Targeting Cancer Pathways: Tumor Metabolism and Proliferation

– [Sean] Hello, everyone, and a very warm welcome to this Science AAAS and Science Signaling webinar entitled Part Three: Targeting Cancer Pathways, Tumor Metabolism and Proliferation Thank you for joining us today I’m Sean Sanders, editor for custom publishing at Science and I’ll be moderating this live webinar event As the title notes, this webinar is the third in a series focusing on the cancer pathways that support human development, the emerging research in identifying and targeting these pathways, and innovations in the development of increasingly efficient cancer therapy options The first two installments on this topic are available at webinar.ScienceMag.org Recent advances in our understanding of cancer have revealed that the disease cannot be understood through simple analysis of genetic mutations within the cancerous cells Instead, tumors should be considered as complex tissues in which the cancer cells communicate directly and indirectly with the surround cellular microenvironment and evolved traits that promote their own survival In this webinar, we will explore how cancer cells are able to reprogram their metabolic pathways to enable energy production and the conditions that are disabling for most normal cells Of particular interest are how tumor’s specific metabolic changes promote on congeneric progression and how these changes can be exploited to develop more efficient treatment options It gives me great pleasure to introduce our speakers to you now They are Dr. M Celeste Simon from the University of Pennsylvania in Philadelphia, and Dr. Nissim Hay from the University of Illinois in Chicago Thank you both very much for being with us on the line today Before we get started, I’d like to share some information for our online viewers At the top right of the screen, you’ll find photographs of today’s speakers and a view bio link, which you can click on to read more details about their background and research Underneath the slide viewer is the resources tab where you can find additional information related to today’s discussion and the link to download a PDF version of the slides After the speakers’ presentations, we will have a short Q and A session during which they will address some of the questions submitted by our live online viewers If you’re joining us live, start thinking about some questions now and submit them at any time by clicking the Ask a Question button below the slide window, typing the question into the message box, and clicking okay You can also log in to your Facebook, Twitter, or LinkedIn accounts during the webinar to post updates or send tweets about the event Just click the relevant icon at the top right of the screen For tweets, you can add the hashtag HashScienceWebinar Finally, thank you to Cell Signaling Techonology and the Koch Institute for sponsoring today’s webinar Now, I’d like to introduce our first speaker, Dr. M. Celeste Simon Dr. Simon is the scientific director of the Abramson Family Cancer Research Institute of the Perelman School of Medicine at the University of Pennsylvania Her research is focused on how cells sense and respond to changes in the availability of molecular oxygen and nutrients with a particular interest on understanding the effects of hypoxia in solid tumors A very warm welcome to you, Dr. Simon – [Dr. Simon] Thank you very much It’s great to be here Shall I begin my presentation? – [Sean] Sure, please, go ahead – [Dr. Simon] Okay, I’d like to use my time to focus on the unique relationship between tumor metabolism and the reality that malignant cells exist in what’s intrinsically a very stressful microenvironment This includes cells being very limited in terms of access to molecular oxygen This is a state that’s known as hypoxia But, hypoxia is almost always coupled with decreased availability of nutrients and so, we can even think of many domains within solid tumors as more like tumor ischemia This is because, as tumors eventually achieve a size that extends beyond the natural diffusion limits of either molecular oxygen or blood-born nutrients, such as glucose, glutamine, and lipids, the natural response would be internal domains, which cells have, as I said before, deficiencies in oxygen and nutrients and this impacts many aspects of disease progression, including the acquisition of blood vessels, which are needed to sustain tumor growth and allow tumor cell dissemination during metastasis Of note, going from the small avascular tumor in the upper left hand corner to the large,

