Pediatr Blood Cancer
Anti-CD3 T Anti-GD2 Bispecific Antibody Redirects T-Cell Cytolytic
Activity to Neuroblastoma Targets
Maxim Yankelevich, MD,1,2** Sri Vidya Kondadasula, MS,1Archana Thakur, PhD,1
Steven Buck,2Nai-Kong V. Cheung, MD, PhD,3and Lawrence G. Lum, MD, DSc1*
Gangliosides are glycosphingolipids that are expressed on the
surface of mammalian cells, and are concentrated in nervous
tissues . The ganglioside GD2 is expressed in human neuro-
blastoma, melanoma, and osteosarcoma as well as certain brain
tumors [2,3]. Because of its tumor-specific and persistent surface
expression, GD2 is an attractive target for cancer immunotherapy
. GD2 has been identified as an important target antigen for
antibody-dependent cellular cytotoxicity (ADCC) for neuroblas-
toma and malignant melanoma cells .
Neuroblastoma is the most common extra-cranial tumor in
children. Stage 4 disease in children more than 18 months of
age at diagnosis is uniformly aggressive and often recurrent fol-
lowing successful induction therapy. Despite the use of intensive
regimens, the survival rates for such patients have remained un-
acceptable for more than two decades. Since the first phase I study
of anti-GD2 monoclonal antibodies (mAb)  to the most recent
randomized trial, GD2 is accepted as a viable tumor target for
immunotherapy . The recent Children’s Oncology Group trial
of anti-GD2 ch14.18 antibody, IL-2, and granulocyte–macrophage
colony stimulating factor combination, following autologous stem
cell transplant in patients with high-risk neuroblastoma was one
of the few rare randomized studies demonstrating the clinical
benefit of antibody immunotherapy in metastatic solid tumors
among children .
A number of mAbs specific for the GD2 (including 14.G2a,
ch14.18, and 3F8) have been used in phase I–II clinical trials [7–
9]. Anti-GD2 mouse mAb 3F8  has shown highly specific
tumor targeting in preclinical studies [11,12] and has shown ob-
jective tumor responses in patients with primary chemotherapy
resistant bone marrow disease .
further development of anti-GD2 therapeutic strategy. Although
survival rates increased in high-risk patients treated with naked
anti-GD2 mAb in combination with cytokines, only 63% of Stage
4 patients remained free of disease at 2 years . Long-term surviv-
al analysis showed that more than 50% of the patients with Stage 4
neuroblastoma developed recurrent disease after treatment with
anti-GD2 mAb alone . The factors limiting the clinical utility
andefficacyofnakedanti-GD2 mAb are mostlyunknown;however,
the deficiencies in number and activity of effector cells mediating
ADCC observed in patients with post-chemotherapy immunosup-
pression may play a role. We therefore hypothesized that arming of
ex vivo activated and expanded cytotoxic T cells with anti-
CD3 ? anti-GD2 bispecific antibody (BiAb) will redirect them to
GD2 positive tumors and result in enhanced cytotoxicity.
In this study, we exploit the non-MHC restricted, perforin/
granzyme-mediated cytotoxic ability of activated T cells (ATC)
by redirecting their cytotoxicity using a BiAb approach. BiAb was
produced by chemically heteroconjugating anti-CD3 (OKT3) and
anti-GD2 (3F8) to generate OKT3 ? 3F8 BiAb (3F8BiAb). The
first antibody is directed at CD3 on T cells and the second targets
GD2 expressed on the surface of the tumor cells. Binding of ex
vivo expanded and BiAb coated (armed) T cells to the tumor
targets, through the tumor-specific portion of the BiAb molecule,
reactivates the T cells. This approach has been used to redirect
Background. The ganglioside GD2 is an attractive target for
immunotherapy of neuroectodermal tumors. We tested a unique
bispecific antibody anti-CD3 ? anti-GD2 (3F8BiAb) for its ability
to redirect activated T cells (ATC) to target GD2-positive neuroblas-
tomas. Procedure. ATC were generated from normal human periph-
eral blood mononuclear cells (PBMC) by stimulating the PBMC with
OKT3 and expanding the T cells in the presence of interleukin 2 (IL-
2) for 14 days. ATC were armed with 3F8BiAb (100 ng/106cells) or
Her2BiAb (50 ng/106cells) prior to use. 3F8 BiAb were tested for its
dual-binding specificity to GD2 expressed on cancer cell lines and
CD3 expressed on ATC. 3F8BiAb-armed ATC were further tested ex
vivo for their cytotoxicity against GD2 positive tumor targets and its
ability to induce cytokine response upon binding to targets. Results.
GD2 expression in neuroblastoma cells was confirmed by FACS
analysis. Specific binding of 3F8BiAb to the tumor targets as well
as to ATC was confirmed by FACS analysis. 3F8BiAb-armed ATC
exhibited specific killing of GD2 positive neuroblastoma cell lines
significantly above unarmed ATC (P < 0.001). GD2BiAb-armed
ATC secreted significantly higher levels of Th1cytokines and che-
mokines compared to unarmed ATC (P < 0.001). Conclusions.
