Colon Cancer Wikipedia

>> right. so, now we move on to the finalact and frank needs no introduction. i can't be bothered to give one so-- >> [inaudible background discussion] >> no. but, i will tell an amusing anecdote.so, it's somewhat irking for me that the biggest talk i ever gave in my life is rememberedfor a comment. well, it's remembered first of all by the fact that there was 10,000 peoplein the audience. the opening plenary session at aacr and i wasn't there. i gave the talkfrom london on-- with my computer on a cardboard box by telephone link. and the aacr beingthe efficient organization they are made me and frank practice the transfer so that theday before we tested the phone link to make

sure everything was working. so, we did thehello frank, hello allen kind of thing and frank said yes and where are you doing thisfrom. and i said, "well, i'm doing it from my office. and just so you know i'm completelynaked." [laughter] and this was the practice and frank said you're not going to say thatare you. and i said-- he said americans don't have much sense of a humor today. don't forgod's sake say that. so, then we the next day we get to the main event and they transferover and frank says al, i know you're there. can you hear me? and i said, "yes i can youloud and clear frank." and frank then said yes, but are you wearing any clothes? [laughter]no one will ever know the truth. that's-- so it was good that that big moment endedup in ridicule as usual. anyway, frank.

>> but, he's a quick thinker. he said an immediateresponse it's a testimonial to the importance of this meeting that i am fully clothed frank.so-- >> [inaudible background comment] >> aw, i was very impressed. [inaudible speakercomments] okay, well thanks for introduction. i'm just going to jump in and talk about somedata from the lab actually almost all of which is unpublished so not from want of tryingi must say. [laughter] and-- but i want to follow up first of all on the topic relatingto neurofibromatosis type 1, nf1, which is a topic that i have been interested in andworking on for a long time. actually i never heard of nf1 before i got a phone call fromben lewin who was then the head of-- the editor

of, "cell" and he called up and said, franki need to review a paper. they found the gene for the nf1 and it's homologous. looks likeit could be a gap for ras so would you review the paper? well, okay i hadn't heard of nf1,but i pretended to be interested. so i said, "okay i'll do it. no problem." he said youdon't quite understand the paper's already on the way to the printer. the only decisionyou can make is completely veto it. so, that was the level of excitement around the discoveryof the gene responsible for this terrible disease neurofibromatosis type 1. and actuallyyou can see here that this paper describing the cloning of the gene was received at, "cell"on the 24th of july 1990 and it appeared in the august issue [laughter] after a thoroughreview on the telephone. and then we followed

up very quickly later and got a cell paperabout november showing that the gap rated domain of nf1 does indeed interact with rasp21 and led to the idea that this disease, which has all these complicated phenotypesshowing at the top, is simply a disease of too much active ras. and this was sort ofa shocking thought to the field really, because as you can see there are many different kindsof manifestations of the disease. but, it also gave hope to the field that we'd prettyquickly be able to find out ways of turning down ras activity and this would be a wayof treating some of the aspects of this disease. but, some-- what is it 15 years later or somethingor 25 years later, 25 jesus, almost 25 years later we still have no therapies for thisdisease. and we really haven't made a terrible

amount of progress in understanding how thisprotein actually works and how the-- what the disease is all about. but, we have madesome progress recently and i want to describe some of that and actually give some hope thatthere may be ways of intervening in this disease, which are more tractable than simply tryingto reduce levels of ras gtp, but are the result of loss of the gap protein neurofibromin.and the breakthrough that led to some new discoveries about a mechanism of the nf1 neurofibrominprotein actually came at a meeting that pete rowlin [assumed spelling] organized in a berkleya few years ago in which eric legious [assumed spelling] presented data that a certain numberof patients with nf1 syndromes actually how a wild type for the nf1 gene. and he had chaseddown the gene, which phenocopies the loss

of nf1 and localized it to a gene called spred1.so, loss of spred1phenocopies nf1 and actually forms a mild form of the nf1 disease. but,this is not a gap and so we thought that we understood the mechanism of action of spred1it might intersect with nf1 and tell us more about how the proteins are regulated. spred1i hadn't heard of that either. but, it is a member of the sprouty family, which is betterunderstood. spred1 and its family member spred2 and also there's spred3 have an evh-1 domain,a kit binding domain and a sprouty domain. and mutations occur across this protein butthey fall into different categories. the pathogenic mutations are either generally at least ofthis time when the first families were analyzed either point mutations in this evh-1 domainor deletions or mutations in the sprouty domain.

