The Combination of Ionizing Radiation and Peripheral
Vaccination Produces Long-term Survivalof Mice
Bearing Established Invasive GL261Gliomas
ElizabethW. Newcomb,1,5Sandra Demaria,1,5Yevgeniy Lukyanov,1Yongzhao Shao,2,5
Tona Schnee,1Noriko Kawashima,1Li Lan,1J. Keith Dewyngaert,3David Zagzag,1,4,5
William H. McBride,6and Silvia C. Formenti3,5
Purpose: High-grade gliomatreatmentincludes ionizing radiationtherapy.Thehighinvasiveness
of glioma cells precludes their eradication and is responsible for the dismal prognosis. Recently,
we reported the down-regulation of MHC class I (MHC-I) products in invading tumor cells in
human and mouse GL261gliomas. Here, we tested the hypothesis that whole-brainradiotherapy
(WBRT) up-regulates MHC-I expression on GL261tumors and enhances the effectiveness of
Experimental Design: MHC-I molecule expression on GL261cells was analyzed in vitro and
in vivo by flow cytometry and immunohistochemistry, respectively. To test the response of
randomlyassigned to four groups receiving (a) no treatment, (b) WBRT in two fractions of 4 Gy,
(c)vaccinationwithirradiatedGL261cells secretinggranulocyte-macrophage colony-stimulating
factor, or (d) WBRTand vaccination. Endpoints were tumor response and survival.
Results: Anionizing radiationdose of 4 Gy maximally up-regulated MHC-Imolecules on GL261
cells invitro. Invivo,WBRT induced the expression of the h2-microglobulinlight chain subunit of
the MHC class I complex on glioma cells invading normal brain and increased CD4+ and CD8+
Tcell infiltration. However, the survival advantage obtained withWBRTor vaccination alone was
minimal. In contrast,WBRT in combination with vaccinationincreased long-term survival to 40%
to 80%, compared with 0% to 10% in the other groups (P < 0.002). Surviving animals showed
antitumorimmunity by rejecting challenge tumors.
Conclusion: Ionizing radiationcanbe successfullycombined withperipheral vaccinationfor the
treatment of establishedhigh-grade gliomas.
Radiation therapy remains the principal approach for the
treatment of adult patients with high-grade glioma and
pediatric patients with brainstem gliomas, generally involving
fractionated radiotherapy with total doses up to 60 Gy (1–4).
The addition of concurrent temozolomide to radiotherapy has
been shown to significantly prolong the life of patients with
glioblastoma, but long-term survivors of this disease remains
extremely rare (5).
Because of the pressing need for adjunctive therapies to
enhance the effectiveness of radiation therapy for glioma, we
have developed an experimental murine GL261 model that
mimics the aggressive and invasive growth observed in human
brain tumors (6–8) and whose growth can be monitored
noninvasively by CT imaging. Using this model, we showed
significant down-regulation of MHC genes in vivo in tumor cells
invading normal brain tissue in comparison with the glioma
cells in the core of the tumor, similar to what we had observed
in human glioma (9). We hypothesized that one of the
surveillance system is down-regulating expression of MHC
molecules at the growing edge of the tumor. The general hypo-
thesis of tumor immune surveillance was originally formulated
by Burnet in 1970 (10). It is now clear that there is a complex
relationship between the host’s immunosurveillance system and
the ability of transformed cells to escape immune recognition
and destruction (reviewed in refs. 11–14). The ability of tumor
cells to modulate the host’s immune response has important
implications for the development of immunotherapies, includ-
ing the design of effective cancer vaccines (15, 16).
Authors’Affiliations: Departments of1Pathology,2Environmental Medicine, and
3Radiation Oncology,4Division of Neuropathology and Neurosurgery,5NewYork
University Cancer Institute, NewYork University School of Medicine, NewYork,
New York; and6Department of Radiation Oncology, Experimental Division,
Universityof California at Los Angeles Schoolof Medicine, Los Angeles, California
Received 3/10/06; revised 4/21/06; accepted 5/4/06.
Grant support: NIH Research grants NS-57829 (E.W. Newcomb), CA-100426
(D. Zagzag), KO8 CA-89336 and American Cancer Society grant RSG-05-145-
01-LIB (S. Demaria); Department of Defense grant W8IXWH-04-1-0905 and
American Cancer Society grantTURSGCCE103174 (S.C. Formenti).
The costs of publicationof this article were defrayedinpartby thepaymentof page
charges.This article must therefore be hereby marked advertisement in accordance
with18 U.S.C. Section1734 solely toindicate this fact.
Note: E.W. Newcomb and S. Demaria contributedequally to this work.
Requests for reprints: ElizabethW. Newcomb, Department of Pathology, MSB-
531, NewYork University School of Medicine, 550 First Avenue, NewYork, NY
10016. Phone: 212-263-8758; Fax: 212-263-8211; E-mail: newcoe01@med.
F2006 American Association for Cancer Research.
Cancer Therapy: Preclinical
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 20064730
Recently, there has been growing interest in the interactions
between ionizing radiation (IR) and the immune system.
Several studies have shown that irradiation of human and
murine tumor cell lines up-regulates the expression of
numerous immunologically relevant molecules, including Fas,
ICAM-1, MHC class I, and the human carcinoma-associated
antigens, CEA and MUC-1 (17–21), which may make them
better targets for immune recognition. In the present study, we
tested the hypothesis that a low dose of whole-brain
radiotherapy (WBRT) of GL261 intracranial (i.c.) tumors may
up-regulate MHC class I expression, as measured by the
surrogate marker, h2-microglobulin, thus providing a target
for a T cell–mediated antitumor immune response elicited by
vaccination. We show significant increases in the long-term
survival of mice with well-established i.c. GL261 gliomas
treated with WBRT and vaccination, concomitant with the
development of antitumor immunity. The growing awareness
that IR can make tumors more amenable to immune
recognition has important implications for the development
of novel immunotherapies. Importantly, the more insidious
invading tumor cells that invariably lead to fatal relapse may be
particularly targeted as a result of this strategy.
