a-Conotoxin analogs with additional positive charge show
increased selectivity towards Torpedo californica and
some neuronal subtypes of nicotinic acetylcholine
receptors
Igor E. Kasheverov
1
, Maxim N. Zhmak
1
, Catherine A. Vulfius
2
, Elena V. Gorbacheva
2
,
Dmitry Y. Mordvintsev
1
, Yuri N. Utkin
1
, Rene
´van Elk
3
, August B. Smit
3
and Victor I. Tsetlin
1
1 Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, Moscow, Russia
2 Institute of Cell Biophysics, Russian Academy of Sciences, Pushchino, Russia
3 Department of Molecular and Cellular Neurobiology, Center for Neurogenomics and Cognitive Research, Vrije Universiteit, Amsterdam,
the Netherlands
a-Conotoxins are a group of relatively short peptides
(12–19 amino acid residues, two disulfide bridges) from
the venom of poisonous marine snails of the Conus
genus [1]. In addition to peptides isolated from venom
new a-conotoxins have recently been identified by
cDNA cloning from venomous glands and have been
Keywords
acetylcholine-binding protein; acetylcholine-
elicited Cl
–
current; a-conotoxin analogs;
identified Lymnaea neurons; nicotinic
acetylcholine receptor
Correspondence
V. I. Tsetlin, Shemyakin-Ovchinnikov
Institute of Bioorganic Chemistry, Russian
Academy of Sciences, Miklukho-Maklaya
str. 16 ⁄10 Moscow, Russia
Tel ⁄Fax: +7 495 335 57 33
E-mail: vits@ibch.ru
(Received 28 March 2006, revised 16 June
2006, accepted 4 August 2006)
doi:10.1111/j.1742-4658.2006.05453.x
a-Conotoxins from Conus snails are indispensable tools for distinguishing
various subtypes of nicotinic acetylcholine receptors (nAChRs), and synthe-
sis of a-conotoxin analogs may yield novel antagonists of higher potency
and selectivity. We incorporated additional positive charges into a-conotox-
ins and analyzed their binding to nAChRs. Introduction of Arg or Lys res-
idues instead of Ser12 in a-conotoxins GI and SI, or D12K substitution in
a-conotoxin SIA increased the affinity for both the high- and low-affinity
sites in membrane-bound Torpedo californica nAChR. The effect was most
pronounced for [D12K]SIA with 30- and 200-fold enhancement for the
respective sites, resulting in the most potent a-conotoxin blocker of the
Torpedo nAChR among those tested. Similarly, D14K substitution in
a-conotoxin [A10L]PnIA, a blocker of neuronal a7 nAChR, was previously
shown to increase the affinity for this receptor and endowed
[A10L,D14K]PnIA with the capacity to distinguish between acetylcholine-
binding proteins from the mollusks Lymnaea stagnalis and Aplysia califor-
nica. We found that [A10L,D14K]PnIA also distinguishes two a7-like
anion-selective nAChR subtypes present on identified neurons of L. stag-
nalis: [D14K] mutation affected only slightly the potency of [A10L]PnIA to
block nAChRs on neurons with low sensitivity to a-conotoxin ImI, but
gave a 50-fold enhancement of blocking activity in cells with high sensitiv-
ity to ImI. Therefore, the introduction of an additional positive charge
in the C-terminus of a-conotoxins targeting some muscle or neuronal
nAChRs made them more discriminative towards the respective nAChR
subtypes. In the case of muscle-type a-conotoxin [D12K]SIA, the contribu-
tion of the Lys12 positive charge to enhanced affinity towards Torpedo
nAChR was rationalized with the aid of computer modeling.
Abbreviations
ACh, acetylcholine; AChBP, acetylcholine-binding protein; IC
50
, ligand concentration at which 50% inhibition is achieved; nAChR, nicotinic
acetylcholine receptor; n
H
, Hill coefficient.
4470 FEBS Journal 273 (2006) 4470–4481 ª2006 The Authors Journal compilation ª2006 FEBS
synthesized chemically [2–5]. a-Conotoxins have become
widely used tools in studies on nicotinic acetylcholine re-
ceptors (nAChRs) [6,7] because they can distinguish
between different nicotinic acetylcholine receptor
(nAChR) subtypes. For example, a-conotoxins GI, MI
and SIA selectively block muscle-type nAChRs, whereas
some others block distinct neuronal nAChRs, e.g.
a-conotoxins ImI and ImII target homo-oligomeric a7
nAChR [8], whereas a-conotoxins MII, PnIA, GIC
block heteromeric nAChR containing a3, a6 and b2
subunits [6]. A change in one or several residues of the
naturally occurring a-conotoxin might result in a change
in its nAChR subtype selectivity [9]. For example, the
A10L substitution in a-conotoxin PnIA switched its
selectivity from the a3b2 to the a7 nAChR [10,11].
