(Received for publication, January 13, 1997, and in revised form, February 14, 1997)
From the Department of Physiology and Department of
Neuroscience, School of Medicine, The Johns Hopkins University,
Baltimore, Maryland 21205
Cloned auxiliary -subunits (e.g.
Kv
1) modulate the kinetic properties of the pore-forming
-subunits of a subset of Shaker-like potassium channels.
Coexpression of the
-subunit and Kv
2, however, induces little
change in channel properties. Since more than one
-subunit has been
found in individual K+ channel complexes and expression
patterns of different
-subunits overlap in vivo, it is
important to test the possible physical and/or functional
interaction(s) between different
-subunits. In this report, we show
that both Kv
2 and Kv
1 recognize the same region on the
pore-forming
-subunits of the Kv1 Shaker-like potassium channels. In
the absence of
-subunits the Kv
2 polypeptide interacts with
additional
-subunit(s) to form either a homomultimer with Kv
2 or
a heteromultimer with Kv
1. When coexpressing
-subunits and Kv
1
in the presence of Kv
2, we find that Kv
2 is capable of inhibiting
the Kv
1-mediated inactivation. Using deletion analysis, we have
localized the minimal interaction region that is sufficient for Kv
2
to associate with both
-subunits and Kv
1. This mapped minimal
interaction region is necessary and sufficient for inhibiting the
Kv
1-mediated inactivation, consistent with the notion that the
inhibitory activity of Kv
2 results from the coassembly of Kv
2
with compatible
-subunits and possibly with Kv
1. Together, these
results provide biochemical evidence that Kv
2 may profoundly alter
the inactivation activity of another
-subunit by either differential
subunit assembly or by competing for binding sites on
-subunits,
which indicates that Kv
2 is capable of serving as an important
determinant in regulating the kinetic properties of K+
currents.
The heterogeneity of voltage-sensitive potassium currents present
in excitable and nonexcitable cells is essential for diverse biological
functions (1-4). In addition to the large number of genes encoding the
channel subunits and posttranslational modulations of channel protein,
the diversity of potassium channels is further enhanced by the
mix-and-match assembly of different subunits (5, 6). Within the large
family of Shaker-like potassium channels, the selective subunit
assembly includes heteromultimer formation of distinct pore-forming
-subunits and/or assembly of different kinds of subunits such as the
-subunits and hydrophilic cytoplasmic
-subunit(s) (7, 8).
Together, these give rise to the vast heterogeneity of K+
currents. Changes in expression of a given subunit may alter the
composition of heteromultimers in vivo, which would allow a
cell to tune its K+ current system(s) during development
and in response to changes in the cellular environment.
There are more than 60 cloned genes encoding functional Shaker-like
-subunits, which have been divided into several subfamilies. Among
them, subunits in the Kv1 to Kv5 subfamily are capable of functional
homomeric channels in heterologous systems, such as Xenopus
oocytes (9-13). Four
-subunits within a given subfamily can form a
functional channel either as a homotetramer or a heterotetramer (14,
15). In the case of auxiliary subunits in animals, four genes encoding
-subunits for Shaker-like potassium channels have been well
characterized: Kv
1, Kv
2, Kv
3 (which has now been suggested to
be a splice variant of Kv
1), and Hk (7, 8, 16-20). These
-subunits share at least 85% amino acid sequence identity in their
COOH-terminal core regions, but differ significantly in length and
sequence of the remaining NH2-terminal regions. Despite the
remarkable sequence similarity among different
-subunits, their
functional effects are quite different. For example, coexpression of
-subunits with certain
-subunits in Xenopus oocytes
induces pronounced alterations in channel kinetic properties, most
noticeably acceleration of fast inactivation by either Kv
1 or Kv
3
(8, 17, 18-20). Kv
2, on the other hand, binds to
-subunits, such as Kv1.2 (or RCK5) (21). However, it has little effect on inactivation of
-subunits such as Kv1.2 (RCK5) (7, 8, 22, 23). Recent data have
shown that Kv
2 is capable of increasing the surface expression of
certain K+ channels in transfected cells (24).
