The modulation of channel activity by direct interaction with membrane lipids is now
an emerging theme in ion channel biology. In particular, phosphoinositides such as
phosphatidylinositol 4,5-bisphosphate (PIP2) are known to regulate the activity of
most major classes of ion channel, as well as a number of other membrane transport
proteins. The regulation of inwardly rectifying (Kir) potassium channels by PIP2 is
one of the most intensively studied and best-defined model systems for studying channel–lipid
interactions. In this review we explore the mechanisms that underlie the complex regulation
of Kir channels by phosphoinositides and other highly charged anionic lipids such
as long chain CoA esters. Intriguingly, these different lipids result in either activation
or inhibition depending on the Kir channel subtype. We describe how recent studies
have provided structural insight into these lipid-induced gating movements, and discuss
how this gating mechanism is central to Kir channel physiology.
Introduction
As pointed out in the preceding perspective by Lundbæk (see p. 421), many lipids and
lipophilic compounds alter ion channel function by affecting the collective physical
properties of the membrane, i.e., the regulation does not depend on specific interactions
with the channel. However, the regulation of potassium channels by highly negatively
charged lipids such as phosphoinositides and long chain acyl-CoA esters is mechanistically
distinct because these lipids bind directly to the channel and any effects mediated
by changes in the physical properties of the bilayer itself are considered to be minimal.
The direct interaction of highly charged anionic lipids with ion channels is now emerging
as a fundamental mechanism of channel regulation. A considerable body of work over
the past decade has shown that members of almost every class of ion channel can be
regulated by phosphoinositides, in particular phosphatidylinositol-4,5 bisphosphate
(hereafter referred to as PIP2) (Hilgemann et al., 2001; Suh and Hille, 2005; Gamper
and Shapiro, 2007a). However, the molecular mechanisms underlying this form of direct
lipid regulation are best understood in the family of inwardly rectifying potassium
(Kir) channels and so will form the focus of this perspective.
Regulation of Kir Channels by Highly Charged Anionic Lipids
In 1996/1997 three papers were published that provided the first evidence that highly
negatively charged anionic lipids can modulate the activity of Kir channels. Hilgemann
and Ball (1996) demonstrated that phosphoinositides such as PIP2 activate native cardiac
KATP channels. In addition, Fan and Makielski (1997) showed that several cloned Kir
channels, (e.g., Kir1.1, Kir2.1, and KATP [Kir6.2/SUR1] channels) are activated by
PIP2. Furthermore, Larsson et al. (1996) demonstrated activation of native KATP channels
by long chain acyl-CoA esters (LC-CoA), which are metabolites of fatty acid metabolism.
It is now clear that all Kir channels are regulated by phosphoinositides (Logothetis
et al., 2007a) and long chain CoA esters (Rapedius et al., 2005), and that the structure
of the highly negatively charged head group present in these lipids is central to
their ability to affect Kir channel activity. The anionic head group is thought to
interact directly with a distinct positively charged binding site on the channel (Logothetis
et al., 2007a) and can induce quite rapid gating motions (Rapedius et al., 2007a).
Thus, the anionic head group appears to gate Kir channels very much like a classical
ligand, with the only distinction that the ligand is tethered to the membrane. Interestingly,
the functional consequences of these head group interactions are markedly different
between members of the Kir channel family. For example, whereas PIP2 activates all
members of the Kir channels, LC-CoA esters inhibit their activity, except for KATP
channels, which are activated by LC-CoA (Rapedius et al., 2005).
The main issues that will be discussed in this perspective center around the following
questions. (a) Where and how do anionic lipids interact with Kir channels? (b) What
is the mechanistic basis for the different effects of PIP2 and LC-CoA on Kir channels?
(c) What are the conformational changes induced by the binding of these lipids? (d)
What is the physiological relevance of this regulation?
