Thank you also to the organizers for the opportunity to come and talk about my thoughts about melanoma.
Now, I have been working on RAF signalling for many years, and it is a particular pleasure for me to talk to this audience because the man who discovered RAF is sitting right here.

For those of you in the melanoma field who don't know Ulf Rapp, he is the great granddaddy of RAF biology.

TOP

Now, we got into melanoma, really, by the back door because we had developed a screen.

Essentially, this is the data from that screen.

We were looking for mutations in cancer, basically, not melanoma. We were just looking for small coding region mutations in cancer.

In our original studies, we found that BRAF was mutated at quite a high frequency. So, we had a look at about 530 tumor cell lines and 380-odd primary tumors and 341 normal cells. We discovered that the BRAF gene was mutated.

I would just like to make a couple of points. First of all, obviously the numbers are quite robust here. We are not looking at small numbers of tumors.

Secondly, although we have looked at 17 out of the 18 exons in BRAF, most of the tumors turned up in exons 11 and 15.

Subsequent studies have really focused on those two exons. So, the data that you see now is almost certainly biased against finding very rare mutations, and I think it needs to be viewed in that fact.

However, having said that, those rare mutations might turn out to be clinically irrelevant or at least not important enough to worry about at the moment.

TOP

What we discovered was that seven percent of those samples has mutations in the BRAF. Really, what jumps out in the headline was that, in melanoma, 70 to 80 percent of melanomas have mutations in the BRAF gene.

Now, subsequent studies have gone on to demonstrate that our numbers were more or less correct, at least for melanoma.

We didn't discover a link to thyroid cancer. That came later by other people.

There weren't very many cancer cell lines in our panel, and so we simply missed them.

We found about 15 to 20 percent in colorectal, and Burt Vogelstein's lab did a much larger number of tumors, and they also found similar numbers.

TOP

Then other people, moving on again -- this was something we thought we would like to do but other people were much faster than us at doing it -- went on to show that these are, in fact, very early events.

Jeff Trent's lab showed that BRAF mutations occur in nevi at about 80 percent. So, this BRAF mutation could be a founder event. If it is not a founder event, it seems to be a very early event in the development of the cancer.

Again, work out of Burt Vogelstein's lab showed that the BRAF mutations were found in all four stages of (??) disease at the same frequency in colorectal cancer. Again, this is arguing that the mutation is either a very early event or, in fact, that it is a founder event.

TOP

Just to get you oriented, this is the sort of reductionist view of how the BRAF proteins operate.

Plasma membrane here, and the RAS proteins are embedded on the inner surface of the cell's plasma membrane.
In this sort of standard textbook view of RAF activation, what happens is that a growth factor will bind to a receptor tyrosine kinase.

That will activate some exchange factors using adaptor proteins and ultimately culminate in the activation of the RAS protein.

Now, the RAF proteins float about in the cytosol but, when RAS becomes activated, they get recruited to the plasma membrane, where they become activated and they then activate MEK, which then activates ERK. ERK regulates a whole bunch of things that regulate cell division.

So, really, RAF is very important because it couples RAS activation to ERK activation. For many years it had been thought, well, RAF is very important, but it had never been shown that RAF was a common human oncogene.

So, this is the way the pathway works.

TOP

Now, there are actually 32 mutations identified in the BRAF gene now. Twenty-nine of them are shown here.

So, for each of them I have indicated in red above the line -- this is the single amino acid code -- the mutations that have been identified.

You can see that they cluster in two regions, the P loop, or the glycine rich loop, which is typified by this motif here, which is characteristic of all kinases and is very highly concerned, and down here in a region called the activation segment. In BRAF, you need to have phosphorylation of this threonine and this serine for activation.

So, BRAF activation is actually quite simple. The protein gets attracted to the plasma membrane by RAS. It becomes phosphorylated on these amino acids, and then it is active and then it can do its job, which is to phosphorylate MEK.

You can see really that these are the two major clusters, and this goes back to the point I made. This is actually exon 11, this is exon 15, if you look at the gene.
So, most people look for the mutations here and here and, of course, they would never find these guys, which are much rarer. That may be unimportant because of what I will tell you a little bit later.

TOP

Now, we have made 24 of these 29 mutations, and we have analyzed them, and I don't propose to tell you everything that we have found.

I just want to make the point that they do fall into different categories. So, what we are looking at here is just a measurement of BRAF kinase activity.

