Content uploaded by Douglas E Brough
Author content
All content in this area was uploaded by Douglas E Brough
Content may be subject to copyright.
JOURNAL OF VIROLOGY,
0022-538X/97/$04.0010Nov. 1997, p. 8221–8229 Vol. 71, No. 11
Copyright © 1997, American Society for Microbiology
Increased In Vitro and In Vivo Gene Transfer by Adenovirus
Vectors Containing Chimeric Fiber Proteins
THOMAS J. WICKHAM,
1
* EDITH TZENG,
2
LARRY L. SHEARS II,
2
PETER W. ROELVINK,
1
YUAN LI,
1
GAI M. LEE,
1
DOUGLAS E. BROUGH,
1
ALENA LIZONOVA,
1
AND IMRE KOVESDI
1
GenVec, Inc., Rockville, Maryland 20852,
1
and Department of Surgery, Presbyterian University Hospital,
University of Pittsburgh, Pittsburgh, Pennsylvania 15213
2
Received 4 April 1997/Accepted 16 July 1997
Alteration of the natural tropism of adenovirus (Ad) will permit gene transfer into specific cell types and
thereby greatly broaden the scope of target diseases that can be treated by using Ad. We have constructed two
Ad vectors which contain modifications to the Ad fiber coat protein that redirect virus binding to either a
v
integrin [AdZ.F(RGD)] or heparan sulfate [AdZ.F(pK7)] cellular receptors. These vectors were constructed by
a novel method involving E4 rescue of an E4-deficient Ad with a transfer vector containing both the E4 region
and the modified fiber gene. AdZ.F(RGD) increased gene delivery to endothelial and smooth muscle cells
expressing a
v
integrins. Likewise, AdZ.F(pK7) increased transduction 5- to 500-fold in multiple cell types
lacking high levels of Ad fiber receptor, including macrophage, endothelial, smooth muscle, fibroblast, and T
cells. In addition, AdZ.F(pK7) significantly increased gene transfer in vivo to vascular smooth muscle cells of
the porcine iliac artery following balloon angioplasty. These vectors may therefore be useful in gene therapy for
vascular restenosis or for targeting endothelial cells in tumors. Although binding to the fiber receptor still
occurs with these vectors, they demonstrate the feasibility of tissue-specific receptor targeting in cells which
express low levels of Ad fiber receptor.
Adenovirus (Ad) has been widely used as a vector to deliver
genes to a number of tissues, including lung, vascular, neuro-
nal, and muscle tissue, in vivo (7, 11, 31, 41). Ad grows to high
titer, its genome is well characterized, and it delivers and ex-
presses genes in nonproliferating cells. However, previous
studies have found a low level of receptor-mediated binding of
Ad to many cells and tissues, including primary bronchial ep-
ithelium (8), smooth muscle (49), endothelium (49), and mac-
rophage and T cells (24). This reduced binding has been found
to result in lowered transduction efficiencies (49). Therefore,
one potential improvement to this system would be to target
the virus to receptors on specific cell types. This improvement
would increase the delivery efficiency to the target tissue and
reduce nonspecific transduction of nontarget tissue, thus al-
lowing lower vector doses to be administered with fewer un-
wanted side effects (7, 35).
Host cell infection by Ad involves two of its coat proteins
which interact with distinct cellular receptors (47). The fiber
protein, alone, mediates viral attachment to recently identified
cellular receptors. Not one but two unrelated cellular receptors
for the fiber have recently been reported (3, 23, 43). One
receptor, termed the coxsackievirus-Ad receptor because it
functions as a receptor for both Ad and coxsackievirus, con-
tains two immunoglobulin domains and has not been previ-
ously identified in any protein databases (3, 43). The a2 do-
main of major histocompatibility complex class I has also been
reported to mediate Ad attachment via the fiber (23). Follow-
ing fiber-mediated attachment to cells, penton base binds via
an RGD motif to a
v
b
3
and a
v
b
5
integrin receptors, which then
mediate virus internalization via receptor-mediated endocyto-
sis (2, 33, 37, 45, 47). The Ad type 5 (Ad5) penton base does
not function in virus attachment to host epithelial cells, which
is likely due to the steric hindrance imposed by the 32-nm-long
fiber protein (40, 47).
Due to the limited expression of coxsackievirus-Ad receptor
in many tissues, Ad tropism can be expanded by targeting its
binding to a broadly expressed receptor (48). For example,
heparan sulfate-containing receptors are broadly expressed on
many cell types and are known to bind to stretches of the
positively charged amino acids lysine and/or arginine (13, 26).
Another approach to modify tropism is to target Ad binding to
a more narrowly expressed or tissue-specific receptor, in order
to limit vector binding and transduction to a particular cell
type. For example, the a
v
integrin receptors are not expressed
in many cells unless they are stimulated by cytokines or growth
factors, which are produced as a result of wounding, infection,
or inflammation (17). For example, granulocyte-macrophage
and macrophage colony-stimulating factors (GM-CSF and G-
CSF) and basic fibroblast growth factor/vascular endothelial
growth factor have been shown to upregulate a
v
integrin ex-
pression in monocytes (9) and endothelial cells (12), respec-
tively. In addition, a
v
integrins can also be aberrantly expressed
in invasive glioblastomas and metastatic melanomas (1, 16).
Other successful strategies to target Ad have used antibodies
to redirect the virus to new receptors (10, 49). We previously
have found that a bispecific antibody which binds to a modified
Ad and directs its binding directly to a
v
integrins enhances
vector binding and transduction of endothelial and smooth
muscle cells (49). However, drawbacks to this method include
the additional production and characterization of the bispecific
antibodies, the potential clearance of virus by the Fc receptor,
and the potential activation of the complement system. The
most direct way to target Ad is to incorporate receptor-binding
motifs into the penton base or fiber coat proteins (36, 46).
While Ad5 vectors containing the fiber proteins from other
serotypes have been developed (14, 29), high-titer Ad vectors
that are targeted to known receptors through the addition of
defined receptor-binding motifs to the fiber protein have not
been reported.
* Corresponding author. Mailing address: GenVec, Inc., 12111
Parklawn Dr., Rockville, MD 20852. Phone: (301) 816-0396. Fax: (301)
816-0440.
8221
In this study, we report a novel system that has been used to
produce two Ads containing modifications to their fiber pro-
tein. These viruses can be grown to high titer and can target
specific receptors through the incorporation of high-affinity
peptide motifs into the fiber protein. One virus contains a
high-affinity, a
v
integrin-binding motif (28) which preferen-
tially increases the transduction of endothelial and smooth
muscle cells. A second vector contains a polylysine, heparin-
binding motif (13, 48) which increases the transduction of
multiple cell types, including macrophage, endothelium,
smooth muscle, glioblastoma, and T cells. These vectors dem-
onstrate the feasibility of designing and producing Ads which
can target tissue-restricted receptors to increase the efficiency
and/or specificity of gene transfer.
MATERIALS AND METHODS
Cells. Human alveolar carcinoma cells (A549), human embryonic kidney cells
(293), bovine endothelial cells (CPAE), primary human intestinal smooth muscle
cells (HISM), rat smooth muscle cells (A-10), mouse melanoma (B16-F1), hu-
man glioblastomas (U-118 and A172), human monocyte-like cells (THP-1 and
U-937), and primary human foreskin fibroblasts (Hs68) were obtained from the
American Type Culture Collection (Rockville, Md.). The a
v
-293 cells were
produced by transfecting 293 cells with the a
v
and b
3
integrin subunit genes.
A549, CPAE, A-10, 293, and a
v
-293 cells were maintained in Dulbecco’s mod-
ified Eagle’s medium (DMEM) supplemented with 5% calf serum (Gibco BRL,
Grand Island, N.Y.). The primary cells, HISM and Hs68, were maintained in
MCDB medium (Gibco BRL) supplemented with 10% fetal bovine serum and
bovine pituitary extract (Gibco BRL). Peripheral blood T lymphocytes were
isolated from buffy coats or leukopacks from normal NIH Blood Bank donors by
Ficoll-Hypaque density gradient sedimentation, plastic adherence, and nylon
wool adsorption (15). Resting cells were obtained from a discontinuous 30 and
40% Percoll gradient (15, 39) as the high-buoyant-density fraction. Fluorescence-
activated cell sorting using monoclonal antibody (MAb) OKT3, specific for the
CD3 receptor, showed that greater than 90% of the cells isolated in this manner
were CD3
1
. Peripheral blood monocytes were isolated from the fraction of
peripheral blood cells adhering to plastic. The monocytes were cultured for 10 to
14 days to allow differentiation into macrophages prior to transduction experi-
ments. All cells and cell lines were maintained in RPMI 1640 supplemented with
10% calf serum (Gibco BRL).