These preclinical findings support the potential of a novel immuno-
therapeutic approach to target T cells to neuroblastoma.
? 2012 Wiley Periodicals, Inc.
bispecific antibody; GD-2; immunotherapy; neuroblastoma; T cells
1Department of Oncology, Wayne State University, Barbara Ann
Karmanos Cancer Institute, Detroit, Michigan;2Division of Pediatric
Hematology/Oncology, Department of Pediatrics, Children’s Hospital
of Michigan, Wayne State University, Detroit, Michigan;3Department
of Pediatrics, Memorial Sloan-Kettering Cancer Center, New York,
Conflicts of interest: Nothing to report.
Sri Vidya Kondadasula and Maxim Yankelevich contributed equally
to this work.
*Correspondence to: Lawrence G. Lum, MD, DSc, Barbara Ann
Karmanos Cancer Institute, 7th Floor, HWCRC, Rm 740.1, 4100
John R., Detroit 48201, MI. E-mail: firstname.lastname@example.org
**Correspondence to: Maxim Yankelevich, MD, Division of Hema-
tology/Oncology, Children’s Hospital of Michigan, 3901 Beaubien,
Detroit 48201, MI. E-mail: email@example.com
Received 3 December 2011; Accepted 24 May 2012
? 2012 Wiley Periodicals, Inc.
Published online in Wiley Online Library
ATC toward Her2/neuþ [14,15], EGFRþ , and CD20þ 
targets. Multiple infusions of targeted T cells in phase I clinical
trials have been shown to be safe in patients with breast cancer .
Since 3F8 mAb treatment showed promising clinical outcomes
in children with high risk neuroblastoma , we produced
3F8BiAb to investigate whether 3F8BiAb-armed ATC can medi-
ate specific cytotoxicity directed at GD2-positive neuroblastoma
cell lines. In this study, we showed that ATC armed with 3F8BiAb
could engage and lyse GD2-positive neuroblastoma targets, pro-
viding preclinical rationale for immunotherapy using 3F8BiAb-
armed ATC in children with neuroblastoma.
MATERIALS AND METHODS
Cell Lines and mAb
The neuroblastoma cell lines (LAN-1, LAN-6, KCNR, and
LHN) were all grown in RPMI-1640 media (Lonza, Walkersville,
MD) supplemented with 10% fetal bovine serum (FBS; Valley
Biomedical, Winchester, VA). The cell lines were a kind gift from
Dr. Leonid Metelitsa (Texas Children’s Cancer Center, Houston,
TX). OKT3 is an anti-CD3 murine IgG2a(Janssen Biotech, Hor-
sham, PA) that is available commercially. 3F8 is also a murine
anti-GD2 mAb of the IgG3 subclass previously described .
Herceptin (trastuzumab) is an anti-HER2/neu, humanized IgG1
(Genentech, Inc., San Francisco, CA) antibody that is available
Preparation of Bispecific Antibodies
Both Anti-CD3 ? Anti-GD2 BiAb and Anti-CD3 ? Anti-
Her2/neu BiAb was prepared by chemical heteroconjugation as
previously described by Sen et al. . Anti-CD3 (OKT3; Cen-
trocor Ortho-Biotech, Raritan, NJ) was cross-linked with Traut’s
reagent (2-iminothiolane HCl; Pierce, Rockford, IL) and Anti-
GD2 was cross-linked with sulphosuccinimidyl 4-(N-maleimido-
linked mAb were desalted on PD-10 columns (Pharmacia,
Uppsala, Sweden) to remove unbound cross-linker. The cross-
linked OKT3 and anti-GD2 were heteroconjugated overnight.
The heteroconjugated product was analyzed by non-reducing
SDS–PAGE (4–20% gradient; Lonza Inc., Walkersville, MA)
and quantified by densitometry using Quantity One software
(Bio-Rad Lab., Hercules, CA).
Generation and Arming of ATC With the BiAb
Armed ATC were prepared as described previously . Brief-
ly, peripheral blood mononuclear cells (PBMC) purified from
whole heparinized blood of healthy donors by ficoll-hypaque
density gradient centrifugation were activated in culture with
20 ng/ml OKT3, and cultured for 14 days in complete RPMI
1640 media containing 100 IU/ml of IL-2 (Chiron Corp., Emery-
ville, CA). The harvested ATC were armed with Her2BiAb or
3F8BiAb at indicated doses/106ATC for 15 minutes at room
temperature. We wash armed ATC at least twice after arming in
complete medium to eliminate any unbound dimer (BiAb), mono-
mers and multimer before using them in vitro experiments. Blood
collection and use of human blood products for research were
conducted under protocols approved by the Internal Review Board
at Wayne State University. Signed consents for blood draws were
obtained from normal healthy donors.
