Binding-induced fluorescence of serotonin transporter ligands: A spectroscopic and structural study of 4-(4-(Dimethylamino)phenyl)-1- methylpyridinium (APP+) and APP+ analogues

Article abstract of DOI:10.1021/cn400230x

The binding-induced fluorescence of 4-(4-(dimethylamino)-phenyl)-1- methylpyridinium (APP⁺) and two new serotonin transporter (SERT)-binding fluorescent analogues, 1-butyl-4-[4-(1-dimethylamino)phenyl]- pyridinium bromide (BPP⁺) and 1-methyl-4-[4-(1-piperidinyl)phenyl]- pyridinium (PPP⁺), has been investigated. Optical spectroscopy reveals that these probes are highly sensitive to their chemical microenvironment, responding to variations in polarity with changes in transition energies and responding to changes in viscosity or rotational freedom with emission enhancements. Molecular docking calculations reveal that the probes are able to access the nonpolar and conformationally restrictive binding pocket of SERT. As a result, the probes exhibit previously not identified binding-induced turn-on emission that is spectroscopically distinct from dyes that have accumulated intracellularly. Thus, binding and transport dynamics of SERT ligands can be resolved both spatially and spectroscopically.

Full text of DOI:10.1021/cn400230x

Research Article
pubs.acs.org/chemneuro
Binding-Induced Fluorescence of Serotonin Transporter Ligands: A
Spectroscopic and Structural Study of 4‑(4-(Dimethylamino)phenyl)-
1-methylpyridinium (APP+) and APP+ Analogues
James N. Wilson,*,† Lucy Kate Ladefoged,‡ W. Michael Babinchak,† and Birgit Schiøtt*,‡
†Department of Chemistry, University of Miami, Coral Gables, Florida 33124, United States
‡inSPIN and iNANO Centers, Department of Chemistry, Aarhus University, 8000 Aarhus C, Denmark
S
* Supporting Information
ABSTRACT: The binding-induced fluorescence of 4-(4-(dimethy-
lamino)-phenyl)-1-methylpyridinium (APP+) and two new seroto-
nin transporter (SERT)-binding fluorescent analogues, 1-butyl-4-[4-
(1-dimethylamino)phenyl]-pyridinium bromide (BPP+) and 1-
methyl-4-[4-(1-piperidinyl)phenyl]-pyridinium (PPP+), has been
investigated. Optical spectroscopy reveals that these probes are
highly sensitive to their chemical microenvironment, responding to
variations in polarity with changes in transition energies and
responding to changes in viscosity or rotational freedom with
emission enhancements. Molecular docking calculations reveal that
the probes are able to access the nonpolar and conformationally restrictive binding pocket of SERT. As a result, the probes
exhibit previously not identified binding-induced turn-on emission that is spectroscopically distinct from dyes that have
accumulated intracellularly. Thus, binding and transport dynamics of SERT ligands can be resolved both spatially and
spectroscopically.
KEYWORDS: Serotonin, SERT, fluorescence, induced fit docking, quantum polarized ligand docking, LeuT
erotonin, or 5-hydroxytryptamine (5-HT), is a chemical
Smessenger with a diverse set of functions in the brain as
well as the periphery.1−3 5-HT distribution and action is
regulated by the serotonin transporter (SERT or 5-HTT),
which is responsible for removing extracellular serotonin from
serum and synaptic clefts following 5-HT release.4,5 SERT
expression and localization has traditionally been ascertained by
immunohistochemical techniques5−7 or radiolabeled ligands.3,8
Recently, fluorescent substrates of SERT9,10 and related
monoamine transporters (MATs) have been reported.11−13
Figure 1. Chemical structures of serotonin (5-HT), ASP+, APP+, and
two new fluorescent SERT ligands, PPP+ and BPP+.
These probes and other fluorescent reporters have proven
useful for assessing MAT activity in mixed cell populations,14
visualizing neurotransmitter release,13,15,16 and studying bind-
ing ratios and uptake kinetics.11,12
high sensitivity of ASP+ is the ability of the two aromatic rings
to twist from coplanarity, that is, increase the dihedral angle
between the aniline and pyridinium rings. Twisting to 90° or a
perpendicular geometry effectively turns the emission of ASP+
off. Twisting can be limited in viscous solvents or in
geometrically confined environments. The binding of ASP+ to
the norepinephrine transporter (NET) places the probe in a
solvent excluding and more restrictive environment compared
with bulk solution.17 This enables visualization of binding and
displacement reported by DeFelice and co-workers.11,12 ASP+
One important group of fluorescent MAT substrates is the
class of the environmentally responsive stilbazolium probes,
such as 4-(4-(dimethylamino)-styryl)-N-methylpyridinium
(ASP+), which exhibit so-called “turn-on” emission. The high
sensitivity of ASP+ toward its chemical microenvironment is the
result of two features of its molecular architecture (Figure 1).
First, the presence of an electron donating dimethyl amino
group coupled opposite the electron withdrawing N-methyl-
pyridinium ring yields a structure with strong charge transfer
character in the excited state. More polar solvents or
environments help to stabilize this charge transfer state
resulting in a shift to lower emission energies as well as
emission quenching. The second feature that contributes to the
Received: December 18, 2013
Accepted: January 24, 2014
Published: January 24, 2014
© 2014 American Chemical Society
296
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
also exhibits emission upon binding to mitochondria following
transport into the cell.
Recently, 4-(4-(dimethylamino)phenyl)-1-methylpyridinium
(APP+, Figure 1) has been reported as a substrate of SERT.9
APP+ shares several structural features with ASP+ and is a
highly sensitive turn-on fluorescent probe. However, in contrast
to ASP+, APP+ was reported to exhibit turn-on fluorescence
neither when binding to SERT9 nor when exposed to cells
expressing the dopamine transporter (DAT).18 Like ASP+,
APP+ does exhibit fluorescence upon binding to mitochondria
and nucleoli after transport into the cells.9,18 This discrepancy
in the emission behavior of ASP+ and APP+ is intriguing and
may reflect alternate binding modes of the two probes,
differences in the structural features of the DAT, SERT, and
NET binding pockets, or unusual spectroscopic features of the
APP+/SERT complex.
