Multiscale Structuring of the E. coli Chromosome by
SUMMARY
As in eukaryotes , bacterial genomes are not randomly folded . 
Bacterial genetic information is generally carried on a circular chromosome with a single origin of replication from which two replication forks proceed bidirectionally toward the opposite terminus region . 
Here , we investigate the higher-order architecture of the Escherichia coli genome , showing its partition into two structurally distinct entities by a complex and intertwined network of contacts : the replication terminus ( ter ) region and the rest of the chromosome . 
Outside of ter , the conden-sin MukBEF and the ubiquitous nucleoid-associated protein ( NAP ) HU promote DNA contacts in the meg-abase range . 
Within ter , the MatP protein prevents MukBEF activity , and contacts are restricted to 280 kb , creating a domain with distinct structural properties . 
We also show how other NAPs contribute to nucleoid organization , such as H-NS , which restricts short-range interactions . 
Combined , these results reveal the contributions of major evolutionarily conserved proteins in a bacterial chromosome organization . 
INTRODUCTION
The genomes of all organisms must be folded to fit within a cell that is typically several 1,000-fold smaller than the size of the DNA molecule itself . 
The overall chromosome fold is a combination of intertwined structural features resulting from differential accessibility , polymer properties , epigenetic modifications , and binding of proteins to the sequence . 
Recent work has highlighted the dynamics and regulation of this complex network and its functional interplay with metabolic processes over time , such as gene expression regulation , chromosome segregation , and repair ( Dekker and Mirny , 2016 ) . 
In bacteria , DNA is efficiently compacted into the nucleoid , a dynamic macromolecular complex where the genetic material and its associated proteins are located . 
Nucleoid folding and compaction result from a combination of processes ( Wang et al. , 2013 ; Kleckner et al. , 2014 ) : DNA supercoiling , formation of ( elusive ) chromatin-like structures by nucleoid-associated proteins ( NAPs ) , condensation by structural maintenance of chromosome ( SMC ) proteins , macromolecular crowding , and out-of-equilibrium processes such as transcription . 
The mostly negatively supercoiled DNA results in branched and plectonemic structures whose precise role ( s ) and distribution remain poorly understood . 
Genetic studies revealed the presence of stochastic boundaries between these structures , allowing DNA interactions in cis between sites located more than 100 kb apart ( Higgins et al. , 1996 ) . 
RNA polymerase blocks the movement of the plectonemic supercoil in the transcribed track , generating supercoil diffusion barriers ( Booker et al. , 2010 ) . 
Those constraints were proposed to segment the chromosome into `` chromosomal interaction domains '' ( CIDs ) ranging in length from 30 to 400 kb and identified through chromosome conformation capture ( 3C/Hi-C ) ( Dekker et al. , 2002 ; Le et al. , 2013 ) because these domains often display highly expressed genes at their boundaries in Caulobacter crescentus , Bacillus subtilis , and Vibrio cholerae ( Le et al. , 2013 ; Marbouty et al. , 2015 ; Val et al. , 2016 ) . 
Co-regulated genes were recently found to generate smaller ( 15 - to 30-kb ) domains in Mycoplasma pneunomiae ( Trussart et al. , 2017 ) . 
NAPs are highly abundant proteins that play diverse roles as chromatin organizers , transcription factors , and , more generally , accessory partners involved in DNA transactions ( Dillon and Dorman , 2010 ) . 
In E. coli , at least ten abundant ( 30,000 -- 60,000 copies per cell ) DNA binding proteins have been found to be associated with the nucleoid . 
These NAPs bend , wrap , or bridge DNA , to which they show different types of affinities ; for instance , high and specific ( e.g. , Fis ) or non-specific and with a preference for AT-rich sequences ( e.g. , HU ) . 
Until recently , NAPs were suspected to contribute to DNA condensation by acting on the structuring and regulation of the chromatin only at a local scale . 
Although a recent 3C experiment unveiled involvement of HU for contacts up to 100 kb in C. crescentus ( Le et al. , 2013 ) , the respective contributions of NAPs to chromatin organization and DNA dynamics in vivo remain poorly understood . 
Higher-order levels of organization of bacterial chromosomes have been described in recent years , involving long-range contact structuring of the genome over large distances . 
The E. coli chromosome is segmented into macrodomains ( MDs ; Niki et al. , 2000 ; Valens et al. , 2004 ; Espeli et al. , 2008 ) , with the DNA binding protein MatP specifying a constrained 800-kb region ( ter ) surrounding the terminus of the replication locus ( Mer-cier et al. , 2008 ) . 
The functional requirement for organizing this ter domain is not completely understood . 
The interaction of MatP with the protein ZapB associated to the divisome promotes anchoring of the ter at the midcell and , therefore , controls chromosome choreography during the cell cycle ( Espéli et al. , 2012 ) . 
The interplay of MatP with MukBEF associated with Topo-isomerase IV may ensure timely chromosome unlinking and segregation ( Nolivos et al. , 2016 ) . 
By performing a structurefunction analysis of MatP , the molecular bases for MatP-medi-ated ter formation were identified . 
