Disrupted apolipoprotein L1-miR193a axis dedifferentiates podocytes through autophagy blockade in an APOL1 risk milieu

Abstract

We hypothesized that a functional apolipoprotein LI (APOL1)-miR193a axis (inverse relationship) preserves, but disruption alters, the podocyte molecular phenotype through the modulation of autophagy flux. Podocyte-expressing APOL1G0 (G0-podocytes) showed downregulation but podocyte-expressing APOL1G1 (G1-podocytes) and APOL1G2 (G2-podocytes) displayed enhanced miR193a expression. G0-, G1-, and G2-podocytes showed enhanced expression of light chain (LC) 3-II and beclin-1, but a disparate expression of p62 (low in wild-type but high in risk alleles). G0-podocytes showed enhanced, whereas G1- and G2-podocytes displayed decreased, phosphorylation of Unc-51-like autophagy-activating kinase (ULK)1 and class III phosphatidylinositol 3-kinase (PI3KC3). Podocytes overexpressing miR193a (miR193a-podocytes), G1, and G2 showed decreased transcription of PIK3R3 (PI3KC3′s regulatory unit). Since 3-methyladenine (3-MA) enhanced miR193a expression but inhibited PIK3R3 transcription, it appears that 3-MA inhibits autophagy and induces podocyte dedifferentiation via miR193a generation. miR193a-, G1-, and G2-podocytes also showed decreased phosphorylation of mammalian target of rapamycin (mTOR) that could repress lysosome reformation. G1- and G2-podocytes showed enhanced expression of run domain beclin-1-interacting and cysteine-rich domain-containing protein (Rubicon); however, its silencing prevented their dedifferentiation. Docking, protein-protein interaction, and immunoprecipitation studies with antiautophagy-related gene (ATG)14L, anti-UV radiation resistance‐associated gene (UVRAG), or Rubicon antibodies suggested the formation of ATG14L complex I and UVRAG complex II in G0-podocytes and the formation of Rubicon complex III in G1- and G2-podocytes. These findings suggest that the APOL1 risk alleles favor podocyte dedifferentiation through blockade of multiple autophagy pathways.

INTRODUCTION

Autophagy is a lysosomal-dependent process that is associated with the sequestration and degradation of intracellular constituents (7). This action selectively degrades unwanted proteins/damaged organelles/intracellular pathogens or nonselectively promotes the degradation of cytosolic proteins to provide nutrition under starved conditions. It involves more than 30 genes (autophagy-related genes, ATGs), including class III phosphatidylinositol (PI) 3-kinase [PI3KC3, vacuolar protein sorting (Vps) 34/p150], LC3 (ATG8), and beclin-1 (Vps30, ATG6). PI3KC3 (hVps34/p150) promotes the conversion of PI to phosphatidyinositol 3-phosphate (PI3P) via subunit binding [regulatory (PIK3R3/PIK3R4/p150) to the catalytic (hVps34 and beclin-1)], resulting in the initiation of autophagy (20, 39, 44). Phosphatidylinositol 3-kinase (PI3K) comprises three configurations, classes 1, 2, and 3 (9, 10). The binding of the regulatory subunit 58 to the catalytic subunit p110 of the class I PI3K (PI3KC1), promotes the formation of phosphatidylinositol 3,4,5-triphosphate, thus activating the mammalian target of rapamycin (mTOR) pathway, resulting in repression of Unc-51 like autophagy activating kinase (ULK1) as well as autophagy (13, 36). Conversely, inhibition of mTOR recruits ULK1 to PI3KC3 complex (beclin-1-PI3KC3) along with either autophagy-related gene (ATG)14L (complex I) or the UV radiation resistance-associated gene product (UVRAG, complex II), resulting in the generation of PI3P and nucleation (ATG14L complex I)/maturation (UVRAG complex II) of autophagosomes.

The beclin-1 protein contains three identifiable domains: a short Bcl-2 homology domain (BH)3 motif, a coiled-coil domain (CCD), and a COOH terminus containing an evolutionarily conserved domain (ECD) (14). Bcl-2 family proteins bind to the BH3 domain, and the CCD is involved in homodimerization with Bcl2 and heterooligomerization with proteins such as UVRAG (29) and ATG14L (40). Both ATG14L and UVRAG regulate the formation of PI3KC3 complexes. The homodimer beclin-1 is metastable and inactive and may serve as a reservoir for beclin-1 heterooligomers. In contrast, the binding of ATG14L and UVRAG to PI3KC3 complexes are mutually exclusive and localize to different subcellular organelles carrying out distinct functions (18); nonetheless, the beclin-1-UVRAG interaction is stronger than the beclin-1-ATG14L interaction (28). Normally, ATG14L (localized in the endoplasmic reticulum) recruits PI3KC3 complex to nascent autophagosomes to phosphorylate PI (PI3P); additionally, ATG14L binds to the syntaxin 17-synaptosomal-associated protein (SNAP29) complex and facilitates fusion of autophagosomes to endosomes (8). In association with the core complex (beclin-1-PI3KC3), UVRAG (UVRAG complex II) facilitates maturation of autophagosomes (18, 30).

An autophagy inhibitor, Rubicon (run domain beclin-1-interacting and cysteine-rich domain-containing protein) binds UVRAG when UVRAG is in a phosphorylated state and thus inhibits autophagosome maturation (21). However, this effect of Rubicon usually occurs in the absence of beclin-1 overexpression; thus, binding of Rubicon with beclin-1 seems to act as a brake on autophagy inhibition (52). Rubicon is localized in endosomes and lysosomes and inhibits the autophagosome-lysosome fusion step. Podocyte-expressing apolipoprotein L1 (APOL1) risk alleles (APOL1G1 and APOL1G2) were recently reported to partially block autophagy flux in the formation of autolysosomes (3). However, the involved mechanism was not investigated. We hypothesized that APOL1 risk alleles would downregulate autophagy flux in podocytes by enhancing the expression of Rubicon. Interventions that potentially downregulate Rubicon could be used as a therapeutic target for promoting autophagy in the setting of APOL1 risk.

We recently reported that expression of APOL1 risk alleles by podocytes was associated with enhanced expression of microRNA 193a (miR193a) (32). Of note, miR193a downregulates the transcription of mTOR through the binding of miR193a-5p to the mTOR mRNA (50); additionally, miR193a binds to the regulatory unit of PI3KC3 (P3KR3), resulting in its downregulation. Therefore, we investigated whether results in deficient mTOR/PIK3R3 transcription in the APOL1 risk milieu decreased the formation of ATG14L and UVRAG complexes.

