Regulation of gene expression by adeno-associated virus


Material Information

Regulation of gene expression by adeno-associated virus
Physical Description:
162 leaves : ill. ; 29 cm.
Labow, Mark Aron, 1960-
Publication Date:


Subjects / Keywords:
Dependovirus   ( mesh )
Gene Expression Regulation   ( mesh )
Transcription, Genetic   ( 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, 1987.
Bibliography: leaves 147-161.
Statement of Responsibility:
by Mark Aron Labow.
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 - 000900674
oclc - 17562099
notis - AEK9473
sobekcm - AA00004840_00001
System ID:

Full Text







This dissertation is dedicated to my mother Audrey whose

support has made this possible and whose love has made this



I would like to thank Dr. Kenneth I. Berns for his

guidance, infinite patience, and love for biology (and

countless jokes and stories) which helped make several years

pass by very quickly.

I would like to thank the faculty of the University of

Florida who have given me a priceless education. In

particular I would like to thank William B. Gurley and

Nicholas Muzyczka for their advice and inspiration over the


I wish to thank my friends Randy Horowitz and Paul L.

Hermanot for their wonderful company, help and advice.

I wish to thank Patricia Burfeind for her excellent

technical assistance and most of all, for her friendship.

Finally, I wish to thank Gwen Tse Wong for her faith,

advice and love.



LIST OF TABLES ... . .. vi





The Parvoviruses .. 1
The Natural Occurrence of AAV 2
Latent Infection by AAV . 4
The AAV Helper Functions . 5
The Physical and Genetic Organization
of the AAV Genome .. 10
Molecular Genetics of AAV .. 16
Perspective . 19


Cells and Culture Conditions .. 21
Viral Stocks . 21
Preparation of Plasmid DNAs .. 21
Enzymes and Linkers . .. .22
Construction of Mutant AAV Genomes 22
Radiolabeling of DNA . .. 23
Diethylaminoethyl (DEAE)-Dextran DNA
Transfections . .. 24
Calcium Phosphate DNA Transfections 25
Genetic Selection of Mammalian Cells 26
Chloramphenicol Acetyltransferase
(cat) Assays . .. 26
RNA Extraction and Analysis .. 27
DNA Isolation . 29
Southern Blotting and Hybridization
of Filters . 30
Nuclear Runoff Transcription (NRT)
Analysis . 30
Immunizations and Antibody Analysis 31
Metabolic Labeling and
Immunoprecipitation Analysis 33






Introduction . .
Results . .
Discussion . .


Introduction . .
Results . .
Discussion . .


Introduction . .
Results . .
Discussion . .


Introduction .
Results .
Discussion .




S . 126
S . 127
S . 137

S . 146

S . 147

S . 162






Table Page



d152-91/neo . . 101






Figure Page

1-1. Structure and genetic organization of the
AAV genome . . 15

3-1. Accumulation of RNAs by two types of rep
mutants . . .. 42

3-2. Accumulation of RNAs by mutant AAV genomes .. 43

3-3. Accumulation of RNAs by ori- AAV genomes ... 48

3-4. Accumulation of RNAs by rep deletion mutant
genomes . .. .. 49

3-5. Transcription of AAV genomes in isolated nuclei. 53

3-6. Complementation of expression of rep mutants 58

3-7. Complementation of expression by left side
deletion mutants . .. 59

3-8. Accumulation of RNAs by left side deletion
mutants. . . .. .. 63

3-9. Phenotypes of left side deletion mutants 65

3-10. Structures of AAV genomes with duplications
of p40 . . .. .. 69

3-11. Expression of AAV genomes containing
duplications of p40 . .. 70

3-12. Stability of AAV RNAs. ... 71

3-13. Transcription of AAV in isolated KB cell nuclei. 76

3-14. Transcription of AAV in isolated HeLa cell
nuclei . . .. 78

4-1. AAV genomes inhibit genetic transformation 95

4-2. Southern blot analysis of G-418r cell clones 96


5-1. Immunoprecipitation analysis with antipeptide
antibody . . .

5-2. Immunoprecipitation analysis of AAV proteins

5-3. Identification of an AAV protein made early
during infection . .

5-4. Identification of phosphorylated capsid
proteins . . .

6-1. Construction of AAV/SV40 hybrid genomes .

6-2. Structure of hybrid AAV/SV40 genomes .

6-3. Expression of AAV plasmids in human cells

6-4. Replication of AAV plasmids in various
cell lines . . .

6-5. Replication of hybrid genomes in cos-7 cells

6-6. Inhibition of DNA replication by the AAV
rep gene . . ..



. 121

. 122

. 133

. 135

* 136

* 139

. 141

S. .142



AAV . adeno-associated virus

AAV-2 AAV type-2

Ad . adenovirus

BPV .. .. bovine papilloma virus

BSA ... bovine serum albumin

C- .. .. carboxyl

CT . calf thymus

cat . choloramphenicol acetyltransferase

cpm .. counts per minute

DEAE .. diethylaminoethyl

DMEM Dulbecco's modified Eagles medium

DNA .. ... deoxyribonucleic acid

EDTA ethelene diaminetetraacetic acid

FCS .. fetal calf serum

G .. .gravity

HAT .. ... Hypoxanthine-aminopterin-thymidine

HSV .. ... herpes simplex virus

kb . kilobase

kd . kilodaltons

KLH ... .Keyhole limpet hemocyanin

md . monomer duplex


mMT .

moi .

mRNA .

mu .

neo .



ori .






ss .

SV40 .



TRs .

... minimal essential medium

S. mouse metallothionein

S. multiplicity of infection

S. messenger RNA

* .. map units

S. neomycin phosphotransferase

S. nuclear runoff transcription

S. open reading frame

S. origin

S. phosphate buffered saline

.* .. ribonucleic acid

S. reticulocyte swelling buffer

... sodium dodecyl sulfate

S. sheep red blood cells

S. single stranded

S. Simian Virus 40

S. tumor-antigen

S. thin layer chromatography

.* .. Terminal Repeats

thymidine kinase

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



Mark Aron Labow

May, 1987

Chairman: Kenneth I. Berns
Major Department: Immunology and Medical Microbiology

The ability of adeno-associated virus (AAV) to affect

gene expression from both its own genome and heterologous

genes has been examined using several different assays. The

expression of both wild type and mutant AAV genomes was

examined in human cells. The effects of the AAV genome on

heterologous genes were examined in stable transformation

and transient gene expression assays. The properties of

hybrid AAV/simian virus 40 (SV40) genomes were also


Mutations of the open reading frame within the AAV p5

and p19 RNAs greatly inhibited both the accumulation of AAV

RNAs and AAV transcription in adenovirus infected human

cells. This defect in gene expression was complemented in

trans indicating that a gene product encoded within the p5

gene activates AAV transcription. Proteins encoded by the


p5 and p19 genes were also identified using antisera to

synthetic peptides

Deletion of certain sequences within the left side of

the AAV genome enhanced the accumulation of the p5 or p19

RNAs. The effect of such deletions occurred only in cis.

Analysis of transcription from the wild type AAV genome

suggests that the cis-active negative regulation occurs

after transcription initiation.

The AAV genome was also shown to inhibit the activity

of heterologous genes. Cotransfection of AAV genomes

inhibited both the transformation by several different

dominant selectable genes and transient expression by

pSV2cat. Both of the inhibitory activities were shown to be

mediated by the AAV p5 and p19 genes.

Finally, hybrid AAV genomes containing the SV40 early

promoter and origin of DNA replication were constructed.

The replication of the hybrid genomes, initiated by the SV40

origin, was inhibited by the p5 and p19 genes in cells which

constitutively produce the SV40 tumor-antigen.



The Parvoviruses

The parvoviruses are among the smallest of the

eukaryotic viruses. The family Parvoviridae consists of

three genera, the densoviruses, the autonomous parvoviruses

and the defective parvoviruses also known as the

dependoviruses (Siegl et al., 1985). The densoviruses

infect several species of insects. The autonomous

parvoviruses infect and cause disease in many vertebrate

species (Siegl, 1984). Although the dependoviruses also

infect many vertebrate species from birds to humans, they

are unique in that they require coinfection with a

genetically distinct helper virus in order to replicate

(Atchison et al., 1965; Melnick et al., 1965; Hoggan et al.,

1966; Blacklow et al., 1967a; Carter and Laughlin, 1984).

The dependoviruses were first discovered as contaminants of

purified Adenovirus (Ad) preparations and are thus referred

to as adeno-associated viruses (AAV). AAV type-2 (AAV-2) is

the most extensively characterized dependovirus and is the

focus of this dissertation.

The AAV virion is typical of all parvoviruses,

consisting of a naked icosohedral shell of approximately 20


nm in diameter. The capsid is composed of three virion

proteins, VP1, VP2, and VP3, with molecular weights of 87

kilodaltons (kd), 72 kd, and 63 kd, respectively (Salo and

Mayor, 1977; Johnson, 1984). The proteins are

immunologically cross reactive and contain extensive

overlapping amino-acid sequences (Johnson et al., 1971; Rose

et al., 1971; Johnson et al., 1975; Buller and Rose, 1978;

Lubeck et al., 1979). VP3 comprises approximately 90% of

total virion protein. Subspecies of VP1 and VP3 which differ

slightly in MW have also been reported (McPherson and Rose,

1983). The AAV virion is extremely stable, being resistant

to heat, ionic detergents, and certain proteases.

The capsid surrounds a linear single stranded (ss)

deoxyribonucleic acid (DNA) genome of approximately 5

kilobases (kb) (Mayor et al., 1969; Rose et al., 1969; Berns

and Rose, 1970; see Berns and Hauswirth, 1984, for review).

Although many autonomous parvoviruses preferentially package

the minus strand of DNA, AAV, as well as the autonomous

parvoviruses LuIII and B19, packages DNA molecules of both

polarities with equal frequency (Berns and Adler, 1972;

Majaniemi et al., 1981; Bates et al., 1984; Cotmore and

Tattersall, 1984).

The Natural Occurrence of AAV

The existence of AAV in nature is extremely widespread.

AAVs have been isolated from nearly all animal species which

have been specifically tested. Although natural hosts of


AAV appear to include humans, several species of monkeys,

cattle, horses, dogs, quails, and chickens, the broad

species specificity of many AAV prevents the actual

determination of species origin. While AAV-4 is only known

to replicate efficiently in vitro in primary African green

monkey cells, AAV-1, -2 and -3 appear to replicate in cells

from a wide range of species (Hoggan et al., 1966; Blacklow

et al., 1968b; see Cukor et al., 1984, for review). However

these AAV are somewhat species specific, in that they

efficiently replicate only in the presence of a helper virus

of the same species origin as that of the host cell. For

example human AAV-1, -2 and -3 will replicate in canine or

chicken cells as long as canine or chicken Ad are provided

as helper.

Infection by AAV-2 and 3 in humans is very common, as

is infection of captive monkeys by AAV-1, -2, and -3

(Blacklow et al., 1968a,b). A fifth primate AAV, AAV-5, has

been isolated from human genital tissue (Bantel-Shaal and

zur Hausen, 1984; Georg-Fries et al., 1984). An estimated

85% of the population of human adults are sero-positive for

AAV (Blacklow et al., 1967b, 1968a,b). Further, the mean

age of seroconversion for AAV appears to be between the ages

of 7 to 8 years of age. Although infection is very common,

no disease in humans or animals is associated experimentally

or epidemiologically with AAV. This benign nature of AAV


infection is consistent with the normal inability of AAV to

replicate in the absence of a helper virus.

Latent Infection by AAV

The defective nature of AAV replication raises the

question of how the virus retains its biological identity in

the absence of helper virus. In this regard, the AAV genome

appears to exist in a latent state as a silent provirus.

The latent state of AAV infection was originally discovered

during a program for screening for persistent viral

infections of primary cell culture lots to be used for

vaccine production. It was discovered that although no cell

lots were positive for AAV structural antigens, infection of

the cells with Ad resulted in the detection of AAV antigen

in 20% of the lots of African green monkey kidney cells and

1-2% of the lots of human embryonic kidney cells (Hoggan et

al., 1972).

Latent infection with AAV can be established

experimentally. Several studies have shown that infection

of human tissue culture cells with AAV results in a carrier

state from which no AAV virion DNA or capsid antigen is

detected but upon superinfection with Ad, AAV is produced

(Hoggan et al., 1972; Handa et al., 1977; Laughlin et al.,

1986). The process of establishing latent AAV infection is

extremely efficient. Up to one third of the cell clones

derived after infection produce AAV after superinfection


with Ad. The ability to rescue AAV is stable for greater

than 100 cell passages and up to one third of the individual

cells produce AAV antigen after superinfection with a helper

virus. The AAV genome exists in the latently infected

cells as an integrated provirus. Examination of the

integration complexes indicates that the AAV genomes are

often present as head to tail concatamers and that

integration occurs at random sites within the host DNA

(Berns et al., 1975; Cheung et al., 1980; Hermonat, 1984;

Tratschin et al., 1985; Laughlin et al., 1986). Further,

the AAV genome is integrated colinearly with the terminal

repeats (TRs) at the junctions of integration, suggesting a

role for the TRs in the integration process.