Tumor cell lines were stained with monoclonal mouse anti-GD2
IgG2a (clone 14.G2a; BD Pharmingen, San Diego, CA) and mono-
clonal mouse anti-human c-erbB-2 IgG1 (clone 9G6; BD Pharmin-
gen) followed by a goat anti-mouse-F(ab0)2 Phycoerythrin (PE)
secondary antibody (Biosource International, Camarillo, CA) and
analyzed with a EPICS-XL-MCLflow cytometer(BeckmanCoulter,
Brea, CA) equipped with an Argon laser. More than 10,000 events
were analyzed in all experiments. The percentage of positive cells
was determined from markers set with matched isotype control anti-
bodies. Analysis was performed using Coulter System II Software.
Flow Cytometry for 3F8 BiAb-Binding Assays
Binding of 3F8 mAb and 3F8BiAb to ATC was evaluated by
incubating 106ATC with either 1 or 10 mg of 3F8 mAb or
3F8BiAb for 1 h at 48C. A goat anti-mouse IgG3-fluorescein
isothiocyanate (FITC; eBioscience, San Diego, CA) was used to
label the 3F8 mAb and the 3F8 fragment of 3F8BiAb. Similarly,
binding of 3F8 BiAb to LAN-6 and KCNR targets was deter-
mined by incubating 106LAN-6 or KCNR cells with either 1 or
10 mg of 3F8 BiAb for 1 h at 48C. The cells were washed and
incubated with goat anti-mouse IgG2a-PE to measure the OKT3
fragment of the BiAb (Becton Dickinson, San Diego, CA). Cells
were analyzed by flow cytometry.
Specific Cytotoxicity Assay
Tumor cell lines were plated into 96-well flat-bottom plates at
cell concentration of 4 ? 104cells/well and incubated overnight
at 378C. The following day, targets were washed once and labeled
with51Cr (2 mCi/well; MP Biomedicals, Irvine, CA) for 4 hours
at 378C. The plates were washed three times with medium after
51Cr labeling and ATC or armed ATC were plated with the target
cells in duplicates at different effector to target (E:T) ratios and
incubated overnight for 18 hours at 378C. 3F8 mAb-mediated
ADCC was evaluated by exposing target cells to PBMC or ATC
plus 100 ng of 3F8 mAb. Her2 BiAb (50 ng/106cells)-armed
ATC were used in some experiments as an irrelevant antibody
control. 0.1 ml aliquots of the supernatants were harvested for
quantification of chromium release. Spontaneous release was de-
termined by incubation of targets with media alone, and maxi-
mum release was determined by lysing the targets with 2%
sodium dodecyl sulfate (SDS) solution. Specific lysis of target
cells was calculated by measuring
using the formula: [% specific lysis ¼ (test ? spontaneous re-
lease)/(maximum release ? spontaneous release) ? 100]. Results
are expressedasa percentage
(mean ? SD) from duplicate wells.
Cytokine Profiling of Co-Cultures
51Cr release in supernatants
Cytokines were quantitated in culture supernatants following
binding of 3F8BiAb armed ATC to GD2 expressing tumor cells.
Unarmed ATC, armed ATC and GD2 positive tumor targets were
co-cultured in a 96-well plate for 24 hours. Tumor cells alone,
unarmed ATC and armed ATC without tumor cells were run as
controls using a 25-plex human cytokine Luminex Array (Invi-
trogen, Carlsbad, CA) on a Bio-Plex system (Bio-Rad Lab.). The
limit of detection for these assays is <10 pg/ml based on detect-
able signal of greater than twofold above background (Bio-Rad
2 Yankelevich et al.
Pediatr Blood Cancer DOI 10.1002/pbc
Lab.). Cytokine concentrations were automatically calculated by
the BioPlex Manager Software (Bio-Rad Lab.).
All statistical analysis were performed using GraphPad Prism 5
for windows (GraphPad software, San Diego,CA). Paired t-test was
used to assess the optimal dose of the BiAb needed to exhibit maxi-
mum cytotoxicity with P < 0.05 considered as significant. Two-way
analysis of variance (ANOVA) was used to analyze results from
Heteroconjugation of 3F8 Bispecific Antibody
The heteroconjugated product of equimolar concentrations of
OKT3 and 3F8 mAb was quantified by Coomassie blue staining
of SDS-gel as shown in Figure 1. Densitometric quantitation of
Lane 4 of the gel showed 73.5% monomer, 17.5% dimer, and 9%
GD2 and Her2 Expression in Neuroblastoma Cells
The expression of GD2 and Her2 proteins on neuroblastoma
cells was quantitated by flow cytometric analysis (Fig. 2). All
neuroblastoma cell lines expressed high levels of GD2 except
for LAN-6 [LAN-1,100%
(MFI ¼ 2555.3); and KCNR, 100% (MFI ¼ 4784.9)], which did
not show any detectable GD2 expression. There was no detectable
expression of Her2 on any of the neuroblastoma cell lines tested.
(MFI ¼ 6667.6);LHN, 100%
lowing chemical heteroconjugation of OKT3 and 3F8 mAb’s as de-
scribed in materials and methods. The product was resolved by SDS–
non-reducing polyacrylamide gradient (4–20%) gel electrophoresis
and detected by Coomassie blue staining. Lane 1, 8 mg of 3F8
mAb; lane 2, 8 mg of OKT3 mAb; lane 3, no sample; lane 4, 8 mg
of 3F8 BiAb. The unconjugated OKT3 and 3F8 (monomers), conju-
gated 3F8 BiAb dimers and multimers are indicated in the figure.
