CME Slides Forum

A joint Congress by ERA-EDTA and ISN

NEW EXPERIMENTAL DATA ON THE PATHOGENESIS OF ANCA ASSOCIATED VASCULITIS

Renate Kain, Vienna, Austria

Chair: R. Kettriz, Berlin, Gemany

Charles Pusey, London, UK

kain

Dr Renate Kain
Institute of Clinical Pathology
Medical University of Vienna
Vienna, Austria

Well, thank you. Ladies and Gentlemen, I would like to thank the organisers for giving me the opportunity to present work

which not only I myself have worked on over the years but many co-workers and collaborators in Vienna and Aberdeen foremost I’ll mention Ricarda, Markus and Reinhard, as well as Carol who have set up the ELISA. Agnes, Ingrid, Oliver, Ecevit and Andrea who do the cell biology work and the Western Blots. Dawn who performed the beautiful animal experiments.

It is work that focuses around the pathogenesis of focal necrotising GN. It’s the renal form of systemic vasculitis, of small vessel vasculitis which initially affects small necrosis, segmental necrosis of the glomeruli which is followed by extracapillary proliferation and then it leads subsequently if not treated, to destruction of the glomeruli and a loss of the filter function of the glomerulus.

Most - almost all cases of small vessel vasculitis in the kidney are associated with the occurrence of ANCA: anti-neutrophil cytoplasmic antibodies. Although today it is without any doubt that antibodies MPO and PR3 are pathogenic, the work that I will present is based on the question whether there might be actually additional targets for the autoantibodies that are present on the surface of endothelia and granulocytes as those are the targets of the injury and neither MPO and PR 3 are expressed on the cell surface of these cells.

In work that has been long published we found a group of autoantibodies that are directed against antigens, membrane glycoproteins on the cell surface of endothelia in the glomeruli, peritubular capillaries and the glomeruli themselves and the granulocytes. These are mainly two antigens, the one we provisionally designed gp130 resides as you can see here in the Immunogold Labelling primarily on the glomerular endothelium, on the cell surface of these cells while the second molecule which is the human lysosomal membrane protein is located as you can see here, in the vesicles indicating the lysosomes but also in up to 30% on the cells surface of these cells.

The human lysosomal membrane glycoprotein 2 which we followed up further seemed a very attractive candidate for an autoantigenic target - not least because it’s present on the cell surface, it’s also present in the lysosome and also because of its function. It was meant originally to protect the lysosome from self-digestion due to its large carbohydrate moiety which can take up to 70% of the protein. It also shuttles between the cell surface where it’s available to the antibody and to the endosomal and lysosomal compartments. It was meant to be an adhesion molecule and more recently it has been discovered as one of the major players in cellular homeostasis, in autophagy and antigen presentation.

We therefore, confirmed whether a larger population of ANCA positive patients did have antibodies to the LAMP-2 molecule and we set up a small research ELISA in which we tested 84 patients about half from Aberdeen and half from Vienna. It turned out that 92% of these patients did have in an active stage of disease either at presentation or at confirmed relapse antibodies LAMP2 which declined when the patient went into remission and were virtually absent in the control groups.

These antibodies occurred, as one would expect, together with antibodies to MPO and PR3 in about 48% and 46%. (Here are all the counter double positives,) we had a few double positive patients that had antibodies to both MPO and PR3 although I have to say they (antibodies to LAMP-2) also occurred in a small group of patients that were negative for both antigens.

The next question that we tried to answer is, since they are obviously present, are they really pathogenic? We tested them in vitro on isolated granulocytes later on also on monocytes and on microvascular endothelium. In this case we incubated the cells not with sera because they might have multiple specificities but with monoclonal antibodies to the LAMP2 molecule and you see here two representative experiments in which is shown the so-called shape change assay in which granulocytes are shown to change their shape in response to the antibody and the actin condensation. When the cells were incubated with 10 µg/ml of LAMP-2 antibody that is called H4B4 one can clearly see they settle down, flatten and change their shape, they are not round anymore, the most pronounced change is after stimulation with TNF-α. Similarly as one would expect, the same happens with antibodies to PR3 and monoclonal antibody to PR3 but not with an antibody to CD4 which is not expressed on granulocytes.

