Genetic analysis and utilization of adeno-associated virus (AAV) as a mammalian cloning vector


Material Information

Genetic analysis and utilization of adeno-associated virus (AAV) as a mammalian cloning vector
Physical Description:
xiv, 155 leaves : ill. ; 29 cm.
Hermonat, Paul Lyle, 1952-
Publication Date:


Subjects / Keywords:
Genetic vectors   ( mesh )
Cloning, Molecular   ( mesh )
Immunology and Medical Microbiology thesis Ph.D   ( mesh )
Dissertations, Academic -- Immunology and Medical Microbiology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1984.
Bibliography: leaves 142-154.
Statement of Responsibility:
by Paul Lyle Hermonat.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030351868
oclc - 11999583
System ID:

This item is only available as the following downloads:

Full Text







Copyright 1984

Paul Lyle Hermonat

This dissertation is dedicated to my mother, sister and brother,
and my friends in New Hampshire.


I would like to express my appreciation to my advisor,

Dr. Nicholas Muzyczka, for introducing me to a subject which I truly

enjoy and for his help during my preliminary examinations. I would

like to thank all the members of my dissertation committee, especially

Drs. Kenneth Berns and Edward Wakeland, for their encouragement and

constructive criticism. I would also like to thank Dr. Dan Rawlins,

Mark Labow, Dr. Richard Samulski and Randy Horwitz for many

enlightening discussions and for their friendship.



ACKNOWLEDGEMENTS ........................................... iv

LIST OF FIGURES ...................................... viii

LIST OF TABLES ............................................ x

KEY TO ABBREVIATIONS ...................................... xi

ABSTRACT .............. ............... .... ............ xiii


ONE INTRODUCTION ...................................... 1

Virus Vectors Which Replicate as Episomes .......... 5
Virus Virions as Vectors .......................... 7
AAV, a Potentially Useful Vector ................... 10

TWO METHODOLOGY ....................................... 13

Cells .............................................. 13
Enzymes and Linkers ................................ 13
Construction of Deletion and Insertion Mutants ..... 14
Construction of AAV/neo, dhfr and Bacteriophage X
Recombinant Plasmids ........................... 15
Plasmid Preparation and E. coli HB101 Transfection 16
Agarose Gel Electrophoresis ........................ 18
Klenow Fragment End Labeling ....................... 18
DNA Transfection into Tissue Culture Cells ......... 18
Mutant and Wild-type AAV Virus Stock Preparation ... 19
Recombinant Virus Stock Preparation ................ 19
Hirt DNA Extraction from Transfected, New Infected
or Latently Infected Tissue Culture Cells ........ 20
Genomic DNA Isolation .............................. 20
Southern Blotting and Hybridization ................ 20
Nick Translation .................................. 21
Transduction of Mammalian Cells with Recombinant
Virus ........................................ .. 22


THREE THE GENETICS OF AAV-2 .............................. 23

Introduction ...................................... 23
AAV Proteins .................................... 24
AAV Transcripts ................................. 24
MV Sequence Analysis ............................ 25
AAV/pBR322 Recombinant Plasmids .................. 29

Results ............................................ 30
Construction and Physical Characterization of
the AAV-2 Mutants .............................. 30
Mutants that are Defective for Duplex DNA
Replication ................................... 32
Mutants that were Defective in Generating
Infectious Virus and Single-stranded Progeny
DNA ............................................ 33
Mutants that Produced Single-stranded Progeny
but Low Virut Yields .......................... 34
Complementation and Recombination Between Mutant
Phenotypes .................................... 35

Discussion ........................................ 57
Rep Mutants ..................................... 58
Cap Mutants ...................................... 60
Li Mutants ..................................... 62

FOUR AAV-2 AS A MAMMALIAN VECTOR ........................ 65

Introduction ...................................... 65
The Advantages of an AAV Vector .................. 65
Strategies for Using AAV as a Vector ............. 67
A Prototype Foreign Gene A Selectable Marker ... 68

Results .......................................... 70
Construction of AAV Recombinant Genomes .......... 70
Preparation of a Recombinant Virus Stock ......... 72
Transduction of Human and Murine Tissue Culture
Cells with d152-91/neo Virus ................... 85
Rescue of d152-91/neo DNA from Transduced
Cells ......................................... 87

Discussion ........................................ 90
Expression of Recombinant AAV Proviruses ......... 93
Rescue of Recombinant Proviruses ................. 94
The Ultimate Host Range Virus .................... 94
Studies of AAV Recombinant Vectors Without
Virion Packaging ............................... 94
Gene Therapy .................................... 95



Introduction ..................................... 97
Latent Infection in Tissue Culture ............... 98
Structure of an AAV Provirus ..................... 99

Results ........................................... 101
Isolation of Neomycin Transduced Cell Clones ..... 101
Rescue of Recombinant Provirus from Transduced
Cell Clones .................................... 102
Restriction Analysis of AAV/neo Provirus
Clones ......................................... 107

Discussion ....................................... 131
AAV Provirus Integrity ........................... 131
A Model for AAV Integration ...................... 132

BIBLIOGRAPHY ........... ................................ 142

BIOGRAPHICAL SKETCH ...................................... 155



3-1. Physical structure of AAV RNAs and open reading
frames ............................................. 28

3-2. Physical structure of AAV mutants .................. 38

3-3. DNA replication phenotypes of AAV mutants .......... 48

3-4. DNA replication phenotypes of additional AAV mutants 50

3-5. Mutants that were partially defective-for DNA
replication ....................................... 52

3-6. Complementation of capsid mutants by rep mutants ... 54

3-7. Infectious virus DNA production by lip mutants ..... 56

4-1. Construction of AAV recombinants ................... 74

4-2. Physical organization of AAV recombinant genomes ... 76

4-3. Comparison of wt and recombinant DNA replication ... 78

4-4. Transducing virus titer determination .............. 82

4-5. Replication and virus production assays of ins 96-x
recombinants ...................................... 84

4-6. Rescue of AAV/neo sequences from transduced cells .. 89

5-1. Rescue of AAV/neo provirus from transduced cell
clones ........................ ....... ........... 104

5-2. Complementation of rescue negative cell clones with
wt AAV ............................................. 106

5-3. Restriction map of d152-91/neo ..................... 110

5-4. Detection of d152-91/neo DNA in genomic sequences of
transduced cell clones by BglII digestion .......... 112

5-5. Detection of head to tail tandem d152-91/neo
proviral genomes in transduced cell clones genomic
DNA ................................................ 116


5-6. Analysis of d152-91/neo proviral internal sequences
by PstI digestion of transduced cell clones genomic
DNA ............................................... 118

5-7. Analysis of d152-91/neo proviral internal sequences
by Ncol digestion of transduced cell clones genomic
DNA ............................................... 120

5-8. Analysis of d152-91/neo proviral sequences by Smal
and Smal, BglII double digestion of transduced cell
clone DNA ........................................ 123

5-9. Proposed AAV/neo proviral clone structures ......... 130

5-10. Models for retrovirus, SV40 and polyoma, and AAV
integration ...................................... 136



3-1. Physical Structure of AAV Mutants .................. 39

3-2. Phenotypes of AAV Mutants .......................... 42

3-3. Complementation of AAV Mutants ..................... 44

3-4. Listing of DNA Construction Transfections into
HB1Ols ............................................. 45

4-1. Transduction Efficiency of d152-91/neo Virus ....... 86

5-1. Hn Clonal Relationship ............................ 113

5-2. Summary of Genomic Digestions ...................... 125

5-3. Hn Provirus Genotype/Phenotype Summary ............. 128


AAV adeno-associated virus

Ad adenovirus

BKV BK virus

bp base pair

BPV bovine papilloma virus

BRL Bethesda Research Laboratories.-

05 a Detroit 6 cell line lately infected with AAV;
also called B7374IIIDV

06 Detroit 6 cells (human)

DEAE diethylaminoethyl

dhfr dihydrofolate reductase

DI defective interfering (particle)

dl deletion mutant

DNA deoxyribonucleic acid

DNase deoxyribonuclease

EBV Epstein-Barr virus

E. coli Escherichia coli

EDTA ethylenediaminetetraacetic acid

EPR early region promoter/enhancer (SV40)

fm form plasmidd)

HAT hypoxanthine-aminopterin-thymidine (media)

HSV herpes simplex virus

IBI International Biotechnologies, Inc.

ins insertion mutant

kb kilobases

kd kilodaltons

LTR long tenninal repeat (retrovirus)

MEM minimum essential medium

MOI multiplicity of infection

mRNA messenger ribonucleic acid

mu map unit

neo neomycin resistance

ORF open reading frame

RNA ribonucleic acid

RNase ribonuclease

SDS sodium dodecyl sulfate

SNV spleen necrosis virus

SV40 simian virus 40

tk thymidine kinase

TR terminal repeat (AAV)

VP viral protein

wt wild type

xgprt xanthine-guanine phosphoribosyl transferase

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Paul Lyle Hermonat

December 1984

Chairman: Nicholas Muzyczka
Major Department: Immunology and Medical Microbiology

Adeno-Associated Virus (AAV) is a helper dependent parvovirus

which can exist as both an integrated provirus and as a lytic virus.

To evaluate AAV's utility as a mammalian cloning vector and elucidate

AAV's biology, I first studied AAV's genetics. This was done by

contracting insertion and deletion mutants in the AAV sequences of an

infectious AAV/pBR322 recombinant plasmid. Three phenotypes were

observed when these mutants were transfected into adenovirus infected

tissue culture cells: i) Mutants located between map units (mu) 11 and

42 were defective in their ability to rescue and replicate the AAV

genome from the recombinant plasmid; identifying a gene (or genes) that

is required for AAV replication, ii) Mutants located between mu 63 and

91 were unable to produce infectious virus and could not produce

single-stranded DNA progeny but were capable of duplex DNA replication.

These mutants are probably defective in the major capsid protein, VP3.

iii) Mutants located between mu 48 and 55 were able to rescue and

replicate normally, producing both duplex and single-stranded DNA

progeny, but produced only low levels of infectious virus particles.

These mutants are probably defective in the minor capsid protein, VP1.

With this genetic map I have developed a strategy for using AAV as

a mammalian cloning vector. I have replaced the capsid gene region of

AAV with a selectable marker, the gene conferring neomycin resistance.

Using another AAV recombinant to provide the missing capsid functions,

I have isolated a stock of the recombinant AAV/neomycin virus, which

will transform human and murine cells to neomycin resistance by

establishing a proviral tLate. A maximum transformation efficiency of

10% has been obtained. Characterization of the AAV/neomycin provirus

clones, by restriction analysis, indicates a variety of proviral

structures. Two of eight clones analyzed contain a head to tail

tandem array of the AAV/neomycin genome, and are positive for rescue

following adenovirus super-infection. The six other clones contain

only single copy proviral insertions, all of which show evidence of

deletions and are negative for rescue. Thus, AAV provirus shows a high

mutation rate.

A model for AAV integration intermediates is proposed, which

involves rolling circle replication of AAV without helper virus.


During the last decade there has been a revolution in biology

which has occurred as a result of recombinant DNA technology. The

ability to use plasmids and bacteriophages as autonomous replicons for

construction of prokaryotic cloning vectors has been essential to this

revolution. Many such prokaryotic vectors are available, each with

particular capabilities and experimental advantages. Taken together,

these vector systems allow the isolation of almost any gene, provided

there is an adequate selection or screening procedure. In addition,

these prokaryotic vectors aid in sequencing of these cloned genes, and

can be used to get expression, in some cases, of eukaryotic genes in

prokaryotic environments. The use of such techniques has resulted in a

large accumulation of nucleotide sequences from both eukaryotes and

prokaryotes for study and has resulted in the identification of several

"consensus sequences" believed to be involved in gene regulation, such

as the TATA box (Goldberg, 1978; Corden et al., 1980) that is found

about 30 nucleotides 5' of the initiation site of most RNA polymerase

II transcribed genes, as well as sequences associated with RNA splicing

and polyadenylation.

Identification of such regions by sequence inspection does not

provide any functional information about such regions. This informa-

tion must be derived from functional assays. Recombinant DNA


technology allows such regions of interest to be cloned and mutated for

functional analysis. The need for an appropriate environment in which

to study these cloned genes in a biologically significant way remains

restrictive. The study of eukaryotic genes in a prokaryotic environ-

ment is unacceptable, since the environment must closely resemble the

natural environment from which the gene was obtained.

In vitro systems have been developed for the study of eukaryotic

transcription initiation and splicing (Kole and Weisman, 1982; Golden-

berg and Hauser, 1983). Although these systems are important for

defining and characterizing the processes as they occur in vivo, their

versatility and validity is open to question. Results obtained from in

vitro versus in vivo systems have been shown to be inconsistent, at

least in the study of the SV40 early region promoter (Benoist and

Chambon, 1981; Mathis and Chambon, 1981). Thus, what is needed is a

technique for reintroducing cloned eukaryotic genes back into their

natural environment; that is, back into a chromosomal or chromatin
structure in the nucleus of eukaryotic cells themselves. The develop-

ment of such vector systems, analogous to plasmids and bacteriophages

in prokaryotes, is of paramount importance in the study of eukaryotic

genes. It seems obvious that animal viruses, with their own replicons,

are the best candidates for such eukaryotic vectors.

