worldwide epidemiology of neutralizing antibodies to adeno ... · worldwide epidemiology of...

Worldwide Epidemiology of Neutralizing Antibodiesto Adeno-Associated Viruses

Roberto Calcedo,a Luk H. Vandenberghe,a Guangping Gao, Jianping Lin, and James M. Wilson

Gene Therapy Program, Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia

Recombinant adeno-associated viruses (AAVs) have unique gene-transfer properties that speak to their potentialas carriers for gene therapy or vaccine applications. However, the presence of neutralizing antibodies to AAV as aresult of previous exposure can significantly limit effective gene transfer. In this study, we obtained 888 humanserum samples from healthy volunteers in 10 countries around the world. Samples were assayed for neutralizingantibodies to AAV1, AAV2, AAV7, and AAV8, as well as to a novel, structurally distinct AAV vector, rh32.33, in anin vitro transduction inhibition assay. Our data revealed that neutralizing antibodies to AAV2 were the mostprevalent antibodies in all regions, followed by antibodies to AAV1. The seroprevalences of antibodies to AAV7and to AAV8 were lower than that for antibodies to AAV1, and neutralization of AAVrh32.33 was only rarelydetected. Our data also indicate a strong linkage of seroreactivity between apparently distinct serotypes that hasnot been predicted previously in animal models.

The adeno-associated virus (AAV) serves as a promising

gene delivery system because of its safety profile, its abil-

ity to transduce both dividing and nondividing cells, and

its proven record of efficacy. AAV has been detected in

many different human tissues [1–3] but has not been asso-

ciated with any disease. Preexisting immunity to AAV can

limit effective gene transfer [4, 5]. Animal studies indicate

this limitation is most likely to occur through antibody-

mediated neutralization of the incoming vector particles in

a serotype-specific manner [5–8].

In the early 1970s, several groups reported frequen-

cies of antibodies to AAV1 and AAV2 ranging from 30%

to 80% among humans [9 –13]. Since then, several other

AAV serotypes and �100 natural AAV variants have been

isolated from tissue specimens of humans and nonhuman

primates [14–16]. In preclinical models, the AAV7 and

AAV8 serotypes have, in recent years, emerged as interest-

ing candidates for gene therapies. Studies using polyclonal

rabbit antiserum have demonstrated only a low level of

cross-reactivity between AAV1, AAV2, AAV7, and AAV8

in terms of the repertoire of humoral responses they elicit

[17]. In addition, because AAV7 and AAV8 were isolated

from nonhuman primates, antibodies to these viruses were

anticipated to have lower seroprevalences among humans.

These data were consistent with the use of AAV7 and AAV8

as vectors in human populations that have high frequencies

of seroreactivity toward AAV2 and AAV1. Analysis of hu-

man IgG (intravenous immunoglobulin [IVIG]) pooled

from large groups of individuals revealed that AAV8 was 10

times more resistant to neutralization than AAV2 [18], yet

to date there is limited information available on the sero-

prevalence of antibodies toward AAV7 and AAV8 in hu-

mans.

The goal of this study was to determine the frequen-

cies of AAV7 and AAV8 specific and cross-reactive neu-

tralizing antibodies (NAbs) among humans and com-

pare them with the frequencies of AAV1 and AAV2

serotypes. In addition, a structurally distinct AAV hy-

brid of the natural isolates rh.32 and rh.33 (L.H.V. and

J.M.W., unpublished data) was included in this sero-

prevalence survey. These AAV variants were isolated

from the spleen of a Rhesus macaque, and the amino

acids encoded by their cap genes are �70% identical

Received 27 May 2008; accepted 5 August 2008; electronically published 10January 2009.

Potential conflicts of interest: L.H.V., G.G., and J.M.W. are inventors on patentslicensed to various pharmaceutical companies. R.C. and J.L.: none reported.

Presented in part: Tenth Annual Meeting of the American Society of GeneTherapy, Seattle, Washington, 30 May–3 June 2007 (abstract 8).

Financial support: National Institutes of Health (grants P30-DK047757, PO1-HL059407, and PO1-HL51746 to J.M.W.); Cystic Fibrosis Foundation (grant R881 toJ.M.W.); GlaxoSmithKline (to J.M.W.).

a R.C. and L.H.V. contributed equally to this article.Reprints or correspondence: Dr. James M. Wilson, Dept. of Pathology and

Laboratory Medicine, University of Pennsylvania, Philadelphia, PA 19104([email protected]).