more vascularized tumor in the right hand corner, can actually take several decades, based on rates of mutation studies performed in pancreatic cancer Nevertheless, once the tumor does obtain blood vessels, it can, in some way, continue to grow and ultimately, metastasize The problem is, as shown on my next slide, that tumor blood vessels differ from their normal counterparts by a number of key features and they’re listed here First of all, they lack the typical arterial capillary-venule hierarchy, they display apparent, somewhat loose associations with pericytes As shown in the slide, they’re highly fenestrated They can be torturous, exhibit excessive branching, they can abruptly open and close, and the result is chaotic blood flow patterns and very poor profusion of the tissue and this is highlighted in the next slide What’s shown here is a section of a fibric sarcoma taken from a patient that was treated at the University of Pennsylvania and administered the Nitroimidazole compound called EF5 There’s several important points to make from this slide First of all, the tumor, indeed, does have a number of blood vessels and this is shown by immunostaining for the endothelial cell protein, alternatively known as PECAM or CD31, as shown by the green fluorescent depicted in many parts of this fibrous sarcoma section Adjacent to the green blood vessels are cells that are reasonably well-oxygenated, but not that far away are cells that are clearly quite hypoxic and that is because the Nitroimidazoles introduce thiol adducts on proteins and peptides within hypoxic cells For the reaction to work, the cells must exist below roughly one to one point three percent oxygen or 10 millimeters of mercury and they are now immunoreactive for an antibody that detects these protein modifications Again, the important point is that, even tumors with blood vessels have dysfunctional blood vessels and the result is domains where cells are limited for oxygen, glucose, glutamine, et cetera Now, the oxygen in nutrient-limited cells will respond in a variety of very sensible ways They will very rapidly limit anabolic metabolism, they will rapidly inhibit an mTOR pathway, which is a pro-growth pathway, they will engage along with distress responses, which include autophagy, the unfolded protein response There are a number of oxygen-consuming biochemical reactions that will change and there’s a growing family of oxygen to oxoglutarate dioxygenases, which regulates DNA methylation, histone methylation, and collagen modification, among other important biological processes who’s CAM for molecular oxygen will indicate that many of them will also be influenced by the oxygen gradients typical of most solid tumors Nevertheless, one of the best characterized oxygen-sensing mechanisms is depicted here This is a large transcriptional response regulated mostly by hypoxia-inducible factors or HIFs and, if you’re not familiar with these factors, they’re dimers They consist of a constitutively expressed beta subunit, also known as ARNT, show as the light yellow ball on the right hand side part of the slide, but they’re also composed of extremely O2 labile alpha subunits, and so, an oxygen replete or normoxic conditions, the alpha subunits are extremely unstable because they are hydroxoyated on proline residues by a family of HIF specific prolyl hydroxylase domain enzymes called PHDs, hydroxylated HIF alpha is now recognized by a multi-protein complex that includes the tumor suppressor protein, von Hippel-Lindau, targeting it for proteasome-mediated degradation As O2 levels decline, the alpha subunit is now reversibly stabilized, enters the nucleus, dimorizes with ARNTs, binds to HREs in the genome and enhances the expression of literally hundreds of genes that contribute to malignant progression They’re involved in metabolism, tumor endogenesis, rates of cell proliferation, and inflammation and among the many things that the HIFs do, its interfaced with metabolic pathways that are active in proliferating cells and what is shown

in the red box region are the domains of metabolic pathways that have been traditionally associated with hypoxia and activated HIFs and that includes excessive or enhanced glucose import because many of the HIF targets include glucose transporters on the surface of cells Most of the glycolytic enzymes are actually regulated by hypoxia or their HIF targets, rates of oxygen consumption by the mitochondria are downstream of the HIFs, glutamine import and glutaminolysis is enhanced by hypoxia in the HIFs to allow TCA cycle anaplerosis to occur, maintaining lipid synthesis under low O2 and so, along with other stress responses, such as autophagy or the unfolded protein response, the HIF pathway has a role in helping to coordinate oxygen and nutrient availability with the rates of these metabolic pathways to, somehow, allow cells, not only to survive in the stressful tumor microenvironment where they’re limited for oxygen and nutrients, but to somehow continue to proliferate and this is what I would like to talk about today Now, HIF1 alpha and the related isoform, HIF2 alpha are found in numerous tumor diagnostic biopsies and typically correlate with poor prognosis, but one form of cancer is particularly dependent on HIFs and this is the renal cancer known as clear cell renal cell carcinoma Kidney cancers can be subdivided based on their histopathology With clear cell renal cell carcinoma being by far and away the most prevalent form Roughly 75 to 80 percent of renal cancers are of this variety They get their name because, upon H and E staining, they have a so-called clear cell cytophil shown in the right hand image in this slide and this is because they’re thought to excessively accumulate lipid and glycogen and, in over 90% of clear cell renal cell carcinomas, the VHL protein is disabled in some fashion, which means that we thought they had essentially full-on HIF activity, making the HIFs central in the initiation and progression of this disease Now, beyond VHL loss, however, they’re genetically very heterogenous and we and others have engaged in extensive molecular characterization This includes