Báo cáo khoa học: Characterization of N-glycosylation consensus sequences in the Kv3.1 channel pot

Characterization of N-glycosylation consensus sequences
in the Kv3.1 channel
Natasha L. Brooks, Melissa J. Corey and Ruth A. Schwalbe
Department of Biochemistry and Molecular Biology, Brody School of Medicine, East Carolina University, Greenville, NC, USA
Voltage-gated K
+
channel (Kv3.1) plays a fundamen-
tal role in neuronal excitability and lymphocyte differ-
entiation [1–6], and belongs to the Kv3 subfamily of
the voltage-gated K
+
channel (Kv) supergene family
[7]. Upon stimulation, the voltage-dependent gate
opens and potassium ions flow out of the cell, indu-
cing negative intracellular voltage, and termination of
excitation [8]. Based on hydropathy plots, Kv3.1 has
six transmembrane segments (S1–S6) and cytoplasmic
N- and C-termini (Fig. 1A). The segments between
S1–S2 and S3–S4 are extracytoplasmic loops, and
those between S2–S3 and S4–S5 are cytoplasmic
loops.
Each of the Kv3.0 family members and their splice
variants contain two conserved, native N-glycosylation
sites in the S1–S2 linker. Rat and human Kv3.1 pro-
teins have two native N-glycosylation consensus
sequences running from amino acid residues 220 to
Keywords
brain; glycosylation; K
+
channel; topology,
trafficking

Correspondence
R. A. Schwalbe, Department of
Biochemistry and Molecular Biology, Brody
School of Medicine at East Carolina
University, 600 Moye Boulevard, Greenville,
NC 27834, USA
Fax: +1 252 744 3383
Tel: +1 252 744 2034
E-mail:
(Received 25 August 2005, revised 18 May
2006, accepted 23 May 2006)
doi:10.1111/j.1742-4658.2006.05339.x
N-Glycosylation is a cotranslational and post-translational process of pro-
teins that may influence protein folding, maturation, stability, trafficking,
and consequently cell surface expression of functional channels. Here we
have characterized two consensus N-glycosylation sequences of a voltage-
gated K
+
channel (Kv3.1). Glycosylation of Kv3.1 protein from rat brain
and infected Sf9 cells was demonstrated by an electrophoretic mobility shift
assay. Digestion of total brain membranes with peptide N glycosidase F
(PNGase F) produced a much faster-migrating Kv3.1 immunoband than
that of undigested brain membranes. To demonstrate N-glycosylation of
wild-type Kv3.1 in Sf9 cells, cells were treated with tunicamycin. Also, par-
tially purified proteins were digested with either PNGase F or endoglycosi-
dase H. Attachment of simple-type oligosaccharides at positions 220 and
229 was directly shown by single (N229Q and N220Q) and double
(N220Q ⁄ N229Q) Kv3.1 mutants. Functional measurements and membrane
fractionation of infected Sf9 cells showed that unglycosylated Kv3.1s
were transported to the plasma membrane. Unitary conductance of

N220Q ⁄ N229Q was similar to that of the wild-type Kv3.1. However, whole
cell currents of N220Q ⁄ N229Q channels had slower activation rates, and a
slight positive shift in voltage dependence compared to wild-type Kv3.1.
The voltage dependence of channel activation for N229Q and N220Q was
much like that for N220Q ⁄ N229Q. These results demonstrate that the
S1–S2 linker is topologically extracellular, and that N-glycosylation influen-
ces the opening of the voltage-dependent gate of Kv3.1. We suggest that
occupancy of the sites is critical for folding and maturation of the func-
tional Kv3.1 at the cell surface.
Abbreviations
CDGS, carbohydrate-deficient glycoprotein syndromes; Endo H, endoglycosidase H; ER, endoplasmic reticulum; G–V plot, conductance–
voltage plot; Kv, voltage-gated K
+
channel; KvAP, voltage-gated K
+
channel of Aeropyrum pernix; PM, plasma membrane; PNGase F,
peptide N glycosidase F; Sf9, Spodoptera frugiperda; TM, tunicamycin.
FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3287
222 (NKT) and from 229 to 231 (NGT) in the S1–S2
linker, and they share 100% sequence identity in this
region (Fig. 1B). A recent X-ray structure of a Kv
from Aeropyrum pernix (KvAP) suggested that the
S1–S2 linker resides in the membrane for all Kvs [9].
Comparison between the S1–S2 linker of KvAP and
mammalian Kvs may be difficult because of differences
in the length of their S1–S2 linkers and the N-glycosy-
lation consensus sequences within this segment for
Kv3.0s and Kv1.0s (Fig. 1C).
N-Glycosylation is a cotranslational and post-trans-
lational modification found on extracellular segments

of membrane proteins and is important for protein
maturation, trafficking, and function [10–13]. The N-
glycosylation consensus sequence is AsnXxxSer ⁄ Thr,
where the central residue cannot be a Pro residue.
Membrane protein segments are only glycosylated
when they are translocated to the luminal side of the
endoplasmic reticulum (ER) membrane, and therefore
occupancy of an N-glycosylation site designates a
region of extracellular topology [11,12]. Defects in the
attachment of oligosaccharides to protein give rise to
mental and psychomotor retardation, dimorphisms,
and blood coagulation defects [14,15]. Carbohydrate-
deficient glycoprotein syndromes (CDGS) I–IV are a
group of disorders characterized by the presence of
abnormal oligosaccharides on many glycoproteins
[16,17]. The occurrence of CDGS emphasizes that
proper glycosylation of both membrane and secretory
glycoproteins are essential for normal development
and health [18,19]. More recently, it has been sugges-
ted that ER stress is linked to several human neuron-
al diseases [20], and therefore it may be that
abnormal glycosylation processing of proteins contri-
butes to these diseased states as well.
Here we have examined whether the native N-gly-
cosylation sites are utilized in rat brain and infected
Sf9 cells, and the role that occupancy of these sites
has in the expression of functional Kv3.1s at the cell
surface of Sf9 cells. Immunoband patterns of wild-
type Kv3.1, N220Q ⁄ N229Q, N229Q, and N220Q, in
the absence and presence of tunicamycin (TM), endo-

glycosidase H (Endo H), or peptide N glycosidase F
(PNGase F), revealed that both sites in Kv3.1 were
occupied by N-linked oligosaccharides. Patch clamp
measurements and cell fractionation showed that the
unglycosylated Kv3.1, N220Q ⁄ N229Q, is targeted to
the plasma membrane, like wild-type Kv3.1. However,
whole cell currents of N220Q ⁄ N229Q revealed slower
activation kinetics and a small positive shift in
voltage dependence compared to wild-type Kv3.1.
The voltage dependence of activation for the partially
glycosylated Kv3.1s, N229Q and N220Q, appeared
similar to that of N220Q ⁄ N229Q. Our findings
demonstrate that the S1–S2 linker of Kv3.1 is in an
extracellular aqueous environment. Additionally, they
demonstrate that N-glycosylation influences the open-
ing of the voltage-dependent gate of Kv3.1, suggest-
ing that vacant sites alter the folding and maturation
of Kv3.1 at the cell surface.
N
C
220
229
S1 S2
S3
S6
S5
S4
A
Rat 210 ETHERFNPIVNKTEIENVRNGTQVRYYREAETEAFLTY
Human 210 ETHERFNPIV

