ã
Oncogene (2000) 19, 3050 ± 3058
2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00
www.nature.com/onc
Two actions of frabin: direct activation of Cdc42 and indirect activation of
Rac
Yuichi Ono1,3,5, Hiroyuki Nakanishi1,2,5, Miyuki Nishimura1,3,5, Mayumi Kakizaki1,3,5,
Kenichi Takahashi1,4,5, Masako Miyahara1,5, Keiko Satoh-Horikawa1,3,5, Kenji Mandai1,2,5 and
Yoshimi Takai*,1,2,5
1
Takai Biotimer Project, ERATO, Japan Science and Technology Corporation, c/o JCR Pharmaceuticals Co. Ltd., 2-2-10
Murotani, Nishi-ku, Kobe 651-2241, Japan; 2Department of Molecular Biology and Biochemistry, Osaka University Graduate
School of Medicine/Faculty of Medicine, Suita 565-0871, Japan
Frabin is an actin ®lament-binding protein which shows
GDP/GTP exchange activity speci®c for Cdc42 small G
protein and induces ®lopodium-like microspike formation
and c-Jun N-terminal kinase (JNK) activation presumably
through the activation of Cdc42. Frabin has one actin
®lament-binding (FAB) domain, one Dbl homology (DH)
domain, ®rst pleckstrin homology (PH) domain adjacent
to the DH domain, one cysteine-rich FYVE domain, and
second PH domain from the N-terminus to the C-terminus
in this order. Di€erent domains of frabin are involved in
the microspike formation and the JNK activation, and the
association of frabin with the actin cytoskeleton through
the FAB domain is necessary for the microspike
formation, but not for the JNK activation. We have found
here that frabin induces the formation of not only
®lopodium-like microspikes but also lamellipodium-like
structures in NIH3T3 and L ®broblasts. We have analysed
the mechanism of frabin in these two actions and found
that frabin induces ®lopodium-like microspike formation
through the direct activation of Cdc42 and lamellipodiumlike structure formation through the Cdc42-independent
indirect activation of Rac small G protein. The FAB
domain of frabin in addition to the DH domain and the
®rst PH domain is necessary for the ®lopodium-like
microspike formation, but not for the lamellipodium-like
structure formation. The FYVE domain and the second
PH domain in addition to the DH domain and the ®rst PH
domain are necessary for the lamellipodium-like structure
formation. We show here these two actions of frabin in the
regulation of cell morphology. Oncogene (2000) 19,
3050 ± 3058.
Keywords: frabin; Rac; Cdc42; ®lopodia; lamellipodia;
membrane ru‚es
Introduction
The Rho family small GTP-binding proteins (G
proteins) regulate a variety of the actin cytoskeleton-
*Correspondence: Y Takai, Department of Molecular Biology and
Biochemistry, Osaka University Graduate School of Medicine/
Faculty of Medicine, Suita 565-0871, Japan
Current addresses: 3KAN Research Institute Inc., 1 ChudojiAwatamachi, Shimogyo-ku, Kyoto 600-8815, Japan; 4 JCR
Pharmaceuticals Co., Ltd., 2-2-10 Murotani, Nishi-ku, Kobe 6512241, Japan
5
Takai Biotimer Project was closed in September 1999
Received 24 January 2000; revised 11 April 2000; accepted 12 April
2000
dependent cell functions, including cell adhesion and
motility (for reviews, see Takai et al., 1995; Hall, 1998;
Symons, 1996). The Rho family consists of three major
subfamilies: the Cdc42, Rac, and Rho subfamilies. In
®broblasts, Cdc42 induces ®lopodium formation; Rac
induces the formation of lamellipodia or ru‚es
(membrane ru‚ing); and Rho regulates assembly of
stress ®bers and focal adhesions. Moreover, Cdc42,
Rac, and Rho are placed in a hierarchical cascade
wherein Cdc42 activates Rac, which in turn activates
Rho. Thus, these small G proteins coordinate the
temporal and spatial changes in the actin cytoskeleton.
However, molecular mechanisms of these connections
among the three subfamilies have not yet been clari®ed.
The Rho family small G proteins act as biotimers
through interconversion between the GDP-bound
inactive form and the GTP-bound active form (Takai
et al., 1995; Hall, 1998; Symons, 1996). The exchange
of GDP for GTP is stimulated by a GDP/GTP
exchange protein (GEP). Many GEPs for the Rho
family small G proteins have been identi®ed and share
two conserved domains, a Dbl homology (DH) domain
and a pleckstrin homology (PH) domain adjacent to
the DH domain. The DH domain is responsible for the
catalytic activity and the role of the PH domain
remains to be established but is suggested to be
involved in the regulation of the activity or determination of its intracellular localization (for a review, see
Cerione and Zheng, 1996).
We have recently identi®ed a novel actin ®lament (Factin)-binding protein, named frabin (Obaishi et al.,
1998). Frabin has one F-actin-binding (FAB) domain,
one DH domain, the ®rst PH domain adjacent to the
DH domain, one cysteine-rich FYVE domain, and the
second PH domain. This domain structure is similar to
that of FGD1, a GEP speci®c for Cdc42, except that
FGD1 lacks the FAB domain but has a proline-rich
domain (Pasteris et al., 1994; Zheng et al., 1996a). The
FGD1 gene is originally determined by positional
cloning to be the genetic locus responsible for
faciogenital dysplasia or Aarskog-Scott syndrome
(Zheng et al., 1996a). We have shown that frabin has
GEP activity speci®c for Cdc42 in a cell-free system
and that frabin induces ®lopodium-like structure
formation in Swiss 3T3 and COS7 cells and c-Jun Nterminal kinase (JNK) activation in COS7 cells
(Obaishi et al., 1998; Umikawa et al., 1999) as
described for a dominant active mutant of Cdc42
(V12Cdc42) and FGD1 (Nobes and Hall, 1995; Kozma
et al., 1995; Zheng et al., 1996a; Nagata et al., 1998).
