1 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES PUBLIC HEALTH SERVICE FOOD AND DRUG ADMINISTRATION CENTER FOR BIOLOGICS EVALUATION AND RESEARCH INTERNATIONAL ASSOCIATION FOR BIOLOGICALS NATIONAL INSTITUTE OF ALLERGY AND INFECTIOUS DISEASES NATIONAL VACCINE PROGRAM OFFICE WORLD HEALTH ORGANIZATION - - - EVOLVING SCIENTIFIC AND REGULATORY PERSPECTIVES ON CELL SUBSTRATES FOR VACCINE DEVELOPMENT - - - WORKSHOP - - - WEDNESDAY, SEPTEMBER 8, 1999 - - - The workshop was held in the Plaza Ballroom, Doubletree Hotel, 1750 Rockville Pike, Rockville, Maryland 20852, at 8:30 a.m., Harry Rubin, DVM, and Martin Myers, M.D., Co-Chairs, presiding. PRESENT: HARRY RUBIN, DVM Co-Chair MARTIN MYERS, M.D., PhD Co-Chair NAOMI ROSENBERG, PhD Session Chair HENRY PITOT, M.D., PhD Speaker ALEX VAN DER EB, PhD Speaker JAMES McDOUGALL, PhD Speaker 2 PRESENT: (continued) STEPHEN BAYLIN, M.D. Speaker WALTER DOERFLER, M.D. Speaker JAMES COOK, M.D. Speaker SATVIR TEVETHIA, PhD Speaker FRANK SISTARE, PhD Speaker MICHAEL FRIED, PhD Speaker SANDRA RUSCETTI, PhD Session Chair CLIVE PATIENCE, PhD Session Chair LEONARD EVANS, PhD Session Chair DAMIAN PURCELL, PhD Session Chair PAUL JOLICOEUR, M.D., PhD Session Chair ALSO PRESENT: DAVID ONIONS, PhD 3 I N D E X Page SESSION 2: Mechanisms of Neoplastic Development and Neoplastic Cells Tumorigenicity: Implications for Cell Substrate Development Introduction: Session Co-Chairs 5 Animals of neoplastic development 5 Henry Pitot, M.D., PhD Multistep carcinogenesis, Harry Rubin, DVM 21 Transformation by DNA viral oncogenes 45 Alex van der Eb, PhD Hit and run transformation leading to 61 carcinogenesis, James McDougall, PhD DNA methylation and epigenetic mechanisms of 75 carcinogenesis, Stephen Baylin, M.D. A new concept in viral oncogenesis 89 Walter Doerfler, M.D. Role of nonspecific NK/macrophage cell host 108 responses in assessing tumorigenicity, James Cook, M.D. Role of CTL host responses and their 125 implications, Satvir Tevethia, PhD Transgenic animals that might be useful in 137 identifying unsuspected oncogenic factors, Frank Sistare, PhD SESSION 2: Panel-Audience Discussion 155 SESSION 3 (Part 1) Viral-Viral and Viral-Cellular Interactions Introduction 229 Generation of MCF retrovirus as a model of 233 viral-viral and viral-cellular interactions Sandra Ruscetti, PhD 4 INDEX (continued) Page MuLV packaging systems as models for estimating/ 248 measuring retrovirus recombination frequency, Clive Patience, PhD In vivo and in vitro effects of mixed retrovirus 264 infections, Leonard Evans, PhD Cross-species pathogenesis of replication- 277 competent retrovirus in immunosuppressed hosts, Damian Purcell, PhD Pathogenesis of defective retroviruses, 296 Paul Jolicoeur, M.D., PhD 5 1 P R O C E E D I N G S 2 Time: 8:00 a.m. 3 CO-CHAIRMAN MYERS: On behalf of the 4 National Vaccine Program, welcome. Our first speaker 5 this morning will be Henry Pitot of the University of 6 Wisconsin, and he's going to start talking about 7 animals of neoplastic development. 8 DR. PITOT: Hopefully, this thing works. 9 Does it? Both Harry and I chose to speak down here. 10 I guess the other speakers can decide whether or not 11 they can see the slides from up here or down there. 12 I think my function this morning was 13 basically to try to cover in about 15 or 18 minutes 14 most of the more commonly used animal models for 15 cancer development, and probably without further ado, 16 I'll just start with the first slide, which is, I 17 think, familiar to most of you. 18 It's probably the most widely used animal 19 model for neoplastic development. It's, obviously, 20 used by the regulatory agencies, and some people have 21 considered it what might be considered the gold 22 standard of carcinogenesis. 23 Unfortunately, it has a lot of drawbacks, 24 as those of us that work in experimental systems know. 25 But looking at the literature, in fact, probably most 6 1 of the data on whether or not a chemical is 2 carcinogenic really comes from this sort of data that 3 you see here, which really is based on studies that 4 went back many years, actually to the 1930s where 5 animals were given a carcinogen for an extended period 6 of time until tumors developed. 7 In this particular system there's 8 actually, as you can see, a fairly standardized type 9 of thing which allows one over a two-year period to 10 determine the development of neoplasm. 11 Now nice as it might look, there are a 12 number of problems, and I don't have the time this 13 morning to go into all of the difficulties. But one 14 of the more interesting facets that's come out of this 15 whole series is this table which I borrowed from Dr. 16 Ames' publication several years ago, which basically 17 demonstrates, looking at a series of some 380 18 different chemicals that were tested, the relationship 19 between whether or not they were carcinogenic or 20 whether or not they were mutagenic. 21 I think you can look at that table and 22 immediately see that perhaps one of both the advantage 23 of this chronic bioassay and the disadvantage is that 24 there are a variety of chemicals which, obviously, are 25 mutagenic and carcinogenic, which one might expect, 7 1 but there are also a large number that are perhaps not 2 mutagenic but still are carcinogenic. 3 So this type of system has led us to the 4 finding that we can do certain things with it, but 5 clearly we cannot begin to study mechanisms and 6 looking at dissecting the whole process of 7 carcinogenesis. 8 So as a result of that, of course, one 9 goes back a number of years, actually to the 1940s 10 when some of the earlier studies on what might be 11 called, what we call today at least, multi-stage 12 carcinogenesis were carried out. 13 This slide is just a classic experimental 14 slide which shows the studies that were done 15 originally by Barren, Blum and Schubik and others in 16 the 1940s which basically was done on the back of a 17 mouse, painting with a chemical, then administering a 18 material which at that time was an irritant called 19 croton oil, but today we call a promoting agent, 20 eventually ending up, if the format of the system is 21 appropriate, with neoplasm. 22 Now what I mean by appropriate is that 23 there is first an initiating agent, this material that 24 is given on the back of the skin, followed by a 25 promoting agent. If you reverse the process, it 8 1 doesn't seem to work, at least in most instances. 2 If you, for example, try just the 3 promoting agent, nothing happens; and the interesting 4 experiment that was done by Bautwell some 20 years 5 later was that, if you change the format of the 6 administration of the promoting agent, then you do not 7 get neoplasm. 8 So here was a system which allowed for 9 carcinogenesis to occur. The endpoint of this 10 particular system is seen in this slide, which is the 11 typical papilloma of the mouse. It is not a malignant 12 neoplasm. It is benign, but it is useful in this 13 particular system, which really tried to understand 14 what might be called the latent period of 15 carcinogenesis. 16 So this system was used, actually, for 17 about 30 years, during which time the chronic bioassay 18 developed. This system never became really important 19 with respect to regulatory agencies or anything of 20 that sort, but it was used primarily in academic 21 circles in trying to understand the whole process of 22 carcinogenesis. 23 It was in the 1970s that Dr. Carl Pareno 24 at the Argon laboratory actually first sort of broke 25 the ice by demonstrating that other tissues besides 9 1 mouse skin showed this exact same phenomenon -- that 2 is, of a stage phenomena in the development of 3 neoplasia. 4 Dr. Pareno's system, which is sort of 5 shown very briefly here, was to administer a 6 carcinogen -- in this case, this is actually a 7 modification of his which is done in our laboratory -- 8 with a mitotic stimulus followed by the administration 9 of some promoting agent. 10 This material, this promoting agent, which 11 did the same thing as the irritant in the skin system, 12 now had become sort of a much more well recognized of 13 what we call today promoting agent. 14 Over the years since Pareno's 15 demonstration, there's been sort of a parallel 16 development of both the mouse skin system and the rat 17 liver system, which is what Pareno studied, in 18 addition to -- and I won't have time to name them all 19 here -- probably a dozen other different tissue 20 specific multi-stage models of carcinogenesis. 21 So it's not just unique as it was a few 22 decades ago just to the mouse skin. It clearly goes 23 well beyond that. 24 So I'll spend just a few minutes talking 25 about the liver system, simply because that's what's 10 1 done in our laboratory. I think potentially it can be 2 utilized perhaps to go on to other model systems as 3 well. 4 One of the big advantages in the liver 5 system is that we think, and there's certainly 6 reasonable evidence to argue, that one may identify 7 initiated cells. That is the very first step in the 8 development of neoplasia. This is just a histologic 9 section which shows in the middle of this section a 10 single cell which now expresses a gene, the piriform 11 of glutathionase transferase, which normal hepatocytes 12 do not express. 13 We have found this to be a very useful 14 what is called a marker for identifying initiated 15 cells and, as you'll see, these cells then under the 16 influence of a promoting agent go on to clonally 17 develop into small colonies in the organ -- this is 18 all done in vivo -- expressing the same particular 19 genetic component. 20 Now this -- It turns out not only is the 21 expression of this particular gene in these cells 22 abnormal, but a whole series of others which have been 23 studied over the years by many, many different 24 laboratories, either as markers or trying to 25 understand what actually is going on during this 11 1 process of promotion. 2 As a result of that, looking to mining 3 years of work and many, many experiments, one can come 4 up with a series of conclusions for the effect of 5 promoting agents. 6 Now I'm sure in the discussion we can 7 spend an awful lot of time on various other aspects, 8 but I'm going to home just in on these two 9 characteristics which I feel are probably the most 10 critical aspects of these agents which cause a 11 selective enhancement of cell replication basically of 12 initiated cells. That is, they cause the replication 13 of these cells different from the normal cells. They 14 will also cause normal cells to replicate, but the 15 initiated cells they cause much more effectively and 16 selectively. That was shown several years ago by 17 Farber and others. 18 The other aspect that they also do has 19 been shown recently by Dr. Schulte Herman in Vienna. 20 They selectively inhibit the programmed cell death of 21 cells. So these two selective actions, actually, I 22 would propose -- and we can discuss it -- actually can 23 explain virtually all of the actions of promoting 24 agents. 25 Now, of course, one easily says what is 12 1 the molecular mechanism of this. I think at the 2 moment we really don't know. We certainly have a lot 3 of ideas about what's going on, but there's still a 4 lot of work to be done. 5 You can also see in this system, perhaps 6 unlike the mouse skin system, that it is possible to 7 quantitate the development of neoplasia. So if one, 8 for example, administers a -- initiates cells by just 9 a single dose, a very small dose of a carcinogen, you 10 can see you increase the number of these putatively 11 initiated cells by three orders of magnitude, which 12 perhaps is not unexpected. 13 Also notice that there are a certain 14 number which occur spontaneously. Then if you promote 15 and cause these to develop into these small colonies, 16 only about one percent of these develop into that; and 17 although I don't have the quantitation on there, one 18 can also go on into neoplasm and, in fact, of this 19 perhaps one-tenth of one percent of these cells 20 develop into neoplasm. 