cancer de colon es curable

[ silence ] >> so jumping right in, our first speakertoday is kevin shannon, one of the organizers of the symposium today. and he is recipientof many awards and recognition which i won't go through. but i'll point that he is notmade a lot of significant contributions to the ras field but he's but he's also servedas scientific advisors to many illustrious bodies including the nci, american societyof hematology, many famous programs, and he served as the msp director until very recentlyas well. so it's nice to see that when you do good works, sometimes you do get widelyrecognized for it. i also want to point out that kevin was my own research mentor andfellowship. i did my postdoc with him. and

so besides his many outstanding qualitiesas scientist, he's also a phenomenal role model as a physician scientist and as a clinician,although he's not doing that anymore. he's really a wonderful pediatric oncologist tohave trained with as well as a hematologist. and the one thing i did want to say aboutworking in his lab is that, you know, i'd had other labs that i could have joined aswell that were, you know, maybe even more illustrious if consider nobel prize, somethingthat is worthwhile. and--but of all the people i talked in my residency programs, you know,there were a lot of medical schools or residency interviews. and in the field of medicine ingeneral, you know, interviews tend to be fairly stayed and boring affairs, everybody is veryserious. but interview with residency with

kevin and--you know, with kevin that was theonly residency interview was ways [phonetic] with profanity for 45 minutes. and i haveto say that put me so at ease i knew at once that this was in fact the place for me tobe. and that initial impression has born out to be true. it's been a wonderful to havetrained with kevin, to continue my collaboration and on going life long training with him.so without no further ado, i'd like to ask kevin to come up and tell us about his studieswith ras signaling in leukemogenesis. >> thanks ben. so i just wanted to do a coupleof things at the start. there's a bit of an illusion in this--at this meeting that frankand i and martin were really the co-organizers. and i would say that i think frank and i wouldagree that if frank and i did 10 percent,

martin did 90 percent of the heavy lifting.and so he really--as expected. so we want to really acknowledge martin as the sort ofmajor force behind this symposium which has been really terrific. i also want to thankpeople from--as martin has, from coming far away from the east coast and even through--evenfurther away many time zones, julian, herbert and allen. i think as californians going toeurope, we have this feeling that we're not quite sure what day it is, what time it is,what month it is and i'm sure that you guys are all experiencing that today and we reallyappreciate you coming. so i'm going to talk today--i was so inspired by yesterday's talkthat i trash the first talk that i had by that noon time and then the second one lastnight. so this may be a little bit rough and

ready, but trying to pick up on some of thethings we heard now. as ben told you, i'm a pediatrician by training. so i found rasas a sort of second calling i supposed or kind of wandered into the field. and i wanderedinto the field because of a rare pediatric disease called juvenile myelomonocytic leukemia.and this is a mild proliferative neoplasm, a bit like cml or cmml. these patients have--areyoung and they have very white--high white blood cell counts, the cells look funny sothey're dysplastic. they have high levels of hemoglobin f. they have little plateletcounts and they get massive organ infiltration with these abnormal cells. this is one ofthe pediatric leukemias with a very poor prognosis. it's really only curable with bone marrowtransplant and only done about half the time.

there's a peculiar hypersensitivity to thegrowth factor gm-csf that was noted in cell biologic essays. but when i started to workon this over 20 years ago, what drew me in was--my postdoc mentor at that time y.w. kanwas happy to have me come and work on leukemia in his lab, which he'd never worked on before,but he said, "you need to have some sort of genetic connection" because he's basically,as many of you know, a geneticist. and so i stumbled on jmml because of the dissociationwith neurofibromatosis. and of course, as a young physician trying to learn how to doscience, doing southern blotting and extracting dna and things, but the network that i hadthrough the children's oncology group allowed me to collect samples from patients. and sortof the big break for in my career i supposed

was when whitton collins [assumed spelling]independently identified nf-1 and run it through those primitive protein databases that existed20 years ago. and asked does this protein look like any other protein? and the computerthought about for probably four or five days. something your mac would now in 30 secondsand said, and yes by the way, it looks like these other proteins, these ira proteins inyeast and this protein that frank and others have identified called gap. and so, that thenlet us to think, well, a dominant familial inherited cancer predisposition, a gene thatlooks like a negative regulator of ras. data from frank's lab and julian's and daglowy[assumed spelling] showing that in fact it was a proper negative regulator of ras. wethought, well, maybe this is a tumor suppressor.

