Synapse formation on neurons born in the adult hippocampus.

Abstract

Although new and functional neurons are produced in the adult brain, little is known about how they integrate into mature networks. Here we explored the mechanisms of synaptogenesis on neurons born in the adult mouse hippocampus using confocal microscopy, electron microscopy and live imaging. We report that new neurons, similar to mature granule neurons, were contacted by axosomatic, axodendritic and axospinous synapses. Consistent with their putative role in synaptogenesis, dendritic filopodia were more abundant during the early stages of maturation and, when analyzed in three dimensions, the tips of all filopodia were found within 200 nm of preexisting boutons that already synapsed on other neurons. Furthermore, dendritic spines primarily synapsed on multiple-synapse boutons, suggesting that initial contacts were preferentially made with preexisting boutons already involved in a synapse. The connectivity of new neurons continued to change until at least 2 months, long after the formation of the first dendritic protrusions.

Full text

Page 1

Synapse formation on neurons born in the
adult hippocampus
Nicolas Toni1, E Matthew Teng1, Eric A Bushong2, James B Aimone1, Chunmei Zhao1, Antonella Consiglio1,
Henriette van Praag1, Maryann E Martone2, Mark H Ellisman2& Fred H Gage1
Although new and functional neurons are produced in the adult brain, little is known about how they integrate into mature
networks. Here we explored the mechanisms of synaptogenesis on neurons born in the adult mouse hippocampus using confocal
microscopy, electron microscopy and live imaging. We report that new neurons, similar to mature granule neurons, were contacted
by axosomatic, axodendritic and axospinous synapses. Consistent with their putative role in synaptogenesis, dendritic filopodia
were more abundant during the early stages of maturation and, when analyzed in three dimensions, the tips of all filopodia were
found within 200 nm of preexisting boutons that already synapsed on other neurons. Furthermore, dendritic spines primarily
synapsed on multiple-synapse boutons, suggesting that initial contacts were preferentially made with preexisting boutons already
involved in a synapse. The connectivity of new neurons continued to change until at least 2 months, long after the formation of
the first dendritic protrusions.
Synapticplasticityconstantly reshapesthecircuitryoftheadultbrainin
response to external stimuli, including learning and memory forma-
tion1. Additionally, an evenmore dramatic source ofnetwork remodel-
ing involves the incorporation of new neurons2. Stem or progenitor
cells reside in the adult central nervous system, and neurons are
constantly added in the brain of mammals, including humans3. In
the adult hippocampus4, new neurons are functional and may be
importantinhippocampus-dependentlearning5–7,suggestingthatnew
neuronsfunctionally integrate intothe hippocampal network. Further-
more, we recently showed that during the second week after cell
division new neurons project axons in the hilus and dendrites in the
molecular layer, whereas dendritic protrusions only appear during the
thirdweekaftercelldivision8.Itisstillunclear,however,ifandhownew
neurons make mature and appropriate synaptic connections and to
what extent they integrate in the adult brain.
In the central nervous system, more than 90% of excitatory synapses
occur on dendritic protrusions with a bulbous head, referred to as
dendritic spines9. Protrusions vary in shape and length, and are
generally separated into three classes: filopodia, which are long and
thin with a fine tip and are very motile10–13; thin spines, which are
long and have a small spine head; and mushroom spines, in which
the diameter of the head is much larger than the diameter of the
neck14. Notably, the morphology of the protrusions is correlated with
function. Indeed, the density of the glutamate AMPA receptor is
correlated with spine-head size, and receptors are more abundant in
large spines than in small spines, whereas they are absent in filo-
podia15,16. Although filopodia are therefore unlikely to participate
in synaptic transmission, increasing evidence suggests that these
protrusions participate in synaptogenesis10,17,18. It is proposed that
these structures sample potential presynaptic partners and may initiate
physical contact with axons, sometimes resulting in the formation
of mature synapses18,19.
In the present study, we examined in situ the maturation of synapses
made on neurons born in the adult dentate gyrus. We used a Moloney
murine leukemia virus (MoMulV)-based retrovirus to express the
green fluorescent protein (GFP) specifically in proliferating cells and
their progeny4,8, and then examined the fine morphology of newly
generatedneurons(dentategranulecells;DGC)usingacombinationof
confocal microscopy, electron tomography, serial section electron
microscopy (ssEM) and live-cell imaging.
We found that, 30 days after viral injection (days post-injection;
d.p.i.), new neurons received a diversity of inputs similar to mature
granule neurons, including axosomatic, axodendritic and axospinous
synapses. Some of this input originated from the entorhinal cortex, as
suggested by close appositions between perforant path axonal boutons
and dendritic spines from new neurons, and by the presence of
axospinous synapses in the outer molecular layer. When analyzed in
three dimensions, the tip of dendritic filopodia was preferentially
associated with axon terminals already synapsing with dendritic spines
originating from different neurons. Furthermore, spines from new
neurons preferentially contacted multiple synapse boutons whereas, as
new neurons matured, spines formed synapses preferentially with
boutons devoid of other synaptic partners. Taken together, these
results indicate that new neurons preferentially contact preexisting
boutons and thattheirconnectivitychanges longafter the formationof
the first connections.
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
Received 30 January; accepted 13 April; published online 7 May 2007; doi:10.1038/nn1908
1Laboratory of Genetics, the Salk Institute for Biological Studies, 10010 N. Torrey Pines Road, La Jolla, California 92037, USA.2National Center for Microscopy and
ImagingResearch,UniversityofCaliforniaSanDiego,9500GilmanDrive,LaJolla,California92093,USA.CorrespondenceshouldbeaddressedtoF.H.G. (gage@salk.edu).
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 1
ARTICLES

