Logo of embojLink to Publisher's site
EMBO J. 2004 Oct 27; 23(21): 4330–4341.
Published online 2004 Oct 7. doi:  10.1038/sj.emboj.7600434
PMCID: PMC524398

Macrodomain organization of the Escherichia coli chromosome


We have explored the Escherichia coli chromosome architecture by genetic dissection, using a site-specific recombination system that reveals the spatial proximity of distant DNA sites and records interactions. By analysing the percentages of recombination between pairs of sites scattered over the chromosome, we observed that DNA interactions were restricted to within subregions of the chromosome. The results indicated an organization into a ring composed of four macrodomains and two less-structured regions. Two of the macrodomains defined by recombination efficiency are similar to the Ter and Ori macrodomains observed by FISH. Two newly characterized macrodomains flank the Ter macrodomain and two less-structured regions flank the Ori macrodomain. Also the interactions between sister chromatids are rare, suggesting that chromosome segregation quickly follows replication. These results reveal structural features that may be important for chromosome dynamics during the cell cycle.

Keywords: bacterial chromosome, chromatids cohesion, chromosome conformation, macrodomain, site-specific recombination


The large size of genomes compared to the cell dimensions imposes condensation of chromosomes but the understanding of chromosome architecture and spatial organization remains unclear. In eukaryotic cells, chromosomes are not distributed in a disorderly manner in the interphase nucleus; instead, each chromosome occupies a defined, mutually exclusive fraction of the nuclear space referred to as a chromosome territory (Parada and Misteli, 2002). Although models of nonrandom nuclear chromosome organization emerge, the nature of territories and the molecular bases that enhance or restrain chromosome movements during the cell cycle are poorly understood. The bacterial chromosome, named a nucleoid, is a compact structure containing independent supercoiling domains (for review, see Higgins, 1999; Sherratt, 2003). Supercoiling domains were recently shown to be smaller and more labile than previously believed (averaging roughly 10 kb) and barriers delimiting domains being created and destroyed in intervals considerably less than a generation time (Deng et al, 2004; Postow et al, 2004). The circular bacterial chromosome has been shown to be organized with a particular orientation inside the cell that preserves the linear order of genes on the DNA (Teleman et al, 1998; Wu and Errington, 1998; Niki et al, 2000; Viollier et al, 2004). Bacteria replicate their chromosomes at a central replisome relatively stationary through which the DNA template passes as it is replicated (Lemon and Grossman, 2000; Espéli et al, 2003). Replication initiates from a single origin, oriC, and progresses bidirectionally. Soon after duplication, the two origins move from a midcell location to positions in the two halves of the cell, whereas the terminus is found at midcell and quarter points in between (Teleman et al, 1998; Niki et al, 2000). In Escherichia coli, a cis-acting 25-bp palindromic site, migS, plays a crucial role in bipolar positioning of oriC (Yamaichi and Niki, 2004). Because chromosome segregation and chromosome folding occur concurrently in bacteria, regions of the chromosome appear to be partitioned soon after their replication even as the remainder of the chromosome awaits replication. Analysis of how the SMC-like protein MukBEF condensates DNA revealed its likely involvement in organizing the chromosome in a series of loops orthogonal to the cell axis (Case et al, 2004), which might account for the orderly arrangement of the chromosome (Breier and Cozzarelli, 2004). In Bacillus subtilis and E. coli, several phenomena have been reported that suggest a spatial organization of the chromosome in the cell. For example, in B. subtilis, specific regions have been shown to be required for chromosome positioning in sporulating cells (Wu and Errington, 2002; Ben-Yehuda et al, 2003). In addition, the Par-like protein SpoOJ contributes to the ori region organization by binding numerous parS sites scattered over 800 kb (Lee et al, 2003 and references therein). In E. coli, FISH analyses have revealed the existence of two macrodomains, the Ori and Ter macrodomains, defined as large regions of about 1 megabase (Mb) that localize precisely in the cell (Niki et al, 2000). Macrodomains are not static and they have been observed to relocate at specific positions during the cell cycle (Niki et al, 2000). The exact nature of macrodomains and the role they play in cell cycle-dependent localization of the chromosome remain to be defined.

Three different genetic systems have been used to probe in vivo the structure of E. coli or Salmonella nucleoid: the γδ Res system (Higgins et al, 1996; Higgins, 1999), the transposon Tn7 immunity system (DeBoy and Craig, 1996) and the λ Int system (Garcia-Russell et al, 2004). The first two systems did not reveal genome-wide DNA interactions, and frequency of interactions was decreasing proportionally with the increase in the distance between the sites. This feature might result from the intrinsic properties of the systems used: in γδ Res, synapsis of recombining sites results obligatory from plectonemic DNA slithering with the trap of three negative supercoils and this feature imposes one-dimensional interactions. Based on 20 intervals, probing the Salmonella chromosome structure with the λ site-specific recombination system has revealed that the Salmonella chromosome is not completely fluid but rather organized in some way.

