Proc. Natl. Acad. Sci. USA
Vol. 93, pp. 12599-12604, October 1996
Fluoxetine-elicited changes in brain neurosteroid content
measured by negative ion mass fragmentography
D. P. UZUNOV*, T. B. COOPERt, E. COSTA*, AND A. GuIDOrrI*
*The Psychiatric Institute, Department of Psychiatry, University of Illinois at Chicago, Chicago, IL 60612; and tAnalytical Psychopharmacology Laboratory,
Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY 10962
Contributed by E. Costa, July 25, 1996
sham-operated or adrenalectomized/castrated (ADX/CX)
male rats dose-dependently (2.9-58 ,umol/kg i.p.) increased
the brain content of the neurosteroid 3c-hydroxy-5a-
pregnan-20-one (allopregnanolone, 3a,5a-TH PROG). The
increase of brain 3a,5a-TH PROG content elicited by 58
,umol/kg fluoxetine lasted more than 2 hr and the range of its
extent was comparable in sham-operated (--3-10 pmol/g) and
ADX/CX rats (2-9 pmol/g) and was associated with a de-
crease (from 2.8 to 1.1 pmol/g) in the 5a-pregnan-3,20-dione
(5a-dihydroprogesterone, 5a-DH PROG) content. The preg-
nenolone, progesterone, and dehydroepiandrosterone content
failed to change in rats receiving fluoxetine. The extent of
3a,5a-TH PROG accumulation elicited by fluoxetine treat-
ment differed in various brain regions, with the highest
increase occurring in the olfactory bulb. Importantly, fluox-
etine failed to change the 3a,5ca-TH PROG levels in plasma,
which in ADX/CX rats were at least two orders of magnitude
lower than in the brain. Two other serotonin re-uptake
inhibitors, paroxetine and imipramine, in doses equipotent to
those of fluoxetine in inhibiting brain serotonin uptake, were
either significantly less potent than fluoxetine (paroxetine) or
failed to increase (imipramine) 3a,5ci-TH PROG brain con-
tent. The addition of 10 ,uM of 5a-DH PROG to brain slices
of ADX/CX rats preincubated with fluoxetine (10IuM,15
min) elicited an accumulation of 3a,5a-TH PROG greater
than in slices preincubated with vehicle. A fluoxetine stimu-
lation of brain 3a,5ai-TH PROG biosynthesis might be oper-
ative in the anxiolytic and antidysphoric actions of this drug.
Fluoxetine administered intraperitoneally to
Neurosteroids such as 3a-hydroxy-5a-pregnan-20-one (allo-
pregnanolone, 3a,5a-TH PROG); pregnenolone (PREG) sul-
fate; and dehydroepiandrosterone (DHEA), sulfate, promptly
decrease or increase brain excitability acting as potent positive
(3a,5a-TH PROG) or negative (PREG sulfate, DHEA sul-
fate) allosteric modulators of y-aminobutyric acid (GABA)
action at GABAA receptors (1-6). These discoveries have
provided a new mechanism for brain GABAergic tone mod-
ification applicable to the treatment of various neuropsychi-
atric disorder symptoms (7, 8). These include the anxiety and
mood changes of the "late luteal dysphoria syndrome," asso-
ciated with low progesterone (PROG) plasma levels (a steroid
participating in 3a,5a-TH PROG biosynthesis) (8, 9).
Though the prospect that neuroactive steroids could be used in
the symptomatic treatment of specific neurological and psychi-
atric disorders has generated some enthusiasm, substantial diffi-
culties prevent the therapeutic use ofneurosteroids. For example,
the systemic administration of 3a,5a reduced derivatives of
PROG or androstenedione acting as positive allosteric modula-
tors of GABA action at GABAA receptors, indicates that the
doses required to elicit a clear anxiolytic, antidysphoric, and
antiepileptic activity may also produce profound sedation, motor
impairment, or ataxia (6, 8, 10). Two additional pharmacological
actions that may limit the protracted therapeutic use of neuro-
active steroids are: (i) their ability to trigger complex DNA
transcription modifications in neuronal (11) and glial cells (12)
and (ii) a possible tolerance liability, which may limit the pro-
tracted therapeutic use of these compounds in sleep disorders or
convulsive syndromes (13-15).
