AUSTRALIAN FRONTIERS OF SCIENCE, 2005
Walter and Eliza Hall Institute of Medical Research, Melbourne, 12-13 April
Histone deacetylase inhibitors: novel drugs for cancer treatment
Dr Ricky Johnstone, Senior Research Fellow, Peter MacCallum Cancer Centre
Introduction by Professor Tony Burgess (Session chair) The therapeutics of cancer is coming up a long way in a very short time. The targets for many of our cancers have been identified over the last 25 years, and the genetic basis of many cancers is well defined, both in the initial stages of the carcinogenesis process and during progression. In diseases like colorectal cancer, we know that there are early events truncating mutations in a particular gene that occur in 85 per cent of all colorectal cancers. And they are accompanied then by changes in other genes that we can understand a little bit about, like the Ras oncogene. In other cancers such as pancreatic cancer, 95 per cent of them involve a mutation to one particular protein, the Ras oncogene. And you can go through the 50 or so cancers and start to point out the mutations that are actually not only associated but probably causative for the final state. Many cancers involve several genes, and they can involve combinations of genes. The thesis is that if you can identify the target and turn it off, you can help some patients. We know that there are some classic examples of that all too few such as the abelsen kinase, and we will soon see, I think, the therapeutic benefits of inhibiting the JAK kinases. But people are solving the structures, they are designing drugs, and more and more we are able to target the therapy to either slow the growth or kill the cells. We have to understand each of the targets in detail, with enough details that we can deliver a drug that will turn off the target. And then we have to be able to measure, in the patients, that the target is turned off. And unfortunately with most of the therapeutic systems that we have dealt with to date, people haven’t bothered to measure whether the target is actually turned off or not. It turns out with Gleevec and Abel, yes, they did, and it was a really very important breakthrough, because they could see a group of patients with a particular type of stomach cancer by doing some functional imaging, and they could tell which patients would really respond. So today we have got two speakers from the Peter MacCallum Cancer Centre. Ricky Johnstone, is going to tell us about a novel target, the histone deacetylase, and the development of those inhibitors. Then Grant McArthur, an oncologist scientist, is actually following the effects of the drugs with the latest technologies and trying to work out the best way to optimise the treatment. So the fitting of those two talks together should be really very interesting for us.
(Click on image for a larger version)
The way I am going to structure this talk is to provide you with a bit of an overview on this new class of chemotherapeutic drugs known as HDAC inhibitors (HDACi). I am going to tell what they are, in general terms, and what they do; and a little bit about the mechanism of action of these compounds and the work that has been done in my lab and others. I will show you just a small bit of data that there is tumour cell selectivity of these compounds, which of course is important for any chemotherapeutic drug; some work that is ongoing now about combination studies and some nice work that has been done by groups all over the world to rationally design combination studies by knowing mechanisms of action of the drugs and knowing what is happening at the oncogenic target; and then, finally, some progress in terms of clinical trials.
(Click on image for a larger version)
The first thing is: what are these compounds, these HDACi, and what do they do? They come in various shapes and sizes, from very simple compounds such as valproic acids and butyrate down to more complex structures such as the cyclic tetrapeptides. And these compounds can be synthetically made, such as SAHA and oxaliplatin, or can be natural products depsipeptide is a bacterial compound.
(Click on image for a larger version)
The target of these compounds: although they are called targeted therapeutics, there are certainly more than one molecular target inside the cell. In fact, there are at least eleven different HDACs that have been reported in mammalian cells and that supposedly do very similar things, and most of the compounds in fact hit all of these proteins, to varying degrees. There haven’t really been very elegant studies yet to determine the various affinities of the compounds for the different HDACs, mainly because they are so difficult to purify in an active form. But certainly it is fair to say that most of the compounds that have been used can target most of these HDACs.