To further describe chromosome organization in E. coli and characterize parameters that may control its folding and DNA dynamics, we have developed a genetic system that reports the relative probabilities that different pairs of loci scattered over the genome collide with one another. The existence of macrodomains or of other organized regions should bias the relative probabilities of collision between distant DNA sites in the chromosome. The site-specific recombination system of bacteriophage λ has been extensively characterized (for review, see Azaro and Landy, 2002) and recombination between λ att sites is appropriate to disclose the conformation of the bacterial chromosome. We have constructed several series of strains containing one defined att site at a fixed position and its att partner site inserted at various locations. We subsequently selected strains that were able to support recombination between att sites. The percentages of recombination observed with sites scattered over the chromosome were recorded and reported as a function of the location on the chromosome map. These results reveal a strong bias in the selection of DNA sites, in support of a highly ordered organization of the bacterial chromosome consisting of four macrodomain-like regions and two less-constrained regions.


Probing chromosome organization by measuring long-range DNA interactions

The analysis of long-range DNA interactions relies on the availability of a system that can record these communications. DNA transactions promoted by λ Int (Azaro and Landy, 2002) are appropriate to reveal DNA collisions in the cell for the following reasons: (i) synapsis of att sites occurs by random collision, therefore the frequency of recombination between att sites will indicate their spatial proximity (Crisona et al, 1999); (ii) Int is a versatile recombinase that can recombine sites in direct or inverse orientation, located on the same molecule or on separate molecules (Figure 1); and (iii) Int mediates phage integration in the chromosome by recombination between phage attP and bacterial attB attachment sites (integrative recombination) generating two hybrid sites, attL and attR. Recombination between attL and attR (excisive recombination) requires the additional presence of excisionase Xis (Figure 1A). In the absence of Xis, the configuration attL–attR is blocked and restoration of the initial state attP–attB is possible only in the presence of Int and Xis. In the presence of Xis, Int-mediated recombination between attP and attB is inhibited insuring an efficient attL–attR reaction. Therefore, the sense of recombination is controlled; inverted fragments are stable and consequently easy to score.

Figure 1
Versatility of the site-specific recombination system of phage lambda and recombination scenarios between att sites located on the E. coli chromosome. (A) Integrative and excisive recombination promoted by ‘Int' and ‘Int+Xis'. ...

This study required the development of two types of tools to generate and quantify DNA recombination. First, att cassettes were designed to detect recombination between att sites: an in-frame fusion of attB in lacZ that retained lacZ function was engineered such that recombination with attP leads to lacZ disruption, and consequently to the formation of cells devoid of β-galactosidase activity. The opposite reaction between attL and attR restores lacZ integrity (Figure 1B). Second, to control recombination, the int gene or xis and int genes were cloned downstream of promoter λ PR, which is under control of the thermosensitive λ cI857 repressor (see Materials and methods). Conditions that provide the optimal amount of recombinase and allow the segregation of recombined chromosomes were experimentally determined (Figure 2 and Materials and methods). Temperature and time of induction for recombinase expression were determined such that, for any combination of att sites, the percentage of recombinants never reached 100% (Figure 2A and Materials and methods). A period of 2 h following the induction of recombinase was required to segregate recombinants (Figure 2B).

Figure 2
Long-distance DNA interactions revealed by excisive recombination. (A) Excisive recombination of 6-kb fragment (strain LR2) promoted by different amounts of recombinase. Int+Xis induction was performed for different times at different temperatures. ...

E. coli cells growing exponentially contain several genomes; the detected recombination events might correspond to recombination reactions within a chromosome, between sister chromatids or between duplicated chromosomes. Directly repeated att sites allow one to detect ‘intermolecular' events (Figure 1C), whereas inversely oriented att sites generate viable recombinants only upon intrachromosome events (Figure 1D).

A total of 14 sets of strains were generated to analyse recombination events (Table I and Supplementary Tables I and II). Each set carried an att site at a fixed position (attR17 indicating for example a fixed attR at 17 min on the chromosome map) and partner att sites inserted at various loci (up to 41 according to the set).

Table 1
Percentage of duplications obtained by excisive recombination

Recombination between duplicated chromatids is a rare occurrence

To determine how frequently att sites located on duplicated chromosomes collide, strains that can support lacZ reconstitution by intermolecular recombination between att sites separated by varying distances (from 10 to 562 kb) were subjected to the recombination test (Figure 1). We used five strain sets, each of them with a defined att site. Intermolecular recombination tests were performed in four different regions of the chromosome, with strains containing as fixed att site attR17, attR17off (orientation of attR17 has been inversed), attL29, attL66 and attL83. No recombinants were detected in any strain at a ratio above 3.3% (Table I). The low occurrence of duplications for these intervals did not result from the impossibility of these particular sites to recombine, since deletions provoked by intramolecular recombination events involving these att sites were detected (see Supplementary data). The low recombination rate obtained with att sites present on duplicated molecules indicates that interactions between sister chromatids or duplicated chromosomes are not frequent.

