Electrostatic effects on ion selectivity and rectification in designed ion channel peptides

Article abstract of DOI:10.1021/ja9629672

To help determine how amino acid sequence can influence ionic conduction properties in α-helical structures, we have synthesized and studied three closely related, channel-forming peptides. The sequences are based on a 21-residue amphiphilic Leu-Ser-Ser-Leu-Leu-Ser-Leu heptad repeat motif and differ in having either neutral, negatively, or positively charged N-termini. The channels formed by the neutral peptide are modestly cation selective and exhibit asymmetric current-voltage curves arising from the partial charges at the ends of the α-helix. Addition of a negatively charged Glu residue converted the channel to a completely cation-selective structure and essentially eliminated its rectification. Addition of a positively charged Arg residue near the N-terminus of the peptide reduced this channel's cation selectivity and increased the extent of rectification. These effects on channel ionic conductance can be explained by a theoretical electrostatic model and provide insights into the workings of more complex channel proteins.

Full text of DOI:10.1021/ja9629672

3212
J. Am. Chem. Soc. 1997, 119, 3212-3217
Electrostatic Effects on Ion Selectivity and Rectification in
Designed Ion Channel Peptides
J. D. Lear,* J. P. Schneider, P. K. Kienker,† and W. F. DeGrado*
Contribution from the Johnson Research Foundation, Department of Biochemistry & Biophysics,
UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6059
ReceiVed August 23, 1996X
Abstract: To help determine how amino acid sequence can influence ionic conduction properties in R-helical structures,
we have synthesized and studied three closely related, channel-forming peptides. The sequences are based on a
21-residue amphiphilic Leu-Ser-Ser-Leu-Leu-Ser-Leu heptad repeat motif and differ in having either neutral, negatively,
or positively charged N-termini. The channels formed by the neutral peptide are modestly cation selective and
exhibit asymmetric current-voltage curves arising from the partial charges at the ends of the R-helix. Addition of
a negatively charged Glu residue converted the channel to a completely cation-selective structure and essentially
eliminated its rectification. Addition of a positively charged Arg residue near the N-terminus of the peptide reduced
this channel’s cation selectivity and increased the extent of rectification. These effects on channel ionic conductance
can be explained by a theoretical electrostatic model and provide insights into the workings of more complex channel
proteins.
Ligand-gated ion channels in the central nervous system have
properties,5 it is now possible to ask how the addition of a
charged residue will affect its ion selectivity and rectification
properties.
significant charge selectivity.1 Excitatory ligand (e.g., acetyl-
choline, glutamate) receptor channels are very selective for
positive ions and promote the generation of nerve impulses by
allowing extracellular cations (Na+, Ca2+) to cross the post-
synaptic membrane. Corresponding inhibitory ligand (e.g.,
γ-amino butyric acid, glycine) receptor channels counteract the
depolarizing effects of positive ion flux by selectively allowing
chloride ions to cross the membrane. Both classes of channels
are important targets in medicinal chemistry, and their structure-
property relationships are of great interest. However, the large
size and extreme complexity of such ion channel proteins make
it difficult to understand their mechanisms at the atomic level.
This problem can be reduced somewhat by studying the
properties of the pore-lining segments of natural ion channel
proteins2 or helix-forming antibiotic peptides.3 However, even
in these simplified systems, considerable complexity remains,
making interpretation difficult. Thus, upon inspection of an ion
channel sequence, it is currently not possible to predict a priori
the contribution of a given amino acid residue to the conductance
properties of the channel. Therefore, we have constructed a
“minimalist” peptide (Ac-(LSSLLSL)3; Ac ) acetyl; C-terminus
) CONH2) that contains the simplest features consistent with
formation of an amphiphilic, channel-forming R-helix.4 After
careful characterization of this peptide’s channel-forming
A wealth of information6 suggests that Ac-(LSSLLSL)3
associates into channels consisting of a bundle of approximately
six parallel, transmembrane helices with apolar Leu residues
(L) projecting toward the bilayer and polar Ser residues (S)
lining the pore. The channels have a moderate degree of
selectivity for cations relative to anions, presumably because
of the orientation of the hydroxyl groups of the seryl side chains
in the pore.7 Although the channels formed by Ac-(LSSLLSL)3
are formally neutral, R-helical peptides have partial charges of
approximately +0.5 and -0.5 near their N- and C-termini,
respectively, due to the alignment of their amide groups.8 This
asymmetric charge distribution strongly influences their channel-
forming and ion-conduction properties. For instance, the
frequency of formation of Ac-(LSSLLSL)3 channels increases
exponentially with the applied transmembrane voltage, which
helps to unidirectionally orient the peptides in a transmembrane
configuration conducive to forming a parallel helical bundle.
In the absence of a transmembrane voltage, these peptides prefer
to lie in the membrane with their helical axes parallel to the
membrane surface. However, the applied voltage aligns a
fraction of the helices in a unique transmembrane orientation.
For instance, if the peptide is added to one side of the bilayer
and the opposite side is held at positive potential, this will attract
the partial negative charge of the C-terminal carboxamide,
causing this end of the helix to cross the bilayer and assume a
transmembrane orientation.
* Authors to whom correspondences should be addressed.
† Department of Physiology & Biophysics, Albert Einstein College of
Medicine, Bronx, NY 10461.
