Content uploaded by Matthew John Benskey
Author content
All content in this area was uploaded by Matthew John Benskey on Mar 09, 2016
Content may be subject to copyright.
Continuous Collection of Adeno-Associated Virus from Producer Cell
Medium Significantly Increases Total Viral Yield
Matthew J. Benskey,
1
Ivette M. Sandoval,
1
and Fredric P. Manfredsson
1,2,
*
1
Department of Translational Science and Molecular Medicine, College of Human Medicine, Michigan State University, Grand Rapids, Michigan;
2
Mercy Health Saint
Mary’s, Grand Rapids, Michigan.
The ability to efficiently produce large amounts of high-titer recombinant adeno-associated virus (AAV) is
a prerequisite to the continued success of AAV as a gene therapy tool targeted toward large-animal
preclinical studies or human clinical therapeutics. Current manufacturing procedures necessitate labo-
rious and time-consuming purification procedures to obtain AAV particles of sufficient titer and purity for
these demanding biomedical applications. The finding that AAV can be harvested and purified from
producer cell medium may represent an efficient alternative to purifying AAV from cellular lysates. Here
we sought to determine the maximum duration of time, and frequency within which AAV can be harvested
from producer cell medium, in order to maximize the yield obtained from a single transfection prepara-
tion. Human embryonic kidney 293T cells were transfected with polyethylenimine to produce AAV2/5
expressing green fluorescent protein (GFP), and cellular medium was harvested every 2 days until a
maximum duration of 19 days posttransfection. AAV2/5-GFP was released into producer cell medium at a
steady state until 7 days posttransfection, at which time titers dropped dramatically. Harvesting medium
every two days resulted in the maximum yield of AAV from a single preparation, and the cumulative yield
of AAV harvested from the producer cell medium was 4-fold higher than the yield obtained from
a traditional purification of AAV from cellular lysates. The AAV2/5 harvested from medium within the 7-
day collection time-course mediated high levels of transduction in vivo, comparable to AAV2/5 harvested
from cellular lysates. AAV purified from cell lysates showed increasing amounts of empty particles at 5
and 7 days posttransfection, whereas AAV purified from cell medium did not show an increase in the
amount of empty particles throughout the 7-day time course. Finally, we extended these findings to AAV2/9,
demonstrating that a comparable ratio of AAV2/9 particles are also released for up to 7 days post-
transfection.
INTRODUCTION
RECOMBINANT ADENO-ASSOCIATED VIRUS (AAV) is one
of the most frequently used viral vector systems for
gene therapy research and clinical applications.
1–3
One major advantage of AAV is the ability to rap-
idly initiate transgene expression in both dividing
and nondividing cells, which can be maintained
in nondividing cells over the lifetime of an individ-
ual.
4,5
Further, AAV has a high biosafety rating
because of its naturally replication-incompetent and
nonpathogenic nature. However, despite these ma-
jor advantages and the success already achieved
with AAV, concerns regarding the viability of AAV
as a gene therapy tool have arisen as a result of the
relatively cumbersome and time-intensive process
necessary to purify high-quality virus from cell ly-
sates.
5
This limitation is particularly relevant to
large-animal preclinical studies and human clinical
applications, which necessitate the ability to pro-
duce large amounts of high-titer AAV in a simple,
efficient, and cost-effective manner.
AAV production laboratories are addressing this
problem by implementing new and innovative
ways to produce AAV in large quantities, such as
the use of insect cells or cell suspensions within
bioreactors.
6,7
However, because of the use of hy-
brid helper viruses or expensive technical equip-
ment, these production techniques may not be
*Correspondence: Dr. Fredric Manfredsson, 333 Bostwick Avenue NE, Grand Rapids, MI 49503. E-mail: fredric.manfredsson@hc.msu.edu
32 jHUMAN GENE THERAPY METHODS, VOLUME 27 NUMBER 1 DOI: 10.1089/hgtb.2015.117
ª2016 by Mary Ann Liebert, Inc.
sufficiently flexible or cost effective to fit the needs
of researchers performing important gene therapy
research. Further, these techniques still rely on
the laborious task of purifying viral particles from
cellular lysates. The discovery that AAV can be
harvested from cellular medium may provide a
solution to these problems. Specifically, recent work
has demonstrated that after the standard co-
transfection technique for AAV production in hu-
man embryonic kidney 293T (HEK 293T) cells, large
amounts of AAV are released into the culture me-
dium.
8–10
The released AAV vectors contain intact
genomes, are infectious, and are easily purified with
standard iodixanol gradient centrifugation.
8–11
The
release of AAV to the cellular medium is serotype
dependent, and in most instances the majority of
total viral particles produced are released into the
medium.
8,9
Owing to the simplicity, flexibility, and
scalability of this novel production technique, it is
possible that purification of AAV from cellular me-
dium may represent a solution to the increasing
demands of gene therapy research. However, the
duration of time, and frequency within which AAV
can be harvested from producer cell medium after a
single transfection has not been fully characterized.
Lock et al.
8
demonstrated that AAV can be har-
vested from the medium at 72 hours posttransfection,
and that by increasing incubation time, the propor-
tion of virus in the medium increases. This finding
prompted us to investigate the maximum time and
frequency within which functional AAV particles can
be harvested from a single transfection preparation
of AAV. HEK 293T cells were transfected to produce
AAV2/5 expressing green fluorescent protein (GFP),
and cellular medium was harvested every 2 days
until a maximum duration of 19 days posttransfec-
tion. After that, AAV was purified from the medium,
and the number of viral genomes at each individual
time point was quantified to determine the optimal
length and frequency that AAV can be harvested
fromasinglepreparation.Finally,theratioofempty
versus full particles of medium-purified AAV was
assessed, and the biological activity was evaluated
in vivo after stereotaxic injection to the central ner-
vous system. From this work, we report the novel
observation that the total yield of intact and bio-
logically active AAV can be increased four-fold by
harvesting particles from producer-cell medium
for seven days posttransfection.
METHODS
Virus preparation
HEK 293T (ATCC) cultures were maintained in
a triple flask (Sigma-Aldrich) in Dulbecco’s modi-
fied Eagle’s medium (DMEM; Life Technologies)
supplemented with 10% fetal bovine serum (FBS;
Atlanta Biologicals) and 1% penicillin/streptomycin
(Life Technologies). Cells were passaged bi-weekly
in order to maintain the cells in the log phase of
growth. A fully confluent triple flask was used
to seed six triple flasks for each individual vector
production group.
AAV5 vectors encoding humanized GFP, or AAV9
vectors encoding a blue fluorescent protein (BFP)
under control of the hybrid chicken b-actin/cyto-
megalovirus (CBA/CMV) promoter, were produced
as described previously.