and this is basically the genetic proof ofthat. this is from a family--here is mom resolving her two [inaudible] this 403 and 395, dadhas homozygotes 399. here's the 399 from dad, the 395 from mom. we know this is the disease[inaudible]. and then all of the cells from the patient, the bone marrow cells that leukemiacells as was of the normal allele as you would expect following the classic paradigm. andi point this out because exactly 20 years ago is when this was published, and it waspublished at the time when i had a job at ucsf in the laboratory space. and so, thefirst six months, i was a faculty member at ucsf. i actually spend at a place called onyxpharmaceuticals. and frank sort of inspired me to follow this along and provided somethe advice and reagents and also thought me

how to interview people and what words tosay when you're interviewing somebody. so, that i think ben later encountered. and thisis what i look like back then. [ laughter & inaudible remarks ] all right. [ laughter ] ok. >> i'd fall in love with that. >> so. and then, you know, we had this connectionwith this growth factor gm-csf which was known from work [inaudible] to signal through thesignal relay protein as an adapters that julian

introduced us to activate ras neurofibromaand this is turned off and some biochemical experiments that i did in onyx with the youngbiochemist there named ginnie boag [assumed spelling]. we actually showed that in thesejmml cells, you actually had abnormal activation of them [inaudible] pathway and daphne andjulian i'm sure will remember those 5 to 10 millicurie orthophosphate labeling experiments.and that's where my hair is a color it is today. and it turned out that that observationabout jmml with nf-1 sort of told us what this disease is all about. and it's probablythe purest example of the human cancer driven exclusively by hyperactive ras. so, afternf-1, we went on and found mutations in kras and nras and about 25, 30 percent of thesepatients. later bruce gelb in our lab [inaudible]

particular discovered ship-ii, dominant oncogenicmutations and the signal really preyed to protein ship-ii, and most recently, miniancybil [assumed spelling]. and about 90 percent of patients where jmml will have a mutationin one of these genes. and jmml genomes are very, very stable. they only have--they haveless than one other genetic abnormality. so, this is really a pretty pure example of adisease driven by hyperactive ras. so, over the years, we've collaborative and luis [assumedspelling] and with tyler and david tuveson and kevin higas [assumed spelling] to generatemodels. and then sort of satisfying result when you either inactivate nf-1 or turn onkras or nras and hemapoietic compartment you get a disease that looks very much like jmmlor cmml the adult equivalent. what's also

interesting is that these three differentlesions in the contact, in the same strain background where the ras alleles are expressedfrom indigenous promoters can be quite different phenotypes. so, oops, i'm sorry. the krasis by far the most aggressive, the nf-1 is intermediate, and nras is quite indolent.and this is of course interesting since nras is really the ras gene that is mutated about75 percent of the time in different blood cancer. so, it's sort of the weakest alleleis the one that seems to cooperate to--across cancer in the hematopoietic compartment. unlikemany others, we've explored the downstream factors as targets again following up on julian'stalk yesterday and use some of the same compounds gdc-0941 from genentech and mk-2206 whichis a mer [phonetic] compound that we had to

synthesize our cells, and then ci-1040 andpd-901 and more recently trametinib, the mek inhibitors. so in the--in the nf model themice develop this myeloproliferative neoplasm. it's fatal, but it takes a long time for themice to die 5 to 10 days. and it becomes sick at about six months and will live out to yearabout with--most of them die between six months in a year. and there's this progressive increasein the hematopoietic compartment elevated white counts more and more organomegaly. andit's easiest to measure by just following the spleen which are heavily infiltrated withthese abnormal infiltrating hematopoietic cells. and so, we have done a number of preclinicaltrials. this is one done by tiffany chang who's in the audience in which we just basicallyrandomly assigned wild type or nf-1 mutant

mice to either treatment with pd-901, themek inhibitor or the vehicle. we enrolled these mice at six months of age about thetime they were getting sick with high white counts. they got 5 milligrams per kilogramand we treated them for 10 weeks. and then just, you know, it's relatively straight forwardin hematopoietic system, we followed blood cell counts and then spleen weights at thetime of sacrifice. and the mek inhibition works very well in this system. what we seeis a very rapid and sustained decrease in the white blood cell count. restoration ofa more normal hemoglobin although it's really hard to see since the nf syndrome doesn'treally cause severe anemia. but we do see is a lot of ineffective erythropoiesis. andthis is characterized by the hemato--by the

