Optical control of muscular nicotinic channels with azocuroniums, photoswitchable azobenzenes bearing two N-methyl-N-carbocyclic quaternary ammonium groups

Article abstract of DOI:10.1016/j.ejmech.2020.112403

By linking two N-methyl-N-carbocyclic quaternary ammonium groups to an azobenzene scaffold in meta- or para-positions we generated a series of photoswitchable neuromuscular ligands for which we coined the term “azocuroniums”. These compounds switched betw

Full text of DOI:10.1016/j.ejmech.2020.112403

Journal Pre-proof
Optical control of muscular nicotinic channels with azocuroniums, photoswitchable
azobenzenes bearing two N-methyl-N-carbocyclic quaternary ammonium groups
Clara Herrera-Arozamena, Martín Estrada-Valencia, Olaia Martí-Marí, Concepción
Pérez, Mario de la Fuente Revenga, Carlos A. Villalba-Galea, María Isabel
Rodríguez-Franco
PII:
S0223-5234(20)30373-1
DOI:
https://doi.org/10.1016/j.ejmech.2020.112403
EJMECH 112403
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 11 February 2020
Revised Date: 17 April 2020
Accepted Date: 27 April 2020
Please cite this article as: C. Herrera-Arozamena, Martí. Estrada-Valencia, O. Martí-Marí, Concepció.
Pérez, M. de la Fuente Revenga, C.A. Villalba-Galea, Marí.Isabel. Rodríguez-Franco, Optical control of
muscular nicotinic channels with azocuroniums, photoswitchable azobenzenes bearing two N-methyl-N-
carbocyclic quaternary ammonium groups, European Journal of Medicinal Chemistry (2020), doi: https://
doi.org/10.1016/j.ejmech.2020.112403.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Masson SAS.

Optical control of muscular nicotinic channels with
azocuroniums, photoswitchable azobenzenes bearing two
N-methyl-N-carbocyclic quaternary ammonium groups
Clara Herrera-Arozamenaa, Martín Estrada-Valenciaa, Olaia Martí-María,
Concepción Péreza, Mario de la Fuente Revengab, Carlos A. Villalba-Galeac,*,
and María Isabel Rodríguez-Francoa,*
aInstituto de Química Médica, Consejo Superior de Investigaciones Científicas (IQM-
CSIC), c/ Juan de la Cierva 3, E-28006 Madrid, Spain.
bDepartment of Physiology and Biophysics, School of Medicine, Virginia Commonwealth
University, Richmond, VA 23298, USA
cPhysiology and Pharmacology Department, University of the Pacific, Stockton, CA, USA.
*Correspondence should be addressed to:
María Isabel Rodríguez-Franco
Instituto de Química Médica (IQM-CSIC)
C/ Juan de la Cierva, 3. 28006
Madrid, Spain.
Email: isabelrguez@iqm.csic.es
ORCID: 0000-0002-6500-792X
Carlos A. Villalba-Galea
Department of Physiology and Pharmacology
Thomas J. Long School of Pharmacy & Health Sciences
University of the Pacific
3601 Pacific Avenue
Stockton, CA 95211, USA
Email: cvillalbagalea@pacific.edu
1

Keywords
Photoswitchable neuromuscular ligands
Muscle-type nicotinic receptors
Electrophysiological studies
Structure-activity relationships
Highlights
•
Azocuroniums: azobenzene plus two N-methyl-N-carbocyclic quaternary
ammonium groups
•
•
Azocuroniums are easily (E)/(Z)-photoisomerized by irradiation at blue or UV-LED
m-Azocuroniums: more potent and selective in muscular nAChR than p-
azocuroniums
•
•
m-Azocuroniums are selective ligands of muscle-type nAChRs vs. neuronal
nAChRs
m-Azocuroniums are muscular nAChR antagonists, except the pyrrolidine
derivative
2

Abbreviations
ACh, acetylcholine; nAChRs: nicotinic acetylcholine receptors; bisQ, 3,3'-bis[α-
(trimethylammonium)methyl]azobenzene; GPCRs, G protein-coupled receptors; LED:
light-emitting diode; NMBAs, neuromuscular blocking agents; TEVC, two-electrode
voltage-clamp technique.
3

ABSTRACT
By linking two N-methyl-N-carbocyclic quaternary ammonium groups to an azobenzene
scaffold in meta- or para-positions we generated a series of photoswitchable neuromuscular
ligands for which we coined the term “azocuroniums”. These compounds switched between
the (E)- and (Z)-isomers by light irradiation at 400-450 nm and 335-340 nm, respectively.
Meta-azocuroniums were potent nicotinic ligands with a clear selectivity for the muscular
nAChRs compared to neuronal α7 and α4β2 subtypes, showed good solubility in
physiologic media, negligible cell toxicity, and would not reach the CNS.
Electrophysiological studies in muscle-type nAChRs expressed in Xenopus laevis oocytes
showed that (E)-isomers were more potent than (Z)-forms. All meta-azocuroniums were
neuromuscular blockers, with the exception of the pyrrolidine derivative that was an
agonist. These new meta-azocuroniums, which can be modulated ad libitum by light, could
be employed as photoswitchable muscle relaxants with fewer side effects for surgical
interventions and as tools to better understand the pharmacology of muscle-type nAChRs.
4

