The Nucleoid: an Overview

AKIRA ISHIHAMA

[SECTION EDITOR: AKIRA ISHIHAMA]

Posted October 15, 2009

Hosei University, Department of Frontier Bioscience, Koganei, Tokyo 184–8584, Japan

Mailing address: Hosei University, Department of Frontier Bioscience, Koganei, Tokyo 184–8584, Japan. Phone/Fax: 81–42-387–6231, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

INTRODUCTION

The genome of Escherichia coli consists of a single molecule of covalently closed circular DNA. From DNA sequencing of two K-12 laboratory strains (MG1655 and W3110), the E. coli genome is now known to be about 4,650,000 bp in size (238). The large size (about 1.54 mm length when linearized or a circle with a diameter of about 0.49 mm) of the genome is packed into a compact body, known as the nucleoid, in a cell of only 1 μm long (approximately, 1,000-fold compaction). Most of the intracellular space is therefore occupied by the nucleoid. The intrinsic physical properties of the genome DNA, its association with various DNA-binding proteins, and its repulsion from the cytoplasmic crowding altogether contribute to the compaction of the nucleoid into a tightly packed form. Despite this high-level compaction, the nucleoid remains functional, being faithfully replicated, divided, segregated, recombined, repaired, and transcribed, and thus accessible to the replication, segregation, recombination-repair, and transcription machineries. Each of a total of 4,453 genes on the E. coli genome is transcribed in regulated manner at the right time and at the right level. Presently, the genome architecture and its spatial organization are not fully understood, and little is known about the coordination of the genome architecture with its functions. Recent research has, however, led to marked progress in our understanding of bacterial nucleoids. The progress includes: (i) the complete genome sequence for several strains of both laboratory and pathogenic E. coli; (ii) resolution of the molecular events underlying replication, recombination, repair, division, and transcription of bacterial genomes; (iii) understanding of the molecular composition of bacterial nucleoids and the function of each nucleoid protein; and (iv) the application of modern cytological techniques for nucleoid analysis such as use of fluorescent probes and observation with electron and atomic force microscopes. It is thus timely to reassess our view of bacterial cellular organization, including the nucleoid. This chapter provides a brief review of our current understanding of the structure-function relationship of the E. coli nucleoid developed after the overview by Pettijohn (225) focusing on the physical properties of nucleoids.

NUCLEOID STRUCTURE

Nucleoid Isolation

Isolation of nucleoids requires suppression of DNA expansion by various procedures. The ability to control the expansion of nucleoids in vitro has led to purification of nucleoids for chemical and physical analyses and for high-resolution imaging. For isolation of bacterial nucleoids, numerous protocols have been developed that are based on treatment of cells with sucrose-lysozyme-EDTA for partial digestion of peptidoglycan layer and subsequent lysis with detergents in the presence of counter ions such as high salts and polyamines (74, 225). Without counter ions, the condensed state of the nucleoid DNA is lost and the DNA unfolds during or soon after lysis. Depending on the lysis conditions, both membrane-free and membrane-bound nucleoids could be isolated (Fig. 1). In the case of membrane-attached nucleoid, specific regions or domains of the nucleoid are thought to be involved in interaction with the membrane (225). In all cases, nucleoid purification from cell lysates involves sedimentation through sucrose or glycerol density gradients containing the stabilizing counter ions.

Fig. 1Electron micrographs of isolated E. coli nucleoids after spreading the DNA on protein monolayers. (Left) Membrane-free nucleoid (bar, 1 μm). (Right) Membrane-associated nucleoid.

Source: Micrographs courtesy of R. Kavenoff.

High-salt nucleoids.

When E. coli cells are lysed under low-salt conditions where the cell membrane is retained, some cytoplasmic proteins are released through the disassembled membrane and, as a result, the nucleoid expands to fill the intracellular space. To overcome such problems, a variety of simple monovalent or divalent salts have been used as stabilizing counter ions. Relatively high concentrations (0.5 to 1.0 M) of simple salts such as NaCl are often used for isolation of “high-salt nucleoids” (74, 225); under these conditions, many DNA-bound proteins become dissociated, resulting in nucleoids becoming detached from membranes. DNA-bound proteins play a role in maintaining the DNA in a condensed state and thus stepwise dissociation of DNA-bound proteins results in expansion of nucleoids. Among the proteins tightly bound even after high-salt treatment is transcribing RNA polymerase, which remains associated with DNA even after equilibrium centrifugation in CsCl or Cs₂SO₄ (199, 200).

Polyamine and polylysine nucleoids.

The addition of low concentrations of polyamines such as 1 to 2 mM spermidine is another choice for stabilization of the nucleoids; under these conditions, most nucleoid-bound proteins remain associated with the “spermidine nucleoids” (195, 196). Polylysine of relatively low molecular weight is also used to produce lysates containing nucleoids that are only a few fold expanded relative to the sizes of in vivo nucleoids (333). The expanded forms of nucleoids can be converted into compact forms similar in dimension to the cellular nucleoids by further addition of polylysine. Compact forms of the “polylysine-spermidine nucleoids” are progressively expanded by exposure to higher salt concentrations (331).

From “high-salt nucleoids,” some DNA-bound proteins are dissociated during nucleoid isolation, resulting in deformation of nucleoids. In the case of “polylysine-spermidine nucleoids,” the nucleoid-associated proteins are partially dissociated, but the nucleoids can be isolated without loss of the compactness in the presence of a macromolecular crowding agent such as polyethylene glycol (330). Several important structural parameters such as DNA packing density, molecular weight of the DNA, specific linking number deficit, and domains of supercoiling appear to be preserved in the polyamine assembly.

Light microscopic observation of “spermidine nucleoids” indicates envelope-associated dense particles, while electron microscopy shows an irregular central core that is partially covered with small, membrane vesicles (196). Addition of polyethylene glycol or dextran sulfate during nucleoid isolation prevents denaturation of the “spermidine nucleoids,” extending the range of stability of the isolated nucleoids (195).

Osmotic nucleoids.

Osmotic shock of E. coli spheroplasts results in nucleoids free from envelope fragments (55). Osmotic shock is also used for preparation of nucleoids that remain compact under the low ionic strength such as 10 mM NaCl. The nucleoids isolated by osmotic shock without using high salt and in the absence of polyamines exhibited, at least for a certain time period, expanded structures with an estimated volume of about 27 μm when measured as a function of the concentration of added polyethylene glycol (56).

Nucleoid Morphology

The morphology of nucleoids has been mainly investigated by using electron microscopy in various ways. Recent progress depends on the employment of nanotechnology such as atomic force microscopy (AFM) and in silico image reconstruction of three-dimensional structures. Use of fluorescent labeling of nucleoid components is permitting breakthroughs in the investigation of in vivo nucleoids.

Isolated nucleoids.

A landmark study of the structure of the nucleoid was its visualization by scanning electron microscopy and transmission electron microscopy after spreading on protein monolayers (225). Under the scanning electron microscope, the particles have the approximate size of the nucleoid visualized in the cell, which is close to two nucleoid structures that are commonly observed in exponentially growing cells (Fig. 1). Thus, isolated nucleoids have a unique DNA packing density similar to that observed in vivo. When the nucleoids are spread for examination by transmission electron microscopy, however, the maximal observed extension of genome DNA is much greater, suggesting that the state of compaction changes markedly depending the spreading conditions.

Nucleoids in vivo.

The high concentration of proteins and RNAs, ranging from 0.3 to 0.4 mg/ml in E. coli cytoplasm (334), induces the phase separation between the nucleoid and cytoplasm. This cytoplasmic crowding plays a role in nucleoid compaction (194). Recent development of cytological technology has made it possible to locate the nucleoid within bacterial cells. The fluorescent-labeled nucleoids can be found at a variety of locations in the bacterial cells. Image reconstruction indicated that the nucleoid in living E. coli cells forms a doublet shape, consisting of two replicated nucleoids joined by a linker (332). Based on the observation of the nucleoids with simultaneous fluorescent labeling at two different sites of the genome, the nucleoid is organized with the left and right arms in separate cell halves (205). The membrane attachment site of nucleoids varies depending on the cell division cycle, but the replication origin (oriC) and terminus (terC) are always maximally apart (for details, see “Replication apparatus,” below).

Domain Organization

The E. coli nucleoid is divided into topologically independent regions, called “domains,” by the action of as yet uncharacterized barriers. Although the precise size and distribution of domains on the E. coli nucleoid is not clear, the nucleoid domain is defined as a region of genomic DNA bounded by topological constraints on the rotation of the DNA double helix. This notion was originally proposed based on the microscopic observation of isolated nucleoids. For instance, the electron microscopic and cryo-substituted electron microscopic observations of thin sections of E. coli cells indicated that the nucleoid is composed of a rosette-like structure with domains of supercoiled DNA loop emanating from a central node (145, 146) (Fig. 2). The nucleoid domains correspond physically to branch projections of the nucleoid that can be visualized by immunostaining techniques (21, 240). The domain organization is considered to constitute transcriptionally active DNA surfaces accessible to transcription apparatus. In concert with such prediction, the arrangement and size of domains are dynamic and variable depending on the cell growth conditions (291). Even using the cryo-substitution methods, however, the nucleoid was irregular and dispersed, making it difficult to show any detailed structure of the nucleoid domains.

Fig. 2Topological domain organization of the E. coli nucleoid. Domain size measurements were performed from electron micrographs of isolated nucleoids (230).

Source: Figures courtesy of L. Postow.

The genome DNA is believed to be organized into long-range architectures to form a series of independent structural domains that expose DNA loops on the surface of nucleoid (225, 259). If the nucleoid domains are topologically equivalent to DNA loops, observed in the electron micrographs of isolated nucleoids, the domain formation contributes to achieving long-range condensation of the genome DNA into an organized compact structure within the nucleoid.

Macrodomains.

Using psoralen photo-affinity cross-linking probes, the average domain size in the E. coli nucleoid was once estimated to be about 100 kbp or about 47 domains of equal size (225). Analysis of the rate of site-specific recombination indicated the restriction of DNA interactions within subregions of the nucleoid, consisting of four macrodomains (MDs) and two less-structured regions (208, 291). Two of the MDs correspond to the Ori and Ter MDs (Fig. 3). Replication of the genome starts at the ori site within the Ori MD and ends at one of ten ter sites within the Ter MD. The Ori MD is flanked by two nonstructural regions, while the Ter MD is flanked by right and left MDs. Genetic analysis indicates that four different MDs do not interact each other during cell growth. Fluorescent microscopic observations not only support the macrodomain concept (81) but also indicate dynamic behaviors of each domain during the cell cycle (82). Motility of genetic markers appears restricted within domains, while markers exhibit greater mobility in nonstructural regions. During the genome, the Ori MD and two flanking nonstructural regions segregate together, concomitantly followed by segregation of right and left MDs. Cell division takes place after segregation of the Ter MD. The organization of Ter MD during the segregation depends on the association of MatP protein at 23 repeats of the 13-bp-long matS site (184).

