Content uploaded by Po-Tsang Huang
Author content
All content in this area was uploaded by Po-Tsang Huang on May 19, 2016
Content may be subject to copyright.
Content uploaded by Po-Tsang Huang
Author content
All content in this area was uploaded by Po-Tsang Huang on May 19, 2016
Content may be subject to copyright.
J Mol Model (2001) 7:20–25
DOI 10.1007/s008940100010
Abstract Rundown is a generally encountered problem
while recording KATP channel activity with inside-out
patches. No assigned structural fragment related to this
mechanism has yet been derived from any of the func-
tional analyses performed. Therefore, based on a com-
bined sequence and secondary structure alignment
against known crystal structure of segments from closely
related proteins, we propose here the three-dimensional
structural model of an intracellular C-terminal domain of
the Kir6.2 subunit in KATP channels. An E. coli CMP-
kinase was suggested as template for the model building.
The subdomain arrangement of this novel kinase domain
and the structural correlation for UDP-docking are de-
scribed. With structural-functional interpretation, we
conclude that the reactivation of KATP channel rundown
by MgATP or UDP is very possibly regulated by this in-
tracellular kinase domain at the C-terminus of Kir6.2
subunit in KATP channels.
Keywords Channel gating · 3D homology modeling ·
Kinase domain · Kir6.2 · Rundown reactivation
Introduction
ATP-sensitive K+-channels (KATP channels or K-ATP
channels) are distributed in a wide variety of tissues,
including brain nerve cells, cardiac and skeletal mus-
cles, and pancreatic β-cells [1, 2, 3]. They play a pivot-
al role in coupling membrane excitation to cellular me-
tabolism. The molecular architecture of KATP channels
is an octameric complex of two structurally unrelated
subunits that assemble with 4: 4 stoichiometry [4, 5, 6].
The ion-pore subunits are members of the inwardly rec-
tifying potassium channel family (Kir), which have two
membrane-spanning domains (M1 and M2) flanking
the K+-selective ion-pore region (H5) and cytoplasmic
amino (N-) and carboxyl (C-) termini [7]. The regulato-
ry subunits are sulfonylurea receptors (SUR), members
of the ATP-binding cassette superfamily which have
two intracellular nucleotide binding folds, NBF1 and
NBF2 [8].
The regulation of KATP channel activity by cytosolic
constituents is extremely complex and is still not fully
understood. In addition to the ATP-dependent channel
closure under physiological conditions, some cytosolic
agents are required to maintain the ability of the channel
to enter the open state, because after formation of an in-
side-out patch the activity of the channel declines with
time. This phenomenon has been described as rundown
for KATP channels [9, 10, 11, 12, 13]. Providing that
channel activity has not completely vanished, it can be
partially restored by brief exposure of the patch to
MgATP or UDP [10, 11, 12, 13, 14, 15, 16, 17]. It has
been proposed that this reactivation might involve pro-
tein phosphorylation on residues other than serine/threo-
nine in Kir6.2 and that the hydrolysis energy of MgATP
seems to be utilized for such reactivation [9, 10, 18, 19,
20]. However, the ability of MgATP to reactivate the
channel activity is variable, declines with time and is in-
effective after complete rundown; it is, therefore, also
suggested that an endogenous kinase responsible for
phosphorylation might exist and is gradually inactivated
or lost from the patch membrane with time [11]. More-
over, trypsin digestion of Kir6.2 leads to the prevention
of rundown, indicating an exposed binding fold responsi-
ble for the deduced conformational change and the run-
down mechanism [11, 21]. On the other hand, UDP can
also reactivate the KATP channel rundown, which is pre-
K.-L. Lou (✉) · H.-C. Chou · Y.-W. Tsai · P.-T. Huang
T.-Y. Chen · Y.-Y. Shiau
Graduate Institute of Oral Biology, College of Medicine,
National Taiwan University, Taipei 10042, Taiwan
e-mail: kllou@ha.mc.ntu.edu.tw
Tel.: +886 2 23562340, Fax: +886 2 23820785
Yu-Shuan Shiau
Institute of Statistical Science, Academia Sinica, Taipei, Taiwan
Robert J. French
Department of Biophysics & Neuroscience Research Groups,
Faculty of Medicine, Health Sciences Centre,
University of Calgary, T2N 4N1, Canada
ORIGINAL PAPER
Kuo-Long Lou · Hsiu-Chuan Chou · Yau-Wei Tsai
Yu-Shuan Shiau · Po-Tsang Huang · Ting-Yu Chen
