Int. J.

Cancer: 108, 136 –145 (2004) Publication of the International Union Against Cancer

© 2003 Wiley-Liss, Inc.

RETARGETING OF ADENOVIRAL INFECTION TO MELANOMA: COMBINING

GENETIC ABLATION OF NATIVE TROPISM WITH A RECOMBINANT
BISPECIFIC SINGLE-CHAIN DIABODY (scDb) ADAPTER THAT BINDS TO
FIBER KNOB AND HMWMAA
Dirk M. NETTELBECK1,*, Angel A. RIVERA1, Jörg KUPSCH2, Detlef DIECKMANN3, Joanne T. DOUGLAS1, Roland E. KONTERMANN4,
Ramon ALEMANY5 and David T. CURIEL1
1
Division of Human Gene Therapy, Departments of Medicine, Pathology and Surgery, and the Gene Therapy Center, University of
Alabama at Birmingham, Birmingham, AL, USA
2
RAFT Institute, Mount Vernon Hospital, Northwood, United Kingdom
3
Department of Dermatology, University of Erlangen-Nuremberg, Erlangen, Germany
4
Institute of Molecular Biology and Tumor Research, Philipps-University of Marburg, Marburg, Germany
5
Gene Therapy Unit, Institut Català d’Oncologia, L’Hospitalet Barcelona, Spain

Gene therapy is an emerging and promising modality for vectors possess critical properties required for this endeavor. These
the treatment of malignant melanoma and other neoplasms include a highly evolved gene transfer mechanism, the stability of
for which conventional therapies are inadequate. Various virus particles and the ease of virus production at high titers.6 Most
therapeutic genes have shown promise for tumor cell killing.
However, successful gene therapy depends on the develop- importantly, the capsid structure, genome and replication cycle of
ment of efficient and targeted gene transfer vectors. Here we adenoviruses, particularly of the most commonly employed sero-
describe a novel strategy for targeting of adenovirus-medi- type 5, have been extensively characterized, which allows for the
ated gene transfer to melanoma cells. This strategy com- molecular modifications required for their utilization as gene trans-
bines genetic ablation of native adenoviral tropism with re- fer vectors. The necessity of such modifications is predicated by
directed viral binding to melanoma cells via a bispecific the observation that the primary receptor for Ad5 and other ade-
adapter molecule, a bacterially expressed single-chain dia-
body, scDb MelAd, that binds to both the adenoviral fiber novirus serotypes, CAR is widely expressed on normal tissues
protein and to the high molecular weight melanoma-associ- resulting in nonspecific susceptibility to adenoviral infection. In
ated antigen (HMWMAA). This antigen is widely and specif- addition, reduced or absent expression of CAR has been reported
ically expressed on the surface of melanoma cells and its previously for several tumor types, including melanoma, indicat-
expression is associated with tumor development and pro- ing resistance to adenoviral infection by tumor cells in situ.7–9.
gression. Our results showed specific and strong binding of These considerations of adenoviral biology are paralleled by the
the anti-HMWMAA scFv RAFT3 and the bispecific adapter
scDb MelAd to melanoma cells. In adenoviral infection ex- observation of limited efficacy and vector-related toxicity in pre-
periments, we demonstrated i) substantially (>50-fold) re- clinical and clinical adenoviral gene therapy studies. Therefore, the
duced infectivity of capsid mutant adenoviruses, ii) restored development of tropism-modified, tumor-targeted adenoviral gene
(up to 367-fold increase), CAR-independent and HMWMAA- transfer vectors is a key endeavor in current gene therapy research.
mediated infectivity of these mutant viruses by scDb MelAd Towards this goal, the native tropism of adenoviruses needs to be
specifically in melanoma cells, and iii) higher levels of trans-
gene expression in melanoma cells by fiber mutant virus ablated (infectivity ablation) and a new, tumor-specific tropism
complexed with scDbMelAd, relative to a vector with wild- needs to be engineered into viral particles (retargeting).
type fibers. We confirmed the utility of this targeting strat-
egy with human primary melanoma cells that represent clin-
ically relevant substrates. These experiments established
Abbreviations: Ad5, adenovirus serotype 5; bp, base pairs; CAR, cox-
that the retargeting strategy mediates up to 54-fold in- sackie-adenovirus receptor; EGF, epidermal growth factor; FGF, fibroblast
creased adenoviral gene transfer to CAR-negative melanoma growth factor; HMWMAA, high molecular weight melanoma-associated
cells compared to the vector with native tropism. Hence, the antigen; IGF, insulin-like growth factor; IMAC, immobilized metal affinity
HMWMAA-targeted adenoviral vector lacking native tro- chromatography; MAb, monoclonal antibody; PAGE, polyacrylamide gel
pism exhibits both enhanced specificity and augmented in- electrophoresis; pfu, plaque-forming units; scDb, single-chain diabody;
fectivity of gene transfer to melanoma cells, suggesting that scFv single-chain Fv antibody fragment; sCAR, extracellular domain of
it is feasible to use this vector to improve gene therapy for CAR; TNF␣, tumor necrosis factor alpha; vp, viral particles.
malignant melanoma.
© 2003 Wiley-Liss, Inc.
Grant sponsor: Deutsche Forschungsgemeinschaft; Grant numbers;
Key words: adenovirus targeting; melanoma, single-chain diabody; NE832/1; Grant sponsor: National Institutes of Health; Grant numbers:
high molecular weight melanoma-associated antigen; CAR/integrin- R01 HL67962, P50 CA89019, R01 CA86881, U19 DK57958; Grant spon-
binding ablation sor: CFAR AIDS; Grant number: P30AI27767; Grant sponsor: UAB
Center for AIDS Research; Grant number: P30 A1-27767

