Slide 1

Thank you David. I would like to start straight away with a traditional view of this feedback regulation that I’m sure everybody here in the audience is familiar with.

Slide 2

The oxygen content of blood which is determined by the haemoglobin concentration and by the oxylate saturation regulates the amount of erythropoietin which is primarily produced by liver and kidneys. EPO then stimulates red cell production by inhibiting apoptosis of red cell precursors and in this way we have a very nice negative feedback regulation which as we all know, maintains the haemoglobin concentration in very narrow limits under normal conditions.

Slide 3

Now, this concept of a feedback regulation of EPO has been established many, many years ago. It can be found in every text book but there are some interesting elements and there’s increasing complexity when we look at some of these aspects in more detail. One of the very interesting aspects is that there is not only, of course, the opportunity to intervene here by recombinant erythropoietin but there’s increasing evidence that the expression of EPO receptors is not limited to the bone marrow but that EPO receptors may also be expressed in a variety of other tissues, that these tissues may also express low levels of erythropoietin. So that we may have to add to this scheme another loop of tissue protection in which EPO can also operate locally and not only as a hormone. I would like to focus on a few aspects in this scheme and for reasons of time, of course, I cannot go into every detail of every aspect but the 3 areas that I would like to touch are the evidence for the expression of erythropoietin receptors in tissues outside the bone marrow. Second some very exciting evidence that erythropoietin receptors may in fact be heterogeneous in different sites. Finally, knowledge that has accumulated more recently about the oxygen sensor which is essential for the operation of this feedback circuit.

Slide 4

So let me start with the extraerythropoietic paracrine effects of EPO and the evidence for EPO receptor expression. I have no time to go through all this evidence but if you go to the literature, you will find that there are many, many papers which have demonstrated in recent years that EPO receptors are expressed outside the bone marrow. A word of caution has to be added here because many of these data are based on immunohistochemical evidence and this evidence is somewhat weak and indeed there is a paper published recently in Blood by colleagues from AMGEN very nicely and convincingly demonstrating that some of the antibodies that were used to demonstrate EPO receptor expression in different tissues are not picking up EPO but that the results of these studies are obviously false positives. So there’s some caution, this data has to be seen with some caution but the concept of EPO receptors outside the bone marrow fortunately does mean one other result on other experimental approaches. One of these approaches which I think is pretty elegant is shown on the next slide. Investigators have knocked out the erythropoietin receptor and of course, this would not be compatible with normal development, the animals would simply die and in order to rescue these animals they have then on the background of this complete knockout of the receptor rescued EPO receptor expression specifically to the bone marrow by using a promoter sequence which is only expressed in the bone marrow. These animals then ideally should only have EPO receptors in the bone marrow and should allow to study the effects of EPO in other non-erythropoietic tissues to see whether there’s really a functional difference.

Slide 5

And the one example that I would like to show here is related to the heart of the paper that was recently published in Circulation Research and as you can see here these wild type animals when compared to those with EPO receptor knockout and then rescued in the bone marrow have a much smaller infarct size than those lacking the EPO receptor in tissues outside the bone marrow including the heart. This increased infarct size was also quantified here and it is functionally relevant and over the time leads to impaired myocardial infarction.

Slide 6

And I think that’s pretty elegant evidence that indeed EPO receptors outside the bone marrow do play a physiological and even more pathophysiological role. Finally, there are a lot of studies which have shown that tissue protection can be achieved using very high doses of erythropoietin.

Slide 7

For this audience I would like to focus exclusively on evidence provided in models of renal injury, acute renal ischemia and this is one of a number of studies which have demonstrated that when you use high doses of EPO, very high doses I should say, 5000 units/kg and give this to animals prior to renal ischemia, these animals they show protection against the ischemic injury. One of the groups that published this data even claimed that this protection can be seen at a macroscopic level as is evidence from this slide. There is less apoptosis in these tissues after the application of erythropoietin and this translates into a functional benefit with less of an increase in the serum creatinine concentration after injury. From a clinical point of view it’s very exciting that in one of the recent issues of Kidney International another study has been published showing that it is not essential to apply erythropoietin before the ischemic insult but that the whole system seems to work also when erythropoietin or darbepoetin are given 6 hours after ischemia and that protection is then still maintained and of course, this brings this approach somewhat closer to clinical application.