One might reduce the complications associated with the
protracted administration of neuroactive steroids by develop-
ing drugs that affect selectively some rate-limiting steps of
brain neurosteroid biosynthesis, which unlike that of periph-
eral endocrine tissues, is not under pituitary control (7).
Recently, we were intrigued by a report that fluoxetine's
beneficial effects in the treatment of "late luteal dysphoria"
symptoms occur after a latency time shorter than that de-
scribed for the treatment of the symptoms of depression (16).
Though the mechanism whereby fluoxetine relieves the symp-
toms of late luteal dysphoria remains uncertain, very likely it
may differ from that which alleviates the symptoms of depres-
sion. Since a decrease in brain availability of PROG metabo-
lites may contribute to the onset of late luteal dysphoria
symptoms (9), we began to investigate whether fluoxetine
could change rat brain levels of 3a,5a-TH PROG and its
precursors, including 5a-pregnan-3,20-dione (5a-dihydropro-
gesterone, 5a-DH PROG) and PROG.
We measured simultaneously in the same small brain area
subpicomole amounts of PROG, its Sa metabolites (3a,5a-TH
PROG and 5a-DH PROG), PREG, and DHEA using gas
chromatography negative ion mass fragmentography (GC/NICI-
MF). In fact, steroid radioimmunoassay technology, although
highly reliable for measures of PREG, PROG, and 3a,5a-TH
PROG in thenanomolar range (9,17), loses its intrinsic specificity
and sensitivity in measuring neurosteroids in the picomolar range
(18, 19). The present report, which documents that fluoxetine
increases 3a,5a-TH PROG brain content, also shows that this
action is unrelated to the inhibition of serotonin uptake (20)
elicited by this drug.
MATERIALS AND METHODS
Normal, sham-operated, andADX/CX Sprague-Dawley male
rats (Zivic-Miller), weighing 220-250 g were used. Food and
water or physiologic saline (ADX/CX rats) were available ad
libitum. ADX/CX and sham-operated rats were used for
experiments 10-15 days after surgery. To monitor the results
ofADX, the plasma levels ofcorticosterone were measured by
radioimmunoassay (ICN). All animal procedures employed
were in strict accordance with the National Institutes ofHealth
Abbreviations: PROG, progesterone; PREG, pregnenolone; DHEA,
dehydroepiandrosterone; 3a,5a-TH PROG, 3a-hydroxy-5a-pregnan-
20-one (allopregnanolone); 5a-DH PROG, 5a-pregnan-3,20-dione
(5a-dihydroprogesterone); GABA, y-aminobutyric acid; GC/NICI-
MF, negative ion chemical ionization-mass fragmentographic; ADX/
CX, adrenalectomized/castrated; HFBA, heptafluorobutyric acid an-
hydride; 5,-DH PROG, 5,-pregnane-3,20-dione; GC, gas chromato-
graph; SSRI, selective serotonin reuptake inhibitor.
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement" in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
Neurobiology: Uzunov et al.
Guide for the Care and Use of Laboratory Animals (21) and
were approved by the Animal Care Committee.
Studies on Brain Slices. ['4C]Serotonin uptake. The brain
uptake of [14C]serotonin was studied ex vivo with a modifica-
tion of the method of Shaskan and Snyder (22) in ADX/CX
rats receiving i.p. equimolar doses offluoxetine, paroxetine, or
imipramine. Brain slices (0.3 x 0.3 mm, -6 mg protein) were
incubated in 2 ml Locke's solution (154 mM NaCl/5.6 mM
KCl/3.6 mM NaHCO3/2.3 mM CaCl2/10 mM glucose/i mM
MgSO4/1 mM Hepes/10 ,uM pargyline/1 mM ascorbic acid,
pH 7.4) at 37°C for 5 min in the presence of 50 nM serotonin
(59.7 mCi/mmol; 1 Ci = 37 GBq).
