Inhibition of Rho-associated kinase relieves C5a-induced proteinuria in murine nephrotic syndrome

Abstract

Childhood nephrotic syndrome is mainly caused by minimal change disease which is named because only subtle ultrastructural alteration could be observed at electron microscopic level in the pathological kidney. Glomerular podocytes are presumed to be the target cells whose protein sieving capability is compromised by a yet unidentified permeability perturbing factor. In a cohort of children with non-hereditary idiopathic nephrotic syndrome, we found the complement fragment C5a was elevated in their sera during active disease. Administration of recombinant C5a induced profound proteinuria and minimal change nephrotic syndrome in mice. Purified glomerular endothelial cells, instead of podocytes, were demonstrated to be responsible for the proteinuric effect elicited by C5a. Further studies depicted a signaling pathway involving Rho/Rho-associated kinase/myosin activation leading to endothelial cell contraction and cell adhesion complex breakdown. Significantly, application of Rho-associated kinase inhibitor, Y27632, prevented the protein leaking effects observed in both C5a-treated purified endothelial cells and mice. Taken together, our study identifies a previously unknown mechanism underlying nephrotic syndrome and provides a new insight toward identifying Rho-associated kinase inhibition as an alternative therapeutic option for nephrotic syndrome.

Electronic supplementary material

The online version of this article (doi:10.1007/s00018-015-1888-0) contains supplementary material, which is available to authorized users.

Introduction

Idiopathic nephrotic syndrome (INS) is a common pediatric renal disease manifested with a constellation of proteinuria, hypoalbuminemia, generalized edema and dyslipidemia. These characteristic features result from a failure of the glomerular capillary barrier to restrict urinary protein loss to less than 1 g of urine protein per square meter of body surface area per day. Although childhood nephrotic syndrome could be a comorbid of certain systemic diseases or secondary to infection or side effect of drugs, most childhood nephrotic syndrome is primary which lacks nephritis features. Minimal change disease (MCD) is the most common histological variant for primary nephrotic syndrome which accounts for over 50 % of the INS in children and responds favorably to corticosteroids treatment [1–3].

INS has been thought to be a T-lymphocyte mediated disease because it is associated with lymphocyte disorders such as lymphoma [4] and Kimura’s disease [5–7], and is occasionally observed in cases receiving interferon therapy [8]. Furthermore, isolated T cells from nephrotic patients have been demonstrated to generate a glomerular permeability factor which could induce transient proteinuria when injected into rats [9] and impair the capability of podocytes to synthesize glycoaminoglycans [10] which have been thought to be an important constituent of the glomerular permeability barrier. It is proposed that the activated T cells secrete a circulating factor which compromises glomerular capillary permeability [11]. The most compelling evidence supporting this theory comes from the experience with renal allografts [12]. In that study, nephrotic range proteinuria recurs in up to 50 % of focal segmental glomerulosclerosis (FSGS, the second common histological variant causing INS in children) victims after transplantation with a median time to recurrence of 14 days suggesting the existence of a circulating factor. However, the identity of this putative critical permeability modifier remains elusive over the years, although various growth factors [13–15] and cytokines [16] have been proposed to exert the pathogenic effect in INS.

Recently, we identified a proteinuric candidate C5a, a fragment generated by activation of complement, in the sera of INS children. C5a is a small soluble peptide fragment regulating many inflammatory pathways, including mast cells degranulation [17], neutrophil chemotaxis [18], and complement activation-related T cell functions such as T lymphocytes infiltration [19]. C5a also plays a role in causing renal injury [20–22]. Nevertheless, C5a has never been implicated in the pathogenesis of INS. We employed both murine glomerular endothelial cells and whole mice model to demonstrate that C5a could indeed elicit proteinuria. Furthermore, our study delineates the signaling pathway leading to this previously unidentified C5a effect and sheds light on potential use of Rho-associated kinase (ROCK) inhibitor for the therapy of INS or other protein losing vasculopathy.

Materials and methods

Patient selection

Twenty-two pediatric patients who were diagnosed as INS according to KDIGO (Kidney Disease Improving Global Outcomes) guideline of proteinuria (urinary protein excretion >40 mg/m²/h, or spotted urine protein >300 mg/dL or 3+ on urine dipstick), hypoalbuminemia (serum albumin ≤2.5 g/dL), generalized edema, and dyslipidemia in National Taiwan University Hospital during 2008–2012 were enrolled in this study after excluding patients with definitive diagnosis of systemic lupus erythematosus, post-streptococcal glomerulonephritis, and hepatitis B virus-related membranous nephropathy. This study was authorized by the Research Ethics Committee (REC) of National Taiwan University Hospital. Informed consents were signed by the patients and the family before blood and urine sampling.

Cytokine array

A human cytokine antibody array (Proteome Profiler™, R&D Systems) was exploited to analyze the cytokine profiles according to the manufacturer’s instruction. Digital imaging system (Bio Pioneer Tech Co., Ltd.) was used to detect the chemiluminescent signals which were further analyzed using ImageJ program.

