Mechanism of Ad5 Vaccine Immunity
and Toxicity: Fiber Shaft Targeting
of Dendritic Cells
Cheng Cheng1[, Jason G. D. Gall2[, Wing-pui Kong1, Rebecca L. Sheets1, Phillip L. Gomez1, C. Richter King2,
Gary J. Nabel1*
1 Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland, United States of America, 2 GenVec,
Incorporated, Gaithersburg, Maryland, United States of America
Recombinant adenoviral (rAd) vectors elicit potent cellular and humoral immune responses and show promise as
vaccines for HIV-1, Ebola virus, tuberculosis, malaria, and other infections. These vectors are now widely used and
have been generally well tolerated in vaccine and gene therapy clinical trials, with many thousands of people
exposed. At the same time, dose-limiting adverse responses have been observed, including transient low-grade fevers
and a prior human gene therapy fatality, after systemic high-dose recombinant adenovirus serotype 5 (rAd5) vector
administration in a human gene therapy trial. The mechanism responsible for these effects is poorly understood.
Here, we define the mechanism by which Ad5 targets immune cells that stimulate adaptive immunity. rAd5 tropism
for dendritic cells (DCs) was independent of the coxsackievirus and adenovirus receptor (CAR), its primary receptor or
the secondary integrin RGD receptor, and was mediated instead by a heparin-sensitive receptor recognized by a
distinct segment of the Ad5 fiber, the shaft. rAd vectors with CAR and RGD mutations did not infect a variety of
epithelial and fibroblast cell types but retained their ability to transfect several DC types and stimulated adaptive
immune responses in mice. Notably, the pyrogenic response to the administration of rAd5 also localized to the shaft
region, suggesting that this interaction elicits both protective immunity and vector-induced fevers. The ability of
replication-defective rAd5 viruses to elicit potent immune responses is mediated by a heparin-sensitive receptor that
interacts with the Ad5 fiber shaft. Mutant CAR and RGD rAd vectors target several DC and mononuclear subsets and
induce both adaptive immunity and toxicity. Understanding of these interactions facilitates the development of
vectors that target DCs through alternative receptors that can improve safety while retaining the immunogenicity of
Citation: Cheng C, Gall JGD, Kong WP, Sheets RL, Gomez PL, et al. (2007) Mechanism of Ad5 vaccine immunity and toxicity: Fiber shaft targeting of dendritic cells. PLoS
Pathog 3(2): e25. doi:10.1371/journal.ppat.0030025
The efficacy of adenovirus vectors as vaccines in many
animal models of infectious diseases [1–3] and their immu-
nogenicity in early clinical evaluation indicate their potential
for human use. The mechanism underlying their strong
immunogenicity and their relationship to adverse responses
has not been well defined, nor has the connection between
immunogenicity and adverse responses [4–6]. To address this
issue, we evaluated the contribution of known receptor
binding domains in the recombinant adenovirus serotype 5
(rAd5) fiber and penton base. The primary receptor
recognition sequence resides in the knob region of the fiber.
This domain has been localized to regions identified
structurally , specifically in the AB loop, the B b-sheet,
and the DE loop of the knob, and interacts with coxsack-
ievirus and adenovirus receptor (CAR) [8–10]. Binding and
internalization are facilitated through an interaction of an
RGD motif in the penton base with integrin receptors [11,12].
To evaluate the contributions of these regions to targeting of
rAd vectors to different cell types, we prepared vectors with
mutations in these domains. Previous studies have shown that
mutations in the CAR-binding domain inhibit infection of
many cell types [10,13], and modifications of the RGD domain
also affect targeting. Investigations of cell culture trans-
duction do not always predict transduction of similar cell
types in vivo. However, analysis of tissue transduction
following intramuscular injection shows that elimination of
both CAR and integrin receptor interactions greatly dimin-
ishes local transduction in muscle . On the other hand,
the long shaft in the fiber of Ad5 determines its hepatic
tropism for systemic administration in mice [15,16]. In this
study, we systematically investigated the contribution of these
domains to the immunogenicity of Ad5-based vaccine vector
when administered intramuscularly in mice and defined the
molecular basis of its toxicity in a rabbit pyrogenicity model.
Editor: John A. T. Young, The Salk Institute for Biological Studies, United States of
Received July 31, 2006; Accepted January 5, 2007; Published February 23, 2007
This is an open-access article distributed under the terms of the Creative Commons
Public Domain declaration which stipulates that, once placed in the public domain,
this work may be freely reproduced, distributed, transmitted, modified, built upon,
or otherwise used by anyone for any lawful purpose.
Abbreviations: BM, bone marrow; CAR, coxsackievirus and adenovirus receptor;
DC, dendritic cell; GFP, green fluorescent protein; PU, particle unit; rAd,
recombinant adenovirus; rAd5, recombinant adenovirus serotype 5; VRC, Vaccine
Research Center; WT, wild-type
* To whom correspondence should be addressed. E-mail: firstname.lastname@example.org
[ These authors contributed equally to this work.