12
Additionally, AAV5 ex-
pressing GFP under control of the synapsin pro-
moter was used for experiments testing whether
saturation of producer cell medium diminishes AAV
release. Viruses were created by co-transfection of
HEK 293T cells with the AAV transgene plasmid
and a helper plasmid encoding capsid proteins for
AAV serotype 5 or 9, as well as adenovirus helper
functions. Transfected HEK 293T cells were main-
tained at 37C for 24 hr in DMEM containing 10%
FBS, after which the culture medium was replaced
with serum-free DMEM for the reminder of incu-
bation. After the change to serum-free DMEM,
culture medium was collected every 24 or 48 hr and
replaced with fresh serum-free DMEM, until the
termination of the respective time course (Fig. 1).
Harvested medium was stored at 4C until further
processing. At the termination of the time course,
cells were harvested and lysed, and virus was pu-
rified using an iodixanol gradient as described pre-
viously.
12
To purify AAV from the medium, the
harvested medium was centrifuged at 4000 ·gfor
5 min to pellet cellular debris, and clarified through
a0.5lm Mini-profile II capsule filter (Pall). The
clarified medium was then concentrated to 12 ml by
tangential flow filtration (TFF), using two parallel
Minimate TFF capsules (Pall) with a 100 kDa mo-
lecular weight cutoff. Clarified and concentrated
medium or cellular lysates were then loaded onto
an iodixanol concentration gradient and purified
by ultracentrifugation as previously described.
12
DNAse 1 (Benzonase; Sigma-Aldrich)-resistant vec-
tor genomes were titered with a dot blot assay,
12
us-
ing a biotinylated probe against the promoter.
13
A
near-infrared-conjugated antibody (IRDye 800; Li-
Cor Biosciences) was used to detect the probe and this
complex was quantified using a Li-Cor Odyssey
scanner (Li-Cor Biosciences).
Animals and viral injection
Experiments were conducted on young adult
(220 g) male Sprague Dawley rats in accordance
with guidelines of the Michigan State University
CONTINOUS PRODUCTION OF AAV 33
Institutional Animal Care & Use Committee (AUF
10/12-196-00). Rats were housed two per cage and
maintained in a light-controlled (12 hr light/dark
cycle; lights on 0600 hr) and temperature-controlled
(22 –1C) room, and provided with food and water ad
libitum. All surgery was performed under 2% iso-
flurane anesthesia. After induction of anesthesia,
rats were placed in a stereotaxic frame and the
surgical site was scrubbed with Betadine before in-
cision. A single incision was made along the ros-
trocaudal axis of the skull, and tissue overlying the
skull was retracted to expose the skull surface. A
Hamilton syringe (Hamilton) with a 30-gauge blunt-
tip needle was fitted with a siliconized pulled glass
micropipette with an opening of 60–80 lmtousefor
injections. Single-site striatal injections were made
at the following coordinates relative to Bregma:
anterior/posterior: 0.0 mm; medial/lateral: –2.7 mm;
dorsal/ventral: -4.0mm.Asthereisnobilateral
connectivity between the rodent striata, individual
striatum from separate hemispheres were used as an
n=1. Thus, each subject received two separate in-
jections (one per hemisphere) of randomized vectors
from each production group. Purified virus from all
production groups was injected at a titer of 1.0 ·10
13
viral genomes (vg)/ml. A total of 2 ll was injected at a
rate of 0.5 ll/min using an automated micropump
(World Precision Instruments). To prevent reflux the
needle was left in place for one minute after the in-
jection, after which the needle was retracted 1 mm,
and thereafter left in place for 4 additional minutes
before completely retracting the needle. The hole in
the skull was filled with sterile bone wax, and the
skin was closed using surgical staples. Rats were
checked daily for signs of infection/distress.
Tissue collection and processing
One month after AAV injections, when transgene
expression reaches peak levels,
14
animals received a
lethal injection of sodium pentobarbital and were
transcardially perfused with Tyrode’s solution fol-
lowed by 4% paraformaldehyde. Brains were re-
moved and postfixed in 4% paraformaldehyde in
tris-buffered saline (TBS) overnight, followed by
cryoprotection in 30% sucrose. Six separate series of
coronal sections (40 lm) encompassing the entire
rostrocaudal axis of the brain were prepared with a
microtome.
Figure 1. Seven-day AAV purification time course. Polyethylenimine was used to co-transfect six confluent triple flasks of HEK 293T cells with the AAV
genome along with a plasmid encoding capsid proteins and all necessary helper functions. One-day posttransfection, cellular medium was replaced with fresh
serum-free DMEM. Then, medium from each group was collected every 2 days and replaced with fresh serum-free medium (3-, 5-, and 7-day groups). To
compare the amount of AAV within the medium to that purified from cellular lysates, producer cells from individual groups were harvested at 3 (3-day group), 5
(5-day group), or 7 days (7-day group), respectively. To determine if allowing the cells to incubate with no medium change would result in an additive
accumulation of AAV in the supernatant, an additional group was transfected in an identical manner, and the same serum-free medium was kept on the
producer cells until 7 days posttransfection, at which point both cells and medium was harvested (terminal group). AAV, adeno-associated virus; DMEM,
Dulbecco’s modified Eagle’s medium; HEK, human embryonic kidney.
34 BENSKEY ET AL.
Immunohistochemistry
Immunohistochemistry (IHC) was performed on
free-floating sections. Sections were washed in TBS
containing 0.25% Triton-X 100 and blocked in 10%
normal goat serum. Blocked sections were then
incubated in a primary rabbit anti-GFP antibody
(AB290; Abcam), followed by a biotin-conjugated,
goat anti-rabbit secondary antibody (ap132b; Mil-
lipore), or a goat anti-rabbit 680LT near-infrared
secondary antibody (Li-Cor). Bound peroxidase
was visualized with 0.05% 3-3¢-diaminobenzidine
tetrahydrochloride (Sigma) with 0.01% hydro-
gen peroxide using an ABC Elite kit (Vector
Laboratories).
Unbiased stereological cell counting
Unbiased stereological counting of striatal GFP-
positive (GFP+) cells was performed as previously
described.
14
In brief, using Stereo-Investigator
software (Version 4.03; Microbrightfield, Inc., 2000),
sections were viewed on a screen at low magnifica-
tion (4·) and the transduction area containing
GFP+cells was delineated through the rostrocaudal
extent of the striatal nuclei. For estimates of GFP+
cells, every sixth section was sampled. GFP+cells
were counted using the optical fractionator meth-
od.
14
Approximately 6–9 sections per animal were
needed to count the entire transduced striatum.