In this contribution, we examine the spectroscopic behavior
of APP+ and two newly synthesized analogues (PPP+ and BPP+,
Figure 1) in detail. Compounds PPP+ and BPP+ were designed
to test the limits of APP+ binding and possess either a bulky
anilino “head” in the case of PPP+ or a longer alkyl “tail” in the
case of BPP+. Both PPP+ and BPP+ exhibit solvent-sensitive
emission as previously reported for APP+ and function
effectively as turn-on fluorescent probes. We next evaluated
our understanding of their photophysical behavior and
potential interactions with SERT utilizing the induced fit
docking protocol (IFD) entailing full ligand and partial protein
flexibility. Both PPP+ and BPP+ were found to share similar
binding poses and effectively sample the same pocket
environment as APP+. Based on the calculated intramolecular
torsions, binding-induced emission was predicted. Confocal
microscopy, using a 405 nm diode laser as an excitation source,
reveals that APP+ and the two analogues exhibit turn-on
emission when exposed to SERT but not when exposed to
DAT or NET. The emission of SERT-bound probes is
spectroscopically distinct from the intracellular pool. Thus,
while APP+, PPP+, and BPP+ exhibit fluorescence in cells
expressing DAT, NET, and SERT, binding to SERT produces a
unique spectroscopic signature. This fact, coupled with the
relatively high activity of ASP+ toward DAT and NET,11
suggests that the combined application of ASP+ and APP+
should enable differentiation of catecholaminergic and
serotonergic cells in complex tissues.
Figure 2. Excitation and emission spectra of PPP+ in octanol and other
solvents show the effect of polarity and visosity, that is, restricted
rotation, on emission; APP+ and BPP+ exhibit nearly identical spectra,
see Supporting Information. Emission is enhanced in more viscous
solvents, such as octanol or PEG, compared with polarity matched
solvents of lower viscosity, for example, AcCN and MeOH; the
temperature dependent emission in octanol also shows the effect of
viscosity, eliminating solvent-specific effects. The emission maximum is
sensitive to solvent polarity and is red-shifted in PEG relative to
octanol.
solvent specific interactions might contribute to the differences
in emission intensities, we also examined the effect of
temperature-induced changes in viscosity. Increasing the
temperature also leads to reduced emission, clearly establishing
the relationship between viscosity and fluorescence intensity.
Based on these results, we predicted that APP+, PPP+, and
BPP+ should exhibit turn-on emission upon binding to SERT,
provided that the probes access a pocket that limits access to
the quenched TICT state.
APP+ and analogues PPP+ and BPP+ are also sensitive to
changes in solvent polarity. For example, the emission
maximum of PPP+ is red-shifted to 505 nm in poly(ethylene
glycol) (PEG). Thus, in cases where turn-on emission is
observed, the wavelength of the emission maximum can report
the average polarity of the chemical microenvironment that the
probes experience. Binding-induced emission to SERT may be
spectroscopically distinct from emission from mitochondria.
Furthermore, if the presence of the bulky head (PPP+) or tail
(BPP+) forces these probes to adopt different binding poses,
this may also be reflected in variations in their emission profiles.
Molecular Docking and Cluster Analysis. We next
examined the binding of APP+, BPP+, and PPP+ to a homology
model of hSERT21 based on the bacterial leucine transporter
homologue, LeuT (PDB 2A65)22 to determine the overall
geometry, the inter-ring torsions, and the chemical nature of
the supporting amino acids and to evaluate whether binding
seems possible. The IFD protocol was chosen to accommodate
the large ligands in the binding site in an unbiased manor. The
resulting poses were clustered according to conformation. A
representative low energy conformation of each cluster was
further evaluated by quantum polarized ligand docking
(QPLD) in order to fully take into account the effect of the
charge distribution of the ligands as affected by the protein in
order to locate the optimal binding interactions between ligand
and hSERT.
RESULTS AND DISCUSSION
■
Photophysical Properties of APP+ and Analogues. The
photophysical properties of APP+ have been examined in detail
by Fromherz and Heilemann.19 Optical spectroscopy reveals
that PPP+ and BPP+ exhibit similar photophysical behavior to
APP+, including high sensitivity to their chemical environment
due to their ability to access a nonfluorescent, twisted
intramolecular charge transfer (TICT) state. By limitation of
torsion between the aniline and pyridinium rings, emission is
enhanced. Both probes are fluorescent in solvents of higher
viscosity, such as octanol (η = 7.24 cP), and quenched in
solvents with lower viscosity, such as acetonitrile (η = 0.34 cP),
even though they are similar in polarity (ET(30) = 46.0 and
48.3 kcal/mol for acetonitrile and octanol, respectively).20
Figure 2 shows the excitation and emission spectra of PPP+ in
octanol (purple lines): PPP+ has an excitation maximum at 426
nm and an emission maximum at 493 nm; the emission
maximum of BPP+ is slightly red-shifted to 498 nm (see
Supporting Information). To eliminate the possibility that
The IFD calculation of APP+ resulted in 95 poses, while the
IFD calculations of PPP+ and BPP+ gave 13 and 15 poses,
respectively. To ensure that the starting conformation of the
piperidine ring structure did not affect the results of the IFD,
four separate IFD calculations were performed starting from
either a chair equatorial, chair axial, twistboat equatorial, or
twistboat axial conformation of the piperidine ring.23 All four
297
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
a
Table 1. Summarized Results of IFD and QPLD Calculations for APP+, PPP+, and BPP+
IFD
QPLD
cluster
poses
average Gscore (kcal/mol)
average E-model (kcal/mol)
poses
average Gscore (kcal/mol)
average E-model (kcal/mol)
APP+
−43.57 [5.15]
−50.24 [4.73]
−59.59 [5.36]
−39.88 [8.54]
−50.00 [6.69]
−50.60 [2.68]
−49.59 [0.68]
A1
9
67
2
−7.80 [0.49]
−7.66 [0.63]
−6.08 [1.06]
−7.16 [1.21]
−4.49 [0.17]
−4.11 [0.34]
−7.69 [0.15]
3
2
1
2
1
1
4
−3.68 [4.31]
−6.38 [2.57]
0.41 [0.00]
−46.70 [4.84]
−49.82 [1.77]
−59.67 [0.00]
−57.37 [0.60]
−55.07 [0.00]
−41.94 [0.00]
−52.81 [1.62]
A2
A3
A4
6
−6.15 [2.47]
−4.80 [0.00]
−1.62 [0.00]
−6.34 [1.79]
A5
2
A6
2
A7
2
outliers
5
PPP+
−56.37 [4.61
−53.78 [6.55]
−56.41 [5.23]
P1
3
3
7
0
−8.75 [0.42]
−9.00 [0.60]
−8.48 [1.22]
2
2
2
−9.05 [0.83]
−5.52 [0.23]
−8.86 [0.17]
−66.05 [2.71]
−63.67 [3.70]
−67.08 [3.77]
P2
P3
outliers
BPP+
B1
15
0
−9.40 [0.50]
−61.22 [4.77]
1
−9.78 [0.00]
−63.65 [0.00]
outliers
a
Standard deviations are in brackets. All data is listed in Table S1 of the Supporting Information.