PLoS Pathogens | www.plospathogens.org February 2007 | Volume 3 | Issue 2 | e250239
Preparation of rAd Vectors
The construction and propagation of the rAd5 vectors with
wild-type (WT) capsid proteins and with mutated CAR- and
integrin-binding motifs were previously described . Total
particle unit (PU) titer was determined by absorbance .
The chimeric rAd5þAd35knob vector was a kind gift of Andre
Leiber and is E1 and E3 deleted with the green fluorescent
Isolation and Culture of Different Dendritic Cells from
Peripheral Blood, Bone Marrow, and Spleen
Human peripheral blood was obtained from the National
Institutes of Health Clinical Center Blood Bank (Bethesda,
Maryland, United States), and mononuclear cells (PBMCs)
were isolated by gradient centrifugation with Ficoll-Paque
PLUS (Amersham Biosciences, http://www.amersham.com) as
buffy coat and cultured in 10% RPMI medium (Invitrogen,
http://www.invitrogen.com). Plasmacytoid dendritic cells
(DCs) were isolated by magnetic cell sorting with BDCA-4
cell isolation kit (Miltenyi Biotec, http://www.miltenyibiotec.
Bone marrow (BM)-derived DCs were obtained from BM of
BALB/c mice and cultured according to published methods
. More than 80% of these cells cultured with mGM-CSF
after 1 wk expressed DC surface markers CD11b and CD11c
as measured by flow cytometry.
Lymphoid DCs (CD8þDCs) and plasmacytoid DCs (B220þ
DCs) were isolated from mouse spleens by magnetic cell
sorting according to the manufacturer’s protocol (Miltenyi
Biotec). More than 90% of these purified cells expressed CD8
or B220 as measured by antibody staining of the cells.
Six- to 8-wk-old BALB/c female mice were used for
immunogenicity studies. Mice were injected once with 100
ll of the specified rAd vectors encoding GFP or HIV Env
(gp140DCFI) as the control vector at the indicated particle
concentration bilaterally in the muscle with the use of needle
and syringe. For each vector and dose, a group of five mice
was injected with vector in PBS. All animal experiments were
reviewed and approved by the Animal Care and Use
Committee, Vaccine Research Center (VRC), National In-
stitute of Allergy and Infectious Diseases (http://www.niaid.
nih.gov/vrc) and performed in accordance with all relevant
federal and National Institutes of Health guidelines and
Cellular Immune Analysis
Three weeks after vaccination, mouse spleens were
removed aseptically, gently homogenized to a single-cell
suspension, washed, and resuspended to a final concentration
of 106cells/ml. Harvested spleen cells (106cells/peptide pool)
were stimulated for 6 h in the presence of 2 lg of anti-CD28
and anti-CD49d MAbs/ml (BD PharMingen, http://www.
pharmingen.com). The last 5 h of stimulation occurred in
the presence of 10 lg/ml brefeldin A (Sigma, http://www.
sigmaaldrich.com), with no stimulation as the background
control or with phorbol myristate acetate (PMA) as the
positive control, or peptide pools having the same amino acid
sequences as GFP, or Ebola GP protein as the negative control
(Figures S2 and S3). All peptides used in this report were 15
mers overlapping by 11 amino acids that spanned the
complete sequence of the protein. Cells were permeabilized
and fixed with Cytofix/Cytoperm and stained with mono-
clonal antibodies (rat anti-mouse cell surface antigens CD3-
PE, CD4-PerCP and CD8-APC; BD PharMingen) followed by
multiparametric flow cytometry to detect the IFN-c (IFN-c-
FITC) and TNF-a (TNF-a-FITC) –positive cells in the CD4þor
CD8þT-cell population. Statistical analyses in observed CD4þ
and CD8þresponses between control-vaccinated and test
article–vaccinated mice were performed by the t-test using
Microsoft Excel software (http://www.microsoft.com).
Samples were assayed on an FACSCalibur instrument using
CELLQuest software (BD Biosciences, http://www.
bdbiosciences.com). The collected data were analyzed with
FlowJo 6.1 software (Tree Star, http://www.flowjo.com).
ELISA plates of 96 wells were coated with 100 ll/well
purified GFP (BD Biosciences) at 2.5 lg/ml and kept overnight
at 4 8C. The GFP was removed, and each well was blocked with
200 ll of PBS containing 10% FBS for 2 h at room
temperature. The plates were washed twice with PBS
containing 0.2% Tween-20 (PBS-T). Then, 100 ll of serum
from vaccinated mice was added to each well at a dilution of
1:100. The plates were incubated for 1 h at room temperature
and washed. Afterward, 100 ll of horseradish peroxidase–
conjugated goat anti-mouse IgG was added to each well. The
plates were again incubated for 1 h at room temperature and
washed. Subsequently, 50 ll of substrate (fast o-phenylenedi-
amine dihydrochloride; Sigma) was added to each well. The
plates were then incubated for 30 min at room temperature.