Materials and Methods
GL261 glioma tumor cell line and reagents.
glioma cell line was obtained from the National Cancer Institute-
Frederick Cancer Research Tumor Repository (Frederick, MD). The cells
were transfected with the plasmid encoding the gene for green
fluorescent protein (GFP) according to the manufacturer’s protocol
(Clontech, Palo Alto, CA) and cultured in 5% CO2and 95% humidified
air atmosphere at 37jC in DMEM (Life Technologies, Grand Island,
NY) supplemented with 10% fetal bovine serum (Atlanta Biologicals,
Norcross, GA), 0.25% gentamicin (Life Technologies) and 1%
L-glutamine (Life Technologies) as described (7). The GL261 cell line
retrovirally transduced to secrete granulocyte-macrophage colony-
stimulating factor (GM-CSF) was a kind gift from Dr. X.O. Breakefield
(22) and was cultured as described for the nontransduced GL261 cells.
Both GL261 glioma cell lines were split every 3 days to ensure
Flow cytometric analysis of the GL261 cell line.
plated (5 ? 105/75 cm2flask) and after 48 hours, the flasks were either
irradiated (0, 2, 4, and 6 Gy) using a60Co source (Theratron 780-C;
AECL Medical, Canada) or treated with IFN-g (150 pg/mL). Cells were
harvested 72 hours after irradiation in cold PBS containing 0.04%
EDTA and processed in parallel for MHC expression by fluorescence-
activated cell sorting (FACS) analysis or for mRNA gene expression by
RT-PCR analysis as described below. For FACS analysis, cells were
washed and incubated with phycoerythrin-conjugated anti-H2-Kb
(clone AF6-88.5), phycoerythrin anti-H2-Db(clone KH95), and
phycoerythrin anti-Fas/CD95 or isotype controls (all antibodies from
BD PharMingen, San Jose, CA). Samples were analyzed using a
FACSCalibur flow cytometer (Becton Dickinson, Bedford, MA) and
FlowJo 4.5.2 software (Ashland, OR).
Flow cytometric analysis of tumor-infiltrating cells.
nature and number of immune cells infiltrating GL261 tumors growing
in the brain of mice, on day 23 postimplantation, two brains were
harvested from each treatment group and single-cell homogenates were
prepared for FACS analysis. Each brain was removed and the tumor-
bearing hemispheres were finely minced with a razor blade and then
homogenized in 5 mL of PBS in a Dounce homogenizer (7 mL, Pyrex, no.
7727-07). The homogenate was drawn up into a 12 mL syringe fitted
with a 21-gauge needle and passaged 10 times through the needle. The
final cell suspension was filtered through a 40-Am nylon cell strainer
The murine GL261
GL261 cells were
To determine the
(BD Falcon, San Jose, CA) into a 50 mL tube. The single-cell suspension
was centrifuged at 400 ? g for 10 minutes at room temperature. The
pellet was resuspended in 4 mL of 30% isotonic Percoll and overlaid on a
Percoll gradient and centrifuged at 500 ? g for 20 minutes as described
(23). Lymphocytes were collected from the 37% to 70% interface, washed
once in PBS and counted in a hemacytometer. Aliquots of cells (1 ? 105)
were blocked in 10% normal mouse serum in PBS for 15 minutes at
room temperature followed by incubation with phycoerythrin-conjugat-
ed antimouse monoclonal antibodies to CD4 T cells (clone GK1.5) and
to natural killer cells (clone 01295A), and Cychrome-conjugated
antimouse monoclonal antibody to CD8 T cells (clone 53-6.7) for 30
minutes on ice. All antibodies were purchased from BD PharMingen.
Cells were washed once in PBS, resuspended in 1% paraformaldehyde,
and stored at 4jC until analyzed by FACS analysis using a FACSCalibur
flow cytometer (Becton Dickinson).
To obtain the total number of lymphocytes, we took the total number
of viable cells isolated from the two pooled brains (control, 3.7 ? 105;
WBRT, 8.0 ? 105) and multiplied by the percentage of cells in the
lymphocyte gate (Fig. 3A; control, 44.5% gated represents 1.65 ? 105;
WBRT, 60.3% gated represents 4.82 ? 105). This number of total
lymphocytes was then multiplied by the percentage of cells in the
lymphocyte gate positive for CD4+, CD8+, or NK1.1 to obtain the total
number of lymphocytes in each subset depicted as a bar graph (Fig. 3B).
RNA isolation and reverse transcription-PCR.
from suspension cultures of GL261 cells (see above for FACS analysis)
using the RNeasy kit (Qiagen Inc., Valencia, CA) according to the
manufacturer’s directions. RNA was quantitated by absorbance at 260 nm.
Two micrograms of total RNA was reverse-transcribed using Super
Script II RNase H reverse transcriptase (Invitrogen, Carlsbad, CA) and
random hexamer primers (Invitrogen) at 25jC for 10 minutes and
42jC for 1 hour for cDNA synthesis. Two microliters of the reverse
transcription product was used as a template for PCR amplification.
PCR was done under standard conditions in a 50 AL reaction mix
containing 1? PCR buffer, 1 unit of Platinum Taq polymerase
(Invitrogen), 200 Amol/L of deoxynucleotide triphosphate mix, 1.5
mmol/L of MgCl2, and 150 nmol/L of H-2Kbprimers (5-GGAAAAG-
GAGGGGACTATGC-3, sense; 5-TCCATAAGGCTCACAGGGAAC-3,
antisense), or 25 nmol/L of h-actin primers (5-CACAGGGGAGGT-
GATAGCAT-3, sense; 5-TCAAGTTGGGGGACAAAAG-3, antisense). The
PCR conditions consisted of 3 minutes of an initial denaturation step
(95jC) followed by 25 cycles of denaturation (95jC, 30 seconds),
annealing (55jC, 30 seconds), and extension (72jC, 50 seconds)
followed by a final elongation step of 7 minutes at 72jC. Twenty
microliters of PCR product was analyzed on 3% agarose gels stained
with ethidium bromide. Quantitation of bands was done with the Bio-
Rad Fluor-S apparatus (Bio-Rad, Hercules, CA) with Quantity One
(version 4.2.1) software.
