HUMAN GENE THERAPY 13:311–320 (January 20, 2002)Mary Ann Liebert, Inc.

Hexon Gene Switch Strategy for the Generation of Chimeric Recombinant Adenovirus

RIMA YOUIL, TIMOTHY J. TONER, QIN SU, MINCHUN CHEN, AIMIN TANG, ANDREW J. BETT, and DANILO CASIMIRO

ABSTRACT

The usefulness of adenovirus as a vehicle for transgene delivery is limited greatly by the induction of neu-tralizing anti-adenoviral immunity following the initial administration, thereby resulting in shorter-term andreduced levels of transgene expression. In this paper, we outline a strategy for the generation of recombinantAd5-based adenovectors that have undergone a complete hexon exchange in an effort to circumvent pre-ex-isting anti-vector humoral immunity. Eighteen different chimeric adenoviral vectors (from subgroups A, B,C, D, and E) have been constructed using a combination of direct cloning and bacterial homologous recom-bination methods. However, only chimeric Ad5-based constructs in which the hexons from Ad1, Ad2, Ad6,and Ad12 are incorporated in place of the Ad5 hexon were successfully rescued into viruses. Despite severalattempts, the remaining fourteen chimeric adenovectors were not rescuable. In vivo rodent studies using trans-genes for human immunodeficiency virus type 1 (HIV-1) gag and secreted human alkaline phosphatase (SEAP)suggest that the Ad5/Ad6-gag chimera (wherein Ad5 hexon was replaced with that of Ad6) is able to evadeneutralizing antibodies generated against Ad5 vector efficiently. However, it appears that cross-reactive cy-totoxic T lymphocytes (CTL) may also play a role in controlling in vivo infectivity of Ad5/Ad6-gag chimera.The Ad5/Ad12 chimera was found to be extremely ineffective in the i.m. delivery and expression of HIV-1gag in mice compared to the Ad5/Ad6 construct. Implications of these results will be discussed.

311

OVERVIEW SUMMARY

Recombinant chimeric adenovectors were constructed by ex-changing the complete hexon gene from an Ad5 vector withhexon genes from alternate adenovirus serotypes. Whereasthe construction of the vectors was possible, many of thesechimeric adenovectors did not rescue. Those that did werehexon exchanges made within the subgroup C viruses as wellas a subgroup A virus (Ad12). These gag-containing chime-ric viruses were analyzed for their ability to circumvent pre-existing Ad5 antibodies in rodents previously vaccinated withfirst generation Ad5 virus. In the background of Ad5 virus,gag expression from the recombinant Ad5-gag vector was to-tally ablated. However, gag expression from the Ad5/Ad6-gag vector was suppressed 10-fold compared with the ex-pression observed from naïve (non-preexposed) mice. In aseparate rodent experiment, SEAP expression from anAd5SEAP virus was completely ablated following preincu-bation of the virus with anti-Ad5 containing sera. However,

SEAP expression was comparable when Ad5SEAP waspreincubated in sera obtained from either naïve mice or thoseimmunized with the Ad5/Ad6-gag chimera. This indicatedthat at the level of humoral immunity, Ad5/Ad6-gag chimeracan in fact circumvent neutralization by Ad5 virus.

INTRODUCTION

REPLICATION-DEFECTIVE ADENOVIRUSES are efficient vehiclesfor the in vivo delivery of a variety of heterologous pro-

teins (Graham, 1990; Robbins et al., 1998; Zheng et al., 2000).Adenoviruses are maintained episomally and do not, suppos-edly, integrate into the host genome. Consequently, turnover ofdifferentiated cells infected with such a recombinant virus willresult in the loss of the viral genome and hence concomitantloss of the transgene. To sustain transgene expression levels,repeat administrations of the respective recombinant adenovi-rus are often required, particularly for genetic therapy applica-

1Merck & Co., Inc., West Point, PA, 19486.

tions. However, the efficacy of subsequent doses may be limitedby pre-existing host immunity generated against the adenovirusvector. The humoral responses are largely directed against theviral capsid (Willcox and Mautner, 1976; Adam et al., 1986; Rus-sell et al., 1991; Gahery-Segard et al., 1998), which consists ofthree major structural proteins—hexon, penton, and fiber. Thehexon protein forms the major part of the virion coat, account-ing for 240 capsomeres out of the 252 subunits that comprise thecapsid. Several reports have suggested that antibodies directedtoward the hexon coat are the strongest and most neutralizing(Wohlfart, 1988; Toogood et al., 1992). In fact, vaccine devel-oped to prevent adenoviral infections has been based on purifiedhexon protein (Couch et al., 1973).

The hexon protein contains determinants for both type- andgroup-specific neutralizing antibodies (Toogood et al., 1992). Theneutralizing capacity of the Ad5 hexon protein has been associ-ated with two main regions, Loop 1 (L1) and Loop 2 (L2) (Jorn-vall et al., 1981; Chroboczek et al., 1992; Crompton et al., 1994).Loop 1 contains six specific regions designated as hypervariable(HVR 1–6) and spans amino acid residues 132–320. L2 containsthe seventh hypervariable region (HVR 7) and it spans aminoacids 408–459. The remaining regions between these hypervari-able regions are highly conserved within subgroup C; in the caseof Ad2 and Ad5 capsids, these regions are nearly 100% identical.The three-dimensional structures of the Ad2 and Ad5 capsidsshow that the L1 and L2 regions are, in fact, present on the sur-face of the virion and subsequently represent the type-specificantigenic determinants of the hexon protein. The similarity in thethree-dimensional structures of subgroup C adenoviruses suggeststhat it would be plausible to construct Ad 5 hexon chimeras bysimply exchanging hexon genes with members from its own sub-group. Gall et al. (1998) have successfully constructed Ad5chimera where the native hexon has been completely replacedwith an Ad2 hexon, but not with that of Ad7 (subgroup B). Hexonproteins from subgroups A, B, D, E, and F, on the other hand,show lower levels of similarity with Ad 5 hexon (Crawford-Mik-sza et al., 1996). This difference subsequently allows for alternateadenovirus to escape pre-existing immunity to the more prevalentgroup C adenovirus.

To test whether anti-viral immunity can be largely circum-vented by replacing the hexon gene, we have generated chi-meric recombinant adenoviruses that contain the E1,E3-deletedAd5 backbone and wherein the Ad5 hexon gene was replacedwith the hexon genes of several alternate serotypes from sub-groups A, B, C, D, and E. This strategy tested the feasibility ofpropagating such chimeras in the existing Ad5 E1-transformedcell lines because the majority of the adenoviral backbone isbased on the Ad5 genome. In addition, we were able to assessthe effect of pre-existing anti-Ad5 fiber and penton antibodieswith respect to their neutralizing ability.