Di€erent domains of frabin are necessary for the
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3051
®lopodium-like structure formation and the JNK
activation in COS7 cells: the FAB domain in addition
to the DH domain and the ®rst PH domain is
necessary for the ®lopodium-like structure formation,
whereas the FYVE domain and the second PH domain
in addition to the DH domain and the ®rst PH domain
are necessary for the JNK activation (Umikawa et al.,
1999).
During the course of the study on the activities of
frabin in cell lines other than Swiss 3T3 and COS7
cells, we have found here that frabin is capable of
inducing not only the direct activation of Cdc42, which
induces ®lopodium-like structure formation, but also
the Cdc42-independent indirect activation of Rac,
which induces the formation of lamellipodium-like
structures or ru‚es, in NIH3T3 and L ®broblasts.
We describe here these two actions of frabin in the
regulation of cell morphology. We use here `®lopodium' and `lamellipodium' for just simplicity as
®lopodium-like structure and lamellipodium-like structure, respectively.
Results
Direct activation of Cdc42 and Cdc42-independent
indirect activation of Rac by frabin in NIH3T3 cells
We have previously shown that frabin induces
®lopodium formation in Swiss 3T3 and COS7 cells
(Obaishi et al., 1998; Umikawa et al., 1999). We
examined here by use of retrovirus-mediated or
liposome-mediated gene transfection whether frabin
also shows this activity in NIH3T3 cells. The expression
vectors used here are designed to produce lower levels
of proteins than the expression vectors used previously
(Obaishi et al., 1998; Umikawa et al., 1999). When
NIH3T3 cells, which were cultured at low and high cell
densities, were transfected with a Myc-tagged, dominant active mutant of Cdc42 (Myc-V12Cdc42), ®lopodia were observed in both the cells which did not form
cell ± cell contacts and the cells which formed cell ± cell
contacts (Figure 1A ± C). In contrast, green ¯uorescent
protein (GFP)-tagged, full-length frabin (frabin-GFP)
formed ®lopodia and lamellipodia at cell ± cell contact
sites and was colocalized with F-actin there (Figure
1D). However, these phenotypes were observed only in
about 2% of the frabin-GFP-expressing cells. Instead
of the formation of these structures, it induced
membrane ru‚ing at free edges, where frabin-GFP
was colocalized with F-actin, in more than 60% of the
frabin-GFP-expressing cells (Figure 1E). The frabinGFP-induced membrane ru‚ing was apparently similar
to that induced by a Myc-tagged, dominant active
mutant of Rac1 (Myc-V12Rac1) (Figure 1F). However,
transfection of Myc-V12Rac1 did not induce the
formation of ®lopodia or lamellipodia at cell ± cell
contact sites which were seen in Figure 1D (data not
shown). These results indicate that frabin mainly
induces membrane ru‚ing in NIH3T3 cells although
it also induces ®lopodium formation at cell ± cell
contact sites in about 2% of its expressing cells. This
activity of frabin in NIH3T3 cells is quite di€erent from
that in Swiss 3T3 and COS7 cells in which frabin
mainly induces ®lopodium formation (Obaishi et al.,
1998; Umikawa et al., 1999).
Figure 1 Frabin-induced membrane ru‚ing in NIH3T3 cells.
Frabin-GFP, Myc-V12Cdc42, or Myc-V12Rac was transfected by
use of the retrovirus-mediated gene transfection. The Myc-tagged
protein was assumed to be coexpressed with GFP alone. The
transfected cells were identi®ed by the detection of GFP. The cells
were stained with rhodamine-phalloidin. (A) Control cells; (B, C)
Myc-V12Cdc42-transfected cells; (D, E) frabin-GFP-expressing
cells; (F) Myc-V12Rac1-transfected cells; (B, E) cells without
cell ± cell contacts; and (C, D) cells with cell-cell contacts. Less
than 2% of the control cells showed membrane ru‚ing. More
than 50% of the Myc-V12Cdc42-transfected cells formed
®lopodia, but membrane ru‚es were observed in less than 2%
of the cells. Only about 2% of the frabin-GFP-expressing cells
formed ®lopodia and lamellipodia at cell-cell contact sites, but
more than 60% of the cells showed membrane ru‚ing at free
edges. More than 60% of the Myc-V12Rac1-transfected cells
showed membrane ru‚ing. Bars, 10 mm
To examine whether Rac is involved in the frabinGFP-induced membrane ru‚ing in NIH3T3 cells, a
Myc-tagged, dominant negative mutant of Rac1 (MycN17Rac1) was cotransfected with frabin-GFP. Cotransfection of Myc-N17Rac1 completely inhibited the
frabin-GFP-induced membrane ru‚ing (Figure 2A,B).
