GSK-3[beta] Regulates Phosphorylation of CRMP-2 and Neuronal Polarity

Abstract

Neurons are highly polarized and comprised of two structurally and functionally distinct parts, an axon and dendrites. We previously showed that collapsin response mediator protein-2 (CRMP-2) is critical for specifying axon/dendrite fate, possibly by promoting neurite elongation via microtubule assembly. Here, we showed that glycogen synthase kinase-3beta (GSK-3beta) phosphorylated CRMP-2 at Thr-514 and inactivated it. The expression of the nonphosphorylated form of CRMP-2 or inhibition of GSK-3beta induced the formation of multiple axon-like neurites in hippocampal neurons. The expression of constitutively active GSK-3beta impaired neuronal polarization, whereas the nonphosphorylated form of CRMP-2 counteracted the inhibitory effects of GSK-3beta, indicating that GSK-3beta regulates neuronal polarity through the phosphorylation of CRMP-2. Treatment of hippocampal neurons with neurotrophin-3 (NT-3) induced inactivation of GSK-3beta and dephosphorylation of CRMP-2. Knockdown of CRMP-2 inhibited NT-3-induced axon outgrowth. These results suggest that NT-3 decreases phosphorylated CRMP-2 and increases nonphosphorylated active CRMP-2, thereby promoting axon outgrowth.

Full text

Cell, Vol. 120, 137–149, January 14, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.cell.2004.11.012
GSK-3␤Regulates Phosphorylation
of CRMP-2 and Neuronal Polarity
neurite/axon is generated by two mechanisms: the
transport of microtubule polymer and the microtubule
assembly at the plus ends of the microtubules. Both
Takeshi Yoshimura,
1
Yoji Kawano,
1
Nariko Arimura,
1
Saeko Kawabata,
1
Akira Kikuchi,
2
and Kozo Kaibuchi
1,
*
1
Department of Cell Pharmacology appear to contribute to axon outgrowth (Baas, 1997).
CRMP-2, which also has been independently identi-Graduate School of Medicine
Nagoya University fied as Ulip2/CRMP-62/TOAD-64/DRP-2, is one of at
least five isoforms (Goshima et al., 1995; Arimura et al.,65 Tsurumai, Showa-ku, Nagoya
Aichi 466-8550 2004). CRMP-2 is expressed exclusively and highly in
the developing nervous system. Mutations in the
2
Department of Biochemistry
Graduate School of Biomedical Sciences UNC-33 gene, a Caenorhabditis elegans homolog of
CRMPs, lead to severely uncoordinated movement andHiroshima University
1-2-3, Kasumi, Minami-ku abnormalities in the guidance of axons of many neurons
(Hedgecock et al., 1985). Previously, we showed thatHiroshima 734-8551
Japan CRMP-2 is enriched in the growing axon of hippocampal
neurons, the overexpression of full-length CRMP-2 in-
duces the formation of multiple axons and elongation
of the primary axon, and the dominant-negative form ofSummary
CRMP-2 inhibits axon formation (Inagaki et al., 2001).
CRMP-2 shows the ability to convert immature neuritesNeurons are highly polarized and comprised of two
structurally and functionally distinct parts, an axon and and preexisting dendrites to axons. Thus, CRMP-2 is
crucial for axon outgrowth and determination of the fatedendrites. We previously showed that collapsin re-
sponse mediator protein-2 (CRMP-2) is critical for of the axon and dendrites, thereby establishing and
maintaining neuronal polarity. In addition, we recentlyspecifying axon/dendrite fate, possibly by promoting
neurite elongation via microtubule assembly. Here, we demonstrated that CRMP-2 binds to tubulin heterodi-
mers to promote microtubule assembly, thereby en-showed that glycogen synthase kinase-3␤(GSK-3␤)
phosphorylated CRMP-2 at Thr-514 and inactivated it. hancing axon elongation and branching (Fukata et al.,
2002). However, the molecular mechanism by whichThe expression of the nonphosphorylated form of
CRMP-2 or inhibition of GSK-3␤induced the formation CRMP-2 is regulated remains unclear.
A ternary complex of PAR-3, PAR-6, and atypical pro-of multiple axon-like neurites in hippocampal neurons.
The expression of constitutively active GSK-3␤impaired tein kinase C (aPKC) functions in various cell polarization
events from worms to mammals (Etienne-Mannevilleneuronal polarization, whereas the nonphosphory-
lated form of CRMP-2 counteracted the inhibitory ef- and Hall, 2003b), including cultured hippocampal neu-
rons (Shi et al., 2003; Nishimura et al., 2004). The PAR-fects of GSK-3␤, indicating that GSK-3␤regulates neu-
ronal polarity through the phosphorylation of CRMP-2. 6-PAR-3-aPKC complex accumulates at the tip of the
axon, and its polarized localization and aPKC activityTreatment of hippocampal neurons with neurotro-
phin-3 (NT-3) induced inactivation of GSK-3␤and de- are important for axon specification (Shi et al., 2003;
Nishimura et al., 2004). aPKC can also phosphorylatephosphorylation of CRMP-2. Knockdown of CRMP-2
inhibited NT-3-induced axon outgrowth. These results GSK-3␤and inactivate its kinase activity, and GSK-3␤
is important for polarization of migrating fibroblastssuggest that NT-3 decreases phosphorylated CRMP-2
and increases nonphosphorylated active CRMP-2, (Etienne-Manneville and Hall, 2003a). However, it re-
mains unknown whether GSK-3␤is involved in neuronalthereby promoting axon outgrowth.
polarity and, if so, what the target substrates of GSK-3␤
are.Introduction
We report here that GSK-3␤phosphorylates CRMP-2
at Thr-514, inactivates the CRMP-2 activity, and partici-The neuron is one of the most highly polarized cells
known and is comprised of two structurally and func- pates in neuronal polarization through CRMP-2. NT-3
and brain-derived neurotrophic factor (BDNF) inhibittionally distinct parts, an axon and dendrites (Craig and
Banker, 1994). The specification of the axon is thought GSK-3␤via the phosphatidylinositol-3-kinase (PI3-
kinase)/Akt (also known as PKB) pathway, and therebyto depend on its length relative to the other minor pro-
cesses, which are called immature neurites (Bradke and reduce phosphorylation levels of CRMP-2 at Thr-514,
leading to axon elongation and branching.Dotti, 2000). Elongation of one of the immature neurites
is necessary for axon specification. Intracellular mecha-
nisms that help to enhance neurite/axon outgrowth evi- Results
dently require reorganization of cytoskeletons including
actin filaments and microtubules. Microtubule assembly GSK-3␤Phosphorylates CRMP-2
occurs in the cell body and the growth cone (Brown It has been previously reported that the phosphorylation
et al., 1992). Formation of the microtubule array in the of CRMP-2 at Thr-514 by unidentified kinases induces
the mobility shift of CRMP-2 (Gu et al., 2000). The optimal
consensus site for the phosphorylation by GSK-3␤is
*Correspondence: kaibuchi@med.nagoya-u.ac.jp

