1
DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR BIOLOGICS EVALUATION AND RESEARCH
BIOLOGICAL RESPONSE MODIFIERS ADVISORY COMMITTEE
OPEN SESSION
Meeting #32
Friday, May 10, 2002
8:10 a.m.
Hilton Hotel
Gaithersburg, Maryland
2
PARTICIPANTS
Daniel R. Salomon, M.D., Acting Chair
Gail Dapolito, Executive Secretary
MEMBERS
Katherine A. High, M.D.
Richard C. Mulligan, Ph.D.
Mahendra S. Rao, M.D., Ph.D.
Alice J. Wolfson, J.D. (Consumer
Representative)
TEMPORARY VOTING MEMBERS
Martin Dym, M.D.
Jon W. Gordon, M.D., Ph.D.
Thomas F. Murray, Ph.D.
Terence Flotte, M.D.
Eric T. Juengst, Ph.D.
R. Jude Samulski, Ph.D.
GUESTS/GUEST SPEAKERS
Valder Arruda, M.D., Ph.D.
Linda Couto, Ph.D.
Mark Kay, M.D.
Stephen M. Rose, Ph.D.
FDA PARTICIPANTS
Jay P. Siegel, M.D.
Philip D. Noguchi, M.D.
Daniel Takefman, Ph.D.
Anne Pilaro, Ph.D.
3
C O N T E N T S
PAGE
Welcome/Administrative Remarks
Dr. Daniel Salomon, Acting Chair 4
Introduction of Committee 5
Conflict of Interest Statement
Gail Dapolito, Executive Secretary 8
FDA Introduction
Potential for Inadvertent Germline Transmission
of Gene Transfer Vectors: FDA Approach for
Patient Follow Up
Daniel Takefman, Ph.D. 13
Guest Presentations
AAV Vector Biology, Jude Samulski, Ph.D. 23
Questions and Answers 46
Germline Transmission by Gene Transfer
Vectors: Assessing the Risk
Jon Gordon, M.D., Ph.D. 61
Questions and Answers 84
A Phase I Trial of AAV-Mediated Liver-Directed
Gene Therapy for Hemophilia B
Mark Kay, M.D., Ph.D. 98
Safety Studies to Support Intrahepatic
Delivery of AAV, Linda Couto, Ph.D. 116
Assessing the Risk of Germline Transmission of
AAV in a Rabbit Model
Valder Arruda, M.D. 130
Questions and Answers 144
Open Public Hearing
Mr. Steven Humes 177
National Hemophilia Foundation
Dr James Johnson, Patient 184
Dr. Kenneth Chahine, Avigen 190
Committee Discussion of Questions 197
4
1 P R
O C E E D I N G S
2
Opening Remarks
3 DR.
SALOMON: Good morning, everybody.
4 Welcome to day two of
the Biological Response
5 Modifiers Advisory
Committee Meeting No. 32. I
6 guess we should call
it 32B. We have got a title.
7 I have been
complaining and I finally got what I
8 wanted a title for
these meetings. This one, this
9 is good - Vector
Pellucida 2002. Not my title,
10 but, you know, you
can't criticize it, I got what I
11 wanted. Thank you.
12 So, welcome
everybody. Today we have
13 changed the scenery
around the table quite a bit.
14 So, to get reoriented,
I think we should go back
15 around again this time
and introduce ourselves, so
16 that both the
audience, as well as each other, has
17 a little sense of who
we are and what we are doing.
18 Just if you
can introduce yourself, we
19 will just go around
the table and give a few
20 sentences on where you
are from and what you do,
21 what kind of expertise
you bring.
22 In front of
you is a button on the thing.
23 It says speaker. If you push it, it turns red.
24 Talk, and then when
you are done, turn it off.
25 Otherwise, there is a
funny feedback. So if I am
5
1 ever looking at you,
gesturing, it means to turn it
2 off. It is one of my big duties.
3 Introduction of Committee
4 DR.
DYM: Martin Dym, Georgetown
5 University. I worked on the testis and
6 specifically on
spermatogonia, which are the male
7 germline stem cells.
8 DR.
FLOTTE: I am Terry Flotte from the
9 University of
Florida. We have been working on AAV
10 biology, AAV vectors
and AAV gene therapy.
11 DR.
JUENGST: I am Eric Juengst. I am in
12 the Department of
Bioethics at Case Western Reserve
13 University and
recently rotated off the RAC is
14 where my last
connection with these issues.
15 DR.
MURRAY: I am Tom Murray. I am from
16 the Hastings Center,
Bioethics, the world's first
17 bioethics research
institute, and my work has been
18 in a variety of
issues, but quite a lot in
19 genetics, parents, and
children.
20 MS.
WOLFSON: I am Alice Wolfson. I am
21 the Consumer
Advocate. In this incarnation, I am a
22 policyholder's lawyer
representing policyholders
23 against their
insurance companies when they don't
24 pay what they are
supposed to pay.
25 In my
previous incarnation, however, I am,
6
1 and was, a women's
health activist and a founder of
2 the National Women's
Health Network.
3 DR.
RAO: My name is Mahendra Rao. I am
4 in the Intramural
Program at the National Institute
5 on Aging. I am also a member of the BRMAC. I work
6 on stem cells, most
parts of the body, I guess.
7 DR.
SALOMON: Jude, we missed you the
8 first time around.
9 DR.
SAMULSKI: I am Jude Samulski from the
10 University of North
Carolina, and work in the area
11 of AAV vectors.
12 DR.
SALOMON: I am Dan Salomon. I have
13 the pleasure of
chairing the committee today. I am
14 from the Scripps
Research Institute in La Jolla,
15 California. I work on cell transplantation,
16 particularly islet
cell transplantation and tissue
17 engineering and
therapeutic gene delivery.
18 MS.
DAPOLITO: Gail Dapolito, Center for
19 Biologics. I am the Executive Secretary of the
20 committee.
21 DR. GORDON: Jon Gordon from Mount Sinai
22 School of
Medicine. I make a lot of transgenic
23 mouse models of
disease and gene therapy for
24 disease. I was on the RAC. I am actually the
25 first person to say
the word "transgenic," if that
7
1 means anything.
2 DR.
SALOMON: It means a lot.
3 DR.
PILARO: I am Anne Pilaro. I am an
4 expert toxicologist in
the Division of Clinical
5 Trials at CBER. I regulate a lot of the gene
6 therapy protocols, in
fact, I think I have 167
7 active right now.
8 DR.
TAKEFMAN: Dan Takefman. I am a gene
9 therapy product
reviewer with the Division of
10 Cellular and Gene
Therapies, CBER.
11 DR.
NOGUCHI: Phil Noguchi. I am director
12 of the Division of
Cell and Gene Therapy at CBER.
13 DR.
SALOMON: Welcome. We will be joined
14 a little bit later by
my colleague to the right,
15 Richard Mulligan from
Harvard Medical School.
16 This is
interesting for two reasons. One
17 is that this is kind
of a revisit to a very
18 important area that
the BRMAC dealt with, not the
19 last time, but I guess
at least two times ago,
20 where we initially
talked about how to address
21 potential regulatory
issues specifically with this
22 Avigen trial, and then
more generally with how to
23 deal with the
potential of infection germline in
24 this case with semen.
25 We got into
the whole discussion about
8
1 semen versus infecting
the motile sperm and what
2 was the evidence, if
any, that you could really
3 infect the germline,
the spermatogonia, or infect
4 the sperm themselves,
and very much tried to deal
5 with some of the
practical issues of what you would
6 demand of any company
of a sponsor in doing this
7 kind of research, and
to do it in such a way that
8 you wouldn't put an
unnecessary hold that could
9 therefore interrupt a
very important trial unless
10 there was awfully good
evidence.
11 It is also
very interesting in that it is
12 an interesting theme
for the two days. In some way
13 I am sorry that some
of you weren't here yesterday
14 where there we were
really talking about another
15 kind of germline
transfer issue, the injection of
16 ooplasm into oocytes
for infertile women, but it is
17 an interesting thing
now to go on to the idea of
18 potentially doing
something like this through
19 therapeutic gene
delivery.
