Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
Cellular/Molecular
Cortactin-Binding Protein 2 Modulates the Mobility of
Cortactin and Regulates Dendritic Spine Formation and
Maintenance
Yi-Kai Chen
1,2
and Yi-Ping Hsueh
1,2
1
Molecular Cell Biology, Taiwan International Graduate Program, Institute of Molecular Biology, Academia Sinica, and Graduate Institute of Life Sciences,
National Defense Medical Center, and
2
Institute of Molecular Biology, Academia Sinica, Taipei, Taiwan 115, Republic of China
Dendritic spines, the actin-rich protrusions emerging from dendrites, are the locations of excitatory synapses in mammalian brains.
Many molecules that regulate actin dynamics also influence the morphology and/or density of dendritic spines. Since dendritic spines are
neuron-specific subcellular structures, neuron-specific proteins or signals are expected to control spinogenesis. In this report, we char-
acterize the distribution and function of neuron-predominant cortactin-binding protein 2 (CTTNBP2) in rodents. An analysis of an
Expressed Sequence Tag database revealed three splice variants of mouse CTTNBP2: short, long, and intron. Immunoblotting indicated
that the short form is the dominant CTTNBP2 variant in the brain. CTTNBP2 proteins were highly concentrated at dendritic spines in
cultured rat hippocampal neurons as well as in the mouse brain. Knockdown of CTTNBP2 in neurons reduced the density and size of
dendritic spines. Consistent with these morphological changes, the frequencies of miniature EPSCs in CTTNBP2 knockdown neurons
were lower than those in control neurons. Cortactin acts downstream of CTTNBP2 in spinogenesis, as the defects caused by CTTNBP2
knockdown were rescued by overexpression of cortactin but not expression of a CTTNBP2 mutant protein lacking the cortactin interac-
tion. Finally, immunofluorescence staining demonstrated that, unlike cortactin, CTTNBP2 stably resided at dendritic spines even after
glutamate stimulation. Fluorescence recovery after photobleaching further suggested that CTTNBP2 modulates the mobility of cortactin
in neurons. CTTNBP2 may thus help to immobilize cortactin in dendritic spines and control the density of dendritic spines.
Introduction
Dendritic spines, the major locations of excitatory synapses in
mammalian brains (Harris and Stevens, 1989), are actin-rich
structures (Fischer et al., 2000). Neuronal activity controls actin
cytoskeleton dynamics and thus modulates dendritic spine mor-
phology and remodeling (Matus, 2000; Pontrello and Ethell,
2009). Cortactin, an actin-binding protein (Wu and Parsons,
1993), promotes branching and stabilization of actin filaments
(for review, see Ammer and Weed, 2008; Ren et al., 2009) and is
highly enriched in the lamellipodia of motile cells (Wu and Par-
sons, 1993) and in the dendritic spines of neurons (Hering and
Sheng, 2003). Knockdown of endogenous cortactin reduces the
spine density in cultured hippocampal neurons (Hering and
Sheng, 2003), indicating a critical role for cortactin in spinogen-
esis. Moreover, NMDA receptor (NMDAR) activation triggers
the redistribution of cortactin from dendritic spines to the den-
dritic shaft (Hering and Sheng, 2003), suggesting that cortactin
participates in neuronal activity-dependent remodeling of den-
dritic spines.
Cortactin interacts with filamentous actin (F-actin) via cen-
tral actin-binding repeats (Weed et al., 2000) and binds to the
Arp2/3 complex with its N-terminal acidic domain (Weed et al.,
2000; Uruno et al., 2001; Weaver et al., 2001). These interactions
stabilize and promote the branching of actin filaments (Uruno et
al., 2001; Weaver et al., 2001). Cortactin thus contributes to the
morphological maintenance of dendritic spines (Hering and
Sheng, 2003). We wondered whether the activity of cortactin in
the regulation of dendritic spinogenesis, a neuron-specific event,
is controlled by a neuron-specific signal or interacting protein.
Several proteins containing a proline-rich sequence that interacts
with the C-terminal Src homology 3 (SH3) domain of cortactin
can direct cortactin’s participation in various cellular events (for
review, see Cosen-Binker and Kapus, 2006; Ammer and Weed,
2008). For instance, the SH3 domain of cortactin interacts with
postsynaptic density cortactin-binding protein 1 (CortBP1) (Du
et al., 1998), also known as Shank (Naisbitt et al., 1999). This
interaction bridges F-actin, the postsynaptic guanylate kinase-
associated protein–PSD-95–NMDAR complex (Naisbitt et al.,
1999), and the Homer–mGluR complex (Tu et al., 1999). In ad-
dition, cortactin-binding protein 2 (CTTNBP2), also known as
CortBP2 (Cheung et al., 2001) or CBP90 (Ohoka and Takai,
1998), was found to interact with the SH3 domain of cortactin
Received Aug. 29, 2011; revised Nov. 26, 2011; accepted Dec. 2, 2011.
Author contributions: Y.-K.C. and Y.-P.H. designed research; Y.-K.C. performed research; Y.-K.C. analyzed data;
Y.-K.C. and Y.-P.H. wrote the paper.
This work was supported by grants from Academia Sinica (AS-100-TP-B09 to Y.-P. H.) and the National Science
Council (NSC 99-2321-B-001-032 and NSC 100-2321-B-001-022 to Y.-P. H.). We thank Dr. Morgan Sheng for the
cortactin constructs; Dr. Roger Tsien for the mCherry plasmid; and the Electrophysiology and Calcium Imaging Core
Facility, Neuroscience Program and Institute of Molecular Biology, Academia Sinica, and Drs. Tzyy-Nan Huang and
Chiung-Ya Chen for technical support.
The authors declare no conflicting financial interests.
Correspondence should be addressed to Yi-Ping Hsueh, Institute of Molecular Biology, Academia Sinica, 128
Academia Road, Section 2, Nankang, Taipei, Taiwan 115, Republic of China. E-mail: yph@gate.sinica.edu.tw.
DOI:10.1523/JNEUROSCI.4405-11.2012
Copyright © 2012 the authors 0270-6474/12/321043-13$15.00/0
The Journal of Neuroscience, January 18, 2012 • 32(3):1043–1055 • 1043
(Ohoka and Takai, 1998). Although CTTNBP2 is exclusively ex-
pressed in the brain (Ohoka and Takai, 1998), its function in
neurons and in the regulation of cortactin activity is still com-
pletely unknown.
In the present study, we hypothesize that CTTNBP2 regulates
neuron-specific cortactin functions, such as dendritic spine for-
mation. The interaction between cortactin and CTTNBP2 in
neurons was confirmed by immunoprecipitation and immuno-
staining. Through a combination of RNAi knockdown, immu-
nostaining, and fluorescence recovery after photobleaching
(FRAP), our investigation suggests that CTTNBP2 regulates the
mobility of cortactin and thus controls the formation and main-
tenance of dendritic spines.
Materials and Methods
Antibodies and reagents. The following antibodies were used in this study:
rabbit polyclonal cortactin (H-191; Santa Cruz Biotechnology); mouse
monoclonal PSD-95 (K28/43; Millipore); mouse monoclonal Myc-tag
(9B11; Cell Signaling Technology); rabbit polyclonal green fluorescent
protein (GFP; A-6455, Invitrogen); chicken polyclonal GFP (ab13970;
Abcam); mouse vesicular GABA transporter (VGAT; 131 011, Synaptic
Systems); and mouse monoclonal
␣
-tubulin (B-5-1-2; Sigma-Aldrich).
CTTNBP2 polyclonal antibody was generated by immunizing rabbits
with glutathione S-transferase (GST)-CTTNBP2 (amino acids 498 –625)
recombinant protein. After passage through a GST-coupled affinity col-
umn, specific antibody was purified with a GST-CTTNBP2 (amino acids
498 – 625)-conjugated column. Alexa Fluor 546-conjugated phalloidin
was purchased from Invitrogen. Tetrodotoxin and bicuculline were pur-
chased from Tocris Bioscience.
DNA constructs. pGW1-CMV-Myc-cortactin and pGW1-CMV-GFP-
cortactin (Hering and Sheng, 2003) were gifts from Dr. Morgan Sheng at
the Howard Hughes Medical Institute and Massachusetts Institute of
Technology, Cambridge, MA (current affiliation: Genentech, South San
Francisco, CA). To construct the CTTNBP2 short-form expression plas-
mid, the 5⬘ (1–1413 bp) and 3⬘ (1393–1893 bp) regions of the coding
sequence (CDS) were amplified by PCR from RIKEN 4732477G22 and
IMAGE 6833056, respectively. The resulting overlapped DNA fragments
were mixed for an assembly PCR in which they primed on each other and
assembled into the CTTNBP2 short-form full-length CDS. Following
PCR with primers carrying additional KpnI and BglII restriction sites, the
full-length CDS was cloned into the vector pGW1–CMV with or without
a Myc tag.