Production of 3F8BiAb. The 3F8BiAb was generated fol-
flow cytometry, using anti-GD2 and anti-c-erbB2 mAb. Data are presented as histograms with their matched IgG isotype antibodies (open
dotted curves) to account for background fluorescence. Solid gray histograms represent cells that are positive for GD2. Percentage of positive
cells is shown as well.
GD2 and Her2 expression in neuroblastoma cell lines. The tumor cells were examined for expression of GD2 and Her2 proteins by
Anti-GD2 ? Anti-CD3 Bispecific Antibody-Armed T Cells3
Pediatr Blood Cancer DOI 10.1002/pbc
Dual-Binding Specificity of 3F8 BiAb
Binding of 3F8 mAb and 3F8BiAb to ATC was determined by
arming ATC with 1 mg of the above antibodies, followed by stain-
ing with a FITC-conjugated anti-mouse IgG3 to measure the
amount of 3F8BiAb bound on the ATC (Fig. 3A). Approximately,
?95% of ATC stained positive for 3F8BiAb binding. In order to
determine the binding of 3F8BiAb to the target cells, GD2-negative
LAN-6 cells and GD2-positive KCNR cells were stained with the
3F8BiAb. The relative amount of OKT3 parental mAb associated
with the 3F8BiAb was quantitated using PE-conjugated IgG2a
(Fig. 3B). KCNR cells showed 100% positive staining for binding
of 3F8BiAb. In contrast, LAN-6 cells did not stain with 3F8BiAb,
confirming the absence of GD2 expression on LAN-6.
Specific Cytotoxicity With Increasing Arming Doses of
the 3F8 BiAb
To determine the optimal arming dose of 3F8BiAb, a dose
titration of 3F8BiAb using
done against KCNR neuroblastoma target cells. ATC obtained
from four normal donors were armed with increasing doses of
3F8BiAb ranging from 25 to 250 ng/106ATC. Unarmed ATC
from the same donors were used as controls. Figure 4A shows
the mean % specific cytotoxicity for each donor ATC armed with
increasing doses of the 3F8BiAb targeted against KCNR tumor
cells at an E:T ratio of 25:1 for 18 hours. Increasing the arming
doses of 3F8BiAb up to 100 ng/106ATC showed dose-dependent
51Cr release cytotoxicity assay was
increase in specific cytotoxicity against the KCNR neuroblastoma
cell line. However, arming doses above 100 ng/106ATC did not
result in any further increase in specific cytotoxicity. Therefore,
the dose of 100 ng/106ATC (P < 0.007) was selected as an
optimal dose for all subsequent experiments. The unarmed ATC
control showed the expected low levels of non-MHC restricted
cytotoxicity against the same targets.
Effect of Effector:Target Ratio on Specific Cytotoxicity
by 3F8BiAb Armed ATC
51Cr labeled KCNR and LAN-1 neuroblastoma cell lines and
3F8BiAb-armed ATC were plated at various E:T ratios ranging
from 3.125:1 to a maximum of 25:1. Figure 4B shows mean %
specific cytotoxicity of armed ATC against KCNR (upper panel)
and LAN-1 (lower panel) targets. Increasing the E:T ratio led to
increasing mean % specific cytotoxicity from 24.2% at 3.125:1 to
55.7% at 25:1 for KCNR specific lysis and from 7.4% at 3.125:1
to 37.2% at 25:1 for LAN-1 specific lysis. Unarmed ATC were
run as controls for the same ratios.
Influence of ADCC on 3F8BiAb-Induced Cytotoxicity
In order to assess the contribution of ADCC to 3F8BiAb-
armed ATC and naked 3F8 mAb-mediated killing, both ATC
and PBMC were tested as potential sources of effector cells
mediating ADCC in KCNR neuroblastoma cell line (Fig. 5A).
PBMC with Her2 mAb were used as an irrelevant antibody con-
trol and Her2BiAb armed ATC as irrelevant BiAb control. The
dose of 3F8 mAb and 3F8BiAb applied to the targets with ATC
was 100 ng/well and 100 ng/106cells, respectively. While we
detected a substantial amount of ADCC-mediated killing of neu-
roblastoma cells with 3F8 mAb plus PBMC (Fig. 5A,B), the
cytotoxicity exhibited by 3F8 mAb plus ATC was not detected
suggesting that ATC were not a significant source of effector cells
mediating ADCC. Also, the cytotoxicity mediated by 3F8BiAb-
armed ATC was higher than ADCC mediated by 3F8 mAb plus
PBMC against neuroblastoma cell lines (P < 0.01; Fig. 5A,B).
Specific Cytotoxicity of 3F8BiAb-Armed ATC Against
Neuroblastoma Cell Lines
ATC derived from six normal subjects were armed with 100 ng
of 3F8BiAb per 106cells and co-cultured for 18 hours with
51Cr labeled neuroblastoma cells at a E:T ratio of 25:1. Control
conditions consisted of unarmed ATC alone and 3F8 mAb (100 ng/
well) with PBMC. Her2BiAb armed ATC from the same donors
were tested on neuroblastoma as irrelevant antibody control.