Under the same conditions, this cotransfection did not
inhibit the ®lopodium formation which was observed
at cell ± cell contact sites in the frabin-GFP-expressing
cells. We have recently found that a Myc-tagged,
dominant negative mutant of Cdc42 (Myc-N17Cdc42)
does not inhibit Cdc42-induced ®lopodium formation
in various types of cells, such as COS7 cells, indicating
that Myc-N17Cdc42 does not act as a dominant
negative mutant under some conditions (K Takaishi
and Y Takai, unpublished observations). Consistently,
cotransfection of Myc-N17Cdc42 did not inhibit the
frabin-induced formation of ®lopodia or ru‚es (data
not shown). Therefore, we used a Myc-tagged, Cdc42Oncogene
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3052
Figure 2 Frabin-induced membrane ru‚ing in a Cdc42-independent manner in NIH3T3 cells. Frabin-GFP plus MycN17Rac1 and/or Myc-N-WASP-CRIB was transfected by use of
the liposome-mediated gene transfection. Myc-N17Rac1 and/or
Myc-N-WASP-CRIB were assumed to be coexpressed with
frabin-GFP. The cells were stained with rhodamine-phalloidin.
(A) Frabin-GFP-expressing cells; (B) cells expressing frabin-GFP
and Myc-N17Rac1; (C) cells expressing frabin-GFP and Myc-NWASP-CRIB; and (D) cells expressing frabin-GFP, MycN17Rac1, and Myc-N-WASP-CRIB. About 10% of the cells
cotransfected with frabin-GFP and N17Rac1 showed ®lopodium
formation at cell-cell contact sites, but no cells showed membrane
ru‚ing. Bars, 10 mm
binding domain of N-WASP (Myc-N-WASP-CRIB),
which has been shown to speci®cally bind the GTPbound form of Cdc42 and to inhibit its activity (Miki
et al., 1998). Cotransfection of Myc-N-WASP-CRIB
did not inhibit the frabin-induced membrane ru‚ing
(Figure 2C), but cotransfection of Myc-N-WASPCRIB and Myc-N17Rac1 inhibited the frabin-induced
formation of ®lopodia and lamellipodia at cell ± cell
contact sites (Figure 2D). We have previously shown
that frabin is active on Cdc42 and inactive on Rac1
and RhoA as a GEP in a cell-free system (Umikawa et
al., 1999). Collectively, it is likely that frabin induces
not only the direct activation of Cdc42, which induces
®lopodium formation, but also the Cdc42-independent
indirect activation of Rac, which induces membrane
ru‚ing, in NIH3T3 cells.
Domains required for the Rac-mediated membrane
ruffling in NIH3T3 cells
To determine the domains of frabin required for the
Rac-mediated membrane ru‚ing, a series of deletion
mutants were expressed in NIH3T3 cells. Consistent
with frabin-GFP, Myc-tagged, full-length frabin (Mycfrabin-a) induced membrane ru‚ing and was colocalized with F-actin there (Figure 3A, see also Figures
1E, 2A and 4). Myc-Frabin-b, lacking the FAB
domain, also induced membrane ru‚ing to a slightly
smaller extent (Figure 3B, see also Figure 4). MycFrabin-b was concentrated with F-actin at the ru‚ing
sites, but it was also di€usely distributed throughout
the cytoplasm. Neither Myc-frabin-c, containing only
the FAB domain, nor Myc-frabin-d, lacking the FYVE
Oncogene
Figure 3 Domains of frabin required for membrane ru‚ing in
NIH3T3 cells. Various Myc-tagged mutants of frabin were
transfected by use of the retrovirus-mediated transfection. The
cells were doubly stained with the anti-Myc antibody and
rhodamine-phalloidin. The anti-Myc antibody was visualized with
an FITC-conjugated anti-mouse antibody. (A) Myc-Frabin-a; (B)
Myc-frabin-b; (C) Myc-frabin-c; (D) Myc-frabin-d; and (E) Mycfrabin-e. More than 60% of the Myc-frabin-a-expressing cells
showed membrane ru‚ing. More than 40% of the Myc-frabin-bexpressing cells showed membrane ru‚ing. Less than 2% of the
cells expressing either Myc-frabin-c, -d, or -e showed membrane
ru‚ing. Bars, 10 mm
domain and the second PH domain, induced membrane ru‚ing, although Myc-frabin-c and -d were
colocalized with F-actin (Figure 3C,D, see also Figure
4). Myc-Frabin-e, containing the DH domain and the
®rst PH domain, did not induce morphological change
and was di€usely distributed throughout the cytoplasm
(Figure 3E, see also Figure 4). These results indicate
that the FYVE domain and the second PH domain in
addition to the DH domain and the ®rst PH domain,
but not the FAB domain, are necessary for the Racmediated membrane ru‚ing in NIH3T3 cells. The
results of the e€ects of these mutants are summarized
in Figure 4.
To determine the involvement of each domain in the
Rac-mediated membrane ru‚ing, a mutant, containing
a small deletion in the DH domain (Myc-frabin-f), the
®rst PH domain (Myc-frabin-g), or the FYVE domain
(Myc-frabin-h), or containing the complete deletion of
the second PH domain (Myc-frabin-i), was examined.
None of these mutants except Myc-frabin-i induced
membrane ru‚ing (Figure 4). Myc-Frabin-i induced
membrane ru‚ing in about 2 ± 5% of its expressing
cells. These results indicate that all the domains other
than the FAB domain are necessary for the Racmediated membrane ru‚ing.