I should point out here that even the term ‘HDAC’ is a misnomer. They were described by their ability to deacetylate histones, but really there are a whole range of different proteins that are regulated by acetylation, and these HDAC enzymes can deacetylate a whole range of different proteins. So we should really change this, and there is a bit of a move in the field now to call them protein deacetylases, PDACs. That is a more fitting term.
(Click on image for a larger version)
But certainly these enzymes do alter the acetylation state of histones, and this has a knock-on effect of regulating transcription. It was brought up in the last talk that DNA is not a single strand that sits inside the cell; it is wrapped around these proteins, histones, and forms a high-order complex known as chromatin. And it is the status of chromatin that regulates whether a gene is turned on or off.
In the situation shown at the left of this slide, where the HDAC enzymes are deacetylating the amino terminal tails of the histones, preferably histone H3 and H4, this results in a condensation of the chromatin, and transcriptional repression. Now, there are opposing enzymes to the HDACs, the so-called HATs, or histone acetyl transferases, that hyperacetylate the histone tails and open up the chromatin. As you can see here, shown schematically, that then allows for putative promoter regions to be bound by transcription factors, by DNA binding proteins, and drive transcription. And certainly this is what HDACs do very well, but as I pointed out there are other proteins that can be regulated in this way. So what the HDAC inhibitors do, of course, is to inhibit these enzymes and drive the reaction in this way.
(Click on image for a larger version)
In terms of the biological effects of HDACi, they have many of the properties that you would like in a chemotherapeutic drug. To a tumour cell, they can drive the cells into a cell cycle arrest that can result in cytostatis or in some cells drive the tumour cells to terminal differentiation.
Normal cells are also affected by HDAC inhibitors, but they tend to just drive into G1 arrest and this effect is transient. So, at least in vitro and in some in vivo experiments that have been done, once you remove the drug the normal cells actually go on and start to divide again.
One of the nice properties of the HDACi is that they induce apoptosis of tumour cells. This really doesn’t happen very well in normal cells until you get to very high concentrations of these drugs. And the effects of the HDACi to drive a tumour cell to cell cycle arrest or to apoptosis are mutually exclusive events. So if the cell responds by going to cytostatis, into a G1 arrest, then this has an inhibitory effect on the apoptotic program. We don’t really understand yet the molecular interplay between cell cycle arrest effect and the apoptosis effect, but I think this is going to be quite important down the track.
The other activities of these compounds that can help in an anti-cancer setting, I guess, are to increase the immunogenicity of the tumour cells by increasing the expression of molecules such as MHC class I and class II and we have just heard a talk about the importance of these in immune responses and also to decrease angiogenesis of the tumours by targeting key genes such as VEGF.
(Click on image for a larger version)
This is some experimental data that I would like to show you, from my lab, where we used the E-myc B-cell lymphoma system that was developed in this Institute. (David Huang spoke about that earlier.) We can take these cells out of the mice ex vivo and treat them with a particular HDACi this is the hydroxamic acid called SAHA and these cells are killed very nicely by this compound at low micromolar concentrations. We get the classic induction of apoptosis, including DNA fragmentation, which is shown here.
(Click on image for a larger version)
SAHA works very nicely in vivo. Shown here is some small animal PET analysis that we have done, where we have injected these E-myc B-cell tumour cells into C57BL/6 mice. After about six days we can image these tumours, using radiolabelled FDG. So this is a marker of metabolism of the tumour cell. And you can see that there are various sites within this mouse that are lighting up, showing the tumour cells that are existing within the lymphoid organs.
When we inject SAHA into these mice and wait 24 hours, and then re-image these same mice, you can see that the metabolism of those tumours is being affected. Now, this is not a read-out for cell death of the tumours; it is a read-out for their metabolism. This is the glucose uptake, if you like. But we know that at least in vivo this drug is having an effect on the metabolic rate of the tumours. This is just the control, to show that if we have the vehicle alone there is no effect in the uptake of the label by the tumour cells.
