Liver-directed strategies and immunologic consequences of recombinant AAV-mediated cross-correction of glycogen storage ...

MISSING IMAGE

Material Information

Title:
Liver-directed strategies and immunologic consequences of recombinant AAV-mediated cross-correction of glycogen storage disease type II
Physical Description:
xvii, 150 leaves : ill. (some col.) ; 29 cm.
Language:
English
Creator:
Cresawn, Kerry Owens, 1976-
Publication Date:

Subjects

Subjects / Keywords:
DNA, Recombinant -- physiology   ( mesh )
DNA, Recombinant -- therapeutic use   ( mesh )
alpha-Glucosidases -- physiology   ( mesh )
alpha-Glucosidases -- secretion   ( mesh )
Glycogen Storage Disease Type II -- prevention & control   ( mesh )
Mice, Knockout -- growth & development   ( mesh )
Mice, Knockout -- physiology   ( mesh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2004.
Bibliography:
Includes bibliographical references (leaves 136-149).
Statement of Responsibility:
by Kerry Owens Crewsawn.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 55897898
ocm55897898
System ID:
AA00011187:00001

Full Text











LIVER-DIRECTED STRATEGIES AND IMMUNOLOGIC CONSEQUENCES OF
RECOMBINANT AAV-MEDIATED CROSS-CORRECTION OF GLYCOGEN
STORAGE DISEASE TYPE II














By

KERRY OWENS CRESAWN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Kerry Owens Cresawn


































This work is dedicated to my mother, Mary Lea Lausch.















ACKNOWLEDGMENTS

There are many colleagues and friends whose technical support, advice, and

encouragement have made this work possible. I would first like to express my gratitude

to my mentor, Dr. Barry J. Byrne. Dr. Byrne's infectious enthusiasm for both patient

care and the science behind it has provided our group with limitless innovative ideas and

a better appreciation for our goals as scientists. I would also like to thank Dr. Byrne for

his understanding and support of my commitments outside of the lab as a mother. I

sincerely appreciate the guidance, insight, and patience of my friend and colleague, Dr.

Cathryn Mah. I would also like to thank my committee members (Drs. Henry Baker,

William Dunn, Alfred Lewin, and Gary Visner) as well as Dr. Mark Atkinson and Clive

Wasserfall for their valuable advice and time.

I greatly appreciate the assistance and support of the many members of the Byrne

Lab. Numerous friendly scientific discussions with my friend and colleague, Dr. Tom

Fraites helped to broaden my knowledge of both gene therapy and Pompe disease, and

provided helpful insight into experimental design and data analysis. I would like to

express my appreciation for the dedication and commendable work ethic of our

histologist, Melissa Lewis. I would also like to thank one of the most talented and skilled

scientists I know, Irene Zolotukhin (whose meticulous technique I have always admired,

although never mastered) for providing me with DNA, virus, and many other helpful

reagents. I greatly appreciate the patience and excellent organizational skills of our

laboratory manager, Denise Cloutier; and the countless hours Stacy Porvasnik has spent









with me, doing animal surgeries and necropsies. I thank all other members of the Byrne

lab, past and present (especially Dr. Mary Rucker, Dr. Priscilla McAulliffe, Christina

Pacak, Gregory Simon, Dr. Yoshihisa Sakai and Larysa Sautina). I would also like to

thank the University of Florida's Powell Gene Therapy Vector Core, and Pathology Core

(particularly Dr. Marth Campbell Thompson and Tina Yanchis) for their collaboration

and technical expertise.

The tireless effort of Laura Miriel and Joyce Conners, who direct graduate students

and faculty on a daily basis, has made the lives of all graduate students remarkably easier.

Without them, this work would not be possible.

I would like to thank our collaborators, Drs. Nina Raben and Paul Plotz at the

National Institutes of Health for developing and providing the animal model used in these

studies (as well as two additional animal models that have been useful in answering

important questions pertaining to this work). Additionally, the early work of Drs.

Rochelle Hirschhorn and Arnold Reuser provided us and others in the field with an

understanding of the molecular and cellular basis of GSDII, which has been necessary for

the development of treatment modalities.

Special thanks go to my friend and mentor during my undergraduate studies at

James Madison University, Dr. Douglas Dennis. It was in Dr. Dennis' lab that I first

learned to think as a scientist and have a passion for the field. I would like to express my

sincere gratitude to the patients and their families for sharing their stories of heartache

and courage. They inspire us to continue our efforts.

Finally, I would like to thank my family for their encouragement and support

during the difficult years of both undergraduate and graduate school. I especially thank









my husband, Steve Cresawn (who has shared the graduate school journey with me) for

his emotional support and continuous faith in my abilities as a wife, mother, and scientist.

I lastly want to thank the person who has changed my life the most during the last 3 years

of graduate school: my daughter, Abigail Cresawn. Her smile, laughter, and hugs have

made every day a gift.
















TABLE OF CONTENTS

Page

ACKNOW LEDGM ENTS............................................................................................... ..... iv

LIST OF TABLES ............................................................................................................ xii

LIST OF FIGURES.......................................................................................................... xiii

ABSTRACT ..................................................................................................................... xvi

CHAPTER

1 INTRODUCTION .................................................................................................... 1

Glycogen Storage Disease Type II............................................................................ 1
Three Variants of GSDII.................................................................................... 1
Cardiac and Skeletal M uscle Involvement ........................................................ 3
Acid Alpha-Glucosidase ........................................................................................... 4
Genotype-Phenotype Correlations..................................................................... 4
Biochem istry and Cell Biology .......................................................................... 6
Anim al M odels of GSDII.......................................................................................... 7
Therapeutic Strategies for GSDII ............................................................................. 9
Enzyme Replacement Therapy .......................................................................... 9
Gene Therapy...................................................................................................... 11
Recombinant Adeno-Associated Viral Vectors ...................................................... 13
Recombinant AAV Serotypes.......................................................................... 14
Liver-Directed rAAV Delivery........................................................................ 16
Liver-Directed versus M uscle-Directed Gene Therapy.......................................... 17
Immune Response to Vector-Derived Transgene Products.................................... 20

2 PRELIMINARY STUDIES TO EVALUATE LIVER-DIRECTED DELIVERY OF
RECOMBINANT AAV VECTORS EXPRESSING HUMAN GAA TO GSDII
M ICE.......................................................................................................................... 25

Background ................................................................................................................ 25
M materials and M ethods............................................................................................ 26
M olecular Cloning of DNA Constructs........................................................... 26
DNA Sequencing of Cloned Constructs.......................................................... 27
Packaging and Purification of rAAV Vectors ................................................. 27
In vitro Analysis of Liver Promoters Driving LacZ Expression......................... 27









A nim als.................................................................................................... ....... 28
Intrahepatic Recombinant AAV Delivery Methods ........................................ 29
T issue Processing ............................................................................................. 29
GAA Enzymatic Activity Assay ...................................................................... 29
ELISA Detection of Anti-GAA Antibodies..................................................... 31
H istology ............................................................................................................. 32
Statistical A analysis ........................................................................................... 32
R esu lts ........................................................................................................................ 3 2
In Vitro Analysis of Four Liver-Specific Promoters ....................................... 32
Superphysiologic Levels of Liver GAA Expression from the rAAV2-DHBV-
hGAA Vector are not Sufficient to Restore Activity in Affected Tissues....... 34
Recombinant AAV Serotype 5 Vector Yields Higher Liver GAA Levels than
Serotype 1 V ector ......................................................................................... 35
Gaa'- Mice Elicit a Humoral Immune Response to Vector-Derived hGAA...... 36
D discussion .................................................................................................................. 39

3 CHARACTERIZATION OF A NEONATALLY TOLERIZED GSDII MOUSE
M O D E L ..................................................................................................................... 4 2

B background .......................................................................................................... 42
M materials and M ethods............................................................................................ 43
Pretreatment of Gaa'- Mice ............................................................................. 43
Intravenous Protein Delivery ........................................................................... 43
ELISA for Detection of Anti-GAA Antibodies............................................... 44
ELISA for Detection of Distinct Isotypes........................................................ 44
Competition ELISA to Determine Antibody Specificity................................. 44
Lymphocyte Proliferation Assay ..................................................................... 45
In vitro GAA Inhibition Assay ........................................................................46
R esu lts........................................................................................................................ 4 7
Neonatal Pretreatment Results in Humoral Tolerance to Intravenous Protein
D delivery ............................................................................................... ......... 47
Antibodies are Specific to GAA ...................................................................... 48
Recombinant Human GAA-Challenged Mice Show no Signs of
Cell-Mediated Immunity.............................................................................. 50
Susceptibility to Tolerization Decreases from 1 to 7 Days after Birth............ 53
Anti-GAA Antibodies in Treated Serum Inhibit GAA Activity In Vitro........... 54
D discussion .................................................................................................................. 57

4 AAV5-MEDIATED CROSS-CORRECTION OF GSDII: EFFECTS OF
INHIBITORY ANTIBODY FORMATION AND IMMUNOMODULATION....... 59

B background .......................................................................................................... 59
Materials and Methods............................................................................................ 60
Cloning and Packaging of rAAV Constructs................................................... 60
ELISA for Detection of Anti-GAA Antibodies............................................... 60
A nim als .................................................................................................... ....... 6 1
T issue Processing ............................................................................................. 61









GAA Enzymatic Activity Assay ...................................................................... 61
Histological Detection of Glycogen ................................................................ 62
Determination of Vector Genome Copy Number............................................ 63
RNA Isolation and Analysis of Human GAA Transcript................................ 64
Immunoblot Detection of the Mannose 6-Phosphate Receptor....................... 64
Statistical A nalysis........................................................................................... 65
R esu lts ........................................................................................................................ 6 5
Liver GAA Activity and Antibody Formation after Intrahepatic Delivery of
rAAV(5)-DHBV-hGAA to Naive Gaad- Mice ............................................. 65
Diaphragm GAA Levels are Dependent on Both Liver GAA Activity and
A ntibody T iter .............................................................................................. 68
Neonatally Pretreated Gaa' Mice do not Form Antibodies to rAAV-Derived
hGAA after Low Dose Vector Delivery ...................................................... 69
Neonatal Tolerance is Broken after High Dose Delivery................................ 71
Lack of Cell-Mediated Immune Response After rAAV(5)-DHBV-hGAA
T reatm ent ............................................................................................ ......... 75
Restoration of GAA Activity in Tolerized Mice Improves Glycogen
Clearance.......................................................................... ........75
Human GAA Protein Observed in Corrected Tissues is Hepatic-Produced....... 78
Assessment of Hepatic Transduction Efficiency ............................................. 79
Poor Correction of Quadriceps Correlates with Lower Levels of
Mannose 6-Phosphate Receptor................................................................... 81
D discussion ............................................................................................................ 83
Impact of Immune Responses to Intrahepatic Delivery of rAAV- hGAA ......... 83
Biochemical and Histological Assessment of Cross-Correction ..................... 85
S um m ary ................................................................................................. ......... 87

5 DEVELOPMENT AND CHARACTERIZATION OF A GSDII MOUSE MODEL
THAT CONDITIONALLY EXPRESSES HUMAN GAA.................................... 88

B background .......................................................................................................... 88
M materials and M ethods................................................................................. .. ......... 89
Generation of the Transgenic Skin-hGAA / Gaa-' Strain................................. 89
Doxycycline Administration............................................................................ 90
Genotyping of Transgenic Lines by PCR........................................................ 90
GAA Staining of Tail Snips and Skin.............................................................. 91
T issue Processing ............................................................................................. 92
GAA Enzymatic Activity Assay ...................................................................... 92
ELISA for Detection of Anti-GAA Antibodies............................................... 93
R esu lts........................................................................................................................ 9 3
Generation of Skin-hGAA / Gaad- Mice .......................................................... 93
Human GAA Transcript Detected in Skin, Spleen and Stomach .................... 96
GAA Activity Detected in All Examined Tissues........................................... 97
Doxycycline-Mediated Repression of GAA Expression Leads to Gaa/_
P henotype ..................................................................................................... 99
Anti-GAA Humoral Response is Completely Inhibited in Skin-hGAA / Gad'
M ice ........................................................................................................... 102









Unexpected Observations of the Skin-hGAA / Gad'- Phenotype .................... 99
C onclusions........................................................................................................ 101

6 BIOCHEMICAL, HISTOLOGICAL, FUNCTIONAL, AND IMMUNOLOGIC
ASSESSMENT OF RECOMBINANT AAV(8)-MEDIATED CROSS-
CORRECTION IN TWO MURINE MODELS OF GSDII.................................. 105

B background ........................................................................................................ 105
M materials and M ethods.......................................................................................... 106
Cloning and Packaging of rAAV8 construct................................................. 106
Anim als and Vector Delivery ........................................................................ 107
Tissue Processing ........................................................................................... 107
GAA Enzym atic Activity Assay .................................................................... 108
ELISA Detection of Anti-GAA Antibodies and Isotypes ............................. 108
H istology........................................................................................................ 109
Western Blot Detection of GAA in Immunoprecipitated Serum................... 109
In Vitro Force Frequency Measurement ........................................................ 110
R esults............................................................................................... ............ ........ 111
Intrahepatic Delivery of rAAV8-DHBV-hGAA Vectors Leads to Higher
Levels of Liver GAA expression than Previously Observed........................ 111
Skin-hGAA/Gaa'- Mice are Tolerant to rAAV8-Derived hGAA................. 112
Recombinant AAV8-DHBV-hGAA Delivery to Pretreated Gaa-' Mice
Results in Early Formation of High Titer Antibodies................................ 112
Lack of Evidence For a Cell-Mediated Immune Response to
rAAV 8-Derived hGAA .............................................................................. 114
Partial Restoration of Heart GAA Activity Observed after
rAAV8-DHBV-hGAA Delivery ................................................................. 116
Superphysiologic Levels of Diaphragm GAA Result in Reversal of
Glycogen Accumulation after rAAV8-DHBV-hGAA Delivery................. 117
Low Levels of GAA Activity Observed in Hind-Limb Skeletal Muscles
Despite High Levels of Liver GAA in Tolerant Mice ............................... 118
110-kDa Circulating GAA Detected in rAAV8-Treated Mice...................... 121
Partial Restoration of Soleus Muscle Function in
rAAV8-DHBV-hGAA-Treated mice.......................................................... 123
Human GAA RNA from the rAAV8-DHBV-hGAA Vector is Expressed in
M multiple T issues ......................................................................................... 124
D discussion .......................................................................................................... 125
Relative Success of rAAV8-hGAA in the Liver ............................................ 125
Immune Responses to rAAV8-Derived Human GAA .................................. 126
Cross-Correction of Affected Tissues............................................................ 127
Correlation Between Human GAA Gene Expression and Cross-Correction ... 129
S um m ary ........................................................................................................ 129

7 CONCLUSIONS AND FUTURE DIRECTIONS................................................ 130

A PPE N D IX ................................................................................... 134









LIST OF REFERENCES.................................................................136

BIOGRAPHICAL SKETCH ................................................................................... 150















LIST OF TABLES

Table page

4-1. Acid ca-glucosidase activity and anti-GAA antibody titer 8 weeks after low dose
vector delivery to naive Gaa'- m ice. .................................................................... 67

4-2. Acid u-glucosidase activity and anti-GAA antibody titer 8 weeks after high dose
vector delivery to naive Gaad' m ice. ..................................................................... 67

4-3. Acid a-glucosidase activity and anti-GAA Ab titer 8 weeks after high dose
vector delivery to pretreated Gad'- mice.............................................................. 73

6-1. Acid a-glucosidase activity and anti-GAA antibody values after intrahepatic
delivery of rAAV8-DHBV-hGAA to two Gaa' mouse models ......................... 119















LIST OF FIGURES


Figure page

2-1. Cloned vector constructs .................................................................................... 33

2-2. In vitro comparison of 4 liver promoters and the ubiquitous CMV promoter....... 34

2-3. Summary of in vivo studies comparing liver promoters by portal vein delivery of
human GAA- expressing rAAV vectors in adult Gaa'- mice............................... 36

2-4. Weekly anti-GAA antibody formation in rAAV1- and
rAAV5-hAAT-hGAA-treated Gad'- mice ............................................................ 38

2-5. Exponential decay relationship between percent normal liver GAA activity
versus terminal anti-GAA antibody titers in rAAV-hAAT-hGAA treated mice..... 39

3-1. Anti-GAA antibody formation in pretreated and naive Gaad -/ mice to rhGAA
pro tein ...................................................................................................................... 4 9

3-2. Competition ELISA of anti-GAA antibody-containing serum............................. 50

3-3. Lymphocyte proliferation assay on splenocytes from pretreated and naYve Gaa'
m ice ......................................................................................................................... 52

3-4. Detection of anti-GAA isotype in naive Gad-' mice after rhGAA delivery ........... 52

3-5. Anti-GAA antibody titer in Gad' mice after pretreatment at different ages
followed by rhGAA challenge.............................................................................. 54

3-6. Diagram of in vitro GAA inhibition assay experimental design.......................... 56

3-7. Affects of anti-GAA antibody-containing serum on GAA activity in vitro ......... 56

4-1. Exponential decay relationship between diaphragm activity normalized to liver
activity versus anti-GAA antibody titer................................................................ 69

4-2. Anti-GAA antibody formation in naive and pretreated Gad' mice over time after
low dose vector delivery....................................................................................... 71

4-3. Anti-GAA antibody formation in naive and pretreated Gaa-' mice over time after
high dose vector delivery...................................................................................... 72









4-4. Summary of percent normal GAA activity 8 wk post-delivery of rAAV5-DHBV-
h G A A ...................................................................................................................... 74

4-5. Isotype analysis of anti-GAA antibodies formed against rAAV5-derived hGAA.. 76

4-6. Semi-quantitative analysis of rAAV5-hGAA vector genome copies.................... 77

4-7. Detection of glycogen by PAS staining on longitudinal sections of cardiac and
skeletal m uscle...................................................................................................... 78

4-8. RT-PCR detection of rAAV5-delivered hGAA RNA.......................................... 79

4-9. Immunhistochemical detection of human GAA-expressing hepatocytes............. 81

