Severed channels probe regulation of gating of cystic fibrosis transmembrane conductance regulator by its cytoplasmic domains

Abstract

Opening and closing of a CFTR Cl(-) channel is controlled by PKA-mediated phosphorylation of its cytoplasmic regulatory (R) domain and by ATP binding, and likely hydrolysis, at its two nucleotide binding domains. Functional interactions between the R domain and the two nucleotide binding domains were probed by characterizing the gating of severed CFTR channels expressed in Xenopus oocytes. Expression levels were assessed using measurements of oocyte conductance, and detailed functional characteristics of the channels were extracted from kinetic analyses of macroscopic current relaxations and of single-channel gating events in membrane patches excised from the oocytes. The kinetic behavior of wild-type (WT) CFTR channels was compared with that of split CFTR channels bearing a single cut (between residues 633 and 634) just before the R domain, of split channels with a single cut (between residues 835 and 837) just after the R domain, and of split channels from which the entire R domain (residues 634-836) between those two cut sites was omitted. The channels cut before the R domain had characteristics almost identical to those of WT channels, except for less than twofold shorter open burst durations in the presence of PKA. Channels cut just after the R domain were characterized by a low level of activity even without phosphorylation, strong stimulation by PKA, enhanced apparent affinity for ATP as assayed by open probability, and a somewhat destabilized binding site for the locking action of the nonhydrolyzable ATP analog AMPPNP. Split channels with no R domain (from coexpression of CFTR segments 1-633 and 837-1480) were highly active without phosphorylation, but otherwise displayed the characteristics of channels cut after the R domain, including higher apparent ATP affinity, and less tight binding of AMPPNP at the locking site, than for WT. Intriguingly, severed channels with no R domain were still noticeably stimulated by PKA, implying that activation of WT CFTR by PKA likely also includes some component unrelated to the R domain. As the maximal opening rates were the same for WT channels and split channels with no R domain, it seems that the phosphorylated R domain does not stimulate opening of CFTR channels; rather, the dephosphorylated R domain inhibits them.

Scheme S1

Scheme S2

Figure 1

Dependence on phosphorylation by PKA of membrane conductance in resting (basal) and activated oocytes expressing WT CFTR and severed CFTR channels [Flag3-835 plus 837-1480 (F835+837), or Flag-cut-ΔR (F633+837)], and influence of RpcAMPS injection. Bars show mean (±SEM) basal and activated (by 50 μM forskolin plus 1 mM IBMX) conductances of oocytes injected with cRNAs (2.5 or 0.25 ng per construct, as indicated) of the CFTR segments cartooned at left: NH₂-terminal Flag epitope (black zigzag); transmembrane domains (blue, cyan); NBD1 (red circle); R domain (green rectangle); NBD2 (yellow circle). Maximal conductances (μS) of oocytes expressing any half-molecule alone were close to that of uninjected oocytes (3 ± 0.3, n = 21): Flag3-633, 2 ± 0.2 (n = 15); Flag3-835, 3 ± 0.4 (n = 10); 634-1480, 5 ± 1 (n = 10); and 837-1480, 3 ± 0.4 (n = 9).

Figure 2

Macropatch CFTR currents recorded before, during, and after exposure to 2 mM MgATP with and without 300 nM PKA. (A) WT channels. (B) Severed 633+634 channels. (C) Severed 835+837 channels; note significant current activated by MgATP alone, and robust stimulation by PKA. (D) Flag-cut-ΔR (F633+837) channels; note strong activation by MgATP alone, and small stimulation by PKA. (E) Summary of currents activated by MgATP alone (black bars), normalized to the currents subsequently measured in the same patches in the presence of PKA (striped bars), for constructs 835+837 (0.19 ± 0.03, n = 12), cut-ΔR (633+837, 0.78 ± 0.01, n = 6), and Flag-cut-ΔR (F633+837, 0.71 ± 0.03, n = 24); normalized basal current of Flag-cut-ΔR channels was unaltered (0.78 ± 0.04, n = 10, gray bar) in patches from oocytes preinjected with RpcAMPS to inhibit endogenous PKA.

Figure 3

Kinetic parameters underlying phosphorylation-dependent changes in channel currents. (A) Representative baseline-subtracted records from excised patches containing small numbers of WT, 633+634, 835+837, cut-ΔR (633+837), or Flag-cut-ΔR (F633+837) channels. (B) Open probability, (C) mean burst duration, and (D) mean interburst duration, from fits to steady state dwell-time histograms for each construct (identified below each column): striped bars give estimates in the presence of 300 nM PKA and 2 mM MgATP; black bars give estimates in MgATP alone, before exposure to PKA, for constructs 835+837, cut-ΔR (633+837), and Flag-cut-ΔR (F633+837); gray bars give estimates in MgATP alone, just after removal of PKA, for WT and 633+634 channels.