Gene therapy is a promising new strategy for treatment of

cancer where conventional therapeutic regimens are often in- *Correspondence to: Department of Dermatology, University of
Erlangen-Nuremberg, Hartmannstrasse 14, 91052 Erlangen, Germany.
adequate. Proof-of-principle has been clearly established for Fax: ⫹49-9131-853-6417.
therapeutic gene transfer aiming at direct tumor cell killing, E-mail: dirk.nettelbeck@derma.imed.uni-erlangen.de
prodrug-activation by tumor cells, mutation compensation, im-
munopotentiation and viral oncolysis following the transfer of
genomes of replication competent viruses to tumor cells.1,2 How- Received 31 March 2003; Revised 30 May 2003; Accepted 4 August
ever, the application of gene therapy in patients is hampered by 2003
insufficient gene transfer efficiencies and by toxic effects mediated
by gene transfer to normal tissues.3 Thus, the development of DOI 10.1002/ijc.11563
efficient and targeted gene transfer vectors is key to the successful Published online 7 October 2003 in Wiley InterScience (www.
clinical translation of gene therapy.4,5 In this regard, adenoviral interscience.wiley.com).
 ScDb FOR RETARGETING OF ADENOVIRUS INFECTION TO MELANOMA 137
Malignant melanoma is characterized by rapidly growing inci- an HMWMAA-binding scFv. Importantly, HMWMAA-binding
dence and mortality rates, metastasis at an early stage of tumor scFvs have been described by us and others.17,39 – 43 Like anti-
development and resistance of metastatic disease to current ther- HMWMAA monoclonal antibodies, these scFvs have been em-
apies. These observations underline the critical need for novel ployed in diagnostic and targeted therapeutic studies,39,43– 45 in-
therapeutic strategies, such as gene therapy, to treat malignant cluding targeting of retroviral vectors.46 For the adenovirus
melanoma. In this regard, the above-mentioned effector strategies targeting strategy developed in our study, we applied a HMW-
have proved effective in initial in vitro and locoregional melanoma MAA-binding scFv, RAFT3, that was optimized for improved
gene therapy studies.10 Nevertheless, tumor-specific gene transfer melanoma binding and strong bacterial expression. Furthermore,
is required to adopt these strategies for disseminated disease. The RAFT3 was successfully targeted to malignant melanoma in
definition of a suitable target molecule and ligand is of key vivo.39
importance for any specific therapeutic intervention. Clearly, the Our results demonstrate that the developed adenovirus retarget-
high-molecular-weight melanoma-associated-antigen (HMW- ing strategy results in selective and enhanced adenoviral infection
MAA) is an attractive target for melanoma-directed gene therapy of melanoma cells. Infectivity of the tropism-modified adenovi-
because of its specific, strong and consistent expression in malig- ruses was mediated by HMWMAA, independent of the adenovirus
nant melanoma and its functional association with tumor progres- receptor CAR, and did not require penton base-binding to integrins
sion.11–17 Because of these favorable attributes, HMWMAA-bind- for virus internalization. Infectivity enhancement was especially
ing monoclonal antibodies have been widely employed for prominent in primary melanoma cells that were devoid of CAR.
diagnostics or targeted therapeutics in preclinical and clinical This suggests that the developed strategy for retargeting of adeno-
studies.18 –24 These considerations clearly established HMWMAA viral infection represents a means to improve gene therapy for
as a promising cell-surface target for a melanoma-directed, tro- malignant melanoma.
pism-modified adenovirus.
The prime objective of the present study was the development of
a new strategy for targeted and enhanced adenoviral gene transfer MATERIAL AND METHODS
to malignant melanoma. The native tropism of adenoviruses is Cell culture
determined by their capsid proteins, fiber and penton base. Ad5 Human tumor cell lines SK-MEL-28 [melanoma, American
binds to CAR via the protruding knob domain of the fiber.25 Viral Type Culture Collection (ATCC), Manassas, VA], MeWo (mela-
internalization is triggered subsequently by the interaction of the noma, kindly provided by Dr. I. Hart, London, UK), A549 (lung
viral penton base with cellular integrins.26 To date, tropism mod- adenocarcinoma, ATCC) and MCF-7 (breast cancer, ATCC) were
ification of adenoviruses has been achieved by genetic capsid cultivated in DMEM (Mediatech, Herndon, VA). Human mela-
engineering or with bispecific adapter molecules.9 However, most noma cell lines Mel888 (kindly provided by Dr. J. Schlom, Be-
genetic strategies have resulted in expanded tropism but not in thesda, MD), A375M (kindly provided by Dr. I.J. Fidler, Houston,
targeting of adenoviral infection to individual tissue types. Partic- TX) and the mouse melanoma cell line B16F10 (ATCC) were
ularly, the genetic incorporation of scFvs into adenoviral capsids cultivated in RPMI1640 (Mediatech). 293 cells (purchased from
remains a major obstacle because of structural constraints. In our Microbix, Toronto, Canada) and OV-4 cells (kindly provided by
study, we describe a strategy for adenoviral capsid modification Dr. T.J. Eberlein, Boston, MA) were grown in DMEM/F12 (50:50;
that combines genetic ablation of viral cell binding with a bispe- Mediatech). 293.HissFv.rec cells47 were propagated in DMEM/
cific adapter molecule for retargeting of infection to melanoma F12 with 500 ␮g/ml G418. All media were supplemented with
cells. Mutations of adenoviral capsid proteins fiber and penton 10% fetal bovine serum (FBS, HyClone, Logan, UT), 2 mM
base that prevent binding to the cellular receptors have been L-glutamine, 100 I.U./ml penicillin, and 100 ␮g/ml streptomycin
described recently27–29 and we exploited corresponding virus mu- (all Mediatech). Primary Melanoma cells were obtained from
tants for the targeting strategy investigated in our study. In addi- surgically removed metastasis. Tissue samples were dissected free
tion, we engineered a recombinant bispecific single-chain diabody of fat and epidermis and incubated in antibiotic/antimycotic solu-
as a targeting adapter for specific transduction of melanoma cells. tion (Life Technologies, Eggenstein, Germany) for 20 min at room
The concept of targeting adenovirus infection with bispecific temperature. Afterwards tissue was washed with PBS and cut into
adapter molecules was originally described using a Fab fragment small pieces with sterile scalpels. These pieces were incubated
of a knob-binding monoclonal antibody chemically linked to a cell overnight with 1 mg/ml collagenase/dispase (Roche, Mannheim,
binding ligand, monoclonal antibody, or growth factor.9,30 The Germany) at 4°C. Next day tissue was washed with PBS and
feasibility of the “adapter approach” for potential clinical applica- incubated for 60 min at 37°C in trypsin/EDTA (Life Technolo-
tions was further underlined by reports on effective modification of gies). Tissue pieces were further minced by an inverted syringe
adenoviral tropism by chemical conjugates in vivo.31,32 However, and passed through a 40 ␮m cell strainer (Falcon, Heidelberg,
the application of bispecific chemical conjugates, especially in Germany). Cells were then either stored frozen in FCS 10%
clinical trials, is hampered by their rather elaborate synthesis and DMSO or cultivated for further experiments in RPMI containing
purification procedures, resulting in a substantial portion of non- 10% FCS (BioWhittaker, Walkersville, MD), 20 ␮g/ml gentamy-
functional and poorly defined molecules. In contrast, recombinant cin (Sigma, Deisenhofen, Germany) and 2 mM glutamine (Bio-
bispecific adapter molecules such as sCAR-ligand fusion proteins Whittaker). Cells were grown at 37°C in a humidified atmosphere
(sCAR, extracellular domain of CAR,33–36 tandem scFvs37 or of 5% CO2.
single-chain diabodies (scDbs38) represent defined molecules that
are produced by standard procedures of protein expression in Recombinant adenoviruses
insect cells, eucaryotic cells or bacteria, respectively, and protein Adenovirus serotype 5-derived vectors AdGFPLuc (wild-type
purification by chromatography. These properties are advanta- capsid) and AdGFPLucY477Ax6H (Fiber mutant, CAR-binding ab-
geous for potential clinical applications. We established previously lated) viruses have been described earlier48 and are here named AdGL
the suitability of bispecific scDbs for targeting of adenovirus and AdGLmF, respectively. AdGLmFP (AdGFPLucY477RAEx6H,
infection to endothelial cells.38 In that study, we generated an also named DATL for Double-Ablated TracLuc) was constructed
scDb, EDGAd, which simultaneously binds to the Ad5 knob and to by site directed mutagenesis with oligo mutRAE (CAT GCC ATT
a cell surface marker of tumor endothelium, endoglin. This scDb CGC GCT GAG ACC TTT GCC AC -RAE residues underlined)
mediated an increased gene transfer by an adenoviral vector with in a plasmid containing a Sac I - Sal I fragment of the penton base
wild-type capsid specifically to endothelial cells. For targeting of (pGEM3Z-PB). This mutation was then incorporated into the viral
melanoma, we sought to direct cell binding of capsid-mutant genome of AdGLmF by homologous recombination. Viruses were
adenoviruses to HMWMAA with a single-chain diabody, MelAd. amplified in 293 (AdGL) or 293.HissFv.rec (AdGLmF, and AdG-
This was derived from an scFv binding to the Ad5 fiber knob and LmFP) cells and purified by 2 rounds of CsCl density gradient
138 NETTELBECK ET AL.