A Variety of Highly Charged Anionic Lipids Activate KATP Channels
The effects of various lipids on cloned Kir6.2/SUR1 (KATP) channels have been evaluated
by their application to the cytoplasmic membrane side in excised patches. From these
studies it became clear that neutral lipids (e.g., phosphatidylethanolamine, phosphatidylcholine)
or weakly negatively charged lipids (e.g., phosphatidylinositol) have little effect
on channel activity (Fan and Makielski, 1997; Krauter et al., 2001), whereas lipids
with highly negatively charged headgroups strongly activated KATP channels. These
were the different phosphoinositides including PI(4)P, PI(4,5)P2, PI(3,4)P2, and PI(3,4,5)P3,
which all restored channel activity when applied after rundown, a process that typically
occurs in excised patches due to the breakdown of endogenous phosphoinositides (Hilgemann
and Ball, 1996; Fan and Makielski, 1997; Baukrowitz et al., 1998; Shyng and Nichols,
1998; Rohacs et al., 2003). However, it is PIP2 that is most important as it represents
the major phosphoinositide in eukaryotic membranes comprising ∼1% of the total phospholipids
and is >100-fold more abundant than the other phosphoinositides such as PI(3,4)P2
and PIP(3,4,5)P3 (Gamper and Shapiro, 2007a).
LC-CoA esters (e.g., oleoyl-CoA, palmityl-CoA) also potently activate KATP channels
at low concentration (<1 μM) (Larsson et al., 1996; Gribble et al., 1998; Liu et al.,
2001; Schulze et al., 2003b; Branstrom et al., 2004) (Fig. 1 C). Here the negative
charge is contained in the CoA headgroup (structure see Fig. 1) and the 3′-ribose
phosphate is particularly important as its removal abolishes any stimulatory effect
on KATP channels (Rapedius et al., 2005). Furthermore, a negatively charged artificial
lipid (DOGS-NTA) also strongly activates KATP channels (Krauter et al., 2001), indicating
that the specific structure of the charged head group is not critical. However, other
Kir channels exhibit much greater specificity (see below).
Figure 1.
Proposed mechanism of phosphoinositide and LC-CoA gating in Kir channels. (A) A homology
model of Kir6.2 in the closed state showing the location of the four bound PIP2 molecules
at the membrane/cytoplasm interface (position of PIP2 was determined by a computational
docking procedure (Haider et al., 2007). The negative charges on the headgroups (oxygen
atoms in red) interact with a positively charged cluster in the cytoplasmic domains,
for clarity the multiple residues that comprise the binding site are not shown, for
details see Haider et al. (2007). (B) Previous models suggested that opening of the
Kir channel is produced by electrostatic attraction between PIP2 and its binding site
that induces a major movement of the cytoplasmic domain toward the membrane. (C) In
addition to being activated by PIP2, PI(3,4)P2, and PI(3,4,5)P3 (not depicted), KATP
channels are activated by oleoyl-CoA, a long-chain acyl-CoA ester as measured in inside-out
patches from Xenopus oocytes. BSA is used to rapidly remove oleoyl-CoA. (D) Kir1.1
is activated by PI(4,5)P2 but inhibited by oleoyl-CoA and PI(3,4)P2 and inhibition
could be reversed by PIP2. (E) In contrast to the original model (B) we propose that
all phosphoinositides and LC-CoAs can bind to the PIP2 binding site on the closed
Kir subunit. However, only certain bound lipids (e.g., PIP2) are subsequently able
to initiate the allosteric changes required for channel opening. Those that cannot
induce channel opening (LC-CoA and PI(3,4)P2) inhibit channel activity by competitive
displacement of PIP2. The chemical structures of PIP2 and LC-CoA are shown at the
top of the figure. The critical phosphate on LC-CoA required for its ability to induce
opening in KATP channels and inhibition in, e.g., Kir1.1 channels is highlighted in
blue. Panel A is adapted from the results of Haider et al. (2007). C and D are adapted
from Rapedius et al. (2005).
Screening of the negative charge in these lipids by divalent cations (e.g., Mg2+)
or their complexation by polycations such as polylysine or neomycin abolishes their
stimulatory effect, suggesting that electrostatic interactions between the negatively
charged headgroup and the channel are critical (Fan and Makielski, 1997; Schulze et
al., 2003a,b). However, application of the head group itself (i.e., IP3 or CoA) does
not activate KATP channels (Shyng and Nichols, 1998; Schulze et al., 2003b), demonstrating
that lipid tethering of the charged head group to the membrane surface is required.
