Reviews

Adenovirus-Derived Vectors for Prostate CancerGene Therapy

Jeroen de Vrij,1 Ralph A. Willemsen,2 Leif Lindholm,3 Rob C. Hoeben,1 and the GIANT Consortium*

Abstract

Prostate cancer is a leading cause of death among men in Western countries. Whereas the survival rate approaches100% for patients with localized cancer, the results of treatment in patients with metastasized prostate cancerat diagnosis are much less successful. The patients are usually presented with a variety of treatment options, buttherapeutic interventions in prostate cancer are associated with frequent adverse side effects. Gene therapy andoncolytic virus therapy may constitute new strategies. Already a wide variety of preclinical studies has demon-strated the therapeutic potential of such approaches, with oncolytic prostate-specific adenoviruses as the mostprominent vector. The state of the art and future prospects of gene therapy in prostate cancer are reviewed, with afocus on adenoviral vectors. We summarize advances in adenovirus technology for prostate cancer treatment andhighlight areas where further developments are necessary.

Introduction

Although many viruses are being evaluated as oncolyticagents, human adenoviruses (HAdVs) are among the

most popular to be developed. There are several good reasonsfor making HAdV vectors such a popular choice. Re-combinant HAdV vectors have a good safety profile as a genetherapy vector, can be produced under GMP conditions, andvarious commercially operating manufacturing facilities areavailable, allowing research groups access to batches ofHAdV vectors that meet the required quality standards. Inaddition, the wild-type HAdVs from which vectors are de-rived are only mildly pathogenic (Van der Vliet and Hoeben,2006). Although HAdV genomes are stable enough to preventthe rapid development of heterogeneous populations ofso-called quasispecies, they are easily amendable by geneticmodification. Genetically modified HAdVs with altered hostrange (‘‘targeted viruses’’) have been generated by engineer-

ing new polypeptide ligands in the capsids of the particles,yielding viruses that preferentially infect specific cell or tissuetypes. In parallel, by engineering mutations at known recep-tor-binding sites in the capsid, ‘‘detargeted’’ HAdV vectorshave been generated to reduce transduction of nontargettissues. Also, HAdVs can be modified to carry therapeuticor reporter transgenes (Bachtarzi et al., 2008). Numerous anddiverse transgenes have been inserted in E1-deleted HAdVsto be exploited as cytolytic agents. Such therapeutic trans-genes include genes encoding prodrug-activating enzymes(e.g., herpes simplex virus thymidine kinase for activatingganciclovir, cytosine deaminase for activating 5-fluorocytosine,bacterial nitroreductase activating CB1954), genes encodingimmune-stimulatory cytokines (interleukin [IL]-2, IL-12,granulocyte-macrophage colony-stimulating factor [GM-CSF], and IL-24), or genes encoding proteins inducing apo-ptosis (e.g., p53). More recently, conditionally replicatingadenoviruses (CRAds) have been developed. Such viruses

1Department of Molecular Cell Biology, Leiden University Medical Center, 2300 RC Leiden, The Netherlands.2Tumor Immunology Group, Department of Medical Oncology, Erasmus MC-Daniel den Hoed, 3075 EA Rotterdam, The Netherlands.3Got-a-Gene, 42930 Kullavik, Sweden.*Members of the GIANT Consortium: Chris H. Bangma,1 Chris Barber,1 Jean-Paul Behr,11 Simon Briggs,2 Robert Carlisle,3 Wing-Shing

Cheng,2 Iris J.C. Dautzenberg,4 Corrina de Ridder,1 Helena Dzojic,2 Patrick Erbacher,12 Magnus Essand,2 Kerry Fisher,13 April Frazier,9

Lindsay J. Georgopoulos,14 Ian Jennings,14 Stefan Kochanek,5 Daniela Koppers-Lalic,4 Robert Kraaij,1 Florian Kreppel,5 Maria Magnusson,7

Norman Maitland,9,14 Patrick Neuberg,13 Regina Nugent,14 Manfred Ogris,10 Jean-Serge Remy,11 Michelle Scaife,14 Ellen Schenk-Braat,1 ErikSchooten,1 Len Seymour,3 Michael Slade,14 Pio Szyjanowicz,14 Thomas Totterman,2 Taco G. Uil,4 Karel Ulbrich,6 Laura van der Weel,1 Wytskevan Weerden,8 Ernst Wagner,10 and Guy Zuber11: 1Erasmus MC University Medical Centre, Rotterdam, The Netherlands; 2Uppsala Uni-versity, Uppsala, Sweden; 3University of Oxford, Oxford, UK; 4Leiden University Medical Center, Leiden, The Netherlands; 5University ofUlm, Ulm, Germany; 6Academy of Sciences of the Czech Republic, Prague, Czech Republic; 7Got-a-Gene, Kullavik, Sweden; 8Scuron,Rotterdam, The Netherlands; 9Procure, York, UK; 10Ludwig-Maximilians-Universitat, Munich, Germany; 11Universite Louis Pasteur deStrasbourg, Illkirch, France; 12Polyplus Transfection, Illkirch, France; 13Hybrid Systems, Oxford, UK; and 14University of York, York, UK.

HUMAN GENE THERAPY 21:795–805 (July 2010)ª Mary Ann Liebert, Inc.DOI: 10.1089=hum.2009.203

795

rely on the lytic replication cycle of HAdVs for tumor cellkilling. Viral replication in the tumor increases the local vectorconcentration and may lead to spread of the virus within thetumor. In addition, HAdV may provide a danger signal thatstimulates an antitumor immune response (Geutskens et al.,2000). Successful preclinical studies have led to various phaseI clinical trials of replication-defective as well as condition-ally replication-competent HAdV vectors in prostate cancer(Figueiredo et al., 2007; Freytag et al., 2007). So far these studieshave confirmed the good safety profile of HAdV vectors.Taken together, these factors have made adenoviral vectors aprime candidate for developing viral oncolytic agents.

Getting the Virus to the Tumor

Intratumoral injection

An important lesson learned from preclinical and clinicalresearch in cancer gene therapy is that efficient transductionof the cancer cells in the tumor is essential for efficacioustreatment. Tumors are heterogeneous and contain a stromalcompartment and extracellular matrix components that formphysical barriers within the tumor (Fig. 1). Therefore evendirect intratumoral administration of viral anticancer agents isoften disappointingly inefficient. Expression of the receptorused by most HAdVs to enter the cell, that is, the coxsack-ievirus and adenovirus receptor (CAR), is often scanty. Thehigh interstitial fluid pressure within most tumors causes aconvective flow from the tumor. This inhibits the passivediffusion of viral particles into the tumor. The extracellularmatrix may form physical barriers that prevent efficientspread of viral vectors (Kuppen et al., 2001; Li et al., 2004; Moket al., 2009). In elegant studies, Jain and co-workers demon-strated that destruction of the matrix by collagenase treatmentor overexpression of matrix metalloproteinases (MMP)-1 and-8 increases the volume distribution of oncolytic herpes-viruses (McKee et al., 2006; Mok et al., 2007). The volumedistribution can also be increased by multiple injections, orby convection-enhanced delivery procedures.

Vascular delivery

Theoretically, vascular delivery of vectors may lead to alarger distribution of viruses within the tumor. Also, it couldprovide an option for transducing (micro)metastatic tumors.However, vascular delivery, too, has been frustratingly inef-ficient so far. This is attributable to a wide variety of factors.Direct contact between malignant cells and the oncolyticHAdV may be difficult to obtain. Often blood vessels areconfined to the tumor stroma, and therefore several layers ofstromal cells must be passed before malignant cells arereached by vascularly applied therapeutic agents (Kuppenet al., 2001; Li et al., 2004).

HAdV vectors may become unavailable to the tumor bypromiscuous association with nontarget tissues such as theliver. The primary receptor of many HAdVs is the CAR. Afterligation of the adenoviral fiber with the CAR an integrin-binding RGD motif in the penton base binds avb3- or avb5-integrins. This promotes adenovirus internalization. BothCAR and the integrins are widely expressed on cells in thehuman body, resulting in transduction of nontarget tissues(Arnberg, 2009). Intriguingly, mutation of the CAR-bindingsite of the fiber and the RGD motif in the penton base was

found not to reduce liver transduction, and on intravascularadministration adenoviruses were still efficiently sequesteredby Kupffer cells in the liver (Di Paolo et al., 2009). Subse-quently it became evident that the interaction of the virus withhost blood cells and plasma proteins is critical. Studies sug-gest that these interactions dictate the particle biodistributionof adenovirus in vivo. Various plasma proteins and in partic-ular vitamin K-dependent coagulation factors IX and X canbind to hexon proteins in the capsid, and bridge the virus toreceptors in the liver (Kalyuzhniy et al., 2008; Waddingtonet al., 2008). In this respect there is variability between humanadenoviral serotypes. Whereas the serotypes commonly usedas vector, that is, HAdV-5 and -2, strongly bind factor X,others such as serotype 26 and 46 do not (Waddington et al.,2007; Alba et al., 2009). Therefore (hexons of) different sero-types, or non-clotting factor-binding derivatives of theHAdV-5 hexon, can be used to decrease the loss of vectorparticles in the liver and to improve bioavailability to thetumors. Indeed, such mutations significantly decrease livertransduction of HAdV-5 vectors in mice (Waddington et al.,2007; Alba et al., 2009). The HAdV-5 fiber harbors a site withhigh affinity for heparan sulfate proteoglycans; however,mutation of the binding motif KKTK barely affects livertransduction in mice (Kritz et al., 2007; Di Paolo et al., 2009).