Biochemical evidence has indicated that there are more than one
-subunit present in each K+ channel complex (21). Given
that cloned
-subunits have different modulatory effects on
-subunits, it would be interesting to test whether different
-subunits can interact with each other, which could be an important
mechanism to increase the diversity of potassium currents. To test this
hypothesis, we have used the yeast two-hybrid system to study the
interaction specificity of Kv
2 with various
- and
-subunits.
The functional consequences of heteromeric
-
and
-
interactions were evaluated by electrophysiological analyses.
Plasmid vector
construction was performed according to standard recombinant DNA
techniques (26). The vectors that express partial cDNA fragments
were constructed by a high fidelity polymerase chain reaction cloning
strategy according to the procedures described by Li et al.
(27). The oligonucleotides used are listed in Table I.
In yeast, the expression of different fusion proteins of -subunits, Kv
1, and Kv
2 was carried out by inserting the corresponding cDNA fragments into the SmaI/NotI,
SalI/NotI, or BglII/NotI
sites of pPC97 and pPC86 vectors (28-31). Construction of tagged
Kv
1, Kv
2, and Kv
2 mutants was carried out by fusing the coding
fragment with a peptide which represents a heart muscle kinase
recognition sequence and the 12CA5 monoclonal epitope (PYDVPDYASL), at
the end of the coding sequences before the stop codon (28). Transient expression and immunodetection of potassium channel subunits were performed according to our published protocol (28, 51).
|
The procedures were
performed according to our published protocol (28, 29) using HF7c yeast
strain (MAT ura3-52 his-200 ade 2-101 lys2-801
trp1-901 leu2-3, 112 gal4-542 gal80-538
LYS2::GAL1UAS-GAL1TATA-HIS3 URA3::GAL4
17mer(x3)-CyclTATA-lacZ)
as host cells (25).
Whole-cell voltage clamp
recordings were carried out according to the published protocol (28,
33). The liquid junction potential was calculated to be 7.2 mV using
JPCalc software (34) and corrected from the holding potential.
Typically, the cell was held at 87 mV, and the holding voltage was
then jumped from this potential up to a test potential of +53 mV in
10-mV increments for 300 ms. Current data were filtered at 1 kHz,
digitized at 100-µs intervals. Data analysis was done using clampfit
software (pCLAMP6, Axon Instrument, Foster City, CA).
A standard formula to compare two
proportions was used to determine the statistical significance of
different pairs of sample sets (35). We have used = 64 ms as the
cutoff to separate populations with or without fast inactivation. In
this analysis, the one-sided z test statistics were
calculated using the following formula z =
1-
2 /{
p(1-
p)[1/n1 + (1/n2)]}0.5, where
1 and
2 are sample proportions that
showed fast inactivation of each given group,
p is the
weighted average of the sample proportions, and
n1 and n2 are sample
sizes.
The interaction between -subunits and Kv
1 has been
studied in more detail. In particular, the
-
complex is
assembled, at least in part, by the association of the conserved core
regions in Kv
1 with NABKv1 of
-subunits, a critical
assembly motif located in the hydrophilic NH2-terminal
domains (28, 29, 36). The formation of an
-
complex presumably
recruits the inactivation particle of Kv
1 close to the
"receptor" site, thereby either accelerating the rate of
inactivation of
-subunits of Kv1.4 (RCK4) and ShB (or H4) or
inducing inactivation of compatible
-subunits which lack intrinsic
fast inactivation, such as ShB
(6-46) (28, 36). Because the
formation of the
-Kv
1 complexes is subfamily-specific, i.e. Kv
1 binds only to the NH2-terminal
domains of Kv1
-subunits (28), this allows Kv
1 to selectively
modulate a subset of
-subunits.