The AAV Helper Functions

Many viruses provide the full helper effect for AAV.

As described above, AAV will replicate in cells of varied

species origins using Ad of that same species. The helper

functions are also provided by herpes simplex virus (HSV) I

and II (Buller et al., 1981). In addition, the replication

of at least one AAV, AAV-5, has been reported to be

supported by vaccinia virus (Schlehofer et al., 1986).

While the plethora of genetically distinct helper viruses

might indicate that only a single common feature of the

helper viruses is needed for AAV replication, studies with

Ad indicate an interplay of many helper virus functions.


Studies using mutant Ad indicate that all the helper

functions are provided by Ad early regions. The first

region of importance is the El region. This region contains

two transcriptional units, Ela and Elb. The two genes

comprise the transforming region of Ad (Graham et al.,

1974). The normal functions of the two genes appear

complex. Ela encoded products normally function to activate

transcription of the other Ad early regions (Berk et al.,

1979; Jones and Shenk, 1979). The normal functions of the

Elb gene products are less clear but appear to involve the

ability to selectively transport viral ribonucleic acid

(RNA) to the cytoplasm while blocking the same process for

cellular messages (Babiss and Ginsberg, 1984; Babiss et al.,

1985; Pilder et al., 1986). Several proteins are produced

from alternatively processed Ela and Elb RNAs. When

studied in human KB culture cells, Ad mutants defective in

the production of either Ela or Elb products were defective

helpers for AAV-2, but provided significant helper effect

when the Ad mutants were used at very high multiplicities

(Ostrove and Berns, 1980; Janick et al., 1981; Laughlin et

al., 1982). However, mutants that were defective for the

production of both Ela and Elb products were completely

defective for helper function at all multiplicities.

The role of the Ela gene products as helper functions

can be explained by the ability of Ela proteins to stimulate

transcription. Transcription of AAV is readily detected in


cells transformed by the El region of Ad (Ostrove et al.,

1981). Specific induction of expression from the AAV p19

promoter by the Ela gene has also been reported (Tratschin

et al., 1984b). The role of the Elb gene is less clear. To

some extent the functions of Ela and Elb appear

interchangeable in that mutants in both El regions are more

defective for AAV helper functions than the individual

mutants. Also microinjection of helper virus genes into

AAV-2 infected Vero cells showed that in addition to the E2a

and E4 genes, helper effect, as measured by AAV capsid

protein synthesis, required both Ela and Elb genes when

injected in low copy number but either Ela or Elb alone when

injected in high copy number (Richardson and Westphal,


The E2a gene encodes the 72 kd ssDNA binding protein

(see Kelly, 1984, for review). This protein is required for

Ad DNA replication both in vivo and in vitro, negative

regulation of E4 transcription and regulation of the

stability of Ad early messenger RNA (mRNA) (Babich and

Nevins, 1981; Handa et al., 1983). The E2a gene product has

been reported to be involved in both the regulation of AAV

gene expression and AAV DNA replication. One group has

reported that although AAV DNA replication and production of

RNA were relatively normal, the accumulation of AAV

structural proteins and progeny ss-DNA were severely

inhibited when a temperature sensitive E2a mutant was used


as helper at the nonpermissive temperature (Myers et al.,

1980; Jay et al., 1981). However, another group reported

that the same mutant was as efficient a helper as wild type

Ad for AAV-2 replication but that the E2a gene product was

responsible for a species specific block in the translation

of AAV-2 capsid proteins, as well as certain late Ad

proteins, in monkey cells (Straus 1976a; McPherson et al.,


In addition, one group has reported that efficient

expression of the E2a gene in cos-1 cells supported AAV DNA

replication in the absence of other viral early genes (Quinn

and Kitchingman, 1986). However this study was done only in

cos-1 cells which produce the Simian Virus 40 (SV40) tumor-

antigen (T-antigen) which may contribute to AAV DNA

replication (see below). The production of AAV capsid

proteins is also enhanced by the Ad VA-RNAs (Janick et al.,

1981; Janick et al., 1982).

Finally, the E4 gene appears to be required for AAV DNA

replication (Carter et al., 1983). Microinjection studies

have shown that AAV DNA replication required the El, E2a,

and E4 genes (Richardson and Westphal, 1981). However,

microinjection of E4 RNA alone was sufficient for AAV DNA

replication, indicating that the function of the other Ad

regions might be indirectly needed to induce expression of

the E4 gene. However, the role of E4 as a helper gene has

not been examined in detail. Also, E4 mRNA has been shown


to act as a complete helper only in one cell line (Vero

cells). This is an important point in that Ad replication

is affected differently by E4 mutations in different cell

lines (Halbert et al., 1985)

The ability of divergent viruses and genes to support

AAV replication suggests that the helper functions act

indirectly by modulating activities of cellular genes. For

example, the Ela gene has been suggested to stimulate

transcription indirectly by enhancing the activity of a

cellular factor which binds Ad promoters (Feldman et al.,

1982; Kovesdi et al., 1986). Also VAI RNAs enhance

translation in Ad infected cells indirectly by preventing

the cell from inactivating a translation initiation factor

(Thimmappaya et al., 1982; Kitajewski et al., 1986). If all

the helper functions act indirectly, then one prediction is

that there would be a certain cell type or physiologic state

which would support AAV growth in the absence of helper


Several groups have reported helper virus independent

replication of AAV. Schlehofer (1983, 1986) originally

reported that certain human and Chinese hamster cell lines

transformed with SV40 become permissive for AAV-5

replication upon addition of a potent carcinogen or

cyclohexamide. Similar results were obtained for AAV-2.

Yakobson et al. (in press) have shown that several cell

lines will replicate AAV, supporting the production of


infectious progeny, when pretreated with potent cell

synchronizing agents. Interestingly, the cells most

efficient at replicating AAV were Chinese hamster cell lines

transformed with SV40. The SV40 genome appeared to be a

contributing factor for AAV replication since monkey CV1

cells did not support AAV replication although cos-1 cells

(SV40-transformed CV-1 cells) supported AAV replication.

However a Chinese hamster cell line that did not contain the

SV40 genome also replicated AAV although at a much reduced


It should be noted that under optimal conditions,

replication without helper virus was at least two logs lower

than that in the presence of helper virus. Further, only a

fraction of treated cells were able to replicate AAV in the

absence of helper virus. Thus AAV replication appears to

require a complex cell milieu which can be modulated by an

array of viral genes and physiologic states.

The Physical and Genetic Organization of the AAV Genome

The AAV genome has been extensively characterized and

completely sequenced (Lusby and Berns, 1982; Srivastava et

al., 1983). One outstanding feature of AAV DNA is the

presence of inverted TRs. The TRs are 145 bases long and

contain several internal palindromes (Gerry et al., 1973;

Kozcot et al., 1973; Fife et al., 1977; Lusby et al., 1980).

Thus AAV DNA can form several different secondary structures

including ss-circles held together with 145 base duplex


"panhandles" (Berns and Kelly 1974). The termini can also

fold upon themselves to form intrastrand T-shaped

structures. The TRs appear to act as the origin (ori) of

AAV DNA replication, apparently acting as primers for DNA

synthesis (Straus et al., 1976b; Hauswirth and Berns, 1977,

1979). Further the AAV TRs also contain a nicking site

allowing both the replication of 5' terminal sequences and

the resolution of replication intermediates. Thus AAV DNA

is replicated in a manner analogous to the model proposed by

Cavalier-Smith (1974) for the replication of linear

chromosomal DNA. The TRs may also be important in the

process of integration into and rescue out of foreign DNA by

the AAV genome.

Sequence analysis of the internal portion of the AAV

genome has revealed several large open reading frames (ORFs)

as illustrated in figure 1-1. The genome can be divided

into two domains, with long ORFs on either side. The two

largest ORFs are the major left and right side ORFs from 6-

47 map units (mu) (bases 320-2186) and from 60-92 mu (bases

2810-4321), respectively. The right side ORF predicts a

protein of size and composition corresponding to the

smallest capsid protein, VP3. Due to the overlapping nature

of the capsid proteins amino-acid sequences, the larger

virion proteins must also contain the body of this reading



All of the ORFs are present in different forms within

the AAV mRNAs. The known AAV RNAs have been characterized

from Ad-infected cells. The genome appears to encode three

sets of transcripts (see Carter et al., 1984, for review).

The RNAs are all overlapping, polyadenylated, and have

coterminal 3' ends at mp 96 (Laughlin et al., 1979b; Green

and Roeder, 1980a,b; Green et al., 1980; Lusby and Berns,

1982; Srivastava et al., 1983). However, three different

5'ends have been located at 5, 19, and 40 mu, giving rise to

unspliced RNAs of 4.2, 3.6, and 2.6 kb respectively. An

intron located between 40 and 48 mu has also been identified

and spliced RNAs of 3.9, 3.3, and 2.3 kb have been


The RNA start sites are each located downstream from

putative RNA polymerase II promoters, p5, p19, and p40 as

indicated by nucleotide sequence. The promoters contain

elements typical of eukaryotic RNA polymerase II promoters

including TATA boxes located approximately 30 bases upstream

of the RNA cap sites as well as CAAT boxes located

approximately 80 bases upstream from the p19 and p40

transcripts. The sequences including and surrounding the

TATA boxes of the p5 and p19 promoters are highly related to

that surrounding the Ela gene of Ad-5 (Lusby and Berns,

1982). Transcription of the AAV promoters is also

sensitive to low concentrations of a-amanitin as expected


for RNA polymerase II genes (Bloom and Rose, 1978; Jay et

al., 1978).

The p40 RNAs are predominantly spliced and comprise

about 90% of the total AAV RNA accumulated in cells

coinfected with AAV and Ad. The larger p5 and p19 RNAs are

accumulated in approximately equal amounts and exist as

predominantly unspliced molecules. All of the AAV RNAs are

found in both the nucleus and cytoplasm.

Several experiments show that the p40 messages encode

the AAV structural proteins. First, the capsid proteins are

still produced by mutant AAV genomes lacking the p5 and p19

promoters (Janik et al., 1984; Hermonat et al., 1984;

Tratschin et al., 1984a). Second, all three of the

structural proteins are translated in vitro from isolated

2.3 kb size AAV RNA (Jay et al., 1981). This observation is

paradoxical in that the sequence and structure of the

spliced p40 message could only account for the production of

the smallest capsid protein VP3. VP2 appears to be

synthesized using additional reading frame encoded upstream

from and in the same reading frame as the major left side

ORF, initiating translation at an upstream ACG codon

(Becerra et al., 1985). The mechanism by which the largest

structural protein, VP1, is synthesized is unclear but has

been suggested to involve the use of alternatively spliced

mRNAs (Hermonat et al., 1984). Several additional ORFs of

a) >*(1) Q)- a) ) ( a) a)

(1 4J. 4-)

44 a) 0) (a P4 V (D

040 4n-)

4-4 4j 0, 0 4-4
4J 0 a
01J0p 4a ()a -H r

fo M Q) 4 4J~
.H-41E-4 P

4j 4 r

'- 4 0 r4. 4-4 -i

X 4-)-H a 04
H(21 M )~~ 0 -H 4J

U ~ u 0 4 -
a) H, *aa ~ t a

(0 9 0~4 r-r.

ro a) rd )1r.W .
f) 4 0
4-) 4 U 0 Z" O M (0 : 4-

0 M Q) 44 W 0)

p ra ro pQ) 4 a)
4-j 4J 4-) *H ,
U) 44 V Q) 0)

a) a 4 H0 '4

(0~ ro 0
P I-q a) Mr- E 4
::I U) ro >I- a)> 1

rA4JU U J M(0'0C
r to '0 ri V, ) u ri :3
P4 0, r. 0 -1 Q) 0
'CTE- 0 r- -q 4 4 -)P 4, U


( rp) 4 *
rhrX rr cinL




unknown function overlap the carboxyl (C)-terminal regions

of both large ORFs.

Molecular Genetics of AAV

The defectiveness of AAV has suggested that all

functions required for AAV replication are provided by both

the host cell and the helper virus and that AAV simply

encodes its capsid proteins. However the existence of the

large left side ORF indicated that AAV might encode

nonstructural proteins which might be required for AAV

replication. This notion was supported by the observation

that AAV defective interfering particles, whose genomes

contain large heterogeneous internal deletions, required

coinfection with wild type AAV in order to replicate

(Laughlin et al., 1979a). However the identification of

specific AAV replication functions was hampered by the

requirement for a cytopathic helper virus; thus AAV could

not be plaqued nor could AAV mutants be selected.

This problem was circumvented by cloning the AAV genome

into bacterial plasmid vectors. The AAV genome was

originally cloned by GC-tailing of reannealed virion DNA and

insertion into pBR322 (Samulski et al., 1982). Clones

containing the intact AAV genome are biologically active

(Samulski et al., 1982; Laughlin et al., 1983). Upon

transfection into helper virus infected human tissue culture

cells, the cloned AAV genomes are rescued from the plasmid

sequences and AAV replication ensues. The process of rescue


and replication from the plasmid vectors is highly

efficient, approaching the yields of AAV DNA and virus

achieved by transfection of wild type virion DNA.