Synthesis of diverse a-conotoxin analogs, mutations
in nAChRs and pair-wise mutation analysis have
enabled the identification of specific a-conotoxin
and ⁄or nAChR residues taking part in ligand–receptor
interactions [12–15]. The crystal structure of the acet-
ylcholine-binding protein (AChBP) from the mollusk
Lymnaea stagnalis, which provides a high-resolution
structure for the extracellular domains of nAChRs
[16,17], has been used to build models for a-conotoxin
binding to distinct nAChRs [18]. Recently, crystal
structures have been solved for AChBP complexes
with two a-conotoxins: [A10L, D14K]PnIA, a double
mutant of a-conotoxin PnIA [19], and for a-conotoxin
ImI [20,21]. These structures provide a solid basis for
modeling the spatial structures of a-conotoxins with
the cognate nAChRs. Modeling may also be a start-
ing point for the rational design of new a-conotoxins
with higher affinity and better selectivity towards
nAChRs.
D14K substitution increased the affinity of the
starting [A10L]PnIA for chicken a7 nAChR and
L. stagnalis AChBP [19]. X-Ray data on the
AChBP)a-conotoxin complex were the basis for con-
structing a model for a7 nAChR complexes with
[A10L]PnIA and [A10L, D14K]PnIA [19]. We used the
X-ray data and cryoelectron microscopy structure of
Torpedo nAChR [22] to build a respective model for
a-conotoxin [D12K]SIA, wherein the Lys12 positive
charge gave the most dramatic increase in the affinity
for T. californica nAChR.
Anion-selective nAChRs in some identified neurons
of the fresh-water snail L. stagnalis and marine mol-
lusk Aplysia californica were found to resemble the a7
nAChRs of vertebrates in terms of their pharmacologi-
cal profile and the response kinetics to acetylcholine
(ACh) [23,24]. To further elucidate the significance of
a positive charge in the C-terminus of a-conotoxins
we compared the action of [A10L]PnIA and
[A10L,D14K]PnIA on a7-like nAChRs in identified
Lymnaea neurons.This is of interest in light of the
recent cloning of a set of nAChR subunits from this
species and electrophysiological analysis of several of
them expressed in Xenopus oocytes [25,26].
Results and Discussion
Synthesis of a-conotoxins
New analogs of a-conotoxins GI, SI and SIA with
arginine, lysine and ⁄or aspartate introduced at position
12 (Table 1) were synthesized using a solid-phase
method with the simultaneous formation of the two
disulfides. For a-conotoxin SIA, which has Asp12 in
this position, an additional D12S analog was also
synthesized. A series of a-conotoxin MI analogs was
similarly synthesized. In this case, we employed Lys-
scanning mutagenesis for the possibly complete set of
MI variants, excluding the substitutions of structurally
important amino acid residues (Cys, Pro). As a result,
three novel analogs of a-conotoxin MI with a lysine
residue introduced at position 5, 7 or 11 were obtained.
Simultaneous formation of the disulfides decreases
the number of stages and usually gives higher peptide
yields, although this is sometimes accompanied by the
production of incorrectly bridged isomers [27]. When
several isomers were formed, the peptide with correctly
formed disulfide bridges was assumed to have a higher
potency to bind to the membrane-bound T. californica
nAChR in the radioligand-binding assay (see below).
It is known that incorrect disulfide formation in
a-conotoxins that target the muscle-type nAChRs
entails a decrease in the affinity [28]. However, the
enhanced affinity of the incorrectly formed isomer of
a-conotoxin AuIB, targeting one neuronal-type
nAChR, was revealed previously [29] and makes the
method less predictive. Therefore, all new synthesized
analogs were also characterized using CD spectroscopy
to detect secondary structure changes in the ‘incorrect’
isomers (see below). In 13 syntheses of muscle-type
conotoxins we found the generation of isomers only in
two cases – one additional minor peak for [S12D]GI
and two for SIA (all peaks had correct molecular
masses).
a-Conotoxin [A10L]PnIA, known to act on a7
nAChR [10,11] was obtained by solid-phase peptide
synthesis using the simultaneous formation of two
disulfides as described previously [30]. In the case of
[A10L,D14K]PnIA, orthogonal protecting groups were
used for correct pair-wise closing of disulfides to
exclude the formation of other isomers (see Experi-
mental procedures). The structures of all synthesized
I. E. Kasheverov et al. Novel a-conotoxin analogs
FEBS Journal 273 (2006) 4470–4481 ª2006 The Authors Journal compilation ª2006 FEBS 4471
peptides were verified by MALDI analysis (Table 1)
and purity by RP-HPLC (data not shown).