ultracentrifugation. Validation of viral genomes (mutations in fiber Flow cytometry analysis

and penton base genes) and exclusion of wild-type contamination Cells were detached from the cell culture dishes with 0.02%
was performed by PCR and restriction digest. The physical particle EDTA. After being washed twice with PBS/1% FBS, cells were
concentration (viral particles (vp)/ml) was determined by OD260 incubated with antibodies diluted in PBS/1%FBS in a volume of
reading and biological particle concentration (plaque-forming 100 ␮l for 45 min at 4°C. For detection of CAR, monoclonal
units (pfu)/ml) was determined by standard plaque assay on antibody RmcB (25, hybridoma cell line purchased from ATCC)
293.HissFv.rec cells. Particle/pfu ratios were 15 for AdGL, 60 for or mouse IgG (Sigma Chemical Co.) as a control were employed
AdGLmF and 72 for AdGLmFP. at a final concentration of 2 ␮g/ml and detected with Alexa488-
Construction of bispecific single-chain diabody MelAd labeled anti-mouse antibody (Molecular Probes, Eugene, OR) di-
luted 1:200. For analysis of scFv or scDb binding, RAFT3 scFv or
The single-chain diabody (scDb) MelAd was derived from the scDbMelAd were applied at 5 ␮g/ml or 10 ␮g/ml, respectively,
HMWMAA-binding scFv RAFT339 and an scFv derived from the and detected with Anti-Tetra-His antibody (Qiagen) diluted 1:250
monoclonal antibody 1D6.14 binding to trimeric Ad5 knob.47 and Alexa488-labeled anti-mouse antibody. Subsequent to the
ScDbMelAd cDNA was constructed by insertion of PCR products antibody incubations, cells were washed twice with PBS, resus-
into pAB1,38 resulting in pAB1MelAd in the configuration SfiI - pended in 0.5 ml PBS and analyzed by flow cytometry (FACS-
VH/RAFT3 scFv - BstEII - VL/anti-knob scFv - AscI - VHR/anti- Calibur; Becton Dickinson, Franklin Lakes, NJ).
knob scFv - SacI - VL/RAFT3 scFv - NotI. Primers VHforAscI
(5⬘-GGT GGG CGC GCC TCG GGC GGA GGT GGC TCA GGC
Adenovirus infection experiments
GGA GGT GGC TCA GAG GTG CAG CTT CAG GAG TCA G)
and VHrevSacI (5⬘-CTG GGT GAG CTC GAT ATC TGA CCC To assess adenoviral transduction, 1 ⫻ 104 cells were plated per
GCC CCC TCC TGA GGA GAC GGT) were employed to am- well of a 96-well plate. The next day 2.5 ⫻ 105 vp (experiments
plify the VH domain and primers VLforBstEII (5⬘-ACC TCG GTC with cell lines) or 5 ⫻ 105 vp (experiments with primary mela-
ACC GTC TCG AGT GGC GGT GGT GGC TCT GAC ATT noma cells) adenovirus in 25 ␮l DMEM/2% FBS were mixed with
GTG) and VLrevAscI (5⬘-GCC CGA GGC GCG CCC ACC GCT 25 ␮l PBS (control) or scDb MelAd at 10 ␮g/ml (cell lines) or 20
GCC ACC GCC TCC TTT CAG CTC CAG CTT GGT CCC AG) ␮g/ml (primary melanoma cells) in 25 ␮l PBS per well and
to amplify the VL domain of the knob-binding scFv with plasmid incubated for 1 hr at 37°C. Mixtures were then added to the cells
pOPE51knobscFv as template (restriction sites are underlined). and incubated for 45 min. Subsequently, virus complexes were
Primers RAFT/V␬rev (5⬘-ATTCAGATCCTCTTCTGAGAT- aspirated, cells were washed with PBS, and 200 ␮l of complete
GAG) and RAFT/V␬forSacI (5⬘-GAT ATC GAG CTC ACC CAG medium was added to each well. To block CAR- or HMWMAA-
TCT CCA TCC TCC CTG TCT) were employed to amplify the mediated adenovirus uptake, cells were incubated for 1 hr with 20
V␬domain and primers RAFT/VHrevBstEII (5⬘-CGA GAC GGT ␮g/ml recombinant knob protein or with 120 ␮g/ml scFv RAFT3
GAC CGA GGT TCC TTG ACC CCA GGA GTC) and RAFT/ at 37°C before virus infection. Luciferase activity of cell lysates
VHforSfiI (5⬘-TAT GCG GCC CAG CCG GCC ATG GCC CAG was determined 48 hr after virus infection using a luciferase assay
ATC) to amplify the VH domain of scFv RAFT3 with plasmid system (Promega, Madison, WI). Experiments were performed in
pUC119RAFT3 as template. The sequence of pAB1MelAd was triplicates; mean values and standard deviations are shown. Sta-
confirmed by sequencing. tistical analysis was performed using a Student’s t-test. Differences
were considered statistically significant for p ⬍ 0.05.
Expression of recombinant proteins in E. coli
The scFv RAFT3 and scDb MelAd both possessing a His-tag at
the C-terminus were purified by immobilized metal-affinity chro- RESULTS
matography. One or 3 liters of LB medium, 100 ␮g/ml ampicillin, A novel strategy for retargeting of adenovirus infection to
0.1% glucose were inoculated with 5 or 15 ml of an overnight melanoma
culture of a clone of E.coli strain TG1 (Stratagene, La Jolla, CA)
The goal of our study was to achieve targeted and efficient
containing the expression plasmid for the recombinant protein and
adenoviral transduction of melanoma cells. Therefore, the modi-
grown to an A600 of 0.8 at 37°C. Protein expression was then
fication of adenovirus tropism is required. To this end, we estab-
induced by adding 1 mM IPTG and the bacteria were grown for an
lished a 2-component strategy, as outlined in Figure 1. This strat-
additional 3 hr at 23°C. The bacteria were harvested by centrifu-
egy combines ablation of the native adenoviral tropism by
gation and resuspended in 30 mM Tris-HCl, pH 8, 1mM EDTA,
mutating the fiber (AdGLmF) or both fiber and penton base (AdG-
20% sucrose. After 15 min, MgSO4 was added to a final concen-
LmFP, not shown) with the introduction of a new tropism for
tration of 5 mM and cells were centrifuged at 8,000 rpm for 15
melanoma cells via a recombinant adapter, a bispecific, bacterially
min. The supernatant was dialyzed against IMAC buffer (50 mM
expressed single-chain diabody (scDb MelAd) that binds to the
sodium phosphate buffer, pH 7.5, 500 mM NaCl) containing 20
adenoviral fiber (both wild-type and mutant) and to HMWMAA.
mM imidazole (IMAC loading buffer) and loaded onto a Ni-NTA
All viruses contained a luciferase reporter gene driven by the CMV
column (Qiagen, Chatsworth, CA) equilibrated with loading
promoter in the E1 region. AdGL was used as control and con-
buffer. After washing with IMAC buffer containing 35 mM imi-
tained wild-type fiber and penton base.
dazole (wash buffer), scFvs or scDbs were eluted with IMAC
buffer containing 100 mM imidazole (elution buffer). Peak frac-
tions were dialyzed overnight against PBS. Soluble Ad5 knob Specific binding of the anti-HMWMAA scFv RAFT3 to
protein49 was purified from cytoplasmic extracts by IMAC as melanoma cells
described above. We analyzed the ability of the anti-HMWMAA scFv RAFT3,
the candidate cell-binding component of the scDb adapter, to bind
SDS-PAGE and Western blot to a panel of 5 melanoma and 3 nonmelanoma cell lines. This scFv
Purified recombinant proteins were analyzed by SDS-PAGE. was derived from the monoclonal antibody LHM240 and was
Formation of trimeric knob was analyzed by SDS-PAGE of boiled further modified by chain shuffling (replacing the mouse V␬ chain
and unboiled samples. For immunoblot analysis of scDb MelAd, by a human V␬ chain) and antibody phage selection resulting in
protein was transferred to a PVDF membrane (Biorad, Hercules, higher affinity binding to the same eptitope.39 By flow cytometry
CA) after SDS-PAGE. The membrane was probed with Anti- analysis. we showed strong binding of the scFv RAFT3 to all
Penta-His monoclonal antibody (Qiagen) and secondary horserad- human melanoma cell lines tested but no binding to lung adeno-
ish peroxidase-conjugated antibody (Sigma Chemical Co., St. carcinoma, breast cancer and ovarian cancer cells (Fig. 2). This
Louis, MO). Bound antibody was detected by enhanced chemilu- result demonstrates melanoma-specific cell binding by the anti-
minescence (NEN Life Science Products, Boston, MA). HMWMAA scFv RAFT3. RAFT3 did not bind to the murine
 ScDb FOR RETARGETING OF ADENOVIRUS INFECTION TO MELANOMA 139

