Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Sep;645(8082):1060-1070.
doi: 10.1038/s41586-025-09396-y. Epub 2025 Aug 13.

Elementary 3D organization of active and silenced E. coli genome

Alexey A Gavrilov #  1   2 , Ilya Shamovsky #  2 , Irina Zhegalova  1   3 , Sergey Proshkin  4 , Yosef Shamovsky  2 , Grigory Evko  1 , Vitaly Epshtein  2 , Aviram Rasouly  2 , Anna Blavatnik  2 , Sudipta Lahiri  2 , Eli Rothenberg  2 , Sergey V Razin  5   6 , Evgeny Nudler  7   8
Affiliations

Elementary 3D organization of active and silenced E. coli genome

Alexey A Gavrilov et al. Nature. 2025 Sep.

Abstract

Unravelling how genomes are spatially organized and how their three-dimensional (3D) architecture drives cellular functions remains a major challenge in biology1,2. In bacteria, genomic DNA is compacted into a highly ordered, condensed state called nucleoid3-5. Despite progress in characterizing bacterial 3D genome architecture over recent decades6-8, the fine structure and functional organization of the nucleoid remain elusive due to low-resolution contact maps from methods such as Hi-C9-11. Here we developed an enhanced Micro-C chromosome conformation capture, achieving 10-base pair (bp) resolution. This ultra-high-resolution analysis reveals elemental spatial structures in the Escherichia coli nucleoid, including chromosomal hairpins (CHINs) and chromosomal hairpin domains (CHIDs). These structures, organized by histone-like proteins H-NS and StpA, have key roles in repressing horizontally transferred genes. Disruption of H-NS causes drastic reorganization of the 3D genome, decreasing CHINs and CHIDs, whereas removing both H-NS and StpA results in their complete disassembly, increased transcription of horizontally transferred genes and delayed growth. Similar effects are observed with netropsin, which competes with H-NS and StpA for AT-rich DNA binding. Interactions between CHINs further organize the genome into isolated loops, potentially insulating active operons. Our Micro-C analysis reveals that all actively transcribed genes form distinct operon-sized chromosomal interaction domains (OPCIDs) in a transcription-dependent manner. These structures appear as square patterns on Micro-C maps, reflecting continuous contacts throughout transcribed regions. This work unveils the fundamental structural elements of the E. coli nucleoid, highlighting their connection to nucleoid-associated proteins and transcription machinery.