copy number variation studies, exome sequencing, transcription profiles, and other assays of the genomic and RNA content of these tumors We recently complemented this with a pan metabolomic assessment What you’re looking at here is a heat map of roughly 420 metabolites comparing metabolites and kidney tumors relevant to adjacent healthy kidney tissue and we saw a number of consistent changes For example, the renal tumor had lower accumulation of urea cycle metabolites, they had increased accumulation of long chain fatty acids, decreased abundance of lysolipids, but the red box regions showed the metabolites, which were the most different The renal tumors had a significant increase in the abundance of metabolites that fall loosely in the category of glycolysis, gluconeogenesis, and glucose-related sugars We integrated the metabolomic assessment with a metabolic gene set analysis based on RNAseq data and what this is depicting is over 2700 metabolic genes, I E, their MRNAs that encode metabolic genes, subdivided into 72 functional groups based on the KEGG classification and what’s immediately obviously and was somewhat surprising to us is that many of these functional groups, if anything, are under-expressed, in the renal tumors relative to healthy tissue The most under-expressed category was one involved in carbohydrate storage The top fit is the catalytic subunit of glucose six phosphatase, phosphoenolpyruvic carboxykinase, or PCK one and then, two isoforms of fructose one six bisphosphatase, or FBP1 Now, all three of these enzymes, as shown in the next slide, oppose glycolysis in central carbon metabolism It’s appreciated that many malignant tumors and cancerous cells engage in enhanced glycolysis to complement oxidative phosphorylation

and the three green enzymes, PCK, FBP, glucose six phosphatase would actually oppose this They engage in gluconeogenesis and glycogen storage Kidney is a good example of a gluconeogenic tissue It’s not surprising the gluconeogenic enzymes would be down regulated in favor of the glycolytic pathway, but FBP is the rate limiting enzyme and we’ve been able to, now, assess over a thousand individual patient samples and have shown that this is a nearly universal metabolic adaptation FBP is universally lost at the RNA protein level, it’s mutated, or I should say deleted, about 40% of the time and so, this is something that the tumors do They wanna opposed FBP1 activity gluconeogensis to promote, instead, glycolysis and glucose carbon flux into the pentose phosphate pathway to produce, not only nucleotides, but reducing equivalents in the form of NADPH Now, this is hardly surprising, but what is surprising is that FBP1 is another example of a metabolic enzyme that has both a daytime job, that is, it has its metabolic activity in the cytosol, but much to my surprise, FBP1 can also be found in the nucleus In the upper left hand corner is healthy kidney tissue whereby IF immunofluorescence, we can show that FBP1 is both in the cytosol and the nucleus Probably more convincing is the subcellular fractionation shown in the right hand side where it’s very clear that FBP1 protein is in the cytosol, but also the clean nuclear fraction Also shown is the fact that we can introduce a wild type FBP1, this is in the lower right hand side, into these cells These are renal cancer cells that don’t normally have FBP1 If we introduce FBP1 in, we actually saw that a number of canonical HIF target genes that I mentioned previously like lytic enzymes, like LDHA, peruvic dehypyruvic dehydrogenase kinase one, the glucose transporter GLUT1 and the endogenic factor, VEGF, are under expressed, so FBP1 is actually opposing HIF transcriptional activity If we now tag FBP1 with the potent nuclear export sequence shown over in the left hand side, so there’s no longer FBP1 available in the nucleus, the ability of FBP1 to attenuate HIF activity is significantly lost The take home message is that FBP1 is found in the nucleus and can actually oppose HIF transcriptional activity In fact, FBP1 can physically associate with both HIF1 alpha and HIF2 alpha, as shown here by immunoprecipitation Not shown here is that, not only does FBP1 physically associate with HIF1 alpha and HIF2 alpha, it does so at the level of chromatin at the HREs of a number of canonical HIF target genes that fall in the category of metabolic enzymes, inflammatory cytokines, androgenic factors, and other outputs To summarize what I’ve shown you today is that hypoxia and responses to the hypoxic nutrient poor microenvironment impose a number of important metabolic changes Many of these have to do with metabolic enzymes themselves, but as shown in the left hand side of this conclusion slide, many metabolic enzymes have surprising roles Once again, FBP1 in the gluconeogenic tissue, such as kidney, has a role in the cytosol where it carefully balances the toggle of glucose carbons in central carbon metabolism through glycolysis in the pentose phosphates pathway versus gluconeogenesis But, surprisingly, and there are a growing number of metabolic enzymes that fit this category, FBP1 can also be found in a nucleus and it’s actually the nucleus, for a number of reasons, including to restrain HIF activity, and so, we thought for some time that the first event in promoting clear cell renal carcinoma is the universal loss of the von Hippel-Lindau tumor protein shown in the right hand image that we now know that the second hit has to be loss of FBP1 Now, because of loss of both VHL and FBP1, we have essentially full-on HIF activity This revs up glucose uptake, enhances glycolysis shown through the pentose phosphate pathway and many other metabolic targets of HIFs

