Tài liệu Báo cáo khoa học: Interaction between very-KIND Ras guanine exchange factor and microtubule-associated protein 2, and its role in dendrite growth – structure and function of the second kinase noncatalytic C-lobe domain docx

The binding core modules critical for the v-KIND–MAP2 interaction were defined within 32 residues of the mouse v-KIND KIND2 and 43 residues of the mouse MAP2 central domain.. Results The

factor and microtubule-associated protein 2, and its role

in dendrite growth – structure and function of the second kinase noncatalytic C-lobe domain

Jinhong Huang1,*, Asako Furuya1, Kanehiro Hayashi1–3and Teiichi Furuichi1,2,4

1 Laboratory for Molecular Neurogenesis, RIKEN Brain Science Institute, Saitama, Japan

2 JST, CREST, Kawaguchi, Saitama, Japan

3 Research Institute of Pharmaceutical Sciences, Musashino University, Tokyo, Japan

4 Faculty of Science and Technology, Tokyo University of Science, Chiba, Japan

Keywords

dendrite growth; KIND domain; MAP2;

protein–protein interaction; RasGEF

Correspondence

T Furuichi, Laboratory for Molecular

Neurogenesis, RIKEN Brain Science

Institute, 2-1 Hirosawa, Wako 351-0198,

Japan

Fax: +81 48 467 6079

Tel: +81 48 467 5906

E-mail: tfuruichi@brain.riken.jp

*Present address

Discovery & Development Laboratory I,

Hanno Research Center, Taiho

Pharmaceutical Co., Ltd, Saitama, Japan

(Received 5 January 2011, revised 19

February 2011, accepted 28 February

2011)

doi:10.1111/j.1742-4658.2011.08085.x

The kinase noncatalytic C-lobe domain (KIND) is a putative protein–protein interaction module Four KIND-containing proteins, Spir-2 (actin-nuclear factor), PTPN13 (protein tyrosine phosphatase), FRMPD2 (scaffold protein) and very-KIND (v-KIND) (brain-specific Ras guanine nucleotide exchange factor), have been identified to date Uniquely, v-KIND has two KINDs (i.e KIND1 and KIND2), whereas the other three proteins have only one The functional role of KIND, however, remains unclear We previously demon-strated that v-KIND interacts with the high-molecular weight microtubule-associated protein 2 (MAP2), a dendritic microtubule-microtubule-associated protein, leading to negative regulation of neuronal dendrite growth In the present study, we analyzed the structure–function relationships of the v-KIND– MAP2 interaction by generating a series of mutant constructs The interac-tion with endogenous MAP2 in mouse cerebellar granule cells was specific to v-KIND KIND2, but not KIND1, and was not observed for the KINDs from other KIND-containing proteins The binding core modules critical for the v-KIND–MAP2 interaction were defined within 32 residues of the mouse v-KIND KIND2 and 43 residues of the mouse MAP2 central domain Three Leu residues at amino acid positions 461, 474 and 477 in the MAP2-binding core module of KIND2 contributed to the interaction The MAP2-binding core module itself promoted dendrite branching as a dominant-negative regu-lator of v-KIND in hippocampal neurons The results reported in the present study demonstrate the structural and functional determinant underlying the v-KIND–MAP2 interaction that controls dendrite arborization patterns Structured digital abstract

l vKIND-KIND2 binds to Map2 by pull down (View interaction)

l vKIND-KIND2 physically interacts with Map2 by pull down (View interaction)

Abbreviations

CD, central domain; DIV, day in vitro; EGFP, enhanced green fluorescent protein; GST, glutathione S-transferase; HMW, high-molecular-weight; KIND, kinase noncatalytic C-lobe domain; KIND1, first kinase noncatalytic C-lobe domain; KIND2, second kinase noncatalytic C-lobe domain; MAP2, microtubule-associated protein 2; GEF, guanine exchange factor; v-KIND, very-KIND.

Protein–protein interactions play important roles in

the molecular recognition and functional modulation

between proteins in many signal transduction

path-ways [1,2] The kinase noncatalytic C-lobe domain

(KIND) was determined to be a putative signaling

domain based on bioinformatic analysis of the

N-ter-minal sequence of the Drosophila protein Spir, an

actin-nucleation factor [3] The KIND domain shows

homology to the C-terminal protein kinase catalytic

fold (C-lobe), although it lacks the sequence similarity

critical for kinase activity [3] Four proteins containing

KIND domains have so far been identified in

mam-mals: Spir [4,5], nonreceptor-type protein tyrosine

phosphatase 13 (PTPN13, or PTP-BL⁄ PTP-BAS) [6,7],

FERM and PDZ-domain-containing 2 (FRMPD2) [8]

and Ras guanine exchange factor (RasGEF)