MatP contains a tripartite fold that includes a four-helix bundle ( corresponding to the DNA binding domain ) , a ribbon-helix-helix ( RHH ) domain ( responsible for the formation of the MatP dimer ) , and a C-termi-nal coiled coil . 
Although the RHH domain promotes the formation of MatP dimers , the coiled-coil regions form a bridged 
MatP tetramer that might flexibly link distant matS sites , prompting a model for a protein-mediated DNA-looping mechanism for ter organization ( Dupaigne et al. , 2012 ) . 
Mutating the residues involved in the tetramerization affects both DNA condensation and the ability of MatP to interact with ZapB ; by contrast , these mutants were still able to specify the ter MD ( Dupaigne et al. , 
2012 ; Espéli et al. , 2012 ) . 
In other species lacking MatP , large domains have also been characterized . 
In B. subtilis , a large 800-kb region overlapping the origin of replication is maintained into a constrained , dense state through the action of SMC proteins , as revealed by super-resolution imaging and 3C ( Marbouty et al. , 2015 ) . 
It was speculated that the condensation of this domain plays a role in the proper completion of the replication and segregation program . 
The disposition of the chromosome within the cell differs between bacterial species . 
In B. subtilis , C. crescentus , and M. pneunomiae , the chromosome has a longitudinal disposition , with the two replication arms aligned along the long axis of the cell ( Le et al. , 2013 ; Marbouty et al. , 2014 , 2015 ; Trussart et al. , 2017 ; Umbarger et al. , 2011 ; Wang et al. , 2015 ) . 
In E. coli , the chromosome has a transversal disposition , with the two replication arms occupying distinct nucleoid halves and the replication origin in between ( Wang et al. , 2006 ) . 
The different chromosome dispositions observed in bacteria suggest that different factors may be involved in chromosomal organization . 
Despite numerous efforts to understand the role of each structural factor in the overall chromosome organization among diverse species , their precise effect is yet to be fully understood . 
Genomic ana-lyses have shown that bacterial species exhibit diverse combinations of organizing factors . 
In E. coli and enterobacteria , these structural factors involve the ubiquitous NAPs such as HU , Fis , and H-NS as well as a specific group that coevolved with Dam methylase , including the condensin complex MukBEF and MatP ( Brézellec et al. , 2006 ) . 
In this study , we explored the higher-order E. coli chromosome organization and reveal the effect of several factors controlling its folding . 
The analysis of high-resolution ( 5 kb ) 3C contact maps of the E. coli chromosome in wild-type ( WT ) and mutant backgrounds reveals a multilevel 3D organization mediated by the major ubiquitous NAPs as well as the influence of transcription on local chromatin structure . 
The teaming up of HU and the condensin MukBEF to promote long-range contacts within chromosome arms and the formation of a specific chromosomal domain through the restraint of condensin activity by MatP provide clues regarding long-range chromosome organization and domain structuring in bacteria . 
RESULTS
A High-Resolution Contact Map of the E. coli Chromosome
3C coupled with deep sequencing was applied to exponentially
growing WT cells to investigate the precise effect of nucleoid structuring factors on chromosome organization in E. coli . 
In agreement with the transverse disposition of the E. coli chromosome , the contact map displayed a single strong diagonal resulting from the enrichment of contacts between neighboring loci ( Figures 1A and S1A ) . 
The absence of a secondary diagonal perpendicular to this main one reflects the lack of contacts between the two replication arms and offers a sharp contrast to bacterial chromosomes characterized so far , such as that of C. crescentus ( Umbarger et al. , 2011 ; Le et al. , 2013 ) or B. subtilis ( Marbouty et al. , 2015 ; Wang et al. , 2015 ) . 
Replicate experiments performed on WT cells produced highly reproducthe robustness of this 3C protocol to investigate the spatial organization and patterns of DNA interactions throughout the E. coli chromosome . 
To facilitate the interpretation of contact maps , we developed a visualization tool dubbed `` scalogram , '' which represents , for each bin , the cumulated contact frequencies as a function of the genomic distance ( STAR Methods ; Figure S1E ) . 
A scalogram therefore displays the accumulated distribution of contacts for each bin with its flanking regions , reflecting the relative tightness of the contact distribution ( Figure 1D ) . 
Abrupt changes in signal along the chromosome reveal three regions of 0.5 -- 1 Mb in size each that exhibits a distinct contact pattern : a single highly constrained domain around ter ( ter ) and two loosely structured regions ( L1 and L2 ) whose loci form contacts over a larger distances . 
This genome segmentation correlated with domains identified by a directional index analysis that reports the degree of upstream and downstream interactions for a genomic region carried out at a scale of 400 kb ( Figure 1E ; STAR Methods ; Dixon et al. , 2012 ) . 
This organization was conserved across three rep-licates ( Figures S1A ) and , in asynchronous E. coli cell populations , growing in different media and at different temperatures ( Figures S1F and S1G ) . 
Overall , these three regions underline the segmentation of the chromosome into six intervals ( Figure 1F , top ) that correlate well with those defined by a genetic recombination assay that identified four ( Ori , ter , left and right ) MDs and two left and right non-structured regions ( Figure 1F , bottom ; Va-lens et al. , 2004 ) . 
