Immune responses to adenovirus and adeno-associated virus in humans

Abstract

Vectors based on human adenovirus (Ad) and adeno-associated virus (AAV) are being evaluated for human gene therapy. The response of the host to the vector, in terms of antigen-specific immunity, will play a substantial role in clinical outcome. We have surveyed cohorts of normal subjects and cystic fibrosis patients for pre-existing immunity to these viruses, caused by naturally acquired infections. A number of humoral and cellular assays to adenovirus serotype 5 (Ad5) and adeno-associated virus serotype 2 (AAV2) were performed from serum and peripheral blood mononuclear cells. Virtually all subjects had Ig to Ad5 although only 55% of these antibodies neutralized virus (NAB). Approximately two of three patients demonstrated CD4+ T cells that proliferated to Ad antigens of which most were of the TH1 subset, based on cytokine secretion. A substantially different pattern of immune responses was observed to AAV2. Although virtually all patients had Ig to AAV2, most of these antibodies were not neutralizing (32% NAB) and only 5% of patients had peripheral blood lymphocytes that proliferated in response to AAV2 antigens. These studies demonstrate marked heterogeneity in pre-existing immunity to Ad5 and AAV2 in human populations. The impact of these findings on outcome following gene therapy will require further study.

Full text

Gene Therapy (1999) 6, 1574–1583
 1999 Stockton Press All rights reserved 0969-7128/99 $15.00
http://www.stockton-press.co.uk/gt
Immune responses to adenovirus and adeno-
associated virus in humans
N Chirmule
1
, KJ Propert
2
, SA Magosin
1
, Y Qian
1
, R Qian
1
and JM Wilson
1,3
1
Institute for Human Gene Therapy and Departments of Molecular and Cellular Engineering, and Medicine;
2
Department of
Biostatistics and Epidemiology, University of Pennsylvania; and
3
Wistar Institute, Philadelphia, PA, USA
Vectors based on human adenovirus (Ad) and adeno- virus (NAB). Approximately two of three patients demon-
associated virus (AAV) are being evaluated for human strated CD4
+
T cells that proliferated to Ad antigens of
gene therapy. The response of the host to the vector, in which most were of the TH
1
subset, based on cytokine
terms of antigen-specific immunity, will play a substantial secretion. A substantially different pattern of immune
role in clinical outcome. We have surveyed cohorts of nor- responses was observed to AAV2. Although virtually all
mal subjects and cystic fibrosis patients for pre-existing patients had Ig to AAV2, most of these antibodies were
immunity to these viruses, caused by naturally acquired not neutralizing (32% NAB) and only 5% of patients had
infections. A number of humoral and cellular assays to peripheral blood lymphocytes that proliferated in response
adenovirus serotype 5 (Ad5) and adeno-associated virus to AAV2 antigens. These studies demonstrate marked het-
serotype 2 (AAV2) were performed from serum and periph- erogeneity in pre-existing immunity to Ad5 and AAV2 in
eral blood mononuclear cells. Virtually all subjects had Ig human populations. The impact of these findings on out-
to Ad5 although only 55% of these antibodies neutralized come following gene therapy will require further study.
Keywords:
adenovirus; AAV; neutralizing antibodies; lymphoproliferation; cytokines; humans
Introduction
Initial paradigms of ex vivo gene therapy, based on trans-
plantation of genetically modified cells, have given way
to in vivo gene therapy where the vector is directly intro-
duced into the recipient. In vivo gene therapy has the
potential of improved efficiency and expands the scope
of applications to a wider array of acquired and inherited
diseases.
1
An example is cystic fibrosis (CF) where a
number of viral and nonviral vectors have been intro-
duced into the airway to correct the genetic defect in res-
piratory epithelial cells. Vectors based on recombinant
forms of human adenoviruses (Ad) and adeno-associated
viruses (AAV) have played a major role in in vivo gene
therapy for CF and a number of other disorders.
1–5
A
number of factors will contribute to the outcome of in
vivo gene therapy with Ad and AAV, including the nat-
ure of the immune response to vector, transgene product
and transduced cell. Genetic factors, such as the genotype
(eg mutation of CF gene) and MHC haplotype, will play
a role. Pre-existing immunity to the virus due to naturally
acquired infections will also contribute to outcome.
6
Human Ad vectors have been created by replacing the
immediate early genes E1a and E1b with a therapeutic
minigene. There exist over 40 serotypes of human adeno-
virus, although most vectors are based on serotypes 2
and 5.
7
Administration of replication-defective Ad2 and
Ad5 vectors expressing the normal CF gene into the air-
Correspondence: N Chirmule, Institute for Human Gene Therapy, Univer-
sity of Pennsylvania Health System, 304 Stellar-Chance Laboratories, 422
Curie Boulevard, Philadelphia, PA, 19104-6100, USA
Received 3 December 1998; accepted 11 May 1999
way has demonstrated gene transfer and partial correc-
tion of function in preclinical and clinical models.
