Virology 392 (2009) 222–229

Contents lists available at ScienceDirect

Virology

j ourna l homepage: www.e lsev ie r.com/ locate /yv i ro

Defining the role of CD46, CD80 and CD86 in mediating adenovirus type 3 fiberinteractions with host cells

Kathryn Hall a, Maria E. Blair Zajdel b, G. Eric Blair a,⁎a Institute of Molecular and Cellular Biology, Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, UKb Faculty of Health and Wellbeing, Sheffield Hallam University, Sheffield S1 1WB, UK

Abbreviations: Ad, Adenovirus; Ad3F, Ad3 fibeAllophycocyanin; BHK, Baby Hamster Kidney; CAR,Receptor; CHO, Chinese Hamster Ovary; CMV, Cytomisothiocyanate; EGFP, Enhanced Green Fluorescentinfection; PE, Phycoerythrin; sBAR, species B AdenovB:2 Adenovirus Receptor; SCR, Short consensus repeat;STP, Serine/threonine/proline.⁎ Corresponding author. Institute of Molecular and

Biological Sciences, Garstang Building, Room 8.52d, UniUK. Fax: +44 1133433167.

E-mail address: g.e.blair@leeds.ac.uk (G.E. Blair).

0042-6822/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.virol.2009.07.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 May 2009Returned to author for revision 17 June 2009Accepted 14 July 2009Available online 13 August 2009

Keywords:Adenovirus serotype 3Fiber proteinCD46CD80CD86

Attachment of human adenoviruses (Ads) to host cells is mediated by the interaction of the fiber protein ofthe capsid with specific cell-surface molecules. For one of the species B adenoviruses, Ad3, the mechanism ofbinding to cells remains to be defined. Several previous reports have proposed CD46, CD80 or CD86 aspossible Ad3 fiber attachment molecules. In this study, CD80 and CD86 were not found to mediate Ad3 fiberbinding or Ad3-EGFP transduction of cells. Low levels of Ad3-EGFP transduction of a CHO cell line expressingrelatively high levels of CD46 were detected which might suggest a role for CD46 in facilitating Ad3: cellinteractions, in the absence of other attachment molecules. Anti-CD46 antibodies and siRNAs had almost noeffect on Ad3 fiber binding or Ad3-EGFP transduction of HeLa cells. However, treatment of A549 cells withCD46 siRNA resulted in some decrease of transduction with Ad3-EGFP.

© 2009 Elsevier Inc. All rights reserved.

Adenoviruses (Ads) are non-enveloped viruses that contain alinear double-stranded DNA genome of approximately 30–38 kb. Atleast 51 human adenovirus serotypes have currently been identifiedand classified into six species, A–F, on the basis of their oncogenicity inrodents, genome organisation and hemagglutination properties(Russell, 2009). The icosahedral adenovirus capsid is composed of252 capsomers of which 240, the hexon protein trimers, are arrangedin facets and 12 capsomers, the pentons, are located at each of thetwelve vertices of the capsid (Fabry et al., 2005). The penton consistsof a pentameric penton base protein from which projects a trimericfiber protein. The fiber comprises an amino-terminal domain which isanchored in the penton base, a variable length shaft domain andcarboxy-terminal globular knob domain (van Raaij et al., 1999;Russell, 2009).

The entry of adenoviruses into human cells is a two-step process inwhich an initial attachment to the cell surface is followed by asecondary interaction, which initiates internalisation of the virus. Theknob domain of the fiber protein mediates the primary interaction

r; Ad11F, Ad11 fiber; APC,Coxsackie B and Adenovirusegalovirus; FITC, FluoresceinProtein; moi, multiplicity ofirus Receptor; sB2AR, speciessiRNA, Small interfering RNA;

Cellular Biology, Faculty ofversity of Leeds, Leeds LS2 9JT,

ll rights reserved.

with the cell, effectively tethering the virus particle to the cell surfacevia a cellular attachment protein (Zhang and Bergelson, 2005). Theinteraction of the fiber protein with the primary attachment moleculeappears to be important in determining the tropism of adenoviruses.A further interaction occurs between the conserved RGD motifpresent in the penton base of most of the human Ad serotypes andcell-surface integrins, principally αvβ3 and αvβ5 (Nemerow et al.,2009). Binding of integrins by the virus induces changes in the actincytoskeleton, promoting endocytic uptake of the virus throughclathrin-coated vesicles to endosomes (Meier and Greber, 2004).Escape from the endosome then enables delivery of partiallydisassembled viruses to the nucleus via nuclear pore complexes(Greber and Way, 2006).

Adenoviruses from all species, except species B and certainserotypes of species D (Ad8, Ad19, Ad30 and Ad37), utilise thecoxsackie B and adenovirus receptor (CAR) as their primary cellularattachment protein (Bergelson et al., 1997; Roelvink et al., 1998;Arnberg et al., 2002; Law and Davidson, 2002; Wu et al., 2004; Zhangand Bergelson, 2005). The species B Ads have been classified into twogroups, the B1 viruses (serotypes 3, 7, 16, 21 and 50) and the B2viruses (serotypes 11, 14, 34 and 35) (Wadell et al., 1980). Thisdivision has been made on the basis of genetic similarity andcorrelates with the tropism of these viruses. The B1 viruses causeinfection of the upper respiratory tract whereas the B2 viruses areassociated with infection of the kidneys and urinary tract. Competi-tion studies of virus binding to cells have suggested the existence oftwo different cell-surface molecules capable of interaction with thespecies B adenoviruses (Segerman et al., 2003a). It has been proposedthat one of them, sBAR (species B Adenovirus Receptor), could

223K. Hall et al. / Virology 392 (2009) 222–229

interact with all species B adenoviruses and a distinct molecule,termed sB2AR (species B:2 Adenovirus Receptor) could only beutilised by the B2 adenoviruses. Subsequently, a molecule involved inthe primary attachment of Ad11 was shown to be CD46, acomplement regulatory protein, which is expressed on all humannucleated cells (Segerman et al., 2003b) and these observations wereextended to other species B adenoviruses, with the exception of Ad3(Gaggar et al., 2003). In contrast to these studies, CD46 was proposedto mediate Ad3 attachment to non-permissive baby hamster kidney(BHK) cells (Sirena et al., 2004) and a human glioma cell line (Ulasovet al., 2006). Adding a further complexity in understanding of theinteractions of species B Ads with human cells, it has been shown thatAd3 and several other members of B1 and B2 subspecies interact withthe co-stimulatory molecules CD80 and CD86 (Short et al., 2004,2006; Ulasov et al., 2007).

