Oncogene (2001) 20, 3217 ± 3225
ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00
www.nature.com/onc
The Gem GTP-binding protein promotes morphological di€erentiation in
neuroblastoma
Alvaro Leone1,3, Nicholas Mitsiades2, Yvona Ward1, Beth Spinelli1, Vasiliki Poulaki2,
Maria Tsokos2 and Kathleen Kelly*,1
1
Cell and Cancer Biology Department, Medicine Branch, Division of Clinical Sciences, National Cancer Institute, Building 10,
Room 3B43, 9000, Rockville Pike, Bethesda, Maryland, MD 20892, USA; 2Pediatric Tumor Biology Section, Laboratory of
Pathology, Division of Clinical Sciences, National Cancer Institute, Bethesda, Maryland, USA
Gem is a small GTP-binding protein within the Ras
superfamily whose function has not been determined. We
report here that ectopic Gem expression is sucient to
stimulate cell ¯attening and neurite extension in N1E115 and SH-SY5Y neuroblastoma cells, suggesting a
role for Gem in cytoskeletal rearrangement and/or
morphological di€erentiation of neurons. Consistent with
this potential function, in clinical samples of neuroblastoma, Gem protein was most highly expressed within
cells which had di€erentiated to express ganglionic
morphology. Gem was also observed in developing
trigeminal nerve ganglia in 12.5 day mouse embryos,
demonstrating that Gem expression is a property of
normal ganglionic development. Although Gem expression is rare in epithelial and hematopoietic cancer cell
lines, constitutive Gem levels were detected in several
neuroblastoma cell lines and could be further induced as
much as 10-fold following treatment with PMA or the
acetylcholine muscarinic agonist, carbachol. Oncogene
(2001) 20, 3217 ± 3225.
Keywords: Gem; neuroblastoma; ganglia; morphology;
RGK proteins; cytoskeleton
Introduction
Gem (Maguire et al., 1994), Rad (Reynet and Kahn,
1993) and Rem (Finlin and Andres, 1997) are members
of a small GTP binding family of proteins within the
Ras superfamily, sometimes referred to as the RGK
(for Rad, Gem, and the mouse ortholog of Gem, Kir)
family. The basic structure of RGK proteins consists of
a Ras-related core, a non-CAAX containing carboxylterminal extension, and amino-terminal extensions. The
G3 motifs (DXXG) of RGK proteins are nonconserved relative to other small GTPases, consistent with
their low intrinsic GTPase activity (Cohen et al., 1994;
*Correspondence: K Kelly
3
Current address: Laboratorio di Istopatologia, ASL San CamilloForlanini, Via Portuense 332, 00152, Roma, Italia
Received 29 June 2000; revised 14 February 2001; accepted 26
February 2001
Finlin et al., 2000). It appears likely that either the
function of RGK proteins is not regulated via GTP
hydrolysis or that there is a unique mechanism for
GTPase activating protein (GAP) catalyzed GTP
hydrolysis relative to other Ras superfamily members.
Other potential mechanisms for modulating RGK
protein activity include transcriptional regulation,
described for Gem (Maguire et al., 1994) and Rem
(Finlin and Andres, 1997), and phosphorylation,
described for Gem, Rad, and Rem (Maguire et al.,
1994; Moyers et al., 1998; Finlin and Andres, 1999).
The RGK proteins share many structural features,
which suggest common mechanisms of regulation.
However, like other Ras superfamily members, Gem,
Rad, and Rem have relatively unique sequences in their
putative e€ector binding regions, implying distinct
protein binding capacities. Overexpression of mouse
Gem was found to induce invasive pseudohyphal growth
in Saccharomyces cerevisiae (Dorin et al., 1995).
Although there is no apparent Gem ortholog in yeast,
this assay most likely re¯ects the interaction of Gem with
a protein common to yeast and mammalian cells. Ges,
the likely human ortholog of mouse Rem, and Rem
recently were described to induce endothelial cell
sprouting (Pan et al., 2000). Speci®c biochemical
functions for the RGK proteins are as yet unknown.
Gem expression is highly regulated; Gem has been
shown to be an immediate early gene in primary
lymphocytes, monocytes, endothelial cells and human
embryonic ®broblasts (Maguire et al., 1994; Vanhove
et al., 1997). Gem RNA has been detected in several
di€erent human and murine tissues including thymus,
spleen, kidney, lung and testis (Maguire et al., 1994).
