12_ANIMAL CELL TECHNOLOGY_707

ADENO-ASSOCIATED VIRAL VECTORS : PRINCIPLES AND IN
VIVO USE

OLIVER DANOS
Gene Therapy Program. Généthon, CNRS URA 1922,
Ibis, rue de l’Internationale, BP 60
91002 Evry, France.

1. Biology of Adeno-Associated Viruses

Small non-enveloped viruses with an icosahedral capsid and a single-stranded DNA
genome are grouped in the family of Parvoviridae. The virions are usually around 20
nm in diameter and are extremely resistant to a variety of harsh physical conditions
(low pH, heat, presence of detergent). Within this family, Dependoviruses where first
identified as contaminants of human adenovirus preparations and shown to be strictly

dependent on the presence of the helper adenovirus for their replication. These Adeno-

associated viruses (AAV) are found in a variety of birds and mammals including
humans. Five human serotypes have been identified, and around 70 % individuals are
seropositive in the general population. However, no pathology has been associated

with AAV infection and although the AAV genome is known to integrate the host
cell genome under certain conditions, no vertical transmission has been documented.
For a complete review on Parvoviruses and AAV, see (1).
AAV particles are able to infect a variety of cell type in culture. Viral capsids
entering the cytosol are rapidly transported to the nucleus were the single stranded

genome is delivered. In the presence of a helper virus, such as Adeno, Herpes or
Vaccinia virus, the genome is actively transcribed and replicated. This step is
performed mostly by the cellular DNA and RNA synthesis machineries and requires
the presence of AAV proteins encoded by the rep gene. It is most efficiently
activated by a co-infecting helper virus, but more generally, it can be seen in
response to a variety of cellular stresses (2). Viral capsids assemble in the nucleus,
package the replicated genome and virions are liberated upon cell lysis (3).
In the absence of helper functions, the AAV genome is not productively transcribed
or replicated. The single stranded (ss) DNA genome is converted into a double
stranded (ds) form and eventually integrates into the cellular DNA. When human
cells are infected with helper free AAV, tandem arrays of the genome are found at a
preferred location on chromosome 19. This site specific integration requires the
presence of Rep proteins (4). The latent genomes can be reactivated and rescued by an
infection with a helper virus (5).
The 4679 nucleotide genome packaged into AAV virions can be of plus or minus
polarity. It contains a 145 nt inverted terminal repeat (ITR) of which the first 125 nt
can form a T-shaped hairpin structure. The ITR contains all the information needed
in cis for replication, packaging and integration of the genome (6). Viral proteins are

493

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 493-502.
© l998 Kluwer Academic Publishers. Printed in the Netherlands.

494

encoded by two genes, rep and cap. Multiple proteins are translated from
differentially initiated or spliced transcripts. There are 4 Rep (MW 78, 68, 52 and 40)
and 3 Cap proteins (VP1, VP2, VP3). The protein composition of AAV virions is
80% VP3.
The smaller Rep proteins are involved in viral RNA maturation and packaging. The
two larger Rep proteins display site specific endonuclease, ATPase and helicase
activities (7) and bind repeated copies of the motif GAGC on double stranded DNA.
Binding sites are found: i) on the A stem of the ITR; ii) within the three AAV
promoter regions (8) and iii) at several loci the human genome, including one close
to the AAV preferential integration site on chromosome 19 (4 , 9). The current view
of the role of the large Rep proteins is that they: i) resolve the closed T-shaped end
of the double stranded replication intermediates and allow for the formation of
monomeric single stranded genomes ; ii) modulate AAV gene expression through
transcriptional regulation (10) ; iii) mediate site specific integration, possibly by
bridging the viral and cellular DNA binding sites (11).

2.AAV-derived Vectors

2.1. PRINCIPLE AND METHODS

AAV vectors only retain the duplicated 145 nt ITR from the viral genome, the rep

and cap genes being removed and replaced by the sequences of interest. Vector

particles can theoretically be produced in the absence of wild type AAV, if the rep

and cap gene products are provided in trans (6). The initial method to achieve this

was to transfect cells with the vector and a rep-cap expression cassette on two

separate plasmids (12, 13). Human 293 cells, which are highly permissive to both

AAV and adenovirus replication and can be efficiently transfected, are of general use

for vector production. Following transfection, helper functions are provided by

infecting the cells with Adenovirus. Since the rep and cap sequences are not linked to

the AAV ITR, only the recombinant construct is replicated and packaged.

Recombinant virions, free of wt AAV can be purified from cell lysates and separated

from the Adenovirus particle by density centrifugation. This procedure can be

optimized to yield recombinant AAV (rAAV) particles per transfected cell,

which is within the range of wild type AAV production (13, 14).

Still, the current method presents several limitations : i) a typical prepartion is very

work intensive, since it involves the transfection of cells, followed by a

week-long purification and assay procedure (13); ii) each viral preparation is made

from a new transfection, and reproducibility problems arise ii) a low level

contamination of the final stock by the helper adenovirus is unavoidable, following

physical separation (14); iii) illegitimate recombination, which may occur during the

transfection, or under the influence of Rep results into structures where ITR

sequences are linked to rep and cap. These chimeric forms can subsequently evolve

into replication competent AAV genomes, through additional recombination (15).

Several significant improvements have been made to the transient transfection

method. Adenovirus infection can now be replaced by the cotransfection of a cloned

Adenovirus genome deleted for the packaging sequence and part of the El region

(unpublished observation). This allows for the production of rAAV particles in the

495

absence of functional Adenovirus. Further deletions in the Adenovirus genome,
eliminating most of the late (L) genes, have also been shown to provide efficient
help for rAAV production (X Xiao and RJ Samulski, personal communication).
Following this method, adenoviral capsids are eliminated and cleaner prepartions
should be obtained. It was also observed that the level of Rep proteins produced after
transfection was critical for obtaining high titer rAAV. Surprisingly, low amounts
of rep expression result in higher rAAV production, probably because of the
negative influence of Rep on cap expression. Helper (rep-cap) plasmids designed to
express low levels of Rep allow for the production of 5-10 fold more rAAV (16). In
achieving high titer preparations, the ability to synthetize large amounts of Cap
proteins is critical (14).
One obvious goal is to obtain stable packaging systems for the production of rAAV.
This would not only simplify the preparation method, but also facilitate the
definition of specifications for the production of vectors to be used in pre-clinical and
clinical experiments. The development of a universal packaging line is underway.
Clones stably expressing rep and cap as well as a vector construct have been used to
produce helper free rAAV, following Adenovirus infection (17). The isolation of
« proto-producer » clones in which rAAV production is triggered by Adenovirus
infection, requires more work initially, but they result in permanent sources of
rAAV that can be mobilized by Ad infection. This approach is likely to become of
general use, once its reproducibility will be established.

2.2.FACTORS INFLUENCING AAV-MEDIATED GENE TRANSFER
EFFICIENCY

The understanding of the replication mechanism of AAV is only partly relevant to
the situation encountered with vectors. Like the parental wild-type (wt) AAV,
recombinant genomes entering the cell nucleus must be converted to a dsDNA in
order to be transcribed. In wt AAV the Rep proteins either directly facilitate second
strand synthesis, or can trigger the (still undefined) cellular functions involved. On
the other hand, vectors do not retain the rep gene and the Rep proteins are not
detectably associated with the released viral particles. Thus the recombinant genomes
uniquely depend on the cell DNA synthesis machinery for second strand synthesis.
This implies that the expression of sequences transferred by an AAV vector will be
highly dependent on the presence of an appropriate cellular environment. Different
situations have been shown to enhance rAAV-mediated gene transfer by allowing for
a more efficient second strand synthesis. One is the expression in the target cell of
the open reading frame 6 from the E4 region of Adenovirus (18, 19), a protein that
can bind p53 (20), and may therefore interfere with cell cycle signals. More
generally, drug or radiation induced genotoxic stresses, have an enhancing effect on
rAAV-mediated gene transfer in cultured cells, or in vivo (20-24). Although a
unifying explanation for these observations is still lacking, the initiation of
unscheduled DNA repair synthesis is probably central for rAAV second strand
synthesis.

3.AAV vectors for direct gene transfer in vivo:

496

A number a features of rAAV make them well suited for direct gene transfer in vivo.
Compared to retroviruses or adenoviruses, the rAAV particles are small and very
resistant, and may therefore remain in the circulation or in an injected tissue for
longer times, and diffuse more efficiently across some the natural barriers within the
organism. In addition, rAAV should, in principle, be able to transduce non-dividing
cells, which constitute the overwhelming majority of targets encountered in vivo.
Limitations include, the size of the transferred sequences and, in the prospect of
human clinical trials, the widespread seropositivity for AAV in human populations.
Over the past couple of years, the methods for rAAV production, have become
efficient enough to yield preparations with titer and quality compatible with in vivo
experimentation. A number of in vivo targets have been explored, including central
and peripheral nervous system (23, 25-30, 39), repiratory epithelium (31), vascular,
cardiac and skeletal muscle (32-36) and liver (24). Here we discuss our experiments
with skeletal muscle and liver (37, 38).

3.1.GENE TRANSFER TO THE SKELETAL MUSCLE

Our interest was to further document the remarkable efficiency of rAAV for gene
transfer into the skeletal muscle and to analyze the structure of the recombinant
genome in the transduced muscle fibers. A secondary goal was to start to analyse the
contribution of the post-mitotic environment found in the terminally differentiated
muscle fiber, to the high-efficiency of AAV-mediated gene transfer.
The efficiency of AAV-mediated gene transfer into the skeletal muscle of mice was
evaluated by performing a single injection of purified recombinant AAV (rAAV)
particles encoding a modified E. coli ß-galactosidase with a nuclear localization
signal, under the control of the CMV promoter. The rAAV preparations were titered
by dot blot hybridizationand (1012 physical particles / ml) and by limiting dilution
infections and X-gal staining of 293 cells
in the absence of presence of adenovirs co-infection, respectively. The stocks were
determined to be free of detectable contamination by adenovirus and by replication
competent AAV less than 1wt AAV / 109 rAAV). Young adult Balb/c mice were
injected into the quadriceps muscle with 30 (m1 of rAAV preparationcontaining

gal infectious units and animals were sacrificed between 2 and 28 weeks later.
Histological sections of the injected muscles, stained for ß-galactosidase activity
revealed between 10 and 70% positive fibers on more than 75% of the muscle
length.
Two weeks after gene transfer, the muscle had a normal histological aspect, whereas
on sections taken at four and eight weeks, infiltrates became conspicuous in the areas
containing positive fibers. Regenerated fibers with centrally located nuclei were also
noted in the infiltrated areas. Remarkably, these fibers still expressed the transgene
and may therefore have arisen through the proliferation and fusion of transduced
satellite myoblasts containing the recombinant AAV genome. The cellular response
was not observed at later time points. In conclusion, this gene transfer procedure is
associated with a transient and limited immune response, without any major impact
on the long term genetic modification of muscle cells.
We then asked whether an enhancement of gene transfer is observed when muscle
satellite cells are induced to proliferate as a consequence of muscle injury. The same
gene transfer procedure was applied to animals pretreated 48 hours before by an intra-

497

muscular injection of barium chloride (BaC12) which provoks a rapid necrosis of the

muscle fibers, followed by tissue regeneration. In animals analysed 2 weeks after

gene transfer, over 85% of the muscle fibers had been regenerated and displayed

centrally located nuclei.