that sustain tumor self-survival and tumor growth in the naturally stressful tumor microenvironment In the future, we will be performing ChIP-seq for FBP1 That might be surprising for a gluconeogenic enzyme, but we wanna see how many of the HIF HREs are occupied or co-occupied by FBP1 We’ve now been able to extend our observations in renal cancer to liver cancer by doing ChIP-seq and also performing proteomics and mass spectrometry, we’d like to identify other nuclear factors that FBP1 may be influencing This may involve, for example, Wnt/beta-catenin signaling We’ve been able to show that there’s a similar relationship between the related factor, FBP2, and muscle-derived tumors, such as sarcomas and, finally, we’ve now established a conditional wheel of FBP1 to explore the relationship of FBP1 and HIFs and metabolic donation in liver and kidney where certain oxygenated regions engage in gluconeogenesis, but purely oxygenated regions instead promote glycolysis and we’d ultimately like to combine VHL loss with FBP1 loss to generate a more faithful replica of clear cell renal cell carcinoma in a genetically engineered mouse model, which is something that has not been a benefit to this field up to this time Thank you for your attention and I’d like to now turn it over to my colleague Nissim Hay – [Sean] Great Thank you so much, Dr. Simon, for that very engaging presentation and a great start to the webinar We are going to move on to our second speaker right now, Dr. Nissim Hay Dr. Hay is currently a distinguished university professor in the department of biochemistry and molecular genetics at the University of Illinois College of Medicine The main body of his research has been to delineate the mechanisms by which the serine/threonine kinase, AKT, affects thermogenesis and metabolism both at the cellular and organismal levels In addition, he investigates the role of glucose metabolism in cell perforation and survival of cancer cells and how it could be exploited for cancer therapy A very warm welcome to you, Dr. Hay – [Nissim] Thank you, Sean for the opportunity to present our studies today and what I’m going to tell you about is how we could exploit glucose metabolism in cancer cells for cancer therapy High glucose rate of metabolic rate of glucose is a feature of cancer cells that distinguish the cancer cells from normal cells and this could be exploited for detection of cancer cells in vivo, selective detection of cancer cells in vivo For example, using the fluorodeoxyglucose, FDG, glucose analog as topically labeled in combination with positron emissions tomography and CT scans What you see in this slide is a patient with esophageal cancer, before and after treatment using this technology If high glycolytic rate is exploited for selective detection of cancer, why couldn’t it be exploited to selectively eradicate cancer? Actually, the FDG PET scan is dependent only on two steps of glucose metabolism The transportation of FDG by the glucose transporters and the phosphorylation of FDG by hexokinase and this is because, unlike glucose, FDG cannot be farther utilized in a glucose metabolism In fact, the FDG PET scan is more dependent on hexokinase activity than on the glucose transporter because glucose transporters are often bi-directional If FDG can not be phosphorylated, it cannot be trapped inside cells By catalyzing the first committed step in glycolysis, hexokinase is determined if lack of glucose is not only into glycolysis, but also into other glycolytic pathway For example, the pentose phosphate pathway, which is required to generate ribonucleotide and NADPH or the exozamine pathway, which is required for glycosylation of proteins

There are four major hexokinases as expressed in mammalian cells encoded by four separate genes, hexokinase to hexokinase four There is also a fifth hexokinase that has not yet been fully characterized yet The hexokinase one to three are high affinity hexokinase whereas hexokinase four is a low affinity hexokinase and it is expressed only in liver and pancreas Hexokinase one and hexokinase two share structural and functional similarities and they have the ability to bind to the outer mitochondrial membranes and utilize, preferentially, ATP derived from mitochondria to phosphorylate glucose to to glucose six phosphate, thereby coupling of oxidative phosphorylation and glycolysis The binding of hexokinase to the outer mitochondrial membrane is also important for cell survival We are interested in hexokinase two in particular and this is because both hexokinase one and hexokinase two are expressed at relatively high levels in embryonic tissue However, while hexokinase two is not expressed in most adult tissues, hexokinase one continues to be expressed What is shown here at the left side of the slide is immunoblot showing the expression of hexokinase one and hexokinase two in different mouse tissues You can see hexokinase two is expressed at relatively high levels in the heart and skeletal muscles and also in fat, which is not shown here