erythroid compartment working very hard producinga lot of new red blood cells which we called reticulocytes in the hematology world. andwhen we start treating with pd-901 what we see is that those elevated abnormal reticulocytescounts returned to normal and the spleen size is diminished. what's interesting about thisis that these all occurs in the context of a 100 percent mutant hematopoietic system.that is these nf-1 mutant cells which are 100 percent mutant when we start treatmentare a 100 percent mutant in these mice that have had this traumatic response at the endof treatment. so what we've done is essentially alleviated to symptoms that these animalshave but have done nothing to reverse the underlying molecular defect in the stem cellcompartment of these persistence or proliferative

advantage of these nf mutant cells. and infact a therapeutic benefit of this in mice is probably visualized best here. and thisis sort of things that hematologist and immunologist talk about that put everybody else to sleepbut just very briefly. red cells go through this orderly maturation where they start expressinga transparent receptor at high levels when they're proliferative. and as they begin todifferentiate they lose transparent receptor expression and acquire this marker throughter-119. and you can put them into boxes basically. and this was sort of a wild type looks likewith a lot of cells over here. in the nf-1 mutant what you see is that the erythroidcells get stuck. and this is why they get trapped in the bone marrow. the blood cellbrought--red blood cell production is abnormal.

and then there are these high reticulocytescounts to try to compensate. when you treat with the mek inhibitor now these cells driveright through and differentiate and that's the--the--essentially promoting differentiation.so, just to summarize where we are with this, the treatment reduces leukocytosis, the highwhite counts. it enhances a much more normal pattern of what red blood cell productionand maturation. and it enhances survival in both the nf1 mutant mice and ben showed previouslyin a kras mutant mice with this jmml like mpm. what we might call an early stage canceror an adenoma if you will. the beneficial effects are only observed with the long actinginhibitor. i haven't shown you this but with ci-1040 we saw absolutely no effect. we reallydo need to--could sustain inhibition of the

target. when you get that sustained mek inhibitionit rebalances hematopoiesis, but doesn't eliminate the mutant cells. the therapeutic effectsare most pronounced when we look where it's occurring. it's not at the stem cell levelbut at the next step up these transient amplifying progenitors. and actually there's a humantrial that minian and tiffany are going to be opening here at ucsf and [inaudible] aroundthe country using the trametinib in 2014, so later this year. so what i wanted to nowturn to is talk about using the system to build more--more models of multi-step cancerand to try to look at drug response and resistance in that setting. and the way we decided todo this was to use a sort of inspired by--by neal copeland and nancy jenkins was to tryto just randomly generate retroviral integrations.

and i must tell you that we first did thisas a discovery sort of experiment. but this kind of technology for discovery, as manyof you know, is sort of now become i think relatively antiquated because we can sequencehuman tumors at great depth. and so we can--we can really ask what the players are in humancancer at a much more precise level. nevertheless we can generate using these retroviruses micethat--or leukemias in mice that have multiple different cooperating events. we can detectby southern blotting or by cloning integrations. and when we've done this experiment threetimes, we've done it in the nf-1 background, jennifer lauchle who's now genentech sortof pioneered this work in the lab. and she got mostly aml out with some t-lineage acutelymphoblastic leukemia. chingly [assumed spelling]

and malic dell [assumed spelling] has donea kras model. and this leads to mostly t-lineage acute lymphoblastic leukemia and some aml.and this data actually sort of predated the discovery of ras mutations in high risk t-lineageacute lymphoblastic leukemia by jim downing. so it's kind of a satisfying result to seetheir paper after having seen this in these kras mice. and in the nras we get a reallylovely aml which very accurately modeled what we see in human aml with nras mutations. sowe begin to think about using these models in a slightly different way and that is touse them to actually study treatment in vivo. and so one of the advantages of this systemis that these primary cancers model the genetic heterogeneity that we see in human cancers.and this is in two different ways both within

individual tumors where you probably havemultiple different clones competing in any given mouse to grow out with one winning,and also across leukemia. so if you generate 20 or 30 they're going to have different retroviralintegrations. in the same way is in pancreas cancer they all have kras but they may havefour, five, or six, or eight other cooperating mutations that are different from tumor totumor. we can use the insertion as sequence tags to track the clonal evolution we canalso clone the integration sites. and when we see things that are revolving sometimesthese new integrations can give us candidate drug resistant genes. and we can keep doingthis over and over again because of the kinds of system this is. that is we can--we cantranspond [phonetic] a given leukemia in say

