Amiloride Reduces Urokinase/Plasminogen-Driven Intratubular Complement Activation in Glomerular Proteinuria

Visual Abstract

Abstract

Significance Statement

Proteinuria predicts accelerated decline in kidney function in CKD. The pathologic mechanisms are not well known, but aberrantly filtered proteins with enzymatic activity might be involved. The urokinase-type plasminogen activator (uPA)-plasminogen cascade activates complement and generates C3a and C5a in vitro/ex vivo in urine from healthy persons when exogenous, inactive, plasminogen, and complement factors are added. Amiloride inhibits uPA and attenuates complement activation in vitro and in vivo. In conditional podocin knockout (KO) mice with severe proteinuria, blocking of uPA with monoclonal antibodies significantly reduces the urine excretion of C3a and C5a and lowers tissue NLRP3-inflammasome protein without major changes in early fibrosis markers. This mechanism provides a link to proinflammatory signaling in proteinuria with possible long-term consequences for kidney function.

Background

Persistent proteinuria is associated with tubular interstitial inflammation and predicts progressive kidney injury. In proteinuria, plasminogen is aberrantly filtered and activated by urokinase-type plasminogen activator (uPA), which promotes kidney fibrosis. We hypothesized that plasmin activates filtered complement factors C3 and C5 directly in tubular fluid, generating anaphylatoxins, and that this is attenuated by amiloride, an off-target uPA inhibitor.

Methods

Purified C3, C5, plasminogen, urokinase, and urine from healthy humans were used for in vitro/ex vivo studies. Complement activation was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, immunoblotting, and ELISA. Urine and plasma from patients with diabetic nephropathy treated with high-dose amiloride and from mice with proteinuria (podocin knockout [KO]) treated with amiloride or inhibitory anti-uPA antibodies were analyzed.

Results

The combination of uPA and plasminogen generated anaphylatoxins C3a and C5a from intact C3 and C5 and was inhibited by amiloride. Addition of exogenous plasminogen was sufficient for urine from healthy humans to activate complement. Conditional podocin KO in mice led to severe proteinuria and C3a and C5a urine excretion, which was attenuated reversibly by amiloride treatment for 4 days and reduced by >50% by inhibitory anti-uPA antibodies without altering proteinuria. NOD-, LRR- and pyrin domain-containing protein 3-inflammasome protein was reduced with no concomitant effect on fibrosis. In patients with diabetic nephropathy, amiloride reduced urinary excretion of C3dg and sC5b-9 significantly.

Conclusions

In conditions with proteinuria, uPA-plasmin generates anaphylatoxins in tubular fluid and promotes downstream complement activation sensitive to amiloride. This mechanism links proteinuria to intratubular proinflammatory signaling. In perspective, amiloride could exert reno-protective effects beyond natriuresis and BP reduction.

Clinical Trial registry name and registration number:

Increased Activity of a Renal Salt Transporter (ENaC) in Diabetic Kidney Disease, NCT01918488 and Increased Activity of ENaC in Proteinuric Kidney Transplant Recipients, NCT03036748.

Introduction

Proteinuria is an independent risk factor for progressive kidney injury and promotes renal inflammation, tissue damage, and fibrosis.^1,2 Besides albumin, plasminogen and complement factors have been identified in urine from patients with glomerular proteinuria across various etiologies.^3–7 After filtration barrier injury, filtered plasminogen is activated by urokinase-type plasminogen activator (uPA) in tubular fluid.^8–10 Activated plasmin contributes to salt and water retention by proteolytic activation of the epithelial sodium channel (ENaC),^10–13 which is alleviated by the diuretic drug amiloride, both directly and likely through off-target inhibition of uPA-mediated plasmin formation.^9,14,15

In vitro, plasmin effectively cleaves complement factors C3 to C3a and C5 to C5a, generating biologically active proinflammatory anaphylatoxins through a noncanonical pathway.^16–20 Moreover, plasmin inhibits convertase formation by inactivating C3b but promotes terminal complement complex (C5b-9) activation independently from normal convertase activity.^20–22 However, data are conflicting because plasmin has been shown to inhibit C5b-9 formation by degrading C5 to smaller fragments.²¹ In vivo, C3a is elevated in the urine of mice with proteinuria and promotes epithelial to mesenchymal transition of proximal tubular epithelial cells and fibrosis via the C3a receptor (C3aR).²³ C3aR is highly expressed in the kidneys, and its distribution is restricted to epithelial cells with both luminal and basolateral expressions in the proximal tubules.²⁴ C5a receptors (C5aR1 and C5L2) are also expressed in proximal tubular epithelial cells,²⁵ and C5a may also be an important driver of tubulointerstitial inflammation.²⁶ In a series of elegant experiments in rats with C6 deficiency, intact C5b-9 formation was important for kidney injury in experimental proteinuria, but not in a range of nonproteinuric kidney injury models.^27–30 Furthermore, at least some complement must originate from systemic circulation because local synthesis in kidney tissue alone is not sufficient for renal C5b-9 formation and deposition.³¹ There is ample evidence for proinflammatory signaling of filtered complement components, but the activation pathway is not clear.

Taken together, uPA-plasminogen could promote kidney inflammation and, in the long term, fibrosis through noncanonical continuous complement activation in tubular fluid. All necessary components are aberrantly filtered from plasma and concentrated in tubular fluid in conditions with proteinuria, as demonstrated by our group and others, in patients with preeclampsia, nephrotic syndrome, and transplant nephropathy.^5,11,32–36 As this is related to the leakage of plasma proteins above a certain molecular size and activation would occur after filtration and not in the glomerulus, it is likely that this mechanism is not linked to any specific kidney disease but a common feature of proteinuria across etiologies.

We hypothesized that in conditions with proteinuria, (1) uPA-plasminogen generates C3a and C5a from filtered C3 and C5, leading to proinflammatory signaling and propagation of the terminal complement pathway in tubular fluid, and (2) that this reaction can be mitigated by amiloride via inhibition of uPA. The hypotheses were tested with purified proteins in vitro and in samples stored from studies in patients with proteinuria treated with amiloride.¹⁵ In addition, data were corroborated in mice with proteinuria by inducible deletion of podocin and subsequent treatment with amiloride and uPA-neutralizing antibodies.³⁷

Methods

Study Populations

This study includes samples from three patient cohorts.

1. Patients with albuminuria: Urine samples from patients with urinary albumin-creatinine ratio (UACR) ≥300 mg/g (n=10) and urine from controls with UACR <30 mg/g (n=6) were collected as part of a cross-sectional study of injury mechanisms in proteinuria, as published earlier.³⁵ Nontransplanted patients, aged 18–75 years, who were not treated with amiloride or a mineralocorticoid receptor blocker were included. Control urine was collected from healthy volunteers. Baseline clinical data, medical history, and kidney disease were not recorded. Urine samples were added cOmplete protease inhibitor cocktail, one tablet per 50 ml urine (Roche, Basel, Switzerland), directly on voiding. The study was approved by the Research Ethics Committee of the Region of Southern Denmark (Project-ID: S-20160020) and the Danish Data Protection Agency (ID: 2018-10-058).