The Journal of Infectious Diseases 2009; 199:381–90© 2009 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2009/19903-0013$15.00DOI: 10.1086/595830

M A J O R A R T I C L E

Seroprevalence of Antibodies to AAV in Humans ● JID 2009:199 (1 February) ● 381

to those of AAV1, AAV2, AAV7, and AAV8 and 82% identical

to those of their closest relative, AAV4 (figure 1A and figure 2,

which is available only in the electronic version of the Jour-

nal). In comparison, the amino acids encoded by the cap gene

of AAV1 are 83% homologous with those encoded by AAV8

cap [16].

MATERIALS AND METHODS

Cell culture. Human hepatoma cell line Huh7 was maintained

in Dulbecco’s modification of Eagle’s medium (DMEM [Cell-

gro]) supplemented with 10% fetal bovine serum (FBS [Hy-

clone]). Cells were cultured at 37°C in an atmosphere of 5% CO2

in air.

Production of AAV vectors. AAV vectors contained the

gene encoding �-galactosidase (LacZ), which was driven by

a cytomegalovirus (CMV) promoter (AAV.CMV.LacZ). All

AAV.CMV.LacZ vectors used in this study were made by the

Vector Core of the University of Pennsylvania (Philadelphia,

PA), as described elsewhere [17]. Recombinant AAV genomes

equipped with AAV2 inverted terminal repeats (ITRs) were

packaged by triple transfection of 293 cells with cis-plasmid, ad-

enovirus helper plasmid, and a chimeric packaging construct in

which the AAV2 rep gene was fused with cap genes of different

AAV1, AAV2, AAV7, AAV8, and rh32.33 serotypes. All recom-

binant vectors were purified by the standard CsCl sedimentation

method. Genome copy titers of AAV vectors were determined by

TaqMan analysis (Applied Biosystems), using probes and prim-

ers targeting a bovine growth hormone polyadenylation signal,

as described elsewhere [19].

Sources of serum samples. A total of 888 persons in 10

different countries across 4 continents participated in this study.

Each person provided 1 serum sample. One hundred persons

were from the Unites States (Department of Pathology and Lab-

oratory Medicine, University of Pennsylvania Hospital [Phila-

delphia]), 100 were from Australia (National Centre for Immu-

nization Research and Surveillance [Sidney]), 81 were from

Greece (Clinical Microbiology, Evangelismos Hospital [Ath-

ens]), 100 were from Belgium (Rode Kruis Vlaanderen [Leu-

ven]), 100 were from Italy (Telethon Institute of Genetics and

Medicine [Napoli]), and 407 were from Africa, with 51 from

Zambia (Lusaka), 51 from Kenya (Nairobi), 112 from South

Africa (Cape Town), 60 from Rwanda (Kigali), and 113 from

Uganda (60 from Kakira and 73 from Entebbe). All serum sam-

ples from persons in Africa were provided by Immunotherapeu-

tics GlaxoSmithKline. No demographic information was pro-

vided for any of the persons from whom samples were obtained.

NAb assay. Serum samples were heat inactivated at 56°C for

35 min. Recombinant AAV.CMV.LacZ (109 genomic copies/

well) was diluted in serum-free DMEM and incubated with

2-fold serial dilutions (initial dilution, 1:20) of heat-inactivated

serum samples on DMEM for 1 h at 37°C. Subsequently, the

serum-vector mixture was added to 96-well plates seeded with

1 � 105 Huh7 cells/well that had been infected 2 h earlier with

wild-type HAdV5 (50 viral particles/cell). After 1 h, each well

was supplemented with an equal volume of 20% FBS DMEM

and incubated for 18 –22 h at 37°C and 5% CO2. Then, cells were

washed twice in PBS and lysed, and the lysate was developed with

the mammalian �-galactosidase assay kit for bioluminescence,

in accordance with the manufacturers’ protocol (Applied Bio-

systems), and measured in a microplate luminometer (Clarity

[BioTek]). The NAb titer was reported as the highest serum di-

lution that inhibited AAV.CMV.LacZ transduction (�-gal ex-

pression) by �50%, compared with the mouse serum control

(Sigma S3509).

Statistical and informatics analysis. Multiple logistic re-

gression analysis was used to derive estimates of the prevalences

of viral types in the continents from which samples were ob-

tained. The analysis included an interaction between viral type

and location to allow for regional differences within continents.

The SEs were corrected to account for correlations between es-

timates for the same individual, and results were presented as

prevalence estimates with 95% CIs. Excel (Microsoft) was used

to measure the effect of sample bias on the linkage between the

seroprevalences of different serotypes in a given population. The

seroreactivity linkage ratio, created to describe the linkage in

seroreactivity between 2 serotypes in a given population, is cal-

culated by dividing the prevalence of serotype A in the subpop-

ulation of persons who are positive for serotype B by the preva-

lence of serotype A in the subpopulation of persons who are

negative for serotype B. Specifically, in each subpopulation, for a

particular threshold of positivity (i.e., antiserotype NAb titers of

1:20 or 1:80), the seroreactivity linkage ratio between 2 serotypes

A and B (denoted by RAB) was calculated as [PAB� � NB]/

[PB � PAB�], where NB denotes the number of serum samples

negative for serotype B, PB denotes the number of samples pos-

itive for serotype B, PAB� denotes the number of samples nega-

tive for serotype B and positive for serotype A, and PAB� denotes

the number of samples positive for serotype B and positive for

serotype A.