1. Introduction
Nicotinic acetylcholine receptors (nAChRs) are ligand-gated ion-channels physiologically
activated by acetylcholine (ACh), which consist of five protein subunits organized around a
central pore. They constitute a superfamily of receptors that are critically involved in
cellular electrical signaling in the nervous and skeletomuscular systems. In general,
nAChRs are found in either neurons or skeletal muscles, confined in synapses where they
relay unidirectional electrical information from one cell to another. Neuronal nAChRs are
made by combinations of alpha (α2-α10) and beta subunits (β2-β4) with variable
stoichiometry or by alpha homo-pentamers (α7-α9). In the mammalian brain the most
common neuronal subtypes are α7 and α4β2 nAChRs, which are involved in numerous
physiological functions such as cognition, learning and memory, physical stimulation,
cerebral blood flow and metabolism [1,2].
In contrast to the neuronal counterparts, muscular nAChRs are composed of four types of
proteins: two α1, one β1 and one gamma (γ) subunits, accompanied by either an epsilon (ε)
or a delta (δ) subunit, depending on the developmental stage of the muscle [3,4].
Embryonic muscle receptors have a δ subunit, whereas in adults the ε type is present [5,6].
In skeletal muscles, nAChRs are confined within the end-plate terminal, which is made of
specialized synapses linking a motor neuron to individual muscle fibers. Activation of
nAChRs initiates the electrical signal that triggers action potentials, leading to muscle
contraction.
Natural extracts containing neuromuscular blocking agents (NMBAs), such as curare
alkaloids, have been used for centuries by South American indigenous tribes for hunting.
5

Many plant extract derived-NMBAs have been identified, purified and used as leads in the
development of muscle relaxants for surgical procedures and other applications, including
cosmetics (Figure 1). Despite the benefits that these compounds may offer in clinic, some
of them are known to produce adverse side-effects in patients. For instance, the
aminosteroid pancuronium is a synthetic curare-mimetic used as a muscle relaxant to
facilitate endotracheal intubation, which causes several undesirable effects such as
tachycardia, increase in blood pressure, respiratory depression, and apnea [7]. In this
context, we recognized that the development of novel NMBAs analogues with manageable
or null side-effects would produce safer new drugs for surgical interventions and also
provide a breakthrough in understanding the pharmacology of muscle nAChRs. Towards
this ideal outcome, we set out to design novel ligands for nAChRs that, in addition to their
selectivity towards muscular subtypes, they have photoswitchable properties so that they
can be activated and deactivated by light irradiation at different wavelengths. The quick
modulation of the biological activity of the photoisomerizable compounds by light allows a
fine adjustment of their spatial and temporal activity, which can hardly be achieved with
non-photoswitchable ligands [8].
Each muscular nAChR has two binding sites for ACh in the extracellular domain, at αδ and
either αε or αγ subunit interfaces. In these interfaces, aromatic residues (tryptophan and
tyrosine) provide negative π-electron clouds that can electrostatically interact with the
positively charged ammonium group [-N+(CH3)3] of the neurotransmitter ACh [9,10]. In
these receptors, the cation-π interaction is essential for binding and action. Regarding the
quaternary ammonium group, it is known that compounds with less sterically hindered
amino groups tend to be agonists, such as carbachol, suxamethonium, and bisQ. In contrast,
6

compounds with larger and hydrophobic substitutions, including N-carbocycle groups tend
to be antagonists, such as the case of benzylisoquinoliniums (e.g., atracurium, mivacurium
and cisatracurium) or aminosteroids (e.g., pancuronium, rocuronium and vecuronium)
(Figure 1) [11,12].
In the early 1970s, the Erlanger’s group reported that bisQ activates nicotinic receptors in
the electroplax of electric fish Electrophorus electricus. As other azobenzene-based
compounds, bisQ is photo-isomerizable, being a potent nAChR activator when it is a (E)-
isomer, but not while being a (Z)-isomer [13,14]. From these pioneer studies many
photoswitchable ligands have been developed and applied to a wide range of biological
targets, such as G protein-coupled receptors (GPCRs), ion channels, kinases, proteases, etc
[15-22]. Therapeutic efficacies of photo-isomerizable compounds have been demonstrated
on the cellular level and some studies have progressed to live animals, such as the vision
restoration with diazene-based photoswitchable drugs in blind mice [23-26].
In recent years, new photoswitchable nicotinic ligands have been developed by adding a
single trimethylammonium head to one extreme of azobenzene scaffolds [27-29]. This
strategy produced preferentially agonists, with selectivity for activating neuronal nAChRs
compared to muscle-type receptors. For example, the trimethylammonium derivative
azocholine is a selective photoswitchable agonist for neuronal α7 nAChRs compared to
muscular receptors [30].
Our group is now interested in developing new neuromuscular blockers that, in addition to
their photoswitchable properties, displayed selectivity towards muscle-type nAChRs
compared to neuronal receptors. From bibliographic precedents, we rationalized that we
7