21 So it allows one in a model system to 22 quantitate the various stages and really get an idea 23 of how effective the various stages are. 24 Now to do this, there have been model 25 systems developed -- this is one of them -- where you 13 1 can actually dissect each of the three stages. One 2 can initiate with this very small dose of a 3 carcinogenic agent, an initiating agent, which by 4 itself will do nothing. 5 One can then promote, as shown with the 6 blue line here, with a promoting agent. Then 7 somewhere down the line after these focal lesions, 8 these small lesions, have occurred, one can administer 9 another agent which Bernie Weinstein coined the term 10 progressor agents, which actually then causes during 11 this period the development of malignant neoplasms. 12 So during this period from the 13 administration of the initiating agent to this point, 14 basically you have these focal lesions. The cells are 15 different phenotypically. They are undoubtedly 16 different genetically, at least from point mutations 17 and other things, but they are not different 18 cytogenetically. They have perfectly normal 19 cytogenetics. 20 It is in the stage of progression that the 21 cytogenetics of cells become abnormal, as we all know. 22 Not only do they become abnormal, but they 23 continuously get worse. That is, they continually 24 evolve in what is called evolving karyotypic 25 abnormalities. 14 1 From this, both this model and others, 2 again one can determine certain characteristics. I've 3 just listed some of the consequences of this major 4 factor of the stage of progression. Four of them 5 here, gene amplification, gene deletion, 6 rearrangement, and a very effective way, certainly 7 cells that are in the stage of progression are much 8 more effective at accepting external genetic 9 information by transection mechanisms than are normal 10 diploid cells. 11 Consequences of -- You might say the 12 functional consequences of this particular phenomenon 13 can be seen in this slide, just some of them. There 14 are probably a number of others. I've listed a few of 15 these here, and probably for the discussion of this 16 particular symposium, one of the most important ones 17 is this particular part right here. 18 It is well known that cells in the stage 19 of progression lose the expression of the MHC 20 determinant which, of course, makes them a very 21 difficult target for the immune system, because there 22 will be no interaction between the MHC system and the 23 T-cell receptors. 24 I'm sure that will be discussed in many 25 other components here, but it does allow one by using 15 1 this multi-stage phenomenon, then to dissect this 2 component into its various units and actually try to 3 determine changes which occur at each of the different 4 stages. 5 Finally, and I've taken this from an 6 article by Ray Tennant from the NTP, the more modern 7 models of carcinogenesis are transgenic models, either 8 transgenesis itself by adding genes in the standard 9 transgenic mechanism or by gene targeting of 10 knockouts. 11 What I've listed here are three -- 12 actually, two of the more common ones, the upper two 13 here, the T-53 knockout animal, both the homozygote 14 and the heterozygote, and the so called TGAC animal in 15 which the viral Harvey ras gene has been associated 16 with either a Zeta hemoglobin promoter or also, more 17 recently, with a keratin promoter, such that one can 18 actually do many of the skin tumor painting without 19 worrying about initiation, because basically one 20 already has initiated cell populations. So promoting 21 agents actually under this system become complete 22 carcinogens, in quotes at least. 23 Just briefly, looking at the upper one, 24 the P53 deletion mutation, this is taken from an 25 article that was recently published. This just shows 16 1 the tumors that develop in these animals, and you'll 2 notice that -- these are mice -- that the normal 3 animals spontaneously, at least out to this number of 4 weeks, develop very few spontaneous neoplasms. But 5 as; you might expect, in both the heterozygotes and 6 the homozygote P53 animals, the spontaneous 7 development of neoplasia is really extremely high. 8 This creates a problem perhaps when one 9 actually is trying to determine mechanisms, because it 10 is rather difficult to add to the system something to 11 perturbate it, while at the same time the system is 12 spontaneously developing a very large number of 13 neoplasms. 14 On the other hand, if you look at the 15 neoplasms that are produced in such a system, you can 16 see that there seems to be a somewhat, at least 17 quantitatively, different spectrum. Perhaps in the 18 homozygotes deletion animal, lymphomas are the 19 predominant component, whereas in the heterozygotes, 20 one can see that the lymphomas still are a significant 21 component of it, but both the soft tissue sarcomas and 22 osteosarcomas, which are really a very small part here 23 -- these soft tissue tumors become very prominent. 24 Now there are -- This is not the least in 25 this long list of modern models. This just shows you 17 1 -- Some of you in the back probably can't read this, 2 and it's really not that critical. It's not complete 3 by any means. There are a whole series of transgenic 4 models which are now developing. 5 The difficulty is, if, of course, one 6 wants to study the development of a specific neoplasm, 7 it may be extremely useless. On the other hand, 8 because the animal is programmed for certain things, 9 it becomes perhaps difficult to study the earlier 10 stages in the development of neoplasia and, rather, 11 what one will be looking at is the later stages 12 themselves. 13 So just to sort of sum up, what I've shown 14 you are three basic models: This chronic bioassay 15 which is perhaps, from the regulatory standpoint, the 16 gold standard, but from the investigatory standpoint 17 it has many problems; the multi-stage model which 18 allows one to look and dissect at the various stages 19 of neoplasia some of its mechanisms and its 20 characteristics; and finally, the newer transgenic and 21 knockout models which, certainly in specific areas, 22 can answer specific questions, but probably they have 23 to be geared and tooled to do this, and it seems that 24 they will probably be most significant in answering 25 questions in this final stage of neoplasia -- that is, 18 1 of progression. 2 So I'll stop there, and hope I kept on 3 time. 4 (APPLAUSE) 5 CO-CHAIRMAN MYERS: We have time for a few 6 questions. If you would come to the microphone and, 7 as we are transcribing the meeting, if people could 8 identify themselves, both by name and institution. 9 DR. LEOWER: Johannes Loewer. Has ever 10 DNA of neoplastic or normal cells been tested as 11 initiator or promoter in these -- If so, what was the 12 outcome? 13 DR. PITOT: I'm not sure of all the 14 experiments. I know it was tested once in mouse skin, 15 but probably not appropriately. It didn't work. But 16 I guess that, certainly, if one can use in -- as we'll 17 turn in the next discussion, in cell culture, one can 18 clearly by transection of DNA get a situation where 19 one can get -- I'm not quite sure what the stage one 20 is dealing with, but certainly a transformation. 21 DR. FRIED: I'm Mike Fried, ICRF, London. 22 You said in the beginning that certain 23 carcinogens were not mutagenic. What is the 24 mechanism, how they work, if it's not a genetic one? 25 DR. PITOT: Well, we can spend another 19 1 three hours on that one. I'll give you my opinion, 2 and I'm sure it can be discussed otherwise. 3 I showed you that in the putatively 4 initiated cells in the liver, there are a number of 5 spontaneous lesions. We well know that spontaneous 6 carcinogenesis is very characteristic of all mammals 7 probably, more so of some than others. 8 So one could make the argument that the so 9 called non-genotoxic carcinogens which are not 10 mutagenic in the chronic bioassay or in the Ames 11 system and other systems are, in fact, promoting 12 agents. What they are doing is they are causing the 13 development of spontaneously initiated lesions through 14 promotion and then spontaneously into progression. 15 That's a fairly simplistic answer, but I 16 think it follows the facts. 17 CO-CHAIR RUBIN: I want to call attention 18 to some recent work by Zarbell and Tilley with -- I 19 think it was NMU carcinogenesis in the mammary gland 20 of rats where they found that this so called mutagenic 21 agent, in fact, was causing mammary cancer, mammary 22 epithelial cancer. But it turned out that all the 23 tumors had a Harvey ras mutation in them, and that 24 Harvey ras mutation preexisted in the mammary 25 epithelium, and the agent -- you could call it then a 20 1 promoter, but in effect what it did was select with a 2 clonal expansion of those clones in the mammary gland. 3 DR. DOERFLER: My name is Walter Doerfler 4 from Cologne in Germany. 5 Isn't it a problem perhaps with the 6 transgenic model as people are now finding applying 7 the DNA array technology that, when one does a 8 knockout, perhaps quite a number of functions are 9 altered, not at all exclusively the functions that you 10 are knocking out, but other functions. 11 So the interpretation must become 12 extremely complex. 13 DR. PITOT: I certainly agree. I think 14 that the use of the gene targeted and transgenic 15 animals may be very useful, but I think it's new 16 enough that we really don't know all of the problems 17 that may occur. 18 AUDIENCE PARTICIPANT: In your multi-stage 19 approach, how specific to the species that you are 20 testing, the fact is, particularly related to the MHC 21 expression, it is clearly published -- the regulation 22 of the expression differs from species to species. 23 So, therefore, what could be changing 24 expression in your rat model may not be applied to 25 humans. You know very well that expression of MHC 21 1 genes is all over the place. It's up-regulation, 2 down-regulation. 3 So how can you apply this overall and 4 generalize your results? 5 DR. PITOT: I think that you see the sort 6 of thing that you're discussing in the human also in 7 the rodent, not only in the rat but the mouse, and the 8 same sort of system. So that you will find neoplasms 9 that will up-regulate the MHC components, others that 10 will down-regulate them. 11 Unfortunately, that it will happen in the 12 same neoplasm, that some cells will up and some will 13 down. So, therefore, you're going to be faced with a 14 problem, I think, from the vaccine question actually 15 of trying to get cells to up-regulate, and there are 16 mechanisms to do this, so that some of the vaccines 17 may well be able to work. 18 I think it is applicable to all different 19 systems, and certainly the human is perhaps the best 20 example. 21 CO-CHAIRMAN MYERS: Thank you very much. 22 Our next paper will be presented by our Co-Chair, Dr. 23 Harry Rubin, Professor of Molecular and Cell Biology 24 at Berkeley, and he's going to talk about multi-step 25 carcinogenesis. 22 1 CO-CHAIRMAN RUBIN: As most of you know 2 may know, our speaker was supposed to be Larry Loeb. 3 He was supposed to be the first speaker. Larry Loeb 4 suffered a back accident trying to climb into a cave, 5 which he shouldn't have done, with his grandson. 6 We tried to find at a late date some 7 alternate speakers. It was too late to get them to 8 come, and so as the designated chairman I volunteered 9 to speak. That will explain the rather primitive 10 state of some of my displays, which were -- Some of 11 them were made at four o'clock this morning. 12 So I would like to talk about -- 13 basically, about the transformation of cells in 14 culture, both the transformation to the neoplastic 15 state of primary cells, if you like, of normal cells 16 obtained from the animal and in some cases even from 17 humans, and then into permanent cell, the highest. 