Not only these interactions with clotting factors thwartefficient tumor cell transduction, but so do neutralizing im-munoglobulins. A majority of humans have preexisting hu-moral neutralizing activity against HAdV-5 and HAdV-2 as aresult of prior exposure to these viruses. Intravascular ad-ministration of adenovirus to recipients with preexisting hu-moral immunity will strongly reduce gene transfer. The use ofvectors derived from HAdV with a low seroprevalence in thegeneral population, or from nonhuman adenoviruses, mayreduce the magnitude of the problem (Abbink et al., 2007).

An unexpected finding came with the observation thathuman, in contrast to murine, erythrocytes bind HAdV-5.Human erythrocytes present CAR at their surface, whichstably interacts with HAdV-5 particles (Carlisle et al., 2009;Seiradake et al., 2009). Also human, but not murine, erythro-cytes present complement receptor-1 (CR1), which bindsHAdV-5 in the presence of antibodies and complement(Carlisle et al., 2009). Transplantation of human erythrocytesinto immune deficient mice extended the blood circulationtime of HAdV-5, reduced liver transduction, and decreasedextravasation of the virus into human xenograft tumors. Si-milarly, HAdV-5 showed extended circulation and decreasedliver transduction in transgenic mice presenting either CAR orCR1 on their erythrocytes. Erythrocytes may therefore restrictHAdV-5 infection in humans, independent of antibody status,presenting another challenge to HAdV-5-based anticancerviruses (Carlisle et al., 2009; Seiradake et al., 2009). Althoughmuch insight has been acquired on the interaction of adeno-viruses with blood cells and plasma proteins, many otherareas of virus–host interactions remain underexplored and ingeneral we understand little of it.

Shielding Vector Particles from Neutralizing Immunity

Although formidable, the challenges of vascular deliverysummarized previously may not be insurmountable. Oneapproach potentially leading to improvements in delivery ofadenoviral vectors for cancer gene therapy involves chemical

796 DE VRIJ ET AL.

coating of the vector particles. This approach is based onestablished and clinically applicable technology of packagingdrugs or therapeutic biologicals in synthetic polymers. It ishoped that this can lead to improved pharmacological pa-rameters, such as improved solubility and stability, reduceddosing frequency, potentially reduced toxicity, and extendedcirculation time. Amongst others, poly[N-(2-hydroxypropyl)-methacrylamide] and polyethylene glycol (PEG) are oftenused to covalently coat therapeutics (Kreppel and Kochanek,2008).

Coating with multivalent polymers based on poly[N-(2-hydroxypropyl)-methacrylamide] abrogated normal HAdV-5infectious tropism (Fisher et al., 2001). In addition, neutrali-zation by antibodies was decreased up to 50-fold, and theresulting polymer-coated adenoviral particles have a greatlyextended plasma circulation time in mice (Fisher et al., 2001;Green et al., 2004). Although normal tropism was blocked,polymer-coated adenovirus accumulated within a solid sub-cutaneous tumor 40 times more efficiently than unmodifiedvirus, and mediated higher levels of transgene expressionwithin tumors. This has been attributed to the enhancedpermeability and retention effect, which leads to the non-specific accumulation of circulating macromolecules withintumors (Matsumura and Maeda, 1986).

After blocking, infectivity can be restored by linking tar-geting peptides onto the surface of the polymer-coated viru-ses. Addition of a synthetic –SIKVAV– peptide, which bindsa6-integrin, can restore viral infectivity of PC-3 cells. Compe-tition assays confirmed that entry of retargeted viruses wasmediated via the incorporated ligand. Intravenous adminis-

tration of retargeted viruses to tumor-bearing mice resulted inslower plasma clearance and greatly reduced liver tropism,and hence toxicity compared with unmodified virus, whilemaintaining reporter gene expression in the tumor (Stevensonet al., 2007). Similarly, polymer-coated HAdV could be tar-geted with cetuximab to target the epidermal growth factor(EGF) receptor (Morrison et al., 2009). The data demon-strate that the polymer-coating technology is compatible withpeptide-based tumor targeting.

Other groups used coating of adenoviral particles withpolyethylene glycol (PEG) molecules for particle shielding,and showed that this allows escape from neutralizing anti-bodies and to some extend allowed vector readministration.In this respect the size of the PEG molecules matters. With theuse of large PEG molecules (e.g., 20-kDa PEG), vector particleswere detargeted from muscle after local delivery and fromliver after systemic delivery in mouse models. Surprisingly,fully detargeted PEGylated adenoviral vectors still inducedstrong cellular and humoral immune responses to vector-encoded transgene products. PEGylation does not affect thekinetics of transgene product-specific cytotoxic immune re-sponses (Wortmann et al., 2008). These data have been cor-roborated by Barry and collaborators, who demonstrated thatPEGylation with 20-kDa PEG was as efficient at detargetingadenovirus from Kupffer cells and hepatocytes as virus pre-dosing and warfarinization (Hofherr et al., 2007, 2008; Weaverand Barry, 2008; Doronin et al., 2009). Bioluminescence im-aging of viral distribution in a xenograft model in nude micedemonstrated that PEGylation with 20-kDa PEG reducedliver infection 19- to 90-fold. Tumor transduction levels were

FIG. 1. Schematic representation of the tumor structure and the hurdles to efficient adenoviral vector-mediated prostatecancer gene therapy. Several efficacy-lowering aspects are encountered on systemic or intratumoral delivery of adenoviralvectors to prostate cancer. Systemic delivery of vector particles to the primary tumor site and to metastasized tumor cells ishampered by innate and humoral immunity, sequestration from the blood stream through the binding of blood cells orplasma proteins, and limited permeability of blood vessels. The heterogeneous composition of the primary tumor massresults in inefficient penetration into the tumor and insufficient transduction of the neoplastic cells.

Ad-DERIVED VECTORS FOR PROSTATE CANCER GENE THERAPY 797

similar for 20-kDa PEGylated and un-PEGylated vectors.Anticancer efficacy after a single intravenous injection wasretained at the level of unmodified vector in large establishedLNCaP prostate carcinoma xenografts, resulting in completeelimination of tumors in all animals and long-term tumor-freesurvival (Doronin et al., 2009). It should be noted that theprotective effects of PEGylation could be more pronounced inthe presence of human erythrocytes, as PEGylation will alsoreduce association with CAR and complement receptorI. Taken together, these data suggest that chemical shieldingof HAdV particles is a powerful approach to prevent inter-action of HAdV particles with blood proteins, erythrocytes,and nontarget tissues, and thereby may increase the bio-availability, and as a result the uptake of viruses into tumors,by enhanced permeability and retention.

Vascular Permeability

To further stimulate extravasation of vector particles,new physiological regulators of vascular permeability (i.e.,vascular endothelial growth factor [VEGF]) may be used.This strategy allows enhanced transduction of striated muscleby combining intravenous AAV6-vector administration withinfusion of VEGF (Gregorevic et al., 2004). Alternatively,strategies are being developed that employ the endothelialreceptor-mediated transcytosis pathway. The transferrin re-ceptor is an example of a receptor that binds transferrin and itsassociated iron, resulting in caveolar uptake of the complex.Via a series of vesicles these complexes are transported acrossthe endothelium and released at the basolateral side. Curieland co-workers provided evidence that this pathway can beused by adenoviruses. By using a bifunctional adaptor, forexample, a soluble CAR–transferrin fusion protein, the par-ticles could be taken up by Caco-2 cells and transported acrossa polarized monolayer (Zhu et al., 2004). These data suggestthat adenoviruses can be redirected to the transcytosis path-way, although it remains to be established how efficiently thisroute can be recruited in the tumor endothelium.