2012 and treat it with the pi-3 kinase inhibitor.2013, we treat the same leukemia with pi-3 kinase inhibitor and mek inhibitor et cetera,et cetera, et cetera. so, i'm going to tell you about studies that we've done in the krassystem today. and the first sets of studies are in the t-all model which monic dale [assumedspelling] has really led all these work. and in the kras system we inject the mice withretroviruses. initially, we then turn on the kras oncogene. these mice will die from thismyeloproliferative disease, jmml disease. and we can rescue the premalignant clonesby transplanting into a sub-lethally irradiated mice. these leukemias acquire notch mutationas sort of [inaudible], and then they died from t-lineage leukemia. we also see frequentnotch mutations in the wild type mice. and

so this is basically how we do this. so, asi've told you we take mice and we inject them with the [inaudible] 4070 virus and they getleukemia. we can vary the genotype. and so in the situation of nf-1 we know this is aninitiating event so we often will turn on in--or inactivate nf-1 and give the virusat the same time. so the integrations are being selected for in the context of the initiatingmutation. in the kras and nras strains what we've done is given the virus waited threeweeks. and these pups their hematopoietic system will expand by 90 percent. set theretroviral integrations. then use mx1-cre to turn on either kras or nras as a secondarycooperating event. we then collect these leukemias, we store them in trial vials, shown here.and then what we can do is we can take these

leukemias out and transplant them into recipients.and we transplant them into the recipients we can then assign these different recipientstransplanted with the same leukemia to receive either a vehicle or a drug. now there aresome really advantages to the system which i think are pretty obvious. number one, they'reprimary cells. these are primary cancers that have never seen plastic. number two is thesemice are essentially identical twins. they're congenic. so they metabolize drugs exactlythe same way. and number three, they're transplanted with exactly the same cancer. and so you basicallycan do these kinds of binary, does a drug work or doesn't work kinds of experimentsthat we scientist like to think about that are really hard to do in the clinic. and thenwe can see if the mice respond. and if they

respond, do they relapse, and if they relapsedoes the relapse clonal show resistance, and what's the conal dynamics of the resistanceprocess and then what's the mechanism underlying the resistance? so in ti while we have someclues from--from human patients where to go, and that is that there really are two sortof interesting driver mutations that occur in t-all high frequency. one is ras pi3-kinase.so, about 10 to 15 percent will have ras gene mutations a little bit of nf-1, but a fairamount of pi3-kinase p10 and akt. so, pointing us down this side of the pathway is beingimportant, this ras effector pathway and very few. in fact no that i'm aware of raf, mek,or erk mutations in t-lineage leukemia patients. and then as i told you notch is the most commongene mutated in this disease. and between

the notch and few7 [phonetic] it's probably65 or 70 percent. and ras pi3-kinase and notch mutation is frequently coexist. and one ofthe important downstream targets it's been implicated in the system in t-lineage leukemiagrowth is met. and there's this--there is one connection between this pathway whichis that activated notch is known during development to activated a protein called hes [phonetic]which negatively regulates p10 and up regulates pi3-kinase signaling. so monic, this is twoand a half years of work or so by monic summarized in one slide essentially. this is 21 differentindependent leukemias transplanted into over 300 mice and then treated for out probablyin 6 to 10 weeks. so, what you see here first of all is that the wild type leukemias dorespond to the pi3-kinase inhibitor. and this

is gdc-0941 that we obtained from [inaudible]at genentech. and that the mek inhibitor helps a little bit. so this statistically betterthan this and this is statistic better than that. the kras leukemia show very differentpattern and that is an absolutely defeat gdc-0941 by itself. but if we combine it with the mekinhibitor we see very nice prolongation and survival. and then of course the key questionis, "are these leukemias that are coming out here after 40 or 45 days are they differentfrom the leukemias that we initially injected?" and the easiest way to ask that question isjust reinject them and retreat them. and when we do that--and i'll just show you this onepanel here with this leukemia. what you see is that here's the initial parental treatedwith vehicle and then treated with drug. if

