Cancer Research Logo

good afternoon. so we'll get started. so already i've been told by one unnamed member of the audience today that this is the golden gate bridge. but you all know that it isn't. [ laughter ] it's the logo of this course,

that's the brooklyn bridge. and today's subject is particularly well suited because the thesis of a bridge, of course, is that things are different on both sides, and when you build the bridge, they become different on both sides again because they become mixed

and so that's the dilemma that we'll be discussing today, the ras ongo gene, the most commonly mutated oncogene in all human cancer, with an enormous amount being known about its structure, mechanisms and so forth. and then on the other side of the bridge, it's mutated in so

many different human cancers, and the problem is how to put it together particularly in some therapeutic manner. so where is -- oh, i have to do this. hmm. so just by way of brief introduction, for those who are

not ras aficionados, ras is the originally named for two rat sarcoma viruses, which were identified by two different investigators, harvey and kirsten, and then the genes responsible for these viruses, for their transforming properties, were discovered here

at nih, mainly by ed scolnick and his associates, and then later they were found by many, many others, in various cancer cells. and they were transforming when put into tissues and cells. and then in 1982, several people, including if i recall

correctly, i think mike widler at cold spring harbor was studying a neuroblastoma and found the third member of the family which is called nras because of that. the three human ras genes, very similar proteins, 188-9 amino acids, are gtaases which control

many aspects of intracellular signaling networks and when overexpressed are associated with growth, differentiation and transformation. now, it's interesting, and this is the challenge, the frequency of hyperactivated mutations of each of these ras genes occurs

in different distributions with different cancer, some like pancreatic, kirsten ras is overactive, over expressed, mutated in almost 100% of the patients. the question is given all this information about, which we'll hear about much more, why is

this diversity in ras genes? how does it happen? what is their potential? what's their differential function? normally and in malignancy. and given this propensity to be overexpressed in different malignant diseases, are they

really therapeutic targets? and if so, where do we stand? so the answer to that is to build a bridge, and the bridge that's being built here is the ras central at frederick, which was started i think a year-and-a-half ago, and frank mccormick, who is the director

of it, is one of our speakers today. this is an effort about which we'll hear more, but which serves to try and answer, bring together, not only from nih but from all over about this important question. so we're very fortunate to have

two experts discussing both sides of this proverbial bridge, so this first speaker is going to be susan bates, who i suspect is known to most of you. susan's the head of the molecular therapeutics groups in the developmental therapeutic branch of the nci.

she graduated from arkansas school of medicine, trained in georgetown, and in oncology here, and she became tenured in 1992. she's very well known for her exciting work on drug resistance and malignant disease. susan is going to discuss the

sort of status of the ras genes with respect to human cancer. and then frank mccormick will take up the theme and for those of you who don't know, frank is emeritus professor at the moment at medicine from uc-san francisco. and from the comprehensive

cancer center there, he received his ph.d. degree in cambridge and then was a fellow at stoneybrook on long island with c marcone and imperial cancer in london with allen smith and when he went to california began his career with three leading biotechnology companies, almost

from their beginning, cetus and carlon and he formed onyx company. now, the goal of all this was to study the malignant process and to develop therapeutic agents, some of which he did develop, particularly saranafibid. a member of the national

academy, received untold honors which i will not go through to relate to you except to say he's been a leader in this field and very productively, and we're very grateful that he is here. in his present position he's oscillating between san francisco and frederick, because

he's the director of the ras initiative which i briefly mentioned before. susan? >> all right. thanks. well, i want to say thank you for inviting me to do this. when he invited me i said, well,

i'm not really an expert in pancreatic cancer or lung cancer, but i said, i'm a medical oncologist, and medical oncologists have been basically ting ras for the entire time we've been medical oncologists. so, i framed this to be basically a leadin.

i'm the warmup band for dr. mccormick's talk because he's working with the ras initiative to come up with therapies for ras cancers. what you're to think about here is what are we going to expect from that initiative, what can we hope for and what can we

expect? so the first question, is it the holy grail? i told him it's the holy grail. he's always questioning things. it's not the holy grail. we had an opposite opinion of ras. you can decide at the end is it

the holy grail, is it not the holy grail. so we're going to cover these things. how often do we find it? in what diseases? what does it teach us for what we find it? what is its role in outcome in

whether patients do well or don't do well, what is it's role in affecting therapies, in the origin of cancer, what approaches have been attempted and how much benefit can we expect from targeting ras? we're going to scratch the surface.

many papers have been written. this is the time line when it got started. this is 50 years ago. we've known about the transforming potential of these sarcoma viruses for 50 years. just to follow through a few details that i wanted to point

out, very early on we knew the viruses, we knew they were transforming. we now dr. varmus actually discovered that a normal mammalian counterpart would be a normal cellular counterpart to the oncogenes, and that they were -- overexpression would

transform n nih3t3 cells, mutations with discovered and they were transforming, so you could either make a cell become a cancer cell, either by putting in a normal -- a lot of normal ras, or by making a mutation in but then there were papers that said you can't actually

transform a primary cell, take a really normal differentiated cell, you can't transform it without some cooperating genes. and then eventually the fact that ras has to be farnesylated, this is a time line to the start of medical oncology for ras.