8–10
A
number of problems attributable to immune responses
have emerged. B cells and T helper cells are activated to
the viral capsid proteins, which leads to the secretion of
neutralizing antibody (NAB) which blocks readminis-
tration of vector. Furthermore, expressed viral genes and
the transgene product are presented by MHC class I
resulting in activation of cytotoxic T cells that extinguish
transgene expression.
6,11
The majority of humans have circulating antibodies
against the common Ad serotypes and infected individ-
uals develop prolonged immunity to the virus.
7
Multiple
factors may contribute to the generation of Ad-specific
immune responses in humans.
12
About 2–5% of respir-
atory infections, 10% of febrile illnesses, and 5–10% of
gastrointestinal infections in children are due to adenovi-
ral infections leading to shedding of virus for weeks and
even years following infections. Lymphoid organs, such
as the tonsils, are probably the site for persistent infec-
tions, where viral DNA has been detected in 25% of sub-
jects. Acute infection with Ad leads to immunity in most
cases. Military recruits administered Ad7 and Ad4 live
vaccines seroconvert and are protected from the disease.
Ad has been associated with fatal disseminated disease in
individuals with primary immune deficiencies, indicating
that cellular immune responses are important for recov-
ery from acute infections.
12,13
Adeno-associated virus is a human parvovirus with six
distinct serotypes that is not associated with disease in
primates. Wild-type AAV can infect non-dividing per-
missive cells and integrate as a proviral genome into a
specific site in chromosome 19. Over 90% of humans

Immune responses to Ad and AAV in humans
N Chirmule
et al
1575
demonstrate antibodies that cross-react to one or more
AAV serotypes.
14–16
Replication-defective forms of AAV
serotype 2 (AAV2) have been generated in which all viral
open reading frames are deleted from the vector and
replaced with a therapeutic gene. Recombinant AAV is
capable of efficient transduction (with integration) in a
number of cells in vivo. Intramuscular injection of AAV
results in transduction of muscle fibers without activating
destructive T cell responses even with transgenes that
encode foreign proteins.
17–19
Parenteral administration of
AAV will elicit neutralizing antibodies that block admin-
istration.
In this study we performed a comprehensive survey of
in vitro measures of B and T cell activation to Ad5 and
AAV2 in normal subjects and patients with CF.
Results
Cohorts of normal subjects (n = 45) and CF patients (n =
29) were analyzed for B and T cell responses to Ad5 and
AAV2 (Table 1). Serum was evaluated for neutralizing
antibody (NAB), Ig isotypes by ELISA and specific bind-
ing to capsid proteins by Western blots. Peripheral blood
mononuclear cells were evaluated for proliferation to
virus (lymphoproliferative response (LPR)) and secretion
of cytokines IL-2, IFN-␥, IL-10 and IL-4. The results
within the CF patients and normal controls for each of
the measures are shown in Table 2.
Immune response to adenovirus
ELISA of serum showed Ig to adenovirus defined as in
71/74 (96%) individuals with IgG consistently greater
than IgM or IgA (Figure 1a and b). There were no differ-
ences between normal subjects and CF patients with
respect to IgM or IgA, although significantly higher levels
of Ad-specific IgG were noted in normal subjects than CF
(Table 2). A wide spectrum of NAB to Ad was measured
(Figure 1c) with 41/74 (55%) showing titers greater than
background level of 1:20.
Western blot analyses were performed to evaluate the
distribution of antibodies against capsid proteins. Figure
2 presents representative Western blots for 67 of 74
patients. All patients except for three had antibodies to
one of the capsid proteins although the relative binding
to different proteins within patients varied as follows:
Table 1 Population demographics
CF Controls Total
No. % No. % No. %
Gender 29 45 74
Male 17 59 13 29 30 41
Female 12 41 32 71 44 59
Race
White 27 93 31 69 58 78
Black 2 7 12 27 14 19
Other 0 0 2 4 2 3
Age (years)
Median 25 29 27
Range 18–52 18–54 18–54
37/67 (55%) had antibodies to hexon, 58/67 (87%) had
antibodies to penton, and 62/67 (92%) had antibodies to
fiber. A significant association was observed between
NAB and antibodies to penton (P = 0.0042) with no
association to hexon or fiber (Table 3). There were no stat-
istically significant differences in NAB or Western pro-
files between normal subjects and CF (data not shown).
LPR responses to Ad antigens were used to evaluate
in vitro measures of activated CD4
+
T cells. The validity
of this in vitro correlate was tested in experiments
presented in Figure 3a. Incubation of peripheral blood
mononuclear cells with inactivated Ad resulted in sub-
stantial proliferation in a number of individuals, as meas-
ured by
3
H-thymidine uptake. The data are presented as
the percentage inhibition of LPR. LPR responses were
completely abrogated with blocking CD4 mAb. Similar
inhibition of proliferation was seen in the presence of
mAb to MHC class II (ie HLA-DR) but not MHC class
I, consistent with the responses reflecting CD4
+
T cells.