2. Kidney transplant recipients (KTRs) with albuminuria: We included urine and plasma samples from a published, prospective, pharmacodynamic study in KTRs with UACR <30 mg/g (n=7) or ≥300 mg/g (n=7).³⁸ Participants were included from November 2017 to September 2021 at the Department of Nephrology, Odense University Hospital, Odense, Denmark. Inclusion criteria were age between 18 and 75 years and KTRs with or without albuminuria, defined as urine UACR ≥300 mg/g or <30 mg/g, respectively, in two consecutive measurements. Exclusion criteria were treatment with amiloride, spironolactone, aldosterone or analogs, tranexamic acid, pregnancy, hyperkalemia (plasma potassium >5.0 mmol/L), eGFR <30 ml/min per 1.73 m², or incapability to cooperate or inability to understand the study protocol. Age, eGFR, and plasma albumin were similar between groups at baseline, but albuminuric KTRs had longer time from transplantation.³⁸ Spot urine samples were added protease inhibitor cocktail (cOmplete) on voiding and stored at −80°C. The study was approved by the Ethics Committee of the Region of Southern Denmark (Project-ID: S-20150015) and the Danish Data Protection Agency and registered at ClinicalTrials.gov (NCT03036748).

3. Patients with diabetic nephropathy (DN) treated with amiloride for 2 days: Biobanked urine and plasma samples were obtained from patients with type 1 diabetes with (n=15) and without (n=12) DN participating in a published, prospective, pharmacodynamic study on the acute natriuretic effect of amiloride.¹⁵ Participants were treated with high-dose amiloride on a fixed sodium intake (200 mmol/d). Amiloride 10 or 20 mg/d was administered for 2 days depending on body weight. The study included patients with albuminuria, defined as UACR ≥300 mg/g or albumin excretion ≥300 mg/d, and without albuminuria, defined as UACR <30 mg/g. In this cohort, urine samples were not added protease inhibitors. The study was approved by the Ethics Committee of the Region of Southern Denmark (Project-IDs: S-20130061 and S-20190093) and registered at ClinicalTrials.gov (NCT01918488). The use of amiloride was approved by the Danish Health and Medicines Authority (2013061668).

All participants gave written informed consent, and the studies were performed in accordance with the Declaration of Helsinki. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the Declaration of Istanbul on Organ Trafficking and Transplant Tourism.

Animal Experiments

Podocin Knockout Mice

We included 24-hour urine samples from a published interventional study in a conditional, inducible podocin gene knockout (KO) mouse model of FSGS (NphS2^lox/lox mice on 129/Svj background crossed with tamoxifen‐inducible UBC‐Cre/ERT2 transgenic mice; Jackson Laboratory, Bar Harbor, ME).^39–41 The interventions were amiloride or an anti-uPA antibody.³⁷ In short, mice were fed standard chow with tap water ad libitum in 12-hour light/dark cycles. Urine was collected directly into tubes containing cOmplete (Roche) protease inhibitor cocktail and stored at −80°C.³⁷ A baseline 24-hour urine sample was collected before induction, and animals were placed in metabolic cages again on day 10 after induction for 10 consecutive days. Amiloride was administered as an intraperitoneal (i.p.) injection twice daily in a step-up fashion between days 13 and 17 (2.5 mg/kg on days 13 and 14; 10 mg/kg on days 15 and 16; n=7). A second group received vehicle (n=7). Intervention was followed by a 3-day washout period to day 20. Samples from baseline, day 12 (before), day 16 (during amiloride), and day 19 (after amiloride) were analyzed. Another series of mice were given an uPA-inhibiting antibody (120 mg/kg mouse/d; n=5) or a similar amount of an isotype-matched control antibody (n=4) i.p. from day 9–20, which effectively blocked uPA.^42,43 Samples from baseline and days 12, 16, and 19 were analyzed. Kidneys were harvested from each mouse at the end of experimentation; one was homogenized for protein analysis and the other perfused and embedded in paraffin, as previously described.³⁷

Mice with DN and Hypertension

We also included kidney tissue from a previously published study in mice with streptozotocin-induced DN and angiotensin II–induced hypertension that received intravenous amiloride infusion 2 mg/kg per day or vehicle for 4 consecutive days.⁴⁴ Kidneys were harvested and snap frozen in liquid nitrogen at day 4 of amiloride infusion, and RNA was extracted as previously described.⁴⁴ These samples were included because the amiloride-treated podocin KO mice had a washout period before kidney harvesting, and there were no tissues left for mRNA extraction in the anti-uPA series. Kidney tissue was available from only n=4 in the vehicle group and n=4 in the amiloride group.

The studies were approved by the Danish Animal Experiment Inspectorate under the Danish Ministry of Justice (approval numbers 2013‐15‐2934‐00890 and 2012‐15‐2934‐00126), and all procedures were executed in accordance with the Danish National Guidelines for handling and care of animals and to the published guidelines of the National Institutes of Health. An overview of the animal experiments is presented in Supplemental Table 1.

Reagents and Buffers

Human C3, C3a, C5, and C5a protein (A113, A118, A120 and A144, Complement Technology, Tyler, TX), uPA (CC4000, Millipore, Temecula, CA), plasmin (Hplasmin, Enzyme Research Laboratories, South Bend, IN), plasminogen (HPG 2001, Enzyme Research Laboratories, South Bend, IN), amiloride hydrochloride (Sigma-Aldrich, St. Louis, MI), benzamil hydrochloride hydrate (Sigma-Aldrich), triamterene (Sigma-Aldrich), aprotinin (Trasylol, Nordic Group, Limhamn, Sweden), phosphate buffered saline (PBS), PBS with 0.05% tween20 (PBStw), Tris buffered saline (TBS), TBS with 0.05% Tween20 (TBStw), human serum albumin (HSA, Statens Serum Institut, Copenhagen Denmark), fetal bovine serum (FBS, Biowest, Nuaillé, France), and skim milk powder (Sigma-Aldrich) were used.

In Vitro Cleavage of C3 and C5 Protein by uPA-Plasminogen and the Effect of Amiloride

uPA (2–10 µg/ml) was incubated with purified human C3 (50 µg/ml) or C5 (50 µg/ml) and plasminogen (50 and 6.3 µg/ml, respectively) for 60 minutes at 37°C in TBS (total volume 40 µl) to assess C3a and C5a fragment release. uPA was preincubated with amiloride (4–4000 µmol/L) or aprotinin (1–1000 KIU/ml) for 10 minutes at room temperature. Reactions were assessed by ELISA for C3a and C5a and sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by Coomassie staining or immunoblotting. Directly after incubation, aprotinin 1000 KIU/ml was added to stop the reaction.