Counting of GFP+cells was performed using a 60·
oil objective on an Olympus BX53 microscope
equipped with a motorized stage. The coefficient of
error for each estimate was calculated and was less
than 0.1 (Gundersen, m=1).
15
Protein quantification
Quantification of GFP protein levels in the
striatum was made indirectly by measuring near-
infrared signal with an Odyssey near-infrared
scanner after IHC using the LiCor 680LT secondary
antibody. After IHC for GFP, mounted and cover-
slipped sections containing the entirety of the
striatum were scanned with the Odyssey infrared
analyzer. The striatum was then outlined using
Odyssey software, and the total signal intensity at
the 680 wavelength within the delineated area was
quantified.
16
Electron microscopy
The ratio of empty to full particles within vector
preparations was quantified with electron mi-
croscopy after negative staining. Formvar-coated
copper mesh grids (Electron Microscopy Sciences)
were loaded with 10 ll of virus from each produc-
tion group. Grids were then washed with dH
2
O
and stained with 1% filtered uranyl acetate
(Electron Microscopy Sciences) before being
viewed on a TEM 1400 plus electron microscope
( JEOL). Ratios of empty versus full particles
were determined by directly counting negatively
stained and total particles from micrographs.
Power analysis was conducted to determine the
necessary sampling size based on an aof 0.05 for
the examined comparisons and the expected
standard error of measurement for viral particle
counting. A sample size of five per group yields a
power of 0.95. Five images were randomly ac-
quired from preparations corresponding to indi-
vidual samples within each production group. A
minimum of 1500 total particles were counted for
each production group.
qPCR analysis of AAV genomes
Quantification of AAV genomes within the stria-
tum was performed as previously described.
17
Total
DNA was extracted from the AAV-injected striatum
by incubating tissue in extraction buffer (200 mM
tris, 250 mMNaCl, 25 mMEDTA, 0.5% SDS) with
proteinase K and RNase A at 55C, followed by
phenol/chloroform extraction and ethanol precipi-
tation of the DNA. DNA concentration was deter-
mined using a Nanodrop. qPCR primers for theAAV
genome (Integrated DNA Technology) and GAPDH
(ABI Research) were conjugated to a FAM or
VIC reporter, respectively. All primers were en-
sured to produce a single product as assessed by
melt curve analysis. GAPDH was used as the ref-
erence gene and detection was similar across
treatment groups. Samples for PCRs were run in
triplicate and averaged to give a single mean value
per sample. Changes between treatment groups
were analyzed by the differences in DC
t
, which
compares the C
t
value of the AAV genome to that of
the GAPDH control gene. Data are presented as fold
change over corresponding control tissue (identical
region of the striatum from noninjected rats) as
determined by the DDC
t
, which represents the DC
t
normalized to a calibrator, in this case noninjected
control tissue.
Statistical analysis
The experimenter was blind to all experimental
conditions during data collection and analysis.
One-way analysis of variance (ANOVA) tests were
used to test for statistical significance between two
or more groups with a single independent variable.
Ap-value of less than or equal to 0.05 was consid-
ered statistically significant. If the ANOVA re-
vealed an interaction of statistical significance,
post-hoc analysis was followed by between-group
comparisons using Tukey’s test.
CONTINOUS PRODUCTION OF AAV 35
RESULTS
Time course of AAV5 release
into cellular medium
Previous studies have shown that functional AAV
can be harvested from the medium of HEK 293T
producer cells in a simple and effective method.
8,9
The majority of protocols describing AAV purifica-
tion from medium (or cellular lysates) have har-
vested medium 3 days posttransfection.
8,9,18
Here
we wanted to advance these findings and deter-
mine the maximum duration of time within which
functional AAV particles could be harvested from
the medium of HEK 293T producer cells. To do so
we began by analyzing the amount of AAV in the
cellular medium from 3 to 7 days posttransfec-
tion. We chose to initially study the time course of
release of AAV5 because of its highly efficient
neuronal tropism that is commonly utilized in our
laboratory. Polyethylenimine (PEI) was used to
transfect six confluent triple flasks of HEK 293T
cells with the AAV genome (containing a GFP
transgene under control of the CBA/CMV hybrid
promoter enhancer, flanked by AAV2 terminal
repeats) along with a helper plasmid encoding all
necessary helper functions and AAV serotype 5
capsid proteins. One day posttransfection, cellu-
lar medium was replaced with fresh serum-free
DMEM. Then, medium from each group was col-
lected every 2 days and replaced with fresh serum-
free medium. To compare the amount of AAV
within the medium to that purified from cellular
lysates, producer cells from individual group were
harvested at 3, 5, and 7 days posttransfection
(Fig. 1). Medium and cells were processed and
AAV was purified as previously described,
9,12,19
after which titers of DNAse-resistant viral ge-
nomes were quantified. Three days posttransfec-
tion there were high numbers of viral genomes
from both the medium and the cell lysate (Fig. 2A).
The amount of AAV detected in the medium
fraction remained constant out to the 7-day time
point, whereas there was a small, nonsignificant
decrease in AAV purified from the cell lysates
at these later time points (Fig. 2A). We observed
no significant degree of cell death within any
groups during the 7-day time course (data not
shown).
The fact that the amount of AAV in the medium
remained constant out to 7 days posttransfection in
the absence of overt cellular lysis, while the me-
dium was harvested and replaced every 2 days,
suggests that there is a steady-state release of AAV
into the cellular medium. To determine if allowing
the cells to incubate with no medium change would
result in an additive accumulation of AAV in the
supernatant, an additional group was transfected
in an identical manner, and the same serum-free
medium was kept on the producer cells until 7 days
posttransfection, at which point both cells and me-
dium were harvested (Fig. 1, terminal group). Sur-
prisingly, the amount of AAV in this ‘‘terminal’’
group did not show the anticipated additive accu-
mulation of virus in the medium. Instead, the de-
tected viral genomes were comparable to that of the
3-, 5-, and 7-day time points (Fig. 2A), suggesting
that the steady-state release of AAV into the me-
dium dissipates with prolonged incubation in the
same medium.
