|
 Where
does that leave us? What I tried to share with you is the idea that
most anticancer drugs activate apoptosis through the mitochondrial
pathway. Now the inadvertent activation of this pathway would have
dire consequences for cells, and so it is not surprising that there
is a wide variety of regulatory mechanisms including Bcl-2 family
members, kinases involved in survival, and direct caspase inhibitors
that are expressed in a wide variety of cells.
Therapeutically,
the challenge is first to define which of these mechanisms are aberrantly
activated in AML and then to start to devise strategies if they
are aberrantly activated in AML.
DR. SCHULMAN:
I am Phil Schulman from North Shore University Hospital. There are
some data from Europe that show that if you culture leukemic cells
for long term or culture marrow derived from patients with AML for
long term that the leukemic clonigenicity markedly decreases. Are
there ongoing spontaneous apoptotic pathways in those cultured samples
that allow for decrease in clonigenicity and therefore a possible
use in autologous transplants?
DR. KAUFMANN:
I cannot respond directly to the question by citing specific data.
All I can do is refer you to a paper in the British Journal of Hematology
by Rick Jones and collaborators. I believe it appeared last year
or maybe late in 1998, showing that if you take leukemia specimens
out of any AML patient and simply put them in culture for a day,
you start to see apoptosis.
Again, I don't
know the exact culture conditions you are referring to, but lots
of leukemia specimens have a certain incidence of spontaneous apoptosis.
Likewise, data from the M. D. Anderson from Estrov et al shows that
caspase-3 is spontaneously activated to a certain extent in AML
specimens from patients with newly diagnosed AML.
DR. ANDREEFF:
Michael Andreeff, M. D. Anderson, a comment on the targets. BAD
is constitutively phosphorylated in essentially all AMLs and that
is a therapeutic target, I think. As far as the IAPs are concerned,
XIAP and survivin both are expressed in essentially all AMLs. So
these are also targets. The XIPs cleave when you treat with Ara-C,
but that is probably an indirect effect. Caspase cleaves XIP into
two parts, etc., and one of them is slightly pro-apoptotic. One
is slightly anti-apoptotic according to John Reed, but these are
clearly targets that have to be addressed in different ways.
DR. ESTEY: Yes,
it is a question to the first three speakers in general. Let us
say you had a drug that targeted one of these activators or inhibitors,
all the things that people were talking about, do you feel comfortable
enough in your knowledge of these things to say, "Okay, I have
a drug, and now, I don't know what dose to use."? Would you
define the dose you are going to use in clinical trials based on
a biologic end point; you know, it inhibits a certain amount of
whatever you think that you are interested in inhibiting, or would
you still do a standard Phase I trial, feeling, gee, I know something
but not enough to base my whole premise of future trials on this,
if you see what I am saying? It is a general question because I
think it is something that comes up in all clinical trials. Is the
end point, targeting something that we think we know biologically,
or is it still the traditional end points feeling, we may know something,
but we really don't know as much as we might think. I don't know
if I made that clear, and it is a general question, but I think
it is a very important question in clinical trial design.
DR. KAUFMANN:
You made it very clear. I would invite the other two speakers from
the first plenary session to address it. I think you have to say
that we are just at the beginning in terms of apoptotic regulators.
We know nothing about their expression in normal stem cells, for
example, and so I don't know whether there is any therapeutic benefit
to be gained by inhibitors of any these apoptotic pathways. I think
only time will tell, and so I think we are stuck saying that we
don't know where we are, but if we have a good inhibitor of XIAP,
for example, we would have to do the clinical trial.
DR. ESTEY: But
the dosage you would use in the trial, would that be the dose that
says, "Oh, 80 percent inhibition," or would you just do
a Phase I study? Here is the dose that produces toxicity, and now
I will correlate response with my presumed target because they are
two very different approaches.
DR. KAUFMANN:
I agree, and I would throw it back to you. I would say that in putting
on my clinical hat I would be very uncomfortable looking for biological
effects.
DR. ESTEY: I
agree with that.
DR. WILLMAN:
It also presupposes, Eli, we really think we know what the target
is.
DR. ESTEY: Right,
that is my point because I think one of the things that we are seeing
at least from the NCI, and maybe Bruce can disagree because I am
probably wrong, is the idea that the dose should target the end
point rather than a Phase I study. In my opinion that is a big mistake,
because I think we think we know more than we really do, and I think
these talks have eloquently demonstrated that.
DR. LARSON:
Clearly I think this is a key question that each of the breakout
session groups is going to have to struggle with. How are we going
to recognize when we have actually hit the target?