Page 2

RESULTS
Protrusions’ density, dimensions and stability increase over time
To determine the time course of synapse formation, we analyzed the
morphology of GFP-expressing (GFP+) new neurons over time after
viral injection (21, 30, 60 and 180 d.p.i.). We compared those cells to
GFP+cells infected at postnatal day 4 (PND4) and analyzed at 56 d.p.i.
and to uninfected neurons (GFP–) from the same mice. Using confocal
microscopy, we observed that dendritic protrusion density increased
with time. At 21 d.p.i., dendrites displayed immature features such as
beading (local enlargement of dendritic diameter) and low dendritic
protrusion density. Dendritic protrusion density increased until 180
d.p.i. (one-way ANOVA, F5,105¼ 92.87, P o 0.001); most of this
increase occurred between 21 and 30 d.p.i., a period during which
about two-thirds of the protrusions were added (post hoc Bonferroni
test, P o 0.001), and between 60 and 180 d.p.i. (post hoc Bonferroni
test, P o 0.05; Fig. 1a,b). The volume of the tips of protrusions
correlates with the amount of glutamate receptors15,16. Using electron
tomographic reconstructions, we measured the volume of the tips of
dendritic protrusions and observed that, at 30 d.p.i., dendritic protru-
sions were on average smaller on GFP+neurons than on GFP–neurons
(Student’s t-test, P o 0.05; Fig. 1c,d).
To determine the nature of the protrusions that were added on new
neurons, we examined dendritic segments of newborn neurons in
hippocampal slices using time-lapse imaging. We imaged GFP+neu-
rons at 21, 28, 90 and 120 d.p.i. every 30 min, for a total of 90 min, and
analyzed the addition or retraction of protrusions during this time
(Fig. 2a). A one-way ANOVA test shows that the growth or the
retraction of protrusions significantly decreased with time after infec-
tion (F2,6¼ 14.57, P ¼ 0.005) and a post hoc Bonferroni test shows that
protrusionsfromGFP+neuronsat21d.p.i.weresignificantlylessstable
than protrusions from neurons at 90–120 d.p.i. (P o 0.005; Fig. 2b).
Whenwemeasuredthemaximaldiameterofthetipsofprotrusions,we
found that the thickness of protrusions predicted their stability:
protrusions with a tip thinner than 0.2 mm were more dynamic than
protrusions with a tip ranging from 0.2 to 0.4 mm (Student’s t-test,
P o 0.001), whichwere more motile than protrusions with a tip larger
than 0.4 mm (Student’s t-test, P o 0.001; Fig. 2c). Thus, the vast
majority of protrusions that were added on new neurons at 30 d.p.i.
were thin protrusions. Together, these data indicate that most protru-
sions were created between 21 and 30 d.p.i.; the protrusions at 30 d.p.i.
presented immature characteristics that we further explored using
ultrastructural approaches.
New neurons receive mature synaptic input
For electron microscopic analyses, GFP+neurons were microinjected
with the fluorophore Lucifer yellow and photoconverted using
3,3¢-diaminobenzidine tetrahydrochloride (DAB) as a substrate. We
found that GFP+DGCs at 30 d.p.i. had a large ovoid and clear nucleus
containing several nucleoli, features typical of granule neurons
(Fig. 3a). Using ssEM, we observed a variety of mature synaptic inputs
including axosomatic and axodendritic synapses, as well as axospinous
synapses (Fig. 3). We identified mature synapses on serial sections and
defined them by the presence of the following features on at least one
section: a postsynaptic density (PSD), more than four presynaptic
vesicles within 100 nm of the presynaptic membrane and a clearly
defined synaptic cleft.
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
21 d.p.i.
180 d.p.i.PND4
GFP+ 30 d.p.i.
GFP+ 30 d.p.i.
GFP–
*
21 d.p.i.
30 d.p.i.
60 d.p.i.
180 d.p.i.
PND4
GFP–
GFP–
***
*
0
0.01
0.02
GFP–
30 d.p.i.60 d.p.i.
Protrusion
volume (µm3)
a
c
bd
0
10
20
Protrusions per 10 µm
21 d.p.i. 28 d.p.i.
90–120 d.p.i.
0
4
8
**
12
Growth or retraction
(% of total protrusions)
Growth or retraction
(% of total protrusions)
b
a
c
0 min
30 min
60 min
90 min
>0.4 <0.2
0
0
***
0.2–0.4
Protrusion tip diameter (µm)
40
80
Figure 2 The motility of protrusions decreases with time after viral
infection. (a) Time-lapse confocal projections of dendritic segments showing
a protrusion retracting (arrow; left), growing (arrow; right) and a stable
protrusion (star). Scale bar, 5 mm. (b) Histogram showing the proportions of
protrusions that are growing or retracting in new neurons at 21 d.p.i. (n ¼ 3
neurons; 260 protrusions), 28 d.p.i. (n ¼ 2 neurons; 131 protrusions) and
90–120 d.p.i. (n ¼ 4 neurons; 737 protrusions). (c) Histogram showing the
distribution of motile protrusions between protrusion size (maximal tip
diameter). **P o 0.005, ***P o 0.001. Values are mean ± s.e.m.
Figure 1 Density and volume of dendritic protrusions on GFP+neurons
increase with time after viral infection. (a) Confocal projections of dendrites
from GFP+neurons at 21, 30, 60, 180 d.p.i., from GFP+neurons infected at
PND4 and from GFP–neurons. Scale bar, 10 mm. (b) Histogram showing the
linear density of dendritic protrusions in GFP+neurons at 21 d.p.i. (n ¼ 27
dendritic segments; 438 protrusions), 30 d.p.i. (n ¼ 16; 3,547 protrusions),
60 d.p.i. (n ¼ 26; 1,502 protrusions), 180 d.p.i. (n ¼ 6; 568 protrusions),
from GFP+neurons infected at PND4 (n ¼ 29; 2,314 protrusions) and
from GFP–neurons (n ¼ 7; 2,420 protrusions). (c) Electron tomographic
reconstructions of spiny dendrites of GFP+neurons at 30 d.p.i. and of
GFP–neurons. Scale bar, 2 mm. (d) Histogram presenting the average
protrusion volume in GFP+neurons at 30 d.p.i. (n ¼ 70 protrusions; 2
neurons) and in GFP–neurons (n ¼ 163 protrusions; 3 neurons). *P o 0.05,
**P o 0.001. Values are mean ± s.e.m.
2ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
ARTICLES

