The guanine nucleotide exchange factor RasGRF1 directly
binds microtubules via DHPH2-mediated interaction
Greta Forlani, Simona Baldassa, Paola Lavagni, Emmapaola Sturani and Renata Zippel
Department of Biomolecular Sciences and Biotechnology, University of Milan, Italy
Microtubules are crucial elements in the generation
and maintenance of neuronal morphology. They play
a role not only in the establishment of neuronal out-
growth during brain development but are also involved
in the remodeling of mature neurons [1].
The dynamics of microtubules as well as their inter-
actions with other cytoskeletal elements are regulated
by microtubule-binding proteins [2]. Some of them,
such as microtubule-associated proteins (MAPs), pro-
mote the assembly of microtubules, whereas others,
such as stathmin, increase microtubule instability.
Other microtubule-binding proteins are involved in the
transport of organelles and cargos along the micro-
tubule network. Microtubule dynamics is modulated by
different extracellular signaling molecules [3], and the
monomeric GTP-binding proteins Rho and Rac are
implicated in these processes [4]. In fibroblasts, Rho
activity induces microtubule stabilization independ-
ently from its effect on actin filaments [5,6]. p21-activa-
ted kinase, one the effectors of Rac, phosphorylates
stathmin, thus inhibiting its destabilizing effect on
microtubules [7].
RasGRF is a family of guanine nucleotide exchange
factors consisting of two members: RasGRF1 [8–10]
exclusively expressed in neurons of the central nervous
system [11] and in b cells of the pancreas [12]; Ras-
GRF2, highly expressed in the brain but also present

the DHPH2 module is largely responsible for RasGRF1–microtubule inter-
action. In vivo colocalization of RasGRF1 and microtubules was also
observed by fluorescence confocal microscopy in nonneuronal cells after
stimulation with an oxidative stress agent and in highly differentiated
neuron-like cells. Identification of microtubules as new binding partners of
RasGRF1 may help to elucidate the signaling network in which RasGRF1
is involved.
Abbreviations
DH, Dbl-homology domain; ERK, extracellular regulated kinase; GEF, guanine nucleotide exchange factor; GST, glutathione S-transferase;
LPA, lysophosphatidic acid; MAP, microtubule-associated protein; MAPK, mitogen-activated protein kinase; MBP, maltose-binding protein;
PH, Pleckstrin homology domain.
FEBS Journal 273 (2006) 2127–2138 ª 2006 The Authors Journal compilation ª 2006 FEBS 2127
guanine nucleotide exchange domain for Ras, and the
Dbl-homology domain (DH) and the Pleckstrin-
homology domain (PH) which are present in the cen-
tral part of the molecule are responsible for Rac
exchange activity [14–17]. The N-terminal region con-
tains a PH domain, a coiled-coil region, and an IQ
domain which binds calmodulin in a calcium-depend-
ent manner [18]. Very recently, a ‘neuronal domain’
located in the central part of RasGRF1, but absent
from RasGRF2, has been identified [19]. This region
has been found to be responsible for the physical inter-
action of RasGRF1 with the NR2B subunit of the
NMDA subtype of glutamate receptor.
RasGRF1 is activated by G-protein-coupled recep-
tors [20,21] and requires both calcium and calmodulin
to exert its activity on Ras [18,21]. The protein is phos-
phorylated on serine, threonine and tyrosine residues
in vivo and is a substrate for different kinases in vitro

Interaction of RasGRFs with microtubules
RasGRF1 is expressed in adult brain and, as previ-
ously shown, it is enriched in postsynaptic densities
[26]. However, RasGRF1 is also present in the cytoso-
lic fraction (Fig. 1A; S100). As well as the expected
140-kDa band, antibodies to RasGRF1 detected in
brain extracts another band of slightly lower molecular
mass which is also present in RasGRF1 knockout
mice. This band was isolated, analyzed by MALDI
TOF, and found to correspond to RasGRF2
(Fig. 1B,C).
In the course of a more general study on proteins
that interact with RasGRF1, we investigated its poss-
ible association with tubulin. Unpolymerized tubulin
did not significantly associate with RasGRF1, as indi-
cated by coimmunoprecipitation experiments on cyto-
solic brain extracts (data not shown). Further
Fig. 1. Distribution of RasGRF1 and Ras-
GRF2 in mouse brain extracts. (A) Equal
amounts of protein from total brain extracts,
particulate fraction (P100) and soluble frac-
tion (S100) were analyzed with antibodies to
RasGRF1. (B) Brain extracts obtained from
wild-type and RasGRF1 knockout mice [28]
were immunoprecipitated with antibodies to
RasGRF1 (lane 2, 4) or with nonrelated IgG
(lanes 1 and 3). Immunoprecipitates were
analyzed by SDS ⁄ PAGE and silver staining.
(C) The lower band present in knockout
mice was isolated and analyzed by MALDI