and we've-- subsequently there are more mutationsthat i'll just show you in just a moment. so, a graduate student of the lab called irmadecided to do mass spec analyses of proteins that bind to wild type spred1 but not to thepathogenic mutants hoping to find proteins that only bind to the wild type and not themutant and that protein would then be a candidate that would be the critical factor of thisprotein. so, she did mass spec on those two proteins side by side and found the differencewas a neurofibromin. so, neurofibromin binds to the adh1 domain of spred1 and doesn't bindto the pathogenic mutants. so, all mutations in evh-1, point mutations in evh-1 are defectiveand neurofibromin binding. she also showed that that sprouty domain mutants retain nf1binding but don't localize total he plasma

membrane and you can actually rescue themby putting a catch box onto one of these mutants and now they turn of ras through nf1. so,this led to a lot of analysis to a simple model in which nf1 binds consistently andwith high affinity actually to the evh-1 domain but is recruited to ras in the plasma membranevia the sprouty domain. i should say we don't really understand how this happens but thesprouty domain does have palmitoylation sites and other potential ways of bringing it tothe plasma membrane. and there's also the possibility of engagement of the kit bindingdomain with specific receptors such as c-kit. and this actually gave us some hope, becauseit gave us the idea perhaps that nf1 normally turns off ras in response to local signalsfrom specific receptors. not just globally,

but when c-kit is activated ras is turnedon locally and then spred1 brings nf1 to the complex and turns off ras. and this againgave the idea that nf1 disease might be caused by hyperactivation of just a few receptorsthat are sort of controlled by nf1 spred and not everything in the cell in which case drugsthat block c-kit and other receptors that are sensitive to this down regulation mightbe effective ways of dealing with some of the phenotypes. and we're still pursuing thatidea although it's gotten slightly more complicated. so, irma left the lab and graduated and thisproject's been taken up by [inaudible] lab who's done a great job in identifying howthese proteins interact. the first obvious question, which part of nf1 binds to spred1?now this is interesting because the nf1 protein

as we've heard already earlier is a giganticprotein with some 2800 amino acids. it has the gap domain in the middle and then mostof the rest of the protein is unknown territory. highly conserved between humans and flies,70 percent across the whole protein and even yeast has an nf1 lookalike. so, this is aprotein that does something fundamental to cell biology metabolism but we just don'tknow what it actually does beyond regulating ras. and our theory is that this protein sensessomething in the cell and then regulates ras accordingly. we just can't figure out whatit is, because the protein has no domains of any-- doesn't give any clues as to whatkind of interactions it undergoes. anyway, so we tried to chop the protein into bitsto find out which bits bind to spred1 and

were somewhat dismayed to find that the bitthat binds is actually the gap domain. so, we didn't find anything outside of the gapdomain. so, the gap domain itself is sufficient to bind very tightly actually to spred1. withsome refinements now then we have a better idea of how it interacts. so, the evh-1 domainbinds to the gap domain of nf1 but it binds to two discontinuous sequences which flankthe actual catalytic region of gap and these are domains that klaus scheffzek who's collaboratingwith us on this has referred to the extra domains and i'll show you why he came up withthat term. so, this is a structure of p120-gap, gap domain bound to ras. so, here is the catalyticdomain of gap. here's the arginine finger poking into the ras site. this is the rasprotein so this is all beautiful and actually

this structure helped understand how gap actuallycapitalizes or converts ras gtp to ras gdp by the arginine finger poking into the activesite of ras gtp here as we heard from kevan and from callora and others. but, klaus alsonoted an extra domain of the protein, which actually is comprised of the n terminal sequenceof this protein and the c terminal sequence, which come together to form a domain. thenhe saw the structure of nf1 gap domain with some 300 amino acids, not with ras for somereason. but, this is the gap domain itself. but, he also noticed an extra domain, whichlooks exactly the same as the p120-gap domain although the amino acid sequence is quitedifferent. so, here is the amino terminus. you go off here and then go through the gapdomain and come back to the c terminus again.

and it turns out that i could look at thesequence alignments of a whole bunch of gaps. you can see extra domains in all of them includingthe yeast ira proteins. so, this brown piece here is the end terminal piece of the extradomain and here is the c terminal piece. and in each case they form a slightly predicativeform. this conserves structure but with different amino acid sequences. so, this extra domainis where spred1 binds. so, we have mutations actually in nf1 patients, point mutationsthat don't bind spred1 as well as mutations of spred1 that don't bind nf1. so, we're workingon the crystal structure of this complex with klaus scheffzek in innsbruck and he's beenable to show that spred1 and nf1 bind with very high affinity, much more than the affinityof nf1 to ras. this binding does not affect

ras itself, doesn't affect the binding constantto ras and is most likely involved in bringing nf1 to ras in a membrane, because withoutspred1, nf1 doesn't work in cells. so, nf1 protein is totally dependent on spred to bringit to the membrane to work on the ras. in fact, the level of nf1 in the membrane seemsto be directly proportional to level of spred in the cell. so, this is, we think, a majorregulator of determinant of how active nf1 is in the plasma membrane. the spred1 alsobinds the c-kit and to the csf1 receptor and flt3 and maybe other receptors. point mutationshave been now identified in the c-kit domain in legius syndrome families. these-- now thisis the latest version of all mutations that are found in the evh-1 domain, the kit bindingdomain and the sprouty domain. so, there are

