The Manchester Group had been actively investigating the pathogenesis of membranous nephropathy (MN) since the 1970s. The global interest in understanding MN led to Manchester hosting the 1st International Symposium on Membranous Nephropathy in 1990 at the Whitworth Art Gallery in Manchester. This was attended by the leading researchers in all aspects of MN from across Europe and USA.
Contents
We still had no clear understanding of the immune mechanism causing MN in man at the turn of the 21st century. The diagnosis of MN was still based on clinical presentation and kidney biopsy findings. An immune component to the glomerular pathology was apparent from the presence of glomerular IgG and C3 immune reactants in biopsies and supported clinically by the induction of complete or partial remission of proteinuria using several types of immunosuppression.
But despite intense study of experimental models in rats and rabbits over the previous 40 years, and some understanding of possible immune mechanisms, no definite pathway or antigen target was apparent in human MN. The role of the IgG and complement present in the glomerulus remained a mystery and the putative influence of a reported HLA Class II genetic association was obscure.
In 2002, Pierre Ronco’s group in Paris identified the first human antigen to cause MN [1]. Although this was a rare genetic case involving a mother’s immune system being stimulated by a foetal antigen (neutral endopeptidase, NEP, on the podocyte) not possessed by the mother, it shed light on the role of IgG and showed that the involved antigen was expressed on the podocyte cell surface. In this scenario, the mother’s IgG could cross the placenta and access the NEP antigen in the baby’s glomerulus to cause in situ immune complex formation and complement activation, initiating glomerular injury in the infant who developed nephrotic syndrome and the histopathological features of MN. Fortunately, NEP was a well-studied protein available in pure recombinant form allowing confirmation of NEP as the target antigen. I’m sure Pierre Ronco’s group would have screened sera from idiopathic MN cases against NEP, we in Manchester certainly did and found no reactivity. Other rare cases involving NEP were subsequently identified but this antigen was not the common antigen of human MN.
One area of lab research becoming increasingly important from the 1990’s was human glomerular cell culture. Human mesangial cells had been successfully grown and passaged But the podocyte was the cell of interest in the study of MN. While primary podocytes could be established in culture, they were reluctant to proliferate to provide a useful large mass of cells for study. The work of Moin Saleem in Peter Mathieson’s group in Bristol in establishing a method to grow primary human podocytes in culture and subsequently differentiate them using a temperature sensitive genetic control element was invaluable in the field of MN research [2]. For the first time, MN researchers had an accessible podocyte antigen source which might contain the elusive MN autoantigen.
In Manchester, since the 1990’s we had made clinical and lab studies of human MN a major focus of our research programme. We had a biobank of MN patient samples – serum and urine collected on a regular basis from the weekly GN clinics by Shelly Williams with the support of consultants and research registrars. On rare occasions, we were able to collect autopsy kidney tissue from MN patients. Our hypothesis was that the immune reactants (antibody or immune complexes) known to be deposited in the glomerulus could be used as specific reagents to identify potential glomerular target antigens. We had tested the practicality of this idea earlier in a case of tumour associated MN [3] and found that by extracting the glomerular eluate and radioactively labelling it, we could locate potential target antigens in the tumour extract when analysed on SDS gels.
By 2007 we had refined this approach by using biotinylated, Protein G purified, glomerular eluates rather than radiolabelled extracts and tested them on MN and normal kidney glomerular extracts generating enough data to present our findings at American Society of Nephrology in San Francisco in 2007 [4]. We had detected antibody eluate binding to high molecular weight antigens from several normal glomerular extracts but only when the antigen was run on the gel in the unreduced state. This implied that the antigen(s) were common to normal kidney and that the structure (epitope) recognised by the IgG depended on a disulphide stabilised conformation in order to interact with the antibody. We had also found that the IgG eluates which were highly enriched with IgG4 antibodies compared to the eluate from normal kidney.
Our frustration at this time was that these gels could only separate small samples; there was insufficient protein in the band to apply the gold standard techniques to determine the amino acid sequence of the antigen. These were in-gel trypsin digestion of the protein bands for mass spectrometry and tryptic peptide classification from large peptide databases.
At the poster session at ASN 2007, we discussed our work with many colleagues in the field. In particular, I remember talking through our data with David Salant and his new clinical research fellow Larry Beck from Boston. They were particularly interested in the gel patterns and size of bands identifiable on the gels. When I explained that we only detected bands on gels run in the absence of reducing agent but that when run in the reduced state we saw nothing, they looked at each other and smiled. I realised that I had said something of significance but at the time, the implications of what I had revealed was unclear!