20 We have to
read the conflict of interest.
21 Gail.
22 Conflict
of Interest Statement
23 MS.
DAPOLITO: I would just like to read
24 for the public record,
the conflict of interest
25 statement for today's
meeting.
9
1 Pursuant to the authority granted under
2 the Committee charter,
the Director of FDA Center
3 for Biologics
Evaluation and Research has appointed
4 Drs. Terence Flotte,
Jon Gordon, Eric Juengst,
5 Thomas Murray, Daniel
Salomon, and Jude Samulski as
6 temporary voting
members for the discussions
7 regarding issues
related to germline transmission
8 of gene therapy
vectors.
9 Dr. Salomon
serves as the Acting Chair for
10 today's session.
11 To determine if any conflicts of
interest
12 existed, the Agency
reviewed the submitted agenda
13 and all financial
interests reported by the meeting
14 participants. As a result of this review, the
15 following disclosures
are being made:
16 In
accordance with 18 U.S.C. 208, Drs.
17 Terence Flotte,
Jonathan Gordon, Daniel Salomon,
18 and Jude Samulski were
granted waivers permitting
19 them to participate
fully in the committee
20 discussions. Dr. Richard Mulligan was granted a
21 limited waiver for
this discussion which permits
22 him to participate in
the committee discussion
23 without a vote. Dr. Katherine High recused herself
24 from this committee
meeting.
25 In regards
to FDA's invited guests, the
10
1 Agency has determined
that services of these guests
2 are essential. The following interests are being
3 made public to allow
meeting participants to
4 objectively evaluate
any presentation and/or
5 comments made by the
guests related to the
6 discussions of issues
of germline transmission of
7 gene therapy vectors.
8 Dr. Valder
Arruda is employed by the
9 University of
Pennsylvania. He is involved in the
10 studies of
adeno-associated virus vectors. Dr.
11 Stephen Rose is
employed by the Office of
12 Biotechnology
Activities, NIH.
13 In the event
that the discussions involve
14 other products or
firms not already on the agenda,
15 for which FDA's
participants have a financial
16 interest, the
participants are aware of the need to
17 exclude themselves
from such involvement, and their
18 exclusion will be noted
for the public record.
19 With respect
to all other meeting
20 participants, we ask
in the interest of fairness
21 that you state your
name, affiliation, and address
22 any current or
previous financial involvement with
23 any firm whose product
you wish to comment upon.
24 Copies of
these waivers addressed in this
25 announcement are
available by written request under
11
1 the Freedom of
Information Act.
2 As a final
note, as a courtesy to the
3 committee discussants
and your neighbors in the
4 audience, we ask that
cell phones and pagers be put
5 in silent mode.
6 Thanks.
7 DR.
SALOMON: Thank you, Gail.
8 What we will
do here is begin with an FDA
9 introduction from Dan
Takefman, will kind of walk
10 us through some of the
key issues that the FDA
11 wants to answer. Remember that part of the dynamic
12 here is that we are an
FDA Advisory Committee.
13 There will
be times when we all, certainly
14 myself as a scientist,
get really interested in
15 some scientific
question, but at some point you
16 will have to forgive me
if we steer away from that
17 since, if we are not
really answering the FDA's
18 question, then, we are
not doing what we are
19 supposed to be doing
here.
20 In the
meantime, though, obviously, to the
21 extent that any of
these scientific issues are
22 relevant to answering
the questions, you know, you
23 obviously are here and
your expertise is greatly
24 welcomed.
25 I guess the
other thing, as long as I am
12
1 giving an introduction
on that score, I will just
2 say that we are going
to try and come to consensus
3 on some of these
questions, but in some instances,
4 there is no consensus,
and there is no effort here
5 on my part to force
this group into consensus, so
6 well-articulated,
minority opinions or even just
7 where we go, I am
sorry, but there is no way we can
8 agree on it, that's
the kind of information that we
9 need to pin down.
10 So it is
important for us to make sure
11 that we have
represented everything as evenly as
12 possible for the
community. The last thing I will
13 say to the audience is
that I feel you also are
14 participants in this
meeting. This is an open
15 public meeting. That mike in the center is open. I
16 welcome all of you, if
you have something to say,
17 to come up during the
meeting during discussion and
18 make your points, and
we will definitely be here to
19 listen to them and try
and make sure that we do an
20 adequate discussion of
this.
21 Dan, you are
on.
22
FDA Introduction
23 Potential for
Inadvertent Germline Transmission of
24 Gene Transfer Vectors: FDA Approach for
Patient
25
Follow Up
13
1
Daniel Takefman, Ph.D.
2 DR.
TAKEFMAN: Thank you. I would like to
3 welcome the committee
and speakers, and thank
4 everyone for
participating in today's meeting.
5 [Slide.
6 The topic
for today is the discussion of
7 potential for
inadvertent germline transmission of
8 gene transfer vectors,
and as Dan said, this has
9 been a topic of
previous discussions and public
10 meetings. Today, we will be discussing the finding
11 of vector sequences in
patient semen and to discuss
12 FDA's current approach
for patient follow up.
13 [Slide.
14 Concerns
regarding inadvertent germline
15 transmission, or IGLT,
are twofold.
16 Societal/ethical
concerns are based on previous
17 public discussions and
publications in which
18 deliberate germline
alteration has been deemed
19 unacceptable.
20
Additionally, there are potential adverse
21 biological effects,
such as genetic disorders,
22 birth defects, and
lethality to developing fetus,
23 just to list a few
which are also of concern.
24 [Slide.
25 What is the
likelihood that IGLT would be
14
1 deleterious? Well, retroviruses have been used as
2 tools to investigate
the role of certain genes
3 which are important in
development. I refer to, in
4 this slide, data
involving retroviral insertion to
5 the germline of mice
and as a specific example, a
6 retrovirus was used to
infect a murine blastocyst.
7 In this case, this
infection resulted in a mouse
8 strain with a lethal
embryonic mutation, which was
9 induced by proviral
insertion into the alpha-1
10 collagen gene. This mutation was recessive, so
11 that the phenotypic
effect required homozygosity.
12 [Slide.
13 So data
exist suggesting that in the case
14 of retroviruses,
deliberate insertion into the
15 germline may be
deleterious, but what about data
16 from preclinical
animal studies regarding the
17 ability of gene
transfer vectors to transmit to the
18 germline?
19 Well, the
FDA does require biodistribution
20 studies with gene
transfer vectors in relevant
21 animal models. These biodistribution studies,
22 performed in support
of clinical trials, have shown
23 evidence of vector
dissemination to gonadal tissue.
24 However, in
most studies, vector sequences
25 have not been detected
in semen samples, and the
15
1 point I need to make
in regards to these
2 preclinical studies is
that they are not always
3 predictive of human
experience.
4 A case in
point is today's topic in which
5 vector sequences were
found in semen from clinical
6 trial subjects,
however, initial preclinical
7 studies, such as those
done in dogs, demonstrated
8 no detectable vector
in semen.
9 Again,
certainly in today's case, animal
10 studies are not always
predictive.
11 [Slide.
12 I would like
to give an update on the kind
13 of current active gene
transfer INDs we currently
14 have in file just to
give you an idea of what is
15 being used in the
clinic.
16 You can see
here in regards to retroviral
17 vectors, they are
predominantly being used in ex
18 vivo types of gene
transfer studies, while
19 adenoviral vectors and
plasmids are often being
20 used in direct in vivo
type of administrations.
21 You will
notice here with AAV vectors,
22 compared to other
systems, FDA has seen relatively
23 few gene transfer
INDs. Of the few we have, they
24 are primarily in vivo,
localized injection type of
25 administrations.
16
1 [Slide.
2 I would like
to go over some of the
3 factors that FDA
considers important for assessing
4 risks of inadvertent
germline transmission of gene
5 transfer vectors.