Yuh-Yuan Shiau · Robert J. French
Involvement of a novel C-terminal kinase domain of Kir6.2
in the K-ATP channel rundown reactivation
Received: 24 October 2000 / Accepted: 1 February 2001 / Published online: 4 April 2001
© Springer-Verlag 2001
21
sented in different kinetic styles between Kir6.1 and
Kir6.2 [13]. Whether a common structural motif or a
similar pathway and/or mechanism is involved in the re-
activation by MgATP and UDP remains controver-
sial [13]. Currently, however, no direct structural evi-
dence is available to indicate the reactivation mechanism
of KATP channel rundown. Therefore, based on a com-
bined sequence and secondary structure alignment
against known crystal structure of segments from closely
related proteins, we propose here a three-dimensional
(3D) structural model of the intracellular part of the
Kir6.2 subunit of KATP channels. A kinase domain was
suggested for the C-terminal residue region. For further
investigation of the molecular catalytic behavior of this
kinase domain, nucleotide-docking simulation was ac-
complished. The residues required for binding contacts
with nucleotides in our structural model are also de-
scribed in this paper.
Methods
Search for templates. The BLAST algorithm was employed to
search in PDB for protein segments whose sequences are similar
to that of mouse Kir6.2 (Genbank accession number S68403) and
whose structures can serve as viable structural templates. The
crystal structures of E coli CMP-kinase (CKE, 6137462) (free and
in complexed forms) [22, 23] were chosen for the determination of
structurally conserved regions (SCRs). The Kir6.2 residues used
for model building are according to their paired sequence (see be-
low) compared to the CKE sequence. In addition, immediately af-
ter the primary sequence comparison with BLAST, residues before
228 were eliminated according to the results.
Paired sequence alignment. The GCG program was used to deter-
mine the equivalent residues. Residue regions of CKE represented
as continuous lines dominantly observed were employed as appro-
priate template regions and the corresponding fragments in Kir6.2
were chosen for alignment. The amino acid sequences of these
Kir6.2 fragments were then included in the multiple sequence
alignment of the appropriate CKE regions to specify the residue
numbers for model building [24].
Model building and docking simulation for nucleotides. Modeling
by homology was performed essentially by following the proce-
dures described by Siezen [24]. Briefly, the residue fragments of
Kir6.2 were chosen according to the results from GCG paired se-
quence alignment. They were then superimposed on to the crystal
coordinates of the Cαatoms of the corresponding SCRs from the
CKE structure. This generated the secondary structure and relative
position of the definite secondary structural elements in the chosen
residue fragments of Kir6.2. Junctions between the secondary
structural elements were individually regularized by energy mini-
mization to give reasonable geometries.
The UDP and CDP molecules were created from small mole-
cule units in a databank via modification with the cvff force field.
Upon docking, the total energy and van der Waals’ contacts be-
tween the complexes of CDP- and UDP-binding to Kir6.2 model
were compared. Distribution of surface charges via electrostatic
potentials was performed by exhibiting the Connolly surfaces for
the binding residues from the kinase domain on Kir6.2.
All the calculations and structure manipulations described
above were performed with the Discover/Insight II molecular sim-
ulation and modeling program (Molecular Simulation, San Diego,
CA, USA; 950 release) on a Silicon Graphics Octane/SSE work-
station.
Results and discussion
Paired sequence and structural alignment
From the BLAST results, the crystal structures of E coli
CMP-kinase (CKE, 6137462) (free and in complexed
forms) [22, 23] were chosen as template protein and for
the determination of structural conserved regions
(SCRs). Sequence comparison and the residue identities
between Kir6.2 and CKE are shown in Fig. 1. Two resi-
due fragments of Kir6.2 were chosen according to the re-
sults from GCG paired sequence alignment, and then
used for structural alignment. These are residues
229–281 and 338–385 (from N- to C-termini of Kir6.2),
for which residues 3–54 and 62–113, respectively, of
CKE are applied to create coordinates (Figs 1 and 2).
