Non-canonical protein-DNA interactions identified by ChIP are not artifacts
© Bonocora et al.; licensee BioMed Central Ltd. 2013
Received: 24 April 2012
Accepted: 1 April 2013
Published: 15 April 2013
ChIP-chip and ChIP-seq are widely used methods to map protein-DNA interactions on a genomic scale in vivo. Waldminghaus and Skarstad recently reported, in this journal, a modified method for ChIP-chip. Based on a comparison of our previously-published ChIP-chip data for Escherichia coli σ32 with their own data, Waldminghaus and Skarstad concluded that many of the σ32 targets identified in our earlier work are false positives. In particular, we identified many non-canonical σ32 targets that are located inside genes or are associated with genes that show no detectable regulation by σ32. Waldminghaus and Skarstad propose that such non-canonical sites are artifacts, identified due to flaws in the standard ChIP methodology. Waldminghaus and Skarstad suggest specific changes to the standard ChIP procedure that reportedly eliminate the claimed artifacts.
We reanalyzed our published ChIP-chip datasets for σ32 and the datasets generated by Waldminghaus and Skarstad to assess data quality and reproducibility. We also performed targeted ChIP/qPCR for σ32 and an unrelated transcription factor, AraC, using the standard ChIP method and the modified ChIP method proposed by Waldminghaus and Skarstad. Furthermore, we determined the association of core RNA polymerase with disputed σ32 promoters, with and without overexpression of σ32. We show that (i) our published σ32 ChIP-chip datasets have a consistently higher dynamic range than those of Waldminghaus and Skarstad, (ii) our published σ32 ChIP-chip datasets are highly reproducible, whereas those of Waldminghaus and Skarstad are not, (iii) non-canonical σ32 target regions are enriched in a σ32 ChIP in a heat shock-dependent manner, regardless of the ChIP method used, (iv) association of core RNA polymerase with some disputed σ32 target genes is induced by overexpression of σ32, (v) σ32 targets disputed by Waldminghaus and Skarstad are predominantly those that are most weakly bound, and (vi) the modifications to the ChIP method proposed by Waldminghaus and Skarstad reduce enrichment of all protein-bound genomic regions.
The modifications to the ChIP-chip method suggested by Waldminghaus and Skarstad reduce rather than increase the quality of ChIP data. Hence, the non-canonical σ32 targets identified in our previous study are likely to be genuine. We propose that the failure of Waldminghaus and Skarstad to identify many of these σ32 targets is due predominantly to the lower data quality in their study. We conclude that surprising ChIP-chip results are not artifacts to be ignored, but rather indications that our understanding of DNA-binding proteins is incomplete.
KeywordsChIP-chip ChIP-seq σ32
ChIP-chip (sometimes referred to as ChIP-on-chip) and ChIP-seq are widely-used genomic methods that combine chromatin immunoprecipitation (ChIP) with microarrays and deep sequencing, respectively, to map protein-DNA interactions in vivo. The genome-wide binding profiles of hundreds of proteins have been mapped using ChIP-chip and ChIP-seq in organisms ranging from bacteria to humans. ChIP-chip/ChIP-seq often identifies non-canonical target regions for DNA-associated proteins, i.e. target regions that are inconsistent with our current understanding of the protein being studied. In many cases, these discoveries have provided new insight into the function of those proteins. In bacteria, many transcription factor (TF) binding sites identified using ChIP-chip/ChIP-seq are located in “unexpected” genomic regions: (i) upstream of genes whose described function is seemingly unconnected to the described function of the TF [2–4], (ii) upstream of genes whose expression does not change detectably when the TF-encoding gene is mutated [2, 4–8], (iii) inside genes [2–4, 9–13], and (iv) far from any DNA sequences that are close matches to the known consensus binding site [2, 3, 8, 14, 15]. In most cases, the significance of these observations is unclear, although they suggest that (i) gene annotations are often incomplete, (ii) TFs often function redundantly, such that expression of the regulated gene does not change unless multiple TF-encoding genes are deleted, (iii) TFs often regulate the expression of non-coding RNAs that initiate within genes , and (iv) TFs often bind DNA cooperatively such that the DNA sequence requirements are altered or relaxed.