While hexokinase one is considerably expressed in most other mammalian tissues, except in the liver, which is expressing only hexokinase four and not hexokinase one or hexokinase two However, when normal cells convert into cancer cells, they start expressing very high levels of hexokinase two in addition to hexokinase one and this is manifested by the FDG PET scan I would like to address three major questions in my presentation today First, is hexokinase two required for tumor initiation and or tumor maintenance? And, I will address that in four types of cancer: lung, breast, prostate, and liver The second question is, is it feasible to systematically ablate hexokinase two and to emulate drug therapy without severe physiological consequences? This is an important question because, in mice, the germ line deletion of hexokinase two is embryonically lethal And third, why is hexokinase two expression induced in cancer cells? To address the first question, we employed a mouse model of lung cancer, developed by Tyler Jacks Laboratory in which oncogenic K-ras is expressed upon exposure of the lung to adenovirus expressing Cre-recombinase When we analyzed two more lung tumors in this mice, we found that hexokinase two is highly expressed in the lung tumors, whereas the normal lung does not express hexokinase two at all Hexokinase one expression doesn’t change in the lung tumors What we did, we crossed this mice with mice that have hexokinase two floxed gene and we asked where the deletion of hexokinase two will impair the development of cancer induced by oncogenic ras In this generated mice, both K-Ras is activated and hexokinase two is deleted In the controlled group, only K-Ras is activated,

whereas in the experimental group, K-Ras is activated together with the deletion of hexokinase two As you can see in the right panel here, at the right panel, the deletion of hexokinase two markedly extended tumor rate survival and the lifespan of the mice was markedly increased The medium lifespan of this mice is 665 days in comparison with 175 days in the controlled group We markedly decrease tumor burden in this mice, as you can see in the slide and average tumor size and proliferation index is measured by BrdU incorporation Hexokinase two is required for the development of lung cancer in this model and this is also true in human lung cancer What is shown here is a tissue micro-array derived from patient with lung cancer and this is immunostaining with hexokinase two antibodies and you can see high level of hexokinase two in the lung cancer samples in comparison to normal lung This is the staining of hexokinase two that is associated with the grade of the tumor and also, high hexokinase two expression is associated with poor prognosis of patients We have generated a human lung cancer cell line in which we can conditionally silence hexokinase two We then subjected the cells to xenograft assay in mice When the tumors were palpable, we then exposed the mice to doxycycline, which then delete hexokinase two and you can see that we markedly attenuated tumor growth in vivo when we deleted hexokinase two We also find high level of expression of hexokinase two in mammary gland tumors in mice In several mouse model of breast cancer ERbB2/Neu and polyoma middle T and the level of hexokinase two expression is almost equivalent to the level we found in a normal fat tissue We therefore used a mouse model in which ErbB2/Neu is expressed in the mammary gland together with Cre-recombinase We crossed this mice with hexokinase two floxed mice and we asked, what happened in the absence of hexokinase two? As you can see here, the deletion of hexokinase two markedly attenuated the tumor formation and extended the tumor-free survival In fact, it is more than that because when we isolated tumors from the mice without hexokinase two, we found that the tumors were derived from cells that escaped a complete deletion of hexokinase two This is true also in human breast cancer This is a tissue micro-array from a patient with breast cancer You can see high level of hexokinase two expression in the tumor samples and this is, again, correlated with the grade of the tumor and high level of hexokinase two is associated with poor prognosis Here, we have a conditionally deleted hexokinase two This is in the bottom part of the slide We have deleted hexokinase two in breast cancer cell line and we also topically inoculated the cells and to developed tumors and you can see here that, in the red line, the tumor growth was markedly attenuated in absence of hexokinase two

