Content uploaded by Hidde J Haisma
Author content
All content in this area was uploaded by Hidde J Haisma
Content may be subject to copyright.
Gene Therapy (2000) 7, 1940–1946
2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00
www.nature.com/gt
VIRAL TRANSFER TECHNOLOGY RESEARCH ARTICLE
Recombinant adenovirus vectors with knobless fibers
for targeted gene transfer
VW van Beusechem
1
, ALCT van Rijswijk
1
, HHG van Es
2
, HJ Haisma
1
, HM Pinedo
1
and WR Gerritsen
1
1
Division of Gene Therapy, Department of Medical Oncology, University Hospital Vrije Universiteit, Amsterdam;
2
Galapagos
Genomics, Leiden, The Netherlands
Adenoviral vector systems for gene therapy can be much
improved by targeting vectors to specific cell types. This
requires both the complete ablation of native adenovirus tro-
pism and the introduction of a novel binding affinity in the
viral capsid. We reasoned that these requirements could be
fulfilled by deleting the entire knob domain of the adenovirus
fiber protein and replacing it with two distinct moieties that
provide a trimerization function for the knobless fiber and
specific binding to the target cell, respectively. To test this
concept, we constructed adenoviral vectors carrying knob-
less fibers comprising the
␣
-helix trimerization domain from
MoMuLV envelope glycoprotein. Two mimic targeting
Keywords:
gene therapy; targeting; adenovirus vector; fiber; trimerization
Introduction
Recombinant adenoviral vectors are being used as gene
delivery vectors in a variety of gene therapy strategies.
While the efficient transduction of many different human
tissues is on the one hand an important attribute of aden-
oviral vectors, this promiscuous tropism represents, on
the other hand, a limiting feature for their use in gene
therapy. In vivo delivery of adenoviral vectors yields
efficient transduction of cells that may not be a target for
the therapy, most notably liver cells.
1,2
Because of this
vector sequestration by non-target cells, high vector
doses are needed for effective gene delivery into target
cells. This imposes an increased risk for unwanted side-
effects of the gene therapy procedure, by direct toxicity
or host immune responses against the vector. Therefore,
adenoviral vectors could potentially be much improved
if specific gene transfer into only the desired target cells
is accomplished.
The primary high-affinity binding of adenovirus to the
host cell surface is mediated by the knob domain of the
fiber protein.
3,4
This knowledge has directed several stra-
tegies for adenoviral vector targeting by genetic modifi-
cation of the viral capsid. Peptide motifs with receptor
binding specificity have been incorporated into the fiber
protein by extension of the carboxy terminus of the fiber
5–8
Correspondence: VW van Beusechem, Division of Gene Therapy, Depart-
ment of Medical Oncology, University Hospital VU, PO Box 7057, 1007
MB Amsterdam, The Netherlands
Received 9 June 2000; accepted 17 August 2000
ligands, a Myc-epitope and a 6His-tag, were attached via a
flexible linker peptide. The targeted knobless fiber molecules
were properly expressed and imported into the nucleus of
adenovirus packaging cells, where they were incorporated
as functional trimers into the adenovirus capsid. Both ligands
were exposed on the surface of the virion and were available
for specific binding to their target molecules. Moreover, the
knobless fibers mediated gene delivery into cells displaying
receptors for the coupled ligand. Hence, these knobless fib-
ers are prototype substrates for versatile addition of tar-
geting ligands to generate truly targeted adenoviruses.
Gene
Therapy (2000) 7, 1940–1946.
and by insertion into the flexible and accessible HI-loop
of the fiber knob.
9,10
These studies have confirmed the
feasibility of the two approaches to generate adenoviral
vectors with altered binding specificities. Importantly,
Roelvink et al
11
recently ablated the native receptor-bind-
ing domain by mutating the adenovirus type 5 (Ad5)
fiber knob. By combining this mutation with the HI-loop
insertion strategy, they developed a truly targeted adeno-
virus vector. However, it appears that incorporation of
targeting ligands in the fiber knob is not always compat-
ible with proper fiber folding, resulting in loss of fiber
functions.
7,8
Deletions in the fiber knob as small as only
two amino acids can result in loss of fiber trimerization,
the second important function of this domain.
12,13
Since
the trimeric fashion of the fiber is essential for its incor-
poration in the virus particle by binding to the penton
base,
12,14
modifications that result in loss of trimerization
function will prevent formation of viable viruses carrying
the modified protein. Clearly, the development of chim-
eric fibers that impose few demands on the structure of
the ligand while retaining their trimeric quaternary struc-
ture would extend the applicability of targeted adeno-
viruses in gene therapy.
Our approach to develop truly targeted adenoviruses
is to delete the complete fiber knob and to replace it with
two distinct protein moieties. The first moiety is a com-
mon oligomerization motif found in many proteins, the
␣-helical coiled-coil,
15
which serves to substitute for the
fiber knob trimerization function. The second moiety is
the target ligand that is coupled to the trimerization
domain via a flexible linker peptide, such as to allow the
Adenoviruses with knobless fibers
VW van Beusechem
et al
1941
two moieties to adopt their functional conformations. To
test this concept, we constructed a first set of knobless
fiber mutants using the ␣-helix domain from the trans-
membrane subunit p15E of Moloney murine leukemia
virus (MoMuLV) envelope glycoprotein
16
and two mimic
targeting ligands, a Myc-epitope and 6His-tag. One of the
chimeric proteins described here exhibited all the func-
tions required for its use in targeted gene delivery. It was
properly incorporated into the adenovirus capsid where
its carboxy-terminal 6His- and Myc-epitopes were access-
ible for specific binding to nickel ions and immobilized
anti-Myc antibodies, respectively. Moreover, these knob-
less fibers mediated gene transfer into cells displaying
artificial receptors for His-tagged adenoviruses. Hence,
the novel knobless fiber molecules described here are
prototype substrates for incorporation of cell type-
specific targeting ligands to generate truly targeted
adenoviruses.
