Genes VII

So far we have dealt with the promoter as a DNA sequence that is competent to bind RNA polymerase, which then initiates transcription. But there are some promoters at which RNA polymerase cannot initiate transcription without assistance from an ancillary protein. Such proteins are positive regulators, because their presence is necessary to switch on the transcription unit. Typically the activator overcomes a deficiency in the promoter, for example, a poor consensus sequence at V35 or V10.

One of the most widely acting activators is a protein that controls the activity of a large set of operons in E. coli in response to carbon nutrient conditions. When glucose is available as an energy source, it is used in preference to other sugars. So when E. coli finds (for example) both glucose and lactose in the medium, it metabolizes the glucose and represses the use of lactose.

This choice is accomplished by preventing expression of several operons, including lac, gal, and ara. The effect is called catabolite repression. It represents a general coordinating system that exercises a preference for glucose by inhibiting the expression of the operons that code for the enzymes of alternative metabolic pathways.

Catabolite repression is set in train by the ability of glucose to reduce the level of cyclic AMP (cAMP) in the cell. Cyclic AMP is synthesized by the enzyme adenylate cyclase. The reaction uses ATP as substrate and introduces a 3′ V5′ link via phosphodiester bonds, generating the structure drawn in Figure 10.22. Mutations in the gene coding for adenylate cyclase (cya ^V) do not respond to changes in glucose levels.

Expression of the catabolite-regulated operons shows an inverse relationship with the level of cyclic AMP, which acts by binding to the product of cap gene. Mutations in the cap gene prevent activation of the operons that normally are expressed in the absence of glucose. The protein is known as CAP (for catabolite activator protein). It is a positive control factor whose presence is necessary to initiate transcription at dependent promoters. The protein is active only in the presence of cyclic AMP, which behaves as the classic small-molecule inducer (see Figure 10.21; upper right).

Figure 10.23 shows that reducing the level of cyclic AMP renders the (wild-type) protein unable to bind to the control region, which in turn prevents RNA polymerase from initiating transcription. So the effect of glucose in reducing cyclic AMP levels is to deprive the relevant operons of a control factor necessary for their expression.

The CAP factor binds to DNA, and complexes of cyclic AMP PCAP PDNA can be isolated at each promoter at which it functions. The factor is a dimer of two identical subunits of 22.5 kD, which can be activated by a single molecule of cyclic AMP. A CAP monomer contains a DNA-binding region and a transcription-activating region (for review see Botsford and Harman, 1992).

A CAP dimer binds to a site of ~22 bp at a responsive promoter. The binding sites include variations of the consensus sequence given in Figure 10.24. Mutations preventing CAP action usually are located within the well conserved pentamer ^TGTGA_ACACT, which appears to be the essential element in recognition. CAP binds most strongly to sites that contain two (inverted) versions of the pentamer, because both subunits of the dimer bind effectively to the DNA. Many binding sites lack the second pentamer, however, and in these the second subunit must bind a different sequence (if it binds to DNA). The hierarchy of binding affinities for CAP helps to explain why different genes are activated by different levels of cyclic AMP in vivo.

The action of CAP has the curious feature that its binding sites lie at different locations relative to the startpoint in the various operons that it regulates. And the TGTGA pentamer may lie in either orientation. The three examples summarized in Figure 10.25 encompass the range of locations:

• Sometimes the CAP-binding site lies within the promoter, as in the gal locus, where the CAP-binding site is centered on V41. It is likely that only a single CAP dimer is bound, probably in quite intimate contact with RNA polymerase, since the CAP-binding site extends well into the region generally protected by the RNA polymerase.
• The CAP-binding site is adjacent to the promoter, as in the lac operon, in which the region of DNA protected by CAP is centered on V61. It is possible that two dimers of CAP are bound. The binding pattern is consistent with the presence of CAP largely on one face of DNA, the same face that is bound by RNA polymerase. This location would place the two proteins just about in reach of each other.
• In other operons, the CAP-binding site lies well upstream of the promoter. In the ara region, the binding site for a single CAP is the farthest from the startpoint, centered at V92. Here the CAP cannot be in contact with RNA polymerase, because another regulatory protein binds in the region between the CAP and RNA polymerase sites.