In in vitro studies, beclin-1 binds equally well to cardiolipin and phosphatidic acid (PA) (51). APOL1 also binds strongly with cardiolipin and PA (11); the latter directly modulates mTOR’s downstream signaling (27). It is likely that the affinity of APOL1 to bind with PA would decrease the availability of PA to activate mTOR’s downstream signaling in the form of repression of ULK1 and retardation of ATG14L/UVRAG complex formation (11).

Adult podocytes possess a high rate of autophagy flux and are especially susceptible to injury caused by an imbalance between protein load and rate of autophagy flux (16). Podocyte functional integrity, including maintenance of slit diaphragm proteins, the organization of actin cytoskeleton, and protein clearance (to prevent filtration barrier clogging) is dependent on a high capacity for endosomal trafficking (1, 35, 38). The role of PI3KC3 (Vps34) as a major regulator of endocytic pathways in podocytes has been well established (2, 4), thus contributing to the disparate onset of kidney disease in mice lacking PI3KC3 (Vps34) and ATG5 knockout mice (16). The role of APOL1 in the endocytic pathway in general, and the autophagy pathway in particular, have been reported in multiple experimental models (23, 24).

In recent reports, APOL1 has been demonstrated to also influence differentiation with a role in the preservation of fully differentiated podocyte molecular phenotype (25, 26, 32). This effect of APOL1 in podocytes was dependent on a bifunctional APOL1-miR193a axis. The contribution of autophagy to the preservation of the podocyte molecular phenotype was reported earlier (12); however, the role of APOL1 was not investigated in those studies. Since APOL1 is an inducer of autophagy in several cell types including podocytes (5, 11, 19, 23, 33, 45), we hypothesized that the reduction of autophagy in the APOL1 risk allele state would lead to a potential loss of the fully differentiated and functional podocyte molecular phenotype. In that scenario, silencing of the molecule(s) causing relative attenuation of autophagy would prevent dedifferentiation of podocytes in the APOL1 risk milieu. We proposed that blockade of autophagy could be the consequence of the disruption of APOL1-miR193a axis in the APOL1 risk state.

In the present study, we have examined the effect of the constitutive ectopic overexpression of nonrisk and risk alleles on autophagy-associated signaling molecules in human podocytes. We evaluated the effect of disrupted APOL1-miR193a axis in autophagy signaling and associated effects on the molecular phenotype of podocytes in APOL1 nonrisk and risk milieus.

MATERIALS AND METHODS

Human Podocyte Embryonic Kidney Cells

Human podocytes (PDs) were conditionally immortalized by introducing temperature-sensitive SV40-T antigen by transfection (34). These cells proliferate at a permissive temperature (33°C) and enter growth arrest after transferring to a nonpermissive temperature (37°C). The growth medium contains RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 1× Pen-Strep, 1 mM l-glutamine, and 1× insulin-transferrin-selenium (ITS, Invitrogen). Undifferentiated (UND) podocytes were seeded on collagen-coated plates and differentiated by preincubation in normal RPMI for 10 days at 37°C (differentiated podocytes, PDs). Before the experimental procedure(s), podocytes were starved overnight to stimulate autophagy. Genotyping of these podocytes revealed expression of APOL1^G0/G0. Differentiated podocytes (DPDs) displayed a robust expression of APOL1.

Human embryonic kidney cells (HEKs) were purchased from ATCC (CRL-1573TM), and cultured in DMEM supplemented with 10% fetal calf serum (FCS) and 25 mM HEPES.

Generation of Stable Cell Lines Expressing APOL1G0, APOL1G1, APOL1G2, and Vector

We selected control podocytes, which display a minimal expression of endogenous APOL1 even after differentiation (to display the maximal effect of ectopically expressed APOL1 G0 and its variants). Stable cell lines ectopically expressing vector, APOL1G0, APOL1G1, and APOL1G2 were generated by retroviral infection as described previously (31). These undifferentiated podocyte-expressing vector and ectopic APOL1G0/APOL1G1/APOL1G2 were seeded on collagen-coated plates and differentiated by incubation in normal RPMI (containing 11 mM glucose) for 10 days at 37°C. We used differentiated vector (V)-, APOL1G0 (G0)-, APOL1G1 (G1)-, and APOL1G2 (G2)-podocytes in the majority of experiments, unless specified otherwise.

Transfection of miR193a Expression Plasmid

miR193a expression plasmid (25 nM; cat. no. SC400232; Origene), and empty vector (25 nM; pCMV-MIR; Origene) were transfected in cells using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific) according to the manufacturer’s protocol. All miRNA products were dissolved in nuclease-free water. Briefly, DPDs were transfected at 70–80% confluence in six-well plates. The Lipofectamine (7.5 µl) and plasmid DNA were diluted in opti-MEM media (125 and 250 µl, Thermo Fisher Scientific) followed by addition of P3000 enhancer reagent (10 µl) to diluted DNA. Diluted DNA (125 µl) was added to diluted Lipofectamine 3000 (125 µl) in the ratio of 1:1 (vol/vol) and incubated for 10 min at room temperature (25°C). After incubation, the DNA-lipid complex was added to the cells and kept at 37°C in opti-MEM media for 48 h. Control and transfected cells were harvested for protein and RNA analyses.

Silencing of Rubicon

Differentiated podocytes were transfected with scrambled siRNA (control, Sc-37007) or Rubicon siRNA (Sc-78326, 20 nM; Santa Cruz Biotechnology) with Lipofectamine RNAiMAX transfection reagent according to the manufacturer’s protocol (Thermo Fisher). Briefly, differentiated podocytes were transfected when they were 60–80% confluent in 6 well plates. Lipofectamine reagent (9 µl) and siRNAs (10 µM, 2–3 µl) were diluted in opti-MEM media (150 µl) (Thermo Fisher Scientific). Diluted siRNA (150 µl) was added to diluted Lipofectamine reagent (150 µl) in the 1:1 ratio (v/v) and incubated for 5 min at room temperature (25°C). After incubation, the siRNA lipid-complex was added to cells and kept at 37°C in opti-MEM media for 48 hrs. The cells were harvested for protein and RNA analyses. Control and transfected cells were used under control and experimental conditions.