could generate photoswitchable antagonists for muscular nAChRs by the replacement of
two methyl groups of both quaternary ammonium groups of the muscle agonist bisQ by N-
carbocycles, commonly present in curare NMBAs. As a proof of principle, here we have
designed, synthesized, and tested a series of novel muscular nAChR ligands, named
azocuroniums (1) due to their structural similarity to curare-like drugs, the activity of which
can be easily modulated by light (Figure 1) and which functional character depends on
some structural features evaluated here.
Figure 1. Structures of muscle-type nAChRs agonists (carbachol, suxamethonium and (E)-
bisQ) and antagonists (atracurium and pancuronium). General structure of the new
azocuroniums (1) on their (E)- and (Z)-isomers.
8

2. Results and discussion
2.1. Synthesis of azocuroniums
Reaction of commercially available meta- or para-nitrobenzyl bromide (1 equiv.) with the
corresponding amine (pyrrolidine, piperidine, azepane, or 1,2,3,6-tetrahydropyridine) (1.2
equiv.) in basic conditions under mw irradiation at 120 °C for 10 min, gave the
corresponding nitro derivative m-2(a-d) and p-2(a,b) in excellent yield (91-99%) (Scheme
1). The subsequent treatment of these nitro compounds with a solution of LiAlH4 in diethyl
ether (5 equiv.) afforded the required azobenzenes m-3(a-d) and p-3(a,b), in one-step and
in moderate to excellent yields (61-97%). Asymmetric azocompound m-4b was synthesized
from 3-(piperidin-1-ylmethyl)aniline [31] and commercial nitrosobenzene in high yield
(93%) by a Mills reaction. All azo intermediates m-3(a-d), p-3(a,b) and m-4b were purified
by preparative thin-layer chromatography (TLC), obtaining the (E)-isomer as the major
isomer, as deduced from HPLC-MS and NMR data [see (E) / (Z)-ratio in Scheme 1].
Then, the intermediate tertiary amines were methylated with CH3I (1.3 equiv.) at 120 °C for
12 min. under mw irradiation, obtaining the desired quarternary ammonium azobenzenes
m-1(a-d), p-1(a,b) and m-5b, which were purified by crystallization from methanol.
Remarkably, these azocuronium salts were isolated in their (E)-forms in high isomeric
purity, equal or greater than 95% [see (E) / (Z)-ratio in Scheme 1]. This fact could be the
result of the sum of a high temperature (120 ºC) in the mw-oven, which could favor the
formation of the thermodynamically more stable (E)-isomer, and the crystallization process,
which could enrich the major species.
9

I
p N
N
CH3
NO2
m
NO2
(a)
o
i
(b)
(c)
N
N
N
N
N
Br
(E)/(Z)-ratio
(E)/(Z)-ratio
H3C
N
N
m-2a (96%)
p-2a (98%)
m-2b (93%)
p-2b (91%)
m-2c (>99%)
m-2d (>99%)
m-3a (90%) 80/20
p-3a (93%) 95/5
m-3b (97%) 75/25
p-3b (95%) 90/10
m-3c (61%) >99/0
m-3d (49%) 90/10
m-1a (89%) 98/2
p-1a (88%) 99/1
m-1b (85%) 95/5
p-1b (82%) 95/5
m-1c (90%) 98/2
m-1d (78%) >99/0
I
p
m'
m
o'
CH3
NO
N
N
o
i
(d)
(c)
+
N
N N
N
N
H2N
I
o''
m-4b (93%)
(E)-purity > 99%
m-5b (58%)
(E)-purity > 99%
m''
p'
6
5
N
N
N
N
N
: a =
b=
c =
d =
Aryl substitution: meta (m) or para (p)
2
4
2
4
2
2
4
3
3
3
3
Scheme 1. Reagents and conditions: (a) Amine (pyrrolidine, piperidine, azepane, or
1,2,3,6-tetrahydropyridine) (1.2 equiv.), K2CO3, acetone, mw, 120 ºC, 10 min; (b) LiAlH4
in Et2O (5 equiv.), N2, -78 °C to rt; (c) CH3I (1.3 equiv.), DMF, mw, 120 ºC, 12 min; (d)
Acetic acid, toluene, 60 ºC, overnight.
All compounds were characterized by their analytical (HPLC, HRMS) and spectroscopic
data (1H NMR, 13C NMR). Complete NMR assignment of their hydrogen and carbon atoms
1
were made by H – 13C two-dimensional diagrams, mainly HSQC (heteronuclear single
quantum correlation) and HMBC (heteronuclear multiple bond correlation) (See
Supplementary Information for further details).
10

