In vivo phosphorylation of CFTR promotes formation of a nucleotide-binding domain heterodimer

Abstract

The human ATP-binding cassette (ABC) protein CFTR (cystic fibrosis transmembrane conductance regulator) is a chloride channel, whose dysfunction causes cystic fibrosis. To gain structural insight into the dynamic interaction between CFTR's nucleotide-binding domains (NBDs) proposed to underlie channel gating, we introduced target cysteines into the NBDs, expressed the channels in Xenopus oocytes, and used in vivo sulfhydryl-specific crosslinking to directly examine the cysteines' proximity. We tested five cysteine pairs, each comprising one introduced cysteine in the NH(2)-terminal NBD1 and another in the COOH-terminal NBD2. Identification of crosslinked product was facilitated by co-expression of NH(2)-terminal and COOH-terminal CFTR half channels each containing one NBD. The COOH-terminal half channel lacked all native cysteines. None of CFTR's 18 native cysteines was found essential for wild type-like, phosphorylation- and ATP-dependent, channel gating. The observed crosslinks demonstrate that NBD1 and NBD2 interact in a head-to-tail configuration analogous to that in homodimeric crystal structures of nucleotide-bound prokaryotic NBDs. CFTR phosphorylation by PKA strongly promoted both crosslinking and opening of the split channels, firmly linking head-to-tail NBD1-NBD2 association to channel opening.

Figure 1

Native cysteines of human epithelial CFTR. Topological cartoon of CFTR showing the 18 cysteines indicated by their position numbers, the transmembrane domains (gray), the R domain (green), and the relative locations of Walker A (red) and ABC signature (purple) motifs in the nucleotide-binding domains, NBD1 (yellow) and NBD2 (cyan).

Figure 2

Expression and function of cysteine-deficient CFTR channels in Xenopus oocytes. (A) Two-microelectrode voltage-clamp current recordings from uninjected oocyte and oocytes expressing WT CFTR (2.5 ng cRNA) or HA-tagged Cys-free CFTR 16CS+C590L/C592L (20 ng cRNA); vertical current deflections monitor conductance, which was transiently increased by brief exposure to 40 μM forskolin (between arrows). (B) Summary of mean±s.d. whole-oocyte conductances determined as in (A), before (‘resting', white bars) and at maximal forskolin effect (‘stimulated', gray bars), 3 days after injection of 20 ng cRNA encoding HA-tagged CFTR 16CS constructs containing native C590 and C592 or substitutions at those positions as indicated; forskolin elicited significant conductance only with C590 and C592 unchanged (131±6 μS, n=3), or replaced by valines (119±6 μS, n=15) or leucines (155±9 μS, n=3). (C) Conductances from oocytes injected with 2.5 ng cRNA, and measured 1 day later for WT (153±17 μS, n=3), or 3 days later for HA-tagged 16CS mutants with C590/C592 (42±10 μS, n=6) or C590V/C592 V (51±3 μS, n=6). This under-represents the functional difference between WT and mutants, as conductance is enhanced by injecting more cRNA or allowing more time for its expression. (D) WT CFTR and Cys-free CFTR (16CS+C590V/C592V) were immunoprecipitated from membranes of oocytes injected with cRNA amounts indicated, and subjected to SDS–PAGE and Western blot analysis; arrows mark core-glycosylated and mature fully glycosylated CFTR.

Figure 3

Homology model of a head-to-tail CFTR NBD1–NBD2 heterodimer, based on crystal structures of human CFTR NBD1 F508A and of ATP- or AMPPNP-bound NBDs of other ABC proteins (Materials and methods). Two ATP molecules (CPK-colored stick structures) are sandwiched between the Walker A and ABC signature sequences of opposing NBDs, NBD1 (yellow) and NBD2 (blue). To test this model by sulfhydryl-specific crosslinking, the residues in spacefill were mutated to cysteines in pairs, chosen so that an interfacial crosslink would span the ‘NBD1' composite site (cyan, orange, and purple residue pairs), or a central region (green residues), or the ‘NBD2' composite site (red residues). Due to sequence and structural differences between CFTR's NBD1 and other NBDs, the position of S434 is uncertain.