In Vitro Activation of Complement by Human Urine

C3 10 µg/ml and plasminogen 40 µg/ml or TBS were added to normal human urine (pooled from seven healthy individuals) and incubated for 45 minutes at 37°C. Similarly, C5 10 µg/ml was incubated with 0.05–50 µg/ml plasminogen or TBS in normal human urine for 70 minutes at 37°C. Inhibition was performed by preincubating normal human urine with amiloride 2 mM, aprotinin 500 KIU/ml, or EDTA 20 mM for 10 minutes at room temperature. Aprotinin 1000 KIU/ml was added to stop the reaction.

ELISA for Complement Activation Split Products

Human C3a was measured using mouse monoclonal anti-HuC3a antibodies GAU 017-01 (2 µg/ml, Bioporto, Hellerup, Denmark) as capture and biotinylated GAU 013-16 (1 µg/ml, Bioporto)^45,46 as detection, and human C5a was measured using mouse monoclonal anti HuC5a GAU 025-05 (5 µg/ml, Bioporto),⁴⁷ and biotinylated rabbit polyclonal anti-HuC5a PA5-35000 (1 µg/ml, Invitrogen, Carlsbad, CA), in a PBS system with aprotinin 200 KIU/ml (Nordic Group) and EDTA 10 mM added to the sample buffer if not stated otherwise. Standard curves were generated using serial dilutions of purified C3a and C5a, respectively.

Human C3dg was measured using monoclonal rat anti-HuC3dg 15-39-06 IgM 2.5 µg/ml,^48–50 biotinylated rabbit polyclonal anti-HuC3d diluted 1:500 (A0063, Agilent-Dako, Santa Clara, CA), and sC5b-9–associated C9 neoantigen was measured using mouse anti-Hu C9 neoantigen WU13-15 2.0 µg/ml (HM2264, Hycult Biotech, Uden, The Netherlands) and biotinylated mouse monoclonal anti-Hu-C9 antibodies (8-12-67 and 8-12-71, Bioporto), both diluted 1:500,⁵¹ in TBS as described previously.³⁵ The procedure for biotin conjugation has been described.⁵² The specificity of rat anti-huC3dg 15-39-06 for plasmin-generated C3 split products was tested by incubating C3 (50 µg/ml) for 15, 60, and 180 minutes with active plasmin (25 µg/ml) in TBS and then subjecting the reacted samples to immunoblotting and ELISA. Biotin conjugation (1 mg N-hydroxydsuccinimide activated biotin/6 mg antibody) was performed as described previously.⁵²

Mouse C3a was measured in ELISA using rat anti-mouse C3a I87-1162 (coat 2.0 µg/ml in PBS pH 6.5, BD Pharmingen, San Diego, CA) as capture and biotin rat anti-mouse C3a I87-419 (1:1000, BD Pharmingen) as detection. Mouse C5a was measured using rat anti-mouse C5a I52-1486 (coat 2.0 µg/ml in carbonate buffer pH 9.6, BD Pharmingen) as capture and biotin rat anti-mouse C5a I52-278 (1:500, BD Pharmingen) as detection. The C5a assay was blocked for 1 hour in PBS with 10% FBS. Samples were diluted (1:10–1:2000) in PBStw with 10 mM EDTA and 200 KIU/ml aprotinin sample buffer. Reagents were diluted in PBStw wash buffer. Normal mouse serum (Invitrogen) added 10 mM EDTA was defined as 1000 units and used to produce standard curves in two-fold dilution for both assays. Single-use aliquots were stored at −80°C. Kidney injury molecule 1 (KIM-1) was measured in urine and tissue using mouse KIM-1 ELISA Kit (ab213477, Abcam, Cambridge, United Kingdom).

Plates were developed with streptavidin-conjugated horseradish peroxidase (1:3000–5000, Strep-HRP, Invitrogen), and 3,3′,5,5′-tetramentylbenzidine (TMB). TMB-ONE (Kementec, Taastrup, Denmark) was used for mouse C3a, and 1-Step Ultra TMB (Pierce, Thermo Fischer Scientific, Rockford, IL) was used for all other ELISAs. TMB reaction was stopped with 0.2 M H₂SO₄, and plates were read at 450 nm on a Vmax microplate reader (Molecular Devices, San Jose, CA). Intra-assay and interassay variations were <10% and <20%, respectively. Samples from the same study were analyzed on the same plate. If this was not possible, groups were distributed evenly between plates. Samples were thawed as few times as possible, and control/intervention samples experienced the same number of freeze/thaw cycles with exception of occasional reruns.

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis and Immunoblotting

Electrophoresis was performed using Bolt 4%–12% Bis-Tris Plus gels and 2-N-morpholinoethanesulfonic acid running buffer using the NuPAGE system (Invitrogen). Samples were mixed with lithium dodecyl sulfate NuPage Sample buffer 4× and heated to 85°C for 10 minutes. Reduction of disulfide bridges was performed with 0.1 M 1,4-dithiothreitol (D9779 Sigma-Aldrich) to better visualize protein cleavage. Novex Sharp unstained or prestained ladders (Invitrogen) were used.

Coomassie staining was performed by incubating gels for 1 hour with Novex SimplyBlue SafeStain (Invitrogen) and destaining overnight in deionized water. Immunoblotting was performed by blotting gels on to 0.2 µm Trans-Blot Turbo transfer membranes (Bio-Rad Laboratories, Hercules, CA) and blocking with PBStw or TBStw with 3% skim milk. Membranes were incubated overnight in PBStw or TBStw with rabbit anti-human C3a whole serum (A218, 1:1000, Complement Technology), rabbit anti-Hu rC5a (A221, 1:1000, Complement Technology), rabbit anti-mouse collagen I (ab21286, 1:2000, Abcam), rabbit anti-mouse α-smooth muscle actin (αSMA, 1:10,000, ab5694, Abcam), rabbit anti-mouse α-tubulin (ABT171, 1:10,000, Millipore, Burlington, MA), rabbit anti-heme oxygenase 1 (HO-1 ADI-SPA-896, 1:1000, Enzo Life Sciences, Farmingdale, NY), rabbit anti-super oxide dismutase 1 (SOD1, ADI-SOD-100, 1:1000, Enzo Life Sciences), rabbit anti-super oxide dismutase 2 (06-984, 1:1000, Upstate Cell Signaling Solutions/Millipore, Lake Placid, NY), rabbit anti-mouse NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3, D4D8T, 1:1000, Cell Signaling Technology, Danvers, MA), mouse anti–E-cadherin (clone 36/E, 1:1000, BD Biosciences, Franklin Lakes, NJ), biotin-conjugated rat anti-HuC3dg 15-39-06 IgM 1 µg/ml, or biotin-conjugated rabbit anti-HuC3d (A0063, 1:5000, Agilent-Dako). After washing, horseradish peroxidase (HRP)–conjugated goat anti-rabbit IgG (P0448, 1:2000, Agilent-Dako), goat anti-mouse IgG-HRP (P0447, 1:2000, Agilent-Dako), or strep-HRP (Invitrogen, 1:5000) was used. Bands were visualized by enhanced chemiluminescence (Perkin Elmer, Waltham, MA) as specified by the producer. Gels and membranes were documented using a ChemiDocXRS+ and Imagelab software v.6 (Bio-Rad Laboratories).