MATERIALS AND METHODS

Growth and viral DNA extraction of alternate wild-type adenoviruses

Wild-type (WT) adenoviruses (serotypes 12 and 18 from sub-group A; 7, 11, 16, and 35 from subgroup B; 1, 2, and 6 fromsubgroup C; 9, 10, 13, 15, 17, 19, 27, and 37 from subgroup

D; 4 from subgroup E) were purchased from the American TypeCulture Collection (ATCC) as lyophilized viruses. Virus waspropagated in HeLa cells and was CsCl purified. Viral DNAwas extracted following protease treatment and phenol/chloro-form extraction.

PCR amplification of hexon genes from alternateadenovirus serotypes

In designing the primers to amplify the hexon genes, the re-striction enzymes Cla I and Nae I were selected as convenientcloning sites based on the conserved amino acids at the 59 and39 end of the hexon genes from a number of fully sequencedhexon genes from subgroups A, B, C, D, and E (Crawford-Mik-sza et al., 1996). The Ad5 hexon sequence was obtained fromGenBank accession #AH007403. The following Ad5 specificPCR primers that contain these specific restriction endonucle-ase recognition sites (underlined) were used for amplificationof Ad5 hexon:

59–TTGCCGCCATGGCTACCCC ATCGATGATGC–39 (Ad5Forward primer)

59–CGTTATGTTGTGGCGTTGCCGGC C–39 (Ad5 Reverseprimer)

The boldface type signifies the silent nucleotide change to gen-erate the Cla I site:

The Ad4 hexon sequence was obtained from GenBank ac-cession #AFO65064. The following Ad4-specific primers wereused for the amplification of Ad4 hexon.

59–TTGCCGCCATGGCCACCCCA TCGATGCT–39 (Ad4 For-ward primer)

59–CGTTATGTGGTGGCGTTGCCGGC T–39 (Ad4 Reverseprimer)

The Ad12 hexon sequence was obtained from GenBank acces-sion #X73487.

The following PCR primers were used for amplification ofAd12 hexon:

59–TTGCCGCCATGGCCACTCCATCGATGATGC–39

(Ad12 Forward primer)59–CGTTAGGTGGTAGCGTTGC CGGCCG–39 (Ad12 Re-

verse primer)

The boldface type signifies the silent nucleotide changes madeto generate the Cla I and the Nae I sites.

Because complete sequence information does not exist forthe majority of the alternate adenoviral hexons, amplificationof the remaining hexons was performed using primers based onAd 5 hexon sequence. This decision was based on the conservedamino acid sequence at the extreme ends of hexon genes asshown in alignments between a number of fully sequenced al-ternate hexon genes (Crawford-Miksza et al., 1996).

PCR was conducted using Advantage®-HF PCR Kit (Clon-tech Laboratories, Inc., CA) and the following buffers were usedto amplify each DNA sample:

100% high-fidelity buffer for Ad 1, 2, 4, 5 10, 11, 12, 13, 15,37, 7, 16, 19, 6

YOUIL ET AL.312

60% high-fidelity buffer / 40 % cDNA buffer for Ad 17, 9, 35100% cDNA buffer for Ad18

Each 50-ml amplification contained , 100 ng viral DNA. Eachreaction also contained , 500 ng of each specific hexon primer(forward and reverse). The PCR conditions were as follows:94°C for 15 sec; 94°C for 15 sec, and 68°C for 3 min (35 cy-cles); the final extension was 68°C for 3 min.

The adenovirus hexon PCR products were gel purified andcloned using Zero Blunt TOPO PCR cloning kit (Invitrogen,CA) following manufacturer’s instructions.

Generation of Ad5 shuttle vector containing a unique Cla I site

The Ad5 shuttle vector was constructed by digesting the first-generation Ad5 backbone with Asp 718 and Hind III and iso-lating the 3870-bp fragment. This fragment, which contains thehexon gene and flanking Ad5 DNA sequences, was cloned di-rectly into Asp 718/Hind III site of pUC18 cloning vector. Thisvector is now referred to as pAd5 hexon shuttle vector.

The incorporation of a Cla I site within the extreme 59 endof the Ad5 hexon gene in the pAd5 shuttle vector allowed forthe removal of the native Ad5 hexon gene from the shuttle vec-tor by digestion with Cla I and Nae I. To create this Cla I site,an Ad5 hexon-specific 59 primer was designed to include a sin-gle-nucleotide change 12 bp from the start of the hexon codingsequence. This nucleotide change produced the required Cla Isite without altering the amino acid at this position (CCT RCCA). All serotypes amplified with the Ad5-based primer setscontained had 59 Cla I and 39 Nae I sites incorporated into theiramplified hexon genes.

Ad4 specific 39 primer was made with the base changeGGT R GGC to allow for generation of the Nae I site. Thisbase change did not alter the coding for the amino acid at thisposition.

The Ad 12 specific 59 primer included the base changeCCC R CCA for generation of the Cla I site. The 39 primer in-cluded two base changes GCGGGT R GCCGGC for genera-tion of the Nae I site. These base alterations did not result inany amino acid changes.

Insertion of the Cla I site at the 59 end of the Ad5 hexongene in pAd5 hexon shuttle vector required amplification oftwo PCR products that spanned the Ad5–59 hexon region us-ing the following primer sets:

Fragment A: Forward primer: 59-CTGGTGACGCAAATAGA-CGAReverse primer: 59-TCATCGATGGGGTAGCCAT

Fragment B: Forward primer: 59-CCCATCGATGATGCCGCAReverse primer: 59-CTTGCTCGTCTACTTCGTCT

PCR conditions for fragment A and B were as follows:

Clontech (high fidelity) PCR kit was used as per the manufac-turer’s instructions. The WT Ad5 viral DNA was used as thetemplate (100 ng). Amplification was 95°C for 5 min fol-lowed by, 35 cycles at 94°C for 50 sec, 60°C for 50 sec, 68°Cfor 50 sec. A final 7-min extension was performed at 68°C.

Fragment A PCR product (440 bp) and fragment B PCRproducts (481 bp) were each cloned directly into TOPO PCRBlunt II (Invitrogen). The TOPO clone containing fragment Awas digested with Cla I and EcoRV and treated with calf in-testinal phosphatase. This digestion resulted in the removal ofa 28 bp fragment. The remaining 4321 fragment was gel puri-fied. TOPO containing fragment B was digested with Cla I andEco RV and the 487-bp fragment was purified. This 487-bpfragment (which is the fragment B product) was then ligatedinto the 4321 bp vector fragment to produce the TOPO (frag-ment A 1 B) plasmid. This plasmid was then digested with EcoRI and the 918-bp fragment containing the A 1 B region wasgel purified. The pAd5 hexon shuttle vector was digested withBss HII and Bsg I enzymes and the 7093-bp fragment was gelpurified. A total of 100 ng of the 7093-bp and the 918-bp frag-ments were co-infected into BJ5183 cells (Chartier et al., 1996;Youil et al., 2001). Positive clones were identified by screen-ing with Cla I and Nae I. This is the shuttle vector designatedpAd5 hexon(Cla I/Nae I) that was used for all subsequentcloning manipulations with alternate hexon genes.