Page 3

The main synaptic input to the dentate gyrus originates in the
entorhinal cortex. To determine whether new DGCs received input
from the entorhinal cortex, we performed tracing experiments using
lentiviral vectors. Lentivirus induces stable expression of the transduced
genes in dividing as well as in quiescent cells. We injected a lentivirus
expressing the red fluorescent protein (RFP) into the entorhinal
cortex of adult mice and a MoMulV expressing GFP into the
dentate gyrus of the same mice. At 30 d.p.i., we observed RFP+axons
in the molecular layer of the dentate gyrus contacting dendritic spines
from newborn neurons (Supplementary Fig. 1 online). At least 10
contacts were identified, in which an apposition between pre- and
postsynaptic profiles was clearly visible on two-thirds of the
optical sections (data not shown). Thus, at 30 d.p.i., new DGCs had
gross morphological features and synaptic input similar to mature
granule neurons.
Geometry of dendritic protrusions predicts their connectivity
To gain insight into the connectivity of new neurons, we analyzed
dendritic protrusions using ssEM and three-dimensional (3D) recon-
structions. At 30 d.p.i., we found that 74.7 ± 4.3% of the dendritic
protrusions displayed a bulbous tip (tip dia-
meterlargerthan0.25mm),atypicalfeatureof
dendritic spines. All of them formed mature
synapses (Fig. 4). The remaining 25.2 ± 4.3%
of the protrusions displayed a fine tip (dia-
meter smaller than 0.25 mm), a typical feature
of filopodia, and all of them were devoid of
synapse at their tip (Figs. 4d and 5a). Con-
sistent with the electron tomography analysis,
which showed that protrusions on GFP+neu-
rons at 30 d.p.i. were on average smaller than
protrusions on GFP–neurons, the proportion
of smallprotrusions (tip smaller
0.25 mm) significantly decreased with matura-
tion, as shown by a one-way ANOVA test
(F4,12¼ 13.78, P o 0.001) and a post hoc
Bonferroni test between 30 and 180 d.p.i. on
protrusionssmaller than0.25mm (P o 0.001)
and between 30 d.p.i. and PND4 on protru-
than
sions ranging between 0.25 and 0.6 mm (P o 0.05; Fig. 4d). Notably,
64.1 ± 2.9% of GFP+dendritic spines contacted axonal boutons
involved in another synapse, thereby forming multiple-synapse bou-
tons(MSBs; Fig. 4b,e). The remaining spines contacted axon terminals
devoid of other synaptic contacts (single-synapse boutons (SSB);
Fig. 4a). MSBs predominantly contacted a total of two spines, but as
many as four spines were observedon a single boutonand the presence
of labeling on onlyone of them suggests that these spines came from at
least two different neurons. A one-way ANOVA test showed that the
proportion of MSBs decreased with time after infection (F4,13¼ 8.757,
Po0.001),andaposthocBonferronitestshowedasignificantdecrease
between 30 and 180 d.p.i. (P ¼ 0.01). The proportion of MSBs at 180
d.p.i. was similar to values found on GFP+neurons from mice injected
at PND4 and analyzed at 40 d.p.i., and on GFP–neurons (Fig. 4e).
Thus, the connectivityofnew DGCschangedbetween30d.p.i. and 180
d.p.i., and remained stable thereafter.
The observations that new protrusions are primarily immature
(Fig. 1) and that there is a shift between MSBs and SSBs occurring
duringmaturation(Fig.4e)promptedustoexaminewhetherfilopodia
participate in MSB formation. We reconstructed filopodia (that is,
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
a
b
c
Figure 3 At 30 d.p.i., GFP+neurons display
ultrastructure and synaptic input typical of
dentate granule neurons. (a) Electron micrograph
of a soma and an axosomatic synapse (square).
A higher-magnification inset shows a single
section micrograph (left inset) and a 3D
reconstruction (right inset) of the synapse in
the square. The panels on the right show two
examples of axo-somatic synapses. Scale bars,
5 mm; 0.5 mm in insets. (b) Electron micrograph
of a cross-section of a dendrite and of an
axodendritic synapse (rectangle). The insets show
a single section micrograph (top inset) and a 3D
reconstruction (bottom inset) of the synapse in the
rectangle. The panels on the far right show two
examples of axo-dendritic synapses. Scale bars,
1 mm; 0.5 mm in insets. (c) Electron micrograph
of a longitudinal section of a dendrite and of an
axospinous synapse (rectangle). The insets show
a single section micrograph (left inset) and a 3D
reconstruction (right inset) of the synapse in the
rectangle. The panels on the right show other
examples of axo-spinous synapses. Scale bar,
5 mm; 0.5 mm in insets.
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION3
ARTICLES

Page 4

protrusions with a tip diameter thinner than 0.25 mm) in three
dimensions and analyzed the immediate environment of their tip.
We found that the tip of each GFP+filopodium was surprisingly close
to anaxonalbouton(Fig.5a).Whenwemeasuredtheshortestdistance
between the tips of filopodia and the nearest bouton, we found that as
many as 84.3 ± 4.6% of the filopodia tips were within 50 nm of a
bouton and 100% were within 200 nm (Fig. 5b,c), regardless of the age
of the DGC they originated from (data not shown). The minimal
distance between filopodia tips and boutons was shorter than that
measured from a random distribution. Indeed, when filopodia were
artificially displaced by approximately 1 mm and 5 mm along the axis of
their dendrite so as to randomly position them in the extracellular
space (Fig. 5b), we measured that only 34 ± 9% of filopodia tips were
found within 50 nm of the closest bouton, and 100% were within
1,500 nm. A one-way ANOVA test showed an effect of position
(F2,6¼ 19.1, P o 0.005) and a post hoc Bonferroni test showed
a significant difference between the actual position and either
displaced position (P ¼ 0.006; Fig. 5c). Thus the association between
filopodia tips and boutons was closer than when measured in a
random distribution.
Finally, when we analyzed the axonal bou-
tons associated with the tips of filopodia (tip
diameter smaller than 0.25 mm), we found
that 96.4 ± 1.9% of the boutons that were
associatedwith a filopodiafrom a new neuron
werealsosynapsingwithatleastoneunlabeled
dendritic spine. Furthermore, spines with
headdiameterof0.25–0.6mmweremorelikely
than spines with heads larger than 0.6 mm
tosynapsewithMSBs.Aone-wayANOVAtest
showed an effect of the protrusions’ geo-
metry (F2,33¼ 69.26, P o 0.001) and a post
hoc Bonferroni test showed a significant dif-
ferencebetweenallgroups(Fig.6a),regardless
of the age of the neuron (data not shown).
Thus, small GFP+protrusions were more
likely to contact a bouton already involved
in a synaptic contact than large protrusions,
whereas large protrusions were more likely
devoidof otherto
This indicates that the morphology of the protrusions predicted
their connectivity.
When we combined our confocal microscopy results with the
classification of protrusions made using ssEM, we observed that the
increase in protrusion density with new neurons’ age was paralleled by
an increase in synapse density, indicating that synaptogenesis occurred
as new neurons matured and their dendrites bore more protrusions
(Fig. 6b). Furthermore, we observed that spines contacted primarily
MSBs at 30 d.p.i. and primarily SSB after 30 d.p.i. However,
the absolute density of spines contacting MSBs remained constant
through all ages.
As the size of the spine head predicted the connectivity of the axonal
boutons they contacted (Fig. 6a), and because spine heads became
larger as neurons matured (Figs. 1d and 4d), we hypothesized that, on
individual MSBs, spines synapsing with the same bouton may be
morphologically related. We analyzed the volume of spine heads
contacting the same MSB and found that large spines (40.05 mm3)
were coupled with spines smaller than 0.03 mm3and that two large
spines never synapsed on the same axonal bouton (Fig. 6c).
contact boutons synapticcontact.
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
a
b
c
d
e
GFP+ 30 d.p.i.GFP–
30 d.p.i.
60 d.p.i.
180 d.p.i.
80
40
0
80
40
0
30 d.p.i.
<0.25
Protrusion trip diameter (µm)
60 d.p.i.
180 d.p.i.
PND4
GFP–
>0.6
Percentage of protrusions
Percentage of MSB
*
*
***
0.25–0.6
PND4
GFP–
Figure 4 New neurons synapse primarily with
MSBs. (a,b) Electron micrographs (left three
panels) and 3D reconstructions (right) of SSB
(a) and MSB (b) synapses between GFP+neurons
and axonal boutons. Scale bar, 0.5 mm. (c) 3D
reconstruction of a dendritic segment from a GFP+
neuron at 30 d.p.i. (left) and a GFP–neuron
(right). Blue, axonal boutons; red, GFP–spines;
green, GFP+dendrites and spines. Scale bar,
1 mm. (d) Histogram showing the proportion
of protrusions in relation to the maximal diameter
of their tip, in GFP+neurons at 30 d.p.i. (n ¼ 3
neurons; 289 protrusions), 60 d.p.i. (n ¼ 3; 210
protrusions), 180 d.p.i. (n ¼ 3; 279 protrusions),
in GFP+neurons labeled at PND4 (n ¼ 4; 238
protrusions) and in GFP–neurons (n ¼ 5; 542
protrusions). (e) Histogram showing the proportion
of protrusions synapsing with MSB in GFP+
neurons at 30 d.p.i. (n ¼ 3 neurons; 224 spines),
60 d.p.i. (n ¼ 3; 161 spines), 180 d.p.i. (n ¼ 3;
267 spines), PND4 (n ¼ 4; 222 spines) and in
GFP–neurons (n ¼ 5; 522 spines). *P o 0.05,
***P o 0.001. Values are mean ± s.e.m.
4ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
ARTICLES

