Cooperative interactions of BRAFV600Ekinase and
CDKN2A locus deficiency in pediatric malignant
astrocytoma as a basis for rational therapy
Emmanuelle Huillarda,b,1,2, Rintaro Hashizumec,1, Joanna J. Phillipsc,d, Amélie Griveaua,b, Rebecca A. Ihrieb,c,3,
Yasuyuki Aokic, Theodore Nicolaidesa,c, Arie Perryc,d, Todd Waldmane, Martin McMahonf, William A. Weissa,c,g,
Claudia Petritschc, C. David Jamesc,4, and David H. Rowitcha,b,c,4
Departments ofaPediatrics,cNeurological Surgery,gNeurology, anddNeuropathology,bEli and Edyth Broad Institute for Stem Cell Research and Regeneration
Medicine and Howard Hughes Medical Institute, andfDepartment of Cellular and Molecular Pharmacology, Helen Diller Family Comprehensive Cancer
Center, University of California, San Francisco, CA 94143; andeDepartment of Oncology, Lombardi Comprehensive Cancer Center, Georgetown University,
Washington, DC 20057
Edited by Webster K. Cavenee, Ludwig Institute, University of California San Diego, La Jolla, CA, and approved April 16, 2012 (received for review October
Although malignant astrocytomas are a leading cause of cancer-re-
lated death in children, rational therapeutic strategies are lacking.
We previously identified activating mutations of v-raf murine sar-
coma viral oncogene homolog B1 (BRAF) (BRAFT1799Aencoding
inhibitor 2A (CDKN2A, encoding p14ARF and p16Ink4a) deletions in
pediatric infiltrative astrocytomas. Here we report that BRAFV600E
expression in neural progenitors (NPs) is insufficient for tumorigen-
esis and increases NP cellular differentiation as well as apoptosis. In
contrast, astrocytomas are readily generated from NPs with addi-
tional Ink4a-Arf deletion. The BRAFV600Einhibitor PLX4720 signifi-
cantly increased survival of mice after intracranial transplant of
genetically relevant murine or human astrocytoma cells. Moreover,
combination therapy using PLX4720 plus the Cyclin-dependent ki-
nase (CDK) 4/6-specific inhibitor PD0332991 further extended sur-
vival relative to either monotherapy. Our findings indicate a rational
therapeutic strategy for treating a subset of pediatric astrocytomas
with BRAFV600Emutation and CDKN2A deficiency.
glioma|protein kinase|tumor suppressor
tumors is second only to leukemia among malignancies af-
including anaplastic astrocytoma World Health Organization
(WHO) grade III and glioblastoma WHO grade IV, is a leading
cause of mortality. However, development of rational targeted
therapies for pediatric malignant astrocytoma has been limited by
inadequate information regarding the relevant underlying genetic
alterations and their signaling pathway consequences. In adults,
basis of gene expression patterns (2, 3) and mutations affecting
spectrum and frequency of mutations in pediatric malignant as-
trocytoma, however, differ from that in adults. For example,
whereas mutations and amplifications of EGFR are common in
adult glioblastomas (GBM), they are observed only rarely in
corresponding pediatric anaplastic astrocytomas or GBM (5–10).
Similarly, mutations of PTEN and IDH1 are relatively uncommon
in pediatric astrocytomas (5). Recent studies have reported re-
current mutations of histone H3 and chromatin remodeling genes
in pediatric GBM (6, 7).
Previously, we used a whole genome approach to identify copy
number variance and mutations in pediatric astrocytomas (11).
This analysis revealed activating mutations of v-raf murine sar-
coma viral oncogene homolog B1 (BRAF) (BRAFT1799Aencoding
BRAFV600E) in 7/31 (23%) cases of pediatric diffuse astrocytoma
(WHO grades II–IV), consistent with subsequent findings from
others, and is in contrast to adult GBMs, which infrequently show
BRAFV600Emutation (12, 13). In our series, BRAFT1799Aoccurred
he incidence of pediatric central nervous system (CNS)
coincident with homozygous deletion of cyclin-dependent kinase
inhibitor 2A (CDKN2A), encoding Ink4a-Arf in 5/7 cases (71%),
suggesting the possibility of mechanistic cooperation between
these two alterations in promoting glial tumor malignancy.
Here we investigated genetic requirements for generation of
malignant astrocytomas in mice and used this information in
preclinical testing. We show that expression of BRAFV600E
mutation in combination with homozygous CDKN2A/Ink4a-Arf
deletion is sufficient for formation of tumors with histology
typical of malignant astrocytomas in humans. Further, we have
used this mouse model, as well as human malignant astrocy-
toma xenografts with genetically faithful BRAFV600Eand Ink4a-
Arf mutations, to test a unique therapeutic strategy. We report
that combination therapy with BRAFV600Eand Cyclin-de-
pendent kinase (CDK) 4/6 inhibitors has remarkable activity
against intracranial tumors in vivo. Together our data suggest
that it is possible to faithfully model the genetics of a subset of
pediatric malignant astrocytomas in mice and to use combined
murine and human tumor models as platforms for testing
Strategy for Faithful Regulation of BRAFV600EExpression in Mouse
Neural Progenitor Cells to Model Pediatric Malignant Astrocytoma.
