Maleimide conjugates of saxitoxin as covalent inhibitors of voltage-gated sodium channels

Article abstract of DOI:10.1021/ja4019644

(+)-Saxitoxin, a naturally occurring guanidinium poison, functions as a potent, selective, and reversible inhibitor of voltage-gated sodium ion channels (Na_Vs). Modified forms of this toxin bearing cysteine-reactive maleimide groups are available through total synthesis and are found to irreversibly inhibit sodium ion conductance in recombinantly expressed wild-type sodium channels and in hippocampal nerve cells. Our findings support a mechanism for covalent protein modification in which toxin binding to the channel pore precedes maleimide alkylation of a nucleophilic amino acid. Second-generation maleimide-toxin conjugates, which include bioorthogonal reactive groups, are also found to block channel function irreversibly; such compounds have potential as reagents for selective labeling of Na_Vs for live cell imaging and/or proteomics experiments.

Full text of DOI:10.1021/ja4019644

Communication
pubs.acs.org/JACS
Maleimide Conjugates of Saxitoxin as Covalent Inhibitors of Voltage-
Gated Sodium Channels
William H. Parsons† and J. Du Bois*
Department of Chemistry, Stanford University, 333 Campus Drive, Stanford, California 94305-5080, United States
S
* Supporting Information
our lab have demonstrated that functionalization of the
ABSTRACT: (+)-Saxitoxin, a naturally occurring guani-
dinium poison, functions as a potent, selective, and
reversible inhibitor of voltage-gated sodium ion channels
(NaVs). Modified forms of this toxin bearing cysteine-
reactive maleimide groups are available through total
synthesis and are found to irreversibly inhibit sodium ion
conductance in recombinantly expressed wild-type sodium
channels and in hippocampal nerve cells. Our findings
support a mechanism for covalent protein modification in
which toxin binding to the channel pore precedes
maleimide alkylation of a nucleophilic amino acid.
Second-generation maleimide-toxin conjugates, which
include bioorthogonal reactive groups, are also found to
block channel function irreversibly; such compounds have
potential as reagents for selective labeling of NaVs for live
cell imaging and/or proteomics experiments.
carbamate moiety in STX can be accommodated with limited
influence on the binding affinity between ligand and receptor.17
As shown in Scheme 1, maleimide conjugates of STX (1, 6, and
Scheme 1. De Novo Synthesis of Maleimide and Succinimide
Conjugates of (+)-Saxitoxin (STX)
n electrically excitable cells, the coordinated action of
voltage-gated sodium channels (NaVs) is responsible for the
I
rising phase of the action potential.1 These integral membrane
protein complexes are comprised of a large, pore-forming α-
subunit, coexpressed with two β-auxiliary glycoproteins. Genes
encoding 10 sodium channel α-subtypes (NaV1.1−1.9, NaX)
have been identified in mammalian cells.2 Protein isoform
expression levels are tightly regulated and varied within the cell
and across different tissues. Chemical agents that act as specific
modulators of NaV function and/or as probes for live-cell
imaging or affinity profiling are sought as tools for studies
aimed at understanding the role of NaV isoforms in shaping
electrical signals in neuronal cells.3,4 The potential utility of
such compounds has motivated the studies described herein
and has led to the design of a unique collection of small
molecule, covalent inhibitors of NaVs.
7) can be obtained through selective N-hydroxysuccinimide
(NHS) ester coupling to aminoalkyl-substituted STXs, starting
materials made available through multistep synthesis. The same
type of reaction has been employed to prepare succinimide (2),
a structural analogue of 1 needed for control experiments.
Inhibition of sodium current by STX-maleimide conjugates
was evaluated through whole-cell electrophysiology measure-
ments. Experiments were performed as voltage-clamp record-
ings on the α-subunit of the rat skeletal muscle channel
(rNaV1.4) heterologously expressed in Chinese hamster ovary
(CHO) cells. Initial recordings using saturating concentrations
(5 μM) of 1 against wild-type rNaV1.4 (CHO cells) gave
unexpected results, as channel conduction was irreversibly
disabled. Following 1 min of incubating cells with 1, continuous
perfusion with toxin-free external solution for 8 min restores
only 51 6% of initial peak current, I0. As shown in Figures
1A,B and S1a,b (Supporting Information), longer incubation
periods of 3 and 6 min lead to demonstrable decreases in %
Guanidinium toxins, which bind reversibly and with low
nanomolar affinity to the outer pore of the sodium channel,
have featured prominently in NaV research.5 Structure−activity
experiments using tetrodotoxin (TTX), saxitoxin (STX), and a
small number of congeners together with protein mutagenesis
have informed modeling studies of the channel pore and toxin
binding site.6−9 Access to modified STXs through de novo
chemical synthesis10−15 enables further examination of the pore
architecture and generation of novel reagents for NaV labeling.