Figure 3. Display of all high-ranked clusters of the IFD into the binding pocket of hSERT, which is colored by atom type. The protein has carbons
shown in gray. (A) Display of clusters A1 in bright green, A2 in cyan, and A7 in yellow-green. (B) Display of cluster A4 in lime positioned within the
primary binding site. (C) Display of clusters P1 in orange, P2 in red, and P3 in purple. (D) Display of cluster B1 in blue.
298
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
Figure 4. Time course images and analyses of APP+ binding and transport in hSERT-overexpressing HEK293 cells. Cells were exposed to 2 μM
APP+ after the first image was captured at t = 0 s. A SERT inhibitor, escitalopram, was added at t = 120 s. The white bar in panel C shows the
position of the cross sections given in panel D and corresponds to 24 μm. (A) The blue channel, with excitation at 405 nm and emission at 425−500
nm, shows binding to hSERT as well as intracellular accumulation; (B) the green channel, with excitation at 458 nm and emission at 500−600 nm,
shows only the intracellular pool with APP+ localized to mitochondria and nucleoli; (C) the red channel shows the membrane counterstain,
CellMask Deep Red, overlaid with blue and green channels. The appearance of emission at t = 24 s in the blue channel colocalizing with the
membrane stain (pink areas in C) indicates binding to SERT; this is confirmed by the complete disappearance of the membrane after addition of
escitalopram. Uptake of APP+ (green and blue channels) is slower and the intracellular pool appears after the initial binding to SERT at the
membrane (shown at t = 96 s). (D) Cross sections of the blue, green and red channels from a single cell quantitatively show APP+ binding to and
displacement from SERT.
calculations gave very similar results indicating that the starting
conformation of the piperidine ring did not bias the IFD
calculation (data not shown). Only the results from the lowest
energy starting conformation of the piperidine ring (chair axial)
will therefore be discussed here. The resulting poses of the IFD
calculation of APP+ were clustered into seven clusters (A1−A7)
and five outliers based on the Conformer cluster script of
Schrodinger suite 2012 (Table 1). Due to the very poor average
XP Gscores of clusters 3, 5, and 6 these were considered
unlikely binding modes but were evaluated by QPLD for
completeness (see Table 1). The resulting poses of the IFD
calculation of PPP+ and BPP+ were clustered into three clusters
(P1−P3) and a single cluster (B1), respectively. Only APP+ was
able to access the primary binding site of hSERT (A4, Figure
3B), and it was found to be in a manor closely resembling that
of previously published data of 1-phenyl-piperazine:23 The
charged nitrogen of APP+ interacts with the polar Asp98, the
backbone of Phe335, or the hydroxyl group of Tyr95, while the
remainder of APP+ primarily interacts with the apolar residues
Phe341, Phe335, Ile172, Tyr176, and Ala173 all of which line
the binding pocket. The three other clusters of APP+ (A1, A2,
and A7), as well as all clusters of BPP+ and PPP+, place the
299
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
ligands in the extracellular vestibule of hSERT close to the S2
site found in LeuT.24 APP+ and PPP+ were able to bind in
either a head-first (A7 and P1) or tail-first fashion (A1, A2, P2,
and P3), while BPP+ was found only to bind in a tail first
fashion (B1). Cluster A2 is positioned approximately 3 Å
further into the primary binding site than A1, but the two
clusters are otherwise highly similar. The clusters of APP+,
PPP+, and BPP+ are placed in the S2 site and are very similarly
positioned with the apolar portion of the ligand being
supported by several aromatic (Phe335, Tyr176, Tyr175,
Phe341, and Trp103) and apolar residues (Ile179, Leu99,
and Ile172), thus placing the ligands between the aromatic lid
of hSERT. The charge of all ligands is supported by interactions
with the polar residues Asp98 or Ser438 or the backbone of
Phe335 in the case of a tail first entry, but in the case of head-
first entry, the charge is left without much support in the
vicinity of Glu493. According to the average XP Gscores, all
clusters were found to be equally likely to be representative
binding modes of APP+, PPP+, and BPP+ in hSERT.
405 nm diode laser was selected as the excitation source
because it most closely matches the excitation maximum of the
dyes (ca. 425 nm) as shown in Figure 2. Longer wavelength
excitation using lines (e.g., 458, 476, 488 nm) of a tunable
argon laser is possible because the excitation spectrum extends
to approximately 490 nm; however, two issues are present with
excitation at these energies: (1) the probes absorb fewer
photons, requiring higher laser intensity, and (2) only a fraction
of the emission profile will be captured (in either band-pass or
long-pass set ups) because the probes emit between 425 and
600 nm.
We imaged the binding, transport, and displacement of APP+
collecting the emission between 425 and 500 nm in the blue
channel (blue) for 405 nm excitation and between 500 and 600
nm in a second channel (green) for 458 nm excitation (Figures
4 and 5). Following addition of APP+ to hSERT-HEK293 cells,
a rapid rise in emission intensity was observed at the cell
membrane (blue channel, Figure 4; blue trace, Figure 5)
The availability of the TICT state was estimated for APP+,
PPP+, and BPP+ when bound to hSERT in either of the two
proposed binding sites, the central S1 site or the vestibular S2
site. The torsional angle between the two aromatic units was
measured to be less than 50° in 96% of poses and less than 45°
in 91% of poses of all ligands indicating that the ligands will not
likely access the TICT state while bound to the protein. The
modest twist angles also suggest that excitation energies should
be relatively high, that is, at shorter wavelengths. While the
cationic pyridinium moiety is supported by polar amino acid
residues or the amide backbone, the aniline moiety interacts
with apolar residues. Thus, redistribution of charge in the
excited state is less favorable and the probes should emit at
relatively high energies, that is, shorter wavelengths, when
bound to hSERT.
