Content uploaded by Reno Debets
Author content
All content in this area was uploaded by Reno Debets on Dec 13, 2015
Content may be subject to copyright.
An Oncolytic Adenovirus Redirected with a Tumor-Specific
T-Cell Receptor
Zsolt Sebestyen,
1
Jeroen de Vrij,
2
Maria Magnusson,
3
Reno Debets,
1
and Ralph Willemsen
1
1
Tumor Immunology Group, Unit of Clinical and Tumor Immunology, Department of Medical Oncology, Erasmus Medical Center-Daniel
den Hoed Cancer Center, Rotterdam, the Netherlands;
2
Virus and Stem Cell Biology Laboratory, Department of Molecular
Cell Biology, Leiden University Medical Center, Leiden, the Netherlands; and
3
Got-a-Gene, Go¨tenburg, Sweden
Abstract
To improve safety and specificity of oncolytic adenoviruses, we
introduced T-cell receptors (TCR) specific for a unique class of
truly tumor-specific antigens into the adenoviral fiber protein.
The adenoviral fiber knob responsible for attachment to the
coxsackie-adenoviral receptor (CAR) on target cells was
replaced by a single-chain TCR (scTCR) molecule with
specificity for the melanoma-associated cancer-testis antigen
MAGE-A1, presented by HLA-A1, and an extrinsic trimeriza-
tion motif in a replicating Ad5 vector (Ad5.R1-scTCR). The
production of the recombinant virus was initiated in a novel
producer cell line that expressed an antibody-based hexon-
specific receptor (293T-AdR) in the cell membrane. This new
production system allowed CAR-independent and target
antigen–independent propagation of Ad5.R1-scTCR. Infection
with adenovirus bearing the scTCR-based fiber resulted in an
efficient killing of target tumor cells. The infection was cell
type specific because only HLA-A1
+
/MAGE-A1
+
melanoma
cells were killed, and thus, this retargeting strategy provides
a versatile tool for future clinical application. [Cancer Res
2007;67(23):11309–16]
Introduction
A prerequisite for the safe and effective application of therapies
that are based on antigen-driven tumor eradication is the nature of
the target antigen, which has to be truly tumor specific. Studying
tumor cell eradication by CTL immunologists identified such
tumor-specific antigens that were targets for CTL while leaving
healthy tissue intact (1, 2). Tumor cells may express a group of
antigens termed ‘‘cancer-testis antigens’’ that are presented as
antigenic peptides by MHC molecules to CTL (3). In fact, cancer-
testis antigens are immunogenic in cancer patients as they may
elicit an anticancer response (4–6). They exhibit highly tissue-
restricted expression and are considered promising target mole-
cules for immunotherapies. To date, 44 cancer-testis antigen gene
families have been identified and their expression has been studied
in numerous cancer types (7). For example, bladder cancer, non–
small lung cancer, and melanoma are high cancer-testis antigen
gene expressers, with 55%, 51%, and 53% of the cancer-testis
antigen transcripts examined by reverse transcription-PCR detected
in 20% or more of the specimens examined, respectively. With the
exception of testis-restricted cancer-testis antigen transcripts, all
remaining cancer-testis antigen transcripts were expressed in
normal pancreas. Other antigens that were shown to elicit potent
antitumor responses in cancer patients include differentiation
antigens, such as the melanoma antigens gp100, Mart-1, and
tyrosinase, or antigens that are overexpressed on tumor cells, such
as p53, Her-2/neu, and WT-1 (7, 8). Both groups of antigens are also
expressed in healthy tissue and may therefore elicit autoimmune
disease when targeted.
Oncolytic adenoviral vectors hold great promise for cancer gene
therapy because they potently eradicate tumor cells (9). Therefore,
efforts are currently invested in improving replication-competent
adenoviruses with respect to safety and specificity to fulfill criteria
for clinical application (10). Among several strategies, genetic
modification of the adenoviral fiber, which is responsible for cell
binding, may result in a logical and preferable site to carry
structures that specifically bind to target antigens of choice,
thereby changing viral tropism.
Application of adenoviruses in a tumor cell–specific fashion
is highly hampered because the natural cellular receptor of
adenovirus is widely expressed on normal tissues and, on the
other hand, often reported to be down-regulated or even absent on
tumor cells (11). To address this issue, one needs strategies to alter
the tropism of adenoviral vectors and retarget them against tumor-
specific antigens.
Recent developments to change the natural tropism of
adenoviral vectors into tumor-specific recognition are based on
the genetic engineering of capsid proteins, such as pIX (12, 13),
hexon, and fiber (14–19).
Genetic modification of the fiber protein has been achieved
either through exchange of the Ad5 fiber knob with the Ad3 knob
(20, 21), knob mutagenesis (22, 23), or incorporation of small
ligands into the knob domain (21, 24). However, it should be noted
that an effective and safe retargeting strategy should include
complete ablation of the natural tropism, which is not guaranteed
by the above-mentioned modifications, and preferably include
deletion of the fiber knob. This can be accomplished by replacing
the fiber knob by new antigen-binding structures and an extrinsic
trimerization signal (18, 21). Antibody or T-cell receptor (TCR)
fragments [e.g., single-chain Fv (scFv) and single-chain TCR
(scTCR)] mediate tumor cell recognition and are able to redirect
T cells (25) and viruses (26) and, as such, are candidate structures
to genetically redirect adenoviruses to tumor cells. Previous
attempts to produce adenoviruses with fibers that include scFv
have failed, most likely as a consequence of improper folding of the
chimeric fiber in the cellular cytoplasm (18).