Targeting Adenoviruses

Many strategies have been pursued to improve genetransfer into CAR-negative cells. In addition to the use of non-CAR-binding serotypes and fiber-swap vectors (Murakamiet al., 2009; Sandberg et al., 2009), recombinant HAdV vectorswith altered tropism have also been generated by engineeringnew ligands for cellular receptors into surface loops of capsidcomponents. The favorite locations have been the C terminusand the HI loop of the knob domain of the fiber, the RGD loopof the penton base protein, and the L1 loop of the hexon.Although effective, the applicability of this approach wasinitially limited by the restricted tolerance for inserting newligands at these positions (Arnberg, 2009). In addition, newligands that are to be incorporated genetically into adenoviralvectors must be able to fold correctly in the reducing envi-ronment of the mammalian cell cytoplasm. This excludesmost ligands dependent on disulfide bond formation forproper folding such as epidermal growth factor and mostsingle-chain variable fragments (Lindholm et al., 2008).Promising candidate ligands are Affibody� molecules (Affi-body AB, Bromma, Sweden), which are affinity proteinsbased on a 58-amino acid three-helix bundle structure, termed‘‘Z,’’ that is derived from the immunoglobulin-binding do-

main of staphylococcal protein A. Display libraries have beenconstructed on the basis of randomization of 13 surface ac-cessible amino acids in the Z domain, from which novel Af-fibody molecules to desired targets have been selected. TheseAffibodies can be efficiently used for genetic retargeting ofadenovirus (Henning et al., 2002; Lindholm et al., 2008). Al-though initial experiments were thwarted by structural con-straints, Lindholm and collaborators managed to insert twoAffibodies in tandem in the HI loop, by connecting them viasmall flexible linkers. This resulted in HAdV-5 vectors ge-netically retargeted with a HER2=neu-specific Affibody mol-ecule inserted in the HI loop of the fiber knob of a CARbinding-ablated fiber (Magnusson et al., 2007). With thistechnology, vectors can be generated by incorporating twoAffibody molecules with different specificities (Myhre et al.,2009). Camelid and human single-domain antibody frag-ments may be applicable in a similar manner (Harmsen andDe Haard, 2007).

In another approach, the knob and shaft of the fiber havebeen replaced by an artificial trimerization domain, whichwas linked to an heterologous receptor-binding ligand(Krasnykh et al., 2001; Magnusson et al., 2001; Schagen et al.,2008). Via this strategy Willemsen and co-workers managedto retarget HAdV-5 to tumor cells by replacing the shaft andknob of HAdV-5 by a single-chain T cell receptor specific forHLA-A1 molecules that present a MAGE-A1 peptide (Sebes-tyen et al., 2007). Similar single-chain T cell receptors havebeen fused to the C terminus of the minor capsid proteinIX (de Vrij et al., 2008). Immunoaffinity studies suggested thatthe C termini of protein IX molecules are positioned nearthe capsid surface (Akalu et al., 1999; Vellinga et al., 2005b).This has been confirmed by cryoelectron microscopy stud-ies, which suggest that the C termini are located near theperipentonal hexons. Here the leucine zipper domain inthe C-terminal part of protein IX interacts with the zipperof other molecules, forming a coiled coil (Scheres et al.,2005; Saban et al., 2006; Fabry et al., 2009). These protein IX–protein IX associations are not necessary for pIX–capsid in-corporation and thermostability of the particles (Vellinga et al.,2005a).

If ligands are to be fused with pIX, a spacer may expose theligands above the outer surface of hexon capsomers to ensureits accessibility to cellular receptors. Vellinga and co-workershave demonstrated that linkers up to a length of 75 A can beadded to the C terminus of protein IX without affecting theincorporation of protein IX into the capsid (Vellinga et al.,2004). Indeed, a wide variety of targeting polypeptides couldbe functionally incorporated by genetic fusion at the C ter-minus of protein IX. In this way small targeting peptides, ahyperstable single-chain Fv, and a single-chain T cell receptorcould be functionally incorporated into the capsid (Vellingaet al., 2004, 2006, 2007; de Vrij et al., 2008). In addition, otherfunctional proteins were incorporated in the adenoviral cap-sid through linkage to pIX, such as fluorescent proteins thatallow particle tracing by fluorescence microscopy (Le et al.,2004; Meulenbroek et al., 2004).

An elegant combination of genetic modification andchemical modification has been developed by Kreppel andco-workers. HAdV-5 vectors were genetically modified tocontain cysteines at solvent-exposed positions in the capsid(Kreppel et al., 2005). The introduced thiol groups are highlyreactive, and procedures were established for their controlled

798 DE VRIJ ET AL.

covalent coupling to protein and nonprotein ligands.Depending on the chemistry used, ligands could be coupledby formation of thioether or disulfide bonds. The lattermethod yields viruses that release the coupled ligand in theendosome. In addition, thiol groups in the fiber knob werestill accessible after amino PEGylation, allowing PEG shield-ing to be combined with targeting by ligand coupling to thethiol groups (Kreppel et al., 2005). Coupling of transferrin toengineered cysteine residues at the C terminus of protein IXallowed targeting of HAdV particles in mice in vivo (Corjonet al., 2008). This validates the applicability of the technique inprocedures involving intravenous administrations of adeno-viral vectors.

Taken together, these data show that HAdV vector tech-nology has matured and constitutes a functional and robustplatform for generating shielded, retargeted vectors that canbe used for developing oncolytic HAdV vectors for cancergene therapy.

Targetable Receptors in Prostate Cancer

With the targeting platform in place, a key question con-cerns which receptors could be targeted in prostate cancer.Extensive target exploration has yielded several cell surfacereceptors that are expressed preferentially or specifically inprostate cancer cells.

A prime candidate is prostate-specific membrane antigen(PSMA). It is expressed both on benign and malignant pros-tate cells. It is a type 2 membrane receptor, which can beefficiently internalized. The ligands that trigger internaliza-tion remain to be identified (Wang et al., 2007). Nearly allprostate tumors and prostate cancer cells express PSMA andincreased expression correlates with aggressive tumors(Tasch et al., 2001). PSMA is also upregulated after androgendeprivation in model systems whereas other markers such asprostate-specific antigen (PSA) are decreased after androgenwithdrawal (Israeli et al., 1994). PSMA can serve as a tissue-specific target for adenoviral vectors (Kraaij et al., 2005). Re-targeting of viral particles to prostate cancer cell lines wasobtained through the attachment of bispecific molecules,which consisted of conjugates between an anti-adenoviralfiber knob Fab0 fragment and anti-PSMA monoclonal anti-bodies (Kraaij et al., 2005).

Besides PSMA, various other cell surface molecules havebeen demonstrated to be upregulated in prostate cancer, suchas prostate stem cell antigen (PSCA). Successful targeting toPSCA-expressing prostate cancer cells has been achieved forgenetically engineered T cells that have been equipped with achimeric T cell receptor recognizing PSCA (Morgenroth et al.,2007). These findings support the exploration of PSCA tar-geting in the context of prostate cancer-targeted oncolyticviruses.

The urokinase-type plasminogen activator receptor (uPAR)has also been exploited for targeting of oncolytic virusesto prostate cancer cells. uPAR is overexpressed in tumors aswell as in stromal cells of multiple malignancies, includingprostate cancer (Romer et al., 2004; Li and Cozzi, 2007). uPARis involved in tumor angiogenesis. For example, tumor cell-conditioned media can upregulate endothelial uPAR expres-sion (Seghezzi et al., 1996). The principle of targeting tumorendothelium by aiming at uPAR, rather than at the cancercells themselves, is highly attractive, because such an ap-

proach may lead to improved viral trafficking from thebloodstream into the tumor tissue. HAdV targeting to uPARis feasible (Drapkin et al., 2000). More recently, tumor andvascular targeting of an oncolytic measles virus has beenachieved ( Jing et al., 2009).

In addition, other cell surface molecules are upregulated inthe tumor vasculature, including members of the vascularendothelial growth factor receptor (VEGFR) family. BothVEGFR1 (Flt-1) and VEGFR2 (Flk-1) are selectively expressedon endothelial cells and are highly upregulated in prolifer-ating (angiogenic) capillary cells of numerous types ofprostate tumor (Kollermann and Helpap, 2001). The potentialof VEGFR2 as a target has been shown in studies on systemictargeting of drug-loaded microspheres to subcutaneousprostate tumors in mice, which demonstrated significant in-hibition of tumor growth after conjugation of anti-VEGFR2antibodies to the microspheres (Lu et al., 2008).

One class of potentially specific receptors is the group oftumor-specific cancer testis (CT) antigens (Costa et al., 2007).Peptides of these CT antigens are presented on the cell surfacein complex with major histocompatibility class (MHC) Imolecules. With the exception of their expression in the testis,an immune-privileged site due to the absence of MHC ex-pression, the CT antigens are expressed exclusively in cancercells. Proof-of-principle of targeting adenoviral vectors to CTantigens has been shown by genetically fusing viral capsidproteins with single-chain T cell receptors, which could spe-cifically recognize the melanoma-specific CT antigen MAGE-A1 in complex with HLA-A1 (Sebestyen et al., 2007; de Vrijet al., 2008). Multiple cancer CT antigens have been found inpatients with prostate cancer, including SSX-2 and MAD-CT-1 and -2 (Hoeppner et al., 2006; Dubovsky and McNeel, 2007).A peptide present in the majority of MAGE-A gene familymembers could serve as an ideal target as most tumors, bothsolid and blood-borne, express at least one member of thisMAGE-A gene family (Scanlan et al., 2002). In addition to cellsurface antigens, intracellular antigens, for example, PSA,have now become available for targeting and warrant furtherexploration as targets for gene therapy vectors (Hoeppneret al., 2006; Dubovsky and McNeel, 2007). Combining the se-lective power of phage display, which allows for the testing oftens of billions of individual clones, with high-throughputselection of Fabs with peptide–MHC complex-binding ca-pacity will yield new human ‘‘T cell receptor (TCR)-like’’ Fabfragments that specifically target viruses to tumor cells ex-pressing intracellular tumor antigens (Willemsen et al., 2008).