To precisely model activating mutations of BRAF found in pedi-
atric astrocytomas, we used BrafCAmice carrying a genetically
engineered knock-in allele of Braf (14). The BrafCAallele
expresses normal Braf in its nonrecombined configuration.
However, upon exposure to bacteriophage P1 cre recombinase,
the BrafCAallele is recombined to encode mutationally activated
BRAFV600Eat normal physiological levels of expression. By using
the BrafCAallele in a heterozygous configuration, we precisely
mimic the status of heterozygous BRAF mutations in human
Author contributions: E.H., R.H., R.A.I., C.P., C.D.J., and D.H.R. designed research; E.H., R.H.,
J.J.P., A.G., R.A.I., Y.A., and C.P. performed research; T.N., T.W., and M.M. contributed new
reagents/analytic tools; E.H., R.H., J.J.P., T.N., A.P., T.W., M.M., W.A.W., C.D.J., and D.H.R.
analyzed data; and E.H., R.H., C.D.J., and D.H.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
1E.H. and R.H. contributed equally to this work.
2Present address: Université Pierre et Marie Curie S975/Institut National de la Santé et de la
Recherche Médicale U975/Centre National de la Recherche Scientifique UMR7225, Centre
3Present address: Vanderbilt University Medical Center, Nashville, TN 37232-6840.
4To whom correspondence may be addressed. E-mail: email@example.com or rowitchd@
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| May 29, 2012
| vol. 109
| no. 22www.pnas.org/cgi/doi/10.1073/pnas.1117255109
pediatric tumors (11). We considered this strategy more accurate
for investigating the expression of BRAFV600Ethan introduction
by lentivirus, in which case the expression of a single splice variant
of mutated BRAF would be driven by a strong viral promoter that
is not subject to regulation by the factors involved in control of
endogenous BRAF expression in normal or malignant cells. In-
deed, the importance of faithful BRAFV600Eregulation is in-
dicated by other recent studies (14–17).
To assess whether BRAF activation alone is sufficient for tu-
mor formation from neural progenitors (NPs), we crossed
hemizygous mice to hGFAP-cre transgenic mice
expressing cre under the control of the hGFAP promoter, which
targets neural stem cells and astroglia in the developing brain
(18). The resulting hGFAP-cre; BrafCAcompound heterozygous
mice displayed an enlarged brain with expanded populations of
type A neuroblasts in the subventricular zone, oligodendrocytes,
and astroglia. Such animals survived until approximately 3 wk of
age but did not demonstrate evidence of tumor development in
the brain. Because of the lethality associated with BRAFV600E
expression in these transgenic mice, we used an alternative sys-
tem of orthotopic transplant of tumorigenic progenitor cells
harvested from embryonic ventral telencephalon (Fig. 1A) (19,
20). hGFAP-cre; BrafCA/+NPs were harvested and expanded for
transplant into SCID mice by intracranial injection. In this ex-
perimental paradigm, mice undergo injection of 2 × 105dis-
associated NP cells and are then monitored for symptoms
indicative of tumor development. As shown in Table 1, of five
mice injected with cells expressing BRAFV600Ealone, we ob-
served no evidence of gross or microscopic tumor formation
after 6 mo. We obtained similar results with NP cells derived
from Olig2-cre; BrafCA/+mice (n = 6, Table 1).
BRAFV600EExpression in Combination with Ink4a-Arf Loss of Function
Is Sufficient for Tumorigenesis. BrafCA/+mice were then crossed
with mice lacking Ink4a-Arf to test the influence of p16Ink4aand
p19Arftumor suppressor function. In contrast to the findings
above, expression of BRAFV600Ein Ink4a-Arf–deficient neuro-
spheres (line 10740) was associated with tumor formation 3–4
mo after intracranial implantation (Table 1 and Fig. 1B). We
obtained comparable results in SCID mice injected with identical
BrafCA/+; Ink4a-Arf−/−cells infected with an adenovirus encoding
cre (Ad-cre; line 10776) (Table 1 and Fig. 1B). These findings
indicate that expression of BRAFV600Ecooperates with Ink4a-
Arf deletion in tumor formation. Our results do not rule out the
possibility of tumor formation at later time points, but none-
theless suggest that expression of BRAFV600Ealone is in-
sufficient to confer tumorigenicity to NPs.
Tumors exhibited restricted infiltration and tended to grow exo-
phytically despite high mitotic index, nuclear pleomorphism, and
expression of cellular markers typical of human pediatric astrocy-
and 10776 were further manipulated by introduction of a luciferase
reportertoallow noninvasive bioluminescence imaging(BLI).After
passage in the murine brain, the tumors took on a highly infiltrative
growth pattern characteristic of malignant astrocytoma with diffuse
invasion throughout the cerebral hemisphere and extension along
white matter tracts (see Fig. 4A). Tumor cells exhibited marked
pleomorphism, had a high mitotic index, and expressed phosphor-
ylated ERK (Fig. S2), which is an expected downstream conse-
quence of BRAF activation. Because line 10776 demonstrated
To further investigate cooperative interactions between
into the subventricular zone of mice carrying BrafCA/+alone or
BrafCA/+; Ink4a-arffl/fl(Fig. 1C). No tumor formation was observed
following injections of BrafCA/+mice for up to 490 d postinjection
(Fig. 1D and Table 1). In contrast, intracranial tumor formation
was observed in all BrafCA/+Ink4a-arffl/flmice injected with Ad:cre
(median survival, 70 d). These findings indicate that heterozygous
BRAFV600Eexpression alone is insufficient to promote tumor de-
velopment in vivo. In contrast, such BRAFV600Eexpression is
transforming when combined with homozygous inactivation
BRAFV600EMutation Combined with Ink4a-Arf Deletion Results in
Reduced Differentiation and Apoptosis, as Well as Enhanced
Proliferation of Neural Progenitors. To further investigate the ap-
parent cooperativity of BRAFV600Eand Ink4a-Arf mutations, we
cultured NPs that harbored these mutations. As shown in Fig. 2,
hGFAP-cre; BrafCA/+; Ink4a-Arf−/−cells have an increased pro-
portion of bromodeoxyuridine (BrdU) positivity and a decreased
apoptotic fraction compared with cells with intact p16Ink4a.