To investigate whether Kv2 alters the electrophysiological
properties of
-subunits in transfected mammalian cells, we have constructed two plasmids that express either Kv
1 or Kv
2 with the
12CA5 monoclonal antibody tag fused at the COOH terminus of each coding
sequence (see "Materials and Methods"). Both plasmids use the
Kv
1 5
-untranslated sequence. Thus, the two expression vectors are
identical except for the amino acid coding sequence. Experiments
utilizing these constructs permit better comparison of Kv
1 and
Kv
2 expression and their ability to modulate
-subunits. By
transient transfection in COS cells, we functionally expressed ShB
(6-46), a mutated ShB potassium channel that lacks the
inactivation gate (37, 38). The Kv
2 effects on this
-subunit were
studied by whole-cell voltage clamp recording. Fig.
1A shows a series of traces obtained by
stepping up from a holding potential of
87 mV to a final test
potential of +33 mV in 20-mV increments. The recorded cells were
transfected with ShB
(6-46) alone (upper panel),
ShB
(6-46) in the presence of either Kv
1 (middle
panel), or Kv
2 (bottom panel). In contrast to the
ShB
(6-46) + Kv
1 cotransfection in which we observed the
Kv
1-mediated inactivation (Fig. 1A, middle panel),
expression of ShB
(6-46) in the presence of Kv
2 resulted in no
detectable changes of the fast inactivation properties (Fig. 1A,
bottom panel). When traces were averaged within the group of
ShB
(6-46) + Kv
2 (n = 12) or ShB
(6-46) alone
(n = 17), we observed little variations of inactivation
properties between these two groups of recorded cells (Fig.
1B). To examine the protein expression of Kv
2 in the
experiments, total cell lysates from the transfected cells were
separated by SDS-polyacrylamide gel electrophoresis. The expression of
Kv
1 and Kv
2 was detected by immunoblot using the 12CA5 monoclonal
antibody (mAb12CA5). Indeed, Kv
2 was found to express in the
transfected COS cells and exhibited higher expression as compared with
that of Kv
1 (Fig. 1C, lanes 1 and 2). The
higher expression of Kv
2 has been reproducible in multiple
transfection experiments.1 The strong
mAb12CA5 binding signal indicates that the failure of Kv
2 to
modulate N-type fast inactivation was not due to the lower protein
expression of Kv
2. Thus, Kv
2 differs from Kv
1 and by itself
fails to induce the fast inactivation of the ShB
-subunit.
The Subfamily-specific Association of Kv
Biochemical characterization supports direct
physical interaction between Kv2 and Kv1.2 (RCK5) (7, 21) as well as
Kv1.4 (RCK4) (39). However, the existing results for Kv
2 binding specificity and region(s) involved are not conclusive (Ref. 23, also
see "Discussion"). In addition, there is no information on whether
Kv
2 binds to ShB. To test the association between Kv
2 and Kv1
-subunits, we expressed various regions of
-subunits and Kv
2
in yeast and used the yeast two-hybrid system to study the potential
interaction(s) (30, 31). In the case of Kv
1, it has been found that
the NH2-terminal domains of the Kv1
-subunits are
involved in the
-Kv
1 interaction (28, 36). Because Kv
1 and
Kv
2 share considerable sequence homology, we first tested the
potential interaction of Kv
2 with the cytoplasmic regions of the
Kv1.4, an
-subunit which has been found to interact with Kv
2
(39). The truncated cytoplasmic fragments, i.e. the
NH2-terminal domain (aa2
1-306) and COOH-terminal domain (aa 566-651), were expressed individually with Kv
2 as GAL4 fusion proteins. If Kv
2 interacts with one or both truncated Kv1.4 fragments, the resultant
interaction(s) should confer the ability of the yeast transformants to
grow on synthetic medium lacking histidine. Fig.
2A shows that when Kv
2 was expressed alone
either as a fusion protein of the GAL4 DNA binding domain (GAL4-DB) or
that of the GAL4 transcription activation domain (GAL4-TA), the yeast
transformants grew on double selection medium supplemented with
histidine, indicating that they carry both plasmids (Fig.
2A, numbers 1 and 2, middle left
panel). When the same number of transformants were tested to grow
on the triple selection medium lacking histidine, they showed no growth
(Fig. 2A, numbers 1 and 2, lower left panel).