Prostate-Targeted ConditionallyReplicative Adenoviruses

HAdVs have been generated that replicate specifically incertain cell types. In HAdV infection, the E1A gene acts as themaster switch that activates the viral gene expression cascade.Therefore, by controlling E1A expression with a tumor- ortissue-specific promoter, viral replication can be restricted tocertain cell types. Along these lines, Essand and collaboratorsdeveloped a series of prostate-specific adenoviruses thatreplicate exclusively in normal and neoplastic prostate epi-thelial cells (Cheng et al., 2006; Dzojic et al., 2007; Danielssonet al., 2008). In these vectors expression of E1A is controlledby the recombinant prostate-specific PPT sequence. The PPTsequence is composed of a prostate-specific antigen (PSA)

Ad-DERIVED VECTORS FOR PROSTATE CANCER GENE THERAPY 799

enhancer, a prostate-specific membrane antigen (PSMA)enhancer, and a T cell receptor g-chain alternate reading frameprotein (TARP) promoter. The PSMA enhancer, which upre-gulates PSMA expression in androgen-depleted prostatecancer cells, also ensures that the PPT sequence is active underandrogen-deprived conditions. The mouse H19 insulator (I),with enhancer-blocking activity, was placed upstream of PPTto protect it from interfering signals from the adenoviralbackbone (Cheng et al., 2006; Dzojic et al., 2007). The mostadvanced version, the so-called Ad[i=PPT-E1A, E3] virus,induced regression of aggressively growing LNCaP tumors,and yielded significantly prolonged survival for treated micecompared with the control groups (Danielsson et al., 2008).

Not only can E1 regulation be used to restrict replicationto cancer cells, but the strategy has also been used to preventexpression of E1A, and thereby expression of other viralgenes, in sensitive nontarget tissues. HAdV5 vectors canmediate significant hepatotoxicity. To prevent viral geneexpression in hepatocytes, multiple binding sites for ahepatocyte-specific microRNA, miR-122, were placed in the30 untranslated region of the E1A gene (Ylosmaki et al., 2008;Cawood et al., 2009). miR-122 is highly and selectively ex-pressed in hepatocytes and this modification might preventexpression of E1A within hepatocytes, and hepatotoxicity,while maintaining its replicative capacity in tumor cells.Animals receiving a lethal dose of wild-type Ad5 (5�1010

viral particles=mouse) showed substantial hepatic genomereplication and extensive liver pathology, whereas inclusionof miR-122 binding sites decreased replication 50-fold andvirtually abrogated liver toxicity, demonstrating the effi-ciency of the approach (Cawood et al., 2009). These examplesdemonstrate that we can endow replicating HAdV vectorswith tumor cell selectivity, while protecting sensitive non-target tissues.

New Directions

A robust technology platform for cancer-targeted and on-colytic HAdV vector generation has been established. Manyelegant tools and techniques have been developed with well-chosen experiments to show their proof-of-concept. However,it seems fair to state that most improvements have been in-cremental and only a few of the vectors that showed promisein preclinical studies have reached the stage of clinical eval-uation. So, where do we go from here? In what fields are newdevelopments necessary to fulfill the promise of efficaciousprostate cancer-targeted HAdV vectors? It may be useful tostep back, reflect, and place our activities in perspective.

Rational design or evolution of new oncolytic viruses?

So far, most vectorologists have followed a ‘‘rational design’’or reverse-genetics approach for building new therapeuticvectors. In other words, on the basis of a priori knowledge ofvirus biology, tumor cell biology, and pharmaceutical pa-rameters, we have built our new vectors to have the desiredphenotype and to perform as anticipated. This approachhas been most useful and has delivered most of the vectors thatare in use in clinical gene therapy to date. Nevertheless,one should realize that this is not the classical approach inmicrobiology.

Classical virology studies viruses by employing selectionstrategies to isolate mutants with desired phenotypes and to

study these to obtain insight into viral genetics and virusbiology. There has been a revival of interest in the classical,more evolutionary approaches involving bioselection strate-gies for developing improved HAdV oncolytic vectors. Yanand co-workers have provided a fine example of the power ofthis approach (Yan et al., 2003). In vitro chemical mutagenesisof HAdV-5 was combined with a bioselection strategy. Thisyielded a mutant virus that replicated more efficient in theHT29 colorectal tumor cells that were used for selection. Themutation truncates an open reading frame in the late i-leadertranscript. Not only in HT29, but also in several other tumorcell lines, replication was enhanced by the causative i-leadermutation at nucleotide 8350 of the HAdV-5 genome (Yan et al.,2003). Along similar lines, Gros and co-workers used in vivobioselection and obtained an E3=19K mutant with enhancedantitumoral potency (Gros et al., 2008). By selecting forHAdVs with large-plaque phenotypes, Subramanian and co-workers isolated a series of mutants, including mutants in thei-leader and in the E3=19K gene (Subramanian et al., 2006).Taken together, these data suggest that the oncolytic activityof wild-type HAdV-5 can be enhanced by mutations.

Hermiston and collaborators took this approach a stepfurther (Kuhn et al., 2008). On the basis of the notion that thereis no evidence that HAdV-5 is the optimal start point for se-lecting more potent oncolytic HAdVs, they pooled an array ofHAdV serotypes. These pools were passaged under condi-tions that invite recombination between serotypes. They iso-lated a mutant, designated coloAd1, which replicates moreefficiently than any wild-type virus in human colon cancercells. Characterization of coloAd1 revealed that it is a complexhybrid between HAdV-3 and -11p (Kuhn et al., 2008). It isevident that none of the bioselected viruses would have beencreated on the basis of preexisting knowledge. In fact, we haveno clear understanding why these mutants replicate so muchbetter in the cell systems that were used for their isolation.This underscores the potential of the classical strategy. It is tobe expected that evolutionary approaches involving biose-lections will become more widely applied. These approacheswill yield new viruses, and the characterization of theseviruses may provide new insights concerning critical aspectsof virus and tumor cell biology. Uil and co-workers presentedan approach that employs modified adenoviral polymeraseswith deficits in the polymerase proofreading function. Thisstrategy will facilitate the isolation of more complex HAdVmutants for cancer gene therapy (Uil et al., 2009).

Taken together, it seems reasonable to anticipate that ad-enoviral mutations that have been isolated by bioselectionprocedures will soon be incorporated in clinically applicableoncolytic adenoviruses.

Cellular delivery of viruses

Many studies have demonstrated that the delivery of vec-tor particles into tumors is inefficient. Both intratumoral andvascular delivery are thwarted by a variety of factors (seeabove), and therefore new delivery methods are essential.An attractive option is to use cells with the capacity to migrateto tumors as delivery vehicles for oncolytic viruses. In thisstrategy the tumor-targeting cells are loaded with viruses andadministered to the patient. After migration the cells shouldhand off the cell-associated viruses or, if the virus replicates inthe delivery cells, their progeny. In this way viruses should be

800 DE VRIJ ET AL.

delivered to tumor cells. Several cell types can migrate totumors in vivo, including cytokine-induced killer cells, tumorantigen-specific T cells, macrophages, endothelial progenitorcells, and mesenchymal stem cells (Power and Bell, 2008). Inaddition, virus-loaded dendritic cells have been used toeradicate tumor cells from tumor-draining lymph nodes (Ilettet al., 2009).

Cellular delivery of viral anticancer agents may also cir-cumvent the effects of neutralizing immunity (Power et al.,2007). This is evidenced in a study employing human re-oviruses as the oncolytic agent. In reovirus-immune mice withB16tk lymph node melanoma metastases, in vivo delivery offree reovirus to the melanoma was ineffective, whereas ef-fective antitumor responses and long-term tumor clearancewere obtained if mature dendritic cells as well as T cells wereused as carriers (Ilett et al., 2009). This and other studiessuggest that cellular delivery is feasible and may be applicablefor the delivery of viral oncolytic agents at tumor sites. Itshould be noted that cellular delivery adds a new level ofcomplexity to clinical gene therapy studies and is logisticallychallenging.

Tumor stem cells as new targets?