Construction of shuttle vectors containing alternate hexon genes

Each of the alternate hexon genes were removed from theTOPO cloning vectors using the following combination of re-striction digestions and then gel purified:

Nae I TURBO (Promega)/Cla I for Ad 1, 2, 4, 12, 7, 11NgoM IV/Cla I for Ad 16, 35NgoM IV (partial)*/Cla I for Ad 10, 13, 27, 37, 9, 15, 19, 17,

18

(*Note: Because there was no available full sequence informa-tion on these hexon genes, it was not possible to predict Nae Isites, until they were cloned into the TOPO vector. To removethese particular hexon genes, partial digestion using NgoM IVwas required.) Each full-length hexon gene fragment was thencloned directly into pAd5 hexon (Cla I/Nae I) shuttle vector,which had been digested with either Nae I TURBO/Cla I orwith the combination NgoM IV / Cla I, to remove the nativeAd5 hexon. NgoM IV was used as an alternative to Nae I fordigestion convenience. The sizes of the hexons were around the2.8- to 3.2-kb range. The hexon genes were each ligated withthe appropriately digested pAd5 shuttle vector (4398-bp frag-ment). Each shuttle vector was screened to determine insertionof the correct-sized fragment and was then sequenced 300–500bp into each end of the hexon genes (at the 59 and 39 Ad5/hexongene flanking region) to confirm correct sequence and readingframe. Where possible, sequences were confirmed with Gen-Bank published sequences. However, as only partial sequenceinformation exists for many of the hexon genes, it was not pos-sible to make such comparisons for all hexon genes. Sequenceor amino acid homologies were made with hexon genes fromwithin the same subgroup as obtained through our personal se-quencing effort or through published data.

The hexon genes Ad12 and Ad18 from our subgroup Achimeras (Ad5/Ad12-gag and Ad5/Ad18-gag) were sequencedin their entirety. Each hexon gene represented a single and com-plete ORF.

RECOMBINANT CHIMERIC ADENOVECTORS 313

Construction of pAd5(no hexon) recombinant adenovector

To produce the hexon-less version of the recombinant Ad5adenovector, a shuttle vector was used that contained a uniqueSwa I site flanked by Ad5 sequences (deleted of the Ad5 hexongene). A shuttle vector, pAd5 (No hexon) was constructed thatcomprised of the WT Ad5 region 18,074–18,830 bp and region21,771–22,621 bp. A unique Swa I site exists between these re-gions. Pac I digestion released the 1637-bp fragment that con-tains the Ad5 hexon flanking sequences. A recombinant Ad5adenovector (E1-/E3-) containing the HIV-1 gag transgene in-serted into the E1 deleted region, was digested with Asc I to re-lease a 9620-bp (which contains the Ad5 hexon) and a 26,090-bp fragment. The 9620-bp fragment was cloned into the Asc Isite of pNEB193 to produce the plasmid pNEB1931Ad5hexon(Asc I). This plasmid was digested with Apa I. The result-ing 9423-bp band was gel purified and co-infected with 100 ngof the 1637-bp Pac I fragment of pAd5(No hexon) plasmid intoBJ5183 cells (Chartier et al., 1996; Youil et al., 2001). Positiveclones [named pNEB1931Ad5(no hexon) Swa I] were selectedby digestion with Swa I and Bgl II.

pNEB1931Ad5(No hexon) Swa I plasmid was digestedcompletely with Asc I, and the 6689-bp band was gel purified.The original recombinant vector (pAd5-gag) was digested withSfi I. The 29,000 bp band was gel purified and co-transfectedwith pNEB1931Ad5(No hexon) Asc I fragment (6689 bp) inBJ5183 cells. Positive clones were identified by digestion withSwa I and EcoR1. This plasmid is the original recombinant vec-tor, which now is devoid of the Ad5 hexon gene and containsa unique Swa I site in its place. It is referred to as pAd5–gag(no hexon) adenovector.

Construction of recombinant adenovectors containingalternate hexon genes

The pAd 5–gag (no hexon)SwaI adenovector was digestedto completion with Swa I and the 32,779-bp fragment was gelpurified. The various pAd5 shuttle vectors containing hexonsfrom alternate serotypes were each digested with Pac I to re-move the ampicillin resistance vector backbone and release thenovel hexon gene flanked by Ad5 sequences. Each of the frag-ments containing the novel hexon genes were co-transformedwith the 32,779-bp Swa I fragment of pAd5-gag(no hexon) intoBJ5183 chemically competent cells (Chartier et al., 1996; Youilet al., 2001). Positive colonies were identified by using the re-striction enzymes Eco RI, Swa I, and Asc I (Fig. 1).

Rescue strategy of chimeric adenovectors

A total of 3.3 mg of each Pac I-digested chimeric aden-ovector was transfected into a 6-cm dish containing PER.C6cells (Fallaux et al., 1998) at , 40–60% confluence using thecalcium phosphate method (CellPhect Transfection kit, Amer-sham Pharmacia Biotech., Cat# 27-9268-01). Four hours fol-lowing transfection, medium was removed and replaced withfresh Dulbecco’s modified Eagle medium (DMEM, GibcoBRLCat# ) containing 10% fetal bovine serum (FBS) defined (Hy-clone, Cat# SH30071.03) and 1% penicillin/streptomycin/glut-amine (GibcoBRL Cat# 10378-016). Cells were incubated at37oC, 5% CO2 until cytopathic effect (CPE) was observed.