3F8BiAb-armed ATC effectively killed GD2-positive neuroblasto-
ma cell lines (KCNR and LHN (P < 0.001), LAN-1 (P < 0.05),
whereas the killing of GD2-negative neuroblastoma (LAN-6) was
minimal (Fig. 5B). Her2Bi-armed ATC did not kill the neuroblasto-
ma cell lines, further verifying the specificity of 3F8BiAb mediated
killing of GD2-positive neuroblastoma cells.
Enhanced Secretion of Cytokines Following Binding of
3F8BiAb-Armed ATC to GD2 Positive Tumors
Since binding of armed ATC to Her2/neu on SK-BR-3 cells
triggered cytokine and chemokine secretion , we tested if
3F8BiAb to ATC. 1 ? 106ATC were armed with 1 mg each of
3F8mAb and 3F8BiAb and the amount of antibody bound to the
surface of the cells was measured by flow cytometry as described
in materials and methods. Solid gray histograms represent cells that
have OKT3 bound to their surface via 3F8 (3F8BiAb). Percentage
positive cells shown in parenthesis were obtained by setting the
markers using appropriate isotype control antibodies (open dotted
lines). B: Binding of 3F8BiAb to tumor cells. LAN-6 (106cells)
and KCNR (106cells) were incubated with 1 mg of 3F8BiAb and
analyzed by flow cytometry. The solid grey histograms show the
binding of 3F8BiAb to the target cells. Matched isotype controls
were used to quantitate the positive cells shown in parenthesis.
Binding of 3F8BiAb to ATC and target cells. A: Binding of
4 Yankelevich et al.
Pediatr Blood Cancer DOI 10.1002/pbc
armed or unarmed ATC stimulated with KCNR tumor cells for
24 hours could trigger cytokine and chemokine production
(Fig. 6). As expected, stimulation of 3F8BiAb armed ATC with
target cells induced higher levels of two key cytokines, IFN-g and
TNF-a which were known to play a key role in cytotoxicity
compared to unarmed ATC. Interestingly, no change in the levels
of IL-4, IL-10 was detected between tumor cells alone or tumor
cell stimulated with ATC or armed ATC (n ¼ 4). The most strik-
ing up-regulation was seen in the chemokine secretion levels of
MIP-1a, MIP-1b and RANTES with significantly higher levels
(P < 0.001) in culture supernatant of 3F8BiAb armed-ATC stim-
ulated with tumor cells compared to co-cultures of unarmed ATC
and tumor cells or tumor cells alone.
In this study, we produced and functionally characterized a
unique BiAb 3F8BiAb that recognizes the tumor-associated gan-
glioside GD2 and the T-cell receptor antigen CD3. Our data show
that 3F8BiAb can activate and redirect non-MHC restricted cyto-
toxic activity of ATC toward GD2-positive neuroblastoma cell
lines. The tumor cell killing was GD2-specific. To the best of
our knowledge, the strategy of targeting neuroblastoma with 3F8
BiAb-armed ATC has not been reported.
Studies using anti-GD2 BiAb with a different design had been
reported. Thus, Bernhard et al.  used Fab’ dimer anti-
CD3 ? anti-GD2 BiAb to target melanoma cell lines, and Man-
zke et al.  used 14G2a in a tetradoma-based anti-GD2 ? anti-
CD3 BiAb in preclinical experiments targeting the human neuro-
blastoma cell line IMR5. Compared to ours, these BiAb produc-
tion methods are labor intensive and require multiple steps of
production and purification of BiAb. Both groups did not coat T
cells with BiAb, but used co-injection of BiAb with effectors. In
our opinion, this may result in less effective arming and subse-
quent redirection of T cells to targets.
Most clinical studies used direct intravenous injection of BiAb
and were limited by cytokine storm induced by BiAb binding to
Fc-receptor bearing cells . Recently, the development of T-
cell engager BiAb (BiTE) and trifunctional bispecific antibodies
(TriFAb) with engineered Fc have met with clinical success in the
treatment of mantle cell lymphoma and ascites in ovarian cancer
trials [22–26]. There are ongoing phase I/II clinical trials that are
250 ng per 1 ? 106cells was measured in51Cr-release assay against KCNR neuroblastoma cells. Controls included unarmed ATC (0 ng). E:T
of 25:1 was used for these experiments. Mean % cytotoxicity ? SEM for four healthy donors (&) is shown for each dose of the BiAb.
?P < 0.01 and??P < 0.007 as analyzed by paired t-test. B: Cytotoxicity of 3F8BiAb armed ATC increases with E:T ratio. ATC
generated from healthy donors (n ¼ 2) were armed with 100 ng/1 ? 106cells of 3F8BiAb ( ) and co-cultured at four different E:T
ratio; overnight at 378C in 96-well flat-bottom plates containing51Cr-labeled KCNR (upper panel) or LAN-1 (lower panel) target cells.