Slide 8

One of the important questions is, of course, considering these diverse effects of erythropoietin and considering the very high doses of erythropoietin which are obviously needed to induce these effects whether there’s heterogeneity of the erythropoietin receptors in different tissues which might explain why different doses are needed and why the response to different doses is so different.

Slide 9

Indeed there’s evidence upcoming suggesting that these two receptors may not be identical. This is work primarily by Doctor Brines and Cerami and I’m quoting their work here and trying to summarise it schematically and what they have shown or provided evidence for is that in addition to the erythropoietin receptor which we normally find in the bone marrow, which consists of two identical EPO receptor subunits there’s a second erythropoietin receptor which also consists of one of these subunits but is a dimer with another type of receptor the gamma beta receptor.

Slide 10

So this is a homodimer and this is a heterodimer and obviously the affinity of these two receptors is very different. The affinity of this receptor for EPO is much higher than of this receptor, in other words much higher concentrations are needed to stimulate this heterodimer as compared to the homodimer. Of course, this heterogeneity in erythropoietin receptor opens the door for a potential selective agonist which may activate one or the other receptor. There’s some evidence for this and what again this group showed is that if you use carbamylated erythropoietin and modified erythropoietin molecule is obliviated as C-EPO, this C-EPO is no longer able to activate the homodimer which is responsible for red cell production but is still able to activate the heterodimer and induce tissue protection and it is obvious that this opens a completely new avenue for the development of tissue protective agents.

Slide 11

In a very recent review article in Kidney International, one of the last issues, these authors actually show some data also indicating that carbamylated EPO, C-EPO is effective in preventing renal ischemic injury as you can see here on the histology sections, as well as when considering the macroscopic view of the kidney. So there’s a lot ongoing here in this field.

Slide 12

Now, the third aspect that I would like to address and you will see that it is also in someway related to the aspect of tissue protection concerns the molecular mechanisms of the oxygen dependent control of this system and central to this is the family of hypoxia inducible transcription factors, HIF.

Slide 13

Now many of you will have followed the elegant presentation by Peter Ratcliffe in one of the plenary sessions who developed very nicely how the knowledge accumulated over years explaining how HIF is regulated and I would just like to summarise very briefly the basic components of this system.

Slide 14

When we talk about HIF, we’re talking about a family of transcription factors and we are in particular talking about two isoforms HIF-1 and HIF-2. Both are heterodimers consisting of an oxygen regulated alpha subunit which is shown here in colour blue or green and a constitutive beta subunit. The main regulation occurs at the level of these alpha subunits and as shown here, in the presence of oxygen these alpha subunits are rapidly degraded by the ubiquitin protein pathway and in order to achieve this degradation a complex of molecules, a ubiquitin ligase complex has to capture the HIF molecules and then transfer them to the proteasome after the addition of the ubiquitin molecules. The essential molecular step here which is required for the binding between this ubiquitin ligase complex and the HIF molecule is a hydroxylation reaction for which oxygen is required as a substrate and this means, very simple that in the absence of oxygen this substrate is lacking, therefore the hydroxylation reaction does not occur, there’s no binding and therefore the degradation process can not be initiated, HIF is stable translocates to the nucleus and can activate its target genes.

Slide 15

A critical component, of course, in this scheme is a family of enzymes which is responsible for these hydroxylation reactions and since these hydroxylations take place on prolyl residues, this family is a family of prolyl hydroxylases. The complexity of this comes from the fact that there are at least three prolyl hydroxylases which have been identified, so far there are two prolyl residues and there’s an additional asparagyl residue which needs to be hydroxylated for full HIF activity. Some of the activities of these enzymes maybe complementary but nevertheless it’s interesting that one of them, prolylhydroxylase 2 appears to be of central importance because as you can see here knocking down this prolylhydroxylase 2 as you can see here using siRNA is sufficient to induce the HIF-1 alpha protein and the same can be seen when we overexpress, when we knockdown different PHD isoforms, the PHD2 knockdown is associated with an increase in the activity of HIF dependent reporter genes.