DR. CIVIN: Curt
Civin from Baltimore. Scott, nice talk. I want to ask you another
question, kind of a vision question like the former one. If you
have by micro array or phosphorylation or whatever you are measuring
a whole bunch of polar apoptotic genes up and down here and you
have a lot of anti-apoptotic genes up and down here, what do you
do, add them all up? How do you tie this into the biology of the
cells, particularly in the patient? I mean at what point do we get
apoptosis or do we inhibit it?
DR. KAUFMANN:
Curt, that question is precisely the reason that I don't work on
BCL family members in my laboratory. I think it is very complicated,
and I don't know that anybody has a simple answer to that right
now. I am a poor dumb cellular pharmacologist, and so my readout
is how many cells have I killed or how much caspase have I activated.
I think you have to go with the functional assay which gets back
to Dr. Estey's question. I think you have to do the functional assay
on the patient, i.e., do the dose-response curve and see what happens
because I think it is extremely complicated.
DR. CIVIN: So
should we do the functional assay in vitro, or is there a way to
do it in the patient who is undergoing the treatment?
DR. KAUFMANN:
Again, this is opinion and it is an opinion of one, and I seem to
be a lightning rod for these questions. But my opinion would be
if there is good rationale for taking the agent forward into AML,
you do the clinical trial, but you would do it properly in the sense
that you would do a full-dose escalation to whatever your MTD is
and you ask the biological question in terms of what happened to
Bcl-2 family members or caspases or whatever at every dose level,
but you keep on going up because I don't know that 80 percent inhibition
of XIAP is the target as opposed to 90 percent or 95 percent or
99 percent.
I think we have
really faced that same issue in terms of the farnesyl transferase
inhibitors where No. 1, when the study started we thought we knew
what the target was, and No. 2, we thought we knew how much we needed
to inhibit the target in order to have biological effect. What we
have discovered while the clinical trials are ongoing is that the
basic scientists were wrong. The target probably is not ras, No.
1, and No. 2, we don't know whether we need 99 percent or 99.99
percent inhibition to get biological effects.
DR. CIVIN: So
until the tumor shrinks?
DR. KAUFMANN:
Exactly.
DR. GORE: Steve
Gore from Johns Hopkins. Scott, just to borrow your own phrase,
I would like to add a cautionary note to our attempts to find the
role of apoptosis in clinical response in leukemia. That is, we
have to be absolutely consistent on our definition of apoptosis
because there are a number of operational assays for apoptosis that
may vary with the stimulus.
Secondly, it
is known that Bcl-2 family members can delay but not ultimately
prevent apoptosis and the timing of monitoring apoptosis is critical.
Thirdly, a lot of studies have demonstrated you can induce mitochondrial
damage and block downstream apoptotic targets. The cell may not
exhibit the classic manifestations of apoptosis, yet be dead from
a reproductive standpoint. So I think all of these things will potentially
confound our attempts to define specific role of apoptosis and clinical
responsiveness.
DR. KAUFMANN:
I think all three of those points are excellent. In fact, on the
last point, the fact is that once cells have released cytochrome
C to the cytoplasm, they might have irreversible damage. Whether
they can activate caspases or not is an important one because it
might explain why even leukemia specimens that have very little
Apaf-1 or very little caspase-9, that those patients still could
potentially get CRs with current antileukemia therapy. So I think
those comments are right on target.
DR. GABRILOVE:
Janice Gabrilove, Mount Sinai Medical Center. This probably reflects
my lack of knowledge, but are there examples of or anything known
about polymorphisms of caspase-9 or AKT since they seem to be critical
points giving rise to more or less functional proteins enzymatically?
If that does occur in the general population, has there been any
effort to look at the presence of those mutations or polymorphisms
in combination with altered transcription factors in terms of partially
knocking out an apoptotic pathway or making it relatively dysfunctional
rather than an enhanced proliferative one?
DR. KAUFMANN:
A great question to which I don't know the answer, and I am going
to ask my colleagues in the audience who also follow this literature
to shake your heads. Basically I don't know of any polymorphisms
that have been defined. I see Kap shaking his head. Steve, do you
know of any polymorphisms?
DR. GORE: BAX
mutation.
DR. KAUFMANN:
Somatic BAX mutation, but in terms of Apaf-1 and caspase-9 I am
unaware of any, but it could be that the WNL meaning iwithin normal
limitsi might also mean iwe never looked,i okay? So there is an
absence of data.