The Myc-frabin-b-induced membrane ru‚ing was
inhibited by cotransfection of Myc-N17Rac1, but not
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3053
Figure 4 Schematic representation of mutant proteins of frabin and their activities. The positions of deletions in each mutant are
described under Materials and methods. ++, more than 60%; +, 40 ± 60%; +, 2 ± 5%, 7, less than 2%; and N.D., not
determined
by cotransfection of Myc-N-WASP-CRIB (Figure
5A ± C). We have previously shown that the FAB
domain in addition to the DH domain and the ®rst PH
domain is necessary for the ®lopodium formation in
COS7 cells (Umikawa et al., 1999). Taken together, it
is likely that in NIH3T3 cells, frabin induces mainly
membrane ru‚ing through the Cdc42-independent
activation of Rac, and that the FYVE domain and
the second PH domain in addition to the DH domain
and the ®rst PH domain, but not FAB domain, are
necessary for the ecient activation of Rac, although
frabin slightly induces the activation of Cdc42 which
induces ®lopodium formation at cell ± cell contact sites.
Direct activation of Cdc42 and Cdc42-independent
indirect activation of Rac by frabin in L cells
We next examined the activities of frabin in another
cell line, L cells. When L cells were transfected with
Myc-V12Cdc42, ®lopodia were observed in more than
80% of the Myc-V12Cdc42-transfected cells (Figure
6A,B). However, the frabin-GFP-expressing L cells,
which did not form cell ± cell contacts at a low cell
density, did not induce ®lopodium formation, but
induced membrane ru‚ing as observed in the MycV12Rac1-transfected cells (Figure 6C,D). Frabin-GFP
was colocalized with F-actin at the ru‚ing sites. The
radial array of actin bundles were also observed in the
frabin-GFP-expressing cells, but not in the MycV12Rac1-transfected cells. In contrast, when the cells
were cultured at a high cell density, more than 70% of
the frabin-GFP expressing cells formed ®lopodia at
Figure 5 Myc-Frabin-b-induced membrane ru‚ing in a Cdc42independent and Rac-dependent manner in NIH3T3 cells. Mycfrabin-b plus Myc-N17Rac1 or Myc-N-WASP-CRIB was transfected by use of the liposome-mediated gene transfection. MycN17Rac1 or Myc-N-WASP-CRIB was assumed to be coexpressed
with Myc-frabin-b. The cells were stained with rhodaminephalloidin. (A) Myc-Frabin-b-expressing cells; (B) cells expressing
Myc-frabin-b and Myc-N17Rac1; and (C) cells expressing Mycfrabin-b and Myc-N-WASP-CRIB. Bars, 10 mm
cell ± cell contact sites (Figure 6E, see also Figure 7A).
In addition, lamellipodia were observed between the
®lopodia. These lamellipodia were not observed in
Myc-V12Cdc42-transfected cells even at cell ± cell
contact sites (Figure 6F), suggesting that the frabinGFP-induced lamellipodium formation is not simply
mediated by Cdc42. Similar results were obtained using
Oncogene
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3054
Figure 7 Involvement of both Cdc42 and Rac in frabin-induced
formation of ®lopodia and lamellipodia in L cells. Frabin-GFP
and Myc-N17Rac1 or Myc-N-WASP-CRIB were transfected by
use of the liposome-mediated gene transfection. Myc-N17Rac1 or
Myc-N-WASP-CRIB was assumed to be coexpressed with frabinGFP. The cells were stained with rhodamine-phalloidin. (A)
Frabin-GFP-expressing cells; (B) cells expressing frabin-GFP and
Myc-N-WASP-CRIB; (C) cells expressing frabin-GFP and MycN17Rac1 (with cell ± cell contacts); (D) cells expressing with frabinGFP and Myc-N17Rac1 (without cell ± cell contacts). Bars, 10 mm
Figure 6 Frabin-induced formation of ®lopodia and lamellipodia in L cells. Frabin-GFP, Myc-V12Cdc42, or Myc-V12Rac1
was transfected by use of the liposome-mediated gene transfection. Myc-V12Cdc42, or Myc-V12Rac1 was assumed to be
coexpressed with GFP alone. The cells were stained with
rhodamine-phalloidin. (A) Control cells; (B) Myc-V12Cdc42transfected cells; (C) Frabin-GFP-expressing cells which did not
form cell ± cell contacts; (D) Myc-V12Rac1-transfected cells: (E)
frabin-GFP-expressing cells which formed cell ± cell contacts; and
(F) Myc-V12Cdc42-transfected cells which formed cell ± cell
contacts. More than 80% of the Myc-V12Cdc42-transfected cells,
which did not form cell ± cell contacts, showed ®lopodium
formation. More than 80% of the frabin-GFP-expressing cells,
which did not form cell ± cell contacts, showed membrane ru‚ing.
More than 80% of the Myc-V12Rac1-transfected cells showed
membrane ru‚ing. More than 70% of the frabin-GFP-expressing
cells, which formed cell ± cell contacts, showed the formation of
®lopodia and lamellipodia at cell ± cell contact sites. More than
80% of the Myc-V12Cdc42-transfected cells, which formed cell ±
cell contacts, showed ®lopodium formation, but lamellipodia were
not observed in these cells. Bars, 10 mm
Myc-frabin-a (see Figure 8A). These results indicate
that frabin induces the formation of not only ®lopodia
but also lamellipodia in L cells, and that the
®lopodium formation is related to cell ± cell contacts.