Surprisingly, most of the galactosidase positive fibers were confined to the

remaining, non regenerated, areas and accounted for less than 20% of all fibers. At
this time point, cellular infiltrates were observed in the regenerated areas, but not in
the intact tissue. At late time points, the infiltrates became much less pronounced,

and only limited areas of positive and mostly regenerated fibers remained. We

concluded that there is no obvious enhancement of gene transfer when the AAV

vector is applied to a regenerating muscle containing proliferating cells. On the
contrary, the terminally differentiated, post-mitotic, muscle fibers appear to be

preferentially permissive to AAV vectors.

Another group of mice received a single intra-muscular injection of a rAAV encoding
murine erythropoietin (Epo, the hormone regulating erythopoiesis), under the control

of the CMV promoter. Each animal was injected with total vector

particles, blood samples were collected over time and Epo production was monitored

either directly by a radio-immunoassay on plasma, or indirectly by measuring the

hematocrit which reflects the number of circulating erythrocytes. The hematocit of

every animal increased during the first four weeks following gene transfer, and

reached a plateau value of 80 to 90%. These high hematocrits were observed until the

animals were sacrificed after 2 to 4 months. They corresponded to 10 to 20 fold

increase in serum Epo levels.
The structure of the rAAV genomes was analyzed by Southern blot on high

molecular weight DNA prepared at different time points from the injected muscles.

Eight weeks after injection, 1 to 3 copies of rAAV per haploid genome were

measured. This material was associated with high molecular weight DNA, under the

form of concatemers. No proof of integration, like the characterisation of junction

between cellular and vector genome, has been obtained. It is possible that the rAAV

genome remains extrachromosomal, under the form of either head-to-tail tandem

repeats or interlocked circles.

The status of the vector DNA was also analyzed at early time points following

injection. After one and two days, a strong signal is found corresponding to the input

ssDNA genome. This signal progressively decreases until 2 weeks. Double stranded

monomers of the genome are also observed during the first days following gene

transfer. Although this could represent an artefactual reassociation of the

complementary strands of DNA delivered by the rAAV particles, it may also reflect

the high permissivity of the muscle fiber. This ds monomers progressively disappear

during the first two weeks, and are chased into the high molecular weight forms. In

conclusion, we observe two successive transformation of the recombinant genome

upon entry into the muscle fiber nuclei : a conversion into dsDNA monomer,
followed by a concatemerisation.

3.2.GENE TRANSFER INTO THE LIVER

The steady state liver contains mostly non dividing cells, 90% of which are
hepatocytes. These hepatocytes are normally quiescent but they are induced to divide

in response to liver injury.

498

The quiescent or regenerating liver can be subjected to gene transfer using a variety
of vectors by portal infusion. We have examined the potential of rAAV in this
system by administrating a preparation of rAAV carrying the human Factor IX
cDNA to adult C57B1/6 mice (37). Factor IX is a component of the blood

coagulation cascade, normally produced by the liver. Its deficiency results in

Hemophilia B. Beetween 2 and vector particles were infused in the portal

vein and the presence of human Factor IX was periodically measured in plasma
samples. The human protein was detectable at low levels during the first week and
increased to steady state concentrations of 250 to 2000 ng/ml. The secreted Factor IX
was active in coagulation assays and the serum levels were dose dependent and

persisted for at least 36 weeks (the duration of the experiment). When extrapolated to

a clinical situation, such concentrations would be relevant for the treatment of

Hemophilia B.
The amount of rAAV genomes present in the infused livers was measured by

Southern blot to be between 1 and 4 copies per haploid cellular genome. In situ

hybridization showed that human Factor IX mRNA could be found in only 1 to 5 %

of the hepatocytes, implying that either a few cells contain and express multiple (30

to 50) copies of the rAAV genome, or that most cells contain vector genomes and
only a minority expresses them. Considering the number of target cells in the mouse

liver and the estimated amount of infused infectious particles it is

unlikely that most cells can be transduced. As a comparison, over: adenoviral

particles are needed to transduce most cells in the liver . We therefore favor the

hypothesis where the gene transfer procedure would deliver the ssDNA genome to 1

to 5 % of the liver cells, and after conversion into the ds form, an amplification step

would take place.

4. Conclusions

AAV vectors have now proven to be very efficient for stable gene delivery into a
number of in vivo targets. At this time, preparation of high quality and high titer
rAAV, although dramatically improved over the past couple of years, remain a
significant bottleneck. Yet, specifications for clinical grade material are starting to be
defined and phase I clinical trials involving rAAV as vector are now underway.

In animal experiments, high gene transfer efficiency is seen in myotubes and
neurones, which are post-mitotic cells, and possibly in a sub-population of
hepatocytes for which the cell cycle status is unknown.
Not all arrested primary cells, however are permissive to rAAV gene transfer. Bone
marrow cells enriched in non cycling CD34+ hematopoietic progenitors will
accumulate rAAV DNA, but mostly fail to express the transfered gene (C.Jordan ,
personal communication). Understanding the determinants of cellular permissivity to
AAV will have important implications for the definitions of optimal gene transfer
targets in vivo.

499

5. References

1. Berns, K. I. (1996) in Fields Virology, eds. Fields, B., Knipe, D. & Howley, P.
(Lippincott-Raven, Philadelphia), Vol. 2, pp. 2173-2220.

2. Schelehofer, J. R., Matthias, E. & Zur Hausen, H. (1986) Virology 152, 110-
117.

3. Wistuba, A., Kern, A., Weger, S., Grimm, D. & Kleinschmidt, J. (1997) J.
Virol 71, 1341-1352.

4. Linden, R. M., Winocour, E. & Berns, K. I. (1996) Proc. Natl. Acad. Sci.
U.S.A. 93, 7966-7972.

5. McLaughlin, S. K., Collis, P., Hermonat, P. L. & Muzyczka, N. (1988) J.
Virol. 62, 1963-1973.

6. Muzyczka, N. (1992) Curr. Top. Microbiol. Immunol. 158, 97-129.
7. Im, D. & Muzycska, N. (1990) Cell 61, 447-457.
8. Me Carty, D. M., Pereira, D., Zolotukhin, I., Zhou, X., Ryan, J. H. &

Muzyczka, N. (1994) J. Virol. 68, 4988-4997.
9. Wonderling, R. & Owens, R. (1997) J. Virol. 71, 2528-2534.
10.Pereira, D., McCarty, D. & Muzyczka, N. (1997) J. Virol 71, 1079-1088.
11.Weitzman, M. D., Kyostio, S. R. M., Kotin, R. M. & Owens, R. A. (1994)

Proc. Natl. Acad. Sci. U.S.A. 91, 5808-5812.
12.Samulski, R. J., Chang, L. S. & Shenk, T. (1989) J. Virol. 63, 3822-3828.
13.Snyder, R., Xiao, X. & Samulski, R. J. (1996) in Current Protocols in Human

Genetics, ed. Haines, J., pp. 12.1.1-121.1.23.
14. Vincent, K., Piraino, S. & Wadsworth, S. (1997) J. Virol. 71, 1897-1905.
15. Allien, J., Debelak, D., Reynolds, T. & Miller, A. (1997) J. Virol 71, 6816-

6822.
16.Li, J., Samulski, R. J. & Xiao, X. (1997) J Virol 71, 5236-5243.
17.Clark, K., Voulgaropoulou, F., Fraley, D. & Johnson, P. (1995) Hum. Gene

Ther. 6, 1329-1341.
18.Ferrari, F., Samulski, T., Shenk, T. & Samulski, R. (1996) J. Virol. 70, 3227-

3234.
19.Fisher, K. J., Gao, G. P., Weitzman, M. D., De Matteo, R., Burda, J. F. &

Wilson, J. M. (1996) J. Virol. 70, 520-532.
20.Dobner, T., Horikoshi, N., Rubenwolf, S. & Shenk, T. (1996) Science 272,

1470-1473.
21. Alexander, L, Russell, D. & Miller, A. (1994) J. Virol. 68 N°12, 8282-8287.
22.Russell, D. W., Alexander, I. E. & Miller, A. D. (1995) Proc. Natl. Acad. Sci.

U.S.A. 92, 5719-23.
23. Alexander, I., Russell, D., Spence, A. & Miller, A. (1996) Hum. Gene Ther. 7 ,

841-850.
24.Koeberl, D. D., Alexander, I. E., Halbert, C. L., Russels, D. W. & Dusty

Miller, A. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 1426-1431.
25.Kaplitt, M., Leone, P., Samulski, R., Xiao, X., Pfaff, D., O'Malley, K. &

During, M. (1994) Nat. Genet. 8.
26.Ali, R. R., Reichel, M. B., Thrasher, A. J., Levinsky, R. J., Kinnon, C.,

Kanuga, N., Hunt, D. M. & Battacharya, S. S. (1996) Hum. Mol. Genet. 5 ,
591-594.

500

27.Lalwani, A. K., Walsh, B. J., Reilly, P. G., Muzyczka, N. & Mhatre, A. N.
(1996) Gene Therapy 3.

28.McCown, T. J., Xiao, X., Li, J., Breese, G. R. & Samulski, R. J. (1996) Brain
Res. 713, 99-107.

29.Peel, A., Zolotukhin, S., Schrimsher, G. W., Muzyczka, N. & Reier, P. J.
(1997) Gene Therapy 4, 16-24.

30.Flannery, J.G., Sergei Zolotukhin, S., Isabel Vaquero, M., LaVail, M.M.,
Muzyczka, N. & Hauswirth, W. (1997) Proc. Natl. Acad. Sci. USA 94, 6916-
6921.

31.Afione, S. A., Conrad, C. K., Kearns, W. G., Chunduru, S., Adams, R.,
Reynolds, T. C., Guggino, W. B., Cutting, G. R., Carter, B. J. & Flotte, T. R.
(1996) J.Virol 70, 3235-3241.

32.Kaplitt, M. G., Xiao, X., Samulski, R. J., Li, J., Ojamaa, K., Klein, I. L.,
Makimura, H., Kaplitt, M. J., Strumpf, R. K. & Diethrich, E. B. (1996) Ann
Thorac Surg 62, 1669-1676.

33.Xiao, X., Li, J. & Samulski, R. J. (1996) J. Virol. 70, N°11, 8098-8108.
34.Kessler, P. D., Podsakoff, G. M., Chen, X., McQuiston, S. A., Colosi, P. C.,

Matelis, L. A., Kurtzman, G. J. & Byrne, B. J. (1996) Proc. Natl. Acad. Sci.

U.S.A. 93, 14082-14087.
35.Fisher, K. J., Jooss, K., Alston, J., Yiping , Y., Haecker, S. E., High, K.,

Pathak, R., Raper, S. E. & Wilson, J. M. (1997) Nature Medicine 3, 306-312.
36.Herzog, R., Hagstrom, J., Kung, S., Tai, S., Wilson, J., Fisher, K. & High, K.

(1997) Proc. Natl. Acad. Sci. USA 94, 5804-5809.
37.Snyder, R., MIiao, C., Patijn, G., Spratt, K., Danos, O., Nagy, D., Gown, A.,

Winther, B., Meuse, L., Cohen, L., Thompson, A. & Kay, M. (1997) Nat.
Genet. 16, 270-276.
38.Snyder, R., Spratt, S., Lagarde, C., Bohl, D., Kaspar, B., Sloan, B., Cohen, L.
& Danos, O. (1997) Hum. Gene Ther. 8, 1891-1900.
39.Jomary, C., Vincent, K.A., Grist, J., Neal, M.J. & Jones, S.E. (1997) Hum.
Gene Ther. 4, 683-690.

Discussion 501
Lupker:
What will happen if someone treated with your AAV factors gets a
Danos: superinfection by Adenovirus?