In conclusion , the WT genome-wide contact map of E. coli validates previous genetic and imaging studies that revealed the non-homogeneous organization of the chromosome and confirmed the existence of a peculiar folding for the 
To determine whether the distribution of contacts made by a chromosomal locus correlated with its dynamics ( STAR Methods ) , the cumulative contact signal at various distances was compared with mean square displacements ( MSDs ) of chromosomal loci at short timescales ( Espeli et al. , 2008 ; Javer et al. , 
2014 ; Figure S1H ) . 
This analysis revealed that cumulative contact signals at 200 kb best correlated with the MSD measured at 10 s by Javer et al. ( 2014 ) or at 180 s by Espeli et al. ( 2008 ) . 
As shown by the strong anti-correlation observed between the cumulated 3C signal at 200 kb and the MSDs of several loci ( Figure 1H ; Figure S1I ) , contact maps can provide insights into the 
Interplay between Transcription and Local Chromatin Structure
Chromosomal structures ranging in size from 15 to 33 kb ( Myco-plasma pneumonia ; Trussart et al. , 2017 ) and 20 to 200 kb ( C. crescentus and B. subtilis ; Le et al. , 2013 ; Marbouty et al. , 2015 ) have been characterized . 
A directionality index ( DI ) analysis performed at a scale of 100 kb ( Le et al. , 2013 ) along the E. coli chromosome identified 31 CIDs ranging in size from 40 to 300 kb in exponential phase ( average , 150 kb ; Figure S2A ) . 
Boundaries were conserved across all exponential growth conditions ( Figure S1G ) and in different genetic backgrounds ( below ) . 
Boundaries were enriched in highly expressed , sometimes long ( ( Le and Laub , 2016 ; Marbouty et al. , 2015 ) transcription units ( 22 of 31 ; Figure S2B ; STAR Methods ) and in genes encoding proteins with an export signal sequence ( signal recognition particle [ SRP ] genes , 9 of 31 ) . 
This diversity suggests that multiple mechanisms may be responsible for defining CID boundaries , including local decompaction of active transcribed regions ( Le and Laub , 2016 ) . 
We next investigated how transcription may affect short-range contacts along the chromosomes . 
The contact frequency between adjacent bins was plotted along the transcription profile at resolutions of 2 kb and 5 kb . 
The strong correlation ( Pearson correlation [ PC ] > 0.5 ) between the two signals suggests that transcription levels correlate with short-range contact frequencies ( Figures S2C and S2D ) . 
Contact frequencies as a function of genomic distance were then plotted for genes pooled according to their expression level ( Figure S2E ; STAR Methods ) , revealing that higher levels of expression correlated in enrichment in short range contacts as well as stronger decay of the slope . 
Remarkably , these correlations were also observed in contact maps of other bacteria generated by different laboratories using different enzymes and crosslinking conditions ( Figures S2C and S2E ) , suggesting the existence of transcription-induced constraints that favor interactions between neighboring loci . 
Different interpretations can be provided to account for this observation . 
For instance , a less mobile locus ( above ) would indeed result in fewer long-range contacts and an increase in short-range contacts . 
Alternatively , a more open fiber may also lead to a stronger decay of the contact slope . 
The correlation was not apparent in C. crescentus , but similar trends were also observed in Drosophila ( Corrales et al. , 2017 ) . 
Contact maps of non-dividing E. coli cells in stationary phase revealed a large-scale chromosomal reorganization with an increase of long-range contacts more pronounced in ter ( Figures S1F and S1J ) . 
A directional index analysis performed at a scale of 100 kb identified 30 CIDs ( Figure S2F ) , and the boundaries identified under this condition were different from those identi-fied in exponential phase ( Figure S2G ) . 
Interestingly , here too a significant correlation ( PC = 0.40 ) was found between transcription levels and short-range contact frequencies ( Figures S2H and S2I ) . 
The causal relationships between these correlations ( transcription , short-range contact decay , and dynamics ) remain to be deciphered . 
Organization of ter and a Role for MatP in ter Insulation
The MatP protein plays a major role in the ter organization of enterobacteria ( Mercier et al. , 2008 ; Dupaigne et al. , 2012 ) . 
To gain insights into the molecular mechanism by which MatP bound to matS sites structures the ter MD , genomic 3C experiments were performed in a matP mutant . 
In the absence of MatP , the pattern of chromosome contacts was conserved across the genome , except for ter , which now appeared similar to the rest of the genome ( Figures 2A , 2B , and S3A ) . 
In the absence of MatP , an enrichment in long-range contacts ( more than 280 kb ) within ter and between ter and its flanking regions appeared , accompanied by a compensatory decrease in contacts under 280 kb within ter ( Figure 2B ) . 
These results are consistent with genetic recombination assays that show , in the absence of MatP , an enrichment of long-range contacts within ter ( Figure 2C ) and between ter and its flanking domains ( Mercier et al. , 2008 ) 
Remarkably , matS sites did not display enriched contacts with each other ( Figures 2D and S3B ) . 
These observations incited us to investigate more precisely the mechanism of MatP-medi-ated organization of ter . 