it is a small gtp-ase, highly conserved, but for this one five prime region. and then the switch that occurs between ras off and ras on is what is actually impaired, so you can't turn it off once you have the mutations that have been discovered in the viruses

and in the cancers. that's what we'll talk more about. so the incidence in cancer, i'm not sure it's our highest mutated cancer but that may be true, in 22% of cancers, k ras is. some of the others are less, h

ras is only in 3% of cancers, and n-ras in 16% of cancers, a differential distribution across them in which we do speak, for example, in hematological malignancy more nras and less k-ras, in solid we see mostly k-ras, and h-ras is much less common, even though that was the

first discovered. here is the mutation, and the point to be shown, in pink are cancer-causing mutations, in green are mutations that we're not going to get into today, maybe dr. mccormick will, but the -- these are what's called ras-opathys, in diseases not

cancer related, these mutations are occurring more across the protein, whereas there are three hot spots, 97% of ras mutations are in these three hot spots, codeon 12, 13 and 61. it's very rare at other residues and other locations. and i would just also, if anyone

wants to do more reading, this is an outstanding review article. the different mutations -- for example, the k-ras isoform is mostly mutated around codon 12. h-ras is more in codon 12 and codon 61. the other interesting thing,

k-ras is mostly codon across cancer mutations but that's not true in every single cancer but in pancreas, for example, it's almost totally codon 12, and that's cigarette smoking. one of the interesting things about was i mentioned is that it's farnyslated, it has to plug

into the cell membrane leading to the first therapy tested for ras, and we'll come back to that in a bit. ras sends signals through at least six validated intracellular pathways and signals at least as far as we know currently for multiple

important things in cancer cells, survival, progression, migration, endocytosis, calcium signaling, really every cellular function is in some way affected by ras being turned on and being left turned on. we'll focus on the mutations that are present, that have been

described in acute leukemia, in colorectal cancer, 33%. acute leukemia is 5% for k-ras, 15% for nras. lung cancer, in about 1/4 of lung cancers, that number is a little bit low, and then pancreatic cancer 60%. so what can we learn from

clinical data in these specific cancer types that will help us decide what benefit my accrue from a successful ras initiative in frederick? this is a study of lung cancer, 670 670 patients with the k-ras mutation.

those who were smokers, currently or former, have a nucleotide transfer, resulting in the glycine to cysteine mutation at codon 12, whereas never-smokers go to aspartic acid. it's interesting that tobacco smoke cause us a typical

mutation and typical outcome, so i think the biology of the differences between the mutations remains to be really worked out. one question, we always thought for years and years that ras made people have much more aggressive cancers, die much

sooner, but there's really not much evidence for that. leukemia and colon cancer, lung cancer, we're not really seeing that. when it's put to a test where you look at a specific cancer type it seems that the -- a specific subset, it seems

there's no real impact. so this is the kaplan meyer curve, this is the way we typically look at the outcome in cancer patients, and here you can see blue for the wild-type and gold for those who have mutations, those curves are overlapping so no impact on

outcome in lung cancer. and then so the question is could it be a therapeutic target in these quarter of patients with lung cancer that have a k-ras mutation? lung cancer in recent years has become incredibly interesting because it's been subsets into

different mutations that are oncogenic drivers, so we have egf receptor, epideterminal growth factor receptor driving a proliferative pathway, and many, many other small percentages but nonetheless where we found, for example, alk is 5%, add an alk inhibitor gives marked benefit,

adding egf receptor gives marked benefit, suggesting we should get marked benefit when we inhibit ras in lung cancer. but we don't know. that's what the question is going to be. now, the epidermal factor is activated in lung cancer, a mek

inhibitors is blocking this signaling pathway. if we have k-ras, i showed you six pathways, but this one anyway is one of the most important because it leads to cell proliferation, and we thought it was important, so what does mek inhibition do?

added to chemotherapy, this is a water fall plot, how we show whether tumors shrank or didn't shrink. here they are increasing in size, decreasing in size, you see selumetinib gave a few responses, not a home run, nothing like the alk or egf

receptor but it is mek, not ras, so we still can hope that we get major benefit from ras inhibition in lung cancer, but this was a little disturbing that we didn't see more from mek inhibition here. if we move to colorectal cancer, one things that's interesting,

it arises from at least familial and many syndromes, at least, from early defects. and so ras doesn't appear as a mutation until later on. we'll pause for the recording. it doesn't appear until a little later on. and then other mutations

accumulate to get the final very invasive cancer. how important are mutations that arise later? can we say which one is truly oncogenic? it's an accumulation of mutations for sure but what does it mean that ras arises about

halfway through that process? another point that says ras may or may not be the most critical thing in the colorectal cancer, although it's prevalent in 30%, is that when people have looked at mutations in lymph nodes, distant metastasize, it's not 100% concordant, you don't

always start with a wild-type and end with a wild-type though generally you do, and you don't always go from mutant to mutant, though generally you do. but there are examples where you have k-ras appearing later in a new clone at the time of relapse, and that's a very

common finding actually in leukemia which we'll get to in a few minutes. what about clinical trials? what can that tell us about how important k-ras is in colorectal cancer? there's two drugs used to inhibit the egf receptor

pathway. are we okay? okay. cetuximab was shown to benefit colorectal cancer, fda approved it in 2004. they approved it with not very many months benefit in terms of disease shrinkage or control but

nonetheless they approved it, it wasn't selected for particular patient subsets. but then four years later a group discovered that in the trial that was done on which the fda approved it, they found that k-ras mutations actually do not give benefit to patients who

have -- are receiving cetuximab, no benefit, it is inhibiting the map kinase signaling pathway, no benefit, in the presence of the k-ras mutation. and if you have a wild-type k-ras, yes, you do get benefit from adding this egf receptor inhibitor cetuximab.

so that was one of the first clues that we have to be careful with inhibiting egf receptor pathway because downstream of that we're going to have alterations that may affect whether you respond or not. so here you have egf receptor signaling through k-ras, b-ras,

mek, and you're going to not be successful if you've got your signaling starting below at the level of k-ras. that's also true if you have mutation in braf. a provisional statement in 2009, maybe you you shouldest testfor k-ras and four years to end up

with an fda approved medical test, a device, testing for k-ras mutations. this is and done and recommended in colorectal cancer in combination. another interesting point is that this k-ras test is fda approved, doesn't test for every

it tests for multiple codon 12s, which is appropriate for many cancers, but not for all cancers, so you can't think i'm going to always use my k-ras test because you'll miss codon 61 and other mutations in this group here but it does pick up the one we talked about, being

in lung cancer. and so this has been reproven several times, even clinical trials combining multi-agent chemotherapy with k ras testing found only patients benefited who are wild-type tumors, so very clear, have you mutations, you should not be treated with

anything that inhibits the pathway above k-ras. so -- and this is not beneficial and could even do harm. i think this fundamental fact that you have this be aeration implies the pathway is critical in proliferation in colorectal cancer.