Quantitative PCR

Quantitative PCR was performed with universal SYBR green supermix (Bio-Rad) on an AriaMx real-time quantitative PCR system (Agilent Technologies, Santa Clara, CA), as described previously with minor modifications.⁴⁴ In short, samples were run for 40 cycles, 95°C 3/15 minutes and 40 cycles of 95°C 15 seconds and 60°C 15 seconds (56°C for fibroblast specific protein 1 and mouse collagen I). The specific primers used are presented in Supplemental Table 2.

Light Microscopy

Paraffin-embedded mouse kidneys were cut in 2-µm slices and processed as previously described.⁵³ Hematoxylin–eosinophil and Periodic acid–Shiff stains were performed following the standard procedure. Sections were analyzed using a BX51 microscope (Olympus, Tokyo, Japan) and documented with a DP26 (Olympus) digital camera.

Statistical Analysis

Student's t test was used to evaluate data from in vitro experiments and humans. Student's t test or two-way ANOVA with repeated measures was used to test for significance in animal data. Missing data were replaced by an averaged value in the vehicle group at baseline (n=1), in the amiloride-treated group on day 19 (n=1), and in the anti-uPA group at day 19 (n=1) for ANOVA. Normal distribution was tested by Shapiro–Wilks or Kolmogornov–Smirnov's test, and data were log-transformed if this yielded normal distribution. The original study cohorts included in this study were powered to investigate a similar biologic mechanism, but sample sizes could not be altered. In vitro experiments were repeated 2–5 times. Prism 8 (GraphPad Software, San Diego, CA) was used. P<0.05 was considered statistically significant.

Results

uPA Promotes Cleavage of C3 and C5 by Activating Plasminogen

In vitro, activation of purified human C3 and C5 depended on the concentration of active plasmin, which induced a shift in the migratory pattern from 110 kDa for C3 α-chain to the predicted approximately 100 kDa α′-chain under reducing conditions (Figure 1A) and for C5 (predicted at approximately 190 kDa) to C5b (180 kDa) under nonreducing conditions (Figure 1B). Plasminogen or uPA alone had no effect on C3 cleavage (Supplemental Figure 1A).

Figure 1.

C3 and C5 are cleaved via uPA-plasminogen cascade in vitro. (A) Active plasmin induces concentration-dependent activation of C3. Nonreduced samples are shown in lanes 1–4. Corresponding samples display a migratory shift from α- to α′-chain under reducing conditions in lanes 5–8, indicating separation of C3a from the α-chain. (B) Plasmin cleaves and activates C5 generating C5b, migrating at approximately 190 and 180 kDa, respectively, under nonreducing conditions (lanes 1–4). Reduced samples are shown in lanes 5–8 and indicate that plasmin further cleaves the C5 α-chain, but not the β-chain, to smaller fragments in a concentration-dependent manner. (C) Activation of C3 is inhibited by increasing concentrations of amiloride. Nonreduced (lanes 1–6) and corresponding reduced samples (lanes 7–12) are shown. SDS–PAGE separated proteins were visualized with Coomassie staining. Experiments were performed four times and representative gels are shown. L, ladder; Pl, plasmin; Plg, plasminogen; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; uPA, urokinase-type plasminogen activator.

Formation of C3a and C5a via uPA-Plasminogen Is Inhibited by Amiloride Acting on uPA

Preincubation of uPA with amiloride (4–400 µM) or aprotinin (1–1000 KIU/ml) resulted in inhibition of uPA-plasminogen–induced activation of C3 in a concentration-dependent manner as evident by the gradual appearance of more C3 α′-chain with decreasing concentration of the inhibitor (Figure 1C and Supplemental Figure 1B). Similarly, western blotting for C3a and C5a showed that anaphylatoxin generation was attenuated by increasing concentrations of amiloride (4–4000 µM) and abolished in the absence of uPA (Figure 2, A and B). Quantification by ELISA verified the titratable inhibitory response to amiloride on C3a and C5a formation (Figure 2, C and D). In this setting, using supraphysiologic levels of uPA (2 µg/ml) for C3 activation, IC₅₀ for amiloride was 100 µM, and 4 mM inhibited the reaction to the same level as the negative control. C5a generation could not be inhibited completely even at high concentrations of amiloride. Preincubating uPA with the structurally different ENaC blockers, benzamil (up to 2 mM) or triamterene (up to 2 mM), had no effect on uPA/plasminogen-driven C3a generation (Supplemental Figure 2).

Figure 2.

Amiloride inhibits uPA-mediated generation of C3a and C5a in vitro. uPA was preincubated for 10 minutes with different concentrations of amiloride or with TBS. Next, pure plasminogen and C3, or C5, were added and incubated at 37°C. The mixture was subsequently subjected to immunoblotting. (A) Immunoblot for C3a (predicted migrating at 9 kDa); lanes 1–4: uPA-plasminogen-C3 with amiloride 4 µM–4 mM, lane 5: full reaction without amiloride, lane 6: plasminogen with C3 (negative control), and lane 7: pure C3a (100 ng, positive control). (B) Immunoblot for C5a (predicted migration at 9 kDa but appears glycosylated at 15 kDa); lanes 1–4: uPA-plasminogen-C5 with amiloride 4 µM–4 mM, lane 5: full reaction without amiloride, lane 6: plasminogen with C5 (negative control), and lane 7: pure C5a (50 ng, positive control). Inhibition of the uPA-plasminogen C3/C5 reactions by amiloride was also assessed by specific ELISA for (C) C3a and (D) C5a. Samples were run in two-fold serial dilution. Representative experiments of n=2–3 are shown. ODF, optical density fluorescence; TBS, Tris buffered saline.