Comparison of amino acid sequences were done with Clus-

talX1.8, and phylogenies were inferred and visualized with

MEGA4.0. All sequences are publicly available in Genbank (ac-

cession numbers AAV1-NP_049542, AAV2-YP_680426, AAV7-

AAN03855, AAV8-AAN03857, and AAVrh.32.33-EU368926).

RESULTS

Serum samples from 888 persons spanning 4 continents and 10

countries were analyzed for the presence of NAbs against AAV1,

AAV2, AAV7, AAV8, and the structurally divergent vector

rh32.33. NAb titers were determined for each sample; data are

classified on the basis of cohort (i.e., geographic region) and the

382 ● JID 2009:199 (1 February) ● Calcedo et al.

prevalence of vector transduction inhibition at serum dilutions

of �1:20 and �1:80. Figure 1 summarizes data for samples from

persons in Australia, Europe, Africa, and the United States, fig-

ure 3 summarizes data for those in various African countries,

and figure 4 summarizes data for those in different European

countries.

Seroprevalence of anti-AAV NAbs, by continent. The data

were initially categorized on the basis of 4 large geographical

areas: Australia (100 samples), Europe (281), Africa (407), and

the United States (100). For all 4 cohorts, seroprevalences at

serum dilutions of �1:20 (figure 1B) and �1:80 (figure 1C) were

highest for anti-AAV2 NAb; the second highest seroprevalences

Figure 1. Phylogenetic tree and prevalence of neutralizing antibodies (NAbs) against different adeno-associated virus (AAV) types in 100 serumsamples from Australia, 281 from Europe, 407 from Africa, and 100 from the United States. A, Neighbor-joining phylogeny was inferred with Poissoncorrection for the protein sequences of the AAV VP1 Cap proteins. Scale, evolutionary distance of the number of substitutions per site. B and C, Sampleswere considered positive if serum dilutions of �1:20 (A) or �1:80 (B) inhibited vector transduction by �50%. **Less prevalent than anti-AAV2 NAb(P � .05); ***less prevalent than anti-AAV2 and anti-AAV1 NAbs (P � .05).


were observed for anti-AAV1 NAb at each titer and for all co-

horts except the �1:80 dilution in the United States. The differ-

ences between the seroprevalences of anti-AAV1 and anti-AAV2

NAbs were significant in Europe, Africa, and the United States at

dilutions of �1:20 and in the United States at �1:80. For each

dilution and all cohorts except the �1:80 dilution in the United

States, for which the seroprevalence of anti-AAV7 NAb was sec-

ond highest, the seroprevalences of anti-AAV7 and anti-AAV8

NAbs were lowest and were indistinguishable from one another

(P � .05). In virtually all data sets, the seroprevalences of anti-

AAV7 and anti-AAV8 NAbs were significantly different from the

seroprevalence of anti-AAV2 NAb (Australia was an exception

at �1:20) and, in most cases, from the seroprevalence of anti-

AAV1 NAb. At �1:20, the highest seroprevalences of antibodies

to each AAV were observed in Africa. The NAb titers in samples

from each country cohort were also analyzed (figure 5). The

trends noted in figure 1 are manifested in figure 5.

Seroprevalence of anti-AAV NAbs, by country. Analysis of

the samples from Africa was further stratified into 6 regions:

Entebbe, Uganda (73 samples), Kakira, Uganda (60), Rwanda

(60), South Africa (112), Kenya (51), and Zambia (51). The se-

roprevalences of anti-AAVs NAbs are specified in figure 3A for

serum dilutions of �1:20 and in figure 3B for serum dilutions of

�1:80. Anti-AAV2 NAb had the highest seroprevalence, fol-

lowed by anti-AAV2 NAb. However, the differences between

anti-AAV2 and anti-AAV1 NAb seroprevalences were statisti-

The figure is available in its entirety in the onlineedition of the Journal of Infectious Diseases.

Figure 2. Alignment of amino acid sequences for capsid protein VP1of adeno-associated virus (AAV) 2, 1, 7, and 8 and rh32.33.