The approach which has been taken for the use of animal viruses as

vectors has been similar to that in the development of bacteriophage

vectors (Williams and Blattner, 1980; Bramner, 1982; Lau and Kan, 1983;

Lau and Kan, 1984). Ideally, the vector genome should be easily

manipulated, well-mapped and its genomic organization known. It should

also contain convenient sites for insertion of foreign DNA, and the

location of these sites in relation to viral regulatory regions must be

well understood. The location of such control sequences is very

important with respect to the expression of inserted genes. The need

for either constitutive or induced expression from both viral and

endogenous regulatory elements can be envisioned. An additional prob-

lem, similar to the bacteriophage vectors, is that animal viruses

usually have limitations in the amount of nucleic acid which can be

encapsidated. This problem is particularly pertinent in the study of

eukaryotic genes, as many of these genes are very large. For example,

the mouse dihydrofolate reductase gene which is 42 kb (Nunberg etal.,

1980). Thus, many genes can only be cloned into viruses which have

large genomes and therefore less size restrictions, e.g. herpes simplex

virus (HSV) or vaccinia. The fact that most animal viruses have a

limited host range also creates a problem in developing a generalized

viral vector. To avoid these last two problems, size constraints and

limited host range, it may be desirable to develop viral vectors which

do not need to be packaged into infectious virions but need only to be

transfected and maintained as replicating episomes. Bovine papilloma

virus (Law et al., 1981), herpes simplex virus (Stow, 1982), BK virus

(Milaneri et al., 1981) and Epstein Barr virus (Yates et al., 1984) are

examples of viruses capable of replicating as episomes. Use of deleted

versions of large genes, such as cDNA clones, can help circumvent the

size restriction problem and allow packaging into smaller, more easily

manipulated viral genomes.

There are additional techniques for introducing genes into eukary-

otic cells besides using viral vectors. Transfection by calcium

phosphate co-precipitation (Graham and Van Der Eb, 1973) is one such

technique, and can be used in conjunction with other methods already

described. An example of this is the use of BPV-1 DNA as a vector,

using the calcium phosphate transfection technique as the method of

introduction in tissue culture cells (Sarver et al., 1981). This

transfection technique can be used to stably introduce genes into cells

even without a viral replicon being present. At low levels, such DNA

transfected into cells will integrate chromosomally and become a stable

element of the host cells genome, but this occurs in only 1 out of

104 cells at best. Microinjection is an alternative technique which

allows for higher stable transformation rates,-but only a limited

number of cells can be transfected and the technique requires consider-

able technical skill (Graessmann et al., 1980; Diacumakos, 1973).

One of the most useful outcomes from the development of animal

cell cloning vectors has been the invention of "shuttle" vectors.

Shuttle vectors can maintain their genome integrity and replicate, or

stably integrate in both animal cells and bacteria. Thus, they gener-

ally have both a prokaryotic and a eukaryotic replicon. This allows

for easy manipulation and amplification in prokaryotic cells and then

transfection into mammalian tissue culture cells for functional analy-

sis. BPV-1 (Sarver et al., 1982), SV40 (Lusky and Botchan, 1981) and

BKV (Milaneri et al., 1984) have been incorporated into bacterial

plasmids and used in this manner.

As useful as vector techniques may be for basic research in study-

ing gene regulation and function, an equally exciting use is their

potential for clinical application. Vaccinia virus, which is highly

immunogenic, may have considerable potential as a vector for inducing

expression of genes derived from other viruses and inserted into its

genome, and thus lead to a potent vaccine (Mackett et al., 1982;

Panicali, et al., 1983). Additionally, viruses which do not lead to

cell lysis or transformation but can be used to latently infect host

cells may be useful as transducing vectors for the treatment of single

gene genetic diseases.

Virus Vectors Which Replicate as Episomes

One problem in using viruses as vectors is the limitation in the

size of the DNA which can be packaged or encapsidated in the virus

capsid. There are two general virus capsid types, icosahedral (roughly

spherical) and filamentous. Both capsid types have practical limits to

the amount of the nucleic acid which can be packaged. A way to circum-

vent this problem is the use of viral replicons which can replicate as

stable episomal elements in eukaryotic cells. Such a vector could be

transfected into cells and not be subject to size limitations of the

capsid. Presumably they could be engineered, by deletion of various

genomic regions, to be unable to produce infectious virions or cause

cell lysis and, upon transfection, lead to establishment of a stable

cell line which expresses the desired gene (Law et al., 1981).

Papillomaviruses have the ability to replicate as episodes with

20-200 viral genones per cell (Law, et al., 1981; Sarver et al., 1982;

Di Maio et al., 1982). These viruses do not integrate into the host

genome unless they are defective. Bovine papilloma virus (BPV) has

been used as a vector to introduce preproinsulin (Sarver et al., 1981),
beta-globin (Di Maio et al., 1982), growth hormones (Kushner et al.,

1982), beta-interferon (Maroteaux et al., 1983) and neomycin resistance

(Law et al., 1983) into mouse cells. It was found that BPV-1, when

cloned into pBR322, required removal of specific pBR322 sequences

removed before the viral sequences would replicate. These so-called

"poison sequences" have been removed from pBR322 in the plasmid pML2

(Lusky and Botchan, 1981) and when this plasmid is linked to BPV-1 the

resulting vector can replicate in both prokaryotic and eukaryotic

(mouse) cells.

The biggest drawback of BPV-1 as a vector is that several genes

which have been cloned have been found to be cis inhibitory to BPV-1's

episomal replication (Law et al., 1983) in mammalian cells. Thus only

certain genes can be used'with BPV-1 and there-is no method for

determining if a specific gene being inserted is incompatible until the

recombinant is assayed. Additionally, the BPV-1 host range for repli-

cation is unknown and at this time limited to mouse (Law et al., 1981),

hamster (Moar et al., 1981) and rat (Lusky and Botchan, 1981). Also

BPV-1 is a transforming virus and therefore is inappropriate to be used


Simian virus 40 (SV40), a polyoma virus, has also been used as an

episomal vector. The SV40 origin of replication, on an 85 bp fragment

(Myers and Tjian, 1980), will allow replication of most DNAs linked to

it in COS cells constructed by Gluzman (1981). COS cells contain an

integrated copy of an origin defective SV40 and express large T antigen

as well as being permissive for SV40 replication. Again, as with

BPV-1, the SV40 recombinant must be constructed in a "poison sequence"

negative plasmid in order to allow efficient replication. This SV40

origin/COS cell system is in wide use today and allows for very high

replication of the recombinant genome, with as many as 4 x 105 copies

per cell (Mellon et al., 1981). Unfortunately, however, this system is

lethal to the host cell, presumably due to the massive amount of

plasmid replication, and therefore can only be used for transient

studies. Recently BK virus, another polyoma virus (80% homologous with

SV40), has been used as an episomal shuttle vector in human cells

(Milaneri et al., 1984). Milaneri et al., 1984, have used BKV to

stably transform human cells with the HSV-tk selectable marker.

HSV and Epstein Barr virus (EBV) also have potential as episomal

vectors. Small fragments containing the origins of replication for

both viruses have been identified and cloned (Stow, 1982; Yates et al.,

1984; Spaete and Frenkel ,1982), and plasmids,containing these

fragments will replicate as episomes when complemented in trans. HSV

episomal vectors do appear to have a size constraint for stable

replication which limits the total plasmid length to about 15 kb (Kwong

and Frenkel, 1984).

Virus Virions as Vectors

One main purpose of the virus capsid is to efficiently carry the

viral nucleic acid into the infected cell. Thus, it is an ideal

structure for introducing foreign DNA into cells, if one can get the

DNA packaged into the virus capsid and thus create a virus vector.

SV40 was the first eukaryotic DNA virus to be sequenced in its entirety

and its transcriptional organization and regulation understood (Tooze,

1981); therefore, it was the logical choice to be used as the first

virus vector. Because SV40 has a small genome it was necessary to make

deletions in it to allow insertion of foreign DNA and get the recombi-

nant packaged. Additionally, the recombinant genome has to be propa-

gated together with a helper virus to provide missing functions.

Initial work used a mixed infection with a recombinant SV40 virus

containing foreign DNA inserted in the late region and a temperature

sensitive early region mutant as helper (Goff and Berg, 1976; Goff and

Berg, 1979; Hamer et al., 1979). By co-transfecting at the non-

permissive temperature a mixed virus stock could be isolated. This

late region insertion strategy has now been used to transduce a number

of genes into tissue culture cells, including rabbit beta-globin

(Mulligan et al., 1979), mouse beta-globin (Hamer and Leder, 1979), rat

preproinsulin (Gruss and Khoury, 1981), human hepatitis-B virus surface

antigen (Moriarty etal.,,1981), and influenzavirus haemagglutinin

(Gething and Sambrook, 1981). Although SV40 early region replacement

vectors have also been constructed, the late region allows for higher

levels of expression. However, pure recombinant virus stocks can be

made of early region insertions by transfection into COS cells which

provide the large T antigen.

Drawbacks of the SV40 vector system are that only small inserts

can be used, approximately 2.6 kb replacing the late or early region,

and that SV40 has a limited host range.

Adenovirus (Ad) has also been used as a vector. It is a much
larger virus than SV40 and consequently is much more difficult to

manipulate. This problem has been solved somewhat by the isolation of

variant adenovirus genomes which are lacking certain restriction sites

(Jones and Shenk, 1978). Adenovirus has been used as a vector for the

transduction of the SV40 large T antigen. The large T antigen gene,

when placed under the control of the adenovirus major late region

promoter, expresses large quantities of the T antigen and therefore

allows easier biochemical characterization (Thummel et al., 1981).

Pox virus may be a very useful vector for making vaccines (Mackett

et al., 1982; Panicali et al., 1983). Unfortunately, poxvirus repli-

cates in the cytoplasm and thus it is unlikely that it can be used as a

transducing vector for producing stably transformed cell lines.

Probably some of the most useful viruses being used today as

vectors are retroviruses. Even though these are single-stranded RNA

viruses, their life cycle includes a double-stranded DNA intermediate

and provirus. The viral DNA integrates into the host cell's chromo-

somes in so as to leave the viral sequences intact to a significant

extent (Varmus, 1982). Their efficiency for integration is very high,

nearing 100%. Additionally, retroviruses usually do not kill infected

cells but these cells can continually shed virus. The integration of

these viruses is in sharp contrast to SV40 and adenovirus that almost

always have rearrangements and deletions upon integration. Retro-

viruses are actually natural transducing vectors integrating cellular

transforming genes into their genomes (Varmus, 1982). Retroviruses

have been studied extensively. Their genomic structure and transcript

organization is well understood (Dhar et al., 1980; Swanstrom et al.,

1981), and their size, 8 kb, is small enough to be easily manipulated.

They have already been used to transduce several genes into tissue

culture cells, including HSV-tk (Weeks et al., 1982), neomycin resis-

tance (Joyner and Bernstein, 1983), and HGRPT (Miller et al., 1983).

Retroviruses have also been used to transduce primary hematopoietic

stem cells in tissue culture (Joyner et al., 1983). Additionally,

retroviruses have a broad host range, and their long terminal repeats

(LTR) contain strong enhancer elements which allow for constitutive

expression of viral or inserted gene. As with SV40 deletions,

recombinant retroviruses must be complemented with helper virus to

provide missing functions. Recently, the need for a helper virus has

been replaced by the construction of helper cell lines analogous to COS

cells (Mann et al., 1983). These cell lines provide all trans

functions but do not produce wild type virus. Thus pure stocks of

recombinant retroviruses can be made.

Retroviruses do have their drawbacks, however. Cell lines created

by retrovirus infection usually continually shed virus, thus creating

contamination as well as safety problems as retroviruses are associated

with disease in animals and man. The most recent association has been

with Acquired Immune Deficiency disease (Gallo et al., 1983). Addi-

tionally, retroviruses do acquire seemingly random deletions and

rearrangements in their genome (Coffin, 1979). Another more specific

type of deletion which occurs is the loss of introns in transduced

genes (Shimotohno and Temin, 1982). An additional drawback of retro-

viruses is that inserted genes are always under the control of LTR's

enhancer elements. This usually results in constitutive expression of

the inserted gene. In many cases this is desirable, but if one wants

to study a gene under its own regulatory control, retroviruses are

unable to provide the appropriate background.

AAV, A Potentially Useful Vector

One virus that has not yet been investigated as a vector is adeno-

associated virus (AAV). It appears to have desirable traits which

would give it a high potential as a transducing vector. AAV is a

helper dependent parvovirus which requires coinfection with its helper

viruses, adenovirus or herpes simplex, in order to replicate (Atchison

et al., 1965; Buller et al., 1981; Carter and Laughlin, 1984). When no

helper virus is present, no AAV specific transcription or DNA replica-

tion can be detected (Carter and Laughlin, 1984). In this helper

deficient state AAV is not totally inert. It will latently infect the

host cell by integrating into the host cell's chromosomes (Cheung et

al., 1980; Berns et al., 1982). Presumably recombinant AAV genomes

would also integrate. Additionally, this integration has been shown to

take place in such a manner as to leave the AAV sequences intact. In

this way AAV latent infection is similar to that of retrovirus. AAV

integration has been shown to be as high as 30% in Detroit 6 cells

infected at high multiplicities (Hoggan et al., 1972). If these

latently infected cells are superinfected with a helper virus, the AAV

genome will be rescued, replicate, and produce infectious virus. This

inducible rescue is a useful feature which retroviruses do not have.

Also, AAV is transcriptionally silent without helper virus, and

therefore presumably would not interfere with the expression of

inserted genes under their own regulatory control. Retroviruses are

unable to provide such a silent background. Additionally, AAV has an

extremely broad host range and has been found to replicate in every

mammalian species tested, given the appropriate helper virus (Casto et

al., 1967). Thus, AAV has several features which give it a high

potential as a transducing vector, and these features will be discussed

in greater detail in Chapter Four.

Parvoviruses are among the smallest of the eukaryotic DNA viruses.

AAV has a linear single-stranded DNA genome of 4,675 bases (Srivastava

etal., 1983), and both the positive and negative strands are encapsi-

dated (Berns and Adler, 1972). AAV has terminal repeats of 145 bases

in length, of which the first 125 are internally palindromic and can

base pair into a "T" shaped cruxiform or hairpin structure. The

terminal repeats can be found in two orientations that represent inver-

sions of each other: "flip" and "flop" (Spear et al., 1977; Lusby et

al., 1980; Samulski et al., 1982). This finding has lead to a model of

replication for AAV replication (Cavalier-Smith, 1974; Straus et al.,

1976) in which AAV replication proceeds from the 3' hairpin structure

to the end of the molecule. The newly completed double-stranded

terminal repeat presumably dissociates, the 3' cruciform structure

reforms, and replication proceeds in a "rolling hairpin" mechanism.

