Strategies Towards Antigen-Specific Treatments for Membranous Nephropathy

Membranous nephropathy (MN) is a rare but potentially severe autoimmune disease and a major cause of nephrotic syndrome in adults. Traditional treatments for patients with MN include steroids with alkylating agents such as cyclophosphamide or calcineurin inhibitors such as cyclosporine, which have an undesirable side effect profile. Newer therapies like rituximab, although superior to cyclosporine in maintaining disease remission, do not only affect pathogenic B or plasma cells, but also inhibit the production of protective antibodies and therefore the ability to fend off foreign organisms and to respond to vaccination. These are undesired effects of general B or plasma cell-targeted treatments. The discovery of several autoantigens in patients with MN offers the great opportunity for more specific treatment approaches. Indeed, such treatments were recently developed for other autoimmune diseases and tested in different preclinical models, and some are about to jump to clinical practice. As such treatments have enormous potential to enhance specificity, efficacy and compatibility also for MN, we will discuss two promising strategies in this perspective: The elimination of pathogenic antibodies through endogenous degradation systems and the depletion of pathogenic B cells through chimeric autoantibody receptor T cells.


INTRODUCTION
Membranous nephropathy (MN) is a rare but potentially severe kidney disease and a major cause of nephrotic syndrome in adults. According to the new KDIGO 2021 Clinical Practice Guidelines, a nephrotic syndrome is defined as proteinuria of more than 3.5 grams per 24 hours or a protein-tocreatinine ratio of more than 3 g/g in combination with low plasma albumin, peripheral edema, and hyperlipidemia (1). The descriptive term "membranous" refers to the prominent change that is classically seen in light microscopy: a diffuse thickening of the glomerular basement membrane (2). Additionally, granular depositions of immunoglobulins and complement components can be detected by immunofluorescence microscopy, suggesting a role of both autoantibodies and the complement system in the pathogenesis of MN. The hallmark findings in electron microscopy include electron-dense deposits in a subepithelial localization, i.e. on the outer aspect of the glomerular basement membrane (GBM), and an extensive effacement of podocyte foot processes.
The current view on the pathogenesis of MN ( Figure 1) is that predisposing factors, such as an underlying genetic disposition and/or immune dysregulation, in combination with an initiating trigger such as an infection, a malignancy or environmental factors lead to loss of tolerance for the respective autoantigen with consecutive activation of B cells and production of autoantibodies (11)(12)(13)(14)(15)(16)(17)(18). B cells differentiate towards plasma and/or memory B cells, which in turn produce large amounts of autoantibodies. These autoantibodies reach the kidney via the circulation and bind to their target antigen, which is assumed to induce damage to podocytes with subsequent loss of plasma proteins to the urine, likely via complementdependent and complement-independent mechanisms (19). Complement-independent mechanisms have been ill-defined so far, but may include modification of cellular signaling, antigen/receptor blocking or stimulation or interference with the antigens' biological function, e.g. enzymatic activity. Importantly, the direct pathogenicity of autoantibodies has been demonstrated by transfer experiments of patient-derived anti-NEP and anti-THSD7A autoantibodies, which cause MN in animals (3,20). Even though such a mechanism has not been demonstrated for anti-PLA2R1 autoantibodies, the association of high autoantibody levels with an unfavorable clinical outcome in patients with PLA2R1-associated MN and the development of MN transgenic mice expressing the murine PLA2R1 after transfer of rabbit anti-PLA2R1 antibodies strongly argue for a pathogenic role of anti-PLA2R1 autoantibodies (21,22). Alternative proposed mechanisms of immune deposit formation in MN include glomerular deposition of preformed immune complexes and binding of circulating antibodies to a planted antigen (23,24), but the relevance of these mechanisms is less clear. FIGURE 1 | Current view on the pathogenesis of membranous nephropathy and potential targets for antigen-specific treatments. Predisposing factors such as underlying genetic dispositions and/or immune dysregulation in combination with an initiating trigger such as an infection, a malignancy or environmental factors may lead to loss of tolerance for the respective autoantigen with consecutive activation of B cells and production of autoantibodies. B cells differentiate to plasma and/or memory B cells, which in turn produce large amounts of autoantibodies. These autoantibodies reach the kidney via the circulation and bind to their target antigen (e.g. PLA2R1 or THSD7A), which induces damage to podocytes with subsequent urinary loss of plasma proteins. Complement-dependent and complementindependent injury mechanisms are likely to be involved. Depletion of autoantibody-producing B cells or elimination of pathogenic autoantibodies constitute potential antigen-specific treatments for MN.