CD spectroscopy
CD spectra were obtained for aqueous solutions of
native a-conotoxins GI, MI, SI, SIA and their analogs
[S12R ⁄K⁄D]GI, [H5K]MI, [S12R]SI, [D12S ⁄K]SIA, as
well as for one isomer of [S12D]GI and two isomers
of SIA, which were produced in noticeable quantities
during peptide syntheses. As an example the spectra of
a-conotoxin GI analogs are presented in Fig. 1. Amino
acid substitutions at position 12 did not result in any
noticeable alterations in peptide secondary structure.
However, the second (minor) isomer of [S12D]GI dis-
played a remarkable change in spectral characteristics
(inset in Fig. 1). Similarly, the spectra of the SI and
SIA analogs with substitution at position 12 (as well
as [H5K]MI) were identical to that of the respective
naturally occurring a-conotoxins. However, both
minor isomers of SIA had spectra resembling that of
minor [S12D]GI isomer (data not shown).
The available literature data indicate that single
amino acid substitutions do not markedly change the
CD curves of a-conotoxins. However, breaking the
Cys–Cys disulfide bonds in a-conotoxin ImI [31] or
x-conotoxin MVIIA [32], or changing the size of the
disulfide-confined peptide loops by introduction of
an additional amino acid residue in a-conotoxin ImI
[33], resulted in a remarkable change in CD spectra,
with shifting of the ellipticity minimum into the
195–200 nm region. This shift resembles that seen for
minor isomers of both [S12D]GI and SIA (see curve
4a in the inset of Fig. 1). Taken together, these results
indicate that analysis of biological activities (Fig. 2)
has been carried out on a series of a-conotoxins with
correctly closed disulfides.
Binding of synthesized a-conotoxin analogs to
membrane-bound Torpedo nAChR
The activity of analogs was evaluated in competition
with radioiodinated a-conotoxins GI or MI for binding
Fig. 1. CD spectra of a-conotoxins GI (1, solid line), [S12K]GI (2,
dotted line), [S12R]GI (3, dash-dot line) and main isomer of
[S12D]GI (4, dash line) in water. Inset: CD spectra of two [S12D]GI
isomers – the main (4, solid line) and minor (4a, dash line) ones.
Table 1. The structures of synthesized naturally occurring a-conotoxins and their analogs. All a-conotoxins have amidated C-termini as well
as disulfide bridges Cys1–Cys3 and Cys2–Cys4. The substituted residues in the analogs are indicated in bold type.
a-Conotoxin Sequence
Mol. mass, MH
+
Calculated MALDI-measured
GI ECCNPACGRHYSC 1438.6 1438.4
[S12R]GI ECCNPACGRHYRC1507.7 1506.6
[S12K]GI ECCNPACGRHYKC1479.7 1480.9
[S12D]GI ECCNPACGRHYDC1466.6 1465.5
SI ICCNPACGPKYSC 1354.6 1353.5
[S12R]SI ICCNPACGPKYRC1423.7 1422.4
SIA YCCHPACGKNFDC 1456.7 1455.6
[D12S]SIA YCCHPACGKNFSC1428.7 1427.5
[D12K]SIA YCCHPACGKNFKC1469.9 1469.3
MI GRCCHPACGKNYSC 1494.7 1494.4
[H5K]MI GRCCKPACGKNYSC 1485.7 1484.6
[A7K]MI GRCCHPKCGKNYSC 1553.2 1554.1
[N11K]MI GRCCHPACGKKYSC 1508.8 1507.7
[A10L]PnIA
a
GCCSLPPCALNNPDYC 1664.7 1664.7
[A10L,D14K]PnIA
a
GCCSLPPCALNNPKYC 1677.8 1677.6
a
Described in Celie et al. [19].
Novel a-conotoxin analogs I. E. Kasheverov et al.
4472 FEBS Journal 273 (2006) 4470–4481 ª2006 The Authors Journal compilation ª2006 FEBS
to membrane-bound T. californica nAChR (Fig. 2).