Cell
138
were cotransfected with GSK-3␤wild-type (wt) and
CRMP-2 wt and subjected to immunoblot analysis with
anti-CRMP-2 antibody (Figure 1B). The mobility shift of
CRMP-2 (asterisk) was observed in the cells cotrans-
fected with GSK-3␤wt and CRMP-2 wt. We then pro-
duced the antibody that specifically recognized phos-
phorylated CRMP-2 at Thr-514 (anti-pT514 antibody).
In the COS7 cells cotransfected with CRMP-2 wt and
GSK-3␤wt, anti-pT514 antibody recognized the upper
bands, which corresponded to the phosphorylated
CRMP-2 at Thr-514. In contrast, the immunoreactive
band was not observed in runs with CRMP-2 T514A
(Thr-514 was replaced by Ala) and GSK-3␤wt. Consis-
tent with the result using anti-pT514 antibody, the mobil-
ity shift of CRMP-2 (asterisk) was not observed in these
cells. These results suggest that GSK-3␤phosphorylates
CRMP-2 at Thr-514.
Next, we examined whether GSK-3␤phosphorylates
CRMP-2 at Thr-514 in vitro (Figure 1C). The phosphoryla-
tion of CRMP-2 at Thr-514 was not observed when
GSK-3␤alone was added. GSK-3␤requires prephos-
phorylation of its substrate (Eldar-Finkelman, 2002).
CRMP-2 has a consensus motif recognized by Cdk5 at
Ser-522 in the vicinity of the phosphorylation site of
GSK-3␤, Thr-514. We found that GSK-3␤phosphory-
lated CRMP-2 that was prephosphorylated by Cdk5
in vitro. When Cdk5 alone was added, CRMP-2 at Thr-
514 was not phosphorylated. These results indicate that
GSK-3␤can phosphorylate CRMP-2 at Thr-514 after
CRMP-2 is phosphorylated by Cdk5.
We then evaluated each phosphorylation site (Thr-
514, Ser-518, and Ser-522). Not only CRMP-2 T514A but
also S518A (Ser-518 was replaced by Ala), S522A (Ser-
522 was replaced by Ala), and AAA (Thr-514, Ser-518,
and Ser-522 were replaced by Ala) blocked the band
shift of CRMP-2 (see Supplemental Figure S1A at http://
www.cell.com/cgi/content/full/120/1/137/DC1/). Con-
sistently, the band immunoreactive with anti-pT514 anti-
body was not observed in runs with T514A, S518A,
S522A, and AAA. It is reported that the phosphorylation
of CRMP-2 at Thr-514 but not Ser-518 or Ser-522 in-
duces the mobility shift of CRMP-2 (Gu et al., 2000).
These results suggest that Ser-518 and Ser-522 are re-
Figure 1. GSK-3␤Phosphorylates CRMP-2 at Thr-514
quired for the phosphorylation of CRMP-2 at Thr-514
(A) Potential phosphorylation sites of CRMP-2 by GSK-3␤and Cdk5.
The numbers denote amino acid positions. The optimal consensus
by GSK-3␤. Ser-522 appears to be phosphorylated by
site for the phosphorylation by GSK-3␤is Ser/Thr-Xaa-Xaa-Xaa-
Cdk5 as a priming kinase, and Ser-518 and Thr-514
Ser/Thr (where Xaa represents any amino acid). The black bar shows
seem to be phosphorylated by GSK-3␤after the phos-
a consensus motif recognized by Cdk5.
phorylation at Ser-522.
(B) COS7 cells were cotransfected with CRMP-2 and GSK-3␤mu-
tants. Samples were subjected to immunoblot analysis with anti-
pT514 (top) and anti-CRMP-2 (bottom) antibodies. The mobility shift
Nonphosphorylated Pool of CRMP-2 Localizes
of CRMP-2 (asterisk) was caused by the phosphorylation of CRMP-2
in Axonal Growth Cone
at Thr-514.
In cultured hippocampal neurons, neurons extend sev-
(C) Kinase assay was performed using purified CRMP-2, GSK-3␤,
eral minor processes during the first 12–24 hr after plat-
and Cdk5 in vitro. Each reaction mixture was subjected to SDS-
ing (stages 1 and 2; Dotti et al. [1988]). Then, one of the
PAGE and immunoblot analysis with anti-pT514 (top) and anti-
processes begins to extend rapidly to form an axon,
CRMP-2 (bottom) antibodies.
resulting in the morphological polarization of the neuron
(stage 3). The remaining processes result in the morpho-
logical features of dendrites (stage 4). By 7 days in vitroSer/Thr-Xaa-Xaa-Xaa-Ser/Thr (where Xaa represents
any amino acid; Frame and Cohen [2001]; Eldar-Finkel- (DIV), the neurons become highly polarized, and the
axon and dendrites continue to mature and subse-man [2002]). Considering the consensus sequences of
GSK-3␤, we speculated that CRMP-2 was phosphory- quently develop (stage 5).
We examined whether CRMP-2 is phosphorylated bylated by GSK-3␤at Thr-514 (Figure 1A). To examine
whether GSK-3␤phosphorylates CRMP-2, COS7 cells GSK-3␤in hippocampal neurons during axon outgrowth.

GSK-3␤and CRMP-2 in Neuronal Polarity
139
Figure 2. Nonphosphorylated CRMP-2 Lo-
calized in the Axonal Growth Cone
(A) Hippocampal neurons were cultured in the
presence of GSK-3 inhibitors (2 mM LiCl,
5␮M SB216763, or 25 ␮M SB415286), buffer,
2 mM NaCl, or DMSO for 48 hr before TCA
treatment. Each sample was subjected to
SDS-PAGE and immunoblot analysis with
anti-pT514 (top) and anti-CRMP-2 (bottom)
antibodies. The asterisk indicates the mobil-
ity shift of CRMP-2. These results are repre-
sentative of three independent experiments.
(B) Nonphosphorylated pool of CRMP-2 lo-
calized in growth cone of growing axon. Hip-
pocampal neurons were fixed at 3 DIV and
then immunostained with anti-pT514 (Ba and
Bd) and anti-CRMP-2 (Bb and Be) antibodies.
The merged (Bc and Bf) and ratio (phosphory-
lated CRMP-2 to total CRMP-2, [Bg]) are
shown. The graph (Bf-2) plots the fluores-
cence intensities of CRMP-2 phosphorylated
at Thr-514 (red) and total CRMP-2 (green) in
the line (Bf, from X to Y). The ratio of fluores-
cence intensities (phosphorylated CRMP-2 to
total CRMP-2) was measured in the indicated
areas of 30 cells (arbitrary units per pixel [Bh]).
Scale bar, 10 ␮m.
The immunoblot analysis with anti-pT514 antibody re- (phosphorylated CRMP-2, red) and anti-CRMP-2 anti-
bodies (total CRMP-2, green; Figure 2B). CRMP-2vealed that CRMP-2 was phosphorylated at Thr-514 in
hippocampal neurons at 3 DIV (more than 75% of neu- (green) was enriched in the distal part of the growing
axon as previously reported (Inagaki et al., 2001). Phos-rons at stage 3; Figure 2A). About 30% of CRMP-2 was
constitutively phosphorylated at Thr-514. When hippo- phorylated CRMP-2 at Thr-514 (red) was enriched in the
distal part of the growing axon without growth cone.campal neurons were cultured in the presence of GSK-3
inhibitors (LiCl, SB216763, or SB415286), the phosphor- The merged images of CRMP-2 and phosphorylated
CRMP-2 at Thr-514 immunofluorescence enable us toylation levels of CRMP-2 at Thr-514 were decreased.
Cdk inhibitor (olomoucine) also decreased the phos- roughly estimate the phosphorylation levels of CRMP-2.
The merged image in the shaft was yellow, whereas thatphorylation levels of CRMP-2 at Thr-514 (see Supple-
mental Figure S1B on the Cell web site). These results in the axonal growth cone was more greenish than that
in the shaft. Thus, the ratio of phosphorylated CRMP-2suggest that GSK-3␤phosphorylates CRMP-2 at Thr-
514 after CRMP-2 is phosphorylated by Cdk5 in hippo- to total CRMP-2 in the axonal growth cone was lower
than that in the shaft.campal neurons.
To investigate the spatial distribution of phosphory- Further, clear evidence was obtained by intensity im-
aging of phosphorylated CRMP-2 and total CRMP-2lated CRMP-2 at Thr-514, hippocampal neurons were
fixed at 3 DIV and then immunostained with anti-pT514 (Figure 2B). The intensity imaging was similar in shafts,

Cell
140
Figure 3. Phosphorylated CRMP-2 Lowers Its Activity to Interact with Tubulin
(A) HeLa cells were transfected with GFP-GST (Aa–Ac), GFP-CRMP-2 wt (Ad–Af), T514A (Ag–Ai), or T514D (Aj–Al). They were then immunostained
with anti-tubulin antibody (DM1A; [Ab], [Ae], [Ah], and [Ak]). HeLa cells having spindles in mitotic phase were chosen to show the CRMP-2 local-
ization.
(B) HeLa cells transfected with GFP-CRMP-2 wt were immunostained with anti-pT514 and tubulin (DM1A) antibodies.
(C) Porcine brain lysate was mixed with glutathione-Sepharose 4B beads coated with purified GST, CRMP-2 wt-GST, or CRMP-2 T514D-

