CME Slides Forum - N.J. Brunskill

A joint Congress by ERA-EDTA and ISN

TUBULAR TOXICITY OF PROTEINURIA

Nigel J. Brunskill, Leicester, UK

Chair: Richard Glassock, Los Angeles, USA

Thierry Hannedouche, Strasbourg, France

brunskill

Prof N.J. Brunskill
Department of Infection, Immunity and Inflammation
University of Leicester
Leicester, United Kingdom

Ok great. Good morning everybody. I want to thank you all for being here actually, there’s always a worry when you give a talk in the last session of a meeting that everybody will be at the airport rather than at the meeting itself. So I hope we don’t let you down. I’m going to talk to you about the tubular toxicity of proteinuria and you’ll see I’ve put a question mark after the title, simply because this remains quite a controversial area.

I’m going to give you some historical background as to how we’ve got where we are with our thoughts about proteinuria as a toxic moiety and then give you some fairly new data from our lab which I think reveals some interesting things about the way proteins interact in the tubule.

So as clinicians we’ve known for many years that there’s something uniquely bad about having protein in your urine and this is some data that was presented in a review by De Jong and Gansevoort recently representing the experience in 90.000 patients in Japan. In essence what it shows is that for whatever stage of CKD you may have, your chance of developing ESRD is roughly a 100 times greater if you have proteinuria. Secondly, if you have CKD 2 with proteinuria, your chance of reaching ESRD is greater than if you have CKD 4 without proteinuria. So something pretty bad about that. We also know that if you reduce proteinuria, it appears to be renoprotective.

So this is some data from the Italian Gisen group which shows that the reduction of proteinuria with ramipril is protective of renal function and the degree of protection appears to be proportional to the degree of proteinuria reduction suggesting but not proving a causal link between proteinuria and progression.

So alongside those clinical observations a laboratory hypothesis was developed that somewhere upstream here is a glomerulus that is damaged and the previous speaker has told you how glomeruli may be damaged and it leaks into the tubule a variety of abnormal macromolecules that are not normally found there. These are normally or most notably proteins and we define clinically abnormalities of glomerular permeability as proteinuria although it’s not just protein, there’s lipid and also a wide variety of molecules that may be carried into the proximal tubule as well. These molecules interact with the tubular cell; they damage the tubular cell and initiate a process that culminates in scarring and fibrosis.

So the most tangible clinical evidence that something like this may happen can be seen in renal biopsies of patients with heavy proteinuria. So this is a proximal tubule and the cells are heavily vacuolated by proteinatious material that has been reabsorbed from the lumen. I think it’s easy to believe that cells and structures with such disordered architecture might also have disordered function.

So over ten years ago we set out to look at the mechanisms of interactions of protein with proximal tubular cells. I’m going to focus on albumin because that’s what we’ve studied largely but I don’t want you to think that that’s the only protein that’s important or even that it’s most the important, it’s just the most studied. We incubated proximal tubular cells in culture with goal albumin and you can see if you look at the panel on the right, what we understand as the paradigm of proximal tubular handling of proteins. Binding of the protein into a kind of apical invagination which draws the proteins into the cell, into this multi-lamellar body which is part of the endosomal pathway and then a concentration in this electron dense structure which is a lysosome. So uptake, endocytosis, breakdown and the presumed reabsorption of the constituent aminoacids.

We asked a supplementary question to that, we wondered whether the binding of proteins particularly albumin into this invagination on the surface of the cell might send signals into the cell and change the way that it behaved. So not just an inert piece of cargo destined for destruction but a signalling molecule in its own right. To cut a long story short over the last 10 years or so work from our laboratory and work from other laboratories has shown that albumin incubated with proximal tubular cells activates a whole variety of signalling pathways. As a consequence it activates a variety of transcription factors.

So that type of work has been criticised and some of the criticisms have been that the experiments have used supra-physiological or even supra-pathophysiological concentrations of albumin. So I just want to take you back to some studies that we did about 10 years ago looking at PI3K activity in cells. So these are proximal tubular cells that have been incubated with recombinant human albumin, it’s not contaminated with anything that may be adsorbed from serum. We did really very careful even tedious experiments actually looking at various PI metabolites and we showed that activation of PI3K occurred at very low concentrations of albumin, physiological concentrations that are found in the normal proximal tubule.

So in fact, the signalling pathways activated by albumin in the tubule are not necessarily provoked only by high concentrations. We know that as a consequence of activation of these signalling pathways and transcription factors the proximal tubular cells will make a whole alphabet soup of different mediators and substances that will provoke an environment that will foster development of scarring and fibrosis.