very-KIND (v-KIND, or kinase noncatalytic C-lobe

domain containing 1) [9] The KIND domain in these

proteins is localized to the N-terminal region, and their

specific functional domains are located in the

C-termi-nal region The C-lobe of protein kinases mediates the

interaction with activators, substrates and regulatory

subunits, implying that the KIND domain, an atypical

noncatalytic C-lobe, is involved in the interaction with

signaling proteins [3,5,10] However, the structural and

functional properties of the KIND domains remain

largely unknown

Within the KIND protein family, v-KIND is unique

because it possesses two tandem-repeated KIND

domains, KIND1 and KIND2, in the N-terminal

region Recently, a heterozygous, nonsynonymous

somatic single nucleotide variation of human v-KIND

(KNDC1), in which Leu799 is changed to a Phe

resi-due, was reported in acute myeloid leukemia genomes

[11] We previously identified v-KIND with a

character-istic spatiotemporal expression pattern during the

post-natal development of mouse brain: it shows low- or

moderate-level expression in the cerebrum,

hippocam-pus and thalamus in the first week after birth, whereas

its highest expression level occurs in cerebellar granule

cells of the internal granular layer by postnatal week 2

and thereafter [12] We showed that v-KIND

overex-pression suppresses and v-KIND knockdown promotes

dendrite growth of cultured cerebellar granule cells

and hippocampal neurons, suggesting that v-KIND

acts as a signaling molecule in controlling or limiting

dendrite growth of neurons during development [12]

We also suggested that the protein–protein interaction

between v-KIND and the high-molecular-weight

(HMW) form, but not the low-molecular-weight form,

of microtubule-associated protein 2 (MAP2) via

KIND2 is critical for this signaling pathway [12] HMW-MAP2 (referred to hereafter as MAP2) is known to modulate polymerization, stability and rear-rangement of microtubules in neuronal dendrites [13–15] and is associated with some neurological and psychiatric disorders [16,17] However, the structure– function relationship of the interaction between v-KIND and MAP2, as well as its biological significance, remains unclear

In the present study, we determined the structural and functional properties of the protein–protein inter-action between v-KIND and MAP2 We defined the binding core regions for the v-KIND–MAP2 interac-tion and showed that the MAP2 binding core is not only critical for targeting of v-KIND to neuronal den-drites, but also is indispensable for the function of v-KIND in negatively controlling dendrite growth and branching

Results The KIND2 domain of v-KIND has a unique ability

to localize to dendrites via MAP2 binding, which is absent in the KINDs from other KIND-containing proteins

To examine the dendrite localization signal domains in v-KIND, we first investigated the subcellular localiza-tion of eight different Flag epitope-tagged v-KIND derivatives with domain deletions (as shown inFig 1A)

in primary cultured mouse cerebellar granule cells, coexpressed with enhanced green fluorescent protein (EGFP) to visualize the protrusion patterns of

transfect-ed neurons As shown in Fig 1B, the full-length v-KIND was specifically localized to dendrites and soma, although not to axons The expression of three other KIND2-containing constructs (DKIND1, DRasN and DGEF) was restricted to dendrites and soma (Fig 1B) On the other hand, two KIND2 domain-lacking constructs, DKIND2 (Fig 1B) and DKIND1 + 2 (Fig 1B), were widely distributed throughout the cells, including the axons Notably, the KIND2 domain has a specific ability to localize to dendrites by itself, whereas the KIND1 domain alone has lost this ability (Fig 1B) The results obtained indicate that the KIND2 domain

is necessary and sufficient for the targeting of v-KIND

to dendrites of neurons

To investigate the MAP2-binding characteristics of the KIND domain protein family, we generated four glutathione S-transferase (GST)-fused proteins of KIND domains (v-KIND KIND1, v-KIND KIND2,

Spir-2 KIND and PTPN13 KIND) (Fig 2B) and

ana-lyzed their ability to interact with MAP2 protein

(Fig 2A) Only GST-fused v-KIND KIND2 was able

to pull down MAP2 from mouse cerebellar lysates

(Fig 2A, left), as well as purified MAP2 protein

sam-ples (Fig 2A, right) Taken together, these data

indi-cate that the direct interaction with the dendritic

MAP2 protein is a unique feature of v-KIND KIND2,

among the four KIND domains tested

Residues 702–745 in the central domain of MAP2

contain the v-KIND binding core module

MAP2 consists of three main structural domains: the

cAMP-dependent protein kinase regulatory subunit

RII binding domain, the central domain (CD) and the microtubule-binding domain (Fig 3A) To deter-mine the v-KIND-binding region in MAP2, we first divided MAP2 into five regions (Fig 3A) and ana-lyzed the bacterially expressed proteins of these con-structs (Fig S1A) for binding to v-KIND in cerebellar lysates by a pull-down assay Only the GST-fused CD2 region (residues 600–1099) could pull down endogenous v-KIND protein (Fig 3B) These results indicate that the region spanning amino acids 600–1099 of MAP2 (i.e around the middle part of the CD) interacts with the endogenous v-KIND in mouse cerebellum

To verify whether the CD2 region of MAP2 specifi-cally binds to KIND2 in v-KIND, we screened the

v-KIND ΔKIND1 ΔKIND2 ΔKIND1+2 ΔRasN ΔGEF KIND1 KIND2

KIND1 KIND2 CC RasN GEF

1 37–217 456–620 1238–1364 1461–1734 1742

ΔKIND2-Flag v-KIND-Flag

dendrite

axon

KIND1-Flag ΔRasN-Flag

KIND2-Flag ΔGEF-Flag

A

B

Fig 1 Domain structure of the

MAP2-asso-ciated RasGEF v-KIND and its dendritic

tar-geting via KIND2 domain (A) Structures of

the v-KIND KIND1, KIND2, coiled-coil (CC),

RasN and RasGEF domains Flag-tagged

v-KIND derivatives: v-KIND, full-length

v-KIND; DKIND1, deletion of KIND1;

DKIND2, deletion of KIND2; DKIND1 + 2,

deletion of both KIND1 and KIND2; DRasN,

deletion of RasN; DGEF, deletion of

RasGEF; KIND1, KIND1 domain; KIND2,

KIND2 domain (B) KIND2 domain anchors

v-KIND to dendrites Flag-tagged v-KIND,

DKIND1, DKIND2, DKIND1 + 2, DRasN,

DGEF, KIND1 or KIND2 together with EGFP

were transfected into primary cultures of

mouse cerebellar granule cells at DIV1.