The cloning of the AAV genome in an infectious form

provides the ability to modify the genome in vitro and to

propagate large amounts of the mutant genomes in bacteria.

The effects of such modifications are then assayed by

transfection of the mutant genomes into Ad-infected culture

cells. This procedure was originally used to examine the

effects of deletions within the TRs on AAV DNA replication

(Samulski et al., 1983; Senapathy et al., 1984; Lefebvre et

al., 1984). Mutants which contained large deletions of both

TRs sequences failed to replicate their DNA. Replication of

these mutants could not be complemented in trans, confirming

the predicted role of the TRs as origins of AAV DNA

replication. Unexpectedly, mutants which contained only one

deleted TR were viable, producing AAV genomes with wild type

TR on both sides of the genome. Thus there appears to be

some mechanism for repairing deleted terminal DNA sequences.

A detailed genetic analysis of the internal AAV

sequences has been done in a similar manner. Deletion and

frameshift mutations have been created in cloned AAV

genomes. Deletion mutations were created by digestion of

cloned AAV genomes with various restriction endonucleases

(Hermonat et al., 1984; Tratschin et al., 1984a).

Frameshift insertion mutations have also been created by


insertion of synthetic nucleotide linkers at many places

throughout the genome. Analysis of the different mutants

revealed at least three distinct phenotypes and have again

divided the genome into two basic domains.

Mutations between 11 and 42 mu severely inhibit the

replication of AAV DNA. The defective replication of the

mutants, referred to as rep mutants, is complemented in

trans by cotransfected AAV genomes. Thus the ORFs contained

uniquely within the p5 and p19 transcripts encode a

protein(s) (rep proteins) absolutely required for AAV DNA

replication. Mutants containing a frameshift at 11 mu are

also defective for replication, indicating that a product

encoded by the p5 ORF is essential for DNA replication. No

mutation which specifically affects the p19 ORF has yet to

be created so the importance of this gene in AAV replication

has not yet been elucidated.

Mutations which affect the C-terminal region of the p5

and p19 ORFs have variable phenotypes. A mutation which

specifically affects the C-terminus of the p5 and p19 ORFs

contained within the spliced mRNAs has no apparent effect on

DNA replication, although a frameshift mutation at 42 mu

within the rep ORF contained within the AAV intron inhibits

AAV DNA replication by 100 fold. Thus potential rep

proteins encoded by the unspliced p5 and p19 ORFs appear to

be critical for AAV DNA replication. Conversely, certain

mutants which contain deletions of the entire intron and C-


region of the rep proteins efficiently replicate their DNA

while other similar mutants do not (Tratschin et al.,

1984b). As such it is unclear what specific functions) the

C-terminal coding regions of the rep ORFs have.

Although mutations within the right half of the genome

do not affect the accumulation of AAV replication

intermediates, these mutations do affect capsid protein

production and function. As expected deletion and

frameshift mutation within the body of the large right side

ORF (cap mutants) completely abolishes capsid production,

the sequestration of ss-DNA, and virus production. Also

mutations between 48 and 55 mu specifically affect the

syntheses of the larger structural proteins, VP1 and VP2 but

not VP3 (Janik et al., 1984). Thus VP3 appears to be a

primary translation product, rather than a proteolytic

cleavage product of the larger virion proteins.

Unexpectedly, mutants with lesions between 48-55 mu and

those with lesions in the body of the right side ORF have

different phenotypes. Mutants containing lesions between 48

and 55 mu (lip mutants) still produce sequestered progeny

ss-DNA but produce virions of low infectivity. Thus the

different virion proteins appear to have separable



The biology of AAV in many senses is similar to that of

latent prokaryotic viruses. Like the lambda bacteriophage,


AAV can efficiently integrate into and be rescued from the

host chromosome. Further similarities exist between AAV and

the defective bacteriophage P2 (Six and Klug, 1973; Calender

et al., 1981). Both exist as latent proviruses which

require coinfection with a specific helper virus in order to

replicate. One hallmark of the bacterial viruses is an

ability to modulate the expression of their genes. Indeed

the study of the "genetic switch" (Ptashne, 1986) between

the lytic and lysogenic states of lambda infection has

proved to be one of the most informative models of the

control of gene expression. We began these studies in the

hope that AAV might similarly provide a useful model of

eukaryotic gene regulation.


Cells and Culture Conditions

Human KB and HeLa monolayer and monkey cos-7 cell

cultures were maintained in Dulbecco's modified Eagles

medium (DMEM) containing 10% fetal calf serum (FCS), 2mM

glutamine, and 500 units/ml of penicillin and streptomycin.

Mouse B78H1 and Ltk- monolayer cultures were maintained in

either minimal essential media (MEM) or DMEM containing 5%

FCS, 5% calf serum, 2mM glutamine, 500 units/ml of

penicillin and streptomycin, 2.5 pg/ml Amphotericin B

(Fungizone, Gibco Laboratories), and 0.1 mg/ml Gentamycin

(Gibco Laboratories). In some experiments cells were

treated with 10 pg of actinomycin D (Sigma)/ml of media for

times indicated in the text.

Viral Stocks

AAV was prepared from lysates of HeLa cells coinfected

with both Ad-2 and AAV-2. Ad-2 was inactivated by heating

of the lysates at 560 for 30 minutes. Ad-2 stocks were also

prepared from lysates of infected HeLa cells.

Preparation of Plasmid DNAs

All plasmid DNAs were propagated in Escherichia coli

HB101 cells. After inoculation of fresh cultures with 1/50

volume of an overnight culture, plasmid containing cells



were grown to a density of 80 Klett units (green filter,

Klett-Summerson colorimeter) in Luria broth (1% tryptone,

0.5% yeast extract, 0.5% NaCl, and 0.5% glucose).

Chloramphenicol was then added to a final concentration of

170 pg/ml and the cultures were grown for an additional 16-

20 hours. Plasmid DNAs were isolated as described by

Birnboim and Doly (1979). Form I plasmid DNA was further

purified by centrifugation in cesium-ethidium bromide

equilibrium gradients as described by Maniatis et al.


Enzymes and Linkers

Restriction enzymes, T4 DNA ligase, T4 DNA polymerase,

T4 polynucleotide kinase, Escherichia coli DNA polymerase I

and Klenow fragment were purchased from either New England

Biolabs, Bethesda Research Laboratories, or International

Biotechnologies Inc. and used as specified by the

manufacturer. Linkers were purchased from either Pharmacia

or New England Biolabs and were phosphorylated as specified

by the manufacturer.

Construction of Mutant AAV Genomes

Plasmid constructions are summarized in chapter III

(table 3-1). Form I pSM620 (Samulski et al., 1982) was

digested with various enzymes (ApaI, NcoI, or XhoI) and

religated to yield various deletion mutants. Certain

deletion mutants were constructed by lighting previously

described plamsids, which had been digested with BglII and


EcoRV. Mutants containing deletions of the TRs (ori-

mutants) were created by insertion of the large SmaI or Ball

fragment of particular genomes into the EcoRV site of pBR322

(Bolivar et al., 1977), pAT153 (Twigg and Sherratt, 1980),

or pFB69 (Barany, 1985). The construction of pLB1701 and

derivatives is described in chapter VI.

All plasmid constructions were placed into and

propagated in HB101. Initially, small amounts of plasmid

DNAs were isolated for structural analysis by the method of

Holmes and Quigley (1981) or Birnboim and Doly (1979).

Plasmid constructs were analyzed by digestion with

appropriate restriction endonucleases and electrophoresis in

agarose and acrylamide gels.

Radiolabeling of DNA

Radiolabeled DNAs to be used for probe were produced

either by nick translation (Rigby et al., 1977) or by the

random primer technique (Feinberg and Vogelstein, 1983) and

were used interchangeably. Unless stated otherwise in the

text, duplex AAV-virion DNA was labeled as probe. Nick

translation reactions contained 1 pg of DNA, 50 mM Tris-HC1

(pH 7.8), 5 mM MgCl, 10 mM 2-mercaptoethanol, 0.1 mM each of

dGTP, dATP, and TTP, 100 uCi of a32P-dCTP, 1 ng DNAase I

(Sigma), and 5 units of DNA polymerase I in a final reaction

volume of 50 pl. Unincorporated deoxyribonucleotides were

removed by chromatography on Sephadex G-75 (Pharmacia Fine

Chemicals) in 10 mM Tris-HCl (pH 8.0). The specific


activity of probe obtained by this procedure was between 2

to 5 x 107 counts per minute (cpm)/pg DNA. Random primer

reactions consisted of 0.1 mg of denatured DNA (boiled for 7

minutes), 0.2 M Hepes (pH 6.6), 0.15 optical density units

of random hexadeoxyribonucleotides (P-L No. 2166), 50 mM

Tris-HCl (pH 8.0), 50 mM MgCl, 10 mM 2-mercaptoethanol, 100

AM each of dATP, dGTP, and TTP, 100 pCi of a32P-dCTP (ICN),

0.2 pg/ml bovine serum albumin (BSA), and 5 units Klenow

fragment. Reactions were incubated at room temperature for

2 hours and stopped by the addition of sodium dodecyl

sulfate (SDS) and ethelene diaminetetraacetic acid (EDTA) to

0.2% and 50 mM, respectively. At this point reactions had

incorporated between 90 to 99% of the label and probes were

used without further purification.

Diethylaminoethvl (DEAE)-Dextran DNA Transfections

The DEAE-dextran transfection procedure was carried out

as dsecribed by McCutchan and Pagano (1968) with some

modifications. Either 100 mm or 150 mm dishes of

approximately 80% confluent KB cells were washed once with

phosphate buffered saline (PBS) and a DNA-dextran mixture

was added and left on the cells at room temperature for 30

minutes. DNA-dextran consisted of variable amounts of each

plasmid as indicated and 500 pg of DEAE-dextran (MW =1 x 103

kd, Pharmacia Fine Chemicals) per ml in MEM without serum.

Plates of cells, either 100 mm or 150 mm diameter, received

either 1 ml or 2.5 ml of the DNA-dextran mixture


respectively. The transfection mixture was then replaced

with fresh media containing 50 pM chloroquine diphosphate.

Experiments involving infection with Ad-2 contained Ad-2 at

a multiplicity of infection (moi) of 10 in the chloroquine

containing media. The cells were incubated for 3 hours at

370C and refed with fresh media.

Calcium Phosphate DNA Transfections

Calcium phosphate coprecipitation transfections were

carried out as described by Graham and van der Eb (1973) and

as modified by Wigler et al. (1979). For stable

transformation and genetic selection, plates were seeded the

night before transfection at densities of 1.25 x 105 cells

or 3 x 105 cells per dish of B78H1 or Ltk- cells,

respectively, while 30-50% confluent plates were used for

transient expression assays. Precipitates were formed by

combining the following in the following order: DNA plasmidd

and carrier DNA for a total of 25 pg of DNA) and H20 to

0.250 ml, 0.250 ml 0.5 M CaC1, and 0.5 ml of 2 x Hepes (42

mM Hepes, 0.27 M NaCl, 10 mM KC1, 1.4 mM Na2HPO4(2H20), pH

7.05). The precipitates were allowed to form at room

temperature for 20 minutes and then added directly to the

media of 100 mm plates of cells. For stable transformation,

the media and precipitate were replaced with fresh media the

following day and again with selective media (see below) the

next day. For transient expression assays, the precipitate

and media were removed at 4 to 6 hours after transfection


and replaced with 5 ml of 12% glycerol in PBS, incubated for

1 minute, and finally replaced with 10 ml of fresh media

containing 4 mM sodium butyrate as described by Gorman et

al. (1983).

Genetic Selection of Mammalian Cells

G-418 resistant (G-418r) cells were selected in media

containing 1 mg/ml G-418 Sulfate (Geneticin, Gibco) as

described by Southern and Berg (1982). Thymidine kinase

(tk) positive cells were selected in hypoxanthine-

aminopterin-thymidine (HAT) media as described by Wigler et

al. (1979) and consisted of 15 pg/ml hypoxanthine, 1 j/g/ml

aminopterin, and 5 Mg/ml thymidine in DMEM. Colonies were

grown for 10-18 days in selective media, after which the

plates were fixed with 10% formaldehyde and stained with

cresyl violet.

Chloramphenicol Acetyltransferase (cat) Assays

Assays for cat activity were carried out essentially as

described by Gorman et al. (1982). Plates of transfected

cells were washed three times with cold PBS, incubated under

1 ml of STE (40mM Tris-HCL (pH7.4), 0.15 M NaCl, 1 mM EDTA)

for 5 minute at room temperature. The cells were harvested

by scraping, pelleted by centrifugation, and resuspended in

110 il of 0.25 M Tris-HCl (pH 7.8) and frozen and thawed

three times. Cell debris was removed by centrifugation and

the supernatant stored at -200C. Half of the supernatants

were used in the assays which consisted of 55 pl extract, 75


Jl 1 M Tris-HCl (pH7.8), 25 pl Acetyl Coenzyme A (4 mg/ml,

made fresh each time), and 0.1-0.25 pC 14C-chloramphenicol

(New England Nuclear Inc.). The reactions were incubated at

37C for 1 hour and extracted with 1 ml of ethyl acetate.