4-10. Immunodetection of the 220-kDA mannose 6-phosphate receptor...................... 82

5-1. Schematics of the tetracycline-mediated gene regulation system (tet-off) and
breeding strategy................................................................................................... 95

5-2. X-Gluc detection of GAA on tails of mice after PCR determination of genotype.. 96

5-3. Genetic and biochemical characterization of skin-hGAA / Gaa1- mice and their
single transgenic litterm ates. ................................................................................. 98

5-4. Biochemical and genetic affects of Dox-mediated repression ........................... 101

5-5. Anti-GAA antibody formation in skin-hGAA/Gaa''-, naive Gad- and neonatally
pretreated G ad m ice......................................................................................... 103

5-6. Photograph of skin-hGAA/Gaa-' mice with single transgenic littermates ......... 105

6-1. Anti-GAA antibody formation in pretreated Gaa mice (1-10) and skin-
hGAA IG aa-'- m ice (S 1-S6).................................................................................. 112

6-2. Hematoxylin and Eosin-stained section of liver 16 wk after
rAAV8-DHBV-hGAA-treatement. ...................................................................... 113

6-3. Relative levels of IgGI, IgG2b, IgG2a and IgE in pretreated Gad'- mice 12 wk
post injection of rAAV8-DHBV-hGAA ............................................................. 114

6-4. Acid a-glucosidase activity values in the 6 distal tissues of tolerant (Ab-) versus
im m une-responsive (Ab+) m ice......................................................................... 118

6-5. Detection of glycogen by PAS staining of diaphragm sections ......................... 119

6-6. Western blot analysis of hGAA in the serum of rAAV8-DHBV-hGAA-treated
m ice 16 w k post injection................................................................................... 121









6-7. Force-frequency relationship of intact soleus muscle 16 wk after intraheptic
delivery of 1 x 1012 vg rAAV8-DHBV-hGAA. ................................................... 122

6-8. RT-PCR of rAAV8-delivered hGAA.................................................................. 123















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

LIVER-DIRECTED STRATEGIES AND IMMUNOLOGIC CONSEQUENCES OF
RECOMBINANT AAV-MEDIATED CROSS-CORRECTION OF GLYCOGEN
STORAGE DISEASE TYPE II

By

Kerry Owens Cresawn

May 2004

Chair: Barry J. Byrne
Major Department: Molecular Genetics and Microbiology

Glycogen Storage Disease Type II (GSDII) is a lysosomal storage disease caused

by a deficiency in the lysosomal enzyme acid a-glucosidase (GAA). The disorder causes

cardiac and skeletal myopathy in infants and is fatal within 2 years of life. Currently,

there are no effective treatments for GSDII patients. The ultimate goal of viral vector-

mediated correction includes transduction of a repository of producer cells, such as

hepatocytes, with secretion and re-uptake of GAA by the affected muscles (in particular

the heart, skeletal, and respiratory muscles) via the mannose 6-phosphate receptor

(M6PR) pathway. Our initial studies using rAAV serotype 2 vectors in a null mouse

model of GSDII (Gaa-l-) demonstrated superphysiologic levels of hepatic GAA without

correction of affected tissues. After observing anti-GAA antibody formation in

rAAV2-treated Gaa'- mice, we generated two models of tolerance in Gaa' mice. The

first involved low-dose subcutaneous delivery of recombinant hGAA (rhGAA) to 1 day-









old Gad'- mice. The second involved generating a Gaa mouse model that expresses

skin-restricted human GAA under control of the tetracycline response element.

Both of these models were shown to be completely tolerant to repeated intravenous

injections of rhGAA at a dose that resulted in 100-fold elevation of anti-GAA antibody

titers in naive Gada mice. While neonatally-induced tolerance was broken in 66% of

mice after challenge with liver-directed delivery of IxlO12 particles of rAAV(5)-hGAA or

rAAV(8)-hGAA tolerance was sustained in 100% of the skin-hGAA-tolerized mice.

Results from these combined studies demonstrate the inhibitory affect of the anti-GAA

humoral response; as significantly higher levels of GAA activity were observed in the

heart, hindlimb muscles, and diaphragm of tolerant mice than of immune-responsive

mice. We also noted that the ability to correct the affected muscles is dependent on both

anti-GAA antibody titer and the level of liver-GAA expression, and that the degree of

impact of these two parameters varies among the different muscle types examined.

Overall, these results demonstrate the feasibility of liver-directed cross-correction of

GSDII using rAAV-mediated gene delivery in the presence of high-expressing vectors

and immunosuppression.


xvii














CHAPTER 1
INTRODUCTION

Glycogen Storage Disease Type II

Glycogen storage disease type II (glycogenosis type II, acid maltase deficiency,

Pompe disease) was first described in 1932 by J.C. Pompe after examining the heart of a

7-month-old girl who died suddenly 2 days after admission to the hospital. While

examining the girl's enlarged heart, Pompe noted massive accumulation of glycogen

within vacuoles of the heart as well as other tissues (97). Within the next 2 decades,

additional reports emerged on the association of glycogen storage with muscular

weakness, hypotonia, and cardiomegaly, with death by the age of 1. The biochemical

link in these observations was not understood until 2 landmark discoveries in the 1960s.

With De Duve's (24) identification and characterization of the lysosome as a membrane-

bound vacuole containing pH-active, hydrolytic enzymes, Hers et al. (47,67) were able to

identify the lysosomal acid a-glucosidase that can release glucose from glycogen; and

recognized that Pompe disease was due to the absence of this enzyme. Since these

findings, over 40 lysosomal storage disorders have been discovered.

Variants of GSDII

Glycogen storage disease type II (GSDII) is inherited as an autosomal recessive

trait. The frequency of GSDII varies among ethnic groups; and while the frequency in

the general population is still debated, estimates range between 1 in 300,000 and 1 in

40,000 live births (52). The clinical presentation of GSDII represents a continuum of

phenotypes that is, for the most part, correlated with age of onset, organ involvement and









degree of residual enzyme activity. The clinical phenotypes have traditionally been

divided into 3 groups.

Adult-onset GSDII is characterized by onset between the third and seventh decade

of life with residual enzyme activity levels around 15 to 20% of normal. These patients

present predominantly with progressive proximal muscle weakness, with truncal and

lower limb involvement. Respiratory insufficiency (which is ultimately the cause of

death) is apparent at the time of presentation in approximately 30% of patients and does

not develop until years later in others (52,109). The juvenile-onset form is characterized

by onset in the first to second decade, with residual enzyme activity levels around 3 to

10% of normal. Juvenile-onset patients present with progressive proximal muscle

weakness, respiratory muscle involvement, mild hepatomegaly and absent or mild

cardiomegaly. The course of juvenile-onset GSDII is more rapidly progressive than the

adult onset form. Death usually occurs before the end of the third decade, and (as in the

adult-onset patients) usually results from respiratory insufficiency. However, age of

death in juvenile-onset patients does not always correlate with age on onset (52,109).

The term Pompe disease generally refers to the third clinical phenotype of GSDII,

infantile-onset. Infantile-onset patients have complete or near complete enzyme

deficiency, and present from birth to 7 months of age (109). The clinical presentation is

that of a floppy baby with rapidly progressive skeletal muscle weakness, cardiomegaly,

respiratory muscle weakness, severe hypotonia, mild hepatomegaly, and macroglossia.

The progressive cardiomegaly involves left ventricular wall thickening, with obstruction

of left ventricular outflow (which can lead to further signals of hypertrophy).

Electrocardiograms reveal a shortened PR interval and large QRS complex (112,118).









The respiratory insufficiency caused by the weakened diaphragm is further complicated

by bronchial compression (caused by the enlarged heart). Pompe patients usually die

within the first 2 years of life of cardio-respiratory failure (52,106,109). Mental

development is normal, despite glycogen accumulation in the central nervous system.

Intellectual decline is not characteristic of the juvenile or adult-onset patients either

(106).

This continuum of diseases has leads to the observation that some patients may

present within the first 6 months of life with symptoms that are similar to traditional early

onset patients, but with less severe cardiomegaly (29,119). Within the group of patients

presenting in the first 6 months of life, there is an overall correlation of increasing age of

onset and absence of cardiomegaly. The less severe cardiac disease has allowed this

group of patients to survive past infancy with assisted ventilation.

Cardiac and Skeletal Muscle Involvement

Acid ax-glucosidase (GAA) is the only enzyme capable of hydrolyzing lysosomal

glycogen. Consequently, GAA deficiency results in lysosomal glycogen accumulation in

all tissues including liver, cardiac, skeletal and smooth muscle, kidney, skin, endothelial

cells, lymphocytes and in the peripheral and central nervous system (52,109). Despite the

global tissue involvement, the cardiac and skeletal muscles are the most susceptible to the

effects of glycogen accumulation. Several theories have been proposed to explain the

severe phenotype specific to these tissues, including the following:

* Lysosomes in the intrafibrillar spaces rupture due to spatial constraints, mechanical
effects of muscle contraction, or osmotic and/or destabilizing effects of glycogen
accumulation. Additionally, the upregulation of other lysosomal hydrolases, some
of which remain active at neutral pH, can exacerbate the effects of lysosomal
rupture by continuing the breakdown of the contractile apparatus (47).









* As the lysosomal membrane is forced to expand, the sarcoplasmic reticulum
membrane is recruited to form the expanding membrane; and its depletion from its
normal function results in abnormal calcium regulation.

* The pathway of some unidentified cellular protein is altered leading to the
accumulation of a toxic metabolite (106).

Acid Alpha-Glucosidase

Genotype-Phenotype Correlations

The gene encoding acid ca-glucosidase is 28 kb long, contains 20 exons, and is

localized to the long arm of chromosome 17 (17q25.2-q25.3) distal to the thymidine

kinase gene at the same localization (44,66). The GAA promoter has characteristic

"house-keeping" features contributing to the ubiquitous gene expression observed (98),

however, there is also evidence from in vitro studies of tissue-specific gene regulation

(56,92,105,139-141). The cDNA (Genbank accession numbers M34424 and Y00839) is

approximately 3.6 kb with 2859 coding nucleotides encoding a protein of 952 amino

acids (53,79). The amino acid sequence shares considerable homology with the amino

acid sequence of the sucrose-degrading enzymes human isomaltase and rabbit sucrase

isomaltase. This similarity includes identical sequence homology around the catalytic

sites, suggesting a common ancestral gene among these 3 enzymes (53,77).

Over 70 different mutations have been identified in the GAA gene. While most

mutations have been reported only in 1 or 2 patients, there is some association between

mutation and ethnicity (106). In general, the genotype is an accurate predictor of the

clinical phenotype. In the early-onset patients all mutations (nonsense, frameshift and

missense) result in lack of GAA transcription or transcription of unstable mRNA with no

detectable protein. It is rare that a mutation results in normal amounts of protein that are

catalytically inactive. Later-onset mutations involve homozygosity of a less deleterious






5


mutation; or heterozygosity for a fully deleterious mutation in combination with a milder

one (109). For example, the most common mutation in the late-onset Caucasian group is

a t to g transition in intron 1 (IVS1 t-13g) that has an allele frequency of 0.46 to 0.67

(54). This mutation results in the generation of several splice variants (including one in

which exon 2 is partially or completely deleted) that leads to complete loss of GAA

activity. However, this mutation is restricted to the later-onset phenotype due to the

presence of low levels of correctly spliced exon 2-containing mRNAs that allow for the

production of low levels of functional protein (106). Alternatively, the A exon 18

mutation results in deletion of the region crucial for maturation and processing of the

enzyme, leading to no residual enzyme activity. As a result, patients homozygous for the

A exon 18 mutation fall into the severe early-onset group. However, patients

heterozygous for IVS1 t-13g and A exon 18 have the less severe, later-onset phenotype

(106).

Recently, more cases have been reported of patients with a more severe genotype,

presenting with less severe, later-onset symptoms (such as a group of German adult-onset

patients homozygous for the A exon 18 mutation associated with the infantile-onset

phenotype) (130). The lack of correlation observed among genotype, degree of residual

activity, and severity of phenotype in some patients (45) suggests that genetic factors

such as genetic background, somatic mosaicism, presence of a second mutation or

polymorphism at the same allele, differential leakage of splice site mutations and/or

environmental factors (such as diet and exercise) could influence the time course and

severity of the disease (106,109).









Biochemistry and Cell Biology

Acid ac-glucosidase catalyzes the hydrolysis of alpha 1, 4- and alpha

1,6-glucosidic linkages of glycogen at an optimal pH of 4.3, resulting in the complete

hydrolysis of lysosomal glycogen to glucose. Muscle glycogen contains about 60,000

glucose residues per glycogen molecule, with alpha 1,4 linkages every 4-10 residues and

alpha 1,6-branches every 10 residues (145). The role of glycogen storage varies

depending on the tissue. Conversion of glycogen (stored in the brain) to glucose provides

emergency supplies during times of hypoglycemia or hypoxia. In the muscle (which

contains 3/4 of the total glycogen in the body), breakdown of glycogen to glucose is

necessary for short-term energy consumption (such as muscle contraction) (145).

Glycogen enters the lysosome either by fusion of glycogen-containing autophagic

vacuoles with the lysosome, to form secondary lysosomes (43); or by direct invagination

of the membrane. The signals responsible for entry of glycogen into the lysosome are not

known.

The GAA precursor is translated as a 110 kDa product, and further undergoes

extensive modifications that are characteristic of lysosomal enzymes. In the endoplasmic

reticulum, GAA is glycosylated at 7 potential sites, at least 2 containing high-mannose

chains (46). In the cis golgi, N-acetylglucosamine (GlcNAc)-1-phosphate is transported

to the high-mannose residues. The GlcNAc group is then removed by the enzyme, N-

acetylglucosamine-1-phosphodiester alpha-N-acetylglucoasminidase resulting in

exposure of the mannose 6-phosphate (M6P) groups (108). The precursor GAA protein

can subsequently be transported to the trans golgi, by binding of the M6P groups to one

of two distinct M6P receptors present on the trans golgi membrane. The M6PR-GAA









complex pinches off via clathrin coating on the external membrane to create

clathrin-coated vesicles containing the M6PR-bound 110 kDa precursor. As the clathrin

disassociates, the uncoated vesicle fuses with a low-pH sorting vesicle (called an

endosome). The low pH of the endosome causes the M6P receptor to dissociate from the

enzyme, allowing the receptor to be recycled either back to the trans golgi or to the cell

surface where it can bind and internalize exogenous GAA. The early endosome

containing the phosphorylated GAA finally fuses with the lysosome via an intermediate

compartment where the phosphate groups are removed. The 110 kDa precursor protein is

processed in the lysosomal compartments to yield an intermediate 95 kDa form and the

two mature, fully active, 76 and 70 kDa, forms of GAA (109).

While 80-90% of GAA in the trans golgi is directed toward this lysosomal

targeting pathway, 10-20% of the 110 kDa form is secreted from the cell (132). This

secreted, M6P-containing GAA can subsequently bind to the M6P receptor on the cell

membrane of the same cell or neighboring cells; and re-traffic to the lysosomes to

produce catalytically active protein. This 10-20% of GAA that enters the secretary

pathway provides the rationale for both enzyme and gene replacement therapies where

GAA-deficient cells can be corrected by uptake of phosphorylated GAA supplied

exogenously (via either intravenous delivery of the enzyme or from a repository of

producer cells).

Animal Models of GSDII

Animal models are valuable tools for studying disease pathophysiology, as well as

treatment modalities. Several naturally occurring animal models of GSDII have been

reported, including cat, Brahman and Shorthorn cattle, Correidale sheep, Lapland dog,

and Nicholas turkey (131). The cat, sheep and dog strains have not been maintained.









The Japanese quail model resembles the adult human form of the disease, and has been

the most extensively used (of the naturally occurring models) to study the efficacy of

various therapies (19). While the small size and obvious clinical phenotype make the

quail a relatively easy disease model, quails are evolutionarily too far distant from

humans; and their usefulness as an accurate disease model is limited by the existence of

two separate GAA-encoding genes (65).

As an alternative, 3 different genetically engineered GSDII mouse models have

been created by targeted disruption of the GAA gene. The two most-extensively used to

study gene or enzyme replacement therapies are the exon 13 knockout (13 4-), created by a

targeted disruption with a Neomycin gene (neo) cassette in exon 13 (7), and the exon 6

knockout (6neo/6 neo), generated by insertion of the neogene cassette into exon 6 (103).

Both of these knockout strategies result in no detectable protein and low levels of

detectable mutant GAA-neo transcript by RT-PCR. After observation of these two

strains for up to 7-9 months of age, both strains showed complete absence of GAA

enzyme activity, and marked glycogen accumulation. However, only the exon 6

knockout showed reduced mobility and muscle strength along with signs of muscle

weakness (104). One likely explanation for the varying phenotypes in these two models

is the different targeting exon suggesting that the targeted allele may be differentially

inactivated. Alternatively, the difference could be due to the backgrounds. While the

exon 6 knockout model is maintained on a 129SvJ x C57BL/6 background, the exon 13

knockout has been maintained on 129SvJ x C57BL/6 or 129SvJ x FVB background.

Raben et al.(104) noted that the same knockout generated on a 129SvJ x C57BL/6

background, backcrossed to the C57BL/6 strain, developed a more-progressive









cardiomyopathy with earlier-onset symptoms than when knockouts on the same

background were backcrossed to the FVB strain. While the heterozygosity in phenotype

(due to mixed genetic background) complicates characterization of the disease

pathophysiology, it does recapitulate the heterozygosity of the patient population,

allowing investigators to better anticipate the variegated responses to therapy. Because

the exon 6 knockout model has the more severe phenotype, it is the chosen model for our

studies. The biochemical phenotype of these mice recapitulates the infantile form of the

disease, allowing investigators to test both enzyme and gene replacement therapies for

restoring GAA activity. Additionally, our group demonstrated the ability to detect

glycogen accumulation from as early as 1 month of age (117); as well loss in skeletal

muscle function evident at 3 months of age (38). These functional phenotypes provide

appropriate outcome measurements for determining the therapeutic efficacy of enzyme

restoration.