we now take the resisting leukemia that'scoming from over here, reinject it and retreat it we now completely lose the treatment effects.so these leukemias and everyone that we've tried that has come out after a significantextension and survival is in fact has a phenotypic drug resistance. they've now developed intrinsicdrug resistance. and this is associated in many of these leukemias with clonal progressionor clonal evolution at the level of southern blotting. and as you can see some of thesenew southern integrations in the resisting leukemia is marked here. make a couple ofpoints about this, one is that what you see is that there is evidence that these leukemiasare all arising from the same ancestor clone that has the same in this case four retroviralintegrations. and that in some cases what

we see is that--so in this r1 clone we seeit only came up once. but this r2 has come up in two different independent mice. whenwe see clones that are coming up in multiple independent mice, that tells you that theclone that is driving the relapse had to be there when you first injected the leukemia.so, it's pre-existing clones that are causing the relapse. and again this has been alsodescribed for adult and now an adult all by tim lay's [assumed spelling] group, washu[assumed spelling], and by mo [assumed spelling] and downing. so we think what we're seeingin this system is very much reminisce of what we get in human leukemia and human cancerwhen it relapses that is sort of darwinian selection for these--for these pre-existingdrug resistant or drug less sensitive clones.

now we can clone the integration sites thatcome up selectively in the resisting leukemias and we can ask what pathways [inaudible].and what struck us initially was that notch was coming up and of course as the pi youalways immediately say, "well, here's--here's a perfect example of one oncogene driver pathwaytaking over for the other." so you got these notch mutations and you've got pi3-kinaseras mutations. you put a pi3-kinase inhibitors on. the tumor gets smart and they amp up notchsignaling to sort of overcome the pi3-kinase inhibitor. so i was quite certain that wasgoing to be true. i was quite wrong. what you see here is biochemistry showing thatin this resisting leukemia that derive from this parental leukemia the activating cleavenotch is gone in the resistant leukemia and

mek is also no longer expressed at elevatedlevels. and the similarly in this other leukemia intent--independent leukemia very high expressionlevels of activated notch and mek. in the parental leukemia, now the resisting clonehas said, "no thanks, we'll turn that--this oncogene and protein and this one off as well."we also encountered a number of leukemias where they've actually lost their notch mutations.that is to say they started in the dominant clone with a notch mutation. and in six independentleukemias we could no longer detect notch mutation in the relapse clone. again, thisgoing along with the laws of the activated notch which comes from the mutation and alsofrom the lost of mek. so the two different genetic mechanisms at least they are contributingone retroviral integrations that are turning

off the notch pathway that's been activatedpreviously. and [inaudible] also evolution of clones that no longer have the notch mutation.and as we--as we--then we just went back and tested everything by western blotting it turnedout that 60 percent of these leukemias actually were losing the activated notch expression.and when they lost activated notch they were down regulating mek. so, what i've told youthen is that these drug resistant leukemias are down regulating what we know is a veryimportant dominant driver oncogenic pathway by multiple different mechanisms. and so webegan to think about why this might happen and why a cancer might want to get rid ofan oncogene mutation, it drove it to become a dominant clone in the first place. and thefirst clue was--what we saw was that in these

resisting leukemias there is consistentlymuch higher basal activation of pi3-kinase signaling. and one might expect this, sincethese are resistant to pi3-kinase inhibitors, but what they really done is broken throughand activated particularly phospho-akt. and when we take these leukemias now and theseare just primary leukemia cells and put them in vitro into culture and ask what does ittake to shut down the pi3-kinase pathway? here's the parental leukemia here and youcan see in this resisting clone takes maybe tenfold more and this resisting clone it takesmaybe a 100-fold more gdc-0941 to shut down the pathway. there's--the pathway still sensitivebut now orders of magnitude. and similarly down here, this is a particularly resistantpair jw-81 quite sensitive in two different