A matPDC20 derivative unable to form MatP tetramers and to interact with ZapB ( Dupaigne et al. , 2012 ; Espéli et al. , 2012 ) did not significantly modify the contact map ( Figure 3A ) , as shown by the ratio of normalized contact maps ( Figure S3C ) or the ratio plot of contact signals ( Figure 3B ) . 
Similarly , a zapB mutation that abolishes the interaction of MatP with the divisome did not affect the contact pattern ( Figures 3C , 3D , and S3D ) . 
Therefore , tetramerization of MatP and the interaction of MatP with the division machinery are not required for maintaining the architecture of ter or for insulating it from flanking MDs . 
These observations do not support the hypothesis that MatP tetramers bridge matS DNA sites into a single 800-kb intertwined domain ( Dupaigne et al. , 2012 ) but suggest instead that MatP binds to matS sites to organize the ter MD into a succession of overlapping subdomains . 
MatP was also described for its ability to maintain together sister ter regions extensively following replication . 
We tested whether intermolecular interactions between chromosome and plasmids carrying matS sites could be revealed by chromosome conformation capture sequencing ( 3C-seq ) . 
The behavior of a plasmid with and without matS sites was investigated in WT , matPDC20 , and zapB mutants . 
In WT cells , targeting of MatP at the septum ring promotes the anchoring of the replicated ter at the midcell . 
As a consequence , plasmids carrying matS sites also colocalize with ter at the midcell ( Espéli et al. , 2012 ) . 
Accordingly , the plasmid displayed enriched contacts with ter compared with a plasmid devoid of matS ( Figures 3E and 3F ) . 
Contacts between the plasmid and the chromosome did not exhibit a discrete pattern that would result from matS-matS contacts ( Figure 3E , inset ) , further confirming that MatP does not specifically connect these sites . 
In the matPDC20 and zapB mutants , plasmids with or without matS sites do not position at the midcell ( Dupaigne et al. , 2012 ; Espéli et al. , 2012 ) , but whether they interact with ter is unknown . 
In these mutants , although ter displays a WT organization ( Figures 3A and 3C ) , contacts between ter and the matS plasmid are lost ( Figures 3G -- 3J ) . 
This demonstrates that MatP-dependent intermolecular contacts between molecules ( or replicons ) carrying matS sites are not involved in ter structuring and that the colocalization of matS sites at the midcell requires ZapB . 
The Condensin MukBEF Promotes Long-Range Chromosome Folding
E. coli has one single SMC complex , MukBEF . 
This complex is essential for correct chromosome segregation and conformation ( Nolivos and Sherratt , 2014 ) . 
In a mukB deletion mutant , 20 % of anucleate cells are produced at a permissive temperature 
( 11 % of right loci in 0.75 of the cell length in WT cells versus 22 % in mukB ; Figure S4C ) . 
To investigate the effect of MukBEF on chromosome organization at the molecular level , a contact map of cells depleted in mukB was generated ( Figures 4A and S5A ) . 
In the absence of condensin , the contact map ratio showed a reduction in long-range ( > 280 kb ) contacts concom-itant with an increase of mid-range contacts up to 280 kb along the chromosome compared with the WT strain , except in ter ( Figures 4B and S5B ) . 
No significant changes were detected in the ter MD of the mukB mutant compared with WT cells ( Figure 4B ) . 
Altogether , these results suggest that MukBEF promotes long-range ( > 280 kb ) contacts within replication arms outside of ter . 
Restriction of MukBEF-Dependent Long-Range Contacts in ter by MatP
Previous work in E. coli showed a physical interaction between MatP and MukB both in vivo and in vitro ( Nolivos et al. , 2016 ) . 
This interaction has been proposed to promote the displacement of MukB out of ter , facilitating the association of MukBEF with the Ori region . 
The unaltered contact pattern observed in ter in muk cells suggests that MukBEF is not active in ter organization . 
To test this hypothesis , the chromosome conformation contact map of a double mukB matP mutant ( Figure S5C ) was compared with a matP single mutant ( Figures S5C and S5D ) . 
These results showed a reduction in long-range contacts over the entire chromosome , including ter , in the absence of MukB ( Figure 4C ) , indicating that MatP impedes MukBEF activity in ter . 
In agreement with these data , the inactivation of MatP allows MukBEF to interact with ter and , hence , to increase long-range interactions ( > 280 kb ) in this region ( Figures 2B and 2C ) . 
Combined , our data reveal that MukBEF promotes long-range contacts along the chromosome , except in ter , where this activity is reduced or alleviated by MatP . 
Therefore , the peculiar structure of ter appears to result from a default of access by MukB instead of active folding promoted by MatP . 
HU Is Also Essential to Promote Long-Range Communication
Although NAPs have long been known to modulate DNA conformation by bending , wrapping , or bridging it , their exact contribution to chromosome folding in vivo is still unknown . 
The involvement of the NAPs Fis , H-NS , and HU in chromosome conformation was therefore investigated . 
NAP mutants were grown under conditions where growth defects were minimalized ( Figure S4 ; STAR Methods ) . 
We first focused on the conserved protein HU . 
