J Physiol Volume 539, Number 2, 331-, March 1, 2002 DOI: 10.1113/jphysiol.2002.017210
Journal of Physiology (2002), 539.2, p. 331
© Copyright 2002 The Physiological Society
DOI: 10.1113/jphysiol.2002.017210
The perplexing challenges of a pump turned channel
John F. Hunt
Department of Biological Sciences, 702A Fairchild Center, MC2434, Columbia University, New York, NY 10027, USA
Although the cystic fibrosis transmembrane conductance regulator (CFTR) functions as an ATP-gated anion channel, it is homologous to a family of ATP-powered solute pumps called ATP-binding cassette (ABC) transporters. This family derives its name from its stereotyped motor domains or ATP-binding cassettes that are called nucleotide-binding domains (NBDs) in the CFTR literature. NBDs contain a structural core ~250 residues in length that shows at least 25 % sequence identity between most proteins in the superfamily. They function as either homodimers or heterodimers in conjunction with a pair of transmembrane domains (TMDs) which share weaker sequence homology. Recently, high-resolution crystal structures have been determined for five NBDs (reviewed by Thomas & Hunt, 2001) and a low-resolution structure has been determined for an intact ABC transporter (Chang & Roth, 2001). This information should provide a foundation for understanding the structural mechanics of the phosphorylation-regulated and ATP-gated channel opening in CFTR.
However, several questions need to be answered before reliable structural models of gating can be developed. (1) Do NBDs form a Rad50-like nucleotide-sandwich complex? The dimeric architecture of ABC transporters, coupled with the twofold positive kinetic cooperativity exhibited by some, led to an assumption that the NBDs form a conserved dimer. However, the NBD crystal structures show no consistent pattern of oligomerization. Nonetheless, the ATPase active site in the monomer is mysteriously solvent exposed for a mechanoenzyme, suggesting that it must be completed by interactions with either another NBD and/or a TMD in intact transporters (Thomas & Hunt, 2001). A solution to this conundrum was offered by the crystal structure of Rad50 (Hopfner et al. 2000), a DNA-repair enzyme that bears remote but unmistakable structural homology to NBDs from ABC transporters. Rad50 dimerizes upon binding ATP, forming a nucleotide-sandwich complex in which dimer interactions complete the active site. This structure looks compelling as the 'LSGGQ' transporter signature sequence, the hallmark of the ABC superfamily, is solvent exposed in NBD monomers but completes the active site in the Rad50 dimer. However, the formation of such a structure by NBDs from an ABC transporter has not been observed, even in the crystal structure of the intact MsbA integral membrane protein (Chang & Roth, 2001), so that its occurrence in ABC transporters remains unproven. (2) Is the interaction between the NBDs positively or negatively cooperative? While Rad50 and some ABC transporters show positive cooperativity and bind two ATP molecules symmetrically, others show negative cooperativity or reciprocating site catalysis, which means that exclusively one of the active sites in the dimer is occupied by nucleotide at any time. Given the ready adaptation of a Rad50-like dimer to function in either way, either model seems possible for CFTR. (3) Does CFTR function as a dimer? This contentious issue (Raghuram et al. 2001 and citations therein) raises complex possibilities in terms of cooperative inter-NBD interactions. (4) Do NBD1 and NBD2 of CFTR interact with each other to form a heterodimer or with themselves to form two homodimers? CFTR contains two tandem NBDs, two tandem TMDs and a regulatory R-domain in a single protein. If CFTR functions as a monomer, NBD1 is likely to interact with NBD2 in a Rad50-like heterodimer complex. However, the possibility that CFTR functions as a dimer raises more complex options. In this case, there could be functional interaction between two Rad50-like NBD1-NBD2 heterodimers, but a fundamentally different organization would also be possible in which NBD1 and NBD2 each interact with themselves to form a pair of Rad50-like homodimers. (5) Does NBD1 of CFTR hydrolyse ATP? Some isolated NBD1 constructs have been reported to have ATPase activity (Howell et al. 2000 and citations therein). However, a Glu at the end of the Walker-B (
4DE with any hydrophobe) functions as the catalytic base activating the hydrolytic water for attack on the 
-phosphate of ATP in other NBDs but is replaced by a Ser in NBD1 of CFTR, raising the possibility that NBD1 may be incapable of hydrolysing ATP in intact CFTR. (6) How does phosphorylation by protein kinase A activate CFTR? The structural mechanism of this required regulatory modification is unknown but could involve modulation of inter-NBD interactions (Howell et al. 2000).