NKTEIENVRNGTQVRYYREAETEAFLTY
B
Kv3.1 P25122 210 ETHERFNPIVNKTEIENVRNGTQVRYYREAETEAFLTY
Kv1.1 P10499 187 ETLPELKDDKDFTGTIHRIDNTT
VIYTSNIFTDP
Kv1.2 P63142 183 ETLPIFRDENEDMHGGGVTFHTYSNST
IGYQQSTSFTDP
Kv1.4 P15385 329 ETLPEFRDDRDLIMALSAGGHSRLLNDT
SAPHLENSGHTIFNDP
Kv1.5 P19024 261 ETLPEFRDERELLRHPPVPPQPPAPAPGINGS
VSGALSSGPTVAPLLPRTLADPF
KvAP Q9YDF8 64 SGEY
C
Fig. 1. Topological model of Kv3.1 and
amino acid sequences of the S1–S2 linker
of Kvs. (A) Topology of a Kv3.1 monomeric
unit. Black circles represent the Asn of
native N-glycosylation consensus sites N220
and N229. Branched structures represent
the attachment of oligosaccharide at native
sites. (B) Sequence identity between Kv3.1
S1–S2 linkers from rat (P25122) and human
(P48547). (C) Comparison of the S1–S2
amino acid sequence of eukaryotic Kvs and
prokaryotic KvAPs. The Kv name corres-
ponding to the adjacent S1–S2 amino acid
sequence is indicated in bold and is fol-
lowed by the accession number. Conserved,
native N-glycosylation sites are shown as
underlined font. The italicized number indi-

cates the first residue of the S1–S2 linker.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.
3288 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
Results
Occupancy of the two native N-glycosylation
sites
Rat brain membranes were digested with PNGase F,
and then analyzed by western blotting. PNGase F is
an enzyme that removes a wide range of N-linked
oligosaccharides from proteins [21]. Native Kv3.1
migrates as a diffuse immunoband (about 109 kDa)
which is much larger than its calculated molecular
mass of 66 kDa (Fig. 2). This migration pattern of
native Kv3.1 suggests that the protein undergoes a
cotranslational or post-translational modification. To
show that the modification was indeed a result of
attachment of N-linked oligosaccharides, rat brain
membranes were incubated with PNGase F. The
Kv3.1 immunoband (about 81 kDa) migrated much
faster, indicating that Kv3.1 undergoes N-glycosylation
in rat brain membranes. To further verify specificity of
the Kv3.1b antibody, membranes isolated from Sf9
cells infected with recombinant baculovirus that enco-
ded the Kv3.1b cDNA were immunoblotted (Fig. 2).
The electrophoretic migration pattern of wild-type
Kv3.1 revealed a predominant immunoband at about
87 kDa and two lower faint bands. The lowest band
migrated to about 77 kDa, and the middle band was
at about 81 kDa. Only the two lowest bands were
detected when Sf9 cell membranes were treated with

PNGase F, suggesting that the top two bands are gly-
cosylated protein. Taken together, these results demon-
strate that Kv3.1 is N-glycosylated in rat and insect
cells, and that the type of N-linked oligosaccharide
differs.
To directly demonstrate that both of the absolutely
conserved N-glycosylation consensus sequences were
utilized, they were removed independently (N229Q
and N220Q) and simultaneously (N220Q ⁄ N229Q) by
conserved substitutions of the Asn residues with Gln
residues. In addition, an M2 FLAGÒ epitope was
attached to the C-terminus and was utilized for purifi-
cation and identification of the various Kv3.1 proteins.
Wild-type Kv3.1, N229Q, N220Q and N220Q ⁄
N229Q were M2 immunoaffinity purified from whole
cell lysates of Sf9 cells infected in the absence and
presence of TM, and then immunoblotted using anti-
++
_
_
Anti-Kv3.1b
rat brain
membranes
Sf9 cell
membranes
PNGase F:
Fig. 2. N-Glycosylation of Kv3.1 in rat brain tissue and Sf9 cell
membranes. Rat brain membranes and partially purified Sf9 pro-
teins were untreated (–) or treated (+) with PNGase F, resolved by
SDS ⁄ PAGE and immunoblotted, as indicated. The arrows of each

panel indicate migration of glycosylated (upper arrow) or unglycosyl-
ated (lower arrow) Kv3.1 protein. Ovals represent Kaleidoscope
TM
protein standards (top to bottom in kDa): 250, 150, 100, and 75.
Endo H:
____
++++
TM
:
Wt
N220Q/
N229Q
N229Q N220Q
Kv3.1:
____
++++
Anti-FLAG
Wt N229Q N220Q
N220Q/
N229Q
+ +++
_
_
_
_
Kv3.1:
TM:
A
B
Anti-Kv3.1b

Fig. 3. Detection of high-mannose oligosaccharides in Sf9 cell
membranes. Sf9 cells were infected with recombinant baculovirus
containing wild-type Kv3.1, N229Q, N220Q or N220Q ⁄ N229Q in
either the absence (–) or the presence (+) of 25 lgÆmL
)1
tunicamy-
cin (TM). Proteins were transferred and probed with anti-Kv3.1b (A)
or anti-FLAG (B, upper panel). Partially purified Kv3.1 protein was
treated (+) with endoglycosidase H (Endo H) (B, lower panel). The
arrows in each panel indicate migration of fully glycosylated (upper),
partially glycosylated (middle) or unglycosylated (lower) Kv3.1
protein. In all cases, proteins were partially purified from whole cell
lysates using M2-agarose. Ovals represent molecular mass stand-
ards (top to bottom in kDa): 250, 150, 100, and 75.
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1
FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3289
Kv3.1b (Fig. 3A) and M2 anti-FLAG (Fig. 3B). TM
inhibits the oligosaccharyltransferase that carries out
the initial step of the N-glycosylation pathway in the
ER lumen [22]. As mentioned above, wild-type Kv3.1
migrates as three immunobands, with the upper band
as the predominant band. When Sf9 cells expressing
wild-type Kv3.1 were treated with TM, only the lowest
immunoband was observed. The single Kv3.1 mutants,
N229Q and N220Q, migrated as doublets which
appear to correspond to the lower two immunobands
of wild-type Kv3.1. In both cases, the upper band was
darker than the lower band. Additionally, the upper
band was not visible in the presence of TM. A single
immunoband was detected for N220Q ⁄ N229Q, which

migrated to a similar position as the lowest faint band
observed for wild-type Kv3.1 and the lower faint band
of the single mutants, and furthermore, the immuno-
band did not shift in the presence of TM. To verify
that wild-type Kv3.1 was modified by a high-mannose
oligosaccharide typical of Sf9 cells, not a complex
oligosaccharide [23], N-linked oligosaccharide was
removed by Endo H treatment of partially purified
Kv3.1 protein (Fig. 3B). When partially purified wild-
type Kv3.1, N229Q and N220Q proteins were digested
with Endo H, the lowest band becomes the predomin-
ant form in all three instances. The band observed for
N220Q ⁄ N229Q does not shift in the presence of Endo
H. These results indicate that the upper band of
wild-type Kv3.1 represents the situation when both
glycosylation sites are occupied by high-mannose-type
oligosaccharides, the middle band represents one occu-
pied site, and the lowest band is the unglycosylated
monomer.
Glycosylated and unglycosylated forms of Kv3.1
are targeted to the plasma membrane
Infected Sf9 cells expressing wild-type Kv3.1 and
N220Q ⁄ N229Q were fractionated into three distinct
fractions [24,25]. Subsequently, Kv3.1 protein was M2
immunoaffinity purified from each fraction (Fig. 4A).
A predominant immunoband was detected for wild-
type Kv3.1 in all three distinct fractions. Two faint
lower bands were clearly observed in the ER fraction,
while in the other two fractions only the lower faint
band was observed. The totally unglycosylated form of

Kv3.1, generated by mutating Asn residues at positions
220 and 229 to Gln residues (N220Q ⁄ N229Q, Fig. 4A)
or by treating Sf9 cells expressing wild-type Kv3.1 with
TM (Fig. 4B), was also observed in the plasma mem-
brane. These results indicate that N-glycosylation is
not required to transport Kv3.1 to the plasma mem-
brane.
Functional unglycosylated Kv3.1 is at the cell
surface
Whole cell currents of infected Sf9 cells expressing
either wild-type Kv3.1 or N220Q ⁄ N229Q were
observed when the membrane potential was depolar-
ized beyond ) 10 mV and current amplitudes reached
saturation at membrane potentials beyond +40 mV
(Fig. 5A,B, top panel). The patterns of these inactivat-
ing, voltage-dependent K
+
currents were typical of a
delayed rectifier, and were similar to those expressed
by wild-type Kv3.1 in Xenopus oocytes [1,3,4,26–28]
and other heterologous expression systems [29–31]. To
show that channel densities at the cell surface for wild-
type Kv3.1 (I
max
⁄ cap is 140 ± 29 pA ⁄ pF, n ¼ 13) and
N220Q ⁄ N229Q (I
max
⁄ cap is 156 ± 36 pA ⁄ pF, n ¼ 11)
were comparable, the maximum current amplitude was
determined and divided by the cell capacitance. Differ-