Enzyme immunoassay

Serum C5a level was determined using C5a EIA kit from R & D Systems according to the manufacturer’s instructions. Each measurement was performed in duplicate.

Antibodies

All the antibodies used in this study are described in supplementary Table 1.

Animal study

Six-week-old male specific pathogen-free ICR mice were used and housed at the Experimental Animal Center, National Taiwan University, at a constant temperature. The ICR mice were treated with normal saline as the control or recombinant mouse C5a (R&D Systems), respectively, with the dose of 10 μg/kg (200 ng per mouse) every 3 days through tail vein injection. Blood and urine samples were collected every 2 days for albumin (ELISA kit Catalog #KA0488. Abnova, Inc.), creatinine (Catalog #KA0849. Abnova, Inc.), total cholesterol (Catalog #ab65390, Abcam, Inc.), and triglyceride (Catalog #ab65336, Abcam, Inc.) assays. Each measurement was performed in duplicate. After 21 days, mice were sacrificed in CO₂ chambers, and their kidneys were isolated for pathology study, light microscopy, electron microscopy, and immunofluorescence microscopy using goat anti-mouse IgG-FITC (sc-2010), rabbit anti-C1q (sc-25856), rat anti-C3 (sc-58926), rat anti-C4 (sc-58930) antibodies from Santa Cruz Biotechnology, Inc, and goat anti-mouse IgA (M8769) from Sigma-Aldrich Co. All the animal experiments were performed in accordance with the protocols approved by Institutional Animal Care and Use Committee (IACUC) of National Taiwan University College of Medicine and were performed in accord with the NIH guide for the care and use of laboratory animals.

Light and electron microscopic examination of murine renal tissue

The murine renal tissue was fixed in 10 % formalin and embedded in paraffin for 3-μm-thick section. These sections were prepared for H&E, PAS and chromotrope silver methenamine stains. Another renal cortical tissue (2 mm³) is fixed in 2.5 % glutaraldehyde for electron microscopic examination. After washing with 0.1 M sodium cacodylate buffer (pH 7.4), the tissue was fixed in 1 % osmium tetroxide for 1 h and then dehydrated for embedding in the epoxy resin. We checked the adequacy of glomeruli in the 1-μm-thin sections by toluidine blue stain, and then examined these thin sections on the grids stained with uranyl acetate and lead citrate for ultrastructural study using the Jeol JEM-1400 (Joel Ltd., Tokyo, Japan) transmission electron microscope.

Mouse renal glomeruli and kidney endothelial cells (KECs) preparations

The glomeruli were isolated from minced kidneys of 20 mice aged between 6 and 8 weeks under sterile conditions by gradually sieving with stainless steel sieves (150, 90, 75, 53, and 32 μm). These glomeruli were washed with 50 ml Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen Technologies) and then treated with 2.5 ml 0.1 % type II collagenase in DMEM at 37 °C for 1.5 h. The cell suspension was washed in 5 ml Hanks’ balanced salt solution containing 10 % fetal calf serum (Invitrogen). The isolated cells were incubated in 1 ml of cold M199 medium (Invitrogen) containing 1 μg/ml rabbit anti-mouse CD31 antibody (sc-1506-R; Santa Cruz Biotechnology) for 30 min with gentle agitation, and followed by addition of 50 μl Dynabeads^® M280 sheep anti-rabbit IgG (10 mg/ml), and incubated further for 30 min. The endothelial cells bound to the magnetic beads were removed from the unbound non-endothelial cells by magnetic isolation using an MPC-1 magnet (Dynal, Oslo, Norway). The cells bound on the beads were re-suspended and cultured in endothelial cell growth medium (Cell Applications, San Diego, CA, USA).

RT-PCR

Total RNA was isolated from the cells using the RNAzol B reagent (Biotecx Laboratories, Houston, TX), and then cDNA was prepared from 2 μg of the total RNA with ImProm-II RT system (Promega, Southampton, UK). The cDNAs for specific target genes were amplified by PCR with ExTaq DNA polymerase (Takara Bio Inc., Shiga, Japan) using the indicated forward and reverse primers as specified in Supplementary Table 2.

Chemicals

ROCK inhibitor Y-27632, blebbistatin and pertussis toxin (a Gαi inhibitor) were from Calbiochem. The vascular endothelial growth factor (VEGF) receptor tyrosine kinase inhibitor SU-5416 was from Sigma-Aldrich (St. Louis, MO). The TAT-C3 inhibitor was from Cytoskeleton Inc. The C5a receptor antagonist W-54011 was from Santa Cruz Biotechnology, Inc.

Western blotting analysis

Cells were lysed with PBS (pH 7.4) containing 1 % Nonidet P-40, 0.5 % sodium deoxycholate, 0.1 % sodium dodecyl sulfate (SDS), and protein inhibitor cocktail (Nacalai Tesque, Kyoto, Japan). The cellular lysates were centrifuged at 13,400×g for 10 min at 4 °C. Then protein concentration was measured by Bradford assay (Bio-Rad, Hercules, CA, USA). Fifty µg of each protein sample were processed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto polyvinylidene difluoride membranes, and immunoblotted with specific primary and secondary antibodies followed by enhanced chemiluminescence development (Amersham) with standard procedure.