Blocking antibody to CD40L (ie 5C8) and B7 (data not
shown) also inhibited proliferation confirming the impor-
tance of these costimulatory pathways.
The majority of individuals (47/74, 64%) had evidence
of activated CD4
+
T cells as detected by LPR stimulation
index ⬎2.0 (Figure 3b). The specific subset of T helper
cells activated to Ad was evaluated by measuring
secretion of cytokines during stimulation (TH
1
cytokines
(IL-2 and IFN-␥) and TH
2
cytokines (IL-4 and IL-10)). Fig-
ure 4 shows the cytokine analysis for normal subjects and
CF. There was substantial patient-to-patient variation in
the secretion of each cytokine with a number of individ-
uals not responding to antigens. IFN-␥ was secreted most
consistently and at the highest levels with remarkably
little IL-4 with IL-2 and IL-10 secreted at intermediate
levels.
Relationships between in vitro measures of humoral
and cellular immunity were evaluated using a number of
statistical analyses, combining the normal subjects and
CF patients (Figure 5 and Tables 3 and 4). A moderate but
significant association (P = 0.016) exists between CD4
+
T
cell activation (LPR) and B cell activation as measured by
NAB (Table 3); the patients with positive LPR responses
had the positive NAB. Cytokine analysis of lymphocytes
from each subject revealed frequent secretion of IFN-␥
with little IL-4 suggestive of a dominant TH
1
response
(Figure 4). There was a highly significant association
between LPR and IL-2 and IFN-␥ secretion confirming
the importance of the TH
1
phenotype (P ⬍ 0.0001, Figure
5 and Table 4). NAB to Ad5 was not associated with the
TH
2
cytokines IL-4 and IL-10 (Table 3).
Immune responses to AAV
AAV-specific antibodies were detected by ELISA in 96%
of subjects (71/74, Figure 6a and b) although only 32%
showed neutralizing antibodies (24/74, Figure 6c). No
specific difference in humoral response between the two
groups was noted except that AAV-specific IgM and IgA
were higher in CF patients than normal subjects (Table
2, P ⬍ 0.0001). Western blot analysis showed antibodies
to each capsid protein (ie VP-1, VP-2, and VP-3) in all
serologically positive patients (Figure 2).
Peripheral blood mononuclear cells were analyzed for
proliferation to AAV by measuring
3
H-thymidine uptake
(Figure 6d). A minority of subjects demonstrated LPR ⬎
2.0, (3/57, 6%); exogenous IL-2 had no effect on these in

Immune responses to Ad and AAV in humans
N Chirmule
et al
1576
Table 2 Comparison of cell immune assays between normal subjects and CF
Measure Unit CF Non-CF P value
n Mean n Mean
Serum immunoglobulins
Ad-IgM OD 29 0.21 ± 0.13 45 0.29 ± 0.20 0.056
Ad-IgG OD 29 1.08 ± 0.67 45 2.20 ± 0.86 ⬍0.0001
Ad-IgA OD 29 0.18 ± 0.11 45 0.19 ± 0.14 0.89
AAV-IgM OD 29 0.36 ± 0.16 45 0.20 ± 0.13 ⬍0.0001
AAV-IgG OD 29 1.16 ± 0.66 45 1.13 ± 0.93 0.87
AAV-IgA OD 29 0.14 ± 0.14 45 0.03 ± 0.05 ⬍0.0001
Lymphoproliferative responses
Adenovirus SI 29 10.60 ± 19.10 45 15.80 ± 23.60 0.32
AAV SI 22 1.09 ± 0.41 30 1.84 ± 3.06 0.26
P aeruginosa SI 24 13.50 ± 15.50 41 9.64 ± 20.70 0.43
PHA SI 29 159 ± 151 45 284 ± 291 0.036
Cytokines
IFNg pg/ml 26 526 ± 750 35 579 ± 705 0.78
IL-2 pg/ml 26 77.10 ± 104 35 88.00 ± 150 0.75
IL-10 pg/ml 26 34.80 ± 52.5 35 49.30 ± 58.8 0.33
IL-4 pg/ml 26 13.90 ± 42.3 35 10.40 ± 31.4 0.71
Various immunological analyses were performed on peripheral blood of CF patients and normal healthy individuals (controls). n denotes
the number of samples analyzed; the mean values from all the samples in the group are presented. The comparison of each test between
CF patients and normal controls were done using two-sample t tests.