Transfected cells fixed at DIV14 were

immunostained with anti-Flag serum Flag

immunoreactivity (red) and EGFP

fluores-cence (green) were observed by confocal

microscopy Open and closed arrowheads

indicate dendrites and axons, respectively.

Scale bar = 50 lm.

GST-CD2 binding ability of six Flag-tagged v-KIND

derivatives (Fig S2A) The pull-down assay with

GST-CD2, followed by immunoblotting with anti-Flag

serum, showed that four KIND2-containing constructs

(v-KIND, DKIND1, DRasN and DRasGEF)

inter-acted with GST-CD2, whereas two constructs lacking

KIND2 (DKIND2 and DKIND1 + 2) failed to bind

to GST-CD2 (Fig S2A) This indicates that the

KIND2 domain of v-KIND is involved in the

interac-tion with the CD2 region of MAP2 Furthermore,

a similar pull-down assay with GST-CD2 to screen

four KIND domain constructs (Spir-2 KIND, PTPN13

showed that only v-KIND KIND2 could bind to

GST-CD2 (Fig S2B) Taken together, these results

reveal the specific protein–protein interaction between

the v-KIND KIND2 domain and the middle CD

region of MAP2

To narrow down the region responsible for v-KIND

binding within the MAP2 CD2 region, we first

subdi-vided the CD2 region (amino acids 600–1099)

into three subregions: CD2-1 (amino acids 600–767),

CD2-2 (amino acids 768–934) and CD2-3 (amino acids

935–1099) (Fig 3C) Next, we analyzed the binding

ability of bacterially expressed proteins of these

con-structs (Fig S1B) to endogenous v-KIND protein from

cerebellar lysates As a result, only GST-CD2-1 pulled

down v-KIND (Fig 3D) To identify the sequence

crit-ical for v-KIND binding within the CD2-1 (amino

acids 600–767) region, we next generated a series of

GST-fused deletions of the CD2-1 region (Fig 3E)

and examined the interactions between bacterially

expressed proteins of these subregions (Fig S1C) and

endogenous cerebellar v-KIND protein All four C-ter-minal subregions (CD2-1-1 to CD2-1-4) failed to pull down v-KIND (Fig 3F, upper), whereas three of the four N-terminal subregions (CD2-1-6, CD2-1-7 and CD2-1-8, but not CD2-1-5) could pull down v-KIND (Fig 3F, middle) Finally, we generated three GST-fused C-terminal truncations of the CD2-1-6 subregion (Fig 3E, lower) The pull-down assay with bacterially expressed proteins of these truncated constructs

GST-CD2-1-6-3, but not GST-CD2-1-6-1, bound to cerebellar v-KIND (Fig 3F, lower) This indicates that residues 702–744 of the smallest construct CD2-1-6-2 contain the core sequence for v-KIND binding

To evaluate the v-KIND KIND2-binding specificity

of residues 702–744 of MAP2, we performed a pull-down assay of combinations of EGFP-fused KIND1

or KIND2 of v-KIND with GST-fused CD2-1-6-1 or CD2-1-6-2 of MAP2 We successfully detected a pull down for the combination of EGFP-KIND2 and GST-CD2-1-6-2, but not for other combinations (Fig S2C), suggesting that the core sequence critical for the specific v-KIND KIND2 binding resides within residues 702–744 of MAP2

Residues 456–487 in the KIND2 domain of v-KIND contain the core MAP2 binding module

To identify the core MAP2-binding site within the KIND2 domain (amino acids 456–620), we analyzed five fused KIND2 domain derivatives, EGFP-KIND2-1 to -KIND2-5 (Fig 4A), by a pull-down assay with GST-CD2 (MAP2 v-KIND binding core)

Input (cerebellum) GST v-KIND KIND1 v-KIND KIND2 Spir-2 KIND PTPN13 KIND Input (MAP2) GST v-KIND KIND1 v-KIND KIND2 Spir-2 KIND PTPN13 KIND

IB:

α-MAP2 kDa

50

37

25

CBB

A

B

Fig 2 Of all members of the KIND family

of proteins, only KIND2 binds to MAP2 Pull-down assay (PD) of the endogenous MAP2 from cerebellar lysates of P21 mice (A, left) and purified MAP2a ⁄ b (A, right) by GST-fused KIND domains (v-KIND KIND1, v-KIND KIND2, Spire-2 KIND and PTPN13 KIND) shown in (B), followed by immuno-blotting with anti-MAP2 serum.

