Clades of Adeno-associated viruses are widely disseminated in human tissues

Abstract

The potential for using Adeno-associated virus (AAV) as a vector for human gene therapy has stimulated interest in the Dependovirus genus. Serologic data suggest that AAV infections are prevalent in humans, although analyses of viruses and viral sequences from clinical samples are extremely limited. Molecular techniques were used in this study to successfully detect endogenous AAV sequences in 18% of all human tissues screened, with the liver and bone marrow being the most predominant sites. Sequence characterization of rescued AAV DNAs indicated a diverse array of molecular forms which segregate into clades whose members share functional and serologic similarities. One of the most predominant human clades is a hybrid of two previously described AAV serotypes, while another clade was found in humans and several species of nonhuman primates, suggesting a cross-species transmission of this virus. These data provide important information regarding the biology of parvoviruses in humans and their use as gene therapy vectors.

FIG. 1.

Tissue distribution of AAV sequences in primate tissues. Total cellular DNAs were extracted from 10 major tissues, including the blood, from nonhuman primates (top) and humans (bottom) and were subjected to PCR analysis for AAV sequences. The x axis indicates the tissues analyzed and the y axis represents the frequencies of detection of AAV sequences. The total number of samples examined for each tissue of human or nonhuman primate origin is listed at the top of each bar.

FIG. 2.

Recombination analysis of AAV2-AAV3 hybrid clade representative hu.2. (Top) Bootscanning analysis of hu.2 versus representatives of all other clades showing a distinct phylogenetic relatedness between the 5′ (A) and the 3′ (B) segments (indicated below) of the entire VP1 capsid sequence. (Bottom) Split decomposition analysis of AAV2, AAV3, and the AAV2-AAV3 hybrid group representative hu.2, with AAV8 used as an outgroup. The left panel shows the analysis of the entire VP1 sequence, while on the right the conflicting phylogenies are resolved for segments A and B. The hamming distances are indicated on the splits. All branches were supported by a minimum of 85% bootstrap values (n = 1,000).

FIG. 3.

Neighbor-joining phylogenies of the VP1 protein sequence of primate AAVs. Major nodes with bootstrap values of <75 are indicated with an “X.” A goose parvovirus and an avian AAV (6) were used as the outgroup. Clades are indicated by name and by vertical lines to the right of the taxa from which they are made. The nomenclature for the taxa is either the serotype name or a reference to the species source (hu, human; rh, rhesus macaque; cy, cynomolgus macaque; bb, baboon; pi, pigtailed macaque; ch, chimpanzee), followed by a number indicating the order in which they were sequenced. Clade C was identified and positively determined to have originated through the recombination of known clades. The AAV2-AAV3 hybrid clade originated after one recombination event, and its unrooted neighbor-joining phylogeny is shown.

FIG. 4.

Evaluation of gene transfer efficiency of novel primate AAV-based vectors in vitro and in vivo. AAV vectors were pseudotyped with capsids as indicated and measured as percentages of GFP⁺ cells. Analyses of in vitro transduction with GFP vectors and in vivo transduction with A1AT vectors (measured as serum A1AT levels, in micrograms per milliliter) were performed as described in Materials and Methods. For in vivo comparisons, the serum A1AT level is indicated at the top of each bar, representing the gene transfer efficiency of each AAV vector.