The hairpinned termini can be resolved by way of a putative site-

specific "nickase" clipping at the internal palindrome junction and

giving rise to a terminal sequence inversion, hence "flip" and "flop".

Only adenovirus early genes appear to be required for productive

AAV infection. Adenovirus mutants in E1A, E1B, E2A and E4 do not

supply the required helper function for AAV replication (Carter and

Laughlin, 1984), although microinjected Ad 5 E4 mRNA appears to fully

help AAV replication and capsid protein production (Richardson and

Westphal, 1981). Possibly E1A and E2A are only necessary in positive

regulation of E4.

My goal was to map the genomic organization of AAV and thus

develop a strategy for using it as a vector. Additionally, AAV has

several unique features and therefore its study would be of biologic





Human Detroit 6 cells (D3405, hereafter called D6) and D5 (a 06

derived cell line which is latently infected with wild type AAV and was

previously called B7374IIIDV) were obtained from M.D. Hoggan (National

Institutes of Health). Human KB cells were agift from H. Ginsberg

(Columbia University); mouse L thymidine kinase negative (tk-) cells

were obtained from S. Silverstein (Columbia University); and HeLa cells

were obtained from Dr. C.H.S. Young. These cell lines were maintained

in minimal essential medium containing 10% fetal calf serum, 1% gluta-

mine, penicillin and streptomycin. Cell line 293-31 (Graham et al.,

1977), a human Ad transformed cell line, were obtained from T. Shenk

(Stony Brook) and were maintained in Dulbecco modified Eagle medium

containing 10% fetal calf serum, 1% glutamine, penicillin and strepto-

mycin. All cells were passed bi-weekly.

Enzymes and Linkers

Restriction enzymes, T4 ligase, T4 DNA polymerase, T4 polynucleo-

tide kinase, E. coli DNA polymerase I and the Klenow fragment were

purchased from New England Biolabs and Bethesda Research Laboratories,

and the enzyme reactions were performed according to the suppliers'

specifications. Bg1II linkers were purchased from New England Biolabs;

BamHI linkers were purchased from Collaborative Research.

Construction of Deletion and Insertion Mutants

Form I pSM620 plasmid DNA was partially digested with HaeIII,

BstN1, Pstl, Ncol or Apal, and completely digested with XhoI. Table

III indicates the restriction enzyme used in the case of each mutant.

The products of the digestions were fractionated on 1.0% agarose gels,

and full-length or nearly full-length linear DNA was extracted by

electroelution (Muzyczka, 1980) and concentrated by passage through

Elutip-d columns (Schleicher and Schuell, Inc.). In the case of the

Xhol and Apal digested DNA, the linear DNA was ligated to circles and

cloned without further modification. Other DNAs had one of two foreign

DNA fragments inserted before circularization and cloning. These

foreign DNAs were either an 8 bp BgII linker or a 263 bp CbI fragment

that contained the gene for a suppressor tyrosine tRNA. The latter

fragment was isolated from the plasmid pSV-tT-2 (Su+) constructed by

Laski et al. (1982) and kindly supplied by P. Sharp (Massachusetts

Institute of Technology). If the termini of the fragment or insert

contained either a 3' or 5' single-stranded DNA overhang, the ends were

repaired with T4 DNA polymerase or the Klenow fragment in the presence

of all four deoxynucleoside triphosphates (Rawlins and Muzyczka, 1980)

before ligation. After blunt end ligation with kinased BglII linkers,

the products of the reaction were digested to completion with BglII and

(to ensure only one linker was inserted) precipitated with ethanol.

The BgII linears were then religated back to circles before cloning.

None of the mutants were sequenced but it was possible to deduce

the probable nature of the mutation on the basis of the enzymes used in

the construction of each mutant (Table 3-1) and the known DNA sequences

of AAV (Srivastava et al., 1983) and the inserts. The final ligation

products were transfected into E. coli HB101 (Rawlins and Muzyczka,

1980) and the individual clones were examined by rapid heat-lysis DNA

preparation (Holmes and Quigley, 1981) and restriction enzyme analysis.

Mutants that contained a net gain of nucleotides were designated inser-

tion mutants (ins) and named by the map position of the insertion.

Mutants that resulted in a net loss of nucleotides were designated

deletion mutants (dl) and named by the range of the deletion map units.

The isolation mutant number was included to avoid confusion between

similar mutants.

Finally, some of the mutants resulted from reconstruction using

preexisting mutants. These were as follows: d152-91 (pHM1320) was

constructed by digesting ins52 and d10-91 with BgII and EcoRV diges-

tion and ligation of the fragments; d149-94 (pHM3305) was constructed

by adding BglII linkers to the 1,700 lip Hinfl fragment of pSM620

(containing AAV mu94 to 100 and adjacent pBR322 sequences) and digest-

ing the products with EcoRV and BglII. This restriction fragment

mixture was then ligated to a BglII-EcoRV double digest of ins49.

Construction of AAV/neo, dhfr and Bacteriophage X Recombinant Plasmids

The construction scheme of d152-91/neo and d152-91/dhfr is illus-

trated in Figure 4-1. The plasmids pBR-neo (Southern and Berg, 1982)

and pSV2-dhfr (Subramani et al., 1981) were obtained from P. Berg

(Stanford University). Ins96 was derived from the plasmid, pSM620

Xbal, kindly supplied to us by R.J. Samulski (Stony Brook). The pSM620

Xbal plasmid is derived from pSM620 by the insertion of an Xbal linker

into the Ncol site at seq 4483 (mu96). This plasmid was digested with

XbaI, T4 polymerase blunt ended, and a BglII linker was inserted to

give ins96 (pHM2904). Bacteriophage X Sau3A fragments were then

inserted into the BglII site to yield ins96/X-R, ins96/X-M and

ins96/X-F with inserts of 550 bp, 1100 bp and 2800 bp respectively.

dl10-37/EPR was constructed by inserting the SV40 early region

promoter, obtained from pSV2-dhfr, into d110-37. To obtain a 400 bp

fragment containing the SV40 EPR with BgII compatible sticky ends the

plasmid pHM2102 (Figure 4-1) was digested with HindIII and a BgII

linker was inserted to give pHM2102- EPR. BamHI-BglII digestion of

this plasmid yield the 400 bp SV40-EPR fragment with 5' BamHI and 3'

BgII sticky ends.

Plasmid Preparation and E. coli HB101 Transfection

Form I DNA was prepared by an altered heat lysis technique (Holmes

and Quigley, 1981). Plasmid containing E. coli HB101 were grown in 500

ml culture in a shaker-incubator at 37C to a Klett value of 80 (green

filter, Klett-Summerson Colorimeter). The culture was induced with

chloramphenical at 1.7 ug/ml and allowed to incubate for 16-20 hours

and pelleted by low speed centrifugation. The pellet was resuspended

in 10 ml of STE (0.1 M NaCl, 10 mM Tris-Cl (pH 8.0), and 0.1 mM EDTA)

and transferred to a 30 ml glass centrifuge tube to which was added 1

ml of a freshly made lysozyme solution of 20 mg/ml in 10 mM Tris-Cl,

pH 8.0. With the use of a clamp the glass tube was held in boiling

water for 35 to 40 seconds and then put on ice for five minutes. The

tube was then spun in a Beckman JA-20 rotor at 12,000 rpm for 40

minutes. The resulting supernatant was then extracted with phenol

twice and precipitated with ethanol. This DNA was then dissolved in 1

x SSC and form I plasmid DNA was purified by equilibrium centrifugation

in a CsC1 gradient containing 1 mg/10 ml of ethidium bromide. The

density of the CsCl solution before centrifugation was 1.56 g/ml and

the gradients were formed by centrifugation at 40,000 rpm in a Ti50

rotor at 4C for 24 to 48 hours.

Plasmid DNA was transfected into HB101 E. coli in the following

manner. E. coli HB101 was grown to a density of 120 to 130 Klett units

(green filter, Klett-Simmerson colorimeter) in Luria broth (1.0%

tryptone, 0.5% yeast extract, 0.05% NaCl, and 0.5% glucose, pH 7.4).

Two ml of the culture was centrifuged at 1500 rpm for ten minutes at

4C and resuspended in 1 ml of 100 mM CaC12 and allowed to incubate

at 4C overnight. DNA, 40 to 100 ng, was then added to these cells,

allowed to incubate for ten to 30 minutes, and then the cells were heat

shocked at 420C for two minutes. The cells were then diluted with 5 ml

of Luria broth for 50 minutes, pelleted and plated on Luria agar plates

containing the appropriate antibiotic. At approximately 24 hours after

incubation at 37C, colonies from these plates were picked for


Plasmid DNA from these HB101 clones was isolated by a rapid heat

lysis method (Holmes and Quigley, 1981) for restriction analysis

(Klenow end labeling), and gel electrophoresis. Ten ml cultures of the

plasmid containing E. coli were centrifuged and resuspended in 400 ul

of STET (8% sucrose, 5% Triton X-100, 50 mM EDTA and 50 mM Tris pH 8.0)

to which was and 5 ul of a lysosome solution of 10 mg/ml. The cells,

after incubation on ice for ten minutes, were boiled for 40 seconds and

centrifuged for ten minutes at 12,000 x G at room temperature. The

resulting supernatant was phenol extracted, ethanol precipitated and

resuspended in 10 fIM Tris, 1 mM EDTA. One ml of cell culture yields

approximately 1 to 2 ug of plasmid DNA.

Agarose Gel Electrophoresis
Electrophoresis through agarose gels was used routinely for prep-
arative and analytical DNA separations. Gels contained 1.0 to 1.4%

agarose (BRL) in TEA buffer (0.04 M Tris, 0.02 M sodium acetate and 2

mM EDTA (pH 7.8) or one half TBE buffer (0.045 M Tris-borate, 0.045 M

boric acid and 0.002 M EDTA). The gels were stained for 15 minutes in

the appropriate buffer containing 1 ug/ml ethidium bromide (Sharp et

al., 1973) and the DNA bands were visualized with short-wave ultra-

violet light. Photographs of the gels were obtained using Polaroid

Type 57 film.

Klenow Fragment End Labeling
DNA prepared by rapid isolation (Holmes and Quigley, 1981), about

100 ng, was digested with the appropriate restriction enzyme in 30 ul

of the appropriate buffer. After digestion, 1 uCi of [a-32P]dCTP

and 0.5 units of Klenow fragment (large fragment E. coli Polymerase I)

was added to the restriction buffer and the solution incubated at room

temperature for 30 minutes. This reaction mixture was then loaded onto

1.0 to 2.0% agarose gels for electrophoresis.

DNA Transfections into Tissue Culture Cells
Tissue culture cells were CaCl2 transfected with 1 to 11 ug of
form I plasmid DNA essentially as described by McCutchan and Pagano

(1968). The DNA transfection solutions contained 500 ug/ml DEAE-
dextran (MW=500,000 d, Pharmacia in 2 mls of MEM buffered with 0.45 mM

HEPES (pH 7.4). The transfected cells were typically at 70% conflu-

ence. The transfected cells were infected with Ad-2 at an MOI of 5 and

then treated with 0.1 mM choroquin diphosphate for 20 minutes to four


Mutant and Wild-type AAV Virus Stock Preparation

Cells were transfected with AAV plasmid via DEAE-dextran method,

infected with Ad-2 and allowed to incubate for 36 to 48 hours. At this

time the medium and cells were frozen and thawed three times and heated

to 560C for ten minutes to heat inactivate the Ad-2.

Mutant lip and wild-type virus titers were determined by immuno-

fluorescence of infected cells with anti-AAV capsid antibody (Carter et

al., 1979), (a kind gift of M.D. Hoggan), or by quantitating the amount

of AAV DNA synthesized in cells infected with samples of the viral

stocks. In the latter method, virus stocks were used to infect human

cells, and the yields of monomer-length duplex DNA in mutant and

wild-type infections were compared.

Recombinant Virus Stock Preparation

Five ug of d152-91/neo or d152-91/dhfr plasmid DNA and 0.5 ug of

ins 96/X-M plasmid DNA were co-transfected into Ad-2 infected (moi=5)

KB or 293-31 cells using the DEAE-dextran method. Alternatively 10 ug

of d152-91/EPR were co-transfected into Ad-2 (MOI=5) KB cells. The

transfected cells were treated with 0.1 mM chloroquin diphosphate

(Luthman and Magnusson, 1983) for 20 minutes to four hours. Two days

after transfection, the cells were frozen and thawed twice and passed

through a 0.45 um filter to remove cellular debris. Contaminating Ad-2

helper virus was inactivated by heating the virus stock at 56C for two


Hirt DNA Extraction from Transfected, Newly Infected
or Latently Infected Tissue Culture Cells

Low molecular weight DNA was isolated from tissue culture cells at

24 to 48 hours after transfection, infection, or rescue from latent

integration by the method of Hirt (1967) as previously described

(Muzyczka, 1980). Medium as removed from the plates and the cells were

lysed with 0.6% SDS, .1 mn EDTA. Pronase was then added, which had

been predigested for one hour at 370C, to a final concentration of 0.8

mg per ml and incubated for 30 minutes at 370C. NaCI was then added to

give a final concentration of 1 M and the DNA solution was incubated at

4C overnight. The next day the high molecular weight DNA precipitate

was pelleted by centrifuging at 12,000 x g for 30 minutes. The

supernatant was then phenol extracted, ethanol precipitated and

redissolved in 10 mM Tris, 1 mM EDTA for further analysis.

Genomic DNA Isolation

Confluent 10 cm plates of tissue culture cells was lysed with 5 ml

of a solution of 0.6% SDS, 1 nM EDTA and 0.8 mg/ml predigested pronase.

After incubation overnight at 370C the viscous solution was phenol

extracted twice, chloroform extracted twice and dialyzed against 10 mM

Tris, 1 mM EDTA (pH 7.4). In some cases, before phenol extraction, the

samples were treated with RNase A at 10 ug/ml for one hour at 370C.

The DNA solution concentration was quantitated by gel electrophoresis

and ethidium bromide stain and/or spectrophotometric analysis.