(Click on image for a larger version)
So what about the therapeutic effects of these compounds against this sort of tumour? As I said, this is a B-cell lymphoma model, and if you don’t treat the mice after about 18 days or so, the mice will succumb to this tumour. You can see massive infiltration of the lymph nodes and the spleen and various lymphoid organs; the blood is packed with basophilic blast cells. If we treat with SAHA, we can see we have a very nice therapeutic effect of this compound: the mice for all intents and purposes look pretty much normal when we open them up, and their blood is relatively clear of tumour cells.
So I have shown you that the HDACi can be varied in their structure, they have various anti-cancer activities and, at least in the laboratory setting, they can have very nice therapeutic effects.
(Click on image for a larger version)
What about their mechanism of action? How are they working?
According to the initial model we would think that what these compounds do is to target transcription, to alter transcription inside the tumour cell. So we and others have asked whether that is an important effect for them to have their anti-cancer effect. What we can do is block de novo protein synthesis by using cycloheximide we can also do this with actinomycin D to block new gene transcription and then look to see what effect that has on the ability of these HDACi to kill the tumour cells.
What you can see here is that, as you increase concentrations of cyclohexamide, you increase the anti-cancer or the apoptotic activity of these compounds, indicating that de novo protein synthesis is going to be important for the activity of these compounds.
(Click on image for a larger version)
One of the models being established at the moment is that these drugs target histones: they alter the chromatin structure by causing histone hyperacetylation, this results in differential gene expression inside the cell, and somehow this is triggering an apoptotic event. I haven’t gone into all the details, but we see all the classical features of cell suicide that David Huang and David Vaux were talking about, earlier in the day.
So one of the questions is: well, what are the genes that are being targeted here, and can we make some sense out of these genes in relation to the triggering of the apoptotic program?
(Click on image for a larger version)
My lab and others have performed microarray analysis, using different HDACi on tumour cells over a timecourse, to look to see what the genes are that are changing. We have found that a large number of genes in the genome are being altered. The initial figures that came out from some of the early studies was that 10 per cent or so of the genes in the genome are affected by these compounds, but we have found by using more robust bioinformatics programs that it is probably more like about 30 or 35 per cent of the genes in the genome that are being affected.
In particular, there is a whole raft of apoptotic genes that are affected by these compounds, and what we have found is that, in general, what happens is that pro-apoptotic genes are being activated and anti-apoptotic genes are being repressed. So here is a pro-apoptotic gene, APAF1; it is being activated; Bcl-2 is being down-regulated. It is not 100 per cent correlation with this effect, but in general most of the genes that we see activated are pro-apoptotic, and most of the genes that we see repressed are anti-apoptotic.
We think there is a cascade of events occurring, pushing the balance from cell survival in the tumour setting to now cell death in the tumour setting. So what we want to do is to be able to interrogate the relative role of any of these genes. We can do that in the laboratory by picking one for example, APAF1 and using SRNA techniques to knock down expression of this gene and then see what effect that has on the activity of our HDAC inhibitor. What you can see here is that if we knock down APAF1 we have a significant effect on the ability of SAHA to kill these tumour cells.
An even better, more robust and more elegant way to do this is to use a genetic approach. This is using the E-myc model, where a particular gene in this case, APAF1 has been knocked out. (These are cells provided by Clare Scott and Andreas Strasser from the WEHI.) Then we can ask whether the particular HDAC inhibitor that we are using can now kill these cells. You see here the E-myc B-cell lines being killed by SAHA very nicely, as I have shown you before, and now if we knock out APAF1 you can see there is an attenuation of the response.
It appears in this case that losing APAF1 alters the kinetics of death induced by this compound, and not the overall effect. Whether this has a therapeutic effect is something we are looking at in the laboratory at the moment, because we have the ability to look at these cells in an in vivo setting.