Figure 3. Prevalence of neutralizing antibodies (NAbs) against different adeno-associated virus (AAV) types in 73 samples from Entebbe, Uganda;60 from Kakira, Uganda; 60 from Kigali, Rwanda; 112 from Cape Town, South Africa; 51 from Nairobi, Kenya; and 51 from Lusaka, Zambia. Sampleswere considered positive if serum dilutions of �1:20 (A) or �1:80 (B) inhibited vector transduction by �50%. **Less prevalent than anti-AAV2 NAb(P � .05); ***less prevalent than anti-AAV2 and anti-AAV1 NAbs (P � .05); ****less prevalent than anti-AAV2, anti-AAV1, and anti-AAV7 NAbs(P � .05).


cally significant in only 2 of 6 countries at a serum dilution of �1:20

and in only 1 of 6 countries at a dilution of �1:80. In all cases, the

seroprevalences of anti-AAV7 and anti-AAV8 NAbs were statisti-

cally significantly lower than that of anti-AAV2 NAb, with more-

dramatic differences at the higher dilution of �1:80 (the only ex-

ception was the seroprevalence of anti-AAV8 NAb at a dilution of

�1:20 in Rwanda, where the difference was not statistically signifi-

cant). Seroprevalences of anti-AAV7 and anti-AAV8 NAbs were

statistically significantly lower than that for anti-AAV1 NAb in most

countries at serum dilutions of �1:20 (in 4 of 6 countries for anti-

bodies against both AAVs) and �1:80 (in 4 of 6 countries for anti-

AAV7 NAb and 6 of 6 countries for anti-AAV8 NAb).

The situation in Europe (i.e., Belgium, Greece, and Italy) was

similar at serum dilutions of �1:20, with the seroprevalence of

anti-AAV2 NAb exceeding that of anti-AAV1 NAb (differences

were statistically significant for samples from Italy and Greece).

Data in Greece and Italy differed from findings in Africa at dilu-

tions of �1:80: the seroprevalences of anti-AAV1 NAbs ex-

ceeded those of anti-AAV2 NAbs (figure 4). In Belgium, the se-

roprevalence of anti-AAV2 NAb was statistically significantly

higher than that of anti-AAV1 NAb. The seroprevalences of anti-

AAV7 and anti-AAV8 NAbs were lower than that of anti-AAV2

NAb in all countries at both serum dilutions. Differences be-

tween anti-AAV8 and anti-AAV2 NAb seroprevalences were sta-

tistically significant in 3 of 3 countries at a serum dilution of

�1:20 and in 2 of 3 countries at a dilution �1:80, whereas dif-

ferences between seroprevalences of anti-AAV7 and anti-AAV2

NAbs were significant in 3 of 3 countries at a dilution of �1:20

and in only 1 of 3 countries at a dilution of �1:80.

Relative degree of seroreactivity between AAV serotypes. We

also wanted to know the percentages of individuals who had anti-

AAV1, anti-AAV7, and/or anti-AAV8 NAb titers that were

greater than the anti-AAV2 NAb titer. This information would

be useful in assessing the relative value of using the novel sero-

types rather than AAV2 as vectors. These data are summarized in

table 1. In most areas, there were remarkably few individuals in

whom the anti-AAV1, anti-AAV7, and/or anti-AAV8 NAb levels

exceeded the anti-AAV2 NAb level.

Figure 4. Prevalence of neutralizing antibodies (NAbs) against different adeno-associated virus (AAV) types in 100 serum samples from Belgium, 81from Greece, and 100 from Italy. Samples were considered positive if serum dilutions of �1:20 (A) or �1:80 (B) inhibited vector transduction by �50%.*Less prevalent than anti-AAV1 NAb (P � .05); **less prevalent than anti-AAV2 NAb (P � .05); ***less prevalent than anti-AAV2 and anti-AAV1 NAbs(P � .05).


Seroprevalence of anti-AAVrh32.33 NAbs. We recently

created a novel AAV vector from a capsid formed as a hybrid

between 2 similar capsids isolated from the spleen of a

healthy-appearing Rhesus macaque. This capsid is called

rh32.33 (figures 1 and 2). The same 888 serum samples that

were used to screen for NAbs against AAV1, AAV2, AAV7,

and AAV8 were evaluated for NAbs against AAVrh32.33.

There was a remarkable absence of NAb against this capsid in

all populations studied (figures 1, 3, and 4). Prevalence rates

in any major region or the component countries were never

Figure 5. Global distribution and magnitude of the neutralizing antibody (NAb) response in serum samples from the United States, Australia, Europe,and Africa in which the NAb titer was �1:20. Each box encloses 50% of the data, with the upper and lower limits denoting the interquartile range;the horizontal line denoting the median value; the whiskers denoting minimum and maximum values in the data set that fell within an acceptable range;and the circles denoting outliers.