In E. coli , HU exists as an heterodimer ( HUab ) or as a homodimer ( HUa2 ) ( Claret and Rouviere-Yaniv , 1997 ) and is one of the most abundant NAPs in exponential phase . 
The conserved HU protein binds non-specifically to DNA with a preference for AT-rich sequences ( Prieto et al. , 2012 ) . 
In the hupAB mutant , E. coli cells present segregation defects , filament formation , and nucleoid compaction ( Figures S4B and S4D ) . 
To determine the role of HU in chromosome conformation , 3C contact maps were produced for a hupAB mutant ( Figure 4D ) and 
Remarkably , this increase in contacts is similar to that observed in the absence of MukBEF activity outside of ter . 
In ter , the absence of HU leads to an increase in contacts in the 5 - to 50-kb range and a reduction in contacts in the 50 - to 280-kb range ( Figures 4E , S5E , and S5F ) . 
These results reveal the existence of multiple mechanisms of DNA folding in ter , with HU favoring contacts in the 50 - to 280-kb range . 
No correlation between the DNA binding profile of HU with the contact map ratio at short scales was found ( Figures S5G -- S5I ) , whereas most CID borders identified in WT cells were retained in hupAB mutants ( 23 of 29 identified under these conditions ; Figure S5J ) . 
Altogether , these results show that HU is required to maintain DNA contacts in the megabase range outside of ter and up to 280 kb within ter . 
The Roles of Fis and H-NS in Chromosome Organization
Fis , the most abundant DNA-binding protein in E. coli , binds to 1,200 sites and modulates the expression of hundreds of genes ( Cho et al. , 2008 ; Kahramanoglou et al. , 2011 ) . 
Fis is also thought to play an important role in shaping nucleoid structure by bending DNA and promoting the branching of plecto-nemes ( Hardy and Cozzarelli , 2005 ; Skoko et al. , 2006 ) . 
In the absence of Fis , cells were longer ( 5.21 ± 1.8 mm versus 3.05 ± 0.74 mm ) , with minor chromosome segregation defects , and nucleoids were more spread out compared with WT cells ( Figures S4B and S4E ) . 
The 3C contact map for fis showed that the overall chromosome conformation remained conserved compared with the WT ( Figure 5A ) , including CID boundaries ( 22 of 31 ; Figure S6A ) . 
However , the ratio of the contact maps ( Figure S6B ) and the ratio plot of contact signals ( Figures 5B and S6C ) between fis and WT cells revealed an enrichment of contacts in the 5 - to 100-kb range , a strong decrease above 200 -- 400 kb , and a strong decrease above 100 kb in ter . 
The contact ratio along the genome did not correlate with the density of Fis binding sites ( Figures S6D -- S6G ) . 
To further investigate the reduction of contacts , a recombination assay was performed ; it confirmed that contacts in the range of 250 kb are reduced outside of ter and that this effect is more pronounced in ter ( Figure 5C ) . 
Although the underlying mechanisms promoted by Fis responsible for this higher-order architecture remain unknown , these results show that this NAP is a global player of chromosome folding by promoting contacts beyond 100 kb without discrimination along the genome . 
The transcriptional repressor H-NS prevents the transcription of horizontally acquired genes in enterobacteria . 
Chromatin immunoprecipitation sequencing ( ChIP-seq ) experiments confirmed that , in vivo , H-NS binds specifically to AT-rich sequences and spreads upon binding ( Kahramanoglou et al. , 2011 ) . 
To investigate the role of H-NS in chromosome organization , contact maps were determined in a deletion mutant . 
The overall conformation of the chromosome ( Figure 5D ) and the distribution of CIDs are highly conserved compared with WT cells ( 26 CID borders of 31 ; Figure S6H ) . 
The ratio of the contact maps ( Figure S6I ) and the ratio plot of contact signals ( Figures 5E and S6J ) between hns and WT contact patterns demonstrated variations in DNA contacts in the absence of H-NS , with the removal of H-NS resulting in a significant enrichment in short-range contacts of H-NS binding regions ( Figures 5E , 5F , and S6C ) . 
To understand this local enrichment , the DNA binding profile of H-NS was correlated with the contact map ratio at short scales and at 5 kb resolution ( Figure S6K ; STAR Methods ) . 
A significant correlation was observed ( PC , 0.5 ; p = 7.89 e 59 ) between the two parameters ( Figure 5G ) , with bins enriched in H-NS binding sites displaying two types of behaviors in the mutant : either an increase in short-range contacts ( 70 % overlap ) or no changes ( STAR Methods ) . 
The same analysis performed with 2-kb binning of the maps led to similar , slightly noisier results with 63 % overlap 
( PC , 0.38 ; p = 2.6 e 79 ; see also Figure S6K ) . 
This result shows that the local binding of H-NS in WT cells prevents a large fraction of its targets from interacting with their neighboring loci . 
The absence of changes for the other targets may result from other processes maintaining the local folding in the absence of H-NS or preventing H-NS to fold the DNA in the WT . 
We did not observe the previously reported H-NS-promoted juxtaposition of H-NS-regulated operons ( Wang et al. , 2011 ; Figures S6I and S6L ) , and no variations in long-range contacts were detected with the recombination assay ( Figure S6M ) . 