ences between the two forms could be identified when
the voltage dependence for channel activation was
analyzed. The membrane conductance vs. applied test
potential indicated that more depolarization was
required for 50% of the N220Q ⁄ N229Q channels (V
0.5
[test potential at which g/g
max
¼ 0.5] is 20.5 ±
0.6 mV, n ¼ 11) to reach activation than for wild-type
Kv3.1:
A
Fraction:
PM Golgi
ER
PM Golgi
ER
Wt
N220Q/N229Q
B
Fraction:
PM Golgi
ER
Wt Kv3.1 +TM
Kv3.1:
Fig. 4. Glycosylated and unglycosylated Kv3.1 proteins are deliv-
ered to the plasma membrane. (A) Plasma membrane (PM), Golgi
apparatus (Golgi) and endoplasmic reticulum (ER) fractions from
Sf9 cells infected with the indicated Kv3.1 baculovirus were isola-
ted by sucrose density gradients. Protein was M2-affinity agarose

purified from each fraction and immunoblotted. (B) M2-agarose
affinity purified Kv3.1 protein from membrane fractions of Sf9 cells
expressing wild-type Kv3.1 in the presence of 25 lgÆmL
)1
tunica-
mycin (TM). The top two arrows denote where glycoforms would
be and the bottom arrow represents the aglycoform.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.
3290 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
Kv3.1s (V
0.5
is 16.6 ± 0.7 mV, n ¼ 13). Additionally,
slightly fewer channels were activated as the applied
voltage was increased for unglycosylated Kv3.1 (slope
of normalized current voltage relationship, dV,is
9.3 ± 0.4 mV for N220Q ⁄ N229Q, n ¼ 11) than for
glycosylated Kv3.1 (dV is 8.1 ± 0.4 mV for wild-type
Kv3.1, n ¼ 13). A range of values for Vm
0.5
from
10 mV to 18 mV, and for dV from 8 mV to 11 mV, of
wild-type Kv3.1 have previously been reported in
various heterologous expression systems [30].
The activation kinetics of wild-type Kv3.1
expressed in vitro and in vivo is quite rapid [30,31].
When the whole cell current tracings were normalized
at each potential from + 20 mV to + 50 mV for
wild-type Kv3.1 and N220Q ⁄ N229Q, and then placed
on top of each other, it was observed that the
activation kinetics were somewhat slower for

N220Q ⁄ N229Q than for wild-type Kv3.1 (Fig. 6A).
AB
25 ms
25 ms
0.5 nA
0.5 nA
g/gmax
Voltage (mV)
C
-40 -20 0 20 40 60 80 100
0.0
0.2
0.4
0.6
0.8
1.0
Wt Kv3.1
N220Q/N229Q
Fig. 5. Functional expression of wild-type Kv3.1 and the
N220Q ⁄ N229Q mutant. Whole cell currents were produced by
depolarizing voltage pulses from a holding potential of ) 50 mV to
levels ranging from ) 40 to +100 mV in 10 mV increments. Repre-
sentative tracings are shown from Sf9 cells infected with (A) wild-
type Kv3.1 and (B) N220Q ⁄ N229Q. (C) Corresponding Boltzmann
plots. Wild-type Kv3.1 (V
0.5
¼ 16.6 ± 0.7, dV ¼ 8.1 ± 0.4, n ¼ 13)
data are represented by (d) and N220Q ⁄ N229Q (V
0.5
¼

20.56 ± 0.6, dV ¼ 9.3 ± 0.4, n ¼ 11) data are represented by n.
The Boltzmann isotherm G ¼ G
max
⁄ [1 + exp(V
0.5
) V) ⁄ q] was used
to fit the data, which represent ± SEM.
A
25 ms
tnerruc
B
20 40 60 80 100
0
10
20
30
40
50
60
Voltage (mV)

Rise times (ms)
20 40 60 80 100
0
10
20
30
40
Volta
g

e (mV)
Activation time constants (ms)
C
Fig. 6. Comparison of activation rates in wild-type Kv3.1 and
N220Q ⁄ N229Q. (A) Whole cell currents from Sf9 cells expressing
wild-type Kv3.1 (solid line) and N220Q ⁄ N229Q (dashed line) were
normalized at 20 mV (red), 30 mV (blue), 40 mV (purple) and 50 mV
(black), and the resulting normalized currents were placed on top of
each other. (B) Rise times and (C) activation time constants are
shown for wild-type Kv3.1 (d) and N220Q ⁄ N229Q (n). Rise times
represent the time required for the current to rise from 10% to
90% of its peak current at the indicated applied voltage. Activation
time constants were determined by fitting the current traces at
each potential to a single exponential. Data represent SEM.
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1
FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3291
In both cases, the activation time decreased as the
applied potential increased, which indicates the volt-
age dependence of channel activation. The time for
the current to rise from 10% to 90% of its maximum
value was less for N220Q ⁄ N229Q than for wild-type
Kv3.1 at the various applied potentials (Fig. 6B).
Time constants for activation at similar potentials
were also determined by fitting each current with a
single exponential (Fig. 6C). Again, it was demonstra-
ted that the activation rate for N220Q ⁄ N229Q is
slower than that for wild-type Kv3.1. The deactiva-
tion kinetics of wild-type Kv3.1 (time constant deacti-
vation, s
off

is 2.4 ± 0.9 at ) 40 mV, n ¼ 3) and
N220Q ⁄ N229Q (s
off
is 3.6 ± 0.3 at ) 40 mV, n ¼ 3)
were rapid, and similar to those previously reported
in heterologous expression systems and neurons [31].
These results indicate that differences in the voltage
dependence of channel activation can be measured
between the glycosylated Kv3.1 (wild-type Kv3.1) and
its unglycosylated counterpart (N220Q ⁄ N229Q).
Previously, it has been reported that whole cell
current recordings of wild-type Kv3.1 in mammalian
expression systems display little saturation in current
amplitude in response to large depolarization steps
[29,32,33]. This noninactivating type of behavior was
also observed for both glycosylated and unglycosylat-
ed Kv3.1s expressed in Sf9 cells (Fig. 7A,B) but
occurred less often than the inactivating currents. In
the case of the noninactivating behavior, the channel
densities for wild-type Kv3.1 (I
max
⁄ cap is 368 ± 29
pA ⁄ pF, n ¼ 11) and N220Q ⁄ N229Q (I
max
⁄ cap is
314 ± 50 pA ⁄ pF, n ¼ 9) were quite similar. How-
ever, both forms of Kv3.1 had higher channel densi-
ties than those that had inactivating behavior. Like
the cells that expressed the inactivating type of
behavior for wild-type Kv3.1 and N220Q ⁄ N229Q,

the rise times at the various potentials were slower
for the unglycosylated Kv3.1 than for glycosylated
Kv3.1 (Fig. 7C). Moreover, the rise times were faster
in those cells that expressed the noninactivating type
of behavior than in those that expressed the inacti-
vating type of behavior for either wild-type Kv3.1 or
N220Q ⁄ N229Q.
Single-channel recordings of wild-type Kv3.1 and
N220Q ⁄ N229Q have long openings, and long and brief
closures (Fig. 8A,B). Current amplitudes and unitary
conductances of wild-type Kv3.1 and N220Q ⁄ N229Q
were virtually identical (Fig. 8C), and quite similar to
those in previous reports of wild-type Kv3.1 in
Xenopus oocytes [26,28] and mammalian cells [29,32].
These results indicate that the current amplitudes and
single-channel conductances are similar for glycosylat-
ed and unglycosylated Kv3.1s.
Partially glycosylated Kv3.1 mutants at the cell
surface are functional
The single N-glycosylation Kv3.1 mutants (N229Q and
N220Q) expressed whole cell currents at applied poten-
tials of ) 10 mV, and current amplitudes increased as
the applied potential increased until currents reached
saturation at membrane potentials beyond + 40 mV
(Fig. 9A,B). The Boltzmann equation indicates that
a little more depolarization is required to activate
50% of the partially glycosylated Kv3.1s (V
0.5
is
18.7 ± 1.2 mV and 20.9 ± 1.3 mV for N229Q and

N220Q, respectively, n ¼ 5) than for the fully glycosyl-
ated Kv3.1 (wild-type Kv3.1; Fig. 9C). The conduct-
ance–voltage (G–V) slopes for N229Q (dV is
8.7 ± 0.6 mV, n ¼ 5) and N220Q (dV is 9.0 ± 0.7 mV,
n ¼ 5) appear to be more similar to those for
N220Q ⁄ N229Q than wild-type Kv3.1, but they were not
statistically different. As for the N220Q ⁄ N229Q channel
(rise times at + 20 mV, 50.5 ± 3.3 ms, and + 40 mV,
29.6 ± 2.6 ms, n ¼ 10; and activation time constants
at + 20 mV, 27.7 ± 3.1 ms, and + 40 mV, 16.0 ±
20 40 60 80 100
0
10
20
30
40
Voltage (mV)
Rise times (ms)
C
AB
1 nA
1 nA
25 ms 25 ms
Wt Kv3.1 N220Q/N229Q
Fig. 7. Whole cell analysis of noninactivating currents from wild-
type Kv3.1 and N220Q ⁄ N229Q. Representative tracings are shown
from Sf9 cells infected with (A) wild-type Kv3.1 and (B)
N220Q ⁄ N229Q. (C) Rise times are shown for wild-type Kv3.1 (n ¼
11, d) and N220Q ⁄ N229Q (n ¼ 8, n). Data represent ± SEM.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.