We then wanted to address the question whether or not it’s possible to emulate drastic therapy, namely to systemically delete hexokinase two whole-body and to inhibit lung cancer progression For this purpose, we used a different mouse model in which oncogenic K-Ras is continuously expressed in the lung and the onset of tumors is about eight weeks after birth We isolated cells from the tumors in this mice and we deleted hexokinase two in-vitro and we markedly inhibited the proliferation of the cells We cross this mice with mice that have hexokinase floxed allele, together with a courier combination that can be induced by test tamoxifen and that is expressed in all cells in the body After eight weeks, after tumor onset, we deleted whole-body hexokinase two by injection of tamoxifen to the mice After 20 weeks, we analyzed the lungs of these mice What we found is that, here, even after tumor onset, the systemic deletion of hexokinase two whole-body decreased tumor burden, number of tumors, and tumor size Most of the tumors developed in the absence of hexokinase two are much smaller, but a subset of tumor have similar size to tumor developed in controlled mice, but when we analyzed these tumors for hexokinase two expression, we found that they are still expressing hexokinase two, namely these tumors were derived from cells that escaped full deletion of hexokinase two Now, I would like to switch to another type of tumor and this is quite interesting here: liver cancer As I told you before, normal liver cells and parenchymal hepatocytes do not express either hexokinase one or hexokinase two, they express glucokinase But, once they develop cancer, when they are concerted to cancer cells, they turn off the expression of glucokinase and start expressing high level of hexokinase This is also true in human liver cancer cell line, as you can see here They are expressing only hexokinase two No glucokinase and not hexokinase one In liver cancer, there is either forms which, from the high KM hexokinase, glucokinase, to the low KM hexokinase two, and we analyzed tissue micro-array in patient with HCC, hepatocellular carcinoma, and we found high levels of hexokinase two, which is correlated with the progression of the disease and this is regardless of the reason why this patient developed HCC, even regardless of hepatitis C or hepatitis B, alcohol, and non-alcohol mediated HCC When we deleted hexokinase two in the liver of this mice, mice are relatively resistant to hepatocarcinogenesis and the incidence of tumors in this mice was markedly reduced When we conditionally silence hexokinase two in human ATCC cell line, we markedly inhibited the proliferation of the cells

and the growth of the cells in vivo, here, as shown in the upper part of the slide We inhibited, also, encouraged independent growth by silencing hexokinase two and this is a hallmark of cancer cells and we could restore encouraged independent growth by really expressing hexokinase two, but not the kinase that form of hexokinase two and not by re-expressing glucokinase The fourth type of cancer is prostate cancer Here, we employed a mouse model for prostate cancer in which the tumor suppressor, Pten, is specifically deleted in the prostate of the mice We crossed these mice with hexokinase two floxed mice and we asked what happened to prostate tumor development As you can see in the western block, here, hexokinase two is highly expressed in the prostate of the Pten null mice and when we deleted hexokinase two, you can see a good deletion of hexokinase two and the deletion of hexokinase two impaired tumor development as manifested by the weight of the prostate and the tumor rate of survival, which is markedly increased When we analyzed the section from the prostate of the mice, we found that the deletion of hexokinase two not only inhibited the proliferation of the cells, but also increased cell death as measured by cleaved caspase three When we conditionally silence hexokinase two in human prostate cancer cell line deficient for Pten, we increase the sensitivity of these cells through other therapeutic drugs and I would like to draw your attention to etoposide This kind of cell lines are relatively resistant to etoposides because they have hyperactive ATP, but once we silence hexokinase two in the cells, we re-sensitize to etoposides This is the summary of what I just told you and I would like to emphasize three major take home messages First, hexokinase two ablation reverses tumorigenesis of human cancer cells despite hexokinase one expression Second, during hepatocarcinogenesis, there is an enzyme switch from glucokinase to hexokinase two and hexokinase two expression is correlated with progress of HCC and most important, systemic whole body deletion of hexokinase two in the mouse does not exert an overt phenotype, but inhibits tumor progression in mouse models for lung and breast cancer and this is a genetic proof of concept that inhibition of hexokinase two is well tolerated and therapeutic for lung cancer I would like to thank the people in my lab who did the work The work was pioneered by Krushna Patra and the work on prostate cancer was done by Veronique Nogueira and the work on liver cancer was done by Dannielle DeWaal These are my collaborators Thank you for your attention – [Sean] Wonderful Thank you very much, Dr. Hay Thank you to both of our speakers for those excellent presentations We’re gonna move right on to questions submitted by our online viewers A quick reminder to those watching live that you can still submit your questions by clicking the Ask a Question button below the slide window, typing the question into the message box and then, clicking okay The first question that I have that I’m actually gonna put to both of you and, let me start with you, Dr. Simon Are there specific cancers associated with specific changes in certain metabolic pathways or do you see the tumorigenic effects on metabolic pathways

