Architectural organization in E. coli nucleoid

Abstract

In contrast to organized hierarchical structure of eukaryotic chromosome, bacterial chromosomes are believed not to have such structures. The genomes of bacteria are condensed into a compact structure called the nucleoid. Among many architectural, histone-like proteins which associate with the chromosomal DNA is HU which is implicated in folding DNA into a compact structure by bending and wrapping DNA. Unlike the majority of other histone-like proteins, HU is highly conserved in eubacteria and unique in its ability to bind RNA. Furthermore, an HU mutation profoundly alters the cellular transcription profile and consequently has global effects on physiology and the lifestyle of E. coli. Here we provide a short overview of the mechanisms by which the nucleoid is organized into different topological domains. We propose that HU is a major player in creating domain-specific superhelicities and thus influences the transcription profile from the constituent promoters. This article is part of a Special Issue entitled: Chromatin in time and space.

1. Introduction

In eukaryotic organisms, DNA is compacted by histones which organize DNA in a hierarchical process into the higher order structure that is chromatin. In contrast, the bacterial chromosome is condensed to a form called the nucleoid. The nucleoid, although compacted, is perceived to lack any higher-order chromosomal organization and represents a comparatively open structure, accessible to DNA-binding proteins, RNA and DNA polymerases throughout the cell cycle [1,2]. The mechanisms used by prokaryotes to organize their chromosomal DNA are poorly understood. In this article, we summarize our view of the structural organization of the E. coli chromosome, and the role that the nucleoid protein HU plays in it.

2. Nucleoid composition

E. coli is a gram negative rod-shaped bacterium (Fig. 1A). Its nucleoid consists of a single, 4.6 kb long circular DNA molecule, several RNA, and over 200 DNA-binding proteins [3-5], among which is a set of so-called histone-like proteins that include HU, H-NS, IHF and FIS [6,7]. The 4.6 kb DNA is compacted approximately 1000 fold in the nucleoid by the nucleoid proteins and supercoiling [8] (Fig. 2A). Nucleoid proteins are referred-to as histone-like not because of sequence or structural similarities to eukaryotic histones, but because of their presumed role in nucleoid compaction which is comparable to that of eukaryotic histones. HU, H-NS, IHF and FIS are abundant in bacterial cells, present at concentrations up to or even exceeding 10 μM depending on growth conditions [9,10]. In addition to their role in nucleoid organization, these proteins function directly or indirectly to control expression of a large number of genes that are essential for cell viability, such as those involved in protein synthesis and metabolism. They also regulate genes whose expression varies in response to different environmental stimuli, such as those that mediate responses to changes in temperature, pH, osmolarity as well as genes affecting virulence [11-15]. HU, H-NS, IHF and FIS bind to DNA to AT-rich DNA targets and shape the local structure of the nucleoid by bending DNA upon binding [16-19]. Independent studies suggest that, in contrast to eukaryotic histones, the binding of bacterial histone-like proteins is biased towards non-coding regions of the genome [20-22]. A bioinformatics search of the E. coli genome for DNA sequences that resemble the binding sites for 55 proteins including FIS, H-NS and IHF identified over 10,000 protein binding sites [20]. Interestingly, although less than 10% of the E. coli genome is non-coding, more than 23% of the predicted targets for most proteins were in the non-coding DNA [20]. Other studies directly measured the binding of histone-like proteins FIS, H-NS and IHF across the entire E. coli genome using a ChIP-chip analysis [21,23]. In these experiments, although some binding sites for nucleoid proteins were found in coding regions of the genome that are transcribed, the majority of sites were also located in the non-coding DNA segments [21].

Fig. 1.

Scanning electron microscopic pictures of E. coli cells. (A) Wild type strain MG1655; (B) HU³⁸⁴² strain SK3842.

Fig. 2.

Thin section electron micrographs of E. coli cells. (A) Wild type strain MG1655;(B) HU³⁸⁴² strain of SK3842.

Using a bioinformatics approach, the overall binding pattern of cellular histone-like proteins to the chromosomal DNA was estimated to be approximately one protein per 100 base pairs of DNA [24]. However, due to moderate affinity of these proteins for DNA in general, the resulting complexes may not be very stable [9]. Thus, the apparent lack of a hierarchy of nucleoid organization in bacteria may be attributed to a low stability of histone-like protein-DNA complexes as well as to the dynamic nature of the bacterial chromosome [25]. In the presumed absence of any hierarchical higher-order structure that imposes restrictions on the accessibility of bacterial promoters, gene expression in bacteria, unlike in eukaryotes, is thought to be regulated by operon-specific transcription factors and easily accessible DNA control elements adjacent to the promoters [26].