A lowest energy representative of each cluster of APP+, PPP+,
and BPP+ was further evaluated by QPLD in order to assess the
validity of the obtained clusters from the IFD based on more
accurate partial charges of the ligands. Clusters A3, A5, and A6
all scored poorly in both the IFD and QPLD calculations and
are therefore not considered plausible binding poses and will
not be discussed further. The lowest energy pose of the rest of
the QPLD output was compared with its input pose, and they
were found to be very similar in both conformation and
position with a maximum RMSD of 1.13 Å in all clusters except
P3 where the lowest energy pose from the QPLD calculations
was postioned upside-down relatively to the input IFD pose.
The XP Gscores awarded to the best poses of the QPLD were
worsened in all clusters except P1, P3, and B1 compared with
the XP Gscores given in the IFD, but the Emodel scores were
improved in all cases suggesting that there is less conforma-
tional strain of the ligands in the poses from QLPD but also
indicating that the IFD calculation overestimated the electro-
static reward of poses due to inaccurate partial charges of the
ligands. However, the best scoring poses of the QPLD all
scored well enough to imply that the clusters are plausible
binding modes of APP+, PPP+, and BPP+. The remaining poses
of the QPLD were few in number, and most scores of second
best poses were poor indicating a rigid binding mode of the
ligands in both binding sites.
Figure 5. (A) Time-dependent emission profiles of the membrane and
cytoplasmic populations of APP+ reveal that APP+ rapidly associates
with SERT at the membrane (blue trace) and is displaced by
escitalopram, an SSRI, or 5-HT, a competitive substrate of SERT.
Transport via SERT is constant until addition of a competitive ligand,
after which the emission intensity (green trace) plateaus. (B) The
binding, displacement and transport kinetics of PPP+ are identical to
those observed for APP+. (C) BPP+ shows binding and displacement
from SERT but also some nonspecific uptake; intracellular
accumulation of BPP+ is only moderately affected by the addition of
escitalopram.
Confocal Microscopy of APP+ Binding and Transport.
We next evaluated the binding and transport of APP+ exposed
to HEK293 cells expressing hSERT.25 Based on the optical
spectra of the probes and the relatively nonpolar binding site, a
300
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
colocalizing with a membrane-specific stain, CellMask Deep
Red (Life Technologies). As time progressed, emission was also
observed within the cells (green channel, Figure 4; green trace,
Figure 5), which gradually increased in intensity over time.
After 120 s, either 5-HT, the native substrate of SERT, or
escitalopram, a selective serotonin reuptake inhibitor (SSRI),
was introduced. The addition of either competitive ligand
resulted in a rapid decrease in emission intensity at the
membrane until the emission intensity within the cell reaches a
plateau. The observation of inhibition of APP+ transport is
consistent with the previous report of APP+ and SERT
interactions;9 however, our observation of membrane-localized
emission is unique and appears to be the result of APP+ binding
to SERT rather than a nonspecific interaction with the cell
membrane. This notion is supported by the fact that no
membrane-localized fluorescence was observed for nontrans-
duced, control HEK293 cells as well as the loss of emission
following addition of a competitive substrate (5-HT) or
inhibitor (escitalopram). Furthermore, the binding induced
emission enhancement appears to be SERT-specific because it
is not observed for HEK293 cells overexpressing hDAT or
hNET.
Figure 6. Comparison of optical spectra and available excitation
sources for APP+; the emission spectra were obtained on intact cells
using a confocal microscope in xyλ mode. The excitation maximum of
APP+ in octanol most closely matches the 405 nm laser line, while 458
nm is also an option. The emission from cellular substructures show
distinct spectral profiles; membrane-localized (i.e., SERT-bound)
APP+ exhibits the shortest wavelength emission maximum.
the binding pocket. Interestingly, the nucleoli pool of APP+ can
be selectively excited using the 458 nm laser line as evident in
Figure 6. APP+ molecules in the nucleoli appear very faint
compared with either the mitochondrial or membrane
population when excited at 405 nm (blue channel), while
they are clearly evident when excited at 458 (green channel).
Conversely, the SERT-bound population of APP+ does not
appear in the green channel with 458 nm excitation and is only
visible with 405 nm excitation. This can be attributed to the
moderate twist between the anilino and pyridinium rings that is
predicted by IFD calculations: increasing the angle between the
two π systems will lead to an increased electronic bandgap and
require higher excitation energies. The distinct photophysical
behavior of SERT-bound and intracellular (i.e., transported)
APP+ may enable these two populations to be differentiated
spectroscopically in high throughput fluorescence assays
without the need for microscopy-based assays and image
analysis.
Binding and Transport of PPP+ and BPP+. All three
probes, APP+, PPP+ and BPP+, share similar binding and
displacement dynamics; however the intracellular accumulation
of BPP+ and PPP+ differs from that observed for APP+ (Figure
4). Upon addition of PPP+ or BPP+ to hSERT-HEK293 cells,
turn-on emission is observed at the membrane while
introduction of either 5-HT or escitalopram reduces the
membrane emission. Intracellular accumulation of PPP+ differs
slightly from that observed for APP+. Following introduction of
the probe, emission is observed to increase at the mitochondria,
but not nucleoli; the rate of transport is significantly attenuated
following addition of a competitive SERT ligand. For BPP+,
different transport dynamics were observed. Addition of 5-HT
or citalopram does not lead to a marked decrease in the rate of
transport suggesting that transport via SERT is not the primary
mode of intracellular accumulation for BPP+. Instead, it appears
that the lipophilic butyl tail renders BPP+ membrane
permeable, which has previously been observed for other
cationic and lipophilic dyes such as the membrane stain FM1-
43 as well as ASP+ derivatives possessing alkyl tails.17 The
differences observed in PPP+ and BPP+ transport dynamics may
be linked to the IFD predicted poses. While APP+ and PPP+
were able to bind in both head-first and tail-first poses, BPP+
was found to only bind tail-first.