Immunoblotting with anti-EGFP serum showed that

three C-terminal truncations (KIND2-1, KIND2-2 and

KIND2-3), but not two N-terminal truncations

(KIND2-4 and -KIND2-5), were pulled down by

GST-CD2 (Fig 4B) This suggests that the core

sequence critical for the specific MAP2 binding resides

within the 32 residues (amino acids 456–487) of the v-KIND KIND2 domain

To examine whether the MAP2 binding core site of v-KIND binds to intact MAP2 protein, we coex-pressed EGFP-fused KIND2-1 and full-length MAP2

in COS7 cells and analyzed their interaction by

Input Mouse IgG -MAP2 GST RII CD1 CD2 CD3 MBD

IB: -v-KIND

250 150

IB: -v-KIND Input GST CD2-1 CD2-2 CD2-3 CD2 PD

250 150

Input GST CD2-1-6-1 CD2-1-6-2 CD2-1-6-3 CD2-1 PD

IB: -v-KIND

250 150

Input GST CD2-1-5 CD2-1-6 CD2-1-7 CD2-1-8 CD2-1 PD

250

Input GST CD2-1-1 CD2-1-2 CD2-1-3 CD2-1-4 CD2-1 PD

250 150

MAP2

MBD

RII CD1 CD2 CD3

1–147 148–599 600–1099 1100–1518 1519–1829

767 934

CD2-1 CD2-2 CD2-3

600

CD2

Inter-action

CD2-1

Inter-action

CD

Inter-action

Fig 3 MAP2 interacts with v-KIND via residues 702–744 within the MAP2 center region of the CD (A) MAP2a was subdivided into five fragments [the cAMP-dependent protein kinase regulatory subunit RII binding domain (RII), the central domain (CD)1-3 and the microtubule-binding domain (MBD)] and corresponding GST fusion proteins were generated (B), Pull-down assay (PD) of v-KIND from mouse cerebellar lysates (input) by bacterially expressed GST-fused MAP2 derivatives, followed by immunoblotting (IB) with anti-v-KIND serum Immunopre-cipitation assay (IP) of v-KIND from cerebellar lysates (input) by anti-MAP2 serum and normal mouse IgG as a positive and negative control, respectively (C) Division of the middle CD2 region (amino acids 600–1099) of the MAP2 CD into three subregions: CD2-1 (amino acids 600– 767), CD2-2 (amino acids 768–934) and CD2-3 (amino acids 935–1099) (D) Pull-down assay (PD) of v-KIND from cerebellar lysates (input) by fused CD2 and its subregions (CD2-1, CD2-2 or CD2-3), followed by immunoblotting (IB) with anti-v-KIND serum (E) The series of GST-fused MAP2 CD2-1 (amino acids 600–767) derivatives: CD2-1-1 (amino acids 600–634), CD2-1-2 (amino acids 600–667), CD2-1-3 (amino acids 600–701), CD2-1-4 (amino acids 600–734), CD2-1-5 (amino acids 735–767), CD2-1-6 (amino acids 702–767), CD2-1-7 (amino acids 668– 767), CD2-1-8 (amino acids 635–767), CD2-1-6-1 (amino acids 702–735), CD2-1-6-2 (amino acids 702–744) and CD2-1-6-3 (amino acids 702– 755) (F) Pull-down assay of v-KIND from cerebellar lysates (input) by the GST-fused CD2 derivatives shown in (E), followed by immunoblot-ting with anti-v-KIND serum.

a co-immunoprecipitation assay (Fig 4C) Using

anti-MAP2 sera, we found that EGFP-KIND2-1 was

co-immunoprecipitated with MAP2 in the cell lysates

Taken together, these data indicate that the

MAP2-binding core site of v-KIND is sufficient for the

v-KIND-MAP2 interaction within cells

The MAP2-binding core module of v-KIND is

involved in targeting to neuronal dendrites and

dendrite growth

We next investigated the MAP2 binding core of

v-KIND in terms of its ability to target to dendrites

and to control dendrite morphology We transfected

three HA-epitope-tagged v-KIND derivatives (v-KIND,

KIND2 domain and KIND 2-1 MAP2-binding core

region) into primary cultured hippocampal neurons at

7 days in vitro (DIV), fixed at DIV21, and analyzed

the dendrite localization of the expressed proteins as

well as the dendrite morphology by immunodetection with anti-HA serum (Fig 5A) Hippocampal neurons were used because the dendrite morphology was much more significant and easy to observe than that in cere-bellar granule cells The KIND2-1 core region was localized to dendrites, as was the endogenous v-KIND (Fig S3), indicating that the MAP2-binding core con-fers the dendritic targeting of v-KIND Overexpression

of v-KIND decreased dendritic branching as reported previously [12], whereas overexpression of KIND2 pro-moted dendrite extension, compared to that of EGFP alone (Fig 5A) Interestingly, overexpression of the KIND 2-1 domain resulted in a complex dendrite mor-phology (Fig 5A) Sholl analysis of the dendrite

PD: GST-CD2 IB: -GFP

Input IB: -GFP

GST-CD2 (CBB stain)