3292 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
1.2 ms, n ¼ 11), the activation kinetics of N229Q (rise
times at +20 mV, 54.9 ± 6.4 ms, and +40 mV,
26.3 ± 2.3 ms; and activation time constants at
+20 mV, 29.6 ± 5.6 ms, and +40 mV, 12.8 ± 1.0 ms,
n ¼ 5) and N220Q (rise times at +20 mV,
54.5 ± 5.2 ms; and +40 mV, 34.4 ± 2.8 ms; and acti-
vation time constants at +20 mV, 30.8 ± 3.8 ms, and
+40 mV, 16.7 ± 0.9 ms, n ¼ 5) were slower than those
of wild-type Kv3.1 (rise times at +20 mV,
41.0 ± 2.7 ms and +40 mV, 19.4 ± 1.1 ms; and acti-
vation time constants at +20 mV, 20.4 ± 1.6 ms, and
+40 mV, 10.5 ± 0.8 ms, n ¼ 13). These results indi-
cate that the absence of one N-linked high-mannose-
type oligosaccharide can produce small changes in the
voltage dependence of channel activation.
Discussion
The findings reported here demonstrate that both of
the conserved N-glycosylation consensus sequences, in
the S1–S2 linker, of Kv3.1 can be occupied by various
types of oligosaccharide. Three distinct immunobands
were identified for wild-type Kv3.1 expressed in Sf9
cells. The upper band was the predominant band, and
the two lower bands were of minor intensity (Fig. 3).
In the presence of PNGase F, Endo H, or TM, the
upper band was no longer observed. When digestion
was complete or cells expressing wild-type Kv3.1 were
treated with TM, only the lowest faint immunoband
was observed (Fig. 3). In addition, when the conserved
N-glycosylation sites were removed either independ-

ently or together, the immunobands corresponded to
the two faint bands of wild-type Kv3.1 or the lowest
faint band, respectively. These results indicate that the
most prominent band of wild-type Kv3.1 expressed in
insect cells represents occupancy of both glycosylation
sites by simple oligosaccharides, the upper faint band
represents occupancy of one site by a simple oligosac-
charide, and the lowest faint band represents vacancy
of both sites.
A
+60 mV
+80 mV
+100 mV
100 ms
1 pA
Wt Kv3.1 N220Q/N229Q
B
O
C
30 60 90 120 150
1
2
3
4
C
Volta
g
e (mV)
Current (pA)
Fig. 8. Single-channel analysis of wild-type Kv3.1 and N220Q ⁄ N229Q. Representative single-channel recordings from infected Sf9 cells

expressing (A) wild-type Kv3.1 ( d ) and (B) N220Q ⁄ N229Q (n) at various test potentials as indicated. The open state of the channel is indica-
ted by O and the closed state is represented by C. (C) Current–voltage relationship of wild-type Kv3.1 and N220Q ⁄ N229Q. Single-channel
conductance was 27 pS (n ¼ 8) for wild-type Kv3.1 and 24 pS (n ¼ 6) for N220Q ⁄ N229Q. Open circles denote wild-type Kv3.1 and closed
circles represent N220Q ⁄ N229Q. Linear regression fit of the data was performed with a dashed line for wild-type Kv3.1 and a solid line for
N220Q ⁄ N229Q. Data represent ± SEM.
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1
FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3293
On the basis of conservation of initial steps in the
N-glycosylation pathway and divergence of this path-
way following synthesis of the common N-glycan inter-
mediate, GlcNAc
2
Man
3
GlcNAc
2
-N-Asn, in insects and
mammals [34], we would expect Kv3.1 to be glycosyl-
ated in native tissue. A diffuse immunoband was
observed in rat brain membranes that migrated much
slower than would be expected from its calculated
molecular mass, or that detected in Sf9 cells (Fig. 2).
Previous studies are in agreement that Kv3.1 in rat
brain migrates much slower than would be expected
from its predicted molecular mass [35], and further-
more, its migration appears to differ in various regions
of the brain [36,37]. This slow migration pattern of
Kv3.1 was not shown to be due to N-glycosylation.
Our results show that digestion of rat brain mem-
branes with PNGase F produces a much faster-migra-

ting band that moves to a similar position as
unglycosylated Kv3.1 expressed in Sf9 cells. Taken
together, we conclude that Kv3.1 isolated from rat
brain is N-glycosylated (Fig. 2), and the oligosaccha-
rides are of either hybrid or complex type in composi-
tion. Additionally, it may be that the composition of
N-linked oligosaccharides is different in various
regions of the brain.
Many glycosylation studies have indicated that in
order for the oligosaccharyltransferase to have access
to an N-glycosylation consensus sequence of a mem-
brane protein, the segment containing the tripeptide
sequence must enter the lumen of the ER, which
becomes the extracellular segment of the protein once
it is transferred to the plasma membrane [11,38,39].
Additionally, if the site is within an extracytoplasmic
loop, the segment must be larger than 30 residues, and
this site must be at least 11 residues away from the
membrane. Utilization of a site is also greater at an
earlier time point during protein synthesis. In conjunc-
tion, glycosylation of sites at positions 220 and 229 of
Kv3.1 confirms the extracellular placement of the S1–
S2 linker identified by hydropathy plots (Fig. 1A). This
finding is also in agreement with utilization of the
native glycosylation site in the S1–S2 linker of other
Kvs (Fig. 1C). For example, the native glycosylation
site in the S1–S2 linker of Shaker H4 [40], Kv1.1 [41],
Kv1.2 [42], Kv1.4 [43] and Kv1.5 [44] were shown to
be occupied by N-linked oligosaccharides. Addition-
ally, utilization of introduced glycosylation sites

throughout the S1–S2 segment of Kv1.2 suggested that
the majority of this segment resides in the extracellular
aqueous environment and that its conformation is flex-
ible [42]. The Kv3.1 results demonstrated that both
glycosylation sites in the S1–S2 linker were utilized,
indicating that both of these regions can accommodate
a conformation accessible to the oligosaccharyltransf-
erase. Moreover, the large hydrophilic oligosaccharides
attached to Asn220 and Asn229 would place the entire
region of the S1–S2 linker outside the lipid bilayer.
The structural model of bacterial KvAP places this
segment in a lipid environment, and suggests that this
was the case for all Kvs [9]. It is possible that the S1–
S2 linker may reside in the membrane of a bacterial
Kv, which would not undergo N-glycosylation, but it
is unlikely that this orientation applies to all eukaryot-
ic Kvs. The S1–S2 linker of KvAP is very short (four
residues) compared to the longer linkers of Kv3.1 (38
residues), Kv1.1 (34 residues), Kv1.2 (39 residues),
Kv1.4 (44 residues) and Kv1.5 (55 residues) (Fig. 1C).
Therefore, this region in the bacterial channel is not
very comparable to that of eukaryotes. The aforemen-
tioned glycosylation reports, along with our report,
provide strong evidence that the conserved S1–S2
linker of Kv1.1, Kv1.2, Kv1.4 and Kv1.5, along with
that of Kv3.1, is topologically extracellular. The
reports also suggest that the orientations of the S1 and
C
-40 -20 0 20 40 60 80 100
0.0