(Click on image for a larger version)
So that is one of the models that I have shown you now, where HDACi are targeting the HDACs to cause histone hyperacetylation, altering the chromatin structure, causing multiple genes to be turned on and some to be turned off. This results in apoptosis in most cases, but in some cases can result in cell cycle arrest.
Now, I don’t want to lead you up the garden path by saying this is the be-all and end-all of this model, because that would not be true. As I said before, these enzymes, these HDACs, are protein deacetylases. And there are some key proteins that are regulated according to their acetylation status, including transcription factors like p53, E2F, Rb. It is possible, though not formally shown yet, that what is happening when you add an HDACi is that you are affecting the activity of these transcription factors, which then in turn regulate gene transcription. It may actually be that there is a combination of events involving the two pathways shown here, and this is something that we are looking at in the laboratory at the moment.
(Click on image for a larger version)
Another model is that other proteins that may not have too much at all to do with transcription are being affected, and it is the regulation of these proteins that can regulate the anti-tumour effects of drugs like HDAC inhibitors.
(Click on image for a larger version)
I just want to draw your attention to one, Ku70. These are results that were recently presented in PNAS. Ku70 is a protein that is predominantly found in the nucleus but can be found out in the cytoplasm, and there was always some conjecture as to what it might be doing out in the cytoplasm.
It turns out that Ku70 can bind to and inhibit the activity of a pro-apoptotic protein called Bax, which David Huang was talking about earlier today. When Ku70 binds to Bax, it sequesters its activity. So this is an anti-apoptotic state here. Now, the ability of Ku70 to bind to Bax is regulated by the acetylation status of Ku70. In its non-acetylated state, Ku70 binds to Bax and inhibits its pro-apoptotic activity; when this protein is acetylated by these HATs, then the Bax can be released, go to the mitochondria and induce the intrinsic apoptotic pathway.
Shown here are some studies that were done by this group to show the involvement of Ku70 in HDAC inhibitor-induced apoptosis. If you use cells that don’t have Ku70, for example, there is an attenuation of the effect. And if you express a mutant form of Ku70 that can’t be hyperacetylated, so always binds Bax in the state I spoke about, there is also an attenuation of the ability of this HDACi, TSA, to kill these cells.
What I am trying to point out at the moment is that we are learning a lot about how these compounds work, but you should not be misled into believing that we know everything about them. The ability of these compounds to work, both in preclinical models and, as I will show you, in the clinic, may actually be a result of the combination of effects throughout the cell and they don’t have a single target and a single pathway through which they work.
 |
 |
| (Click on images for larger versions) | |
I have just one slide here to show you that there is tumour cell selectivity or specificity of these compounds. Once again we have gone back and used the E-myc model to look at this. This is an extremely nice model to work with. It is a syngeneic model. What we can do is to take the lymphoma cells out of these E-myc mice and inject them into a strain of mouse that is congenic to the C57BL/6 mouse, in that it has the same genetic background except for one allele. We can then follow that by flow photometry.
So we inject these cells into the donor cells, we allow these mice to form lymphomas, and then we inject them with our HDAC inhibitors. We then take out the cells from these mice and analyse them by flow photometry. Here you can see the tumour cells in a mouse that we inject with the vehicle alone. They form the bulk of the lymphoid organ that we are looking at, be it the lymph nodes or the spleen. And here, then, are the normal cells from the PTPRCA mice.
After we treat with SAHA and just eight hours later, you can see that there is a massive effect, a massive reduction in the number of tumour cells in this lymphoid organ, whereas the normal cells remain relatively unharmed. We have done this by various methods now, to show that there is selectivity of these HDACi for the tumour cells in vivo.
 |
 |
| (Click on images for larger versions) | |
So what about combination studies? I think this is an area in oncology that is really going to grow and expand over the next few years the use of different compounds in combination. My honest opinion is that the HDAC inhibitors will be best in combination therapies rather than as single agents. There are various reports in the literature of using HDACi with various other compounds, and to a certain degree it has mainly been done in somewhat of an ad hoc way, where you take one cytotoxic or one anti-cancer agent, stick it in with another and hope that it works better. But there has been some rational use, or rational design of trials testing HDACi.