With the aid of such new technologies, that is, thebioselection-based strategies for vector improvement, andcell-based methods for vector delivery, new cancer genetherapeutics and therapeutic strategies may be developed thatensure efficient vector delivery into the tumors. It is knownthat tumors are usually markedly heterogeneous, consistingof complex mixtures of cancer cells with various grades ofdifferentiation. A cell type that has attracted much attentionis the tumor-initiating cell, or cancer stem cell. The tumor-initiating cell is a cell with the capacity to self-renew anddifferentiate into any of the lineages of cancer cells thatcomprise a tumor (Lobo et al., 2007). Presumably the tumor-initiating cells are derived from organ stem cells (Collins andMaitland, 2009). The latter are long-lived and change over thecourse of time by accumulating (epi)genetic alterations. In thismodel, the characteristics of a cancer-initiating cell, andtherefore of the resulting tumor, depend (in part) on thesealterations. The cancer stem cell population in a tumor maygovern crucial tumor processes such as progression, invasion,and metastasis. Therapy resistance may be the consequence ifa particular treatment does not effectively eradicate the cancerstem cell population. It may therefore be important to ensurethat new oncolytic agents have the capacity to transduce andkill cancer stem cells.

Nonviral delivery of viral genomes

Systemic approaches for cancer gene therapy have beenfocused on delivering viruses to the tumor. With advances inthe field of nonviral gene delivery, an alternative approachbecame feasible (Carlisle et al., 2006). Rather than deliveringintact viruses to the tumor cells, a strategy is followed inwhich viral genomes are delivered. If transferred into cells,HAdV DNA can yield replicating virus. Although not par-ticularly efficient, infectious HAdV can be reproducibly re-covered from cell cultures on transfer of naked DNA.

The field of nonviral gene transfer has seen impressiveadvances in increased tumor accumulation of transgenes(Wolff and Rozema, 2008), but so far the therapeutic conse-

quences remain to be improved. Individual steps in genedelivery need further improvement, especially those relatingto intracellular trafficking of the complexes. This includes thetimely release of the DNA complex from the targeting ligand,efficient escape from the endosomal compartment, and properdelivery of the genes into the nucleus (Ogris et al., 2007; Russet al., 2008; Schwerdt et al., 2008). Furthermore, innate im-munity must be evaded. Integration of controlled-releasetechnologies into targeted gene delivery systems will providemore effective gene delivery systems (Meyer and Wagner,2006; Philipp et al., 2008; Schaffert and Wagner, 2008).

With the delivery of viral genomes we could benefit fromthe best of two worlds. Nonviral vectors may be easier toproduce and formulate, and may deliver their payload morereproducibly at the tumor site than viral vectors. If used forthe delivery of viral genomes, tumor-selective replicatingviruses may be generated on site, which can spread in thetumor and exert their therapeutic action, without causingcollateral damage to nontarget tissues. It remains doubtful,however, that HAdV is the best choice of virus for suchstrategies.

New therapeutic genes

New transgenes in the vectors may enhance therapeuticefficacy. An intriguing class of prodrug-activating genes isbased on deoxyribonucleoside kinases (dNKs). These en-zymes catalyze the phosphorylation of deoxyribonucleosidesto deoxyribonucleoside monophosphates and thereby pro-vide the cell with deoxyribonucleoside triphosphates. ThedNKs catalyze the first, and often rate-limiting, step of nu-cleoside analog activation. These enzymes are thereforepromising candidates to be used in combination with nucle-otide analogs as prodrug–enzyme combinations. Bioselectionyielded mutants of the Drosophila melanogaster-derived dNKwith enhanced sensitivity to a range of clinically approvednucleotide analogs, such as 30-azido-30-deoxythymidine(AZT), arabinosylcytosine (AraC), ddA, and ddC (Knechtet al., 2000, 2007). These mutants efficiently sensitized glio-blastoma, osteosarcoma, and breast cancer cells to clinicallyaccepted drugs (Knecht et al., 2007). The use of approvednucleoside analogs may facilitate swift acceptance of thestrategy.

Future prospects

We have witnessed an enormous expansion of the genetransfer technology required for cancer gene therapy. Initialclinical safety studies have demonstrated the validity of theconcept, and the feasibility of current approaches (Schenket al., 2009). The viral vectors used in the clinical studies re-ported so far have been well tolerated and safe. Robust tech-nology platforms have been established and many of thefactors currently thwarting prostate cancer gene therapy havebeen identified. New platforms for preclinical evaluation ofnew oncolytic vectors are available (Maitland et al., 2009).Combining the technologies and building on new insightsfrom prostate cancer biology and virology will facilitate thegeneration of new vectors that will, it is hoped, be as safe ascurrent vectors, but more efficacious. With these we may keepthe promise of gene therapists to provide new and effectivetreatment for malignant neoplastic disease in general andprostate cancer in particular.

Ad-DERIVED VECTORS FOR PROSTATE CANCER GENE THERAPY 801

Acknowledgment

This work was supported by the European Union throughthe 6th Framework Program GIANT (contract no. 512087).

Author Disclosure statement

Jeroen de Vrij, Ralph A. Willemsen, and Rob C. Hoebendeclare no competing financial interests. Leif Lindholm is ashareholder of Got-a-Gene, which has intellectual propertyon adenovirus targeting technology.

References

Abbink, P., Lemckert, A.A., Ewald, B.A., Lynch, D.M., Denholtz,M., Smits, S., Holterman, L., Damen, I., Vogels, R., Thorner,A.R., O’Brien, K.L., Carville, A., Mansfield, K.G., Goudsmit, J.,Havenga, M.J., and Barouch, D.H. (2007). Comparative ser-oprevalence and immunogenicity of six rare serotype recom-binant adenovirus vaccine vectors from subgroups B and D.J. Virol. 81, 4654–4663.

Akalu, A., Liebermann, H., Bauer, U., Granzow, H., and Seidel,W. (1999). The subgenus-specific C-terminal region of proteinIX is located on the surface of the adenovirus capsid. J. Virol.73, 6182–6187.

Alba, R., Bradshaw, A.C., Parker, A.L., Bhella, D., Waddington,S.N., Nicklin, S.A., van Rooijen, N., Custers, J., Goudsmit, J.,Barouch, D.H., McVey, J.H., and Baker, A.H. (2009). Identifi-cation of coagulation factor (F)X binding sites on the adeno-virus serotype 5 hexon: Effect of mutagenesis on FXinteractions and gene transfer. Blood 114, 965–971.

Arnberg, N. (2009). Adenovirus receptors: Implications for tro-pism, treatment and targeting. Rev. Med. Virol. 19, 165–178.

Bachtarzi, H., Stevenson, M., and Fisher, K. (2008). Cancer genetherapy with targeted adenoviruses. Expert Opin. Drug Deliv.5, 1231–1240.

Carlisle, R.C., Briggs, S.S., Hale, A.B., Green, N.K., Fisher, K.D.,Etrych, T., Ulbrich, K., Mautner, V., and Seymour, L.W. (2006).Use of synthetic vectors for neutralising antibody resistantdelivery of replicating adenovirus DNA. Gene Ther. 13, 1579–1586.

Carlisle, R.C., Di, Y., Cerny, A.M., Sonnen, A.F., Sim, R.B.,Green, N.K., Subr, V., Ulbrich, K., Gilbert, R.J., Fisher, K.D.,Finberg, R.W., and Seymour, L.W. (2009). Human erythro-cytes bind and inactivate type 5 adenovirus by presentingcoxsackie virus–adenovirus receptor and complement recep-tor 1. Blood 113, 1909–1918.

Cawood, R., Chen, H.H., Carroll, F., Bazan-Peregrino, M., vanRooijen, N., and Seymour, L.W. (2009). Use of tissue-specificmicroRNA to control pathology of wild-type adenoviruswithout attenuation of its ability to kill cancer cells. PLoSPathog. 5, e1000440.

Cheng, W.S., Dzojic, H., Nilsson, B., Totterman, T.H., and Es-sand, M. (2006). An oncolytic conditionally replicating ade-novirus for hormone-dependent and hormone-independentprostate cancer. Cancer Gene Ther. 13, 13–20.

Collins, A.T., and Maitland, N.J. (2009). Prostate cancer:Regeneration of interest in the prostate. Nat. Rev. Urol. 6, 184–186.

Corjon, S., Wortmann, A., Engler, T., van Rooijen, N., Kochanek,S., and Kreppel, F. (2008). Targeting of adenovirus vectors tothe LRP receptor family with the high-affinity ligand RAP viacombined genetic and chemical modification of the pIX cap-somere. Mol. Ther. 16, 1813–1824.

Costa, F.F., Le, B.K., and Brodin, B. (2007). Concise review:Cancer=testis antigens, stem cells, and cancer. Stem Cells 25,707–711.

