Fluoxetine targets early progenitor cells
in the adult brain
Juan M. Encinas, Anne Vaahtokari, and Grigori Enikolopov*
Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724
Communicated by James D. Watson, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, March 15, 2006 (received for review January 19, 2006)
Chronic treatment with antidepressants increases neurogenesis in
the adult hippocampus. This increase in the production of new
neurons may be required for the behavioral effects of antidepres-
sants. However, it is not known which class of cells within the
neuronal differentiation cascade is targeted by the drugs. We have
generated a reporter mouse line, which allows identification and
classification of early neuronal progenitors. It also allows accurate
quantitation of changes induced by neurogenic agents in these
distinct subclasses of neuronal precursors. We use this line to
demonstrate that the selective serotonin reuptake inhibitor anti-
depressant fluoxetine does not affect division of stem-like cells in
the dentate gyrus but increases symmetric divisions of an early
progenitor cell class. We further demonstrate that these cells are
the sole class of neuronal progenitors targeted by fluoxetine in the
adult brain and suggest that the fluoxetine-induced increase in
new neurons arises as a result of the expansion of this cell class.
This finding defines a cellular target for antidepressant drug
hippocampus ? neural stem cells ? neurogenesis ? dentate gyrus ?
a wide spectrum of mood disorders in adults (1); they also are
the cellular basis for the action of SSRIs is not clear. In addition to
its effects on neurotransmission, SSRI fluoxetine increases gener-
ation of new neurons in the dentate gyrus (DG) of the adult brain
a causative factor in the behavioral effects of this class of antide-
understanding depression and designing new therapeutic drugs.
However, the step within the neuronal differentiation cascade
targeted by SSRIs remains unknown. Particular targets (e.g., stem
cells vs. early progenitors vs. advanced neuroblasts) may imply
different molecular mechanisms of controlling cell division and
survival, different circuits affected by the drugs, and different
insights on the behavioral action of the drugs.
proliferation-differentiation cascade is the imprecision in quanti-
Accurate enumeration of changes in distinct subpopulations of
neuronal precursors by immunocytochemistry is problematic: High
of evaluating changes in particular subclasses of neuronal precur-
sors (e.g., in contrast to BrdU or thymidine labeling of cell nuclei,
where great precision can be achieved); this problem is particularly
acute in the young brain, where the number of neural stem and
progenitor cells is particularly high. Likewise, functional in vitro
assays for identifying neural stem and progenitor cells (e.g., for-
mation of neurospheres) are unable to provide confident measures
of changes on a scale commensurate with the action of antidepres-
sants [or many other reported inducers of neurogenesis (10, 11)
that, in many cases, induce a 30–40% increase in the number of
ntidepressant drugs of the selective serotonin reuptake inhib-
itor (SSRI) class (e.g., fluoxetine) are commonly used to treat
newly generated cells); furthermore, such assays presently cannot
be performed for small subregions of neurogenic areas.
To circumvent these problems, we have generated a reporter
mouse line that allows a quantitative assessment of changes in the
stem?progenitor cell compartment of the adult brain. We used this
discernable steps. We then used this reporter line to show that the
in the adult brain, increasing symmetric divisions of a particular
early neural progenitor class in the DG.
Defined Steps in the Neurogenesis Cascade in the DG. Expression of
nestin marks neural stem and progenitor cells; the regulatory
elements of the nestin gene direct reporter gene expression to the
neuroepithelium of the embryo and to stem and progenitor cells of
the adult brain (12–16). We used these elements to generate a
transgenic mouse line in which the reporter, a cyan fluorescent
protein, is fused to a nuclear localization signal (CFPnuc). The
developing nervous system and in the neurogenic areas of the adult
brain [the DG, subventricular zone (SVZ), rostral migratory
stream, and olfactory bulb]. Importantly, the distribution of the
stem?progenitor cells in the neurogenic areas of these mice can be
visualized as a dotted pattern corresponding to the nuclei of these
cells. This nuclear representation of stem?progenitor cells greatly
reduces the complexity of their distribution pattern and permits
their unambiguous enumeration (thus capturing the power of
BrdU- or thymidine-based enumeration of nuclei). Fig. 1 A–F
compares the structures of the SVZ and DG as revealed by
nestin-GFP (13). Whereas we were unable to generate accurate
counts of nestin- or nestin-GFP-positive cells, we were able to
unambiguously enumerate (by using confocal stereology) all of the
labeled nuclei in the SVZ and DG of the nestin-CFPnuc mice.
