Manuscript Title: Podocyte-specific knockout of the neonatal Fc receptor (FcRn) results in differential protection depending on the model of immune-mediated kidney disease. Authors and Affiliations:

Podocytes have been proposed to be antigen presenting cells (APCs). In traditional APCs, the neonatal Fc receptor (FcRn) is required for antigen presentation and global knockout of FcRn protects against immune-mediated kidney disease. Since podocytes express FcRn, we sought to determine whether the absence of podocyte FcRn ameliorates immune-mediated disease. We examined MHCII and costimulatory markers expression in cultured wild type (WT) and FcRn knockout (KO) podocytes. Interferon gamma (IFN) induced MHCII expression in both WT and KO podocytes but did not change CD80 expression. Neither WT nor KO expressed CD86 or inducible costimulatory ligand (ICOSL) at baseline or with IFN Using an antigen presentation assay, WT podocytes but not KO treated with immune complexes induced a modest increase in IL-2. Induction of the anti-glomerular basement membrane (anti-GBM) model resulted in a significant decrease in glomerular crescents in podocyte-specific FcRn knockout mouse (podFcRn KO) versus controls but the overall percentage of crescents was low. To examine the effects of the podocyte-specific FcRn knockout in a model with a longer autologous phase, we used the nephrotoxic serum nephritis (NTS) model. We found that the podFcRn KO mice had significantly reduced crescent formation and glomerulosclerosis compared to control mice. This study demonstrates that lack of podocyte FcRn is protective in immune mediated kidney disease that is dependent on an autologous phase. This study also highlights the difference between the anti-GBM model and NTS model of disease.


Introduction
Despite immune-mediated nephritis being the third leading cause of end stage kidney disease in the United States(1), treatment options are limited and typically involve systemic immunosuppressive medications which are variably effective and have multiple side effects.
Treatment of these diseases is limited, in part, by the incomplete understanding of how the immune system recognizes and targets the kidneys.
Prior studies have demonstrated that immune-mediated kidney diseases are dependent on CD4 + T cell and the adaptive immune system. (2)(3)(4) Stimulation of the CD4 + T cells have been shown to be carried out by intrinsic renal cells that express major histocompatibility complex class II (MHCII).(5) Goldwich et al published data showing that podocytes express MHCII and suggesting that podocytes act as non-hematopoietic professional antigen presenting cells (APCs) that can stimulate CD4 + T cells. (6) Podocytes are able to express CD80, which is a well-known costimulatory marker needed for activation of T cells, and CD80 expression has been associated with certain glomerular diseases. (7,8) Given these findings, podocytes have been proposed as candidates for the intrinsic renal cells that trigger an immune response and lead to immunemediated kidney disease. However, a better understanding of the mechanism by which podocytes can cause an immune response is needed to provide targeted therapies for these devastating disorders.
A promising area of study is the neonatal Fc receptor (FcRn). FcRn is a major histocompatibility class I-like protein that was initially described and discovered as a means for infant enterocytes to obtain passive immunity through breast milk. (9,10) Since its initial description, FcRn has been found to play important roles in albumin and IgG recycling and has been shown to be expressed in several other cell types. (9,(11)(12)(13)(14)(15)(16)(17)(18) In the kidneys, FcRn is expressed in endothelial cells, podocytes and proximal tubular cells (13-15, 17, 19).
FcRn also plays an important role in adaptive immunity. (12,(20)(21)(22)(23)(24) Studies in dendritic cells have established that FcRn is necessary for trafficking antigen-antibody complexes for degradation as well as presentation on MHCII. (10,20,25,26) Furthermore, studies have shown that when FcRn is absent, dendritic cells cannot present antigen for antigen presentation. (25) Interestingly, when FcRn is globally knocked out of mice, immune-mediated kidney disease is attenuated. (27) However, it remains unclear whether lack of FcRn in dendritic cells, endothelial cells, or podocytes provides this protective effect. Knowledge of precisely in which cells FcRn is required for induction of immune-mediated kidney disease would allow for the development of targeted therapies.
In this study, we investigated whether podocytes can function as APCs and whether knockout of FcRn specifically in podocytes attenuates the progression of anti-glomerular basement membrane (anti-GBM) nephritis and nephrotoxic serum nephritis (NTS). We chose these models as both are well-characterized models of immune mediated kidney disease. In addition, using the anti-GBM model, Goldwich et al. had previously found that podocyte-specific knockout of MHCII attenuated renal disease but whether this can be observed in other disease models and the exact mechanism is unknown.(28) Since in dendritic cells, FcRn is required for antigen presentation via MHCII, we hypothesized podocyte-specific knockout of FcRn would ameliorate antibody-mediated immune kidney disease.
The anti-GBM model is characterized by a heterologous phase which lasts 4 -5 days post anti-GBM antibody injection and is characterized by neutrophil invasion and complement deposition. (29,30) The initial heterologous phase is followed by an autologous phase in which T helper cells enter the kidney in response to autologous antibodies produced to the anti-GBM antibody. In the anti-GBM model, the autologous antibody phase only lasts 4-6 days as the anti-GBM model is a severe model of disease with mice not surviving beyond 10 days. (31,32) We examined the effects of podocyte-specific FcRn KO at both day 3 and day 8 after induction of anti-GBM nephritis to determine whether lack of FcRn had an effect on either phase of the disease.
The NTS model has relatively mild heterologous phase and is more dependent on autologous antibody production. (33,34) Since this model is more dependent on the humoral immune system and takes longer to develop, we examined the effects of the podocyte-specific FcRn KO at 28 days after induction to ensure adequate time for humoral response.

