SLIDES & TRANSCRIPTS
Tuesday, February 1, 2000

Signal Transduction Abnormalities in Leukemia
Jerald Radich, MD

DR. WILLMAN: I would like to thank Gary. That was a perfect talk and exactly what we want to do, Gary. So thank you very much.

Our next speaker is Dr. Jerry Radich from the Fred Hutchinson Cancer Center, who is going to give us an overview of signal transduction pathways and signal transduction abnormalities that we believe are critical in acute myeloid leukemia as well.

DR. RADICH: What I want to talk today about is looking at an overview of signal transduction, and really emphasizing abnormalities that occur in AML, and the theme is going to be that technology often doesn't work. The theme is going to be that signal transduction abnormalities may actually be a fairly common unifying theme in AML and while people have looked at various of their favorite genes in single transduction abnormalities, ras, fms, etc., if you actually start looking at the whole picture in a given subset of patients, abnormalities of one of the pathways may, in fact, be quite common.

So what I want to talk about is signal transduction abnormalities in AML, and the bottom line is I am going to try to weave a picture where abnormalities are actually a unifying concept of AML. This is sort of signal transduction for poets. What you really want in the system is a way to sense external stimuli, and could we translate that into changes in the internal environment to signal new genes and have this so that when you turn on the signal you quickly flip it off. Once you give the signal to go and you have new genes involved in proliferation and differentiation and the like, you can quickly stop that mechanism so that it doesn't go on uncontrolled.

TOP

Here are the pieces that we are going to be talking about (Referring to Slide).

We have the extracellular and intracellular membranes with transmembrane tyrosine kinase receptors. Grb-2 is an adaptor protein that we will talk about. SOS is a guanine nucleotide exchange protein that will transfer GTP for GDP in ras-DGP. F is a farnesyl unit that is basically put post-translationally onto ras so that ras can associate with the plasma membrane, and GAP is the GTPase activating protein. We will talk about the components of these together. When this signal is activated, it gives signals for proliferation, differentiation and signals about cell survival.

TOP

This is what happens with activation. We have a stimulus. Some type of growth factor comes on board and binds to tyrosine kinase. When that happens, the tyrosine kinase is dimerized, causing autophosphorylation of the of tyrosine kinase domains.

Once those are phosphorylated, they become a site for the adaptor Grb to bind (via SH2 domains which bind phosphorylated tyrosines). Then the proline rich region of SOS binds to to the SH3 domain of Grb2, linking Sos to ras-GDP. Via farnesylation, ras becomes associated with the plasma membrane. SOS causes GTP to replace GDP in ras, leading to activation. When that happens, ras changes its conformation and that is the signal for these downstream effectors. So that is the on switch, and now GAP basically hydrolyzes one of the phosphates and then returns to switch the off formation. Ras has an intrinsic GTPase activity but GAP accelerates this intrinsic GTPase function. GAP increases the GTP activity by about a hundred-fold.

TOP

So let us talk about some of the downstream effectors. Once this is going, and the stimulus is going through, there are several different signaling pathways that get activated. This is just a summary of three of them.

The main one is raf which is a MAP kinase that actually directly associates with ras GTP and that is the MAP kinase system. It activates ERK which then stimulates the jun kinase activity which activates jun. That binds with fos. You get an AP1 complex and you drive transcription.

It probably also activates cytoskeletal function through rac and rho, and also probably influences survival. There is some evidence now that ras when activated goes through the fos 3 kinase pathway, and maybe raf directly goes in to activate BCL2 which is basically promoting cell survival. So once ras is on, it activates multiple signaling pathways for gene expression, cytoskeletal changes and survival changes, and again, you have to treat this and lock very quickly. So GAP turns off the signals and these are tightly, tightly controlled.

TOP

Letis talk about the first set of linchpins in this pathway, the ras oncogenes. There are three types, N, K and H. K and H were found in avian retroviruses. N-ras is not. They are virtually all GTP-GDP binding proteins that have intrinsic GTPase activity. That activity is increased by GAP. These are expressed in all tissues but certain tissues express more than others do, for instance with N-ras mostly in hematopoietic cells and in the brain. Point mutations in these three genes are all activating -- very specific point mutations.

TOP

So this is what we are looking at here. The box is the model that we are looking at right now, and what we are talking about is mutations in ras, and just to make the emphasis that this is some conformational change, I put this kind of wart on ras GTP.

