JOURNAL OF NEUROCHEMISTRY | 2019
14661
University of Alberta, Edmonton, AB, Canada
Abstract
Inflammatory insult to the central nervous system (CNS) can
lead to development of depression, and subsequently depres-
sion is the most frequent psychiatric comorbidity following
ischemic stroke, often limiting recovery and rehabilitation in
patients. The initiators of inflammatory pathways in the CNS
are microglia activated in response to acute ischemic stress,
and anti-depressants have been shown to have anti-inflam-
matory effects in the CNS, promoting neuronal survival
following ischemic insult. We have previously shown that the
selective serotonin reuptake inhibitors (SSRIs) fluoxetine and
citalopram promote neuronal survival after oxygen-glucose
deprivation, an in vitro model of ischemia, by attenuating the
release of glutamate and D-serine from activated microglia.
Interestingly, we found that fluoxetine-treated microglial cul-
tures contained fewer numbers of cells compared to other
groups and hypothesized that fluoxetine and citalopram atten-
uated the release of glutamate and D-serine by promoting the
apoptosis of microglia. The present study aimed to test and
Depression is a prevalent and highly debilitating mental
disorder affecting up to 15% of the population at least once in
their lifetime (Masi and Brovedani 2011). Although depressive
symptoms are well defined, the mechanisms involved in the
onset and progression of depression are widely debated. Many
hypotheses have been proposed to describe the neurobiology
of depression; however, hypotheses focused on monoamines,
Received September 25, 2018; revised manuscript received December
17, 2018; accepted December 20, 2018.
Address correspondence and reprint requests to Kathryn G. Todd,
Neurochemical Research Unit, Department of Psychiatry, University of
Alberta, 116th St and 85th Ave NW Edmonton, AB T6G2R3 Canada.
E-mail: kgtodd@ualberta.ca
Abbreviations used: 5HT, 5-hydroxytryptamine; ANOVA, analysis of
variance; BBB, blood brain barrier; CC3, cleaved caspase 3; CNS,
central nervous system; DIV, days in vitro; DMEM/F12, Dulbecco’s
modified Eagle medium/Ham’s F12 nutrient mix (1:1); EDTA, ethylene
diamine tetraacetic acid; EGTA, ethylene glycol-bis(b-aminoethyl ether)-
N,N,N0 ,N0 -tetraacetic acid; ELISA, enzyme linked immunosorbent assay;
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
doi: 10.1111/jnc.
compare antidepressants from three distinct classes (tricyclics,
monoamine oxidase inhibitors, and SSRIs) on microglial
apoptosis. Primary microglia were treated with 1 lg/mL
lipopolysaccharide and/or 10 lM antidepressants, and various
apoptotic markers were assayed. Fluoxetine and its metabolite
norfluoxetine decreased protein levels in cell lysates,
decreased cell viability of microglia, and increased the
expression of the apoptotic marker cleaved-caspase 3 in
microglia. Live/dead nuclear staining also showed that fluox-
etine- or norfluoxetine-treated cultures contained greater
numbers of dying microglial cells compared to vehicle-treated
cultures. Cultures treated with citalopram, phenelzine, or
imipramine showed no evidence of inducing microglial apop-
tosis. Our results demonstrate that fluoxetine and norfluoxetine
induce the apoptotic death of microglia, which may serve as a
mechanism to attenuate the release of glutamate and D-serine
from activated microglia.
Keywords: antidepressant, fluoxetine, microglia, norfluoxetine.
J. Neurochem. (2019) https://doi.org/10.1111/jnc.14661
the hypothalamic-pituitary-adrenal axis, and neurogenesis are
not sufficient to describe these mechanisms entirely. Consid-
erable work has demonstrated that inflammatory immune
processes may contribute to the etiology of depression, and
interact with many of the neurobiological and neurochemical
mechanisms described in the monoamine, hypothalamic-
pituitary-adrenal axis, and neurogenesis hypotheses. The
FBS, fetal bovine serum; GC-MS, gas chromatography mass spec-
troscopy; HPA, hypothalamic-pituitary-adrenal axis; HRP, horseradish
peroxidase; Iba1, ionized calcium-binding adapter protein-1; IETD, Z-
Ile-Glu(OMe)-Thr-Asp(OMe)-fluoromethylalketone; IL1 b, interleukin-
1b; LPS, lipopolysaccharide; MAOI, monoamine oxidase inhibitor;
MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide;
NEC, necrostatin-1; NED, naphthylethylenediamine dihydrochloride;
NO, nitric oxide; OGD, oxygen-glucose deprivation; PBS, phosphate
buffered saline; RRID, research resource identifier; SDS, sodium dodecyl
sulfate; SSRI, selective serotonin reuptake inhibitor; STS, staurosporine;
TCA, tricyclic antidepressant; TEEN, tris-EDTA-EGTA-NaCl buffer;
TNF, tumor necrosis factor; WB, western blotting.
1
2 K. S. Dhami et al.
innate immune cells responsible for initiating inflammation in
the central nervous system (CNS) are microglia. In the healthy
adult brain, microglia are found in a ramified state of active
surveillance, characterized by a rod-shaped body and dynamic,
elongated processes. In addition to their surveillance roles,
microglia carry out maintenance and homeostatic roles and
contribute to ongoing plasticity of the CNS by regulating
synaptic refinement, neurogenesis, and synaptic plasticity
(Davalos et al. 2005 and Nimmerjahn et al. 2005). Microglial
cells undergo morphological and functional changes in
response to acute tissue damage, as in traumatic brain injury,
infection, or stroke, and may be chronically activated in
various neurodegenerative diseases (Ling et al. 2001). Once
activated, microglia acquire an amoeboid morphology,
actively phagocytose, and secrete various cytokines and pro-
inflammatory factors including interleukin 1-beta (IL-1b)
(Kim et al. 2004 and Lai and Todd 2008), tumor necrosis
factor (TNF) (Jung et al. 2007 and Park et al. 2007), and nitric
oxide (NO) (Bi et al. 2005 and Wu et al. 2007).
Depression is a common psychiatric co-morbidity accom-
panying diseases such as Parkinson’s and Alzheimer’s
disease and insults such as stroke (Tanaka et al. 2006; Even
and Weintraub 2010; Raskin and Durst 2010; Weintraub
et al. 2010; Arbus et al. 2011 and El Husseini et al. 2012),
wherein it is also a major factor limiting recovery and
rehabilitation in stroke patients (Li et al. 2012). Acute stroke
causes irreversible damage to neurons within the ischemic
core and progressive, spreading injury to vulnerable neurons
in the surrounding penumbra, which is the pharmacological
target for neuroprotective treatment in acute ischemic stroke
(Liu et al. 2010). The primary objective in treatment of
ischemic stroke is to salvage the penumbra as much and as
early as possible, and it has been reported that roughly half of
all acute ischemic patients show potentially treatable penum-
bra on magnetic resonance imaging (Rivers et al. 2006).
The primary treatment for depression before or after a stroke
is selective serotonin reuptake inhibitor (SSRI) antidepres-
sants; however, many investigations have shown that various
classes of antidepressants also attenuate the pro-inflammatory
profile of microglia in the CNS (Obuchowicz et al. 2006;
Hashioka et al. 2007; Tynan et al. 2012). Previously, we have
shown in an in vitro co-culture model of ischemia that cultured
cortical neurons subjected to oxygen-glucose deprivation were
protected from inflammatory injury when microglia were pre-
treated with the SSRI antidepressants fluoxetine and citalo-
pram (Dhami et al. 2013). Various monoamine oxidase
inhibitor (MAOI) and tricyclic (TCA) class antidepressants,
including phenelzine, imipramine, and clomipramine were
also investigated; however, these antidepressants failed to
rescue the ischemic injured cortical neurons. Fluoxetine and
citalopram inhibited glutamate and D-serine release from
activated microglia, whereas the other antidepressants tested
failed to inhibit this release. Glutamate and D-serine released
from microglia can lead to excitatory neurotoxicity primarily
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
mediated by N-methyl-D-aspartic acid receptor signaling (Mat-
sui et al. 1995; Wu et al. 2004; Takeuchi et al. 2005, 2006).
Surprisingly, we observed a reduction in cell number and
protein levels of cultured microglia following fluoxetine
treatment, and hypothesized that fluoxetine and citalopram
induced the cellular death of microglia, reducing the release of
glutamate and D-serine from these inflammatory cells. In the
current exploratory study, we compare the effects of antide-
pressants from the TCA, MAOI, and SSRI classes on apoptotic
cellular death of cultured microglia. We present the novel
finding that treatment with fluoxetine and its metabolite
norfluoxetine induces the apoptotic cellular death of microglia.
This may serve as a mechanism to reduce the release of
neurotoxic glutamate and D-serine from activated microglia.
Methods and materials
Primary mixed glial cultures
All animal procedures were performed in accordance with the
University of Alberta Animal Care and Use Committee’s regulations
under approved Animal Use Protocol AUP#0000343. Mixed glia
were prepared from whole brains of postnatal day 1 Sprague-
Dawley rats (Charles River Laboratories, Montreal, QC, Canada,
CD 001, RRID: RGD_734476) acquired from the Science Animal
Support Services colony at the University of Alberta. Day 1 rat pups
were killed by conscious decapitation and meninges were removed
and cells were dissociated by enzymatic digestion with 0.25%
trypsin-EDTA (Gibco, Rockville, MD, USA; 25200056) for
25 min, followed by mechanical trituration in basal media [Dul-
becco’s Modified Eagle Medium (DMEM/F-12, Gibco 11320033)
supplemented with 10% fetal bovine serum (Gibco 12483) and 2%
penicillin/streptomycin (P/S, Gibco 15140122)]. Mixed glia were
plated on poly-L-lysine-coated (2 lg/mL, Sigma P0879) 12- or 24-
well plates at a density of 1 9 105 cells viable per well in basal
media, determined via trypan blue exclusion method using a
hemocytometer. Mixed glia were maintained at 37°C in a 5% CO2
humidified incubator with complete media changes every 3–4 days
in vitro (DIV) with basal medium. Mixed glia were cultured for 21
DIV to confluence prior to isolation of microglia.
Primary microglial culture
Microglia were isolated after 21 DIV from mixed glial cultures via
trypsinization (Saura et al. 2003). Briefly, media was removed from
mixed cultures and the wells were washed with serum-free DMEM/F-
12, trypsin (diluted 1:3 in DMEM/F-12) was added and plates were
returned to incubator until the adherent cell layer detached (approx-
imately 30 min). Trypsin and debris were aspirated and fresh basal
media was added to the wells and incubated for 10 min to inactivate
trypsin. Following inactivation, medium was aspirated and experi-
mental media (serum-free DMEM/F12 supplemented with 2% P/S)
was added to the wells. Cultures were incubated overnight at 37°C
with 5% CO2 and used for experiments the following day.
Microglial treatments
Representative antidepressants from three drug classes were used in
pretreatment of microglia prior to co-culture: imipramine (5–40 lM as