Fig. 3Organization of macrodomains within the E. coli nucleoid. Boccard and colleagues (81, 82) proposed the organization and dynamic behaviors of the macrodomains.

Source: Figures courtesy of F. Boccard.

In agreement with the macrodomain organization of the E. coli nucleoid, an in vivo genomic fragmentation assay mediated by DNA gyrase in the presence of oxolinic acid also indicated the presence of long-range segments ranging from about 1,000 kbp to 50 kbp (34).

Microdomains.

Biochemical measurements of the torsional tension and long-range mobility of the genome DNA in E. coli cells, however, suggested that the nucleoid contains approximately 40 topologically isolated smaller domains (264, 274, 313). Involvement of topoisomerases in maintenance of the E. coli nucleoid intact has been suggested based on the observations that treatment with topoisomerase inhibitors generates large DNA fragments of 50 to 100 kbp in length (118, 324). When E. coli embedded in agarose is lysed in situ with a detergent and protease, the typical E. coli nucleoid forms a rosette, in which 20- to 50-kb-long loops of DNA emanate from a dense node of DNA core (114). Using the expression of more than 300 supercoiling-sensitive genes to gauge local genome supercoiling, the distance between two supercoiling boundaries was estimated, on average, to be about 10 kbp (or a total of 470 units), much smaller than the macrodomain size of 50 to 100 kbp, implying that there are microdomains within each macrodomain (230) (see Fig. 2).

The fully extended circular DNA of E. coli genome would have a diameter of about 490 μm, but when organized into 50 fully extended equal-sized domains (about 98 kbp per domain), for instance, it would be about 20 μm in maximal dimension. Thus, the domain organization provides a 25-fold condensation of the loop length to compact the DNA into a nucleoid-sized structure. For accommodation of nucleoid into a single E. coli cell of about 1 μm length, more compaction is needed possibly after fully covering the nucleoid with nucleoid proteins or binding of a specific protein with DNA-condensing activity and through macromolecular crowding of cytoplasm. The entire genome can be folded into the cell size, if the domain size is as small as about 10 kbp.

Domain Formation

Several models have been proposed to explain the formation of the domain structure of the nucleoid. Both supercoiling of DNA and association of DNA-binding proteins are involved in domain formation. One of the factors facilitating DNA compaction is a propensity of the DNA molecule for folding due to its intrinsic curvature originating from the nucleotide sequence. The sequence correlations in the genome reflect such a propensity (289). For instance, the A tract is a sequence factor inducing DNA curvature (288). There is a 10- to 12-bp periodicity in A-tract positioning, indicating that A tracts are phased with respect to the DNA helical repeat. The phased A tracts are organized in approximately 100-bp-long clusters, which may serve as binding sites for the nucleoid proteins that have affinities for curved DNA such as HU, H-NS, Hfq, and Rob (see below). The 100-bp-long clusters of the phased A tracts may constitute the structural code for DNA compaction by both providing the long-range intrinsic curvature and increasing the stability of DNA complexes with architectural proteins.

DNA Supercoiling

Isolated E. coli genomes display a number of individually intertwined supercoiled loops emanating from a central core. Loops are formed by short- and/or long-range interactions between pairs of DNA loci along the bacterial chromosome (254). These supercoiled loops are a factor in forming long-range topological domains. A fundamental aspect of the folding and compaction of genome DNA is thus its negative supercoiling (313), with an estimated specific linking deficit (σ value) of about −0.05 (225, 228). This estimation differs by about 50% from estimates of the effective superhelical density measured in living bacteria. The negative supercoiling is critical for the numerous functional processes of genome DNA, such as transcription and replication, that require unwinding of DNA. Supercoiled DNA is equilibrated with torsional constraint in the double helix, which greatly influences important protein-DNA interactions. It has long been considered that two types of the topoisomerase counteracted each other to maintain the optimum level of DNA supercoiling. Type-I topoisomerase (topo I and topo III) removes negative supercoils while type-II topoisomerases (topo II and topo IV) introduce negative supercoils (76, 324). Topo II or DNA gyrase plays a role in forming DNA superhelixes, thereby restraining DNA loops and defining nucleoid domains (74). This possibility has been reexamined by using pulsed-field gel electrophoresis to resolve large DNA fragments of the nucleoid cut in vivo at sites where DNA gyrase is associated (50, 266). Results indicate that DNA gyrase-bound sites are grouped in 5- to 10-kbp clusters that are mostly 50 to 100 kbp apart. Noteworthy is that the average separation is similar to the average size of the macrodomain in E. coli. This finding raises the speculation that DNA gyrase (one of type-II topoisomerases; see “Replication apparatus,” below) may play a role in forming the barriers that define nucleoid domains.

Protein-Induced Condensation

A number of proteins restrain specific DNA topologies, including negative DNA supercoils and DNA bends, to different extents. In general, their binding to the DNA would reduce the effective unrestrained superhelical density. Major nucleoid proteins HU, IHF, and H-NS have higher affinity for bent DNA and/or bend DNA (see “Molecular composition,” below). Among these major nucleoid proteins, HU is believed to play a major role in condensing the nucleoid by maintaining DNA supercoiling. The nucleoid proteins provide mechanisms for distorting or introducing the flexibility in the short-range structure of DNA within the nucleoid. Estimates of the unrestrained superhelical density of plasmids in vivo indicate that the effective superhelical density is about half that of isolated plasmid after removal of some nucleoid proteins (18, 228). For instance, mutant E. coli cells that lack HU protein have plasmids with reduced superhelical density, suggesting that HU protein actually restrains negative DNA supercoils in vivo (106, 321). However, HU mutants with mutations in the N-terminal oligomerization domain form a high degree of DNA condensation by wrapping right-handed DNA, which restrains positive supercoils (107, 141) (for details, see “Molecular composition,” below).

Domain Barriers

A number of nucleoid proteins participate in the organization of short-range domains, and, in addition, an as yet unidentified protein(s) must be involved in positioning the domain barriers, resulting in domain loop formation when they are encountered (Fig. 4). The topological barriers arise either at random in the E. coli nucleoid (112) or at specific local constraints as observed with Salmonella nucleoids (92). The domain barriers are, however, not placed stably at fixed sites on the genome, but instead both the location of domain barriers and the size of domains vary depending on the culture conditions, indicating dynamic changes in the domain organization.

Fig. 4A model of domain organization.

The domain barrier model requires that the genome DNA contain a repeated similar DNA sequence element(s), which is recognized by the protein(s) that defines the boundary of nucleoid domains. One candidate is the repetitive extragenic palindromic (REP) or palindromic unit (PU) sequence (97, 276), which is an extragenic palindromic sequence of approximately 30 bp in length. This type of sequence is repeated hundreds of times throughout the entire E. coli genome, totaling about 0.5% of the genome DNA. DNA knots and tangles formed by as yet unsolved unique sequences may be other candidates forming domain barriers along the genome (112). Since DNA gyrase specifically recognizes REP sequences and the recognition is stimulated by HU protein (319), it raises the possibility that DNA gyrase-bound REP sequences form the domain barriers. The stable association of DNA gyrase at the barriers should be reasonable for control of DNA superhelicity at each domain loop in response to the dynamic and differential transcription between a total of about 400 short-range domains. However, the total number of REP sequences is more than that of nucleoid domains, and the total number of DNA gyrase molecules is much less than that of REP sequences (50).

Genomic fragmentation in vivo by endogenous DNA gyrase in the presence of oxolinic acid generates DNA fragments of a minimum size of about 50 kbp, equivalent to the domain size (34), indicating that the DNA gyrase is bound at the boundary of each domain. Besides DNA gyrase, several candidates have been proposed to be responsible for fixing the barriers between region-specific nucleoid domains, including MukBEF SMC (structural maintenance of chromosome) complex and nucleoid proteins Fis and H-NS (67). The MukBEF SMC proteins are considered to be involved in the condensation of genome DNA into the compact nucleoid for partition into daughter cells during cell division, and thus the bacterial condensin is another candidate for targeting the domain barriers (209, 318).

Domain Functions

Metabolic processes of the DNA double-helix lead to three types of topological constraints that all cells must resolve to survive: linking number, catenates, and knots. The linking number represents how many times the two strands of the DNA are interwound. In E. coli cells, the linking number is generally lower than would be most thermodynamically favored and as a result, the DNA forms negative supercoil.

The domain organization prevents spreading or propagation of looping and of relaxation of one genome segment into the entire genome. In principle, separate domains make it possible to regulate superhelical tension independently between different domains. Along this line, the concept of a domain organization of nucleoids has biological significance. The domains have the effect of achieving long-range compaction of the nucleoid DNA into an organized structure. A DNA segment within a single domain could be relaxed while the rest of the genome DNA remained supercoiled (65).

The domain organization is not static but represents dynamic entities resulting from transcription and translation complexes creating topological barriers (168). For instance, DNA enters RNA polymerase through one channel and RNA emerges from another (191), indicating that the nucleoid is frequently remodeled at the sites of transcription (67). Likewise, the nucleoid organization is dynamic during the cell cycle (208). Direct interaction of the genome DNA with the cell membrane and the dynamic state resulting from coupled transcription and translation form bridges with the transcription and/or translation apparatus on the membrane.

Nucleoid Fibers

In eukaryotes, the compaction of genomic DNA is achieved by wrapping DNA around nucleosomes (fundamental repeating units of eukaryotic chromatin consisting of about 200-bp-long DNA and four species of histones) and then wrapping the nucleosomes into higher-ordered structures to create coiled nucleosome fibers of about 30 nm in diameter, which are assembled into chromatin of about 300 nm in diameter (143, 234). The chromatin is subject to reversible covalent and noncovalent modifications that render specific genes accessible to the transcription machinery in response to specific regulatory signals. Recent progress in structural studies of bacterial nucleoids indicates that both prokaryotic nucleoids and eukaryotic chromatins are composed of thin nucleoprotein fibers as fundamental structural units, suggesting the conservation of nucleoid compaction systems.

A one-molecule-imaging technique, AFM, was applied to observe the folded genome in E. coli cells on a cover glass after successive treatment with detergents. This experiment discerned a fibrous nucleoid composed of 80-nm-diameter strands, which in turn were further dissected into a 40-nm fiber (149) (Fig. 5). The E. coli nucleoid dynamically changes its structure during cell growth; the 80-nm fibers releasable from the cell are transformed into a tightly packed state during transition from exponential-growth phase to stationary phase. The hierarchy of 40 nm, 80 nm, and thicker fibers depends on the molecular composition of nucleoid proteins. For instance, the formation of thicker fibers in stationary-phase E. coli depends on the appearance of Dps (DNA-binding protein in starved cells) (see “Stationary-phase nucleoid,” below). Single-molecule techniques have also been successfully employed to understand correlation between DNA metabolism and genome configuration. For instance, the mechanism of DNA topology alteration by topoisomerases has been solved by using single-molecule techniques (64).