For long-form CTTNBP2, the 3⬘ coding sequences were amplified
from mouse genomic DNA (1393–2038 bp) and RIKEN 6430526E05
(2020 – 4947 bp) and were assembled by PCR. For the intron form CTT-
NBP2, the 3⬘ coding sequences (1393–2133 bp) were amplified from
mouse genomic DNA. To generate the CTTNBP2 long-form and intron-
form expression constructs, the resulting 3⬘ coding sequences were indi-
vidually cloned into the short-form expression vector using NheI and
EcoRI to replace the 3⬘ segment of the short-form CDS.
For miRNA knockdown, the linearized vector pcDNA6.2–GW/
EmGFP-miR was purchased from Invitrogen. The nucleotide sequence
of base pairs 1362–1382 of the CTTNBP2 CDS was then used to construct
pcDNA6.2-GW/EmGFP-miR-BP2 (BP2-miR) according to the manu-
facturer’s instructions. Plasmid cDNA6.2-GW/EmGFP-neg control
(Ctrl-miR), which expresses a miRNA that was predicted to not target
any gene in mammalian genomes, was used as the negative control in the
knockdown experiments. Both miR-BP2 and Ctrl-miR coexpress
EmGFP, which was used to outline cell morphology. To obtain miRNA
vectors coexpressing mCherry, the miRNA cassettes in BP2-miR and
Ctrl-miR were individually amplified and cloned into the pGW1–CMV–
mCherry vector using EcoRI. Consequently, the miRNA cassettes were
integrated into the 3⬘ untranslated region of the mCherry transcripts.
The original pRSETB–mCherry construct was kindly provided by Pro-
fesspr Roger Tsien at the University of California, San Diego, CA. The
mCherry coding region was PCR amplified and subcloned into the vector
GW1–CMV.
For generating the CTTNBP2 constructs PA1 (P540A/P543A), PA2
(P599A/P602A), and a silent mutant resistant to the CTTNBP2 miRNA,
site-directed mutagenesis was performed with the following oligonucleo-
tides: PA1, 5⬘-CAGAGGAAATCCTGCTCCTATCGCTCCCAAAAAGC
CAG-3⬘; PA2, 5⬘-CTAAGTCGTCCTCCGCTCAGCTGGCACCAAAACC
GTCC-3⬘; silent mutant, 5⬘-GGGCAATGCAAATGATCCTGACCAAA
ATGGAAATAACACT-3⬘. The bases in italics indicate the mutated sites.
Reverse transcription-PCR. Total RNA was isolated from embryonic
day 14.5 mouse brain using Trizol (Invitrogen), followed by treatment
with DNase I (Sigma-Aldrich). Reverse transcription-PCR (RT-PCR)
was performed with the One-Step RT-PCR Kit (Genemark) according to
the manufacturer’s instructions. To discriminate between the various
splicing forms of CTTNBP2 (see Fig. 1 A), three oligonucleotide primers
weredesigned:A,5⬘-CCTCCCTCTACTTTGCCACA-3⬘;B,5⬘-GCCATCTTCG
CAGGAGTAAT-3⬘;C,5⬘-AAGAAATGAGGAAGTGGGTGAA-3⬘.
Animals. All animal experiments were performed with the approval of
the Academia Sinica Institutional Animal Care and Utilization Commit-
tee. For primary culture, pregnant rats were killed by CO
2
inhalation;
E18 –E19 fetal pups were then isolated and killed by decapitation. To
prepare brain extracts for biochemical study, adult rats of either sex were
killed by decapitation with a guillotine. For immunohistochemistry, 2- to
3-month-old mice of either sex were first anesthetized by intraperitoneal
injection with a mixture of ketamine (8.7 mg/100 g of body weight) and
xylazine (1.3 mg/100 g of body weight) and intracardiacally perfused
with ⬎50 ml of PBS containing heparin (10 U/ml) followed by ⬎50 ml of
4% paraformaldehyde in PBS to fix brain tissue. After dissection from
skulls, brains were post-fixed in 4% paraformaldehyde overnight at 4°C
for slicing at a later time.
Immunoprecipitation. To obtain a soluble synaptosome fraction, the
brains of 2- to 4-month-old rats of either sex were homogenized in lysis
buffer (10 m
M Tris, pH 7.4, 320 mM sucrose, 2 mM dithiothreitol, 2
g/ml
leupeptin, 2
g/ml pepstatin-A, 2
g/ml aprotinin, 1 mM tosylphenyla-
lanylchloromethane, and 2 m
M phenylmethylsulfonyl fluoride) and cen-
trifuged at 800 ⫻ g for 10 min at 4°C. The supernatant was centrifuged at
9200 ⫻ g for 15 min to collect the synaptosomal fraction (P2), which was
then resuspended and incubated in 20 m
M Tris, pH 7.4, 5 mM ethylene-
diaminetetraacetic acid, 200 m
M NaCl, and 1% Triton X-100 for1hat
4°C. After centrifugation at 35,000 ⫻ g for 40 min, the supernatant was
collected for further experiments. For immunoprecipitation, the soluble
synaptosome fraction was diluted fivefold in 20 m
M Tris buffer and in-
cubated with antibody preadsorbed Sepharose resin at 4°C overnight.
The resin was washed three times with 20 m
M Tris, pH 7.4, 40 mM NaCl,
and 0.2% Triton X-100, and immunoblotted.
Primary rat hippocampal neuron cultures and immunofluorescence. At
embryonic day 18 –19, rat hippocampal neurons were dissociated by
trypsinization, resuspended in growth medium (50% Neurobasal Me-
dium and 50% DMEM supplemented with 2% B27 supplement, 0.5 m
M
glutamine, and 12.5
M glutamate), and plated in 12-well culture plates
containing glass coverslips coated with poly-
L-lysine (1 mg/ml) at a den-
sity of 200,000 neurons per well. Transfection was performed at 12 DIV
using calcium phosphate precipitation. For immunofluorescence, cells
were fixed with 4% paraformaldehyde and 4% sucrose in PBS, followed
by permeabilization with 0.2% Triton X-100 in PBS at 18 DIV. To ex-
plore the role of CTTNBP2 in the maintenance of dendritic spines, trans-
fection was performed at 20 DIV and immunostaining was performed at
26 DIV. After blocking with 10% bovine serum albumin, cells were in-
cubated with primary antibodies diluted in PBS containing 3% bovine
serum albumin at 4°C overnight. Following PBS washes, the cells were
incubated with secondary antibodies conjugated with Alexa Fluor 488,
555, and/or 647 (Invitrogen) for 2 h. DNA was counterstained with
4⬘,6⬘-diamidino-2-phenylindole dihydrochloride (DAPI). Vectashield
mounting medium (H-1000; Vector Laboratories) was used to mount
the samples for imaging. Images were acquired using a confocal micro-
scope (LSM510-Meta or LSM700; Carl Zeiss) equipped with a 63⫻/NA
1.4 oil (Plan-Apochromat; Carl Zeiss) objective lens and LSM 3.2 or Zen
2009 (Carl Zeiss) acquisition and analysis software. All fixed cells were
imaged at 20–22°C. For publication, the images were processed with
Photoshop (Adobe) with minimal adjustment of brightness or contrast
applied to the whole images. Quantitation of spine morphology and
1044 • J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
density was performed using ImageJ 1.45 (NIH). Density and size were
manually quantitated along 20
m of dendrite starting 20
m away from
the soma. Some experiments were repeated blind to minimize the effect
of bias. Statistical analysis of spine density was performed with unpaired
Student’s t tests (see Fig. 4 D) or with one-way ANOVA and Tukey’s post
hoc test (see Fig. 5C) using GraphPad Prism 5.0 (GraphPad Software).
Spine width and length were analyzed with the Kolmogorov–Smirnov
test using SPSS 10.0 (SPSS).
Immunohistochemistry. The 50-
m-thick adult mouse brain sections
were collected with a vibratome and incubated with CTTNBP2 antibody
at 1
g/ml in PBS containing 3% horse serum, 2% bovine serum albu-
min, and 0.3% Triton X-100 at 4°C for 2 d. After washing, the brain
sections were incubated with Alexa Fluor 488-conjugated secondary an-
tibody, Alexa Fluor 546-conjugated phalloidin, and DAPI at room tem-
perature for 2 h. Images were acquired as described above.
