|
Figure 1. . (Top) Cartoon of CFTR's domain topology, comprising two transmembrane domains (TMDs), a regulatory (R) domain, and NBD1 (green) and NBD2 (blue). (Below) Ribbon representations of homology models of NBD1 (green) and NBD2 (blue), based on the crystal structure of HisP-ATP (Hung et al., 1998), showing exposed positions of ATP molecules (yellow), Walker A lysines (red), and NBD2 Walker B aspartate (green). Homology models were built using the automated comparative modeling server “Swiss-model” (http://www.expasy.ch/swissmod/SWISS-MODEL.html; Guex and Peitsch, 1997) with alignments optimized on the basis of a ClustalW multiple sequence alignment of 20 bacterial and human NBDs.
|
|
Figure 2. . [MgATP] dependence of rates of opening to, and closing from, a burst of WT and mutant CFTR channels in inside-out patches. (A) WT channels, prephosphorylated by 300 nM PKA and 5 mM MgATP, shown during exposure to 5 mM, then 50 μM, and again 5 mM MgATP, as indicated; expanded segments at 5 mM and 50 μM MgATP are shown below. (B and C) Representative traces for prephosphorylated K464A and D1370N channels. Relative opening (D) and closing (E) rates (mean ± SEM, 2 ≤ n ≤ 7) from analysis of records as in A–C for WT (blue circles), K464A (red triangles), and D1370N (green squares) channels at 10 μM ≤ [MgATP] ≤ 5 mM, plotted on semilogarithmic axes. Opening rate and closing rate at each test [MgATP] were normalized to the mean values measured for bracketing segments at 5 mM MgATP (procedure (b), materials and methods). Curves in D show Michaelis-Menten fits, yielding K0.5 of 56 ± 5, 807 ± 185, 391 ± 118 μM, and rCOmax of 1.02, 1.16, and 1.08, for WT, K464A, and D1370N, respectively. For display, the data were further normalized to these rCOmax values. Mean absolute opening rates at 5 mM MgATP are given in Table I B. In E, the measured relative closing rates at each [MgATP] have been multiplied by the mean absolute closing rate for each construct in 5 mM MgATP (Table I B).
|
|
Figure 3. . The K1250A mutation strongly shifts the [MgATP] dependence of Po to higher [MgATP]. (A) steady state level of macroscopic current of prephosphorylated WT CFTR channels was ∼2-fold lower at 50 μM MgATP than during bracketing exposures to 5 mM MgATP (as expected from Fig. 2); lines below traces mark MgATP applications. Rapid current decay on MgATP washout gave (exponential fit lines superimposed on traces) τ = 0.45 s, τ = 0.40 s, τ = 0.38 s, from left to right (mean τ = 0.54 ± 0.04 s, n = 21, pooled from all [MgATP]). (B) Macroscopic current of K1250A channels was reduced ≥2-fold on lowering [MgATP] from 5 to 1 mM. Superimposed exponential fit lines show slower current decay (note 10-fold contracted time scale relative to A) with, from left to right, τ = 28 s, τ = 30 s, τ = 32 s (mean τ = 39 ± 5, n = 9, from all [MgATP]). (C) Semilog plot of Po versus [MgATP]. Steady currents (averaged over final ≥20 s) at each [MgATP], normalized to the mean bracketing level at 5 mM MgATP, yielded least-squares. Michaelis fit parameters for WT: Po max = 1.04 ± 0.01, K0.5 = 57 ± 2 μM; for K1250A: Po max = 2.45 ± 0.88, K0.5 = 6.5 ± 4.8 mM; for display, WT (circles) and K1250A (inverted triangles) data (mean ± SD, 3 ≤ n ≤9) were renormalized to these Po max values. Because 10 mM, the highest [MgATP] used, was still far from saturating for K1250A channels, the fit for this mutant is less accurate, evident from large errors on fit parameters.