�2% in serum dilutions of �1:20 and were never detected in

dilutions of �1:80.

Seroreactivity linkage analyses. An interesting question is

whether there is a linkage in seroreactivity between 2 different

AAVs within individuals. In other words if you are seroposi-

tive against one AAV, are you more likely to be seropositive

against a second AAV? This analysis was performed as follows

and is summarized in table 2. For example, we determined if

a linkage existed between AAV7 and AAV8 in the United

States by first dividing all those in the population into 2

groups: AAV7 seronegative and AAV7 seropositive (for pur-

poses of this example the distribution was made based on

titers �1:20). We then determined the frequency of seroreac-

tivity at �1:20 to AAV8 in these 2 populations (AAV7 posi-

tive and AAV7 negative). Linkage is calculated by dividing

frequency of AAV8-positive individuals in the AAV7 positive

population by the frequency of AAV8-positive individuals in

the AAV7-negative population (see Material and Methods). If

there is no linkage for or against cross seroreactivity the ratio

should be 1. A ratio of �1 suggests a positive linkage (i.e.,

seroreactivity to one virus is associated with a higher fre-

quency of seroreactivity to another virus), whereas a ratio of

�1 indicates a negative association (i.e., seroreactivity to one

virus excludes reactivity to a second virus). In this example,

for AAV7 versus AAV8, the ratio is 4.3. The reciprocal rela-

tionship (i.e., the increase in likelihood that AAV8-sero-

positive individuals are positive to AAV7 in comparison to

the AAV8-seronegative individuals) yields a ratio of 12.8. A

very different result is achieved when correlating seroreactiv-

ity between AAV2 and 7. In subjects from the United States

their evaluation of AAV7-positive individuals who are posi-

tive to AAV2 yields a ratio of 1.1 with the reciprocal evalua-

tion producing a ratio of 1.0. We concluded from this analysis

that there is a high correlation between seroreactivity to

AAV7 and AAV8 but not AAV7 and AAV2 in the United

States. This analysis was performed for individuals from the 4

major geographical areas as presented in table 2 and further

discussed below.

Europe and Australia produced high correlations in seroreac-

tivity between different AAV serotypes. Very similar findings

emerged from the analysis of samples from Africa with high cor-

relations between all pairwise comparison except AAV2 in AAV1

positive (ratio, 2.2), AAV2 in AAV7 (ratio, 2.1) and AAV2 in

AAV8 (ratio, 1.8). The data for the United States was indeed

quite different, with most comparisons showing no correlations

(AAV2 in AAV1 positive, AAV1 in AAV2 positive, AAV2 in

AAV7 positive, AAV7 in AAV2 positive, AAV1 in AAV7 positive,

and AAV7 in AAV1 positive). The comparisons of AAV2 and

AAV8 yielded modest correlations (ratios, 3.0 and 1.8) with

higher correlations between AAV7 and AAV8 as noted above

(ratios, 4.3 and 12.8). It was interesting that there was a negative

correlation between AAV1 and AAV8 (ratios, 0.3 for both pair-

wise comparisons).

DISCUSSION

Preexisting immunity to a specific AAV serotype due to a previ-

ous exposure from a natural infection can limit effective thera-

peutic gene transfer and efficacy of a genetic vaccine. For this

reason, the study of the prevalence of NAbs to AAVs in human

populations is critical for clinical vector development and the

design of gene transfer and genetic vaccine regimens. This study

is the largest published survey of seroprevalence of antibodies to

different AAV types in worldwide human populations.

Our work demonstrated that the prevalence of anti-AAV2

NAb was significantly higher than that of other evaluated anti-

Table 1. Serum samples in which the prevalence of neutralizing antibodies (NAbs)against adeno-associated virus (AAV) 1, AAV7, or AAV8 exceeded the prevalence ofanti-AAV2 NAb.

Region AAV1, % of samples AAV7, % of samples AAV8, % of samples

Australia 11 6 2Europe

Belgium 6 3 3Greece 17 1 1Italy 3 2 4

AfricaUganda 12 4 7Rwanda 18 10 11South Africa 30 4 5Kenya 21 6 11Zambia 14 8 1

United States 13 10 5

NOTE. Data indicate the percentage of samples in which anti-AAV1, anti-AAV7, or anti-AAV8 NAbtiters were �1:20 and anti-AAV2 NAb titers were �1:20.


