Genome-Wide Mapping of the Escherichia coli PhoB Regulon Reveals Many Transcriptionally Inert, Intragenic Binding Sites

Abstract

Genome-scale analyses have revealed many transcription factor binding sites within, rather than upstream of, genes, raising questions as to the function of these binding sites. Here, we use complementary approaches to map the regulon of the Escherichia coli transcription factor PhoB, a response regulator that controls transcription of genes involved in phosphate homeostasis. Strikingly, the majority of PhoB binding sites are located within genes, but these intragenic sites are not associated with detectable transcription regulation and are not evolutionarily conserved. Many intragenic PhoB sites are located in regions bound by H-NS, likely due to shared sequence preferences of PhoB and H-NS. However, these PhoB binding sites are not associated with transcription regulation even in the absence of H-NS. We propose that for many transcription factors, including PhoB, binding sites not associated with promoter sequences are transcriptionally inert and hence are tolerated as genomic "noise." IMPORTANCE Recent studies have revealed large numbers of transcription factor binding sites within the genes of bacteria. The function, if any, of the vast majority of these binding sites has not been investigated. Here, we map the binding of the transcription factor PhoB across the Escherichia coli genome, revealing that the majority of PhoB binding sites are within genes. We show that PhoB binding sites within genes are not associated with regulation of the overlapping genes. Indeed, our data suggest that bacteria tolerate the presence of large numbers of nonregulatory, intragenic binding sites for transcription factors and that these binding sites are not under selective pressure.

Keywords: ChIP-seq; PhoB; pho regulon; transcription factors.

FIG 1

Partially reduced activity of the C-terminally FLAG₃-tagged PhoB. qRT-PCR was used to measure levels of the pstS RNA relative to the minD RNA control in wild-type MG1655/pBAD24 (wild type), MG1655 ΔphoB (CDS091)/pBAD24 (ΔphoB), or MG1655 phoB-FLAG₃ (DMF34)/pBAD24 (PhoB-FLAG₃) for cells grown under low-phosphate conditions. Values are the average of three independent biological replicates; error bars represent ±1 standard deviation.

FIG 2

ChIP-seq identifies PhoB binding sites. (A) ChIP-seq data for (i) an untagged control under low-phosphate conditions, (ii) PhoB-FLAG₃ under low-phosphate conditions, and (iii) PhoB-FLAG₃ under high-phosphate conditions. Three genomic regions are shown, with one data set from two independent biological replicates. Values on the x axis represent genome positions. Values on the y axis represent normalized sequence read coverage, with positive values indicating sequence reads mapping to the forward strand and negative values indicating sequence reads mapping to the reverse strand. y axis scales differ between the three genomic regions but are matched for the three data sets for any given genomic region. (B) Significantly enriched DNA sequence motif derived from 100-bp regions surrounding each ChIP-seq peak. The number of sites contributing to the motif and the E value determined by MEME are indicated. (C) Analysis of the position of inferred PhoB binding sites relative to the position of ChIP-seq peak centers. For each of the binding sites contributing to the motif determined by MEME (see panel B), we determined the position of the binding site relative to the associated ChIP-seq peak center. The x axis indicates positions relative to ChIP-seq peak centers. The y axis indicates the number of binding sites that cover any given position. (D) Pie chart showing the genome context of PhoB binding sites identified by ChIP-seq. Sites designated “intragenic and upstream” are intragenic but <200 bp upstream of an annotated gene start.

FIG 3

Comparison of ChIP-seq and ChIP-chip data sets. (A) Significantly enriched DNA sequence motif derived from 100-bp regions surrounding each ChIP-seq peak for regions shared between the ChIP-seq data set and a published ChIP-chip data set (43). The number of sites contributing to the motif and the E value determined by MEME are indicated. (B) Significantly enriched DNA sequence motif derived from 100-bp regions surrounding each ChIP-seq peak for regions unique to the ChIP-seq data set, i.e., not found in the published ChIP-chip data set (43). The number of sites contributing to the motif and the E value determined by MEME are indicated.

FIG 4

RNA-seq analysis of E. coli wild-type and ΔphoB strains. Scatter plot showing relative RNA levels for all genes in wild-type (MG1655/pBAD24) or ΔphoB (CDS091/pBAD24) cells. Each data point corresponds to a gene. Triangular data points represent genes previously reported to be in the pho regulon, with red fill indicating that the transcript has an upstream PhoB site identified by ChIP-seq and gray fill indicating no upstream site. Red circle data points represent genes not previously reported to be in the pho regulon but with upstream PhoB sites identified by ChIP-seq. Blue circle data points represent genes with internal PhoB sites identified by ChIP-seq. All other genes are represented by gray circle data points.