(Click on image for a larger version)
One is the use of HDAC inhibitors in conjunction with DNA demethylating agents. One of the things that HDACi can’t do is that they can’t reactivate genes that are hypermethylated, where the DNA is methylated. So if you have got a hypermethylated gene and you add an HDACi, that gene is not activated. And there are some key tumour-suppressor genes such as p16 that are hypermethylated in cancer and are thought to be causative. So if you added an HDACi to those cells, you wouldn’t reactivate a tumour-suppressor gene; therefore, you might not have an anti-cancer effect.
However, it is known that if you demethylate that gene, the gene is now responsive to an HDACi. So a combination therapy, of a DNA demethylating agent and an HDACi, would supposedly reactivate this gene. This is shown experimentally here, where we are using a read-out of a hypermethylated gene: when we add TSA this is the histone deacetylase inhibitor alone it is not reactivated. If you add the demethylating agent it is not reactivated. If you add both, in combination, now you strongly reactivate that gene. And there are indeed now clinical trials ongoing, involving demethylating agents in conjunction with HDACi for exactly this purpose.
(Click on image for a larger version)
Another nice example of the rational use of this compound in combination is in acute promyelocytic leukaemia, that result from chromosomal translocation, that cause the expression of this fusion protein, or this hybrid protein, involving the retinoic acid receptor and PLZF. Now, usually retinoic acid receptors respond very nicely to retinoic acid, to reactivate the genes that may be repressed by retinoic acid receptors. But in this case, because PLZF brings in, by itself, an HDACi so both parts of the fusion protein bring in HDACi to repress cellular genes the retinoic acid receptor, while it can remove one HDAC inhibitor, still retains the HDACi brought in by the other part of the fusion protein.
So it was decided that if you could use retinoic acid in conjunction with an HDACi you might be able to reactivate this gene. And it is certainly known that this fusion protein and the repression of the target genes of this fusion protein are causative of particular leukaemias. Shown here is the experimental evidence that that can work. When you have got cells that express this fusion protein PLZF-RAR, if you just add retinoic acid alone there is no effect on the cells this is cellular differentiation, actually, that is being read out here. If you add TSA alone there is no real effect. But if you add both, in combination, there is a very nice effect.
 |
 |
| (Click on images for larger versions) | |
Finally, two slides to show you how these compounds are working in the clinic. This slide is of a schematic to show you all of the different trials that are ongoing at the moment I think there are about 30 different clinical trials all over the world, including two at the Peter MacCallum Cancer Institute, ongoing at the moment.
(Click on image for a larger version)
And really this is encouraging for someone who works in the laboratory on these compounds, in that it shows they can have a clinical effect.
This is a patient with cutaneous T-cell lymphoma, who has not responded to a whole raft of different classic chemotherapeutic regimes, including etoposide and cyclophosphamide and doxorubicin, I believe. This patient was treated with an IV dose of depsipeptide, which is an HDACi that we use in the laboratory. After four rounds, I think, of therapy, his tumours are basically disappearing from the face and from the neck, and also from the legs. There are also very nice CT responses. This patient went on to be tumour-free at the end of his therapy.
(Click on image for a larger version)
So I hope that in the 20 minutes that I had I have given you a bit of a feel for this new class of anti-cancer compounds. I have told you what they are and what they can do, and something about the mechanisms of action of the compounds, although we have obviously still got a long way to go with that. I have shown you that they are selective for tumour cells; they can be used very nicely in combination and I think that the rational use of these compounds in combination is something that is really going to be moving forward in the next few years. And the clinical trials are so far showing some very good, promising results, at least for certain disease states such as cutaneous T-cell lymphoma.