But it’s also important to remember that proteins are not the only substances that accumulate in the proximal tubule in proteinuria. The previous speakers have already talked about the accumulation of lipid in the proteinuric glomerulus and lipid also accumulates in proximal tubular cells. It gets there largely bound to albumin actually because albumin carries fatty acids with it. The clinical manifestation of that is the appearance of fat bodies in the urine.

These are epithelial cells loaded with fat that have been killed by the fat because it’s toxic and I haven’t got time to show you the data but if you load albumin with fat, it’s more toxic to the cell than fat free albumin and these cells appear in the urine.

So you might imagine that if the uptake of these materials by the cells in the tubule was injurious, that blocking that process, the endocytic process, would be beneficial. So we came into some experiments based on the observation that some patients treated with statins developed proteinuria. We looked for the mechanism of that and we fund that statins inhibit the reabsorption of proteins by proximal tubular cells. This is a class effect so in some patients who were given statins they excrete proteins because the normal reabsorption is blocked. We also know actually that if we treat some proximal tubular cells with different thiazolidinediones that also blocks protein endocytosis.

So therefore, does that block the production of the abnormal injurious mediators that I’ve talked to you about? There’s a bit of a disconnect here actually. So if we incubate proximal tubular cells with statins, they block endocytosis and they block the production of MCP1 and I just used that as an illustration. However, if we block endocytosis with thiazolidinediones, it has no effect on the production of mediators.

We looked at this in a slightly different way with some colleagues in London using two different cell types the ok cell which shows very avid protein endocytosis and the HKC-8 cell which is a human proximal tubular cell, immortalised cell line which shows really trivial endocytosis. But both of these cell types respond to albumin by the production of TGF-β.

So the suggestion may be that at least endocytosis is not obligatory for these cells to demonstrate an injurious phenotype.

So if that’s the point if you don’t need to have endocytosis, how could these injurious effects be mediated? The previous speaker’s already talked to you a little bit about megalin. Megalin is the receptor in tandem with cubilin in the proximal tubule for albumin and for a number of other filtered proteins. Cubilin is almost entirely extracellular, it’s just anchored to the membrane. The megalin has a cytoplasmic tail which dangles in the cytoplasm and is larger than the other members of the LDL receptor family of which megalin is a part. If you look at the cytoplasmic tail of megalin, it has some interesting motifs that may be involved in signalling. So it has MPXY domains that are important for interaction with clathrin and endocytosis. But it has SH3 and SH2 domains and it has consensus phosphorylation sites for kinases like PKC and PKA.

So we set out on a number of experiments to see if there were any signs that the megalin cytoplasmic tail might be involved in signalling in proteinuria. We asked whether the megalin cytoplasmic tail was phosphorylated and if it was, was it agonist mediated? And if it was, was it mediated by megalin ligand interaction, or interactions with agonist acting via other receptors? What are the kinases and what might be the functional consequences?

We have the answers to some of those questions. So we took a megalin cytoplasmic tail GST fusion protein and we incubated the fusion protein with lysates from proximal tubular cells that had been stimulated with a variety of relevant agonists.

So we chose albumin and EGF and -- as a stimulator of PKC and we incubated them in the presence of P32 and then we washed the GST fusion proteins, run then out on a gel and also autoradiographed it. This is phosphorylation under non-stimulated conditions and this is stimulated phosphorylation of the megalin cytoplasmic tail in response to a variety of agonists. We then wanted to see if we could immunoprecipitate phosphorylated tail from intact cells. These were formidably difficult experiments. We tried to immunoprecipitate the full length megalin which is an enormous protein and we found that impossible.

So we developed a protocol where we clipped off the extracellular components of megalin by light trypsinization of the cells, we then loaded them with P32, stimulated them and pulled down the cytoplasmic tail with anti-sera that we raised in our laboratory.

We were able to show that the cytoplasmic tail of megalin in intact cells is phosphorylated. After the post-hoc who was doing that work had irradiated himself sufficiently with the enormous quantities of P32 he needed to use for those studies, we decided we’d use an alternative approach and we developed a chimeric mini-receptor whereby we fused the cytoplasmic tail of megalin to the transmembrane and extracellular domain of a protein called CD8 which we know is correctly targeted when transfected into proximal tubular cells. This is what it looks like.