AAV NAbs in 10 countries across 4 continents. The prevalence

ranged from 60% in Africa to 30% in the United States. A similar

frequency for anti-AAV2 NAb was described previously in a US

population [8, 20, 21]. Although the seroprevalence of antibod-

ies to AAV1 was lower than that for anti-AAV2 NAb, it was still

higher than those for anti-AAV7 and anti-AAV8 NAbs in most

regions. This was clearly observed for samples with higher NAb

titers. These data demonstrate a moderate yet significant advan-

tage of AAV7 and AAV8 over AAV1 and AAV2 in overcoming

preexisting B cell immunity in humans. Similar studies using

Table 2. Linkage of seroreactivity between adeno-associated virus (AAV) serotypes inhumans from 4 geographic areas.

NAb titer,region,seroprevalent serotype

Seroreactivity linkage ratio

AAV1�/AAV1� AAV2�/AAV2� AAV7�/AAV7� AAV8�/AAV8�

�1:20Africa

AAV1 . . . 3.5 4.5 4.0AAV2 2.2 . . . 2.2 1.8AAV7 17.2 9.7 . . . 6.4AAV8 10.7 5.7 14.8 . . .

EuropeAAV1 . . . 26.3 29.4 11.5AAV2 7.0 . . . 6.8 5.0AAV7 184.5 126.2 . . . 17.3AAV8 51.8 35.3 57.9 . . .

AustraliaAAV1 . . . 26.0 9.8 8.9AAV2 9.3 . . . 5.3 6.8AAV7 11.2 8.9 . . . 10.4AAV8 13.4 11.6 14.1 . . .

United StatesAAV1 . . . 1.3 1.2 0.3AAV2 1.3 . . . 1.1 3.0AAV7 1.2 1.0 . . . 12.8AAV8 0.3 1.7 4.3 . . .

�1:80Africa

AAV1 . . . 5.2 8.1 6.0AAV2 4.4 . . . 4.8 4.3AAV7 51.1 35.9 . . . 14.6AAV8 31.7 23.6 61.0 . . .

EuropeAAV1 . . . 25.1 13.2 9.4AAV2 16.5 . . . 9.7 7.8AAV7 NA NA . . . 32.1AAV8 56.6 50.2 95.8 . . .

AustraliaAAV1 . . . 14.2 9.0 6.7AAV2 10.7 . . . 7.5 7.0AAV7 NA NA . . . 40.4AAV8 32.0 31.9 72.0 . . .

United StatesAAV1 . . . 2.2 NA NAAAV2 1.8 . . . 1.3 1.8AAV7 NA 1.0 . . . 47.5AAV8 NA 1.3 13.3 . . .

NOTE. See Materials and Methods and Results for details on calculating ratios. NA, not available(because the denominator in the ratio was zero).


human pooled IgG (IVIG from 50,000 human plasma samples)

have shown a 10-fold greater resistance of AAV8 to neutraliza-

tion than AAV1 and AAV2 [22] .

The low seroprevalence of antibodies to AAVrh32.33 is re-

markable. AAV4, the relative of closest homology, was found

equally resistant to human IVIG neutralization as AAV [23].

When AAVrh32.33 was tested against human IVIG, a 16-fold

lower neutralizing activity than that for AAV8 was detected

(table 3). Only samples with a wide breadth of neutralization

to all tested serotypes and an AAV2 neutralizing activity of

�1:2560 were reactive to this new AAV variant (data not

shown). These data highlight the potential of this new hybrid

AAV variant for human applications in broad populations.

Although its properties are encouraging for development as

gene therapy vectors, further studies are warranted to fully

characterize its transduction properties and vector host inter-

actions.

The remarkably higher frequency of anti-AAV2 NAb seropos-

itivity observed in Africa than in other regions is surprising. This

higher prevalence of NAb to AAV2 was accompanied by greater

prevalences of NAbs to AAV1, AAV7, and AAV8 serotypes but

not to AAVrh32.33. In the absence of clear demographic corre-

lates to these data, it remains unclear whether living conditions,

population density, hygenic conditions, MHC background, or

other factors are involved in this phenomenon.

We observed a significant amount of linkage in seropositivity

toward distinct serotypes in several evaluated populations and

for several serotypes. One hypothesis to explain these results and

the lack of a monospecific serological response is that the anti-

serum raised in humans following infection with a single AAV

serotype has a wide, cross-reactive repertoire. The different pat-

tern of linkage in the United States, compared with the rest of the

world, and the lack of serological cross-reactivity across sero-

types of serum raised against AAVs in several animal models are

not explained by this hypothesis. In another possible explana-

tion, suggested by Halbert et al. [20], the cooccurrence of NAbs

to multiple serotypes in the same individual may be the result of

multiple subsequent or simultaneous infections with various

AAV types in subpopulations at relatively higher risk for AAV

infection. The apparent higher titers of anti-AAV2 NAb and

other anti-AAV NAbs might then be explained by higher inher-

ent immunogenicity [20] or multiple subsequent infections with

the same serotype. Another hypothesis for these results is that

humans generate a serological response of larger breadth that

may be due to a rapid molecular evolution of multiple coinfect-

ing parental viruses in each individual: this could lead to the

generation of antibodies directed against a wide spectrum of ho-

mologous antigens. This last hypothesis is supported by the wide

distribution of heterogeneous AAV sequences retrieved from

human tissue and the propensity of AAVs to evolve through

various mechanisms of molecular evolution [14].