Therapeutic Strategies for GSDII

Enzyme Replacement Therapy

Early attempts at enzyme replacement therapy (ERT) in the 1960s-1970s involved

administering unphosphorylated GAA from Aspergillus niger (5) or human placenta (23)

to infantile-onset patients. In both studies, an increase in enzyme activity was observed

in the liver, but there was no reduction in stored glycogen. Additionally, the protein

derived from Aspergillus niger was highly immunogenic, causing the study to be

terminated. In 1984, Reuser and co-workers (110) demonstrated that uptake by

fibroblasts of phosphorylated GAA from bovine testis and unphosphorylated GAA from

human placenta occurred via M6PR-mediated endocytosis. However, uptake of the

bovine testis GAA was 200-fold more efficient. These studies also showed that skeletal









muscle cells isolated from GSDII patients do express M6PR on the surface and can be

completely corrected by uptake of phosphorylated GAA. A later study showed that

uptake of dephosphorylated bovine testis GAA was not detectable (110,127). These

findings demonstrated that the extent of phosphorylation of mannose groups directly

correlates with uptake efficiency and paved the way for developing more-effective

recombinant enzyme for enzyme replacement therapy. Since these findings, recombinant

enzyme has been produced in Chinese hamster ovary (CHO) cell lines (62,129); the milk

of transgenic rabbits (8); and milk from transgenic mice (7). Transfer of enzyme activity

and glycogen clearance was observed in transgenic mice after administering purified

GAA from the milk of rabbits and mice. Similar results were also observed after

administration of CHO cell-derived GAA in the Japanese quail model (62).

These promising results led to several clinical trials around the world, including

two ongoing trials initiated by Genzyme Corporation (Cambridge, MA): One Phase II

trial for subjects less than 6 months of age, and one Phase I/II trial for subjects between 6

and 36 months of age (89). These trials are currently evaluating the efficacy of a

recombinant human GAA (rhGAA), isolated from CHO cell-lines and subsequently

modified in vitro to manipulate the degree of phosphorylation. While no data are

available on these patients (to date), results from preclinical studies in an immune-

tolerant tolerant GSDII mouse model show 75% clearance of skeletal muscle glycogen

and complete clearance of cardiac muscle glycogen at a dose of 100 mg/kg. However, at

a dose of 20 mg/kg, there was only partial cardiac muscle correction and no skeletal

muscle correction. When naive GSDII mice were given the 20 mg/kg dose, 100% of the









mice died from anaphylaxis after the seventh injection, demonstrating the potential need

for immunosuppression in the null patient population (101).

Gene Therapy

Gene therapy strategies offer the potential for providing a permanent, endogenous

source of a therapeutic protein both by direct correction of transduced cells and

cross-correction of distal cells, via uptake of the secreted protein after single

administration of vector. Many obstacles have thwarted progress toward clinical

application of gene therapy, including low and/or transient levels of gene expression;

immunogenicity of viral vectors (particularly the adenoviral-based vectors); germ-line

transmission of the vector; and the potential for vector integration into regions of the

genome crucial to homeostasis (resulting in events such as activation of oncogenes).

Recent advances in vector development and an understanding of the immune response to

vector and transgene have made gene therapy for GSDII a realistic goal for the future.

The first study to demonstrate correction of GAA-deficient cells from a viral

vector involved in vitro transduction of myoblasts and fibroblasts isolated from a Pompe

patient (146). Using a retroviral vector carrying the GAA cDNA, there was an increase in

GAA activity and a decrease in glycogen content in both of these cell types. More

importantly, the transduced cells were able to cross-correct GAA-deficient myoblasts by

secretion of GAA.

The first adenovirus studies for correction of GSDII performed by our group and

others (2,91,94,95) demonstrated the ability of recombinant El-deleted adenovirus

carrying the human GAA (hGAA) cDNA (rAd-hGAA) to yield expression of GAA in

deficient fibroblasts. It was also demonstrated that these cells could then serve as

producer cells of human GAA that could subsequently be recaptured by untransduced









acceptor cells via M6PR-mediated endocytosis (94). Furthermore, intrahepatic

administration of the same adenoviral vector carrying the murine GAA (mGAA) cDNA

(94)or hGAA cDNA to mice resulted in enzyme secretion and increase in enzyme

activity in cardiac and skeletal muscle (27). While these studies demonstrated the

feasibility of recombinant viral vector delivery of GAA, adenoviral-mediated gene

delivery has been shown to be highly immunogenic. Problems arising from adenoviral

delivery consist of a humoral response to both the transgene product and vector, and

potentially a cell-mediated response to the vector. Such a response was shown to be fatal

in one adenovirus-mediated gene therapy trial conducted at the University of

Pennsylvania in 1999 (36,58,107).

Non-viral gene delivery systems are being evaluated for treatment of genetic

disorders in hopes of avoiding the safety issues presented by viral vectors. Martinuik and

co-workers (78) have reported success using "gene gun" delivery of human GAA plasmid

DNA to Gaa~- mice. Although results were somewhat promising, expression was

transient, as is common with plasmid-based delivery methods; and repeated dosing was

required.

As safety is becoming the primary concern of viral-mediated gene delivery,

recombinant adeno-associated viral vectors (rAAV) are emerging as a favorable gene

delivery system. Fraites et al. (39) demonstrated that intramyocardial or intramuscular

injection of rAAV serotype 2 vectors carrying the murine GAA cDNA results in

restoration of GAA activity levels to near normal in the injected tissues, with partial

restoration of skeletal muscle function after intramuscular delivery. These studies are the

first and only (to date) to show GAA expression from a rAAV vector in GSDII mice.









However, it appears that higher levels of expression (requiring higher doses, stronger

promoters, and/or alternative serotypes) will be needed for successful cross-correction in

GSDII mice from a rAAV vector.

Recombinant Adeno-Associated Viral Vectors

Adeno-associated virus (AAV) is a member of the dependovirus genus of the

Parvoviridae family of small DNA animal viruses. As a dependovirus, AAV requires

helper virus (such as adenovirus or herpes virus) for a productive viral infection. In the

absence of a helper virus, AAV establishes a latent infection (6). AAV was first

discovered as a satellite contaminant in human and simian cell cultures infected with

adenovirus (6); and subsequently found as human, simian and avian isolates causing no

apparent disease (10). This lack of association with human disease makes AAV an

attractive vector for gene delivery. Other favorable characteristics include the following:

Recombinant AAV particles can be produced that completely lack AAV-encoding
sequences, contributing to the ability of these vectors to escape host-defense
mechanisms.

AAV can infect both dividing and quiescent cells.

AAV has a broad host and tissue tropism.

AAV particles are relatively stable (resistant to chemical and physical treatment),
which is important for vector production and commercialization (51,135).

AAV serotype 2 has been the most-extensively studied serotype; and the isolation

of a molecular clone for AAV2 by Samulski et al. in 1983 (113) allowed for the

development of recombinant AAV (rAAV) serotype 2 vectors for gene transfer. The

AAV2 genome consists of an approximate 4.7 kb of single-stranded DNA of plus or

minus polarity, including two 145 bp inverted terminal repeats (ITRs) located at the 5'

and 3' ends. The palindromic regions of the ITRs can form hairpin structures in the









single-stranded DNA, with the 3' hairpin acting as a primer for DNA replication as well

as a binding site for the Rep protein. This special structural feature of the AAV ITRs

makes them the only cis-acting viral sequences necessary for replication, packaging and

rescue (135). Therefore, recombinant AAV vectors can be produced by deleting viral

gene sequences and adding a therapeutic gene expression cassette. The wild-type AAV

viral proteins (3 structural Cap proteins and 4 nonstructural Rep proteins) must be

supplied in trans in order to make recombinant AAV virions (136).

Wild-type AAV2 has been shown to preferentially integrate into a region of the q

arm of human chromosome 19 (AAVS1) (63,86,114,136). This site-specific integration

is restricted to wild-type AAV2 as it is catalyzed by Rep proteins, which are absent in an

infection with recombinant AAV. The extent to which rAAV integrates into the host

genome is still under debate (85). The ubiquitously expressed heparin sulfate

proteoglycan receptor was shown to be the primary receptor for AAV2 (124), with the

fibroblast growth factor and (vP35 integrin acting as co-receptors (100,124). The use of

rAAV2 as a gene delivery vector for some diseases is limited by the delayed onset of

gene expression compared to other vector systems. While adenoviral gene expression is

detected within 1 week after delivery, gene expression from rAAV2 vectors takes

approximately 4 weeks. The primary rate-limiting step remains in discussion, with

evidence suggesting roles of conversion of the single-stranded genome to double-

stranded DNA (35) as well as the nuclear entry process (4).

Recombinant AAV Serotypes

As the gene therapy field advances and the number of disease applications for

rAAV-mediated gene delivery increases, interest in achieving higher levels of









tissue-enhanced transgene expression and earlier gene expression has become a focus of

AAV biologists, who have set out to isolate new AAV serotypes that can be developed

for gene therapy. As of March 2004, eight distinct AAV serotypes have been isolated

from primates, with 7 being distinct serotypes based on antibody cross-reactivity studies

(40). The standard approach to generating rAAV viruses of alternative serotypes is

pseudotyping, which involves packaging the gene expression cassette with AAV2 ITRs

in the capsid of the alternative serotype. Pseudotyping is a favorable approach to creating

recombinant AAV as there is more experience and information concerning the safety,

chromosomal integration efficiency, and specificity with AAV2 ITRs in animal models

and humans (150). Numerous studies have been performed to compare transgene

expression levels in specific tissues from different serotypes cross-packaged with AAV2

ITRs. Serotype 6 (which appears to be a recombinant of serotypes 1 and 2) and serotype

1 have both been shown to be far more efficient at muscle transduction than serotype 2

(15,111,134,135). Additionally, Fraites et al. (39) observed transgene expression from an

rAAV1 vector at 1 week post injection (3 weeks earlier than observed for rAAV2).

Serotype 5, isolated from a human condylmatous wart, improved transgene expression in

the liver by 3 to 10-fold compared to serotype 2 vectors (40,82). Serotypes 7 and 8,

recently isolated by Gao et al. (40) from monkey tissue, appear to have the greatest

divergence in capsid proteins; and are not neutralized by heterologous anti-AAV sera

raised against any other serotype. Serotype 7 (compared with serotypes 1 and 6) was

shown to have similarly high levels of expression in muscle. Serotype 8 (compared with

serotype 5) was shown to yield up to 15-fold higher transgene levels in the liver.









Liver-Directed rAAV Delivery

The liver is the largest gland in the body, comprising 2% of the total body weight.

Additionally, 1/5 of the total cardiac output passes through the liver each minute. The

large mass, high vascularity, and protein production and secretion efficiency make the

liver an ideal target organ for correction of multiple tissue beds after single-vector

delivery.

Delivery of rAAV vectors to the liver via intraportal or intravenous injection in

experimental animals has resulted in stable hepatocyte transduction, with therapeutic

levels of transgene expression in the circulation and/or target tissue

(14,16,22,48,59,60,84,120-122,133,138). Additionally, liver-directed delivery of rAAV

vectors is being evaluated in clinical trials for the treatment of hemophilia B (60).

Numerous studies over the past 10 years have answered important questions concerning

the potential for liver-directed, rAAV-mediated gene therapy. In 1997, Snyder et al.

(121) noted that only 2 to 5% of hepatocytes were successfully transduced 11 weeks post

portal vein injection; and that this number was sufficient to supply therapeutic

concentrations of transgene product. In 1999, Nakai et al. (85) observed (after portal

vein delivery of rAAV vectors) episomal forms of head-to-tail circular intermediates as

well as host-hepatocyte genome integration of head-to-tail high-molecular weight

concatemers at two different locations. In 2000, Miao et al. (80) demonstrated that while

only approximately 5% of hepatocytes express the rAAV-delivered transgene, the

rAAV genome is taken up by most hepatocytes. The authors also noted that the small

population of hepatocytes permissive to transduction is a changing population and that

cell cycling did not influence the cell's ability to undergo stable transduction.









Despite this small percentage of hepatocytes permissive to transduction by rAAV

vectors, high levels of transgene expression can be produced from the liver in a dose-

responsive manner. In 2002, Nakai et al. (88) reported that a 2-log range linear

dose-response curve of transgene expression was obtained from 3.7 x 109 to 3.0 to 1011

vector genome (vg) per mouse. Above 3.0 x 1011 vg/mouse there was a disproportionally

smaller increase in both the level of transgene expression and number of hepatocytes

transduced, with saturation at doses above 1.8 x 1012 vg/mouse. However, there was a

linear increase in the number of vector genomes per cell (up to 1.8 x 1012 vg/mouse).

These findings overall are important for understanding the feasibility of liver-targeted

rAAV transduction. The low percentage of hepatocytes susceptible to successful rAAV

transduction and loss of dose-response at higher doses may present obstacles for

achieving superphysiologic levels of hepatic transgene expression.

Liver-Directed versus Muscle-Directed Gene Therapy

An alternative target organ for cross-correction from a gene delivery vector is

skeletal muscle which is also highly vascularized and efficient at protein production and

secretion. Phase I clinical trials of liver- or muscle-directed gene transfer with human

factor IX-expressing AAV-2 vectors are currently ongoing (50,60). The liver-directed

trials for hemophilia B were halted for 3 months because of detection of AAV in the

gonads of treated patients (49). The increased risk of germline transmission of vector

after intrahepatic delivery versus intramuscular delivery was also demonstrated in mice

and rabbit studies (3). Along with this increased risk of germline transmission, another

potential disadvantage of the liver-directed approach is the invasive nature of portal vein

or hepatic artery delivery.









Despite these issues, there are two important advantages of intrahepatic delivery

versus intramuscular delivery: higher transgene expression levels and decreased

immunogenicity. Because the liver is naturally more efficient at protein production and

secretion, lower doses may achieve the same therapeutic effect in the liver as those seen

in the muscle after significantly higher doses (75). The only direct comparison of

intrahepatic versus intramuscular delivery in humans is a clinical trial that began in 2000,

evaluating delivery of rAAV-human factor IX (hFIX) by either direct muscle or portal

vein injection in hemophilia patients. At a dose of 2 x 1012 vg, patients receiving

intramuscular delivery failed to show any significant elevation in serum hFIX levels,

despite having tolerated the vector; while one intrahepatic patient had therapeutic levels

of serum hFIX. Unfortunately, these levels dropped within 1 week to sub-therapeutic

levels (51,76).

The lack of consistent and significant differences in the level of secreted

transgene product observed in liver versus muscle-directed rAAV gene therapy in mice

and/or humans encouraged Raben et al. (102) to generate two alternative Gaa-' mouse

models to compare the two approaches. In their study, two mouse models were generated

using the Gaa1 background: one expressing hGAA from the liver, and one expressing

hGAA from the muscle. In both cases, the hGAA gene was under control of the

tetracycline response element (TRE) promoter and could therefore be regulated in

response to administration of tetracycline in the food, water, or subcutaneous pellets.

Tissue-specific expression was conferred by the promoter used to drive expression of the

tetracycline transactivator (tTA) (the element that binds the TRE promoter to initiate

expression of hGAA). Therefore, these mice were transgenic for 3 genotypes: the Gaa "









genotype (as previously described); the hGAA gene, under control of the ubiquitously

expressed TRE promoter; and the tTA gene, under control of either the albumin or muscle

creatine kinase promoter. Raben et al. showed that cross-correction from liver-produced

hGAA required 12-fold normal liver GAA enzyme activity levels; whereas

cross-correction from muscle-produced hGAA required 100-fold normal muscle GAA

activity levels. While these data suggest that the liver is a more optimal target organ for

achieving cross-correction in GSDII mice, it should be noted that different tissues were

analyzed for comparing cross-correction outcome. Additionally, the different promoters

driving tissue-specific expression of the transactivator may have resulted in significantly

higher levels of transactivator, and potentially higher levels of hGAA in the liver,

although there is no evidence to support this.

The second potential advantage of intrahepatic delivery versus intramuscular

delivery is decreased immunogenicity of the transgene product as demonstrated by two

groups, both using rAAV-hFIX. In the first study by Ge et al. (42), rAAV vectors

expressing hFIX were delivered to the liver or muscle of immunocompetent mice at

identical doses. While all of the intramuscular-treated mice developed a robust humoral

immune response blocking hFIX expression in the serum, none of the intrahepatic-treated

mice elicited an immune response coinciding with significant levels of serum hFIX.

Results from this study suggest that the likelihood of an immune response was directly

related to the target tissues, rather than the level of protein expression. The reason for

increased immunogenicity observed after intramuscular delivery is not fully elucidated.

However, naked DNA delivery studies have demonstrated that expression of transgene

product from muscle leads to binding of the protein to the extracellular matrix. This









results in high levels of locally concentrated protein which may lead to more efficient

cross-presentation of antigen to antigen-presenting cells (21,28,126). Because more of

the total transgene product is secreted from the liver than from the muscle, the liver-

produced transgene product may not reach high enough local levels of antigen, and may

therefore remain undetectable by the immune system (57).

In the second study by Mingozzi et al. (81), immune tolerance was induced in

immunocompetent mice by intrahepatic administration of rAAV-delivered hFIX.

Tolerance of hFIX was challenged by administration of recombinant hFIX protein in

adjuvant after pretreatment with intrahepatic delivery of either rAAV-hFIX or

rAAV-GFP. Mice pretreated with rAAV-hFIX developed no anti-hFIX antibodies and

had reduced in vitro T cell responses. However, mice pretreated with rAAV-GFP

developed anti-FIX antibodies 14 days after injection of rhFIX along with significantly

higher in vitro T cell responses. Results from their study and others (16) suggest that the

combination of intrahepatic delivery of high levels of sustained expression of a

rAAV-expressed foreign transgene is sufficient to evade the immune response; and also

induce tolerance to subsequent challenge by recombinant protein. The advantages and

disadvantages of intrahepatic versus intramuscular delivery will have to be weighed for

each disease, and will likely depend on factors such as immunologic background,

minimal level of expression required for correction, and any associated diseases that may

complicate targeting of a specific tissue (such as the increased incidence of hepatitis

observed in hemophilia patients) (51).

