Unlike prokaryotic ABC proteins, NBD1 and NBD2 of SUR1 show significant sequence differences: thus, the ATPase activity of the isolated recombinant NBD homodimers will not necessarily refl
Trang 1Heidi de Wet1, Michael V Mikhailov1, Constantina Fotinou1, Mathias Dreger1, Tim J Craig1,
Catherine Ve´nien-Bryan2and Frances M Ashcroft1
1 Henry Wellcome Centre for Gene Function, Department of Physiology, Anatomy and Genetics, University of Oxford, UK
2 Laboratory of Molecular Biophysics, Department of Biochemistry, University of Oxford, UK
ATP-sensitive potassium (KATP) channels couple cell
metabolism to membrane excitability and
transmem-brane ion fluxes In pancreatic b-cells, they are of
crucial importance for regulating insulin secretion [1]
At substimulatory glucose concentrations, KATP chan-nels are open and generate a negative potential that
Keywords
ABCC8; ATPase activity; K ATP channel;
nucleotide-binding domain; sulfonylurea
receptor
Correspondence
F M Ashcroft, University Laboratory of
Physiology, Parks Road, Oxford OX1 3PT,
UK
Fax: +44 1865 285812
Tel: +44 1865 285810
E-mail: frances.ashcroft@physiol.ox.ac.uk
(Received 4 April 2007, revised 8 May 2007,
accepted 14 May 2007)
doi:10.1111/j.1742-4658.2007.05879.x
The ATP-sensitive potassium (KATP) channel couples glucose metabolism
to insulin secretion in pancreatic b-cells It comprises regulatory sulfonyl-urea receptor 1 and pore-forming Kir6.2 subunits Binding and⁄ or hydro-lysis of Mg-nucleotides at the nucleotide-binding domains of sulfonylurea receptor 1 stimulates channel opening and leads to membrane hyperpolari-zation and inhibition of insulin secretion We report here the first purifica-tion and funcpurifica-tional characterizapurifica-tion of sulfonylurea receptor 1 We also compared the ATPase activity of sulfonylurea receptor 1 with that of the isolated nucleotide-binding domains (fused to maltose-binding protein to improve solubility) Electron microscopy showed that nucleotide-binding domains purified as ring-like complexes corresponding to 8 momomers The ATPase activities expressed as maximal turnover rate [in nmol
PiÆs)1Æ(nmol protein))1] were 0.03, 0.03, 0.13 and 0.08 for sulfonylurea receptor 1, nucleotide-binding domain 1, nucleotide-binding domain 2 and
a mixture of nucleotide-binding domain 1 and nucleotide-binding domain 2, respectively Corresponding Km values (in mm) were 0.1, 0.6, 0.65 and 0.56, respectively Thus sulfonylurea receptor 1 has a lower Km than either of the isolated nucleotide-binding domains, and a lower max-imal turnover rate than nucleotide-binding domain 2 Similar results were found with GTP, but the Kmvalues were lower Mutation of the Walker A lysine in nucleotide-binding domain 1 (K719A) or nucleotide-binding domain 2 (K1385M) inhibited the ATPase activity of sulfonylurea recep-tor 1 by 60% and 80%, respectively Beryllium fluoride (Ki16 lm), but not MgADP, inhibited the ATPase activity of sulfonylurea receptor 1 In con-trast, both MgADP and beryllium fluoride inhibited the ATPase activity of the nucleotide-binding domains These data demonstrate that the ATPase activity of sulfonylurea receptor 1 differs from that of the isolated nucleo-tide-binding domains, suggesting that the transmembrane domains may influence the activity of the protein
Abbreviations
ABC, ATP-binding cassette; BeF, beryllium fluoride; CFTR, cystic fibrosis transmembrane conductance regulator; DDM, dodecylmaltoside; DMPC, 1,2-dimyristoyl-sn-glycero-phosphocholine; EM, electron microscopy; KATP, ATP-sensitive potassium; MBP, maltose-binding protein; MRP1, multidrug resistance protein 1; NBD, nucleotide-binding domain; SUR, sulfonylurea receptor; SUR1F, full-length flagged-tagged SUR1;
WA, Walker A; WB, Walker B.