2.2. Evaluation of thermodynamic solubility, in vitro CNS-penetration (PAMPA-BBB
assay), and cell toxicity
The thermodynamic solubility of azocuronium salts m-1(a-d), p-1(a,b) and m-5b was
determined in pH 7.4 phosphate 45 mM buffer following described protocols [32,33]
(Table 1). As expected and due to their ionic character, in physiological medium all
azocuronium salts showed high solubility values (5.5 – 38.4 mM). Remarkably, all meta-
derivatives were found to be more soluble than their para- counterparts, probably due to the
higher polarity of the former.
Table 1. Thermodynamic solubility (mM) in water at pH 7.4 and in vitro CNS permeability
data of azocuroniums m-1(a-d), p-1(a,b) and m-5b.a
Solubility at pH 7.4 (mM)
PAMPA-BBB (Pe, 10-6 cm s-1)
m-1a
p-1a
m-1b
p-1b
m-1c
m-1d
m-5b
36.3 ± 3.1
14.8 ± 1.2
38.4 ± 3.5
34.9 ± 3.2
5.9 ± 0.5
< 1.0 (cns -)
< 1.0 (cns -)
< 1.0 (cns -)
< 1.0 (cns -)
< 1.0 (cns -)
16.3 ± 0.1
5.5 ± 0.1
< 1.0 (cns -)
2.0 ± 0.2 (cns +/-)
aResults are the mean ± SD of three independent experiments. bPredictive CNS penetration:
cns- denotes compounds that are no able to penetrate into the CNS; cns +/- denotes
uncertain CNS penetration.
11

To check if new azocuroniums m-1(a-d), p-1(a,b) and m-5b could be able to reach the
CNS, we used the in vitro parallel artificial membrane permeability assay to model the
blood-brain barrier (PAMPA-BBB) described by Di et al. [34], and partially modified and
validated by us [35-40]. The passive CNS-permeation of azocuroniums through a lipid
extract of porcine brain was measured at rt. In each experiment, 11 commercial drugs of
known brain permeability were also tested and their permeability values normalized to the
reported PAMPA-BBB data (See Table S1 in the Supporting Information). According to Pe
values previously described [34], compounds with Pe exceeding 4·10-6 cm s-1 would be able
to cross the BBB (cns+), whereas those displaying Pe less than 2·10-6 cm s-1 would not
reach the CNS (cns-) via passive diffusion. Between these two values the prediction is
uncertain (cns +/-). While the asymmetric derivative m-5b with only one positive charge
displayed an uncertain CNS penetration, the azocuronium salts with two positive charges,
m-1(a-d) and p-1(a,b), were predicted not able to cross the BBB (Table 1).
Cytotoxicity of azocuroniums salts m-1(a-d), p-1(a,b) and m-5b was determined in Raw
264.7 macrophages, using the quantitative colorimetric assay MTT (3-[4,5-dimethylthiazol-
2-yl]-2,5-diphenyltetrazolium bromide) [41]. Compounds at a concentration of 10 μM were
incubated with the cell line at 37 ºC during 24 h and then, the mitochondrial activity of
living cells was measured by the absorbance change at 540 nm. All tested compounds
showed cell viabilities in the same range than the basal experiment, pointing out that none
of them showed any significant cytotoxic effect in this model.
As required for further development as potential therapeutic or research tools, we
demonstrate that the new azocuronium salts m-1(a-d) and p-1(a,b) are soluble in
12

physiological medium, lack toxicity in vitro and are predicted to be CNS-impermeable via
passive diffusion.
2.3. Photochemical characterization by UV/vis and 1H-NMR
To study the photochemical behavior of new azocuroniums, we performed their
spectroscopic characterization by UV/vis and 1H-NMR.
UV/vis spectra of azocuroniums m-1(a-d), p-1(a,b) and m-5b were recorded in water at rt
and the photochemical characterization of m-1b is given as an example (see Figure S1 in
Supplementary Information for further details). The thermally equilibrated spectrum of m-
1b (Figure 2B, black line) showed the two absorption bands characteristic of azobenzene
derivatives. The most intense band maxima at 315 nm was assigned to the π-π* electronic
transition of the (E)-azobenzene moiety, while the other maxima around 425 nm is due to
the forbidden n-π* electronic transition. After irradiation with 335-340 nm light using a
LED-based source (UV LED-light), the amplitude of the 315-nm band decreased along
increasing irradiation time, while the amplitude of the 425-nm peak increased (red line). As
previously reported, this spectral change is due to the (E)-to-(Z) isomerization of
azobenzene molecules, which was reverted by irradiating with light of 400-450 nm (blue
LED, blue line in Figure 2B) [14]. The insert in Figure 2B shows details of the UV/vis
spectra between 375 nm and 525 nm.
To check the stability of new azocuroniums in the photoswitching process, we performed
three consecutive cycles of interconversion, monitoring the absorption at 315 nm of m-1b
13

throughout alternating irradiation with blue LED and UV LED (Figure 2C). After each
cycle, each species showed its characteristic absorption value namely, 0.54 AU for the (E)-
isomer and 0.19 AU for the (Z)-form. Therefore, no evidence of photo-fatigue was
observed under conditions close to functional studies in muscle-type nAChRs expressed in
Xenopus laevis oocytes, where no more than two photoswitching cycles were used (see
Section 2.5).
Figure 2. Photoswitched properties of m-1b. A) (E)-to-(Z) Isomerization of the azobenzene
scaffold. B) UV/vis spectra of m-1b (50 μM) in water at rt, thermally relaxed (black line),
irradiated with UV LED (red trace) and blue LED (blue trace). The inset shows a
magnification of the n−π* band. C) Reversibility study of the photoisomerization of m-1b
over three cycles, monitoring the absorption at 315 nm throughout alternating irradiation
with blue LED (blue circle) and UV LED (red circle). D) Expansion of 1H-NMR spectra of
azocuronium m-1b at the indicated conditions.
14