Immune Response to Vector-Derived Transgene Products

Gene transfer, in general, has been shown to be an efficient way to elicit humoral

and cellular immune responses to foreign proteins in the development of vaccines for









diseases such as cancer (25,69,83), HIV (137), and HSV (74). While this immunologic

property is an attractive feature for vaccine development, it limits the application of gene

transfer for the treatment of diseases where prolonged transgene expression is necessary.

Immune responses to recombinant adenoviral-derived transgene products have been the

most extensively studied. Vigorous cellular and humoral immune responses to

recombinant adenoviral-derived cytosolic beta galactosidase (13-gal) was observed

following delivery to the liver, lung, and muscle and (in many cases) led to the

destruction of transduced cells, inflammation, and loss of the lacZ transgene expression.

(41,128,142-144).

In effort to understand earlier studies demonstrating that rAAV vectors did not

elicit immune responses to foreign transgene products (while delivery of the same protein

by recombinant adenovirus and naked DNA did), Jooss et al. (58) evaluated the

mechanisms by which rAAV evades immunologic response. These studies found that

adoptive transfer of dendritic cells isolated from murine spleen (and transduced by

rAAV-lacZ) did not elicit transgene-specific immune responses in vivo. Transduction of

the same cell type by adenoviral-lacZ vectors, followed by adoptive transfer, did lead to

immune-mediated elimination of transduced cells. These results were explained by the

fluorescent in situ hybridization experiments showing rAAV genomes in the perinuclear

space of dendritic cells (compared to the intranuclear localization of adenoviral vectors).

The conclusions from this important study were that AAV can enter dendritic cells, but

post-entry blocks interfere with efficient transduction and antigen presentation. Later

studies suggested that the efficiency of dendritic cell transduction, by rAAV, depends

heavily on the number of dendritic cells and maturation state of the dendritic cell









population at the time of rAAV encounter (147). It is not known whether the number of

immature dendritic cells required to elicit a cell-mediated response after rAAV-LacZ

transduction in vitro is realistic in vivo. The general consensus (to date) concerning

rAAV and immunity is that rAAV vectors are incapable of recruiting enough immature

antigen presenting cells to the site of vector administration to lead to T cell activation

(57). These important studies were all performed using rAAV serotype 2 vectors, and

similar studies will need to be done with alternative serotypes.

While the studies described above suggest that rAAV-mediated gene delivery is

not likely to result in cell-mediated immunity to transgene products, the potential for a

humoral immune response to rAAV-derived and secreted transgene products remains a

considerable concern for many investigators. The likelihood of rAAV-derived transgene

products eliciting a humoral immune response depends on several factors, including the

genetic background of the host; amount of foreign protein secreted by transduced cells;

presence or absence of residual mutant protein (13); structural differences between the

replacement and mutant protein; and the route of vector administration (12,20,81).

While delineation of the immune response to secreted transgene products is

complicated by the presence of both the vector and therapeutic protein as potential

antigens, several in depth studies have been able to elucidate the mechanisms of immune

activation of rAAV-derived foreign proteins (33,34,42). Quantitative assays for detecting

various cytokines, isotypes of B-cell secreted antibodies and cytolytic activity, along with

histological detection of lymphocyte infiltration can all be used to further understand the

immunologic consequences of any treatment. T lymphocytes consist of both T helper

(Th) lymphocytes and cytotoxic T lymphocytes (CTLs). Within the Th class exist Th









and Th2 cells. The Th 1 subset characteristically secretes IL-2 and IFN-y and is capable

of stimulating both proliferation of cytotoxic T lymphocytes (leading to destruction of

antigen-expressing tissues) as well as proliferation of B cells secreting antibodies of the

IgG2a isotype. The Th2 subset characteristically secrete IL-4 and IL-10 and can activate

B cells leading to production of antibodies of the IgG1 and IgG2b, and IgE isotype. In

the context of rAAV-mediated delivery of therapeutic transgenes, Fields et al. (34) found

that the immune response to secreted factor IX after rAAV-mediated delivery was

characterized by activation of the Th2 subset of T helper cells (determined by cytokine

profile). This resulted in the production of predominately IgG1 antibodies and absence of

Th subset activation coinciding with lack of both CTL stimulation and inflammatory

cell infiltrates. In contrast, factor IX delivery by recombinant adenoviral vectors resulted

in activation of both Th2 and Thi cells. This resulted in IgG2a antibody formation, CTL

stimulation, inflammation and rejection of transduced muscle fibers. In a later study,

Sarukhan et al. (34,116) demonstrated that rAAV-derived transgene products can elicit a

cell-mediated immune response leading to destruction of transduced cells provided the

foreign gene product is highly immunogenic (such as the influenza virus hemagglutinin

(HA) protein). Consistent with the previous study, these authors found that while the

immune response to rAd-HA resulted from direct transduction of dendritic cells as well

as cross-presentation of transgene product, the immune response to rAAV-HA delivery

was only activated by the latter mechanism.

Summary

Clinical evaluation of enzyme replacement therapy for the treatment of GSDII is

underway. However, long-term goals for GSDII treatment involve gene replacement









therapy using rAAV-mediated delivery of GAA. To understand the full potential of this

therapy, we evaluated one of the proposed gene therapy approaches: rAAV-mediated

delivery of GAA to the liver to achieve cross-correction of a mouse model of GSDII.

Preliminary results from this proposal led us to investigate the potential for and

consequences of an immune response to rAAV-derived GAA and determine the

necessary elements for achieving varying levels of correction in GSDII mice. These

findings provide us with both a new appreciation for the potential pitfalls of gene

replacement therapy for GSDII (as well as other similar diseases) and an understanding

of how these obstacles can be overcome.














CHAPTER 2
PRELIMINARY STUDIES TO EVALUATE LIVER-DIRECTED DELIVERY OF
RECOMBINANT AAV VECTORS EXPRESSING HUMAN GAA TO GSDII MICE

Background

Achieving high levels of sustained transgene expression is a critical component for

the success of gene therapy. The level of transgene expression necessary to reverse the

disease pathophysiolgy varies for each disease. In the case of GSDII, adult-onset patients

have approximately 20% of normal GAA activity. With this level of GAA, these patients

have a relatively mild presentation with no cardiac involvement. It has therefore

generally been accepted in the field that restoring at least 20% of normal enzyme activity

to infantile-onset (Pompe) patients would likely ameliorate most of the symptoms

associated with the Pompe patients, particularly those associated with cardio-respiratory

failure (109). However, the ability to restore cardiac and respiratory function with this

percentage of normal enzyme activity will depend on the age at treatment and severity of

phenotype and may be different between the GSDII mouse model and human patients.

While this level of enzyme restoration (20%) is relatively low, achieving this in

non-transduced tissue will require high, yet undefined, levels of GAA expression in the

transduced tissue (particularly because only 10-20% of GAA in transduced cells is

secreted for uptake by distal cells). The aim of the following studies was to optimize

hepatic GAA expression by the use of high-expressing, liver-specific promoters. After

confirming hepatocyte-specific activity in vitro from four promoters using the

P-galactosidase (f-gal) reporter gene, 3 promoters were chosen for evaluating the ability









to drive high levels of hepatic GAA expression after intrahepatic delivery of recombinant

AAV (rAAV) vectors in Gaa mice. Candidate promoters were restricted to a size of no

greater than 1.0 kb in order to produce a final plasmid of less than 4.9 kb, the upper limit

for packaging of rAAV vectors.

Materials and Methods

Molecular Cloning of DNA Constructs

The 3.1 kb human GAA (hGAA) cDNA was constructed as described (95). The

full-length cDNA was first subcloned under the transcriptional control of the

cytomegalovirus (CMV) immediate early promoter in the mammalian expression

plasmid, pCI (Clonetch, Palo Alto, CA), to yield pCI-hGAA. The expression cassette

was subcloned in the p43.2 plasmid containing the two rAAV serotype 2 inverted

terminal repeats (ITRs) via EcoRI-Xbal digestion to yield p43.2-hGAA. The pGHP3

plasmid containing 350 bp of the Duck Hepatitis B Virus (DHBV) core promoter was

constructed as described (68) and received as a gift from C. Liu (University of Florida,

Gainesville, FL). The pAT2 plasmid containing the 950 bp albumin promoter and 3

upstream hepatocyte nuclear factor-1 enhancer (HNF-1) elements was constructed as

described (70) and received as a gift by K. Zarret (Fox Chase Cancer Center,

Philadelphia, PA). The phAAT-hAAT del IRES (p547) containing the 400 bp human

alpha-1-antitrypsin (hAAT) promoter and the pTTR-CAT (p481) construct containing the

300 bp human transthyretin (TTR) promoter and enhancer were received as gifts from K.

Ponder (Washington University, St Louis, MO). The DHBV, albumin, hAAT and TTR

promoters were all subcloned from the parent vector into the p43.2hGAA plasmid by the

following methods: 5' BglIH and 3'HindlII sites were added to the promoter sequence by

PCR and the amplified product was subcloned into the PCR2.1 Topo Cloning vector









(Invtirogen, Carlsbad, CA). The Bglll-Hindlll-flanked promoters were subsequently

subcloned in the p43.2-GAA plasmid via BgllI-HindJII digestion to yield the following

pTR-Liverp-hGAA constructs: pTR-DHBV-hGAA, pTR-hAAT-hGAA and

pTR-Alb-hGAA. A schematic of the constructs is shown in Figure 2-1. The pTR-LacZ

plasmids (pTR-DHBV LacZ, pTR-hAAT LacZ, pTR-Alb LacZ, pTR-TTR LacZ) were

constructed by replacing the hGAA cDNA in the pTR-Liverp-hGAA plasmids with the

LacZ cDNA from the pAAV-LacZ plasmid described previously (61) via NheI-NotI

digestion.

DNA Sequencing of Cloned Constructs

DNA samples of cloned constructs along with sequencing primers flanking the

promoter regions were sequenced using the fluorescent dideoxy terminator method of

automates sequencing by to the Interdisciplinary Center for Biotechnology Research

DNA Sequencing Core at University of Florida.

Packaging and Purification of rAAV Vectors

Recombinant AAV2 vectors and rAAV pseudotype vectors (rAAV2 rep / 5 cap

and rAAV2 rep / 1 cap) were packaged by the University of Florida Powell Gene

Therapy Vector Core as previously described (136,149,150). Purified rAAV stocks were

characterized by SDS/polyacrylamide gel electrophoreses with silver-stain and particle

count. Total particle titer was determined by dot blot method as previously described

(149). The infectious titer of rAAV2 serotype vectors was determined by infectious

center assay (149).

In vitro Analysis of Liver Promoters Driving LacZ Expression

Cultured cells were maintained in humidified air, 5% CO2 at 370C. HepG2 (ATCC

#HB-8065) or HEK 293 (ATCC #CRL-1573) cells plated at 80% confluency in 12-well









plates were transfected with 2 plg pTR-Liverp-LacZ or p43.2-CMV-LacZ plasmids using

Lipofectamine PLUS Transfection Reagent (Invitrogen, Carlsbad, CA) according to the

manufacturer's instructions. Mock-treated wells were transfected with lipofectamine

reagent alone. Cells were harvested from the dish in 500 pL of lx phosphate buffered salt

solution (PBS) at 48 hrs post-transfection. The cells were centrifuged and the cell pellet

was resuspended in 1X PBS. Cell lysates were subjected to three freeze-thaw cycles and

centrifuged. Clarified lysate (10 gtL) was used in the Galacto-Star chemiluminescent

reporter gene assay system for detection of P3-galactosidase activity (Tropix Inc., Bedford,

MA.). Protein concentrations of cell lysates were determined using a standard Bradford

assay (11) based the binding properties detection of Coosmassie Brilliant Blue G-250 dye

reagent (Bio-Rad, Hercules, CA). The Coomassie dye allows for an increasingly intense

blue color to be detected at 620 nm wavelength with increasing amounts of total protein

in the well. A standard curve was generated using seven concentrations of bovine serum

albumin and corresponding A620 values.

Animals

The Gaa-- mouse model used in these studies was generated from a targeted

disruption of exon 6 and is maintained on a mixed C57BL/6 x 129X1/Svj background, as

described previously (103). Age-matched C57B6/129 were used as controls. All mice

were housed in the University of Florida SPF animal facility and all animal procedures

were done in accordance with the University of Florida's Institutional Animal Care and

Use Committee approval guidelines









Intrahepatic Recombinant AAV Delivery Methods

Animals were anesthetized with 2% isoflurane and restrained supine on a warmed

operating surface. A midline incision was made and the abdominal skin and muscles

were retracted. To expose the portal vein, the intestines were moved out of the

abdominal cavity, covered with sterile gauze sponge and irrigated with warm saline

(0.9%). Using a 29-gauged tuberculin syringe, 100-200 lpL rAAV-hGAA diluted in

Lactated Ringer's Saline (Baxter Healthcare Corporation, Deerfield, IL) was injected into

the portal vein. Pressure was applied to the portal vein using a cotton applicator to

prevent bleeding. The intestines were placed back into the abdominal cavity. The

abdominal muscles were sutured using 5-0 prolene. The skin was closed using 5-0 vicryl.

Prior to awakening from the anesthetic, the animal received 500 pL subcutaneous fluids

consisting of 0.9% sodium chloride, ampicillin (50 mg/kg) and buprenorphine (0.1

mg/kg).

Tissue Processing

Eight weeks post injection, animals were euthanized by intraperitoneal injection of

pentobarbital sodium solution (150 mg/kg) and extracted tissues (liver, heart, quadriceps,

and diaphragm) were immediately cut and frozen in liquid nitrogen. Quadriceps muscle

was first crushed to a powder using a hemostat in liquid nitrogen prior to

homogenization. Tissues were homogenized in sterile water on ice using a PowerGen 35

Homogenizer (Fisher Scientific, Pittsburg, PA) and centrifuged. Homogenized

supernatants were subjected to three freeze-thaw cycles and centrifugation.

GAA Enzymatic Activity Assay

Enzymatic activity assay for GAA was determined by measuring cleavage of the

synthetic substrate 4-methylumbelliferyl-a-D-glucoside (4-MUG) (Sigma M9766,









Sigma-Aldrich, St Louis, MO). Cleavage of 4-MUG by active GAA results in a

fluorescent product, 4-methylumbelliferone (4-MU), that emits a fluorescent signal at 448

nm when excited at 360 nm. A 75 mM 4-MUG stock solution is prepared with dimethyl

sulfoxide (DMSO). Twenty paL of clarified cell lysate was added to one well of a black

96-well plate. To each well, 1.6 pL of 75mM 4-MUG diluted in 38.4 pgL of 200 mM

sodium acetate (pH 3.6) was added. The plate was then incubated at 370C for 1 h. The

enzyme reaction was stopped by adding 200 paL of 0.5 M sodium carbonate (pH 10.7).

Six standards of 4-MU (Sigma-Aldrich) were prepared from a 1 mM stock of 4-MU

diluted in sterile water. Twenty p.L of each standard were added to individual wells

along with 40 pl sodium acetate and 200 p.L sodium carbonate. Fluorescence was

measured at 460 nm using the Flx800 microplate fluorescence reader (Bio-Tek

Instruments, Winooski, VT). A standard curve was generated using a linear regression of

the 4-MU standards ranging from 0 pgm to 500 pam.

Protein concentrations were determined using the Bio-Rad DC Protein Assay kit

(Bio-Rad, Hercules, CA). Clarified lysate was diluted 1:5 or 1:20 depending on the

tissue type. Standards ranging from 0 to 5 tpg of bovine serum albumin (BSA) were

prepared in sterile water. Five pl of diluted lysate or standard were added to individual

wells of a 96-well plate along with 225 pL total volume of Bio-Rad Coomassie Brilliant

Blue dye reagents. After 15-minute incubation, the plate was read at 750 nm. A standard

curve was generated using a linear regression of the BSA standards. For both the activity

and protein assays, samples that generated fluorescence or absorbance values outside of

the linear range were subsequently diluted and re-calculated. GAA activity was

calculated as nmol 4-MUG cleaved per hour per milligram of total protein in the lysate.









Data are represented as the percent of normal GAA activity observed in tissues from

age-matched and sex-matched normal (C57BL/6/129) mice processed and assayed

side-by-side with treated mice.

ELISA Detection of Anti-GAA Antibodies

Serum samples were obtained weekly by tail vein bleeds of anesthetized animals.

Whole blood was collected in Microtainer tubes (Becton Dickinson, Franklin Lanes, NJ),

spun at 5500 rpm for 10 min and serum was removed and stored at -200C. Immulon

microtiter plates (Thermo Labsystems, Franklin, MA) were coated overnight at 40C with

200 l1 of 0.5 mg/mL human GAA in 0.1M NaHCO3 (pH 8.2). Wells were washed 3

times with 300 AL PBS containing 1% Tween20 (PBS/T) and blocked with 300 ApL 10%

fetal bovine serum (FBS) in PBS/T for 2 h at room temperature. Wells were washed 3

times and serum samples (diluted 1:80 in the blocking reagent) were added to the wells in

a total volume of 100 tl. Serial dilutions of rabbit-anti-human GAA antibody (from

1:500 to 1: 5 x 106) were used to derive the standard curve. Samples and standards were

incubated for Ih at room temperature. Washing was repeated and 100 AtL sheep of

anti-mouse IgG-HRP linked antibody or donkey-anti-rabbit-IgG-HRP-linked antibody

(Amersham Pharamcia Biotech, Piscataway, NJ) diluted 1:10,000 was added to the

sample wells or standard wells respectively and incubated for 30 min at room

temperature. After incubation, washing was repeated and 100 il of Tetramethyl

benzidine (Sigma-Aldrich) was added to the wells for 1-3 minutes. The reaction was

stopped with 100 pl of IN H2SO4 and absorbance was measured at 450 nm. A standard

curve was generated (to control for assay variability) with the dilution of rabbit

anti-human GAA antibody as the abscissa and the corresponding A450 values as the









ordinate (R2=0.95-0.98). Standardized values for serum of treated mice were divided by

mean standardized values for serum from untreated Gaa- mice (n=3). Final values were

reported as fold over background anti-GAA antibody titer.