Immunohistochemistry. Single and double label immunohistochem-
istry was done using the NexES automated immunostainer and
detection systems (Ventana Medical Systems, Tucson, AZ). Formalin-
fixed, paraffin-embedded, 6 Am sections were deparaffinized in xylene
(three changes), rehydrated through a graded series of alcohols, and
rinsed in distilled water. All incubations were carried out at 37jC,
unless noted otherwise.
To detect MHC class I expression in mouse tissue, sections were
immunostained with rabbit anti-human/mouse h2-microglobulin
antibody (Novocastra Laboratories, Newcastle upon Tyne, United
Kingdom). This antibody reacts with the light chain component of
MHC class I molecules (9). Antigen retrieval was done by proteolytic
enzyme digestion (alkaline endopeptidase, 0.5 units/mL; Ventana
Medical Systems) for 4 minutes. Endogenous peroxidase was blocked
by the application of hydrogen peroxide for 4 minutes at 37jC. Anti-
h2-microglobulin was diluted 1:1,000 and h2-microglobulin antibody
was detected by the application of a secondary biotinylated goat
anti-mouse (Ventana Medical Systems) for 8 minutes, followed by
the application of streptavidin-horseradish peroxidase for 8 minutes.
The chromogen, 3,3¶-diaminobenzidine/hydrogen peroxide mix was
Total RNA was isolated
Combination of Radiation andVaccination for Glioma
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 20064731
applied for 8 minutes and then enhanced with copper sulfate for
4 minutes. Slides were then counterstained in hematoxylin, dehydrated,
and mounted with Permount.
GL261 glioma cells expressing GFP were used to detect tumor cells
invading the brain adjacent to the tumor. Tissue sections were
immunostained with mouse anti-GFP (clone GFP01; Neomarkers,
Fremont, CA). Antigen retrieval was done by microwaving tissue
sections in 0.01 mol/L (pH 6.0) citrate buffer for 20 minutes in a 1,200
W microwave oven. Slides were allowed to cool to room temperature,
then washed in 0.05 mol/L Tris-HCl (pH 7.6) containing 0.3 mol/L
NaCl. Endogenous peroxidase was blocked by the application of hydro-
gen peroxide for 4 minutes at 37jC. For double label immunohisto-
chemistry, tumor sections were first stained with h2-microglobulin.
Heat-induced antigen retrieval was done as described above. Next,
slides were incubated with anti-GFP antibody diluted 1:300 and
incubated overnight at room temperature. GFP antibody was detected
as described above using alkaline phosphatase fast red as chromogen.
To evaluate lymphocyte populations in brain tissue, tissue sections
were immunostained with mouse anti-human CD3 (clone PS2,
Novocastra Laboratories). Antigen retrieval was done by microwaving
tissue sections in 0.01 mol/L (pH 6.0) citrate buffer for 10 minutes as
described above. Endogenous peroxidase was blocked by the
application of hydrogen peroxide for 4 minutes at 37jC. Anti-CD3
was diluted 1:200, incubated for 30 minutes at room temperature, and
then anti-CD3 antibody was detected by the application of a secondary
biotinylated goat anti-mouse (Ventana Medical Systems) for 8 minutes,
followed by the application of streptavidin-horseradish peroxidase for
8 minutes. The chromogen, 3,3¶-diaminobenzidine/hydrogen peroxide
mix was applied for 3 minutes.
Animals and i.c. tumor implantation.
obtained from Taconic (Germantown, NY) and maintained under
aseptic conditions in microisolator cages in facilities approved by the
American Association for Accreditation of Laboratory Animal Care in
accordance with the current regulations and standards of the U.S.
Department of Agriculture, The Department of Health and Human
Services, and the NIH. Animal studies were done under an approved
protocol by the Institutional Animal Care and Use Committee. The
animals were allowed food and water ad libitum. GL261 cells were
cultured to subconfluence, trypsinized, washed twice in DMEM without
serum, and resuspended in DMEM for inoculation into animals.
To establish i.c. tumors, GL261 glioma cells were implanted in the
brains of 10- to 12-week-old female C57BL/6 mice (20 g) as described
(24). Briefly, animals were anesthetized with an i.p. injection of
xylazine (80 mg/kg) and ketamine (10 mg/kg) and a burr hole was
drilled in the skull 0.1 mm posterior to the bregma and 2.3 mm lateral
to the midline. GL261 cells (5 ? 107/mL) in 2 AL of medium were
inoculated stereotactically using a head frame (David Kopf Instru-
ments, Tujunga, CA) in the defined location of the caudate/putamen
(0.1 mm posterior to the bregma, 2.3 mm lateral to the midline) using
a 10 AL Hamilton syringe (Reno, NV) with a 1-in. 30-gauge needle
attached and inserted into a Kopf microinjection unit (Model 5000
with Model 5001 Hamilton syringe holder). The needle was advanced
to a depth of 2.3 mm from the brain surface and the cell suspension
was delivered slowly over the course of 3 to 4 minutes. Following
injection, the needle was left in place for 2 minutes, after which time, it
was raised to a depth of 1.5 mm below the brain surface and left in
place for an additional 1 minute. Upon withdrawal of the needle, the
burr hole was immediately sealed with bone wax and the incision
sutured. Animals were randomly assigned into control and treatment
groups (n = 5/group). Animals were observed weekly and when they
showed signs of neurologic deficit (lethargy, failure to ambulate, or
lack of feeding resulting in loss of >20% body weight) they were
sacrificed. Prior to sacrifice, the animals were anesthetized and then
perfused transcardially with PBS followed by 4% paraformaldehyde.
Brains were removed and placed in cold 4% paraformaldehyde
overnight, then sliced into 2 mm coronal sections prior to processing
and embedding in paraffin. Tumors were measured grossly in three
Female C57BL/6 mice were
orthogonal axes to determine tumor volumes. H&E-stained coronal
sections were used as a check of tumor volumes and for overall tumor
CT imaging.Animals were anesthetized with xylazine and ket-
amine, as described above, prior to imaging the brain using a CT
scanner (GE LightSpeed Qx/i). Animals received an intraorbital
injection of contrast agent (100 AL; Conray, Mallinckrodt, Inc., St.
Louis, MO) and within 10 minutes of contrast injection, axial images of
the brain were acquired using the CT scanner to detect the presence of
brain tumors. Contrasted tumors appear as regions of brighter intensity
(similar to that of the bone tissue, in the animal skull) in the CT scan
compared with surrounding brain tissue which appears dark due to
Preparation of the GL261 vaccine.