Trang 2keeps voltage-gated Ca2+ channels closed and
abol-ishes Ca2+ influx Because a rise in intracellular Ca2+
is needed to stimulate insulin granule release, this
pre-vents insulin secretion When plasma glucose levels
increase, glucose uptake and metabolism lead to
chan-ges in the intracellular concentrations of adenine
nucleotides that close KATP channels, triggering Ca2+
channel opening, Ca2+influx, elevation of intracellular
Ca2+and insulin release
The b-cell KATP channel is a large octameric
com-plex that comprises a central tetrameric Kir6.2 pore
surrounded by four sulfonylurea receptor (SUR) 1
subunits [2] Both Kir6.2 and SUR1 subunits are
involved in the metabolic regulation of channel
activ-ity: ATP binding to Kir6.2 causes channel inhibition
[3], whereas interaction of Mg-nucleotides (MgATP
and MgADP) with SUR1 stimulates channel opening
[4–6] Impairment of nucleotide interactions with either
subunit can lead to neonatal diabetes or its converse,
congenital hyperinsulinism [1]
SUR belongs to the ATP-binding cassette (ABC)
protein superfamily [7] It has 17 transmembrane
heli-ces and two large cytosolic loops, which contain the
nucleotide-binding domains (NBDs) NBD1 and
NBD2 As in all ABC proteins, each NBD contains a
highly conserved Walker A (WA) and Walker B (WB)
motif involved in ATP binding and hydrolysis, an
invariant ‘signature sequence’, and several other
con-served residues Crystallization of a number of
pro-karyotic NBDs and ABC proteins indicates that they
associate in a sandwich dimer conformation [8–11], in
which residues from the WA and WB motifs of one
NBD interact with the signature sequence of the other
NBD to form separate ATP-binding sites, with distinct
properties Each ATP-binding site therefore contains
contributions from both NBD1 and NBD2 Evidence
of physical interaction between the NBDs, and
molecular modeling studies, support the idea that
SUR1 also conforms to the sandwich dimer model
[12,13] Functional studies demonstrate that formation
of such a sandwich dimer is critical for driving gating
of cystic fibrosis transmembrane conductance regulator
(CFTR) channels [14], but this has not yet been
dem-onstrated for KATPchannels
There are two genes that encode SUR, ABCC8
(SUR1) and ABCC9 (SUR2) [15–17] The latter exists
in several splice variants, the most important being
SUR2A and SUR2B Differences in the SUR subunit
contribute to the variable metabolic sensitivities of
KATP channels in different tissues For example, even
when heterologously expressed in the same cell,
recom-binant Kir6.2–SUR2A channels open less readily on
metabolic inhibition than Kir6.2–SUR1 channels [18]
It has been suggested that this may relate to differ-ences in the ATPase activity of SUR1 and SUR2 [19] The ATPase activity of full-length SUR1 has not been measured directly to date However, MgATP hydrolysis has been measured directly for recombinant proteins in which either NBD1 or NBD2 of SUR was fused to the maltose-binding protein [19–21] ATPase activity of native SUR (i.e containing both NBDs and transmembrane domains) has also been inferred by comparing covalent labeling with 8-azido-[32P]ATP[aP] and 8-azido-[32P]ATP[cP] [22] In these studies, how-ever, hydrolysis by NBD2, but not NBD1, was detec-ted Unlike prokaryotic ABC proteins, NBD1 and NBD2 of SUR1 show significant sequence differences: thus, the ATPase activity of the isolated recombinant NBD homodimers will not necessarily reflect that of the NBD heterodimer expected for native SUR1 Fur-thermore, the presence of the transmembrane domains
in SUR1 may influence ATPase activity We have therefore purified SUR1 and compared its capacity to hydrolyze ATP and GTP with that of isolated NBD1,
or NBD2, of SUR1 [fused to maltose-binding protein (MBP)] We also measured the ATPase activity of a mixture of NBD1 and NBD2 proteins In addition, the effects of the inhibitors beryllium fluoride (BeF) and MgADP were explored
Results
Purification and characterization of SUR1 MBP–NBDs and SUR1F
SDS⁄ PAGE and Coomassie staining revealed a single major band following purification of full-length flagged-tagged SUR1 (SUR1F), MBP–NBD1 and MBP–NBD2 (Fig 1A) For simplicity, we refer to these proteins subsequently as SUR1, NBD1 and NBD2 MALDI-TOF analysis of total purified pro-teins confirmed their identities as well as the absence
of any other contaminating ATPases (data not shown) Additional bands visible on these gels were identified
as degradation products by MALDI-TOF analysis of gel cut-outs
Gel filtration of SUR1 yielded two fractions (Fig 1Ba) The smaller peak corresponds to the molecular mass expected for monomeric SUR1 and the larger peak runs as expected for a mixture of tetra-meric and oligotetra-meric species