X Abstract published in AdVance ACS Abstracts, March 15, 1997.
(1) Nicoll, R. A.; Malenka, R. C.; Kauer, J. A. Physiol. ReV. 1990, 70,
513-565.
The electrostatic asymmetry of a parallel helical bundle can
also account for the highly asymmetric, single channel current-
voltage curves (rectification) observed for Ac-(LSSLLSL)3.
Current-voltage curves for this peptide are generally obtained
by first applying a potential to orient the peptides in the bilayer
and initiate channel formation. Once a single channel is
observed, the voltage is rapidly (10 mV/ms) ramped from
approximately -200 mV to +200 mV to allow the measurement
of a current-voltage curve on a single channel. When the
lifetime of a single channel is long enough, this technique allows
current-voltage curves to be measured at both positive and
(2) Montal, M. Current Opin. Struct. Biol. 1995, 5, 501-506.
(3) Sansom, M. S. P. Prog. Biophys. Mol. Biol. 1991, 55, 139-235.
(4) DeGrado, W. F.; Wasserman, Z. R.; Lear, J. D. Science 1989, 243,
622-628.
(5) Kienker, P. K.; Lear, J. D. Biophys. J. 1995, 68, 1347-1358.
(6) A° kerfeldt, K. S.; Kienker, P. K.; Lear, J. D.; DeGrado, W. F. In
ComprehensiVe Supramolecular Chemistry; Reinhoudt, D. N., Ed.; Perga-
mon: London, 1996; Vol. 10, pp 659-686.
(7) A° kerfeldt, K. S.; Lear, J. D.; Wasserman, Z. R.; Chung, L. A.;
DeGrado, W. F. Acc. Chem. Res. 1993, 26, 191-197.
(8) Hol, W. G. J. Progr. Biophys. Mol. Biol. 1985, 45, 149-195.
S0002-7863(96)02967-8 CCC: $14.00 © 1997 American Chemical Society

Ion Channel Peptides
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3213
account for the modest cation selectivity observed for Ac-
(LSSLLSL)3. This is because a parallel bundle of Ac-
(LSSLLSL)3 helices has equal-but-opposite partial charges on
either side of the pore (Figure 1). As a result, the electrostatic
potential associated with a cation entering from one end of the
channel will always be equivalent to that experienced by an
anion entering on the other (oppositely charged) end of the
channel. Thus, the moderate charge selectivity observed for
this peptide appears to be related to the lining of the pore rather
than the partial charges at the ends of the R-helices.
In this paper we further test and extend this model by asking
how the addition of a formally charged residue to just one end
of the helix will affect the ion selectivity and rectification of
the Ac-(LSSLLSL)3 channel. To investigate this feature, two
peptides of sequence Ac-EW-(LSSLLSL)3 and Ac-RW-(LSS-
LLSL)3 were prepared, and the conductance properties of their
corresponding ion channels were characterized for comparison
to those of the Ac-(LSSLLSL)3 peptide. Addition of a
negatively charged Glu (E) residue at the N-terminus should
decrease the electrostatic asymmetry of the peptide, converting
it to a structure with terminal charges of -0.5 at both ends of
the helix (assuming the charge of the initial helix was 0.5 and
-0.5 at the N- and C-termini, respectively). In contrast, addition
of a positively charged Arg (R) at the N-terminus increases the
overall electrostatic asymmetry, giving a peptide with an
N-terminal charge of 1.5 and a C-terminal charge of -0.5. In
each peptide, the charged residue lies on the polar, Ser-
containing side of the R-helix, and the adjacent Trp (W) provides
a convenient spectroscopic tag. As in Ac-(LSSLLSL)3, the
C-terminus was blocked with a formally neutral carboxamide
to facilitate its insertion into bilayers for channel formation.9
The addition of a charged residue to just the N-terminus of
the helix should have a large effect on both rectification and
charge selectivity, because this change should affect the
conductance of ions flowing from the N-terminal portion of the
pore without significantly affecting conductance of ions from
the other, uncharged end. For instance, the addition of a
negative charge as in Ac-EW-(LSSLLSL)3 would be expected
to lead to an increase in cation-selectivity and a decrease in
rectification; now the electrostatic environment at the entry of
the channel should favor conductance of cations over anions
irrespective of the side from which they enter (Figure 1). Also,
the current-voltage asymmetry should be reduced in concert with
the decrease in the electrostatic asymmetry of the peptide. By
analogous reasoning, adding a positive charge to the N-terminus
as in Ac-RW-(LSSLLSL)3 should result in an increase of
current-voltage curve asymmetry and a decrease in the cation
selectivity.
Figure 1. Schematic diagram of the conduction of ions through the
Ac-(LSSLLSL)3 channel. For clarity, only two helices of the proposed
hexameric bundle are shown. δ+ and δ- refer to the partial charges
near the N- anc C-terminus of a formally neutral R-helix. The Ac-
(LSSLLSL)3 helices are preoriented in a parallel bundle with the
C-terminus of each helix positioned on the zero (ground) potential side
of the membrane by the preramp holding potential. (A) The current
observed when the applied potential is negative is comprised of two
distinct fluxes: the flow of cations from the channel C-terminus toward
the N-terminal side and the flow of anions from the N-terminus toward
the C-terminus. (B) When the applied voltage is changed quickly to
positive, the direction of both ion fluxes is reversed with respect to the
fixed channel orientation. The heights of the bars next to the schematic
channels are proportional to the total current (the sum of the cation
and anion fluxes) observed at (100 mV for the Ac-(LSSLLSL)3 peptide
channels in 1 M KCl.
negative potentials on a channel with a defined orientation.
