The proteasome modulates endocytosis specifically in glomerular cells to promote kidney filtration

Abstract

Kidney filtration is ensured by the interaction of podocytes, endothelial and mesangial cells. Immunoglobulin accumulation at the filtration barrier is pathognomonic for glomerular injury. The mechanisms that regulate filter permeability are unknown. Here, we identify a pivotal role for the proteasome in a specific cell type. Combining genetic and inhibitor-based human, pig, mouse, and Drosophila models we demonstrate that the proteasome maintains filtration barrier integrity, with podocytes requiring the constitutive and glomerular endothelial cells the immunoproteasomal activity. Endothelial immunoproteasome deficiency as well as proteasome inhibition disrupt the filtration barrier in mice, resulting in pathologic immunoglobulin deposition. Mechanistically, we observe reduced endocytic activity, which leads to altered membrane recycling and endocytic receptor turnover. This work expands the concept of the (immuno)proteasome as a control protease orchestrating protein degradation and antigen presentation and endocytosis, providing new therapeutic targets to treat disease-associated glomerular protein accumulations.

Fig. 1. Glomerular cell-specific distribution of proteasomal and lysosomal proteins in human and murine glomeruli.

A Schematic depiction of the ubiquitin proteasome system (UPS) and autophagosome-lysosome pathway (ALP) with individual marker proteins used for the expression analyses within glomerular cell types. Ubiquitin = polypeptide involved in dynamic regulation of protein function, localization, and stability; Rpt5 = proteasome regulatory subunit 6A (Psmc3) of the proteasome 19S regulatory particle; β5c (Psmb5) = main proteolytic subunit of the constitutive 20S core particle; LC3 = microtubule-associated protein 1A/1B-light chain 3; Lamp1 and Lamp2 = lysosomal-associated membrane proteins 1 or 2; Limp2 (Scarb2) = lysosomal integral membrane protein 2. Distribution of marker proteins (green) by high-resolution confocal images in a healthy human (B) and murine (C) glomerulus in relation to the slit diaphragm protein nephrin (red) and DNA (blue); pc podocyte, mc mesangial cell, ec glomerular endothelial cell, pec parietal epithelial cell, white arrows point toward endothelial lining filled with Lamp2-positive lysosomes, n = 3 individuals. D–F Podocytes (PC), mesangial cells (MC) and glomerular endothelial cells (EC) were bulk-isolated from glomeruli. D Proteomic analyses depict molecular properties of glomerular cell types as shown by the radar plot, whereby two-fold changes of distinct uniprot key words are plotted. Protein values were obtained by label-free quantification results using the MaxQuantLFQ algorithm. E Relative transcript levels quantified via qRT-PCR of Psmb5 (encoding for β5c) and Scarb2 (encoding for Limp2) normalized to 18S as home keeper in relation to total glomerular transcript levels (dashed line), mean ± SEM, *p = 0.0292 (PC Psmb5), *p = 0.048 (MC Scarb2), one-way ANOVA with Bonferroni post-test for multiple comparisons, n = 12 of 2 pooled independent experiments. F Protein abundance from isolated glomerular cell types determined by immunoblot, equal loading was ensured by loading equal numbers of FACS-sorted PCs, MCs, and ECs, n = 3 independent experiments. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 2. Glomerular cell-specific proteasomal activities.

Glomerular cell types were analyzed for the main proteasomal β5-subunit abundance and activity. Micrographs were analyzed from 3 individual experiments, with 3 micrographs per group. A Proteomic label-free quantification of proteasome subunit abundance additionally normalized to Psma1 of the structural proteasome 20S core particle; podocytes (PC), mesangial cells (MC) and glomerular endothelial cells (EC). B Scheme depicting localization of the proteolytic β-subunits within the constitutive proteasome (c20S) and the immunoproteasome (i20S). Abundance of the c20S subunit β5c, of the i20S subunit β5i, and of the (c + i)20S subunit α2 from cell number adapted cells by immunoblotting. C Expression of Psmb8 (encoding for the β5i) quantified via qRT-PCR, mean ± SEM, **p = 0.0096, one-way ANOVA with Bonferroni post-test for multiple comparisons, n = 8 of 2 independent experiments. D Distribution of the β5i-subunit (green) in healthy human and murine glomerular cells analyzed by high-resolution confocal images in relation to the slit diaphragm protein nephrin (red) and DNA (blue), white arrows point toward β5i expressing GEnCs (ec), pc podocyte, n = 3 individual experiments. E In-gel activity assay from cell number adapted cells determining the proteolytic c20S and i20S β-subunit activity using the pan-reactive activity-based probe MVB003 and the β5i-subunit-specific activity-based probe GB514. Immunoblots to α2 and/or β-actin from the same activity gels are shown. Graph: quantification of β5c/β5i activity via MVB003, mean ± SEM, **p = 0.009, ***p = 0.0004, one-way ANOVA with Dunn’s post-test for multiple comparisons (n = 6 PC, MC, n = 8 EC). F Proteasomal activity assay in isolated pig glomeruli by ex vivo incubation with MVB003 for 1 h with or without the proteasome inhibitor epoxomicin. Glomerular cell types were demarcated using CD31 (red) for GEnCs and nephrin (turquoise) for podocytes, DNA (white); pc podocyte, mc mesangial cell, ec glomerular endothelial cell. Note strongest activity in GEnCs followed by podocytes (enlarged panel 1). MCs exhibit lowest proteasomal activity (enlarged panel 2). G In-gel activity measurement of the in (F) visualized pig glomerular samples to control for specificity of the (by confocal microscopy) depicted proteasomal activity. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 3. Endothelial cell-specific β5i-deficiency results in morphological alterations of the glomerular filtration barrier.

