Flavonoids, flavonoid metabolites, and phenolic acids inhibit oxidative stress in the neuronal cell line HT-22 monitored by ECIS and MTT assay: A comparative study

Article abstract of DOI:10.1021/np400518k

A real-time and label-free in vitro assay based on electric cell-substrate impedance sensing (ECIS) was established, validated, and compared to an end-point MTT assay within an experimental trial addressing the cytoprotective effects of 19 different flavo

Full text of DOI:10.1021/np400518k

Article
pubs.acs.org/jnp
Flavonoids, Flavonoid Metabolites, and Phenolic Acids Inhibit
Oxidative Stress in the Neuronal Cell Line HT-22 Monitored by ECIS
and MTT Assay: A Comparative Study
Beata Kling,† Daniel Bucherl,† Peter Palatzky,‡ Frank-Michael Matysik,‡ Michael Decker,†,§
̈
Joachim Wegener,‡ and Jorg Heilmann*,†
̈
†Institut fur Pharmazie, Universitat Regensburg, Universitatsstraße 31, 93053 Regensburg, Germany
‡Institut fur Analytische Chemie, Chemo- und Biosensorik, Universitat Regensburg, Universitatsstraße 31, 93053 Regensburg,
̈
̈
̈
̈
̈
̈
Germany
§Institut fur Pharmazie und Lebensmittelchemie, Julius-Maximilians-Universitat Wurzburg, Am Hubland, 97074 Wurzburg, Germany
̈
̈
̈
̈
S
* Supporting Information
ABSTRACT: A real-time and label-free in vitro assay based
on electric cell-substrate impedance sensing (ECIS) was
established, validated, and compared to an end-point MTT
assay within an experimental trial addressing the cytoprotective
effects of 19 different flavonoids, flavonoid metabolites, and
phenolic acids and their methyl esters on the HT-22 neuronal
cell line, after induction of oxidative stress with tert-butyl
hydroperoxide. Among the flavonoids under study, only those
with a catechol unit and an additional 4-keto group provided
cytoprotection. The presence of a 2,3-double bond was not a
structural prerequisite for a neuroprotective effect. In the case of the phenolics, catechol substitution was the only structural
requirement for activity. The flavonoids and other phenolics with a ferulic acid substitution or a single hydroxy group showed no
activity. Electrochemical characterization of all compounds via square-wave voltammetry provided a rather specific correlation
between cytoprotective activity and redox potential for the active flavonoids, but not for the active phenolics with a low molecular
weight. Moreover this study was used to compare label-free ECIS recordings with results of the established MTT assay. Whereas
the former provides time-resolved and thus entirely unbiased information on changes of cell morphology that are unequivocally
associated with cell death, the latter requires predefined exposure times and a strict causality between metabolic activity and cell
death. However, MTT assays are based on standard lab equipment and provide a more economic way to higher throughput.
mbued by the dramatically increasing incidence of
IAlzheimer’s disease (AD) worldwide, the search for
compounds with neuroprotective activity has been enhanced
in the last years. As AD is correlated with neuronal damage in
different brain areas and is still incurable, protection of neuronal
cells by inhibition or retardation of the destruction process is a
promising therapeutic concept.1
Limited defense strategies against oxidative stress play an
important role in neuronal damage and the generation of AD.2
In vitro assays to evaluate the neuroprotective effects of anti-
AD drug candidates are therefore often based on studying the
impairment of neuronal cell lines after exposure to different
stress factors such as radicals, glutamate, and amyloid-β
peptide.3−5 The viability of the damaged cells is frequently
quantified by well-established assays based on tetrazolium dyes
(MTT, XTT, or WST-1) or lactate dehydrogenase release
(LDH) and compared with an untreated control.5−8 The
decrease of cell viability caused by the toxic agents is reversed
by the test compounds, resulting in an increase in cell viability
after co-incubation of the noxious agent and the test
compound.
Over the last twenty years, results from several in vitro and in
vivo studies indicated that plant phenols are an important class
of defense antioxidants, showing cytoprotective effects for
different cell types.9,10 Cumulative evidence has shown that
phenols can act as radical scavengers, antioxidants, and metal-
chelating compounds.11,12 Furthermore, their cytoprotective
effects may be mediated via direct interference with signaling
cascades, the enhanced expression of enzymes with antioxidant
activity including catalase and glutathione reductase, or other
cellular defense strategies such as the GSH/GSSG ratio.13
Phenols are widespread in fruits and vegetables and include
mainly flavonoids, phenolic acids, and tannins. Whereas various
in vitro and in vivo studies document protective effects of
different flavonoids on neuronal cells,5,14,15 data for the
neuroprotective activity of procyanidins, flavonoid metabolites
such as phase-II conjugates or C-ring cleavage products, and
Special Issue: Special Issue in Honor of Otto Sticher
Received: June 27, 2013
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
phenolic acids and their esters are scarce. Furthermore,
comprehensive investigations addressing structural require-
ments of flavonoids and related phenolic compounds to
become neuroprotective are available only for individual stress
factors such as glutamate.4,5
mitochondrial enzyme inhibition,26 the inhibition of lipid
peroxidation in rat liver microsomes,27 and a series of
spectrometric cell-free antioxidant capacity assays28 such as
the DPPH and TEAC assays, using cyclic voltammetry for the
electrochemical measurements. In contrast to cyclic voltamme-
try and differential pulse voltammetry (DPV), which have also
been reported in the literature, SWV allows very short analysis
times and is often more sensitive (e.g., in the case of fast
electron transfer rates) than DPV. In addition, undesired
adsorption effects of analyte molecules at the electrode surface
can be diminished.23 Comparative SWV measurements of the
phenolic compounds studied in the present investigation were
carried out by the use of screen-printed carbon electrodes. It
has been reported that flavonoids with a low positive oxidation
potential possess a high radical-scavenging activity, i.e.,
representing higher receptiveness to electrochemical oxida-
tion.27,29 It was the aim of the present work to identify putative
correlations between the oxidation potentials obtained in SWV
experiments and the antioxidative, neuroprotective effects on
HT-22 cells for compounds 1−19.