Hundreds of channel openings are recorded and averaged to
improve the signal-to-noise ratio. Figure 1 illustrates how the
partial charges of the parallel helical bundle give rise to
rectification. At the beginning of a recording, the Ac-(LSS-
LLSL)3 helices are oriented with the N-terminus of each helix
positioned on the negative potential side of the membrane. The
overall current observed when the voltage is negative on the
N-terminal side is comprised of the flow of anions from the
channel N-terminus toward the C-terminal (ground potential)
side and the flow of cations from the C-terminus toward the
N-terminus. When the applied voltage is positive on the
N-terminal side the ions flow in the reverse directions. The
total current at positive potential is smaller than the current at
negative potential because at positive potential both cations and
anions must traverse an electrically repulsive vestibule as they
enter the N- and C-terminal sides of the channel, respectively.
Conversely, with an applied negative potential, both cations and
anions experience an electrostatic attraction as they enter the
channel at the C- and N-termini, respectively, so total current
is increased. Thus, for a channel such as this, in which ion
entry appears to be rate-determining, the electrostatic environ-
ment of the mouth of the channel can have a very large effect
on conductance.
Results
Peptides. The peptides were prepared by segment condensa-
tion (Experimental Section), providing compounds that could
(9) The parallel helix bundle model for the Ac-(LSSLLSL)3 peptide
channels is based largely on the observed dependence of channel open
probability on transmembrane voltage and on the sign of the voltage
dependence relative to the helical dipole. The Ac-EW- peptide, with an
ionized Glu side chain, appears to have no net dipole moment so we cannot
infer the orientation from the sign of the applied potential alone. However,
we have observed that peptides with a single N-terminally charged residue
(including Ac-RW-(LSSLLSL)3 and H2N-(LSSLLSL)3) appear to form
channels by preferentially inserting their formally uncharged C-terminal
carboxamide into the bilayer. We speculate that this is because insertion of
the more polar charged group would have a much higher energetic barrier
to overcome. Thus when these peptides are added to only one side of a
pre-formed bilayer, channels are formed at significantly greater rates when
the potential is positive on the side opposite to that of peptide addition
(trans-positive) as opposed to a trans-negative potential. The trans-positive
potential presumably attracts the partial negative charge of the peptide helix
C-terminus. Similar data are observed for Ac-EW-(LSSLLSL)3, suggesting
that it also inserts its formally neutral C-terminus in preference to its
N-terminus which contains the formally charged Glu residue.
It is interesting that although the partial charges on the helices
can account for the observed current rectification, they cannot

3214 J. Am. Chem. Soc., Vol. 119, No. 14, 1997
Lear et al.
at negative potential compared to the Ac-(LSSLLSL)3 channel,
attributable to parallel changes in the flux of cations and anions,
respectively. Conversely, the Ac-RW-(LSSLLSL)3 channel
shows the opposite behavior as would be predicted from the
qualitative model. A more quantitative analysis of the curves
will be discussed below.
Charge Selectivity. The charge selectivity of an ion channel
is generally evaluated by measuring its conductance with
different concentrations of the same electrolyte on either side
of the bilayer. In the absence of an applied potential, the ions
tend to move through the channel down their concentration
gradients. If the channel is nonselective (i.e., it does not
kinetically discriminate between anions versus cations), no net
current is observed in the absence of an applied potential,
because the oppositely charged ions move through the channel
at equal rates. On the other hand, if the channel shows charge
selectivity, a net electrical conductance is observed. Conven-
tionally, the selectivity is measured by determining the “reversal
potential” (Erev), which is the transmembrane electrical potential
required to achieve zero current with a given asymmetric salt
concentration. If the channel is nonselective the reversal
potential is zero, whereas if the channel is perfectly selective
the reversal potential will be equal to the difference in the
chemical potential of the permeant ion on the two sides of the
bilayer. More generally, the ionic selectivity of a channel is
frequently defined by the ion permeability ratio (in this case,
PK/PCl) calculated by fitting reversal potentials to the Goldman-
Hodgkin-Katz eq 1
Figure 2. Current-voltage curves for Ac-(LSSLLSL)3 (circles), Ac-
EW-(LSSLLSL)3 (squares), and Ac-RW-(LSSLLSL)3 (triangles). The
lines represent theoretical curves generated from eq 2 fit to the weighted
data at ∼1 mV intervals. Data points are shown at wider voltage
intervals for clarity.
readily be purified to homogeneity. The Trp residue provides
a convenient fluorescent tag for monitoring incorporation of the
peptides into membranes and also to evaluate their position in
the bilayer. An emission maximum of 332 nm was observed
for Ac-EW-(LSSLLSL)3, consistent with the value obtained
from earlier work10 on Trp-substituted derivatives of (LSS-
LLSL)3 suggesting that Ac-EW-(LSSLLSL)3 binds in an
orientation analogous to (LSSLLSL)3.