DR. SCHIFFER:
Ever since early in medical school I have found that I was intimidated
by slides that had lots of arrows, and then arrows with double heads
on both sides, then arrows that are in color which are really something,
and you and Jerry showed lots of arrows, and you didn't even touch
the surface of the number of inhibitory pathways that go parallel
and the number of pathways that bypass ras, etc. We all know that
cancer cells want to live, and they are really successful at that
and have redundancy in so many of these pathways that allow them
to escape most everything that we toss at them. And it is worse
in epithelial cells. In looking at all of these pathways and all
of these potential targets and whether there are really actually
well-defined choke points, how do we in our lifetime sort of get
past all the confusion to really target something that may represent
something that is really critical to a cell? With BCR-ABL we have
something that perhaps gets us 75 or 80 percent of the way there,
but in these much more complex situations how can we do it because
I joked that I am intimidated by those slides but obviously in trying
to select agents and strategies it makes it a very formidable task.
DR. KAUFMANN:
Charlie, you have stated the same problem that I think we face in
terms of drug transporters as well. We thought that there was one,
and now we see that there are tens of them, scores of them. I wish
I had an answer to that. I am going to be anxious to see what others
have to say in the breakout session, and again, I will throw it
back to the audience. Dr. Andreeff, how do you pick which of these
regulators to focus on because sitting in my vantage point I find
it very difficult to say that you ought to focus here as opposed
to here. We don't know.
DR. SCHIFFER:
Survivin sounds like a good one!
DR. LOWENBERG:
Bob Lowenberg, Rotterdam. What we have seen so far this morning
is an extremely complex situation of genetic mechanisms leading
to leukemia, and wouldn't it be an important addition to try to
build models of the key steps that determine the leukemic phenotype?
If we would be able to refer to that particular step selectively,
that could give a clue to where to go for treatment intervention.
I think this also relates to the question that Eli raised.
If we are going
to enter clinical trials, many of these drugs may affect secondary
events that will not be very meaningful from a clinical point. So
maybe we should also invest in models and try to select that target,
in vivo models.
DR. LARSON:
We will take just two more questions and then I think we will have
to give Dr. Kaufmann a rest so he doesn't have to defend the whole
field.
DR. COTTER:
Finbarr Cotter, London. You didn't mention the benzidine receptor.
There has been some quite interesting work recently looking at the
mitochondria. It very much controls the transition pool which appears
to be one of the very critical gatekeepers to whether apoptosis
does or doesn't take place, and it appears to work possibly downstream
from the Bcl-2 family of genes. So it looks quite interesting therapeutically.
I think Dieter Kramer had a nice paper showing that you can inhibit
it with small molecules. I wonder if you would like to comment on
that because I think it is quite an important area for therapeutic
targets.
DR. KAUFMANN:
Two comments. First of all, our goal as clinicians is not to inhibit
apoptosis but in fact to facilitate it. So I am not sure that Dr.
Kramer's papers, while very interesting, lead us anywhere therapeutically,
and the second concern I have is that if one goes to a mitochondrial
meeting as opposed to an apoptosis meeting, all of that work on
the role of the mitochondrial permeability is said to be subject
to fatal flaws, paper by paper. So I think this whole area of how
cytochrome C gets out of mitochondria is incredibly controversial
in the field. Guido Kramer has one opinion. The people who have
spent their careers working on mitochondria cannot reproduce or
perform key experiments that support this model. Let us just say
that everybody agrees that cytochrome C leaks out, but one of the
black boxes is how does it leak out, and how could you selectively
trigger that if you wanted to?
DR. COTTER:
There have been a few papers actually looking at inhibition of the
antagonist to the benzdiazapine receptor which has repeated some
of his work as actually inducing apoptosis by sensitizing drugs
or the use of drugs in that area.
DR. GRANT: Scott,
you know as one looks at a pathway from the beginning to the end
and in particular in apoptosis, I am wondering if you could comment
upon the fact that, at least in laboratory models, adding ZVAD or
DEBD which would inhibit one of the very last steps of the pathway
can in fact very efficiently block apoptosis generated by a whole
variety of signals. Do you think, therefore, that if we try to get
to the bottom of the pathway and some of the final steps, that it
may be more efficient than starting at the top of the pathway where
you may, as Charlie Schiffer said, get into the different colored
arrows?
DR. KAUFMANN:
To summarize the question for the audience that isn't familiar with
the data there is a series of 4-0-methyl ketone derivatives that
have cleavage sites that correspond to the recognition sequences
of some of the caspases. DEVD is the cleavage site for caspases
3 and 7 and then ZVAD is a relatively non-selective inhibitor of
these caspases. The question is if you put in these inhibitors,
you abrogate the apoptotic phenotype after a variety of stimuli.