Figure 1 Serum adenovirus immunoglobulins in humans. Sera from humans (a, normal; b, CF) were analyzed for the presence of adenovirus-specific
IgM, IgG and IgA by ELISA as described in the Materials and methods. Each point denotes a single individual and the bar represents the mean. (c)
Serial dilutions of serum were evaluated for their ability to block the infection of Hela cells by a LacZ expressing adenovirus. Antibody titers are expressed
as the dilution of serum that result in a 50% reduction of Ad infection in HeLa cells without the addition of serum. This panel shows the distribution
of the individuals with different NAB titers to adenovirus vector.

Immune responses to Ad and AAV in humans
N Chirmule
et al
1577
Figure 2 Western blot analyses to adenovirus and AAV. Sera from donors were used to probe Western blots. Adenovirus (top panel) and AAV (bottom
panel) protein preparation were separated on a 10% acrylamide gel, and subjected to Western blot analyses. Serum from a rabbit immunized with
adenovirus was used as a positive control for the top panel while a monoclonal antibody that detects all AAV2 capsid proteins was used as a control
in the bottom panel. Positions of hexon, penton and fiber were determined by using antibodies to peptides of these proteins as controls (top panel). AAV
capsid proteins were identified based on relative molecular weight (bottom panel).
vitro assays (data not shown). Supernatants from stimu-
lated lymphocytes were largely negative for cytokines,
consistent with the LPR assays (data not shown). IFN-␥
was detected in 3/57 (6%) of patients and IL-10 in 7/57
(12%) of patients; IL-2 and IL-4 were not detected (data
not shown).
Discussion
The impetus for this study was the realization that host
immune responses will play a significant role in the out-
come of in vivo gene therapy. Experiments with adenovi-
ral vectors clearly demonstrate the development of neu-
tralizing antibodies that block readministration, and
cellular responses that extinguish gene expression.
Humoral responses are also relevant to problems of AAV
readministration. Strategies to prevent these confounding
responses include transient immune blockade to prevent
CD4
+
T cell activation and re-engineered vectors to dim-
inish CD8
+
T cell responses.
20–23
Most of this work has
been performed in animal models who are naive to the
virus. This will not be the case in humans, many of whom
have been exposed to Ad or AAV due to a naturally
acquired infection. Our survey of normal subjects and CF
patients revealed interesting findings of pre-existing
immunity to these viruses that are of potential relevance
to the outcome of in vivo gene therapy.
Cellular immune responses to Ad were studied using
lymphoproliferative responses to capsid proteins. This
assay detected significant proliferation of T cells to anti-
gen in 64% of individuals, which is lower than what was
observed in a previous study where all subjects
responded.
24,25
Further study confirmed that Ad-specific
responses are MHC class II restricted and mediated by
CD4
+
T cells, findings consistent with previous studies
using wild-type Ad antigens. We further showed the
requirement of CD40L–CD40 and B7–CD28 interactions
in CD4
+
T cell activation. Different levels of responsive-
ness to Ad could reflect variations in prior exposure to a
naturally acquired infection or HLA heterogeneity.
The qualitative nature of T cell responses in terms of
T helper subsets was evaluated by measuring secreted
cytokines in response to Ad. IFN-␥ and IL-2 were used
as markers of a TH
1
response while IL-10 and IL-4

Immune responses to Ad and AAV in humans
N Chirmule
et al
1578
Table 3 Relationship of various immunologic parameters to neutralizing antibodies to adenovirus
Measure Value Neutralizing antibody Total P value
Negative Positive
n% n% n%
LPR Negative 17 52 10 24 27 36 0.016
Positive 16 48 31 76 47 64
Hexon Negative 14 47 16 43 30 45 0.780
Positive 16 53 21 57 37 55
Penton Negative 8 27 1 3 9 13 0.0042
Positive 22 73 36 97 58 87
Fiber Negative 4 13 1 3 5 8 0.100
Positive 26 87 36 97 62 92
IL-4 Negative 21 81 25 71 46 75 0.400
Positive 5 19 10 29 15 25
IL-10 Negative 9 35 9 26 18 30 0.450
Positive 17 65 26 74 43 70
Associations for various parameters (column 1) with positive or negative neutralizing antibodies to adenovirus (defined as ⬍20). P
values for association between NAB and other measures are based on a ␹
2
text on one degree of freedom.
Figure 3 Characterization of adenovirus-vector-mediated LPR responses in normal and CF individuals. (a) Peripheral blood lymphocytes from five
individuals from a separate cohort, who responded to adenovirus antigens, were cultured with inactivated Ad vector (1 × 10
9
particles/ml) in the presence
of various concentrations of antibodies directed to HLA-DR (anti-HLA-DR mAb), CD4 (OKT4A mAb), CD40L (5C8), or MHC class I (W6–32 mAb).