The uptake was terminated by filtration through GF/B glass
fiber filters. A parallel assay at 0°C accounts for the [14C]sero-
tonin uptake by passive diffusion. The uptake of [14C]serotonin
detected in the presence of 1 ,tM to 1 mM fluoxetine (=30%) was
considered to be non-specific [due to the uptake of[14C]serotonin
by other monoaminergic uptake systems (22)] andwas considered
to be a background, which was subtracted from the total uptake
Conversion of 5a-DH PROG into 3ca,5a-TH PROG by rat
brain slices. The brain slices (=10 mg/protein in triplicate)
were preincubated in 1 ml of Locke's solution at 37°C for 15
min with fluoxetine prior to adding 10 ,tM 5a-DH PROG. The
incubation was terminated by the addition of4 ml ofice-chilled
ethyl acetate. Fluoxetine and 5a-DH PROG were dissolved in
dimethyl sulfoxide (final concentration of dimethyl sulfoxide
in the medium was 0.01%). Incubation without tissue was
performed to establish the nonenzymatic conversion of5a-DH
PROG, PREG, PROG, and 3a,5a-TH PROG, which ac-
counted for less than 2% of the total metabolites produced
when brain slices were incubated with the above-mentioned
Neurosteroid Determination. Extraction, HPLC, and TLC
purification. Whole rat brains, dissected brain structures, brain
slices, or plasma samples were homogenized in 5 vol of distilled
water containing 2 fmol/ml of the [3H]neurosteroid ofinterest.
The ethyl acetate extraction and the HPLC purification were
carried out as described by Cheney et al. (23). The recovery of
tritiated steroids by means of the HPLC purification ranged
from 83 to 94%.
An additional purification step on silica gel TLC was
required for the fractions containing 5a-DH PROG to mini-
mize possible contamination with cholesterol. The TLC plates
were developed twice in one direction with chloroform/ethyl
acetate/ethanol (30:1:0.1). The retardation factor values for
5a-DH PROG and cholesterol were 0.71 and 0.52, respectively.
Standards (50 ,tg each) were run in parallel and were sprayed
with 2% perchloric acid and heated at 120°C for 15 min to
allow spot visualization of cholesterol and 5a-DH PROG. The
area of the TLC plate corresponding to the 5a-DH PROG
standard was extracted four times with 2 ml of methanol. The
3H-5a-DH PROG recovery through this procedure was cal-
culated and ranged between 90 and 95%.
Sample derivatization for GC/MF analysis. 3a,5a-TH
PROG, PREG, DHEA, and the internal standard 3,3-hydroxy-
5a-pregnan-20-one (3f3,5a-TH PROG) were derivatized with
heptafluorobutyric acid anhydride (HFBA) as described (23).
PROG, 5a-DH PROG, and the internal standard 513-
pregnane-3,20-dione (5,B-DH PROG) were derivatized with
O-(pentafluorobenzyl) hydroxylamine HCl (FLOROX re-
agent) according to the protocol provided by Pierce. The
amount of internal standard added to the HPLC and the TLC
fractions containing the individual neurosteroids extracted
from the brain or plasma was of the order of magnitude of the
neurosteroid concentration expected to be present in the
specimen to be analyzed.
GC/NICI-MF analysis. GC/MF analysis of the HFBA- and
FLOROX-derivatives was carried out on an HP 5988B mass
spectrometer coupled to an HP 5890 gas chromatograph (GC)
equipped with a J & W Scientific (Folsom, CA) capillary
column (DB-5, length 30 m, i.d. 0.25 mm, film thickness 0.25
,um). Helium was used as the carrier gas. Mass fragmentation
was performed with NICI using methane/ammonia (95:5) as
the reaction gas. Samples were injected at a column temper-
ature of 80°C. The oven temperature was programmed to
increase at a rate of 30°C per min until it reached 240°C, and
this temperature was maintained until the end of the chro-
matographic run. In the mass spectrometer the derivatized
steroids of interest, subjected to NICI, yielded negative ions in
the mass range between m/z 100 and m/z 700.