Generally, CPE occurred between 6 and 12 days, after which thetotal cell culture was collected. After three rounds of freeze/thaw,the cell culture was spun down at 4,000 rpm at 4°C to removecellular debris. The cell lysate was then used to infect into pas-sage 2. For P2, 1 ml of P1 was infected into a 6 cm dish con-taining PER.C6 cells at 80–90% confluence. CPE was generallyobserved after 48 hr (for Ad5) and longer (3–4 days) for thechimeras. Following CPE, the total cell culture was collected andthe cells were disrupted by three rounds of freeze/thaw. The cul-ture was centrifuged at 4000 rpm at 4°C to remove the cellulardebris. One milliliter of the cell lysate was then used to infectinto a 15-cm dish containing PER.C6 at 80–90% confluence. Thecell culture was then collected and the cells disrupted by threerounds of freeze/thaw. This cell lysate was then used to expandthe virus for CsCl banding and dialysis. For banding, clarifiedcell lysate was digested with benzonase for 30 min at 37°C andapplied to a two-gradient CsCl gradient for ultracentrifugationfor 4 hr. The band was collected and applied to a continuousCsCl gradient for overnight ultracentrifugation. The viral bandwas collected using standard techniques and then subjected totwo rounds of dialysis (in 10 mM Tris-Cl, pH 7.5; 1 mM MgCl2;10% glycerol), changing buffer every 4 hr. The particle numberwas determined spectrophotometrically using the conversion fac-tor of 1.1 3 1012 virus particle (vp) per absorbance unit at A260nm

(Maizel et al., 1968). For genomic analysis of the virus, the vi-ral DNA was extracted by first treating with Pronase (2 hr at37°C) followed by phenol/chloroform extraction and isopropanolprecipitation. The viral DNA was then analyzed by radioactivelylabeling restriction digestions of the viral DNA with Klenow en-zyme and [33P]dATP.

Rodent immunization

Thirty BALB/c mice were vaccinated with Ad5 empty virus(i.e. containing no transgene) at 1 3 108 vp/mouse at weeks 0and 3. At week 6, the mice were separated into three cohortsof 10 mice/cohort. Each cohort was vaccinated with 1 3 108

vp/mouse of either the Ad5-gag or Ad5/Ad6-gag chimera, or4 3 108 vp/mouse of the Ad5/Ad12-gag chimera. A higher par-ticle dose was used for the Ad5/Ad12-gag to compensate forthe four-fold lower transgene expression level (see Results). Atweek 10, 5 of 10 mice from each cohort received a second doseof the same virus they received at week 6. In all cases, the viruswas administered in a total volume of 100 ml; a 50-ml aliquotwas injected into each quadricep muscle using insulin syringes(Becton Dickinson, NJ). As controls, another 30 BALB/c miceremained Ad5 naïve until week 6, at which point, cohorts of n 5 10 mice were given 1 3 108 vp/mouse of Ad5-gag orAd5/Ad6-gag, or 4 3 108 vp/mouse of Ad5/Ad12-gag. At week10, half of each cohort was injected with the same virus theyhad received at week 6. Blood samples were taken at weeks 9and 20.

Anti-p24 enzyme-linked immunosorbent assay

Anti-p24 titers were obtained following standard secondaryantibody-based enzyme-linked immunosorbent assay (ELISA).Maxisorp plates (NUNC, Rochester, NY) were coated byovernight incubation with 100 ml of 1 mg /ml HIV-1 p24 pro-tein (Austral Biologicals, CA) in phosphate-buffered saline(PBS). The plates were washed with PBS and 0.05% Tween 20

YOUIL ET AL.314

using a Titertek MAP instrument (Hunstville, AL) and incu-bated for 2 hr with 200 ml/well of blocking solution containingPBS, 0.05% Tween, and 1% bovine serum albumin (BSA). Aninitial serum dilution of 100-fold was performed followed by4-fold serial dilution. Then 100-ml aliquots of serially dilutedsamples were added per well and incubated for 2 hr at roomtemperature. The plates were washed, and 100 ml of 1/1000-di-luted horseradish peroxidase (HRP)-rabbit anti-mouse im-munoglobulin G (IgG) (ZYMED, San Francisco, CA) wereadded with 1 hr of incubation. The plates were washed thor-oughly and soaked with 100 ml of 1,2-phenylenediamine dihy-drochloride/hydrogen peroxide (DAKO, Norway) solution for15 min. The reaction was quenched by adding 100 ml of 0.5 MH2SO4 per well. OD492 readings were recorded and end point

titers were defined as the highest serum dilution that resultedin an absorbance value of greater than or equal to 0.1 OD492

(2.5 times the background value).

In vivo Ad5SEAP study

A total of 1 3 109 vp Ad5SEAP (an adenovirus vector thatexpresses a secreted form of the human placental alkaline phos-phatase) were mixed with 1–10 volume sof serum samples col-lected from mice that received two doses of either Ad5gag (1 3

108 vp) or Ad5/Ad6-gag (108 vp) or from untreated mice. Atleast 30 min after mixing, the vector was injected i.m. intoBALB/c (n 5 5). Serum samples were collected at 3, 6, 10, and17 days after administration. Then 5-ml aliquots of each serum

RECOMBINANT CHIMERIC ADENOVECTORS 315

FIG.1. Strategy used to generate chimeric Ad5-gag adenoviruses.

were assayed in duplicates for SEAP levels using a chemilu-minescent detection kit from Tropix (Bedford, MA) and refer-enced to a human protein standard (Sigma, MO).

RESULTS

Ad5-chimeric plasmid rescues in PER.C6 cell line

To maintain consistency between the chimeras and theWTAd5, we used the Ad5 construct that contained the Cla Isite within the hexon gene. The growth properties of this viruswere identical with the native Ad5 (i.e., non-Cla I containingAd5), as expected. Ad5/Ad1-gag, Ad5/Ad2-gag, and Ad5/Ad6-gag chimeras were all rescued into virus. The viruses were sub-sequently expanded and purified following five passages. Eachof the remaining viruses was continually propagated for sevenpassages by collecting total cells from each 15-cm dish andpreparing a concentrated cell lysate for infection into the nextpassage to ensure highest multiplicity of infection (MOI) pos-sible. Despite passage incubation times of 4–5 days, no CPEwas observed except for Ad5/Ad12-gag, which did show CPEafter passage 6 in duplicate infections. After passage 7, a gagELISA assay was performed on the supernatants from each ofthe chimeras to enable detection of any virus. No gag expres-sion was detected in any of the chimeras, with the exception ofAd5/Ad1-gag, Ad5/Ad2-gag, Ad5/Ad6-gag, and Ad5/Ad12-gag.

Rescue of transgene-less chimeric adenovectors

The rescues of these remaining 14 chimeras were unsuc-cessful despite repeated attempts using PER.C6 cells and 293cells. To address whether the gag transgene had an adverse ef-fect on the rescue and propagation of these viruses, 10 chime-ric adenovectors (Ad1, Ad2, Ad4, Ad5, Ad7, Ad9, Ad12, Ad13,Ad17, Ad35) that did not contain a transgene were constructed.As in the case of gag-containing chimeras, neither Ad4-, Ad7-,Ad9-, Ad13-, Ad17-, nor Ad35-based viruses could be propa-gated successfully.