In contrast to primary cells, Gem expression was not
detectable in a variety of human hematopoietic tumorderived lines including T cells, B cells, myeloid, and
erythroid cell lines, either before or after treatment
with PMA (Maguire et al., 1994). Lack of expression in
hematopoietic cancers may result from cells being in
the wrong di€erentiation state or from a loss of
signaling pathways associated with transformation.
Contrary to a lack of Gem expression in epithelial
and hematopoietic cancer cell lines, here we describe
the constitutive expression of Gem in neuroblastoma
cell lines. In clinical samples of neuroblastoma, the
Gem promotes neurite extension
A Leone et al
3218
highest levels of Gem expression were observed in cells
undergoing gangliocytic di€erentiation. Signi®cantly,
we show that one aspect of neuronal morphological
di€erentiation, neurite extension, is induced in N1E115 and SH-SY5Y neuroblastoma cells following
ectopic Gem expression.
Results
CHP-126
Because Gem is known to be transcriptionally
regulated by receptor-mediated activation in lymphoid
and endothelial cells, we assessed Gem expression
following treatment of SH-SY5Y neuroblastoma cells
with various ligands which bind neuronal receptors
including NMDA, kainate, and carbachol. Carbachol,
an acetylcholine analog that has been demonstrated to
be an activator of G protein coupled muscarinic
receptors (Wess et al., 1997), increased Gem protein
above the constitutive expression level in various
experiments between 4- and 10-fold (Figure 2a). Gem
protein levels varied in parallel to Gem RNA (not
shown). Gem was also induced approximately 10-fold
by PMA treatment of SH-SY5Y (Figure 2a). Gem was
similarly induced following PMA treatment of IMR-
KCNR
We searched for a cell type that constitutively expresses
detectable Gem in order to identify a system in which
Gem function could be uniquely investigated. Constitutive Gem RNA expression was identi®ed in ®ve
out of ®ve human neuroblastoma cell lines (Figure 1a
and not shown) and two of three melanoma cell lines
(Figure 1a), suggesting a frequent association of Gem
expression with tumors of neuroectodermal origin.
Gem was only sporadically detected in various
epithelial cancers (Figure 1a). Gem protein expression
was con®rmed in the neuroblastomas, SH-SY5Y, SKN-SH, SMS-KCNR, and CHP-126 (Figure 1b). Gem
SK-N-SH
Muscarinic receptor induction of Gem expression in
neuroblastoma cells
SH-SY5Y
Constitutive Gem expression in melanoma and
neuroblastoma cells in culture
Figure 1 Gem expression in cultured neuroblastoma cells. (a)
Northern blot of Gem expression in total RNAs isolated from
various cell lines and ethidium bromide staining of the
corresponding blot. RNA isolated from various tissues was
separated on a 1% agarose formaldehyde gel, blotted onto a
Nytran ®lter, and hybridized with a 32P labeled Gem cDNA
probe. (b) Western blot of Gem protein in various human
neuroblastoma cell lines. Gem protein was immunoprecipitated
from cellular extracts normalized for a constant amount of
protein and subsequently visualized by Western blotting as
described in Materials and methods. (c) Time course of Gem
protein expression in DMSO (1%)-treated N1E-115 cells. Gem
protein was immunoprecipitated from cellular extracts normalized
for a constant amount of protein and subsequently visualized by
Western blotting. Morphological di€erentiation of the cells
reached a plateau on day 4
Oncogene
protein was also present in the mouse neuroblastoma
N1E-115 (Figure 1c). Treatment of N1E-115 cells with
DMSO resulted in morphological di€erentiation of the
cells that reached a plateau on day 4 (not shown) as
previously described (Kimhi et al., 1976). Following
DMSO treatment, Gem protein transiently decreased
and subsequently increased above nontreated levels,
peaking on day 4 in parallel with morphological
di€erentiation.
Figure 2 Neurotransmitter-induced expression of Gem in neuroblastoma cells. (a) Gem was immunoprecipitated from a
constant protein amount in extracts of SH-SY5Y cells and
subsequently visualized by Western blotting. Cells were treated
for 4 h with kainic acid (5 mM), NMDA (5 mM) or carbachol
(1 mM) or for 2 h with PMA (10 ng/ml). (b) Time course of Gem
protein expression following 500 mM carbachol addition to SHSY5Y cells. Gem protein levels were assayed as in (a). (c) Gem
protein levels were assayed 4 h after carbachol-treatment of SHSY5Y cells in the presence of various inhibitors. One hundred mg
of total protein in the lysates was assayed using immunoprecipitation and Western blotting as detailed in Materials and methods.
