Gene Therapy (2001) 8, 1248–1254
 2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00
www.nature.com/gt
RESEARCH ARTICLE
Self-complementary recombinant adeno-associated
virus (scAAV) vectors promote efficient transduction
independently of DNA synthesis
DM McCarty1,2, PE Monahan1,3 and RJ Samulski1,4
1
UNC Gene Therapy Center, 2School of Pharmacy, 3Department of Pediatrics, 4Department of Pharmacology University of North
Carolina at Chapel Hill, NC, USA
Adeno-associated virus (AAV) vectors package singlestranded genomes and require host-cell synthesis of the
complementary strand for transduction. However, when the
genome is half wild-type size, AAV can package either two
copies, or dimeric inverted repeat DNA molecules. Dimeric,
or self-complementary molecules (scAAV) should spontaneously reanneal, alleviating the requirement for host-cell
DNA synthesis. We generated and characterized scAAV
vectors in order to bypass the rate-limiting step of secondstrand synthesis. In vitro, scAAV vectors were five- to 140fold more efficient transducing agents than conventional
rAAV, with a 5.9:1 particle to transducing unit ratio. This
efficiency is neither greatly increased by co-infection with
Ad, nor inhibited by hydroxyurea, demonstrating that transduction is independent of DNA synthesis. In vivo, scAAV
expressing erythropoietin resulted in rapid and higher levels
of hematocrit than a conventional single-stranded vector.
These novel scAAV vectors represent a biochemical intermediate in rAAV transduction and should provide new
insights into the biology of vector transduction. Gene
Therapy (2001) 8, 1248–1254.
Keywords: AAV; vector; dimer; double-stranded DNA
Introduction
Adeno-associated virus (AAV) is a nonpathogenic,
helper-dependent member of the parvovirus family. One
of the identifying characteristics of this group is the
encapsidation of a single-stranded DNA (ssDNA) genome.1–3 In the case of AAV, the separate plus or minus
polarity strands are packaged with equal frequency, and
either is infectious.4,5 At each end of the ssDNA genome,
a palindromic terminal repeat (TR) structure base-pairs
upon itself into a hairpin configuration.6 This serves as
a primer for cellular DNA polymerase to synthesize the
complementary strand after uncoating in the host cell.
Adeno-associated virus generally requires a helper virus
for a productive infection. Although adenovirus (Ad)
usually serves this purpose, treatment of AAV infected
cells with UV irradiation or hydroxyurea (HU) will also
allow limited replication.7–10
Recombinant AAV (rAAV) gene delivery vectors also
package ssDNA of plus or minus polarity, and must rely
on cellular replication factors for synthesis of the complementary strand. While it was initially expected that this
step would be carried out spontaneously, by cellular
DNA replication or repair pathways, this does not appear
to be the case. Early work with rAAV vectors revealed
that the ability to score marker gene expression was dra-
Correspondence: DM McCarty, Rm 7119 Thurston Building, CB#7352,
Chapel Hill, NC 27599, USA
Received 24 January 2001; accepted 22 May 2001
matically enhanced when cells were co-infected with
adenovirus, or transiently pretreated with genotoxic
agents.11 This enhancement correlated with the formation
of duplex DNA from the single-stranded virion DNA
(vDNA).12,13 Similar induction of rAAV vectors has been
observed in vivo following treatment with Ad, ionizing
radiation, or topoisomerase inhibitors.13–15 However, the
effect was highly variable between different tissues and
cell types. It has more recently been suggested that reannealing of complementary vDNA from separate infecting
rAAV particles may be an important pathway for
rAAV transduction.16
The requirement for complementary-strand synthesis,
or recruitment, is now considered to be a limiting factor
in the efficiency of rAAV vectors. The transduction rate
for rAAV in mouse liver has been estimated at approximately 5% of hepatocytes after portal vein infusion of
4.2 × 1010 particles.17 Subsequent experiments revealed
that the rAAV vDNA had been taken up into the nuclei
of virtually all of the liver hepatocytes, and that the transduction potential of these genomes could be rescued by
co-infection with adenovirus.18 This is consistent with an
earlier report of up to 25% of mouse hepatocytes transduced by 1010 particles of rAAV in the presence of coinfecting adenovirus.13 Expression from rAAV in liver
tissue coincides with the formation of duplex DNA and
the vDNA appears to be lost if not converted to duplex
within 5–13 weeks.19 Further experiments suggest that a
subpopulation of mouse hepatocytes is transiently
permissive for rAAV tranduction in vivo.18
Rather than rely on potentially variable cellular mech-
rAAV double-stranded vector
DM McCarty et al
anisms to provide a complementary-strand for rAAV
vectors, we have found that we can circumvent the problem by packaging both strands as a single DNA molecule.