We were already moving away from using glomerular extracts to using extracts of cultured human podocytes as the antigen source on which to test our glomerular antibody eluates by SDS gel analysis. Podocyte extracts could be better controlled using enzyme inhibitors during extraction to minimize protein degradation. In particular, we were investigating native preparative SDS gels that allowed us to load 10-100 times the amount of sample with the hope of retrieving enough target bands to analyse by mass spectrometry.
David Salant identifies the major autoantigen in human MN
In July 2009, a paper from David Salant’s group in Boston was published [5]. David had won the race to identify PLA2R as the main MN autoantigen!
There had been three main research groups trying to identify the autoantigen in MN; Pierre Ronco’s group in Paris, David Salant’s group in Boston and our Manchester group. All three were using a similar approach; employing an antigen source (glomerular extract or podocyte culture extract) separated on SDS gels followed by western blotting with patient serum or glomerular eluate. Candidate bands identified on gels were to be identified by in gel trypsinisation and mass spectrometry. Having spoken to Pierre and David since the discovery of PLA2R as the antigen, it seemed to me that David Salant it seemed had the most sensitive mass spectrometry analysis so he was able to get peptide sequence from SDS gels when Pierre Ronco and I could not. As the antigen, PLA2R, had already been described and cloned in 1995 [6] in a different research context , the PLA2R peptide sequences were already in the mass spec databases and, therefore, identification of PLA2R was rapid. What helped David to confirm PLA2R as the antigen was the availability of pure PLA2R antigen and antibodies to PLA2R from Gerald Lambeau, who had originally cloned PLA2R.
So, what was the nature of PLA2R, the main autoantigen in MN? It was a large (180kDa) multidomain podocyte membrane receptor whose native domain structure was dependent on multiple disulphide bonds. It had very limited expression in other tissues and its function in human cells was uncertain. PLA2R had been studied in rabbit and bovine cells and had been shown to bind PLA2. It is assumed in man to be a receptor for PLA2 based on the structural homology but no primary binding data for the human receptor has been published and the possibility exists that it serves a different function in human podocytes. So back in 2007, we had been on the right track in looking for a normal glomerular antigen whose structure was dependent on disulphide bonding.
Once David Salant’s 2009 paper [5] was published, we realised we would be able to confirm his findings in the UK using two different approaches, a genomic study, and also an ELISA bioassay for the autoantibody.
A genomic strategy had already been in progress in UK since 2008. UK nephrologists had obtained MRC funding in 2003 to establish the MRC Glomerulonephritis DNA Bank. A Renal Association consortium led by Prof Andy Rees, collected DNA, serum and a short clinical data set on at least 300 cases of each of the main patterns of glomerulonephritis (including MN) from renal centres across England and Wales. In 2008, Robert Kleta, Peter Mathieson and I established the UK MN Consortium to use the MN DNA samples (336 cases) in the first genome wide association study (GWAS) of SNP’s in MN. At that time, the consensus view was that several thousand cases were required if a GWAS was to have sufficient statistical power. However, Robert Kleta’s view was that MN exhibited several features that would allow sufficient power with much smaller case numbers. MN was a rare disease with a known HLA link, with a clear biopsy proven classification which minimised mis-classification and therefore he argued only several hundred cases would be required to identify risk SNPs. While Robert’s group at UCL started investigating the MRC MN cases, we contacted European colleagues – Jack Wetzels in Nijmegen, who contributed 146 cases, and Pierre Ronco in Paris, 75 cases. (This collaboration became the European MN Consortium). This allowed Robert to test his hypothesis in the Dutch, French and UK patients separately and in the combined group of 556 cases. Robert was correct and the two risk genes, PLA2R and DQA1, were detectable even in the smallest , the odds ratio for homozygosity of both risk alleles in the combined group a staggering 10-80 ! [7]. So ,in July 2009, when David Salant published PLA2R as the MN autoantigen, we already had the genomic signature of two risk genes PLA2R and DQA1. The MN Consortium paper was published by 2011 [7]. The UK MN group now represented the major nephrology centres researching MN, including Manchester (Paul Brenchley) Bristol (Peter Mathieson), London (Robert Kleta), Oxford (Ian Roberts) and obtained an MRC Project grant in 2013 to explore the new science around PLA2R and MN. The study enabled the largest collection of new MN patient DNA (over 800 cases) from most nephrology centres in UK, delivered by Jean Winterbottom, research nurse employed on the study. These new UK cases were used in an a global GWAS of 3000 cases led by Krzystof Kiryluk and Robert Kleta, which confirmed our 2011 paper, identified two additional risk genes NFKB1 and IRF4 and showed that in European MN cases, DQA1 0501 was the HLA risk gene but that in Asian MN cases this was DRB1 1501 [8].