Monitoring neuroprotection assays in real time using label-
free readout technologies offers many methodological advan-
tages over traditional end-point assays, since the former provide
continuous information over the entire length of the experi-
ment. Moreover, the independence of labels allows for cellular
or molecular follow-up experiments with the same cell
population. Among the different types of label-free approaches,
impedance analysis is the most well developed. Impedance-
based cellular assays were described initially by Giaever and
Keese16,17 and are now referred to as electric cell-substrate
impedance sensing (ECIS). The technique is based on growing
the cells of interest on a pair of gold-film electrodes (the
working and the counter electrode), which are deposited on the
bottom of a cell culture dish.16−18 The cell culture medium
serves as an electrolyte and provides the electrical connection
between the electrodes. During measurement, the impedance of
the cell-covered electrodes is recorded continuously at a
predefined set of AC frequencies. The cell bodies, surrounded
by their dielectric plasma membranes, behave like insulating
particles, forcing the current to flow through the small
electrolyte-filled channels beneath and between the cells at
most frequencies. This restriction of current flow within these
channels creates extra impedance that is dependent on the
three-dimensional shape of the cells on the electrode surface. If
the cells change their morphology, this alters the geometry of
the channels and consequently impedance readings. Thus,
ECIS reports on cell shape changes when the cells are exposed
to biological, chemical, or physical stimuli with a resolution that
is well below optical microscopy.17,19 As adherent cells tend to
swell and eventually lose the insulating properties of their
plasma membranes during necrosis, whereas they shrink and
round up during apoptosis, both forms of cell death are
sensitively monitored by ECIS recordings. Membrane rupture
during necrosis allows the current to flow rather unimpeded
through the cells, whereas apoptotic cell shrinkage provides
open spaces between cells. In either case, a severe decrease of
impedance down to values of a cell-free electrode is observed.52
In this study, ECIS was applied to monitor continuously the
neuronal cell line HT-22 upon exposure to flavonoids,
flavonoid metabolites, and phenolic acids and their methyl
esters (1−19) after induction of oxidative stress with tert-butyl
hydroperoxide (t-BOOH). The organic hydroperoxide, t-
BOOH, is able to induce oxidative stress in cells via the
generation of alkoxyl and peroxyl radicals that accelerate lipid
peroxidation.20,21 This is associated with several intracellular
stress factors such as increased ROS, decreased GSH levels, and
DNA strand breaks, as observed for C6 glial cells.22 The ECIS
assay was run for 24 h, and a traditional MTT assay was
performed for all compounds at the end of the exposure time.
Besides their biological activity, all test compounds (1−19)
were studied with respect to their electrochemical redox
behavior using square-wave voltammetry (SWV). SWV is a
pulsed voltammetric measuring technique that allows for a
highly sensitive quantification of redox-active flavonoids and the
determination of their individual oxidation potentials.23−25 So
far, several correlations have been established in the literature
for flavonoids between their redox potentials and different
biological activities, such as antioxidative processes and
RESULTS AND DISCUSSION
■
Addition of t-BOOH is a well accepted in vitro model to mimic
oxidative stress in neuronal cells.22,38,39 In recent studies,
several different neuronal cell lines, for example, PC-12 (rat
pheo-chromocytoma),38 C6 (rat glial),22 and HT-22 (mouse
hippocampal) neuronal cells,39 have been exposed to individual
concentrations of t-BOOH. Thus, the optimal t-BOOH
concentration was initially determined to induce oxidative
stress in HT-22 cells by monitoring the dose-dependent
impairment of cell viability via the MTT assay and time-
resolved ECIS recordings (Figure 1). Concordantly, both assays
showed only a partial reduction of HT-22 viability at
B
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 1. Dose-dependent neurotoxicity of t-BOOH on HT-22 cells monitored by (A) time-resolved ECIS readings and (B) MTT end-point assays
after 24 h of incubation. (A) The red arrow indicates addition of t-BOOH at different concentrations. The impedance magnitudes |Z| of three
independent ECIS experiments have been averaged and were normalized to the last value of |Z| recorded before the addition of t-BOOH. (B) Cells
were incubated with t-BOOH for 24 h on confluent cells, and the viability was quantified by a modified MTT assay. Results of cell viability are
expressed as percentage to untreated control cells. Results of three independent experiments were averaged and are presented as means SD. Data
were subjected to one-way ANOVA followed by Dunnett’s multiple comparison post-test using GraphPad Prism 5 software (level of significance,
***p < 0.001).
concentrations of 10 or 100 μM t-BOOH. Incubation with 200
and 300 μM t-BOOH led to a strong reduction of cell viability,
as observed in both the MTT assay and the ECIS readings. The
latter t-BOOH concentration of 300 μM was chosen as the
standard concentration in all protection assays conducted, since
it showed an earlier and stronger effect in the ECIS
measurements and more pronounced significant toxic effect
in the MTT assay at 300 μM compared to 200 μM (Figure 1).
The flavonol quercetin (1) is one of the best investigated
cytoprotective flavonoids. It has proven antioxidative and
radical-scavenging activity in various tetrazolium and LDH
release-based assays.5,22,40 Therefore, it was chosen as a positive
control for the cytoprotection assays used herein. In accordance
with literature studies, it showed a significant and concen-
tration-dependent neuroprotective effect at a concentration
range between 10 and 50 μM (Figure 2).
The protective effect of 1 against oxidative stress in HT-22 cells
was saturated at 40 μM, as revealed by ECIS readings (Figure
2). Some of the “protective” test compounds did indeed exhibit
cytotoxic effects at concentrations above 40 μM (data not
shown) but not at 40 μM. Except for 11, none of the phenolics
significantly influenced the cell viability of HT-22 cells at 40
μM.