Zhang: To get a re-mobilisation of the vector, you need to have products
Danos: of the rev gene. So if the AV preparation is free of rev genes or
wild type AV, then nothing will happen. If you have a cell with an
Zhang: integrated recombinant pro-virus and you infect this with an AV,
Danos: you do not rescue the recombinant structure. You need the AV
functions as well.
Ostrove:
It is puzzling that in your slides you showed a gradual decrease in
the DNA but at the same time you showed that the virus was
stable. Can you comment on this?

There is a decrease in the overall DNA signal but it does not
disappear. Only a minority of the genomes are being stabilised and
converted into whatever form is needed for them to be expressed.
So probably only 1% of the input ends up being expressed. At the
beginning you see a lot of the input DNA, and it is difficult to see
anything on the blots.

What happens if you do a Northern Blot - do you see a stable
messenger RNA?

We do not know when the expression really starts. One can see
the expression going progressively up over time. The number of
cells and the intensity of staining is much less. So there is a
progressive recruitment of the single stranded DNA into an active
form. The problem we have is to define this active form. One
obligative pathway for this active form is double stranded
conversion.

A question regarding the gal experiment. You showed that the
addition of adenovirus increased your transduction efficiency by
approximately 2 logs. Can you comment on that observation, and
is it true for in vivo as well as in vitro, or are all the data in vitro?

502 Most of the data are for in vitro. I discussed adenovirus helper
Danos: function in terms of production of AV particles. Adenovirus can
also help gene expression from the recombinant AV. This has been
Berthold: mapped now in the E4 region of the virus. There is one particle
Danos: open-reading frame, a protein, that is central for that. There is
debate over transduction with AV because of possible
contamination with adenovirus. There is now no problem as you
can prepare AV without any adenovirus by using mini-plasmids.
Using these preparations you do see transductions in vivo. If you
add adenovirus to liver you do see an increase and there is a 10
fold increase in Factor IX production transiently, which then drops
totally. This is due to an immune reaction against the adenovirus
which eliminates the transduced hepatocytes.

A question on the safety aspects: you described a transient
inflammatory response - is this related to non-mouse sequences, or
to the virus infection? Also, how do you make sure that other
components of your production system are not replicating as well?

You do get an immune response against the particles but this is
only seen when you re-administer. It is a cellular immune response
seen after 4 weeks. It is not an acute inflammation. We have
looked for wild type AV, and wild type adenovirus, or El deleted
adenovirus. We do not detect any gross contamination, or
anything that can create a significant response in this animal
system. If we use this in the clinic we will have to be much more
thorough with the characterisation of the preparation.

NOVEL RETROVIRAL PACKAGING CELL LINES: IMPROVED
VECTOR PRODUCTION FOR EFFICIENT HEMATOPOIETIC
PROGENITOR CELL TRANSDUCTION

Sean Forestall, Richard Rigg, Jonathan Dando, Jingyi Chen, Barbara Joyce,
Robert Tushinski, Chris Reading, and Ernst Böhnlein

SyStemix Inc., 3155 Porter Drive, Palo Alto, CA 94304, USA

1. Introduction

Safe and efficient means to transfer therapeutic genes stably must be developed
for successful hematopoietic cell-based gene therapy. Currently, retroviral vectors offer
the only practical way to integrate genes stably, although retroviral gene transfer to
primary hematopoietic stem and progenitor cells has been low [1-2]. While amphotropic
vectors have been used in the majority of studies, gibbon ape leukemia virus (GaLV)
enveloped vector has been reported to transduce hematopoietic progenitor cells and
peripheral blood lymphocytes (PBL) more efficiently than amphotropic vector [3-4],
suggesting that targeting different retroviral receptors may improve gene transfer
efficiencies into these lineages.

High efficiency gene transfer into human hematopoietic progenitor cells has
been achieved with retroviral vector generated from the novel ProPak packaging cell
lines. The ProPak cell lines have complementary tropisms producing either murine
leukemia virus (MLV) xenotropic (ProPak-X cells) or amphotropic envelope (ProPak-A),
and were designed to minimize the risk of replication competent retrovirus generation [5-
6]. Vector supernatants from ProPak or existing packaging cell lines producing different
pseudotyped particles (amphotropic MLV, xenotropic MLV, or gibbon ape leukemia
virus) were compared for the ability to transduce clinically relevant human hematopoietic
cells. All vector types transduced primary human CD34-positive cells regardless of
tropism. However, consistently higher transduction of target cells was achieved with
ProPak-derived amphotropic vector than with PA317-packaged amphotropic vector.
Improved conditions for the production of vector supernatants by the coculture of the
complementary ProPak packaging cell lines in a packed-bed bioreactor are presented that
result in 100% transduction efficiencies of human hematopoietic progenitor cells.

2. Materials and Methods

Packaging and producer cell lines(proPak-A, ProPak-X, PA317, PG13) were
grown in Dulbecco's modified Eagle’s medium (high glucose DMEM: JRH Biosciences,

503

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 503-508.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

504

Lenexa, KS, USA) supplemented with fetal bovine serum (FBS: HyClone Laboratories
Inc., Logan, UT, USA). Vector supernatants were produced in a 500 ml packed-bed
bioreactor with 10 g of Fibra Cell discs (New Brunswick Scientific, Edison, NJ, USA).
All supernatants used for gene transfer experiments were free of RCR as determined by
the extended S+L- assay on PG4 cells (CRL 2032; ATCC). The vector used in all
experiments (LMiLy: LTR-RevM10-IRES-Lyt2) is described elsewhere [7].

PEL were isolated from peripheral blood mononuclear cells and depleted of
CD8-positive cells. The CD34-positive (CD34+) fraction was isolated from mobilized
peripheral blood (MPB) or adult bone marrow (ABM) using a positive selection device.
Further purification to obtain the fraction carrying the Thy1 antigen (CD34+/Thy 1+)
was achieved by high-speed flow cytometric sorting. Hematopoietic stem and progenitor
cells were cultured in a 1:1 mixture of IMDM and RPMI media (JRH Biosciences)
containing 10% FBS, interleukin-3 (IL-3) and IL-6 (20 ng/ml each; Sandoz Pharma,
Basel, Switzerland), and stem cell factor (SCF; 100 ng/ml) or leukemia inhibitory factor
(LIF; Sandoz) (100 ng/ml). Cell lines were inoculated with vector once at unit gravity
for 3 h at 37°C. Primary cells were exposed to vector supernatants under centrifugation,
termed spinoculation, at 2550 g for 3 or 4 h. Polybrene (8 mg/ml; Sigma) or protamine
sulfate (4 mg/ml; Sigma) was added for cell lines or primary cells, respectively.

Transduction efficiencies achieved with Lyt2-encoding vector supernatants were
quantitated as the proportion of target cells that expressed the Lyt2 antigen 2 or 3 days
following inoculation. Integration of the RevM10 gene into clonogenic progeny (CFU-
C) was determined using aDNA PCR assay [8]. At least 92 colonies were analyzed for
each condition to allow precision in experimental comparisons. Colonies were scored
positive if both RevM10 and the endogenous beta-globin sequences were detected.

3. Results

ProPak Packaging Cells

The derivation of the ProPak packaging cell lines has been described elsewhere
[5-6]. Briefly, separate minimal env and gag-pol expression plasmids with heterologous
promoters were stably transfected into an MLV-free cell line (human 293). These steps
were taken to minimize the chances of recombination which could lead to RCR. Using a
vector known to reproducibly give rise to RCR in PA317 producer cells, ProPak cells
were shown to be RCR-free using conditions favorable to RCR formation (ping-pong
amplification of the vector by coculture of both the amphotropic and xenotropic producer
cells). Further, vector produced by either amphotropic or xenotropic ProPak producer
cells is resistant to inactivation by human serum making it suitable for in-vivo gene
transfer applications.

Improved Retroviral Vector Supernatant Production in a Bioreactor

We have previously shown that a packed-bed bioreactor is applicable to the
production of retroviral vector using PA317-based producer cells [9]. We have since

505
determined that the packed-bed bioreactor is suitable for vector production with all 3T3-
and 293-based producer cell lines tested. In comparison with T-flask cultures, bioreactor
supernatants have yielded 2- to 20-fold higher transduction of cell lines (data not shown).
Similar to results on cell lines, bioreactor supernatants have yielded approximately 3-fold
higher transduction of CD34+ cells than T-flask supernatants as shown in Figure 1.

Primary Cell Transduction with Different Vector Tropisms
Using the packed-bed bioreactor, vector supernatants from PA317, PG13,

ProPak-A and ProPak-X producer cell clones, all carrying the LMiLy vector, were
produced. These vector supernatants with different tropisms were compared for their
ability to transduce CD4+ PBL and primary human hematopoietic progenitor and stem
cells (defined as CD34+ or the more highly purified CD34+ Thy1+ expressing cells).
Primary cells were exposed to vector once to directly compare transduction efficiencies
which are presented as Lyt2 surface marker expression relative to the expression achieved
with PA317-packaged vector (Table 1). Regardless of tropism, all vector types
successfully transduced CD34-positive cells isolated from ABM or MPB, or CD4+ PBL
(Table 1). This implies that functional receptors for all three vector types are expressed
on these cells. While no single vector tropism appeared to mediate significantly higher
levels of transduction than any other, the highest transduction efficiencies were achieved
with ProPak-A or ProPak-X supernatants (Table 1). Most significantly, amphotropic
vector supernatants from the human ProPak-A cell lines consistently transduced a higher
proportion of target cells than amphotropic vector prepared from PA317-based producer
cells.

506

Table 1. Transduction of primary human hematopoietic cells with vector supernatants
of different tropisms from LMiLy-based producer cells cultured in the packed-bed
bioreactor. Transduction efficiencies were measured as the proportion of cells expressing
Lyt2, and values have been normalized to that achieved with PA317-based supernatants.
Duplicate samples from 4 different tissues were inoculated as indicated.

Vector Production in Producer Cell Cocultures

Next we attempted to increase transduction efficiency by producing vector
supernatants by coculture of the complementary ProPak-X.LMiLy and ProPak-A.LMiLy
producer cells in the packed-bed bioreactor. This technique, known as ping-pong
amplification, results in higher titers as a result of increased vector copy number [10-12].
The problem that has arisen in the past with coculture production is the generation of
RCR [11]. We have already shown that no RCR arises during extended coculture of
ProPak-A and ProPak-X cells carrying the BC140revM10 vector [6], an event that is
even less likely with the LMiLy vector which contains the shorter psi packaging
sequence, and therefore lacks sequences overlapping with the gag/pol structural genes.

Using the coculture supernatant, we investigated different cytokine
combinations and multiple rounds of inoculation in an effort to optimize gene transfer.
CD34+ cells isolated from MPB were spinoculated with PP-A/X.LMiLy vector once,
twice on the same day, or once each day on two consecutive days in the presence of IL-3,
IL-6, and either LIF or SCF. The results in Table 2 show that while gene transfer was
higher in cultures treated with SCF after a single inoculation, gene transfer into CFU-C
was 100% for cultures inoculated twice on consecutive days in either LIF or SCF.
Furthermore, in both cytokine cocktails the Lyt2 transgene was expressed in up to 40%
of the total cell population, and 21% of the total cell population co-expressed the Lyt2
transgene and CD34 antigen as shown in Figure 2. Not surprisingly, the gene transfer
efficiency measured by DNA PCR is consistently higher than that measured by transgene
expression. This is likely due to the differences in the assays, and the inability of
quiescent hematopoietic cells to express the transgene until activated. We are, however,
confident that the gene marking is biologically relevant as parallel studies from our
group show that transduction of hematopoietic progenitor cells leads to sustained

marking and expression in lymphoid and myeloid progeny [8].

