 Great. Thank you, Roberto, and thank you organizers for inviting me. I'm going to tell you a little bit today about some of our work, much like Billy Kim talked about yesterday using synthetic lethal screens, except instead of using SIR and A libraries, we use small molecule libraries. And the idea was that we could interrogate the whole genome to undercover genetic drivers using this approach. And when we have identified these, then we would already have candidate molecules to inhibit them. So the left-hand side is just a very simplistic cartoon taken from really the work done in yeast where the concept of synthetic lethal is derived from these lower organisms where you have two genes, A and B, that when either one of them individually is lost, there's no effect. But when both of them are lost, you get that lethality. So the right-hand side is a cartoon from a review that we wrote several years ago that was how we did the screen where we mixed renal cancer cells that were genetically matched for VHL, one containing two different fluorescent markers so that they were internally controlled. And then what the predicted outcomes would be that the vast majority of these would either have no effect on either cell type or they would kill both cell types or that you would have ones that would specifically kill cells that lack VHL, which we were interested in, or those that would kill the wild type. And just for your interest, we've published a number of these studies right now. So some of what I'm going to talk about, you can look up if you're interested. So we screened 130,000 small molecules. Again, these were in multiple cell lines that were genetically matched for VHL. And we were looking for selective killing of those cells that lacked VHL. So out of this 130,000, we've had about 400. From those 400, we no longer were able to use robotics to do the screens. Instead now a single graduate student took these and arrayed them and did secondary assays to make sure that these actually what we thought they were. And I will tell you about two of these molecules. One called STF61, which is an autophagy inducer. It's inducing cell death in renal cancer cells by inducing an autophagic process. Second is called STF31, of which we've got several candidates that came out. And this inhibits the Glute 1 transporter. And then just, I added this just to be complete, is that we have molecules that would selectively kill the VHL wild type cells and have no effect on the null cells as you might expect. And that's called STF51. So the reason why we did this, and I'm sure you're all aware of is if you looked at renal cancer cells that either lack VHL or her wild type VHL, and you looked at classical chemotherapeutic agents, what you find is that it kills both cell types. So there's no differential in terms of VHL killing or it has no effect in either cell type. Instead, what we really want to maximize the ability to get a therapeutic index is to develop small molecules that would selectively kill those cells that lack VHL, and killing here is defined in the black, versus having no effect in those cells that are wild type VHL. And here's several examples from this screen. So here is STF6247. And this is a cell survival curve. This is a log log base scale. And what you can see here is the wild type VHL cells are really unaffected by this molecule. And the VHL deficient cells are very potently affected. If you take a renal cancer cell that has wild type VHL and you put an SH for VHL in, you can now sensitize the cell to killing, indicating that this sensitivity is based on VHL. And here's an example if you actually looked in the wells and looked at increasing concentrations of this drug, you see is that you see a nice dose dependent increase in cell killing in the VHL deficient, whereas in the VHL wild type there's no effect. So very interestingly though, this molecule's toxicity is hip independent. So either by inhibiting HIF or by conditionally expressing a non-degradable HIF molecule, you can see that in either case you're not changing the sensitivity. And in the bottom here is the appropriate controls to demonstrate that we're getting the effects that we thought. But instead what's killing the cell appears to be an autophagic type cell death. So these are electron micrographs. This is vehicle treated. This is a six series treated. You can see the formation of these double membranes, classical feature of autophagy. And then in the VHL deficient cells, we saw this pattern a great deal. We saw the formation of lysosomes and acidic vesicles and we never really saw the fusion of the two. So if we look now at a series of analogs, so kind of in this case trying to do almost kind of like a chemical genetic derivative screen to prove this point is that here's a series of analogs. They're all inducing autophagy to sub-degree as detected by LC3, the myosin light chain, lipidated form. And you can see here is FTF62247, which has a 25-fold differential between VHL wild type and deficient cells. You can see there's this very large increase in the amount of these acidic organelles. And if you look at other molecules here that are also quite potent in terms of killing, so for example this one here is about 20-fold, you can see that it also is very potent. Whereas those that are not inducing any autophagy or have minor effects on autophagy, you're not seeing any increase in this formation of acidic vesicles. So the way we think this works is that 6247 interferes with ER Golgi trafficking, in that we found a series of genes that are VHL regulated which are involved in this process. And that this then initiates an autophagy process. And this autophagy process occurs both in VHL wild type and VHL null cells. But the difference is that in the VHL deficient cells, you're not getting this efficient diffusion of these acidic vesicles with the lysosomes. And this is resulting in cell death, whereas the VHL wild types you have cell survival. So the second molecule is this STF31. So this is looking at a colony formability. And you can see here in VHL deficient cells you're able to wipe out these colonies. And these colonies are dying by a necrotic form of cell death. So if we look at changes in glucose uptake, here's the VHL wild type cells and you can see that there's really no change with increasing concentration of drug, whereas the VHL deficient, there's a very precipitous drop with drug concentration. If you look at ATP levels, which follows this decrease in glucose, ATP levels go down. And if you look at clonogenic survival, you see that also decreases. Again, a log-log base scale, whereas the VHL wild type cells are not affected. So glute one belongs to the solute carrier to a transporter family. It's