Danielsson, A., Dzojic, H., Nilsson, B., and Essand, M. (2008).Increased therapeutic efficacy of the prostate specific oncolyticadenovirus Ad[I=PPT-E1A] by reduction of the insulator sizeand introduction of the full-length E3 region. Cancer GeneTher. 15, 203–213.

de Vrij, J., Uil, T.G., van den Hengel, S.K., Cramer, S.J., Koppers-Lalic, D., Verweij, M.C., Wiertz, E.J., Vellinga, J., Willemsen,R.A., and Hoeben, R.C. (2008). Adenovirus targeting to HLA-A1=MAGE-A1-positive tumor cells by fusing a single-chainT-cell receptor with minor capsid protein IX. Gene Ther. 15,978–989.

Di Paolo, N.C., van Rooijen, N., and Shayakhmetov, D.M. (2009).Redundant and synergistic mechanisms control the seques-tration of blood-born adenovirus in the liver. Mol. Ther. 17,675–684.

Doronin, K., Shashkova, E.V., May, S., Hofherr, S., and Barry,M.A. (2009). Chemical modification with high molecularweight polyethylene glycol reduces transduction of hepato-cytes and increases efficacy of intravenously delivered onco-lytic adenovirus. Hum. Gene Ther. 20, 975–988.

Drapkin, P.T., O’Riordan, C.R., Yi, S.M., Chiorini, J.A., Cardella,J., Zabner, J., and Welsh, M.J. (2000). Targeting the urokinaseplasminogen activator receptor enhances gene transfer tohuman airway epithelia. J. Clin. Invest. 105, 589–596.

Dubovsky, J.A., and McNeel, D.G. (2007). Inducible expressionof a prostate cancer-testis antigen, SSX-2, following treatmentwith a DNA methylation inhibitor. Prostate 67, 1781–1790.

Dzojic, H., Cheng, W.S., and Essand, M. (2007). Two-step am-plification of the human PPT sequence provides specific geneexpression in an immunocompetent murine prostate cancermodel. Cancer Gene Ther. 14, 233–240.

Fabry, C.M., Rosa-Calatrava, M., Moriscot, C., Ruigrok, R.W.,Boulanger, P., and Schoehn, G. (2009). The C terminal do-mains of adenovirus serotype 5 protein IX assemble into anantiparallel structure on the facets of the capsid. J. Virol. 83,1135–1139.

Figueiredo, M.L., Kao, C., and Wu, L. (2007). Advances in pre-clinical investigation of prostate cancer gene therapy. Mol.Ther. 15, 1053–1064.

Fisher, K.D., Stallwood, Y., Green, N.K., Ulbrich, K., Mautner, V.,and Seymour, L.W. (2001). Polymer-coated adenovirus permitsefficient retargeting and evades neutralising antibodies. GeneTher. 8, 341–348.

Freytag, S.O., Stricker, H., Movsas, B., and Kim, J.H. (2007).Prostate cancer gene therapy clinical trials. Mol. Ther. 15, 1042–1052.

Geutskens, S.B., van der Eb, M.M., Plomp, A.C., Jonges, L.E.,Cramer, S.J., Ensink, N.G., Kuppen, P.J., and Hoeben, R.C.(2000). Recombinant adenoviral vectors have adjuvant activityand stimulate T cell responses against tumor cells. Gene Ther.7, 1410–1416.

Green, N.K., Herbert, C.W., Hale, S.J., Hale, A.B., Mautner, V.,Harkins, R., Hermiston, T., Ulbrich, K., Fisher, K.D., andSeymour, L.W. (2004). Extended plasma circulation time anddecreased toxicity of polymer coated adenovirus. Gene Ther.11, 1256–1263.

Gregorevic, P., Blankinship, M.J., Allen, J.M., Crawford, R.W.,Meuse, L., Miller, D.G., Russell, D.W., and Chamberlain, J.S.(2004). Systemic delivery of genes to striated muscles usingadeno-associated viral vectors. Nat. Med. 10, 828–834.

802 DE VRIJ ET AL.

Gros, A., Martinez-Quintanilla, J., Puig, C., Guedan, S., Mollevi,D.G., Alemany, R., and Cascallo, M. (2008). Bioselection of a gainof function mutation that enhances adenovirus 5 release andimproves its antitumoral potency. Cancer Res. 68, 8928–8937.

Harmsen, M.M., and De Haard, H.J. (2007). Properties, pro-duction, and applications of camelid single-domain antibodyfragments. Appl. Microbiol. Biotechnol. 77, 13–22.

Henning, P., Magnusson, M.K., Gunneriusson, E., Hong, S.S.,Boulanger, P., Nygren, P.A., and Lindholm, L. (2002). Geneticmodification of adenovirus 5 tropism by a novel class ofligands based on a three-helix bundle scaffold derived fromstaphylococcal protein A. Hum. Gene Ther. 13, 1427–1439.

Hoeppner, L.H., Dubovsky, J.A., Dunphy, E.J., and McNeel,D.G. (2006). Humoral immune responses to testis antigens insera from patients with prostate cancer. Cancer Immun. 6, 1.

Hofherr, S.E., Mok, H., Gushiken, F.C., Lopez, J.A., and Barry,M.A. (2007). Polyethylene glycol modification of adenovirusreduces platelet activation, endothelial cell activation, andthrombocytopenia. Hum. Gene Ther. 18, 837–848.

Hofherr, S.E., Shashkova, E.V., Weaver, E.A., Khare, R., andBarry, M.A. (2008). Modification of adenoviral vectors withpolyethylene glycol modulates in vivo tissue tropism and geneexpression. Mol. Ther. 16, 1276–1282.

Ilett, E.J., Prestwich, R.J., Kottke, T., Errington, F., Thompson,J.M., Harrington, K.J., Pandha, H.S., Coffey, M., Selby, P.J.,Vile, R.G., and Melcher, A.A. (2009). Dendritic cells and T cellsdeliver oncolytic reovirus for tumour killing despite pre-existing anti-viral immunity. Gene Ther. 16, 689–699.

Israeli, R.S., Powell, C.T., Corr, J.G., Fair, W.R., and Heston,W.D. (1994). Expression of the prostate-specific membraneantigen. Cancer Res. 54, 1807–1811.

Jing, Y., Tong, C., Zhang, J., Nakamura, T., Iankov, I., Russell,S.J., and Merchan, J.R. (2009). Tumor and vascular targeting ofa novel oncolytic measles virus retargeted against the uroki-nase receptor. Cancer Res. 69, 1459–1468.

Kalyuzhniy, O., Di Paolo, N.C., Silvestry, M., Hofherr, S.E., Barry,M.A., Stewart, P.L., and Shayakhmetov, D.M. (2008). Adeno-virus serotype 5 hexon is critical for virus infection of hepato-cytes in vivo. Proc. Natl. Acad. Sci. U.S.A. 105, 5483–5488.

Knecht, W., Munch-Petersen, B., and Piskur, J. (2000). Identifi-cation of residues involved in the specificity and regulation ofthe highly efficient multisubstrate deoxyribonucleoside kinasefrom Drosophila melanogaster. J. Mol. Biol. 301, 827–837.

Knecht, W., Rozpedowska, E., Le, B.C., Willer, M., Gojkovic, Z.,Sandrini, M.P., Joergensen, T., Hasholt, L., Munch-Petersen,B., and Piskur, J. (2007). Drosophila deoxyribonucleoside kinasemutants with enhanced ability to phosphorylate purine ana-logs. Gene Ther. 14, 1278–1286.

Kollermann, J., and Helpap, B. (2001). Neuroendocrine differ-entiation and short-term neoadjuvant hormonal treatment ofprostatic carcinoma with special regard to tumor regression.Eur. Urol. 40, 313–317.

Kraaij, R., van Rijswijk, A.L., Oomen, M.H., Haisma, H.J., andBangma, C.H. (2005). Prostate specific membrane antigen(PSMA) is a tissue-specific target for adenoviral transductionof prostate cancer in vitro. Prostate 62, 253–259.

Krasnykh, V., Belousova, N., Korokhov, N., Mikheeva, G., andCuriel, D.T. (2001). Genetic targeting of an adenovirus vectorvia replacement of the fiber protein with the phage T4 fibritin.J. Virol. 75, 4176–4183.

Kreppel, F., and Kochanek, S. (2008). Modification of adenovirusgene transfer vectors with synthetic polymers: A scientificreview and technical guide. Mol. Ther. 16, 16–29.

Kreppel, F., Gackowski, J., Schmidt, E., and Kochanek, S. (2005).Combined genetic and chemical capsid modifications enableflexible and efficient de- and retargeting of adenovirus vectors.Mol. Ther. 12, 107–117.

Kritz, A.B., Nicol, C.G., Dishart, K.L., Nelson, R., Holbeck, S.,Von Seggern, D.J., Work, L.M., McVey, J.H., Nicklin, S.A., andBaker, A.H. (2007). Adenovirus 5 fibers mutated at the puta-tive HSPG-binding site show restricted retargeting with tar-geting peptides in the HI loop. Mol. Ther. 15, 741–749.