6 Certainly,
integration potential of the
7 vectors is important
to consider. Of the current
8 vectors being used in
the clinic, FDA is
9 considering both
retroviral and AAV vectors as
10 vectors with potential
to integrate. Certainly
11 with retroviruses, as
well as lentiviral vectors,
12 they are known to have
efficient abilities to
13 integrate and host
genomes.
14 In terms of
AAV vectors, this system is
15 not as clearly worked
out as in other systems, such
16 as retroviruses. FDA is currently considering AAV
17 vectors as having a
low, but potential to integrate
18 in vivo, and I
specifically refer here to a couple
19 of papers from Nakai's
lab in which he showed low
20 levels of integration
in mouse livers.
21 [Slide.
22 The risk of
inadvertent germline
23 transmission is also
likely highly dependent upon
24 route of
administration. An ex vivo gene transfer
25 would likely represent
a minimal risk in terms of
17
1 IGLT, while at the
other end of the spectrum, a
2 systemic injection
would represent a relatively
3 higher risk in terms
of transfer to the germline
4 via hematogenous
spread.
5 [Slide.
6 As Dr.
Salomon mentioned, IGLT has been a
7 topic of discussion, and
I would like to go over
8 some of the previous
public discussions in order to
9 put today's meeting in
a little perspective.
10 Beginning
with the March 1999 RAC meeting,
11 here, there was a
focused discussion on preclinical
12 data which
demonstrated gonadal distribution. It
13 was the consensus from
this meeting that despite
14 this preclinical data,
the probability of
15 inadvertent germline
transmission occurring during
16 a gene transfer
clinical trial was low.
17 However,
further discussion became
18 necessary at the
November 2000 BRMAC meeting. At
19 this meeting, we heard
data from a trial which
20 involved I.V.
administration of a gammaretroviral
21 vector which contained
the factor VIII gene for
22 treatment of
hemophilia A.
23 I should
point out this was the first
24 trial under IND which
involved I.V. administration
25 of a gammaretroviral
vector. Data was presented in
18
1 which 1 out 12
subjects treated had vector
2 sequences transiently
present in semen.
3 In the one
patient, vector sequences were
4 detected at only one
time point by DNA-PCR.
5 [Slide.
6 Then, at a
recent meeting of the RAC, a
7 trial was presented,
which will also be presented
8 today, which involved
an AAV vector, which contains
9 the factor IX gene for
the treatment of hemophilia
10 B. This is the first trial under IND which
11 involved
administration of an AAV vector into the
12 hepatic artery.
13 Data was
presented in which vector
14 sequences were found
in semen of the first two
15 patients treated. The first patient had positive
16 PCR signal at multiple
time points for up to 10
17 weeks post
administration, and the implication here
18 is that all patients
treated in this trial may test
19 positive for vector
sequences in semen samples.
20 [Slide.
21 So to
summarize some of the consensus from
22 these public
discussions, there was a consensus
23 from the RAC meeting
on preclinical data that the
24 probability of
inadvertent germline transmission is
25 low and that the use
of a fertile subject
19
1 population was
acceptable.
2 From the
BRMAC meeting, the committee
3 agreed with FDA's
approach to institute a clinical
4 hold when vector
sequences are detected in semen
5 samples from study
subjects.
6 There was a
consensus from both the RAC
7 and the BRMAC that
there is a need to determine if
8 vector is associated with
sperm cells. Using
9 fractionation methods,
such as density separation,
10 potential
contaminating transduced white blood
11 cells can be removed
from sperm cell fractions.
12 You are going to hear
more later on from Avigen on
13 their fractionation assays.
14 [Slide.
15 I would like
to turn now to FDA's approach
16 for patient follow up,
which has been modified in
17 response to these
public discussions and from data
18 regarding this current
trial.
19 Prior to
initiation of the trial, of
20 course, if during
preclinical animal studies,
21 vector is found in
gonadal tissue, this finding and
22 the potential for
germline alterations should be
23 included in informed
consent documents.
24 [Slide.
25 As for FDA's
current approach for patient
20
1 follow up, if semen
from clinical trial subjects
2 tests positive for
vector sequences, the clinical
3 trial will be allowed
to continue, however, FDA
4 will request timely
follow-up testing of
5 fractionated
semen. As has been in the case in the
6 past, barrier
contraception is requested until
7 three consecutive
samples test negative.
8 [Slide.
9 Now, if the
motile sperm fraction tests
10 positive for vector
sequences, FDA will institute a
11 clinical hold and
subject enrollment will be
12 stopped until it is
determined that the signal from
13 the motile sperm
fraction is transient, and
14 specifically, we are
asking for serial fractionated
15 samples to test
negative three times over three
16 consecutive monthly
intervals.
17 [Slide.
18 I would like
to turn now to some of the
19 concerns that FDA
has. Specifically, the finding
20 of vector sequences in
semen may become more
21 common. Certainly with subject from trials
22 involving systemic or
intrahepatic administration
23 of AAV, such as in
this trial, every patient
24 treated might have
vector sequences found in semen
25 samples.
21
1
Additionally, we have new vector classes
2 on the horizon, such
as lentiviral vectors, which
3 we know have a high
potential to integrate, and
4 there is also new
production technologies which
5 allow for higher titer
viruses to be produced and
6 new clinical
applications of gene delivery systems
7 designed to increase
transduction efficiency, all
8 of which may make the
detection of vector sequences
9 in subject semen more
prevalent in future clinical
10 trials.
11 [Slide.
12 Of
particular concern, the fact that
13 patient follow up is
difficult with certain
14 populations. Obviously, there are technical
15 limitations in the
ability to monitor women and
16 certain men who are
unable to repeatedly supply
17 adequate samples. There is technical limitations
18 to monitor these
subject populations for evidence
19 of germline
alterations.
20 The specific
concern will be re-presented
21 in the form of a
question to the committee for
22 discussion in the
afternoon session.
23 [Slide.
24 To
summarize, FDA's primary concern of
25 inadvertent germline
transmission of gene transfer
22
1 vectors is with
systemic administration of
2 integrating vectors.
3 A clinical
hold is instituted only if
4 vector sequences are
detected in motile sperm
5 fractions, and the
inability to monitor certain
6 patient populations is
a concern and warrants
7 further discussion.
8 I will end
here and just remind everyone
9 that there is a number
of background talks and
10 still data on the
clinical trial and preclinical
11 studies to be
presented, so I would request that we
12 limit the majority of
discussion of patient follow
13 up until the afternoon
session, but I will be happy
14 to answer a few
questions at this time for
15 clarification.
16 DR. SALOMON: Thank you, Dan.
17 Are there
any questions from the committee
18 to the FDA regarding
the overall umbrella charge
19 that we have for
today? Okay.
20 The next are
two presentations. It is a
21 pleasure to start with
Jude Samulski from the
22 University of North
Carolina to talk to us about
23 the biology of AAV
vectors.
24
Guest Presentations
25
AAV Vector Biology
23
1
Jude Samulski, Ph.D.
2 DR.
SAMULSKI: It is a pleasure to be
3 here. I want to thank Daniel for asking me to come
4 up. He requested that I give some type of overview
5 of AAV biology and try
to focus a little bit on our
6 understanding of the
potential for integration and
7 mechanisms.
8 I think what
I am going to do is offer you
9 an opinion of a
consensus of what we think is
10 happening in the
field, point you in the direction
11 of probably papers
that are relevant, that start to
12 show trends that are
happening, but more than
13 likely I am going to
end up with the conclusion
14 that Daniel has
already described, is that AAV is
15 somewhere on that
curve as a vector that can
16 integrate, the
efficiency is not well established,
17 but the potential is
there.
18 I will start
off by introducing you to the
19 life cycle of this virus. In the laboratory, an
20 AAV particle can have
a lytic component or a latent
21 component, so we refer
to it as a biphasic life
22 cycle.