Time-lapse recording and FRAP. Time-lapse recording was performed
using a confocal microscope (LSM700; Carl Zeiss) with a 63⫻/1.4 oil
objective (Plan-Apochromat; Carl Zeiss) at 37°C supplied with 5% CO
2
.
One day after transfection, COS cells were trypsinized and replated on
poly-L-lysine (0.1 mg/ml)-coated glass coverslips, followed bya4hin-
cubation at 37°C. Before recording, culture medium was replaced with
prewarmed HBSS (Invitrogen). The GFP and mCherry signals were ac-
quired every 5 s for 5–10 min.
For FRAP, hippocampal neurons grown on poly-
L-lysine (1 mg/ml)-
coated glass coverslips were transfected at 12 DIV and recorded at 18
DIV. During FRAP, the image series was captured before and immedi-
ately after photobleaching with 0.5 s intervals and a scan speed of ⬃0.2
s/scan. Spines of interest were photobleached 10 times with a 488 nm
laser at 100% output. ImageJ was used to measure fluorescence intensity
in the image series. For each image, a region without transfected cells was
measured as background, and its intensity was subtracted from the in-
tensity of the region of interest. These raw data were divided by the
intensity measured at unbleached dendrites to correct for fluorescence
loss during image acquisition. Finally, all intensity data were normalized
to the average fluorescence intensity of 10 scans acquired just before
bleaching. GraphPad Prism was used for curve fitting with the following
one-phase exponential equation: y ⫽ a(1 ⫺ exp(⫺bx)).
Electrophysiology. Cultured rat hippocampal neurons were transfected
at 12 DIV, and whole-cell patch-clamps were performed at 18 DIV to
record miniature EPSCs (mEPSCs). Neurons were incubated in extracel-
lular solution containing 145 m
M NaCl, 3 mM KCl, 10 mM HEPES, 3 mM
CaCl
2
,2mM MgCl
2
,8mM glucose, 0.001 mM tetrodotoxin, and 0.02 mM
bicuculline. The intracellular solution contained 136.5 mM K-gluconate,
9m
M NaCl, 17.5 mM KCl, 10 mM HEPES, 0.2 mM ethylene glycol tet-
raacetic acid, 4 m
M Mg-ATP, and 0.3 mM Na-ATP. Neurons were
voltage-clamped at ⫺70 mV, and mEPSCs were recorded with the Axon
Axopatch 200B amplifier (Molecular Devices) and filtered at 1 kHz. The
Clampfit 9 software (Molecular Devices) was used to detect mEPSCs
from the raw data with an amplitude threshold of 4.5 pA. The mEPSC
recording was performed blind at the Electrophysiology and Calcium
Imaging Core Facility, Neuroscience Program and Institute of Molecular
Biology, Academia Sinica. Statistical analyses of amplitude and frequency
were then performed with unpaired Student’s t tests using GraphPad
Prism.
Results
The CTTNBP2 short form is the major gene transcript in
the brain
Rat CTTNBP2 was originally identified as a protein product
smaller than 90 kDa from a pulldown assay with a GST-
cortactin fusion protein (Ohoka and Takai, 1998). Later, hu-
man CTTNBP2 was predicted to encode a 1663 aa product
(Cheung et al., 2001). Through analysis of an Expressed Se-
quence Tag database, we identified three transcripts encoded
by CTTNBP2: short, long, and intron forms (Table 1). This
variation is due to alternative splicing between exon 4 and exon
5ofCTTNBP2 (Fig. 1A). Two RNA splice donor sites are present
at the end of exon 4 of CTTNBP2. When the first splice site is
used, the transcript encodes the short form of CTTNBP2. When
the second splice site is chosen, the long protein is produced.
Retention of intron 4 causes the production of the intron form;
the polypeptide chain terminates early at the alternative stop
codon in intron 4. Thus, the C-terminal amino acid sequences of
these three forms are variant (Fig. 1B).
To investigate expression of these three forms in neurons, we
first performed RT-PCR using RNA extracted from mouse brain,
which suggested that the short-form transcripts were the pre-
dominant products of CTTNBP2 in the mouse brain (Fig. 1C).
To confirm this observation, we generated CTTNBP2-specific
antibodies to analyze the protein products of CTTNBP2 (Fig.
1D); only the fragment containing the proline-rich domain suc-
cessfully generated CTTNBP2-specific antibodies (Fig. 1D).
Compared with cell extracts prepared from COS cells transfected
with the short, long, and intron forms individually, the short
form appeared to be the dominant protein product of CTTNBP2
in the brain (Fig. 1E). We therefore only included the short form
in our subsequent investigations.
CTTNBP2 interacts with cortactin in COS cells and
in neurons
A previous study demonstrated the interaction between cor-
tactin and CTTNBP2 by a fusion protein pulldown assay
(Ohoka and Takai, 1998). To confirm the interaction of cor-
tactin and CTTNBP2 in cells, we transfected GFP-cortactin
and mCherry-CTTNBP2 into COS cells; 1 d after transfection,
cells were replated and analyzed by time-lapse recording ⬃4h
after replating. Under these conditions, high cell mobility facil-
itates the observation of cytoskeleton dynamics. The live imaging
clearly showed the colocalization and comigration of cortactin
and CTTNBP2 at the cell cortex and intracellular puncta of COS
cells (Fig. 2 A,B). Some CTTNBP2/cortactin puncta associated
with intracellular vesicles [Fig. 2A(inset),B]. Additionally, cor-
tactin antibodies precipitated cortactin as well as CTTNBP2 from
rat brain extracts (Fig. 2C), further supporting the interaction of
CTTNBP2 and cortactin in neurons.
Cortactin interacts with F-actin via its central repeat do-
mains and binds to CTTNBP2 through its C-terminal SH3
domain. CTTNBP2 therefore likely associates with F-actin
through the interaction with cortactin. To test this possibility,
COS cells were transfected with Myc-tagged CTTNBP2 and
immunostained 1 d after transfection using an anti-Myc anti-
body and phalloidin. In fixed cells, F-actin and CTTNBP2
were colocalized at the cell cortex (Fig. 2D, arrowheads), re-
flecting the association of CTTNBP2 with the cortactin-F-
actin cytoskeletons.
Table 1. Mouse CTTNBP2 expressed sequence tag clones
GenBank
accession no. Clone ID Source tissue
Containing
full exon 4
Containing
intron
Splicing
form
BC141407 IMAGE 9056020 Brain Yes No Long
AK173254 Pancreatic islet, adult Yes No Long
BC068156 IMAGE 30362957 Brain No No Short
BQ769661 IMAGE 5697958 Brain, E12.5 No No Short
CB526439 IMAGE 6848778 Brain, embryo No No Short
AK032356 RIKEN 6430526E05 Olfactory bulb, adult No No Short
CB244938 IMAGE 6833056 Brain, embryonic No No Short
AK028980 RIKEN 4732477G22 Skin, 10 d neonate Yes Yes Intron
BQ961104 IMAGE 6439978 Mammary gland
tumor, 5-month-
old female
Yes Yes Intron
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1045
CTTNBP2 is highly concentrated at synapses in cultured
neurons and brains
Since cortactin is concentrated at the dendritic spines and regu-
lates spine morphology, we wondered whether CTTNBP2 also
localized to dendritic spines in neurons. Myc-tagged CTTNBP2
and GFP were coexpressed in neurons, and immunostaining in-
dicated that Myc-tagged CTTNBP2 was highly concentrated at
dendritic spines (Fig. 3A). CTTNBP2-specific antibody was then
used to examine the distribution of endogenous CTTNBP2. In
Myc-tagged CTTNBP2-transfected COS cells, the immuno-
reactivities of the CTTNBP2 antibody colocalized well with Myc-
tag immunoreactivities (Fig. 3B), supporting the specificity of
our CTTNBP2 antibody. Triple staining with anti-CTTNBP2,
postsynaptic marker PSD-95 antibody, and phalloidin in cul-
tured hippocampal neurons at 23 DIV demonstrated that the
CTTNBP2 immunoreactivities colocalized very well with those of
PSD-95 and F-actin in mature cultured neurons (Fig. 3C). In
adult mouse brains, CTTNBP2 immunoreactivities were also co-
localized with F-actin (Fig. 3D), and higher magnification re-
vealed the punctate patterns of CTTNBP2 and F-actin, which
colocalized along dendrites (Fig. 3E). The CTTNBP2/F-actin
double-positive puncta did not colocalize with the inhibitory
synapse marker VGAT (Fig. 3E, enlarged images), suggesting that
the CTTNBP2/F-actin puncta are not the location of inhibitory
synapses. In addition to dendritic spines, CTTNBP2 antibody
also accumulated in a punctate signal in the soma; however, these
somatic CTTNBP2 puncta were not colocalized with F-actin (Fig.