The reaction was stopped by the addition of 100 ll of 1(N)
H2SO4, and the optical density was read at 450 nm.
Neutralization of Adenovectors with Human Sera
One hundred samples of human sera from volunteers
enrolled in VRC-sponsored HIV trials in the United States
PLoS Pathogens | www.plospathogens.orgFebruary 2007 | Volume 3 | Issue 2 | e250240
Ad5 Fiber Mediates DC Targeting/Toxicity
Recombinant adenovirus (rAd) vectors are remarkable for their
ability to stimulate potent immune responses and to mediate highly
efficient gene transfer. These vectors have been used extensively in
human studies with generally acceptable tolerability. As with many
bioactive compounds, adenoviruses can also cause potentially
serious side effects, as observed in a human gene therapy trial
several years ago that led to a fatality. The first manifestation of this
toxicity is fever, but the relation of this side effect to the ability of
the vector to stimulate immunity was unknown. We show that
targeting of rAd vectors induces vaccine responses and toxicity
through a previously unrecognized mechanism related to its
attachment and entry into cells. We find that both adaptive
immunity and fever are mediated by targeting of the rAd vector
to dendritic cells and some monocytes, independent of the
coxsackievirus and adenovirus receptor and RGD binding domains,
mediated instead by the fiber shaft. This finding suggests that a
distinct receptor present on dendritic and mononuclear cells
mediates both effects. The immunogenicity of rAd vectors is
dependent on targeting of virus through a specific fiber region
and mediates rAd toxicity, which has implications for vaccine and
gene therapy vector design that may help to improve rAd safety and
were obtained from the VRC Immunology Core Laboratory.
The sera were diluted with Dulbecco’s modified Eagle’s
medium (1:12) and mixed with the indicated rAd vector
encoding GFP for 1 h at room temperature. The neutralized
virus was used to infect 293 cells at 500 PU/cell for 2 h, and
GFP expression was analyzed by flow cytometry at 24 h post
The 11- to 16-wk-old New Zealand White rabbits were
administered 1012particles of vector intramuscularly. This
dose ensured that 100% of the animals injected with WT
capsid rAd5 would have elevated body temperature because
1011particles induced fevers in 60% of the animals
(unpublished data). Body temperature was measured using
subcutaneously implanted thermometers (BMDS transponder
IPTT-200; BioMedic Data Systems Inc, http://www.bmds.com)
at the nape of the neck.
Transduction of Different Cell Types by WT and CAR?RGD?
To identify the viral component responsible for gene
transfer into specific cell types, mutant adenoviral vectors
were constructed and tested in vitro for gene transfer and
expression. The CAR?RGD?rAd vector failed to mediate
gene transfer into a number of epithelial and fibroblast cell
lines of human or mouse origin that were readily transduced
with a similar amount of WT rAd expressing a GFP reporter
(Figures 1A, left, and S1). The specificity of CAR?RGD?rAd
was evaluated further in DC subsets and mononuclear cells
Figure 1. Specificity of WT or CAR?RGD?rAd Vectors for Different Cell
Types: Mutant Virus Retains Ability to Infect DCs but Not Other Cell Types
(A) GFP expression following transduction by rAd5 vectors with WT
capsid proteins or mutated fiber and penton base proteins at 1,000 viral
particles per cell. Transduction of human squamous epithelial cell A549,
mouse BM-derived DCs, and human CD11cþPBMCs was assessed by flow
cytometry at 48 h post transduction.
(B) Transduction by the indicated rAd vectors expressing luciferase was
evaluated in mouse DCs. BM cells were isolated from BALB/c mice and
cultured for 1 wk in the presence of GM-CSF . The cells were
transduced with indicated viruses at 3,000 viral particles per cell, and at
24 h postinfection, cells were stained with anti-CD19 and -CD11c
antibodies and sorted into CD19?and CD11cþcells with FACSDiVa (BD
Biosciences). Mouse plasmacytoid and lymphoid DCs were isolated with
magnetic cell sorting and transduced with the indicated viruses at 10,000
viral particles per cell. Luciferase expression was measured at 24 h after
transduction. The average of three infections is shown, along with
Figure 2. Transduction Efficiencies of WT and Mutant rAd Vectors in DCs
and Sensitivity of Mutant Adenoviral Vector Infectivity to Heparin
Murine BM DCs (A) or human plasmacytoid DCs (B) isolated from PBMCs
using MACS magnetic cell separation (Miltenyi Biotech Inc) were
transfected with the indicated viral vectors expressing luciferase and
luciferase expression were analyzed at 24 h postinfection. The control
consisted of cells mock-transfected with culture medium. The data are
the representatives of three independent transfections.