genetically distinct resisting clones withthese that are very resistant to inhibition by the inhibitor. so this got us to think,well, maybe there's some sort of crosstalk between the notch and pi3-kinase pathwaysin this leukemia such that notch is normally sending some sort of a negative growth signalto the pi3-kinase pathway. but for most growth enough that the leukemias role will acceptthis as a fair trade. but now when we impose clonal selection, now this activated notchbecomes much more deleterious. and so we wanted to test that and ask the question of doesnotch negatively regulate pi3-kinase signaling. and so, what we did is we got a number ofdifferent human and mice cell lines that were p-10 null. so we could sort of take p10 outof the equation. and we had this particular

cell line which is an important control doesnot express cleave notch. and you can see there's no effect as we add compound e whichis a gamma-secretase inhibitor that turns off chemically the activated notch. but inall of the other cell lines where there's high level expression of activated notch,the nicd, we treat with this gamma-secretase inhibitor. the notch expression is suppressedand as we suppressed notch expression up comes phospho-akt. and using then either nicd itself,enforce expression of nicd itself, we have a dominant negative mastermind, what we seeis we now can change the--does response curve of [inaudible] t-lineage leukemia cells topi3-kinase inhibition by basically manipulating levels of notch expression with higher levelsof notch expression rendering these cells

more sensitive. and when we turn off notchthey become more resistant. so giving us a sort of rationale for why these leukemiaswould now give up the oncogene and we don't think they're really giving it up. what wethink we're doing is driving selection for initiated clones that are preleukemic or maybegrow as well under basal conditions. but then when we impose the selection of the pi3-kinaseinhibitor they are the ones that survive and grow out. so one question we've been sortinterested in lately is can the system readout drugs that we know can be use to treat inthe clinic. drugs that are fda approved and have been used. and so this is--i thoughtthis was a good idea the first i thought of it is. the experiment was going i thoughtmaybe this wasn't such a good idea. because

this is one of those experiments where onlythree things--where three things can happen and only one of them is good, right. so onething it can happen that would--one thing is that you can cure these leukemias witha standard drug. that wouldn't be good because it would say that these leukemias don't reallymodel real patients because we can't cure standard multi-step cancer with single drugs.the other thing is that these leukemias wouldn't respond at all to the standard drug. and thenthey would be just some useless mouse model that doesn't really reflect what happens inpatients. so what you really want to see is that they actually do despond to the standarddrug at some level and then as a single agent where relapse as they do to the targeted inhibitors.and [inaudible] one was done these experiments

so these are monic's data for wild type leukemiashowing gdc by itself and then gdc plus 901. and when she then takes a drug that we oncologistare very familiar with dexamethasone which we've use to treat leukemia and lymphoid malignancyfor 50 years, this is what she sees. the single best drug that we've try in the system isin fact the corticosteroid dexamethasone. it's better than a pi3-kinase inhibitor. it'sbetter than a pi3-kinase inhibitor plus a mek inhibitor. it's the best we have and itworks even better when we combine it with the pi3-kinase inhibitor. and when these micebecome resistant, we do pullout drug resistant clones and they have different retroviralintegrations than the drug resistant clones that we're pulling out of the pi3-kinase plusmek or pi3-kinase alone. so indicating sort

of the clonal heterogeneity in the systemand how the different treatment modulates that and we're just now figuring out whatthese different new integration are and hoping it collaborate with kiki yamamoto [assumedspelling] to try to sort out what's been a big mystery in clinical oncology for 50 yearswhich is how these cells really become resistant to glucocorticoids. so the other side of thisscreen was an aml patient, some of the--some of mice develop aml about 10, 15 percent.so i've told you about the t-all side. but the other side is that some of the mice developmyeloid leukemia. and we've been interested in those as well. this is treatment with pd-901in aml and in contrast the t-lineage leukemia the amls are completely refractory to gdc-0941and when we added it to pd-901 it just add

toxicity. so there's a very different preferencefor which rats affect your pathway seems to be important depending on whether these aremyeloid or lymphoid leukemias. in the myeloid leukemias we do get a nice extension of survivalwith pd-901 and this is all driven by a subset of about 10 to 20 percent of these leukemiasthat respond extremely well. and this is one of those leukemias again mike burgess [assumespelling] has gone back and here's the initial leukemia injected and treated with a big advantageand survival. here's the resisting leukemia retransplanted now showing phenotypic resistancein this new retroviral integration here. this new retroviral integration was a perfect absolutelyslam dunk. this has to be the drug resistant chain. it inserted into a gene called gng12which a orphan gpcr. so we thought driving

