Molecular Paradigims/Mechanisms in Acute Myeloid Leukemia by Gary Gilliland, MD

SLIDES & TRANSCRIPTS
Tuesday, February 1, 2000

Molecular Paradigims/Mechanisms in Acute Myeloid Leukemia
Gary Gilliland, MD, PhD

Slide 1:

DR. WILLMAN: Good morning everyone. As Dr. Larson has mentioned, what we hope to do in these two plenary sessions is to really set the stage and tenor for the breakout sessions for the rest of the day. Recognizing that you are an audience of experts, these are not meant to be review talks but presentations that highlight certain processes, biological pathways or mechanisms that we think are scientifically critical in the study of acute myeloid leukemia, and that we also think might be exploited for new therapeutic interventions. We have asked three speakers this morning to give us overviews of critical science pathways. After a break, we will go into a second plenary session that really highlights new means of therapy and delivery of therapy.

So without further ado, I would like to introduce our first speaker, whom I know is well known to all of you. Dr. Gary Gilliland from Brigham and Women's Hospital is going to talk to us about molecular paradigms and mechanisms in acute myeloid leukemia, and I would like to thank Gary for accepting this task on such short notice.

DR. GILLILAND: I would like to thank Dr. Willman and Dr. Larson for this opportunity to provide an overview of molecular mechanisms of leukemia in 15 minutes. Obviously, I will have to cover some of the topics in less detail as a consequence of that.

Most acute leukemias in humans are the consequence of acquired somatic mutations in hematopoietic progenitor cells. These typically take the form of balanced reciprocal translocations. If I could have the projector on and the first slide, please?

They are typified by the lack of loss of genetic material. Rather over the last 10 years, it has been determined through molecular cloning

Slide 2:

that the consequence of these chromosome translocations in most cases is the generation of a fusion transcript that can be causally implicated in disease pathogenesis.

Slide 3:

Over the past decade more than 50 different chromosome translocations have been cloned and characterized, and we can begin to break these translocations down into specific subsets and in acute myeloid leukemias have determined that in most cases these translocations target transcription factors that are important in hematopoietic development.

Slide 4:

These include core binding factor, the retinoic acid receptor alpha, members of the HOX gene family, transcriptional modulatory proteins which include MLL and finally transcriptional co-activating proteins such as CREB binding protein or CBP and p300.

For the sake of establishing a paradigm, I will spend most of my time today discussing what we know of the role of core binding factor in acute myeloid leukemias and offer some possible approaches for utilizing these insights into developing new therapeutic approaches for leukemias.

Slide 5:

Before I do that, I will remind you briefly with these simplified diagrams about mechanisms of transcriptional activation and repression that occur in hematopoietic cells, as in all mammalian cells.

Transcription factors in hematopoietic cells bind to their cognate DNA binding sequences. The binding of transcription factors to these sequences is facilitated in part by proteins such as MLL which can bind to the minor groove of DNA.

On binding to promoters, transcription factors are, also, frequently capable of recruiting co-activating proteins such as CBP. CBP is associated with histone acetylase activity which serves in effect to open chromatin structure and can, also, contact the basal machinery of transcription which is comprised of more than 40 different proteins but is represented simply here as RNA polymerase 2.

The consequence of transcriptional activation is the expression of hematopoietic target genes that are important for the normal development of hematopoietic cells.

In addition to turning on genes at appropriate times during the ontogeny of hematopoietic cells, it is also important to be able to turn them back off again,

Slide 6:

and there are well-characterized mechanisms of transcriptional repression for this purpose.

In this case transcription factors recruit nuclear co-repressor proteins such as N-CoR or SMRT and through adaptor molecules including SIN-3A bring histone de-acetylase to the promoter which has the outcome of closing chromatin structure and abrogating transcription.

So this is a delicately balanced regulation between activation and repression, delicately orchestrated during hematopoietic development, and we will come back to these points in the context of mechanisms of action of core binding factor.

Slide 7:

There is now a spectrum of at least 10 different translocations that involve core binding factor. I have listed three here because these are epidemiologically significant in acute leukemias.

The translocation 8;21 and inversion 16 account for about 25 percent or so of acute myeloid leukemias in adults, and the 12;21 translocation is the most common gene rearrangement in childhood cancer and accounts for about 25 percent of pediatric acute lymphoblastic leukemias.

.

