Angiogenesis in Schwannomas
through a Rac1/Semaphorin
Hon-Kit Wong*,2, Akio Shimizu†,
Nathaniel D. Kirkpatrick*, Igor Garkavtsev*,
Annie W. Chan‡, Emmanuelle di Tomaso*,3,
Michael Klagsbrun†and Rakesh K. Jain*
*The Steele Lab of Tumor Biology, Department of Radiation
Oncology, Massachusetts General Hospital and Harvard
Medical School, Boston, MA, USA;†Departments of
Surgery and Pathology, Children’s Hospital Boston and
Harvard Medical School, Boston, MA, USA;‡Department
of Radiation Oncology, Massachusetts General Hospital
and Harvard Medical School, Boston, MA, USA
Neurofibromatosis type 2 (NF2) is an autosomal-dominant multiple neoplasia syndrome that results from mutations
in the NF2 tumor suppressor gene. Patients with NF2 develop hallmark schwannomas that require surgery or radia-
in particular, tumor blood vessels—of schwannomas may be an important therapeutic target. Furthermore, although
much has been done to understand how merlin, the NF2 gene product, functions as a tumor suppressor gene in
regulates angiogenesis to support schwannoma growth is largely unexplored. Here we report that the expression of
semaphorin 3F (SEMA3F) was specifically downregulated in schwannoma cells lacking merlin/NF2. When we reintro-
duced SEMA3F in schwannoma cells, we observed normalized tumor blood vessels, reduced tumor burden, and ex-
tended survival in nude mice bearing merlin-deficient brain tumors. Next, using chemical inhibitors and gene
knockdownwithRNAinterference, wefound thatmerlin regulatedexpressionof SEMA3F through RhoGTPase family
member Rac1. This study shows that, in addition to the tumor-suppressing activity of merlin, it also functions to main-
tain physiological angiogenesis in the nervous system by regulating antiangiogenic factors such as SEMA3F. Restor-
ing the relative balance of proangiogenic and antiangiogenic factors, such as increases in SEMA3F, in schwannoma
microenvironment may represent a novel strategy to alleviate the clinical symptoms of NF2-related schwannomas.
Neoplasia (2012) 14, 84–94
Abbreviations: HMVECs, human microvascular endothelial cells; HUVECs, human umbilical vascular endothelial cells; NF2, neurofibromatosis type 2; NRP, neuropilin;
ROCK, Rho-associated coiled-coil forming protein serine/threonine kinase; SEMA, semaphorin; TSP2, thrombospondin 2; VEGF, vascular endothelial growth factor
Address all correspondence to: Rakesh K. Jain, PhD, Harvard Medical School and Edwin L. Steele Laboratory for Tumor Biology, Department of Radiation Oncology,
Massachusetts General Hospital, 100 Blossom St, Cox 7, Boston, MA 02114. E-mail: email@example.com
1This work was supported in parts by grants from the National Institutes of Health (P01-CA80124 to R.K.J. and CA37392 and CA45548 to M.K.), Federal Share/National
Cancer Institute Proton Beam Program Income Grant (to R.K.J.), the Claflin Award (to E.d.T.), and the Flight Attendant Medical Research Institute (A.W.C.). The authors
have no conflicts of interest to declare.
2Present address: Center for Neurologic Diseases, Brigham and Women’s Hospital and Harvard Medical School, Harvard Institutes of Medicine, Boston, MA 02115.
3Present address: Novartis Institute of Biomedical Research, Cambridge, MA 02139.
Received 16 November 2011; Revised 3 February 2012; Accepted 3 February 2012
Copyright © 2012 Neoplasia Press, Inc. All rights reserved 1522-8002/12/$25.00
Volume 14 Number 2February 2012 pp. 84–94
Neurofibromatosis type 2 (NF2) is an autosomal-dominant multiple neo-
plasia syndrome that results from mutations in the NF2 tumor suppressor
to nerves of the peripheral nervous system), and eye and skin lesions to
a lesser extent . Although most patients with sporadic schwannomas
benefit from surgery, patients with NF2 have limited options owing to
the multiplicity and bilateral location of their tumors . The hallmark
feature of NF2, vestibular schwannomas, is particularly challenging sur-
gically because of the potential for brainstem or cranial nerve damage—
hearing loss and tinnitus are very common in these patients.
Our group recently provided encouraging clinical evidence that
bevacizumab, an anti–vascular endothelial growth factor (VEGF) anti-
body treatment, improved hearing and stabilization of tumor growth in
patients with bilateral vestibular schwannoma  and subsequently de-
lineated the mechanisms by which anti-VEGF treatment might have
produced such clinical outcomes . Keeping in line with the pub-
lished data that VEGF is expressed in schwannomas [5–8], our preclin-
ical and clinical results clearly show that VEGF and tumor angiogenesis
clinical symptoms. More importantly, during the course of analysis, we
discovered that the expression of class 3 semaphorins (SEMA3), partic-
ularly SEMA3A and SEMA3F, was markedly decreased in NF2-related
schwannomas, whereas that of VEGF remains unchanged. This observa-
tion led to a hypothesis that the relative balance of VEGF and SEMA3
expression regulates physiological angiogenesis in normal nerve tissues.
the ratio of VEGF levels to SEMA3s increases and contributes to abnor-
mal angiogenesis, tumor growth, and progression . However, a causal
link between merlin/NF2 and SEMA3s as well as the role of SEMAs
in pathologic schwannoma angiogenesis remains not established.
SEMA3s are secreted proteins that were first shown to regulate axon
guidance in the developing nervous system [10–13] and subsequently
found to also regulate both physiological and pathologic angiogenesis
[14–17]. In general, SEMA3s first bind to neuropilin (NRP) and plex-
instotransduce intracellularsignalsandexerttheirantiangiogenic func-
tions . Specifically, SEMA3F, in an NRP2-dependent manner, can
inhibit cell adhesion and cell migration in vitro and tumor angiogenesis
and metastasis in vivo [19,20]. Furthermore, ithasbeen recently shown
that ABL2 and RhoA play key roles in mediating SEMA3F-induced
depolymerization of F-actin and the subsequent cytoskeleton collapse
in tumor cells and endothelial cells .
In this study, we report that merlin/NF2 upregulates SEMA3F
expression through Rho GTPase Rac1. Furthermore, we show that
SEMA3F is indeed an essential molecule to regulate angiogenesis and
normalizevessels in orthotopic preclinicaltumor modelsof NF2through
to tumor angiogenesis through modulation of SEMA3F expression.
Interestingly, merlin/NF2 also induced the expression of other anti-
in NF2-mediated angiogenesis needs to be investigated.