The second approach for confirmation of the role of PLA2R as the MN autoantigen was to develop a quantitative ELISA for the autoantibody to PLA2R. Despite PLA2R having been cloned in 1995 there was no commercial or research supply of PLA2R. Having recombinant PLA2R would also facilitate the production of monoclonal and polyclonal antibodies to PLA2R which could be used in future research projects. I discussed the possibility of cloning and expressing PLA2R from scratch with my colleague Eddie McKenzie at University of Manchester. I had known Eddie for about 15 years and we had previously worked together on heparanase projects. Eddie was a brilliant molecular biologist and had experience of expressing the most difficult of proteins. His group cloned the full length PLA2R gene and by the end of 2011 had good levels of protein expression from mammalian cells, producing milligram amounts of protein. PLA2R is a large protein, it had proved difficult to express and we had heard rumours that other competing groups had tried without success. So we were able to establish the first ELISA for anti-PLA2R autoantibodies using full length recombinant PLA2R sequence and published the first papers on quantitative anti-PLA2R levels in different clinical states of MN from our own biobank samples in Manchester [9] and from samples provided by the European Consortium [10]. Over the next couple of years we were able to use the ELISA to support multiple research studies of anti-PLA2R in MN including the MENTOR Clinical Trial in North America [11] and studies by the European MN Consortium [12].
It became clear that anti-PLA2R is an important biomarker for over 80% of MN cases. It is specific for diagnosis of MN, and anti-PLA2R levels parallel disease activity although falling before reductions in proteinuria, albeit running on a different time base compared to proteinuria, and downregulation of anti-PLA2R consequent to therapy or spontaneous immune remission is associated with disease remission. This is the best evidence one can achieve in confirming pathogenicity of an autoantibody in man. Experimental models may in future define pathogenicity further, but for now are hampered by the absence of PLA2R in rodent glomeruli.
Availability of assays for anti-PLA2R took a different route from previous autoantibody assays in nephrology. In the 1980’s and 1990’s, assays for anti-GBM and ANCA, initially using Western blotting and then by ELISA were developed by numerous research groups who compared assays and standards at International Workshops in order to understand the complexity, sensitivity and specificity of such assays. This iterative process did not happen for anti-PLA2R autoantibodies. The University of Boston had patented assays for anti-PLA2R in 2009 and licenced the patent to a company which by 2013 had commercialised an immunofluorescence test, and later an ELISA [13], first in USA and eventually globally. The advantage of this approach meant that research groups across the world were using the same assay and therefore results were easily comparable. The disadvantage was that little information was available to researchers about the autoantibody serum used for standardising and calibrating the assay, how often the standard changed, which epitopes were recognised by the standard serum, and the avidity of the standard serum for antigen. Nor was it known how the standard serum calibrated an unknown serum containing high affinity antibodies to the dominant epitope on the one hand, or a serum with lower affinity antibodies to perhaps five different epitopes on the other hand. Unfortunately, in the absence of any international workshops on anti-PLA2R these questions remain unanswered. Even now the sensitivity of the assay for clinical diagnosis is still discussed as the company recommends a value for top of the normal range at 20u/ml but some clinicians use a level as low as 4u/ml.
Since 2014, when the commercial anti-PLA2R ELISA assay became globally available , many clinical studies have reported the usefulness of anti-PLA2R antibody levels for diagnosis of MN and for monitoring disease activity and response to therapy. The MENTOR clinical trial looked in detail at the usefulness of anti-PLA2R monitoring for predicting response to therapy using both the Manchester ELISA and the commercial assay [14] and confirmed the importance of levels of anti-PLA2R post therapy in predicting response to therapy.