Thus , in the absence of H-NS , short-range contacts increase in many cases , suggesting that the local binding of H-NS in 
WT cells prevents these discrete regions from interacting with neighboring loci . 
Cooperation of MukBEF and NAPs for Long-Range Chromosome Organization in E. coli Collectively , our results reveal complex , intertwined levels of higher-order organization of the E. coli nucleoid , with different players having contrasting roles in chromosome architecture ( Figures 6A and 6B ) . 
The three proteins Fis , HU , and MukBEF promote long-distance DNA contacts in the megabase range outside of ter in WT cells and on the whole chromosome in the matP mutant . 
In fis-deficient cells , long-range contacts above 
300 -- 500 kb decreased , whereas , in the absence of HU or MukB , contacts below 280 kb are enriched . 
In ter , MatP ( or MatPDC20 ) maintains contacts up to 280 kb by restricting the action of MukBEF . 
In the absence of HU , contacts in ter in the 5 - to 50-kb range increased concomitantly with a decrease in the 50 - to 280-kb range . 
Finally , in the absence of Fis , most contacts in ter occur in the 5 - to 100-kb range . 
Our results support a model in which MukBEF and HU cooperate to promote DNA contacts in the megabase range along the chromosome arms , MatP prevents MukBEF activity in ter , Fis favors DNA communications , and H-NS has only local effects on DNA conformation . 
Therefore , MatP appears to insulate ter from 
DISCUSSION
Global Organization of the E. coli Chromosome
The understanding of chromosome structuring and dynamics remains fragmented for most prokaryotic and eukaryotic spe-cies , hampering the study of their functional roles . 
Using 3C contact maps of the E. coli genome , we disclose the multilayer organization of the chromosome . 
The E. coli chromosome contact map points to an absence of contacts between arms , as expected from a transversal chromosome disposition . 
DNA collisions along the genome are not uniform , suggesting the existence of processes that modulate the probabilities of contacts at different scales . 
At a lower scale ( 40 -- 300 kb ) , CIDs are also present , as in C. crescentus , B. subtilis , and V. cholera . 
Th invariable nature of these structures in multiple growth and mutant conditions suggests local imprinting of a mark resulting in the systematic generation of boundaries within 3C datasets . 
Some of these boundaries appear to result from topological constraints ( Le et al. , 2013 ) . 
Abrupt changes of long-range contact frequencies confirm the existence of larger domains ( 0.5 to 
Partitioning of the E. coli Chromosome
The present analyses reveal a partitioning of the E. coli genome into two entities , ter and the rest of the chromosome . 
In agreement with previous work ( Nolivos et al. , 2016 ) , our data support a model in which MatP impedes MukBEF activity from ter and reveal that MukBEF is required for long-range DNA contacts 
( A ) Schematic of the bipartite structure of the E. coli chromosome , representing the differential contacts inside and outside of ter . 
Outside of ter ( nonter ) , contacts occur up to the megabase range . 
Upon inactivation of either MukB ( yellow line ) or HU ( red line ) , contacts are limited to 280 kb . 
Fis inactivation ( cyan ) variably affects the contact range along the genome . 
Inside ter , contacts are limited to 280 kb . 
In the absence of MatP , they become similar to those observed in non-ter ( dashed gray line ) . 
Upon inactivation of HU ( red line ) or Fis ( cyan ) , contacts above 50 kb and 100 kb are decreased , respectively . 
( B ) Recapitulation of contact ranges ( CRs ) observed in different mutants for the ter region ( dashed blue box ) or the rest of the genome ( dashed gray box ) illustrates the role of MatP as a regulator of chromosomal partition . 
( C ) Schematic of putative chromosome folding in ter and outside of ter . 
The arrows in the vicinity of MukBEF outside of ter represent a process not yet characterized , allowing MukBEF to promote long-range contacts . 
For clarity , plectonemic structures resulting from DNA supercoiling were omitted . 
interplay with the MukBEF complex cements the role of MatP as an important player in chromosome organization ; this activity does not require the formation of tetramers or its anchoring at the septum of division . 
MatP has already been shown to confer specific properties to ter by interacting with ZapB and localizing ter at the midcell ( Espéli et al. , 2012 ) . 
Here we demonstrate that the MatP-ZapB interaction favors intermolecular contacts between plasmids and the ter region , unveiling how sister ter MDs might be in contact before their segregation . 
MatP therefore specifies ter by two different activities : first by connecting the chromosome with the divisome through determinants in the C terminus of the protein and second by inhibiting MukBEF activity through other determinants . 
Whether only the promotion of long-range contact by MukBEF is excluded from the ter or whether other activities associated with the condensin complex are excluded as well remains unknown . 
A functional link between the role held by Topoisomerase IV ( TopoIV ) in decatenation and the facilitation of chromosome segregation by MukBEF has been proposed ( Zawadzki et al. , 2015 ) . 
Modulation of TopoIV recruitment to ter by MatP could thus control the extent of sister chromatid colocalization and coordinate the late stage of chromosome segregation with cell division ( Nolivos et al. , 2016 ) . 