FIGURE 1 – Outline of the strategy employed for transductional

targeting of adenoviral vectors to melanoma cells: modification of the
adenoviral capsid by combining a fiber mutation conferring ablation of
CAR-binding with a single-chain diabody (scDb) adapter for retarget-
ing of adenoviral infection (left). Structure of the bispecific single-
chain diabody MelAd derived from the anti-Ad5 knob scFv cloned
from MAb 1D6.14 and from the anti-HMWMAA scFv RAFT3 (right).
VH, VL, variable domain of the IgG heavy or light chain; HMWMAA,
high molecular weight melanoma-associated antigen; Y477A, point
mutation of the fiber to ablate CAR binding.

melanoma cell line B16F10, which might indicate species-speci-

ficity of antigen recognition by this scFv.
Construction, bacterial expression and specific cell binding of
single-chain diabody MelAd
Encouraged by the melanoma-specific cell binding properties of
the scFv RAFT3, we generated a recombinant bispecific adapter
molecule, the single-chain diabody MelAd, by combining RAFT3
with an scFv binding to Ad5 knob.47 For this purpose, the variable
domains of scFv RAFT3 were genetically fused to those of the
knob-binding scFv by defined nucleotide linkers as indicated in
Figure 1. Recombinant protein was expressed in E.coli and puri-
fied from the periplasm by IMAC. Analysis of the purified protein
by SDS-PAGE followed by Coomassie staining and Western blot
showed a protein band of the expected size of 57 kDa (Fig. 3a). By
flow cytometry analysis, we demonstrated that the engineered
scDb MelAd binds to melanoma cells but not to lung adenocarci-
noma cells (Fig. 3b). Thus, the scDb MelAd retains the melanoma-
specific cell-binding properties of the scFv RAFT3.
FIGURE 2 – FACS analysis: Evaluation of binding of the anti-
Infectivity of fiber and penton base mutant adenoviruses to HMWMAA scFv RAFT3 to human melanoma cell lines SK-MEL-28,
melanoma cells is restored by scDb MelAd A375M, Mel888 and MeWo, to the murine melanoma cell line
We next determined whether the scDb MelAd could mediate B16F10 and to the nonmelanoma cell lines A549 (lung adenocarcino-
efficient infection of melanoma cells by adenovirus mutants with ma), MCF-7 (breast cancer) and OV-4 (ovarian cancer). Single-chain
Fv RAFT3: thick line, control: gray.
ablated native tropism. For this purpose, SK-MEL-28 melanoma
cells (which express high levels of CAR, see below) were trans-
duced by capsid-mutated adenoviruses AdGLmF or AdGLmFP, or transduction by AdGL but not by AdGL complexed to scDb
by AdGL, the vector that contains a wild-type capsid. Viruses were MelAd. Most importantly, transduction of SK-MEL-28 cells by
applied with or without scDb MelAd. Also, cells infected by AdGLmF was strongly increased by the scDb MelAd, resulting in
AdGL were preincubated with recombinant Ad5 knob to block a luciferase activity similar to AdGL-infected cells. These obser-
CAR or with buffer only. The reporter assay (Fig. 4) demonstrated vations clearly establish that scDb MelAd binds to the fiber mutant
50-fold reduced luciferase activity in SK-MEL-28 cells after in- and that infection of scDb MelAd-complexed adenovirus is CAR-
fection by the fiber mutant virus AdGLmF relative to AdGL. A independent. Preincubation of viruses with scFv RAFT3 did not
similar inhibition of transduction by AdGL was observed when result in increased adenoviral transduction (not shown). Thus, we
CAR was blocked by preincubation of the cells with recombinant could exclude the possibility that increased adenoviral transduc-
Ad5 knob (47-fold reduction in luciferase activity). The mutation tion by scDb MelAd results from a cellular response to the inter-
of both the CAR-binding domain of the fiber and the integrin- action of the antibody with HMWMAA. The scDb MelAd medi-
binding motif within the penton base in AdGLmFP resulted in ated increased transduction of melanoma cells by the double
further reduced luciferase activity (230-fold reduced relative to mutant AdGLmFP, indicating that HMWMAA-mediated adenovi-
AdGL). The scDb MelAd mediated a modest increase of AdGL- ral infection does not require binding of the penton base to inte-
mediated luciferase expression when CAR was not blocked (1.5- grins for internalization. However, overall levels of gene transfer
to 3-fold). However, blocking of CAR resulted in strongly reduced were reduced for both uncomplexed and scDb MelAd-complexed
140 NETTELBECK ET AL.

FIGURE 4 – Ablation of native adenoviral tropism and retargeting of

adenovirus infection to melanoma cells. Luciferase assay after trans-
duction of SK-MEL-28 cells with AdGL (wild-type capsid), AdGLmF
(fiber mutated, thus CAR-binding ablated) or AdGLmFP (fiber and
penton base mutated, thus CAR- and integrin-binding ablated) with or
without bound scDb MelAd. Infection with AdGL with or without
scDb MelAd was also performed after blocking cellular CAR with
recombinant knob protein. Mean RLUs and standard deviations (bars)
are shown. *p ⬍ 0.05; ** p ⬍ 0.05 for comparison to AdGL; *** p ⬍
0.05 for comparison to AdGLmF with scDb.

FIGURE 5 – Transduction of CAR-binding ablated adenoviruses

complexed to scDb MelAd is melanoma-specific and HMWMAA-
FIGURE 3 – Expression of the bispecific scDb MelAd and analysis of mediated. (a) Luciferase assay after infection of melanoma cells (SK-
its cell binding properties. SDS-PAGE analysis of the bacterially MEL-28, Mel888 and A375M) or nonmelanoma cells (A549, MCF-7
expressed and IMAC-purified scDb MelAd: (a) Coomassie staining and OV-4) with AdGLmF with or without bound scDb MelAd. Num-
(lane 1) and immunoblot with an anti-His-tag antibody (lane 2). M, bers indicate the ratio of mean RLUs with scDb to mean RLUs without
molecular weight marker; numbers indicate molecular weight in kDa. scDb. (b) Luciferase assay after infection of SK-MEL-28 cells with
(b) Purified scDb MelAd was analyzed by flow cytometry for binding AdGLmF without scDb, AdGLmF with scDb or AdGLmF with scDb
to melanoma cells (SK-MEL-28) and nonmelanoma cells (A549). after blocking cellular HMWMAA with scFv RAFT3. Mean RLUs
Single-chain diabody MelAd: thick line, control: gray. and standard deviations (bars) are shown.