Results
Construction of knobless fiber genes
Two chimeric genes encoding the entire Ad5 fiber tail
and shaft domains and a trimerization domain derived
from the MoMuLV envelope glycoprotein were made
and were designated TSC and TSFLC, respectively
(Figure 1). In both molecules, Ad5 fiber sequences are
included that encode Met-1 to Thr-403, where Thr-403 is
the last residue of the highly conserved TLWT motif that
delineates the start of the fiber knob. The trimerization
domain in both genes covers the 33-residue trimeric
coiled-coil from Asp-515 to Leu-547 in the MoMuLV
envelope glycoprotein.
16
This domain has no known
binding function of its own. In TSC, the fiber shaft and
trimerization domains are separated by the sequence Gly-
Ser-Gly, in TSFLC these domains are linked via a classical
(Gly
4
Ser)
3
linker commonly used in single-chain anti-
bodies. Thus, TSC and TSFLC represent fusion proteins
with a linkage between the fiber-shaft and ␣-helix
domains that allows minimal or maximal folding free-
dom, respectively. TSC and TSFLC each have a carboxy-
terminal (Gly
4
Ser)
2
flexible linker extension with a unique
BamHI restriction site to allow targeting ligand addition.
To test these knobless fibers for properties relevant to
their use in genetically targeted adenovirus we added a
carboxy-terminal Myc-epitope and 6His-tag via the
BamHI site, producing TSCmychis and TSFLCmychis,
respectively.
Figure 1 Structure of knobless fiber mutants TSCmychis and
TSFLCmychis. T, Ad5 fiber tail-domain; S, Ad5 fiber shaft-domain; C,
MoMuLV p15E coiled-coil domain; Myc-His, Myc/6His-tag from
pcDNA3.1(−)mychis-A. The protein length is given in number of amino
acids and the predicted molecular weight in kilodalton. Relevant amino
acids and restriction sites are indicated. Numbers refer to the correspond-
ing amino acids in the parental proteins.
Gene Therapy
Nuclear import of knobless fibers in mammalian cells
Expression of TSCmychis and TSFLCmychis knobless
fiber proteins was initially evaluated by transient trans-
fection of eukaryotic expression plasmids in 911 packag-
ing cells.
17
This allowed the analysis of nuclear import in
E1-complementing cells in the absence of a cytopathic
effect (CPE). As a control, a vector expressing the bac-
terial LacZ gene with C-terminal Myc/6His-peptide was
used. Twenty-four hours after transfection, Myc-epitope
containing proteins could be detected by immunocyto-
chemistry allowing analysis of the intracellular localiz-
ation of the chimeric proteins (Figure 2). As was
expected, the control protein LacZmychis was detected
predominantly in the cytoplasm of transfected cells. In
contrast, TSCmychis and TSFLCmychis proteins accumu-
lated in the cell nuclei, where adenovirus capsids
assemble. Thus, the nuclear localization signal in the Ad5
fiber tail is functionally intact and correctly targets the
knobless fiber molecules with their carboxy-terminal pep-
tide-ligands to the cell nucleus.
Trimerization of knobless fibers and accessibility of the
C-terminal tag for specific binding
To produce knobless fiber-expressing adenovirus vectors
on conventional 293 packaging cells
18
we cloned
expression cassettes for TSCmychis and TSFLCmychis in
the E1-region of an adenovirus vector also carrying an
expression cassette for enhanced green fluorescent pro-
tein (GFP). The resulting vectors AdGFP-TSCmychis and
AdGFP-TSFLCmychis co-express the knobless fiber vari-
ants with wild-type fibers. As a negative control, AdGFP
virus with GFP as the only E1-insert was produced.
Protein lysates were prepared from adenovirus vector-
infected 293 cells and subjected to Western analysis using
anti-Myc and anti-fiber knob antibodies (Figure 3a). This
showed that TSCmychis and TSFLCmychis knobless fib-
ers were expressed, albeit at much lower levels than the
wild-type fiber. The oligomeric structure of the knobless
fiber molecules was assessed by comparing their electro-
phoretic mobility under semi-native versus denaturing
conditions. As can be seen in Figure 3a, both TSCmychis
and TSFLCmychis exhibited the expected molecular
weight of approximately 50 kDa. Under semi-native con-
Figure 2 Nuclear import of knobless fiber proteins in 911 packaging cells.
Cells transfected with CMV-driven expression constructs were stained
with anti-Myc MoAb after 24 h. (a) mock transfection (empty
pcDNA3.1(−)mychis-A vector); (b) pcDNA3.1(−)mychisLacZ; (c)
pCMVtpl-TSCmychis; (d) pCMVtpl-TSFLCmychis.