Dependence on CAP is related to the intrinsic efficiency of the promoter. No CAP-dependent promoter has a good V35 sequence and some also lack good V10 sequences. In fact, we might argue that effective control by CAP would be difficult if the promoter had effective V35 and V10 regions that interacted independently with RNA polymerase.

There are in principle two ways in which CAP might activate transcription: it could interact directly with RNA polymerase; or it could act upon DNA to change its structure in some way that assists RNA polymerase to bind. In fact, CAP has effects upon both RNA polymerase and DNA (for review see Kolb, 1993).

Binding sites for CAP at most promoters resemble either gal (centered at V41 bp) or lac (centered at V61). When the distance is changed from either of these two standards, the ability of CAP to activate transcription is reduced. A unifying model for the ability of CAP to activate transcription in spite of its varying distance from the startpoint is suggested by the correlation that activation occurs only when this distance is an integral number of turns of the double helix. This suggests that CAP must be bound to the same face of DNA as RNA polymerase.

When the a subunit of RNA polymerase has a deletion in the C-terminal end, transcription appears normal except for the loss of ability to be activated by CAP. This suggests that CAP acts directly upon the a subunit to stimulate RNA polymerase. Experiments using CAP dimers in which only one of the subunits has a functional transcription-activating region shows that, when CAP is bound at the lac promoter, only the activating region of the subunit nearer the startpoint is required, presumably because it touches RNA polymerase. This offers an explanation for the lack of dependence on the orientation of the binding site: the dimeric structure of CAP ensures that one of the subunits is available to contact RNA polymerase, no matter which subunit binds to DNA and in which orientation. The activating region in CAP consists of a small exposed loop, in which point mutations block the interaction with RNA polymerase.

The effect upon RNA polymerase binding depends on the relative locations of the two proteins. When CAP binds within the promoter (as in gal), it increases the rate of transition from the closed to open complex. When CAP binds adjacent to the promoter (as in lac) its predominant effect is to increase the rate of initial binding to form a closed complex. This suggests that the exact effects of the interaction depend upon the geometry of the proteins at the individual promoter. (The interactions at the ara promoter may involve different interactions, involving more proteins.)

The structure of the CAP-DNA complex is interesting: the DNA has a bend. Proteins may distort the double helical structure of DNA when they bind, and several regulator proteins induce a bend in the axis.

Figure 10.26 illustrates a technique that can be used to measure the extent and location of a bend. A dimer of the target sequence is made, and it is cut with different restriction enzymes to generate a set of circularly permutated fragments each containing a monomeric length of DNA. The protein-binding site therefore lies at a different location in each of these fragments.

The fragments move at different speeds in an electrophoretic gel, depending on the position of the bend. (If there is no bend, all fragments move at the same rate.) The greatest impediment to motion, causing the lowest mobility, happens when the bend is in the center of the DNA fragment. The least impediment to motion, allowing the greatest mobility, happens when the bend is at one end.

The results are analyzed by plotting mobility against the site of restriction cutting. The low point on the curve identifies the situation in which the restriction enzyme has cut the sequence immediately adjacent to the site of bending.

For the interaction of CAP with the lac promoter, this point lies at the center of dyad symmetry. The bend is quite severe, >90 X, as illustrated in the model of Figure 10.27. There is therefore a dramatic change in the organization of the DNA double helix when CAP protein binds. The mechanism of bending is to introduce a sharp kink within the TGTGA consensus sequence; the two kinks in each copy present in a palindrome cause the overall 90 X bend. It is possible that the bend has some direct effect upon transcription, but it could be the case that it is needed simply to allow CAP to contact RNA polymerase at the promoter (Gaston et al., 1990).

Whatever the exact means by which CAP activates transcription at various promoters, it accomplishes the same general purpose: to turn off alternative metabolic pathways when they become unnecessary because the cell has an adequate supply of glucose. Again, this makes the point that coordinate control, of either negative or positive type, can extend over dispersed loci by repetition of binding sites for the regulator protein.