3E). Together, these immunofluorescence experiments identified
the synaptic distribution of CTTNBP2 in cultured neurons and
mouse brains and suggested an association of CTTNBP2 with
actin cytoskeletons.
Knockdown of CTTNBP2 reduces the density and size of
dendritic spines
To explore the role of CTTNBP2 in dendritic spine morphogenesis,
we generated an artificial CTTNBP2 miRNA construct (BP2-miR)
to knock down CTTNBP2 expression in cells. Nonsilencing Ctrl-
miR predicted not to target any gene in mammalian genomes was
used as a negative control. BP2-miR reduced the expression of
cotransfected Myc-tagged CTTNBP2 in COS cells (Fig. 4 A). We
next examined the effect of BP2-miR in cultured hippocampal neu-
rons; since our miRNA constructs coexpressed EmGFP, the EmGFP
signals labeled transfected neurons and outlined cell morphology.
Immunostaining with GFP and CTTNBP2 antibodies demon-
strated that the CTTNBP2 immunoreactivities were lower in BP2-
miR-transfected neurons than in Ctrl-miR-transfected neurons or
neighboring untransfected neurons (Fig. 4B). High-magnification
images revealed that BP2-miR expression impaired dendritic spine
morphology (Fig. 4C). The spine density of CTTNBP2-knockdown
neurons (5.1 ⫾ 0.3 spines/10
m) was significantly lower than that
Figure 1. Expression ofCTTNBP2 in thebrain. A, Schematicof genomic structure and splicing forms of CTTNBP2 transcripts. Arrowheads pointto the firstand second splicedonor sites at the end
of exon 4. Arrows denote the positions of RT-PCR primers a– c. Primer a hybridizes before the first splicing site, primer b binds in exon 5, and primer c corresponds to the sequence in intron 4. B,
Alignment of the predicted C-terminal amino acid sequences of the long, short, and intron forms of CTTNBP2. The corresponding alternative splice sites are indicated by arrowheads. Amino acid
residues in the gray box are encoded by exon 5. Residues in the black box are encoded by intron 4. For the long and short forms, the amino acid sequence encodedby exon5 resultsfrom aframeshift
caused byusage ofa different spice site. Forthe long form, the amino acid sequencebeyond residue 715 is omitted. C, RT-PCRusing mRNA purified from E14.5 mouse brain.D, Domain structures of
thethree CTTNBP2forms. Thelengthsof the threeformsare indicated,as istheregion (498⬃625 aa) usedasthe immunogenfor theproduction ofspecificantibodies. CC,Coiled-coil; P,proline-rich;
Ank, ankyrin repeat. E, Adult rat brain extract and whole-cell lysates prepared from COS cells transfected with CTTNBP2 isoforms were immunoblotted for CTTNBP2, suggesting that the shortform
predominates in the brain.
1046 • J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
of neurons transfected with Ctrl-miR (7.9 ⫾ 0.3 spines/10
m) (Fig.
4D). The width of the spine heads was also reduced in CTTNBP2-
knock-down neurons (mean widths: BP2-miR, 0.55
m; Ctrl-miR,
0.68
m). However, the lengths of the dendritic spines were not
affected by CTTBP2 knockdown (Fig. 4D). The same conclusions
were also obtained in blind experiments (data not shown).
To corroborate these morphological changes in the dendritic
spines, we conducted blind experiments to further measure the
frequency and amplitude of mEPSCs in BP2-miR-transfected
neurons. Consistent with the change in dendritic spine density,
the mEPSC frequency of BP2-miR-transfected neurons was
lower than that of control neurons (Fig. 4E,F ); however, the
amplitude of the mEPSC did not differ significantly between
CTTNBP2 knock-down neurons and control neurons (Fig.
4E, G). Knockdown of CTTNBP2 in neurons thus impairs the
density and morphology of dendritic spines and the electrophys-
iological response of neurons.
In typical cultured hippocampal neurons, dendritic filop-
odia actively emerge from dendrites at ⬃12–14 DIV and then
transform to dendritic spines (Ziv and Smith, 1996; Chao et
al., 2008). Therefore, the experiments performed during DIV
12–18 favor a role of CTTNBP2 in spinogenesis. In fully ma-
ture cultured hippocampal neurons, dendritic spines are rela-
tively stable along dendrites (Ziv and Smith, 1996). Newborn
spines are difficult to detect in cultured hippocampal neurons at
⬃3 weeks in vitro (H.-W. Chao, unpublished data). To investi-
Figure2. CTTNBP2interacts with cortactin.A, Colocalization offluorescence protein-tagged CTTNBP2and cortactinin living cells.GFP-cortactin andmCherry-CTTNBP2 (mCherry-BP2) transiently
cotransfected COS cellswere replatedon glasscoverslips 4h before recording.Images wererecorded every5sfor⬃10 min.A representativeimage ata single timepoint isshown. Theinsets arethe
local enlargement of the area indicated by the arrow and illustratethe colocalization ofCTTNBP2 and cortactinsurrounding an intracellularvesicle. B, Enlargedtime-lapse images correspondingto
the regionindicated bythe arrowin A.C, Coimmunoprecipitation of cortactin and CTTNBP2 from rat brainwith anti-cortactinand nonimmunerabbit IgG.The precipitates were immunoblotted (IB)
with CTTNBP2 or cortactin antibodies as indicated. The arrowhead indicates the position of CTTNBP2 (left) or cortactin (right). D, Overlapping distribution of CTTNBP2 and F-actin at the cell cortex
(arrowheads). COScells expressingMyc-tagged CTTNBP2(Myc-BP2) were fixed and stained with anti-Myc, DAPI (tolabel nuclei),and phalloidin(to label F-actin). Scale bars: A,10
m; B,2
m; D,
20
m.
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1047
gate whether CTTNBP2 influences the maintenance of estab-
lished dendritic spines in fully mature neurons, transfection was
performed at 20 DIV and immunostaining was performed at 26
DIV. Similar to the results collected at 18 DIV, knockdown of
CTTNBP2 starting at 20 DIV also reduced the spine density at 26
DIV (Fig. 4H), although the widths and lengths of the dendritic
spines were not affected by CTTNBP2 knockdown (Fig. 4H).
These data suggest that CTTNBP2 is likely also involved in the
maintenance of dendritic spines.
The interaction of CTTNBP2 and cortactin is required for
regulation of spine density by CTTNBP2
Since CTTNBP2 interacts with cortactin in neurons and since cortactin
also regulates spine morphology and density, we wished to explore the
Figure 3. Synaptic distribution of CTTNBP2 in neurons. A, Rat hippocampal neurons were transfected with Myc-tagged CTTNBP2 (Myc-BP2) and GFP at DIV 12 and immunostained with Myc
antibody at DIV 18. Bottom panels show higher magnification of dendrites. B, Specificity of CTTNBP2 antibody in immunostaining. Myc-tagged CTTNBP2-transfected COS cells were analyzed by
double immunostaining with Myc antibody and CTTNBP2 antibody. C, Synaptic distribution of CTTNBP2 (BP2) in rat hippocampal neurons. Fluorescence staining using PSD-95 and CTTNBP2
antibodies and phalloidin was performed at DIV 23. D, E, Immunohistochemistry of adult mouse hippocampus with CTTNBP2 antibody. D, Low-magnification image covering a part of CA1 and the
dentate gyrus of the hippocampus. E, High-magnification image of the dentate gyrus. CTTNBP2 and VGAT antibodies were used for fluorescence immunostaining. F-actin and nuclei were labeled
with phalloidinand DAPI,respectively. Theenlarged individualimages arealso shown in the bottom panel. VGAT immunoreactivity isnot obviouslyoverlapping oradjacent tothe CTTNBP2puncta.
O, Stratum oriens; R, stratum radiatum; L, stratum lacunosum; M, stratum moleculare; DG, dentate gyrus; H, hilus. In C, yellow arrowheads in the individual panels and white arrowheads in the
merged panelshighlight examples ofcolocalization. In E, arrowheads indicate the colocalization of CTTNBP2 and F-actin,while arrows denotethe positions of VGAT alone. Scale bars: A–C,20
m;
D, 200
m; E,30
m.