PubMed Disclaimer

Conflict of interest statement

Competing interests: The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. Ultra-high-resolution Micro-C maps reveal the complex 3D organization of the E. coli genome.
a, Agarose gel electrophoresis of DNA after different steps of the Micro-C procedure, demonstrating the fragmentation pattern required for ultra-high-resolution mapping (for the uncropped gel, see Supplementary Fig. 1). As E. coli lacks nucleosomes, MNase does not produce the characteristic 200-bp DNA ladder but instead generates a smear of fragments. All Micro-C experiments were tested to show similar separation patterns. b, Examples of Micro-C contact maps showing OPCIDs, CHINs and CHIDs alongside transcription activity profiles and operon tracks. c–e, Comparison between Micro-C and Hi-C data: Micro-C maps generated from WT E. coli cells under normal growth conditions (c); Hi-C contact maps generated using the standard Hi-C protocol on the same cells (d); and previously published, Hi-C maps for E. coli MG1655 (e). The left panels show the whole-genome view. Note that reducing bin sizes in Hi-C (below 500 bp) does not significantly enhance resolution, whereas Micro-C maps reveal much finer details, particularly at sub-kilobase resolution.
Fig. 2
Fig. 2. Transcription-driven OPCIDs on active operons and transcription-independent CHINs and CHIDs in repressed genomic regions.
a, Average contact map for operons grouped by transcription activity, based on the Red-C signal. b, Transcription level around CHINs, CHIDs, OPCIDs, inactive operons (zero Red-C signal) and active operons (top 10% by Red-C signal). The curves above the piles-up represent median signal, and the shaded areas denote the bootstrap standard error. c, Transcription level within specified genomic features, normalized by feature length. Solid and dashed lines represent the median and 25th to 75th percentiles, respectively. d, Contact maps for a region harbouring the heat shock (HS) operon clpX-lon in normal and HS conditions. e, Average contact maps for σ32 operons under normal and HS conditions. n = 26. f, Schematic illustrating the assembly of an operon into OPCID on transcription activation under HS conditions. g, Frequency of promoter–terminator contacts for operons grouped by Red-C signal into non-transcribed (zero Red-C signal), lowly transcribed (bottom 50% excluding zero), moderately transcribed (50–90%) and highly transcribed (top 10%). n represents the number of operons in each group. Also shown is the frequency of contacts between internal operon regions: one located 20% downstream of the TSS and another 20% upstream of the TES. The boxplots extend from the 25th to 75th percentiles, whiskers are drawn down to the 10th and up to the 90th percentiles, and centre lines represent the median. The P values were calculated using one-way analysis of variance followed by Tukey’s multiple comparisons test. **P < 0.01 and ****P < 0.0001. Exp, expected; NS, not significant; obs, observed. The dashed line indicates the obs/exp contact frequency level of 1. h, Correlation between operon transcription level and the frequency of promoter–terminator contacts. The black line shows linear regression, and the shaded area denotes 95% CI. i, Contact maps for two representative regions harbouring an OPCID and CHINs, shown before and after treatment with 25 µg ml−1 or 750 µg ml−1 Rif. j, Average contact maps for OPCIDs, CHINs and CHIDs before and after exposure to low (25 µg ml−1) or high (750 µg ml−1) concentrations of Rif, illustrating the effect of transcriptional inhibition on these genomic structures. Source data
Fig. 3
Fig. 3. Organization of CHINs and CHIDs directed by H-NS and other NAPs.
a, Example contact map with profiles of NAPs distribution, GC content, transcription signal and operon track. b, Distribution of H-NS around CHINs, CHIDs, OPCIDs, inactive operons and active operons. Annotations are consistent with Fig. 2b. c, H-NS motif from the PRODORIC database (left) alongside the CHIN motif identified by HOMER (right). d, A model depicting the organization of a single CHIN, with CHIN length distribution shown (right; n represents the number of CHINs assessed). dsDNA, double-stranded DNA. e, A model illustrating the interaction between two CHINs, where they unfold and fuse to form a stem of a large loop. The distribution of distances at which CHINs interact is also shown (right; n represents the number of CHIN contacts assessed). The boxplots extend from the 25th to 75th percentiles, the whiskers are drawn down to the 10th and up to the 90th percentiles, and the centre line represents the median (d,e). f, Localization of H-NS within the stem regions of CHINs, showing peaks at CHIN stems. g, Number of CHINs overlapping with H-NS peaks. h, Number of HTGs overlapping with CHINs. The P values were calculated using two-sided shuffle test: P < 0.001 (g,h). i–r, Contact maps of two representative CHIN-harbouring regions in WT E. coli cells (i) and the same regions in strains with deletions of the indicated genes (j–n) or in WT cells treated with antibiotics (o–r), illustrating the CHIN reorganization in Δhns cells and complete CHIN disassembly in ΔhnsΔstpA and netropsin-treated cells. The average contact map for CHINs under these conditions is also shown (right). Source data
Fig. 4
Fig. 4. Activation of HTGs on H-NS and double hns stpA knockout.