Slide 16

One of the most interesting aspects of this work is that although it all started by identifying the molecular mechanism of erythropoietin regulation it is clear meanwhile that this is not limited to the regulation of erythropoietin but HIF is an ubiquitously expressed transcription factor, it does regulate a large ray of hypoxia sensitive genes and that overall is responsible for mechanisms that increase the oxygen supply or hypoxia tolerance with erythropoietin being one important aspect for this but many other additional aspects need to be considered here as well.

Slide 17

This is a list of HIF target genes published a couple of years ago by Patrick Maxwell in a nice review article and as you can see here, the list is large, it is increasing and the genes that are listed here are very heterogeneous but they can be grouped to some extent into genes responsible for blood oxygen carrying capacity, for vascular architecture and tone, for energy metabolism but also cell proliferation, pH regulation and many other aspects. Overall, most of these genes do in some way contribute to the adaptation to hypoxia either by improving the oxygen supply or by improving hypoxia tolerance at either the cellular, regional or the systemic level.

Slide 18

We developed a particular interest recently in trying to understand better why we have two HIF isoforms and what the specific roles of these two different HIF isoforms may be. This is again an siRNA knockdown experiment in which either HIF-1 alpha or HIF-2 alpha were knocked down in cell culture and combining this with an ephimetric shift approach we have been able to identify a number of genes which are only induced or only upregulated or down regulated by HIF-2 alpha as compared to a larger number of genes which are only upregulated or downregulated by HIF-1 alpha. There’s some overlap here, interesting overlap between both groups but there’s also separation. From the aspect of the feedback regulation of erythropoietin the most interesting observation is that erythropoietin appears to be a HIF-2 target rather than a HIF-1 target.

Slide 19

This is obvious already when looking at immunohistochemistry in the kidney and in the liver the two sites which produce erythropoietin where those cells which are responsible for erythropoietin production, the peritubular fibroblasts in the kidney and the hepatocytes in the liver express HIF-2 alpha and not HIF-1 alpha. In addition to that at a cellular level when cells are investigated which can express both HIF-1 alpha and HIF-2 alpha and one or the other downregulated erythropoietin is sensitive to a knockdown of HIF-2 alpha but not of HIF-1 alpha, complementary evidence that HIF-2 alpha is obviously important.

Slide 20

So you might the question this is all nice but is this in anyway clinically relevant for understanding the feedback regulation of erythropoietin production or oxygen dependant control of genes if we know these molecular mechanisms?

Slide 21

Well, my answer would be definitely yes and one of the interesting aspects is that the knowledge about these molecular interactions allows us to explain a number of rare polycythemias in which there’s obviously over expression of erythropoietin in this feedback circuit. In fact, alterations in any of the molecules that I’ve introduced here which are important for the degradation of HIF under normoxic conditions can lead to stabilisation of HIF and upregulation of erythropoietin production. The most prominent example are mutations in the Von Hippel-Lindau protein which maybe associated with the Von Hippel-Lindau syndrome but also occur in the majority of sporadic renal clear cell carcinomas. Inhibition of the functional activity of Von Hippel-Lindau inhibits the interaction with hydroxylated HIF and this inhibits HIF degradation, leads to upregulation of HIF in virtually every single cell, nucleus in the tumour here where the neighbouring normal or not tumourous kidney tissue is not affected and indeed a number of these cancers, not all of them but a number of them do express, do overexpress erythropoietin. Of course, we know that from a clinical point of view only a minority of them, of those patients are actually polycythemic, so there must be additional mechanisms to explain this discrepancy, the mechanism of the anaemia of cancer might sort of reduce the incidence of polycythemia in those overexpressing EPO and there must also be some additional regulatory steps between HIF overexpression and EPO which are yet poorly understood.