Southern Blotting and Hybridization

The DNA was fractionated by electrophoresis on 1.4% agarose gels

and transferred to nitrocellulose by method of Southern (1975). The

DNA was denatured by soaking the gel twice for 0.15 minutes in 0.5 N

NaOH, 0.5 M NaCl and then twice for 15 minutes in 0.5 M Tris-HCl (pH

7.4), 0.5 M NaC1. The DNA was transferred from the gel onto a

nitrocellulose filter (Schleicher and Schuell) by blotting in an upward

direction with 6 x SSC (pH 7.0) for 16-24 hours. After blotting, the

nitrocellulose filter was baked for two hours at 80C in a vacuum oven.

The nitrocellulose filter was then prehybridized for four to six hours

at 680C in 15 to 100 ml of 6 x SSC, 1 x Denhardt (0.2% polyvinylpyro-

lidine, 0.02% ficoll, 0.02% BSA). Hybridization was carried out for 12

to 24 hours at 680C in 10 ml of the same solution containing 0.1 mg/ml

denatured herring sperm DNA and 0.1 ug/ml denatured (32P) DNA.

When genomic DNA was being probed, dextran sulfate (Pharmacia Fine

Chemicals) was added to the hybridization solution to a final concen-

tration of 10%. Following hybridization, the filters were washed at

68C once each for 30 minutes in 20 ml prehybridization solution and in

100 mM KPO4 and twice in 20 ml of 0.1 x SSC, 0.6% SOS. The filters

were then air dried and autoradiographed at -700C with Cronex-4 X-Ray

film (Dupont) or in the case of genomic blots XAR-5 x-ray film (Kodak)

and a Dupont HiPlus intensifying screen.

Nick Translation

The DNA was labeled with 32P by a modified in vitro nick-

translation procedure (Rigby et al., 1977). The reaction mixture (200

ul) contained 0.5 ug DNA, 50 mM Tris-HC1 (pH 7.8), 5 mM MgCl, 10 mM
2-rercaptoethanol, 0.1 mM each of dGTP, dATP and dTTP, 0.5 uM
(a-32p) dCTP (>400 Ci/mmol, Amersham Corp.), 1 ng DNase I and two

units of DNA polymerase I. The reaction was incubated at 17C for one

to six hours and was then extracted with phenol. Unincorporated deoxy-

ribonucleotides were removed by chromatography on Sephadex G-75

(Pharmacia Fine Chemicals) in 10 m Tris-HCl (pH 8.0). The specific

activity of the probe obtained was usually in the range of 0.5 to 2.0 x

108 cpm/ug DNA.

Transduction of Mammalian Cells with Recombinant Virus

06, KB or Ltk- cells were plated at 102 to 105 cells/dish.

Cells were infected with recombinant virus stocks 12 to 24 hours after

plating and selected for drug resistance at 12 to 24 hours or seven

days post-infection. D6 and KB cells were selected with 1 mg/ml Gene-

ticin, G-418 sulfate (Binetruz et al., 1982) (Gibco); Ltk- cells were

selected at 0.4 mg/ml. These G-418 antibiotic concentrations killed

all uninfected control cells in six to ten days. The media was changed

every three to four days. At eight to 14 days post-selection cells

were fixed with ethanol, stained with Giemsa (Fisher) and counted.



Adeno-associated virus appears to have several useful character-

istics which indicate that it could be a used as a mammalian cloning

vector: i) AAV can readily set up a latent infection when no helper

virus is present (Ad or HSV) (Hoggan etal., 1972). ii) It appears to

integrate into chromosomal sequences in such a manner as to leave its

own sequences intact (Cheung et al., 1980; Berns et al., 1982).

iii) AAV can be rescued from its integrated state by superinfecting the

latent cell with one of its helper viruses (Atchison et al., 1965;

Buller et al., 1981; Carter et al., 1983). Before I could determine if

it was possible to use AAV as a vector I first needed to map the func-

tional regions of its genome. The determination of these regions would

allow me to develop an AAV vector strategy. It was previously deter-

mined that the terminal repeats were cis required for AAV rescue and

replication. Therefore, the next step was to dissect the internal

sequences by mutation mapping and determine what regions could be

deleted and what functions would be lost by such deletions. I assumed

that the AAV capsid had a limited capsid packaging capacity of near

wild-type size (4.7 kb) and therefore the limit to the size of foreign

DNA I could insert would be determined by the size of the deletions I

could construct and still have the minimum cis-required sequences

needed for rescue, replication, packaging and integration.

Although AAV is a defective virus and its genetics could not be

studied by conventional methods, there was a considerable amount of

information which could be gleaned from the study of its protein,

transcripts and DNA sequence.

AAV Proteins

There appear to be five new polypeptides synthesized in vivo in

AAV and Ad infected cells when compared with cells infected with only

Ad (Buller and Rose, 1978; Janik et al., 1984). Three of these

proteins are capsid proteins, which can be immunoprecipitated with

anti-capsid antibodies. These are VP1 (85-90 kd), VP2 (72-73 kd) and

VP3 (60-67 kd). VP3 comprises about 85% of the capsid, while VP1 and

VP2 comprise about 7% each. Tryptic peptide mapping of these three

proteins indicates that most of their amino acid sequences are the same

(Johnson et al., 1977). Additionally, the smallest AAV transcript, 2.3

kb, when translated in vitro, gives rise to all three capsid proteins

(Jay et al., 1981). Therefore the genes encoding these capsid proteins

have a high degree of sequence overlap. Two other proteins identified,

50 kd and 67 kd, were not immunoprecipitated by anticapsid antibody and

their function is unknown (Janik et al., 1984).

AAV Transcripts

There are six known RNA transcripts made by AAV during productive

infection (Green and Roeder, 1980; Green et al., 1980; Carter et al.,

1984; Lusby and Berns, 1982) (Figure 3-1). All are polyadenylated and

transcribed from the minus (-) strand. These transcripts can be

categorized into three sets. Members of each set have the same tran-

scription start (5') and stop (3') site but differ as to whether there

is a splice or not. The spliced transcripts all have the same splice

between map units 41 and 49. All six transcripts have the same poly-

adenylation site at map unit 96. The three transcription start sites

map to map units (mu) 6, 19 and 39 and thus give rise to unspliced

messages of 4.2, 3.6 and 2.6 kb in length and spliced messages of 3.9,

3.3 and 2.3 kb in length (Figure 3-1). The smallest transcript, 2.3
kb, is the most abundant, comprising 50% of all AAV messengers. As

this transcript gives rise to all capsid proteins upon in vitro trans-

lation the capsid genes must map within the 49 to 96 map unit region.

AAV Sequence Analysis

The DNA sequence of AAV has been analyzed and three large open

reading frames and several smaller ones have been identified

(Srivastava et al., 1983) (Figure 3-1). There is a major open reading

frame on the right side of the genome which maps between mu60 and 93

and has the potential to code for a protein of about 63 kd. The amino

acid sequence derived from this open reading frame agrees well with the

amino acid composition of VP-3 (Srivastava et al., 1983; Rose et al.,

1971). Therefore, all indications are that this open reading frame is

the VP-3 gene and that additional sequences are used in addition to

this open reading frame to produce the larger capsid proteins VP1 and

The left side of the AAV genome contain one large open reading

frame (ORF). The indication that this ORF represents two different

putative genes is the finding that there are two transcripts coining

from this region and that two transcription start sites (map units 5

and 19) map just 5' to translation initiation sites (map units 7 and

22). The complete sequence characterization of these putative genes is

confused by the fact that there is a splice present (map units 41 to

49) in the 3' region of the reading frame. This gives rise to two

different carboxy terminal coding sequences, one ending in the splice

at map unit 48 and the other extending just 3' of the splice. There-

fore, taken together, there is a potential for four different large

proteins being encoded by the left side of the AAV genome.

It was previously suggested that this left hand region could code

for a protein(s) required for replication (Laughlin et al., 1979).

During productive AAV infection naturally occurring deletion mutants,

called defective-interfering (DI) particles, arise. It was found that

certain types of DI particles could not replicate without the presence

of wild-type AAV. Additionally, an in vitro constructed deletion

mutant (d138-43) was also found to be replication negative (Cheung,

1979). Unfortunately, DI particles are of a heterogeneous population

and the in vitro deletion effected all AAV transcripts so there was no

way of knowing where the putative replication gene was located.

There are also several smaller open reading frames which have been

identified in the sequence (Srivastava et al., 1983). One is located

within the splice, located from map units 41 to 46. Another is located

just upstream of the VP-3 open reading frame (mu60-93) at map units 47

to 54. A third open reading frame is located out of phase, but over-

lapping, with the VP-3 open reading frame at map units 84 to 94. The

functions of these open reading frames are unknown, but it is

Figure 3-1. Physical structure of AAV RNAs and open reading frames.
The solid lines above the map position line indicate the structure of the AAV spliced
(2.3, 3.3 and 3.9 kb) and unspliced (2.6, 3.6 and 4.2 kb) RNAs. The solid boxes indicate
the major open reading frames. The empty box entending to mu 48 from the mu 60 to 93 open
reading frame indicates an open reading frame which does not have an ATG start codon but may
be used as coding sequences.

2.3 Kb


3.9 Kb


3.6 Kb

4.2 Kb
0 10 20 30 40 50 60 70 80 90 100

speculated that they may be the additional coding sequences required

for VP-1 and VP-2 production. Another possible coding sequences for

possible coding sequence for VP-1 and VP-2 may be a 5' extension of the

mu60 to 93 open reading frame, extending upstream to mu48, which does

not have an ATG start codon (Figure 3-D) but could be connected to an

upstream start codon by an unmapped spliced transcript.

AAV/pBR322 Recombinant Plasmids

The largest problem to be circumvented in studying AAV has been

its nature as a defective virus. AAV, by itself, is not a productive

or lytic virus, thus mutants cannot be isolated and studied (plaque

purified) like conventional lytic viruses. Fortunately, a technique

has been found for circumventing this problem. It has been found that

AAV cloned into pBR322 not only allows for prokaryotic amplification of

AAV sequences, but also such recombinant plasmids can be transfected

into adenovirus infected tissue cultures and AAV productive infection

will occur (Samulski et al., 1982). This indicates that such plasmids

may be an accurate model for studying chromosomal rescue from latent

infection. Thus, it is possible to mutate the AAV sequences of the

plasmid by standard molecular genetic techniques and then to observe

the altered phenotypes by simply transfecting the mutant plasmid into

Ad infected cells. This technique has already been used to study the

terminal repeat sequences of AAV and determine their cis-active

requirements for rescue and replication (Samulski et al., 1983;

LeFebvre et al., 1984). Using this approach I have mutated the

internal sequences of AAV to map the functional regions of its genome.

This was done by constructing a large number of insertion and deletion

mutations and determining their altered phenotypes.


Construction and Physical Characterization of the AAV-2 Mutants

The AAV insertion mutants were constructed as described previously

(Heffron et al., 1978). The wild-type plasmid, pSM620, was cleaved to

linear form by partial digestion by one of several multiple cut

restriction enzymes. The full length or close to full length molecules

were isolated by electroelution after agarose gel electrophoresis. On

those DNAs cut by restriction enzymes which left an overhang, either T4

polymerase or the Klenow fragment front E. coli polymerase was used to

blunt end the duplex DNA (Rose et al., 1971). An 8 bp BgII linker

fragment was ligated into the resulting blunt ended restricted site.

Many of these mutants constructed by this procedure were expected to be

frameshift mutants but those constructed by partial digestion with

BstN1 were expected to produce an in-frame 9 bp insertion. In addition

to the insertion mutants being produced we also isolated numerous

deletion mutants. These mutants were deleted between two restriction

sites in the AAV genome and had an 8 bp BglII linker insertion. The

BglII linker was chosen for these mutant constructions as there is no

site present in the wild-type plasmid and the 4 bp single-stranded

overhang left by BglII digestion is compatible with those left by

restriction enzymes BamHI, BclI, Sau3A and Mbol. The use of this

linker simplified the restriction enzyme mapping of the mutants (not

shown) and provided a unique restriction site into which we could

insert foreign DNA.

Typically, these linker insertion and deletion mutants were

characterized by at least three different restriction enzyme

digestions. PstI, BglII double digestions were performed on all

mutants. Single PstI digestion was sometimes also performed. PstI

cleaves the AAV genome away from pBR322 and into four fragments.

Further double digestions of BglII in conjunction with a single cut or

a double cut restriction enzyme for the wild type pSM620 plasmid were

also performed. The single cut enzymes used were PvuII and EcoRV which

cut once in pBR322. The double cut enzymes used were HindIII and BamHI

which both cut once each in the pBR322 and AAV sequences of the

plasmid. Further digestions were performed if appropriate.

Three additional classes of mutants were also constructed. One

class, constructed by Mark A. Labow, were simple deletions between

adjacent restriction sites made by a particular restriction enzyme

(e.g., Xhol digestion d148-52). A second class of mutants was

constructed and were similar to the BglII linker inserts but instead

contained an insert of a 263 bp DNA fragment that contained a tyrosine

suppressor tRNA gene (Laski et al., 1982) (e.g., ins 80 tsup). These

two mutants were constructed by Richard Wright and Phil Collis. A

third class of mutants were constructed by using already constructed

mutants. The mutant pHM1320 (d152-91) was constructed from ins52 and

dlO-91. This was done by digesting ins52 and d10-91 with BglII and

EcoRV and lighting the resulting fragments. The mutant pHM3305,

d149-94, was constructed by lighting BglII linkers onto the 1700 bp

Hinfl fragment of pSM620. This fragment contains AAV sequences from

mu94-100 and adjacent pBR322 sequences. The fragment was then digested

with BglII and EcoRV and then ligated with fragments from a BglII and

EcoRV digestion of pHM1523 (ins49).

For simplification, insertion mutants are designated by ins and

deletion mutants by dl followed by the map unit position of the

insertion or range of the deletion (Table 3-1).

The construction of these mutants is summarized in Table 3-1 and

shown in Figure 3-2. Table 3-4 is a listing of the plasmid transfec-

tion done into HB10ls. None of these mutants have been sequenced. The

nucleotide numbers included are based on restriction enzyme analysis,

the particular restriction enzyme used and the AAV sequence, and the

method used to repair the overhang.