C3 and C5 Are Activated by Urine from Healthy Persons in the Presence of Exogenous Plasminogen

To examine whether the reactions occur in urine with a physiologic level of endogenous uPA, C3 was incubated in urine from healthy persons for 1 hour at 37°C. Incubation of C3 in urine alone had negligible effect, but when exogenous plasminogen was added, C3a was generated (Figure 3A). This reaction was inhibited by amiloride (2 mM) and by the serine protease inhibitor aprotinin (500 KIU/ml), but not by the calcium chelator EDTA (20 mM, Figure 3B). When C5 was incubated in urine with increasing concentrations of plasminogen, a gradual increase in C5a generation was observed (Figure 3C). This was reduced significantly by aprotinin but not by amiloride (Figure 3D).

Figure 3.

Urine from healthy humans promotes anaphylatoxin generation from C3 and C5 only in the presence of exogenous plasminogen. ELISA for C3a showing the effect of (A) ex vivo addition of C3 (10 µg/ml) to normal human urine in the presence or absence of plasminogen (40 µg/ml). (B) C3a generation from C3 in the presence of exogenous plasminogen in urine preincubated with amiloride (500 µM), aprotinin (500 KIU/ml), or EDTA (20 mM) (experiment n=5 run in parallel). ELISA showing C5a generation from (C) C5 (10 µg/ml) in urine when increasing amounts of plasminogen (0.05–50 µg/ml) were added and (D) in urine preincubated with amiloride, aprotinin, or EDTA when C5 (10 µg/ml) and plasminogen (50 µg/ml) were added (n=4 per group). (D) C5a generation from C5 in the presence of exogenous plasminogen in urine preincubated with amiloride (500 µM), aprotinin (500 KIU/ml), or EDTA (20 mM, experiment n=4 run in parallel). Means with scatter are presented in (B) and (D). *P < 0.05; ***P < 0.001.

Next, in urine samples from two different patient cohorts (patients and KTRs with albuminuria and UACR ≥300), C3a/creatinine ratios were significantly increased compared with controls with UACR <30 mg/g (Figure 4A). In KTRs, plasma C3a concentration was significantly increased (Figure 4B). When data from both cohorts were pooled, urine C3a correlated log-linearly to urine albumin in individuals with UACR ≥300 mg/g (Figure 4C). Similarly, urine C5a/creatinine ratios were significantly higher in the albuminuria groups (Figure 4D), as was KTR plasma C5a (Figure 4E). Urine C5a concentration also related directly and significantly to urine albumin (Figure 4F).

Figure 4.

Urine excretion of anaphylatoxins in humans relates to proteinuria. (A) C3a/creatinine ratio in urine from healthy controls (n=6), in patients with albuminuria (n=10), and KTRs with (n=7) or without (n=7) clinically significant albuminuria. (B) Corresponding plasma concentrations for KTRs. (C) Urine C3a and albumin concentrations from pooled individuals with albuminuria (n=17) were related in a log–log linear fashion: slope with 95% CI: 2.4 (1.7 to 3.8). (D) C5a/creatinine ratio for healthy controls, patients with albuminuria, and KTRs with or without albuminuria. (E) C5a plasma concentration in KTRs. (F) Log–log linear relationship between pooled urine C5a and albumin concentrations: slope 2.3 (1.6 to 4.0). Medians with scatter are presented in (A), (B), (D), and (E). Log–log lines are plotted in (C) and (F). Unpaired t test *P < 0.05; **P < 0.01; ****P < 0.0001. CI, confidence interval; KTR, kidney transplant recipient; UACR, urine albumin-creatinine ratio.

High-Dose Amiloride for 2 Days Lowers C3dg and sC5b-9 Excretion in Patients with DN

To examine if similar inhibitory effects on complement cleavage could be observed with amiloride in vivo in patients, stored spot urine and plasma samples from patients with and without DN treated with amiloride for 2 days were used.¹⁵ As published previously, amiloride treatment reduced urine excretion of active plasmin in these patients.¹⁵

In nonprotease inhibited urine, C3dg was detectable in at least one of two paired samples from five of 15 patients, and in those patients, C3dg/creatinine decreased significantly (Figure 5A). sC5b-9–associated C9 neoantigen was detectable in ten of 15 patients, and the C9 neoantigen/creatinine ratio was significantly reduced in response to amiloride treatment (Figure 5B). In most controls, urine C3dg/creatinine and C9 neoantigen/creatinine ratios were below detection range, and no effect of amiloride was, therefore, observed (Figure 5, A and B). In corresponding plasma samples, no significant changes were detected in C3dg and C9 neoantigen concentration between groups or in response to amiloride (Figure 5, C and D). UACR (median with interquartile range) in the nephropathy group was 207 (138–577) mg/g and 7 (1–10) mg/g in controls, and as published previously, albumin excretion, creatinine clearance, BP, and plasmin excretion in urine decreased after amiloride treatment in the nephropathy group.¹⁵ The source of C3dg in urine was evaluated in vitro by reacting plasmin with purified C3, which generated C3dg-like, C3d, and C3d-like fragments, as reported by others.²² The fragments had signals with similar intensity as C3dg in human plasma (diluted 1:100) as demonstrated by immunoblotting with polyclonal anti-C3d (A0063) with slightly different migration patterns, whereas monoclonal anti-C3dg 15-39-06 resulted in only very weak signals for the plasmin-generated fragments, even after prolonged exposure of the membrane (Supplemental Figure 3A). ELISA for C3dg did not result in any signal in the same samples diluted 1:20 (equivalent of 0.25 µg C3 per 100 µl and corresponding to the approximate content of nonactivated C3 in plasma diluted 1:400) (Supplemental Figure 3B).

Figure 5.

Effect of high-dose amiloride for 2 days in patients with type 1 diabetes with and without diabetic nephropathy (DN). Graphs show (A) urinary C3dg/creatinine ratio and (B) urinary C9 neoantigen (C9 neo)/creatinine ratio for controls and patients with DN. In the DN group, only patients with urine complement within measurable range were included in the statistical analysis and plotted (C3dg: n=5 of 15, C9 neoantigen: n=10 of 15). All controls are shown. (C) Corresponding plasma concentrations of C3dg and (D) C9 neoantigen. Paired plasma samples from four controls and four cases were missing from the biobank and were thus not measured. *P < 0.05.

24-Hour C3a and C5a Excretion Rates Are Attenuated by Inhibition of uPA in Mice with Proteinuria

To discriminate between a direct effect of amiloride on anaphylatoxin formation and a beneficial therapeutic effect on the glomerular filtration barrier, we analyzed samples from mice with inducible podocin KO. In this model of FSGS, proteinuria developed progressively, beginning 10 days after Cre induction, and the degree of albuminuria was similar with or without amiloride treatment as reported earlier.³⁷ In these mice, 24-hour urine excretion of anaphylatoxins were below the limit of detection at baseline (Figure 6, A and B). Twelve days after Cre induction, C3a and C5a were detectable in the urine, and excretion rates were not different between groups. At day 16, the urine excretion of C3a and C5a had increased further but was significantly attenuated by treatment with amiloride for 4 days (P = 0.024 and P = 0.031 for C3a and C5a, respectively). After 3 days of washout without amiloride (day 19), differences in urine excretion were no longer significant (Figure 6, A and B).