To con®rm that the frabin-induced ®lopodium
formation is mediated through the activation of
Cdc42 in L cells, frabin-GFP was cotransfected with
Myc-N-WASP-CRIB. This cotransfection of Myc-NWASP-CRIB completely inhibited the frabin-induced
®lopodium formation (Figure 7A,B). However, lamellipodia were still observed in the cells cotransfected with
Myc-N-WASP-CRIB. Cotransfection of Myc-N17Rac1
completely inhibited the frabin-GFP-induced lamellipodium formation, but not the ®lopodium formation
(Figure 7C). Consistent with the results obtained with
NIH3T3 cells, it is likely that frabin induces not only
Oncogene
the direct activation of Cdc42, which induces ®lopodium formation, but also the Cdc42-independent
indirect activation of Rac, which induces lamellipodium formation.
Cotransfection of frabin-GFP and Myc-N17Rac1
showed ®lopodium formation in L cells even when the
cells did not form cell ± cell contacts (Figure 7D). This
result, together with the results shown in Figure 6C,
and D, suggests that the Cdc42-induced ®lopodia are
apparently masked by the Rac-induced lamellipodia or
that Rac inhibits the Cdc42-induced ®lopodium
formation in an unknown manner.
Domains required for the formation of filopodia and
lamellipodia in L cells
We determined the domains of frabin required for the
formation of ®lopodia and lamellipodia in L cells. A
series of the Myc-tagged deletion mutants were
transiently expressed in L cells. Myc-Frabin-a induced
the formation of ®lopodia and lamellipodia at cell ± cell
contact sites and was colocalized with F-actin there
(Figure 8A, see also Figure 4). Consistent with our
previous report that the FAB domain is necessary for
®lopodium formation in COS7 cells (Umikawa et al.,
1999), Myc-frabin-b did not induce the formation of
®lopodia or lamellipodia and was di€usely distributed
throughout the cytoplasm (Figure 8B, see also Figure
4). However, in about 2 ± 5% of the cells expressing
Myc-frabin-b at a high level, ®lopodia and lamellipodia
were observed (see Figure 4). Myc-Frabin-c did not
induce morphological change and was colocalized with
F-actin (Figure 8C, see also Figure 4). Myc-Frabin-d
did not induce the formation of ®lopodia or ru‚es
(Figure 8D, see also Figure 4). However, in the cells
expressing Myc-frabin-d at a high level, ®lopodia were
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
observed not only at cell ± cell contact sites but also at
free edges (Figure 8E, see also Figure 4). Myc-Frabin-d
was colocalized with F-actin. In this case, lamellipodia
were not formed between the ®lopodia. Myc-Frabin-e
did not induce morphological change and was di€usely
distributed throughout the cytoplasm (Figure 8F, see
also Figure 4). These results indicate that the FAB
domain in addition to the DH domain and the ®rst PH
domain is necessary for the activation of Cdc42, which
induces ®lopodium formation in L cells, and that the
FYVE domain and the second PH domain in addition
to the DH domain and the ®rst PH domain are
necessary for the ecient activation of Rac, which
induces lamellipodium formation in L cells.
To further examine the involvement of each domain
in the formation of ®lopodia and lamellipodia, small
deletions were introduced into the most conserved
regions of respective domains. The mutant containing a
small deletion in the DH domain (Myc-frabin-f) or the
®rst PH domain (Myc-frabin-g) did not induce the
formation of ®lopodia or lamellipodia (see Figure 4).
Filopodia and lamellipodia were observed in about 2 ±
5% of the cells expressing the mutant containing a
small deletion in the FYVE domain (Myc-frabin-h).
These results indicate that all of these domains are
necessary for the ecient formation of ®lopodia and
lamellipodia in L cells.
Figure 8 Domains of frabin required for the formation of
®lopodia and lamellipodia in L cells. Each Myc-tagged mutant of
frabin was transfected by use of the liposome-mediated gene
transfection. The cells were doubly stained with the anti-Myc
antibody and rhodamine-phalloidin. The anti-Myc antibody was
visualized with an FITC-conjugated anti-mouse antibody. (A)
Myc-frabin-a; (B) Myc-frabin-b; (C) Myc-frabin-c; (D) Mycfrabin-d; (E) Myc-frabin-d at a high expression level; and (F)
Myc-frabin-e. Bars, 10 mm
Figure 9 Subcellular localization of various mutants of frabin.
Each Myc-tagged mutant was transfected in L cells by use of the
liposome-mediated gene transfection. The cell lysates were
fractionated by centrifugation. A comparable amount of each
fraction was subjected to SDS ± PAGE (10% polyacrylamide gel),
followed by Western blot analysis using the anti-Myc antibody.
low TX insol., low speed Triton X-100-insoluble fraction; high
TX insol., high speed Triton X-100-insoluble fraction; and TX
soluble, Triton X-100-soluble fraction
3055
Involvement of the FYVE domain and the second PH
domain of frabin in the association with the cytoskeleton
The observation, that the mutant lacking the FAB
domain (Myc-frabin-b) was colocalized with F-actin
at the ru‚ing sites, suggests that frabin contains a
domain interacting with the cytoskeleton other than
the FAB domain. To investigate this possibility,
subcellular fractionation experiments were performed
using the deletion mutants of frabin. L cells were
transfected with each Myc-tagged mutant of frabin
shown in Figure 4 and lysed in a bu€er containing
Triton X-100. Each lysate was subsequently fractionated, followed by Western blot analysis using the
anti-Myc antibody. Although a major portion of
Myc-frabin-a was detected in the Triton X-100soluble fraction, a signi®cant portion was recovered
in the low speed Triton X-100-insoluble fraction but
not in the high speed insoluble fraction (Figure 9).