Materials and Methods
Chemical Reagents and Antibodies
Recombinant C3 transferase (CT03) was purchased from Cyto-
skeleton (Denver, CO). Y-27632 (688001) and Rac1 (553502)
inhibitor were purchased from Calbiochem (Darmstadt, Germany),
and Evans Blue dye was purchased from Fisher Scientific (Waltham,
MA). Lentiviral particles expressing merlin short hairpin RNA (shRNA;
L271) and scramble (Lmag) were kind gifts from Dr Marianne F. James
(Centre for Human Genetic Research, Massachusetts General Hospital,
Richard B. Simches Research Building). The following antibodies were
used in notated dilutions: β-tubulin (1:10,000, MAB3408; Chemicon,
Billerica, MA), merlin (1:1000, sc-331/A-19; Santa Cruz Biotechnology,
Santa Cruz, CA), Rac1 (1:1000, ARC03; Cytoskeleton), semaphorin
3F (SEMA3F, 1:3000; GenScript, Piscataway, NJ), TSP2 (1:2000,
611150; BD Transduction Laboratories, Sparks, MD), and V5 (1:5000,
46-0705; Invitrogen, Grand Island, NY).
The pMOWSdSV-dsred retroviral construct was a kind gift from
Prof Brian Seed (Department of Genetics, Massachusetts General
Hospital and Harvard Medical School) . Murine SEMA3F or
TSP2 was cloned into pMOWSdSV using PacI and SalI sites. Murine
SEMA3F (in pCMV-SPORT6, MMM1013-65024) and TSP2 (in
IL) and Addgene (Cambridge, MA), respectively. SEMA3F was amplified
from pCMV-SPORT6 using forward 5′-GCGCTTAATTAAACC
ATGGTTGTCACTGCCTTCATC and reverse 5′-GCGCGTCGACT-
TACCTGTGTCCGGAGGGTGGTGGCG primers. TSP2 was
amplified from pcDNA3 using forward 5′-GCGCTTAATTAAAC-
CATGGTCTGGGCACTGGCCCTG and reverse 5′-GCGCGTC
TAGGCTTACCGGCATCTCT GCACTCATACTT. All the cloning-
related polymerase chain reaction (PCR) amplifications were done with
PfuUltra DNA polymerase (Agilent Technologies, Santa Clara, CA)
using a standard protocol. Sequencing was performed by MGH DNA
Core Facility (Boston, MA).
Generation of Rac1 shRNA Retroviruses
ShRNA against mouse Rac1 (5′-GCATTTCCTGGAGAGTA-
CA-3′ ) was cloned into pSilencer 5.1-U6 construct (AM5782)
according to the standard instructions (Ambion, Grand Island, NY),
with pSilencer 5.1-U6 Retro scrambled (5783G) as a control. pSilencer
plasmids into HEK293T cells to produce retroviruses (see next section
for the transfection procedures).
Cell Lines and Generation of Stable Cell Lines
The original clone of human HEI193 [24,25], which contained a
mutated copy of merlin/NF2, was a gift from Dr Xandra Breakefield.
Subsequent analysis of our HEI193 revealed that our clone did not
contain a wild-type copy of merlin/NF2 . The generation of murine
Nf2−/− tumor Schwann cells has been described previously , and
these cells were kindly provided by Drs Andrea I. McClatchey and
Annie Chan. We verified that the Nf2−/− cells were merlin deficient
through Western blot analysis. Both cell lines were kept in Schwann
cell medium supplemented with Schwann cell supplement, 5% fetal
bovine serum, and penicillin/streptomycin (ScienCell 1701, Carlsbad,
CA). Human umbilical vein endothelial cells (HUVECs) were main-
tainedinendothelialgrowth medium(EGM) supplementedwithbovine
brain extract (CC-3024A; Lonza, Basel, Switzerland), human micro-
vascular endothelial cells (HMVECs) were maintained in 1× EBM-2
(CC-3156; Lonza) supplemented with EGM-2 MV (CC-4147; Lonza),
Neoplasia Vol. 14, No. 2, 2012Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.
and U87MG was maintained in 1× Dulbecco modified Eagle medium
(10-014-CV; CellGrow, Manassas, VA) supplemented with 10% fetal
bovine serum. To generate Nf2−/− cell lines overexpressing SEMA3F
or TSP2, pMOWSdSV-SEMA3F or pMOWSdSV-TSP2 was trans-
fected into HEK293T cells with GAG/POL and VSV in a 3.3:2.3:1
ratio using Lipofectamine 2000 (Invitrogen). Culture medium contain-
ing retroviruses was collected 1 and 2 days after transfection. Nf2−/−
cells were infected with SEMA3F or TSP2 retroviruses multiple times
to ensure virtually 100% expression levels.
Tube Formation Assay
The effect of conditioned medium from various genetically mod-
ified Nf2−/− cells on the endothelial cell structures and morphologies
was investigated with endothelial cell tubes formed with HUVECs.
Growth factor–reduced Matrigel (354230; BD Biosciences, Sparks, MD)
was first plated in a 24-well plate and allowed to solidify for 30 minutes
at 37°C with 5% CO2. One hundred thousand cells were then seeded
on top, and tubes were allowed to form for 2 hours. Conditioned me-
dium was added, and the effect on tube structure and morphology was
examined with bright-field microscopy for 36 hours.
To recapitulate as closely as possible the microenvironment of
vestibular schwannomas (intracranial but arising from a cranial nerve),
we modified the previously published model for gliomas. In brief,
Nf2−/− cells were implanted between the pia and arachnoid meninges
in nude mice (8 weeks old) bearing transparent cranial windows . In
the second model, we reproduced the nerve microenvironment using
the mouse sciatic nerve . This environment had previously been
shown to be suitable for schwannoma growth. In both models, approx-
imately a million cells were injected with a 28.5-gauge needle to gen-
Mouse Angiogenesis PCR Array
RNA was extracted from tumor cells with or without merlin/NF2
using RNeasy Mini Kit (Qiagen, Valencia, CA), DNase-treated
(Promega, Madison, WI), and retrotranscribed by RT2First Strand
Kit (SA Bioscience, Valencia, CA). Mouse Angiogenesis RT2Profiler
PCR Array and RT2Real-time SyBR Green/ROX PCR Mix were
purchased from SA Biosciences. PCR was performed on either
Stratagene Mx3000P or Mx3005P (Agilent Technologies). We used
the ΔCtmethod for data analysis. For each gene fold, changes were cal-
culated as difference in gene expression between cells expressing merlin/
NF2 and that without. A positive value indicates gene up-regulation and
a negative value indicates gene down-regulation.
DNA Microarray Analysis
Total RNA was isolated from merlin/NF2 cells (RNeasy Mini Kit;
Qiagen) according to recommended Affymetrix (Santa Clara, CA)
protocols. Total RNA was checked for RIN score using an Agilent
2100 Bioanalyzer with an RNA Nano 2000 chip using methods
according to the Agilent Bioanalyzer manual. Samples were processed
using 20 μg of starting total RNA and were amplified and labeled
using the GeneChip Expression 3′-Amplification Reagents One-
Cycle complementary DNA (cDNA) synthesis kit (P/N 900431 from
Affymetrix) followed by the GeneChip Expression 3′-Amplification
sample was then fragmented using 5× fragmentation buffer contained
cocktail containing Control Oligo B2, biotinylated hybridization
controls, acetylated bovine serum albumin, herring sperm DNA, 2×
hybridization buffer in an Affymetrix Model 640 Hybridization oven
for ∼16 hours at 45°C with rotation at 60 RPM. The chips were then
washed using buffers A and B in a Model 450Gene Chip Fluidics
station and then scanned using an Affymetrix Model 3000 GeneChip
Scanner with Autoloader. Scanning and analysis were carried out using
Affymetrix GCOS software. The raw data were then normalized using
MAS5.0 and scaled to a factor of 500. Comparative analysis was then
performed by using a two-sided t test with a minimum SD of 1.0. We
used false discovery rate (FDR) adjustment to account for multiple
Total RNA was extracted from Nf2−/− or HEI193 cells using the
RNeasy Mini Kit (Qiagen) according to the instruction manual.