Neuroprotection by the flavonols 1−3, the flavones 5 and 6,
and the flavanones 7 and 8 were evaluated at a standard
concentration of 40 μM by the ECIS method. As a result, the
presence of a 3′,4′-dihydroxy substituent (catechol substruc-
ture) in the B-ring seems to be an essential requirement for a
strong neuroprotective effect by flavonoids in HT-22 cells after
exposure to t-BOOH (Figures 3 and 4). Neither B-ring
monosubstitution with a OH-4′ group nor B-ring disubstitution
with 3′-methoxy- and 4′-hydroxy substituents provided any
neuroprotection. The inability of 3 to exert a neuroprotective
effect is noteworthy, as phenolic derivatives with a feruloyl
substructure are also often strong radical scavengers/antiox-
idants and only slightly less effective in comparison to the
corresponding derivatives with a catechol moiety.41 Con-
sequently, protection against oxidative stress cannot be
explained entirely by radical scavenging or the antioxidant
effect of flavonoids with 3′,4′-dihydroxy substitution. This is
supported by the fact that two compounds with a catechol
moiety, namely, 9 and 10, also did not show any effect in the
neuroprotection assay (Figures 3 and 4, data for 10 in Figure 3
not shown). Both compounds are also reported as strong
radical scavengers and cytoprotective phenolics.42,43 Thus,
according to the ECIS-based protection assays conducted, it is
apparent that an additional oxo group at C-4 is a further
prerequisite for the neuroprotective activity of flavonoids in t-
BOOH-treated HT-22 cells in vitro.
To exclude false negative results due to any inherent
cytotoxicity of the putative protective substances, all test
compounds were first investigated for their influence on HT-22
viability at a fixed concentration of 40 μM, which was selected
as a standard test concentration for the subsequent neuro-
protection assay (Figure 2S of the Supporting Information).
Phenolic derivatives with a low molecular weight, which are
derived biosynthetically from the shikimate pathway, are
widespread in the plant kingdom. Several studies have reported
on the biological effects of caffeic acid (13) and ferulic acid
(15) and their ester derivatives.44,45 In the neuroprotection
assay with t-BOOH, neither 13 nor 15 showed any in vitro
effect at 40 μM (Figure 4, data for 13 and 15 in Figure 3 not
shown). In contrast, the corresponding methyl ester of caffeic
acid (14) proved to be as active as the flavonoids 1, 6, and 8,
Figure 2. Dose-dependent neuroprotective effect of quercetin (1)
against t-BOOH-induced oxidative stress on HT-22 cells using time-
resolved ECIS readings. The black arrow indicates the addition of 1 at
different concentrations for a preincubation period of 3 h. The red
arrow indicates addition of t-BOOH at a final concentration 300 μM.
The impedance magnitude |Z| at any time of the measurement was
normalized to the last value of |Z| recorded before the addition of t-
BOOH.
C
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 3. Neuroprotection of different flavonoids, flavonoid metabolites, or related phenolics against t-BOOH-induced oxidative stress on HT-22
cells using time-resolved ECIS readings. The black arrow indicates the addition of each test compound at the final concentration 40 μM for a
preincubation period of 3 h. The red arrow indicates addition of t-BOOH at the final concentration 300 μM. Results are presented as means SD of
three independent experiments. The impedance magnitude |Z| at any time of the measurement was normalized to the last value of |Z| recorded
before the addition of t-BOOH. In each graph, curves of two test compounds were compared to an untreated or t-BOOH-treated control.
whereas the methyl ester of ferulic acid (16) was also inactive,
in accordance with the negative result obtained for 3. The
dissociation of 13 and 15 to the corresponding carboxylates
presumably prevents the free acids from permeating through
the cell membrane. Thus, the observed inactivity of these
molecules is a further hint that the neuroprotective effect
requires an intracellular mechanism and is not the result of
direct scavenging reactions on the cell surface with the alkoxyl
and peroxyl radicals generated by the decay of t-BOOH. This
assumption was supported by the results obtained for 17, an
important C-ring cleavage product of the in vivo metabolism of
1, its glycosides, and several other flavonols and flavones.46 The
free acid is reported to be a strong radical scavenger in PMNs
stimulated with formyl-methionyl-leucyl-phenylalanine
(FMLP) or opsonized zymosan,47 but showed no neuro-
protective effect. As expected, esterification to 18 restored the
neuroprotective activity (Figures 3 and 4). Furthermore, 12,
which is also a common C-ring cleavage product of flavonoids
in vivo,46 and 11, a metabolic product of salicortin,48 were
active (Figures 3 and 4).
A very interesting investigation on the cytoprotective effect
of flavonoids toward neurons was performed on HT-22 cells
after exposure to glutamate,5 which is a method of activating
stress cascades in neurons. As these cells lack ionotropic
glutamate receptors, addition of extracellular glutamate in high
concentrations inhibits cystine (as oxidized form of cysteine)
transport via the cystine/glutamate antiporter.49 This neuronal
oxidative stress induced toxicity (referred to as oxytosis) leads
to intracellular cysteine and therefore glutathione depletion,
D
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
be essential when studying self-toxic effects of compounds since
cytotoxic effects were found to be different according to the cell
density (the glutamate assay was performed with unconfluent
cells,4,34,37 while the modified MTT assays under ECIS
conditions were performed on confluent cells for direct
comparability with ECIS experiments). The glutamate assay
reproduced the neuroprotective effect of 2 at 5, 10, and 25 μM
in concordance with Ishige et al.5 (Figure 5B), revealing that 2
targets a special molecular pathway of the glutamate stress
response, but not the pathways downstream of t-BOOH.
However, 2 by itself significantly reduced the cells’ viability at
25 and 50 μM (Figure 5A), which may explain the decrease of
its protective effect from 10 to 25 μM in the glutamate assay.
Ishige et al. considered an unsaturated C-ring as a structural
requirement for the neuroprotective activity of flavonoids in the
glutamate toxicity model.5 Nevertheless, they did not include in
their investigation the flavanone 8, which was found to be
protective in the t-BOOH assay (Figures 3 and 4). Compound
8 was tested in the glutamate toxicity model (Figure 5B) and
showed a significant protective activity as quantified by MTT
and ECIS, indicating that an unsaturated flavonoid C-ring is not
a structural prerequisite for neuroprotective activity. However,
similar to 2, compound 8 showed cytotoxicity at increasing
concentrations on unconfluent cells under the glutamate assay
conditions (Figure 5A), demonstrating a concentration-
dependent loss of its protective effects at increasing
concentrations (Figure 5B).
Besides C-ring cleavage, deglycosylation, absorption, and
phase-II metabolism of intact flavonoids are key steps in the in
vivo flavonoid metabolism,51 leading to flavonoid glucuronides
and sulfates. The quercetin phase-II derivative quercetin 3-O-β-
glucuronide (4) showed no activity in the neuroprotection
assay, despite its catechol moiety, either with MTT or with the
ECIS readout. To evaluate whether dissociation of the
glucuronic acid unit is the reason for a reduced permeation
through the plasma membrane, the effect of the methyl ester of
4 was also tested (data not shown). The methyl ester of 4 did
not show any neuroprotective activity, as the molecule is
probably unable to penetrate the membrane, too.