Use of a maleimide-conjugated STX (e.g., 1) to covalently
modify a cysteine mutant channel was envisioned as a strategy
for examining the fidelity of our toxin pore model and for
targeting selectively a single NaV subtype.16 Prior studies from
current recovery after equivalent washout periods (36
and 18
5%
3% of I0, respectively).18 The inability to restore
initial current levels following perfusion of the cell with toxin-
free solution is in marked contrast to the fully reversible
Received: February 23, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja4019644 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
the presence of the electrophilic unsaturated dicarbonyl moiety,
which can react with a nucleophilic amino acid group.
To explore further the activity of 1 as an irreversible NaV
antagonist, whole-cell electrophysiology recordings were
performed against two additional channel isoforms, rNaV1.2
and hNaV1.5. The measured inhibitory constant for 2 against
rNaV1.2 (IC50 = 38 5 nM) is comparable to that of rNaV1.4
but is substantially higher for the TTX-resistant channel,
hNaV1.5 (IC50 = 4.5
0.8 μM) (Figure S4, Supporting
Information). As shown in Figure 2, application of 5 μM 1 to
Figure 1. (A) Current recordings, elicited by a 10 ms voltage step
from −100 to 0 mV, upon 3 min application and wash-off of 5 μM
maleimide 1 to CHO cells expressing rNaV1.4. (B) Representative
wash-off time course for 1. (C) Current recordings upon 3 min
application and wash-off of 5 μM succinimide 2. (D) Representative
wash-off time course for 2. Average current recovery after washout was
36 5% of I0 for 1, compared to 101 6% for 2. Data represent the
average of 3−5 cells SEM. (E) Current recovery following 8 min
washout (incubation time for each compound shown in parentheses).
Data represent the average of 3−7 cells SEM.
Figure 2. Representative time courses of wash-off of 5 μM 1 from
CHO cells expressing rNaV1.2 (A) and hNaV1.5 (B). Average current
recovery after washout was 25 1% of I0 on rNaV1.2, compared to
100 3% on hNaV1.5. (C) Current recordings, elicited by a 10 ms
voltage step from −80 to 0 mV, upon 6 min application and wash-off
of 5 μM maleimide 1 to embryonic rat hippocampal neurons. (D)
Representative wash-off time course of 5 μM 1 from embryonic rat
hippocampal neurons. Average current recovery after washout was 25
2% of I0. Data represent the average of 3−5 cells SEM.
binding behavior of STX (Figure S1c, Supporting Information).
The unique activity of 1 appears to be use-independent, as 3
min application and washout of this compound without active
stimulation of the channel does not alter performance (current
CHO cells expressing rNaV1.2 completely blocks sodium
conduction; following a 3 min incubation period, subsequent
washout of the toxin conjugate results in only partial restoration
of channel current (25
1% of I0). Comparable results are
recovery = 35
3% of I0) (Figure S1d, Supporting
obtained when 1 is perfused onto CHO cells coexpressing
NaVβ1 and the α-subunit of NaV1.2 (19 4% of I0, Figure S5b,
Supporting Information); these findings are consistent with
previous reports demonstrating that accessory subunit proteins
do not affect STX binding.19 By contrast, 5 μM solutions of 1
administered to CHO cells expressing hNaV1.5 give only partial
current block, and complete restoration of peak current is
achieved following washout with toxin-free external solution
(100 3% of I0). These data establish that high affinity binding
of the toxin to the channel is a prerequisite for achieving
efficient, covalent protein cross-linking.