Genomic DNA was digested using the RNase-free DNase I (Qiagen)
to ensure the complete removal of DNA contamination. The amount
of total RNA was quantified by absorbance at 260 nm (NanoDrop,
Wilmington, DE). Five hundred nanograms of RNA from each sam-
ple was transcribed to cDNA using SuperScript III RNase H−Reverse
Transcriptase (Invitrogen). PCR was performed in a thermal cycler
(DYAD). PCR products were then resolved in a 2% agarose gel by
electrophoresis,andDNAimages werecapturedwith theUVtransillumi-
nator (Bio-Rad, Hercules, CA). Glyceraldehyde-3-phosphate dehydro-
genase (Gapdh) gene was amplified in each PCR experiment as an
internal loading control.
Determination of Rac1 Activity
Rac1 activity was measured with Rac1 (BK035) activation assay
biochem kit (Cytoskeleton) according to the manufacturer’s protocol.
Cells were first washed with phosphate-buffered saline (PBS) and then
lysed with lysis buffer supplemented with protease inhibitors. About
100 μg of lysates was precleared, and 10 μl of PAK-PBD beads was
added for pull-down of activated Rac1. Positive (GTPrS) and negative
(GDP) controls were included toshow thefunctionality of theactivity
assay. After rocking at 4°C for 1 hour, we washed the beads once and
boiled them at 100°Cfor 2minutes. Protein lysateswere subsequently
resolved in a 8% to 16% gradient gel (Invitrogen) and transferred
them to a polyvinylidene fluoride membrane, and Rac1 expression
(corresponds to its activity) was detected using standard Western blot
analysis. Ten nanograms of His-tagged Rac1 recombinant protein was
loaded as a control.
Western Blot Analysis
Cells were rinsed with cold PBS and lysed in an appropriate volume
of lysis buffer. Protein concentration was measured with bicinchoninic
acid (BCA) protein assay kit (Pierce, Rockford, IL) according to the
manufacturer’s instructions. Appropriate volumes of protein lysates
were then mixed with 4× SDS gel-loading buffer, boiled at 100°C
for 5 minutes, and resolved in 8% to 16% gradient polyacrylamide gels
(Invitrogen). Proteins were subsequently transferred onto polyvinylidene
fluoride membrane (0.45 μm; Thermo Scientific, Rockford, IL), and
with Tween 20 for 1 hour. Blots were then incubated with appropri-
ately diluted primary and horseradish peroxidase–conjugated secondary
Merlin Regulates Angiogenesis through Rac1/SEMA3FWong et al.Neoplasia Vol. 14, No. 2, 2012
antibodies, each for 1 hour at room temperature. Finally, blots were
developed with enhanced chemiluminescence solutions and exposed
onto HyBlot CL (Denville Scientific, Metuchen, NJ).
Mice were perfusion fixed with 4% formaldehyde in PBS. Perfused
tumors were extracted, postfixed in formaldehyde at 4°C for 4 to
5 hours, and incubated with 30% sucrose overnight at 4°C with con-
stant rocking. Sciatic nerve tumors were then embedded in optimal
cutting temperature freezing medium and frozen for sectioning.
Sections (10 μm thick) were first air-dried for 30 minutes at room
temperature and postfixed with acetone for 5 minutes at −20°C.
Double staining was performed using CD31 (1:150, MAB 1398Z;
Chemicon) and desmin (1:300, A0611; Dako, Carpinteria, CA).
Detection was carried out using secondary antibodies (or streptavidin)
labeled with fluorophores of different wavelength (fluorescein isothio-
cyanate, Cy3, and Cy5). Confocal images were acquired using a
FluoView 500 confocal microscope (Olympus, Center Valley, PA).
A constant surface area (0.4 mm2) was imaged in all cases.
Blood Vessel Permeability Measurement
Changes in vessel permeability in sciatic nerve tumors was exam-
ined with Evans Blue dye extravasation assay. Mice were first anes-
thetized with ketamine (10 mg/kg)/xylazine (1 mg/kg) mix and then
injected with 1 mg/kg 3% Evans Blue dissolved in 0.9% saline. After
20 minutes, mice were perfusion fixed with 4% formaldehyde in
PBS, and sciatic nerve tumors were extracted. Tumors were weighed
(wet weight), cut, and immersed in 1 ml of formamide at 56°C over-
night to extract the Evans Blue dye that was extravasated out of the
vessels. The following day, tissues were removed from the formamide,
and the absorbance (OD620nm) of the samples was measured with a
spectrophotometer. Results were expressed as microgram of Evans
Blue per milligram of tumor tissue based on a standard curve of Evans
Blue and the wet weight of the tumor.
Quantification of Vessel Area and Perivascular Cell Coverage
Quantification of the immunohistochemistry (IHC) sections for
vessel area and pericyte coverage was completed with a custom image
process program written in MATLAB (MathWorks). On the basis of
CD31 fluorescence intensity, images were thresholded to achieve vessel
segmentation. To determine the vessel area (a surrogate for vessel den-
the segmented images, the average vessel diameter was measured using
automated protocols. Finally, the segmented CD31 images were dilated
by 4 μm and colocalized with desmin fluorescence images to determine
the amount of desmin+ cells in vessel regions (pericytes coverage).
We used the unpaired, two-tailed Student’s t test for comparison
between two samples. One-way analysis of variance Fisher test followed
by Tukey honestly significant difference test was used for multiple
comparisons with a 95% confidence level. For survival rate, we plotted
the survival distribution curve with the Kaplan-Meier method followed
by log-rank testing (XLSTAT software, New York, NY). We considered
the difference between comparisons to be significant when P < .05 for
all the statistical analysis. All values were presented as mean ± SEM.
Class 3 Semaphorin Expression Levels Are Specifically
Downregulated in the Absence of Wild-type Copies
We have previously shown that expression levels of SEMA3A and
SEMA3F are significantly downregulated in both NF2-related and
sporadic schwannomas . To examine more systematically whether
secreted class 3 SEMAs are specifically affected in tumor cells that do
not express wild-type merlin/NF2, we compared the expression levels
of class 3 SEMAs between primary human and mouse Schwann cells
with HEI193 (human merlin-deficient Schwann cells) and Nf2−/−
(mouse merlin-deficient Schwann cells), respectively. This comparison
consistentlyrevealedthat theexpression ofmostofSEMA3s was largely
downregulated in the absence of merlin (Figure 1A) without affecting
that of NRP or plexins (data not shown). Knocking down the expres-
sion of merlin in primary mouse Schwann cells recapitulated similar
observations (Figure 1B). When we reintroduced merlin into mouse
Nf2−/− schwannoma cells [the Nf2−/−(pWZL-NF2)], we specifically
restored expression of SEMA3B and SEMA3F (Figure 1C). Of note,
SEMA3F, in particular, is a potent antiangiogenic factor [19,21].