Amplification and infectivity of Ad5 chimerascompared with WT Ad5

In vitro infections were performed at a MOI of 1 and 10 tostudy growth rates compared with Ad5. Infectivity of the re-combinant adenovirus preparations was analyzed in two celllines, 293 and PER.C6. Six-well plates were seeded with 5.6 3106 cells/well of 293 or 7.15 3 107 cells/well of PER.C6. Ad5-gag and the gag-containing chimeras (Ad5/Ad1-gag, Ad5/Ad2-gag, Ad5/Ad12-gag) were infected at MOI 1 and MOI 10 intoeach set of plates. The growth was observed under microscopefor CPE at 48, 72, 96,120, and 144 hr post infection. 293 cellswere far more susceptible to infection than PER.C6 cells as seenby a more rapid CPE formation in 293. At a MOI of 10, it tookAd5-gag, Ad5/Ad1-gag, and Ad5/Ad2-gag 96 hr to reach com-plete CPE in PER.C6 compared to only 48 hr in 293 cells.Within the same cell line, it was only at the lower MOI that wenoticed differences in growth kinetics between the chimericadenoviruses and Ad5-gag. Ad5/Ad1-gag and Ad5/Ad2-gagshowed slower progression to CPE compared with native Ad5-

gag. Ad5/Ad12-gag chimera took the longest to reach CPE. Thisdecreased growth was evident during propagation of the viruswhich consistently produced (at least 10-fold) lower yield.

We observed differences in the expression levels of gag fol-lowing infection of COS cells with Ad5-gag and chimericviruses. In all cases, the cell pellet was tested for gag levels byELISA assay (Coulter HIV-1 p24 Antigen assay, ImmunotechInc.). Infection at a MOI of 1 showed equivalent levels of gagexpression between Ad5-gag and Ad5/Ad6-gag. However, theAd5/Ad12-gag chimera consistently showed four-fold lowerlevels of gag expression at the same MOI (data not shown).This decreased expression level indicates a loss in infectivityof the chimeric virus. In a separate study, Ad5/Ad1-gag andAd5/Ad2-gag showed similar (if not slightly reduced) levels ofgag expression to Ad5-gag (data not shown).

In vivo rodent study to test for circumvention of pre-existing Ad5 immunity

Two chimeric constructs, Ad5/Ad6-gag and Ad5/Ad12-gag,were analyzed for their ability to evade pre-existing vector im-munity in rodents using anti-gag IgG response as a marker. Pre-vious studies have established that antibody levels correlate wellwith amounts of gene expression following i.m. genetic injec-tion (Chastain et al., 2001). BALB/c mice were treated withone or two doses of either Ad5-gag, Ad5/Ad6-gag, orAd5/Ad12-gag vector. A single 1 3 108 vp dose of Ad5-gag orAd5/Ad6-gag resulted in cohort mean titers of .3000 and asecond immunization with either vector boosted the levels ofanti-gag IgG by 30- to 50-fold (Fig. 2). In contrast, no signifi-cant anti-gag antibody response (cohort geometric mean titer ofapprox. 1000) was observed following two doses of Ad5/Ad12-gag vector in BALB/c mice (data not shown). When BALB/cmice were treated i.m. with two doses of 1 3 108 vp “empty”Ad5 virus to generate high levels of Ad5 circulating antibod-ies prior to Ad-gag immunization, circulating gag-specific IgGlevels were at the limit of detection following one or two dosesof the Ad5-gag adenovirus. A single 1 3 108 vp dose ofAd5/Ad6-gag in the pre-exposed animals elicited anti-gag lev-els that were at least 10-fold higher than those with the Ad5-gag construct in similarly pre-exposed mice. A second immu-nization with Ad5/Ad6-gag vector resulted in 100-fold increasein anti-gag IgG; these levels were at least 1000-fold higher thanthose achieved with a second dose of Ad5-gag in pre-exposedanimals. However, the anti-gag titers achieved following oneor two doses of Ad5/Ad6-gag in pre-exposed animals were 10-to 30-fold lower, respectively, compared to the levels observedin non-pretreated animals. No significant anti-gag antibodieswere detected in pre-exposed animals immunized with theAd5/Ad12-gag construct (data not shown).

A separate experiment was designed to address directly theextent to which antibody-mediated virus neutralization is at-tributed to anti-hexon antibodies. Specifically, serum samplesobtained from BALB/c mice that received two equivalent dosesof either the Ad5-gag or Ad5/Ad6-gag virus were incubatedwith Ad5 vector that expresses human SEAP. BALB/c micewere treated i.m. with these Ad5SEAP/serum combinations;circulating SEAP levels were determined in serum samples col-lected from the treated animals at various times. As controls,the Ad5SEAP was incubated with serum from naïve animals or

YOUIL ET AL.316

with saline solution prior to injection into naïve mice. Results(Fig. 3) clearly showed that anti-Ad5 antibodies derived fromanimals vaccinated with Ad5-gag were enough to totally ablateSEAP expression from the Ad5SEAP virus. Pre-incubation ofthe Ad5SEAP virus with serum obtained from BALB/c micethat received two doses of the Ad5/Ad6-gag virus, in contrast,did not affect the levels and kinetics of transgene expressionfrom the Ad5SEAP virus. This indicated that antibodies di-rected toward Ad6 hexon and Ad5 non-hexon proteins were noteffective at neutralizing the Ad5SEAP virus.

DISCUSSION

One of the most significant drawbacks to the usefulness ofhuman adenoviruses for clinical applications is the prevalenceof established immunity against the vector. Several approacheshave been described to allow circumvention of pre-existingadenovector immunity. They include formulating the virus withprotective polymers such as polyethylene glycol (PEG) (O’Ri-ordan et al., 1999), poly-lactic-glycolic acid (PLGA) micro-spheres (Mathews et al., 1999), or lipids (Meunier-Durmort etal., 1997; Lee et al., 2000). An obvious way to overcome anti-adenovirus neutralizing immunity is to convert to the adminis-tration of an alternate, rare human serotype (Mastrangeli et al.,1996; Mack et al., 1997) or non-human adenoviruses (Mittal etal., 1995; Kass-Eisler et al., 1996; Hofmann et al., 1999; Mi-chou et al., 1999; Reddy et al., 1999).

Although alternate serotype switching (particularly of therarer serotypes) may provide the best means of circumventing

pre-existing humoral immunity, in practice it provides a num-ber of concerns. First, serotypes other than Ad5, have not beenas studied extensively. Hence greater knowledge of their safetyand tropism profiles need to be generated before they can besafely administered for human use. Second, sequencing and an-notation of these serotypes must be performed to generate thereplication-incompetent adenovectors with deletion of their E1and/or E3 (or other) genes. Third, alternate serotypes have thepotential of targeting different cell types (Arnberg et al., 1997;Chillon et al., 1999). Different serotypes use different cellularreceptors and binding strategies for attachment (Defer et al.,1990; Stevenson et al., 1995; Roelvink et al., 1996; Freimuthet al., 1999; Arnberg et al., 2000). Fourth, use of alternateserotypes (outside of subgroup C) will most likely require gen-eration of new complementing cell lines to support their prop-agation.