Viruses. The E1- and E3-deleted Ad, AdZ, contains the b-galactosidase gene
under a cytomegalovirus (CMV) promoter (GenVec, Inc., Rockville, Md.). AdZ
was grown in human embryonic kidney (293) cells which contain the comple-
mentary E1 region for virus growth (19). AdZ.F(pK7) and AdZ.F(RGD) were
derived from the E4-minus vector, AdZ.11A, to incorporate the additional
amino acids present on the C termini of their fiber proteins (see below).
AdZ.11A was constructed by exchanging the E1 expression cassette of
AdCFTR.11A to a CMV expression cassette expressing b-galactosidase.
AdCFTR.11A and AdZ.11A contain a complete deletion of E4 and inclusion of
the E4 spacer element as previously described (5). These vectors express a high
level of b-glucuronidase from the E4 promoter in 293 or 293-ORF6 cells.
All viruses were purified from infected 293 cells at 2 days postinfection by
three freeze-thaw cycles followed by three successive bandings on CsCl gradients
(30). Purified virus was dialyzed into 10 mM Tris–150 mM NaCl (pH 7.8)
containing 10 mM MgCl
2
and 3% sucrose and frozen at 280°C until required for
use. Radiolabeled Ad was made by infecting cells at a multiplicity of infection
(MOI) of 5 and adding 50 mCi of [
3
H]thymidine (Amersham, Arlington Heights,
Ill.) per ml to the medium of infected cells at 20 h postinfection. The infected
cells were then harvested at 60 h postinfection, and the virus was purified as
described above by two successive bandings on CsCl gradients. The activity of the
labeled viruses was approximately 10
4
virus particles/cpm. Infectious particle titer
(in fluorescent focus units [FFU]) was determined by using a fluorescent focus
assay on 293 cells (42).
The virus yield from infected 293 cells was determined by infecting 10
6
293
cells with 0.2 ml of virus for1hin6-cm-diameter plates at an MOI of 10 on day
0. The cells were harvested on 1, 2, and 3 days postinfection. The cells were spun
down and resuspended in 1 ml of phosphate-buffered saline (PBS) for AdZ and
AdZ.F(RGD). AdZ.F(pK7) was lysed in 1 ml of PBS containing 2 M NaCl to
facilitate release of virus particles from the cells.
Construction of transfer plasmids. The transfer plasmid, pNS 83-100, was
constructed by cloning from pGBS 59-100 the Ad5 NdeI-to-SalI fragment, which
spans the region of the Ad5 genome from map units 83 to 100, into plasmid
pNEB193 (New England Biolabs, Beverly, Mass.). The NdeI-MunI fragment was
replaced with a synthetic oligonucleotide comprising a BamHI site, which was
flanked by a 59NdeI site and a 39MunI site to facilitate cloning. The double-
stranded synthetic oligonucleotide fragment was created from the overlapping
synthetic single-stranded sense and antisense oligonucleotides, i.e., primers TAT
GGA GGA TCC AAT AAA GAA TCG TTT GTG TTA TGT TTC AAC GTG
TTT ATT TTT C and AAT TGA AAA ATA AAC ACG TTG AAA CAT AAC
ACA AAC GAT TCT TTA TTG GAT CCT CCA, respectively. The ends of the
overlapping oligomers were made to have overhangs compatible for direct clon-
ing into the NdeI and MunI sites. The resultant vector, pNS (DF), lacks all of the
coding sequences for the fiber gene but contains the entire Ad E4 coding
sequence. The plasmid retains the AATAAA polyadenylation signal (boldface)
included in the synthetic NdeI/MunI oligonucleotide and also incorporates a new
BamHI restriction site (underlined).
The transfer plasmid, pNS (F5*), which contains a mutated fiber gene with a
BamHI site between the last fiber protein codon and the fiber protein stop
codon, was constructed from pNS (DF). The mutated fiber gene was incorpo-
rated into the fiber-minus (DF) pNS plasmid, using synthetic sense and antisense
oligonucleotide primers to amplify the fiber gene by PCR while incorporating a
BamHI site following the last codon of the fiber gene to create the mutant fiber
gene. This BamHI site also serves to code for the amino acids glycine and serine.
The primers used to amplify from the NdeI site to the C-terminal coding regions
of the fiber gene from Ad5 genome DNA were antisense primer T CCC CCC
GGG TCT AGA TTA GGA TCC TTC TTG GGC AAT GTA TGA (stop site
in boldface; BamHI site underlined) and the sense primer CGT GTA TCC ATA
TGA CAC AGA (NdeI site underlined). The PCR product was then cut with
NdeI and BamHI and cloned into the NdeI/BamHI sites of pNS (DF).
The mutant transfer plasmids containing sequences encoding an amino acid
glycine-serine (GS) repeat linker, a targeting sequence, and a stop codon were
made by cloning synthetic oligonucleotides into the BamHI site of pNS (F5*).
The overlapping synthetic oligonucleotides used to make the transfer plasmid
pNS (F5) pK7 were GA TCA GGA TCA GGT TCA GGG AGT GGC TCT
AAA AAG AAG AAA AAG AAG AAG TAA G (sense) and GA TCC TTA
CTT CTT CTT TTT CTT CTT TTT AGA GCC ACT CCC TGA ACC TGA
TCC T (antisense). The oligonucleotides used to make the transfer plasmid pNS
(F5) RGD were GA TCA GGA TCA GGT TCA GGG AGT GGC TCT GCC
TGC GAC TGT CGC GGC GAT TGT TTT TGC GGT TAA G (sense) and GA
TCC TTA ACC GCA AAA ACA ATC GCC GCG ACA GTC GCA GGC AGA
GCC ACT CCC TGA ACC TGA TCC T (antisense). The sense and antisense
oligonucleotides were mixed in equimolar ratios and cloned into the BamHI site
of pNS (F5*) to create pNS (F5) pK7 and pNS (F5) RGD. Sequencing in both
directions across the region of the inserts verified that the clones contained the
appropriate sequence.
A third version of pNS (F5*) was also created to allow multiple targeting
sequences to be inserted following a preexisting poly(GS) spacer. The sense and
antisense oligonucleotides used to make the vector pNS pGS were GA TCC
GGT TCA GGA TCT GGC AGT GGC TCG ACT AGT TAA A and GA TCT
TTA ACT AGT CGA GCC ACT GCC AGA TCC TGA ACC G, respectively.
This sequence encoded amino acids GSGSGSGSGSTS and contained an SpeI
site in the TS codons, which facilitated the direct cloning of targeting sequences
following the poly(GS) spacer region. Other oligonucleotide sense and antisense
pairs were synthesized to clone into the SpeI site of pNS pGS. The sense and
antisense pairs encoding the poly(arginine-glycine-aspartate [RGD]) sequence
(GRGDTF)
3
were CT AGT GGA AGA GGA GAT ACT TTT GGC CGC GGC
GAC ACG TTC GGA AGG GGG GAT ACA TTT T and CT AGA AAA TGT
ATC CCC CCT TCC GAA CGT GTC GCC GCG GCC AAA AGT ATC TCC
TCT TCC A, respectively. The sense and antisense pairs encoding the poly
(YIGSR) sequence (GYIGSR)
3
designed to target the high-affinity laminin
receptor (18) were CT AGT GGA TAC ATC GGC AGT CGC GGT TAC ATT
GGG TCC CGA GGA TAT ATA GGC TCA AGA T and CT AGA TCT TGA
GCC TAT ATA TCC TCG GGA CCC AAT GTA ACC GCG ACT GCC GAT
GTA TCC A, respectively. The sense and antisense pairs encoding the E-
selectin-binding sequence DITWDQLWDLMK (32) were CT AGA GAC AAT
ACC TGG GAC CAG CTT TGG GAC CTT ATG AAG A and CT AGT CTT
CAT AAG GTC CCA AAG CTG GTC CCA GGT AAT GTC T, respectively.
