Genes VII

Although bacteria do not display structures with the distinct morphological features of eukaryotic chromosomes, their genomes nonetheless are organized into definite bodies. The genetic material can be seen as a fairly compact clump or series of clumps that occupies about a third of the volume of the cell. Figure 18.4 displays a thin section through a bacterium in which this nucleoid is evident (for review see Brock, 1988).

When E. coli cells are lysed, fibers are released in the form of loops attached to the broken envelope of the cell. As can be seen from Figure 18.5, the DNA of these loops is not found in the extended form of a free duplex, but is compacted by association with proteins.

Several DNA-binding proteins with a superficial resemblance to eukaryotic chromosomal proteins have been isolated in E. coli. What criteria should we apply for deciding whether a DNA-binding protein plays a structural role in the nucleoid? It should be present in sufficient quantities to bind throughout the genome. And mutations in its gene should cause some disruption of structure or of functions associated with genome survival (for example, segregation to daughter cells). None of the candidate proteins yet satisfies the genetic conditions (for review see Drlica and Rouviere-Yaniv, 1987).

Protein HU is a dimer that condenses DNA, possibly wrapping it into a bead-like structure. It stimulates DNA replication (see 13 DNA replication). It is related to IHF (integration host factor), another dimer, which has a structural role in building a protein complex in some specialized recombination reactions, including the integration and excision of phage lambda (for which it is named; see 14 Recombination and repair). Null mutations in either of the genes coding for the subunits of HU (hupA,B) have little effect, but loss of both functions causes a cold-sensitive phenotype and some loss of superhelicity in DNA. These results raise the possibility that HU plays some general role in nucleoid condensation.

Protein H1 (also known as H-NS) binds DNA, interacting preferentially with sequences that are bent. Mutations in its gene have turned up in a variety of guises (osmZ, bglY, pilG), each identified as an apparent regulator of a different system. These results probably reflect the effect that H1 has on the local topology of DNA, with effects upon gene expression that depend upon the particular promoter.

We might expect that the absence of a protein required for nucleoid structure would have serious effects upon viability. Why then are the effects of deletions in the genes for proteins HU and H1 relatively restricted? One explanation is that these proteins are redundant, that any one can substitute for the others, so that deletions of all of them would be necessary to interfere seriously with nucleoid structure. Another possibility is that we have yet to identify the proteins responsible for the major features of nucleoid integrity.

The nucleoid can be isolated directly in the form of a very rapidly sedimenting complex, consisting of ~80% DNA by mass. (The analogous complexes in eukaryotes have ~50% DNA by mass; see later.) It can be unfolded by treatment with reagents that act on RNA or protein. The possible role of proteins in stabilizing its structure is evident. The role of RNA has been quite refractory to analysis.

The DNA of the compact body isolated in vitro behaves as a closed duplex structure, as judged by its response to ethidium bromide. This small molecule intercalates between base pairs to generate positive superhelical turns in "closed" circular DNA molecules, that is, molecules in which both strands have covalent integrity. (In "open" circular molecules, which contain a nick in one strand, or with linear molecules, the DNA can rotate freely in response to the intercalation, thus relieving the tension.)

In a natural closed DNA that is negatively supercoiled, the intercalation of ethidium bromide first removes the negative supercoils and then introduces positive supercoils. The amount of ethidium bromide needed to achieve zero supercoiling is a measure of the original density of negative supercoils.

Some nicks occur in the compact nucleoid during its isolation; they can also be generated by limited treatment with DNAase. But this does not abolish the ability of ethidium bromide to introduce positive supercoils. This capacity of the genome to retain its response to ethidium bromide in the face of nicking means that it must have many independent domains; the supercoiling in each domain is not affected by events in the other domains.

This autonomy suggests that the structure of the bacterial chromosome has the general organization depicted diagrammatically in Figure 18.6. Each domain consists of a loop of DNA, the ends of which are secured in some (unknown) way that does not allow rotational events to propagate from one domain to another. There are ~100 such domains per genome; each consists of ~40 kb (13 µm) of DNA, organized into some more compact fiber whose structure has yet to be characterized.

The existence of separate domains could permit different degrees of supercoiling to be maintained in different regions of the genome. This could be relevant in considering the different susceptibilities of particular bacterial promoters to supercoiling (see 9 Transcription).

Supercoiling in the genome can in principle take two forms:

• If a supercoiled DNA is free, its path is unrestrained, and negative supercoils generate a state of torsional tension that is transmitted freely along the DNA within a domain. It can be relieved by unwinding the double helix, as described in 14 Recombination and repair. The DNA is in a dynamic equilibrium between the states of tension and unwinding.
• Supercoiling can be restrained if proteins are bound to the DNA to hold it in a particular three-dimensional configuration. In this case, the supercoils are represented by the path the DNA follows in its fixed association with the proteins. The energy of interaction between the proteins and the supercoiled DNA stabilizes the nucleic acid, so that no tension is transmitted along the molecule.

Are the supercoils in E. coli DNA restrained in vivo or is the double helix subject to the torsional tension characteristic of free DNA? Measurements of supercoiling in vitro encounter the difficulty that restraining proteins may have been lost during isolation. Various approaches suggest that DNA is under torsional stress in vivo, although it is difficult to quantitate the level of supercoiling.

A direct approach is to use the crosslinking reagent psoralen, which binds more readily to DNA when it is under torsional tension. The reaction of psoralen with E. coli DNA in vivo corresponds to an average density of one negative superhelical turn / 200 bp (σ = V0.05).

Another approach is to examine the ability of cells to form alternative DNA structures; for example, to generate cruciforms at palindromic sequences. From the change in linking number that is required to drive such reactions, it is possible to calculate the original supercoiling density. This approach suggests an average density of σ = V0.025, or one negative superhelical turn / 100 base pairs.

So supercoils do create torsional tension in vivo. There may be variation about an average level, and although the precise range of densities is difficult to measure, it is clear that the level is sufficient to exert significant effects on DNA structure, for example, in assisting melting in particular regions such as origins or promoters.

Many of the important features of the structure of the compact nucleoid remain to be established. What is the specificity with which domains are constructed Xdo the same sequences always lie at the same relative locations, or can the contents of individual domains shift? How is the integrity of the domain maintained? Biochemical analysis by itself is unable to answer these questions fully, but if it is possible to devise suitable selective techniques, the properties of structural mutants should lead to a molecular analysis of nucleoid construction.