Long-distance DNA interactions are restricted to within subregions of the chromosome

The analysis of long-distance DNA interactions was performed by determining the amount of excisive recombination in 11 strain sets carrying a fixed att site and partner's att sites inserted at various positions (Supplementary Table I). To determine the limits of the competent zones for recombination and select the criteria for recognizing putative domains, the amount of excisive recombination was first recorded for the set of 41 strains carrying attR17 and attL inserted at various positions (Supplementary Table I) with different amounts of recombinase. The percentage of recombinants was reported as a function of the position of the variable attL site (Figure 2C). The comparison of the profiles obtained with the three amounts of recombinase showed clearly that attL sites located in the interval 3′–26′ interacted preferentially with attR17. In the presence of a nonsaturating amount of recombinase (20 min induction at 36°C), only attL sites located in the region between 3′ and 26′ of the chromosome could support inversion with attR17 at a rate above 3%. The recombination rate decreased with the distance between attL and attR sites; however, this decrease was not symmetrical. In the clockwise (CW) direction, the most distal attL sites that gave rise to inversion at high frequency are found at 420 kb, whereas in the counterclockwise (CCW) direction, sites located up to 646 kb interacted with attR17 (Table II and Supplementary Table I). The number of recombinants varied from 57% when the recombination interval was 6-kb long to about 3% with longer intervals within the competent region, while the frequency varied from less than 0.1 to 3% for sites located in other regions of the chromosome. When a nearly saturating amount of recombinase (10 min induction at 37°C) was provided, the same chromosomal region (3′–26′) remained competent for inversion from attR17 (Figure 2C). When a saturating amount of recombinase (20 min induction at 37°C) was provided, higher recombination frequencies were observed in strains carrying attL sites, both within the 3′–26′ interval and outside this interval, with percentages between 10 and 20% for many of the latter strains (Figure 2C).

Table 2
Extent of competent zones for inversion

The low percentage of recombinants in strains carrying attL outside the interval 3′–26′ might reveal a low probability of collisions between attR17 and these sites, or might originate from a biased recovery due to poor viability of strains that underwent a large inversion event. If the low percentage of recombinants resulted from a low collision frequency between partner att sites, increased amounts of recombinase should allow the detection of sporadic collisions between attR17 and attL sites located outside the 3′–26′ interval. On the other hand, effects on viability caused by a large inversion should affect the ratio of blue to white colonies compared to that of recombined DNA. To discriminate between these two possibilities, a real-time quantitative PCR (qPCR) assay was carried out to compare the amount of recombined DNA and the percentage of recombinants. The qPCR assay was used to estimate the proportion of recombined chromosomes obtained with two amounts of recombinase, for five strains having different recombination frequencies (Figure 2D). For strains with attL in the 3′–26′ interval (LR2 and LR59 in Figure 2D), recombined DNA was readily detected in the presence of nonsaturating amounts of recombinase (indicated by (+)), and the percentage of recombined chromosomes by qPCR was similar to that determined by the genetic test. For strains with attL outside the 3′–26′ interval (LR1, LR6 and LR8 in Figure 2D), the amount of recombined DNA was low with nonsaturating amounts of recombinase, but increased in saturating conditions of recombinase (indicated by (+++)); with both amounts of recombinase, the genetic and qPCR determinations gave concordant results (Figure 2D). The qPCR assay indicates that the determination of recombinants was a satisfactory reflection of the probability of collision between att sites and that the low percentage of recombinants with strains carrying attL outside the interval 3′–26′ originated from a low probability of collisions of these sites with attR17.

The results obtained from attR17 with three amounts of recombinase allow one to define limits of competent zones for inversion (approximately between 3′ and 26′; Figure 2C): in nonsaturating conditions of recombinase, the limits correspond to the intervals located between the most distal site recombining above 3% (LR35 and LR161; Supplementary Table I) and the most proximal site recombining below 3% (LR11 and LR21 (set attR17off); Supplementary Table I). In subsequent analyses, the 3% cutoff level of recombination will be applied to define the limits of the competent zones for inversion. In these conditions of nonsaturating recombinase, the percentage of recombinants was reproducible from experiment to experiment (Supplementary Figure 1). In agreement with the prediction that these recombination events resulted from intrachromosomal events, we observed the same pattern of recombination in cells grown in minimal medium (data not shown), which contain less chromosome units per cell (Akerlund et al, 1995).

Chromosomal organizational features limit long-range DNA interactions

The extent of the region colliding with attR17 does not appear to be symmetrical with regard to attR17. To uncover the parameters that determine the range of distant DNA collisions, the limits of the competent zone for inversion from another fixed position in this region, that is, attR22, were identified. The inversion percentages of a series of att intervals were measured and reported as a function of the position of the variable att site with two different amounts of recombinase (Figure 3A). The results indicated that only attL sites located in the region between 6′ and 26′ of the chromosome could support inversion with attR22 in the presence of a nonsaturating amount of recombinase (20 min induction at 36°C). In the CW direction, the most distal attL sites that gave rise to inversion at high frequency are found at 171 kb, whereas in the CCW direction, a site located at 738 kb still collided with attR22. As seen from attR17, increasing the amount of recombinase (10 min induction at 37°C) revealed that the 3′–26′ region was found competent for inversion from attR22 (Figure 3A): in the CCW direction, sites as distant as 895 kb collide with attR22 (LRHK35 in Supplementary Table I), whereas the limit of the competent zone did not change in the CW direction. These results indicate that sites located in the 17′–22′ region interact with sites located up to 3′ in the CCW direction but do not collide with sites located beyond 26′ in the CW direction.

Figure 3
Chromosomal organization features limit long-range DNA interactions. (A) Percentage of recombinants obtained by excisive inversion in two sets of strains carrying one fixed att site (attR17: black; attR22: red) and variously inserted partner att sites ...