Moreover, BrafCA/+cells lacking Ink4a-Arf have an impaired ca-
pacity to differentiate into neurons, oligodendrocytes, and astro-
when subjected to differentiation culture conditions, as shown by
their high Ki67-proliferative labeling indices (Fig. 2 C and D).
CDK4/6 Activity Is Required for Cell Cycle Progression of BRAFV600E;
Ink4a-Arf−/−Murine Astrocytoma Cells. p16 suppresses the activity
of CDK4 and CDK6, which in turn antagonize retinoblastoma 1
(RB) activity to promote cell cycle progression. PD0332991 is an
orally administered CDK4/6 inhibitor in clinical trials to assess
efficacy against several types of cancer (22). We decided to test
whether elevated CDK4/6 activity is necessary to sustain cell cycle
progression in BRAFV600E; Ink4a-Arf−/−NP cells. PD0332991
treatment significantly reduced BrdU incorporation and in-
creased the G1 phase fraction of BRAFV600ENP cells that were
Ink4a-Arf−null, but not Ink4a-Arf intact (Fig. 3 A and B), sug-
gesting that sustained CDK4/6 activity is necessary for cell cycle
progression. Use of ImageJ to analyze and compare phosphory-
lated Rb (pRb) band intensity in treated vs. untreated cells, for
replicate experiments and multiple Western blot analyses, con-
sistently revealed 50–80% decrease in pRb for PD0332991-trea-
tedInk4a-Arf−/−cells,relative tountreated Ink4a-Arf−/−cells,and
0 50100 150
Days post injection
Days post injection
0100 200300 400
malignant astrocytoma from mouse neural progenitors. (A) Mouse model
for pediatric malignant astrocytoma. GE, ganglionic eminence; ctx, cortex.
(B) Kaplan–Meier survival curves of SCID mice transduced with hGFAP-cre
BrafCA/+Ink4a-Arf−/−(n = 6) or Ad:cre BrafCA/+Ink4a-Arf−/−(n = 5) neural
progenitor cells (P = 0.64). (C) Injection of an adenovirus encoding ubiqui-
tous cre expression into BrafCA/+or BrafCA/+Ink4a-arfLoxP/LoxPadult mice. (D)
Kaplan–Meier survival curves of BrafCA/+(n = 6) and BrafCA/+Ink4a-arfLoxP/LoxP
(n = 9) mice injected with Ad:cre into the lateral ventricle (P = 0.0004).
Concurrent BRAFV600Eactivation and Ink4a-arf deletion induces
Huillard et al.PNAS
| May 29, 2012
| vol. 109
| no. 22
following normalization of pRb signal against either corre-
results are shown in Fig. 3C.
Dual BRAFV600Eand CDK4/6 Inhibition Shows Additive Effects Against
Murine 10776 Astrocytoma Progenitors in Vivo. PLX4720 is a tool
clinical results in patients with BRAFV600Emetastatic melanoma
(23). We found significant therapeutic benefit of PLX4720 mono-
therapy against murine 10776 astrocytoma, as indicated by bio-
luminescence monitoring and survival of treated mice (Fig. 4 B and
C). Similarly, PD0332991 monotherapy provided a significant ben-
efit (Fig. 4 B and C). To determine possible additive antitumor
activity, a treatment group was included to evaluate PLX4720 +
PD0332991 combination therapy. Bioluminescence monitoring of
growth rate, relative to either monotherapy (Fig. 4B), as well as
increased survival benefit (Fig. 4C). Analysis of intratumor Ki67
staining showed reduced proliferation in mono and dual PLX4720-
and PD0332991-treated samples compared with control tumors
(Fig. 4 D and E). There were significantly fewer Ki67+cells in the
samples treated with the combination therapy compared with
Human GBM Cell Line with BRAFV600EMutation and Deletion of
CDKN2A Responds to Dual Inhibitor Therapy. We next extended
results to a human malignant astrocytoma xenograft model,
which used intracranial xenografts of the DBTRG05-MG astro-
cytoma cell line, previously determined as harboring BRAFV600E
(24), and for which we subsequently detected homozygous de-
letion of CDKN2A (25). As with the 10776 cells, DBTRG05-MG
cells were modified with luciferase to enable bioluminescence
monitoring. DBTRG05-MG xenografts showed features of high-
grade astrocytoma, such as invasion, anaplastic tumor cell mor-
phology, Nestin expression, and high mitotic index (Fig. 5A).