Slide 22

A second example is Chuvash polycythemia. This is a rare condition but it’s the most frequent type of congenital erythrocytosis, it’s endemic in the Chuvash province in Russia and the majority of affected individuals likely originate from a single founder event that occurred many, many years ago. The mutation is known and it’s a single point mutation leading to an aminoacid exchange in the Von Hippel Lindau protein which is not associated with a higher incidence of tumours but is associated with endemic polycythemia.

Slide 23

Finally, the third and very interesting observation that was very recently published in PNAS that also mutations in these prolyl hydroxylases in particular here in prolylhydroxylase 2 may lead to polycythemia. This is a family where the father, daughter and son are affected suggesting an autosomal dominant trait and a point mutation was identified in prolylhydroxylase 4 which leads to the loss of binding affinity between prolylhydroxylase 2 and the HIF alpha protein and thereby induces this polycythemia in the infected individuals.

Slide 24

The second aspect and this will be the final aspect that I will touch here concerns the use of HIF as a therapeutic target to intervene with this feedback circuit and to induce either erythropoietin or tissue protection.

Slide 25

In order to understand this we need to go back for a minute to the chemical reaction which I illustrated before, the hydroxylation reaction, it is listed up here and it requires not only oxygen as a substrate but in addition to that also oxoglutarate as a co-substrate. This opens a nice interface into cell metabolism but the important thing here from a therapeutic perspective is that since oxoglutarate is required for this reaction, oxoglutarate analogues which may compete with oxoglutarate in this reaction can actually be used as inhibitors of this family of prolylhydroxylases and thereby inhibits the hydroxylation of HIF, stabilizes HIF and induces HIF target genes.

Slide 26

So for the first time we have a second tool now to interfere with this feedback and either to induce erythropoietin or even induce other hypoxia sensitive genes.

Slide 27

This is not just theory but clinical reality has been demonstrated already a year ago at the previous congress where Andrzej Wiecek provided data showing that one of these inhibitors a compound called FG-2216, a prolylhydroxylase inhibitor which can be given orally to the patients is able to induce the endogenous erythropoietin production and this compound is being in clinical development as an antianaemic agent.

Slide 28

Possibly even more importantly if we look at this feedback again and do not limit ourselves to erythropoietin but remember that we’re talking about a ubiquitous feedback mechanism, regulation oxygen balance and ischemia tolerance, there are many more applications for these kinds of compounds.

Slide 29

Again I would like to concentrate here on the kidney and showing experiments from our own lab that have recently been published where we have used this approach to prevent ischemic injury. In this experiment a single dose of such a HIF stabiliser was applied to animals a couple of hours before nephrectomy of the right kidney and subsequent ischemia of the left kidney. This HIF stabiliser obviously is able to induce HIF in virtually every tubular cell here in the kidney cortex, this is a picture taken from the nephrectomised right kidney and subsequently during the ischemic injuries the left kidney which of course shows equal upregulation of HIF was protected and showed much lower increase in the serum creatinine concentration and this was associated also with much less histologic damage.

Slide 30

So there are obviously two ways here to intervene with this system and the question is what is the advantage and disadvantage of either of them. Now, in general interfering with single target genes has turned out to be somewhat problematic and there are many, many examples for that. Simply because single genes have a limited relevance, there’s biological redundancy and there maybe compensatory changes in the activity of other genes which limit the effect. On the other hand, if we interfere with HIF, this can be considered as a master switch affecting a multiple target genes many of them known, some may not even be known so far and thereby the response is likely to be a coordinated response that mimics the physiological response to hypoxia.

Slide 32

So, I will try to sum this up in a schematic drawing again. What I’ve tried to show you is that what started as a very specific feedback control of erythropoietin production and regulation of red cell production has now turned into a somewhat broader feedback control of adequate oxygenation with hypoxia inducible genes, and the HIF family of transcription factors taking a major part in this with erythropoietin definitely playing a major role in this but not the only role and with its main effect on erythropoiesis but additional effects on tissue protection that may contribute here.
With at least two possibilities for therapeutic intervention either by using erythropoietin, as we have been doing for 20 years now, by using modified compounds which maybe specific for one or the other aspects or by actually using HIF stabilisers to intervene at an earlier level. Which of these two approaches will eventually be more successful in clinical practice is open so far. I personally have a bias and believe that this is a more elegant approach for tissue protection but I admit that this is a bias

Slide 33

And I’d like to thank you for your attention.