Similar results we obtained when we incubated glomerular and microvascular blood endothelial cells with the antibody H4B4. You see here the results of the blood endothelial cells. It is clearly shown that these cells have an expression of the stimulation, upregulate E-selectin. But what we also observed is an increase in cell death more clearly visible here in the chart where one can see that TNF-α 10 ng/ml TNF-α induces cell death.

These are the caspase 3 positive cells per 10 to the 5th cells but even more so (are present after) LAMP-2 (treatment). CD4 again not expressed in endothelial cells was negative also PR3 antibodies did not have an effect as it is not expressed on microvascular endothelium.

We don’t have an idea yet as entirely how these antibodies cause the damage to the endothelial cells but what we do know is that these antibodies are actually actively taken up into the cells. The cells were over a time course of up to 20 hours incubated with H4B4 antibodies and after fixed time points starting at 0 minutes up to 20 hours fixed and directly stained with anti-mouse IgG intracellularly and the red points here indicate the antibody that had been taken up and had been detected with the fluorescent labelled secondary antibody. This seems to be a specific process because CD4 antibody, which could potentially be unspecifically taken up, is not taken up and of course, untreated cells and cells with TNF-α did not show any staining intracellularly for mouse IgG either.

So the next logical step was to test the hypothesis that these antibodies are injurious in an in vivo model. We injected WKY rats that are prone to glomerular injury with this time a polyclonal antibody and IgG to LAMP-2 and what we saw after 24 hours were first small areas of necrosis in the glomeruli which then at about 48 hours were followed by crescent formation, proliferation of Bowman’s capsule and epithelia and after 5 days they started to change into fibrous crescents - so a similar progress of disease as one would expect in humans. None of the control animals showed similar changes.

That this process was really pauci-immune is illustrated in the next picture. There is no specific deposition of immunoglobulin either in the mesangium or in the periphery of the capillaries. What one sees after 2 hours is just unspecifically trapped rabbit IgG in the capillary loops and here there is a capillary with a mononuclear cell which has apparently, similar to the in vitro studies, taken up the antibody to the LAMP2. After 24 hours there’s just a minor background staining still visible which is completely absent after 48 hours or in controls.

So having confirmed that the antibodies to LAMP-2 cause injury it was now interesting to ask ourselves the question, are there specific epitopes on the LAMP-2 molecule that are recognised by patients’ antibodies? To make a long story short we did several assays, peptide mapping, epitope mapping and we found in patients in active disease mainly two epitopes and I’ll show you only one because it’s the one on which the further work will focus. We called it the peptide P41-49 according to its position of the aminoacid sequence in the mature chain of the LAMP2 molecule.

Using a synthetic peptide we tried to inhibit the binding of positive LAMP2 sera to the antigen and we could inhibit with synthetic peptides confirming that this epitope is really recognised by the patient sera. The controls of course, were negative in this test.

Already with the idea of molecular mimicry in mind we investigated this epitope for the presence of any epitopes or sequence homologies shared with pathogenic bacteria. Both -- Wegener who in ‘36 already suggested for his Wegener’s granulomatosis that there might be a bacterial association, later on also others confirmed the staphylococcal association with Wegener’s granulomatosis. But interestingly enough there were no stringent homologies to any bacterial proteins other than one that was rather surprising because it’s a bacterial protein of rather common pathogens. It’s the fimbrial adhesin FimH of gram negative bacteria that bear type 1 fimbriae. It’s present most likely in the pathogenic E.coli, Klebsiella, Proteus and Salmonella as well. It has a very stringent upregulation on the cell surface of these cells depending on their virulence.

Here is a schematic diagram of the protein, it is the last, on the tip of the type 1 fimbriae or pili that these bacteria have on a series of sequentially expressed proteins (in E.coli). You see here a 3D model in 2 orientations of the tip of FimH in which in green you see modelled the homologous sequence which is really lying in the so-called mannose-binding pocket on the surface of the mannose-binding pocket indicating that it is exposed to the immune system and available for binding. With these mannose-binding pockets the bacteria adhere to mannose residues on proteins on the host cell surface.