Consistent with results obtained with the murine allograft model,
we found that treatment of DBTRG05-MG tumors with each
monotherapy resulted in reduced tumor growth rate and con-
ferred substantial survival benefit, and combination therapy sig-
nificantly outperformed either monotherapy (Fig. 5 B and C).
Results from the analysis of tumor Ki67 positivity also revealed
consistency with those obtained with the murine model: all
therapies significantly reduced malignant astrocytoma xenograft
cell proliferation relative to the vehicle treatment group (Fig.
5D). The treatment of a second human glioma line, AM-38, also
showed combination therapy as being most effective at inhibiting
tumor growth and extending animal survival (Fig. S3).
To further address the importance of tumor cell BRAFV600E
and CDKN2A-RB status as key determinants of tumor response
to these small molecule inhibitors, we treated mice with in-
tracranial GS2 malignant astrocytomas, that are wild type for
BRAF, express p16Ink4a, but lack RB protein (Fig. S4), with
mono- and combination therapies as before with the other tumor
models. GS2 tumors showed no response to either inhibitor (Fig.
5 E and F), demonstrating specificity of responses to PLX4720
and PD0332991, respectively.
To gain insight into the additional benefit of PLX4720 and
PD0332991, we analyzed the signaling pathway responses to
monotherapies and combination therapy (Fig. S4). As expected,
treatment of BRAFV600ECDKN2A−/−cells with the CDK4/6 in-
hibitor PD0332991 reduced RB phosphorylation, and treatment
with the BRAF inhibitor PLX4720 suppressed phosphorylation of
extracellular signal-regulated kinase (ERK) and MAPK/ERK ki-
nase (MEK). Surprisingly, however, treatment of BRAFV600E
CDKN2A−/−cell lines with PLX4720 alone resulted in elevated
returned to basal level phosphorylation when cells were cotreated
% c-casp3+ cells
differentiated than hGFAP-cre; BrafCA/+cells. (A) Immunocytochemical anal-
ysis of proliferation (BrdU) and apoptosis (cleaved caspase 3) markers in
BrafCA/+and BrafCA/+Ink4a-Arf−/−neural progenitors cultures. (Scale bars, 100
µm.) (B) Quantification of the numbers of BrdU+and Cleaved-caspase 3+cells.
(C) Corresponding analysis of the differentiation markers βIII-tubulin (neu-
ronal), GFAP (astrocyte), O4 (oligodendrocyte), and the proliferation marker
Ki67 in neural progenitor cultures under differentiation culture conditions.
(Scale bars, 100 µm.) (D) Quantification of the numbers of cells expressing the
differentiation markers (mean ± SEM). ***P < 0.0005, **P < 0.005.
hGFAP-cre; BrafCA/+Ink4a-Arf−/−cells are more proliferative and less
Table 1. Summary of the orthotopic transplants of murine cells and in vivo injections
No. of cell
No. of animals
Adult BrafV600Efl/+ Ink4a-Arffl/fl
*Mice were killed at 180 d postinjection and did not show signs of tumor or hyperplasia.
†Three mice were killed at 86–163 d postinjection and did not show signs of tumor or hyperplasia.
| www.pnas.org/cgi/doi/10.1073/pnas.1117255109 Huillard et al.
with PD0332991. This unexpected result, showing combination
treatment suppression of paradoxical stimulation of AKT by
PLX4720 treatment alone, provides unique insight regarding the
molecular basis of combination therapy antiproliferative effects on
Although malignant pediatric astrocytomas are particularly le-
thal CNS cancers, advances in therapy have been minimal, im-
paired by a poor grasp of the relevant underlying oncogenic
signaling pathways. We have recently reported that ∼20% of
pediatric astrocytomas (WHO grades II–IV) carry the activating
mutations in BRAF, which occurs commonly in combination with
homozygous deletion of the CDKN2A locus, encoding p19ARF
and p16Ink4a(11). Here we extend these findings by showing pro-
tumorigenic interactions of pathways regulated by BRAFV600E
Cooperative Interactions of BRAFV600EExpression and Ink4a-Arf Loss
of Function. Known BRAF alterations in human pilocytic astro-
cytomas include copy number gains, BRAF-KIAA1549 fusion (13,
26–28), and the activating BRAFV600Emutation (13). Combined
expression of BRAFV600Ewith silencing of CDKN2A has been
observed in other cancers such as melanoma (29, 30). To in-
vestigate the relationship between BRAFV600Eand Ink4a-Arf
deficiency in the development of malignant astrocytoma, we took
advantage of a mouse model with a cre-conditional (floxed) al-
lele of Braf in which the activating mutation is introduced in
response to cre recombinase activity. This approach allows
a precise recapitulation of BRAFV600Eexpression by endoge-
nous cis-acting regulatory sequences.