indicated in figure legends, TCA, Sigma I7379), phenelzine (5–40 lM
as indicated in figure legends, MAOI, Sigma P6777), and fluoxetine (5–
40 lM as indicated in figure legends, SSRI, Eli Lilly F03072),
citalopram (5–40 lM as indicated in figure legends, SSRI, H-
Lundbeck Lu10-171B), and norfluoxetine (5–40 lM as indicated in
figure legends, SSRI, Eli Lilly 141296). Staurosporine (Sigma S6942)
was used as a positive inducer of apoptosis at 100 ng/mL, while TNF (R
& D systems DY510, 20 ng/mL) was used as a positive inducer of the
extrinsic apoptotic pathway (Boldin et al. 1996; Fernandes-Alnemri
et al. 1996; Muzio et al. 1996). The caspase 8 inhibitor Z-Ile-Glu
(OMe)-Thr-Asp(OMe)-fluoromethylalketone (IETD, Enzo Life
Sciences ALX-260-144, 50 lM) was used to inhibit extrinsic pathway
activity and the RIP kinase inhibitor necrostatin-1 (30 lM, Sigma
N9037) was used to reverse this inhibition. All drugs were prepared as
100X stocks in aqueous solutions and diluted to the final concentrations
as indicated immediately prior to each experiment.
Assessment of microglial viability and protein levels
Microglial viability was assessed using the 3-(4,5-dimethylthiazol-
2-yl)-2,5-diphenyltetrazolium bromide (MTT, Sigma M5655) col-
orimetric assay. Briefly, 100 lL of 5 mg/mL MTT were added to
microglial wells and incubated for 40 min at 37°C, 5% CO2. Media
was aspirated and adherent cells were lysed with 210 lL of
dimethyl sulfoxide. Absorbance was read at 570 nm on a microwell
plate reader (SpectraMax M3, Molecular Devices).
Protein was assayed by the bicinchoninic acid method (BCA assay,
Pierce, Rockford, IL, USA; 23223) according to the manufacturer’s
protocol. Media was collected and adherent cells were gently washed
with phosphate buffered saline (PBS) (3x), and lysed with lysis buffer
(0.8% Triton X-100 and 0.2% sodium dodecyl sulfate (SDS) in tris-
EDTA-EGTA-NaCl buffer (50 mM Tris, 1 mM EDTA, 1 mM
EGTA, and 150 mM NaCl) and scraping. Lysates were measured in
duplicate by addition of bicinchoninic acid reagent [copper (II) sulfate
solution diluted in bicinchoninic acid (1 : 50; Pierce, as manufac-
turer’s protocol), after incubation at 37°C for 30 min and measure-
ment of absorbance at 562 nm.
Immunocytochemistry
Following experimental treatments, cells were washed (all subse-
quent wash steps were with PBS) and fixed with 10% phosphate-
buffered formalin for 10 min. After three washes with PBS, the cells
were incubated with a blocking solution [10% horse serum (Gibco
16050122) in PBS with 0.1% Triton X-100] for 1 h, washed, and
incubated and co-labeled with Iba-1 (1 : 1000, Wako RRID:
AB_839504) or cleaved-caspase 3 (1 : 500, Cell Signaling Tech-
nology, Beverly, MA, USA; RRID:AB_2070042) primary antibod-
ies (diluted in PBS with 1% horse serum) overnight at 4°C. After
subsequent washing, appropriate fluorescent-conjugated secondary
antibodies (Alexa Fluor 488 or 647, 1 : 200, Invitrogen RRID:
AB_2536183, RRID:AB_162542, RRID:AB_2535792, and RRID:
AB_141607) were diluted in PBS for 1 h at 21°C. After a final
wash, wells were mounted using aqueous mounting medium
(Vectashield, Vector Laboratories, Burlingame, CA, USA). Images
were acquired on a Leica DM-6000 inverted fluorescence micro-
scope (Leica Microsystems, Wetzlar, Germany). Nine images were
captured per well for each filter, image overlays were created using
Image-J. Image acquisition and analysis was carried out by a user
blinded to the experimental conditions.
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Fluoxetine induces microglial apoptosis 3
Nitrite assay
Nitrite levels, an indirect measure of NO, were determined by the
Griess Reaction as described in previous reports (Green et al. 1982;
Lai and Todd 2008). Briefly, differentially treated media was
collected and nitrite measured in duplicate by incubation with equal
parts 1% sulfanilamide in 3N HCl and 0.1% naphthylethylenedi-
amine dihydrochloride in distilled water. Absorbance (540 nm) was
read on a plate reader (SpectraMax M3, Molecular Devices, Palo
Alto, CA, USA). Values were calculated from measurement of
parallel nitrite standards (Sigma, St Louis, MO, USA). Extrapolated
nitrite values were normalized to protein measured in the corre-
sponding cell lysate.
Enzyme-linked immunosorbant assay (ELISA)
Quantification of TNF was performed using ELISA kits (R & D
Systems, Minneapolis, MN, USA; DY510) according to the
manufacturer’s protocol. Briefly, plates were coated with capture
antibody (4 lg/mL anti-TNF in PBS) overnight, blocked with 1%
bovine albumin (BSA) in PBS for 1 h, and samples and standards
loaded overnight at 4°C. After washing with PBS, signal was
detected with sequential incubation of detection antibody [225 ng/
mL biotinylated anti-TNF in PBS + 1% BSA, 2 h at 21°C],
horseradish peroxidase-conjugated streptavidin [diluted 1 : 200 in
PBS + 1% BSA, 20 min at 21°C], and color developed with
tetramethylbenzidine (TMB) (270 mg of TMB in 40% methanol,
100 mM sodium acetate, 18 mM citric acid, Sigma) with 0.03%
hydrogen peroxide. The reaction was stopped by adding 1.8 N
sulfuric acid to the wells and the absorbance was read at 450 nm
with background correction at 570 nm. Values were normalized to
protein levels of the corresponding cell lysate.
Live/dead staining of microglia
Live/dead assay was performed on cultured microglia using the
fluorescent nuclear stains SYTO 13 and SYTOX Orange (Life
Technologies, Grand Island, NY, USA; S7575 and S11368)
without fixation. Following treatments media was removed and
replaced with fresh DMEM/F-12 with SYTO 13 to a final
concentration of 5 lM and incubated at 37°C at 5% CO2 for
110 min, followed by SYTOX orange (0.1 lM) for 10 min. The
media was then removed, and wells were washed two times with
PBS and cover-slipped using aqueous mounting medium. Images
were acquired on a Leica DM-6000 inverted fluorescence micro-
scope (Leica Microsystems). Nine images were captured per well
for each filter, and merged images were created by Image-J
software.
Western blotting (WB)
WB was used to determine cleaved-caspase 3 (CC3), caspase 3
(C3), caspase 8 (C8) and b-actin expression cell lysates from vehicle
and antidepressant-treated microglia. Briefly, 9–12 microglial wells
(in 12-well plates) were pooled and lysed with tris-EDTA-EGTA-
NaCl buffer (0.8% Triton X-100 and 0.2% SDS) after 24 h of
antidepressant treatment. 15% acrylamide gels (37.5 : 1 bis-
acrylamide) were loaded with 15–22 lg of protein per well. The
gels were run for 2 h under a voltage of 120 volts using BioRad
(Hercules, CA, USA) western blotting equipment in running buffer
(25 mM Tris, 250 mM glycine, and 3.5 mM SDS). Gels were then
placed in transfer buffer (40 mM glycine, 50 mM Tris, and 1 mM
4 K. S. Dhami et al.
SDS) and proteins were transferred to polyvinylidene fluoride
membrane (Millipore Corporation, Bedford, MA, USA) at 4°C
overnight at 0.08 amps. The following day, the membranes were
washed with wash buffer (20 mM Tris, 140 mM NaCl, and 1 : 50
Tween) and blocked for 1 h with 10% skim milk powder (diluted in
wash buffer) with gentle agitation at 21°C. After washing, the primary
antibodies [rabbit anti-CC3 (1 : 1000, Cell Signaling RRID:
AB_2070042), rabbit anti-C3 (1 : 1000, Cell Signaling RRID:
AB_2069872), rabbit anti-C8 (1 : 1000, Cell Signaling RRID:
AB_10545768), or mouse anti-rat b-actin (1 : 5000, Sigma RRID:
AB_476744)] at the indicated dilutions (in PBS with 1% bovine
serum albumin and 0.02% sodium azide in wash buffer) were
incubated overnight on a shaker at 4°C. Subsequently, horseradish
peroxidase-conjugated 2° antibodies (diluted 1 : 10000 in PBS with
5% skim milk solution, Santa Cruz RRID:AB_641181, RRID:
AB_641170) were incubated for 1 h at 21°C and detected using
enhanced chemiluminescence (ECL, Amersham), exposed to Kodak
film for varying amounts of time, and developed using a Kodak
processor.
Gas chromatography-mass spectrometry (GC-MS)
Fluoxetine and norfluoxetine levels in medium were measured by
GC-MS using a modification of an assay described previously in
our laboratory (Rotzinger et al. 1997). Standards were prepared in
naive culture medium. Media collected from fluoxetine-, norfluox-
etine-, or vehicle-treated microglia (450 lL) was used in the
analytical procedure. Fluvoxamine (100 ng) was added as internal
standard to the samples, which were basified (to pH 8.5) by the
addition of 50 lL potassium bicarbonate. A solution of 2 lL
pentafluorobenzoyl chloride in 2 mL toluene was added to each
sample tube. The tubes were capped and shaken vigorously for
15 min and centrifuged at 1800 g for 5 min in a benchtop
centrifuge. The organic phase was pipetted into 100 mm 9
13 mm screw cap culture tubes and taken to dryness in a Savant
evaporator (Speed Vac SC 110, Fisher Scientific). Toluene
(100 lL) was added to each tube and the tubes were vortexed.
The toluene was transferred to a silanized glass insert and 2 lL was
injected onto an Agilent 6890 GC coupled to a 5973n mass
spectrometer with an electron ionzation source. Data were collected
in the single ion monitoring mode. The column used was HP-5MS
(5% phenyl methyl siloxane; 30 9 0.25 m, 0.25 lM), with a gas
flow rate of 1 mL/min helium.
Statistical analysis
Significance levels for data sets were calculated using a two-tailed t-
test for pairwise comparisons or a two-way ANOVA followed by
Bonferroni post hoc test for multivariate comparisons as appropriate.
Normality testing was not performed as the number of replicates
(< 10) is insufficient for such testing, however parametric tests were
used given the nature of sampling (biochemical sampling of cell
populations) assumes normality of the underlying biological analyte.
No tests for outliers were performed on data. Differences were
considered significant at a p < 0.05, denoted by * and †. For all
molecular analyses each ‘n’ represents an independent culture
preparation from a separate animal with three technical replicates
per condition, and appropriate numbers were determined based on
prior published examples using this model (Lai and Todd 2008;
Dhami et al. 2013).
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Results
Fluoxetine decreases the viability of primary cultured
microglia
Having previously demonstrated decreased microglial release
of D-serine and glutamate after treatment with either
fluoxetine or citalopram (Dhami et al. 2013), we sought to
investigate the possible mechanisms behind this result. Upon
close examination of data, we found that fluoxetine (10 lM)
significantly decreased the protein levels from microglial cell
lysates compared to vehicle-treated microglia and that this
effect was dose-dependent, as higher doses of fluoxetine (30
and 40 lM) lowered these protein levels even further
(Fig. 1a). As protein levels in the cell lysate are generally
proportional to the number of cells in culture, we hypoth-
esized that a possible mechanism behind the attenuation of
glutamate and D-serine release from activated microglia is by
the induction of microglial apoptosis after fluoxetine or
citalopram treatment. Measurement of microglial viability
after SSRI treatment showed that the addition of fluoxetine
similarly decreased the viability of microglia compared to
vehicle-treated controls in a dose dependent manner
(Fig. 1b). When examining the morphology of fluoxetine-
treated microglia under a phase-contrast microscope, we
observed that those cells treated with fluoxetine (10, 20, 30,
and 40 lM) showed an increased number of spherical-like
cells compared to vehicle-treated controls, which showed
more elongated cells and few spherical cells (data not
shown). The morphological transformation of microglia from
elongated to spherical is one classic hallmark of apoptosis.
Phenelzine, imipramine, and citalopram do not decrease the
viability of microglia
As phenelzine, imipramine, or citalopram have been shown
to exert anti-inflammatory effects on microglia, and citalo-
pram attenuated the release of glutamate and D-serine from
lipopolysaccharide (LPS)-activated microglia comparable to
fluoxetine [self-cite], we sought to determine if comparable
effects were seen on microglial viability. As expected,
phenelzine and imipramine did not decrease the level of
protein in cell lysates (Fig. 1c) or the viability of microglia
(Fig. 1d) at any dose tested, though contrary to our
expectations, citalopram did not affect the protein levels
(Fig. 1c) or viability of microglia (Fig. 1d) at any dose
tested. This disparity between the two SSRI antidepressants
fluoxetine and citalopram suggests that the effect of
fluoxetine on microglial viability may be unrelated to their
ability to inhibit serotonin reuptake.
Fluoxetine, but not citalopram, phenelzine, or imipramine,
increases the number of compromised microglial cells and
reduces the percentage of live cells.
In order to verify that the observed decrease in microglial
protein levels and viability following fluoxetine treatment
was a result of cellular death we used a SYTO/SYTOX live/