CCh (carbachol), PTX (pertussis toxin), CTX (cholera toxin)
Gem promotes neurite extension
A Leone et al
32, SK-N-SH, SMS-KCNR, and CHP-126 (not
shown).
A time course of carbachol-treatment (Figure 2b)
revealed a transient pattern of Gem expression peaking
at approximately 4 h and returning to baseline by 24 h.
In order to determine the speci®city of carbacholinduced Gem expression, SH-SY5Y cells were stimulated with carbachol for 4 h in the presence or absence
of various drugs and subsequently assayed for Gem
protein levels. Atropine, which is a competitive
antagonist of muscarinic cholinergic synapses in the
central and peripheral nervous system, potently
blocked carbachol-induced Gem expression suggesting
receptor speci®city (Figure 2c). Neither cholera toxin,
an inhibitor of Gas, nor pertussis toxin, an inhibitor of
Gai family members and Gao, demonstrated a
reproducible e€ect on Gem induction stimulated by
carbachol (Figure 2c), implying the activation of toxin
insensitive G alpha subunits such as Gaq. In contrast
to SH-SY5Y cells, carbachol treatment did not e€ect
the level of Gem protein expression in N1E-115 cells,
possibly re¯ecting di€erences in the signaling pathways
engaged downstream of muscarinic receptors in the
two cell lines.
Pattern of Gem protein expression in situ in clinical
specimens of neuroblastoma
In order to gain a broader understanding of the pattern
of Gem expression in neuroblastoma, we used
immunohistochemistry to investigate Gem expression
in clinical specimens of neuroblastoma, ganglioneuroblastoma, and ganglioneuroma. Speci®city of the antiGem antibody was assessed on Western blots by
comparing reactivity toward Gem and its related
family members, Rem (Finlin and Andres, 1997) and
Rad (Reynet and Kahn, 1993). As shown in Figure 3,
the anti-Gem antibody bound strongly to Gem, did not
bind detectably to Rad, and bound barely detectably to
Rem. By comparison, antibodies directed against GST
bound equally to the recombinant Gem, Rem, and Rad
fusion proteins (Figure 3). Quantitation of the signals
generated following binding of the Gem-directed
antibodies showed at least a 100-fold increased binding
to Gem as compared to Rem.
Diagnoses (Table 1) of the neuroblastoma cases were
established on the basis of the histopathologic criteria
recommended by the International Neuroblastoma
Pathology Committee (Shimada et al., 1999). Seven
cases were diagnosed as neuroblastoma with variable
gangliocytic di€erentiation (undi€erentiated, poorly
di€erentiated, and di€erentiating). The eighth case
showed, in addition, a prominent Schwann cell
component accompanying mature neurites and was
classi®ed as ganglioneuroblastoma, intermixed. In
another section, this latter case exhibited entirely
mature elements, i.e. ganglion cells, Schwann cells
and neurites conforming to a ganglioneuroma. Gangliocytic di€erentiation is de®ned by the presence of cells
with variable but usually abundant pink cytoplasm and
eccentrically localized nuclei with prominent nucleoli
3219
Figure 3 The speci®city of Anti-Gem antibodies. Recombinant
GST-Gem, Gst-Rad, and GST-Rem proteins were puri®ed and
equal amounts of protein were separated on an SDS gel and
subsequently assayed in a Western blot for reactivity to anti-GST
or anti-Gem antibodies which had been absorbed and anity
puri®ed as detailed in Materials and methods
Table 1 Immunohistochemical staining of neuroblastoma tissues for
Gem
1
2
3
4
5
6
7
8
Cases
Neuroblasts
Gangliocytes
Spindle cells
Neuroblastoma
Neuroblastoma
Neuroblastoma
Neuroblastoma
Neuroblastoma
Neuroblastoma
Neuroblastoma
Ganglioneuroblastoma
Ganglioneuroma
7
7
7
7
7
7
+
7
NP
+
+
NP
+
+
+
NP
+
+
+
NP
NP
+
NP
NP
NP
7
7
NP=not present; +=strong staining; +=weak staining; 7=not
detectable
(Figure 4a,b). In all specimens examined in this study,
immunostaining for Gem was observed speci®cally and
almost exclusively in cells with gangliocytic di€erentiation (Figure 4c). The latter exhibited a universally
strong, granular pattern of staining (inset Figure 4c).