Comparison of Ac-(LSSLLSL)3, Ac-EW-(LSSLLSL)3, and
Ac-RW-(LSSLLSL)3 Peptide Channels. Each of these pep-
tides in 1 M KCl, formed channels in planar bilayers with
lifetimes sufficiently long to allow current-voltage curves to be
measured on single channels. At 1 M salt concentration, the
PK [aCl]V - [aCl]0 exp(FErev/RT)
)
(1)
PCl
[aK]V exp(FErev/RT) - [aK]0
quality of the single channel recordings was similar to that
where the a’s denote thermodynamic activities on the applied
potential (V) and ground (0) sides of the membrane, F is
Faraday’s constant, and RT is thermal energy.12
6
previously described for Ac-(LSSLLSL)3
.
At lower salt
concentrations, it was difficult to measure single channel
properties of the Ac-RW-(LSSLLSL)3 channel, possibly because
charge repulsion between the Arg side chains led to low channel
lifetime or membrane disruption. In 1 M KCl, the mean lifetime
of channels formed from Ac-(LSSLLSL)3 is approximately 200
ms;11 Ac-EW-(LSSLLSL)3 has a shorter lifetime (∼100 ms),
and the lifetime of Ac-RW-(LSSLLSL)3 is also shorter (∼50
ms). The current-voltage curves for the neutral Ac-(LSSLLSL)3,
the negatively charged Ac-EW-(LSSLLSL)3, and the positively
charged Ac-RW-(LSSLLSL)3 peptide channels show that the
N-terminal charge indeed has a large influence on current-
voltage curve asymmetry (Figure 2). The Ac-(LSSLLSL)3
channel is about 2-fold more conductive at -125 mV than at
+125 mV, indicative of its inherent electrostatic asymmetry as
described above. Addition of an Arg to the N-terminus (Ac-
RW-(LSSLLSL)3) markedly increases the rectification; at -125
mV this channel is 4- to 6-fold more conductive than at +125
mV. On the other hand, addition of a negatively charged Glu
to the N-terminus (Ac-EW-(LSSLLSL)3) decreases the rectifying
behavior; this channel is nearly as conductive at +125 mV as
it is at -125 mV. Further, the degree of rectification for this
peptide is pH-dependent. At pH 4.5, near the intrinsic pKa of
the Glu carboxylate, the rectification falls to about half that
observed at pH 9. At pH 2, it becomes the same as that
observed for the Ac-(LSSLLSL)3 channel.
For Ac-EW-(LSSLLSL)3 (Figure 3a), dilution of the KCl
electrolyte on the ground side of the bilayer (corresponding to
the C-terminal end of the channel;9 Figure 1) reduces the current
at negative potential and shifts the reversal potential toward more
negative values. This is indicative of cation selectivity, as
previously observed for the uncharged Ac-(LSSLLSL)3 chan-
nel.5 However, the degree of the shift with dilution (Figure
3c) is significantly greater for the Ac-EW-(LSSLLSL)3 channel.
The calculated value of the cation selectivity (PK/PCl) for the
Ac-EW-(LSSLLSL)3 channel approaches infinity, indicative of
a perfectly cation selective channel. By comparison, PK/PCl
for the Ac-(LSSLLSL)3 channel under these experimental
conditions is approximately 10.
In contrast to the cation selectivity observed for the uncharged
and negatively charged peptides, the Ac-RW-(LSSLLSL)3
channel (PK/PCl ) 1.1) displays reduced currents at both positive
and negative potentials upon dilution of the KCl on the
C-terminal side of the channel (Figure 3b), and there is very
little shift in the reversal potential (Figure 3c). Both of these
features indicate that this channel conducts cations and anions
about equally well over the range of conditions examined. Thus,
the sign of the N-terminal charge can predictably modulate the
modest intrinsic charge-selectivity of Ac-(LSSLLSL)3 to make
it either highly cation selective or non-selective. These changes
are easily rationalized according to the model in Figure 1.
These results are in excellent qualitative agreement with our
model. In particular, the Ac-EW-(LSSLLSL)3 channel shows
an increase in total current at positive potential and a decrease
(12) Hille, B. Ionic Channels of Excitable Membranes, 2nd ed.; Sinauer
Associates, Inc.: Sunderland, MA, 1992.
(10) Chung, L. A.; Lear, J. D.; DeGrado, W. F. Biochemistry 1992, 31,
6608-6616.
(11) Kienker, P. K.; DeGrado, W. F.; Lear, J. D. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 4859-4863.
(13) Parsegian, V. A. Ann. N. Y. Acad. Sci. 1975, 264, 161-174.
(14) The effect on the curve-fitting results of adding 3 Å to the channel
length is small. For example, using L ) 30 Å instead of L ) 33 Å in fitting
the Ac-EW-(LSSLLSL)3 data increased the fitted value of Zn by only 4%.