To determine whether other angiogenic factors in addition to
SEMA3F were regulated by merlin, we compared the messenger
RNA (mRNA) expression profile of 87 angiogenic factors between
Nf2−/−(pWZL) and Nf2−/−(pWZL-NF2) cells with mouse angio-
genesis array. We found that the expression of TSP2, a well-described
antiangiogenic factor, was also prominently upregulated with merlin
reintroduction. There was a 7.04 ± 1.85-fold increase (P < .02) of thbs2
mRNA that encodes TSP2 protein. This increase is specific for TSP2
because there was only a slight but insignificant increase for that of
thbs1 (2.3 ± 0.78, P = .6, n = 3). Interestingly, we did not detect any
significant change of VEGF or PDGFa expression, two very potent
proangiogenic factors (0.9 ± 0.56- and 1.4 ± 0.36-folds, respectively).
These data suggest that reintroduction of merlin/NF2 specifically tips
the angiogenesis balance to antiangiogenic by increasing antiangiogenic
factors rather than decreasing proangiogenic factors. We then con-
firmed up-regulation of TSP2 and SEMA3F at the mRNA level by
DNA microarray analysis(TSP2: 7.9 ± 0.7-fold increase and SEMA3F:
15.3-fold increase). Finally, we determined that the up-regulation of
SEMA3F and TSP2 was seen in the protein level (Figure 1D). These
data indicate that merlin regulates the expression of two secreted anti-
angiogenic factors (SEMA3F and TSP2) that could potentially modify
the tumor microenvironment of schwannomas.
Merlin/NF2 Regulates Expression of SEMA3F and
TSP2 through a Rac1-Dependent Pathway
To investigate the mechanism(s) whereby merlin/NF2 regulates
expression of SEMA3F and TSP2, we used protein inhibitors and
retroviral RNA interference in our schwannoma cells. One likely
candidate modulating this pathway could be Rho GTPase because 1)
Rac1  and cdc42  activity and intracellular localization were
altered in human schwannoma cells deficient in merlin expression, 2)
Rac1 was found to be able to regulate the expression of TSP2 in endo-
thelial cells , and (3) RhoA and Rho-associated coiled-coil forming
protein serine/threonine kinase (ROCK) were found to mediate the
collapsing activity of SEMA3F . Thus, in our system, we examined
Rac1 activity. Compared with the cells that were reintroduced with
wild-type merlin, we found that Rac1 activity is indeed increased in
the merlin-deficient counterpart (Figure 2A). Next, we treated Nf2−/
Neoplasia Vol. 14, No. 2, 2012 Merlin Regulates Angiogenesis through Rac1/SEMA3FWong et al.
−(pWZL-NF2) cells with exoenzyme C3 transferase (an ADP ribosyl
transferase that selectively ribosylates RhoA, RhoB, and RhoC proteins
on asparagine residue 41, rendering them inactive), Y-27632 (a specific
inhibitor of the ROCK family of protein kinases), or a Rac1 inhibitor
to examine which member in the Rho family GTPase is responsible for
the induction of SEMA3F and TSP2 on the reintroduction of merlin/
NF2. We found that Rac1 inhibition itself was enough to almost com-
pletely abrogate the induction of SEMA3F and TSP2 (Figure 2B
and quantified in Figure 2C). Blocking RhoA/B/C or ROCK did
not produce any significant and detectable suppression of either
SEMA3F or TSP2.
To further confirm that Rac1 is the specific member in the Rho
GTPase family regulating SEMA3F and TSP2 expression, we knocked
down the expression of Rac1 genetically in the Nf2−/−(pWZL-NF2)
cells by Rac1 shRNA retroviruses achieving a 60% reduction of Rac1
expression (Figure 2D). This knockdown resulted in a marked suppres-
sion of SEMA3F and TSP2 production, similar to the data that we
obtained with Rac1 inhibitors. To investigate the consequence of this
with conditioned medium collected from Nf2−/−(NF2 + Rac1 −
scramble) or Nf2−/−(NF2 + Rac1 − shRNA). In endothelial tubes
treated with the Rac1 shRNA conditioned medium, we observed a
significant maintenance of tube structure compared with endothelial
tubes treated with scramble conditioned medium where we found tube
collapse (Figure 2D, 5.3 ± 0.47 more tubes in HUVECs treated with
the Rac1 shRNA conditioned medium compared with the scrambled
control). These data suggest that an increase in Rac1 activity in the
absence of merlin leads to a subsequent decrease in SEMA3F and
TSP2, thereby reducing the antiangiogenic effects of schwannoma cells.
SEMA3F Normalizes Nf2−/− Tumor Vessels Resulting in
Reduced Tumor Size and Improvement of Overall Survival
To determine whether this merlin-mediated regulation of SEMA3F
and TSP2 observed in vitro has a functional role in schwannoma angio-
genesis and tumor progression in vivo, we generated Nf2−/− cell lines
that constitutively express high levels of either SEMA3F or TSP2 com-
pared with control cells (Figure 3A). We also confirmed that these cell
lines continued to express SEMA3F or TSP2 in an orthotopic model of
brain tumor (Figure 3B). Finally, to further confirm that the secreted
SEMA3F is biologically active in these cells, we used a cytoskeleton
Figure 1. Merlin regulates expression of class 3 semaphorins
and TSP2. (A) Merlin-deficient tumor cells have significantly lower
SEMA3 expression compared to their primary counterparts. Semi-
quantitative PCR was done to examine the expression profile of
SEMA3s in primary human Schwann cells (PHSC), primary mouse
Schwann cells (PMSC), mouse schwannoma cells (Nf2−/−), and
human schwannoma cells (HEI193). Expression levels of SEMA3s
in primary cells were normalized with an input control (gapdh), and
that of tumor cells was compared accordingly and expressed as
relative expression levels. n = 3 and *P < .01. (B) Merlin expression
positively correlates with SEMAs in primary Schwann cells. Merlin
expression was downregulated with lentiviral particles expressing
merlin shRNA. Total RNA was then extracted and reverse tran-
scribed, and the expression of merlin and SEMA3s was examined
with semiquantitative PCR. Scramble-infected primary Schwann
cells were used as a control for comparison (set to 1). n = 3 and
*P < . 01 compared with the expression in scramble control. (C)
Reintroduction of wild-type merlin restores expression of SEMA3B
and SEMA3F. Wild-type merlin (pWZL-NF2) was reintroduced into
Nf2−/− cells with retroviruses. Infected cells were selected with
hygromycin B at 200 μg/ml for several weeks, and single clones
were selected. Expression of SEMAs in pWZL-NF2 was compared
with a control schwannoma cell line that was infected with an
empty vector pWZL. n = 3 and *P < .01. (D) Reintroduction of
merlin upregulates SEMA3F and TSP2 protein expression in
schwannoma cells. Protein lysates and conditioned medium from
Nf2−/−(pWZL) and Nf2−/−(pWZL-NF2) were resolved in a gradi-
ent gel and expression of SEMA3F, and TSP2 was examined by
Western blot analysis. Merlin blot confirmed the successful reintro-
duction, and β-tubulin was used as an input control. n = 3.
Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al. Neoplasia Vol. 14, No. 2, 2012
collapse assay that directly shows the activity of SEMA3F . In cells
treated with conditioned medium produced from Nf2−/−(SEMA3F),
almost 90% of U87MG cells collapsed within 5 hours, whereas cells
treated with control Nf2−/−(dsred) conditioned medium remained
unchanged (Figure 3C). We also found that conditioned medium
obtained from Nf2−/−(TSP2) decreased proliferation of HMVECs by
more than 50% in 4 days after treatment compared with that produced
from Nf2−/−(dsred) (Figure 3D).
Given a functional role of SEMA3F and TSP2 in our cell lines, we
implanted Nf2−/−(SEMA3F), Nf2−/−(TSP2), or Nf2−/−(dsred) into
Figure 2. Merlin regulates expression of SEMA3F and TSP2 through a Rac1-dependent pathway. (A) Rac1 activity and expression
increases in schwannoma cells that are deficient in merlin. PAK-PBD-Rac1 indicates active Rac1, and Rac1 represents the total form.
PAK indicates p21 activated kinase 1; PBD, p21 binding domain. (B) Inhibiting Rac1 activity diminishes the expression of SEMA3F and
TSP2 induced by merlin. Rho GTPases (cdc42, Rac1, and ROCK) were inhibited with specific inhibitors at increasing concentrations. One
day after incubation, both medium and lysates were collected, and the expression levels of SEMA3F and TSP2 as well as merlin were
analyzed with Western blots. (C) Quantification of SEMA3F and TSP2 expression levels in Rac1-inhibited conditions. Expression levels of
SEMA3F and TSP2 were measured and normalized with β-tubulin. The normalized values in the control line (pWZL) were then set to 1,
and the expression levels of SEMA3F and TSP2 in other conditions were compared with the control accordingly. HC indicates high
concentration; LC, low concentration. n = 3 independent Western blots. *P = .012. **P = .043. ***P = .05. (D) Suppressing Rac1
expression by RNA interference specifically reduces SEMA3F and TSP2 expression and results in a more intact network of endothelial
tubes. Rac1 shRNA (or scramble [scr]) was transduced into Nf2−/−(pWZL-NF2) cells by retroviruses, and the infected cells were
selected by puromycin at 1 μg/ml for 3 days. Conditioned medium from scr (i) or Rac1 shRNA (ii) was then added to the endothelial
network formed by HUVECs. Phase-contrast images were captured 36 hours after the addition of conditioned medium to examine the
biologic effect of Rac1 knockdown. The tube formation per field was calculated to be 5.3 ± 0.47 times higher in Rac1 knockdown
conditioned medium wells compared with control conditioned medium.
Neoplasia Vol. 14, No. 2, 2012Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.
the sciatic nerve of nude mice and examined the effect of these mole-
cules on the progression of schwannoma. We found that nerve tumors
with ectopic SEMA3F expression were much smaller compared with
dsred control tumors (Figure 4A). Nevertheless, we were unable to ob-
serve any prominent antitumor effect of TSP2 in the development of
these tumors (data not shown). When we analyzed the tumor vessels in
SEMA3F reduced total number of CD31+vessels per area, a surrogate
for vessel density, by 25% without affecting the diameter of the vessels
(Figure 4B), whereas TSP2 did not affect vessel density or diameter.
Moreover, SEMA3F was able to restore pericyte coverage on tumor
vessels—we measured a 2.5-fold increase in pericyte coverage based
on desminstaining(Figure 4C). Accompanying theincrease ofpericyte
coverage, there was a 30% reduction of permeability in sciatic nerve
tumors with SEMA3F overexpression (Figure 4D). When we implanted
tumors using another orthotopic model of NF2 in the brain, Nf2−/
−(SEMA3F) brain tumors also grew more slowly and were smaller in
size, and the mice bearing these brain tumors had a significantly better
overall survival rate (Figure 4E; dsred, 17.9 ± 1.5 days; SEMA3F, 26.5 ±
1.3 days; P < .01). However, consistent with the lack of effects in the
nerve tumor, Nf2−/−(TSP2) tumor-bearing mice had no improvement
in survival over control tumor-bearing mice. Taken together, although
reintroduction of merlin upregulates both SEMA3F and TSP2- in
schwannoma microenvironment, increased expression of TSP2 merely
does not result in any observable angiogenic changes. It is more likely
that merlin-SEMA3F signaling keeps the angiogenic balance in check.
The role of TSP2 in Schwann cells or the schwannoma microenviron-
ment remains to be further investigated.
Schwannomas typically progress slowly and are not considered to be
angiogenic tumors. However, our recent review of surgical archival
specimen revealed a proangiogenic profile defined by number and
size of vessels combined with quantification of VEGF expression
and SEMA3F loss . Furthermore, in a recent preclinical study,
we showed that loss of the Nf2 gene was accompanied by the loss
of SEMA3F and that treatment with anti-VEGF therapies resulted in
sustained decreases in growth rate and vessel permeability . These
Figure 3. CharacterizationofexpressionandfunctionofSEMA3Fand
TSP2 in genetically modified schwannoma cells. (A) Schwannoma
cells that are reintroduced with SEMA3F or TSP2 secrete these pro-
teins in a time-dependent manner. Nf2−/− cells were infected with
retroviral particles expressing either pMOWSdSV-SEMA3F or -TSP2.
Almost 100% of the cells were positive for the reintroduction after
two overnight infections, as evident by the red fluorescence from
the dsred control. Conditioned medium (free of serum) was col-
lected from each cell line, and 20 μl of medium was resolved in a
gradient gel for the detection of SEMA3F and TSP2. V5 blots were
used to confirm the expression of exogenously introduced protein.
(B) Schwannoma cells continueto express high levels ofSEMA3F or
TSP2 in the brain environment. Schwannoma cells expressing either
SEMA3F or TSP2 were implanted into the brain parenchyma, and
tumors wereallowed to form. Tumors wereextracted whenreached
10 × 10 mm in size, and protein lysates were used for the detection
of SEMA3F or TSP2. B indicates brain; T, tumor. (C) Conditioned
medium collected from SEMA3F expressing schwannoma cells is
able to collapse U87MG cells. Nf2−/−(SEMA3F) conditioned me-
dium was added to U87MG cells at a 1:10 dilution (ii and ii′) and
Nf2−/−(dsred) conditioned medium was used as a control (i and i′).