Gel filtration revealed that NBD1, NBD2 or a 1 : 1 mixture of NBD1 and NBD2 eluted as a single sharp peak corresponding to a single oligomeric species (Fig 1Bb) Calculated molecular masses gave approxi-mate sizes of 8 monomers for NBD1 and 9
Trang 3mono-mers for NBD2 No larger aggregates or other protein
species were detected SDS⁄ PAGE analysis of the
pro-teins in the respective gel filtration fractions confirmed
their identities as MBP–NBD1 and MBP–NBD2
Because gel filtration suggested that NBD1 and NBD2
associate as oligomers, we collected the peak eluates
and examined them by negative stain electron
micros-copy (EM) This revealed that both proteins formed
ring-like oligomers For NBD1, the outer diameter of
the projected structure was between 120 and 140 A˚
and the inner diameter was 40–75 A˚ (Fig 1C) A
sim-ilar structure was observed for NBD2 and for a 1 : 1
mixture of NBD1 and NBD2 Oligomerization was
independent of the presence of MgATP (data not
shown) MBP alone did not form ring-like oligomers
(data not shown), suggesting that the interaction is mediated by the NBD part, rather than the MBP part,
of the MBP–NBD fusion proteins
Nucleotide hydrolysis by SUR1 Recombinant full-length SUR1 hydrolyzed MgATP very slowly, with a maximal turnover rate of 0.03 s)1 (Table 1 and Fig 2A) and a Km of 0.1 mm
No ATPase activity was detected in the absence of
Mg2+ or from protein purified from cells transfected with an SUR1 construct lacking the FLAG tag used for affinity purification (Fig 2A) Because KATP channels are stimulated by GTP, via interaction with the NBDs
of SUR1 [32], we also investigated the ability of SUR1
Octamer Tetramer
Monomer
Oligomer
A
250
150
100
75
50
kDa
1 2
SUR1
188 98
62 49 38 28
kDa
1 2 3
NBD1 NBD2
Fig 1 Purification of SUR1, NBD1 and NBD2 (A) Coomassie-stained denaturing gels of purified SUR1 (left, lane 2), MBP– NBD1 (right, lane 1), and MBP–NBD2 (right, lane 2) Molecular mass markers, lane 1 (left) and lane 3 (right) Samples shown are the purified eluates from affinity resins, and and were not subjected to further purifica-tion by gel filtrapurifica-tion (Ba) Gel filtrapurifica-tion analy-sis of purified SUR1 (Bb) Gel filtration analysis of purified MBP–NBD1 (Ca) Negat-ive stain electron micrograph of MBP– NBD1 The scale bar is 100 nm Black arrows indicate ring-like structures The white arrow points to a stack of rings (Cb) Ten different classes of particle The size of the boxes is 280 A ˚
Trang 4to hydrolyze GTP Figure 2B shows that GTP was also
hydrolyzed, but with a much higher Km (> 1 mm),
which suggests that GTP binds to SUR1 with a lower
affinity than ATP The turnover rate, estimated by
fit-ting the data to the Michaelis–Menten equation, was similar to that of ATP
Mutation of residues in the Walker motifs of SUR impairs channel activation by MgATP and MgADP [5] Mutating the WA lysine in NBD1 of SUR1 to alanine (K719A) reduced ATPase activity by approxi-mately 60%, whereas mutating the WA lysine in NBD2 to methionine (K1385M) inhibited ATPase activity by about 80% (Fig 2C) when compared to wild-type controls assayed in parallel (n¼ 2) Neither mutation affected the Km for ATP These data are consistent with the idea that these mutations affect the rate of hydrolysis but do not influence ATP binding
Table 1 ATPase activities and kinetic constants.
Construct
Turnover rate
(nmol P i Æs)1Æ
nmol)1protein)
V max
(nmol P i Æmin)1Æ
mg)1)
K m
MBP–NBD1 0.03 ± 0.003 23.8 ± 2.40 0.6 ± 0.09 14
MBP–NBD2 0.13 ± 0.01 103.81 ± 8.70 0.65 ± 0.13 12
MBP–NBD1 + 2 0.08 ± 0.01 61.22 ± 6.78 0.56 ± 0.11 10
[ATP] (m M )
[ATP] (m M )
Wt of KA KM KA
Wt of KM
[inhibitor] (m M ) 0.001 0.01 0.1 1 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
0.0
0.2
0.4
0.6
0.8
1.0
A
0
2
4
6
8
10
12
MgATP
No FLAG tag
Mg free
C
D
[GTP] (m M )
0 2 4 6 8 10
12
B
Fig 2 ATPase activity of SUR1 (A) ATPase activity in the presence (filled circles, n ¼ 6) or absence of Mg2+(crosses, n ¼ 2) of purified SUR1, at 37 C Membranes from cells expressing SUR1 lacking a FLAG tag and purified as usual show no ATPase activity (open circles,
n ¼ 1) (B) GTPase activity of SUR1 (n ¼ 4) The line is fitted to the Michaelis–Menten equation, with an estimated V max and Kmof 10 nmol
P i Æmin)1Æmg)1and 0.86 m M , respectively (C) ATPase activity of SUR1 containing the mutations K719A (triangles, n ¼ 2) or K1385M (dia-monds, n ¼ 2) Data are expressed as a fraction of the turnover rate for wild-type SUR1 assayed in parallel (D) Inhibition of ATPase activity
at 1 m M MgATP by ADP (open circles, n ¼ 4) and by BeF (filled circles, n ¼ 3) Data are expressed as a fraction of the turnover rate in the absence of inhibitor.
