(Received for publication, June 13, 1996, and in revised form, September 17, 1996)
From the Laboratoire de Neurobiologie Cellulaire,
CNRS, 31 Chemin J. Aiguier, 13402 Marseille Cedex 20, France and
the § Centre d'Immunologie, INSERM-CNRS de
Marseille-Luminy, Parc Scientifique de Luminy, 13288 Marseille Cedex 9, France
The ATP binding cassette transporter ABC1 is a 220-kDa glycoprotein expressed by macrophages and required for engulfment of cells undergoing programmed cell death. Since members of this family of proteins such as P-glycoprotein and cystic fibrosis transmembrane conductance regulator share the ability to transport anions, we have investigated the transport capability of ABC1 expressed in Xenopus oocytes using iodide efflux and voltage-clamp techniques. We report here that ABC1 generates an anion flux sensitive to glibenclamide, sulfobromophthalein, and blockers of anion transporters. The anion flux generated by ABC1 is up-regulated by orthovanadate, cAMP, protein kinase A, and okadaic acid. In other ABC transporters, mutating the conserved lysine in the nucleotide binding folds was found to severely reduce or abolish hydrolysis of ATP, which in turn altered the activity of the transporter. In ABC1, replacement of the conserved lysine 1892 in the Walker A motif of the second nucleotide binding fold increased the basal ionic flux, did not alter the pharmacological inhibitory profile, but abolished the response to orthovanadate and cAMP agonists. Therefore, we conclude that ABC1 is a cAMP-dependent and sulfonylurea-sensitive anion transporter.
ATP binding cassette (ABC)1 transporters are implicated in the vectorial movement of a wide variety of substrates across biological membranes (1, 2). Most of the mammalian ABC transporters identified so far have been associated with clinically relevant phenotypes (2). The human P-glycoprotein confers resistance to chemotherapeutic drugs on tumor cells (3). Persistent hyperinsulinemic hypoglycemia of infancy is associated with mutation of SUR, the receptor for sulfonylureas (4). Cystic fibrosis is caused by mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR), a cAMP-dependent chloride channel (5, 6).
The basic structural unit of an ABC transporter consists of a pair of
nucleotide binding folds (NBF) and two transmembrane domains, each
composed as a rule of six transmembrane spanners (1, 2). Their activity
as transporters is dependent on their interaction with ATP at the NBFs
followed by its hydrolysis (see Fig. 1A) (7-16), and in
some cases evidence has been provided for a further regulation via
phosphorylation of serine/threonine residues in the region linking the
two symmetric halves (6, 17-19). The NBF domains contain the highly
conserved phosphate binding loop (20) that forms intimate contacts with
the - and
-phosphates of bound ATP (21) and an additional
diagnostic motif, the active transport signature, whose function is so
far unknown.
We have recently reported on the functional characterization of a novel ABC transporter, ABC1 (22, 23). This molecule, whose expression during embryonic development correlates with the occurrence of programmed cell death, is required by macrophages during engulfment of cells undergoing apoptosis (23, 24). From a structural standpoint, ABC1 possesses all of the typical features of ABC transporters, i.e. the two symmetric halves each equipped with six transmembrane spanners and an NBF. In addition, a long charged region, reminiscent of the regulatory domain of CFTR but unique in that it is equipped with an extra hydrophobic segment, links the two halves of the molecule (23). The purpose of this work was to analyze the ionic transport capability of ABC1 transporter using radiotracer efflux and electrophysiological techniques.
The ABC1 full-length cDNA was
constructed from the 13C,
10F, and
8a4 overlapping phage
clones (22) using the prokaryotic cloning vector pBluescript
KS+/
(Stratagene). The NgoMI-EcoRI
fragment from the
13C clone, which extends in the 5
region of ABC1
(EMBL accession number X75926[GenBank], nucleotides 1-2567), was juxtaposed to
the EcoRI-HindIII fragment from
10F
(nucleotides 2568-5665) and the HindIII-XbaI
fragment of
8a4 (nucleotides 5666-6916). The final construct,
pABC1KS, was excised by NotI-ApaI in the vector
polylinker and then cloned into the pSP6TN poly(A) vector (pABC1TN)
modified from pSP64T poly(A) (a generous gift of A. Ceriotti, Milano,
Italy) (25) by the insertion of an oligonucleotide linker
(GATCTGGGCCCACTCGAGTTAACGCGGCCGCA) conferring ApaI,
XhoI, HpaI, and NotI as additional
unique sites. The CFTR cDNA construct pACF23 was provided by J. R. Riordan, Scottsdale, AZ, (26).