Because the low speed insoluble fraction contains
large cytoskeletal components, this result suggests
that full-length frabin is associated with the cytoskeleton. Myc-Frabin-b, lacking the FAB domain, was
recovered in the low speed Triton X-100-insoluble
fraction as well as the high speed insoluble fraction.
Although a little portion of Myc-frabin-e, containing
the DH domain and the ®rst PH domain, was
recovered in the low speed insoluble fraction, the
ratio of Myc-frabin-e recovered in the low speed
insoluble fraction was far less than that of Mycfrabin-b. Furthermore, Myc-frabin-n, containing the
Oncogene
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3056
FYVE domain and the second PH domain, was
recovered in the low speed Triton X-100-insoluble
fraction. These results indicate that frabin contains
another domain(s) interacting with the cytoskeleton in
the C-terminal region in addition to the FAB
domain.
To further determine the mininum domain of the Cterminal region for the association with the cytoskeleton, other mutants were analysed by the same assay.
Myc-Frabin-1, lacking the FAB domain and the
FYVE domain, was recovered in the low speed Triton
X-100-insoluble fraction (Figure 9). Myc-Frabin-m,
lacking the FAB domain and the second PH domain,
was also recovered in this fraction. These results
suggest that both the FYVE domain and the second
PH domain are involved in the association with the
cytoskeleton.
Discussion
We have previously shown that frabin induces
®lopodium formation and JNK activation in Swiss
3T3 and COS7 cells as described for Cdc42 (Obaishi et
al., 1998; Umikawa et al., 1999). We have not shown
the direct evidence that these activities of frabin are
mediated by Cdc42, but suggested it on the basis of
two lines of evidence (1) that these activities of frabin
are similar to those observed by a dominant active
mutant of Cdc42 (V12Cdc42) (Nobes and Hall, 1995;
Kozma et al., 1995; Nagata et al., 1998); and (2) that
the fragment containing the DH domain and the ®rst
PH domain of frabin shows GEP activity speci®c for
Cdc42 in a cell-free system (Umikawa et al., 1999). We
have ®rst shown here that the frabin-induced ®lopodium formation is inhibited by cotransfection of the
Cdc42-binding domain of N-WASP (N-WASP-CRIB)
in NIH3T3 and L cells. This result has provided
additional evidence that the frabin-induced ®lopodium
formation is mediated by Cdc42.
We have moreover shown here in NIH3T3 and L
cells that frabin induces the formation of not only
®lopodia but also lamellipodia or ru‚es. The frabininduced ®lopodium formation is inhibited by NWASP-CRIB, but not by a dominant negative mutant
of Rac1 (N17Rac1), suggesting that this e€ect of frabin
is mediated by Cdc42, but not by Rac. The frabininduced formation of lamellipodia or ru‚es is inhibited
by N17Rac1, but not by N-WASP-CRIB. Moreover,
this e€ect of frabin is not mimicked by a dominant
active mutant of Cdc42 (V12Cdc42) which induces
®lopodium formation. These results suggest that the
frabin-induced formation of lamellipodia or ru‚es is
mediated by Rac, but not by Cdc42. In Swiss 3T3 cells,
Cdc42 induces the activation of Rac in a still unknown
mechanism (Nobes and Hall, 1995; Kozma et al.,
1995), but in NIH3T3 and L cells the frabin-induced
activation of Rac is not simply mediated by Cdc42.
We have previously shown that at least the fragment
containing the DH domain and the ®rst PH domain is
active on Cdc42 but inactive on RhoA and Rac1 in a
cell-free system, and that the FAB domain in addition
to the DH domain and the ®rst PH domain is
necessary for ®lopodium formation in COS7 cells
(Umikawa et al., 1999). We have con®rmed here this
latter result in L cells also. In contrast, the FYVE
Oncogene
domain and the second PH domain in addition to the
DH domain and the ®rst PH domain, but not the FAB
domain, are necessary for the formation of lamellipodia and ru‚es in L and NIH3T3 cells, respectively.
These results suggest that the two actions of frabin are
mediated through its di€erent domains as schematically
shown in Figure 10. Deletion analysis of frabin has
revealed that the DH domain of frabin is necessary for
the activation of both Cdc42 and Rac. Because frabin
is active on Cdc42, but not on Rac1 (Umikawa et al.,
1999), this result suggests that frabin induces the
activation of Rac through Cdc42, but is apparently
inconsistent with another result that the frabin-induced
activation of Rac is not inhibited by N-WASP-CRIB.
The reason for this inconsistency is currently unknown,
and further studies are necessary to understand how
frabin indirectly induces the activation of Rac.
Most GEPs for the Rho family small G proteins
contain protein- or lipid-interaction domains in addition to the PH domain adjacent to the DH domain
(Cerione and Zheng, 1996). In some cases, proper
cellular localization of GEPs through these domains is
necessary for their catalytic activity (Whitehead et al.,
1995; Zheng et al., 1996b; Stam et al., 1997; Glaven et
al., 1999). For example, membrane localization of
Tiam1, a GEP speci®c for Rac (Habets et al., 1994), is
required for membrane ru‚ing (Michiels et al., 1997).