The interaction of Ad3 with human cells is clearly complex. It ispossible that Ad3 can utilise one or several of these three attachmentmolecules or other, as yet unidentified cell-surfacemolecules, perhapsdepending upon the target cell type. Ad3 is a serotype of growingimportance, as Ad5 vectors modified to carry Ad3 fibers showsignificantly higher transduction efficiency when compared to Ad5in several cell systems (Stoff-Khalili et al., 2007; Tsuruta et al., 2008;Volk et al., 2003). As a result of the lack of clarity regarding theidentity of the molecules used by Ad3 to enter cells and theimportance of this serotype in terms of improving vectors designedfor gene therapy, we have investigated the roles that CD46, CD80 andCD86 may play in mediating Ad3 attachment to a range of humancells, as well as rodent cells engineered to over-express each of thesecell-surface molecules.

Results

Interaction of Ad3 fiber with cell-surface CD46, CD80 and CD86

To assess whether there was a relationship between the levels ofcell-surface CD46, CD80 and CD86 and the extent of Ad3 fiber (Ad3F)binding, expression of each molecule was determined by flowcytometry in several human and transfected rodent cell lines. Fiberbinding assays were performed using recombinant His-tagged fiber.The use of purified fiber enabled the primary interaction of Ad3 withtarget cells to be studied, separate from the penton base interactionwith integrins.

Recombinant Ad3F was purified from E. coli using an immobilisedmetal affinity chromatography resin. SDS-PAGE followed by Coomas-sie Blue staining showed that the Ad3F consistedmainly of fiber trimerwith some monomer in the absence of heat treatment; followingboiling of the sample prior to electrophoresis, all of the Ad3Fmigratedas a monomer (Supplementary Fig. 1A). Densitometric analysisrevealed an approx. 3:1 ratio of Ad3 fiber trimer to monomer (resultsnot shown). Western blotting using an anti-penta-his antibodyconfirmed the presence of the his tag in both trimers and monomers(Supplementary Fig. 1B). In addition to Ad3 fiber attachment, Ad11fiber (Ad11F) binding was also performed for comparison, since thesefibers are derived from species B1 (Ad3) and species B2 (Ad11) andthere is a consensus from several studies that CD46 interacts withAd11 (Gaggar et al., 2003;Marttila et al., 2005; Segerman et al., 2003b).Ad11F was also purified from E. coli as a trimer, converted to amonomer after boiling of the sample prior to electrophoresis(Supplementary Fig. 1B). Thus both Ad3 and Ad11 fibers adopted anative conformation. The biological activity of recombinant Ad3F wasdetermined by its ability to block the entry of Ad3-EGFP virus intoHeLacells in a concentration-dependent manner (Supplementary Fig. 1C).

The highest levels of surface CD46 were detected on HeLa (whichare cervical carcinoma cells and are widely used for adenoviruspropagation) and A549 cells (lung epithelial carcinoma cells whichare permissive for Ad3 replication and also represent the known

tropism of Ad3 for airway cells), while cell lines of lymphoid origin,Daudi and Raji, expressed significantly lower levels of this protein(Fig. 1A). In particular, Daudi cells possessed only about 10–15% of theCD46 present on either HeLa or A549 cell lines. There wasconsiderable variation in the levels of surface CD46 on hamster CHOcell lines engineered to express human isoforms of this molecule (Fig.1B). Notably, two clones of different origin expressing the BC2 isoform(CHO-BC2a and CHO-BC2b) differed by two-fold in the level of cell-surface CD46 (Fig. 1B). Cell-surface CD80 and CD86 were detectableon Raji and Daudi cells, but at much lower levels than on CHO cellsengineered to express human forms of these proteins (Figs. 1C and D).HeLa and A549 cells did not possess cell-surface CD80 and CD86,although they displayed the highest levels of Ad3 fiber binding. Ad3Fbinding was not detected to Daudi and Raji cells (Fig. 2), althoughthey expressed all three proposed attachment molecules. Ad11Fstrongly associated with HeLa and A549 cells and also bound to Daudiand Raji cells, albeit at much lower levels (Fig. 2). Expression of eitherhuman CD80 or CD86 on CHO cells did not result in a significantdifference in Ad3F or Ad11F binding (Fig. 2). Ad11F, but not Ad3F,bound to all CHO cell lines expressing CD46 isoforms (Fig. 2).

Transduction of cell lines expressing CD46, CD80 and CD86 by Ad3-EGFPvirus

To extend the fiber binding studies, transduction of cell lines wasperformed using a replication-deficient Ad3-EGFP. This permits Ad3entry to be assayed bymeasuring EGFP expression. Entry of Ad3-EGFPwas fiber-dependent, since it was greatly reduced by pre-incubationof HeLa cells with recombinant Ad3F (Supplementary Fig. 1C) andpre-treatment of Ad3-EGFP with anti-Ad3 (but not anti-Ad5) fiberserum in a concentration-dependent manner (Supplementary Fig.1D). Cell lineswhich displayed the greatest level of Ad3F bindingweretransducedmost efficiently by Ad3-EGFP, namely HeLa and A549 (Fig.2). However, no EGFP expression was detected when Ad3-EGFP wasused to transduce Daudi or Raji cells.

The efficiency of transduction of CHO cell lines expressing themajor CD46 isoforms (CHO-C1, CHO-C2 and CHO-BC1), as well as twodifferent CHO clones expressing BC2 was also compared. Very lowlevels of transduction of these cell lines were detected, with theexception of CHO-BC2a (Fig. 2). This cell line could be transducedwithconsiderable efficiency, but at three- to four-fold lower level thanHeLa or A549 cells, regardless of comparable expression of cell-surface CD46. Interestingly, the CHO-BC2b cell line was not trans-duced at the level observed for CHO-BC2a. This disparity could be dueto the lower level of CD46 expressed by CHO-BC2b, compared to CHO-BC2a (Fig. 1B). Expression of CD80 and CD86 in CHO cells did notresult in Ad3-EGFP transduction (Fig. 2).