(C) Murine BM DCs were preincubated with heparin or heparan sulfate
for 30 min at room temperature and then transfected with the
CAR?RGD?rAd at 100 viral particles per cell, and relative luciferase
activity was evaluated in a dose-response fashion to heparin or heparan
sulfate at 24 h postinfection. The data shown are representative of three
(D) BM-derived DCs were transduced with the CAR?RGD?rAd or
CAR?RGD?rAd with Ad35 shaft at 3,000 viral particles per cell, and
luciferase expression was analyzed at 24 h postinfection.
PLoS Pathogens | www.plospathogens.org February 2007 | Volume 3 | Issue 2 | e250241
Ad5 Fiber Mediates DC Targeting/Toxicity
derived from alternative tissues. Murine BM cells were
isolated and incubated with mGM-CSF to promote differ-
entiation into BM DCs . The CAR?RGD?rAd readily
transduced these cells as measured with a GFP reporter,
although transduction of this mixed population was more
efficient with WT virus (Figure 1A, center). Unseparated
human CD11cþmononuclear cells derived from peripheral
blood were also transduced by the mutants with similar
efficiency to the WT virus (Figure 1A, right). When these cells
were purified to yield murine BM DCs by sorting
CD19?CD11chighcells, the CAR?RGD?rAd vector transduced
these cells with similar efficiency to the WT virus, as
determined with a luciferase reporter (Figure 1B, left). The
CAR?RGD?rAd vector was able to transduce other DC
subsets from alternative tissues: both mouse B220þ(plasma-
cytoid) DCs and CD8þ(lymphoid) DCs derived from spleen
cells were readily transduced by the mutant virus (Figure 1B,
middle, right). A titration of input vector showed slightly
lower transduction efficiencies by the CAR?RGD?rAd vector,
as measured by slightly reduced luciferase reporter activity in
murine BM-derived and plasmacytoid DCs; nonetheless, the
transduction was comparable to the WT capsid vector over a
2-log range of multiplicities of infection (Figure 2A and 2B).
These findings indicate that transduction of several DC and
mononuclear subsets is independent of CAR and integrin
Adenoviral infection mediated by the shaft region of the
fiber protein has been reported for certain cell types. Fiber
shaft structure is specific to different serotypes, with respect
to both the specific amino acid sequence and shaft length as
measured by the number of repeats in each fiber. A KKTK
motif has been identified in the Ad5 shaft that mediates
transduction of some cells, and transduction mediated by this
motif is sensitive to inhibition by heparin [20–22]. To test the
role of the KKTK motif, we evaluated transduction of DCs in
the presence of heparin sulfate or heparan sulfate, an
alternative proteoglycan that served as a negative control.
Heparin sulfate nearly completely inhibited CAR?RGD?rAd
gene transfer to murine BM-derived DCs, in contrast to
heparan sulfate (Figure 2C), which suggests that ionic
interactions contained within the fiber shaft are necessary.
We also substituted the Ad5 shaft with the Ad35 shaft, which
lacks the KKTK motif, in the CAR?RGD?mutant. Gene
transfer to murine BM-derived DCs by this mutant was
substantially reduced (Figure 2D).
Immunogenicity of Mutated Ad5 Vectors in Mice
To define the Ad5 viral determinants of immunogenicity in
vivo, mice were injected with increasing amounts of different
recombinant viruses expressing GFP as the antigen. When the
WT and CAR?RGD?rAd were analyzed, there was no
significant difference (p . 0.05) in the peak levels of cellular
immune response elicited by the WT and double-mutant
vectors (Figure 3). Thus, although the specificity of the
CAR?RGD?adenoviral vector differs markedly from the WT
vector, its ability to bind to DCs remains unchanged, and it is
potently immunogenic in vivo. Finally, to confirm the role of
the shaft in stimulating immune responses by rAd vectors, the
immunogenicity of rAd5 was compared to rAd5 with Ad35
knob transposed onto the fiber. Both vectors stimulated
responses that were significantly above those from a vector
containing a control insert (Figure 4A). Instead of binding to
CAR, the Ad35 knob normally binds to CD46 in humans, but
CD46 is absent from nearly all mouse tissues . Con-
sequently, the rAd5 vector containing the Ad35 knob is
expected to have little cellular binding dependent on the
knob. However, both vectors can utilize the Ad5 shaft and
showed comparable CD4 and CD8 intracellular cytokine
staining and increased antibody titers (Figure 4A), confirming
the role of the Ad5 shaft independent of the Ad5 knob in T-
Neutralization of WT and Chimeric Ad5 and Ad35 Vectors
by Human Sera
Because the Ad5 shaft had a significant role in the adaptive
immune response, it was possible that the shaft was a target