Here, we show that an oncolytic adenovirus bearing chimeric
fibers, comprising an extrinsic trimerization signal and scTCR with
HLA-A1–restricted MAGE-A1 specificity, can be produced. To this
end, we generated a novel producer cell line expressing an anti-
hexon receptor, which was needed to initiate production of virus
Requests for reprints: Ralph Willemsen, Tumor Immunology Group, Unit of
Clinical and Tumor Immunology, Department of Medical Oncology, Erasmus Medical
Center-Daniel den Hoed Cancer Center, Groene Hilledijk 301, Rotterdam 3075 EA, the
Netherlands. Phone: 31-10-4391574; E-mail: r.a.willemsen@erasmusmc.nl.
I2007 American Association for Cancer Research.
doi:10.1158/0008-5472.CAN-07-0739
www.aacrjournals.org
11309
Cancer Res 2007; 67: (23). December 1, 2007
Research Article
that specifically infects HLA-A1/MAGE-A1
+
melanoma cells but not
MAGE-A1
or HLA-A1
target cells. The presented strategy to
produce genetically retargeted oncolytic adenoviruses holds great
promise to develop clinically applicable anticancer agents.
Materials and Methods
Cells and Antibodies
Target cell lines used in this study are the melanoma cell line MZ2-mel
3.0, MZ2-mel 2.2, and MEL.2A (kindly provided by T. Boon and P. Coulie,
Ludwig Institute for Cancer Research, Brussels, Belgium). The melanoma
cell line 9303-A, 518-A1, and FM-3; the renal cell carcinoma cell line Nemeth
(kindly provided by Dr. E. Oosterwijk, University Medical Center Nijmegen,
Nijmegen, the Netherlands); and the human embryonic kidney cell line
293T and its derivative 293T-AdR ( further described in the Results section)
were grown in DMEM (Life Technologies) supplemented with 10% fetal
bovine serum (FBS; Hyclone). HLA-A1/MAGE-A1 and coxsackie-adenovirus
receptor (CAR) expression status of the target cell lines is further described
in Table 1 (see Results).
Antibodies used in this study were against fiber tail (4D2.5; NeoMarkers),
hexon (clone BOD604, FITC conjugated; Biodesign), rabbit polyclonal
anti-Ad5 (Abcam), TCR Va12.1 (FITC or nonconjugated; Endogen), CAR
(USBiological), and c-Myc (9E10, FITC conjugated; Convance). FITC-
conjugated rabbit anti-mouse IgG Fab fragment (Jackson ImmunoResearch)
or horseradish peroxidase–conjugated goat anti-rabbit IgG (Becton Dick-
inson Biosciences) was used as secondary antibodies.
Infectivity Assay with Fiberless Virus
293T and 293T-AdR cells were transfected with a fiberless pAdeasy-EGFP
construct (a kind gift of Wim Jongmans, University Medical Center,
Nijmegen, the Netherlands) using the CellPhect Transfection kit (Amersham
Biosciences). The expression of the reporter gene EGFP was monitored
using a Leica DMIL inverted fluorescence microscope (Leica Microsystems).
At time points indicated, culture supernatant was collected and virus
release (particle count) was analyzed using the IDEIA Adenovirus ELISA kit
(DakoCytomation).
DNA Constructs
Ad5.R1-scTCR adenoviral DNA was generated as described (18). Briefly,
recombinant fiber genes were constructed using methods based on ligation,
PCR, and splicing by overlap extension. The gene encoding the Ad5
wild-type (WT) fiber was obtained from pAB26 (Microbix, Inc.) by PCR
introducing an upstream Bam HI and downstream XhoI site, respectively.
The knob domain in recombinant fibers was deleted and replaced by a
36-amino acid extrinsic trimerization motif derived from the neck region
peptide (NRP) of human lung surfactant protein D (18). The NRP sequence
followed by a linker sequence from Staphylococcus protein A was ligated to
the COOH-terminal end of fiber shaft with one repeat and named R1, and
the scTCR VaVhCh was added to the COOH-terminal end of the Staph-A
linker. The resulting R1-scTCR fiber was then cloned into a fiberless Ad5
genome as described (18).
Retroviral vectors encoding the HLA-A1 gene, MAGE-A1 complete cDNA,
or MAGE-A1 minigene (encoding the 9-amino acid antigenic epitope
EADPTGHSY) were generated as follows: HLA-A1 and MAGE-A1 cDNA
cloned in pCDNA-3 (a kind gift from Pierre van der Bruggen, Ludwig
Institute for Cancer Research, Brussels, Belgium) were reamplified to
introduce NcoI and XhoI sites and cloned into the retroviral vector pBullet.
The MAGE-A1 minigene was introduced into a version of pBullet that
contains a signal sequence from the G250 antibody heavy chain (27) by
ligation of a small linker encoding the MAGE-A1 minigene next to the signal
sequence.
To construct the membrane-bound adenovirus-specific receptor (AdR),
first an scFv was generated from the hexon-specific hybridoma 2Hx-2
(American Type Culture Collection). In short, RNA isolated from the 2Hx-2
hybridoma was reverse transcribed using SuperScript II (Invitrogen) and
amplified using Ig variable heavy and variable light chain primers
(Amersham scFv module, Amersham Biotech). The variable heavy and
variable light chain DNA fragments were then reamplified to fuse them
together by introducing a linker sequence between the two fragments and
to introduce SfiI and NotI restriction sites. The resulting scFv was then
introduced into the retroviral expression cassette pBullet-CD4g, and
pBullet-cMyc/~, which allows for membrane expression of the scFv (27, 28).