18 The first observation that was made in a 19 semi-systematic way about the transformation of normal 20 mouse cells into sarcoma cells beginning with fiber 21 blast was made, of course, by Wilton Earle of the 22 National Cancer Institute, beginning in about 1943, 23 and working together with Sanford they investigated 24 this problem for many years. 25 Of course, as many of you know, at first 23 1 what they were working with was carcinogen induction 2 of transformation in mouse cells, but they found that 3 their controls were transforming at the same rates as 4 the carcinogen treated cells were, and that started 5 the whole study of spontaneous transformation of cells 6 in culture. 7 Probably the earliest really systematic 8 investigation of the transformation of mouse cells in 9 culture, although that's not what it was called in the 10 particular paper, was by Todaro & Green in 1963, 11 Journal of Cell Biology, and a similar systematic 12 investigation was carried out by Paul Kraemar and 13 associates in New Mexico with Chinese hamster cells. 14 What I've listed here are the basic -- not 15 necessarily the sequence of changes, but an indication 16 of a variety of changes that occur during spontaneous 17 transformation of rodent cells in culture. So let me 18 just run down them. 19 You take the mouse fibroblast, in the case 20 of Todaro & Green, and while they multiply fairly well 21 when they're first put in cell culture, the rate of 22 multiplication decreases with every passage. They go 23 through a crisis. If you're transferring a small 24 number of cells -- in that case like three times 105 - 25 - you may, in fact, lose the culture, in a sense like 24 1 you lose human cells eventually. 2 If you have enough cells at each transfer, 3 every third day you go through a crisis period, which 4 means it looks like the cell culture is dying out, but 5 then there begin to appear variance in the culture, 6 and gradually you get an increase in cloning 7 efficiency. You get an increase in the rate of 8 multiplication back to the original rate and perhaps, 9 in some cases, an increased rate. 10 Then you find that the growth -- the cells 11 start to multiply to a higher saturation density. 12 Roughly about the same time, they will grow in low 13 serum concentrations or lower serum concentrations 14 than they would grow in before. 15 Then later on at a later step the cells 16 will grow in suspension in soft agar or methyl 17 cellulose. This is very often associated with an 18 ability to produce tumors in the animal, but not 19 always. It is not an invariant accompaniment of 20 ability to produce tumors. 21 Anywhere along these lines -- and we'll go 22 into this in a few minutes -- you begin to get 23 transformed colony morphology and transformed foci in 24 the culture. Some of these things such as the 25 transformed foci or production of tumors are spoken 25 1 about in a qualitative way when, in fact, they're 2 quantitative, progressive increases in the neoplastic 3 transformation of the cells. 4 You see that, I think -- In particular, 5 this was shown by the group that worked with Chinese 6 hamster cells at Los Alamos, Kraemer & Kramm, 7 etcetera. They found that their hamster cells went 8 through many of these stages, but as they were going 9 along, they were continually testing them for ability 10 to produce tumors in the isologous animals, in the 11 Chinese hamsters. 12 What they found, that even though all 13 these changes were occurring -- there were tissue 14 culture representations of changes in the behavior of 15 the cells after they had gone through their crisis 16 also -- they would not produce tumors, and that was 17 the case even where every cell in the culture had 18 shown chromosome aberrations. After the 20th passage 19 in culture, every cell in the culture was 20 chromosomally abnormal. 21 They were still not producing tumors when 22 injected back into Chinese hamsters, but about the 23 40th or 50th passage they could produce tumors, but 24 they could produce tumors only under certain 25 conditions. 26 1 The particular conditions were (a) to use 2 10 million cells -- a lot of cells -- and instead of 3 inoculating the cells directly subcutaneously into the 4 hamster, they were inoculated into gelatin sponges 5 that had been implanted subcutaneously. Under those 6 conditions, they produced tumors of the gelatin 7 sponge, which up to that time didn't worry the hamster 8 too much, but in fact they spread into the rest of the 9 hamster after that. 10 Then with further passages when chromosome 11 aberrations really became quite severe in all of the 12 cells, the tumors could be produced by direct 13 subcutaneous inoculation, and even later tumor 14 production would occur with smaller numbers of 15 inoculated cells. 16 Now the cells would also change -- this 17 was shown first, I think, in the mouse -- would change 18 after they produced the tumor in the mouse, if you 19 take the tumor out of the mouse, and the tumor 20 required the inoculation, let's say, of a million 21 cells. Once you got the tumor, you could produce a 22 tumor with 1,000 cells. 23 So that there was continuous progression. 24 What you see here is really a complexity of 25 progression, that each one of these stages that we're 27 1 talking about is, in fact, an indication of another 2 change in the population of cells. 3 Could I have the next transparency? Well, 4 this I did draw at about 4:30 this morning, and 5 they're from memory. So if someone has a better 6 memory than I do, please inform me. 7 This is taken from Todaro & Green's paper 8 in 1963, which is real classic. What's shown in these 9 top two graphs here is a plot on the horizontal axis 10 of the number of passages of the cells in culture, and 11 the increase in the cell population every three days, 12 n-zero being the number inoculated. Sometimes they 13 used n-1, that if they counted cells at one day and 14 then they counted them in three days when they passage 15 them. 16 At first, you could start out with a 17 population that would increase about eightfold in 18 those three days, doubling every day, but with each 19 successive passage the cells would multiply slower. 20 You saw a lot of pathology in the culture. 21 Then at about the tenth passage it looked 22 like the whole culture is going to die out. This is 23 in passaging three times 105 cells, which at least at 24 that time was thought to be a low inoculant of cells. 25 Then with further passages variance appeared. That 28 1 could be distinguished morphologically from the 2 original cells. 3 They continued to grow faster and faster, 4 and they reached a plateau, which was a relatively low 5 plateau. These cells would not produce tumors in 6 immunocompromised mice, for example, which is usually 7 said to be the case, but in fact if you wait three, 8 four, five months, very often a tumor will appear. 9 Again, when you get that tumor and 10 reinoculated it into mice, it will produce a tumor 11 very fast. So all that time these cells have been 12 incubating in an occult stage in the animal, and 13 finally a variant appears there, just like it appeared 14 in cell culture. 15 Now if one increases the concentration of 16 cells at every passage -- and this is a 3T12 passage 17 which means 12 times 105 or 1.2 times 106 cells were 18 passaged at each passage; three simply means they were 19 passaged every third day -- then you get a decline in 20 the rate of multiplication of the cells, but it's a 21 much shallower decline, for one thing. Although 22 there's something that resembles a crisis, it does not 23 appear to be a very serious crisis. 24 Then you start to get the upswing in the 25 rate of multiplication of the cells, and they end up 29 1 multiplying faster than these -- maybe not much faster 2 -- but what they do do is multiply to a much higher 3 saturation density. 4 These cells, when inoculated into mice, 5 will now produce a tumor fairly quickly. The only 6 difference here is that these cells have been passaged 7 at four times higher concentration than the 3T3 cells. 8 So there's something about the higher population 9 density of the cells that furthers the transformation. 10 In fact, a few years later in 1968-1969 11 Aaronson & Todaro found that if you took cells 12 directly from the mouse, start passaging them in 13 culture, and inoculating them back into mice every few 14 passages, that the higher the density of the cells 15 that you made your passages at, the quicker the tumor 16 would start to show up in the mouse. So there was 17 something either selective or inductive about high 18 population density in inducing tumors. It is 19 obviously the selective aspect that you select for 20 cells and that you grow at high density. 21 Now I want to say a word about human 22 cells. This is a paper by Smith & Hayflick in 1974, 23 and Hayflick can correct me on this. I'm sure I've 24 got some of it wrong. Again, it's from memory. 25 Basically, this was an experiment with 30 1 clones of human fibroblasts. I think it was the WY38 2 line. So what you're looking at here is the 3 proportion of cells that go through a certain number 4 of divisions in culture. 5 So I think it's widely thought that human 6 cells will go through 50 divisions and sort of fall 7 off the end of the cliff. That's based on work with 8 large populations of cells, but when you study clones 9 and their capacity to continue multiplication in 10 culture, the capacity drops off in some clones right 11 away, and right at about roughly the tenth passage or 12 so -- I can't really give you an exact figure -- about 13 half the cells have lost the ability to multiply. 14 So it looks like the loss of capacity to 15 multiply is a stochastic event. It's a random loss of 16 capacity to multiply. What you appear to end up with 17 is the last surviving clone. That's the clone that 18 goes 50 plus or minus ten divisions in culture. 19 That result was basically confirmed by 20 Peter Rabinovitch of Seattle in 1983. He used a new 21 technique for labeling the DNA of the cells, 22 gromodeoxyuridine label. He worked with whole 23 populations of cells, and he found that the percentage 24 of human cells that were cycling decreased linearly 25 with the number of divisions that the cells went 31 1 through in culture. 2 Now if you look at published literature 3 also, it looks as though human cells will not undergo 4 spontaneous transformation. There are many ways of 5 eliciting long term growth in culture, but there's 6 been some unpublished work, the senior author of which 7 was a cytogeneticist, Tamara Ignatova, was working in 8 Marguerite Vogt's lab at the Salk Institute a few 9 years ago -- unpublished work using Li-Fraumeni cells, 10 which are unstable -- relatively unstable human 11 fibroblasts. 12 Again, what she found was that, if you 13 kept the cells in a condition of confluence, which as 14 we all know results in contact inhibition of the cells 15 -- but I think it's not often appreciated, there's 16 also a good deal of cell death that occurs at 17 confluence -- then after a month or two of leaving the 18 cells intact in a confluent layer, you start to see 19 large scale chromosome rearrangements and indefinite 20 growth. 21 So these cells apparently have undergone 22 some kind of genetic changes involving chromosomal 23 changes that are fairly easy to see that results in 24 indefinite growth. That is, in effect, spontaneous 25 transformation. Unfortunately, as far as I know, this 32 1 work has never been published, and I think there's 2 some disagreement among the group that's done it about 3 aspects of it should ultimately get published. 4 Do I have another? That's it? Well, I 5 want to turn to some slides now. I want to deal now 6 for a few minutes with a cell line, but what I'm going 7 to say about this cell line, I think, can be 8 generalized to cell lines in general. 9 So what I want to talk about is the NIH 10 3T3 cell line, which is famed in modern molecular 11 genetics as the first line in which there was 12 demonstrated a transformation of an animal cell line 13 by DNA extracted from a human tumor, the EJ bladder 14 carcinoma. 