Mutants that are Defective for Duplex DNA Replication

The mutants were first tested for their ability to rescue and

replicate the AAV sequences by DEAE-dextran transfection (McCutchan and

Pagano, 1979) into adenovirus-2 infected cells. Human Ad-5 transformed

293 cells, D6 cells or HeLa cells were used as the host. The Hirt

supernatant DNA (Hirt, 1967) was harvested at 36 to 48 hours post-

infection and analyzed by Southern blotting (Southern, 1975) with
32P-labeled AAV DNA being used as the probe. A representative

selection of the mutants are shown in Figure 3-3 and Figure 3-4 and a

summary of the results from all the mutants is listed in Table 3-2.

It was found that the left half of the AAV-2 genome was involved

in rescue and replication of AAV DNA. Mutants located at and in

between map units 11 and 42 were defective in their ability to produce

duplex AAV monomer DNA when transfected into Ad infected cells. These

mutants were called rep mutants for this replication defective pheno-

type. Usually only input plasmid DNA forms were detected with no

evidence of any rescue or replication, but in two cases there were low

levels of rescue and replication taking place. These two mutants

pHM1505 (ins23) and pHM1411 (ins42) were both able to produce low

levels of monomer duplex AAV DNA at 0.1 to 1.0% from that seen in a

wild-type plasmid transfection (Figure 3-5). One of these mutants,

pHM1505 (ins23), was also capable of producing virus at detectable

levels, 0.1% (Table 3-2). A 400 bp bacteriophage A fragment was

inserted into the BgII site of pHM1505 (ins23) and the resulting

mutant was completely defective for replication to the limit of


Mutants in the right half of the AAV genome (mu48-96) were, with

one exception, able to rescue and replicate at near normal levels for

production of monomer duplex DNA. The mutant which did not replicate

as well was pLB314 (d180-96). This mutant, assayed by Mark Labow,

produced only 5% the amount of monomer duplex DNA as that of wild-type.

The reason for this mutant phenotype is probably a result of the

missing polyadenylation signal at mu95 which is deleted. Deletion

mutants in the right half of the genome produced monomer duplex progeny

of the appropriate size.

Mutants that were Defective in Generating Infectious Virus and
Single-stranded Progeny DNA

As discussed in the Introduction, mutants which were located in

the presumed VP-3 coding region (mu60-93) would probably not be able to

produce infectious virus. As shown in Figure 3-3 and Figure 3-4, these

mutants could replicate as duplex DNA. These mutants were tested for

virus production by a two passage transfection experiment. Plasmid DNA

was DEAE-dextran transfected (McCutchan and Pagano, 1968) in 293, HeLa

or 06 cells infected with Ad and allowed to productively replicate for

48 hours. At that time the plates were frozen and thawed three times,

heated to 56C for 20 minutes, then an aliquot of 3 ml of the media

transferred to a fresh plate of Ad infected cells. AAV mutant virus

produced in the first plasmid transfected plate would infect and

multiply in the second. The Hirt supernatant DNA (Hirt, 1967) was

harvested from the second plate at 36 hours post-infection and analyzed

by Southern blot (Southern, 1975). None of the mutants tested from

nu63 to 91 could produce detectable virus (Table 3-2, Figure 3-6).

AAV virions contain single-stranded DNA and during wild-type AAV

infection single-stranded progeny can be detected. I could detect only

duplex monomer DNA replicative forms of the mutants mapping between

60-91; no single-stranded DNA progeny was evident (Table 3-2 and Figure

3-3). Because these mutants are probably defective in the major capsid

protein, VP-3, they were called cap mutants.

Mutants that Produced Single-stranded Progeny but Low Virus Yields

AAV mutants between map units 48 and 55 were able to replicate

monomer duplex DNA. These mutants, unlike the cap mutants, also

produced single-stranded progeny (Table 3-2). These mutants are

represented by ins52 (pHM347) and d148-52 (pLB202) in Figure 3-3. The

virus yield of these mutants was tested in a similar manner, namely,

two plate passage, as the cap mutants and compared with that of log

dilutions of an equivalent amount of wild-type plasmid being
transfected. They gave an approximate infectious titer of 0.1 to 1%

that of wild-type (Figure 3-7). The mutant stocks were also tested by

indirect immunofluorescence using anti-AAV capsid antibodies (data not

shown). Both titering techniques gave similar results (data not

shown). These mutants appeared to produce a defective capsid resulting

in a low infectious particle yield and were therefore called lip


Complementation and Recombination Between Mutant Phenotypes

I next wanted to determine if I could complement one of our three

mutant phenotypes with another. We first tried complementing the

replication defective rep mutants with mutants of the lip or cap

(replication positive) phenotype. In order to do this I mixed plasmids

of the two mutant phenotypes together and transfected the mixture into

Ad infected cells. The Hirt extracted DNA was then harvested at 36 to

48 hours, examined by agarose gel electrophoresis and Southern blotted.

When two BglII linker insertion mutants were mixed together for analy-

sis the DNA extracts were digested with BglII so I could differentiate

the amount of duplex progeny produced (Figure 3-3) by each mutant. In

all cases, when a rep mutant was complemented with a replication

positive lip or cap mutant both genomes were found to replicate (Figure

3-3). For example, in the ins42/ins63 complementation (Figure 3-3),

five strong BglII restriction fragments were seen. These bands were:

a full length uncut monomer AAV band (4.7 kb, 100%); a 63% (of full

length AAV) fragment and a 37% fragment which were derived from ins63

digestion; and a 42% and 58% fragment which were derived fran ins42

digestion. The full length AAV monomer DNA which remained uncut by
BgII is probably due to the reannealing of the single-stranded DNA of

ins42 and ins63 to give duplex DNA and would be resistant to Bg1II

digestion at both sites.

Also noteworthy is the appearance of a light additional band, not

predicted by BgII digestion of the input genomes. This was a faint

21% fragment and it was seen in all ins42/ins63 complementations. It

was apparent that this band was derived from BgII digestion of an

ins42/ins63 double mutant. Therefore there was recombination taking

place between these two replicating mutants and the yield of this 21%

fragment represents an estimate of the recombination frequency during a

lytic cell cycle or approximately 1% (Figure 3-3). Such double mutant

recombinant bands were seen during all replication positive complemen-

tation when digested with BglII. As an example, complementation of rep

mutant insll with cap mutant ins63 resulted in the appearance of a 52%

size fragment characteristic of such a double mutant (Figure 3-3).

These data indicates that both lip and cap mutants can complement

replication defective rep mutants. I also wanted to determine if

mutants which were defective for production of wild-type virus (lip and

cap) but were replication positive could be complemented by using rep

mutants which appeared to leave the right side of the AAV genome

intact. As in the complementation of the rep mutants, again I mixed

representative mutants and transfected. I altered the protocol to a

two plate passage experiment, similar to the analysis of the lip

mutants, to determine if indeed the lip and cap mutants could be

packaged into infectious capsids. The results are shown in Figure 3-6

for the complementation of capsid mutant ins63. All three rep mutants,

insll, d110-37, and ins42 could complement ins63 to various levels to

be packaged into infectious AAV capsids. Ins63 alone produced no

detectable AAV virus. In addition, d110-37 has been found to

complement lip-cap mutant d152-91 in a similar manner, Table 3-3.

Figure 3-2. Physical structure of AAV mutants.
Above the map position line are AAVs two large open reading
frames. Below the map position line, the arrows indicate the positions
of insertion mutations and the interrupted lines indicate the positions
of AAV sequences missing in the deletion mutants.



insl Insll ins23 ins32 ins42ins ins55 ins63 ins78 ins86 ns96
insl3 ins27 ins52 ins80 ins91
t sup tsup

dl3-22 -
dl3-49 -
dl3-94 -
dl10-37 ---
d113-96 ----
dl23-28 ---
dl27-41 ------



dl 72-77
dl -11l
dlll-I0 ---
dl 91-100



Physical Structure of AAV Mutants

Mutant+ SiteA Bp




insl (pHM351) HaeIII 0 8 40

inslltsup(PHM1412) PstI 4 8 499

insl3 (pWM108) Ncol 0 263 625

ins23 (pHM1505) BstNI 0 9 1059

ins27 (pHM322) HaeIII 0 8 1278

ins32 (pHM326) HaeIII 0 8 1488

ins42 (pHM1411) PstI 4 8 1962

ins49 (pHM1523) BstNI 0 9 2282

ins52 (pHM347) HaeIII 0 8 2416

ins55 (pHM1551) BstNI 0 9 2561

ins63 (pHM805) HaeIII 0 8 2945

ins78 (pHM1508) BstNI 0 9 3639

ins86 (pHM1536) BstNI 0 9 4002

ins80tsup(pWM106) Ncol 0 263 3762

ins91 (pHM1410) PstI 4 8 4258

ins96 (pHM2904) Ncol 0 8 4483

d13-23 (pHM1515) BstNI 915 9 144-1059

d13-49 (pHM1558) BstNI 2138 9 144-2282

d13-94 (pHM3902) BstNI-HinfI 4247 9 144-4391

d110-37 (pHM334) HaeIII 1251 8 486-1737

d111-94 (pHM4003) PstI-HinfI 3892 8 499-4391

d113-96 (pHM1008) Ncol 3859 8 625-4484

d122-25 (pHM1520) BstNI 113 9 1059-1172

d123-28 (DHM3241 HaeIII 236 8 1086-1322




TABLE 3-1 continued

Mutant+ SiteA Bp Deletedz Bp Insertedz

d125-55 (pHM1504) BstNI 1390 9

d127-41 (pHM502) Rsal 669 8

d148-52 (pLB202) Xhol 186 0

d149-55 (pHM1549) BstN1 279 9

d149-86 (pHM1561) BstNI 1720 9

d149-94 (pHM3305) BstN1, Hinfl 2108 8

d152-91 (pH11320) HaeeII, Pstl 1842 8

d158-77 (pHM383) HaeIII 904 8

d158-87 (pHM401) HaeIII 1351 8

d163-86 (pLB101) Apal 1097 0

d168-86 (pHM1503) BstNI 822 9

d172-78 (pHM1545) BstNI 252 9

d178-86 (pH.1156',l BstNI 363 9

d180-86 (pLB102) Apal 281 0

d180-96 (pLB314) NcoI 722 0

dl0-11 (pHM714) PstI 499 8

d10-42 (pHM702) PstI 1059 8

d10-91 (pHM707) Pstl 4258 8

dill-100 (pHM706) PstI 4176 8

d142-100 (pHM713) PstI 3616 8

d191-100 (pHM722) PstI 417 8
























TABLE 3-1 continued

+Each mutant was designated as an insertion (ins) or deletion (dl) at
a particular map position. Numbers in parentheses are laboratory
isolation numbers which are included in the mutant name to prevent
possible confusion between two mutants isolated at the same position.

ARestriction site indicates the initial restriction enzyme which was
used in the construction of the mutant.

EBp deleted and/or inserted were estimates based on the restriction
enzyme analysis of the mutant and the mode of construction. For
example, d123-28 (pHM324) was apparently the result of the deletion of
the HaeIII fragment containing nucleotides 1086 to 1322 followed by
the insertion of the 8 bp BglII linker fragment.

*Nucleotide numbers are those of Srivastava et al. (68) except for
mutants d180-86. The Apal site at map positTon 80 was not present in
the DNA sequence and its position was estimated by restriction enzyme
analysis (Labow and Berns, unpublished). The numbers given for each
mutant indicate the nucleotide numbers of the restriction site(s) used
for the construction of each mutant.

Phenotypes of AAV Mutants

Duplex DNA Single-stranded Virus Yield
Mutants Replicatorsz ProgenyE (%)A Phenotype

insi (pHM351) + + ND wt

ins11 (pHM1412) ND rep

insl3tsup (pWMI08) ND rep

ins23 (pHM1505) + (0.1-1.0) + 0.1 rep

ins27 (pHM322) -ND rep

ins32 (pHM326) -ND rep

ins42 (pHM1411) + (0.1) ND 0a rep

ins49 (pHM1523) + ND 0.1-1.0 lip

ins52 (pHM347) + + 0.1-1.0 lip

ins55 (pHM1551) + + 0.1-1.0 lip

ins63 (pHM805) + 0 cap

ins78 (pHM1508) + -ND cap

ins80tsup (pWM106) + ND cap

ins86 (pHM1536) + ND cap

ins91 (pHM1410) + ND 0 cap

ins96 (pHM2904) + + 100 wt

d13-23 (pHM1515) ND rep

d110-37 (pHM334) 0 rep

d123-28 (pHM324) ND rep

d148-52 (pLB202) + + 0.1-1.0 lip

d149-55 (PHM1549) + ND 0.1-1.0 lip

d149-94 (pHM3305) + 0 cap

d152-91 (pHM1320) + 0 cap

d158-87 (pHM401) + ND cap

TABLE 3-2 continued

Duplex DNA Single-stranded Virus Yield
Mutants Replicatorsz Progenyz (%)A Phenotype
d163-86 (pLB01) + ND cap

d180-86 (pLB1O2) + ND cap

d180-96 (pLB314) + (5) ND ND rep

EDuplex or single-stranded DNA replication was judged by the presence
of monomer single-stranded or duplex AAV DNA in Hirt supernatant DNA
following transfection into 293 or Hela cells with Ad2 as the helper
virus. See legends to Figures 2, 3, and 4 and Materials and Methods.
Numbers in parentheses indicate the yield of monomer duplex AAV DNA as
a percent of wild type levels (see Fig. 3).

AVirus yield refers to the yield of infectious virus from cells trans-
fected with mutant plasmids and harvested at 48 hrs post infection.
(See Materials and Methods for additional details.) The yield is
expressed as the per cent of the wild type plasmid (pSM620) yield.
+ND means the determination was not done; in the case of single-
stranded progeny DNA, the DNA isolation was not done under conditions
which inhibit reannealing.
aThe limit of detection was a virus yield 0.01% that of wild type.