CURRENT TREATMENTS FOR MN
The clinical outcome in patients with MN is highly variable with about one-third of patients experiencing spontaneous remission within one year after diagnosis, whereas another 20-30% develop end-stage renal disease within 10 years (25). The identification of patients with an unfavorable outcome, which would benefit from early immunosuppressive treatment, represents a major clinical challenge. Classical treatment regimens for MN include the use of steroids in combination with alkylating agents or calcineurin inhibitors (26)(27)(28)(29). Recently, two large prospective clinical trials have investigated the use of rituximab in the treatment for MN. The MENTOR trial compared rituximab with cyclosporine and found rituximab to be superior to cyclosporine in maintaining remission for up to 24 months (30). The RI-CYCLO trial found comparable remission rates with the use of rituximab and cyclophosphamide and no significant differences in the safety profiles of these medications (31).
The novel KDIGO guidelines define clusters of patients according to the risk of disease progression and loss of renal function. Categories comprise low risk, moderate risk, high risk, and very high risk, depending on proteinuria, serum albumin, estimated GFR (eGFR), and anti-PLA2R1 antibody levels (1). Patients with a low or moderate risk of disease progression are usually treated with optimal supportive care (e.g. antiproteinuric therapy with renin-angiotensin aldosterone system inhibition (RAAS) and blood pressure control) and monitored for 3-6 months. In case of worsening proteinuria, eGFR or antibody levels, patients should be evaluated for treatment with rituximab or calcineurin inhibitors. Patients with a high risk of disease progression should be treated with rituximab or cyclophosphamide plus steroids or a calcineurin inhibitor plus rituximab and patients with a very high risk should receive cyclophosphamide plus steroids (1).
While calcinerin inhibitors and cyclophosphamide are broad immunosuppressants, rituximab is targeting B cells expressing the surface marker CD20. Initially developed as a lymphoma treatment, it acts through different mechanisms of action: i) Antibody-dependent cellular cytotoxicity (ADCC) through either NK cells, monocytes or granulocytes (32-34) (ii) apoptosis of B cells through caspase 3 activation (35) and iii) complement-dependent cytotoxicity (36,37).
In summary, the targets of autoimmunity in MN have been increasingly understood and therapies have shifted from broad immunosuppression using alkylating agents and calcineurin inhibitors towards a more pathogenesis-based treatment targeting autoantibody-producing B cells using rituximab. However, rituximab treatment may entail resistance towards apoptosis, ADCC, CDC and downregulation or loss of CD20 (38), which can decrease treatment efficacy. It also affects preexisting protective antibodies, reduces the body's ability to generate an immune response against foreign organisms, and greatly decreases the response to vaccination. These are undesired effects of B cell-targeted treatments, in particular in the setting of a worldwide pandemic (39), and may limit the use of agents such as rituximab in the treatment of antibodymediated diseases. Additionally, the increasing knowledge on specific antigens in MN principally enables antigen-specific treatment strategies, which would ideally target the immunological disease mechanisms while sparing protective immunity. Such antigen-specific therapies would have enormous potential to enhance specificity, efficacy and compatibility. Hence, there is a considerable gap between the increasing knowledge on the pathogenic role of autoantibodies and autoantigens in MN on the one side and the currently available treatments with limited specificity on the other side.
In the following, we will discuss two promising strategies that could be applied in the field of MN: The elimination of pathogenic antibodies through endogenous degradation systems and the elimination of autoreactive B cells through chimeric autoantibody receptor T cells (Figure 1).

NO SWEEPING UNDER THE RUG: THE SWEEPING ANTIBODY TECHNOLOGY
Physiologically, immunoglobulins (IgG) are constantly taken up by endothelial cells and shuttled to the sorting endosome. At pH 6 inside the sorting endosome IgG binds to the neonatal Fc receptor (FcRn) and is transported to the cell surface. Now at pH 7.4 the IgG is released from the FcRn back into the circulation due to a reduced affinity at neutral pH. This system prolongs the half-life of IgG in the circulation to approximately 30 days (40). Consequently, antibodies may also prolong the half-life of their target antigen by constant recirculation from endothelial cells back to the plasma.