23 It has been
established that it is
24 dependent on a helper
virus in order to go through
25 a productive lytic
cycle, and in this setting, the
24
1 virus goes in,
reproduces, and progeny comes back
2 out.
3 What was
established in the laboratory in
4 the early seventies
was that if you took AAV
5 particles and put them
in cells in the absence of
6 the helper, you could
see this persistence, what
7 was referred to as
"latency," and in this setting,
8 it was determined that
the virus was establishing
9 an integration event
in the chromosome, and in this
10 integration event, it
appeared to be targeting,
11 meaning it was going
to a specific locus in the
12 human genome.
13 This was all
done in vitro and tissue
14 culture cells, and to
complete the biological life
15 cycle, if you take
these cells and now superinfect
16 them with adenovirus,
AAV has the ability to come
17 back out of the
chromosome and reenter its lytic
18 component.
19 So in the
laboratory, it was established
20 the mechanism in which
we could argue how AAV,
21 which was found in
nature in clinical isolates of
22 adenovirus, how these
two would co-persist, but we
23 could also explain a
question of what is the
24 consequences of AAV
infecting the cell in the
25 absence of its
helper. Is that genetic suicide?
25
1 That answer was no,
the virus has a mechanism of
2 persistence.
3 I should
argue that there is absolutely
4 zero data of AAV
integration in humans. This is
5 all established in
vitro, and it is inferred that
6 this mechanism can
take place.
7 I should also mention that the early
8 studies of AAV showing
up in clinical isolates, it
9 has only been isolated
in adenovirus, although
10 herpes can supply the
same helper function. There
11 has never been a
clinical isolate of herpes that
12 has had a
contamination of AAV.
13 So what you
should be asking yourself is
14 that we can mimic a
paradigm in tissue culture and
15 substitute other
viruses, but what appears to be
16 out there in nature is
this co-relationship. This
17 was established in
vitro, and it is presumed that
18 this can also happen
in vivo.
19 The genome
is fairly simple. It is about
20 5,000 base pairs, and
what is of importance today
21 is paying a little bit
of attention to what is
22 referred to as the Rep
genes and the inverted
23 terminal repeats of
the virus, which are the
24 origins of
replication, the packaging signal, and
25 what appear to be the
break points that join
26
1 recombination events
with the chromosome.
2 Of the Rep
genes that are made, it has
3 been shown that it is
the large Rep proteins, Rep
4 78 and 68, that appear
to be responsible for the
5 integration
events. I just want to point out that
6 in AAV, these are
identical proteins. They only
7 differ by a splice
variate, and in the absence of
8 adenovirus, this is
the dominant protein that you
9 see in the presence of adenovirus. This comes on
10 first and then it
switches over to Rep 68.
11 They all
have enzymatically identical
12 activities. They bind
to the AAV terminal repeat
13 and what is called a
Rep binding element. They
14 have a site-specific,
strand-specific endonuclease
15 activity where they
can nick this molecule, and
16 they have helicase
activity which allows it to
17 unravel to DNA.
18 So we see a
relationship with the Rep
19 proteins were the key
element on the virus, which
20 is the origin of
replication, showing that it has a
21 binding site, a
nicking site, and enzymatic
22 activities to allow
this virus to replicate.
23 So the first
evidence of AAV integrating
24 site specifically was
generated in Ken Burns' lab
25 in 1996, and in this
study, what they did was
27
1 pulled out some
junctions, sequenced the junctions,
2 and went back and used
those sequences as probes.
3 This is just
a representative example from
4 our lab that shows
that if you look at your
5 chromosome 19 locus in
a control cell, it is about
6 a 2.6 kilobase fragment,
but after you integrate
7 and establish
independent clones, you can find
8 variance that show
evidence that the chromosome
9 sequence now has a
rearrangement suggestive of an
10 insertion, and some of
these are multiple fragments
11 showing that there is
amplification and
12 rearrangement.
13 If you take
a blot like this and strip off
14 the chromosome 19
probe and then come back with the
15 viral probe, you can
see there is co-segregation of
16 these viral sequences
with these chromosome 19
17 rearranged, so this
was the data that said there
18 was a preferred site
of integration, a
19 rearrangement of
chromosome 19 and a
20 co-localization of
these sequences with chromosome
21 19 sequences.
22 Ken Burns
and others looked in detail to
23 bring to try to
understand why was this virus going
24 to this specific
locus, and from that study came
25 the following
information.
28
1 There is an
identical Rep binding site and
2 a nicking site located
on human chromosome 19, so
3 what we had was a
mechanism that is virtually of
4 viral origin sitting
on chromosome 19, that gave a
5 putative reason for
why this site is preferred as
6 an integration locus
over any other sequence in the
7 human genome.
8 What I
should point out is that further
9 studies have shown
that not only is the Rep binding
10 required, the spacing
between this binding site to
11 the nicking site and
the nicking site itself, so if
12 you take these
sequences and count them up, there
13 are over 15 base
pairs.
14 It is argued
that a sequence over 15
15 nucleotides is only
represented one time in the
16 human genome. This is probably why this virus is
17 only targeting this
locus. This element is present
18 in about 200,000
copies in the human genome, which
19 would argue that the
Rep protein is sitting on lots
20 of spots on the human
chromosome, but it is only
21 when it is this
context that it can initiate the
22 event to promote the
integration step.
23 So we have a
model and a mechanism that is
24 being supported both
in vitro and in vivo.
25 A group in
Italy went on to show that the
29
1 site has an open
chromatin confirmation and that it
2 is not a closed site,
so it is not a site that is
3 unaccessible. All of these things are beginning to
4 support the type of
DNA structure that AAV needs to
5 see in order to go
into the chromosome.
6 A number of
labs, including our own, have
7 gone after looking at
these integration events, and
8 most of you are pretty
well aware, that if you look
9 at retroviral
integration event, it is a fair
10 precise cut and paste
mechanism in which it cuts
11 the chromosome,
integrates its genome, and there is
12 like a 3 to 5
nucleotide duplication on either
13 side.
14 When you
looked at these AAV proviral
15 structures, what we
saw was there were a lot of
16 tandem repeats,
amplification events, and all of
17 these things were
supporting a type of integration
18 that was completely
different than the
19 well-characterized
retrovirus integration.
20 This has
been consistent both in cell
21 lines, as well as
episomal integration events, as
22 well as in vitro
systems, so there is a mechanism
23 for integration that
is not consistent with a cut
24 and paste. It is referred to as a non-homologous
25 amplification
mechanism.
30
1 Our lab and
others went on to look at the
2 break points between
the viral terminal repeat,
3 which I showed you has
this origin activity, and
4 this hairpin
structure, and the junctions between
5 that and chromosome
19.
6 What you can
see was there was very little
7 fidelity and
conserving the integrity of the
8 terminal repeat. You would get break points that
9 were scattered
throughout these hairpins, and these
10 are just positioned
here on the sequence to give
11 you an impression that
there is no fixed break
12 point between the
viral sequence and the chromosome
13 19. They cluster around this hairpin element,
but
14 other than that, you
can virtually find break
15 points throughout
these sequences.
16 If you look
at that from a biological
17 point of view, it
again suggests that AAV may have
18 a problem in retaining
its integrity as a virus if
19 it's indiscriminately
breaking these hairpins and
20 going into the
chromosome, but this virus has a
21 phenomenal ability of
carrying out a step code gene
22 correction.
23 There is
technically two copies of every
24 sequence in the
hairpin, and since there is two
25 hairpins, there is the
total of four copies on the
31
1 virus, so between all
of these copies, the virus
2 will gene convert back
and forth and regenerate
3 these sequences with
fair efficiency, so you always
4 get a wild-type virus
coming back out even though
5 what is integrated in
the chromosome may be
6 somewhat fragmented.
7 Because the
virus also integrates in what
8 appears to be
head-to-tail concatemers, it is
9 preserving the
integrity of these hairpins
10 internally, and again
allowing it to use it as a
11 template to amplify
and come back out of the
12 chromosome.