are more general? – [Dr. Simon] Well, I think that’s a really excellent question and I’m afraid the answer is, it depends One thing that’s very interesting about renal cancer, or at least the flavor that we’re looking at, is that, even though they’re genetically very heterogenous, in fact, even in a single patient, sub-domains of the tumors have extensive genetic and transcriptional heterogeneity, we are finding common metabolic adaptations In fact, the loss of SVP one appears to be nearly universal Another pathway that’s almost equally down regulated are components of the urea cycle and this is something that we’re seeing in each of the nearly thousand patient samples we’ve been able to query in this regard However, work from Ralph DeBerardinis’ Lab on non-small cell lung carcinoma shows that, even in a single tumor, let alone tumor to tumor, their dependence on glucose versus glutamine versus fatty acid beta oxidation can be quite variable Coming back to some of the things that we’re looking at with regard to, for example, urea cycles, enzyme changes, a very significant inhibition of gluconeogenesis, which we think are very particular to this tumor and then, another area that we’re working on is their excessive accumulation of lipid droplets The lipid droplet cenotype is in sort of an extreme case, so we think it’s somewhat unique to renal cancer and forms of ovarian cancer and sarcoma, but not necessarily as pronounced in other kinds of cancers The effects on gluconeogenesis, we think, are more particular to tissues that tend to be more proactive in this regard I’m afraid, as is very typical the case, some things are nearly universal I think many cancers are gonna be somewhat glucose dependent, but others are gonna be more glutamine dependent It’s gonna depend on the oxygenic background It’s also gonna depend on, in a single tumor, the oxygen and nutrient gradients I tried to emphasize I think the honest answer is some things are gonna be more universal, some are gonna be more particular to the tissue type and the oncogenic stress and then, even within a single tumor, it’s gonna depend on local oxygen and nutrient availability – [Sean] Dr. Hay, any comments from you on your experience? – [Dr. Hay] Yes Actually, hexokinase two is highly expressed during hypoxia It is a target of HIF and, I guess, the cell would be dependent on hexokinase two expression under hypoxic condition in particular because they switch to glycolysis and they don’t have much oxidative phosphorylation – [Sean] Right Another question that I’m gonna put to both of you Let me stay with you, Dr. Hay How do you imagine your work translating into potential treatments and what sort of time scale do you think you’re looking at? – [Dr. Hay] Well, I don’t know the exact answer to that This is dependent on pharmaceutical companies who are interested in developing specific drugs The major challenge is to distinguish between hexokinase one and hexokinase two, but I think this is possible – [Sean] Dr. Simon? – [Dr. Simon] Well, there is very intense activity in therapeutic interventions in a number of very widely studied metabolic adaptations A good example is in the kinds of cancer, AML, glioma, chondrosarcoma, angiosarcoma Sorry I can’t remember the name of this tumor Anyway, let’s just stick with AML, glioma, chondrosarcoma, which have mutations in two different forms of isocitrate dehydrogenase Very potent selective drugs have been developed by Agios and other companies They’re in clinical trials and seeing promising, very impressive responses, even complete responses in a subset of patients Those kinds of things are happening

as well as C-myc and N-myc dependent tumors Glutaminase inhibitors have been developed They look very promising in pre-clinical models and they’re making their way through a number of early phase clinical trials In the setting of renal cancer, I think it’s a very interesting point because, as I mentioned, many of these metabolic pathways, if anything, are down regulated, FBP1 and the urea cycle enzymes being very good examples They clearly contribute to the tumor cenotype, but they’re gone and so, we’re pressed with the conundrum of how do you drug something that isn’t there? And, there are ways to do it, what you have to find are consequences of the absence of these metabolic enzymes? But now, one thing we’re engaging in, which I think is a very innovative approach is there are large drug screens that have been conducted on a number of cell lines with expression profiles available So, can we find a drug that provides an MRA expression profile that mimics FBP1 add back, for example, in renal cancers? Can we come up with a drug that mimics FBP1 add back or reintroduction of urea cycle enzymes? As in the case of hexokinase two or mutations in isocitrate dehydrogenase or ovary lines and contaminates are available, those things can be drugged If we’re looking for something that’s missing, we have to do something a little bit more innovative, I would say, but those means are available now – [Sean] Fantastic One of the viewers has made the statement that many of the same metabolic features found in cancer cells can also be found in fast dividing normal cells and, obviously, this is current challenge in chemotherapy If metabolic pathways are to be targeted for cancer therapy, is there a way that one could avoid patterning normal tissue through maybe more specific targeting? Dr. Hay, I’ll come to you with that one – [Dr. Hay] Okay, if you could see from my presentation, when we systemically deleted