This localization is mediated by the ®rst PH domain at
the N-terminal region (Stam et al., 1997; Michiels et
al., 1997). In the case of frabin, its association with the
actin cytoskeleton through the FAB domain is
necessary for the ®lopodium formation (Umikawa et
al., 1999). In contrast, the FAB domain is not required
for the activation of Rac, but the FYVE domain and
the second PH domain are necessary for the ecient
activation of Rac. Subcellular fractionation experiments have revealed that these domains are involved in
the association of frabin with the cytoskeleton. Thus,
frabin may be localized at the di€erent intracellular
Figure 10 A model for the signaling pathways of frabin. The Nterminal region of frabin containing the FAB domain, the DH
domain, and the ®rst PH domain, induces ®lopodium formation
through the direct activation of Cdc42. The C-terminal region of
frabin containing the DH domain, the ®rst PH domain, the
FYVE domain, and the second PH domain, induces the
formation of lamellipodia and ru‚es through the indirect
activation of Rac
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
sites by two independent cytoskeleton-association
domains and regulate di€erent signaling pathways at
the respective sites. By these two actions, frabin may
induce the formation of two di€erent actin-based
structures, ®lopodia and lamellipodia.
We have analysed here the actions of frabin by use
of the two cDNA transfer methods: the retrovirusmediated gene transfer and the vector transfection. The
expression vectors (pMX) of frabin used here are
designed to produce lower levels of proteins than the
expression vectors (pCMV) used previously (Obaishi et
al., 1998; Umikawa et al., 1999). We have previously
shown that the high expression level of Myc-frabin-b,
lacking the FAB domain, or Myc-frabin-d, lacking the
FYVE domain and the second PH domain, induces the
formation or ®lopodia or ru‚es in COS7 cells,
respectively (Umikawa et al., 1999). However, these
morphological changes are dependent on their expression levels. The high expression level of Myc-frabin-b
induces lamellipodium formation in about 2 ± 5% of its
expressing L cells, but the low expression level does
not. Similarly, the high expression level of Myc-frabin-d
induces ®lopodium formation in L cells, but the low
expression level does not. The FAB domain or the
FYVE domain and the second PH domain may be
necessary for the ecient activation of Rac or Cdc42,
respectively.
Materials and methods
Construction of expression vectors
Expression vectors of frabin were constructed in pMXIIMyc. pMXII was constructed by ligating the oligonucleotides, TCG AGC TCG AGG CGG CCG CAC GCG TAA
GCT TGA ATT CGT CGA CCT CGA GG/GAT CCC
TCG AGG TCG ACG AAT TCA AGC TTA CGC GTG
CGG CCG CCT CGA GC, into BamHI/SalI sites of pMX, a
Moloney murine leukemia virus promoter driven retrovirus
expression (Onishi et al., 1996). pMXII-Myc was constructed
by ligating the oligonucleotides, GAT CGG CCA CCA TGG
AGC AGA AGC TTA TCA GCG AGG AGG ACC TGG
AAT TCC CGG GGA TCC GTC GAC CTG CAG GCG
GCC GCT GAC TGA CTG AGG/GGC CCC TCA GTC
AGT CAG CGG CCG CCT GCA GGT CGA CGG ATC
CCC GGG AAT TCC AGG TCC TCC TCG CTG ATA
AGC TTC TGC TCC ATG GAG GCC, into BamHI/NotI
sites of pMXII. Various pMXII-Myc constructs of frabin
shown in Figure 4 contained the following amino acid
residues (aa): Myc-frabin-a, aa 1 ± 766 (full-length); Mycfrabin-b, aa 151 ± 766; Myc-frabin-c, aa 1 ± 150; Myc-frabin-d,
aa 1 ± 539; Myc-frabin-e, aa 169 ± 539; Myc-frabin-f, aa 1 ±
766 carrying an internal deletion of aa 353 ± 362; Myc-frabing, aa 1 ± 766 carrying an internal deletion of aa 442 ± 456;
Myc-frabin-h, aa 1 ± 766 carrying an internal deletion of 11
582 ± 593; Myc-frabin-i, aa 1 ± 626; Myc-frabin-j, aa 151 ± 766
carrying an internal deletion of aa 353 ± 362; Myc-frabin-k,
aa 151 ± 766 carrying an internal deletion of aa 442 ± 456;
Myc-frabin-1, aa 151 ± 766 carrying an internal deletion of aa
582 ± 593; Myc-frabin-m, aa 151 ± 626; and Myc-frabin-n, aa
540 ± 766. An expression vector of Myc-V12Cdc42, MycN17Cdc42, Myc-V12Rac1, or Myc-N17Rac1 was constructed
in pMX-IRES-GFP to produce both each Myc-tagged
protein and GFP. IRES-GFP sequence was ampli®ed from
pIRES-EGFP (Clontech) and inserted into pMX, yielding
pMX-IRES-GFP. A cDNA fragment encoding EGFP was
ampli®ed from pEGFP-N1 (Clontech) and cloned into
pMXII, yielding pMXII-EGFPN. A cDNA fragment encoding full-length frabin was cloned into pMXII-EGFPN at the
N-terminus of EGFP. An expression vector of N-WASPCRIB, corresponding to aa 124 ± 274 of N-WASP, was
constructed in pEF-BOS-Myc (Komuro et al., 1996).
3057
Cells and transfection
NIH3T3 and L cells were cultured in Dulbecco's modi®ed
Eagle's medium supplemented with 10% fetal calf serum.