Our published ChIP-chip study of σ32, an alternative σ factor in E. coli, led to the identification of 22 putative σ32 binding sites within genes . These represent ~25% of all the σ32 binding sites we identified. All but 2 of the gene-internal promoters are >300 bp from an annotated translation start codon. We proposed that RNA polymerase (RNAP) associated with σ32 (RNAP:σ32) often binds to promoter elements within genes and initiates transcription of non-coding RNAs in either the sense or antisense orientation. We confirmed this for three examples that we examined in more detail. Furthermore, five of the σ32 binding sites within genes are immediately adjacent to genes identified in previous studies as being upregulated by σ32, but for which no promoter could be identified in the upstream region [17, 18]. Our ChIP-chip data also permitted identification of 65 σ32 binding sites in intergenic regions, 26 of which are not associated with genes identified in either of two transcriptomic studies of σ32[17, 18]. Thus, many of the sites of σ32 association we identified are non-canonical.
In a recent study published in this journal, Waldminghaus and Skarstad describe modifications to the standard ChIP-chip procedure . The key modifications are avoiding the use of Spin-X filter columns during immunoprecipitation (IP) wash steps, including an RNase treatment following the IP, and collecting reference material after the IP rather the traditional “input” starting chromatin. Waldminghaus and Skarstad propose that the standard ChIP-chip method results in identification of false positives that are eliminated when using the modified method. Waldminghaus and Skarstad demonstrated their modified ChIP-chip procedure by performing ChIP-chip of E. coli σ32. They identified many fewer target regions for σ32 than our earlier study. We will refer to the 46 σ32 target regions identified in our previous study but not by Waldminghaus and Skarstad as “Disputed σ32 targets” (DSTs). DSTs are enriched for non-canonical σ32 binding sites. Specifically, 16 of the 46 DSTs are located inside genes or between convergently transcribed genes, and 21 DSTs are located in intergenic regions but are not associated with genes identified in transcriptomic studies of σ32[17, 18]. We have reanalyzed our published ChIP-chip datasets and those of Waldminghaus and Skarstad. This reanalysis demonstrates low reproducibility in the datasets of Waldminghaus and Skarstad. We also used targeted ChIP/qPCR to directly compare the standard and modified ChIP methods. We demonstrate that non-canonical targets of σ32 are real and that the lower data quality and deficiencies in the modified ChIP method are sufficient to explain the absence of DSTs in the list of σ32 targets generated by Waldminghaus and Skarstad.
Results and discussion
Existing evidence that DSTs are genuine sites of σ32association
Nine of the DSTs (mfd, phoP, ldhA, recF, narP, holC, glnS, ileS, and yfjN) are σ32 targets identified in independent studies that did not involve ChIP [17, 18]. With the exception of the DSTs inside yfjN and recF, these would all be considered canonical σ32 binding sites, i.e. located in an intergenic region upstream of a gene known from previous studies to be transcribed by σ32 [17, 18].
Our previous study included validation of three non-canonical DSTs (between tdk and ychG, within dhaM, and within ydeP) using ChIP/qPCR . This method does not involve amplification of ChIP DNA using Klenow DNA polymerase. Furthermore, we demonstrated heat shock-dependent increases of σ32 association with all three regions .
Although many DSTs are located inside genes, there are significantly more DSTs located in intergenic regions than expected by chance (Binomial Test p = 0.00033).
Note that, for all the analyses described herein, we have excluded the two DSTs that are located in repetitive sequence (yibA and yrdA; see Conclusions).
Comparison of data quality between our data and those of Waldminghaus and Skarstad
The two replicate datasets for our study correlate very well (Spearman Correlation Coefficient of 0.93) whereas those of Waldminghaus and Skarstad correlate less well (Spearman Correlation Coefficient of 0.64).
One of the datasets of Waldminghaus and Skarstad has a substantially lower dynamic range than the other. Several of the targets identified in both studies have z-scores within the noise for this replicate.
Although the dynamic range of one Waldminghaus and Skarstad dataset is high, the vast majority (~98.5%) of the probes have z-scores lower than 3, suggesting that these datasets are effective at identifying strong protein-DNA interactions but not weaker interactions.
Although they were not called as targets, DSTs have significantly higher z-scores for the datasets of Waldminghaus and Skarstad than expected by chance (Mann Whitney U Test p < 1e-30 for each replicate dataset).
We conclude that our ChIP-chip data are of substantially higher quality with respect to both dynamic range and reproducibility. Figure 1C-H shows normalized ChIP-chip data for replicate datasets from both studies for six selected genomic regions. These data further demonstrate the differences in reproducibility and dynamic range between the two studies. The genomic regions shown include DSTs and non-canonical targets (inside genes and/or no detectable regulation in transcriptomic studies).
