cAMP gives CAP a twist

The catabolite activator protein (CAP), which plays a significant role in telling bacterial metabolism to digest sugars other than glucose, is a classic example of allosteric activation. Binding of the small signaling molecule cyclic AMP (cAMP) switches CAP into an active state that recruits RNA polymerase to certain metabolic genes. The biochemistry of cAMP activation is well understood, but the structural basis is not as clear, because a structure of the inactive protein was not available. This week in PNAS, researchers from Rutgers University and the University of Wisconsin-Madison report a structure of the free state of CAP that illuminates this allosteric effect.

Many excellent structural studies have examined CAP in its activated and DNA-bound form. CAP is a dimer, and each monomer has two domains: a DNA-binding domain (DBD) that recognizes its specific sequence, and a cAMP-binding domain (CBD). The monomers bind to each other through a coiled-coil interaction between two long helices. When CAP is activated it binds to DNA, with two helices (the recognition helices) sliding into the major groove and specifically identifying the sequences to which it should recruit the transcription apparatus. Without cAMP bound, CAP can still interact with DNA, but this interaction is of low affinity and not specific for any particular sequence. There are a number of ways this could conceivably happen, but it's difficult to be certain about any model in the absence of a structure of the free (apo-) protein.

In order to determine the structure, Popovych et al. used NMR. The 50 kDa size of the dimer means that it requires some extra effort for NMR work, but it is still well within the capabilities of the technique. The fact that the protein is a symmetric homodimer makes assigning the spectra somewhat easier, as the researcher only needs to deal with 209 residues rather than 418. The authors determined the structure using short-range distance restraints from nOe experiments, long-range restraints from paramagnetic relaxation enhancement, and angular restraints from residual dipolar couplings (RDCs). These angular restraints allowed the authors to unambiguously determine the relative orientation of the DBD and CBD in each monomer.

Getting that orientation right is key to the story here, as you can see from the image to the left. Here I'm showing you the DBD and coiled-coil helix (lower left) of a single monomer in the two different states. The activated CAP is in green (PDB code: 1G6N, and the apo- structure is in red (PDB code: 2WC2). You're looking down the coiled-coil, and the recognition helix is in a brighter color right at the front. If you superpose these structures on the bottom end of the coiled-coil, you can see that the recognition helix is rotated by 60° when cAMP binds. This twist of the DNA binding domain gives the recognition helices the right orientation and spacing to slide into the major groove of DNA and identify genes to activate. In the apo- state, these helices cannot both fit into the major groove simultaneously, explaining the low affinity and lack of specificity in that state.

Although the DBDs undergo a radical change in position following cAMP binding, they don't actually have any direct interactions with the signaling molecule, which binds down in the CBD near the coiled-coil helix. This helix, which links the CBD to the DBD, turns out to be key to communicating the allosteric signal. In the apo- state, the top part of this helix (near the DBD) is actually somewhat disordered and loop-like, not helical. Binding of cAMP to the CBD forms several contacts and causes several structural shifts that result in the formation of regular helical structure at the top of the coiled-coil. This in turn swings the DBDs around so that the recognition helices are in position to interact with target sequences (the authors provide a short movie of this in the Supplementary Information). A similar molecule, cGMP, that does not activate CAP, fails to make the key contacts with T127 and S128 that mediate this structural change.

The fact that the upper part of the coiled-coil is unstructured suggests that CAP may sample a range of conformations in the apo- state. This possibility is supported by one of the mutational studies in the paper. As you can see from Figure 5, a G141S mutation and binding of various effectors to the mutant causes the NMR resonances of DBD residues to shift on a line between the WT apo- and WT cAMP-bound states. This, in conjunction with the broadening of those intermediate peaks, suggests that the DBDs are exchanging between the two states on a timescale of microseconds. It seems quite likely that one or both DBDs in the inactive dimer occasionally samples the active conformation. In this model, the function of cAMP would be to stabilize, rather than enable the active conformation. The negative cooperativity of cAMP binding may help keep CAP switched "off" in the face of this conformational heterogeneity.

This study only dealt with a single protein, but the results are likely to be applicable to a number of systems. The allosteric mechanism described here seems to fit observations in at least some other members of the protein family to which CAP belongs. As such, this structural work and the dynamics investigations that will probably ensue are likely to provide important insights into a number of regulatory pathways in bacteria.

Popovych, N., Tzeng, S., Tonelli, M., Ebright, R., & Kalodimos, C. (2009). Structural basis for cAMP-mediated allosteric control of the catabolite activator protein Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0900595106