Kuhn, I., Harden, P., Bauzon, M., Chartier, C., Nye, J., Thorne, S.,Reid, T., Ni, S., Lieber, A., Fisher, K., Seymour, L., Rubanyi,G.M., Harkins, R.N., and Hermiston, T.W. (2008). Directedevolution generates a novel oncolytic virus for the treatmentof colon cancer. PLoS One 3, e2409.

Kuppen, P.J., van der Eb, M.M., Jonges, L.E., Hagenaars, M.,Hokland, M.E., Nannmark, U., Goldfarb, R.H., Basse, P.H.,Fleuren, G.J., Hoeben, R.C., and van de Velde, C.J. (2001).Tumor structure and extracellular matrix as a possiblebarrier for therapeutic approaches using immune cells oradenoviruses in colorectal cancer. Histochem. Cell Biol. 115,67–72.

Le, L.P., Everts, M., Dmitriev, I.P., Davydova, J.G., Yamamoto,M., and Curiel, D.T. (2004). Fluorescently labeled adenoviruswith pIX-EGFP for vector detection. Mol. Imaging 3, 105–116.

Li, Y., and Cozzi, P.J. (2007). Targeting uPA=uPAR in prostatecancer. Cancer Treat. Rev. 33, 521–527.

Li, Z.Y., Ni, S., Yang, X., Kiviat, N., and Lieber, A. (2004).Xenograft models for liver metastasis: Relationship betweentumor morphology and adenovirus vector transduction. Mol.Ther. 9, 650–657.

Lindholm, L., Henning, P., and Magnusson, M.K. (2008). Novelstrategies in tailoring human adenoviruses into therapeuticcancer gene-therapy vectors. Future Virol. 3, 45–59.

Lobo, N.A., Shimono, Y., Qian, D., and Clarke, M.F. (2007). Thebiology of cancer stem cells. Annu. Rev. Cell Dev. Biol. 23,675–699.

Lu, J., Jackson, J.K., Gleave, M.E., and Burt, H.M. (2008). Thepreparation and characterization of anti-VEGFR2 conjugated,paclitaxel-loaded PLLA or PLGA microspheres for the sys-temic targeting of human prostate tumors. Cancer Chemother.Pharmacol. 61, 997–1005.

Magnusson, M.K., Hong, S.S., Boulanger, P., and Lindholm, L.(2001). Genetic retargeting of adenovirus: Novel strategy em-ploying ‘‘deknobbing’’ of the fiber. J. Virol. 75, 7280–7289.

Magnusson, M.K., Henning, P., Myhre, S., Wikman, M., Uil,T.G., Friedman, M., Andersson, K.M., Hong, S.S., Hoeben,R.C., Habib, N.A., Stahl, S., Boulanger, P., and Lindholm, L.(2007). Adenovirus 5 vector genetically re-targeted by an Af-fibody molecule with specificity for tumor antigen HER2=neu.Cancer Gene Ther. 14, 468–479.

Maitland, N.J., Chambers, K., Georgopoulos, L., Simpson-Holley, M., Leadley, R., Evans, H., Essand, M., Danielsson, A.,van Weerden, W., de Ridder, C., Kraaij, R., Bangma, C.H., andmembers of the GIANT FP6 Consortium. (2009). Gene transfervectors targeted to human prostate cancer: Do we need betterpreclinical testing systems? Hum. Gene Ther. 21, 815–827.

Matsumura, Y., and Maeda, H. (1986). A new concept for mac-romolecular therapeutics in cancer chemotherapy: Mechanismof tumoritropic accumulation of proteins and the antitumoragent smancs. Cancer Res. 46, 6387–6392.

McKee, T.D., Grandi, P., Mok, W., Alexandrakis, G., Insin, N.,Zimmer, J.P., Bawendi, M.G., Boucher, Y., Breakefield, X.O.,and Jain, R.K. (2006). Degradation of fibrillar collagen in a

Ad-DERIVED VECTORS FOR PROSTATE CANCER GENE THERAPY 803

human melanoma xenograft improves the efficacy of an on-colytic herpes simplex virus vector. Cancer Res. 66, 2509–2513.

Meulenbroek, R.A., Sargent, K.L., Lunde, J., Jasmin, B.J., andParks, R.J. (2004). Use of adenovirus protein IX (pIX) to dis-play large polypeptides on the virion: Generation of fluores-cent virus through the incorporation of pIX-GFP. Mol. Ther. 9,617–624.

Meyer, M., and Wagner, E. (2006). Recent developments in theapplication of plasmid DNA-based vectors and small inter-fering RNA therapeutics for cancer. Hum. Gene Ther. 17,1062–1076.

Mok, W., Boucher, Y., and Jain, R.K. (2007). Matrix metallo-proteinases-1 and � 8 improve the distribution and efficacy ofan oncolytic virus. Cancer Res. 67, 10664–10668.

Mok, W., Stylianopoulos, T., Boucher, Y., and Jain, R.K. (2009).Mathematical modeling of herpes simplex virus distribution insolid tumors: Implications for cancer gene therapy. Clin.Cancer Res. 15, 2352–2360.

Morgenroth, A., Cartellieri, M., Schmitz, M., Gunes, S., Weigle, B.,Bachmann, M., Abken, H., Rieber, E.P., and Temme, A. (2007).Targeting of tumor cells expressing the prostate stem cell an-tigen (PSCA) using genetically engineered T-cells. Prostate 67,1121–1131.

Morrison, J., Briggs, S.S., Green, N.K., Thoma, C., Fisher, K.D.,Kehoe, S., and Seymour, L.W. (2009). Cetuximab retargetingof adenovirus via the epidermal growth factor receptor fortreatment of intraperitoneal ovarian cancer. Hum. Gene Ther.20, 239–251.

Murakami, M., Ugai, H., Belousova, N., Pereboev, A., Dent, P.,Fisher, P.B., Everts, M., and Curiel, D.T. (2009). Chimeric ad-enoviral vectors incorporating a fiber of human adenovirus 3efficiently mediate gene transfer into prostate cancer cells.Prostate 70, 362–376.

Myhre, S., Henning, P., Friedman, M., Stahl, S., Lindholm, L.,and Magnusson, M.K. (2009). Re-targeted adenovirus vectorswith dual specificity: Binding specificities conferred by twodifferent Affibody molecules in the fiber. Gene Ther. 16, 252–261.

Ogris, M., Kotha, A.K., Tietze, N., Wagner, E., Palumbo, F.S.,Giammona, G., and Cavallaro, G. (2007). Novel biocompatiblecationic copolymers based on polyaspartylhydrazide beingpotent as gene vector on tumor cells. Pharm. Res. 24, 2213–2222.

Philipp, A., Meyer, M., and Wagner, E. (2008). Extracellulartargeting of synthetic therapeutic nucleic acid formulations.Curr. Gene Ther. 8, 324–334.

Power, A.T., and Bell, J.C. (2008). Taming the Trojan horse:Optimizing dynamic carrier cell=oncolytic virus systems forcancer biotherapy. Gene Ther. 15, 772–779.

Power, A.T., Wang, J., Falls, T.J., Paterson, J.M., Parato, K.A.,Lichty, B.D., Stojdl, D.F., Forsyth, P.A., Atkins, H., and Bell,J.C. (2007). Carrier cell-based delivery of an oncolytic viruscircumvents antiviral immunity. Mol. Ther. 15, 123–130.

Romer, J., Nielsen, B.S., and Ploug, M. (2004). The urokinasereceptor as a potential target in cancer therapy. Curr. Pharm.Des. 10, 2359–2376.

Russ, V., Elfberg, H., Thoma, C., Kloeckner, J., Ogris, M., andWagner, E. (2008). Novel degradable oligoethylenimine acry-late ester-based pseudodendrimers for in vitro and in vivo genetransfer. Gene Ther. 15, 18–29.

Saban, S.D., Silvestry, M., Nemerow, G.R., and Stewart, P.L.(2006). Visualization of a-helices in a 6-angstrom resolutioncryoelectron microscopy structure of adenovirus allows re-

finement of capsid protein assignments. J. Virol. 80, 12049–12059.

Sandberg, L., Papareddy, P., Silver, J., Bergh, A., and Mei, Y.F.(2009). Replication-competent Ad11p vector (RCAd11p) effi-ciently transduces and replicates in hormone-refractory met-astatic prostate cancer cells. Hum. Gene Ther. 20, 361–373.

Scanlan, M.J., Gure, A.O., Jungbluth, A.A., Old, L.J., andChen, Y.T. (2002). Cancer=testis antigens: An expandingfamily of targets for cancer immunotherapy. Immunol. Rev.188, 22–32.

Schaffert, D., and Wagner, E. (2008). Gene therapy progress andprospects: Synthetic polymer-based systems. Gene Ther. 15,1131–1138.