Ion Channel Peptides
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3215
thermal energy, and L is the length of the pore (estimated as 30
Å for the Ac-(LSSLLSL)3 channel). The potential term, φ(x,V)
(eq 3), describes the variation in electrical potential with respect
to the displacement (x) through the channel as a sum of three
contributions:
φ(x,V) ) φapp(x,V) + 180[φec(x) + φimage(x)]
(3)
The center of the channel is taken to be at x ) 0, and the factor
of 180 mV/Å gives φ(x,V) in mV when x, a, and L are expressed
in Å units. The individual electrical potentials in eq 3 arise
from three sources shown diagramatically in Figure 4a. Equa-
tion 3a gives the portion of the total variation in the electrical
potential with respect to displacement through the channel that
is associated with the applied transmembrane voltage:
φapp(x,V) ) V(x + L/2)/L
(3a)
Equation 3b describes the contribution arising from the channel’s
end-charges:
φec(x) ) 6[zc[a2 + (x + L/2)2]-0.5 + zn[a2 + (x - L/2)2]-0.5
]
(3b)
The terms zc and zn are the values in electron charge units of
each of six noninteracting charges at each of the two ends of
the pore (zc at x ) -L/2, zn at x ) +L/2) arranged radially at
a distance of a ) 4 Å from the cylindrical axis. Equation 3c
describes the repulsive potential associated with the ion passing
from water into a region bounded by a lower dielectric constant
medium (the image charge13)
φimage(x) ) {f(K)/a - ln[2/(1 + K)]/KL} exp(-x2/2σ2) (3c)
in which K ) ꢀw/ꢀp, ꢀw ) wall dielectric constant, ꢀp ) pore
dielectric constant, f(K) is a tabulated function13 describing the
dependence of the induced image charge on the dielectric ratio
(K), and σ describes the range of the potential in Å. These
parameters were held constant at values found previously to
provide excellent fits to the Ac-(LSSLLSL)3 channel’s current-
voltage curves in 1 M KCl. Those values were ꢀw ) 12 and ꢀp
) 80, giving f(K)) 1.47, and σ ) 12.4 Å.11
Equation 2 was used to fit each set of current-voltage data in
Figure 2. The effective N-terminal charges (zn) and P were
the only fitting variables.14 The computed potentials due to the
end charges for each of the three peptides are shown in Figure
4a-c. The determined values of zn were +0.11, -0.10, and
+0.35 for Ac-(LSSLLSL)3, Ac-EW-(LSSLLSL)3, and Ac-RW-
(LSSLLSL)3, respectively, while the values of P were nearly
identical among the three peptides (1.4 ( 0.1 × 10-5 cm2/s).
The interpretation of the values of zn in terms of the actual
N-terminal charges requires knowledge of the electrolyte
screening factor. Because the two N-terminally charged pep-
tides differ from Ac-(LSSLLSL)3 by the addition of either a
negative or positive unit charge to the N-terminus, it is possible
to determine both the screening factor and the effective end
charge of the parent peptide directly from the experimentally
determined values of zn. The effective N-terminal charge zn
on each of the peptides is
Figure 3. Representative current-voltage curves for asymmetric KCl
concentrations for channels formed by (A) Ac-EW-(LSSLLSL)3 and
(B) Ac-RW-(LSSLLSL)3. Successive dilutions were made of the
initially symmetric 1 M KCl electrolyte on the zero (ground) holding
potential side of the bilayer membrane using KCl-free buffer. The ratios
of KCl activities (calculated from the measured solution conductivities
and standard data tables) on the ground- vs the holding-potential side
were (A) i ) 1.00, ii ) 0.289, iii ) 0.0323 and (B) i ) 1.00, ii )
0.53, iii ) 0.122. Panel C shows reversal potentials, determined by
fifth order polynomial interpolation of current-voltage data from
experiments such as (and including) those in panels A and B and
corrected for liquid junction potentials. Values for Ac-EW-(LSSLLSL)3
(circles) and Ac-RW-(LSSLLSL)3 (squares) are plotted against the base
10 logarithm of the activity ratio. Previously published data for Ac-
(LSSLLSL)3 (triangles) are included for comparison.
Theoretical Analysis of Current-Voltage Curves. The
current-voltage curves in Figure 2 were quantitatively analyzed
to determine whether the magnitude of the differences between
the conductances of the three peptides were consistent with their
N-terminal end charges. This approach has also allowed us to
estimate the values of the partial charges at the ends of the
R-helix. The curves were analyzed according to an elementary
continuum electrodiffusion model in which ions diffuse in a
nonlinear electrical field according to eq 2,12 which was
previously employed to describe the Ac-(LSSLLSL)3 channel11
PC[exp(VF/RT) - 1]
I(V) ) πa2F
(2)
zn(Ac-(LSSLLSL)3) ) S[zR] ) 0.11
zn(Ac-EW-(LSSLLSL)3) ) S[zR - 1] ) -0.10 (5)
zn(Ac-RW-(LSSLLSL)3) ) S[zR + 1] ) 0.35 (6)
(4)
+L/2
exp[φ(x,V)F/RT]dx
∫
-L/2
in which I is the current, V is the applied voltage, a is the pore
radius (estimated as 4 Å for the Ac-(LSSLLSL)3 channel), F is
Faraday’s constant, P is the sum of cation and anion perme-
abilities in the pore, C is the electrolyte solution activity, RT is
in which S is the screening factor, zR is the partial N-terminal
charge arising from the orientation of the amides in the R-helix,

3216 J. Am. Chem. Soc., Vol. 119, No. 14, 1997
Lear et al.
rectification and increased cation selectivity of this particular
channel. The same argument can be applied to deduce the
underlying cause of the increased rectification and decreased
cation selectivity in the positively charged peptide channel.