Spectroscopic Resolution of SERT-Bound and Intra-
cellular APP+. Intracellular APP+ localizes to mitochondria
and nucleoli where the probes also are emissive.9,18 IFD studies
(see above) placed APP+ in a primarily hydrophobic binding
pocket supported by aromatic residues. Thus, APP+ should
experience a distinct chemical environment when bound to
SERT compared with the interactions with nucleic acids or
subcompartments of mitochondria. Because the optical proper-
ties of APP+ are highly dependent on polarity of its
microenvironment,19 we expected that emission from these
different populations would vary accordingly. Figure 6 depicts
the emission spectra of APP+ at the membrane, mitochondria,
and nucleoli. The emission maximum of SERT-bound APP+
differs by approximately 40 nm from that observed from the
mitochondria- or nucleoli-bound APP+. The emission is at
shorter wavelengths, that is, higher energies, which likely
reflects the presence of mainly nonpolar aromatic moieties in
CONCLUSION
■
Binding induced turn-on emission is an attractive feature of
fluorescent probes and ligands because this behavior minimizes
background emission from unbound probes and eliminates the
need for rinsing steps when imaging. We have demonstrated
that APP+ and analogues BPP+ and PPP+ exhibit turn-on
emission when binding to SERT. IFD studies reveal that SERT
presents a geometrically restrictive ligand binding pocket in
which these probes bind amidst mainly aromatic and nonpolar
residues. This environment influences the optical properties of
the dyes allowing for selective excitation at shorter wavelengths.
We have also demonstrated that emission of the SERT bound
population is distinct from probes bound to other cellular
components, enabling spectroscopic deconvolution of binding
and transport without the need for spatial resolution. Based on
these results, APP+ perfectly compliments the binding,
transport, and spectroscopic behavior of ASP+. Employed in
tandem, these two probes should enable differentiation of
serotonergic and adrenergic cells in tissue presenting both cell
types with applications in pharmacological profiling of selective,
dual, and triple reuptake inhibitors also possible.
301
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
SP scoring function, and the redocking was performed using Glide
XP.39 Up to 100 poses were saved within an energy window of 50
kcal/mol of the lowest scoring pose. All other settings were default
settings. Two scoring functions were used to distinguish poses in this
study, the Gscore and the E-model score. The Gscore is an empirically
determined score that considers the interaction energy between ligand
and protein as well as the entropic contribution thus approximating
the ligand binding affinity.37 The E-model score is similar to the
Gscore but also takes into account ligand strain and handles the
nonbonded interactions more explicitly.
Quantum Polarized Ligand Docking. The lowest energy pose of
each cluster from the IFD for each ligand was further evaluated by
QPLD in order to fully assess the correct binding mode of each ligand.
QPLD is a QM/MM approach that reassigns new partial charges to a
ligand using Jaguar based on its position in the binding site thereby
acknowledging the polarization experienced by the ligand from the
protein, where after the ligand is redocked into the binding site using
Glide.40 Glide version 5.8 and Jaguar version 7.9 was employed by the
Schrodinger suite.35,41 The selected poses from the IFD were used as
input directly into the QPLD workflow whereafter the partial charges
were recalculated using the 6-31G*/LACVP* basis set and the B3LYP
density functional. The redocking was performed with Glide XP, and a
maximum of 20 poses were retained. The rest of the settings applied
were default settings.
METHODS
■
Experimental Methods. APP+ was synthesized as previously
described18 and characterized by H NMR, 13C NMR, and HRMS
spectroscopies.
1
Synthesis of 1-Butyl-4-[4-(1-dimethylamino)phenyl]-pyridinium
Bromide (BPP+). 4-[4-(1-Dimethylamino)phenyl]-pyridine27 (100
mg) was combined with 200 μL of 1-bromobutane and 8.0 mL of
toluene in a 20 mL scintillation vial. The reaction mixture was heated
at 80 °C for 24 h as a yellow precipitate formed. After the reaction
mixture cooled, the crude product was isolated by suction filtration
and rinsed with EtOAc. The crude product was crystallized from
1
MeOH/EtOAc to yield 67 mg (52%) of a yellow powder. H NMR
(500 MHz, DMSO-d6): δ (ppm) 0.92 (3H, t, J = 7.5 Hz), 1.31 (2H,
quin, J = 7.5 Hz), 1.87 (2H, quin, J = 7.5 Hz), 3.07 (s, 6H), 4.48 (2H,
t, J = 7.5 Hz), 6.86 (2H, d, J = 5.0 Hz), 8.03 (2H, d, J = 5.0 Hz), 8.33
(2H, d, J = 7.2 Hz), 8.85 (2H, d, J = 7.2 Hz). 13C NMR (500 MHz,
DMSO-d6): δ (ppm) 13.8, 19.3, 40.1, 59.0, 112.6, 119.2, 121.4, 130.0,
144.1, 153.5, 154.2. IR νmax (cm−1): 715.7, 766.38, 811.9, 837.2, 941.5,
990.0, 1043.0, 1066.6, 1116.0, 1168.0, 1233.6, 1310.9, 1440.0, 1474.2,
1504.9, 1557.9, 1590.0, 2919.8, 2958.4, 3428.2. HRMS (ESI+)
C17H23N2 [M+]: calculated 255.1859, found. 255.1869. Mp: 249−
251 °C.
Synthesis of 1-Methyl-4-[4-(1-piperidinyl)phenyl]-pyridinium Io-
dide (PPP+). 4-(4-Fluorophenyl)-1-methylpyridinium iodide26 (100
mg) was combined with 0.5 mL of piperidine, 100 mg of K2CO3, and 2
mL of DMSO in a 20 mL scintillation vial. The reaction mixture was
heated at 50 °C for 12 h, developing into a bright yellow solution.
After cooling, the solution was filtered, and the solvent was removed
under vacuum. The crude product was taken up in the minimum
volume of MeOH (∼1 mL) and precipitated in EtOAc. Twenty-three
mg (17% yield) of a yellow powder was collected by suction filtration
(care must be taken because the product is hydroscopic) and dried
Clustering and Analysis of Poses. The poses from the IFDs were
clustered using the Conformer Cluster script available through the
Schrodinger Web site. This script was the latest update to the
Schrodinger 2012 program suite. The script divides poses into clusters
based on the hierarchical clustering algorithm used in Canvas,42 and
the average linking method was chosen to determine the dissimilarity
between clusters.