Fig. 5A model of structural hierarchy of the E. coli nucleoid. (Right panels) E. coli cells growing at different phases were gently lysed and directly observed with an atomic force microscope (149). (Left panel) A model of fundamental architecture of the E. coli nucleoids predicted from AFM patterns.

Source: AFMs courtesy of K. Takeyasu.

Stationary-Phase Nucleoid

In rapidly growing E. coli cells, the nucleoid domains disperse, but upon entering into stationary phase, the E. coli nucleoid undergoes a marked reorganization to transform from dispersed morphology into an ordered assembly of ring-shaped structure (88) in the cell center while ribosomes segregate into the periphery (312). In the early stage of growth-phase transition, the conformational alteration in the nucleoid takes place by changes in intracellular conditions such as salt composition and energy balance. In the absence of an energy supply, DNA toroids represent metabolically quiescent forms. In the stationary phase or under starved culture conditions, the nucleoid fibers are transformed, in a nonenzymatic manner, into tightly packed and highly ordered assemblies without using energy (89). This compaction is mainly due to the replacement of nucleoid-bound protein species from Fis to Dps (9, 149).

Dps is a very abundant sequence nonspecific DNA-binding protein produced in stationary-phase E. coli cells, and is involved in converting genomic DNA into a crystalline-like array (89). Apparently Dps-DNA cocrystals are formed, converting the overall nucleoid configuration into tightly packed forms (see Fig. 10). The finely tuned and fully reversible phase transition results in sequestration and genetic protection of the genome in starved cells. In parallel with the drastic change in nucleoid organization in the stationary phase and reflecting the transcription-translation coupling in prokaryotes, the conformation of ribosomes, translational apparatus, exhibits a marked change by forming the translationally inert 100S dimers linked by a stationary-phase-specific protein ribosome modulation factor (RMF) (296).

MOLECULAR COMPOSITION

Among a total of about 4,000 proteins encoded by the E. coli genome, approximately 10% (more than 400 protein species) can be classified as being in the family of DNA-binding proteins, of which only a small set of low-molecular-weight DNA-binding proteins is stably associated with the genome and is involved in the architecture of the genome structure, forming the nucleoid. This superfamily of low-molecular-weight DNA-binding structural proteins is hereafter referred to “nucleoid proteins.” However, the other DNA-binding proteins are involved in various functions of the nucleoid such as DNA replication and partition (about 40 species), recombination and repair (about 40 species), and in transcription and regulation (about 350 species). This group of proteins involved in genome functions associate transiently with DNA for expression of their activities.

Common characteristics of the nucleoid core proteins are the low molecular mass in monomeric forms, their ability to form homo- or heterodimers or oligomers for generating functional forms, their ability to induce DNA bending or curvature as originally demonstrated in vitro, and their participation in vivo in forming the supercoiled state of the genome to a suitable level depending on the environmental conditions. In addition, it should be noted that these specific nucleoid proteins also affect DNA functions and could be classified as general or global regulatory factors for replication and transcription. Such functional dichotomy is a common characteristic of the nucleoid proteins. On the other hand, some DNA-binding proteins involved in replication-partition, recombination-repair, and transcription also contribute to altering the nucleoid architecture suitable for expression of genome functions.

When the nucleoid is isolated under mild conditions at low temperatures, large amounts of membrane proteins are also associated in addition to the nucleoid proteins (225), reflecting the association of nucleoids to membranes. Under high-salt conditions, however, membrane-free nucleoids can be isolated (see “Nucleoid structure,” above) but some nucleoid proteins are also dissociated during the isolation of high-salt nucleoids. In contrast, polyamine nucleoids isolated from E. coli at low salt concentrations retain a set of functional proteins that are involved in replication and transcription. Irrespective of the difference in isolation procedure, the major nucleoid proteins can be always identified, forming the nucleoid cores (126).

Nucleoid Protein Composition

The major species of nucleoid core protein share functional properties with eukaryotic histones forming chromatin; even the structures are different from histones. Like the core histones from eukaryotes, some core nucleoid proteins form homo- or heterodimers, and others form heterooctamers (73, 76, 182, 245, 306). Accordingly the fundamental fiber structure of prokaryotic nucleoids is similar to that of eukaryotic nucleosome (149) (see “Nucleoid structure,” above), implying that the basic molecular mechanisms of genome packing may be common to both prokaryotes and eukaryotes.

The nucleoid from exponentially growing E. coli cells contains four major nucleoid core proteins: Fis (factor for inversion stimulation), HU (heat unstable nucleoid protein), H-NS (histone-like nucleoid structural protein), and IHF (integration host factor) (Table 1). In addition, several nucleoid proteins such as StpA (identified as a suppressor of thymidylate synthase mutant phenotype A) always associate with nucleoids, albeit at lower levels. The protein Hfq (host factor for phage Qβ-replication) is known to be an RNA-binding protein with RNA chaperon function, but a part of the Hfq protein is also associated with the nucleoid directly or indirectly through nucleoid-associated RNAs (11). Hfq is also associated with RNA polymerase (278). Abundant nucleoid-associated core proteins, HU, HNS, Fis, and IHF, are likely to achieve most compaction of the nucleoid in growing cells (225) while in stationary-phase cells, Dps (DNA-binding protein from starved cells) becomes the most abundant nucleoid protein.

These nucleoid core proteins were previously considered to be entirely sequence nonspecific in DNA binding, but recent studies indicate that most nucleoid core proteins prefer to bind relatively unique DNA sequences or unique conformation such as curved DNA. Apparent requirements for specific DNA sequences by the major core nucleoid proteins are, in most cases, due to the generation of specific DNA configurations, which are recognized by these proteins. Nucleoid proteins, HU and IHF, are members of a family of prokaryotic proteins that interact with the DNA minor groves in a sequence-specific (IHF) or nonspecific (HU) manner to induce and/or stabilize DNA bending. Thus, both HU (292) and its homolog IHF (1) play a role in reducing the effective volume occupied by genome DNA. Cocrystal structure of Anabaena HU with DNA indicated the DNA bending by associated HU (281). Cocrystal structure of E. coli IHF with kinked DNA showed essentially the same conformation with HU-DNA complexes (237). Sharper bending is correlated with longer DNA-binding sites and smaller dihedral angles. Fis rather prefers to bind sets of specific sequences, while H-NS binds to preformed curved DNA regions.

Intracellular Levels

By using quantitative immunoblot measurements, the level of each nucleoid protein was determined for E. coli at various phases of cell growth (Fig. 6) (9). The order of abundance in the exponential growing cells is: Fis >HU >H-NS >IHF, whereas in the stationary phase, this order changes to Dps >IHF >HU >H-NS >Fis (9). The orderly change in the nucleoid protein composition is accompanied by repression of the growth-related genes in the stationary phase (126).

Fig. 6Growth phase-coupled change in nucleoid protein composition. Intracellular levels of nucleoid proteins were measured at various phases of E. coli growth (11, 126).

Silencing of the overall genome functions correlates with condensation of the nucleoid by replacing the nucleoid protein species (173). Fis is the most abundant nucleoid protein in exponentially growing E. coli cells (9). Fis is a sequence-specific DNA-binding protein (10), which plays not only a structural role in forming nucleoid compaction, but also regulatory roles in genome functions such as replication, recombination, transposition, and transcription (see below) (182, 225). However, Fis disappears during the transition from exponential growth to stationary phase, whereas Dps starts to be synthesized during the transition from exponential to stationary phase (9, 10, 13, 68).

Nucleoid Core Proteins

Eukaryotic histones play dynamic roles in the remodeling of eukaryotic chromatin, thereby controlling the access of RNA polymerase and transcription factors to promoters (80, 295). Bacterial nucleoid proteins play similar roles, or more important roles, affecting various biological processes involved in DNA recombination, replication, or transcription (76, 87, 125, 126, 182, 250). Quick and controlled alteration of the nucleoid conformation is necessary for rapid adaptation of bacteria, which are directly exposed to environments in nature. For instance, the nucleoid core proteins are involved in remodeling the nucleoid to either enhance or inhibit transcription (7, 14, 87, 182). In addition, the nucleoid proteins can also function as either repressors or activators of transcription of some promoters. As in the case of specific transcription factors, the nucleoid proteins activate transcription by either supporting the binding of RNA polymerase or transcription factors to specific promoters and/or subsequent steps of transcription initiation (Fig. 7). For instance, H-NS inhibits transcription by limiting access of either RNA polymerase and/or transcription factors to target DNA sites (28, 38, 144, 252). At least in the cases of Fis and IHF, the direct protein-protein contact with RNA polymerase is considered to be involved (79, 128, 129). Thus, the expression of regulatory functions depends on the site of DNA binding relative to promoters (Fig. 7).

Fig. 7Regulation of E. coli promoters by multiple transcription factors. Some E. coli promoters are regulated by multiple transcription factors including nucleoid proteins. References for each regulon: carA (261); focA (138); marR (176); narG (327); narK (22); nirB (31); ndh (130); gltB (223); bglG (70); sulA (322); gadA (301); acsP2 (27); rrnBP1 (5, 61).

HU.

Nucleoid protein HU is composed of two homologous polypeptides, Huα (HupA, HU-2) and Huβ (HupB, HU-1), each having 90 amino acid residues that are closely related but differ by 28 residues (140). HU forms dimers with a compact, positively charged body and two long β-ribbon arms that track along the DNA minor groove (107, 233, 283) (Fig. 8). The αβ-heterodimer is predominant in E. coli, but α- or ββ-homodimers are also formed under conditions where one component is overproduced. The hupA gene is under the control of a single promoter whereas the hupB gene carries three promoters, indicating that the two HU genes are under different modes of regulation. For instance, Fis activates hupA but represses hupB (47). Accordingly in exponential-growth phase, transcription of hupA is higher than that of hupB, and thus the HUα homodimer prevails whereas transcription of hupA is blocked in stationary phase. Shutoff of hupA transcription also takes place following a sudden temperature shiftdown (95). These observations altogether indicate the different functional roles between HU homo- and heterodimers.

Fig. 8Molecular structure of the major nucleoid proteins. Molecular structures from Protein Data. (A) HU (107). (B) HU-DNA complex (233, 283). (C) IHF-DNA complex (237). (D) Fis (43, 155). (E) H-NS (83). (F) Dps (105).