Lymphoproliferation was measured by
3
H-thymidine incorporation and presented as stimulation index of c.p.m. from adenovirus stimulated cultures
per c.p.m. in the presence of medium alone. A stimulation index of ⬎2.0 was considered positive. The y-axis shows percent inhibition of LPR responses
by the antibodies, compared with control responses in the absence of any antibody. Each point denotes mean and s.d. of five separate donors. (b) LPR
measurements from normal subjects and CF (P = 0.32, two-tailed t test).
reflected TH
2
response. We did this with the disclaimer
that the TH
1
/TH
2
dichotomy developed in mice may not
precisely reflect the situation in humans.
26
Our cohort of
subjects did show a significant association between LPR
responses and secretion of IL-2 and IFN-␥. Induction of
IL-4 was observed only in some individuals; no signifi-
cant correlation was observed with LPR responses. This
was surprising in light of the more extensive prevalence
of memory B cells as suggested by NAB in 55% of
patients. It is possible that other cytokines or T helper
subsets contribute to B cell activation in this system.
Alternatively, this may reflect limitations of the IL-4
ELISA.
No clear associations of humoral and cellular immu-
nity were noted with Ad-stimulated secretion of IL-10.
We focused specifically on any relationship with LPR
because of recent work, which implicated IL-10 on inhi-
bition of T cell functional responses.
27
IL-10 was found
across a spectrum of LPRs from stimulation index of 1 to
100. Further analysis of non-responding IL-10-secreting
cells indicated they express IFN-␥ but not IL-2. Such cells
have been recently classified as TH
3
cells, which play a
role in regulation of immune reactivity to tolerant anti-
gens in infectious agents.
28,29
Antibodies reactive to Ad capsid proteins were present
in 97% of individuals based on Western and ELISA

Immune responses to Ad and AAV in humans
N Chirmule
et al
1579
Figure 4 Cytokine secretion patterns induced by adenovirus. Peripheral blood lymphocytes were cultured in the presence or absence of inactivated Ad
for 48 h and culture supernatants assayed for the presence of cytokines (IFN-
␥
(a), IL-2 (b), IL-4 (c) and IL-10 (d)) by commercial ELISA kits. Each
point denotes a single individual. The data are presented for both normal subjects and CF as concentration of cytokine (pg/ml).
assays. These were primarily IgG, which recognized pen-
ton, hexon and fiber at variable levels. Interestingly,
serum from only 55% of subjects actually neutralized Ad
infection in vitro. Why the discrepancy between seroposi-
tivity, based on Western and ELISA, and neutralization
of virus? Each individual likely has a different history of
infections with more than one of the over 40 Ad sero-
types. The circulating antibodies resulting from these
memory B cell responses will bind with varying
efficiencies to Ad serotypes within and between aden-
ovirus groups. The Western and ELISA assays are prob-
ably more sensitive in detecting antibodies across all ser-
otypes, while the neutralizing assay is more stringent in
terms of serospecific responses. An important question is
which in vitro assay will predict interference of gene
transfer in vivo? In this respect, a recent study has demon-
strated that pre-existing humoral immune responses did
not preclude gene transfer.
30
Our strategy to date in
human trials is to include all patients in trials, irrespec-
tive of in vitro measures of pre-existing immunity, and
to develop correlations between the in vitro assays and
efficiency of in vivo gene therapy.
Profiles of immune activation to AAV were quite dif-
ferent than what was observed to Ad. ELISA and West-
ern did detect antibodies in 96% of individuals, with only
32% capable of neutralizing an in vitro AAV infection.
The high prevalence of seropositivity to AAV is consist-
ent with previous epidemiologic studies.
1
The three sub-
jects who had no evidence of immunity to Ad were also
negative for AAV responses confirming the symbiotic
relationship of these viruses in humans. Interestingly,
only 5% of subjects showed proliferative responses to
AAV antigens. It is unclear why the substantial discor-
dance between CD4
+
T cell responses and memory B cell
activity. It is possible that a component of AAV humoral
response is T cell-independent.
The implications of our studies on AAV immunity in
humans is unclear. Some humans clearly have significant
levels of neutralizing antibody to AAV2 which could
impact on the efficiency of in vivo gene engraftment.
Prevalence of NAB to AAV2 is lower than NAB to Ad5,
however. The nature of the B cell response to AAV2
needs clarification to help design strategies to prevent it
from occurring in therapies that require readministration
of vector.
Materials and methods
Human subjects
Normal volunteers (n = 45), and CF patients (n = 29) were
analyzed for immune reactivity to adenovirus and AAV.

Immune responses to Ad and AAV in humans
N Chirmule
et al
1580
Figure 5 Correlation between IFN-
␥
, IL-2, IL-4 and IL-10 secretion with LPR responses to adenovirus. Stimulation index (x-axis) was compared with
the adenovirus-induced IFN-
␥
(a), IL-2 (b), IL-4 (c) and IL-10 (d) secretion (y-axis, pg/ml). The Pearson correlation coefficients for panel a = 0.48; panel
b = 0.65; panel c = 0.10; panel d =−0.01.