The sense and antisense pairs encoding the laminin receptor-binding sequence
(SIKVAV)
2
(27) were CT AGT GCC GCC AGC ATT AAG GTG GCT GTC
TCG ATC AAA GTT GCG GTA TAA GAC GT and C TTA TAC CGC AAC
TTT GAT CGA GAC AGC CAC CTT AAT GCT GGC GGC A, respectively.
The sense and antisense pairs encoding the FLAG peptide sequence DYKDDD
DK were CT AGA GAC TAC AAG GAC GAC GAT GAT AAG A and CT
AGT CTT ATC ATC GTC GTC CTT GTA GTC T, respectively. Sequencing in
both directions across the region of the inserts verified that the clones contained
the appropriate sequences.
Construction of chimeric Ad vectors. The plasmid DNAs from pNS (F5) pK7,
pNS (F5) RGD, and the other pNS transfer plasmids were linearized with SalI,
purified, and transfected by using calcium phosphate into 293 cells (20). The
transfected 293 cells had been preincubated with the E1-, E3-, and E4-deleted
construct AdZ.11A (GenVec) at an MOI of 1 FFU per cell, 1 h prior to
transfection with the pNS plasmids (5). Recombination of the E4
1
pNS plasmid
with the E4-deleted vector resulted in the rescue of an E1
2
E3
2
E4
1
vector
capable of replication in 293 cells. The resultant vectors, AdZ.F(pK7) and
AdZ.F(RGD), and the other modified vectors were isolated in two successive
rounds of plaque purification on 293 cells. Each vector was verified to contain
the correct insert by sequencing PCR products derived from virus DNA template
by using primers that span the region of the insert DNA. Restriction analysis of
Ad DNA from each of the viruses showed that the viruses were pure and
contained the BamHI or SpeI restriction site unique to the correctly constructed
virus (22).
8222 WICKHAM ET AL. J. VIROL.
Expression of recombinant fiber proteins by using the baculovirus expression
system. Recombinant fiber and penton base were produced in baculovirus as
previously described (47). High-titer recombinant fiber and penton base bacu-
lovirus stocks were used to produce the recombinant protein in TN 5 cells.
Baculovirus-infected cells were pelleted at 3 days postinfection and resuspended
in PBS containing the protease inhibitors leupeptin (5 mg/ml), aprotinin (5
mg/ml), and phenylmethylsulfonyl fluoride (1 mM). The cell suspension was
subjected to three freeze-thaw cycles and then cleared by centrifugation at
15,000 3gfor 15 min. The recombinant proteins were then purified from the
cellular proteins as previously described (47).
Immunoblot analysis of virus particles. Purified virus particles (2 310
10
)ina
volume of 10 ml were diluted 1:1 in running buffer and loaded onto a 0.1%
sodium dodecyl sulfate (SDS)–9% acrylamide gel. The gel was run at 150 mV,
and the protein was then transferred to nitrocellulose. The nitrocellulose was
blocked with 5% dry milk and then probed with a combination of rabbit poly-
clonal antiserum against denatured Ad5 virions (1:1,000) and against fiber pro-
tein (1:5,000). Proteins were then detected by using anti-rabbit-peroxidase (1:
5,000) and the ECL Western blotting analysis system (Amersham). A control
blot probed with only polyclonal serum to the fiber protein verified that the
shifted bands in the previous immunoblot were fiber protein.
Binding assays using recombinant Ad vectors. Confluent monolayers of 293,
CPAE, and A-10 cells in collagen-coated 24-well plates were preincubated with
300 ml of DMEM–20 mM HEPES containing recombinant fiber protein (5
mg/ml), heparin (3 mg/ml), penton base (50 mg/ml), a combination of fiber plus
heparin, a combination of fiber plus penton base, or no competitor for1hat4°C.
3
H-labeled AdZ, AdZ.F(RGD), or AdZ.F(pK7) (5,000 to 25,000 cpm) in a
volume of 10 ml was added to the wells, which were then rocked for 90 min at
room temperature. The cells were washed three times with PBS containing 1 mM
MgCl
2
, solubilized in 200 ml of a 1% SDS solution, and counted in a scintillation
counter. The bound counts were then expressed as a percentage of the radioac-
tive counts added per well.
Gene expression assays. Approximately 0.5 310
6
to 1.0 310
6
A549, HISM,
CPAE, Hs68, or macrophage cells were seeded onto 6-cm-diameter plates 1 to 2
days prior to experiments. Increasing doses of 10
7
,10
8
,10
9
,or10
10
AdZ,
AdZ.F(RGD), or AdZ.F(pK7) virus particles were incubated with the cells in
a volume of 0.2 ml for 1 h. For the human high-density T cells, the experiments
were performed by incubating the cells in suspension in a volume of 0.3 ml for
1 h. The cells were then washed three times with DMEM and cultured in DMEM
containing 10% calf serum at 37°C. After 2 days of culture, the cells were lysed
in1mlof13reporter lysis buffer containing 10 mM EDTA (Promega, Madison,
Wis.). Galactosidase activity in the cell lysates was then assayed by using a
Galactolight fluorometric assay kit (Tropix, Bedford, Mass.).
In vivo gene expression. Male domestic pigs (12 to 15 kg; Walter Whippo,
Enon Valley, Pa.) were placed under general anesthesia, and bilateral common
iliac arteries were exposed through a midline laparotomy incision. After obtain-
ing distal and proximal vascular control, arterial injury was performed with a
4-French Fogarty catheter, inserted through a distal side branch of the common
iliac artery, and inflated to 2 atm for 5 min. Ad solution (2 310
10
particles/ml,
approximately 500 ml/iliac artery) was instilled for 30 min. After infection, the Ad
solution was evacuated, the side branch was ligated, and blood flow was rees-
tablished. AdZ.F(pK7) was instilled in the right common iliac artery, while AdZ
was instilled in the left. No anticoagulation or antiplatelet agents were admin-
istered.
Three days after viral transduction, animals were euthanized with potassium
chloride and sodium pentothal overdose, and the iliac arteries were collected.
The arteries were opened on their long axes, fixed in 2% paraformaldehyde in
PBS for 1 h, and then stained with 5-bromo-4-chloro-3-indolyl-b-D-galactopy-
ranoside (X-Gal) overnight. Vessels were then imaged by light microscopy (Ni-
kon light microscope) for the amount of blue-staining cells on the luminal
surface. In addition, some vessels were cross-sectioned (10-mm thickness) after
X-Gal staining and counterstained with eosin prior to imaging.
RESULTS
Construction of Ad particles containing chimeric fiber pro-
teins. An Ad transfer plasmid was constructed to allow the
convenient addition of ligands onto the C terminus of the Ad
fiber protein. The transfer plasmid contained the Ad5 DNA
sequence from map units 83 to 100 of the Ad5 genome, in-
cluding most of the fiber gene as well as the complete E4
region. A unique BamHI restriction site was inserted between
the last amino acid codon and the fiber protein termination
codon to allow the cloning of sequence encoding receptor-
binding ligands directly onto the end of the fiber.
Multiple peptide motifs were chosen to add onto the C
terminus (Table 1). Of these motifs, at least two were known to
interact with high affinity to heparin sulfate proteoglycans or to
a
v
integrins (Fig. 1). One motif contained an RGD integrin-
binding sequence, ACDCRGDCFCG, which has been shown
to bind to a
v
integrins with approximately 100-fold-higher af-
finity than the motif GRGDSP from fibronectin (28). The
second motif incorporated a string of seven lysines (KKKK
KKK) which contains multiple overlapping consensus motifs
that allow high-affinity binding to heparin and heparan sulfate
FIG. 1. Diagram depicting the linker and two ligand sequences used to target
Ad binding to adhesion molecules. The vector AdZ.F(RGD) contains a targeting
sequence with high affinity for a
v
integrins. The vector AdZ.F(pK7) contains a
stretch of seven lysines.