0.2
0.4
0.6
0.8
1.0
g/gmax
Voltage (mV)
25 ms
0.5 nA
A

0.5 nA
25 ms
B
N220Q
N229Q
Fig. 9. Functional expression of N-glycosylation single mutants.
Whole cell currents are shown for Sf9 cells infected with (A)
N229Q and (B) N220Q. (C) Corresponding Boltzmann plots for
N229Q (. solid line; V
0.5
¼ 18.7 ± 1.2, d V ¼ 8.7 ± 0.6, n ¼ 5) and
N220Q (m dashed line; V
0.5
¼ 20.9 ± 1.3, dV ¼ 9.0 ± 0.7, n ¼ 5)
are shown here.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.
3294 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
S2 segments are similar in all eukaryotic Kv1.0 and
Kv3.0 subfamilies, but not necessarily the same

between prokaryotic and eukaryotic domains.
N-Glycosylation, a cotranslational and post-transla-
tional process, of proteins may influence protein fold-
ing, maturation, stability, trafficking, and consequently
cell surface expression of functional channels [10–12].
Cell fractionation results of unglycosylated Kv3.1
produced by elimination of the two native sites,
N220Q ⁄ N229Q mutant, or by treating cells expressing
wild-type Kv3.1 with TM, demonstrated that targeting
to the plasma membrane was not abolished. In addi-
tion, whole cell current measurements of unglycosylat-
ed Kv3.1 (N220Q ⁄ N229Q) and partially glycosylated
Kv3.1 (N229Q and N220Q) show that they form func-
tional channels at the cell surface. These results indi-
cate that N-glycosylation is not required for transport
of Kv3.1 to the cell surface, and thus suggest that the
protein can fold and assemble into a stable functional
homomultimer. However, the small changes in the acti-
vation kinetics suggest that when the sites are vacant,
Kv3.1 is folded slightly different in its mature structure
compared to when the sites are occupied. Previous
studies of Kv1.4 have also demonstrated that glycosy-
lation of the native site in the S1–S2 linker was
required for proper trafficking and stability [43]. On
the other hand, glycosylation of the native site in the
S1–S2 linker of Kv1.0s, including Shaker [40] and its
mammalian homologue, Kv1.1 [41], did not affect cell
surface expression.
In this article, we demonstrate that the voltage
dependence of Kv3.1 activation is altered by the

absence of N-linked oligosaccharides. More depolar-
ization (about 6 mV) was required to activate 80% of
the unglycosylated Kv3.1s (N220Q ⁄ N229Q; V
>80%
is
38 mV) than to activate 80% of the glycosylated
Kv3.1s (wild-type Kv3.1; V
>80%
is 32 mV). Addition-
ally, the fraction of channels that were activated as the
applied voltage increased was significantly lower for
unglycosylated Kv3.1s than for their glycosylated
counterparts. The time required for the unglycosylated
Kv3.1s to reach their peak current was also less at the
various applied potentials relative to the glycosylated
Kv3.1s. Activation kinetic values of the partially gly-
cosylated Kv3.1s (N229Q and N220Q) appeared to be
more similar to those of unglycosylated Kv3.1 than to
that of the glycosylated channel. Thus, these results
indicate that the occupancy of the N-glycosylation sites
is a determining factor for the voltage-dependent acti-
vation kinetics of Kv3.1.
A recent report on Kv1.1 indicated that N-glycosyla-
tion did alter its gating function, and this effect was
shown to result from sialic acid residues attached to
the oligosaccharide [45]. It was suggested that when
the composition of the N-linked oligosaccharide was
of a simple type, there was a positive shift in voltage
dependence of activation and slower activation kinetics
than when the oligosaccharide was of a complex type

[45,46]. Therefore, it is possible that the maturation
processing of the high-mannose glycan of Kv3.1 to a
complex carbohydrate may cause a negative shift in
voltage dependence, and an increase in the rate of acti-
vation. When the channel activation of wild-type
Kv3.1s expressed in Sf9 cells is compared to that
expressed both in vivo and in mammalian heterologous
expression systems (V
>80%
is 30 mV; t
on
at +40 mV
is 3.4 ms; t
on
at +20 mV is 3–4 ms), the voltage
dependence of activation appears to be different
[31,47]. The results of the aforementioned studies of
Kv3.1, along with our report, would suggest that the
composition of the N-linked oligosaccharides may
influence the voltage dependence of Kv3.1 activation
in a similar manner as occupancy of the glycosylation
sites.
In conclusion, our study indicates that decreases in
the occupancy of N-glycosylation sites that may occur
in patients suffering from CDGS [15,19] or ER stress-
related neurodegenerative diseases [20] could alter the
expression of K
+
currents at the cell surface of neu-
rons that express Kv3.1 at high densities. Future stud-

ies will be needed to determine whether different
compositions of the N-linked oligosaccharide alter the
channel activation of Kv3.1.
Experimental procedures
Materials
Spodoptera frugiperda (Sf9) cells were obtained from
PharMingen, San Diego, CA, USA. Hink’s TNM-FH
insect medium was bought from MediaTech, Inc., Herndan,
VA, USA, FBS was from Invitrogen, Carlsbad, CA, USA,
and Pluronic F-68 solution, as well as gentamicin, came
from Sigma Chemical Co., St Louis, MO, USA. The TA
cloning kit and restriction enzymes were acquired from
Invitrogen. The BaculoGold transfection kit was purchased
from BD Biosciences, San Diego, CA. Plasmid purification
columns were obtained from Qiagen, Valencia, CA, USA.
Anti-FLAGÒ M2-agarose gel and mouse anti-FLAGÒ M2
were purchased from Sigma, and goat anti-mouse IgG,
alkaline phosphatase-conjugated, was purchased from MP
Biomedicals, Inc., Irvine, CA, USA. Rabbit anti-Kv3.1b
was purchased from Alomone Laboratories, Jerusalem,
Israel. Protease inhibitor cocktail set III and Triton X-100
were from CalBiochem, San Diego, CA, USA. Precast
SDS ⁄ PAGE gels were procured from Bio-RAD, Hercules,
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1
FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3295
CA, USA. Ultracentrifuge tubes for SW41 and 70Ti rotors
were purchased from Beckman, Palo Alto, CA, USA. All
other chemicals used in this study were ordered from Sigma
or Fisher Scientific, Co., Hampton, NH, USA.
Mutant constructs

PCR was used to attach the FLAG sequence (DY-
KDDDDK) to the 3¢-end of Rattus rattus Kv3.1 cDNA
(accession number P25122), referred to as 3¢FLAG-Kv3.1.
The 3¢FLAG-Kv3.1–pCRII recombinant vector was kindly
provided by B. Wible and A. M. Brown (Rammelkamp
Center for Education and Research, Case Western Univer-
sity, Cleveland, OH, USA). The N220Q ⁄ N229Q mutant
was constructed by PCR overlap extension [48] using
3¢FLAG-Kv3.1–pCRII as template. Forward and reverse
primers were designed to contain nucleotide mismatches
that eliminated the two native N-glycosylation sites at posi-
tions 220–222 (NKT) and 229–231 (NGT) of Kv3.1 cDNA.
PCR products were subcloned into pCR2.1 for amplifica-
tion and sequencing. The N220Q ⁄ N229Q was then sub-
cloned into EcoRI-digested baculovirus transfer vector,
pACSG2. To generate the N220Q and N229Q single
mutants, the QuikChange site-directed mutagenesis kit
(Stratagene, La Jolla, CA, USA) was used (manufacturer’s
protocol). Mutagenic forward and reverse primers were
designed to contain nucleotide mismatches that eliminated
either the first native N-glycosylation site at position
220–222 (NKT) or the second native site at position
229–231 (NGT) of Kv3.1 cDNA, respectively. The dsDNA
template was Kv3.1-pacSG2. DNA sequences were verified.
Standard procedures were followed for subcloning, and
DNA amplification, isolation and sequencing [49].
Cell culture and recombinant baculoviruses
Sf9 cells were maintained in Hink’s TNM-FH medium
containing 10% FBS, 10 lgÆmL
)1

gentamicin, and 0.1%
Pluronic F-68 at 27 °C as previously described [50]. Mono-
layer Sf9 cultures were used to maintain Sf9 cells and were
passaged about twice a week. Suspension cultures of Sf9
cells were seeded from monolayer cultures and stirred at a
constant rate of 80–120 r.p.m. Fresh suspension cultures
were prepared every 3–5 days. Recombinant baculoviruses
were produced by cotransfection of recombinant Baculovi-
rus transfer vectors and BaculoGold viral DNA [modified
Autographa californica nuclear polyhedrosis virus
(AcNPV)]. The manufacturer’s instructions were followed
for this procedure (BD Biosciences). Viral seed stocks of
intermediate viral titer were generated using monolayer cul-
tures. High viral titers were produced in suspension cultures
(0.8–0.9 · 10
6
viable cellsÆmL
)1
) using an aliquot of the
intermediate viral titer supernatant. Expression of recom-
binant proteins required the addition of high viral titer
supernatant to a suspension culture (1.1 · 10
6
cellsÆmL
)1
),
followed by an incubation period of 24 h at 27 °C. When
necessary for studying the occupancy of N-glycosylation
sites, TM (25 lgÆmL
)1