Several factors likely contribute to the difference in data quality between the two studies. First, we used a TAP-tagged derivative of σ32 whereas Waldminghaus and Skarstad used an antibody raised against the native protein. Second, our heat shock conditions (50°C for 10 minutes) were different to those of Waldminghaus and Skarstad (43°C for 5 minutes). Third, as described below, the modifications to the ChIP method reduce the sensitivity of the assay.
ChIP/qPCR validation of DSTs
We compared the standard ChIP method with the modified method proposed by Waldminghaus and Skarstad. Importantly, ChIP with the modified method used the same sonicated, cross-linked cell extracts as the standard method. Using the modified method, we detected significant σ32 association with the region upstream of dnaK (Figure 2B), and association increased significantly following heat shock (Figure 2B). However, the absolute ChIP signal was substantially lower than that observed using the standard ChIP method (Figure 2A). Thus, the modified ChIP method has a decreased sensitivity relative to the standard method. Using the modified ChIP method we detected significant association of σ32 following heat shock with three of the four DSTs tested (Figure 2B). We also observed a significant reduction in σ32 association in the absence of heat shock at two of these DSTs (Figure 2B). Thus, even with the decreased sensitivity of the modified ChIP method, three of the four DSTs tested were validated as genuine sites of σ32 association. We believe that we were unable to detect significant association of σ32 with the fourth DST, ybjX, due to the substantial decrease in sensitivity relative to the standard ChIP method. We note that the ChIP signal for ybjX was the lowest of all the regions tested using the standard method (Figure 2A). We conclude that the reduced sensitivity of the modified ChIP method prevented Waldminghaus and Skarstad from identifying DSTs as sites of σ32 association. This is consistent with the observation that DSTs have above average ChIP-chip scores in the Waldminghaus and Skarstad datasets (Figure 1B).
ChIP method comparison for AraC
We conclude that Waldminghaus and Skarstad failed to identify DSTs not because of an improvement in the ChIP methodology, but because of lower data quality. Consistent with this, the majority of DSTs showed relatively low association of σ32 in our study: when ranked by the level of σ32 association, 36 of the bottom 43 targets are DSTs (Figure 1A) . Furthermore, DSTs have significantly higher signal in the Waldminghaus and Skarstad datasets than expected by chance (p < 1e-30; Figure 1B), consistent with the idea that these regions represent true binding sites for σ32 but fall below the detection threshold of this analysis. We note that Waldminghaus and Skarstad did not present any σ32 ChIP data generated using the standard methodology, precluding direct comparison of our work, nor did they use ChIP/qPCR with their modified method to measure association of σ32 with specific target regions . Furthermore, Waldminghaus and Skarstad demonstrated a dramatic improvement in ChIP-chip data for SeqA using the modified ChIP method , but their data is very similar to that generated using the standard ChIP method by another group .
Our comparison of ChIP-chip datasets highlights the importance of data quality for correct identification of protein-DNA interactions. Guidelines for ChIP-chip and ChIP-seq experimental and analytical approaches have been described previously [26, 27]. Key components of these methods that are especially relevant to our own study are the comparison of replicates, the choice of control, and the importance of repetitive sequence. Current guidelines for ChIP-seq recommend the use of only two independent biological replicates , but also stress the importance of reproducibility. As shown in Figure 1B, the poor reproducibility of the Waldminghaus and Skarstad datasets is likely to be a major cause of their failure to identify DSTs as regions truly bound by σ32. Recommended controls are either input DNA or ChIP-enriched DNA from an untagged strain (when using an epitope-tagged protein). Waldminghaus and Skarstad instead used DNA left in the supernatant after the initial IP, acknowledging that this DNA would be de-enriched for target regions. While this may increase the apparent signal, we caution against this approach as the ChIP-chip or ChIP-seq signals may not accurately reflect the actual level of binding. Finally, Waldminghaus and Skarstad highlighted the importance of treating repetitive DNA sequences with caution when interpreting ChIP-chip (or ChIP-seq) datasets. In the case of σ32, two of the ChIP peaks identified in our earlier study overlap repetitive regions. It is impossible to determine from ChIP-chip data alone whether σ32 associates with one or all of the repetitive regions. Since this caveat applies to repetitive sequences in any ChIP-chip or ChIP-seq experiment, we echo the sentiment expressed by Waldminghaus and Skarstad and caution against analysis of sequences in these regions.