Page 5

DISCUSSION
In the present study we took advantage of
viral-mediated expression of GFP in dividing
cells and their progenies to identify neurons
born in the dentate gyrus of the hippocampus
and to analyze in situ their fine morphology
and connectivity. We observed that at 30
d.p.i., new neurons received axosomatic, axo-
dendriticandaxospinousinput.Dendriticprotrusiondensityincreased
until 180 d.p.i., and some of the axo-spinous afferents probably
originated from the entorhinal cortex, as revealed by tracing experi-
ments. This variety of synaptic input suggests that new neurons fully
integrate into the hippocampal circuitry, a finding that is consistent
with recent electrophysiological observations of both GABAergic and
glutamatergic input on new neurons4,20,21. Furthermore, the timing of
axospinous synapse formation is consistent with electrophysiological
findings showing that glutamatergic input on new neurons appears
between 14 and 28 d.p.i. and continues to mature thereafter8,20,21.
Together, these data indicate that, although the first glutamatergic
synapses appear during the third week after division, full maturity of
theexcitatory inputisonlyreachedafter60d.p.i.,anobservationthatis
supported by recent electrophysiological analyses22.
We observed that the size and the density of dendritic protrusions
made by GFP+neurons increase with time after viral injection, a
finding that has implications for the functional integration of new
neurons. Indeed, ultrastructural and electrophysiological reports indi-
cate that spine head size is correlated with the size of the PSD, the
numberofglutamatereceptors,thenumberofdockedvesiclesandthus
with the probability of release15,16,23,24. Therefore, at early develop-
mental stages (21–30 d.p.i.), new neurons may receive fewer and
weaker inputs than later in their development, which is consistent
with the involvement of silent synapses in the maturation of glutama-
tergic neurotransmission25.
Notably, the bulk of synaptogenesis is concomitant with a critical
period for the survival of newborn neurons. Indeed, we observed that
most dendritic protrusions are formed between 21 and 30 d.p.i.8
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
100
0
Cumulative proportion
of protrusions
100
200
Distance to bouton (nm)
300
500
1,000
1,500
Actual position
Displaced ~1 µm
Displaced ~5 µm
a
cb
**
**
** **
Figure 5 The tips of filopodia are preferentially
associated with axonal boutons. Protrusions
thinner than 0.25 mm (that is, filopodia) were
reconstructed in three dimensions, and the
immediate environment surrounding their tip
was analyzed. (a) Electron micrographs and
corresponding 3D reconstructions (on the right of
each micrograph) of two photoconverted GFP+
filopodia, whose tips are in close contact with
axonal boutons. Blue, axonal boutons; red, GFP–
spines; green, GFP+filopodia. Scale bar, 0.5 mm
and 0.1 mm in inset. (b) Each filopodium (green)
was artificially displaced along the axis of its
dendrite to a new position (red), approximately
1 mm, and then 5 mm away from the original
position, so as to place the filopodia into the
extracellular space. The distance between the tip
of the filopodia and the closest bouton was then
measured in both the actual position and the two
displaced positions. A total of 118 filopodia were
measured in n ¼ 3 groups of 2–5 neurons each.
(c) Data are reported in the histogram. **P o
0.01. Values are mean ± s.e.m.
30 d.p.i. 60 d.p.i.
180 d.p.i.
PND4
GFP–
***
*
Protrusions associated
with a synapsing bouton
(%)
a
c
>0.6<0.25
Protrusion tip diameter (µm)
0
0.25–0.6
GFP+ spine volume (µm3)
GFP– spine volume
(µm3)
40
80
Protrusion per 10 µm
b
0
123
10
20
MSB spine
SSB spine
SSB-associated filopodia
Bouton-associated filopodia
0
0.1
0.2
1
2
0.10.2 0.3
3
0.3
Figure 6 The connectivity of protrusions is related to their geometry.
(a) Histogram presenting the proportion of protrusions associated with
a bouton already synapsing with another neuron, in relation to the
diameter at their tip (n ¼ 12 neurons; 1,020 protrusions). *P o 0.05,
***P o 0.001. (b) Histogram showing the density of protrusions
separated into four categories as defined by ssEM: filopodia associated
with boutons devoid of synaptic contact (bouton-associated filopodia),
filopodia associated with boutons already involved in a synapse
(SSB-associated filopodia), spines synapsing with boutons devoid of
other synapse (SSB spine) and spines synapsing with boutons involved
in other synapses (MSB spine). (c) 3D volumetric analyses of 82 MSBs
showing the volume of GFP+spine heads and the volume of the GFP–
spine heads synapsing on the same axonal bouton. The three populations
of MSB circled in the histogram are schematized on the right. Red,
GFP–spines; green, GFP+spines; blue, axonal boutons. Values are
mean ± s.e.m.
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION5
ARTICLES