In this study, we have taken a conservative approach to over-come pre-existing or acquired anti-adenoviral immunity byswapping the entire Ad5 hexon with those of differentserotypes. Infectious chimeric adenoviruses were rescued andpropagated when the hexon switch was performed within thesame subgroup. Recombinant chimeric adenoviruses were gen-erated where the Ad5 hexon gene was totally replaced with theAd1, Ad2, and Ad6 hexon genes. Only one non-subgroup Chexon (Ad12, subgroup A) was found to replace the Ad5 hexonand produce viable viruses effectively. Another subgroup A(Ad18) did not rescue along with the four serotypes from sub-group B, eight serotypes from subgroup D, and Ad4 (subgroupE). In each case, the rescue was performed using fully con-structed recombinant adenovector plasmid, so failure of recov-

RECOMBINANT CHIMERIC ADENOVECTORS 317

FIG. 2. Anti-gag IgG titers in BALB/c mice treated with one or two 1 3 108 vp doses of Ad5-gag (left panel) or Ad5/Ad6-gag (right panel) in the face of pre-existing anti-Ad5 immunity. Individual IgG titers are shown (open symbols) for animals thatwere either pretreated with non-gag-encoding Ad5 vector (open diamonds) or not (open circle). The geometic mean titers foreach cohort of 5 mice are indicated (solid symbols) along with the standard errors of the mean (bar). A nonresponder for this as-say is given a score of 50 or half the lowest serum dilution used. The responses seen in untreated or naïve animals are also shown(0 dose).

ery was not due to the lack of homologous recombination tak-ing place in the cell line. Furthermore, the Ad12 and Ad18hexon genes from each of our chimeric adenovectors were se-quenced in their entirety. Both hexon genes were present as asingle and complete ORF, hence eliminating the possibility ofPCR induced errors being the cause behind our failure to res-cue the Ad5/Ad18-gag chimera. A more likely possibility forthe inability to rescue certain alternate serotypes may be due toa lack of structural compatibility between the alternate hexonproteins and cognate proteins during process of assembly (e.g.,100-kDa scaffolding protein) and/or in the final capsid (e.g.,penton, fiber, protein IX).

Herein, evaluation of anti-gag immune responses followingi.m. injection of the gag-encoding chimeric viruses in Ad5-preimmunized animals suggests that the simple exchange of thehexon coating (Ad6 for Ad5) of the viral capsid provides a fea-sible avenue for partially circumventing pre-established anti-Ad5 immunity. Gall et al. (1998) had similarly shown that liverchloramphenicol acetyltransferase (CAT) expression followingintravenous injection of Ad5 vector was reduced only 100-foldwhen the animals were preimmunized with Ad2-hexon-bearingAd5 chimera instead of 10,000-fold using an Ad5 vector bear-ing the Ad5 hexon. Similar qualitative results were observedby Roy et al. (1998) using an b-galactosidase-encodingAd5/Ad12 construct. It is clear from these studies and ours thatother components of anti-Ad immunity contribute to loweredvirus efficacy; these would include antibodies directed againstthe penton and fiber proteins (Wohlfart et al., 1985; Gahery-

Segard et al., 1998) and cross-reactive Ad-specific cytotoxic Tcells (Smith et al., 1998).

In this paper, we have addressed the relative extent to whichantibodies directed to hexon versus non-hexon proteins con-tribute to virus neutralization. In vitro neutralization experimentstudies (data not shown) indicated that the chimeric adenovec-tors Ad5/Ad6-gag and Ad5/Ad12-gag were not neutralizedwhen mixed with serum containing Ad5 antibodies. A rodentstudy (Fig. 3) also showed that serum antibodies elicited by theAd5/Ad6-gag chimera were incapable of neutralizing (even par-tially) the Ad5SEAP virus; such antibodies in principle shouldinclude anti-Ad6 hexon and anti-Ad5 fiber/penton activities.Gall et al. (1998) also reported that replacement of Ad5 hexonprotein was 100-fold more effective in improving the efficacyof a repeat administration than a swap of the fiber with thatfrom an alternative serotype. All of these studies strongly sup-port that the hexon contains the major antigenic determinantsfor neutralizing antibodies against the virus. These further im-ply that cross-reactive Ad5-specific CTLs (Smith et al., 1998;Smyth and Trapani, 1998) play a very significant role in mod-ulating the efficacy of the Ad5/Ad6-gag chimera in Ad5-pre-immunized animals; these CTLs would include those directedat non-hexon proteins as well as epitopes common to both Ad5and Ad6 hexons. To address the contributions of cross-reactiveCTL directly, it is necessary to develop quantitative assays tomeasure them. We extrapolate that the Ad5/Ad1-gag andAd5/Ad2-gag chimeric vectors would behave similarly as theAd5/Ad6-gag vector.

YOUIL ET AL.318

FIG. 3. Kinetics of SEAP expression in BALB/c mice vaccinated i.m. with 1 3 109 vp Ad5SEAP mixed with serum samplesfrom animals previously vaccinated with Ad5-gag or Ad5/Ad6-gag. At 30 min prior to administration, 1 3 109 vp Ad5SEAP wasmixed with serum samples from mice with 1 3 108 vp Ad5-gag (filled diamonds), mice vaccinated twice with 1 3 108 vpAd5/Ad6-gag (open squares), or untreated mice (open diamonds). As contrals, BALB/c mice were either not treated with Ad5SEAP(filled circles) or treated with 1 3 109 vp Ad5SEAP in the absence of any sera (open circles). Shown are the cohort geometricmeans and the associated standard means.

The Ad5/Ad12-gag chimera showed significant loss of in-fectivity on the basis of reduced growth kinetics, lower titersper viral particles, and lower gag expression per viral particles.These results could be due to defective hexon/penton interac-tions which could, in turn, lead to improper association of thefiber protein into the capsid structure and hence, reduced viralbinding with the cell receptor molecules. The poor anti-gag im-munogenicity following i.m. injection of the Ad5/Ad12-gag(about 1000-fold lower titers than those of Ad5-gag orAd5/Ad6-gag) suggests that this chimera has an even lower in-trinsic infectivity of muscle cells compared to COS cells; in thelatter, levels of in vitro gag expression were only four-foldlower than those of Ad5-gag or Ad5/Ad6-gag. Roy et al. (1998)have constructed an Ad5/Ad12-gag chimeric adenovector, inwhich only a portion (L1–L4) of the Ad5 was exchanged withthat of Ad12. The yields of this chimera were likewise signif-icantly lower (100-fold) than those of Ad5; however, its parti-cle-to-pfu ratio was reportedly similar to that of Ad5, perhapssuggesting less perturbation of the capsid interactions with themore limited L1–L4 swap. Nevertheless, it should be noted thatpoor overall growth properties of these Ad5/Ad12 chimericviruses present a downstream problem in terms of large-scalemanufacturing yields for any clinical application. In addition,there may be potential genetic instability associated with theserecombinant adenoviruses that will further impact their practi-cal use.