The organic layer was isolated and dried in a Savant speed

vacuum centrifuge for 30 minutes. The desiccated samples

were resuspended in 25 pl of ethyl acetate and spotted on

thin layer chromatography (TLC) plates (Kodak silica or

Bakerflex silica plates), and eluted in a TLC chamber

equilibrated with chloroform and methanol (95:5). Cat

activity was determined by autoradiography of the TLC plates

followed by scintillation counting of spots cut from the


RNA Extraction and Analysis

At 33 to 36 hours post transfection, cells were

harvested by scraping, washed with cold PBS, and swelled on

ice in reticulocyte swelling buffer (RSB, 10 mM Tris-HC1 (pH

7.4), 10 mM NaCl, 5 mM MgCl and 2 mM DTT) for ,10 minutes.

The cells were lysed with 12 strokes of a Dounce homogenizer

and the nuclei removed by centrifugation. RNA was then

extracted from the cytoplasmic homogenate by a modification

of the procedure described by Lusby and Berns (1982).

Briefly the cytoplasmic extract was added to 4 ml of ETS (10

mM EDTA, 0.,2% SDS, and 10 mM Tris-HCl (pH 7.4), and 5 ml of

phenol heated to 70C, and placed in a 700C water bath for

15 minutes with occasional shaking. After quick chilling on


ice, the aqueous phase was isolated, reextracted with 4 ml

of phenol/chloroform (1:1) and the RNA precipitated with


For some experiments total cellular RNA was isolated

using the guanidinium/CsCl method as described by Maniatis

et al. (1982). Cell pellets were lysed by addition of 2.5

ml of a solution containing 4 M guanidinium isothiocyanate,

5 mM sodium citrate (pH7.0), 0.1 M 8-mercaptoethanol and

0.5% sarkosyl. One-gram of CsCl was then added and the

solution layered onto a 1.2 ml cushion of 5.7 M CsC1

containing 0.1 M EDTA (pH7.5) in Beckman SW 50.1 tubes

followed by centrifugation at 35,000 rpm for 12-16 hours at

200C. The supernatants were discarded and the pellets

resuspended in 400 pl of a solution containing 10 mM Tris-

HC1 (pH7.4), 5 mM EDTA, and 1% SDS. This solution was then

extracted twice with chloroform and 1-butanol (4:1) and,

after addition of 0.1 volume of 3 M sodium acetate (pH 4.8),

precipitated with ethanol. Concentrations were determined

by A260 and by intensity of ribosomal RNA bands after

shadowing with UV light on cellulose TLC (Analtech, catalog

number 08011). RNA samples were prepared and fractionated

in 1.1% agarose gels containing 2.2 M formaldehyde as

described by Maniatis et al. (1982) using vertical slab gels

run for 5 to 8 hours at 100 volts. RNAs were transferred

onto nitrocellulose filters in 20 x SSC without further


treatment of the gels. The filters were then handled as

described below.

DNA Isolation

Low molecular weight DNAs were isolated by a

modification of the method of Hirt (1967). Cell monolayers

were washed in cold PBS and harvested by scraping in 1 ml of

Hirt buffer (0.6% SDS and 50 mM EDTA, pH7.4). Sodium

chloride was added to 1 M after incubation for 20 minutes at

room temperature and the samples incubated overnight at 40C.

The samples were then centrifuged for 45 minutes in an

Eppendorf centrifuge, the supernatants were collected,

extracted 2-3 times with an equal volume of

phenol:chloroform (1:1), and the nucleic acid precipitated

with ethanol. In some experiments this procedure was used

to isolate DNA directly from isolated nuclei.

Genomic DNAs were isolated by a modification of a

procedure reported by Cheung et al. (1980). Nuclei isolated

as described above were resuspended in 1 ml of 1 x SSC and

pronase and SDS were added to concentrations of 1 mg/ml and

0.1%, respectively, followed by incubation overnight at

370C. The solution was then gently extracted twice with an

equal volume of phenol:chloroform (1:1) and precipitated

with 2.5 volumes of ethanol. Nucleic acid was then gently

resuspended in 10 mM Tris-HCl (pH7.4) and incubated with 1

mg of RNase A (Sigma) for 60 minutes at 370C, followed by

one extraction with phenol:chloroform and ethanol


precipitation. DNAs were concentrations were determined by


Southern Blotting and Hybridization of Filters

DNA gels (nondenaturing) were prepared and blotted to

nitrocellulose as described by Southern (1975). After

transfer, the filters were baked at 800C for two hours,

placed in a heat sealable bag with 15 ml of prehybridization

buffer (50% formamide, 5 x SSC, 0.5% SDS, 0.1% PVP, 0.1%

BSA, 0.1% Ficoll, 0.15 mg/ml sheared calf thymus (CT) DNA)

and incubated at 420C for 4 hours, and the buffer replaced

with fresh buffer containing denatured DNA probe and CT DNA,

and then incubated at 420C for 12 to 16 hours. The filters

were then washed twice for 15 minutes in 2 x SSC and 0.5%

SDS at 680C, twice for 60 minutes in 0.1 x SSC and 0.1% SDS

at 680C and once at room temperature for 20 minutes in 0.1 x

SSC and 0.5% SDS. Filters were dried and placed against

film for autoradiography.

Nuclear Runoff Transcription (NRT) Assays

Cells were harvested at 24 hours postinfection and

nuclei were obtained as described above except that DTT was

added to 2 mM to the RSB and KCl added to 140 mM immediately

after homogenization. Nuclei were isolated by centrifugation

and prepared and analyzed as described (Falk-Pedersen et

al., 1985). Nuclei from 3 x 107 cells were resuspended 1 ml

NRT buffer (20 mM hepes (pH 7.9), 20% glycerol, 140 mM KCl,

10 mM MgCl, 2 mM DTT), repelleted at 2,000 times gravity (G)

for 4 minutes, and resuspended in 200 4l of transcription

cocktail (60 4l 32P-UTP (200 pC), 1 1l of a solution

containing 1 mM each of GTP, ATP, and CTP, 2 mM DTT, 80 pl

NRT buffer), incubated on ice for 3 minutes then incubated

at 30C for 15 minutes. Reactions were terminated by adding

1 ml of HSB (0.01M Tris-HCl (pH 7.4) 0.5 M NaCI, 0.05 M

MgCl, 2 mM CaC1) and 0.1 mg DNAase I and then incubated at

room temperature for 1 minute followed by the addition of 4

ml of ETS and 5 ml of 700C phenol. Labeled RNA was isolated

as described for cytoplasmic RNA. RNA was further purified

by precipitation in 2 M LiCl at 0C overnight. The amount

of incorporated label was determined by chromatography on

PEI cellulose TLC using 1 M HC1. Equal amounts of labeled

RNA for each experiment (approximately 1x107 or 5x107 acid

precipitable cpm for KB cells and HeLa cells, respectively)

were used as probes of southern blots containing restriction

digested AAV-2 and Ad-2 DNAs. Blots were hybridized and

washed as described above except that an additional wash was

carried out in 0.1 x SSC, 0.5% SDS, 10 mM NaPO4, 1% Triton

X-100, and 1% Tween-20 for 1 hour at 680C.

Immunizations and Antibody Analysis

Synthetic peptides were prepared by Dr. Ben Dunn as

described by Dunn et al. (1983). Peptides contained 3H-

amino acids so all steps could be quantitated. Peptides

were coupled to either BSA or keyhole limpet hemocyanin

(KLH) with glutaraldehyde as described by Kagan and Glick


(1979). Glutaraldehyde (1 ml, 20 mM) was added dropwise to

a 2 ml solution containing 10-20 mg of BSA or KLH and a 30

molar excess of peptide in 0.1 M sodium phosphate (pH 7.5)

and incubated at room temperature with stirring for 2 hours.

Conjugated peptide was purified by chromatography on

Sephadex G-25 followed by dialysis against PBS for 12-16

hours with several changes of buffer. Conjugates contained

an approximate carrier to peptide molar ratio of 1:15 to


Two-month-old New Zealand white rabbits were immunized

with conjugates containing a total of 300 jg of peptide

emulsified in 2 ml of Freund's complete adjuvant by

intradermal injections at multiple sites across the back.

Animals were boosted with subcutaneous injections containing

200 pg of peptide/conjugate in 2 ml of Freund's incomplete

adjuvant at approximate 4 week intervals. Antisera were

drawn approximately 7 days after injections. Antisera were

further purified by chromatography on peptide columns

created by coupling of peptide to CnBr-activated Sepharose

4B prepared as specified by the manufacturer (Pharmacia Fine


The presence of antipeptide antibody was determined by

hemagglutination assays using peptide coupled to sheep red

blood cells (SRBCs). SRBCs (0.2 ml, pelleted) were washed

three times with 2 ml of 0.1 M NaCl, pelleted by

centrifugation at 1000 x G for 5 minutes each time and then


resuspended in cold PBS (pH 8.2) containing 1%

glutaraldehyde. The cells were rotated gently at 4C for 30

minutes, washed 5 times in cold 0.15M NaCl, 5 times in cold

water, diluted to 2.5% in PBS (pH 7.2) and mixed with an

equal volume of a 1:60,000 solution of tannic acid in PBS

(pH 7.2). The cells were gently rotated at 40C for 20

minutes, washed twice with PBS (pH 7.2), resuspended as a

2.5% suspension in PBS (pH 6.4), incubated with 400 ug of

peptide at 37C for 2 hours, washed twice with 1% normal

rabbit serum in PBS (pH 7.2), resuspended in the same

solution as a 2.5% suspension and frozen by emulsion in

liquid nitrogen in small aliquots and stored at -700C.

Hemagglutination assays were carried out as described.

All sera tested were preincubated at 370C to inactivate

complement. Reactions contained 25 ul of 0.5% coupled SRBCs

(diluted in 1% agammaglobulin horse sera) and 50 il of

diluted rabbit antisera. Reactions were incubated at room

temperature for 4 hours in 96 well round bottom microtiter


Metabolic Labeling and Immunoprecipitation Analysis

KB cells (approximately 1 x 106 cells) were infected

with either Ad-2 (moi of 10) or coinfected with Ad-2 and

AAV-2 (moi of 20), washed twice with methionine-free or

phosphate free MEM, and then incubated for 1 hour in 0.5 ml

of the same media containing 2% dialyzed FCS and either 250

pC 35S-methionine or 500 pCi 32P-phosphate. Labeled cells


were then washed twice in cold PBS on the plates and lysed

by scraping in 2.5 ml PLB (phospholysis buffer; 10mM sodium

phosphate (pH7.5) 100 mM NaCl, 1% Triton X-100, 0.5% sodium

desoxycholate, 0.1% SDS) containing 1 mM PMSF (Sigma). Cell

debris were removed by centrifugation at 1000 xG for 10

minutes and the lysates cleared by centrifugation at 100,000

x G for 1 hour at 4C. Samples were then diluted two fold

with PLB containing 10 mg/ml BSA.

Equal amounts of labeled proteins (approximately 5 x

107 cpm acid precipitable counts) were incubated overnight

on ice with 5 1p of antibody. Protein-A sepharose (Pharmacia

Fine Chemicals) was preincubated in PLB and BSA overnight at

40C with rocking, added (100 il of a 10% mixture) and

reactions rocked gently at 4C for 45 minutes. Bound immune

complexes were then pelleted and washed 8 to 12 times in PLB

with BSA and twice in PLB, resuspended in SDS sample buffer

and boiled to elute antibody and antigen.

Samples were then fractionated on discontinuous

polyacrylamide gels as described by Laemmli (1970). Gels

consisted of a 4% stacking gel and 10% running gel and

samples were electrophoresed for 4 to 6 hours at 125 volts.

Samples were run next to prestained protein MW markers

(Bethesda Research Laboratories). Proteins were then fixed

in 50% methanol, 10% acetic acid and 1% commasie brilliant

blue for 20 minutes followed by destaining overnight in 10%

acetic acid. Gels were photographed, rinsed with water for


30 minutes and treated with 1 M sodium salicylate for 45

minutes (for 35S- labeled proteins only) before autoradiography.



As described in chapter I, all aspects of AAV

macromolecular synthesis are regulated by various helper

functions. However, these studies also show a great

plasticity in the requirement for specific helper virus

genes. Although AAV gene expression is positively regulated

by the Ela, Elb, E2a, and VA genes, none of these genes

appears absolutely essential for AAV gene expression. The

recent observation that AAV can replicate in certain cells

in the absence of helper functions supports the idea that no

helper virus genes are directly involved in or essential for

AAV gene expression. Thus the regulation of AAV gene

expression may largely depend on both cellular and AAV

encoded functions whose activities may be modulated by

helper virus gene products.