To confirm the restrictions to DNA interactions from the 17′–22′ region beyond 26′, the amount of excisive recombination was recorded from a site located beyond 26′, that is, attL29. We selected site attL29, inserted between Ter sites and located outside the non-divisible zones (NDZs) and the Dif activity zone, because this location avoids deleterious effects associated with inversions involving these loci (for review, see Capiaux et al, 2001). The inversion frequencies for strains carrying as fixed site attL29 and attR inserted at various loci were determined and reported as a function of the position of the variable attR site (Figure 3A). All attR insertion sites that gave rise to efficient recombination were located in one subregion of the chromosome, between 26′ and 47′ (Figure 3A and Supplementary Table I). Therefore, in the CCW direction, the most distal site that gave rise to inversion at a significant rate was found at 150 kb, whereas in the CW direction, a site up to 823 kb away still recombined above 3%. This result confirms that the recombination frequency was determined not just by the size of the interval between att sites. Rather, the probability of collision between sites straddling 26′ is low.

Genetic characterization of a macrodomain

Interestingly, the limits of the competent region from attR17 or attR22 (3′ and 26′; Figure 3A) coincide with the limits of the Ori and Ter macrodomains defined by Niki et al (2000). Similarly, the limits of the competent region defined from attL29 (26′ and 47′; Figure 3A) coincide with the limits of the Ter macrodomain (Niki et al, 2000). To further analyse the 3′–26′ region, we determined the limits of the competent zone for inversions from another fixed position in this region, that is, attL7. The inversion percentages of a series of att intervals were measured and reported as a function of the position of the variable att site (Figure 4A). The pattern obtained indicated that the zone competent for inversion from attL7 was wider than that obtained from attR17 or attR22 (1.75 versus 1.08 Mb; Table II). With attL7, whereas the limit in the CW direction was unchanged, around 26′, the limit in the CCW direction is now located at 1183 kb, close to 81′ (Figure 4A), that is, the other limit of the Ori macrodomain. Remarkably, the percentage of recombinants is as great for a site 1183 kb away (10.1%) as for sites 599 kb (10.9%) or 668 kb away (12.8%, LC4R30 versus LC4R32 or LC4R45 in Supplementary Table I), indicating that the large size of the competent inversion zone from attL7 does not result from an increase in reactivity of this site but rather from the possibility to collide with more distant DNA sites.

Figure 4
Genetic characterization of a macrodomain. (A) Percentage of recombinants obtained by excisive inversion in two sets of strains carrying one fixed att site (attR17: blue; attL7: red) and variously inserted partner att sites. The x-axis indicates the nt ...

These results suggested that the interval between 3′ and 26′ contains two different zones, one exemplified by attR17 and attR22 from which collisions with partner sites are restricted to the 3′–26′ part and one exemplified by attL7 from which interactions are possible with sites included both in the Ori macrodomain and in the 3′–26′ region. To confirm this hypothesis and map the limits of these two regions, a number of attL–attR combinations located in the 94′–14′ were analysed (Figure 4C); collisions between attL inserted at various locations (attL8, attL12 and attL14) were recorded with different attR sites (attR94, attR97 and attR6). Whereas inversions between attL8 or attL12 were readily detected with attR6, attR97 or attR94 (Figure 4C), attL14 interacted only with attR6 and not with attR94 or attR97 (located in the Ori macrodomain; Supplementary Table I). These results confirmed the existence of two regions in the 3′–26′ interval (Figure 4D): DNA sites located in the 3′–12′ interval can collide with sites present in the Ori macrodomain and with sites in the 3′–26′ interval. In contrast, sites in the 14′–26′ interval can interact only with sites present in the 3′–26′ interval. Altogether, these results indicated that collisions between distant DNA sites are restricted to different subregions of the chromosome, which we call hereafter as macrodomains. From the interval 14′–26′, defined hereafter as Right macrodomain, the probability to collide with sites outside the 3′–26′ interval is low. From the 3′–12′ interval, collisions with sites present in both Ori and Right macrodomains are high, suggesting the existence of a less-structured region that can interact with both flanking macrodomains. The E. coli chromosome thus has a defined conformation in the cell.

Four macrodomains and two less-structured regions in the E. coli chromosome

To further probe the conformation of the E. coli chromosome and define other macrodomains, we analysed the inversion patterns from att sites inserted at three different positions in other regions of the chromosome (53′, 70′ and 87′ in Figure 5A). As described above for patterns from attL7, attR17, attR22 or attL29, we observed that long-distance interactions were restricted to subregions of the chromosome from each of the reference att sites. The sizes of the competent zones for inversion varied from 0.92 to 2.15 Mb (Table II) and the different patterns observed supported the existence of four macrodomains and two less-structured regions in the E. coli chromosome.

Figure 5
Genetic characterization of macrodomains. (A) Percentage of recombinants obtained by excisive inversion in different sets of strains carrying one fixed att site and variously inserted partner att sites. The x-axis indicates the nt coordinates of the chromosome. ...

The pattern of inversions obtained with the set of strains carrying attL70 was reminiscent of that obtained with the set of strains containing attL7: DNA collisions occurred between attL70 and sites located between 51′ and 97′, that is, in both the Ori macrodomain and the interval between Ter and Ori macrodomains (Figure 5A). No interactions were detected with sites present in the 1′–44′ interval, indicating that the CW limit of the zone competent for inversion from attL70 coincides with the CW limit of the Ori macrodomain whereas the CCW limit is close to 47′, the CW limit of the Ter macrodomain defined above (Table II). These results indicated that attL70 is located in a region that is analogous to the 3′–12′ less-structured region and that can interact with both the Ori macrodomain and the interval between Ori and Ter macrodomains (Figure 5A and see below).