In 2015 the Manchester group identified the dominant epitope on PLA2R in terms of the 3D amino acid sequence [15]. This work was a team effort with major contributions from Maryline Fresquet, Tom Jowitt, Eddie McKenzie and Rachel Lennon at the University of Manchester. Later, other groups identified additional minor epitopes throughout the PLA2R molecule. The idea of “epitope spreading” over time as disease progresses was proposed [16] but current evidence suggest that with sensitive assays, these antibodies to minor epitopes can be only be detected early in the clinical course. The precise role of these additional antibodies remains uncertain; perhaps they lead to enhanced deposition of antibodies in the sub-epithelial target space in the glomerulus which is more efficient at complement activation leading to greater damage and heavier proteinuria? We pursued our studies of the structure of PLA2R using cryo-EM, light scattering and amino acid substitution experiments. Improving on the 2015 paper, we identified that the domains containing epitopes fold together to form an immunogenic patch on the molecule [17]. Why this “patch” activates the immune system to produce autoantibodies remains a mystery.
Since the discovery of PLA2R, other podocyte antigens have been shown to account for a small percentage of MN cases. THSD7A accounts for 2-3% of cases and we described a polypeptide structure of the main epitope in THSD7A that resembles that of PLA2R [18].
With the accumulated evidence that reducing and removing circulating anti-PLA2R brings remission of proteinuria in MN, we carried out the first immunoadsorption study to remove anti-PLA2R in 12 cases of MN. [19]
Getting to grips with the epidemiology of MN in the human population has been challenging. We used the UK BioBank population for a life-time epidemiology study of MN pathogenesis, since all participants have had genome sequencing, so subjects with the risk alleles for MN can be identified. Within this group we can identify those that have developed proteinuria and those who have biopsy proven MN [20].
It remains in the future to add the anti-PLA2R status to the cohort and to follow those who are currently anti-PLA2R negative in order to identify what triggers anti-PLA2R positivity. Long term follow up of the cases with the high risk alleles for MN will show the strength of the genetic risk in causing MN.
Familial MN in twins was first reported in Manchester [21] Although rare, Familial MN cases continued to be reported in the literature with a male preponderance in about 65% of cases. The UCL genetics team led by Kleta and Stanescu had collected DNA and serum on several large families with multiple cases of MN. We found no circulating anti-PLA2R in these families but there was a risk locus on chromosome Xp11.3-11.22 in affected cases. The precise nature of this genetic risk remains to be defined [22].
Recurrent MN post transplantation occurs in up to 40% of cases and it became clear following the discovery of anti-PLA2R that return of the autoantibody was a likely cause of autoimmune damage to the graft. A European collaboration led by Pierre Ronco collected serum and DNA on donor and recipients including from patients in UK [23]. It was found that seven SNPs located between DRB1 and DQA1 and three SNPs in PLA2R1 region predicted disease recurrence. This offers the hope that in future, donor organs may be genetically screened to avoid such risk of recurrence.
Despite the rapid progress since 2009, many questions remain unanswered about the development of autoimmunity to PLA2R, the role of complement and the natural immune regulation evidenced by spontaneous remission in some with MN. Our hypothesis is that anti-PLA2R MN develops in genetically susceptible individuals (those having the pathological DQA1 alleles) consequent to an environmental immune trigger (possibly a common microbial agent) [21]. We described similar peptide sequences in PLA2R that are present in microbial strains of Candida and Clostridium that are likely to generate natural IgM antibodies in all humans as their immune system matures in infancy. This is normal protective immunity at work. However, as immune regulation degrades with aging in those who are genetically susceptible, loss of tolerance generates an antibody class switch from IgM to IgG resulting in a circulating IgG antibody able to bind to PLA2R. Since MN cases on presentation already display IgG4 antibodies to PLA2R, the class switch has occurred months to years before presentation. This chronic phase of antibody production also allows time for affinity maturation as we showed previously that anti-PLA2R antibodies display high affinity [15].
Anti-PLA2R measurement has improved diagnosis of MN (some clinicians now question the need to biopsy the kidney for diagnosis). The assay provides the ideal biomarker to monitor the outcome of treatment for the individual patient and for use in clinical trials. The challenge now is to translate this new knowledge of the autoimmune nature of MN into specific new therapies that improve outcomes for patients while reducing risk of comorbidity. My personal view is that the therapeutic goal should be to re-establish immune tolerance to PLA2R, which is probably what happens naturally during spontaneous remission.
Following my retirement in 2019, MN research continues in Manchester led by Patrick Hamilton and Durga Kanigicherla who seek to develop antigen-specific therapies to reduce and remove anti-PLA2R antibodies in MN.
Paul Brenchley. First posted July 2024
Last Updated on August 5, 2024 by neilturn