Methods
Cell culture: Generation of conditionally immortalized WT and FcRn KO podocytes: Podocytes were isolated from WT or global FcRn KO mice and then underwent immortalization using a thermosensitive polyomavirus simian virus 40 (SV40) T antigen as previously described. (35)(36)(37) Primary podocytes were allowed to replicate at 33 o C. To induce differentiation, podocytes were placed at 37 o C for 8 -10 days. To verify expression of podocyte markers, podocytes were stained with podocin or WT1 and synaptopodin expression was checked by Western blot. Conditionally immortalized WT or KO podocytes were allowed to differentiate at 37 °C for 9 -10 days prior to being used in experiments. to sacrifice urine and blood were collected. Urine albumin was measured using the Albuwell assay (Exocell), urine creatinine was measured using the assay and BUN was measured on an Alpha Wasserman auto analyzer. Titers of rabbit IgG in control and podFcRn KO mice were measured as previously described (39). Briefly, a 96 well plate was coated with 10 ug/ml rabbit IgG. Coated wells were incubated with serum from control or podFcRn KO mice diluted 1:10,000. Mouse antirabbit total IgG titers were measured by ELISA (Biolegend).

Nephrotoxic serum nephritis model:
The nephrotoxic serum nephritis model is a newer but increasingly utilized model of glomerulonephritis in mice. (33,(40)(41)(42)(43)(44)(45) Briefly, 8-12 week old podFcRn KO or control mice were given an intraperitoneal injection with 100μL of sheep anti-Rat glomerular basement membrane antibody made by Probetex Inc. (San Antonio, TX) at day 0. Mice were sacrificed on day 28 after injection. Prior to sacrifice, urine and blood were collected. Urine albumin, urine creatinine, and serum BUN were measured using the same methods as described above under the anti-GBM nephritis model.

Histology:
Three micrometer (m) sections were cut from paraffin embedded tissue and stained using the periodic acid Schiff reagent. Crescent formation, defined as the presence of two or more layers of cells in Bowman's space, was evaluated in 20 -30 random glomeruli for each mouse.
Glomerulosclerosis was evaluated using a semi-quantitative scoring system. The glomerular score was determined from a mean of at least 20 -30 glomeruli sampled at random. The severity of glomerulosclerosis was graded on a scale (0 to 4) as follows: grade 0, normal glomerulus; grade 1,mild hyalinosis/sclerosis involving < 25% of the glomerulus; grade 2, moderate segmental hyalinosis/sclerosis involving < 50% of the glomerulus; grade 3, diffuse glomerular hyalinosis/sclerosis involving more than 50% of the glomerulus; grade 4, diffuse glomerulosclerosis with total glomerular obliteration. Histologic analysis was performed using Aperioscope.
Immunofluorescence: Confocal microscopy images were acquired using Zeiss 780 laserscanning confocal/multiphoton-excitation fluorescence microscope with a 34-Channel GaAsP QUASAR Detection Unit and non-descanned detectors for two-photon fluorescence (Zeiss, Thornwood, NY). The imaging settings were initially defined empirically to maximize the signalto-noise ratio and to avoid saturation. In comparative imaging, the settings were kept constant between samples. Images were captured with a Zeiss C-Apochromat 40x/1.2NA Korr FCS M27 water-immersion lens objective. The illumination for imaging was provided by a 30mW Argon Laser using excitation at 488 nm, Helium Neon (HeNe) 5mW (633 nm) and 1mW (543 nm). Image processing was performed using Zeiss ZEN 2012 software. Images were analyzed in Image J software (NIH, Bethesda, Maryland). Data analysis: Data are presented as means ± SE. Statistical analysis was performed using t-tests for two groups and one-way analysis of variance for 3 or more groups, using Prism software (GraphPad, San Diego, CA). Tukey's post hoc test was applied to the ANOVA data. Values were considered statistically significant when p < 0.05.