Original studies on wild-type (wt) AAV determined that
defective interfering particles (DI) often carried atypical
genomes comprised of inverted repeats.20 Subsequent
studies have also revealed that rAAV DNA of less than
half the wtAAV genome length can be packaged as a
dimer or a diploid monomer, mimicking a DI particle.21–23
The replication scheme of AAV suggests that the dimeric
DNA molecules would be in the inverted repeat configuration.24,25 These self-complementary molecules
(scAAV) should have the ability to re-fold into doublestranded DNA (dsDNA) templates for expression with
pseudo-first order kinetics upon uncoating. In this study,
we engineered a number of vector constructs, which
packaged genomes varying in size relative to the wtAAV
genome, and evaluated their ability to transduce in vitro
and in vivo. We observed an increased efficiency of transduction from scAAV vectors over conventional rAAV in
HeLa cells (5- to 140-fold). More importantly, unlike conventional single-stranded AAV vectors, inhibitors of
DNA replication did not affect transduction from scAAV
vectors. In addition, scAAV vectors displayed a rapid
onset and a higher level of transgene expression in mouse
hepatocytes in vivo. All of these biological attributes support the generation and characterization of a new class
of AAV vectors (delivering duplex DNA) that should
significantly contribute to the ongoing development of
parvovirus-based gene delivery systems.
1249
Figure 1 Virion DNA content of rAAV vectors. The drawing illustrates
the DNA content of the rAAV vectors used in this study and the predicted
conformation that they adopt upon release from the virions. The transgenes expressed from the cytomegalovirus immediate–early promoter
(CMV) are: green fluorescent protein (GFP), ␤ galactosidase (LacZ),
mouse erythropoietin (nEpo). Neomycin phosphotransferase (neo) is
expressed from the SV40 early promoter (SV40). The size, in nucleotides
(nt) of each packaged DNA molecule is indicated. The self-complementary
(scAAV) GFP dimer and mEpo dimer vectors fold into a complete duplex
DNA with one extra copy of the terminal repeat while the GFPneo, LacZ,
and mEpo␭ vectors require cell-mediated DNA synthesis of the
complementary strand.
Results
Generation of scAAV vector
A rAAV plasmid construct (pTR-CMV-GFP), with a
replicon size of 2299 nucleotides, was used to generate a
viral vector stock (rAAV-GFP) by conventional methods.
The predicted size of the dimeric replicative form of this
vector was 4474 nucleotides (Figure 1), which was 95.6%
of the wt AAV genome length. The viral vectors were
fractionated by isopycnic gradient centrifugation in CsCl
and the vDNA content of each fraction was analyzed on
alkaline agarose gels (Figure 2). PhosphoImager scans
were used to quantify the vDNA-specific bands from
each fraction. Under denaturing conditions, the selfcomplementary dimer DNA (Figure 2a, fractions 10–13)
ran at approximately twice the length of the monomeric
genome. The hybridizing material in fractions 2–4 is
unpackaged replicative form DNA that sediments at the
bottom of the gradient. Athough a DNase step was
included in the vector purification (see Methods), the
treatment was not intended to be exhaustive and this
material proved to be DNase sensitive in subsequent
experiments while the material in fractions 10–14 was
DNase resistant (data not shown). Vectors containing
mostly dimeric DNA genomes (fractions 10 and 11) were
designated as ‘self-complementary rAAV’ (scAAV). The
inverted repeat structure of these molecules was confirmed by restriction enzyme digestion (data not shown).
Two additional rAAV vectors (Figure 1) were generated and purified in parallel, and analyzed in the same
manner (Figure 2b and c). The first, rAAV-GFPneo, contained a neo gene in addition to the GFP and had a replicating genome length of 3398 nucleotides. This was
Figure 2 Recombinant AAV vectors fractionated on CsCl gradients.