1
under vacuum. H NMR (500 MHz, DMSO-d6): δ (ppm) 1.60 (6H,
bs), 3.43 (bs, 4H), 4.22 (3H, s), 7.07 (2H, d, J = 8.3 Hz), 7.99 (2H, d,
J = 8.3 Hz), 8.32 (2H, d, J = 5.6 Hz), 8.75 (2H, d, J = 5.6 Hz). 13C
NMR (500 MHz, DMSO-d6): δ (ppm) 24.4, 25.4, 46.7, 48.2, 114.6,
120.5, 121.5, 130.1, 14.0, 153.8(0), 153.8(3). IR νmax (cm−1): 667.0,
745.2, 798.0, 848.4, 918.4, 988.9, 1020.3, 1044.9, 1122.7, 1186.4,
122.4, 1246.0, 1340.6, 1386.3, 1444.8, 1470.9, 1497.3, 1584.3, 1642.0,
2849.6, 2928.0, 3010.8, 3445.1. HRMS (ESI+) C17H21N2 [M+]:
calculated 255.1699, found 253.1716.
Optical Spectroscopy. Spectroscopy and HPLC grade solvents
were utilized for spectroscopic measurements; all path lengths were 1
cm. Emission and excitation spectra were obtained on a PerkinElmer
LS55 fluorometer with probe concentrations of 1 μM.
Cell Culture. Parental and HEK293 cells stably expressing SERT
(hSERT-HEK293)25 were cultured as previously described in sterile
T-75 flasks.17 Cells were maintained in DMEM (Dulbecco’s
modification of Eagle’s medium with 4.5 g/L glucose, without L-
glutamine and sodium pyruvate) containing 10% dialyzed FBS, 2 mM
glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, and 250
μg/mL Geneticin (hSERT-HEK293 cells only) at 37 °C and 5% CO2.
For imaging experiments, cells were seeded at a density of 105 cells/
cm2 in 96-well microtiter plates and incubated for at least 48 h until a
visible monolayer was established. Prior to imaging, DMEM was
removed from the cells and Leibovitz’s L-15 modified (pH 7.3) media
with 5% FBS was added to each well. Working solutions were prepared
in L-15 media from 1 mM DMSO stocks to concentrations of 100 μM
for the fluorescent probes and 10 μM for escitalopram.
Confocal Microscopy. Imaging was performed on a Leica SP5
confocal microscope housed within the UM Biology Imaging Core
Facility. All imaging experiments were performed at room temperature
(22−25 °C) under normal atmosphere. Excitation was achieved using
the 405 nm diode laser or the 458 and 488 nm lines of an argon laser
for APP+ and derivatives as noted in the text; the membrane stain,
CellMask Deep Red (CMDR, Life Technologies), was excited at 633
nm. Emission was captured on separate PMTs at wavelengths noted in
the text; CMDR emission was captured between 645 and 700 nm. For
time course experiments (in xyt mode), cells were placed on the
imaging stage and brought into focus using phase contrast or the
emission of CMDR. Solutions of fluorescent probes or inhibitors were
added and gently mixed with pipetting action to produce final
concentrations of 2 μM for probes and 2 μM for inhibitors. Images
were collected between 2 and 15 s as noted in the text for up to 10
min. For xyλ scans, images of whole cells were captured at 5 nm
increments with 10 nm band-pass. Images were analyzed using Fiji/
ImageJ software (NIH, USA). For time course images, three
independent experiments were performed and three or four cells in
a field of view were analyzed; error bars show SE.
Computational Methods. Protein Preparation. The homology
model of hSERT used in this study was based on an alignment28 of the
bacterial leucine transporter homologue22,29 (LeuT) with further
optimizations to the extracellular loop 2 of hSERT as described
previously.21,30 This model includes two sodium ions and one chloride
ion close to the binding site and has been optimized to accommodate
5-HT.21
Ligand Preparation. Ligands APP+, PPP+, and BPP+ were built in
Maestro 9.331 and were subjected to a conformational search using the
OPLS-2005 force field32 and implicit water in MacroModel 9.9.33
A
minimization step was performed after each iteration of the
conformational search using a conjugate gradient algorithm in 5000
steps or until convergence.
Induced Fit Docking. The lowest energy conformation of each
ligand was docked using the IFD protocol. This method allows for
complete flexibility of the ligand and flexibility of the protein near the
binding site, allowing for docking of all ligands into a highly movable
protein. First the ligand is docked into the protein using a softened van
der Waals (vdW) potential using Glide. This is followed by an
optimization of the side chains of the protein near the binding site by
Prime, and finally the ligand is redocked using a full vdW potential into
the optimized protein using Glide.34 This protocol employs Glide
version 5.8 and Prime version 3.1 from the Schrodinger suite.35−37
The binding site was defined by residues Gly442, Asp98, Leu431,
Thr439, and Ile172.29,38 The initial docking was performed using the
302
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
(5) Blakely, R. D., De Felice, L. J., and Hartzell, H. C. (1994)
Molecular physiology of norepinephrine and serotonin transporters. J.
Exp. Biol. 196, 263−281.
(6) McLane, M. W., McCann, U., and Ricaurte, G. (2011) Identifying
the serotonin transporter signal in Western blot studies of the
neurotoxic potential of MDMA and related drugs. Synapse 65, 1368−
1372.
(7) Giang, T., Ritze, Y., Rauchfuss, S., Ogueta, M., and Scholz, H.
(2011) The serotonin transporter expression in Drosophila mela-
nogaster. J. Neurogenet. 25, 17−26.
(8) Schenk, S., Hely, L., Lake, B., Daniela, E., Gittings, D., and Mash,
D. C. (2007) MDMA self-administration in rats: Acquisition,
progressive ratio responding and serotonin transporter binding. Eur.