The presence and the limits of two less-structured regions flanking the Ori macrodomain were confirmed by the inversion profile obtained with the set of strains containing as fixed att site attL87. From attL87 that is located near the middle of the Ori macrodomain, interactions were found with attR sites located in the interval 62′–12′ (Figure 5A), that is, in the Ori macrodomain (from 81′ to 1′) and in the two less-structured flanking regions (from 62′ to 81′ and from 3′ to 12′, respectively) as predicted from Figure 4, profile attL7 and profile attL70. These results also indirectly indicated the existence of a Left macrodomain: DNA sites located in the 47′–62′ interval interact with attL70 but not with attL87. The existence of the Left macrodomain was confirmed by the inversion profile obtained with a set of strains containing as fixed att site attR53. As predicted, the CCW limit was found between 41′ and 47′, that is, coincides with the CW limit of the Ter macrodomain. In the CW direction, attR53 collides with sites as distant as 911 kb, setting the CW limit close to 75′. As observed with attR22 (Figure 3), an increase of the amount of recombinase allowed interaction with sites located up to 76′ but not with sites present beyond 80′, that is, in the Ori macrodomain (data not shown). Altogether, these results indicate that the E. coli chromosome contains two less-structured regions flanking the Ori macrodomain and two structured regions, the Left and Right macrodomains, flanking the Ter macrodomain (Figure 5B).

Integrative inversions reveal the same long-distance DNA interactions

Biochemical analyses have revealed that integrative (attB–attP) and excisive (attL–attR) recombinations differ in several ways. First, integrative recombination is dependent on the level of attP supercoiling, whereas excisive recombination does not require supercoiled att sites (Azaro and Landy, 2002). Second, topological analyses have revealed that the two reactions promote the formation of products with different topologies (Crisona et al, 1999). Third, we have found in this assay that excisive recombination is far more efficient than integrative recombination (see Materials and methods). To determine whether long-range DNA collisions revealed by integrative recombination were similar to those uncovered by excisive recombination, the pattern of interactions deduced from integrative recombination frequencies was performed (data not shown). The amount of integrative inversion was recorded for the set of 33 strains carrying attB17 inserted in-frame in lacZ and attP inserted at various loci. The inversion frequencies for each of these clones were determined (Supplementary Table II); attP sites able to support inversion above 3% were found in the 3′–26′ interval. Because the two types of reactions rely on different processes to perform synapsis of att sites, the findings that zones competent for inversion from att17 were similar for the two types of reactions indicated that the assays reflected the probability of collisions between distant DNA sites.


Organization of the chromosome into macrodomains

We have shown that the E. coli chromosome appears to be subdivided in large regions referred to as macrodomains in which DNA interactions occurred preferentially. DNA interactions between these different macrodomains are highly restricted. The principle of the method was to select many pairs of inversely oriented att sites that can give rise to inversions of the intervening fragment, to provide a limiting amount of recombinase that will not saturate the recombination reaction between any combination of sites and to measure the extent of recombination. These three steps have potential limitations for this study and are discussed below.

By using a system of programmed genetic DNA transactions, we were able to characterize 211 combinations of att sites that support recombination. This method is different from other approaches that have been used previously (Rebollo et al, 1988; Segall et al, 1988). In these systems, viable, detrimental or forbidden inversions were revealed but there was no direct measure of spatial chromosome organization. Our experimental design eliminated combinations of sites that gave rise to nonviable or forbidden inversions since we selected only clones able to support recombination. Among clones competent for inversions, the measure of recombination rates allowed the identification of macrodomains and does not preclude the existence of nonviable or forbidden inversions in macrodomains.

The second critical feature of the method was the amount of recombinase that was provided; in pilot tests, we set up conditions that provide the optimal amount of recombinase taking into account that an excessive amount of recombinase, saturating some combinations of sites, will diminish the differential of interactions (Figure 2).

Finally, by using a qPCR assay, we unambiguously showed that the percentage of recombinant clones reflected satisfactorily the fraction of DNA that recombined, indicating that no bias was introduced in the assay and that the measure of recombinant clones was appropriate to disclose the conformation of the bacterial chromosome (Figure 2).

For all profiles, interactions were restricted to subregions of the chromosome that included the reference att site and regions contiguous to this reference site. In many cases, we observed a decrease in the recombination frequency with distance. In nonsaturating conditions of recombinase, we did not detect any preferential interactions with a noncontiguous region or even with a region diametrically opposed relative to the oriCdif axis or to the Ori–Ter macrodomain axis. At first sight, these results appear consistent with a model for in vivo synapis assembly involving tracking from one att site rather than relying on random collision of att sites. However, although this model is attractive, it can be ruled out for several reasons. First, in a tracking model, the level of recombination should decrease with distance, but this is not always the case. For example, the percentage of recombinants is similar for sites 599, 668 or 1183 kb away from attL7. Second, we have observed that in cells containing specific rearranged chromosomes, preferential interactions were detected with noncontiguous regions (Valens et al, in preparation). Instead of resulting from a tracking mechanism, the observed small decrease in recombination with distance, within macrodomains, might originate from the condensed ring organization of the chromosome (see below) because increasing the distance between the att sites will diminish the probability of collision of these sites in space.