(a)
(c)
Fig. 1 Fluoxetine decreases protein levels in microglial cell lysates
and decreases microglial viability in a concentration-dependent
manner. (a) All fluoxetine-treated microglial cells showed a decrease
in the protein levels in cell lysates compared to vehicle-treated
controls. Microglia treated with 30 and 40 lM fluoxetine showed a
greater reduction in protein levels compared to 5 or 10 lM treatments,
n = 10. (b) All fluoxetine-treated microglial cells, excluding 5 lM
treatments, showed a decrease in MTT absorbance (viability) com-
pared to vehicle-treated controls. Microglia treated with 30 and 40 lM
fluoxetine showed a greater reduction in viability compared to 5 or
dead staining protocol to distinguish healthy intact cells from
compromised damaged cells. SYTO is a membrane perme-
able nuclear stain, while SYTOX is membrane impermeable,
thus cells with compromised plasma membranes will co-label
while healthy cells will only retain the green SYTO dye. We
found that the number of co-stained cells (dead cells) in the
imipramine (Fig. 2c), phenelzine (Fig. 2d), or citalopram
(Fig. 2e) groups did not differ compared to vehicle-treated
controls (Fig. 2a). As a positive control for the initiation of
cell death, we treated microglia with 100 nM of stau-
rosporine. After the treatment with fluoxetine (Fig. 2b) or
staurosporine (Fig. 2f), we found a significant increase in the
percentage of co-stained cells compared to vehicle-treated
controls (Fig. 2g). Therefore, we found that fluoxetine
induces cell death in microglia, whereas the other antide-
pressants tested did not have this effect.
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Fluoxetine induces microglial apoptosis 5
(b)
(d)
10 lM treatments n = 10. (c) Phenelzine, Imipramine, or citalopram
treatments of 5 or 10 lM showed no differences in the protein levels in
cell lysates compared to vehicle-treated controls, n = 8. (d) Phenel-
zine at 5 lM increased microglial viability compared to vehicle-treated
controls; whereas all the other concentrations of phenelzine, imipra-
mine, or citalopram tested had no effect, n = 8. Each ‘n’ represents an
independent culture preparation from a separate animal. Statistical
significance, *p < 0.05 (relative to control), †p < 0.05 (relative to
indicated condition).
Fluoxetine, but not citalopram, phenelzine, or imipramine,
increases cleavage of caspase 3 in microglia
As cell death can occur through a number of distinct
mechanisms, we measured the expression of CC3 as a
marker of apoptotic activity in antidepressant-treated micro-
glia. Apoptosis involves the cleavage of the protease caspase
3 within the cytoplasm to produce CC3; therefore, apoptotic
cells show decreases in caspase 3 and increases in CC3
levels. CC3 was first visualized in cultured microglia via
immunocytochemistry to measure the increase in apoptotic
activity in microglia following fluoxetine treatment. Com-
pared to vehicle-treated controls (Fig. 3a), microglia treated
with 10 lM fluoxetine displayed a significantly greater
number of cells co-labeled with both ionized calcium-
binding adapter protein-1 (Iba1) and CC3 (Fig. 3b). The
treatment of microglia with 10 lM of imipramine (Fig. 3c),
6 K. S. Dhami et al.