Here, you see BRAF that has been activated with RAS. So, this is fully activated and BRAF. You can see that each of these mutations here -- one, two, three, four, five, six, seven of these mutations -- all have more activity in the unstimulated state than wild type BRAF has when it is activated.

Then, experimentally defined, we have got another class of mutants here. You can see these. Again, I think there are six of these.

This is wild type BRAF, unactivated, this is BRAF activated. So, the comparison is this here equals this here.

Here, you can see that these mutants all have activity that is intermediate, between activated and unactivated BRAF.

So, we went through the panel and we tried to define them based on these sorts of experimental considerations.

TOP

Now, as far as we can tell -- and I have to admit, we haven't tested them all -- those mutants that are activating are oncogenic in NIH3T3 cells. What that means is you can do transformation assays, you can develop foci per microgram of DNA.

This is what the data looks like. If you take NIH3T3 cells and you transfect in wild type BRAF, you don't see any foci. If you put in any of the activated mutants, you do see foci.

It is very important to note that actually, compared to RAS, BRAF is a pretty pathetic oncogene. Here, we have Harvey RAS, five nanograms of DNA transfected into the cells.

This is one of the active mutants. This is another one of the most activated mutants. Here, we have used 250 nanograms of DNA, so 50 times more DNA and, really, we are struggling to get as many foci as you get from RAS.

So, it is an oncogene but, by the usual sort of experimental criteria, it is pretty pathetic.

TOP

I think the last thing -- and this really goes back to these rare mutations that I keep on talking about is, this is our data for the frequency of mutations. Here is the 599E mutation.

You can see that, if you just look at the original 69 mutations that we identified -- again, this data has been borne out by mostly other people's work -- this mutation accounts for about 80 percent of all the mutations that have been identified.

Now, it turns out that, in melanoma, the data are even more startling because most of these rare mutations do not occur in melanoma.

In melanoma, this mutation accounts for about 95 percent of all the BRAF mutations that you see. So, if you think about it, BRAF accounts for about 5,400 Americans or, if the data have any value, then melanoma kills about 5,400 Americans every year, and those melanomas carry BRAF mutations.

So, if you could get an effective cure or an effective treatment for this particular mutation, you could reduce the number of deaths in this country by about 70 percent for this particular disease.

I am going to put my cards on the table, and I am going to argue strongly that BRAF is a good target for chemotherapy for melanoma.

TOP

I would just like to say a little bit about the co-incidence of BRAF and RAS mutations in those 1,000 cell lines or samples that we looked at.

We didn't see very many RAS and RAF mutations.

Particularly, the 599E mutation never occurs co-incident with RAF mutations in cell lines.

Now, you might find it co-incident in nevi. Certainly, Jeff Trent's data showed that there was a co-incidence there, and you may even find it in tumors, but you can't obviously tell if it is in the same cell.

At least if you look in cell lines, you appear never, at least, to find these BRAF and RAS mutants co-incident.

TOP

Importantly, if you look at cell lines and ask if they have got mutant BRAF, what is their dependence on RAS signalling?

Then you can do this sort of experiment, where you actually inject an antibody that neutralizes RSA function and you ask whether the cells will continue to grow.

What we are measuring here is inhibition of growth. So, a big number means that the cells have stopped growing.

You can see that, for most cell lines, RAS is important. If you inject RAS, you block their growth. clearly, there are a few outliers -- this one and this one, which we don't understand -- but the point is that, for all the mutants that carry the 599E mutation, RAS signalling is not required. So, you can knock out RAS activity in these cells, and yet, the cells continue to grow.

This would argue -- this goes back to Lynda's point -- this would argue that actually RAS therapies are unlikely to be of much value against melanoma, at least for those that carry this particular mutations. There are other ways to interpret the data, of course.

TOP

I just want to finish up with a couple of the rare mutations. Again, this goes back to our own data and the mutations we identified.

The majority of them, as you can see, are the 599E mutation. We have one 599D mutation.

This is more or less the same thing. You put an acidic amino acid in the place of this valine.

We had a couple here, 465A and 465E. These are actually amino acids that are required to get ATP into the ATP binding pocket. You would predict that these mutants would be inactivating.

So, these are two of the mutants we know. We tested them.

TOP

In fact, here you can see that both of these mutations are inactivating.

So, we have a kind of conundrum here. We have mutants which are activating and which presumably stimulate ERK activity in the cells, and we do have a lot of data to support that hypothesis.