Figure 4

Expression (A, B) and function (A, C) of split CFTR channels containing introduced cysteines. Oocytes were injected with 5+5 ng cRNA encoding NH₂-terminal (1–633), and COOH-terminal (634–1480) 9CS (Ser replacing Cys at positions 647, 832, 866, 1344, 1355, 1395, 1400, 1410 and 1458), half channels containing either no introduced cysteine in NBD1 or NBD2 (‘no C' label in A, B), or a single introduced cysteine at the residue position indicated beside each point in (A), above each lane in (B), and above each recording in (C). (A) Summary of resting and stimulated whole-oocyte conductances (μS) normalized to relative expression level, plotted against levels of expression (normalized to the highest measured value). (B) Expression levels were measured from a Western blot (anti-R-domain antibody) of ∼25 μg of membrane proteins from 30 to 32 resting oocytes as gray level intensities of the broad bands of fully glycosylated COOH-terminal half channels (arrow); the lower sharp bands near 85 kDa are core-glycosylated COOH-terminal half channels. (C) Three examples of conductance measurements, at rest and after stimulation with 40 μM forskolin, in oocytes containing an introduced cysteine in NBD1 (top), or in NBD2 (middle), or in both NBDs (bottom).

Figure 5

The absence of efficient crosslinking when no, or only one, engineered cysteine is present. Oocytes injected with 5+5 ng cRNA encoding NH₂-terminal (1–633), and COOH-terminal (634–1480) 9CS, half channels containing no (background), or just one, introduced cysteine in either NBD1 or NBD2, at the position indicated above each panel. Oocytes were untreated, or pretreated with 40 μM forskolin plus 1 mM IBMX and then incubated with 300 μM BMOE or 600 μM BMH at room temperature, as indicated. Membrane proteins were analyzed by SDS–PAGE and blotted with antibody against the R domain (COOH-terminal half-channel, upper blots) or antibody against the NH₂-terminus (lower blots). Note that upon forskolin treatment, phosphorylation of the R domain slows mobility of the COOH-terminal half channel (Csanády et al, 2005) resulting in the appearance of a sharp band (at ∼90 kDa) just above the ∼85 kDa core-glycosylated band. No strong high-molecular-mass band, reflecting crosslinked product, is evident in any lane, but arrows mark weak, dispersed, unidentified BMOE- and/or BMH-induced bands discernible in some blots with antibody against the NH₂-terminus.

Figure 6

Crosslinking across the ‘NBD2' composite catalytic site, between position 1248 in NBD2 Walker A and position 549 in NBD1 LSGGQ. Western blots identify the NH₂-terminal half channel (1–633), S549C (left panel; lower arrow), the COOH-terminal half channel (634–1480) 9CS+S1248C (right panel; core-glycosylated, ∼85–90-kDa, bands; fully glycosylated, lower arrow), and crosslinked product (both panels; arrows labeled X-link). Incubation temperature, presence or absence of 300 μM BMOE or 600 μM BMH, and/or treatment with 40 μM forskolin plus 1 mM IBMX (‘fsk') are indicated below each lane. Forskolin increased the yield of crosslinked product four-fold for BMOE and two-fold for BMH. Samples in lanes 8 and 16 are from uninjected control oocytes.

Figure 7

Crosslinking between central region residues, 605 of NBD1 and 1374 of NBD2. Western blots show CFTR half channels (1–633) S605C (left panel; lower arrow), (634–1480) 9CS+A1374C (right panel; core-glycosylated, ∼85–90 kDa, bands; fully glycosylated, lower arrow), and crosslinked product (both panels; arrows labeled X-link). Incubation temperature, presence or absence of 300 μM BMOE or 600 μM BMH, and/or treatment with 40 μM forskolin plus 1 mM IBMX (‘fsk') are indicated below each lane. Forskolin increased the yield of crosslink product 870-fold for BMOE and four-fold for BMH.