This modeled part of the molecule is located on the in-
tracellular C-terminus of Kir6.2.
Overall structural features and comparison
with the template
Structural information from our model suggests a novel ki-
nase domain at the intracellular C-terminus of the Kir6.2
subunit with the subdomain arrangement that is very simi-
lar to those observed in the crystal structure of E. coli
CMP-kinase (CKE) and of other NMP kinases [22, 23].
Fig. 2 shows a comparison of the folding patterns of CKE
and this novel kinase domain of Kir6.2 C-terminus. A
binding cleft for nucleotides can be observed in the upper
middle part of the structure in both molecules (Fig. 2).
Fig. 1 Sequence alignment of Kir6.2 and CKE. The residues of
Kir6.2 (Genbank accession number S68403) identical to those of
template E coli CMP-kinase (CKE, 6137462) are in red, whereas
the conservative substitutions are in green. Two corresponding
residue fragments chosen according to the results of paired se-
quence alignment are compared. The residue identity for each
fragment: (upper part) 33.96% for residues 229 to 281; (lower
part) 27.08% for residues 338 to 385 (rmsd=0.349 and 0.354, re-
spectively). All amino acid residues are represented with single-
letter abbreviation
22
Nucleotide binding
For further investigation of the molecular catalytic behav-
ior of this kinase domain, or, to confirm the role this nov-
el deduced kinase domain may play in Kir6.2, the nucleo-
tide-docking simulation was accomplished. Energy mini-
mization gave a very stable conformation for both UDP
and CDP molecules upon binding to this kinase domain
Fig. 2 Comparison of Cα-tracing between CKE crystal structure
(left) and the structural model of Kir6.2 intracellular kinase do-
main (right). The residue numbers indicated in the kinase domain
structure are to distinguish the connections for the two separate
residue fragments. Both diagrams are viewed with the binding
cleft facing upside. Drawings in all the figures (Fig. 2, Fig. 3,
Fig. 4, Fig. 5) were made with Insight II software package
Fig. 3 3-D structural model of the intracellular kinase domain of
Kir6.2 in docking with the UDP molecule. The kinase domain is
drawn as yellow ribbons and the N- and C-termini are indicated.
The UDP molecule is drawn with CPK spheres in different colors
according to the type of atoms. Viewing is directly towards the
binding cleft
(CDP-form/UDP-form: –625.77/–621.45 (kcal mol–1) for
total energy; 253.42/253.28 for van der Waals’ contact;
rmsd=0.110/0.103). Fig. 3 illustrates the binding of UDP
with our structural model. The UDP molecule is located
and properly oriented in the binding cleft flanked by sev-
eral α-helices of which the residue side-chains protrude
towards UDP to form binding contacts. The details of the
side-chain contacts are shown in Fig. 4.
Because such structural models do not give sufficient
detail to enable us to measure the hydrogen-bonding dis-
tances, we can merely depict here what we have ob-
served from the structure in Fig. 4 regarding the interac-
tion between the corresponding atoms from residue side-
chains of Kir6.2 and from the UDP molecule. These resi-
dues of Kir6.2 are: Arg-365, Arg-367, Lys-373 on one
side of UDP and Glu-241 on the other side. They form
putative hydrogen-bonds between UDP and Kir6.2 in the
appropriate orientation and in a reasonable range of spa-
Fig. 4 Close view of the binding cleft with the UDP molecule.
The Kir6.2 kinase domain is drawn as the yellow ribbon, whereas
UDP molecule is depicted as ball-and-stick models in various col-
ors according to the atom types. Side-chains of the Kir6.2 residues
that might interact or form putative hydrogen bonds with the UDP
molecule are shown as sticks in colors according to the types of
atoms, with residue number indicated
23
tial distance. This might imply and emphasize the utili-
zation of UDP by this kinase domain on Kir6.2.
To describe further the binding ability and the resi-
due contacts, the electrostatic potential surfaces for the
binding pocket on Kir6.2 are depicted in Fig. 5. This
confirms again the appropriate and reasonable binding
contacts between the Kir6.2 cleft and UDP upon dock-
ing.