Trang 5BeF (BeF3 and BeF42–) is a potent inhibitor of
ATP hydrolysis by many ABC proteins, including
P-glycoprotein [33] and the isolated NBD2 of SUR2A
[20] It acts by arresting the ATPase cycle in the
pre-hydrolytic conformation [34] The ATPase activity of
SUR1 was potently and completely inhibited by BeF
(Ki16 lm; Table 2; Fig 2D) Previous studies have
shown that MgADP blocks the ATPase activity of
NBD2 of SUR2A [20] by trapping the ATPase cycle in
the posthydrolytic conformation Unexpectedly,
how-ever, no effect of ADP on the ATPase activity of
SUR1 was observed (Fig 2D)
The ATPase activity of SUR1 is about 10-fold less
than what we previously reported for the complete
octameric KATP channel complex, SUR1–Kir6.2
(turn-over rate 0.4 ± 0.03 s)1 calculated from Mikhailov
et al [2])
Nucleotide hydrolysis activities of NBD1 NBD1 displayed low maximal ATPase activity, similar
to that of SUR1, but the Kmwas about six-fold larger (0.6 mm, P < 0.005) (Table 1 and Fig 3A) This sug-gests that ATP binds more tightly to full-length SUR1 than to the isolated NBD1 GTP was hydrolyzed with
a Kmhigher than that for ATP (Fig 3B) There was a very small, but significant, apparent hydrolysis of ADP (Table 3, Fig 3B) Negligible ATP hydrolysis was observed in the absence of Mg2+ or in protein prepa-rations from cells expressing MBP alone (Fig 3A) BeF blocked ATP hydrolysis at NBD1 with a
Ki of 33 lm (Table 2 and Fig 3C) Unlike with SUR1, however, inhibition appeared to be incom-plete, and the maximal block was 76% In marked contrast to SUR1, MgADP inhibited ATP hydrolysis
[Nucleotide] (m M )
0 5 10 15 20 25 30
ADP GTP
[ATP] (m M )
0
5
10
15
20
25
A
C
B
[Inhibitor] (m M )
0.0
0.2
0.4
0.6
0.8
1.0
1.2
BeF
ADP
MBP
–Mg2+
+Mg2+
Fig 3 ATPase activity of NBD1 (A) ATPase activity of NBD1 in the presence (filled circles, n ¼ 14) or absence (open circles,
n ¼ 2) of Mg 2+ , at 37 C MBP control (open triangles, n ¼ 2) (B) GTP (filled circles, n ¼ 3) and ADP (filled triangles, n ¼ 3) hydrolytic activity The line is fitted to the Michaelis–Menten equation, through the GTP data points with an estimated Vmaxand
Kmof 39 nmol PiÆmin)1Æmg)1and 2.6 m M , respectively (C) Inhibition of ATPase activity
at 1 m M MgATP by ADP (open circles, n ¼ 3) or BeF (filled triangles, n ¼ 3) Data are expressed as a fraction of the turnover rate
in the absence of inhibitor.
Table 2 Inhibition constants for ADP and BeF ND, not detected n indicates the number of different protein preparations.
Construct
IC 50 (ADP) (m M )
K i (ADP)
IC 50 (ADP) (m M )
K i (BeF)
Trang 6at NBD1, with a Ki of 2.1 mm (Table 2 and
Fig 2C)
Nucleotide hydrolysis activities of NBD2
The maximal ATPase activity of NBD2 was about
four-fold greater than that of either SUR1
(P < 0.001) or NBD1 (P < 0.001) The Km was
sim-ilar to that of NBD1 and about six-fold larger than
that of SUR1 (Table 1, Fig 4A (P < 0.005) These
results suggest that the ATP-binding affinities of
NBD1 and NBD2 are similar, but that the hydrolytic
step occurs more rapidly in NBD2, and that,
com-pared with SUR1, the ATP-binding affinity of NBD2
is less but hydrolysis is faster GTP was also
hydro-lyzed, with a Km higher than that for ATP (Fig 4B)
MgADP was hydrolyzed at very low rate, but this was
about three-fold greater than that of NBD1 (Table 3, Fig 4B) Negligible ATP hydrolysis was observed in the absence of Mg2+
As found for NBD1, MgATP hydrolysis (1 mm) was inhibited by both BeF and MgADP (Table 2 and Fig 4C) The Kifor BeF inhibition (19 lm) was lower than that for NBD1, and maximal inhibition was 86% ATP hydrolysis was also inhibited by MgADP, with a Kiof 0.84 mm (Table 2)
Nucleotide hydrolysis activities of NBD1 + NBD2 Interactions of the isolated MBP–NBDs of the CFTR, or the multidrug resistance protein MRP1, have been demonstrated previously [35,36] We there-fore examined ATP hydrolysis in a 1 : 1 mixture of
NBD1 + NBD2 mixture had a maximal turnover rate intermediate between that of NBD1 and NBD2 alone, and the Km(0.56 mm) was not significantly dif-ferent from that of either NBD1 or NBD2 (Table 1, Fig 5A) Thus, as found for MRP1 [36], mixing the two NBDs did not have a major impact on the cata-lytic activity of either NBD
GTP was hydrolyzed by the NBD1 + NBD2 mix-ture with Km> 1 mm (Fig 5B) As observed for the
[ATP] (m M )
0 20 40 60 80 100 120 140
A
C
B
[Nucleotide] (m M )
0 20 40 60 80 100
120 GTP ADP
[Inhibitor] (m M )
0.0 0.2 0.4 0.6 0.8 1.0 1.2
ADP BeF
-Mg2+
+Mg2+
Fig 4 ATPase activity of NBD2 (A) ATPase
activity of NBD2 in the presence (filled
cir-cles, n ¼ 12) or absence (open circles, n ¼
2) of Mg 2+ , at 37 C (B) GTP (filled circles,
n ¼ 3) and ADP (filled triangles, n ¼ 3)
hydrolytic activity The line is fitted to the
Michaelis–Menten equation, with an
estima-ted Vmaxand Kmof 153 nmol PiÆmin)1Æmg)1
and 2.2 m M , respectively (C) Inhibition of
ATPase activity at 1 m M MgATP by ADP
(open circles, n ¼ 3) or BeF (filled triangles,
n ¼ 4) Data are expressed as a fraction of
the turnover rate in the absence of inhibitor.