Generation of the Recombinant Virus
293T-AdR cells were transfected with PacI-digested recombinant
adenovirus plasmid (Ad5.R1-scTCR), and after 3 days, culture supernatant
was harvested and used immediately for infection or further analysis.
Adenovirus particle count (semiquantitative) was determined by IDEIA
Adenovirus ELISA kit. Infectious adenovirus particle number [multiplicity
of infection (MOI)] was determined by the Adeno-X Rapid titer kit (BD
Clontech) on 293T-AdR cells.
Analysis of Adenoviral Particles
Electron microscopy. For electron microscopy, MZ2-mel 3.0 cells were
fixed in 1.5% glutaraldehyde in 0.1 mol/L cacodylate buffer for 1 h at room
temperature, postfixed in 1% OsO
4
in the same buffer for 1 h at 4jC,
dehydrated in a graded ethanol series, and embedded in epon. Ultrathin
sections were poststained with uranyl acetate and lead citrate and viewed
with a Tecnai 12 electron microscope at 80 kV (FEI).
Flow cytometry. Supernatant derived from Ad5.R1-scTCR–producing
293T-AdR cells (12.5 mL containing 10
8
particles/mL) was incubated
overnight with magnetic beads (Dynal Biotech ASA) that were loaded with
in vitro–generated HLA-A1 complexes (1 Ag total) presenting the MAGE-A1
nonapeptide (EADPTGHSY) or an irrelevant peptide derived from influenza
virus A nucleoprotein (CTELKLSDY). After three wash steps with PBS, the
beads were incubated with a saturating concentration of anti-hexon
FITC
monoclonal antibody (mAb) and incubated for 30 min at 4jC. Specific
binding of Ad5.R1-scTCR virus to the beads was then analyzed by flow
cytometry on a Cytomics FC-500 flow cytometer (Beckman Coulter).
Western blotting. Ad5.R1-scTCR virus bound to the HLA-A1/MAGE-A1–
coated magnetic beads was eluted from the beads by addition of high-
affinity Fab fragments that specifically bind to HLA-A1/MAGE-A1 (15 min
at room temperature, 39 Ag total in 1 mL PBS; ref. 29). Excess high-affinity
Fab fragments were then removed by addition of Ni-NTA agarose (Qiagen)
that binds to the 6
His tag present in the Fab fragment. Purified Ad5.R1-
scTCR virus was then separated on SDS-PAGE, immobilized on a
nitrocellulose membrane, and detected with fiber tail–specific mAb (4D2).
Detection of Adenoviral Infection
Flow cytometry of infected cells. One million cells were infected at
indicated virus particle-to-cell ratios using virus supernatant diluted in
DMEM supplemented with 10% FBS for 2 h at 37jC/5% CO
2
. After infection,
cells were seeded in six-well plates. Cells were harvested 2 days after
Table 1. Antigen expression and infection of melanoma
and renal cell carcinoma cell lines by Ad5.R1-scTCR
Target cell HLA-A1 MAGE-A1 CAR Infection by
Ad5.R1-scTCR*
MZ2-mel 3.0 + + +
9303-A + + + +
518-A1 + + + +
MZ2-mel 2.2 +
MEL.2A + +
FM-3 +
Nemeth +
293T +
*Represents infection of target cells analyzed by methylene blue
staining. + or is based on qualitative comparison with noninfected
control cells.
Cancer Research
Cancer Res 2007; 67: (23). December 1, 2007
11310
www.aacrjournals.org
infection by scraping, after which they were spinned and permeabilized in
FACSPerm2 solution (Becton Dickinson). Following a PBS wash, cells were
incubated in the presence of FITC-hexon mAb (1:10 dilution) for 30 min at
room temperature in the dark, washed again, and analyzed on a Cytomics
FC-500 flow cytometer.
Methylene blue staining of infected cells. Half a million cells were
infected using virus supernatant as described above. After infection, cells
were seeded in gelatin-coated (0.1% gelatin in PBS) six-well plates in the
presence of 2 mL 1.25% agar in DMEM culture medium. Cells were stained
with methylene blue 5 days (after infection) and photographed (LEICA
DMIL inverted microscope).
Expression and localization of adenoviral proteins. A quarter of a
million cells were infected using virus supernatant (at virus particle-to-cell
ratio of 50) as described above. After infection, cells were seeded in 24-well
plates and cultured for the indicated times. Cells were carefully washed with
PBS and fixed with a 1:1 solution of ice-cold methanol and acetone for
10 min on ice. After repeated washing steps with PBS, cells were blocked
using 1% bovine serum albumin in PBS for 30 min at room temperature.
Cells were then shortly air dried and stained with primary and secondary
antibodies (diluted in blocking buffer). Kinetics of expression and cellular
localization of fluorescently labeled adenoviral proteins were monitored
(LEICA DMIL inverted fluorescence microscope).