The Lys1892 in
the second ATP cassette of ABC1 was mutated to methionine by polymerase
chain reaction amplification of the wild-type cDNA with a mutated
5 primer spanning nucleotides 5854-5874 and a vector-specific 3
primer downstream from the multiple cloning sites linker of pSP64TN
poly(A). The 5
sense primer whose sequence is
GGAGCTGGGATGTCA includes the AAG (lysine)
to ATG (methionine) single mutation (in bold) and the wild-type
HpaI restriction site (underlined). The final construct,
K1892M, was obtained after insertion into pABC1TN of the amplification
product digested by HpaI-ApaI and verified by
sequencing.
10 µg of linearized DNA template was transcribed in vitro by Sp6 RNA polymerase following standard protocols (27) for 1 h at 37 °C. After removal of DNA template by treatment with DNase I, the capped cRNA was precipitated and resuspended in diethyl pyrocarbonate-water. The final concentration was adjusted to 200 ng/µl for ABC1 and K1892M cRNA and to 100 ng/µl for CFTR cRNA after analysis on formaldehyde-agarose gel.
Oocyte Expression StudiesStage V Xenopus laevis oocytes were defolliculated by collagenase and microinjected with 50 nl of water or aqueous solutions of cRNA encoding ABC1, K1892M, or CFTR. Oocytes were incubated in modified Barth's saline medium (88 mM NaCl, 1 mM KCl, 0.41 mM CaCl2, 0.33 mM Ca(NO3)2, 0.82 mM MgSO4, 2.4 mM NaHCO3, and 10 mM HEPES, pH 7.4) containing 5 IU/ml penicillin/streptomycin (Life Technologies, Inc.). The protein expression was monitored systematically for every batch of injected oocytes. After overnight metabolic labeling the oocytes were lysed in 0.1 M NaCl, 0.1 M Tris, pH 8.0, 10 mM EDTA, 1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride. The proteins of interest were immunoprecipitated according to the standard protocols (28) and eluted from protein A- or protein G-Sepharose in 20 µl of sample buffer (supplemented with 8 M urea) for 2 min at 50 °C. ABC1 and K1892M were immunoprecipitated using Ab16, a rabbit polyclonal antiserum recognizing the first NBF domain of ABC1 (dilution 1:250) (23), and CFTR with 1 µg of monoclonal mouse anti-human regulatory domain-specific antibody (Genzyme, Cambridge, MA) (28).
Transport AssaysGroups of five oocytes were incubated for
30 min in a 6.5-mm diameter porous bottom dish
(Transwell®, Costar) in 1 ml of ND96 buffer (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, and 5 mM HEPES titrated with NaOH to pH 7.4) containing 1 µM KI (1 µCi of Na125I/ml, DuPont) at room
temperature (adapted from Refs. 29 and 30). After the loading period,
the oocytes were sequentially moved at 1-min intervals through a series
of wells containing 1 ml of buffer. 125I released from the
oocytes in each well was measured by radioactivity counting (Compu
Gamma, LKB). The first two wells were used to establish a stable base
line in ND96 buffer (time 0). After eight 1-min washes, the oocytes
were solubilized in 1 M NaOH and the residual iodide
content was measured. The total amount of 125I (800-1200
cpm/oocyte) at time 0 was calculated as the sum of radioactivity
recovered in each 1-min sample plus the residual iodide content. Efflux
curves were constructed by plotting the percentage of total
radioactivity released in the medium versus time. The
percentage of total at 8 min (T8) was used for
further analysis. The appropriate drug or modified ND96 buffers
(Cl-free ND96 medium, all external Cl
was
replaced by gluconate; Na+-free ND96 medium, 96 mM NaCl was replaced by 96 mM LiCl) was applied
after the stable base line in ND96 buffer was obtained. Oocyte ion
conductance was measured by conventional two-electrode voltage clamp
(26) and bathed in ND96 medium. Data are presented as the mean ± S.D. of n observations. Statistical significance was
assessed at the 95% confidence level with Student's t
test.
Ab16 immunoprecipitates from
samples of 10 unlabeled oocytes were equilibrated in 50 mM
Tris, pH 7.5, 10 mM MgCl2, and 100 mg/ml bovine
serum albumin and then allowed to react in the presence of 50 ng of
catalytic subunit of protein kinase A (Sigma) and 10 µCi of [-32P]ATP for 1 h at 30 °C. The
reaction was stopped by the addition of sample buffer. The patterns of
phosphorylation in oocytes injected either with ABC1 or water were
analyzed by SDS-polyacrylamide gel electrophoresis on 7.5%
polyacrylamide gels and autoradiography.