In this study, the regulation of the AAV genome has

been investigated. Mutant genomes created in vitro have

been studied in order to define the genetic elements

involved in the autoregulation of AAV gene expression. The

results show that autoregulation occurs on at least two

levels; first, a product encoded by the AAV rep gene is

essential for activation of AAV transcription, and second,



elements within the coding regions for the rep gene products

negatively regulate expression of the rep genes.


Structure of mutant AAV genomes. The structures of

mutant AAV genomes created during this study are summarized

in table 3-1. Many of the cloned genomes were derived from

previously described mutants (Hermonat et al., 1984). The

name of the plasmid genome indicates the type of mutation.

Frameshift insertion mutations are indicated by ins followed

by the position of the insertion, indicated in mu. The

insertions consist of 8 base synthetic oligonucleotides

containing a BqlII restriction site. Deletion mutations are

indicated by dl followed by the location of the deletions.

Mutants containing nonrepairable deletions of both TRs are

designated as ori-. Certain mutants contain a tandem

duplication of AAV DNA and are indicated by dp and position

of the duplicated region. Other mutants contain an

insertion of the SV40 early promoter and are indicated by

/SV following the initial mutant description.

Accumulation of AAV RNAs by two types of rep mutants.

To determine whether an AAV nonstructural protein encoded by

the p5 or p19 genes regulates AAV gene expression, the

accumulation of RNAs by two types of replication-defective

AAV genomes was compared. The first types contain 8 base

frameshift insertions at either 11 or 32 mu and thus are

defective for the production of the trans-acting p5 and p19


Plasmid Description

d103-11 (pLBl415o) Ligation of BqlII/EcoRVa digested
d103-23 and insll

dill-23 (pLBl415b) Ligation of BglII/EcoRV digested
d103-23 and insll

d123-37 (pLB1406) Ligation of BglII/EcoRV digested
d123-28 and d110-37
dl63-86 (pLB101) AaI deletion of pSM620

d163-80 (pLB102) ApaI deletion of pSM620

d148-52 (pLB202) XhoI deletion of pSM620

d115-86 (pLB306) NcoI deletion of pSM620

dl80-96 (pLB314) NcoI deletion of pSM620

inv15-86 (pLB315) Inversion of NcoI A of pSM620

pAAV/0-11 (pLB710) pSM620 PstI C into PstI site of pBR322

pAAV/11-42 (pLB709) pSM620 PstI B into PstI site of pBR322

pAAV/42-91 (pLB705) pSM620 PstI A into PstI site of pBR322

pAAV/91-100 (pLB702) pSM620 PstI D into PstI site of pBR322

620ori- pSM620 SmaI A into EcoRV site ofpBR322

insllori- (pLB1207) insll SmaI A into EcoRV site of pBR322

ins32ori- ins32 SmaI A into EcoRV site of pBR322

ins32ori-b ins32 Ball A into EcoRV site of pBR322

ins42ori- (pLB3411) ins42 SmaI A into EcoRV site of pBR322


TABLE 3-1 (continued)


ins78ori- (pLB2402)

d103-23ori- (pLB1302)

dl10-37ori- (pLB1102)

d163-86ori- (pLB801)

d158-870ri- (pLB2904)

dE28-49 (pLBp40A)

d_37-58 (pLBp40B)

d103-23/SV (pLB605)

d103-05/SV (pLB1701)


dl03-05/SVori~ (pLB1801)

dl03-11/SV (pLB4104)

d103-20/SV (pLB4004)

ins78 SmaI A into EcoRV site of pFB69

d103-23 SmaI A into the EcoRV site of

d110-37 SmaI A into the EcoRV site of

ApaI deletion of 620ori-

d158-87 SmaI A into the EcoRV
site of pAT153

Ligation of BqlII/EcoRV digested
d149-94 and d123-28

Ligation of BglII/EcoRV digested
d158-87 and d110-37

SV40 early promoter into BqlII
digested d103-23

Insertion of Aval/BstEII of pSM620
into BgiII/BstEII digested d103-23/SV

Ligation of SacI/BstEII digested ins32
and d103-05/SV

d103-05/SV SmaI A fragment into the
EcoRV digested pBR322

Ligation of BglII/BstEII digested
d103-23/SV and insll

Ligation of BlgII/EcoRV digested
d103-23/SV and BclI/EcoRV digested


encoded rep gene products. The second type is represented

by a mutant containing deletions of both TRs, 620ori-. The

replication of 620ori- is blocked by cis-active defects of

the origins of DNA replication. This plasmid was used as a

control rather than the parental pSM620 plasmid so that the

effects) of mutations within the rep genes can be directly

analyzed. Because mutations within the rep genes abolish

AAV DNA replication, the use of an ori~ control eliminates

possible differences due to increased copy number of

replicating genomes or by differing conformations of

replicating and nonreplicating transcription templates. The

RNAs accumulated by the AAV genomes were analyzed after

transfection into Ad2-infected KB cells. Total cytoplasmic

RNAs were isolated from the transfected cells at 30 hr post-

transfection. The RNAs wer fractionated by formaldehyde gel

electrophoresis and analyzed by blotting and hybridization

to radiolabeled AAV DNA as shown in figure 3-1.

The 620ori~ mutant (lane 1) produced readily detectable

levels of all the AAV transcripts except for the spliced p5

RNA. This RNA is normally present in relatively low

abundance and is difficult to resolve from the unspliced p19

RNA. The RNAs produced by 620ori- were in the same

stoichiometry as that produced by wild type AAV (Green and

Roeder, 1980a; Laughlin et al., 1979b), indicating that DNA

replication was not required for qualitatively normal

expression of the AAV genes. The most abundant RNA


accumulated was the spliced 2.3 kb p40 transcript. The

larger p5 and p19 RNAs were accumulated in significantly

lower amounts.

Although the nonreplicating 6200ri- genome produced

readily detectable levels of transcripts, the nonreplicating

ins mutants did not (lanes 2 and 3). Both insertion mutants

appeared equally defective for the production of RNAs. The

subtle nature of the insertion mutations suggests that a

product encoded within rep genes is required for efficient

expression of all the AAV genes.

Similar results were obtained when total cellular RNAs

were analyzed as shown in figure 3-2. The amounts of

transcripts accumulated by the rep mutants insll and ins32

were significantly lower than by 620ori~, again indicating a

critical role for the p5 gene product in gene expression.

The amount of p40 RNA detected was higher for insll than for

ins32 in this experiment, although this result was not

always reproducible. Low amounts of p5 RNA from both

insertion mutants were also detected.

The accumulation of RNAs by a deletion mutant, d103-23,

was also analyzed in this experiment. D103-23 accumulated

an intermediate level of p40 RNA, higher than by either rep

insertion mutant but significantly lower than by 620ori-.

The overproduction of the p40 RNA by left side deletion

mutants was reproducible and is discussed below.





Figure 3-1. Accumulation of RNAs by two types of
rep mutants. KB cells were transfected with 20 pg of either
620ori- (lane 1), ins32 (lane 2), or insll (lane 3) and
infected with Ad-2. At 30 hours postinfection, cytoplasmic
RNAs were isolated, and 20 pg of RNA from each experiment
were fractionated on a formaldehyde-agarose gel and
subsequently analyzed by blotting and hybridization to 32p_
labeled AAV virion DNA. The sizes of the RNAs are as shown
in figure 1 and indicated here are the 4.2 kb unspliced p5
RNA, the 3.6 kb unspliced (top of caret) and 3.3 kb spliced
(bottom of caret) p19 RNAs, and the 2.3-kb spliced p40 RNA.
The 2.6-kb unspliced p40 RNA is seen directly above the 2.3
kb RNA.

1 2 3 4
4 B


2.32 *

Figure 3-2. Accumulation of RNAs by mutant AAV
genomes. KB cells were transfected with either insll (lane
1), ins32 (lane 2), 6200ri- (lane 3), or dl03-23 (lane 4)
and infected with Ad-2 as described for figure 3-1. 20 ug
of total cellular RNA from each experiment were analyzed as
described for figure 371. The sizes of the unspliced p5
(4.2 kb), p19 (3.6 kb), and spliced p40 RNAs (2.3 kb) are
indicated on the right.


Accumulation of AAV RNAs by rep- ori- double mutants.

The AAV genomes used in the experiments described above

contained either of two types of mutations, both of which

blocked DNA replication. The rep mutants (ins and dl)

contained trans-complementable defects in the AAV rep gene,

while the defective replication of 620ori- was

noncomplementable due to deletion mutations within the TRs

(which act as origins for AAV DNA replication). Thus the

difference in the accumulation of transcripts by the mutants

might be explained either by the loss of a positive

activator of expression in the rep mutants or by the removal

of a cis-active repressor of expression within the TRs of


To distinguish between these possibilities, we analyzed

the accumulation of transcripts by mutant AAV genomes

containing similar deletions of the TRs. The RNAs produced

by the genomes upon transfection into Ad2-infected cells are

shown in figure 3-3. Both rep insertion mutants (insllori"

and ins32ori~) produced greatly reduced amounts of

transcripts compared with 620ori- (lane 1). Thus the

defective expression of the ins mutants was due to the

lesions in the rep gene.

An ori- capsid deletion mutant, dl63-86ori~ accumulated

normal amounts and proportions of its deleted transcripts

(lane 4). This indicates that AAV capsid proteins have no

role in the accumulation of AAV RNAs.


Although most mutations within the rep gene abolish AAV

DNA replication, a mutant containing an insertion at 42 mu

is partially defective, replicating at about the 0.1% level

of wild type genomes (Hermonat et al., 1984). To determine

if this mutant is also defective in the accumulation of

RNAs, an ori- version of ins42 (lane 6) was constructed and

analyzed. This mutant genome produced intermediate levels

of RNAs compared to ins32ori- (lane 7) and 620ori- (lane 5).

While the mutation at 42 mu reduces DNA replication by 100-

1000 fold (Hermonat et al., 1984) the accumulation of RNAs

was only inhibited by approximately 10 fold. This may

suggest that different regions of the AAV rep proteins are

required for DNA replication and regulation of gene


The predominant RNA accumulated by ins42ori- was the

spliced p40 message, demonstrating that the mutation did not

severely affect the splicing efficiency of AAV messages.

Thus the defects caused by mutation at 42 mu do not appear

to be caused by defects in RNA splicing but by lesions

within the coding region contained in the AAV intron,

supporting the suggestion that products encoded by the

unspliced p5 and p19 transcripts are essential for AAV

replication (Hermonat et al., 1984).

The results presented above indicate that a product

encoded within the AAV p5 and p19 genes is required for the

efficient accumulation of AAV RNAs in helper virus infected


cells. In order to determine if this positive autoregulation

by the rep gene product occurred in the absence of helper

virus, the RNAs accumulated by cloned genomes upon

transfection into uninfected KB cells were analyzed. As

shown in figure 3-3 equally low levels of the 2.3 kb p40

transcript were accumulated by 620ori- (lane 8) and

ins32ori-(lane 9) in the absence of Ad while only 620ori-

produced high levels of transcripts in the presence of Ad.

Thus efficient activation of expression only appeared to

occur in Ad infected cells.

We also compared the accumulation of RNAs by ori-

genomes which have deletions on the left side of the AAV

genome. A blot comparing the RNAs produced by d103-23ori~,

dl10-37ori-, 620ori-, and ins32ori- is shown in figure 3-4.

As expected, genomes containing mutant rep genes accumulated

much lower amounts of transcripts than 620ori". However,

d103-23ori" produced higher levels of the 2.3 kb p40 message

than by the other rep mutants. Thus both ori- and ori+

versions of d103-23 show enhanced accumulation of the p40

RNA suggesting that upstream sequences negatively regulate

transcription from p40. D110-37ori~ did not overproduce the

p40 RNA suggesting that deletion of sequences between 03-10

mu was required for the enhanced accumulation of the p40


'd0 -p-4 a ) 0 )
0 0 a) 04 t k

a o
w-.o o H

0 tD

1 o 1 D 44 ) 44.4.
I r-4 .O *4 H rg
M OI .0_ .

r. P r. :, to i
W 0 m J 0--,

St4 L4 4
4) M0 -
C r- I 3 g T

0 0) r. 4J -

*' a 44 Q r >

o (0 ,0 4 4-)y
0 r- J 04
(0 r-l (4 a 1W4

i g4 *4
aC ) r-43 o

4J F Ol Cl
o .o t .
4 I r-l O l O

r a) 0U) C
O n$ 1 O I
S41"' o

CI mr

4 -It0- n ( 4 4
V rU 0o (U (n

5 O O
r r-1 INP .0 0 04
i3 W4I 4 -r (03

u) r- i t o C --I
M I O (0 M

00 *H ) 4-4 Q-H x



)00) 0

0. 0.