Adenoviruses with knobless fibers
VW van Beusechem
et al
1942
Gene Therapy
ditions, approximately 5–10% was found as oligomers, as
estimated by comparing band intensities of the high mol-
ecular weight and monomeric protein species. This
shows that the MoMuLV trimerization domain is func-
tional in both knobless fiber variants. The apparent mol-
ecular weight of the knobless fiber oligomers was larger
than expected for homotrimers. However, analysis of the
same samples with an antibody recognizing the trimeric
wild-type fiber showed that most of this protein also
migrated at a larger apparent molecular weight
(Figure 3a). This is a well-described phenomenon that can
be explained by partial unfolding of the fiber tail and
shaft under laboratory conditions.
19
We next investigated if the 6His-targeting ligand at the
C-terminus of the knobless fiber molecules was accessible
for binding. To this end, cell lysates were incubated with
nickel-nitrilotriacetic acid (Ni-NTA) metal-affinity matr-
ices. Unbound material was washed away and specifi-
cally bound 6His-containing proteins were eluted by
Figure 3 Analysis of knobless fiber proteins expressed in adenovirus vec-
tor-infected packaging cells. Cell lysates were subjected to Western analy-
sis using an anti-Myc or anti-fiber knob antibody as indicated. Sizes of
molecular weight markers are shown. (a) Knobless fiber protein oligomer-
ization assessed by semi-native (unboiled) vs denaturing (boiled) SDS-
PAGE. Lanes 1 and 5, uninfected 293 cells; lanes 2, 6 and 9, AdGFP
infected; lanes 3, 7 and 10, AdGFP-TSCmychis infected; lanes 4, 8 and
11, AdGFP-TSFLCmychis infected. (b) Ni-NTA binding of knobless fiber
proteins. Lysates (total) of 293 cells infected with AdGFP-TSCmychis (left
panel) or AdGFP-TSFLCmychis (right panel) were loaded on to Ni-NTA
beads, unbound material was collected and analyzed (unbound), as well
as material eluted at low stringency (30 mmimidazol; wash) and at high
stringency (300 m
m
imidazol; elution).
competition with imidazol. Individual fractions were
subjected to Western analysis. As can be seen in
Figure 3b, TSCmychis and TSFLCmychis proteins bound
to Ni-NTA matrices, confirming that their C-terminal
ligand is accessible for binding.
Incorporation of knobless fibers in adenovirus capsids
To investigate if the knobless fiber molecules are incor-
porated in complete adenovirus capsids, high-titer virus
stocks of vectors AdGFP, AdGFP-TSCmychis and
AdGFP-TSFLCmychis were purified by CsCl banding
and subjected to Western analysis for wild-type and
knobless fiber variants (Figure 4). As expected, wild-type
fiber trimers were detected on the capsids of all three
viruses and knobless fibers were not seen on AdGFP par-
ticles. TSCmychis molecules were only inefficiently co-
purified with intact adenovirus particles. In contrast, the
knobless fiber protein TSFLCmychis was reproducibly
detected on AdGFP-TSFLCmychis particles.
To corroborate the capsid incorporation of
TSFLCmychis molecules further, we investigated if func-
tional AdGFP-TSFLCmychis viruses could bind with
specificity to mimic receptors for the knobless fiber mol-
ecule. Accessibility of the 6His-tag on the virus capsid
was tested by binding the virus to Ni-NTA beads. Elution
of virus particles at increasing stringency of imidazol
competition was quantified by measuring the functional
GFP-vector titer (Figure 5a). The wild-type fiber-express-
ing AdGFP virus was gradually washed out upon sub-
sequent incubation steps, indicating that as expected it
did not bind to Ni-NTA with specificity. In contrast,
approximately 9.5% and 1.3% of the total recovered
AdGFP-TSFLCmychis virus was found in the 50 mmand
250 mmimidazol elution fractions, respectively. In
addition, full-length viral DNA could be isolated from
the 250 mmfraction and visualized on EtBr-stained aga-
rose gel (not shown). Hence, intact infectious AdGFP-
TSFLCmychis viral particles bound to Ni-NTA with high
affinity via their 6His-tagged knobless fibers.
We also tested functional exposure of the Myc-tag on
the virus capsid. To this end, AdGFP or AdGFP-
TSFLCmychis virus was allowed to bind to immobilized
anti-Myc antibody or to anti-fiber knob antibody as a
control. Target cells were exposed to the bound virus and
GFP-vector titers were measured (Figure 5b). Both
viruses bound efficiently to the anti-fiber knob antibody
Figure 4 Protein analysis of purified adenovirus particles. Approximately
7×10
9
CsCl-purified virus particles were subjected to Western analysis
for wild-type fiber using anti-fiber knob MoAb or for knobless fiber
mutants using anti-Myc MoAb as indicated. A total protein lysate of 293
cells infected with AdGFP-TSCmychis virus was used as a control (cell
lysate). Sizes of molecular weight markers are indicated. One of three
Western analyses performed is shown. TSCmychis molecules were
only detected in this experiment, TSFLCmychis molecules were found
reproducibly.
Adenoviruses with knobless fibers
VW van Beusechem
et al
1943
(functional titers of 2.4 ×10
8
and 2.6 ×10
8
per 1 ×10
9
input virus particles, respectively). Background binding
of the negative control AdGFP virus to the anti-Myc anti-
body was very low (ie only 0.2% of the anti-fiber knob
binding). In contrast, approximately 2.3% functional
AdGFP-TSFLCmychis viruses bound to the anti-Myc
antibody, with 0.6% of these viruses adhering to the
control plates.