Protein permeability assay on endothelial monolayer

The isolated mouse KECs or human umbilical vein endothelial cells (HUVECs) (Cell Applications, San Diego, USA) were cultured to confluency on Transwell filters (0.4 µm pore-size polycarbonate filters; Costar Corp., Cambridge, MA, USA). After serum fasting overnight in M199 medium, recombinant C5a and various pharmacological inhibitors with the indicated concentration were applied as specified. After incubation, the cells were washed, and then horseradish peroxidase (HRP) molecules (Type VI-A, 44 kDa; Sigma) at a concentration of 0.126 μM were added to the upper compartment. After further incubation for 1 h, the medium from the lower compartment was assessed for enzymatic activity using a photometric guaiacol substrate assay (Sigma).

Immunofluorescence and F-actin staining

Cells were fixed with 3.7 % paraformaldehyde for 20 min and permeabilized with 50 mM NaCl, 300 mM sucrose, 10 mM PIPES (pH 6.8), 3 mM MgCl₂ and 0.5 % Triton X-100 for 5 min. Cells were blocked with PBS plus 10 % goat serum, 1 % BSA and 50 mM NH₄Cl for 1 h, and then incubated with primary antibodies for 1 h. Cells were then incubated with the appropriate fluorescein isothiocyanate (FITC) or rhodamine-conjugated secondary antibodies for 1 h. Polymerized actin was detected using FITC-conjugated phalloidin (Invitrogen), which was diluted in PBS (2 unit/ml) with the other secondary antibody and applied to the specimens in the dark for 1 h, before being mounted with Vectashield mounting medium (Vector Laboratories). The images were viewed using Leica DMR fluorescent microscope.

Fresh frozen section preparation for confocal microscopic examination

Isolated mouse kidneys were washed in cold PBS, then transferred to 4 % paraformaldehyde in PBS for overnight incubation at 4 °C, and then subsequently transferred to 30 % sucrose in PBS for overnight again at 4 °C. The kidneys were embedded in OCT for 1 h at 4 °C and stored at −20 °C for frozen sectioning. Sections of 7.5 μm thickness were used in the following staining process. Frozen renal tissue sections were warned up in a humidity chamber at 4 °C for 48 h and rehydrated in PBS at room temperature for 10 min before staining. Sections were blocked in PBS containing 1 % bovine serum albumin, 0.1 % triton X-100 for 30 min. Primary antibodies were applied to the sections at a concentration of 1 µg/ml in the blocking buffer and incubated in humidity chambers for 1 h. The sections were washed three times with PBS containing 1 % BSA followed by incubation with florescence labeled secondary antibodies for 30 min. Then, the sections were washed three times with PBS containing 1 % BSA, wet-mounted in Prolong™ antifade (Molecular Probes Inc., Eugene, OR), sealed with nail varnish, and stored at 4 °C in the dark before processing of confocal imaging.

GTP-Rho pull-down assay

KECs were grown to near confluency before serum fasting for 16 h. Recombinant C5a (50 ng/ml) was added for 1 h before the cells were lysed with cold lysis buffer (50 mM Tris–HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5 mM MgCl2, 10 % glycerol, 50 mM NaF, 1 mM Na₃VO₄, 1 mM dithiothreitol, 50 µg/ml phenylmethylsulfonyl fluoride, 10 µg/ml each of leupeptin and pepstatin and 1 % Nonidet P-40). The cell lysates were clarified by centrifugation at 10,000g at 4 °C for 5 min and the supernatants were collected. While 5 % of the supernatants was saved as input control, the rest of the supernatants were incubated with 60 µg of glutathione S-transferase/Rhotekin Rho-binding domain fusion protein conjugated with glutathione beads (Cytoskeleton Inc., catalogue No. RT02) for 1 h at 4 °C. The beads bound with Rho-GTP were washed twice with lysis buffer, and were processed for SDS-PAGE and Western blotting analysis using mouse anti-RhoA (Santa Cruz 26C4) and goat anti-RhoC (Santa Cruz G-12) antibodies, respectively.

Statistical analysis

The data were expressed as mean ± SD. Comparisons for continuous data between relapse and remission in INS patients were made by paired t test. In animal studies, a two-tailed t test was used to evaluate statistically significant differences in biochemical data between two groups. One-way ANOVA test with post hoc analysis by LSD method was used to compare biochemical data of four groups. In cell study, a two-tailed t test was used to evaluate statistically significant differences between two groups. A P < 0.05 was selected as the statistically significant cut-off value.