EGFP EGFP-KIND2-1 EGFP-KIND2-2 EGFP-KIND2-3 EGFP-KIND2-4 EGFP-KIND2-5 EGFP-KIND2 EGFP-KIND1

KIND2-4 488

KIND2

v-KIND

+ + + – –

Interaction

EGFP EGFP- KIND2-1

Input IB: -GFP

IP: -MAP2 IB: -GFP

IP: -MAP2 IB: -MAP2

A

B

C

Fig 4 Residues 456–487 within the v-KIND KIND2 domain contain

the MAP2-binding core site (A) Division of the KIND2 domain

(456–620) into five subregions: KIND2-1 (amino acids 456–487),

KIND2-2 (amino acids 456–521), KIND2-3 (amino acids 456–555),

KIND2-4 (amino acids 488–620) and KIND2-5 (amino acids

589–620) These KIND2 subregions were fused with EGFP at the

N-termini and expressed in COS7 cells (B) Pull-down assay (PD)

of EGFP-fused KIND2 derivatives (middle) by GST-fused CD2

(bottom), followed by immunoblotting (IB) with anti-GFP serum

(top) KIND2 and KIND1 constructs were used as positive and

negative controls, respectively (C) Immunoprecipitation assay (IP)

of COS7 cells co-expressing MAP2 full-length (middle) and

GFP-fused KIND2-1 (bottom) with anti-MAP2 serum, followed by

immunoblotting with anti-GFP serum (top).

16 18

EGFP

***

6 8 10 12 14

EGFP v-KIND KIND2 KIND2-1

0 2 4

Distance from soma (µm)

A

B

Fig 5 KIND 2-1 targets the dendrite and its overexpression sup-pressed dendrite growth (A) Flag-tagged v-KIND, v-KIND KIND2 (amino acids 456–620) or KIND2-1 (amino acids 456–487) together with EGFP were cotransfected into primary cultured mouse hippo-campal neurons at DIV7 Cells fixed at DIV21 were immunostained with anti-HA serum HA immunofluorescence and its colocalization with GFP fluorescence were observed by confocal microscopy Scale bar = 100 lm (B) Sholl analysis of the dendrite complexity of individual neurons Data were obtained from three independent transfection experiments (for each experiment, n = 15 per con-struct) The results are the mean ± SEM ***P < 0.001 for v-KIND versus KIND2-1.

branching patterns showed that the number of

den-drites within 130 lm from the soma was decreased in

neurons overexpressing v-KIND, but increased in

neu-rons overexpressing either KIND2 or KIND2-1,

com-pared to neurons expressing EGFP (Fig 5B) Neurons

overexpressing KIND2-1 had an increased number of

dendrite branches of < 70 lm compared to neurons

overexpressing KIND2 Interestingly, the number of

dendrites > 300 lm was slightly increased by

overex-pressing v-KIND compared to the overexpression of

EGFP, KIND2 or KIND2-1 (Fig 5B) These results

suggest that KIND2 and KIND2-1, overexpressed

in neurons, act as dominant-negative regulators of

v-KIND and suppress the negative regulation of

den-drite growth by endogenous v-KIND

Three conserved Leu residues are critical for the

v-KIND and MAP2 interaction

We examined the 32 residues of the MAP2-binding

core (amino acids 456–487) of mouse v-KIND by

com-paring them with those of human and Gallus v-KIND,

as well as the corresponding residues of mouse

v-KIND KIND1, Spire-2 KIND and PTPN13 KIND

(Fig 6A) The amino acid sequences of the core

regions of mouse, human and Gallus v-KIND shared

65.6% homology, and 17 identical and four function-ally similar residues, including seven conserved Leu residues, were identified By contrast, the correspond-ing sequence in the three other KIND domains from mouse Spir-2, v-KIND and PTPN13 shared 46.9% homology and had only two or three conserved Leu residues out of seven residues It is notable that four Leu residues (amino acids 461, 474, 477 and 482) and one Thr residue (amino acid 487) were well conserved

in the KIND2 of v-KIND in all species analyzed, although not in the other KINDs To investigate the possible involvement of these conserved Leu residues

in the interaction between v-KIND and MAP2, we generated EGFP-tagged KIND2 mutants with an Ala substitution at Leu461, 474, 477, 482 or 485 (L461A, L474A, L477A, L482A or L485A) and conducted a pull-down assay with the v-KIND-binding core mod-ule CD2-1-6-2 of MAP2 (Fig 6B) KIND2 L461A, L474A and L477A mutants failed to bind to

CD2-1-6-2 of MAPCD2-1-6-2, whereas the L48CD2-1-6-2A and L485A mutants did bind to this module These results suggest that the Leu461, 474 and 477 residues conserved in the KIND2 domain alone are indispensable for the v-KIND– MAP2 interaction

The Leu474 residue of the KIND2 MAP2-binding core module is important for dendrite growth of hippocampal neurons

To clarify the biological significance of the v-KIND MAP2-binding core module, we generated full-length v-KIND containing the L474A substitution mutation and analyzed its role in dendrite growth by cotransfection with EGFP in hippocampal neurons (Fig 7) When compared with neurons transfected with wild-type v-KIND, which decreased the number of dendrites, neu-rons transfected with the L474A mutant had more com-plex dendritic arborization (Fig 7A) Sholl analysis showed that the L474A mutant-transfected neurons had

an increased number of dendrites within 150 lm from the soma compared to wild-type v-KIND-transfected neurons (Fig 7B) In addition, L474A-transfected neu-rons had more dendritic branches at 90–140 lm from the soma compared to neurons transfected with EGFP alone, although the two transfectants had similar num-bers of proximal dendrites < 80 lm and distal dendrites

> 150 lm from the soma In addition, the maximal length of each dendrite in v-KIND-overexpressed neu-rons was significantly greater than that of EGFP- or L474A-overexpressed neurons (Fig 7C) These results suggest that the Leu474 residue contributed to the func-tional interaction between v-KIND and MAP2 in the regulation of neuronal dendrite growth

A

B

Fig 6 Conserved Leu residues are important for the interaction

between v-KIND and MAP2 (A) Alignment of the KIND2-1 region in

human, mouse and Gallus (top) and the corresponding mouse

v-KIND KIND1, Spir-2 KIND and PTPN13 KIND domains (bottom).