Statistical analyses of all graphs: Two-sided Mann–Whitney U, mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n ≥ 3 per group, pooled values from 1 (EM) to 3 independent experiments (rest). A A tamoxifen-inducible endothelial cell-specific Lmp7 (β5i) knockout mouse was generated with the cre-lox system. Naïve mice were analyzed 5–50 weeks after induction of knockout. B Knockout was evaluated by high-resolution confocal microscopy to β5i (green) in relation to nephrin (red) and DNA (blue) in glomeruli from Lmp7^ΔEnC and control littermates, arrows point toward β5i-expressing endothelial cells within (glomerular endothelial cell, white arrow) and outside (peritubular endothelial cell, red arrow) of the glomerulus. C Immunoblot of β5c and β5i abundance in bulk-isolated glomerular endothelial cells (GEnCs) in comparison to non-endothelial kidney cells (non-EnC) of the preparations from Lmp7^ΔEnC (ΔEnC) compared to control (Ctrl) kidneys; lower graph exhibits relative densitometric analysis of β5c levels in Lmp7-deficiency compared to control, n = 6 per group, **p = 0.0022. D Flow cytometry plot depicting surface levels of H-2Kb MHC class I alloantigen in GEnCs in relation to controls. Lower graph exhibits quantification. E Cell size determination of GEnCs isolated by FACS-sort using the mean FSC-A, *p = 0.0485. F Cell cycle analysis with propidium iodide staining via flow cytometry in bulk-isolated GEnCs, pie charts: % of total cells in different cell cycle stages. G EM analyses exhibit loss of endothelial fenestrations (empty arrow heads, quantification lower left graph) in GEnCs, podocyte foot process effacement (red arrows, quantification lower middle graph), and focal splitting of the glomerular basement membrane with irregularity in thickness (filled arrow heads, quantification lower right graph) and electron dense depositions in Lmp7^ΔEnC mice; c capillary lumen, u urinary space, gbm glomerular basement membrane, overview in Supplementary Fig. 8, *p = 0.0222. H Immunoblot of lysine (K)48-polyubiquitinated proteasome substrates (K48pUb) in bulk-isolated GEnCs from Lmp7^ΔEnC (ΔEnC) compared to control (Ctrl) kidneys. I Albuminuria measured by ELISA to albumin and normalized to corresponding creatinine. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 4. Proteasome inhibition results in morphological and functional alterations of the glomerular filtration barrier.

Mice were treated with the irreversible proteasome inhibitor epoxomicin (0.5 µg/g bodyweight), the lysosomal inhibitor leupeptin A (40 µg/g bodyweight) or equal volumes of DMSO (vehicle, 125 µl) on four consecutive days. Urine, serum, and kidneys were collected and analyzed. Micrographs were analyzed from 3 individual experiments, with 3 micrographs per group. A Blood-urea nitrogen (BUN) measurement to assess renal function, n = 11 per group. B Albuminuria measured by ELISA to urinary albumin and normalized to corresponding creatinine to assess for glomerular filtration barrier functionality, n = 8 (veh, leup), n = 10 (epox). C Glomerular tuft area and absolute or relative podocyte number were quantified by immunohistochemical staining for the podocyte marker p57 in kidney paraffin sections (n ≥ 16 per group). All shown statistical analyses: one-way ANOVA with Dunn’s multiple comparison test, mean ± SEM, *p < 0.05, ***p < 0.001, n ≥ 11 per group, pooled data from 3 independent experiments. D Glomerular morphology assessed by PAS staining, *magnetic bead originating from glomerular isolation procedure, pc podocyte, ec glomerular endothelial cell. E High-resolution confocal micrograph of immunofluorescent staining for the slit diaphragm protein nephrin (white) and DNA (blue, Hoechst) to assess for podocyte foot process effacement. F–I Electron microscopical ultrastructural analyses exhibit focal podocyte foot process effacement and loss of endothelial fenestrations (empty arrow heads) as well as focal splitting of the glomerular basement membrane with accumulation of electron dense material (filled arrow heads) in epoxomicin-treated mice. c capillary lumen, u urinary space. F, G In leupeptin A treated mice (H, I) mesangial cells (mc) show electron dense lysosomal storage (asterisks). Source data are provided as a Source Data file.