In summary, our results show that the hexon exchange strat-egy is a practical approach to overcoming pre-existing neutral-izing immunity so long as the exchange occurs within mem-bers of the same subgroup. Whether such an approach will beuseful to circumvent virus-specific cellular-mediated immunityrequires further study.

ACKNOWLEDGMENTS

We would like to thank Sara Case and Jing Zhao for sup-plying the DNA for the alternate adenoviruses, Dr. Ling Chenfor supplying the pAd5-gag adenovector used in this study, andDr. Michael Chastain for helpful discussions.

REFERENCES

ADAM, E., LENGYEL, A., TAKACS, M., ERDEI, J., FACHET, J.,and NASZ, I. (1986). Grouping of monoclonal antibodies to adeno-virus hexons by their cross-reactivity. Arch. Virol. 87, 61–71.

ARNBERG, N., MEI, Y-F., and WADELL, G. (1997). Fiber genes ofadenoviruses with tropism for the eye and the genital tract. Virology227, 239–244.

ARNBERG, N., EDLUND, K., KIDD, A.H., and WADELL, G. (2000).Adenovirus Type 37 uses sialic acid as a cellular receptor. J. Virol.74, 42–48.

CHARTIER, C., DEGRYSE, E., GANZTER, M., DIETERLE, A.,PAVIRANI, A., and MEHTALI, M. (1996) Efficient generation ofrecombinant adenovirus vector by homologous recombination inEscherichia coli. J. Virol. 70, 4805–4810.

CHASTAIN, M., SIMON, A. J., SOPER, K. A., HOLDER, D. J.,MONTGOMERY, D., SAGAR, S. L., CASIMIRO, D. R., and MID-DAUGH, C. R. (2001). Antigen levels and antibody titers after DNAvaccination. J. Pharm. Sci. 90, 474–484.

CHILLON, M., BOSCH, A., ZABNER, J., LAW, L., ARMENTANO,D., WELSH, M.J., and DAVIDSON, B.L. (1999). Group D aden-oviruses infect primary central nervous system cells more efficientlythan those from Group C. J. Virol. 73, 2537–2540.

CHROBOCZEK, J., BIEBER, F., and JACROT, B. (1992). The se-quence of the genome of adenovirus type 5 and its comparison withthe genome of adenovirus type 2. Virology 186, 280–285.

COUCH, R.B., KASEL, J.A., PEREIRA, H.G., HAASE, A.T., andKNIGHT, V. (1973). Induction of immunity in man by crystallineadenovirus Type 5 capsid antigens (37438). Proc. Soc. Exp. Biol.Med. 143, 905–910.

CRAWFORD-MIKSZA, L., and SCHNURR, D.P. (1996). Analysis of15 adenovirus hexon proteins reveals the location and structure ofseven hypervariable regions containing serotype-specific residues. J.Virol. 70,1836–1844.

CROMPTON, J., TOOGOOD, C.I.A., WALLIS, N., and HAY, R.T.(1994). Expression of a foreign epitope on the surface of the adeno-virus hexon. J. Gen. Virol. 75,133–139.

DEFER, C., BELIN, M-T., CAILLET-BOUDIN, M-L., andBOULANGER, P. (1990). Human adenovirus-host cell interactions:comparative studies with members of subgroups B and C. J. Virol.64, 3661–3673.

FALLAUX, F.J., BOUT, A., VAN DER VELDE, I., VAN DEN WOL-LENBER, D.J., KEEGAN, K.M., AUGER, C., CRAMER, S.J., VANORMONDT, H., VAN DER EB, A.J., VALERIO, D., and HOEBEN,R.C. (1998). New helper cells and matched early region 1-deletedadenovirus vectors prevent generation of replication-competent ade-noviruses. Hum. Gene Ther. 9, 1909–1917.

FREIMUTH, P., SPRINGER, K., BEARD, C., HAINFELD, J., BEW-LEY, M., and FLANAGAN, J. (1999). Coxsackievirus and adeno-virus receptor amino Terminal immunoglobulin V-related domainbinds adenovirus type 2 and fiber knob from adenovirus type 12. J.Virol. 73, 1392–1398.

GAHERY-SEGARD, H., FARACE, F., GODFRIN, D., GASTON, J.,LENGAGNE, R., TURSZ, T., BOULANGER, P., and GUILLET, J-G. (1998). Immune Response to recombinant capsid proteins of ad-enovirus in humans: antifiber and antipenton base antibodies have asynergistic effect on neutralizing activity. J. Virol. 72, 2388–2397.

GALL, J.G.D., CRYSTAL, R.G., and FALCK-PEDERSEN, E. (1998).Construction and characterization of hexon-chimeric adenoviruses:specification of adenovirus serotype. J. Virol. 72,10260–10264.

GRAHAM, F.L. (1990) Adenoviruses as expression vectors and re-combinant vaccines. Trends Biotechnol. 8, 85–87.

HOFMANN, C., LOSER, P., CICHON, G., ARNOLD, W., BOTH,G.W., and STRAUSS, M. (1999). Ovine adenovirus vectors over-come preexisting Humoral immunity against human adenoviruses invivo. J. Virol. 73, 6930–6936.

JORNVALL, H., AKUSJARVI, G., ALESTROM, P., VON BAHR-LINDSTROM, H., PETTERSSON, U., APPELLA, E., FOWLER,A.V., and PHILIPSON, L. (1981). The adenovirus hexon protein:The primary structure of the polypeptide and its correlation with thehexon gene. J. Biol. Chem. 256, 6181– 6186.

KASS-EISLER, A., LEINWAND, L., GALL, J., BLOOM, B., andFALCK PEDERSEN, E. (1996). Circumventing the immune re-sponse to adenovirus mediated gene therapy. Gene Ther. 3, 154–162.

LEE, S.G., YOON, S.J., KIM, C.D., KIM,K., LIM,D.S., YEOM, Y.I.,SUNG, M-W., HEO, D.S., and KIM, N.K. (2000). Enhancement ofadenoviral transduction with polycationic liposomes in vivo. CancerGene Ther. 7, 1329–1335.