Phase contrast images were captured 5 hours after the addition of
SEMA3F conditioned medium and the number of collapsed cells
were counted and presented graphically in lower panel. To define
a cell that was collapsed, it had to possess a shrunken cell body
and multiple extended processes (ii′). Controls cells were normally
flat without many processes (i′). CM indicates conditioned medium.
n = 3. *P < .01. (D) Conditioned medium collected from TSP2
expressing schwannoma cells inhibits proliferation of HMVECs.
Schwannomacells were lysedwiththe same volume ofprotein lysis
buffer, and 5 μl of precleared lysate was quantified by BCA assay.
The total amount of cells in each condition was represented as the
absorbance at 562 nm. SF indicates serum-free medium. n = 3.
*P < .01.#Not significant.
Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.Neoplasia Vol. 14, No. 2, 2012
Figure 4. SEMA3F normalizes schwannoma vessels and reduces tumor burden. (A) SEMA3F reexpression reduces sciatic nerve tumor
burden. Animals were perfusion fixed, and tumors were extracted approximately 2 weeks after implantation. Tumor volume was mea-
sured and compared accordingly. Scale bar, 1 cm. O/E indicates overexpression. n = 10. *P = .05. (B) SEMA3F reexpression reduces
CD31+area without affecting the vessel diameter. Frozen sections of sciatic nerve tumor generated from each schwannoma line as
indicated in the diagram were immunostained with anti-CD31 to reveal tumor vessels. Overexpression of SEMA3F reduced vascularity by
25% compared with the dsred control, without influencing the blood vessel diameter. n = 7. *P = .043. (C) SEMA3F reexpression mark-
edly increases pericyte coverage. Frozen sections of sciatic nerve tumor generated from Nf2−/−(dsred) (i and i″) and Nf2−/−(SEMA3F)
(ii and ii″) were costained with anti-CD31 and anti-desmin. The overlapping of desmin and CD31 area was quantified to indicate pericyte
coverage. Scale bar, 100 μm. n = 7. *P = .028. (D) SEMA3F reexpression reduces tumor vessel permeability. Evans Blue dye extra-
vasation assay was used to examine vessel permeability. Amount of Evans Blue dye in tumor tissues was expressed as micrograms
of dye per gram of tissue. n = 3. *P = .039. (E) SEMA3F extends overall survival in mice bearing brain tumors. SEMA3F reexpression
extends overall survival in mice bearing brain tumors, whereas TSP2 does not show any beneficial effect. n = 7 for dsred, n = 8 for
SEMA3F, and n = 10 for TSP2. *P < .01 by log-rank test.
Neoplasia Vol. 14, No. 2, 2012Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.
findings led us to propose that the loss of merlin could contribute to an
abnormal tumor vasculature by disrupting the delicate balance of
proangiogenic and antiangiogenic factors.
In the current study, we indeed found that merlin regulates expres-
sion of class 3 SEMAs through the Rho GTPase Rac1 specifically, but
not the other close family members such as cdc42 or Rho A. Rac1
expression levels were found upregulated as well as Rac1 protein trans-
located to the membrane ruffles in human schwannoma cells lacking
wild-type copies of merlin . Merlin was subsequently found to
regulate cell proliferation and contact inhibition dominantly by inter-
fering with the guanine exchange of Ras and Rac and its downstream
signaling cascades involving Raf and ERK . In addition, Rac1 ac-
tivity has been reported to modulate growth cone collapse induced by
SEMA3A [30–32]. Our findings established a signaling cascade con-
nection whereby Rac1 works downstream of merlin and upstream of
SEMA3F (Figure 5). However, as blocking the expression and activity
of Rac1 was able to virtually erase SEMA3F expression, it is likely that
the regulation involves multiple steps. Histone deacetylase may rep-
resent an intermediate step between Rac1 and SEMA3F because its
activity directly correlated with active Rac1 [33,34] and histone dea-
cetylase inhibition markedly stimulated the activity of SEMA3F .
By reintroducing SEMA3F into merlin-deficient schwannoma
cells, we further confirmed the role of SEMA3F as an antiangiogenic
factor in our experimental settings. Our data are consistent with the
published data that expression of SEMA3F was specifically down-
regulated in NF2-driven schwannomas  as well as in tumors
of the pancreatic and cervical origins  and that reexpressing
SEMA3F is able to decrease tumor vascularity and restore pericyte
coverage in vivo [19,36]. On the basis of this potent angioinhibitory
effect, we are the first to show that orthotopic schwannoma tumors
overexpressing SEMA3F were smaller in size and the lower tumor
burden resulted in extension of overall survival, similar to the effect
observed in mice carrying pancreatic tumors overexpressing SEMA3A
. Furthermore, we showed that reexpressing SEMA3F contrib-
uted to an effective decrease in permeability. Although much less is
known about the blood-nerve barrier, permeability has been shown
to play a role in schwannomas and its control might have a direct
In addition to SEMA3F, we also found that TSP2 was highly up-
regulated in Nf2−/− schwannoma cells that were reintroduced with
wild-type merlin. This up-regulation is consistent with the idea that
Rac1 is working upstream because it has been reported that a con-
stitutively active mutant of Rac1 (RacV12) is able to regulate TSP2
expression at the transcription level in endothelial cells . However,
we did not observe any detectable antiangiogenic properties with
merlin-deficient schwannoma cells that overexpress TSP2 in vivo
(Figure 4). In general, thrombospondins are antiangiogenic .
However, because TSP1 and TSP2 are composed of multiple func-
tional domains and each has its own receptors, the consequence of
TSP expression may not always be simply antiangiogenic. In fact, there
is ample evidence that TSP1 is not antiangiogenic and even pro-
angiogenic [38–41]. In all these studies, it can be concluded that the
overall angiogenic outcome initiated by TSPs is largely determined by
1) specific domain interaction with appropriate receptors, (2) absolute
concentration of TSPs, and/or (3) the unique microenvironment of
the experimental models. Because TSP2 has an equivalent domain
structure to TSP1 and the microenvironment of nerve tumors is not
completely explored, it is plausible to assume that the domain/receptor
interactionsbetween the tumor-secretedTSP2 and the stroma could be
complex, and thus, the overall angiogenic outcome is difficult to pre-
dict. Moreover, because TSP2 has pleiotropic effects and is involved in
processes as disparate as bone growth, homeostasis, and foreign body
response , itisalsopossiblethatTSP2isnotsimply angioinhibitory
in the nerve microenvironment. Nevertheless, it is essential to establish
a causal correlation between TSP2 and merlin in clinical samples of
NF2-related schwannomas so that the precise role of TSP2 can be
investigated in the pathogenesis of schwannomas.
Furthermore, there are few data on the regulation and interaction
between SEMA3F, RAC1, and TSP2. Our data suggest SEMA3F is
but the reverse is not true. SEMA3F works upstream of RhoA and
ROCK because this signaling cascade leads to the activation of cofilin
and actin depolymerization in both glioma and endothelial cells .