CSF into the medium, cells (5 ? 105) were plated in a 60 mm dish in
2.5 mL of medium. After 24 hours, the supernatant was harvested and
either stored at ?80jC or tested directly by ELISA for GM-CSF using a
commercial ELISA kit (R&D System, Minneapolis, MN) according to the
manufacturer’s instructions. The GM-CSF transduced cells that we
obtained secreted 100 ng/106cells/24 hours GM-CSF, similar to the
original report (22). For vaccine preparation, GL261 cells were irradiated
with 25 Gy in 75 cm2flasks, using a60Co source (Theratron 780-C, AECL
Medical), trypsinized, washed, and inoculated s.c. into the left and right
flanks of mice (2.5 ? 106/100 AL/injection in the inguinal regions).
WBRT combined with GL261 peripheral vaccine treatment.
first experiment, on day 12 post-i.c. implantation, animals were
randomly assigned to four treatment groups (n = 5/group): (a) control
with sham radiation, (b) WBRT given in two fractions of 4 Gy, 48 hours
apart on days 12 and 14, (c) vaccination treatment given on both flanks
with irradiated (25 Gy) GL261 tumor cells secreting GM-CSF on day 14
after i.c. tumor cell implantation, (d) WBRT + vaccination treatment,
where peripheral vaccination was given a few hours after the animals
had received their second fraction of WBRT on day 14. WBRT was
delivered to the head of the mouse centered in a 5 cm radiation field
using a60Co radiation source (Theratron 780-C, AECL Medical) on days
12 and 14 postimplantation.
In the second experiment, CT on day 15 post-i.c. implantation, when
tumors were large enough to be reliably measured, was used to
compute tumor volumes. Images were digitally transferred to a Varian
Eclipse radiation planning system workstation and tumor volumes were
calculated using the formula for ellipsoid volume. Two dimensions (in
frontal plane) were measured directly and the third dimension was
measured using three-dimensional reconstructing feature of the
planning system. Animals were assigned to four treatment groups
(n = 5/group) so that the mean tumor volumes were 3.5 mm3(range,
2.1-4.8) in all groups. The groups consisted of (a) control with sham
irradiation, (b) WBRT, given in two fractions of 4 Gy, 48 hours apart on
days 15 and 17, (c) vaccination as described above on day 17, and (d)
WBRT + vaccination, where peripheral vaccination was given a few
hours after the animals had received their second fraction of WBRT on
day 17. On day 28, tumor volume was assessed by CT. On day 60,
control ‘‘naBve’’ 6- to 8-week-old female C57BL/6 mice or surviving
animals were challenged by s.c. injection in the hind limb of 2 ? 106
cells in 0.05 mL of serum-free media. We chose to evaluate antitumor
immunity in the long-term survivors using s.c. rather than i.c. tumor
rechallenge in order to assess their systemic state of immunity. Tumor
growth was measured twice weekly using calipers and tumor volumes
were calculated using the formula (length ? width2) / 2, where the
length represents the longest axis and the width was measured at right
angles to the length. Animals were observed for an additional 60 days
and/or sacrificed when tumor volumes reached 900 to 1,200 mm3.
Brains from ‘‘cured’’ animals were removed and processed for
Statistics. Kaplan-Meier survival curves with a log-rank test were
used to compare the survival proportions of the four treatment groups
(control, WBRT, vaccination, and WBRT + vaccination). Differences
were considered significant at P < 0.05.
To analyze the secretion of GM-
Cancer Therapy: Preclinical
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 2006 4732
IR up-regulates MHC class I expression on GL261 glioma cells
in vitro. We recently reported that invading GL261 glioma
tumor cells in the brain down-regulate MHC class I expression,
a molecule required for recognition by CTLs (9). We inves-
tigated whether irradiation could reverse this low expression,
as has been reported for other cell lines in vitro (17), GL261
cells were irradiated with different doses of IR (0, 2, 4, or
6 Gy) or treated with IFN-g (150 pg/mL) as a positive control
Irradiation of GL261 cells increased surface MHC class I
H-2Kbexpression 72 hours after irradiation (Fig. 1A). Treatment
with 4 or 6 Gy induced similar increases in H-2Kbexpression,
whereas 2 Gy was less effective (data not shown). Expression of
H-2Dbon GL261 was undetectable on untreated cells and was
not induced by radiation. Similarly, there was no induction
of Fas/CD95 molecules after irradiation (data not shown). As
expected, IFN-g markedly increased the surface expression of
both H-2Kband H-2DbMHC class I molecules. Up-regulation of
the H-2KbMHC protein complex at the cell surface by IR or
IFN-g treatment was accompanied by increased levels of H-2Kb
mRNA gene expression as assessed by RT-PCR (Fig. 1B).
IR up-regulates b2-microglobulin expression on GL261 glioma
cells in vivo. GL261-GFP tumor cells were used to identify
glioma cells in the brain and to test whether irradiation would
increase the expression of MHC in vivo. WBRT was delivered to
the brains of mice on days 12 and 14 after implantation, when
tumors were f4 F 1 mm3in volume (24). On the basis of our
in vitro MHC expression data, we selected two 4-Gy doses
48 hours apart for the in vivo treatments. Brains were harvested
48 hours following the second fraction of WBRT and processed
for immunohistochemistry as described previously (9). We
used double-labeled immunohistochemistry to visualize GFP-
positive tumor cells invading the brain adjacent to tumor
(Fig. 2A) and expression of the light chain of the h2-
microglobulin complex to assess MHC class I expression.