a–d, Gene expression changes following deletion of hns and stpA (a), exposure to 200 µg ml−1 netropsin (b), deletion of hns (c) and deletion of stpA (d) as determined by RNA sequencing. The distribution of HTGs and non-HTGs that are upregulated, downregulated or insignificantly altered in Δhns, ΔstpA, ΔhnsΔstpA and netropsin-treated E. coli cells is also shown. The P values were calculated using two-sided Wald test and adjusted for multiple testing using Benjamini–Hochberg correction with false discovery rate set to 0.05. FC, fold change. e, Expression levels of genes overlapping with CHINs and those not overlapping with CHINs in WT E. coli cells (green) compared with the expression levels of the same gene groups in Δhns (blue), ΔstpA (white) and ΔhnsΔstpA (red). f, Expression changes of CHIN-associated and CHIN-non-associated genes in response to Δhns, ΔstpA and ΔhnsΔstpA. g,h, Similar to panels e and f, but focusing specifically on HTGs. The number of genes (n) is indicated. The boxplots extend from the 25th to 75th percentiles, whiskers are drawn down to the 10th and up to the 90th percentiles, and the centre lines represent the median. The P values were calculated using the two-sided Kolmogorov–Smirnov test: **P < 0.01 and ****P < 0.0001. The black dashed line indicates the fold change value of 1. i, Changes in expression of selected HTGs upon Δhns, ΔstpA and ΔhnsΔstpA, as determined by RT–qPCR. Signals are normalized to 16S rRNA, with the expression level in WT cells set to 1. Data are presented as mean ± s.d. j, Micro-C contact maps of regions harbouring HTGs selected for RT–qPCR analysis in WT, ΔstpA, Δhns and ΔhnsΔstpA cells. k, Growth rates of WT, ΔstpA, Δhns and ΔhnsΔstpA E. coli cells measured using Bioscreen. Data are presented as mean ± s.d. from n = 7 biological replicates for WT, ΔstpA and Δhns, and n = 14 for ΔhnsΔstpA. OD600, optical density at 600 nm. Source data
Fig. 5
Fig. 5. Elementary and higher-order structures captured by Micro-C in the E. coli genome.
a, Example contact maps from WT E. coli cells showing various contact patterns. b, Interpretation of the different contact patterns. c, OPCIDs and CHIN networks, as mapped in this study, can facilitate RNAP recycling by bringing promoters (TSSs) and terminators (TESs) into close proximity (i and ii). They can also establish spatial insulation of active operons (ii), enable recombination between distant HTGs (iii) and promote HTG silencing (iv).
Extended Data Fig. 1
Extended Data Fig. 1. Comparison of Micro-C data at different resolutions and detection of CHINs in Micro-C and Hi-C data.
a, Micro-C contact map of a region harboring two CHINs displayed at different map resolution. b, Examples of CHINs seen in Hi-C contact maps obtained in this study, alongside corresponding Micro-C contact maps of the same regions. All maps are provided at a resolution of 200 bp. c,d, Average contact map for CHINs (c) and the top10% of active operons (d) generated from Micro-C contact maps with resolution of 10 and 200 bp. e, Representative examples of Micro-C contact maps showing the positions (centers) of CHINs, annotated either manually or using the Chromosight program. f, Intersection of CHINs mapped manually and those identified by the Chromosight program. g, Distribution of H-NS around CHINs, as mapped manually and by the Chromosight program.
Extended Data Fig. 2
Extended Data Fig. 2. Plaid patterns observed in contact maps.
Examples of Micro-C contact maps for wild-type E. coli cells in normal growth conditions are shown.
Extended Data Fig. 3
Extended Data Fig. 3. Effect of heat shock, rifampicin, and bleomycin treatment on OPCIDs, CHINs, and CHIDs.
a, Example of OPCID appearance under HS. b,c, Examples of OPCID disappearance under HS. d, A region harboring three OPCIDs, left (at gntXY) appearing under HS, middle (at malPQ) disappearing under HS, and right (at malT) weakening under HS conditions. e,f, Examples of OPCIDs that are invariant between normal and HS conditions. g,h, Preservation of CHIN structures under HS conditions. i, Examples of Micro-C contact maps for non-treated and rifampicin-treated E. coli cells. j, Examples of Micro-C contact maps for non-treated and bleomycin-treated E. coli cells.
Extended Data Fig. 4
Extended Data Fig. 4. Distribution of proteins (a-k) and GC content (l) around different genomic features in wild-type E. coli cells.
All designations are as in Fig. 2b.
Extended Data Fig. 5
Extended Data Fig. 5. CHIN validation in vivo and in vitro.
a-d, Detection of CHIN assembly in vivo using proximity ligation assay (PLA). a, Schematic of the PLA approach. Two His-tagged dCas9 molecules are targeted to the CHIN bases via specific guide RNAs. Mouse and rabbit anti-His primary antibodies are added, followed by oligonucleotide-conjugated anti-mouse and anti-rabbit secondary antibodies. Each His-tag is recognized by only one antibody at a time, so the two dCas9 molecules are bound by either homotypic (mouse-mouse or rabbit-rabbit) or heterotypic (mouse-rabbit or rabbit-mouse) antibody pairs. If the dCas9 molecules are brought in close spatial proximity (<40 nm), the oligonucleotides on the heterotypic antibody pair are annealed to connector oligonucleotides, enabling their ligation into a circular DNA template. This DNA circle is then amplified by rolling circle amplification (RCA), generating a micrometer-sized DNA bundle with multiple binding sites for a fluorophore-labeled oligonucleotide, which is subsequently visualized by HILO microscopy. b, Micro-C contact map of the studied E. coli genomic region harboring a strong CHIN, with positions of the guide RNAs indicated. The inferred CHIN configuration and guide RNA locations are shown below. c, Representative PLA images of E. coli cells expressing His-tagged dCas9 and the indicated guide RNA pairs. d, Quantification of cells showing PLA foci under indicated conditions. Data are presented as mean values from