RNA Isolation and qPCR Studies

Total RNA was isolated from control and experimental differentiated podocytes with TRIzol reagent (Invitrogen). A 20-µl reaction mix was prepared containing iTaq Universal SYBR Green reaction mix (2×, 10 µl), iscript reverse transcriptase (0.25 µl), forward and reverse primers (2 µl), RNA (4 µl), and nuclease-free water (3.75 µl). Real-Time PCR was performed using the one-step iTaq Universal SYBR Green kit (Bio-Rad Laboratories) according to the manufacturer’s instructions and using specific primers obtained from Thermo Fisher Scientific. GAPDH forward (FW) 5′-CCC ATC ACC ATC TTC CAG GAG-′3, reverse (Rev) 5′-GTT GTC ATG GAT GAC CTT GGC-′3; PIK3R3 FW 5′-GAGAGGGGAATGAAAAGGAGA-3′, Rev 5′-TCATGAATCTCACCCAGACG-3′; Rubicon FW 5′-AACCTCACCCACCATCTTCTTAGCGT-3′ Rev 5′-CACAGAGTTAAGTGCATAATTGGCATAAAGG-3′.

The qPCR conditions were as follows: 50°C for 10 min at 95°C for 1 min, followed by 40 cycles of 95°C for 15 s at 60°C for 1 min. Quantitative PCR was performed using the Roche 480 Light Cycle system, and relative quantification of gene expression was calculated using the ΔΔC_T method. Data are expressed as relative mRNA expression in reference to the control, normalized to the quantity of RNA input by performing measurements to the endogenous reference gene GAPDH.

MicroRNA Assay

For miRNA quantification, the total RNA was isolated from control and experimental podocytes with a miRVana miRNA isolation kit, and 1 µg of RNA was reverse transcribed using miR193a and U6-small nuclear RNA (U6snuRNA)-specific RT primers to generate first-strand cDNA from mRNA using a TaqMan microRNA Reverse Transcription kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. For cDNA, a 15-µl PCR reaction was prepared containing 100 mM dNTP mix (0.15 µl), multiscribe RT enzyme 50 U/µl (1 µl), 10× RT buffer (1.5 µl), RNase inhibitor 20 U/µl (0.19 µl), nuclease-free water (4.16 µl), RNA (5 µl), and primers (3 µl). The PCR conditions followed were as follows: 16°C for 30 min, at 42°C for 30 min, 85°C for 5 min, and 4°C until stopped. Real-time PCR was performed using TaqMan-based PCR master mix and detection primers miR-193a and U6snuRNA (Thermo Fisher) in an ABI 7500 (Applied Biosystems). For real-time PCR a 10-µl reaction mix was prepared containing TaqMan PCR master mix II (5 µl), cDNA (2 µl), nuclease-free water (2 µl), and primer (1 µl). The qPCR conditions were as follows: 50°C for 2 min at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s at 60°C for 1 min. U6 was used as an internal control. Relative quantification of gene expression was calculated using the ΔΔC_T method, and the results were normalized to U6-snuRNA expression.

Western Blot Studies

Western blot studies were carried out as described previously (25, 26). Briefly, control and experimental cells were harvested, lysed in RIPA buffer containing 50 mM Tris-Cl (pH 7.5), 150 mM NaCl, 1mM EDTA, 1% NP-40, 0.25% deoxycholate, 0.1% SDS, 1× protease inhibitor cocktail (Calbiochem, Cocktail Set I), 1 mM PMSF, and 0.2 mM sodium orthovanadate. Protein concentration was determined using the Bio-Rad Protein Assay kit (Pierce, Rockford, IL). Total protein lysed extracts (30 μg/lane) were loaded onto a 10% polyacrylamide (PAGE) precast gels (Bio-Rad, Hercules, CA) and after transfer onto PVDF membranes were processed for immunostaining with primary antibodies against APOL1 (cat. no. 66124-I-IG, Proteintech), ATG14L (cat. no. Ab139727, Abcam), UVRAG (cat. no. 19571-I-AP, Proteintech), p62 (cat. no. SC-28359, Santa Cruz Biotechnology), LC3-II (cat. no. 2775S, Cell Signaling), PI3KC3 (cat. no. I3857S, Cell Signaling), phosphorylated (p-)ULK1 (cat. no. SC-390904, Santa Cruz), p-PI3KC3 (cat. no. 13857S, Cell Signaling), p-mTOR (cat. no. 2971S, Cell Signaling), CD2AP (cat. no. SC-25272; Santa Cruz Biotechnology), nephrin (cat. no. ABb235903, Abcam), WT1 (cat. no. AB15249, Abcam), podocalyxin (cat. no. PA-1-46170, Invitrogen), Rubicon (cat. no. 7151-S, Santa Cruz Biotechnology), podocin (cat. no. SC-22296, Santa Cruz Biotechnology), and beclin-1 (cat. no. 3738S, Cell Signaling), followed by treatment with horseradish peroxidase-labeled appropriate secondary antibodies. Characterization of antibodies is described in Supplemental Table S1 (see Supplemental Material: https://doi.org/10.6084/m9.figshare.8066420.v1). Equal protein loading and the protein transfers were confirmed by immunoblotting for determination of actin/GAPDH protein using a monoclonal β-actin/GAPDH antibody (Biotechnology SC-47724, Santa Cruz Biotechnology) on the same (stripped) Western blots. The blots were developed using a chemiluminescence detection kit (Pierce) and scanned on the Bio-Rad ChemiDoc MP imaging system using Image Laboratory software. The images captured were then converted to either a TIFF or JPEG format and were further subjected to Adobe Photoshop/Microsoft Power Point software. For quantification, we used either the Gel Doc Image laboratory software or ImageJ software, and further analyses were performed on a Microsoft Excel Sheet.

Immunofluorescence Detection of Vacuoles and Autophagosomes

Control and experimental podocytes were stained with acridine orange and fixed as described previously (48). For immunolabeling, cells were fixed and permeabilized with a buffer containing 0.02% Triton X-100 and 4% formaldehyde in PBS. Fixed cells were washed three times in PBS and blocked in 10% BSA for 60 min at 37°C. Subsequently, cells were labeled with an anti-LC3 antibody (cat. no. 2775S, Cell Signaling). DAPI was used for nuclear localization. Control and experimental cells were examined under a confocal microscope.

Immunoprecipitation Studies

Lysates from V- and G0-podocytes were first immunoprecipitated following the addition of 5 μg of anti-APOL1 (Proteintech), 5 μg of anti-ATG14L (Abcam), 5 μg of anti-Rubicon (Sana Cruz Biotechnology), or 5 μg of anti-UVRAG (Proteintech) antibodies. The immune complexes were then collected using 25 μl of protein A + G Sepharose beads (GE Health Care, Life Sciences) in RIPA buffer. Immunoprecipitation (IP) was carried out at 4°C for 4 h on a rotating platform. Following this, precipitated A/G proteins were pelleted down by centrifugation at 4,500 rpm for 10 min at 4°C. Next, the protein pellet was washed three times, each time with 1 ml of cold RIPA lysis buffer followed by centrifugation each time for 10 min at 2,500 rpm in a microfuge. After washings, beads were resuspended in 100 μl of lysis buffer to which SDS-PAGE sample buffer (50 μl) was added, and samples were boiled at 100°C, followed by SDS-PAGE and immunoblotted using specific antibodies as indicated.