substantially increased efficiency if compared to uncomplexed

adenovirus after mutation of the penton base RGD motif. Further CAR-binding-ablated virus (93- to 367-fold increase in lucif-
analysis of the specificity of scDbMelAd-mediated adenovirus erase activity). For nonmelanoma cell lines, the increase in
infection was performed with AdGLmF. transduction was only minimal, resulting in a tissue-specificity
index (increase of luciferase activity in melanoma vs. nonmela-
ScDb MelAd-mediated adenovirus infection is specific for noma cell lines) of 16- (Mel888 versus OV-4) to 280-fold
melanoma cells and mediated by HMWMAA (A375M versus A549). In order to demonstrate that scDb
Next we analyzed the specificity of scDb MelAd-mediated MelAd-mediated adenoviral infection of melanoma cells is
adenoviral transduction. First, we transduced 3 human mela- HMWMAA-dependent, we blocked this antigen by pre-incuba-
noma cell lines (SK-MEL-28, Mel888 and A375M) and 3 tion of cells with the scFv RAFT3 (Fig. 5a). Indeed, blocking of
human nonmelanoma cell lines (A549 lung adenocarcinoma, HMWMAA inhibited transduction by AdGLmF complexed to
MCF-7 breast cancer and OV-4 ovarian carcinoma) with AdG- the scDb MelAd by more than 3-fold. Thus, adenoviral trans-
LmF with or without scDb MelAd (Fig. 5a). The scDb MelAd duction of scDb MelAd-complexed adenoviruses is mediated by
complexed adenovirus transduced all melanoma cell lines with HMWMAA.
 ScDb FOR RETARGETING OF ADENOVIRUS INFECTION TO MELANOMA 141
ScDb MelAd mediates transductional targeting and infectivity reported in our study is that the ablation of the native tropism is
enhancement of adenovirus for primary melanoma cells independent of the interaction between adapter and adenovirus.
We further analyzed the developed targeting strategy with Hence, a wild-type fiber virus but not a fiber mutant virus can still
freshly purified (primary) human melanoma cells as clinically infect cells via CAR-binding, if only a subset of fibers are bound
more relevant substrates. In this regard, it has been reported that to a targeting adapter. These considerations are especially relevant
melanoma cells in situ or freshly purified from biopsies express for in vivo applications, where adapters might dissociate from viral
CAR at low levels only, or not at all.8 However, the established particles over time. Adapters derived from knob-binding scFvs and
melanoma cell lines analyzed in this report were highly transduc- cell-binding scFvs have been described in 2 formats. One is the
ible by adenoviruses and transduction was reduced after blocking tandem scFv (format scFvA-scFvB or scFvB-scFvA37) and the
virus binding to CAR (Fig. 4 and data not shown). Thus we other is the more compact scDb (format VHA-VLB-VHB-VLA or
analyzed levels of CAR-expression of established melanoma cell VLA-VHB-VLB-VHA, with VH and VL, indicating the variable
lines and of 2 primary cell cultures by flow cytometry (Fig. 6a). domains of the heavy or light antibody chain, respectively38).
Our data showed strong expression of CAR by all established cell Tandem scFvs have been produced in eucaryotic cells and applied
lines. Melanoma cells freshly purified from one patient (patient 1) as crude cell culture supernatant. Single-chain diabodies might
did not express any detectable CAR, whereas those from another differ from tandem scFvs in their stability due to their more
patient (patient 2) express CAR at a level somewhat lower if compact structure. They were produced in bacteria and applied as
compared to established cell lines. Importantly, FACS analysis chromatography-purified preparations. Thus, like the sCAR-de-
with scFv RAFT3 revealed strong expression of HMWMAA in rived adapters, which were produced in insect cells, these mole-
primary melanoma cells from both patients (Fig. 6b). We then cules are homogenous proteins and are relatively easy to generate
transduced primary melanoma cells and the established cell line by procedures compatible with upscaled production for clinical
SK-MEL-28 with AdGL, AdGLmF and AdGLmF complexed to applications. Importantly, knob-binding scFv-derived adapters can
scDb MelAd (Fig. 6c). The CAR-expressing primary melanoma bind to fiber-mutant adenovirus if the mutation does not modify
cells (patient 2) showed a pattern of transduction comparable to the epitope to which the antibody fragment binds. We have proven
SK-MEL-28, although at a somewhat higher level. In contrast, the this concept here for the Y477A fiber mutant and the Ad5 knob-
CAR-deficient primary melanoma cells (patient 1) were only binding MAb 1D6.14-derived scDb MelAd. Gerritsen and co-
poorly transduced by adenoviruses with wild-type capsid. Trans- workers have recently reported on binding of a tandem scFvs that
duction of these cells by AdGLmF complexed to the scDb MelAd contain a library-derived anti-Ad5 knob scFv, S11, to a different
resulted in enhanced transduction compared to both AdGLmF (TAYT deletion in FG loop) fiber mutant.51,52 S11 also binds to the
(170-fold) and AdGL (54-fold). Thus, the scDb MelAd mediates Y477A mutant (RA, unpublished observation). Yet another fiber
specific and strongly enhanced adenoviral transduction of mela- mutant that contains 4 point mutations in the AB loop was not
noma cells that lack CAR. feasible for adapter-mediated targeted adenoviral transduction.51
Notably, the strategy developed in our study resulted in a restored
DISCUSSION
infectivity of fiber mutant adenoviruses for melanoma cells. By
binding scDb MelAd to fiber mutant adenovirus AdGLmF, we
Our study describes a new strategy for retargeting of adenoviral achieved a 93- to 367-fold increased transduction of various mel-
vectors to melanoma cells that recognizes the need to ablate the native anoma cells lines relative to AdGLmF alone. Moreover, blocking
adenoviral tropism and simultaneously introduce a new and targeted of HMWMAA by incubation of melanoma cells with scFv RAFT3
tropism. Towards this goal, we combine a genetic strategy to ablate resulted in inhibition of transduction by scDb MelAd-complexed
native adenoviral cell binding with a bispecific adapter derived from adenovirus. Finally, the scDb MelAd mediated no or minimally
an scFv antibody fragment that binds to a melanoma cell surface increased transduction of nonmelanoma cell lines. Because
antigen. For infectivity ablation, we mutated amino acid 477 of the RAFT3 does not bind to these cells, minimal nonspecific trans-
adenoviral capsid fiber DE loop by replacing tyrosine for alanine.35 duction via scDb MelAd might be attributable to a change in the
This mutation was described recently to eliminate binding to the surface charge of viral particles by binding to scDbs. In conse-
primary adenovirus receptor CAR.27–29 Our results show a substan- quence, these observations demonstrate clearly that the developed
tially reduced infectivity of the mutant virus AdGLmF relative to retargeting strategy results in specific, HMWMAA-mediated, and
wild-type AdGL, which is in agreement with previous reports.48 As CAR-independent infection of melanoma cells. CAR-indepen-
expected, the reduction in transduction efficiency by fiber mutation or dence is also reflected by our finding that melanoma cell transduc-
by blocking CAR with recombinant knob protein were nearly iden- tion by AdGL, but not by AdGL/scDb MelAd is inhibited by
tical (50-fold or 47-fold, respectively). This result confirms that the recombinant Ad5 knob. The scDb MelAd likewise restored the
fiber mutation Y477A completely abolishes CAR-mediated adenovi- transduction of melanoma cells by the doubly ablated adenovirus
ral infection, as required for the targeting strategy developed in our AdGLmFP, indicating that binding of the penton base to cellular
study to avoid transduction of normal cells. However, the interaction integrins is not required for internalization of the HMWMAA-
of the viral penton base with cellular integrins, required for internal- targeted adenovirus. Similar results have been reported for adeno-
ization of native viral particles, was shown to mediate residual ade- viral vectors targeted to different cell surface molecules.38,51 We
novirus infection in the absence of CAR binding.50 We generated a hypothesize that the nature of the target molecule, specifically its
double mutant virus, AdGLmFP, with an additional mutation of the capability for internalization after ligand binding, determines
penton base RGD motif that ablates binding to integrins26 and resulted whether tropism-modified adenoviruses can infect cells indepen-
in a further decrease of adenoviral transduction (230-fold relative to dent of the interaction between penton base and integrins. It is
AdGL). noteworthy that HMWMAA itself has been shown to interact with
To complete the adenovirus retargeting strategy, we directed ␣4␤1-integrins.53 Thus, it is possible that internalization of the
these native tropism-ablated adenovirus mutants to melanoma cells scDb MelAd-bound, doubly ablated adenovirus is triggered by
by a novel recombinant bispecific protein, scDb MelAd. We and binding of HMWMAA to cellular integrins. The ability of HMW-
others have previously described recombinant, bispecific adapter MAA to mediate adenovirus internalization, directly or indirectly,
molecules.33–38 Targeting adapters derived from sCAR (mono-, is advantageous for the adenovirus retargeting strategy described
di- or trimeric) have been shown to block adenoviral infection of in our study. This is because the transduction of nontarget cells is
irrelevant cells and to direct infection to cells to which the fused minimized if penton base-mutated adenoviruses are employed,
ligand binds.33–36 However, in contrast to the scDb MelAd de- thus avoiding nonspecific, penton base-mediated cell binding.
scribed in the present study, sCAR-derived adapters naturally However, penton base-mediated cell binding might also be respon-
cannot bind to fiber mutant viruses with ablated native tropism. In sible for the observed decrease in transduction of melanoma cells
this regard, an important advantage of the retargeting strategy by both scDb-complexed and uncomplexed adenoviruses after
142 NETTELBECK ET AL.