A Conserved Binding Site for Highly Charged Anionic Lipids in Kir Channels
Systematic mutagenesis of basic residues in the cytoplasmic domains of Kir channels
has identified several potential PIP2-interacting residues that are highly conserved
among the different Kir channels (Logothetis et al., 2007a). Neutralization of these
residues results in channels with a low open probability that increases upon application
of exogenous PIP2. Furthermore, these mutant channels show increased sensitivity to
inhibition by polylysine and neomycin, as well as faster inhibition by PIP2-specific
antibodies, suggesting that these mutations reduce the apparent PIP2 affinity (Huang
et al., 1998; Lopes et al., 2002; Schulze et al., 2003a). When mapped onto structural
models of Kir channels these basic residues cluster in a region close to the membrane
interface suggestive of a binding site (for details see Haider et al., 2007; Logothetis
et al., 2007a). Each of the four subunits of these tetrameric channels likely contains
one binding site. A recent study on Kir2.1 channels suggest that binding of PIP2 to
one of the four sites is sufficient to promote channel opening, and PIP2 interaction
with the remaining sites further increases the open probability in a cooperative fashion
(Xie et al., 2008).
The structure of the PIP2 binding site is therefore beginning to emerge. However,
extensive mutagenesis studies have shown that a number of other residues located at
positions quite distant from this site can affect the apparent PIP2 affinity (Logothetis
et al., 2007a). Some of these residues are uncharged and either alter the structure
of the PIP2 binding site allosterically, or interfere with the transduction mechanism
that couples PIP2 binding and channel opening. The latter effect would also change
the apparent PIP2 affinity as measured with macroscopic currents that are unable to
distinguish between changes in ligand affinity versus changes in ligand efficacy.
Interestingly, mutation of the conserved PIP2-interacting residues within the PIP2
binding site also reduce the apparent affinity for LC-CoA in KATP channels (Fox et
al., 2003; Schulze et al., 2003b). In agreement, biochemical experiments on the C-terminal
domains of Kir channels show that LC-CoA can competitively antagonize the binding
of PIP2 (Rapedius et al., 2005). In addition, the functional effects of PIP2 and LC-CoA
on KATP channels are similar since both lipids strongly antagonize ATP inhibition,
as well as inhibition by sulphonylureas (e.g., glibenclamide) and activation by channel
openers (e.g., diazoxide) (Schulze et al., 2003b). These pharmacological compounds
affect KATP (Kir6.2/SUR) channels via interaction with the SUR subunit. These results
suggest that PIP2 and LC-CoA probably regulate KATP activation via the same mechanism
and the same interaction site.
Bidirectional Regulation of Kir Channels by Phosphoinositides and LC-CoA Esters
As stated above, all tested highly charged anionic lipids activate KATP channels with
similar potency. However, the KATP channel is the exception, as all other Kir channels
show a bidirectional regulation by these different lipids, i.e., while PIP2 activates
all Kir channel subtypes, LC-CoA as well as some phosphoinositides have the opposite
effect and inhibit Kir channel activity (Rohacs et al., 1999, 2003; Rapedius et al.,
2005; Shumilina et al., 2006). To determine the apparent affinity (i.e., potency to
activate) of a particular phosphoinositide Logothetis and coworkers established an
assay using short chain fatty acids phosphoinositides such diC8(octanoly)-PIP2 (Rohacs
et al., 1999, 2003). These short chain phosphoinositides are soluble in aqueous solution
as monomers and rapidly establish a water–membrane partitioning equilibrium allowing
different diC8-phosphoionositides to be compared at a similar membrane concentration.
This assay revealed that PIP2, PI(3,4)P2 and PI(3,4,5)P3 activate KATP channels with
a similar potency, whereas all other Kir channels exhibit varying degrees of phosphoinositide
specificity. For example, PI(3,4,5)P3 produces little activation, and PI(3,4)P2 no
activation, of Kir1.1 and Kir2.1 channels when applied to closed channels (i.e., after
rundown) (Rohacs et al., 1999, 2003). Intriguingly, PI(3,4)P2 when applied to active
channels (i.e., before rundown) produced marked inhibition of Kir1.1 and Kir2.1 that
could be reversed by PI(4,5)P2 (Rapedius et al., 2005) (Fig. 1 D). Even more strikingly,
LC-CoA esters such as oleoyl-CoA potently inhibit all Kir channels (except KATP channels)
in a way that can be reversed by PIP2 (Rapedius et al., 2005) (Fig. 1 D). Furthermore,
removal of the 3′-ribose phosphate in LC-CoA that prevented activation in KATP channels
also abolished inhibition in Kir1.1 and Kir2.1 (Rapedius et al., 2005), indicating
that this phosphate group is critical for Kir channel interaction.