We next wanted to determine whether the lack
of accumulation of viral particles in the medium of
the terminal group was because of impaired release
of AAV, as the result of nutrient starvation or sat-
uration of the medium with viral particles. In this
experiment HEK 293T producer cells were trans-
fected as described above to produce AAV2/5-CBA-
GFP. However, one day posttransfection, groups of
two triple flasks of HEK 293T producer cells received
(1) fresh serum-free medium (control group), (2) used
3-day-old serum-free medium transferred from mock-
transfected cells (used medium group), (3) condi-
tioned 3-day-old serum-free medium transferred from
cells that were transfected and actively releasing
AAV2/5 containing a GFP transgene under control of
the synapsin promoter (transfected-used medium
group), or (4) fresh serum-free medium injected with
8.45 ·10
11
viral particles per flask (which is the av-
erage amount of virus released into the medium of
producer cells at 3 days posttransfection) of purified
AAV containing a GFP transgene under control of
thesynapsin(AAV-freshmediumgroup).Medium
was harvested at 3 and 5 days posttransfection, at
which point the medium was replaced with fresh
medium, used medium, transfected-used medium,
or AAV-fresh medium, identical to those described
above. If the cessation of virion accumulation
within the medium of HEK 293T producer cells
was because of a lack of nutrients, we would expect
to see a decreased total number of viral particles
produced in groups receiving used medium. If the
cessation of virion accumulation in the medium of
HEK 293T producer cells was because of a satu-
ration of the producer cell medium with viral
particles, we would expect to see a decrease in the
total number of viral particles produced in groups
receiving medium transferred from cells actively
releasing AAV, and in cells receiving fresh me-
dium injected with purified AAV. Medium was
processed as described above, and the total num-
ber of AAV2/5-CBA-GFP viral particles purified
36 BENSKEY ET AL.
was quantified using a probe specific to the CBA
promoter (i.e., will not detect the AAV/2/5-synapisn-
GFP virus used to treat the cells). Supplementary
Fig. S1 (Supplementary Data are available online
at wwwliebertpub.com/hgtb) shows that there
was no change in the total number of viral parti-
cles purified from any group at any time point. To
confirm the presence of viral particles in the me-
dium used to treat cells in the transfected-used
medium group, we also quantified the number of
AAV2/5-synapsin GFP genomes, using a probe
specific to the synapsin promoter. The average
number of AAV2/5-synapsin-GFP viral genomes
within the transfected-used medium was 7.33 ·10
11
,
9.72 ·10
11
, and 7.1 ·10
11
at the 3-, 5-, and 7-day time
point, respectively. This indicates that neither nu-
trient deficiency nor saturation of medium with viral
particles can account for the lack of an additive ac-
cumulation of virions observed in the terminal group
(Fig. 2A).
Figure 2. Characterization of the time course of AAV5-GFP release into cellular medium. AAV5-GFP was produced in HEK 293T cells by polyethylenimine
transfection of the AAV-GFP genome and a plasmid encoding capsid proteins for AAV serotype 5, and all necessary helper functions. Each individual
production group consisted of six triple flasks of HEK 293T cells. (A) Medium from HEK 293T producer cells was harvested at 3, 5, and 7 days posttransfection.
Individual AAV production groups were terminated and cells harvested at 3, 5, and 7 days posttransfection, respectively. An additional group of producer cells
was incubated for the entirety of time course with no medium change, and cells were harvested 7 days posttransfection (terminal group). AAV was purified and
total DNase1-resistant viral genomes were quantified. Columns represent total viral genomes collected at each respective time point +1 SEM (n=3/group) after
purification from cellular lysates (black column) or cellular medium (white columns). (B) To determine the duration in which viable AAV particles are released
into the cellular medium, medium from HEK 293T producer cells was harvested every 2 days, after the switch to serum-free medium (alternate days group, open
circle). Medium was harvested until the cells were no longer adherent. To determine the frequency at which viable AAV particles can be harvested, cellular
medium was collected every day from a separate group of HEK 293T producer cells, after the switch to serum-free medium (daily group, closed circle). AAV
was purified from medium at all time points and total DNase1-resistant viral genomes were quantified. Individual data points represent the total number of viral
particles collected at each time point, averaged over experimental replicates –1 SEM (n=2–3/group) (C) The cumulative yield of AAV5-GFP was determined by
quantifying the total amount of virus harvested from a 3-day lysate preparation (black column), a 7-day medium collection paradigm in which medium was
harvested daily (gray column), a 7-day medium collection paradigm in which medium was harvested on alternate days (white column), or a 13-day medium
collection paradigm in which medium was harvested on alternate days (white striped column). Columns represent total viral genomes collected at each
respective time point +1 SEM (n=2–3/group). Collecting and purifying AAV5-GFP from medium that was harvested on alternate days for 7 days posttransfection
resulted in significant 4-fold increase in the number of AAV genomes detected over that of AAV purified from 3-day cellular lysates. There was no significant
increase in the number of viral genomes detected by collecting the cellular medium for 13 days as compared with 7 days. *Significantly different from the 3-day
lysate group ( p<0.05). GFP, green fluorescent protein.
CONTINOUS PRODUCTION OF AAV 37
Because virion accumulation dissipates when
medium remains on producer cells forlonger periods
of time, we next wanted to investigate the maximum
duration that AAV could be collected from producer
cell medium when medium was harvested and re-
placed at regular intervals. Six triple flasks of HEK
293T cells were transfected as describe above, and
medium was collected and replaced every 2 days
until the point when the cells were no longer ad-
herent. Peak levels of AAV purified from the me-
dium were observed at 7 days posttransfection, after
which point there was a drop in the number of viral
genomes detected (Fig. 2B, open circles). Despite
this drop after the 7-day time point, there was still
an appreciable amount of virus purified from the
medium from 9 to 13 days posttransfection, after
which titers dropped precipitously. These results
demonstrate that a significant amount of intact viral
particles can be successfully harvested and purified
from the medium of producer HEK 293T cells for
almost two weeks posttransfection.
Because there is a steady-state release of AAV
when medium is replaced every 2 days, and that this
steady-state accumulation of AAV into the medium
ceases when the same medium was left on producer
cells for 7 days, we next wanted to determine the
effects of harvesting and replacing medium every
day. When medium from transfected HEK 293T
producer cells was harvested every day, there was a
similar steady-state release of AAV into the medi-
um; however, after the 4-day time point the number
of genomes detected when the medium was collected
every day was less than half of thegenomes detected
in AAV purified from the medium that was collected
on alternate days (Fig. 2B). Further, during daily
collections, AAV release into the medium dropped
precipitously after 7 days posttransfection, mirror-
ing the drop observed in the alternate-day collection
paradigm (Fig. 2B).
Together, these data demonstrate that intact
AAV5 particles are released into the medium at a
steady state where they can be regularly harvested
on alternate days until approximately 7 days post-
transfection, after which there is only a minuscule
increase in total viral yield. This ability to harvest
virus from the medium of producer cells on a regular
basis drastically increases the yield obtained from a
single preparation of AAV. For example, a compar-
ison of the total yield of AAV5 from a traditional
cell lysate preparation harvested at 3 days post-
transfection (the prototypical AAV production par-
adigm
12
) versus the total yield obtained from an
identical preparation in which medium was har-
vested every 2 days until 7 days posttransfection
results in an approximately 4-fold increase in total
viral particles (Fig. 2C). There is no significant ac-
cumulation of total viral particles beyond the 7-day
posttransfection time point, and daily collections
result in a slightly decreased yield compared with
the alternate-day collection paradigm (Fig. 2C).