FIGURE 6 – Expression of CAR and HMWMAA on primary melanoma cells and scDb MelAd-mediated infection of primary melanoma cells:
transductional targeting and infectivity enhancement. (a) Detection of CAR expression on established melanoma cell lines (SK-MEL-28, A375M
and Mel888) and primary melanocytes isolated from 2 different patients (Pat. 1 and 2) by flow cytometry after staining of cells with the anti-CAR
monoclonal antibody RmcB (thick line) or isotype control (gray). (b) Analysis of HMWMAA expression on primary melanoma cells by flow
cytometry after staining with scFv RAFT3 (thick line) or control (gray). (c) Luciferase assay after transduction of the melanoma cell line
SK-MEL-28 and of primary melanoma cells by AdGL (wild-type capsid), or by AdGLmF (fiber mutant) with or without scDb MelAd. Numbers
indicate ratios of mean RLUs of AdGLmF with scDb to AdGLmF without scDb (upper numbers) or of AdGLmF with scDb to AdGL without
scDb (lower numbers). Mean RLUs and standard deviations (bars) are shown.

mutation of the penton base RGD motif, because melanoma cells a result of the artificial cell culture environment. For this reason,
strongly, but not specifically, express integrins. we examined the developed strategy for retargeting of adenoviral
Tropism-modification of adenoviral vectors for gene therapy infection with freshly purified, primary melanoma cells that rep-
and viral oncolysis is mandated whenever expression of the ade- resent clinically more relevant substrates. Indeed, flow cytometry
novirus receptor CAR is variable, low or absent on tumor cells as analysis of 2 isolates from different patients showed a complete
described for several tumor types including melanoma.8 In fact, paucity of CAR expression in 1 and intermediate CAR expression
recent studies describe CAR as a tumor suppressor gene.54 There- in the other specimen. These results are in accord with a previous
fore, strong expression of CAR in established cell lines might be study of Hemmi et al,8 in which a large number of freshly purified
 ScDb FOR RETARGETING OF ADENOVIRUS INFECTION TO MELANOMA 143

tumor cells and tissue-sections was analyzed for CAR expression. The strategy for retargeting of adenoviral infection established
In contrast, all established melanoma cell lines showed strong in our study embodies a platform for further therapeutic develop-
CAR expression. Importantly, we could clearly show with these ment. Clearly, scFvs represent attractive tools for vector targeting
primary cell cultures, representing CAR-negative or intermediate because novel scFvs that bind to a target of interest can be derived
CAR expressing melanoma cells, that the targeted adenovirus by existing technologies such as the selection of phage libraries or
results in specific and substantially enhanced adenoviral infectiv- cloning from monoclonal antibodies. Given that the scDb-targeting
ity. Melanoma cells from both patients showed strong expression adapter is derived from a cell-binding scFv antibody fragment, the
of HMWMAA and AdGLmF complexed to scDb MelAd resulted targeting strategy investigated in our study is flexible and can be
in 54-fold increased transduction of CAR-negative melanoma cells applied to different tumors or targets by switching this scFv
relative to the vector with wild-type capsid. component. Furthermore, to avoid immunogenicity it might be
necessary to replace nonhuman components of the scDb with fully
Our results are in accord with recent reports51,52 on retargeting
human sequences. In this regard, the V␬chain of the HMWMAA-
of adenovirus infection with capsid mutant viruses and a distinct
binding scFv was already replaced with a human sequence and a
adapter molecule, the tandem-scFv, indicating a broad applicabil- similar procedure could be applied to the residual domains of the
ity and underlining the potential of the capsid mutant/adapter scDb. Notably, the scDb MelAd can be employed for gene therapy
approach. The study of van Beusechem and co-workers51 exploited of HMWMAA-positive tumors other than melanoma, such as
an EGFR-binding adapter for transduction of brain tumors. Hei- leukemias, osteosarcomas or gliomas. In addition, the specificity of
demann et al. 52 targeted gastric cancers with an adapter binding to our approach can be further increased by incorporation of selective
epithelial cell adhesion molecule. In contrast to our strategy for promoters for transcriptional targeting of gene expression.58 – 60
HMWMAA-targeted adenoviral gene transfer, the former study This concept of combined transductional and transcriptional tar-
shows increased selectivity and efficacy of EGFR-targeted adeno- geting has been described previously with bispecific chemical
viral gene transfer after ablation of the penton-integrin interaction. conjugates targeted to pulmonary endothelium or osteosarco-
These observations indicate that the role of the penton-integrin ma.61,62 Finally, transductional targeting of adenoviral infection
interaction in targeted gene transfer depends on the targeted cell will be pivotal also for our recently described approach of mela-
type (i.e., its integrin expression) and/or the target molecule. noma cell killing by adenoviral oncolysis.63 In this regard, we
Based on the proof-of-concept provided by our study, future observed that CAR-negative primary melanoma cells were resis-
investigations will have to determine the feasibility of the de- tant to cell killing by melanoma-targeted conditionally replicating
scribed adenovirus retargeting strategy in vivo. In previous reports, adenoviruses or even to wild-type adenoviruses (DMN, AAR,
targeting of adenoviral vectors to tumors has been demonstrated unpublished observations).
after loco-regional application of the virus.31,55 However, for tar- In conclusion, we describe in this report a novel strategy for
geting of gene transfer to metastatic melanoma adenoviruses need retargeting of adenoviral infection to malignant melanoma via a
to be applied systemically. Whereas adenoviral gene transfer has tumor antigen and independent of the adenovirus receptor CAR.
been successfully targeted to pulmonary endothelium after sys- Importantly, we demonstrate specific transduction of melanoma
temic application32 the analysis of tumor-targeted vectors in a cells with an efficacy up to 54-fold superior to wild-type adeno-
systemic context is more complex. In this regard, xenograft tumor virus. Thus, the adenovirus retargeting strategy developed in this
models have shown recently that the efficient transduction of report offers a new means to improve efficiency and safety of gene
tumor cells in vivo not only requires tumor-targeted vector tropism therapy for malignant melanoma and other cancers.
but also a means to overcome physical barriers, such as the
extracellular matrix surrounding tumor nests.56 Moreover, mouse ACKNOWLEDGEMENTS
models of spontaneous tumorigenesis, which are currently being We are grateful to Dr. T.J. Eberlein, Dr. I. Hart and Dr. J.
developed for malignant melanoma,57 might be the preferred Schlom for cell lines. Special thanks to Dr. J. Blackwell, Dr. A.
means for in vivo analysis of tumor-targeted gene transfer. As well, Kanerva, T. Korn, Dr. A. Pereboev, Dr. L. Rivera, Dr. T. Seki, Dr.
in vivo studies are required to analyze the stability and immuno- A. Steinkasserer, W. Stöhr, Dr. T. Strong, Dr. K. Suzuki and Dr.
genicity of adapter-targeted adenoviruses after systemic injection A. Volk for their valuable contributions. We would like to ac-
and would furthermore allow for comparison of the different knowledge the Center For Aids Research Flow Cytometry Facility
adapter formats, such as sCAR-ligand fusion proteins, tandem (support provided by CFAR AIDS Core Grant #P30AI27767) and
scFvs and scDbs in this respect. However, analysis of targeted Marion Spell for providing flow cytometry equipment and exper-
adenoviruses in mouse models requires targeting ligands that are tise. DNA sequencing was performed in the DNA Sequencing and
cross-reactive with the corresponding mouse antigen, unlike the Analysis Core of the UAB Center for AIDS Research, supported
scDb MelAd described in this report. by grant P30 A1-27767.