Histology

Tissues were fixed in 10% buffered formalin and embedded in paraffin and

sectioned. Sections were stained with hematoxylin and counterstained with eosin (H&E).

H&E-stained liver sections were examined for histopathology by a veterinary pathologist

at the University of Florida Pathology Core Lab.

Statistical Analysis

Unpaired Student's t-test (Sigma Plot 2001, SPSS Inc, Chicago Illinois) was used

for statistical analysis comparing treatment groups with statistical significance considered

if P<0.05.

Results

In Vitro Analysis of Four Liver-Specific Promoters

The relative strengths and hepatic-specificities of four liver-specific promoters

(albumin, hAAT, DHBV, and TTR) were evaluated in vitro by transient transfection of ]3-

galactosidase expressing plasmids (Figure. 2-1). The LacZ reporter cDNA was used in

place of the hGAA cDNA for these studies due to the high background levels of GAA

activity in hepatocyte cell lines. The p43.2-CMV- LacZ plasmid was used as a positive

control as the CMV promoter is known to yield high levels of expression in most cell

types in vitro.








ITR Promoter intron human GAA or LacZ A(n) ITR
CMV
750 bp

j- Albumin D
1.0 kb


400 bp


370 bp


300 bp


Figure 2-1. Cloned vector constructs. All constructs contain the 5' and 3' rAAV2 ITRs,
promoter (with size designated below),'chimeric intron, the human GAA
cDNA (used for in vivo studies) or LacZ cDNA (used for in vitro studies) and
a poly A tail. Drawings are not to scale.

After transfection of the human hepatoma cell line, HepG2, the TTR (transthyretin)

promoter yielded the highest levels of 0-galactosidase activity followed by the hAAT

(human a-l-antitrypsin) promoter (Figure 2-2). Additionally, all four liver promoters

yielded undetectable or very low levels of 0-galactosidase activity in the HeLa cell line

(a cervical cancer cell line consisting predominately of epithelial cells). While these

results were valuable in confirming hepatocyte-enhanced activity from all four

promoters, it has become apparent, based on work from our laboratory and others

(unpublished) that accurate predictions of transgene expression from various gene

regulatory elements in animals cannot be made based on in vitro transfection data. Three

promoters, DHBV, albumin and hAAT were chosen for in vivo analysis of hGAA

activity. The TTR promoter, which yielded the highest transgene levels in vitro was not









evaluated for in vivo studies at this time based on consistent results from other (obtained

after our in vitro studies were performed) demonstrating poor activity in vivo (99).

7e+5

6e+5

5e+5
0
4e+5

3e+5

_j 2e+5
cc:
le+5

Albumin hAAT TTR DHBV CMV

mmm HepG2
HeLa

Figure 2-2. In vitro comparison of 4 liver promoters and the ubiquitous CMV promoter
by transient transfection of LacZ expressing constructs in HepG2 and HeLa
cell lines. Background luminescence in mock-transfected wells was subtracted
from all values. Bars represent mean standard error of the mean (n=3 wells
per plasmid).

Superphysiologic Levels of Liver GAA Expression from the rAAV2-DHBV-hGAA
Vector are not Sufficient to Restore Activity in Affected Tissues of Gaa'"1 Mice

The albumin promoter has been one of the most well characterized liver-specific

promoters and numerous studies have reported high levels of transgene expression under

the control of this promoter (64,73,115,133). While the DHBV promoter was also shown

to have strong hepatic cell type specificity in tissue culture (68), it had not been

previously tested as a candidate promoter for gene delivery studies. In effort to

determine whether the DHBV promoter could yield higher levels of hepatic transgene

expression in vivo than the well-studied albumin promoter, the pTR-Alb-hGAA and pTR-

DHBV-hGAA constructs were packaged in rAAV serotype 2 viruses and administered via

intrahepatic delivery to 8 wk-old Gaa-' mice. Eight weeks after portal vein delivery of









3 x 1012 vector genomes (vg) of rAAV2- hGAA, liver GAA expression was 6-fold greater

from the DHBV promoter than from the albumin promoter (180 21.7% versus 30 +

3.4% of normal; P=0.003) (Figure 2-3). Despite these superphysiological levels of GAA

observed in the livers of mice receiving the rAAV-DHBV-hGAA vector, we observed no

increase in GAA enzyme activity levels in the cardiac or skeletal muscle (not shown).

At this time, the use of alternative rAAV serotypes was reported and results

consistently demonstrated that rAAV serotype 5 yielded significantly higher transgene

levels in the liver than serotype 2 (82). However, results comparing the relative strength

of serotype 1 versus serotype 2 in the liver were not consistent. The second in vivo study

was designed to evaluate rAAV-mediated delivery of human GAA, under control of the

hAAT promoter in Gaa-' mice using serotypes 1 and 5.

Recombinant AAV Serotype 5 Vector Yields Higher Liver GAA Levels than
Serotype 1 Vector

After portal vein delivery of 5 x 10" vg of rAAV1-hAAT-hGAA or

rAAV5-hAAT-hGAA to 8 wk old Gaa' mice, liver GAA activity levels were 2-fold

higher with rAAV5 (P=n.s.) (Figure 2-3). However, these levels (30.9 11.0% of

normal) were again insufficient to restore GAA activity in cardiac or skeletal muscle.

Although comparison of the DHBV promoter used in the serotype 2 study and the hAAT

promoter used in this study is complicated by both the dose and serotype variables, if the

doses are accounted for (3 x 1012 vg and 5 x 10" vg respectively), expression from the

hAAT promoter appears to be equal to expression from the DHBV promoter. That is,

both the dose and liver GAA levels from rAAV2-DHBV-hGAA vector are 6-fold greater

than from the rAAV5-hAAT-hGAA vector. However, data from numerous groups

reporting that liver transgene levels from rAAV5 vectors are consistently 5 to 10-fold









greater than from rAAV2 vectors (40,82) suggests that the promoter strengths may not be

equal. Future studies would benefit from evaluating the DHBV promoter using the

rAAV5 vector.

250


200
E
S150-
0


2_ 100


50



Promoter Albumin DHBV hAAT hAAT
Serotype 2 2 5 1
Dose (P) 3 x 1012 3 x 1012 5 x 1011 5 x 1011


Figure 2-3. Summary of in vivo studies comparing liver promoters by portal vein delivery
of human GAA-expressing rAAV vectors in adult Gaa- mice. Background
activity in untreated Gaa- mice was subtracted from all values. Bars
represent mean standard error of the mean (n=3 for albumin and DHBV
vectors and n=5 for hAAT vectors) P=<0.01.

Gaa-d' Mice Elicit a Humoral Immune Response to Vector-Derived Human GAA

Before planning future studies to optimize hepatic GAA expression, it was

necessary to determine if lack of cross-correction was solely due to insufficient levels of

hepatic GAA expression or if Gaa-d mice were forming anti-GAA antibodies that could

potentially inhibit uptake of liver-secreted GAA by other tissues. Serum samples

collected weekly from rAAV1- and rAAV5-hAAT-hGAA treated mice were assayed for

the presence of anti-GAA antibodies by ELISA. Formation of anti-GAA antibodies was

observed beginning at 5 weeks post injection (Figure 2-4). Variability in the degree of









immune response was observed between mice from both groups. In the rAAV1-treated

group, 2 of 5 mice had elevated antibody titers peaking at 2.4 and 3.1-fold over

background, respectively, while the remaining 3 had lower titers of 1.2 0.06-fold over

background. Antibody titers overall (as well as the degree of variability) were lower in

the rAAV5-treated group, with 2 of 5 mice having elevated antibody titers of 1.6- and

2.2-fold over background, and the remaining 3 having background titers.

The significance of these values was difficult to assess, as there was no cross-

correction in any of the mice, regardless of their immune response. However, we did

observe an overall inverse relationship between liver GAA activity and antibody titer

(Figure 2-5). In mice with liver-GAA values of < 10% of normal, anti-GAA antibody

titers were between 2 to 4-fold over background. In contrast, mice with higher levels of

liver GAA expression (between 16% to 140% of normal) had anti-GAA antibody titers

near background. The exponential decay relationship of these two variables suggests the

possibility of a threshold effect where liver GAA expression values greater than 10% of

normal are able to induce immune-tolerance. While the mechanism behind the idea of

high dose tolerance has not been elucidated, it has been demonstrated by others studying

liver-directed, rAAV-mediated gene replacement for hemophilia and Fabry disease

(81,148).

Alternatively, the negative relationship observed between liver GAA activity and

anti-GAA antibody titer could suggest the presence of a cell-mediated immune response

(CMI). If higher antibody titers are an indication of a CMI response in these mice, then

we would expect to see a loss of liver GAA activity due to destruction of transduced

hepatocytes by cytotoxic T-lymphocytes in these mice. To begin evaluating this theory,









liver tissues from rAAV-hAAT-hGAA-treated mice and saline-treated mice were

sectioned, stained with hemotoxylin and eosin (H&E), and evaluated for histopathology

(not shown). While no signs of infiltration or other hepatocellular changes were observed

in the terminal liver tissues, it is theoretically possible that an earlier cytotoxic response

would have been cleared and become undetectable at 8 wk post injection. More in depth

immunologic studies will be necessary in future experiments to confirm this.


2.8 AAV1 (n=2)
o- AAV1 (n=3)
2" 2.4 AAV5 (n=2)
3 -9 AAV5 (n=3)
< )2.0

0 1.6

< 1.2


0.8
0 2 4 6 8 10
Time
(weeks post-injection)
Figure 2-4. Weekly anti-GAA antibody formation in rAAV1- and rAAV5-hAAT-hGAA
treated Gaa-- mice.









4.5

4.0

S-o 3.5

p 3.0


2 1.5
( -0 2.0-


1.0 ,

0.5 ,
0 10 20 30 40
Liver GAA Activity
(percent of normal)
Figure 2-5. Exponential decay relationship between percent normal liver GAA activity
versus terminal anti-GAA antibody titers in rAAV-hAAT-hGAA treated mice.

Discussion

Recombinant AAV-mediated delivery of human GAA to Gaa mice was evaluated

using 4 promoter/serotype combinations. Results from in vitro DNA transfection studies

showed transgene expression from the hAAT promoter to be 2.6-fold greater than the

DHBV promoter, which was approximately equal to the levels of expression observed

from the albumin promoter. Conversely, in vivo studies showed expression from the

DHBV promoter to be 6-fold greater than the albumin promoter. Surprisingly, liver

transgene expression with a 6-fold lower dose using the hAAT promoter with serotypes 1

and 5 were very low. While a direct comparison of all three promoters using the same

serotype and dose would be necessary to determine the relative promoter strengths, these

results suggest that the hAAT promoter is weaker in the liver than both the DHBV and

albumin promoters.









These initial studies to evaluate liver-directed rAAV gene therapy for GSDII were

disappointing, showing up to 200% of normal liver GAA activity levels with no

correction of distally affected tissues. There are two likely hypotheses for the lack of

cross-correction observed in Gaa-- mice:

* Levels of liver GAA expression were not sufficient to provide necessary amounts
of secreted GAA for uptake by distal cells.

* Gaad mice, recognizing the rAAV-derived human GAA as a foreign protein,
formed anti-GAA antibodies, which inhibited liver-secreted human GAA
preventing uptake by distal cells.

The former hypothesis can be tested only by generating higher-expressing rAAV-hGAA

vectors, and future studies will benefit from evaluating intrahepatic delivery of

rAAV5-DHBV-hGAA in Gaad mice. The latter hypothesis is not likely to be the sole

explanation of the observed results, as 6 of the 10 rAAV-hAAT-hGAA-treated mice had

near-background to background antibody titers with no restored activity in distal tissues.

It is possible that a combination of both insufficient liver GAA levels and inhibitory

antibody formation is responsible for lack of cross-correction, and it will likely be

necessary to overcome both of these obstacles to achieve systemic correction.

Immunologically naYve treated Gaa'- mice showed considerable heterogeneity not

only in the levels of hepatic GAA expression from identical treatments, but also in the

anti-GAA antibody response. This heterogeneity in hepatic expression levels could be

attributed to the interplay of unidentified modifier genes, variable success of vector

administration between mice, or physiological state of the mouse at the time of treatment.

While it is difficult to control for variable hepatic expression levels after identical

treatments, we may be able to control for the variegated immune response. In order to

determine the potential for cross-correction from liver-produced, rAAV-derived hGAA, a






41


mechanism for inhibiting the immune response to human GAA in Gaa-d mice will have

to be developed.














CHAPTER 3
CHARACTERIZATION OF A NEONATALLY TOLERIZED GSDII MOUSE MODEL

Background

Initial studies evaluating liver-directed rAAV-mediated delivery of hGAA to Gaa'

mice resulted in up to 200% of wild-type liver GAA levels with no restoration of activity

in the cardiac or skeletal muscle. At this time, numerous reports were beginning to

emerge investigating the immune response to rAAV-derived transgene products,

(particularly for the treatment of hemophilia). While anti-GAA antibody titers were

observed beginning at five weeks post delivery of rAAV-hAAT-hGAA to Gaa-' mice, the

peak titers were only 1.0 to 3.1-fold over background. Because of the lack of information

concerning the presence of anti-GAA antibodies to hGAA (delivered by gene or enzyme

replacement therapy), it was necessary to determine if the observed antibody titers were

inhibitory. One approach to this is comparison of treatment outcomes in Gaa'' mice with

and without anti-GAA immune responses. To better control for the anti-GAA response,

we generated a tolerant Gad'- mouse model. There are a variety of approaches to

generating tolerance to a therapeutic foreign protein. We chose to generate a murine

model of neonatal tolerization by administering a small dose of highly purified

recombinant hGAA (rhGAA) to 1-day-old Gaa' mice. This technique is modified from

the protocol described by Pittman et al. (96), where a low dose of various antigenic forms

of hFVIII was delivered intraperitoneolly to 1 day-old mice for the purpose of studying

relative antigenicty of the different isoforms.









For our studies, human GAA tolerance was confirmed by lack of antibody

formation after challenge with repeated injections of a therapeutic dose of the same

protein. This tolerant GSDII model will be valuable for determining the effects of

anti-GAA humoral immunity and for evaluating the full potential of both gene and

enzyme replacement therapies in the absence of anti-GAA antibodies.

Materials and Methods

Pretreatment of Gaa-" Mice

All mice were house in the University of Florida SPF animal facility and all animal

procedures were done in accordance with the University of Florida's Institutional Animal

Care and Use Committee approval guidelines. Recombinant hGAA (25 ftg), diluted in 50

IxL of 0.9% saline, was injected subcutaneously in the scruff of the neck of 15-30 h old

Gaa-- mice using a 0.5 cc insulin syringe. Mice were immediately returned to the cage.

Intravenous Protein Delivery

Mice were anesthetized via intraperitoneal delivery of avertin (2,2,2-

tribromoethanol) prepared as a 1.2% solution and used at a dose of 0.2 mL/10 g body

weight. The fur surrounding the neck area was removed using a depilatory cream,

followed by cleaning with 10% povodone-idodine solution. A small incision was made

in the skin to provide access to the internal jugular vein, and rhGAA was injected at a

dose of 10 mg/kg (diluted to a total volume of 100 pL in 0.9% saline) using a 0.5 cc

insulin syringe. The incision was closed and analgesics were delivered subcutaneously.

ELISA for Detection of Anti-GAA Antibodies

Serum samples were obtained weekly via tail bleed of anesthetized mice. Serum,

diluted 1:80, was assayed for the presence of anti-GAA antibodies by ELISA (described

in chapter 2). Results were standardized to serial dilutions of rabbit anti-human GAA









antibody and reported as fold over background with background equal to the anti-GAA

antibody titer in untreated Gaa-- mice.

ELISA for Detection of Distinct Isotypes

The ELISA protocol for measuring distinct antibody isotypes was performed as

described in chapter 2 for the anti-GAA antibody ELISA with the following

modification: After incubation of serum (diluted 1:80) on coated plates and washing,

rabbit-anti mouse- IgGI, IgG2a, Ig2b or IgM (Zymed, San Francisco, CA) (diluted

1:10,000 in 10% FBS), was added to the wells for 30 min at room temperature. The

remaining ELISA steps were performed as previously described.

Competition ELISA to Determine Antibody Specificity

Immulon microtiter plates (Thermo Labsystems, Franklin, MA) were coated

overnight at 4C with 200 tL 0.5 mg/ml human GAA in 0.1M NaHCO3 (pH 8.2). Wells

were washed three times with 300 pL PBS containing 1% Tween20 (PBS/T) and blocked

with 300 p.L 10% fetal bovine serum (FBS) diluted in PBS/T for 2 h at room temperature.

After incubation, washing was repeated, and the competing rhGAA was added to

triplicate wells in a total volume of 100 tL (diluted in 10% FBS) at the following

concentrations: 0, 0.1, 0.5, 5.0, 10, and 20, pg/mL. Serum (pooled from 3 naive Gaa'

mice 1 wk after intravenous delivery of 10 mg/kg rhGAA) was diluted 1:80 and added to

wells containing competing rhGAA. After 1 h at room temperature, wells were washed 3

times to remove antigen-bound serum. Sheep-anti-mouse IgG-HRP linked antibody

(Amersham Pharamcia Biotech, Piscataway, NJ), diluted 1:10,000 in 10% FBS, was

added to the wells in a total volume of 100 pLL. After 30 min incubation at room

temperature, washing was repeated, and 100 [tL tetramethyl benzidine (Sigma-Aldrich,









St. Louis, MO) was added to the wells for 1-3 min. The reaction was stopped with

100 pL IN H2SO4, and absorbance was measured at 450 nm. The absorbance values

obtained in the presence of competing antigen were divided by the absorbance values

obtained in the absence of competing antigen to obtain the percent decrease in signal.

Lymphocyte Proliferation Assay

All techniques were performed under sterile conditions. Spleens were harvested

from treated and control animals and placed in a Petri dish containing IX Hanks

Balanced Salt Solution (HBSS) (Cellgro, Herndon, VA). The spleens were then placed in

70 pm nylon mesh filters over a 50 mL conical tube. Using a 27-gauge needle and 3 cc

syringe, HBSS was injected directly into the spleen to break up the tissue. The spleens

were ground on the surface of the nylon mesh filter using the plunger from the 3 cc

syringe. The membrane was rinsed with HBSS using a disposable transfer pipette.