The residues conserved across all species in the KIND2 domain

and in any other KIND domain are highlighted in gray, whereas

those conserved only in the KIND2 domain are highlighted in black.

The Leu residues at positions 461, 474, 477 and 482 were

substi-tuted by Ala (B) Pull-down assay (PD) of EGFP-fused KIND2,

L461A, L474A, L477A, L482A or L485A mutant of v-KIND KIND2

by the v-KIND-binding core module CD2-1-6-2 of MAP2 EGFP

alone was used as a control Asterisks in (A) show the Leu

residues essential for binding.

The present study revealed the structural and

func-tional determinants of the protein–protein interaction

between v-KIND and MAP2 (i.e the MAP2-binding

core module in v-KIND and the v-KIND-binding core

module in MAP2) across a range of amino acid

resi-dues, and provided evidence that these novel protein–

protein interaction core modules play a pivotal role in

regulating dendrite growth and branching of cerebellar

granule cells and hippocampal neurons

Among the KIND domains in four KIND-containing

proteins (Spir-2, PTPN13, FRMPD2 and v-KIND),

only the KIND2 domain of v-KIND specifically bound

to MAP2 It is noteworthy that the v-KIND

MAP2-binding core polypeptides of 32 residues expressed in

hippocampal neurons were very effective in promoting

dendrite branching, which is opposite to the effect of the

overexpression of v-KIND, but similar to the effect of

knockdown of v-KIND [12], thereby suggesting that the

32 amino acids core polypeptide acts as a

dominant-neg-ative molecule by competing with endogenous v-KIND

for MAP2 binding However, v-KIND-overexpressing

neurons showed decreased dendrite branches, but

formed slightly longer dendrites than the control

neurons and KIND2- or binding core-overexpressing

neurons A previous in vitro study indicated that the

RasGEF activity of v-KIND induces the

phosphoryla-tion of MAP2 by JNK1 and⁄ or ERK via the activation

of the Ras–Raf–MAP kinase pathway [12] These results

appear to be in agreement with those of previous studies

showing that the phosphorylation of MAP2 by JNK1 (such as downstream of v-KIND overexpression [12]) enhances the activity of MAP2 to bind to microtubules and promote their assembly [18], and that the inhibition

of JNK1 (such as downstream of v-KIND knockdown [12]) increases the number of dendrite branches, but decreases the mean dendrite length [19] Taken together, these results indicate that the KIND2 domain regulates dendrite complexity via targeting of v-KIND RasGEF (an activator of the Ras pathway) to MAP2 associated with the dendritic microtubule cytoskeleton

The present study indicates that the conserved amino acid sequence of the MAP2 binding core module in the mouse, human and Gallus v-KIND is indispensable for the v-KIND–MAP2 interaction The Ala substitution for Leu at amino acids 461, 474 or 477, which are con-served among the mouse, human and Gallus KIND2 but not in the mouse Spir-2 and PTPN13 KIND, as well as the mouse v-KIND KIND1, impaired the interaction with MAP2 In addition, neurons transfected with v-KIND bearing the L474A mutation induced more complex dendritic arborization patterns than those of neurons transfected with wild-type v-KIND, which exhibited a decrease in total number of dendrites and an increased mean length of dendrites These findings dem-onstrate the structural and functional importance of the Leu474 residue in the v-KIND–MAP2 interaction-medi-ated regulation of dendrite growth

The interaction with v-KIND is specific to HMW-MAP2, but not to LMW-HMW-MAP2, which lacks the CD [14] Although the functional property of the CD, the

EGFP v-KIND L474A

20 50 80 110 140 170 200 230 260 290 320 350

14

12

10

8

6

4

2

0

***

Distance from soma (µm)

0 100 200 300 400

EGFP v-KIND L474A

A

Fig 7 Contribution of the Leu474 residue

in the MAP2-binding core of v-KIND to den-drite growth and branching in hippocampal neurons (A) HA-tagged v-KIND or the L474A mutant was cotransfected with EGFP into primary-cultured mouse hippo-campal neurons at DIV7 Transfection with EGFP alone was used as a control Cells fixed at DIV21 were immunostained with anti-HA serum Cell morphology images with EGFP fluorescence were obtained by confocal microscopy Scale bar = 100 lm (B) Sholl analysis of dendrite complexity of individual neurons Data were obtained from three independent transfection experiments (for each experiment, n = 15 per construct) The results are the mean ± SEM *P < 0.05 and ***P < 0.001 for v-KIND versus L474A (C) Mean dendrite length in individual neurons The results are the mean ± SEM.