15 A problem with that original finding -- 16 there are many problems with the original finding 17 which three laboratories reported back in 1981 -- was 18 the fact that again, if you take the original line of 19 NIH 3T3 cells and you let them sit at confluence for 20 a couple of weeks, that you start to see spontaneous 21 transformation. 22 In fact, probably the outstanding 23 characteristic of Jane Hill, Todaro and Aaronson's 24 original line of NIH 3T3 cells is the ease of 25 spontaneous transformation. All you have to do is 33 1 leave the cells for two weeks at confluence and 2 transfer them once, and foci start to show up 3 throughout the culture. I'll give you some examples 4 of that. 5 So again there's something about 6 surprisingly enough from conventional genetic thinking 7 where one thinks that at the highest rate of mutation 8 or genetic change is during cell division when DNA is 9 being replicated. Actually, it appears that the 10 highest rate of chromosomal change of mutational 11 change -- we're not sure which, or maybe all of them 12 together -- occurs when the cells are inhibited at 13 confluence. 14 You might ask why. Well, there's 15 something else which I already inferred, that there is 16 a considerable amount of cell death when these cells 17 or many cells are left at confluence for an extended 18 period of time. So there's considerable damage done 19 to the culture. 20 So let me show you a few slides of the 21 kind of observations that one makes. Let's go back to 22 the first one. Okay. 23 What you see here is a culture of NIH 3T3 24 cells which ordinarily would look like this, a mono- 25 layer of cells, nontransformed cells, but when left 34 1 for a couple of weeks or if left for a couple of weeks 2 and transferred, you get a high density population. 3 This is a transformed focus where the cells are no 4 longer in any regular arrangement. They're criss- 5 crossing. You can't identify individual cells, 6 because they're piled so thickly. 7 This is an unstained slide, which was one 8 of the first observations we made back in about 1988- 9 1987, which surprised us that the cells which people 10 were talking about as being used as targets for 11 transformation by the Harvey ras oncogene would 12 transform by themselves. 13 Now another feature is, when one looks at 14 independent transformations -- and now we're getting 15 to another point in transformation. So you start out 16 with a single culture, and you start splitting it, and 17 you let each one of them go to confluence various 18 numbers of times. You get independent 19 transformations. 20 If you look at the foci in each one of 21 these dishes -- these are the cuts from individual 22 dishes -- in a rough sense, the foci, which are these 23 thick aggregations of cells on a background of mainly 24 monolayers of cells -- each of the groups of foci look 25 different from one another. 35 1 So you can't simply talk about 2 transformation of cells by producing foci. These -- 3 If you collect the cells from any of these foci and 4 inoculate them into mice, they'll produce tumors, and 5 the tumors in the mice will all look like sarcomas. 6 But in culture where you can find a distinction in the 7 appearance of the cells, each one of the independent 8 transformations is different from every other one, 9 which means that the genetic changes that are 10 occurring in these cells are different in each one of 11 these cases. 12 So there are genetic changes, chromosomal 13 rearrangements, deletions, etcetera, that are going on 14 in these cultures that cause that kind of variation, 15 which is reminiscent of what pathologists were telling 16 us way back in the Thirties and Forties, that no two 17 tumors look exactly the same. 18 Well, okay, now I want to show you want 19 will happen with the NIH 3T3 cells if you culture them 20 in a certain way. So what you see here are a series 21 of cells that have been through what we call a primary 22 assay, which means two weeks at confluence, and every 23 two weeks they are transferred to become confluent 24 again, and this is a tertiary assay, and this is after 25 eight such sequences. 36 1 You start out with a culture that has no 2 foci. By the third round of confluence, even when you 3 place only 1,000 cells on a background of 4 nontransformed cells, you see these thick foci. By 5 the eighth passage at confluence, even 200 cells are 6 producing about 100 foci. So that's a high rate of 7 transformation. 8 Here is a culture derived from this same 9 initial culture, but transferred originally about 100 10 times at low density. You go through the same 11 operations, and you see much less evidence of 12 transformation. 13 So there's continuous selection and change 14 going on in these established cultures, and one cannot 15 rely on them by saying, well, they're going to remain 16 the same as long as we continue passaging them. It's 17 very dependent, just like the original transfer of 18 cells directly from the mouse -- dependent on 19 population density. Even the established lines depend 20 on population density. 21 If you look at another criterion of what 22 was happening to these cells, in these cultures if you 23 took the undiluted cell populations, you got a very 24 rapid rise in saturation density, which remained 25 roughly constant after that; whereas, in this case 37 1 even by the seventh round of confluence, there was 2 only a minimal increase in saturation density. 3 Another important aspect of these 4 observations, contrary again to what you might expect 5 -- You might think the cells which are becoming 6 transformed would certainly grow more at high density. 7 If you would grow them at low density, would they grow 8 any faster? Quite the opposite is true. 9 Actually, at low density these cells grow 10 slower and slower, which is another indication that 11 what confluence -- extended confluence is doing to 12 these cells is damaging them, and apparently damaging 13 their DNA. The problem is that occasionally in some 14 of the cells, that particular type of damage results 15 in transformation. 16 Now if you take the original line of NIH 17 3T3 cells and clone it out, what you see here is 18 three different clones. We've done this with very 19 large numbers of clones, but these are illustrative of 20 the heterogeneity you get within any single 21 population. 22 So this clone 1A actually started to show 23 even in the first round of confluence very tiny dense 24 foci. By the second round, it was showing these huge 25 foci. Then we had to dilute it out, and it continued 38 1 to produce large foci. 2 This clone of cells transformed more 3 gradually, and here you can see progressive 4 transformation within a clone. At first, light foci, 5 broad foci are produced. Then denser foci begin to 6 show up, and finally you get foci with this clone that 7 are just about the same roughly as this clone. 8 You take clone 4B here, and in spite of 9 many rounds of confluence, here you seem to get one 10 light focus, but in the sister dish which was being 11 transferred and not being fixed, apparently were no 12 foci. Even after five rounds of confluence, there's 13 only a minimal amount of transformation. 14 So what this means is every time you are 15 working with a mixed population or an uncloned 16 population, it's always a heterogeneous combination of 17 clones that behave differently from one another in 18 transformation. 19 Okay. Now I want to talk about one other 20 point. I tried to emphasize the point that an 21 important aspect in transformation, both of primary 22 fibroblast or at least explants directly from the 23 animal, and in established cell lines is high density, 24 is contact inhibition of the cells. 25 I might mention parenthetically, it's now 39 1 been clearly established in bacteria, which is always 2 our great sounding board for genetic change -- Warner 3 Arbor who won a Nobel prize working with bacteria that 4 led to very important discoveries found that, if he 5 left a culture, several cultures of bacteria sitting 6 around for 17 years so they hadn't grown at all during 7 that time except the first day that they were sitting 8 there, that there were enormous chromosomal 9 rearrangements that occurred even in a single 10 chromosome of bacterium. 11 So cells that are being starved, in 12 effect, are great candidates for genetic change, both 13 in bacteria and apparently in animal cells. 14 Now what I'm going to illustrate in this 15 last pair of slides is something that came as a great 16 surprise to me. Having worked with cells for about 50 17 years now, I thought it was pretty hard to surprise 18 me, but these cells are pretty clever. 19 It was this. We were working with this 20 subline of NIH 3T3 cells which was very difficult to 21 transform by keeping them at confluence, and we 22 decided, well, we ought to look at what happens with 23 clones of those cells to see how heterogeneous they 24 are. 25 What we found was quite a shock to us. 40 1 Here we have two clones -- they're representative 2 clones, clone 1A of this resistant line and clone 2E. 3 Then the parental population, which we're calling 4 large A prime here. 5 Again, we go through one round of 6 confluence, and then this is the fourth round of 7 confluence and the sixth round of confluence. The 8 shock was that, while the parental culture -- which 9 is, after all, made up of all of these clones -- was 10 exhibiting very little, if any, transformation or the 11 barest minimum, the clones were showing a lot of 12 transformation. 13 So what does this mean? I won't go into 14 this point here. So we quantitated that on the next 15 slide. This is the last slide. And we made a scale 16 of transformation, which is shown here. 17 So none of them transformed to the extent 18 of producing these really thick foci, but they would 19 produce distinguishable foci, and they would produce 20 foci even on a background of nontransformed cells. 21 Just look at this first panel here. We 22 don't have to go into the second one. It's sort of a 23 cumulative observation of relatively heavily 24 transformed cells -- that is, this type here -- or an 25 accumulation of those plus more moderately transformed 41 1 ones -- these here -- or more moderately and lightly 2 transformed ones. 3 So the combination of all of them after 4 five rounds of confluence had involved about 80 5 percent of the clones that we had. At that time, 6 shown on the scale at the top, the parental culture 7 had shown no transformation at all, in spite of the 8 fact that it was made up of thousands of those clones. 9 So could I have the lights, please? So 10 the right panel is a repetition of the same 11 experiment. What that says to us, that when you get 12 a heterogeneous mixture of clones from the parental 13 population, there is some kind of mutual protection 14 against transformation. 15 We later learned that back in 1981 George 16 Post had found something very similar to this with 17 melanoma cells in the mouse, that parental populations 18 retained their consistency of behavior; whereas, the 19 clones obtained from those populations underwent a 20 great deal of variation. 21 What we have to think about in that case 22 is what that means for progression in the animals, 23 because the general model that Henry Pitot was 24 presenting to you, and general model that Wallace 25 Clarke, the great melanoma specialist, has emphasized, 42 1 that in many systems in humans, including the 2 formation of melanomas, I think, of liver cancers, of 3 colorectal cancers, of skin cancers, all seem to go 4 through a sequence, let's say, -- colorectal is 5 probably the best known these days -- of the formation 6 of foci of cells developing into polyps, into 7 adenomas, all of which are of monoclonal origin. 8 It's from those monoclonal, benign tumors, 9 whether they are moles or adenomas or warts or, in the 10 case of the liver cancer models in animals, these 11 altered hepatic foci, that the next step is most 12 likely to occur. 