GSK-3␤and CRMP-2 in Neuronal Polarity
141
whereas that of phosphorylated CRMP-2 was lower than Nonphosphorylated CRMP-2 Promotes Axon
Outgrowth and Induces the Formation
that of total CRMP-2 in the axonal growth cone. The
of Multiple Axon-Like Neurites
intensity of total CRMP-2 and phosphorylated CRMP-2
To examine the effect of CRMP-2 mutants, hippocampal
in the dendrites were lower than those in the axon. These
neurons were transfected with Myc-CRMP-2 wt, T514A,
results indicate that there is a nonphosphorylated
or T514D and fixed at 3 DIV. CRMP-2 wt enhanced axon
CRMP-2 pool at Thr-514 in the growing axonal growth
elongation and branching as previously reported (Fukata
cone. We confirmed that GSK-3 inhibitor decreased the
et al., 2002; Figures 4A and 4B). CRMP-2 T514A pro-
phosphorylation levels of CRMP-2 by immunostaining
moted axon outgrowth and branching more than CRMP-2
with anti-pT514 antibody (Supplemental Figure S1C).
wt, whereas CRMP-2 T514D had the weaker activity.
These results are consistent with the observation that
CRMP-2 T514D showed lower activity to interact with
Phosphorylation of CRMP-2 Lowers its Activity
tubulin. CRMP-2 S518A, S522A, and AAA promoted
for Interaction with Tubulin
axon outgrowth as well as CRMP-2 T514A (Supplemen-
We recently found that the phosphorylation of CRMP-2
tal Figure S2A).
at Thr-555 by Rho-kinase turns off the ability of CRMP-2
To examine the effect of CRMP-2 mutants on axon
to bind tubulin (N. Arimura, C. Menager, Y.K., Y. Fukata,
formation, hippocampal neurons transfected with Myc-
M. Amano, Y. Goshima, N. Morone, J. Usukura, and K.K.,
CRMP-2 wt, T514A, or T514D were fixed at 6 DIV. To
unpublished data). Here, we tried to prepare CRMP-2
visualize secondary axons, neurons at 6 DIV are better
phosphorylated by GSK-3␤in vitro but found that the
than those at 3 DIV because the secondary axons appear
stoichiometry of the phosphorylation was less than 40%.
later (Inagaki et al., 2001). CRMP-2 wt induced the for-
Then, we produced and characterized a phosphomimic
mation of multiple axon-like neurites as previously re-
CRMP-2 mutant. CRMP-2 T514D (Thr-514 was replaced
ported (Inagaki et al. [2001]; Figure 4D). CRMP-2 T514A
by Asp) is expected to mimic the phosphorylated form
increased the percentage of neurons that had multiple
of CRMP-2 (Kamisoyama et al., 1994; Sweeney et al.,
long neurites and multiple Tau-1-positive neurites (Fig-
1994; Bresnick et al., 1995). Because ectopic CRMP-2
ures 4C and 4D). The effect of CRMP-2 T514A on the
was diffusely overexpressed in neurons, it was difficult
formation of multiple axon-like neurites was stronger
to examine the colocalization of ectopic CRMP-2 with
than that of CRMP-2 wt, whereas CRMP-2 T514D
microtubules in detail at axons. Instead, we used HeLa
showed the weaker activity (Figure 4D). The consistent
cells, in which green fluorescent protein (GFP)-tagged results were obtained by using other axonal markers,
CRMP-2 was uniformly distributed along microtubules such as synapsin I and synaptophysin for axon and
(Fukata et al., 2002). GFP-CRMP-2 wt was clearly local- MAP2 for dendrite (Fletcher et al., 1991; Supplemental
ized along the mitotic spindle (N. Arimura, C. Menager, Figure S2B). CRMP-2 S518A, S522A, and AAA induced
Y.K., Y. Fukata, M. Amano, Y. Goshima, N. Morone, J. the formation of multiple axon-like neurites as well as
Usukura, and K.K., unpublished data; Figure 3A, Supple- CRMP-2 T514A (Supplemental Figure S2C).
mental Figure S1D). GFP-CRMP-2 T514A, S518A,
S522A, and AAA were also localized along the mitotic
spindle, whereas GFP-CRMP-2 T514D, S518D, S522D, GSK-3␤Regulates Neuronal Polarity via CRMP-2
and DDD resulted in lower ability to localize along the Inhibition of GSK-3␤results in enhanced neurite growth
mitotic spindle (Figure 3A, Supplemental Figure S1D). in rat cerebellar granule neurons and DRG neurons (Mu-
Under the same conditions, phosphorylated CRMP-2 noz-Montano et al., 1999; Jones et al., 2003). To investi-
was diffusely distributed throughout the cytoplasm and gate the effect of inhibition of GSK-3␤on axon outgrowth
not localized along the mitotic spindle (Figure 3B). and formation, hippocampal neurons were cultured in
Next, we directly compared the binding activity of the presence of GSK-3 inhibitors (LiCl, SB216763, or
CRMP-2 wt and T514D to tubulin in vitro. Porcine brain SB415286). These inhibitors slightly enhanced axon
lysate, which contained tubulin, was mixed with glutathi- elongation (Figure 5A) and branching (data not shown)
one-Sepharose 4B beads coated with GST, CRMP-2 wt- at 3 DIV and increased the percentage of neurons that
GST, or CRMP-2 T514D-GST. Tubulin bound to CRMP-2 had multiple long neurites and multiple Tau-1-positive
wt-GST as previously described (Fukata et al., 2002; neurites at 6 DIV (Figures 5B and 5C). The consistent
Figure 3C). The stoichiometry of bound tubulin to results were obtained by using other axonal markers,
CRMP-2 wt was about 0.3, whereas the stoichiometry such as synapsin I and synaptophysin for axon and
of bound tubulin to CRMP-2 T514D was about 0.05. The MAP2 for dendrite (Supplemental Figure S3A).
binding activity of CRMP-2 T514D to tubulin was weaker To confirm these results, we used hairpin short in-
than that of CRMP-2 wt. Taken together, these results terfering RNA (siRNA) construct for GSK-3␤(Yu et al.,
suggest that the binding activity of CRMP-2 to tubulin 2003). Previous applications of the same hairpin siRNA
is decreased by the phosphorylation of CRMP-2 by construct in mammalian cells have shown efficient and
specific inhibition of GSK-3␤(Yu et al., 2003), and weGSK-3␤.
GST in vitro. The bound proteins were coeluted with GST fusion proteins by the addition of buffer containing glutathione. Portions of start
samples (top) and eluates (middle) were subjected to SDS-PAGE followed by immunoblot analysis with anti-tubulin antibody (DM1A). CRMP-2
wt-GST and CRMP-2 T514D-GST were immobilized in comparable quantities (bottom; Coomassie brilliant blue staining). These results are
representative of three independent experiments.

Cell
142
Figure 4. Nonphosphorylated CRMP-2 Enhanced Axon Elongation and Branching and Induced the Formation of Multiple Axon-Like Neurites
(A) Hippocampal neurons were transfected with Myc-GST (Aa), Myc-CRMP-2 wt (Ab), T514A (Ac), or T514D (Ad). Neurons were fixed at 3 DIV
and then immunostained with anti-Myc antibody. Scale bar, 100 ␮m.
(B) Axon length (Ba) and the number of branch tips per axon (Bb) were measured at 3 DIV neurons transfected with the indicated plasmids.
n⫽50 per experimental condition. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t
test; *p ⬍0.05; **p ⬍0.01).
(C) Hippocampal neurons were transfected with Myc-GST (Ca), Myc-CRMP-2 T514A (Cb). Neurons were fixed at 6 DIV and then immunostained
with anti-Myc and Tau-1 antibodies. The enlarged images of the neurites (1, 2, 3) are shown. The neuron transfected with Myc-GST (Ca) had
one Tau-1-positive neurite (1). The neuron transfected with Myc-CRMP-2 T514A (Cb) had multiple Tau-1-positive neurites (1, 2, 3). Scale bar,
100 ␮m.
(D) The percentages of the cells that had multiple long neurites (Da) and multiple Tau-1-positive neurites (Db) were measured at 6 DIV neurons
transfected with the indicated plasmids. Fifty cells for each plasmid were measured by tracing images of immunofluorescence staining with
anti-Myc antibody. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; *p ⬍0.05;
**p ⬍0.01).