The one neuroblastoma in this series that lacked
ganglion cell di€erentiation was negative for Gem in
this assay, consistent with the observation that poorly
di€erentiated neuroblastic cells were negative for Gem
on almost all histologic sections. The only exception
was case seven, in which poorly di€erentiated tumor
cells stained weakly positive for Gem. Interestingly,
ganglion cells at their ®nal stages of di€erentiation
showed weaker positivity for Gem (Figure 4d,f), in
contrast to their di€erentiative counterparts. This was
more obvious in the ganglioneuroma section, in which
only weak ganglion cell staining for Gem was observed
(Figure 4e). In two neuroblastomas, Gem-positive
spindle cells were additionally observed in the thin
septal portion of the supportive stroma that surrounds
the neuroblastic lobules (Figure 4f). The immunostaining results are summarized in Table 1. No Gem
immunoreactivity was observed in specimens stained
Oncogene
Gem promotes neurite extension
A Leone et al
3220
Figure 4 Ganglionic di€erentiation and immunohistochemical staining for Gem in neuroblastoma. The neuroblastic nodule shown
in (a) consists mostly of poorly di€erentiated neuroblastoma cells. In addition, ganglion cells with ample pink cytoplasm and
eccentric nuclei with prominent nucleoli are present (Hematoxylin and Eosin X 165). (b) and (c) represent serial sections of the same
area in a neuroblastoma, stained with Hematoxylin and Eosin (b) or an antibody against Gem (c). Cells with ganglionic
di€erentiation stain strongly for Gem. Selected ganglionic cells are shown with arrows for orientation. The pattern of staining was
granular cytoplasmic, as shown in the insert of (c) (b and c6100, insert6220). Cells with more advanced ganglionic di€erentiation
showed weaker staining for Gem as shown in (d) (6270), and terminally di€erentiated ganglion cells stained even weaker as shown
in a ganglioneuroma in (e) (680). (f) In addition to ganglion cell staining for Gem (thin arrow), there is staining of spindle cells in
the septa (thick arrows). The latter are thought to represent cells with Schwannian di€erentiation (6270)
with anti-Gem antibody mixed with an eightfold excess
of recombinant Gem protein (data not shown),
con®rming the speci®city of the technique.
Gem protein expression in developing mouse ganglia
The high levels of Gem expressed in tumor cells that
had begun to di€erentiate to a ganglionic morphology
suggested that Gem might normally be expressed in
di€erentiating or di€erentiated neurons. Therefore, we
used immunohistochemistry to investigate Gem protein
expression in the mouse in adult and fetal adrenal
glands and in developing ganglia. There was no
detectable Gem expression in either adult or fetal
adrenal medulla (not shown), consistent with the low
level of expression in undi€erentiated neuroblastoma.
Serial transverse sections through the heads of 12.5
day mouse embryos were stained with antibodies
Oncogene
directed against Gem or tyrosine hydroxylase (TH)
(Figure 5). In the adult PNS, TH expression occurs in
sympathetic ganglia, sensory ganglia, a subpopulation
of dorsal root ganglia neurons, some parasympathetic
neurons, and neuroendocrine adrenal chroman cells
(Son et al., 1996). The embryonic PNS contains
additional cells that transiently express TH, including
cells within the cranial sensory and dorsal root ganglia.
Speci®c Gem expression was detected in various
developing structures within the head, some of which
were ganglia as determined by morphology and
positive staining with antibody directed to TH. The
trigeminal (shown in Figure 5a ± d) and dorsal root
ganglia were clearly identi®ed adjacent to the pars
anterior of the pituitary gland (Figure 5a) and the
rostral portion of the central canal, respectively. Gem
was readily detectable in the trigeminal ganglia (Figure
5c) and less so in the dorsal root ganglia (not shown).
Gem promotes neurite extension
A Leone et al
3221
Figure 5 Immunohistochemical staining for Gem in the head of 12.5 day mouse embryos. (a) The trigeminal ganglion (boxed) is
shown adjacent to the ¯oor of the IV ventricle (Hematoxylin and Eosin 6120) and in a higher magni®cation in (b) (H & E6220).
Positive cells within the trigeminal nerve following immunohistochemical staining with antibodies directed against Gem (c) or TH
(d) are shown in serial sections (c,6330, and d,6340)
Therefore, Gem is expressed in a population of normal
developing PNS neurons.
Ectopic Gem expression stimulates neurite extension in
neuroblastoma cells
The potential mechanistic importance of Gem expression in di€erentiating neuroblastoma was investigated
in a mouse neuroblastoma, N1E-115 cells, that are
permissive for ecient gene transfer and are poised to
undergo neurite remodeling (Hirose et al., 1998; Katoh
et al., 1998). N1E-115 cells adhere only loosely to glass
or plastic, and therefore, we found it dicult to
reproducibly perform immunohistochemistry with such
cells. As an alternative, we co-transfected N1E-115
cells with DNA encoding Green Fluorescent Protein
(GFP) as a marker of transfection along with Gem.