The quantitative fitting of data using the electrodiffusion
model gave parameters consistent with the physical model. The
assumption of a parallel, hexamer bundle of R-helices, together
with the total end charge values calculated from the curve fits,
yields a value of 0.54 for the effective N-terminal charge of
the R-helix, in excellent agreement with accepted values.8,15,16
The screening factor calculated from these data (S ) 0.23)
indicates somewhat more attenuation of charge effects than
would be calculated (S ) 0.36) from simple Debye-Hu¨ckel
theory for 1 M ionic strength and a 4 Å distance, but this is
hardly surprising given the complexity of the ionic atmosphere
near the entrance of the channel.
This electrostatic treatment provides remarkably good agree-
ment with the data, considering the small number of adjustable
parameters. Moreover, the fitted values of the parameters are
physically quite reasonable, especially the values of the intrinsic
partial end charges for the peptide helix. This encourages us
to accept the gross features of the structural model involving a
parallel bundle of R-helices. However, more detailed theories
are clearly required to quantitate the current-voltage curves
measured using asymmetric salt solutions.17 Also, electrostatic
theories based on the actual atomic coordinates of the peptide
will be needed to further determine whether a six-stranded coiled
coil geometry is uniquely consistent with our results or whether
five-and/or seven-stranded structures might also account for the
data.
The experimental results here, along with their simple
interpretation, complement and deepen our understanding of
studies involving mutants of large ion channel proteins. Site-
directed mutagenesis can help to define which of the many
charged residues present in a channel are important for ion
permeation. For instance, in the acetylcholine and GABA
receptors, site directed mutagenesis has been used to show which
of a number of potentially important groups actually influence
ion conduction under physiological conditions.18-20 However,
without high resolution structures, it is very difficult to interpret
these results in terms of detailed geometric models. The results
of the model studies presented here should be relevant to the
interpretation of natural ion channel mutation studies. For
example, the uncharged channel’s cation selectivity suggests a
similar bias might be found to underlie the selectivity observed
in natural channels. Perhaps this is why the cation selectivity
of the acetylcholine receptor is significantly greater than the
anion selectivity of the GABA receptor.21 Also, the large
changes in rectification and selectivity we observed due to
changing our model peptide’s end charges should be magnified
under physiological salt concentration (approximately 0.15 M)
where the end charges would be less well screened. Thus, our
results support and help quantify previous suggestions22 that
end charges play a very important role in modulating the charge
selectivity of excitatory and inhibitory receptor ion channels.
Figure 4. Calculated potential profiles for each of the three peptidic
ion channels in this study: (A) Ac-(LSSLLSL)3, (B) Ac-EW-(LSS-
LLSL)3, and (C) Ac-RW-(LSSLLSL)3. Each panel shows the total
potential (heavy solid line) for a permeating cation at -50 mV applied
transmembrane voltage together with lighter lines showing the isolated
contributions of the applied potential (dots), the helix amide end charges
(solid), and the repulsive image potential (dashes). The image potential
is the same for all ions in each channel, and the other potentials differ
only in sign for cations (positive) and anions (negative).
and -1 and +1 in eqs 5 and 6 refer to the unit charge on the
N-terminal Glu and Arg residue, respectively. Solution of these
eqs gives S ) 0.23 ( 0.02 and zR ) 0.54 ( 0.02.
Discussion
These results show how charged residues near the mouth of
an ion channel can significantly affect its ionic conductance.
The neutral Ac-(LSSLLSL)3 channel shows a moderate degree
of rectification arising from the helical dipole as well as a
moderate degree of cation selectivity postulated to arise from
the preferential orientation of the Ser side chains within the
pore.6 Formally charged residues can modulate these properties,
giving rise to increases or decreases in both rectification and
selectivity. In the model channels studied here, the ion flux is
largely determined by ion entry rates (the current in these
channels is approximately proportional to the electrolyte solution
activity between 0.1 and 1 M KCl). Consequently, addition of
the N-terminal Glu should increase current due to cations at
positive potential and decrease current due to anions at negative
potential. Both of these effects contribute to the decreased
(15) Honig, B. H.; Hubbell, W. L.; Flewelling, R. F. Ann. ReV. Biophysics
Biophys. Chem. 1986, 15, 163-193.
(16) Lockhart, D. J.; Kim, P. S. Science 1993, 260, 198-202.
(17) Chen, D. P.; Lear, J. D.; Eisenberg, R. S. Biophys. J. 1997, 72,
97-116.
(18) Imoto, K.; Busch, C.; Sakmann, B.; Mishina, M.; Konno, T.; Nakai,
J.; Bujo, H.; Mori, Y.; Fukuda, K.; Numa, S. Nature 1988, 335, 645-648.
(19) Kienker, P.; Tomaselli, G.; Jurman, M.; Yellen, G. Biophys. J. 1994,
66, 325-334.
(20) Galzi, J.-L.; Devillers-Thie´ry, A.; Hussy, N.; Bertrand, S.; Changeux,
J.-P.; Bertrand, D. Nature 1992, 359, 500-505.
(21) Franciolini, F.; Petris, A. Biochim. Biophys. Acta 1992, 1113, 1-11.