Recent unexpected results from the STEP HIV vaccine effi-

cacy study led Moore et al. [23] to propose the existence of het-

erogeneity in the study populations with respect to the ability to

mount an effective immune response, which may be relevant to

our data. They pointed out an inverse correlation in the placebo

group between preexisting NAb to adenovirus serotype 5 and

acquisition of HIV infection and proposed that the level of an-

tiadenovirus NAb is a surrogate measure of genetically deter-

mined stronger immune responses across a broad array of

pathogens. The hypothesis of Moore et al. [23] suggests that

individuals capable of generating more-robust immune re-

sponses would likely have more-robust NAb responses to AAV

infections. Such infections, if prevalent (as suggested by our mo-

lecular epidemiology data), would lead to individuals with a

broad profile of anti-AAV NAbs. The ability of the host to re-

spond to infectious insults would be influenced greatly by ge-

netic factors and therefore may exhibit regional variation, which

is what we observed when comparing samples from the United

States with samples from the rest of the world. It is also possible

that linkage between AAV serotypes was not observed in the

United States, because the overall frequency of serologic activity

to AAV was lower.

Although the root of this wide breadth in NAb responses to

AAV1, AAV2, AAV7, and AAV8 in humans is unclear, cross-

reactivity toward AAVrh32.33 was only rarely observed, and

when it was observed, the magnitude of the NAb response was

much lower. AAVs that were isolated from mouse and rat were

also more resistant than AAV2 to human antibody neutraliza-

tion when tested by IVIG; of interest, vectors based on theses

viruses have limited tropism in human cell lines [18]. A novel

caprine AAV that is capable of transducing human cell lines has

been recently characterized by its resistance to neutralization by

IVIG, but only to levels obtained by AAV4 and AAV8 [22].

Although in vitro detection of NAbs does not always mimic

the mechanism by which NAbs may exert their effect in vivo, in

this study we have demonstrated the minimal prevalence of NAb

to a novel AAV vector, rh32.33, in several populations across the

world. In addition, the lower frequency of NAbs to AAV7 and

AAV8 serotypes may warrant their progression toward clinical

application in light of the dominating global seroprevalence of

antibodies to AAV2.

Table 3. Neutralizing antibody (NAb)titers of human pooled IgG in response todifferent adeno-associated virus (AAV)types.

AAV type NAb titer

AAV1 1:640AAV2 1:2560AAV7 1:640AAV8 1:320AAVrh32.33 1:20


Acknowledgments

We thank the following individuals for providing serum specimens: DonL. Siegel and Maria Limberis (Department of Pathology and LaboratoryMedicine, University of Pennsylvania [Philadelphia, PA]); Laurette Steens-sens (Dienst Voor Het Bloed Rode Kruis Vlaanderen [Leuven, Belgium]); JoBackhouse (National Centre for Immunization Research and Surveillance[Sidney, Australia]); Myrsini Moschou-Parara (Department of Clinical Mi-crobiology, Evagelismos Hospital [Athens, Greece]); Matthew Cleveland(Immunotherapeutics, GlaxoSmithkline [Stevenage, United Kingdom]); Al-berto Auricchio (Telethon Institute of Genetics and Medicine [Napoli, It-aly]); Matthew Price and Nicholas Cleveland (International AIDS VaccineInitiative [IAVI] [United States]); Peter Mugyenyi, Cissy Kityo Mutuluuza,and Pontiano Kaleebu (IAVI [Uganda]); Etienne Karita (IAVI [Rwanda]);Job Bwayo, Walter Jaoko, and Omu Anzala (IAVI [Kenya]); and Linda-GailBekker and Heather Jaspan (IAVI [South Africa]).

References

1. Smith-Arica JR, Thomson AJ, Ansell R, Chiorini J, Davidson B, Mc-Whir J. Infection efficiency of human and mouse embryonic stemcells using adenoviral and adeno-associated viral vectors. CloningStem Cells 2003; 5:51– 62.

2. Friedman-Einat M, Grossman Z, Mileguir F, et al. Detection of adeno-associated virus type 2 sequences in the human genital tract. J Clin Mi-crobiol 1997; 35:71– 8.

3. Auricchio A, Rolling F. Adeno-associated viral vectors for retinal genetransfer and treatment of retinal diseases. Curr Gene Ther 2005; 5:339 –48.