Table 4 Relationship of various immunologic parameters to adenovirus LPR
Measure Value LPR Total P value
Negative Positive
n% n% n%
IFN␥ Negative 6 35 0 0 6 10 ⬍0.0001
Positive 11 65 44 100 55 90
IL-2 Negative 13 77 7 16 20 33 ⬍0.0001
Positive 4 23 37 84 41 67
IL-4 Negative 15 88 31 72 46 75 0.15
Positive 2 12 13 28 15 25
IL-10 Negative 6 35 12 27 18 30 0.54
Positive 11 65 32 73 43 70
Some assays were not done in all the samples due to low
cell numbers, lack of samples, or invalid assay conditions.
The numbers of individuals tested for each assay are
specified in Table 2. Individuals were derived from the
local CF clinic and members of the University of Pennsyl-
vania community. The study was approved by the Insti-
tutional Review Board. Demographics of the individuals
in the study are shown in Table 1. The median age was
27 years (range 18 to 54 years) and the age distribution
was similar in the two groups. An additional 10 individ-
uals were analyzed for adenovirus-specific lymphopro-
liferative responses. All these samples were pretreated

Immune responses to Ad and AAV in humans
N Chirmule
et al
1581
Figure 6 Humoral and cellular immune responses to AAV. Sera from normal subjects (a) and CF patients (b) were analyzed for the presence of AAV-
specific IgM, IgG and IgA by ELISA as described in Materials and methods. Each point denotes a single individual. (c) Serial dilutions of serum were
evaluated for their ability to block the infection of 84–31 cells by LacZ expressing AAV. Antibody titers are expressed as the dilution of serum that
result in a 50% reduction of infection of AAV infected 84–31 cells without the addition of serum. The Figure shows the distribution of the individuals
with different NAB titers to adenovirus vector. (d) Peripheral blood lymphocytes were cultured in the absence or presence of AAV for 7 days. Lymphopro-
liferation was measured by
3
H-thymidine incorporation and presented as stimulation index of c.p.m. from adenovirus stimulated cultures per c.p.m. in
the presence of medium alone. A stimulation index of ⬎2.0 was considered positive.
with either medium alone, or various antibodies. Five of
10 individuals responded strongly in this assay, and were
used for the data in Figure 3a.
Recombinant viruses
The structure, production, and measurement of titer of
E1-deleted recombinant adenovirus virus based on sero-
type 5 (H5.010CMVlacZ, hence called Ad-lacZ) that
express ␤-galactosidase (CMV promoter in sub360
backbone) have been described previously. Recombinant
adeno-associated virus (AAV2lacZ, hence called AAV),
deleted of rep and cap genes was generated by plasmid
transfections in 293 cells infected with E1-deleted adeno-
virus as described earlier.
Lymphoproliferative assays
Peripheral blood mononuclear cells (PBMC) were har-
vested from 40 ml of heparinized blood and isolated fol-
lowing Ficoll–Hypaque density gradient centrifugation,
washed in PBS and resuspended in RPMI 1640 and sup-
plemented with 10% FCS, penicillin, streptomycin, and
10
−5
m 2-mercaptoethanol. Triplicate cultures PBMC (100
␮lof1× 10
6
cells/ml) were cultured with either inacti-
vated Ad-lacZ (MOI in particles = 10) or rAAV (MOI in
particles = 100) supplemented with 0.5 ␮g/ml phytohem-
agglutinin (PHA, Difco, Detroit, MI, USA) or medium
alone. PHA-stimulated cultures were harvested on day 3
and antigen-stimulated cultures were harvested on day
6. In some cases, exogenous recombinant IL-2
(Immunotech IC, Westbrook, ME, USA) or antibodies to
CD4 (Leu3a mAb); HLADR, MHC class I (HLA-A,B-C,
Becton Dickinson, Mountain View, CA, USA), CD40L
(5C8, Biogen, Boston, MA, USA), or B7–1 (Pharmingen,
San Diego, CA, USA) were added to the cultures. Pro-
liferation was measured by a 16-h
3
H-thymidine (1 ␮Ci
per well) pulse on a Wallach liquid scintillation counter
(Gaithersburg, MD, USA). Results are presented as
stimulation indexes, which denote the ratio of c.p.m. of
stimulation cultures to c.p.m. of unstimulated cultures.
Cytokine release assays
PBMC were cultured with or without antigen (ie inacti-
vated Ad or AAV) for 48 h in a 24-well plate. Cell free
supernatants were collected and analyzed for presence
of IL-2, IL-4, IFN␥ and IL-10 by commercial ELISA kits
(BioSource, Rahway, NJ, USA) using manufacturer’s
protocols.