Although NP with heterozygous BRAFV600Eactivation alone
failed to form tumors after 6 mo postimplantation in SCID
0 200 400 600 800 1000
0 200 400 600 800 1000
% cell population
200 400 600 800 1000
200 400 600 800 1000
Arf−/−genetic alterations for cell cycle progression in murine astrocytoma
cells. BrafCA/+; Ink4a-Arf+/+and BrafCA/+; Ink4a-Arf−/−cells were incubated in
the presence of vehicle (sodium lactate) or 10 μM PD0332991 for 48 h. (A)
Flow cytometry analysis of treated BrafCA/+; Ink4a-Arf+/+(+/+) and BrafCA/+;
Ink4a-Arf−/−(−/−) cells, pulsed with BrdU for 1 h and costained with FITC-
conjugated BrdU and 7-amino-actinomycin D (7-AAD). The intensity of cell
incorporated BrdU (logarithmic mode) vs. total DNA content with 7-AAD
(linear signal amplification mode) is shown. (B) Cell cycle distribution of
control and treated BrafCA/+; Ink4a-Arf+/+(+/+) and BrafCA/+; Ink4a-Arf−/−(−/−)
cells. (C) Immunoblotting analysis of pRb and Rb in treated BrafCA/+; Ink4a-
Arf+/+and BrafCA/+; Ink4a-Arf−/−cells.
CDK4/6 activity is required in the context of dual BRAFV600E; Ink4a-
Days after implantation
Ctrl PLX PDPLX
Days after implantation
5 10 1520 25
Ink4a-Arf−/−murine cells using BRAFV600Eand CDK4/6 inhibitors. (A) Histo-
logical characterization of luciferase-modified BrafCA/+Ink4a-Arf−/−ortho-
topic allografts. (Scale bars, 20 µm.) (B–E) BrafCA/+Ink4a-Arf−/−orthotopic
allografts were treated with PD0332991, PLX4720, or PLX4720 + PD0332991
for 14 consecutive days (pink area). (B) Bioluminescence imaging (BLI); P =
0.0110 for control vs. PLX + PD, P = 0.0341 for PLX vs. PLX + PD, P = 0.0222
for PD vs. PLX + PD. There are no statistically significant differences in tumor
growth rate between control vs. PLX (P = 0.1866) or PD (P = 0.1971). (C)
Kaplan–Meier survival curves; P = 0.0057 for control vs. PLX, P = 0.0069 for
control vs. PD0332991, P = 0.0005 for control vs. PLX4720 + PD0332991.
Although there is a trend for the combination being superior, there are no
statistically significant differences in survival between PD or PLX vs. PD/PLX
treatment groups: P = 0.0637 for combination vs. PLX only, and P = 0.0698
for combination vs. PD only. (D) Immunostaining of the proliferation marker
Ki67 in control vs. treated (PLX, PD, and PLX/PD) tumors. (Scale bars, 20 µm.)
(E) Quantification of the percentage of Ki67+cells in untreated as well as
treated tumor sections (mean ± SEM); ***P < 0.001, **P = 0.006.
Mono- and combination therapy of mice transduced with BrafCA/+
Huillard et al. PNAS
| May 29, 2012
| vol. 109
| no. 22
recipient animals, it is possible that higher BRAFV600Eexpression
of BRAFV600Eusing retrovirus, which might drive higher levels of
BRAF activity or target higher numbers of suitable progenitors, is
sufficient to produce pilocytic astrocytomas in mice (31).
In contrast, we found that heterozygous BRAFV600Eexpression
combined with Ink4-Arf deficiency abrogated NP cell cycle arrest
and conferred tumorgenicity to BRAFV600E-expressing NPs
when transplanted orthotopically into SCID mice or when acti-
vated in subventricular zone NP in situ. These results are con-
sistent with observations reported in a recent study where
BRAFV600Ewas overexpressed in Ink4a-Arf−/−neonatal brain
using a retroviral vector (32) and indicate that the coincidence of
BRAF-activating mutation and deletion of CDKN2A may rep-
resent obligatory steps during initiation and/or malignant pro-
gression of a subset of pediatric astrocytomas. One possibility is
that BRAFV600Einduces p16 expression, promoting cell cycle
arrest, as has been reported in human neural stem cells and
v-raf-1 murine leukemia viral oncogene homolog 1 (RAF)-
expressing human fibroblasts (33–35). From these data we con-
clude that expression of BRAFV600Etriggers an oncogene-in-
duced senescence-like (OIS) response, which is mediated by
elevated p16 expression (Fig. S5). If so, when BRAFV600Eex-
pression is triggered in Ink4a-Arf deficient cells there is no OIS
and tumor progression is permitted.
Our previously published survey of gene alterations in diffuse
pediatric astrocytomas, which included analysis of tumor copy
number variation as well as investigation of selected cancer-as-
sociated genes, such as BRAF and P53, revealed no other gene
alteration with BRAFT1799Aother than homozygous CDKN2A
deletions (11). In addition, the DBTRG05-MG and AM-38 as-
trocytoma lines that we used here do not carry additional
alterations other than PTEN inactivation in DBTRG05-MG
(25). These results do not exclude the possibility of other gene
alterations acting in combination with BRAFV600Eto promote
tumorigenesis, such as the missense mutation of histone H3.3
encoding H3F3A, which has been reported in a significant pro-
portion of pediatric GBMs (6, 7), but the results do support
CDKN2A inactivation as being especially effective in cooperating
with BRAFV600Efor promoting tumor development.
We noted that hGFAP-cre; BRAFV600E; Ink4a-Arf−/−mouse-
derived astrocytomas initially showed a propensity for exophytic
growth. Interestingly, in a recent study (13), BRAFV600Emutations
a superficial human astrocytic tumor that can grow exophytically.
