Functional roles of nonconserved structural segments in CFTR's NH2-terminal nucleotide binding domain

Abstract

The cystic fibrosis transmembrane conductance regulator (CFTR), encoded by the gene mutated in cystic fibrosis patients, belongs to the family of ATP-binding cassette (ABC) proteins, but, unlike other members, functions as a chloride channel. CFTR is activated by protein kinase A (PKA)-mediated phosphorylation of multiple sites in its regulatory domain, and gated by binding and hydrolysis of ATP at its two nucleotide binding domains (NBD1, NBD2). The recent crystal structure of NBD1 from mouse CFTR (Lewis, H.A., S.G. Buchanan, S.K. Burley, K. Conners, M. Dickey, M. Dorwart, R. Fowler, X. Gao, W.B. Guggino, W.A. Hendrickson, et al. 2004. EMBO J. 23:282-293) identified two regions absent from structures of all other NBDs determined so far, a "regulatory insertion" (residues 404-435) and a "regulatory extension" (residues 639-670), both positioned to impede formation of the putative NBD1-NBD2 dimer anticipated to occur during channel gating; as both segments appeared highly mobile and both contained consensus PKA sites (serine 422, and serines 660 and 670, respectively), it was suggested that their phosphorylation-linked conformational changes might underlie CFTR channel regulation. To test that suggestion, we coexpressed in Xenopus oocytes CFTR residues 1-414 with residues 433-1480, or residues 1-633 with 668-1480, to yield split CFTR channels (called 414+433 and 633+668) that lack most of the insertion, or extension, respectively. In excised patches, regulation of the resulting CFTR channels by PKA and by ATP was largely normal. Both 414+433 channels and 633+668 channels, as well as 633(S422A)+668 channels (lacking both the extension and the sole PKA consensus site in the insertion), were all shut during exposure to MgATP before addition of PKA, but activated like wild type (WT) upon phosphorylation; this indicates that inhibitory regulation of nonphosphorylated WT channels depends upon neither segment. Detailed kinetic analysis of 414+433 channels revealed intact ATP dependence of single-channel gating kinetics, but slightly shortened open bursts and faster closing from the locked-open state (elicited by ATP plus pyrophosphate or ATP plus AMPPNP). In contrast, 633+668 channel function was indistinguishable from WT at both macroscopic and microscopic levels. We conclude that neither nonconserved segment is an essential element of PKA- or nucleotide-dependent regulation.

Figure 1.

Crystal structure of NBD1 of murine CFTR. (Left) Domain topology of CFTR comprising two transmembrane domains (TMD1 and 2), two cytoplasmic nucleotide binding domains (NBDs, marked 1 and 2), and unique cytosolic R domain. (Right) Ribbon diagrams (created using Swiss-PdbViewer v3.7b2) of crystal structure of mouse CFTR NBD1 with bound MgATP (Lewis et al., 2004; accession code 1R0X), in two orientations: top, view from the side of the ATP-binding pocket (top view); bottom, view from an angle roughly perpendicular to the top and along the plane of the F1-like parallel β-sheet (back view). Color coding: ABC-specific NH₂-terminal antiparallel β-sheet, cyan; F1-like ATP-binding core, green; ABC-specific α-helical domain, gold; nonconserved insertion and extension, red; ATP, magenta; side chains of serines 660 and 670, blue. Dotted red line within the nonconserved insertion represents disordered residues 412–428, not resolved in the crystal structure, and is simply intended to indicate the continuity of the peptide chain. β-Strands and α-helices discussed in the text are marked by arrows.

Figure 2.

Obligatory dependence on phosphorylation by PKA of macroscopic currents of severed CFTR constructs. Macropatches containing tens or hundreds of (A) WT, (B) 633+668, (C) 414+433, or (D) F633(S422A)+668 (with NH₂-terminal Flag tag), channels were superfused with 2 mM MgATP and, after ∼1 min, transiently with 300 nM PKA catalytic subunit (bars); the 20-s time bar applies to all four panels A–D, which show recordings obtained at −80 mV. (E) Mean currents in MgATP alone, before (pre, black bars) or after (post, gray bars) exposure to PKA, normalized to the mean of the steady current in the presence of PKA (PKA, striped bars). Neither pre-PKA (P > 0.32) nor post-PKA (P > 0.11) currents were significantly different from those of WT for any of the three severed constructs.

Figure 3.