Homology Modeling

We retrieved the sequences of APOL1 (Uniprot Id O14791), beclin-1 (Uniprot Id Q14457), Bcl-2 (Uniprot Id P10415), PI3KC3 (Uniprot Id Q8NEB9), ATG14L or Bakor (Uniprot Id Q6ZNE5), UVRAG (Uniprot Id Q9P2Y5), and Rubicon (Uniprot Id Q92622) from the Uniprot database. The sequence of APOL1 that is the G0 variant (398 aa residues) has different domains including an amino-terminal signal peptide (1–59 aa), pore-forming domain (60–237 aa), and membrane-addressing domain (238–304 aa) and the COOH-terminal region (305–398 aa). There are two APOL1 variants, i.e., APOL1G1 and APOL1G2 (associated with chronic kidney disease) which show mutations in the COOH-terminal region of APOL1. The APOL1G1 variant has two site-specific mutations, S342G and I384M, whereas, the APOL1G2 variant has two deletions, N388 and Y389. These APOL1 variants have significant functional roles in kidney-associated diseases, so we modeled all APOL1G0, APOL1G1, and APOL1G2 variants and other associated proteins in the autophagy-regulating complexes. The 3D structure models of beclin-1, Bcl-2, APOL1G0, APOL1G1, APOL1G2, PI3KC3, ATG14L, UVRAG, and Rubicon were modeled using an Iterative Threading ASSEmbly Refinement (ITasser) server (49). Models were generated based on templates identified by threading approach to maximize percentage identity, sequence coverage, and confidence.

Model Refinement

All of the structure models were refined using GalaxyRefine (37). The refinement process is based on repetitive relaxations by short molecular dynamics simulations for mild (0.6 ps) and aggressive (0.8 ps) relaxations, with 4 fs time step after structure perturbations. The refinement of models improved certain parameters, for example, an increase in Rama favored residues and decrease in poor rotamers.

Docking

Docking of APOL1G0 with other molecules in the ATG14L complex I such as beclin-1, PI3KC3, and ATG14L; APOL1G0 with other molecules in the UVRAG complex II such as beclin-1, PIK3C3, and UVRAG; APOL1G1/APOLG2 in the Rubicon complex III and IIIb such as PI3KC3 and UVRAG, using 3D structure models were performed using the GRAMM-X web interface (43), which uses the FFT (Fast Fourier Transformation) methodology by utilizing the smoothed potentials, refinement stage, and knowledge-based scoring.

Protein-Protein Interaction Analyses

We analyzed the protein-protein interaction properties of all complexes, including ATG14L complex I, UVRAG complex II, and Rubicon complex IIIa and IIIb, using PDBePISA (22) which is a calculation method based on chemical thermodynamics of macromolecular complexes.

Visualization

All structural models were visualized, and figures were generated using a PyMol PDB viewer (42).

Statistical Analyses

Statistical comparisons were performed with the program PRISM using the Mann–Whitney U-test for nonparametric data and the unpaired t-test for parametric data. For comparison in multiple groups, one-way analyses of variance were followed by the use of Tukey multiple comparison tests. A P value ≤ 0.05 was considered statistically significant.

RESULTS

Podocytes Expressing APOL1 Nonrisk and Risk Alleles Display Enhanced Accumulation of Autophagosomes

Overt expression of APOL1 nonrisk alleles has been shown to enhance autophagy in several cells (5, 11, 19, 23, 33, 45). Recently, overt expression of risk alleles in podocytes has been reported, with the accumulation of autophagosomes as a consequence of inhibition of fusion between autophagosomes and lysosomes (3).

To determine autophagy profiles in the cells used in our experimental protocols, V (Vector)-, G0-, G1-, and G2-podocytes were grown on coverslips and stained with acridine orange as described previously (48). Cells were examined under a confocal microscope. Representative fluoromicrographs are shown in Fig. 1A. Cumulative distribution data of vacuole accumulation (orange fluorescence) in control and experimental conditions (n = 6 replicates) are shown in a dot plot (Fig. 1B). G0-podocytes displayed a higher number of orange vacuoles compared with V-podocytes. However, G1- and G2-podocytes showed several-fold increase in accumulated vacuoles.

Fig. 1.

Autophagosome formation and autophagy markers in podocytes expressing apolipoprotein L1 (APOL1) nonrisk and risk alleles. A: vector- (V-), G0-, G1-, and G2-podocytes grown on coverslips were stained with acridine orange (n = 6). Cells were examined under a confocal microscope. Representative fluoromicrographs (autophagosomes displaying orange fluorescence) are shown. Scale bar, 50 µm. B: cumulative distribution data of vacuole accumulation (orange fluorescence) in control and experimental conditions (n = 6) are shown in a dot plot. *P < 0.05 vs. V. **P < 0.01 vs. V and G0. C: HEK cells grown on coverslips were transfected with V (VHEK), APOL1G0 (G0HEK), and APOL1G1 (G1HEK). After 48 h, cells labeled for light chain 3-II (LC3-II; green fluorescence) and nuclei were stained with DAPI (blue fluorescence); subsequently, cells were examined under a confocal microscope. Representative fluoromicrographs are displayed. Scale bar, 50 µm. D: podocytes stably overexpressing APOL1G0 (G0) and V were differentiated (n = 4–5). Protein blots from different cellular lysates of V and G0 were probed for beclin-1, p62, and APOL1; the same blots were reprobed for GAPDH. Gels from different cellular lysates are shown. i: Distribution densitometric data of beclin-1/GAPDH ratios from D are shown in a dot plot. *P < 0.05 vs.V. ii: Distribution densitometric data of p62/GAPDH ratios from D are shown in a dot plot. *P < 0.05 vs. V. iii: Distribution densitometric data of APOL1/GAPDH ratios from D are shown in a dot plot. *P < 0.05 vs. V. E: RNAs were extracted from different sets of V- and G0-podocytes (n = 5) and assayed for miR193a. Distribution data are shown in a dot plot. *P < 0.05 vs. V.