Results
Construction of a replication-competent adenovirus with
HLA-A1/MAGE-A1 specificity. A knobless fiber containing fiber
tail plus NH
2
-terminal first shaft repeat (R1, 61 amino acids), an
extrinsic trimerization motif (NRP, 36 amino acids) from lung
surfactant protein D, a linker derived from Staphylococcus protein A
(13 amino acids), and scTCR VaVhCh (377 amino acids), specific
for the melanoma antigen MAGE-A1, presented by HLA-A1 (which
replaced the natural fiber knob), was constructed (Fig. 1) and intro-
duced into replication-competent adenovirus serotype 5 essentially
as described (17, 18). To construct the scTCR, TCR a and h chains
were cloned from an HLA-A1–restricted, MAGE-A1–specific CTL
clone, MZ2-82/30, and reformatted into the scTCR VaVhCh as
described (30). Specific binding of the scTCR was verified by
expression on primary human T lymphocytes, which showed
scTCR-directed immune functions such as specific tumor cell kill
and cytokine production. The apparent molecular weight of the R1-
scTCR fiber (54 kDa) is similar to that of the WT Ad5 fiber (59 kDa).
Generation of 293T-AdR cells to propagate fiber-modified
adenoviruses. As a consequence of ablating the natural tropism,
we expected that recombinant Ad5-scTCR virus would require the
presence of its target antigen, HLA-A1/MAGE-A1, on the surface of
the producer cell line for the primary attachment and entry.
Therefore, we generated 293T cells expressing the HLA-A1/MAGE-
A1 antigen. 293T cells were infected with retroviral vectors pBullet
HLA-A1 and pBullet MAGE-A1 ( full-length cDNA and minigene).
Next to these antigen
+
293T cells, we also used MZ2-mel 3.0
melanoma cells, which naturally present MAGE-A1 in the context
of HLA-A1. Neither the antigen-transduced 293T cells nor MZ2-mel
3.0 cells were able to initiate production of the recombinant virus
starting with transfection of the adenoviral DNA (data not shown).
To support initiation of viral production and propagation of fiber-
modified adenoviruses that depend neither on CAR nor on MAGE-
A1/HLA-A1 antigen (or any ligands of interest for that matter),
we generated a novel producer cell line based on the introduction
of an adenovirus-binding antibody into 293T cells.
To this end, we constructed two membrane-anchored anti-
adenovirus receptors, AdR and AdR-cMyc/~, from hybridoma cells
producing a hexon-specific antibody (Fig. 2A) that cross-reacts
with many adenovirus subtypes and introduced it via retroviral
transduction into 293T cells. Due to a lack of antibodies binding to
the scFv directly, demonstration of cell surface expression of the
anti-adenovirus receptors on 293T cells was only possible for AdR-
cMyc/~ using anti-c-Myc mAb (Fig. 2B).
The ability of AdR and AdR-cMyc/~ to serve as universal
receptors for Ad5 was analyzed in 293T cells. 293T cells with AdR,
termed 293T-AdR, 293T cells with AdR-cMyc/~, termed 293T-AdR-
cMyc/~, or parental 293T cells were transfected with a fiberless Ad5
vector encoding the EGFP gene. We observed a severely impaired
propagation of fiberless adenovirus in 293T cells in line with
previous reports and most likely due to a lack of CAR-fiber knob
interactions (21, 31). We hypothesized that the presence of the
AdR receptor would at least in part restore the ability of 293T cells
to produce fiber-deleted viruses. As shown in Fig. 2C,onday1
following transfection, the expression of EGFP was comparable in
both 293T and 293T-AdR cells. The ratio of EGFP in normal 293T
cells did not improve on day 2 or 3. However, in 293T-AdR cells, we
observed a robust spread of the reporter gene together with comet-
like formation that was most significant on day 3 after transfection.
Production of virus particles in culture supernatant was confirmed
by an adenovirus-specific ELISA (data not shown). When 293T cells
were stably expressing the AdR-cMyc/~ receptor, we also observed
an increase in reporter gene expression and release of viral
particles. However, the ability of 293T-AdR-cMyc/~ to induce
adenovirus production was significantly less than that of 293T-AdR
cells (data not shown).
Figure 1. Diagram of WT Ad5 fiber and
R1-scTCR fiber. Distinct domains of the
Ad5 WT fiber and R1-scTCR fiber as well
as the amino acid (a.a.) composition of the
distinct domains. Va, variable domain of
TCRa chain; Vb, variable domain of TCRh
chain; Cb, constant domain of TCRh chain.
TCR-Redirected Tumor-Specific Adenovirus
www.aacrjournals.org
11311
Cancer Res 2007; 67: (23). December 1, 2007
We then introduced the Ad5.R1-scTCR construct into 293T-
AdR cells and showed that, 3 days following transfection, virus
was produced at a titer of 6
10
7
particles/mL (= physical
particles, determined by ELISA), starting from 3
10
6
293T-AdR
cells.
Ad5.R1-scTCR virus is produced, specifically binds to HLA-
A1/MAGE-A1 complexes, and replicates in HLA-A1
+
/MAGE-
A1
+
tumor cells. To show production, specific binding to
peptide/MHC complexes, and fiber incorporation of adenoviral
particles that incorporate the chimeric R1-scTCR fiber, we did
the following experiments: (a) electron microscopy, to show
presence of viral particles in MZ2-mel 3.0 cells (Fig. 3A); (b) flow
cytometry analysis of HLA-A1/MAGE-A1–specific binding of Ad5-
R1-scTCR (Fig. 3B); (c) Western blot analysis, to show
incorporation of the R1-scTCR fiber (Fig. 3C); and (d) ELISA,
to show production of Ad5.R1-scTCR in 293T AdR and MZ2-mel
3.0 cells.