Figure 6.

Urine anaphylatoxins in conditional podocin KO mice with severe proteinuria. Twenty-four–hour urine excretion of anaphylatoxins was measured in two series of mice with conditional KO of podocin, amiloride for 4 days or vehicle, or an anti-uPA–inhibiting antibody or a control antibody. A baseline urine sample was collected on day −7, and Cre induction with tamoxifen was performed on days −4 to 0. Mice in the amiloride trial were divided in two groups receiving either amiloride (■ and dashed line) or vehicle (● and full line) by two daily i.p. injections³⁷ and kept in metabolic cages from days 10–20 during the experiment. Amiloride was administered in a step-up fashion: 2.5 mg/kg on days 13 and 14 and 10 mg/kg on days 15 and 16 (shaded area in A and B), followed by a 3-day washout period.³⁷ Twenty-four–hour urine excretion rates are shown for (A) C3a and (B) C5a, in amiloride- (n=7) and vehicle- (n=7) treated podocin KO mice. As published earlier, albumin excretion did not change between groups in amiloride-treated animals.³⁷ In the second series, inhibiting anti-uPA antibodies (120 mg/kg per day) or a similar dose of isotype-matched control IgG antibodies were injected i.p. from days 9–20 (shaded areas in C–E).³⁷ Urine excretion rates are presented for (C) C3a, (D) C5a, and (E) albumin in anti-uPA-mice (□ and dashed line, n=5) and isotype control antibody-mice (○ and full line, n=4). Urine was collected directly into tubes containing protease inhibitor cocktail, and excretion rates are normalized to body weight measured daily. Kidney tissue protein concentrations are shown for (F) C3a and (G) C5a. Data are presented as mean with SEM in (A–G) and mean with SD in (F) and (G). *P < 0.05 compared with vehicle by Student's t test. #P < 0.05 by 2w-ANOVA between groups. Mixed effects analyses with missing values were also tested and produced similar results. 2w-ANOVA, two-way ANOVA; i.p., intraperitoneal; KO, knockout; ns, not significant.

In a second series of experiments, podocin KO mice were treated with uPA-inhibiting antibodies (n=5) or a similar amount of an IgG-isotype control antibody (n=4) for 11 days by i.p. injection.³⁷ This dose blocked uPA and formation of active plasmin from plasminogen in the urine.^37,42 Inhibition of uPA caused significant reductions in 24-hour urine C3a and C5a excretion rates (two-way ANOVA: P = 0.012 and P = 0.019, respectively) (Figure 6, C and D). As for amiloride, the effect of inhibiting uPA on anaphylatoxin generation was not associated with any change in 24-hour urine albumin excretion (Figure 6E). In kidney tissue homogenates, there were no differences in kidney tissue anaphylatoxin levels (Figure 6, F and G). NLRP3 inflammasome-protein was significantly lower in uPA-inhibited mice, but oxidative stress markers heme oxygenase 1 and super oxide dismutase 1 and 2 were similar between groups (Figure 7, A and B). Immunoblotting demonstrated bands with similar intensity for E-cadherin in both groups but more intense bands for αSMA and collagen I in anti-uPA–treated mice (Figure 7, C and D). The epithelial injury marker KIM-1 increased significantly in urine from induction to day 19, but there was no difference between groups (Figure 7E). Kidney tissue KIM-1 protein was modestly but significantly higher in anti-uPA–treated mice (Figure 7F). Hematoxylin–eosin and Periodic acid–Shiff stains showed increased morphologic kidney injury with cell infiltration, and tubular atrophy and dilatation in podocin KO mice compared with the control, but morphology was not grossly different between isotype and anti-uPA groups (Supplemental Figure 4).

Figure 7.

Kidney inflammation and fibrosis in anti-uPA–treated podocin KO mice. Kidneys were harvested from podocin KO mice after 11 days of treatment with either anti-uPA antibodies or isotype control antibodies, and 24-hour urine samples from baseline and days 12, 16, and 19 were analyzed.³⁷ (A) Immunoblots for oxidative stress markers HO-1 (expected migration 33 kDa and approximately 25 kDa in truncated form), SOD1 (20 kDa), and SOD2 (27 kDa) and α-tubulin. (B) Semiquantified data from immunoblots for HO-1, SOD1, SOD2, and NLRP3 normalized to α-tubulin. (C) Immunoblots for E-cad (expected migration 110 kDa), αSMA (expected migration 42 kDa), collagen I (expected migration of procollagen I approximately 250 kDa; collagen I 138 kDa but observed approximately 55 kDa, weak bands also appear at approximately 150 and approximately 100 kDa), and α-tubulin (expected migration 50 kDa). Wild-type mouse kidney tissue was used as negative control (Neg), and kidney tissue from a wild-type mouse subjected to unilateral ureter obstruction was used as positive control (Pos). (D) Semiquantified data from immunoblots for E-cadherin, αSMA, and collagen I normalized to α-tubulin. (E) Urine excretion rate for KIM-1 in anti-uPA–treated mice (□ and dashed line, n=5) and isotype control antibody–treated mice (○ and full line, n=4). (F) Kidney tissue levels of KIM-1 protein normalized to kidney weight after 11 days of treatment. Data are presented as mean with SD. *P < 0.05. αSMA, α-smooth muscle actin; E-cad, E-cadherin; HO-1, heme oxygenase 1; KIM-1, kidney injury molecule 1; NLRP3, NOD-, LRR-, and pyrin domain-containing protein 3; SOD1, super oxide dismutase 1; SOD2, super oxide dismutase 1.

Because the amiloride-treated podocin KO mice had a 3-day washout period before termination, gene transcript abundances of inflammation and fibrosis markers were analyzed in kidney tissue from a third series of previously published mouse experiments.⁴⁴ In this setting, streptozotocin-angiotensin II diabetic-hypertensive mice treated with continuous infusion of amiloride (2 mg/kg per day) for 4 days up to termination showed no relative differences compared with vehicle-treated animals in any of the investigated inflammation or fibrosis markers (Supplemental Figure 5).