How Can the Same Lipids Act as Both Activators and Inhibitors of Different Kir Channels?
As described above, LC-CoA appears to bind to the same conserved phosphoinositide
binding site that is found in all Kir channels. In a previous study we proposed a
mechanism whereby LC-CoA competes with PIP2 for the same binding site, but that LC-CoA
binding does not activate the channel, thereby inhibiting channel activity by displacement
of PIP2 (Rapedius et al., 2005). This hypothesis has important structural implications
for the mechanism of PIP2 activation in Kir channels as it suggests that phosphoinositides
and LC-CoA bind to the closed state of the channel. This idea is different to some
previously held views about PIP2 activation (as depicted in Fig. 1 B), which suggested
that opening of the channel is achieved by a movement of the positively charged cytoplasmic
domains toward the membrane that is driven by electrostatic attractions between the
phosphoinositide binding site and membrane bound PIP2 (Fan and Makielski, 1997; Enkvetchakul
and Nichols, 2003). If this were the case then the simple electrostatic interaction
between the binding site and PI(3,4)P2 or LC-CoA should also be able to stabilize
the open state in all Kir channels.
These findings argue for a relatively nonspecific electrostatic interaction between
all phosphoinositides (or LC-CoA) and the binding site on the channel and suggest
that specificity is achieved (at least in part) in more subtle structural changes
subsequent to binding. In this model the binding of PIP2 is capable of inducing pore
opening, but the binding of either PI(3,4)P2 or LC-CoA is not (as depicted in Fig.
1 E). Indeed, point mutations in Kir2.1 have been found that allow LC-CoA to activate
the channel probably by permitting those specific interactions in the transduction
mechanism required to induce channel opening (Rohacs et al., 2003).
To understand this in structural terms will ultimately require a high resolution structure
of a eukaryotic Kir channel with different phosphoinositides (or LC-CoA) bound, a
goal that may not be such a dream given the recent high-resolution structures of KirBac/Kir3.1
and Kv2.1/Kv2.1 chimeras produced by the MacKinnon laboratory that resolve some bound
lipids and detergents, although not yet PIP2 (Long et al., 2007; Nishida et al., 2007).
However, recently, computational ligand docking studies have been used to define the
location of PIP2 and the PIP2 binding site in a homology model of Kir6.2 (Haider et
al., 2007). This site appears to satisfy most of the available functional data and
also allows the fatty acids to reside in the membrane (Fig. 1 A). This latter point
is important because the Kir channel homology models are all in the closed state and
therefore suggest that it is not necessary for the phosphoinositide binding site to
translocate significantly toward the membrane before binding can occur. This is therefore
consistent with the proposed hypothesis in which PI(3,4)P2 and LC-CoA can bind to
the closed state of the channel. Furthermore, even though some of the highly conserved
positive charges found within the PIP2 binding site are missing in bacterial Kir channels,
these channels exhibit inhibition by PIP2, demonstrating that the KirBac closed state
can bind PIP2 (Enkvetchakul et al., 2005).
Kir channels also differ markedly in their apparent PIP2 affinity as determined by
their sensitivity to inhibition by various PIP2 complexing agents (e.g., neomycin,
polylysine, PIP2 antibodies) (Huang et al., 1998; Rohacs et al., 1999, 2003). For
example, Kir4.1 and Kir2.1 show very low sensitivity to neomycin (i.e., they have
a high PIP2 affinity), Kir1.1, Kir2.2, and KATP (Kir6.2/SUR) show an intermediate
sensitivity (i.e., intermediate PIP2 affinity), whereas the G protein–regulated Kir3.x
channels and heteromeric Kir4.1/Kir5.1 channels display a high sensitivity (i.e.,
they have a lower PIP2 affinity) (Rapedius et al., 2007a,b). The molecular basis of
these differences is currently unknown and might not be determined solely by differences
in PIP2 binding affinity, but also by the equilibrium between the open and closed
state that is shifted toward the open state by PIP2 binding (as implied by the term
apparent PIP2 affinity).