To show production of Ad5.R1-scTCR particles, MZ-2-mel 3.0
cells were infected with viral supernatant and analyzed by electron
microscopy. Figure 3A shows the presence of viral particles in the
nucleus of MZ2-mel 3.0 cells 72 h after infection.
To show specific binding to HLA-A1/MAGE-A1, supernatant
from 293T-AdR cells producing Ad5-R1-scTCR was incubated with
magnetic beads that were loaded with HLA-A1/MAGE-A1 com-
plexes or HLA-A1 complexes that present an irrelevant peptide
derived from influenza virus A nucleoprotein.
Figure 2. A, diagram of an expression
vectors encoding the ‘‘adenoreceptor.’’ SS,
Ignsignal sequence; c, intracellular domain
of FcqR1-g chain. B, surface expression
of AdR-cMyc/~ on 293T cells. 293T
cells were retrovirally transduced with
adenoreceptor fused to a c-Myc tag
(AdR-cMyc/~). Cells were stained with
FITC-conjugated c-Myc–specific mAb
(9E10) and samples were measured by
flow cytometry. Histograms represent
nontransduced 293T cells (thin line) and
receptor-positive 293T cells (thick line).
C, infection of 293T cells by fiberless
adenovirus requires the expression of AdR
on the cell surface. 293T and 293T-AdR
cells were seeded and transfected with
fiberless pAdeasy EGFP construct.
Kinetics of green fluorescent protein
expression was monitored by fluorescence
microscope at days 1, 2, and 3 after
transfection. Representative images
(
10 magnification) from one of three
experiments.
Cancer Research
Cancer Res 2007; 67: (23). December 1, 2007
11312
www.aacrjournals.org
As shown, Ad5.R1-scTCR virus only bound to HLA-A1/MAGE-A1
complexes and not to HLA-A1 complexes presenting an irrelevant
influenza virus peptide (Fig. 3B).
Ad5.R1-scTCR virus bound to HLA-A1/MAGE-A1–coated mag-
netic beads was then analyzed by Western blotting using fiber
tail–specific mAb 4D2. As shown in Fig. 3C, chimeric scTCR fibers
were incorporated into adenoviral particles.
To determine whether Ad5.R1-scTCR virus is able to infect HLA-
A1
+
/MAGE-A1
+
tumor cells, we incubated MZ2-mel 3.0 melanoma
cells with supernatant obtained from 293T AdR cells transfected
with either Ad5.WT or Ad5.R1-scTCR DNA. MZ2-mel 3.0 cells lack
CAR expression (Table 1), making them refractory to infection by
WT virus. As shown in Fig. 3D, Ad5.R1-scTCR virus produced by
293T AdR cells infected MZ2-mel 3.0 cells and was able to replicate
Figure 3. Characterization of Ad5.R1-
scTCR particles. A, electron microscopic
image of Ad5.R1-scTCR in MZ2-mel
3.0 cells. Presence of Ad5.R1-scTCR
particles in the nucleus of MZ2-mel
3.0 cells was shown by electron
microscopy 72 h after incubation of
MZ2-mel 3.0 cells with viral supernatant.
Right, a five times enlargement of a region
of the left. B, Ad5.R1-scTCR particles from
293T-AdR supernatant specifically bind
to HLA-A1/MAGE-A1 complexes only.
Tissue culture supernatants derived from
293T-AdR cells producing Ad5.R1-scTCR
particles were incubated with HLA-A1/
MAGE-A1 or irrelevant HLA-A1/Flu
complex-coated magnetic beads. Ad5.R1-
scTCR particles were detected by flow
cytometric analysis using anti-hexon
FITC
mAb. C, Ad5.R1-scTCR viral particles
incorporate the R1-scTCR fiber. Ad5.R1-
scTCR virus bound to HLA-A1/MAGE-A1
complex-coated beads and cesium
chloride–purified WT Ad5 were loaded
on 7% SDS-PAGE, transferred to
nitrocellulose membrane, and detected
with anti-fiber tail antibody 4D2. Lane 1,
Ad5.WT; lane 2, Ad5.R1-scTCR.
D, adenovirus expressing the scTCR fiber
infects and replicates in HLA-A1/MAGE-
A1
+
melanoma cells. MZ2-mel 3.0 cells
were incubated with viral supernatant
obtained from 293T AdR cells transfected
with either Ad5.R1-scTCR or Ad5.WT
DNA. Adenoviral titer in tissue culture
medium was detected by ELISA after
(a) transfection of the producer cell
293T-AdR with Ad5.WT (black columns)
or Ad5.R1-scTCR (gray columns) and (b )
3 d after infection of MZ2-mel 3.0 cells
by crude lysates from a.
TCR-Redirected Tumor-Specific Adenovirus
www.aacrjournals.org
11313
Cancer Res 2007; 67: (23). December 1, 2007
in these cells, shown by the presence of viral particles in the tissue
culture supernatant 3 days after infection. In contrast, WT virus at
comparable virus particle-to-cell ratio did not result in adenoviral
infection.
Cellular localization of Ad5.R1-scTCR epitopes during
replication. It has been suggested that only those recombinant
fibers that assemble correctly in the nucleus may be incorporated
into an infectious adenoviral particle (24). To analyze intracellular
localization of adenoviral proteins, we infected MZ2-mel 3.0 cells
with Ad5.R1-scTCR and did intracellular immunofluorescent
staining with mAbs specific for hexon, fiber tail, and scTCR. Hexon
and fiber tail molecules showed a comparable cellular distribution,
localized almost exclusively to the nucleus 24 h after infection
(Fig. 4, top). Interestingly, at that time point, hexon molecules were
detected in the cytoplasm to some extent, whereas fiber molecules
were not. In control experiments, we observed that adenoviruses
displaying WT fiber showed similar cellular localization of hexon
and fiber tail (data not shown). In contrast, staining for the R1-
scTCR fiber with the anti-TCR-specific antibody did not result in
any detectable fluorescent signal at 24 h after infection (Fig. 4, top).