HBSS was added to bring the final volume to 25 mL. The cells were centrifuged at 500 g

for 10 min. Supernatant was removed, and cells were resuspended in the remaining small

volume of media. Two mL of cold NH4Cl Solution (Stem Cell Technologies, Vancouver,

BC) was added to the cells and kept on ice for 2 min (to induce lysis of the red blood

cells). The reaction was stopped by adding HBSS to a final volume of 25 mL. Cells

were centrifuged at 500 g for 10 min. Supernatants were removed, and cells were

resuspended to a final volume of 5 mL in RPMI media (Cellgro) supplemented with 10%

FBS, 1% penicillin/streptomycin and 2.5 % IM HEPES. Cells were counted on a

hemacytometer and splenocytes were transferred to a 96-well round-bottom dish (1 x 105

cells/well in a total volume of 100 gpL). Conconavalin A (Con A) (Sigma) was diluted in

RPMI media and added to positive control wells at 6 final concentrations ranging from









0.1 to 5 [tg/mL. Recombinant human GAA was diluted in RPMI media and added to

sample wells at 4 concentrations ranging from 0.5 to 10.0 jtg/mL. Cells were incubated

in humidified air, 5% CO2 at 370C. Five days after treatment, tritiated thymidine was

added to wells at a final concentration of 0.2 Ci/well. Cells were incubated at 370C for

7 h. Cells were harvested onto glass fiber multiscreen plates (Millipore, Billerica, MA)

and washed 3 times with sterile water. After drying, 25 RtL scintillation fluid was added

to wells and radioactivity was counted using a Trilux beta counter (Wallac, Boston, MA).

Responses were calculated as a stimulation index by dividing cpm of the stimulated wells

by the cpm of the unstimulated wells.

In vitro GAA Inhibition Assay

HeLa cells were cultured in IX DMEM (Cellgro) supplemented with 10% fetal

bovine serum and maintained in 5% CO2 at 370C. Cells were seeded to 90% confluency

in 12-well plates. For each treatment sample, pooled serum from 3 mice was used.

Serum-free DMEM (760 RiL) was added to 40 IxL serum and 2 RlL rhGAA (0.15 tg/p.L in

serum-free DMEM). The 800 pL mixture was incubated for I h at 4C on an inverting

platform. A control sample of rhGAA alone was also mixed for 1 h at 4C to control for

loss of activity. Medium was removed from cells, which were then washed with IX PBS.

The serum-rhGAA mixture was added to cells, and cells were incubated at 370C, 5% CO2

for 3.5 h. Cells were harvested in 400 tL IX PBS per well. Cells were centrifuged and

the cell pellet was resuspended in 120 pL IX PBS. The cell lysates were subjected to 3

freeze-thaw cycles followed by centrifugation to pellet cell debris. Cell lysate (20 |gL)

was used in the 4-MUG-cleavage assay for determination of GAA enzymatic activity as

described in chapter 2.









Results

Neonatal Pretreatment Results in Humoral Tolerance to Intravenous Protein
Delivery

To induce tolerance to hGAA, Gaa- mice were administered 25 gg rhGAA

subcutaneously between 15-30 h after birth. Eight weeks after rhGAA administration,

pretreated and na've Gaad mice were administered the same protein intravenously at a

dose of 10 mg/kg (which was the intermediate dose being evaluated at this time for

preclinical enzyme replacement therapy trials). A second dose was delivered 4 wk after

the first to boost the immune response, and serum samples were collected weekly out to 4

wk after the second injection. Anti-GAA antibody formation in naive mice followed a

typical humoral response pattern to repeated antigen exposure; antibodies were first

detected 2 wk after the first injection and peaked 1 wk after the second. Antibody titers

gradually fell to approximately 5-fold over background by the end of the experiment

(Figure 3-1). In contrast, there was no detectable anti-GAA antibody formation for the

duration of the study in pretreated Gaad' mice.

Antibodies are Specific to GAA

A competition ELISA was performed to confirm the specificity of antibodies

detected in serum from treated mice for GAA (Figure 3-2). Six titrations of rhGAA were

added to respective wells, along with serum from treated mice, to compete with

well-bound rhGAA for binding of IgG in the serum. The antigen specificity of IgG is

determined by the degree to which the competing GAA prevents serum IgG from binding

the well-bound GAA and is measured by a loss of signal after colorometric detection of

the HRP-conjugated anti-mouse IgG anitbody. After adding 0.01 [tg rhGAA, an 8%

decrease in signal was observed. This was followed by a 30% decrease in signal in the






48

presence of 0.05 lpg competing GAA. Percent signal reduction reached a steady state

with 0.5 pXg of competing GAA. This exponential curve suggests high specificity of IgG

detected in the serum of rhGAA-challenged Gaa-- mice for GAA.


100




10


0 2 4 6 8 10
Time
(weeks post-injection)
Figure 3-1. Anti-GAA antibody formation in pretreated and naive Gaa' mice to rhGAA
protein. Recombinant hGAA was injected via the jugular vein at a dose of 10
mg/kg at 0 and 4 wk (8 and 12 wk of age respectively). Anti-GAA antibodies
were measured by ELISA (n=8/group). Data points represent mean anti-GAA
antibody titer standard error of the mean.


Naive Gaa"/
-o- Pre-treated Gaa-/-










60




30


S20
0) 5u
C.l


0-



0.0 0.5 1.0 1.5 2.0 2.5
Competing rhGAA (lg)

Figure 3-2. Competition ELISA of anti-GAA antibody-containing serum. Serum samples
were pooled from 3 nalve Gada'- mice 1 wk after the second injection of 10
mg/kg rhGAA. Data points represent the percent decrease in serum to plate
binding after competition with the amounts of rhGAA designated on the
X-axis.

Recombinant Human GAA-Challenged Mice Show no Signs of Cell-Mediated
Immunity

We previously reported in chapter 2 that naive Gad-1 mice show no detectable signs

of cell-mediated immunity to rAAV-derived hGAA after intraheptic delivery. Because

the development of a tolerant GSDII model will be useful for both gene and enzyme

replacement therapies studies, it was necessary to determine if naive Gaa-'- mice elicited

a cell-mediated immune (CMI) response to direct rhGAA infusion. If a CMI response is

expected, it will be necessary to confirm that the chosen approach to tolerization reduces

or abolishes the CMI as well as the humoral immune response. In order to evaluate the

Thl-driven CMI response to hGAA in Gaa-' mice, we examined liver histopathology,

proliferation of splenocytes from challenged mice in response to rhGAA, and formation

of the Th l-indicating isotype, IgG2a.









No lymphocyte infiltration was observed on hematoxylin and eosin-stained liver

sections from naive or pretreated Gad- mice after rhGAA challenge (data not shown).

After stimulation of harvested splenocytes from treated mice, no proliferation was

observed in rhGAA treated wells in contrast to 25-fold proliferation observed in the

positive control (ConA)-treated wells (Figure 3-3).

While proliferating Th2 cells characteristically activate B cells leading to the

production of IgGI, Thl cells (which stimulate proliferation of cytoxic T-lymphocytes)

can also activate B cells resulting in IgG2a production (34). Therefore, IgG2a can be

considered a Thl-indicating isotype. Anti-GAA antibodies in all samples consisted

predominantly of IgGI with 3- to 6-fold lower levels of IgG2b and background levels of

IgG2a (Figure 2-6). This lack of IgG2a suggests the absence of significant Thl- cell

proliferation and presumably a lack of CMI. Note that IgM formation follows an

expected pattern with a primary response forming at approximately 1 wk prior to

class-switching to IgG followed by a secondary reaction immediately after the second

dose. The results suggest a lack of CMI after two monthly doses of 10 mg/kg rhGAA

and therefore tolerance to this dosing regimen of hGAA will not require inhibiting

Th 1-mediated responses.
























5



Pre-treated Naive

Figure 3-3. Lymphocyte proliferation assay on splenocytes from pretreated and naive
Gaa--mice 8 wk after challenge with 10 mg/kg rhGAA. Bars represent the
mean (n=6 mice per treatment each assayed in quadruplicate) standard error
of the mean.


35

t 30
*-.
'25

S20

15

.9 10

< 5


0 2 4 6 8
Time
(weeks post first injection)

Figure 3-4. Detection of anti-GAA isotypes in naive Gad' mice after rhGAA delivery.
Naive Gada' mice were administered rhGAA (10 mg/kg) via the jugular vein
at 0 and 4 wk (8 and 12 wk of age respectively). Symbols represent the mean
standard error of the mean (n=3).


z









Susceptibility to Tolerization Decreases from 1 to 7 Days after Birth

To use this neonatal tolerant GSDII model for future studies evaluating gene

delivery and consequential immune response, we needed to determine the window of

opportunity for successful induction of tolerance in the Gaa' model. The theoretical

window of opportunity for neonatal tolerance varies depending on the mode of tolerance

induction and nature of the toleragen. Most reported neonatal tolerance studies induce

tolerance within 24 h after birth. However, we were interested in the possibility of later

tolerance induction; as the immunologic development in the mouse is thought to occur

throughout the first week of life (55).

Six mice were pretreated at 1, 3, or 7 days after birth and subsequently challenged

with 10 mg/kg of rhGAA 8 wk later. Three weeks after challenge, serum was collected

and assayed for the presence of anti-GAA antibodies (Figure 3-5). Six of 6 mice

pretreated at 1 day-old had less than 2-fold over background anti-GAA antibody titers.

Mice pretreated at 3 days-old showed a slightly less susceptibility to tolerance: Four of 6

mice had antibody titers of less than 2-fold over background while 2 of 6 mice had

intermediate antibody titers (between 2- to 4-fold over background). Mice pretreated at 7

days-old showed even less susceptibility to tolerance: One of 6 mice had background

antibody titers, 3 of 6 mice had intermediate titers and the remaining two had higher titers

(4- to 5-fold over background). Despite the high antibody titers observed in 7 day

pretreated mice, antibody formation was inhibited to some degree; as anti-GAA antibody

titers in naive mice receiving the same protein challenge were approximately 9-fold

background at 3 weeks post-challenge (Figure 3-1).









6




2 4

< 3-
< >
0. 2

4 1 o
< 0---o


0
1 3 7
Time of Pre-Treatment
(days after birth)
Figure 3-5. Anti-GAA antibody titer in Gaa/ mice after pretreatment at different ages
followed by rhGAA challenge. Gaa-- mice were pretreated with 25 ig
rhGAA at 1, 3, or 7 days after birth and subsequently challenged with 10
mg/kg rhGAA 8 wk later. Serum samples assayed were obtained 3 weeks post
injection. Data points represent individual mice. Each group contains 6 data
points.

Anti-GAA Antibodies in Treated Serum Inhibit GAA Activity In Vitro

Antibodies generated to a foreign protein can inhibit protein activity by a variety of

mechanisms including binding of the active site, preventing uptake of the

antigen-antibody complex at the receptor, and altering intracellular distribution.

Alternatively, it is possible that antibodies can stabilize a foreign protein by increasing

the half-life, or have no detectable effect at all (particularly at low antibody to antigen

ratios). In effort to determine if anti-GAA antibodies observed in treated serum from

Gaa-- mice are inhibitory, an in vitro inhibition assay was performed (Figure 3-6).

Recombinant hGAA was mixed at 4C for 1 h with 1 of 5 different samples:

1. Serum from pretreated Gaad'- mice after challenge with 10 mg/kg rhGAA
2. Serum from naive Gaad- mice after challenge with 10 mg/kg rhGAA
3. Serum from naive Gaa-- mice after challenge with 5 mg/kg purified 3-galactosidase









4. Serum from naive Gaa-- mice (not challenged)
5. Serum-free DMEM (used for diluting all serum samples) alone.

After incubation, the rhGAA-serum or rhGAA-media mixture was added to HeLa cells.

Cellular GAA activity was assayed 3.5 h after treatment. GAA activity in cells receiving

rhGAA-serum treatment was normalized to cells receiving rhGAA-media alone. Due to

the non-specific binding activity of serum, we expected a decrease in GAA activity after

incubation with untreated serum when compared to cells treated with rhGAA-media

alone. A 28 4% decrease in GAA activity was observed after incubation with serum

from untreated mice. In contrast, a 60 5% decrease in GAA activity was observed after

incubation with serum from naive Gaa-' mice challenged with rhGAA (corresponding to

an anti-GAA titer of approximately 15-fold background). Incubation with serum from

pretreated Gaa'- mice receiving the same protein challenge (determined to have

background anti-GAA antibody titers) resulted in only a 26% 14% decrease in GAA

activity, similar to that observed with control serum. To control for the presence of other

immune-regulatory components generated from a humoral response to a foreign protein,

serum from mice challenged with (3-galactosidase was included as a negative control.

These results demonstrate that the 15-fold background anti-GAA antibody titers observed

in naive Gaa-' mice after repeated rhGAA challenge, were sufficient to significantly

inhibit GAA activity by approximately 30% in vitro (P=<0.01).











O .

naive Gaa'
untreated


naive Gasa
5 mg/kg p-gal


pre-treated Gaa&'
10 mg/kg rhGAA


naive Gaa'
10 mg/kg rhGAA


Collect serum lwk post-challenge #2


Mix with rhGAA 40C 1hr


HeLa cells

Assay cellular GAA activity

Figure 3-6. Diagram of in vitro GAA inhibition assay experimental design. The colored
boxes correspond to the colored bars in Figure 3-7.


70

60

W 50

1540

'- *30

6 .- 20

10

0


S Naive serum (untreated)
SNaive serum (B-gal-challenged)
lm Pre-treated serum (rhGAA-challenged)
-- Naive serum (rhGAA-challenged)


Figure 3-7. Affects of anti-GAA antibody-containing serum on GAA activity in vitro.
Bars represent the decrease in GAA activity of HeLa cells after treatment with
rhGAA mixed with serum relative to cells treated with rhGAA alone (n=3
wells). P-<0.05









Discussion

For the purposes of evaluating the potential for intrahepatic rAAV-hGAA delivery

to lead to systemic cross-correction in Gaad mice without the interference of an

anti-GAA humoral response, a tolerant Gaa-- mouse model was generated by

introduction of rhGAA antigen to neonatal Gaad- mice during a period of immunologic

immaturity. Neonatal tolerance was first described in 1945 by Owen (93) and later

demonstrated experimentally by Billingham et al. (9). Since these studies, the

immunologic mechanism responsible for neonatal tolerance has remained controversial.

Originally, it was believed that neonatal exposure to antigen results in clonal deletion

and/or inactivation of T-cells. More recently, alternative mechanisms have been

proposed including T-cell suppression, which could result in reversible tolerance, and

immune deviation, leading to a skewed T helper2 (TH2) response (1,18,37). Multiple

mechanisms may play a role in how neonatal tolerance is induced and could possibly

depend on factors such as dose, route and immunogenicity of toleragen.

Neonatally pretreated Gaad- mice were shown to be 100% tolerant to hGAA after 2

injections of rhGAA at a dose that resulted in up to 100-fold over background anti-GAA

antibody titers in naive Gaad- mice. While no signs of CMI were observed in these

studies, we speculate that higher and more frequent doses would eventually lead to CMI

as demonstrated by Raben et al (101) who observed 100% anaphylaxis after the 7th

injection at a dose of 20 mg/kg.

We demonstrated that susceptibility to neonatal tolerization decreases between 1 to

7 days after birth. This is important for the use of this model in future studies evaluating

rAAV-mediated treatment as inaccurate estimation of birth date by 1 or 2 days will likely

skew results. Preliminary in vitro studies suggest that anti-GAA antibodies generated









against rhGAA inhibit GAA activity. These results provide a rationale for further

investigation of inhibiting anti-GAA immunity for successful gene transfer. The results

from this study, however, provide no information as to the relationship between titer and

degree of inhibition and what consequences the observed degree of inhibition will have

on Gaa-' mice. Determining how anti-GAA antibody formation will impact the ability to

achieve cross-correction from liver-directed gene transfer will best be evaluated by

comparing relative levels of GAA in distal tissues of treated naive Gaa/ mice versus

treated tolerant Gaa-1- mice.














CHAPTER 4
AAV5-MEDIATED CROSS-CORRECTION OF GSDII: EFFECTS OF INHIBITORY
ANTIBODY FORMATION AND IMMUNOMODULATION

Background

Achieving systemic correction in GSDII mice after liver-directed delivery of

rAAV-human GAA (hGAA) will likely require high levels of liver-GAA expression along

with inhibition of anti-GAA antibody formation. Previous experience with rAAV

serotype 2 vectors encoding hGAA driven by the DHBV promoter resulted in

superphysiologic levels of liver GAA activity with no biochemical correction of distally

affected tissues. Recombinant AAV serotype 5 vectors have been shown to yield up to

10-fold greater liver transgene expression (82), however, our experience with

rAAV-mediated delivery of hGAA under control of the hAAT promoter resulted in

similar yet insufficient levels of liver GAA expression for cross-correction as with

rAAV2. We therefore generated a rAAV vector using serotype 5 with the DHBV

promoter driving expression of human GAA.

To address the obstacle of anti-GAA antibody formation observed in preliminary

studies, we generated a tolerant Gaa d model by neonatal pre-treatment with low dose

antigen. The anti-GAA humoral response to rhGAA, observed in naYve Gaa mice, was

abolished in the pretreated Gaa mice. However, the affect of neonatal pretreatment on

anti-GAA immunity to rAAV-derived hGAA has not been evaluated. We challenged

tolerant Gaa- mice with rAAV5-DHBV-hGAA with the goal of achieving higher levels

of liver GAA than previously observed; inhibition of anti-GAA antibody formation; and









subsequent correction of the cardiac, respiratory and skeletal muscles. Nafve Gaa- mice

were identically treated. Comparing outcomes in both the presence and absence of

anti-GAA antibodies should answer important questions about the significance of anti-

GAA humoral immunity and provide further information about the necessary approach

toward clinical applications.