Figure 4. Depicted are the results of the MTT end-point assay. A
colorimetric MTT test was additionally performed in the ECIS array
directly after the ECIS measurement (Figure 3) was terminated
(which was 20 h after the addition of t-BOOH). Results of cell viability
are expressed as percentages to untreated control cells. The results of
three independent experiments have been averaged and are presented
as means SD. Data were subjected to one-way ANOVA followed by
Dunnett’s multiple comparison post-test using GraphPad Prism 5
software (level of significance, ***p < 0.001).
which induces intracellular ROS accumulation and cell
injuries.50
Ishige et al. used a set of test compounds based on flavonoids
with a lower number of hydroxy groups in comparison to the
present investigation, but flavonoids 1, 2, 5−7, and 9 as well as
13 and 14 were also tested.5 The conformity of the present
results was very high, as 1, 6, and 14 were active, whereas 5, 7,
9, and 13 were not. Molecular investigations have shown the
influence of different flavonoids on diverse molecular pathways.
Nevertheless, one of the most interesting points is the diverging
neuroprotective behavior of 2 in different types of stress-related
experiments. Whereas 2 is not active in the t-BOOH work
(Figures 3 and 4), the glutamate-based assay revealed
significant activity.5 Therefore, the glutamate assay was used
according to standard procedures4,34,37 to study cytoprotective
potential toward HT-22 neurons against glutamate-induced
oxidative stress, but also potential self-toxic effects of 2 were
studied at exactly the same conditions and the same seeding
density applied for the cytoprotection assay. This was found to
All results obtained from experiments that were monitored
by continuous ECIS readings were confirmed by an MTT test
Figure 5. Dose-dependent self-toxicity (A) and dose-dependent neuroprotection against glutamate-induced oxidative stress (B) of kaempferol (2)
and eriodictyol (8) on HT-22 cells using a modified MTT test on unconfluent cells according to standard protocols as described in the Experimental
Section. Briefly, HT-22 cells were incubated for 24 h either in the absence (A) or presence (B) of glutamate with compounds 2 and 8 at the
indicated concentrations. Results of cell viability were expressed as percentages to untreated control cells and presented as means SD of three
independent experiments each performed in quadruplicate. Data were subjected to one-way ANOVA followed by Dunnett’s multiple comparison
post-test using GraphPad Prism 5 software (level of significance, **p < 0.01, ***p < 0.001).
E
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
performed directly (after ECIS data acquisition was termi-
nated) in the wells of the electrode array (Figure 4), showing
that the impedance measurements match the results of the
tetrazolium-based method. An additional comparison study
between the time-resolved ECIS measurement curves (Figure
3) and the MTT test results (Figure 4) can be found in Figure
3S of the Supporting Information. Briefly, viability of HT-22
cells was studied for compounds 1−19 under the exact
experimental ECIS conditions, but with MTT reaction 4.5 h
(Figure 3S A) and 9 h (Figure 3S B) after exposure to t-BOOH.
After 4.5 h only three (namely 1, 12, and 18) of the seven
active compounds of Figures 3 and 4 were able to reduce t-
BOOH-induced toxicity in this MTT setup, and also the
toxicity of t-BOOH is relatively low, at 70% (Figure 3S A).
After 9 h the results match the data of Figure 4, indicating that
at least 9 h is required for t-BOOH incubation to detect
neuroprotective properties sufficiently with MTT quantification
(Figure 3S B of the Supporting Information).
Summing up the results from a structural point of view,
flavonoids showed a neuroprotective activity when t-BOOH is
used, provided they possess a catechol substructure and a 4-
keto group (1, 6, and 8), whereas a 2,3-double bond is not
necessary. In the case of low molecular weight phenolics, the
only prerequisite for neuroprotective activity is a catechol
substitution (11, 12, 14, and 18). The flavonoids and other
phenolics with a ferulic acid substitution or with only one
hydroxy group are not protective against t-BOOH. This is a
striking difference in comparison with a glutamate model in
which kaempferol (2) is active according to the literature and to
our results. All phenolic compounds with a free and dissociable
carboxylic acid group did not obey the general trend
independent from the substitution pattern, as they showed in
general no activity, maybe due to insufficient permeation across
the plasma membrane into the cells.
Table 1. Peak Potentials Determined by Square-Wave
Voltammetry
a
peak 1
(mV)
peak 2
(mV)
peak 3
(mV)
peak 4
(mV)
compound
quercetin (1)
+40 (p) +189
(w)
+361
(w)
+919
(m)
kaempferol (2)
− 33
+611
(w)
(m)
isorhamnetin (3)
quercetin 3-O-glucuronide (4)
apigenin (5)
−109
+391
(w)
(p)
+105
(p)
+804
(m)
+298
(w)
+540
(m)
luteolin (6)
+64
(vp)
+835
(m)
naringenin (7)
+391
(w)
+677
(m)
eriodictyol (8)
+82 (p) +851
(p)
catechin (9)
+205
(m)
+506
(m)
procyanidin B1 (10)
catechol (11)
+242
(m)
+490
(m)
+213
(m)
4-methylcatechol (12)
caffeic acid (13)
+163
(m)
+161
(m)
caffeic acid methyl ester (14)
ferulic acid (15)
+94
(vp)
+205
(m)
+702
(w)
ferulic acid methyl ester (16)
+141
(p)
3,4-dihydroxyphenylacetic acid
(17)
+276
(m)
3,4-dihydroxyphenylacetic acid
methyl ester (18)
+197
(m)
+591
(w)
The oxidation propensity of flavonoids has been considered
an important parameter describing their antioxidative potential;
therefore, it has often been used in studies that evaluate the
antioxidant activity.26−29 Thus, SWV was used to measure the
oxidation potential of all compounds under study (1−19) to
identify putative relationships between oxidation potential and
an ability to serve as an antioxidant in neuroprotection assays.