Page 6

(Fig. 1), and previous studies indicate that most newborn neurons die
during the first month aftercell division26. The possibility that survival
is input-dependent is supported by the observation that hippocampal
activityregulates new neuron
may be mediated by neurotransmitters20,28,29. In support of
this possibility, we recently showed that the survival of new neurons
is mediated by the expression of NMDA receptor, in a cell-specific
manner30. Taken together, these results raise the possibility that
the early events of synaptogenesis and the integration (or not) of
new neurons in the excitatory circuit may result in cell survival
(or death).
Contacting presynaptic partners is a crucial step in the functional
integration of newborn neurons in the network, and the mechanisms
controlling synaptogenesis have implications for the connectivity of
newborn neurons. Dendritic spines from new neurons may develop in
the absence of presynaptic partners and new axonal boutons may be
formed concomitantly with spine formation, as was observed
in vitro18,31,32. Alternatively, new neurons may synapse on preexisting
axonal boutons. Two observations support the latter possibility: first,
the observation that virtually all filopodia are preferentially associated
with boutons already synapsing on a dendritic spine supports the view
that new protrusions have a preference for preexisting boutons19, and
second, the high incidence of MSBs at 30 d.p.i. suggests that newly
formed spines preferentially synapse with preexisting boutons.
Filopodiaare very motile, and most ofthe dendritic protrusions that
are added on new neurons are either filopodia or small spines. Thus,
MSBsmayresultfromthecontactbetweenafilopodiaandapreexisting
synapse, and the subsequent transformation of this filopodium into a
dendritic spine. This hypothesis does not exclude other models for
MSB formation, such as the fusion of varicosities33, but nonetheless
suggests thatneighboringsynapsesinfluence filopodiaorientationand/
or extension, and that filopodia remain in proximity of synapse-
bearing axonal bouton(s). In support of this hypothesis, previous
reports have indicated that filopodia growth is regulated by synaptic
activity34–36and that filopodia extend toward sources of glutamate13,
effects that may be mediated by glutamate spillover from active
synapses37. Thus, our results suggest that initials contacts are prefer-
entially made with preexisting boutons, supporting the view that
synaptic activity may regulate the integration of newborn neurons.
The numberofMSBs contactedby a newbornneuronmay influence
its connectivity and may have profound functional implications.
Indeed, as MSBs synapse with at least two different neurons, they
may contribute to spreading the information into the network and
synchronizing the activity of the newborn neuron with the other
contacted neuron(s)38. Furthermore, MSBs may increase network
plasticity. Learning is believed to depend on synaptic remodeling1
that involves weight changes (modifications in the efficacies of existing
synapses) or wiring changes (modification of the wiring pattern
between neurons). MSBs that synapse with several neurons may
represent a substrate on which modifications of the relative efficacy
of synapses on the same axonal bouton may produce modifications in
the connectivity of the bouton, thereby combining weight and wiring
changes, a combination that is believed to increase storage capacity39.
Thus, by controlling the relative size of dendritic spines synapsing on
the same axonal bouton, MSBs may regulate the relative activity of
postsynaptic neurons. Furthermore, such ‘switchable’ synapses may
quickly modulate the synaptic integration of a newborn neuron into
the adult network. Although hypothetical, such a mechanism may
provide the network with increased computational power and flex-
ibility, which could account for the increased synaptic plasticity found
in new neurons40,41.
survival27,28,a processthat
Are MSBs stable over time? As neurons matured and spine density
increased, the proportion of spines synapsing on MSBs indeed
remained stable, whereas the total number of spines synapsing on
SSBs increased. There are two possible explanations for this finding:
either MSBs are stable, or their turnover is steady. The first possibility
implies that synapses formed during the first month after cell division
involve primarily MSBs, and synapses formed thereafter contacted
almost exclusively boutons devoid of other synaptic partners. This
possibility implies that the probability that a new protrusion synapses
with an MSB depends on the age of the neuron from which it extends,
and that boutons devoid of synaptic contact are created de novo18,42or
are readily available43. This possibility, however, is not supported by
our observations. Indeed, filopodia were present in all examined
neurons, regardless of their age, and more than 90% of them were
closelyapposedtoboutonsalreadybearingasynapse.Thus,iffilopodia
maturedintodendriticspines,MSBswouldbeformedeveninthemost
mature neurons. The second possibility is that MSBs are transformed
into SSBs. This possibility implies that MSB formation is independent
of neuronal age and that MSB turnover is proportional to synapse
formation. Indeed, several studies report that the occurrence of MSB is
transiently increased by manipulations inducing synaptogenesis, such
as long-term potentiation induction44, learning45and the formation of
ocular dominance columns in the visual cortex46. Furthermore, we
found that the geometry of the protrusions (that is, the size of the
protrusion tip) was a better predictor of their connectivity (synapsing
with or being associated with an MSB or an SSB) than the age of the
neuron (Fig. 6a). Also, in individual MSBs, the size of spines synapsing
with the same axonal bouton are related, which suggests that as new
protrusions enlarge, the spine synapsing with the same bouton may
decrease or even retract, an effect that may be mediated by steric
hindrance or by direct synaptic competition. Together, these results
support the possibility that, as their geometry matures from filopodia
to large spines, protrusions gradually modify their connectivity from
MSB to SSB. Further experiments using live-cell imaging are now
underway to test this possibility.
Taken together, these results indicate that MSBs may represent a
morphologicalcorrelateofsynaptogenesisintheadultbrain,whichmay
have a fundamental role in the integration of new neurons into the
preexistingcircuitry.Furthermore,ourfindingsshow that,althoughthe
first glutamatergic synapses are likely to appear during the first month
after cell division, full maturation of the connectivity of new neurons is
reached only between 60 and 180 days after cell division. These results
havefundamentalimplicationsforourunderstandingofthefunctionof
newborn neurons and the formation of neuronal networks.
METHODS
Viral vectors production. We used a retroviral vector based on the Moloney
murine leukemia virus (MoMulV) expressing GFP. A stable packaging cell line
was generated as described previously4. Supernatant containing the virus was
collected from clone 293gp/NIT-GFPc11 after transfection with pVSVG and
was concentrated as described previously4. Final virus titers were 5 ? 108viral
particles ml–1, as measured by G418-resistant colony formation on NIH 3T3
cells. High-titer VSV-pseudotyped lentiviral vector stocks were produced in
293T cells as previously described47. The mRFP1 protein was expressed under
the control of the CMV promoter. Expression titers, determined on HeLa cells
by fluorescence-activated cell sorting analysis, were 5 ? 109to 1 ? 1010
transducing units ml–1with an HIV-1 p24 concentration of 100 mg ml–1. For
live-cell imaging experiments, we used a MoMulV expressing GFP from the
CAG promoter, as described previously8.
Subjects and stereotaxic surgery. Six- to seven-week-old female C57Bl/6 mice
(Jackson Laboratories) were housed in cages with a running wheel to enhance
neuronal stem cell proliferation48. After 1 week, mice were anesthetized
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
6 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
ARTICLES