The effect of polyclonal anti-CD46 antibodies on Ad3-EGFP transductionand Ad3 fiber binding to HeLa cells

To investigate the role that CD46 might play in Ad3 entry, theeffect of pre-treating HeLa cells with antibodies against CD46 on Ad3-EGFP transduction and Ad3F binding was determined. To evaluate theeffect on fiber binding, HeLa cells were pre-incubated with polyclonalanti-CD46 antibodies at 4 °C for 30 min, Ad3F was then added and thelevel of fiber binding to the cells was determined by flow cytometryusing Alexa 488-labelled anti-penta-his antibody. The binding of Ad3Fto HeLa cells was reduced by approximately 10% in the presence ofanti-CD46 antibodies (Figs. 3A and B). This was not significant (p-value 0.5888), whereas Ad11F binding was significantly reduced byapproximately 70% (p-value 0.0003). To determine the effect on Ad3-EGFP transduction, HeLa cells were pre-incubated with the sameantibody for 30 min at room temperature. Ad3-EGFP was then addedto the cells and incubated for 2 h at 37 °C. Cells were then washed andgrowth medium was added to each well. After 24 h, cells were

Fig. 1. Expression of cell-surface CD46, CD80 and CD86 on human and hamster cell lines. The levels of cell-surface CD46, CD80 or CD86 on a range of cell lines were determined byflow cytometry using either FITC conjugated anti-CD46 antibody, PE-labelled anti-CD80 or APC anti-CD86 (BD Pharmingen). Results are presented as net geometric meanfluorescence, following subtraction of fluorescence due to binding of an isotype-matched control antibody. Antibody binding was performed in duplicate in at least threeindependent experiments. A. CD46 expression on human cell lines. B. CD46 expression on a series of hamster CHO cells engineered to express isoforms of CD46; CHO-BC2a is a BC2-expressing cell line that is a separate clonal isolate from the cell line denoted CHO-BC2b; CHO and R-CHO are the control CHO cell lines for CHO-BC2a, and CHO-C1, C2, BC1 and BC2brespectively. C. CD80 expression on human cell lines and a CHO cell line engineered to express CD80; CHO is the control cell line. D. CD86 expression on human cell lines and a CHOcell line engineered to express CD86; CHO is the control cell line.

224 K. Hall et al. / Virology 392 (2009) 222–229

analysed for EGFP expression using flow cytometry. Pre-incubationwith anti-CD46 antibodies reduced EGFP expression in HeLa cells byapproximately 20%, which was of low significance (p-value 0.014). Incontrast, entry of Ad3-EGFP to CHO-BC2a cells was reduced by morethan 90% (p-value 1.44×10−5) by anti-CD46 antibodies.

The effect of CD46 siRNA treatment of HeLa and A549 cells on Ad3 fiberbinding and Ad3-EGFP transduction

The potential involvement of CD46 in mediating Ad3 fiber:cellinteractions was analysed using siRNA-mediated knock-down in HeLaor A549 cells. Reduction of 70% to 90% in cell-surface CD46 wasobtained 48 h after transfection (Fig. 4). This resulted in anapproximately 50% reduction in Ad11F binding to HeLa and A549cells. However, despite the great reduction in cell-surface CD46, therewas no change in Ad3F binding to either cell line. This suggests thatCD46 is not required for Ad3 fiber attachment to those cells.Interestingly, transduction of both HeLa and A549 cells by Ad3-EGFPwas more discernably reduced following knock-down of CD46 thanAd3F binding. EGFP expression in HeLa cells was reduced by 16%,however this was not significant (p=0.179) whereas the reduction inAd3-EGFP entry to A549 cells of 30% was regarded to be of somesignificance (p=0.0125).

Discussion

Knowledge of Ad3 cell attachment protein(s) will advance theunderstanding of the pathogenicity of adenoviruses and expand their

application as vectors in gene therapy. The complement regulatoryprotein, CD46 and the co-stimulatory molecules CD80 and CD86 haveall been reported to mediate Ad3 interactions with cells. Howeverconflicting data has been published. For the first time, we analyse thepotential role of all three proposed candidates in mediating Ad3:hostcell interactions in one study.

CD80 and CD86 are expressed mainly on antigen-presenting cells,however their presence on certain tumour cells has also been reported(Ulasov et al., 2007). Results from this study, using fiber binding andAd3-EGFP transduction, showed that there was no relationshipbetween the expression of CD80 or CD86 and Ad3F binding or Ad3entry. This provides evidence against the notion that CD80 and CD86act as primary attachment molecules for Ad3 and is in contrast toseveral previously published reports (Short et al., 2004, 2006; Ulasov etal., 2007). If these co-stimulatory molecules facilitate Ad3 (or Ad11)attachment to or infection of human cells, thiswould require additionalmolecule(s) as, in this study, expression of CD80 or CD86 alone did notappear to be sufficient to mediate Ad3F (or Ad11F) attachment or Ad3entry. Interestingly, a lack of CD80 and CD86 expression in HeLa cells(which are efficiently transduced by Ad3-EGFP) has also been reported(Tuve et al., 2008), in agreement with our data.

CD46 has been identified as an attachment molecule for otherspecies B adenoviruses (Gaggar et al., 2003; Marttila et al., 2005), aspecies D adenovirus, Ad37 (Wu et al., 2004), measles virus (Dorig etal., 1993), human herpesvirus 6 (HHV-6) (Santoro et al., 1999) andNeisseria species (Kallstrom et al., 1997). CD46 belongs to a familyof proteins that regulate the activation of complement, preventingdestruction of host tissues (Seya et al., 1990). In nucleated cells,

Fig. 2. Adenovirus fiber binding and Ad3-EGFP transduction of human and hamster cell lines expressing CD46, CD80 and CD86. Ad3 and Ad11 fiber binding was determined by flow cytometry following binding of recombinant fiber to cells.Bound fiber was detected with Alexa Fluor 488-conjugated mouse monoclonal anti-penta-his antibody. Transduction of cells was performed using a replication-deficient Ad3-EGFP, as described in Materials and methods. Expression of theEGFP reporter gene following entry of the Ad3-EGFP virus into cells was measured by flow cytometry. Results are presented as net geometric mean fluorescence, following subtraction of fluorescence detected in the absence of fiber binding orin mock-transduced cells, respectively. Fiber binding assays and Ad3-EGFP transductions were performed in duplicate in at least three independent experiments.