Information). Other maleimide conjugates with extended linker
groups (6 and 7) exhibit similar irreversible binding behavior
(Figure S2, Supporting Information). Collectively, these results
are in accord with reaction of 1 and a nucleophilic residue
positioned near the toxin binding site in the wild-type rNaV1.4
channel.
Support for the covalent modification of wt-rNaV1.4 by 1 has
been gained using a nonreactive, succinimide analogue 2.
Concentration−response measurements (Figure S4, Support-
ing Information) give an IC50 value of 54
7 nM for 2,
evidence that modification of the carbamate unit does not
significantly perturb toxin affinity for the channel. Application
of this compound at 5 μM concentration to rNaV1.4-expressing
CHO cells affords complete channel block. Perfusion of toxin-
free buffer, however, results in rapid recovery of peak current
(101 6% of I0), behavior that parallels STX binding (Figure
1D). Thus, irreversible block by 1 appears to be dependent on
To validate the utility of 1 for covalent modification of
endogeneous NaVs in primary cells, we have performed
electrophysiology experiments with embryonic rat hippocampal
neurons (Figure 2C,D).20 Following a six minute application of
5 μM 1 to these cells, only a fraction of the initial current (25
2% of I0) is recovered after extended washout. Accordingly, the
irreversible blocking behavior of 1 is not unique to one cell type
B
dx.doi.org/10.1021/ja4019644 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
and does not appear to be influenced by auxiliary proteins
known to associate with the NaV α-subunit.21
Scheme 2. Synthesis of C2-Substituted Maleimide
Derivatives
a
Experiments with 8, a maleimide conjugate prepared from β-
saxitoxinol, offer additional support for our hypothesis that
toxin association with the channel pore occurs prior to
nucleophilic attack on the maleimide (Figure 3A,B).22 β-
a
Figure 3. (A) Structure of maleimide-conjugated β-saxitoxinol 8. (B)
Representative time course of wash-off of 5 μM 8 from CHO cells
expressing NaV1.4. Current recovery after washout was 94 7% of I0,
which represents the average of 5 cells SEM.
Conditions: (a) glyoxylic acid·H2O, morpholinium hydrochloride,
dioxane/H2O, 100 °C, 65%; (b) Dess−Martin periodinane, CH2Cl2,
71%; (c) β-alanine, AcOH, 23 → 105 °C, 77%; (d) HgSO4, H2SO4,
MeOH, 59%; (e) N,N′-disuccinimidyl carbonate, i-Pr2NEt, CH2Cl2,
62%; (f) 3, aq. CH3CN, pH 8.5, 58%.
Saxitoxinol itself is a reduced form of the natural product that is
3 orders of magnitude less potent against rNaV1.4.23,24
Incubation of CHO cells expressing the 1.4 subtype for 3
min with 5 μM 8 affords only partial current block, and nearly
full recovery of I0 is observed following perfusion with external
solution (94
7% of I0). Similarly, application of a 5 μM
solution of a maleimide tethered to a single guanidinium group
to rNaV1.4 gives neither transient nor prolonged block of
channel current (Figure S7b, Supporting Information).
Examination of the primary sequences of rNaV1.2 and
rNaV1.4 as well as our homology model of the channel pore
shows three native cysteines, C763, C1546, and C1561 (NaV1.4
numbering), within proximity of the proposed toxin binding
site. On the basis of our homology model of the p-loop region,9
C763 appears to be the most likely candidate nucleophile
(Figure S8, Supporting Information). To test this prediction,
we have constructed two amino acid mutants, C763S and
C763A, of the rNaV1.4 isoform. Expression of these mutant
proteins in both CHO and tsa201 cells failed to produce
macroscopic current levels comparable to wild-type (0−0.3 nA
for all n = 32 cells tested). Such low current densities were
deemed insufficient for definitive characterization of the
blocking behavior of 1. Our present efforts are dedicated
toward identifying the amino acid residue(s) modified by 1
through alternative means, which include protein isolation and
mass spectrometric sequencing.
The surprising behavior of 1 against wt-NaV isoforms can be
exploited to mark endogenous channels in primary cells (see
Figure 2C,D). To this end, we have prepared unique, C2-
substituted maleimide groups bearing an attendant alkyne 10 or
ketone moiety 11 (Scheme 2).25 Selective coupling of activated
ester forms of these reagents to STX-amines such as 3 proceeds
without event. Modification at C2 on the maleimide core has a
limited influence on electrophilic reactivity with cysteine
nucleophiles (see Figure S9, Supporting Information, for a
comparison of reaction rates).