Unarmed ATC were used as controls (&). Mean % cytotoxicity was calculated and error bars represent SD from two experiments.
A: Arming dose titration of 3F8BiAb. Cytotoxicity mediated by ATC armed with the 3F8BiAb at doses of 25, 50, 100, 200, and
Anti-GD2 ? Anti-CD3 Bispecific Antibody-Armed T Cells5
Pediatr Blood Cancer DOI 10.1002/pbc
promising using a number of different BiAb that target solid
tumor and hematologic malignancies (reviewed in ). Studies
from our group have used BiAb armed ATCs to increase cytotox-
icity directed at both solid tumors and hematologic malignancies
[16,17,21,27]. However, none of our studies have targeted neuro-
Our in vitro data showed that 3F8BiAb armed ATC secreted
increased levels of IFN-g and TNF-a when they specifically
engaged tumor targets but there was no change in IL-4 and IL-
10 secretion by 3F8BiAb-armed ATC stimulated with tumor cells.
Infusions of armed ATC may reverse tumor tolerance by polariz-
ing the tumor microenvironment towards a Th1 condition rich in
IFN-g and TNF-a, which are known to be tumoricidal and capa-
ble of inducing local immunization and systemic anti-tumor
responses . Shifting the in vivo immune responses to
Th1 was consistent with our observations of PBMC from women
who received multiple infusions of armed ATC and developed
specific cytotoxicity directed at breast cancer cell lines that per-
sisted at least 4 months after immunotherapy . In addition to
cytokines, chemokines such as RANTES, MIP-1a, and MIP-1b
were significantly upregulated (P < 0.001) by 3F8BiAb-armed
ATC upon engagement with tumor cells. RANTES, MIP-1a,
and MIP-1b are known to modulate T-cell migration, degranula-
tion, co-stimulation, and tumor-specific cytolytic activity [30,31].
In addition, these soluble factors can recruit antigen-presenting
cells and naive T cells to the site, and may facilitate antigen
presentation and endogenous T-cell activation [30–32].
The use of GD2 as an immunological target was confirmed in
clinical trials that reported objective clinical responses and im-
proved survival in children with neuroblastoma who received
infusions of anti-GD2 mAbs [6,8,9,13]. In vitro studies show
that ADCC is the major mechanism of tumor cell killing mediated
by the anti-GD2 mAb with NK-cells and granulocytes being the
key effector cells [4,33,34]. In the current study, we addressed the
hypothesis whether the cytotoxicity mediated by 3F8BiAb armed
ATC is comparable or better than 3F8 mAb-mediated ADCC.
The potential clinical advantage of 3F8BiAb-armed ATC over
naked 3F8 mAb includes a significantly lower dose of 3F8BiAb to
arm ATC compared to intravenously injected dose of 3F8 mAb
(10 mg/m2/dose ? 5 days). GD2 expression in normal human
tissues is restricted to the CNS, peripheral nerves, and skin mel-
anocytes . While CNS neurons are protected from anti-GD2
mAb (3F8) by blood brain barrier, their binding to peripheral
nerve fibers causes pain, which represents one of the major
dose-limiting toxicities of 3F8. All patients receiving 3F8 require
morphine administration. Fevers, urticarial rash, hypotension, and
development of human anti-mouse antibodies (HAMA) responses
have been reported as the other common side effects of 3F8 .
The actual dose of 3F8 required to arm 2.5 ? 109ATC (a targeted
cell dose per one weekly infusion in 20 kg child) would be
0.25 mg, but with a 20% arming efficiency, the actual amount
of antibody bound to the ATC after washing would be only
0.05 mg/2.5 ? 109ATC. This is 200 times less than a single
dose of naked 3F8 used in phase II–III trials. Since most of the
3F8 side effects were dose dependent, and rarely noted at dosages
of <1 mg/m2, we do not expect any significant toxicities
related to 3F8 component of BiAb-armed ATC therapy.
In summary, our results showed that (i) 3F8BiAb-armed ATC
induced cytotoxic activity directed at GD2 positive neuroblastoma
cells; and (ii) 3F8BiAb-armed ATC secreted higher levels of
tumoricidal cytokines IFN-g and TNF-a and chemokines MIP-
1a, MIP-1b, and RANTES compared to tumor cells alone or ATC
stimulated with tumor cells. This approach may provide a viable
alternative or synergistic addition to mAb therapies alone for the
pediatric population with high risk neuroblastomas. The in vitro
targeting studies provide a strong rationale for the use of 3F8BiAb
armed ATC for the initiation of phase I/II clinical trials in patients
with GD2 positive tumors. This approach with potentially low
51Cr labeled target cells were co-cultured in the presence of unarmed
ATC, 3F8 mAb plus ATC, 3F8 mAb plus PBMC and armed ATC with
3F8 BiAb, controls comprised of Her2 mAb in the presence of PBMC
and Her2BiAb-armed ATC. A standard 18-hour chromium release cy-
totoxicity assay was performed as described in materials and methods.