This indicates that Kv
2 itself does not exert any endogenous
activity that permits the yeast transformants to grow on the selection
medium. By contrast, the coexpression of Kv
2 and the
NH2-terminal domain (Fig. 2A, number 3), not the
COOH-terminal domain (Fig. 2A, number 4) of Kv1.4, resulted
in growth on the selection medium lacking histidine. Consistent results
have been obtained using a
-galactosidase assay (data not shown).
Thus, similar to Kv
1, Kv
2 interacts with the
NH2-terminal domain of the Kv1.4
-subunit.
The ability of Kv2 to interact with the NH2-terminal
domain of Kv1.4 suggests that the resultant association may be
essential for Kv
2 to interact with
-subunits, as seen in
biochemical copurification. In the case of Kv
1, its
subfamily-specific association with the NH2-terminal
domains of Kv1
-subunits has been shown to be essential for the
Kv
1-mediated inactivation (28, 36). Coimmunoprecipitation of
K+ channel polypeptides in rat brain has indicated that
Kv
2 interacts with Kv1.2 and Kv1.4, but not Kv2.1 (24, 39). To
further test the specificity of the Kv
2, pairwise combinations of
Kv
2 and the NH2-terminal domains of eight different
-subunits were analyzed with the yeast two-hybrid system. The eight
-subunits included were: Shaker B (40-42), Shabll, Shaw2, and Shal2
from Drosophila (9); Kv1.4 (or RCK4) (43), Kv2.1 (or DRK1)
(44), Kv3.1 (or NGK2b) (45), and Kv4.2 (or rShal1) (46, 47) from rat. These genes belong to the four major subfamilies, one fly gene and one
rat gene for each subfamily. Among the selected
NH2-terminal domains, Kv
2 interacts only with the
NH2-terminal domains of ShB and Kv1.4 (Fig.
3B, numbers 1 and 5), both of
which belong to the Kv1 subfamily. Furthermore, the Kv
2 interacting
site was mapped to aa 174-306 within the NH2-terminal
domain of Kv1.4 (data not shown), which coincides precisely with the
domain that interacts with Kv
1 (28, 36). Thus, both Kv
1 and
Kv
2 interact subfamily-specifically with the Kv1
-subunits and
share the same binding site on the
-subunits.
Formation of Homo- and Heteromultimeric
Based on the hydrodynamic estimates, the
-dendrotoxin acceptor (or Kv1.2) complex contains more than one
Kv
2 subunit per complex (21). It is not clear, however, whether
Kv
2 can form an oligomeric complex in the absence of
-subunits.
Additionally, since expression patterns of Kv
1 and Kv
2 overlap
in vivo (8, 22, 39, 52), it would be interesting to examine
whether different
-subunits can interact with each other to form
heteromultimers. Fig. 3 shows that Kv
2 can indeed associate to form
homomultimers as the yeast transformants grow in the selection medium
lacking histidine (Fig. 3, number 3). The known potassium
channels in yeast have distinctive topology and belong to a subclass
different from the Shaker-like potassium channels (48). Therefore, the above result supports that Kv
2 is capable of interacting with itself
in the absence of
-subunits.
Both Kv1 and Kv
2 are expressed in rat brain with overlapping
expression patterns (8, 22, 39). Because Kv
2 forms multimers in the
absence of
-subunits (Fig. 3, number 3) and has
considerable overall sequence homology (73%) to Kv
1, we tested the
possible interaction between Kv
1 and Kv
2 and found that Kv
1
and Kv
2 can also interact (Fig. 3, number 4). This
implies that Kv
1 and Kv
2 can form heteromultimers in the absence
of the pore-forming
-subunits.
Both
Kv2 and Kv
1 interact with the Kv1
-subunits by recognizing the
same region in the Kv1
-subunits (Fig. 2 and Ref. 28). Additionally,
Kv
2 interacts with itself and/or Kv
1 to form homo- and/or
heteromultimers (Fig. 3). Because Kv
1, not Kv
2, induces the fast
inactivation of the Kv1
-subunits that lack fast inactivation (Fig.