Virion DNA (vDNA) was extracted from CsCl gradient fractionated
CMV-GFP (a), GFPneo (b), and LacZ (c) rAAV vectors. Alkaline agarose
gels of the vDNA were Southern blotted and hybridized with a CMVGFP DNA fragment. Markers at the left end of panel a were the excised
vector sequences from the plasmids used to generate the viral vectors (see
Results). The number of unit length, ssDNA, vector copies per molecule
are indicated by 1×, 2×, and 4×. The viral vectors used in the experiments
depicted in Figures 3 and 4 were from fractions a-11 or a-10 for scAAV
CMV-GFP (as indicated in the Figure legends), fraction a-13 or a-14 for
the monomer CMV-GFP, fraction b-13 for GFPneo, and fraction c-12
for LacZ.
Gene Therapy
rAAV double-stranded vector
DM McCarty et al
1250
72.6% of the wtAAV genome size, and was too large to
be packaged as a dimer. The second was a 4898 nucleotide rAAV-CMV-LacZ construct, which was slightly
larger (104.7%) than wtAAV genome size, but within the
limit for efficient packaging.22 The lower density, higher
mobility hybridizing material in fractions 14 and 15
(Figure 2, panel c) comprised genomes which had undergone deletions and these fractions were not used in
subsequent experiments.
Transduction with dimeric versus monomeric rAAV
vectors and effects of Ad co-infection
The transducing efficiency of the scAAV-GFP (Figure 2a,
fraction 11) was compared with the homologous monomer (fraction 13), as well as the GFPneo and LacZ vectors
(Figure 2b and c, fractions 13 and 12, respectively), in
HeLa cells infected at low multiplicity (Figure 3). The
particle numbers were calculated from the specific, fulllength vDNA PhosphorImager signals in each fraction on
the Southern blot, after correction for monomeric versus
dimeric DNA copy number. Thus, each scAAV contains
two copies of the transgene as a single molecule, in the
inverted repeat orientation, while each monomeric
particle contains one single-stranded copy.
The scAAV CMV-GFP vector (fraction 11), containing
approximately 90% dimer virus, yielded a 5.9:1 ratio of
physical particles to transducing units, thus bearing out
our prediction of high transducing efficiency. Fraction 13,
from the same gradient, conversely containing approximately 80–90% monomer virus, had a 24.6:1 particle to
transducing unit ratio. This four-fold difference in
efficiency represented a minimum difference when it was
considered that the dimer contamination in the monomer
fraction would have a greater impact on its transducing
potential than the monomer component would contribute
Figure 3 Transduction efficiency of scAAV versus conventional rAAV
vectors in the absence and presence of co-infecting adenovirus. The
efficiencies of the three CsCl fractionated vectors (Figure 1) were compared
in rapidly dividing HeLa cells infected with scAAV-GFP fraction 11 (0.5
particles per cell), monomer CMV-GFP fraction 13 (0.5 particles per cell),
rAAV-GFPneo fraction 13 (2 particles per cell), or rAAV-LacZ fraction
12 (0.5 particles per cell). Transduction was quantified at 24 h after infection by counting GFP-positive cells using fluorescence microscopy, or by
fixing the cells and X-gal staining. The transducing efficiency was graphed
as the number of physical particles per transducing unit, as determined
by the number of cells scoring positive for GFP or LacZ expression. Dark
grey bars indicate transducing efficiency in the presence of Ad co-infection
at 5 p.f.u. per cell.
Gene Therapy
to the dimer fraction. In contrast, the monomeric ssDNA
GFPneo and LacZ vectors had particle to transducing
unit ratios of 125:1 and 828:1, respectively, comparable to
previously reported efficiencies for these vectors.13,26
The transducing efficiency of conventional rAAV vectors can be greatly enhanced (up to 100-fold) by co-infection with Ad, or by treatment with DNA damaging
agents or other types of cell stress. This enhancement had
been associated with the cell-mediated transformation of
the ssDNA genome into active ds-DNA transcription
templates. Because the dimeric rAAV contains the two
complementary strands packaged as a single molecule,
we predicted that transduction would be independent of
enhancement by adenovirus. This was largely the case
when HeLa cells were co-infected with the scAAV vectors and 5 infectious units per cell of adenovirus
(Figure 3). The number of GFP-positive cells in the
scAAV infected cultures was increased by only 1.6-fold,
an effect which could be attributed to the transcriptional
effects of adenovirus infection on the activity of the CMV
promoter as previously reported.27,28 The monomer vector transduction rate was increased 2.4-fold by Ad coinfection, while the GFPneo and LacZ vectors were
induced 6.0-fold and 12.8-fold, respectively.