Western blot analyses veri®ed increased Gem expression following transfection (data not shown). As a
control of the system, we also cotransfected DNA
encoding GFP and ROK that previously has been
shown to induce neurite retraction in N1E-115 cells
(Hirose et al., 1998; Katoh et al., 1998). Twenty-four
hours following transfection, ¯uorescent cells were
scored for morphology in the presence of serum as
either round cells, ¯at cells with short neurites, or cells
with long neurites (Figure 6a). N1E-115 cells normally
are heteromorphic; ¯uorescent cells cotransfected with
empty vector alone demonstrated approximately half
the cells in a rounded state and half the cells in a
¯attened state, with a small number of cells (4%)
displaying long neurites. Importantly, Gem expression
stimulated neurite extension. There was a signi®cant
loss of rounded cells from 48 to 13% and a
concomitant increase in cells with short and long
neurites. The speci®city of Gem in this assay is
supported by the ®nding that transfection of Rem, a
related protein, had no e€ect despite its expression as
determined by Western blotting (Figure 6a).
RasS17N is a widely used point mutant with
dominant negative activity. Based upon the conservation of a homologous serine at positions 17 and 89 in
Ras and Gem, respectively, we constructed GemS89N.
Baccuolvirus-produced recombinant GemS89N had
reduced guanine nucleotide anity (not shown), as
does RasS17N (Cool et al., 1999). However, GemS89N
maintained functional activity in the neurite extension
assay (Figure 6a). The dominant negative activity of
RasS17N appears to be particularly dependent upon an
inability to interact with e€ectors (Cool et al., 1999), a
property that may not have been lost in GemS89N.
As expected, ectopic expression of ROK resulted in
neurite retraction and cell rounding in virtually all of
the transfected cells (Figure 6a). An epistasis assay with
cotransfected Gem and ROK showed a reversal of
ROK-mediated cell rounding, suggesting that Gem acts
downstream of or in parallel with ROK (Figure 6a).
We conclude that in N1E-115 cells, increased Gem
Oncogene
Gem promotes neurite extension
A Leone et al
3222
Figure 6 Neurite induction by ectopic Gem expression in neuroblastoma cells. (a) N1E-115 cells were co-transfected with DNA
encoding green ¯uorescent protein and either vector alone, wt Gem, GemS89N, Rem, RoK, or wt Gem plus RoK. The amount of
DNA transfected was kept constant and normalized with vector DNA. Twenty-four hours later, minimums of 400 ¯uorescent cells
per sample were scored for morphology. Representative morphologies are shown. The data presented is the average of seven
separate experiments. (b) SH-SY5Y cells were cotransfected with DNA encoding GFP and wt Gem, RoK, or Gem plus RoK as
described above in (a). The data presented is the average of four separate experiments
Oncogene
Gem promotes neurite extension
A Leone et al
expression sustained over 24 h is sucient to stimulate
neurite outgrowth.
Similarly, Gem transfection of human SH-SY5Y
cells (Figure 6b) increased morphological di€erentiation as measured by cell ¯attening and neurite
extension. Up to 60% of untreated or vectortransfected SH-SY5Y cells are ¯at. Gem expression
e€ectively stimulated long neurites in these cells and
partially reversed ROK-mediated cell rounding.
Discussion
The physiological role of Gem is unknown. We
describe here a functional assay for Gem, i.e. the
stimulation of neurite outgrowth by ectopic expression
of Gem in N1E-115 and SH-SY5Y neuroblastoma
cells. These results strongly imply that one function of
Gem relates to cytoskeletal reorganization and/or
morphological di€erentiation of neurons. We were led
to consider such an assay by the observation that high
Gem expression correlates with the di€erentiation of
immature cells to a gangliocytic morphology in clinical
specimens of human neuroblastoma. In neuroblastoma,
ganglioneuroblastoma and ganglioneuroma, Gem is
easily detectable in cells with ganglionic morphology,
being absent or minimally expressed in the precursor
tumor cells with undi€erentiated morphology. Importantly, Gem is also expressed in ganglia of the
developing mouse embryo, demonstrating that Gem
expression in ganglia is a property of normal
development and not a unique property of neuroblastoma.