Inaddition,RhoA and ROCKareabletosuppressTSP1through Myc,
a close member of TSP2 . Interestingly, Myb, a nuclear transcrip-
tion factor in the same class of Myc, has been shown to suppress TSP2
specifically . Therefore, we speculate that SEMA3F works up-
stream of TSP2, possibly through RhoA, ROCK, and Myb.
In summary, this study revealed that merlin/NF2, a tumor sup-
pressor gene, on top of its antiproliferative function on contact inhi-
bition, also influenced tumor angiogenesis by regulating expression of
SEMA3F, a potent angioinhibitory factor, and this regulation worked
through Rac1. This study further advanced our knowledge that angio-
genesis is essential in benign tumor progression, and by modulating
the relative balance of proangiogenic and antiangiogenic components
Figure 5. The proposed mechanism by which merlin loss supports
neurofibromatosis angiogenesis. In the presence of wild-type merlin,
Rac1 activity is maintained at a normal state, as well as the levels of
SEMA3F/TSP2.The physiologicalbalanceofproangiogenicand anti-
angiogenic factors supports normal vascularity that is necessary for
healthy neural system functions (A). However, in Schwann cells that
are deficient in merlin, the increased Rac1 activity leads to an en-
hanced suppression of SEMA3F/TSP2 expression—the augmented
relative VEGF levels and the subsequent pathologic changes thus
are able to drive abnormal angiogenesis to further support the nutri-
tious need for the unlimited proliferation of schwannoma cells (B). It
is likely that the mechanism(s) by which Rac1 suppresses SEMA3F
and TSP2 involves multiple steps, and the precise regulatory mech-
anisms remain to be investigated.
Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.Neoplasia Vol. 14, No. 2, 2012
in these tumors, one can successfully control tumor development.
This study provides a mechanism for the potential “angiogenic
switch” in NF2-driven schwannoma. Benign tumors such as schwan-
nomas progress very slowly, and understanding the mechanism be-
hind the modulation of the proangiogenic and antiangiogenic
balance will contribute to improving treatment of these tumors.
Although treatment with anti-VEGF therapies has proven successful
so far, the long-term treatment of these patients necessitates the iden-
tification of new therapeutic regimens. To date, although SEMA3F
is well established as an inhibitor of angiogenesis, the regulatory
mechanisms of SEMA3F, or other secreted SEMAs that possess po-
tent antiangiogenic properties, are not completely understood. In fact,
small molecules that can stimulate SEMA3s expression or activity
have not been developed. We strongly believe that more effort should
be focused on finding drugs/small molecules that can enhance the
SEMA-NRP-plexin signaling event. Modulating angiogenic balance
by inhibiting proangiogenesis (blocking VEGF ), or promoting
antiangiogenesis (reintroducing SEMA3s, this study), proved to be
a successful way to control schwannoma progression and should be
considered as a novel approach to treat neurofibromatosis-related
tumors similar to malignant tumors .
The authors thank Sylvie Roberge and Carolyn Smith for their tech-
nical expertise, Padera T. Timothy and Lei Xu for discussions and
guidance, and Andrew B. Gladden, Andrea I. McClatchey, and Brian
Seed (Massachusetts General Hospital and Harvard Medical School,
Boston, MA) for providing critical reagents.
 Asthagiri AR, Parry DM, Butman JA, Kim HJ, Tsilou ET, Zhuang Z, and
Lonser RR (2009). Neurofibromatosis type 2. Lancet 373, 1974–1986.
 Mrugala MM, Batchelor TT, and Plotkin SR (2005). Peripheral and cranial
nerve sheath tumors. Curr Opin Neurol 18, 604–610.
 Plotkin SR, Stemmer-Rachamimov AO, Barker FG II, Halpin C, Padera TP,
Tyrrell A, Sorensen AG, Jain RK, and di Tomaso E (2009). Hearing improve-
ment after bevacizumab in patients with neurofibromatosis type 2. N Engl J Med
 Wong HK, Lahdenranta J, Kamoun WS, Chan AW, McClatchey AI, Plotkin SR,
Jain RK, and di Tomaso E (2010). Anti–vascular endothelial growth factor
therapies as a novel therapeutic approach to treating neurofibromatosis-related
tumors. Cancer Res 70, 3483–3493.
 Caye-Thomasen P, Baandrup L, Jacobsen GK, Thomsen J, and Stangerup SE
(2003). Immunohistochemical demonstration of vascular endothelial growth
factor in vestibular schwannomas correlates to tumor growth rate. Laryngoscope
 Caye-Thomasen P,Werther K, NallaA, Bog-Hansen TC,Nielsen HJ, Stangerup
SE, and Thomsen J (2005). VEGF and VEGF receptor-1 concentration in ves-
tibular schwannoma homogenates correlates to tumor growth rate. Otol Neurotol
 Koutsimpelas D, Stripf T, Heinrich UR, Mann WJ, and Brieger J (2007).
Expression of vascular endothelial growth factor and basic fibroblast growth
factor in sporadic vestibular schwannomas correlates to growth characteristics.
Otol Neurotol 28, 1094–1099.
 Uesaka T, Shono T, Suzuki SO, Nakamizo A, Niiro H, Mizoguchi M, Iwaki T,
and Sasaki T (2007). Expression of VEGF and its receptor genes in intracranial
schwannomas. J Neurooncol 83, 259–266.
 Jain RK (2005). Normalization of tumor vasculature: an emerging concept in
antiangiogenic therapy. Science 307, 58–62.
 Yazdani U and Terman JR (2006). The semaphorins. Genome Biol 7, 211.
 Chen H, He Z, and Tessier-Lavigne M (1998). Axon guidance mechanisms:
semaphorins as simultaneous repellents and anti-repellents. Nat Neurosci 1,
 Luo Y, Raible D, and Raper JA (1993). Collapsin: a protein in brain that
induces the collapse and paralysis of neuronal growth cones. Cell 75, 217–227.
 Puschel AW, Adams RH, and Betz H (1995). Murine semaphorin D/collapsin
is a member of a diverse gene family and creates domains inhibitory for axonal
extension. Neuron 14, 941–948.
 Miao HQ, Soker S, Feiner L, Alonso JL, Raper JA, and Klagsbrun M (1999).
Neuropilin-1 mediates collapsin-1/semaphorin III inhibition of endothelial cell
motility: functional competition of collapsin-1 and vascular endothelial growth
factor-165. J Cell Biol 146, 233–242.
 Kessler O, Shraga-Heled N, Lange T, Gutmann-Raviv N, Sabo E, Baruch L,
Machluf M, and Neufeld G (2004). Semaphorin-3F is an inhibitor of tumor
angiogenesis. Cancer Res 64, 1008–1015.
 Neufeld G, Lange T, Varshavsky A, and Kessler O (2007). Semaphorin signal-
ing in vascular and tumor biology. Adv Exp Med Biol 600, 118–131.
 Bielenberg DR and Klagsbrun M (2007). Targeting endothelial and tumor cells
with semaphorins. Cancer Metastasis Rev 26, 421–431.