Structural integrity of AAV5 released
into the medium
Previous reports have demonstrated that physi-
cally intact AAV can be harvested from producer cell
medium; however, the effects of prolonged harvest
on the ratio of empty versus full particles have not
been investigated. As such, we next sought to con-
firm that the ratio of empty versus full particles of
the AAV5 released into the medium was both con-
sistent throughout the 7-day collection period, and
was on a comparable level to that of AAV5 produced
through conventional means, that is, purified from
cellular lysates harvested at 3 days posttransfec-
tion. Empty vector particles can be identified after
negative staining with uranyl acetate using electron
microscopy.
8,11
Empty vectors appear darker because
of the presence of an electron-dense core resulting
from the pooling of uranyl acetate on top of the empty
capsid (arrowhead in Fig. 3B). Negatively stained
empty particles and total particles were counted di-
rectly from electron micrographs obtained from vec-
tor preparations purified from cell lysates and
medium harvested at 3, 5, and 7 days posttransfec-
tion. The table in Fig. 3 shows the mean ratio of empty
to full particles and the corresponding percentage of
empty particles per production group. There was no
increase in the ratio of empty to full particles in AAV
purified from producer cell medium throughout the 7-
day time course (Fig. 3 table and E–H). Surprisingly,
there was a progressive increase in the ratio of empty
versus full particles observed from 3 to 7 days post-
transfection in AAV purified from cellular lysates
(Fig. 3 table and A–D). Curiously, this increase was
not observed in the terminal pellet group (which was
also harvested at 7 days posttransfection with no
medium changes; Fig. 3 table and D).
Biological activity of AAV5 released
into the medium
Finally, we wanted to analyze the biological ac-
tivity of AAV5 purified from cellular medium. Again,
although AAV harvested from medium has demon-
strated infectivity,
8–11
the effects of continual har-
vest on the biological activity of AAV are unknown.
We chose to target stereotaxic injections of AAV to
theratstriatum,asthisallowsforsimplequantifi-
cation of transduced cells within a contained area.
We performed striatal injections of AAV5 purified
from cellular lysates or medium harvested at 3, 5,
38 BENSKEY ET AL.
and 7 days posttransfection. Adult male rats re-
ceived 2 ll of AAV5-GFP (normalized to 1 ·10
13
vg/
ml) stereotactically injected into the striatum, and
were sacrificed 1 month postsurgery. The number of
cells expressing the GFP transgene was quantified
using unbiased stereological cell counting. All
AAV5 vectors purified from the medium or lysate
displayed equal levels of transduction, with no
significant differences in the number of striatal
GFP+cells detected between any groups analyzed
(Fig. 4). Representative images of striatal GFP+
cells after transduction with AAV5-GFP purified
from cell lysates or medium harvested at 3, 5, or
7 days posttransfection are shown in Fig. 4B–I.
Counting the number of GFP+cells does not pro-
vide any information as to how much transgene
product is made by each cell (i.e., the multiplicity of
infection [MOI] or efficiency of intracellular pro-
cessing within each cell). Consequently, in addition
to quantifying the number of GFP+cells, we also
analyzed GFP protein expression in the striatum by
quantifying the total signal at the 680 wavelength
after immunohistochemical detection of GFP using
a LiCor 680 secondary antibody. There was no sig-
nificant difference in the overall GFP protein levels
detected in the striatum after delivery ofAAV5-GFP
purified from cell lysates or medium harvested at 3,
5, or 7 days posttransfection (Fig. 5A–I). That being
said there was a nonsignificant trend toward in-
creased GFP protein expression in animals that had
received AAV5-GFP purified from the cell lysate as
compared with AAV5-GFP purified from the cellu-
lar medium (Fig. 5A–I). To confirm that there was
no difference in the infectivity of AAV derived from
medium versus cellular lysate, we also quantified
the number of viral genomes present in the striatum
of the same animals using qPCR (Fig. 5J). There
was no difference in the number of intact viral ge-
Figure 3. The ratio of empty to full particles of AAV harvested from cellular lysates and medium. AAV purified from the cellular lysate or cellular medium
collected at 3, 5, and 7 days posttransfection was negatively stained with 1% uranyl acetate and viewed under a transmission electron microscope. Empty
particles are easily identified by an electron-dense circle at the center of the capsid (arrow head in B). The number of empty and full particles was counted
directly from electron micrographs. Five images were acquired at random from each sample preparation, and the ratio of empty particles versus full particles
was quantified. The table lists the number of full particles, empty particles, and total particles counted, as well as the average empty-to-full particle ratio and
average percentage of empty particle per preparation (listed as mean –1 SEM; n=5/group). Representative images of electron micrographs from each
production group are shown in (A–H), respectively. Scale bar in (H) is 100 nm and applies to (A–G). *Significantly greater than all medium groups and the
3-day and terminal lysate groups (p<0.05). **Significantly greater than all groups (p<0.05).
CONTINOUS PRODUCTION OF AAV 39
nomes detected in the striatum of animals injected
with AAV-GFP derived from either medium or cel-
lular lysates (Fig. 5J). Taken together, these data
indicate that the biological activity of AAV derived
from medium and traditional cellular lysate purifi-
cation is equal.
Time course of AAV9 release
into cellular medium
Here we have detailed the maximum duration
and optimal harvesting frequency to maximize
AAV5 yield from cellular medium after a single PEI
transfection preparation in HEK 293T cells. How-
ever, although other AAV serotypes are released
into producer cell medium,
8–11
they may not dis-
play the same prolonged release of intact viral
particles into cellular medium. To confirm that the
ability to harvest virus from cellular medium for up
to one week posttransfection was not limited to
AAV5-GFP, we performed a similar experiment by
packaging an AAV genome expressing BFP into the
AAV9 capsid. Groups of two triple flasks were
transfected as described above and medium was
collected every 2 days after the change to serum-
Figure 4. Quantification of GFP-positive cells in the striatum after transduction with AAV5-GFP purified from cellular lysates or cellular medium. Adult male
rats received 2 ll unilateral injections of AAV5-GFP (1 ·10
13
vg/ml) into the striatum. One month postsurgery, animals were sacrificed, and brains were
removed, sectioned, and stained for GFP. Numbers of GFP-positive (GFP+) cells were estimated using unbiased stereology. Columns in (A) represent mean
numbers of GFP+cells, +1 SEM (n=6/group), in animals receiving AAV5-GFP purified from cellular lysates (black columns) or cellular medium (white columns)
collected at the 3-, 5-, and 7-day posttransfection or terminal groups, respectively. Representative images of striatal transduction for each vector production
group are shown in (B–I). Insets in (B–I) are high-magnification images of the area within the white box of the transduced striatum. Scale bar in the inset in
panel I represents 50 lm and applies to all other insets. Color image available online at www.liebertpub.com/hgtb