Mutations occur usually in codon 12, 13 or 61. These are again activating point mutations. In AML, N-ras mutations are more common than K-ras mutations, and H-ras mutations are relatively rare.

All these mutations tend to put ras in the on position, and if you look at the intrinsic gap activity, this is decreased. So these mutations are activating and prevent the hydrolysis back to GDP. So you basically get an inappropriate signal flowing this way. The signal goes on, but there is no easy way to turn it off.

TOP

Now, you can look for ras mutations in patient samples in a number of ways. This is our preferred way, using SSCP -- single stranded conformational polymorphisms. We do a PCR of the region of interest and then run on special gel conditions, and you can discriminate changes of 1 nucleotide base pair. So these are AML samples. These are the normal single strands, and you can see in this one, THP1 which has a mutation in codon 12 there is a shift in this band which is a conformational change due to the nucleotide substitution. Here are a couple of other samples here and here which also show shifts in bands. So there are heterozygous mutations for ras, and this is a way you can quickly screen for mutations in multiple samples.

TOP

Ras mutations are pretty common in myeloid leukemia. In AML, about 15 to 30 percent have mutations. In MDS, it has been reported anywhere from 5 to 40 percent. The question is whether this depends on the phase of disease. There have been numerous studies that show as you go from RA into transformation and into frank AML that these cases may accumulate ras mutations along the way and be involved in the transformation.

In juvenile CML, at least those that are wild type, NF-1 ras mutations probably occur in 20 to 30 percent, in CMML 30 to 50 percent, but in CML they are rare.

Now, these mutations aren't random. If you think about the mutations you would expect if this were fully random, there would be twice as many transversions as transitions because if you have a purine you can go to a pyrimidine. So there are at least twice as many chances for a transversion, but in fact the opposite is true. There are at least twice as many transitions than transversions in these mutations, and GA is the one that happens the most.

TOP

That is the ras part. Let us go back and look at upstream mutations because changes in tyrosine kinase would also cause inappropriate signaling this way and give you functionally the same effect as a ras mutation. I am going to center on this part right now.

TOP

There are three well-documented tyrosine kinase mutations. The oldest is fms, which is the receptor for M-CSF which is found expressed mostly in monocytes. It is found that point mutations in either codon 301 or 969 are activating for fms and are present in about 10 to 20 percent of AMLs, especially M4 and M5 types.

Mutations at kit were originally found in mast cells and especially mast cell leukemia cell lines. We are now looking at unselected AML cells. It is found that they actually may have mutations, either point mutations, deletions or insertions at somewhere between 5 to 10 percent.

One that we are particularly interested in now is the FLT-3 mutations which have a unique type of mutation called internal tandem duplication. I will show you a cartoon of this in a minute. These appear to occur in at least 15 to 30 percent of AML patients.

TOP

iFlti stands for i the fms-like tyrosine kinase,i a Class III tyrosine kinase, which is expressed in earlier progenitor cells especially CD34 positive cells and immature lymphocytes, and it has been found in virtually all AML samples.

TOP

This is its structure. It has five immunoglobulin-like domains in the external region, a transmembrane domain, and this juxtamembrane domain which kind of peeks through the membrane, and this is the area that has the internal tandem duplications.

TOP

These are repeats of anywhere from usually 18 to 200 base pairs. They are sometimes a little bit smaller than that. They are put end to end and in all cases found so far have been both in frame and activating. These are found in about 15 percent of pediatric samples and maybe up to 23 percent of AMLs. They are associated with high white counts and may be associated with worse prognosis. This is a cartoon of what it looks like. If you have a normal exon 11 and exon 12 of FLT-3, and let us say this highlighted area in blue is the area of concern, what you find in the ITD mutant is this is put end to end. So you get a duplication of this area right here, and if it is not cleaved, if it is done right at the end of the codon by some process we don't understand, there will actually be an insertion of a couple of nucleotides that keep this whole protein in frame. This is how you can screen for it just by PCR assays just on size. You just take your samples and do PCR across those exons and you get your normal size and you get these mutations from the ITDs which can be in various sizes.

TOP

It looks like the prevalence of this ITD goes up as the age of the patient goes up, so that in kids this seems to happen about 14 percent of the time, in adults about 23 percent, and we have looked at elderly AMLs. Over 30 percent of them will actually have the FLT-3 ITD. So it is very common.

TOP

So those are some of the upstream signals of inappropriate signaling. We talked about ras. Now let us talk about another theoretical place where you could have disturbances and that is in GAP function because if you just decrease GAP function entirely, signaling will go this way, and again the signal will be on and there will be no way to release it.