Brains from one representative control and one WBRT animal
of a group of four animals are shown (Fig. 2). In untreated
mice, there was no detectable expression of h2-microglobulin
by the invading tumor cells (Fig. 2B). In contrast, following
WBRT, there was strong surface staining of the invading tumor
cells, consistent with up-regulation of MHC class I complexes
IR increases lymphocytes infiltrating GL261 tumors. A possi-
ble consequence of the administration of IR is the induction of
an inflammatory response (27–30). To test this, mice were
randomly assigned to receive mock treatment or WBRT given
in two doses of 4 Gy on days 15 and 17 post-i.c. tumor
implantation. Brains were harvested 7 days later and single-cell
suspensions were prepared and partially purified on Percoll
gradients for the analysis of lymphocyte subsets by flow
cytometry (Fig. 3). Representative data from one of two groups
of control or WBRT animals are shown. The contamination
with GFP-positive tumor cells in the lymphocyte gate was <1%
in all samples (data not shown). The numbers of CD4+, CD8+,
and natural killer cells recovered from brain tumors of control
and irradiated mice are shown in the bar graph (Fig. 3B). There
was a 4-fold increase in the number of CD4+ and a 4.5-fold
increase in CD8+ T cells in the brain tumors of irradiated mice,
whereas natural killer cells showed a small increase compared
with brain tumors of control mice.
Fig.1. IRinduces up-regulation of MHC expressionin GL261glioma cells in vitro
as shown by (A) FACS analysis and (B) RT-PCR. Using FACS analysis, GL261
cells treated with 4 Gy or IFN-g show up-regulation of MHC expression (black
histogram) compared withuntreated cells (grayhistogram).The up-regulation of
cell surface antigen expression for MHC was correlated withincreased gene
Fig. 2. IRinduces up-regulation of h2-microglobulin expressionin GL261glioma
cells in vivo. A, invasive growth pattern of GL261tumors.Tissue sections were
stained for GFP to reveal GL261tumor cells (red label; magnification,?100).
C, double labelimmunohistochemistry analysis of MHC expression (brown label)
on GFP-positive GL261tumor cells (red label). B, representative brains from a
control animal or 48 hours afterWBRT treatment with two fractions of 4 Gy are
shown (magnification,?100). Insets, one representative GFP-positive GL261tumor
cellinvading thebrainadjacent to tumor (BAT) incontrol (redlabel only) compared
with one representative GFP-positive GL261tumor cellinvading the BAT from a
WBRTanimal showing coexpression of GFPand h2-microglobulin (double red and
brown label; magnification,?400).
Combination of Radiation andVaccination for Glioma
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 2006 4733
Increased tumor-free survival of animals bearing i.c. GL261
tumors treated with WBRT combined with peripheral vaccina-
tion. Vaccination with irradiated autologous tumor cells
transduced to secrete GM-CSF has been shown to induce a T
cell–mediated antitumor response against GL261 glioma (31).
The elicited immune response, which is mediated, at least in part,
by CD8+ T cells can inhibit the growth of early brain tumors
(31). To test whether WBRT can increase the susceptibility of
established invasive GL261 glioma to the antitumor immune
response elicited by vaccination, we used the treatment protocol
shown in the schema (Fig. 4A). At day 12 after i.c. implantation,
animals were randomly assigned to four treatment groups: (a)
control, (b) WBRT given in two fractions of 4 Gy, (c) vaccination
with irradiated GL261 cells transduced to secrete the cytokine
GM-CSF, and (d) WBRT + vaccination. The mean survival time
was 33 F 7 days for control animals, 55 F 24 days for animals
treated with fractionated WBRT alone, and 45 F 21 days for
animals treated with vaccination only (Fig. 4B). In the group
treated with WBRT + vaccination, 80% of the animals had long-
term survival times of >75 days compared with only 10% in the
other treatment groups. This difference in overall survival was
statistically significant (P < 0.002).
To test whether the long-term survivors had developed
effective systemic antitumor immunity, all surviving animals
were rechallenged with viable GL261 cells in the hind limb and
observed for tumor development for an additional 8 weeks
(Fig. 4C). All five of the naBve animals developed tumors by day
28. In contrast, none of the long-term survivors from the
different treatment groups developed tumors, indicating that
these animals had developed immunologic memory capable of
rejecting GL261 tumor.
Radioresponse of established i.c. GL261 tumors treated with
WBRT combined with peripheral vaccination. In a subsequent
experiment, CT scanning was used to image the i.c. tumors and
follow the kinetics of tumor response to treatment. CT scanning
was first done on day 15 after implantation, at which time,
tumors were usually large enough to be reliably visualized.
Animals were then assigned to four treatment groups (control,
WBRT, vaccination, or WBRT + vaccination), such that the
tumor volumes were comparable between groups. Treatment
was administered as described above (Fig. 4A) except that
WBRT was given on days 15 and 17, and vaccination on day 17.
Representative CT scans from animals prior to treatment on day
15 and after treatment on day 28 are shown (Fig. 5A, arrows
indicate tumors) together with the Kaplan-Meier survival curves
(Fig. 5B). The mean survival was 33 F 3 days for control
animals, 37 F 7 days for animals treated with fractionated
WBRT alone, and 34 F 6 days for animals treated with
vaccination only. The survival advantage observed following
WBRT or vaccination alone was less than in the previous
experiment (Fig. 4B), suggesting that a delay of only a few days
in treatment and correspondingly larger tumors dramatically
reduces the effectiveness of vaccination and radiation as single
modalities. Importantly, although the group treated with
WBRT + vaccination had an overall reduced survival advantage
as compared with the previous experiment (Fig. 4B), 40% of
Fig. 3. IRincreases tumor-infiltrating lymphocytes into GL261tumors. GL261cells
were implantedi.c. on day 0, and mice were either left untreated (Control) or given
WBRTof 4 Gy on days15 and17 (WBRT). On day 23, brains were harvested and
the lymphocyte-enriched fractions were collected as describedin Materials and
Methods and stained with phycoerythrin-Cy5 anti-CD8, and phycoerythrin
anti-CD4 oranti-NK1.1monoclonalantibodies, followedby flowcytometry analysis.
A, the lymphocyte gate was set based on the scatter plots of spleen cells.
Representative data from one of two groups of mice. B, the totalnumber of CD4,
CD8, andnaturalkillercells were calculatedbasedonthetotalviable cells recovered
from each group, and multipliedby the percentage of gated lymphocytes as
described in Materials and Methods; columns, number of cells in each subset.
Cancer Therapy: Preclinical
www.aacrjournals.org Clin Cancer Res 2006;12(15) August1, 20064734
the animals experienced long-term survival as compared
with none of the animals in the other treatment groups.