lanes 1 and 2).
RasGRF association with microtubules is
mediated by neither motor proteins nor MAPs
It has been demonstrated that RasGRF1 binds
IB2 ⁄ JIP2 [33], a scaffold protein for the Jun N-ter-
minal kinase signaling pathway. JIP proteins also link
the motor proteins kinesins with the cargo complex to
be transported along the microtubules [34]. In vitro
microtubule-binding motor proteins can be released
from microtubules by treatment with high concentra-
tion of ATP [32]. To verify that motor proteins medi-
ate RasGRF–microtubule interaction, mouse brain
high-speed supernatant was subjected to microtubule
assembly in the presence of either ATP or the non-
A
B
Fig. 2. RasGRFs cosediment with in vitro assembled microtubules.
(A) Cytosolic high-speed supernatant of a mouse brain homogenate
(Input) was used for the microtubule cosedimentation assay (see
Experimental procedures). Microtubules were polymerized at
37 °C, then loaded on a sucrose cushion and sedimented by cen-
trifugation. The microtubule pellet (MT) was resuspended in the
same volume as the supernatant (SN), and an equal volume of
each fraction and the Input were resolved by electrophoresis, and
analyzed by western blotting. (B) Mouse brain high-speed supernat-
ant was used in a microtubule cosedimentation assay, following
different procedures: microtubules were induced to polymerize at
37 °C with (lanes 1, 4) or without (lanes 2, 5) 10 l
M taxol. To avoid
tubulin polymerization, the extracts were maintained on ice (lanes

Also MAPs seem not to be involved in the in vitro
association of RasGRFs with microtubules. In fact
treatment of isolated taxol-stabilized microtubules with
high salt concentration, a condition reported to disso-
ciate MAPs [35], did not affect the amount of Ras-
GRFs in the pellet (Fig. 3B). Urea also did not
dissociate RasGRFs from microtubules.
These results indicate that neither motor proteins
nor MAPs mediate RasGRF interaction with micro-
tubules and suggest that this association is very tight.
The DHPH2 module directly interacts with
microtubules
To verify whether pure microtubules are also able to
bind RasGRF1, microtubules preassembled from pure
commercial tubulin were incubated with a small
aliquot of extracts (5 lg) of Hek293 cells expressing
RasGRF1. Microtubules were then recovered by
centrifugation and analyzed by western blotting for
RasGRF1. Figure 4 shows that most of the RasGRF1
cosedimented with pure microtubules. As expected in
the absence of preassembled microtubules, RasGRF1
was found in the supernatant.
The latter experiment shows that RasGRF1 interacts
with pure microtubules but its direct association is not
yet proven, as proteins present in Ras-GRF1 extracts
could mediate this interaction. To investigate this
point, purified tagged recombinant proteins coding for
different regions of RasGRF1 were prepared (Fig. 5A).
Microtubules preassembled from pure commercial tub-
ulin were incubated with purified proteins, and both

the right summarizes the results of binding assays shown in (B).
(B) Purified proteins MBP-PHCCIQ, MBP-DHPH2, MBP-DH, GST-
PH2, GST-Cat, MBP and GST (0.2 l
M) were incubated with (+)
or without (–) pure preassembled microtubules (10 l
M, relative to
tubulin dimers) in a microtubule binding assay (see Experimental
procedures). Input, pellet (P) and supernatant (SN), were analyzed
by western blotting with antibodies to MBP, GST or a-tubulin.
RasGRF1 directly binds microtubules G. Forlani et al.
2130 FEBS Journal 273 (2006) 2127–2138 ª 2006 The Authors Journal compilation ª 2006 FEBS
centrifugation, both supernatants and microtubule pel-
lets were analyzed by SDS ⁄ PAGE and Coomassie Brilli-
ant Blue staining (Fig. 6A). The intensity of the bands
was then determined by densitometry. Plots of the con-
centration of MBP-DHPH2 in the pellets versus total
MBP-DHPH2 protein added to the reaction mixture
are reported in Fig. 6B, and Scatchard analysis is shown
in Fig. 6C. The data reveal that DHPH2 binds microtu-
bules with high affinity, showing an estimated dissoci-
ation constant of  2 lm. The stoichiometry of the
interaction at saturating conditions of MBP-DHPH2 is
one MBP-DHPH2 molecule per two tubulin dimers.
An in vitro competition experiment was performed
to test whether the DHPH2 module is the only region
responsible for the interaction of the entire RasGRF1
molecule with microtubules. Pure microtubules were
incubated for 20 min with RasGRF1-containing cell
extracts (as in Fig. 4) and with increasing concentra-
tions of purified DHPH2 protein. As shown in Fig. 7,

