3DStructure of the Kv4'2KChIP2 Complex Using Electron Microscopy L' Kim1, J' Furst2, N' Grigorieff2,

1
3D-Structure of the Kv4.2-KChIP2 Complex Using
Electron MicroscopyL. Kim1, J. Furst2, N.
Grigorieff2, and S. Goldstein11 Boyer Center for
Molecular Medicine, Department of Cellular and
Molecular Physiology, Yale University School of
Medicine, New Haven, CT 06511, USA 2 HHMI and
Department of Biochemistry, Brandeis University,
Waltham, MA 02454, USA
ABSTRACT In the brain and heart rapidly
inactivating (A-type) voltage-gated potassium
currents (Kv) regulate the excitability of
neurons and cardiac myocytes. The Kv4, or
Shal-related, family of quickly inactivating
potassium channels have been found in mammalian
heart and brain and is thought to mediate A-type
current activity. In addition, Kv4 potassium
channels interact with a family of ß-subunits
known as the Kv channel-interacting proteins
(KChIPs). KChIPs are small (200 residue)
cytoplasmic peptides found to modulate the
inactivation kinetics, recovery from
inactivation, and surface expression of Kv4
subunits. Co-expression of Kv4 and KChIPs in
heterologous expression systems have been shown
to recapitulate the native behavior of A-type
currents found in vivo. Here we have engineered
a charybdotoxin sensitive variant of human Kv4.2
tagged with the 1D4 epitope. We have further
developed a system to quantify and produce
significant quantities of Kv4.2 with KChIP2 in
the COS-7 expression system. Using this
expression system we have purified Kv4.2 and
KChIP2 by immunoaffinity purification with 1D4
antibodies. Purified Kv4.2-KChIP2 complexes were
imaged using transmission electron microscopy at
60,000 x magnification. Structures were
determined by single particle analysis revealing
a structure similar to the previously resolved
Shaker potassium channel structure. However, the
cytoplasmic domain containing the T1 region
appears larger in our structure suggesting a
location for binding of KChIP2 on the channel.
Figure 8 10 SDS PAGE analysis of protein
purification Purified elutions of the
Kv4.2-KChIP2 complex were examined using SDS
PAGE. Eluted fractions E1-E4 were silver stained
to examine purity. Western gels show lysate S1
(before the addition of 1D4 beads), S2 (after the
addition of 1D4 beads), elutions E3 and E4.
Immunoblots were stained with anti-1D4 and
donated anti-KChIP2 antibodies
Figure 4 Kv4.2 mutants are sensitive to
charybdotoxin These are representative raw traces
of Kv4.2 block with application of 10 nM
charybdotoxin. Note the increased block with
respect to the Kv4.2-GV mutant with the addition
of an aspartic acid in the Kv4.2-GVD mutant. 50
block is achieved at 10 nM for Kv4.2-GV and 1
nM for Kv4.2-GVD
Figure 5 Kv4.2-KChIP2 expression can be assayed
using a radioactively labeled CTX. In order to
purify the Kv4.2-KChIP2 complex it was necessary
to determine the level of expression in the COS-7
cell expression system. A radioactively labeled
charybdotoxin was produced and used to assay and
count the number of channels expressed using a
plasmid construct containing an ADV promotor and
standard transfection techniques using
Lipofectamine 2000.
Figure 1 Predicted topology of KChIP2 and
Kv4.2 KChIP2 is a 252 amino acid cytoplasmic
protein with four distinct EF hand like motifs.
KChIPs have about 40 amino acide similarity to
the recoverin/NCS subfamily of calcium-binding
protein. KChIPs modulate the density,
inactivation kinetics, and rate of recovery from
inactivation of Kv4 channels in heterologous
expression systems. KChIP is thought to bind the
Kv4 channels at the N-terminus. The
stoichiometry of KChIPs to pore forming subunits
is currently unknown. Kv4.2 is a 630 residue
voltage gated potassium ion channel. Kv4.2 ?
subunits consist of six transmembrane domains and
the canonical pore forming P loop. Kv4.2
generates a rapidly activating and inactivating
K current. Four alpha subunits form an intact
potassium channel.
Figure 9 Purified Kv4.2-KChIP2 complex are
negatively stained and imaged Purified protein
complexes are placed onto carbon coated copper
grids washed and stained with 1 uranyl acetate.
Images were taken with a Philips CM12 electron
microscope at acceleration voltages of 120 keV at
a magnification of 60000x with 1.5 mm defocus.
Figure 6 Kv4.2 expression compared against
Kv4.2-KChIP2 As expected from the
electrophysiology, Kv4.2 expression is increased
when co-expressed with Kv4.2. Expression was
examined using the pMT3 plasmid. Different
amounts and ratios of DNA were examined. We
further examined expression under different
storage conditions (frozen/not frozen) and
different times for expression. The non-frozen
condition with 2.55 mg of DNA yielded about 5
pmol per 100 mm plate after 48 hours.
Figure 11 Preliminary structure refined with
FREALIGN Using FREALIGN, images were corrected
for the contrast transfer function (CTF) error
introduced by the electron microscope. FREALIGN
uses an automatic weighting scheme based up on
the phase residual of the image against the
reference structure to further refine the
structure. Particle images above a certain phase
residual are discarded. The first reference
structure was the IMAGIC 3-D structure. After 75
iterations of FREALIGN we arrived at a
self-consistent structure, with a tentative
resolution of 2.5 nm. However, further work must
be accomplished to obtain more particles to
further validate this preliminary structure. In
the structure above, we believe the smaller
domain (in blue) represents the transmembrane
domain of Kv4.2, based upon its predicted size.
Also, the cytoplasmic domain may be separated
into three parts the N-terminus (in red), the
C-terminus (in yellow), and the KChIP2 (in
magenta). More experiments must be performed to
verify the location of each domain.