13 So to get to
the mechanism, Matt Weitzman
14 in Roland Owens' lab
did an experiment in the early
15 nineties that said
that they could show that the
16 Rep protein of AAV
could form a complex between the
17 terminal repeat of the
virus and this
18 pre-integration site.
19 Again, this
made logical sense because
20 there was the same Rep
binding element on both of
21 these sequences. This
is just an illustration from
22 Sam Young's data
showing the Rep protein bound to
23 the terminal repeats
of an AAV vector. It has an
24 extremely high
affinity for the sequence and a Rep
25 complex binding to the
same element on chromosome
32
1 19. It was data like this and other that began
to
2 propose a model that
the virus express its Rep
3 protein, it binds to
this element on chromosome 19.
4 In vitro,
Rob Cotton showed that this is
5 sufficient to start a
synchronized single-stranded
6 DNA replication. So now you have this region of
7 chromosome 19 serving
as an origin. Since the Rep
8 protein is terminally
attached to this chromosomal
9 sequence, and you can
reinitiate, we feel that
10 there is a number of
initiation events that are
11 taking place on this
region of chromosome 19.
12 It should be
understood that there is an
13 enzyme called Fen-1
which is a host enzyme, that
14 actually repairs this type
of repeated initiation
15 event, however, if you
have a hairpin or a protein
16 attached to this, it
doesn't have the ability to
17 correct these
sequences.
18 So what
happens is you see recombination
19 events taking place to
resolve these molecules. It
20 has been suggested
that the AAV genome, which has
21 Rep, allows for
Rep-Rep tethering mechanism, as
22 Weitzman showed, and
at this point it is all host
23 enzymes that are
involved in inserting this
24 sequence into the host
genome, and this type of
25 tandem repeat,
head-to-tail type of format.
33
1 This is data
that was provided to me by
2 Regina Hildabraun. It is not published. It is
3 coming out in a
journal Virology. She has
4 developed a real-time
PCR assay to look at the
5 efficiency of AAV
viruses to go to chromosome 19.
6 It is a PCR assay that
look at the terminal repeat
7 and a locus on
chromosome 19.
8 What I think
is important to see here is
9 that she can score
integration events taking place
10 over the first 72
hours or so, but the most
11 important thing is
that the wild-type virus, which
12 she is seeing an
integration event for about 1,000
13 particles, so it is
suggest about 0.1 percent of
14 all the AAV virus is
capable of carrying out
15 integration.
16 This is
completely different than like the
17 retroviruses where it
is 100 percent integration.
18 As Daniel
said, there is a propensity for
19 the virus to
integrate. The efficiency is what
20 needs to be look at in
this setting.
21 This is a
paper that was published by
22 Ernst Winocour. I think this is of importance
23 because what I am
going to suggest to you is this
24 is another parvovirus
called minute virus in mice.
25 It's an autonomous
parvovirus. Nowhere is its life
34
1 cycle does it
establish latency. It has no
2 mechanism. There has never been any data
3 supporting it.
4 But what
Ernst was able to do was show
5 that these viruses
also have terminal repeats, they
6 also have Rep-like
proteins, and that he could take
7 an episome substrate
and show that this virus could
8 also integrate into a
target sequence if the Rep
9 protein on this minute
virus was present and if the
10 subsequent sequences
were available.
11 So what I
think this is suggesting is that
12 the parvoviruses have
proteins that are involved in
13 replication that are
able to carry out nicking and
14 helicase activity on
substrates. In the case of
15 minute virus of mice,
there is no target in the
16 genome.
17 In the case
of AAV, there is an origin
18 identical to AAV
sitting on chromosome 19. So the
19 question may be, does
AAV really set up a latency
20 or is this an
interaction between Rep proteins and
21 target sequences, and
1 percent begins to suggest
22 that it is not a very
efficient mechanism.
23 I am going
to shift gears and now talk to
24 you about vectors
because I think this is where
25 most of the interest
is. In the laboratory, a
35
1 number of people
generate vectors by different
2 procedures.
3 In our lab, we
use plasmids to start to
4 make the vector, so
now we only retain the terminal
5 repeats. The gene of interest is in the middle.
6 You have a helper
plasmid carrying the Rep and
7 capture genes, and
another plasmid carrying the
8 essential sequences
from adenovirus to activate all
9 of these steps.
10 What happens
when all of these are in the
11 cell, you produce a
single virus particle, which is
12 an AAV particle
carrying the foreign gene of
13 interest. If you take these viruses and put them
14 in tissue culture
cells, and put them under
15 selection, what you
see is if you go to the
16 chromosome 19 region
and look at individual clones
17 that had the vector
integrated in the human genome,
18 you don't see a
significant rearrangement under
19 chromosome 19
sequence.
20 So unlike
wild type where it appeared that
21 70 to 90 percent of
the integrations were targeting
22 this locus, the
vectors have lost this ability to
23 go to chromosome 19.
It has been shown by a number
24 of labs that if you
add Rep back to this reaction,
25 these vectors will go
to chromosome 19 and
36
1 integrate.
2 So it is
fairly well established now that
3 AAV vectors have no
targeting capacity and that
4 what they do have is
the capacity to integrate into
5 the chromosome under
these selected conditions.
6 This is an approach that Charley Yang took
7 in the lab about seven
years ago, in which he made
8 AAV vectors that were
carrying a plasmid origin and
9 ampicillin sequence,
as well as a selectable
10 mechanism to look at
selection in eukaryotic cells.
11 He made this
into a virus, allowed it to
12 integrate into the
chromosome, and he used enzymes
13 that were cut outside
of the viral DNA, closed this
14 up into a circle, and
pulled out these so-called
15 cellular junctions, and when he characterized
16 these, he came up with
the following results.
17 The break
points of the terminal repeat
18 and the chromosome
were almost identical to what we
19 saw with wild
type. They clustered around the
20 hairpin structure, but
there was no defined break
21 point in any of these
vectors.
22 When we
looked at the location that they
23 were going into, they
appeared to be random on
24 chromosome 17, 7,
1. We had two examples of it
25 integrating on
chromosome 2. But what we were
37
1 seeing was that all of
the characteristics of
2 integration were
identical to wild type. It is
3 just that their targeting ability was lost.
4 Instead of going to
19, it was random.
5 If you look
at the vectors, they were
6 again consistent with
this head-to-tail mechanism
7 and amplification
event or rearrangement event. I
8 should mention that
David Russell has just
9 published a little
paper in Nature Medicine that
10 has shown another
clustering of these things pulled
11 out of HeLa cells, and
we have generated the exact
12 same information. There is breakage and
13 duplication and some
type of random repeats that
14 are being generated.
15 So I want to
point out because I think we
16 get misled a lot when
we think about AAV's
17 integration and that
it is something special. This
18 ability to form
concatemers is something that was
19 documented a number of
years ago by Schimke's lab.
20 In fact, if you look
at any transgenic animal that
21 has ever been
generated, it is always generated in
22 a head-to-tail
concatemer formation.
23 If you look
at virtually any cell line
24 that is established by
plasmids to give stability,
25 it is typically a
head-to-tail concatemer, that is
38
1 going into the
chromosome. So what we see is that
2 AAV is probably using
host enzymes to generate
3 these concatemers that
eventually go into the
4 chromosome.
5 As I
mentioned to you, without the Rep
6 protein, there is no
targeting capability. This
7 integration appears to
be random. The insertion
8 that takes place at
the integration site is not a
9 cut and paste
mechanism, it's a deletion,
10 amplification,
rearrangement, illegitimate type of
11 recombination.
12 This is just
our data showing all of the
13 break points that we
have generated both with
14 vectors with wild type
AAV as far as the junctions
15 that are generated
between the terminal repeats and
16 the chromosome, and
you can see that again there
17 are preferred
clustering sites, but there is no
18 distinct break point
that takes place between AAV
19 molecule and the
chromosomal DNA sequence.