Transduction with dimeric rAAV vectors in the absence
of host-cell DNA synthesis
Because the vDNA of the scAAV vectors contained both
DNA strands on a single molecule, allowing efficient
renaturation upon uncoating, we predicted that these
vectors would obviate the role of host-cell DNA synthesis
in transduction. We compared the scAAV-GFP vector
with the homologous monomer, and the GFPneo vector,
in HeLa cells pretreated with hydroxyurea (HU) 24 h
before infection to inhibit host cell DNA synthesis. Hydroxyurea treatment was continued, uninterrupted, at the
same concentrations following infection and maintained
on the cells for the following 24 h, until GFP transduction
was scored.
Transduction from the scAAV-GFP was stimulated by
up to 1.9-fold in response to increasing concentrations of
HU (Figure 4a). This stimulation, similar in magnitude to
that observed with Ad co-infection, was probably affected through a combination of transcriptional transactivation of the CMV promoter brought about by cell stress,
and the accumulation of GFP in the nondividing cells.
In contrast, transduction from the homologous monomer
vector fraction was stimulated at the lowest HU concentration and inhibited at higher concentrations. The
residual transducing activity from the monomer vector at
higher HU concentrations, at a level approximately fivefold lower than that of the scAAV fraction, is consistent
with the 10–20% contamination of the monomer fractions
with dimer containing particles (Figure 2a). The rAAVGFPneo vector transduction was inhibited greater than
10-fold under the same conditions. Similar results were
obtained by treatment with aphidicolin, a polymerase
␣/␦ specific inhibitor (Figure 4b). This confirmed our
hypothesis that scAAV transduction was independent of
host-cell DNA synthesis.
Transduction by scAAV in vivo
A different reporter was used for the comparison of
scAAV and conventional single-stranded rAAV
efficiency in vivo. The dimer-producing construct con-
rAAV double-stranded vector
DM McCarty et al
1251
Figure 5 In vivo transduction of mouse liver tissue with scAAV or singlestranded rAAV vectors. Ten-week-old Balb-c ByJ mice were infused with
2 × 1010 particles of either scAAV-CMV-mEpo, (䉬) (n = 4), or full-length
single-stranded rAAV-CMV-mEpo␭, (왖) (n = 5), in 200 ␮l normal saline
by portal vein injection. One group of control mice was infused with normal saline, (䊐) (n = 4), and a single mouse, (䊊) was phlebotomized at
the same 7-day intervals without prior surgery. Blood hematocrit was used
as a functional measure of mEpo expression.
Figure 4 Transduction with scAAV and conventional rAAV in the presence of DNA synthesis inhibitor. (a) HeLa cell cultures at 30% confluence
were treated with the indicated concentrations of hydroxyurea 24 h before
infecting with 3.8 × 106 particles of the scAAV-GFP, (䉬), (Figure 2a,
fraction 10), the homologous monomer, (䊉) (Figure 2a, fraction 14), or
rAAV-GFPneo, (왖), (Figure 2b, fraction 13). The HU treatment was
maintained until transduction was assayed at 24 h after infection. Each
data point was calculated from the mean of the number of GFP-positive
cells in 10 random fields independently of the total cell number, which
was variable due to the effect of hydroxyurea on cell division. (b) The same
procedure was used to evaluate transduction in the presence of the indicated concentrations of aphidicolin. Only the scAAV and homologous
monomer (fractions 10 and 14) were compared.