1), these data suggest that one potential function of Kv
2 would be
to alter the efficacy of the Kv
1-mediated inactivation. One
predicted outcome would be that Kv
2 weakens the ability of Kv
1 to
inactivate, as Kv
2 may compete with Kv
1 for the binding site on
-subunits and/or associate with Kv
1 to form Kv
1-Kv
2
heteromultimers containing fewer inactivation particles.
One experiment to test this hypothesis would be to coexpress Kv1 and
a compatible
-subunit in the presence or absence of Kv
2 and ask
whether Kv
2 alters the ability of Kv
1 to inactivate. We
cotransfected COS cells with noninactivating ShB
(6-46) and Kv
1
in a 1:6 plasmid ratio of
/Kv
1. Fig. 4A
shows three representative traces, one from each group, that were
superimposed and normalized. These traces were recorded by stepping up
the holding potential from
87 mV to a test potential of +13 mV for a
duration of 300 ms. ShB
(6-46) alone produced a trace with fast
activating kinetics lacking N-type fast inactivation. When Kv
1 was
included in the transfection, we observed a majority of transfected
cells that show the Kv
1-mediated fast inactivation (Fig.
4A). If, however, both Kv
1 and Kv
2 were included in a
plasmid ratio of
/Kv
1/Kv
2 of 1:6:5, much fewer transfected
cells showed fast interaction induced by Kv
1 (see below and Fig.
4B).
The lack of fast inactivation by Kv1 in the presence of Kv
2 in
the example shown in Fig. 4A could have resulted from
variable transfection rates of the three plasmids. To address this, we recorded 17 Shaker-positive cells for the ShB
(6-46) transfection, 40 cells for the ShB
(6-46) + Kv
1 transfection, 44 cells for the
ShB
(6-46) + Kv
1 + Kv
2 transfection. Since recorded cells were
selected solely based on the presence of Shaker current and the
expression of CD4 antigen that was cotransfected in all
experiments, the percentage of recorded cells in a given transfection
shown, fast and/or slow inactivation can then be determined. Among the recorded traces, some exhibited both fast and slow inactivation, the
others showed only slow inactivation. The traces were fit by a double
exponential function, and the resultant inactivation constants were
plotted against the cell number in percentage (Fig. 4B).
Fig. 4B shows plots using one inactivation constant per
recorded cell, i.e. if the inactivation consists of two
(fast and slow) components, only the fast inactivation constant was
plotted. For the ShB
(6-46) transfection, we observed that all
recorded cells lacked the fast inactivation and gave a slow
inactivation constant (
2) larger than 128 ms (Fig.
4B, top panel). In the 40 recorded cells for ShB
(6-46) + Kv
1 (Fig. 4B, middle panel), we observed that more than
60% of recorded cells possessed a fast inactivation component
(
1 = 1-64 ms), while about 35% of cells showed no fast inactivation, which presumably is due to low or no expression of
Kv
1. By contrast, we observed that only 17% of cells showed fast
inactivation when Kv
2 was coexpressed (Fig. 4B, bottom
panel). As the amount of plasmid input for both ShB
(6-46) and
Kv
1 was identical in both transfections, it is unlikely that the
population of the recorded cells, which showed no fast inactivation, is
due to the lack of Kv
1 plasmids. Furthermore, statistical analysis shows that the difference was statistically significant (see legend to
Fig. 4B and "Materials and Methods").
Despite the constant plasmid input of Kv1 in both experiments, the
inhibition of the Kv
1-mediated inactivation by Kv
2 can be
subjected to several interpretations. For example, it is not known
whether the presence of Kv
2 decreases the expression of other
subunits at the protein level and/or at the level of channel surface
expression, thereby resulting in the inhibition of the Kv
1-mediated
inactivation. To examine whether coexpression of
-subunits
differentially altered the channel surface expression, we plotted the
current amplitude of ShB
(6-46) in the presence of Kv
1 and Kv
1 + Kv
2. The result indicates that the averaged current amplitudes
were similar: 6.5 nA for the ShB
(6-46) + Kv
1 transfection and
6.7 nA for the ShB
(6-46) + Kv
1 + Kv
2 transfection (p > 0.2, Student's t test). Furthermore,
there was no obvious correlation between current amplitude and
inactivation properties (data not shown). To examine the expression
level of Kv
1 in the presence or absence of Kv
2, aliquots of the
transfected cells in the experiment (Fig. 4A) were
collected, and the total cell lysates were prepared and separated by
SDS-polyacrylamide gel electrophoresis. Indeed, the Kv
1 expression
was found to be comparable, regardless of the presence of Kv
2 (Fig.