Fig. 5. Proteasome impairment results in glomerular IgG accumulation.

A Scheme depicting the experimental setup. Mice were treated with the proteasome inhibitor epoxomicin (epox, 0.5 µg/g bodyweight), the lysosomal inhibitor leupeptin A (leup, 40 µg/g bodyweight) or equal volumes of DMSO (veh, vehicle, 125 µl) on four consecutive days following an initial injection with 160 µl of unspecific rabbit IgG (rbIgG) or PBS. Mice were perfused and kidneys were removed on day 10. Micrographs were analyzed from 3 individual experiments, with 3 micrographs per group. B Immunoblot quantification of deposited rbIgG following rbIgG-pulldown from isolated glomeruli. Graph exhibits densitometric analysis, mean ± SEM, n = 6 (veh), n = 7 (leup), n = 8 (epox), pooled data from 2 independent experiments, *p = 0.013, one-way ANOVA with Bonferroni’s post-test for multiple comparisons. C Confocal micrographs depicting rbIgG (white) deposition pattern in glomeruli on day 10, note the linear deposition at the GFB (white arrows) of the epoxomicin-treated mouse. D High-resolution confocal micrographs resolving rabbit IgG (green) accumulation pattern in epoxomicin-treated mice at the GFB. White arrows point toward linear meandering rbIgG accumulations in the subepithelial space. The glomerular basement membrane is marked by collagen type 4 (red) and DNA is depicted in blue (Hoechst). E Lmp7^ΔEnC and control littermates were analyzed after induction of β5i knockout. Following kidney perfusion with PBS, glomerular (GEnCs) and non-endothelial kidney cells were FACS-sorted. Immunoblot quantification of msIgG, densitometric analysis relative to control littermate cell populations, mean ± SEM, n = 6 mice of 2 pooled independent experiments, *p = 0.0152, two-sided Mann–Whitney U. Equal loading was ensured by loading equal numbers of FACS-sorted endothelial cell populations between genotypes. F Experimental kidneys of Lmp7^ΔEnC and controls were additionally perfused with AF647-Lycopersicon esculentum lectin (here: red) to demarcate the endothelial glycocalyx. Following zinc-fixation, sections were stained for msIgG (green) and the podocyte (pc) foot process protein THSD7A to enable localization of msIgG, DNA (white, Hoechst). Red arrows point toward msIgG accumulation underneath the glycocalyx within GEnCs (ec) in the Lmp7^ΔEnC mouse, mc mesangial cell. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 6. Proteasome functionality affects podocyte endocytosis.

Micrographs were analyzed from 3 individual experiments, with 3 micrographs per group. A Primary podocytes were outgrown for 5 days and assessed for their endocytic activity. Cells were pretreated with either 10 nM epoxomicin or DMSO for 24 h, synchronized on ice prior to the addition of endocytic substrates. Cells were harvested after time point (t) 0, 30, and 60 min, fixed with 4% PFA and stained for synaptopodin (white) to demarcate podocytes. Internalization of FITC-transferrin (green) and Cy3-rabbit IgG (rbIgG, red) was assessed by immunofluorescence. Green and red arrows point toward endocytic cargo that was taken up; the dotted line indicates the cell border. B FITC-albumin endocytosis was assessed in isolated garland nephrocytes from Drosophila after pretreatment with 2 µM epoxomicin (epox) or equivalent amounts of DMSO (veh). High-resolution confocal images of FITC-albumin in nephrocytes, graph exhibits quantification of relative mean fluorescent intensity per nephrocyte area to vehicle, mean ± SEM, pooled data of 3 independent experiments with n = 25 per group, *p = 0.0409, two-sided Mann–Whitney U. C Human podocytes were pretreated with 1 µM epoxomicin or equivalent amounts of DMSO (vehicle) for 6 h. Wheat germ agglutinin (WGA)-rhodamine was applied at timepoint 0 min to the medium to mark glycoproteins of the plasma membrane. Time-lapse live-cell images were taken at 37°C and 5% CO₂ using a Nikon Spinning Disc microscope, a frame every 5 min is shown (live-imaging films are within the supplement). Note the reduced appearance of WGA-positive vesicles in epoxomicin pretreatment. The large vesicle depicted (white arrow) exhibits a reduced motility, ec extracellular space, ic intracellular space. D Following live-cell imaging, human podocytes were fixed and processed for high-resolution confocal microscopy for localization of the in vivo applied rhodamine-WGA (red) in conjunction with filamentous (F)-actin (green) and lysine (K)-48-polyubiquitinated proteins (K48-pUb, turquoise). Note the strong WGA signal as well as the accentuated signal for K48-pUb at the plasma membrane (white arrows) and at the thin processes (yellow arrows in [1]) of the epoxomicin-exposed human podocyte. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 7. Plasma membrane abundance of endocytic receptors is modulated by proteasome functionality.