tained only the mouse erythropoietin gene (mEpo) transcribed from the CMV promoter. The size of the replicating element of this minimal vector was 2248
nucleotides. The dimeric form of this molecule, 4372
nucleotides in length (Figure 1), was 93% of the wtAAV
genome size, and was readily packaged. A second construct contained the identical transgene, with the addition
of a downstream heterologous sequence (␭ phage) to
bring the size of the recombinant vector to 4570 nucleotides, or 98% of the wtAAV genome size. Previous studies have used lambda phage DNA as a stuffer without
deleterious effects on the vector.21 Both vectors were purified by heparin-agarose chromotography. The smaller
vector was additionally purified on a CsCl gradient to
isolate dimeric DNA-containing virions (not shown). The
two vector stocks were quantified using Southern blots
from alkaline agarose gels to determine the number of
DNA-containing particles. In this case, approximately
25% of the particles in the dimer fraction contained two
separate monomer genomes. Because they could not be
separated from true dimer by density, and because their
behavior has not been characterized, we counted these as
dimer particles, for the purpose of comparison to the fulllength vector, such that the dimer effect might only be
underestimated rather than overestimated.
Equal numbers of physical rAAV particles (2 × 1010 per
animal in 200 ␮l normal saline) were administered to
mice by portal vein injection. The expression of the mEpo
gene was evaluated by observing changes in hematocrit
at 7-day intervals. Control mice received either intraportal saline injections or were not operated, but phlebotomized at 7-day intervals. Mice receiving the dimeric vector
responded with a rapid increase in hematocrit (Figure 5),
and with continuing increases over the following 2
weeks. Considering the lag time between expression of
erythropoietin and the production of red blood cells, this
suggested that the scAAV vector was expressed at high
levels within the first week. Mice which received the fulllength, ssDNA vector did not show a significant increase
in hematocrit until 21 days after injection, and did not
reach levels comparable to the scAAV-treated animals
over the course of the experiment.
Discussion
We have set out to circumvent the requirement for target
cell-mediated complementary-strand synthesis for rAAV
transduction. In doing so we expected to achieve higher
rates of transduction and to identify additional parameters, which might effect these rates. The means of providing cells with a duplex rAAV virion DNA was to take
advantage of the AAV rolling-hairpin mode of replication in which monomeric vDNA is generated from
dimeric inverted repeat intermediates.24,25 When the genGene Therapy
rAAV double-stranded vector
DM McCarty et al
1252
ome is sufficiently small, the dimeric inverted repeats
themselves can be encapsidated as a self-complementary
DNA molecule (scAAV).
In cultured HeLa cells, the scAAV vector was greater
than four-fold more efficient than the homologous vector
containing only a monomeric ss-DNA genome. We
expect that this difference would be greater if not for the
approximately 10–20% contamination of monomer fractions with dimer vectors. Consistent with this interpretation, the scAAV was 20-fold more efficient than a conventional rAAV-GFPneo vector and 140-fold more
efficient than a rAAV-LacZ vector, although we cannot
rule out the possibility of additional inhibition of the latter vectors due to specific cis or trans effects of the different marker genes.
We have further demonstrated the increased efficiency
of the scAAV in vivo by transducing mouse hepatocytes.
While the CMV promoter is not known to be particularly
active in liver tissue,28 we chose to target liver in this
experiment because it is feasible to achieve a relatively
homogeneous distribution of vector particles over a large
tissue. Infecting mice with scAAVmEpo leads to a faster
response, and a greater rise in hematocrit, than the fulllength ssDNA vector carrying the same gene. This supports our observations in cultured cells and is consistent
with the view that the dimeric vectors are ready to
express the transgene immediately upon uncoating and
entry into the nucleus. The higher levels of expression
ultimately achieved might reflect either the inability of
many infected cells to form dsDNA from conventional
rAAV, or the loss or degradation of ss vDNA before the
formation of duplex.19
Our indirect assessment of transduction in vivo,
through increase in hematocrit, provides only a limited
range of response to increasing levels of erythropoitin. It
therefore remains possible that the differences that we
observed between the two vectors represents a minimum
difference in the number of transduced hepatocytes. If
indeed it is only complementary-strand synthesis that is
limiting rAAV in these cells, we would expect to achieve
near 100% transduction, based on previous observations
of rAAV delivery to hepatocytes at similar doses.18
Future experiments using heterologous marker genes in
liver and other potential target tissues in vivo will provide
a more accurate assessment of the potential of the scAAV
vectors in cells which appear to be refractory to rAAV
transduction.