Sweeping antibodies as potential therapeutics to remove soluble antigens from plasma were first described by Igawa et al. (41). Sweeping antibodies are engineered IgG that remove soluble antigens from the circulation, which is enabled by two distinct modifications: a pH-dependent binding to the antigen and an increased affinity of the antibody Fc part to an Fc receptor (FcR) (41). A pH-dependent antigen binding can be achieved by histidine mutagenesis, meaning that amino acids in the Fab region of the antibody are substituted for histidines. In the acidic environment of the sorting endosome, the protonated histidines lead to conformational changes in the Fab region of the sweeping antibody, which results in weakening of the binding to the antigen and finally dissociation of antibody and antigen. Enhanced immune complex uptake into the endosome is achieved by mutations in the Fc region, increasing the binding to either FcRn or FcgRIIB.
After injection, a sweeping antibody binds its target molecule in the circulation. The resulting immune complex is then taken up into endothelial cells by pinocytosis or FcR-mediated endocytosis. In the sorting endosome, the pH shifts from neutral to slightly acidic (pH 6) and the antigen is released from the sweeping antibody. While the sweeping antibodybound by the FcRnis shuttled to the cell surface and released there, the antigen is degraded in the lysosome. The enhanced affinity to the FcR allows efficient recirculation of the therapeutic sweeping antibody from endothelial cells to the plasma, which in turn leads to further removal of antigens from the circulation, creating the "sweeping" effect. Several different mutations have been described that enhance active internalization of antigenantibody complexes via FcRn, leading to efficient degradation of the antigen (42)(43)(44).
Sweeping can also occur through the FcgRIIB (45). FcgRIIB is the only inhibitory Fc receptor with an immunoreceptor-based inhibitory motif (ITIM) and normally has a very low affinity for IgG monomers (IgG1<IgG2a=IgG2b<IgG3) (46). FcgRIIB is, for example, expressed on B cells regulating their activation (47), and on myeloid-derived cells modulating endocytosis through clathrin-coated pits (48). Liver sinusoidal endothelial cells (LSECs) are scavenger cells specialized to clear blood from smaller immune complexes through pinocytosis (49). In mice, three quarters of all FcgRIIB are expressed in the liver and 90% of liver FcgRIIB is found on the surface of LSECs (50). Interestingly, Ganesan et al. could demonstrate that small immune complexes are cleared via FcgRIIB-mediated uptake into the liver (50). Binding to FcgRIIB can be dramatically enhanced by mutating the Fc part of the sweeping antibody, leading to efficient clearing of soluble antigens (51).
Recently, it could be demonstrated that the sweeping antibody principle can also be applied for the elimination of antigen-specific antibodies. Devanaboyina et al. (52) used therapeutic antibodies including fragments of either the myelin oligodendrocyte glycoprotein (MOG), an antigen in multiple sclerosis, or HER2, a tumor protein (52). The therapeutic antibodies specifically bound to anti-MOG and anti-HER2 antibodies, which lead to rapid clearance of the resulting immune complex in mice. In a follow-up study, the authors could demonstrate a therapeutic effect using similar constructs in a mouse model of multiple sclerosis (53).
The increasing knowledge on pathogenic autoantibodies in MN and their target antigens offers excellent conditions to apply the sweeping antibody concept also in this disease. Sweeping antibodies with the Fab region substituted for an antigen fragment could be generated and tested for their capacity to bind and eliminate specific autoantibodies. As virtually 100% of patients with PLA2R1-associated MN have antibodies against the most N-terminal domain (the cysteine-rich domain) of PLA2R1 (54-56), a sweeping antibody containing this part of PLA2R1 may be a promising candidateyet therapeutic antibodies could be engineered for any MN antigen and tailored to the immune and epitope profile of each individual patient. After administration, the therapeutic sweeping antibodies bind their targets, the pathogenic autoantibodies, in the circulation. The resulting immune complex is then taken up by liver LSECs through enhanced affinity for FcgRIIB or by other endothelial cells via FcRn, resulting in lysosomal degradation of the pathogenic autoantibodies and potentially recirculation of the therapeutic sweeping antibody (Figure 2A). Fewer autoantibodies could bind their target antigen in the kidney, potentially leading to amelioration of disease activity. This can improve the course of MN without affecting parts of the immune system that are essential for the body's defense e.g. against infection. In summary, the elimination of pathogenic autoantibodies through endogenous degradation systems represents a promising therapeutic strategy for patients with MN.