GSK-3␤and CRMP-2 in Neuronal Polarity
143
obtained the similar result in rat 3Y1 cells (Supplemental (PIP
3
) and phosphoinositide-dependent kinase (PDK;
Scheid and Woodgett [2001]). Activated Akt phosphory-Figure S3B), indicating that the hairpin siRNA construct
for GSK-3␤is effective in rat cells. Rat hippocampal lates GSK-3␤at Ser-9 and inactivates its kinase activity
(Grimes and Jope, 2001).neurons were cotransfected with hairpin siRNA for
GSK-3␤and Myc-GST. Inhibition of GSK-3␤by hairpin We examined whether NT-3 regulates the phosphory-
lation levels of CRMP-2 (Figure 6A). Hippocampal neu-siRNA enhanced axon elongation at 3 DIV (Figure 5A)
and increased the percentage of neurons that had multi- rons were stimulated by NT-3 for 3, 10, 30, or 90 min.
A decrease of the phosphorylation levels of CRMP-2ple long neurites and multiple Tau-1-positive neurites
at 6 DIV (Figures 5B and 5C). The consistent results was not observed during first 3 min. After the 10 min
stimulation, the phosphorylation levels of CRMP-2 werewere obtained by using other axonal markers, such as
synapsin I and synaptophysin for axon and MAP2 for decreased. The decrease was sustained until at least
90 min after the stimulation. BDNF had a similar effectdendrite (data not shown).
Next, hippocampal neurons were cotransfected with on CRMP-2 phosphorylation (Figure 6B). Treatment with
NT-3 and BDNF increased the phosphorylation levels ofHA-GSK-3␤wt, S9A (Ser-9 was replaced by Ala; consti-
tutively active form), or KD (Lys-85 was replaced by Akt at Ser-473 and GSK-3␤at Ser-9, whereas NGF had
no obvious effects on the phosphorylation levels ofMet; kinase dead mutant) and Myc-GST. The ectopic
expression of GSK-3␤wt and S9A inhibited axon elonga- CRMP-2, GSK-3␤, and Akt (Figure 6B). PI3-kinase inhibi-
tor (wortmannin) inhibited NT-3- and BDNF-inducedtion (Figures 5D and 5E) and branching (data not shown)
at 3 DIV, whereas GSK-3␤KD had no apparent effect. decrements of the phosphorylation levels of CRMP-2
and increments of the phosphorylation levels of GSK-3␤Most of the cells transfected with Myc-GST (about 75%)
displayed normal polarity with a single neurite, which and Akt. Because the phosphorylation of GSK-3␤at
Ser-9 is known to inactivate GSK-3␤, our findings sug-was stained with Tau-1 antibody, and some minor pro-
cesses (Nishimura et al., 2004). The remaining (about gest that NT-3 and BDNF decrease the phosphorylation
levels of CRMP-2 at Thr-514 and increase nonphosphor-25%) neurons expressing Myc-GST remained in stage
2 of development. The expression of GSK-3␤wt and ylated active CRMP-2 via the PI3-kinase/Akt/GSK-3␤
pathway.S9A increased the percentage of neurons that have no
Tau-1-positive neurites (Figures 5D and 5E). GSK-3␤To examine the effect of NT-3 on the spatial distribu-
tion of phosphorylated CRMP-2 at Thr-514, hippocam-S9A showed slightly higher activity than GSK-3␤wt in
this capacity. GSK-3␤KD had no obvious effect. The pal neurons were fixed at 3 DIV after the treatment of
NT-3 for 30 min and then immunostained with anti-pT514consistent results were obtained by using other axonal
markers, such as synapsin I and synaptophysin for axon (phosphorylated CRMP-2, red) and anti-CRMP-2 anti-
bodies (total CRMP-2, green; Figure 6C). The phosphor-(Supplemental Figure S3C). Further, the effect of GSK-3␤
S9A on axon formation was examined at 6 DIV, and the ylation levels of CRMP-2 at Thr-514 (red) were de-
creased not only in the axonal growth cone but also inconsistent results were obtained (Supplemental Figure
S3D). These results suggest that the overexpression of the shaft as compared with nontreated cells (Figures 2B
and 6C). In the cell body, the obvious decrease of theGSK-3␤impairs neuronal polarity, presumably by sup-
pressing neurite elongation. phosphorylation levels of CRMP-2 was not observed.
We then examined whether CRMP-2 is involved inNext, we investigated whether the expression of the
nonphosphorylated form of CRMP-2 compensates for NT-3- and BDNF-induced axon outgrowth and branching.
NT-3 and BDNF promoted axon elongation andthe GSK-3␤-induced defect. The expression of CRMP-2
wt or CRMP-2 T514A counteracted the effects of GSK-3␤branching as previously described (Tucker [2002]; Fig-
ures 6D and 6E). We used siRNA to directly test whetherS9A on axon outgrowth and neuronal polarity (Figure
5F), but CRMP-2 T514D had no obvious effects (data endogenous CRMP-2 is required for axon elongation
and branching by NT-3 and BDNF (Figure 6E). Expres-not shown). Taken together, these results suggest that
GSK-3␤regulates neuronal polarity through the phos- sion of CRMP-2 was markedly inhibited by CRMP-2
siRNA in 10%–20% cells as revealed by immunocyto-phorylation of CRMP-2, though we can not neglect the
possibility that GSK-3␣is involved in neuronal polarity. chemistry as described (Nishimura et al. [2003]; data
not shown). Knockdown of CRMP-2 caused a marked
inhibition of NT-3- and BDNF-induced axon elongation
NT-3 and BDNF Regulate CRMP-2 Phosphorylation
and branching, indicating that CRMP-2 is necessary for
via the PI3-Kinase/Akt/GSK-3␤Pathway
NT-3- and BNDF-induced axon outgrowth and branching.
NT-3 and BDNF but not NGF enhance axon elongation
Taken together, these results suggest that NT-3 de-
and branching in hippocampal neurons (Ip et al., 1993;
creases the phosphorylation levels of CRMP-2 and in-
Morfini et al., 1994; Labelle and Leclerc, 2000). GSK-3␤
creases nonphosphorylated active CRMP-2 to promote
is known to be constitutively active, and its activity can
axon outgrowth and branching.
be inhibited by treatment with NT-3 and BDNF (Huang
and Reichardt, 2003; Segal, 2003). This inhibitory mech-
anism is thought to be mediated by the PI3-kinase/Akt Discussion
pathway (Markus et al., 2002; Huang and Reichardt,
2003; Segal, 2003). Several groups, including ours, re- Phosphorylation of CRMP-2 by GSK-3␤
In the present study, we found that CRMP-2 was phos-ported that PI3-kinase inhibitors inhibited axon elonga-
tion (Shi et al., 2003; Menager at al., 2004). PI3-kinase phorylated at Thr-514 by GSK-3␤in vitro and in vivo
(Figures 1B, 1C, 2A, and 2B; Supplemental Figure S1C).activates Akt by the phosphorylation of Akt at Thr-308
and Ser-473 via phosphatidylinositol-3,4,5-triphosphate GSK-3␤alone did not phosphorylate CRMP-2 in vitro

Cell
144
Figure 5. GSK-3␤Regulates Neuronal Polarity via CRMP-2
(A) Neurons transfected with Myc-GST were cultured in the presence of GSK-3 inhibitors (2 mM LiCl, 5 ␮M SB216763 or 25 ␮M SB415286),
buffer, 2 mM NaCl, or DMSO for 48 hr before fixation (Aa). Hairpin siRNA construct for GSK-3␤or control vector was cotransfected into
neurons with Myc-GST (Ab). Neurons were fixed at 3 DIV and then immunostained with anti-myc antibody to measure axon length. n ⫽50
per experimental condition. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; *p ⬍
0.05; **p ⬍0.01).
(B) Neurons transfected with Myc-GST were cultured in the presence of GSK-3 inhibitor (5 ␮M SB216763) for 5 days before fixation (Ba).
Neurons were cotransfected with hairpin siRNA for GSK-3␤and Myc-GST (Bb). The neurons were fixed at 6 DIV and then immunostained
with anti-Myc and Tau-1 antibodies. The enlarged images of the neurites (1, 2) are shown. The neuron treated with GSK-3 inhibitor (5 ␮M