MACK, C.A., SONG, W-R., CARPENTER, H., WICKHAM, T.J.,KOVESDI, I., HARVEY, B-G., MAGOVERN, C.J., ISOM, O.W.,ROSENGART, T., FALCK PEDERSEN, E., HACKETT, N.R.,CRYSTAL, R.G., and MASTRANGELI, A. (1997). Circumventionof anti-adenovirus neutralizing immunity by administration of an ad-enoviral vector of an alternate serotype. Hum. Gene Ther. 8, 99–109.

MAIZEL, J.V.J., WHITE, D.O., and SCHARFF, M.D. (1968). The

RECOMBINANT CHIMERIC ADENOVECTORS 319

polypeptides of adenovirus. I. Evidence for multiple protein compo-nents in the virion and a comparison of types 2, 7A, and 12. Virol-ogy 36,115–125.

MASTRANGELI, A., HARVEY, B.G., YAO, J., WOLFF, G.,KOVESDI, I., CRYSTAL, R.G., and FALCK-PEDERSEN, E.(1996). “Sero-switch” adenovirus mediated in vivo gene transfer: cir-cumvention of anti-adenovirus humoral immune defenses against re-peat adenovirus administration by changing the adenovirus serotype.Hum. Gene Ther. 7, 79–87.

MATHEWS, C., JENKINS, G., HILFINGER, J., and DAVIDSON, B.(1999). Poly-L-lysine improves gene transfer with adenovirus for-mulated in PLGA microspheres. Gene Ther. 6,1558–1564.

MEUNIER-DURMORT C., PICART, R., RAGOT, T., PERRI-CAUDET, M., HAINQUE, B., and FOREST, C. (1997) Mechanismof adenovirus improvement of cationic liposome-mediated genetransfer. Biochim. Biophys. Acta 1330, 8–16.

MICHOU, A-I., LEHRMANN, H., SALTIK, M., and COTTON, M.(1999). Mutational analysis of the avian adenovirus CELO, whichprovides a basis for gene delivery vectors. J. Virol. 73, 1399–1410.

MITTAL., S.K., PREVEC, L., GRAHAM, F.L., AND BABIUK, L.A.(1995) Development of a bovine adenovirus type 3-based expressionvector. J. Gen. Virol. 76, 93–102.

O’RIORDAN, C.R., LACHAPELLE, A., DELGADO, C., PARKES,V., WADSWORTH, S.C., SMITH, A.E., and FRANCIS, G.E.(1999). PEGylation of adenovirus with retention of infectivity andprotection from neutralizing antibody in vitro and in vivo. Hum. GeneTher. 10, 1349–1358.

REDDY, P.S., IDAMAKANTI, N., CHEN, Y., WHALE, T., BABIUK,L.A., MEHTALI, M., and TIKOO, S.K. (1999). Replication-defec-tive bovine adenovirus type 3 as an expression vector. J. Virol. 73,9137–9144.

ROBBINS, P.D., TAHARA, H., and GHIVIZZANI, S.C. (1998). Vi-ral vectors for gene therapy. Trends Biotechnol. 16, 35–40.

ROELVINK, P.W., KOVESKI, I., and WICKHAM, T.J. (1996) Com-parative analysis of adenovirus fiber-cell interaction: adenovirus type2 (Ad2) and Ad9 utilize the same cellular fiber receptor but use dif-ferent binding strategies for attachment. J. Virol. 70, 7614–7621.

ROY, S., SHIRLEY, P.S., MCCLELLAND, A., and KALEKO, M.(1998). Circumvention of immunity to the adenovirus major coat pro-tein hexon. J. Virol. 72, 6875–6879.

RUSSELL, W.C., BARTHA, A., DEJONG, J.C., FUJINAGA, K.,GINSBERG, H., HIERHOLZER, J.C., LI, Q.G., MAUTNER, V.,NASZ, I., and WADELL, G. (1991). Adenoviridae. In Classificationand Nomenclature of Viruses Fifth Report of the International Com-mittee on Taxonomy of Viruses. R.I.B. Francki, C.M. Gauqet, D.L.Knudson, and F. Brown, eds. (Springer, Vienna and New York) pp.140–144. (Arch. Virol. [Supp]2).

SMITH, C.A., WOODRUFF, L.S., ROONEY, C., and KITCHING-

MAN, G.R. (1998). Extensive cross-reactivity of adenovirus-specificcytotoxic T cells. Hum. Gene Ther. 9, 1419–1427

SMYTH, M.J., and TRAPANI, J.A. (1998) The relative role of Lym-phocyte granule exocytosis versus death receptor-mediated cytotox-icity in viral pathophysiology. J. Virol. 72, 1–9.

STEVENSON, S.C., ROLLENCE, M., WHITE, B., WEAVER, L., andMCCLELLAND, A. (1995) Human Adenovirus serotype 3 and 5bind to two different cellular receptors via the fiber head domain. J.Virol. 69, 2850–2857.

TOOGOOD, C.I.A., CROMPTON, J., and HAY, R.T. (1992) An-tipeptide antisera define neutralizing epitopes on the adenovirushexon. J. Gen. Virol. 73, 1429–1435.

WILLCOX, N., and MAUTNER, V. (1976a). Antigenic determinantsof adenovirus capsids. I. Measurement of antibody cross reactivity.J. Immunol. 116, 19–24.

WILLCOX, N., and MAUTNER, V. (1976b). Antigenic determinantsof adenovirus capsids. II. Homogeneity of hexons, and accessibilityof their determinants, in the virion. J. Immunol. 116, 25–29.

WOHLFART, C. (1988). Neutralization of adenovirus: kinetics, stoi-chiometry and mechanisms. J. Virol. 62, 2321–2328.

WOHLFART, C.E.G., SVENSSON, U.K., and EVERITT, E. (1985).Interaction between HeLa cells and adenovirus type 2 virions neu-tralized by different antisera. J. Virol. 56, 896–903.

YOUIL, R., TONER, T.J., SU, Q., and KASLOW, D.C. (2001). Rapidmethod for the isolation of full length adenoviral genomes by bac-terial intermolecular homologous recombination. J. Virol. Methods92, 91–97.

ZHENG, C., GOLSMITH, C.M., O’CONNELL, B.C., and BAUM, B.J.(2000). Adenoviral vector cytotoxicity depends in part on the trans-gene encoded. Biochem. Biophys. Res. Commun. 274, 767–771.

Address reprint requests to:Dr. Rima Youil

Merck & Co., Inc.Virus and Cell Biology

WP44L-206B770 Sumneytown Pike

P.O. Box 4West Point, PA, 19486

E-mail: rima_youil@merck.com

Received for publication July 2, 2001; accepted after revisionDecember 12, 2001.

Published online: January 11, 2002.

YOUIL ET AL.320