Gene transfer to animal cells - Principles of Gene Manipulation - page 189

gene-therapy treatment underlines the necessity for
rigorous safety testing (Marshall 1999). A number
of strategies are being developed to control the activ-
ity of the immune system when the vectors are first
introduced into the host (reviewed by Benihoud et al .
1999).
Most early adenoviral vectors were replication-
deficient, lacking the essential E1a and E1b genes
and often the non-essential gene E3. These first-
generation 'E1 replacement vectors' had a max-
imum capacity of about 7 kb and were propagated
in the human embryonic kidney line 293. This is
transformed with the leftmost 11% of the adenoviral
genome, comprising the E1 transcription unit, and
hence supplies these functions in trans (Graham et al .
1977). Although these vectors have been used
with great success, they suffer from two particular
problems: cytotoxic effects, resulting from low-level
expression of the viral gene products, and the
tendency for recombination to occur between the
vector and the integrated portion of the genome,
resulting in the recovery of replication-competent
viruses. Higher-capacity vectors have been devel-
oped, which lack the E2 or E4 regions in addition to
E1 and E3, providing a maximum cloning capacity
of about 10 kb. These must be propagated on com-
plementary cell lines providing multiple functions,
and such cell lines have been developed in several
laboratories (e.g. Brough et al . 1996, Gao et al .
1996, Gorziglia et al . 1996, Zhou et al . 1996). The
use of E1/E4 deletions is particularly attractive, as
the E4 gene is responsible for many of the immuno-
logical effects of the virus (Gao et al . 1996, Dedieu
et al . 1997). Unwanted recombination has been
addressed through the use of a refined complement-
ary cell line transformed with a specific DNA frag-
ment corresponding exactly to the E1 genes (Imler
et al . 1996). An alternative strategy is to insert
a large fragment of 'stuffer DNA' into the non-
essential E3 gene, so that recombination yields a
genome too large to be packaged (Bett et al . 1993).
Gutless adenoviral vectors are favoured for in vivo
gene transfer, because they have a large capacity
(up to 37 kb) and minimal cytotoxic effects (reviewed
by Morsey & Caskey 1999). Therefore, transgene
expression persists for longer than can be achieved
using first-generation vectors (e.g. see Schneider
et al . 1998). Complementary cell lines supplying all

adenoviral functions are not available at present, so
gutless vectors must be packaged in the presence of a
helper virus, which presents a risk of contamination.

Adeno-associated virus

AAV is not related to adenovirus, but is so called
because it was first discovered as a contaminant in
an adenoviral isolate (Atchison et al . 1965). AAV
is a single-stranded DNA virus, a member of the
parvovirus family, and is naturally replication-
defective, such that it requires the presence of
another virus (usually adenovirus or herpesvirus)
to complete its infection cycle. In adenovirus- or
herpesvirus-infected cells, AAV replicates lytically
and produces thousands of progeny virions (Buller
et al . 1981). However, in the absence of these
helpers, the AAV DNA integrates into the host cell's
genome, where it remains as a latent provirus (Berns
et al . 1975). In human cells, the provirus integrates
predominantly into the same genetic locus on chro-
mosome 19 (Kotin et al . 1990). Subsequent infec-
tion by adenovirus or herpesvirus can 'rescue' the
provirus and induce lytic infection.
The dependence of AAV on a heterologous helper
virus provides an unusual degree of control over
vector replication, making AAV theoretically one of
the safest vectors to use for gene therapy. Proviral
integration is considered advantageous for increas-
ing the persistence of transgene expression, while at
the same time the site specificity of this process the-
oretically limits the chances of insertional mutagen-
esis. Other advantages include the wide host range,
which encompasses non-dividing cells (reviewed by
Muzyczka 1992, Rabinowitz & Samulski 1998).
The AAV genome is small (about 5 kb) and
comprises a central region containing rep (replicase)
and cap (capsid) genes flanked by 145-base inverted
terminal repeats (Fig. 10.6). In the first AAV vec-
tors, foreign DNA replaced the cap region and
was expressed from an endogenous AAV promoter
(Hermonat & Muzyczka 1984). Heterologous pro-
moters were also used, although in many cases
transgene expression was inefficient because the
Rep protein inhibited their activity (reviewed by
Muzyczka, 1992). Rep interference with endogenous
promoters is also responsible for many of the cyto-
toxic effects of the virus. Several groups therefore