Discussion

This study demonstrates that the uPA-plasminogen cascade promotes generation of the anaphylatoxins C3a and C5a from native complement C3 and C5. This activation is significantly attenuated by inhibition of uPA in vitro and in vivo in podocin KO mice. We show that endogenous uPA in urine from healthy humans is necessary and sufficient to activate and generate anaphylatoxins in the presence of exogenous plasminogen and that this is sensitive to the potassium-sparing diuretic amiloride, an off-target uPA antagonist. Urine samples from patients and KTRs with albuminuria contain active plasmin^5,11,15,34 and have significantly elevated C3a and C5a levels in direct relation to the degree of albuminuria. In patients with DN, urine excretion of C3dg and sC5b-9–associated C9 neoantigen decreased after 2 days of treatment with high-dose amiloride, without changes in plasma concentrations but with concomitant changes in BP, eGFR, and grade of albuminuria.¹⁵ Together, it can be concluded that uPA-plasmin–driven activation of complement is a postfiltration, likely intraluminal, tubular event, not reflected in plasma. In vivo in mice with severe proteinuria, lower urine anaphylatoxin excretion was associated with less inflammasome abundance in kidney tissue, but markers of fibrosis were not lower but tended to be higher.

Plasmin has been described both as a complement activator and inhibitor.^20,21 Data support local generation of functional anaphylatoxins by plasmin in plasma at sites of coagulation.^16–20 The intraluminal milieu in the tubular system and in the urinary tract progressively differs from plasma, but data presented here document that activation of complement by uPA-plasmin can occur also under the prevailing conditions in urine. The fact that C3a and C5a are present in urine only from patients with proteinuria and that plasmin is the dominant protease in proteinuric urine^11,13 suggest that plasmin generates biologically active anaphylatoxins inside the tubular lumen. This links aberrant uPA-plasmin activity in urine to anaphylatoxin-mediated downstream proinflammatory signaling and ultimately to promotion of nephron loss. In mice, plasminogen deletion attenuates kidney fibrosis.⁵⁴ The profibrotic role of anaphylatoxins is documented for C5/C5a/C5aR and C3a/C3aR.^23,24,26,55 Moreover, anaphylatoxin receptors C3aR, C5aR1, and C5L2 are expressed on tubular epithelial cells.^24,25,56,57 The exception to this concept is perhaps minimal change disease, but the lack of inflammation and fibrosis in this form of nephrosis resulting from isolated podocyte dysfunction could be explained by the hypothetical need for a second trigger, e.g., tubular stress, for proinflammatory effector functions to be activated. Furthermore, steroids suppress complement signaling in general, and steroid-resistant minimal change disease often progresses to FSGS over time.

Amiloride is a small-molecule diuretic that is concentrated in tubular fluid and inhibits uPA activity off-target by binding to its catalytic site in a competitive manner.^58–60 This occurs in tubular fluid, but likely not in plasma where micromolar concentrations are not reached.^{14,37,60–62} In consequence, it inhibits activation of plasminogen to plasmin because uPA is the dominant activator of soluble plasminogen not bound to surfaces where tPA is relevant.^14,15,63,64 Our finding that normal human urine generates C3a from C3 in the presence of plasminogen and that this is inhibited by amiloride and aprotinin but not EDTA lends to the hypothesis that endogenous uPA in urine drives noncanonical complement activation via plasminogen. A similar pattern was observed for C5a generation, bypassing both the C3 and C5 convertases. Amiloride reaches peak urine concentrations 3–20 µM after oral dosage of 20 mg in humans⁶² and two-fold higher in rats with proteinuria compared with nonproteinuric control animals.⁹ In vitro, we found IC₅₀ for C3a generation to be 100 µM, using uPA at concentrations approximately 50 times higher than the physiologic levels found in human urine,⁶⁵ suggesting that the amiloride concentration reached in tubular fluid is sufficient for inhibition in a clinical setting. ENaC inhibitors triamterene and benzamil were without effect on uPA-mediated C3a generation, which implies that this trait is specific for amiloride and likely caused by the six-chloro group site in amilorid.^59,66 Of note, amiloride inhibits murine and human forms of uPA to a similar degree.⁵⁹

The capacity of amiloride to attenuate intratubular complement activation was documented in vivo in patients with DN where amiloride lowered plasmin,¹⁵ C3dg, and C9 neoantigen excretions. Concomitant reduction in eGFR and albuminuria could have contributed. The urine samples in this intervention study were not protease inhibited at collection and were subjected to freeze-thaw cycles, which obscured determination of anaphylatoxins. Another factor to consider is that plasmin promotes C3b inactivation to C3c-like and C3dg-like fragments, similarly but not identically to factor I, and independently of cofactors.^21,22 Consistently, our N-terminal specific anti-HuC3dg antibody did not recognize C3dg generated exclusively by plasmin. The relatively weak signal for C3dg in patient urine could thus be explained by the fact that the antibody overlooks plasmin-generated C3d/dg isoforms.

To overcome some of these constraints, human data were corroborated in the controlled setting of an inducible FSGS mouse model with massive proteinuria. Here, amiloride lowered urine anaphylatoxin excretion reversibly despite high continuous albuminuria. A decisive role for uPA in this reaction was confirmed by inhibiting uPA specifically with a monoclonal antibody, which reduced C3a and C5a urine excretion significantly without changes in albumin excretion. As urine plasmin formation was abolished by the anti-uPA antibody in these mice,³⁷ the observation strongly indicates that active plasmin contributes to the generation of anaphylatoxins in tubular fluid in vivo. The reductions in urine C3a and C5a excretion were only partial (75% and 50%, respectively), and it is likely that plasmin is not the only protease activating complement in tubular fluid during proteinuria. Blocking one pathway might provide more substrate for other proteases that could potentially mediate intratubular complement activation. This is also a possible explanation for the limited effect in the human experiment. Proteases with complement-activating abilities are thrombin,^18,67 coagulation factor Xa,²⁰ kallikrein,^5,68 and renin,⁶⁹ to name a few.

We found that the inflammasome protein NLRP3 was significantly downregulated in kidney tissue from anti-uPA–treated podocin KO mice, linking C3aR signaling and C5b-9 formation to less inflammasome activity.⁷⁰ In kidney tissue KIM-1, αSMA and collagen I increased despite reduced urine anaphylatoxins.³⁷ This is in contrast to findings by others that C3aR KO mice with adriamycin-induced proteinuria had significantly less interstitial fibrosis.²³ However, this could also be explained by significantly less albuminuria in the C3aR KO mice,²³ as we demonstrate here that the concentration of C3a in urine correlates to albumin. The inducible podocin KO model used in this study has the advantage that the intervention could be performed after induction of proteinuria and that albuminuria was similar between groups. It is, however, possible that proteinuria in the podocin KO model is so severe that other injury mechanisms, such as protein overload, overshadow the effect of reduced complement activation. This led us to examine gene expression in a less severe mouse model treated continuously with i.v. amiloride for 4 days. In this setting, the mRNA levels of αSMA, collagen I, and III were nominally, but not significantly, lower after treatment with amiloride. KIM-1 urine excretion in healthy humans is increased by saline infusion and alleviated by furosemide,⁷¹ implying that alterations in solute transport and changes in oxygen consumption within physiologic range alone can affect this parameter. Plasmin activates ENaC but inhibits Ca²⁺ transport in the tubules.^11,72 Thus, altered solute transport and oxygen consumption along the tubules could affect the expression of KIM-1 and explain why we observe more tissue KIM-1 in uPA-inhibited animals. KIM-1 might be protective in some settings (AKI) and harmful in others (CKD).^73–75

The uPA-plasminogen cascade contributes to noncanonical intratubular complement activation and anaphylatoxin generation in vivo in mice and humans, and this reaction can be attenuated by targeting constitutive endogenous uPA activity with amiloride. This mechanism links aberrantly filtered plasminogen to continuous complement activation in conditions with glomerular proteinuria. The long-term significance for inflammation, nephron loss, and fibrosis remains to be determined.