(a) i iii

ii

(b)
i iii

ii

(c) (d)

Fig. 2 Live/dead assay of antidepressant-

treated non-activated microglia. (a) SYTO
green (i), SYTOX orange (ii), and the
merged image (iii) of STYO green and
SYTOX orange in vehicle-treated control
cells. (b) SYTO green (i), SYTOX orange
(e) (f) (ii), and the merged image (iii) of STYO
green and SYTOX orange of fluoxetine-
treated cells. (c–f) The merged images of
SYTO green and SYTOX orange of
imipramine (c), phenelzine (d), citalopram
(e), and staurosporine (f) treated cells. (g)
Quantification of the percentage of co-
stained microglia for the live and dead
stains across different treatment
conditions. Fluoxetine (10 lM) significantly
increased the number of microglial cells co-
(g) stained with the live and dead stain
compared to vehicle-treated controls;
whereas phenelzine, imipramine, and
citalopram did not have this effect at the
same concentration tested. Staurosporine,
a positive apoptosis-inducing control,
showed similar increases in the number of
microglial cells co-stained with the live and
dead stain compared to fluoxetine-treated
microglia. Scale bars = 80 lm. For each
condition n = 4 independent culture
preparations from a separate animal with
pictures taken from 9 fields per well within
each experiment. Statistical significance,
*p < 0.05 (relative to control), †p < 0.05
(relative to indicated condition).

© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661

 Fluoxetine induces microglial apoptosis 7

(a) i iii Fig. 3 Immunofluorescence microscopy of Iba1 and cleaved-caspase

3 (CC3) expression in antidepressant-treated non-activated microglia.
(a) Iba1 labeling (i, red), CC3 labeling (ii, green), and the merged
image (iii) of Iba1 and CC3 labeling in vehicle-treated control cells. (b)
Iba1 labeling (i, red), CC3 labeling (ii, green), and the merged image
ii (iii) of Iba1 and CC3 in fluoxetine-treated cells. (c–f) The merged
images of Iba1 and CC3 labeling in imipramine (c), phenelzine (d),
citalopram (e), and staurosporine (f) treated cells. (g) Quantification of
the percentage of co-labeled microglia for both Iba1 and CC3 across
different treatment conditions. Fluoxetine significantly increased the
(b) number of cells co-expressing Iba1 (red) and CC3 (green) in microglia
i iii
compared to vehicle-treated controls; whereas imipramine, phenelzine,
or citalopram did not increase the number of cells co-expressing Iba1
and CC3 in microglia compared to vehicle-treated controls. Scale bar =
80 lm. For each condition n = 4 independent culture preparations
ii from a separate animal, with pictures taken from 9 fields per well.
Statistical significance, *p < 0.05 (relative to control), †p < 0.05
(relative to indicated condition). Examples of co-expressing microglia
are shown in merged images with arrows.

(c) (d)
phenelzine (Fig. 3d), or citalopram (Fig. 3e) showed no
differences in co-labeled cells compared to vehicle-treated
controls (Fig. 3a). As with the live/dead assay 100 nM of
staurosporine was used as a positive control for apoptosis
(Fig. 3f), and was found to significantly increase the number
of CC3 positive microglia (Fig. 3g). Western blot analysis
further showed that fluoxetine and staurosporine treatment
(e) (f) significantly increased CC3 levels as well as significantly
decreased caspase 3 levels within microglia compared to
vehicle-treated controls (Fig. 4a,d). Citalopram, imipramine,
or phenelzine treatment had no significant effect on CC3 or
caspase 3 expression levels (Fig. 4b and c). The cleavage of
caspase 3 to CC3 suggest that fluoxetine increased cell death
by inducing microglial apoptosis, whereas citalopram,
phenelzine, and imipramine do not.

Norfluoxetine attenuates the release of pro-inflammatory

(g)
factors from activated microglia and reduces microglial
viability
In a human subject, fluoxetine is primarily metabolized in the
liver to form norfluoxetine, a SSRI as potent as fluoxetine at
inhibiting 5-HT reuptake and with a much longer elimination
half-life than the parent compound (Sager et al. 2014). Since
norfluoxetine is highly prevalent in the depressed brain follow-
ing fluoxetine treatment, we investigated if norfluoxetine had
similar effects as fluoxetine in attenuating pro-inflammatory
factors released from microglia and reducing their cell viability.
First, we questioned whether microglia were able to convert
fluoxetine to norfluoxetine. To answer this question, we
measured the levels of fluoxetine and norfluoxetine in the
media of fluoxetine-treated microglia with GC-MS. If
microglia treated with fluoxetine converted the drug into
norfluoxetine, then we would measure a significant amount of
norfluoxetine within the medium compared to vehicle-treated
controls. GC-MS analysis 24 h after treatment with 10 lM

© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661

8 K. S. Dhami et al.
Fig. 4 Western Blot analysis for cleaved-caspase 3 and caspase 3
expression in microglial cell lysates following antidepressant treat-
ments. (a) Fluoxetine increased the expression of cleaved-caspase 3
(19 kDa) and decreased caspase 3 (35 kDa) expression compared to
vehicle-treated controls. (b) Citalopram did not alter the expression
levels of cleaved-caspase 3 or caspase 3 compared to vehicle-treated
controls. (c) Imipramine and phenelzine did not alter the expression
levels of cleaved-caspase 3 (19 kDa) or caspase 3 (35 kDa) compared
to vehicle-treated controls. (d) Staurosporine (a positive control)
fluoxetine showed that there were insignificant levels of
norfluoxetine present in the culture media, equivalent to only
0.09% conversion (Fig. 5a). As norfluoxetine is the major
metabolite present in the CNS during fluoxetine treatment we
measured the release of pro-inflammatory factors from non-
activated and LPS-activated microglial cells following norflu-
oxetine treatment. Norfluoxetine significantly attenuated the
release of NO and TNF from LPS-activated microglia (Fig. 5b
and c), and decreased the viability of microglia in a dose-
dependent manner comparable to fluoxetine treatment
(Fig. 5e). Additionally, norfluoxetine decreased the protein
levels within microglial cell lysates (Fig. 5d). Thus, while
microglia do not convert fluoxetine to norfluoxetine to any
significant extent, both of these SSRIs have similar effects in
decreasing microglial reactivity and viability.
Norfluoxetine increases the number of compromised
microglia and CC3 expression in microglia
To determine if norfluoxetine affects microglial viability by
induction of apoptosis as seen with fluoxetine we first
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
increased the expression of cleaved-caspase 3 (19 kDa) and
decreased the expression of caspase 3 (35 kDa) compared to
vehicle-treated controls. In all conditions, b-actin (43 kDa) expression
was measured as a loading control. (e) Histograms displaying cleaved
caspase 3 and (f) caspase 3 expression from western blotting (WB)
analysis, n = 5. Each ‘n’ represents an independent culture prepara-
tion from a separate animal. Statistical significance, *p < 0.05 (relative
to control), †p < 0.05 (relative to indicated condition).
co-stained 10 lM norfluoxetine-treated microglia with
SYTO and SYTOX live/dead stains. Norfluoxetine (Fig. 6a)
significantly increased the percentage of co-stained (dead)
cells compared to vehicle-treated controls (Fig. 6b), In
addition, we measured the number of microglial cells co-
labeled against Iba1 and CC3 following 10 lM norfluoxetine
treatment and found that norfluoxetine (Fig. 6c) significantly
increased the percentage of co-labeled cells compared to
vehicle-treated controls (Fig. 6d). Western blot analysis
showed that 10 lM norfluoxetine treatment increased the
expression of CC3 and decreased the expression of caspase 3
in microglia compared to vehicle-treated controls comparable
to the effects of fluoxetine (Fig. 6e–g).
Fluoxetine and norfluoxetine affect microglial apoptosis via
the extrinsic pathway
As apoptosis can occur in response to external (extrinsic
pathway) or internal stimuli (intrinsic pathway), we sought to
determine the site of effect of fluoxetine and norfluoxetine on
apoptosis. We measured the expression levels of caspase 8 in
 Fluoxetine induces microglial apoptosis 9