Materials and Methods

Cloning and Packaging of rAAV Constructs

Cloning of the pTR-DHBV-hGAA plasmid is described in chapter 2. The

recombinant AAV5 vector (rAAV2 rep / rAAV 5 cap) was generated, purified and

tittered (described in chapter 2) by the University of Florida Powell Gene Therapy Center

Vector Core.

ELISA for Detection of Anti-GAA Antibodies

Serum samples were obtained weekly via tail bleed of anesthetized mice. Serum,

diluted 1:80, was assayed for anti-GAA antibodies by ELISA (described in chapter 2).

Results were standardized to serial dilutions of rabbit anti-human GAA antibody and

reported as fold over background with background equal to the standardized value of

serum from untreated Gaa- mice.

Animals

All mice were housed in the University of Florida SPF animal facility and all

animal procedures were done in accordance with the University of Florida's Institutional

Animal Care and Use Committee approval guidelines. The pretreated Gaa mouse

model was generated by subcutaneous delivery of 25 lag rhGAA at 1 day of age

(described in chapter 3). Mice were treated at 10 wk of age with 5 x 10" or 1 x 1012 vg









rAAV5-DHBV-hGAA via portal vein delivery of rAAV as described in chapter 2. Age-

matched C57B6/129 mice were used as controls.

Tissue Processing

Eight weeks post injection, animals were sacrificed and extracted tissues were

immediately cut and frozen in liquid nitrogen. Skeletal muscle tissue was first crushed to

a powder using a hemostat in liquid nitrogen prior to homogenization. Tissues were

homogenized in sterile water with lysing matrix D using the Fast Prep Instrument

(Qbiogene, Carlsbad, CA). Homogenized lysates were centrifuged to pellet the lysing

matrix. Supernatants were subjected to three freeze-thaw cycles and centrifugation.

GAA Enzymatic Activity Assay

The enzymatic activity assay for GAA and the Bradford assay for total protein

quantification were performed as described in chapter 2. Briefly, 20 1pL of homogenized

tissue lysate was added to one well of a black 96-well plate with 1.6 gL of 75mM 4-

MUG (diluted in 38.4 [tL of 200 mM sodium acetate, pH 3.6). After 1 h incubation at

370C, the reaction was stopped with 200 tL of 0.5 M sodium carbonate (pH 10.7). Six

standards of 4-MU ranging from 0 to 500 p~m were used to generate the standard curve.

Fluorescence was measured at 460 nm. Protein concentrations were determined using the

Bio-Rad DC Protein Assay kit (Bio-Rad, Hercules, CA) according to the manufacturer's

instructions. GAA activity was calculated as tM 4-MU /h/mg protein. Values are

reported as percent of GAA activity in normal tissues from C57BL/6 / 129 mice (n=3)

(processed and assayed at the same time) minus background.









Histological Detection of Glycogen

Tissues were fixed in 10% buffered formalin, embedded in paraffin and sectioned

(4-+m thickness). For liver histopathology, paraffin-embedded liver sections stained

with hematoxylin and counterstained with eosin were examined and scored for

histopathology by a Veterinary Pathologist at the University of Florida Pathology Core

Lab. For glycogen staining, paraffin-embedded muscle sections were stained with

Periodic Acid Schiff (PAS) reagent for glycogen detection (Richard Allen, Kalamazoo,

Michigan) according to the manufacturer's instructions and counterstained with toluidine

blue. For immunohistochemistry, paraffin-embedded liver sections were immunostained,

by the University of Florida Pathology Core Lab, using the Vectastain ABC Kit (Vector

Laboratories, Burlingame, CA) as per manufacturer's instructions with the rabbit anti-

human GAA antibody (1:10,000 dilution). Images were taken using a Zeiss Axioskop

motorized Plus with Axiocam HR color camera. Relative transduction efficiency was

determined as described by Nakai et al. (88). The number of positively stained cells and

total number of cells were counted. Five fields of approximately 200 nuclei were

counted for each section.

Determination of Vector Genome Copy Number

Quantitative PCR followed by chemiluminescent detection was performed as

described by Mingozzi et al. (82). Briefly, using 1.25 jtg genomic liver DNA as a

template and biotinylated primers, co-amplification of a 2.1 kb region of the DHBV-

human GAA DNA (5'-CCAACACATGCGCAATATCC and 5'-

CCGGTCTCGTTGGTGATGAAA-3') and a 1.0 kb region of the endogenous HPRT

gene (5'-GCTGGTGAAAAGGACCTCT-3' and 5'-CACAGGACTAGAACACCTGC-









3') was performed. In addition to the genomic DNA, standard controls included 0.01,

0.05, 0.1, 0.5 and 1.0 pg pTR-DHBV-human GAA plasmid DNA spiked into 1.25 pig

genomic liver DNA from saline-treated Gaa-1 mice. All reactions were performed at the

following conditions: hot start denaturation at 94C for 5 min, followed by 30 cycles of

denaturation at 94C for 1 min, annealing at 570C for 1 min, and extension at 720C for 2

min. A final 10 min extension was done at 720C. Products were separated by

electrophoresis on a 1.0% agarose gel followed by transfer to a nylon membrane and

visualized using the Southern-StarTM system (Applied Biosystems, Bedford, MA) as per

the kit protocol. Densitometric analysis of resulting bands was performed using Scion

Image Release Beta 4.0.2 software (Scion Corporation, Frederick, Maryland) and ratios

of human GAA/HPRT band intensity were calculated. Gene copy numbers were

estimated using the following calculations: Picrograms of hGAA cDNA used in the

standards were converted to copies of hGAA per cell based on 1 pg of double-stranded

DNA of 1000bp = 0.9 x 107 copies. Total copies in each standard were divided by the

number of cells in the standard based on 1 cell = 15 pg DNA (71). A standard curve was

generated using hGAA/HPRT mean pixel density ratio versus copies of human GAA per

cell. Copies of human GAA in the samples were determined using the equation from the

standard curve (y = 0.464x + 0.20, R2=0.973).

RNA Isolation and Analysis of Human GAA Transcript

RNA was isolated using the Rneasy kit (Qiagen, Valencia, CA) according to the

manufacturers instructions for RNA isolation from the respective tissue. The RNA was

treated with DNAseI (Ambion, Austin, TX) for 1 h. First strand cDNA synthesis was

primed from 2.5 pg of DNAseI-treated RNA using the First-strand cDNA Synthesis Kit









(Amersham Pharmacia Biotech) and 40 pmol of the reverse 13-actin and human GAA

primers described below. PCR amplification of a 270 bp region of the human GAA

cDNA (1 gl) was done using the primer pair 5'-CCTTTCTACCTGGCGCTGGAGGAC-

3' / 5'-GGTGATAGCGGTGGAGGAGTA-3'. A separate PCR reaction for

amplification of a 300 bp region of the P-actin cDNA (1 gl) was performed under the

same conditions using the following primer pair:

5'TCTAGGCACCAAGGTGTGAT-3' /5'- GTGGTACGACCAGAGGCATA-3'.

Immunoblot Detection of the Mannose 6-Phosphate Receptor

Protein concentrations of tissue lysates were determined by a standard Bradford

assay using two dilutions of each lysate and performed in triplicate. Absorbance values

were calculated using a standard curve from 8 serial dilutions of BSA. All values fell

within the linear range of the standard curve. Tissue lysates (100 gpg) were

electrophoresed through a non-denaturing SDS-8% polyacrylamide gel and transferred to

nitrocellulose membranes. The membrane was blocked for 1 h in 5% nonfat dry milk

followed by three 10-min. washes. The membrane was probed with a 1:2500 dilution of a

rabbit-anti-human M6P receptor antibody (gift of W. Dunn, University of Florida,

Gainesville, FL) overnight at 4C, washed and probed with a 1:10,000 dilution of horse

radish peroxidase (HRP)-conjugated, donkey-anti-rabbit IgG secondary antibody

(Amersham Pharamacia Biotech, Piscataway, NJ). After washing, hybridization was

detected using the ECL Plus Western blotting detection system (Amersham Pharmacia

Biotech).









Statistical Analysis

Unpaired t test and unpaired t test with Welch's correction (GraphPad Instat

version 2.0, San Diego, CA) were used for statistical comparison between groups with

statistical significance considered if P<0.05.

Results

Liver GAA Activity and Antibody Formation after Intrahepatic Delivery of
rAAV(5)-DHBV-hGAA to Naive Gaa'" Mice

To evaluate the hepatic GAA expression levels obtainable under control of the

DHBV promoter packaged in serotype 5, two doses of rAAV(5)-DHBV-hGAA were

delivered to the portal vein of 10 wk old, female, naive Gaad mice. After low dose (5 x

1011 vg) delivery, liver GAA activity levels were 324% 17.5% of normal (normal)

(Table 4-1). This level is 1.8-fold greater than the liver activity levels observed with the

rAAV2-DHBV-hGAA vector at a 6-fold higher dose (3 x 1012vg), which is consistent

with observations by others (40,82) that hepatic transgene expression from rAAV5 is up

to 10-fold greater than from rAAV2. Despite the presence of 3-fold normal hepatic GAA

activity levels, there was no significant activity restoration in the affected distal tissues

(heart, diaphragm and quadriceps).

After high dose (1 x 1012 vg) delivery of rAAV5-DHBV-hGAA to naive Gaa-'

mice, 3-fold greater hepatic GAA activity levels were observed than from the low dose

vector (930% 200% of normal; P=0.01) (Table 4-2). GAA activity was partially

restored in the heart of 1 animal and the diaphragms of all 6 mice (Table 4-2). Like the

low dose group, there was no significant activity observed in the quadriceps after high

dose treatment. Because GAA activity levels in this group of mice receiving identical

doses varied by 15-fold in the heart and 8-fold in the diaphragm, and because the immune









response in naive Gaa'- mice is similarly heterogeneic, we hypothesized that the

variability in GAA activities of distal tissues was, in part, determined by the

heterogeneity of the anti-GAA immune response.

Antibodies to GAA were detected in 5 of 6 high dose-treated mice and 3 of 6 low

dose-treated mice after intrahepatic delivery of rAAV(5)-DHBV-GAA to naive Gad'

mice (Tables 1 and 2). The 3 non immune-responsive mice from the low dose group

showed no significant cardiac activity (Table 1). On the other hand, the 1 non immune-

responsive animal from the high dose group (#5) showed restoration to 28.7% of normal

activity levels in the heart and was, in fact, the only high dose mouse to have therapeutic

levels of cardiac activity (Table 2). Furthermore, while there was no diaphragm

correction in any of the 6 low dose mice (Table 1), all 6 high dose mice showed some

level of restored diaphragm activity (26% to 304%). Interestingly, the one

non-immune-responsive mouse in the high dose group (#5) had 4-fold greater diaphragm

GAA activity levels than the five that elicited an immune response (Table 2).

Table 4-1. Acid a-glucosidase activity and anti-GAA antibody titer 8 weeks after low
dose vector delivery to naive Gaad- mice.
Mouse Anti-GAA Ab Percent of normal GAA activity
titer (fold- Liver Diaphragm Heart Quad.
background)
1 16.5 304.3 6.8 1 0.0
2 1.0 354.3 NA 2.5 0.0
3 1.0 310.3 0.0 1.4 0.0
4 5.6 276.3 0.0 9.7 0.0
5 1.0 395.3 0.0 0.3 0.0
6 3.6 304.3 6.8 1 0.0
mean. iL .4.8i *"4 *' .. ;324 17.5 ." 2;7 .6.11.6.4 6 4'!0.0.:+: 0"-.0
a Background activity averaged from 3 saline-treated Gaa mice was subtracted from all
values.











Table 4-2. Acid a-glucosidase activity and anti-GAA antibody titer 8 weeks after high
dose vector delivery to naive Gaa-'- mice.
Mouse Anti-GAA Ab Percent of normal GAA activity
titer (fold-over Liver Diaphragm Heart Quad.
background)
1 3.2 833.3 40.8 2.4 0.0
2 8.5 1403.3 26.8 0.0 0.0
3 1.9 1597.3 165.8 2.1 0.0
4 1.5 406.3 52.8 2.2 0.0
6 5.0 762.3 26.8 5.3 0.0

58 1.0 487.8 304.8 28.7 3.1
a Background activity averaged from 3 saline-treated Gaa-- mice was
subtracted from all values.
b Mouse 5 is separated from mice 1-4 to demonstrate difference in both antibody titer and
GAA activity of diaphragm, heart and quadriceps.


Diaphragm GAA Levels are Dependent on Both Liver GAA Activity and Antibody
Titer

Careful examination of liver and diaphragm GAA activities along with the

respective antibody titers of individual mice reveals an important relationship. Graphical

representation of diaphragm GAA activity as a function of either antibody titer or liver

GAA activity alone resulted in no consistent trend. However, because the liver can be

considered the source of the diaphragm GAA, the ratio of diaphragm activity was

normalized to liver GAA activity and the result of graphing these values as a function of

antibody titer is an exponential decay curve (Figure 4-1). This graph suggests both that

the level of diaphragm cross-correction is dependent on liver GAA activity and antibody

titer and, there is a small range of antibody titer where diaphragm correction is optimal.

For example, diaphragm activity (normalized to liver activity) dropped by 5-fold between

mouse 5 with a background antibody titer and mouse 4 with only a 1.5-fold over

background titer. The effect of antibody titer on diaphragm correction is less dramatic at








antibody titers of greater than 1.5-fold over background. Within this higher antibody titer

range, we observed that substantially higher liver GAA levels can compensate for

relatively high antibody titers. For example, high dose mouse 2 and mouse 6 achieved

the same levels of diaphragm correction (26.8% of normal) with nearly 2-fold different

liver GAA activity levels (14-fold and 7.6-fold over background respectively), and this is

likely attributed to the nearly 2-fold lower anti-GAA antibody titer observed in mouse 6.

(Table 4-2).


0.7

< 0.6

0.5

--0.4

0 0.3

p 0.2
CL.
-s01 0

S0.0


0 2 4 6 8 10 12 14
Anti-GAA Ab titer
(fold background)
Figure 4-1. Exponential decay relationship between diaphragm activity normalized to
liver activity versus anti-GAA antibody titer. Data points represent individual
naive Gaa '- mice in the high-dose group. This relationship demonstrates the
significant impact of very low antibody titers (1.5-fold background) on
diaphragm activity.









Neonatally Pretreated Gaa-" Mice do not Form Antibodies to rAAV-Derived hGAA
after Low Dose Vector Delivery

Based on the previous studies suggesting that the presence of antibodies to

hepatic-produced GAA inhibits correction of distal tissues, we repeated the experiment in

a group of Gaa-' mice that had previously been tolerized to human GAA protein by

neonatal subcutaneous injection (described in chapter 3). Ideally, using a tolerized mouse

model should provide more information regarding the relationship between dose, immune

response, and correction in Gaa- mice. As in the initial experiment, neonatally

pretreated mice received a high (1 x 1012 vg) or low (5 x 10"1 vg) dose of

rAAV5-DHBV-hGAA via the portal vein at 10 wk of age. Anti-GAA antibody formation

was followed weekly for the duration of the study (Fig. 4-2). While 50% of nafve mice

had background antibody titers after low dose delivery (Table 4-1), 100% of the

pretreated mice had background antibody titers (Figure 4-2). Despite inhibition of the

humoral response in all 4 low dose mice, we did not see any restoration of GAA activity

in the heart, diaphragm, or quadriceps (Figure 4-5). This is consistent with data from the

na've group showing that the absence of an immune response in 3 of the 6 mice after low

dose delivery also had no impact on correction outcome in any of the tissues examined

despite having 300% to 400% of normal liver GAA levels.









18
16
"14


S10
._| 8
Sl 6
4
2
0 -- .------I--- -- -
0 2 4 6 8
Time
(weeks post-injection)
Figure 4-2. Anti-GAA antibody formation in naive and pretreated Gaa- mice over time
after low dose vector delivery. Black lines represent the 6 individual naive
Gaa- mice. The red line represents the mean and standard error of the mean
of the 4 tolerized Gaa/' mice.

Neonatal Tolerance is Broken after High Dose Delivery

The trend of anti-GAA antibody formation observed over time for all of the

immune responsive mice described thus far is consistent with earlier experiments using a

weaker rAAV-hGAA vector, where the most significant increase in antibody titer occurs

between 4-5 weeks (Figure 2-2). This is not surprising based on the finding that

transgene expression from an rAAV5 vector is not observed until 4 wk post injection

(15,82) and formation of the primary immune response should follow within 4 to 7 days.

The immune response in tolerized mice after high dose rAAV-hGAA delivery was

different from the low dose tolerized group. Interestingly, after pretreatment of Gaad

mice and subsequent challenge with the high dose vector, we see a variation in the

immune response among the 5 mice (Figure 4-3). Four of the 5 high dose mice elicited

an immune response (Table 4-3). However, anti-GAA antibody formation in these mice









was not detected until 6 to 8 wk post injection (1 to 3 weeks later than in naive mice)

suggesting the possibility of a break in tolerance (Figure 4-3).

In the pretreated group, high dose mouse 2 was the only 1 of 5 without antibody

formation. This may be due to the heterogeneity of the immune response observed in the

naive mice (mouse 2 was naturally non-immune-responsive and would have not required

tolerance) or this mouse was tolerized for the 8 wk duration of the study and tolerance

may have been broken at a later time point. In the absence of any anti-GAA antibody

formation, mouse 2 had the highest levels of biochemical correction in both the

diaphragm (249%) and heart (110%). More importantly, this was the only mouse to

show significant levels of GAA in the skeletal muscle with restoration to 72% of normal

activity (Table 4).

18
16
14








2

0 2 4 6 8
Time
(weeks post-injection)
Figure 4-3. Anti-GAA antibody formation in naive and pretreated Gaa' mice over time
after high dose vector delivery. Red lines represent the 5 individual tolerized
Gad-' mice. The black line represents the mean and standard error of the mean
of the 6 naive Gaa mice.