Amiloride could have anti-inflammatory effects by off-target complement inhibition in tubular fluid,^23,28,29,76 which may account for its BP-independent renoprotective effect reported previously in proteinuria.^15,64 If so, renoprotective beneficial effects of amiloride would likely be additive to other renoprotective drugs. Amiloride should be tested for these effects as primary outcomes. At a mechanistic level, uncovering the molecular inner workings of noncanonical complement activation via uPA-plasminogen-complement interaction and potentially additional proteases might be a significant step toward understanding the driving force behind progressive loss of nephrons, and ultimately kidney function, in proteinuria.

Supplementary Material

jasn-35-410-s001.pdf ^{(952.7KB, pdf)}

Disclosures

H. Andersen reports employment with EFKT. H. Birn reports consultancy for AstraZeneca, Boehringer Ingelheim, Bayer, Galapagos, and GlaxoSmithKline (GSK); research funding from GlaxoSmithKline (GSK) and Vifor Pharma; honoraria from AstraZeneca; a patent application under review and not filed at this point; and advisory or leadership role as President of the Danish Society of Nephrology and a member of working groups under the Danish Health Authority. C. Bistrup reports employment with Odense University Hospital, Odense, Denmark and University of Southern Denmark, Odense, Denmark. J.E. Henriksen reports employment with Odense University Hospital, Denmark. G.R. Hinrichs and B.L. Jensen report employment with University of Southern Denmark. I.K. Lund reports employment with and ownership interest in Novo Nordisk A/S. K. Weyer reports employment with Draupnir Bio; consultancy for Astex Pharmaceuticals, Muna Therapeutics, Novo Holdings, and Novo Nordisk Foundation; ownership interest in Denali Therapeutics, Draupnir Bio, Muna Therapeutics, Novo Nordisk, Teitur Trophics, and Verve Therapeutics; research funding from Novo Nordisk; patents or royalties from Aarhus University, Draupnir Bio, and Teitur Trophics; advisory or leadership role for Draupnir Bio; and other interests or relationships with Aarhus Lifescience and Biotech Alliance (ALBA) and Danish Nephrology Society. All remaining authors have nothing to disclose.

Funding

B.L. Jensen: Augustinusfonden Novo Nordisk Fonden (NNF17OC0028972), Det Frie Forskningsråd (8020-00212B). H. Birn: Karen Elise Jensens Fond. G.R. Hinrichs: Region Syddanmark, Helen og Ejnar Bjørnows Fond, and Dansk Nefrologisk Selskab. G.L. Isaksson: Odense Universitetshospital (PhD fund), Syddansk Universitet, and Dansk Nefrologisk Selskab. H. Andersen: Hjerteforeningen.

Author Contributions

Conceptualization: Henrik Birn, Claus Bistrup, Gitte R. Hinrichs, Gustaf L. Isaksson, Boye L. Jensen, Kirsten Madsen, Yaseelan Palarasah.

Data curation: Henrik Andersen, Marie L. Bach, Gitte R. Hinrichs, Gustaf L. Isaksson.

Formal analysis: Gustaf L. Isaksson.

Funding acquisition: Henrik Andersen, Henrik Birn, Jan Erik Henriksen, Gitte R. Hinrichs, Gustaf L. Isaksson, Boye L. Jensen, Yaseelan Palarasah, Kathrin Weyer.

Investigation: Henrik Andersen, Marie L. Bach, Henrik Birn, Jan Erik Henriksen, Gitte R. Hinrichs, Gustaf L. Isaksson, Boye L. Jensen, Rikke Zachar, Kirsten Madsen, Yaseelan Palarasah, Kathrin Weyer.

Methodology: Henrik Andersen, Henrik Birn, Claus Bistrup, Jan Erik Henriksen, Gitte R. Hinrichs, Gustaf L. Isaksson, Boye L. Jensen, Rikke Zachar, Ida K. Lund, Kirsten Madsen, Géraldine Mollet, Yaseelan Palarasah, Kathrin Weyer.

Project administration: Claus Bistrup, Gustaf L. Isaksson, Boye L. Jensen, Kirsten Madsen, Yaseelan Palarasah.

Resources: Jan Erik Henriksen, Boye L. Jensen, Yaseelan Palarasah.

Software: Marie L. Bach, Gustaf L. Isaksson.

Supervision: Henrik Birn, Claus Bistrup, Boye L. Jensen, Kirsten Madsen, Yaseelan Palarasah.

Validation: Henrik Birn, Gustaf L. Isaksson, Boye L. Jensen, Ida K. Lund, Géraldine Mollet, Kathrin Weyer.

Visualization: Gustaf L. Isaksson, Yaseelan Palarasah.

Writing – original draft: Gustaf L. Isaksson, Boye L. Jensen, Yaseelan Palarasah.

Writing – review & editing: Marie L. Bach, Henrik Birn, Claus Bistrup, Gitte R. Hinrichs, Gustaf L. Isaksson, Boye L. Jensen, Rikke Zachar, Yaseelan Palarasah, Kathrin Weyer.

Data Sharing Statement

Previously published data were used for this study. Partial restrictions to the data and/or materials apply. Anonymized data will be shared on request.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E579.

Supplemental Table 1. Quantitative PCR primer overview.

Supplemental Table 2. Animal experiment overview.

Supplemental Figure 1. Cleavage of pure C3 in vitro: sodium-dodecyl-sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and visualization by Coomassie staining.

Supplemental Figure 2. Urokinase-type plasminogen activator (uPA)/plasminogen-mediated generation of C3a is inhibited by amiloride but not benzamil or triamterene.

Supplemental Figure 3. Plasmin cleavage of iC3b to C3dg/C3d.

Supplemental Figure 4. Light microscopy images of paraffin-embedded kidney tissue from podocin knockout (KO) mice.

Supplemental Figure 5. mRNA abundance of inflammation and fibrosis markers in streptozotocin-angiotensin II (STZ-ANGII) mice treated with amiloride.