 Takahashi T and Strittmatter SM (2001). Plexina1 autoinhibition by the plexin
sema domain. Neuron 29, 429–439.
 Bielenberg DR, Hida Y, Shimizu A, Kaipainen A, Kreuter M, Kim CC, and
Klagsbrun M (2004). Semaphorin 3F, a chemorepulsant for endothelial cells,
induces a poorly vascularized, encapsulated, nonmetastatic tumor phenotype.
J Clin Invest 114, 1260–1271.
 Coma S, Amin DN, Shimizu A, Lasorella A, Iavarone A, and Klagsbrun M
(2010). Id2 promotes tumor cell migration and invasion through transcriptional
repression of semaphorin 3F. Cancer Res 70, 3823–3832.
 Shimizu A, Mammoto A, Italiano JE Jr, Pravda E, Dudley AC, Ingber DE, and
Klagsbrun M (2008). ABL2/ARG tyrosine kinase mediates SEMA3F-induced
RhoA inactivation and cytoskeleton collapse in human glioma cells. J Biol Chem
 Ketteler R, Glaser S, Sandra O, Martens UM, and Klingmuller U (2002).
Enhanced transgene expression in primitive hematopoietic progenitor cells
and embryonic stem cells efficiently transduced by optimized retroviral hybrid
vectors. Gene Ther 9, 477–487.
 Gualdoni S, Albertinazzi C, Corbetta S, Valtorta F, and de Curtis I (2007).
Normal levels of Rac1 are important for dendritic but not axonal development
in hippocampal neurons. Biol Cell 99, 455–464.
 Hung G, Li X, Faudoa R, Xeu Z, Kluwe L, Rhim JS, Slattery W, and Lim D
(2002). Establishment and characterization of a schwannoma cell line from a
patient with neurofibromatosis 2. Int J Oncol 20, 475–482.
 Prabhakar S, Messerli SM, Stemmer-Rachamimov AO, Liu TC, Rabkin S,
Martuza R, and Breakefield XO (2007). Treatment of implantable NF2
schwannoma tumor models with oncolytic herpes simplex virus G47Δ. Cancer
Gene Ther 14, 460–467.
 Kaempchen K, Mielke K, Utermark T, Langmesser S, and Hanemann CO
(2003). Upregulation of the Rac1/JNK signaling pathway in primary human
schwannoma cells. Hum Mol Genet 12, 1211–1221.
 Flaiz C, Kaempchen K, Matthies C, and Hanemann CO (2007). Actin-rich
J Neuropathol Exp Neurol 66, 608–616.
 Lopes N, Gregg D, Vasudevan S, Hassanain H, Goldschmidt-Clermont P, and
Kovacic H (2003). Thrombospondin 2 regulates cell proliferation induced by
Rac1 redox-dependent signaling. Mol Cell Biol 23, 5401–5408.
 Morrison H, Sperka T, Manent J, Giovannini M, Ponta H, and Herrlich P
(2007). Merlin/neurofibromatosis type 2 suppresses growth by inhibiting the
activation of Ras and Rac. Cancer Res 67, 520–527.
 Jin Z and Strittmatter SM (1997). Rac1 mediates collapsin-1–induced growth
cone collapse. J Neurosci 17, 6256–6263.
 Kuhn TB,Brown MD,WilcoxCL,Raper JA, and Bamburg JR (1999). Myelin and
collapsin-1 induce motor neuron growth cone collapse through different pathways:
inhibition of collapse by opposing mutants of rac1. J Neurosci 19, 1965–1975.
 Vastrik I, Eickholt BJ, Walsh FS, Ridley A, and Doherty P (1999). Sema3A-
induced growth-cone collapse is mediated by Rac1 amino acids 17-32. Curr Biol
 Kim YB, Yu J, Lee SY, Lee MS, Ko SG, Ye SK, Jong HS, Kim TY, Bang YJ,
and Lee JW (2005). Cell adhesion status–dependent histone acetylation is reg-
ulated through intracellular contractility–related signaling activities. J Biol Chem
 Gao YS, Hubbert CC, Lu J, Lee YS, Lee JY, and Yao TP (2007). Histone
deacetylase 6 regulates growth factor–induced actin remodeling and endocytosis.
Mol Cell Biol 27, 8637–8647.
Neoplasia Vol. 14, No. 2, 2012 Merlin Regulates Angiogenesis through Rac1/SEMA3FWong et al.
 Kusy S, Potiron V, Zeng C, Franklin W, Brambilla E, Minna J, Drabkin HA,
and Roche J (2005). Promoter characterization of semaphorin SEMA3F, a
tumor suppressor gene. Biochim Biophys Acta 1730, 66–76.
 Maione F, Molla F, Meda C, Latini R, Zentilin L, Giacca M, Seano G, Serini G,
Bussolino F, and Giraudo E (2009). Semaphorin 3A is an endogenous angio-
genesis inhibitor that blocks tumor growth and normalizes tumor vasculature in
transgenic mouse models. J Clin Invest 119, 3356–3372.
 Bornstein P (2009). Thrombospondins function as regulators of angiogenesis.
J Cell Commun Signal 3, 189–200.
 Chandrasekaran L, He CZ, Al-Barazi H, Krutzsch HC, Iruela-Arispe ML,
and Roberts DD (2000). Cell contact–dependent activation of α3β1integrin
modulates endothelial cell responses to thrombospondin-1. Mol Biol Cell 11,
 Qian X, Wang TN, Rothman VL, Nicosia RF, and Tuszynski GP (1997).
Thrombospondin-1 modulates angiogenesis in vitro by up-regulation of matrix
metalloproteinase-9 in endothelial cells. Exp Cell Res 235, 403–412.
 StaniszewskaI,ZaveriS,Del ValleL,Oliva I,RothmanVL,Croul SE,RobertsDD,
Mosher DF, Tuszynski GP, and Marcinkiewicz C (2007). Interaction of α9β1
integrin with thrombospondin-1 promotes angiogenesis. Circ Res 100, 1308–1316.
 Taraboletti G, Morbidelli L, Donnini S, Parenti A, Granger HJ, Giavazzi R, and
Ziche M (2000). The heparin binding 25 kDa fragment of thrombospondin-1
promotes angiogenesis and modulates gelatinase and TIMP-2 production in
endothelial cells. FASEB J 14, 1674–1676.
 Bornstein P, Armstrong LC, Hankenson KD, Kyriakides TR, and Yang Z
(2000). Thrombospondin 2, a matricellular protein with diverse functions.
Matrix Biol 19, 557–568.
 Watnick RS, Cheng YN, Rangarajan A, Ince TA, and Weinberg RA (2003). Ras
modulates Myc activity to repress thrombospondin-1 expression and increase
tumor angiogenesis. Cancer Cell 3, 219–231.
 Bein K, Ware JA, and Simons M (1998). Myb-dependent regulation of
thrombospondin 2 expression. Role of mRNA stability. J Biol Chem 273,
Merlin Regulates Angiogenesis through Rac1/SEMA3F Wong et al.Neoplasia Vol. 14, No. 2, 2012