507

4. Discussion
We have constructed safe, new retroviral vector packaging cell lines which

stably produce particles that target distinct receptors on human cells, are resistant to
inactivation by human serum, and provide efficient transduction for gene therapy
applications. In our experiments, we observed consistently higher transduction
efficiencies with ProPak-A-derived particles compared with amphotropic vector packaged
in PA317 cells. There are at least two possible explanations for this difference. Other
reports have shown that murine cells secrete proteoglycans that inhibit transduction [13],
and the possibility that ProPak cells secrete less or no such inhibitor(s) is currently
being examined. Our own studies suggest that PA317 vector supernatants contain a viral

508

inhibitor which is envelope-specific [9]. It may be that the ProPak cells, which were
constructed with a split genome, produce a more favorable ratio of Env to Gag
components. Intriguingly, in comparison to PA317 supernatants, ProPak-A supernatants
contain less Env protein and similar levels of Gag protein (unpublished observations).
Using ProPak-A/X vector produced in a packed-bed bioreactor, we achieved 100%
marking of CFU-C derived from CD34+ cells purified from MPB, and demonstrated
expression of the Lyt2 transgene in 40% of the CD34+ cells 2 days post inoculation.
Previous applications of the ping-pong amplification technique were hampered by the
generation of RCR, and employed ecotropic producer cell partners, generating particles
which cannot transduce human cells. However, the demonstrated safety of the
complementary ProPak cell lines allows for RCR-free ping-pong amplification.

5. References

1. Miller, AD. Progress towards human gene therapy. Blood 1990; 76: 271-278.
2. Xu, LC. et al. Poor transduction efficiency of human hematopoietic progenitor cells

by a high-titer retrovirus producer cell clone. J. Virol. 1994; 68: 7634-7636.
3. von Kalle, C. et al. Increased gene transfer into human hematopoietic progenitor cells

by extended in vitro exposure to a pseudotyped retroviral vector. Blood 1994; 84:
2890-2897.
4. Bunnell, BA. et al. High-efficiency retroviral-mediated gene transfer into human and
nonhuman primate peripheral blood lymphocytes. Proc. Natl. Acad. Sci. USA 1995;
92: 7739-7743.
5. Rigg, RJ. et al. A novel human amphotropic packaging cell line: high titer,
complement resistance, and improved safety. Virol. 1996; 218: 290-295.
6. Forestell, SP. et al. Novel retroviral packaging cell lines: complementary tropisms
and improved vector production for efficient gene transfer. Gene Ther. 1997; 4: 600.
7. Rigg, RJ. et al. Detection of intracellular HIV-1 Rev protein by flow cytometry. J.
Immunol. Meth. 1995; 188: 187-195.
8. Plavec, I. et al. Sustained retroviral gene marking and expression in lymphoid and
myeloid cells derived from transduced hematopoietic progenitor cells. Gene Ther.
1996; 3: 717-724.
9. Forestell, SP. et al. Retroviral end-point titer is not predictive of gene transfer
efficiency: implications for vector production. Gene Ther. 1995; 2: 723-730.
10. Bestwick, RK., Kozak, SL., Kabat, D. Overcoming interference to retroviral
superinfection results in amplified expression and transmission of cloned genes.
PNAS USA 1988; 85: 5404-5408.
11. Bodine, DM. et al. Development of a high-liter retrovirus producer cell line capable
of gene transfer into rhesus monkey hematopoietic slem cells. Proc. Natl. Acad. Sci.
USA 1990; 87: 3738-3742.
12. Cosset, FL. et al. Use of helper cells wilh iwo hosl ranges to generate high-titer
retroviral vectors. Virol. 1993; 193: 385-395.
13. Le Doux, JM., Morgan, JR., Snow, RG., Yarmush, ML. Proteoglycans secreted by
packaging cell lines inhibit retrovirus infection. J. Virol. 1996; 70: 6468-6473.

SCALE-UP AND OPTIMISATION ISSUES IN GENE THERAPY
VIRAL VECTOR PRODUCTION

J.E. BOYD. L. BORLAND. C.E. FISHER
Q-One Biotech Ltd
West of Scotland Science Park, Todd Campus, Glasgow G20 0XA

1. Scale-Up

To produce viral vectors for gene therapy, it is essential, as in conventional
biopharmaceutical GMP systems to maximise and maintain production levels. The growth
and production characteristics of cells and vector should be investigated in a range of scale-
up scenarios e.g., roller bottles, microcarriers and fermenters. However, these systems vary
in their potential for manufacturing scale-up.

Roller bottles I can

be used very effectively to yield high cell numbers when bulking of a cell line is required

for cell banking or virus/viral vector propagation is required. This system is disposable so

no cleaning validation is required. The average cell yield from a roller bottle is

viable cells. A large scale production of 100 roller bottles may require up to 15 litres of

medium and requires a large number of manipulations upon set up and harvest. Variable
cell growth is observed and growth curve evaluation of each cell line should be carried out.

There are no dissolved oxygen or pH controls in place in this system.

While most surfaces used in cell culture possess a specific density of small charged

molecules to promote attachment and growth of cells, Cytodex 3 microcarriers have a
surface layer of denatured collagen covalently bound to a matrix of cross-linked dextran

spheres. The microcarriers are prepared by swelling and sterilised by autoclaving and they
are disposable. The surface area of cytodex 3 is of microcarriers. It can

therefore be calculated that 18.5g of these microcarriers would be equal to 100 roller
bottles in 6.2 litres of medium. Technology may be used for production of virus / viral

vector which is directly scalable but has no dissolved oxygen or pH control. However,

microcarriers may also be considered as a matrix for cellular growth and virus production
in the New Brunswick Celligen System using a double screen cell lift impeller.

The Celligen system is designed for growth of anchorage dependent ( microcarriers and
fibrous polyester disks ) and suspension cells in working volumes of 1.251 to 51. The
fibrous polyester disks are disposable with 70g of disks being equivalent to 100 roller
bottles but in a volume of 1.41 medium. This system has the advantages of dissolved
oxygen and pH control with medium circulation and feeding for optimal growth conditions

509

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 509-513.

© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

510

in a fully controllable temperature bench top unit. It requires few manipulations once set up
and the system is directly scalable. One important factor to consider is the cleaning
validation that would be required for use of this system.

In conclusion it can be noted that a smaller set up volume is required using the microcarrier
and fibrous disk systems promoting a more efficient use of medium, particularly important
if expensive serum-free or selection medium is required. The Celligen system allows easy
monitoring of medium exchange, whereas in the roller bottle system medium change is
labour intensive and carries the potential risk of contamination. The microcarrier and
Celligen systems can readily alter culture conditions e.g., from serum to serum-free
medium.

2. Recovery of Viable Cells After Freezing In Liquid Nitrogen

Cryopreservation is the storage of cells at very low temperatures such that any metabolic /
enzymic activity is virtually zero. Genetically manipulated cell lines used in gene therapy
often have altered freezing and recovery properties from that of their parent cell line. In
order to investigate freezing and recovery parameters a range of cell lines namely: 293;
vero; HeLa; NIH 3T3, 3T3 (retroviral vector producing) derivative cell line and PALidSN
cells were cultured in DMEM, 10% FCS/NBCS. These cells were harvested and then
frozen down in vapour phase liquid nitrogen at various cell densities using the freezing
down medium formulation of 95% complete medium, 5% DMSO.

These studies showed no distinct pattern of recovery. In some cases freezing at
yielded poor results (38% recovery for the 3T3 derivative) whereas for
some cell lines , 293, the percentage recovery was consistently good. Freezing at a higher

cell density viable cells/4.5ml freezing down medium yielded results ranging

511

from 56% to 111%. Freezing and recovery studies are therefore vital before laying down a
Master or Working Cell Bank (MCB/WCB).

3. Growth Curve Parameters

Mammalian cells can be characterised in culture by four growth phases; Lag, (resting) Log,
(exponential growth) Plateau and Decline. The cell lines 293 and Vero were cultured in
order to investigate growth curve parameters. The cells were grown in DMEM [4,500mg/l
glucose with sodium pyruvate] containing 10% FCS, and growth curves were generated
from roller bottles, tissue culture flasks culture systems and thirdly a
Cytodex 3 microcarrier system (using spinner vessels containing cultures of 293 cells on
574mg of Cytodex 3 microcarriers).

Generations of exponential growth, doubling time and viable cell yield per can be
calculated for each system studied. Analysis of supernatant generated during the culture
process for glucose content will give some indication of when the cultures should be fed.
All this information is required to determine accurately the requirements for each
individual cell line including tissue system to be used, average cell yield and feeding and

harvesting patterns.

4. Retrovirus Production

Retrovirus production schedules have been varied to maximise vector yield but it can be

demonstrated that if long term cultures are used the vector stability, rather than the

production rate may be limiting. PALidSN is a cell line which expresses the

iduronidase (IDUA) gene in a retroviral vector which contains a neo marker enabling

positive selection of cells containing the neo gene to occur using the G418 geneticin

selection system. PALidSN cells cultured in DMEM [4,500mg/l glucose], 10% FCS were

harvested and roller bottles were initiated and the supernatant harvested either;

512

daily, after four days incubation at then daily or after 7 days incubation and then

daily. The PALidSN roller bottle supernatant were titrated on NIH 3T3 cells with
Geneticin selection.

Roller bottle 2 displayed low viral tires for the first 3 days with viral production increasing

on day 4. Roller bottle 1 was incubated for four days and then harvested daily. The viral

titre on day 4 was the same as that for roller 2 but the highest titre obtained, harvested on
day 5, was double that of roller 2. Leaving the cell culture for 7 days prior to harvest saw

the titre drop proving that this viral vector is not stable at for prolonged periods of

time and therefore, should be harvested daily after the cells have grown to confluency.

5. Retrovirus Stability

Stability of the vector and Amphotropic Murine Leukaemia Virus (A-MLV) supernatant

were investigated at various temperatures over 24 hours and long term storage over a

number of days at Over this time, samples were removed for titration

on NIH 3T3 and PG4 respectively. No significant change in viral titre was observed after
eight hours storage at each temperature for either virus or vector. However, after 24 hours,

a log decrease in titre was observed for vector stored at and A-MLV stored

at and A-MLVincubated at for 24 hours retained only 0.4% of the initial
titre, a reduction in units. At vector titre displayed greatest stability,

decreasing only by units over 24 hours. Both vector and A-MLV were completely

513

inactivated by storage at 37°C for 3 days. However, vector supernatant was more stable

than A-MLV over long term storage at 4°C A-MLV was inactivated after 37 days whereas

vector supernatant titre decreased by after 37 days at this temperature.

6. Downstream Processing

Downstream processing capacity is an important factor particularly for adenoviruses which
require separation from inactive particles. Current practices for adenovirus vector
purification involve at least two rounds of density gradient centrifugation. This technique
separates cellular debris and defective particles from infectious adenovirus in the first
round and further separates defective from infective in subsequent rounds of centrifugation.
However this technology is not readily scalable and losses may be observed depending on
skill and accuracy when removing the bands. De-salting and further purification is carried
out using chromatography techniques. Investigation into alternative downstream processes
including filtration and chromatography techniques will be carried out in a view to

optimising this and other processes.