To clarify whether the vacuoles in G0- and G1-podocytes are indeed likely to be autophagosomes, we turned to transient transfection in HEK cells. HEKs grown on coverslips were transfected with either vector, APOL1G0, or APOL1G1. After 48 h, cells were labeled for LC3, and nuclei were stained with DAPI (blue fluorescence); subsequently, cells were examined under a confocal microscope. Representative fluoromicrographs are displayed in Fig. 1C. HEKs transfected with APOLG1 displayed a higher number of LC3 labeled vacuoles (green fluorescence) when compared with HEKs transfected with APOL1G0.

Overexpression of APOL1G0 Enhances Autophagy

APOL1 has been reported to enhance autophagy in several cell types including podocytes (5, 11, 19, 23, 33, 45). To determine the effect of overexpression of APOL1 on autophagy, podocytes stably overexpressing APOL1G0 and vector were differentiated (n = 4–5). Protein blots for V- and G0 podocytes from independent cellular lysates were probed for APOL1, beclin-1, and P62. The same blots were reprobed for GAPDH. Gels from independent cellular lysates are displayed in Fig. 1D. Raw data are shown in Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.8066600.v1. Densitometric distribution data are shown in dot plots (Fig. 1D, i–iii. G0-podocytes displayed enhanced (P < 0.05 compared with respective V-podocytes) expression of beclin-1 and APOL1 but downregulation (P < 0.05 compared with respective V-podocytes) of p62. These findings suggest stimulation of autophagy in podocytes overtly expressing APOL1G0.

APOL1G0 has been reported to downregulate the expression of miR193a in podocytes (32). To measure miR193a expression, RNAs were extracted from V- and G0-podocytes and assayed for miR193a (n = 5). Cumulative data are shown in a dot plot (Fig. 1E). G0-podocytes displayed a 50% decrease (P < 0.001) in miR193a expression compared with V-podocytes. These findings confirm that APOL1G0 expression maintains an inverse relationship with miR193a.

Role of PI3KC3 Complexes in APOL1-Induced Autophagy in Podocytes

Both ATG14 and UVRAG regulate the formation of PI3KC3 complexes, and the binding of ATG14L and UVRAG to PI3KC3 complexes is mutually exclusive (18). To determine the composition of these complexes, we utilized bioinformatics analyses, including protein-protein interaction and docking studies.

Protein-Protein Interaction Complexes

Beclin-1-Bcl-2 homodimer.

The analysis of protein-protein interaction (PPI) interface of beclin-1-Bcl-2 homodimer (Fig. 2A) indicated that the surface area (Ų) of the PPI interface was 2,636.9 and the value of solvation-free energy gain upon the formation of the interface (ΔⁱG) was −47.2 kcal/mol. The ΔG^int was −47.2 kcal/mol, entropy change at dissociation (TΔS^diss) was 17.1 kcal/mol, and the free energy of assembly dissociation, ΔG^diss, was 32.8 kcal/mol.

Fig. 2.

Formation of autophagy complexes utilizing bioinformatics studies. A: beclin1-Bcl2 homodimer: beclin-1 (bright orange) and Bcl2 (raspberry) form a homodimer in which Bcl2 binds at the BH3 domain of beclin-1. ATG14L complex I consists of beclin-1 (bright orange), APOL1G0 (marine blue), PI3KC3 (wheat), and ATG14L (magenta). UVRAG complex II consists of beclin-1 (bright orange), APOL1G0 (marine blue), PI3KC3 (wheat). and UVRAG (cyan). B: schema of protein-protein interaction is shown. In brief, beclin-1 dimerizes with Bcl2 to form an inactive complex. Activation of JNK1 phosphorylates Bcl2, causing release of beclin-1. Oligomerization of beclin-1 allows it to form PI3KC3 complexes along with ATG14 (complex I) or UVRAG (complex II). C: Rubicon complex IIIa consists of APOL1G1 (marine blue), PI3KC3 (wheat), UVRAG (magenta), and Rubicon (brown). Rubicon complex IIIb consists of APOL1G2, PI3KC3 (wheat), UVRAG (magenta), and Rubicon (brown). D: schema of Rubicon complex III is shown. Enhanced expression of Rubicon in APOL1G1 and APOL1G2 milieus lead to their binding with UVRAG and PI3KC3 in a relatively lower concentration of beclin-1. APOL1, apolipoprotein L1; BH3, Bcl-2 homology domain 3; CCD, coiled-coil domain; ECD, evolutionarily conserved domain; PI3KC3, class III phosphatidylinositol 3-kinase (PI3KC3); Rubicon, run domain beclin-1-interacting and cysteine-rich domain-containing protein; ATG14L, autophagy-related gene 14L; UVRAG, UV radiation resistance-associated gene; JNK1, c-Jun NH₂-terminal protein kinase-1.

ATG14L complex I.

The ATG14L complex I (Fig. 2A) includes beclin-1, PIK3C3, ATG14L, and APOL1G0. The analysis of PPI interfaces suggested that the interactions between beclin-1-ATG14L, ATG14L-APOL1G0, and APOL1G0-PI3KC3 were stronger than beclin-1-PIK3C3 and beclin-1-APOL1G0 based on the surface area of interfaces and binding energy. The surface area (Ų) of PPI interfaces of beclin-1-ATG14L, ATG14L-APOL1G0, and APOL1-PIK3C3 were 2,002.7, 1,583.0, and 1,510.1 respectively, whereas the Ų of beclin-1-PIK3C3 and beclin-1-APOL1G0 were less, i.e., 843.0 and 282.7, respectively. In terms of binding energy, the PPI complexes between beclin-1-ATG14L, ATG14L-APOL1G0, and APOL1G0-PIK3C3 were energetically favorable, as the values of solvation-free energy gain upon formation of the interface (ΔⁱG) in kcal/mol were −24.6, −23.8, and −22.6, respectively, whereas, the PPI complexes between beclin-1-PIK3C3 and beclin-1-APOL1G0 were energetically unfavorable, as their values of binding energy were 2.6 and −0.3 kcal/mol, respectively. Overall, the free energy gain upon the formation of assembly (complex I), ΔG^int was −68.7 kcal/mol, TΔS^diss was 17.0 kcal/mol, and the free energy of assembly dissociation, ΔG^diss, was 7.2 kcal/mol. The positive value of ΔG^diss indicates that an external driving force should be applied to dissociate the assembly. Therefore, assemblies with ΔG^diss> 0 are thermodynamically stable.

UVRAG complex II.