was monitored for further 30 min. As shown in
Fig. 8 (depolymerization), stathmin caused a large
decrease in the steady-state level of polymerized tubulin,
and DHPH2 did not prevent stathmin-induced micro-
tubule depolymerization. Moreover the simultaneous
A
B
C
Fig. 6. Kinetics of DHPH2 binding to microtubules. (A) Constant
amounts of pure taxol-stabilized microtubules (5 l
M, relative to tub-
ulin dimer) were incubated with increasing concentrations of MBP-
DHPH2 (1–5 l
M). Input, microtubule pellet (MT) and supernatant
(SN) were resolved by SDS ⁄ PAGE and stained with Coomassie Bril-
liant Blue. (B) Plots of the amounts of MBP-DHPH2 in the pellets
(bound MBP-DHPH2) (l
M) as a function of total MBP-DHPH2 pro-
tein added to the binding assays (total MBP-DHPH2) (l
M) shown in
(A). The amounts of MBP-DHPH2 were quantitated by densitom-
etry. The intensity of single bands was compared with that calcul-
ated for known amounts of BSA used as standard control and was
expressed as micromolar. Results of the Scatchard analysis are
reported in (C).
Fig. 7. DHPH2 domain competes with RasGRF1 for microtubule
binding. Constant amounts of pure taxol-stabilized microtubules
(5 l
M, relative to tubulin dimer) were incubated with extracts (5 lg)
of HEK293 cells expressing RasGRF1 and increasing concentrations

stimuli have been reported to activate different kinases
of the MAPK family (reviewed in e.g. [41–45]). After
treatment with LPA and A23187 (30 min) or with
sodium arsenate (1 h), cells were fixed and immuno-
stained using monoclonal antibodies to tubulin and
polyclonal antibodies to RasGRF1 or the N-terminal
region of RasGRF1 [28]. Immunofluorescence was
then analyzed by confocal microscopy. In unstimulated
cells, we were unable to detect a significant colocaliza-
tion of microtubules with either full-length RasGRF1
or its N-terminal region. Moreover, treatment with the
calcium ionophore or LPA also did not have any dis-
cernible effect (data not shown). Conversely, when
cells were treated with sodium arsenate, the N-terminal
region of RasGRF1 clearly associated with microtu-
bules in a large proportion of the transfected cells
(compare Fig. 9C with Fig. 9F). Identical results
were obtained with sodium arsenite, another arsenic
Fig. 8. DHPH2 tandem does not affect in vitro microtubule dynam-
ics. In vitro tubulin polymerization ⁄ depolymerization assay. Tubulin
(40 l
M) polymerization was performed at 37 °C for 30 min, and
microtubule assembly was monitored at A
350
(polymerization).
Time-course of tubulin assembly in the presence of 10 l
M MBP-
DHPH2 (A) or 10 l
M MBP recombinant proteins (B). The effect of
stathmin (20 l

found in neurons. Thus SO5 differentiated cells were
stained for Ras-GRF1 and tubulin. As shown in
Fig. 10B, tubulin has the typical microtubule organiza-
tion of a neuronal cell. Ras-GRF1 staining was distri-
buted in the cell body and along the neurites and
excluded from the nucleus (Fig. 10A). In these cells,
RasGRF1 was found to partially colocalize with
microtubules mainly within the proximal region of
cellular processes (Fig. 10C), along those neurites in
which tubulin was well organized in microtubule bun-
dles (Fig. 10F), in the tips and in the varicosities. No
colocalization could be depicted in thinner neurites
with a less organized microtubular structure (Fig. 10F
left). We did not detect colocalization of RasGRF1
with the actin network (not shown).
Discussion
In this study, we provide evidence that RasGRF1
interacts both in vivo and in vitro with microtubules.
Both RasGRF1 and RasGRF2 present in the cytosolic
fraction of brain extracts bind microtubules, whereas
other proteins involved in the Ras pathway do not.
Neither high salt nor urea dissociates RasGRFs from
microtubules, indicating that both electrostatic and
hydrophobic interactions are involved in this tight
association. In particular, the lack of effect of high salt
suggests that MAPs are not involved in the interaction.
Also motor proteins, for instance kinesin, do not
appear to mediate this interaction, so that it is unlikely
that RasGRFs use microtubules as tracks for its trans-
port to the neurites.