It is noteworthy that MatP and MukBEF belong to a group of proteins that coevolved in enterobacteria along with Dam methylase and ten other proteins ( Brézellec et al. , 2006 ) , whose potential role in chromosome organization remains to be investigated . 
Crystal structures have revealed the dimerization of MatP di-mers bound to matS sites . 
In vitro microscopic observations of MatP-dependent loops of DNA molecules carrying multiple matS suggest that ter organization is mediated by the bridging of distant matS sites ( Dupaigne et al. , 2012 ) . 
However , 3C analyses did not unveil any discrete in vivo intrachromosomal matS-matS contacts . 
Furthermore , the chromosome and plasmids carrying matS sites were not brought together in the absence of ZapB but in the presence of MatP . 
The same plasmids in the presence of ZapB contacted ter , but no discrete matS-matS interactions could be identified . 
Combined , these results show that MatP does not promote DNA bridges between matS sites , either in cis or in trans , but promotes the formation of a chromosomal domain by exclusion of a condensin complex . 
They also reveal that MatP-ZapB interactions at the divisome are responsible for the clustering of distinct DNA molecules carrying matS sites . 
A future challenge is to achieve a full understanding of how such DNA complexes are dynamically organized during the cell cycle . 
In the absence of MatP or in the absence of both MukBEF and 
MatP , ter shows the same range of interactions as the rest of the chromosome . 
These results reveal that the absence of MukB is the major determinant of the 3C signal in ter compared with the rest of the genome and that MatP itself has little effect on ter DNA contacts . 
However , this effect in ter is exacerbated in the absence of HU or Fis , indicating a counteraction between 
MatP and these two NAPs.
NAPs Contribute in Diverse Ways to E. coli Chromosome Conformation
Decades of biochemical and genomic studies have been carried out for the three important NAPS HU , Fis , and H-NS . 
They typically cover 10 -- 30 bp of DNA at their binding sites , and , depending on local binding properties or additional interactions with other protein-bound DNA complexes , they can organize DNA into various conformations . 
However , the relation between local DNA binding and the in vivo organization of chromosomal DNA over long scales remains to be defined . 
This study provides important insights into distinct activities of three major NAPs in the control of chromosome conformation , with H-NS affecting short-range contacts , whereas HU and Fis promote long-range contacts in different ways . 
The H-NS effect is in agreement with its modus operandi of silencing extensive regions of the bacterial chromosome by binding first to nucleating high-affinity sites and then spreading along AT-rich DNA ( Lang et al. , 2007 ) . 
HU is required along with MukBEF to promote a megabase range of communications in the chromosome . 
Thus , one can speculate that either HU cooperates with MukBEF to promote long-range interactions or that MukBEF activity builds on DNA properties generated by HU . 
Surprisingly , the inactivation of HU in E. col has opposite effects as those observed in C. crescentus ( Le et al. , 2013 ) . 
This difference may either result from the presence in C. crescentus of SMC , a class of bacterial condensins different from MukBEF that would not require HU for its activity , or from the presence of another NAP in C. crescentus that would play the role of E. coli HU . 
The absence of Fis is less dramatic than the absence of HU in E. coli because the decrease of long-range DNA communication varies along the chromosome and may depend on local DNA properties resulting from the absence of Fis interacting with its targets . 
Finally , in ter , both HU and Fis are required for optimal contacts up to 280 kb , presumably by counteracting an effect of MatP . 
How MatP may 
Higher-Order Organization of the E. coli Chromosome
Our results reveal two modes of DNA communication in the E. coli chromosome ( Figures 6B and 6C ) . 
First , there is a long-range mode , homogeneous throughout most of the chromosome outside of ter , that depends on both MukBEF and HU action . 
Although the precise interplay between 
HU and MukBEF is unknown , these results provide significant insights into the organization of bacterial chromosomes . 
The effect of the MukBEF complex in E. coli appears to be radically different from that of SMC in B. subtilis ( Marbouty et al. , 2015 ; Wang et al. , 2015 ) . 
Instead of aligning the chromosome arms from a centromere-like locus , MukB promotes DNA contacts in the megabase range within each replication arm . 
How these structurally related proteins promote such different processes remains unknown , but it may involve a similar mechanism . 
As invoked for B. subtilis arm bridging ( Marbouty et al. , 2015 ; Wang et al. , 2015 ) , DNA loop extrusion , a model by which mo-lecular motors actively generate loops ( Alipour and Marko , 
2012 ; Dekker and Mirny , 2016 ) , could also account for the MukBEF-dependent formation of dynamic long-range cis contacts along the E. coli chromosome arms . 
Interestingly , HU appears as a key cofactor of the DNA management process by MukB . 
In hupA hupB mutants , outside of ter , the absence of HU mimics the absence of MukBEF , suggesting a direct link between the activities of the two proteins . 
So far , no such general role for an accessory protein has been uncovered for bacterial condensins . 
The second mode of DNA communication is revealed in the absence of MukBEF and corresponds to enriched homogeneous DNA contacts within 280 kb that could result either from a condensation process bringing together distant loci or from dynamic process ( es ) resulting in frequent and transient collisions between these sites . 
Several factors can cooperate to generate such contacts . 