TABLE 3-3(a)
Complementation of AAV Mutants

Complementing ins11 ins23 ins27 d110-37 ins32 ins42 ins52 ins63 ins78 ins86 ins91 d152-91/neo d149-94
genone (rep) (rep) (rep) Trep) (rep) (rep) lip) (cap) (cap) (cap) (cap) Tlip-cap) (lip-cap)

ins96/xs or
wild type + + +(b) +(b)

ins11 (rep) +

ins23 (rep) +

dl10-37 (rep) + +(b)

ins32 (rep)

ins42 (rep) +

ins52 (lip) + +

ins55 (lip) +

ins63 (cap) + + + + + +

ins78 (cap) + + +

ins86 (cap)

ins91 (cap)

(lip-cap) +

(a) The table indicates whether wild type or a mutant plasmid DNA indicated in the left column could
complement the defect in the mutants listed at the top of the table. When the partially viable rep
mutants ins 23 and ins 42 were used to complement other rep mutants, both genomes replicated at the
level of the partiaTTyviable mutant. However, no evidence was seen for either intergenic or
intragenic complementation.
(b) These experiments are discussed in Chapter Four.

Listing of DNA Construction Transfections into HB1Ols

pHM 100s Bg1II linker into HinclI sites of pSM620

pHM 200s Bg1II linker into hindIII site of pSM620

pHM 300s BgII linker into HaeIII sites of pSM620

pHM 400s Same as 300s

pHM 500s BgII linker into RsaI sites of pSM620

pHM 600s Bg1II linker into HincIl sites of pSM620

pHM 700s BglII linker into PstI sites of pSM620

pHM 800s BgIII linker into HaeIII sites of pSM620

pHM 900s B1II linker into BstNI sites of pSM620

pHM 1000s BglII linker into NcoI sites of pSM620

pHM 1300s pHM347 (ins55) and pHM 707 (dl0-91), BglII, EcoRV
digestion, ligation

pHM 1400s Bg1II linker into PstI sites of pSM620

pHM 1700s BgLII linker into NcoI sites of pSM620

pHM 2100s BamHI linker into PvuII site of PSV-2 dhfr

pHM 2300s Ligation of HindIII, BamHI double digestion of pBR neo
and pSV-2 dhfr

pHM 2400s BglII linker into HindIII site of pDR 540 (tac
promoter) (Russel and Bennett, 1982)

pHM 2500s BglII linker into EcoRI site of pDR 540 (tac promoter)

pHM 2600s BamHI linker into PvuII site of 2302

pHM 2700s BgII linkers into BstNI sites of pSM620

pHM 2900s BgII linker into Ncol sites of pSM620

pHM 3100s pBR322 derivatives with BamHI linker into PvuII site
with BamHI-PvuII deletion

pHM 3200s pBR322 derivatives with BillI linker into PvuII site

pHM 3300s d149-94 construction

TABLE 3-4 continued

pHM 3400s BglII linker insertion into BstNI sites of pHM704

pHM 3500s BgII linker insertion into Hinfl sites of pHM704

pHM 3700s pHM1523 BgII, EcoRV double digestion, ligation onto
PstI partial digestion, EcoRV digestion of pWM103

pHM 3800s pHM1515, pHM334, pHM3305 BgII, EcoRV double digestion,

pHM 3900s pHM1515, pHM3305 BgIII, EcoRV double digestion,

pHM 4000s pHM1412, pHM3305 BglII, EcoRV double digestion,

pHM 4100s Bal31 digestion of pHM351 with BglII linker insertion

pHM Is 1.4 kb Ld fragment into HindIII site of pHM2102

pHM 2s BgII linker into Xbal site of pHM1A

pHM 3s 2.1 kb Ld fragment into BglII site of pHM2A

pHM4s BamHI linker into PvuII site of pHM3Q

pHM 1320-neo BamHI digestion of pHM2609 and insertion into BgII site
of pHM1320

pHM 1320-dhfr BanHI digestion of 2102 and insertion into BglII site of

pHM 4003/Ld BamHI digestion of pHM4A and insertion into BgII site
of pHM4003

pHM 1320-neo-rep EcoRI digestion and religation of pHM1320-neo

pHM 1505-x pHM1505 was digested with BglII and ligated with a Sau3A
of bacteriophage

pHM 2904-x pHM2904 was digested with BgII and ligated with a Sau3A
digestion of bacteriophage

Figure 3-3. DNA replication phenotypes of AAV mutants.
One microgram of form I supercoiled plasmid DNA was transfected
into 293 cells in single plasmid assays. When two plasmids were
transfected together 10 ug of the replication defective mutants (insll,
ins42, d11O-37) were mixed with 1 ug of ins63 prior to transfection.
The celT were subsequently infected with Ad2 helper virus at a MOI of
5. At 36 to 48 hr, cells were lysed, treated with pronase, and low
molecular weight DNA was isolated by Hirt (1967) extraction. One tenth
of each DNA extract was then fractionated on a 1.4% agarose gel,
transferred to nitrocellulose by the method of Southern (1975) and
hybridized to nick-translated 32P-labeled AAV DNA. Where indicated
the extracts were digested with BglII prior to electrophoresis.
0152-91 appeared to replicate at a lower level in this experiment but
other experiments (not shown) indicated that its replication was
similar to that of other lip and cap mutants. Ins42 appeared not to
replicate in this experiment but was subsequently found to replicate at
very low levels; see Figure 3. The positions of the dimer duplex
replicative form (dd), monomer duplex replicative form (md), and
single-stranded form (ss) are indicated. Also, the sizes of the
fragments in the ins42 and ins63 experiment (lane 6) are indicated as a
percent of the full-length AAV size.


8 Ofl D (D
c -- c
. r C



0) cC
i cc

dd- -


- 63




Figure 3-4. DNA replication phenotypes of additional AAV mutants.
Procedures were the same as those for Figure 3-3 except ins55 and
ins78 where used as the replication positive complementor in place of
ins63. Aberrant bands seen in ins55 lane may be AAV DI DNA (natural
deletions) replicating. Aberrant bands seen in mock lane may be
spilled over from adjacent lane but no replicating monomer DNA is seen.


CO = to

o .- C) ) n () cn l
E ~ .E .E .E E


I *




Figure 3-5. Mutants that were partially defective for DNA
The procedures were the same as those in Fig. 2. Ten fold dilu-
tions of the wild type (pSM620) plasmid extract were electrophoresed in
order to quantitate the level of rescue and DNA replication of mutants
ins23 and ins42. The two bands which were observed in d110-37 trans-
fections represented fmII (nicked circular) and fmIII (linear) input
plasmid DNA.

o 'o M
0 00 r -
- X X 0




Figure 3-6. Complementation of capsid mutants by rep mutants.
Virus production was assayed by repeating the complementation
experiment as in Figure 2 with the following changes. At 48 hr post-
infection, plates were frozen and thawed three times and heated to 560C
for 30 min. One third of the virus stock was then used to infect a
single 10 cm plate of 293 cells and the plates were subsequently
infected with Ad2 at an MOI of 5. At 36 hr post-infection, DNA was
harvested and analyzed as in Figure 2. Each lane represents 10% of the
DNA recovered from one 10 cm dish. Wherever indicated the DNA extracts
were digested with B gII prior to electrophoresis. The sizes of the
fragments in the ins42 and ins 63 lane are indicated as a percent of
full length AAV dupTex DNA.



1+ n
C -) O

o -((
0 m 0) -0
E .CE C,

md 100

a 63

S 42

Figure 3-7. Infectious iirus DNA production by lip mutants.
Human 06 cells which had been transfected with mutant plasmids
or the wild type plasmid (pSM620) were harvested at 48 hr post-
transfection, frozen and thawed three times, and heated at 560 for
30 min. One third of the virus stocks were then used to infect 10 cm
dishes of 06 cells. At 36 hr post-infection, DNA was harvested from
the infected dishes and analyzed as in Figure 2. Mutant lanes repre-
sent 10% of the DNA recovered from one 10 cm dish (lanes 4,5,6). Lanes
1, 2, and 3 represent the indicated dilutions of 10% of the DNA
recovered in the wild type extract. Numbers at right of figure
indicate the position of full length monomer duplex (100%) and the
BglII restriction fragments in the ins55 lane as a percent of the
full-length genome.

Io q
0 0

+ -



- 5 55




As some of the rep mutants were capable of replication at low

levels it was speculated that there may be two complementation groups

located within the rep region. Therefore I mixed various combinations

of rep mutants and found no evidence for intra-rep region complementa-

tion. As shown in Table 3-3, none of the total and partial rep defec-

tive mutants could complement other mutants of the same phenotype.

From protein, transcript and DNA sequence analysis it was predic-

ted that VP-1, VP-2 and VP-3 all used much of the same coding sequences

(mu60-93). As the lip mutants also appeared to be capsid mutants it

was probable that lip and cap mutants could not complement each other

due to multiple use of the VP-3 reading frame. To determine if indeed

this was the case I attempted complementation between lip and cap

mutants. As shown in Table 3-3, these two phenotypes could not

complement one another. Thus, although I could identify two distinct

phenotypes among the mutants in the map unit 48 to 91 region, I could

not identify two different complementation groups. It was concluded

that lip mutants must share the VP-3 coding region (mu60-93) with the

cap mutants and that the lip-cap region (mu48-93) must be intact to

produce a normal yield of infectious virus.


Deletion and insertion mutants have been constructed in all of the

major open reading frames of the AAV genome. The mutants were con-

structed in an infectious AAV/pBR322 recombinant plasmid (Samulski et

al., 1982) and could be analyzed by transfection into Ad infected

cells. Using these mutants we have been able to identify three regions

displaying distinct phenotypes: 1) mutants located between mull and 42

are defective in their ability to rescue and replicate the AAV genome

from the recombinant plasmid (rep mutants), 2) mutants located between

mu63 and 91 are unable to produce infectious virus and can not produce

single-stranded DNA progeny but are capable of duplex DNA replication

(cap mutants), and 3) mutants located between mu48 and 55 are able to

rescue and replicate normally, producing both duplex and single

stranded DNA progeny, but can produce only low levels of infectious

virus particles when compared to wild type AAV (lip mutants).

Rep mutants

AAV is a virus which relies heavily on Ad helper and host cell

functions during productive infection. Because of this, it was uncer-

tain whether AAV encoded for any proteins required for its own DNA

replication. Because some kinds of defective interfering particles

(naturally occurring AAV deletion mutants) were unable to replicate in

the absence of wild-type AAV, it was suggested AAV may indeed code for

a rep function (Laughlin et al., 1979). The DI virus stocks used
consisted of a population of heterogeneous internal AAV deletions;

therefore, it was unclear which region of the AAV genome might code for

the rep function. Additionally, an AAV deletion mutant (d138-42) was

constructed in vitro and when transfected into Ad infected cells it was

found to be DNA replication negative (Cheung, dissertation), unless

complemented with wild-type AAV. As this deletion removed one of the

AAV splice junctions it was unclear, again, where the rep gene was
located. With the discovery that the left half of the AAV genome

contained a large open reading frame of no known function it was

suggested that this reading frame (or its smaller in-phase counterpart)

might code for a protein involved in replication (Srivastava et al.,

1983). The mutant analysis revealed that AAV does have at least one

gene required for DNA replication and that it is located in the left

half of the AAV genome between map units 11 and 42 (Figure 3-3).

A second question is not whether there is a rep gene, but how many

rep genes there are. There are, in fact, four transcripts (4.2 and 3.6

unspliced and 3.9 and 3.3 kb spliced RNAs) which are synthesized from

the left half of the genome (Green and Roeder, 1980; Green et al.,

1980; Carter et al., 1984; Lusby and Berns, 1982). This indicates that

there may be as many as four large proteins (each >50 kd) coming from

the rep region. Because the left side reading frames overlap exten-

sively and are in phase with one another it was impossible, using the

present construction techniques, to mutate the smaller open reading

frame (from the 3.6 and 3.3 kb transcripts) without also mutating the

larger open reading frame as well. All mutants between mu21 and 48

would inactivate both large and small reading frames. Therefore, even

though there is more than one protein being produced by the rep region

there is no possibility of identifying them through complementation

because of the presence of overlapping in phase open reading frame

phenomenon. In fact, my intra-rep region complementation attempts were

negative (Table 3-3); but this does not rule out the possibility that

there may be more than one rep gene. Also worthy of note is the fact

that there may be separate genes required for rescue and replication.

In order for mutants to be analyzed as being replication positive, both

rescue from the plasmid and then replication must take place. If the
larger reading frame encoded for the rescue protein then a mutant in

the 11-20 mu region would still be scored as rep even though it might

be able to replicate.

An additional question is whether the spliced or unspliced

versions of the rep genes are being used, or possibly is the small

reading frame entirely within the splice being used. Two mutants give

evidence that it may indeed be the unspliced reading frames which are

actually used. These are the partially viable rep mutants, ins42 and

the lip mutant d148-52. Ins42 is located in the intron (mu 41-48) and

was expected to be viable. As the replication of this mutant is

severely depressed (0.1%) this indicates that one of the three reading

frames present in the splice codes for a functional protein or alter-

natively that the mutation effects RNA splicing even though it is

located well within (50 bp) the intron.

The lip mutant d148-52 is missing the last four carboxy terminal

amino acids of the spliced versions of the left side reading frames and

adds on an additional 77 amino acids not normally present. In spite of

this large alteration in the spliced rep open reading frame d148-52 has

a normal rep+ phenotype. Therefore, this is added evidence that it

is not the spliced 3.9 and 3.3 kb RNAs that are needed for replication

but the unspliced 4.2 and 3.6 kb transcripts. There is, of course, the

possibility that the carboxy terminus of the rep proteins(s) is not

required for the rescue and replication of AAV.