microtubules from depolymerization induced by stath-
min. Moreover, expression of the DH-PH tandem did
not affect in vivo microtubule reorganization following
recovery after nocodazole washout (data not shown).
Thus, we can reasonably assert that the DHPH2 mod-
ule of RasGRF1 does not directly affect microtubule
stability. However, there is the possibility that Ras-
GRF1 may act as scaffolding for other proteins that
modulate microtubule dynamics. We are now investi-
gating this aspect using both the yeast two hybrid sys-
tem and affinity-based chromatography.
In vivo studies have shown that arsenic compounds,
which are known oxidative stress agents [38,39,44,45,
51,52], induce the interaction between the N-terminal
part of RasGRF1 and microtubules in COS7 cells,
whereas other stimuli, known to activate RasGRF1,
such as LPA and a calcium ionophore [18,21], do not.
In this regard, it can be recalled that the agents used
(ionophore and LPA on one side and arsenic com-
pounds on the other) activate different pathways and
most probably lead to different modifications of either
RasGRF1 or microtubule structure.
It is, however, difficult to understand the different
behavior of the entire RasGRF1 and its N-terminal
region after arsenate treatment. One possible explan-
ation is that RasGRF1 with its catalytic region
strongly activates the Ras pathway, leading to inhibi-
tion of the interaction. Interestingly, in differentiated
neuron-like cells in which microtubular organization is
very different, colocalization of the entire RasGRF1

was kindly provided by A. Colombatti (University of Udine,
Udine, Italy).
Cell culture and transfections
Human SK-N-BE neuroblastoma cells expressing cDNA
for HA-tagged-Ras-GRF1 (SO5 clone) [46] were cultured in
RPMI 1640 medium supplemented with fetal bovine serum
(Gibco, Invitrogen Corporation, Carlsbad, CA, USA).
HEK293 cells and COS7 cells grown in Dulbecco’s modi-
fied Eagle’s medium supplemented with 10% fetal bovine
serum were transfected using the Lipofectamine (Invitrogen,
Life Technologies) method according to the manufacturer’s
instructions. All media were from Gibco, Invitrogen
Corporation.
Immunofluorescence analysis
For immunofluorescence studies, COS7 cells, plated on
glass coverslip, were transfected with different constructs
RasGRF1 directly binds microtubules G. Forlani et al.
2134 FEBS Journal 273 (2006) 2127–2138 ª 2006 The Authors Journal compilation ª 2006 FEBS
and serum deprived over night before stimulation with
sodium arsenate (0.5 mm) for 1 h. SO5 cells were treated
with 10 lm retinoic acid. Seven days later, cells were used
for immunofluorescence analysis.
For tubulin and Ras-GRF1 staining, cells were fixed and
double-stained with polyclonal antibodies to Ras-GRF1
(Santa Cruz Biotechnology, Santa Cruz, CA, USA) or the
N-terminal region [28] and with monoclonal antibodies to
a-tubulin. Alexa Fluor 488-conjugated anti-rabbit IgG and
Alexa Fluor 594-conjugated anti-mouse IgG (Molecular
Probes, Eugene, OR, USA) were used as secondary anti-
bodies. Microscopic analysis was performed with a Leica

MgSO
4
,1mm dithiothreitol, 0.5 mm GTP) supplemented
with EDTA-free protease inhibitor cocktail (Roche Phar-
maceuticals, Basel, Switzerland) at a ratio of 1 mL per g
brain tissue. The homogenate was centrifuged at low speed
(1000 g, 10 min, 4 °C) followed by a high-speed (100 000 g,
60 min, 4 °C) step. The supernatant (Input), was then incu-
bated at 30 °C for 30 min in the presence of 20 lm taxol,
loaded on a 13% (w ⁄ v) sucrose cushion in the above buffer
and centrifuged at 45 000 g, for 30 min at 30 °C. Superna-
tant, usually 200 lL total volume, was collected and sup-
plemented with 4 · SDS sample buffer, and the pellet was
resuspended in SDS sample buffer in the same total volume
of supernatant. Input and equal volumes of the supernatant
and pellet were separated by SDS ⁄ PAGE and blotted on
nitrocellulose membranes (Schleicher & Schuell Bioscience,
Dassel, Germany), for western blotting. Bound antibodies
were visualized by enhanced chemiluminescence (ECL)
detection (Amersham Pharmacia Biotech, Milan, Italy)
using horseradish peroxidase-conjugated secondary antibo-
dies (Jackson Immunoresearch Laboratories, West Grove,
PA, USA).
Binding assays
For binding assays, microtubules were prepared from pure
bovine brain tubulin according to the protocol described
by the manufacturer (Cytoskeleton, Denver, CO, USA).
Extracts of RasGRF1-expressing cells or purified GST and
MBP fusion proteins were then incubated with preas-
sembled, taxol-stabilized, pure microtubules for 20 min at