1048 • J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
Figure 4. Knockdown of CTTNBP2 impairs dendritic spine density and reduces spontaneous neuronal activity. A, Knockdown of CTTNBP2 assayed in COS cells cotransfected with
BP2-miR orCtrl-miR and Myc-tagged wild-type CTTNBP2(Myc-BP2) or a CTTNBP2 silentmutant resistant toBP2-miR (Myc-BP2-resc). Whole-cell extracts wereimmunoblotted with Myc
antibody and tubulin antibody. B–F, The effect of expression of the BP2-miR in cultured neurons. Rat hippocampal neurons were transfected at 12 DIV with BP2-miR or Ctrl-miR. GFP
expressed by the miRNA vectors highlights the transfected cells. Neurons were harvested at DIV 18 for analysis. B, BP2-miR reduces the endogenous CTTNBP2 protein level in rat
hippocampal neurons. Transfected neurons were fixed and immunostained with CTTNBP2 antibody. C, CTTNBP2 knockdown decreases spine density and reduces spine width. Repre-
sentative images of GFP signal are shown. D, Quantification of the number of protrusions per 10
m of dendrites (left) and width and length of protrusions (right two panels). A total of
20 neurons were collected from two independent experiments for each group; ⬎50 dendrites and 400 spines for each group were assayed. E, mEPSCs were recorded on transfected
hippocampal neurons. F, G, Quantification of mEPSC frequency (F ) and amplitude (G) in transfected neurons (n ⫽ 23 for Ctrl-miR and n ⫽ 27 for BP2-miR). H, CTTNBP2 participates in
the in maintenanceof dendritic spines. Culturedrat hippocampal neurons were transfected with Ctrl-miR and BP2-miR at 20 DIV and were fixed for staining at 26 DIV. Error bars indicate
mean ⫾ SEM. *p ⬍ 0.05; **p ⬍ 0.01; ***p ⬍ 0.005. Scale bars: B,20
m; C,2
m.
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1049
relationship between CTTNBP2 and cortactin
in spinogenesis. A previous study had identi-
fied the proline-rich domain ofCTTNBP2as a
domain that interacted with the cortactin SH3
domain (Ohoka and Takai, 1998). We further
identified two candidate SH3-binding do-
mains in CTTNBP2: amino acids 538 –547
(-NPPPIPPKKP-) and amino acids 597-
606 (-SSPQLPPKPS-). Two proline-to-
alanine mutations were introduced into
each stretch to identify the region critical
for cortactin binding. These mutants were
designated PA1 for the P540A/P543A
double mutant and PA2 for the P599A/
P602A double mutant. Compared with
wild-type CTTNBP2, both the PA1 and
PA2 mutations reduced the interaction
between cortactin and CTTNBP2 in COS
cells; the PA1 mutation almost completely
abolished the interaction (Fig. 5A).
Since the PA1 mutation more strongly
interfered with the interaction between cor-
tactin and CTTNBP2, we used this mutant
to examine the role of the interaction of cor-
tactin and CTTNBP2 in dendritic spinogen-
esis. In CTTNBP2 knock-down neurons,
expression of the silent mutant resistant to
BP2-miR (BP2-resc) rescued the effect of
BP2-miR on spine density (Fig. 5B,C).
However, the PA1 mutant also resistant to
BP2-miR (BP2-PA1-resc) did not rescue the
spine density phenotype of the CTTNBP2
knockdown (Fig. 5B,C), suggesting that the
interaction with cortactin is required for
CTTNBP2 to control spine density. Consis-
tent with this possibility, cortactin overex-
pression also rescued the phenotype of the
CTTNBP2 knockdown (Fig. 5B,C). To-
gether, these observations suggest that cor-
tactin functions downstream of CTTNBP2
to regulate dendritic spine density.
Dendritic spine length was unaffected by
treatment with BP2-resc, BP2-PA1-resc, or
cortactin (Fig. 5D, right), consistent with
our conclusion that CTTNBP2 does not
regulate spine length. For the widths of the spine heads, cortactin
overexpression rescued the spine-head phenotype of the CTTNBP2
knockdown (Fig. 5B,D). Unexpectedly, although the BP2-PA1-resc
mutant did not rescue the effect of BP2-miR on spine density (Fig.
5C), it restored the spine-head phenotype of the CTTNBP2 knock-
down (Fig. 5B,D), suggesting that there are two pathways down-
stream of CTTNBP2 to control the size of spine heads, one cortactin
dependent, the other cortactin independent. Activation of either
pathway is sufficient to maintain the size of spine heads. Alternatively,
since the effect of BP2-miR on the spine widths is weaker than that on
the spine density, it is also possible that the spine width phenotype is
easier to rescue with BP2-PA1-resc, although the interaction between
cortactin and BP2-PA1-resc is reduced to a very low level.
Cortactin, but not CTTNBP2, redistributes into the dendritic
shaft after glutamate treatment
A previous study showed that glutamate treatment induces the
redistribution of cortactin and actin from dendritic spines to the
dendritic shaft (Hering and Sheng, 2003), which may contribute
to the regulation of activity-dependent remodeling of dendritic
spines. Since our data suggest that cortactin acts downstream of
CTTNBP2 to control spine density, we investigated whether
CTTNBP2 also redistributed to the dendritic shaft after gluta-
mate stimulation. Two treatments were used to stimulate cul-
tured hippocampal neurons at ⬃21–24 DIV: 50
M glutamate
treatment for 2 min (plus 8 min recovery) or 15 min.
Similar to the previous report (Hering and Sheng, 2003), F-actin
and cortactin were redistributed to the dendrites after glutamate
treatment in both conditions (Fig. 6A,B, data not shown for the
distribution of cortactin after 15 min glutamate treatment). In con-
trast, CTTNBP2 remained at the dendritic spines after glutamate
stimulation (Fig. 6 A,B). As previously (Hering and Sheng, 2003),
glutamate-induced F-actin and cortactin redistribution relied on
NMDAR but not AMPAR, since the effect of NMDA on F-actin and
cortactin is similar to that of glutamate (Fig. 6, compare C, A). KCl
stimulation also induced redistribution of F-actin and cortactin (Fig.
Figure5. Cortactinfunctionsdownstream of CTTNBP2in theregulation of dendriticspine density. A,Coimmunoprecipitation of
cortactinandCTTNBP2 (BP2) mutants.Whole-cell extracts ofCOS cells transfectedwith cortactin andwild-type or mutantCTTNBP2
(PA1, P540A/P543A; PA2, P599A/P602A) were precipitated with CTTNBP2 antibody. Immunoblotting (IB) was then performed to
assess the presence of cortactin and CTTNBP2 in the precipitates. The arrowhead indicates the position of cortactin (top) or
CTTNBP2 (bottom). The asterisk indicates a nonspecific signal. B–D, Rat hippocampal neurons were transfected with control
miRNA (Ctrl-miR) or CTTNBP2 miRNA (BP2-miR) along with the Myc-tagged CTTNBP2 silent mutant (BP2-resc), the PA1 mutant
(BP2-PA1-resc),or cortactin at12 DIV.Neurons wereharvested forstaining withGFP andMyc-tag antibodies atDIV 18.B, Cortactin
and the CTTNBP2 silent mutant restore the spine density in CTTNBP2 knock-down neurons. GFP signals were used to outline
dendrite and spine morphology. Scale bar, 2
m. C, Quantification of the density of protrusions. Eighteen neurons and ⬎50
dendriteswereassayed for each group.D, The widthand length of protrusionsassessed from ⬎400 spinesin each group.Error bars
indicate mean ⫾ SEM. *p ⬍ 0.05; ***p ⬍ 0.005.
1050 • J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
6C); in contrast, regardless of KCl, NMDA, or AMPA treatment,
CTTNBP2 localized to the dendritic spines (Fig. 6C). Together, these
observations suggest that, unlike cortactin and F-actin, the synaptic
distribution of CTTNBP2 does not shift to the dendritic shaft in
response to neuronal activation.
CTTNBP2 regulates the mobility of cortactin in
dendritic spines
We observed that CTTNBP2 stably resides at dendritic spines and
that CTTNBP2 acts upstream of cortactin to regulate spine den-
sity and morphology. FRAP was therefore used to test whether
CTTNBP2 regulated the mobility of cortactin at dendritic spines.