225K.H

allet

al./Virology

392(2009)

222–229

Fig. 3. Effects of polyclonal anti-CD46 antibodies on Ad3 fiber binding and Ad3-EGFP transduction. HeLa cells were detached with versene, pre-incubated with rabbit polyclonalantibodies to CD46 (or normal control rabbit immunoglobulins) and Ad3F or Ad11F binding assayed. For transduction experiments, HeLa or CHO-BC2a cells in six-well plates werepre-incubated with rabbit polyclonal anti-CD46 antibodies or control rabbit immunoglobulins and treated with Ad3-EGFP as described in Materials and methods. A. Flow cytometryprofiles are showing the level of Alexa 488-conjugated anti-penta-his antibody binding in the absence of fiber (dotted line), the level of fiber binding under control conditions (greypeak) and the level of Ad3F or Ad11F binding to HeLa cells when cells were pre-incubated with polyclonal anti-CD46 antibodies (black line). B. The effects of polyclonal anti-CD46antibodies on fiber binding or transduction are shown relative to the control with normal rabbit immunoglobulins (100%). The following geometric means correspond to 100%: Ad3Fbinding=20.29; Ad11F binding=62.23; Ad3-EGFP/HeLa=271.94; Ad3-EGFP/CHO-BC2a=17.02. p values ⁎b0.05; ⁎⁎⁎b0.001; ns, not significant. Fiber binding or transductionassays were performed in duplicate in at least three independent experiments. Controls are shown as solid bars and samples treated with anti-CD46 as dotted bars.

226 K. Hall et al. / Virology 392 (2009) 222–229

alternative RNA splicing results in the synthesis of four major CD46isoforms, C1, C2, BC1 and BC2 which are differentially expressed,although all four isoforms are expressed in HeLa cells (Buchholz et al.,1996). The various isoforms have either alternative cytoplasmic tails(type 1 or 2) or differences in the serine/threonine/proline (STP)domains. BC1 and BC2 are longer isoforms and have STPB domains inaddition to STPC. In contrast, the C1 and C2 isoforms contain only STPC

regions (Post et al., 1991). All of the major isoforms express four shortconsensus repeat (SCR) domains, each of which is a structural motif ofapproximately 60 amino acids. SCR2 has been shown to facilitate theinteraction with both Ad11 and Ad35 fiber proteins (Fleischli et al.,2005; Gaggar et al., 2005). The SCR domains have also been reportedto mediate the interaction of measles virus and HHV-6 with CD46(Maisner et al., 1996; Greenstone et al., 2002).

Investigations into the role of CD46 in facilitating Ad3 attachmentto cells have produced conflicting data (Sirena et al., 2004; Marttila etal., 2005; Tuve et al., 2006). In this study, there did not appear to be adirect relationship between the level of Ad3F binding and levels ofcell-surface CD46. Ad3F did not bind to several cell lines expressingCD46, for example Daudi and Raji cells. In addition, Ad3F bound toHeLa cells but there was no discernible binding to CHO-BC2a cells,although they expressed similar levels of CD46. Interestingly, thisparticular clone of CHO-CD46 (CHO-BC2a) was transduced by Ad3-EGFP whereas another CHO clone expressing the same CD46 isoform(CHO-BC2b), but at a considerably lower level, was not transduced.This might suggest that threshold levels are required for cell-surfaceCD46 to interact with Ad3, which is further stabilised by interactionswith other cell-surface molecules e.g. integrins. A recent studysuggested that Ad3F would be expected to have a very low affinityfor CD46 due to two additional residues in the HI loop within the fiberknob (Pache et al., 2008). Other studies have also shown that certainamino acids within the HI loop of the fiber knob domain, which aresubstituted in Ad3, are critical for CD46 binding (Gustafsson et al.,2006; Wang et al., 2007). This proposed low affinity might be thereason that Ad3F could require critical levels of CD46 for inter-action. Low affinity of Ad3 for CD46may also explain why certain cells

(e.g. Daudi and Raji), which expressed relatively low levels of CD46and also lacked expression of the other putative Ad3 attachmentprotein(s), were not transduced by Ad3-EGFP. Interestingly, nosignificant changes were detected in Ad3F binding or Ad3-EGFPtransduction following an approximately 90% reduction of CD46 onthe surface of HeLa cells by RNA interference. Furthermore there wasno significant reduction in Ad3F binding or Ad3-EGFP entry followingpre-incubation of HeLa cells with anti-CD46 antibody. This suggeststhat Ad3F does not interact with CD46 on HeLa cells, or if thisinteraction takes place, it has a minor effect on Ad3 entry. If Ad3 doesnot utilise CD46 on HeLa cells, this suggests the presence of anotherattachment molecule(s), as previously proposed (Tuve et al., 2006).This high affinity interaction would be the primary factor determiningAd3 entry into HeLa cells. We suggest that Ad3 might weakly interactwith CD46 and that such interactions may only be observed in theabsence of the, as yet, unidentified attachment molecule(s) whichcould interact much more strongly than CD46 with Ad3. The relativecontributions of the unidentified molecule(s) and CD46 to Ad3attachment could vary, depending on their ratio and distribution onthe cell surface in different cell types. For example, in A549 cells, CD46appeared to play a more discernible role in Ad3 entry than in HeLacells, as judged by RNA interference experiments (Fig. 4). In additionto the interaction mediated by the knob domain, the length andflexibility of the fiber have also been shown to play a significant role indetermining tropism (Wu et al., 2003). It is probable that the bindingof Ad3 to cells will be a combination of interactions, which mayinvolve other capsid components in addition to the fiber. We have sofar identified, in common with others (Tuve et al., 2006), that themoiety that mediates Ad3 attachment is most likely to be a protein(i.e. it is trypsin-sensitive) and that the interaction has a strongdependence on calcium ions (unpublished results). Further studieshave suggested a role for heparan sulfate glycoproteins (HSPGs) inAd3 (and Ad35) attachment to human cells (Tuve et al., 2008).However since these molecules have been described to mediate lowaffinity interactions with target cells, high affinity target molecule(s)for Ad3 on the cell-surface remain to be identified (Tuve et al., 2008).

Fig. 4. Effects of reduction of cell-surface CD46 by RNA interference on Ad3 fiberbinding and Ad3-EGFP transduction of HeLa and A549 cells. Transfection of cells withsiRNA targeted against CD46 or control siRNA was performed as described in Materialsand methods. Forty-eight hours after transfection, cells were either detached withversene and analysed for CD46 expression or Ad3F or Ad11F binding or transduced withAd3-EGFP. Solid bars show the level of CD46 expression, fiber binding or Ad3-EGFPtransduction in cells transfected with control siRNA (100%). Dotted bars show resultsobtained following treatment with CD46-specific siRNA. All assays were performed induplicate in three independent experiments. p values ⁎b0.05; ⁎⁎⁎b0.001; ns, notsignificant. Results shown are from three independent experiments.

227K. Hall et al. / Virology 392 (2009) 222–229

Identification of cell-surface proteins which mediate Ad3 attachmentwill contribute to a greater understanding of the interaction of thisserotype and other species B adenoviruses with human cells.