The (E)-(Z) conversion of azocuroniums was also quantified by 1H-NMR technique, using
deuterium oxide (D2O) as solvent. Solutions were irradiated under UV or blue LED-light
1
into a round-bottom flask and then transferred to a NMR tube for registering H-NMR
spectra under darkness. The isomeric percentage was calculated from the integral of signals
attributable to common protons in each species. The photo-induced (E)-to-(Z) isomerization
of azocuroniums is more appreciable in the aromatic area, as explained below for m-1b (an
1
expansion of this area is shown in Figure 2D). Firstly, the H-NMR spectrum of the non-
irradiated m-1b showed two groups of aromatic signals integrated for 8 protons, which
were assigned to the (E)-azobenzene isomer (upper spectrum in Figure 2D). After LED
irradiation under UV light at 335-340 nm, the photostationary state (PSSUV) containing
96% of (Z)-isomer was reached (see Supplementary Information, page S23). Afterward, the
solution was irradiated at 400-450 nm (blue LED) obtaining again the (E)-isomer in around
90% yield.
2.4. Radioligand binding assays at nAChRs
(E)-Azocuroniums m-1(a-d), p-1(a,b) and m-5b were assayed in human nicotinic receptors
in radioligand-displacement experiments, using muscular nAChR and two neuronal-type
nAChRs (α7 and α4β2) [42]. The radioligands used were [125I]α-bungarotoxin (for muscle-
type and α7 nAChRs) and [3H]cytisine (for α4β2 nAChRs). The percentage of radioligand
displacement of compounds at a single concentration (10 µM) was calculated for each
receptor type and then, binding constants (Ki) were calculated for azocuroniums that
produced a displacement percentage above 60%, using a range of 5 different concentrations
15

in 3 independent experiments (Table 2 and Figure S2 for further details). Standard
reference compounds were α-bungarotoxin (muscle-type nAChR), epibatidine (α7 nAChR)
and nicotine bitartrate (α4β2 nAChR), which were tested in each experiment at several
concentrations to obtain competition curves, from which Kis were calculated.
Table 2. Binding constants at human nAChRs of muscular and neuronal-type (α7 and
α4β2), or percentage of radioligand displacement (in brackets) at the specified
concentration, of azocuroniums m-1(a-d), p-1(a,b) and m-5b.a
Ki (nM)
Muscle-type
Neuronal α7
Neuronal α4β2
m-1a
42 ± 4
200 ± 20
35 ± 3
2500 ± 210
930 ± 80
>10,000 (31%)
>10,000 (45%)
>10,000 (7%)
1,500 ± 130
>10,000 (29%)
>10,000 (32%)
>10,000 (39%)
n.d.
p-1a
m-1b
910 ± 90
p-1b
1,100 ± 100
220 ± 20
100 ± 9
>10,000 (16%)
1,900 ± 180
730 ± 70
m-1c
m-1d
m-5b
>10,000 (45%)
1.0 ± 0.1
n.d.
>10,000 (48%)
n.d.
α-bungarotoxin
epibatidine
nicotine
120 ± 10
n.d.
n.d.
n.d.
1.5 ± 0.1
aResults are the mean ± SEM of three independent experiments. nd: not determined.
16

In general, meta-substitution in azocuronium salts favored the interactions with muscle-
type nicotinic receptors (Ki = 10-8 M) as compared to neuronal subtypes α7 (Ki = 10-6 – 10-7
M) and α4β2 (Ki > 10-5 M). Azocuroniums bearing para-substitutions were less active in all
nAChRs and showed no selectivity towards the muscular type. For instance, the pyrrolidine
derivative p-1a that was the most active para-azocuronium compound in muscular nAChR
(Ki = 200 nM) showed α7-nAChR affinity in the same order of magnitude (Ki = 930 nM).
Similarly, para-piperidine azocuronium p-1b was not selective either, this time between
muscle-type and neuronal α4β2, as its binding constants were in the same range (Ki = 1,100
and 1,400 nM, respectively).
In muscle-type nAChRs, the most potent ligands were the meta-pyrrolidine m-1a and the
meta-piperidine m-1b azocuroniums, with binding affinities in the nanomolar range (Kis =
42 nM and 35 nM, respectively). These ligands also showed a clear muscle-type selectivity
as their affinities were in the micromolar range for α7 (Kis = 2,500 and 910 nM,
respectively), and above in the case of α4β2 (Kis >10,000 nM). Thus, m-1a and m-1b
displayed high binding affinity and selectivity ratios towards muscle-type nAChRs
compared to the neuronal subtypes α7 (60- and 26-fold) and α4β2 (>240- and >290-fold).
Relative to the piperidine-substituted m-1b, both the ring expansion in the meta-azepane
derivative m-1c or the introduction of a double bond to provide the meta-1,2,3,6-
tetrahydropyridine derivative m-1d, decreased the binding affinity in muscle-type nAChR
(Kis = 220 and 100 nM, respectively) and also diminished selectivity indexes between
muscular and neuronal nAChRs. Finally, asymmetric derivative m-5b was poorly active in
all nicotinic receptors assayed (Table 2), pointing out the importance of the presence of two
cationic moieties for a successful nAChRs binding.
17