here is the braf result where you have treating. they have looked for other types of mutations in the k-ras wild-type groups, trying to see could we have had a better response rate if we got rid of downstream mutations and found in the wild-type group are a

number of other your mutations, braf and n.r.a.f so you don't have ras mutations in the same they looked at response in these patients who had aggressively removing. you remove the braf mutated disease, remove those with nras

mutations, and remove pik3ca, you can double the response rate if you basically define your mutation spectrum in your cancer cell. this is where the idea of personalized medicine really starts to be -- this is true, you should not give patuximab if

you that these mutations. this is different from colorectal and lung cancer, when people tried to look very early, just the dysplastic epithelium, in situ, the very earliest point, you already see k-ras mutations. this is either true in the first

mutation or something hasn't been found that needs to be found. in pancreatic cancer the first mutation is the k-ras mutation. in fact early neoplastic lesions, 95% have k-ras mutations, 95%, the vast majority.

when you look at progressively worsening intraductal lesions there's not much of an increase which all of that is consistent with the idea k-ras is the important oncogenic driver and source of pancreatic cancer. so that suggests that it is fundamental to pancreatic cancer

onco-genesis and this is the proving ground. so if the ras initiative is going to come up with a drug, it better work in pancreatic that's a huge challenge because this is one of the most drug-resistant cancers we have. so then aml brings us a really

and different story. aml, this is a variety of patients, you don't have all these mutations in every patient obviously, but they found a number of different mutations and then have classified these mutations in acute leukemia, so mutations that affect dna

methylation, chromatin remodeling, chromatin modifies, transcription factors, across the board, and mutations in ras as i've alluded to. and also mutation in k-ras, more rare, but also pit, a kinase with activated signaling versus these are epigenetic mutations

that people like to say affect the landscape. so it's interesting when they have done this kind of circos diagram that the signaling mutations are linked -- you never have two signaling mutations, they are non-overlaps but they are linked to these

more epigenetic, the landscape mutations are what people are calling them. you need an aberrant landscape, in cooperation with t signaling pathway mutations, so every time you have a signaling path way mutation in leukemia, you have a landscape mutation.

and so what a number of people, tim lay and others have done, the landscape mutations are occurring first. a cup of lines of evidence for that, in elderly patient who do not have aml they find landscape mutations and clonal populations.

so almost suggesting if patients live long enough they will get leukemia, rising in those clonal populations that have the landscape mutations. and here is an example of a patient with leukemia, so they are calling this landscape mutation that founding set and

the proliferative signal is coming from the flt3 mutation, the subclones are going to outgrow and the patient is going to have frank leukemia from. and when you treat the patient, and then after 30 days patients are in remission, the proliferative genes that are

mutated disappear, they are gone, and this is a number of patients, but the proliferative genes disappear but the landscape epigenetic modifiers, those persist. so those pieces of evidence, and here is a different paper showing basically the same

thing. you have pre-leukemic clones, you have idh 1, 2, these are epigenetic mutations, this is something that affects ribosome biogenesis, histone chaperones, these are there. these are the signaling and those are just never found in

those early pre-leukemic ones. the thinking is you have the founder, landscape mutation and then that expands and then you get a proliferative mutation and then that expands, and then you have frank acute leukemia. so that suggests, differently from pancreatic cancer, that ras

is not really the precursor lesion. you could get a great benefit, but you probably are not going to likely cure a patient unless you can address the landscaping and then there's another point that makes it more difficult, and that is that there's a whole

super-family of genes that also are small gta-ases, and i would like someone to tell me that they are not involved in cancer because i don't think that's been proven yes or no yet. and so quite a number of them, ras, rho, arf, each as a family. the was family and rho family

here, for example, so we don't know to what extent they are involved in cancers. but they are gth-ases and could serve like ras and testifies recently reported evidence of roh mutation in t-cell lymphoma, where the epigenetic lesions why known but can we find out which

came first? what about therapeutic targets? as we mentioned earlier, ras mediated signaling through six pathways is well known, the map kinase which is the one we studied longest, braf, we've done some work around this already by using an inhibitor

downstream and mek inhibitors and we've not seen home runs in that's not to say they have been used in well-defined studies, but they have been used, so that's concerning for how well we're going to do. but one has to hope when we inhibit ras at the top of all

this we'll be inhibiting all the pathways and we'll have more success. so what have we done clinically? we had a transferase inhibitor, i mentioned you have to have farnesylation to allow the k-ras has a separate and nras a

separate pathway. they were taken to the clippic and studied intensity. both of these made it into phase 3 testing, in the clinic, and it's remarkable in the oncology literature because there's paper after paper after paper and then it stops.

it just comes to a big halt. now why did that happen? so what they found early in aml was there were responses in 10 of 34 patients, none of those patients had ras mutations but they said, well, maybe this drug works anyway. we don't really care if it hits

the target or some other target, we're going to go ahead with it. but it did show inhibition of farnesylation. they used surrogate substrates and were able to confirm inhibition of farnesylation. the phase 3 style that randomized could supportive care

could include it, response rate 8%. for all that drug development, complete response rate 8%. in leukemia if you don't get a complete response it's not a good situation. so 11% overall.