E:Tof 25:1 was used. Representative donor for neuroblastoma (KCNR)
cells is shown. Error bars represent SD of two replicates for each
condition.??P < 0.01 for KCNR cells versus 3F8mAb plus PBMC
control. B: 3F8BiAb-armed ATC kill GD2-expressing neuroblastoma
cells in co-culture. LAN-6, LAN-1, LHN, and KCNR neuroblastoma
cell lines were exposed to ATC armed with 100 ng/1 ? 106cells of
3F8BiAb ( ) in a standard51Cr release assay. ATC alone (&), 100 ng/
well of 3F8 mAb combined with PBMC ( ) and Her2Bi-armed ATC
( ) were used as controls. E:Twas 25:1. Mean % cytotoxicity ? SEM
of six donors with each condition done in duplicates is shown.
?P < 0.05 and???P < 0.001 versus the unarmed ATC control.
A: 3F8 mAb dependent ADCC against neuroblastoma cells.
6 Yankelevich et al.
Pediatr Blood Cancer DOI 10.1002/pbc
toxicity profile may be used in combination with adjuvant surgery,
chemotherapy, or high dose chemotherapy and stem cell trans-
plant to enhance overall survival and quality of life.
This study was supported in part by R01 CA 092344 (L.G.L.),
R01 CA 140314 (L.G.L.). M.Y. is a Hyundai Hope on Wheels
1. Modak S, Cheung NK. Disialoganglioside directed immunotherapy of neuroblastoma. Cancer Invest
2. Wu ZL, Schwartz E, Seeger R, et al. Expression of GD2 ganglioside by untreated primary human
neuroblastomas. Cancer Res 1986;46:440–443.
3. Longee DC, Wikstrand CJ, Mansson JE, et al. Disialoganglioside GD2 in human neuroectodermal
tumor cell lines and gliomas. Acta Neuropathol 1991;82:45–54.
4. Honsik CJ, Jung G, Reisfeld RA. Lymphokine-activated killer cells targeted by monoclonal antibodies
to the disialogangliosides GD2 and GD3 specifically lyse human tumor cells of neuroectodermal
origin. Proc Natl Acad Sci USA 1986;83:7893–7897.
5. Cheung NKV, Lazarus H, Miraldi FD, et al. Ganglioside Gd2 specific monoclonal antibody-3F8—A
phase-I study in patients with neuroblastoma and malignant-melanoma. J Clin Oncol 1987;5:
armed ATC with tumor (T þ aATC), tumor alone (T), and tumor þ ATC (T þ ATC) obtained from four normal donors were tested for Th1
cytokines IFN-g and TNF-a (upper panel), Th2cytokines IL-4 and IL-10 (middle panel), and chemokines MIP-1a, MIP-1b, and RANTES
Cytokine profile of culture supernatants detected by multiplex luminex system. Co-culture supernatants of ATC alone (ATC), 3F8BiAb
Anti-GD2 ? Anti-CD3 Bispecific Antibody-Armed T Cells7
Pediatr Blood Cancer DOI 10.1002/pbc
6. Yu AL, Gilman AL, Ozkaynak MF, et al. Anti-GD2 antibody with GM-CSF, interleukin-2, and Download full-text
isotretinoin for neuroblastoma. N Engl J Med 2010;363:1324–1334.
7. Frost JD, Hank JA, Reaman GH, et al. A phase I/IB trial of murine monoclonal anti-GD2 antibody
14.G2a plus interleukin-2 in children with refractory neuroblastoma: A report of the Children’s Cancer
Group. Cancer 1997;80:317–333.
8. Cheung NK, Kushner BH, Yeh SD, et al. 3F8 monoclonal antibody treatment of patients with stage 4
neuroblastoma: A phase II study. Int J Oncol 1998;12:1299–1306.
9. Kushner BH, Kramer K, Cheung NK. Phase II trial of the anti-G(D2) monoclonal antibody 3F8
and granulocyte-macrophage colony-stimulating factor for neuroblastoma. J Clin Oncol 2001;19:
10. Cheung NK, Saarinen UM, Neely JE, et al. Monoclonal antibodies to a glycolipid antigen on human
neuroblastoma cells. Cancer Res 1985;45:2642–2649.
11. Cheung NK, Berger N, Coccia P, et al. Murine monoclonal-antibody (Mab) specific for Gd2 Ganglio-
side—A phase-I trial in patients with neuroblastoma, melanoma and osteogenic-sarcoma. Proc Am
Assoc Cancer Res 1986;27:318.
12. Miraldi FD, Nelson AD, Kraly C, et al. Diagnostic-imaging of human neuroblastoma with radiolabeled
antibody. Radiology 1986;161:413–418.
13. Simon T, Hero B, Faldum A, et al. Long term outcome of high-risk neuroblastoma patients after
immunotherapy with antibody ch14.18 or oral metronomic chemotherapy. BMC Cancer 2011;11:
14. Sen M, Wankowski DM, Garlie NK, et al. Use of anti-CD3 ? anti-HER2/neu bispecific antibody for
redirecting cytotoxicity of activated T cells toward HER2/neu Tumors. J Hematother Stem Cell Res
15. Grabert RC, Cousens LP, Smith JA, et al. Human T cells armed with Her2/neu bispecific antibodies
divide, are cytotoxic, and secrete cytokines with repeated stimulation. Clin Cancer Res 2006;12:
16. Reusch U, Sundaram M, Davol PA, et al. Anti-CD3 ? anti-epidermal growth factor receptor (EGFR)
bispecific antibody redirects T-cell cytolytic activity to EGFR-positive cancers in vitro and in an animal
model. Clin Cancer Res 2006;12:183–190.