2.5. Functional characterization in nAChRs
We evaluated the effect of meta-azocuroniums in muscular nAChR expressed in Xenopus
laevis oocytes. To activate nAChRs, a solution of ACh (50 μM, 2.3-18.4 nL) was delivered
on the surface of the oocytes using a nano-injector (Figure 3A). This small volume of ACh
(“puff”) was washed out by perfusing the oocytes with a recirculating recording solution
propelled by a pump attached to the chamber (Figure 3A). Above the oocyte, two LEDs
were placed to drive photo-isomerization. These LEDs, hereafter referred to as “UV LED”
and “blue LED”, had their emission spectra centered at 335-340 nm and 400-450 nm,
respectively. Receptor activity was assessed by measuring ionic currents through the
plasma membrane mediated nAChR using the two-electrode voltage-clamp (TEVC)
technique (Figure 3A). For current recordings, the membrane potential was held at -60 mV.
Thus, nAChR activation will result in a downward deflection of the current trace, as the
result of the net inward current through the receptors.
Under voltage-clamp, oocytes have a basal conductance, yielding relatively small
background currents in the order of 0.05-0.2 μA. Oocytes displaying higher background
currents were discarded. An ACh puff was applied after the recirculation pump was turned
off. This allowed ACh to reach the membrane which activated nAChR, resulting in an
increase of the membrane conductance (Figure 3B). Then, the pump was turned on,
washing ACh from the oocyte’s vicinity, deactivating nAChR. The deactivation of nAChR
was made apparent by the decrease of the inward current amplitude (Figure 3B). Although
the decay of the current amplitude was due to removal of ACh by activation of the pump,
we cannot rule out the possibility that a small fraction of this is due to receptor
desensitization. To address this we observed that with “no pumping” the decay of the
18

current was much slower (see Figure S3 in Supplementary Information), indicating that
desensitization plays a minor role in our observations. Thus, we concluded that this
approach allows us to evaluate the effect of the meta-azocuronium salts on muscular
nAChRs.
Figure 3. A) TEVC technique. Ionic currents were measured in Xenopus oocytes
expressing embryonic mouse muscular nAChR. The membrane potential was held at –60
mV. Two LEDs (UV and blue) were used to drive isomerization, a nano-injector to deliver
ACh (2.3-18.4 nL, 50 µM, 46 nL/s) on the surface of the oocytes, and a pump to circulate
the solution and wash out ACh from oocytes. B) Record obtained by adding an ACh puff
while the pump is turned off, as a control (no compound). The current increases due to the
receptor activation.
2.5.1. Piperidine-derived azocuronium salt m-1b blocks muscular nAChR
Under voltage-clamp, inward currents were observed in oocytes expressing muscle-type
nAChR upon application of an ACh puff (Figure 4A, green trace). This response was
inhibited by the addition of the meta-piperidine azocuronium m-1b (5 μM) to the recording
19

solution. Such inhibition was observed by a decrease in the maximum amplitude of the
currents (Figure 4A, black trace). The inhibitory effect of m-1b on current amplitude was
partially reverted when oocytes and the surrounding solution were irradiated with the UV
LED for 5 minutes before applying a ACh puff and during the recording (Figure 4A, red
trace). The inhibitory effect of m-1b was recovered by irradiating with the blue LED. These
observations suggested that m-1b acted on the deactivated (closed) receptors.
To understand whether compound m-1b could act on activated (open) receptors, nAChR
currents were recorded in the presence of m-1b (0.5 μM) which was held in its (Z)-isomer
by constant irradiation with the UV LED. We performed this assay with no perfusion to
prolong the activating effect of ACh on the receptor. Following nAChR activation and blue
LED irradiation [(E)-isomer], a faster decay of the ACh-induced current was observed
(Figure 4B, red trace) with respect to the case when constant UV radiation was applied
(Figure 4B, black trace). Since the perfusion pump was switched off, the decay observed
during constant UV LED was likely due to receptor desensitization. Swapping the UV LED
for the blue LED radiation [(E)-isomer] caused a faster decrease in the current amplitude
(Figure 4B, red trace), indicating that m-1b was like affecting receptors in any isomeric
form.
20