So why not turn that around and say that what we need to do is activate
these same caspases. Is that in essence the question? Okay, now,
two problems. First of all the inhibitors you are referring to,
including the DEVD FMK, are all promiscuous inhibitors that show
relatively little selectivity for one caspase versus another.
In fact, that
data is published in the JBC in August this past year. So DEVD FMK
not only inhibits caspases 3 and 7, it also very effectively inhibits
caspases 8 and 9, and that data is in that paper.
I think the
inhibitor data are misleading us because we have been taught that
those are selective inhibitors, and they are not, but the second
issue really is how many caspases do you need to turn on in order
to get the apoptotic phenotype, and I think the jury is out on that.
It is possible that if you simply turn on caspase-8 or caspase-9
in some cells, perhaps those cells that have less XIAT or whatever
will die.
Other cells
can tolerate low-level caspase activation. There was a very nice
paper by Craig Thompson's group about 5 years ago now in PNAS showing
that caspase-8 could be activated in certain cells under certain
conditions and yet the cells would tolerate that. What we don't
know is how individual leukemias are hard wired, and it might vary
from leukemia to leukemia as to whether you need all of the downstream
pathway activated or not.
TOP
|
DR.
WILLMAN: Our next speaker is Dr. Scott Kaufmann from the Mayo Clinic
who is going to give us an update and quick overview of apoptotic
pathways and ways that this pathway is both perturbed in leukemogenesis
and again might be exploited for therapy. Again, I would like to
thank Scott for agreeing to do this on very short notice.
I
would like to talk very briefly about the apoptotic machinery and
then talk about the multiple levels of regulation of that machinery
and then end up with some speculations and some unanswered questions,
particularly as they relate to AML.
Now,
to get us all on the same page I simply want to remind the audience
that many of the aspects of the apoptotic phenotype can now be explained
by caspase-mediated proteolytic cleavages. Basically those of you
who haven't been following this field know apoptosis as a process
that has morphological changes and gives DNA ladders.
Now
all of these cleavages are mediated by a protease that is called
caspase and caspases are intracellular proteases. They are cysteine
dependent. They cleave on the C-terminal side of aspartate which
is a relatively unique cleavage site for mammalian proteases. They
are synthesized as proenzymes that have low activity, and they are
activated at least in part by cleavage themselves at aspartate residues,
and the fact that these cleave at aspartate and are activated by
cleavage at aspartate gives you the possibility of protease cascades.
Over
the last 3 or 4 years two major pathways of caspase activation have
been worked out. I will be going through each of these in turn,
but on the left hand side you have a protease cascade where the
initiator protease is caspase-9 and it then can cleave procaspase-3
and procaspase-7 to give you the active caspases that give you all
the cleavages I showed you a couple of slides ago.
Now
let us first go to the death receptor pathway. This is a pathway
that is activated by ligation of certain cell surface protein receptors.
On this slide I have outlined what happens when the fas receptor
interacts with the ligand, and basically from right to left what
you have is a multimeric fas ligand that interacts with a monomeric
receptor and causes multimerization of the receptor. This receptor,
once it trimerizes, then acquires affinity for an intracellular
adaptor protein called FADD. FADD binds to the receptor, undergoes
a conformational change and recruits procaspase-8. Upon recruitment,
procaspase-8 is then proteolytically activated and activates the
downstream part of the pathway.
In
fact, over the last few years there have been six different death
receptors identified, the tumor necrosis factor receptor 1, fas,
death receptor 3. Two are receptors for the activating ligand TRAIL
and there is also a DR6 that was identified last year.
Let
us turn for just a minute to the other pathway, the mitochondrial
pathway. This is a pathway that is activated by pro-apoptotic BCL2
family members. So polypeptides such as BAX or BAK, in some way
that is currently not well understood, cause release of cytochrome
C from the mitochondria, from the intermembrane space of the mitochondria
to the cytosol.
One
of the questions that comes up is which of these pathways is activated
by anticancer drugs because if you are thinking about these pathways
as potential contributors to drug resistance, the mechanisms you
would think about over here, if drugs act through the death receptor
pathway, are actually quite different than if the drugs work through
the mitochondrial cytochrome C pathway.
This
question of which pathway is activated by anticancer drugs has received
quite a bit of attention over the last 3 years. To summarize about
50 papers, probably the cleanest data come from mouse knockout experiments
where you can knock out the gene for FADD or the gene for procaspase
8 and you see diminished apoptosis mediated by the death receptors
but you see a normal response to doxorubicin and etoposide. So at
least for those two drugs, it appears as if the death receptor signaling
pathway is not required for them to trigger apoptosis.