several mutations in the kit binding domainand this is one of them and this mutation abolishes binding to c-kit or csf1 or flt3.so, spred1 binds directly to some receptors, directly to nf1 and is localized to the membranethrough the sprouty domain, so it's sort of getting more complicated. but, we think thisis potentially interesting from the point of view of sort of therapeutic intervention,because we found recently that if you introduce into cells activated mutations in say egfror other tyrosine kinases these signals disrupt spred1, nf1 interaction and therefore preventfeedback to turn off ras. and this is a partial solution to a sort of conundrum we've beenwondering about for a while, how does a mutant egfr drive proliferation in the face of feedbackfrom nf1 and other negative regulators. it

would turn ras off. and it turns out thatone mechanism involves disruption of the complex between spred proteins and nf1. so, activatedmutations of egfr or mec tyrosine kinases reduce the binding of spred proteins to nf1and thus disable this aspect of the feedback. there are other angles to this and my friendand colleague [inaudible] in the audience is also working how egfr can maintain levelsof ras gtp through that which relate to this, but i haven't got time to discuss that today.the signal that disrupts nf1 and spred binding most likely is coming into spred rather thannf1. i'm not sure of that yet but yea this-- these signals disrupt binding just of thegrd to spred1, not necessarily the whole protein. and spred1 or spred2 actually in this caseis a phosphor tyrosine protein in cells expressing

mutant egfr. so, our guess is phosphorylationof spred proteins disrupts binding to the grd and this again prevents feedback frommutant egfr. now, we don't know if this is a direct phosphorylation event or if thismodel is actually right. but, if it all pans out it does suggest that if we could blockthat phosphorylation event then nf1 spred would come back to life and turn off ras andwould be another way potentially of shutting down proliferation of cells driven by tyrosinekinases. that's speculation for the future. anyway, so we have finally made some progresson understanding how the nf1 protein is regulated but really just getting started. now, to comeback to this more general issue about targeting kras in human cancers, we've already heardthat this is a complicated network. and we

heard from daphner about the role of whiletype ras proteins in kras cells. we have found likewise that in the kras cancer wild krasmay feed maybe low levels of signals persistently into this network. these cells still respondto growth factors through the wild type, hras and nras. and the combined result of thissignaling through wild type proteins plus the persistent signaling from the mutant allintegrate into this network to bring about transformation. so, we could block proliferationof kras cancers by knocking down hras and nras as effectively as by knocking down krasso you need both signals for various reasons including the beautiful work that daphnerdescribed earlier on. but, all feedback loops which are turned on by persistent signalingdriven by the mutant kras. so, this-- these

signals are attenuated but they're still necessary.and again, this attenuation means that ras cancers are less responsive to inhibitionof say egf receptor than even a wild type cell, but they still need this signal but--so mutant ras cancers need wild type ras proteins before transformation. but, the question hascome up to what extent the k-- the mutant proteins themselves respond to upstream signals.i think that question came up earlier on in the meeting. does mutant kras actually respondto egf at all? this is a tough question to figure out in these kind of complex systems.so, my brightest student cameron pitt, who i hope is in the audience, you there cameron? >> yes.

>> alright, okay, very quietly in the back.again, he's a brilliant graduate student who had the idea of dissecting these complicatedpathways by using a much simplified system. and that is a system, which he's derived basedon again brilliant work from mariner barbaset [assumed spelling] of showing that you canmake cells, mass and culture, which are totally rasless. these rasless cells are made frommice that lack nras and hras and have a kras allele, which can be removed by addition oftamoxifen to the region, because it's a [inaudible] allele, which pops out when you add tamoxifen.so, you can take [inaudible] and make them rasless just by adding tamoxifen. and thesecells stop growing. they stop migrating but they don't die. they just sit there and they'llsit there for quite a while until they get

contaminated and that's the end of it. [laughter]but, they can be rescued. this is a fake diagram by the way. but, this is a cartoon. but, theycan be rescued as mariano showed by any single ras allele, wild type or mutant. i think heshowed that, but any single wild type allele. also by an activated version of craf or brafor an activate version of mec or an activated version of urc. so, anything in this pathwaywill rescue proliferation and migration also. so, that linear pathway seems to be all youneed for each of those phenotypes. now, you can also rescue proliferation by knockingout b or p21 or p53 but that's a more complicated story that mariner shared at the meeting recentlyin florida. he showed us the stunning result that loss of p53 rescues proliferation butnot just by relieving a checkpoint and allowing