This difference in overall survival was statistically significant
(P < 0.002). Again, the long-term survivors failed to develop
tumors upon rechallenge with GL261 glioma cells in the hind
leg after 60 days compared with the controls.
Our previous work has shown that the murine GL261 cell
line can be used to develop a valid transplantable orthotopic
model of human glioma. Traditionally, mouse models of glioma
have primarily used xenografts of human glioma cell lines
implanted i.c. into immunoincompetent mice (32). As a result,
they have been unable to test immunotherapeutic strategies (33).
Furthermore, such tumors do not show the invasive phenotype
characteristic of human gliomas that is often associated with
treatment failure (33). In contrast to the xenograft models,
GL261 tumors show extensive invasion into the brain parenchy-
ma adjacent to the tumor with visible single or multiple tumor
cells advancing into the brain along vascular channels (7, 9). The
derivative of GL261 expressing GFP was used to readily identify
invading tumor cells (7). Although GFP can be immunogenic,
our data show that it does not change the tumorigenicity of this
poorly immunogenic glioma (7). Moreover, using neuroimaging
techniques, including perfusion MRI, we have documented a
correlation between tumor growth, relative cerebral blood
volume, microvascular density, and histologic features associated
with the highly angiogenic nature of the GL261 tumors (34).
Contrary to the earlier belief that the brain was refractory to
infiltration with immune cells, there is now ample evidence
that immune cells can traffic to the brain and affect the growth
of experimental gliomas (35–37). More than a decade ago, it
was shown that peripheral vaccination with cytokine-secreting
autologous tumor cells could induce an immune response
against the intracerebral tumor in experimental animals (38).
In the current study, we used GL261 glioma cells transduced
to secrete the cytokine GM-CSF as a vaccine. Vaccination with
tumor cells modified to secrete GM-CSF has been shown to be
effective in several preclinical cancer models, as well as in
clinical trials in generating a systemic and durable antitumor
Fig. 4. Increased tumor-free survival of animals bearing i.c. GL261tumors treated
with whole brain fractionated radiation combined with peripheral vaccination.
A, schema outlining the treatments usedin the experiments. B, Kaplan-Meier
survival curves of animals in the different treatment groups. C, animals‘‘cured’’ by
the different treatments were rechallenged on day 75 of i.c. implantation with
s.c. injection of 2?106GL261cells in the hind limb. Normal growth of GL261
tumors was monitored by injection of cells into naI«ve mice.The overall survival
of naI«ve and rechallenged animals was followed for 8 weeks.
Fig. 5. Radioresponse of establishedi.c. GL261tumors treated with whole brain
fractionated radiation combined with peripheral vaccination. A, CTscans of
representative animals at day15 prior toinitiationof treatments andagainonday28
of the same animals showing their response to a given treatment. Arrows,
contrasted tumors. B, Kaplan-Meier survival curves of animals in the different
Combination of Radiation andVaccination for Glioma
www.aacrjournals.org Clin Cancer Res 2006;12(15) August1, 2006 4735
immune response (reviewed in ref. 39). In an experimental
setting of well-established, radiologically detectable i.c. tumors,
the combination of the vaccine with WBRT was superior to
either treatment alone. Similar data were obtained by Lum-
niczky et al. (31). However, in their study, treatment was given
3 days following i.c. implantation of GL261, at a time when the
tumor was unlikely to be well established. To better mimic the
clinical setting, we elected to study GL261 glioma tumors when
they were firmly established, as documented by CT imaging.
Time after implantation and tumor size are critical factors
affecting the therapeutic response to vaccination, as shown by
the fact that 40%, 10%, and 0% of mice treated with
vaccination alone achieve long-term survival when vaccine is
given on days 3, 14, and 17, respectively (ref. 31; Figs. 4 and 5).
Although we do not know whether delayed vaccination elicits
an antitumor immune response of the same magnitude as
vaccination given earlier in the course of tumor growth, the
response is sufficient to obtain a therapeutic effect when
combined with WBRT. GL261 gliomas are relatively radio-
resistant tumors (TCD50= 65 Gy; ref. 8). Consequently, the
modest dose of radiation used in our experiments is unlikely to
have resulted in sufficient cell death to explain the clinical
response observed (Fig. 5B), suggesting that its major effect may
indeed have been to facilitate infiltration and/or recognition of
tumor cells by antitumor T cells. Our studies identify one
possible mechanism whereby this could occur in that radiation
treatment up-regulated the expression of MHC class I H2-Kb
and h2-microglobulin molecules on GL261 glioma cells in vitro
and in vivo, respectively. In vitro, treatment with 4 Gy was
sufficient to up-regulate the expression of H2-Kbbut not H2-Db
molecules on GL261 cells. Differential regulation of H2-K and
H2-D locus products by radiation has been previously reported
in the B16 melanoma model (40). Other studies using
sublethal doses of radiation on a wide variety of human cancer
cell lines also showed up-regulation of one or more surface
molecules involved in T cell–mediated immune responses,
including MHC class I, Fas, ICAM-1, CEA, and MUC-1 (17–21).
Increase of MHC class I expression on GL261 cells has been
shown to confer greater sensitivity to tumor-specific CTLs
in vitro, as shown by using T cells specific for the melanoma-
associated antigen gp100 expressed by GL261 cells (41).
Significantly, irradiation induced the expression of h2-
microglobulin on invading GL261 glioma cells in vivo
(Fig. 2C), a surrogate marker used in this study for MHC class
I expression. The finding of very low MHC class I expression on
glioma cells invading the normal brain tissue (9) of nonirra-
diated mice could, at least in part, be responsible for the
resistance of established tumors to antitumor T cells elicited by
vaccination (Fig. 5B). Therefore, it is possible that the
significant tumor response, as determined by CT imaging and
increased survival, among the mice treated with WBRT and
vaccination is due to the increased susceptibility of irradiated
GL261 to antitumor T cells (Fig. 5B). Nevertheless, other
radiation-induced changes in brain tumor microenvironment
are likely to contribute to the therapeutic benefit of the
combination treatment. For instance, the observed influx of
inflammatory cells, including CD4+ and CD8+ T cells in
response to local radiation (Fig. 3) may be triggered by local
production of cytokines and other inflammatory signals in
addition to the up-regulation of vascular adhesion molecules
(reviewed in ref. 30). Moreover, we cannot exclude the
possibility that the effects of radiation not only on the tumor
but also on normal brain tissue, which was included in the
radiation field, contribute to the observed results. MHC class I
antigens are usually not expressed in the central nervous system,
but they might be induced by radiation. However, by immuno-
histochemistry, we did not detect the expression of h2-micro-
globulin in the normal brain adjacent to the irradiated tumor,
suggesting that an autoimmune response is unlikely to be
generated and a true therapeutic benefit might result from the
combination of radiation treatment with a vaccination strategy.