Vector-specific immunoglobulins
Serum (diluted 1:200) samples from donors were ana-
lyzed for Ad- and AAV-specific isotype specific immuno-
globulins (IgM, IgG, IgA) by ELISA. For the ELISA, 96-

Immune responses to Ad and AAV in humans
N Chirmule
et al
1582
well flat bottomed, high-binding ELISA plates (Costar,
Cambridge, MA, USA) were coated with 100 ␮l Ad-anti-
gen (5 × 10
9
particles/ml) or AAV (1 × 10
9
particles/ml)
in PBS overnight at 4°C, washed four times in PBS/0.05%
Tween and blocked in PBS/1% BSA for 1 h at 37°C.
Appropriately diluted samples were added to antigen-
coated plates and incubated for 4 h at 37°C. Plates were
washed four times in PBS/0.05% Tween and incubated
with peroxidase conjugated goat anti-human IgM, IgG or
IgA (1:2000 dilution, Sigma Chemical, St Louis, MO,
USA) for 2 h at 37°C. Plates were washed as above and
ABST substrate (Kirkegaard and Perry, Gaithersburg,
MD, USA) was added. Optical densities were read at 405
nm on a MRX microplate reader (Dynatech Laboratories,
Chantilly, VA, USA).
Western blot analyses
Serum samples were analyzed for reactivity to various
Ad proteins by Western blot as described earlier. Whole
virus particles were boiled in sample buffer and used as
a source of adenovirus antigens from Ad and AAV. A
polyclonal rabbit antiserum (for Ad) and monoclonal
antibody that recognizes shared epitopes on VP1, VP2,
VP3 (clone B1, American Research Products, Piscataway,
NJ, USA) as a positive control. The antibodies to hexon,
penton and fiber were identified by probing the nitrocel-
lulose blots with polyclonal antibodies generated against
hexon, penton and fiber specific peptides. Briefly, Ad
antigens were electrophoresed on a 10% SDS-PAGE and
transferred to nitrocellulose membrane (Hybond-ECL,
Amersham, Pistcataway, NJ, USA). The reactivity of the
serum was measured by incubating test serum, followed
by peroxidase anti-human IgG antibody. The reaction
was detected by the ECL chemiluminiscence kit
(Amersham).
Anti-adenovirus and AAV neutralizing antibodies
Neutralizing antibody titers were analyzed by assessing
the ability of serum antibody to inhibit transduction of
reporter virus, Ad-lacZ, or AAVlacZ into Hela or 84–31
cells, respectively. The 84–31 line is a subclone of 293 cells
that stably expresses E4 of Ad that renders it permissive
for AAV transduction. Various dilutions of antibodies
pre-incubated with reporter virus for 1 h at 37°C, were
added to 90% confluent cell cultures. Cells were incu-
bated for 16 h and expression of lacZ, was measured by
staining with X-gal substrate. The neutralizing titer of
antibody was calculated by the highest dilution with
which 50% of the cells stained blue.
Statistical methods
Continuous measures were compared between groups
using two-sample t tests. Chi-square tests were used to
evaluate measures of association between dichotomous
outcomes. Pearson correlation coefficients were used to
evaluate linear relationships among continuous outcome
variables. For some analyses and outcomes, measures
were further analyzed by first dichotomizing into ‘nega-
tives’ and ‘positives’ according to standard cut-offs. The
cut-off values used for these outcomes were immuno-
globulins 0.050; LPR stimulation index 2.0; cytokines 0.0;
neutralizing antibodies 20. Analyses utilized the S-Plus
and StatXact statistical analyses packages.
Acknowledgements
The cooperation of the human subjects involved in this
study was greatly appreciated. We thank the Vector and
Immunology Cores of the Institute for Human Gene
Therapy for their help, and Biogen for providing the 5C8
antibody. This work was supported by the CF Foun-
dation and NIH (P50DK49136 and P30DK47757) and
Genovo, Inc, a biotechnology company, which Dr J
Wilson founded and in which he holds equity.
References
1 Wilson J. Gene therapy for cystic fibrosis: challenges and future
directions. J Clin Invest 1995; 96: 2547–2554.
2 Wilson JM. Adenoviruses as gene-delivery vehicles. New Engl J
Med 1996; 334: 1185–1187.
3 Crystal R. Transfer of genes to humans: early lessons and
obstacles to success. Science 1995; 270: 404–410.
4 Flotte TR, Carter BJ. Adeno-associated virus vectors for gene
therapy. Gene Therapy 1995; 2: 357–362.
5 Koeberl DD et al. Persistent expression of human clotting factor
IX from mouse liver after intravenous injection of adeno-asso-
ciated virus vectors. Proc Natl Acad Sci USA 1997; 94: 1426–1431.
6 Bromberg JS, Debruyne LA, Qin L. Interactions between the
immune system and gene therapy vectors: bidirectional regu-
lation of response and expression. Adv Immunol 1998; 69: 353–
409.
7 Schmitz H, Wigand R, Henrich W. Worldwide epidemiology of
human adenovirus infections. Am J Epidemiol 1983; 117: 455–466.