Among the high dose-treated mice from both the pretreated and naive groups that

elicited an immune response, there was overall higher level of GAA activity observed in
elicited an immune response, there was overall higher level of GAA activity observed in









distal tissues of pretreated mice (Figure 4-6). Higher liver GAA activities in these mice

(1461% 160% versus 1000% 218% respectively; P=0.048) may have contributed to

the higher average levels of distal tissue correction. However, comparison of individual

retreated and naive mice with similar liver GAA activity levels and terminal antibody

titer, yet significantly different levels of cardiac correction, suggests the role of an

additional factor (such as the 1-3 week delay in antibody formation) (Tables 2 and 3).

The detailed analysis of individual mice represented on all 3 tables is represented as

treatment groups in Figure 4-4. Due to the inability to successfully induce permanent

tolerance in Gaa ^mice to high dose (1 x 1012 vg) rAAV-derived human GAA, only 2 of

11 high dose-treated mice showed no antibody formation to rAAV-derived human GAA.

However, these results show a clear relationship between absence of antibody formation

and higher levels of activity in all three distal tissues examined (outliers in Fig. 4-5

represent single non-immune responsive animals).

Table 4-3. Acid a-glucosidase activity and anti-GAA Ab titer 8 weeks after high dose
vector delivery to pretreated Gada-' mice.
Mouse Anti-GAA Ab Percent of normal GAA activity
titer (fold over Liver Diaphragm Heart Ouad.
background)
1 9.4 1068.3 1189.8 39.9 2.1
3 2.3 1843.3 60.2 23 4
4 16.6 1396.3 244.6 49.2 4.7
5 6.1 1537.3 139.6 9.1 11.8
[Dean..:: ; .6 .0 .146.1 + 160 ,. 408, 263,i ;;.3.0%,8.9. 5,6 2.1
2b 1.0 1646.3 2497.8 110.3 72.1
a Background activity averaged from 3 saline-treated Gaa mice was
subtracted from all values.
b Mouse 2 is separated from 1, 3-5 to demonstrate difference in both antibody
titer and GAA activity of diaphragm, heart and quadriceps.














Liver

"


Diaphragm


I Naive
In Pre4reaked


(0
*5 0
-C
<0"


2.


10000



<> 100


10

1


80


5 0
C
<0*40

20


High Low


Quadriceps


Figure 4-4. Summary of percent normal GAA activity 8 wk post delivery of rAAV5-
DHBV-hGAA in the liver, diaphragm, heart and quadriceps of all high and
low dose-treated mice from both the naive and pretreated Gaa- groups. For
the diaphragm, heart and quadriceps graphs, outlier symbols (*) represent the
individual high dose mice from the naive (mouse 5) and pretreated (mouse 2)
groups that had background anti-GAA antibody titers. Bars represent mean
(n=6 for both naYve groups, n=5 for high dose, pretreated group and n=4 for
low dose, pretreated group) standard error of the mean. For the liver graph,
all mice in each group were grouped together to determine the mean and
standard error of the mean.

Lack of Cell-Mediated Immune Response After rAAV(5)-DHBV-hGAA Treatment

In order to further characterize the immune response to rAAV-derived hGAA, it

was necessary to rule out a cell-mediated immune (CMI) response. To assess a potential

CMI response, hemotoxylin and eosin (H&E)-stained liver sections were independently

evaluated by the University of Florida Pathology Core for inflammatory responses.


I I r I
High Low


Heart

Naive
c Pre-treated





*A


CU
.5 0
C

<0

V


-k Naive
Pre-lreated







* ir"









Additionally, proliferation of rAAV-human GAA-treated splenocytes in response to

rhGAA protein was measured. No signs of infiltration were observed on H&E-stained

liver sections taken at sacrifice. Furthermore, there was an absence of proliferation of

splenocytes from rAAV-DHBV-hGAA- treated mice after exposure to rhGAA compared

to a 30-fold stimulation index after ConA exposure (not shown). These two assays,

however, evaluate CMI in terminal tissues only and may not reveal an earlier response.

Serum samples taken bi-weekly from rAAV-DHBV-hGAA-treated mice were assayed for

the presence of IgGI, IgG2b and the Thl-indicating isotype IgG2a. We observed

predominant levels of IgGI, followed by IgG2b and no detectable IgG2a (Figure 4-5) for

the duration of the experiment suggesting a Th2 and not a Th -mediated response.

As an alternative method to rule out the possibility of lower liver GAA activity

levels being attributed to a CMI response, we estimated the vector genome copy number

per cell using methods described by Mingozzi et al. (82) and looked for a correlation

between copy number and terminal liver GAA activity. Liver DNA from high

dose-treated mice was shown to have an average of 0.74 0.05 vector genome copies per

diploid genome (Fig. 4-6). Within the high dose-treated mice from both the naYve and

pretreated groups, there was a strong correlation between copy number and liver GAA

activity (P=<0.01) suggesting that relative levels of GAA activity in the transduced liver

were more likely attributed to the varying degree of vector copy number per cell than to

loss of activity due from a CMI response.










1000


"3
0o



- 10
0

0
*0


0 2 4 6
Time
(weeks post-injection)


Figure 4-5. Isotype analysis of anti-GAA antibodies formed against rAAV5-derived
hGAA. Symbols represent the mean standard error of the mean (n=3).

pDHBV-human GAA plasmid (pg)
0.01 0.05 0.1 0.5 1.0


2.1 kb GAA


1.0 kb HPRT I



2.1 kb GAA



1.0 kb HPRT

copies/cell 0.92


13333








0.79 0.84 0.49 073 062 079


liver GAA activity 1843
(% of wild type)


1537 1646 487 1403 762 1597 untreated


Figure 4-6. Semi-quantitative analysis ofrAAV5-hGAA vector genome copies per cell
by Southern blot ofhGAA and endogenous HPRT after chemiluminescent detection of
PCR-amplified region of the vector.


--- IgG1
-o- IgG2a
- gG2b









Restoration of GAA Activity in Tolerized Mice Improves Glycogen Clearance

In order to evaluate whether the levels of GAA activity observed were sufficient for

clearance of lysosomal glycogen, tissue sections were stained with Periodic acid-Schiff's

reagent (PAS). On longitudinal muscle sections, relative amounts of glycogen can be

determined by the overall intensity of the pink staining. The intensity of PAS positive

staining is decreased to normal levels in the heart and quadriceps of pretreated mouse 2

(Fig. 7). This mouse had the highest levels of GAA activity in both tissues (110% in

heart and 72% in quadriceps) (Table 3). After restoration to 49% of normal heart activity

in pretreated mouse 4, there is no detectable difference in PAS staining (not shown).

This is not surprising as the mice in this study were 3-months old (2-months past the age

at which glycogen accumulation is first detected) and tissues were examined only 4

weeks after initiation of GAA expression.



























Figure 4-7. Detection of glycogen by PAS staining on longitudinal sections of cardiac
and skeletal muscle. Tissues were harvested 8 wk post delivery of 1 x 1012 vg
rAAV5-DHBV-hGAA and stained. Cardiac muscle is shown in A-C and
quadriceps femoris muscle is shown in D-F. A and D represent an untreated,
age-matched Gaa' mouse. B and E represent pretreated mouse 2. C and F
represent an untreated, age-matched normal (C57BL6/129) mouse. Original
magnification is 100X for A-C and 200X for D-F.

Human GAA Protein Observed in Corrected Tissues is Hepatic-Produced

To confirm that hGAA activity observed in distal tissues of high dose-treated mice

is derived from hepatic-produced human GAA, RT-PCR of a 270 bp region of human

GAA RNA was performed on corrected tissues from pretreated mouse 2 (shown),

pretreated mouse 1 and naive mouse 5. The resulting transcript was readily detected in

liver RNA and was undetectable in RNA from heart, diaphragm and quadriceps (Fig. 4-

8). More sensitive, quantitative RT-PCR experiments would exclude the possibility of

leaky promoter activity in other tissues. However, these preliminary RT-PCR results

suggest that GAA activity observed in distal tissues is liver-derived and this is consistent

with the observed correlation between hepatic GAA activity and distal tissue activity

after accounting for the immune response.










Human
RT(-) GAA B-actin
II II I I
L D H L D H L D H Q





Figure 4-8. RT-PCR detection of rAAV5-delivered hGAA RNA in liver (L), Diaphragm
(D), Heart (H) and Quadriceps (Q) by RT-PCR. Reverse transcription of
P3-actin was performed as an internal control. RT-PCR of human GAA in the
absence of the reverse transcriptase enzyme was performed to control for
DNA contamination in the RNA preparation.

Assessment of Hepatic Transduction Efficiency

Successful inhibition of the anti-GAA immune response may be beneficial only if

sufficient levels of hepatic-GAA are produced. As only approximately 10-20% of GAA

is secreted for uptake by distal tissues and high levels of activity are required for reversal

of glycogen accumulation, systemic correction will rely heavily on the ability to

maximize hepatic expression at both the levels of gene expression and cell transduction.

In effort to begin characterizing the transduction efficiency of the vector used in these,

the percentage of human GAA-expressing hepatocytes was estimated by

immunohistochemical staining of liver sections from treated mice followed by cell

counting as described by Nakai et al. (88) (Figure 4-10). In high dose-treated livers, 86%

+ 2.5% (78-93%) of total hepatocytes stained positive for human GAA. This number

closely reflects the vector copy number per cell data for high dose-treated mice (Fig. 6B)

showing 0.74 0.05 copies per cell. In low dose-treated livers, we observed 29% 1.3%

(26-32%) of positive cells. This 3-fold difference in GAA-expressing cells between the









high and low dose groups correlates with the 3-fold difference in liver GAA activity

observed (Table 1 and 2).


- *I I
"' 0...i'e ; ..

'bi. .- o ,
1 I **


C *6
4 S


*W U*a. '1f
'0 01
;. ac '
*' *n b." Y I
a. *r 0 -.'. -


SA s' -. "* A "
.'r m .,, ,, ",- -'. .9.. r

2-', V *" .. *'
....0.. : : ,t, 'I '"K e '
vi,. I- ..s' -., ., i'.... *',$
\, i' ., ,. ..r
'.*N ^yjS
^'^*^^~ r'''; *-,'




^a't -.-- -^.^ .
'a .: r. -, ." ; ,. .", ,, ,
,. -., ,,..e. '*_ ^ '
l A 4
,,., .I ., ... .



..V-, -,


Figure 4-9. Immunhistochemical detection of human GAA-expressing hepatocytes of (A)
saline-treated Gaa-' mouse, (B) low dose-treated Gaad mouse (naive 1), and
(C) high dose-treated Gad-' mouse (naive 2). Original magnification is 400X.









Poor Correction of Quadriceps Correlates with Lower Levels of Mannose
6-Phosphate Receptor

The differences in the ability to correct diaphragm versus cardiac versus

ambulatory muscles has been observed in enzyme replacement therapy studies in the

Gaa- mouse model (101) and is recapitulated in these rAAV-mediated gene replacement

studies. One likely explanation is the difference in the M6P receptor population. After

examining M6P amounts (relative to total protein) in the diaphragm, heart, and

quadriceps by immunoblot, we find lower levels in the quadriceps than in the heart and

diaphragm (Fig. 4-10). Based on the cross-correction results of this study, we would

expect to see higher relative M6P receptor levels in the diaphragm than in the heart.

Other factors may influence diaphragm correction, such as the high surface area and

vascularity to mass ratios, allowing for maximal exposure to and uptake of circulating

GAA. While more sensitive assays need to be performed to look at M6P receptor

expression levels, the Immunoblot analysis data does suggest that the reason for varying

susceptibility to correction in the three tissues examined may be, at least partially, due to

the receptor availability.





Quadriceps Diaphragm Heart


270 kDa ...




Figure 4-10. Immunodetection of the 220-kDA mannose 6-phosphate receptor in the
quadriceps, diaphragm and heart tissue lysate (100 pg) from 4 wk-old Gaa
mice.









Discussion

Impact of Immune Responses to Intrahepatic Delivery of rAAV-hGAA

Earlier studies evaluating cross-correction after rAAV-hGAA delivery to Gad

mice led to local correction without correction of distally affected tissues. These results

were likely attributed to both insufficient hepatic GAA expression and anti-GAA

antibody formation. Using the high expressing, rAAV5-DHBV-hGAA vector we were

able to observe, for the first time, restoration of proposed therapeutic levels of GAA in

the heart, diaphragm and quadriceps of Gaa-' mice after intrahepatic delivery of a rAAV

vector.

At a high dose (1 x 1012 vg), the rAAV5-DHBV-hGAA vector produced between 4-

fold to 16-fold normal hepatic GAA levels in naive Gaa-' mice, whereas the low dose

produced 2.7-fold to 3.5-fold normal levels. These results correlated with the percentage

of GAA-expressing hepatocytes. Upon closer examination, similar liver GAA activities

observed in the higher expressing low dose mice and the lower expressing high dose mice

could not be accounted for by the 2-fold difference in dose. However, as these values

represent terminal activity values only (8 weeks post injection), more significant variation

may have occurred between 4 to 7 weeks post injection, particularly if the kinetics of

expression are dose-dependent.

In the high dose-treated naYve Gaa-d group, the one mouse that failed to elicit an

immune response had the highest level of GAA activity in distal tissues, despite having

the lowest hepatic GAA activity. The naive Gaa-' mice that did elicit an immune

response to rAAV-derived human GAA had varying levels of diaphragm correction, and

these levels correlated with both the anti-GAA antibody titer and hepatic GAA activity.









The variability in hepatic GAA activity levels between mice receiving high dose

vector is likely due to the observed range in vector genome copy number. The variegated

immune response within both the high and low dose groups is likely attributed to

unidentified modifier genes in the mixed background of the Gaa--strain. A better

understanding of how both hepatic GAA expression levels and immune response

cooperate to impact cross-correction can be gained from controlling for the variability in

at least one of these two parameters.

We attempted to control for the immune response by tolerizing Gaa'- mice to

human GAA protein using a model of neonatal tolerance. The failure to successfully

induce tolerance in Gaa-1 mice was surprising as preliminary studies testing this method

of tolerance by challenge with rhGAA protein resulted in complete tolerance in 8 out of 8

mice. One potential explanation for a break in tolerance in 4 of the 5 pretreated mice is

greater overall immunogenicity of rAAV-derived human GAA versus direct enzyme

delivery. This theory is unlikely as anti-GAA antibody titers were approximately 6-fold

over background after high dose delivery versus the 100-fold over background titer

observed after direct enzyme delivery.

A second possible explanation would involve the difference in time lapse between

pre-treatment and protein challenge (8 weeks) versus pre-treatment and vector challenge

(10 weeks). If neonatal tolerance was occurring by T-cell suppression or inactivation,

and not deletion, it is possible that the absence of a stimulus for a sustained period of time

can re-activate suppressed or inactivated T-cells. A protocol involving multiple

injections of toleragen throughout the first two weeks may have been required to sustain

tolerance as demonstrated by others (31,32). Conversely, it is possible that neonatal









antigen exposure primed the TH2 response, as observed by others (37) prior to vector

delivery. If this were the case, we would have expected to see significantly higher

antibodies titers at the same time or earlier as in naive Gaa-' mice. A 2.5-fold higher

antibody titer was observed in 1 of the 4 pretreated mice that elicited an immune

response, however, this difference is within the range of variability observed within

treatment groups and is therefore not to likely a relevant indicator of T-cell priming.

A third potential explanation would involve structural differences between the

directly infused rhGAA and the vector-produced, hepatic-secreted hGAA protein. While

the amino acid sequence of the hGAA used in both the transfection of CHO-cells to

produce rhGAA and in the rAAV vector are identical, post-translational modifications of

GAA in CHO cells versus murine liver cells may vary. Additionally, pharmacological

modifications were made in the development of rhGAA to increase the degree of

phosphorylation for more efficient receptor-mediated uptake of circulating GAA.

Biochemical and Histological Assessment of Cross-Correction

Along with developing a further understanding of the immune response to

hepatic-secreted human GAA, we also wanted to assess the feasibility of rAAV-mediated

cross-correction from the liver for the treatment of GSDII. We were able to demonstrate

GAA activity levels that were > 20% of normal in the diaphragms of 100% of high dose-

treated mice. We were also able to achieve these proposed therapeutic levels of GAA

activity in the hearts of 45% of high dose-treated mice and the quadriceps of one mouse.

The relationship between percent restoration of activity and ability to clear

lysosomal glycogen after short-term studies in the mouse model is not yet fully

understood. While 100% normal heart activity resulted in obvious reduction in positive

PAS staining for glycogen, 50% normal heart activity levels led to no detectable









improvement. Due to difficulties in processing of diaphragm sections, histochemical

staining of the treated diaphragms is not available. However, Mah et al. (72)

demonstrated near complete reversal of glycogen accumulation in the diaphragm with

120% normal GAA activity levels obtained 6-weeks after direct delivery of rAAV-GAA

to the diaphragms of the same Gaa-' mouse model. A better understanding of the

relationship between age of treatment, levels of activity restoration and ability to reduce

accumulated lysosomal glycogen in different tissues will be necessary for both outcome

assessment during clinical trials and further defining corrective levels for Gaa-d mice.

The more stringent requirements for correction of the quadriceps and potentially

other skeletal muscles may be attributed to decreased levels of M6P receptor in hind-limb

skeletal muscles versus the cardiac and diaphragm muscle. Raben et al. (101) suggested

that, in skeletal muscle, type I fibers (slow-twitch) have greater levels of M6P receptor

than type II (fast-twitch). The diaphragm predominately consist of type I fibers, and

cardiomyocytes behave more like type I than type II skeletal muscle fibers. However, the

quadriceps are a mixture of type I and type II fibers. Examining the ability to correct

different types of skeletal muscles to confirm this theory will be beneficial for future

cross-correction studies.

Summary

In summary, we showed that the immune response to rAAV-derived hGAA in nafve

Gaa-' mice is inhibitory. We found that in naive Gaa'' mice, a high dose (1 x 1012 vg) of

vector is required for diaphragm correction. Additionally, we observed that diaphragm

correction is primarily dependent on antibody titer; presence of relatively low antibody

titers can substantially decrease GAA activity levels in the diaphragm. However GAA

activity can still be restored in the diaphragm in the presence of antibodies provided