Limited chromatid interactions

With the use of directly repeated sites, we could monitor duplication that occurs necessarily by intermolecular recombination. These results show that ability to support intermolecular recombination is rare in the chromosome. Fluorescence approaches gave contradictory results about the extent of cohesion between the two oriC-proximal halves of the E. coli chromosome after replication: Sunako et al (2001) observed a long period of cohesion, whereas in two different reports cohesion was estimated to be very short (Li et al, 2002; Lau et al, 2003). Our results of interchromatid recombination support the extrusion capture model (Lemon and Grossman, 2001).

Macrodomains and spatial organization of the chromosome

Our data clearly demonstrated that, in E. coli, sites do not interact equally with the different parts of the chromosome and the great number of att pair's combinations analysed allowed us to map macrodomains and less-constrained regions. Our results must be compared with those obtained using FISH by Niki et al (2000). We were able to predict four macrodomains and two less-structured regions (Figure 5B), whereas FISH disclosed only two macrodomains. Ori and Ter macrodomains defined here remarkably coincide with Ori and Ter FISH macrodomains respectively, indicating that the structuring mechanism revealed by the genetic approach is stable enough to resist cell biology techniques.

The observations reported here indicate that the left and right intervals between Ori and Ter macrodomains are not linker DNA, but that a part of these exists also as organized Left and Right macrodomains, respectively. The absence of communications between an att site inserted at 17′ and sites diametrically opposed relative to an axis oriCdif suggests a strict spatial localization of the Right and Left macrodomains. From attR17, collision occurs with sites distant of 650 kb, but interactions with the opposite replichore are very limited, although these regions were seen in FISH experiments along the cell's long axis (Niki et al, 2000). It is also interesting to note that although the region present between the Right and Ori macrodomains appeared to be less structured because it can interact with both flanking macrodomains, it does not interact with other parts of the chromosome, indicating limited flexibility of this region.

The biased interactions detected in this study highlight two different phenomena: condensation of the DNA molecule and sequestration of macrodomains. It is remarkable that within a macrodomain, the condensation of the DNA molecule allows the interaction of sites more than 500 kb away. On the other hand, macrodomain sequestration inhibits collision of sites situated in different macrodomains. Altogether, these results indicate that the chromosome is organized as a ring composed of four macrodomains and two less-structured regions with flexibility limited to the flanking macrodomains (Figure 6A).

Figure 6
A model for chromosome organization in E. coli. (A) The chromosome is organized as a ring composed of four macrodomains (Ori, Ter, Right and Left) and two less-structured regions (NS) with flexibility limited to the flanking macrodomains. (B) Two models ...

Two major processes could account for this chromosome structuring (Figure 6). First, macrodomains may be separated by DNA structures that disfavour interactions (Figure 6B, top). Alternatively, it is possible that the different macrodomains are insulated in different parts of the cell and that the specific localization maintains DNA separation; sequestration of large regions would imply the presence of determinants present in the macrodomain (Figure 6B, bottom). Three different arguments indicate that the second scenario is more likely to apply to the E. coli chromosome (Figure 6B, bottom). First, since synapsis of att sites occurs by random collision, it is hard to conceive from the first hypothesis how structures on the DNA molecule could inhibit collisions between distant DNA sites. Second, the condensation of a large region harbouring distant DNA binding sites has been described in B. subtilis (Lee et al, 2003) and it is conceivable that different proteins may help localize different macrodomains of the E. coli chromosome. Third, the structuring processes did not impede inversions from attL7 or attL70 with sites located in both flanking macrodomains implying the absence of fixed barriers. Further experiments will be required to characterize the role and the origin of macrodomains in the cell. A possible reason for this organization is to orchestrate chromosome movements that occurred during the cell cycle. As observed in the B period by Niki et al (2000), repositioning of the chromosome seems to be required at specific step(s) in the cell cycle. Orchestrating the movement of a limited number of ‘organizing' proteins might be easier than the manipulation of 4.6 Mb of DNA. The characterization of the E. coli chromosome organization reveals for the first time the structuring of a complete chromosome into large macrodomains. This study will enable the search for determinants that are responsible for this structuring and may uncover processes that control spatial organization of chromosomes in living cells.

Materials and methods

Strains and plasmids

E. coli K12 strains are all derivatives of MG1655. Standard transformation and transduction procedures used were as described before (Espéli et al, 2001). Constructions of strains and plasmids are described in Supplementary data. Plasmids and strains with relevant genotypes are described in Table III.

Table 3
Strains and plasmids

Selection of strains supporting excisive and integrative recombination

To select for strains that support excisive recombination, 13 sets of strains carrying one fixed att site and the partner att site inserted at various loci were constructed. For four of the sets carrying one fixed att site and variously inserted mobile partner sites (sets of strains carrying attR17, attL29, attR53, attL82), a large number of clones were individually transformed with pTSA29-CXI, a plasmid expressing int and xis under the control of cI857 repressor; in the absence of repressor in the cell, introduction of this plasmid by transformation resulted in transient synthesis of Int and Xis and allowed the detection of strains supporting recombination. The location and orientation of mobile att sites were then determined by sequence analysis. For the other nine sets, specific combinations of att sites were constructed using phage P1 transduction (see Supplementary data). To select for strains that support integrative recombination, several hundred derivatives of FBG140 carrying attB17 and variously inserted attP sites were transformed with pTSA29-CXI-AK, a plasmid expressing int under the control of cI857 repressor; incubation at 37°C resulted in synthesis of Int and allowed the detection of strains supporting recombination.