Page 7

(100 mg ketamine, 10 mg xylazine in 10 ml saline per gram) and 1.5 ml of
MoMulV was infused into the right dentate gyrus (spatial coordinates from
Bregma: anteroposterior ¼ –2 mm, lateral ¼ 1.5 mm, ventral ¼ 2 mm). In the
experiments described in Supplementary Figure 1, 1.5 ml of recombinant
MoMulV was injected into the right dentate gyrus at the coordinates described
above, together with 1 ml of lentivirus diluted to a titer of 5 ? 103. In the same
mice, 1.5 ml of the RFP lentivirus was injected in the right entorhinal cortex, at
the following coordinates: anteroposterior ¼ –4.25 mm from bregma, medio-
lateral ¼ 3.5 mm, ventral ¼ 4.2 mm. Three days after surgery, mice were
housed in standard cages and anesthetized 21, 30, 60 or 180 d after surgery.
As a control, we labeled neurons born at early postnatal stage by injecting
recombinant MoMulV in 4-day-old neonate pups (PND4). Brains were
subsequently processed for confocal or electron microscopy as described below.
The animal protocols were approved by the Salk Institutional Animal Care and
Use Committee.
Confocal microscopy. Sixteen mice were perfused with saline, followed by a
solution of phosphate-buffered saline (pH 7.4) containing 4% paraformaldehyde.
We mounted 100-mm-thick hippocampal sections and imaged dendrites with a
40 ? oil lens and a z-step of 0.5 mm using a Bio-Rad radiance 2100 confocal
microscope. We analyzed the distal portion of the dendritic tree, after the second
branching point. As a control, we injected the fluorescent tracer Lucifer yellow
into GFP–DGCs from the outermost layer of the granule cell layer of MoMulV
infected mice. To avoid possible confusion between GFP+cells and GFP–cells
injected with Lucifer yellow, GFP–cells were injected in consecutive sections.
Live-cell imaging. Acute brainslicesfromMoMulV-infectedmicewereprepared
as described previously8. After recovery in artificial cerebrospinal fluid (CSF)
solution for 30 min, the brain slices were transferred to either the closed-dish
(FCS2; Bioptechs, Butler) or open-dish temperature-controlled chamber (Delta
T4 Culture dish system; Bioptechs) and perfused continuously with oxygenated
artificial CSF solution at 30 1C. The system was mounted on an inverted Nikon
TE2000 microscope or an upright Olympus BX51 (Olympus) connected to the
Bio-Rad R2100 confocal system for time-lapse imaging. Images of dendritic
processes identified at the outer molecular layer were taken every 30 min for
2h,andfour timepointsat 30-minintervalsforthefirst90minwereselectedfor
quantitative analyses. The image files were subjected to five iterations of
deconvolution (AutoDeblur), and the image stacks were aligned with the
alignment function (Image-Pro). Maximum intensity projections were created
with Image-Pro for each time point. An image stack was then created for these
four time points and aligned manually with the dendritic shaft as reference. The
maximal diameter of the tip of protrusions was measured in projected images.
Selection of neurons for analysis. During the course of our study, we
compared GFP+DGCs at 21, 30, 60 and 180 d.p.i. and neurons infected in
4-day-old neonate pups (PND4) with GFP–DGCs that were injected with the
fluorophore Lucifer yellow. Although the age of GFP–neurons was not possible
to determine with our procedure, the relative scarcity of new neurons implies
that most of the GFP–neurons were more than 60 days old. BrdU studies
indicate that 1,000 new neurons are created every day49and that cell survival
one month after cell division is about 30% (ref. 26). Furthermore, GFP–cells
were selected in the outer granule cell layer, which includes less than 10% of all
new neurons26. We therefore estimated that fewer than 1 in 1,000 GFP–
neurons was between 30 and 60 days of age. Consistent with this estimation,
all GFP–neurons injected (n ¼ 15) had extensive dendritic branches and a
dendritic spine density greater than on GFP+neurons at 60 d.p.i.
Electron microscopy. We perfused 10 mice with saline, followed by a solution
of phosphate-buffered saline (pH 7.4) containing 4% paraformaldehyde and
0.2% glutaraldehyde, and cut slices at a thickness of 100 mm. We injected 1–2
cells per mouse under a fluorescence microscope with 5% aqueous Lucifer
yellow (Sigma). GFP–cells were injected in slices of infected brains analyzed at
30 and 60 d.p.i. To avoid confusion between GFP+and GFP–cells, we injected
one cell per slice. Slices were then incubated with 2.8 mM DAB and 6 mM
potassium cyanide and irradiated under conventional epifluorescence using a
75-W Hg to induce photoconversion of DAB into an electron-dense residue.
Slices were then postfixed overnight in a solution of 3% glutaraldehyde and
processed conventionally for electron microscopy. For ssEM, 45–200 serial
sections were cut at a thickness of 40 nm. Labeled dendritic segments were cut
longitudinally and imaged with a Megaview III camera mounted on a JEOL
100CXII electron microscope at a 19,000? magnification. Images were con-
trasted and stitched using Adobe Photoshop, aligned using the Align software
(provided by J. Fiala, Boston University) and stacks were analyzed using
Metamorph 6.2r. The small-fold procedure was used to determine section
thickness and an average value of 37.1 ± 4.5 mm was used for reconstruction. In
injected neurons, we observed that all of the protrusions were labeled, avoiding
ambiguity concerning the identity of unlabeled structures. Any protrusion or
contacted axonal bouton that extended beyond the limits of a series was
excluded from analysis because not all of their potential partners could be
confirmed. Mature synapses were identified on serial sections and defined by
the presence of the following features on at least one section: PSD, more than
four presynaptic vesicles within 100 nm of the presynaptic membrane and a
clearly defined synaptic cleft. Although the accumulation of photoconverted
DAB hindered the PSD on some sections, PSDs were usually identifiable on
consecutive sections. Potential synapses for which the PSD or any other
characteristic was not visible on any of the serial sections represented less than
1% of all synapses, and were discarded from statistical analyses. The maximal
thickness of the tip of protrusions was measured on serial sections using the
Trace software (developed by J. Fiala). For electron tomography, 2-mm-thick
sections were analyzed at a magnification of 5,000? using a JEOL JEM-4000EX
intermediate voltage microscope operated at 400 keV. The specimen was tilted
in 21 increments from –601 to +601, and electron micrographs were recorded at
each tilt. The 3D structure of labeled dendrites was then calculated from the
transmitted intensities using methods described previously50.
Statistics and data presentation. For all analyses, the observer was blind to the
identity of the sample. Data are presented as mean ± s.e.m., with n indicating
the number of neurons or protrusions analyzed, as stated. Statistical analyses on
data containing two groups were performed using the Student’s t-test.
Statistical analyses on data containing more than two groups were performed
using the one-way ANOVA test, followed by Bonferroni post hoc analyses, to
account for multiple comparisons.
Note: Supplementary information is available on the Nature Neuroscience website.
ACKNOWLEDGMENTS
We thank K. Harris and J. Fiala for 3D reconstruction software (http://
www.synapses.bu.edu), R. Tsien (University of California, San Diego) for the gift of
mRFP1, Y. Jones and J. Jepsen for technical assistance, J. Simon for artwork, and
T. Shikorski and M.L. Gage for comments on the manuscript. N.T. is supported by
the Human Science Frontier Program Organization and the Swiss National Science
Foundation, F.H.G. is supported by the US National Institutes of Health (NIH)
NS050217-02, and a research grant from the Picower and the Lookout
Foundations. Some of this work was conducted at the National Center for
Microscopy and Imaging Research supported by NIH RRP41-04050 to M.H.E.
AUTHOR CONTRIBUTIONS
N.T. conceived the study, performed the experiments, analyzed the data and
prepared the manuscript. E.M.T. helped with confocal microscopy and electron
tomography data collection and analyses. E.A.B. provided technical expertise
for intracellular injections. J.B.A. provided help with the filopodia study. C.Z.
performed the live imaging experiment. A.C. provided the lentivirus. H.v.P.
provided the Moloney virus construct and technical expertise. M.E.M.
contributed to the analyses of the tomography data. M.H.E. contributed to the
tomography experiment and provided support for all the electron microscopy
experiments. F.H.G. discussed the experiments and the data, revised the
manuscript and provided financial support.
COMPETING INTERESTS STATEMENT
The authors declare no competing financial interests.
Published online at http://www.nature.com/natureneuroscience
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Lamprecht, R. & LeDoux, J. Structural plasticity and memory. Nat. Rev. Neurosci. 5,
45–54 (2004).
2. Altman, J. & Das, G.D. Autoradiographic and histological evidence of postnatal
hippocampal neurogenesis in rats. J. Comp. Neurol. 124, 319–335 (1965).
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION7
ARTICLES