So this is the challenge. This is just immunostaining for the transfected receptor and this is a composite z series which has just been rotated a little bit. I think you can get a feel for the 3D structure and that we do have expression of that cytoplasmic tail on the cell membrane. Then we immunoprecipitated the mini-receptor from P32 loaded cells and we were again able to show that this was phosphorylated in a similar way.

We now know having mapped the phosphorylation sites on the megalin cytoplasmic tail that it’s phosphorylated on something there, that serine, that serine, that serine that threonine and that threonine in an agonist dependent way. So we now have the possibility of developing phosphoro specific antibodies to study the activation state of megalin in proteinuric conditions and we haven’t been able to that in any kind of human samples as of yet. Some studies from proteinuric mice that have a kidney directed megalin knockout were also very interesting. So the kidney specific megalin knockout is a mosaic, so some proximal tubular cells have megalin and others do not.

So that’s quite an interesting model for studying adjacent cells that express megalin and those that don’t. If you make these mice proteinuric, the megalin expressing cells show stress responses in proteinuria with the expression of heme-oxygenase, they express MCP1 and they show apoptosis. One observation from this study that hasn’t been widely picked up on is that actually up to 15% of the megalin negative cells still showed albumin endocytosis.

So the question is what could be mediating that? If these cells don’t have megalin, they must have something else that’s mediating uptake. I just want to spend the last few minutes of the talk telling you about some new evidence of a new receptor. We considered two candidates as potential albumin receptors in the proximal tubule in addition to megalin. We thought about the neonatal Fc receptor which binds albumin in endothelial cells and it directs albumin to a transcytotic pathway and increases the half-life of albumin in animals. It’s involved in protein endocytosis by podocytes where it mediates uptake of immunoglobulin, it’s expressed in the proximal tubule and it’s been shown to bind immunoglobulin in the proximal tubule.

We considered CD36 which is another multi-ligand scavenger receptor which is also expressed in the proximal tubule. It binds fatty acids and we wondered sort of intuitively whether CD36 or a molecule like it could potentially bind fatty acids that were stuck to albumin and help to bring albumin into the cell in that way. CD36 is upregulated in proteinuria in the proximal tubule and it also has some signalling functions.

So in fact, we went on and chose CD36 and we established some stably transfected cells expressing a FLAG tagged CD36 and here are a number of clones expressing different levels of CD36. This is quite a useful tool we have a kind of dose response of CD36 expression to study.

If we look at the binding of albumin to these cells, we find that the binding of albumin is increased in proportion to the amount of CD36 over expression. This is shown in the graphs and we can also see the increased binding with the immunofluorescence on this side.

Wild type cells show fairly low grade albumin binding to the cells, no CD36 expression, whereas the CD36 transfected cells show much more uptake, very profound expression of CD36 and co-localisation of binding with CD36. If we look at the kinetics of uptake that’s driven by CD36 in these cells, it’s different actually to that in the wild type. If you just look at the graph at the top here which represents this early part of the curve, you can see that the shape of that uptake curve is rather different. This is a much more avid system for bringing albumin into the cell. The affinity of albumin is similar but the transport process appears to be different.

So I want to summarise just by saying that tubular luminal macromolecules alter tubular cell function. I don’t think there’s intuitively any difficulty in believing that really, I’m sure that tubular cells respond to altered microenvironment just the same as many other cell types do in their particular environment. I think the tubular cell response to proteinuria is related to the quality and the quantity of proteinuria involved. I don’t think endocytosis is necessarily a prerequisite. I think the tubular protein receptors are probably also signalling receptors and there are probably receptors other than those already described. Thank you very much for your attentions.

Question: In minimal change glomerulopathy we’ve got heavy proteinuria, heavy albuminuria, lipiduria and still we rarely have glomerulosclerosis. So do you think that these factors need special pre-conditioning to --- the kidney if we compare with minimal change?

Prof. Brunskill: This is a common question actually and it’s a very obvious clinical question. There is some early data from Italy actually, about 20 years ago that showed the proteinuria in minimal change disease was qualitatively different. So in fact the albumin in minimal change disease appears to have quite a different fatty acid composition to the albumin that is excreted in the urine in other types of nephropathy. I do believe actually that the lipid component of the filtered protein is very important in directing the tubular response. So I think that’s certainly an element. Many patients with minimal change disease obviously have proteinuria of relatively short duration and of course, some people believe that minimal change disease in some individuals is kind of an early stage of FSGS. So there are some patients who may not respond to immunosuppression who actually don’t have minimal change disease, who have something that’s more aggressive.