Materials and methods

Cell culture and virus growth

A549, HeLa monolayer and Chinese Hamster Ovary (CHO) cellswere obtained from the European Collection of Animal Cell Cultures,Salisbury, UK. Raji and Daudi cell lines were provided by Dr G. Doody(Leeds Institute of Molecular Medicine, Leeds, UK). HeLa, A549 andCHO cells were grown in DMEM (Sigma, Dorset, UK) supplementedwith 10% FCS (Biosera Ltd, Ringmer, East Sussex, UK), 0.1 mg/mlstreptomycin, 100 U/ml penicillin and 2 mM L-glutamine (Sigma,Dorset, UK). Raji and Daudi cell lines were cultured in RPMI-1640(Sigma, Dorset, UK), supplemented with 10% FCS, 0.1 mg/mlstreptomycin and 100 U/ml penicillin. The CHO-BC2a cell line,which expresses the BC2 isoform of CD46 (Loveland et al., 1993)was a kind gift from Dr J. Schneider-Schaulies, University ofWürzburg, Germany. CHO-CD80 and CHO-CD86 cells were kindlyprovided by Dr D.M. Sansom, MRC Centre for Immune Regulation,University of Birmingham Medical School, UK (Sansom et al., 1993).These three CHO cell lines expressing human CD46, CD80 or CD86 andcorresponding control cell lines were maintained in DMEM, supple-mented as described for HeLa cells. CHO cell lines expressing each of

the four major isoforms of CD46 (CHO-C1, CHO-C2, CHO-BC1 andCHO-BC2, termed here CHO-BC2b) and a control CHO cell line (R-CHO) were kindly provided by Dr K. Liszewski, WashingtonUniversity School of Medicine, USA (Liszewski and Atkinson, 1996).These CHO cell lines were cultured in Ham's F12 nutrient mixture(Sigma, Dorset, UK) supplemented with 10% FCS, 0.1 mg/mlstreptomycin, 100 U/ml penicillin, 2 mM L-glutamine and 0.5 mg/ml geneticin. All cell lines were maintained at 37 °C in a humidifiedatmosphere containing 5% CO2.

Ad3-EGFP virus, an E1-deleted Ad3 vector expressing an EGFPreporter gene from a CMV promoter, was kindly provided by Dr S.Hemmi, University of Zurich, Switzerland (Sirena et al., 2005). Viruswas propagated in a complementing 293-2V6.11 cell line whichexpresses the Ad5 E1B-55K protein constitutively and the Ad5 E4-34Kprotein under the control of an ecdysone-inducible promoter(Mohammadi et al., 2004). Twelve to twenty-four hours prior to theaddition of virus, cells were treated with 1 μg/ml Ponasterone A(Axxora (UK) Ltd, Nottingham, UK), an ecdysone analog. Virus waspurified using CsCl centrifugation, as described (Tollefson et al., 1999).The concentration of physical virus particles (vp) per ml wasdetermined by UV spectrophotometry (Mittereder et al., 1996).

Antibodies

Alexa 488-conjugated mouse monoclonal anti-penta-his antibodywas purchased from Qiagen (Crawley, UK). Directly conjugatedmonoclonal antibodies against CD46 (FITC-labelled anti-CD46, cloneE4.3), CD80 (PE-labelled anti-CD80, clone L307.4) and CD86 (APC-labelled anti-CD86, clone BNI3) and their respective isotype-matchedcontrols were obtained from BD Pharmingen (BD Biosciences, UK).Rabbit polyclonal anti-CD46 antibody was a kind gift from Dr ClaireHarris, Cardiff University, UK. This was raised against whole CD46protein isolated from human platelets. Rabbit polyclonal anti-Ad3Fand anti-Ad5F antibodies were raised against recombinant fiberproteins (GEB, unpublished results).

Expression and purification of recombinant adenovirus fiber proteins

The Ad3 fiber (Ad3F) sequence comprising the knob and two shaftrepeats was amplified by PCR using an Ad3 fiber genomic plasmid astemplate (provided by Dr G. Akusjarvi, Uppsala, Sweden) and forward(5′-CGGCATATGAAACTTTGCAGTAAACTC-3′) and reverse (5′-CGCGTCGACTCATCTTCTCTAATATA-3′) primers. The PCR productwas cloned in the bacterial expression vector pET33b (Novagen,Nottingham, UK) and expressed as an amino-terminal hexa his-tagged sequence. DNA sequencing confirmed that the cloned Ad3fiber was of wild-type sequence (Genbank DQ086466.1; results notshown). The entire Ad11 fiber (Ad11F) gene, cloned into the pRSETvector as an amino-terminal hexa his-tagged sequence, was kindlydonated by Dr Ya-Fang Mei, University of Umea, Sweden (Mei andWadell, 1996). Recombinant fiber proteins were expressed in the E.coli strain BL21 DE3 (Promega, Southampton, UK). Bacterial cells wererecovered from culture by centrifugation at 3000 rpm for 10 min anddisrupted by sonication. Recombinant proteins were isolated from thesoluble fraction of cell lysates by His-tag affinity column purificationusing Talon resin, a cobalt-based agarose (Clontech Europe, France).Protein was eluted from the column using increasing concentrationsof imidazole. Recovered protein was dialysed overnight in PBS. Thesubunit structure of fiber proteins was determined by SDS-PAGE(Hong and Engler, 1996) and Western blotting using a mousemonoclonal anti-penta-his antibody (Qiagen, Crawley, UK).

Flow cytometric analysis of fiber binding and cell-surface proteins

Adherent cells were detached from flasks using Versene (Gibco,Paisley, UK). Cells grown in suspension, namely Raji and Daudi, were

228 K. Hall et al. / Virology 392 (2009) 222–229

pelleted directly from culture at 1500 rpm. Cells were washed inbinding buffer (PBS with 1% BSA, 1 mM CaCl2, 1 mM MgCl2),centrifuged and resuspended in the same buffer. For fiber binding,cells (0.3×106 cells per 100 μl) were incubated with a saturatingamount of fiber protein (as determined by titration studies,approximately 0.4 μg of Ad3F and 1 μg Ad11F, data not shown) for30 min at 4 °C. Cells were then washed and resuspended in bindingbuffer with Alexa 488-conjugated mouse anti-penta-his antibody(Qiagen, Crawley, UK) at 4 °C. After 30 min, cells were centrifuged,washed, resuspended in ice-cold PBS and analysed by flow cytometry.A total of 20,000 cells were counted per sample using a FACSCaliburflow cytometer (Becton Dickinson, Oxford, UK). Analysis of cell-surface expression of CD46, CD80 and CD86 was also performed asabove, except that 10 μl of each fluorescently-conjugated antibodywas incubated with 0.3×106 cells for 1 h (following the manufac-turer's instructions). Background binding was determined using cellsamples incubated with either Alexa 488-conjugated mouse anti-penta-his antibody alone or by comparison with an isotype-matchedcontrol antibody. All flow cytometry data was analysed usingCellQuest software (Becton Dickinson, Oxford, UK).