SWV abbreviates a pulsed voltammetric measurement
technique that enables a highly sensitive quantification of
redox-active flavonoids and the determination of their
corresponding oxidation potential.24 In terms of electrode
setup and square-wave voltammetric experimental conditions,
there is no uniform protocol to study the compounds of
interest. Only a few flavonoids were characterized in this
manner in previous reports.23−25 In the present study, all
substances were investigated by applying the same electro-
chemical protocol to enable a comparative evaluation. Addi-
tionally, some phenolic acid derivatives and metabolites have
been investigated in this study that have not been evaluated
with square-wave voltammetry previously. A potential range of
0.2−1.5 V was chosen to explore the putative radical-
scavenging activities of all compounds under physiological pH
conditions. The peaks of interest are the first peaks of the
measurements occurring at negative or low positive potentials
that indicated the susceptibility of a molecule to be oxidized by
free radicals. Compounds showing a similar electrochemical
behavior were grouped with respect to their pharmacological
activity (Table 1).
3-hydroxyphenylacetic acid (19) +413
(m)
a
The small letters in brackets specify the peak height in the square-
wave voltammograms at the indicated peak potentials as very
pronounced (vp), pronounced (p), moderate (m), or weak (w).
Peak heights of very pronounced peaks are around 300 μA, of
pronounced peaks in the range 150−200 μA, of moderate peaks in the
range 50−130 μA, and of weak peaks below 50 μA.
The first observation that could be made is that all active
flavonoids with a catechol moiety and a 4-oxo-group produce
very pronounced and sharp peaks (with a peak width of
approximately 200 mV) at low positive potentials: 40 mV for 1,
64 mV for 6, and 82 mV for 8. This was true also for caffeic acid
methyl ester (14), which gave rise to a similar peak at 94 mV.
The molecule has a catechol moiety connected to an oxo group
as a part of an ester functionality. All of these three flavonoids
but not 14 exhibited a second oxidation reaction in a very
narrow range between 835 (6) and 919 mV (1). Quercetin 3-
O-β-glucuronide (4) showed pronounced electrochemical
activity at 105 mV, and another oxidation reaction occurred
at 804 mV. Two very weak oxidation peaks observed for 1 at
190 and 360 mV were not present for 4. Thus, no substantial
impact due to the glucuronide substituent could be observed
with respect to the redox properties and antioxidant activity of
1. Thus, its low activity in the cell-based protection assays used
is most likely due to reduced membrane permeability.
F
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
murine hippocampal tissue.33 HT-22 cells were grown in high-glucose
Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen, Karlsruhe,
Germany) supplemented with 10% (v/v) heat-inactivated fetal calf
serum (FCS, Biochrom, Berlin, Germany).34 Cells were kept under
standard cell culture conditions (37 °C, 5% CO2) in a humidified
incubator. Cells were subcultured three times a week.
Isorhamnetin (3) and kaempferol (2) showed moderate
oxidation peaks at very low potential ranges (−109 and −33
mV), but exhibited no activity in the t-BOOH-based neuro-
protection assay. Moreover, both molecules were characterized
by a second oxidation reaction, which appeared as a weak peak
for 3 at 391 mV and for 2 at 611 mV, respectively.
Impedimetric Monitoring Using ECIS. Time-resolved impedi-
metric monitoring of HT-22 cells was performed using the ECIS
1600R instrument purchased from Applied BioPhysics Inc. (Troy, NY,
USA). Disposable electrode arrays of type 8w1e (Applied BioPhysics)
were used in all experiments. They consisted of eight-well cell culture
dishes with gold electrodes deposited on the bottom of the wells. Each
well contained a small working electrode (area 5 × 10−4 cm2) and a
larger counter electrode (area 0.15 cm2). Due to this difference in
surface area of the electrodes, the total impedance of the system was
dominated by the impedance of the small working electrode.19 Thus,
the ECIS readout mirrored the averaged response of approximately
100 cells fitting on the surface of the small electrode. Data were
recorded at 23 selected frequencies between 25 and 105 Hz (MFT
mode), but only the impedance at 32 kHz was used for analysis.
Previous studies have shown that impedance readings at 32 kHz are
the most robust indicator for cytotoxicity.35 The complete ECIS device
is depicted in Figure 1S (Supporting Information).
Prior to cell seeding, the array was precoated with a layer of cross-
linked gelatin to provide better attachment conditions to the HT-22
cells, as these cells easily detach on regular cell culture surfaces when
grown to confluence. Therefore, the wells were incubated with 200 μL
of a 0.5% (w/v) aqueous gelatin solution for 1 h at room temperature.
After aspiration, 200 μL of 2% (v/v) glutaraldehyde solution was
added and the incubation was continued for another 15 min. The wells
were then thoroughly washed 10 times with sterile Millipore water to
remove any cytotoxic glutaraldehyde. Cells were then seeded in a
density of 6 × 105 cells per well and grown to confluency over 2−3
days.
Catechol (11) and 4-methylcatechol (12) were also active in
the neuroprotection assay but showed different electrochemical
characteristics in the SWV experiments conducted in
comparison to 1, 6, and 8. They showed moderate oxidation
peaks at relatively low potentials of 163 (12) and 213 mV (11).
No further oxidation reactions were detected for this pair of
compounds. A similar characteristic was observed for the
inactive compounds caffeic acid (13) and 3,4-dihydroxyphenyl-
acetic acid (17). Esterification led to a shift of ca. −65 to −80
mV from the free phenolic acids to the ester as observed for the
pairs 13/14, 15/16, and 17/18.
The electrochemical profile of the active compound 18
showed no characteristic pattern. This compound behaved in a
similar manner to the inactive 9 and 10 and belongs to a group
for which two oxidation reactions were detected. The first was
observed as a moderate oxidation reaction at 200 and 240 mV
for 9 and 10 and at 197 mV for 18. The second oxidation
reaction was observed at around 591 mV. The peak for 9 and
10 was again moderate, but for 18 less intensive than its first
peak.
It can be concluded from these physicochemical results that
biological activity is not associated directly with one specific
electrochemical profile. Nevertheless, for one group of
flavonoids (1, 6, and 8) with a characteristic electrochemical
pattern, neuroprotective activity was indicated. It would be of
interest to evaluate whether structurally different compounds
with the same electrochemical profile as observed for these
flavonoids are also active in in vitro neuroprotective assays.