N-(2-(p-Bromocinnamylamino)ethyl)-5-isoquinolinesulfonamide (H89) and okadaic acid were from RBI (Natick, MA), and A27187, forskolin, and cpt-cAMP were from Boehringer Mannheim. The drugs were dissolved in dimethyl sulfoxide stock solutions and used at a final dimethyl sulfoxide concentration of 0.1%. All other reagents were from Sigma.
After injection of cRNA, the synthesis of ABC1 at the predicted size (220 kDa) was monitored after immunoprecipitation by Ab16 (Fig. 1B, lane 2). The expression of K1892M, an ABC1 mutant in which the Lys1892 of the Walker A motif in the second nucleotide binding fold (NBF2) has been replaced by a methionine, was monitored in the same way (Fig. 1B, lane 3). No variations with respect to the wild-type protein were detected either in the temporal onset of protein synthesis or in the ratio of produced protein to injected cRNA. The expression of CFTR in oocytes was monitored by immunopurification with a monoclonal mouse anti-human regulatory domain-specific antibody (Fig. 1B, lane 5).
ABC1 Generates an Anion Efflux in Xenopus OocytesSince the
activity of CFTR and P-glycoprotein is associated with chloride
transport (6, 18, 19, 26, 31), we wished to determine the properties of
ABC1 expressed in Xenopus oocytes using the radiotracer
iodide efflux technique. First, we analyzed the activity of two well
characterized Cl channels in Xenopus oocytes.
The endogenous Ca2+-dependent Cl
channel can be activated by promoting Ca2+ influx using
A23187 (29, 32) as shown in Fig. 1C. Indeed, the presence of
10 µM A23187 in the buffer increased (
3-fold) the
iodide efflux in water-injected oocytes (efflux at 8 min, T8 = 68 ± 9%, n = 4)
compared with basal oocytes (T8 = 21 ± 5%, p < 0.01, n = 4). After
expression in Xenopus oocytes, CFTR generates a
cAMP-dependent chloride current (26, 29). In our
experiments the iodide efflux in CFTR-expressing oocytes was increased
(p < 0.01) upon addition of cAMP agonists (Fig.
1D, CFTR basal (T8 = 31 ± 5%, n = 5) and CFTR + cAMP
(T8 = 53 ± 11%, n = 5)).
Having determined that our technique offers a convenient way to
evaluate the activity of both endogenous and expressed chloride
channels in oocytes, we then investigated whether an iodide efflux is
detectable in ABC1-injected oocytes. Fig. 1, E and
F, shows that the expression of ABC1 in oocytes is
associated with an increase (p < 0.01) of the iodide
efflux amplitude (T8 = 49 ± 7%,
n = 12) greater than that seen in water-injected
oocytes (T8 = 19 ± 7.5%,
n = 14) in Ca2+-free ND96 medium. Similar
results were obtained in ND96 medium containing 2 mM
Ca2+ (ABC1, T8 = 50 ± 3%,
n = 7; water, T8 = 20 ± 6%, n = 9, p < 0.01). The amplitude
of the iodide efflux appeared to be directly dependent on the amount of
injected ABC1 mRNA because 10-fold less mRNA led to a reduced
efflux (T8 = 36 ± 7%, n = 6).
The equimolar substitution of extracellular chloride with the nonpermeant anion gluconate did not affect the amplitude of the A23187-activated efflux from water-injected oocytes (gluconate + A23187, T8 = 67.4 ± 7%, n = 2; ND96 + A2318, T8 = 68 ± 9%, n = 4 (not shown)). In contrast, a similar procedure decreased the magnitude of the efflux from ABC1-injected oocytes (gluconate, T8 = 34.2 ± 6%, n = 8 (Fig. 1F)), whereas substitution of extracellular sodium with lithium had no effect (T8 = 49.8 ± 5%, n = 5 (Fig. 1F)).
Pharmacological SpecificityCompared with the basal activity
of ABC1 (Fig. 2B, noted as DMSO,
T8 = 51 ± 4%, n = 5), the
Cl channel blockers DIDS (500 µM,
n = 6 (Fig. 2A)), diphenylamine-2-carboxylic acid (500 µM, n = 3), and flufenamic acid
(500 µM, n = 4) blocked 50-80%
(p < 0.01) of the iodide efflux generated by the
expression of ABC1 (Fig. 2B) (T8 = 22.4 ± 4, 29 ± 1, and 33 ± 8%, respectively). ABC1
activity was also sensitive to bumetanide (T8 = 42 ± 6%, 200 µM, n = 3, p < 0.05) and furosemide (T8 = 33 ± 7%, 200 µM, n = 3, p < 0.01 (Fig. 2B)). The prostaglandin
transporter (33) inhibitor sulfobromophthalein (BSP) (500 µM, n = 8) inhibited
80% of the ABC1
activity (T8 = 24 ± 6%, p < 0.01 (Fig. 2, A and B)). The sulfonylurea
compound glibenclamide (100 µM), which is an inhibitor of
the activity of KATP (34) and CFTR (35) channels, almost completely
blocked the activity of ABC1 (T8 = 21 ± 7%, n = 6, p < 0.01 (Fig. 2,
A and B)) and CFTR (CFTR + cAMP,
T8 = 53 ± 11%, n = 5;
CFTR + cAMP + glibenclamide, T8 = 31 ± 5%, n = 4, p < 0.01) in oocytes.