To label miRNA expression in GFP-cortactin-positive neurons,
the miRNA cassettes in the miRNA constructs were cloned into
the mCherry expression vector to generate BP2-miR(Cherry)
and Ctrl-miR(Cherry). At 12 DIV, GFP-cortactin was cotrans-
fected with BP2-miR(Cherry) and Ctrl-miR(Cherry) into cul-
tured hippocampal neurons; only neurons double labeled with
GFP and mCherry (Fig. 7A) were subjected to FRAP analysis at 18
DIV. After bleaching, time-lapse recording was continued to
monitor the recovery of GFP-cortactin (Fig. 7B). The recovery
rate of GFP-cortactin at the dendritic spines of CTTNBP2-
knockdown neurons was faster than that in neurons transfected
with Ctrl-miR(Cherry) (Fig. 7C), with recovery half-times of 2.18
and 4.22 s, respectively, suggesting that cortactin mobility is reg-
ulated by CTTNBP2. Moreover, recovered GFP-cortactin fluo-
rescence reached prebleaching levels in CTTNBP2-knock-down
neurons (Fig. 7C). By contrast, the plateau of recovered GFP-
cortactin fluorescence in control neurons maximized at 85% of
the prephotobleaching levels (Fig. 7C), suggesting that ⬃15% of
the cortactin stably resided in dendritic spines due to the presence
of CTTNBP2, likely through the interaction of CTTNBP2 and
cortactin.
To ensure that the mobility of GFP fluorescence was con-
trolled by cortactin, rather than by GFP itself, we also performed
FRAP with cultured hippocampal neurons transfected with GFP.
The recovery rate of GFP alone was much faster than that of
GFP-cortactin in control and CTTNBP2 knock-down neurons
(Fig. 7C), supporting the relevance of the GFP-cortactin FRAP
assay.
Knockdown of CTTNBP2 reduces the distribution of
cortactin at dendritic spines
To further confirm the role of CTTNBP2 in controlling the syn-
aptic distribution of cortactin, we performed immunostaining of
endogenous cortactin in CTTNBP2 knock-down neurons and
control neurons. Consistent with a previous study (Hering and
Sheng, 2003), cortactin was highly enriched at dendritic spines in
control neurons. In CTTNBP2 knock-down neurons, the density
and size of the dendritic spines was reduced (Fig. 8A), with
Figure 6. CTTNBP2 stably resides at dendritic spines after glutamate stimulation. At 21⬃24 DIV, cultured rat hippocampal neurons were stimulated with glutamate (50
M), NMDA (100
M),
AMPA (100
M), and KCl (75
M) for 15 min (A), 15 min (C), or 2 min plus 8 min of recovery (B) in normal growth medium. A, Triple staining with PSD-95 antibodies, CTTNBP2 (BP2) antibodies, and
phalloidin. Yellow arrowheadsin the individual panels and white arrowheads in the merged panels point to the puncta containing overlapping PSD-95 and CTTNBP2. B, C, Distributions of PSD-95,
CTTNBP2, F-actin,and cortactin along dendrites withor without stimulation. Due totechnical limitations, we were unableto visualize PSD-95, CTTNBP2, cortactin,and F-actin in the sameneurons.
Images shown in B and C were not captured from the same cells. Scale bars, 5
m.
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1051
mature spines preserved in few CTTNBP2 knock-down neurons.
In these residual dendritic spines, the enrichment of cortactin was
reduced (Fig. 8 A). A line scan starting from the tip of the den-
dritic spine and ending on the other side of the dendritic shaft was
then performed to quantify the intensity of cortactin at spines
and dendrites. Our quantitative analysis revealed a reduction in
the cortactin protein levels at the dendritic spines in CTTNBP2
knock-down neurons (Fig. 8B). We noticed that the global inten-
sity of cortactin in CTTNBP2 knock-down neurons seemed
lower than that in control neurons. It is possible that knockdown
of CTTNBP2 influences the levels of cortactin protein in neurons.
We then compared the ratio of the cortactin protein level at the
spines to that at the dendrites. In control neurons, the ratio
was 3.8 (the peak of cortactin intensity/the mean of cortactin
intensity from a distance of 2.5–3.5
m ⫽ 149.4/39.6). When
CTTNBP2 was knocked down, the ratio dropped to 2.4 (67.9/
28.2). Therefore, the global reduction in cortactin protein lev-
els is unlikely to be the main cause of the reduction in cortactin
enrichment at the spines. Together, these data support the
notion that CTTNBP2 regulates the dendritic spine distribu-
tion of cortactin in neurons.
Discussion
Here we have characterized the distribution and function of
CTTNBP2 in neurons. CTTNBP2 proteins, which are specifically
expressed in neurons (Ohoka and Takai, 1998), are highly con-
centrated at the dendritic spines of cultured hippocampal neu-
rons and brains. Knockdown of CTTNBP2 reduced the density
and size of dendritic spines, supporting a role for CTTNBP2 in
dendritic spine formation and maintenance. Our observations
also suggest that CTTNBP2 interacts with and immobilizes cor-
tactin at dendritic spines and functions upstream of cortactin to
control dendritic spine density.
Through interactions with F-actin and the Arp2/3 com-
plex, cortactin is believed to facilitate actin polymerization
and branching (Weaver et al., 2001). It thus controls enlarge-
ment of the dendritic spines and maintenance of dendritic
spine morphology (Hering and Sheng, 2003). When neurons
were treated with glutamate, cortactin dispersed to the den-
dritic shafts. However, CTTNBP2 stably resided in dendritic
spines and controlled the mobility of cortactin. Shank, an-
other protein that interacts with the SH3 domain of cortactin,
also remains at the postsynaptic density after glutamate treat-
ment (Naisbitt et al., 1999; Tao-Cheng et al., 2010). Although
it is unclear whether Shank also regulates the mobility of cor-
tactin in dendritic spines, it is possible that multiple binding
partners participate in this regulation.
Through these interactions with binding partners, cortactin
targets various upstream protein complexes and receives signals
to influence dendritic spine formation. Since NMDAR activation
triggers translocation of cortactin from the dendritic spines to the
dendritic shaft, unknown signals downstream of NMDAR may
regulate the interaction of cortactin and CTTNBP2. Given that
Figure 7. GFP-cortactin is more mobile in the absence of CTTNBP2. Rat hippocampal neu-
rons were transfectedat 12 DIV withGFP or GFP-cortactin along with control miRNA (Ctrl-miR)
or CTTNBP2 miRNA(BP2-miR). The plasmidexpressingmiRNA coexpressesmCherry,which was
therefore used to label miRNA-expressingneurons. FRAP measurementswere performed at18
DIV. A, Expression of GFP-cortactin and mCherry in transfected neurons. B, FRAP of GFP-
cortactin in the spines of Ctrl-miR or BP2-miR transfected neurons. Images depict the same
spine before (Pre) and 0, 1, 3, 5, 10, 15, and 25 s after photobleaching. The arrowheads indicate
thetime ofphotobleaching. Scalebar,1
m.C, FRAPanalysis ofGFPand GFP-cortactinover the
35 s period after photobleaching. The average of the fluorescence intensities of 10 scans ac-
quired before photobleaching was set to 100%, and the curves were fit with one-phase expo-
nential equations. A total of 10 neurons for each group and three spines for each neuron were
analyzed. Error bars indicate mean ⫾ SEM.
Figure 8. Knockdown of CTTNBP2 reduces the dendritic spine distribution of cortactin. Cul-
tured rat hippocampal neurons weretransfected withCtrl-miR(mCherry) orBP2-miR(mCherry)
at 12 DIV and fixed for staining at 18 DIV using cortactin antibody. Cortactin was visualized by
Alexa Fluor 488. mCherry signals were used to outlined neuronal morphology. A, Representa-
tive images of the distribution of cortactin along dendrites. The 3-pixel-width lines in the en-
larged images indicate the paths for line scanning, which start from the tip of dendritic spines
and then cross the dendritic shaft. Scale bar, 5
m. B, Quantitative analysis of cortactin distri-
bution by line scan. More than 35 spines collected from seven neurons were analyzed for each
group. The means ⫾ SEM of fluorescence intensity along the path from dendritic spine to
dendritic shaft are provided.
1052 • J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
cortactin is highly phosphorylated on tyrosine, serine, and thre-
onine residues by Src family kinase and ERK (Martin et al., 2006),
and that phosphorylation modulates cortactin function (Huang
et al., 1998; Ammer and Weed, 2008; Ren et al., 2009; Kelley et al.,
2010), protein phosphorylation may regulate the interaction be-
tween cortactin and CTTNBP2.
A previous proteomics study indicated that CTTNBP2 N
terminus-like (CTTNBP2NL), a molecule sharing 54% amino
acid similarity with CTTNBP2, associates with the serine/threo-
nine protein phosphatase 2A (PP2A) protein complex in
HEK293 cells (Goudreault et al., 2009). Since the PP2A complex
contains striatins (Goudreault et al., 2009), which are highly en-
riched at dendritic spines (Gaillard et al., 2006), it seems possible
that CTTNBP2 associates with the PP2A complex in neurons.