20 We concluded
from this study that when AAV
21 vectors go into cells,
it is cellular recombination
22 pathways that are
responsible for the integration
23 of that, and that
there is no viral participation
24 in this enzymatic
step, it is all carried by
25 cellular
recombination.
39
1 If you look
at the data that has been
2 generated, it falls
under the category of an
3 illegitimate,
non-homologous recombination. This
4 would be true if you
put in plasmid DNA,
5 oligonucleotides, any
piece of DNA that ends up
6 going into the
chromosome. It is following a
7 pathway that supported
cellular enzymes carrying
8 out the integration step.
9 I want to
just summarize this and then I
10 am going to switch to
the last third of the talk,
11 which is going to just
talk about information
12 generated with vectors
in animals.
13 Right now,
AAV vectors do not target
14 chromosome 19. They are identical to wild type
15 with respect to the
terminal repeat break points.
16 They are essentially
identical at this level. The
17 head-to-tail
orientation of vector proviruses, you
18 can find tail-to-tail
and head-to-head, but this is
19 pretty much the
dominant species you will see.
20 They
rearrange to chromosome integration
21 site. There is not a
cut and paste mechanism.
22 There is always some
type of deletion,
23 amplification, and
rearrangement that takes place
24 at the integration
locus.
25 So by all
these criteria, AAV fits the
40
1 conditions of an
insertional mutagen. It has the
2 ability to go into the
chromosome, and the critical
3 question is at what
frequency does it carry out
4 this insertion event.
5 This is
where I think we began to
6 accumulate data in the
field that drifted us away
7 from all that
information that was derived in
8 vitro, and you should
understand that the data was
9 derived in vitro was
under selected conditions with
10 a gene, such as G418
or neomycin, so that you are
11 only looking at the
integration events.
12 In vivo, the
first data that began to
13 suggest that this may
not be consistent with what
14 was happening in vitro
was actually carried out in
15 Terry Flotte's lab
where they were looking at
16 adeno-associated
viruses in monkeys after
17 administration for
airway gene delivery.
18 When they
characterized this, they saw
19 that the virus was
persisting for a period of time
20 and the virus could be
rescued completing all of
21 those steps that we
talked about in the life cycle,
22 but it was showing up
as an episome. There was
23 very little data
suggesting that this type of
24 persistence was taking
place as an integration
25 event.
41
1 This is a
paper that I would like to
2 direct people to,
because I think buried in this
3 paper is some really
important information. This
4 was a study carried
out in Jim Wilson's lab where
5 what he virtually did
was an in vivo selection like
6 what we do with in
vitro selection with G418, in an
7 animal model that had
a disease for the liver, so
8 the AAV vector was
transducing a gene and to
9 deliver, that he could put a selective
pressure on.
10 This
selective pressure meant that if this
11 liver was to survive,
the virus had to integrate.
12 After it integrated,
you could see nodules begin to
13 grow of liver cells. He characterized those
14 nodules. He showed they had integration events in
15 them. They were similar to what I have just
16 described for in
vitro.
17 They were
tandem repeats, rearrangements,
18 and an illegitimate
recombination mechanism, but if
19 you go into the paper
and dig at the multiplicity
20 of virus that he was
putting into the liver, 1012
21 particles per liver,
he was only getting about 0.1
22 percent of the liver
cells showing an integration
23 event.
24 So I think
what Daniel was referring to is
25 where does AAV fit on
this curve of an obligated
42
1 integration event
versus the potential to
2 integrate, and this
study, under selective
3 pressure, there was a
frequency that was derived,
4 which I think may be
telling to the type of numbers
5 that may happen in the
absence of selection.
6 I point to
these last two papers only
7 because it has been
characterized in extensive
8 detail in muscle, and
I bring up Phil Johnson's
9 study because he now
has an abstract that is going
10 to be presented as
ASGT, where he is showing that a
11 majority of what I
think he calls 98.5 percent of
12 all the vectors that
are in skeletal muscle are
13 persisting in episomal
form.
14 He does a
real-time PCR assay. I am not
15 going to try to
describe his data, it is written in
16 an abstract form, but
I think it is something that
17 the field in general
will want to look at and see
18 if this will be
something that can be used for
19 other target tissues.
20 But it is
consistent with the theme. What
21 I did not talk about
here today was any of the data
22 that Mark and Kathy
have generated, because I know
23 they are going to
speak later and they can tell you
24 specifically what has
been derived in their hands,
25 but I think the theme
is we see what these vectors,
43
1 they have the
propensity to set up a persistence,
2 the data that has been
generated in liver, muscle,
3 lung, and brain is
that episomal forms that are
4 predominantly seen,
but there is always the
5 potential and evidence
for integration.
6 This is the
last paper that I am going to
7 point you to, and I am
going to just mention this
8 because I think this
is going to give us a starting
9 place to begin to
understand AAV integration in
10 whole animal.
11 Terry Flotte
and his lab have generated
12 some data showing that
the DNA-dependent protein
13 kinase, the gene that
has mutated in SCID mice,
14 seems to have an
impact on the molecular phase of
15 AAV genomes.
16 Again, I am
going to paraphrase what
17 Terry's data says, and
he can speak to it in more
18 detail because he has
got new data that is a little
19 bit more
extensive. It appears that if you knock
20 out this protein
kinase, which is involved in
21 immunoglobulin
rearrangement as one example of its
22 role in the human
cell, the virus appears to
23 integrate more efficiently
into the chromosome.
24 This is an
enzyme that plays a role in
25 end-to-end joining,
and it seems that if you lose
44
1 the ability of these
host enzymes to form the
2 so-called concatemer
structure that we all
3 characterize, you can
see an increase in
4 integration event
takes place.
5 So it
appears that if you are defective in
6 one pathway, AAV will
just follow another host
7 mechanism for
persistence, which is an integration
8 mechanism.
9 Again, if
there are any specific
10 questions, I will ask
you to direct them to Terry
11 where he can give you
the details of what is going
12 on, but what this data
tells me is that we probably
13 we will be able to
identify these so-called
14 cellular recombination
pathways that are
15 influencing AAV
vectors when they go into so-called
16 non-dividing tissue.
17 I am going
to conclude by trying to
18 reemphasize the
following points. Wild type and
19 AAV vector integration
is not very efficient, and
20 this fairly well
documented in vitro. It is
21 something that seems
to be a theme that is
22 recurring in vivo.
23 If you look
at the ability of the virus to
24 target chromosome 19,
it is absolutely dependent on
25 a viral protein called
Rep. The mechanism is now
45
1 well understood
because they are identical binding
2 sites to facilitate
this targeting.
3 AAV vectors,
which do not have Rep
4 protein, do not have
the ability to go to
5 chromosome 19 into the
site-specific manner. If
6 you look at the
proviral structure of wild type AAV
7 and vector DNA, they
are essentially identical at
8 all levels.
9 The break
points and the terminal repeats,
10 the amplification, the
concatemerization, and the
11 rearrangement under
chromosome sequence is
12 identical whether it's
on chromosome 19 or randomly
13 inserted throughout
the genome.
14 Finally,
with the limited number of
15 studies that are being
done, it appears that in
16 non-dividing cells in
vivo, the AAV vectors exist
17 predominantly in an
episomal form, and again, I
18 will conclude.
19 Daniel
basically summarized the AAV field
20 by saying it has the
propensity to integrate into
21 the chromosome, where
it fits on that rheostat as
22 being very efficient
or not efficient, I think it
23 is going to be
dependent on more studies in vivo in
24 which we can continue
to accumulate data.
25 But as of
today, what we keep seeing is
46
1 some propensity for
this episomal form, but the
2 risk is still there,
and I will stop there and take
3 questions.
4 DR.
SALOMON: Thank you very much. Very
5 interesting.
6
Q&A
7 I have a
couple of questions that kind of
8 occurred to me in the
setting of thinking about
9 this thing riskwise.