REFERENCES

1. Gomez-Navarro J, Curiel DT, Douglas JT. Gene therapy for cancer. with efficient adenovirus-mediated transgene expression in human
Eur J Cancer 1999;35:2039 –57. melanoma cell cultures. Hum Gene Ther 1998;9:2363–73.
2. Alemany R, Balague C, Curiel DT. Replicative adenoviruses for 9. Barnett BG, Crews CJ, Douglas JT. Targeted adenoviral vectors.
cancer therapy. Nat Biotechnol 2000;18:723–727. Biochim Biophys Acta 2002;1575:1–14.
3. Nabel GJ. Development of optimized vectors for gene therapy. Proc 10. Sotomayor MG, Hu H, Antonia S, Sotomayor EM, Pardoll DM.
Natl Acad Sci U S A 1999;96:324 – 6. Advances in gene therapy for malignant melanoma. Cancer Control
4. Verma IM, Somia N. Gene therapy: promises, problems and pros- 2002;9:39 – 48.
pects. Nature 1997;389:239 – 42. 11. Wilson BS, Imai K, Natali PG, Ferrone S. Distribution and molecular
5. Vile RG, Russell SJ, Lemoine NR. Cancer gene therapy: hard lessons characterization of a cell-surface and a cytoplasmic antigen detectable
and new courses. Gene Ther 2000;7:2– 8. in human melanoma cells with monoclonal antibodies. Int J Cancer
6. Zhang WW. Development and application of adenoviral vectors for 1981;28:293–300.
gene therapy of cancer, Cancer Gene Ther 1999;6:113–38. 12. Natali PG, Imai K, Wilson BS, Bigotti A, Cavaliere R, Pellegrino
7. Miller CR, Buchsbaum DJ, Reynolds PN, Douglas JT, Gillespie GY, MA, Ferrone S. Structural properties and tissue distribution of the
Mayo MS, Raben D, Curiel DT. Differential susceptibility of primary antigen recognized by the monoclonal antibody 653.40S to human
and established human glioma cells to adenovirus infection: targeting melanoma cells. J Natl Cancer Inst 1981;67:591– 601.
via the epidermal growth factor receptor achieves fiber receptor- 13. Harper JR, Reisfeld RA. Inhibition of anchorage-independent growth
independent gene transfer. Cancer Res 1998;58:5738 – 48. of human melanoma cells by a monoclonal antibody to a chondroitin
8. Hemmi S, Geertsen R, Mezzacasa A, Peter I, Dummer R. The pres- sulfate proteoglycan. J Natl Cancer Inst 1983;71:259 – 63.
ence of human coxsackievirus and adenovirus receptor is associated 14. Berd D, Herlyn M, Koprowski H, Mastrangelo MJ. Flow cytometric
144 NETTELBECK ET AL.