Taken together, these findings explicitly demonstrate
that the TSFLCmychis knobless fiber is incorporated in
complete adenovirus vector capsids and exposes both the
6His-tag and the Myc-epitope for specific binding. The
relative incorporation of knobless fibers compared with
wild-type fibers was low. This is most probably a reflec-
tion of the relative abundance of the two fiber species in
the packaging cells.
Figure 5 Accessibility of Myc and 6His-tags on knobless fiber carrying
adenovirus particles. (a) Binding of adenovirus particles to Ni-NTA. 10
12
virus particles were incubated with Ni-NTA beads and unbound particles,
and particles eluting from the beads after competition with 5, 50, and
250 m
m
imidazol, respectively, were analyzed. Functional GFP virus tit-
ers were determined by limiting dilution titration. The bars depict the
recovered virus in each individual Ni-NTA fraction as a percentage of
the total recovered virus. The results shown are the average ±s.d. of two
individual experiments each performed in duplicate. (b) Binding of adeno-
viruses to MoAbs. Left panel, binding of virus to anti-fiber knob MoAb
per 10
9
input particles. Right panel, percentage virus bound to the anti-
Myc MoAb, relative to the positive control binding to anti-fiber knob
MoAb. The figure shows the result of a representative experiment perfor-
med in triplicate. Values given are average ±s.d. No second antibody,
negative control plates coated with RbaMIgG only.
Gene Therapy
Targeted gene transfer by knobless fiber-carrying
adenoviruses into cells displaying an artificial receptor
Knobless fiber-mediated gene transfer was demonstrated
using 293.HissFv.rec cells.
20
293.HissFv.rec cells display
an anti-His single-chain antibody variant on their surface
that functions as an artificial receptor for 6His-tagged
adenoviruses.
20
Hence, these cells can be used to test the
ability of 6His-tagged knobless fibers to function as
primary binding molecules for CAR-independent
adenovirus-mediated gene transfer.
293.HissFv.rec cells were infected with AdGFP or
AdGFP-TSFLCmychis vectors at various MOI and GFP
expression was measured the next day (Figure 6). To dis-
criminate between wild-type and targeted infection, the
alternative fiber–receptor interactions were blocked with
neutralizing anti-fiber knob and anti-Myc antibodies,
respectively. When the CAR-binding site on the wild-
type fiber was blocked with anti-fiber knob antibody,
gene transfer by AdGFP was efficiently inhibited to
approximately 2% background infection. In contrast,
depending on the MOI, AdGFP-TSFLCmychis virus exhi-
bited 6–12% residual gene transfer in the presence of the
anti-fiber knob antibody. Thus, part of the gene transfer
by AdGFP-TSFLCmychis vector was wild-type fiber-
independent. To confirm that this gene transfer was
mediated by the TSFLCmychis knobless fiber, we added
an anti-Myc antibody that binds to the carboxy terminus
of the knobless fiber. This significantly reduced the gene
delivery by AdGFP-TSFLCmychis vector, whereas it had
no effect on the gene transfer by AdGFP. Hence, the
TSFLCmychis knobless fibers on the AdGFP-
TSFLCmychis capsid mediate targeted gene transfer
through binding of their carboxy-terminal peptide to cell
surface receptors.
Figure 6 Knobless fiber-mediated gene transfer into cells displaying an
artificial receptor. 293.HissFv.rec cells were subjected to infection with
AdGFP or AdGFP-TSFLCmychis at MOI of 30, 100 or 300 particles per
cell. Gene transfer was quantified by GFP fluorescence measurement.
Wild-type fiber-dependent gene transfer was blocked with anti-fiber knob
MoAb, knobless fiber-mediated gene transfer was blocked with anti-Myc
MoAb. Results are given as average percent GFP expressing cells ±s.d.
from a representative experiment performed in triplicate. Open bars,
unblocked control infections; black bars, with anti-fiber knob MoAb; gray
bars, with anti-fiber knob MoAb and anti-Myc MoAb. Relative percentage
transduced cells in the presence of neutralizing anti-fiber knob MoAb com-
pared with unblocked controls is indicated above the corresponding bars.
Adenoviruses with knobless fibers
VW van Beusechem
et al
1944
Gene Therapy
Discussion
Genetic modification of adenovirus capsid proteins to tar-
get new cell surface receptors is a potentially efficient
way to overcome the promiscuous tropism of wild-type
adenovirus vectors. The key role of the fiber knob in the
adenovirus infection pathway makes this domain a
rational target for such an endeavor. However, little is
known about the structural requirements for successful
ligand incorporation into the complex fiber knob
domain.
21
The currently available data suggest that both
the carboxy-terminal extension and the HI-loop insertion
strategies allow inserts that are either very small or exhi-
bit a high degree of protein flexibility. Here, we report
on the generation of a new class of chimeric fibers in
which the knob is replaced by a much less complex struc-
ture that complements for the trimerization function of
the knob, but that is completely devoid of its wild-type
binding property. To this end, we selected the trimeric
␣-helical coiled-coil that is found in the extraviral portion
of the p15E envelope protein of MoMuLV
16
and made
two knobless fiber variants that differed only in the nat-
ure of the peptide linkage between the fiber shaft and
trimerization domain. Upon expression in adenovirus
packaging cells, both knobless fibers with their peptide-
tags properly accumulated in the cell nucleus. Further-
more, the two chimeric fibers formed oligomers. In differ-
ent independent experiments performed, however, the
efficiency of trimerization detected on semi-native gels
varied considerably and was sometimes even undetect-
able. We attribute this to a rather weak stability of the
molecules under laboratory conditions, which was not
entirely unexpected, because p15E coiled-coils have a low
thermostability.