Slide 8:

Core binding factor is actually a heterodimeric transcription factor that is comprised of CBF alpha and beta subunits. The CBF alpha subunit, which is also known as AML1, contacts DNA and is a weak transactivator, but its ability to transactivate expression is greatly potentiated by CBF beta. CBF alpha/CBF beta is responsible for the coordinated expression of a variety of genes that are important in hematopoietic development. More than 50 such genes that have been identified. I have only listed a few here to emphasize the point that these are proteins that are important both in myeloid and in lymphoid development, including IL-3, GM-CSF, the M-CSF receptor, TCR beta and the immunoglobulin enhancer promoter.

Slide 9:

It has also become apparent that as with other transcription factors, CBP is recruited by the core binding factor to activate transcription. We will come back to that point in a moment.

So with this brief background on core binding factor's normal function in hematopoietic cells, how do core binding factor related translocations affect hematopoietic development?

Slide 10:

This is the 8;21 translocation that results in expression of the AML and ETO fusion transcript, and I would like to point out on this slide that when we think about how the aberrant fusion transcripts might contribute to pathogenesis of leukemia, we need to account not only for the function of that transcript but also for how it impacts the function of the wild type protein which is expressed perfectly normally and in a regulated fashion from the non-derivatized Chromosome 21.

Slide 11:

There is a broad spectrum of evidence, much of which has been developed by people in this room, Jim Downing, Nancy Speck, Paul Liu and others, that the AML1/ETO fusion protein is a dominant negative inhibitor of transcription mediated by the wild type AML1 protein. This is to say that not only is it not effective as a transcriptional activating protein, but it is able to impair the ability of the residual allele to activate transcription. There is a broad spectrum of evidence that I won't elaborate both at the level of transcriptional activation and in elegant experiments of the role of AML1 in mammalian development that demonstrates this dominant negative activity.

Slide 12:

Scott Hiebert and others have demonstrated during this past year that the way that AML1/ETO does this is to in effect recruit the native or the normal transcriptional repression mechanisms, such as the nuclear co-repressor complex and histone de-acetylase to CBF promoters, and it does so in part at least through the ability of the ETO protein to recruit N-CoR to this complex. This results in lack of transcription of hematopoietic target genes, and I think you can envision that if it is capable of doing this, it could at least explain the blocking differentiation that we see in acute leukemias of the M2 subtype that harbor this translocation.

Slide 13:

It turns out that this is a generally applicable mechanism to all of the CBF rearrangements in leukemias that have been thus far characterized. They include the 8;21 translocation that I described for you in a bit of detail, the inversion 16 that involves the heterodimeric partner, CBF beta, and the TEL/AML1 fusion that is a consequence of the 12;21 translocation.

Each of these fusion proteins can be shown to act as a dominant negative inhibitor of the native CBF and this inhibition in each case is due in part at least to the recruitment of the nuclear co-repressor complex.

How can we take advantage of these insights towards developing novel therapeutics approaches? Clearly that is not going to be an easy problem to address, but I think we can all take hope that this may be tenable in the next several years based on the experience that everyone in this room is well aware of

TOP

Slide 14:

with the 15;17 translocation associated with acute promyelocytic leukemia. This translocation results in the expression of the PML-RAR fusion transcript from the derivative Chromosome 15 and

Slide 15:

we know that all-trans retinoic acid can induce complete remission in these patients.

The molecular mechanism appears to be that the PML/RAR alpha fusion protein, like the CBF fusion protein, recruits the nuclear co-repressor complex to RAR promoters and blocks differentiation at the promyelocyte stage through its ability to attract these co-repressor complexes.

The addition of all-trans retinoic acid, which binds to it as a ligand, binds to its native receptor RAR alpha to release the co-repressor complex and allows for normal maturation and differentiation of these cells, and that is what happens with ATRA therapy in promyelocytic leukemia. This ATRA doesn't kill the cells. It appears to allow them to grow and mature normally, to differentiate and to apoptose.

So it is plausible, at least, that similar strategies or small molecules could be screened that would have similar effects on differentiation in the context of CBF-related translocations.

It is also clear that all-trans retinoic acid doesn't cure these diseases, that these individuals require chemotherapy to induce long-standing complete remissions, and I will come back to that point in a moment.

Slide 16:

So I will summarize what we know about chromosome translocations in acute myeloid leukemia by again noting that they target the transcriptional machinery in hematopoietic cells.

These include hematopoietic transcription factors such as CBF, RAR alpha and HOX gene, and MLL gene fusions. They can involve fusions directly involving the co-activating complex such as the MLL-CBP and MOZ-CPB fusions, and there are more than 20 different MLL gene fusions that have been cloned in association with leukemia.