Since cortactin is regulated by phosphorylation, if CTTNBP2 in-
teracts with the PP2A complex, it is likely that PP2A dephospho-
rylates cortactin or CTTNBP2 and thus
modulates the activity of cortactin and
CTTNBP2 in controlling F-actin polym-
erization and actin branching (Fig. 9A).
Actin polymerization and branching
have been suggested to play an essential
role in the enlargement of dendritic spines
(for review, see Ethell and Pasquale, 2005;
Tada and Sheng, 2006). Rac1, a small GT-
Pase protein, is known to promote actin
branching and polymerization through the
pathway containing the insulin receptor
substrate p53 (IRSp53), neural-Wiskott–
Aldrich syndrome protein or WASP family
verprolin-homologous protein, and the
Arp2/3 complex (for review, see Ethell
and Pasquale, 2005). Recently, cortactin
was also shown to regulate cellular mo-
bility and actin cytoskeleton dynamics
via regulation of Rac1 (Lai et al., 2009).
Thus, cortactin may use two mecha-
nisms to regulate actin cytoskeleton dy-
namics: direct binding to F-actin and
the Arp2/3 complex, and via control of
Rac1 activity (Fig. 9A ).
Our data suggest that CTTNBP2 is re-
quired for anchoring cortactin at den-
dritic spines. CTTNBP2 may thus direct
cortactin-dependent actin dynamics at
dendritic spines and control spine mor-
phology and density. When CTTNBP2
protein levels are reduced, endogenous
cortactin is unable to target dendritic
spines to maintain the structure of these
spines. When cortactin is overexpressed
in CTTNBP2 knock-down neurons, cor-
tactin ectopically distributes into the den-
dritic spines, thus rescuing the spine
defects caused by CTTNBP2 knockdown
(Fig. 9B).
It is not yet clear how CTTNBP2 stably
resides at dendritic spines. If CTTNBP2 as-
sociates with the striatin–PP2A protein
complex, similar to CTTNBP2NL (Gaillard
et al., 2006; Goudreault et al., 2009), it will be
interesting to examine whether the PP2A
complex is involved in the CTTNBP2
distribution in dendritic spines. On the other hand, it is also
possible that synaptic distribution of the PP2A complex is
dependent on CTTNBP2.
Synapses, the sites where neurons transmit signals to other cells,
may be compromised in autism or other psychiatric disorders. The
Shank protein family interacts with cortactin to regulate the size
and/or density of dendritic spines (Sala et al., 2001; Hung et al., 2008;
Durand et al., 2011). Human genetic studies have demonstrated the
association of Shank3 mutations with autism or other psychiatric
disorders (Bonaglia et al., 2001; Durand et al., 2007; Moessner et al.,
2007). Interestingly, chromosome 7q31, the location of CTTNBP2,
is an autism candidate region. The contribution of CTTNBP2 to
dendritic spine formation highlights the possibility that CTTNBP2
participates in controlling cognitive functions related to autism or
other psychiatric disorders.
Figure 9. Model of the function of CTTNBP2 at dendritic spines. A, Protein–protein interactions of CTTNBP2 and cortactin and
thepotential regulatory signalsof actin cytoskeletondynamics. Cortactininteracts withthe Arp2/3 complexthrough itsN-terminal
regionand binds F-actinvia themiddle repeatdomains. TheC-terminal SH3domain ofcortactin isthe bindingsite forCTTNBP2. Src
and ERK phosphorylation of cortactininfluences theactivity ofcortactin. In addition to directlybinding tothe F-actincytoskeleton,
cortactinmay alsoregulate F-actinbranching throughthe Rac1-IRSp53-WAVEpathway.CTTNBP2NL, amolecule sharingsimilarity
with CTTNBP2, associates with the PP2A protein complex, which contains PP2A, striatins, Mob3, and FAM40A/B. Striatins are
highly enriched at dendritic spines (Gaillard et al., 2006), and the Mob protein family has been shown to regulate synapse
formation inDrosophila (Schulteet al.,2010). FAM40Aand FAM40Bhave recentlybeen shown to modulate the actin cytoskeleton
(Bai et al., 2011). It is unclear whether CTTNBP2 also associates with the PP2A complex in neurons; if so, the PP2A complex likely
dephosphorylates thecortactin–CTTNBP2 complexand regulatesF-actin dynamics.B, CTTNBP2regulates cortactindistribution at
dendritic spines and thus maintains spine structure. The presence of CTTNBP2 targets cortactin to the dendritic spines. When
CTTNBP2 is knocked down, cortactin is not efficiently targeted to dendritic spines, leading to shrinkage of the spines. When
cortactinisoverexpressedinCTTNBP2 knock-down neurons, the ectopic distributionofcortactin into the dendritic spinesmaintains
the spine structure and rescues the defects caused by CTTNBP2 knockdown. Thus far, it is unclear what mechanism or molecule
anchors CTTNBP2 at the dendritic spines.
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1053
Although the full-length human CTTNBP2 sequence encodes
a large protein of up to 1663 aa residues (Cheung et al., 2001), we
identified three splicing variants of CTTNBP2 and demonstrated
that the short form is the major protein product in the brain. The
sequence around the alternative splicing sites is highly conserved
from human to chicken (data not shown), suggesting that expres-
sion of CTTNBP2 splicing variants is conserved. To explore the
role of CTTNBP2 in autism, it will be critical to focus on muta-
tions located before exon 5 of CTTNBP2, since the C-terminal
half of the CTTNBP2 long-form protein is not present or exists in
limited amounts in neurons.
Exogenous CTTNBP2 proteins in COS cells are colocalized
with cortactin and enriched at the cell cortex (lamellipodia
and intracellular puncta); a similar distribution of GFP-
cortactin was reported in NIH3T3 cells (Kaksonen et al.,
2000). Cortactin has been implicated in endocytosis because
cortactin localizes with clathrin-coated pits (Cao et al., 2003,
2010) and because the knockdown of cortactin impairs endo-
cytosis (Zhu et al., 2005). Cortactin may bind dynamin and
thus couple the F-actin beneath the plasma membrane to en-
docytic vesicles, providing physical force to separate endocytic
vesicles from the plasma membrane (Zhu et al., 2005). A re-
cent investigation has demonstrated that cortactin is required
for endosomal segregation from early endosomes to late/recy-
cling endosomes (Ohashi et al., 2011). We also observed punc-
tate colocalization of CTTNBP2 and cortactin, particularly in
time-lapse imaging. Intracellular vesicles always associated
with cortactin/CTTNBP2 puncta in COS cells, and thus CTT-
NBP2 is likely also involved in endocytosis in neurons.
References
Ammer AG, Weed SA (2008) Cortactin branches out: roles in regulating
protrusive actin dynamics. Cell Motil Cytoskeleton 65:687–707.
Bai SW, Herrera-Abreu MT, Rohn JL, Racine V, Tajadura V, Suryavanshi N,
Bechtel S, Wiemann S, Baum B, Ridley AJ (2011) Identification and
characterization of a set of conserved and new regulators of cytoskeletal
organisation, cell morphology and migration. BMC Biol 9:54.
Bonaglia MC, Giorda R, Borgatti R, Felisari G, Gagliardi C, Selicorni A, Zuf-
fardi O (2001) Disruption of the ProSAP2 gene in a t(12;22)(q24.1;
q13.3) is associated with the 22q13.3 deletion syndrome. Am J Hum
Genet 69:261–268.
Cao H, Orth JD, Chen J, Weller SG, Heuser JE, McNiven MA (2003) Cor-
tactin is a component of clathrin-coated pits and participates in receptor-
mediated endocytosis. Mol Cell Biol 23:2162–2170.
Cao H, Chen J, Krueger EW, McNiven MA (2010) SRC-mediated phos-
phorylation of dynamin and cortactin regulates the “constitutive” endo-
cytosis of transferrin. Mol Cell Biol 30:781–792.
Chao HW, Hong CJ, Huang TN, Lin YL, Hsueh YP (2008) SUMOylation of
the MAGUK protein CASK regulates dendritic spinogenesis. J Cell Biol
182:141–155.
Cheung J, Petek E, Nakabayashi K, Tsui LC, Vincent JB, Scherer SW (2001)
Identification of the human cortactin-binding protein-2 gene from the
autism candidate region at 7q31. Genomics 78:7–11.
Cosen-Binker LI, Kapus A (2006) Cortactin: the gray eminence of the cyto-
skeleton. Physiology (Bethesda) 21:352–361.