1 2 3 4



a A

Figure 3-4. Accumulation of RNAs by rep deletion
mutant genomes. Ad-2 infected KB cells were transfected
with 20 pg of 620ori~ (lane 1), ins32ori- (lane 2), d110-
37ori- (lane 3), or d103-23ori- (lane 4). A 20 pg amount of
RNA from each transfection was analyzed as described in the
legend to figure 3-1. Only the position of the spliced 2.3
kb p40 RNA is indicated.


Transcription of AAV cenomes in isolated nuclei. The

data presented above showed that mutants with defective p5

and p19 ORFs were defective in the accumulation of AAV

transcripts. In order to determine if the defect was at the

transcriptional or post-transcriptional level, the rates of

transcription from transfected AAV genomes were measured in

isolated nuclei (figure 3-5). Nuclei from Ad2-infected HeLa

cells transfected with either 620ori~ (panel A), ins32ori~

(panel B), or mock transfected (panel C) were isolated and

incubated to allow runoff elongation of preinitiated

transcription complexes.

Radiolabeled transcripts were isolated and equal

amounts hybridized to Southern blots containing 620ori- DNA

which had been digested with PstI and HindIII (lane 1).

This digestion produces four fragments. Fragment a (3.8 kb)

contains pBR322 sequences in addition to 0.4 kb of AAV DNA

(91-98 mu). Fragments b and c contain only AAV DNA from 42-

91 mu and 11-40 mu, respectively. A fourth fragment, d,

contains AAV DNA from 2-11 mu as well as a small amount of

pBR322 DNA. Each blot also contained PstI digests of cloned

Ad5 DNA, either pXHOC (Ela and Elb sequences, lane 2) or

pXBAC (E4 sequences, lane 3) to confirm that equivalent

amounts of labeled nuclear RNA were used in each experiment.

Although equal amounts of labeled RNAs were used in

these experiments, the level of Ad transcription as

indicated by hybridization signal in panel B (ins32ori-)


appears to be approximately two-fold higher than in panel A

(620ori~). That this represents a consequence of 620ori~

expression seems unlikely since only a fraction of cells can

be expected to receive AAV DNA after transfection while

virtually all cells are expected to be infected with Ad.

However, an inhibition of Ad transcription by AAV is

documented below.

Although the level of Ad transcription detected was

higher upon transfection with ins32ori-, hybridization

signal to the AAV-specific fragments was observed only when

labeled nuclear RNA from 620ori- transfected cells was used.

No AAV-specific transcription from ins32ori- was detected in

isolated nuclei in this experiment. Thus, expression of the

rep insertion mutant was defective at the level of

transcription initiation.

Complementation of rep mutant gene expression. The

experiments described above suggest that an AAV gene

products) encoded in the p5 gene is required for activation

of AAV transcription. If so, the defect should be

complementable in trans. To test this, cotransfection

experiments were carried out using two rep- ori- double

mutants, ins32ori- and dll0-37ori-. Ori- plasmids were used

so that any induction of expression would be assayed for in

the absence of an increase in potential template number due

to DNA replication. The plasmids were transfected into Ad-2

infected KB cells or cotransfected with a capsid deletion

Q. H -PX .9r
Sr o 0)H U 0- -
0 4cp H0 l~i r. C
$2 P 4 I rO '
4 En 0H H Z H ( 0
.U) =I 44 44 rd 0


Ut HZ I=
a) u

(pi -po r3cO 4M 4J Q P

0 r
z Cfi
* HU 0) 0 g Z-
Ho (o-i OH q (1)
HU .,A 0 p 0 $4 -H t
I0) -H En 0 44 04 0
W 0O UH 0 94
0. 4J >4 9o ~ m 9:
cr 0 tp 4 %0 m o F
C.4 g r P4a l a
9 m (n 'c .
.Hl rj), rq 4J tp 44\ o
Z 0 4J 4j g: (00 ri tt
Q) a) .,i~ Q) U 4-4I
(W4 P 44

Q uO

0 0 cl

44 `4 r. Q 4J 4 0
0 a) A tAto r z -zr :
r IE
$41 ro 0 U 4
0 0 (1) ., f 0 Z cc
.r.4 Caj u U gi H (d
M 4 W o z 0~
4-) %0 U :j 4 Q) t
.rj a ) Q U 444-

u H o J (a > q4

E-1 rc$ n a< "ci >4 r

4- r-I 4j 4Jr j 0 (
(D 0) 9 r: 0 U 44 a)(
tA m :3 ca rol 0 > t3) d O
(a 04 "t aU 4 r l P
rl kQ, O C to 44 F-i >4
(1) -H 4J r-4 H 0 4J HCO~a
uu4 ka 5 in u

. l I



S .
.^^aal i:





- ; YlP~...i A~

of 1

mutant, dl49-94. Total cytoplasmic RNAs were analyzed by

Northern blotting and hybridization to a labeled fragment of

AAV DNA (52-91 mu) so that the complementing genomes

transcripts were not visualized. As shown in figure 3-6, no

transcripts were detected upon transfection of either

ins32ori- (lane 2), d110-37ori- (lane 5) or, as expected due

to the specificity of the probe, upon transfection of the

capsid deletion mutant (lane 6). Cotransfection of the

capsid deletion mutant with ins32ori~ resulted in the

production of detectable levels of a 2.3 kb RNA, predicted

for the p40 message of the rep insertion mutant. As

expected, the larger RNAs of the insertion mutant were

present in much lower amounts than the p40 message.

A 2.3 kb transcript was also detected in a

cotransfection experiment using dl10-37ori- (lane 4). In

this case, a second transcript slightly slower in migration

was also detected in approximately equal amounts to that of

the 2.3 kb RNA. This extra band appeared to represent a p5

transcript since dl10-37ori~ contains a p5 promoter which

would produce a highly truncated spliced transcript of

approximately 2.6 kb. The p5 transcript is normally

accumulated at levels of 5-10% of the p40 transcript. Thus,

the p5 RNA of dl10-37ori- appeared to be overproduced.

One explanation for these data was that expression from

the p5 promoter is artificially high due to transcription

off a nonreplicating template. Alternatively, expression


may have been enhanced by the fortuitous construction of a

positive regulatory element (e.g. an enhancer) during the

cloning procedures. The identification of the 2.6 kb

species as a p5 transcript in the above experiment is also

difficult due to the low amount of transcripts produced by

complementation of the nonreplicating genomes. Therefore

several experiments were done using various ori+ re

deletion mutants.

In the first of these experiments the production of

transcripts by left side deletion mutants upon

complementation with capsid deletion mutants was examined.

Capsid deletion mutants d158-87 and d163-86 (figure 3-7,

lanes 1 and 2) produced readily detectable amounts of their

deleted transcripts while the rep deletion mutants did not

(lanes 3 and 6). Cotransfection of d103-23 with d163-86

resulted in the accumulation of large amounts of 2.3 kb p40

transcripts (lane 4) which must have been produced from

d103-23 since the capsid mutant does not produce a 2.3 kb

message. Further, the lack of wild type size p5 and p19

messages indicated that the 2.3 kb message did not arise

from a wild type genome generated by recombination.

D110-37 produced high levels of 2.3 kb p40 RNA when

complemented by d158-87, as well as an equally abundant 2.6

kb RNA. Because d110-37 contains only a p5 promoter in

addition to the p40 promoter contained within d103-23, this

2.6 kb RNA is most likely a p5 transcript. The 2.6 kb RNA


also hybridizes to a p5 specific probe (see below). Thus

the p5 RNA was overproduced by replicating and

nonreplicating genomes.

One explanation for the increased accumulation of the

deleted p5 transcript is the juxtaposition of p5 to

regulatory elements of p40. In order to determine if

overproduction of p5 RNA required sequences within p40, a

deletion mutant containing only the p5 promoter, d115-86,

was constructed and analyzed. RNAs accumulated by this

genome upon transfection with and without pSM620 (to supply

p5 and p19 rep gene products) are also shown in Figure 3-7.

A high level of the deleted p5 transcript was detected upon

complementation. The structure of the deleted RNA also

eliminates the possibility that it was produced by a

recombinant genome. This RNA was produced in approximately

equal amounts with the pSM620 p40 transcript and in

approximately 16 fold excess with the pSM620 p5 transcript

(as determined by counting the radioactive bands cut out

from the nitrocellulose blot). Thus, overproduction of the

p5 message was independent of regulatory sequences within

the p40 promoter. Further, overproduction of the p5 message

is not unique to a single mutation, indicating that

expression was not enhanced by the inadvertent construction

of a positive regulatory element.

These complementation experiments clearly demonstrate

positive trans-activation by rep gene products, although we


cannot rule out the possibility that the high level of

transcript accumulation in these experiments may in part be

a dosage effect due to DNA replication and concomitant

increases in potential transcription templates.

The overproduction of p5 RNA by genomes containing

deletions between 15 and 37 mu suggests the existence of a

negative regulatory element. Since the presence of this

putative negative regulatory sequence in wild type genomes

does not appear to eliminate expression from p5, but rather

reduces its accumulation the putative regulatory element

will be referred to as a depressor (dep) of expression and

the mutant phenotype as a dep- phenotype. Because

complementation of dep- mutants did not affect the ratio of

transcripts produced by the complementing genome, the dep

mutations must be cis-active.

In order to more precisely map the putative dep

sequence, additional deletion mutants were constructed and

analyzed as shown in figure 3-8. In this experiment several

left side deletion mutants were cotransfected with a capsid

deletion mutant, d163-86. No AAV RNA was detected after

transfection of the left side mutants without

complementation (data not shown). Complementation of d123-

28 resulted in the production of the expected rep mutant

RNAs. The p5 and p19 RNAs of d123-28 were produced at

normally low levels compared with the 2.3 kb p40 RNA. Thus

dep does not appear to reside between 23-28 mu. Similarly



i, i l : ... :. -- ^


Figure 3-6. Complementation of expression of rep
mutants. KB cells were transfected with 20 pg of either
620ori~ (lane 1), ins32ori- (lane 2), ins32ori~ and d149-94
(lane 3), dl10-37ori- and d149-94 (lane 4), dl10-37ori-
(lane 5), or d149-94 (lane 6) and infected with Ad-2. RNA
was prepared and analyzed as described in figure 3-1 except
that the blot was probed with a 32P-labeled fragment of AAV
DNA from 52-91 mu. The position of RNAs indicated on the
left are the wild type spliced p40 RNA (2.3 kb) and on the
right the unique p5 transcript of dl10-37ori- seen in lane 4
(2.6 kb).

1234 56







Figure 3-7. Complementation of expression by left side
deletion mutants. KB cells were transfected with 20 ig of
either d158-87 (lane 1), d163-86 (lane 2), d103-23 (lane 3),
d103-23 and d163-86 (lane 4), d110-37 and d158-87 (lane 5),
d110-37 (lane 6), pSM620 (lane 7), d115-86 and pSM620 (lane
8), or d115-86 (lane 9) and subsequently infected with Ad-2.
The locations of specific RNAs originating from capsid
deletion mutants are indicated by a superscript c. Only the
spliced p40 and unspliced p19 and p5 RNAs are indicated.
The bottom of the caret points to the d158-87 RNAs, and the
top of the caret points to the d163-86 RNAs. The truncated
RNAs of the rep deletion mutants are indicated with the
superscript r.


complementation of either d103-11 and dlll-23 also resulted

in production of rep mutant RNAs with wild-type


Although both d110-37 and dlll-23 are missing the p19

promoter, only d110-37 overproduced a p5 RNA (lane 5). Thus

the dep phenotype is not simply a reflection of reduced

competition due to the deleted promoter, as seen for the

SV40 early promoter (Kadesch and Berg, 1986). The 2.6 kb

RNA produced upon complementation of dllO-37 was shown

clearly to be a p5 transcript by its hybridization to a p5

specific probe (figure 3-8B, lane 5).

Because neither d110-23 nor d123-28 overproduced any

RNAs, the dep sequence can be predicted to reside between

28-37 mu. Indeed, d123-37 exhibited a unique phenotype upon

complementation. The truncated p5 transcript of d123-37 was

produced at low levels compared to the 2.3 kb p40 transcript

and was predominantly unspliced (based on the size of the

RNA). However, complementation of the mutant resulted in

the accumulation of an RNA equal in abundance and slightly

larger in size than 2.3 kb. This RNA was a size predicted

for the spliced p19 message of d123-37. Thus d123-37

appears to accumulate excess amount of the p19 RNA but

normal amounts of the p5 RNA.

The results of the complementation experiments are

summarized in figure 3-9. Deletion of sequences between 23-

37 mu was sufficient for overproduction of p19 RNA but


insufficient for overproduction of p5 RNA. Thus a

functional dep sequence for p19 appears to reside between

23-37 mu. Further because deletion from 23-28 mu had no

effect on RNA levels, one dep region must reside between 28-

37 mu. However the functional dep sequence for p5 appears

to also require an element between 10-23 mu. These

conclusions are discussed below.

Another point worth noting is that although the p5 and

p19 messages are normally accumulated in predominantly

unspliced form, the p5 RNA or p19 RNAs overproduced by dll0-

37 or d123-37, respectively, were a size predicted for

spliced messages. Thus the dep mutations appeared to

simultaneously increase both the inherent splicing

efficiency of an RNA as well as the abundance of that RNA.