|
|
Figure 4. . Burst duration distributions are similar for WT (A–C) and K464A (D–F) channels under comparable conditions, as indicated (PKA present for left column only). Fitted curves (through data points) show single exponentials from maximum-likelihood fits. Improvement of the fit by inclusion of a second exponential component was judged using the algorithm described in Csanády et al. (2000). Only for K464A at μM MgATP (F) could the likelihood be significantly increased by including a second component, though with a shorter (but not longer; Ikuma and Welsh, 2000) mean: τ1 = 30 ms, a1 = 0.17; τ2 = 263 ms, a2 = 0.83; increase in log likelihood, ΔLL = 8.3; number of bursts fitted, M = 263; giving (ΔLL − ln(2M) = 2.0). The small differences between means at mM and μM MgATP (B vs. C, E vs. F) may be only apparent, as the mean τb, estimated by multichannel kinetic fits, from these same stretches of record at μM MgATP is not significantly different from that during intervening stretches in 5 mM MgATP (for WT: τbμM/τb5mM = 1.03 ± 0.07, n = 9; for K464A: τbμM/τb5mM = 0.95 ± 0.13, n = 7). (G and H) Representative traces showing gating of K464A and D1370N channels at 15 μM MgATP (after PKA removal). Prolonged bursts of K464A channels (Ikuma and Welsh, 2000) are not evident. Though variability among the four patches containing sufficiently few D1370N channels precluded pooling the data for burst distribution analysis, in none of those patches (analyzed separately) did introduction of a second component significantly improve the maximum likelihood fit.
|
|
Figure 5. . Mutations at the NBD1 catalytic site do not markedly alter open burst duration of channels exposed to 5 mM MgATP and 300 nM PKA. Patches contained one WT (A), K464A (B), or S573E (D) channel, or more than one D572N (C) channel. (E) Summary of mean (±SEM) τb values at 5 mM MgATP and 300 nM PKA (n = 30, 21, 9, and 7 for WT, K464A, D572N, and S573E, respectively).
|
|
Figure 6. . Mutations at the NBD2 catalytic site slow channel exit from open bursts. WT (A), D1370N (B), K1250A (C), and E1371S (D) CFTR channels were activated by 5 mM MgATP plus PKA as indicated: burst termination (∼0.4-pA downward steps) after nucleotide washout was slowed for NBD2 mutants, relative to WT. Note persistence of brief (intraburst) closures during K1250A and E1371S bursts, long after nucleotide withdrawal.
|
|
Figure 7. . Opening of WT CFTR channels by poorly hydrolyzable ATP analogs. (A) Representative recordings from four patches containing many channels, previously phosphorylated by PKA; each patch was exposed to 10 μM MgATP, to (mM) analogue, and to 5 mM MgATP (shown for only two patches). Maximum likelihood fits using C-O-CF model gave, for MgATPγS τb = 2.0 ± 0.6 s (n = 5), for MgAMPPNP τb = 1.6 ± 0.2 s (n = 32), and for MgAMPPCP τb = 0.36 ± 0.05 (n = 11). (B) Summary of rCO values for nonhydrolyzable analogs normalized to rCO at 10 μM MgATP (which is ∼11% of maximum, Fig. 2 D) in the same patch.
|
|
Figure 8. . MgAMPPNP inhibits MgATP-induced opening of WT channels. (A) Representative record of prephosphorylated channels, exposed alternately to 50 μM MgATP or to 50 μM MgATP plus 200 μM MgAMPPNP, and then to 200 μM MgAMPPNP alone. (B) Summary of rCO values (mean ± SEM, n = 3) normalized to the average of that obtained from the stretches in 50 μM MgATP.
|
|
Figure 9. . Exit from MgAMPPNP-locked burst states is slower when bursts are initiated in the presence of MgATP. Patches with hundreds of prephosphorylated WT CFTR channels were repeatedly subjected to ∼30-s long exposures to nucleotides (as in inset), in varied sequence. Each trace in the main figure is the sum of 21 recordings, synchronized upon nucleotide washout (arrow; also in inset), from 12 patches, each exposed to 0.5 mM MgATP, 5 mM MgAMPPNP, or 0.5 mM MgATP + 5 mM MgAMPPNP alternately, an equal number of times. Exponential decay fit parameters are: after MgATP, a = 33 pA, τ = 0.8 s; after AMPPNP, single a = 8 pA, τ = 6.8 s; double af = 6 pA, as = 6 pA, τf = 0.7s, τs = 8.8 s; after MgATP + MgAMPPNP, af = 20 pA, as = 18 pA τf = 2 s, τs = 36.6 s. As solution exchange time was 0.5–1s, fast components do not accurately reflect channel closing.