13 So that's a new parameter that we have to 14 deal with where we get an association between the 15 changes in the cell culture reflecting, we think, the 16 effect of clonal expansion in the organism. Thank 17 you. 18 (APPLAUSE.) 19 DR. ONIONS: Harry, I think you've proved 20 that you don't need much warning to -- 21 CO-CHAIRMAN RUBIN: Oh, I didn't go over. 22 That's a first. 23 DR. ONIONS: Is there time for one or two 24 questions? 25 DR. COFFIN: John Coffin, Tufts. What is 43 1 known, if anything, about the molecular changes that 2 are associated with these different colony 3 morphologies, for example? 4 CO-CHAIRMAN RUBIN: If you're asking me 5 what have we done, nothing. 6 DR. COFFIN: Well, what is known 7 elsewhere? 8 CO-CHAIRMAN RUBIN: Well, let me tell you 9 something negative. This is work Stuart Ansen did 10 with the NIH 3T3 cells before we ever got them. 11 So he encountered this spontaneous 12 transformation as well. So he tested the spontaneous 13 transformation by extracting DNA and doing the 14 classical transfection experiment into nontransformed 15 NIH 3T3 cells. 16 He found essentially none of them were -- 17 the DNA of none of them was able to transform more 18 rapidly than as spontaneous transformation went on. 19 So there's no indicating, at least, that the Harvey 20 ras gene had mutated there. 21 What we know from classical genetics of 22 cells in culture, actually, was done with other cell 23 lines that was independent of transformation is that 24 when you get chromosome rearrangements or deletions, 25 like the deletions that you get in loss of 44 1 heterozygosity where you lose real chunks of the 2 genome, is you end up with cells that will grow slower 3 than the original cell. 4 If you get point mutations, it's very rare 5 that that slows down the rate of multiplication, at 6 least in the thymidine kinase gene, which has been 7 looked at. 8 So we think that the correlation that we 9 see of reduced growth rate at low density of the 10 transformed cells parallels the classical genetic 11 findings, and it's likely that what we're seeing is a 12 lot of chromosome rearrangements and deletions in the 13 cell, but we don't have any proof of it. 14 DR. COFFIN: If I may raise another 15 question. 16 CO-CHAIRMAN MYERS: Make it short. 17 DR. COFFIN: Yes. Can you separate the 18 slow growth property from the transformation property 19 by forcing rapid passages to transform cells, for 20 example? 21 CO-CHAIRMAN RUBIN: Yes. Well, you always 22 then select for faster growers, and it varies with 23 whatever population you use. In one population you 24 can select from the faster growers, and they turned 25 out to be less transformed, and in another population 45 1 you don't. But you can do it. 2 Not all cells that grow slower are 3 transformed. That seems to be a more general finding 4 that involves the whole population. It's only a 5 subset of those, presumably particular chromosomal 6 rearrangements, deletions, etcetera, that result in 7 transformation. But a large proportion of those 8 changes result in a slowing down of the growth of the 9 cell at low density. Okay? Thank you. 10 CO-CHAIRMAN MYERS: Thank you. Our next 11 paper is the transformation by DNA viral oncogenes by 12 Dr. Alex van der Eb from Leiden University. 13 DR. VAN DER EB: So as you have heard last 14 night and you will hear in the coming thoughts during 15 this meeting, diploid human cells have a finite life 16 span in vitro. They divide a certain number of times 17 and then they stop dividing. 18 This property limits to some extent their 19 usefulness for the production of viral vaccines or 20 production of viral factors for gene therapy. 21 Now cell lines, continuous cell lines, are 22 immortal and are, therefore, more suitable in certain 23 aspects. However, there may be certain risks 24 associated with the use of continuous cell lines. 25 Therefore, it would be helpful if we would be able to 46 1 immortalize diploid human cells without transforming 2 them. 3 So far the only means of immortalizing 4 diploid human cells is by -- reproducibly 5 immortalizing is by transforming them with a DNA 6 virus. Whereas spontaneous transformation or 7 spontaneous cancer is a multi-step process which 8 requires a large number of different gene mutations or 9 changes or alterations in genes that accumulate over 10 the years, transformation by DNA virus apparently 11 seems to be a one-step event. 12 This is due to the fact that the viral 13 transforming genes are -- or gene is a multifunctional 14 protein that alters simultaneously a number of 15 different regulatory pathways in the cells. 16 I would like to discuss briefly 17 transformation and immortalization by DNA tumor 18 viruses, and particularly focus on SV40, an adenol 19 virus, and also say a few words about HPV, but that 20 will be more extensively dealt with by Dr. McDougall 21 later during this meeting. 22 If I can have the first slide, please. 23 This slide shows the SV40 large T antigen, 24 which is the major transforming gene of the SV40 25 virus, and it also shows the three main transforming 47 1 domains in this gene. 2 On the righthand side you see the P53 3 binding sites which overlaps with ATPase binding site, 4 which is very important for transformation by SV40. 5 A second site is more to the left, and that coincides 6 with CR1, although CR2, the conserved regions 1 and 2, 7 also is important. That is a site that's responsible 8 for binding of the RB protein and the RB family 9 proteins as well as the coactivator B300 or CBP. 10 In addition -- So this is the second 11 important part for transformation, and in addition a 12 third part is the N terminus which is the DNAJ-like 13 protein -- like domain which resembles the DNAJ 14 proteins of E. coli that have an important role in 15 molecular chaperons for -- in conjunction with HSB 16 proteins. 17 How the DNAJ and terminus of SV40 18 contribute to transformation is still unclear, but it 19 is clear that P53 binding and inactivation of the RB 20 proteins or P300 really very importantly disrupts the 21 main growth control pathways in the cell. 22 Adenovirus transforms basically in a 23 rather similar way, and the next slide shows the 24 adenovirus E1A gene, which is the major transforming 25 gene of adenovirus. As you can see here, there are 48 1 two again conserved regions here. One is CR2 which 2 are important for transformation. 3 The third region, CR3, is not important 4 for transformation, but another region, the N terminus 5 which is not conserved among adenoviruses, is on the 6 other hand again essential for transformation. 7 So the N terminus, CR1 and CR2 are 8 important for transformation of primary cells, as well 9 as association to cellular proteins. The N terminus 10 and CR1 are responsible for binding to P300 and CBP as 11 well as the P400 coactivator which is similar to -- 12 more or less similar to P300, but as CR2 and CR1 are 13 important for binding to the retinoblastoma protein 14 and the related pocket proteins, so again basically in 15 a rather similar way compared to the SV40. 16 Now the adenovirus E1A gene confers such 17 a strong growth promoting effect on cells that these 18 cells apparently -- certainly, if one has primary 19 cells in which E1A is expressed to a reasonable 20 extent, that these cells respond by activating their 21 P53, and this will lead to either growth arrest or 22 apoptosis. 23 So to counteract this growth arrest or 24 apoptosis, the E1B region is necessary. So you see 25 here the E1A region with the two proteins -- with the 49 1 two RNAs of proteins that I just showed you with CR1 2 and CR2, as well as the N terminus which is located 3 here. But the E1B region calls for two proteins, a 4 large protein and a small protein, and these are 5 essential for neutralizing the apoptotic and cell 6 cycle arrest activity of the E1A region. 7 So E1A and E1B are needed in order to 8 transform cells, because of the effect of E1A on P53. 9 Therefore, it is almost impossible to 10 transform -- obtain E1A transformed cells alone, and 11 very few cells -- and those are rodent cells that are 12 obtained that are transformed by E1A alone -- express 13 E1A to very low levels. 14 HPV, the human papilloma viruses, 15 basically transform again in a similar way. Next 16 slide, please. 17 The two transforming genes or papilloma 18 viruses are E6 and E7. E6 targets P53 and causes its 19 degradation, but E7 again targets the other growth 20 regulatory proteins, PRB, the RB protein and its 21 family members as well as probably P300, but that 22 we'll hear later in more detail. So they resemble 23 again the adenoviruses which have basically the same 24 properties. 25 Now not all these cells react in the same 50 1 way to these transforming genes or viruses. Next 2 slide, please. Here I show you SV40, how SV40 3 transforms cells, human fibroblasts, diploid 4 fibroblasts or epithelial cells and keratinocytes. 5 Fibroblasts are very efficiently 6 transformed by SV40. Immortalization, however, is 7 very rare, at least 10-7 events per senescent cell, 8 and there is a pronounced extended life span. 9 Epithelial cells, on the other hand -- not 10 on the other hand, but epithelial cells more or less 11 at the same weight are responding more or less at the 12 same rate as SV40. There is morphological 13 transformation. There is also rare immortalization. 14 It's not completely clear how much their extended life 15 span is, but I believe that there is an extended life 16 span again in these epithelial cells. 17 The situation is different for adenovirus 18 E1. Forget E1B. This is an old slide where I thought 19 that I had E1A transformed human cells, but this is 20 not the case. 21 Fibroblasts or epithelial cells appear to 22 be surprisingly resistant to transformation by 23 adenovirus. Frank Rijm in 1974 has done many attempts 24 to transform human embryonic retinal cells with DNA of 25 adenovirus, and he was initially completely 51 1 unsuccessful until he found one clone, and this clone 2 gave rise to the well known 293 cell, and that is 3 really the only clone that he has seen transfected 4 with adenovirus. 5 Fibroblasts similarly do not show 6 transformation, and I mean now transformation 7 according to the focus assay. You can introduce by 8 transfection the adenovirus E1 genes into human cells 9 where they are expressed, but we have the impression 10 that expression is lost after a while. There is no 11 morphological alteration. 12 These transformed cells -- and I'm talking 13 now about 293 cells -- They immortalize probably after 14 short crisis periods. Efficiency of immortalization 15 is not so clear, because there's only, as far as I 16 know, only one example. It's not so clear whether 17 there is an extended life span. 18 So we then switch to embryonic retinal 19 cells, because we had heard from work of Phil Denimore 20 and others that neural cells, cells of neural origin, 21 could be transformed more easily also in the rate 22 system, the rodent system; and we switched to 23 embryonic -- primary embryonic retinal cells, and they 24 can be transformed quite efficiently by adenovirus E1, 25 although still at relatively low numbers compared to 52 1 transformation of rat kidney cells, for example, but 2 it is reproducible. 3 Also there is a high frequency apparently 4 of immortalization, and there is no apparent crisis, 5 which is rather surprising. 6 I will not say much about HPV E6 and E7. 7 There is little interaction with fibroblast. There 8 may be reported very rare immortalizations by E6, E7, 9 but epithelial cells at keratinocytes are immortalized 10 more frequently, and apparently there is no distinct 11 crisis. 12 So transformation of human cells by SV40 13 has been studied most extensively. So I will briefly 14 turn now -- go back now to SV40. 15 If diploids and diploid fibroblasts -- If 16 diploid fibroblasts are transformed by SV40, then a 17 number of changes occur, and these are depicted here 18 in this slide. 