We have used this nestin-CFPnuc reporter line to define discrete
steps in the neuronal differentiation cascade in the DG (leading
from stem?progenitor cells to differentiated granule neurons),
based on the morphology of the cells, the marker proteins that they
express, and their mitotic activity (measured by BrdU incorpora-
tion). We identify six classes of cells in the neuronal lineage in the
DG of nestin-CFPnuc mice (Figs. 1 and 2).
The first class is represented by glial fibrillary acidic protein
(GFAP)-positive nestin-CFPnuc cells. The triangular soma and the
nuclei of these cells reside in the subgranular zone (SGZ); they
extend a single- or double-apical process radially across the granule
Conflict of interest statement: No conflicts declared.
Freely available online through the PNAS open access option.
Abbreviations: ANP, amplifying neuroprogenitor; DG, dentate gyrus; Dcx, doublecortin;
1 neuroblasts; NB 2, type 2 neuroblasts; NeuN, neuronal nuclei; Prox-1, homeobox pros-
neuroprogenitor; SGZ, subgranular zone; SSRI, selective serotonin reuptake inhibitor; SVZ,
*To whom correspondence should be addressed. E-mail: email@example.com.
© 2006 by The National Academy of Sciences of the USA
May 23, 2006 ?
vol. 103 ?
no. 21 ?
cell layer (GCL), terminating with elaborated arbors of very fine
leaf-like processes in the molecular layer (Fig. 1 G–K; see also ref.
13). This characteristic apical process is easily visualized with
antibodies to GFAP, nestin, vimentin, and brain fatty acid-binding
protein. Cells of this class have been described in detail (13, 17–21),
and they correspond to the most primitive, stem-like population in
the DG; note, however, that not all of the criteria of stem cells, e.g.,
ability to self-renew, have been demonstrated for these cells (22–
24). Only a small fraction of these cells (?2%) can be labeled by
BrdU after a short (2-h) pulse, indicating their low rate of division
and consistency with the quiescent state of these cells (18, 20); we
therefore designate these cells as quiescent neural progenitors
(QNP). We have not been able to detect instances of symmetric
plane of division perpendicular to the SGZ); however, these cells
can be seen undergoing asymmetric divisions (below).
The second class is represented by small (somatic diameter ?10
?m) round or oval cells located in the SGZ (Fig. 1 H–K). Similar
protein, and Sox2, but they do not stain for GFAP or vimentin and
stain very weakly for nestin (which may indicate that CFPnuc
protein persists in these cells longer than nestin, or that nestin is
doublecortin (Dcx), polysialic-acid neural cell adhesion molecule
(PSA-NCAM), or for markers of differentiated neurons [ho-
meobox prospero-like protein (Prox-1), ?III-tubulin, neuronal nu-
frequency (20–25% 2 h after a single injection of BrdU), indicating
cells as amplifying neural progenitors (ANP). They are often seen
in clusters extending along the SGZ (Fig. 1L); when the plane of
division of cells in these clusters is visible, it is most often perpen-
dicular to the SGZ, such that the daughter cells remain in the SGZ
(Fig. 1L). Importantly, a fraction of these cells are seen separating
from QNPs after mitosis; in each case, the division plane is parallel
or slightly oblique to the SGZ such that the daughter cell is
deposited beneath the QNP cell (Fig. 1 H–K) (the plane of division
may explain why these cells do not inherit GFAP or nestin, which
are predominantly localized to the apically positioned processes of
the QNPs but not to their soma). Together, our results suggest that
QNP cells, by undergoing asymmetric divisions, give rise to ANP
cells, which then propagate in the SGZ through a series of sym-
The next class of precursor cells, still located in the SGZ, ceases
to express nestin, Sox2, brain fatty acid-binding protein, or CFPnuc
and starts to express Dcx and PSA-NCAM (Fig. 1 M and N). A
small subclass (?1% of cells in this class) morphologically resem-
bles ANPs, carries short (1–5 ?m) horizontal processes (Fig. 1M),
and is the final population in the differentiation cascade that is
labeled by BrdU (19). Most of the cells in this class are represented
by larger (10–15 ?m somatic diameter) cells that extend longer
(10–30 ?m) horizontal processes in the plane of the SGZ and do
not incorporate BrdU (Fig. 1N). These cells stain for Dcx, PSA-
NCAM, Prox-1, and ?III-tubulin but do not express NeuN. Thus,
the bulk of this class is represented by postmitotic neuronal
precursors; we designate them as type 1 neuroblasts (NB1).