Complex II (Fig. 2A) includes beclin-1, PIK3C3, APOL1G0, and UVRAG. The analysis of PPI interfaces indicated that the surface area (Ų) of beclin-1-APOL1G0, APOL1G0-PIK3C3, PIK3C3-UVRAG, beclin-1-PIK3C3, beclin-1-UVRAG, and APOL1G0-UVRAG were 1,407.4, 475.7, 1,235.2, 1,406.5, 1,787.8, and 1,487.5, respectively. The solvation energy gain upon interface formation (ΔⁱG) in kcal/mol were −20.6, −9.3, −14.2, −19.3, −22.0, and −20.9, respectively. The free energy of assembly dissociation, ΔG^diss, indicated that a thermodynamically stable complex could be formed.

Schematic representation of protein-protein interaction is shown in Fig. 2B. In brief, beclin-1 dimerizes with Bcl-2 and forms an inactive complex. Activation of c-Jun NH₂-terminal protein kinase (JNK)1, in starved conditions, phosphorylates Bcl2, causing the release of beclin-1 (46). ULK1 (as a consequence of downstream signaling of inhibited mTOR) is recruited to beclin-1-PI3KC3 complexes through binding to either ATG14 L (complex I) or UVRAG (complex II), phosphorylating beclin-1 and PI3KC3, resulting in the generation of PI3P.

Rubicon complex IIIa and complex IIIb.

The Rubicon complex IIIa consists of APOL1G1, PIK3C3, UVRAG, and Rubicon, while the Rubicon complex IIIb (Fig. 2C) consists of APOL1G2, PIK3C3, UVRAG, and Rubicon. The analysis of PPI interfaces suggested that the surface area (Ų) of Rubicon-APOL1G1, APOL1G1-PIK3C3, Rubicon-PIK3C3, and UVRAG-Rubicon were 1,609.4, 1,184.4, 777.2, and 1,764.3, respectively, in the Rubicon complex IIIa. The solvation free energy gain upon interface formation (ΔⁱG) in kcal/mol were −22.2, −12.9, −11.3, and −34.3, respectively. The energy value indicates that Rubicon interacts with other associated proteins via UVRAG.

The PPI interface of Rubicon complex IIIb indicated that the Ų of Rubicon-PIK3C3, Rubicon-APOL1G2, and UVRAG-Rubicon were 1,681.5, 1,487.5, and 1,764.3, respectively. Also, the solvation free energy ΔⁱG were −22.9, −20.9, and −34.3 kcal/mol, respectively. Overall, the free energy gain upon the formation of assemblies (Rubicon complex IIIa and IIIb), ΔG^int, was −40.5 and −39.5 kcal/mol. Moreover, the positive values of free energy of assembly dissociation, ΔG^diss, indicated that a thermodynamically stable complex could be formed.

A schematic representation of Rubicon complex III is shown in Fig. 2D. In brief, enhanced expression of Rubicon in APOL1G1/APOL1G2 milieu led to its binding with UVRAG and PI3K3C3 in the relative absence of beclin-1. Formation of Rubicon complex will inhibit the fusion of autophagosome to lysosomes. Additionally, enhanced miR193a expression in APOL1 risk milieu will downregulate mTOR as well as the transcription of PIK3R3 (the regulatory unit of PI3KC3); the latter would attenuate the formation of PI3KC3 complexes, resulting in inhibition of autophagosome maturation. Additionally, inhibition of mTOR would suppress reformation of lysosomes, which are required to fuse with autophagosomes.

To confirm the composition of these complexes, immunoprecipitation (IP) studies were carried out to analyze the components of ATG14L complex I. Protein blots of cellular lysates (n = 6) from differentiated V- and G0 podocytes were probed for ATG14L and reprobed for UVRAG, beclin-1, PI3KC3, APOL1, and GAPDH. Representative gels are displayed in Fig. 3A; raw data are displayed in Supplemental Fig. S3. Cumulative densitometric data in the form of protein variable/GAPDH ratios are shown in dot plots (Fig. 3A, i–v). Cellular lysates showed the presence of both ATG14L and UVRAG in addition to the expressions of PI3KC3, beclin-1, and APOL1.

Fig. 3.

Analysis of composition of autophagy complexes by immunoprecipitation (IP) studies. A: protein blots of cellular lysates (n = 3) from differentiated vector- (V-) and G0-podocytes were probed for ATG14L and reprobed for UVRAG, beclin-1, PI3KC3 (VPS34), APOL1, and GAPDH. Representative gels are displayed. i: Distribution densitometric data of ATG14L/GAPDH ratios from A are shown in a dot plot. ii: Distribution densitometric data of UVRAG/GAPDH ratios from A are shown in a dot plot. iii: Distribution densitometric data of beclin-1/GAPDH ratios from A are shown in a dot plot. iv: Distribution densitometric data of PI3KC3/GAPDH ratios from A are shown in a dot plot. v: Distribution densitometric data of APOL1/GAPDH ratios from A are shown in a dot plot. *P < 0.05 vs. V. B: cellular lysates were immunoprecipitated with an anti-ATG14L antibody (n = 6). Protein blots of IP fractions were probed for ATG14L and reprobed for beclin-1, PI3KC3, and APOL1. IgG expression was used as a loading marker. Gels from three different IP fractions are displayed. i: Distribution densitometric data of ATG14L/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V. ii: Distribution densitometric data of beclin-1/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V. iii: Distribution densitometric data of PI3KC3/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V; iv: Distribution densitometric data of APOL1/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V. C: cellular lysates of V- and G0-podocytes from A were immunoprecipitated with anti-UVRAG antibody (n = 6). Protein blots were probed for UVRAG, beclin-1, PI3KC3, APOL1, and IgG. Gels from three independent IP fractions are shown. i: Distribution densitometric data of UVRAG/IgG ratios from C are shown in a dot plot. *P < 0.05 vs. V. ii: Distribution densitometric data of beclin-1/IgG ratios from C are shown in a dot plot. *P < 0.05 vs. V. iii: Distribution densitometric data of PI3KC3/IgG ratios from 3C are shown in a dot plot. *P < 0.05 vs. V. iv: Distribution densitometric data of APOL1/IgG ratios from C are shown in a dot plot. *P < 0.05 vs. V. D: protein blots of cellular lysates of V- and G0-podocytes were probed for APOL1, PI3KC3, and beclin-1 and reprobed for actin (n = 6). Cellular lysates from three independent lysates are shown. i: Distribution densitometric data of APOL1/actin ratios from D are shown in a dot plot. *P < 0.05 vs. V. ii: Distribution densitometric data of PI3KC3/actin ratios from D are shown in a dot plot. *P < 0.05 vs. V. iii: Distribution densitometric data of beclin-1/actin ratios from D are shown in a dot plot. *P < 0.05 vs. V. E: cellular lysates of V- and G0-podocytes were immunoprecipitated with anti-APOL1 antibody. IP fractions were probed for APOL1, PI3KC3, beclin-1, and IgG. Gels from three independent IP fractions are displayed. i: Distribution densitometric data of APOL1/IgG in ratios from E are shown in a dot plot. *P < 0.05 vs. V. ii: Distribution densitometric data of PI3KC3/IgG ratios from 3E are shown in a dot plot. iii: Distribution densitometric data of beclin-1/IgG ratios from E are shown in a dot plot. *P < 0.05 vs. V. APOL1, apolipoprotein L1; PI3KC3, class III phosphatidylinositol 3-kinase; ATG14L, autophagy-related gene 14L; UVRAG, UV radiation resistance-associated gene.