so they packaged this up and went to the fda, to oncology drug advisory committee who said not enough activity to be worth fda approval this is 2005. the company continued development. they said maybe we can successfully do this, we're

inhibiting farnesylation of other targets, but ultimately someone came up with a two-gene signature, this reflects sensitivity to tipifarnib or inhibition of farnesylation and took the two-gene signature and then they took tipifarnib to clinical trial, selecting

patients based on the two-gene signature, cr rate 11%, it went up by 3 over unselected. notice they still didn't pick h-ras or nras. they picked the two-gene signature. so the company decided to terminate further development,

but the light at the end of the tunnel, now we're learning so much more, they have licensed this out for development in hepatitis delta virus, which one says, oh, yes, i know about this. it turns out that another -- the other advanced stage farnesyl

transferase inhibitor has a new life in inhibiting hepatitis delta virus, orally activity, inhibitor, and the enzyme is required for hepatitis delta virus infection, and it uses that cellular process in order to complete its life cycle. so when patients received

lonafarnib you have a marked decrease in hepatitis delta virus. what we could see happening is if we come back and we say we understand this process a little better, we could end up repurposing these drugs but that's sort of a dream.

we're almost done. so what about that gold standard i said pancreatic cancer? we ought to be able to inhibit was if we're going to do anything, target ras, it's got to be in pancreatic cancer. so the farnesyl transferase inhibitor tipifarnib, no

benefit. in combination, no benefit in terms of survival, duration of benefit, or in response rate. so no wonder the company decided to stop development. so the bottom line, it's not clear that ras mutation confers a worst outcome but it does

interfere with egfr signaling blockade, when you have a ras mutation it's not good. beyond that, its role in onco-genesis is being worked out. in the early days it was not able to transform primarily, and then the new data.

k-ras mute and not mutant subclones often coexist. it's important but inhibition may not be sufficient in many cancer types. so we need definitive k-ras blockade to answer many of our questions about the role of ras in origin and maintenance of

cancer, you have your mission. the k-ras initiative, all of nih, nih has been involved in ras development all along. i want to acknowledge great collaborators at the nih and helping me understand biology of [applause] thank you.

>> susan, that was a great summary of a very difficult and complex field. maybe we h few questions if someone has a burning question they would like to ask right now? otherwise we'll have more questions at the end after

frank. yes? wait a minute. we need a microphone. no, no. >> i'll wait. >> he took mine. thank you. >> hello.

so you showed that ras is an initiating event in pancreatic cancer but appears to be accumulating in others. what is it about the pancreas that selects for ras mutation right off the bat? >> i don't know. one idea batted around in

literature, due to bile duct acid, in that area. [low audio] >> well, right. well, it does. you have gallbladder cancer and bili-rate cancer. they do have ras mutations but that's not an explanation,

that's just one thing that i read. very good question, why is pancreatic cancer so susceptible. on know. >> that's another area of interest.

inflammation. i remember the day the paper came out with the sarcoma virus injected into a chicken, it didn't get cancer unless you clipped a wing and it got big inflammation and cancer. in human cancer that's been clear for a long time.

>> could i ask, when any of these tumors as lines or as primaries have been put into animals, do they respond to the farnesylation inhibitors? >> for many of these, the animal data, animal models, we use a xenograph model, put into the skin, the response is better

than in actual patients, much better. also a work in progress, developing better models. >> you showed us that in lung cancer if patients have ras mutations and you're targeting egf you don't get a good response.

>> in colorectal. >> if you do the same, and get a good response, when the patients relapse with resistant disease do they have ras mutations in the relapsed tumors? >> that's been reported. you can detect ras in dna, it's been reported, after the

treatment in colorectal cancer. some people dispute those data. >> susan, what is the thinking about frequent coexistens of mutation in ras. >> what's the thinking about it? i have no idea how the ras mutation occurs, but perhaps this is a hot spot.

perhaps there's something in the chromatin that this is a place that is not easily -- that is easily upset so you end up with a mutation, and then once you have this abnormal chromatin, it's even more easily damaged. so i don't have an answer for that at all.

leukemia data is very striking i think. >> does inflammation upregulate alter ras expression? >> do you have the answer for that? right. >> okay. i think, frank, if you'll

continue and we'll have time for questions at the end of dr. mccormick's talk. that was great. >> that was a great introduction. thank you very much. thank you for inviting me to come down and talk.

okay, thank you. i'm going to continue the same think but focusing on the molecular biology and chemistry. to develop therapy to target ras directly we needd the function of the protein in exquisite detail. i'll try to point out the mutant

ras proteins that play a role in human cancer have unique an distinct properties which offer opportunities for therapeutic intervention, that will be one of the themes of my presentation. let's see. so this is the lab at frederick,

which is the ras central hub, as you described it. has anybody in this audience been to this facility yet? nobody at all? it's not that i suggest you -- if not, i suggest you visit. this is my lab at ucsf where i

spend the other half of my life. as you've already heard, ras plays a major role in human cancer, really k-ras is the major player. and the reason we don't really understand because as we heard from susan, k-ras, nras and h-ras are similar proteins,

actually expressed in all tissues, so tumor driven by k-ras also expressed nras and h-ras and so on. there's not that this individual tissue types express selective types of ras, they are expressed everywhere, a bias towards different tumor types in terms

of which ras gets mutated. we don't understand that at all. the k-ras is a major player in human cancer. h-ras is a minor player. playing a small role in terms of numbers, bladder cancer and thyroid cancer. we also heard from susan h-ras

is a target for farnesyltransferase drugs, nras and k-ras are not, there are renewed efforts to target h-ras-driven bladder and thyroid cancer by pre-selection of patients with mutations followed by treatment with farnesyl transferration inhibitors, they

will provide benefit to these patients and teach us effects of targeting k-ras. this is the ras protein, which has been described as the beating heart of signal transduction, because the protein is heart shaped, and also because it changes between