TABLE 1. Construction and isolation of Ad vectors containing fiber C-terminal peptides
Peptide addition Target No. of white
plaques
picked
No. of
plaques
positive
Binding
of target
(GS)
5
ACDCRGDCFCG a
v
Integrins 5 5 Yes
(GS)
5
KKKKKKK Heparan sulfate 5 5 Yes
(GS)
5
TRDYKDDDDKTS Anti-FLAG MAb 5 5 Yes
a
(GS)
5
TS(GRGDTF)
3
SS a
v
Integrins 5 5 No
b
(GS)
5
TS(GYIGSR)
3
SS Laminin receptor 5 5 No
c
(GS)
5
TRSDITWDQLWDLMKTS E-selectin 0 0 ND
d
(GS)
5
TSAA(SIKVAV)
2
Laminin receptor 0 0 ND
a
293 cells infected by this virus were positive by immunofluorescence with the anti-FLAG MAb M2.
b
Compared to unmodified virus, this virus did not increase the transduction of a
v
-expressing 293 cells preincubated with fiber. The other RGD-modified virus
comprising the sequence ACDCRGDCFCG did increase transduction in these cells preincubated with fiber.
c
Compared to unmodified virus, this virus did not increase the transduction of B16-F1, macrophage, HISM, or A549 cells preincubated with fiber.
d
ND, not determined.
VOL. 71, 1997 TARGETING OF ADENOVIRUS TO CELLS 8223
(13). Oligonucleotides encoding either the RGD motif or the
polylysine motif plus a short spacer encoding five GS repeats
were synthesized and cloned onto the end of the fiber gene.
The spacer sequence was encoded before the targeting se-
quence to optimize accessibility of the ligand to a target re-
ceptor. The other receptor or ligand-binding sequences chosen
to incorporate into the C terminus of the fiber protein are
shown in Table 1. These sequences have been reported to
mediate binding to the anti-FLAG MAb (49), E-selectin (32),
and laminin receptors (18, 27), as well as a sequence that was
comprised of three tandem repeats of a lower-affinity RGD
motif.
Recombinant vectors containing the fiber chimeras were
efficiently produced by using an E4 rescue technique (Fig. 2).
An E4-deleted vector, AdZ.11A, containing the lacZ gene
driven by the CMV promoter was used to infect 293 cells that
had been transfected with the transfer plasmids containing the
F(RGD), F(pK7), or other chimeric fiber genes. A single,
homologous recombination event between the E4-deleted Ad
and the E4-positive plasmid produced a recombinant virus that
was replication competent in the 293 cells. The resultant re-
combinant vector was isolated with high efficiency following
plaque purification on 293 cells. Because only E4-expressing
vectors replicate in 293 cells, this method resulted in highly
efficient recovery of the recombinant vector (Table 1). In ad-
dition, the AdZ.11A vector contains the b-glucuronidase gene
under control of the E4 promoter. Consequently, only the
recombinant plaques remain white after the infected-trans-
fected cell lysates are plaqued on 293 cells in the presence of
the substrate X-Glu, which turns glucuronidase-expressing
cells blue. Six of the eight infections-transfections resulted in
white-plaque isolates at the 10
22
dilution of the infection-
transfection cell lysate. Of those that were further analyzed, all
were shown by PCR to contain the C-terminal addition to the
fiber gene. In addition, there was no detectable b-glucuroni-
dase activity in 293 cells infected with isolates that had been
twice plaque purified. Two of the constructs could not be
isolated despite repeated attempts. The inability to isolate
these constructs suggests that the peptides were incompatible
with the correct folding of the fiber protein. These constructs
encoded C-terminal additions that were five or six amino acids
shorter than the two longest constructs that were successfully
made. These observations suggest that the targeting peptide
composition, rather than absolute peptide length, is most crit-
ical for making short peptide additions to fiber C terminus.
Characterization of the chimeric fiber proteins incorporated
into Ad vectors. Plaque isolates of the two recombinant vectors
AdZ.F(RGD) and AdZ.F(pK7) were amplified and character-
ized to verify that they encoded the desired receptor-binding
ligands on the C terminus of the fiber gene. Sequencing of viral
DNA confirmed that AdZ.F(RGD) and AdZ.F(pK7) encoded
the correct insert sequence. To confirm the increased size of
the modified fiber proteins on the virus, equal numbers of
CsCl-purified AdZ, AdZ.F(RGD), and AdZ.F(pK7) particles
were loaded on an SDS gel and evaluated by immunoblot
analysis (Fig. 3). The proteins were probed with a combination
of antibodies directed against whole Ad5 and against fiber.
The combination of antibodies detected the major coat pro-
teins, hexon, penton base, and fiber. The mobilities of the
hexon and the penton proteins were the same in AdZ.F(RGD)
and AdZ.F(pK7) as in AdZ. However, the decreased mobility
of the fiber proteins of AdZ.F(RGD) and AdZ.F(pK7) relative
to AdZ were consistent with AdZ.F(RGD) and AdZ.F(pK7)
containing the appropriate additions. Immunoblot analysis us-
ing an antifiber polyclonal antibody confirmed that the shifted
bands were fiber (data not shown).
Characterization of vector growth kinetics. Virus growth
kinetics of the AdZ.F(RGD), AdZ.F(pK7), and AdZ vectors
were assessed to determine whether the ligand additions to the
fiber protein had any adverse effects on active virus particle
assembly (Fig. 4). 293 cells were infected at an MOI of 5 active
FIG. 2. Schematic of the E4 rescue vector construction method used to
create Ad vectors containing modified fiber proteins. A base transfer plasmid
comprising the Ad5 sequence from the NdeI site (map unit 83) to the SalI site
(map unit 100) and a chimeric fiber gene encoding the fiber, a linker sequence,
and a targeting sequence was transfected into 293 cells by using calcium phos-
phate. The cells were infected at a low MOI with the E4-deleted construct
AdZ.11A (AdZ.DE4). Homologous recombination produced a vector containing
the E4 region plus the modified fiber gene. Recombinant E1
2
E4
1
vectors were
isolated by plaquing on 293 cells.
FIG. 3. Immunoblot analysis showing the mobility shift of chimeric fiber
proteins compared to wild-type fiber. Equal particle numbers of all vectors (2 3
10
10
particles/lane) were diluted in Laemmli running buffer, boiled, and run on
a 0.1% SDS–9% acrylamide minigel. The proteins were transferred to nitrocel-
lulose and detected by using a rabbit polyclonal antibody to denatured Ad
particles. Lane 1, AdZ; lane 2, AdZ.F(pK7); lane 3, AdZ.F(RGD).
FIG. 4. Virus growth kinetics of AdZ, AdZ.F(pK7), and AdZ.F(RGD) in
293 cells. On day 0, 1 million 293 cells were infected with AdZ, AdZ.F(pK7), or
AdZ.F(RGD) at an MOI of 5 FFU/cell. At 1, 2, or 3 days postinfection, the cells
were harvested and freeze-thawed three times. The number of active particles
(FFU) produced per cell was determined.
8224 WICKHAM ET AL. J. VIROL.
virus particles/cell with either AdZ, AdZ.F(RGD), or AdZ.F
(pK7), and the number of infectious particles (FFU) produced
per cell was determined at 1, 2, and 3 days postinfection. The peak
titers of AdZ.F(RGD) and AdZ.F(pK7) were 80 and 56%, re-
spectively, of that of AdZ. These results indicate that the growth
kinetics of the two modified vectors were not dramatically af-
fected by the addition of ligands onto the end of the fiber protein.
Gene delivery by AdZ, AdZ.F(pK7), or AdZ.F(RGD) to mul-
tiple cell types. The relative transduction efficiencies of the
modified vectors were compared on different cell types (Fig.
5). Increasing total particle doses of AdZ, AdZ.F(pK7), or
AdZ.F(RGD) were incubated with cells, and the resulting b-
galactosidase gene expression was determined 2 days later.
A549 cells were used as a control to verify that each vector
equivalently transduced cells expressing high levels of the fiber
receptor. All three vectors equally transduced A549 cells, in-
dicating that the activity of the vector particles was not nega-
tively affected by the addition of ligands onto the end of the
fiber protein (Fig. 5). At the highest dose of vector, corre-
sponding to approximately 10,000 virus particles/cell, the trans-
duction did not linearly increase with vector concentration.