GSK-3␤and CRMP-2 in Neuronal Polarity
145
(Figure 1C). GSK-3␤requires a priming phosphate to phorylation levels of CRMP-2 at Ser-522 and thereby
promote growth cone collapse.phosphorylate its substrates. CRMP-2 has a consensus
motif recognized by Cdk5 at Ser-522 in the vicinity of
Thr-514 (Figure 1A). We found that treatment of Cdk Neuronal Polarity and GSK-3␤
Previously, we found that CRMP-2 is crucial for de-inhibitor (olomoucine) decreased the phosphorylation
levels of CRMP-2 in COS7 cells (data not shown) and termining the fate of the axon and dendrites, thereby
establishing and maintaining neuronal polarity (Inagakihippocampal neurons (Supplemental Figure S1B) and
that GSK-3␤phosphorylated CRMP-2 that was pre- et al., 2001). CRMP-2 appears to be critical for axon
formation by promoting elongation of one of the imma-phosphorylated by Cdk5 in vitro (Figure 1C). Therefore,
it is likely that Cdk5 phosphorylates CRMP-2 at Ser- ture neurites, which is the future axon and within which
CRMP-2 is enriched. CRMP-2 binds to tubulin hetero-522, whereby GSK-3␤phosphorylates CRMP-2 at Thr-
514 in vivo. Ser-518 is also a potential phosphorylation dimers and promotes microtubule assembly to enhance
axon growth and branching (Fukata et al., 2002). Re-site of CRMP-2 by GSK-3␤.
Phosphomimic CRMP-2 was not colocalized with the cently, it has been reported that the polarized distribu-
tions of PAR-3/PAR-6 and aPKC activity are importantmitotic spindle in vivo (Figure 3A, Supplemental Figure
S1D). Phosphorylated CRMP-2 at Thr-514 was not asso- for axon specification in hippocampal neurons and that
the PI3-kinase activity at the axon is required for polar-ciated with the mitotic spindle (Figure 3B). The binding
activity of phosphomimic CRMP-2 to tubulin was weaker ized localization of PAR-3 (Shi et al., 2003; Nishimura et
al., 2004). We recently reported that the local contactthan that of CRMP-2 wt in vitro (Figure 3C). These results
indicate that the phosphorylation of CRMP-2 at Thr-514 of the immature neurites with adhesion molecules such
as laminin induces the rapid production of PIP
3
at thelowers its binding activity to tubulin. Phosphorylated
CRMP-2 at Thr-514 was enriched in the growing axon, tip of the neurite through the action of PI3-kinase and
that PIP
3
is involved in axon specification, possibly bywhereas the phosphorylation levels of CRMP-2 were
low in the growth cone (Figure 2B), making it convenient stimulating elongation of an immature neurite (Menager
et al., 2004). Elongation of one of the immature neuritesfor CRMP-2 to copolymerize with tubulin dimers into
microtubules in the growth cone (Arimura et al., 2004). is necessary for axon specification (Bradke and Dotti,
2000). The accompanying paper (Jiang et al., 2005 [thisIn shafts, the phosphorylation of CRMP-2 at Thr-514
may prevent the copolymerization of CRMP-2 with tu- issue of Cell]) and we have now shown that the expres-
sion of constitutively active GSK-3␤suppresses axonbulin dimers into microtubules until reaching the growth
cone, or it may induce the dissociation of CRMP-2 from formation and that inhibition of GSK-3␤induces the for-
mation of multiple axons in hippocampal neurons (Fig-microtubules. It may be noted that the phosphorylation
of CRMP-2 at Ser-518 and Ser-522 is implicated in the ures 5B–5E; Supplemental Figures S3A, S3C, and S3D).
We also found that the inhibitory effect of GSK-3␤onformation of degenerating neurites in Alzheimer’s dis-
ease (Gu et al., 2000). Phosphorylated CRMP-2 was neuronal polarity is counteracted by nonphosphorylated
CRMP-2 (Figure 5F). Taken together, these results indi-identified as an antigen for 3F4 monoclonal antibody,
which was raised against partially purified paired helical cate that GSK-3␤regulates neuronal polarity through
the phosphorylation of CRMP-2.filaments and labeled neurofibrillary tangles and some
plaque neurites. The level of 3F4 antigen is increased in We propose the role of GSK-3␤and CRMP-2 in axon
specification as follows (Figure 7). The activation of PI3-the soluble fraction of the brains affected by Alzheimer’s
disease. These observations raise the possibility that kinase at the selective immature neurite produces PIP
3
,
thus activating Akt and recruiting the PAR-3/PAR-6/hyperphosphorylation of CRMP-2 is involved in the de-
velopment of neurofibrillary tangles and plaque neurites. aPKC complex at the growth cone. Activated Akt and
aPKC inhibit GSK-3␤by its phosphorylation, wherebyInhibition of GSK-3␤prevents semaphorin-3A (Sema3A)-
induced growth cone collapse (Eickholt et al., 2002). nonphosphorylated CRMP-2 is increased in the growth
cone. Nonphosphorylated active CRMP-2 promotes mi-Uchida et al. (2005) have recently found that Sema3A
induces the phosphorylation of CRMP-2 at Ser-522 by crotubule assembly and Numb-mediated endocytosis
of cell adhesion molecules to enhance elongation of theCdk5 followed by the phosphorylation at Thr-509 and
Ser-518. Thus, Sema3A appears to increase the phos- immature neurite for axon specification (Fukata et al.,
SB216763) or transfected with hairpin siRNA for GSK-3␤had multiple Tau-1-positive neurites (1, 2). Scale bar, 100 ␮m.
(C) The percentage of the cells that had multiple long neurites (Ca) and multiple Tau-1-positive neurites (Cb) were measured at 6 DIV neurons
treated with the indicated compounds or transfected with the indicated plasmids. n ⫽50 per experimental condition. Data are means ⫾SD
of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; *p ⬍0.05; **p ⬍0.01).
(D) Hippocampal neurons were cotransfected with HA-GSK-3␤S9A and Myc-GST. Neurons were fixed at 3 DIV and then immunostained with
anti-myc (Da and Dd) and axonal marker Tau-1 (Db and De) antibodies. The merged images (Dc and Df) are shown. Arrows indicate Tau-1-
positive neurites, and arrowheads denote Tau-1-negative neurites in the transfected cells. Scale bar, 50 ␮m.
(E) Hippocampal neurons were cotransfected with HA-GSK-3␤wt, S9A, or KD and Myc-GST. Neurons were fixed at 3 DIV and then immuno-
stained with anti-myc and axonal marker Tau-1 antibodies. The axon length (Ea) and the percentage of the cells that had no Tau-1-positive
neurites (Eb) were measured. Fifty cells for each plasmid were measured by tracing images of immunofluorescence staining with anti-Myc
antibody. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; **p ⬍0.01).
(F) Cotransfection of HA-GSK-3␤wt or S9A with CRMP-2 T514A was performed in hippocampal neurons. The axon length (Fa) and the
percentage of the cells that had no Tau-1-positive neurites (Fb) were measured at 3 DIV neurons. n ⫽50 per experimental condition. Data
are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; *p ⬍0.05; **p ⬍0.01).

Cell
146
Figure 6. NT-3 and BDNF Regulate CRMP-2 Phosphorylation via the PI3-Kinase/Akt/GSK-3␤Pathway
(A) 3 DIV hippocampal neurons were stimulated with NT-3 for 3, 10, 30, or 90 min after neurons were cultured in neurobasal medium for 2 hr.
The cell lysates were resolved by SDS-PAGE and immunoblotted with anti-pT514 and anti-CRMP-2 polyclonal antibodies. The relative levels
of CRMP-2 phosphorylation at Thr-514 were calculated with those of untreated control cells.
(B) Hippocampal neurons were treated with BDNF, NT-3, or NGF for 30 min after neurons were cultured in neurobasal medium for 2 hr.
Neurons were treated with PI3-kinase inhibitor (100 nM wortmannin) or DMSO during the last 2 hr. Immunoblot analyses were performed with
anti-pT514, anti-CRMP-2 polyclonal, anti-phospho-GSK-3␤(Ser-9), anti-GSK-3␤, anti-phospho-Akt (Ser-473), and anti-Akt antibodies. The
asterisk shows the mobility shift of CRMP-2 induced by the phosphorylation of CRMP-2 at Thr-514.
(C) NT-3 decreased the phosphorylation levels of CRMP-2 at Thr-514 in the axonal growth cone and the shaft. Hippocampal neurons were
fixed at 3 DIV after the treatment of NT-3 for 30 min and then immunostained with anti-pT514 (Ca) and anti-CRMP-2 (Cb) antibodies. The
merged (Cc) is shown. The fluorescence intensities of 30 cells were measured in growth cones (Cd), shafts (Ce), and cell bodies (Cf). Scale
bar, 50 ␮m. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance (Student’s t test; *p ⬍0.05; **p ⬍0.01).