First , DNA supercoiling could influ-ence the likelihood of distant loci to collide with each other by promoting the sliding of branched plectonemic structures ( Staczek and Higgins , 1998 ) . 
Second , NAPs may play an important role in modulating the ability of a locus to make contacts with flanking sequences . 
The 3C results reported here indicate that both HU and Fis promote higher-order DNA organization . 
Further studies will aim to characterize the precise organization of DNA in the regions that contribute to these contacts . 
members of the R.K. , O.E. , and F.B. laboratories for fruitful discussions and advice . 
We thank the I2BC genomic facility for high-throughput sequencing . 
This research was supported by funding from the European Research Council under the 7th Framework Program ( FP7/2007 -2013 , ERC Grant Agreement 260822 to R.K. ) , by the Agence Nationale pour la Recherche ( HiResBac 
ANR-15-CE11-0023-03 to R.K. , O.E. , and J.M. ) , by the Fondation pour la Re-cherche Médicale ( to V.S.L. ) , and by the Agence Nationale pour la Recherche ( ANR-12-BSV8-0020-01 to F.B. ) . 
AUTHOR CONTRIBUTIONS
Conceptualization , V.S.L. , A.C. , O.E. , F.B. , and R.K. ; Methodology , V.S.L. , A.C. , M.M. , and J.M. ; Investigation , V.S.L. , A.C. , and S.D. ; Writing -- Draft , V.S.L. , A.C. , F.B. , and R.K. ; Writing -- Review & Editing , O.E. , F.B. , and R.K. ; Funding Acquisition , F.B. and R.K. ; Supervision , F.B. and R.K. 
Dillon, S.C., and Dorman, C.J. (2010). Bacterial nucleoid-associated proteins, Niki, H., Yamaichi, Y., and Hiraga, S. (2000). Dynamic organization of chromo- nucleoid structure and gene expression. Nat. Rev. Microbiol. 8, 185–195. somal DNA in Escherichia coli. Genes Dev. 14, 212–223.
Nolivos , S. , and Sherratt , D. ( 2014 ) . 
The bacterial chromosome : architecture and action of bacterial SMC and SMC-like complexes . 
FEMS Microbiol . 
Rev. 38 , 380 -- 392 . 
Nolivos , S. , Upton , A.L. , Badrinarayanan , A. , Müller , J. , Zawadzka , K. , Wiktor , J. , Gill , A. , Arciszewska , L. , Nicolas , E. , and Sherratt , D. ( 2016 ) . 
MatP regulates the coordinated action of topoisomerase IV and MukBEF in chromosome segregation . 
Nat . 
Commun . 
7 , 10466 
Prieto , A.I. , Kahramanoglou , C. , Ali , R.M. , Fraser , G.M. , Seshasayee , A.S.N. , and Luscombe , N.M. ( 2012 ) . 
Genomic analysis of DNA binding and gene regulation by homologous nucleoid-associated proteins IHF and HU in Escherichia coli K12 . 
Nucleic Acids Res . 
40 , 3524 -- 3537 . 
Skoko , D. , Yoo , D. , Bai , H. , Schnurr , B. , Yan , J. , McLeod , S.M. , Marko , J.F. , and Johnson , R.C. ( 2006 ) . 
Mechanism of chromosome compaction and looping by the Escherichia coli nucleoid protein Fis . 
J. Mol . 
Biol . 
364 , 777 -- 798 . 
Staczek , P. , and Higgins , N.P. ( 1998 ) . 
Gyrase and Topo IV modulate chromosome domain size in vivo . 
Mol . 
Microbiol . 
29 , 1435 -- 1448 . 
Tjaden , B. ( 2015 ) . 
De novo assembly of bacterial transcriptomes from RNA-seq data . 
Genome Biol . 
16 , 1 . 
Trussart , M. , Yus , E. , Martinez , S. , Baù , D. , Tahara , Y.O. , Pengo , T. , Widjaja , M. , Kretschmer , S. , Swoger , J. , Djordjevic , S. , et al. ( 2017 ) . 
Defined chromosome structure in the genome-reduced bacterium Mycoplasma pneumoniae . 
Nat . 
Commun . 
8 , 14665 . 
Umbarger , M.A. , Toro , E. , Wright , M.A. , Porreca , G.J. , Baù , D. , Hong , S.-H. , Fero , M.J. , Zhu , L.J. , Marti-Renom , M.A. , McAdams , H.H. , et al. ( 2011 ) . 
The three-dimensional architecture of a bacterial genome and its alteration by ge ¬ 
Val , M.-E. , Marbouty , M. , de Lemos Martins , F. , Kennedy , S.P. , Kemble , H. , Bland , M.J. , Possoz , C. , Koszul , R. , Skovgaard , O. , and Mazel , D. ( 2016 ) . 
A checkpoint control orchestrates the replication of the two chromosomes of Vibrio cholerae . 
Sci . 
Adv. 2 , e1501914 . 
Valens , M. , Penaud , S. , Rossignol , M. , Cornet , F. , and Boccard , F. ( 2004 ) . 
Macrodomain organization of the Escherichia coli chromosome . 
EMBO J. 23 , 4330 -- 4341 . 
METHODS