We describe detectable constitutive expression of
Gem in neuroblastoma cell lines and its transient
upregulation by PMA and by the muscarinic receptor
agonist, carbachol in the neuroblastoma cell line, SHSY5Y. Thus, Gem appears to be a neurotransmitterresponsive immediate early gene in neuronally related
tissues, consistent with the highly regulated nature of
Gem expression in hematopoietic and endothelial cells.
Although Gem expression is upregulated, SH-SY5Y
cells do not morphologically di€erentiate following
carbachol treatment. This is in contrast to the neurite
extension induced in N1E-115 and SH-SY5Y cells by
ectopic Gem expression. Perhaps the timing of Gem
expression is critical, and sustained Gem expression, as
would be anticipated following transfection, is important. Alternatively, carbachol treatment may lead to
additional signals that block morphological di€erentiation.
After induction of gangliocytic di€erentiation, Gem,
consistent with its nature as an activation-related early
response gene, may return to normal low levels, as
observed in terminally di€erentiated ganglion cells
derived from neuroblastoma (Figure 4c ± e). The
observed Gem-positive perilobal spindle cells coincide
topographically and morphologically with the S-100positive putative Schwann cells in neuroblastoma,
which have been associated with a better prognosis
(Shimada et al., 1985; Hachitanda and Tsuneyoshi,
1994) and have been implicated by the induction of
neuroblastic di€erentiation (Ambros and Ambros,
1995). How do Gem levels vary with the normal
di€erentiation of neurons? A detailed time course of
Gem expression and/or Gem promoter activity during
embryological development of PNS ganglia may
provide interesting insights into the potential role of
Gem in the developmental process.
Rem promotes endothelial cell (EC) sprouting, a
type of cytoskeletal reorganization observed as elongated or dendrite-like morphology of EC (Pan et al.,
2000). Thus, Gem and Rem may both function in roles
associated with cytoskeletal changes and morphological
di€erentiation. A putative dominant negative form of
Rem, but not Gem, inhibited matrigel-induced morphological di€erentiation of endothelial cells, implying
cell type speci®city (Pan et al., 2000). Consistent with
this speci®city, Rem did not induce neurite extension in
neuroblastomas (Figure 6a).
An important question is the possible mechanistic
relationship of Gem expression to cytoskeletal reorganization and morphological di€erentiation. Ectopic
Gem expression stimulated cell ¯attening and neurite
extension in N1E-115 cells (Figure 6). In this system,
activation of RhoA and/or its e€ector Rho kinase (also
known as ROCK or ROK) are necessary and sucient
to stimulate neurite retraction (Hirose et al., 1998;
Katoh et al., 1998). Inhibition of Rho kinase has been
shown to lead to neurite extension in N1E-115 cells
(Hirose et al., 1998; Katoh et al., 1998). Signi®cantly,
we have demonstrated that Gem interacts with ROK in
yeast two hybrid analyses (unpublished). Consistent
with its retention of neurite-inducing activity,
GemS89N also bound ROK. Studies are currently
ongoing to investigate the mechanistic importance of
the Gem-ROK interaction. The anticipation from the
neurite extension observed following Gem transfection
is that the activity of ROK or one of its required
substrates is blocked by Gem.
Maturation in neuroblastoma is thought to be one
mechanism associated with spontaneous regression. In
addition, tumors with maturing neuroblasts and
Schwannian stromal development (ganglioneuroblastoma, intermixed and ganglioneuroma) have a favorable
prognosis (Shimada et al., 1999). The induction of new
gene expression, especially in a cell type-speci®c way,
and consequent modulation of signaling pathways that
impact upon morphological di€erentiation is a therapeutic modality to be considered in neuroblastoma
and other types of cancer. The metastatic potential of
neuroblastoma expressing ectopic Gem currently is
being investigated in a mouse model.
3223
Materials and methods
Cell lines and materials
M Tsokos (NCI, NIH) provided the SH-SY5Y, IMR32, SKN-SH, SMS-KCNR, and CHP126 neuroblastoma cell lines.
The prostate cancer cell lines Siha-Pro and D3-14 in addition
Oncogene
Gem promotes neurite extension
A Leone et al
3224
to the melanomas, Mel-397, Mel-4 and Mel-501 were a gift
from Dr F Marincola (NCI, NIH). The cell lines MCF-10
(immortalized mammary epithelial), MCF-7 and MDA-231
(mammary cancer), Hela (cervical cancer) and OVCAR-3
(ovarian cancer) were provided by Dr P Steeg (NCI, NIH).