Results

C5a was elevated in children during the active phase of idiopathic nephrotic syndrome which elicited minimal change disease in mice

To characterize the cytokine change in children with INS, a commercial antibody array was employed to profile the cytokines in paired sera collected from nephrotic children (Fig. 1a). Three of the four tested paired samples showed increased C5a levels in relapse compared to those during remission (Fig. 1b). Further validation of these array data was performed for several cytokines of interest, including C5a, by ELISA analysis. Twenty-two INS children were recruited (Table 1) and the result of ELISA assay confirmed that serum C5a significantly increased during relapse (Fig. 1c). To confirm that C5a in patient sera is indeed pathogenic, we set up in an in vitro model to test whether C5a receptor antagonist treatment could attenuate the protein losing phenotype. Because our subsequent in vitro studies indicated endothelial cells, but not glomerular podocytes, possessed C5a receptor (Fig. 4a), we chose HUVECs to test the effect of patient sera in the presence of C5a receptor antagonist. While patient sera prepared from the relapse phase elicited more protein permeability than the respective patient sera collected during remission, both samples showed diminished effects upon treatment at 56 °C for 30 min (Fig. 1d), a standard procedure to inactivate serum complement components. The addition of C5a receptor antagonist W-54011 decreased the effect of sera from patients during the active stage which indicates a specific pathogenic role of C5a in the manifestation of proteinuria.

Fig. 1.

C5a increased in children with idiopathic nephrotic syndrome when the disease flared up. a A representative cytokine array data of the paired sera samples from a patient during relapse and remission was presented. Marked area denotes the duplicated spots where C5a antibody was dotted. b The pixel density of each duplicated C5a spot was measured and the average calculated. The ratio of the C5a spots intensity for paired samples during relapse and remission from four patients were presented. c Serum C5a changes from a cohort of 22 children with idiopathic nephrotic syndrome during relapse and remission were determined by ELISA. The result shows that C5a level significantly increased in this population when nephrotic syndrome was manifested (C5a = 44.30 ± 32.95 ng/ml during relapse vs. 22.31 ± 14.62 ng/ml during remission; mean ± SD, N = 22. **P = 0.005). d Sera were collected from 5 children afflicted with nephrotic syndrome whose serum C5a values were assessed to be the highest among the cohort as discussed in (c) during the relapse stage. Then undiluted sera (filled square, relapse; open square, remission) from these five children were each incubated with HUVECs for 60 min to perform the permeability assays as described in the “Materials and methods” section. In a tested group of Transwells, the assays were performed in the presence of C5a receptor antagonist W-54011 (10 nM). Serum samples which had been heated at 56 °C for 30 min served as negative controls. The data shown represent mean ± SD. *P < 0.05

Table 1.

The clinical characteristics during relapse and remission of 22 children diagnosed as idiopathic nephrotic syndrome in 2008–2012

	Relapse	Remission	P
Onset age (years)	5.87 ± 4.81
WBC (K/μL)	12.63 ± 4.67	11.67 ± 6.50	0.552
Hb (g/dL)	14.05 ± 3.32	13.92 ± 1.12	0.883
Platelet (K/μL)	398.90 ± 116.94	345.7 ± 82.46	0.116
BUN (mg/dL)	12.92 ± 2.70	12.36 ± 3.96	0.769
Creatinine (mg/dL)	0.44 ± 0.09	0.47 ± 0.08	0.299
Albumin (g/dL)	2.18 ± 0.59	3.70 ± 0.50	0.000
TG (mg/dL)	257.50 ± 259.95	299.83 ± 74.95	0.760
Cholesterol (mg/dL)	443.33 ± 157.59	296.83 ± 74.95	0.054
Urine protein (mg/dL)	557.14 ± 108.94	24.29 ± 79.68	0.000

All the values are expressed as mean ± SD

Fig. 4.

C5a-induced permeability change towards protein in kidney endothelial cells. Kidney endothelial cells (KECs) and glomerular podocytes were isolated by antibody-coated magnetic beads form mouse kidney tissue as described in the “Materials and methods” section. a The isolated KECs (E) and podocytes (P) were further processed by semi-quantitative RT-PCR (mRNA) and Western blotting analysis (protein) using the indicated markers to confirm the purity of each cellular population. b The homogeneity of the KECs was further verified by immunostaining with endothelial cell-specific surface marker CD31 followed by flow cytometry (empty shadow as a negative control in which isotype specific control antibody was added). The percentage of endothelial cells was quantified to be 96.56 % positive for CD31. c Phase contrast image of KECs with ×200 magnifications. Scale bars 10 μm. d KECs were cultured on Transwell plates and treated with the indicated dosages of C5a for 60 min before permeability assays were performed as described in “Materials and methods” section. The data shown represent mean ± SD (N = 5). *P < 0.05; **P < 0.001

To further examine whether the elevated C5a found when the disease was active contributed to the proteinuria in vivo, we injected mice with recombinant C5a parentally. This treatment resulted in persisted proteinuria and hypoalbuminemia followed by hyperlipidemia, which is a characteristic biochemical feature of nephrotic syndrome, but is missing in other nephrotic syndrome animal models, such as puromycin aminonucleoside nephrosis in rats (Fig. 2) [23, 24]. Histopathological examination of kidneys from C5a-injected mice did not reveal any significant change by hematoxylin and eosin (H&E) stain (Fig. 3a), periodic acid-Schiff (PAS) (Fig. 3b), and Roussel-Silver staining (Fig. 3c), compatible with minimal change disease as the predominant type of manifestation in childhood INS. Furthermore, immunofluorescence studies using IgG, IgA, C1q, C3 and C4 specific antibodies did not disclose any immunocomplex deposition in the kidneys (unpublished observations) also congruent with the pattern of minimal change nephrotic syndrome. Instead, some of the normally well-interdigitated podocytes appeared effaced under electron microscopic examination (Fig. 3d). These results indicate that the increased C5a observed in the children with INS plays a significant pathophysiological role.