4C, lanes 2 and 3). Together, these results
suggest that when both Kv
1 and Kv
2 are expressed in the cells,
Kv
2 is capable of inhibiting the Kv
1-mediated inactivation.
Because the Kv
1 and ShB
(6-46) expression remained relatively
constant, both in the presence and absence of Kv
2, the inhibition by
Kv
2 is likely to be caused by differential subunit assembly,
i.e. Kv
2 competes with Kv
1 for the binding site on
-subunits and/or Kv
1 and Kv
2 form heteromultimeric complex(es).
If the
differential subunit assembly indeed plays a role in inhibiting the
Kv1-mediated inactivation, then the binding of Kv
2 to
-subunits and/or Kv
1 would be essential for the Kv
2-mediated inhibition.
Deletion analysis for Kv1 has shown that the conserved core region
of Kv
1, i.e. amino acids 73-401, is sufficient for the Kv
1-
interaction (28). Among the 47 positions that harbor different residues between Kv
1 and Kv
2, 27 are conservative changes. Additionally, both Kv
1 and Kv
2 recognize the same subset of
-subunits, i.e. the Kv1
-subunits. These suggest
that the corresponding region in Kv
2 may serve a similar function in
binding to the
-subunits. Indeed, when a truncated fragment
representing the conserved core region of Kv
2 was subjected to the
yeast two-hybrid test, the core region is sufficient to interact with
the NH2-terminal domain of ShB (NShB) (Fig.
5).
One possibility for inhibiting the Kv1-mediated inactivation is that
Kv
2 can somehow interact with the inactivation particle of Kv
1,
which is located at the NH2 terminus (8). This interaction may lead to the defective activity of Kv
1 to inactivate the Kv1
-subunits. Taking this into consideration, we chose to first carry
out deletion analysis using Kv
1 to map the region involved in the
Kv
1-Kv
2 interaction to test whether the inactivation particle is
necessary for the heteromultimeric
-
interaction. A total of 11 Kv
1 deletion mutants were constructed (Fig. 5A). Their
ability to interact with Kv
2 was tested by the yeast two-hybrid system using growth selection. The minimal region in Kv
1 required for interacting with Kv
2 was mapped to the conserved core region, which indicates that the NH2-terminal 72 residues of Kv
1
are not required for the Kv
1-Kv
2 interaction in the yeast
two-hybrid test. Similar to Kv
1, the corresponding core region of
Kv
2 is sufficient for interacting with Kv
2 (Fig. 5B).
Thus, the conserved core region of
-subunits is involved in both
-
and
-
interactions.
The results in Fig. 5 suggest that the formation of
heteromultimeric -Kv
2 and/or Kv
1-Kv
2 complexes is
responsible for Kv
2 to inhibit the Kv
1-mediated inactivation. If
this is true, one prediction is that the truncated Kv
2 polypeptide
containing the mapped interacting region, i.e. the conserved
core region, should be capable of inhibiting the Kv
1-mediated
inactivation. To test this, we constructed two Kv
2 deletion mutants:
Kv
2(39-367) and
Kv
2(39-316) (Fig.