Murine neuroblastoma cells (N1E-115) were obtained from
the ATCC (Manassas, VA, USA). All the cell lines except the
SH-SY5Y were maintained in DMEM (high glucose) with
10% FCS in an atmosphere of 5% CO2. The cell line SHSY5Y was routinely cultured in RPMI (high glucose) in 10%
FCS in an atmosphere of 5% CO2.
Carbachol, atropine, pertussis toxin, and cholera toxin
were purchased from Calbiochem (La Jolla, CA, USA).
Kainic acid, NMDA and all-trans retinoic acid were
purchased from Sigma (St. Louis, MO, USA). GST-Rad
and GST-Rem clones were the kind gifts of Ronald Kahn
and Douglas Andres, respectively. DNA encoding Rhokinase in the pCAG vector was kindly provided by Dr
Narumiya (Kyoto University, Japan).
Drug treatments of neuroblastoma cells
Unless speci®cally mentioned, treatments to SH-SY5Y cells
were performed in RPMI with 10% FCS. The SH-SY5Y cells
were grown in 6-well plates to subcon¯uency, fresh medium
containing di€erent inhibitors was added, and 1 h later
carbachol was added at the concentration and for the length
of time described in the ®gure legends. Inhibitors were used
at the following concentrations: 0.5 mM atropine sulfate,
0.5 mg/ml Pertussis toxin, and 1 mg/ml Cholera toxin.
Induction of morphological di€erentiation in N1E-115 cells
with DMSO (1% v/v) was performed as described (Kimhi et
al., 1976). Maximal morphological di€erentiation which
consisted of enlarged refractile cells with long, branched
processes was complete by day 4.
Northern blots
Total RNA from cell lines was isolated with RNAzol (TelTest, Inc., Friedswood, TX, USA). Approximately 10 mg of
total RNA were separated on a 1% agarose formaldehyde
gel, transferred to Nytran ®lters (Schleicher and Schuell,
Keene, NH, USA) and probed with a full-length Gem cDNA
insert, 32P-labeled with the random priming method. The ®nal
wash was performed in 0.1% SSC, 0.1% SDS at 708C.
Immunoprecipitations and Western blots
Total protein was extracted in lysis bu€er (25 mM HEPES
pH 7.4, 1% Triton X-100, 150 mM NaC1, 10% glycerol
5 mM EDTA, 1 mM PMSF, 10 mg/ml Leupeptin, 10 mg/ml
Aprotinin, 25 mM b-glycerolphosphate). Immunoprecipitations were performed by incubating approximately 100 mg
of total protein lysate in 500 ml of lysis bu€er with 2.5 mg of
anti-Gem monoclonal antibody P2D10 for 3 h at 48C with
mixing. The immunocomplexes were then incubated with
20 ml of packed rProtein G agarose beads (Gibco-BRL,
Bethesda, MD, USA) for 2 h at 48C and washed three times
in 20 volumes of lysis bu€er. The immunoprecipitates were
resuspended in loading bu€er (60 mM TRIS-HCl pH 6.8, 2%
SDS, 0.5% glycerol, 0.5% DTT, 0.002% Bromophenol blue),
boiled for 2 min and separated on a 10% SDS polyacrylamide gel. Separated proteins were transferred to PVDF
membrane (Costar, Cambridge, MA, USA) probed with an
anti-Gem polyclonal antibody (270 PS) and visualized by
chemilluminescence (Pierce, Rockford, IL, USA). The blots
shown are representative of experiments that were repeated a
minimum of three times.
Immunohistochemistry
Immunohistochemistry was performed on clinical samples of
neuroblastoma and staged mouse embryos. Ten neuroblastoma tissue specimens from eight patients were randomly
selected from the ®les of the Laboratory of Pathology, NCI.
All cases were unlinked to patient clinical identi®ers. The
heads from 12.5 day old mouse embryos were ®xed in 4%
paraformaldehyde overnight, dehydrated with a graded series
of ethanol and embedded in paran. Serial, ®ve micron
transverse sections starting in the upper eye region were cut.
Brie¯y, 5 mm paran sections were deparanized, dehydrated, treated for 30 min in methanol containing 0.5%
H2O2, microwaved for 40 min in 10 mM citrate bu€er and
then incubated for 1 h in 16% normal goat serum and
overnight with anity-puri®ed anti-Gem (270PS) antibody
(1 : 200 dilution), or with anti-tyrosine hydroxylase antibody
(1 : 100) (Pel-Freeze, Rogers, Arkansas). The secondary
antibody was then applied for 1 h at room temperature,
followed by the Vectastain Elite ABC Reagent (Vector
Laboratories, Burlingame, CA, USA) for 30 min. The
peroxidase reaction was developed with diaminobenzidine
and the slides were counterstained with hematoxylin.