4. Moskalenko M, Chen L, van Roey M, et al. Epitope mapping of humananti-adeno-associated virus type 2 neutralizing antibodies: implicationsfor gene therapy and virus structure. J Virol 2000; 74:1761– 6.

5. Manning WC, Zhou S, Bland MP, Escobedo JA, Dwarki V. Transient im-munosuppression allows transgene expression following readministrationof adeno-associated viral vectors. Hum Gene Ther 1998; 9:477–85.

6. Fisher KJ, Jooss K, Alston J, et al. Recombinant adeno-associated virusfor muscle directed gene therapy. Nat Med 1997; 3:306 –12.

7. Halbert CL, Standaert TA, Aitken ML, Alexander IE, Russell DW,Miller AD. Transduction by adeno-associated virus vectors in therabbit airway: efficiency, persistence, and readministration. J Virol1997; 71:5932– 41.

8. Xiao W, Chirmule N, Berta SC, McCullough B, Gao G, Wilson JM. Genetherapy vectors based on adeno-associated virus type 1. J Virol 1999; 73:3994 – 4003.

9. Erles K, Sebokova P, Schlehofer JR. Update on the prevalence of serumantibodies (IgG and IgM) to adeno-associated virus (AAV). J Med Virol1999; 59:406 –11.

10. Blacklow NR, Hoggan MD, Rowe WP. Serologic evidence for humaninfection with adenovirus-associated viruses. J Natl Cancer Inst1968; 40:319 –27.

11. Sprecher-Goldberger S, Thiry L, Lefebvre N, Dekegel D, de Halleux F.Complement-fixation antibodies to adenovirus-associated viruses, cy-tomegaloviruses and herpes simplex viruses in patients with tumors andin control individuals. Am J Epidemiol 1971; 94:351– 8.

12. Mayor HD, Drake S, Stahmann J, Mumford DM. Antibodies to adeno-associated satellite virus and herpes simplex in sera from cancer patientsand normal adults. Am J Obstet Gynecol 1976; 126:100 – 4.

13. Parks WP, Boucher DW, Melnick JL, Taber LH, Yow MD. Seroepide-miological and ecological studies of the adenovirus-associated satelliteviruses. Infect Immun 1970; 2:716 –22.

14. Gao G, Vandenberghe LH, Alvira MR, et al. Clades of adeno-associated viruses are widely disseminated in human tissues. J Virol2004; 78:6381– 8.

15. Mori S, Wang L, Takeuchi T, Kanda T. Two novel adeno-associatedviruses from cynomolgus monkey: pseudotyping characterization ofcapsid protein. Virology 2004; 330:375– 83.

16. Gao G, Alvira MR, Somanathan S, et al. Adeno-associated viruses un-dergo substantial evolution in primates during natural infections. ProcNatl Acad Sci U S A 2003; 100:6081– 6.

17. Gao GP, Alvira MR, Wang L, Calcedo R, Johnston J, Wilson JM. Noveladeno-associated viruses from rhesus monkeys as vectors for humangene therapy. Proc Natl Acad Sci U S A 2002; 99:11854 –9.

18. Lochrie MA, Tatsuno GP, Arbetman AE, et al. Adeno-associated vi-rus (AAV) capsid genes isolated from rat and mouse liver genomicDNA define two new AAV species distantly related to AAV-5. Virol-ogy 2006; 353:68 – 82.

19. Gao G, Qu G, Burnham MS, et al. Purification of recombinant adeno-associated virus vectors by column chromatography and its perfor-mance in vivo. Hum Gene Ther 2000; 11:2079 –91.

20. Halbert CL, Miller AD, McNamara S, et al. Prevalence of neutralizingantibodies against adeno-associated virus (AAV) types 2, 5, and 6 incystic fibrosis and normal populations: implications for gene therapyusing AAV vectors. Hum Gene Ther 2006; 17:440 –7.

21. Chirmule N, Propert K, Magosin S, Qian Y, Qian R, Wilson J. Immuneresponses to adenovirus and adeno-associated virus in humans. GeneTher 1999; 6:1574 – 83.

22. Arbetman AE, Lochrie M, Zhou S, et al. Novel caprine adeno-associatedvirus (AAV) capsid (AAV-Go.1) is closely related to the primate AAV-5and has unique tropism and neutralization properties. J Virol 2005; 79:15238 – 45.

23. Moore JP, Klasse PJ, Dolan MJ, Ahuja SK. AIDS/HIV. A STEP intodarkness or light? Science 2008; 320:753–5.


worldwide epidemiology of neutralizing antibodies to adeno ... · worldwide epidemiology of...

Documents