cade in the DG. (A) Expression of endogenous
nestin, detected by using a monoclonal antibody,
in the DG of nestin-GFP transgenic mice; the pat-
tern was the same in WT animals. (B) Expression of
GFP in the DG of nestin-GFP transgenic mice; note
that endogenous nestin in A is seen mostly in the
processes, whereas GFP is present in the processes,
the cytoplasm, and the nucleus. Tight packing of
cells in the SGZ prevents accurate enumeration
of nestin-GFP expressing cells. (C) Expression of
the SGZ of nestin-CFPnuc mice; transgene-express-
ing cells now are represented by their nuclei, thus
making accurate cell counts even in densely
packed areas possible. (D–F) Expression of nestin
(D) and GFP (E) in the SVZ of nestin-GFP mice and
CFPnuc (F) in the SVZ of nestin-CFPnuc mice. Note
that densely packed SVZ cells, which cannot be
accurately counted in D or E, can be quantified
easily in F. (G) GFP-expressing neural progenitor
cells in the DG of the nestin-GFP mice. The soma of
carry vertical processes, which cross the granule
cells layer and end as elaborated arbors in the
an antibody to GFAP). ANP cells lack the processes;
seen in close contact with QNP (note a QNP and an
ANP cell above and beneath the dashed line).
the DG generates ANP cells. (H) After BrdU labeling, cells with GFAP-labeled processes can be seen dividing (note the horizontal plane of division, dashed line)
and generating daughter cells that are deposited below, do not carry processes or arbors, and do not stain for GFAP. (I–K) Staining for GFAP (blue; I), BrdU (red;
J), and CFPnuc (green; K). (L) A QNP cell (arrow) generates an ANP cell (arrowhead) through an asymmetric division, with the plane of division parallel to the
(M and N) ANPs differentiate into NB1 cells, which start to express markers of young neurons. NB1 cells still are located in the SGZ, cease to express nestin or
nestin-CFPnuc, and start to express PSA-NCAM (green), Dcx, and Prox-1. A small subclass of these cells (M) resembles the ANP cells morphologically and is the
last cell population to incorporate BrdU. The majority of the NB1 cells (N) extends horizontal processes and does not incorporate BrdU. (O) NB1 cells evolve
grows vertically or obliquely and extends into the GCL. These cells express PSA-NCAM (green), Dcx, Prox-1, and NeuN. (P) NB2 cells progress into IN. The
morphology of these cells resembles that of mature granule neurons. They extend a single vertical apical process and have their soma in the granule cell layer.
They express NeuN and Prox-1 but still express PSA-NCAM (green) and Dcx. (Scale bars: A–F, 20 ?m; G, H, L–P, 5 ?m.)
Defining neuronal differentiation cas-
www.pnas.org?cgi?doi?10.1073?pnas.0601992103Encinas et al.
Cells of the next class, type 2 neuroblasts (NB2), are larger than
NB1 cells (somatic diameter ?15 ?m) and remain confined to the
SGZ. They extend longer (20–40 ?m) processes horizontally and
obliquely to the plane of the SGZ (Fig. 1O). They do not express
QNP or ANP markers (nestin, GFAP, vimentin, Sox2, brain fatty
acid-binding protein, or CFPnuc), and express Dcx, PSA-NCAM,
Prox-1, ?III-tubulin, and NeuN.
The next class of cells corresponds to immature neurons (IN).
They are larger than the cells of the previous classes (somatic
diameter 15–20 ?m), and their morphology resembles that of
mature granule cells of the DG (Fig. 1P). Their soma is round or
oval and can be found both in the SGZ and, mainly, in the GCL.
These cells carry a single apical process that branches in its distal
Prox-1, ?III-tubulin, and NeuN.
The next class represents differentiated granule neurons, with
cease to express PSA-NCAM and Dcx but express Prox-1, ?III-
tubulin, NeuN, and markers of mature granule neurons (e.g.,
calbindin; ref. 25).