(22) Green, W. N.; Andersen, O. S. Annu. ReV. Physiol. 1991, 53, 341-
359.

Ion Channel Peptides
J. Am. Chem. Soc., Vol. 119, No. 14, 1997 3217
Ac-Glu(t-Bu)-Trp-OPfp (3). A dry 50 mL, round-bottomed flask
was charged with pentafluorophenol (0.74 g, 4.02 mmol), and then a
solution of 2 (1.65 g, 3.82 mmol in 20 mL anhydrous CH2Cl2) was
added via syringe. The solution was stirred on an ice bath and EDC
(0.77 g, 4.02 mmol) was added at once under a blanket of N2. The
reaction was stirred for 20 min, then allowed to warm to room
temperature, stirred for an additional 45 min, transferred to a separatory
funnel, washed with 5% HCl (3 × 15 mL) and water (1 × 15 mL),
and dried (Na2SO4). The CH2Cl2 was evaporated affording a yellow
solid which was flash chromatographed on silica (ethyl acetate/hexane,
9:1) yielding 1.33 g (2.23 mmol, 58%) of 3 as a yellow solid. Mass
spectrum (FAB) m/z 598.1974 [(M + H)+, calcd for C28H29N3O6F5,
598.1977]; analytical C18 HPLC employing a linear gradient from 20%
to 80% solvent B over 30 min, r.t. ) 26.4 min.
Ac-Glu-Trp-OPfp (4). A dry 25 mL, round-bottomed flask was
charged with 3 (0.5 g, 0.84 mmol), and then 2.7 mL of 10% thioanisole
in TFA at 0 °C was added via syringe. The reaction was stirred on an
ice bath for 1.5 h after which time, the TFA was evaporated by passing
a stream of N2 over the cold solution. The resulting oil was triturated
with cold diethyl ether affording 259 mg (0.48 mmol, 57%) of 4 as a
light brown solid. Mass spectrum (FAB) m/z 542.1367 [(M + H)+,
calcd for C24H21N3O6F5, 542.1351]; analytical C18 HPLC employing a
linear gradient from 20% to 80% solvent B over 30 min, r.t. ) 19.5
min.
Experimental Section
General Methods and Material. All the reactions were carried
out under a dry N2 atmosphere. Ethyl acetate, acetonitrile, chloroform
(CHCl3), acetic acid, hexane, diethyl ether, anhydrous dichloromethane
(CH2Cl2), anhydrous dimethylformamide (DMF), anhydrous methanol,
N-hydroxysuccinimide, dicyclohexylcarbodiimide (DCC), diisopropyl-
ethylamine (DIEA), acetic anhydride, 10% Pd on activated carbon (Pd/
C), pentafluorophenol, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide
hydrochloride (EDC), thioanisole, and the trifluoroacetic acid (TFA)
were purchased from Aldrich and used without further purification.
Z-Glu(t-Bu)-OH and tryptophan were purchased from Advanced
ChemTech and used without further purification. Mass determinations
were carried out on a VG 70-VSE high-resolution mass spectrometer.
Analytical HPLC was performed employing an HP-1090 series system
equipped with either a Vydac C18 peptide/protein column (4.6 mm ID
× 25 cm L, 5 µm) or a Hamilton PRP-1 column (4.1 mm ID × 25 cm
L, 10 µm). Solvent A was composed of water and 0.1% TFA and,
unless stated otherwise solvent B consisted of 90% acetonitrile, 10%
water, and 0.1% TFA.
Peptides. Ac-(LSSLLSL)3 was prepared as described previously,11
and Ac-RW-(LSSLLSL)3 was prepared using standard stepwise solid
phase methods.10 Ac-EW-(LSSLLSL)3 was prepared by coupling the
preformed dipeptide fragment, Ac-Glu-Trp-OPfp (4), to (LSSLLSL)3.
Ac-EW-(LSSLLSL)3-CONH2. A dry vial was charged with H2N-
(LSSLLSL)3-CONH2 (8.0 mg, 3.7 µ mol), 4 (6.0 mg, 11.1 µ mol), and
140 µL of anhydrous DMF. The solution was stirred on an ice bath
and DIEA (0.7 µL, 4.07 µmol) was added via syringe. The reaction
was stirred for 1 h, then allowed to warm to room temperature, and
stirred for an additional 4 h after at which time cold diethyl ether was
added to the solution resulting in the formation of a precipitate. The
solid was collected by filtration and purified by analytical PRP-1 HPLC
employing isocratic conditions of 80% B (solvent B is composed of
60% acetonitrile, 30% isopropyl alcohol, 10% water, and 0.1% TFA).
The yield as determined by HPLC was nearly quantitative; a small
sample (200 µg) was purified for ion channel measurements. Mass
spectrum (ESI) 2513.6 [(M + H)+, calcd 2514.4].