for neutralizing antibody. A panel of rAd35 vectors engi-
neered with Ad5 fiber domains were compared to rAd5 for
susceptibility to neutralization by human serum samples
Figure 3. Immunogenicity of WT and Mutant rAd Vectors
Dose-response analysis of WT or CAR?RGD?mutant rAd immunogenicity in vivo. Cellular (CD4, CD8) and humoral (IgG) immune function was analyzed
at the indicated doses of rAd vector. Spleen cells were stimulated with GFP peptides or Ebola virus glycoprotein (EB peptides) as the negative control, or
peptide dissolving solvent as background control, and stained for the indicated cell surface markers and intracellular cytokines using fluorochrome-
conjugated monoclonal antibodies. The percentage of intracellular IFN-cþ(IFN-c-FITC) and TNF-aþcells in total CD4þor CD8þin GFP peptide-stimulated
samples was subtracted from the background control samples (no peptide stimulation). The samples were also stimulated with an irrelevant peptide
pool derived from Ebola virus glycoprotein (Figure S2). The average of data from five mice in each group and the standard deviations are shown. For
CD4þand CD8þresponses, at each dosage, there is no significant difference between WT and the mutant (p . 0.05, t-test by Microsoft Excel). Mouse
sera were diluted at 1:100, and anti-GFP IgG was measured by ELISA. The averages of five mouse sera in each group with standard deviations are
presented. Mutant generated less IgG response compared with WT at each dosage (p , 0.05, t-test by Microsoft Excel).
PLoS Pathogens | www.plospathogens.orgFebruary 2007 | Volume 3 | Issue 2 | e250242
Ad5 Fiber Mediates DC Targeting/Toxicity
obtained from healthy adults. Transposition of the Ad5 shaft
to rAd35 did not affect neutralization; however, the presence
of the Ad5 knob alone changed the neutralization profile of
rAd35 to that of rAd5 (Figure 4B). Thus, the shaft region did
not contain human neutralization epitopes.
Pyrogenicity of Ad5-Based Vectors in a Rabbit Model
To determine whether the Ad5 shaft contributed to the
pyrogenicity observed after the administration of rAd
vectors, rabbits were injected with a high dose, 1012particles,
of rAd5 vectors. All vectors containing the rAd5 shaft,
Figure 4. Ad5 Shaft Does Not Contain Human Neutralization Epitopes and Contributes to In Vivo Immunogenicity and Pyrogenicity of rAd5 Vector
(A) Mice were injected with rAd5 or a chimeric rAd5 containing the knob region from Ad35 at 1010viral particles (Ad5 35knob). Cellular immune
responses were assessed by analysis of intracellular cytokine staining in CD4þor CD8þcells, and antibody responses were assessed by ELISA in the same
animals (n ¼ 5). The samples were also stimulated with an irrelevant peptide pool derived from Ebola virus glycoprotein (Figure S3).
(B) Neutralization of rAd vectors with 100 individual human sera. Recombinant Ad vectors with chimeric rAd5-rAd35 fiber proteins were incubated with
serum diluted 1:12 for 1 h and added to the cells, and the GFP fluorescence intensity arising from infection was determined by flow cytometry. Sera that
inhibited GFP expression by more than 50% were considered positive for neutralization.
(C) Pyrogenicity of rAd5 vectors evaluated in rabbits. Rabbits were injected intramuscularly with 1012particles of the indicated rAd5-luciferase vectors,
and body temperature was measured at 24 h postinjection for each animal (n¼5 for all the groups except n¼3 in the control group of the right graph).
The control groups were injected with the storage and dilution buffer for rAd vectors. The rAd5 vectors in the left panel had WT Ad5 capsid proteins
(WT), a deletion of the RGD motif in the penton base protein (DRGD), a mutation in the fiber knob that prevented CAR binding (DCAR), or both capsid
protein mutations (DRGDDCAR). The two rAd5 vectors in the right panel were the WT capsid vector (WT) and a DRGDDCAR vector with the Ad5 shaft
region replaced with the Ad35 shaft region (DRGDDCAR 35shaft).
PLoS Pathogens | www.plospathogens.orgFebruary 2007 | Volume 3 | Issue 2 | e25 0243
Ad5 Fiber Mediates DC Targeting/Toxicity
including those with mutations in the penton base RGD and
fiber CAR-binding domain, induced a statistically significant
increase in body temperature (Figure 4C, left), including
those in which both the CAR and integrin interactions had
been ablated. In contrast, the group of animals that received
the CAR?RGD?rAd5 vector without the Ad5 shaft had
significantly lower mean body temperature relative to the
group with the Ad5 shaft (Figure 4C, right). Thus, the ability
of a vector to induce a potent pyrogenic response was also
associated with the Ad5 shaft, consistent with the observation
that the Ad5 shaft is associated with transduction of DCs and
stimulation of adaptive immunity.