Table 3 ADPase activities and kinetic constants.
Construct
Turnover rate
(nmol P i Æs)1Æ
nmol)1protein)
V max
(nmol P i Æ min)1Æmg)1)
K m
MBP–NBD1 0.005 ± 0.001 3.83 ± 1.26 0.17 ± 0.08 3
MBP–NBD2 0.014 ± 0.003 11.26 ± 2.40 0.07 ± 0.02 3
MBP–NBD1 + 2 0.012 ± 0.002 9.80 ± 2.69 0.06 ± 0.01 3
Trang 7individual domains, the NBD mixture apparently
hydrolyzed MgADP but with a very low turnover rate
(Table 3, Fig 5B), and no hydrolysis was observed in
the absence of Mg2+
MgATP hydrolysis was inhibited by BeF with a Ki
of 20 lm (Table 2, Fig 5C), which is not significantly
different from that of either NBD1 (30 lm) or
NBD2 (19 lm) alone However, the Ki for MgADP
inhibition of ATP hydrolysis was significantly less
than that of NBD1 (P < 0.05): it was also lower than
that of NBD2, although this difference was not
significant (Table 2) A value intermediate between
those of NBD1 and NBD2 would be expected if
the NBDs were functionally independent: thus, these
data suggest that the NBDs may functionally interact
when mixed This finding is consistent with the
idea that at least some heterodimers of NBD1
and NBD2 are present in the NBD1 + NBD2 mixture
Discussion
Nucleotide handling by SUR1
The ATPase activity of SUR1 was 10-fold lower, and
the Km three-fold larger, than that measured for the
purified complete KATP channel (Kir6.2–SUR1)
com-plex [2] This suggests that ATP binds with lower
affinity and the rate of ATP hydrolysis is faster in the KATP channel complex than in SUR1 alone CryoEM analysis revealed that the KATP channel associates as a large octameric complex in which the individual SUR1 subunits are tightly packed around
a central Kir6.2 tetrameric pore [2] The higher ATPase activity of the KATP channel complex might therefore result from interactions between adjacent SUR1 subunits, and⁄ or between SUR1 and Kir6.2, that enhance cooperativity and⁄ or crosstalk between the NBDs
The ATPase activity of purified SUR1 (Vmax,
9 nmol PiÆmin)1Æmg)1) is at the lower end of the range found for MRP1: from 5 to 10 nmol PiÆ min)1Æmg)1 [37] to 470 nmol PiÆmin)1Æmg)1 [38] It is less than that reported for CFTR (60 nmol PiÆ min)1 Æmg)1 protein [39]), or P-glcyoprotein (320–3900 nmol PiÆ min)1Æmg)1 [40,41]), but higher than that found for ABCR (1.3 mol PiÆmin)1Æmg)1 [42]) It is possible that the lower ATPase activity of SUR1 is related to the unique role of this ABC protein as a channel regulator rather than a transporter It is worth noting that the activity of other ABC transporters, including the closely related MRP1 [38], are stimulated by their substrates It is possible that mechanism by which Kir6.2 stimulates the ATPase activity of SUR1 resem-bles this substrate activation
0
20
40
60
80
A
C
B
0 10 20 30 40 50 60
[Inhibitor] (m M )
0.0
0.2
0.4
0.6
0.8
1.0
1.2
ADP
ADP GTP
BeF
Fig 5 ATPase activity of NBD1 + NBD2 (A) ATPase activity of the NBD1 + NBD2 mixture in the presence (filled circles, n ¼ 10) or absence (open circles, n ¼ 2) of
Mg2+, at 37 C (B) GTP (circles, n ¼ 3) and ADP (triangles, n ¼ 3) hydrolytic activity of NBD1 + NBD2 The line through the GTP data points is fitted to the Michaelis– Menten equation, with an estimated Vmax and Kmof 70 nmol PiÆmin)1Æmg)1and 1.7 m M , respectively (C) Inhibition of ATPase activity at 1 m M MgATP by ADP (open circles, n ¼ 3) or BeF (filled triangles,
n ¼ 4) Data are expressed as a fraction of the turnover rate in the absence of inhibitor.