Many ChIP-chip studies have revealed the existence of unexpected protein-DNA interactions. For example, ChIP-chip studies in bacteria have demonstrated that transcription factors often bind to sites within genes, sites without a recognizable motif, and sites that are not associated with described regulation by the transcription factor . This is one of the great strengths of ChIP-chip and ChIP-seq, since these non-canonical binding sites often cannot be identified using other genomic approaches such as transcription profiling. In the case of σ32, our data provide strong evidence that RNAP:σ32 initiates transcription of many RNAs from within genes, and our original study described three such examples in greater detail . The function of intragenic transcripts in bacteria is poorly understood, although several antisense transcripts have been shown previously to regulate expression of the overlapping mRNA . Our own studies have revealed pervasive antisense transcription in E. coli, and this has since been observed in several other bacterial species . Intriguingly, many ChIP-chip studies of bacterial DNA-binding TFs have revealed sites of association inside genes [10, 15], suggesting regulation of intragenic transcripts. Similar phenomena have been observed in eukaryotes, including human cells [29, 30]. Other types of non-canonical transcription factor binding sites, i.e. sites without a recognizable motif and sites that are not associated with described regulation by the transcription factor, are also poorly understood. However, sites without a recognizable motif could be explained by indirect association with DNA (detectable using ChIP) [15, 31] or cooperative interactions with other DNA-binding proteins . Sites that are not associated with described regulation by the transcription factor could be explained by combinatorial regulation by multiple, redundant transcription factors. In the case of σ32, our data suggest that many σ32 promoters are not associated with detectable regulation using transcriptomic approaches due to a high basal level of transcription, or a specific requirement for heat shock conditions.
It is important to note that Waldminghaus and Skarstad identified many non-canonical σ32-target regions in their study. Specifically, Waldminghaus and Skarstad detected σ32 association upstream of four genes whose expression was not detectably upregulated by overexpression of σ32 in either of two transcriptomic studies (yafU, rpsL, yjhI, and fimB) [17, 18], and six sites of σ32 association within genes or between convergently transcribed genes (yfbM/yfbN, yfjU, ypjA, sbcD, cycA, and macB) . Waldminghaus and Skarstad suggest that “surprising”, non-canonical protein-DNA interactions are often artifacts. We caution against this dogmatic approach. Artifacts can arise from ChIP-chip and ChIP-seq experiments; however, with the appropriate experimental and analytical methods, and with the appropriate controls, it is possible to identify protein-DNA interactions with high confidence. Atypical binding sites identified using these methods may indicate novel functions for well-studied proteins. These binding sites should not be dismissed, but rather should be the focus of additional studies.
Strains and plasmids
E. coli MG1655 rpoH-NFLAG containing the rpoH gene at its native chromosomal location fused to three FLAG tags was constructed using FRUIT . Primer sequences are available on request. Construction of MG1655 with C-terminally FLAG-tagged AraC (AMD187) will be described elsewhere (Stringer, A.M., Currenti, S.A., Bonocora, R.P., Baranowski, C., Petrone, B.L., Singh, N., Palumbo, M.J., Reilly, A.E., Zhang, Z., Erill, I. and Wade, J.T.: Comprehensive genomic analysis of the Escherichia coli and Salmonella enterica AraC regulons; in preparation).
List of oligonucleotides used in this work
For heat shock ChIP experiments, 100 ml LB was inoculated with 1 ml of fresh overnight culture of MG1655 rpoH-NFLAG and cells were grown at 30°C at 225 rpm to an OD600 of 0.5-0.6. Cultures were split (40 ml each) for further incubation at either 30°C or 50°C for 10 minutes. For ChIP experiments involving overexpression of σ32, 40 ml LB supplemented with 30 μg/ml chloramphenicol was inoculated with 0.4 ml of a fresh overnight culture of MG1655 containing either pRB1 or pBAD18-Cm. Cells were grown at 37°C at 225 rpm to an OD600 of 0.7-0.8. Expression of rpoH from pRB1 was induced by the addition of 0.2% arabinose and further incubation at 37°C for 10 minutes. For ChIP of AraC, AMD187 was grown in LB at 37°C at 225 rpm to an OD600 of 0.6-0.8.