• Figure 2
• KChIP2 co-expression modulates the function of
  Kv4.2
• Raw traces of Kv4.2 current with and without
  KChIP2. Voltage pulses were applied from -80 mV
  to 60 mV in increments of 10 mV from a holding
  potential of -80 mV. Here cells were bathed in
  (mM) 91 NaCl, 4 KCl, 0.3 CaCl2, 0.7 MgCl2 5
  HEPES, pH 7.5. Currents were recorded 3 days
  post-injection of Xenopus Laevis oocytes with
  Kv4.2 and KChIP2 RNA as appropriate.
• Representative traces of Kv4.2 and Kv4.2-KChIP2
  currents at 60 mV normalized to peak current.
• KChIP2 increases the rate of inactivation
  approximate two-fold in oocytes.
• Peak current from -80 to 60 mV relationship with
  and without KChIP2, reveals increased current
  density with co-expression of KChIP2. (n 4-5)
• Relative current recovery from inactivation for
  Kv4.2 and Kv4.2-KChIP2.

Figure 7 Kv4.2-KChIP2 purification scheme Using
the charybdotoxin sensitive Kv4.2 mutant, a 1D4
immunoaffinity tag was attached to the C-terminus
of Kv4.2. This mutant was co-expressed with
KChIP2. Kv4.2-KChIP2 complexes were expressed in
COS-7 cells and solubilized in 2.5 CHAPS for 1
hour at 4 C in the presence of 0.5 mg/ml of
E.Coli lipid, 1 mM DTT, 0.2 mM leupeptin/pepstatin
, 1 mM PMSF, 2 mM EDTA, 40 mM HEPES-KOH (pH 7.4),
80 KCl, and 100 mM NaCal. Insoluble material was
pelleted at 100,000 x g for 45 minutes. The
extract was then loaded onto a 1D4 immunoaffinity
column in the presence of 10 gylcerol and
incubated for 2 hours, washed with wash buffer
containing 300 mM NaCl and eluted in 1D4 peptide
with 100 mM NaCl.
Figure 10 Single particle processing of EM images
using IMAGIC 30 micrographs were digitized on a
SCAI scanner (Zeiss) with a 7 mm pixel size.
Groups of 4 x 4 pixels were averaged to give 28
mm per pixel on the micrograph or 0.47 nm on the
specimen. Over 4,000 single-particle images were
hand selected and image processing was performed
as implemented in IMAGIC. The images were
filtered and normalized, and then rotationally
and translationally aligned. A new reference was
calculated using the aligned images and used as a
reference for further alignment and averaging
through five iterations. Aligned images were
subject to multivariate statistical analysis
(MSA) for classification. The best classes were
then used for multi reference alignment (MRA) for
further refinement of the individual particles.
We performed four more cycles of MRA, MSA, and
classification. Using angular reconstitution,
the relative orientations of the final classes
were used to determine a 3-D reconstruction with
imposed four-fold symmetry. Re-projections of
the 3-D structure were further used for MRA and
iterated till a stable structure was obtained.
CONCLUSIONS Here we have developed a system to
express Kv4.2 and KChIP2 in sufficient levels to
purify and examine using electron microscopy. We
have engineered a charybdotoxin binding site into
Kv4.2 to use as an assay to determine the amount
of channel expression. Purification was
performed using the 1D4 epitope and
immunoaffinity purification. Purified material
was then imaged using electron microscopy. The
preliminary structure reveals a large cytoplasmic
domain containing the N and C-termini of Kv4.2
and the KChIP2. Further work must be done to
verify the location of these domains. In
addition, the stoichiometry of the KChIPs must be
determined biochemically to validate the
imposition of four-fold symmetry.
Interestingly, the KChIPs appear to attach to
the sides of Kv4.2 as opposed to bottom like the
Shaker b subunits. In addition, the question of
whether the T1 domains of Kv4.2 maintain their
tetrameric state must be examined. It is known
that the KChIPs bind the T1 domain, this method
may be used to illuminate conformational changes
induced by KChIP binding. This structure also
suggests that the modulation of channel kinetics
by KChiP may be mediated by its interaction with
both the N and C-termini. Thus, channel
inactivation may involve both ends of Kv4.2,
suggesting a unique mechanism of Kv4.2
inactivation. Certainly, further work must be
accomplished to detail the location of each of
these domains. Future work may involve using
cyro-EM techniques to increase the resolution.
Figure 3 Mutations in Kv4.2 confer charybdotoxin
sensitivity In order to quantify the expression
of channels in our COS-7 expression system we
introduced a charybdotoxin binding site into
Kv4.2. In addition, the charybdotoxin can be
used to verify the presence of intact channels in
our purification. A radioactively labeled
charybdotoxin is used to assay and quantify
channel production. Charybdotoxin sensitivity
has been previously conferred to Shaker and
Kv2.1. Mutations near the pore, underlined in
Shaker, and bold in Kv2.1 confer sensitivity to
charybdotoxin. The asterisk marks phenylalanine
425 in shaker found to be critical in conferring
charybdotoxin sensitivity. Mutations were
introduced into Kv4.2 to similarly increase toxin
affinity.