J. Neurosci. 26, 3229−3236.
ASSOCIATED CONTENT
■
S
* Supporting Information
1H and 13C NMR spectra of BPP+ and PPP+, summary of IFD
and QPLD results, images of BPP+ and PPP+ binding to
hSERT-HEK293 cells, and control images of APP+, BPP+, and
PPP+ with parental HEK293 cells. This material is available free
of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
*Email: jnwilson@miami.edu.
*E-mail: birgit@chem.au.dk.
(9) Solis, E., Zdravkovic, I., Tomlinson, I. D., Noskov, S. Y.,
Rosenthal, S. J., and De Felice, L. J. (2012) 4-(4-(Dimethylamino)-
phenyl)-1-methylpyridinium (APP+) is a fluorescent substrate for the
human serotonin transporter. J. Biol. Chem. 287, 8852−8863.
(10) Oz, M., Libby, T., Kivell, B., Jaligam, V., Ramamoorthy, S., and
Shippenberg, T. S. (2010) Real-time, spatially resolved analysis of
serotonin transporter activity and regulation using the fluorescent
substrate, ASP+. J. Neurochem. 114, 1019−1029.
Author Contributions
J.N.W. and B.S. planned the project. W.M.B. and J.N.W.
synthesized the fluorescent probes and carried out the optical
spectroscopy and imaging experiments. L.K.L performed the
docking experiments, and B.S. and L.K.L analyzed the results.
The manuscript was written through contributions of all
authors.
(11) Schwartz, J. W., Blakely, R. D., and DeFelice, L. J. (2003)
Binding and transport in norepinephrine transporters. Real-time,
spatially resolved analysis in single cells using a fluorescent substrate. J.
Biol. Chem. 278, 9768−9777.
(12) Schwartz, J. W., Novarino, G., Piston, D. W., and DeFelice, L. J.
(2005) Substrate binding stoichiometry and kinetics of the
norepinephrine transporter. J. Biol. Chem. 280, 19177−19184.
(13) Gubernator, N. G., Zhang, H., Staal, R. G., Mosharov, E. V.,
Pereira, D. B., Yue, M., Balsanek, V., Vadola, P. A., Mukherjee, B.,
Edwards, R. H., Sulzer, D., and Sames, D. (2009) Fluorescent false
neurotransmitters visualize dopamine release from individual pre-
synaptic terminals. Science 324, 1441−1444.
(14) Beikmann, B. S., Tomlinson, I. D., Rosenthal, S. J., and Andrews,
A. M. (2013) Serotonin uptake is largely mediated by platelets versus
lymphocytes in peripheral blood cells. ACS Chem. Neurosci. 4, 161.
(15) Hettie, K. S., Liu, X., Gillis, K. D., and Glass, T. E. (2013)
Selective catecholamine recognition with NeuroSensor 521: A
fluorescent sensor for the visualization of norepinephrine in fixed
and live cells. ACS Chem. Neurosci. 4, 918−923.
(16) Klockow, J. L., Hettie, K. S., and Glass, T. E. (2013) ExoSensor
517: A dual-analyte fluorescent chemosensor for visualizing neuro-
transmitter exocytosis. ACS Chem. Neurosci. 4, 1334−1338.
(17) Wilson, J. N., Brown, A. S., Babinchak, W. M., Ridge, C. D., and
Walls, J. D. (2012) Fluorescent stilbazolium dyes as probes of the
norepinephrine transporter: Structural insights into substrate binding.
Org. Biomol. Chem. 10, 8710−8719.
(18) Karpowicz, R. J., Jr., Dunn, M., Sulzer, D., and Sames, D. (2013)
APP+, a fluorescent analogue of the neurotoxin MPP+, is a marker of
catecholamine neurons in brain tissue, but not a fluorescent false
neurotransmitter. ACS Chem. Neurosci. 4, 858−869.
(19) Fromherz, P., and Heilemann, A. (1992) Twisted internal
charge transfer in (aminophenyl)pyridinium. J. Phys. Chem. 96, 6864−
6866.
(20) Reichardt, C. (1979) Empirical parameters of solvent polarity as
linear free-energy relationships. Angew. Chem., Int. Ed. 18, 98.
(21) Koldsø, H., Christiansen, A. B., Sinning, S., and Schiøtt, B.
(2013) Comparative modeling of the human monoamine transporters:
Similarities in substrate binding. ACS Chem. Neurosci. 4, 295−309.
(22) Yamashita, A., Singh, S. K., Kawate, T., Jin, Y., and Gouaux, E.
(2005) Crystal structure of a bacterial homologue of Na+/Cl−-
dependent neurotransmitter transporters. Nature 437, 215−223.
(23) Severinsen, K., Kraft, J. F., Koldsø, H., Vinberg, K. A., Rothman,
R. B., Partilla, J. S., Wiborg, O., Blough, B., Schiøtt, B., and Sinning, S.
(2012) Binding of the amphetamine-like 1-phenyl-piperazine to
monoamine transporters. ACS Chem. Neurosci. 3, 693−705.
(24) Shi, L., Quick, M., Zhao, Y., Weinstein, H., and Javitch, J. A.
(2008) The mechanism of a neurotransmitter:sodium symporter–
Funding
This work was supported by the Bankhead-Coley Biomedical
Research Program, Grant 3BN08 (J.N.W.), the Danish
National Research Foundation, Grant DNRF59, and the
Lundbeck Foundation (B.S.).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
HEK293 cells expressing SERT were a generous gift from Dr.
Randy Blakely, Vanderbilt University.
ABBREVIATIONS
■
5-HT 5-hydroxytryptamine; APP+ 4-(4-(dimethylamino)-phe-
nyl)-1-methylpyridinium; ASP+ 4-(4-(dimethylamino)-styryl)-
N-methylpyridinium; BPP+ 1-butyl-4-[4-(1-dimethylamino)-
phenyl]-pyridinium bromide; CT charge transfer; DAT
dopamine transporter; IFD induced fit docking; LeuT leucine
transporter; NET norepinephrine transporter; QPLD quantum
polarized ligand docking; PPP+ 1-methyl-4-[4-(1-piperidinyl)-
phenyl]-pyridinium; SERT serotonin transporter; SSRI selec-
tive serotonin reuptake inhibitor; TICT twisted intramolecular
charge transfer; vdW van der Waals
REFERENCES
■
(1) Baganz, N. L., and Blakely, R. D. (2013) A dialogue between the
immune system and brain, spoken in the language of serotonin. ACS
Chem. Neurosci. 4, 48−63.