22
In any event, the efficiency of trimeriz-
ation was sufficient for incorporation of the chimeric fib-
ers into the viral capsid. Several independent lines of evi-
dence explicitly demonstrated that the TSFLCmychis
knobless fiber variant was incorporated in complete and
functional adenovirus vectors and exposed both its car-
boxy-terminal tags for specific binding. Capsid incorpor-
ation of TSCmychis molecules was much less efficient,
suggesting that flexibility of the linkage between the fiber
shaft and trimerization domain is important. This was a
surprising observation, because trimerization and nuclear
import efficiencies of the two variants seemed compara-
ble. In addition, recent knowledge on the structure of the
fiber shaft suggest that residues 393–398 that were
included near to the linkage site in both knobless fiber
variants already form a flexible linker.
23
Importantly, TSFLCmychis knobless fibers mediated
CAR-independent gene transfer into cells displaying an
artificial receptor for His-tagged adenoviruses. Thus,
TSFLCmychis molecules were functional in providing
primary attachment to alternative cell surface receptors.
The efficiency of targeted gene transfer reached approxi-
mately 10% of wild-type fiber-mediated transduction.
Since the virus binding studies had indicated that
AdGFP-TSFLCmychis particles carried only few
TSFLCmychis molecules compared with wild-type fibers,
we may conclude that TSFLCmychis knobless fiber-
mediated gene delivery is very efficient.
In conclusion, the TSFLC knobless fiber variant exhib-
ited all functions required for its use in truly targeted
gene transfer. It is therefore a prototype substrate to
derive targeted adenovirus vectors for a variety of cell
types that are important targets for gene therapy. To
determine the true value of this approach vectors lacking
wild-type fibers will have to be made. For this purpose,
it is of note that both carboxy-terminal peptide tags of
the knobless fiber were accessible for functional binding.
This will allow the incorporation of a cell type-specific
binding ligand in place of the Myc-epitope, while the
6His-tag can be employed for vector propagation on
293.HissFv.rec cells. Such completely knobless virus par-
ticles may perhaps encounter similar limitations as have
been observed with completely fiberless particles.
24,25
As
yet unidentified domains of the fiber have been shown
to play a role in particle maturation
24,26
and intracellular
trafficking.
27
If either of these functions is mediated by
the fiber knob, completely knobless fiber particles may
exhibit reduced infectivity. Finally, the current generation
of knobless fibers may be further improved by enhancing
their thermostability. To this end, teachings on structural
design of trimeric coiled-coils are at hand.
28,29
The stab-
ility of the current generation of chimeric fibers was
shown to be sufficient for incorporation of peptide tar-
geting-ligands, but a better stability could perhaps
become important when more complex ligands, such as
single-chain antibodies, are added.
Materials and methods
Construction of recombinant plasmids and adenoviral
vectors
The knobless Ad5 fiber genes TSC and TSFLC were con-
structed using PCR techniques. To construct TSC, Ad5
sequences 31042–32250 were amplified using primers T-
for (5′-CTAATACGACTCACTATAGGCTCGAgccaccAT-
GAAGCGCGCAAGACCGTC-3′) and CS-rev (5′-CATC
TCCGGAACCGGTCCACAAAGTTAGCTTATC-3′). T-
for contains a XhoI site (underlined) and sequences fulfil-
ling the Kozak consensus (small case) upstream of the
Ad5 nt 31042–31061. CS-rev contains the antisense
sequences of MoMuLV nt 7316–7319, three codons for
Gly-Ser-Gly (bold), and Ad5 nt 32230–32250. The
MoMuLV p15E helix was amplified using primers TC-for
(5′-GTGGACCGGTTCCGGAGATGATCTCAGGGAGGT
TGA-3′)andC-rev(5′-GCTAGGATCCTCCACCTCC
GGAACCTCCCCCTCCTTCTTTTAGAAATAAC-3′). TC
-for contains Ad5 nt 32244–32250, codons for Gly-Ser-Gly
(bold) and MoMuLV nt 7316–7335. Antisense primer C-
rev contains codons for (Gly
4
-Ser)
2
(bold) including a
restriction site for BamHI (underlined) and MoMuLV nt
7414–7435 (the first two Gly residues of the flexible linker
are from the MoMuLV envelope). Next, both PCR pro-
ducts were mixed and amplified in an assembly PCR
with sense primer T-for and antisense primer XFL-rev (5′-
GCTCTAGAGCTAGGATCCTCCACCTCC-3′), contain-
ing a XbaI site (underlined). To construct TSFLC, Ad5 nt
31042–32250 were amplified using primers T-for and FLS-
rev(5′-GCTATCCTCCGGAA CCGCCTCCACCGGTCCAC
AAAGTTAGCTTATC-3′). FLS-rev contains the antisense
sequences of Gly
4
-Ser-Gly
2
(bold) including a BspEI site
(underlined), and Ad5 nt 32230–32250. The MoMuLV
p15E helix was amplified using primers FLC-for (5′-
GGTTCCGGAGGAGGAGGATCAGGTGGTGGTGG
ATCAGATGATCTCAGGGAGGTTGA-3′) and C-rev.