This corresponding decrease in transduction efficiency was as-
sociated with an apparent cytotoxic effect caused by all three of
the vectors not observed at the lower doses. This cytotoxic
effect was noted despite each vector containing less than 1
replication-competent Ad in 10
8
active particles (not shown).
When the three vectors were tested on the other cell lines,
the results were dramatically different. AdZ.F(pK7) and
AdZ.F(RGD) transduced the endothelial cell line CPAE
approximately 100-fold more efficiently than AdZ. These
results indicate that CPAE cells lack high levels of fiber
receptor but do express sufficient levels of the receptors
recognized by the polylysine and RGD-containing ligands
present in AdZ.F(pK7) and AdZ.F(RGD), respectively.
The transduction levels achieved by the three vectors were
all different on primary smooth muscle cells. The AdZ.F(pK7)
was 650- and 170-fold more efficient than AdZ in transducing
the smooth muscle cells at approximately 100 and 1,000 parti-
cles per cell, respectively. The transduction by AdZ.F(RGD)
was between that by AdZ and AdZ.F(pK7) and averaged sev-
enfold higher than that by AdZ at 100, 1,000, and 10,000
particles per cell. These results suggest that the level of recep-
tor expression plays a role in the level of transduction that is
observed.
AdZ.F(pK7), but not AdZ.F(RGD), transduced human pri-
mary macrophages over 100-fold more efficiently than AdZ at
the two highest doses of vector. The lack of increase by the
AdZ.F(RGD) vector indicates that macrophages lack high lev-
els of a
v
integrins. This result is not unexpected, as macro-
phages isolated from peripheral blood do not normally express
high levels of a
v
integrins, except in response to GM-CSF and
G-CSF (9). This assumption seemed to be correct, as culturing
human macrophages in a combination of GM-CSF and G-CSF
increased their transduction by AdZ.F(RGD) 4.3-fold over
that by AdZ. In addition, activation of the macrophage-like
cell lines THP-1 and U-937 by phorbol myristate acetate, which
has been shown to upregulate a
v
integrin expression in these
cells, likewise increased transduction by AdZ.F(RGD) relative
to AdZ 1.6- and 2.0-fold, respectively (data not shown).
The transduction of resting peripheral blood T lymphocytes,
fibroblasts (Hs68), and glioblastoma cells (U-118 and A172)
with 100 particles per cell was also evaluated (data not shown).
Relative to the AdZ vector, AdZ.F(RGD) increased transduc-
tion of the Hs68 cells 6.6-fold but did not increase the trans-
duction of the glioblastoma cells. The AdZ.F(pK7) vector in-
creased the transduction of T, Hs68, A172, and U-118 cells 40-,
109-, 5.5, and 4.2-fold, respectively.
Gene delivery by the other targeted vectors to multiple cell
types. The four other vectors that could be made were also
evaluated for binding to their target molecules (Table 1). Im-
munofluorescence studies showed that AdZ.F(FLAG)-infect-
ed, but not AdZ-infected, 293 cells were recognized by the
anti-FLAG MAb M2, similar to results of previous studies with
an Ad vector containing the FLAG peptide in the penton base
(49). Furthermore, the FLAG peptide expressed on the fiber
C terminus was also able to mediate direct binding of AdZ.F
(FLAG) to a
v
integrins in assays using a bispecific antibody
with specificities for the FLAG peptide and for a
v
integrins,
similar to previous results with the penton-modified vector
(data not shown) (49).
Two other vectors, AdZ.F(GRGDTF)
3
and AdZ.F
(GYIGSR)
3
, each containing C-terminal additions of 32 amino
acids, could be efficiently grown to high titer. However, neither
vector appeared to recognize the targeted receptor with high
enough affinity to mediate increased virus binding and trans-
duction (Table 1). AdZ.F(GRGDTF)
3
, which contains three
lower-affinity RGD motifs in tandem at the end of each fiber
monomer, did not increase delivery to a
v
integrin-expressing
cells relative to the unmodified vector, AdZ. This result was in
contrast to that for the vector AdZ.F(RGD), which contains a
single high-affinity RGD motif and which increased the trans-
duction of a
v
integrin-expressing endothelial cells (Fig. 5).
Similarly, the vector expressing three tandem YIGSR motifs,
which have been shown to interact with the high-affinity lami-
nin receptor, did not increase the transduction of B16-F1 mel-
anoma cells or other cell lines known to express high levels of
the high-affinity laminin receptor. These results suggest that a
critical receptor-ligand affinity is required to achieve transduc-
tion via the targeted receptor.
FIG. 5. Comparison of transduction of different cell lines by AdZ, AdZ.F
(pK7), and AdZ.F(RGD). The indicated cells (0.5 310
6
to 1.0 310
6
per 6-cm-
diameter plate) were seeded 1 day prior to incubation with increasing concen-
trations of virus in a total volume of 0.2 ml for1hat37°C. The cells were washed
three times with DMEM and cultured for 2 to 3 days in DMEM–10% fetal
bovine serum. The cells were lysed in 1.0 ml of reporter lysis buffer, and b-
galactosidase activity was determined by using a chemiluminescence assay. En-
zyme activity is reported in relative light units (RLU). Reported RLU represent
the average of duplicate measurements. (A) A549 alveolar epithelial cells; (B)
CPAE endothelial cells; (C) human primary smooth muscle cells; (D) human
peripheral blood macrophages.
VOL. 71, 1997 TARGETING OF ADENOVIRUS TO CELLS 8225
AdZ.F(pK7) binding to cells expressing high or low levels of
fiber receptor. Virus binding assays were performed to deter-
mine whether the increases in transduction observed on cer-
tain cell lines using AdZ.F(pK7) and AdZ.F(RGD) were due
to increased binding mediated by heparan sulfate proteogly-
cans and a
v
integrins, respectively (Tables 2 and 3). The bind-
ing properties of radiolabeled AdZ.F(RGD), AdZ.F(pK7),
and AdZ were evaluated on cells expressing high levels of fiber
receptor and a
v
integrins (a
v
-293 cells), cells expressing very
low levels of fiber receptor (CPAE endothelial cells), and cells
expressing low but detectable levels of Ad fiber receptor (A-10
cells). For AdZ.F(pK7), the specificity of binding to the fiber
receptor or to heparan sulfate-containing receptors was as-
sessed through competition with soluble recombinant fiber
protein, soluble heparin, or a combination of fiber and heparin
(Table 2). The inhibition of AdZ.F(pK7) binding by these
competitors was measured relative to their inhibition of AdZ
binding. Overall, the binding to 293 cells by AdZ.F(pK7) com-
pared to AdZ was comparable in the absence of inhibitors,
although AdZ.F(pK7) binding was approximately double that
of AdZ. Fiber, but not heparin, inhibited the binding of AdZ.
However, fiber had only a marginal effect on AdZ.F(pK7)
binding, whereas heparin alone reduced binding by over two-
thirds. Only the combination of heparin plus fiber was able to
completely abrogate the binding of AdZ.F(pK7) to 293 cells.
These results strongly suggest that AdZ.F(pK7) binds to cells
via two interactions: the fiber receptor interaction blocked by
soluble fiber and a second class of receptors whose interaction
is blocked by soluble heparin.
CPAE cells bound 10-fold-higher levels of AdZ.F(pK7) than
AdZ. Soluble fiber had no significant effect on the binding of
AdZ, confirming that CPAE cells lack detectable levels of
functional fiber receptor. These results demonstrate that the
polylysine modification in AdZ.F(pK7) significantly increases
its binding to cells that express low or undetectable levels of
functional fiber receptor. Furthermore, the increased binding
is blocked by heparin, indicating that the polylysine addition to
the fiber mediates Ad binding to a heparin- or heparan-con-
taining cellular receptor. These results demonstrate that the
increased transduction of the fiber receptor-deficient cells by
AdZ.F(pK7) is facilitated through the increased binding of the
virus via its polylysine-modified fiber protein.
AdZ.F(RGD) binding properties to cells expressing high or
low levels of fiber receptor. AdZ.F(RGD) specificity of binding
to the fiber receptor or to a
v
integrins was assessed through
competition with soluble recombinant fiber protein, soluble
penton base, or a combination of fiber and penton base (Table
3). Penton base was used as a competitor since it has been
shown to block a
v
integrin interaction via its RGD motif. Like
AdZ.F(pK7), the inhibition of AdZ.F(RGD) binding by these
competitors was determined relative to their inhibition of AdZ
binding.