cells to sort of drift into s phase but bysomehow activating raf kinase. so, rescue by p53 is raf dependent. so, if you knockout p53 somehow raf comes to life and is able to drive the bus even without any ras proteinspresent at least not h, n and k. and this work wasn't interesting enough for the editorto sell apparently. but, he needs-- obviously looking for a mechanism by which that happensbut i digress. [laughter] and the point was that cameron's been able to rescue proliferationof these cells by introducing each of the ras alleles that we're interested in in thewild type configuration or the mutants. so, now we have cells that for example, only haveq61l or r or any one of these mutants. and now we're in a position to look at differencesin downstream signaling and response to growth

factors and so on in a very clean isogenicsystem. so, cameron is working through that and i'll just show you some data, which isnot a complete story yet. he hopes to graduate soon but you'll see he has plenty of workahead of him to figure out. [ laughter ] >> got to get back to the lab. >> yea, okay. for example, the issue cameup today or i think yesterday actually in julia's talk and stuff to what extent growthfactors utilize ras to activate pi-3 kinase. and i haven't done a complete analysis ofthis but certainly egf doesn't care about ras. so, if you add egf to these rasless cellsyou see full activation of akt without any

ras protein involvement at all. but, you don'tsee any activation of raf. so, activation of raf as you probably would have expectedis totally raf dependent. activation of pi-3 kinase is-- by egf is totally ras independent.so, i'll just sort of set the stage. and so the next thing we want to ask okay if youhave a cell alignment only has mutant ras does it actually respond to growth factors?and if i had to take money from the audience i would probably bet most people would sayno. the ras proteins are fully gtp bound and they-- that gtp loading doesn't change whenyou add growth factors and that is true. so, these proteins are heavily loaded with gtpand the loading doesn't-- is not-- doesn't change measurably when you add egf for example.however, you see activation of map kinase.

and so this is wild type on the left so additionof egf activates phospho mac, phospho urc and to my astonishment the g12d and g12v allelesshow very similar time course of mac activation. so, first obvious response the cells weremixed up but of course, they weren't. so, what this tells us is that egf is providinga second signal on top of the requirement for ras, because this doesn't happen withoutras to activate downstream signaling from even mutant ras alleles, g12d and d12v inthis case. so, the 61 mutations are less responsive to egf receptor however so just on the leftis getting rasless cells are growing with either g12d, qct1l or qc1r. they just lookat phospho mac plus or minus egf receptor. they start off with a high baseline leveland it really doesn't change significantly

when you add egf whereas g12d is clearly stimulatedby egf. so, the degree to which these mutants depend on a second signal is different foreach of the mutants. this is why i think cameron's going to take a while to graduate. [laughter]he's also trying to probe which aspect of egf signaling is responsible for the secondsignal. and one of the things that he's tested, which looks most interesting is the effectof inhibition of phospholipase c and this could be measured or assessed by treatingcells with a drug that inhibits phospholipase c to be induced by egf. and what you can seehere is that the presence of this inhibitor does prevent phospholip activation in thecase of gct1l. but, for d12v these are pretty much indifferent. so, we see differences inrequirement for a second signal and even differences

in signaling between different mutants. so,t12b it doesn't read, is unresponsive to inhibition by [inaudible], but the others are responsive.and then now it starts to get really complicated. so, on that i'm just going to show you this,because it just gives you a sense of how different these proteins are and also the beauty ofthe system of analyzing them. so, cameron here, ips ras proteins from q51lr, d12c andd12d just to look at the signaling complexes that come down with ras. and your eye willimmediately move to this point here to show you that ksr1 [inaudible] protein allegedly,which brings raf and mec together in a complex and urc is very definitely associated withq61r than say g12d. and in fact, this association is affected by the plc inhibitor but thisis not different levels of phospho urc and

differences across the board. so, the signalingcomplexes are quite different between the different mutants. who'd have thought ipingraf itself and looking at proteins associated with that it again shows the complexity ofthis. so, q61r is heavily associated with ksr2 but the other mutants really are not,you know differences in level of mac and in craf activation. so, i won't say can of wormsbut your pandora's box but cameron has opened up a whole new set of issues here that needto be resolved, because these are very seriously different signaling complexes in the presenceof isogenic cells that only differ in mutant rass that differ by one amino acid. so, thisis where we are today. i believe cameron's gone a little further on this but-- and hasa model sort of brewing. but again, this speaks

to the complexity of the system and the needto redissect different signaling pathways downstream from different proteins. so again,we think that ras is essential to activate raf presumably by recruiting it to the membraneas debbie showed in her beautiful talk and somehow promoting dimerization possibly becauseras is already a dimer. and that sort of catalyzes raf dimerization as charlyn suggested. but,in any case some of the mutants require a second kick in the butt and this can comefrom tpa or egf. in some cases its plc gamma dependent. the q51 mutants are less dependenton this than the codon 12s, which is interesting and there maybe differences in the signalinginput into raf. so, that's where we are with that. okay, now i want to go back to our issueabout differences between kras signaling relative