Both tracers bound specifically to the Torpedo receptor
with equal high affinity: K
d
values for
125
I-labeled GI
and MI were 24 ± 3 and 28 ± 6 nm, respectively. By
contrast to a-conotoxin GI and M1, a-conotoxin SI
has an equal potency to both sites in the Torpedo
nAChR [34], whereas a-conotoxin SIA binds to only
one site [35] as revealed by competition with
125
I-labe-
led a-bungarotoxin. That is why we did not prepare the
radioactive forms of these peptides, and
125
I-labeled GI
was used as a tracer to test the SI and SIA analogs. In
these experiments the synthetic a-conotoxins GI, SI,
SIA and MI were used as controls. The respective lig-
and concentrations at which 50% inhibition is achieved
(IC
50
values) are presented in Table 2.
The introduction of a positively charged amino acid
residue instead of a neutral one in position 12 of
a-conotoxins GI and SI resulted in a three- to seven-
fold increase in the affinity to both binding sites of
the Torpedo nAChR (Fig. 2A,B; Table 2). The most
remarkable was the D12K mutation in a-conotoxin
SIA: the binding efficiencies to the high- and low-affin-
ity sites increased for the [D12K]SIA analog by 35 and
260 times, respectively (Fig. 2C; Table 2). This increase
was due mainly to removal of the negatively charged
amino acid residue in this position, because substitu-
tion with neutral Ser also resulted in affinity enhance-
ment to both sites (25 and 65 times, respectively).
Conversely, the introduction of a negative charge in
position 12 of a-conotoxin GI caused a considerable
decrease in the affinity for the receptor (Fig. 2A;
Table 2). However, the introduction of an additional
positive charge (Lys) at position 11 of a-conotoxin MI
(which corresponds spatially to residue 10 of a-cono-
toxins GI, SI and SIA) affected the peptide activity
only slightly, whereas H5K or A7K mutations wor-
sened the binding characteristics of these analogs
(Fig. 2D; Table 2).
It should be noted that all three minor isomers
(of [S12D]GI and SIA) showed more than tenfold
Fig. 2. Inhibition of
125
I-labeled a-conotoxins GI (A–C) and MI (D) binding to membrane-bound Torpedo nAChR with indicated a-conotoxins
and their analogs. Final concentrations of the radioligand and toxin-binding sites of receptor were 280 and 230 nM, respectively. The data
shown are the averages of two independent experiments. The inhibition curves were fitted using ORIGIN 6.1 (MicroCal Software Inc.) in the
frames of a two-site competition model for all peptides (with one exception for [A7K]MI). The respective IC
50
values are presented in
Table 2.
I. E. Kasheverov et al. Novel a-conotoxin analogs
FEBS Journal 273 (2006) 4470–4481 ª2006 The Authors Journal compilation ª2006 FEBS 4473
decreased efficacies, compared with the major com-
pounds, in competition with radiolabeled a-conotoxin
GI for the T. californica nAChR binding (data not
shown).
Both PnIA variants at concentrations of up to
100 lmwere inactive in competition with
125
I-labeled
GI for binding to the membrane-bound Torpedo
nAChR (data not shown).
We synthesized mainly the modified a-conotoxins
targeting the muscle-type nAChR. Literature data on
the role of charged residues in this group of a-cono-
toxins are in part contradictory. Several researchers
have shown that charged groups at the N-termini of
a-conotoxins GI, MI and SI exert only a weak influence
on the activity [33,35–39]. The important role of Arg9
in the interaction with a high-affinity a-conotoxin-
binding site on the Torpedo nAChR has been convin-
cingly demonstrated: R9P and R9A substitutions in
a-conotoxin GI resulted in a two to three order of
magnitude loss in the affinity for the a⁄csite, whereas
the reverse substitutions P9R and P9K in a-conotoxin
SI enhanced the affinity for this site [34,35,40]. How-
ever, when Ala or Pro residues in a-conotoxin MI were
substituted for the Lys10, whose spatial disposition is
close to that of Arg9 in a-conotoxin GI, the interac-
tion with the high-affinity a⁄c-binding site was affected
to a much less degree [38–40]. In addition, acylation of
Lys10 with azidobenzoyl or benzoylbenzoyl groups
practically did not change the capacity of the respect-
ive derivatives to interact with the membrane-bound
Torpedo nAChR [41].