While fiber blocked binding of AdZ to a
v
-293 cells, it had no
significant effect on the binding of AdZ.F(RGD) to these cells.
This result indicated that AdZ.F(RGD) bound to cells via an
additional interaction. Penton base had no significant effect on
AdZ.F(RGD) binding; however, the combination of fiber
plus penton base significantly blocked AdZ.F(RGD) bind-
ing. Fiber had no effect on AdZ.F(RGD) binding to CPAE
cells, whereas penton base significantly blocked binding close to
the levels observed for AdZ. These results demonstrate that
AdZ.F(RGD), unlike AdZ, can bind directly to a
v
integrins on
cells expressing or lacking fiber receptor expression. In addi-
tion, compared to AdZ, the additional interaction of AdZ.F
(RGD) with a
v
integrins significantly increases virus binding to
cells which express little or no detectable fiber receptor. These
results demonstrate that the increased transduction of the en-
dothelial and smooth muscle cells by AdZ.F(RGD) is due to
increased binding of the virus to a
v
integrins which are ex-
pressed on these cells.
Increased in vivo transduction by AdZ.F(pK7). The greatly
increased transduction efficiency of smooth muscle cells in tis-
sue culture using the AdZ.F(pK7) vector suggested that it should
increase LacZ gene delivery to vascular smooth muscle cells in
vivo. To investigate this, the left and right pig iliac arteries were
subjected to balloon catheter injury and then transduced for 30
min with either AdZ.F(pK7) or AdZ (Fig. 6). After transduc-
tion, blood flow was reestablished in the arteries; after 3 days,
the arteries were removed and stained for b-galactosidase ex-
pression. Arteries transduced with AdZ.F(pK7) (Fig. 6C and D)
showed increased expression compared to the corresponding
arteries transduced with AdZ (Fig. 6A and B). These results
demonstrate that the AdZ.F(pK7) vector transduces smooth
muscle cells, in vivo, with higher efficiency than the unmodified
vector, AdZ.
DISCUSSION
Depressed fiber receptor expression has been shown to re-
duce gene transfer by decreasing viral adsorption. To over-
come this problem, we have shown that high-titer stocks of
modified Ad particles which increase adsorption and gene
transfer efficiency to cells both in vitro and in vivo can be made.
While Ad has been shown to transduce a number of tissues and
cell types, this study and previous studies have identified mul-
tiple cell types which express low levels of fiber receptor, in-
TABLE 3. Comparison of AdZ and AdZ.F(RGD)
binding to three cell lines
a
Protein
Bound counts
b
a
v
-293 CPAE A-10
AdZ AdZ.F(RGD) AdZ AdZ.F(RGD) AdZ AdZ.F(RGD)
Control 7.6 12.7 0.19 0.84 0.72 1.68
Fiber 1.7 12.3 0.22 1.06 0.23 1.40
Penton base 9.0 9.7 0.20 0.37 0.80 0.62
Fiber/penton
base 1.0 3.7 0.21 0.46 0.20 0.41
a
Confluent cell monolayers in collagen-coated 24-well plates were preincu-
bated with the indicated competitors for1hat4°C.
3
H-labeled AdZ or AdZ.F
(RGD) was then added to cells for 90 min at room temperature.
b
Expressed as a percentage of the radioactive counts added per well. Error for
all values was less than 10%.
TABLE 2. Comparison of AdZ and AdZ.F(pK7)
binding to three cell lines
a
Protein
Bound counts
b
a
v
-293 CPAE A-10
AdZ AdZ.F(pK7) AdZ AdZ.F(pK7) AdZ AdZ.F(pK7)
Control 7.6 18.2 0.19 2.32 0.72 9.90
Fiber 1.7 13.3 0.22 2.07 0.23 8.30
Heparin 9.0 5.1 0.06 0.13 0.53 0.41
Fiber/heparin 0.3 0.9 0.05 0.13 0.10 0.20
a
Confluent cell monolayers in collagen-coated 24-well plates were preincu-
bated with the indicated competitors for1hat4°C.
3
H-labeled AdZ or AdZ.F
(pK7) was then added to cells for 90 min at room temperature.
b
Expressed as a percentage of the radioactive counts added per well. Error for
all values was less than 10%.
8226 WICKHAM ET AL. J. VIROL.
cluding endothelium, smooth muscle, melanoma, glioblas-
toma, primary bronchial epithelium, macrophages, and T cells
(8, 24, 25, 40, 49). Therefore, modified Ad vectors with in-
creased adsorption and transduction efficiency could have a
profound impact upon the diseases treatable by gene therapy.
Vascular smooth muscle cells are a primary target for gene
therapy to prevent aberrant smooth muscle cell proliferation
(6, 21, 44). Smooth muscle cell proliferation has been impli-
cated in the pathogenesis of atherosclerosis and in clinically
significant restenosis in 30 to 40% of patients following balloon
angioplasty of the coronary arteries (34). High levels of Ad
particles have typically been required to achieve significant
transduction of the smooth muscle. Our results demonstrate
that smooth muscle transduction by Ad vectors can be dramat-
ically increased by using the modified polylysine vector AdZ.F
(pK7). Decreased doses of therapeutic Ad in smooth muscle
gene therapies are likely to decrease side effects often associ-
ated with Ad gene therapy.
Ad coat proteins can be modified to redirect the virus to
ubiquitously expressed receptors which then mediate the effi-
cient binding and transduction by Ad to a broad spectrum of
tissues. Polylysine has been shown to interact with heparan
sulfate (13) as well as other polyanions such as DNA (8).
Because heparan sulfate and other polyanionic molecules such
as chondroitin sulfate and mucins are broadly expressed on
cells, the pK7 vector is universal in the cell types that it can
transduce. Thus, in terms of its cell tropism the pK7 vector is
actually nontargeted because the receptor targeted by the virus
is broadly expressed. The significant increases in binding and
transduction of cells lacking high fiber receptor expression
levels shown here demonstrate that the addition of peptide
receptor-binding motifs into the coat proteins of Ad can ex-
pand the range of tissues amenable to efficient adenovirus-
mediated gene therapy.
A vector similar to AdZ.F(pK7) has previously been con-
structed by incorporating a frameshift mutation into the fiber
stop codon, which resulted in polylysine addition to the fiber C
terminus (48). The absence of an in-frame stop codon permits
translation to proceed into the poly(A) tail of the message
which encodes polylysine. However, the frameshift mutation
also significantly depresses fiber protein synthesis which is as-
sociated with a 100-fold decrease in virus titer (unpublished
results). Here we have shown that the addition of a defined
peptide linker and polylysine sequence onto the C terminus
results in both efficient transduction via the polylysine and
relatively unaffected vector growth properties.
Coat protein modifications can also be made to direct the
virus to tissue-specific receptors. Such modifications then allow
the virus to efficiently bind and transduce only select tissues.
The a
v
integrins are promising receptors for targeted gene
transfer. While many cells in the body are capable of express-
ing a
v
integrins, they are often highly expressed in response to
only certain cytokines and growth factors which are produced
during infection, wounding, or inflammation. This characteris-
tic of a
v
integrins makes them an ideal receptor for targeting
injured vasculature following angioplasty or for targeting pro-
FIG. 6. Comparison of in vivo gene transfer efficiencies in assays using AdZ and AdZ.F(pK7). Pig iliac arteries were subjected to balloon catheter injury and then
transduced for 30 min with 2 310
10
particles of either AdZ or AdZ.F(pK7) per ml. After transduction, viral solutions were evacuated and blood flow was reestablished
through the iliac arteries. Animals were euthanized on postoperative day 3, and the iliac arteries were fixed and stained with X-Gal. (A and B) AdZ-treated vessels;
(C and D) AdZ.F(pK7)-treated vessels. Panels A and C are en face images (magnification, 34) of the lumenal surface of the arteries, while panels B and D are
cross-sections (magnification, 310) of the stained vessels. Blue-staining cells represent positively transduced cells.