It is interesting to note that:
1. the two parts of Kir6.2 residues forming H-bonds
with UDP come from the two separate residue frag-
ments mentioned previously in the structural align-
ment section; and
2. they are all charged residues.
Such facts or observations indicate that they are indeed
appropriate to contribute to the construction of the bind-
ing moiety and form H-bonds with nucleotides upon
docking. Such information might, therefore, enable us to
make decisions on selecting reasonable candidates for
site-directed mutagenesis and then further to verify the
role of this kinase domain in the reactivation of KATP
channel rundown for investigation in the future.
Structural–functional interpretation
The most important structural-functional interpretation
provided by this 3-D structural model is with respect to
the reactivation of KATP channel rundown. As has been
described in the introduction, providing the channel ac-
tivity has not completely vanished, it can be partially re-
stored by brief exposure of the patch to MgATP and
UDP [10, 11, 13, 19]. Such reactivation might involve
protein phosphorylation on residues other than ser-
ine/threonine in Kir6.2 and the hydrolysis energy of
MgATP seems to be utilized. In addition, the ability of
MgATP to reactivate the channel activity declines with
time and is ineffective after complete rundown. All this
suggested that an endogenous kinase responsible for sub-
sequent phosphorylation should exist and is gradually in-
activated or lost from the patch membrane with
time [11].
One explanation of these results is that the KATP chan-
nel complex can exist in an active, phosphorylated state
and an inactivated, dephosphorylated state. Dephosphor-
ylation by membrane-associated phosphatases causes the
channel to enter a closed (or, here, inactivated) state
from which it can only exit on phosphorylation by an en-
dogenous kinase. Thus in the absence of MgATP, chan-
nel activity will slowly run down as more channels enter
the dephosphorylated closed state.
We propose that such phosphorylation should occur
via this novel kinase domain at the C-terminus of Kir6.2.
The following statements might clarify and support such
a hypothesis. One should not forget that the rundown of
KATP channels can be also reactivated by exposure of the
patch to UDP. This kinase domain provides an obvious
link between the MgATP- and UDP-reactivation of chan-
nel rundown. The template protein, CMP-kinase, was
proposed to be CMP-specific in prokaryotes and
CMP/UMP-dispensable in eukaryotes [22]. If we look
into the reversible reaction of this kinase, a spectrum of
agents required for channel gating could be observed.
This kinase uses MgATP to produce UDP and MgADP
when UMP is present. Vice versa, when UDP is applied,
MgADP can be used to generate MgATP. This explains
at least in part why MgATP and UDP can both reactivate
the channel rundown, and, as a consequence, describes
the utilization of the hydrolysis energy of MgATP for
channel phosphorylation through an endogenous path-
way. Some aspects still should be clarified, however. For
example:
1. the exact phosphorylation site(s) which can allow
channels to enter the activatable state via MgATP hy-
drolysis and subsequent phosphorylation; and
2. the involvement of other membrane-associated or
non-associated components contributing to phosphor-
ylation/dephosphorylation, e.g. PIP2, Ca2+ and PKC,
instead of this kinase domain alone, in the rundown
and in its reactivation.
The second aspect will bring the emphasis to the mutu-
al interactions among those components, i.e., the sig-
naling pathway and the mechanism evoking the slow-
phase rundown via Kir6.2 [25]. In addition, it is also
very important to identify the structural requirement
that serves as the rest part of the nucleotide-binding
moiety. Regarding this part, the N-terminal residues of
Kir6.2 seem to satisfy such purpose upon considering
the tetrameric assembling structural characteristics of
Fig. 5 Electrostatic potential surfaces for the UDP-binding pocket
of Kir6.2. Red corresponds to an electrostatic potential of
≤–5 kBT/e, white an electrostatic potential 0 kBT/e and blue an
electrostatic potential of ≥+5 kBT/e. The UDP molecule is drawn
as sticks in color according to the types of atoms. The orientation
of molecules is almost the same as observed in Fig. 4.
24
these K-channel family members [26, 27], and, there-
fore, an interaction involving both N- and C-termini of
Kir6.2 with respect to the complexity of gating can be
expected.