NICI/MF of PREG, PROG, 5a-DH PROG, 3a,c5a-TH
PROG, and DHEA. The gas chromatographic elution profile
and the negative ion mass fragmentation pattern of HFBA-
PREG, HFBA-3a,5a-TH PROG, FLOROX-PROG, and
FLOROX-Sa-DH PROG derivatives are shown in Figs. 1 and
2, respectively. In addition to the GC retention time charac-
teristic of each steroid, the structural identification of each
neurosteroid assayed was provided by its unique mass frag-
mentation pattern. By operating the mass spectrometer in the
single ion monitoring mode, we focused on the most abundant
ion fragment of each steroid derivative: m/z 474 for HFBA-
3a,5a-TH PROG, 472 and 492 for HFBA-PREG, 464 for
HFBA-DHEA, 491 for FLOROX-5a-DH PROG, 489 for
FLOROX-PROG, 474 for HFBA-3,3,5a-TH PROG, and 491
for FLOROX-513-DH PROG. Despite the fact that ion frag-
ments with the same 474m/z value were selected for the single
ion monitoring quantification of 3a,5a-TH PROG and its
internal standard, each steroid was reliably identified by virtue
of its different GC retention time.
Following derivatization with FLOROX, the 3-pentafluoro-
benzyl-oximes of 5a-DH PROG and PROG yielded syn and
anti isomers, which can be completely separated by GC (Fig.
1), but have an identical mass fragmentation pattern (Fig. 2).
Quantitative analysis ofneurosteroids by NICI/MF. Aliquots
of the HPLC and the TLC fractions corresponding to 5-20 mg
of brain or 10-20 ,ul of plasma and the respective internal
standards were derivatized. The standard curve of HFBA-
3a,5a-TH PROG prepared using 313,5a-TH PROG as the
internal standard is shown in Fig. 3. The area under the peak
ofeach known quantity of3a,5a-TH PROGwas divided by the
area under the peak of the internal standard. This ratio was
plotted against the quantity of 3a,5a-TH PROG, which was
used to generate the standard curve of Fig. 3. The detection
limit of HFBA-3a,5a-TH PROG, HFBA-PREG, and HF-
BA-DHEAwas -1.5 fmol, whereas that ofPROG and 5a-DH
PROG FLOROX derivatives was =3 fmol. In establishing the
maximal sensitivity of the assay we considered only peaks that
have a signal-to-noise ratio greater than five.
The precision of the method was estimated from the calcu-
lated concentration divided by the actual concentration per-
centage. The difference was less than 2% for each steroid
analyzed injected in quadruplicate. For 3a,5a-TH PROG, the
inter- and the intra-assay coefficients of variation were 3 and
5%, for pregnenolone 4 and 6%, for progesterone 7 and 9%,
and for 5a-DH PROG 5 and 8%, respectively. The inter- or
intra-assay variation of each steroid failed to reach statistical
significance (P < 0.05) using a one-way ANOVA.
In samples without tissue, but containing trace amounts of
the [3H]neurosteroid of interest the neurosteroid content was
below the detection limit.
Drugs and Reagents. PREG, PROG, Sa-DH PROG,
3a,5a-TH PROG, and DHEA were from Steraloids (Wilton,
NH). [14C]serotonin and all tritiated steroids with the excep-
tion of 3H-Sa-DH PROG were from New England Nuclear.
Tritiated Sa-DH PROG was enzymatically synthesized by
incubating 3H-3a,5a-TH PROG (50 Ci/mmol) in C6-2B cells
for 4 hr. 3H-Sa-DH PROG was purified from the cell extract
by HPLC and TLC using the conditions described above for
the purification of neurosteroids (see Materials and Methods).