two biological replicates, with at least 200 cells analysed per condition in each replicate. e-j, Transmission electron microscopy (TEM) analysis of H-NS complexes with plasmids containing E. coli genomic regions either prone or not prone to CHIN formation. e, Micro-C contact maps of the E. coli genomic regions inserted into plasmids. f, Schematic of the plasmids. The CHIN-containing plasmid harbors region I, which shows a strong CHIN signal in the Micro-C map (same region as analysed in a-d). The control plasmid lacks CHINs and contains regions II and III, which do not exhibit CHIN signatures. g, Representative TEM images of the CHIN-containing plasmid incubated with H-NS, showing nucleoprotein filaments (red arrows). Plasmid DNA is marked by black arrows. h, TEM image of the control plasmid (no CHINs) after incubation with H-NS. No nucleoprotein filaments are detected. i, TEM image of the CHIN-containing plasmid in the absence of H-NS. No filaments are observed. j, Quantification of plasmids displaying nucleoprotein filaments under the indicated conditions. Data are presented as mean values from two independent experiments, with at least 100 plasmids analysed per condition in each experiment. Source data
Extended Data Fig. 6
Extended Data Fig. 6. Colocalization of CHINs and CHIDs with HTGs.
a, Examples of colocalization of HTGs (dark green filled arrows on the gene tracks) with individual CHINs and CHIN networks. b, CHINs and CHIDs formed on HTGs near repetitive elements. c, Colocalization of CHIDs with HTG clusters. Shown are Micro-C contact maps for wild-type E. coli cells in normal growth conditions.
Extended Data Fig. 7
Extended Data Fig. 7. Validation of knockout strains and correlation of Micro-C data between biological replicates.
a, Micro-C contact maps of wild-type and deletion mutant E. coli cells for several genomic regions spanning deleted genes. The white cross over the deleted gene in the corresponding cell indicates successful deletion. b, Stratum-adjusted correlation coefficients between samples, assessing data consistency across replicates and the degree of similarity of Micro-C matrices. c,d, Replicate Micro-C contact maps of wild-type E. coli cells in normal (c) and heat shock (d) conditions. e-g, Interaction-versus-distance (scaling) plot for various samples.
Extended Data Fig. 8
Extended Data Fig. 8. Intensification of CHIN spatial contacts in Δhns cells.
Examples of Micro-C contact maps for wild-type, Δhns, ΔstpA, and ΔhnsΔstpA E. coli cells are presented. Below the wild-type contact maps, the distribution profiles for H-NS and StpA in wild-type E. coli are provided alongside the operon track.
Extended Data Fig. 9
Extended Data Fig. 9. Micro-C control experiments and optimization of MNase digestion.
a, Micro-C contact map for wild-type E. coli cells processed according to the standard Micro-C protocol, but without the addition of DNA ligase. b-d, Comparison of Micro-C contact maps for wild-type E. coli cells grown at 37 °C and fixed with formaldehyde at room temperature, as per the standard Micro-C protocol (b), those grown and fixed at room temperature (c), and those grown and fixed at 37 °C (d). e, In vivo protein-protein crosslinks induced by treatment of E. coli cells with DSG and DSS, assessed by mass spectrometry. Protein names are according to UniProt. f, Distribution of DNA between soluble (super) and insoluble (debris) fractions following treatment with increasing amounts of MNase of wild-type E. coli cells fixed with formaldehyde alone (upper gel) or with formaldehyde and DSG (lower gel). g, Gel shows a typical MNase titration experiment. An amount of 5 U was selected as optimal, producing the desired fragment sizes range. h, Gels show four different samples tested after MNase digestion and ligation for two biological replicates. Sample 4 (non-relevant to the present study) indicates over-digestion, leading to excessive fragmentation, in replicate 1. For uncropped gels, see Supplementary Fig. 1. i, Micro-C contact maps for replicate 1 and 2 of sample 4, demonstrating similarity between contact maps (stratum-adjusted correlation coefficient 0.978).
Extended Data Fig. 10
Extended Data Fig. 10. Inferring transcription level from Red-C data.
RNA-DNA (Red-C) and DNA-DNA (Micro-C) contact maps for the same genomic region are shown, along with the operon track, gene track, and transcription activity profile. The transcription activity is determined by summarizing, for each DNA bin, the number of contacts that RNA produced from this bin makes within ±10 kb from the bin. For example, the transcription activity of the cydAB operon is visualized by the signal in the dotted area of the RNA-DNA contact map. The red vertical line in the RNA-DNA contact map marks tRNA contacts.
See this image and copyright information in PMC

References

    1. Marti-Renom, M. A. et al. Challenges and guidelines toward 4D nucleome data and model standards. Nat. Genet.50, 1352–1358 (2018). - DOI - PubMed
    1. Dekker, J. et al. Spatial and temporal organization of the genome: current state and future aims of the 4D nucleome project. Mol. Cell83, 2624–2640 (2023). - DOI - PMC - PubMed
    1. Kleckner, N. et al. The bacterial nucleoid: nature, dynamics and sister segregation. Curr. Opin. Microbiol.22, 127–137 (2014). - DOI - PMC - PubMed
    1. Dame, R. T. & Tark-Dame, M. Bacterial chromatin: converging views at different scales. Curr. Opin. Cell Biol.40, 60–65 (2016). - DOI - PubMed
    1. Verma, S. C., Qian, Z. & Adhya, S. L. Architecture of the Escherichia coli nucleoid. PLoS Genet.15, e1008456 (2019). - DOI - PMC - PubMed

Publication types

MeSH terms

LinkOut - more resources