6A). The
Kv
2(39-367) mutant contains the intact interacting region mapped by the yeast two-hybrid analysis (Fig. 5), while 51 residues COOH-terminal to the interacting region of
Kv
2 were truncated in
Kv
2(39-316). The resultant mutant can
no longer interact with either Kv
1 or NH2-terminal
domains of the Kv1
-subunits (Fig. 5). Because the ability to
interact should correlate with the activity in inhibiting the
Kv
1-mediated inactivation, we carried out experiments similar to
those in Fig. 4, which involved cotransfecting ShB
(6-46) and Kv
1
in the presence of either
Kv
2(39-367) or
Kv
2(39-316), and
examined the subsequent inactivation properties. Fig. 6B
shows that the protein levels of Kv
2 and the two deletion mutants
are comparable. Among the 39 recorded Shaker-positive cells transfected
in the presence of
Kv
2(39-367), we found only 22.5% of cells
that showed fast inactivation (Fig. 6, C and D, panel
b). Indeed,
Kv
2(39-367) acts similarly to Kv
2 by
decreasing the number of cells that exhibit the fast inactivation (Fig.
6, C and D, panels a and b). The
difference between Kv
1 alone and Kv
1 +
Kv
2(39-367) was statistically significant (p < 0.001 and see legend to
Fig. 6D). In contrast, among the 49 recorded cells that were
transfected in the presence of
Kv
2(39-316), 61% of cells showed
fast inactivation. The distribution of inactivation constant from this
group of cells is similar to that obtained from cells transfected by
ShB
(6-46) + Kv
1 (Fig. 6D, panels c and d).
Thus, the ability of the mutated Kv
2 to interact with ShB
(6-46)
and/or Kv
1 directly correlates with their ability to inhibit the
Kv
1-mediated inactivation.
Using transient expression of different combinations of
-subunits and
-subunits, we have observed a functional role of
Kv
2 in modulating channel inactivation. Since the activity of Kv
2 to inhibit the Kv
1-mediated inactivation directly correlates with
the ability of Kv
2 to associate with
-subunits and Kv
1, differential subunit assembly is a likely mechanism responsible for the
Kv
2 activity. This subunit interaction may be important for tuning
the K+ channel diversity in vivo.
Consistent with the biochemical data from copurification of Kv1.2 and
Kv2 (7, 22), our results (Fig. 2) show that Kv
2 interacts with
the NH2-terminal domains of
-subunits in a
subfamily-specific manner. Curiously, a recent report suggested that
both Kv
1 and Kv
2 interact with Kv1 and Kv4
-subunits (23).
This is inconsistent with results from biochemical binding and
electrophysiological analysis (28, 36). Because Kv
2 plays a role in
both channel expression (24) and channel properties (this report),
important future experiments would be to investigate the in
vivo specificity of
-
interaction.
The formation of heteromultimeric complexes as a mechanism to increase
K+ current diversity has been studied in several channel
systems. This includes the formation of functional channels by
different pore-forming subunits (e.g. nicotinic
acetylcholine receptor, voltage-gated K+ channels, etc.)
and by assembly of pore-forming subunits with various auxiliary
subunits (e.g. voltage-gated sodium or calcium channels)
(see review by Catterall (49)). The heteromultimeric oligomerization by
auxiliary subunits in the absence of pore-forming subunits has not been
reported. The biochemical evidence of two homologous auxiliary subunits
"competing" with same set of -subunits and the formation of
heteromultimeric auxiliary subunits in the absence of pore-forming
subunits suggests yet another potential mechanism to create and tune
their electrical diversity. The ability to form homo- or
heteromultimers of
-subunits in the absence of
-subunits does not
necessarily imply the
-
and
-
subunit assembly that happens
in separate steps in vivo. Future experiments are needed to
determine the physiological existence of heteromeric complexes of
-subunits and molecular cascade for assembling
-
heteromultimers.
The results reported here demonstrate that Kv2 is an active player
in determining the fast inactivation mediated by other
-subunits.
How does Kv
2 inhibit the Kv
1-mediated inactivation? Our data can
be best explained by the following mechanism (Fig. 7A). In the absence of Kv
2, the expression
of Kv
1 will permit it to coassemble with compatible
-subunits.
Depending upon whether the interacting
-subunits contain an
inactivation gate, Kv
1 either accelerates or induces fast
inactivation (Fig. 7A, I). In the presence of the high
concentration of Kv
2, Kv
2 occupies most of the sites on
-subunits as homomultimeric Kv
2 complexes. As a result, it
prevents (or removes) Kv
1 from interacting with
-subunits (Fig.