You have been very straight
10 about it. What is interesting is a lot of times
11 when it is introduced
for the first time, people
12 talk about OAB, it's a
parvovirus, it has been in
13 humans for a really
long time, and it has been
14 extremely safe in the
sense that it is not
15 associated with any
known disease entity, and the
16 implication is many
times that therefore, AAV gene
17 therapy as a vector is
going to be similarly safe.
18 However, I
think what you very clearly
19 point out in all the
molecular biology that has
20 been done with the
vector is that an AAV vector
21 really isn't anything
like a wild-type AAV in the
22 sense that now what
you have got mainly is
23 episomes, it is not
integrating in chromosome 19,
24 so there is a lot of
assurance that one might take
25 from the first part of
the data that it is probably
47
1 not reasonable to
carry forward into thinking about
2 AAV vectors.
3 DR.
SAMULSKI: Right. I will give
4 opinions on both
sides. I think if you look at the
5 biology of the virus,
it falls in the biological
6 features, so that we
don't see significant immune
7 response generated
from AAV infections. You don't
8 see that with wild
type.
9 You don't
see the virus taking over the
10 host cell as a lytic
virus does, so there is
11 consistency in that
aspect of saying AAV is more
12 like its features of
being non-pathogenic, but I
13 think you only need to
hear what Phil and them
14 mentioned at the RAC
probably every time AAV is
15 discussed, you know,
this is not normal. You are
16 putting in 1012
viruses into a focal injection,
17 hundreds of particles,
lots of genomes. This is
18 something that doesn't
happen in nature, and so it
19 shouldn't be
considered as the viral life cycle,
20 because in that
setting, we can't reproduce the
21 viral life cycle. We are not getting a systemic
22 infection that is
disseminating and maybe setting
23 up latency.
24 We are
inducing an artificial way of
25 getting
persistence. So I think you are right
on
48
1 the money there. I
think what will go back and
2 forth between these
systems is how much does the
3 vector mimic wild
type. As far as integration they
4 are identical, it is
just one is on 19, the other
5 one is random.
6 So there is
some ability to go back and
7 forth as to what is
happening.
8 DR.
SALOMON: So the second question I had
9 was I don't know a lot
about chromosome 19, so I
10 apologize for what I
am certain are stupid
11 questions to the
geneticists here, but is it clever
12 that the virus chose
this area in chromosome 19, is
13 that a safe area to
integrate in that?
14 I guess the
follow-up question here would
15 be maybe one thing to
think about, has anyone
16 thought about it, is
if you add the Rep gene back
17 and let it integrate
into a place that we know is
18 safe instead of having
all this episomal DNA that
19 we have no idea what
it is doing.
20 DR. SAMULSKI: Your question is something
21 that you would discuss
at a cocktail hour, why does
22 AAV go to 19. We could say mechanistically, there
23 is a viral origin
sitting on 19. Did the virus
24 pick it up from 19 and
retrofit it into its life
25 cycle or is that a
remnant, some integration event
49
1 that took place who
knows when.
2 It is only
conserved in monkeys and
3 humans, so it is a
sequence that is not found, so
4 there may be some
selective pressure for why that
5 took place. Is it a safe site? In tissue culture,
6 we are in HeLa cells,
there are 19 chromosomes, 3
7 copies in 19, we can
get latency all the time. In
8 vivo, there hasn't
been the kind of studies you
9 would want to see, and
if AAV integrates in 19, is
10 that going to be an
adverse event.
11 I would
argue 19 in liver cells may not be
12 essential, but 19 in
another tissue like neuronal
13 cells may be
essential, but to get back to your
14 question, which I
think is more directed to what is
15 on that locus, there
is no gene located at that
16 region.
17 Michael
Linden has argued that there is a
18 transcript that can go
through this region that is
19 related to a muscle
transcript, but from our and
20 other studies, there
has never been an integration
21 event that has
disrupted that gene or the potential
22 for the gene, but
again, there are all tissue
23 culture cells, so I
think it is an interesting
24 biology.
25 When we
first saw this, what is clustered
50
1 on chromosome 19 were
a lot of genes we would have
2 liked to have seen it
go into, the receptor for
3 polio virus, a
receptor for a lot of other viruses,
4 and we thought, oh,
maybe, AAV will integrate, give
5 the host cell a
mechanism of protection from
6 another infections
agent, and there would be a
7 reason for why it
targets, but this locus is not by
8 those type of genes,
although it would have been a
9 nice story. So it is an unknown.
10 DR. SALOMON: I had one last question, and
11 that is when it
integrates and then almost sort of
12 kind of does its
version of concatemerization in
13 that area -- that is
not quite exactly what
14 happens, but -- what
does it do to the promotor
15 regions in the ITR, is
the payload gene still
16 promoted, or does it
destroy the promoter region,
17 so you basically have
dead genes there?
18 DR.
SAMULSKI: AAV is not like the
19 retrovirusus where it
has a promoter, a strong
20 promoter in the
LTR. It has promoter-like
21 activity, but all the
cassettes have the promoter
22 built in between the
terminal repeats, and so the
23 gene remains intact,
the break points seem to be in
24 this buffering area in
the terminal repeats.
25 So, again,
all of these things are skewed.
51
1 They are put under
selection so you insert the
2 genes that go in
intact, and they rescue them out.
3 We can only see the
products that E. coli will
4 tolerate, so you have
to realize that head-to-head
5 and tail-to-tail
formations are not very stable in
6 E. coli, so we are
getting a biased opinion every
7 time we pull these
out.
8 The PCR
reaction is extremely biased
9 because that is Mother
Nature's best primer, it's
10 an 80 percent GC
hairpin structure. If you try to
11 prime through that
region, you will generate
12 deletions, so we even
think a lot of our data
13 showing break points
is an artifact of pulling out
14 junctions.
15 The only
data that begins to support that
16 if you have a real
controlled Rep expression, you
17 don't see as much
amplification rearrangement. The
18 group in Italy put the
Rep gene on the regulatable
19 promoter, and they
actually dosed in the amount of
20 Rep, and what they was
the integrations were more
21 well behaved.
22 So I would
say that we have not been able
23 to mimic what probably
the virus does very well,
24 but we can score all
the downstream events. It
25 goes in a chromosome,
it looks like this, and so
52
1 forth.
2 So I would
be hesitant about taking my
3 opinion about this
field and turning it into this
4 is the fact of all it
all happened.
5 For the
vectors where there is no Rep, and
6 you do see the integration,
it is cellular
7 mechanisms that are
putting it into the chromosome.
8 DR.
SALOMON: Dr. Rao and then Dr.
9 Mulligan.
10 DR.
RAO: Is there any evidence of
11 mobilization of the
integrated thing, wild-type
12 infection?
13 DR.
SAMULSKI: That is a good point.
14 There is the risk of
mobilization if you get an
15 added infection and a
wild-type AAV infection, so
16 you need a two-hit
kinetics to move the vector out
17 of the chromosome.
18 In the
laboratory, if you do those
19 experiments, wild-type
dominates the product that
20 comes out, because
there are more elements that
21 ensure packaging, and
they are not in the vectors,
22 but you do mobilize it
if you get a two-hit
23 kinetic.
24 DR.
RAO: Is there a rough percentage on
25 that? I know wild-type predominates, but --
53
1 DR. SAMULSKI: Wild-type plate
2 90-something percent
of all the virus that comes
3 out, and if you cycle
it, it is the only virus that
4 you see. The vector doesn't compete very well in
5 that setting, but the
risk is there, in an in vivo
6 setting.
7 DR.
MULLIGAN: In the in vivo case, the
8 integration question
is complicated by all the free
9 copies, and I think it
is important that people
10 that are not experts
here get a sense of if you had
11 very efficient
integration in the sense that you
12 had one copy for large
number of cells, but then
13 you had hundreds of
unintegrated copies, that would
14 confuse your
interpretation, so can you
15 characterize for
people how you get at the issue,
16 that is, if you just
look at the sum of
17 unintegrated copies,
and that is a large number,
18 and then the sum of
integrated copies, and that is
19 a small number, then,
one conclusion is that you
20 have mainly
unintegrated gene transfer, but in
21 principle, on a
cell-by-cell basis, you could have
22 very efficient
integration, while on top of it you
23 could have a large
amount of unintegrated copies.