From the Kinetics of PIP2 Gating to the Structural Mechanism of PIP2 Activation
To gain insight into the mechanism of PIP2 gating we recently established an assay
that allowed measurement of PIP2 gating kinetics using fast application of either
diC8-PIP2 or neomycin to inside-out membrane patches (Rapedius et al., 2007a). These
measurements revealed that PIP2 rapidly binds to closed Kir channels (τ < 300 ms)
and then induces a much slower conformational change leading to the open state of
the channel (Fig. 1 E). We found that the speed of this conformational change differs
dramatically between Kir channels and that this depends on the presence (or absence)
of a single lysine residue at the base of the first transmembrane domain (TM1). Kir
channels possessing a lysine at this TM1 position (e.g., Kir1.1, Kir4.1) display slow
PIP2 activation, and mutation of this residue (e.g., Kir1.1[K80M], Kir4.1[K67M]) dramatically
accelerates PIP2 activation (∼50-fold) (Fig. 2 D). In agreement with the role of this
lysine residue, Kir channels lacking a lysine at this position (e.g., Kir2.1) display
fast PIP2 activation kinetics that are dramatically slowed upon mutation of this site
to a lysine (e.g., Kir2.1[M84K]). However, this positively charged residue does not
interact directly with PIP2. Instead, homology modeling and molecular dynamics (MD)
simulations of a Kir1.1 closed state structure suggest that the -NH3
+ group of this TM1 lysine forms a hydrogen bond with the backbone carbonyl group
of A177 in TM2 at the helix bundle crossing (Rapedius et al., 2007a) (Fig. 2, A and
B).
Figure 2.
Gating model of Kir1.1. (A) Side view of a closed state homology model of Kir1.1 with
one of the four subunits highlighted. The highlighted residues at the helix bundle
crossing (K80 and A177) are proposed to form an H bond, which stabilizes the closed
state. In this and all other panels TM1 is shown in green and TM2 in yellow with the
slide helix in red. (B) An expanded view shows the proposed H bonding interaction
between residues in the -NH3
+ group of K80 and the backbone carbonyl group of A177. (C) A bottom up view through
the pore of a closed state (top) and open state (bottom) model of Kir1.1 (for details
see Rapedius et al., 2007a) illustrating a major movement of TM1 and TM1 induced by
PIP2 binding that would rupture the TM1–TM2 H bond. The highlighted residues are K80
and A177. (D) Rapid application of 3 μM diC8-PIP2 on Kir1.1-WT and Kir1.1-K80V channels
subsequent to rundown measured in inside-out patches. When H bonding at the helix
bundle crossing is disrupted in Kir1.1-K80V channels, PIP2 activation is much faster
than in wild-type channels. A–D are modified from Rapedius et al. (2007a).
A recent model of the open state KirBac3.1 suggests a large rotation and tilt of the
TM1 and TM2 helices during channel opening (Domene et al., 2005; Kuo et al., 2005)
that would rupture this TM1–TM2 H-bond (as depicted in Fig. 2 C), thereby explaining
why this H bond determines the time course of channel opening induced by PIP2 as its
rupture would contribute to the activation energy of the closed to open state transition.
In support of this model, a study on cysteine accessibility in Kir2.1 channels also
indicates conformational changes of the pore at the helix bundle crossing during PIP2
gating (Xiao et al., 2003). Taken together these results suggest that PIP2 gating
in Kir channels involves a major conformational change at the helix bundle crossing.
However, the mechanism by which PIP2 triggers this conformational change remains unknown.