Standard ChIP method
Cells were crosslinked by the addition of formaldehyde to a final concentration of 1% for 20 minutes. Formaldehyde was quenched with glycine (0.5 M final concentration) and cultures were pelleted by centrifugation. Pellets were washed twice with Tris-buffered saline (TBS; pH 7.5) and resuspended in 1 ml FA lysis buffer (50 mM Hepes-KOH, pH 7, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS) supplemented with 4 mg/ml lysozyme. After a 30 minute incubation at 37°C, cells were chilled on ice and sonicated in 30 second on/off pulses for 30 minutes at 100% output using a BioRuptor Sonicator. Lysates were centrifuged for five minutes to pellet cell debris. The supernatant was transferred to a new tube, brought up to a final volume of approximately 2 ml, and frozen in 0.5 ml aliquots. 0.5 ml crosslinked, sonicated cell lysate was brought up to a final volume of 0.8 ml with FA lysis buffer. A 20 μl aliquot was removed for “input” DNA control sample. 25 μl of protein A-Sepharose beads (50% slurry in TBS) and either 1 μl anti-RNA polymerase beta subunit (Neoclone) or 2 μl anti-FLAG (M2 monoclonal; Sigma) was added to the lysate and incubated for 90 minutes at room temperature with gentle rotation. Beads were pelleted at 4000 rpm in a microcentrifuge for one minute and the supernatant was removed. Beads were resuspended in 700 μl FA lysis buffer, transferred to a Spin-X column (Corning) and washed for three minutes by rotation, centrifuged for 1 minute at 4,000 rpm in a microcentrifuge and the flow through discarded. The beads were washed in a similar fashion with 750 μl of each of the following: FA lysis buffer, FA lysis buffer 500 mM NaCl, ChIP wash buffer (10 mM Tris–HCl, pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5% Nonidet-P40, 0.5% sodium deoxycholate) and TE (10 mM Tris–HCl, pH 8.0, 1 mM EDTA). The Spin-X column was transferred to a fresh tube and the chromatin was eluted from the beads by addition of 100 μl ChIP elution buffer (50 mM Tris–HCl, pH 7.5, 10 mM EDTA, 1% SDS) and incubation at 65°C for 10 minutes. The eluate was collected by centrifugation for 1 min at 4,000 rpm in a microcentrifuge. Crosslinks were reversed for both the eluate and the input samples by incubation for 10 minutes at 100°C. DNA was purified using QIAgen PCR purification kit followed by elution in either 50 μl or 200 μl for the IP samples or 200 μl for the input samples. For AraC ChIP, Spin-X columns were omitted from this procedure when indicated in the figure. Note that data shown for AraC ChIP/qPCR with Spin-X columns will be presented elsewhere (Stringer, A.M., Currenti, S.A., Bonocora, R.P., Baranowski, C., Petrone, B.L., Singh, N., Palumbo, M.J., Reilly, A.E., Zhang, Z., Erill, I. and Wade, J.T.: Comprehensive genomic analysis of the Escherichia coli and Salmonella enterica AraC regulons; in preparation).
Modified ChIP method described by Waldminghaus and Skarstad
ChIP was performed as above but with the following modifications: (i) 100 μl of post-immunoprecipitation supernatant was substituted for the “input” control DNA sample, (ii) no Spin-X columns were used, (iii) 1 μl RNase A (30 mg/ml) was added after elution and incubated for 2 hours at 42°C for both the input and immunoprecipitated DNA samples, (iv) 80 μl TE and 20 μl proteinase K (20 mg/ml) was added incubated for 2 hours at 42°C, (v) crosslinks were reversed by incubation overnight at 65°C, and (vi) DNA was purified by phenol/chloroform/isoamyl alcohol and chloroform/isoamyl alcohol extraction followed by ethanol precipitation. Note that aliquots from the same sonicated, crosslinked cell extract were used for both the standard and modified ChIP methods.
ChIP and input samples were analyzed by quantitative real time PCR using an ABI 7500 Fast real time PCR machine, as described previously . Enrichment of ChIP samples was calculated relative to a control region within the transcriptionally silent bglB gene, and normalized to input DNA. Occupancy units represent background-subtractedfold-enrichment. Oligonucleotides used for real time PCR were JW125/JW126 (bglB), JW1610/JW1611 (dnaK), JW1612/JW1613 (ygcI), JW1614/JW1615 (ybjX), JW1616/JW1617 (tdk-ychG), JW1622/JW1623 (b2084), JW071/JW072 (araB), JW073/JW074 (araE), JW075/JW076 (araF), JW389/JW390 (ytfQ), JW1312/JW1313 (dcp), and JW393/JW394 (ydeN; Table 1). Note that primers for ytfQ produced primer dimers in qPCR for ChIP with an untagged strain (Additional file 1: Supplementary Data), so we were not able to assess enrichment of this region.