Recombination assays

The recombination test (Figure 2) includes a transient incubation at a higher temperature to inactivate cI857 repressor and to promote a pulse of Int (integrative recombination) or Int and Xis (excisive recombination) synthesis, respectively. Because several lines of evidence indicated that reactions of excisive inversion were more efficient than those of integrative inversion, the conditions for Int and Int+Xis induction were different. For excisive inversions and duplications, inductions were performed at 36°C during 20 min (unless otherwise stated), conditions that provide a nonsaturating amount of recombinase and give highly reproducible recombination rates (Supplementary Figure 1). Overnight cultures were diluted 100-fold, grown to an OD600 of 0.3 in L medium and submitted to heat shock. A control sample was kept at 30°C. Cultures were kept at 30°C during 120 min before plating cells on L medium containing ampicillin (50 μg/ml) and X-Gal (80 μg/ml). Between 200 and 300 colonies were counted to estimate the recombination rate. For the integrative recombination assay, we noticed that strains carrying different constructions expressing int always showed a background level of inversion activity (data not shown). Because of this background level of recombination, clones selected for their ability to support fragment inversion between inversely oriented attP and attB sites were isolated, streaked twice at 30°C and individual blue clones were inoculated in 10 ml culture of Lennox medium. Because integrative recombination was less efficient (data not shown), the induction was performed at 42°C for 10 min when cultures reach OD600 0.3 and a 2 h incubation time was allowed for the segregation of recombinant chromosomes before plating.

Quantitative PCR

For the quantitative analysis of recombination, fluorescence real-time PCR was performed using dsDNA dye SYBR Green I (Roche Diagnostics). Total DNA was extracted immediately after recombinase induction. Primer pairs were 5′-TTACGCGCCGGAGAAAACCG-3′ and 5′-TCAACCACCGCACGATAGAG-3′ for lacZ; 5′-CGACTACCTTGGTGATCTCG-3′ and 5′-CGACATTGATCTGGCTATCTTG-3′ for aadA. The amount of recombined lacZ DNA detected was normalized with the control aadA values.

Identification of the insertion points of att sites

Insertion points of the different minitransposons were determined by direct sequencing of chromosomal DNA using the Big dye terminator Version 3 kit (ABI Prism). Each sequencing reaction contained 10 μg of chromosomal DNA, 40 pmol of primers in a volume of 20 μl and was submitted to 99 PCR cycles (95°C 30 s, 55°C 30 s, 60°C 4 min). Primers used to determine att site insertions were 5′-GCAACGAACAGGTCACTATCAGTC-3′ or 5′-TTCCCAGTCACGACGTTGTAAA-3′ (attL), 5′-ATGTTCTAGAGGATCTGTGA-3′ (attR) and 5′-TGATGCCTCTAGCACGCGTA-3′ (attP). Sequencing reactions were analysed on a 3100 Genetic analyser (Applied Biosystems).

Supplementary Material

Supplementary Figure 1

Supplementary Tables I and II

Supplementary data


We thank Alexandra Gruss, Bénédicte Michel and Linda Sperling for critical reading of the manuscript, Martial Marbouty and Corentin Laulier for performing some inversion experiments, Bob Weisberg for the kind gift of strains and phages, and Laurent Moulin and Olivier Espéli for helpful discussions. This work was supported by the Centre National de la Recherche Scientifique, the Association pour la Recherche sur le Cancer and the Program Microbiologie PRMMIP 295007.