Now it turns out that there is actual GAP and there is another homologue which is NF-1.

TOP

The NF-1 gene was cloned. Neurofibromin (NF-1) has very close structural homology with GAP including an area called the GAP-related domain, the GRD. NF-1, like GAP, is involved in pathways that are integral to myeloid development.

If you look at kids with Neuroblastoma, they have a high propensity to get myeloid malignancies. Kevin Shannon at UCSF looked at kids who had myeloid malignancies, especially juvenile CML, and found that they all had lost heterozygosity of the normal NF-1 allele if one retained the mutant allele.

Now if it was really true that NF-1 and ras mutations have functionally the same activity and that they allow inappropriate signaling, if you looked at a series of juvenile CML patients and those that have the NF-1 mutation, you wouldn't expect to have any ras mutations and vice versa. In fact, that is what we found. Ras mutations in wild type NF-1 pediatric myeloid leukemias happen in 25 percent, but if you looked at NF-1 patients that had juvenile CML or other myeloid malignancies, none of them have ras mutations and that makes some intrinsic sense.

TOP

Lastly, where some of these frequent translocations in myeloid malignancies tie in are also in this ras signaling pathway. When you look at bcr-abl in CML, it turns out that there are domains on bcr that serve to grab GRB2 and activate this pathway. If you look at the tel-PDGF, tel is a transcription factor and has several helix domains. These seem to cause inappropriate dimerization and allow again for signaling through grb and SOS. So even the most common signaling domains also seem to feed into the system.

TOP

In summary, look at all the events that may be taking place here: tyrosine kinase mutations occur in probably at least 20 percent, the translocations, ras abnormalities in over 20 percent, maybe some abnormalities in GAP and NF-1. We haven't even looked at things that might increase the expression of N-ras. That might be another pathway even irrespective of these that might drive the signal to inappropriate proliferative responses.

TOP

One of the things you want to ask then is most of this stuff has been done on individual patients. Someone looks at FMS. Someone looks at their RAS. If this is really a unifying theme, if you looked at the same population and sort of marched up and down the pathway, maybe in fact you could find abnormalities in an overwhelming majority of AML patients.

We took 140 AML patients and are starting now to kind of march through the pathway. If you disregard the p53 stuff but look at RAS and just FLT-3, we found in these 140 patients about 20 percent had ras mutations, and if you take the same samples and say, how many have FLT-3, you get another 30 percent. So just by looking at these two components, we are almost up to half the AMLs that have disturbances in one of these two genes, and we haven't even gotten to the other tyrosine receptors and some of the other possible partners in the pathway. So it looks like you may accumulate a lot of lesions in this pathway.

TOP

There are a number of potential sites of attack that you might be able to harness with this pathway. One is farnesyl transferase inhibitors, and we will talk about those in a second. Others are peptides, now that we are getting more crystal structure, that could inhibit the SOS binding site to ras, inhibit the GTP binding or inhibit the binding to raf. The underlying theme given the mouse data is that maybe while you want to have molecules that would target oncogenic N-ras and spare normal N-ras, maybe you don't really need to do that. Maybe you could knock out all ras function and maybe the H-and-K-ras will take over in somatic cells.

TOP

First, let us talk about internalization of ras. Ras to function normally needs to have fat added onto it, and in this way I feel one with ras. We need fat to function, and what happens normally is that a 15-carbon farnesyl group is hooked onto this residue of ras. There is an alternate pathway of adding fat, the geranylgeranyl fat moieties which are 20 carbon moieties. It turns out farnesyl transferase is very important in H-, K- and N-ras, but in N-ras this pathway can be used very well. So it may well be that the farnesyl transferase that drugs are knocking out might be very important for H-ras. There is an alternate pathway for N-ras that may make it actually a little bit more difficult to target, but there are various types of inhibitors that are used. Basically, they are targeting this interaction of this carbon and this chain here. There are peptidomimetic molecules that basically mimic this and so the transferase hooks onto this molecule.

There are monoterpenes which basically are 15-carbon moieties that then hook on here and make ras inactive and stay in the cytosol, and there are bisubstrate molecules. Most of the studies that have been done are on solid tumors, which are predominantly H-ras, and it is unclear whether or not these types of inhibitors are actually going to do much for N-ras because of the alternate pathways that it can utilize.

TOP

Peptides probably are fairly promising in that we know now there are map areas for SOS-ras interaction and raf-ras interactions, and it may well be that with small peptide molecules you can block those interactions and essential strangle the downstream signals of ras.