to n and h. in the introduction yesterdayi mentioned very briefly the sort of debate as to whether or not these different frequenciesrepresent different biochemical properties of kras proteins relative to say hras or differencesin the locus between kras and hras and nras in the cells of origin that lead to thesetumors or if different codon usage or whatever. so, [inaudible] who came to the lab a littlewhile ago to test the hypothesis that kras has unique stem like properties relative tohras and nras. and she felt this was likely, because kras is essential for mass developmentwhereas hras and nras are not. so, she set about testing the possibility that kras activatesa stem cell program by mechanisms which are distinct from the classical effect of pathwaysthat hras and nras turn on. and to do this

you went to the sort of classic system ofgoing back to isogenic 3t3 cells transformed by hras versus kras. and she chose pools orclones of cells, which look identical to the eye and also have very similar levels of phosphoakt and phospho urc and similar levels of ras gtp measured by pull downs or by expressionanalysis you see that different clones have very similar levels of induction of map kinasetarget genes. so, in 2d culture these cells grow very similarly and they have similarlevels of the canonical ras pathways not map kinase and pi3 kinase. but, then she movedto a stem cell like assay asking these cells to form spheres along the lines that we heardfrom erica. so, in this assay kras does really well, forms very nice spheres. hras is verypoor. the kras spheres can be regrown in cell

culture and then reform spheres again. andnow you see a really dramatic difference between kras and hras. so, in this assay kras is muchmore efficient and-- from small numbers of can form these gigantic healthy spheres veryefficiently but hras cannot do that. if you ask these cell lines to form tumors in miceat low cell numbers-- with low cell numbers we see that kras is much more potent at doingthis than hras. so, here's a mouse with the kras tumors on one side and h on the otherand clearly kras is much better at establishing tumors in mice relative to hras especiallyat low density. at high density they both work more or less the same. so, kras has thissort of unique feature of standards to use a term loosely. so, to identify factors orpathways that could be responsible for this

manzoo did expression analysis and lookedat expression arrays that are focused on genes involved in stemness. and out of this jumpedthe most obvious genes up regulated by kras but not hras is leukemia inhibitory factor,which lif, which maintains stem cells in an undifferentiated state. i'll come back tothat in just a moment. she also saw up regulation of notch, which has a role in development,cmyk, which does everything. twist promotes emt and blocks differentiation and also veryinterestingly genes involved in drug resistance, because in parallel to this analysis she hadfound that the kras, even the 3t3 cells, are much more drug resistant than hras to a wholebunch of different drugs. so-- and they up regulate drugs involved-- genes involved inpumping drugs. they down regulate frizzled-8,

which is a major driver of non-canonical windsignaling gle-2 and other genes. but, we've really focused on lif and frizzled-8 for thetime being. lif we like because it's a soluble cytokine. it's a member of the il-6 familythat activates staff three signaling. and this is from wikipedia. [laughter] i knowokay i've never heard of lif either. but is that a wikipedia? so, if you-- in this hierarchyof pluripotent stem cell system if you withdraw lif from these systems you get differentiation.so, lif maintains pluripotency and actually i believe in the original yamanaka cocktailfor genes and proteins which keep mouse cells in a pluripotent state lif is one of the factorswhich does just that. in human systems cmyk seems to do something similar and you don'tneed lif, you have cmyk. anyway lif certainly

is has a track record or a history of maintainingpluripotency. so, kras transformed cells put out lif into the medium and this signals tothe cells themselves and presumably to the stromal cells and sends signals that maintainor induce stemness. we've tried to validate lif as a potential target first of all using[inaudible] and knockdown lif and then asking the cell form spheres and knocking down lifcertainly knocks sphere forming activity. ann and manzoo have some early attempts tomake cell lines expressing shr neg against lif and show that these cells are reducingtumor formation. and in mice that do form tumors you see less metastasis. now, theseare very preliminary data, which is now in the process of substantiating with betterstable cell lines from human systems instead

of mice and also with underclonal antibodiesthat neutralize lif, which we've obtained recently. so, our feeling though based onthe preliminary data is that inhibiting lif inhibits stemness in kras mouse transformcells and in pancreatic cancer cells and in mouse models. but again, we're trying to getpublishable data to support that. so, how could kras-4b possibly turn on a whole newprogram when it interacts with the same effect as hras and nras? well, the clue from literaturewhich we think is sending us in the right direction came from this group who had shownpreviously that calmodulin binds to kras-4b and this is a specific interaction. it doesn'thappen with 4a or hras or nras. so, we don't know exactly where this interaction occursbut carla will soon be telling us, because

she's working on the co-structure with-- >> it's a tough, tough structure. >> oh here we go. here we go. it's a toughstructure, okay. she needs more protein, correct. >> needs more protein. >> it's always the same refrain. but, we doknow from some-- >> [inaudible audience comment] >> yea, yea, yea. we do know from by physicalanalysis and from mutagenesis studies that calmodulin it depends on gtp binding. so,it's essentially another affect or a ras. its binding is affected by mutations of codon,switch 2 and also regions around 160 and probably

some component to this polybasic sequenceand it's sensitive to phosphorylation of serum 181, which our own trevor bevona [assumedspelling] has shown is a substrate for protein kinase c. so, phosphorylation of serum [inaudible]calmodulin off kras. i'll show you that in just a moment. so, we've basically confirmedin our own hands that in the presence of calcium you can easily pull down calmodulin boundto kras but not hras from cells. and this interaction can be reproduced and-- >> it's a gift [inaudible]. >> i thought you were yawning. >> okay, sorry. i look forward to being yourslave in the future.

>> anyway, with kras but not hras interactionin a gdp dependent manner to calmodulin. and we prefer not using biophysical analysis you'reusing recombinant proteins as well as pull downs from cell extracts. so, our workingmodel is that calmodulin binds to kras and in doing so inhibits cam kinase as well ascalcineurin, another downstream modulator regulated by calmodulin. by cam kinase seemsto be the most important from what we've seen so far. so, cam kinase, calmodulin dependentkinase regulates the number of pathways including suppression of beta catenin and tcf throughthis pathway and activation of frizzled-8. so, high cam kinase gives you high frizzled-8and low tcf. and calmodulin also regulates nuclear translocation of nfat to affect tissuepolarity. so, when calmodulin binds to kras

you lose cam kinase activity, which is definitelytrue as i'll show you and you lose localization of nfat to turn on genes involved in thesepathways. and you up regulate the tcf by relieving inhibitions by cam kinase. and these datajust support each of those. so, these transform cells have lower cam-- phosphor cam kinasethan hras transform cells, which actually elevate it. nfat no longer accumulates inthe nucleus in the kras versus the hras. and you see kras turns on top flash. we read outfor beta catenin and signaling by reading the negative effects of cam kinase. hras doesthe opposite and this actually i think is significant, because h and kras go in differentdirections. so, kras basically suppresses non-canonical wind signaling and turns onthe readout from canonical wind signaling,

that is tcs signaling but also has multipleother effects from this pathway. so, this we've done in cell lines and we wanted tocompare tumors driven by hras and kras to see if we can see the same phenomenon certainlystarting off in mice. the first problem that hras and kras tumors in mice tend to be differenttissue types so it'd be difficult to make a head to head comparison. so, in collaborationwith allan balmain and his group manzoo took advantage of a knock-in mouse in which thehras gene had been knocked in to the kras allele. and then these two-- these mice formedtumors in response to carcinogen and the tumors form a similar frequency. but, anecdotallythe kras tumors look more as emt like and more spindly than the hras ones but the numbersare the same more or less. but, when we analyze

cam kinase activity measured by phosphor camkinase from the kras driven tumors compared with hras you can see that the kras ones justlike the cell lines have suppressed cam kinase whereas hras ones are on fire. and this iskras versus hras. so, the same locus, only difference is hras versus kras, same carcinogen.we get the same number of tumors but the kras tumors are more malignant looking and theyhave reduced cam kinase and relative to hras. so, the biochemistry thing holds up for whatwe've seen in [inaudible]. now, to transfer that calmodulin binding is necessary. we madea phosphor [inaudible] of serums as done by matt holderfield in collaboration with manzoowhere we converted the serum to aspartate or gluten, glutamate. and this mutant theserum 181d no longer binds calmodulin. so,

now we have a nice control and we've gonethrough the whole set of experiments again to show that this mutant of kras, althoughit can transform cells in 2d, looks similar in 2d and no longer forms spheres and no longeractivates beta catenin. nfat goes to the nucleus and for-- most importantly its way reducedin its ability to form tumors in mice. so, it's simply preventing calmodulin binding,takes away the stemness features of kras, makes it look more like hras. and reciprocallywe can take hras transform cells in which cam kinase is blazing and inhibit cam kinasewith a small molecule inhibitor of cam kinase and now we see that we can actually reversethe situation and make hras more stem like including forming spheres and activating thesame pathways that kras would normally activate.

so, that event, either inhibiting cam kinaseor preventing calmodulin binding is a sort of binary switch, which converts hras to krasand back again. frizzled-8 is something which manzoo is really focusing on, because thisseems to be a major component of this whole story. frizzled-8 was shown previously tobe of all the wind receptors the most active at turning on non-canonical wind signaling.and this is the one that's turned down by kras. so, kras again suppresses non-canonicalwind signaling through cam kinase. knock down kras, frizzled-8 activity goes up even inhuman pancreas, tumor cell lines, pang1 and pang2 and you basically see a reversal ofall the effects of cam kinase and so on as we described earlier. on the other side ofthe coin if you overexpress frizzled-8 in

kras transform cells or in pancreas cellswith mutant kras you suppress the-- all the stemness effects. so, high levels of frizzled-8shut down lif expression, turn on cam kinase and basically make cell-- make the kras ratherequivalent to actually knocking out kras itself in terms of reducing ability to form growthin [inaudible] and these downstream reporters. so, frizzled-8 can override the stemness featuresof kras and even to the point of preventing tumor formation by-- pancreas tumor cellsjust by simply overexpressing frizzled-8. in human tumors we see low expressions offrizzled-8 in metastatic pancreas tumors relative to normal tissue giving us the sense thatduring development of pancreas tumors the initiating event is kras and most likely iskras signaling gets cranked on by increased

expression. we start to see sequestrationof calmodulin and loss of frizzled-8 expression and increase in beta catenin signaling asa result of inhibition of cam kinase and suppression of frizzled-8. so, this leads us to potentialnew therapeutic approaches to kras cancers either based on lif expression or preventingkras binding to calmodulin or possibly going downstream in these pathways to interferewith the stemness features that kras turns on. and that's something which we're pursingalone. final topic i would mention briefly is rather orthogonal to the rest of the talkand that is the possibility of treating kras cancers using delivery of sirna and nanoparticles.and this is something which i've been very interested in for several years through collaborationswith joe gray and through an arrangement with

mark davis at cal tech who's a brilliant chemicalengineer and biologist now actually who has devised some very nice particles, which areable to deliver sirna into tumor cells. these particles have been optimized for the optimalsize to leak out of blood vessels into tumor beds. the right charge, so that they don'tget hung up on membranes such as in the kidney. and they've also been engineered so that whenthe particles get into the endosomes they neutralize the endosomes and release sirnainto the cytoplasm. so, in each of the steps in delivery of this sirna these particleshave been optimized to successfully deliver sirna into tumors. and they've actually beeninto clinical trials and shown to be safe and as mark is able to show, knock down ofa target gene in humans by systemic delivery

of these kinds of particles. but, what gotmy attention really was studies that we did in collaboration with mark and joe gray testingthis sort of best case scenario where you have tumor, which has a very high level ofexpression of a surface protein, which can be targeted. and this is [inaudible] in thiscase in the bg474 breast cancer model. they got a ton of [inaudible] on the surface andthey depend on [inaudible] for survival. so, we have a-- we know exactly what the dependencyis and how to target these tumors. so, mark made these nanoparticles with herceptin onthe surface and loaded them up with sirna against [inaudible]. and after three shotsof this material in the bloodstream of these mice we were able to completely eliminatethese tumors in almost all the mice. herceptin

alone looked pretty good until we stoppedtreating and then it came rebinding back, rebounding back. target particles with scrambledsirna really didn't work or some effect, they really didn't work. and particles with notargeting agent really didn't work either. so, the combination of a good targeting antibodyand a good payload was able to actually eliminate these tumors quite dramatically. so, thisencouraged us to think for kras cancers and other cancers actually if we knew the rightproteins to target and we have sirna but super potent against kras or other components ofthe pathway this could be a therapeutic approach. and it was in collaboration with scott loweand christoph felman [assumed spelling] and gene lowe at nc and the-- sorry [inaudible]at nci. tina a brilliant person up in the

lab has developed super potent sirnas thatmake it possible to do systemic treatment. these are sirnas against all members of thecomponents of the downstream elements of the ras pathway optimized by christoph's algorithm.most of them knocked down the protein of interest by at least 80 percent, thereabouts. so, theseare all extremely potent. this is really important, because having super potent sirnas makes itpossible to load up several of them into the same particle or to knock down a half a dozengenes at once, which you can't do if the sirnas are not potent enough. also by looking atthe effects of these sirnas in rasless cells actually against kras tina was able to showthat super potent sirnas have virtually no off target effects relative to off the shelfones in that kind of system. and she's able

to show that you can knock down in this caseseven genes at once all in one simple transfection and all with equal efficiency as with a singlesirna itself, because they don't compete for the sirna machinery. so, we can do great combinationsof sirnas, because they're super potent and clean. and in mice we can see knock down ofkras. so, these mice have a red dye in all their cells and gfp in their tumors and wetreat these with nanoparticles that contain sirna against dsred. you can see with eachtreatment you get a dip in expression of dsred but no real effect on the tumor growth. but,if you also include a kras sirna you see it again the dips of red showing the tumor cellsgot transduced. but, you also see a decrease in tumor size as you can see also by lightimaging. so, we can get significant knock

down of kras tumors just by delivering sirnaagainst kras itself. and tina has been able to optimize this by adding additional genes,which give it an extra kick. and most recently she's found the kras plus a combination ofpi3-kinase alpha and beta significantly increase the efficacy of kras particles in-- nanoparticlesin a head to head comparison, which is shown here. so, part of this is these are obviouslyearly days but we could imagine engineering particles that are efficiently transducedinto kras tumors and contain a cocktail of kras itself and potential other genes thatwe're worried about that might come up as resistance genes or adaptive mechanisms thatwe can sort of preempt and knock down as well such as [inaudible] 1 for example. so, wethink this technology in the long run might

be one way to solve the problem of attackingkras cancers and we're doing this in parallel to our other more traditional attempts byidentifying targets as i've just described. with that i want to stop and thank the peoplein the lab whose work i've described. this is manza wong [assumed spelling] who did thestemness work. irma did the first work on identifying that spred1 binds to nf1. andellen has followed this up by showing the mapping where they bind and working on thecrystal structure. jackie has been very helpful working with manzoo and others in the lab.and matt helped manzoo on the calmodulin binding mutant. cameron pitt had the idea of usingisolated mecs rescued by different isoforms of ras to dissect signaling. and this is tinawho did the other work. i have one project

i haven't mentioned from stephan who's inthe audience i think today is tethering attack on the c terminal tax box of kras using thetechnology that kevan described in the first talk of the meeting. we've got compounds thatattack the sustain. we've got electrophilic derivatives or we can get a crystal structureso we know how the whole thing works. so, when we get that stephan can come and presentit. and with that i'll thank all of you and be happy to take questions, thank you. [ applause ] [ silence ] carla matos [assumed spelling].

>> yea. i was delighted to see you have cameron'sresults on boosting the g12v with the egf, because that's completely consistent withwhat we found comparing the g12v with the q61l structure and biochemistry. if you rememberwe have g12v boosting that ras [inaudible] pathway but not completely. and g12v is ableto hydrolyze slowly, not as efficiently as wild type ras-- >> right. >> whereas the q61l completely saturates thatpathway. so, you can imagine that the q61l would absolutely not respond to further egsstimulation but that the g12v might have some room for boosting that up a little bit. so,i think that's a really, really awesome result--

>> thank you. >> that you get that completely correlateswith what we're seeing-- >> excellent. >> in the biochemistry and structure. >> okay. >> the other-- yea. >> i'll take awesome, okay. >> awesome. >> thanks. so, along those lines i had a questionas to whether you had examined any other cell

types to see if the complexes form by thevarious mutants with different rtks and so forth was different in different cell types. >> no. and don't suggest that to cameron,okay. [laughter] >> i know he wants-- >> [inaudible] >> to graduate. >> yea, but it's a great question. >> yea. >> in fact, i was just discussing this yesterdaywith debbie that we've done some expressed

analysis on raf family members from tcga lookingat the [inaudible] there. in pancreas cancer we're surprised to see that it's you knowmostly craf and araf and hardly any braf. and i presume in different tumor types thatratio will be different. i'm sure that's going to be true of probably all the componentsof these systems. so, i would be willing to bet that these complexes are very differentin different cell types but to be determined. that's a great question and melanoma is themost different of all. >> frank, over here. >> oh. >> hey. i had a question about the first part,the spred nf1. so, i was wondering how much

activity is sort of constitutive to keep likea constant break on the ras signal. is that part of it or do you think it's a feedbackloop that comes later on and then dampens the signal? >> yea, that's a great question, yea. yeai'm starting to think that there's a constant effect of nf1 you know a really atonic suppression. >> because they're associated-- >> together all the time, right. >> in fact, it's [inaudible] 1 and the membrane.as a matter of fact its-- the amount of nf1 in the membrane is sort of set by the amountof spred1. so, all the cartoons suggested

it translocates. that's what happens whenyou overexpress spred1. it moves it into membrane. but, in real life it may be there all thetime constantly keeping ras into the gdp state. >> and then when you-- we have a signalingevent then you move it aside to allow ras gdp to accumulate, which i think is actuallymore likely to be true, because ras proteins are constantly gdp bound normally. >> so, there's normally excess of gap overany incoming signal, right. the [inaudible] would not be engaged probably very much atthat point. so, i think it's more likely what you suggest. there's a tonic down regulation,which is relieved during signaling. >> actually that's consistent with karen jakowski's[assumed spelling] data that shows that tpa

causes complete destruction of nf1 withinfive minutes after tpa addition to cells. >> right underneath, how there is that increasein ras gdp, yea, yea. >> so, my question actually is related tothe lif connection. so, p53 is a noted regulator of lif. >> it is. >> and i'm wondering if there's any role forp53 in the regulation of lif by kras that you see in that context? >> yea, that's a great question. i've hadthat discussion with arnie levine over a drink. yea, he's extremely interested in lookinginto this further. we haven't done anything

on that at all. but, we do know that fromthe literature that lif expression correlates with bad outcome in lung cancer. that couldbe because it tracks with kras, but it could also be tracking with p53's data, so we haven'taddressed that yet, but really need to. it's a great point. okay. >> [inaudible] off to the airport. >> okay. you miss anything, wrap up or? >> no. i think wrap it up. >> okay, well okay goodbye everybody. [laughter]and thanks to jordan. [applause]