Fig. 2.

C5a elicited nephrotic syndrome like manifestation in mice. ICR mice were injected with 200 ng recombinant C5a (10 μg/kg) every 3 days through tail vein. Urinary albumin/creatinine (A/C) ratio (a), serum albumin (b), triglyceride (c), and total cholesterol (d) levels were monitored every other day in the experimental group (filled square) and control mice which were injected with normal saline (open square). N = 5 for each group. The data shown represent mean ± SD. *P < 0.05; **P < 0.001

Fig. 3.

The pathological analysis of kidney tissues from C5a-treated mice manifested a pattern compatible with minimal change disease. ICR mice were treated with 200 ng recombinant C5a (10 μg/kg) every 3 days. Kidney tissues from mice killed on day 0, 6, 15 and 21 were dissected and processed for H&E (a), PAS (b), Roussel-Silver staining (c), and electron microscopic examination (d) in which mild focal effacement of some podocytes was noted (asterisk sign). Scale bars 10 μm for panels a, b, and c; scale bars 0.5 μm for panel d

C5a receptor was detected in the endothelial cells, but not in glomerular podocytes, isolated from mouse renal tissues

To investigate the potential cellular targets leading to the deteriorating effect of C5a on protein sieving function observed in both the human subjects and experimental mice (Figs. 1, 2), we prepared kidney endothelial cells (KECs) and glomerular podocytes from dissected mouse renal cortex tissues using magnetic beads coated with CD31 and nephrin-specific antibodies, respectively. The purity of endothelial population was confirmed by positive platelet endothelial cell adhesion molecule (PECAM-1/CD31) expression and negative evidence of podocyte markers, while the purity of podocytes was examined by the expression of both nephrin and podocin, but not CD31 (Fig. 4a). The expression of C5a receptor was confirmed in the CD31 positive KECs (Fig. 4a). By contrast, podocytes apparently lacked the expression of C5a receptor (Fig. 4a) which precluded the possibility of glomerular podocytes as the primary target of C5a. The purity of the isolated endothelial population was additionally confirmed by flow cytometry using endothelial specific marker CD31 (Fig. 4b) and the cobblestone appearance when cell culture condition reached confluence (Fig. 4c). Furthermore, a homogenous VE-cadherin staining pattern also demonstrated the purity of the isolated KECs (Fig. 5b). The isolated KECs were plated to confluency and HRP was applied as a surrogate to assess the cross-monolayer permeability toward protein after C5a application. A dose-dependent effect of C5a on increasing the permeability of KECs to HRP (Fig. 4d) was observed which corroborated our earlier finding that C5a could compromise the glomerular barrier function towards proteins. All these results taken together indicate that C5a likely elicits proteinuria through endothelia, but not podocytes in the renal glomeruli.

Fig. 5.

C5a-induced disassembly and redistribution of adherens junctional proteins. Mouse KECs were grown to confluency. The culture medium was replaced with M199 medium for serum fasting overnight before the indicated concentration of C5a (ng/ml) was applied to the cells for 60 min. After phase contrast images of KECs were acquired with CCD camera (a), the KECs were processed for immunofluorescence using mouse anti-VE cadherin (b) and rabbit anti-β-catenin (c) antibodies. The actin cytoskeleton organization was revealed by TRITC-rhodamine labeled phalloidin to stain for polymerized actin (d). Scale bars 10 μm

The C5a-dependent increase of protein permeability was accompanied with breakdown of adherens junction, increase of actin stress fiber formation, and activation of Rho kinase pathway in kidney endothelial cells

After serum fasting overnight, C5a treatment acutely induced drastic morphological change in mouse KECs. While control cells and KECs maintained in 1 ng/ml C5a maintained well intercellular contacts, cells treated with 10 ng/ml C5a became elongated and gross gaps were noticed between cells (Fig. 5a). KECs treated with 50 ng/ml C5a displayed pronounced fusiform phenotype with wide separations between the retracted cells. In certain 50 ng/ml C5a-treated KECs, there were visible paralleled fibrous structures running along the long axis of the elongated cells which were reminiscent of stress fibers. These morphological changes and stress fiber like structures were even more obvious when cells were grown at lower density which could be inhibited by ROCK inhibitor Y27632 (Fig. 6A). This phenotypic change of C5a-treated KECs implied a defective cell–cell adhesion. Indeed, immunofluorescence study disclosed diminished VE-cadherin and β-catenin staining from the cell–cell contacts (Fig. 5b, c). In addition, at higher C5a concentration, there was also an unusual accumulation of VE-cadherin signals inside a large cytosolic punctate (Fig. 5b).