19 The normal diploic fibroblasts grow to a 20 certain saturation density during a number of 21 population doublings, and then they stop dividing, and 22 an irreversible arrest, growth arrest, which is called 23 senescence, and it has been mentioned before. 24 No immortalization will ever occur, as far 25 as I know, from these senescent cells. They do not 53 1 die, but they sit there just sometimes for many 2 months. 3 If you transform the culture at this stage 4 for a -- with SV40, then the saturation density 5 increases, and the transformed cells seem to ignore 6 this senescence arrest phase and just go on for a 7 number of -- for a large number of passages sometimes, 8 which causes an extended life span. However, in the 9 end the cell -- the whole protein will die and enter 10 into so called crisis where the cells -- the 11 transformed cells really die, unlike the situation in 12 senescence. 13 Only in a very few cases, very rare cases, 14 an immortalization will occur after a shorter or 15 longer period, and this will etherize to a cell which 16 is the same as the SV40 transformed cell before the 17 crisis, but now they are immortal and can grow 18 indefinitely. 19 So what is the basis of the appearance of 20 the senescence and the crisis? That has to do with 21 the telomeres. In contrast to the germ cells which 22 have telomeres active in these cells, the normal 23 somatic cells are telomeres repressed and have no 24 telomeres activity. 25 So during population doublings, t he 54 1 telomeres are lost, become smaller and smaller until 2 a certain stage is reached, which may be about two- 3 thirds of the length of the telomeres, approximately 4 two-thirds, but certainly not all telomeres have been 5 used up, and at that stage the cells start senescence. 6 This senescence is called M1 or mortality 7 phase 1. If you, in contrast, inform with SV40, again 8 the M1 senescence is ignored, and the cells continue 9 dividing, and also the telomeres become shorter and 10 shorter until they are almost completely disappeared. 11 That coincides with the crisis which is 12 also called mortality stage 2. So the M2. If there 13 is no immortalization, the cells -- all cells will die 14 here. Immortalized cells have now in most cases 15 active telomerase. 16 So this is then the extended life span 17 between senescence and crisis, and it's also this 18 period which is characterized by the onset of 19 chromosomal abnormalities. So there are -- and that 20 effect already was also seen in the normal cells. 21 When they approach crisis, there is chromosomal 22 rearrangements. There are dicentrics formed, and so 23 on. 24 That is probably due to the fact that 25 during senescence the P53 pathway, P53-P21 pathway, is 55 1 activated for some reason. So what happens during 2 senescence is that at a certain critical length of the 3 telomeres -- we don't know exactly what the signal is 4 -- will cause activation of preexisting P53. There is 5 probably not more transcription of the P53 gene, and 6 this P53 causes accumulation of active P53 which 7 causes activation of transcription of the P21 gene, 8 the WAF-1 gene, which in turn is an inhibitor of 9 cyclin/CDK and inhibits the cell cycle. So this 10 triggers apparently a certain minimum length of 11 telomeres. 12 Now catalytic subunits of the human 13 telomerase has been isolated. The question can be 14 raised would it be possible just to introduce 15 telomerase in cells and immortalize them in diploid 16 cells and immortalize them in that way? 17 Indeed, there has been reports that this 18 is indeed the case. Dr. Hayflick has already 19 mentioned an example yesterday. Also Dr. McDougall 20 will probably talk about this later, but as Botnar and 21 coworkers have also found, that fibroblasts as well as 22 retinal epithelial cells become apparently immortal 23 when the human telomerase catalytic subunit is 24 introduced into the cells. 25 Rather, Piono et al. showed that mammary 56 1 epithelial cells as well as keratinocytes do not 2 immortalize just with only introduction of the 3 telomerase, and this works that immortalization is 4 obtained when, in addition, the E7 gene is added to 5 the cells. That, I think, will be discussed later in 6 much more detail by Dr. McDougall. 7 So this means that, if you inactivate the 8 RB, then that would lead to immortalization. This may 9 be due to the fact that there is apparently a third 10 type of senescence, and that third type of senescence 11 is called N0 and occurs usually before M1. 12 It is known that, unlike fibroblasts that 13 can be found a long time, 50 to 60, 70 or sometimes 80 14 population doublings, depending on the cell strain and 15 also depending on the lab, I have the impression, that 16 many other cells when taking a tissue culture form 17 human tissues will not grow very long, and after ten 18 or maybe 20 passages they just stop dividing. This is 19 long before the telomeres have reached the size, the 20 minimum size which corresponds to the senescence, M1 21 senescence. 22 Now Weinberg has suggested that this M0 23 state is due to a kind of physiological stress and is 24 caused somehow by suboptimal growth conditions that 25 occur in vitro. So introduction of telomerase in 57 1 cells that still have to undergo an M0 may not work, 2 because M0 is cause for physiological stress and has 3 nothing to do with the size of the telomeres. On the 4 other hand, cells that have no M0 but only an M1, 5 there it may work, but this is all theory, and we'll 6 see if it is really true. 7 So the last slide then summarizes here for 8 the two senescent stages that occurred in certain 9 cells, including the mammary epithelial cells but 10 probably many more cells, and that is that M0 occurs 11 after -- in the case of mammary epithelial cells, 12 after about 20 population doublings, and may be 13 activated by physiological stress. 14 It's controlled by the RB P60 pathway, and 15 can be bypassed by HPV. It must be E7. This must be 16 E7, right? It's wrong. Jim, it's wrong. It is E7. 17 I'm sorry. 18 This was made just before I left, and 19 there are more mistakes here, as you can see. The 20 computer did a trick and did not exactly what it was 21 supposed to do. 22 Anyway, it's bypassed by HPV E7. One, on 23 the other hand, may be triggered by certain extent of 24 telomere shortening, and which is actually the mitotic 25 clock, and is controlled by the P50-CP21 pathway, and 58 1 that is bypassed by HPV E6. 2 M2 is called crisis leading to cell death 3 and is probably caused by extensive telomere 4 shortening. It has nothing to do with all these 5 things. 6 So I think I'll stop here. Thank you. 7 (APPLAUSE.) 8 CO-CHAIRMAN RUBIN: We can have a few 9 questions now. 10 CO-CHAIRMAN MYERS: Could we have the 11 lights, please. 12 CO-CHAIRMAN RUBIN: Lights, please. 13 Could we have the lights? Yes, thank you. 14 DR. HAYFLICK: Hayflick, UCSF. Unless I 15 misunderstood your introductory remarks, there are 16 indeed two other ways in which one can transform 17 normal human fibroblasts to an immortal cell 18 population. Namely, we have done this with exposure 19 to cobalt 60 radiation, producing a cell line called 20 SUS M1. 21 It's also been done with chemical 22 carcinogens, producing a cell line called KMS T6, in 23 addition to H terp which you've mentioned. I also 24 should mention that, contrary to popular belief, there 25 are several publications in the open scientific 59 1 literature reporting spontaneous transformation of 2 normal human cells. 3 For those of you who would like those 4 references, please contact me, and I'll be happy to 5 supply you with them. Thanks. 6 DR. VAN DER EB: Thank you. Yes, you can 7 immortalize -- transform or immortalize human cells 8 also by radiation or chemical carcinogens. I think 9 this is often quite difficult. You have to expose the 10 cells many times, as far as I know at least, to the 11 chemicals or to the radiation, and it is a lot more 12 easy just to take SV 40 large-T for fibroblast. But 13 you are completely right, yes. 14 DR. HAYFLICK: Well, the reason I 15 emphasize that -- 16 CO-CHAIRMAN RUBIN: Can you get to the 17 microphone? 18 DR. HAYFLICK: The reason I thought it was 19 worth emphasizing is that the thrust of this meeting 20 is the development or the consideration of immortal 21 cell populations other than those that might be 22 currently used for vaccine production. 23 Two of the obvious choices, if one wants 24 to avoid the overt introduction of viruses or their 25 fragments into cells, is by the use of radiation or 60 1 chemical carcinogens. 2 DR. VAN DER EB: I know from cells that 3 are spontaneously immortalized at least and also 4 clearly transformed that they have lost quite a lot of 5 things like P53 and also the RB pathway. So what you 6 do then is basically the same as what occurs or 7 happens in spontaneous transformation, I think. 8 CO-CHAIRMAN RUBIN: One more question. 9 DR. BERKOWER: Ira Berkower from the FDA. 10 One concept that we've been struggling 11 with is the notion of cells that are more malignant 12 and less malignant, more transforming and less 13 transforming, with the idea that if vaccines could be 14 made in less transformed cells, that will be safer. 15 Does that make any sense in terms of these 16 molecular mechanisms that you're studying, that there 17 would be a more transformed phenotype? 18 DR. VAN DER EB: What is a more 19 transformed phenotype? 20 DR. BERKOWER: Or a more malignable 21 phenotype? 22 DR. VAN DER EB: This is difficult, I 23 think. If I see it now -- If I look at the 24 literature, then I have the impression that if you 25 take immortalization as the important final step that 61 1 you want to reach, that then HPV genes may be more 2 suitable. 3 I have the impression that they are -- The 4 cells look less transformed at least than in the case 5 of SV 40, for example, but also in the case of adeno. 6 But there's very little I can say to that, I think. 7 It's so difficult. What is immortal? Which cell is 8 more transformed than the other? 9 You would have to compare things like 10 tumorigenicity. I should say that cells transformed 11 by viruses are often not yet immortalized, I showed 12 you, but also if they are immoral, they are not yet 13 immediately tumorigenic. 14 So maybe tumorigenicity would be more 15 important than just looking at the cells, how they 16 grow. 17 CO-CHAIRMAN RUBIN: Okay. We'll have to 18 move on to the next speaker, who is Dr. James 19 McDougall from the Fred Hutchinson Cancer Research 20 Center at Seattle, Washington. 21 DR. McDOUGALL: Well, as Alex has already 22 pointed out, the two things that I would rather talk 23 about are human papilloma virus and telomerase 24 immortalization. But when Andy called me -- Andy 25 Lewis called me -- he said that somebody else had 62 1 dropped out of talking about hit and run, and so I got 2 lumbered with the job. 3 It's interesting. It's interesting, but 4 it's controversial, without doubt. I'll just put my 5 first slide on. In the olden days, and there are a 6 few of us that were there in the olden days, one of 7 the most interesting studies that was going on in 8 terms of looking at how viruses might transform cells, 9 and particularly how they might contribute to human 10 cancer, was the belief that herpes viruses might very 11 well be responsible for, for example, cervical cancer. 12 There were good reasons for believing 13 this. For example, serology gave very good evidence 14 that women with cervical cancer had high levels of 15 antibodies for herpes simplex, and it became clear 16 that this was herpes simplex Type 2. 