Fig. 5 Norfluoxetine effects on the release

of pro-inflammatory factors from activated
microglia and on cell viability. (a) Gas
chromatography revealed high levels of
fluoxetine in the medium of 10 lM
fluoxetine-treated microglia, whereas
norfluoxetine levels were virtually
nonexistent. Therefore, microglia were not
converting fluoxetine to norfluoxetine to any
significant extent in culture, n = 3.
Norfluoxetine (10 lM) significantly
attenuated the release of nitric oxide (NO)
(B) and tumor necrosis factor (TNF) (C) from
lipopolysaccharide (LPS)-activated
microglia, n = 10. Norfluoxetine decreased
microglial cell viability (E) and protein levels
(D) within cell lysates at all doses tested,
n = 10. Each ‘n’ represents an independent
culture preparation from a separate animal.
Statistical significance, *p < 0.05 (relative
to control), †p < 0.05 (relative to indicated
condition).

fluoxetine- or norfluoxetine-treated microglia to determine if norfluoxetine-, and staurosporine-treated groups relative to

the extrinsic pathway in apoptosis was activated. Western
blots revealed that both 10 lM fluoxetine and 100 nM
staurosporine decreased the level of caspase 8 in microglia
compared to vehicle-treated controls (Fig. 7a and b). As a
positive control, microglia were treated with 20 ng/mL TNF
to induce the extrinsic pathway of apoptosis. TNF similarly
decreased the level of caspase 8, but 10 lM norfluoxetine
failed to affect caspase 8 levels in treated microglia (Fig. 7a
and b). A decrease in caspase 8 levels is consistent with
cleavage of caspase 8 upstream to caspase 3 cleavage in the
extrinsic pathway of apoptosis. In contrast, CC3 levels were
significantly increased in fluoxetine-, norfluoxetine-, and
staurosporine-treated groups compared to vehicle-treated
controls (Fig. 7a, Fig. 4e, Fig. 6e and f), while caspase 3
levels were significantly decreased in TNF-, fluoxetine-,
control (Fig. 7a, Fig. 4f, Fig. 6e and g).
Interestingly, Fricker et al. (2013) found paradoxically
that inhibition of caspase 8 with IETD resulted in the cell
death of microglia and necrostatin-1, an inhibitor of a
downstream kinase involved in apoptotic death, prevented
microglial caspase 8 inhibition-induced death. Therefore,
caspase 8 activity may have a unique role microglial cell
death compared to other cell types such as neurons. Given
these findings, we hypothesized that fluoxetine may directly
inhibit the activity and expression of caspase 8 to induce the
death of microglia. We inhibited microglial activity of
caspase 8 with IETD and found a significant reduction in
MTT cell viability compared to vehicle-treated controls that
was reversed by treatment with necrostatin-1 (Fig. 7c).
Microglial cultures treated with fluoxetine also showed a

© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661

10 K. S. Dhami et al.
(a) i iii
ii
(c)
i iii
ii
(e)
Fig. 6 Immunofluorescence microscopy and Western blot analysis of
cell death markers in norfluoxetine-treated microglial cultures. (a)
SYTO green (i), SYTOX orange (ii), and the merged image (iii) of
STYO green and SYTOX orange of 10 lM norfluoxetine-treated cells.
(b) Quantification of the percentage of co-stained microglia for the live
and dead stains across different treatment conditions. (c) Iba1 labeling
(i, red), CC3 labeling (ii, green), and the merged image (iii) of Iba1 and
CC3 in 10 lM norfluoxetine-treated cells. (d) Quantification of the
percentage of co-labeled microglia for Iba1 and CC3 across different
treatment conditions. For each condition n = 4 independent culture
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
(b)
(d)
(f)
(g)
preparations from a separate animal, and each n represents pictures
taken from 9 fields per well. Scale bar = 80 lm. (e) Western Blot
analysis for CC3 and caspase 3 expression in microglial cell lysates
following 10 lM norfluoxetine or 100 nM staurosporine-treatment. (f–g)
Histograms displaying CC3 and caspase 3 expression following 10 lM
norfluoxetine treatment from western blotting (WB) analysis, n = 5.
Each ‘n’ represents an independent culture preparation from a
separate animal. Statistical significance, *p < 0.05 (relative to control),
†p < 0.05 (relative to indicated condition).

significant reduction in MTT cell viability compared to
vehicle-treated controls, but necrostatin-1 failed to reverse
this effect (Fig. 7d); therefore, it is unlikely the mechanism
of inducing death by fluoxetine is through the inhibition of
caspase 8 activity and expression in microglia.
Discussion
Given prior observation that treatment with fluoxetine and
citalopram decreased glutamate and D-serine release from
LPS-activated microglia, we hypothesized this effect was due
to induction of cell death SSRI antidepressants. Fluoxetine-
treated microglial cultures contained significantly lower
protein levels and reduced cell viability by MTT assay
compared to vehicle-treated cultures after 24 h treatment. In
addition, using permeability-based live-dead fluorescent
staining, fluoxetine-treated microglial cells showed a greater
number of co-stained cells, comparable with cells treated
with staurosporine as a positive control for cell death. These
effects were not seen with cultures treated with citalopram or
other antidepressants from the MAOI or TCA classes. The
expression of the apoptotic marker CC3, measured with
immuno-fluorescence microscopy and Western blot, was
significantly increased by fluoxetine but not citalopram
(SSRI), imipramine (TCA) or phenelzine (MAOI). Fluox-
etine induced the apoptotic death of microglia, and this is a
possible mechanism by which this drug attenuates the release
of glutamate and D-serine from these cells.
Contrary to our initial hypothesis, citalopram treatment did
not induce microglial apoptosis. As prior experiments have
demonstrated that both fluoxetine and citalopram attenuate
the release of glutamate and D-serine from activated
microglia, and as fluoxetine induces microglial apoptosis,
we hypothesized that citalopram would likewise induce
microglial apoptosis. Citalopram-treated microglial cultures
did not show any changes in the expression levels of cleaved-
caspase 3 or caspase 3 with Western blot or immuno-
fluorescence and citalopram had no effect on live-dead
staining compared to vehicle-treated controls. These unex-
pected results led to questions about the mechanisms of
citalopram’s effects on glutamate and D-serine release that
still require investigation.
After oral administration in human subjects, fluoxetine is
largely metabolized through type I metabolism in the liver,
with less than 10% excreted unchanged or as fluoxetine N-
glucuronide (Benfield et al. 1986). The remainder is metab-
olized to the active metabolite norfluoxetine and other
apparently non-active metabolites primarily by the enzymes
cytochrome P450 2D6 and 2C9 in the liver (Whirl-Carrillo
et al. 2012). Since the majority of fluoxetine is metabolized
into norfluoxetine in the human, we investigated if norflu-
oxetine also induced apoptosis of microglia in culture. We
found that norfluoxetine increased the expression of the
apoptotic marker CC3 while decreasing caspase 3, increased
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Fluoxetine induces microglial apoptosis 11
the number of microglia co-labeled with CC3 and Iba1 as
measured using immunofluorescence microscopy, and
increased the number of cells co-labeled with both the live
and dead stains. In addition, norfluoxetine decreased the
levels of protein within microglial cell lysates and reduced
MTT cell viability. These results raised an important issue in
that microglia may be converting fluoxetine into norfluox-
etine, and that norfluoxetine may be the active component
responsible for inducing apoptosis. To test this, we treated
microglia with fluoxetine and measured the amounts of both
norfluoxetine and fluoxetine in the cell media using GC-MS.
We found relatively insignificant amounts of norfluoxetine in
these lysates compared to the doses of norfluoxetine required
to induce cell death, demonstrating that the cultured
microglia converted fluoxetine into norfluoxetine to only a
minute degree. We conclude that both fluoxetine and
norfluoxetine can independently induce the apoptotic cell
death of microglia, without a requirement for conversion into
norfluoxetine by microglia in vitro.
Fluoxetine and norfluoxetine may induce apoptosis by one
of two primary pathways, the extrinsic or intrinsic pathway
(Slee et al. 2001). To investigate if these antidepressants
initiated the extrinsic pathway, we measured the level of
caspase 8 in microglial cell lysates by western blotting.
Caspase 8 associates intracellularly with death receptors that
are members of the TNF receptor gene superfamily and is
cleaved in response to receptor activation, going on to
activate other downstream caspases such as caspase 3
(Kischkel et al. 1995; Locksley et al. 2001). We found that
fluoxetine- or staurosporine-treated cells expressed lower
levels of caspase 8 compared to vehicle-treated controls.
Norfluoxetine did not decrease caspase 8 levels but decreased
caspase 3 levels while increasing cleaved-caspase 3 expres-
sion. When we treated microglia with TNF, a positive
inducer of extrinsic apoptosis, we found a decrease in
caspase 8 expression compared to vehicle-treated controls.
One unexpected result in this experiment is that TNF
treatment (positive control of extrinsic apoptosis) did not
show increases in CC3 expression, but did decrease caspase
3 levels in microglia. One possible explanation is that the
concentration of TNF may not be sufficient to induce
apoptosis, though the observed decreased caspase 8 and
caspase 3 levels suggests activation of the extrinsic pathway.
Interestingly, there is evidence that caspase 8 activity may
have a unique function in microglia in terms of inducing cell
death compared to other cell types such as neurons.
Pharmacological inhibition of caspase 8 in microglia results
in the cell death of microglia in a caspase 8 inhibition-
induced form of apoptosis not seen in other cell types.
Necrostatin-1, an inhibitor of a downstream kinase involved
in this death, prevents microglial caspase inhibition-induced
death (Fricker et al. 2013). We investigated the possibility
that fluoxetine induced microglial death by acutely inhibiting
the activity of caspase 8 with IETD. Inhibition of microglial
12 K. S. Dhami et al.

© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661


Fig. 7 Western Blot analysis for CC3, caspase 3, and caspase 8
expression in microglial cell lysates following fluoxetine and norfluox-
etine-treatments and the effects of caspase-8 inhibition (by Z-Ile-Glu
(OMe)-Thr-Asp(OMe)-fluoromethylalketone (IETD), fluoxetine, and
necrostatin-1 on microglial cell viability. (a) Representative Western
blots showing levels of cleaved caspase 3 (19 kDa), caspase 3 (35
kDa), b-actin (43 kDa, loading control), and caspase 8 (60 kDa) from
staurosporine, tumor necrosis factor (TNF), norfluoxetine, and fluox-
etine treated microglia cell lysates. (b) Histograms displaying caspase
caspase 8 significantly reduced cell viability compared to
vehicle-treated controls and viability was rescued by necro-
statin-1 treatment; however, necrostatin-1 failed to rescue
microglial cells following fluoxetine treatment. It is thus
unlikely that fluoxetine induces microglial death through the
inhibition of caspase 8, however, further experiments are
required to clarify the specific cell death pathway initiated by
fluoxetine.
In previous studies, we have shown that glutamate and D-
serine release from LPS-activated microglia are factors that
reduce the viability of oxygen-glucose deprivation-injured
cortical neurons. Fluoxetine is an antidepressant that
decreases the release of glutamate and D-serine from LPS-
activated microglia and attenuates the loss in neuronal
viability. In this study, we have shown that fluoxetine
induces apoptosis in microglia and this could be a mecha-
nism behind the attenuation in glutamate and D-serine
release. The treatment with SSRI antidepressants such as
fluoxetine can be either beneficial or detrimental, depending
on context of co-morbid depression following stroke. In
favor of beneficial effects of fluoxetine, microglia are in a
highly reactive and proliferative state as they release a variety
of inflammatory and toxic factors following ischemia, and
anti-inflammatory drugs such as minocycline and fluoxetine
have been shown to protect against ischemic insults (Stoll
et al. 1998; Schroeter et al. 1999; Yrjanheikki et al. 1999;
Lim et al. 2009). The frequency of depression following a
stroke is highest after 1 month and the risk remains elevated
for several years. The treatment of stroke patients with SSRI
antidepressants has been shown to reduce the incidence of
depression (Gothe et al. 2012; Loubinoux et al. 2012). For
instance, the SSRI fluvoxamine has been shown to alleviate
the diminished mood and sleep disturbances in patients with
post-stroke depression (Sunami et al. 2012). In addition,
chronic treatment with the SSRI citalopram has been shown
to delay the degeneration of dopaminergic neurons in the
midbrain, prevent striatal atrophy, and reduce the behavioral
symptoms in rodents following middle cerebral artery
occlusion (Kronenberg et al. 2012). One interpretation of
our findings is that the SSRIs fluoxetine and citalopram
attenuate the reactive and proliferative state of microglia by
reducing their release of inflammatory and toxic factors, most
importantly glutamate and D-serine, and ultimately reducing
neuronal toxicity after an ischemic insult (Dhami et al.
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Fluoxetine induces microglial apoptosis 13
8 expression by WB analysis, n = 5. (c) Inhibition of caspase 8 with
IETD (50 lM) significantly reduced MTT cell viability compared to
vehicle-treated controls and necrostatin-1 (30 lM) reversed this effect,
n = 8. (d) Fluoxetine treatment (10 lM) significantly reduced MTT cell
viability compared to vehicle-treated controls, similar to caspase 8-
inhibited conditions, however, necrostatin-1 failed to reverse this effect,
n = 8. Each ‘n’ represents an independent culture preparation from a
separate animal. Statistical significance, *p < 0.05 (relative to control),
†p < 0.05 (relative to indicated condition).
2013). In addition, the SSRI fluoxetine may induce
microglial apoptosis to help regulate the numbers of reactive
and proliferative microglia in a region of ischemic insult.
The effects of these SSRIs may be beneficial following
stroke, but there is evidence supporting a detrimental effect
of SSRI treatment prior to an ischemic event. In a 9-year
longitudinal study, Li et al. (2012) found that patients
diagnosed with depression and prescribed SSRI antidepres-
sants had higher rates of stroke in later life, and this finding
has been replicated by a number of studies (Bushnell 2011;
Andrews et al. 2012; Castro et al. 2012; Simmons et al.
2012). Smoller (2011) reported that depressed post-
menopausal women had a 45% increased risk of stroke if
prescribed SSRI antidepressants and the risk of fatal stroke
was doubled. In support of this, Ried et al. (2011) reported
that depressed patients prescribed SSRI antidepressants were
three times more likely to die following a stroke. The data in
our studies may help explain these findings. In the absence of
microglia, tissue damage has been shown to be exacerbated
following a stroke, supporting a beneficial role of microglia
(Lalancette-Hebert et al. 2007). We found that fluoxetine
induces microglial apoptosis in vitro, and if fluoxetine had
this same effect in vivo we would expect less microglia
available at times of insult and stress. This reduced
availability in microglia may cause the increased risk of
fatal stroke in these depressed patients.
In our study, we found that citalopram did not induce
microglial apoptosis, but others have found that citalopram
does increase the susceptibility to fatal stroke in depressed
patients (Bushnell 2011; Mikami et al. 2011). One limitation
in our study was that we only investigated the effects of
citalopram treatment on microglial apoptosis for 24 h. It is
possible that longer or chronic treatments of citalopram may
eventually induce microglial apoptosis. Similarly, as we did
not investigate the effects of the metabolites of citalopram on
the induction of microglial apoptosis, there remains a
potential for effects of metabolites on apoptosis. Citalopram
is metabolized into desmethylcitalopram by the enzymes
cytochrome P450 2C19 and 3A4 and desmethylcitalopram is
further metabolized into didesmethylcitalopram by cyto-
chrome P450 2D6 in the liver (Sangkuhl et al. 2011). These
metabolites are significantly less potent reuptake inhibitors of
5-HT, but we cannot rule out the possibility that these
metabolites induce microglial apoptosis. Although these
14 K. S. Dhami et al.
metabolites cross the blood brain barrier poorly compared to
their parent citalopram (Sangkuhl et al. 2011), studies should
investigate the effects of these metabolites on microglial
apoptosis.
Acknowledgments and conflict of interest
disclosure
The authors gratefully acknowledge funding from Alberta Health
Services and the Davey Endowment for Brain Research. The authors
have no financial conflicts to declare. This manuscript is part of a
PhD thesis uploaded to https://era.library.ualberta.ca/items/8222be
4c-015f-43b2-bc11-ffaf3cebd55d.
All experiments were conducted in compliance with the ARRIVE
guidelines.
Open science badges
This article has received a badge for *Open Materials*
because it provided all relevant information to reproduce the
study in the manuscript. The complete Open Science
Disclosure form for this article can be found at the end of
the article. More information about the Open Practices
badges can be found at https://cos.io/our-services/open-scie
nce-badges/.
References
Andrews P. W., Thomson J. A., Jr, Amstadter A. and Neale M. C.
(2012) Primum non nocere: an evolutionary analysis of whether
antidepressants do more harm than good. Front Psychol. 3,
117.
Arbus C., Gardette V., Cantet C.E., Andrieu S., Nourhashemi F., Schmitt
L., Vellas B. and Group R.F. (2011). Incidence and predictive
factors of depressive symptoms in Alzheimer’s disease: the
REAL.FR study. J Nutr Health Aging. 15, 609–617.
Benfield P., Heel R. C. and Lewis S. P. (1986) Fluoxetine. A review of
its pharmacodynamic and pharmacokinetic properties, and
therapeutic efficacy in depressive illness. Drugs 32, 481–508.
Bi X. L., Yang J. Y., Dong Y. X., Wang J. M., Cui Y. H., Ikeshima T.,
Zhao Y. Q. and Wu C. F. (2005) Resveratrol inhibits nitric oxide
and TNF-alpha production by lipopolysaccharide-activated
microglia. Int. Immunopharmacol. 5, 185–193.
Boldin M. P., Goncharov T. M., Goltsev Y. V. and Wallach D. (1996)
Involvement of MACH, a novel MORT1/FADD-interacting
protease, in Fas/APO-1- and TNF receptor-induced cell death.
Cell 85, 803–815.
Bushnell C. (2011) Depression and the risk of stroke in women: an
identification and treatment paradox. Stroke 42, 2718–2719.
Castro V. M., Gallagher P. J., Clements C. C. et al. (2012) Incident user
cohort study of risk for gastrointestinal bleed and stroke in
individuals with major depressive disorder treated with
antidepressants. BMJ Open 2, e000544.
Davalos D., Grutzendler J., Yang G., Kim J. V., Zuo Y., Jung S.,
Littman D. R., Dustin M. L. and Gan W. B. (2005) ATP mediates
rapid microglial response to local brain injury in vivo. Nat.
Neurosci. 8, 752–758.
Dhami K. S., Churchward M. A., Baker G. B. and Todd K. G. (2013)
Fluoxetine and citalopram decrease microglial release of glutamate
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
and D-serine to promote cortical neuronal viability following
ischemic insult. Mol. Cell Neurosci. 56, 365–374.
El Husseini N., Goldstein L. B., Peterson E. D., Zhao X., Pan W., Olson
D. M., Zimmer L. O., Williams J. W., Jr, Bushnell C. and
Laskowitz D. T. (2012) Depression and antidepressant use after
stroke and transient ischemic attack. Stroke 43, 1609–1616.
Even C. and Weintraub D. (2010) Case for and against specificity of
depression in Alzheimer’s disease. Psychiatry Clin. Neurosci. 64,
358–366.
Fernandes-Alnemri T., Armstrong R. C., Krebs J. et al. (1996) In vitro
activation of CPP32 and Mch3 by Mch4, a novel human apoptotic
cysteine protease containing two FADD-like domains. Proc. Natl
Acad. Sci. USA 93, 7464–7469.
Fricker M1, Vilalta A, Tolkovsky AM and Brown GC. (2013). Caspase
inhibitors protect neurons by enabling selective necroptosis of
inflamed microglia. J. Biol. Chem. 288(13):9145–52.
Gothe F., Enache D., Wahlund L. O., Winblad B., Crisby M., Lokk J.
and Aarsland D. (2012) Cerebrovascular diseases and depression:
epidemiology, mechanisms and treatment. Panminerva Med. 54,
161–170.
Green L. C., Wagner D. A., Glogowski J., Skipper P. L., Wishnok J. S.
and Tannenbaum S. R. (1982) Analysis of nitrate, nitrite, and
[15N]nitrate in biological fluids. Anal. Biochem. 126, 131–138.
Hashioka S., Klegeris A., Monji A., Kato T., Sawada M., McGeer P. L.
and Kanba S. (2007) Antidepressants inhibit interferon-gamma-
induced microglial production of IL-6 and nitric oxide. Exp.
Neurol. 206, 33–42.
Jung K. H., Kim M. J., Ha E., Kim H. K., Kim Y. O., Kang S. A., Chung
J. H. and Yim S. V. (2007) The suppressive effect of beta-glucan
on the production of tumor necrosis factor-alpha in BV2 microglial
cells. Biosci. Biotechnol. Biochem. 71, 1360–1364.
Kim W. K., Jang P. G., Woo M. S., Han I. O., Piao H. Z., Lee K., Lee
H., Joh T. H. and Kim H. S. (2004) A new anti-inflammatory agent
KL-1037 represses proinflammatory cytokine and inducible nitric
oxide synthase (iNOS) gene expression in activated microglia.
Neuropharmacology 47, 243–252.
Kischkel F. C., Hellbardt S., Behrmann I., Germer M., Pawlita M.,
Krammer P. H. and Peter M. E. (1995) Cytotoxicity-dependent APO-
1 (Fas/CD95)-associated proteins form a death-inducing signaling
complex (DISC) with the receptor. EMBO J. 14, 5579–5588.
Kronenberg G., Balkaya M., Prinz V. et al. (2012) Exofocal
dopaminergic degeneration as antidepressant target in mouse
model of poststroke depression. Biol. Psychiatry 72, 273–281.
Lai A. Y. and Todd K. G. (2008) Differential regulation of trophic and
proinflammatory microglial effectors is dependent on severity of
neuronal injury. Glia 56, 259–270.
Lalancette-Hebert M., Gowing G., Simard A., Weng Y. C. and Kriz J.
(2007) Selective ablation of proliferating microglial cells
exacerbates ischemic injury in the brain. J. Neurosci. 27, 2596–
2605.
Li C. T., Bai Y. M., Tu P. C., Lee Y. C., Huang Y. L., Chen T. J., Chang
W. H. and Su T. P. (2012) Major depressive disorder and stroke
risks: a 9-year follow-up population-based, matched cohort study.
PLoS ONE 7, e46818.
Lim C. M., Kim S. W., Park J. Y., Kim C., Yoon S. H. and Lee J. K.
(2009) Fluoxetine affords robust neuroprotection in the
postischemic brain via its anti-inflammatory effect. J. Neurosci.
Res. 87, 1037–1045.
Ling E. A., Ng Y. K., Wu C. H. and Kaur C. (2001) Microglia: its
development and role as a neuropathology sensor. Prog. Brain Res.
132, 61–79.
Liu S., Levine S. R. and Winn H. R. (2010) Targeting ischemic
penumbra: part I—from pathophysiology to therapeutic strategy. J.
Exp. Stroke Transl. Med. 15, 47–55.