Micrographs were analyzed from 3 individual experiments, with 3 micrographs per group. High-resolution confocal micrographs of glomeruli from (A) Lmp7^ΔEnC and control kidneys after induction of β5i-knockout or from (B) BALB/c mice treated with epoxomicin or vehicle for 4 consecutive days demonstrates abundant FcRN (green) expression in GEnCs (ec, endomucin (red)) in a vesicular/granular pattern (red arrows) and in podocytes (pc, synaptopodin (turquoise)) in a perinuclear vesicular/granular pattern (white arrows), DNA (white, Hoechst). C Primary podocytes were assessed for FcRN trafficking, experimental procedure depicted in the scheme (corresponding panels for rbIgG and transferrin Fig. 6A). Cells were harvested after time point (t) 0, 30, and 60 min, fixed with 4% PFA and stained for FcRN. White arrows: FcRN trafficking to or from the plasma membrane, the dotted line indicates the cell border. D–F Glomeruli and glomerular cell types of epoxomicin- compared to vehicle-treated mice were evaluated for protein abundance of Mrc2 and nephrin. D Immunoblot quantification of glomerular Mrc2 and nephrin protein levels to vehicle, mean ± SEM, n = 10 (Mrc2), n = 15 (nephrin veh) n = 14 (nephrin epox) mice, data from 5 independent experiments, **p = 0.0014, two-sided Mann–Whitney U test. E High-resolution confocal micrographs of glomeruli derived from proteasome-inhibited mice depicting the localization of Mrc2 (green), nephrin (red) and DNA (blue, Hoechst), note the prominent Mrc2 expression within the podocyte (pc) cytoplasm (yellow arrows) and at foot processes lining and forming the GFB (white arrows) in the vehicle mouse, which is reduced upon proteasome inhibition. F In vivo biotinylation of glomeruli from epoxomicin- or vehicle-treated mice after streptavidin pulldown of biotinylated membrane proteins. Immunoblots depict Mrc2 and nephrin abundance within the biotinylated plasma membrane protein fraction, Mrc2 and nephrin densitometric analysis normalized to the amount of total precipitated biotinylated proteins, mean ± SEM, n = 9 (Mrc2), n = 5 (veh nephrin), n = 8 (epox nephrin) mice, pooled data from 2 independent experiments, *p = 0.0179, **p = 0.0078, two-sided Mann–Whitney U test. Scheme was created with BioRender.com. Source data are provided as a Source Data file.

Fig. 8. Summary of the effects of 20S proteasome modulation in glomerular filtration barrier functionality.

A Glomerular cell type proteostasis differentially depends on the proteasome and lysosome system. Conserved between species, podocytes and glomerular endothelial cells exhibit a prominent expression and activity of the proteasome. Proteasome constitution differs: Podocytes exhibit a prominent abundance and activity of the constitutive (c20S) proteasome, whereas GEnCs exhibit a prominent abundance and activity of the immuno (i20S) proteasome. Varying between species, abundance of lysosome subtypes (Limp2 versus Lamp2-positive) differs. As a common theme, lysosome abundance is highest in mesangial cells and GEnCs. B Modulation of β5i-subunit expression in GEnCs affects the morphology of GEnCs as well as of podocytes and the GBM. GEnC cell size increases, fenestrations are lost and molecules such as IgG accumulate intracellularly. The GBM is thickened and focally split. Podocyte foot processes are effaced. GEnCs exhibit decreased MHC-I at the surface and altered expression of FcRN (endocytic receptor to IgG and albumin) by histology but not total protein level, suggesting alterations in membrane protein trafficking (1.). C Global inhibition of the constitutive as well as of the immune β-subunit activities results in proteinuria. MCs show little alterations, whereas podocytes, GEnCs, and GBM are affected. Podocytes accumulate proteasome substrates intracellularly, and immunoglobulins in the subepithelial space accentuated at the slit diaphragm. The GBM is thickened and split. GEnCs lose their fenestration. Surface levels of the slit diaphragm protein nephrin (2.) and of the endocytic receptor to collagen Mrc2 (3.) are decreased, as is the uptake of substrates by endocytosis (4.). Podocyte plasma membrane dynamics and vesicle movement are reduced. Proteasome impairment in MCs is compensated by the lysosomes, compensation of proteasome impairment in podocytes by lysosomes is scant. Scheme was created with BioRender.com.