Proc. Natl. Acad. Sci. USA 93(1996)
Proc. Natl. Acad. Sci. USA 93 (1996)
of3a,5a-TH PROG; B, theHFBA derivative ofPREG; C, theFILOROX
derivative of5a-DH PROG; D, theFLOROX derivative ofPROG. Total
ion current generated by approximately 100 pmol of each derivatized
steroid is recorded. The FLOROX derivative of 5a-DH PROG and
PROG yielded syn and anti isomers, which were separated by the GC.
Gaschromatographic retention times. A, theHFBAderivative
The purity of 3H-5a-DH PROG was assessed by TLC, dem-
onstrating that all the radioactivity in the sample comigrated
with authentic 5a-DH PROG. Imipramine was from Sigma.
Fluoxetine-HCl was kindly donated by Eli Lilly. Paroxetine-
HCI (BRL-29060-A) was a gift from SmithKline Beecham.
HFBA and FLOROX reagent were purchased from Pierce.
Unless otherwise specified, all organic solvents were ofHPLC
grade and were purchased from Fisher Scientific. Drugs for in
vivo studies were dissolved by initially mixing with 1-2 drops
of Tween 80 (Sigma) and then gradually adding deionized
distilled water to the volume required. The concentration of
Tween 80 was less than 0.1%. If required, sonication was
applied to obtain a clear solution.
Statistical Analyses. All results represent mean ± SEM.
Data were subjected to ANOVA followed by Duncan multiple
range post hoc comparison (24).
Brain and Plasma PREG, PROG, 5a-DH PROG, and
3a,5ca-TH PROG Content in ADX/CX and Sham-Operated
Male Rats Following Fluoxetine or Saline Treatment. Thirty
minutes after fluoxetine treatment (58 ,imol/kg i.p.) the brain
content of 3a,5a-TH PROG increased 4-fold compared with
vehicle-treated ADX/CX rats, whereas that of5a-DH PROG
decreased (Fig. 4). In contrast, the brain content of PROG,
PREG, andDHEAvirtually failed to change significantly (Fig.
4). Furthermore, fluoxetine failed to alter the 3a,5a-TH
PROG plasma level, which consistently was less than 1/20 of
the brain content (0.34 ± 0.047 SEM in saline and 0.39 ± 0.050
SEM pmol/ml in fluoxetine-treated rats, n = 6).
The extent of the selective brain 3a,5a-TH PROG increase
elicited by fluoxetine in ADX/CX rats was comparable to that
detected in sham-operated rats (Fig. 4), even though the brain
content of 3a,5a-TH PROG was significantly higher in sham-
operated than in ADX/CX rats (see Fig. 4 Inset).
As illustrated in Fig. 5, the accumulation of 3a,5a-TH
PROG in the brain ofADX/CX rats receiving fluoxetine was
proportional to the dose of the drug injected and peaked
(about 4-fold over the control level) at 15 min following an
injection of58 ,umol/kg offluoxetine. The brain concentration
of 3a,5a-TH PROG.thereafter slightly decreased from the
peakvalue but remained 2-3 times higher than that ofcontrols
at 120 min after the injection of fluoxetine.
The administration of fluoxetine to ADX/CX rats resulted
in an uneven increase of 3a,5a-TH PROG and in an uneven
decrease of 5a-DH PROG in the various brain areas studied
(Table 1). The olfactory bulb was the brain area with the
highest basal 3a,5a-TH PROG content, and the brain area
with the highest increase in 3a,5a-TH PROG content follow-
ing fluoxetine treatment. The brain content of 5a-DH PROG
had a regional distribution similar to that of3a,5a-TH PROG.
Itwas highest in the olfactory bulb and lowest in the brain stem
(Table 1). The frontal cortex and the cerebellum were the only
structures in which fluoxetine decreased significantly the
5a-DH PROG content.
Accumulation of Brain 3a,5a-TH PROG and Inhibition of
'4C[Serotonin] Uptake Following Administration of Fluox-
etine, Paroxetine, or Imipramine. Fluoxetine is a potent and
selective serotonin reuptake inhibitor (SSRI) (20). Thus, we
studied other serotonin reuptake inhibitors, such as parox-
etine, which is an SSRI 10-fold more potent than fluoxetine
(26), and imipramine, which is equipotent to fluoxetine in
inhibiting serotonin reuptake but is less selective because in
higher doses it also blocks norepinephrine reuptake (26).