Then we have this other rather awkward situation where you get mutations and they are inactivating and the question is really, what do these things mean for cancer and what do they mean for the disease, and what do they mean for therapy.

You could argue that, actually, it is a bad thing to inhibit BRAF, because you might get a situation where you get inactivating mutations or you block activity, and this could be a good thing for the tumor.

So, we really don't understand the biology, but I am just throwing it on the table because I like to be honest, if I can.

TOP

So, very quickly, BRAF is mutated in a lot of cancer. This is probably the most important clinically, the 599E mutation.

The question really is, is BRAF a good target for human therapies, and particularly is it a good target for melanoma.

TOP

So, if one looks at the KIT and ABL situations, KIT is mutated is about 85 percent of gastrointestinal stromal tumors, or GISTS. ABL is activated through a fusion to the BCR gene in about 95 percent of CMLs. In both tumor types, these proteins appear to regulate cell growth, differentiation, and possibly cell death.

Both of these compounds are inhibited by Gleevec or, as it is now called, Imatinib. So, really, this whole story, which I am sure you are all familiar with, serves as a proof of principle that signal transduction inhibitors can work in the clinic.

Of course, there are issues about long-term cures and resistance with these diseases, but they do serve as a proof of principle that they can work, and they argue very strongly, I think, for looking for a very good BRAF inhibitor, or an inhibitor of the mutant form of BRAF.

TOP

If you compare these things more closely, BRAF versus KIT and ABL, well, they all seem to be either founder events, or they appear to happen very early in the disease.

You can find these mutants in pre-malignant tumors for all the diseases that they are associated with.

All of them are activating mutants although, in the case of KIT and BRAF, they seem to be activating mutants within the kinase domain whereas, in the case of ABL, it is a fusion protein. Clearly, these things have different shapes, if you like.

The ABL is much more likely to look like an inactive kinase, whereas these things are much more likely to look like activator kinases.

So, if you are developing therapies, the question really becomes, what should you be targeting. It is possible that a compound like Gleevec would not work against BRAF, because BRAF is mutated and held in an active conformation, whereas ABL presumably still can adopt the inactive conformation, and there is a lot of data to suggest that Gleevec is inhibiting ABL by trapping it in the inactive conformation.

The question has to be, can you ever trap mutated BRAF in that conformation. That is something that I think we ought to consider.

All three seem to mediate proliferation, differentiation and apoptosis. Again, following on from Lynda's work, they are probably required to maintain tumors.

Now, we don't have the elegant sort of in vivo data that Lynda has generated, but we do have some cell line work using siRNA that would argue that BRAF is required for the continued growth of at least melanoma cell lines, whatever that means.

TOP

I mentioned the long-term cures and the resistance. Then, the other thing we really need to consider is the pathway for the inhibition.

So, in a situation where you have ABL and KIT, again, if you look at the reductionist view of the pathway, you have BRAF talking to MEK and ERK, ABL talking to RAS and then RAS talked to BRAF. Then, a lot of other things talk to RAS and, therefore, to BRAF.

KIT and ABL, if you take those out, presumably these other things can continue to signal, but if you take out BRAF, which is at this important node, then even if KIT and ABL are normal in the cells, there is a danger that you might knock out all the signalling through this pathway.

So, I think that there are reasons for believing and arguing that BRAF is like KIT and ABL, but there are other reasons for saying, well, it is much more dangerous to target BRAF, because it occupies the central nodal position in the pathway. I think that is something we need to consider.

TOP

I just bring to your attention that Bayer Pharmaceuticals and ONYX have this compound, which is a RAF-1 inhibitor.

I can tell you that it targets BRAF. I can tell you that it is being used in a clinical trial, both in Philadelphia and at the Royal Marsden Hospital in London against melanoma.

I don't think it is a secret to say that the data are very spectacular, or the data are not very spectacular, at least. Again, this is something else which I think ought to be discussed.

AUDIENCE: [Question off microphone].

I think it depends who you talk to. The trouble is that there is no reporter on this stuff, and the rumors go around. I am not part of the trial. Maybe I should not have gone down this street.

I am not part of the trial, and I think that it is something that perhaps ought to be discussed, as to whether this is an important compound that can be used for melanoma.

I am actually involved in sequencing the DNA from the patients, but I don't know what the clinical data is like. I am a scientist.

TOP

These are the people who did the work. Thank you very much.

TOP