7A, II), thereby inhibiting the Kv
1-mediated inactivation. The coexpression experiments in the present study show
that Kv
2 inhibited the Kv
1-mediated inactivation, consistent with
the results of immunoblot analyses in which we found that the
expression of Kv
2, despite its comparable plasmid input, was
considerably higher than that of Kv
1 (Figs. 1C,
4C, and 6B). Thus, under the conditions of our
experiments, most
-subunits presumably were occupied by
homomultimeric Kv
2 complexes, which prevents the limited numbers of
Kv
1 subunits to bind, thereby eliminating the Kv
1-mediated
inactivation without altering the protein expression of Kv
1 (Fig.
4D).
Conceivably, one should observe an intermediate situation where Kv1
and Kv
2 are present in a comparable concentration (or different
concentrations with an appropriate ratio for heteromultimeric interaction, since the Kv
1 and Kv
2 may have considerable
difference in affinity for subunit interaction). Under such conditions,
most compatible
-subunits should be interacting with
heteromultimeric
-complexes to produce intermediate effects,
i.e. Kv
2 weakens but does not remove the Kv
1-mediated
inactivation. In an attempt to test this, we performed experiments with
lower inputs of Kv
2 plasmid. Thus far, we have not identified an
input plasmid ratio that gives rise to currents with a prominent
intermediate inactivation constant. A different approach to alter
protein expression level for
-subunits might be necessary to test
this hypothesis. The failure to identify intermediate inactivation
could also be interpreted by other mechanism(s). For example, compared
with Kv
1, Kv
2 has considerably higher affinity for
-subunits
and/or itself. In this case, the higher affinity and/or avidity for
-Kv
2 and/or Kv
2-Kv
2 interactions would yield predominantly
-Kv
2 complex(es).
In mammalian systems, two genes encoding three forms of mammalian
-subunits, i.e. Kv
1, Kv
3 (a putative splice variant
of Kv
1), and Kv
2, have been found expressed in brain with an
overlapping expression pattern (7, 8, 39). Although Kv
3 has not been demonstrated directly to bind either
-subunits or
-subunits, the
fact that the Kv
3 amino acid sequence within the interacting region
is identical to that of Kv
1 suggests that Kv
3 is likely to share
features for subunit interaction found in Kv
1 and Kv
2 (17, 18,
20). Based on the results of studying subunit interaction between
-subunits and
-subunits, an interesting new regulatory pathway is
emerging. All cloned
-subunits consist of a conserved core region
critical for
-
and
-
interaction and variable NH2-terminal domains that determine the modulatory activity
(Fig. 7B). Their distinct modulatory effects on
-subunits
and ability to form heteromultimers have revealed yet another potential
in vivo mechanism for creating and tuning the diversity of
potassium currents. Although our present study probes the formation of
heteromultimers by examining the protein-protein interaction and the
resultant alterations of inactivation properties, there is increasing
evidence suggesting that the functional roles of
-subunits may not
be restricted to modulation of inactivation (16, 24, 50). Thus, future
experiments should be focused on addressing whether the heteromultimeric assembly of
-subunits is indeed present in
vivo and whether these heteromultimeric
-
combinations could
specify a wide spectrum of modulatory activities, which may include,
but are not limited to, tuning inactivation properties and surface expression of
-subunits in vivo.
We thank Magdalena Bezanilla for the
Kv
2(39-316) construct; Dr. Weifeng Yu for help on patch clamp
techniques; Dr. Lily Jan for ShB and rshal1 (Kv4.2); Dr. R. Joho and
Dr. A Brown for DRK1; Dr. D. Mckinnon for RCK1; Dr. L. Salkoff for
fshal2, fshaw2, and fshabl1; and Dr. R. Aldrich for the ShB
(6-46)
mutant. We also thank Magdalena Bezanilla, Gerda Breitwieser, Xiaodong
Li, Rajini Rao, and Weifeng Yu for critical reading of this
manuscript.