COMPARISON OF MANUFACTURING TECHNIQUES FOR ADENOVIRUS
PRODUCTION

Jeffrey M. Ostrove* , Paddy Iyer, Jon Marshall and Dominick Vacante
MAGENTA Corp.; 9900 Blackwell Road, Rockville MD 20850 and MAGENTA
Services, Ltd; Innovation Park, Hillfoots Road, Stirling FK9 4NF Scotland.

Keywords: 293 cells, adenovirus, bioreactor, serum-free medium, microcarriers

1. Abstract
We have compared three different production methods which may be suitable for the large-scale
production of adenovirus vectors for human clinical trials. The procedures compared 293 cells
adapted to suspension growth in serum-free medium in a stirred tank bioreactor, 293 cells on
microcarriers in serum containing medium in a stirred tank bioreactor, and standard tissue culture
plasticware. With a given virus, yields varied between 2000 and 12,000 infectious units (iu)/cell.
The stirred tank bioreactor routinely produced between 4000 and 7000 iu/cell when 293 cells were
grown on microcarriers. 293 cells adapted to suspension growth in serum free medium in the same
stirred tank bioreactor yielded between 2000 and 7000 iu/cell. Yields obtained from standard
tissue culture plasticware were up to 12,000 iu/cell. Cell culture conditions were monitored for
glucose consumption, oxygen utilization, lactate production, and ammonia accumulation. Oxygen
utilization rate, glucose consumption and lactate accumulation correlated well with the cell growth
parameters. Ammonia production does not appear to be significant. Based on virus yields, ease of
operation and linear scalability, large scale adenovirus production seems feasible using 293 cells
(adapted to suspension/serum free medium or on microcarriers in serum containing medium) in a
stirred tank bioreactor.

2. Introduction

Adenovirus vectors have various applications in the field of gene therapy and as viral oncolytics
(1). Methods of adenovirus production for phase I clinical trials have relied on standard tissue
culture plasticware for the most part. Apart from being labor intensive, these methods have limited
potential for scale-up. We have compared three different production methods for the generation of
adenovirus vectors. Adenovirus production in a stirred tank bioreactor using both anchorage
dependent 293 cells and 293 cells adapted to growth in suspension using a proprietary serum free
medium were compared to cultures grown on tissue-culture plasticware. 293 cells in bioreactor
cultures were infected at a multiplicity of infection (MOI) of 5-10 when the cell density reached

cells/ml. At 48 to 72 hours post-inoculation, viral infected cultures were harvested
and virus quantitation was performed using standard 50% tissue culture infectious dose

* Corresponding author. Tel: (301) 738-3936; FAX: (301) 738-1036; e-mail: [email protected]
515

O. - W. Merten el al. (eds.), New Developments and New Applications in Animal Cell Technology, 515-521.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

516

assay (2). Glucose consumption, lactate accumulation and ammonia production from the cultures
were monitored using a Kodak metabolite analyzer.

3. Methods

HEK 293 cells obtained from ATCC and grown in fetal bovine serum containing medium
(DMEM, 10%FBS) in tissue culture flasks were used to seed microcarrier cultures. 293
cells adapted to growth in suspension using serum-free medium in 100 ml shaker flasks were used
to seed suspension bioreactor cultures in serum-free medium. A benchtop stirred tank bioreactor
(Artisan Inc., Waltham, MA) with a 4L working volume and height/diameter ratio of 2:1 was
used. Oxygen, air and was supplied by headspace aeration and agitation provided by a marine
impeller. In the bioreactor, the temperature was controlled at Dissolved Oxygen (D.O.) as a
percent of air saturation was controlled at 50% and the was controlled using or 7.5%
sodium bicarbonate at 7.3. Total cell densities for microcarrier cultures were determined using
nuclei staining method (3), while viable cell densities for suspension cultures were determined by
the Trypan Blue dye-exclusion method. Cell densities were determined on days 1, 3, and 6 post-
seeding and glucose consumption, lactate and ammonia production was monitored using a Kodak
metabolite analyzer. When the cell densities reached c/ml, a one volume fed-batch
exchange was performed. After the fed-batch exchange, an adenovirus with Ad-5 backbone and
E1 A region deleted was introduced into the culture at a MOI of 5-10. After infection, conditions
were maintained as above for cell growth. 48 and 72 hours post virus infection, samples were

collected for virus titer assays and glucose, lactate and ammonia analysis. Virus quantitation was

performed using the assay on the clarified infected cell lysate.

For experiments using plasticware, HEK 293 cells grown in DMEM with 10%FBS in
tissue culture flasks were used to seed tissue culture plasticware Day 1 post
seeding, an adenovirus with Ad-5 backbone and E1 A region deleted was introduced into the
culture at a MOI of 5-10. Samples were collected 48 hours post infection and virus quantitation

was performed.

4. Results and Discussion:

HEK 293 cells in serum containing medium were grown in 2L and 4L microcarrier bioreactor
cultures. The cells seeded at c/ml typically by day 6 or 7 of the
culture. During this time, increased glucose consumption and lactate production was noticed
(Figure 1). No trend toward increased ammonia production in the same time period was detected.
This seems to indicate that glutamine at levels present in the medium may not be significant as an
energy source. A fed-batch one volume exchange on day 6 or 7 of the culture led to a rebound in
glucose levels and a drop off in lactate levels (Figure 1). These conditions were considered ideal
for the infection phase and yielded between 4000 and 7000 infectious units per cell (Table 1).

HEK 293 cells were adapted to suspension growth using a proprietary formulation of serum free
medium. Once adapted, these cells have been maintained for more than sixty days and passaged

517

518

nine times in 100 ml shaker cultures with consistent cell growth parameters and cell viabilities
(Figure 2). 293 cells adapted to suspension growth were used to seed 2L bioreactor suspension
cultures. The cells seeded at reached by day 6 or 7 of the culture. As in the

serum containing microcarrier cultures, increased glucose consumption and lactate production was
noted. Adenovirus added after a one volume fed-batch exchange led to virus yields between 2000

and 7000 infectious units per cell (Table 2). Standard tissue culture plastic ware Nunc

trays) provided the highest adenovirus yields. Virus yields from such an optimized system were
upto 12,000 infectious units per cell (based on assay).

519

5. Conclusion
Standard tissue culture plasticware provided the highest adenovirus yields. However, these
methods are labor intensive and scale limited. We have shown that with a stirred tank bioreactor
and anchorage dependent HEK. 293 cells, it is possible to produce adenovirus yields that is linearly
scaleable. Further, we have successfully adapted 293 cells to grow in suspension culture in serum-
free medium in a stirred tank bioreactor. Virus yields from suspension 293 cells in serum-free
medium, show promise for large scale adenovirus production given the potential ease of scale-up
operation. This also will provide a less complex feedstream for downstream processing. Efforts
are ongoing to optimize virus yields for such a system.
6. Acknowledgements
The authors would like to thank the following for technical support: Umme Habiba and Marcia
Meseck.
7. References

1. Crystal, R. Science 270: 404-409, 1995.
2. Schmidt, N. and Emmons, R. in Diagnostic Procedures for Viral, Rickettsial and Chlamydial Infections,

18-21, 1989.
3. Nahapetian, A., Thomas, J. and Thilly, W. J. Cell Sci. 81: 65-103, 1986.

520 If you take your suspension adapted serum-free cells and put them
Discussion back into a T flask, do you get back your 12,000 infectious units?
Aunins:
Ostrove: We have not tried this, but it is worth doing.
Southwick:
About the growth of your 293 cells in shaker flasks; you say that
Ostrove: you keep them growing for 60 days - do you maintain a single cell
suspension during that time, or do you find the cells differ in the
Guillaume: way they grow over that period?
Ostrove:
We can maintain fairly good single cell suspensions. 293 cells
Singhvi: historically have clumping problems but we have worked out
conditions now, in both the shaker flask and bioreactor, to maintain
single cell suspension.

With regard to different productivities with different adenoviruses,
do you have any comments on the construct of the virus?

When we do multiple manufacturing runs of the same virus our
yields are very comparable, within a factor of two. When you look
at the 6 different viruses we used, we consistently find that in many
cases certain viruses which express certain transgenes have a lower
productivity within the cell. Whether this is due to the transgene
expression itself or the way the virus was constructed, is unknown.
When manufacturing cell lines expressing retroviruses, we
recommend that anybody making an adenovirus infectious virus
through transfection procedures should pick multiple plaques and
try to obtain a virus that has a higher yield. We do see differences
due to the recombination events.

You maintained cells for 60 days and then used them in a
production run. Did you look at the population doubling levels to
see if it had any impact on your productivity? Were you able to do
production runs at different PDL levels without running into
problems?

Ostrove: 521
We were concerned about population doublings. The rumour is
that as the passage of 293 cells increases, the yield decreases and
they even lose the ability to plaque. In our case we start with a

Master Cell Bank certified at a passage number of 31. Even after
60 days, the passage level in suspension is relatively low and we
did not see an effect on yield. We freeze these cells and bring them
back up into suspension to see how far we can take them and
whether there is a point that yield does decrease.

SAFETY CONSIDERATIONS IN THE DEVELOPMENT OF NEW
RETROVIRAL AND ADENOVIRAL VECTORS FOR GENE THERAPY

D.MORGAN1, I. FORGIE1, J.OSTROVE2 and M.H.WISHER1
MA BioServices
1Innovation Park, Stirling, Scotland, FK9 4NF
29900 Blackwell Road, Rockville, MD 20850, USA
1. Introduction
As the development of new gene therapy vectors progresses and the number of disorders
which may be treated grows, it is crucial that safety is addressed from the development
and preclinical stages, through the manufacturing processes and into clinical trials. This
paper will deal with adenoviral and retroviral vectors as they have comprised the
majority of approved clinical trials to date.
2. Historical Perspective
Since regulatory aspects of gene therapy are well-documented [1], it may be questioned
whether or not safety issues have in fact been resolved. To illustrate that safety is still an
issue, Table 1 summarises data accumulated over recent years of testing cell banks and
their products from research laboratories and production facilities for the presence of
mycoplasma.

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 523-529.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

524

This shows that a considerable number of cell banks sourced from research laboratories
are contaminated with mycoplasmas. This is an important observation because many of
the cell lines used in gene therapy have been derived from academic research groups.
The lower level of contamination observed in the products of these cell banks may be
due to purification processes which eliminate mycoplasmas. The cell banks derived
from production facilities are also contaminated with mycoplasma at a significant level.

3. Potential Hazards

3.1 DURING DEVELOPMENT

Table 1 illustrated the hazard of mycoplasma contamination arising from cell banks.
Other contaminants which may arise from cell banks are bacteria, fungi, viruses,
retroviruses and prions. All of these may also arise from media components or operators
and are all cause for concern both during development and at later stages.

3.2 DURING PRODUCTION

As cell growth is scaled up to production levels, so the number of hazards increases.
The contaminants which are a hazard in development continue to be a risk. Using
uncontrolled and/or unvalidated manufacturing procedures may lead to further problems
of contamination.

Replication competent viruses may be generated at this stage when the producer cell line
is expanded for the production of the vector. Problems may also arise if the vector is
unstable. A less effective product may result if degradation of the vector takes place.

3.3 DURING CLINICAL USE

Stability of the product continues to be a problem at this stage. Storage conditions
should be validated to ensure stability and the buffer formula should be optimised. In
addition, the vector may not target the appropriate tissue or it may not be expressed
correctly. Incorrect expression may arise from expression in the wrong tissue or lead to
the wrong product in the correct tissue. This may be a particular problem with retroviral
vectors since retroviruses integrate into the host cell genome in a random fashion. This
has the potential to lead to activation of oncogenic genes or inactivation of tumour
suppressor genes. Another problem which may occur during clinical use is
inappropriate or unexpected immune response at the delivery site. This may lead to a
loss of efficacy of the vector.