Estimating null distributions for ChIP-chip datasets to calculate z-scores
Previous studies have analyzed ChIP-chip datasets based on the assumption that the distribution of actual ChIP-chip signals below the modal value closely matches the null distribution, and fits a normal distribution [20, 21]. We determined the modal value for each ChIP-chip dataset and used all probes scoring below the mode to estimate the standard deviation of a null distribution, treating the mode as the mean. We used these mean and standard deviation estimates to calculate z-scores (i.e. number of standard deviations from the mean) for each probe.
Assessment of the number of DSTs in intergenic regions
88% of the E. coli genome is genic. Of the 46 DSTs, 15 have peak probe coordinates that fall in intergenic regions. Note that some additional DSTs were classified as being “intergenic” due to the stringent criterion used in our earlier work  to account for incomplete probe coverage on the microarray. We used a Binomial Test to determine the probability that 15 of 46 DSTs would be located in intergenic regions if their genomic position was unbiased with respect to genes.
Comparison of DST z-scores to those of all z-scores for waldminghaus and skarstad datasets
For each replicate dataset, we determined the z-score for each DST peak probe. We then determined z-scores for 1,000 randomly-selected probes from the complete dataset. We used a Mann–Whitney U Test to determine the probability that the z-scores for DST peak probes are not larger than those of randomly-selected probes.
Disputed σ32 Targets
We thank Todd Gray and David Grainger for comments on the manuscript. We thank Todd Gray, David Grainger, Stephen Busby, Kevin Struhl and Evgeny Nudler for helpful discussions. This work was supported by National Institutes of Health (NIH) Grant 1DP2OD007188. DMF was supported by NIH training grant T32AI055429.
- Park PJ: ChIP-seq: advantages and challenges of a maturing technology. Nat Rev Genet. 2009, 10: 669-680.PubMed CentralView ArticlePubMedGoogle Scholar
- Wade JT, Reppas NB, Church GM, Struhl K: Genomic analysis of LexA binding reveals the permissive nature of the Escherichia coli genome and identifies unconventional target sites. Genes Dev. 2005, 19: 2619-2630. 10.1101/gad.1355605.PubMed CentralView ArticlePubMedGoogle Scholar
- Partridge JD, Bodenmiller DM, Humphrys MS, Spiro S: NsrR targets in the Escherichia coli genome: new insights into DNA sequence requirements for binding and a role for NsrR in the regulation of motility. Mol Microbiol. 2009, 73: 680-694. 10.1111/j.1365-2958.2009.06799.x.View ArticlePubMedGoogle Scholar
- Eichenberger P, Fujita M, Jensen ST, Conlon EM, Rudner DZ, Wang ST, Ferguson C, Haga K, Sato T, Liu JS: The program of gene transcription for a single differentiating cell type during sporulation in bacillus subtilis. PLoS Biol. 2004, 2: e328-10.1371/journal.pbio.0020328.PubMed CentralView ArticlePubMedGoogle Scholar
- Danielli A, Roncarati D, Delany I, Chiarini V, Rappuoli R, Scarlato V: In vivo dissection of the helicobacter pylori Fur regulatory circuit by genome-wide location analysis. J Bacteriol. 2006, 188: 4654-4662. 10.1128/JB.00120-06.PubMed CentralView ArticlePubMedGoogle Scholar
- Molle V, Nakaura Y, Shivers RP, Yamaguchi H, Losick R, Fujita Y, Sonenshein AL: Additional targets of the bacillus subtilis global regulator CodY identified by chromatin immunoprecipitation and genome-wide transcript analysis. J Bacteriol. 2003, 185 (6): 1911-1922. 10.1128/JB.185.6.1911-1922.2003.PubMed CentralView ArticlePubMedGoogle Scholar