VOL. 71, 1997 TARGETING OF ADENOVIRUS TO CELLS 8227
liferating endothelial cells in tumors (4, 12). Bispecific anti-
bodies have been previously shown to target adenovirus to a
v
integrin-expressing cells (49). However, drawbacks to this ap-
proach include the additional production and characterization
of the bispecific antibodies, the potential clearance of virus by
the Fc receptor, and the potential activation of the comple-
ment system. By expressing the high-affinity RGD ligand as a
fiber fusion protein, it is possible to produce comparable titers
of the targeted vector in standard packaging cell lines without
the drawbacks associated with bispecific antibodies.
The affinity of the C-terminal peptide motif for its receptors
appears to be a critical factor in directing significant virus
binding to cells via the targeted receptor. AdZ.F(RGD), which
incorporates multiple, linear RGD motifs, did not mediate
detectable binding to cells via a
v
integrins. We have similarly
found that incorporating multiple YIGSR motifs into the fiber
C terminus does not increase binding to cells expressing the
high-affinity laminin receptor. The ACDCRGDCFCG motif,
that was incorporated into the AdZ.F(RGD) vector, was pre-
viously identified by using phage display technology (28). This
motif has an approximately 100-fold-higher affinity than other
RGD motifs and is believed to form a tight loop through a pair
of disulfide bonds between the cysteines. Therefore, although
several receptor-binding peptide motifs have been identified,
the low affinities of many of these motifs suggest that few will
mediate significant increases in adenovirus binding to cells
expressing the targeted receptor. Furthermore, vector binding
to cells does not necessarily indicate that the vector will effi-
ciently internalize into the cells.
The E4 rescue technique that we used to produce fiber-
modified viruses was much more efficient and reliable than
other vector construction methods, and contaminating virus
background was virtually absent. This system also allowed blue/
white selection of the appropriate plaques by using the X-Glu
substrate. This selection is possible because the AdZ.11A virus
expresses b-glucuronidase from the E4-deleted region (blue
plaques), while the E4-wild type recombinant does not contain
the b-glucuronidase gene (white plaques). Using some peptide
motifs in this system, we found that we could not isolate viable
recombinant vectors. The most likely explanation for this is
that certain peptide motifs might prevent proper folding of the
fiber protein or interfere with trimerization.
While the AdZ.F(RGD) vector is not truly targeted to a
v
integrins because it can still bind to the fiber receptor, the
increased transduction by Ad vectors with expanded receptor-
binding repertoires demonstrates the feasibility of targeting
vectors to specific cell types via tissue-specific receptors. In
fact, the potential of tissue-specific targeting using high-affinity
peptides has recently been demonstrated. Phage displaying the
same high-affinity RGD motif as in AdZ.F(RGD) were found
to home to tumor endothelial cells when injected intravenously
into tumor-bearing mice (38). Therefore, the efficient targeting
of a single cell type or tissue will necessitate Ad vectors which
lack fiber receptor-binding activity. Such vectors may likewise
require special receptor-expressing cell lines in order to prop-
agate them. In any case, it will be important to determine the
levels of Ad receptors in tissues targeted for gene therapy. If
fiber receptor expression is low in a given tissue, targeting Ad
to receptors that are expressed by the tissue is likely to increase
the efficiency and specificity of gene transfer.
ACKNOWLEDGMENTS
We thank Faith Beams, Vicki Kulesa, and Angela Appiah for ex-
cellent technical assistance.
REFERENCES
1. Albelda, S. M., S. A. Mette, D. E. Elder, R. Stewart, L. Damjanovich, M.
Herlyn, and C. A. Buck. 1990. Integrin distribution in malignant melanoma:
association of the b3 subunit with tumor progression. Cancer Res. 50:6757–
6764.
2. Bai, M., B. Harfe, and P. Freimuth. 1993. Mutations that alter an Arg-Gly-
Asp (RGD) sequence in the adenovirus type 2 penton base protein abolish
its cell-rounding activity and delay virus reproduction in flat cells. J. Virol.
67:5198–5205.
3. Bergelson, J. M., J. A. Cunningham, G. Droguett, E. A. Kurt-Jones, A.
Krithivas, J. S. Hong, M. S. Horwitz, R. L. Crowell, and R. W. Finberg. 1997.
Isolation of a common receptor for coxsackie B viruses and adenoviruses 2
and 5. Science 275:1320–1323.
4. Brooks, P. C., R. A. F. Clark, and D. A. Cheresh. 1994. Requirement of
vascular integrin avb3 for angiogenesis. Science 264:569–571.
5. Brough, D. E., A. Lizonova, C. Hsu, V. A. Kulesa, and I. Kovesdi. 1996. A
gene transfer vector-cell line system for complete functional complementa-
tion of adenovirus early regions E1 and E4. J. Virol. 70:6497–6501.
6. Chang, M. W., E. Barr, J. Seltzer, Y. Q. Jiang, G. J. Nabel, E. G. Nabel, M. S.
Parmacek, and J. M. Leiden. 1995. Cytostatic gene therapy for vascular
proliferative disorders with a constitutively active form of the retinoblastoma
gene product. Science 267:518–522.
7. Crystal, R. G., N. G. McElvaney, M. A. Rosenfeld, C. S. Chu, A. Mastrangeli,
J. G. Hay, S. L. Brody, H. A. Jaffe, N. T. Eissa, et al. 1994. Administration of
an adenovirus containing the human CFTR cDNA to the respiratory tract of
individuals with cystic fibrosis. Nat. Genet. 8:42–51.
8. Curiel, D. T., E. Wagner, M. Cotten, M. L. Birnstiel, S. Agarwal, C. Li, S.
Loechel, and P. Hu. 1992. High-efficiency gene transfer mediated by adeno-
virus coupled to DNA-polylysine complexes. Hum. Gene Ther. 3:147–154.
9. De Nichilo, M. O., and G. F. Burns. 1993. Granulocyte-macrophage and
macrophage colony-stimulating factors differentially regulate av integrin ex-
pression on cultured human macrophages. Proc. Natl. Acad. Sci. USA 90:
2517–2521.
10. Douglas, J. T., B. E. Rogers, M. E. Rosenfeld, S. I. Michael, M. Feng, and
D. T. Curiel. 1996. Targeted gene delivery by tropism-modified adenoviral
vectors. Nat. Biotechnol. 14:1574–1578.
11. French, B., W. Mazur, N. Ali, R. Geske, J. Finnigan, G. Rodgers, R. Roberts,
and A. Raizner. 1994. Percutaneous transluminal in vivo gene transfer by
recombinant adenovirus in normal porcine coronary arteries, atherosclerotic
arteries, and two models of coronary restenosis. Circulation 90:2402–2413.
12. Friedlander, M., P. C. Brooks, R. W. Shaffer, C. M. Kincaid, J. A. Varner,
and D. A. Cheresh. 1995. Definition of two angiogenic pathways by distinct
a
v
integrins. Science 270:1500–1502.
13. Fromm, J. R., R. E. Hileman, E. E. Caldwell, J. M. Weiler, and R. J.
Linhardt. 1995. Differences in the interaction of heparin with arginine and
lysine and the importance of these basic amino acids in the binding of
heparin to acidic fibroblast growth factor. Arch. Biochem. Biophys. 323:279–
287.
14. Gall, J., A. Kass-Eisler, L. Leinwand, and E. Falck-Pedersen. 1996. Adeno-
virus type 5 and 7 capsid chimera: fiber replacement alters receptor tropism
without affecting primary immune neutralization epitopes. J. Virol. 70:2116–
2123.
15. Garrido, M. A., P. Perez, J. A. Titus, M. J. Valdayo, D. A. Winkler, S. A.
Barbieri, J. R. Wunderlich, and D. M. Segal. 1990. Targeted cytotoxic cells
in human peripheral blood lymphocytes. J. Immunol. 144:2891–2898.
16. Gladson, C. L., and D. A. Cheresh. 1991. Glioblastoma expression of vitro-
nectin and the avb3 integrin: adhesion mechanism for transformed glial
cells. J. Clin. Invest. 88:1924.
17. Gladson, C. L., and D. A. Cheresh. 1994. The av integrins, p. 83–99. In Y.
Takada (ed.), Integrins: the biological problems. CRC Press, Boca Raton,
Fla.