FIG 5

Differences in RNAP (β subunit) occupancy in genes that are potential members of the pho regulon. RNAP (β subunit) occupancy measured by ChIP-qPCR (see Materials and Methods for details of how occupancy units are calculated) in wild-type MG1655 (dark-colored bars) and MG1655 ΔphoB (CDS091; light-colored bars) for regions within genes that are potential members of the pho regulon. Schematics on the left show genes with upstream or internal PhoB sites (red vertical lines). The horizontal bars indicate the positions of PCR amplicons used in ChIP-qPCR, with black bars indicating amplicons within genes that have upstream PhoB sites, green bars indicating amplicons upstream of intragenic PhoB sites, and blue bars indicating amplicons downstream of intragenic PhoB sites. Values are the average of three independent biological replicates; error bars represent 1 standard deviation.

FIG 6

Differences in σ⁷⁰ occupancy around PhoB binding sites between wild-type and ΔphoB cells. The scatter plot shows normalized σ⁷⁰ occupancy in wild-type MG1655 and MG1655 ΔphoB (DMF84) for the 400-bp regions surrounding PhoB binding sites identified by ChIP-seq. Each data point represents a PhoB binding site. Intergenic binding sites are indicated by red data points, and intragenic binding sites are indicated by blue data points. Genes associated with PhoB binding sites are labeled with the gene name in cases where σ⁷⁰ occupancy differs >2-fold between wild-type and ΔphoB cells. Values are the average of two independent biological replicates; error bars represent ±1 standard deviation.

FIG 7

H-NS suppresses transcription from many promoters. (A) Scatter plot showing normalized σ⁷⁰ occupancy in wild-type MG1655 and MG1655 Δhns (AMD565a) for the 400-bp regions surrounding PhoB binding sites identified by ChIP-seq. Each data point represents a PhoB binding site. The color of each data point indicates the level of H-NS occupancy at the corresponding site (49). Intragenic PhoB sites are represented by crosses; intergenic PhoB sites are represented by circles. (B) Scatter plot showing normalized σ⁷⁰ occupancy in wild-type MG1655 and MG1655 Δhns (AMD565a) for all σ⁷⁰ binding sites identified by ChIP-seq from MG1655 Δhns (AMD565a) cells. The color of each data point indicates the level of H-NS occupancy at the corresponding site (49). Intragenic PhoB sites are represented by crosses; intergenic PhoB sites are represented by circles. For both panels A and B, values are the average of two independent biological replicates; error bars represent ±1 standard deviation.

FIG 8

H-NS does not suppress PhoB-dependent effects on recruitment of initiating RNA polymerase. The scatter plot shows normalized σ⁷⁰ occupancy in wild-type MG1655 Δhns (AMD565a) and MG1655 Δhns ΔphoB (DMF85) for the 400-bp regions surrounding PhoB binding sites identified by ChIP-seq. Each data point represents a PhoB binding site. The color of each data point indicates the level of H-NS occupancy at the corresponding site (49). Intragenic PhoB sites are represented by crosses; intergenic PhoB sites are represented by circles. Genes associated with PhoB binding sites are indicated in cases where σ⁷⁰ occupancy differs >2-fold between MG1655 Δhns (AMD565a) and MG1655 Δhns ΔphoB (DMF85) cells. Values are the average of two independent biological replicates; error bars represent ±1 standard deviation.

FIG 9

Conservation of PhoB binding sites across gammaproteobacterial species. A heat map shows conservation of PhoB binding sites across selected gammaproteobacterial species. Columns represent PhoB binding sites from E. coli, divided into known pho regulon binding sites, intergenic sites, and intragenic sites. The associated genes are indicated above each column. Rows represent species for different gammaproteobacterial genera, as indicated to the left of each row. The color of each square indicates the predicted strength of the best-scoring putative PhoB binding site in a region that is homologous to the corresponding region in E. coli. Binding site strength was predicted using a position weight matrix derived from the E. coli PhoB binding site motif (Fig. 2B). The color scale is shown below the heat map, with yellow indicating stronger predicted binding site strength and blue indicating weaker predicted binding site strength. White indicates the absence of a homologous region in the indicated species.