STRIKING THE EVIL AT ITS ROOTS: CHIMERIC AUTOANTIBODY RECEPTOR T CELLS
Chimeric antigen receptor (CAR) T cells are a promising treatment for cancer, used with remarkable success for example in refractory or relapsed B cell lymphoma (57)(58)(59). To obtain such CAR T cells, peripheral blood mononuclear cells from the patient's blood are isolated and stimulated with interleukin 2 and anti-CD3 antibodies, leading to their proliferation. The T cells are transduced with a construct encoding for the CAR of interest, e.g. the antigen-binding domain of an anti-CD19 antibody fused to a transmembrane and several intracellular signaling domains such as 4-1BB and CD3z (60). Upon transfusion back to the patient, the CAR T cells eliminate cells expressing the target antigen, in this case CD19, which is a marker of B cells. A huge advantage of this treatment is the generation of long-term memory CAR T cells, which offer the opportunity for constant elimination of newly emerging target cells without the need for repetitive dosing.
This strategy can also be applied for the treatment of antibodymediated autoimmune diseases, with one essential modification: the antigen-binding domain of a conventional CAR is replaced by a part of the autoantigen of interest, resulting in a chimeric autoantibody receptor (CAAR). A T cell expressing this CAAR (CAAR T cell) will bind to and eliminate B cells that express the corresponding B cell receptor (BCR), a membrane-bound immunoglobulin matching the antibody produced by this particular cell clone. This approach was firstly tested in an animal model of pemphigus vulgaris (PV). T cells expressing a CAAR consisting of the PV autoantigen desmoglein 3 fused to CD137-CD3z specifically eliminated anti-desmoglein 3-specific autoreactive B cells (61). Moreover, the CAAR T cells erased their targets even in the presence of circulating anti-desmoglein 3 antibodies and did not show significant off target effects. In a follow-up study, the authors preclinically examined the pharmacodynamics and toxicity of this CAAR and showed that the desmoglein 3-CAAR T cells specifically deplete primary human desmoglein 3-specific B cells from PV patients and is effective in an active animal model of PV (62).
Applying the CAAR T cell strategy for the treatment of MN would be a pioneering approach. The antibody binding sites in the antigens PLA2R1 and THSD7A have been already characterized, in parts even down to the level of single antigen domains and smaller epitopes (54)(55)(56)(63)(64)(65). Given this knowledge, to achieve an optimal intermembrane distance of the immunologic synapse (the space between the CAAR T cell and the target B cell) and minimize potential off-target effects, it appears reasonable to fuse smaller fragments of PLA2R1 or the other MN antigens to the chimeric receptor ( Figure 2B). In case of PLA2R1 and THSD7A the most N-terminal regions are considered as immunological hot spots, as most reactivity with patient autoantibodies is found in these areas (56,63,64). Therefore, a CAAR containing only the N-terminal region of the respective autoantigen might be sufficient to eliminate a number of autoreactive B cells large enough to ameliorate disease. It would also be possible to perform an antibody mapping first, e.g. using domain-specific ELISAs, and tailor the CAAR strategy to the individual epitope profile.
Although the CAR T cell therapy has revolutionized the treatment of malignancies, it has several potential flaws. First, the manufacturing of personalized CAR T cells is very complex, time-consuming and expensive. Second, CAR T cell therapy could involve serious adverse effects, such as the cytokine release syndrome (CRS), caused by the activation of CAR T cells and their production of proinflammatory cytokines (66,67), or neurotoxicity, which often accompanies CRS, possibly due to cerebral endothelial dysfunction (68). Third, it is questionable whether the generation of memory T cells is always desirable in light of many patients achieving full remission of disease after some time.