this gpcr it's confirming resistance. mikespend about six months of his life trying to prove that and it was a complete bust.gng12 does not change the drug sensitivity of these leukemias in any way. so in desperationwe then resorted to genomics. and we in collaboration with berry teller [assume spelling] we dida whole exom [phonetic] sequencing but what really turned out to be very helpful is thatwe notice on the resistant leukemia that the resistant leukemia has an extra copy of mousechromosome six and all of the genes on mouse chromosome six that we're able to measureare expressed at about 1.5 higher levels, and krass on mouse chromosome six as thisgng12 insertion [phonetic]. so mike knocked down kras in these primary leukemia cellsnot an easy feat. transplanted them into mice

and was looking for hairpins that would knockeddown by about 50 percent and showed that in fact when you just reduce the amount of krasexpression you could very dramatically increase survival in these mice. so we thought, well,ok, this extra copy of chromosome six is likely to be a duplication of oncogenic kras andthat the leukemia now has more ras signaling, it is more aggressive and that's why it'sresistant. so, that was a good hypothesis like many in our lab. it was wrong. the krasaml-101, this is a sensitive leukemia. what you see here is in fact that leukemia doesn'tshow any wild type kras at all. it's under--it has only mutant transcripts whereas the resistantleukemia retains a wild type signal. and so mike--you know, being derick cady [assumedspelling] was a phd student with charles stowyers

[assumed spelling] who probably beat him mercilesslywent in sequence about a hundred different independent clones from these different leukemias.and what you see is in the kras 101, 82 out 83 clones were mutant. and in the resistingleukemia about a third as you would expect based on the three copies of chromosome six,about a third of wild-type and about two-thirds of mutant. so, the three copies are reallythere. ben just gave me the moment--ask me to speed up so i will. and this is--i'm sorrymichelle obav [assumed spelling] showed us that they're actually are three copies ofthree kras chromosome homologous in the resistant leukemia and also did 100 metaphases and thiswas really helpful because when you see now it's in the sensitive leukemia is a littlebit of the resisting clone and when you imply

selection you now get outgrowth of this drugresisting clone. again going quickly now it turns out there's many uniparental disomyevent in this leukemia. so this is the leukemia that is sensitive, the one with the two copiesof oncogenic kras was also wild type, 100 percent of the chromosome six material frommost of the chromosome is from this mutant kras chromosome which in the systems sincewhen [inaudible] 129 and then what we see is 2 to 1 ratio in the resistant leukemia.so, one of the things that we can do in the system is we can test a conventional wisdomthat the drug resistant clones evolve independently with the resistant emerging later and wouldhave been dominant but to just enough time to catch up. the alternative idea though thatwe really interested in is that the sensitive

clone may evolve from the resisting cloneand--but actively suppress it in the absence of drug treatment. so, what we find is whenwe put in the resisting clone it grows much faster on its own and induces high white countsand kills the mice sooner. but importantly when we injected with the resisting clone--withthe drug sensitive clone we see something entirely different. so here we just labeled--thisis the 101 clone. so this is the drug sensitive initial dominant clone. and if you label themwith two different fluorescent markers and follow the mice the proportion of green versusred cells stays about the same. if you now take this leukemia but inject it with theresisting leukemia with the one that's grown out, what you see is that in fact this leukemianow suppresses the resisting clone in vivo.

so the resisting leukemia starts out as beinga much higher proportion and is remarkably suppressed in vivo by the sensitive leukemia.so, the idea here is that leukemia that is sensitive is actually emerged from the leukemiathat's resistant because it hasn't inherit intrinsic growth advantage in the absenceof drug. and it's evolve that way by losing the wild type copy of kras and retaining twomutant copies of kras. and so this is got us thinking a bit about this whole idea ofhyperactive ras and advanced cancers. and i'd intentionally don't show this is happeningin any particular order but you have this ras pathway mutation you have cooperatingmutations [inaudible] and at that i think will allow for increased to ras output overwhat the basal would be in--say in nf mutant