In this paper, we have evaluated the contribution of the
known receptors of adenoviral vectors as vaccines. Previous
studies have shown that the CAR-binding region of the
adenoviral knob and the integrin-binding motif RGD in the
penton base protein are responsible for targeting the
adenovirus to a variety of cell and tissue types in vivo
[7,12,24–26]. In skeletal muscle, the CAR?RGD?rAd was
previously shown to transduce 100-fold less efficiently
compared to a WT capsid vector . The present study
evaluates the contribution of these receptors to adenovirus
infectivity and immunogenicity. Mutations in the CAR- and
integrin-binding domains were not required for immunoge-
nicity, despite their significant effect on targeting of the virus
to many cell types including skeletal muscle. The CAR?RGD?
vector was clearly still competent for transduction, but the
cell types transduced following an intramuscular adminis-
tration must have been a subset of those transduced by the
WT capsid vector. It has been previously suggested that DCs
likely play a critical role in the ability of the adenovirus to
elicit immunity in vivo [27,28], but the mechanism by which
rAd5 targets these cells was unknown. It has been suggested
that immature cells are infected by the virus, leading to
differentiation to more mature DCs that may more effectively
present antigen in vivo [29–33]. In vitro, adenovirus can
induce maturation of BM DCs, and the penton RGD has been
shown to be involved in such stimulation [31,34]. Our data
suggest that adenovirus can utilize pathways other than the
RGD–integrin interaction to mature DCs in vivo, as deletion
of the penton RGD has no effect on immunogenicity of the
vector. An alternative proposal is that adenovirus targets
more mature DCs more effectively in vivo. Because of the
multiple cell-binding specificities of adenoviral vectors, the
relative contributions of these determinants to immunoge-
nicity and pyrogenic toxicity provide important information
relevant to vaccine design for gene-based and other modes of
antigen delivery in vivo. This study suggests that the shaft
contributes to the targeting of the adenovirus to DCs, which
likely mediate antigen presentation and enhance immune
reactivity in vivo. Within the shaft sequence is a repetitive
heparin-binding motif, KKTK, that may mediate this effect.
The shaft domain has also been implicated in Ad5 targeting
to hepatic cells and cytokine release when administered
through intravenous injection in mice [15,16]; however, its
relation to the pyrogenic response could not be defined in
this model. Although CAR?RGD?virus can bind to DCs and
gene transfer is blocked by heparin, but not heparan sulfate,
it is not clear that the KKTK domain alone mediates
interaction with DCs; nonetheless, it is clear that the shaft
domain is involved and that this interaction is dependent on
a heparin-like receptor interaction. This finding suggests that
modifications of the adenoviral fiber may allow design of
targeted adenovirus vectors modified to avoid toxicity and
reactogenicity. Fever that was observed as a component of a
serious adverse event in a human gene therapy trial  after
intraportal artery infusion of high doses of rAd5 could
possibly have been mediated by fiber interactions. Selective
modification of the shaft region, for example, by substitution
with the Ad35 shaft (which does not have putative heparin
sulfate proteoglycan-binding motifs), as shown here, may
assist in avoiding this complication. In addition, the immune
response to vector capsid proteins has limited the repetitive
use of adenoviral vectors for vaccine-induced immune
responses. The ability to define specific motifs within the
adenoviral fiber that facilitates DC binding and entry may
assist in the development of synthetic vectors that target
these cell types specifically. Interestingly, it has been difficult
to identify antibodies directed to the shaft region of
adenoviral vectors (Figure 4B), raising the possibility that
this motif may be protected from immune recognition and
could be retained in chimeric rAd from different serotypes or
with synthetic vectors. This knowledge may also assist in the
design of Ad vectors that could target DCs by other receptors
that may not lead to the pyrogenic response. For example,
DC-specific ligands, such as DC-SIGN or Langrin, can be
incorporated into the detargeted CAR?RGD?vector with the
shaft region from Ad35 that does not cause a pyrogenic
response, to build alternative DC-targeted adenovirus vec-
tors. These studies therefore lend insight into the mecha-
nisms of adenoviral immune targeting at the same time they
suggest possible means for targeting specific cells in vivo for
future gene-based vaccines.
Figure S1. Ability of WT or CAR?RGD?rAd Vectors to Infect
Alternative Cell Lines
The indicated WT and mutant fiber and penton rAd vectors
expressing GFP were evaluated as in Figure 1. Transduction of
murine NIH 3T3 fibroblasts (10,000 viral particles per cell), Hepa1–6
human hepatocellular carcinoma (500 viral particles per cell), MEL-
12 mouse erythroleukemia (2,000 viral particles per cell), and MM-14
mouse multiple myeloma cells (2,000 viral particles per cell) were
analyzed by flow cytometry.
Found at doi:10.1371/journal.ppat.0030025.sg001 (1.5 MB EPS).