Slide 34

Prof Goldsmith: Thank you Kai-Uwe for a very lucid and elegant exposition of what’s a phenomenally complicated area although when you explain it, it doesn’t seem that way. While we’re thinking of questions I’d like to ask you one. You showed us the nice data from Johnson from Kidney International where you could actually give erythropoietin 6 hours after an injury, renal injury and get protection. Now, if you go to HIF stabilisation, there are various steps before you actually get the protective mechanism to happen. Do you not think in clinical practice something that you could give after the injury which I think is possibly less likely with HIF stabilisers to be successful would be a better strategy?

Prof Eckardt: Well, that’s a very important question David and I agree with you that considering the mechanisms that gene expression is required for the final effect, it is less likely to be effective after the injury, nevertheless there are some data around that it is still effective after the injury and our own experiments in this field are ongoing, so I wouldn’t exclude the possibility that the stabilisers could also be effective when given after the insult.

Prof Goldsmith: Any other questions? I’ll ask you another question. Obviously HIF stabilisation, as you say could be a major switch to a whole panoply of genes in different areas. Is there any mechanism we know that says if you stabilise HIF by a certain means or in a certain way, you can preferentially switch on some of those gene families and not others?

Prof Eckardt: I mean that’s a critical issue of course. Obviously not all the genes respond to the system with the same dose response relationship and so it is possible, although it has not been demonstrated so far that some of the genes are more sensitive and it’s definite that erythropoietin is one of the more sensitive ones if you look at all of these congenital disorders which lead to polycythemia but no obvious other abnormalities. So considering the dose response relationship and considering the complexity with different prolylhydroxylases, different HIF isoforms and other elements which I have not demonstrated here during my presentation there is some possibility that it might be possible to achieve specificity to some extent. I’m not sure whether it will be possible to actually achieve 100% specificity but some preferential induction of a certain set of genes should be possible.

Question: Thank you for that wonderful talk. So, now we’re starting to understand that erythropoietin is part of a paracrine mechanism for tissue defence in the setting of hypoxia. But in the kidney erythropoietin has really been set up as a hormone, so it’s made in the kidney for export. I’d like your thoughts on why it was the kidney that was chosen to be this important organ and you can reflect on that and then I’ll just give you a hint on some of my biases. How do you think the regulation of erythropoietin in the kidney is linked to the fluid and electrolyte balancing of the kidney as such that both these two fluid, the red cell mass and the plasma volume have to come together to form a hematocrit which in itself is a regulated factor?

Prof Eckardt: Well, that’s, of course, a question that has puzzled investigators over many, many decades and there’s no really convincing answer to it. One of the aspects of an answer could be that from a developmental point of view it’s interesting that some lower organisms actually produce red blood cells in the kidney and that this is some sort of the residual of that, that in mammalian organisms we do no longer produce red blood cells in the kidney but still have the major control mechanism located in the kidney. Of course it doesn’t answer your question as to why then the lower organisms had red cell production and volume homeostasis concentrated in one organ. A second possible explanation which is also not entirely satisfactory is that the kidney to some extent, let me put it this way, to some extent changes in blood supply of the kidney will alter both the amount of oxygen delivered to the kidney but also the amount of oxygen that is consumed because oxygen consumption depends on sodium reabsorption thus is linked to kidney reperfusion forgetting a moment about the autoregulation. This is unique situation. In many other organs when you increase blood flow, you increase oxygen availability and from a theoretical point of view it makes the kidney a bit less sensitive to changes in blood flow in terms of its tissue oxygenation.

Prof Goldsmith: Ok that’s fantastic. Thank you very much Kai-Uwe for a very, very wonderful talk.