two configurations off state and on state. ras proteins are binary switches turn on and off, according to how they are bound, and one changes in confirmation, where the downstream effectors jump on and bind. we have a good understanding of

the molecular details of ras and its interaction with some of its partners. this shows a diagram of how this binary switch actually works. again on the left side this is another way of looking at the ras structure which binds tightly to ddp or gdp and this

binding determines whether it's on or off. when the biochemistry was first realized, it was realized it was a gdp binding protein people suggested using analogs that would block the mutant protein selectively. but it turns out the ras protein

binds to nucleotides with extremely high affinity, very slow off rate. on top of that, cells have a high concentration of gtp and g.d.p. the idea of finding a drug to compete and get into the ras active site was abandoned many

years, although some people are rethinking it with clever nuances but that doesn't team to be a tractable way to attach the protein, it's already full of nucleotides and doesn't seem to be an easy way to get an analog into that site. anyway, again the activity of

the protein depends on whether you have gtp or gdp, this is how the switch works on the right. so here is gdp, gamma phospate actually holds the part in a rigid configuration. this is where ras and kinase and ralgds bind to the raf. if this comes off, this spring

mechanism opens up and both switch one and switch two become disorganized, and now the proteins don't bind, raf and kinase and ralgds don't bind. you cut the bond, and then it unraveled and no longer can interact with downstream targets.

that's how the molecular switch functions to engage downstream proteins, but things are complicated as i'm sure you're already gathered. downstream are pathways, and even the conversion is a highly controlled process, not sporadically through losing or

binding gdp, the exchange is controlled by large and complicated proteins, only kicking ras. kinases are the best known signal input into the system. actually there are many other signals, other exchange factors to activate in cells.

this is a highly controlled process, it only happens when signals come into the cells. the off state is equally highly regulated by another complicated family of proteins called gaps. there are a number of gaps in the cell, most importantly

things in human cancer, the nf-1 protein, the nf-1 gene is frequently mutated in human cancers. the off switch is defective, so ras can accumulate. another way of getting ras, breaking the off mechanism. the process is extremely highly

regulated because this is the fundamental switch which determines whether cells will divide or not, and many other fundamental biological properties. so oncogenic mutations in ras, codo in 12, 13, 61 and some others, basically lock the ras

protein in the active state making them resistant to the gap. they can bind mutant protein but the switch is broken. trying to fix it, tricking these gap proteins into turning ras gdp into the off state. sounds crazy but that's the most

elegant way of fixing the problem. so when proteins can't be turned off by gap, they get stuck in that state and become less dependent. not totally but way less dependent on upstream signals

and therefore signaling downstream. so the problem is then that the switch is stuck in the on state, signaling, and we have to figure out a way of fixing that, either by fixing it's switch directly or understanding as it turns off downstream signaling.

just a little bit on the actual mechanism of the off switch. so the structure of the ras and the catalytic domain of gap from the neurofibromin shows us how the switch works in rats with wild-type. it's a cool mechanism i must say.

gap proteins bind to@and create an active site for gtp hydrolysis. when gap binds to ras it creates an active site by poking a finger, so this shows the finger from gap, the active site of ras, you see coordination of the 51 residue, frequently mutated

in human cancer, and a phospate of gtp and water molecules. the finger reorganized ras protein to make it into enzyme to hydrolize gtp, lines up 61 and thin attacks the gamma phospate. that's what gap does when it binds to the ras protein.

you could easily see if you take 61, this won't work. those mutations are dead for likewise, if you look at different view of the protein, this is ras in yellow, here is the finger from gap, poking in here, again lighting up all the residues of the active site if

you replace glycine with any other amino acid it distorts the site and prevents the finger from lining up. and that's the problem. these side groups prevent the finger from correctly making the active site and therefore it doesn't work as an off switch.

that's why glycine 12 and 13 and 61 are the key residues in human cancer, they all break the off switch, ras gets stuck in the active gtp-bound state, that's the problem. it causes a million cancer deaths a year caused by k-ras mutations, all the lung cancers

worldwide. that simple defect in the switch is responsible for a tremendous number of cancers. again, in pancreas and lung cancer, ras mutations are the initiating event. we're pretty sure of that. in colorectal cancers they are

definitely not the primary initiating event, that's true also in other cancers. what of our efforts at the frederick national lab, one is to stall the peak structures. when we started this project we were somewhat -- many people were surprised to see these

structures never got solved. they sue the structure in mutants and the field moved on. we don't have structures at all of any oncogenic ras protein. people in the field moved on. these structures were never solved. we don't know whether it's

possible to fix the prone switch because we've never seen it in a complex. that's one of our main goals to solve the structures of mutant proteins bound to gap and ras and see if there are no ways to find drugs to turn off the ras protein, that's a major effort.

again, no structure of k-ras has ever been solved. we also don't have any structures on full length ras protein with the lipid that susan described. a tremendous amount of information is missing if we want to attack this as a proper

drug discovery effort. in hindsight everybody rushed in in the '80s because it was the biggest oncogene, it seemed simple. and then it failed because of the backup system, and people moved on basically to target kinases and never accumulated

all the data you need to make this into a proper drug discovery program, that's what we plan to do at frederick. this is an updated version of the piechart susan showed, in the mutations that cause lung cancer, the unknown pie is getting smaller, as more

mutations are under to. all the mutations found in lung cancer map to the receptor targeting kinase ras pathways, driver mutations. maybe 25% now, 75% of lung, adenocarcinoma, are driven by activation of the rtk ras the newcomer on this pie here is

the nf-1 protein, neurofibronin protein, 8% now of all lung adenocarcinoma. you putting the nf-1 up there. loss of nf-1 leads to hyperactivation of wild-type rates with the same consequence as mutating the ras by mutation, in the active state.