) was added to infected cells 15 min
postinoculation.
Cell fractionation and M2-agarose affinity
purification
Sf9 cell fractionations were carried out as previously des-
cribed [24,25]. Adjustments of the cell fractionation proto-
col involved reducing the starting material and the volume
of the sucrose layers but maintaining the relative ratios of
the sucrose layers. Infected Sf9 cells (about 6.6 · 10
7
) were
harvested by centrifugation at 1204 g in a Beckman SX
4250 rotor for 10 min at 4 °C. The pellet was washed once
in 15 mL of ice-cold NaCl ⁄ P
i
(50 mm Na
2
HPO
4
; 150 mm
NaCl) at pH 7.4, and recentrifuged under the same condi-
tions. The pellet was frozen at ) 20 °C for at least 1 h to
lyse the cells. Ice-cold homogenizing buffer (250 mm
sucrose, 10 mm Tris, pH 7.4; 1 lLÆmL
)1
protease inhibitor
cocktail set III from Calbiochem; added volume, about
1.8 mL) was used to resuspend the thawed cell pellet. The
cells were disrupted in a dounce homogenizer (continuous
strokes for at least 10 min) and centrifuged at 500 g for

15 min at 4 °C. All centrifugation was performed using an
Eppendorf 45-30-11 rotor, unless otherwise specified. An
equal volume of sucrose adjustment buffer (2.55 m sucrose,
10 mm Tris, 1 mm EDTA, pH 7.4) was added to the
cleared lysate to increase the sucrose concentration to
1.4 m. The sucrose gradient [2.0 m sucrose, 920 lL; 1.6 m
sucrose, 1840 lL; 1.4 m sucrose (sample), about 3680 lL;
1.2 m sucrose, 3680 lL; 0.8 m sucrose, 1840 lL] was pre-
pared and centrifuged at 83 472 g in a Beckman SW41
rotor for 2.5 h at 4 °C. Following the centrifugation,
plasma membrane (PM), Golgi apparatus (Golgi), and ER
fractions were removed and added to ultracentrifuge tubes,
where they were diluted by addition of about 10 mL of imi-
dazole buffer (25 mm imidazole, 1 mm EDTA, pH 7.4),
and centrifuged at 117 734 g in a Beckman 70Ti rotor for
1.5 h to concentrate the fractions. The pellet was solubilized
in 200 lL of resuspension buffer (50 mm Na
2
HPO
4
, 0.3 m
KCl, pH 7.4, 0.5% Triton X-100) and transferred to micro-
centrifuge tubes. The tube was rinsed with 800 lL of pellet
resuspension buffer to recover any remaining sample. Next,
20 lL of M2-agarose affinity gel (1 : 1 NaCl ⁄ P
i
and M2-ag-
arose gel slurry) was added to each solubilized sample (total
volume, about 1 mL) and incubated on a rotator for 1 h at
room temperature. The resin was washed three times with

1 mL of high-K
+
salt buffer (50 mm Na
2
HPO
4
, 0.3 m KCl,
pH 7.4) by centrifugation at 425 g for 3 min. The resin was
washed with 1 mL of NaCl ⁄ P
i
three times by centrifugation
(same as previous step), and the washed resin was resus-
pended in 100–125 lLof2· SDS ⁄ PAGE sample buffer
(62.5 mm Tris, pH 6.8, 2% SDS, 25% glycerol, 0.01%
Bromophenol Blue) containing 200 mm dithiothreitol.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.
3296 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
Samples were incubated overnight at room temperature and
separated on 10% SDS ⁄ PAGE gels. The electrophoresed
proteins were transferred to Immobilon P membranes and
detected with mouse anti-FLAGÒ M2 and alkaline phos-
phatase-conjugated goat anti-mouse [50].
M2-agarose affinity purification of whole
cell lysate
Infected cells (4.4 · 10
7
cells) were resuspended in 1 mL of
lysis buffer (50 mm Na
2
HPO

4
, 0.3 m KCl, pH 7.5, 1% Tri-
ton X-100), sonicated until solubilized, and centrifuged at
500 g for 15 min at 4 °C. M2-FLAGÒ resin (40 lL, 1 : 1
NaCl ⁄ P
i
:M2-agarose gel slurry) was added to the superna-
tant, and samples were subsequently rotated for 1 h at
room temperature and centrifuged at 425 g for 3 min at
4 °C. The resin was washed three times with 1 mL of high-
K
+
salt buffer and centrifuged under the previous condi-
tions. The resin was then washed three times with 1 mL of
NaCl ⁄ P
i
. An equal volume of 2· SDS ⁄ PAGE sample redu-
cing buffer (about 125 lL) was added, and samples were
then incubated overnight at room temperature. For Endo
H and PNGase F experiments, the resin was resuspended
in denaturing buffer (120 lL). Resuspended samples were
then denatured at 100 °C for 10 min. Denatured samples
were split, and appropriate amounts of supplied buffer
(10·) were added to each sample. Glycosylated proteins
were digested with 4 lL of Endo H or PNGase F at 37 °C
for 14–16 h, at which time 125 lLof2· reducing SDS sam-
ple buffer was added. In all cases, a control was run in par-
allel, which involved substituting the enzyme with 4 lLof
water. Samples digested with enzyme and control were
loaded side by side on 10% SDS gels and subjected to

electrophoresis. Electrophoresed proteins were separated,
transferred and probed as described in the above section.
Endo H, PNGase F and buffers were supplied by New
England BioLabs Inc., Ipswich, MA, USA.
Preparation of rat brain membranes
Adult rat brains were a kind gift from S. N. Pennington,
Department of Biochemistry and Molecular Biology, Brody
School of Medicine, East Carolina University. Brain tissue
was ground with a mortar and pestle under liquid nitrogen
into a fine powder and transferred to a sterile tube. The
brain powder was homogenized in 10 mL of lysis buffer
(4 mm Hepes, pH 7, 320 mm sucrose, 5 mm EDTA, pH 8,
protease inhibitor cocktail 1 : 500) and centrifuged at
2000 g for 10 min at 4 °C. The supernatant was transferred
to a clean tube and centrifuged at 100 000 g in a Sorvall
TH641 rotor for 1 h at 4 °C, whereas the pellet was resus-
pended in 10 mL of lysis buffer, homogenized, and centri-
fuged at 2000 g in an Eppendorf F-45-30-11 rotor for
10 min at 4 °C. This supernatant was transferred to a clean
tube and centrifuged at 100 000 g in a Sorvall TH641 rotor
for 1 h at 4 °C. Pellets from high-speed spins were resus-
pended in about 4 mL of lysis buffer. Protein concentration
was measured using a modified Lowry assay. A 200 lL
aliquot of crude brain membranes was removed and resus-
pended in 200 lLof2· reducing SDS sample buffer. Sam-
ples were heated to 80 °C for 3 min and resolved by SDS
gel and western blotting. For PNGase F digestions, 48 lL
of crude brain membranes was mixed with the following:
40 lLofH
2