Du Y, Weed SA, Xiong WC, Marshall TD, Parsons JT (1998) Identification
of a novel cortactin SH3 domain-binding protein and its localization to
growth cones of cultured neurons. Mol Cell Biol 18:5838 –5851.
Durand CM, Betancur C, Boeckers TM, Bockmann J, Chaste P, Fauchereau F,
Nygren G, Rastam M, Gillberg IC, Anckarsa¨ter H, Sponheim E, Goubran-
Botros H, Delorme R, Chabane N, Mouren-Simeoni MC, de Mas P, Bieth
E, Roge´B,He´ron D, Burglen L, et al (2007) Mutations in the gene en-
coding the synaptic scaffolding protein SHANK3 are associated with au-
tism spectrum disorders. Nat Genet 39:25–27.
Durand CM, Perroy J, Loll F, Perrais D, Fagni L, Bourgeron T, Montcouquiol
M, Sans N (2011) SHANK3 mutations identified in autism lead to mod-
ification of dendritic spine morphology via an actin-dependent mecha-
nism. Mol Psychiatry. Advance online publication. Retrieved December
15, 2012. doi:10.1038/mp.2011.57.
Ethell IM, Pasquale EB (2005) Molecular mechanisms of dendritic spine
development and remodeling. Prog Neurobiol 75:161–205.
Fischer M, Kaech S, Wagner U, Brinkhaus H, Matus A (2000) Glutamate
receptors regulate actin-based plasticity in dendritic spines. Nat Neurosci
3:887–894.
Gaillard S, Bailly Y, Benoist M, Rakitina T, Kessler JP, Fronzaroli-Molinie`res
L, Dargent B, Castets F (2006) Targeting of proteins of the striatin family
to dendritic spines: role of the coiled-coil domain. Traffic 7:74– 84.
Goudreault M, D’Ambrosio LM, Kean MJ, Mullin MJ, Larsen BG, Sanchez A,
Chaudhry S, Chen GI, Sicheri F, Nesvizhskii AI, Aebersold R, Raught B,
Gingras AC (2009) A PP2A phosphatase high density interaction net-
work identifies a novel striatin-interacting phosphatase and kinase com-
plex linked to the cerebral cavernous malformation 3 (CCM3) protein.
Mol Cell Proteomics 8:157–171.
Harris KM, Stevens JK (1989) Dendritic spines of CA 1 pyramidal cells in
the rat hippocampus: serial electron microscopy with reference to their
biophysical characteristics. J Neurosci 9:2982–2997.
Hering H, Sheng M (2003) Activity-dependent redistribution and essential
role of cortactin in dendritic spine morphogenesis. J Neurosci 23:
11759–11769.
Huang C, Liu J, Haudenschild CC, Zhan X (1998) The role of tyrosine phos-
phorylation of cortactin in the locomotion of endothelial cells. J Biol
Chem 273:25770 –25776.
Hung AY, Futai K, Sala C, Valtschanoff JG, Ryu J, Woodworth MA, Kidd FL,
Sung CC, Miyakawa T, Bear MF, Weinberg RJ, Sheng M (2008) Smaller
dendritic spines, weaker synaptic transmission, but enhanced spatial
learning in mice lacking Shank1. J Neurosci 28:1697–1708.
Kaksonen M, Peng HB, Rauvala H (2000) Association of cortactin with dy-
namic actin in lamellipodia and on endosomal vesicles. J Cell Sci
113:4421–4426.
Kelley LC, Hayes KE, Ammer AG, Martin KH, Weed SA (2010) Cortactin
phosphorylated by ERK1/2 localizes to sites of dynamic actin regulation
and is required for carcinoma lamellipodia persistence. PLoS One
5:e13847.
Lai FP, Szczodrak M, Oelkers JM, Ladwein M, Acconcia F, Benesch S, Auinger
S, Faix J, Small JV, Polo S, Stradal TE, Rottner K (2009) Cortactin pro-
motes migration and platelet-derived growth factor-induced actin reor-
ganization by signaling to Rho-GTPases. Mol Biol Cell 20:3209 –3223.
Martin KH, Jeffery ED, Grigera PR, Shabanowitz J, Hunt DF, Parsons JT
(2006) Cortactin phosphorylation sites mapped by mass spectrometry.
J Cell Sci 119:2851–2853.
Matus A (2000) Actin-based plasticity in dendritic spines. Science
290:754–758.
Moessner R, Marshall CR, Sutcliffe JS, Skaug J, Pinto D, Vincent J, Zwaigen-
baum L, Fernandez B, Roberts W, Szatmari P, Scherer SW (2007) Con-
tribution of SHANK3 mutations to autism spectrum disorder. Am J Hum
Genet 81:1289 –1297.
Naisbitt S, Kim E, Tu JC, Xiao B, Sala C, Valtschanoff J, Weinberg RJ, Worley
PF, Sheng M (1999) Shank, a novel family of postsynaptic density pro-
teins that binds to the NMDA receptor/PSD-95/GKAP complex and cor-
tactin. Neuron 23:569–582.
Ohashi E, Tanabe K, Henmi Y, Mesaki K, Kobayashi Y, Takei K (2011)
Receptor sorting within endosomal trafficking pathway is facilitated by
dynamic actin filaments. PLoS One 6:e19942.
Ohoka Y, Takai Y (1998) Isolation and characterization of cortactin iso-
forms and a novel cortactin-binding protein, CBP90. Genes Cells
3:603–612.
Pontrello CG, Ethell IM (2009) Accelerators, brakes, and gears of actin dy-
namics in dendritic spines. Open Neurosci J 3:67–86.
Ren G, Crampton MS, Yap AS (2009) Cortactin: coordinating adhesion and
the actin cytoskeleton at cellular protrusions. Cell Motil Cytoskeleton
66:865–873.
Sala C, Pie¨ch V, Wilson NR, Passafaro M, Liu G, Sheng M (2001) Regulation
of dendritic spine morphology and synaptic function by Shank and
Homer. Neuron 31:115–130.
Schulte J, Sepp KJ, Jorquera RA, Wu C, Song Y, Hong P, Littleton JT (2010)
DMob4/Phocein regulates synapse formation, axonal transport, and mi-
crotubule organization. J Neurosci 30:5189–5203.
1054
• J. Neurosci., January 18, 2012 • 32(3):1043–1055 Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation
Tada T, Sheng M (2006) Molecular mechanisms of dendritic spine morpho-
genesis. Curr Opin Neurobiol 16:95–101.
Tao-Cheng JH, Dosemeci A, Gallant PE, Smith C, Reese T (2010) Activity
induced changes in the distribution of Shanks at hippocampal synapses.
Neuroscience 168:11–17.
Tu JC, Xiao B, Naisbitt S, Yuan JP, Petralia RS, Brakeman P, Doan A, Aakalu
VK, Lanahan AA, Sheng M, Worley PF (1999) Coupling of mGluR/
Homer and PSD-95 complexes by the Shank family of postsynaptic den-
sity proteins. Neuron 23:583–592.
Uruno T, Liu J, Zhang P, Fan Yx, Egile C, Li R, Mueller SC, Zhan X (2001)
Activation of Arp2/3 complex-mediated actin polymerization by cortac-
tin. Nat Cell Biol 3:259–266.
Weaver AM, Karginov AV, Kinley AW, Weed SA, Li Y, Parsons JT, Cooper JA
(2001) Cortactin promotes and stabilizes Arp2/3-induced actin filament
network formation. Curr Biol 11:370–374.
Weed SA, Karginov AV, Schafer DA, Weaver AM, Kinley AW,Cooper JA, Parsons JT
(2000) Cortactin localization to sites of actin assembly in lamellipodia requires
interactions with F-actin and the Arp2/3 complex. J Cell Biol 151:29 – 40.
Wu H, Parsons JT (1993) Cortactin, an 80/85-kilodalton pp60src substrate,
is a filamentous actin-binding protein enriched in the cell cortex. J Cell
Biol 120:1417–1426.
Zhu J, Zhou K, Hao JJ, Liu J, Smith N, Zhan X (2005) Regulation of cortac-
tin/dynamin interaction by actin polymerization during the fission of
clathrin-coated pits. J Cell Sci 118:807– 817.
Ziv NE, Smith SJ (1996) Evidence for a role of dendritic filopodia in synap-
togenesis and spine formation. Neuron 17:91–102.
Chen and Hsueh • CTTNBP2 Regulates Dendritic Spine Formation J. Neurosci., January 18, 2012 • 32(3):1043–1055 • 1055