Virus transduction

Approximately 0.4×106 adherent cells were seeded in each wellof a six-well plate and grown until they were 90% confluent. Cellswere washed with PBS and approximately 450 vp per cell of Ad3-EGFP were added in 1 ml serum-free DMEM. This multiplicity ofinfection (moi) was sufficient to transduce more than 90% of HeLacells. Cells were incubated with virus for 1 h at 37 °C and 1 ml ofDMEM containing 10% FCS, 0.1 mg/ml streptomycin, 100 U/mlpenicillin and 2 mM L-glutamine was added to each well. After 24 h,cells were detached with trypsin and analysed for EGFP expressionby flow cytometry. For cells grown in suspension (Raji and Daudi),approximately 0.8×106 cells were resuspended in DMEM containingAd3-EGFP at an moi of approximately 450 vp/cell and maintained at37 °C for 1 h in the wells of a six-well plate. One ml of DMEMcontaining 10% FCS, 0.1 mg/ml streptomycin, 100 U/ml penicillinand 2 mM L-glutamine was then added to each well. After 24 hincubation at 37 °C, cells were removed from wells and analysed byflow cytometry.

Inhibition of Ad 3-EGFP transduction by recombinant fiber protein

HeLa cells (0.4×106 per well) were seeded in six-well plates andgrown until 90% confluent. The appropriate well was incubated with2 μg or 6 μg of Ad3F protein in binding buffer at room temperature for30 min. Control cells were incubated with binding buffer only.Approximately 450 vp/cell of Ad3-EGFP were added directly to eachwell in 1 ml serum-free DMEM and incubated at 37 °C for 2 h. Cellswere washed with PBS and 2 ml of DMEM containing 10% FCS,0.1 mg/ml streptomycin, 100 U/ml penicillin and 2 mM L-glutaminewas added to each well. After 24 h incubation at 37 °C, cells weredetached with trypsin and analysed for EGFP expression by flowcytometry.

Inhibition of Ad3-EGFP transduction by anti-fiber antibodies

Approximately 1.5×108 vp of Ad3-EGFP virus in 20 μl DMEMwereincubated with rabbit anti-Ad3 fiber or anti-Ad5 fiber serum on ice for30 min. Undiluted (0.5 μl) or ten-fold serial dilutions (1 μl) were used.Virus:antibody mixtures were then added to 90% confluent HeLa orCHO-BC2a cells in 1 ml of serum-free DMEM and incubated for 1 h at37 °C. One ml of DMEM containing 10% FCS, 0.1 mg/ml streptomycin,100 U/ml penicillin and 2 mM L-glutamine was then added to eachwell. After 24 h incubation at 37 °C, cells were detached with trypsinand analysed for EGFP expression by flow cytometry.

Inhibition of fiber binding by anti-CD46 antibodies

Cells (0.3×106 in 100 μl) were incubated with rabbit polyclonalanti-CD46 antibody for 30 min at 4 °C and washed in ice-cold PBS.Fiber binding assays were performed as described above. A series oftitrations was performed using anti-CD46 antibody, ranging from1.9 μg to 19 μg. No further significant decrease in fiber binding wasdetected using masses of antibody greater than 9.5 μg (data notshown), therefore this mass of antibody was used in all analyses.Equivalent masses of normal rabbit antibodies were used as control.

Inhibition of Ad3-EGFP entry using anti-CD46 antibodies

Cells (0.4×106 per well) were seeded in six-well plates and grownuntil 90% confluent. The appropriate well was incubated with 19 μg ofpolyclonal anti-CD46 antibody (or normal rabbit immunoglobulins)in binding buffer for 30min at room temperature. Ad3-EGFP in DMEM(approx. 450 vp/cell) was added to each well and incubated at 37 °Cfor 2 h. Unbound virus was removed by washing with PBS. DMEMcontaining 10% FCS, 0.1 mg/ml streptomycin, 100 U/ml penicillin and2 mM L-glutamine was added to each well and incubated for 24 h.EGFP expression was determined by flow cytometry.

RNA interference

HP validated CD46 siRNA was purchased from Qiagen (Cat no:1027400) and prepared according to the manufacturer's instructions.Forward and reverse sequences were as follows:

Sense r(CCA AAA CCC UAC UAU GAG A)dTdTAntisense r(UCU CAU AGU AGG GUU UUG G)dTdT.

Transfections were performed using Oligofectamine (Invitrogen,UK). HeLa or A549 cells were seeded in six-well plates at 0.3×106 cellsper well and grown until approximately 70% confluent. A mixture of30 μl Optimem-1 and 7 μl Oligofectamine was incubated with amixture of 10 μl of 20 μM siRNA and 250 μl Optimem for 20 min atroom temperature. Pre-formed complexes were then added to PBS-washed cells for 5 h at 37 °C. Following transfection, 1 ml of DMEMcontaining 30% FCS and 2 mM L-glutamine was added to each well.After 24 h, an additional 1 ml of DMEM containing 30% FCS and 2 mML-glutamine was added. Cells were then incubated for a further 24 h at37 °C. Following this incubation, cells were transducedwith Ad3-EGFPorwashed and detached using versene for analysis of CD46 expressionand fiber binding. The mouse monoclonal anti-CD46 antibody used toassess the level of CD46 was clone E4.3 (BD Pharmingen), whichrecognises all isoforms of CD46 (Purcell et al., 1991).

Acknowledgments

This research was supported by the Biotechnology and BiologicalSciences Research Council (BBSRC) and Yorkshire Cancer Research.We thank Aadil El-Turabi for the Ad3 fiber expression plasmid,Rebecca Caygill for fiber protein analysis, Graham Bottley and GarethHowell for flow cytometry advice and Joan Jarvis for the technicalassistance.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.virol.2009.07.010.

References

Arnberg, N., Pring-Akerblom, P., Wadell, G., 2002. Adenovirus type 37 uses sialic acid asa cellular receptor on Chang C cells. J. Virol. 76, 8834–8841.