Figure S2. Cellular Response to Irrelevant Peptide Pool Derived from
Ebola Viral Glycoprotein
Spleen cells were stimulated with peptide derived from Ebola viral
glycoprotein, and T-cell response was measured as shown in Figure 3.
Found at doi:10.1371/journal.ppat.0030025.sg002 (426 KB EPS).
Figure S3. Cellular Response to Irrelevant Peptide Pool Derived from
Ebola Viral Glycoprotein
Spleen cells were stimulated with peptide derived from Ebola viral
glycoprotein, and T-cell response was measured as shown in Figure
Found at doi:10.1371/journal.ppat.0030025.sg003 (388 KB EPS).
We thank Ati Tislerics for manuscript preparation, Brenda Hartman
and Toni Garrison for preparation of figures, members of the Nabel
PLoS Pathogens | www.plospathogens.orgFebruary 2007 | Volume 3 | Issue 2 | e250244
Ad5 Fiber Mediates DC Targeting/Toxicity
lab for discussions and comments, and Judy Stein and Charla
Andrews for advice and assistance regarding the rabbit toxicity
Author contributions. CC, JGDG, and WK performed virologic and
immunological studies in murine models and were involved in the
design, planning, data analysis, and writing of the paper. RLS and
PLG were involved in experimental design and data analysis. CRK
and GJN were involved in the design, planning, data analysis, and
writing of the paper.
Funding. This research was supported in part by the Intramural
Research Program of the National Institutes of Health, VRC, National
Institute of Allergy and Infectious Diseases.
Competing interests. The authors have submitted an intellectual
property application based on findings reported in this study.
1.Barouch DH, Nabel GJ (2005) Adenovirus vector-based vaccines for human
immunodeficiency virus type 1. Hum Gene Ther 16: 149–156.
2. Shiver JW, Emini EA (2004) Recent advances in the development of HIV-1
vaccines using replication-incompetent adenovirus vectors. Annu Rev Med
3. Tatsis N, Ertl HC (2004) Adenoviruses as vaccine vectors. Mol Ther 10: 616–
4. Muruve DA, Barnes MJ, Stillman IE, Libermann TA (1999) Adenoviral gene
therapy leads to rapid induction of multiple chemokines and acute
neutrophil-dependent hepatic injury in vivo. Hum Gene Ther 10: 965–976.
5. Raper SE, Yudkoff M, Chirmule N, Gao GP, Nunes F, et al. (2002) A pilot
study of in vivo liver-directed gene transfer with an adenoviral vector in
partial ornithine transcarbamylase deficiency. Hum Gene Ther 13: 163–
6.Raper SE, Chirmule N, Lee FS, Wivel NA, Bagg A, et al. (2003) Fatal systemic
inflammatory response syndrome in a ornithine transcarbamylase deficient
patient following adenoviral gene transfer. Mol Genet Metab 80: 148–158.
7. Xia D, Henry LJ, Gerard RD, Deisenhofer J (1994) Crystal structure of the
receptor-binding domain of adenovirus type 5 fiber protein at 1.7 A
resolution. Structure 2: 1259–1270.
8. Bergelson JM, Cunningham JA, Droguett G, Kurt-Jones EA, Krithivas A, et
al. (1997) Isolation of a common receptor for coxsackie B viruses and
adenoviruses 2 and 5. Science 275: 1320–1323.
9. Tomko RP, Xu R, Philipson L (1997) HCAR and MCAR: The human and
mouse cellular receptors for subgroup C adenoviruses and group B
coxsackieviruses. Proc Natl Acad Sci U S A 94: 3352–3356.
10. Roelvink PW, Mi LG, Einfeld DA, Kovesdi I, Wickham TJ (1999)
Identification of a conserved receptor-binding site on the fiber proteins
of CAR-recognizing adenoviridae. Science 286: 1568–1571.
11. Wickham TJ, Mathias P, Cheresh DA, Nemerow GR (1993) Integrins alpha v
beta 3 and alpha v beta 5 promote adenovirus internalization but not virus
attachment. Cell 73: 309–319.
12. Nemerow GR (2000) Cell receptors involved in adenovirus entry. Virology
13. Jakubczak JL, Rollence ML, Stewart DA, Jafari JD, Von Seggern DJ, et al.
(2001) Adenovirus type 5 viral particles pseudotyped with mutagenized
fiber proteins show diminished infectivity of coxsackie B-adenovirus
receptor-bearing cells. J Virol 75: 2972–2981.
14. Einfeld DA, Schroeder R, Roelvink PW, Lizonova A, King CR, et al. (2001)
Reducing the native tropism of adenovirus vectors requires removal of
both CAR and integrin interactions. J Virol 75: 11284–11291.
15. Shayakhmetov DM, Li ZY, Ni S, Lieber A (2004) Analysis of adenovirus
sequestration in the liver, transduction of hepatic cells, and innate toxicity
after injection of fiber-modified vectors. J Virol 78: 5368–5381.