Equimolar doses of fluoxetine, paroxetine, and imipramine
administered to ADX/CX rats produced in 30 min an almost
complete inhibition of serotonin reuptake measured ex vivo in
100 pmol of each steroid were used in the experiment.
Mass fragmentographic spectra of HFBA-3a,5a-TH PROG, and -PREG and FLOROX-5a-DH PROG and -PROG. Approximately
Neurobiology:Uzunov et al.
Neurobiology: Uzunov et al.
tions of 3a,5a-TH PROG, as indicated in the abscissa, we added before
derivatization a constant 20 fm amount of 313,5a-TH PROG as internal
standard. The ratio 3a,5a-THPROG/303,5a-THPROG on the ordinate
represents the ratio of the area under the peak of each known quantity
of3a,5a-TH PROG (m/z 474) divided by the area under the peak of the
internal standard (m/z 474). Each point represents mean ± SEM of at
least five determinations; r = 0.9998; P < 0.01.
3a,5a-TH PROG standard curve. To increasing concentra-
brain slices (Table 2). During this time, however, fluoxetine
induced a 4-fold increase in brain 3a,5a-TH PROG content,
whereas paroxetine induced a 2-fold increment of 3a,5a-TH
PROG content and imipramine failed to change 3a,5a-TH
PROG concentration (Table 2).
Fluoxetine Increases 3a,5ct-TH PROGAccumulation in Brain
Slices ofADX/CX Rats. Incubation of brain slices ofADX/CX
rats with fluoxetine (10 ,M; 15 min) followed by the addition of
10AMof 5a-DH PROG, the immediate precursor of 3a,5a-TH
PROG, resulted in a significant time-dependent greater accu-
mulation of 3a,5a-TH PROG than in vehicle-pretreated slices
(Fig. 6). It should be noted that rat brain slices ofADX/CX rats
loaded with 5a-DH PROG extensively convert 5a-DH PROG
into 3a,5a-TH PROG (-15%/hr in vehicle-treated slices and
25%/hr in fluoxetine-treated slices), reaching in 1 hr a 10-fold
3a,5a-TH PROG accumulation over the level measured at the
beginning of the incubation with 5a-DH PROG (Fig. 6).
Prompted by the recent report that fluoxetine is beneficial for
the treatment ofthe late luteal dysphoria symptoms (14), using
the GC/NICI-MF methodology we examined whether fluox-
etine can change the content of PROG, 5a-DH PROG,
3a,5a-TH PROG, or DHEA in the whole brain, in discrete
FL LOXETINE, AtinoIkga, i.p.
DHEA content in the brain of ADX/CX rats treated with vehicle or
fluoxetine (58 ,umol/kg i.p., 30 min). (Inset) The 3a,5a-TH PROG
brain content ofsham-operated rats treated with vehicle or fluoxetine.
Data represent the mean ± SEM of 5 to 6 rats; *, P < 0.05; **, P <
0.01 when vehicle-treated rats are compared with fluoxetine-treated
rats; oo, P < 0.05 when the value of3a,5a-TH PROG inADX/CX rats
is compared with that of sham-operated rats. To avoid circadian
variations of the neurosteroid content, the experiments were consis-
tently conducted between 2 and 4 p.m.
PREG, PROG, 3a,5a-TH PROG, 5a-DH PROG, and
brain regions, and in plasma of ADX/CX or sham-operated
The extraordinary sensitivity and reliability, plus the struc-
tural information achieved with GC/NICI-MF analysis of the
neurosteroids has allowed us to measure femtomole quantities
of 3a,5a-TH PROG and 5a-DH PROG and their precursors
or metabolites in 5-10 mg of brain tissue and 0.01 ml ofplasma
for the first time ever with a precise structural identification.