40 BENSKEY ET AL.
free medium, out to 7 days posttransfection. We
also analyzed a ‘‘terminal group,’’ in which the
same serum-free medium was kept on the producer
cells until 7 days posttransfection, at which point
both cells and medium were harvested. Similar to
AAV5, large quantities of AAV9 were detected in
the medium of producer cells out to 7 days post-
transfection (Fig. 6). Further, there was no signif-
icant difference between the amount of AAV
purified from cellular medium versus lysate at any
time point. Finally, again similar to AAV5, there
was no additive accumulation of AAV9 in producer
cell medium in the terminal group. These results
indicate that the optimized method for continuous
collection of AAV described herein with AAV5 is
also applicable to other AAV serotypes.
DISCUSSION
The discovery that a substantial portion of AAV
is released into the cellular medium of production
HEK 293T cells during packaging presents an op-
portunity to increase the efficiency and yield of
vector production.
8–10
However, to date there has
been no investigation into the maximum duration
of time within which functional AAV particles can
be harvested from a single AAV vector preparation.
The novel work presented here details the time
course of AAV5 release into the cellular medium,
including an analysis of the structural and biolog-
ical integrity of released viral particles. Specifi-
cally, we have determined that structurally intact
and biologically active AAV5 particles are released
Figure 5. Quantification of GFP protein and AAV genomes in the striatum after transduction with AAV5-GFP purified from cellular lysates or cellular medium.
Adult male rats received unilateral injections of AAV5-GFP (1 ·10
13
vg/ml) into the striatum. (A–I) One month postsurgery, animals were sacrificed, and brains
were removed, sectioned, and stained for GFP using a 680LT secondary antibody. Stained sections were scanned on an Odyssey infrared analyzer and the area
of transduction was delineated and signal at the 680 wavelength quantified. Columns represent the mean total signal intensity +1 SEM (n=6/group) in the
striatum of animals receiving AAV5-GFP purified from cellular lysates (A, black columns) or cellular medium (A, white columns) collected at the 3-, 5-, and 7-
day posttransfection or terminal groups, respectively.(J)To confirm that the infectivity of AAV purified from cellular medium is equal to that purified from
cellular lysates, the number of viral genomes within the striatum was quantified by qPCR. One month postsurgery, genomic DNA was extracted from tissue
sections containing the striatum adjacent to the injection site. Quantification of AAV genomes was performed by qPCR analysis using probes directed against
the promoter of the AAV genome and GAPDH. Levels of AAV genomes were normalized to GAPDH. Columns represent fold change (as determined by the DDCt
method) over striatal tissue from noninjected control animals, +1 SEM (n=6/group). Color image available online at www.liebertpub.com/hgtb
CONTINOUS PRODUCTION OF AAV 41
into the medium at a steady state until approxi-
mately 7 days posttransfection. The released AAV5
can be harvested on alternate days, combined, and
purified in a simple and efficient manner in order to
increase total vector yield by approximately 4-fold
over a conventional preparation of AAV purified
from cellular lysates. Although we have detailed
the time course of release of AAV5 and shown
that AAV9 is also released for up to 7 days post-
transfection, alternative serotypes have displayed
differential release profiles,
9
and as such the opti-
mal duration and frequency of collection for other
serotypes may differ and should be determined
empirically.
There were no differences in the ratio of empty
versus full viral particles observed between cellular
lysate AAV purified at 3 days posttransfection and
AAV purified from medium at any time point ex-
amined. However, with the exception of the termi-
nal lysate group, there was a progressive increase
in the ratio of empty versus full particles in AAV
purified from cellular lysates after the 3-day post-
transfection time point. The reason why the num-
ber of empty particles would increase in the lysate-
derived AAV but not in the medium-purified AAV
is not clear. Presumably, the presence of empty
particles at longer time points represents the de-
crease in the finite number of AAV genomes trans-
fected into the producer cells, after packaging into
capsids. With this in mind, it is possible that the
release of AAV into the cellular medium is a non-
random event, favoring genome-containing capsids
over empty capsids. However, it must be noted that
EM has been criticized to lack an absolute quanti-
tative nature, and as such these observations may
be viewed as qualitative or semiquantitative at best.
Despite the increased number of empty particles
in the 5- and 7-day lysate AAV preparation, AAV
purified from cell lysates and medium (at all time
points examined) mediated efficient transduction
in vivo, and there was no difference in the number
of transduced cells between any of the groups ex-
amined. This result may seem surprising in the
face of the increased ratio of empty versus full
particles in the 5- and 7-day lysate-purified AAV;
however, the amount of virus injected was nor-
malized based on the number of genome copies.
Thus, although there may have been an increased
number of empty particles within these groups, all
animals received the same number of full, genome-
containing virions, accounting for the equal num-
ber of cells transduced. Although it is possible to
achieve equal levels of transduction by normalizing
the number of genome copies injected, because of
the increased number of empty particles within the
5- and 7-day lysate-derived AAV, there must have
been a greater overall number of capsids injected,
increasing the possibility of receptor saturation or
eliciting an immune response. As such, though it is
possible to harvest viable AAV from cellular lysates
at the 5- and 7-day time points, this AAV should be
used with caution.
Although there were no significant changes be-
tween the medium- or cell lysate-purified AAV in
terms of the numbers of cells transduced or the
GFP protein expression after transduction, there
was a nonsignificant trend toward an increase in
GFP protein levels in the striatum of animals re-
ceiving AAV purified from cell lysates. As all vec-
tors used the same promoter, the slight increase
in the GFP protein, in the presence of an equal
number of transduced cells, suggests that either
the MOI or the efficiency of intracellular processing
of the cell lysate-derived virus may be increased.
This result was surprising as this has not been
reported in other publications,
8,9
and there was
actually an increase in the number of empty par-
ticles in the AAV purified from cellular lysates at 5
and 7 days posttransfection. To address this sur-
prising finding we quantified the number of viral
genomes within the striatum of the same animals
Figure 6. Quantification of AAV9 particles purified from producer cell
medium for 7 days posttransfection. AAV9-BFP was produced in HEK 293T
cells by polyethylenimine transfection of the AAV-BFP genome and a
plasmid encoding capsid proteins for AAV serotype 9, and all necessary
helper functions. Each individual production group consisted of two triple
flasks of HEK 293T cells. Medium from HEK 293T producer cells was har-
vested at 3, 5, and 7 days posttransfection. Individual AAV production
groups were terminated and cells harvested at 3, 5, and 7 days post-
transfection, respectively. An additional group of producer cells was in-
cubated for the entirety of time course with no medium change, and cells
were harvested 7 days posttransfection (terminal group). AAV was purified
and total DNase1-resistant viral genomes were quantified. Columns rep-
resent total viral genomes collected at each respective time point +1 SEM
(n=2–3/group) after purification from cellular lysates (black column) or
cellular medium (white columns).