ilizing activity of stathmin, microtubules were assembled as
described above and when the steady-state was reached
recombinant stathmin (20 lm) was added. Absorbance was
monitored for a further 30 min.
Antibodies and chemicals
Polyclonal antibodies to RasGRF1 (sc-224), Sos1 (sc-256),
Erk2 (sc-154), and K-Ras (sc-30) and monoclonal anti-
G. Forlani et al. RasGRF1 directly binds microtubules
FEBS Journal 273 (2006) 2127–2138 ª 2006 The Authors Journal compilation ª 2006 FEBS 2135
bodies to GST (sc-138) were from Santa Cruz Biothec-
nology (Santa Cruz, CA, USA). Monoclonal antibodies to
a-tubulin (B-5-1-2) and kinesin heavy chain (clone IBII)
were from Sigma (Milan, Italy). Polyclonal antibodies to
MBP were from New England Biolabs. Polyclonal anti-
bodies to the N-terminal region of RasGRF1 have been
previously described [28]. Chemicals were from Sigma
unless otherwise indicated.
Acknowledgements
We are grateful to N. Gnesutta for critical reading of
the manuscript, to U. Fascio for technical assistance
with the confocal microscope, and to G. Cappelletti
and M. V. Schiaffino for their valuable advice. This
work was supported by grants from Ministero dell’Ist-
ruzione, dell’Universita
`
e della Ricerca to R.Z. (COFIN
2003) and by FIRST 2003-4 to R.Z.
References
1 Diaz-Nido J (1997) The Cytoskeleton. BIOS Scientific
Publisher, Oxford.

analysis of human cDNAs encoding a 140-kDa brain
guanine nucleotide-exchange factor, Cdc25GEF, which
regulates the function of Ras. Gene 151, 279–284.
11 Zippel R, Orecchia S, Sturani E & Martegani E (1996)
The brain specific Ras exchange factor CDC25 Mm:
modulation of its activity through Gi-protein-mediated
signals. Oncogene 12, 2697–2703.
12 Font de Mora J, Esteban LM, Burks DJ, Nunez A,
Garces C, Garcia-Barrado MJ, Iglesias-Osma MC,
Moratinos J, Ward JM & Santos E (2003) Ras-GRF1
signaling is required for normal beta-cell
development and glucose homeostasis. EMBO J 22,
3039–3049.
13 Fam NP, Fan WT, Wang Z, Zhang LJ, Chen H &
Moran MF (1997) Cloning and characterization of Ras-
GRF2, a novel guanine nucleotide exchange factor for
Ras. Mol Cell Biol 17, 1396–1406.
14 Kiyono M, Satoh T & Kaziro Y (1999) G protein beta
gamma subunit-dependent Rac-guanine nucleotide
exchange activity of Ras-GRF1 ⁄ CDC25 (Mm). Proc
Natl Acad Sci USA 96, 4826–4831.
15 Kiyono M, Kaziro Y & Satoh T (2000) Induction of
rac-guanine nucleotide exchange activity of Ras-
GRF1 ⁄ CDC25 (Mm) following phosphorylation by the
nonreceptor tyrosine kinase Src. J Biol Chem. 275,
5441–5446.
16 Innocenti M, Zippel R, Brambilla R & Sturani E (1999)
CDC25 (Mm) ⁄ Ras-GRF1 regulates both Ras and Rac
signaling pathways. FEBS Lett 460, 357–362.
17 Fan WT, Koch CA, de Hoog CL, Fam NP & Moran