Localization of both hexon and fiber proteins changed at
72 h after infection from a prominent nuclear localization to
accumulation at the cell periphery, most likely at the plasma
membrane. At this stage of replication, presumably on virus release,
R1-scTCR could be detected with anti-TCR antibody and also
located to the plasma membrane or to its proximity, indicating an
identical cellular compartmentalization of hexon, fiber tail, and
scTCR (Fig. 4, bottom).
Infection by Ad5.R1-scTCR is epitope specific. Specificity of
infection of Ad5.R1-scTCR was analyzed by infecting the melanoma
cells: MZ2-mel 3.0 (HLA-A1
+
/MAGE-A1
+
) and MEL.2A (HLA-A1
+
/
MAGE-A1
). Also included were 293T and 293T-AdR cells. Target
cells were infected at different virus particle-to-cell ratios and
monitoring the production of hexon protein at 2 days after
infection. In this assay, cells expressing the hexon molecule
represent infected cells and constitute an indirect readout for viral
titers as an alternative to plaque assay. As shown in Fig. 5, Ad5.R1-
scTCR virus reached maximum infectivity at f20 virus particle-to-
cell ratio when infecting antigen
+
MZ2-mel 3.0 cells and 293T-AdR,
approximately corresponding to a MOI of 4. When using antigen
MZ2-mel 2.2 cells, Ad5.R1-scTCR infectivity remained low even at
high virus particle-to-cell ratio and showed a similar titration curve
when using 293T cells.
Specificity studies were expanded by the use of a larger panel of
target cells, including the following melanoma cell lines: MZ2-mel
3.0; 9303-A; 518-A2; MZ2-mel 2.2, a MAGE-A1 antigen lost mutant
obtained from MZ2-mel 3.0; MEL.2A; and FM-3. We also included
Nemeth renal cell carcinoma cell lines and the 293T cells. Five days
after infection, surviving tumor cells were stained with methylene
blue. HLA-A1/MAGE-A1 expression as well as infectivity data of all
target cells are summarized in Table 1. As shown in Table 1 only
HLA-A1
+
/MAGE-A1
+
melanoma cells were infected.
Discussion
Our aim of this study was to provide oncolytic adenoviral vectors
with a truly tumor cell specificity to improve safety and efficacy,
which are major criteria for the clinical use of replicating vectors.
TCR recognizing MHC-restricted cancer-testis antigens, such as
MAGE-A1, which are distributed in a highly tumor tissue–specific
manner, may constitute promising molecules to retarget adenovi-
ruses. This study shows for the first time that even complex
molecules such as TCR can genetically replace the fiber knob and
be expressed on the adenoviral fiber, thereby ablating its natural
tropism. The Ad5 fiber knob was replaced by an extrinsic
trimerization motif and an HLA-A1/MAGE-A1–specific scTCR.
Critical to the experimental use of this recombinant adenovirus
was the generation of a novel producer cell line, 293T-AdR,
which was successfully used to initiate production, starting by
transfection with adenoviral DNA, and supported propagation of
adenovirus, which depended neither on CAR nor on HLA-A1/
MAGE-A1. Our failure to use 293T expressing the HLA-A1 and
MAGE-A1 cDNA for the initiation of adenovirus production was
most likely attributed to unstable expression of the HLA-A1 gene.
Figure 4. Localization of adenoviral
proteins during replication. MZ2-mel 3.0
cells were infected with Ad5.R1-scTCR
at virus particle-to-cell ratio of 50, and
cell-associated adenoviral proteins
were stained with mAbs against hexon
(a-hexon), fiber tail (4D2.5), and scTCR Va
domain (Va12.1) at 24 and 72 h after
infection. Images were collected with
fluorescent microscope.
Cancer Research
Cancer Res 2007; 67: (23). December 1, 2007
11314
www.aacrjournals.org
Within 3 days after infection with retrovirus encoding the HLA-A1
gene, HLA-A1 molecules disappeared from the cell surface,
resulting in lack of MAGE-A1 antigen presentation to the chimeric
fibers (data not shown). There may be more reasons why we were
unable to initiate adenovirus production in MZ2-mel 3.0 cells. In
general, the production of adenovirus after DNA transfection is
inefficient because plasmid DNA does not contain the protein
binding to the viral inverted terminal repeats and therefore
requires several rounds of amplification (32). When DNA
transfection efficiencies become limiting, as might be the case
when using MZ2-mel 3.0 cells, low numbers of virus-producing
cells may result in undetectable levels of virus even after serial
amplification. In addition, MZ2-mel 3.0 cells may produce lower
numbers of viral particles than, for example, 293 cells.
Importantly, the retargeted viruses specifically bound to relevant
HLA-A1/MAGE-A1 complexes only (Fig. 3B), specifically infected
target cells expressing HLA-A1–restricted MAGE-A1 antigen (Fig. 5;
Table 1), and killed these melanoma cells.