FLC-for contains codons for Gly-Ser-(Gly
4
Ser)
2
(bold)
including a BspEI site, and MoMuLV nt 7316–7335. Both
Adenoviruses with knobless fibers
VW van Beusechem
et al
1945
PCR products were (partially) digested with BspEI,
mixed, ligated, and amplified using primers T-for and
XFL-rev.
TSC and TSFLC were cloned into pcDNA3 vector
(Invitrogen, San Diego, CA, USA) using the flanking XhoI
and XbaI sites, yielding constructs pCMV-TSC and
pCMV-TSFLC, respectively. Correct construction of the
fusion genes was confirmed by sequencing. pCMV-
TSCmychis and pCMV-TSFLCmychis were made by
replacing the NotI–BamHI fragment from pcDNA3.1
(−)/Myc-His/LacZ (Invitrogen) with the NotI–BamHI
fragment from pCMV-TSC or pCMV-TSFLC. In the ORF
of pCMV-TSCmychis and pCMV-TSFLCmychis the C-
terminal (G
4
S)
2
linker is followed by the 29 amino acid
sequence ELGTKLGPEQKLISEEDLNSAVDHHHHHH,
where the Myc-epitope and 6His-tag are underlined.
The Ad2 tripartite leader (TPL) sequence was amplified
from pMad5
30
using primers X-TPL (5′-TGCTCTA-
GACTCTCTTCCGCATCGCTG-3′), containing a XbaI site
(underlined), and TPL-E (5′-CAGGAATTCTTGC
GACTGTGACTGGTTAG-3′), with an EcoRI site
(underlined). The XbaI and XhoI (TPL nt 172) digested
PCR product was inserted into XbaI/XhoI digested
pCMV-TSCmychis or pCMV-TSFLCmychis, creating
pCMVtpl-TSCmychis and pCMVtpl-TSFLCmychis,
respectively.
The Ad2 major late promoter (MLP) and TPL were
amplified from pMad5 using primers N-MLP (5′-CTAA-
GAATGCGGCCGCGAGCGGTGTTCCGCGGTC-3′), con-
taining a NotI site (underlined) and TPL-E. The PCR pro-
duct was NotI/EcoRI (blunt) inserted into pBluescript II
SK(−) (Stratagene, La Jolla, CA, USA) upstream of the
XhoI (blunt)/PmeI fragment of pCMVtpl-TSCmychis or
pCMVtpl-TSFLCmychis and the BamHI (blunt)/HindIII
SV40-pA fragment from pTet-Off (Clontech, Palo Alto,
CA, USA). The correct sequence of the MLP-TPL PCR
product was confirmed. The NotI–ClaI inserts were iso-
lated and the ClaI sites were made blunt-end. These frag-
ments were cloned into pAdTrack
31
digested with
NotI/KpnI (blunt), giving constructs pAdTrackMLP-
TSCmychis and pAdTrackMLP-TSFLCmychis, respect-
ively.
Adenovirus vector production
Recombinant adenoviruses were produced by homolo-
gous recombination in 293 cells.
18
Adenovirus backbone
plasmid pAdEasy-1
31
was digested with PacI and cotrans-
fected together with PacI/PmeI-digested pAdTrack, pAd-
TrackMLP-TSCmychis or pAdTrackMLP-TSFLCmychis
by the Lipofectamine PLUS (Life Technologies, Paisley,
UK) method according to the manufacturer’s guidelines.
Virus was prepared and further expanded on 293 cells
using standard procedures.
Purified virus stocks were prepared by two rounds of
CsCl banding and dialysis against 10 mmHepes pH 7.4,
10% glycerol, 1 mmMgCl (storage buffer; SB) and were
stored at −80°C until use. Virus identity was confirmed
by PCR analysis. Virus particle titers were determined by
OD260 measurement after lysis in PBS, 1% SDS, 1 mm
EDTA at 55°C. Functional titers in infectious units were
determined by end-point dilution infection on
293.HissFv.rec cells
20
and detecting GFP-expressing cells
6 days after infection. The absence of replication-com-
petent adenovirus was confirmed by infection of A549
(ATCC No. CCL-185) cells.
Gene Therapy
Analysis of knobless fiber expression
Transfection on 911 cells
17
was performed using Lipofec-
tamine PLUS reagent (Life Technologies) according to the
manufacturer’s instructions. Expression analysis by
immunocytochemistry was performed at 24 h after trans-
fection, using anti-Myc MoAb 9E10,
32
RbaMIgG-AP con-
jugate (Dako, Glostrup, Denmark) and BCIP/NBT sub-
strate (Dako). Cells were counterstained with nuclear
fast red.