*P < 0.05 and ***P < 0.001.

largest domain of MAP2 (1372 amino acids,

account-ing for 70% of the total size) is not yet fully

under-stood, the data obtained in the present study indicate

that the 43 residues (amino acids 702–744 in mice) that

reside in the middle region of the CD act as the

v-KIND binding core The v-KIND-binding core

mod-ule of MAP2 is also well conserved among human,

mouse and Gallus, and contains six conserved Leu

resi-dues (Fig S4) Thus, it would be interesting to

deter-mine whether a hydrophobic interaction between the

Leu residues from v-KIND and MAP2 contributes to

the interaction between the two proteins

In conclusion, the present study has clarified the

structural and functional importance of the v-KIND

and MAP2 interaction core modules in the regulation

of dendrite growth and branching in hippocampal

neu-rons and cerebellar granule cells Further studies of

these newly-identified protein–protein interaction core

modules, including tertiary structural analyses, will

shed light on the molecular mechanism by which the

v-KIND–MAP2 interaction regulates the dendrite

arborization patterns that are critical for shaping

neu-ronal circuits, and also may provide a clue to the

understanding of some MAP2-associated

neurodegen-erative and psychiatric disorders [16,17]

Materials and methods

Animals

Mice (ICR) were purchased from Nihon SLC (Hamamatsu,

Japan) and used in accordance with protocols approved by

the Animal Care and Use Committee of RIKEN

Plasmid construction and expression

Plasmid construction and expression in Escherichia coli or

African green monkey kidney cell line COS7 cells were

per-formed essentially as described previously [12] Mouse

v-KINDcDNA [12] and its derivatives were cloned into a

mammalian cell expression vector, pCAG, with Flag or HA

tags at the C-terminal ends The fragments of the KIND1

and KIND2 domains were generated by PCR and inserted

into pEGFP-C1 (Clontech Laboratories, Inc., Mountain

View, CA, USA) and pGEX-4T-2 (GE Healthcare UK Ltd,

Little Chalfont, UK) vectors for EGFP and GST fusion

con-structs, respectively HMW MAP2 cDNA and its derivatives

generated by PCR were cloned into pGEX-4T-2 The KIND

domains of Spir-2 and PTPN13 were generated by RT-PCR

and inserted into pEGFP-C1 and pGEX-4T-2 The N- and

C-terminal regions of the v-KIND KIND2 and the MAP2

CD were generated by PCR and cloned into appropriate

expression vectors Site-directed mutagenesis for substitution

of conserved Leu residues with Ala in the KIND2 domain of

v-KIND was performed using the QuikChange XL Site Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA)

Pull-down assay

GST-fusion proteins were prepared and used for the pull-down assays, as described previously [12] Briefly, E coli expressing GST fusion proteins were lysed in ice-chilled lysis buffer (50 mm Tris-HCl pH 7.4, 25% sucrose, 1% Triton X-100 and 5 mm MgCl2) Then, 10 lg of E coli lysates containing GST fusion protein was coupled to glutathione-sepharose (GE Healthcare UK Ltd) by rotating for 1 h at

4C After washing with lysis buffer, the GST fusion protein coupled with sepharose was mixed with 1 mg of protein lysates prepared from mouse cerebella After rotating for 1 h

at 4C, the GST fusion protein complex was washed with lysis buffer and subjected to immunoblot analysis

Immunoprecipitation

Immunoprecipitation was performed as described previ-ously [20] Briefly, COS7 cells or mouse cerebella were lysed and homogenized in ice-chilled lysis buffer (50 mm Hepes,

150 mm NaCl, 10% glycerol, 1% Triton X-100, 1.5 mm MgCl2, 1 mm EGTA, 100 mm NaF, 1 mm Na3VO4,

10 lgÆmL)1 aprotinin, 10 lgÆmL)1 leupeptin and 1 mm phenylmethanesulfonyl fluoride) After centrifugation at

1000 g for 10 min at 4C, the supernatants were mixed with the antibody and incubated on ice for 1 h, followed by rotation with protein A-sepharose or protein G-sepharose (GE Healthcare UK Ltd) at 4C Proteins immunoprecipi-tated with the antibody–protein A- or G-sepharose com-plexes were washed with lysis buffer and subjected to immunoblot analysis

Immunoblotting

SDS⁄ PAGE and immunoblotting were performed essen-tially as described previously [12] The anti-(rabbit v-KIND) serum [12] was used at 0.5 lgÆmL)1 Antibodies against MAP2a⁄ b (catalog number: AP20; Sigma-Aldrich,

St Louis, MO, USA), MAP2 (catalog number: M1406; Sigma-Aldrich), EGFP (catalog number: SC-9996; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), Flag (catalog number: 3165; Sigma-Aldrich) and HA (catalog number: 1867423; Roche Diagnostics, Basel, Switzerland) were used at dilutions of 1 : 1000 for immunoblotting,

1 : 200 for immunocytochemistry and 1 lgÆmL)1 for co-immunoprecipitation

Primary cultures, transfection and imaging of hippocampal and cerebellar neurons

Hippocampal and cerebellar dissociated primary cultures were prepared from ICR mice (Nippon SLC, Hamamatsu,