The initiation of production when starting from recombinant
Ad5 DNA seemed to be a critical step during propagation. As a
consequence of ablation of native Ad5 tropism, the recombinant
Ad5.R1-scTCR is not able to use the natural CAR-mediated
cellular entry pathway during propagation in conventional
packaging cell lines, such as 293 or 911. Although there are
reports on possible solutions to overcome the limitations of fiber-
modified adenoviral vector production (31, 33), there is no
precedent on virus retargeting via complex molecules, such as
TCR or Igs, and an alternative strategy had to be developed. A
major factor that can hamper genetic retargeting of adenoviral
vectors, especially when including complex molecules into the
viral genome, is the proper folding of these new molecules in the
nucleus. According to Pecorari et al. (34), the formation of
intrachain disulfide bridges, which are crucial for the correct
folding and stability of IgG or TCR, is suboptimal in the reducing
environment of the cytoplasm and nucleus. One could address
this issue by using small molecules, such as affibodies, which
possess the binding properties of an antibody but do not require
intrachain disulfide bounds (16). Magnusson et al. (18) reported
that, although the scTCR fiber was able to form homotrimers and
bound its ligand, the recombinant fiber protein misfolded, thereby
possibly explaining the unsuccessful propagation of Ad5.R1-scTCR
in MZ2-mel 3.0 melanoma cells. However, our studies on scTCR
fiber expression suggest that at the proximity to the cell
membrane (Fig. 4, bottom) proper folding of the scTCR fiber
occurs, resulting in exposure of a fully functional scTCR fiber on
the virus particle. We assume that at this stage of virus assembly
restricting intracellular conditions no longer limits the formation
of sulfide bridges. Furthermore, we now also succeeded in the
initiation of production of two other recombinant viruses, one
with a scTCR fiber and a virus equipped with an affibody fiber
(data not shown), which shows the universal applicability of this
production system.
These findings open new and safer strategies for cell-specific
retargeting of oncolytic adenoviruses, providing a versatile tool for
future clinical application.
Acknowledgments
Received 3/6/2007; revised 8/15/2007; accepted 9/25/2007.
Grant support: European Union grant QLK3-1999-01262.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Prof. Rob Hoeben (Virus and Stem Cell Biology Laboratory, Department
of Molecular Cell Biology, Leiden University Medical Center, Leiden, the Netherlands)
for helpful suggestions and critical reading of the manuscript, Mirjam Heuveling for
technical assistance, and Ronald Limpens and Mieke Mommaas-Kienhuis for doing
the electron microscopic analysis (Section Electron Microscopy, Leiden University
Medical Center).
Figure 5. Ad5.R1-scTCR specifically
infects melanoma cells expressing the
HLA-A1/MAGE-A1 epitope at low virus
particle-to-cell ratio. Cells (5
10
5
)of
MZ2-mel 3.0 (solid line with black box ),
MEL.2A (solid thin line), 293T (solid line
with triangle ), and 293T-AdR (dashed line)
were infected at various virus particle-to-
cell ratios of Ad5.R1-scTCR. Cells were
harvested 2 d after infection and stained
with FITC-labeled mAb against hexon
molecule. Percentage of hexon-expressing
cells analyzed by flow cytometry.
References
1. Zinkernagel RM, Doherty PC. The discovery of MHC
restriction. Immunol Today 1997;18:14–7.
2. van der BP, Traversari C, Chomez P, et al. A gene
encoding an antigen recognized by cytolytic T lympho-
cytes on a human melanoma. Science 1991;254:1643–7.
3. Boon T, Coulie PG, Van den Eynde B. Tumor antigens
recognized by T cells. Immunol Today 1997;18:267–8.
4. Schultz ES, Schuler-Thurner B, Stroobant V, et al.
Functional analysis of tumor-specific Th cell responses
detected in melanoma patients after dendritic cell-
based immunotherapy. J Immunol 2004;172:1304–10.
5. Couly G, Creuzet S, Bennaceur S, Vincent C, Le
Douarin NM. Interactions between Hox-negative ce-
phalic neural crest cells and the foregut endoderm in
patterning the facial skeleton in the vertebrate head.
Development 2002;129:1061–73.
6. Lonchay C, van der Bruggen P, Connerotte T, et al.
Correlation between tumor regression and T cell
responses in melanoma patients vaccinated with a
MAGE antigen. Proc Natl Acad Sci U S A 2004;101 Suppl
2:14631–8.
7. Scanlan MJ, Simpson AJ, Old LJ. The cancer/testis
genes: review, standardization, and commentary. Can-
cer Immun 2004;4:1.
8. Renkvist N, Castelli C, Robbins PF, Parmiani G. A
TCR-Redirected Tumor-Specific Adenovirus
www.aacrjournals.org
11315
Cancer Res 2007; 67: (23). December 1, 2007
Cancer Research
Cancer Res 2007; 67: (23). December 1, 2007
11316
www.aacrjournals.org
listing of human tumor antigens recognized by T cells.
Cancer Immunol Immunother 2001;50:3–15.
9. Nemunaitis J, Khuri F, Ganly I, et al. Phase II trial of
intratumoral administration of ONYX-015, a replication-
selective adenovirus, in patients with refractory head
and neck cancer. J Clin Oncol 2001;19:289–98.
10. Kanerva A, Hemminki A. Modified adenoviruses for
cancer gene therapy. Int J Cancer 2004;110:475–80.
11. Kanerva A, Wang M, Bauerschmitz GJ, et al. Gene
transfer to ovarian cancer versus normal tissues with
fiber-modified adenoviruses. Mol Ther 2002;5:695–704.