hexokinase two whole-body in the mouse, the mice did not have any adverse renal type This was a proof of concept that an enzyme could be deleted whole-body without adverse physiological consequence and yet, could be therapeutic for cancer In terms of hexokinase two, as I mentioned, it is expressed only in limited number of advanced tissues and because it’s a high affinity hexokinase, even if we ablate 90% of its activity, it will be sufficient for normal cells, but this will not be sufficient for cancer cells – [Dr. Simon] Sean, my I weigh in? – [Sean] Sure – [Dr. Simon] Okay, ‘cuz one thing that I neglected to mention is that there are nodes upstream of these metabolic pathways and the HIFs represent a good example, and this is very much along the lines of what Nissim was just mentioning We have found, through extensive genetic studies in our hands, albeit mostly with mice, is that the HIFs are extremely critical prenatally, but postnatally, unless you develop a severe pathological situation, for example, we can globally delete either HIF1, HIF2, or both postnatally in mice and the mice, again, very reasonably can tolerate this Now, these are transcription factors, which have been traditionally more difficult to target, but through very extensive structural work, at least HIF2 alpha dimerization with ARNT now has been very flexibly targeted The form of cancer I just discussed, clear cell renal carcinoma, is uniquely dependent on HIF2 alpha and so, it’s gonna be, I think, quite successful to inhibit HIF2 alpha upstream of many of the metabolic pathways that I’ve just been discussing I mentioned the lipid droplet accumulation It interfaces with FBP1 loss and changes in urea cycle and lipid droplet accumulation and so, we actually, and this is being born out in early phase clinical trials, even though this is a central factor, we believe there is a reasonable therapeutic window Of course, when you have the, unfortunately, somewhat more rare case, but a number of patients

affected by, for example, isocitrate dehydrogenase mutation, fumarate hydratase mutations, succinate dehydrogenase mutations Now, you have the unique opportunity, more or less like what was accomplished with BCR-ABL and Gleevec, where there was an activity that was very specific to the malignant cells, not shared with the normal – [Sean] I’d like to stay with you quickly, Dr. Simon for one specific question One viewer says that in their experiments, they see an increase in hypoxia, but not an increase in HIF1 or HIF2 protein levels and they ask whether hypoxia always correlates with a changing expression of the HIFs in your experience? – [Dr. Simon] In my experience, the answer would absolutely be yes HIF1 alpha is ubiquitously expressed HIF2 alpha is more tissue-restricted By and large, this is not a change in MRNA level, it’s a very rapid reversible change in stability where the half life goes from less than three minutes to 150 minutes, but it depends on a level of O2 In our hands, O2 levels have to decline below, shall we say, three percent O2 to really see a robust response I’m wondering if, perhaps, in the experimental conditions outlying, hypoxia hasn’t really been achieved – [Sean] Right then, Dr. Hay, a couple questions for you Have you determined the subset of the localization of HK2 and, specifically, this viewer is interested in lung cancer cells where the HK2 is associated with the alpha mitochondrial membrane? – [Dr. Hay] Yes, it is associated with the alpha mitochondrial membrane in most cells In fact, we found that increase in the HCC, if we ablate androgenous hexokinase two and express a mitochondrial deficient mutant that cannot bind to the mitochondria, it could not restore the proliferation of the cell, suggesting that mitochondrial binding is also important – [Sean] Right, another question for you, Dr. Hay Most of the resistance through AKT is caused by negative feedback to proteins upstream of AKT, according to this viewer, and they ask what is the relationship between the HK2 and AKT and does HK2 play a role in feedback to AKT? – [Dr. Hay] Okay AKT can phosphorylate hexokinase two and it could mediate interaction with the outer mitochondrial membrane In fact, when you activate AKT, there is more association of hexokinase with the mitochondria and this is, in part, a mechanism by which AKT could promote self-survival – [Sean] Perfect Well, that actually brings us to the end of the webinar because, unfortunately, we are out of time It just remains for me to thank both of our speakers very much for being with us, Dr. Nissim Hay from the University of Illinois and Dr. M. Celeste Simon from the University of Pennsylvania Please, go to the URL now at the bottom of your slide viewer to learn more about resources related to today’s discussion and look out for more webinars from Science and Signaling, available at webinar.ScienceMag.org This particular webinar will be made available to view again as an on-demand presentation within about 48 hours from now We would love to know what you thought of the webinar Send us an email at the address now up in your slide viewer, webinar at triple S dot O-R-G Again, thank you so much to our panel for their wonderful presentations and Q and A discussions and also to Cell Signaling Technology and the Koch Institute for their kind sponsorship of today’s educational seminar Goodbye