Verapamil (T8 = 49 ± 9%, 200 µM, n = 3), a P-glycoprotein inhibitor
(31), and the potassium channel inhibitor tetraethylammonium (10 mM, T8 = 53 ± 5%,
n = 2) had no significant effect on the ABC1 activity
(Fig. 2B).
Regulation of ABC1 Activity
Since CFTR can be regulated by
cAMP (6, 26, 36), we analyzed the effects of cAMP on ABC1. Fig.
3 shows that cpt-cAMP (500 µM) increased
(p < 0.01) the magnitude of the iodide efflux in ABC1
oocytes (T8 = 63 ± 6%, n = 6; basal T8 = 49 ± 7%). These observations were confirmed using forskolin (T8 = 68 ± 5.7%, 10 µM, n = 2 (Fig.
3B)) or a mixture containing cpt-cAMP (500 µM) and IBMX (1 mM, T8 = 67.5 ± 7%, n = 2 (Fig. 3B)). The protein kinase A
inhibitor H89 (20 µM) decreased (p < 0.01) the basal activity of ABC1 (T8 = 37 ± 3.5%, n = 4 (Fig. 3B)). The protein phosphatase inhibitor okadaic acid (10 µM) slightly
increased (p < 0.05) the ABC1 activity
(T8 = 54 ± 3%, n = 3 (Fig. 3B)). In water-injected oocytes, exposure of cAMP
agonists failed to generate a significant anion transport
(T8 = 22 ± 5%, n = 3 (Fig. 3B)) (see Refs. 26 and 29). The cAMP stimulation may
be due to the direct phosphorylation of ABC1 as shown by in
vitro phosphorylation of immunopurified ABC1 by protein kinase A
(Fig. 3C).
Role of NBF2
ABC1 in which the Lys1892 in NBF2
has been replaced by a methionine (K1892M mutant) and expressed in
oocytes (see Fig. 1B) generated an iodide efflux that was
increased (T8 = 57 ± 6.7%,
n = 11) compared with wild-type ABC1
(T8 = 49 ± 7%, n = 12, p < 0.01 (see Fig. 3B)). DIDS
(T8 = 27 ± 7%, 500 µM,
n = 4), BSP (T8 = 28 ± 1%, 500 µM, n = 4), and glibenclamide
(T8 = 26 ± 3%, 100 µM,
n = 6) inhibit (p < 0.01) the
transport activity of K1892M (Fig. 4, A-D). Fig. 4B shows that K1892M failed to respond to 500 µM cpt-cAMP (T8 = 55.6 ± 4.6%, n = 8). Whereas the phosphate analog
orthovanadate increased (p < 0.01) the amplitude of
the efflux of wild-type ABC1 (1 mM, n = 6 (Fig. 3B)) to T8 = 63 ± 5.7%
(an effect inhibited by glibenclamide, 100 µM,
n = 2 (Fig. 3B)), the activity of K1892M was
reduced (T8 = 46 ± 3%, n = 11, p < 0.05 (Fig. 4B)).
ABC1 Is Not a Chloride Channel
We then examined the
electrophysiological properties of oocytes expressing ABC1 compared
with water- and CFTR-injected oocytes (Fig. 5). The membrane potential
(Em) was not affected by the expression of the
different transporters (ABC1, Em = 42 ± 5 mV, n = 13; CFTR, Em =
45 ± 5 mV, n = 5; water, Em =
43 ± 6 mV, n = 10). The holding current
was similar in water (
13 ± 3.4 nA, n = 7) and
ABC1 (
14 ± 4 nA, n = 12) oocytes voltage
clamped at
60 mV and bathed in ND96 medium. The
Ca2+-activated Cl
channel activity induced by
A23187 (10 µM,
126 ± 48 nA at
60 mV,
p < 0.01, n = 5) and inhibited by DIDS
(500 µM,
30 ± 8 nA at
60 mV, n = 3) was not affected by the presence of ABC1 (A23187,
114 ± 50 nA at
60 mV, n = 6; A23187 + DIDS,
18 ± 4 nA
at
60 mV, n = 3 (Fig. 5A)). When cpt-cAMP
(500 µM), IBMX (1 mM), or forskolin (10 µM) was added to the perfusion medium, no modification of
the holding current was observed in the presence of ABC1 (
15 ± 4 nA, n = 14) or in its absence (
14.5 ± 6 nA,
n = 20). In contrast, and as already reported (26, 29),
CFTR generated a cAMP-dependent current (n = 10 (Fig. 5B)) inhibited by glibenclamide (100 µM, n = 4 (not shown)). Like wild-type
ABC1, K1892M also failed to generate a current either under resting
conditions or in the presence of cAMP agonists (data not shown).