This absolute identification of steroids, even when they differ
by small substitutions in the chemical structure, such as PREG,
PROG, 5a-DH PROG, 3a,5a-TH PROG, and DHEA can be
obtained at concentrations that are at least two orders of
magnitude lower than those detected by radioimmunoassay or
by GC/MF using the mass spectrometer in the electron impact
mode [compare the present results with those of Cheney et al.
In rats, a single i.p. injection of fluoxetine produces a
dose-related conspicuous and protracted increase of brain
3a,5a-TH PROG content. This action of fluoxetine does not
require the presence of adrenals and gonads, it is not related
the whole brain ofADX/CXrats receiv-
ing i.p. injections of fluoxetine. (A) Dose
responseat 30 min followingfluoxetine.
(B) Time course followingadministration
of 58,Lmol/kgof fluoxetine. Datarepre-
sent the mean ± SEM of5 to 6 rats; *,P <
0.05;**,P < 0.01 when fluoxetine-treated
rats are compared with vehicle-treated
rats; oo,P < 0.05 when3a,5a-THPROG
content at 15 min following fluoxetine
injection (58 ,umol/kg i.p.) is compared
with the brain 3a,5a-TH PROG content
at other time points.
3a,5a-TH PROG content in
Proc. Natl. Acad. Sci. USA 93(1996)
Proc. Natl. Acad. Sci. USA 93 (1996)
to an increase in the circulating levels of3a,5a-TH PROG, and
as shown in Fig. 3, is independent from an increase in brain
content of PREG, PROG, 5a-DH PROG, or DHEA.
Thus, it is presumed that the effect of fluoxetine on the brain
levels of 3a,5a-TH PROG derives from a direct action of
fluoxetine on brain steroidogenesis rather than being the conse-
quence of a generalized increase of steroid biosynthesis in the
endocrine tissues. Because the fluoxetine-induced increase of
brain 3a,5a-THPROG content differs invarious brain structures
and appears related to the rank order of the endogenous
3a,5a-TH PROG concentration ofeach structure, the possibility
that fluoxetine influences either 3a,5a-TH PROG biosynthesis
or degradation rates in different brain areas must be entertained.
In the central nervous system, PROG is rapidly metabolized
by the action of Sa-reductases into 5a-DH PROG, which in
turn is transformed into 3a,5a-TH PROG by the action of
3a-hydroxysteroid oxidoreductases (3a-HSORs) (28, 29).
Both enzymes are nonuniformly distributed in the brain; the
highest content ofthese enzymes (27, 30) and of5a-DHPROG
and 3a,5a-TH PROG (see Table 1) is found in the olfactory
bulb. Thus, the nonuniform increase in brain 3a,5a-TH PROG
following fluoxetine (Table 1), which in ADX/CX rats occurs
without concomitant changes of PREG or PROG content
(Fig. 4), could be due to an action offluoxetine on the activities
of either Sa-reductases or 3a-HSORs expressed in the various
brain areas. To elucidate whether either Sa-reductases or
3a-HSORs are the target of fluoxetine action, we measured
the content of 5a-DH PROG in different brain areas, includ-
ing the olfactory bulb. Interestingly, not only did the 5a-DH
PROG content fail to increase in most brain regions, but also
in many brain areas it may even decrease (see in Table 1,
5a-DH PROG content in cortex and cerebellum). This sug-
gests that the mechanism by which fluoxetine increases brain
3a,5a-TH PROG content is not via the activation of 5a-
reductases but rather via a complex interaction with the
activity of the 3a-HSORs.