3. DNA organization

Images by electron microscopy showed isolated chromosome of E. coli cells lysed under gentle conditions as a rosette with a compact central core from which about 100 hundred DNA loops radiate [27-29] (Fig. 3). This “rosette” model was the first indication of any higher-order organization of the bacterial chromosome. It was proposed that a central core comprised segments of the bacterial chromosome that are held together by multiple interactions between DNA, proteins and some uncharacterized nucleoid RNA. This results in multiple topologically independent DNA loops or “domains of supercoiling” which radiate from the core have independent superhelicity. This was further supported by the finding that a single strand nick in one of the loops eliminates the superhelicity only in that domain, leaving the rest of the domains unaffected [30]. The existence of domains of supercoiling in the chromosome has been also confirmed by several additional studies [31-33].

Fig. 3.

Electron micrograph of E. coli chromosome after cell lysis (adapted from ref.[28]).

4. The HU protein

There are several characteristics that make HU, unlike other nucleoid proteins, unique. They include a high degree of conservation of HU protein in eubacteria, ability of HU to bind RNA [34,35] and effects of a HU mutation on DNA superhelicity [36]. Most bacterial species do not encode homologs of all of the nucleoid proteins. HU proteins, however, appear to be encoded by all eubacteria, and some bacterial species encode more than one homolog ofHU. HU homologs are also found in organelles such as chloroplasts, where they appear to serve similar functions in DNA organization [37,38].

In E. coli, HU is a basic protein composed of two very similar, but nonidentical subunits, HUα and HUβ [39-41]. In vitro, HU binds with a relatively low affinity to linear DNA fragments with a saturation density of one dimer per 9 bp regardless of sequence [42], but it binds more avidly to negatively supercoiled than to relaxed DNA [43]. It has also been shown that HU binds much more tightly to specific DNA structures such as nicked DNA, junction DNA, bent DNA, single-stranded and double-stranded DNA forks, or DNA 3’-overhangs [44,45]. In addition to DNA, HU also binds to poly(rU) homopolymer (unpublished data quoted in ref. [35]), rpoS mRNA [34] and DsrA sRNA [35]. Null mutations in either of the HU-encoding hupA and hupB genes have very little phenotype. Presumably, each of the HU subunits can substitute for the other for visible phenotypes, requiring the double mutant to display an HU- phenotype. A strain lacking both of the HU proteins shows a decodensed nucleoid, a cold-sensitive phenotype, increased sensitivity to UV light, instability of F plasmids, and relaxation of reporter plasmid DNA superhelicity [46-54]. More importantly, deficiency of HU causes poor growth that is readily suppressed by mutations in gyrB, suggesting an involvement of HU in DNA superhelicity [55]. HU appears to be essential in Bacillus subtilis and other gram-positive organisms [46,50,56-62]. This is likely due, at least in part, to a lack of redundancy of other nucleoid-associated proteins in gram-positive organisms.

We have shown that a gain-of-function HU mutant of E. coli, HUα³⁸⁴², has peculiar characteristics with a profound change in the gene expression profile [63]. A large number of genes are down-regulated and a significant number of genes that are normally silent in the wild type cells are derepressed in the mutant. The mutant shows a rod-to-coccoid morphological conversion (Fig. 1B) as well as transformation of the chromosome into amore condensed one (Fig. 2B). The HUα³⁸⁴² mutant contains two changes in the amino acid sequence–E38K and V42L–in the HUα subunit. In vitro transcription studies of representative affected genes demonstrated that the altered gene expression profile is because of an altered transcription profile [64].

5. DNA supercoiling

How does alteration in HU change the global transcription profile of the cell? HU is known to generate superhelicity in plasmid DNA in vitro [65]. Together with Topoisomerase I, HU can convert a relaxed DNA molecule into a negatively supercoiled DNA. The proposed mechanism is shown in Fig. 4. This mechanism of generating superhelicity in DNA by HU is supported by x-ray structure determination of the wild type E. coli HU protein [66]. The results showthat four HU dimmers associate to forman octameric crystal unit which is capable of forming multimers with both left-handed and right-handed spirals (Fig. 5). In this model, HU exposes the DNA binding β-loops in the multimer which interact with the minor grooves of DNA, resulting in a left- or right-handed toroidal DNA. In the in vitro supercoiling assay starting with relaxed DNA, as mentioned before, wild type HU generates negative supercoil. In the same assay, the mutant HU generates positive supercoils [36] (Fig. 6). We suggest that HU exists in two multimeric conformations: negative and positive spirals in equilibrium. In the wild type, the equilibrium favors the negative spirals whereas in the mutant protein equilibrium shifts toward right-hand spirals. By this mechanism, the mutant HU affects global transcription profile. It should be noted that in the structure generated by wrapping DNA around an HU multimer, it is not known whether each superhelical DNA turn is generated around an HU octamer or the DNA turn is continuous around multimeric HU (Fig. 7). We proposed that HU-induced negative or positive superhelicity in DNA regulates the promoters. We have confirmed this idea by demonstrating that negatively supercoiled DNA template isolated from wild type E. coli is capable of transcribing in vitro a promoter whose transcription is facilitated by negative supercoiling, but not one which is active only in positively supercoiled DNA [36]. The opposite is true for positively supercoiled DNA templates isolated from the HU mutant host; the former promoter is silent and the later is very active in in vitro transcription in conformity with the in vivo results. Thus HU induced superhelicity does control the adequacy of the promoters for transcription.