Trang 8The Km for ATP hydrolysis by SUR1 (0.1 mm) was
similar to that reported for purified CFTR and MRP1
(0.1–3 mm) [37–39], but significantly lower than we
measured for the isolated NBDs of SUR1 We
specu-late that the presence of the transmembrane domains in
SUR1 induces conformational changes in the NBDs, or
in their association, that influences ATP binding The
Km of the purified KATP channel complex (SUR1F–
Kir6.2) was 0.4 ± 0.2 mm [2], which is not significantly
different from that found for the isolated NBDs, but is
somewhat greater than that of SUR1 Thus, it seems
possible that the presence of Kir6.2 within the KATP
complex may further modify interactions between the
NBDs that occur in SUR1 alone Both the NBDs and
the transmembrane domains of SUR1 are known to
interact with the cytosolic and transmembrane domains
of Kir6.2, respectively [23,25,43]
Mutation of the WA lysines (K719A, K1385M)
reduced ATPase activity by 70–80% Mutation of the
equivalent residues in full-length CFTR [44], or the
isolated NBD2 of SUR2A [45] and NBD1 or NBD2
of CFTR [46, 47], also reduces, but does not fully
abolish, ATPase activity Nevertheless, these mutations
completely ablate the ability of MgADP to stimulate
KATP channel activity [5] Thus, WA mutations in
SUR1 may also influence nucleotide binding [48]
and⁄ or the mechanism by which nucleotide
bind-ing⁄ hydrolysis is coupled to channel activity
ATPase activity of the isolated NBDs
As previously reported, MBP-fusion proteins of
isola-ted NBD1 and NBD2 domains hydrolyzed ATP The
Kmfor ATP hydrolysis ( 600 lm) did not vary
signi-ficantly between the isolated NBDs (NBD1, NBD2 or
the NBD1–NBD2 mixture) Previous studies yielded
somewhat lower values of 290 lm and 350 lm for
NBD1 and NBD2, respectively [19] For comparison,
values for the NBDS of SUR2A were 220 lm for
NBD1 [19] and ranged from 370 lm [19] to 4.4 mm
[45] for NBD2 The rates of ATP hydrolysis that we
observed are about two-fold (NBD1) and up to
five-fold (NBD2) higher than those previously reported for
the isolated NBDs of SUR1 [19] It is possible that
these differences reflect differences in the sequence of
isolated domains used in the different studies Mixing
NBD1 and NBD2 did not alter ATPase activity This
is similar to what is found for MRP1 [36], the ABC
protein most closely related to SUR1, but contrasts
with the NBDs of CFTR, where the activity of NBD1
is enhanced by heterodimerization with NBD2 [35,49]
Like other ABC proteins, including MRP1 [38],
SUR1 and both of its isolated NBDs had a broad
nuc-leotide specificity and hydrolyzed GTP as well as ATP There appeared to be a small amount of hydrolysis of MgADP by both NBD1 and NBD2, which contributed less than 10% of the ATP hydrolysis rate It is possible that SUR1 exhibits adenylate kinase activity, as has been suggested for CFTR [50,51] In this case, hydroly-sis of ATP generated from ADP (by adenylate kinase activity) might account for the increase in free phos-phate that we observed
Inhibition of ATPase activity
A decrease in the ATPase activity of SUR1 was observed when the conserved lysine in the WA motif was mutated either in NBD1 or NBD2 Mutation of the WA motif in NBD1 reduced ATPase activity by about 60% If we assume that the relative extent of ATPase activity at NBD1 and NBD2 remains the same in full-length SUR1 (i.e that of NBD1 is
20% of that of NBD2), then the marked inhibition
of ATPase activity of SUR1 suggests that the WA mutation in NBD1 also reduced hydrolysis at NBD2 This might indicate possible interactions between the NBDs The fact that the same mutations did not affect MgADP binding to NBD2 [52] suggests that it
is the hydrolytic capacity that is affected Mutation
of the WA lysine at NBD2 blocked ATPase activity
of SUR1 by about 80% Although this would be con-sistent with inhibition of NBD2 alone, it may also reflect a partial decrease in hydrolysis at both NBD1 and NBD2
The ATPase activity of SUR1 was potently inhibited
by BeF, which traps ABC proteins in a prehydrolytic ATP-bound conformation [33,34] Inhibition by BeF (1 mm) has previously only been reported for the iso-lated NBD2 of SUR2A [20,21]
MgADP also inhibited ATP hydrolysis by isolated NBDs, albeit with very low affinity The lowest value
of Ki (0.6 mm) was found for the NBD1 + NBD2 mixture The inability of ADP to block ATP hydroly-sis by SUR1 is surprising: possible explanations for this finding include a lower ADP affinity for SUR1 or
a higher adenylate kinase activity We presume that this effect is ameliorated in the KATPchannel complex,
as MgADP stimulates channel activity, and reverses channel inhibition by ATP, via interaction with the NBDs of SUR1 [4,5]
Oligomerization of the NBDs Gel filtration indicated that MBP–NBD1, MBP–NBD2 and a 1 : 1 mixture of the two purified as a multimer
of around eight or nine monomers When viewed by
Trang 9EM, the proteins formed ring-like structures with an
outer diameter of 120–140 A˚ This is similar to the
outer diameter of the purified octameric KATPcomplex