For immunoblot analysis of adenovirus vector-infected
293 cells, cells in full CPE were harvested, washed in PBS,
and lysed by four freeze–thaw cycles and three times 10 s
sonification. Western samples were prepared by adding
2 volumes Laemmli loading buffer with 2% SDS, 5% -
mercaptoethanol and 5 min heating at 95°C (denaturing
condition) or a modified loading buffer with 0.2% SDS
and lacking -mercaptoethanol, without heating (semi-
native condition). Samples were separated by 7.5% SDS-
PAGE and transferred to PVDF membrane (Sequiblot;
Bio-Rad, Hercules, CA, USA). Immunoblots were pro-
cessed according to standard procedures, using anti-Myc
MoAb 9E10 or anti-fiber knob MoAb 1D6.14,
33
RbaMIgG-
HRPO conjugate (Dako) and Lumilight
PLUS
chemilumi-
nescence detection reagent (Boehringer Mannheim,
Mannheim, Germany). Fibers on adenovirus particles
were analyzed under semi-native conditions as described
above for cell lysates, starting from CsCl purified virus
stocks in SB.
Ni-NTA purification of His-tagged proteins and viruses
Cell lysates were prepared from adenovirus vector-
infected 293 cells as described above and cleared by cen-
trifugation. Ni-NTA Superflow resin slurry (Qiagen,
Hilden, Germany) equilibrated in PBS with 300 mmNaCl
(PBS-300) and 5 mmimidazol was added to reach 10%
v/v beads. After 1 h incubation at 4°C, Ni-NTA resin was
spun down and the unbound material was harvested.
The beads were washed twice with 3 volumes PBS-
300/5 mmimidazol (discarded) and once with 3 volumes
PBS-300/30 mmimidazol. Finally, specifically bound
material was eluted by incubation with 3 volumes PBS-
300/300 mmimidazol. The eluted proteins were concen-
trated using Ultrafree-0.5 centrifugal filters with Biomax-
10 membrane (Millipore, Bedford, MA, USA) according
to the manufacturer’s instructions.
Virus particles (10
12
) in SB were mixed with 0.1 volume
Ni-NTA beads equilibrated in SB and incubated for 5 h
at 4°C by end-over-end rotation. Beads were sedimented
by gravity on ice and unbound material was aspirated.
Beads were washed twice with 9 volumes SB/5 mmimid-
azol, twice with 9 volumes SB/50 mmimidazol and once
with 9 volumes SB/250 mmimidazol. The unbound frac-
tion and first-step elutions at 5 mm,50mm, and 250 mm
imidazol were kept for analysis.
Antibody-mediated virus binding assay
Ninety-six-well Microlon 200 ELISA plates (Greiner,
Frickenhausen, Germany) were coated with 10 g/ml
RbaMIgG MoAb (Dako) in PBS for 2 h at 37°C. After two
washes with PBS, plates were incubated for 1.5 h at 37°C
with PBS/1% BSA as a negative control or with 1 g/ml
anti-fiber knob MoAb 1D6.14 or anti-Myc MoAb 9E10 in
PBS/1% BSA. After three washes with PBS, limiting
dilution particle titrations of CsCl-purified virus were
made in triplicate in DMEM/F12 medium (Life
Adenoviruses with knobless fibers
VW van Beusechem
et al
1946
Gene Therapy
Technologies) with 2% fetal bovine serum (FBS) and
allowed to bind for 1 h at 37°C. Unbound virus was
removed by five washes with PBS. Next, 293.HissFv.rec
cells in DMEM/F12 medium with 10% FBS were seeded
into the wells. After overnight culture at 37°C, cells were
harvested, washed in PBS, fixated in PBS with 2.5% for-
maldehyde and analyzed for GFP fluorescence on a FAC-
Scan (Becton Dickinson, San Jose, CA, USA) according to
standard procedures. Titer of functional virus bound was
calculated according to the following equation: % GFP-
expressing cells x number of cells seeded ×virus dilution.
Targeted virus infection assay
293.HissFv.rec cells were seeded at 1 ×10
5
cells per well
in 24-well tissue culture plates 24 h before infection. Virus
was diluted in triplicate in DMEM/F12 with 1% FBS with
or without 50 g/ml 1D6.14 MoAb and/or 25 g/ml
9E10 MoAb, and incubated at RT for 30 min. Next, the
virus/antibody mixtures were added to the cells at MOI
ranging from 30 to 300 particles per cell. The virus was
allowed to bind for 30 min at RT, following which the
medium was replaced by DMEM/F12 with 10% FBS.
After 24 h incubation at 37°C, cells were harvested and
analyzed for GFP expression on a FACScan according to
standard procedures.
Acknowledgements
We thank David Curiel and Joanne Douglas, Gene Ther-
apy Center UAB, Birmingham, Alabama for providing
the 293.HissFv.rec cells and 1D6.14 MoAb; and Frits Fal-
laux, Leiden University Medical Center, The Netherlands
for plasmid pMad5. This work was supported by the
Netherlands Organization for Scientific Research
(Spinoza Award 1997).
References
1 Herz J, Gerard RD. Adenovirus-mediated transfer of low den-
sity lipoprotein receptor gene acutely accelerates cholesterol
clearance in normal mice. Proc Natl Acad Sci USA 1993; 90:
2812–2816.
2 Huard J et al. The route of administration is a major determinant
of the transduction efficiency of rat tissues by adenoviral recom-
binants. Gene Therapy 1995; 2: 107–115.
3 Louis N et al. Cell-binding domain of adenovirus serotype 2
fiber. J Virol 1994; 68: 4104–4106.
4 Henry LJ et al. Characterization of the knob domain of the aden-
ovirus type 5 fiber protein expressed in Escherichia coli. J Virol
1994; 68: 5239–5246.