Mobilisation of the vector may occur if target cells are infected with a replication
competent virus. Recombination may occur leading to the formation of a replication
competent virus carrying the gene from the replication incompetent virus.

525

4. Assessing the Risks

There are a number of factors to be considered when an assessment of the risk is being
made and decisions required to minimise the risk. Some of these are fairly easily
defined such as the phase of development, the patient population and the clinical
prognosis of the patients to be treated. For example, the level of safety testing required
may be different when a Phase 1 trial is being planned on terminally ill patients
compared to that required when a late-phase trial is being planned with patients who are
not terminally ill.

There are other factors which may also be considered when the risk is being assessed
which are rather less-well defined. For example, the perception of liability may be very
different for an academic group when compared to that perceived by a pharmaceutical
company.

4.1. QUESTIONS TO BE ASKED

There are a number of questions which should be asked in order to minimise the risks.

How targeted is the gene?
Will the gene enter non-target sequences and if so, what will be the
consequences?
What are the likely toxic effects of the vector?
How long will the gene persist in the cell?
How long is the gene likely to be expressed ie. will multiple administration be
required?

It is possible that the answers to all these questions may not be available so it is
imperative that steps are taken to minimise the risks.

4.2. MINIMISING THE RISKS

When viral vectors are being produced, the risk of microbial and viral contamination is
minimised by preparing viral vectors using Good Manufacturing Practice (GMP).

This starts by banking cells under GMP which includes a thorough screening package to
test cells and all raw materials for contaminants. Vectors are then produced using a
controlled production system with appropriate tests performed on process intermediates
and final products.

It is important to note that most viral vector production does not include any significant
downstream purification, so it is important to detect possible contaminants upstream.

Another stage where risk may be minimised is during development when the packaging
cell line is being selected. This is described below.

526
5. Replication-Competent Retrovirus (RCR) Testing
The current US Food and Drug Administration (FDA) recommendations for the
detection of RCR suggest co-cultivation with Mus Dunni Cells testing 1% or cells,
whichever is less and 5% of the supernatant at the Master Cell Bank stage. At the
Manufacturer’s Working Cell Bank stage, it is recommended to test either supernatant or
cells.

6. Replication-Competent Adenovirus (RCA) Testing
The rate of detection of RCAs is significantly higher than that of RCRs as is shown in
Table 3. It should be noted that these results have been obtained using 293 cells and the
development of new packaging cell lines may reduce the number of RCAs.

527

The titre of adenovirus tested is generally dependent on the dosage planned for clinical
treatment and as Table 3 shows, the rate of RCAs is in excess of 1 in 3 samples tested
when the dosage is greater than

7. Development of Safer Vectors
Factors to be considered during design and development of any vector include ensuring
the vector will be well-targeted, that the gene expression will be well-controlled and that
the formula ensures a low immune response. The use of purification steps and
optimisation of the formulation are also required to ensure a stable vector.
7.1. RETROVIRAL VECTORS
Recombination between the genes encoding the structual proteins of the retrovirus
within the packaging cell line and the vector genome leads to the formation of RCRs.
Viral sequences endogenous to the murine host cell line ( NIH-3T3) have been shown to
participate to form RCR [3][4].
A system that uses human 293 cell lines as the vector producing cell line and therefore
eliminates the presence of homologous endogenous retroviral sequences has been
developed [5]. This has failed to show any evidence of RCR after stringent testing.
However, it has been shown that even in a vector producer line with the lowest
theoretical rates of RCR generation, generation of RCR may occur which cannot be
explained [6].
7.2 ADENOVIRAL VECTORS
RCAs may not be considered to be as high a safety risk as RCRs because they are not
integrated into the target cell genome. The principle area of concern with adenoviruses
is that since they generally require repeated treatments in order to maintain an effect, the
likelihood of an immune response is increased.
To reduce the likelihood of adenoviral antigens on the target cell surface, recombinant
adenoviruses have been constructed which lack the major late promoter (MLP) and
therefore should not express late viral genes [7].

8. Patient Monitoring for RCR
Current FDA recommendations for patient monitoring suggest that blood samples
should be tested 4-6 weeks after treatment then every three months for the first year and

annually thereafter. This testing should comprise serological assays and/or PCR as
appropriate taking into consideration the mode of vector administration and the patient
population. The data should be submitted to the FDA with the annual progress report.

528

Accumulation of this data by the FDA should lead to a great deal of information on the
safety of retroviral vectors. However proposals are under consideration to reduce the
follow-up testing to a level which is less stringent.

9. Summary

Safety in the gene therapy field is a complex area with many issues which have to be
considered. This field is growing very quickly and as data is accumulated, it will
become more evident where problems occur.

It is essential that everyone in the field works towards safer vectors and new production
techniques. It may also be important that testing is standarised and streamlined so that
the minimum amount of testing material gives the maximum amount of information.

10. References

1. Wilson, A.C., Ng, T-H. & Miller, A.E. (1997) Evaluation of Recommendations for
Replication-Competent Retrovirus Testing Associated with Use of Retroviral Vector,
Human Gene Therapy, 8, 869 - 874.
2. Donahue, R.E., Kessler, S.W., Bodine, D., McDonagh, K., Dunbar, C., Goodman,S.,
Agricola,B., Byrne, E., Raffeld, M., Moen, R., Zsebo, K.M. & Nienhuis, A.W. (1992)
Helper Virus Induced T Cell Lymphoma in Non-Human Primates after Mediated Gene
Transfer, J. Exp. Med, 176, 1125 - 1135.
3. Otto, E., Jones-Trowler, A., Vanin, E.F., Stambaugh, K., Mueller, S.N., Anderson,
W.F. & McGarrity, G.J. (1994) Characterisation of a Replication-Competent Retrovirus
Resulting from Recombination of Packaging and Vector Sequences, Human Gene
Therapy, 5, 567 - 575.
4. Purcell, D.F.J., Broscuis, C.M., Vanin, E.F., Buckler, C.E., Nienhuis, A.W. &
Martin, M.A. (1996) An Array of Murine Leukemia Virus-Related Elements is
Transmitted and Expressed in a Primate Recipient of Retroviral Gene Transfer, J. Virol.,
70, 887 - 897.
5. Rigg, R.J., Chem, J., Dando, J.S., Forestall, S.P., Plavec, I. & Bohnlein, E. (1996) A
Novel Human Amphotropic Packaging Cell Line: High Titer, Complement Resistance,
and Improved Safety, Virology, 218, 290 - 295.
6. Chong, H. & Vile, R.G. (1996) Replication-Competent Retrovirus Produced by a
“Split-Function” Third Generation Amphotropic Packaging Cell Line, Gene Therapy, 3,
624 - 629.
7. Schneider, S.D., Rusconi, S., Seger, R.A. & Hosslew, J.P. (1997)
Adenovirus-Mediated Gene Transfer into Monocyte-Derived Macrophages of Patients
X-Linked Chronic Granulomatous Disease: Ex Vivo Correction of Deficient Respiratory
Burst, Gene Therapy, 4, 524 - 532.

Discussion 529
Goodhall:
Could you briefly outline how you test for your mycoplasma
Morgan: contamination?
We use 2 methods. We co-culture the test article withVero cells
and use the Hoescht stain. We also try to grow the mycoplasma
directly on suitable media.

GENE THERAPY FOR NEUROMUSCULAR DISORDERS:
MACROPHAGES AS SHUTTLES FOR WIDESPREAD TARGETING

E. PARRISH, E. AND L. GARCIA cedex, France
INSERM U421, 8 rue du 94010

Abstract

Gene therapy as a treatment for neuromuscular diseases is an ever-developing concept
based on the use of DNA as the therapeutic agent. In the search for appropriate
strategies, however, a bottleneck exists concerning the targeting of the therapeutic gene
to all pathologic sites. These diseases are often characterised by multiple widespread
lesions spread over a large area, rendering administration by local injection into tissues,
clinically irrelevant. We are therefore confronted with the need to seek a method of
targeting which uses the systemic pathway, but nevertheless specifically limits the
delivery of the therapeutic agent to the pathologic sites only. To this end, we have
proposed that circulating cells which home naturally to inflammatory lesions, could be
used as shuttles able to track down every damaged site. Using a murine model of
muscular dystrophy (mdx), we have validated the possibility of using engrafted
monocyte-macrophages to perform this function. The objective is now to engineer them
into ‘monocyte-derived cargo cells’ (MDC-cells) able not only to target, on demand, any
pathological site, but also to locally deliver a given therapeutic agent. In muscular
dystrophies, particularly DMD, the genetic defect results in the sporadic necrosis of
muscle fibres, leading to muscle degeneration. For a period of time, however, a certain
plasticity allows the compensation of this degenerative process by at least a partial
regeneration, giving us two points of attack to limit muscle wasting. The first is
preservation of the tissue by either protecting the fibres or enhancing regeneration. This
would require ‘secreting’ MDC-cells engineered to deliver a factor appropriate to this
purpose. The second aims to mobilise a corrective gene from the MDC-cells into muscle
cells, a process of in situ cell to cell gene transfer which could be accomplished using a
retroviral vector, since the regeneration process involves the proliferation of muscle
precursors before they fuse to form replacement fibres. For this, MDC-cells must be
rendered capable of packaging retroviral vectors.

Secreting MDC-cells might also be suitable for targeting, via the systemic pathway,
the diffuse pathological sites occurring in many CNS diseases and delivering a
neuroprotective factor. They could be used not only as ‘patrollers’ summoned in acute
reaction to inflammatory episodes, as proposed for the muscular dystrophies, but
additionally, as ‘sleepers’, having colonised the CNS from the hematopoietic tissue via
the natural turnover of resident macrophages (i.e. microglia).

531

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 531-539.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

532

Introduction

Gene therapy as a treatment for genetic disorders is an ever-developing concept based
on the use of DNA as the therapeutic agent. For any given disease, the identification of
the implicated gene is, in principal, a prerequisite, although only the beginning of a
multistep process requiring the successful vectorisation of the curative gene and its
efficient targeting to the damaged tissue. Indeed, in the search for appropriate
therapeutic strategies, the latter represents a bottleneck common to the therapy of many
disorders, in particular, the different muscular dystrophies and degenerative diseases of
the central nervous system. In the case of these neuromuscular disorders we are faced
not only with the problem of the widespread and sometimes inaccessible nature of the
affected sites, but also with the scale of the territory, as a whole, to be treated. To
overcome this difficulty, a targeting strategy needs to be sought, therefore, which is
adapted to disseminated lesions and so uses the systemic pathway, but nevertheless
specifically limits the delivery of the therapeutic agent to the pathologic sites.