|
|
Figure 10. . The K464A mutation speeds exit from locked open burst states. (A) Macroscopic WT channel current activated by a mixture of 0.5 mM MgATP and 5 mM MgAMPPNP (+PKA) decays slowly upon removal of nucleotides. (B) Current decay is much faster for the K464A mutant in the same conditions. Blue fit lines in A and B show only the slow components of double exponential fits, with τs = 67.8s, as = 0.92 for WT, and τs = 8.7s, as = 0.79 for K464A. (C and D) Summaries of fractional amplitude, as (C), and time constant, τs (D), of the slow component from 18 WT and 16 K464A experiments. In controls with no MgAMPPNP, closure after exposure to MgATP and PKA yielded τ = 1.9 ± 0.2 s (n = 35) for WT and τ = 1.0 ± 0.1 s (n = 34) for K464A, and both constructs sometimes showed a small amplitude slower component: for WT, τs = 7.6 ± 1.7 s, as = 0.1 ± 0.03 (in 13/35 patches); for K464A, τs = 5.9 ± 0.8 s, as = 0.24 ± 0.04 (20/24 patches). (E) Macroscopic K1250A currents, activated by 5 mM MgATP + PKA, decay slowly on nucleotide withdrawal. (F) The additional K464A mutation accelerates channel closure from bursts: for the traces shown, τ = 71.7s (K1250A) and τ = 29.7s (K464A/K1250A). (G) Mean time constants of all 9 K1250A and 9 K464A/K1250A relaxations, each well fit by a single exponential.
|
|
Figure 11. . Gating of prephosphorylated K464A channels by poorly hydrolyzable ATP analogs, as indicated. Unlike WT (Fig. 7 A), K464A burst duration was not increased during exposure to MgAMPPNP (A and B, τb = 270 ± 50 ms, n = 8), and was only slightly increased during exposure to ATPγS (C, τb = 655 ± 170 ms, n = 8), compared with bursts in MgATP (τb = 276 ± 21 ms, n = 16) in the same patches. Note that, due to the lower apparent affinity of K464A for MgATP (Fig. 2), the relative opening rate of mutant channels at 10 μM MgATP averaged only 2.3 ± 0.8% (n = 3) of that in saturating MgATP (compared with ∼11% for WT), so the opening rate of K464A channels was similar in the presence of millimolar concentrations of the poorly hydrolyzable analogs or of 10 μM MgATP.
|
|
Figure 12. . (A) Simplified scheme illustrating proposed linking of steps in WT CFTR channel gating, and nucleotide binding and hydrolytic cycles. Yellow ovals depict MgATP complexes, smaller red oval is inorganic phosphate, Pi (colored orange in prehydrolysis complex on open channel); the CFTR protein is represented as a green (NBD1) and blue (NBD2) semicircle, with shape altered (signifying induced-fit conformational changes in the NBDs; Karpowich et al., 2001) upon nucleotide binding. “C” represents closed interburst states of the channel and “O” symbolizes the collection of states during open bursts. Thickness and length of arrows indicate relative rates of individual steps. There is no evidence for strict sequential binding of the two MgATP complexes, but the alternative pathway to the doubly occupied closed state, in which nucleotide binds first at NBD2, probably occurs infrequently in WT CFTR (though not necessarily in mutants) and so was omitted for clarity. (B) Cartoon illustrating a possible physical interpretation of the scheme in A, in which NBD dimerization couples ATP binding and hydrolysis at the catalytic sites to opening and closing of the channel pore. Two semicircles represent NBD1 (green) and NBD2 (blue), and the transmembrane domains are represented by straight-line segments connected to the NBDs. Closed and Open channels are indicated by converging or near-parallel transmembrane domains, respectively. Catalytic sites and Cl− permeation pathway are structurally connected such that NBD dimer formation results in opening of the channel pore.
|
|
Figure .
|