HU dimers further form octomers even in the absence of DNA (Fig. 9), but oligomerization is enhanced in the presence of DNA because of cooperative interaction between dimers on DNA. The structure is different between homo- and heterodimers. X-ray diffraction study indicates that HU heterodimers are arranged in four layers in the octomer and the 30% sequence difference is responsible to forming the asymmetric heterodimeric structure, consisting of stacked dimers with octomeric repeats in three different forms (107) (Fig. 9). Two spiral multimers wrap around the superhelical DNA. This finding also indicates that the homodimers and αβ-heterodimer play different functional roles.

Fig. 9Structure of HU octomers. Three different states have been identified for αβ HU heterodimer (107).

Source: Figures courtesy of S. Adhya.

HU binds nonspecifically to dsDNA, and induces its conformational alteration. The action mode of HU is concentration dependent. At low HU concentrations below 100 nM, individual HU dimers induce in vitro flexible bends that are responsible for DNA compaction of up to 50% (292). A single bend angle induced by HU monomer is 70° as determined by a FRET assay (116) while two bend angles of 105 to 140° are formed within a HU dimer-DNA complex, leading to the induction of negative twisting in the bound DNA (281). At high concentrations, a rigid nucleoprotein filament is formed (292), in which HU induces a helical bundle of DNA. HU is not a simple DNA-binding protein but more an architectural protein because it wraps DNA around. In HU-DNA complexes, the DNA is characterized by having the same number of superhelical turns as eukaryotic nucleosomes. Thus, with respect to the mode of DNA-protein interaction, HU is analogous to the eukaryotic HMG-1 and HMG-2 proteins (217, 229). Like the HMG-1 and HMG-2 proteins, HU bends DNA or binds preferentially to bent or kinked DNA. HU also prefers to bind DNA in cruciform structures or in four-way junctions, which at their strand intersections can resemble bent DNA (229, 284). The pattern of DNase protection of DNA complexes with HU also indicates a possible wrapping of the DNA by HU (33).

One of the three structures, structure III, of HU octamers forms spiral filaments with left-handed rotations (107), which may provide the structural platform for wrapping DNA (Fig. 9). By providing a foundation for negative superhelicity in DNA, HU αβ-heterodimer is able to form the nucleosome-like structure. At near saturation, HU prefers to bind G/C-rich single-stranded DNA (ssDNA) whereas HU binds and stabilizes double-stranded DNA (dsDNA), displaying a marked preference to cruciform DNA with sequence motifs AAG, AGA, and AAGA (23, 150). As in the case of IHF, HU binds to the DNA minor groove and thus the apparent preference to AT-rich sequences is attributable to the formation of narrow minor groove with the groove width suitable for keeping HU bound (282). Dual architectural roles of HU have been revealed by using magnetic tweezers and atomic force microscopy (292). HU binds to DNA fragments with a stoichiometry of about one HU dimer per 10 bp; however, the total number of HU molecules in exponential-phase cells is about 10,000 molecules for 465 Mbp genome (or one HU per 300 to 400 bp) (9).

Several lines of evidence indicate that nucleoid compaction, coupled with changes in DNA superhelicity, generate marked changes in the overall pattern of genome transcription. Involvement of HU in regulation of the overall pattern of genome transcription was first demonstrated by using a genetically altered HU, which transformed the loosely packed nucleoid into a densely condensed configuration. HU^E38A mutation at the site critical for the formation of left-handed spiral protein leads to defect in DNA supercoiling; this mutation is located within the N-terminal oligomerization domain (107). A double mutant HU^E38K,V42L within the same region induces positive supercoil locally in the overall negative-supercoiled genome, thereby improving growth characteristics (141, 142). The pattern of genome transcription is totally different between E. coli with wild-type HU inducing negative superhelicity and that with the HU mutant inducing positive supercoils as detected by the microarray assay. This finding supports the notion that HU is involved in the overall regulation of genome expression. In the HU mutant, a number of constitutive housekeeping genes are repressed but quiescent genes are activated, indicating that the role of HU on transcription regulation is generally nonspecific and passive, but in some cases, HU binds at specific sites near promoters and plays as active a role as specific transcription factors.

One class of preferred binding sites for HU is at regulatory regions of certain genes where HU can bind cooperatively with regulatory proteins. Studies of this process demonstrate that HU can serve as an auxiliary protein to facilitate the cooperative binding of the lac repressor and CRP (cAMP receptor protein or catabolite activator protein) (16). Effective repression of the gal operon by GalR repressor requires HU, which induces DNA curvature to bring two GalR repressors close together to form the repression loop within the gal promoter (164, 166, 255). The available evidence suggests that HU facilitates short-range structural transitions (such as DNA bends) in the target sequence recognized by the transcription regulatory protein. HU can also function in facilitating other types of protein-DNA interactions, including those involved in site-specific recombination and the initiation of DNA replication at oriC (69, 132). During initiation of the genome replication, HU binds to oriC and interacts with DnaA for enhancement of DNA opening (46). These interactions are believed to occur via structural deformations in the DNA facilitated by HU.

IHF.

Integration host factor (IHF) is a heterodimeric protein composed of IHF-α (99 amino acid residues) and IHF-β (94 amino acid residues), encoded by himA and hip (or himD), respectively. IHF has 30% amino acid sequence identity with HU (76). It is noteworthy that, in structure, IHF is similar to eukaryotic TBP, the promoter TATA-binding protein (201). As in the case of HU, two IHF subunits are intertwined to form dimers with a compact body, from which two long β-ribbon arms extend to wrap around the minor groove of DNA (237) (Fig. 8). Like HU, IHF facilitates DNA bending, but unlike HU, it apparently recognizes a specific DNA sequence. HU alters the conformation of DNA backbone, but in the IHF complex, the structural perturbations encompass both the deoxyribose-phosphate backbone and the bases of the specific binding sequences (17). The bend angle produced by IHF is about 160° as determined by crystallography (237). Since IHF is known to bind in the minor groove of DNA, the apparent preference for specific DNA sequences by IHF is attributable to the induction of appropriate space along the DNA minor groove formed by these DNA sequences. Temperature-jump measurements demonstrate that the binding of IHF to its cognate DNA site involves an intermediate state with partially bent DNA, with a DNA bending rate of approximately 2 ms⁻¹ at 37°C corresponding to the activation energy of 14 kcal/mol (160). Based on these kinetic data, it is predicted that the thermal disruption of a single base pair is sufficient to overcome the free-energy barrier to form the IHF-DNA intermediate complex.

IHF was originally recognized as a host protein that influences phage DNA integration into the host genome to become a prophage. In this case, IHF binds at the λ-attachment phage site of the E. coli chromosome as well as the attachment site in the λ-genome, specifically bending both DNA sites, and thereby facilitating the interaction of other proteins bound on each side of the bend to promote site-specific recombination of lambda DNA (91, 202, 293). IHF influences the expression levels of many genes, and thus IHF is now known to be involved in the regulation of various genes through its specific interaction near respective promoters (see Fig. 7). Proteome analysis by two-dimensional gel electrophoresis indicates a difference in about 15 to 20% (150 to 200 protein species) of the total detectable proteins between an IHF mutant and its isogenic wild-type parent (90). On the other hand, statistical analyses of transcriptome indicated that among a total of 4,290 genes examined, 27 genes decreased more than 2-fold and 95 genes increased more than 2-fold (less than 3%) (6), indicating that more proteins are affected than promoters. Even though only a population of IHF molecules bound at specific sites near promoters affects transcription (263), each promoter usually controls multiple genes organized in a single polycistronic operon. Since results of transcriptome and proteome analysis include a number of genes indirectly affected by IHF deletion, additional evidence such as in vitro and in vivo examination of the transcription regulation of individual target promoters by IHF are needed to identify the targets under its direct control.

Transcription of certain E. coli promoters is under the control of multiple transcription factors including the nucleoid proteins as in the case of eukaryotic transcription (see Fig. 7). For instance, the promoter of nirBDC, encoding cytoplasmic NADH-dependent nitrite reductase, is repressed by a combination of three nucleoid proteins, Fis, IHF, and H-NS, and activated by FNR (an anaerobically triggered transcription global regulator) (30, 32). NarL/NarP (TCS [two-component system] transcription activators triggered by nitrite and nitrate) relieve IHF- and Fis-mediated repression. Likewise, the nrf operon for cytochrome C nitrite reductase is activated by FNR and NalL/NalP, but is repressed by IHF and Fis (31). The acs gene, encoding acetyl-coenzyme A synthase, is also repressed by IHF and Fis, and activated by CRP (27) (Fig. 7). In this case, both nucleoid proteins function directly as antiactivators of CRP, each working independently under different culture conditions. The gene products of sufABCDSE operon are involved in the assembly of Fe-S clusters and their induction under oxidative stress conditions depends on the presence of OxyR, IHF, and IscR (322) (Fig. 7). Another global transcription factor, leucine-response regulatory protein (Lrp), binds to three sites, 252, 216, and 152 bp, far upstream from the promoter of gltDEF encoding glutamate synthase. For transcription activation, however, the binding of IHF between the gltB promoter and upstream-bound Lrp is needed, suggesting that the direct contact between Lrp and RNA polymerase is mediated by IHF-induced DNA bending (222, 223).

The phase variation of type 1 fimbriation in E. coli is controlled in response to availability of the branched amino acids, Ile, Val, and Leu, and is mediated through the site-specific recombination of a short invertible element including the fimAICDFGH promoter. IHF regulates the genes for this site-specific recombination by switching of the promoter segment to the opposite orientation (162, 268). A group of stationary-phase-specific genes including dps, sra, and pstSCAB operons is transcribed by RNA polymerase RpoS holoenzyme only in the presence of IHF (3, 285) while others such as osmY and ihfA are negatively regulated by IHF (8, 49). IHF is required for in vitro recombination of phage P22 (267), but it is dispensable in vivo (45). This apparently contradictory observation could be due to a difference in the superhelical density of target DNA and/or to a different complement of DNA bending proteins.

In some cases, HU and IHF can substitute for one another in vitro as well as in vivo (320). In others, they may also complement mutations in one another (183). Transcription of ihfA is under negative autoregulation, in which IHF binds to three sites near its own promoter (8). The intracellular level of IHF in growing E. coli cells is about one third that of HU, but increases about 3- to 5-fold in stationary phase, reaching to a level that is higher than HU (11, 68, 126). In agreement with growth-phase-dependent level control, transcription of the ihf gene is under the control of ppGpp and RpoS sigma (8).

H-NS.