O, 0.2 lL of protease inhibitor, and 10 lLof
denaturation buffer. This mixture was incubated at 100 °C
for 10 min. The following was added to this mixture: 11 lL
of 10% NP-40, 12 lL of G7 buffer, and 4 lL of PNGase F
(4 lLofH
2
O was used here for the control). The digestion
was incubated at 37 °C for 4–6 h. Next, 125 lLof2· redu-
cing SDS sample buffer was added, samples were heated at
80 °C for 3 min, and 10–20 lL was loaded per well
(25–50 lg per well) on 10% SDS gels.
Patch clamp recordings
For patch clamp studies, Sf9 cells were inoculated onto
35 mm culture dishes, containing small glass chips, for 24 h
at 27 °C and then kept at room temperature for the dur-
ation of the patch clamp experiments. Small glass chips
were removed from the dish and placed in the recording
chamber for measuring whole cell and single-channel cur-
rents, as needed. Whole cell and single-channel current
measurements were done in the whole cell and cell-attached
modes of the patch clamp technique, respectively. The bath
solution for whole cell recordings contained a low-K
+
solu-
tion (5 mm potassium aspartate, 135 mm sodium aspartate,
1mm MgCl
2
hexahydrate, 10 mm Mes, 60 mm mannitol,
pH 6.3, mOsm 315–335), and that for single-channel
recordings contained a high-K

+
solution (140 mm potas-
sium aspartate, 1 mm MgCl
2
hexahydrate, 10 mm Mes,
60 mm mannitol, pH 6.3, mOsm 315–335). The intracellular
K
+
solution (140 mm potassium aspartate, 10 mm EGTA,
5mm MgCl
2
hexahydrate, 10 mm Hepes, 40 mm mannitol,
pH 7.2, mOsm 320–340) was used to fill patch pipettes.
Patch pipettes were pulled with a Flaming–Brown micropi-
pette puller (Model P-97, Sutter Instrument Co., Novato,
CA, USA) from borosilicate glass tubing and fire-polished
with a microforge (MF-830, Narishige, McHenry, IL,
USA) before use. Pipette resistances ranged from 3 to
8MW for whole cell measurements and from 5 to 10 MW
for single-channel recordings. A silver ⁄ silver chloride wire
united the bath and the headstage through a salt bridge
(plastic pipette tip filled with 3% agarose in either 0.3 m
KCl or 1 m KCl), forming the reference electrode. Whole
cell currents were measured at room temperature with an
Axopatch 200B amplifier with a Digidata 1322 A interface
(Axon Instruments, Sunnyvale, CA, USA) with membrane
capacitance and resistance compensations of cells. Activa-
tion protocols involved evoking currents by depolarizing
voltages pulses (100 ms) from a holding potential of
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1

FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3297
) 50 mV to levels ranging from ) 40 to +100 mV in
10 mV increments. Whole cell current recordings used for
activation kinetic analysis had stable and tight seals
(‡ 500 MW), had maximum current amplitudes of
‡ 1100 pA, and showed minimal residual cell capacitance.
Additionally, the specific record from a cell that was selec-
ted for analysis was the one that displayed the largest cur-
rent amplitude at the various positive potentials. Whole
current recordings from cells expressing H
+
⁄ K
+
-ATPase
b-subunit, as well as uninfected cells, under similar record-
ing conditions produced current amplitudes of less than
100 pA at any given test potential. Other controls have pre-
viously been conducted for Sf9 cells using similar extracel-
lular and intracellular solutions, but different voltage
protocols in which current amplitudes were reported to be
less than 100 pA [13,32]. Deactivation protocols were con-
ducted by stepping to +40 mV (25 ms) and then stepping
from ) 110 mV to 0 mV (200 ms) in 10 mV increments
from a holding potential of ) 50 mV. Time constants for
deactivation were determined by fitting current at the var-
ious potentials with a single exponential. Single-channel
recordings were conducted using an Axopatch 200 A ampli-
fier with a Digidata 1200 interface [51]. The pclamp6.0 and
pclamp9.0 software suites were used for voltage pulse pro-
tocols, and data acquisition for single-channel and whole

cell recordings, respectively. The pclamp9.0 software suite
was used for data analysis. Data were filtered at 1 kHz and
subsequently sampled at 10 kHz. Group data are presented
as mean ± SE. origin7.5 was used for graphics and statis-
tical analysis.
Acknowledgements
We wish to thank Mr John Skaggs, a research techni-
cian in Dr Pennington’s laboratory, for providing
adult rat brains. This work was supported by a
National Institute of Health Grant DK52990 (to
RAS), and with resources and the use of facilities at
the Brody School of Medicine at East Carolina Uni-
versity.
References
1 Luneau CJ, Williams JB, Marshall J, Levitan ES, Oliva
C, Smith JS, Antanava
´
ge J, Folander K, Stein RB,
Swanson R et al. (1991) Alternative splicing contributes
to K
+
channel diversity in the mammalian central ner-
vous system. Proc Natl Acad Sci USA 88, 3932–3936.
2 Drewe JA, Verma S, Frech G & Joho RH (1992) Dis-
tinct spatial and temporal expression patterns of K
+
channel mRNAs from different subfamilies. J Neurosci
12, 538–548.
3 Grissmer S, Ghanshani S, Dethlefs B, McPherson JD,
Wasmuth JJ, Gutman GA, Cahalan MD & Chandy

KG (1992) The Shaw-related potassium channel gene,
Kv3.1, on human chromosome 11, encodes the Type 1
K
+
channel in T cells. J Biol Chem 267, 20971–20979.
4 Rettig J, Wunder F, Stocker M, Lichtinghagen R, Mas-
tiaux F, Beckh S, Kues W, Pedarzani P, Schro
¨
ter KH,
Ruppersberg JP et al. (1992) Characterization of a
Shaw-related potassium channel family in rat brain.
EMBO J 11, 2473–2486.
5 Dubois JM & Rouzaire-Dubois B (1993) Role of potas-
sium channels in mitogenesis. Prog Biophys Mol Biol 59,
1–21.
6 Joho RH, Ho CS & Marks GA (1999) Increased c- and
decreased d-oscillations in a mouse deficient for a potas-
sium channel expressed in fast-spiking interneurons.
J Neurophysiol 82, 1855–1864.
7 Rudy B & McBain CJ (2001) Kv3 channels: voltage-
gated K
+
channels designed for high-frequency repetit-
ive firing. Trends Neurosci 24, 517–526.
8 Yellen G (2002) The voltage-gated potassium channels
and their relatives. Nature 419, 35–42.
9 Jiang Y, Lee A, Chen J, Ruta V, Cadene M, Chait BT
& MacKinnon R (2003) X-ray structure of a voltage-
dependent K
+

channel. Nature 423, 33–41.
10 Bause E (1983) Structural requirements of N-glycosyla-
tion of proteins. Studies with proline peptides as con-
formational probes. Biochem J 209, 331–336.
11 Kornfeld R & Kornfeld S (1985) Assembly of aspara-
gine-linked oligosaccharides. Annu Rev Biochem 54,
631–664.
12 Rademacher TW, Parekh RB & Dwek RA (1988) Gly-
cobiology. Annu Rev Biochem 57, 785–838.
13 Schwalbe RA, Wang Z, Wible BA & Brown AM (1995)
Potassium channel structure and function as reported
by a single glycosylation sequon. J Biol Chem 270,
15336–15340.
14 Jaeken J & Carchon H (1993) The carbohydrate-defi-
cient glycoprotein syndromes: an overview. J Inherit
Metab Dis 16, 813–820.
15 Freeze HH & Aebi M (1999) Molecular basis of carbo-
hydrate-deficient glycoprotein syndromes type I with
normal phosphomannomutase activity. Biochim Biophys
Acta 1455, 167–178.
16 Jaeken J, Eggermont E & Stibler H (1987) An apparent
homozygous X-linked disorder with carbohydrate-defi-
cient serum glycoproteins. Lancet 2, 1398.
17 Orlean P (2000) Congenital disorders of glycosylation
caused by defects in mannose addition during N-linked
oligosaccharide assembly. J Clin Invest 105, 131–132.
18 Carchon H, Van Schaftingen E, Matthijs G & Jaeken J
(1999) Carbohydrate-deficient glycoprotein syndrome
type IA (phosphomannomutase-deficiency). Biochim
Biophys Acta 1455, 155–165.