Figure 4. Effects of m-1b in the electrical signal at indicated conditions. A) A “puff” of
ACh (green trace) evoked currents in oocytes that decreased after adding (E)-m-1b,
showing an almost complete inhibition of the receptors (black trace). After that, UV LED
lighting [(Z)-isomer] partially reversed the effect of (E)-isomer (red trace). B) Currents
evoked by the (Z)-isomer (black trace); exchange the UV LED for the blue LED radiation
[(E)-isomer] caused a faster decrease in the current amplitude (red trace).
To further characterize the action of m-1b on nAChR, we evaluate the relationship between
drug concentration and receptor inhibition. For this purpose, we performed recordings of
ACh activity in the presence of the thermally relaxed m-1b [(E)-isomer]. As the
concentration of m-1b increased, the response to ACh puff decreased (Figure 5A, left). This
inhibitory effect was partially reverted upon irradiation with UV LED for 5 minutes [(Z)-
isomer] (Figure 5A, right). To assess the inhibitory effect of m-1b in detail, we performed
recordings in the presence of a range of concentrations of m-1b (0.05-20 μM) thermally
relaxed [(E)-isomer] (Figure 5B, black trace) and after UV radiation [(Z)-isomer] (Figure
5B, red trace) and we plotted the average of current amplitude normalized with respect to
21

recording in the absence of the drug. The current-vs-dose plots were fitted to a simple one-
site binding model, yielding an apparent half-maximum inhibitory constant (Ki) of 0.46 ±
0.06 μM (n = 24) for the thermally relaxed m-1b (Figure 5B, black trace) and 2.6 ± 0.5 μM
(n = 24) for the UV-irradiated m-1b (Figure 5B, red trace).
Figure 5. Compound m-1b is a light-dependent antagonist. A) AChR currents recorded
upon application of an ACh “puff” (50 μM, 9.2 nL) in the absence (black trace) and
presence of 0-5 μM m-1b thermally relaxed (left) or while irradiating with the UV LED. B)
The maximum amplitude of the currents was normalized with respect to the maximal
amplitude observed in the absence of m-1b when it was thermally relaxed (black symbols)
or after irradiation at 340 nm (red symbols).
We noticed that the variability in current amplitude for the UV-irradiated drugs was slightly
higher than for their thermally relaxed counterparts. Although we consider this issue to be
beyond the scope of the present study, we argue that such increase in variability could be
22

attributed to variations in the effective intensities of the UV radiation that reached oocytes.
We did not evaluate this issue systematically, yet we noticed that it depended on the angle
of the incidence of the LED beam and the recording chamber. For this reason, we opted to
skip to further analyzing the (E)-to-(Z) isomerization time course for this study.
2.5.2. Pyrrolidine-based azocuronium salt m-1a acts as a muscular nAChR agonist
Next, we proceed to evaluate the compound m-1a which differed from m-1b on the smaller
ring size of the quaternary amine. We observed that exposing the oocytes to thermally
relaxed m-1a [(E)-isomer] produced the activation of the nAChR current and thus, we
proceeded with a different approach. Voltage-clamped oocytes were exposed to m-1a under
constant UV irradiation to keep the compound as (Z)-isomer. To test the effect of the (E)-
isomer, a 5-second pulse of blue radiation was applied while turning off the UV LED. After
5 seconds, the UV LED was turned back on [(Z)-isomer]. Turning on the blue LED
activated nAChRs, as a robust inwards current was detected (Figure 6A). This agonist
behavior was observed at different concentrations of m-1a, ranging from 0.5 to 10 μM.
Activation of the current showed a clear concentration dependence, with an apparent half-
maximum activity constant of 4.2 ± 0.4 μM (n = 4) (Figure 6B).
23

Figure 6. Azocuronium m-1a is a light-dependent agonist of the nAChR. A) Current
recordings in the presence of 0.5-10 µM m-1a. Constant irradiation with the UV LED made
m-1a inactive as agonist. Inducing the isomerization to the (E)-isomer with a 5-second
pulse of blue LED resulted in a robust increase of the nAChR current. This activation was
readily reverted by turning back to irradiation with the UV LED. B) Average normalized
current measure following 5 seconds of activation showed a clear concentration-
dependence for activation.
Per our initial hypothesis, the hydrophobic character of the molecule’s quaternary amine
seems to determine whether the agonist/antagonist activity profile. To this point, the
piperidine-based m-1b functions as an antagonist of nAChR, while the pyrrolidine-based
m-1a behaves as an agonist.
24

2.5.3. Azepane-derived azocuronium salt m-1c behaves as a muscular nAChR antagonist
The compound m-1c bears a larger hydrophobic moiety (azepane fragment) associated to
the quaternary amine. Consistent with our initial hypothesis, m-1c inhibited ACh-induced
current in oocytes expressing nAChRs. Fitting the normalized current-vs-dose plots to a
one-site binding model for the thermally relaxed m-1c [(E)-isomer] yielded a Ki of 0.33 ±
0.04 μM (n = 26) (not shown). However, the plot for the UV-irradiated drug was not
properly fitted to the one-site model, leading us to consider the fact that muscular nAChR
have two distinct ACh binding sites with different pharmacology. Therefore, we chose to fit
plots to a Hill equation, from which we calculated Kis of 0.30 ± 0.02 μM and 2.6 ± 0.5 μM
for the (E) and (Z)-isomer, respectively (Figure 7). As expected, the Hill coefficient for the
thermally relaxed m-1c was near one (0.97 ± 0.05), consistent with the previous fit. In the
case of the UV-irradiated m-1c, the Hill coefficient was 0.56 ± 0.07. It is well-known that
ACh binding and activation of the muscular nAChR is highly cooperative, where the
binding of one ACh leads to an increase in affinity for the second ACh molecule [9,43].
Based on this we speculate that, assuming that m-1c binds to the ACh sites, the interaction
of m-1c on one site induces a reduction of the affinity of the other site for this drug. Since
both sites are allosterically coupled, we speculate that there were at least two m-1c
molecules bound to each nAChR, with the first molecule inducing a conformational change
that leads to a decrease in the affinity of the receptor for the second one. Accordingly, we
speculate that the (E)-isomer of m-1c (and m-1b) to occupy one of the ACh binding sites,
hindering further steps in the activation of the receptor. However, the (Z)-isomer of m-1c
seems to be competed out by ACh so that the binding of ACh allosterically decreases the
affinity of the receptor for m-1c. Notwithstanding the importance of determine the
25