Locksley R. M., Killeen N. and Lenardo M. J. (2001) The TNF and TNF
receptor superfamilies: integrating mammalian biology. Cell 104,
487–501.
Loubinoux I., Kronenberg G., Endres M., Schumann-Bard P., Freret T.,
Filipkowski R. K., Kaczmarek L. and Popa-Wagner A. (2012)
Post-stroke depression: mechanisms, translation and therapy. J.
Cell Mol. Med. 16, 1961–1969.
Masi G. and Brovedani P. (2011) The hippocampus, neurotrophic factors
and depression: possible implications for the pharmacotherapy of
depression. CNS Drugs 25, 913–931.
Matsui T., Sekiguchi M., Hashimoto A., Tomita U., Nishikawa T. and
Wada K. (1995) Functional comparison of D-serine and glycine in
rodents: the effect on cloned NMDA receptors and the extracellular
concentration. J. Neurochem. 65, 454–458.
Mikami K., Jorge R. E., Moser D. J., Arndt S., Jang M., Solodkin A.,
Small S. L., Fonzetti P., Hegel M. T. and Robinson R. G. (2011)
Increased frequency of first episode poststroke depression after
discontinuation of escitalopram. Stroke 42, 3281–3283.
Muzio M., Chinnaiyan A. M., Kischkel F. C. et al. (1996) FLICE, a
novel FADD-homologous ICE/CED-3-like protease, is recruited to
the CD95 (Fas/APO-1) death-inducing signaling complex. Cell 85,
817–827.
Nimmerjahn A., Kirchhoff F. and Helmchen F. (2005) Resting
microglial cells are highly dynamic surveillants of brain
parenchyma in vivo. Science 308, 1314–1318.
Obuchowicz E., Kowalski J., Labuzek K., Krysiak R., Pendzich J. and
Herman Z. S. (2006) Amitriptyline and nortriptyline inhibit
interleukin-1 release by rat mixed glial and microglial cell
cultures. Int. J. Neuropsychopharmacol. 9, 27–35.
Park Y. K., Chung Y. S., Kim Y. S., Kwon O. Y. and Joh T. H. (2007)
Inhibition of gene expression and production of iNOS and TNF-
alpha in LPS-stimulated microglia by methanol extract of
Phellodendri cortex. Int. Immunopharmacol. 7, 955–962.
Raskin S. and Durst R. (2010) Bupropion as the treatment of choice in
depression associated with Parkinson’s disease and its various
treatments. Med. Hypotheses 75, 544–546.
Ried L. D., Jia H., Feng H., Cameon R., Wang X., Tueth M. and Wu S.
S. (2011) Selective serotonin reuptake inhibitor treatment and
depression are associated with poststroke mortality. Ann.
Pharmacother. 45(7–8), 888–897.
Rivers C. S., Wardlaw J. M., Armitage P. A., Bastin M. E., Carpenter T.
K., Cvoro V., Hand P. J. and Dennis M. S. (2006) Do acute
diffusion- and perfusion-weighted MRI lesions identify final infarct
volume in ischemic stroke? Stroke 37, 98–104.
Rotzinger S., Todd K. G., Bourin M., Coutts R. T. and Baker G. B.
(1997) A rapid electron capture gas chromatographic method for
the quantification of fluvoxamine in brain tissue. J. Pharmacol.
Toxicol. Methods 37, 129–133.
Sager J. E., Lutz J. D., Foti R. S., Davis C., Kunze K. L. and Isoherranen
N. (2014) Fluoxetine- and norfluoxetine-mediated complex drug-
drug interactions: in vitro to in vivo correlation of effects on
CYP2D6, CYP2C19, and CYP3A4. Clin. Pharmacol. Ther. 95(6),
653–662.
Sangkuhl K., Klein T. E. and Altman R. B. (2011) PharmGKB
summary: citalopram pharmacokinetics pathway. Pharmacogenet.
Genomics 21, 769–772.
© 2019 International Society for Neurochemistry, J. Neurochem. (2019) 10.1111/jnc.14661
Fluoxetine induces microglial apoptosis 15
Saura J., Tusell J. M. and Serratosa J. (2003) High-yield isolation of
murine microglia by mild trypsinization. Glia 44, 183–189.
Schroeter M., Jander S., Witte O. W. and Stoll G. (1999) Heterogeneity
of the microglial response in photochemically induced focal
ischemia of the rat cerebral cortex. Neuroscience 89, 1367–1377.
Simmons A., Pimentel R. and Lakkireddy D. (2012) Sudden cardiac
death in women. Rev Cardiovasc Med. 13, e37–e42.
Slee E. A., Adrain C. and Martin S. J. (2001) Executioner caspase-3, -6,
and -7 perform distinct, non-redundant roles during the demolition
phase of apoptosis. J. Biol. Chem. 276, 7320–7326.
Smoller J. W. (2011) Do antidepressants raise the risk of stroke? Am. J.
Psychiatry 168, 457–459.
Stoll G., Jander S. and Schroeter M. (1998) Inflammation and glial
responses in ischemic brain lesions. Prog. Neurobiol. 56, 149–171.
Sunami E., Usuda K., Nishiyama Y., Otori T., Katsura K. and Katayama
Y. (2012) A preliminary study of fluvoxamine maleate on
depressive state and serum melatonin levels in patients after
cerebral infarction. Intern. Med. 51, 1187–1193.
Takeuchi H., Mizuno T., Zhang G., Wang J., Kawanokuchi J., Kuno R.
and Suzumura A. (2005) Neuritic beading induced by activated
microglia is an early feature of neuronal dysfunction toward
neuronal death by inhibition of mitochondrial respiration and
axonal transport. J. Biol. Chem. 280, 10444–10454.
Takeuchi H., Jin S., Wang J., Zhang G., Kawanokuchi J., Kuno R.,
Sonobe Y., Mizuno T. and Suzumura A. (2006) Tumor necrosis
factor-alpha induces neurotoxicity via glutamate release from
hemichannels of activated microglia in an autocrine manner. J.
Biol. Chem. 281, 21362–21368.
Tanaka S., Ide M., Shibutani T., Ohtaki H., Numazawa S., Shioda S. and
Yoshida T. (2006) Lipopolysaccharide-induced microglial
activation induces learning and memory deficits without neuronal
cell death in rats. J. Neurosci. Res. 83, 557–566.
Tynan R. J., Weidenhofer J., Hinwood M., Cairns M. J., Day T. A. and
Walker F. R. (2012) A comparative examination of the anti-
inflammatory effects of SSRI and SNRI antidepressants on LPS
stimulated microglia. Brain Behav. Immun. 26, 469–479.
Weintraub D., Mavandadi S., Mamikonyan E. et al. (2010) Atomoxetine
for depression and other neuropsychiatric symptoms in Parkinson
disease. Neurology 75, 448–455.
Whirl-Carrillo M., McDonagh E. M., Hebert J. M., Gong L., Sangkuhl
K., Thorn C. F., Altman R. B. and Klein T. E. (2012)
Pharmacogenomics knowledge for personalized medicine. Clin.
Pharmacol. Ther. 92, 414–417.
Wu S. Z., Bodles A. M., Porter M. M., Griffin W. S., Basile A. S. and
Barger S. W. (2004) Induction of serine racemase expression and
D-serine release from microglia by amyloid beta-peptide. J.
Neuroinflammation 1, 2.
Wu C. F., Bi X. L., Yang J. Y., Zhan J. Y., Dong Y. X., Wang J. H.,
Wang J. M., Zhang R. and Li X. (2007) Differential effects of
ginsenosides on NO and TNF-alpha production by LPS-activated
N9 microglia. Int. Immunopharmacol. 7, 313–320.
Yrjanheikki J., Tikka T., Keinanen R., Goldsteins G., Chan P. H. and
Koistinaho J. (1999) A tetracycline derivative, minocycline, reduces
inflammation and protects against focal cerebral ischemia with a
wide therapeutic window. Proc. Natl Acad. Sci. USA 96, 13496–
13500.
 Open Practices Disclosure