H-NS, histone-like nucleoid structural protein, consisting of 137 amino acid residues, was originally identified as a heat-stable protein present in RNA polymerase preparations (131). Later, H-NS was identified as one of the major core proteins of the nucleoid (270, 294). H-NS family proteins exist in a wide variety of bacteria and their phylogenetic distribution agrees well with that of 16S rRNA-based phylogenies (287). H-NS is composed of two distinct structural domains joined by a flexible linker region (Fig. 8). Three-dimensional structures of the N-terminal and C-terminal domains have been solved (20, 83, 262). H-NS forms dimers (84) or tetramers (41) in its native state. The N-terminal domain of H-NS is mainly α helical and forms a short coiled-coil motif that contributes to the formation of dimers or tetramers, which are further converted into large oligomers, which are dissociated in vivo in response to increased osmolarity or high temperature. The C-terminal domain with DNA-binding activity contains an antiparallel β-sheet and an α-helix.

StpA belongs to the H-NS family of proteins in E. coli with 60% amino acid sequence identity to H-NS (326). StpA is able to form heterodimers with H-NS as well as homodimers. The N-terminal protein-protein contact domain of H-NS is involved in the heteromeric interaction with StpA (136, 307). Degradation of StpA by Lon protease is protected by the formation of heterodimers with H-NS (307). H-NS also forms complexes with each of two H-NS-binding proteins, Hha (hemolysin expression-modulating protein) (207) and YdgT (renamed to Cnu) (150, 224). Hha was originally identified as a protein that influences the expression of virulence factors (207), but later the Hha/YmoA family proteins including Cnu were recognized as H-NS-binding proteins (206, 224). H-NS-Hha complexes bind to specific regions, including regulatory sequences of the target genes. Interestingly, the structure of Hha protein is similar to the N-terminal domain of H-NS protein (206), which is involved in protein-protein interaction for formation of H-NS homodimers and H-NS-StpA heterodimers. These observations altogether indicate that the Hha family proteins modulate the oligomerization pattern of H-NS/StpA family proteins. As in the case of Hha, Cnu consists of three α-helices in its N-terminal domain (12). In both Hha and Cnu, the helix 3 provides a common structural platform for H-NS binding. Some H-NS functions require the formation of heterodimers. The formation of mixed heterooligomers between the two family proteins, H-NS/StpA and Hha/YmoA, widens the mode of regulation in response to environmental conditions (135, 172, 206, 224). H-NS also interacts with Hfq (host factor for phage Qβ), an RNA chaperone protein (280), implying that H-NS forms RNA complexes via Hfq. In certain cases, H-NS directly interacts with RNA, thereby influencing the stability of RNA (26). Accordingly both H-NS and SptA are now recognized as RNA chaperones (187, 307, 328).

H-NS employs a different mechanism of DNA recognition than HU and IHF do. H-NS recognizes and binds to intrinsically curved (or bent) AT-rich DNA sequences (25, 59, 163, 239, 315). Apparent preference for specific DNA sequences by H-NS is because the formation of curved DNA conformation depends on the specific AT-rich sequence motif (163, 315). The effect of H-NS on gene expression is global in target selection and negative in action mode because curved sequences are often located near promoters (25, 133). Promoter-associated curved structure provides H-NS with the initial contact site to express its silencing function of transcription. The major source for the gene selectivity in transcription silencing by H-NS is the presence of such a binding motif. In fact, chromatin immunoprecipitation-on-chip (ChIP-chip) analysis indicates high-level colocalization of H-NS and RNA polymerase (216). One H-NS dimer binds to each 15 to 20 bp of DNA (4), and bridges adjacent tracts of DNA, leading to the induction of compaction of short-range DNA (270). After binding to such a high-affinity nucleation site, H-NS binding spreads to neighboring regions along AT-rich DNA, ultimately resulting in the formation of a supercoiled filament containing two DNA duplexes joined by a protein bridge (163). Accordingly, H-NS represses transcription of a large number of genes located within extensive regions of the genome. Thus, H-NS is sometimes called a transcription silencer (98, 204).

Single-molecule technology was applied to observe the H-NS-DNA complex formation (58, 60). Reflecting the step-binding model consisting of the initial local binding for nucleation and the packaging the DNA by spreading through protein-protein interaction (163), two different levels of DNA condensation were observed by AFM: lateral condensation of large regions and formation of large globular structures. StpA also has the ability to bridge DNA (60). At physiological concentrations, a large number of H-NS molecules associate, and form polymers along DNA and cover extended tracts of DNA leading to an apparent increase in DNA width 2-fold greater than naked DNA (4). Overproduction in vivo of H-NS leads to the highly condensed nucleoid and is lethal (273), while the absence of H-NS in an hns deletion mutant results in an increased degree of negative supercoiling (186).

H-NS is a typical nucleoid protein with functional dichotomy (7, 120) and is recognized as a universal regulator controlling a number of genes in various directions and various extents (reviewed in reference 72). Mutations in the hns gene affect the expression of many different genes of apparently unrelated functions (cited in reference 287), including de-repression of β-glucoside metabolism, increased resistance to low pH and high osmolarity, and a loss of motility. DNA microarray analysis indicated that the expression of as many as 5% of the E. coli genes is affected in an hns mutant (117). In many cases, the effect of H-NS on gene expression is direct, being mediated by the binding of H-NS and the changing DNA topology of the promoter region (7, 120). In fact, the curved DNA with H-NS-binding activity often exists near promoters. In these cases, H-NS functions directly as a repressor and interferes with the formation of RNA polymerase-promoter complex (290). For instance, H-NS binds at two regions, one overlapping with promoter −10 and −35, and another at about the −100 region, on the promoters of gadA encoding glutamate decarboxylase and gadX encoding an activator for the gad operon (96, 301) (see Fig. 7). There is a functional competition between H-NS and GadX on the gadX promoter.

However, in good agreement with the spreading mechanism of H-NS binding along DNA, transcription repression extends to wider regions of the genome than the initial binding site (98, 163, 204). Thus, H-NS is sometimes referred to as a transcription silencer. In the case of leuO silencing, H-NS first binds at an AT-rich sequence located upstream of the promoter and then forms a cis-spreading H-NS filament toward the target promoter (42). Likewise, for silencing of proU, one H-NS molecule first binds to one of two high-affinity sites, thereby leading to cooperative binding of the second H-NS molecule for induction of structural changes in DNA superhelical density (24). In most cases, transcription activators antagonize the repressive activity by H-NS, and thus are referred to as antirepressors or antisilencers (39, 85, 137, 304). Transcription repression by H-NS is, however, not always due to interference of the binding of RNA polymerase and/or transcription factors. In the case of rrnB P1 repression, H-NS is assembled into RNA polymerase-promoter open complexes and stabilizes these complexes, thus preventing promoter clearance (61). The silencing of rRNA promoter by H-NS is supported by leucine-responsive protein (LRP) and could be relieved by displacement of DNA-bound LRP by leucine (232). The relief from silencing could also be mediated by displacement of H-NS by other nucleoid proteins such as Fis (61).

The location of H-NS binding sites along the genome has been analyzed by using ChIP-chip technology (170, 216). Results indicate that H-NS-associated sites overlap with promoters of at least 250 loci or 1,000 genes. Interestingly, RNA polymerase is often present nearby, supporting the model that H-NS interferes with the functioning of RNA polymerase, but not the binding of the latter. It is noteworthy that the gene-silencing role of H-NS appears to be used as a defense system, which controls horizontally transferred foreign genes. Consequently the virulence by pathogenic E. coli increases in hns mutants (190). Functional interplay has been observed between H-NS and SlyA (53, 167), a transcriptional regulator, which is required for expression of the virulence and bacterial survival in phagocytic host cells (2002 272). Since AT-rich motifs with H-NS-binding activity are often identified in control regions of pathogenicity islands or virulence genes, which are acquired by lateral transfer from other pathogens (170, 204), an attractive hypothesis has been proposed that H-NS is a key regulator in silencing horizontally transferred genes (71).

Transcription of a limited number of genes is known to be activated by H-NS. In particular, H-NS controls approximately 70% of the temperature-regulated genes (305). For instance, stringently controlled promoters are activated by H-NS because these promoters require a supercoiled state for function (134). In the case of the flhDC master regulator of the flagella regulon, H-NS bound to upstream represses transcription by interfering with the binding of CRP, but the region downstream of the transcription initiation site plays a crucial role in the positive control by H-NS (269). Furthermore, besides the regulation of flagella biogenesis, H-NS interacts with the FliG motor protein and modulates the function of the flagella motor (152). As in the case of DNA-binding transcription factors, the mode of regulation between activation and repression by H-NS varies depending on the site of H-NS binding along promoters.

The intracellular concentration of H-NS is constant throughout the growth phase at a level of about 20,000 copies per genome equivalent (11) under the autogenous control by H-NS and StpA (see Fig. 6). The promoter of hns is regulated by Fis, another abundant nucleoid protein in involved in growth-phase-dependent control of nucleoid conformation, thereby controlling the balance between nucleoid core proteins. The synthesis of H-NS is also under the control of cell cycle through positive correlation between hns transcription and DNA replication (115). The hns promoter is also stimulated by cold shock through the CspA protein (161). The synthesis of H-NS increases at elevated hydrostatic pressure (302).

Fis.

A 98-amino-acid DNA-binding nucleoid protein, Fis forms a homodimer in its native state. Fis (factor for inversion stimulation) was initially identified by its ability to stimulate DNA inversion (87). The crystal structure of Fis (43, 155, 323) indicates that each subunit contains a β hairpin in its N-terminal domain (residues 11 to 26) followed by four α-helices (A to D) separated by short turns in its C-terminal domain (Fig. 8). Helices C and D form the motif for DNA binding and bending.

Fis stimulates both excisive and integrative recombination. Stimulation of inversion is mediated by specific binding of Fis to regions that are strongly bent upon interaction with Fis (119). These regions contain two to three Fis-binding sites, sometimes called recombinational enhancers. Although Fis binds DNA in a site-specific manner, its consensus sequence is highly degenerate, consisting of poorly conserved AT-rich 7- to 11-bp palindromic sequences separated by 11 bp (110), implying that, as in the case of HU and IHF, Fis recognizes a structural feature(s) rather than the primary sequence. In contrast to HU and IHF, however, Fis interacts not only with phosphate backbone, but also with specific bases on both DNA strands within the core-binding site (257). The amino acid residues including R85, N84, T87, and R89 in the DNA-binding C-terminal domain are involved in the base recognition (257). By employing a ChIP-chip system, a total of 894 Fis-associated sites were identified on the E. coli genome (44). About one-third of the Fis-associated sites were located within open reading frames. An average of two Fis-binding sites was identified for each of the supercoiling domains. The overlapping Fis sites are, however, not occupied in vitro by two Fis dimers, but instead only one of these sites is bound by Fis at a time. In supercoiled DNA, Fis binds at multiple sites in a nonrandom fashion and increases DNA branching, leading to reshaping the genome DNA (251). Homodimers of Fis bind to and deflect DNA from 40° to 90° (220). AFM was employed to observe the global DNA-reshaping effect of Fis, which induces branched plectonemes (251).