Fig. 4.

Generation of supercoil in DNA by HU and Topoisomerase I (adapted from ref.[65]).

Fig. 5.

X-ray crystal structure of HU.Left, structures of HUαβ; middle, models of HUαβ-DNA complex; right, models of DNA without HU. Top, left handed spiral of HU; bottom, right handed spiral of HU.

Fig. 6.

Identification of DNA supercoiling. A relaxed DNA circle was treated with wild type HU followed by Topoisomerase I and then subjected to 2-D gel electrophoresis (adapted from ref. [65]). Left panel, no treatment; middle panel, treated with wild type HU and Topoisomerase I; right panel, treatment with HU³⁸⁴² and Topoisomerase I. As discussed in the text, wild type HU generates negative superhelicity and HU³⁸⁴² generates positive superhelicity in DNA.

Fig. 7.

DNA superhelical wrapping around A) HU octamer or B) HU multimer as discussed in text.

We propose that the chromosome is divided into segments, each segment having characteristic superhelicity–both qualitatively and quantitatively (Fig. 8A). These segments are functionally equivalent to topologically independent DNA domains in the “rosette” model. Each segment with a given supercoiling state guides the fate of constituent promoters. The mutant HU changes the supercoiling patterns, and thus the expression of the constituent promoters. This model can be tested by determining the in vivo chromosomal supercoiling map and correlating with the promoter activity map in the wild type and HU mutant cells.

Fig. 8.

Models of E. coli chromosomal DNA structure as discussed in text. (A) Independent domains of relaxed or super helical DNA. As shown the superhelicity could be negative or positive, plectonemic or toroidal. The toroid form may have multimeric HU (shown in purple) around which DNA wraps. Shown in red are the boundary elements located between topologically independent DNA segments. The length of each segment must vary. (B) The boundary elements shown in A coalesce with each other to generate a “rosette-like” structure. (C) In the alternative “membrane” model of chromosome structure discussed in text, the boundary elements attach to membranes making DNA segments topologically independent. As also described in text, cell wall lysis ruptures the continuous membranes resulting in the fusion of boundary elementsbound membrane fragments as in B. In this model, the rosette structure is an in vitro artifact.

6. Boundary elements

The nature of the boundary elements that would keep each DNA segment topologically independent and how they work needs further study. We have been interested in establishing the biochemical basis of the domain organization and what constitutes the boundaries that keep the domains topologically independent. In the “rosette” model, some unidentified DNA sequences around the chromosome are brought together by proteins and RNA generating DNA loops closing the DNA domains to be topologically independent of each other in superhelicity (Fig. 8B). The boundary elements coalesce in generating the rosette observed by electron microscopy (Fig. 3). We propose an alternative working hypothesis for maintaining domain specific superhelicity. It is known that by a process called “transertion” [67] inner membrane targeting proteins inserts into membranes while being nascent and being translated by a nascent mRNA bound to RNA polymerase and template chromosome [68-70]. In some cases, it is the mRNA not the encoded protein that drags the machinery to the membranes [71]. At this stage, the “transertion” structure, being anchored in the membrane through a nascent protein and/or RNA, would serve as a boundary element for chromosomal domains. The genes encoding proteins that go through transertion are distributed around the chromosome. Thus, in the transertion model of domain creation, the intermediate chromosomal segment between two transertion genes would be topologically independent (Fig. 8C). By definition, the life of such domains by the transertion model would be dynamic because membrane “attachment” of the DNA would cease after completion of the membrane protein synthesis. It has been suggested from independent experiments that the topological domains are dynamic [8]. We must note that this model also accommodates the rosette state of the chromosome observed by microscopy (Fig. 3). It is easy to see how degradation of cell wall during lysis in the absence of isotonic buffer would rupture the cell membrane and thus the spheroplasts. Subsequent fusion of the membrane fragments to each other while attached to the transertion complexes would generate a “rosette”-like structure (Fig. 8C). One can distinguish between an in vivo rosette model and a transertion model by differentially labeling and locating appropriate DNA segments in the cell by fluorescent markers.

7. Role of RNA in nucleoid structure

A role of RNA in nucleoid condensation has been suggested by Pettijohn and colleagues. They found that that treatment of the nucleoid with ribonuclease causes unfolding of the nucleoid [72,73]. Since they also found that isolated condensed intact chromosomes from gently lysed cells are organized into topologically independent domains of negatively supercoiled DNA [72], they proposed that RNA might be involved in establishing the topological domains. Later, it was demonstrated that treatment of cells with rifampicin, an antibiotic that inhibits transcription, leads to nucleoid decondensation [74]. We are currently attempting to identify the RNA(s) biochemically determining the full profile of E. coli nucleoid RNAs using a microarray approach, and then analyzing whether one or more of them participates in any step of the chromosome organization and if HU or another nucleoid protein is involved in this step.