MHCII and costimulatory marker expression in WT and FcRn KO podocytes.
Podocytes were isolated from wild type (WT) and global FcRn KO (KO) mice and immortalized using the thermosensitive polyomavirus simian virus (SV40) T antigen as described previously (36,37). Since podocytes are not known to express MHCII at baseline, we used interferon gamma In addition to T cells binding to the APC's MHCII and costimulatory markers, CD4+ T cells need autocrine cytokine production (such as interleukin 2 (IL2)) to act as a third signal for T cell activation. To determine whether podocytes can act as APCs, an in vitro antigen presentation assay was performed using WT and KO podocytes. CD4 + T cell activation was measured by determining the amount of IL2 produced by the T cells when WT or KO podocytes were used as APCs. WT podocytes treated with media alone and immunoglobulin G (IgG) alone induced minimal IL2 production when co-cultured with CD4 + T cells. When WT podocytes were treated with IgG + ovalbumin (immune complexes, ICs) there was a significant increase in IL2 production by CD4 + T cells (6.8 +/-0.9 pg/ml versus 3.3 +/-0.7 pg/ml for WT + IC vs. WT + media, p < 0.05, Figure   2). T cell production of IL2, however, was significantly less when WT podocytes were used as APCs compared to splenocytes (Supplemental Figure 1), suggesting that podocytes are inefficient APCs. KO podocytes treated with media or IgG alone also induced minimal IL2 production by CD4 + T cells. In contrast to WT podocytes, when FcRn KO podocytes were used as APCs, there was no significant increase in IL2 production by CD4 + T cells (Figure 2), suggesting that KO podocytes are unable to present antigen.

Podocyte-specific KO of FcRn did not alter neutrophil invasion or complement component 3 (C3) deposition after induction of anti-GBM nephritis
In order to determine whether podocyte-specific knockout of FcRn would protect against induction of immune-mediated kidney disease, we generated podocyte-specific FcRn KO (podFcRn KO) mice by crossing podocin-Cre mice with FcRn floxed mice to create FcRn fl/fl:Podocin-Cre/+ mice ( Figure 3) as previously described (37). FcRn floxed mice lacking the Cre transgene (FcRn fl/fl;+/+) served as littermate controls. Podocyte-specific knockout of FcRn did not alter renal histology at 3 months of age (for the anti-GBM model mice were used between 8 and 12 weeks of age; Figure 3A). Since FcRn is known to play an important role in IgG recycling, rabbit IgG titers were checked in podFcRn KO mice and controls 3 days after injection of rabbit anti-mouse anti-GBM antibodies to ensure that antibody levels were similar. There was no difference in anti-rabbit IgG titers in controls versus podFcRn KO, indicating that knockout of FcRn in podocytes does not significantly alter IgG metabolism ( Figure 3B).  Figure 4H). Of note, although the anti-GBM model resulted in significant albuminuria in both control and podFcRn KO mice, the percentage of crescents seen in the control mice was relatively low. These findings are in accord with some other studies of this model in mice, (49)(50)(51)(52) making this model less ideal to study immune mediated kidney disease.