Although this study did not address CDKN2A status, deletion of
CDKN2A has independently been identified in pediatric pleo-
morphic xanthoastrocytomas (36–38). Future work is needed to
determine whether the BRAFV600E, p16-deficient astrocytomas we
describe may initially be tropic for superficial growth.
Combinatorial Therapy May Be Efficacious for a Genetic Subset of
Pediatric Astrocytomas. The finding of cooperativity between
BRAFV600Eactivation and Ink4a-Arf deficiency in murine and
human cells has therapeutic implications; namely, that combined
inhibition of CDK4/6 and BRAFV600Ecould prove efficacious in
treating malignant astrocytomas with these gene alterations. In
the present study, we found that both murine and human
BRAFV600Emutant, Ink4a-Arf–deficient astrocytomas respond
to the BRAFV600Einhibitor PLX4720, as indicated by a reduced
rate of intracranial tumor growth and extended survival. These
results are consistent with our previous report using PLX4720 to
inhibit RAF activity in nonpediatric astrocytoma models (39)
and suggest that PLX4720 has a favorable biodistribution for use
in treating BRAFV600E-driven brain cancers. Importantly, when
we combined PLX4720 with the CDK4/6 inhibitor PD0332991,
we observed additive antitumor effects. Our report uniquely
indicates that the two inhibitors can be used in combination for
increasing antitumor effect, due to therapeutic targeting of dis-
tinct enzymatic activities. Interestingly, the paradoxical increase
in AKT phosphorylation following PLX4720 treatment can be
reversed by cotreatment with PD0332991. This result lends fur-
ther support to the use of combination therapy for treating
BRAFV600Eand CDKN2A-deficient tumors. Taken together,
our data support the importance of tumor genotype for pre-
dicting response to each therapy, used singularly or in combi-
nation. We note that no animal subjects with intracranial
BRAFV600E-p16 null tumors, from any of the three in vivo efficacy
experiments (Figs. 4 and 5 and Fig. S3), died while receiving
combination therapy. This observation raises the question of how
long survival can be extended by continuous combination ther-
apy, and related studies are underway to address this question.
Appropriate genetic diagnostic tests are available for identi-
fying tumors with BRAF mutation and also lacking CDKN2A, and
their routine application to pediatric and adult malignant astro-
cytomas would reliably identify patients who might benefit from
this combination therapy. The frequency of BRAFV600Emutations
in pediatric glioblastomas is estimated between 6 and 18% (11,
13), whereas the frequency in adult glioblastomas is estimated to
be 1–3% (13, 40). In pediatric anaplastic astrocytomas (WHO
grade III), we found BRAFV600Ein 3 of 23 cases (13%, ref. 39).
The frequency of BRAFV600Ein pediatric grade III astrocytomas
has not been published by others. For adult anaplastic astrocy-
tomas, no instances of V600E were determined among 51 cases.
In total, these results suggest that BRAFV600Eoccurs at a signifi-
cant frequency among WHO grade III and grade IV pediatric
astrocytomas, but is rare among corresponding adult tumors.
Our data, although suggestive, need to be further examined
in a preclinical setting to address length of benefit from this
Days after implantation
10 20 30 40 50 60
Days after implantation
Days after implantation
50100 150 200
Days after implantation
Ctrl PLXPD PLX
20 2530 3540 45
termining orthotopic xenograft response to BRAF and CDK4/6 treatment.
Tumors generated from the DBTRG05-MG astrocytoma line were treated
with PD0332991, PLX4720, or PLX4720 + PD0332991 therapies for 14
consecutive days (pink area). (A) Histological features of DBTRG xeno-
grafts. (Scale bars, 20 µm.) (B) Bioluminescence imaging (BLI); P = 0.0013
for control vs. PLX, P = 0.0120 for control vs. PD, P < 0.0001 for control vs.
PLX + PD combination therapy on day 77. P = 0.0018 for PLX vs. PLX + PD,
P = 0.0018 for PD vs. PLX + PD. (C) Kaplan–Meier survival curves; P =
0.0028 for control vs. PLX, P = 0.0005 for control vs. PD0332991; P < 0.001
for PLX or PD vs. PLX4720 + PD0332991 combination therapy. (D) Quan-
tification of the number of Ki67+cells in treated tumors (mean ± SEM);
***P < 0.0001; NS, not significant. (E and F) BLI and survival curves of mice
transplanted with the GS2 astrocytoma line, which has wild-type BRAF and
Importance of BRAFV600Emutation and Ink4a-Arf deletion in de-
| www.pnas.org/cgi/doi/10.1073/pnas.1117255109Huillard et al.
therapy, as well as possible tumor adaptation to therapy. None- Download full-text
theless, our preliminary results are promising for this approach
providing effective treatment for appropriately diagnosed CNS
Materials and Methods
Neural Progenitor and Glioblastoma Cultures. Mouse neurosphere cultures
were established from E14.5 basal progenitors, as previously described (20),
with the modification that cells were grown in the presence of EGF and basic
Japan Health Sciences Foundation Health Science Research Resources Bank;
DBTRG05-MG human glioblastoma cells were obtained from the American
Type Culture Collection. Cells were propagated in Eagle’s MEM supplemented
with 10% FBS and nonessential amino acids. GS2 cells were obtained from
Manfred Westphal, University Hospital Eppendorf, Hamburg, Germany, and
maintained as neurosphere cultures, as previously described (41).