Of all known muscle-type a-conotoxins, only
a-conotoxin SIA interacts exclusively with one a⁄csite
on the Torpedo nAChR [35]. Interestingly, this peptide
contains a negatively charged residue (Asp12) in the
C-terminal part of the molecule (Table 1) whose role
has not been examined previously. D12S substitution
resulted in a 25- and 65-fold increase in the affinity for
the high- and low-affinity binding sites, respectively
(Fig. 2C; Table 2). Introduction of a positive charge
(Lys) at this position resulted in an additional fourfold
increase in affinity for the low-affinity site (Fig. 2C;
Table 2). Substitution of Lys or Arg for Ser12 in
a-conotoxins GI and SI gave a reliable enhancement
(three- to sevenfold) of the affinity for both binding
sites (Fig. 2A,B; Table 2). By contrast, introduction of
a negative charge at this position (S12D) in a-conotox-
in GI brought about a marked decrease in the affinity
(Fig. 2A; Table 2). It is noteworthy that use of
125
I-
labeled a-conotoxin GI in these experiments, instead
of the usual
125
I-labeled a-bungarotoxin [34,35,39,40],
revealed the differences in potency to two sites for
a-conotoxin SI and made possible the detection of a
low-affinity binding site for a-conotoxin SIA in the
Torpedo nAChR. From literature data it is known that
the affinities of muscle-type a-conotoxins to the Tor-
pedo nAChRs binding sites (tested in competition with
125
I-labeled a-bungarotoxin) vary from one to three
orders of magnitude [12,34,35,39]. The given explan-
ation for this scatter is the influence of the receptor
state, test conditions, etc. It is therefore not surprising
that using a different tracer in the radioligand assay
may result in different binding parameters for a-cono-
toxins. This was shown previously for one a-conotoxin
GI analog on the Torpedo receptor [41]. There is
convincing evidence (the crystal structures of the
a-cobratoxin and a-conotoxin complexes with acetyl-
choline-binding proteins) [19–21,42] that the binding
sites for these two groups of competitive antagonists
overlap, but are not identical.
Grafting positive charges on to other positions
of a-conotoxin amino acid sequence resulted in
Table 2. Activity of a-conotoxins and their analogs tested in compe-
tition binding assays. Using the membrane-bound Torpedo nAChR,
the inhibitory activities of a-conotoxins GI, SI, SIA or MI and their
analogs were evaluated in competition with
125
I-labeled a-conotox-
ins GI (GI, SI, SIA and their analogs) or MI (MI and its analogs): see
respective inhibition curves presented in Fig. 2. IC
50
values were
calculated using ORIGIN 6.1 in the frames of both one- and two-site
models using the joint data from two or three independent experi-
ments for each a-conotoxin. The choice was made in favor of the
model giving the minimal ‘reduced chi-squared’ parameter comple-
mented with reasonable SE values and taking into consideration
the Hill coefficients (n
H
). For all muscle-type conotoxins (with one
exception), a two-site model was found the best. In the case of
[A7K]MI analog the program failed to fit the data to a two-site
model, so the respective IC
50
value was generated in the frames of
one-site model and ascribed to both sites. Both PnIA variants were
inactive in competition with
125
I-labeled a-conotoxins GI at 100 lM
(4 ± 2% of inhibition).
a-Conotoxin n
H
IC
50
,lM
high affinity site low affinity site
GI 0.64 ± 0.04 1.6 ± 0.7 9.3 ± 3.7
[S12R]GI 0.68 ± 0.04 0.49 ± 0.25 1.7 ± 1.0
[S12K]GI 0.56 ± 0.06 0.29 ± 0.13 3.2 ± 1.1
[S12D]GI 0.52 ± 0.07 12.0 ± 2.9 230 ± 50
SI 0.53 ± 0.06 4.0 ± 1.2 58 ± 25
[S12R]SI 0.59 ± 0.04 1.0 ± 0.3 8.2 ± 3.6
SIA 0.32 ± 0.06 3.5 ± 1.3 440 ± 150
[D12S]SIA 0.46 ± 0.03 0.13 ± 0.02 6.6 ± 0.6
[D12K]SIA 0.52 ± 0.05 0.10 ± 0.05 1.7 ± 0.7
MI 0.55 ± 0.04 0.26 ± 0.07 6.6 ± 0.7
[H5K]MI 0.53 ± 0.04 9.1 ± 1.8 130 ± 60
[A7K]MI 0.83 ± 0.06 54 ± 4 54 ± 4
[N11K]M 0.68 ± 0.06 0.24 ± 0.16 3.7 ± 0.7
[A10L,D14K]PnIA – – 100
[A10L]PnIA – – 100
Novel a-conotoxin analogs I. E. Kasheverov et al.
4474 FEBS Journal 273 (2006) 4470–4481 ª2006 The Authors Journal compilation ª2006 FEBS