- Laub MT, Chen SL, Shapiro L, McAdams HH: Genes directly controlled by CtrA, a master regulator of the caulobacter cell cycle. Proc Natl Acad Sci USA. 2002, 99 (7): 4632-4637. 10.1073/pnas.062065699.PubMed CentralView ArticlePubMedGoogle Scholar
- Grainger DC, Aiba H, Hurd D, Browning DF, Busby SJ: Transcription factor distribution in Escherichia coli: studies with FNR protein. Nucleic Acids Res. 2007, 35: 269-278.PubMed CentralView ArticlePubMedGoogle Scholar
- Grainger DC, Hurd D, Harrison M, Holdstock J, Busby SJ: Studies of the distribution of Escherichia coli cAMP-receptor protein and RNA polymerase along the E. Coli chromosome. Proc Natl Acad Sci USA. 2005, 102: 17693-17698. 10.1073/pnas.0506687102.PubMed CentralView ArticlePubMedGoogle Scholar
- Shimada T, Ishihama A, Busby SJ, Grainger DC: The Escherichia coli RutR transcription factor binds at targets within genes as well as intergenic regions. Nucleic Acids Res. 2008, 36: 3950-3955. 10.1093/nar/gkn339.PubMed CentralView ArticlePubMedGoogle Scholar
- Wade JT, Roa DC, Grainger DC, Hurd D, Busby SJW, Struhl K, Nudler E: Extensive functional overlap between σ factors in Escherichia coli. Nat Struct Mol Biol. 2006, 13: 806-814. 10.1038/nsmb1130.View ArticlePubMedGoogle Scholar
- Reppas NB, Wade JT, Church G, Struhl K: The transition between transcriptional initiation and elongation in E. coli is highly variable and often rate-limiting. Mol Cell. 2006, 24: 747-757. 10.1016/j.molcel.2006.10.030.View ArticlePubMedGoogle Scholar
- Tomljenovic-Berube AM, Mulder DT, Whiteside MD, Brinkman FS, Coombes BK: Identification of the regulatory logic controlling Salmonella pathoadaptation by the SsrA-SsrB Two-component system. PLoS Genet. 2010, 6: e1000875-10.1371/journal.pgen.1000875.PubMed CentralView ArticlePubMedGoogle Scholar
- Molle V, Fujita M, Jensen ST, Eichenberger P, Gonzalez-Pastor JE, Liu JS, Losick R: The Spo0A regulon of Bacillus subtilis. Mol Microbiol. 2003, 50 (5): 1683-1701. 10.1046/j.1365-2958.2003.03818.x.View ArticlePubMedGoogle Scholar
- Wade JT, Struhl K, Busby SJ, Grainger DC: Genomic analysis of protein-DNA interactions in bacteria: insights into transcription and chromosome organization. Mol Microbiol. 2007, 65: 21-26. 10.1111/j.1365-2958.2007.05781.x.View ArticlePubMedGoogle Scholar
- Dornenburg JE, DeVita AM, Palumbo MJ, Wade JT: Widespread antisense transcription in Escherichia coli. mBio. 2010, 1: e00024-00010.PubMed CentralView ArticlePubMedGoogle Scholar
- Nonaka G, Blankschien M, Herman C, Gross CA, Rhodius VA: Regulon and promoter analysis of the E. Coli heat shock factor, sigma 32, reveals a multifaceted cellular response to heat stress. Genes Dev. 2006, 20: 1776-1789. 10.1101/gad.1428206.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhao K, Liu M, Burgess RR: The global transcriptional response of Escherichia coli to induced sigma 32 protein involves sigma 32 regulon activation followed by inactivation and degradation of sigma 32 in vivo. J Biol Chem. 2005, 280: 17758-17768.View ArticlePubMedGoogle Scholar
- Waldminghaus T, Skarstad K: ChIP on chip: surprising results are often artifacts. BMC Genomics. 2010, 11: 414-10.1186/1471-2164-11-414.PubMed CentralView ArticlePubMedGoogle Scholar
- Li XY, MacArthur S, Bourgon R, Nix D, Pollard DA, Iyer VN, Hechmer A, Simirenko L, Stapleton M, Luengo Hendriks CL: Transcription factors bind thousands of active and inactive regions in the drosophila blastoderm. PLoS Biol. 2008, 6: e27-10.1371/journal.pbio.0060027.PubMed CentralView ArticlePubMedGoogle Scholar