270PS was raised in rabbits by repeated immunization with
soluble recombinant GST-Gem protein. The antisera were
absorbed with recombinant GST and recombinant GST-Rad
(Reynet and Kahn, 1993) coupled to beads. Flow-through
from the above columns was bound to a GST-Gem anity
column, eluted with glycine bu€er (pH 3.5) and dialyzed
against PBS. Speci®city of the anti-Gem antibody was shown
in Western blots (see Figure 2) or by the inhibition of
immunohistochemical labeling in the presence of an eightfold
excess of recombinant Gem protein.
Neurite remodeling assay
N1E-115 or SH-SY5Y cells in 35 mm dishes were cotransfected with 0.2 mg pEGFP-N1 (Clontech) and 2.5 mg
eukaryotic expression vector encoding Gem or Rho kinase
using the standard Lipofectamine Plus (BRL) protocol.
Twenty-four hours following transfection, morphology of
the cells expressing Green Fluorescence Protein (GFP) was
examined on a Zeiss Axiovert microscope using the Zeiss
Achrostigmat 3260.40 PH1 objective. At least 400 cells for
each transfection were counted in a single experiment.
Acknowledgments
A Leone was partially supported by the Italian Association
of Neuroblastoma Research.
References
Ambros IM and Ambros PF. (1995). Eur. J. Cancer, 4, 429 ±
434.
Cohen L, Mohr R, Chen YY, Huang M, Kato R, Dorin D,
Tamanoi F, Goga A, Afar D, Rosenberg N and Witte O.
(1994). Proc. Natl. Acad. Sci. USA, 91, 12448 ± 12452.
Oncogene
Cool RH, Schmidt G, Lenzen CU, Prinz H, Vogt D and
Wittingho€er A. (1999). Mol. Cell. Biol., 19, 6297 ± 6305.
Dorin D, Cohen L, Del Villar K, Poullet P, Mohr R,
Whiteway M, Witte O and Tamanoi F. (1995). Oncogene,
11, 2267 ± 2271.
Gem promotes neurite extension
A Leone et al
Finlin BS and Andres DA. (1997). J. Biol. Chem., 272,
21982 ± 21988.
Finlin BS and Andres DA. (1999). Arch. Biochem. Biophys.,
368, 401 ± 412.
Finlin BS, Shao H, Kadono-Okuda K, Guo N and Andres
DA. (2000). Biochem. J., 347, 223 ± 231.
Hachitanda Y and Tsuneyoshi M. (1994). Hum. Pathol., 25,
67 ± 72.
Hirose M, Ishizaki T, Watanabe N, Uehata M, Kranenburg
O, Moolenaar WH, Matsumura F, Maekawa M, Bito H
and Narumiya S. (1998). J. Cell. Biol., 141, 1625 ± 1636.
Katoh H, Aoki J, Ichikawa A and Negishi M. (1998). J. Biol.
Chem., 273, 2489 ± 2492.
Kimhi Y, Palfrey C, Spector I, Barak Y and Littauer UZ.
(1976). Proc. Natl. Acad. Sci. (USA), 73, 462 ± 466.
Maguire J, Santoro T, Jensen P, Siebenlist U, Yewdell J and
Kelly K. (1994). Science, 265, 241 ± 244.
Moyers JS, Zhu J and Kahn CR. (1998). Biochem. J., 333,
609 ± 614.
Pan JY, Fieles WE, White AM, Egerton MM and Silberstein
DS. (2000). J. Cell. Biol., 149, 1107 ± 1116.
Reynet C and Kahn CR. (1993). Science, 262, 1441 ± 1444.
Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV and
Roald B. (1999). Cancer, 86, 349 ± 363.
Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV,
Roald B, Stram DO, Gerbing RB, Lukens JN, Matthay
KK and Castleberry RP. (1999). Cancer, 86, 364 ± 372.
Shimada H, Aoyama C, Chiba T and Newton Jr WA. (1985).
Hum. Pathol., 16, 471 ± 476.
Son JH, Min N and Joh TH. (1996). Molecular Brain
Research, 36, 300 ± 308.
Vanhove B, Hofer-Warbinek R, Kapetanopoulos A, Hofer
E, Bach FH and de Martin R. (1997). Endothelium, 5, 51 ±
61.
Wess J, Liu J, Blin N, Yun J, Lerche C and Kostenis E.
(1997). Life Sci., 60, 1007 ± 1014.
3225
Oncogene