Muscular dystrophies

Gene therapy for muscular dystrophies typically seeks to transfer a healthy or corrective
gene into muscle fibres. This transfer has, so far, been achieved through injection of
recombinant viral vectors (Quantin et al., 1992; Ragot et al., 1993; Vincent et al., 1993),
preparations of naked DNA (Wolff et al., 1990; Acsadi et al., 1991), or lethally
processed murine packaging cells (Fassati et al., 1995), directly into the affected tissues.
Delivery by local injection is, however, of limited clinical use in such diseases where the
sites to be treated are, as already discussed, multiple and disseminated throughout the
skeletal muscle as a whole. One way to circumvent this obstacle has been suggested to
us by considering the pathological process itself. Basically, a muscle lesion, whether
accidental or as the result of a genetic defect, triggers a complex response which can be
broken down into three arbitrary phases. First, the fibres degenerate. Second, at each site
of damage, there is a subsequent rapid infiltration of mononucleated cells, particularly
macrophages. Third, if the muscle plasticity allows, neighbouring satellite cells are
activated, proliferate and fuse, giving rise to new replacement muscle fibres (figure 1).
In view of the specific recruitment of macrophages to these sites of lesion, we have
proposed that this cell type and its blood precursor, the monocyte, could be used as a
cell carrier to shuttle a given therapeutic agent to each afflicted area (Parrish et al.,
1996). We have demonstrated the possibility of using transplanted immortalised
monocytes as naturally homing shuttles able to target multiple disseminated lesions in
skeletal muscle diseases. These cells, injected directly, intravenously, into mice,
successfully attained experimentally induced necrotic sites in muscle, showing that a
one-off administration of cells could rapidly target a given pre-existing muscle injury
and probably any inflammatory zone. However, because inherited muscle dystrophies
are in fact characterised by the sporadic occurrence of muscle fibre degeneration,
throughout the lifespan of the individual, we have also demonstrated the possibility of

533

creating, using bone marrow transplantation, a constitutive reservoir of genetically
modified cells able to infiltrate any spontaneous site of muscle damage, as it occurs.

The aim now is to modify the exogenous macrophages, or their precursors, to
produce circulating ‘monocyte-derived cargo cells’ (MDC-cells), able to work as
therapeutic patrollers, targeting 'on demand' any pathological site as it arises. The
engineering required for this modification (developed later) depends upon the choice of
therapeutic approach. The pathogenic process in muscular dystrophies is summarised in
figure 2. The genetic defect results in the sporadic necrosis of muscle fibres leading,
eventually, to muscle degeneration. However, the dystrophic muscle maintains, for a
period of time, a certain plasticity allowing the compensation of this degenerative
process with at least a partial regeneration. This gives us two points of attack to limit
muscle wasting. The first is based on the possibility of preserving the muscle tissue by
either protecting the fibres, thus slowing the rate of degeneration, or enhancing the
regeneration, thus boosting the compensation. The second deals with the possibility of
escaping from this vicious circle by genetically correcting the replacement fibres as they
form. Two types of MDC-cell are therefore required.

For myopreservation, the macrophage must be engineered to secrete, locally, a factor
acting either on the existing fibres themselves, or on their precursors, the satellite cells.
This idea remains essentially theoretical however, since no such candidate molecule is
currently available. Nonetheless, it is not unreasonable to imagine that in the not too

534

distant future, protection against the lack of dystrophin might perhaps be achieved by
factors stimulating alternative related proteins such as utrophin (Deconinck et al., 1997;
Grady et al., 1997). In addition, in vitro studies have suggested that macrophages
themselves could directly participate in muscle fibre regeneration by exerting mitogenic

and chemotactic effects on muscle precursor cells (Robertson et al., 1993; Cantini et al.,

1995). The anticipated identification of the factors responsible may well provide
candidate molecules, more appropriate than the non-specific LIF and FGFs, for
enhancing the regeneration.

For correction, the aim is to mobilise the corrective gene from the blood borne
macrophage into target muscle cells, a process of in situ cell to cell gene transfer which
could be accomplished using a retroviral vector, since the regeneration process
occurring at the site of damage involves the proliferation of muscle precursors before
they fuse to form replacement fibres. This then, requires that the macrophage be
engineered into a ‘packaging cell’ containing both a replication deficient retrovirus
carrying the gene of interest, and an helper genome (gag-pol-env) needed for its
packaging and secretion. In this way, monocyte shuttles distributing the therapeutic
retrovirus to all widespread sites of necrosis-regeneration, would allow the ‘correction’
of an ever increasing number of fibres as they regenerate. The corrective gene in this
strategy could encode either a functional truncated dystrophin (e.g. mini-dystrophin), a
trans-activating factor inducing the expression of a dystrophin-related protein (e.g.
utrophin), or sequences allowing exon skipping to restore the endogenous dystrophin, in
truncated form.

CNS degenerative disorders
As things stand, the application of gene therapy to degenerative diseases of the CNS at
first sight poses a problem, since very few of the genes implicated have been identified,
and even when this is the case (Lefebvre et al., 1995; Melki et al., 1996), there is not
always an obvious solution for delivering a therapeutic gene into the subpopulation of
cells concerned. In the face of this difficulty an alternative strategy, based on phenotypic
compensation, has emerged from the ability of neurotrophic factors to act on the
survival adult neurones. The interest for these molecules in this context, stems from an

535

ever increasing body of evidence demonstrating their ability to exert neuroprotective
effects both on a wide range of neuronal populations and in diverse lesion paradigms
(Arakawa et al., 1990; Sendter et al., 1990 et 1992a,b; Oppenheim et al, 1991 et 1995;
Martinou et al., 1992; Louis et al., 1993; Mitsumoto et al. 1994; Sariola et al., 1994;
Yan et al. 1995). The presence of the blood-brain barrier imposes the administration of
these factors directly into the CNS and to date, this has been achieved either by using
mini-pumps (Hagg et al., 1992; Anderson et al., 1996) or encapsulated BHK producer
cells implanted into the lateral ventricle (Emerich et al., 1996). However, the very
nature of most neurodegenerative diseases, progressive and occurring over a protracted
period of time, would seem to call for a system of long term delivery. This could be
achieved by genetically modifying the cells in the target area, to produce themselves the
therapeutic molecule. Most viral vectors, adenovirus (Akli et al., 1993; Le Gal la Salle
et al., 1993; Davidson et al., 1993; Bilang-Bleuel et al., 1997), herpes virus (Lawrence
et al., 1996), AAV (Peel et al., 1997) and lentivirus (Naldini et al., 1996) have proved to
be efficient tools for such a gene transfer in situ in the CNS, when administered by
stereotaxic injection. However, this route of local administration is not appropriate for
the targeting of the widespread lesions characteristic of most neurodegenerative
diseases, where a large part of the CNS as a whole needs to be treated. As proposed for
dystrophic muscle, therefore, ‘secreting MDC-cells’ might also be suitable for targeting,
via the systemic pathway, the diffuse pathological sites occurring in CNS diseases, and
delivering, locally, a neuroprotective factor (figure 3). Therapeutic MDC-cells could be
used in two ways: as ‘patrollers’, summoned on demand, in acute reaction to
inflammatory episodes such as those seen in multiple sclerosis for example, or
additionally, as ‘sleepers’, having colonised the target tissue from the hematopoietic
tissue, via the natural turn-over of a proportion of the resident macrophages (i.e.
microglia) (Krall et al., 1994). The sleepers could act either on demand, by induction of
secretion of the therapeutic agent, or chronically by its constitutive production. These
two approaches might well determine the use of multiple therapeutic factors, their
secretion being governed by the state of differentiation of the MDC-cells.

536

Engineering of monocyte-macrophages

To achieve their purpose, the monocyte-macrophages need, therefore, to be genetically
engineered into either ‘secreting’ or ‘packaging’ MDC-cells, depending upon their
therapeutic goal. Secreting MDC-cells could originate either from a transplanted
classically engineered bone marrow, or from primary macrophages transduced using
different defective viral vectors (adenovirus, herpes simplex virus, lentivirus) able to
infect post-mitotic cells. In either case, the aim is simply to efficiently introduce, into
the cell, a cassette containing sequences coding for a secretable therapeutic factor, under
the control of a specific promoter. In contrast to secreting MDC-cells, whose
engineering can be envisaged using a single step transduction procedure, the creation of
packaging MDC-cells entails a double modification of the monocyte-macrophages with
incorporation into the cell of i) the replication deficient retrovirus carrying the gene of
interest, and ii) the helper genome (gag-pol-env) required for packaging and secretion of
the retrovirus. This could be achieved in two ways: a multistep transduction using more
than one viral vector or a single step transduction using one complex vector (figure 4).

The multistep procedure would require sequential transduction with: a) a defective

viral vector (matrix vector), able to transduce post-mitotic cells, carrying the sequences

entirely encoding the provirus to be mobilised (which carries the therapeutic gene); b) a

537
defective viral vector (assembling vector), again able to transduce post-mitotic cells,
carrying a defective MuLvs gag-pol-env helper genome for transcomplementation of the
provirus. In the single step procedure, on the other hand, monocyte-macrophages would
be transduced by a single complex viral vector (master vector), able to transduce post-
mitotic cells, this time carrying both the provirus and a defective gag-pol-env helper
genome, for ciscomplementation of the therapeutic provirus.

So far, we have tested the idea of assembling vector driven mobilisation, in vitro, by
the double transduction of a monocyte cell line with a defective retrovirus (MoMuLV)

carrying the lacZ reporter gene and an HSV-amplicon carrying the Moloney gag-pol-env
sequences (kindly provided by A. Epstein - Lyon), respectively. Mobilisation of
MoMuLV -lacby the resulting engineered monocytes was demonstrated by the
successful transduction of gliobastoma cells (of the 9L cell line) using engineered
monocyte conditioned medium (figure 5).

Conclusion
Our first phase results show that monocyte-macrophages are appropriate cell vectors
either to target widespread inflammatory sites on demand, or to colonise a target tissue
via a process of natural cell turn-over. In addition, it is possible to extemporaneously
engineer these cells into ‘packaging monocyte-macrophages’ able to mobilise defective
retroviral vectors. The next phase is now to engineer these cells into ‘secreting’ or
‘packaging’ MDC-cells able to confer a therapeutic benefit in appropriate animal
models. Eventually of course, the aim is to specifically restrict the secretion of the
therapeutic agent to the target sites. This is a problem which, to a certain extent, could
be addressed by a judicious choice of the promoters controlling its expression. It is not
unreasonable to envisage the possibility of limiting secretion using appropriate
macrophage differentiation-specific promoters, such as CD68, which should restrict
expression of the therapeutic substance to sites of inflammation. In addition, since

538

resident macrophages display considerable phenotypic heterogeneity, to such an extent
that have been individually defined (e.g. Kupffer cell, alveolar macrophage, osteoclast,
microglia), one can speculate that, in the future, their molecular characterisation might
also provide us with candidate promoters for tissue specific expression.

This targeting strategy, although developed here in the context of neuromuscular
diseases, could theoretically be applied to any pathology where the sufferance or death
of individual or groups of cells induces the recruitment of blood borne macrophages. In
addition to the applications discussed in this paper therefore, it is perhaps also worth
considering the possible application of this method of targeting to strategies for the gene
therapy of certain cancers, in particular glioblastoma, in which macrophages naturally
encircle the tumour. Such a situation might allow the possibility of containing the
tumour and limiting its invasion.

References

Akli S. et al., Nature Gen. 3: 224-228, 1993. Sendtner M. et al., Nature 345: 440-441, 1990.
Anderson K. et al., Proc.Natl.Acad.Sci.USA 93: Sendtner M. el al., Nature 358: 502-504, 1992a.
7346-7351, 1996. Sendtner M. et al., Nature 360: 757-759, 1992b.
Arakawa Y. et al., J. Neurosci. 10: 3507-3515, 1990. Vincent N. et al., Nature Genet. 5: 130-134, 1993.
Acsadi G., et al., Nature. 352: 815-818, 1991. Wolff J. et al., Science. 245: 1465-1468, 1990.
Bilang-Bleuel A. et al., Proc.Natl.Acad.Sci.USA 94: Yan Q et al., Nature 373: 341-343, 1995.
8818-8823, 1997.
Cantini M., & Carraro U. J. Neuropathol. Exp.
Neurol. 54: 121-128, 1995.
Davidson B. et al., Nature Gen. 3: 219-223, 1993.
Deconinck A. et al., Cell. 90: 717-727, 1997.
Emerich P. et al., J. Neurosci. 16: 5168-5181, 1996.