Immunotherapy of brain tumors is rapidly emerging as a
viable clinical option (42), but its use as a first-line approach
remains controversial (43). At least two ‘‘proof of principle’’
studies have been reported using vaccines prepared from the
patients’ autologous irradiated glioma tumor cells admixed
with GM-CSF or IL-4-secreting fibroblasts (44, 45). In a phase I
clinical trial using tumor vaccination plus adoptive T cell
immunotherapy for patients with newly diagnosed malignant
gliomas, several objective clinical responses were observed
without any long-term toxicities (44). More recently, several
groups have developed dendritic cell vaccines, using either acid-
eluted peptides from glioblastoma multiforme cell cultures
developed from the surgical specimen or from crude tumor
lysates (45–48). In several patients, a systemic CTL response
was detected, and in some cases, tumor-infiltrating lympho-
cytes were also observed in the resection specimen obtained at
relapse. Despite the occurrence of sporadic objective responses
and some modest increase in the patient’s survival, no long-
term survivors have emerged from these approaches (45–48).
Presumably, established invasive gliomas escape immune
recognition through multiple mechanisms. One of them is low
expression of MHC class I molecules, as we have previously
reported in all grade 4 human gliomas (9). Findings from
studies in other tumor types also emphasize the importance
of MHC class I expression by the tumor cells in eliciting an
effective immune response. A recent study of 124 stage IV
melanoma patients shows that infiltration of tumors by CD4+
and CD8+ T cells correlated with the expression of MHC class I
by the tumor cells and was associated with a better prognosis
(49). Similar findings were reported in another study of head
and neck tumor patients, in which a positive correlation was
found between MHC class I expression, the presence of tumor-
infiltrating lymphocytes, and a better prognosis (50). These
data support the hypothesis that a strategy to enhance the
expression of MHC class I molecules on tumor cells may
increase the efficacy of cancer vaccines.
In summary, our results show that IR increases the expression
of MHC class I on GL261 glioma cells in vitro and increases the
expression of h2-microglobulin in vivo. When radiotherapy is
combined with peripheral vaccination, it achieves significant
long-term survival rates in animals bearing established,
measurable i.c. tumors with an invasive phenotype. These
results also show that the GL261 glioma model is a relevant
preclinical model for further exploring the mechanisms
associated with radiation-induced antitumor immunity. The
preclinical evidence presented offers promise to a clinical
translation. The fact that modest doses of radiation were
required to recover MHC class I expression supports the safety
of combining radiotherapy in trials of immunotherapy for
patients with recurrent high-grade gliomas who are likely to
have already received first-line radiotherapy.
Cancer Therapy: Preclinical
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 2006 4736
Combination of Radiation andVaccination for Glioma
www.aacrjournals.orgClin Cancer Res 2006;12(15) August1, 2006 4737
of postoperative radiotherapy and combined postop-
ciplinary management of malignant gliomas. A Joint
RadiationTherapy Oncology Group and Eastern Co-
operative Oncology Group Study. Cancer 1983;52:
2. Laperriere N, Zuraw L, Cairncross G. Radiotherapy
tematic review. Radiother Oncol 2002;64:259^73.
3. Jallo GI, Biser-Rohrbaugh A, Freed D. Brainstem gli-
omas. Childs Nerv Syst 2004;20:143^53.
4. Broniscer A, GajjarA. Supratentorial high-grade as-
trocytomas and diffuse brainstem glioma: two chal-
lenges for the pediatric oncologist. Oncologist 2004;
tumorandradiotherapygroups: National CancerInsti-
tute of Canada ClinicalTrials Group. Radiotherapy plus
concomitant and adjuvant temozolomide for glioblas-
6. Zagzag D, Armirnovin R, Greco MA, et al.Vascular
larization: a novel concept for glioma growth and an-
giogenesis. Lab Invest 2000;80:837^49.
cent protein immunohistochemistry as a novel experi-
mental tool for the detection of glioma cell invasion
in vivo. Brain Pathol 2003;13:34^7.
8. Newcomb EW, Lymberis SC, LukyanovY, et al.Radi-
ation sensitivity of GL261murine glioma model and
enhanced radiation response by flavopiridol. Cell
9. ZagzagD, Salnikow K, Chiriboga L, et al.Downregu-
lation of major histocompatibility complex antigens in
10. Burnet FM. The concept of immunological surveil-
lance. Prog ExpTumor Res1970;13:1^27.
11. Schreiber H,WuTH, NachmanJ, et al.Immunodomi-
nanceandtumorescape. Semin Cancer Biol 2002;12:
12. CabreraT, Lopez-Nevot MA, GaforioJJ, et al. Anal-
ysis of HLA expression in human tumor tissues. Can-
cer Immunol Immunother 2003;52:1^9.
13. Bubenik J. Tumour MHC class I down-regulation:
tumour escape from immune surveillance? IntJOncol
14. Dunn GP, OldLJ, SchreiberRD.Theimmunobiology
of cancer immunosurveillance and immunoediting.
15. Emens LA, Jaffee EM. Leveraging the activity of
tumor vaccines with cytotoxic chemotherapy. Cancer
16. Pure E, Allison P, Schreiber RD. Breaking down the
17. Garnett CT, Palena C, Chakraborty M, et al. Suble-
thal irradiation of human tumor cells modulates phe-
notype resulting in enhanced killing by cytotoxicT
lymphocytes. Cancer Res 2004;64:7985^94.
18. Klein B, Loven D, Lurie H, et al.The effect ofirradia-
tion on expression of HLA class I antigens in human
brain tumors in culture. JNeurosurg1991 ;80:1074^7.