Not only are these potential targets for therapeutic intervention, but the identification and characterization of shared target genes through strategies such as expression profiling and microarrays may also identify targets of these transcription factor fusions that could be considered for therapeutic intervention.

As requested by the moderators of this conference, I would like to spend the next several minutes talking about future directions and other problems that we might be able to explore that may help us gain additional insights into the biology of these diseases and potential therapeutic applications.

Slide 17:

To open this topic, I will point out the small dark and dirty secret that we are all aware of which is that chromosome translocations that we can see are not the only thing that is going on in leukemia.

More than one mutation is required to cause acute leukemia. That is not to say that the 8;21 is not important or that it is not necessary. This simply says that it is not sufficient to cause these diseases.

Slide 18:

This has several important implications. Multistep pathogenesis of acute leukemia means that there may be second mutations that provide novel targets for therapy, and as importantly, our therapies may be ineffective unless all mutations that are present are targeted. Perhaps this is the reason why all-trans retinoic acid, although it seems to very effectively address the biological function of the PML-RAR fusion protein, in itself is not curative of acute promyelocytic leukemias.

There are several lines of evidence that support the fact that leukemias are multi-step diseases just as are all other human cancers, and I will show you a couple of examples of that.

Slide 19:

One of the pieces of evidence that supports this assertion is that expression of the CBF translocations that I just described for you is not sufficient to cause leukemia. At least at this juncture that appears to be the case.

Jim Downing has some very elegant conditional expression models characterizing the AML1 and ETO expression, but it appears in animal models as well as in other contexts that its expression alone does not cause leukemia or transform cells.

Paul Liu has a very nice CBF beta/MYH 11 knock-in model with expression in chimeric mice. Expression in itself is not adequate to cause leukemia, but Paul can generate it when he potentiates the expression of this gene by adding mutagenic agents, and we and others have had extensive experience with the TEL/AML1 fusion and have shown that it alone does not cause leukemia.

Slide 20:

For example, TEL/AML1 does not confer factor independent growth to the hematopoietic cell lines. It does not transform primary murine bone marrow cells. It does not cause leukemia in bone marrow transplant models, and it does not cause leukemia in transgenic mice of which the TEL/AML1 fusion protein is expressed in the lymphoid lineage cells under the direction of the immunoglobulin enhancer promoter.

So if these aren't sufficient, how can we try to address what the other mutations are that contribute to the pathogenesis of the disease.

Slide 21:

For TEL/AML1 we have some clues because we know genotypically that children that have TEL/AML1 rearrangements almost invariably will have deletion of the other allele. That deletion almost invariably includes the residual TEL allele, as well as the P27 tumor suppressor, a cyclin dependent kinase inhibitor. We know from work at St. Jude's that about 12 percent of TEL/AML1 leukemias are also P16 deficient, and we have identified several cases recently in which TEL/AML1 positivity is associated with myc gene rearrangements. So these provide some clues for us to try to develop animal models to determine which of these are important targets for leukemogenesis, and we and others are in the process of developing such models.

Slide 22:

Another line of evidence that supports the assertion that these are multi-step diseases is rare inherited leukemia syndromes. These are just like colon cancer syndromes or breast cancer syndromes in which you have a mutation that you harbor in your germ line and acquisition of the second mutation during life gives rise to cancer.

One such disorder is the FPD/AML familial platelet disorder with propensity to develop leukemia. This is an autosomal dominant congenital thrombocytopenia characterized by platelet aggregation abnormalities, and affected individuals in these pedigrees develop acute myeloid leukemias.

Slide 23:

I will show you an example of one of these pedigrees from this perspective in which 9 of 22 individuals that are affected have developed leukemia. These are varying flavors of leukemias, AML, CML, myelodysplastic syndromes, and the second mutations are sort of all over the map cytogenetically and include 5q-, 11q- and monosomy 7.

Slide 24:

The molecular mechanism interestingly comes back to our old friend core-binding factor.

It turns out that this inherited disorder as determined by positional cloning strategies is caused by loss of one of the copies of the AML1 gene in the germ line in this family.

The loss of a single copy predisposes to the development of leukemia as well as causing the autosomal dominant platelet disorder, which suggests that AML1 may have tumor suppressor function and perhaps the dominant negative activity of the AML1 fusion proteins also confers a susceptibility to acquisition of second mutations.

We have also demonstrated in collaboration with a number of folks here at this meeting that there are lots of function mutations in AML1 that occur in sporadic cases of pediatric ALL, myelodysplastic syndrome, and acute myelogenous leukemias.