Page 8

3. Eriksson, P.S. et al. Neurogenesis in the adult human hippocampus. Nat. Med. 4,
1313–1317 (1998).
4. van Praag, H. et al. Functional neurogenesis in the adult hippocampus. Nature 415,
1030–1034 (2002).
5. Shors,T.J.etal.Neurogenesisintheadultisinvolvedintheformationoftracememories.
Nature 410, 372–376 (2001).
6. Saxe, M.D. et al. Ablation of hippocampal neurogenesis impairs contextual fear
conditioning and synaptic plasticity in the dentate gyrus. Proc. Natl. Acad. Sci. USA
103, 17501–17506 (2006).
7. Madsen, T.M., Kristjansen, P.E., Bolwig, T.G. & Wortwein, G. Arrested neuronal
proliferation and impaired hippocampal function following fractionated brain irradiation
in the adult rat. Neuroscience 119, 635–642 (2003).
8. Zhao,C.,Teng,E.M.,Summers,R.G.,Jr.,Ming,G.L.& Gage,F.H.Distinctmorphological
stages of dentate granule neuron maturation in the adult mouse hippocampus.
J. Neurosci. 26, 3–11 (2006).
9. Harris, K.M. & Kater, S.B. Dendritic spines: cellular specializations imparting
both stability and flexibility to synaptic function. Annu. Rev. Neurosci. 17, 341–371
(1994).
10.Petrak, L.J., Harris, K.M. & Kirov, S.A. Synaptogenesis on mature hippocampal
dendrites occurs via filopodia and immature spines during blocked synaptic transmis-
sion. J. Comp. Neurol. 484, 183–190 (2005).
11.Dunaevsky, A., Tashiro, A., Majewska, A., Mason, C. & Yuste, R. Developmental
regulation of spine motility in the mammalian central nervous system. Proc. Natl.
Acad. Sci. USA 96, 13438–13443 (1999).
12.Fischer, M., Kaech, S., Knutti, D. & Matus, A. Rapid actin-based plasticity in dendritic
spines. Neuron 20, 847–854 (1998).
13.Portera-Cailliau, C., Pan, D.T. & Yuste, R. Activity-regulated dynamic behavior of early
dendritic protrusions:evidence for differenttypes ofdendriticfilopodia.J.Neurosci.23,
7129–7142 (2003).
14.Harris,K.M.,Jensen,F.E.&Tsao,B.Three-dimensionalstructureofdendriticspinesand
synapses in rat hippocampus (CA1) at postnatal day 15 and adult ages: implications for
the maturation of synaptic physiology and long-term potentiation. J. Neurosci. 12,
2685–2705 (1992).
15.Matsuzaki, M. et al. Dendritic spine geometry is critical for AMPA receptor expression in
hippocampal CA1 pyramidal neurons. Nat. Neurosci. 4, 1086–1092 (2001).
16.Takumi, Y., Ramirez-Leon, V., Laake, P., Rinvik, E. & Ottersen, O.P. Different modes of
expression of AMPA and NMDA receptors in hippocampal synapses. Nat. Neurosci. 2,
618–624 (1999).
17.Fiala, J.C., Feinberg, M., Popov, V. & Harris, K.M. Synaptogenesis via dendritic filopodia
in developing hippocampal area CA1. J. Neurosci. 18, 8900–8911 (1998).
18.Ziv, N.E. & Smith, S.J. Evidence for a role of dendritic filopodia in synaptogenesis and
spine formation. Neuron 17, 91–102 (1996).
19.Knott, G.W., Holtmaat, A., Wilbrecht, L., Welker, E. & Svoboda, K. Spine growth
precedes synapse formation in the adult neocortex in vivo. Nat. Neurosci. 9, 1117–
1124 (2006).
20.Ge, S.etal.GABAregulates synapticintegrationofnewly generated neurons intheadult
brain. Nature 439, 589–593 (2006).
21.Esposito, M.S. et al. Neuronal differentiation in the adult hippocampus recapitulates
embryonic development. J. Neurosci. 25, 10074–10086 (2005).
22.Laplagne,D.A.etal.Functionalconvergenceofneuronsgeneratedinthedevelopingand
adult hippocampus. PLoS Biol. 4, e409 (2006).
23.Harris, K.M. & Stevens, J.K. Dendritic spines of CA 1 pyramidal cells in the rat
hippocampus: serial electron microscopy with reference to their biophysical character-
istics. J. Neurosci. 9, 2982–2997 (1989).
24.Schikorski, T. & Stevens, C.F. Quantitative ultrastructural analysis of hippocampal
excitatory synapses. J. Neurosci. 17, 5858–5867 (1997).
25.Wu,G.,Malinow,R.&Cline,H.T.Maturationofacentralglutamatergicsynapse.Science
274, 972–976 (1996).
26.Kempermann, G., Gast, D., Kronenberg, G., Yamaguchi, M. & Gage, F.H. Early
determination and long-term persistence of adult-generated new neurons in the
hippocampus of mice. Development 130, 391–399 (2003).
27.Gould, E., Beylin, A., Tanapat, P., Reeves, A. & Shors, T.J. Learning enhances adult