The eight-well electrode array was placed in a humidified cell
culture incubator at 37 °C with 5% CO2 and connected to the
electronic devices located outside the incubator. The commercial ECIS
software was used for data acquisition, storage, and analysis. Prior to
treatment, cells were allowed to equilibrate for 1 h to provide stable
baseline data. Subsequently, either medium or medium containing the
test compounds was added to the wells and allowed to preincubate for
3 h. Then t-BOOH was added to the wells in 300 μM concentration,
and impedance data were recorded for a further 20 h. The impedance
magnitude |Z| at any time of the measurement was normalized to the
last value of |Z| recorded before the addition of t-BOOH.
EXPERIMENTAL SECTION
Chemicals and Reagents. Quercetin 3-O-β-glucuronide (4) and
procyanidin B1 (10) were kind gifts from Prof. Dr. I. Merfort
■
(Universitat Freiburg, Germany) and Prof. Dr. A. Nahrstedt
̈
(Universitat Munster, Germany), respectively. Kaempferol (2),
̈
̈
luteolin (6), and racemic eriodictyol (8) were purchased from
Extrasynthese (Genay Cedex, France, purity each with ≥99.9%
determined by HPLC). Isorhamnetin (3, purity 99.9%, determined
by HPLC), apigenin (5, purity 99.9%, determined by HPLC), racemic
naringenin (7, purity ≥95%, determined by HPLC), caffeic acid (13,
purity ≥95%, determined by HPLC), and ferulic acid (15, purity
≥95%, determined by HPLC) were obtained from Roth (Karlsruhe,
Germany). 3-Hydroxyphenylacetic acid (19), monosodium-L-gluta-
mate, and glutaraldehyde were purchased from Merck (Darmstadt,
Germany; purity of each ≥99% determined by HPLC). Caffeic acid
methyl ester (14, 99.9% HPLC) and ferulic acid methyl ester (16,
purity 99%, determined by GC) were purchased from Santa Cruz
Biotechnology (Heidelberg, Germany). 3,4-Dihydroxyphenylacetic
acid (17, purity ≥98% determined by HPLC) was obtained from
Lancaster Synthesis (Frankfurt am Main, Germany). Quercetin (1,
purity ≥98% determined by HPLC), catechol (11, purity ≥99%,
determined by HPLC), 4-methylcatechol (12, purity ≥95%,
determined by HPLC), and ( )-catechin (9, purity ≥96%, determined
by HPLC) as well as all other chemicals used were purchased from
Sigma (Steinheim, Germany). 3,4-Dihydroxyphenylacetic acid methyl
ester (18) was synthesized by dropwise addition of SOCl2 to 17 in
methanol at 0 °C (purity ≥98%, determined by HPLC).30 NMR
spectroscopic and mass spectrometric data were in agreement with
those reported in the literature for this compound.30
All compounds tested were dissolved in DMSO and diluted with
fresh medium, with the DMSO concentration always below 0.1% (v/
v). The aqueous t-BOOH solution was directly diluted with fresh
medium.
MTT Cell Viability Assay. This modified 3-(4,5-dimethylthiazol-2-
yl)-2,5-diphenyl tetrazolium bromide (MTT, Sigma, Steinheim,
Germany) assay36 was applied to study t-BOOH-induced dose-
dependent neurotoxic effects. Furthermore, the phenolic compounds
were prior tested with this MTT assay for their cytotoxicity against
HT-22 cells to exclude any cytotoxic response to the test compounds
during the ECIS measurements. For comparability to ECIS measure-
ments, the MTT assay was performed on a confluent layer of HT-22
cells. Therefore, the seeding density was atypically high when
compared to standard protocols, and the MTT reagent incubation
time was shortened. Briefly, a 96-well plate was precoated with 50 μL
of a 0.5% (m/v) aqueous gelatin solution for 1 h at room temperature.
After removal, cells were seeded in 96-well plates at a density of 5 ×
104 cells per well and cultured to confluence over 24 h. Subsequently,
cells were incubated for another 24 h with either medium or the test
compounds. The MTT solution (4 mg/mL in PBS) was diluted 1:10
with medium, and the mixture was added to the wells after removal of
the previous medium. After 1 h of incubation, the supernatant was
removed and 100 μL of lysis buffer (10% SDS, pH 4.1) was added to
the wells. Absorbance was determined at 560 nm on the next day with
a multiwell plate photometer (Spectra Fluor Plus, Crailsheim,
Cell Culture. All experiments in this study were performed using
the HT-22 neuronal cell line,31,32 which was originally derived from
G
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Germany). The results of cell viability studies were expressed as a
percentage to untreated control cells.
using GraphPad Prism 5 software (levels of significance *p < 0.05; **p
< 0.01; ***p < 0.001).
Dose-Dependent Neurotoxicity of t-BOOH. To identify a
suitable concentration of t-BOOH capable of inducing a pronounced
cytotoxic response of HT-22 cells, 10, 100, 200, and 300 μM t-BOOH
were evaluated in both the ECIS- and MTT-based assays with the
conditions described above.
Dose-Dependent Neuroprotective Effect of Quercetin (1) in
ECIS. To establish a suitable positive control, the dose-dependent
effect of quercetin was evaluated at concentrations of 10, 20, 30, 40,
and 50 μM in the ECIS assay as described above.
ASSOCIATED CONTENT
* Supporting Information
A schematic overview of the experimental setup to perform
ECIS measurements with 8W1E arrays and further MTT assay
results in addition and for comparison to the ECIS experiments
are available free of charge via the Internet at http://pubs.acs.
org.
■
S
Evaluation of Neuroprotection from ECIS Readings. All test
compounds were evaluated at a 40 μM final concentration in the ECIS
assay as described above.
AUTHOR INFORMATION
Corresponding Author
*Tel: +49 941 943 4759. Fax: +49 941 943 4990. E-mail: joerg.
heilmann@chemie.uni-regensburg.de.