Fluorescence Measurements. The peptide’s tryptophan emission
spectrum was determined using a Spex Fluorolog instrument. Excita-
tion was at 280 nm in 1 cm quartz cells. Spectra were determined at
5 µM peptide concentration in pH 7.6, 10 mM phosphate buffer
containing POPC lipid vesicles at 150 µM. The emission spectrum,
corrected for instrument wavelength response and buffer plus vesicle
scattering background, showed a maximum at 332 nm, consistent with
the Trp being located in a moderately apolar environment on the
hydrophobic face of a membrane-associated peptide R-helix.10
Ion Channel Measurements. Peptides were incorporated from
methanolic solutions into diphytanoyl phosphatidylcholine planar
bilayers and single channel current-voltage (I-V) curves were deter-
mined as previously described.5 The electrolyte contained 1 M KCl,
0.2 mM EDTA, and 5 mM HEPES at pH 7.2 (Ac-(LSSLLSL)3, Ac-
RW-(LSSLLSL)3), or 25 mM CHES at pH 9.2 (Ac-EW-(LSSLLSL)3).9
Data points for current-voltage curves were determined by averaging
baseline-subtracted currents measured using 10 V/s voltage ramps
triggered by single channel openings. Only channels whose current
(at the negative holding voltage) lay within the dominant open current
histogram peak were used for averaging. At least 20 ramps were
averaged for each peptide. Data (weighted using standard deviations
of currents at each voltage point) were fit using the program MLAB
(Civilized Software, Bethesda MD) to a previously described fitting
function11 based on a parallel hexamer bundle model of the channel
with ion permeation described by a mean-potential electrodiffusion
model12 and parameters determined as described in the Results section.
Potentials measured in asymmetric solutions were corrected for liquid
junction potentials as described previously.5 No corrections for osmotic
pressure differences were made because previous measurements for
the Ac-(LSSLLSL)3 channel5 showed this effect to be negligible.
Briefly, the synthesis of 4 begins with coupling tryptophan to Z-Glu-
(t-Bu) via the succinimide ester affording Z-Glu(t-Bu)-Trp-OH (1) in
96% yield. Hydrogenation of 1 in the presence of acetic anhydride
yields Ac-Glu(t-Bu)-Trp-OH (2) in 77% yield. Ac-Glu(t-Bu)-Trp-OPfp
(3) was synthesized in 58% yield by reacting 2 with pentafluorophenol
employing EDC activation. Finally, the tert-butyl protecting group of
3 was removed with 10% thioanisole in TFA affording 4 in 57% yield.
Z-Glu(t-Bu)-Trp-OH (1). A dry 100 mL, round-bottomed flask
was charged with Z-Glu(t-Bu)-OH (2 g, 5.93 mmol), N-hydroxysuc-
cinimide (683 mg, 5.93 mmol), and 12 mL of anhydrous CH2Cl2. This
solution was stirred on an ice bath, and DCC (1.23 g, 5.93 mmol) was
added under a blanket of N2. The reaction was stirred for 20 min and
then allowed to warm to room temperature and stirred for an additional
4 h. The reaction was filtered to remove the urea byproduct and
subsequently evaporated to 1/5 its original volume The reaction solution
was then cooled to 4 °C, allowed to sit for 2 h to precipitate the
remaining urea byproduct, filtered, and evaporated to yield Z-Glu(t-
Bu) succinimide ester as a white solid, which was dried under high
vacuum. The Z-Glu(t-Bu) succinimide ester was then dissolved in 24
mL of anhydrous DMF and tryptophan (1.33 g, 6.52 mmol) was added
at once under a blanket of N2. This mixture was stirred on an ice
bath, and DIEA (1.03 mL, 5.93 mmol) was added via syringe. After
20 min, the reaction was allowed to warm to room temperature, stirred
for an additional 14 h, filtered, transferred to a separatory funnel, and
partitioned between 10 mL of 5% HCl and 15 mL of ethyl acetate.
The organic phase was collected, and the aqueous phase was extracted
with ethyl acetate (1 × 15 mL). The combined ethyl acetate extracts
were washed with 5% HCl (3 × 10 mL), water (1 × 10 mL), dried
(Na2SO4), and evaporated to yield an oil which, when evaporated from
diethyl ether, yielded 2.97 g (5.67 mmol, 96%) of 1 as a white foam
after drying under high vacuum. Mass spectrum (FAB) m/z 524.2392
[(M + H)+, calcd for C28H34N3O7, 524.2397]; analytical C18 HPLC
employing a linear gradient from 20% to 80% solvent B over 30 min,
r.t. ) 22.0 min.
Ac-Glu(t-Bu)-Trp-OH (2). A dry 200 mL, round-bottomed flask
was charged with 1 (2.75 g, 5.26 mmol), 73 mL of anhydrous methanol,
acetic anhydride (0.65 mL, 6.83 mmol), and 10% Pd/C (0.29 g). The
N2 atmosphere was replaced with a continuous flow of H2 bubbled
through the reaction mixture. After stirring for 3 h, the Pd/C was
removed by filtration and washed with warm methanol (2 × 20 mL).
The methanol was evaporated yielding a pink oil which was flash
chromatographed on silica (CHCl3/methanol/acetic acid, 8.5:1.5:1)
affording 1.75 g (4.05 mmol, 77%) of 2 as a white solid after drying
on high vacuum. Mass spectrum (FAB) m/z 432.2112 [(M + H)+,
calcd for C22H30N3O6, 432.2135]; analytical C18 HPLC employing a
linear gradient from 20% to 80% solvent B over 30 min, r.t. ) 13.8
min.
Acknowledgment. This work was supported by the United
States Office of Naval Research.
JA9629672