TOP

As far as research directions go, we need to explore the differences of structure and function of N-, K-, and H-ras. So far, most of the work done with crystal structure and pathways has been done with H-ras. There appear to be differences between N-, K- and H-ras in both the downstream signals and their actual crystal structure. We have to interrogate whether there might be an alternate pathway for mutant ras because, if there is, it may actually give us a unique pathway that we can target. We need to look at expression in a more wholesale way to investigate differences between the different ras pathways and mutant ras pathways. One way this might be done is by using expression arrays where you can look at wholesale changes in 5, 10, 50 thousand genes, and be able to look at the downstream pathways of each of these ras under normal conditions and at mutants. This is something that we have talked about doing with Dr. Willman.

TOP

This is just to demonstrate that you can actually do this stuff. Our work chiefly in my lab is looking at the progression of chronic phase myeloid leukemia to blast crisis, and this just shows the power of these expression arrays. These are four chronic phase patients going this way and three blast crisis patients going this way. Each of these lines represents one gene. This is a 10,000 gene chip and there is a mathematical algorithm that clusters the genes by expression patterns, red up, green down. One of the things you can see is that the chronic phase patients cluster very well together. They are very different than the blast phase patients which cluster to each other but it is different than chronic phase. In this context this is just an example that you actually may be able to map pathways fairly specifically.

So it would be fascinating to look at a inormali AML patient versus an N-ras mutant and an N-ras mutant versus a K-ras mutant to get a better idea which of these pathways are really involved in the leukemogenesis.

TOP

To summarize, signaling abnormalities involving ras are quite common in AML. Again, we have just started to look. We can find mutations in 50 percent of AML patients by just looking at two genes so far.

There may be common choke sites to blunt the effect of these mutations downstream. Especially there may be ways to block with peptides interactions between some of these, like SOS and raf, to blunt the signal.

Ras pathway redundancy may give us more flexibility to treatment options. If the mouse model is true, we may be able to eliminate N-ras function entirely and hematopoietic cells may do well, and the last thing is there is lots more to do.

DR. GRANT: Steve Grant from the Medical College of Virginia. A very interesting and provocative discussion. My question is, and I don't know if there is an answer to this, but what factors would cause you to focus on upstream targets rather than say potentially more specific downstream targets of ras? For example, raf which can be inhibited or interfered with pharmacologically with vanomycin or even further downstream using inhibitors which are becoming available? For example, AML cells are characterized by increased activity of MAP kinase. So what theoretical advantage might ensue from inhibiting upstream targets rather than going to the specific downstream targets?

DR. RADICH: I think looking upstream is mostly just to really quantify how often the pathway is involved. I think where the money is, where the chokeholds are, are all downstream. So if we pursue this stuff with AML patients and find that, in fact, if you start adding ras and FLT and we look at fms and kit, all that is going to tell us is how often that pathway is potentially involved. I think as far as the inhibition goes downstream is where you want to do it.

DR. STONE: Rich Stone from Boston. Could you comment on your views based on the preclinical data about how a ras-specific drug like FTI inhibitor might affect cells that have a ras mutation versus those that don't, and which do you think would be more likely to be clinically effective?

DR. RADICH: I would say that would be very hard to know, and the reason is that the new literature that has just come out, it looks like a substantial amount of the ras pathway or the ras activation may be going to pull you around the diagrams that I showed, that it may actually be a direct interaction bypassing the MAP kinase pathway entirely and going straight to jun. So I think that would be easy to find out. I don't harbor even a guess.

DR. GILLILAND: Gary Gilliland. Could you comment on cytogenetic abnormalities, if any, in the FLT-3 mutant AMLs?

DR. RADICH: Yes. So far where we have looked in the adults and in kids, it is fairly well distributed over the whole panel of low and high risk cytogenetic groups. So they do have other cytogenetic abnormalities, but if you look at clustering between low risk and high risk, at least so far it hasn't broken out in this style.

DR. GILLILAND: These are all pediatric AMLs?

DR. RADICH: The ones I showed were adults, but I have done it also in pediatrics.

DR. GILLILAND: I was specifically asking about correlation with cytogenetic abnormalities like 8;21 or inversion 16.

DR. RADICH: So far we have not found it in the kids at least, and in the adults, there have been a couple of 8;21s but not many at all, and I don't think any inversion 16s.

TOP

TOP

TOP

TOP