Notes
■
End-Point MTT Assay after ECIS Readings. Directly after ECIS
data acquisition was terminated at the end of the t-BOOH exposure
time, the viability of cells in the wells of the electrode array was
determined additionally using an MTT assay.36 The MTT solution (4
mg/mL in PBS) was diluted 1:10 with medium, and the mixture was
added to the wells after removal of previous medium. After 1 h of
incubation, the supernatant was removed and 400 μL of lysis buffer
(10% SDS, pH 4.1) was added to the wells. Absorbance was
determined at 560 nm on the next day using a multiwell plate
photometer (Spectra Fluor Plus) after transferring the test solutions
into a fresh 96-well plate. The results of cell viability were expressed as
percentage to untreated control cells.
Neuroprotection against Glutamate-Induced Toxicity. This
test was performed as reported before.34,37 Briefly, HT-22 cells were
seeded in 96-well plates at a density of 5 × 103 per well and cultured
for 24 h. Subsequently, cells were incubated for another 24 h with
either medium or test compounds with potential cytoprotective
activity in either the absence (to test for toxic effects) or the presence
(to test for the protective potential against glutamate-induced
oxidative stress) of 5 mM glutamate. MTT solution (4 mg/mL in
PBS) was diluted 1:10 with medium, and the mixture was added to the
wells after removal of the previous medium. The plates were then
incubated for another 3 h. Then, the supernatant was removed, and
100 μL of lysis buffer (10% SDS, pH 4.1) was added to the wells.
Absorbance at 560 nm was determined on the next day using a
multiwell plate photometer (Spectra Fluor Plus). The results of these
cell viability assays were expressed as percentages relative to untreated
control cells. All compounds were dissolved in DMSO and diluted
with fresh medium, with the DMSO concentration always below 0.1%
(v/v).
Square-Wave Voltammetry. All voltammetric measurements
were performed using a μAutolab III potentiostat (Metrohm,
Switzerland), in combination with GPES software (version 4.9). The
screen-printed electrodes used for SWV were obtained from Dropsens
(Llanera, Spain) and consisted of a 4 mm diameter carbon-based
working electrode. The working electrode was surrounded by an
auxiliary electrode from the same material, and all potentials were
determined relative to a silver pseudoreference electrode. A 50 μL
aliquot of sample solution was sufficient to cover the three-electrode
arrangement. Square-wave voltammetry was carried out applying a
pulse amplitude of 50 mV, a step potential of 1.95 mV, and a
frequency of 180 Hz, according to Adam et al.24 Scanning the potential
window from −0.2 to 1.5 V was performed at a scan rate of 350 mV/s.
The phenolic test compounds were dissolved in DMSO at a
concentration of 50 mM. The stock solution was further diluted to
a concentration of 1 mM in a mixture of 75 mM phosphate buffer (pH
7.4)/DMSO (90:10, v/v) to prevent the phenolics from precipitating.
Additionally, higher background conductivity for SWV experiments
was achieved by using an aqueous buffer system. For data evaluation a
SWV scan from the blank buffer/DMSO mixture was recorded and
subtracted from SWV scans recorded in the presence of phenolic
compounds.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Prof. Dr. I. Merfort (Universitat Freiburg, Germany) and Prof.
̈
Dr. A. Nahrstedt (Universitat Munster, Germany) are gratefully
̈
̈
acknowledged for providing quercetin 3-O-β-glucuronide and
procyanidin B1, respectively. Special thanks are due to Dr. B.
Kraus (Universitat Regensburg, Germany) for introducing B.K.
̈
to cell culture work and for fruitful discussions. Appreciation is
expressed to Dr. X. Chen (Universitat Wurzburg, Germany) for
̈
̈
the synthesis of 3,4-dihydroxyphenylacetic acid methyl ester.
Sincere gratitude is extended to Professor S. Elz (Universitat
Regensburg, Germany) for financial support.
̈
DEDICATION
■
Dedicated to Prof. Dr. Otto Sticher of ETH-Zurich, Zurich,
Switzerland, for his pioneering work in pharmacognosy and
phytochemistry.
REFERENCES
■
(1) Choi, D. Y.; Lee, Y. J.; Hong, J. T.; Lee, H. J. Brain Res. Bull.
2012, 87, 144−153.
(2) Su, B.; Wang, X.; Nunomura, A.; Moreira, P. I.; Lee, H.-G.; Perry,
G.; Smith, M. A.; Zhu, X. Curr. Alzheimer Res. 2008, 5, 525−532.
(3) Behl, C. J. Neural Transm. 2000, 107, 393−407.
(4) Gursoy, E.; Cardounel, A.; Kalimi, M. Neurochem. Res. 2001, 26,
15−21.
(5) Ishige, K.; Schubert, D.; Sagara, Y. Free Radical Biol. Med. 2001,
30, 433−446.
(6) Weir, L.; Robertson, D.; Leigh, I. M.; Panteleyev, A. A. Anal.
Biochem. 2011, 414, 31−37.
(7) Choi, Y. W.; Takamatsu, S.; Khan, S. I.; Srinivas, P. V.; Ferreira,
D.; Zhao, J.; Khan, I. A. J. Nat. Prod. 2006, 69, 356−359.
(8) Schmidt, K.; Riese, U.; Li, Z.; Hamburger, M. J. Nat. Prod. 2003,
66, 378−383.
(9) Duthie, G.; Crozier, A. Curr. Opin. Lipidol. 2000, 11, 43−47.
(10) Bisht, K.; Wagner, K.-H.; Bulmer, A. C. Toxicology 2010, 278,
88−100.
(11) Pietta, P. G. J. Nat. Prod. 2000, 63, 1035−1042.
(12) Perron, N. R.; Brumaghim, J. L. Cell. Biochem. Biophys. 2009, 53,
75−100.
(13) Valsecchi, A. E.; Franchi, S.; Panerai, A. E.; Rossi, A.; Sacerdote,
P.; Colleoni, M. Eur. J. Pharmacol. 2011, 650, 694−702.
(14) Youdim, K. A.; Spencer, J. P. E.; Schroeter, H.; Rice-Evans, C.
Biol. Chem. 2002, 383, 503−519.
(15) Singh, M.; Arseneault, M.; Sanderson, T.; Murthy, V.;
Ramassamy, C. J. Agric. Food Chem. 2008, 56, 4855−4873.
(16) Giaever, I.; Keese, C. R. Proc. Natl. Acad. Sci. 1984, 81, 3761−
3764.