As we have demonstrated by pre-treatment of cells
with HU, transduction with the scAAV vector is independent of host-cell DNA synthesis. The ability to transduce cells in the absence of DNA synthesis represents a
fundamental departure in the biology of the scAAV vectors from the parent virus, allowing them to function
under circumstances where conventional rAAV vectors
would normally fail. Certain cell types are extremely inefficient for rAAV transduction, ostensibly due to the
inability to synthesize or recruit a complementary
strand.13,14,19 The scAAV will suffer no such limitation
and can be used with marker genes to directly determine
whether a cell is permissive for rAAV transduction in all
other steps irrespective of DNA synthesis. In light of the
drastic treatments which have been proposed to promote
rAAV transduction efficiency in vivo (ie gamma radiation,
HU, topoisomerase inhibitors)14,29 it will be prudent to
evaluate whether such lengths would be either necessary
Gene Therapy
or effectual using scAAV in preliminary studies. Regardless of the ability of the target cell to make the rAAV
complementary strand, it is clear that these reagents provide an alternative AAV delivery system for genes that
may require rapid onset. More importantly, our data suggest that scAAV vectors achieve overall higher levels of
therapeutic product when identical number of particles
are administered. Thus, scAAV vectors will prove most
useful where a more timely, robust, or quantitative
response to vector dose is required, ie in endostatin delivery for anti-cancer therapy. This potential for attaining
critical levels of transgene expression at a minimal dose
is important considering the level of vector production
required for clinical trials.
Apart from the potential utility of the scAAV as a vector, we can also regard these new reagents as a biochemical intermediate in the rAAV transduction pathway. This
allows the dissection of events taking place before and
after DNA synthesis, such as the CMV promoter activation that we observed in the presence of Ad co-infection. This kind of effect would otherwise have been
masked by the enhancement of transduction due to the
promotion of complementary-strand synthesis. We anticipate that use of these vectors in specific cell types will
help determine the impact of promoter choice independently of the efficiency of vector complementary-strand
recruitment. Finally, the scAAV vectors are ideally suited
to the transfer of small transgenes, which might include
blood clotting factor IX and X, erythropoietin, superoxide
dismutase, lysosomal enzymes, and most antisense RNA
or ribozyme strategies. For therapies demanding larger
genes, the scAAV may have an important role in pilot
studies for target cell permissiveness for AAV vectors.
Considering that cell cycling per se has not been an accurate predictor of rAAV permissivity,18 the direct comparison of transduction with scAAV versus the homologous
monomer may be the best predictor of target cell
potential for rAAV DNA synthesis.
Materials and methods
Plasmids
The rAAV plasmids expressing green fluorescent protein
(GFP) were constructed from the previously described
pTRBSUF-2 (a gift from Nick Muzyczka, University of Florida, Gainesville, FL USA). First, the humanized GFP
coding sequence was replaced with the enhanced GFP
(eGFP) (Clonetech, Palo Alto, CA, USA) to create the
plasmid, pTR-CMV-GFPneo. This plasmid generated the
rAAV-GFPneo vector. Second, the SalI fragment containing the neo coding region and SV40 promoter was deleted
to create pTR-CMV-GFP. The vector from this plasmid
was referred to as rAAV-GFP in this report.
The plasmid, p43mEpo, a gift from Barry Byrne
(University of Florida, Gainesville, FL, USA) contained
the mouse erythropoietin gene under the control of the
CMV promoter and generated a rAAV replicon
(rAAVmEpo) of less than half the wtAAV length. A
longer version of this construct (pmEpo-␭) was made by
inserting the 2.3 kb HindIII fragment from ␭ phage into
a ClaI site between the polyadenylation signal and the
downstream AAV terminal repeat. The rAAV-LacZ vector was generated from pDX11-LacZ, which has been
described elsewhere.30
rAAV double-stranded vector
DM McCarty et al
Viral vectors
These were generated in 293 cells (108–109 cells per
preparation) by co-transfecting three plasmids containing: (1) the specific rAAV construct, (2) the AAV rep and
cap genes (pACG), or (3) the essential adenovirus helper
genes (pXX-6).31 At 40 h after transfection, the cells were
scraped into the media and lysed by three freeze–thaw
cycles. The lysates were incubated at 37°C with 2 ␮g/ml
DNase I until flocculent debris was dispersed. The lysates
were cleared by centrifugation and rAAV was precipitated using amonium sulfate.32 The virus precipitate was
resuspended with 8 ml 10 mm Tris pH 8.0, 1 mm MgCl2
and cesium chloride was added to reach a final density
of 1.4 g/cm3 and a final volume 12.5 ml. The solution was
centrifuged for 36 h at 38 k r.p.m. in an SW41 rotor. Fractions (0.75 ml) were collected by puncturing with a hypodermic needle at the bottom of each tube and pumping
the liquid to a fraction collector. The vectors were stored
at 4°C in cesium chloride.