There is an intriguing parallel between PIP2 gating and the mechanism of pH inhibition
seen in Kir channels. Kir channels with H bonding at the helix bundle crossing (e.g
Kir1.1 and Kir4.1) display high sensitivity to inhibition by low intracellular pH
whereas Kir channels lacking this H bonding are markedly less sensitive (Rapedius
et al., 2007a). Moreover, the time course of recovery from pH inhibition corresponds
strikingly well to their pH sensitivity and H bonding ability as it is fast in channels
lacking H bonding and slow in channels with H bonding. This suggests that although
the initial events in these gating processes are different (i.e., PIP2 binding versus
protonation of a currently unknown H+ sensor), the structural changes at the helix
bundle crossing underlying both pH gating and PIP2 gating appear to be conserved.
However, it is unlikely that pH inhibition simply results from promoting PIP2 to unbind
because the rate of PIP2 activation is much slower than the recovery rate from H+
inhibition. Therefore, if H+ inhibition were to cause the dissociation of PIP2, then
the rebinding of PIP2 (i.e., the PIP2 activation rate) should govern the time course
of recovery from pH inhibition, which is clearly not the case (Rapedius et al., 2007a).
Nevertheless, there is a general trend that Kir channels with a lower PIP2 affinity
show a higher pH sensitivity, indicating a possible relationship between the PIP2–channel
interaction and pH sensitivity (Leung et al., 2000; Schulze et al., 2003a; Du et al.,
2004), but the mechanistic basis of this remains elusive.
Physiological Implications of PIP2 and LC-CoA Regulation
PIP2 can be regarded as an indispensable cofactor for all eukaryotic Kir channels
because no channel activity is seen in its absence. Not surprisingly therefore, inherited
mutations that reduce PIP2 affinity result in a variety of disease states such as
Type II Bartter's Syndrome (Kir1.1), Andersen's Syndrome (Kir2.1) (Lopes et al., 2002),
and congenital hyperinsulinaemia (Kir6.2) (Lin et al., 2006). This therefore raises
the question as to the biological benefit of having PIP2 as a cofactor. One hypothesis
put forward by Hilgemann suggested that this PIP2 dependence might be a mechanism
to keep Kir channels inactive in intracellular membrane compartments (where PIP2 levels
are thought to be low) because active Kir channels might be harmful to the ion homeostasis
of these compartments (Hilgemann, 2007b). Furthermore, the PIP2 dependence may have
allowed the evolution of mechanisms that control Kir channel activity by modulation
of the interaction with PIP2 (for reviews see Logothetis et al., 2007b; Xie et al.,
2007) For example, G protein activation of Kir3.x channels requires PIP2 and is thought
to result from increasing the channel PIP2 affinity from a level that is otherwise
too low for the endogenous PIP2 levels to support activity (Huang et al., 1998; Zhang
et al., 1999). By contrast, PIP2 strongly antagonizes ATP inhibition of KATP channels,
and vice versa (Baukrowitz et al., 1998; Shyng and Nichols, 1998). Thus, ATP inhibition
can be regarded as a reduction in the apparent PIP2 affinity. The molecular mechanisms
behind this are still controversial and could result either (or both) from the direct
displacement of PIP2 by ATP (MacGregor et al., 2002; Haider et al., 2007) (the binding
sites show some overlap), or an allosteric mechanism by which ATP stabilizes a conformation
that prevents PIP2 binding (Enkvetchakul and Nichols, 2003). Furthermore, the regulation
of Kir1.1 and Kir3.x by PKA and PKC also probably involves changes in PIP2 affinity
(Liou et al., 1999; Zeng et al., 2003; Logothetis et al., 2007b; Xie et al., 2007).
Therefore, understanding the molecular mechanisms that underlie PIP2 gating may well
represent the key to understanding Kir channel gating in general.
The issue of whether Kir channels are regulated by cellular phosphoinositide metabolism
in vivo, and the physiological relevance of such regulation, is still controversial
(Gamper and Shapiro, 2007a) due to the high PIP2 affinity of many Kir channels (e.g.,
Kir1.1, Kir2.1, Kir4.1). For example, stimulation of the principal enzyme involved
in PIP2 hydrolysis (PLC), can be used as a mechanism to deplete cellular PIP2 levels.