GSK-3␤and CRMP-2 in Neuronal Polarity
147
phatase activity toward CRMP-2 to accelerate the de-
phosphorylation of CRMP-2. Further studies are neces-
sary to address these issues.
One can expect that overdosage of NT-3 or BDNF
induces the formation of multiple axons. However, we
found that this was not the case. One possible explana-
tion is that NT-3 and BDNF receptors (TrkC and TrkB)
may be downregulated or desensitized during the action
of NT-3 and BDNF (Frank et al., 1996). Prolonged inhibi-
tion of GSK-3␤may be necessary for multiple axon for-
mation.
GSK-3␤phosphorylates MAP1B and the adenoma-
tous polyposis coil gene product (APC; Grimes and Jope
[2001]; Frame and Cohen [2001]). The phosphorylation
of MAP1B by GSK-3␤suppresses detyrosination of mi-
crotubules and decreases the numbers of stable micro-
tubules (Goold et al., 1999; Gordon-Weeks and Fischer,
2000). It is possible that NT-3 and BDNF inhibit GSK-3␤
and thereby increase the stability of microtubules to
enhance axon outgrowth through the dephosphoryla-
tion of MAP1B. The binding of APC to microtubules
Figure 7. Model Schema to Regulate the Phosphorylation of
CRMP-2 by GSK-3␤
increases microtubule stability, and the interaction of
APC with microtubules is decreased by the phosphory-
NT-3, BDNF, and adhesion molecules are thought to activate PI3-
kinase, thereby producing PIP
3
. PIP
3
activates Akt via PDK. Acti-
lation of APC by GSK-3␤(Zumbrunn et al., 2001). Inacti-
vated Akt phosphorylates and inactivates GSK-3␤. The binding ac-
vation of GSK-3␤by NT-3 and BDNF may induce the
tivity of CRMP-2 to tubulin is decreased by the phosphorylation by
interaction of APC with microtubules, thereby stabilizing
GSK-3␤. Nonphosphorylated CRMP-2 binds to tubulin heterodimers
microtubules. Thus, GSK-3␤appears to regulate dy-
to promote microtubule assembly, thereby enhancing axon elonga-
namics of microtubules through the phosphorylation of
tion and branching.
specific microtubule-associated proteins.
2002; Nishimura et al., 2003). The relation between Akt
Experimental Procedures
and PAR-3/PAR-6/aPKC remains to be clarified.
Materials and Chemicals
Inhibition of CRMP-2 Phosphorylation by NT-3
cDNA encoding human CRMP-2 was obtained using the methods
and BDNF
of Arimura et al. (2000). pCAGGS vector was provided by Dr. M.
Nakafuku (Cincinnati Children’s Hospital Medical Center, Cincinnati,
We demonstrated that the phosphorylation of CRMP-2
OH). Hairpin siRNA for GSK-3␤construct, pU6-GSK-3␤HP2, was
was suppressed by NT-3 and BDNF (Figures 6A–6C).
provided by Dr. David L. Turner (University of Michigan, MI; Yu
Knockdown of CRMP-2 with siRNA inhibited NT-3- and
et al. [2003]). The following antibodies were used: anti-CRMP-2
BDNF-induced axon outgrowth and branching (Figure
monoclonal antibody (C4G), kindly provided by Dr. Y. Ihara (Univer-
6E). These results suggest that NT-3 and BDNF de-
sity of Tokyo, Tokyo, Japan); anti-CRMP-2 polyclonal antibody
crease the phosphorylation levels of CRMP-2 and pro-
raised against MBP-CRMP-2; monoclonal anti-␣-tubulin (DM1A,
Sigma, St. Louis, MO); monoclonal anti-unique ␤-tubulin (TUJ1,
mote axon outgrowth and branching via nonphosphory-
Berkeley Antibody Company, Richmond, CA); polyclonal anti-c-Myc
lated CRMP-2. NT-3 and BDNF are thought to activate
(A-14, Santa Cruz Biotechnology, Inc., Santa Cruz, CA); anti-GSK-3␤
PI3-kinase, thereby producing PIP
3
. PIP
3
activates Akt
(Transduction Laboratories, Lexington, KY); anti-phospho-GSK-3␤
via PDK. Activated Akt phosphorylates and inactivates
(Ser-9; Cell Signaling Technology Inc., Beverly, MA); anti-Akt (New
GSK-3␤. Treatment of hippocampal neurons with NT-3
England BioLabs, Beverly, MA); anti-phospho-Akt (Ser-473) (New
and BDNF resulted in the activation of Akt and the inacti-
England BioLabs); monoclonal Tau-1 (Chemicon, Temecula, CA);
vation of GSK-3␤(Figure 6B). Activated Akt and inacti-
anti-synapsin I (Calbiochem, San Diego, CA); anti-synaptophysin
(Chemicon); anti-MAP2 (Sigma); and anti-actin (Chemicon) antibod-
vated GSK-3␤are localized in the growth cone of axons
ies. Recombinant human NT-3 and BDNF were purchased from
(Eickholt et al., 2002). These findings suggest that NT-3
PeproTech EC LTD (London, UK). NGF was from Upstate Biotech-
and BDNF inhibit GSK-3␤via the PI3-kinase/Akt path-
nology, Inc. (Charlottesville, VA). Recombinant His-tagged Cdk5 and
way, thereby decreasing the phosphorylation levels of
GST-tagged p35 were from Upstate. GSK-3 inhibitors (SB216763 and
CRMP-2 at Thr-514 (Figure 7). It is also possible that
SB415286) were from Tocris Cookson, Inc. (Ellisville, MO). Cdk inhib-
NT-3 and BDNF inhibit GSK-3␤via the PAR-3/PAR-6/
itor (olomoucine) was from Sigma. PI3-kinase inhibitor (wortmannin)
was from Wako (Osaka, Japan).
aPKC pathway, and NT-3 and BDNF increase the phos-
(D) Hippocampal neurons were transfected with Myc-GST and then treated with NT-3 or BDNF. Theywere fixed at 3 DIV and then immunostained
with anti-Myc antibody. Scale bar, 100 ␮m.
(E) Hippocampal neurons were cotransfected with CRMP-2 siRNA and Myc-GST and then treated with NT-3 or BDNF. They were fixed at 3
DIV and then immunostained with anti-Myc antibody. Axon length (Ea) and the number of branch tips per axon (Eb) were measured at 3 DIV
neurons. n ⫽50 per experimental condition. Data are means ⫾SD of triplicate determinations. Asterisks indicate statistical significance
(Student’s t test; **p ⬍0.01).