To evaluate whether ATG14L formed a complex I without UVRAG, cellular lysates of V- and G0-podocytes were immunoprecipitated with an anti-ATG14L antibody (n = 6). Protein blots of IP fractions were probed for ATG14L and reprobed for UVRAG, beclin-1, PI3KC3, and APOL1. IgG expression was used as a loading marker. Gels from three independent IP fractions are displayed in Fig. 3B; raw data are shown in Supplemental Fig. S3. ATG14L complex showed the presence of beclin-1, PI3KC3, and APOL1. Cumulative densitometric data of protein variables and IgG are shown as dot plots (Fig. 3B, i–v).

To analyze the proteins interacting with UVRAG in the complex II, cellular lysates of V- and G0-podocytes (from Fig. 3A) were immunoprecipitated with an anti-UVRAG antibody (n = 6). Protein blots from IP fractions were probed for UVRAG, beclin-1, PI3KC3, APOL1, and IgG. Gels from three IP fractions are shown in Fig. 3C; raw data are shown in Supplemental Fig. S3. Cumulative densitometric data of protein variables and IgG are shown in dot plots (Fig. 3C, i–iv). UVRAG complex showed the presence of beclin-1, PI3KC3, and APOL1.

Activation of PI3K class 3 (phosphorylation of PI3 class 3 at position 3 in PI3 complexes) is critical for autophagosome nucleation as well as for maturation (9, 10, 20, 39). We asked whether APOL1 was one of the constituents of PI3KC3 complexes (APOL1-beclin-1-PI3KC3). Protein blots of cellular lysates of V- and G0-podocytes were probed for APOL1, PI3KC3, and beclin-1 and reprobed for actin (n = 6). Cellular lysates from three independent lysates are shown in Fig. 3D; raw data are shown in Supplemental Fig. S3. Cumulative densitometric data of protein variables and actin are shown in dot plots (Fig. 3D, i–iii). V-podocytes displayed minimal expression of APOL1, PI3KC3, and beclin-1, but G0-podocytes showed a higher expression of APOL1, PI3KC3, and beclin-1.

To confirm whether APOL1 was a part of PI3KC3 complex, cellular lysates of V- and G0-podocytes were immunoprecipitated (IP) with an anti-APOL1 antibody. IP fractions were probed for APOL1, PI3KC3, beclin-1, and IgG (n = 6). Gels from three independent lysates are displayed in Fig. 3E; raw data are shown in Supplemental Fig. S3. Cumulative densitometric data of protein variables and IgG are shown in dot plots (Fig. 3E, i–iii). G0-podocytes displayed a higher expression of APOL1 and beclin-1. These findings suggest that APOL1 is one of the constituents of PI3KC3 complexes.

Recently, Rubicon has been demonstrated to inhibit the fusion of autophagosomes with lysosomes (21, 52). To validate the role of APOL1 risk alleles on the expression of Rubicon in podocytes, cellular lysates of V-, G0-, G1-, and G2-podocytes were electrophoresed, and protein blots were probed of UVRAG, Rubicon, PI3KC3, and APOL1 and reprobed for GAPDH. Gels from three independent lysates are displayed in Fig. 4A; raw data are shown in Supplemental Fig. S4. Cumulative densitometric data of protein variables and GAPDH are shown in dot plots (Fig. 4A, i–iv). G1- and G2-podocytes displayed a higher (P < 0.05) expression of Rubicon and PI3KC3 when than V- and G0-podocytes.

Fig. 4.

APOL1 risk milieus induce Rubicon complex formation in podocytes. A: protein blots of cellular lysates of vector- (V-), G0-, G1-, and G2-podocytes were probed for UVRAG, Rubicon, PI3KC3,and APOL1 and reprobed for GAPDH. Gels from three independent lysates are displayed. i: Distribution densitometric data of UVRAG/GAPDH ratios from A are shown in a dot plot. ii: Distribution densitometric data of Rubicon/GAPDH ratios from A are shown in a dot plot. *P < 0.05 vs. G0. iii: Distribution densitometric data of PI3KC3/GAPDH ratios from A are shown in a dot plot. *P < 0.05 vs. G0. iv: distribution densitometric data of APOL1/GAPDH ratios from A are shown in a dot plot. *P < 0.05 compared vs. V. B: cellular lysates of V-, G0-, G1-, and G2-podocytes from 4A were immunoprecipitated with an anti-Rubicon antibody. Immunoprecipitation (IP) fractions were probed for UVRAG, Rubicon, PI3KC3, and APOL1. IgG was used as a loading marker. Gels from three independent IP fractions are displayed. i: Distribution densitometric data of UVRAG/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V and G0. ii: Distribution densitometric data of Rubicon/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V and G0. iii: Distribution densitometric data of PI3KC3/IgG ratios from B are shown in a dot plot. iv: Distribution densitometric data of APOL1/IgG ratios from B are shown in a dot plot. *P < 0.05 vs. V. C: RNAs were extracted from different sets of V-, G0-, G1-, and G2-podocytes (n = 5). cDNAs were amplified with a specific primer for Rubicon. Cumulative data are shown in a dot plot. ***P < 0.001 vs. V and G. APOL1, apolipoprotein L1; PI3KC3, class III phosphatidylinositol 3-kinase; Rubicon, run domain beclin-1-interacting and cysteine-r UVRAG, UV radiation resistance-associated gene.

To confirm whether Rubicon complex was composed of UVRAG, PI3KC3, and APOL1 in G1- and G2-podocytes, cellular lysates of V-, G0-, G1-, and G2-podocytes were immunoprecipitated with anti-Rubicon antibody, and IP fractions were probed for UVRAG, Rubicon, PI3KC3, and APOL1. IgG was used as a loading marker. Gels from three independent IP fractions are displayed in Fig. 4B; raw data are displayed in Supplemental Fig. S4. Cumulative densitometric data of protein variables and IgG are shown in dot plots (Fig. 4B, i–iv). IP fractions displayed enhanced expression of UVRAG, Rubicon, PI3KC3, and APOL1.