We report here on the characterization, pharmacological profile,
and regulation of ABC1 as a cAMP-dependent and
glibenclamide-sensitive anion transporter. Among the ABC proteins
involved in anion transport, CFTR is a chloride channel (5, 6, 26, 36)
and the P-glycoprotein regulates an endogenous chloride channel (2, 18,
19, 31). In this paper, we provide evidence that ABC1 shares with CFTR (5, 6, 26, 36) the ability to transport anions and to be up-regulated
by cAMP, IBMX, okadaic acid, and vanadate. However, as opposed to CFTR,
ABC1 behaves as an electroneutral anion exchanger since no significant
chloride current was detected from oocytes expressing ABC1 and the
ABC1-generated anion efflux is dependent on extracellular chloride
concentration. In fact, the substitution of extracellular chloride by
gluconate reduces the efflux from ABC1-injected oocytes as would be
expected from a parallel anion antiport (e.g.
Cl/HCO3
exchanger) (37,
38), whereas in similar conditions the activity of the
A23187-stimulated efflux through the
Ca2+-dependent Cl
channel is not
affected.
Like CFTR (6, 29, 35), ABC1 is inhibited by glibenclamide, flufenamic acid, and diphenylamine-2-carboxylic acid. In addition, ABC1 is also inhibited by the organic anion sulfobromophthalein and by the broad acting blocker of anion transporters, DIDS.
We provide evidence that activation of protein kinase A increases the activity of ABC1 and that ABC1 is phosphorylated in vitro by protein kinase A, which suggests that direct phosphorylation of the protein might take place in vivo. On the other hand, the activity of ABC1 is unaffected by extracellular Ca2+ concentration or by Ca2+ entry triggered by A23187. Therefore, ABC1 appears to be up-regulated by cAMP-dependent protein kinases and by the phosphatase inhibitors okadaic acid and vanadate. Our observations are in line with the current opinion that kinases modulate their activities by phosphorylating specific sites of ABC transporters (6, 17, 19, 36).
In other ABC transporters, mutating the conserved lysine in NBF was found to severely reduce or abolish hydrolysis of ATP, which in turn alters the activity of the transporter (1, 2, 14). For example, mutation of the Lys1250 into the NBF2 of CFTR modifies the kinetic parameters of the channel activity (14-16). The corresponding mutation in ABC1 (i.e. K1892M) produces a transporter with an increased basal ionic flux and a conserved pharmacological inhibitory profile (i.e. inhibition by glibenclamide, DIDS, and BSP). In contrast, the K1892M mutant is unaffected by cAMP, and a reduced efflux is observed after vanadate treatment. A reasonable interpretation of these results is that vanadate in the wild-type ABC1 primarily impairs the function of NBF2, as is the case in the K1892M mutant. The reversed effect of vanadate treatment on the mutant might result from other actions of this compound either at NBF1 and/or as a phosphatase inhibitor. Similar effects of vanadate on CFTR and its mutants have been reported and led to a model of the transport cycle in which hydrolysis at NBF2 plays an inhibitory role (14-16, 36).
The identification of ABC1 as a cAMP-dependent anionic transporter does not provide any information on the nature of the transported substrate, nor does it shed light on the role played by ABC1 during engulfment of apoptotic corpses. Nonetheless, the finding of an ABC1-dependent anionic flux with a clear-cut pharmacological profile provides an extremely useful experimental tool. We can in fact readily investigate how modulations of ABC1 transporter activity in phagocytic cells, like mouse peritoneal macrophages, affect their responses to specific physiopathological challenges.
We gratefully acknowledge M. Crest, G. Jacquet, and M. F. Luciani for technical assistance, Aldo Cerriotti for having introduced A. B. to oocyte techniques, J. R. Riordan for the generous gift of full-length CFTR cDNA, and P. Golstein for critical reading of the manuscript.