this is a sidebar, i find this fascinating that the nf1 gene is the gene responsible for neurofibromatis type 1, half of cases transmitted through the germline, half are sporadic. it's a pretty common disease, affecting 1 in 3500 people these individuals inherit or

acquire a defective nf1 gene that can't turn ras off. many phenotypes are due to hyperactivation of ras through loss of nf1, leading to benign and malignant tumors, growth defects and other disability. this was cloned in 1990 based on family lineage, which this runs

in the family, found to be related to the class of enzymes we discovered to turn ras off. the level of excitement when this was cloned and identified this paper was submitted to "cell" on the 24th of july, 1990. and it appeared in the august

issue of "cell," it meant all these heterogeneous phenotypes well known to be associated with neurofibromatosis can be put down in theory to hyperactive and then we followed up quickly showing this protein does in fact enter act with ras and negatively regulate it.

this lead to tremendous optimism in the nf1 community. we finding ways of turning ras down as a therapeutic strategy but 25 years later with no approved therapy for any aspect of the nf1 disease, we haven't been able to figure out how to turn ras down and provide any

clinical benefit to patients suffering from this disease. the nfis protein is gigantic, cancer because any mutation which leads to a defective protein actually has a strong phenotype, hyperactivation of so the region of nf1 is this yellow box, that's the structure

i showed you earlier, with the finger there, the rest is unknown. there are no domains here except the membrane domain that shows what it does. there is a paralog here, this must be a protein in something more fundamental than signal

transdeduction. i'm determined to figure out what this protein does but haven't figured out a strategy for doing this. you mutate this and get a strong phenotype. we recently got clues by looking at another disease which is

often -- or was actually associated with nf1, part of that disease. this is an nf1 disease where the gene is wild-type. eric legius in belgium decided there must be another gene with loss of nf1 and he found the gene responsible for these

phenotypes. it turned out to be mutations in a protein called thread 1, loss of nf1. this is a great truth for us, we thought, okay, if we can trig out what that does, that will help us figure out what nf1, we did mass spec on wild-type and

pathogenic mutants and found it binds to nf1. they work together as a complex. these are now all the pathogenic mutations from what is now called legius syndrome and they affect nf1. spredis takes the protein from cytoplasm to

ras in the membrane. without the spred1 protein the nf1 protein doesn't work. we have clues. another sidebar, to remind me, the nf1 disease is a very common disease, but it's also part of a whole group of diseases called

raf-opathy, usually either inherited or sporadic mutations that lead to all kinds of developmental defects through hyperactivation of the ras pathway in every case. remarkably, there are kids out there who have costello syndrome, these mutations are

full blown oncogenic mutations in h-ras, each has a full oncogenic mutation. these are other versions of the same problem. pathway to some degree is compatible with life but it leads to a large number of developmental issues.

one particularly interesting was the cfc is the result of activation of b-raf, some kids have germline mutations in b-raf melanoma but they don't get melanoma. the same mutation in the same gene could be compatible with life on one hand and be a major

driver of malignant melanoma in adults. getting back to the issue of mutations in k-ras in major diseases, this is a number of patients affected by these alleles. the frederick group, we focus on leed proteins.

that is evidence in colorectal cancer, g13d do respond in contrast to g12, analysis suggests there's some benefit in treating those patients. we have no idea why g13 mutations could be responsive to egf but g-12 is not. there's chemistry we don't

understand, but it leads to who should be treated more efficiently. also, these mutations are not equivalent. we thought these were different ways of locking ras into the gdp state. clinically, you can see patients

with mutations at g12z and c have a worse clinical outcome than b12d, and some speculate this activates. we have to think of disdifferently in severity and qualitatively. this doesn't make sense, the old school structural biologists

looked at structures of ras and thought the mutations occurred in the working of a gdp switch, the outside surface won't be any different. these clinical data suggest that in fact there must be differences in the way the mutant proteins engage

downstream proteins. that's another reason we need to get the structures solved. if they are different, it must be a unique opportunity to intervene. we can't explain the differences at all right now. this shows the structure of the

ras protein, the g-domain, the beating heart, almost identical between each of the ras proteins and differences at the c termal region from the unique sequences protest differently for each of the ras isoforms, you might think this is the differences must be, k ras has a different

biology, the effect of all these proteins are identical, they bind to the same proteins, the same affinity. differences must come from the back end. so we've undertaken an analysis of different ras isoforms and different alleles of ras, using

the sort of reductionist system of making cells which are identical in every way except ras protein that drives their growth. we've think the system developed in madrid, he was able to make fibroblasts ras-less by making the k ras locus surrounded by

sites enabling it to be slipped out using recombinase. you can take k-ras out, now cells have no ras at all. and they stop growing and sit there, sitting on the plate for months until they get infected. they are remarkably stable. we can rescue these cells by

introducing h-ras, nras, k-ras in two versions, all using versions of any one of these. at the frederick lab in collaboration with my lab we made a panel. this is enables us to really understand the differences between the proteins in detail

because there's no confusing elements of having multiple different ras proteins. this is the cartoon to show how it all works. we've been able to screen compounds in the pilot study done at the frederick lab, with the isogenic backgrounds,

focusing on compounds that get to the heart of the matter. this just shows you what example of the ongoing -- this shows the ras proteins are different, i was asked not to show too much primary data but this is pretty cool. here we've taken the single

isoform with these mutations, and then simply precipitated proteins, proteins that could bind to ras. if you compare this particular protein that holds ras and mek together, they are loaded up. a simple bichemical level you can see these proteins engage

different complexes, and the ras proteins are different. each of these mutants has different properties. that's really important if you think about trying to treat tumors in which you have k-ras mutations with a drug that affects mek.