Statistics. Data of all MTT tests were expressed as means SD of
at least three independent experiments. Data were subjected to one-
way ANOVA followed by Dunnett’s multiple comparison post-test
H
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(17) Giaever, I.; Keese, C. R. Nature 1993, 633, 591−592.
(18) Wegener, J.; Keese, C. R.; Giaever, I. Exp. Cell Res. 2000, 259,
158−166.
(19) Giaever, I.; Keese, C. R. Proc. Natl. Acad. Sci. 1991, 88, 7896−
7900.
(20) Van der Zee, J.; Van Steveninck, J.; Koster, J. F.; Dubbelman, T.
M. Biochim. Biophys. Acta 1989, 980, 175−180.
(21) Palozza, P.; Luberto, C.; Ricci, P.; Sgarlata, E.; Calviello, G.;
Bartoli, G. M. Arch. Biochem. Biophys. 1996, 325, 145−151.
(22) Gitika, B.; Ram, M. S.; Sharma, S. K.; Ilavazhagan, G.; Banerjee,
P. K. Free Radical Res. 2006, 40, 95−102.
(23) Oliveira-Brett, A. M.; Ghica, M.-E. Electroanalysis 2003, 15,
1745−1750.
(24) Adam, V.; Mikelova, R.; Hubalek, J.; Hanustiak, P.; Beklova, M.;
Hodek, P.; Horna, A.; Trnkova, L.; Stiborova, M.; Zeman, L.; Kizek, R.
Sensors 2007, 7, 2402−2418.
(25) Maoela, M. S.; Arotiba, O. A.; Baker, P. G. L.; Mabusela, W. T.;
Jahed, N.; Songa, E. A.; Iwuoha, E. I. Int. J. Electrochem. Sci. 2009, 4,
1497−1510.
́
(26) Hodnick, W. F.; Milosavljevic, E. B.; Nelson, J. H.; Pardini, R. S.
Biochem. Pharmacol. 1988, 37, 2607−2611.
(27) Van Acker, S. A. B. E.; van den Bern, D.-J.; Tromp, M. N. J. L.;
Griffioen, D. H.; van Bennekom, W. P.; van der Vijgh, W. J. F.; Bast, A.
Free Radical Biol. Med. 1996, 20, 331−342.
(28) Zhang, D.; Chu, L.; Liu, Y.; Wang, A.; Ji, B.; Wu, W.; Zhou, F.;
Wei, Y.; Cheng, Q.; Cai, S.; Xie, L.; Jia, G. J. Agric. Food Chem. 2011,
59, 10277−10285.
(29) Yang, B.; Kotani, A.; Arai, K.; Kusu, F. Anal. Sci. 2001, 17, 599−
604.
(30) Zhang, Z.; Xiao, B.; Chen, Q.; Lian, X.-Y. J. Nat. Prod. 2010, 73,
252−254.
(31) Davies, J. B.; Maher, P. Brain Res. 1994, 652, 169−173.
(32) Maher, P.; Davies, J. B. J. Neurosci. 1996, 16, 6394−6401.
(33) Morimoto, B. H.; Koshland, D. E. Neuron 1990, 5, 875−880.
(34) Darras, F. H.; Kling, B.; Heilmann, J.; Decker, M. ACS Med.
Chem. Lett. 2012, 3, 914−919.
(35) Hofmann, U.; Michaelis, S.; Winckler, T.; Wegener, J.; Feller,
K.-H. Biosens. Bioelectron. 2013, 39, 156−162.
(36) Mosmann, T. J. Immunol. Methods 1983, 65, 55−63.
(37) Chen, X.; Zenger, K.; Lupp, A.; Kling, B.; Heilmann, J.; Fleck,
C.; Kraus, B.; Decker, M. J. Med. Chem. 2012, 55, 5231−5342.
(38) Palomba, L.; Guidarelli, A.; Scovassi, A. I.; Cantoni, O. Eur. J.
Biochem. 2001, 268, 5223−5228.
(39) Duncan, R. S.; Chapman, K. D.; Koulen, P. Mol. Neurodegener.
2009, 4, 50.
(40) Riesenhuber, A.; Kasper, D. C.; Vargha, R.; Endemann, M.;
Aufricht, C. Pediatr. Nephrol. 2007, 22, 1205−1208.
(41) Ishimoto, H.; Tai, A.; Yoshimura, M.; Amakura, Y.; Yoshida, T.;
Hatano, T.; Ito, H. Biosci. Biotechnol. Biochem. 2012, 76, 395−399.
(42) Saint-Cricq de Gaulejac, N.; Provost, C.; Vivas, N. J. Agric. Food
Chem. 1999, 47, 425−431.
(43) Shi, J.; Yu, J.; Pohorly, J. E.; Kakuda, Y. J. Med. Food 2003, 6,
291−299.
(44) Piazzon, A.; Vrhovsek, U.; Masuero, D.; Mattivi, F.; Mandoj, F.;
Nardini, M. J. Agric. Food Chem. 2012, 60, 12312−12323.
(45) Akyol, S.; Ozturk, G.; Ginis, Z.; Armutcu, F.; Yigitoglu, M. R.;
Akyol, O. Nutr. Cancer 2013, 65, 515−526.
(46) Aura, A.-M. Phytochem. Rev. 2008, 7, 407−429.
(47) Merfort, I.; Heilmann, J.; Weiss, M.; Pietta, P.; Gardana, C.
Planta Med. 1996, 62, 289−292.
(48) Knuth, S.; Schubel, H.; Hellemann, M.; Jurgenliemk, G. Planta
̈
̈
Med. 2011, 77, 1024−1026.
(49) Murphy, T. H.; Miyamoto, M.; Sastre, A.; Schnaar, R. L.; Coyle,
J. T. Neuron 1989, 2, 1547−1558.
(50) Tan, S.; Schubert, D.; Maher, P. Curr. Top. Med. Chem. 2001, 1,
497−506.
(51) Walle, T. Free Radical Biol. Med. 2004, 36, 829−837.
(52) Arndt, S.; Seebach, J.; Psathaki, K.; Galla, H.-J.; Wegener, J.
Biosens. Bioelectr. 2004, 19, 583−594.
I
dx.doi.org/10.1021/np400518k | J. Nat. Prod. XXXX, XXX, XXX−XXX