Initially we tested in crude bacterial lysates to see if we could inhibit the binding to LAMP-2 and we used fimbriated pathogens like E.coli 12900 and Klebsiella and Proteus and non-fimbriated pathogens like non-toxicogenic and non-pathogenic E. coli and Staphylococcus aureus. And, if I may focus on the first line, those are the LAMP2 positive sera of active patients. Here is the percentage of inhibition that we could achieve by pre-incubating the sera with the crude bacterial lysates before the (ELISA) assay was done. You could see this E.coli 12900 and the other FimH bearing bacteria inhibit but not the non-FimH bearing bacteria. This was later then confirmed by recombinant FimH and we could also show that antibody binding to MPO and PR3 was not inhibited.

Last but not least again the question does it work in vivo? What we did is we immunised again WKY rats with FimH. We chose rats because we cannot use mice because in mice this homologous epitope does not exist, the LAMP2 in mice has a completely different sequence.

Not unexpectedly our rats responded nicely to the immunisation with FimH. We made an ELISA test and - not unsurprisingly and as we had hoped for, 6 of the 8 rats that are shown here also developed antibodies that cross-reacted with the LAMP-2. It is shown that the antigen is apparently localised in granulocytes most likely corresponding to LAMP-2. So we get a positive c-ANCA fluorescence.

One thing that we did as well - we used TitreMax for immunisation because it did not contain any other bacteria. It’s an adjuvant free of bacteria. So the only bacterial protein that the animals have seen with the immunisation is FimH. The controls remained negative. What we had hoped happened again in 6 of the animals - namely they developed antibodies that not only reacted with the human LAMP-2 but apparently there was a breaking of tolerance to the rat LAMP-2 and they also had positive rat LAMP-2 titres.

Now the question was: do they bind in vivo to the structures they should do? Just to show you before hand a quick picture that the antibodies against FimH are cross-reactive with LAMP-2, with the homologous epitope and they bind in human glomeruli to the cell surface and in rats they cause injury. This is (the result) of a 5 week experiment: after 4 immunisation (with FimH) one sees (glomerular) necrosis and the formation of crescents.

Noteworthy that 2 of the animals also developed evidence of lung injury, these 2 had either neutrophilic capillaritis, vasculitis in the lung or evidence of lung bleeding as you can see here in Prussian blue staining which indicates the iron taken up by the macrophages in the lung.

So to conclude, the experimental data that I have presented to you today, we believe that in some predisposed individuals there is a possibility that an exposure to fimbriated pathogens, to an infection with fimbriated pathogens breaks the tolerance to the human lysosomal membrane protein 2, that these antibodies are then available to bind to the antigens on the cell surface, activate endothelial cells and granulocytes where the antigen is present and cause either alone or most likely in concert with MPO and PR3 the injury seen in vasculitis. Thank you.

Chairman: Thank you very much for the great presentation and also for staying in time. Are there questions from the auditorium? Before everyone gets ready can I ask you a question? You were saying you tested 40 something MPO positive patients who also were LAMP2 positive is that right?

Chairman: The immunofluorescence seemed to be a c-ANCA pattern for the LAMP2 antibodies, so how does that fit?

Dr. Kain: This is a very old story. The fluorescence that we see with the LAMP2 is a c-ANCA pattern. When one looks at granulocyte preparations that are not cytospun but in which the granulocytes are actually allowed to settle on the cover slips, you can when you visualise them under fluorescence see a mixed pattern of c- and p-ANCA that this was originally described in ‘95 the cytospin seems to predispose the granulocytes either for the c- or the p-ANCA pattern more likely. That is the reason why depending on the preparation of the granulocytes you see both patterns in humans.

Question: Well people are thinking have you seen the anti-LAMP2 antibodies in patients with pauci-immune GN who are actually ANCA negative by fluorescence or ELISA?

Dr. Kain: Yes, we have a small group of these patients and I did not have a chance for time constraints to get into that. It seems actually that they do have antibodies to LAMP 2 and one can visualise them again on the granulocytes but those are patients that apparently recognise a glycoepitope on LAMP2 which the other patients don’t do or predominantly a glycoepitope and when one removes the carbohydrate cyto-chains from the granulocytes unspecifically one gets all of a sudden a granulocyte c-ANCA staining which can be attributed to it. There are also half of the patients 5% of the patients in total, so half of the patients are ANCA negative.

Question: Is there a difference in the phenotype of the patients between those who are LAMP2 positive and ANCA negative?

Dr. Kain: I can’t tell you, it seems to be but we have to look at the clinical presentation more clearly that those are renal limited patients.

Chairman: Are there any more questions? If not then I think we’ll go on. Thank you very much again for this exciting data.