5 Michael SI, Hong JS, Curiel DT, Engler JA. Addition of a short
peptide ligand to the adenovirus fiber protein. Gene Therapy
1995; 2: 660–668.
6 Wickham TJ, Roelvink PW, Brough DE, Kovesdi I. Adenovirus
targeted to heparan-containing receptors increases its gene
delivery efficiency to multiple cell types. Nat Biotechnol 1996; 14:
1570–1573.
7 Wickham TJ et al. Increased in vitro and in vivo gene transfer by
adenovirus vectors containing chimeric fiber proteins. J Virol
1997; 71: 8221–8229.
8 Yoshida Y et al. Generation of fiber-mutant recombinant adeno-
viruses for gene therapy of malignant glioma. Hum Gene Ther
1998; 9: 2503–2515.
9 Krasnykh V et al. Characterization of an adenovirus vector con-
taining a heterologous peptide epitope in the HI loop of the fiber
knob. J Virol 1998; 72: 1844–1852.
10 Dmitriev I et al. An adenovirus vector with genetically modified
fibers demonstrates expanded tropism via utilization of a
coxsackievirus and adenovirus receptor-independent cell entry
mechanism. J Virol 1998; 72: 9706–9713.
11 Roelvink PW et al. Identification of a conserved receptor-binding
site on the fiber proteins of CAR-recognizing adenoviridae.
Science 1999; 286: 1568–1571.
12 Novelli A, Boulanger PA. Deletion analysis of functional
domains in baculovirus-expressed adenovirus type 2 fiber.
Virology 1991; 185: 365–376.
13 Hong JS, Engler JA. Domains required for assembly of aden-
ovirus type 2 fiber trimers. J Virol 1996; 70: 7071–7078.
14 Santis G et al. Molecular determinants of adenovirus serotype 5
fibre binding to its cellular receptor CAR. J Gen Virol 1999; 80:
1519–1527.
15 Cohen C, Parry DAD. ␣-Helical coiled coils and bundles: how
to design an a-helical protein. Proteins Structure Function, Gen-
etics 1990; 7: 1–15.
16 Fass D, Harrison SC, Kim PS. Retrovirus envelope domain at
1.7 angstrom resolution. Nat Struct Biol 1996; 3: 465–469.
17 Fallaux FJ et al. Characterization of 911: a new helper cell line
for the titration and propagation of early region 1-deleted
adenoviral vectors. Hum Gene Ther 1996; 7: 215–222.
18 Graham FL, Smiley J, Russell WC, Nairn R. Characteristics of a
human cell line transformed by DNA from human adenovirus
type 5. J Gen Virol 1977; 36: 59–74.
19 Mitraki A et al. Unfolding studies of human adenovirus type 2
fibre trimers. Evidence for a stable domain. Eur J Biochem 1999;
264: 599–606.
20 Douglas JT et al. A system for the propagation of adenoviral
vectors with genetically modified receptor specificities. Nat
Biotechnol 1999; 17: 470–475.
21 Xia D, Henry LJ, Gerard RD, Deisenhofer J. Crystal structure of
the receptor-binding domain of adenovirus type 5 fiber protein
at 1.7 A resolution. Structure 1994; 2: 1259–1270.
22 Fass D, Kim PS. Dissection of a retrovirus envelope protein
reveals structural similarity to influenza hemagglutinin. Curr
Biol 1995; 5: 1377–1383.
23 van Raaij MJ, Mitraki A, Lavigne G, Cusack S. A triple beta-
spiral in the adenovirus fibre shaft reveals a new structural
motif for a fibrous protein. Nature 1999; 401: 935–938.
24 Legrand V et al. Fiberless recombinant adenoviruses: virus
maturation and infectivity in the absence of fiber. J Virol 1999;
73: 907–919.
25 Von Seggern DJ, Kehler J, Endo RI, Nemerow GR. Comp-
lementation of a fibre mutant adenovirus by packaging cell lines
stably expressing the adenovirus type 5 fibre protein. J Gen Virol
1998; 79: 1461–1468.
26 Falgout B, Ketner G. Characterization of adenovirus particles
made by deletion mutants lacking the fiber gene. J Virol 1988;
62: 622–625.
27 Miyazawa N et al. Fiber swap between adenovirus subgroups B
and C alters intracellular trafficking of adenovirus gene transfer
vectors. J Virol 1999; 73: 6056–6065.
28 Harbury PB, Zhang T, Kim PS, Alber T. A switch between two-,
three-, and four-stranded coiled coils in GCN4 leucine zipper
mutants. Science 1993; 262: 1401–1407.
29 Harbury PB, Kim PS, Alber T. Crystal structure of an isoleucine-
zipper trimer. Nature 1994; 371: 80–83.
30 Toes RE et al. Protective anti-tumor immunity induced by vacci-
nation with recombinant adenoviruses encoding multiple
tumor-associated cytotoxic T lymphocyte epitopes in a string-
of-beads fashion. Proc Natl Acad Sci USA 1997; 94: 14660–14665.
31 He TC et al. A simplified system for generating recombinant
adenoviruses. Proc Natl Acad Sci USA 1998; 95: 2509–2514.
32 Chan S et al. A novel tumour marker related to the c-myc oncog-
ene product. Mol Cell Probes 1987; 1: 73–82.
33 Douglas JT et al. Targeted gene delivery by tropism-modified
adenoviral vectors. Nat Biotechnol 1996; 14: 1574–1578.