Fis is involved in global control of the genome functions. Molecular flip-flops between nearby sites could be employed for switching between transcription and replication. Fis is one of the key global regulators of transcription controlling a number of genes for adaptation to external conditions such as the availability of oxygen and nutrients. For instance, Fis is involved as a repressor in regulation of the promoters for the nir operon encoding cytoplasmic NADH-dependent nitrite reductase (27, 29), the nrf operon encoding cytochrome C nitrite reductase (31), the acs gene encoding acetyl-coenzyme A synthase (27), and the guaB gene encoding GMP synthesis (121) (see Fig. 7). The most important functional role of Fis is the regulation of transcription of many genes involved in translation, and many of these genes are highly transcribed in the exponential-growth phase. Thus, Fis plays a critical role in coordination of the major reconfiguration of cellular processes upon nutritional upshift. For instance, Fis stimulates transcription of a total of more than 41 genes (cited in reference 147), which are mostly essential for cell growth such as rRNA, r-protein, and tRNA genes (210, 244) (Fig. 7). Fis contributes to the activation of rRNA transcription by binding to Fis recognition sites in the rRNA P1 promoter region (5). Fis mutants have been isolated that retain the ability to bind DNA, but are unable to activate rRNA transcription (100), indicating that, in addition to DNA binding, contact with RNA polymerase is needed for transcription activation. The loop region of Fis between helices C and D in its C-terminal domain is supposed to be involved in RNA polymerase contact (43, 100). Direct contact between Fis and RNA polymerase on the tyrT promoter has been visualized by using atomic force microscopy (AFM) (179). Individually, Fis and RNA polymerase wrap 80- and 150-bp DNA of the tyrT promoter, each being partially overlapped. Direct contact between Fis and RNA polymerase is mediated by the C-terminal domain (CTD) of RNA polymerase α-subunit (192). The rRNA operons also contain an UP element located between the Fis-binding sites and the promoter −35 region, which also contributes to transcription activation of rRNA operons by interacting with the α-subunit of RNA polymerase (243). Transcription factors are now recognized as regulatory proteins with dual functions, switching between activators and repressors depending on the site of DNA binding relative to promoters. In fact, a set of at least 17 genes is negatively regulated by Fis (314; also cited in reference 147).

Fis expression spikes sharply during the exponential phase and decreases precipitously as cells enter the stationary phase (13). In exponentially growing cells, Fis is the most abundant nucleoid protein reaching to a peak of 50,000 molecules per genome (or one molecule per 250 bp DNA), but it decreases in stationary-phase cells to a nearly undetectable level (11). In keeping with its role as a key regulator of cell growth, Fis expression is governed by multiple control mechanisms (174, 203). The fis gene forms an operon together with the upstream-located dusB gene for tRNA-dihydrouridine synthase B, and only a single promoter located upstream of dusB is under the control of IHF, CRP, and Fis itself (15, 174). Transcription of fis is enhanced by an increase in negative supercoiling and by the nucleoid protein IHF and several transcription factors such as cAMP-CRP, but is repressed by the stringent factor ppGpp (211, 297). Fis regulates crp transcription while CRP controls fis transcription, thereby resulting in cross talk between two global regulators (203). The fis promoter is among the few that predominantly initiate transcription with CTP (297). This unique feature of initiation nucleotide selection is a functional determinant of its growth rate control (298). The intracellular concentration of CTP influences the level of fis transcription. Fis expression is also controlled at the level of translation by GTP-binding factor BipA or TypA, which binds to ribosomes at a site that coincides with that of elongation factor G (218). The translation regulation activity of BipA is regulated by tyrosine phosphorylation (86).

Dps.

Upon entering the stationary phase, the E. coli nucleoid transforms from a dispersed morphology into an assembly of ring-shaped structures (88). The starvation-induced DNA-binding protein, Dps (DNA-binding protein from starved cells), consisting of 168 amino acid residues, was identified in late stationary cultures of E. coli (2). Dps is induced during the transition from exponential- to stationary-phase growth, reaching up to 200,000 molecules per cell (11). Fis, the most abundant nucleoid protein in exponentially growing E. coli cells, binds to the dps gene promoter and represses, together with H-NS, its transcription by RNA polymerase containing RpoD sigma factor (102, 253).

The transformation of nucleoids in the stationary phase largely depends on the association of Dps, which covers almost two-thirds of the E. coli genome (11). Dps-dependent supercompaction of the nucleoid is triggered concomitant events, the appearance of Dps and the disappearance of Fis. The nucleoid transformation during the growth transition takes place sequentially. The initial event is the conformational alteration in genome DNA by changes in intracellular conditions such as salt composition and energy balance, and afterward, the second-step change takes place by binding Dps (88, 236). Since Dps homologs have been identified in a variety of distantly related bacteria and their sequences are highly conserved, Dps appears essential for bacterial survival in stationary phase.

The E. coli Dps monomer with the molecular mass of 18 kDa has a four-bundle core that is almost identical with the core of the bacterioferritin (Bfr) monomer (105). The first two helices are connected to each other by a short loop, while the last two helices are also connected to each other, two pairs of helices being connected by a long loop (Fig. 10). Bfr forms a 24-mer with 432 symmetry, a hollow core, and pores at the threefold axis. Likewise Dps forms a spherical dodecamer with the molecular mass of 205 kDa and an interior core diameter of 4.5 nm (312) (see Table 1). Scanning electron microscopy showed a ring-like structure about 9 nm in diameter. Dps dodecamers further form a hexagonally packed multilayered structure (88). Electron microscopic analysis of Dps-DNA complexes showed highly ordered and tightly packed Dps-DNA co crystals (2, 88, 89) (Fig. 10). Both self-aggregation and DNA condensation depends on the N-terminal domain of Dps (40). Crystalline structure of Dps-DNA complexes with similar lattice spacing as formed in vitro are observed in vivo in growing E. coli cells with overproduced Dps or in starved wild-type cells (312).

Fig. 10Dps-DNA cocrystals in stationary-phase E. coli cells. Electron microscopic observation of stationary-phase E. coli cells. Highly produced Dps forms cocrystals with the genome DNA (88, 89).

Source: Figures courtesy of A. Minsky.

The transition from active growth to stationary phase entails a coordinated process of sequential transformation of the nucleoid (149). Within the condensed nucleoid, the genome DNA is effectively protected by means of structural sequestration, and the sequestration of macromolecules in crystalline assemblies provides an efficient means for protection (88). DNA-Dps cocrystallization represents a binding mode that provides wide-range protection of DNA by sequestration. Both the induction of Dps synthesis and the prevention of its degradation contribute to the huge accumulation of Dps in the stationary phase (11). The Dps accumulated in stationary-phase E. coli is, however, degraded rapidly upon reinitiation of cell growth. This turnover depends on the ClpPX protease but not on the ClpAX protease (275). Dps proteolysis is a regulated process such that Dps is stabilized upon carbon starvation, but the addition of glucose in starved culture results in rapid resumption of Dps proteolysis. Dps proteolysis is independent of the proteolytic targeting factor RssB or SprE (275).

Dps is required not only for long-term survival in the stationary phase, but also during the exponential phase; it is also needed to protect cells from external stresses (178). Initial studies of a dps-disrupted mutant indicated its role in protection against oxidative stress (2). The lethality of oxidative agents like H₂O₂ is a result of the formation of highly reactive hydroxy radicals in the presence of free Fe(II) ions by Fenton's reaction. Dps protects against DNA damage by masking the exposed DNA surface (312) as well as by binding toxic free Fe(II) ion to convert it to less toxic Fe(III) (123, 329). During the stationary phase, Dps protects the cell not only from oxidative stress, but also from UV and X ray irradiation, iron and copper toxicity, thermal stress, and acid and base shock (198). The protective roles of Dps are achieved through a combination of nucleoid compaction, metal chelation, ferroxidase activity, and regulation of genome expression. In addition to the heme-containing bacterioferritins, a group of ferritins without heme display sequence similarity with Dps, suggesting an evolutionary relationship between Dps and ferritins. In agreement with the structural similarity, Dps has been shown to bind iron and carry out ferroxidation (123, 329). E. coli Dps is able to bind up to 500 iron atoms within its central cavity.

Nucleoid Protein Interplay

Genetic studies of E. coli suggest that the nucleoid proteins have partly overlapping functions (76, 227, 250). Thus, mutants that are unable to produce two nucleoid proteins, H-NS and IHF or H-NS and HU, are not lethal, but a triple mutant lacking all three proteins, H-NS, IHF, and HU, could not be constructed, suggesting that the proteins share a common function (321). DNA gyrase-mediated in vivo fragmentation is not significantly altered in strains defective in one of the major nucleoid proteins (H-NS, HU, Fis, and IHF) (93). These observations together indicate that the overall architecture of the nucleoid is retained in these single mutants if the other nucleoid proteins are present. Upon missing one nucleoid protein, the vacant space on the nucleoid might be occupied by another nucleoid protein with similar function. Systematic analysis is needed to get insights into the network of interplay between nucleoid proteins.

Nucleoid proteins play both regulatory roles of the genome functions and the architectural roles in the nucleoid formation. The nucleoid proteins play key roles in nucleoid dynamics (250). In the regulation of genome transcription, the nucleoid proteins act as the global regulators controlling the overall structure of nucleoid and as the specific transcription factors controlling individual genes or operons. Reflecting the multiple roles of nucleoid proteins, mutations affecting the nucleoid-associated proteins are pleiotropic, exhibiting various modes of influence depending on the model system under study (57).

Intracellular Localization

Immunogold staining of the HU protein revealed that HU is localized at the nucleoid surface (77). However, immunofluorescence studies of fluorescently tagged HU protein introduced into permeabilized E. coli cells showed that the added HU bound throughout the nucleoid (264). Fluorescein-tagged proteins that do not bind to DNA, such as albumin and insulin, remained primarily in the cytoplasm of permeabilized cells, and little could be detected within the nucleoid volume. Thus, it appears that there is no intrinsic restriction on the binding of HU within the nucleoid interior. Similar studies of H-NS showed that this protein is also distributed throughout the nucleoid (78), which suggested that H-NS is the major nucleoid protein involved in packaging DNA in the nucleoid interior.

The indirect immunofluorescent staining of nucleoid proteins in E. coli cells indicated two types of localization (9). The group-I proteins, including the major nucleoid-structural proteins, H-NS, HU, IHF, StpA, and Dps, are distributed uniformly within the entire nucleoid while the group-II proteins formed 2 (SeqA), 3 to 4 (Rob and CbpB), and 6 to 10 (Fis and IciA) immunostained dots within the nucleoid. Each immunostained dot may represent either the association of 100 to 1,000 molecules of each DNA-binding protein at a specific locus of the genome DNA or the assembly of protein-associated DNA segments from different domains of the folded genome. At the concentrations of DNA that exist in the nucleoid, DNA occupies less than 5% of the nucleoid volume, yet proteins that do not bind DNA enter this volume negligibly, if at all. On the other hand, DNA-binding proteins such as HU enter the nucleoid volume quickly, implying a rapid DNA-to-DNA exchange mechanism not available to non-DNA-binding proteins (9, 197).