Discussion
Immune-mediated kidney diseases have long provided therapeutic challenges and efforts are ongoing to understand the mechanisms involved in these diseases. Within the kidney, podocytes have been proposed to act as APCs since they express some of the molecular machinery required for antigen presentation (MHCII, CD80, FcRn) and podocyte-specific KO of MHC II has been shown to attenuate anti-GBM disease.(6) Here we show that in vitro, WT podocytes have the ability to express MHCII and CD80, which are necessary for T cell stimulation, but that WT podocytes have limited ability to activate T cells in vitro. Though inefficient, antigen presentation by podocytes is dependent on FcRn since the KO podocytes were not able to stimulate T cells.
Our in vivo data demonstrated that the absence of FcRn in podocytes did not protect against anti-GBM nephritis in the early heterologous stage of the disease. In the autologous phase, podocytespecific knockout of FcRn resulted in a significant decrease in crescent formation without a significant change in albuminuria, BUN or glomerulosclersosis score. One possibility is that podocyte FcRn is not required for induction of anti-GBM nephritis but may play a role in the resolution phase of the disease. Alternatively, the damage caused by the heterologous phase of the anti-GBM nephritis model may have been too extreme and irreversibly impacted the albuminuria, BUN, and glomerulosclerosis scores despite a protective effect during the autologous phase provided by knocking out FcRn.
By utilizing the NTS model, we were able to demonstrate that podocyte FcRn plays a significant role in glomerulonephritis diseases that are more autologous in nature. Since there was significantly less albuminuria, glomerulosclerosis, and crescents in the podocyte-specific FcRn knockout mice using the NTS model suggests that the albuminuria and glomerulosclerosis observed in our anti-GBM nephritis model was driven mostly by the heterologous phase. Our nephrotoxic serum nephritis findings also suggest that podocyte FcRn is mainly involved in autologous phase and not during the induction phase.
These findings are significant because it demonstrates a significant difference between the anti-GBM model and the NTS model when investigating the role podocytes play in glomerulonephritis.
It further supports the notion that therapies need to be tailored not only to the specific disease but also the particular mechanism underlying each disease process. We found that the anti-GBM model induced a rapid illness in the mice that was often fatal (up to 50% of animals) in a relatively short period of time. The severity of the anti-GBM model may result from the injection with normal rabbit IgG in Freund's complete adjuvant prior to the anti-GBM antibody. This presensitization step may lead to a more severe disease not only due to the immune system being primed to react to the rabbit IgG but also due to the non-specific inflammatory effect from Freund's complete adjuvant.
The NTS model does have a heterologous phase, as reported by the manufacturer, which behaves similarly to the anti-GBM in that it results in IgG and C3 deposition in the glomerulus and proteinuria within a few days. However, contrary to the anti-GBM model, the NTS model had a much lower mortality rate, which could be due to the lack of the pre-sensitization step, thus allowing us to evaluate the disease in the autologous phase. Given the more indolent nature of most immune-mediated kidney diseases in humans, NTS seems like the more clinically similar model to study for human application.
Additionally, it has been suggested that podocytes are involved in making glomerular crescents. (53)(54)(55)(56)(57) Our observed decrease in glomerular crescents in the podocyte-specific FcRn knockout mice in both models is provocative as it would suggest FcRn is involved in crescent formation. To our knowledge, no prior study has investigated the role of FcRn in crescent formation. The exact mechanisms by which the absence of FcRn prevents crescent formation may have to do with impaired podocyte mobility, alterations in actin dynamics or impaired intracellular signaling. The mechanism underlying how FcRn may affect these pathways is unknown but we have found that knockout of FcRn KO in cultured podocytes alters mobility and actin dynamics (37). In addition, when comparing the relative quantity of the crescents between the two models, we found significantly fewer glomeruli in the anti-GBM had crescents compared to the NTS model, which is suggestive that crescent formation is more dependent on the autologous phase than the heterologous phase of immune mediated kidney disease. This information may prove valuable when trying to determine the exact mechanism leading to crescent formation.
A potential area for future study could be aimed at investigating if there is a difference in other models of disease. An example of this might be achieved by breeding our podFcRn KO mouse onto a spontaneous lupus nephritis model which could yield interesting findings. Another area of study would be to create an inducible knockout of podocyte-specific FcRn. This would allow investigation into whether knocking out or blocking FcRn in podocytes after disease onset would modulate the disease course and thus suggest a therapeutic option to treat glomerulonephritis.
In summary, we have shown that induction of anti-GBM nephritis is not dependent on podocyte FcRn but that podocyte-specific knockout of FcRn leads to a significant reduction in crescent formation. Furthermore, we have shown that podocyte-specific knockout of FcRn can significantly reduce albuminuria, glomerulosclerosis, and crescent formation in the autologous phase of NTS, suggesting podocyte FcRn plays a significant role in disease progression.

Disclosure
The authors declare no competing interests.