LB medium. The inoculum was M182 (solid line) or M182hns::kan (dashed line) transformed with the pSR plasmid carrying d) wild type yccE or e) yccE with internal promoter -10 elements mutated. Cells were grown at 37°C. The experiment was done in triplicate. Error bars show standard deviation from the mean.

Figure 3. H-NS represses intragenic transcription and associated fitness costs at many loci.

a) Transcription initiation within the coding regions of H-NS target genes in vitro. An in vitro transcription assay using different AT-rich genes, cloned upstream of the λoop terminator in plasmid pSR, as a template. The different DNA constructs are illustrated above the gel. The cloned genes have an AT-content of 65% (yfdF), 63% (ykgH), 63% (yjgN) and 68% (yjgL). For each cloned gene a solid arrow represents the wild type DNA sequence whereas an open arrow is a derivative where intragenic promoter -10 elements have point mutations. Note that the fepA gene was used as a control and has an AT-content of 55%. The positions of transcripts generated by RNA polymerase (400 nM) from the fepE control (lane 1) or the other cloned genes (lanes 2-9) are labelled. The in vitro transcription assays were run on three separate occasions. Data are representative. b) Increased transcription in cells lacking H-NS frequently requires intragenic promoters in vivo. The panel illustrates a series of DNA constructs where different gene coding regions have been cloned upstream of lacZ (red arrow). For each cloned gene a solid arrow represents the wild type DNA sequence whereas an open arrow is a derivative where intragenic promoter -10 elements have point mutations. The cloned genes have an AT-content of 65% (yfdF), 63% (ykgH), 63% (yjgN) and 68% (yjgL). Note that the fepA gene is used as a control and has an AT-content of 55%. For each lacZ fusion β-galactosidase activity was measured in lysates of M182 (grey bars) and M182Δhns (open bars) cells. Assays were done in triplicate and error bars show standard deviation from the mean. c,d) The toxicity of many AT-rich genes is a consequence of spurious intragenic transcription. The figure illustrates growth of M182 (solid line) or M182hns::kan (dashed line) cells transformed with the pSR plasmid carrying different AT-rich genes. Panel c) shows wild type gene derivatives and d) shows derivatives with internal promoter -10 elements mutated (open arrows). Experiments were done using M9 minimal media at 30°C. The experiment was done in triplicate and error bars show standard deviation from the mean.

Figure 4. Most transcription is uniformly downregulated in cells lacking H-NS.

a,b) Most transcription is uniformly downregulated in cells lacking H-NS. a) the plot illustrates changes in global transcription caused by loss of H-NS. Data points represent H-NS bound (red) and unbound (black) genes. Genes with unaltered transcription should fall on the diagonal blue line. Data are from duplicate RNA-seq experiments. b) The basally expressed fad genes whose transcription is reduced in the absence of H-NS. Genes bound by H-NS are in red and other genes are black. Graphs show H-NS binding and RNA abundance (RNA-seq with spiked in control; this work). Data are representative. c) RNA polymerase is redistributed in cells lacking H-NS. RNA polymerase distribution in wild type (top) and hns::kan cells (middle). Each heat map shows the average position of DNA-bound (i.e. transcribing or interacting with a promoter) RNA polymerase molecules within the cell. The bottom panelshows the average position of Ter as determined by visualising a TetR-mYpet fusion bound at an array of Ter proximal tetO sequences. Each distribution was generated from 100 cells between 3.5 and 4.5 μm in length. Each square is 1/624 of total cell area. d) The σ⁷⁰ G424D mutation hinders constitutive but not activator dependent NM501 promoter activity in vitro. KMnO4 footprinting reactions analysed on a denaturing polyacrylamide gel. Bands are indicative of open complex near the transcription start site (+1). RNA polymerase added at concentrations of 200, 250, 300 or 350 nM and CRP was 1.0 μM. The experiment, done three times, is representative. e,f) The σ⁷⁰ G424D mutation hinders constitutive but not activator dependent promoter activity in vivo. Different promoter DNA fragments were cloned upstream of lacZ (red arrow). For each promoter the location of key DNA sequence elements is represented by a box. In each case, the box is coloured according to the relationship between the DNA sequence and the consensus sequence for that element: perfect (dark blue), imperfect (pale blue) or completely absent (white). For each lacZ fusion β-galactosidase activity was measured in lysates of JCB387rpoD::kan carrying pVRσ (white bars) or pVRσ^G424D (black bars). The experiment was done in triplicate. Error bars show standard deviation from the mean.