underlying process, pinpointing the mechanism responsible for this reduction in the Hill
coefficient was beyond the scope of this study.
Figure 7. Inhibition of nAChR current by m-1c when thermally relaxed (black squares) and
under irradiation with UV LED (red circles). The current-vs-[m-1c] plots were fitted to a
Hill equation, yielding a Ki value of 0.30 ± 0.02 μM and 2.6 ± 0.5 μM, respectively. The
corresponding Hill coefficients were 0.97 ± 0.05 and 0.56 ± 0.07 (n = 26).
2.5.4. The rigidity of the N-cycle is important for activity
Next, we explored the effect of the rigidity of the N-cycle on the action of the drug. For this
purpose, we used piperidine derivative m-1b as reference and compared it to the double-
bond bearing piperidine compound m-1d. The latter N-cycle has less flexible, more planar
geometry than the one of m-1b. Like this reference compound, m-1d behaved as an
antagonist of nAChR. Both the thermal relaxed and UV-irradiated m-1d showed very
similar Ki values of 0.84 ± 0.09 μM and 0.88 ± 0.11 μM (n = 15), respectively (Figure 8).
26

In spite of the similarity in the receptor’s affinity for both isomers, the efficacy for drug
action was isomer-dependent. Per the fitting analysis, the fractions of active receptors at
saturating concentration of m-1d were 0.11 ± 0.04 and 0.35 ± 0.03 for the thermally relaxed
and UV-irradiated m-1d, respectively. This indicated that this drug was likely operating
with a distinct mechanism that those ones described above. Furthermore, it is noteworthy,
that Hill coefficients fitted for both isomers were similar, 3.5 ± 1.3 and 2.9 ± 1.1, and
higher than those of the drugs tested here. These values indicated that a highly cooperative
process underlies the inhibition by m-1d. Understanding the nature of this process was
beyond the scope of this study.
Figure 8. Inhibition of nAChR current by thermally relaxed m-1d (black squares) and
under irradiation with UV LED (red circles). Fitting to a Hill equation, yielded a Ki of 0.84
± 0.09 μM and 0.88 ± 0.11 μM, respectively. The current fraction at maximum inhibition
was 0.11 ± 0.04 and 0.35 ± 0.03 for the thermal relaxed and UV-irradiated m-1d,
respectively. The corresponding Hill coefficients were 3.5 ± 1.3 and 2.9 ± 1.1 (n = 15).
27

3. Conclusions
Azocuroniums were designed by adding two N-methyl-N-carbocyclic quaternary
ammonium groups to an azobenzene scaffold in meta- or para-position. These new
photoswitchable compounds showed good solubility in physiologic media, negligible
toxicity in vitro and as expected, they would not penetrate into the CNS. They can be easily
photoisomerized between the (E)- and (Z)-forms by irradiation at 400-450 nm (blue LED)
and 335-340 nm (UV-LED), respectively.
In radioligand binding assays at nAChRs, meta-substituted azocuroniums were more potent
and selective towards neuromuscular receptors than their para-substituted counterparts.
Derivatives with smaller cationic heads, namely the meta-pyrrolidine m-1a and the meta-
piperidine m-1b azocuroniums emerged as the most potent and selective ligands in muscle-
type nAChRs, with binding affinities in the nanomolar range (Kis = 42 nM and 35 nM,
respectively). Moreover, they presented lower affinities for α7 (Kis = 2,500 and 910 nM,
respectively) and were almost inactive in α4β2 (Kis >10,000 nM), showing interesting
subtype selectivity. In contrast, azocuroniums with increased volume or rigidity in the N-
carbocycle, namely the meta-azepane and the meta-1,2,3,6-tetrahydropyridine derivatives
(m-1c and m-1d), showed reduced binding constants in muscular-type nAChR (Kis = 220
and 100 nM, respectively) with diminished selectivity indexes between muscular and
neuronal nAChRs.
Using the two-electrode voltage-clamp technique, we evaluated the functional activity of
meta-azocuroniums m-1(a-d) in muscular nAChR expressed in Xenopus laevis oocytes. By
irradiation with blue-LED or UV-LED while recording electrical currents, in all cases the
28