Recently, studies by several groups [17, 25] have
successfully demonstrated that both fast-phase (short-
term) and slow-phase (long-term) types of rundown
may exist. The former is induced by interaction with
SUR subunit and can be recovered by the regulation of
MgADP/MgATP [28]. The latter is supposed to result
from either subunit uncoupling [29, 30] or direct de-
phosphorylation on Kir6.2 via, for instance, Ca2+, and
can be reactivated by addition of PIP2and MgATP
through pathways involving kinases [25, 31]. Before
the patch excision or channel rundown, the KATP chan-
nel activity can be maintained by the regulation of
SUR via subunit coupling [25, 29] or the deletion of
26 C-terminal residues [30] in the absence of physio-
logical concentrations of ATP, which binds to the chan-
nels to induce the inhibition. Meanwhile, the phosphor-
ylation on Kir6.2 by endogenous kinases maintains the
channels in the activatable state. Excision of the patch
results in the time-dependent loss of endogenous mem-
brane-associated kinases and the activatable state for
channels cannot be maintained. The channels become
gradually dephosphorylated by phosphatases. Once
K-ATP channels enter the long-term rundown phase,
such dephosphorylated, inactivated state of Kir6.2 can
be reactivated only by those pathways inducing kinase
phosphorylation, for instance, the addition of specif-
ic/non-specific kinases in patch solution, upon utiliza-
tion of MgATP, then back to phosphorylated and activa-
table state [25]. It is, therefore, quite obvious that
MgATP has its own dual effects, either inhibits through
SUR regulation or direct binding on Kir6.2, or reacti-
vates the channels in the rundown events. In addition,
the direct binding of MgATP to Kir6.2 can be prevent-
ed by PKC through electrostatic effect (unpublished da-
ta). However, the existence of a kinase domain at
Kir6.2 C-terminus, in all cases, confirms the possibility
for and provides an alternative for or assists the utiliza-
tion of ATP. Furthermore, our results supplemented the
information required to explain the findings by Noma’s
group [17] in which the long-term rundown phase can
be recovered more efficiently in the presence of UDP.
All this provides indeed a very fresh insight into the
KATP channel rundown phenomenon. However, whether
it is economic for Kir6.2 to include an intrinsic kinase
domain, instead of using another coupling molecule, re-
mains to be clarified.
Conclusion
Our model has for the first time described three-dimen-
sional structural information for subdomain arrangement
of the residues at the intracellular C-terminus of Kir6.2
and the structural correlation for nucleotide-binding, as
well as the functional implication of such a putative ki-
nase domain. Because of the limitation of predictive
structural model, only the structural correlation derived
from the reported functional results related to rundown
reactivation was discussed in this paper. Nevertheless,
such information provides directions for further func-
tional assays and choices of the appropriate candidate for
site-directed mutagenesis before the crystal structure of
Kir6.2 is determined.
Acknowledgements The authors would like to thank Ting-Lin
Chien and Yi-Chun Tsai at Hitron Technology for their technical
advice. We are also very grateful to Drs W. F. van Gunsteren and
Daniela Pietrobon for their advisory discussion and helpful sugges-
tions. This work was supported in part by Research Grants from
NTUMC R&D Committee: 89-N4;CNB 88–03 (88.1.1.-88.12.31.)
and 89D118 (89.4–8), and from National Sciences Council of
Taiwan: NSC 89–2314-B-002–258 (1999.8.1–2000.7.31) and
89–2320-B-002–234 (2000.8.1.-2001.7.31.) for L.K.L.
References
1. Ashcroft, F. M.; Ashcroft, S. J. H. Cell. Signal. 1990,2,
197–214.
2. Ashcroft, F. M.; Gribble, F. M. TINS 1998,21, 288–294.
3. Inagaki, N.; Tsuura, Y.; Namba, N.; Masuda, K.; Gonoi, T.;
Horie, M.; Seino, Y.; Mizuta, M.; Seino, S. J. Biol. Chem.
1995,270, 5691–5694.
4. Inagaki, N.; Gonoi, T.; Clement, J. P., IV; Namba, N.;
Inazawa, J.; Gonzalez, G.; Aguilar-Bryan, L.; Seino, S.; Bry-
an, J. Science 1995,270, 1166–1170.