some mutants drive signaling through mek, some less, that's an important distinction which has not been made, some of these may respond differently to inhibitors with differing pathways. in any case, you want to understand why the differences

exist. so we talked a little bit -- susan talked in the previous discussion about going downstream of ras and blocking ras and mek. in the old days when a lot of these drug efforts started in 1991 this is what the pathway

looked like. that's probably true in primary mek. this is a major pathway that ras turns on. but as seen from this, in the cancer cell with mutant rats, okay, you lost upstream signaling because ras is stuck

in the gtp state you would have thought blocking with more molecules would have done the same thing since it's linear, or, b, would have shut down ras signaling, okay? and you would have been wrong on both counts because these drugs don't work effectively on ras

humans, as you just heard. if anything, it makes it worse. ras inhibitors actually activate the pathways. so the bottom line was this pathway is ludicrously simple. based on genetic systems, even mammalian systems, a hierarchy of protein cells but not the

details of how it's wired together. for example, if you take five different ras inhibitors, totally different structures or mechanisms of action and put them on a k-ras driven cell line and look for phospho-mek, you can see a dose dependent manner

the ras inhibitors activate ras, exactly raf, the opposite of what we expected. the pathway is not on fire, as we expected. when you add raf inhibitors the pathway comes to life. and this is a drug we developed,

kinase is activated by the raf inhibitors. nobody expected that. and it took several cell papers and nature papers to understand. the pathway looks simple, it's much more complicated than we for those who care, this is what was down at the frederick lab.

this is activated in the test tube, pure kinase is activated by the raf inhibitors. the reason this happens is, raf proteins get recruited as dimers, when it is recruited you get a nice signal of activation of mek, although we don't understand how raf activates

it recruits and activates it to get a nice signal. then especially with mutant proteins, the off switch is broken, these raf proteins talk to each other and turn each other off, so the kinases phosphorylate and deregulate. now you have basically inactive

raf. you come along with the raf inhibitor, the first thing you do is relief phosphorylation, a negative regulator, so you relieve autoinhibition, leading to increase in raf kinase. if you had enough drug you could shut it down completely.

you get a bell-shaped curve. this is why raf inhibitors have not worked for raf cancers because of the bizarre paradox. raf inhibitors have been affected in malignant melanoma where raf is mutated and drives the bus in an independent way, driving melanomas.

we can treat patients with raf inhibitors, very high doses, because in normal tissue we don't turn raf off. if anything we activate it. the therapeutic window is infinite. you can really cover these tumors with raf inhibitor

without worrying about them because of its paradoxical activation they are not inhibited. that gives you a good therapeutic window, essential for benefit in malignant melanoma, you don't get 95% inhibition, you don't work.

you can get a strong inhibition without any damage on normal tissue. that's a pretty unique situation. not true of mek inhibitors. it acts like a proper kinase, inhibited, activity goes down.

unfortunately when you inhibit mek, as shown here, mek inhibitor, you actually activate upstream signaling. mek inhibitors activate the kinase pathway through the egf receptor. that's because raf suppresses upstream signaling, as soon as

you shut down the pathway, any level actually, you get hyperactivation of egf receptors and cells can then activate downstream signaling and then help them survive. so i know this is horribly complicated, simple linear pathway, feedback loops put the

brakes on upstream, leading to suppression of the kinase signaling, you block this pathway with mek inhibitors, this off sets the drug. that's something else we didn't know. you could say if enough mek inhibitors would shut down

everything but it's essential in normal tissue, so mek inhibitors, there's a narrow therapeutic window. you can't shut down its pathways with mek inhibitors, unlike raf where you have the unexpected ability to do so. so going downstream has been

thought. it's not over yet. people are still testing combinations of drugs to prevent the rebound effect and so on, we're chasing our tail a little bit because fundamentally we can't shut down this network effectively without being toxic

to normal tissue. it's back to trying to understand basic biology of ras proteins themselves as best places to intervene. this is one vignette. we have to stop at 5:30. we tested the hypothesis k-ras is particularly frequent in

human cancer relative to h-ras because it has additional properties to make it more malignant, the properties we've discovered are the k-ras makes cells more stemlike, turning on a pathway to may be them drug resistant, able to grow at single cells and metastatsis.

h-ras does not do that. it's a unique property of k-ras. and we've prepared transform cells, everything is identical except the c terminal ligand. the same level of ras and asked are they different in growth in vivo or in 3d and see a huge difference, both on the bottom

line here, k-ras is able to make single cells grow with spheres in a stem cell assay h-ras doesn't do that. h-ras is unable to form tumor efficiently when you implant a small number of tumor cells in a mouse. k-ras is good at it.

you see the tumors here growing massively faster. k-ras turns on drug resistant genes, h-ras doesn't. this is a new aspect that has therapeutic possibilities. we looked at genes regulated, and a member of the il 6 family stands out, to prevent embryonic

stem cells from differentiating. i never heard of it when my postdoc discovered it. i went to wikipedia and found this diagram, it shows it is necessary. k-ras cancer cells pump out this

cytokine. if we knock down li, if we reduce initiation of pancreatic tumors. we think lif is an important element to establishing pancreatic cancers and other cancers and neutralizing it with reduce many aspects of

pancreatic tumors. neutralizing lif makes pancreatic tumor cells sensitive to gemphytabine. another aspect, we think stem is derived from ability to k-ras to bind to calmodulin. binding inhibits cam kinase and prevents downstream pathways

that affect stemness. the point of all this, if we can prevent calmodulin binding to k ras we can reverse the features on human cells. this interaction is regulated by phosphorylation on 181, beater bloomberg at nci has shown, there are a number of ways of

activating phosphorylation of this to prevent calmodulin binding. you can make a mutation and make it look like a phospate, it doesn't bind carmodul irkn anymore. i n anymore.