Schagen, F.H., Graat, H.C., Carette, J.E., Vellinga, J., van Geer,M.A., Hoeben, R.C., Dermody, T.S., and van Beusechem, V.(2008). Replacement of native adenovirus receptor-bindingsites with a new attachment moiety diminishes hepatic tro-pism and enhances bioavailability in mice. Hum. Gene Ther.19, 783–794.

Schenk, E., Essand, M., Bangma, C.H., and members of theGIANT FP6 Consortium. (2009). Clinical adenoviral genetherapy for prostate cancer. Hum. Gene Ther. 21, 807–813.

Scheres, S.H., Marabini, R., Lanzavecchia, S., Cantele, F., Rutten,T., Fuller, S.D., Carazo, J.M., Burnett, R.M., and San Martin, C.(2005). Classification of single-projection reconstructions forcryo-electron microscopy data of icosahedral viruses. J. Struct.Biol. 151, 79–91.

Schwerdt, A., Zintchenko, A., Concia, M., Roesen, N., Fisher,K.D., Lindner, L.H., Issels, R.D., Wagner, E., and Ogris, M.(2008). Hyperthermia induced targeting of thermosensitivegene carriers to tumors. Hum. Gene Ther. 19, 1283–1292.

Sebestyen, Z., de Vrij, J., Magnusson, M., Debets, R., and Will-emsen, R. (2007). An oncolytic adenovirus redirected with atumor-specific T-cell receptor. Cancer Res. 67, 11309–11316.

Seghezzi, G., Marelli, R., Mandriota, S.J., Nolli, M.L., Mazzieri,R., and Mignatti, P. (1996). Tumor cell conditioned mediumstimulates expression of the urokinase receptor in vascularendothelial cells. J. Cell Physiol. 169, 300–308.

Seiradake, E., Henaff, D., Wodrich, H., Billet, O., Perreau, M.,Hippert, C., Mennechet, F., Schoehn, G., Lortat-Jacob, H.,Dreja, H., Ibanes, S., Kalatzis, V., Wang, J.P., Finberg, R.W.,Cusack, S., and Kremer, E.J. (2009). The cell adhesion molecule‘‘CAR’’ and sialic acid on human erythrocytes influence ade-novirus in vivo biodistribution. PLoS Pathog. 5, e1000277.

Stevenson, M., Hale, A.B., Hale, S.J., Green, N.K., Black, G.,Fisher, K.D., Ulbrich, K., Fabra, A., and Seymour, L.W. (2007).Incorporation of a laminin-derived peptide (SIKVAV) onpolymer-modified adenovirus permits tumor-specific target-ing via a6-integrins. Cancer Gene Ther. 14, 335–345.

Subramanian, T., Vijayalingam, S., and Chinnadurai, G. (2006).Genetic identification of adenovirus type 5 genes that influ-ence viral spread. J. Virol. 80, 2000–2012.

Tasch, J., Gong, M., Sadelain, M., and Heston, W.D. (2001). Aunique folate hydrolase, prostate-specific membrane antigen(PSMA): A target for immunotherapy? Crit. Rev. Immunol. 21,249–261.

Uil, T.G., Vellinga, J., de Vrij, J., van den Hengel, S.K., Rabelink,M.J., Cramer, S.J., Ariyurek, Y., Van Galen, M., den Dunnen,J.T., Van der Vliet, P.C., and Hoeben, R.C. (2009). Mutantadenovirus polymerases for bioselection of oncolytic adeno-viruses. In M. Benko and B. Harrach, eds. Abstracts of the 9thInternational Adenovirus Meeting. (Hungarian Academy ofSciences, Budapest, Hungary) pp. 76.

804 DE VRIJ ET AL.

Van der Vliet, P.C., and Hoeben, R.C. (2006). Adenoviruses. InM.L. DePamphilis, ed. DNA Replication and Human Disease.(Cold Spring Harbor Laboratory Press, Cold Spring Harbor,New York) pp. 645–661.

Vellinga, J., Rabelink, M.J., Cramer, S.J., van den Wollenberg,D.J., Van der Meulen, H., Leppard, K.N., Fallaux, F.J., andHoeben, R.C. (2004). Spacers increase the accessibility ofpeptide ligands linked to the carboxyl terminus of adenovirusminor capsid protein IX. J. Virol. 78, 3470–3479.

Vellinga, J., van den Wollenberg, D.J., Van der Heijdt, S.,Rabelink, M.J., and Hoeben, R.C. (2005a). The coiled-coildomain of the adenovirus type 5 protein IX is dispensablefor capsid incorporation and thermostability. J. Virol. 79,3206–3210.

Vellinga, J., Van der Heijdt, S., and Hoeben, R.C. (2005b). Theadenovirus capsid: Major progress in minor proteins. J. Gen.Virol. 86, 1581–1588.

Vellinga, J., Uil, T.G., de Vrij, J., Rabelink, M.J., Lindholm, L.,and Hoeben, R.C. (2006). A system for efficient generation ofadenovirus protein IX-producing helper cell lines. J. GeneMed. 8, 147–154.

Vellinga, J., de Vrij, J., Myhre, S., Uil, T., Martineau, P., Lind-holm, L., and Hoeben, R.C. (2007). Efficient incorporation ofa functional hyper-stable single-chain antibody fragmentprotein-IX fusion in the adenovirus capsid. Gene Ther. 14,664–670.

Waddington, S.N., Parker, A.L., Havenga, M., Nicklin, S.A.,Buckley, S.M., McVey, J.H., and Baker, A.H. (2007). Targetingof adenovirus serotype 5 (Ad5) and 5=47 pseudotypedvectors in vivo: Fundamental involvement of coagulationfactors and redundancy of CAR binding by Ad5. J. Virol. 81,9568–9571.

Waddington, S.N., McVey, J.H., Bhella, D., Parker, A.L., Barker,K., Atoda, H., Pink, R., Buckley, S.M., Greig, J.A., Denby, L.,Custers, J., Morita, T., Francischetti, I.M., Monteiro, R.Q.,Barouch, D.H., van Rooijen, N., Napoli, C., Havenga, M.J.,Nicklin, S.A., and Baker, A.H. (2008). Adenovirus serotype 5hexon mediates liver gene transfer. Cell 132, 397–409.

Wang, X., Yin, L., Rao, P., Stein, R., Harsch, K.M., Lee, Z., andHeston, W.D. (2007). Targeted treatment of prostate cancer.J. Cell Biochem. 102, 571–579.

Weaver, E.A., and Barry, M.A. (2008). Effects of shielding ade-noviral vectors with polyethylene glycol (PEG) on vector-

specific and vaccine-mediated immune responses. Hum. GeneTher. 19, 1369–1382.

Willemsen, R., Chames, P., Schooten, E., Gratama, J.W., andDebets, R. (2008). Selection of human antibody fragments di-rected against tumor T-cell epitopes for adoptive T-cell therapy.Cytometry A 73, 1093–1099.

Wolff, J.A., and Rozema, D.B. (2008). Breaking the bonds:Non-viral vectors become chemically dynamic. Mol. Ther. 16,8–15.

Wortmann, A., Vohringer, S., Engler, T., Corjon, S., Schirmbeck,R., Reimann, J., Kochanek, S., and Kreppel, F. (2008). Fullydetargeted polyethylene glycol-coated adenovirus vectors arepotent genetic vaccines and escape from pre-existing anti-adenovirus antibodies. Mol. Ther. 16, 154–162.

Yan, W., Kitzes, G., Dormishian, F., Hawkins, L., Sampson-Johannes, A., Watanabe, J., Holt, J., Lee, V., Dubensky, T.,Fattaey, A., Hermiston, T., Balmain, A., and Shen, Y. (2003).Developing novel oncolytic adenoviruses through bioselec-tion. J. Virol. 77, 2640–2650.

Ylosmaki, E., Hakkarainen, T., Hemminki, A., Visakorpi, T.,Andino, R., and Saksela, K. (2008). Generation of a condi-tionally replicating adenovirus based on targeted destructionof E1A mRNA by a cell type specific MicroRNA. J. Virol. 82,11009–11015.

Zhu, Z.B., Makhija, S.K., Lu, B., Wang, M., Rivera, A.A., Preuss,M., Zhou, F., Siegal, G.P., Alvarez, R.D., and Curiel, D.T.(2004). Transport across a polarized monolayer of Caco-2 cellsby transferrin receptor mediated adenovirus transcytosis.Virology 325, 116–128.

Address correspondence to:Dr. Rob C. Hoeben

Department of Molecular Cell BiologyLeiden University Medical Center=mail stop S1-P

P.O. Box 96002300 RC Leiden, The Netherlands

E-mail: r.c.hoeben@lumc.nl

Received for publication November 6, 2009;accepted after revision November 20, 2009.

Published online: March 19, 2010.

Ad-DERIVED VECTORS FOR PROSTATE CANCER GENE THERAPY 805