17 So as time went on, it looked reasonable 18 to try and sort out whether or not this virus had 19 fragments, had subgenomic fragments that could 20 actually transform cells. That was one of the studies 21 we concentrated on in the early 1980s. 22 The studies were carried out mostly in rat 23 cells or hamster cells, and lo and behold, we were 24 very pleased to find that we could actually transform 25 these cells and, if we grew these cells up from an 63 1 initial transformation by exposing them to herpes 2 virus DNA and to fragments of herpes virus DNA -- and 3 I'll show you the specific fragments in a minute -- 4 that those cells would then produce tumors in their 5 rat or mouse host. 6 So this looked very good, and we felt we 7 were on the track of what might be the key genes that 8 were involved in the development of an important human 9 tumor, in this case cervical carcinoma. 10 So over the years there were a lot of 11 examples of how herpes viruses might perhaps have a 12 hit and run type of effect rather than a similar 13 situation that we see with the adeno viruses and SV 40 14 and human papilloma viruses where it's clear that very 15 specific fragments persist in the cells that are 16 either immortalized or transformed by those. 17 I just put up a series of papers that all 18 refer to the hit and run phenomenon related to herpes 19 viruses. 20 So if we look at the herpes virus genome - 21 - and here is the genome along the top here, and what 22 I really want to bring your attention to are these 23 fragments here that map in this region of the genome. 24 We initially show that this one called 25 morphological transforming region 1 would carry out 64 1 that transformation of both rat and mouse cells at a 2 reasonably efficiently level, almost at the same level 3 as SV 40 would transform these cells. 4 Subsequently, there were other studies 5 which identified two other regions which also were 6 capable of transforming, in this case, rat cells -- in 7 both of these cases, rat cells. 8 Now that was somewhat surprising in that 9 there would be more than one region of this genome 10 that would actually transform or immortalize cells 11 but, of course, this is a very large genome anyway. 12 So that that might not be too unreasonable. 13 So we continued with a large series of 14 experiments to try and track down exactly what the 15 region of the genome might be that was producing this 16 transformation effect. Unfortunately, we were able to 17 make smaller and smaller fragments of this MTR1 region 18 until we got down to a region of that DNA which, in 19 fact, was too small to have its own open reading 20 frame, and immediately one has got to start worrying. 21 If there's not an open reading frame, what 22 in fact, is affecting the cell? Is there a protein 23 being produced? In this case, clearly not. So we 24 began to worry what the mechanism might be in this 25 situation. 65 1 So one of the mechanisms we thought about 2 was that there might be just random insertion of 3 sequences into the host cell genome. So we looked at 4 these MTR1 fragments, and particularly the MTR1 5 fragment of herpes simplex virus type 2, and by 6 sequencing that we were able to show there was a 7 structure in there very similar to a bacterial 8 insertion sequence in its structure, and felt that it 9 was highly likely that this was perhaps integrating 10 and then excising randomly from cells, and this might 11 in fact have a mutagenic effect upon the cell. 12 Now interestingly, we were able to take 13 HSV1 and take a similar -- the identical region from 14 the genome of HSV1 and look at the sequence and 15 structure of that. 16 What we found was that that structure, 17 although for half of this loop, is pretty well 18 identical. The other loop did not pair sufficiently 19 to make an insertion sequence-like structure. That 20 region of HSV1 will not transform cells, whereas this 21 one will transform cells at a reasonable efficiency. 22 So this seemed to be a reasonable 23 suggestion, and perhaps we should look at, in that 24 case, mutagenesis by this fragment, the MTR1 fragment. 25 Now while we were doing these experiments, 66 1 we also realized that there was another herpes virus 2 that we should look at, and that was the human 3 cytomegalovirus. The genome is shown up here. 4 Again, the region that we were 5 particularly interested in was this early region, the 6 immediate early region of the genome, which is 7 transcribed very early after infection by the virus. 8 So one of the experiments we did was to try and work 9 out whether we could actually, first of all, transform 10 cells with this region, and the answer was yes, and 11 this was repeated in two or three different labs. 12 Then we wanted to find out, if we used 13 very sensitive PCR experiments, could we in fact 14 identify regions of that CMV immediate early region 15 present in cells that were transformed by CMV. By 16 going through two rounds of PCR to pick this up, we 17 were able to in fact identify a fragment from CMV that 18 persisted in some of the cell lines, but really in a 19 very low percentage. 20 Most of the cell lines, just like the 21 herpes cell lines, had lost the DNA that we 22 transfected into them. 23 Again, when we looked at the transforming 24 region of CMV and made lots of deletion mutants of 25 that region -- so we started off with a region here 67 1 which was quite a reasonable sized piece of DNA, 2.9 2 kilobases, and just made deletions of this all the way 3 down. 4 Again, to our surprise, we could get right 5 down to here, and we were still capable of 6 transforming cells with that region. So again, this 7 suggested to us that maybe we were looking at the same 8 situation with CMV that we had seen with herpes 9 simplex type 1. 10 In fact, in that very small region of CMV 11 DNA, again you find this same structure present, and 12 in most cases that structure was lost from the cells, 13 that it had been put in by transfection, and we 14 actually carried out not only transfection with DNA 15 but made retroviral constructs and put that region in, 16 could again show that we could transform cells with 17 that region, but that it was lost from the cells on 18 passage and in tumors produced by those cells. 19 So the DNA clearly had been present there, 20 but had been lost fairly on in the process, despite 21 the fact that these cells are capable of making 22 tumors. 23 Obviously, one of the ways to look at this 24 is that maybe these viral DNAs are acting in some way 25 as mutagens, and there was certainly a history of the 68 1 fact that herpes virus infection would produce damage 2 to chromosomes. 3 I must qualify this, of course, by saying 4 that in general herpes virus is lytic to cells, and so 5 you don't expect cells to survive. However, in most 6 herpes virus populations there's a lot of defective 7 virus present, which might very well allow some cells 8 to survive in an infection. 9 We could also show experiments done in Zur 10 Hausen's lab in Germany that you could find mutations 11 in host cell genes after transfection of herpes virus 12 DNA-N. So clearly the virus is acting as a mutant, 13 and this is just a demonstration of some thioguanine 14 resistance with a number of these herpes virus 15 transforming regions, showing that there is in fact a 16 mutation frequency that is measurable in these 17 experiments. 18 So this immediate early region in CMV 19 became a region of interest to us, but we really 20 stopped those experiments because they had been 21 started out of an interest in cervical carcinoma, as 22 I initially said. We stopped those, because human 23 papilloma viruses raised their beautiful heads, and 24 obviously we now know that this is the prime 25 initiating cause of anogenital carcinomas. 69 1 So we really stopped looking at this, but 2 about ten years later it gave us a lot of pleasure 3 when Tom Shent suddenly discovered the idea of hit and 4 run, and reported in the literature that they've used 5 immediate early genes of CMV and had essentially come 6 to the same answer, that these regions would produce 7 increased foci, increased transformation of cells, but 8 in fact the DNA was then lost from the cells, and so 9 this again looked like a hit and run phenomenon. 10 I'll just go past this one for the moment. 11 And the type of experiment that came from Tom Shent's 12 lab, which were published in 1997 in PNAS, was that we 13 heard from Alex van der Eb already about the effects 14 of adenovirus E1A, and what they showed was that if 15 you combined E1A with the CMV early regions, E1 or 16 preferable the E1 and E2, you could get a much higher 17 number of foci produced in those transfected cells 18 than if you had just E1A alone, and they showed this 19 in a number of different experiments. 20 So clearly, the CMV region, which was not 21 then retained in these cells, was nevertheless 22 contributing to the transformation of the cells, the 23 same sort of picture that we've seen earlier. 24 The other picture that they also found was 25 that there was an increased mutation frequency, 70 1 looking at APRT in this case and HPRT in this case, 2 that again if you put these immediate early genes in - 3 - and you can do this on their own or in combination 4 with E1A again -- you could demonstrate that there 5 clearly was a mutation frequency resulting from the 6 transfection of this DNA into, in this case, rat cells 7 or mouse cells. 8 So, clearly, there's no question that 9 these small fragments of viral DNA that can transform 10 cells most likely act as mutagens and are lost from 11 the cells upon passage of these cells, but the cells 12 retain their transformed phenotype. That's probably 13 the most important thing to remember out of this. 14 So in conclusion for this talk, I'd just 15 like to make these points: That it's clear that 16 subgenomic fragments of herpes simplex virus and 17 cytomegalovirus can transform cells in vitro. 18 Now these experiment are not conducted in 19 human cells, but in rodent cells, as I've already 20 said. It's clear from a number of experiments that 21 have been carried out now in certainly more than one 22 or two labs that that virus DNA, which is capable of 23 transformation, may not persist over the long term in 24 the transformed cells. 25 Although you can detect early on -- If you 71 1 put a complete open reading frame in, you can detect 2 the expression of viral proteins. You only see these 3 early, and they do not persist, again suggesting that 4 there is not only a loss of that sequence but perhaps 5 a loss of sequence over time out of these transformed 6 cells. 7 Lastly, it's clear that these viral 8 transforming regions can indeed be mutagenic, shown in 9 a number of experiments both by our labs and by Tom 10 Shent's labs. So that still leaves the possibility 11 that herpes viruses can be responsible for some -- the 12 development of some tumors, and there are still 13 experiments being looked at and some that have been 14 published showing that, even in cells that are 15 immortalized by human papilloma virus, the presence of 16 some of these morphological transforming regions 17 results in a more rapid conversion of those cells to 18 tumorigenicity. 19 So hit and run may seem an old phenomenon, 20 but it's not dead. Thank you. 21 (APPLAUSE.) 22 CO-CHAIRMAN RUBIN: Questions? 23 DR. KUNG: Hsing-Jien Kung from UC-Davis. 24 I'm just wondering, do you have some 25 hypothesis whether this is due to insertion of 72 1 mutagenesis or due to the fact it knocked out P53, 2 make the genome just generally unstable; therefore, it 3 takes its own course. 4 DR. McDOUGALL: Well, I don't think P53 is 5 not found in these cells. So I still believe the 6 hypothesis, that what you're seeing is integration of 7 this and then a random excision of it is probably the 8 reason, and that the integration is random, the 9 excision is random, and probably the way to get at 10 this, if we wanted to go back to these experiments, is 11 really to do much more detailed sequencing of these 12 regions and look at the flanking areas on any DNA that 13 might still be there. 14 It is very clear that this DNA is 15 generally lost from these cells. 16 DR. KUNG: But you do have some residual, 17 right, in some of the recent experiments? 18 DR. McDOUGALL: Some of the experiments 19 with CNV there is some material that remains there, 20 and we could go back to that and ask those questions, 21 yes. 22 DR. FRIED: Did you -- 23 CO-CHAIRMAN RUBIN: Identify yourself. 