The CFTR ion channel belongs to the ATP binding cassette (ABC) protein superfamily. All ABC proteins share a similar domain organization with a “core” functional unit comprising two cytosolic nucleotide binding domains (NBDs) and two transmembrane domains (TMDs) (Fig. 1A). NBDs bind and hydrolyze ATP. Structurally they comprise a RecA-like subdomain (“head”), and a helical subdomain (“tail”). In the presence of ATP, NBDs can associate to form a tight “head-to-tail” dimer with two composite ATP binding sites at the interface (Fig. 1B). The ATP molecules provide crucial molecular contacts that stabilize the dimer. It is generally believed that NBD dimerization – driven by ATP binding – and dimer disengagement – following ATP hydrolysis – couple ATPase cycles to functionally important conformational changes in the TMDs. Most ABC proteins are transporters while CFTR, alone, is an ion channel. However, it was proposed that CFTR gating is “driven” by ATP binding and hydrolysis employing the molecular mechanisms used by transporters: ATP binds to NBD1 and NBD2 which associate to form an intramolecular “dimer” with composite sites 1 and 2 buried at the interface; ATP at site 2 is then hydrolyzed, triggering dimer dissociation; NBD dimer formation and dissociation are sensed by the TMDs where they result in opening and closing, respectively, of the pore (Fig. 1C). In CFTR, consensus motifs involved in ATP binding and hydrolysis are conserved only in site 2. In site 1, non-conservative substitutions at key residues result in tight ATP binding, but no efficient hydrolysis. Non-canonical substitutions in the same key positions, presumably resulting in reduced or absent hydrolysis at one composite site, are also seen in other important human (e.g. TAP1/TAP2 and SUR) as well as in numerous yeast and prokaryotic transporters. The latter group includes TM287/288, a heterodimeric transporter whose high resolution structure has been very recently solved [1]. At site 1 in this crystal, an ATP molecule is bound and NBDs maintain contact but dimer interface is largely solvent accessible. Early biochemical studies revealed that ATP remains bound, unhydrolyzed, at CFTR’s site 1 for periods of time much longer than a channel gating cycle (~1 s). This finding was confirmed in a recent functional study [2]. The high-affinity ATP analogue N6-(2-phenylethyl)-ATP (P-ATP) was found to speed up opening and to slow down closing of CFTR channels. The onset of these two effects, following sudden replacement of ATP with P-ATP, followed distinct time courses. Whereas the effect on opening rate appeared immediately, likely reflecting P-ATP binding to the rapid-turnover site 2, the onset of slowed closure was delayed by 30-50 s, consistent with an action of P-ATP at slow-turnover degenerate site 1. Furthermore, because nucleotide exchange rate at site 1 was affected by mutations in the tail of NBD2, the authors suggested that composite site 1 remained associated in closed channels. Our thermodynamic studies [3], examining state-dependent changes in energetic coupling between pairs of residues on opposing faces of site 1, supported this idea. A lack of change in energetic coupling for three such pairs was interpreted to be consistent with the interface around composite site 1 remaining “closed” throughout the gating cycle. A more detailed analysis of the changes in gating kinetics caused by two site 1 perturbations has, however, lead us to question this initial interpretation. We focused on the effects of P-ATP dependent on site-1 occupancy and of the H1348A mutation in the NBD2 tail. Both site-1 perturbations had similar effects on gating: a 2-3 fold decrease in the non-hydrolytic closing rate (k-1, Fig. 1C) and in the rate of the O1→O2 transition (k1), consistent with the changes in site 1 causing a selective stabilization of state O1 with respect to both posthydrolytic O2 and closed C1. The pore-closing O2→C2 step was not affected. Could the lack of histidine/presence of phenylethyl moiety in site 1 be altering the energetics of other parts of the protein (e.g. TMDs and site 2) without site 1 actually moving during opening and hydrolysis? This interpretation would require that the site 1 perturbations (which cause an increase in the energetic barriers the channel must overcome to exit the O1 state) have different effects on the O1 ground state and on the transition states of the opening and hydrolysis steps. It is difficult to envisage how such differential allosteric effects could occur if the region around the perturbation remained static. A simpler interpretation of the data is that the physicochemical environment around the 1348 histidine side chain and around the N6 group on the adenine is not identical in O1 and in those transition states. In other words, this region of site 1 moves during opening and coincident with hydrolysis at the active site 2.