The differentiation cascade in the DG of nestin-CFPnuc mice
thus can be divided into discrete steps based on the expression
of markers, morphology, and mitotic activity (Fig. 2).
Fluoxetine Increases Symmetric Divisions of Early Progenitors in the
potentially reflect changes in stem?progenitor cells, advanced neu-
roblasts, immature neurons, or in some combination of these
classes. We used our nestin-CFPnuc reporter line to investigate
changes induced by fluoxetine in each of the classes we identified
in the DG. We treated the animals with fluoxetine for 15 days,
labeled dividing cells with BrdU, and monitored selected cell
populations in the DG after 24 h by using confocal stereology
by 40.9% (538 ? 51 vs. 758 ? 58; P ? 0.013) after fluoxetine
administration, in line with previous reports on the effects of
found that after treatment, the number of CFPnuc-positive cells
(i.e., QNPs and ANPs together) increased by 24.7% (8,356 ? 622
vs. 10,422 ? 646; P ? 0.037; Fig. 3 D–F). When these cells were
divided into QNP and ANP classes based on expression of GFAP,
the QNP class showed no change (4,516 ? 582 vs. 4,675 ? 518; Fig.
431 vs. 5,745 ? 506; P ? 0.012; Fig. 3H). No change was detected
in the volume of the GCL, including the SGZ, between the control
and experimental animals (0.468 ? 0.039 vs. 0.483 ? 0.052 mm3).
The number of PSA-NCAM-positive cells (which include NB1,
NB2, and IN cells, Fig. 4 A and B) was increased by 26.5 ? 7.2%
(8,936 ? 577 vs. 11,298 ? 719; P ? 0.022) (identical changes were
seen for Dcx-positive cells; note that Dcx and PSA-NCAM colo-
calized in both control and fluoxetine-treated animals; data not
shown). When these cells were subdivided further by using the
criteria described above, the number of NB1 cells was increased by
42.1% (4,918 ? 418 vs. 6,988 ? 538; P ? 0.089; Fig. 4C), and the
number of NB2 and IN cells remained unchanged (3,110 ? 209 vs.
3,452 ? 413 and 908 ? 11 vs. 858 ? 88, respectively) (Fig. 4 D and
E), compatible with the notion that the wave of increased prolif-
eration and differentiation has not reached those cell classes.
Thus, the earliest class affected by fluoxetine is the ANP cells
which are progeny of stem-like QNP cells. Importantly, the QNPs
themselves do not increase in number, consistent with the lack of
but only give rise to daughter ANP cells while keeping their own
number constant) or (ii) increased symmetric division of ANP cells
(i.e., the same number of ANPs may be born from QNPs, but they
then divide more frequently). To distinguish between these possi-
the neuronal differentiation cas-
cade in the DG. QNPs generate,
through asymmetric divisions, the
ANPs that, after several rounds of
symmetric divisions, exit the cell cy-
cle within 1–3 days and become
postmitotic NB1 cells. Within next
15–21 days, NB1 cells mature into
NB2 and then into IN with apical
processes and basal axons and the
soma located in the GCL. After an
additional 10–15 days, INs acquire
the characteristics of mature gran-
ule neurons, develop extensive
branching, and send long axonal
processes, forming the mossy fiber.
A schematic summary of
Encinas et al.
May 23, 2006 ?
vol. 103 ?
no. 21 ?
bilities, we counted the number of BrdU-labeled QNPs and ANPs.
We used triple labeling (CFPnuc, BrdU, and GFAP) to discrimi-
(Fig. 5). The number of BrdU-labeled QNPs was not affected by
the number of BrdU-labeled ANPs was increased 46.4% (280 ? 36
QNPs (Fig. 5C) and ANPs (Fig. 5D) did not change. These results
indicate that the rate of QNP cell division is unchanged and that
fluoxetine increases symmetric divisions of ANP cells. When con-
sidered together with the data on other cell classes, these results
suggest that ANPs are the only class of precursor cells in the DG
that directly respond to fluoxetine.
We also analyzed the changes in the SVZ, another major
neurogenic region (Fig. 7A, which is published as supporting
information on the PNAS web site). We did not observe changes in
the number of BrdU-labeled cells (10,058 ? 766 vs. 9,550 ? 769;
Fig. 7 B, D, and E), in agreement with the previous observations in
Fig. 7 C–E), or in their density (648 ? 55 vs. 687 ? 64 ? 103mm3),
Together, our data indicate that the fluoxetine-induced increase in
the number of early progenitor cells is specific for the DG and does
not affect the SVZ.