Fassati A. et al., H.Gene Therapy. 7: 595-602, 1996
Grady R. et al., Cell. 90: 729-738, 1997.
Hagg T. et al., Neuron 8: 145-158, 1992.
Krall W. et al., Blood. 83: 2737-2748, 1994.
Lawrence et al., J. Neurosci. 16: 486-496, 1996.
Lefebvre et al., Cell 80: 155-165, 1995.
Le Gal la Salle G. et al., Science. 259: 988-990,
1993.

Louis JC. et al., Science. 259: 689-692, 1993.
Martinou JC. et al., Neuron. 6: 737-744, 1992.
Melki J. et al., Genomics 32 (3): 479-482, 1996.
Mitsumoto H. et al., Science. 265: 1107-1110, 1994.
Naldini L. et al., Science. 272: 263-268, 1996.
Oppenheim R. et al., Science. 252: 1616-1618,
1991.

Oppenheim R. et al., Nature. 373: 344-346, 1995.
Parrish E. et al., Gene Therapy, 3: 13-20, 1996
Peel A. et al., Gene Therapy. 4: 16-24, 1997.
Quantin B. et al., Proc. Natl. Acad. Sci. USA, 89:
2581-2584, 1992.
Ragot T. et al., Nature. 361: 647-650, 1993.
Robertson T. et al., Exp. Cell. Res. 207: 321-331,
1993.

Sariola H. et al., Ann. Med. 26: 355-363, 1994.

Discussion 539
Massie:
Since the cells you are trying to engineer are good presenting cells
Garcia: for antigen, are you not concerned about immune reactions if you
engineer the cell to express these antigens: for example, packaging
function of retroviruses?

Yes, it is a problem. The range of time for these experiments is 2-3
days, so we can expect that the cells were still able to reach their
target site and carry out their specific function. We can also
control the expression of the alpha genome by using constitutive
promotors. We also use differentiation control promotors which
give transient expression when they just arrive at the necrotic site,
or when they begin to phagocytose. Maybe we can decrease the
extent of presentation to some degree.

GENE TRANSFER INTO DOG MYOBLASTS

S. Braun, C. Thioudellet, C. Escriou*, M-C, Claudepierre,
F. -Rouault, E. Jacobs, R. Bischoff, D. Elmlinger,
H. Homann, Y. Poitevin, M. Lusky, M. Mehtali, F. Perraud and
A. Pavirani
Transgene S.A., Strasbourg, France. *ENV, Alfort,

France.

Keywords: Gene transfer, dog skeletal muscle, culture.

1. Introduction

Duchenne Muscular Dystrophy (DMD) is a lethal X-linked genetic disease.
The disorder is manifest in early childhood with progressive skeletal muscle weakness
and wasting. In a predictable fashion, affected boys loose muscle strength and ability to
walk. Death usually occurs due to respiratory and cardiac complications.

DMD muscle degeneration is caused by absence of dystrophin, a large rod-
like cytoskeletal protein found at the inner membrane surface of muscle fibers, which is
thought to play an important role in maintaining muscle fiber integrity.
Gene therapy of DMD aims at the expression of a functional dystrophin gene in skeletal
muscles and myocardium.

The GRMD dystrophic dog is the animal model which clinically resembles the
human disease. Various vectors and strategies of dystrophin or minidystrophin cDNA
delivery to muscles need to be evaluated: plasmids, non-viral (synthetic) vectors,
adenovirus, retrovirus, engineered satellite cell transplantation. The vectors are being
validated in vitro and in vivo and

2. In Vitro tranfection

Among the in vitro models that we have developped, myoblasts cultures of
healthy and dystrophic dogs (6 to 8 month old) are being used.The principle of this
type of culture is based on the ability of skeletal muscle fibers to regenerate from
mononucleated satellite (myoblast) cells. When cells reach confluency, myoblasts fuse
and form multinucleated myotubes.

Naked plasmid DNA is unable to transfect cells in vitro Using the calcium
phosphate precipitation method, up to 50 % dog myoblasts are transfected (LacZ
reporter gene).
High levels of luciferase activity (109 RLU/mg protein) were obtained after
transfection with plasmids encoding this firefly enzyme. Synthetic vectors were also
effective, the polyamines spermidine and spermine isomers-based cationic lipid/DNA
complexes being very potent vectors.

541

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 541-543.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

542

Interestingly, the efficiency of the various isomers varies depending on the transfected
cell type. Expression of dystrophin was also obtained in GRMD dog myoblasts as
demonstrated by Western Blot and immunocytochemistry after plasmid transfection
with various synthetic vectors and with the calcium phosphate precipitation method.

Infection of dog myoblasts with an adenovirus carrying the LacZ reporter
gene, led to up to 90 % of blue-stained cells (as in the case of other primary cultures,
the viral multiplicity of infection had to be elevate, i.e.200 for maximal efficiency).
Retroviruses infect dog myoblast cultures also well ( after 3 cycles of infection 75% of
cells stained blue). No differences in infection were seen between healthy and
dystrophic dog myoblasts and between types of muscle.

3. Myoblast graft

In vitro transfection or infection can be used in view of myoblast graft.Cells
are transduced in vitro before reimplantation into dog muscle. The engrafted cells
would allow regeneration of new muscle fibers expressing dystrophin or
minidystrophin. In a preliminary assay, we injected retrovirus ( ß-Gal)-transduced
autologous myoblasts in two 8 month old litermate dogs (1 healthy and 1 dystrophic
dog). During anesthesia, extensor digitory communis muscles were visualized after
skin and fascia incision. Three million myoblasts were injected in 3 different sites. Two
non-resorbable suturae were left to show the injection sites. Muscle biopsies were
retrieved 8 days and 1 month after myoblast graft, respectively,
cryosections showed some positive fibers in muscle biopsies in both animals. One
month after graft, no positive staining was observed, even though plasmid DNA was
detected by PCR.

4. Conclusions

Primary dog myoblast cultures represent a useful model for in vitro evaluation
of viral and non viral vectors. The cells can also be used for cell grafting with
applications in DMD, vaccination or production of circulating proteins.

Supported by the Association Francaise contre les Myopathies.

Discussion 543
Petrie:
Crespo: In the chairman’s overview he mentioned 1,400 gene therapy
clinical trials, including AIDS and cystic fibrosis. How may of
Ostrove: these give really broad positive effects to the patient?
The development of gene therapy is in a very early stage so only
Perraud: very few clinical trials are reaching phase III. More than 200
clinical protocols have been presented but, unfortunately, very few
are in clinical phase II. More time is needed for this new
technology with the complexity of manufacturing and safety issues
to be developed.
You showed that in order to have 75% transduction efficiency in
the myoblasts with the adenoviral vector, you needed an MOI
between 100 and 800. Are those particles or infectious units, and
can you comment on the need for such an MOI, and is there a
practical way of getting around this problem?
Infectious unit. 100 is minimal for myoblasts as primary cultures
are more difficult to infect than routinely used cell lines. Clearly
this is a problem for in vivo clinical trials as we are limited by the

bulk we can inject into the patient.

BIODISTRIBUTION ANALYSIS OF A GENE THERAPY VECTOR USING
THE POLYMERASE CHAIN REACTION (PCR) TECHNIQUE

Joanne Proffitt*, Martha Leibbrandt#, Keri Jarvis* and Carl Martin*

*Covance Laboratories #Chiron Corporation,
Otley Road Emeryville,
Harrogate, California 94608
North Yorkshire, USA
HG3 1PY

Introduction
The gene therapy T7(3)TK plasmid contains the herpes simplex thymidine kinase
(HSV-TK) gene under the control of the T7 RNA polymerase promoter (T7RNAP).
Rats were injected with T7(3)TK plasmid, at either low dose (group 2) or high dose
(group 3), and the T7 RNA polymerase protein (Table 1). A control group of rats
(group 1) were injected with formulation buffer.

The study was performed to determine the biodistribution of HSV-TK originating
from the T7(3)TK plasmid in certain organs and germ line tissue isolated from the
treated rats. DNA extracted from tissue samples taken from treated rats was tested by
the polymerase chain reaction (PCR) technique (Refs: [1] and [2]) using primers
designed specifically to the HSV-TK DNA sequence. Analysis of a single aliquot of
sample which may contain a DNA sequence at extremely low levels potentiates the
likelihood of a false negative result. Therefore, to circumvent the possibility of
missing a positive signal in a single sample and hence to ensure validity of the results
obtained, each sample was analysed in quadruplicate. In addition 10 plasmid copies
of T7(3)TK plasmid DNA were used to spike duplicate aliquots of each DNA sample

545

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 545-548.

© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

546

hence providing a positive control specific to each DNA sample. The presence of
HSV-TK DNA sequence in any test article DNA samples (identified by PCR analysis)
was confirmed by Southern blot hybridisation (Ref: 3) using a labelled HSV-TK
specific probe.

Assay Regime
DNA was isolated from each tissue sample. Each DNA sample was analysed in
replicates of 4 with additional duplicate aliquots spiked with T7(3)TK positive control
plasmid containing the HSV-TK sequence (Figure 1).

An example template scheme for PCR is given:

1) Purified water: Blank reaction

2) Purified water: Sentinel control

3) Negative control DNA: 0.1 µg DNA

4) Test DNA: 0.1 µg DNA

5) Test DNA: 0.1µg DNA

6) Test DNA: 0.1 µg DNA

547

7) Test DNA: 0.1 µg DNA

8) Spiked test DNA: 0.1 µg DNA spiked with 10 plasmid copies

9) Spiked test DNA: 0.1 µg DNA spiked with 10 plasmid copies

10) Positive control DNA: 1 plasmid copy

11) Positive control DNA: 10 plasmid copies

12) Positive control DNA: 50 plasmid copies

NB One set of assay controls (assay blank, sentinel, negative and positive controls)
were used per run of PCR. The positive control samples (1, 10 and 50 plasmid
copies) ensured consistency of PCR sensitivity between different runs of PCR. To

demonstrate reproducibility of data, PCR was repeated on any samples which were
positive in one or more, but not all of the quadruplicate DNA samples. The sentinel
control functioned as a monitor for airborne contaminants and therefore remained
open during the preparation steps, until initiation of PCR. An aliquot of each PCR
reaction was analysed by agarose gel electrophoresis.

Results

PCR analysis of the positive control T7(3)TK plasmid, established that the sensitivity
for HSV-TK detection was routinely 1 plasmid copy. The blank and sentinel controls
containing purified water only, and the DNA negative control samples were all
negative for the presence of HSV-TK DNA hence demonstrating that no cross-
contamination had occurred and that the primers used for PCR were specific to the
HSV-TK sequence. Spiking the DNA samples with positive control plasmid provided
a “mixed pool” positive control sample specific to each DNA preparation. In addition
it ensured that the DNA preparations were free from the presence of any factors
capable of inhibiting the amplification process.

Table 2

_____Group____________Negative___________Positive____

1 42/42 0/42

2 42/42 0/42

_______3________________39/42______________3/42_____

All the DNA samples (0.1 µg DNA) prepared from tissues isolated from animals in

group 1 (negative control group) and group 2 (low dose T7(3)TK plasmid) were
determined as negative for the presence of HSV-TK DNA (Table 2). Three DNA

samples (0.1 µg DNA) prepared from tissues isolated from animals in group 3 (high