19 Jaeken J, Stibler H & Hagberg B (1991) The carbohy-
drate-deficient glycoprotein syndrome. A new inherited
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.
3298 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS
multisystemic disease with severe nervous system invol-
vement. Acta Paediatr Scand Suppl 375, 1–71.
20 Lindholm D, Wootz H & Korhonen L (2006) ER stress
and neurodegenerative diseases. Cell Death Differ 13,
385–392.
21 Takahashi N (1977) Demonstration of a new amidase
acting on glycopeptides. Biochem Biophys Res Commun
76, 1194–1201.
22 Elbein AD (1981) The tunicamycins ) useful tools for
studies on glycoproteins. Trends Biochem Sci 6, 219–
222.
23 O’Reilly DR, Miller LK & Luckow VA (1994) Baculo-
virus Expression Vector: a Laboratory Manual. Oxford
University Press, New York.
24 Yang B, van Hoek AN & Verkman AS (1997) Very
high single channel water permeability of aquaporin-4
in baculovirus-infected insect cells and liposomes recon-
stituted with purified aquaporin-4. Biochemistry 36,
7625–7632.
25 Hu YK & Kaplan JH (2000) Site-directed chemical
labeling of extracellular loops in a membrane protein.
The topology of the Na,K-ATPase alpha-subunit. J Biol
Chem 275, 19185–19191.
26 Yokoyama S, Imoto K, Kawamura T, Higashida H,
Iwabe N, Miyata T & Numa S (1989) Potassium chan-
nels from NG108-15 neuroblastoma–glioma hybrid cells.

Primary structure and functional expression from
cDNAs. FEBS Lett 259, 37–42.
27 McCormack T, Vega-Saenz de Miera EC & Rudy B
(1990) Molecular cloning of a member of a third class
of Shaker-family K
+
channel genes in mammals. Proc
Natl Acad Sci USA 87, 5227–5231.
28 Taglialatela M, Vandongen AMJ, Drewe JA, Joho RH,
Brown AM & Kirsch GE (1991) Patterns of internal
and external tetraethylammonium block in four homo-
logous K
+
channels. Mol Pharmacol 40, 299–307.
29 Critz SD, Wible BA, Lopez HS & Brown AM (1993)
Stable expression and regulation of a rat brain K
+
channel. J Neurochem 60, 1175–1178.
30 Coetzee WA, Amarillo Y, Chiu J, Chow A, Lau D,
McCormack T, Moreno H, Nadal MS, Ozaita A,
Pountney D et al. (1999) Molecular diversity of K+
channels. Ann NY Acad Sci 868, 233–285.
31 Rudy B, Chow A, Lau D, Amarillo Y, Ozaita A,
Saganich M, Moreno H, Nadal MS, Hernandez-Pineda
R, Hernandez-Cruz A et al. (1999) Contributions of
Kv3 channels to neuronal excitability. Ann NY Acad Sci
868, 304–343.
32 Kanemasa T, Gan L, Perney TM, Wang L-Y &
Kaczmarek LK (1995) Electrophysiological and pharma-
cological characterization of a mammalian Shaw chan-

nel expressed in NIH 3T3 fibroblasts. J Neurophysiol
74, 207–217.
33 Wang L-Y, Gan L, Perney TM, Schwartz I & Kacz-
marek LK (1998) Activation of Kv3.1 channels in
neuronal spine-like structures may induce local potas-
sium ion depletion. Proc Natl Acad Sci USA 95, 1882–
1887.
34 Altmann F, Staudacher E, Wilson IB & Marz L (1999)
Insect cells as hosts for the expression of recombinant
glycoproteins. Glycoconj J 16, 109–123.
35 Weiser M, Bueno E, Sekirnjak C, Martone ME, Baker
H, Hillman D, Chen S, Thornhill W, Ellisman M &
Rudy B (1995) The potassium channel subunit KV3.1b
is localized to somatic and axonal membranes of specific
populations of CNS neurons. J Neurosci 15 , 4298–4314.
36 Du J, Zhang L, Weiser M, Rudy B & McBain CJ
(1996) Developmental expression and functional charac-
terization of the potassium-channel subunit Kv3.1b in
parvalbumin-containing interneurons of the rat hippo-
campus. J Neurosci 16, 506–518.
37 Gelband CH, Warth JD, Mason HS, Zhu M, Moore
JM, Kenyon JL, Horowitz B & Sumners C (1999)
Angiotensin II type 1 receptor-mediated inhibition of
K+ channel subunit kv2.2 in brain stem and hypotha-
lamic neurons. Circ Res 84, 352–359.
38 Landolt-Marticorena C & Reithmeier RA (1994) Aspar-
agine-linked oligosaccharides are localized to single
extracytosolic segments in multi-span membrane glyco-
proteins. Biochem J 302, 253–260.
39 Nilsson I & von Heijne G (1993) Determination of the

distance between the oligosaccharyltransferase active site
and the endoplasmic reticulum membrane. J Biol Chem
268, 5798–5801.
40 Santacruz-Toloza L, Huang Y, John SA & Papazian
DM (1994) Glycosylation of Shaker potassium channel
protein in insect cell culture and in Xenopus oocytes.
Biochemistry 33, 5607–5613.
41 Deal KK, Lovinger DM & Tamkun MM (1994) The
brain Kv1.1 potassium channel: in vitro and in vivo stu-
dies on subunit assembly and posttranslational proces-
sing. J Neurosci 14, 1666–1676.
42 Zhu J, Watanabe I, Poholek A, Koss M, Gomez B,
Yan C, Recio-Pinto E & Thornhill WB (2004) Allowed
N-glycosylation sites on the Kv1.2 potassium channel
S1–S2 linker: implications for linker secondary structure
and the glycosylation effect on channel function.
Biochem J 375, 769–775.
43 Watanabe I, Zhu J, Recio-Pinto E & Thornhill WB
(2004) Glycosylation affects the protein stability and cell
surface expression of Kv1.4 but not Kv1.1 potassium
channels. A pore region determinant dictates the effect of
glycosylation on trafficking. J Biol Chem 279, 8879–8885.
44 Philipson LH, Malayev A, Kuznetsov A, Chang C &
Nelson DJ (1993) Functional and biochemical charac-
terization of the human potassium channel Kv1.5 with a
transplanted carboxyl-terminal epitope in stable mam-
malian cell lines. Biochim Biophys Acta 1153, 111–121.
45 Watanabe I, Wang H-G, Sutachan JJ, Zhu J, Recio-
Pinto E & Thornhill WB (2003) Glycosylation affects
N. L. Brooks et al. Characterization of glycosylation sites in Kv3.1

FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS 3299
rat Kv1.1 potassium channel gating by a combined sur-
face potential and cooperative subunit interaction
mechanism. J Physiol 550, 51–66.
46 Thornhill WB, Wu MB, Jiang X, Wu X, Morgan PT &
Margiotta JF (1996) Expression of Kv1.1 delayed
rectifier potassium channels in Lec mutant Chinese
hamster ovary cell lines reveals a role for sialidation in
channel function. J Biol Chem 271, 19093–19098.
47 Hernandez-Pineda R, Chow A, Amarillo Y, Moreno H,
Saganich M, Vega-Saenz de Miera EC, Hernandez-Cruz
A & Rudy B (1999) Kv3.1–Kv3.2 channels underlie a
high-voltage-activating component of the delayed recti-
fier K+ current in projecting neurons from the globus
pallidus. J Neurophysiol 82, 1512–1528.
48 Ho SN, Hunt HD, Horton RM, Pullen JK & Pease LR
(1989) Site-directed mutagenesis by overlap extension
using the polymerase chain reaction. Gene 77, 51–59.
49 Sambrook J, Fritsch EF & Maniatis T (1992) Molecular
Cloning: a Laboratory Manual. Cold Spring Harbor
Laboratory, Cold Spring Harbor, NY.
50 Schwalbe RA, Rudin A, Xia S-L & Wingo CS (2002)
Site-directed glycosylation tagging of functional Kir2.1
reveals that the putative pore-forming segment is extra-
cellular. J Biol Chem 277, 24382–24389.
51 Schwalbe RA, Wingo CS & Xia S-L (2002) Mutations
in the putative pore-forming segment favor short-lived
wild-type Kir2.1 pore conformations. Biochemistry 41,
12457–12466.
Characterization of glycosylation sites in Kv3.1 N. L. Brooks et al.

3300 FEBS Journal 273 (2006) 3287–3300 ª 2006 The Authors Journal compilation ª 2006 FEBS