  • Akerlund T, Nordstrom K, Bernander R (1995) Analysis of cell size and DNA content in exponentially growing and stationary-phase batch cultures of Escherichia coli. J Bacteriol 177: 6791–6797 [PMC free article] [PubMed]
  • Azaro MA, Landy A (2002) Integrase and the Int family. In Mobile DNA II, Craig NL, Craigie R, Gellert M, Lambowitz AM (eds) pp 118–148. Washington, DC: ASM Press
  • Ben-Yehuda S, Rudner DZ, Losick R (2003) RacA, a bacterial protein that anchors chromosomes to the cell poles. Science 299: 532–536 [PubMed]
  • Breier AM, Cozzarelli NR (2004) Linear ordering and dynamic segregation of the bacterial chromosome. Proc Natl Acad Sci USA 101: 9175–9176 [PMC free article] [PubMed]
  • Capiaux H, Cornet F, Corre J, Guijo MI, Perals K, Rebollo JE, Louarn JM (2001) Polarization of the Escherichia coli chromosome. A view from the terminus. Biochimie 83: 161–170 [PubMed]
  • Case RB, Chang YP, Smith SB, Gore J, Cozzarelli NR, Bustamante C (2004) The bacterial condensin MukBEF compacts DNA into a repetitive, stable structure. Science 305: 222–227 [PubMed]
  • Crisona NJ, Weinberg RL, Peter BJ, Sumners DW, Cozzarelli NR (1999) The topological mechanism of phage lambda integrase. J Mol Biol 289: 747–775 [PubMed]
  • DeBoy RT, Craig NL (1996) Tn7 transposition as a probe of cis interactions between widely separated (190 kilobases apart) DNA sites in the Escherichia coli chromosome. J Bacteriol 178: 6184–6191 [PMC free article] [PubMed]
  • Deng S, Stein RA, Higgins NP (2004) Transcription-induced barriers to supercoil diffusion in the Salmonella typhimurium chromosome. Proc Natl Acad Sci USA 101: 3398–3403 [PMC free article] [PubMed]
  • Espéli O, Levine C, Hassing H, Marians KJ (2003) Temporal regulation of topoisomerase IV activity in E. coli. Mol Cell 11: 189–201 [PubMed]
  • Espéli O, Moulin L, Boccard F (2001) Transcription attenuation associated with bacterial repetitive extragenic BIME elements. J Mol Biol 314: 375–386 [PubMed]
  • Garcia-Russell N, Harmon TG, Le TQ, Amaladas NH, Mathewson RD, Segall AM (2004) Unequal access of chromosomal regions to each other in Salmonella: probing chromosome structure with phage lambda integrase-mediated long-range rearrangements. Mol Microbiol 52: 329–344 [PubMed]
  • Higgins NP (1999) DNA supercoiling and its consequences for chromosome structure and function. In Organization of the Procaryotic Genome, Charlebois RL (ed) pp 189–202. Washington, DC: ASM Press
  • Higgins NP, Yang X, Fu Q, Roth JR (1996) Surveying a supercoil domain by using the gamma delta resolution system in Salmonella typhimurium. J Bacteriol 178: 2825–2835 [PMC free article] [PubMed]
  • Lau IF, Filipe SR, Soballe B, Okstad OA, Barre FX, Sherratt DJ (2003) Spatial and temporal organization of replicating Escherichia coli chromosomes. Mol Microbiol 49: 731–743 [PubMed]
  • Lee PS, Lin DC, Moriya S, Grossman AD (2003) Effects of the chromosome partitioning protein Spo0J (ParB) on oriC positioning and replication initiation in Bacillus subtilis. J Bacteriol 185: 1326–1337 [PMC free article] [PubMed]
  • Lemon KP, Grossman AD (2000) Movement of replicating DNA through a stationary replisome. Mol Cell 6: 1321–1330 [PubMed]
  • Lemon KP, Grossman AD (2001) The extrusion-capture model for chromosome partitioning in bacteria. Genes Dev 15: 2031–2041 [PubMed]
  • Li Y, Sergueev K, Austin S (2002) The segregation of the Escherichia coli origin and terminus of replication. Mol Microbiol 46: 985–996 [PubMed]
  • Nash HA (1996) Site specific recombination: integration, excision, resolution, and inversion of defined DNA segments. In Escherichia coli and Salmonella, Neidhardt FC (ed) pp 2363–2376. Washington, DC: ASM Press
  • Niki H, Yamaichi Y, Hiraga S (2000) Dynamic organization of chromosomal DNA in Escherichia coli. Genes Dev 14: 212–223 [PMC free article] [PubMed]
  • Parada L, Misteli T (2002) Chromosome positioning in the interphase nucleus. Trends Cell Biol 12: 425–432 [PubMed]
  • Postow L, Hardy CD, Arsuaga J, Cozzarelli NR (2004) Topological domain structure of the Escherichia coli chromosome. Genes Dev 18: 1766–1779 [PMC free article] [PubMed]
  • Rebollo JE, Francois V, Louarn JM (1988) Detection and possible role of two large nondivisible zones on the Escherichia coli chromosome. Proc Natl Acad Sci USA 85: 9391–9395 [PMC free article] [PubMed]
  • Rossignol M, Basset A, Espeli O, Boccard F (2001) NKBOR, a mini-Tn10-based transposon for random insertion in the chromosome of Gram-negative bacteria and the rapid recovery of sequences flanking the insertion sites in Escherichia coli. Res Microbiol 152: 481–485 [PubMed]
  • Segall A, Mahan MJ, Roth JR (1988) Rearrangement of the bacterial chromosome: forbidden inversions. Science 241: 1314–1318 [PubMed]
  • Sherratt DJ (2003) Bacterial chromosome dynamics. Science 301: 780–785 [PubMed]
  • Sunako Y, Onogi T, Hiraga S (2001) Sister chromosome cohesion of Escherichia coli. Mol Microbiol 42: 1233–1241 [PubMed]
  • Teleman AA, Graumann PL, Lin DC, Grossman AD, Losick R (1998) Chromosome arrangement within a bacterium. Curr Biol 8: 1102–1109 [PubMed]
  • Viollier PH, Thanbichler M, McGrath PT, West L, Meewan M, McAdams HH, Shapiro L (2004) Rapid and sequential movement of individual chromosomal loci to specific subcellular locations during bacterial DNA replication. Proc Natl Acad Sci USA 101: 9257–9262 [PMC free article] [PubMed]
  • Wu LJ, Errington J (1998) Use of asymmetric cell division and spoIIIE mutants to probe chromosome orientation and organization in Bacillus subtilis. Mol Microbiol 27: 777–786 [PubMed]
  • Wu LJ, Errington J (2002) A large dispersed chromosomal region required for chromosome segregation in sporulating cells of Bacillus subtilis. EMBO J 21: 4001–4011 [PMC free article] [PubMed]
  • Yamaichi Y, Niki H (2004) migS, a cis-acting site that affects bipolar positioning of oriC on the Escherichia coli chromosome. EMBO J 23: 221–233 [PMC free article] [PubMed]

Articles from The EMBO Journal are provided here courtesy of The European Molecular Biology Organization
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...