Negative regulation of p40 expression by the dep

region. The expression of deletion mutants suggests that

the p5 and p19 genes are controlled by negative regulatory

elements, one of which appears to exist between 28-37 mu of

the AAV genome. To determine if dep is a separable genetic

element, hybrid AAV genomes were created which placed a dep

region downstream from a p40 promoter. This was done

because expression from p40 appeared unaffected by the dep

mutations. Genomes which contain a tandem duplication of

sequences including p40 and which either include (dp28-49)

or do not include (dp37-58) the putative dep region were

created and analyzed. The structures of the genomes are

(L) m r.n Enr ) 1 () I a)
Zc C(N 0 t0 0 N W't rOM04r
0 1 -*H4J ~-H a)E-4
HII rml 4J CO tt U ) -4M
co r-I 0 Z 0) 1 U

ul 41) Q p4) -H p p m
% \D 014 04 (0 04 44
N H' 4-) -4H
4JM M o a 1 c

g~~ r- 4J Q 4-H H 40q
rt lQ) 4 0

cnnc tor 0a

0~ (1) Z
Q01rb) 0 44 0)

44 44 'o IV-
+'+, tllC 9: t7.k

a $0 r a)

V D 4J 4- 0 .ii 4-)
.H tn~a) r >

to 1-,4 4 cr
a) k ~- ~, 0 '04. m
coa g CDa CD
4J LO fc 0
a) (1) ko 9 a 4-)
= tz -ri V

CI) E4 V
r- 4.4 0O 441 (C 0 D0 HH
-o 0 *4-

f1C0' 04J4 %: r
tn 000) >4 0 -r-

(1O CN to
() M 134 fn"aa,~ a
4)~ C%4 1 4) -Hc r
44 M (), ,~ Er~~+ CN
(0 4 P rO 44 p (0 r
0 4J p r
ro (L) 44 0
(1) 0) r4 () to
U') 44 :1 0 4-
0 4) -'r 4J 4
4) r (1 ).o
0 M rmi 44 a V 0 C
r. a) a) u

ck C Oa r;4 V :
0 Hul o4 H

m E-4 r m (1) ml 0

P4 r 4) ) (1 -rq0 1

4 -- 9i 4) W 0 J P E


...,.... ...

.......................... .. .
,mgI 3 "

CV )*
o 0t



U 0


(0 >4 PO n l4-4 3 Q) (
l i -.- 4J 0 0 -4 0)
0 r z rI41(a
z 04 (0 E n t/I
(Q)W04Jr-4 -44W
trc U ~ a
M i) 0Q)
, .ri r4 43 0
Prd0C 4J 0 0 -
-H 44 43 > --4 4

03a)- 0H 4
E-04 3 M r-4 43 (L >

o00-,1 4) U) U
P~ 134 (nU 4 --4 (1)
4- 043& P 0
0 0) 4-) 4-1 E
a)~ 4 9i .4 k
id -I z v 0
(($ 00
0J r. C 0 4J 0)
*H0 4) -4*04 43

0 0 0 r ... 0 0
O) Z 041 aW m.c
Va 0 pU Vo t
a~~cu~0 -'-4'

43H 04 rV
4J 4) 1 0 aT4 C

P 44
(44 -4n 4 0C
Q) 043
444 04 a) M H

tn >4 0-1 u
QrH-41 (a v0 0x

04-r- r-4 r V 0

4- r 4j4 0 4C
c~0 4-I X 0 0
rI P i9:a
P4~ (0 043 r-40
V 0 Z (1 4)
oN V434 etC)rI
1~ 0 o,- Q
M 0 Vw r- -I0
FO 0 W z p
-H43U 043 'U0
p V) (1) V 9 4) Nr
0 rC% 4j r- P C LO
44 -r- p () P
w m p 0 rAm

--14- tP(04-) U



:: -m = x z

I .I X I X

jI j =



a a

Na N C
'O If


diagramed in figure 3-10. Each genome contains an upstream

and downstream p40 promoter (p40u and p40d, respectively).

The structure of the p40d transcription unit is identical

for both genomes thus the levels of this transcript (wild

type 2.3 kb in size) can be used for comparison with that of

the upstream p40. However, the upstream p40 promoters

differ, being placed upstream from different but overlapping

duplicated portions of the genome. Only in the case of

d_28-49 does the duplicated region contain the dep


As shown in figure 3-11, both genomes produced

approximately equal amounts of the downstream p40 message.

However dD28-49 accumulated significantly lower amounts of

the p40u transcript compared with its own p40d transcript or

with that of d_37-58. The ratio of the p40u to the p40d

transcripts of dp28-49 was qualitatively similar to the

p5/p19 to p40 ratio of transcripts accumulated by wild type

AAV indicating that the dep element inhibited the

accumulation of the p40 transcript when placed within the

transcriptional unit.

As noted in the figure 3-10, several different sized

messages could be produced from the p40u promoter of either

of the two genomes depending on the number of splicing

events in a particular message. Interestingly, the

predominant p40u message produced by d_28-49 was a 3.5 kb

unspliced message while the predominant message produced by


dp37-58 appeared to be the 2.9 kb doubly spliced message.

Thus the inclusion of the dep sequence in dE28-49 also

appeared to inhibit splicing of the message. These data

again support the notion that dep functions both to affect

the steady state levels and splicing efficiency of

particular messages.

Stability of AAV mRNAs. Several mechanisms could explain

the action of the dep region. One possibility includes the

specific destabilization of the RNA. If true this would

predict that the wild type p5/p19 and p40 messages would

have very different half lives. In order to examine the

stability of the AAV RNAs, AAV infected cells were treated

with actinomycin D (to block transcription) for various

times, and the RNAs analyzed by northern blotting.

As shown in figure 3-12, most of the AAV RNAs remained

constant in level up to 4 hours after addition of

actinomycin D. Although the 2.6 kb p40 message decreased

rapidly in abundance, the levels of the p5 and p19 RNAs

remained relatively constant. These data suggest that only

the 2.6 kb message is unstable. Alternatively, the

disappearance of the 2.6 kb RNA may reflect a continued

processing of the message to the spliced 2.3 kb species

after the addition of actinomycin D, although this seems

unlikely since the levels of unspliced p5 and 19 RNAs

remained constant.

M W0 )4-4

4.)42 Cli4J j0
S0 0V 4Jt-
g. g. (r4 4-4 u)0
W 0 0 0
44 ts::(U Z '

0 o(% 0)( Qr .4JZ

1:1. 1" 2 t nW .
., ~o 4.) 0 oi* I : i0
P '.340 a)4-'
E~~dEc- ~ (

u 10 0

:j 0N .4 4J 0
ro ,t g 4J (U c 7

42 00 p- 44 0 a) ~ -H 4 S
C1N4 0) 4 0) 0 w1

P-H 0 0
U)0)d4)J 0)420 4
E-4rd r4Co

0 (1) 0 to )..HO

0. 4S4T43 4
~r -4- 44

041 -40

*H44 4 4 ) r. 0.
0 U) p- ~' 0 0) ,

0 E4 4 4 tV'I4J )
04 tp 0, 4J t

4) 4' 0 0 0)
0 4J W ..0 04 Q))
0 t 0 )0 o0
0 rq 0 004 J r

S) C) 4- 0 )

0 r442
H 44(044) "U4
0 4J Q, -) 0
w 01 P 4-), a)
a) 0 r k Q) M En X:
P -4 (a x ro Q o 0) 4

44 9 07 U 0 0 Zl
0 u-,4 b) ra 0 to

to r- r-q Q) 0,
4 O 04 (V ) M

u 9: :j 4 4 0 ) -rq(L
0 W r-I O 4J4J PW t


0 No
! (Ko

















Figure 3-11. Expression of AAV genomes containing
duplications of p40. Ad-2 infected KB cells were infected
with AAV (lane 1), mock transfected (-), or transfected with
dE28-49 (lane 2) or dp37-58 (lane 3). The wild type size
messages are indicated on the right and RNAs originating
from the upstream p40 promoters (2.9 and 3.5 kb) and the
downstream p40 message (2.3 kb), are indicated on the left.






Figure 3-12. Stability of AAV RNAs. KB cells were
coinfected with Ad-2 and AAV-2. Actinomycin-D was added to
separate plates at 24 hours postinfection. Cytoplasmic RNA
was then isolated at 0 (lane 1), 30 (lane 2), 60 (lane 3),
or 240 (lane 4) minutes after addition of the drug. The
positions of specific RNAs are indicated on the left side of
the figure.


Transcription of AAV in isolated nuclei. Another

mechanism by which the dep region might act would be through

negative regulation of transcription initiation. This

predicts that the p40 promoter and downstream sequences

would be preferentially transcribed. Transcription rates of

different regions of the AAV genomes were measured by NRT

analysis using nuclei from Ad-2 or Ad-2 and AAV-2 infected

cells, followed by hybridization to restriction digested AAV

DNA blotted to nitrocellulose. The cloned AAV DNAs used are

identical to that described for figure 3-5 and thus

recognize either p5 specific transcription (fragment d), p5

and p19 transcription (fragment c), or transcription from

all AAV transcription units (fragments a and b).

In the first experiment infected KB cells were used

(figure 3-13). No AAV transcription was detected from Ad-2

infected cells while a large amount of hybridization signal

to AAV DNA was detected using RNA transcribed in nuclei from

coinfected cells (panel B, lane 3). Fragment b contains AAV

DNA from 42-91 mu and should hybridize to all complete AAV

transcripts. Fragment c contains AAV DNA from 11-40 mu and

should be specific for the p5 and p19 messages.

Although the steady state levels of p40 RNA (which consists

of mostly fragment b sequences) are 10-20 fold higher than

that of the p5 and p19 RNAs, more transcriptional activity

was detected with the p5 and p19 specific probe (fragment c)

than that detected with probe complementary to all AAV RNAs.


Thus the rate of initiation from the p5 and or p19 promoters

appears to be at least as high as that from the p40


A high level of transcription from the p5 promoter is

specifically indicated by comparison of hybridization signal

to fragments a and d. Both fragments contain approximately

0.2 kb of complementarity to AAV RNAs either at the 3' end

of all AAV transcripts in fragment a or to the 5' end of the

p5 transcript in fragment d. An equal amount of signal was

detected with either fragment indicating that the rate of

transcription for the p5 promoter is very high and is not

proportional to the amount of p5 RNA relative to p40 RNA.

Similar results were obtained using nuclei from

infected HeLa cells (figure 3-14). Transcription rates of

different regions of the AAV genomes were measured by NRT

analysis using nuclei from Ad-2 or Ad-2 and AAV-2 infected

cells followed by hybridization to restriction digested AAV

DNA blotted to nitrocellulose. The cloned AAV DNAs used are

the same as in the previous figure. Transcription was

measured at both 12 (A and B) and 24 hours (C and D)

postinfection. Although the signal detected with fragment b

was somewhat higher than for fragment c in this experiment,

the transcription rate per nucleotide of probe was

approximately equal. In this particular experiment the

fragment d was run off the gel.

Thus, in experiments using either KB or HeLa cells,

transcription from the p5 and p19 sequences appears

approximately equal to that of p40 specific sequences.

These data indicate that the overall ratios of AAV RNA

accumulated must be determined in part by mechanisms after

transcription initiation. Further because the rates of

transcription of the left half of AAV (11-40 mu) appeared at

least as high as of the right half of the genome (42-91

mu), it is suggested that the p5 and p19 messages are being

prematurely terminated.

Although transcription rates have not yet been measured

for dp mutants, the accumulation of RNAs by these mutants

is proportional to the observed transcription rates of the

wild type genome. It is proposed that the dep sequences act

posttranscriptionally and are responsible for the observed

difference between transcription rates and the steady state

levels of RNA p5 and p19 RNAs.

Another observation made in the last two experiments is

that the amount of Ad transcription detected was

significantly lower in cells coinfected with AAV. Ad

transcription detected using probes for either the El, E4,

or late regions was significantly reduced, although the

inhibition of transcription appeared greater in KB cells

(approximately 10 fold) than in HeLa cells (approximately 3

fold). The inhibition observed in HeLa cells was similar at

both early (12 hours) and late (24 hours) times

postinfection. Thus, although the AAV rep gene products

activate AAV transcription, AAV gene expression can be

correlated with an inhibition of Ad transcription.

AAV coinfection is known to inhibit Ad replication.

Interestingly, AAV appears to inhibit the production of

progeny Ad to a much greater extent than Ad DNA replication.

Thus the inhibition of Ad transcription may be a major

effector of the inhibition of the helper virus replication.

Whether the inhibition of Ad transcription is a consequence

of a specific AAV gene product or due to competition for

limited transcriptional machinery is not known.


In this study cloned AAV genomes have been analyzed in

order to delineate genetic elements involved in the

autoregulation of AAV gene expression. These data

illustrates that autoregulation occurs both by trans-active

AAV proteins and by cis active elements.