Cell
148
Plasmid Constructs plemented with B-27 supplement (Invitrogen) and 1 mM glutamine.
Neurons were seeded on coverslips with PDL only for 6 DIV toCRMP-2 mutants were generated with a site-detected mutagenesis
kit (Stratagene, La Jolla, CA). The cDNA fragments encoding visualize secondary axons. Neurons were transfected using a cal-
cium phosphate method before plating to analyze the morphologyCRMP-2 wt and mutants were subcloned into pRSET-C1 (Invitrogen,
Carlsbad, CA) to obtain the construct of CRMP-2 tagged with histi- (Nishimura et al., 2003; Menager et al., 2004). Neurons were fixed
at 3 DIV or 6 DIV with 3.7% formaldehyde in PBS for 10 min at roomdine (His) at the N terminus of protein, pB-GEX (rearranged vector
from pGEX; Amersham Pharmacia Biotech, Buckinghamshire, UK) temperature, followed by treatment for 10 min with 0.05% Triton
X-100 on ice and 10% NGS in PBS for 1 hr at room temperature.to obtain the construct of CRMP-2 tagged with GST atthe C terminus
of protein, pCAGGS-myc, and pEGFP-C1 (Clontech, Palo Alto, CA) Neurons were then immunostained with indicated antibodies and
observed with a confocal laser microscope (LSM510 Carl Zeiss,vectors, respectively. pCGN-GSK-3␤, pCGN-GSK-3␤S9A, and
pCGN-GSK-3␤kinase dead (KD; K85N) and pGEX-2T-GSK-3␤were Oberkochen, Germany) built around an Axiovert 100 M (Carl Zeiss).
The length of a longest neurite was measured as that of an axon atconstructed as described by Tanji et al. (2002). All fragments were
confirmed by DNA sequencing. 3 DIV. The percentage of neurons with the second longest neurite
whose length was longer than a half-length of the longest neurite
was counted as that of multiple long neurites. Neurotrophins were
Protein Purification and Preparation of Anti-pT514 Antibody
added to the medium at a concentration of 100 ng/ml after the
GST- and His-tagged proteins were purified from E. coli on glutathi-
transfection. Neurons were cultured for 2 days and were then grown
one-Sepharose 4B beads (Amersham) and Ni-NTA agarose (Qiagen,
with fresh neurotrophins (100 ng/ml) for 1 day. For some experi-
Hilden, Germany) according to the manufacturers’ protocols. Rabbit
ments, neurons were treated with 10% (w/v) TCA. The resulting
polyclonal antibody against CRMP-2 phosphorylated at Thr-514
precipitates were subjected to SDS-PAGE and immunoblot analysis.
(anti-pT514 antibody) was raised as described by Amano et al.
(2003). As the antigen, the phosphopeptide Gly-Cys- Thr
509
-Pro-Lys-
Thr-Val-phosphoThr
514
-Pro-Ala-Ser-Ser-Ala
519
for CRMP-2 was Acknowledgments
chemically synthesized by Biologica Co. (Aichi, Japan). The antise-
rum obtained was then affinity purified against the respective phos- We thank Drs. Y. Ihara (University of Tokyo, Tokyo, Japan) and David
phopeptide. L. Turner (University of Michigan, MI) for their kind gifts of materials;
Drs. Y. Rao (Washington University, MO), Y. Goshima (Yokohama
City University, Kanagawa, Japan), M. Amano, M. Fukata, Y. Fukata,
Culture of COS7 and Rat 3Y1 Cells for Immunoblot Analysis
S. Taya, C. Menager, T. Nishimura, Miss K. Fujii, and Mr. A. Hattori
COS7 and rat 3Y1 cells were seeded on a 60 mm dish in Dulbecco’s
for helpful discussion; Miss K. Yamada for preparing some materials
modified Eagle’s medium (DMEM) with 10% fetal bovine serum
and technical assistance; and Mrs. T. Ishii for secretarial assistance.
(FBS) and cultured overnight at 37⬚C in an air/5% CO
2
atmosphere
This research was supported in part by grants-in-aid for scientific
at constant humidity. Transfections were carried out using lipofec-
research from the Ministry of Education, Culture, Sports, Science
tamine reagent (Invitrogen) according to the manufacturer’s proto-
and Technology of Japan (MEXT); grant-in-aid for Creative Scientific
col. Cells were grown in DMEM with 10% FBS for 1 day and then
Research from MEXT; The 21st Century Centre of Excellence (COE)
in DMEM for 1 day. Cells were treated with 10% (w/v) trichloroacetic
Program from MEXT; special coordination funds for promoting Sci-
acid (TCA). The resulting precipitates were subjected to SDS-PAGE
ence and Technology (SCFPST); and the Pharmaceuticals and Medi-
and immunoblot analysis.
cal Devices Agency (PMDA).
Culture of HeLa Cells for Immunofluorescence Analysis
HeLa cells were seeded on coverslips in DMEM with 10% FBS and Received: April 12, 2004
cultured overnight at 37⬚C in an air/5% CO
2
atmosphere at constant Revised: September 4, 2004
humidity. Transfections were carried out using lipofectamine re- Accepted: November 2, 2004
agent. After 2 days’ culture, HeLa cells were fixed with 3.7% formal- Published: January 13, 2005
dehyde in phosphate-buffered saline (PBS) for 10 min at room tem-
perature followed by treatment for 10 min with 0.2% Triton X-100 References
on ice and 10% normal goat serum (NGS) in PBS for 1 hr at room tem-
perature. Amano, M., Kaneko, T., Maeda, A., Nakayama, M., Ito, M., Yamauchi,
T., Goto, H., Fukata, Y., Oshiro, N., Shinohara, A., et al. (2003).
Kinase Assay Identification of Tau and MAP2 as novel substrates of Rho-kinase
The kinase reaction for GST-GSK-3␤was carried out in 50 ␮lof and myosin phosphatase. J. Neurochem. 87, 780–790.
kinase buffer (50 mM Tris-HCl [pH 7.5], 0.7 mM EDTA, 12 mM MgCl
2
,Arimura, N., Inagaki, N., Chihara, K., Menager, C., Nakamura, N.,
1 mM dithiothreitol, 100 nM calyculin A) containing 200 ␮M ATP, Amano, M., Iwamatsu, A., Goshima, Y., and Kaibuchi, K. (2000).
recombinant kinases (4.7 ␮M GST-GSK-3␤; 66 nM His-tagged Cdk5 Phosphorylation of collapsin response mediator protein-2 by Rho-
with GST-tagged p35), and substrates (12 ␮M His-tagged CRMP-2). kinase: evidence for two separate signaling pathways for growth
After incubation for 1 hr at 30⬚C, the reaction mixtures were boiled cone collapse. J. Biol. Chem. 275, 23973–23980.
in SDS sample buffer and subjected to SDS-PAGE and immu- Arimura, N., Menager, C., Fukata, Y., and Kaibuchi, K. (2004). Role
noblot analysis. of CRMP-2 in neuronal polarity. J. Neurobiol. 58, 34–47.
Baas, P.W. (1997). Microtubules and axonal growth. Curr. Opin. Cell
In Vitro Binding Assay
Biol. 9, 29–36.
GST alone, CRMP-2 wt-GST, or CRMP-2 T514D-GST was separately
immobilized onto glutathione-Sepharose 4B beads (Amersham). The Bradke, F., and Dotti, C.G. (2000). Establishment of neuronal polarity:
immobilized beads were incubated with porcine brain lysate in buffer lessons from cultured hippocampal neurons. Curr. Opin. Neurobiol.
A (20 mM Tris-HCl, 1 mM EDTA, 50 mM NaCl, 1 mM dithiothreitol, 10, 574–581.
0.1% NP-40 [pH 7.5]) for 1 hr at 4⬚C. The beads were washed six Bresnick, A.R., Wolff-Long, V.L., Baumann, O., and Pollard, T.D.
times with buffer A containing 150 mM NaCl, and the proteins bound (1995). Phosphorylation on threonine-18 of the regulatory light chain
to the washed beads were eluted by the addition of buffer A con- dissociates the ATPase and motor properties of smooth muscle
taining 10 mM glutathione. myosin II. Biochemistry 34, 12576–12583.
Brown, A., Slaughter, T., and Black, M.M. (1992). Newly assembled
Culture of Hippocampal Neurons microtubules are concentrated in the proximal and distal regions of
Culture of hippocampal neurons prepared from E18 rat embryos growing axons. J. Cell Biol. 119, 867–882.
using papain was performed as described by Inagaki et al. (2001).
Craig, A.M., and Banker, G. (1994). Neuronal polarity. Annu. Rev.
Neurons were seeded on coverslips or dishes with poly-D-lysine
Neurosci. 17, 267–310.
(PDL; Sigma) and laminin (Iwaki, Tokyo, Japan) for 3 DIV to estimate
axon length and branching in neurobasal medium (Invitrogen) sup- Dotti, C.G., Sullivan, C.A., and Banker, G.A. (1988). The establish-