Figure 5

Dependence on [ATP] of opening and closing rates of WT and Flag-cut-ΔR in small patches containing few channels. (A) Representative baseline-subtracted recordings: quasi-stable activity of WT channels (top) after initial rapid drop in P _o on removal of PKA; Flag-cut-ΔR (F633+837) channels (bottom) before exposure to PKA. (B and C) Summary of opening and closing rates from the segments at test [ATP] (50 μM, 100 μM, and 1 mM) normalized to the mean of the estimates obtained from the bracketing segments at 2 mM ATP, from experiments like those in A, for WT (▴) and Flag-cut-ΔR channels (F633+837; ▿). For WT, relative rates at 2 μM ATP were estimated from three tests in one macropatch, on the basis of reasonable assumptions (e.g., P _o = 0.36 in PKA and 2 mM ATP) to obtain channel number, and using the mean closing rate at 2 mM ATP from all small patches. Solid and dotted lines in B are Michaelis fits to the data of WT and Flag-cut-ΔR, giving K _m values of 46 ± 13 and 39 ± 7 μM, respectively.

Figure 4

Apparent affinity for ATP as reflected by P_o, after PKA removal for WT, 633+634, and 835+837 channels, and before applying PKA for cut-ΔR and Flag-cut-ΔR channels. (A) Macroscopic currents from representative patches containing WT channels in response to step application and removal of 2 mM [ATP] with intervening 10–30-s test exposures to 2, 10, 20, 50, 100, and 500 μM, and 1 mM [ATP], respectively; scale bars indicate 5 pA and 10 s.(B) Records from representative patches containing cut-ΔR (633+837) channels; protocol and labeling as in A. (C) Summary of activation of macroscopic current by [ATP] for WT (•), cut-ΔR (▿), and Flag-cut-ΔR (▴) channels, from experiments like those in A and B. Average amplitude of steady current near the end of each 10–30-s test exposure was normalized to the mean of the average sizes of the steady currents at 2 mM ATP in that patch just before and after the test. The solid line is a Michaelis fit to the data for WT channels; K _m = 51 ± 2 μM. The dotted line is the fit for Flag-cut-ΔR channels; K _m = 23 ± 1 μM. The fit for cut-ΔR channels overlies the dotted line and was omitted for clarity; K _m = 25 ± 1 μM. Fitting the same data to the Hill equation yielded Hill coefficients of 0.98 ± 0.06 for WT, 0.84 ± 0.05 for cut-ΔR, and 0.99 ± 0.08 for Flag-cut-ΔR channels. (D) Summary of relative current at 50 μM ATP (I_50μM/I_2mM) for WT CFTR and all the severed constructs; the ratio I_50μM/I_2mM provides a rough measure of apparent affinity, because 2 mM ATP was a saturating concentration for each construct (I_1mM/I_2mM ratios were: 1.34 ± 0.10 for 633+634; 1.03 ± 0.02 for 835+837; ∼1.0 for the other three constructs, C). *I_50μM/I_2mM value significantly higher (P < 0.01) than for WT channels.

Figure 6

AMPPNP added with ATP causes slow current relaxation after removal of nucleotides in WT and severed channels. Channels at ∼24°C (23°–25°C) were activated by 2 mM MgATP with or without 300 nM PKA, and then locked open with 1 mM AMPPNP plus 0.1 mM ATP in PKA; brief stimulation of Ca²⁺-activated Cl⁻ channels with 2 mM Ca sulfamate indicated solution exchange time. (A) WT CFTR; blue line is a single exponential, with τ = 47 s. (B) 633+634 channels; blue line is a single exponential, with τ = 38 s. (C) 835+837 channels; fit to current decay after AMPPNP is double exponential, with time constants and amplitudes, τ₁ = 337 ms, τ₂ = 11 s, and a ₁ = 63 pA, a ₂ = 501 pA. (D) cut-ΔR (633+837) channels; fit to current decay after AMPPNP is double exponential, with τ₁ = 439 ms, τ₂ = 4.7 s, and a ₁ = 57 pA, a ₂ = 59 pA. (E) Flag-cut-ΔR (F633+837) channels; fit to current decay after AMPPNP is double exponential, with τ₁ = 364 ms, τ₂ = 5.2 s, and a ₁ = 41 pA, a ₂ = 40 pA.

Figure 9

Enhanced stimulation of cut-ΔR (633+837) channel current by AMPPNP at 20°C without exposure to PKA; the current decay after nucleotide removal was well fit by a single exponential (fit line), τ = 18 s.