Figure 4. (A) Structures of C2-substituted maleimide (12) and
succinimide (13) derivatives of STX. (B) Representative time course
of wash-off of 5 μM 12 from CHO cells expressing rNaV1.2. (C)
Representative time course of wash-off of 5 μM 13 from CHO cells
expressing rNaV1.2. Average current recovery after washout was 22
5% of I0 for 12, compared to 106 6% for 13. Data represent the
average of 3−5 cells SEM.
the succinimide analogue 13 (Figure 4). Alkylation of wt-NaVs
with 12 presents a ketone functional group on the extracellular
protein surface that is suitable for bioorthogonal modification
with a fluorescent dye or biotin cofactor.26 Such probes could
enable tracking of channel endocytosis or facilitate isolation and
identification of NaVs expressed at the cell surface.27
In summary, we have prepared a unique collection of
maleimide-bearing STX derivatives that display irreversible
block of wild-type NaVs expressed in heterologous cells and in
hippocampal neurons. Our experiments support a mechanism
in which toxin binding precedes maleimide cross-linking with a
proximal amino acid, the most likely candidate of which is
C763 on the basis of our homology model of the channel pore
and toxin binding site. These unique reactive probes should
Both maleimide 12 and succinimide 13 conjugates of STX
have been prepared (Scheme 2, Figure 4) and evaluated against
rNaV1.2. As with toxin derivatives 1 and 2, incubation of cells
with 5 μM concentrations of 12 and 13 and subsequent
washout reveals partial current recovery with the former (22
5% of I0) but rapid washout and complete restoration of I0 with
C
dx.doi.org/10.1021/ja4019644 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
(15) Sawayama, Y.; Nishikawa, T. Angew. Chem., Int. Ed. 2011, 50,
7176−7178.
(16) For related studies with a voltage-gated potassium channel:
Morin, T. J.; Kobertz, W. R. ACS Chem. Biol. 2007, 2, 469−473.
(17) Andresen, B. M.; Du Bois, J. J. Am. Chem. Soc. 2009, 131,
12524−12525.
(18) As shown in Figure S3 (Supporting Information), maximal
irreversible inhibition is ∼85% upon either extended incubation times
or iterative application of 1; see Supporting Information for additional
details. A similar inability to completely saturate current levels using a
high affinity, covalent toxin against nAChR has been noted previously;
see: Ambrus, J. I.; Halliday, J. I.; Kanizaj, N.; Absalom, N.; Harpsøe, K.;
Balle, T.; Chebib, M.; McLeod, M. D. Chem. Commun. 2012, 48,
6699−6701.
have general applicability for live-cell investigations of NaV
function, studies of which are ongoing and will be reported in
due course.
ASSOCIATED CONTENT
* Supporting Information
■
S
Experimental synthetic procedures and characterization data for
all novel compounds, details of cell culture and electro-
physiology procedures, additional supplementary electrophysio-
logical data, and full discussion of NMR experiments. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(19) Isom, L. L.; Scheuer, T.; Brownstein, A. B.; Ragsdale, D. S.;
Murphy, B. J.; Catterall, W. A. J. Biol. Chem. 1995, 270, 3306−3312.
(20) Prior studies indicate that the TTX-sensitive isoforms NaV1.2,
1.3, 1.6, and 1.7 are expressed in embryonic rat hippocampal neurons,
see: Mechaly, I.; Scamps, F.; Chabbert, C.; Sans, A.; Valmier, J.
Neuroscience 2005, 130, 389−396. We have determined an IC50 of 2.9
0.4 nM for STX against these cells (Figure S6, Supporting
Information).
AUTHOR INFORMATION
Corresponding Author
jdubois@stanford.edu
■
Present Address
†Department of Chemical Physiology, The Scripps Research
Institute, 10550 North Torrey Pines Road, La Jolla, California
92037, United States.
(21) For a general review, see: Vacher, H.; Mohapatra, D. P.;
Trimmer, J. S. Physiol. Rev. 2008, 88, 1407−1447.