(180 A˚) [2], and is consistent with the idea that the
ring-like structures represent eight MBP–NBDs that
coassemble as a tetramer of dimers The inner diameter
of the NBD ring was 40–75 A˚ This space is expected
to be occupied by Kir6.2 in the native KATP channel
complex The widest diameter of cytoplasmic domain
of the related Kir channels Kir3.1 and Kir3.2 was
80 A˚ in the crystal structure [53,54] Thus, the NBDs
are likely to pack somewhat less tightly in the KATP
complex than in the ring-like structures that we
observed for the isolated NBDs
These results suggest that the NBDs may be involved
in physical subunit–subunit associations within the
KATP channel complex, and raise the possibility that
they may also be involved in functional interactions
between subunits Previous studies have also suggested
that NBD1 and NBD2 can physically interact [23,24]
and that purified NBD1 of SUR1 can exist as a tetramer
[26] Interaction of isolated recombinant NBDs to form
functional heterodimers has also been reported for
several other ABC proteins [35,36,55] Such
hetero-dimerization enhanced the ATPase activity of some
ABCC proteins (e.g CFTR) [35,48], attenuated ATPase
activity in others (e.g ABCR [55]), or was without effect
(e.g MRP1 [36]), as we found for SUR1
Implications for channel gating
Unlike those of other ABC proteins, the functional
role of SUR1 is that of a channel regulator, and ATP
hydrolysis by SUR1 plays an important role in the
metabolic regulation of the KATP channel [23] In
electrophysiologic studies, both MgATP and MgADP
stimulate KATP channel activity [3–6] However,
cur-rent evidence suggests that it is the presence of
MgADP at NBD2 that results in KATPchannel
open-ing, and that MgATP must be hydrolyzed to MgADP
in order for channel activation to occur [20]
It is difficult to measure the EC50 for MgATP
acti-vation of wild-type KATP currents in
electrophysiolog-ical studies, due to simultaneous inhibition via the
ATP-binding site of Kir6.2 Coexpression of SUR1
with Kir6.2 carrying mutations in the ATP-binding
site, however, suggests that half-maximal channel
acti-vation is produced by MgATP concentrations of
around 0.1 mm or greater [6] This is in agreement
with the results we report here for SUR1 and those
found previously for the KATPcomplex [2]
Mutation of the WA lysines markedly decreased but
did not completely abolish ATP hydrolysis by SUR1,
in agreement with the electrophysiological data The same mutations shifted the IC50for ATP block of the
KATP channel to a value (13–16 lm) [6]) intermediate between that seen for wild-type channels in the pres-ence ( 30 lm) [6] and abspres-ence (6 lm) [6] of Mg2+ One might expect that a mutation which abolished MgATP binding⁄ hydrolysis would have an IC50similar
to that found in Mg-free solution for wild-type chan-nels The fact that this is not the case suggests that binding⁄ hydrolysis of MgATP is not entirely abolished
by WA mutations Interestingly, the same mutations completely abolished the ability of MgADP to stimu-late KATPchannel currents [5]
Conclusion
SUR1 is unique among ABC proteins in that it serves
as a channel regulator, forming a tightly associated octameric KATPchannel complex in which four Kir6.2 subunits form a central pore surrounded by four SUR1 subunits [2] The fact that the isolated NBDs of SUR1 associate in tetrameric ring-like structures even when Kir6.2 is not present suggests that these domains possess some intrinsic capacity for stable association and that this may contribute to formation of the octa-meric KATP channel complex Here we show that the ATPase activity of SUR1 alone differs from those of both the isolated NBDs and of the octameric KATP channel complex This suggests that the ATPase activ-ity of the NBDs is influenced both by the presence of the transmembrane domains of SUR1 and by the tetrameric Kir6.2 pore Thus, just as SUR1 influences the channel activity of Kir6.2, so Kir6.2 appears to modulate the ATPase activity of SUR This may be considered analogous to the way in which substrates stimulate the activity of other ABC proteins
Experimental procedures
Protein expression and purification
A FLAG-tag was inserted into the extracellular loop between transmembrane helices 11 and 12 of rat SUR1 (GenBank L40624), as previously reported [2] This full-length construct of SUR1 (SUR1F) was expressed in insect cells (Sf9) using a baculovirus expression system (Invitrogen, Paisley, UK), and expression was verified and quantified by [3H]glibenclamide binding [23] Cells were grown and harvested as previously described [2], dis-rupted using a Stansted TC5W homogenizer (Stansted Fluid Power Ltd, Stansted, UK) at a pressure of
10 000 lbÆin)2, and centrifuged at 200 g for 10 min using
a Beckman Allegra 6KR centrifuge with S/N02E3297
Trang 10rotor The supernatant was loaded on a step sucrose
gra-dient (10%⁄ 46%) and centrifuged at 100 000 g for 1 h
using a Beckman L7 centrifuge with SW28 rotor The
intermediate phase was collected and diluted four times
with 50 mm Tris⁄ HCl (pH 8.8) and 200 mm NaCl
Do-decylmaltoside (DDM) (0.5% w⁄ v) was then added, and
membranes were solubilized for 20 min at room
tempera-ture, and then centrifuged at 48 400 g for 20 min using a
Beckman Avanti J-20XP centrifuge with JA-25.50 rotor
Anti-FLAG M2 affinity gel (Sigma, Poole, UK) was
added to the supernatant and incubated for 2 h The
suspension was washed with 20 volumes of 50 mm
Tris⁄ HCl (pH 8.8), 150 mm NaCl and 0.1% DDM
Pro-tein was then eluted with 100 lm 3-FLAG peptide
(Sigma), 50 mm Tris⁄ HCl (pH 8.8), 150 mm NaCl, 0.2%
DDM, 0.05% 1,2-dimyristoyl-sn-glycero-phosphocholine
(DMPC) All procedures were carried out at 4C The
purified protein yield ranged between 50 and 100 lgÆL)1
The identity and purity of SUR1F was confirmed by
MALDI-TOF MS All assays were performed on freshly
prepared SUR1F
Rat NBDs were cloned into the pMAL-c2X vector (New
England Biolabs, Hitchin, UK) to yield MBP-fusion
con-structs, in which MBP is attached to the N-terminal end of
the NBD This strategy was employed because the NBDs
alone are known to be poorly soluble [24] The nucleotide
sequence used for NBD1 was Val608 to Leu1004, and that
used for NBD2 was Lys1319 to Lys1581 Plasmids were
transformed into BL21-CodonPlus Escherichia coli cells
(Stratagene, La Jolla, CA, USA) One liter of Terrific Broth
(Sigma) in baffled flasks was inoculated with 50 mL of
transformed BL21-CodonPlus, grown to a D600 of 1 and
induced with 0.4 mm isopropyl thio-b-d-galactoside Cells
were harvested after 4 h, at 200 g for 20 min They were
re-suspended in 30 mL of buffer A (50 mm Tris⁄ HCl, pH 7.5,
150 mm NaCl, 2 mm dithiothreitol and 1% protease
inhibi-tors; all Sigma) Cells were lysed by two passages through a
Stansted TC5W homogenizer at 12 000 lbÆin)2and kept on
ice throughout Insoluble debris was pelleted by
centrifuga-tion for 30 min at 48 400 g using a Beckman Avanti
J-20XP with JA-25.50 rotor, and the supernatant was
incu-bated by rotation for 1 h at 4C with 2 mL of amylose resin
Unbound protein was eluted by washing with 2· 10 mL of
buffer A, and bound protein was eluted after 15 min of
rota-ting incubation with 4 mL of elution buffer (buffer A with
10 mm maltose and 20% glycerol) The identity of proteins
of expected sizes for NBD1 and NBD2 were confirmed using
antibody to NBD1 [25] and an antibody to MBP (rabbit
polyclonal; New England Biolabs), respectively
Yields were typically 6 mgÆL)1 for NBD1–MBP and
0.8 mgÆL)1 for NBD2–MBP, and comprised > 95% of
total purified protein When not used fresh, purified
pro-teins were stored at) 80 C in 20% glycerol Proteins were
separated on 4–12% gradient Bis⁄ Tris gels and visualized
by Coomassie staining (Invitrogen)
MALDI-TOF MS MALDI-TOF MS (MS and MS⁄ MS) was performed using
a Bruker Ultraflex TOF⁄ TOF mass spectrometer (Bruker Daltonics, Coventry, UK) equipped with a nitrogen laser Gel bands of interest were digested in-gel according to standard procedures [26] using proteomics-grade Trypsin (Sigma-Aldrich) MS was performed using a-cyano-4-hy-droxycinnamic acid as matrix Peptide mass fingerprint spectra were matched against the NCBI nonredundant protein database using the search engine mascot (Matrix Science, London, UK) via an in-house license MS⁄ MS spectra were taken using the LIFT method (Bruker Dal-tonics) The accuracy of MS spectra was typically better than 50 p.p.m.: the accuracy of MS⁄ MS-generated frag-ment ions was in the range of ± 0.2 Da
Gel filtration Purified NBD–MBPs were diluted to 0.5 lgÆlL)1, and
300 lL was analyzed on a Superdex-200 gel filtration column (Tricon, Amersham Biosciences, Little Chalfont, UK) in AT-Pase buffer (50 mm Tris⁄ HCl, pH 7.2, 150 mm NH4Cl,
10 mm MgCl2) Relative protein size was calculated using Bio-Rad Gel Filtration standards (Cat no 151-1901) Peaks were collected and analyzed by EM, and for ATPase activity Purified SUR1F was mixed with DMPC⁄ DDM to final concentrations of 0.05% w⁄ v for DMPC and 0.1% w ⁄ v for DDM It was further concentrated to 1.45 mgÆmL)1, and
250 lL was loaded on a Superdex 200 (10⁄ 30) gel filtration column pre-equilibrated with 50 mL of buffer containing
50 mm Tris (pH 8.8), 150 mm NaCl, 0.05% w⁄ v DMPC and 0.1% w⁄ v DDM
EM and image processing For EM, protein samples were diluted to a concentration
of between 0.05 and 0.1 mgÆmL)1, applied to EM grids coa-ted with carbon film and stained with 2% uranyl acetate Preparations were examined using a CM120 electron micro-scope (FEI, Eindhoven, the Netherlands) with an acceler-ation voltage of 100 kV Electron micrographs were taken
at a magnification of · 45 000 Selected images were digit-ized with a step size of 25 lm on a Nikon Super Coolscan
9000 (Nikon, London, UK) The web and spider software packages [27] were used for all image processing In total,
524 particles were windowed, subjected to reference-free alignment, and sorted into 10 classes using the partitional method (K-means method) of clustering [28]
Nucleotide hydrolysis ATPase activity was normally measured for proteins puri-fied as above, but without the gel filtration step These