Manuscript Title: Fluoxetine and its metabolite norfluoxetine induce microglial apoptosis
Corresponding Author: Kathryn G. Todd

Articles accepted to Journal of Neurochemistry after 01.2018 are eligible to earn badges that
recognize open scientific practices: publicly available data, material, or preregistered research
plans. Please read more about the badges in our author guidelines and Open Science Badges
page, and you can also find information on the Open Science Framework wiki.

[X] Please check this box if you are interested in participating.

To apply for one or more badges acknowledging open practices, please check the box(es)
corresponding to the desired badge(s) below and provide the information requested in the
relevant sections. To qualify for a badge, you must provide a URL, doi, or other permanent path
for accessing the specified information in a public, open-access repository. Qualifying public,
open-access repositories are committed to preserving data, materials, and/or registered
analysis plans and keeping them publicly accessible via the web in perpetuity. Examples
include the Open Science Framework (OSF) and the various Dataverse networks. Hundreds of
other qualifying data/materials repositories are listed at http://re3data.org/. Preregistration of an
analysis plan must take place via a publicly accessible registry system (e.g., OSF,
ClinicalTrials.gov or other trial registries in the WHO Registry Network, institutional registration
systems). Personal websites and most departmental websites do not qualify as
repositories.

Authors who wish to publicly post third-party material in their data, materials, or preregistration
plan must have the proper authority or permission agreement in order to do so.

There are circumstances in which it is not possible or advisable to share any or all data,
materials, or a research plan publicly. For example, there are cases in which sharing
participants’ data could violate confidentiality. If you would like your article to include an
explanation of such circumstances and/or provide links to any data or materials you have made
available—even if not under conditions eligible to earn a badge—you may write an alternative
note that will be published in a note in the article. Please check this box if you would like your
article to include an alternative note and provide the text of the note below:

[] Alternative note:
[] Open Data Badge
1. Provide the URL, doi, or other permanent path for accessing the data in a public, open-
access repository:

[] Confirm that there is sufficient information for an independent researcher to reproduce

all of the reported results, including codebook if relevant.

[] Confirm that you have registered the uploaded files so that they are time stamped
and cannot be age.

[X] Open Materials Badge

1. Provide the URL, doi, or other permanent path for accessing the materials in a
public, open-access repository: all relevant information is provided in the manuscript and
custom-made materials will be provided upon reasonable request.

[X] Confirm that there is sufficient information for an independent researcher to

reproduce all of the reported methodology.

[] Confirm that you have registered the uploaded files so that they are time stamped
and cannot be age.

[] Preregistered Badge
1. Provide the URL, doi, or other permanent path to the registration in a public,
open-access repository*:

2. Was the analysis plan registered prior to examination of the data or observing the outcomes?
If no, explain.**

3. Were there additional registrations for the study other than the one reported? If yes, provide
links and explain.*
*No badge will be awarded if (1) is not provided, or if (3) is answered “yes” without strong
justification

**If the answer to (2) is “no,” the notation DE (Data Exist) will be added to the badge, indicating
that registration postdates realization of the outcomes but predates analysis.

By signing below, authors affirm that the above information is accurate and complete, that any
third-party material has been reproduced or otherwise made available only with the permission
of the original author or copyright holder, and that publicly posted data do not contain
information that would allow individuals to be identified without consent.

Date: 2018-12-21____________

Name: Matthew Churchward (for Kathryn Todd)

Signature: _____________________________