ATP dependence of macroscopic current is intact for severed CFTR constructs. Currents from macropatches containing (A) WT, (B) 633+668, (C) 414+433, and (D) Flag-tagged 633(S422A)+668 channels superfused with test concentrations of MgATP ranging from 5 μM to 1 mM, bracketed by exposures to 2 mM MgATP (bars; numbers indicate test [ATP] in μM). L-shaped scale bars in each panel indicate 2 pA and 10 s. (E) Mean currents in test [MgATP], normalized to the average of the currents in the 2 mM MgATP bracketing segments, plotted against test [ATP]. Lines show fits to the Michaelis-Menten equation, giving K_m values printed.

Figure 4.

Single-channel kinetics in the presence of PKA and following its removal. Patches containing (A) two WT, (B) a single 633+668, or (C) two 414+433 channels were superfused with 2 mM MgATP; after ∼1 min 300 nM PKA catalytic subunit was transiently added (bars). Note absence of openings during exposure to ATP alone, before addition of PKA. (D) Open probabilities, (E) mean burst (τ_b), and (F) interburst durations (τ_ib), in the presence of PKA+ATP (left) or in just ATP after PKA removal (right), extracted from multichannel patches (materials and methods). P_o and τ_b values in PKA are given in the text; τ_ib in PKA was 1190 ± 177 ms (n = 11) for WT, 1861 ± 850 ms (n = 5) for 633+668, and 2165 ± 1136 ms (n = 6) for 414+433. After PKA removal, P_o was 0.14 ± 0.03 (n = 7), 0.12 ± 0.03 (n = 4), and 0.10 ± 0.05 (n = 5); τ_b was 339 ± 64 ms (n = 8), 314 ± 77 ms (n = 6), and 282 ± 44 ms (n = 10); and τ_ib was 2090 ± 268 ms (n = 7), 3004 ± 736 ms (n = 4), and 3005 ± 847 ms (n = 5) for WT, 633+668, and 414+433, respectively. Asterisks in D and E indicate significantly smaller P_o (P = 0.017) and significantly shorter bursts (P = 0.03), respectively, for 414+433 compared with WT.

Figure 5.

ATP dependence of single-channel gating kinetics is intact for severed CFTR constructs. Currents from patches with few active (A) WT, (B) 633+668, and (C) 414+433 channels exposed to 5 μM MgATP, bracketed by superfusion with 2 mM MgATP (bars). Note prolonged closed (interburst) periods in 5 μM ATP, evident in the inset with expanded time scale in A. (D–F) Mean burst (τ_b) and interburst (τ_ib) durations were extracted (materials and methods) from segments of record in various [ATP], and closing and opening rates defined as 1/τ_b and 1/τ_ib. P_o (D), closing (E) and opening rates (F) in test [ATP], normalized to the average of the same parameters in the bracketing control segments, are plotted against [ATP]. Curves show fits of P_o (D) and opening rate (F) to the Michaelis-Menten equation, giving K_m values printed.

Figure 6.

Relaxation of macroscopic current upon removal of PPi+ATP or AMPPNP+ATP is faster for severed 414+433 channels. Macropatches expressing hundreds of (A) WT, (B) 633+668, or (C) 414+433 channels were superfused with either 2 mM or 0.1 mM (together with PP_i) MgATP, 300 nM PKA catalytic subunit, and 2 mM Mg-PP_i, as indicated by bars; the 40-s time bar applies to all three panels A–C. In A and B, smooth lines through data are single-exponential fits, τ = 56 s for WT (A), τ = 30 s for 633+668 (B). In C, all three instances of current decay were fitted (smooth green lines); first and third relaxations were fit with single exponentials yielding τ = 422 ms and τ = 353 ms, respectively; the relaxation following PP_i removal was fit with two exponentials yielding time constants and fractional amplitudes of τ₁ = 400 ms, τ₂ = 6.5 s, a₁ = 0.56, a₂ = 0.44. (D) Average τ for WT and 633+668 from single-exponential fits, and τ₂ for 414+433 from double-exponential fits, to decaying currents after removal of PP_i+ATP. (E) Average τ₂ for WT, 633+668, and 414+433 from double-exponential fits to decaying currents after removal of AMPPNP+ATP. Asterisks indicate significantly shorter slow time constants for 414+433 compared with WT, in D (P = 10⁻⁵) and in E (P = 0.049).

Figure 7.

Single-channel conductance is unaltered in severed-NBD1 constructs. Representative examples of single-channel current–voltage (I–V) relationships in symmetrical 140 mM Cl⁻ for WT, 633+668, and 414+433 CFTR channels. Slopes of straight lines fitted by linear regression yield single-channel conductances; mean values are given in the text.