To investigate whether the fluoxetine-induced increase in pro-
administration. Fluoxetine increases the number of BrdU-positive cells (A). (B
with vehicle (B) and fluoxetine (C); dashed lines in B, C, E, and F outlines the
external limits of the DG. Exposure to fluoxetine also increases the number of
nestin-CFPnuc cells in the SGZ (D, histogram; E, section of the DG of a control
animal; F, section of the DG of a fluoxetine-treated animal). Within total
nestin-CFPnuc cells, the number of ANPs (H), but not QNPs (G), increases in
response to fluoxetine. (Scale bars: 50 ?m.) In all histograms, white bars
correspond to the vehicle injections (V), and gray bars to the fluoxetine
group in this figure) are shown as black dots.*, P ? 0.05.
Fluoxetine increases cell proliferation in the adult DG. Chronic (15
staining for PSA-NCAM (green), and nestin-CFPnuc (red). Two cell types are
distributed throughout the SGZ, often in close apposition to each other;
however, they do not overlap, as illustrated in B (PSA-NCAM cell is red, and
nestin-CFPnuc nuclei are green; note that colors are switched at low magni-
fication for better visualization). (C–G) Postmitotic precursors in the fluox-
increases the number of NB1 (C) but not of more advanced NB2 (D) or IN (E)
cells. V, vehicle; F, fluoxetine. n ? 8 per group.**, P ? 0.01. F and G are
representative photomicrographs of DG from control (injected with vehicle)
(F) and fluoxetine-treated (G) animals. (Scale bars: A, 20 ?m; B, 5 ?m; F and G,
Fluoxetine increases NB1 cells in the adult DG. (A and B) Immuno-
www.pnas.org?cgi?doi?10.1073?pnas.0601992103Encinas et al.
treatment and BrdU labeling as described above but killed the
animals 30 days (instead of 1 day) later. In this setting, the number
of BrdU-labeled cells was 46.2% higher in the fluoxetine-treated
group (234 ? 28 vs. 342 ? 24; P ? 0.037; Fig. 6A). The number of
BrdU-labeled NeuN-positive neurons also was higher, by 46.3%, in
did not change (92.7 ? 1.2 vs. 92.8 ? 1.6%; Fig. 6C); note that the
high percentage of BrdU-labeled cells that also stain for NeuN
indicates that with or without fluoxetine, the majority of surviving
newborn cells in the DG become granule neurons. No change was
detected in the volume of the GCL, including the SGZ,
between the control and experimental animals (0.496 ? 0.041
vs. 0.512 ? 0.050 mm3).
We also examined changes in the defined classes of precursor
cells in mice killed 30 days after the end of the treatment with
(Fig. 6 F–J), suggesting that once the exposure to fluoxetine ends,
experiments). (A–C) Chronic fluoxetine treatment of adult mice, analyzed 30
days after BrdU administration. Fluoxetine increases the number of BrdU-
cells (B); the fraction of such cells among total BrdU-positive cells remains the
same (C). (D and E) Representative photomicrographs of DG from control
(vehicle) (D) and fluoxetine-treated (E) animals show that new cells became
neurons, shown by immunostaining for BrdU (green) and NeuN (red). The
treatment. The histograms show the data for the QNP (F), ANP (G), NB1 (H),
NB2 (I), and IN (J) cells. Changes did not reach the level of significance in none
of the categories. V, vehicle; F, fluoxetine. n ? 6 per group.*, P ? 0.05.
Fluoxetine increases neurogenesis in the adult DG (30-day survival
Treatment with fluoxetine does not change the number of dividing (BrdU-
labeled) QNPs (A) but increases division of ANPs (B). The fraction of BrdU-
labeled QNP or ANP cells among total QNP or ANP cells, respectively, remains
A cluster of BrdU-positive ANP cells between two QNPs in the DG of a
fluoxetine-treated animal. QNP cells are identified by the presence of GFAP-
positive processes. CFPnuc is shown in green (F), BrdU in red (G), and GFAP in
blue (H). (Scale bar: 5 ?m.)
Fluoxetine increases proliferation of ANP cells in the DG. (A–D)
Encinas et al.
May 23, 2006 ?
vol. 103 ?
no. 21 ?