Notes
(22) Experiments in which either 8 or a monoguanidine-maleimide
construct S1 (Figure S7c, Supporting Information) are applied in
combination with 5 μM STX show no irreversible block of current.
These data argue against the possibility that toxin binding simply
promotes nonspecific alkylation. For reasons that are not evident,
however, simultaneous application of a 20-fold excess of either STX or
TTX with 1 does not significantly attenuate irreversible block of
channel current.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Dr. T. R. Cummins (Indiana University) for
providing the expression plasmid for NaV1.5, Dr. L. L Isom
(University of Michigan) for providing the expression plasmid
for NaVβ1, and Dr. W. A. Catterall (University of Washington)
for the gift of a NaV1.2 stably expressed cell line. We are
grateful to Dr. Bianxiao Cui (Stanford University) and Wenjun
Xie for providing embryonic rat hippocampal neurons for our
experiments. We are indebted to Dr. M. Maduke for her helpful
discussions and for generous use of her lab space. W.H.P. is
supported by a Stanford Interdisciplinary Graduate Fellowship
(SIGF). This work was financially supported by National
Institutes of Health R01 NS045684, R21 NS070064.
(23) Koehn, F. E.; Ghazarossian, V. E.; Schantz, E. J.; Schnoes, H. K.;
Strong, F. M. Bioorg. Chem. 1981, 10, 412−428.
(24) Andresen, B. M. Ph.D. Dissertation, Stanford University, 2010.
(25) Preparation of C2-substituted maleimides adapted from Miles,
W. H.; Yan, M. Tetrahedron Lett. 2010, 51, 1710−1712.
(26) For preparation of a biotinylated saxitoxin derivative through
semisynthesis, see: Robillot, C.; Kineavy, D.; Burnell, J.; Llewellyn, L.
E. Toxicon 2009, 53, 460−465.
(27) Cravatt, B. F.; Wright, A. T.; Kozarich, J. W. Annu. Rev. Biochem.
2008, 77, 383−414.
REFERENCES
■
(1) Aidley, D. J. The Physiology of Excitable Cells; Cambridge
University Press: Cambridge, UK, 1998; Chapter 5.
(2) Catterall, W. A.; Goldin, A. L.; Waxman, S. G. Pharmacol. Rev.
2005, 57, 397−409.
(3) Nardi, A.; Damann, N.; Hertrampf, T.; Kless, A. ChemMedChem
2012, 7, 1712−1740.
(4) Leterrier, C.; Brachet, A.; Fache, M.-P.; Dargent, B. Neurosci. Lett.
2010, 486, 92−100.
(5) Fozzard, H. A.; Lipkind, G. M. Mar. Drugs 2010, 8, 219−234.
(6) Lipkind, G. M.; Fozzard, H. A. Biophys. J. 1994, 66, 1−13.
(7) Tikhonov, D. B.; Zhorov, B. S. Biophys. J. 2005, 88, 184−197.
(8) Scheib, H.; McLay, I.; Guex, N.; Clare, J. J.; Blaney, F. E.; Dale, T.
J.; Tate, S. N.; Robertson, G. M. J. Mol. Model. 2006, 12, 813−822.
(9) Walker, J. R.; Novick, P. A.; Parsons, W. H.; McGregor, M.;
Zablocki, J.; Pande, V. S.; Du Bois, J. Proc. Natl. Acad. Sci. U. S. A.
2012, 109, 18102−18107.
(10) Tanino, H.; Nakata, T.; Kaneko, T.; Kishi, Y. J. Am. Chem. Soc.
1977, 99, 2818−2819.
(11) Jacobi, P. A.; Martinelli, M. J.; Polanc, S. J. Am. Chem. Soc. 1984,
106, 5594−5598.
(12) Fleming, J. J.; McReynolds, M. D.; Du Bois, J. J. Am. Chem. Soc.
2007, 129, 9964−9975.
(13) Iwamoto, O.; Shinohara, R.; Nagasawa, K. Chem.Asian J.
2009, 4, 277−285.
(14) Bhonde, V. R.; Looper, R. E. J. Am. Chem. Soc. 2011, 133,
20172−20174.
D
dx.doi.org/10.1021/ja4019644 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX