Zebrafish as a Model for Human Cancer by Leonard Zon, MD

my name is Karen Schakowsky I’m from Brigham and Women’s Hospital and the first speaker would be dr. Lenz on from Children’s Hospital he’ll be talking about a zebrafish as a model for human cancer and therapeutic development great well it’s a really a pleasure to be here and thanks for coming back from lunch even though we started a little bit late I’ve really enjoyed this conference it’s a really a great intersection of science and medicine and so a great job of definitely putting it together and I’m gonna teach you all about zebrafish genetics so let’s think about cancer in a developmental context you know we all have think about stem cells and how it may relate to cancer biology and how embryogenesis is certainly a period of time when cells are growing and differentiating and so if you look red thing okay great perfect excellent so if you look at a human embryo at 28 days in a zebrafish embryo at 19 hours you can see that they’re very similar in their development and if you think about developmental pathways that are involved in how this embryos are put together notch win and BMP that’s certainly involved in cancer has been demonstrated by somatic mutations and so really the zebrafish has an attribute of being able to study cancer in some unique angles just from a developmental context but the other part about zebrafish is that you can make a lot of them so every mother has 200 to 300 babies every week and we’ve recently made a new tank called the ice bond and we can get about 10,000 embryos in ten minutes so it’s almost at the level that we’ll be able to scale this up to screen equivalent to a cell line screen and I’m gonna talk almost all about chemicals and so you can put embryos into 96-well plates and you can have libraries of chemicals and add them to these wells and then screen for chemicals that could affect signaling pathways to Regenesis and this may have something to do with cancer and adulthood and there’s a time lapse of video microscopy that you can do there’s some micro fluidics that is available for zebrafish and then we have fish that are transgenic fish that have green organs and so you can screen for particularly developmental pathways or in situ hybridization which is to look at where the RNA is expressed or antibodies so actually just before this talk Bruce said you really need to solve the Rast problem so we’re trying I would say that we don’t have a major hit but I want to tell you about how we’re going about it so um Dave Lang and out in my lab but now he’s at Mass General had found that if you overexpress K wrasse in a specific muscle compartment during embryogenesis you get fish that have rhabdomyosarcoma x’ and he was able to take these rhabdomyosarcoma x’ and we made a symmetric strips from fish we call these fish and chips and we were able to do microarrays and we were able to show very similar to what Tyler has shown that there’s a wrasse signature and our signature and Tyler signature is very very similar to each other and we then wanted to see if this was relevant to development and we had a fish that has the heat-shock promoter driving h wrasse and so you could heat up the fish and wrasse would get activated and we took some of the genes from our wrath signature and we studied their expression in the presence of this heat so it would be activating wrasse and we found a number of genes that are significantly up regulated and then as I said zebrafish a great chemical biology system so we incubated those in a pan RAF inhibitor and most of those were specifically downregulated and one of the ones that struck us is this dust 6 gene which was tremendously activated a transcriptional level and then was suppressed by a raff inhibitor so um what is dust 6 so it’s a transcriptional target and it’s also a negative regulator of Rast map kinase pathway it’s a dual specificity phosphatase six and it functions to D phosphorylate map kinase kind of as a negative regulator so map kinase gets activated this thing gets activated it then stops the pathway from

working and so if we take our heat shock fish and here’s no heat shock but then if we heat them up you can see all throughout the entire embryo dust six is transcriptionally activated so this gives us a model to actually do a screen so the screen that we’ve undertaken is to take a heat shock wrasse line and we heat it up and normally what would happen is this gives a burst of expression of rass and then you’ll follow this by bursts of expression of dust 6 and the screen will be to look for chemicals that interfere with dust 6 expression and you would expect that this would get typical map kinase inhibitors or it would also get heat shock or transcriptional in activators and we were able to rule those out okay so we are able to screen 2500 chemicals of known action and then added in a variety of other chemicals from libraries that we had available to us and basically we have categories of expressions so essentially a level 3 hit is what’s a normal fish would look like we found some chemicals that activate the map kinase pathway and we found other chemicals that will be inhibited so these are the chemicals that we found we found 18 chemicals that are confirmed suppressors of hrs and it’s really true I mean they’re very robust and they are completely suppressed and these are a list of the chemicals and it’s not too surprising that a number of these are map kinase inhibitors and other chemicals but as you might expect it’s almost too many things because not all these would actually block cancer and so we wanted to use the zebrafish to see if we could block cancer so we returned to dave’s rhabdomyosarcoma model and i just want to tell you a little bit about this model you basically inject cay wrasse into one cell embryo and seven days later you have cancer so it’s a very very fast asset and you can see this a lot of the fish died and dave also developed a KO injection technique so if you just go inject two plasmids one with K wrasse and one with red then all the tumors will actually be red so Jun Ying Lee a graduate student in the lab developed this assay where essentially she started looking at the tumors at day seven and then she measured the growth on day ten and day thirteen using microscopy for the red color and over here you can see the growth of the tumor again and day seven and then in green day tan and then day thirteen would be the in yellow so this worked incredibly well and were able to monitor the growth of these tumors and so she then tested each of the 18 chemicals to see which would inhibit the tumor and really only two of them do this one of them is an anti map kinase chemical and the other I’ll talk about so this is again the map kinase inhibitor and you can see the growth rates being suppressed with this PD drug and we also look at the length of the fish and make sure that that’s not suppressed by this so it’s not just a complete antiproliferative effect and I think this is very similar to work by Tyler where he found a chemical that would inhibit K wrasse induced intestinal hyperplasia and so I think that the message is very similar that if you can find chemicals that block the map kinase pathway this would be very helpful well this TP CK chemical also specifically slows growth and doesn’t slow the growth of the fish and so this one was pretty interesting because it was originally designed as a protease inhibitor and it’s known that topically applied TP CK was shown to inhibit tumor initiation in a mouse skin tumor model that was induced by DN ba and it specifically inhibits s6 kinase based on some of the work that was done by John Glenn is his lab but not Macker akt and we were able to take zebrafish embryos and actually grind them up and again similarly you can see the PD drug will inhibit map kinase and this TP CK will not inhibit map kinase or a ket but specifically inhibits the phosphorylation of the s6 kinase phosphorylation step so one of the questions we had is what would happen if you combine the map kinase inhibitor and this TP CK and we can basically at a certain dose regimen show that synergistically they actually will will suppress tumor growth and you can see here compared to a control how the tumor growth is actually suppressed this T P CK was there was some attempts to actually get it into the clinic and unfortunately it failed because of the toxicity profile but nevertheless it has a unique activity in this s1 kinase and

it’s probably a little bit downstream of mTOR so I’m not gonna go into the mechanistic studies of this but and I just wanted to use it as an illustration but you know we’ve seen this diagram so many times in this meeting it’s but simply by inhibiting Meccan and inhibiting the other pathways I think that the synergism will be very valuable and this asset can be used to study really any set of chemicals so I think that the fish has a role now perhaps in in pre-screening your drugs if you have too many that you want to don’t want to do mice ok now what I’d like to do is to switch and tell you about a model that we have for melanoma in the lab and also a new drug that we’ve discovered that we think is interesting for treatment of melanoma so in this melanoma model what we did is we took v600e human be RAF and we injected it into fish and when we do that we get a very dark fish but none of the fish ever get melanoma but if we inject this into a p53 mutant fish then all the fish get melanoma and we’ve done microarrays on this fish at melanoma and it looks very similar to human melanoma so rich white who’s a he Munk fellow who’s in the lab was looking at the embryos for these different genotypes and there is a set of embryonic melanocytes and what’s surprising is even though this fish is mutated in p53 and it’s activated b-raf and all of its melanocytes it has a normal number of melanocytes and this was quite surprising to rich and he said I think there’s something wrong with this fish it’s just that we don’t know what it is so he decided to take these embryos and do micro arrays and what he was able to find again was a set of genes that are specifically upregulated in the b-raf p53 combination and so this turned out to be about 377 genes or so and then what he did was to take the zebrafish melanomas and find a set of genes that were specific to the zebrafish melanomas and did an overlap of these two circles which produced 123 seen gene signature which predicts which embryos are going to get adult melanoma which is quite interesting and when he looked at the list of these genes they were almost all neural crest progenitor genes so this includes a gene called Creston sox10 edn are be the end of feel and B receptor and also a few melanocytes genes so this set up a model that he wanted to see what was going on during embryogenesis and we have this marker in zebrafish called Creston and this marks the entire neural crest and what you can see here is that this fish has extra neural crest cells you can see in the head and the cells that are coming down over the brain there’s more cells that are coming down and then later at about 72 hours of development only in this genotype you can see these extra arrests of cells and so he developed a model where basically maybe this cell is the precursor of the melanoma and so perhaps you could think of melanoma as a stem cell disease and that you have an extra set of progenitors this might give you a higher chance for during replication to basically get mutations and so it could be true than in fair-skinned individual they just have a higher pool of stem cells well we went on with Scot grantor to look at human melanomas and there’s a lot of neural crest genes that are expressed in human melanomas this includes the end of feel and B receptors and and it’s villain itself and also sox10 so we’re seeing a large subset of melanomas that actually Express the neural crest progenitor genes that he had found in his acid so then what rich said is well if neural crest progenitors are expanded in melanoma perhaps I should look for a drug that gets rid of the neural crest and maybe that would be a new treatment for melanoma so what this would represent is something that attacks the cell fate of the cancer okay so it’s a little bit different than attacking the signaling but it’s the fate of the tumor okay so what he did is he took wild-type zebrafish put them into Wells and now we’ve got our robotic in situ hybridization for creston so you basically look for a fish that doesn’t have any neural crest cells so he found one chemical that completely erased the neural crest okay and it was this chemical here and it was a novel structure we had no idea what it was and

luckily there are some new chemo informatica bases particularly one from UCSF and we actually put it in and it said that this chemical is a dye hydro orotate dehydrogenase inhibitor now I have to say I did go to medical school I never heard of this enzyme but but what is true is there’s a drug which is fda-approved for arthritis called laughlin amide and laughlin amide is a d h OD h inhibitor and it has no structural similarity to our chemical but we decided to test the flute amide and it also completely erases the neural crest so d h OD h had never been implicated in vivo for neural crest development and I’ll tell you later that this is the and one of the enzymes that’s involved in producing uridine and so it’s a very strange result that you would have such a specific finding in the absence of your Dean okay we wanted to find out if this affected mammalian neural crest development and we worked with Jack Moser who’s in Sean Morrison’s lab and Sean is one of the world’s expert in neural crest stem cells and he can grow these neural crest spheres and basically we plated them with and without la flute amide now because we’ve just decided to use that as our chemical and then evaluated secondary nurse fear assays as a role of looking at self renewal and what we see here is that lafoon amide in a dose-dependent inhibition of self renewal so it’s clear that this chemical can affect mammalian cells also and specifically in the neural crest okay now again for some biochemistry for all of you we have the CAD d h OD h pathway okay so you have CAD which is an enzyme also I didn’t hear about and then it eventually goes to D H OD H as you make uridine and this is actually a highly regulated pathway so actually if you look on your microarrays I bet a lot of in cancer this is a highly regulated pathway but it’s certainly relatively ubiquitous and for a while rich and I were questioning each other about whether to take the project even for because it’s hard to know how your diene would be involved in neural crest development but we were helped by two observations so the first is that there’s actually a zebrafish mutant called perplexed it’s called perplexed because it doesn’t have a jaw and it has defects in craniofacial cartilage and small eyes and so that was very exciting because it gave us some feeling that this were true but nothing was better than when this paper came out I think in January so it turns out that in they needed a case to do exome sequencing of the human genome and so the first time that anybody’s ever cloned a mutation from a novel disease using exome sequencing was for Miller syndrome and when they sequenced Miller syndrome who has craniofacial cartilage small eyes in cold all neural crest defects you basically found that was mutated in dho th so clearly what we had found is related to the human condition so we felt that what we were doing was relevant and the problem was you needed to know how this worked and I had a model which was based on UDP dependent glycosylation you may know that neural crest cells need to migrate it’s known in some literature that they have specific UDP Glick nak UDP glucose all those things are used for migration and rich tried to phenocopy this effect with some glycosylase a–‘s and things and nothing would work but and he tried some DNA replication agents also didn’t give an absence of neural crest and so he found one paper in the literature which had an effect of uridine on transcription and that was this paper from 1997 which is that you need uridine to have your transcription elongation to occur so transcription starts with initiates and then it pauses for a little bit and then it needs to elongate and it’s this elongation step that requires your edema and in this paper you could actually take an elongation factor called SBT v and add it in vitro to the reaction and now you would have a rescue of elongation so it seems that this would regulate transcriptional elongation so this SBT v is an interesting factor it’s also known as D si F and D si F is the pausing factor but then there’s a kinase that comes in which is called peat FB and PTFE will phosphorylate the pausing factor and the polymerase and this will allow the elongation step to occur and I think this is a great place to think about cancer therapeutics

actually at this elongation step so in our lab for another project we happen to have an SBT v null mutant and so we stained it for neural crest and it has no neural crest and we actually went to the trouble of doing microarrays on our Liffe luna my treated embryos and the SBT 5 null allele and they are exact FINA copies of each other so you can see here that there were a little flute imide 223 genes down 183 of those genes were found in the SBT 5 and then even the genes that are up our shared overlap now luckily zebrafish is a genetic system so you could look for genetic interactions with the drug and so we have in our lab a hypo morphic allele of SBT 5 and you can see that compared to wild types it’s a little hype opieop Laughlin amide you can convert this to an animal that has no neural crest demonstrating genetic interaction and if you look at the charts here in a wild type or heterozygote there’s no pigmentation phenotype but in the mutants there’s a light phenotype but essentially an a dose dependent reaction with love flute amide you can convert either the mutant to stop having neural crest or even the heterozygote now will stop having neural crest if you give this chemical we wanted to demonstrate that transcription elongation was actually being regulated and our first way of doing this was to actually make primers towards the 5 prime end and the 3 prime end of a cDNA and what we found is that a gene like MIT F which is the master regulator the melanocytes in age does not elongate and what’s strange about it and we don’t understand why is all neural crest genes will not elongate and all notch genes will not elongate but everything else elongates fine so there’s some specificity to this that we don’t quite understand but it’s interesting so then we wanted to see how global this was and we collaborated with with P trawl and Charles Lin with rick young’s lab who’s one of the pioneers in in this transcription elongation field and this was an amazing experiment we actually did chips seek in human melanoma cells these are two independent human melanoma cells with a pole two antibody and happens is is that this is a typical gene and so normally you have initiation and then you would elongate but you can see here for many many genes that the elongation doesn’t happen so it really demonstrates across the genome that some genes are not elongating and then we wanted to find out what are those genes that aren’t elongating and it turns out that most of those genes are Mik targets and so this interact time is is demonstrating that these factors are bound by Mik and this was interesting because rick had had a paper that mick actually regulates transcription pause release and so it’s one of the major factors that’s a that is regulating in cancer these along and in embryonic stem cells this this elongation effect so with all that we decided to see if these chemicals would work so low fluid amide on its own actually is a pretty good anti melanoma drug but we also tested it with the plex akan drug that’s the v600e drug that’s in a trial and you can see clear synergy in three different human melanoma cells and so then we decided to do zina grafts and what we can see here is that at a certain dose the plexi con drug will have an effect laughlin amide will also have an effect and the two together shown here have a substantial inhibition of tumor growth so we think this is an interesting drug that actually is an fda-approved drug that might be able to be put into the clinic and if you think about it the plexi con drug is probably affecting cell proliferation and the laughlin amide is attacking the fate of the actual melanoma cell and so the two of them together might be an interesting combination so just in closing i hope i’ve convinced you that zebrafish chemical genetics is a high-throughput way of finding inhibitors of neural crest lineages in vivo we think we have a new role for transcription elongation in the the neural crest and that this inhibition of DHA odh blocks neural crest development and cooperates with b-raf a blockade to abrogate melanoma growth and I just want to thank the people who’ve done the work this is the new improved version we’ve now added

stripes ok so so thank you very much and I’m happy to take questions the inhibitor is particularly potent I’ll make targeting transcription elongation but why it’s so specific for homicides I thought yeah I mean I’d say that there’s there’s two parts of this so it’s definitely neural crest genes and notched argit’s so some people think McQuaid be part of the notch pathway or wind pathway you know a lot of people have that thought but though we in a separate story we had published in cell in July we had a similar story in blood in the hematocrit extends ELLs and and erythroid cells that the cell specific transcription complex actually binds to the elongation complex and so our expectation is something like MIT F itself or perhaps socks 10 it would give the neural crest specificity we don’t know what would give the Mik effect or the notch effect and so we would but in theory it would be some type of regulator that controls that pathway has anything to do with your your a Dean or your a delight pools yeah so we’re pretty sure that it does but one of the things we try to do is to rescue with your delay unfortunately didn’t work because we were it was so toxic to the embryos that did you try Yura Dean itself yeah yeah it’s that transported yeah and it didn’t work it didn’t work but there are other pigment mutants that are also not rescued by your Dean that are in this pathway but it’s pretty specific so I you know I think that it’s likely to be right but we don’t have the definitive evidence because the your Dean wouldn’t rescue yeah and is there any attempt now to try to refine the molecule to make it more potent and more specific yeah so there’s a couple companies who’ve made derivative chemicals Sanofi is one of them that i think and so we’d like to you know work with the most active chemical we can and then put it into our model and see how it works but my view is that this one is pretty good but it probably could be better yeah yeah then I know your theory about the elongation as imposing as an important step towards the terminal differentiation so if you were to take an inverse approach and ask yourself what are the elongation components which are tissue-specific and would you be able to drag them and would you be able to generalize these not doing genetics but now candidate gene or protein approach yeah so so far there isn’t a self specific regulator of transcription elongation and I think that in this cell paper we had it it explains it in that things that are the transcription factor complexes that are cell specific are actually binding to the actual elongation machinery conferring the cell specificity so there may be I mean it’s definitely true for instance in transcription initiation Bob Tegan has found some really interesting cell specific regulators so it could be that they would be found but so far to date and we haven’t seen them so I think it’s this coupling of the intrinsic cell specific machinery to the elongation machinery that’s going to generate the specificity I should also say that probably in every single tissue this is happening so for instance it’s known that my OD and a muscle will actually commune to precipitate with PT FB and likely regulate elongation and this work which say in melanocytes it’s the same thing we have also in blood so I think the cell specificity comes from the original master regulator oh right right yes to find out that yeah yeah that’s a good point I don’t know anything directly about how the chemistry would work to fashion this in a cell specific manner but what’s surprising is that this uridine deficiency you know we’ve looked at other organs in the embryo and they’re pretty much normal so it seems to be very specific in embryogenesis for neural crest cells and maybe a couple other tissues but it’s so there is some specificity that’s intrinsic which we don’t understand yeah question on the side effects or toxicities of a double inhibition I think this is fascinating do you sacrifice the fish or you know do you do some studies no regional pathways or organ development yeah so we’ve done that actually and it’s true that if you in particular in the first story where we had the tpc K and we lower the dose we got a much less toxicity if you gave

the combination and so I think that we haven’t evaluated toxicity preps and b-raf and we’ll flu to my LeFleur might and my understanding is it has a toxicity of liver damage that’s one of its major toxicities but otherwise is tolerated pretty well so so again I think looking at combinations and then being able to section the fish look at pathology is looking gene expression is certainly an option absolutely okay thank you very much