Slide 25:

By way of illustration, I will show you two mutations that we have identified in the AML1 gene both in inherited leukemias as well as in sporadic cases that target these two arginine residues of 166 and 201. This is the NMR solution phase structure of the AML1 DNA binding domain, and this loop up here binds to DNA through strong nuclear Overhauser(?) interactions between these arginines.

When these are substituted with glutamine, you lose DNA binding activity, and this is the defect that is present both in sporadic cases of leukemia as well as in inherited leukemias.

We and others are in the process of knocking these mutations into the germ line of mice so that we can try to develop animal models of leukemia for studying the function of this mutation but also for studying progression of leukemia in this context.

Slide 26:

Finally, I will leave you with some fascinating insights about multi-step pathogenesis that are related to acquisition of cytogenetic abnormalities associated with progression of diseases like CML or myelodysplasia to acute myeloid leukemias.

Slide 27:

I will focus on chronic myeloid leukemias, and I have loosely grouped several syndromes together under this rubric including CML, CMML and atypical CML. We would all agree that these are distinct entities which have very similar phenotypic manifestations including asymptomatic leukocytosis at presentation in many instances -- presumably asymptomatic because there is normal maturation and function of leukocytes, and these are all characterized by progression to acute leukemia.

Slide 28:

These are all of the translocations that we are aware of that are associated with chronic myelogenous leukemia, the 9;22 translocation giving rise to BCR/ABL that was cloned nearly 15 years ago and these other five translocations that we and our collaborators have cloned and characterized over the past several years.

Slide 29:

It is somewhat gratifying that since the phenotype is so similar in these diseases that the genotype or at least the structure-function relationships of the fusion transcripts is also quite similar. In each case there is a tyrosine kinase on the 3 prime end of these molecules and there is a multimerization motif on the 5 prime end that serves to constitutively activate the tyrosine kinase.

I will also point out in the context of Dr. Parkinson's discussion later today in which I anticipate he will talk about the new therapies for CML based on specific kinase inhibitors that that same inhibitor is equally effective in inhibiting the PDGF beta receptor and most of the chronic myeloid leukemias that we see can be accounted for by mutations affecting either ABL or PDGF beta receptor.

Now, if you take these fusions, and I will show you one example, the TEL/PDGF fusion, and introduce these into hematopoietic cells in mice using retroviral gene transfer

Slide 30:

as shown here in which we harvest bone marrow after 5-FU treatment to bring progenitors into cycle and infect these cells with murine ecotropic retroviruses that express genes like the TEL/PDGF receptor fusion gene, we can then reinfuse these cells that have the stably integrated retrovirus expressing TEL/PDGF beta receptor into lethally irradiated syngeneic recipients.

Slide 31:

You can take any one of those fusion proteins that I showed you on this previous slide and do this experiment,

Slide 32:

and this is the phenotype that you will get.

You get a myeloproliferative syndrome that is quite similar in many ways to the disease in humans. It includes extramedullary hematopoiesis, very high white blood cell count, and you can appreciate that there is normal maturation and differentiation of these neutrophil lineage cells in these mice. You can tell these are mice because of these curious ring-shaped neutrophilic forms that they have.

So how can we take advantage of these observations in thinking about disease progression from CML to acute myeloid leukemias?

Slide 33:

Well, in CML as I have noted there is a proliferative capacity of these cells, but there appears to be essentially normal differentiation of cells. With a transition to acute myeloid leukemias there is also remarkable proliferative capacity of cells, but there appears to be a block in differentiation.

Slide 34:

The other intriguing aspect of this is that we can now correlate these phenotypes with genotypes. For example, the TEL/PDGF beta receptor fusion in one case in which this was cloned progressed to an acute myeloid leukemia in a patient that acquired a friend, the AML and ETO fusion transcript with an 8;21 translocation. It is also clear that BCR/ABL can undergo similar transitions to acute myeloid leukemias or blast crisis.

The most common second cytogenetic abnormality in these cases is the 3;21 translocation also involving the AML1 gene. So an intriguing hypothesis is that the tyrosine kinases provide a proliferative surge but don't really affect differentiation and that the fusion proteins such as AML1/ETO which we know are not sufficient to cause leukemia but contribute to pathogenesis inhibit differentiation of these cells giving rise to the full-blown phenotype.