Japan) at embryonic day 16 and postnatal day 0,

respec-tively Cells were transfected with Flag- or HA-tagged

full-length v-KIND or mutant constructs together with the

EGFP vector as a cell morphology marker, essentially as

described previously [12] Briefly, hippocampal neurons

(1· 105

cellsÆcm)2) at DIV7 were transfected using

Lipofec-tamine 2000 reagent (Invitrogen, Carlsbad, CA, USA) and

cultured in Opti-MEM (Invitrogen) Cerebellar neurons

(5· 105cellsÆcm)2) at DIV1 were transfected by the Ca2+

-phosphate-mediated method using a CellPhect Transfection

kit (GE Healthcare UK Ltd) and were cultured in

serum-free Eagle’s minimum essential medium (Nissui

Pharmaceu-tical Co., Ltd, Tokyo, Japan) Transfected cells were

visual-ized by EGFP fluorescence and immunocytochemical

staining with anti-Flag or anti-HA sera Cell images were

acquired by confocal microscopy (LSM510; Carl Zeiss,

Inc., Oberkochen, Germany)

Morphometric analysis of dendritic arborization

patterns

To quantify dendrite growth and branching, 15 neurons

transfected with each construct were randomly chosen for

each experiment, and EGFP fluorescent images of their

dendrites were analyzed with neurolucida software (MBF

Bioscience, Williston, VT, USA) Image data were

statisti-cally quantified by repeated-measures analysis of variance

with Bonferroni post-hoc analysis

Acknowledgements

We thank Dr N Cowan (NYU Medical Center) and

Dr J Miyazaki (Osaka University) for their kind gifts

of the MAP2 cDNA and the pCAG expression vector,

respectively The present study was supported by

grants from the Japanese Ministry of Education,

Culture, Sports, Science and Technology; the Japan

Society for the Promotion of Science; and the Japan

Science and Technology Agency

References

1 Pawson T & Nash P (2000) Protein–protein interactions

define specificity in signal transduction Genes Dev 14,

1027–1047

2 Pawson T & Nash P (2003) Assembly of cell regulatory

systems through protein interaction domains Science

300, 445–452

3 Ciccarelli FD, Bork P & Kerkhoff E (2003) The KIND

module: a putative signalling domain evolved from the

C lobe of the protein kinase fold Trends Biochem Sci

28, 349–352

4 Otto IM, Raabe T, Rennefahrt UE, Bork P, Rapp UR

& Kerkhoff E (2000) The p150-Spir protein provides a

link between c-Jun N-terminal kinase function and actin reorganization Curr Biol 10, 345–348

5 Kerkhoff E (2006) Cellular functions of the Spir actin-nucleation factors Trends Cell Biol 16, 477–483

6 Erdmann KS (2003) The protein tyrosine phosphatase PTP-Basophil⁄ Basophil-like Interacting proteins and molecular functions Eur J Biochem 270, 4789–4798

7 Wansink DG, Peters W, Schaafsma I, Sutmuller RP, Oerlemans F, Adema GJ, Wieringa B, van der Zee CE

& Hendriks W (2004) Mild impairment of motor nerve repair in mice lacking PTP-BL tyrosine phosphatase activity Physiol Genomics 19, 50–60

8 Stenzel N, Fetzer CP, Heumann R & Erdmann KS (2009) PDZ-domain-directed basolateral targeting of the peripheral membrane protein FRMPD2 in epithelial cells J Cell Sci 122, 3374–3384

9 Mees A, Rock R, Ciccarelli FD, Leberfinger CB, Borawski JM, Bork P, Wiese S, Gessler M &

Kerkhoff E (2005) Very-KIND is a novel nervous system specific guanine nucleotide exchange factor for Ras GTPases Gene Expr Patterns 6, 79–85

10 Quinlan ME, Hilgert S, Bedrossian A, Mullins RD & Kerkhoff E (2007) Regulatory interactions between two actin nucleators, Spire and Cappuccino J Cell Biol 179, 117–128

11 Ley TJ, Mardis ER, Ding L, Fulton B, McLellan MD, Chen K, Dooling D, Dunford-Shore BH, McGrath S, Hickenbotham M et al (2008) DNA sequencing of a cytogenetically normal acute myeloid leukaemia gen-ome Nature 456, 66–72

12 Huang J, Furuya A & Furuichi T (2007) Very-KIND,

a KIND domain containing RasGEF, controls dendrite growth by linking Ras small GTPases and MAP2

J Cell Biol 179, 539–552

13 Georges PC, Hadzimichalis NM, Sweet ES &

Firestein BL (2008) The yin-yang of dendrite morphol-ogy: unity of actin and microtubules Mol Neurobiol 38, 270–284

14 Sanchez C, Diaz-Nido J & Avila J (2000) Phosphoryla-tion of microtubule-associated protein 2 (MAP2) and its relevance for the regulation of the neuronal cytoskel-eton function Prog Neurobiol 61, 133–168

15 Farah CA & Leclerc N (2008) HMWMAP2: new per-spectives on a pathway to dendritic identity Cell Motil Cytoskeleton 65, 515–527

16 Iqbal K, Liu F, Gong CX, Alonso AC & Grundke-Iq-bal I (2009) Mechanisms of tau-induced neurodegenera-tion Acta Neuropathol 118, 53–69

17 Arnold SE (2000) Cellular and molecular neuropathol-ogy of the parahippocampal region in schizophrenia Ann N Y Acad Sci 911, 275–292

18 Chang L, Jones Y, Ellisman MH, Goldstein LS & Karin M (2003) JNK1 is required for maintenance of neuronal microtubules and controls phosphorylation of microtubule-associated proteins Dev Cell 4, 521–533