19. Santin AD, HiserodtJC, FruehaufJ, et al. Effects of
irradiation on the expression of surface antigens in
ovarian cancer. Gynecol Oncol1996;60:468^74.
20. OgawaY, Nishioka A, Hamada N, et al. Expression
of fas (CD95/APO-1) antigen induced by radiation
therapy for diffuse B-cell lymphoma: immunohisto-
chemical study. Clin Cancer Res1997;3:2211^6.
21. Olschowka JA, Kyrkanides S, Harvey BK, et al.
ICAM-1inductionin the mouse CNS followingirradia-
tion. Brain Behav Immun1997;11:273^85.
22. Herrlinger U, Kramm CM, Johnston KM, et al. Vac-
cination for experimental gliomas using GM-CSF-
transduced glioma cells. Cancer GeneTher 1997;4:
23. Campanella M, Sciorati C,Tarozzo G, et al. Flow
cytometric analysis of inflammatory cells in ischemic
rat brain. Stroke 2002;33:586^92.
24. Newcomb EW,Tamasdan C, Entzminger Y, et al.
Flavopiridol inhibits the growth of GL261gliomas
invivo: implications for malignant glioma therapy.Cell
25.Williams A, Au Peh C, ElliottT. The cell biology of
MHC class I antigen presentation. Tissue Antigens
26. Insug O, Ku G, Ertl HCJ, et al. A dendritic vaccine
induces protective immunity to intracranial growth of
glioma. Anticancer Res 2002;22:613^22.
27. Demaria S, Kawashima N,Yang AM, et al. Immune-
mediatedinhibition of metastases following treatment
withlocal radiationand CTLA-4-blockade in a mouse
model of breast cancer. Clin Cancer Res 2005;85:
28. Friedman EJ. Immune modulationby ionizingradia-
tion and its implications for cancer immunotherapy.
Curr Pharm Des 2002;8:1765^80.
29. McBrideWH, Chiang C-S, OlsonJL, et al. A sense
of danger from radiation. Radiat Res 2004;162:1^19.
30. Demaria S, Bhardwaj N, McBrideWH, et al. Com-
bining radiotherapy with immunotherapy: a revived
partnership. Int J Radiat Oncol Biol Phys 2005;63:
31. Lumniczky K, Desaknai S, Mangel L, et al. Local
tumor irradiation augments the antitumor effect of
cytokine-producing autologous cancer cell vaccines
in a murine glioma model. Cancer GeneTher 2002;9:
32. Hu B, Guo P, Fang Q, et al. Angiopoietin-2 induces
human glioma invasion through the activation of
matrix metalloprotease-2. Proc Natl Acad Sci U S A
33.WeissWA, Israel M, Cobbs C, etal. Neuropathology
of genetically engineered mice: consensus report and
recommendations from an international forum. Onco-
34. Cha S, Johnson G,Wadghiri YZ, et al. Dynamic,
contrast-enhanced perfusion MRI in mouse gliomas:
correlation with histopathology. Magn Reson Med
tumor cells stimulates CD8+ cell-mediated immunity
against tumors located in the ‘‘immunologically
36. HickeyW.Basicprinciples ofimmunologicalsurveil-
lance of thenormalcentralnervous system. Glia 2001 ;
37. Prins RM, Liau LL. Cellular immunity and immu-
notherapy of brain tumors. Front Biosci 2004;9:
38. Dranoff G, Jaffee E, LazenbyA, et al.Vaccination
with irradiated tumor cells engineered to secrete mu-
rine granulocyte-macrophage colony-stimulating
factor stimulates potent, specific, and long-lasting
anti-tumor immunity. Proc Natl Acad Sci US A1993;
39. Dranoff G. GM-CSF-based cancer vaccines.Immu-
nol Rev 2002;188:147^54.
40. Hauser SH, Calorini L,Wazer DE, et al. Radiation-
enhanced expression of major histocompatibility
complexclass Iantigen H-2Dbin B16 melanomacells.
Cancer Res1991 ;53:1952^5.
41. Prins RM, Odesa SK, Liau LM. Immunotherapeu-
tic targeting of shared melanoma-associated anti-
gens in a murine glioma model. Cancer Res 2003;
42. Steiner HH, Bonsanto MM, Beckhove P, et al. Anti-
tumor vaccinationof patients with glioblastomamulti-
forme: a pilot study to assess feasibility, safety, and
clinical benefit. JClin Oncol 2004;22:4272^81.
43. Fine HA. Toward a glioblastoma vaccine: promise
and potentialpitfalls. JClin Oncol 2004;22:4240^3.
44. Plautz GE, Miller DW, Barnett GH, et al.Tcell adop-
tive immunotherapy of newly diagnosed gliomas. Clin
Cancer Res 2000;6:2209^18.
45. Okada H, Lieberman FS, Edington HD, et al. Autol-
ogous glioma cell vaccine admixed with interleukin-4
gene transfected fibroblasts in the treatment of recur-
rent glioblastoma: preliminary observations in a
patient with a favorable response to therapy. JNeuro-
46. Liau LM, Prins RM, Kiertscher SM, et al. Dendritic
cell vaccination in glioblastoma patients induces sys-
temic and intracranial T-cell responses modulated by
the local central nervous system tumor microenviron-
ment. Clin Cancer Res 2005;11:5515^25.
47. Rutowski S, DeVleeschouwer S, Kaempgen E, et al.
cination for patients with relapsed malignant glioma, a
feasibility study. BrJCancer 2004;91:1656^62.
48. Yu JS, Liu G,Ying H, et al.Vaccination with tumor
lysate-pulsed dendritic cells elicits antigen-specific,
cytotoxicT-cells in patients with malignant glioma.
Cancer Res 2004;64:4973^9.
49. Al-Batran S-E, Rafiyan M-R, Atmaca A, et al. Intra-
tumoralT-cellinfiltrates and MHC class I expressionin
patients with stage IV melanoma. Cancer Res 2005;
50. HaldJ, Rasmussen N, Claesson MH. Invivo infiltra-
tion of mononuclear cells in squamous cell carcinoma
of the head and neck correlates with the ability to
expand tumour-infiltratingTcells in vitro and with the
expression of MHC class I antigens on tumour cells.
Cancer Immunol Immunother1994;39:383^90.