neurogenesis in the hippocampal formation. Nat. Neurosci. 2, 260–265 (1999).
28.Cameron, H.A., McEwen, B.S. & Gould, E. Regulation of adult neurogenesis by
excitatory input and NMDA receptor activation in the dentate gyrus. J. Neurosci. 15,
4687–4692 (1995).
29.Balazs, R., Jorgensen, O.S. & Hack, N. N-methyl-D-aspartate promotes the survival of
cerebellar granule cells in culture. Neuroscience 27, 437–451 (1988).
30.Tashiro, A., Sandler, V.M., Toni, N., Zhao, C. & Gage, F.H. NMDA-receptor-mediated,
cell-specific integration of new neurons in adult dentate gyrus. Nature 442, 929–933
(2006).
31.Okabe,S.,Miwa,A.&Okado,H.Spineformationandcorrelatedassemblyofpresynaptic
and postsynaptic molecules. J. Neurosci. 21, 6105–6114 (2001).
32.Majewska, A.K., Newton, J.R. & Sur, M. Remodeling of synaptic structure in sensory
cortical areas in vivo. J. Neurosci. 26, 3021–3029 (2006).
33.Shepherd, G.M., Raastad, M. & Andersen, P. General and variable features of varicosity
spacing along unmyelinated axons in the hippocampus and cerebellum. Proc. Natl.
Acad. Sci. USA 99, 6340–6345 (2002).
34.Trachtenberg, J.T. et al. Long-term in vivo imaging of experience-dependent synaptic
plasticity in adult cortex. Nature 420, 788–794 (2002).
35.Kirov, S.A. & Harris, K.M. Dendrites are more spiny on mature hippocampal neurons
when synapses are inactivated. Nat. Neurosci. 2, 878–883 (1999).
36.Richards, D.A. et al. Glutamate induces the rapid formation of spine head protrusions in
hippocampal slice cultures. Proc. Natl. Acad. Sci. USA 102, 6166–6171
(2005).
37.Kullmann, D.M. & Asztely, F. Extrasynaptic glutamate spillover in the hippocampus:
evidence and implications. Trends Neurosci. 21, 8–14 (1998).
38.Harris, K.M. How multiple-synapse boutons could preserve input specificity during an
interneuronal spread of LTP. Trends Neurosci. 18, 365–369 (1995).
39.Chklovskii, D.B., Mel, B.W. & Svoboda, K. Cortical rewiring and information storage.
Nature 431, 782–788 (2004).
40.Schmidt-Hieber, C., Jonas, P. & Bischofberger, J. Enhanced synaptic plasticity
in newly generated granule cells of the adult hippocampus. Nature 429, 184–187
(2004).
41.Jessberger, S.& Kempermann,G.Adult-born hippocampal neurons mature intoactivity-
dependent responsiveness. Eur. J. Neurosci. 18, 2707–2712 (2003).
42.De Paola, V. et al. Cell type-specific structural plasticity of axonal branches and boutons
in the adult neocortex. Neuron 49, 861–875 (2006).
43.Shepherd, G.M. & Harris, K.M. Three-dimensional structure and composition of CA3-
CA1 axons in rat hippocampal slices: implications for presynaptic connectivity and
compartmentalization. J. Neurosci. 18, 8300–8310 (1998).
44.Toni, N., Buchs, P.A., Nikonenko, I., Bron, C.R. & Muller, D. LTP promotes formation of
multiple spine synapses between a single axon terminal and a dendrite. Nature 402,
421–425 (1999).
45.Geinisman, Y., Berry, R.W., Disterhoft, J.F., Power, J.M. & Van der Zee, E.A. Associative
learningelicitstheformationofmultiple-synapseboutons.J.Neurosci.21,5568–5573
(2001).
46.Friedlander, M.J., Martin, K.A. & Wassenhove-McCarthy, D. Effects of monocular
visual deprivation on geniculocortical innervation of area 18 in cat. J. Neurosci. 11,
3268–3288 (1991).
47.Consiglio, A. et al. Robust in vivo gene transfer into adult mammalian neural stem cells
by lentiviral vectors. Proc. Natl. Acad. Sci. USA 101, 14835–14840 (2004).
48.van Praag, H., Christie, B.R., Sejnowski, T.J. & Gage, F.H. Running enhances neuro-
genesis, learning, and long-term potentiation in mice. Proc. Natl. Acad. Sci. USA 96,
13427–13431 (1999).
49.Kempermann, G., Kuhn, H.G. & Gage, F.H. Genetic influence on neurogenesis
in the dentate gyrus of adult mice. Proc. Natl. Acad. Sci. USA 94, 10409–10414
(1997).
50.Capani, F., Martone, M.E., Deerinck, T.J. & Ellisman, M.H. Selective localization of high
concentrations of F-actin in subpopulations of dendritic spines in rat central nervous
system: a three-dimensional electron microscopic study. J. Comp. Neurol. 435,
156–170 (2001).
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
8 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
ARTICLES

Page 9