274, 37379–37384.
24 Shou C, Wurmser A, Suen KL, Barbacid M, Feig LA
& Ling K (1995) Differential response of the Ras
exchange factor, Ras-GRF to tyrosine kinase and G
protein mediated signals. Oncogene 10, 1887–1893.
25 Zippel R, Gnesutta N, Matus-Leibovitch N, Mancinelli
E, Saya D, Vogel Z & Sturani E (1997) Ras-GRF, the
activator of Ras, is expressed preferentially in mature
neurons of the central nervous system. Brain Res Mol
Brain Res 48, 140–144.
26 Sturani E, Abbondio A, Branduardi P, Ferrari C, Zip-
pel R, Martegani E, Vanoni M & Denis-Donini S
(1997) The Ras Guanine nucleotide Exchange Factor
CDC25Mm is present at the synaptic junction. Exp Cell
Res 235, 117–123.
27 Giese KP, Friedman E, Telliez JB, Fedorov NB, Wines
M, Feig LA & Silva AJ (2001) Hippocampus-dependent
learning and memory is impaired in mice lacking the
Ras-guanine-nucleotide releasing factor 1 (Ras-GRF1).
Neuropharmacology 41, 791–800.
28 Brambilla R, Gnesutta N, Minichiello L, White G, Roy-
lance AJ, Herron CE, Ramsey M, Wolfer DP, Cestari
V, Rossi-Arnaud C, et al. (1997) A role for the Ras sig-
nalling pathway in synaptic transmission and long-term
memory. Nature 390, 281–286.
29 Tonini R, Franceschetti S, Parolaro D, Sala M, Mancin-
elli E, Tininini S, Brusetti R, Sancini G, Brambilla R,
Martegani E, et al. (2001) Involvement of CDC25Mm ⁄
Ras-GRF1-dependent signaling in the control of neuro-
nal excitability. Mol Cell Neurosci 18, 691–701.

407, 111–115.
38 Porter AC, Fanger GR & Vaillancourt RR (1999) Sig-
nal transduction pathways regulated by arsenate and
arsenite. Oncogene 18, 7794–7802.
39 Tanaka-Kagawa T, Hanioka N, Yoshida H, Jinno H
& Ando M (2003) Arsenite and arsenate activate
extracellular signal-regulated kinases 1 ⁄ 2 by an epider-
mal growth factor receptor-mediated pathway in
normal human keratinocytes. Br J Dermatol 149,
1116–1127.
40 Ivanov PA, Chudinova EM & Nadezhdina ES (2003)
Disruption of microtubules inhibits cytoplasmic ribonu-
cleoprotein stress granule formation. Exp Cell Res 290,
227–233.
41 Kranenburg O & Moolenaar WH (2001) Ras-MAP
kinase signaling by lysophosphatidic acid and other G
protein-coupled receptor agonists. Oncogene 20, 1540–
1546.
42 Agell N, Bachs O, Rocamora N & Villalonga P (2002)
Modulation of the Ras ⁄ Raf ⁄ MEK ⁄ ERK pathway by
Ca(
2+
), and calmodulin. Cell Signal 14, 649–654.
43 Seger R & Krebs EG (1995) The MAPK signaling cas-
cade. FASEB J 9, 726–735.
44 Huang C, Li J, Ding M, Wang L, Shi X, Castranova V,
Vallyathan V, Ju G & Costa M (2001) Arsenic-induced
NFkappaB transactivation through Erks- and JNKs-
dependent pathways in mouse epidermal JB6 cells. Mol
Cell Biochem 222, 29–34.

50 Krendel M, Zenke FT & Bokoch GM (2002) Nucleotide
exchange factor GEF-H1 mediates cross-talk between
microtubules and the actin cytoskeleton. Nat Cell Biol
4, 294–301.
51 Huang C, Bode AM, Chen NY, Ma WY, Li J, Nomura
M & Dong Z (2001) Transactivation of AP-1 in AP-1-
luciferase reporter transgenic mice by arsenite and
arsenate. Anticancer Res 21, 261–267.
52 Gao N, Shen L, Zhang Z, Leonard SS, He H, Zhang
XG, Shi X & Jiang BH (2004) Arsenite induces HIF-
1alpha and VEGF through PI3K, Akt and reactive oxy-
gen species in DU145 human prostate carcinoma cells.
Mol Cell Biochem 255, 33–45.
53 Giasson BI, Sampathu DM, Wilson CA, Vogelsberg-
Ragaglia V, Mushynski WE & Lee VM (2002) The
environmental toxin arsenite induces tau hyperphos-
phorylation. Biochemistry 41, 15376–15387.
RasGRF1 directly binds microtubules G. Forlani et al.
2138 FEBS Journal 273 (2006) 2127–2138 ª 2006 The Authors Journal compilation ª 2006 FEBS