5. Inagaki, N.; Gonoi, T.; Seino, S. FEBS Lett.1997,409,
232–236.
6. Shyng, S.-L.; Nichols, C. G. J. Gen. Physiol. 1997,110,
655–664.
7. Isomoto, S.; Kondo, C.; Kurachi, Y. Japan. J. Physiol. 1997,
47, 11–39.
8. Aguilar-Bryan, L.; Nichols, C. G.; Wechsler, S. W.; Clement,
J. P., IV; Boyd, A. E., III; González, G.; Herrera-Sosa, H.;
Nguy, K.; Bryan, J.; Nelson, D. A. Science 1995,268,
423–426.
9. Findlay, I. Pflüg. Arch. 1987,410, 313–320.
10. Ohno-Shosaku, T.; Zünkler, B. J.; Trube, G. Pflüg. Arch. 1987,
408, 133–138.
11. Proks, P.; Ashcroft, F. M. Pflüg. Arch., 1993,424, 63–72.
12. Takano, M.; Qin, D.; Noma, A. Am. J. Physiol. 1990,258,
H45–50.
13. Takano, M.; Xie, L.-H.; Otani, H.; Horie, M. J. Physiol. (Lon-
don) 1998,512.1, 395–406.
14. Findlay, I.; Dunne, M. J. Pflüg. Arch. 1986,407, 238–240.
15. Misler, S.; Falke, L. C.; Gillis, K.; McDaniel, M. L. PNAS
USA 1986,83, 7119–7123.
16. Trube, G.; Hescheler, J. Pflüg. Arch. 1984,401, 178–184.
17. Xie, L.-H.; Takano, M.; Kakei, M.; Okamura, M.; Noma, A. J.
Physiol. (London) 1999,514, 655–665.
18. Findlay, I. J. Physiol. (London) 1987,391, 611–631.
19. Furukawa, T.; László, V.; Furukawa, N.; Sawanobori, T.;
Hiraoka, M. J. Physiol. (London) 1994,479.1, 95–107.
20. Kozlowski, R. Z.; Ashcroft, M. L. J. Proc. R. Soc. Lond. [biol]
1990,240, 397–410.
21. Proks, P.; Ashcroft, F. M. J. Physiol. (London) 1993,459,
240P.
22. Briozzo, P.; Golinelli-Pimpaneau, B.; Gilles, A. M.; Gaucher,
J. F.; Burlacu-Miron, S.; Sakamoto, H.; Janin, J.; Barzu, O.
Structure 1998,6, 1517–1527.
23. Bucurenci, N.; Sakamoto, H.; Briozzo, P.; Palibroda, N.;
Serina, L.; Sarfati, R. S.; Labesse, G.; Briand, G.; Danchin, A.;
Barzu, O.; Gilles, A. M. J. Biol. Chem. 1996,271, 2856–2862.
28. Gribble, F. M.; Tucker, S. J.; Ashcroft, F. M. EMBO J. 1997,
16, 1145–1152.
29. Koster, J. C.; Sha, Q.; Nichols, C. G. J. Gen. Physiol. 1999,
114, 203–213.
30. Tucker, S. J.; Gribble, F. M.; Zhao, C.; Trapp, S.; Ashcroft, F.
M. Nature 1997,387, 179–182.
31. Ribalet, B.; Ciani, S.; Eddlestone, G. T. J. Gen. Physiol. 1989,
94, 693-717.
25
24. Siezen, R. J.; Rollema, H. S.; Kuipers, O. P.; de Vos, W. M.
Prot. Engineer. 1995,8, 117–125.
25. Ribalet, B.; John, S. A.; Weiss, J. N. J. Gen. Physiol. 2000,
116, 391–409.
26. Morais-Cabral, J. H.; Lee, A.; Cohen, S. L.; Chait, B. T.; Li,
M.; Mackinnon, R. Cell 1998,95, 649–55.
27. Proks, P.; Gribble, F. M.; Adhikari, R.; Tucker, S. J.; Ashcroft,
F. M. J. Physiol. (London) 1999,514,19–25.