Intracranial Injections of Tumor Cells and Cre-Expressing Adenovirus. Tumor
cells were transplanted into the brains of SCID mice (ICRSC-M; Taconic) as
previously described (20), at the following coordinates, according to Bregma:
1.0 mm (anterior), 2.0 mm (lateral), and 3.0 mm (deep). Alternatively, cells
were transplanted into athymic mice (5-wk-old female, nu/nu genotype,
BALB/c background; Simonsen Laboratories) as previously described (42).
BrafCA/+and BrafCA/+Ink4a-arffl/fl60-d-old mice were injected with 2 μL of
adenovirus (109multiplicity of infection, MOI) in which cre recombinase
expression is regulated by a ubiquitous promoter (Vector Biolabs). The
following coordinates, according to Bregma, were used: 1.0 mm (anterior),
1.0 mm (lateral), and 1.8 mm (deep).
38 cell lines were randomly assignedto vehicle control (DMSO or 50 mM sodium
lactate, pH 4, for PLX or PD0332991, respectively), PLX treatment (PLX4720;
Plexxikon), PD treatment (PD0332991, Pfizer), or a combination of PLX and PD
mice were monitored every day for the development of symptoms related to
tumor growth and twice weekly by BLI. Mice were killed when they exhibited
symptoms indicative of significant impairment to neurological function.
ACKNOWLEDGMENTS. Theauthors thank MaxwellW. TomandSistaSugiarto
for technical assistance, Ron DePinho for providing the Ink4a-Arflox/loxmouse
line, Peter Hirth and Brian West (Plexxikon) for providing PLX4720, and Pfizer
Brain Tumor Foundation, the Pediatric Low-Grade Astrocytoma Foundation,
theRally Foundation for Childhood Cancer Research, the Campini Foundation,
theMcDonnell Foundation,theClinical and Translational Science Institute, the
University of California at San Francisco, the American Association for Cancer
Research/National Brain Tumor Society (R.A.I.), the American Brain Tumor As-
sociation (A.G.), and National Institutes of Health Grants K08NS065268 (to
T.N.), 5K08NS063456 (to J.J.P.), R01CA159467 (to T.W. and C.D.J.), R01CA131261
(to M.M), 1RO1CA164746 (to C.P.), CA097257 (to W.A.W, C.P., and C.D.J.), R01
NS040511 (to D.H.R.), and 2R01NS057727-05 (to D.H.R). D.H.R. is a Howard
Hughes Medical Institute Investigator.
1. Central Brain Tumor Registry of the United States (2004–2007) Primary Brain and
Central Nervous System Tumors Diagnosed in the United States in 2004–2007, Sta-
tistical Report. Available at http://www.cbtrus.org/reports/reports.html. Accessed
September 11, 2011.
2. Phillips HS, et al.; (2006) Molecular subclasses of high-grade glioma predict prognosis,
delineate a pattern of disease progression, and resemble stages in neurogenesis.
Cancer Cell 9:157–173.
3. Verhaak RG, et al.; Cancer Genome Atlas Research Network (2010) Integrated ge-
nomic analysis identifies clinically relevant subtypes of glioblastoma characterized by
abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17:98–110.
4. Brennan C, et al. (2009) Glioblastoma subclasses can be defined by activity among
signal transduction pathways and associated genomic alterations. PLoS ONE 4:e7752.
5. Bax DA, et al. (2010) A distinct spectrum of copy number aberrations in pediatric high-
grade gliomas. Clin Cancer Res 16:3368–3377.
6. Wu G, et al.; St. Jude Children’s Research Hospital–Washington University Pediatric
Cancer Genome Project (2012) Somatic histone H3 alterations in pediatric diffuse
intrinsic pontine gliomas and non-brainstem glioblastomas. Nat Genet 44:251–253.
7. Schwartzentruber J, et al. (2012) Driver mutations in histone H3.3 and chromatin
remodelling genes in paediatric glioblastoma. Nature 482:226–231.
8. Liang ML, et al. (2008) Tyrosine kinase expression in pediatric high grade astrocytoma.
J Neurooncol 87:247–253.
9. Paugh BS, et al. (2010) Integrated molecular genetic profiling of pediatric high-grade
gliomas reveals key differences with the adult disease. J Clin Oncol 28:3061–3068.
10. Pollack IF, et al.; Children’s Oncology Group (2006) Rarity of PTEN deletions and EGFR
amplification in malignant gliomas of childhood: Results from the Children’s Cancer
Group 945 cohort. J Neurosurg 105(5, Suppl):418–424.
11. Schiffman JD, et al. (2010) Oncogenic BRAF mutation with CDKN2A inactivation is
characteristic of a subset of pediatric malignant astrocytomas. Cancer Res 70:512–519.
12. Dias-Santagata D, et al. (2011) BRAF V600E mutations are common in pleomorphic
xanthoastrocytoma: Diagnostic and therapeutic implications. PLoS ONE 6:e17948.
13. Schindler G, et al. (2011) Analysis of BRAF V600E mutation in 1,320 nervous system
tumors reveals high mutation frequencies in pleomorphic xanthoastrocytoma, gan-
glioglioma and extra-cerebellar pilocytic astrocytoma. Acta Neuropathol 121:
14. Dankort D, et al. (2007) A new mouse model to explore the initiation, progression,
and therapy of BRAFV600E-induced lung tumors. Genes Dev 21:379–384.