Cap Mutants

The cap region maps between mu63 to 91 and evidence is strong that

this region codes for the major AAV capsid protein, VP-3 (Srivastava et

al., 1983; Jay et al., 1981). Because this protein comprises 85% of

the AAV capsid it is unlikely that functional AAV capsids are produced

by cap mutants. AAV replication is believed to be a strand displace-

ment mechanism from either terminus resulting in the production of

equal amounts of both plus and minus DNA strands (Berns and Hauswirth,

1979; Hauswirth and Berns, 1977; Lusby et al., 1981; Lusby et al.,

1980). The cap mutants appear to synthesize normal amounts of monomer

duplex DNA progeny but no single-stranded progeny; therefore it appears

that a pre-formed capsid is not directly involved in the strand dis-

placement synthesis. It would be unlikely that monomer duplex DNA

would accumulate at all in the cap mutants if this were the case. The

inability of the cap mutants to produce single stranded progeny is

probably due to the need of a pre-formed capsid in which to sequester

single-stranded progeny. If no capsid is present the DNA stays within

the replicating DNA pool and appears as duplex DNA.

Carter and his colleagues, during genetic studies of the helper

functions supplied by adenovirus to AAV, discovered that mutant

Ad5ts125, when helping AAV during coinfection at the nonpermissive

temperatures, produced no infectious AAV virus (Myers et al., 1980).

The mutation in Ad5tsl25 affects the adenovirus 72 kd DNA binding

protein which is required for adenovirus DNA replication (Van der Vliet

et al., 1975), and also affects the expression of other adenovirus

genes (Carter and Blanton, 1978; Klessig and Grodzicker, 1979; Nevins

and Winkler, 1980). It was found that the Ad5tsl25 did partially help

AAV. Normal amounts of duplex monomer DNA were produced, but synthesis

of AAV capsid proteins and the production of single-stranded DNA

progeny was severely depressed (Carter and Laughlin, 1984; Myers and

Carter, 1981; Jay et al., 1981). Thus Carter and his colleagues

anticipated the cap phenotype when they suggested that if capsid

proteins were not produced single-stranded progeny would not accumulate

either. I have isolated mutants in the major AAV capsid gene which

support and extend their observations. The knowledge that AAV codes

for a rep protein(s) also indicates that there is differential

regulation of the AAV gene, when taken with the Ad5tsl25 mutant data.

Expression of all of the AAV mRNAs requires expression of Ad early

region 1A or 1B or both. It is apparent from the ts125 observations

that the 72 kd DNA binding protein is needed for AAV capsid gene

expression but it is apparently not needed for duplex DNA replication.

Thus, differential regulation of the AAV genes exists.

It is also possible that VP-3 production (sequestering single-

stranded DNA) is not the sole mechanism by which control is exerted on

the replicating AAV DNA intermediate pool. This is indicated by the

fact that cap mutants produce approximately the same amount of duplex

DNA progeny in spite of the fact that no single-stranded progeny is

being withdrawn from the replicating pool. Therefore some other

regulation of AAV DNA replication probably is taking place.

Lip Mutants

Lip mutants do produce infectious virus (Figure 3-7), but the

level of infectious units produced is about 10-2 that of wild-type.

Therefore, the question is raised whether this phenotype is due to low

numbers of infectious virus particles or lower infectivity per virus

particle. I have not answered this question. Our finding that lip

mutants can not complement cap mutants suggests that the lip phenotype

may be due to a defective minor capsid protein. Supportive data by

Janik et al. (1984) indicate that one of the lip mutants, an indepen-

dently isolated d148-52, is actually a VP-1 mutant (the 85-90 kd minor

capsid protein). The VP-1 protein may be a prominent component

required in the adsorption and penetration of the virus into the host

cell. Additionally, Janik et al. (1984) found that the d148-52 muta-

tion did not effect the VP-2 or VP-3 proteins. Our data, taken with

the findings by Janik et al. indicate that additional AAV mRNAs not yet

observed may exist. The lip mutants map between mu48 and 55, which

suggests that the small open reading frame (ORF) located between mu47

and 55 is used as additional coding sequences for VP-1 (Srivastava,

1983). The known 2.3 kb mRNA, with a single well-mapped intron between

sequence 1906 and 2227 (mu41 and 48), does not contain a sufficiently

long ORF which could code for VP-1. An mRNA with an intron which

brings the mu47 to 55 ORF in phase with the extended ORF which codes

for VP-3 (mu60 to 92) may exist. An alternative explanation is that

the extended version of the mu60 to 93 ORF, which starts at mu48

without a start codon, may be brought in-phase with the mu47 to 55

reading frame by a reading frame shift during translation. A similar

phenomenon has been seen with the bacteriophage MS-2 lysis gene.

Although the lip phenotype appears to be due to an altered VP-1 product

an additional but unlikely explanation may be that the lip mutants are

in fact affecting a sequence in this region which is needed for

efficient packaging of the AAV genome.

In conclusion, our series of deletion and insertion mutations

affect all of the significant AAV open reading frames. However, AAV

has several overlapping reading frames, and our mutants can not affect


these individually. Thus I estimate that several additional AAV

phenotypes (and genes) remain to be identified.



The present state of knowledge of gene regulation and expression

has been attributable in great part to the ability to introduce foreign

genes into eukaryotic cells. The ability to introduce new genetic

material into cells also,allows for the possibility of treating genetic

disease by augmenting defective genes with functional ones. Viral

transducing vectors have become increasingly popular as a method for

transforming eukaryotic cells. The abilities of many of the viruses

used as vectors is described in Chapter One.

The Advantages of an AAV Vector

As stated earlier (see Chapter One) of this dissertation, AAV

shares many of the same advantages that retroviruses have for being

useful as a transducing vector. Briefly, AAV is a helper dependent

parvovirus which, like retroviruses, can exist as both a lytic virus

and as an integrated provirus. AAV requires coinfection with either

adenovirus or herpes simplex virus (its helpers) for lytic virus pro-

duction (Atchison et al., 1965; Buller et al., 1981; Carter and

Laughlin, 1984). When no helper virus is present AAV can integrate

into the host cells sequences and remain latent as a provirus (Cheung

et al., 1980; Berns et al., 1982). It was apparent froa earlier


studies that when AAV integrated it did so in a manner that left its

own sequences intact.

It was apparent that AAV had most of the advantages of retroviral

vectors but also had additional positive features. First, AAV is both

replication and transcription negative without the presence of a helper

virus, this being true for both AAV virus infection or proviral latent

infection (Carter et al., 1984; Berns et al., .1982; Berns et al.,

1975). This feature should allow for the study of transduced gene

regulation under the genes own enhancer and promoter elements, unlike


Second, AAV is not known to cause any disease in animals and man

(Cukor et al., 1984). In fact, it is known to inhibit the replication

and oncogenicity of adenovirus and the oncogenicity of herpes simplex

(Hoggan et al., 1966; Carter et al., 1979; Kirschstein et al., 1968;

Blacklow et al., 1978; Ostrove et al., 1981). Additionally, it has

been found that certain groups of cancer patients have no or lower

titers of anti-AAV antibodies than the general population (Mayor et

al., 1976). This non-pathogenic behavior makes AAV unique as an ideal

potential clinical vector from the standpoint of lack of viral disease

complications. The basis by which AAV inhibits its helper virus is


Third, AAV's potential as a cloning vector is enhanced by its

ability to be rescued and amplified by the superinfection of the latent

cell with adenovirus (Hoggan et al., 1972; Samulski et al., 1982;

Samulski et al., 1983). The word "vector" implies unidirectionality

but AAV can potentially be used as a bi-directional vector, for getting

DNA both into and out of the cell.

Fourth, AAV has an unusually broad host range for productive

replication. It has been found to replicate in cells of every

mammalian species tested when the appropriate helper virus was present

(e.g. mouse Ad with mouse cells, canine Ad with canine cells) (Casto et

al., 1967; Cukor et al., 1984). The study of AAV's host range of

integration has so far been limited to humans and African green monkeys

(Hoggan et al., 1972).

Strategies for Using AAV as a Vector

The initial work I did was to map the functional regions of AAV,

discussed in Chapter Three. This was done by mutating the internal

sequences of AAV in the infectious AAV/pBR322 recombinant plasmid

pSM620 (Samulski et al., 1982) and then transfecting the mutant

plasmids into Ad infected cells to observe their phenotype. The

existence of three phenotypically distinct regions in AAV's genome were

demonstrated. These are: the rep region, required for replication

(mu7 to 47); the cap region (mu60 to 92), coding for the main AAV

capsid protein; and the lip region (mu48 to 55), which appears to code

for part of at least one of AAV's two minor capsid proteins. Also,

very importantly, I learned that there was no absolute cis-required

packaging sequence located between null and 37 (Figure 3-6, Table 3-3),

nor between mu49 and 94, as both pHM334 (dlll-37) and pHM3305 (d149-94)

could be packaged into infectious capsids when complemented. Addition-

ally, there appeared to be no packaging sequence located in the region

between these two large deletions as pHM1411 (ins42) could also be

packaged. This information suggests that all sequences between mull

and 91 could be replaced by foreign DNA inserts. It is believed only

the AAV terminal repeats are required for integration (Cheung etal.,

1980; Berns et al., 1982) and packaging.

Although it appeared possible to replace the entire internal AAV

sequences we concluded that our initial strategy would be a smaller

deletion, limited to the lip-cap region. Leaving the recombinant

vector replication positive, would allow me two advantages for obtain-

ing data which I needed to evaluate AAV's utility as a transducing

vector. First, it would allow me to obtain an approximate titer of the

transducing virus stock. This can be done by comparing the amount of

replicating AAV recombinant DNA present from an infection of human

cells by an aliquot of the AAV recombinant virus stock against that of

a wild-type infection from an aliquot of AAV viral stock of known

titer. Secondly, because the AAV recombinant is replication positive

I could easily analyze presumably transformed cells by scoring for

rescue and replication of the AAV recombinant upon superinfection with


A Prototype Foreign Gene A Selectable Marker

The choice of a prototype foreign gene to insert in place of the

jl.-cap region in the recombinant AAV vector was easily made. Labora-
tories already using viral vectors provided us with considerable infor-

mation on what would be the most appropriate dominant selectable marker

to use. One of the main concerns was the size constraint afforded by

the limited deletion we could make in the lip-cap region of the viral

genome. I had two large deletion mutants which might be used: pHM1320

(d152-91) and pHM3305 (d149-93). Both of these mutants were replica-

tion positive and contained deletions of 1842 bp and 2108 bp

respectively. As I did not know the packaging limitations of the AAV

capsid I believed it necessary to limit the total recombinant genome

size to as close to wt as possible, 4.7 kb.

Several selectable markers were available for our use: 1) The

herpes simplex virus thymidine kinase (tk) gene was available, on

which much of the pioneer work of such selectable markers was done

(Shimotohno and Temin, 1981; Wei et al., 1981). This gene was too

large to insert (3.4 kb) into the lip-cap region withgut extended

construction, could only be used in tk- cell lines, and required

selection in HAT media. ,Additionally, I previously assayed an AAV/tk

recombinant (constructed by Stuart Kupfer) which contained the 3.4 kb

tk gene in the BamHI site (mu23) of AAV. This recombinant was found to

be defective for replication upon complementation. 2) The mammalian

dihydrofolate reductase (dhfr) gene was of small enough size but was,

like the tk gene, limited in usefulness as it could only be used in

dhfr- cell lines (Subramani et al., 1981). 3) The E. coli xanthine-

guanine phosphoribosyltransferase gene (xgprt) was of small enough size

but it caused rearrangements, concatemerization and possible chromoso-

mal integration of recombinant bovine papilloma virus vectors (Mulligan

and Berg, 1980; Law et al., 1982; Law et al., 1983) and also required

selection in HAT media. 4) The prokaryotic dhfr genes appeared to be

viable candidates for use as selectable markers in AAV but selection by

methotrexate was slow, requiring four weeks for colony formation, and

it was not clear if multiple copies were required for transformation
(O'Hare, 1980). 5) The prokaryotic Tn5 (Colbere-Garapin et al., 1981;

Southern and Berg, 1982) neomycin resistance gene was chosen as the

desirable selectable marker for the following reasons: a) it was small

enough (1.4 kb) to fit into the lip-cap deletion mutants along with the

SV40 early promoter (0.4 kb); b) its expression did not require an

intron; c) it had been shown that only one copy was required for trans-

formation; d) it had a simple selection scheme using the neomycin-like

antibiotic G-418; and finally, e) it had been proven a useful select-

able marker in other viral vectors such as SV40 (Southern and Berg,

1982), retroviruses (Joyner and Bernstein, 1983) and bovine papilloma

virus (Law et al., 1983).


Construction of AAV Recombinant Genomes

The strategy which I decided to follow was to replace to capsid

gene region (lip-cap) with the neomycin resistance gene and leave the

rep region intact. Leaving the rep region intact would allow me to

titer the recombinant stock by quantitating its replication during

infection as compared with that of a titered wild-type AAV stock. The

construction scheme for d152-91/neo (pHM1320-neo) is shown in Figure

4-1. I chose to put the neomycin resistance gene under SV40 early

region promoter (EPR) control as this promoter is very well character-

ized (Benoist and Chambon, 1981) and appears to work well in most cell

lines tested (Spandidos and Wilkie, 1983). The SV40 EPR sequences were

obtained from the plasmid pSV2-dhfr (Subramani et al., 1981) and the

neomycin resistance sequences were obtained from the plasmid pBR-neo

(Southern and Berg, 1982). These sequences were inserted into the

plasmid d152-91 (pHM1320), which contains a 1.84 kb lip-cap deletion,

in such a manner that transcription from the SV40-EPR would be in the

same direction as the transcription from AAVs own promoters located as

mu5, 19 and 40. Presumably the neomycin resistance gene would be

expressed from transcripts which originated in the SV40-EPR and termin-

ated at the AAV polyadenylation sequence located at mu95. Due to the

size of the lip-cap deletion and the size of the inserted sequences the

resulting AAV/neo recombinant was essentially the same size as the

wild-type AAV genome. This recombinant, although replication positive,

would be unable to produce infectious virus because it lacks the capsid

genes, Figure 4-1. A replication negative version of d152-91/neo was

also constructed by deleting the sequences between the two EcoRI sites

at mu 38 and 43 and resulting in d138-43/52-91-neo. This replication

negative AAV/neo recombinant would be useful for determining if the rep

region was required for integration.