16. Koizumi N, Kawabata K, Sakurai F, Watanabe Y, Hayakawa T, et al. (2006)
Modified adenoviral vectors ablated for coxsackievirus-adenovirus recep-
tor, alphav integrin, and heparan sulfate binding reduce in vivo tissue
transduction and toxicity. Hum Gene Ther 17: 264–79.
17. Mittereder N, March KL, Trapnell BC (1996) Evaluation of the concen-
tration and bioactivity of adenovirus vectors for gene therapy. J Virol 70:
18. Shayakhmetov DM, Eberly AM, Li ZY, Lieber A (2005) Deletion of penton
RGD motifs affects the efficiency of both the internalization and the
endosome escape of viral particles containing adenovirus serotype 5 or 35
fiber knobs. J Virol 79: 1053–1061.
19. Inaba K, Inaba M, Romani N, Aya H, Deguchi M, et al. (1992) Generation of
large numbers of dendritic cells from mouse bone marrow cultures
supplemented with granulocyte/macrophage colony-stimulating factor. J
Exp Med 176: 1693–1702.
20. Dechecchi MC, Tamanini A, Bonizzato A, Cabrini G (2000) Heparan sulfate
glycosaminoglycans are involved in adenovirus type 5 and 2-host cell
interactions. Virology 268: 382–390.
21. Dechecchi MC, Melotti P, Bonizzato A, Santacatterina M, Chilosi M, et al.
(2001) Heparan sulfate glycosaminoglycans are receptors sufficient to
mediate the initial binding of adenovirus types 2 and 5. J Virol 75: 8772–
22. Smith TA, Idamakanti N, Rollence ML, Marshall-Neff J, Kim J, et al. (2003)
Adenovirus serotype 5 fiber shaft influences in vivo gene transfer in mice.
Hum Gene Ther 14: 777–787.
23. Holers VM, Kinoshita T, Molina H (1992) The evolution of mouse and
human complement C3-binding proteins: Divergence of form but
conservation of function. Immunol Today 13: 231–236.
24. Stevenson SC, Rollence M, Marshall-Neff J, McClelland A (1997) Selective
targeting of human cells by a chimeric adenovirus vector containing a
modified fiber protein. J Virol 71: 4782–4790.
25. Bewley MC, Springer K, Zhang YB, Freimuth P, Flanagan JM (1999)
Structural analysis of the mechanism of adenovirus binding to its human
cellular receptor, CAR. Science 286: 1579–1583.
26. Kirby I, Davison E, Beavil AJ, Soh CP, Wickham TJ, et al. (1999) Mutations
in the DG loop of adenovirus type 5 fiber knob protein abolish high-affinity
binding to its cellular receptor CAR. J Virol 73: 9508–9514.
27. Basak SK, Kiertscher SM, Harui A, Roth MD (2004) Modifying adenoviral
vectors for use as gene-based cancer vaccines. Viral Immunol 17: 182–196.
28. Worgall S, Busch A, Rivara M, Bonnyay D, Leopold PL, et al. (2004)
Modification to the capsid of the adenovirus vector that enhances dendritic
cell infection and transgene-specific cellular immune responses. J Virol 78:
29. Rea D, Schagen FH, Hoeben RC, Mehtali M, Havenga MJ, et al. (1999)
Adenoviruses activate human dendritic cells without polarization toward a
T-helper type 1-inducing subset. J Virol 73: 10245–10253.
30. Dietz AB, Vuk-Pavlovic S (1998) High efficiency adenovirus-mediated gene
transfer to human dendritic cells. Blood 91: 392–398.
31. Philpott NJ, Nociari M, Elkon KB, Falck-Pedersen E (2004) Adenovirus-
induced maturation of dendritic cells through a PI3 kinase-mediated TNF-
alpha induction pathway. Proc Natl Acad Sci U S A 101: 6200–6205.
32. Miller G, Lahrs S, Shah AB, DeMatteo RP (2003) Optimization of dendritic
cell maturation and gene transfer by recombinant adenovirus. Cancer
Immunol Immunother 52: 347–358.
33. Tan PH, Beutelspacher SC, Xue SA, Wang YH, Mitchell P, et al. (2005)
Modulation of human dendritic cell function following transduction with
viral vectors: Implications for gene therapy. Blood 105: 3824–3832.
34. Hensley SE, Giles-Davis W, McCoy KC, Weninger W, Ertl HC (2005)
Dendritic cell maturation, but not CD8þT cell induction, is dependent on
type I IFN signaling during vaccination with adenovirus vectors. J Immunol
PLoS Pathogens | www.plospathogens.orgFebruary 2007 | Volume 3 | Issue 2 | e250245
Ad5 Fiber Mediates DC Targeting/Toxicity