Figure 8

Crosslinking across the ‘NBD1' composite site. Incubation temperature, presence or absence of 300 μM BMOE or 600 μM BMH, and/or treatment with 40 μM forskolin plus 1 mM IBMX (‘fsk') are indicated below each lane in all panels. Left panels show NH₂-terminal half channels (lower arrows), right panels show COOH-terminal half channels (core-glycosylated, ∼85–90-kDa bands; fully glycosylated, lower arrows), and all panels show crosslinked product (upper arrows labeled X-link); some other BMOE- and/or BMH-induced bands, mostly in the blots with anti-NH₂-terminal antibody, likely reflect crosslinking of NH₂-terminal half channels (with nine native Cys) to unknown proteins. (A) CFTR half channels (1–633) A462C (left panel) and (634–1480) 9CS+S1347C (right panel). Forskolin increased the yield of crosslinked product >1000-fold for both BMOE and BMH; lanes 8 and 16 are samples from uninjected oocytes. (B) CFTR half channels (1–633) S459C (left panel) and (634–1480) 9CS+V1379C (right panel). Forskolin increased crosslink product yield nine-fold for BMOE and 5-fold for BMH. (C) CFTR half channels (1–633) S434C (left panel) and (634–1480) 9CS+D1336C (right panel), as well as crosslinked product (arrow labeled X-link, both panels). Forskolin increased crosslink product yield 9-fold for BMH. Crosslink yield for BMOE was negligibly small.

Figure 9

Tests of crosslinking between NBD1 and NBD2 using other combinations of the target cysteines. The presence or absence of 300 μM BMOE or 600 μM BMH and/or treatment with 40 μM forskolin plus 1 mM IBMX (‘fsk') are indicated for each lane in all panels. (A) Oocytes were injected with cRNAs encoding NH₂-terminal (1–633) and COOH-terminal (634–1480) 9CS half channels, each containing one introduced cysteine, in the combinations indicated above each panel. Incubations were at room temperature. Western blots identify NH₂-terminal half channels (lower rows) and core- and fully glycosylated COOH-terminal half channels (upper row). (B) Western blots show CFTR half channels (1–633) S549C (left panel, lower arrow) and (634–1480) 9CS+A1374C (right panel, core-glycosylated, ∼85–90-kDa bands; fully glycosylated, lower arrow), as well as crosslinked product (both panels, arrows labeled X-link). Incubation temperature is indicated below each lane. Forskolin increased the yield of crosslink product eight-fold for BMOE and two-fold for BMH.

Figure 10

Functional consequence of crosslinking S549C to S1248C. (A–C) Currents activated by 5 mM ATP and 300 nM PKA catalytic subunit in thousands of split CFTR channels in inside-out patches excised from oocytes expressing NH₂-terminal (1–633), and COOH-terminal (634–1480) 9CS, half channels containing only one target cysteine, S549C (A) or S1248C (B), or both S549C and S1248C (C). Current decay due to channel closure on washout of ATP and PKA was unaltered by Cu(II)(o-phenanthroline)₂ when channels contained only one target cysteine (A, B), but resulted in a stable nucleotide-independent, but DTT-sensitive, residual current I₀ (C, black arrow). (D) Average (±s.e.m.) amplitude of I₀ normalized to the maximal (ATP-plus PKA-activated) current, I_max, for split channels as in (A–C); number of measurements is indicated for each column. For S549C/S1248C channels I₀/I_max is significantly different (^***P⩽0.0005, Student's t-test) after closure in the presence of Cu(II)(o-phenanthroline)₂ compared to its absence (bath solution).

Figure 11

Summary of crosslinking results. Residues replaced by target cysteines in NBD1 (left column) are paired with those in NBD2 (right column) such that horizontal connections between boxes of identical color represent the crosslinks formed between the original five target pairs. Red connecting lines join positions crosslinked with BMH, and blue lines represent crosslinks with the shorter BMOE. Thinner lines, for example, between A462 and S1347, indicate weaker crosslink signals. Black dashed lines connect pairs of positions that could not be crosslinked with either BMOE or BMH.