we can turn k-ras off from stemness point of view by inhibiting calmodulin. and the couple slides at the end. the therapeutic value of this that we're now testing is we've got hold of a compound which is identified as an anti-viral

agent. the last slide. an anti-viral agented identified by the people of samoa, a group of investigators from the nci discovered an active ingredient, a pkc agonist, it's actually heading toward clinical trials as an anti-viral through its

discovery, it activates phosphorylation of c was, preventing pancreatic cancer in mice. here we have an anti-viral which may be useful in treating pancreatic cancer, something we're currently testing in the lab.

bottom line is that although we thought we knew a lot about ras, a lot of details we didn't understand at the protein or signaling level, even the biological level, we're seeing the new insights can be translated into new ways of treating cancers.

thank you very much. [applause] >> what would be your preferred method to target ras? you alluded to half a dozen. what do you think is most feasible and most likely to succeed and have broader

accessibility? >> well, as of today, i would say based on what we know today, i think the result would be the same with -- it's an orally available drug, which is safe and does exactly what we hoped it would do. that's something which is brand

new, not even published yet but that kind of approach it seems will go to clinic in the next few years, it should have real neutralizing lif also looks good to me. longer term understanding structures of ras proteins and complexes has best opportunity

for finding molecules that will kill ras by preventing ras signaling. then there are also the approaches of targeting k-ras, that's another potential opportunity, or enhancing the immune response to k-ras tumor i think we're looking at

probably seven or eight different options, all pretty early for the most part anyway, but we're betting on four or five at the frederick national lab, we'll assess the situation in a couple years and see which is looking most promising and also hope other people in the

biotech and pharma world will work with us on picking up different approaches. >> i saw that you suggested different mutations in ras will select for certain downstream effectors, might be be occurring in tumors like oncogene affected by the k-ras as opposed to genes

that are no longer relying on it? >> you're raising an interesting question, which tumors are addicted to k-ras, so far in pancreatic, experiments have been done where it's sticking out, i would argue that at least the k-ras and lung cancer, they

remain addicted to k-ras really. cell lines maybe not. in fact, there are cell lines i know even pancreas cell lines that don't care about k-ras in cell culture but put them in 3d or mice and they don't grow at all because they are addicted. it's a very interesting question

whether or not addiction will be different with different alleles, that hasn't been done. they have only been done on two or three alleles, addiction experiments. we don't know yet. it's a great question. >> i was wondering, the

phosphorylation of ras could be targeted because phosphorylation at some point -- [low audio] is there any evidence of that? >> as i mentioned, debbie morrison at nci frederick is planning on doing screens for compounds that prevent ras

dimerization, essential for activation, the screens are not difficult to do technically. that's a very interesting approach. that is a method underway. >> i wonder whether in regenerating tissue or rapidly

turning over tissue to bone marrow, the liver after partial removal, is there a phase where ras, any ras, is expressed at a time particularly early on when in like the regenerating liver or the primordial cells in the bone marrow, when they show drug resistance and stemness-like

characteristics, is there a relationship to ras presence or >> it's easy to say yes because i'm sure that's true, but we don't have evidence to support it. the ras proteins are expressed in all tissue. the question of the different

forms active in those states, whether that activity translates into effect on stemness, i just can't tell you at this moment. >> but i mean are they overexpressed, underexpressed, in relationship to this kind of rapid proliferation? >> i don't think so, no.

i think in most tissues ras proteins in the off stage are gdp pound, the capacity to be affected by signaling, ras proteins are active by nucleotide binding, not increased expression for the most part, difficult to measure than a couple levels of protein,

i think it's a harder question to answer because they've got to measure ras activity. in mice, people are now doing, different tissues depend on different ras by genetic ablation. that's one way to get at your question.

>> great question. yeah, nf1 certainly, about 10% or so of patients with nf1 succumb to malignant tumors, it's not 100 by any means. there is it a higher frequency of rhabdo sarcoma, and one other tumor type i've forgotten what it is, but the braf mutants

don't. it's bizarre. newton's syndrome is caused by activating mutations, those have a higher frequency of you've mile leukemia. generally yes with odd exceptions. >> something that would be able

to block ras, what is the rationale to combine different drugs -- >> well, it all makes sense in the point of ras cancer, but the problem is where is the that is the problem. i think a large number of trials have been tested with egfr

inhibitors, mek, every combination thereof, and we heard from susan there are some signs of some activity but you can't dose enough to shut down the whole network without affecting normal tissues. you can in mice, targeting to go forward, but toxicity of

combinations have been a neil rosen and others that have theories as to how to get around that by different dosing schedules to get around the rebound effect but i don't think those combinations are going to get us where we want to be because we're working against

the therapeutic gradient. ras tumors downregulate everything. >> related to that, the old days, there was work many years ago, schlessinger, showing some ras mutations -- >> yeah. >> how do you relate to that?

>> that's a very good point. we don't have time to go into this in detail but it turns out the mutations do have intrinsic activity, in the rats with mek, they respond to egf, and they respond by gtp loading from relatively high baseline, 20% to 100%.

they depend on upstream glutinin 61 are gdp bound but the mice have a window of activation, so my feeling is there are only partially active and need help upstream they depend on egf less than normal cells do, they still depend on it but it's a

therapeutic window, going the wrong direction, so it has to be more specific than just that. great question. >> a strategy several groups look the at is synthetic screens to look at pathways that would kill cells that have a mutant i guess they have initial

comments but it faded away, have they been revisited, other solutions? >> we're revisiting. we put out an rfa based on the discussion at the nci, the frederick national lab, to get people to come back. the first round really didn't

work. all kinds of issues, partly sirna, cell types with enzymes, idiosyncratic, people have reapplied to take that to the next level, reagents are much cleaner and we learned a lot from the first wave, now let's try going to the second wave

using better libraries, crispr instead of sirna, targeting cells in 3d instead of 2d, the next 2.0 version will soon be tested hopefully. the idea of finding an unbiased way genes essential to k-ras survival is a great idea. we just haven't found it yet.

well, i want to thank you both for a very exciting