Positive regulation in trans by an AAV rep gene

product. Several experiments show that at least one AAV

protein encoded within the p5 gene is essential for trans-

activation of AAV transcription. First, cloned genomes

containing small lesions within the ORF unique to the p5 and

p19 transcripts accumulated extremely low levels of all the

AAV RNAs compared to genomes containing wild type p5 and p19

genes. Virtually all left side mutations had a similar

effect. The subtle nature of certain mutations (8 base

1 2


I 2 3








Figure 3-13. Transcription of AAV in isolated KB cell
nuclei. KB cells were infected with Ad-2 (panel A) or
coinfected with Ad-2 and AAV-2 (panel B). Nuclei were
isolated at 24 hours postinfection and transcription
reactions were carried out for 10 minutes. Equal amounts of
RNA were then hybridized to southern blots containing PstI-
digested Ad-5 Ela and Elb DNA from pXHOC (lane 1), E4 DNA
from pXBAC (lane 2) or PstI- and HindIII-digested 620ori-
DNA (lane 3). The AAV fragments indicated are as described
for figure 3-5.


0-9 Q 0 W
..4 I P Z o o
0) 4J 41) (1) r1 0
.0 c '0
4) c a 1 41 -r- r
*Q) 00
-4 ) U- ) C 0 0

0) ) 4- C

S) 0 c

o U

ino o
*M -m .* ( 4

o, a

0 (0 gW .U


> 0 I .

,CH *' *H
H'0 0

N*H 4. ) 0) 0
r4 0 r-4 <0 ri .

)- u 0 I. o

c 3 H 4,a
p r to n o u

-j (0 o4-> I 0

Q) *- w) 0 r-i W Q )
U k -rA 04Q c4 (n

0 0Q U

a I

C4 6

. .l ill4~L~L4~4~


frameshift insertions) indicates that the defects in

expression were due to mutational effects on the rep


Supporting this notion is the observation that the

defect in the accumulation of RNA was complementable in

trans. However the amount of transcripts accumulated by the

rep-/ori- mutants upon complementation by an ori+ genome was

significantly lower than that of 620ori-. This may reflect

a competition by the ori+ complementing AAV genome for

limited transcriptional machinery as noted by other workers

(see below). Complementation of expression of ori+ rep

mutants was very efficient.

Activation of AAV expression did not require DNA

replication. This was illustrated both by the efficient

expression of 620ori- and by the ability to complement

expression of ori~ rep mutants. The stoichiometry of the

different messages was also identical to that produced by

replicating molecules. Thus AAV gene expression is not

divisible into simple early and late stages separated by DNA


Evidence that activation by the rep proteins is at the

level of transcription initiation has also been presented.

When transcription rates of transfected genomes in isolated

cell nuclei were measured, transcription from a rep+ genome

(620ori~) was detected, although transcription from a rep

genome (ins32ori-) was not. Thus, the rep mutant was

defective at the level of transcription initiation.

One question that may be addressed is whether

transactivation by rep is specific for AAV genes or is

promiscuous in activity as observed for the Ad Ela (Green et

al., 1983: Treisman et al., 1983; Gaynor et al, 1984; Stein

and Ziff, 1984; Kao et al., 1985) and herpes ICPO and ICP4

proteins (Everett, 1984; O'Hare and Hayward, 1985a,b). The

observation that AAV infection inhibits transcription of the

Ad genome rather than activating it strongly suggests that

transactivation is specific for the AAV genome although this

point requires further investigation.

Because a mutation which specifically blocked

production of p5 gene products (insll) inhibited the

accumulation of AAV RNAs, a role for a p5 gene products) in

transactivation of expression is specifically indicated. It

is not yet possible to determine if p19 encoded proteins are

also required for this process. Although transactivation

appears independent of DNA replication, a p5 gene product is

required for both AAV DNA and RNA synthesis. Whether the

two functions of the rep gene products represent one or

multiple activities is not known.

However the phenotype of ins42 suggests that different

functional domains of the rep gene products might be

involved in RNA and DNA synthesis. Although the mutation at

42 mu inhibits DNA synthesis by 100-1000 fold, only a 10


fold reduction in RNA synthesis is seen. Thus the mutation

at 42 mu appears to differentially affect DNA replication

and transcription.

The AAV p5 gene products) appears to be functionally

analogous to the SV40 T-antigen. Both gene products are

required for viral DNA replication and transactivation of

viral transcription (Hartzell et al., 1984). As discussed

above, the AAV p5 gene product may also be composed of

separate functional domains as is the SV40 T-antigen.

Several workers have recently noticed significant amino acid

homology between the predicted p5 and SV40 T-antigen

proteins in a region conserved between most DNA dependent

ATPases (N. Muzyzka, C. Astell personal communications).

This suggests that the different proteins have similar

enzymatic activities.

It should be noted that all eukaryotic DNA virus

families, including Herpes, Ad, papilloma, and polyoma

viruses, examined in detail appear to encode transcription

activator proteins (Jones and Shenk, 1979; Everett, 1984;

Spalholz et al., 1985). Thus the ability to activate

expression may be as essential a feature to these exogenous

replicons as are origins of replication. This feature is of

obvious importance when one thinks of the competition for

limited transcription machinery between viral genes and the

vast excess of cellular genes.


Evidence for trans-active regulation of AAV gene

expression by the rep proteins has also been recently

reported by another laboratory. Tratschin et al. (1986)

have reported that expression of the cat gene, inserted

within the AAV genome under the control of the p40 promoter,

is enhanced by the rep gene. While this study did not

determine at what level activation occurred, our results are

consistent, although the results presented here indicate

that transcription of all AAV promoters is activated by rep.

In addition this group identified a negative regulatory

function of the rep gene products on expression of the p40

cat gene. This negative regulation occurred in human 293

cells, which constitutively produce Ad-5 Ela and Elb gene

products, but not in HeLa cells. Although the results

presented in this section do not address this phenomenon, a

negative regulatory activity of rep on heterologous genes is

described in chapter IV.

Rhode (1985) has also reported that expression of the

cat gene under the control of the p38 capsid gene promoter

of an autonomous parvovirus, H1, is enhanced by an H1 p4

gene product. Again, no work at the RNA level was done by

Rhode. Both Rhode and Tratschin et al. (1986) reported that

expression of genomes was most efficient when the

transactivator genes were carried on the same plasmid

compared to cotransfection of separate plasmids. This was

suggested to represent competition for limited transcription


machinery, possibly the parvoviral transactivator proteins


Cis active negative regulation of expression.

Although the p5 gene product is associated with positive

regulation of AAV transcription, several experiments

indicate that sequences within the p5 ORF itself act as

negative regulators of p5 and p19 gene expression. Several

left side deletion mutants accumulated significantly greater

than normal amounts of the p5 or p19 transcripts upon

complementation with rep+ genomes. Normally the p5

transcript is accumulated at 5-10% the level of the p40

transcripts at late times in Ad infected cells. However,

the dep mutants accumulated equal amounts of either the p5

or p19 and p40 transcripts. The effect of the dep mutations

were clearly cis-active as the expression of the

complementing genomes were qualitatively unaffected.

The relative increases in p5 or p19 transcripts

(compared to the levels of p40 transcripts) accumulated by

the dep mutants were not due to reduced expression from p40.

First, the amounts of p40 RNA accumulated by all left side

mutants upon complementation were similar. Second, ori+ den

mutant genomes produced as much or greater amounts of p40

transcript upon complementation than the complementors

themselves. Thus the dep phenotype was due to enhanced

accumulation of the affected RNAs.


A genome lacking all promoters except for p5 (d115-86),

accumulated a highly truncated p5 message in excess of the

complementor's p5 message. The high level of p5 expression

from this mutant demonstrates that the dep phenotype was not

a result of the close proximity of the p5 and p40 promoters

in the mutants. The unique structure of the truncated p5

RNA also eliminates the possibility that it was produced

from a recombinant genome of unknown structure.

Various deletion mutants were used to define the region

of the genome responsible for cis-active negative

regulation. The first mutant discovered to overproduce the

p5 transcript was d110-37 whose deletion removed the p19

promoter. However, deletion of the p19 promoter was not

sufficient to reproduce the dep phenotype as dlll-23

accumulated normal amounts of p5 transcripts. Thus one

would predict that the negative regulatory sequence would be

contained between 23-37 mu.

One deletion mutant, d123-37, exhibited a unique

pattern of accumulated transcripts. In this case the

truncated p5 transcript was still produced at low levels but

an additional band was detected which was slightly slower in

mobility and approximately equal in amount to the 2.3 kb p40

message. It is assumed that this species is a spliced p19

RNA based on the predicted size of that message. Thus

removal of 23-37 mu was required but not sufficient for

overproduction of the p5 transcript but sufficient for


overaccumulation of the p19 transcript. In explanation for

these observations it is proposed that two dep elements

exist within the AAV genome, one between 15-23 mu and

another between 28-37 mu. However since the presence of the

upstream dep region (between 15-23 mu) in one mutant, d123-

37, does not appear to inhibit the production the p19 RNAs,

the negative regulatory element most likely resides upstream

from the promoter (upstream from 19 mu).

The regulatory effect of sequences between 28-37 mu was

not specific to the p5 and p19 promoters. When the p40

promoter, whose expression was unaffected by the dep

mutations, was placed upstream from the 28-37 mu sequences

the accumulation of the upstream p40 transcript was strongly

inhibited relative to a second downstream p40. The

inhibition was not due to the competition by the second

downstream p40 since removal of 28-37 mu from the

duplication significantly enhanced the expression of the

upstream but not the downstream p40 transcript.

Although the studies presented here have not elucidated

the mechanism by which the putative dep sequences act,

several aspects are noteworthy. First the action of the dep

sequences appear directional, acting on upstream

transcription units. Second, the 28-37 mu sequences can act

on at least one heterologous promoter (p40).

Several mechanisms could explain the observed negative

regulation of p5 and p19 expression. One explanation is


that the putative dep elements specifically destabilize

RNAs. Initial results would indicate that the AAV RNAs do

not differ significantly in cytoplasmic stability; however

the messages may differ in nuclear stabilities.

Alternatively the dep elements might act as binding sites

for repressors of transcription initiation. If so, one

would predict that transcription rates of the wild type p5

and p19 genes would be much lower than that for the p40

transcript. However, when transcription rates were measured

using different DNA probes, an approximately equal amount of

transcription for all AAV genes was detected.

An additional mechanism might involve premature

termination of transcription. Transcription of SV40 and an

autonomous parvovirus, Minute Parvovirus of Mice, has been

suggested to be regulated by some method of transcription

attenuation (Ben-Asher and Aloni, 1984). This mechanism

appears to be the most consistent with the observations made


One other aspect of regulation by dep appears to be an

inhibition of splicing of the affected message. This

conclusion is based on the observation that in all cases an

increased splicing efficiency cosegregates with the dep

phenotype. The dep sequences also appear to inhibit splicing

of p40 transcripts when placed within a normally

predominantly spliced p40 transcription unit. These

observation suggest that dep exerts its effects in cis, by


altering the structure of RNAs. If so, such structural

changes can easily be envisioned to affect either

transcription termination or RNA stability as discussed


Finally we have not addressed the question of which

protein(s) may act to facilitate regulation by dep or why

such regulation even exists. Because dep does not appear to

affect expression of the p40 capsid gene, it is interesting

to speculate that dep is needed to modulate the amount of

rep protein expression at a time after DNA replication has

begun and when large amounts of capsid proteins are needed.

Conversely, expression of the p40 capsid gene appears to be

negatively regulated by rep in 293 cells (Tratschin et al.,

1986). 293 cells only express the Ela and Elb genes of Ad;

resulting in a physiologic state similar to that present

early during Ad infection. Thus, although AAV transcription

cannot yet be divided into early and late phases, such a

distinction may be made at the posttranscriptional level.

Such multiple levels of control are strikingly similar

to the regulation of the SV40 genome. Early in infection

the T-antigen gene is highly expressed (Benoist and Chambon,

1981; Buchman et al., 1984) while expression of the late

region is repressed (Omilli et al., 1986). T-antigen

synthesis leads to both DNA replication and transactivation

of SV40 transcription although the synthesis of early RNA is

kept relatively low compared to late RNAs by repression of


early region transcription by T-antigen itself (Tijan,

1982). The ability to negatively autoregulate expression

has been shown to be a common feature of viral transcription

transactivators, including the Ad Ela, Herpes ICP4, and SV40

T-antigen gene products (Tijan et al., 1982; O'Hare and

Hayward, 1985a,b; Hearing and Shenk, 1985; Smith et al.,

1985). It will be interesting to determine if regulation by

dep is also mediated by a rep gene product.

In summary, despite the small size of the AAV genome,

AAV gene expression appears to be controlled by a variety of

mechanisms. Thus the small size of the genome becomes a

great advantage due to the simplicity with which it can be

manipulated and analyzed. Further, the AAV genome is

subject both to regulation by cellular and helper virus

functions. Thus, the study of AAV gene expression should be

of pivotal importance in the elucidation of the interplay

between complex viral regulatory elements and cellular