24 DR. FRIED: Fried from ICRF. Do you take 25 the region of that secondary structure that you 73 1 pointed to, and did that stop the mutagenesis? 2 DR. McDOUGALL: No. What we did was just 3 to use the HSV1 region as a control. So we didn't use 4 APH -- We didn't do anything to the HSV2 or CMV 5 regions. We just used the HSV1, which is, as you see, 6 a tail. It does not produce a nice insertion 7 sequence. We used that as our control. 8 DR. FRIED: And in Shent's experiments, 9 were there any specificity to the type of mutations 10 that were formed, based changes or -- 11 DR. McDOUGALL: I confess, I don't 12 remember the answer. 13 CO-CHAIRMAN RUBIN: We go to the back now. 14 Someone has been waiting there. 15 DR. KRAUSE: Phil Krause, FDA. To your 16 knowledge, has anybody ever taken MQR regions and 17 injected them into nude mice or something like that 18 and looked for an in vivo tumorigenicity endpoint? 19 DR. McDOUGALL: Not to my knowledge, no. 20 DR. COFFIN: John Coffin. Stem loop 21 structures such as you drew must, of course, be very 22 common, you know, kind of DNAs, and actually -- I 23 mean, is this effect specific for herpes virus DNAs or 24 have they been studied so much more than almost any 25 other DNA of similar kinds or complexities, if you 74 1 take bacteria for HDNA or -- 2 DR. McDOUGALL: Well, in the bacterial 3 systems, right. I don't know of many studies in this 4 sort of viral system, and I don't know of any that 5 have been done very recently either. 6 DR. COFFIN: So, in fact, it could be that 7 all kinds of DNAs may carry regions -- 8 DR. McDOUGALL: I wouldn't be at all 9 surprised. 10 DR. COFFIN: And it's just the herpes 11 viruses that have been looked at. 12 DR. McDOUGALL: It just happens that that 13 was a good candidate. 14 CO-CHAIRMAN RUBIN: One more question, and 15 we'll have to move on. 16 DR. BROKER: Jim, has anyone taken 17 synthetic stem loop structures or is it specific to 18 the particular sequences that these various herpes 19 viruses that are produced? 20 DR. McDOUGALL: Well, in a way, that's the 21 same as John's question. I think the answer is no. 22 AT least, I don't know of any experiments like that. 23 As I say, we got hopelessly diverted by papilloma 24 virus. 25 CO-CHAIRMAN RUBIN: Could you please give 75 1 us your name? 2 DR. BROKER: Oh, Tom Broker, UAB. A good 3 choice, Jim. 4 CO-CHAIRMAN RUBIN: We'll have to go to 5 the next speaker. Thank you very much. 6 The next speaker is Stephen Baylin, Johns 7 Hopkins University School of Medicine, who will talk 8 about DNA methylation and epigenetic mechanisms of 9 carcinogenesis. 10 DR. BAYLIN: Thank you very much. I'd 11 like to thank the organizers for inviting me. As I 12 hear some of the talks going before, with respect to 13 the area our work is in and our group works on, I 14 think of another kind of change other than the classic 15 genetic changes in DNA that we've been hearing about, 16 which is also heritable and involves DNA probably 17 through chromatin structures which are variable. 18 In terms of contribution to oncogenesis 19 right now, I think one of the interesting aspects of 20 that which has emerged is that process as it regards 21 epigenetic gene silencing during all phases of tumor 22 progression. If I can have the first slide, please. 23 Now for those of you who don't work in 24 this particular area of epigenetics, let me just 25 quickly remind you that in the eukaryotic genome and 76 1 certainly in humans and rodents, essentially the only 2 place that we methylate our DNA is at the base 3 cytosine and only at those cytosines that occur 5 4 prime to a guanosine. 5 We talk about the CPG dinucleotide as the 6 substrate for this event. Now it's an event which is 7 catalyzed by what up until recently was thought to 8 just be one mammalian enzyme, certainly DNA methyl 9 transferase. 10 There are at least two others now that 11 have activity, and this is going to probably emerge as 12 a very important part of the story in tumorigenesis, 13 the interplay and the roles of these different DNA 14 methyl transferases. But they each catalyze a 15 reaction in which acidenosine methionine is used as 16 the methyl donor group, and the methyl group is placed 17 on the 5 position of the ring of cytosine. 18 The other important aspect of this for 19 tumorigenesis which is reduced to real simplicity here 20 in terms of what's being understood about methylation 21 and its role in the genome is that in the eukaryotic 22 genome the CPG dinucleotide has been drastically 23 reduced over evolution to a small fraction of its 24 predicted frequency. 25 So most of the genome has a reduced CPG 77 1 content. However, in most of the genome there is a 2 high percentage of those CPGs that actually get 3 methylated in rodents and humans. So we have heavy 4 methylation, as shown by the yellow dots, in most of 5 the genome. 6 This methylation may play various roles, 7 one of which may be to participate in chromatin that 8 is transcriptionally repressive for repeat elements in 9 areas of unwanted transcription in the bulk of the 10 genome. 11 Now the exception to that type of 12 distribution occurs in and around about half of the 13 genes in our genome where the CPG dinucleotide 14 frequency maintained at its predictive level, and 15 these are the so called by Adrian Byrd and others CPG 16 islands. 17 For most genes -- say, for inactive genes 18 on the X chromosome of the female and selected 19 silenced alleles of imprinted genes, these CPG islands 20 are maintained free of methylation, and this is 21 thought to be a permissive state. It doesn't matter 22 whether the gene is actively transcribed or not, but 23 it's a permissive state for active transcription of 24 the gene. 25 What's being more and more frequently 78 1 recognized in the cancer cell genome is an increasing 2 list of genes in which this balance is disrupted, such 3 that in the face of loss of methylation from large 4 regions of the genome other than the promoter regions, 5 CPG islands around an increasing list of genes 6 actually have become methylated. 7 Whether it's the cause or the result is 8 still being worked out. This correlates with a 9 chromatin organization around that gene which is 10 unfavorable for transcription of the gene. 11 Now the impact for this on carcinogenesis 12 and tumorigenesis, I think, can be looked at from the 13 types of genes where this has now been defined. For 14 those genes which are known tumor suppressor genes by 15 virtue of the fact that when they're mutated in the 16 germ line of families, those families have inherited 17 forms of cancer. 18 About half of those genes, as shown in the 19 yellow, in somatic forms of cancer, noninherited 20 cancer, have had some frequency now shown and well 21 worked out of this promoter region CPG 22 hypermethylation. 23 I would point to you that one of the most 24 frequent times when this is seen is in the cyclin DRB 25 pathway for the P16 gene, and now in a number of 79 1 experimental systems, including mammary epithelial 2 cells, loss of M0 that we just heard about, at least 3 experimentally and probably in several natural tumors, 4 may be in association with hypermethylation of this 5 cyclin dependent kinase inhibitor gene P16. 6 So it can be a very early event in 7 carcinogenesis and a very important one, as shown by 8 just that one gene involvement. Next slide, please. 9 Other genes other than the classic tumor 10 suppressor genes are emerging, and these are just a 11 few, and they play a role in various kinds of 12 processes, some for enzymes that guard against DNA 13 damage such as methyl transferase, 06NGMT GSP pi. 14 Recently, for the P53-like gene where 15 mutations of -- or P73 where mutations have not been 16 found, very frequent to see hypermethylation of this 17 gene in lymphomas. So an increasing list of genes 18 which can play a potentially important read in 19 oncogenesis. 20 Just to remind you, that in many of the 21 instances where this has been defined, or in several, 22 that gene might not have been recognized through 23 classic mutational changes in the coding region for a 24 role in tumors, because several of these genes such as 25 P73, in some tumors even P16 like colon cancer, don't 80 1 seem to be altered through mutational change but have 2 a high frequency of this promoter change. Next slide, 3 please. 4 Just to remind you again how frequent this 5 is, -- this is not to meant to be read -- this is a 6 survey of major human tumors where Jim Harmon in our 7 laboratory and others have been working out these 8 hypermethylation events, and just to show you that 9 virtually every kind of human neoplasm has multiple 10 genes, not every tumor in every patient but, for 11 example, in breast cancer, if you take the list of 12 genes here shown to be hypermethylated and examine a 13 series of breast cancers, you will find that every one 14 of those breast cancers has one or more of these genes 15 hypermethylated. This now includes in about a quarter 16 of them actually in the nonfamilial type of breast 17 cancer the BRCA1 gene. Next slide, please. 18 Now in terms of what this might mean for 19 therapy of cancers, the difference, of course, here 20 from coding region mutations is that theoretically, if 21 there are no base changes that inactivate the gene on 22 a permanent basis, this is a potentially reversible 23 situation, although indeed, if you look at it in 24 culture and you look at it over time, for the most 25 part it is a heritable event, the maintenance of this 81 1 kind of gene silencer. But it certainly has raised the 2 question can reactivation of these hypermethylated 3 genes serve as some sort of therapeutic target for 4 cancer. 5 This is not a new concept. There has been 6 a drug around for a long time -- next slide, please -- 7 Peter Jones and others introduced years ago when they 8 found that drugs and congeners of 5 azacytidine can 9 cause demethylation within the genome and reactive 10 genes. So this has been the drug that people have 11 concentrated on. 12 It's only more recently that, I think, 13 people are thinking about it in terms of specific gene 14 reactivation for events in cancer, like John talked 15 about. Various other series of molecules, antisense 16 again, so called DNA methyl transferase 1 and other 17 molecules directed at the major mammalian DNA methyl 18 transferase are being tried. 19 It must be remembered that these other DNA 20 methyl transferases that have recently been described 21 are probably going to also play important roles in 22 their separate genes with separate sequences, and so 23 that's going to have to be taken into account. Next 24 slide. 25 In our laboratory, it just shows you that 82 1 most of the genes where you see the CPG island 2 methylation in culture, you can achieve at least 3 partial reactivation of any or all of these genes 4 through administration of the drug 5 deoxy 5 azacytidine. Next slide. 6 And you can bring back functional events 7 in the cells by reactivating these genes. The classic 8 example has been the loss of the mismatch repair gene 9 hMLH1, which is probably the leading cause for the 10 microsatellite instability phenotype in colon cancer. 11 Most of those colon cancers have this epigenetic 12 change at the promoter of this gene. 13 If you throw different mismatches at 14 cultured cancer cells -- This is a Hela cell in Tom 15 Kunkle's lab. Actually, this is the 100 percent for 16 each of the three different bases in a Hela cell which 17 has effective repair. 18 Here's a hypermethylated colon cancer cell 19 before azacytidine and, if you give 5 azacytidine to 20 that cell, you can restore considerable mismatch 21 function -- repair function to the cell by 22 reactivating the mLH1 gene. Next slide. 23 In terms of manipulating the genome, I 24 think what's become exciting in this area is to 25 consider holistically the role of DNA methylation in 83 1 chromatin which is transcriptionally repres