GSK-3␤and CRMP-2 in Neuronal Polarity
149
ment of polarity by hippocampal neurons in culture. J. Neurosci. Markus, A., Patel, T.D., and Snider, W.D. (2002). Neurotrophic factors
and axonal growth. Curr. Opin. Neurobiol. 12, 523–531.8, 1454–1468.
Eickholt, B.J., Walsh, F.S., and Doherty, P. (2002). An inactive pool Menager, C., Arimura, N., Fukata, Y., and Kaibuchi, K. (2004). PIP3
is involved in neuronal polarization and axon formation. J. Neuro-of GSK-3 at the leading edge of growth cones is implicated in Sema-
phorin 3A signaling. J. Cell Biol. 157, 211–217. chem. 89, 109–118.
Morfini, G., DiTella, M.C., Feiguin, F., Carri, N., and Caceres, A.Eldar-Finkelman, H. (2002). Glycogen synthase kinase 3: an emerg-
ing therapeutic target. Trends Mol. Med. 8, 126–132. (1994). Neurotrophin-3 enhances neurite outgrowth in cultured hip-
pocampal pyramidal neurons. J. Neurosci. Res. 39, 219–232.
Etienne-Manneville, S., and Hall, A. (2003a). Cdc42 regulates
GSK-3␤and adenomatous polyposis coli to control cell polarity. Munoz-Montano, J.R., Lim, F., Moreno, F.J., Avila, J., and Diaz-
Nido, J. (1999). Glycogen synthase kinase-3 modulates neurite out-Nature 421, 753–756.
growth in cultured neurons: possible implications for neurite pathol-
Etienne-Manneville, S., and Hall, A. (2003b). Cell polarity: Par6, aPKC
ogy in Alzheimer’s disease. J. Alzheimers Dis. 1, 361–378.
and cytoskeletal crosstalk. Curr. Opin. Cell Biol. 15, 67–72.
Nishimura, T., Fukata, Y., Kato, K., Yamaguchi, T., Matsuura, Y.,
Fletcher, T.L., Cameron, P., De Camilli, P., and Banker, G. (1991).
Kamiguchi, H., and Kaibuchi, K. (2003). CRMP-2 regulates polarized
The distribution of synapsin I and synaptophysin in hippocampal
Numb-mediated endocytosis for axon growth. Nat. Cell Biol. 5,
neurons developing in culture. J. Neurosci. 11, 1617–1626.
819–826.
Frame, S., and Cohen, P. (2001). GSK3 takes centre stage more
Nishimura, T., Kato, K., Yamaguchi, T., Fukata, F., Ohno, S., and
than 20 years after its discovery. Biochem. J. 359, 1–16.
Kaibuchi, K. (2004). Role of the PAR-3/KIF3 complex in the establish-
Frank, L., Ventimiglia, R., Anderson, K., Lindsay, R.M., and Rudge, ment of neuronal polarity. Nat. Cell Biol. 6, 328–334.
J.S. (1996). BDNF down-regulates neurotrophin responsiveness,
Scheid, M.P., and Woodgett, J.R. (2001). PKB/AKT: functional in-
TrkB protein and TrkB mRNA levels in cultured rat hippocampal
sights from genetic models. Nat. Rev. Mol. Cell Biol. 2, 760–768.
neurons. Eur. J. Neurosci. 8, 1220–1230.
Segal, R.A. (2003). Selectivity in neurotrophin signaling: theme and
Fukata, Y., Itoh, T.J., Kimura, T., Menager, C., Nishimura, T., Shiro-
variations. Annu. Rev. Neurosci. 26, 299–330.
mizu, T., Watanabe, H., Inagaki, N., Iwamatsu, A., Hotani, H., and
Kaibuchi, K. (2002). CRMP-2 binds to tubulin heterodimers to pro- Shi, S.H., Jan, L.Y., and Jan, Y.N. (2003). Hippocampal neuronal
polarity specified by spatially localized mPar3/mPar6 and PImote microtubule assembly. Nat. Cell Biol. 4, 583–591.
3-kinase activity. Cell 112, 63–75.
Goold, R.G., Owen, R., and Gordon-Weeks, P.R. (1999). Glycogen
synthase kinase 3␤phosphorylation of microtubule-associated pro- Sweeney, H.L., Yang, Z., Zhi, G., Stull, J.T., and Trybus, K.M. (1994).
Charge replacement near the phosphorylatable serine of the myosintein 1B regulates the stability of microtubules in growth cones. J.
Cell Sci. 112, 3373–3384. regulatory light chain mimics aspects of phosphorylation. Proc. Natl.
Acad. Sci. USA 91, 1490–1494.
Gordon-Weeks, P.R., and Fischer, I. (2000). MAP1B expression and
microtubule stability in growing and regenerating axons. Microsc. Tanji, C., Yamamoto, H., Yorioka, N., Kohno, N., Kikuchi, K., and
Kikuchi, A. (2002). A-kinase anchoring protein AKAP220 binds to
Res. Tech. 48, 63–74.
glycogen synthase kinase-3␤(GSK-3␤) and mediates protein kinase
Goshima, Y., Nakamura, F., Strittmatter, P., and Strittmatter, S.M.
A-dependent inhibition of GSK-3␤. J. Biol. Chem. 277, 36955–36961.
(1995). Collapsin-induced growth-cone collapse mediated by an in-
tracellular protein related to UNC-33. Nature 376, 509–514. Tucker, K.L. (2002). Neurotrophins and the control of axonal out-
growth. Panminerva Med. 44, 325–333.
Grimes, C.A., and Jope, R.S. (2001). The multifaceted roles of glyco-
gen synthase kinase 3␤in cellular signaling. Prog. Neurobiol. 65, Uchida, Y., Ohshima, T., Sasaki, Y., Suzuki, H., Yanai, S., Yamashita,
N., Nakamura, F., Takei, K., Ihara, Y., Mikoshiba, K., et al. (2005).391–426.
Semaphorin-3A signaling is mediated via sequential Cdk5 and
Gu, Y., Hamajima, N., and Ihara, Y. (2000). Neurofibrillary tangle-
GSK3␤phosphorylation of CRMP2: implication of common phos-
associated collapsin response mediator protein-2 (CRMP-2) is
phorylating mechanism underlying axon guidance and Alzheimer’s
highly phosphorylated on Thr-509, Ser-518, and Ser-522. Biochem-
disease. Genes Cells, in press.
istry 39, 4267–4275.
Yu, J.Y., Taylor, J., DeRuiter, S.L., Vojtek, A.B., and Turner, D.L.
Hedgecock, E.M., Culotti, J.G., Thomson, J.N., and Perkins, L.A.
(2003). Simultaneous inhibition of GSK3␣and GSK3␤using hairpin
(1985). Axonal guidance mutants of Caenorhabditis elegans identi-
siRNA expression vectors. Mol. Ther. 7, 228–236.
fied by filling sensory neurons with fluorescein dyes. Dev. Biol.
111, 158–170. Zumbrunn, J., Kinoshita, K., Hyman, A.A., and Nathke, I.S. (2001).
Binding of the adenomatous polyposis coli protein to microtubules
Huang, E.J., and Reichardt, L.F. (2003). Trk receptors: roles in neu-
increases microtubule stability and is regulated by GSK3␤phos-
ronal signal transduction. Annu. Rev. Biochem. 72, 609–642.
phorylation. Curr. Biol. 11, 44–49.
Inagaki, N., Chihara, K., Arimura, N., Menager, C., Kawano, Y., Mat-
suo, N., Nishimura, T., Amano, M., and Kaibuchi, K. (2001). CRMP-2
induces axons in cultured hippocampal neurons. Nat. Neurosci. 4,
781–782.
Ip, N.Y., Li, Y., Yancopoulos, G.D., and Lindsay, R.M. (1993). Cultured
hippocampal neurons show responses to BDNF, NT-3, and NT-4,
but not NGF. J. Neurosci. 13, 3394–3405.
Jiang, H., Guo, W., Liang, X., and Rao, Y. (2005). A critical role for
glycogen synthase kinase-3␤in determining axon-dendrite polarity
of neurons. Cell 120, this issue, 123–135.
Jones, D.M., Tucker, B.A., Rahimtula, M., and Mearow, K.M. (2003).
The synergistic effects of NGF and IGF-1 on neurite growth in adult
sensory neurons: convergence on the PI 3-kinase signaling pathway.
J. Neurochem. 86, 1116–1128.
Kamisoyama, H., Araki, Y., and Ikebe, M. (1994). Mutagenesis of the
phosphorylation site (serine 19) of smooth muscle myosin regulatory
light chain and its effects on the properties of myosin. Biochemistry
33, 840–847.
Labelle, C., and Leclerc, N. (2000). Exogenous BDNF, NT-3 and NT-4
differentially regulate neurite outgrowth in cultured hippocampal
neurons. Brain Res. Dev. Brain Res. 123, 1–11.