Virion DNA (vDNA) was extracted from 10 ␮l of each
fraction by digestion in 50 ␮l reactions containing
0.4 mg/ml protease K, 1% sarkosyl and 10 mm EDTA at
50°C for 1 h, followed by phenol/choroform extraction.
The samples were diluted three-fold with water and precipitated with ethanol for analysis by alkaline agarose gel
electrophoresis and Southern blot hybridization.
Cells and infections
HeLa and HEK 293 cells were grown in DMEM media
containing 10% FBS and penicillin/streptomycin. Viral
vector stocks were diluted in media before adding to subconfluent cultures and left on the cells until GFP transduction was observed by fluorescence microscopy at 24 h
after infection.
For expression of erythropoietin from rAAV in mouse
livers, 200 ␮l normal saline, containing 2 × 1010 physical
particles of either conventional rAAV or scAAV, was
injected directly into the portal veins of 10-week-old
Balb-c ByJ mice (Jackson Laboratory, Bar Harbor, ME,
USA). Blood samples were collected by retro-orbital
phlebotomy at the time of infection and at 7-day intervals
for determination of hematocrit.
Acknowledgements
We acknowledge the technical assistance of Jennifer Naspinski with scAAV studies in vivo, and Xiaohuai Zhou
from the UNC Vector Core for construction of
rAAV/epo/lambda vector. This work was supported in
part by NIH grants HL 48347, 51818. Doug McCarty was
supported in part by NIH grant HL 51818 and PEM
receives research support from the National Hemophilia
Foundation and from NIH HL 03960.
References
1 Cukor G, Blacklow NR, Hoggan D, Berns KI. Biology of adenoassociated virus. In: Berns KI (ed.). The Parvoviruses Plenum
Press: New York, 1984, pp 33–66.
2 Berns KI. Parvorviridae and their replication. In: Fields BN,
Knipe DM (eds). Virology, 2nd edn. Raven: New York, 1990,
pp 1743–1763.
3 Rose JA, Berns KI, Hoggan MD, Koczot FJ. Evidence for a singlestranded adenovirus-associated virus genome: formation of a
DNA density hybrid on release of viral DNA. Proc Natl Acad Sci
USA 1969; 64: 863–869.
4 Samulski RJ, Chang L-S, Shenk T. A recombinant plasmid from
which an infectious adeno-associated virus genome can be
excised in vitro and its use to study viral replication. J Virol 1987;
61: 3096–3101.
5 Berns KI, Adler S. Separation of two types of adeno-associated
virus particles containing complementary polynucleotide
chains. J Virol 1972; 5: 693–699.
6 Lusby E, Bohenzky R, Berns KI. Inverted terminal repetitions in
adeno-associated virus DNA: independence of the orientation
at either end of the genome. J Virol 1981; 37: 1083–1086.
7 Yakobson B, Hrynko TA, Peak MJ, Winocour E. Replication of
adeno-associated virus in cells irradiated with UV light at 254
nm. J Virol 1989; 63: 1023–1030.
8 Yakobson B, Koch T, Winocour E. Replication of adeno-associated virus in synchronized cells without the addition of a helper
virus. J Virol 1987; 61: 972–981.
9 Yalkinoglu AO et al. DNA amplification of adeno-associated
virus as a response to cellular genotoxic stress. Cancer Res 1988;
48: 3123–3129.
10 Rose JA, Koczot FJ. Adenovirus-associated virus multiplication:
VII. Helper requirement for viral deoxyribonucleic acid and
ribonucleic acid synthesis. J Virol 1972; 10: 1–8.
11 Alexander IE, Russell DW, Miller AD. DNA-damaging agents
greatly increase the transduction of nondividing cells by adenoassociated virus vectors. J Virol 1994; 68: 8282–8287.