However, PLC-mediated inhibition of Kir2.1 channels (which have high PIP2 affinity)
can only be obtained when their PIP2 affinity is reduced by site-directed mutagenesis
(Hardie et al., 2004). By contrast, PLC-mediated inhibition of Kir3.x channels (low
intrinsic PIP2 affinity) has been reported in native tissues (Cho et al., 2005) and
possibly contributes to channel desensitization upon cardiac vagus stimulation by
muscarinic receptors (Kobrinsky et al., 2000).
It is also still unclear as to whether such regulation occurs by a global depletion
of plasma membrane PIP2, or whether this occurs in microdomains (Gamper and Shapiro,
2007b; Hilgemann, 2007a; Xie et al., 2007). The latter mechanism seems more appealing
because global PIP2 depletion should have many potential side effects given that PIP2
is involved in so many other cellular mechanisms including calcium signaling, endo-/exocytosis,
organization of the cytoskeleton, and the regulation of many other ion channels and
transporters. A physiological role for Kir channel regulation by other phosphoinositides
such as PI(3,4)P2 or PI(3,4,5)P3 is rather unlikely because even when their production
is stimulated (e.g., by growth hormones via PI-3 kinases) their global concentration
is still very low compared with PIP2 (Gamper and Shapiro, 2007a). Nevertheless, these
different phosphoinositides have been mechanistically insightful in dissecting the
structural basis of phosphoinositide specificity and the mechanism of PIP2 activation.
LC-CoA esters are obligate intermediates of fatty acid metabolism that are formed
in the cytoplasm and then shuttled into the mitochondria to fuel β-oxidation. In excised
membrane patches, Kir channels can be markedly activated (or inhibited) by oleoyl-CoA
at concentration in range of 100 to 1000 nM (Liu et al., 2001; Branstrom et al., 2004;
Rapedius et al., 2005). However, the concentration of LC-CoA in the cytoplasm is strongly
buffered by specific acyl-CoA binding proteins (Knudsen et al., 2002) and calculations
suggest that the concentration of free LC-CoA is <10 nM under physiological conditions
and should not exceed 200 nM even in pathophysiological conditions (Knudsen et al.,
2002). This raises the question as to whether cytoplasmic levels of LC-CoA can rise
sufficiently to affect Kir channels in vivo. However, studies have shown that extracellular
application of oleic acids to cells can produce KATP channel activation and Kir2.1
inhibition via the cytoplasmic formation of LC-CoA under conditions (e.g., high glucose)
that promote LC-CoA accumulation (Branstrom et al., 2004; Shumilina et al., 2006).
Such activation of KATP channels in pancreatic β cells could contribute to a reduction
in insulin secretion in type II diabetes patients (Riedel et al., 2003; Branstrom
et al., 2004), and the activation of cardiac KATP during ischemia (which is known
to promote LC-CoA accumulation) could represent a potential cardio-protective mechanism
(Liu et al., 2001). Furthermore, LC-CoA activation of KATP channels has been suggested
to be involved in the sensing of circulating fatty acids by hypothalamic neurons (Lam
et al., 2005).
A physiological relevance for the inhibitory effect of LC-CoA seen in most other Kir
channels has so far yet to be established. However, a direct link between Kir channel
activity and fatty acid metabolism could be of potential relevance to many tissues
given the widespread expression of different Kir channels throughout the body and
the general occurrence of LC-CoA in cells. Interestingly, a recent study demonstrated
that the sodium–calcium exchanger, which is also regulated by PIP2, displays similar
regulation by LC-CoA (Riedel et al., 2006). It will therefore be interesting to see
whether other PIP2-regulated channels or transporters also exhibit regulation by LC-CoA.
In conclusion, it is now clear that PIP2 serves as an obligate cofactor for Kir channel
activity and it is difficult to examine any form of Kir channel regulation without
considering the role of PIP2. Work over the past decade has provided significant insight
into the physiological aspects of phosphoinositide-mediated regulation and its structural
mechanisms. It will also be interesting to see whether the lessons we have learned
from Kir channels about, e.g., the regulation of PIP2 affinity, phosphoinositide (or
LC-CoA) specificity and gating efficacy, as well as the kinetics of PIP2 gating can
provide an insight into the molecular mechanisms by which so many diverse classes
of other ion channels and transporters are regulated by phosphoinositides.