24 Now, in
vitro, I know that is not the case
25 because you can
actually directly assess that, but
54
1 how have the various
tests actually ruled out that
2 that is not the case?
3 DR.
SAMULSKI: I think that is a good and
4 hard question. I think Mark has generated data
5 that begins to look at
that where he has put virus
6 in hepatocytes, and he
will probably discuss this,
7 and then did a partial
hepatectomy to let the liver
8 cells grow, and tried
to score how many of those
9 regenerated liver
cells still carry a copy
10 suggesting that that
fraction had integration, and
11 the ones that lost it
were primarily episomal.
12 I will let
him describe that, but I don't
13 think there is any
good way to assess that
14 question.
15 DR.
MULLIGAN: I would think that now that
16 there is these, in
human cells, outlaw PCR
17 approaches, the
question is can you actually
18 directly calculate the
total absolute number of
19 integrations
independent of how much total DNA is
20 there?
21 DR.
SAMULSKI: I don't know how I would do
22 that. I think this is what Phil Johnson is doing
23 in his abstract. He is looking at ALU real-time
24 PCR going across
genomes and stuff like that.
25 DR.
MULLIGAN. Has anyone looked, like
55
1 Ernest Whittaker, like
his system if you have an
2 adeno-infection or HIV
infection, and you all of a
3 sudden do an AAV
infection, is the propensity for
4 integration of AAV
into, say, HIV, a higher
5 integration because
it's unintegrated initially
6 than it would be to go
in the chromosome?
7 DR.
SAMULSKI: I think that is another
8 good question, that
is, if you are in a cell that
9 has substrates, what
is the fate of AAV to those
10 substrates, will it go
into them, or a more
11 preferred event. I don't think anyone has an
12 answer to that, but
it's a good question. It is
13 something that has got
to begin to be looked at.
14 I think I
would like to just emphasize
15 that AAV in the early
days was put in the bone
16 marrow stem cells with
a lot of efficiency, and
17 then it was shown that
as you tried to amplify
18 these cells, they
weren't very good and I think it
19 was speaking directly
to the fact that it wasn't
20 integrating and
therefore, you could transduce them
21 and get positive
cells, but once they are asked to
22 divide, you lost that.
23 So I think
why AAV has been such a niche
24 virus for the
so-called non-dividing cells is
25 because is can set up
this persistence. I think
56
1 the integration
frequency is probably going to be
2 determined by do
non-dividing cells carry out
3 illegitimate
recombination, at what rate compared
4 to a dividing
cell. That is going to be an
5 important number that
is going to influence the
6 outcome in these type
of studies.
7 DR.
GORDON: I have a couple of very quick
8 questions that are
just simple factual answers.
9 Where in the
life cycle of AAV does the
10 uncoating of the
genome take place? That is one.
11 The second question is
you said that when you add
12 Rep back to the
vectors, then, you get chromosome
13 19 integration
again. How is it added back, as a
14 gene or as a protein?
15 DR.
SAMULSKI: The answer to the first
16 question is the
parvovirus are argued to go into
17 the nucleus and uncoat
to release their DNA into
18 the nucleus. There is probably a capsic component
19 still associated with
the virus that is sitting on
20 those terminal repeats
that either prevents it
21 from, you know, being
naked DNA, but at the same
22 time may recruit other
factors to the origin.
23 As far as
the second question that you had
24 -- I forgot it already
--
25 DR.
GORDON: Adding Rep back.
57
1 DR.
SAMULSKI: That's my senior moment
2 there.
3 Rep protein
has been added both as
4 plasmids, as physical
protein injectate, and as
5 inducible protein in
the cell line, and all of
6 those will take
vectors and allow it to go to
7 chromosome 19.
8 The last
thing I will mention is that both
9 the Italian group and
our lab have generated a
10 mouse that carries the
chromosome 19 locus, and in
11 our case, it is
sitting on the X chromosome. When
12 we put wild-type virus
into that, it goes to that
13 chromosome 19 locus
even though it's on the X
14 chromosome, again
suggesting it's the cis elements
15 that are driving where
it goes, and not that it
16 happened to be on 19
in humans, and stuff like
17 that.
18 DR.
DYM: I think you alluded to my
19 question, but i am
going to ask it anyways. Can
20 you clarify or comment
on the ability of the AAV to
21 get into dividing
cells versus non-dividing cells,
22 and, of course, in the
testis, the spermatogonia
23 are very actively
dividing, the sperm are not.
24 DR.
SAMULSKI: I think there is no
25 difference between AAV
going into dividing or
58
1 non-dividing
cells. If the receptor is present, it
2 will bind, and then I
think the mechanism for
3 internalization is
clathrin-coated pits, endosome
4 release, and traffic.
5 If you can
carry out those steps, it is
6 indistinguishable
whether it's a dividing cell or
7 non-dividing
cell. In the very early days, it was
8 suggested that AAV
preferred dividing cells, but
9 that was in vitro
looking at selection and
10 therefore you were
biasing the system.
11 I think once
people went in vivo, they
12 realized that all of
that was probably misleading a
13 little bit.
14 DR.
MULLIGAN: You didn't mention about
15 other AAV serotypes,
so in principle, the
16 efficiency of the
intervention would depend upon
17 just the virus titer.
18 Do you have
any sense that AAV-1, for
19 instance, which in
muscle is much, much more
20 efficient, would
potentially be better at infecting
21 germ cells?
22 DR.
SAMULSKI: I think Richard's point is
23 a really interesting
one because we and others have
24 seen that the other
serotypes have better propisms,
25 are more
efficient. The question is what are
their
59
1 integration
mechanisms.
2 The only one
that we have data on is Type
3 4. Type 4, which is camana monkeys, will target
4 monkey cells and
integrate, will target human cells
5 and integrate in the
chromosome 19, so the
6 wild-type virus will
capitulate exactly what the
7 human virus is.
8 The other
four, 1, 3, and 5, it is
9 unknown, but they are
so homologous, about 80 to 90
10 percent homologous,
they all bind to the terminal
11 repeats, they all can
package each other's DNA.
12 Chances are they will
do the same type of
13 integration.
14 There are
differences in these terminal
15 repeats if you look at
them. Type 5 is different
16 than Type 2, and if
that is a substrate, that may
17 be more prone for
recombination enzymes, you may
18 see an integration
frequency that is different.
19 DR.
MULLIGAN: I just meant the capsid,
20 looking at risk for
germline infection, if it
21 happens just
proportionately, it much better
22 infects that cell and
even though integration is
23 very efficient, then
you get more efficiency.
24 DR.
SAMULSKI: I misunderstood. I think
25 if the virus has a
more efficient tropism in those
60
1 kind of cells, chances
are the integration
2 frequency is going to
be higher. That is kind of a
3 given.
4 DR.
SALOMON: Sort of a follow-up question
5 here is -- and you may
have answered this, and I
6 apologize if you did
-- if you have a cell that is
7 actively dividing or
is activated, let's say, so it
8 has a lot of open chromatin structures, it
is more
9 likely to integrate in
that setting than in, let's
10 say, a stable cell
that is not activated?
11 Obviously,
where I am going is in, you
12 know, if you had an
injury or inflammation, or
13 something, are those
areas in which the rules might
14 be different?
15 DR.
SAMULSKI: Sure. I think that is
16 exactly what the data
are supporting. This virus
17 looks for open
chromatin contacts. Events that
18 were scored appeared
to be in genes, promoter
19 regions in the
gene. I think they are all because
20 of the same reason,
these were open chromatin. If
21 it's condensed
chromatin, there is probably no
22 mechanism, because
again it's a cellular event and
23 it is going to be
acting on cellular regions of the
24 DNA, better
accessible.
25 DR.
SALOMON: That was great. Thank you.
</