determination of the frequency and heterogeneity of expression of Ryan PC, Kaleko M, Stevenson SC. Targeting adenoviral vectors by
human melanoma-associated antigens. Cancer Res 1989;49:6840 – 4. using the extracellular domain of the coxsackie-adenovirus receptor:
15. Burg MA, Grako KA, Stallcup WB. Expression of the NG2 proteo- improved potency via trimerization. J Virol 2002;76:1892–903.
glycan enhances the growth and metastatic properties of melanoma 36. Ebbinghaus C, Al-Jaibaji A, Operschall E, Schoffel A, Peter I, Greber
cells. J Cell Physiol 1998;177:299 –312. UF, Hemmi S. Functional and selective targeting of adenovirus to
16. Kageshita T, Kuriya N, Ono T, Horikoshi T, Takahashi M, Wong GY, high-affinity Fcgamma receptor I-positive cells by using a bispecific
Ferrone S. Association of high molecular weight melanoma-associ- hybrid adapter. J Virol 2001;75:480 – 489.
ated antigen expression in primary acral lentiginous melanoma lesions 37. Haisma HJ, Grill J, Curiel DT, Hoogeland S, van Beusechem VW,
with poor prognosis. Cancer Res 1993;53:2830 –3. Pinedo HM, Gerritsen WR. Targeting of adenoviral vectors through a
17. Kupsch JM, Tidman NH, Kang NV, Truman H, Hamilton S, Patel N, bispecific single-chain antibody. Cancer Gene Ther 2000;7:901– 4.
Newton Bishop JA, Leigh IM, Crowe JS. Isolation of human tumor- 38. Nettelbeck DM, Miller DW, Jérôme V, Zuzarte M, Watkins SJ,
specific antibodies by selection of an antibody phage library on Hawkins RE, Müller R, Kontermann RE. Targeting of adenovirus to
melanoma cells. Clin Cancer Res 1999;5:925–31. endothelial cells by a bispecific single-chain diabody directed against
18. Siccardi AG, Buraggi GL, Callegaro L, Mariani G, Natali PG, Abbati, the adenovirus fiber knob domain and human endoglin (CD105). Mol
A, Bestagno M, Caputo V, Mansi L, Masi R, et al. Multicenter study Ther 2001;3:882–91.
of immunoscintigraphy with radiolabeled monoclonal antibodies in 39. Kang N, Hamilton S, Odili J, Wilson G, Kupsch J. In vivo targeting
patients with melanoma. Cancer Res 1986;46:4817–22. of malignant melanoma by 125Iodine- and 99mTechnetium-labeled
19. Spitler LE, del Rio M, Khentigan A, Wedel NI, Brophy NA, Miller single-chain Fv fragments against high molecular weight melanoma-
LL, Harkonen WS, Rosendorf LL, Lee HM, Mischak RP, et al. associated antigen. Clin Cancer Res 2000;6:4921–31.
Therapy of patients with malignant melanoma using a monoclonal 40. Kupsch JM, Tidman N, Bishop JA, McKay I, Leigh I, Crowe JS.
antimelanoma antibody-ricin A chain immunotoxin. Cancer Res Generation and selection of monoclonal antibodies, single-chain Fv
1987;47:1717–23. and antibody fusion phage specific for human melanoma-associated
20. Yang, HM, Reisfeld RA. Pharmacokinetics and mechanism of action antigens. Melanoma Res 1995;5:403–11.
of a doxorubicin-monoclonal antibody 9.2.27 conjugate directed to a 41. Neri D, Natali PG, Petrul H, Soldani P, Nicotra MR, Vola R, Rivella
human melanoma proteoglycan. J Natl Cancer Inst 1988;80:1154 –9. A, Creighton AM, Neri P, Mariani M. Recombinant anti-human
21. Eary JF, Schroff RW, Abrams PG, Fritzberg AR, Morgan AC, Kasina melanoma antibodies are versatile molecules. J Invest Dermatol 1996;
S, Reno JM, Srinivasan A, Woodhouse CS, Wilbur DS. Successful 107:164 –70.
imaging of malignant melanoma with technetium-99m-labeled mono- 42. Noronha EJ, Wang X, Desai SA, Kageshita T, Ferrone S. Limited
clonal antibodies. J Nucl Med 1989;30:25–32. diversity of human scFv fragments isolated by panning a synthetic
22. Lamki LM, Zukiwski AA, Shanken LJ, Legha SS, Benjamin RS, phage-display scFv library with cultured human melanoma cells.
Plager CE, Salk DF, Schroff RW, Murray JL. Radioimaging of J Immunol 1998;161:2968 –76.
melanoma using 99mTc-labeled Fab fragment reactive with a high 43. Wang B, Chen YB, Ayalon O, Bender J, Garen A. Human single-
molecular weight melanoma antigen. Cancer Res 1990;50:904s–908s. chain Fv immunoconjugates targeted to a melanoma-associated chon-
23. Ghose T, Ferrone S, Blair AH, Kralovec Y, Temponi M, Singh M, droitin sulfate proteoglycan mediate specific lysis of human mela-
Mammen M. Regression of human melanoma xenografts in nude mice noma cells by natural killer cells and complement. Proc Natl Acad Sci
injected with methotrexate linked to monoclonal antibody 225.28 to U S A 1999;96:1627–32.
human high molecular weight-melanoma associated antigen. Cancer 44. Abken H, Hombach A, Heuser C, Reinhold U. A novel strategy in the
Immunol Immunother 1991;34:90 – 6. elimination of disseminated melanoma cells: chimeric receptors en-
24. Fodstad O, Faye R, Hoifodt HK, Skovlund E, Aamdal S. Immuno- dow T cells with tumor specificity, Recent Results Cancer Res 2001;
bead-based detection and characterization of circulating tumor cells in 158:249 – 64.
melanoma patients. Recent Results Cancer Res 2001;158:40 –50. 45. Tordsson JM, Ohlsson LG, Abrahmsen LB, Karlstrom PJ, Lando PA,
25. Bergelson, JM, Cunningham JA, Droguett G, Kurt-Jones EA, Krithi- Brodin TN. Phage-selected primate antibodies fused to superantigens
vas A, Hong JS, Horwitz MS, Crowell RL, Finberg RW. Isolation of for immunotherapy of malignant melanoma, Cancer Immunol Immu-
a common receptor for Coxsackie B viruses and adenoviruses 2 and 5. nother 2000;48:691–702.
Science 1997;275:1320 –3. 46. Martin F, Kupsch J, Takeuchi Y, Russell S, Cosset FL, Collins M.
26. Wickham, TJ, Mathias P, Cheresh DA, Nemerow GR. Integrins alpha Retroviral vector targeting to melanoma cells by single-chain anti-
v beta 3 and alpha v beta 5 promote adenovirus internalization but not body incorporation in envelope, Hum Gene Ther 1998;9:737– 46.
virus attachment. Cell 1993;73:309 –19. 47. Douglas JT, Miller CR, Kim M, Dmitriev I, Mikheeva G, Krasnykh
27. Roelvink PW, Mi Lee G, Einfeld DA, Kovesdi I, Wickham TJ. V, Curiel DT. A system for the propagation of adenoviral vectors with
Identification of a conserved receptor-binding site on the fiber proteins genetically modified receptor specificities. Nat Biotechnol 1999;17:
of CAR-recognizing adenoviridae. Science 1999;286:1568 –71. 470 –5.
28. Kirby I, Davison E, Beavil AJ, Soh CP, Wickham TJ, Roelvink PW, 48. Alemany R, Curiel DT. CAR-binding ablation does not change bio-
Kovesdi I, Sutton BJ, Santis G. Mutations in the DG loop of adeno- distribution and toxicity of adenoviral vectors. Gene Ther 2001;8:
virus type 5 fiber knob protein abolish high-affinity binding to its 1347–53.
cellular receptor CAR. J Virol 1999;73:9508 –14. 49. Krasnykh VN, Mikheeva GV, Douglas JT, Curiel DT. Generation of
29. Kirby I, Davison E, Beavil AJ, Soh CP, Wickham TJ, Roelvink PW, recombinant adenovirus vectors with modified fibers for altering viral
Kovesdi I, Sutton BJ, Santis G. Identification of contact residues and tropism. J Virol 1996;70:6839 – 46.
definition of the CAR-binding site of adenovirus type 5 fiber protein. 50. Einfeld DA, Schroeder R, Roelvink PW, Lizonova A, King CR,
J Virol 2000;74:2804 –13. Kovesdi I, Wickham TJ. Reducing the native tropism of adenovirus
30. Douglas JT, Rogers BE, Rosenfeld ME, Michael SI, Feng M, Curiel vectors requires removal of both CAR and integrin interactions.
DT. Targeted gene delivery by tropism-modified adenoviral vectors. J Virol 2001;75:11284 –91.
Nat Biotechnol 1996;14:1574 – 8. 51. van Beusechem VW, Grill J, Mastenbroek DC, Wickham TJ, Ro-
31. Gu DL, Gonzalez AM, Printz MA, Doukas J, Ying W, D’Andrea M, elvink, PW, Haisma HJ, Lamfers ML, Dirven CM, Pinedo HM,
Hoganson DK, Curiel DT, Douglas JT, Sosnowski BA, Baird A, Gerritsen WR. Efficient and selective gene transfer into primary
Aukerman SL, Pierce GF. Fibroblast growth factor 2 retargeted ade- human brain tumors by using single-chain antibody-targeted adeno-
novirus has redirected cellular tropism: evidence for reduced toxicity viral vectors with native tropism abolished. J Virol 2002;76:2753– 62.
and enhanced antitumor activity in mice., Cancer Res 1999;59:2608 – 52. Heideman DA, Van Beusechem VW, Offerhaus GJ, Wickham TJ,
14. Roelvink PW, Craanen ME, Pinedo HM, Meijer CJ, Gerritsen WR.
32. Reynolds PN, Zinn KR, Gavrilyuk VD, Balyasnikova IV, Rogers BE, Selective gene transfer into primary human gastric tumors using
Buchsbaum DJ, Wang MH, Miletich DJ, Grizzle WE, Douglas JT, epithelial cell adhesion molecule-targeted adenoviral vectors with
Danilov SM, Curiel DT. A targetable, injectable adenoviral vector for ablated native tropism, Hum Gene Ther 2002 13:1677– 85.
selective gene delivery to pulmonary endothelium in vivo. Mol Ther 53. Iida J, Meijne AM, Oegema TR, Jr., Yednock TA, Kovach NL, Furcht
2000;2:562–78. LT, McCarthy JB. A role of chondroitin sulfate glycosaminoglycan
33. Dmitriev I, Kashentseva E, Rogers BE, Krasnykh V, Curiel DT. binding site in alpha4beta1 integrin-mediated melanoma cell adhe-
Ectodomain of coxsackievirus and adenovirus receptor genetically sion. J Biol Chem 1998;273:5955– 62.
fused to epidermal growth factor mediates adenovirus targeting to 54. Okegawa T, Pong RC, Li Y, Bergelson JM, Sagalowsky AI, Hsieh JT.
epidermal growth factor receptor-positive cells. J Virol 2000 74: The mechanism of the growth-inhibitory effect of coxsackie and
6875– 84. adenovirus receptor (CAR) on human bladder cancer: a functional
34. Kashentseva EA, Seki T, Curiel DT, Dmitriev IP. Adenovirus target- analysis of car protein structure. Cancer Res 2001;61:6592– 600.
ing to c-erbB-2 oncoprotein by single-chain antibody fused to trimeric 55. Rancourt C, Rogers BE, Sosnowski BA, Wang M, Piche A, Pierce
form of adenovirus receptor ectodomain. Cancer Res 2002;62:609 – GF, Alvarez RD, Siegal GP, Douglas JT, Curiel DT. Basic fibroblast
16. growth factor enhancement of adenovirus-mediated delivery of the
35. Kim J, Smith T, Idamakanti N, Mulgrew K, Kaloss M, Kylefjord H, herpes simplex virus thymidine kinase gene results in augmented
 ScDb FOR RETARGETING OF ADENOVIRUS INFECTION TO MELANOMA 145
therapeutic benefit in a murine model of ovarian cancer. Clin Cancer 60. McCart J, Wang Z, Xu H, Hu Y, Park B, Alexander H, Bartlett D.
Res 1998;4:2455– 61. Development of a melanoma-specific adenovirus, Mol Ther 2002;6:471.
56. Shayakhmetov DM, Li ZY, Ni S, Lieber A. Targeting of adenovirus 61. Reynolds PN, Nicklin SA, Kaliberova L, Boatman BG, Grizzle WE,
vectors to tumor cells does not enable efficient transduction of breast Balyasnikova IV, Baker AH, Danilov SM, Curiel DT. Combined
cancer metastases, Cancer Res 2002;62:1063– 8. transductional and transcriptional targeting improves the specificity of
57. Tietze MK, Chin L. Murine models of malignant melanoma, Mol Med transgene expression in vivo. Nat Biotechnol 2001;19:838 – 42.
Today 2000;6:408 –10. 62. Barnett B, Tillman B, Curiel D, Douglas J. Dual targeting of adenoviral
58. Vile RG, Hart IR. Use of tissue-specific expression of the herpes vectors at the levels of transduction and transcription enhances the spec-
simplex virus thymidine kinase gene to inhibit growth of established ificity of gene expression in cancer cells. Mol Ther 2002;6:377.
murine melanomas following direct intratumoral injection of DNA. 63. Nettelbeck DM, Rivera AA, Balague C, Alemany R, Curiel DT.
Cancer Res 1993;53:3860 – 4. Novel oncolytic adenoviruses targeted to melanoma: specific viral
59. Nettelbeck DM, Jérôme V, Müller R. Gene therapy: designer promot- replication and cytolysis by expression of E1A mutants from the
ers for tumour targeting.Trends Genet 2000;16:174 – 81. tyrosinase enhancer/promoter. Cancer Res 2002;62:4663–70.