229K. Hall et al. / Virology 392 (2009) 222–229

Bergelson, J.M., Cunningham, J.A., Droguett, G., Kurt-Jones, E.A., Krithivas, A., Hong, J.S.,Horwitz, M.S., Crowell, R.L., Finberg, R.W., 1997. Isolation of a common receptor forCoxsackie B viruses and adenoviruses 2 and 5. Science 275, 1320–1323.

Buchholz, C.J., Gerlier, D., Hu, A., Cathoman, T., Liszewski, K., Atkinson, J.P., Cattaneo, R.,1996. Selective expression of a subset of measles virus receptor-competent CD46isoforms in human brain. Virology 217, 349–355.

Dorig, R.E., Marcil, A., Chopra, A., Richardson, C.D., 1993. The human CD46 molecule is areceptor for measles virus (Edmonston strain). Cell 75, 295–305.

Fabry, C.M., Rosa-Calatrava, M., Conway, J.F., Zubieta, C., Cusack, S., Ruigrok, R.W.,Schoehn, G., 2005. A quasi-atomic model of human adenovirus type 5 capsid. EMBOJ. 24, 1645–1654.

Fleischli, C., Verhaagh, S., Havenga, M., Sirena, D., Schaffner, W., Cattaneo, R., Greber, U.F., Hemmi, S., 2005. The distal short consensus repeats 1 and 2 of the membranecofactor protein CD46 and their distance from the cell membrane determineproductive entry of species B adenovirus serotype 35. J. Virol. 79, 10013–10022.

Gaggar, A., Shayakhmetov, D.M., Lieber, A., 2003. CD46 is a cellular receptor for group Badenoviruses. Nature Med. 9, 1408–1412.

Gaggar, A., Shayakhmetov, D.M., Liszewski, M.K., Atkinson, J.P., Lieber, A., 2005.Localization of regions in CD46 that interact with adenovirus. J. Virol. 79,7503–7513.

Greber, U.F., Way, M., 2006. A superhighway to virus infection. Cell 124, 741–754.Greenstone, H.L., Santoro, F., Lusso, P., Berger, E.A., 2002. Human herpesvirus 6 and

measles virus employ distinct CD46 domains for receptor function. J. Biol. Chem.277, 39112–39118.

Gustafsson, D.J., Segerman, A., Lindman, K., Mei, Y.F., Wadell, G., 2006. The Arg279Gln[corrected] substitution in the adenovirus type 11p (Ad11p) fiber knob abolishesEDTA-resistant binding to A549 and CHO-CD46 cells, converting the phenotype tothat of Ad7p. J. Virol. 80, 1897–1905.

Hong, J.S., Engler, J.A., 1996. Domains required for assembly of adenovirus type 2 fibertrimers. J. Virol. 70, 7071–7078.

Kallstrom, H., Liszewski, M.K., Atkinson, J.P., Jonsson, A.B., 1997. Membrane cofactorprotein (MCP or CD46) is a cellular pilus receptor for pathogenic Neisseria. Mol.Microbiol. 25, 639–647.

Law, L.K., Davidson, B.L., 2002. Adenovirus serotype 30 fiber does not mediatetransduction via the coxsackie-adenovirus receptor. J. Virol. 76, 656–661.

Liszewski, M.K., Atkinson, J.P., 1996. Membrane cofactor protein (MCP; CD46). Isoformsdiffer in protection against the classical pathway of complement. J. Immunol. 156,4415–4421.

Loveland, B.E., Johnstone, R.W., Russell, S.M., Thorley, B.R., McKenzie, I.F., 1993.Different membrane cofactor protein (CD46) isoforms protect transfected cellsagainst antibody and complement mediated lysis. Transpl. Immunol. 1, 101–108.

Maisner, A., Alvarez, J., Liszewski, M.K., Atkinson, D.J., Atkinson, J.P., Herrler, G., 1996.The N-glycan of the SCR 2 region is essential for membrane cofactor protein (CD46)to function as a measles virus receptor. J. Virol. 70, 4973–4977.

Marttila, M., Persson, D., Gustafsson, D., Liszewski, M.K., Atkinson, J.P., Wadell, G.,Arnberg, N., 2005. CD46 is a cellular receptor for all species B adenoviruses excepttypes 3 and 7. J. Virol. 79, 14429–14436.

Mei, Y.F., Wadell, G., 1996. Epitopes and hemagglutination binding domain on subgenusB:2 adenovirus fibers. J. Virol. 70, 3688–3697.

Meier, O., Greber, U.F., 2004. Adenovirus endocytosis. J. Gene Med. 6 (Suppl. 1),S152–S163.

Mittereder, N., March, K.L., Trapnell, B.C., 1996. Evaluation of the concentration andbioactivity of adenovirus vectors for gene therapy. J. Virol. 70, 7498–7509.

Mohammadi, E.S., Ketner, E.A., Johns, D.C., Ketner, G., 2004. Expression of theadenovirus E4 34k oncoprotein inhibits repair of double strand breaks in thecellular genome of a 293-based inducible cell line. Nucleic Acids Res. 32,2652–2659.

Nemerow, G.R., Pache, L., Reddy, V., Stewart, P.L., 2009. Insights into adenovirus hostcell interactions from structural studies. Virology 384, 380–388.

Pache, L., Venkataraman, S., Reddy, V.S., Nemerow, G.R., 2008. Structural variations inspecies B adenovirus fibers impact CD46 association. J. Virol. 82, 7923–7931.

Post, T.W., Liszewski, M.K., Adams, E.M., Tedja, I., Miller, E.A., Atkinson, J.P., 1991.Membrane cofactor protein of the complement system: alternative splicing ofserine/threonine/proline-rich exons and cytoplasmic tails produces multipleisoforms that correlate with protein phenotype. J. Exp. Med. 174, 93–102.

Purcell, D.F., Russell, S.M., Deacon, N.J., Brown, M.A., Hooker, D.J., McKenzie, I.F., 1991.Alternatively spliced RNAs encode several isoforms of CD46 (MCP), a regulator ofcomplement activation. Immunogenetics 33, 335–344.

Roelvink, P.W., Lizonova, A., Lee, J.G., Li, Y., Bergelson, J.M., Finberg, R.W., Brough, D.E.,Kovesdi, I., Wickham, T.J., 1998. The coxsackievirus-adenovirus receptor proteincan function as a cellular attachment protein for adenovirus serotypes fromsubgroups A, C, D, E, and F. J. Virol. 72, 7909–7915.

Russell, W.C., 2009. Adenoviruses: update on structure and function. J. Gen. Virol. 90,1–20.