42 BENSKEY ET AL.
using qPCR. Quantification of viral genomes pres-
ent in striatal tissue one month postinjection likely
reflects the relative number of infectious particles
that have delivered their genome to the nucleus, as
the majority of AAV particles remaining in the
cytosol or outside of the cell will be degraded or
transported away from the injection site.
20,21
We
did not detect any differences between the numbers
of intact viral genomes within the striatum of ani-
mals injected with AAV purified from cellular me-
dium versus cellular lysate, confirming that the
infectivity of AAV derived from these two purifi-
cation procedures is commensurate.
The steady-state accumulation of AAV in the
medium was optimal when the medium was col-
lected on alternate days. The total yield of AAV5-
GFP was reduced when collections occurred daily,
and allowing AAV to accumulate in the same me-
dium for a prolonged period of time (7 days) resulted
in an apparent cessation of release. That the total
viral yield was reduced when medium was collected
daily could simply be because of physical agitation
of the cells during medium changes, as we observed
the cells to detach at a much earlier time when
medium was changed daily. This is reflected by the
drop in titers observed around the 5–7-day time
point during daily collections, as compared with a
commensurate drop in titer at the 11–13-day time
point during alternate-day collections. Currently,
the mechanism by which AAV is released into the
medium is unknown. AAV is not associated with
any cytopathogenic effects (as evidenced by the lack
of cell death during the production time course),
and currently there is no known active egress
pathway for AAV to escape the cell. Recent studies
have identified AAV within exosomes; however, the
amount of AAV harvested within these exosomes is
a miniscule fraction of total AAV produced (ap-
proximately 0.01–0.2%),
22
and thus cannot account
for the large amount of free AAV purified from the
medium. It is possible that AAV is released in exo-
somes, which subsequently rupture, resulting in
free AAV in the culture medium. However, this
possibility remains unfounded and further research
into the mechanism(s) by which functional AAV is
released into the culture medium is needed. Ac-
cordingly, because the mechanisms regulating AAV
release into the medium are poorly understood, we
currently do not understand why the additive ac-
cumulation of AAV in the medium breaks down
upon prolonged incubation. We reasoned that this
could be the result of viral particles saturating the
medium and preventing further release, or it could
result from a lack of nutrients and accumulation of
metabolic byproducts when the medium is not re-
freshed, slowing vector production and release.
Here we tested the possibility that the cessation of
virus accumulation within the medium was caused
by a saturation of the medium or a lack of essential
nutrients. Through this experimentation we found
that neither lack of nutrients nor saturation of the
medium with AAV slowed the release of AAV5 into
producer cell medium over a 7-day time course. Ac-
cordingly, it is likely that the cessation of virion ac-
cumulation upon prolonged incubation in the same
medium reflects active breakdown of virion caused
by proteolytic activity in the medium. Based on these
data,wehavedeterminedthattheoptimalfre-
quency for continual harvest of AAV is an alternate-
day collection paradigm.
The ability to continually harvest and purify vi-
able AAV particles from the medium of production
cells will most certainly act as a boon to the field of
AAV gene therapy. Although AAV is a commonly
used viral vector system that has witnessed great
success in both preclinical and clinical settings,
current manufacturing procedures necessitate la-
borious and time-consuming purification proce-
dures in order to obtain AAV particles of sufficient
titer and purity for biomedical applications. Speci-
fically, evolving gene therapy studies and thera-
peutics targeted toward large animals or humans
require large amounts of high-titer AAV. Accord-
ingly, efficient manufacturing of high-quality AAV
vectors has been foreseen as a potential rate-
limitingstep to the future success of AAV in the field
of gene therapy.
5
We believe that the ability to
continually harvest viral particles from the medium
of producer cells may represent a novel solution to
this potential roadblock. As this production tech-
nique increases the yield of intact viral particles by
several fold, with minimal effort and virtually no
added cost, the continuous harvest of AAV particles
will also be of tremendous benefit to small
laboratory-scale production efforts. Further, the
biological activity of the released AAV is at least
commensurate to that of conventional cell lysate-
purified AAV, and in other reports releasedAAV has
actually shown increased biological activity.
8,9
The
ability to harvest AAV from the medium offers
several unique safety advantages over AAV purified
from cell lysates. Because of the large amount of
virus released into the medium, it is not completely
necessary to harvest intracellular virus. This would
obviate the need to separate virus from cellular
contaminants such as lipids, proteins, or carbohy-
drates that increase the potential to elicit an im-
mune response if not purified correctly. Further,
harvesting virus from serum-free medium elimina-
tes the potential contamination of viral preparations
CONTINOUS PRODUCTION OF AAV 43
from zoonotic proteinacious infections particles de-
rived from serum used in medium preparations.
Beyond safety considerations, AAV purified from
medium also has practical production benefits.
Previous methods for purifying AAV have almost
exclusively relied on cellular lysis because of the
canonical belief that AAV is not normally released
from producer cells in any significant amount.
However, the freeze–thaw cycles that are normally
used to lyse producer cells decrease the yield of
viral preparations.
23
As such, the ability to avoid the
freeze–thaw cycles (or other methods of cellular
lysis) is an added benefit to the purification of AAV
from cellular medium. The purification of virus from
medium could be improved by altering the culture
conditions (osmolarity, pH, salt content) in order to
maximize virus release.
8,24
Alternatively, the na-
ture of the AAV capsid itself can be manipulated in
order to maximize release. For example, abolishing
the heparin-binding capacity of AAV vectors in-
creases the proportion of vector collected from the
medium.
9
Finally, the purification of AAV vectors
from medium is scalable, and yield and efficiency
could potentially be maximized using a system to
continually perfuse and collect virus-laden medium
from cells, or alternatively the use of a bioreactor for
very large batches of AAV. Indeed a recent report
has described the use of a nonadherent HEK 293 cell
suspension system in which medium is collected
continuously to purify GMP-grade AAV.
25
In conclusion, here we have extended prior work
showing that AAV is released in culture medium.
Previous publications have demonstrated that
structurally intact and infectious AAV particles
can be harvested from HEK 293 producer cell me-
dium.
