Genes VII

The ability of RNA polymerase (or indeed any protein) to recognize DNA can be characterized by footprinting. A sequence of DNA bound to the protein is partially digested with an endonuclease to attack individual phosphodiester bonds within the nucleic acid. Under appropriate conditions, any particular phosphodiester bond is broken in some, but not in all, DNA molecules. The positions that are cleaved are recognized by using DNA labeled on one strand at one end only. The principle is the same as that involved in DNA sequencing; partial cleavage of an end-labeled molecule at a susceptible site creates a fragment of unique length.

As Figure 9.15 shows, following the nuclease treatment, the broken DNA fragments are recovered and electrophoresed on a gel that separates them according to length. Each fragment that retains a labeled end produces a radioactive band. The position of the band corresponds to the number of bases in the fragment. The shortest fragments move the fastest, so distance from the labeled end is counted up from the bottom of the gel.

In a free DNA, every susceptible bond position is broken in one or another molecule. But when the DNA is complexed with a protein, the region covered by the DNA-binding protein is protected in every molecule. So two reactions are run in parallel: a control of pure DNA; and an experimental mixture containing molecules of DNA bound to the protein. When a bound protein blocks access of the nuclease to DNA, the bonds in the bound sequence fail to be broken in the experimental mixture.

In the control, every bond is broken, generating a series of bands, one representing each base. There are 31 bands in the figure. In the protected fragment, bonds cannot be broken in the region bound by the protein, so bands representing fragments of the corresponding sizes are not generated. The absence of bands 9 V18 in the figure identifies a protein-binding site covering the region located 9 V18 bases from the labeled end of the DNA. By comparing the control and experimental lanes with a sequencing reaction that is run in parallel it becomes possible to "read off" the corresponding sequence directly, thus identifying the nucleotide sequence of the binding site.

As described previously (see Figure 9.10), RNA polymerase initially binds the region from V50 to +20. The points at which RNA polymerase actually contacts the promoter can be identified by modifying the footprinting technique to treat RNA polymerase Ppromoter complexes with reagents that modify particular bases. The common feature of all the types of modification is that they allow a breakage to be made at the corresponding bond in the polynucleotide chain. The site of breakage can be identified by the same approach used in footprinting. We can perform the experiment in two ways:

• The direct analogy with footprinting is to treat an RNA polymerase PDNA complex with a modifying agent and to compare its susceptibility with that of free DNA. Some bands disappear, identifying sites at which the enzyme has protected the promoter against modification. Other bands increase in intensity, identifying sites at which the DNA must be held in a conformation in which it is more exposed.
• The reverse experiment can be performed by modifying the DNA first; then it is bound to RNA polymerase. Those DNA molecules that cannot bind RNA polymerase are recovered and treated in the usual way to generate strand breaks whose positions can be identified. This locates points at which prior modification prevents RNA polymerase from binding to DNA.

These changes in sensitivity reveal the geometry of the complex, as summarized in Figure 9.16 for a typical promoter. The regions at V35 and V10 contain most of the contact points for the enzyme. Within these regions, the same sets of positions tend both to prevent binding if previously modified, and to show increased or decreased susceptibility to modification after binding. Although the points of contact do not coincide completely with sites of mutation, they occur in the same limited region.

It is noteworthy that the same positions in different promoters provide the contact points, even though a different base is present. This indicates that there is a common mechanism for RNA polymerase binding, although the reaction does not depend on the presence of particular bases at some of the points of contact. This model explains why some of the points of contact are not sites of mutation. Also, not every mutation lies in a point of contact; they may influence the neighborhood without actually being touched by the enzyme (500).

It is especially significant that the experiments with prior modification identify only sites in the same region that is protected by the enzyme against subsequent modification. These two experiments measure different things. Prior modification identifies all those sites that the enzyme must recognize in order to bind to DNA. Protection experiments recognize all those sites that actually make contact in the binary complex. The protected sites include all the recognition sites and also some additional positions, which suggests that the enzyme first recognizes a set of bases necessary for it to "touch down," and then extends its points of contact to additional bases.

The region of DNA that is unwound in the binary complex can be identified directly by chemical changes in its availability. When the strands of DNA are separated, the unpaired bases become susceptible to reagents that cannot reach them in the double helix. Such experiments implicate positions between V9 and +3 in the initial melting reaction. The region unwound during initiation therefore includes the right end of the V10 sequence and extends just past the startpoint.

Viewed in three dimensions, the points of contact upstream of the V10 sequence all lie on one face of DNA. This can be seen in the lower drawing in Figure 9.16, in which the contact points are marked on a double helix viewed from one side. Most lie on the coding strand. These bases are probably recognized in the initial formation of a closed binary complex. This would make it possible for RNA polymerase to approach DNA from one side and recognize that face of the DNA. As DNA unwinding commences, further sites that originally lay on the other face of DNA can be recognized and bound.

The importance of strand separation in the initiation reaction is emphasized by the effects of supercoiling. Both prokaryotic and eukaryotic RNA polymerases can initiate transcription more efficiently in vitro when the template is supercoiled, presumably because the supercoiled structure requires less free energy for the initial melting of DNA in the initiation complex.

The efficiency of some promoters is influenced by the degree of supercoiling. The most common relationship is for transcription to be aided by negative supercoiling. We understand in principle how this assists the initiation reaction. But why should some promoters be influenced by the extent of supercoiling while others are not? One possibility is that the dependence of a promoter on supercoiling is determined by its sequence. This would predict that some promoters have sequences that are easier to melt (and are therefore less dependent on supercoiling), while others have more difficult sequences (and have a greater need to be supercoiled). An alternative is that the location of the promoter might be important if different regions of the bacterial chromosome have different degrees of supercoiling.

Supercoiling also has a continuing involvement with transcription. As RNA polymerase transcribes DNA, unwinding and rewinding occurs, as illustrated in Figure 9.3. This requires that either the entire transcription complex rotates about the DNA or the DNA itself must rotate about its helical axis. The consequences of the rotation of DNA are illustrated in Figure 9.17 in the twin domain model for transcription. As RNA polymerase pushes forward along the double helix, it generates positive supercoils (more tightly wound DNA) ahead and leaves negative supercoils (partially unwound DNA) behind. For each helical turn traversed by RNA polymerase, +1 turn is generated ahead and V1 turn behind (506).

Transcription therefore has a significant effect on the (local) structure of DNA. As a result, the enzymes gyrase (introduces negative supercoils) and topoisomerase I (removes negative supercoils) are required to rectify the situation in front of and behind the polymerase, respectively. If the activities of gyrase and topoisomerase are interfered with, transcription causes major changes in the supercoiling of DNA. For example, in yeast lacking an enzyme that relaxes negative supercoils, the density of negative supercoiling doubles in a transcribed region. A possible implication of these results is that transcription is responsible for generating a significant proportion of the supercoiling that occurs in the cell.

A similar situation occurs in replication, when DNA must be unwound at a moving replication fork, so that the individual single strands can be used as templates to synthesize daughter strands. Solutions for the topological constraints associated with such reactions are indicated later in Figure 14.15.