Sansom, D.M., Wilson, A., Boshell, M., Lewis, J., Hall, N.D., 1993. B7/CD28 but not LFA-3/CD2 interactions can provide ‘third-party’ co-stimulation for human T-cellactivation. Immunology 80, 242–247.

Santoro, F., Kennedy, P.E., Locatelli, G., Malnati, M.S., Berger, E.A., Lusso, P., 1999. CD46 isa cellular receptor for human herpesvirus 6. Cell 99, 817–827.

Segerman, A., Arnberg, N., Erikson, A., Lindman, K., Wadell, G., 2003a. There are twodifferent species B adenovirus receptors: sBAR, common to species B1 and B2adenoviruses, and sB2AR, exclusively used by species B2 adenoviruses. J. Virol. 77,1157–1162.

Segerman, A., Atkinson, J.P., Marttila, M., Dennerquist, V., Wadell, G., Arnberg, N., 2003b.Adenovirus type 11 uses CD46 as a cellular receptor. J. Virol. 77, 9183–9191.

Seya, T., Hara, T., Matsumoto, M., Sugita, Y., Akedo, H., 1990. Complement-mediatedtumor cell damage induced by antibodies against membrane cofactor protein(MCP, CD46). J. Exp. Med. 172, 1673–1680.

Short, J.J., Pereboev, A.V., Kawakami, Y., Vasu, C., Holterman, M.J., Curiel, D.T., 2004.Adenovirus serotype 3 utilizes CD80 (B7.1) and CD86 (B7.2) as cellular attachmentreceptors. Virology 322, 349–359.

Short, J.J., Vasu, C., Holterman, M.J., Curiel, D.T., Pereboev, A., 2006. Members ofadenovirus species B utilize CD80 and CD86 as cellular attachment receptors. VirusRes. 122, 144–153.

Sirena, D., Lilienfeld, B., Eisenhut, M., Kalin, S., Boucke, K., Beerli, R.R., Vogt, L., Ruedl, C.,Bachmann, M.F., Greber, U.F., Hemmi, S., 2004. The human membrane cofactorCD46 is a receptor for species B adenovirus serotype 3. J. Virol. 78, 4454–4462.

Sirena, D., Ruzsics, Z., Schaffner, W., Greber, U.F., Hemmi, S., 2005. The nucleotidesequence and a first generation gene transfer vector of species B human adenovirusserotype 3. Virology 343, 283–298.

Stoff-Khalili, M.A., Rivera, A.A., Stoff, A., Michael Mathis, J., Rocconi, R.P., Matthews, Q.L.,Numnum, M.T., Herrmann, I., Dall, P., Eckhoff, D.E., Douglas, J.T., Siegal, G.P., Zhu, Z.B., Curiel, D.T., 2007. Combining high selectivity of replication via CXCR4 promoterwith fiber chimerism for effective adenoviral oncolysis in breast cancer. Int. J.Cancer 120, 935–941.

Tollefson, A.E., Hermiston, T.W., Wold, W.S.M., 1999. Preparation and titration of CsClbanded adenovirus stock. Adenovirus Methods and Protocols. Methods inMolecular Medicine 21. Humana Press Inc, Totowa, NJ.

Tsuruta, Y., Pereboeva, L., Breidenbach, M., Rein, D.T., Wang, M., Alvarez, R.D., Siegal, G.P., Dent, P., Fisher, P.B., Curiel, D.T., 2008. A fiber-modified mesothelin promoter-based conditionally replicating adenovirus for treatment of ovarian cancer. Clin.Cancer Res. 14, 3582–3588.

Tuve, S., Wang, H., Ware, C., Liu, Y., Gaggar, A., Bernt, K., Shayakhmetov, D., Li, Z., Strauss,R., Stone, D., Lieber, A., 2006. A new group B adenovirus receptor is expressed athigh levels on human stem and tumor cells. J. Virol. 80, 12109–12120.

Tuve, S., Wang, H., Jacobs, J.D., Yumul, R.C., Smith, D.F., Lieber, A., 2008. Role of cellularheparan sulfate proteoglycans in infection of human adenovirus serotype 3 and 35.PLoS Pathog. 4, e1000189.

Ulasov, I.V., Tyler, M.A., Zheng, S., Han, Y., Lesniak, M.S., 2006. CD46 represents atarget for adenoviral gene therapy of malignant glioma. Human Gene Ther. 17,556–564.

Ulasov, I.V., Rivera, A.A., Han, Y., Curiel, D.T., Zhu, Z.B., Lesniak, M.S., 2007. Targetingadenovirus to CD80 and CD86 receptors increases gene transfer efficiency tomalignant glioma cells. J. Neurosurg. 107, 617–627.

van Raaij, M.J., Mitraki, A., Lavigne, G., Cusack, S., 1999. A triple beta-spiral in theadenovirus fibre shaft reveals a new structural motif for a fibrous protein. Nature401, 935–938.

Volk, A.L., Rivera, A.A., Kanerva, A., Bauerschmitz, G., Dmitriev, I., Nettelbeck, D.M.,Curiel, D.T., 2003. Enhanced adenovirus infection of melanoma cells by fiber-modification: incorporation of RGD peptide or Ad5/3 chimerism. Cancer Biol. Ther.2, 511–515.

Wadell, G., Hammarskjold, M.L., Winberg, G., Varsanyi, T.M., Sundell, G., 1980. Geneticvariability of adenoviruses. Ann. N. Y. Acad. Sci. 354, 16–42.

Wang, H., Liaw, Y.C., Stone, D., Kalyuzhniy, O., Amiraslanov, I., Tuve, S., Verlinde, C.L.,Shayakhmetov, D., Stehle, T., Roffler, S., Lieber, A., 2007. Identification of CD46binding sites within the adenovirus serotype 35 fiber knob. J. Virol. 81,12785–12792.

Wu, E., Pache, L., Von Seggern, D.J., Mullen, T.M., Mikyas, Y., Stewart, P.L., Nemerow, G.R., 2003. Flexibility of the adenovirus fiber is required for efficient receptorinteraction. J. Virol. 77, 7225–7235.

Wu, E., Trauger, S.A., Pache, L., Mullen, T.M., von Seggern, D.J., Siuzdak, G., Nemerow, G.R., 2004. Membrane cofactor protein is a receptor for adenoviruses associated withepidemic keratoconjunctivitis. J. Virol. 78, 3897–3905.

Zhang, Y., Bergelson, J.R., 2005. Adenovirus receptors. J. Virol. 79, 12125–12131.