- Gibbons FD, Proft M, Struhl K, Roth RP: Chipper: discovering transcription factor targets from chromatin immunoprecipitation microarrays using variance-stabilization. Genome Biol. 2005, 6: R96-10.1186/gb-2005-6-11-r96.PubMed CentralView ArticlePubMedGoogle Scholar
- Lefrancois P, Euskirchen GM, Auerbach RK, Rozowsky J, Gibson T, Yellman CM, Gerstein M, Snyder M: Efficient yeast ChIP-Seq using multiplex short-read DNA sequencing. BMC Genomics. 2009, 10: 37-10.1186/1471-2164-10-37.PubMed CentralView ArticlePubMedGoogle Scholar
- Vega VB, Cheung E, Palanisamy N, Sung WK: Inherent signals in sequencing-based chromatin-ImmunoPrecipitation control libraries. PLoS One. 2009, 4: e5241-10.1371/journal.pone.0005241.PubMed CentralView ArticlePubMedGoogle Scholar
- Auerbach RK, Euskirchen G, Rozowsky J, Lamarre-Vincent N, Moqtaderi Z, Lefrançois P, Struhl K, Gerstein M, Snyder M: Mapping accessible chromatin regions using sono-Seq. Proc Natl Acad Sci USA. 2009, 106: 14926-14931. 10.1073/pnas.0905443106.PubMed CentralView ArticlePubMedGoogle Scholar
- Sánchez-Romero MA, Busby SJ, Dyer NP, Ott S, Millard AD, Grainger DC: Dynamic distribution of SeqA protein across the chromosome of Escherichia coli K-12. mBio. 2010, 1: e00012-00010-View ArticleGoogle Scholar
- Buck MJ, Lieb JD: ChIP-chip: considerations for the design, analysis, and application of genome-wide chromatin immunoprecipitation experiments. Genomics. 2004, 83 (3): 349-360. 10.1016/j.ygeno.2003.11.004.View ArticlePubMedGoogle Scholar
- Landt SG, Marinov GK, Kundaje A, Kheradpour P, Pauli F, Batzoglou S, Bernstein BE, Bickel P, Brown JB, Cayting P: ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res. 2012, 22: 1813-1831. 10.1101/gr.136184.111.PubMed CentralView ArticlePubMedGoogle Scholar
- Georg J, Hess WR: cis-antisense RNA, another level of gene regulation in bacteria. Microbiol Mol Biol Rev. 2011, 75: 286-300. 10.1128/MMBR.00032-10.PubMed CentralView ArticlePubMedGoogle Scholar
- ENCODE pc: An integrated encyclopedia of DNA elements in the human genome. Nature. 2012, 489: 57-74. 10.1038/nature11247.View ArticleGoogle Scholar
- Cawley S, Bekiranov S, Ng HH, Kapranov P, Sekinger EA, Kampa D, Piccolboni A, Smentchenko V, Cheng J, Williams AJ: Unbiased mapping of transcription factor binding sites along human chromosomes 21 and 22 points to widespread regulation of non-coding RNAs. Cell. 2004, 116: 499-509. 10.1016/S0092-8674(04)00127-8.View ArticlePubMedGoogle Scholar
- Neph S, Vierstra J, Stergachis AB, Reynolds AP, Haugen E, Vernot B, Thurman RE, John S, Sandstrom R, Johnson A: An expansive human regulatory lexicon encoded in transcription factor footprints. Nature. 2012, 489: 83-90. 10.1038/nature11212.PubMed CentralView ArticlePubMedGoogle Scholar
- Belyaeva TA, Wade JT, Webster CL, Howard VJ, Thomas MS, Hyde EI, Busby SJ: Transcription activation at the Escherichia coli melAB promoter: the role of MelR and the cyclic AMP receptor protein. Mol Microbiol. 2000, 36: 211-222. 10.1046/j.1365-2958.2000.01849.x.View ArticlePubMedGoogle Scholar
- Stringer AM, Singh N, Yermakova A, Petrone BL, Amarasinghe JJ, Reyes-Diaz L, Mantis NJ, Wade JT: FRUIT, a scar-free system for targeted chromosomal mutagenesis, epitope tagging, and promoter replacement in Escherichia coli and salmonella enterica. PLoS One. 2012, 7: e44841-10.1371/journal.pone.0044841.PubMed CentralView ArticlePubMedGoogle Scholar
- Guzman L-M, Belin D, Carson MJ, Beckwith JR: Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol. 1995, 177: 4121-4130.PubMed CentralPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.