In partial support for this mechanism of action of fluoxetine is
the demonstration reported in Fig. 6 that in brain slices of
ADX/CX rats preincubatedwith fluoxetine and then loadedwith
5a-DH PROG, the rapid accumulation of 3a,5a-TH PROG is
accelerated when 10 ,uM offluoxetine is present. Importantly, 10
,tM of fluoxetine is approximately the concentration of this drug
found in the the brain 5 min after the administration of 34
,umol/kg i.p. of fluoxetine to rats (31). Since in brain slices the
degradation of 3a,5a-TH PROG to metabolites different from
5a-DH PROG is very slow (see legend of Fig. 6 and ref. 27) the
only mechanism by which, in brain slices, fluoxetine can increase
3a,5a-TH PROG content at the rate shown in Fig. 5 is either by
accelerating the reduction of5a-DH PROG to 3a,5a-TH PROG
or by inhibiting the oxidation of 3a,5a-TH PROG to 5a-DH
Time of Incubation with 5ca-DH PROG (min)
sion into 3a,5a-TH PROG in brain slices ofADX/CX rats. Fluoxetine
(10 ,uM) was preincubated for 15 min with slices before the addition
of 10 ,lM of 5a-DH PROG. The amount of 5a-DH PROG taken up
from the brain slices in 5 min of incubation was -90% of the amount
added in the medium (10 nmol) and was identical in the presence and
absence of fluoxetine. Vehicle was 100 ,ll of Locke's solution con-
taining 0.1% dimethyl sulfoxide. At time 0 are represented the levels
of 3a,5a-TH PROG in slices in which 10AMof Sa-DH PROG was
added and the reaction was immediately stopped by the addition of 4
ml of cold (0-2°C) ethyl acetate. Each bar is the mean ± SEM of 3
experiments. *,P < 0.01 when fluoxetine-treated slices were compared
with vehicle-treated slices; oo,P < 0.01 when 3a,5a-THPROG content
at time 0 is compared with the 3a,5a-TH PROG content at 15, 30, and
60 min. In experiments in which 3H-5a-DH PROG (0.5 ,uCi/10 nmol)
was added to brain slices, the HPLC column chromatography (for
conditions, see ref. 27) of the extract revealed two major radioactive
peaks, one corresponding to 5a-DH PROG and one corresponding to
3a,5a-TH PROG. No other major radioactive peaks were eluted from
the HPLC column under the experimental conditions described.
Fluoxetine-induced facilitation of Sa-DH PROG conver-
PROG. We have obtained preliminary evidence that the rate of
conversion of 3H-3a,5a-TH PROG to 3H-5a-DH PROG in rat
cortical brain slices is greatly reduced if the slices are pretreated
with fluoxetine (10 ,uM or higher).
In rats fluoxetine metabolism is rather slow (20); thus, the
ability of fluoxetine to increase and maintain for longer than
2 hr high levels of 3a,5a-TH PROG in the brain ofADX/CX
rats is in keeping with a prolonged decrease of the conversion
rates of3a,5a-TH PROG into 5a-DHPROG by a direct action
on brain 3a-HSOR activity by fluoxetine.
The mechanisms whereby fluoxetine alleviates depression
have been linked to the ability of fluoxetine to increase
receiving fluoxetine or vehicle
3a,5a-TH PROG and 5a-DH PROG content in brain structures of ADX/CX rats
10 ± 1.9*
2.9 ± 0.48t
4.6 ± 0.39t
3.4 ± 0.29t
1.1 ± 0.21
1.9 ± 0.29
31 ± 4.1**
5.4 ± 0.74**
12 ± 2.1**
7.1 ± 0.96t
1.4 ± 0.26
1.9 ± 0.57
13 ± 2.3*
4.1 ± 0.58t
5.8 ± 0.72t
3.9 ± 0.69
2.4 ± 0.17
1.4 ± 0.13
15 ± 2.4
2.2 ± 0.47t
6.4 ± 2.4
3.2 ± 1.2
0.8 ± 0.09**
1.4 ± 0.18
Neurosteroid content was measured 30 min after the i.p. injection of vehicle or 58 ,umol/kg of
fluoxetine. Brain parts were dissected as described by Glowinsky and Iversen (25). Data represent the
mean ± SEM of three experiments.
*P < 0.01 when the olfactory bulb is compared with all the other brain parts.
tP < 0.05 and **P < 0.01 when the neurosteroid content in brain regions of vehicle-treated rats is
compared with the corresponding regions of fluoxetine-treated rats.
tp < 0.05 when compared to the values of cerebellum and brain stem.
Neurobiology:Uzunov et al.