12. Vellinga J, Rabelink MJWE, Cramer SJ, et al. Spacers
increase the accessibility of peptide ligands linked to the
carboxyl terminus of adenovirus minor capsid protein
IX. J Virol 2004;78:3470–9.
13. Dmitriev IP, Kashentseva EA, Curiel DT. Engineering
of adenovirus vectors containing heterologous peptide
sequences in the C terminus of capsid protein IX. J Virol
2002;76:6893–9.
14. Gaden F, Franqueville L, Magnusson MK, et al. Gene
transduction and cell entry pathway of fiber-modified
adenovirus type 5 vectors carrying novel endocytic
peptide ligands selected on human tracheal glandular
cells. J Virol 2004;78:7227–47.
15. Henning P, Andersson KME, Frykholm K, et al.
Tumor cell targeted gene delivery by adenovirus 5
vectors carrying knobless fibers with antibody-binding
domains. Gene Ther 2005;12:211–24.
16. Hong SS, Magnusson MK, Henning P, Lindholm L,
Boulanger PA. Adenovirus stripping: a versatile method
to generate adenovirus vectors with new cell target
specificity. Mol Ther 2003;7:692–9.
17. Magnusson MK, Hong SS, Boulanger P, Lindholm L.
Genetic retargeting of adenovirus: novel strategy
employing ‘‘deknobbing’’ of the fiber. J Virol 2001;75:
7280–9.
18. Magnusson MK, Hong SS, Henning P, Boulanger P,
Lindholm L. Genetic retargeting of adenovirus vectors:
functionality of targeting ligands and their influence on
virus viability. J Gene Med 2002;4:356–70.
19. Nilsson M, Ljungberg J, Richter J, et al. Development
of an adenoviral vector system with adenovirus
serotype 35 tropism; efficient transient gene transfer
into primary malignant hematopoietic cells. J Gene
Med 2004;6:631–41.
20. Krasnykh VN, Mikheeva GV, Douglas JT, Curiel DT.
Generation of recombinant adenovirus vectors with
modified fibers for altering viral tropism. J Virol 1996;70:
6839–46.
21. Wickham TJ, Tzeng E, Shears LL, et al. Increased
in vitro and in vivo gene transfer by adenovirus vectors
containing chimeric fiber proteins. J Virol 1997;71:
8221–9.
22. Jakubczak JL, Rollence ML, Stewart DA, et al.
Adenovirus type 5 viral particles pseudotyped with
mutagenized fiber proteins show diminished infectivity
of coxsackie B-adenovirus receptor-bearing cells. J Virol
2001;75:2972–81.
23. Kirby I, Davison E, Beavil AJ, et al. Mutations in the
DG loop of adenovirus type 5 fiber knob protein abolish
high-affinity binding to its cellular receptor CAR. J Virol
1999;73:9508–14.
24. Borovjagin AV, Krendelchtchikov A, Ramesh N, Yu
DC, Douglas JT, Curiel DT. Complex mosaicism is a
novel approach to infectivity enhancement of adeno-
virus type 5-based vectors. Cancer Gene Ther 2005;12:
475–86.
25. Willemsen RA, Debets R, Chames P, Bolhuis RLH.
Genetic engineering of T cell specificity for immuno-
therapy of cancer. Hum Immunol 2003;64:56–68.
26. Peng KW, Holler PD, Orr BA, Kranz DM, Russell SJ.
Targeting virus entry and membrane fusion through
specific peptide/MHC complexes using a high-affinity
T-cell receptor. Gene Ther 2004;11:1234–9.
27. Willemsen RA, Debets R, Hart E, Hoogenboom HR,
Bolhuis RL, Chames P. A phage display selected fab
fragment with MHC class I-restricted specificity for
MAGE-A1 allows for retargeting of primary human T
lymphocytes. Gene Ther 2001;8:1601–8.
28. Turatti F, Figini M, Alberti P, Willemsen RA, Canevari
S, Mezzanzanica D. Highly efficient redirected anti-
tumor activity of human lymphocytes transduced with a
completely human chimeric immune receptor. J Gene
Med 2005;7:158–70.
29. Chames P, Willemsen RA, Rojas G, et al. TCR-like
human antibodies expressed on human CTLs mediate
antibody affinity-dependent cytolytic activity. J Immu-
nol 2002;169:1110–8.
30. Willemsen RA, Weijtens ME, Ronteltap C, et al.
Grafting primary human T lymphocytes with cancer-
specific chimeric single chain and two chain TCR. Gene
Ther 2000;7:1369–77.
31. Von Seggern DJ, Kehler J, Endo RI, Nemerow GR.
Complementation of a fibre mutant adenovirus by
packaging cell lines stably expressing the adenovirus
type 5 fibre protein. J Gen Virol 1998;79:1461–8.
32. van Bergen BG, van der Ley PA, van Driel W, van
Mansfeld AD, van der Vliet PC. Replication of origin
containing adenovirus DNA fragments that do not carry
the terminal protein. Nucleic Acids Res 1983;11:1975–89.
33. Douglas JT, Miller CR, Kim M, et al. A system for the
propagation of adenoviral vectors with genetically
modified receptor specificities. Nat Biotechnol 1999;17:
470–5.
34. Pecorari F, Tissot AC, Pluckthun A. Folding, hetero-
dimeric association and specific peptide recognition of a
murine ah T-cell receptor expressed in Escherichia coli.
J Mol Biol 1999;285:1831–43.