18. Graf, J., U. Iwamoto, M. Sasaki, G. R. Martin, H. K. Kleinman, F. A. Robey,
and Y. Yamada. 1987. Identification of an amino acid sequence in laminin
mediating cell attachment, chemotaxis, and receptor binding. Cell 48:989–
996.
19. Graham, F. L., J. Smiley, W. C. Russell, and R. Nairu. 1977. Characteristics
of a human cell line transformed by DNA from human adenovirus type 5.
J. Gen. Virol. 36:59–77.
20. Graham, F. L., and A. J. van der Eb. 1973. A new technique for the assay of
infectivity of human adenovirus 5 DNA. Virology 52:456–467.
21. Guzman, R. J., E. A. Hirschowitz, S. L. Brody, R. G. Crystal, S. E. Epstein,
and T. Finkel. 1994. In vivo suppression of injury-induced vascular smooth
muscle cell accumulation using adenovirus-mediated transfer of the herpes
simplex virus thymidine kinase gene. Proc. Natl. Acad. Sci. USA 91:10732–
10736.
22. Hirt, B. 1967. Selective extraction of polyoma DNA from infected mouse cell
cultures. J. Mol. Biol. 26:365–369.
23. Hong, S. S., L. Karayan, J. Tournier, D. T. Curiel, and P. A. Boulanger. 1997.
Adenovirus type 5 fiber knob binds to MHC class I a2 domain at the surface
of human epithelial and B lymphoblastoid cells. EMBO J. 16:2294–2306.
24. Huang, S., R. I. Endo, and G. R. Nemerow. 1995. Upregulation of integrins
a
v
b
3
and a
v
b
5
on human monocytes and T lymphocytes facilitates adenovi-
8228 WICKHAM ET AL. J. VIROL.
rus-mediated gene delivery. J. Virol. 69:2257–2263.
25. Huang, S., T. Kamata, Y. Takada, Z. M. Ruggeri, and G. R. Nemerow. 1996.
Adenovirus interaction with distinct integrins mediates separate events in
cell entry and gene delivery to hematopoietic cells. J. Virol. 70:4502–4508.
26. Jalkanen, M., S. Jalkanen, and M. Bernfield. 1991. Binding of extracellular
effector molecules by cell surface proteoglycan, p. 1–30. In J. A. McDonald
and R. P. Mecham (ed.), Receptors for extracellular matrix. Academic Press,
Inc., New York, N.Y.
27. Kanemoto, T., R. Reich, L. Royce, D. Greatorex, S. H. Adler, N. Shiraishi,
G. R. Martin, Y. Yamada, and H. K. Kleinman. 1990. Identification of an
amino acid sequence from the laminin A chain that stimulates metastasis and
collagenase IV production. Proc. Natl. Acad. Sci. USA 87:2279–2283.
28. Koivunen, E., B. Wang, and E. Ruoslahti. 1995. Phage libraries displaying
cyclic peptides with different ring sizes: ligand specificities of the RGD-
directed integrins. Bio/Technology 13:265–270.
29. Krasnykh, V. N., G. V. Mikheeva, J. T. Douglas, and D. T. Curiel. 1996.
Generation of recombinant adenovirus vectors with modified fibers for al-
tering viral tropism. J. Virol. 70:6839–6846.
30. Lawrence, W. C., and H. S. Ginsberg. 1967. Intracellular uncoating of type 5
adenovirus deoxyribonucleic acid. J. Virol. 1:851–867.
31. Le Gal La Salle, G., J. J. Robert, S. Berrard, V. Ridoux, L. D. Stratford-
Perricaudet, M. Perricaudet, and J. Mallet. 1993. An adenovirus vector for
gene transfer into neurons and glia in the brain. Science 259:988–990.
32. Martens, C. L., S. E. Cwirla, R. Y.-W. Lee, E. Whitehorn, E. Y.-F. Chen, A.
Bakker, E. L. Martin, C. Wagstrom, P. Gopalan, et al. 1995. Peptides which
bind to E-selectin and block neutrophil adhesion. J. Biol. Chem. 270:21129–
21136.
33. Mathias, P., T. Wickham, M. Moore, and G. Nemerow. 1994. Multiple
adenovirus serotypes use a
v
integrins for infection. J. Virol. 68:6811–6814.
34. Mazur, W., N. M. Ali, A. E. Raizner, and B. A. French. 1994. Coronary
restenosis and gene therapy. Tex. Heart Inst. J. 21:104–111.
35. McCoy, R. D., B. L. Davidson, B. J. Roessler, G. B. Huffnagle, S. L. Janich,
T. J. Laing, and R. H. Simon. 1995. Pulmonary inflammation induced by
incomplete or inactivated adenoviral particles. Hum. Gene Ther. 6:1553–
1560.
36. Michael, S. I., J. S. Hong, D. T. Curiel, and J. A. Engler. 1995. Addition of
a short peptide ligand to the adenovirus fiber protein. Gene Ther. 2:660–668.
37. Nemerow, G. R., D. A. Cheresh, and T. J. Wickham. 1993. Adenovirus entry
into host cells: a role for av integrins. Trends Cell Biol. 4:52–55.
38. Pasqualini, R., E. Koivunen, and E. Ruoslahti. 1997. av integrins as recep-
tors for tumor targeting by circulating ligands. Nat. Biotechnol. 15:542–546.
39. Phillips, J. H., N. L. Warner, and L. L. Lanier. 1983. Correlation of bio-
physical properties and cell surface antigenic profile of Percoll gradient-
separated human natural killer cells. Nat. Immun. Cell Growth Regul. 3:73–
86.
40. Roelvink, P. W., I. Kovesdi, and T. J. Wickham. 1996. Comparative analysis
of adenovirus fiber-cell interaction: adenovirus type 2 (Ad2) and Ad9 utilize
the same cellular fiber receptor but use different binding strategies for
attachment. J. Virol. 70:7614–7621.
41. Rosenfeld, M. A., K. Yoshimura, B. C. Trapnell, K. Yoneyama, E. R.
Rosenthal, W. Dalemans, M. Fukayama, J. Bargon, L. E. Stier, et al. 1992.
In vivo transfer of the human cystic fibrosis transmembrane conductance
regulator gene to the airway epithelium. Cell 68:143–155.
42. Thiel, J. F., and K. O. Smith. 1967. Fluorescent focus assay of viruses on cell
monolayers in plastic petri plates. Proc. Soc. Exp. Biol. Med. 125:892–895.
43. Tomko, R. P., R. Xu, and L. Philipson. 1997. HCAR and MCAR: the human
and mouse cellular receptors for subgroup C adenoviruses and group B
coxsackieviruses. Proc. Natl. Acad. Sci. USA 94:3352–3356.
44. Tzeng, E., L. L. Shears, P. D. Robbins, B. R. Pitt, D. A. Geller, S. C. Watkins,
R. L. Simmons, and T. R. Billiar. 1996. Vascular gene transfer of the human
inducible nitric oxide synthase: characterization of activity and effects on
myointimal hyperplasia. Mol. Med. 2:211–225.
45. Varga, M. J., C. Weibull, and E. Everitt. 1991. Infectious entry pathway of
adenovirus type 2. J. Virol. 65:6061–6070.
46. Wickham, T. J., M. E. Carrion, and I. Kovesdi. 1995. Targeting of adenovirus
penton base to new receptors through replacement of its RGD motif with
other receptor-specific peptide motifs. Gene Ther. 2:750–756.
47. Wickham, T. J., P. Mathias, D. A. Cheresh, and G. R. Nemerow. 1993.
Integrins avb3 and avb5 promote adenovirus internalization but not virus
attachment. Cell 73:309–319.
48. Wickham, T. J., P. W. Roelvink, D. E. Brough, and I. Kovesdi. 1996. Ade-
novirus targeted to heparan-containing receptors increases its gene delivery
efficiency to multiple cell types. Nat. Biotechnol. 14:1570–1573.
49. Wickham, T. J., D. M. Segal, P. W. Roelvink, M. E. Carrion, A. Lizonova,
G. M. Lee, and I. Kovesdi. 1996. Targeted adenovirus gene transfer to
endothelial and smooth muscle cells by using bispecific antibodies. J. Virol.
70:6831–6838.
VOL. 71, 1997 TARGETING OF ADENOVIRUS TO CELLS 8229