Recently ChIP-chip has been applied, together with high-density microarrays, for systematic analysis of the location of nucleoid proteins along the E. coli genome (103). The distribution of nucleoid core proteins appears biased to intergenic regions. Interestingly, some of Fis and H-NS molecules are colocalized with RNA polymerase, supporting the concept that these nucleoid proteins function as regulatory factors of some specific promoters (see above).

Minor Nucleoid Proteins

A number of hitherto unidentified nucleoid proteins have been identified by using a combination of proteome analysis and mass spectrometry. YaiD (or RdgC) and YejK are rather abundant proteins associated with the spermidine nucleoid, which retains even weakly associated nucleoid proteins (193). RdgC is required for the proper replication of genome DNA in cells deficient in the recombination enzymes, RecABC and SbcCD (247), implying that RdgC is one of the components of replication apparatus.

In summary, the E. coli nucleoid contains three proteins, HU, IHF and N-NS, as the major architectural components, and an additional protein Fis for cells in exponential growth phase and Dps for stationary-phase cells. All the major nucleoid proteins are now believed to have functional dichotomy, controlling the nucleoid functions. At present, however, the functions remain unidentified for about half of the genes in the E. coli genome, which are not needed to grow under laboratory culture conditions, but may play certain roles in survival under stressful conditions in nature. Along this line, it is not unlikely that novel nucleoid proteins are produced under as yet uncharacterized conditions as Dps was detected in stationary-phase cells.

NUCLEOID FUNCTIONS

The E. coli genome is tightly packed into the nucleoid, but, at each cell division, the genome must be faithfully replicated, divided, and segregated. Each of a total of 4,453 genes must be transcribed at the right time and at the required level. Nucleoid activities such as transcription, replication, recombination, and repair are all affected by the structural properties and the special conformations of nucleoid. A set of generally abundant proteins with high affinity to DNA is stably associated with the nucleoid, thereby contributing to various degrees in compaction of the genome. Besides these nucleoid architectural proteins, a number of proteins are transiently and reversibly attached to the nucleoid with different binding affinities, including the components and factors involved in genome replication and division, recombination and repair, and transcription.

Replication Apparatus

Replication of the genome DNA is a universal process that proceeds in distinct stages, from initiation to elongation and finally to termination. Each stage involves multiple transient interactions of DNA-binding proteins. Genome replication in E. coli starts from a single locus of about 250 bp in length, called oriC, and proceeds in both direction. This DNA sequence for replication initiation, oriC, contains the binding sites for a number of unique proteins that participate in the initiation of genome replication (Table 2), including eight DnaA boxes for binding of the initiator protein DnaA (271), AT-rich 13-bp repeats for binding of IciA (122), Rob box for binding of Rob (right origin-binding protein or CbpA) (265), 26-bp sequence for Cnu (oriC-binding nucleoid-associated protein) (150), and the sites for nucleoid proteins IHF, Fis, and HU (246, 309).

The SeqA protein, a negative regulator of replication initiation, binds to hemimethylated GATC Dam (DNA adenine methylation) sequences at two sites within the newly replicated oriC region (169). Controlling the activity of DnaA initiator protein by SeqA regulates the timing and frequency of genome replication. During DNA replication, the DnaA protein is stored at several sites outside the oriC site. DnaA-associating proteins, such as IciA, Hda, Cnu, and DiaA (124), are all involved in this control. For the precise localization of the oriC region along the cell membrane, a set of specific proteins is involved, including the ParA/ParB family of proteins (308). Binding of DNA to the bacterial cytoplasmic membrane is important for most of the genome functions. In particular, the binding of the oriC region to the cell membrane is critical for the genome replication to take place (215, 256).

DNA replication takes place in bidirectional fashion and thus divides the circular genome into equal arms. Sequential replication segregates two nascent nucleoids to locate into two cell halves (300). During DNA replication, the genome is always associated at specific sites on the cell membrane, where the replication apparatus or replication factory is associated (51). The main component of the replication apparatus is DNA polymerase III (Pol III), which is composed of 10 different subunits and catalyzes DNA polymerization (Table 2). The replication of both clockwise and anticlockwise forks are located at a single site, suggesting that the genome DNA is fed through the fixed replication complex, extruding the daughter replication duplexes from a single site toward opposite cell pole (66). Thus, each replication apparatus contains two molecules of DNA polymerase III, one for the leading strand and one for the lagging strands. The newly replicated oriC regions of the daughter genomes move apart abruptly, far enough for them to be equally spaced when the cell has doubled in length and the next round initiated. The replication of E. coli genome is also terminated at a specific site, called terC, and the termination takes place near midcell (where the division septum forms). Termination factor Tus recognizes the terC signal and prevents further migration of the replication apparatus along the genome (113, 153) (Table 2). After replication, the active movement of the terC region takes place as is the case for oriC (99). In the newly divided cells, both oriC are terC are located close to the newly formed cell pole, and then move rapidly to the midcell. The movement of terC is due to reorganization of the nucleoid driven by the oriC movement (258).

Overall more than 30 proteins are transiently associated for regulated replication of the genome DNA (Table 2). During the elongation stage of DNA replication, approximately half of these proteins migrate together with the replication apparatus, while the rest of the replication-related proteins are associated at either the oriC or ter regions for control of the timing and site of replication initiation and termination (66).

Replication of the genome DNA generates intertwined DNA intermediates, which must be decatenated for subsequent genome segregation. Catenates, two or more DNA molecules interlinked, are important intermediates in DNA replication and recombination. DNA catenates and knots are, however, problematic to the DNA functions if they are left unresolved after replication and recombination. Two type II topoisomerases (DNA gyrase and topoisomerase IV) play the major role in regenerating the steady state of negative supercoil of the newly replicated genomes (299). DNA gyrase removes positive supercoils caused by DNA replication while Topo IV plays a role in both decatenation of daughter genomes after replication and removing positive supercoils, both together leading to generation of the steady-state negative supercoils in the genome (63, 109, 277, 324, 325). Negative supercoiling favors any process that requires double-helix opening, including DNA replication, transcription, and recombination (139).

In the middle of nucleoid division and segregation, MukBEF condensin participates in the condensation of the newly replicated genome into the compact nucleoid (209, 318), which is different from the nucleoid in nondividing cells. The mukB mutants exhibit less-condensed nucleoids and defects in nucleoid segregation (209), indicating that MukB is acting as a nucleoid-condensing factor.

Nucleoid Segregation Apparatus

The newly duplicated daughter genomes separate from each other, forming two nucleoids. The nucleoid segregation ensures the spatial separation of two sister genomes into daughter cells prior to cell division (258). Nucleoid partitioning during the cell division consists of two different events: movement of the daughter oriC from the parental cell center to the pole-proximal positions, and condensation of the newly replicated genome (310). In sharp contrast to eukaryotes, replication and partition in bacteria are parallel processes. Overall, the segregation of the newly formed nucleoid moves slowly, but the oriC and ter loci move faster into daughter cells. The low diffusion coefficient of the nucleoid is due to association of many DNA-binding proteins with the genome.

A specific set of proteins is involved in the partitioning, condensation, transport, and segregation of the oriC locus and the newly formed genome into daughter cells (94). Membrane-bound Par proteins are involved in the formation of the nascent nucleoid and its association with the membrane. For separation of the newly formed and condensed nucleoid, the prokaryotic mitotic apparatus, which is composed of the actin homolog MreB, is supposed to participate (157). MreB forms helical filaments, as does the eukaryotic actin. Faithful chromosome segregation during mitosis in eukaryotic cells is carried out by the tubulin-based cytoskeleton, while in E. coli this task is accomplished by the MreB filament (157, 158). MreB is also required for the nonspherical shape of bacterial cells.

After the partition of the newly replicated genome, septum formation starts with the FtsZ ring formation. The application of cytological methods in conjunction with fluorescence microscopy allows localization of specific genome segments within the nucleoid (286). For separation of the nucleoid during cell division, the cell division protein FtsZ assembles into a ring-like structure known as the Z ring along the inside of the membrane at the site of cell division (18). In addition to FtsZ protein, the Z ring involves at least 10 more associated proteins (35, 242). Positioning of the cell division constriction depends on nucleoid replication activity and thus Z rings tend to form in nucleoid-free regions (188, 189). The daughter nucleoids interfere with Z-ring assembly (311), by close contact with the inside surface of the cell membrane. The nucleation of FtsZ polymerization therefore takes place in the middle of cell, where the nucleoid segregation creates a space.

Nucleoids are capable of topologically excluding FtsZ rings, leading to restricted assembly of the Z ring only at the nucleoid-free areas (279). The cell cycle is determined by the cooperative activity of min operon products, MinC, MinD, and MinE. These cell-cycle-related proteins undergo more marked movements: the MinD protein oscillates from one cell pole to another cell pole on a timescale of seconds (235), for positioning the FtsZ ring at the sites other than the midcell. Formation of phospholipid domains within the cell membranes provides the attachment site for MinD, together forming the proteolipid tube to provide the contact surface of FtsZ ring (212). The mechanism that prevents FtsZ ring formation at the poles involves the MinCD complex as well as a topological specificity (62).

The MinE protein also exhibits remarkable dynamism, migrating quickly from one pole to another reflecting the oscillation of MinD (108). As MinE moves toward the opposite pole, the MinCD complex is ejected from the membrane and moves to the newly formed MinE band. This cycle takes place within a few seconds so that multiple oscillations take place during each round of cell division. One possibility of the rapid movement of Min complexes is to prevent incorrect assembly of the division machinery near the cell poles, but promote the assembly at the midcell sites close to the replication machinery (62).

The nucleoid is associated with the bacterial cytoplasmic membrane. Not only direct membrane-DNA interactions, but also binding of nascent membrane proteins to the membrane could provide topological restrains on the DNA of the nucleoid (151). Membrane-associated nascent protein-template mRNA complexes can act to anchor the associated RNA polymerase, thereby restricting rotation of the bound DNA template (171). This anchored site of transcription therefore has the potential of defining the boundary of a nucleoid domain. Plasmids carrying genes specifying membrane proteins are tightly associated with membranes, indicating that nascent membrane proteins are the possible linkers between the nucleoids and membranes (52, 171, 231). This type of interaction would create localized domains in the nucleoid. However, several lines of evidence indicate that the formation of about 100-kbp topological domains of nucleoids measured in vivo is independent of nascent membrane proteins and their associated mRNA chains. These findings led to the proposal that similar interactions in the nucleoid at sites of genes specifying membrane proteins could result in the segregation of the chromosome into topological domains.