Slide 35:

We can test this experimentally, and we have initiated experiments to do this using retroviral gene transfer with retroviruses that have both the TEL/PDGF beta receptor fusion and the AML1/ETO fusion protein expressed in the same context. When we do this instead of getting a differentiated myeloproliferative phenotype,

Slide 36:

we get undifferentiated leukemias, and we also get undifferentiated lymphomas.

We think that this provides some rationale for thinking about therapies that could be targeted to not just AML1/ETO but potentially in acute leukemias towards whatever the proliferative signal is, and we cannot see the 5;12-like translocations in most cases of 8;21 leukemias. Often it is the only cytogenetic abnormality in these cases, but I would propose a somewhat radical but testable hypothesis that kinase activation contributes to the pathogenesis of these diseases, and it is even plausible that agents that inhibit the ABL kinase or PDGF beta receptor kinase could be useful and therapeutic approaches to acute leukemias.

Slide 37:

I will conclude by saying that potential therapeutic targets in AML could include the nuclear co-repressor complex, core binding factor and other transcription factor fusions, the target genes of these transcription factor fusions identified by microarrays and other strategies and finally by targeting second mutations when they can be identified and potentially implicating the use of kinase inhibitors.

I will stop there and I will thank you for your attention.

(Applause.)

DR. LARSON: Are there questions for Dr. Gilliland?

DR. SCHIFFER: Charles Schiffer from the Karmanos Cancer Institute. I suspect what we have been seeing or are beginning to see is effects on more undifferentiated cells. So would there be differences in cells at different stages?

DR. GILLILAND: One of the issues may also be the cells in which the transcript and the protein are actually expressed. There is some debate about whether BCR/ABL is expressed in the most primitive hematopoietic progenitors, and perhaps that is why it is difficult to target those cells with agents that are ABL specific. So that may be one issue, and I wouldn't suggest for a moment that targets to AML1/ETO and/or to BCR/ABL or TEL/PDGF in these contexts would necessarily be the only therapy that would be necessary. It may be also important, as with ATRA therapy, to include chemotherapy for remission induction.

DR. BHALLA: Kap Bhalla from the University of Miami. Gary, I was very intrigued by your last comment, but you went to gene therapy to address this issue. We have now small molecules that inhibit both the kinases and small molecules that can, perhaps downstream to the transcription factor, either inhibit histone de-acetylase or cause histone acetylation. So perhaps the small molecule recombinations to test your hypothesis could be tested -- for example, phenylbutyrate, alone or combined with something like an ABL kinase inhibitor.

DR. GILLILAND: Yes, I think small molecule screens or testing small molecules is a good idea, and I don't want to go anywhere with gene therapy at least in the context of the news the last couple of weeks.

DR. NIMER: Steve Nimer from Sloan-Kettering. Can you comment upon two of the mouse model abnormalities? One is the ability using tetracycline inducible systems to generate leukemias that can then go into remission by withdrawing or by adding the tetracycline, and secondly, some of the evidence, for instance, from Pier Paulo Pandolofi that the fusion between PML and RAR alpha may itself constitute more than one hit, that it actually hits both RAR alpha and PML and that there may be some translocations that themselves constitute two important hits?

DR. GILLILAND: In response to the first point which is a good one, Dan Tenen has developed a very nice model that is BCR/ABL mediated. It is actually a lymphoblastic lymphoma that is tetracycline inducible. It can induce expression of the fusion gene and develop leukemia, and it will turn it off with tetracycline and get rid of the leukemia, and those may be useful models for studying therapeutic targets.

With regard to the second point, I agree that for all translocations that are balanced and reciprocal you could argue that that is multi-step pathogenesis. You are targeting both genes. You are making a forward transcript and a reverse transcript but even in Pier Paulo Pandolfi and Tim Ley's models where they express both fusion transcripts, there is still a latency. They get the right disease but there is still a latency of about 6 months in those animals when transmitted in the germ line. So I would argue even in that context that there are mutations in addition to the translocation that are important in the animal model for progression.

DR. EVANS: Bill Evans from St. Jude Hospital. Gary, could you speak to the favorable prognosis associated with either the TEL/AML1 in ALL or the AML1/ETO in AML, why that is the case, and what clues that might provide?

DR. GILLILAND: It is true that they all have favorable prognoses and that fact alone can be useful in planning or modifying therapies. The ideas that I like the most are some of those that have been espoused by Dr. Willman and others that perhaps these genes regulate the expression of resistance markers for disease so that if AML1/ETO represses the expression of MDR, for example, that that might make these cells more susceptible to therapies or less likely to relapse, but there are lots of hypotheses along those lines that could potentially be tested.