12 Ferrari FK, Samulski T, Shenk T, Samulski RJ. Second-strand
synthesis is a rate limiting step for efficient transduction by
recombinant adeno-associated virus vectors. J Virol 1996; 70:
27–34.
13 Fisher KJ et al. Transduction with recombinant adeno-associated
virus for gene therapy is limited by leading-strand synthesis. J
Virol 1996; 70: 520–532.
14 Alexander IE, Russell DW, Spence AM, Miller AD. Effects of
gamma irradiation on the transduction of dividing and nondividing cells in brain and muscle of rats by adeno-associated
virus vectors. Hum Gene Ther 1996; 7: 841–850.
15 Halbert CL et al. Transduction by adeno-associated virus vectors
in the rabbit airway: efficiency, persistence, and readministration. J Virol 1997; 71: 5932–5941.
16 Nakai H, Storm TA, Kay MA. Recruitment of single-stranded
recombinant adeno-associated virus vector genomes and intermolecular recombination are responsible for stable transduction
of liver in vivo. J Virol 2000; 74: 9451–9463.
17 Snyder RO et al. Persistent and therapeutic concentrations of
human factor IX in mice after hepatic gene transfer of recombinant AAV vectors. Nat Genet 1997; 16: 270–276.
18 Miao CH et al. Nonrandom transduction of recombinant adenoassociated virus vectors in mouse hepatocytes in vivo: cell cycling does not influence hepatocyte transduction. J Virol 2000; 74:
3793–3803.
19 Miao CH et al. The kinetics of rAAV integration in the liver. Nat
Genet 1998; 19: 13–15.
20 de la Maza LM, Carter BJ. Molecular structure of adeno-associated virus variant DNA. J Biol Chem 1980; 255: 3194–3203.
21 Muzyczka N. Use of adeno-associated virus as a general transduction vector for mammalian cells. Curr Top Microbiol Immunol
1992; 158: 97–129.
22 Dong JY, Fan PD, Frizzell RA. Quantitative analysis of the packaging capacity of recombinant adeno-associated virus. Hum
Gene Ther 1996; 7: 2101–2112.
23 Hirata RK, Russell DW. Design and packaging of adeno-associated virus gene targeting vectors. J Virol 2000; 74: 4612–4620.
24 Cavalier-Smith T. Palindromic base sequences and replication
of eukaryotic chromosome ends. Nature 1974; 250: 467–470.
25 Straus SE, Sebring ED, Rose JA. Concatemers of alternating plus
and minus strands are intermediates in adenovirus-associated
virus DNA synthesis. Proc Natl Acad Sci USA 1976; 73: 742–746.
26 Zolotukhin S et al. Recombinant adeno-associated virus purifiation using novel mehtods improves infectious titer and yield.
Gene Therapy 1999; 6: 973.
1253
Gene Therapy
rAAV double-stranded vector
DM McCarty et al
1254
27 Clesham GJ et al. High adenoviral loads stimulate NF kappaBdependent gene expression in human vascular smooth muscle
cells. Gene Therapy 1998; 5: 174–180.
28 Loser P, Jennings GS, Strauss M, Sandig V. Reactivation of the
previously silenced cytomegalovirus major immediate–early
promoter in the mouse liver: involvement of NFkappaB. J Virol
1998; 72: 180–190.
29 Russell DW, Alexander IE, Miller AD. DNA synthesis and topoisomerase inhibitors increase transduction by adeno-associated
virus vectors. Proc Natl Acad Sci USA 1995; 92: 5719–5723.
Gene Therapy
30 McCown TJ et al. Differential and persistent expression patterns
of CNS gene transfer by an adeno-associated virus (AAV) vector. Brain Res 1996; 713: 99–107.
31 Xiao X, Li J, Samulski RJ. Production of high-titer recombinant
adeno-associated virus vectors in the absence of helper adenovirus. J Virol 1998; 72: 2224–2232.
32 Snyder RO, Xiao X, Samulski RJ. Production of recombinant
adeno-associated virus vectors. In: Dracopoli N et al (eds). Current Protocols in Human Genetics. John Wiley & Sons: New York.
1996, pp 12.1.1–12.2.23.