Fig. 6.

C5a-induced cellular contraction, myosin contraction and actin stress fiber formation of KECs in a ROCK dependent way. a KECs were serum fasted overnight in M199 medium before C5a (50 ng/ml) was applied to the cells for 60 min with (right) or without (left) 10 μM Y27632 pretreatment for 30 min before phase contrast images were taken. b The culture condition was the same as in (a) except the cells were maintained at a confluent density before staining with non-muscle myosin II heavy chain B antibody (MYHb, red) and FITC-labeled phalloidin (green). c C5a effect on the actin cytoskeleton in vivo was shown. The intensive phalloidin staining in microvilli of the surrounding renal tubules (arrowheads) serves as a reference point to compare the intensities of phalloidin staining in the glomeruli (arrows) on fresh frozen renal sections from control or C5a (50 ng/ml) treated mice. d Confocal sectioning result following immunofluorescence study using VE-cadherin labeling on fresh frozen renal sections from control or C5a (50 ng/ml) treated mice was shown. Scale bars 10 μm

Phalloidin staining revealed not only an overall increase in the intensity of polymerized actin, but also thick paralleled actin stress fibers upon acute C5a application (Fig. 5d). This increased actin polymerization was accompanied by an increase of myosin staining at both the actin stress bundles and cell–cell contact sites, which could be eliminated dramatically by Y27632 (Fig. 6b). To further examine the C5a effect on cellular junction and cytoskeleton components in vivo, we prepared fresh frozen renal sections from mice 4 days after C5a injection when the proteinuria was first evident (Fig. 2a). These specimens were processed for confocal microscopic examination after phalloidin labeling and immunofluorescence study against VE-cadherin. Although the phalloidin staining intensity was weaker in the glomeruli of control mice compared to that in the microvilli, an actin-rich organelle, of the surrounding renal tubules, the phalloidin staining in the glomeruli of C5a-treated mice was comparable to that in microvilli of renal tubules (Fig. 6c). The fluorescent signals in the glomeruli of C5a-treated mice also appeared thicker than that in control mice (Fig. 6c). This result is compatible with the increased stress fiber formation we observed for purified KECs after C5a treatment (Figs. 5d, 6b). Furthermore, VE-cadherin staining was more fragmented, discontinuous, and diffusely distributed in the glomeruli of C5a-treated mouse while this junctional adhesive molecule looked tightly circumscribed along the capillary lumen in the glomeruli of control mouse (Fig. 6d). To better appreciate the geometrical distribution of VE-cadherin associated junctional complex in renal tissues, we reconstructed the three-dimensional (3D) images using a serial of confocal z-sections. The result showed that VE-cadherin staining was sharp and more continuously distributed inside the glomerular tufts of control mouse (supplementary video 1) in contrast to a more fragmented, discontinuous, and diffuse staining pattern in the glomeruli of C5a-treated mouse (supplementary video 2). This 3D reconstruction imaging study clearly illustrated the detrimental effect elicited by C5a on the integrity of endothelial cellular junction.

The increase of actin stress fiber formation and myosin staining at the contractile actins implied activation of Rho/ROCK/myosin light chain (MLC) pathway [25–27]. We then explored the biochemical evidence for the activation of Rho, ROCK and MLC in KECs after C5a treatment. The results demonstrated an activation of RhoA as well as RhoC (Fig. 7a). In addition, C5a-activated ROCK1 as demonstrated by an increase of phosphorylation at serine 1333 of ROCK1 (Fig. 7b). Furthermore, C5a also caused an increase in myosin regulatory light chain phosphorylation which could be inhibited by ROCK inhibitor Y27632, but not by pertussis toxin (Fig. 7c). Cell lysates prepared from isolated glomeruli also demonstrated that RhoA was activated after C5a treatment (Fig. 7d). These data suggest that C5a induces an increase of actomyosin contractility through the activation of Rho/ROCK signaling in KECs.

Fig. 7.

C5a-induced RhoA/ROCK1-dependent myosin light chain activation which led to an increase in the permeability of KECs. KECs were serum fasted in M199 medium for 3 h. Cells were further incubated for 60 min in the presence of the indicated concentration of recombinant C5a before cell lysates were harvested for Rho activation assay (a), Western blotting analysis using the antibodies recognizing total and phospho-specific ROCK1 (b), and total myosin light chain (MLC) and phosphoserine-specific MLC at the amino acid position 19 (c). Y, 10 μM Y27632; PTX, 100 ng/ml pertussis toxin pre-treated for 30 min. (d) Renal glomeruli were isolated from mice treated with 50 ng/ml recombinant C5a as the procedure described in the “Materials and methods” when the proteinuria was first evident. Cell lysates were prepared in NP-40 buffer for RhoA activation assay. e, f KECs were cultured on Transwell plates and pretreated with the indicated inhibitors for 30 min. Then, 50 ng/ml recombinant C5a was applied into the transwell plates for an additional 60 min before permeability assays were performed. The data shown represent mean ± SD (N = 4). *P < 0.05; **P < 0.001