Figure 7

(A and C) Baseline-subtracted traces showing delayed closure of WT (A) and cut-ΔR (633+837) (C) channels after washout (at ∼24°C) of 0.1 mM MgATP and 1 mM AMPPNP 5 s before the start of each trace. Red lines show unlocking events reconstructed using algorithm in Methods. (B and D) Idealized sequences of unlocking events for WT (B) and cut-ΔR (D) channels, constructed by summing results of two WT and five cut-ΔR experiments like those at left. Note different time scales in A and B vs. C and D.

Figure 8

Distributions of burst durations of WT and severed channels. Dwell times of individual bursts, after omitting flickery closures, were ranked by duration in descending order, and rank numbers, divided by the total number of events, were plotted against duration to yield survivor functions. Parameters for fits (solid lines) by single or double exponentials were obtained directly from the events lists by maximum likelihood; the need for a second component was evaluated from its improvement of the likelihood. WT and 633+634 channels were analyzed separately in PKA (+) and after PKA removal (−); for 835+837, cut-ΔR (633+837), and Flag-cut-ΔR (F633+837) channels, analysis used pooled data that included segments with and without PKA. Time constants (τ) and fractional amplitudes (a) are printed in each panel and, for 835+837, cut-ΔR, and Flag-cut-ΔR, τ_sh, τ_l, a _sh, and a _l, are also listed (with their error estimates; half-widths of 0.5 unit likelihood intervals) in Fig. 11 as observed parameters. For WT and 633+634, τ in PKA is given (with error estimates) in Fig. 11 as τ_l; after PKA removal, τ = 263 ± 9 ms for WT and τ = 214 ± 27 ms for 633+634.

Figure 10

Walker-A mutant cut-ΔR(K1250A) [633+837 (K1250A)] channels show prolonged open bursts. (A) Representative baseline-subtracted record of single cut-ΔR(K1250A) channel in 2 mM MgATP, no PKA, at 25°C. (B) Current relaxation of cut-ΔR(K1250A) channels after removal of 2 mM MgATP (no PKA), constructed by summing synchronized decay currents from nine experiments, at 25°C; single-exponential fit (solid line) to quasi-macroscopic current decay gave τ = 6.7 s. (Inset) Similar trace constructed from seven experiments at 20°C with fit (solid line) τ = 10.1 s. (C) Survivor function of burst durations, after exclusion of flickery closures, of cut-ΔR(K1250A) channels in 2 mM MgATP, constructed from events isolated from a total of 16 min of recordings suitable for such analysis, including 10 min from a single channel. The distribution was fit (solid line) significantly better by two exponential components than by one.

Figure 11

Model fit, using Fig. 1, to results from WT and severed CFTR channels in the presence of PKA. The set of rate constants (s⁻¹) giving the closest overall fit to the indicated set of observed parameters is printed on Fig. 1 for each construct (rounded to two significant digits); measured and predicted parameters are compared on the right. Parameters a _sh, a _l, τ_sh, and τ_l are fractional amplitudes and time constants of exponential components describing the distributions (Fig. 8) of burst durations (note a _l = 1 − a _sh is not a free parameter); τ_b is mean burst duration, measured in the presence of PKA for WT and 633+634, or pooled from all experiments for 835+837, cut-ΔR (633+837), and Flag-cut-ΔR (F633+837); τ_AMPPNP and a _locked are the time constant and fractional amplitude of the slowly relaxing macroscopic current component after removal of AMPPNP and ATP (see Table ); τ_relax and a _relax, analogous, after just ATP, for cut-ΔR(K1250A) [633+837(K1250A); Fig. 10 B]. For step C₁ ↔ C₂, K _Pois printed instead of K _d. Errors for observed parameters a _sh, a _l, τ_sh, and τ_l are half-widths of 0.5 unit likelihood intervals, calculated separately for each parameter (reasonably symmetric around the optimum, given the numbers of fitted events); for τ_b, τ_AMPPNP, and a _locked errors are ±SEM. The value of k ₁ was obtained as the reciprocal of τ_ib in the presence of 2 mM MgATP and PKA (except for cut-ΔR(K1250A) channels, for which all data were obtained in the absence of PKA; apparent affinity was not measured for this construct). *Rate k ₂ for WT and 633+634 may be much lower than 200 s⁻¹. From simulations, k ₂ > 20 s⁻¹ is large enough to account for the apparently uniform distribution of burst durations of these two constructs in the presence of PKA. **Rate k ₋₁ for WT and 633+634 channels was fixed to the reciprocal of the mean burst duration observed after removal of PKA, on the assumption that k ₂ << k ₋₁ under those conditions. ***Rate k ₃, representing a compound step including ATP hydrolysis at NBD2, was set to zero for cut-ΔR(K1250A).