15. Corcoran RB, et al. (2010) BRAF gene amplification can promote acquired resistance
to MEK inhibitors in cancer cells harboring the BRAF V600E mutation. Sci Signal
16. Dankort D, et al. (2009) Braf(V600E) cooperates with Pten loss to induce metastatic
melanoma. Nat Genet 41:544–552.
17. Charles RP, Iezza G, Amendola E, Dankort D, McMahon M (2011) Mutationally acti-
vated BRAF(V600E) elicits papillary thyroid cancer in the adult mouse. Cancer Res 71:
18. Brenner M, Kisseberth WC, Su Y, Besnard F, Messing A (1994) GFAP promoter directs
astrocyte-specific expression in transgenic mice. J Neurosci 14:1030–1037.
19. Bachoo RM, et al. (2002) Epidermal growth factor receptor and Ink4a/Arf: Convergent
mechanisms governing terminal differentiation and transformation along the neural
stem cell to astrocyte axis. Cancer Cell 1:269–277.
20. Ligon KL, et al. (2007) Olig2-regulated lineage-restricted pathway controls replication
competence in neural stem cells and malignant glioma. Neuron 53:503–517.
21. Otero JJ, Rowitch D, Vandenberg S (2010) OLIG2 is differentially expressed in pedi-
atric astrocytic and in ependymal neoplasms. J Neurooncol 104:423–438.
22. National Institutes of Health Registry of Clinical Trials Website. Available at http://
clinicaltrials.gov/ct2/results?term=pd-0332991. Accessed August 23, 2011.
23. Smalley KS (2010) PLX-4032, a small-molecule B-Raf inhibitor for the potential
treatment of malignant melanoma. Curr Opin Investig Drugs 11:699–706.
24. Davies H, et al. (2002) Mutations of the BRAF gene in human cancer. Nature 417:
25. Michaud K, et al. (2010) Pharmacologic inhibition of cyclin-dependent kinases 4 and 6
arrests the growth of glioblastoma multiforme intracranial xenografts. Cancer Res 70:
26. Jones DT, et al. (2008) Tandem duplication producing a novel oncogenic BRAF fusion
gene defines the majority of pilocytic astrocytomas. Cancer Res 68:8673–8677.
27. Korshunov A, et al. (2009) Combined molecular analysis of BRAF and IDH1 dis-
tinguishes pilocytic astrocytoma from diffuse astrocytoma. Acta Neuropathol 118:
28. Pfister S, et al. (2008) BRAF gene duplication constitutes a mechanism of MAPK
pathway activation in low-grade astrocytomas. J Clin Invest 118:1739–1749.
29. Dhomen N, et al. (2009) Oncogenic Braf induces melanocyte senescence and mela-
noma in mice. Cancer Cell 15:294–303.
30. Garraway LA, et al. (2005) Integrative genomic analyses identify MITF as a lineage
survival oncogene amplified in malignant melanoma. Nature 436:117–122.
31. Gronych J, et al. (2011) An activated mutant BRAF kinase domain is sufficient to in-
duce pilocytic astrocytoma in mice. J Clin Invest 121:1344–1348.
32. Robinson JP, et al. (2010) Activated BRAF induces gliomas in mice when combined
with Ink4a/Arf loss or Akt activation. Oncogene 29:335–344.
33. Michaloglou C, et al. (2005) BRAFE600-associated senescence-like cell cycle arrest of
human naevi. Nature 436:720–724.
34. Raabe EH, et al. (2011) BRAF activation induces transformation and then senescence
in human neural stem cells: A pilocytic astrocytoma model. Clin Cancer Res 17:
35. Zhu J, Woods D, McMahon M, Bishop JM (1998) Senescence of human fibroblasts
induced by oncogenic Raf. Genes Dev 12:2997–3007.
36. Dougherty MJ, et al. (2010) Activating mutations in BRAF characterize a spectrum of
pediatric low-grade gliomas. Neuro-oncol 12:621–630, 10.1093/neuonc/noq007.
37. Forshew T, et al. (2009) Activation of the ERK/MAPK pathway: A signature genetic
defect in posterior fossa pilocytic astrocytomas. J Pathol 218:172–181.
38. Weber RG, et al. (2007) Frequent loss of chromosome 9, homozygous CDKN2A/p14
(ARF)/CDKN2B deletion and low TSC1 mRNA expression in pleomorphic xanthoas-
trocytomas. Oncogene 26:1088–1097.
39. Nicolaides TP, et al. (2011) Targeted therapy for BRAFV600E malignant astrocytoma.
Clin Cancer Res 17:7595–7604.
40. Knobbe CB, Reifenberger J, Reifenberger G (2004) Mutation analysis of the Ras
pathway genes NRAS, HRAS, KRAS and BRAF in glioblastomas. Acta Neuropathol 108:
41. Günther HS, et al. (2008) Glioblastoma-derived stem cell-enriched cultures form dis-
tinct subgroups according to molecular and phenotypic criteria. Oncogene 27:
42. Hashizume R, et al. (2010) Morphologic and molecular characterization of ATRT
xenografts adapted for orthotopic therapeutic testing. Neuro-oncol 12:366–376.
Huillard et al. PNAS
| May 29, 2012
| vol. 109
| no. 22