A DNA segment carrying a cluster of recognition sites by DNA-binding domains gathers a number of DNA-binding proteins to form patches, initially through DNA-protein interactions followed by protein-protein interactions. Sequence-specific and nonspecific interaction between a long DNA segment and a patch of DNA-binding proteins may represent the gene sequestration model of gene silencing (148).

Recombination-Repair Apparatus

A total of about 40 DNA-binding proteins are involved in the pathways of recombination and repair (Table 3). RecA recombinase plays a major role in homologous recombination because recA is the only gene whose inactivation completely blocks homologous recombination (241). RecA protein polymerizes into spiral filaments on ssDNA in the presence of SSB (single-strand binding protein) in the 5′-to-3′ direction. The resulting RecA filaments serve as substrates for homologous pairing and strand exchange, the most critical step in recombinational repair. The RecA polymerization and depolymerization depend on other Rec proteins such as RecBCD, RecF, RecO, and RecR (48). These proteins work prior to RecA in the recombination pathways. RecBC carries the activities of DNA end-specific helicase, ATP-dependent ssDNA and dsDNA exonucleases, and ATP-dependent ssDNA endonuclease. RecF, RecO, and RecR proteins stimulate RecA polymerization on ssDNA (156).

The products of ruvABC and recG genes work after the RecA-catalyzed steps. RuvABC resolvasome binds four-way DNA junctions (Holiday junctions) and isomerizes them from the folded conformation to the square planar conformation, thereby resolving the Holiday junctions (303). RecG helicase also participates in DNA junction removal (180). Replication fork assembly by RecA recombinase complex is followed by the subsequent PriA-dependent initiation of DNA synthesis. Thus, the events after DNA strand exchanges in homologous recombination depend on the genes whose products participates in DNA replication such as PolA, DnaE, DnaB, PriA, and TopA (see Table 2 and Table 3).

The genetic requirements for the DNA damage-induced repair recombination are essentially the same as those for the homologous recombination. DNA damage takes place generally only on one DNA strand, which is effectively removed by several excision repair systems. The repair system is induced after treatment of E. coli with DNA-damaging agents, and this SOS system is stored for a certain period, keeping the repair-proficient state for better survival after subsequent damaging treatment (54).

Transcription Apparatus

The absence of nuclear membrane, together with the evidence for close coupling of transcription and translation, is in sharp contrast to their striking separation into distinct compartments in eukaryotes. Transcription is thus the major step of regulation of gene expression in prokaryotes. Transcription of the E. coli genome is carried out by a single species of RNA polymerase core enzyme with the subunit structure α₂ββ′ω but for promoter recognition, one of seven species of E. coli σ-subunit is needed (36, 125, 219) (Table 4). Each of the seven E. coli σ-subunits recognizes a specific promoter sequence but only when bound to the core enzyme. Except for RpoN sigma (σ⁵⁴), the σ-subunits alone are unable to form stable promoter complexes.

The activity and functional specificity such as promoter recognition of RNA polymerase is further modulated by interaction with about 300 species of transcription factor (Table 5), of which the regulatory functions remain unidentified for about one third (125, 127, 177). DNA-binding transcription factors from E. coli can be classified into 54 families depending on their protein motifs (Table 5). Identification of the regulatory roles for these putative transcription factors is one of the important subjects in E. coli research. Transcriptome and/or proteome analyses with use of mutant E. coli lacking each one of the factor genes are sometimes useful, but the discrimination between direct and indirect effects after gene disruption is often difficult because the transcription factors form complex interplay networks (125, 177). Instead the genomic SELEX (systematic evolution of ligands by exponential enrichment) screening in vitro for DNA recognition sequences by the purified transcription factors is useful for quick and direct search of the regulation targets (for instance see references 213, 214, 260, and 261). The microarray assay using the “promoter chip” is another choice for direct identification of target promoters recognized by each transcription factor (317). On the other hand, the ChIP-chip is a useful method for in vivo search of the target genes or promoters by transcription factors (104, 111).

Most of the transcription factors control one particular gene or operon, forming simple regulons, but some specific sets of genes or operons form complex regulons in which multiple transcription factors are involved for regulation (see Fig. 7). In eukaryotes, a number of transcription factors are involved in promoter regulation, including general transcription factors, specific factors, and mediators (or coactivators) connecting general and specific factors. Recently, however, the multifactor promoters have been identified in prokaryotes. For instance, more than 10 different transcription factors are involved in regulation of the promoter for the csgD gene encoding the master regulator of curli fimbriae synthesis, thereby controlling biofilm formation (cited in reference 213). The biofilm formation must be controlled by sensing a variety of environmental signals, each being recognized by a different transcription factor, which are all involved in the regulation of csgD promoter.

Global regulators were defined on the basis of the number of genes under the control of test factors (101). CRP (cAMP receptor protein or catabolite gene activator CAP) was the first example of this type of regulator, which was originally identified as a factor involved in cAMP-mediated catabolite repression (134, 154, 221). The number of experimentally identified target genes under the control of CRP, LexA (SOS regulon regulator), PurR (purine regulon regulator), Cra (catabolite repressor activator or FruR) and ArcA (aerobic pathway regulator) are more than 40, 20, 20, 15, and 10, respectively. After systematic genomic SELEX search for recognition DNA sequences by these global factors, additional regulation targets can be identified besides the hitherto identified genes (213, 214, 260, 261). Nucleoid proteins, Fis, IHF, H-NS, and HU, belong to this global regulator family. Often these nucleoid proteins work together with gene-, operon-, or regulon-specific transcription factors (see Fig. 7). Marr et al. (175) propose a classification of the mode of transcription regulation into two types: digital control by dedicated regulators binding specific DNA sites and analog control by nucleoid proteins binding multiple sites along the genome, both together leading to generate the marked flexibility in genetic adaptation of single-cell bacteria to environments in nature. Cooperative binding of global and specific regulators increases the sensitivity in response to environmental changes. Transcription factors form complex interplay networks, in which one transcription factor at the top of the network hierarchy controls the gene for another transcription factor. By forming such a hierarchy, transcription factors are able to amplify the range of regulation.

The mode of transcription regulation by the transcription factors is correlated with their DNA-binding sites relative to the promoters (or RNA polymerase-binding sites) of target genes or operons. In general, transcription factors that bind upstream from promoters activate transcription while downstream-bound factors repress transcription. For transcription activation or repression, the DNA-bound factors interact with RNA polymerase at specific sites on one of the subunits (125, 126). Simultaneous binding of both transcription factors and RNA polymerase within the narrow region is needed to increase the local protein concentrations to achieve effecting protein-protein interaction. In the absence of DNA, both components are unable to form stable complexes. A small fraction of the transcription factors, including Rho and NusA termination factors, however, bind to RNA polymerase in the absence of DNA and migrate together with the elongation complex for control transcription termination and/or antitermination (126). A limited number of specific transcription factors such as N4 ssDNA-binding protein also directly interact with RNA polymerase (185).

The intracellular concentrations of transcription factors vary depending on the culture conditions and the cell growth phase. For instance, the intracellular concentration of CRP ranges from 500 to 2,000 molecules per cell or almost half the level of RNA polymerase core enzyme (125, 126). In addition to the level control, the DNA-binding activity of a set of transcription factors is controlled by interaction with specific inducers or corepressors. A total of about 30 transcription factors are organized in two-component systems (TCSs) and their activities are controlled through reversible phosphorylation by the respective cognate sensor kinases (316). Overall, the level of functional transcription factors markedly fluctuates depending on the culture conditions and cell growth phase.

The central part of growing E. coli cells is rich in RNA polymerase, implying that most of the RNA polymerase is associated with the genome (37, 125, 126), whereas the peripheral zone of the cells is enriched in ribosomes, presumably engaged in translation coupled to transcription (165). Immunogold staining with antibodies directed at RNA polymerase indicates tight association of most, if not all, of the RNA polymerase with the nucleoid (146). By using a functional rpoC-gfp gene fusion inserted into the E. coli genome, RNA polymerase is exclusively located within the nucleoid forming fluorescent foci, which correspond to the site of actively transcribed rRNA genes in growing cells (37). Biochemical measurements also indicated that the majority of RNA polymerase is associated with the isolated nucleoid (249). RNA polymerase core enzyme subunits, α, β, and β′, are prominent proteins in nucleoids isolated in both low- and high-salt solutions or in the presence of stabilizing polyamines. These RNA polymerase subunits are the major proteins of nucleoids isolated in the presence of 1.0 M NaCl. The high salt dissociates many weakly associated proteins that would otherwise be more abundant than the RNA polymerase. These are functioning RNA polymerase molecules in the actively transcribing RNA (74, 226). The transcription initiation complex resists to dissociation even after equilibrium centrifugation in CsCl and Cs₂SO₄ (199, 200), indicating that the affinity of RNA polymerase for DNA is strongest during the initiation stage.

Since transcription and translation are usually coupled, it seems that little transcription could occur on DNA sequences located deep in the nucleoid. It has therefore been considered likely that transcriptionally active DNA is located near the nucleoid surface or on DNA loops extending from the nucleoid. In fact, electron microscopic autoradiography of sections of bacteria indicates the association of pulse-labeled nascent RNA with the nucleoid (248). Analysis of thin sections of E. coli cells indicates that the RNA polymerase is localized primarily on the nucleoid surface. The surface location of RNA polymerase is compatible with the suggested location of transcription in the surface of the nucleoid. If transcription were indeed localized in this manner, it would require that the nucleoid DNA have a dynamic organization, so that all genes could potentially cycle from the interior onto the surface of an extruded loop. A dynamic organization was also suggested from earlier studies of nucleoid structure (266).

THE FUTURE

While it is apparent that much has been learned about the nucleoid, it is also evident that the fundamental interactions organizing the structure of DNA in the nucleoid still need to be clearly defined. The available information reviewed above indicates that the final story will involve a set of DNA-protein and protein-protein interactions. More than 400 species of E. coli proteins, among a total of about 4,000 potential protein-coding gene products, can be classified as in the DNA-binding protein family. A small fraction, most notably the nucleoid proteins, is stably associated with the genome while most of other functional proteins transiently associate with the genome. The different DNA-bending proteins, which are involved in organizing short-range structure, seem to have partly overlapping functions, and there are several different players in this system. The molecular interactions organizing long-range structure are even less clear but seem to involve multiple kinds of interactions. The eventual solution of these intriguing mysteries about nucleoid structure will certainly help our overall understanding of the storage and expression of genetic information.

ACKNOWLEDGMENTS

I thank Sankar Adhya and Cathy Squires for careful reading of the manuscript and valuable comments.