Cells expressing various exogenous proteins were prepared by
use of retrovirus-mediated or liposome-mediated gene
transfection. The retrovirus-mediated gene transfection was
done as described (Onishi et al., 1996) with a slight
modi®cation. Brie¯y, to produce retroviruses, EcoPack-293
cells (Clonetech) were transfected with the constructs
described above using Lipofectamine reagent (GIBCO-BRL)
according to the manufacturer's protocol. After the cells were
cultured for 48 h, the retroviral supernatant of the culture
medium was collected and used to infect NIH3T3 cells. After
the retrovirus infection, the cells were cultured for 30 h,
replated onto glass coverslips, further cultured for 16 h, and
subjected to immuno¯uorescence microscopy. The liposomemediated gene transfection was done as described (Takahashi
et al., 1999). Brie¯y, cells were transfected with the constructs
described above by use of Lipofectamine reagent. After the
cells were cultured for 24 h, the cells were replated onto glass
coverslips, further cultured for 16 h, and subjected to
immuno¯uorescence microscopy. Immuno¯uorescence microscopy was done as described (Takahashi et al., 1999).
Triton X-100 fractionation
L cells were transfected by use of Lipofectamine reagent
(GIBCO-BRL). After 2-day culture, the cells from each 10cm dish were lysed with 300 ml of 20 mM HEPES/NaOH at
pH 7.4 containing 1% (w/v) Triton X-100 on ice for 5 min.
The lysate was centrifuged at 15 000 g for 5 min and the
pellet was taken as `low speed Triton X-100-insoluble'
fraction. The supernatant was centrifuged at 100 000 g for
30 min. The resulting supernatant represents the `Triton X100-soluble' fraction and the pellet represents `high speed
Triton X-100-insoluble' fraction. A comparable amount of
each fraction was subjected to SDS ± PAGE, followed by
Western blot analysis with the monoclonal anti-Myc antibody (American Type Culture Collection).
Acknowledgments
We thank Drs Sh Tsukita and A Nagafuchi (Kyoto
University, Kyoto, Japan) for providing us with L cells,
and Drs T Takenawa and H Miki (Tokyo University,
Tokyo, Japan) for N-WASP cDNA. We are also grateful
to Dr T Kitamura (Tokyo University, Tokyo, Japan) for
the pMX vector.
References
Cerione RA and Zheng Y. (1996). Curr. Opin. Cell Biol., 8,
216 ± 222.
Glaven JA, Whitehead I, Bagrodia S, Kay R and Cerione
RA. (1999). J. Biol. Chem., 274, 2279 ± 2285.
Habets GG, Scholtes EH, Zuydgeest D, van der Kammen
RA, Stam JC, Berns A and Collard JG. (1994). Cell, 77,
537 ± 549.
Hall A. (1998). Science, 279, 509 ± 514.
Oncogene
Frabin-induced activation of Cdc42 and Rac
Y Ono et al
3058
Oncogene
Komuro R, Sasaki T, Takaishi K, Orita S and Takai Y.
(1996). Genes Cells, 1, 943 ± 951.
Kozma R, Ahmed S, Best A and Lim L. (1995). Mol. Cell.
Biol., 15, 1942 ± 1952.
Michiels F, Stam JC, Hordijk PL, van der Kammen RA,
Ruuls-Van Stalle L, Feltkamp CA and Collard JG. (1997).
J. Cell Biol., 137, 387 ± 398.
Miki H, Sasaki T, Takai Y and Takenawa T. (1998). Nature,
391, 93 ± 96.
Nagata K, Driessens M, Lamarche N, Gorski JL and Hall A.
(1998). J. Biol. Chem., 273, 15453 ± 15457.
Nobes CD and Hall A. (1995). Cell, 81, 53 ± 62.
Obaishi H, Nakanishi H, Mandai K, Satoh K, Satoh A,
Takahashi K, Miyahara M, Nishioka H, Takaishi K and
Takai Y. (1998). J. Biol. Chem., 273, 18697 ± 18700.
Onishi M, Kinoshita S, Morikawa Y, Shibuya A, Phillips J,
Lanier LL, Gorman DM, Nolan GP, Miyajima A and
Kitamura T. (1996). Exp. Hematol., 24, 324 ± 329.
Pasteris NG, Cadle A, Logie LJ, Porteous MEM, Schwartz
CE, Stevenson RE, Glover TW, Wilroy RS and Gorski JL.
(1994). Cell, 79, 669 ± 678.
Stam JC, Sander EE, Michiels F, van Leeuwen FN, Kain
HE, van der Kammen RA and Collard JG. (1997). J. Biol.
Chem., 272, 28447 ± 28454.
Symons M. (1996). Trends. Biochem. Sci., 21, 178 ± 181.
Takahashi K, Nakanishi H, Miyahara M, Mandai K, Satoh
K, Satoh A, Nishioka H, Aoki J, Nomoto A, Mizoguchi A
and Takai Y. (1999). J. Cell Biol., 145, 539 ± 549.
Takai Y, Sasaki T, Tanaka K and Nakanishi H. (1995).
Trends. Biochem. Sci., 20, 227 ± 231.
Umikawa M, Obaishi H, Nakanishi H, Satoh-Horikawa K,
Takahashi K, Hotta I, Matsuura Y and Takai Y. (1999). J.
Biol. Chem., 274, 25197 ± 25200.
Whitehead I, Kirk H, Tognon C, Trigo-Gonzalez G and Kay
R. (1995). J. Biol. Chem., 270, 18388 ± 18395.
Zheng Y, Fischer D, Santos MF, Tigri G, Pasteris NG,
Gorski JL and Xu Y. (1996a). J. Biol. Chem., 271, 33169 ±
33172.
Zheng Y, Zangrilli D, Cerione RA and Eva A. (1996b). J.
Biol. Chem., 271, 19017 ± 19020.