The generation of CAR NK cells represents an alternative strategy to overcome some of these undesired effects. Due to their limited lifespan, CAR NK cells show a relatively low toxicity towards normal tissue of the recipient (69). They offer the possibility of "off the shelf" therapy as they can be obtained from peripheral blood mononuclear cells of healthy donors, umbilical cord blood, induced pluripotent stem cells, and even NK92 cell lines (70). Currently, most of the CAR NK constructs derive from CAR T cell approaches, but there are attempts to design also CARs tailored specifically for NK cells (71). To date, CAR NK cell therapy seems to be a safe additional approach for cancer therapy, with several clinical trials running (72,73). In conclusion, CAAR T cell therapy and, alternatively, the safer and likely more cost-efficient and easier to manufacture CAAR NK cell therapy represent an elegant and promising therapeutic strategy for patients with MN.
Notably, is also possible to approach T cell therapy from the opposite site, by regaining immune tolerance. To achieve this, the function of one key player of immune tolerance has to be  modified: the so called regulatory T cells or Tregs. About 5-7% of all CD4+ T cells in the human body are Tregs. The identity and function of Tregs are characterized by the expression of various markers such as the cytotoxic T lymphocyte-associated protein-4 (CTLA-4), the interleukin 2 receptor subunit a (CD25), the transcription-maintaining factor STAT5, and the forkhead box P3 (FOXP3). FOXP3 has a key function for the maintenance of immune tolerance (74). Tregs can be transduced with so called BARs (chimeric B celltargeting antibody receptors) containing an extracellular domain consisting of the immunodominant parts of the antigen, which means they are basically comparable to CAARs on cytotoxic T cells. Instead of killing the B-cell upon binding, such BAR Tregs suppress antigen-specific B cells directly without affecting the T cell response (75).

CONCLUSION
MN is an antibody-mediated autoimmune disease and several autoantigens have been identified over the past years. The direct pathogenicity of the involved autoantibodies has either been shown or is strongly suspected (20,22), and treatment strategies have shifted towards targeting of B cells (30,31,76). In light of these developments, the establishment of antigen-specific treatments represents the consequential continuation of MN therapy. We are aware that the therapeutic concepts discussed in this article have not been tested in MN, not even at an experimental level. However, future research should take into account such potential strategies, particularly as antigen-specific animal models become more and more available (22,77). Clearly, the possible adverse effects of such treatments should not be overlooked, but there is a realistic chance that they would not be as severe as the ones of currently used therapeutics. Especially in case of CAR T cells, huge efforts are made to reduce side effects and make them more controllable, e.g. through transient transduction, suicide genes or elimination markers (78).
The actual charm of the two treatment approaches that we presented here is the option to combine and apply them as a therapeutic package tailored for the individual patient. As an example, one could imagine the following procedure: After confirming the diagnosis of PLA2R1-associated MN by kidney biopsy, the antibody profile is analyzed using a domain-specific ELISA, revealing reactivity with the cysteine-rich domain. Subsequently, sweeping antibodies containing the cysteine-rich domain are applied to eliminate the circulating autoantibodies. This acutely reduces binding of pathogenic antibodies at the glomerular filtration barrier and additionally paves the way for CAAR T or NK cells carrying a CAAR which contains the cysteine-rich domain. The CAAR T or NK cells thus eliminate autoreactive B cells, which prevents further production of pathogenic antibodies. Without previous clearance of antibodies, CAAR cells may in fact be neutralized by the circulating antibodies binding to the CAAR, potentially making them ineffective in eliminating autoreactive B cells. As an alternative or additive strategy, PLA2R1-specific immune tolerance could be restored by suppression of autoreactive B cells using Tregs expressing a BAR that contains the cysteinerich domain.
In summary, the increasing knowledge about the targets of autoimmunity in MN offers a huge potential for the application of antigen-specific treatment strategies, which would ideally spare protective immunity.

AUTHOR CONTRIBUTIONS
SK and LS contributed equally to this review. NT conceptualized the work. All authors discussed the concept and revised the manuscript. All authors approved the final version of the manuscript.

FUNDING
This work was funded by the Deutsche Forschungsgemeinschaft as part of the Emmy Noether Programme "Molecular Mechanisms of Membranous Nephropathy" (TO10-13) to NT and as part of the Sonderforschungsbereich 1192 to NT (project B2). This work was additionally supported by the Deutsche Forschungsgemeinschaft (DFG) as part of the grant ZA163/12-1 to GZ.