cell or in a cell that just had a kras andnras mutation. and in these advanced cancers as a result of that, they're more addictedto ras etp and a more sensitive to drug. and i think this elevated ras output is likelyalthough we don't have formal proof contributing in an important way to the degree of addiction.resistance then is actually due to clonal heterogeneity in this leukemias where younow begin to select for clones that are less addicted. and one of the things we're wonderingand it's hard to test, but we're thinking about ways of doing it is whether in factsome of these resisting clones actually are characterized by basally less ras output whichallows them to duck under the radar when you treat them with target inhibitors. so, i'lljust stop there. i think i've acknowledge

everybody along the way. i particularly wantedto just acknowledge again monic dell [assumed spelling] for the work in the t-all resistancesystem and mike burgess for his--for the work of the kras [inaudible]. so i'll stop thereand thanks. [ applause ] >> time for few questions. >> question kevin. what's the mechanism [inaudible]means? >> what's the mechanism of the >> nic an nic, n-i-c >> oh, nicd.

[ inaudible remark ] and so we've looked--it's a good question.they look for reverse in mutations and we've not--so what we think it is and we can saythis for certain when we get the same resistant clone out of multiple different recipientsthat it has to be a pre-existing clone that never we think had the--had the notch mutationand expressed nicd. so, we think basically that in the system that notch is sort of conditionaloncogene. it basically it's good and cell will tolerate it and use it as a growth advantageif there's plenty of ras pi3-kinase signaling. when the cell is threatened it will do anythingit can to restore the ras pi3-kinase part of the access and even give up a mutationthat is otherwise growth promoting.

>> so, kevin in the clones the mouse thathave lost the nicd expression? >> right >> if you re inject that population cellsback into a mouse that is no longer treated with drug, do you end up with the nicd expressingclones growing out from that population? >> yeah, we haven't done that experiment it'sa really good idea because i think your point and it's probably correct is that we probablywe haven't completely eradicated the initial drug sensitive clones. it's probably some--it'sprobably been markedly suppressed than if we were to reinject. we actually have to dosome reinjections for some of those leukemias so we'll take a look at that for sure. it'sa good question.

>> so, when the clones lose the nicd expressionso, you know, notch is obviously involved in cell fate determination, do you see anydifferences and like differentiation state of those two [inaudible]? >> yeah. that's a really good question. thisis the notch boys, you know, john aster [assumed spelling] and warren pare [assumed spelling]they were all over us about that. i mean, every--it must had a hundred different--theyask me the same question a hundred different ways, you know. it's like parents asking theirteenagers are your having sex. they ask it in a hundred different ways, you know. andwe did as--everything we could possibly do. we look at these cells. we did-- cells arefor [inaudible] marker panels as near as we

can tell, you know, the cd4, cd8 expression.they've most--a lot of their global gene profiling, a lot of their morphology they look exactlythe same. it doesn't look like they're stuck. and one of the things we have thought aboutwith this system though is that although not mutations are oncogenic drivers in leukemiaparticularly in t-all, as many of you know, their loss of function in cell tumors. andone idea that we have which we haven't tested because we're not solid tumor people is thatone effective--why--maybe the drive to lose notch in some solid tumor is in fact to activatepi3-kinase. if it's hooked up the same way, maybe a way that solid tumors actually amplifya pi3-kinase signal by losing notch. yeah, in the head and neck--head and neck,yeah, yeah, yeah.

>> --tune up or down their kras signaling,i wonder going back this discussion about-- --in their outputs did you ever see--12--[inaudible]over 13 or 61 under selective pressure? >> we haven't. but, you know, to be honest[inaudible] we haven't looked. it's a really good question and i do think that i do--idon't know if i'm agreeing with you but you're certainly raising the possibility that thereisn't, you know, certain mutations like q61 are selected in certain context versus g12dor v. i think it's exactly how much of a ras signal were particular terminating [phonetic]cell actually tolerate to gain a growth advantage and it maybe different and i know we'll hearsome data about in melanoma about q61s versus the g12ds. and i think that, you know, theability of cell to tolerate this--i also think

that jmml is probably an example of this becauseadults and older patients [inaudible] are not predisposed to leukemia so there mustbe something about that [inaudible] hematopoietic stem progenitor cell that will tolerate moreras signaling in clonal advantage versus what might happen in adult stem cells. so i thinkthere's a lot of things cell context developmental context that are contributing to what--