Another recombinant was constructed containing the selectable

marker mouse dihydrofolate reductase resulting in d152-91/dhfr (Figures

4-1 and 4-2). This was done in a manner similar to the construction of

dl52-91/neo. pHM2102 was used as the source of the dhfr transcription

cassette. This gene would be expressed by transcripts originating in

the SV40 EPR and terminating at the SV40 polyadenylation sequences

located at the 3' end of the inserted SV40 sequences (Subramani et al.,

1981). This construct also contained an SV40 small T splice sequence

to insure proper processing and expression of the marker gene. Such a

sequence is not needed for expression of the neomycin resistance gene

(Colbere-Garapin et al., 1981). Because the size of the dhfr tran-

scription cassette was larger than the deleted AAV sequences, the

dl52-91/dhfr recombinant was approximately 300 bp or 6% larger than the

wild-type AAV genome, Figure 4-2.

These two recombinant plasmids, d152-91/neo and d152-91/dhfr, were

tested for their ability to replicate by transfection into Ad-2

infected human tissue culture cells. They were both found to replicate

at approximately the same level as the wild-type AAV genome (Figure

4-3). The Hirt DNA (Hirt, 1967) from the cells transfected with

d152-91/neo was digested with BglII, which does not cut wild-type AAV,

to confirm that the structure of neo recombinant was intact.

An additional vector I constructed contained a deleted version of

the murine class I histocompatibility gene H-2Ld (Evans et al.,

1982). The construction of this vector is described in Figure 4-2.

This recombinant has not been shown to be positive for replication upon

complementation and it is included here only for the sake of


Preparation of a Recombinant Virus Stock

As d152-91/neo lacked capsid genes it was necessary to provide

these functions in "trans" to obtain a recombinant viral stock. There-

fore d152-91/neo was co-transfected with wild-type AAV to provide these

missing functions. Unfortunately the resulting viral stock was found

to be almost entirely wild-type AAV virus. Only minute amounts of the

AAV/neo recombinant virus was evident (data not shown). The conclusion

from this result was that there was a strong packaging bias which

favored encapsidation of the wild-type genome over the recombinant.

To compensate for this packaging bias the approach I took was to

construct complementing AAV genomes which would be too large to be

packaged. This was done by inserting Sau3A fragments from bacterio-

phage X DNA into a nonessential region of AAV (R.J. Samulski and T.

Figure 4-1. Construction of AAV recombinants.
The diagram illustrates the intermediate steps in the construction of d152-91/neo (or G418r).
Arrows indicate the derivation of each plasmid as well as the restriction enzymes which were used at
each step. For example, pHM2302 was the product of a ligase reaction which contained the BamHI (B)
and HindIII (H) double digest fragments of pBR-neo (Southern and Berg, 1982) and pSV2-dhfr
(Subramani, 1981). Other enzyme designations are EcoRI (E) and PvuII (P). Stippled boxes are SV40
sequences, open boxes are mouse dhfr sequences and striped boxes are neo sequences. Solid lines and
small open boxes represent AAV coding and terminal sequences, respectively; dotted lines represent
pBR322 sequences. D152-91/dhfr was constructed by inserting the appropriate BamHI fragment from
pHM2102 into d152-91 (not shown).

I G-_418R
SO/ (Neo)

/ ----- E -E
SpSV2 B /
P,Bam \
p HI linkers

SBgl 1


dl 52-91
(pHM 1320) j
\ ,-/

Figure 4-2. Physical organization of AAV recombinant genomes.
Solid boxes indicate the position of the coding sequences for the rep, lip, and cap
functions of AAV. These were determined from the position of AAV mutants (Chapter Three).
D152-91/neo contains the SV40 origin and promoter sequences (SV40 mu 71-65, stippled box) and
the neonycin resistance gene sequences (striped box) inserted in d152-91. In the process of
constructing d152-91/neo the BgllII site the vector was lost (see Fig. 2). The sequences
between the EcoRI sites at mu 38 and 43 were removed to give dl 38-43, 52-91/neo. The inter-
ruption in the dl 10-37 line indicates the extent of deletion in this mutant. The SV40 origin
and promoter sequences were inserted in dl 10-37/EPR. The arrow indicates the single BglII
site (within the neor sequence) which is present in the construct. D152-91/dhfr was
constructed by inserting the SV40/dhfr sequences from the plasmid pSV2-dhfr (Subramani, 1981)
into d152-91 (see Fig. 2). Stippled boxes represent SV40 sequences, the open box represents
the mouse dhfr coding sequence, and the arrow indicates a unique BglII site. Ins96/X-R, M and
F were constructed by inserting a 0.5, 1.1 and 2.8 kb bacteriophage A fragment -pen box) into
the single BglI site of ins96 (see Methods). D111-94/Ld-J was constructed by inserting the
1.4 kb BamHI-XbaI fragmenint d the 2.1 kb BamHI-PvuII fragment; containing the Ld coding
sequences (Evins et al., 1982) together behind the SV40 EPR, in the pla-'ilid pHM4A. This Ld
transcription cassette with BamHI ends was then inserted into dill-94/L0-1. No positive
data was obtained on this construct and it is included here for completeness.

0 10 20 30 40 50 60 70 50 90 100

rep lip cop

dl52-91/neo 0
dl38-43,52-91/neo C--
dl52-91/dhfr 11
d110-37 0---
dlIO-37/EPR O -
ins96/X-R 0 W
ins96/X-M C0 ---"--as SwKMr -0
ins96/X-F 0- 'iCFF C
dl -94/L -J C ----

Figure 4-3. Comparison of wt and recombinant DNA replication.
One microgram of the inii.ated formal supercoiled plasmid DNA was
transfected into KB cell. The cells were subsequently infected with
Ad2 helper virus at an MOI of 5. At 24 hours, cells were lysed,
treated with pronase and low molecular wt DNA was isolated by Hirt
(1967) extraction. One tenth of each DNA extract was then fractionated
on a 1.4% agarose gel, transferred to nitrocellulose by the method of
Southern (1975) and hybridized to nick-translated d152-91/neo DNA.
Where indicated the extract was digested with BgII. The large BglII
fragment (3.1 kb) of d152-91/neo has approximately the same amount of
sequence homology to the probe as the full size band of wt AAV DNA
(4.7 kb). Bands which are larger than 4.7kb are normal concatemeric
replication intermediates which accumulate during AAV DNA replication
or input plasmid species. The identity of the d152-91/neo DNA was
confirmed by digestion with restriction enzymes other than BglII (not






4 .7

II 3.1

..- I.6

Shenk, personal communication) at mu96. This was done by inserting a

BglII linker into the XbaI site of the plasmid pSM620 XbaI, provided by

R.J. Samulski. This plasmid contains an Xbal linker inserted into the

Ncol site at mu96. The resulting plasmid, ins96 (pHM2904), was then
digested with BglII and Sau3A fragments from bacteriophage A were

ligated into this site. Three bacteriophage A inserts were chosen for

further study. These were ins96/A-R (pHM2904-X-R), ins96A-M (pHM2904-
A-M) and ins96/A-F (pHM2904-X-F) with inserts of 550, 1100 and 2800 bp

respectively. These were shown to replicate at levels approximately

10% of that of wild-type AAV, Figure 4-5. Additionally, it was found

that ins96-A/R could be packaged into virions to a significant extent,

and ins 96/X-M to a minute amount (Figure 4-5). A minute amount of a
deleted form of ins96/X-M was also seen to be packaged. These comple-

menting AAV plasmids contained all of the AAV coding regions intact and

could therefore provide all AAV products in trans. When d_52-91/neo or
d152-91/dhfr were co-transfected with ins96/A-M at a ratio of 10:1 (5

ug/0.5 ug) a recombinant AAV/neo or AAV/dhfr viral stock was obtained,

(Figure 4-4). Although neither of these stocks contained ins96/A-M

virus there was a significant amount of wild-type AAV virus present.
This wild-type genome presumably resulted from recombination between

the replicating AAV species. This recombination phenomenon, between
AAV mutants, had been observed during the genetic studies described in

Chapter Three and was therefore expected. The two other complementing

plasmids ins96/A-R and ins96/A-F could also be used in complementing
the AAV selectable marker recombinants to prepare a recombinant virus

stock. Stocks prepared in this way also contained wild-type AAV as

well as the recombinant. Stocks prepared with ins96/X-R also contained

virus with the encapsidated ins96/A-R genome (data not shown)

presumably because this genome was found to be packageable.

An additional type of complementing plasmid which consisted of rep

deletions of AAV were also used to prepare recombinant virus stocks.

These complementing rep deletion plasmids were d110-37 (pHM33-4) and

d110-37/EPR (pHM334-EPR). This latter plasmid was constructed by first

inserting a BgII linker into the HindIII site of pHM2102, to give

pHM2102-EPR. Then the BamHI-BglII fragment (approximately 400 bp),

containing the SV40 EPR from pHM2102-EPR, was ligated into d152-91 in

the orientation that would allow expression of the capsid genes from

the SV40 EPR. Both of these rep complementors, when co-transfected

with d152-91/neo at a 20:1 to 5:1 ratio, were found to result in a

dl52-91/neo recombinant virus stock (data not shown) at approximately

106 infectious units of d152-91/neo per ml. Again, these stocks

contained wild-type virus, presumably due to recombination, but did not

contain an equal amount of the complementing rep genome. Apparently

the packaging bias was even less favorable toward packaging rep


The d152-91/neo and dl52-91/dhfr viral stocks were titered by

comparing the yield of replicating recombinant AAV DNA resulting from

infection with these stocks of human tissue culture cells compared with

that of infections using a wild-type AAV viral stock of known titer,

Figure 4-4. The wild-type AAV stock was titered by immunofluorescent

focus formation using anti-AAV capsid antibody (Carter et al., 1979).

This indirect method of titering the recombinant stocks was the only

method available to us as AAV does not plaque by itself and no anti-

bodies are presently available which are directed against the left rep

Figure 4-4. Transducing virus titer determination.
Ad2-infected D6 cells (107/plate) were either infected with
known quantities of wt MV virus or 100 ul of one of three recombinant
virus stocks. At 48 hours DNA was harvested, digested with BgITT, an
analyzed as in Figure 3. Each lane represents 10% of the DNA reco:,reed
from one 10 cm dish. The lane marked mock contains DNA from uninfected
cells. D152-91/neo (KB) and d152-91/neo (293) were virus stocks grown
in KB and 293 cells respectiv-ey. The lanes marked AAV contain DNA
from cells which had been infected with the indicated number of PFUs as
determined by immunofluorescense. The probe was nick-translated
d152-91/neo plasmid DNA.

v gN
0 0 .

< C

't L
S0 00o 10 10

I 4.7

i 53.1


Figure 4-5. Replication and virus production assays of ins96-x
In both assays one microgram of form I supercoiled plasmid DNA was
transfected into D-6 cells which were subsequently infected with Ad-2
helper virus at a multiplicity of infection of 5. At 36 hours, in the
replication assay, were lysed and treated with pronase, and low-
molecular-weight DNA was isolated by Hirt (1967) extraction. In the
virus production assay, at 36 hours cells were frozen and thawed twice
and heated to 560C for 30 minutes. One-third of the virus stock
(media) was then used to infect a second plate of 0-6 cells which were
infected with Ad-2. At 36 hours post-infection DNA was isolated by
Hirt extraction (1967) and analyzed as in Figure 4-3. All lanes
represent 10% of the DNA recovered from one 10 cm dish.

ins 96/X ins 96/X virus
replication production

-5.2 kb
4.7 kb

I -- 5.2 kb
S -4.7 kb

region AAV products. The titer of the d152-91/neo and d152-91/dhfr

stock depended on the host cell used, Figure 4-4, and the ratio of the

complementor used. Stocks of d152-91/neo prepared with ins96/X-l, in KB

cells consistently contained approximately 106 infectious units/mi of

d152-91/neo virus and variable amounts of wild-type AAV, Figure 4-4.

Stocks of d152-91/neo prepared using d110-37/EPR as a complementor also

contained a similar recombinant titer (data not shown). Stocks made of

d152-91/dhfr contained a low titer of the recombinant virus, presumably

due to its larger than wild-type size. Virus stocks of d138-43,

52-91/neo (pHM1320-neo-rep) were also prepared using ins96/X-M as a

complementor. These stocks were untiterable due to the replication

defective nature of this recombinant.

Transduction of Human and Murine Tissue Culture Cells with d152-91/neo

102 to 105 human 06 cells were infected at various MOI

(0.1-1000) of the d152-91/neo virus and then selected with the anti-

biotic G-418. When selection was applied at 12 to 24 hours post-AAV/

neo infection the frequency of transduction of D6 cells was approxi-

mately 1% (Table 4-1). This transduction frequency was relatively

stable throughout a wide range of MOI. A 104 fold increase in MOI

produced only an 8 fold increase in transduction. 05 cells, a cell

line known to be latently infected with wild-type AAV (Cheung et al.,

1980), also could be transduced at similar levels. Two other cell

lines, human KB cells and mouse L tk- cells, were also transducable,

but at lower levels, Table 4-1.

Table 4-1
Transduction Efficiency of d152-91/neo Virus

Transducing Selection Transduction*
Cell Line Virus (MOI) Time (days) Efficiency (%)

D6 0.1 1 0.4 (.04)

06 1 1 0.6 (0.4)

06 100 1 1.3 (1.3)

D6 1000 1 3 (3)

06 1000 7 10 (10)

KB 1 1 0.1 (.07)

Ltk- 0.4 1 0.05 (.02)

*Transduction efficiency equals the number of G418 resistant colonies
divided by the number of infected cells. At low MOI, the number of
infected cells was calculated by using the Poisson distribution.
Numbers in parentheses indicate the number of transductants divided by
the number of cells seeded on the plate regardless of the MOI.

EThe indicated cells were seeded at 102, 103 or 105 cells/dish
and infected at the indicated MOI. G418 selection was applied at the
indicated time after infection (1 mg/ml for human cells or 0.4 mg/ml
for murine cells). Colonies were stained with Giemsa and counted at
days 8-14.

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E26LPJY92_SKEFUO INGEST_TIME 2013-02-14T17:21:00Z PACKAGE AA00012987_00001