So
much for the overview of the apoptotic machinery. The thing that
I want you to take home actually is that there are multiple levels
at which this machinery is activated, and that the activation is
a fairly complex process.
We
will focus primarily on the mitochondrial pathway because that appears
to be the major pathway in drug-induced apoptosis. I will talk about
regulation of caspase expression, regulation of the release of cytochrome
C from mitochondria, and regulation of caspase activation and activity
because these are at least three points at which the pathway can
be modified.
Originally,
those of us in the field working on caspase had thought of these
polypeptides as constitutively expressed with the genes always turned
on, but George Stern's group published a very nice paper in Science
a few years ago showing that, if they took fibroblasts from STAT1
knockout mice, in fact those fibroblasts lack caspase-3, caspase-2
and were resistant to apoptosis induced by several different inducing
agents, and if they put STAT1 back into those fibroblasts, caspase
expression came back on and the cells became sensitive to the induction
of apoptosis.
The
release of cytochrome C is also a highly regulated process. Work
from several laboratories including that of Kap Bhalla who is here
in the audience showed that Bcl-2 and BCL-X act in part by inhibiting
release of cytochrome C from mitochondria and that contributes to
their anti-apoptotic effect.
but
we now know that the Bcl-2 family consists of at least 17 related
polypeptides several of which are anti-apoptotic including Bcl-2,
BCL-X and Mcl-1, a protein I will return to in a few minutes, and
several more are pro-apoptotic including not only BAX and BAK but
proteins like BID and BAD. 
It
turns out that the kinase which does this, Akt, or protein kinase
B, not only phosphorylates BAD but also in some people's hands,
phosphorylates procaspase-9, inactivating it and keeping it from
being activated. AKT also appears to act in other ways to inhibit
cytochrome C release from mitochondria.
Now
let me turn for a second to another group of proteins. These are
proteins that are less familiar to this audience, and these are
the IAP proteins, inhibitor of apoptosis proteins. They are a conserved
family of apoptotic regulators that are found all the way from viruses
to mammalian cells. What is conserved is the so-called "BIR
domain, or baculovirus inhibitor of apoptosis repeat domain. This
domain in these various proteins encodes a polypeptide sequence
that is able to inhibit active caspases.
so
can these CIAP proteins.
Okay,
the organizers of this symposium asked that we talk a bit about
what the unanswered questions are, and I think in terms of apoptotic
regulation there are lots of unanswered questions in AML.
There
are two technical questions that I think this audience needs to
keep in mind when it gets to the breakout sessions. One is the
issue of shipped samples versus immediate processing because I
think that the question needs to be addressed. Are you selecting
for a certain subpopulation of cells in terms of apoptotic capabilities
if you ship cells overnight and work them up the next day somewhere
else, as opposed to working them up immediately, and so until
that is answered in a fairly definitive manner, it is very difficult
to envision addressing some of these questions by shipping samples
to a central laboratory.
With
that as a disclaimer let us walk through some of these potential
explanations for why apoptosis might be inhibited in resistant leukemia.
what
we found was that in fact Apaf-1 and procaspase-9 were expressed
in the vast majority of those leukemia specimens. Levels were higher
or lower but were detectable in virtually every specimen, and to
our disappointment, there was no correlation between response to
therapy, pluses being CRs and minuses being no response, and expression
of those apoptotic regulators.
How
about elevated levels of Bcl-2 family members, anti-apoptotic Bcl-2
family members? Where do you start? There are 17 family members.
How do you look at them combinatorily? There has been a lot of emphasis
on Bcl-2 and more recently BCL-XL, and suffice it to say that the
three or four papers that are out there don't give a clean answer
as to whether high levels of expression of those two regulators
correlate with clinical outcome. 
Again, in collaboration with my former colleagues at Johns Hopkins
as well as John Reed's lab, we looked at Bcl-2, BCL-XL, BAX and
Mcl-1, and what we found was that if you compare samples harvested
at diagnosis and at relapse, we could not see any systematic changes
in either Bcl-2, BAX or BCL-XL, but in 50 percent of the cases,
we saw at least a twofold increase in Mcl-1.
With
regard to elevated levels of IAP, as far as I can tell there are
no papers out there looking at IAPs in AML at the present time.
I
simply want to remind you what Dr. Radich already showed you and
that is that activating ras mutations are fairly common in hematological
malignancies and so downstream of these one would have activation
of Akt.