(2) Amireault, P., Sibon, D., and Cote, F. (2013) Life without
peripheral serotonin: Insights from tryptophan hydroxylase 1 knock-
out mice reveal the existence of paracrine/autocrine serotonergic
networks. ACS Chem. Neurosci. 4, 64−71.
(3) Arango, V., Underwood, M. D., Boldrini, M., Tamir, H., Kassir, S.
A., Hsiung, S., Chen, J. J., and Mann, J. J. (2001) Serotonin 1A
receptors, serotonin transporter binding and serotonin transporter
mRNA expression in the brainstem of depressed suicide victims.
Neuropsychopharmacology 25, 892−903.
(4) Andersen, J., Stuhr-Hansen, N., Zachariassen, L., Toubro, S.,
Hansen, S. M., Eildal, J. N., Bond, A. D., Bøgesø, K. P., Bang-Andersen,
B., Kristensen, A. S., and Strømgaard, K. (2011) Molecular
determinants for selective recognition of antidepressants in the
human serotonin and norepinephrine transporters. Proc. Natl. Acad.
Sci. U. S. A. 108, 12137−12142.
303
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304

ACS Chemical Neuroscience
Research Article
inward release of Na+ and substrate is triggered by substrate in a
second binding site. Mol. Cell 30, 667−677.
(25) Qian, Y., Galli, A., Ramamoorthy, S., Risso, S., DeFelice, L. J.,
and Blakely, R. D. (1997) Protein kinase C activation regulates human
serotonin transporters in HEK-293 cells via altered cell surface
expression. J. Neurosci. 17, 45−57.
(26) Wimalasena, D. S., Perera, R. P., Heyen, B. J., Balasooriya, I. S.,
and Wimalasena, K. (2008) Vesicular monoamine transporter
substrate/inhibitor activity of MPTP/MPP+ derivatives: A struc-
ture−activity study. J. Med. Chem. 51, 760−768.
(27) Vella, S. J., Tiburcio, J., Gauld, J. W., and Loeb, S. J. (2006)
Push−pull [2]pseudorotaxanes. Electronic control of threading by
switching ON/OFF an intramolecular charge transfer. Org. Lett. 8,
3421−3424.
(28) Beuming, T., Shi, L., Javitch, J. A., and Weinstein, H. (2006) A
comprehensive structure-based alignment of prokaryotic and eukary-
otic neurotransmitter/Na+ symporters (NSS) aids in the use of the
LeuT structure to probe NSS structure and function. Mol. Pharmacol.
70, 1630−1642.
(29) Leyla, C., Sinning, S., Severinsen, K., Hansen, C. G., Møller, M.
S., Bols, M., Wiborg, O., and Schiøtt, B. (2008) Binding of serotonin
to the human serotonin transporter. Molecular modeling and
experimental validation. J. Am. Chem. Soc. 130, 3853−3865.
(30) Koldsø, H., Severinsen, K., Tran, T. T., Celik, L., Jensen, H. H.,
Wiborg, O., Schiøtt, B., and Sinning, S. (2010) The two enantiomers
of citalopram bind to the human serotonin transporter in reversed
orientations. J. Am. Chem. Soc. 132, 1311−1322.
(31) Suite 2012: Maestro, version 9.3, Schrodinger, LLC, New York,
̈
NY, 2012.
(32) Kaminski, G. A., Friesner, R. A., Tirado-Rives, J., and Jorgensen,
W. L. (2001) Evaluation and reparametrization of the OPLS-AA force
field for proteins via comparison with accurate quantum chemical
calculations on peptides. J. Phys. Chem. B 105, 6474−6487.
(33) Schrodinger Suite 2012, MacroModel, version 9.6, Schrodinger,
̈
̈
LLC, New York, NY, 2012.
(34) Sherman, W., Day, T., Jacobson, M. P., Friesner, R. A., and
Farid, R. (2006) Novel procedure for modeling ligand/receptor
induced fit effects. J. Med. Chem. 49, 534−553.
(35) Schrodinger Suite 2012, Glide, version 5.8, Schrodinger, LLC,
̈
̈
New York, NY, 2012.
(36) Schrodinger Suite 2012, Prime, version 3.1, Schrodinger, LLC,
̈
̈
New York, NY, 2012.
(37) Friesner, R. A., Banks, J. L., Murphy, R. B., Halgren, T. A., Klicic,
J. J., Mainz, D. T., Repasky, M. P., Knoll, E. H., Shelley, M., Perry, J. K.,
Shaw, D. E., Francis, P., and Shenkin, P. S. (2004) Glide: A new
approach for rapid, accurate docking and scoring. 1. Method and
assessment of docking accuracy. J. Med. Chem. 47, 1739−1749.
(38) Sinning, S., Musgaard, M., Jensen, M., Severinsen, K., Celik, L.,
Koldsø, H., Meyer, T., Bols, M., Jensen, H. H., Schiøtt, B., and Wiborg,
O. (2010) Binding and orientation of tricyclic antidepressants within
the central substrate site of the human serotonin transporter. J. Biol.
Chem. 285, 8363−8374.
(39) Friesner, R. A., Murphy, R. B., Repasky, M. P., Frye, L. L.,
Greenwood, J. R., Halgren, T. A., Sanschagrin, P. C., and Mainz, D. T.
(2006) Extra precision glide: Docking and scoring incorporating a
model of hydrophobic enclosure for protein-ligand complexes. J. Med.
Chem. 49, 6177−6196.
(40) Cho, A. E., Guallar, V., Berne, B. J., and Friesner, R. (2005)
Importance of accurate charges in molecular docking: quantum
mechanical/molecular mechanical (QM/MM) approach. J. Comput.
Chem. 26, 915−931.
(41) Schrodinger Suite 2012, Jaguar, version 7.9, Schrodinger, LLC,
̈
̈
New York, NY, 2012.
(42) Schrodinger Suite 2012, Canvas, version 1.5, Schrodinger, LLC,
̈
̈
New York, NY, 2012.
304
dx.doi.org/10.1021/cn400230x | ACS Chem. Neurosci. 2014, 5, 296−304