8–11
However, the frequency and duration
within which viable, infectious particles can be ef-
ficiently harvested has not previously been deter-
mined. Here we have detailed the time course of
release of functional AAV5, and described the novel
observation that viable, biologically active parti-
cles can be harvested at regular intervals from
cellular medium for up to 7 days posttransfection.
AAV9 was also collected from the medium for up to
7 days posttransfection, although the maximum
duration and optimal interval for harvesting AAV9
from cellular medium must be empirically deter-
mined. In line with the rigor previously used to
ensure the intact physical nature and high biolog-
ical activity of purified AAV,
8–11
we have provided
a thorough and previously unreported character-
ization of the relative infectivity and empty-to-full
particle ratios for both cell lysate- and cell medium-
derived AAV for the 7-day collection time course.
The data presented herein represent an advance in
AAV production as the ability to dramatically in-
crease the yield of viral production through a sim-
ple and efficient method will save both time and
resources, and can expand the potential of AAV in
the field of gene therapy.
ACKNOWLEDGMENTS
We would like to acknowledge Mark Potter at
the University of Florida and Kevin Nash at the
University of South Florida for their assistance in
the establishment of a large-scale AAV manufactur-
ing facility within our laboratory.
AUTHOR DISCLOSURE
The authors have no competing financial
interests.
REFERENCES
1. Flotte TR. Gene therapy progress and prospects:
Recombinant adeno-associated virus (rAAV) vec-
tors. Gene Ther 2004;11:805–810.
2. Daya S, Berns KI. Gene therapy using adeno-
associated virus vectors. Clin Microbiol Rev 2008;
21:583–593.
3. Mueller C, Flotte TR. Clinical gene therapy using
recombinant adeno-associated virus vectors. Gene
Ther 2008;15:858–863.
4. Rivera VM, Gao GP, Grant RL, et al. Long-term
pharmacologically regulated expression of eryth-
ropoietin in primates following AAV-mediated
gene transfer. Blood 2005;105:1424–1430.
5. Monahan PE, Samulski RJ. AAV vectors: Is clinical
success on the horizon? Gene Ther 2000;7:24–30.
6.UrabeM,DingCT,Kotin RM. Insect cells as a
factory to produce adeno-associated virus
type 2 vectors. Hum Gene Ther 2002;13:1935–
1943.
7. Durocher Y, Pham PL, St-Laurent G, et al. Scalable
serum-free production of recombinant adeno-
associated virus type 2 by transfection of 293
suspension cells. J Virol Methods 2007;144:32–40.
8. Lock M, Alvira M, Vandenberghe LH, et al. Rapid,
simple, and versatile manufacturing of recombi-
nant adeno-associated viral vectors at scale. Hum
Gene Ther 2010;21:1259–1271.
9. Vandenberghe LH, Xiao R, Lock M, et al. Efficient
serotype-dependent release of functional vector
into the culture medium during adeno-associated
virus manufacturing. Hum Gene Ther 2010;21:
1251–1257.
10. Okada T, Nonaka-Sarukawa M, Uchibori R, et al.
Scalable purification of adeno-associated virus se-
rotype 1 (AAV1) and AAV8 vectors, using dual ion-
exchange adsorptive membranes. Hum Gene Ther
2009;20:1013–1021.
11. Grieger JCJ, Samulski RJR. Packaging capacity of
adeno-associated virus serotypes: Impact of larger
genomes on infectivity and postentry steps. J Virol
2005;79:9933–9944.
12. Zolotukhin S, Potter M, Zolotukhin I, et al. Production
and purification of serotype 1, 2, and 5 recombinant
adeno-associated viral vectors. Methods 2002;28:
158–167.
44 BENSKEY ET AL.
13. Morganti JM, Nash KR, Grimmig BA, et al. The
soluble isoform of CX3CL1 is necessary for
neuroprotection in a mouse model of Parkinson’s
disease. J Neurosci 2012;32:14592–14601.
14. Reimsnider S, Manfredsson FP, Muzyczka N,
Mandel RJ. Time course of transgene expression
after intrastriatal pseudotyped rAAV2/1, rAAV2/2,
rAAV2/5, and rAAV2/8 transduction in the rat.
Mol Ther 2007;15:1504–1511.
15. Gundersen HJ, Jensen EB. The efficiency of sys-
tematic sampling in stereology and its prediction.
J Microsc 1987;147:229–263.
16. Gombash SE, Manfredsson FP, Mandel RJ, et al.
Neuroprotective potential of pleiotrophin over-
expression in the striatonigral pathway compared
with overexpression in both the striatonigral and ni-
grostriatal pathways. Gene Ther 2014;21:682–693.
17. Benskey MJ, Kuhn NC, Galligan JJ, et al. Targeted
gene delivery to the enteric nervous system using
AAV: A comparison across serotypes and capsid
mutants. Mol Ther 2015;23:488–500.
18. Zolotukhin S, Byrne BJ, Mason E, et al. Re-
combinant adeno-associated virus purification using
novel methods improves infectious titer and yield.
Gene Ther 1999;6:973–985.
19. Lockhart PJ, O’Farrell CA, Farrer MJ. It’s a double
knock-out! The quaking mouse is a spontaneous
deletion of parkin and parkin co-regulated gene
(PACRG). Mov Disord 2004;19:101–104.
20. Cearley CN, Wolfe JH. A single injection of an adeno-
associated virus vector into nuclei with divergent
connections results in widespread vector distribution
in the brain and global correction of a neurogenetic
disease. J Neurosci 2007;27:9928–9940.
21. Douar AM, Poulard K, Stockholm D, Danos O. In-
tracellular trafficking of adeno-associated virus vec-
tors: Routing to the late en dosomal compartment and
proteasome degradation. J Virol 2001;75:1824–1833.
22. Maguire CA, Balaj L, Sivaraman S, et al.
Microvesicle-associated AAV vector as a novel
gene delivery system. Mol Ther 2012;20:960–971.
23. Howard DB, Fortuno LV, Harvey B. Stability and
inactivation of AAV serotype 1 vectors. Eighteenth
Annual Meeting of the American Society for Gene
and Cell Therapy, 2015.
24. Atkinson EM, Takeya R, Aranha I. Methods for
generating high titer helper-free preparations of
released recombinant AAV vectors. U.S. Patent No.
0266567 2005.
25. Grieger JC, Soltys SM, Samulski RJ. Production of re-
combinant adeno-associated virus vectors using sus-
pension HEK293 cells and continuous harvest of vector
fromtheculturemediaforGMPFIXandFLT1clinical
vector. Mol Ther 2015. doi: 10.1038/mt.2015.187.
Received for publication August 17, 2015;
accepted after revision January 22, 2016.
Published online: January 25, 2016.
CONTINOUS PRODUCTION OF AAV 45