To examine whether this biochemical pathway is responsible for the protein losing phenotype observed in C5a-treated mice, C5a-induced permeability change was assessed in purified KECs pre-treated with various inhibitors. Y27632 strongly inhibits C5a-induced permeability change to HRP in KECs with Rho inhibitor TAT-C3 and myosin ATPase inhibitor Blebbistatin similarly demonstrating an interfering effect (Fig. 7e). By contrast, pertussis toxin and vascular endothelial growth factor receptor tyrosine kinase inhibitor SU-5416 lacked any interfering effect against C5a in KECs (Fig. 7e). The results showed that Rho/ROCK/Myosin signaling mediates the C5a-induced permeability change in KECs. While Y27632 inhibited C5a action on the endothelial permeability in a dose-dependent manner, the application of ROCK inhibitor per se elicited negligible effect on the integrity of KECs (Fig. 7f).

ROCK inhibition alleviated the protein losing nephropathy by C5a administration in mice

To explore the potential therapeutic benefit of ROCK inhibition in reversing the urinary protein loss induced by C5a, we applied again the murine model as described previously. Y27632 were administrated once daily since day 6 after C5a injection and throughout the study period. The mice apparently tolerated Y27632 well, and more importantly the urinary albumin/creatinine ratio abated significantly and serum albumin increased in the group treated with Y27632 compared to those with only C5a treatment (Fig. 8).

Fig. 8.

Y27632 inhibited the C5a-elicited proteinuria in a murine model. ICR mice were injected with 200 ng recombinant C5a (10 μg/kg) every 3 days through tail vein for 21 days, and Y27632 were administrated 0.6 mg (30 mg/kg) orally once daily since day 6. Urinary albumin/creatinine ratio (a), serum albumin (b), triglyceride (c), and total cholesterol (d) levels were monitored every 3 days. The data shown represent mean ± SD for each treatment group (N = 5). *P < 0.05

Discussion

Complement component C5 is cleaved by C5-convertase into C5a and C5b. While C5b participates in the formation of membrane attack complex (MAC) to facilitate target cell lysis, C5a is actively linked to anaphylactoid and inflammatory responses. C5a stimulates degranulation of histamine and TNFα in mast cells and increases surface adhesion molecule expression of endothelial cells to help recruitment of phagocytic cells to the site of infection or inflammation. C5a is also implicated in increasing vascular permeability under a number of pathological situations such as asthma and allograft rejection [20, 28, 29]. C5a activation has been demonstrated to be crucial in experimental sepsis to increase albumin leakage into extravascular space in lung [30], but the molecular mechanism has never been studied in detail. Although the C5b-9 containing MAC has been implicated in the pathogenesis of both rat experimental and adult nephrotic syndromes [31, 32], the importance of C5a in nephrosis has yet to be addressed. The result of using C5a receptor antagonist W-54011 to attenuate the protein permeability change (Fig. 1d) distinguishes a pathogenic role of C5a from the other possible active components, such as C5b-9 MAC, in the sera of child nephrotic patients.

Nephrotic syndrome in children is mainly caused by MCD with conventional wisdom reporting minimal renal abnormality such as effacement of glomerular podocytes. The importance of podocytes as the main gatekeeper in maintaining protein sieving function is best confirmed by elucidation of a pivotal role of podocyte-associated proteins in several congenital or inherited nephrotic syndrome [33–37] with nephrin mutation as the prototype among these molecular anomalies [38, 39]. Although these earlier studies defined the glomerular filtrating apparatus composed of podocin-nephrin as the critical machinery in a number of genetic disorders manifesting heavy proteinuria, they were all characterized by focusing on congenital, inherited, and profound disorders involving other organs besides renal diseases. Furthermore, the renal pathologies depicted in these earlier studies were typically more severe than MCD. Since these landmark reports, many of the followed up studies based on these earlier conclusion also emphasized studies in familial or steroid resistant form of nephrotic syndrome [40–42]. However, most childhood nephrotic syndrome cases are without familial history, sporadic, steroid sensitive, and presumably with MCD as the principal pathological variant. Although there are convincing evidences to validate the central importance of podocytes in maintaining normal filtering function of kidneys, it is still unclear why many acquired and congenital nephropathies manifest heavy protein loss as a hallmark of their clinical presentation yet lacking evidence of primary defect in podocytes. For example, there is no known antibody or inhibitory factor causing the defective function of podocytes in INS. Therefore, it is imperative to search for other pathogenesis to explain the majority of childhood nephrotic patients.