17. Gall JM, Davol PA, Grabert RC, et al. T cells armed with anti-CD3 ? anti-CD20 bispecific antibody
enhance killing of CD20þ malignant B-cells and bypass complement-mediated Rituximab-resistance
in vitro. Exp Hematol 2005;33:452–459.
18. Cheung NKV, Kushner BH, Cheung IY, et al. Anti-GD(2) antibody treatment of minimal residual stage
4 neuroblastoma diagnosed at more than 1 year of age. J Clin Oncol 1998;16:3053–3060.
19. Bernhard H, Karbach J, Strittmatter W, et al. Induction of tumor-cell lysis by bi-specific antibody
recognizing ganglioside Gd2 and T-cell antigen Cd3. Int J Cancer 1993;55:465–470.
20. Manzke O, Russello O, Leenen C, et al. Immunotherapeutic strategies in neuroblastoma: Antitumoral
activity of deglycosylated ricin A conjugated anti-GD2 antibodies and anti-CD3xanti-GD2 bispecific
antibodies. Med Pediatr Oncol 2001;36:185–189.
21. Thakur A, Lum LG. Cancer therapy with bispecific antibodies: Clinical experience. Curr Opin Mol
22. Wolf E, Hofmeister R, Kufer P, et al. BiTEs: Bispecific antibody constructs with unique anti-tumor
activity. Drug Discov Today 2005;10:1237–1244.
23. Molhoj M, Crommer S, Brischwein K, et al. CD19-/CD3-bispecific antibody of the BiTE class is far
superior to tandem diabody with respect to redirected tumor cell lysis. Mol Immunol 2007;44:1935–
24. Topp MS, Kufer P, Gokbuget N, et al. Targeted therapy with the T-cell-engaging antibody blinatumo-
mabof chemotherapy-refractory minimal residual disease in B-lineage acute lymphoblastic leukemia
patients results in high response rate and prolonged leukemia-free survival. J Clin Oncol
25. Burges A, Wimberger P, Kumper C, et al. Effective relief of malignant ascites in patients with
advanced ovarian cancer by a trifunctional anti-EpCAM ? anti-CD3 antibody: A phase I/II study.
Clin Cancer Res 2007;13:3899–3905.
26. Sebastian M, Passlick B, Friccius-Quecke H, et al. Treatment of non-small cell lung cancer patients
with the trifunctional monoclonal antibody catumaxomab (anti-EpCAM ? anti-CD3): A phase I study.
Cancer Immunol Immunother 2007;56:1637–1644.
27. Lum LG, Thakur A. Bispecific antibodies for arming activated T cells and other effector cells for tumor
therapy. In: Kontermann RE, editor. Bispecific antibodies. Berlin, Heidelberg, Stuttgart: Springer-
Verlag; 2011. pp. 243–271.
28. Brunda MJ, Luistro L, Hendrzak JA, et al. Role of interferon-gamma in mediating the antitumor
efficacy of interleukin 12. J Immunother 1995;17:71–77.
29. Thakur A, Schalk D, Al-Khadimi Z, et al. ATh1 cytokine-enriched microenvironment enhances tumor
killing by activated T cells armed with bispecific antibodies and inhibits the development of myeloid-
derived suppressor cells. Cancer Immunol Immunother 2012;61:497–509.
30. Schall TJ, Bacon K, Toy KJ, et al. Selective attraction of monocytes and T lymphocytes of the memory
phenotype by cytokine RANTES. Nature 1990;347:669–671.
31. Schall TJ, Bacon K, Camp RD, et al. Human macrophage inflammatory protein alpha (MIP-1 alpha)
and MIP-1 beta chemokines attract distinct populations of lymphocytes. J Exp Med 1993;177:
32. Taub DD, Conlon K, Lloyd AR, et al. Preferential migration of activated CD4þ and CD8þ T cells in
response to MIP-1 alpha and MIP-1 beta. Science 1993;260:355–358.
33. Munn DH, Cheung NK. Interleukin-2 enhancement of monoclonal antibody-mediated cellular cyto-
toxicity against human melanoma. Cancer Res 1987;47:6600–6605.
34. Barker E, Mueller BM, Handgretinger R, et al. Effect of a chimeric anti-ganglioside GD2 antibody on
cell-mediated lysis of human neuroblastoma cells. Cancer Res 1991;51:144–149.
35. Chan JK, Hamilton CA, Cheung MK, et al. Enhanced killing of primary ovarian cancer by retargeting
autologous cytokine-induced killer cells with bispecific antibodies: A preclinical study. Clin Cancer
8 Yankelevich et al.
Pediatr Blood Cancer DOI 10.1002/pbc