an aqueous tunnel within a cell membrane. It binds directly to glute one, this STF31. And we have no interactions with other glutes that we can detect, either detecting this by affinity chromatography, differential expression assays, or genetically matched glute one cells. So what's the downside of glute one inhibitor? And that downside would be its normal tissue tox, but in fact if you actually look at normal tissues, the largest cell type that expresses glute one are erythrocytes. So here we're looking at human and mouse. Human erythrocytes actually utilize glute one, mouse utilizes glute three. And each one of these panels, there's either red blood cells or mouse RBCs. The left, both panel here is red blood cells incubated with STF31, the middle is the vehicle, and the right is a RBC lysis buffer. And what you can see here is that there was no effect, at least in terms of three days, if not even longer, in terms of RBC effect. So RBCs, well, they have high levels of glute one, and glute one is inhibited in these RBCs, that's not inducing cell lysis. What about other tumor types as well as renal cell cancer? So on the right hand side, we're looking at glute one expression. And at the RNA level, is that there's been several reports saying that the messenger RNA is very high. But in fact, as everybody in this room knows, very rarely do you ever use PET CT scanning really in staging for renal cell cancer, maybe for metastatic disease, but rarely for primary. So if you look though at renal cell cancer, in terms of glute one expression, what you see is that about 40 percent of these tumors have significant levels of glute one at the protein level. So what about in vivo? This is the molecule unoptimized from the screen. This is a PET CT slice. Here is very high glucose uptake of the tumor is that after three treatments, three daily treatments of STF 31, you see that this tumor is significantly decreased in terms of its FDG uptake. And in addition, there's an increase in its necrotic core. Here's seven mice demonstrating the decrease in glucose uptake in all the seven mice. And here's looking at tumor growth delay. Here is the vehicle treated tumors, and here is the glute one inhibitor treated tumors. So there is significant efficacy even with our hit, let alone the optimized molecule. So the way we think this works is that cells that have wild type VHL, they're not dependent on glute one for their glucose uptake and are able to perform both glycolysis and a TCA cycle. Cells that lack VHL is that glute one levels are elevated. They're highly dependent upon glute one. In addition, they have increased expression of a kinase called pyruvate dehydrogen kinase, which prevents pyruvate ultimately from going to acetyl CoA. And therefore, since they're not really very efficient at performing TCA cycle and their metabolic requirements are more dependent on glycolysis, they're dying. So I'm going to end by just telling you the newest gene that we think is a potentially very good target for renal cell cancer is one called null three or arc. And this is a gene that contains something called a card domain, which is a cascase activating and recruitment domain. And this gene arc inhibits both the extrinsic pathway and the intrinsic pathways of apoptosis. So extrinsic, meaning those induced by death ligands such as FAS or TNF, or the intrinsic where apoptosis is being signaled from the mitochondria through cytochrome C release. So it's inhibiting both pathways. So here's data to show you that arc is induced by hypoxia. These are cacky one cells. And as a reference, here's phosphoglucocinase. Here is a variety of different cell types. There's lung cancer, melanoma, cervical, renal cancers. And you can see that in all these cases, arc is very nicely induced by hypoxia. So this arc induction to our surprise was turned out to be a HIF-1 dependent versus HIF-2 dependent. So here we're looking at hypoxia induction. Here's SH to HIF-1 alpha and SH to HIF-2 alpha. If you knock out HIF-1 alpha, you can see you've abolished the induction of arc under hypoxic conditions. Whereas with HIF-2 alpha, there's no effect. If you use genetically matched cells that have HIF-1 alpha deleted from the genome, again, you see there's no induction of arc. If you overexpress a HIF-1 alpha or HIF-2 alpha constituentally active molecule, you can see that HIF-1 alpha is very potently able to induce arc and HIF-2 alpha not so much. We identified the binding site in the arc promoter that's binding the HIF molecule. And very interesting enough is that this is a alteration that occurs of the onset of mammals because as you see here when you go from the marmoset to the lemur, this HRE is lost. So when you look at mouse and rat, you don't see this induction. So arc expression is elevated in human renal cell cancer. Here's normal kidney and clear cell renal cell cancer. All four of these panels demonstrate that point. Panel E here shows that arc and CA9 marker for HIF activity co-localized very nicely. And on the right is actual quantification of that co-localization, demonstrating that both arc and CA9 are very coincidental in terms of their expression. What happens if you knock down arc? So here is under normoxic conditions. Here's the change in cell number. If you knock down arc, what you see here is that inhibition of growth under normoxic conditions as well as an inhibition of growth and if not cell lethality under hypoxic conditions. Here is looking at colony forming ability, same thing. And what's happening is that these cells are dying by apoptosis as evidence here by the activation of the cleavcast base 3. So most importantly, getting back to my initial comments, is that here are multiple VHL cell lines, RCC4 and RCC10 genetically matched for VHL. Knocking down arc alone in these cells is a very potent effect, has a very potent effect on cell survival. And in addition, it takes these cells and makes them now extremely sensitive to chemotherapeutic agents such as cisplatinum. If you try to grow tumors in which arc is stably knocked down, they do not grow very well, if at all. So the way this works is that in cells that have a wild type VHL, arc is at low levels and has very little inhibitory effect in terms of Caspace activity, but in cells in which VHL is lost and probably this is an early event with the initial loss of VHL and both HIF1 and HIF2 are still present in renal cell cancers, is that arc levels goes up and inhibits both the intrinsic and extrinsic pathways. So most important slide, Patrick Suffin, Denise Chan, and Sandra Turkov were responsible for the studies in the synthetic lethal screens and Olga Razoranova and Olga and Lara Castellini responsible for the studies with arc. Thank you very much.