8 Crystal RG et al. Administration of an adenovirus containing the
human CFTR cDNA to the respiratory tract of individuals with
cystic fibrosis. Nat Genet 1994; 8: 42–51.
9 Knowles M et al. A double-blind vehicle controlled study of
adenoviral vector-mediated gene transfer in the nasal epi-
thelium of patients with cystic fibrosis. New Engl J Med 1995;
333: 823–831.
10 Rosenfeld MA et al. In vivo transfer of the human cystic fibrosis
transmembrane conductance regulator gene to the airway epi-
thelium. Cell 1992; 68: 143–155.
11 Gahery-Segard H et al. Phase I trial of recombinant adenovirus
gene transfer in lung cancer: longitudinal study of the immune
responses to transgene and viral products. J Clin Invest 1997;
100: 2218–2226.
12 Wold SM, Tollefson AE, Hermiston TW. Strategies of immune
modulation by adenoviruses. In: G McFadden (ed). Viroceptors,
Virokines and Related Immune Modulators Encoded by DAN Viruses.
RG Landes: 1994, pp 147–185.
13 Takafuji ET, Gaydos JC, Allen RG, Top FH. Simultaneous
administration of live, enteric coated adenovirus 4, 7 and 21 vac-
cines: safety and immunogenecity. J Infect Dis 1979; 140: 48–53.
14 Atchinson R. Adenovirus-associated defective virus particles.
Science 1965; 149: 754–756.
15 McKeon C, Ramulski RJ. NIDDK workshop on AAV vectors:
gene transfer into quiescent cells. Hum Gene Ther 1996; 7:
1615–1619.
16 Kotin RM. Prospects for the use of adeno-associated virus as a
vector for human gene therapy. Hum Gene Ther 1994; 5: 793–801.
17 Xiao X, Li J, Samulski RJ. Efficient long-term gene transfer into
muscle tissue of immunocompetent mice by adeno-associated
virus vector. J Virol 1996; 70: 8098–8108.
18 Fisher KJ et al. Recombinant adeno-associated virus for muscle
directed gene therapy. Nature Med 1997; 3: 306–312.
19 Jooss K, Yang Y, Fisher KJ, Wilson JM. Transduction of dendritic
cells by DNA viral vectors directs the immune responses to
transgene products in muscle fibers. J Virol 1998; 72: 4212–4223.
20 Yang Y, Greenough K, Wilson JM. Transient immune blockade
prevents formation of neutralizing antibody to recombinant
adenovirus and allows repeated transfer to mouse liver. Gene
Therapy 1996; 3: 412–420.

Immune responses to Ad and AAV in humans
N Chirmule
et al
1583
21 Guibinga G-H et al. Combinatorial blockade of calcineurin and
CD28 signaling facilitates primary and secondary therapeutic
gene transfer by adenovirus vectors in dystrophic (mdx) mouse
muscles. J Virol 1998; 72: 4601–4609.
22 Zhang H-G et al. Application of a fas ligand encoding a recombi-
nant adenovirus vector for prolongation of transgene
expression. J Virol 1998; 72: 2483–2490.
23 Ilan Y et al. Insertion of the adenoviral E3 region into a recombi-
nant viral vector prevents antiviral humoral and cellular
immune responses and permits long-term gene expression. Proc
Natl Acad Sci USA 1997; 94: 2587–2592.
24 Flomenberg P, Piaskowski V, Truitt RL, Casper JT. Characteriz-
ation of human proliferative T cell response to adenovirus. J
Infect Dis 1995; 171: 1090–1096.
25 Smith CA, Woodruff LS, Kitchingman GR, Rooney CM. Aden-
ovirus-pulsed dendritic cells stimulate human virus-specific T-
cell responses in vitro. J Virol 1996; 70: 6733–6740.
26 Abbas AK, Murphy KM, Sher A. Functional diversity of helper
T lymphocytes. Nature 1996; 383: 787–793.
27 Ma Y et al. Inhibition of collagen-induced arthritis in mice by
viral IL-10 gene transfer. J Immunol 1998; 161: 1516–1524.
28 Fukaura H et al. Induction of circulating myelin basic protein
and proteolipid protein-specificc transforming growth factor-
␤-1 secreting Th3 T cells by oral administration of myelin in
multiple sclerosis patients. J Clin Invest 1996; 98: 70–77.
29 Pohl-Koppe A et al. Identification of a T cell subset capable of
both IFN-␥ and IL-10 secretion in patients with chronic Borrelia
burgdorferi infection. J Immunol 1998; 160: 1804–1810.
30 Molnar-Kimber KL et al. Impact of pre-existing and induced
humoral and cellular immune responses in an adenovirus-based
gene therapy phase I clinical trila for localized mesothelioma.
Hum Gene Ther 1998; 9: 2121–2133.