In an article in this issue of The Journal of Physiology, Powe et al. (2002) study the gating kinetics of CFTR channels in which the conserved Lys in the Walker-A (GxxGxGKT with x any residue) has been replaced by Ala in NBD1, NBD2 or both NBDs. Their data demonstrate an asymmetrical yet cooperative functional interaction between NBD1 and NBD2 that seems roughly consistent with the simplest possible gating model whereby positively cooperative ATP binding induces formation of a Rad50-like NBD1-NBD2 heterodimer that stabilizes the open state of the channel. However, their results are also consistent with a variety of more complex structural models.
Efforts to use these data to evaluate structural models is complicated by the kinetic behaviour exhibited by most ATPase active site mutations. The Walker-A binds the 
- and 
- phosphates of the nucleotide in many different kinds of NTPases, with its Lys forming a salt-bridge to the - and sometimes -phosphates, thereby providing both binding energy and electrostatic stabilization of the transition state for hydrolysis. Therefore, mutation of the Lys generally reduces both the nucleotide-binding affinity and also the catalytic rate of the NTPase (i.e. increased KM and decreased kcat). The reduction in nucleotide-binding affinity implies a decrease in the rate of binding (i.e. rate of formation of the Rad50-like dimer) and/or an increase in the rate of release (i.e. rate of dissociation of the Rad50-like dimer). Because hydrolysis to ADP probably triggers dissociation of the Rad50-like dimer, the reduction in kcat is very likely to slow its dissociation. However, this effect could be offset by more rapid dissociation of the dimer before hydrolysis due to reduced nucleotide affinity, so that the net effect of this mutation on the dissociation rate of the Rad50-like dimer is unpredictable. Further complications arise from the fact that earlier efforts using non-hydrolysable analogues to distinguish the role of ATP binding vs. hydrolysis in CFTR are clouded by the fact that these analogues are not a faithful mimic of ATP in a significant fraction of ATPases and often behave differently even in closely related enzymes.
Therefore, the data of Powe et al. (2002) do not definitively answer of any of the questions posed above. Nonetheless their approach, involving careful characterization of simultaneous mutations in NBD1 and NBD2, establishes an important model to be followed in future work directed at elucidating the mode of inter-NBD interaction and its role in controlling gating.
 |
REFERENCES |
| CHANG, G. & ROTH, C. B. (2001). Science 293, 1793-1800 |
|
| HOPFNER, K. P., KARCHER, A., SHIN, D. S., CRAIG, L., ARTHUR, L. M., CARNEY, J. P. & TAINER, J. A. (2000). Cell 101, 789-800 |
[Medline] |
| HOWELL, L. D., BORCHARDT, R. & COHN, J. A. (2000). Biochemical and Biophysical Research Communications 271, 518-525 |
[Medline] |
| POWE, A. C. JR, AL-NAKKASH, L., LI, M. & HWANG, T.-C. (2002). Journal of Physiology 539, 333-346 |
[Abstract/Full Text] |
| RAGHURAM, V., MAK, D. D. & FOSKETT, J. K. (2001). Proceedings of the National Academy of Sciences of the USA 98, 1300-1305 |
[Abstract/Full Text] |
| THOMAS, P. J. & HUNT, J. F. (2001). Nature Structural Biology 8, 920-923 |
[Medline] |