The glomerular endothelial cells had been denied as a major contributor to the barrier function toward macromolecular filtrate because of the presence of large fenestrations which lack bridging diaphragm. However, the glomerular endothelial role as a primary target in protein losing nephropathies has recently been re-considered. A complex glycocalyx formed of proteoglycans and glycoproteins has been identified over the fenestrated area of glomerular endothelial cells and ultimately been demonstrated to serve a function in restricting transendothelial albumin passage [43]. Genetic ablation of endothelial nitric oxide synthetase in diabetic mice results in heavy proteinuria and structural changes of podocytes similar to those observed in MCD [44]. Indeed, conditioned medium from glomerular endothelial cells exposed to shearing stress could impair barrier function of podocytes in vitro [45]. Based on emerging evidences that corroborate a dynamic signaling network across the renal filtration barrier, an endothelial–podocyte crosstalk is hypothesized to explain the role of endothelial dysfunction in diabetic nephropathy [46], which displays protein losing as the mainstay of disease manifestation. Compatible with the idea that endothelia may be an essential modulator of trans-glomerular protein transport, a study on idiopathic nephrotic syndrome patients demonstrates the presence of global endothelial defect in maintaining vascular protein permeability [47]. Indeed, our findings that acute application of C5a induced minimal change nephrosis and the demonstration of C5a receptor expression in glomerular endothelia, but not in podocytes, indicate strongly that the C5a may induce protein loss by acting primarily on renal endothelial cells.

C5aRs are widely expressed in inflammatory [48] and non-inflammatory cells [49–51] including murine dermal, pulmonary, and human umbilical vascular endothelial cells [52–54]. The effect of C5a on eliciting proinflammatory responses such as a strong up-regulation of IL-8, IL-1beta, and RANTES in HUVECs could be observed using physiological concentrations of C5a [54]. That concentration is equivalent to what we used to treat the isolated KECs for protein permeability assay which is at nanomolar range (Figs. 4d, 7) and also compatible with the serum concentration of C5a found among the nephrotic children recruited in our current study (mean C5a concentration = 44.3 ng/ml during the relapse phase, Fig. 1c). This not only justifies the C5a dosage employed in our in vivo (animals) and in vitro (cells) studies, but also signifies that the increased C5a during complement activation when disease flares up could substantially bring forth a detrimental effect on human glomerular endothelial cells as we observed in experimental animals.

Rho family small GTPases impinge their effect on cell–cell adhesion through modulating the recycling activity of the cadherin as well as the cytoskeleton organization associated with the cadherin complex. RhoA GTPase and its effector ROCK have complicated effect on cell adhesion. Endothelial permeability is compromised by dominant-negative ROCK constructs and ROCK inhibitor [55], while ROCK inhibitor Y27632 increases epithelial permeability [56], which indicates ROCK might operate differently on regulating cell adhesion between endothelial and epithelial cells. RhoA sub-family GTPase plays a dual role in governing the integrity of adherens junctions (AJ) through diaphanous-related formins Dia proteins, which stabilize cadherin and α-catenin at AJ, and ROCK, which destabilizes AJ by generating myosin dependent contractile force on actin cytoskeleton. The result that both RhoC and RhoA were activated at the C5a dosage we applied in glomerular endothelial cells might explain why the overall effect of C5a on junctional permeability was compromised instead of maintained as RhoC has been shown to have a stronger affinity toward ROCK than RhoA and more disruptive to adherens junctional integrity [57].

Isolated primary KECs further clarify the signaling pathway of C5a action is through Rho/ROCK/myosin contraction as the inhibitors of Rho/ROCK/myosin, C3 toxin, Y27632, and blebbistatin all alleviated the C5a increased permeability of KECs. Intriguingly, although C5a receptor has previously been reported to mediate its several inflammatory functions through the Gi2/3 subtype guanine nucleotide-binding proteins [58–60], the effect of C5a on permeability change of KECs is not inhibited by pertussis toxin in our study. The apparent reason for such a lack of inhibition by pertussis toxin is not clear at this moment, but it implies that there may be a cryptic Gα subunit subtype which mediates the action of C5a. Although Gq16 has been shown to transduce the phosphoinositol release effect of C5a receptor activation [61], that study employed over-expression of both C5aR and Gq16 (human orthologue of mouse Gq15) and no evidence exists so far to demonstrate a physical interaction between these two proteins under more physiological situation.

An important finding of our study raises the possibility of using ROCK inhibitor Y27632 in mitigating the protein losing phenotypes of C5a or other pathogenic factors which act through the same pathway in eliciting protein losing disorders such as SLE. There has been a great enthusiasm towards C5a/C5a receptor blockade since the documented beneficial effects of either C5a antibody or C5a receptor antagonist in treating experimental sepsis. As more efficient Y27632 derivatives are clinically or experimentally tested in other medical situation such as ocular and renal vascular hypertension [62, 63] and these drugs seem to be well tolerated, our findings not only shed lights on a new mechanistic understanding of the basic pathogenesis, but also a novel direction of therapeutical approach for childhood nephrotic syndrome, or similar complement mediated protein losing vasculopathies.

Electronic supplementary material

Below is the link to the electronic supplementary material.