Next, we asked whether APOL1 risk milieu carried the potential to enhance the transcription of Rubicon in podocytes. For this, RNAs were extracted from independent sets of V-, G0-, G1-, and G2-podocytes (n = 5). cDNAs were amplified with a specific primer for Rubicon. Cumulative data are shown in a dot plot (Fig. 4C). G1- and G2-podocytes displayed enhanced expression of Rubicon. In summary, it appears that multiple factors contribute to attenuated autophagy flux in G1- and G2-podocytes.

Expression of APOL1 Risk Alleles Is Associated with Relatively Reduced Autophagy in Podocytes

Recent reports demonstrated that there was an accumulation of autophagosomes in podocytes expressing APOL1 risk alleles (3); this effect of APOL1 risk alleles was considered a consequence of a defect in the fusion of autophagosomes and lysosomes. However, the mechanism involved was not clear. Since inhibition of mTOR is known to enhance autophagy but has also been reported to slow down the reformation of lysosomes in podocytes (6), we hypothesized that APOL1 risk alleles could cause accumulation of autophagosomes via downregulation of the mTOR pathway as one of the contributing factors. To determine the effect of APOL1 risk and nonrisk alleles, protein blots (from independent sets of cellular lysates) of differentiated V-, G0-, G1-, and G2-podocytes were probed for phospho (p)-mTOR, LC3-II, p62, PI3KC3, and GAPDH (n = 4–6). The same cellular lysates were probed for APOL1 and reprobed for GAPDH. Gels from three independent cellular lysates are displayed in Fig. 5, A, i–v, and Bi; raw data are shown in Supplemental Fig. S5. Densitometric distribution data are shown in dot plots. Both APOL1 risk alleles and nonrisk alleles downregulated the expression of p-mTOR compared with V-podocytes. On the other hand, G1- and G2-podocytes displayed a lower expression of p-mTOR compared with G0-podocytes. Although both APOL1 risk alleles and nonrisk alleles showed enhanced expression of LC3-II, only G0-podocytes showed an attenuated expression of p62, suggesting the completion of the accelerated autophagy flux. On the other hand, G1- and G2-podocytes displayed enhanced expression of p62, suggesting a comparative attenuation of autophagy flux. Interestingly, only G0-podocytes displayed enhanced expression of PI3KC3 facilitating the formation of PI3KC3 complexes. In contrast, G1- and G2-podocytes displayed an attenuated expression of PI3KC3; this disparity in the expression of PI3KC3 could have contributed to the lack of PI3P generation resulting in downregulation of autophagosome maturation in APOL1 risk milieu.

Fig. 5.

Expression of APOL1 risk alleles is associated with autophagy blockade in podocytes. A: protein blots of independent sets of differentiated vector- (V-), G0-, G1-, and G2-podocytes were probed for phospho (p)-mTOR, LC3-II, p62, PI3KC3, and GAPDH (n = 6). Gels from two independent lysates are shown. i: Densitometric distribution data of p-mTOR/GAPDH ratios from A are shown in a dot plot. *P < 0.05 vs. V and G0. ii: Densitometric distribution data of PI3KC3/GAPDH ratios from A are shown in a dot plot. *P < 0.05 and **P < 0.01 vs. V. iii: Densitometric distribution data of p62/GAPDH ratios from A are shown in a dot plot. *P < 0.05 vs. V. iv: Densitometric distribution data of LC3-I/GAPDH ratios from A are shown in a dot plot. **P < 0.01 and ***P < 0.001 vs. V. v: Densitometric distribution data of LC3-II/GAPDH ratios from A are shown in a dot plot. *P < 0.05 and **P < 0.01 vs. V. B: cellular lysates from A were probed for APOL1 and reprobed for GAPDH. Gels from two independent cellular lysates are displayed. i: Densitometric distribution data of APOL1/GAPDH ratios from B are shown in a dot plot. **P < 0.01 and ***P < 0.001 vs. V. C: protein blots from independent sets of V-, G0-, G1-, and G2-podocytes were probed for p-ULK1 and reprobed for p-PI3KC3, and GAPDH (n = 6). Gels from three independent lysates are displayed. i: Densitometric distribution data of p-ULK/GAPDH ratios from C are shown in a dot plot. *P < 0.05 compared with other variables. ii: Densitometric distribution data of p-PI3KC3/GAPDH ratios from C are shown in a dot plot. *P < 0.05 vs. other variables. APOL1, apolipoprotein L1; LC3, light chain 3; PI3KC3, class III phosphatidylinositol 3-kinase; mTOR, mammalian target of rapamycin; ULK1, Unc-51 like autophagy-activating kinase-1.

To determine the effect of APOL1 on downstream to the mTOR pathway, protein blots (independent sets of cellular lysates) of V-, G0-, G1-, and G2-podocytes were probed for phospho (p)-ULK1 and reprobed for p-PI3KC3 and GAPDH (n = 6). Gels from three independent lysates are displayed in Fig. 5C; raw data are displayed in Supplemental Fig. S5. Densitometric analyses are shown in dot plots (Fig. 5C, i and ii). G0-podocytes showed enhanced expression of p-ULK1, indicating inhibition of the mTOR pathway. G1- and G2-podocytes showed a decreased expression of phospho-ULK1 compared with G0-podocytes, but it was higher than V-podocytes. Interestingly, only G0-podocytes showed the phosphorylation of PI3KC3.

Role of miR193a in the Regulation of Autophagy in Human Podocytes

3-Methyladenine (3-MA) is a specific inhibitor of PI3KC3 under the starved condition (47). We and other investigators (12, 48) have shown previously the inhibitory effect of 3-MA on podocyte autophagy. We asked whether 3-MA was also modulating the expression of miR193a in podocytes. To determine the effect of 3-MA, independent sets of differentiated podocytes were treated with either vehicle (C) or with 3-MA (3 mM) for 24 h under starved conditions (n = 5). RNAs were extracted and assayed for miR193a. Cumulative data are shown in a dot plot (Fig. 6A). 3-MA-treated podocytes showed a threefold increase (P < 0.001) in miR193a expression. Since 3-MA is known to inhibit PI3KC3, it is possible that 3-MA would downregulate PI3KC3 through the enhanced expression of miR193a.

Fig. 6.

Role of micro- (mi)R193a in the regulation of autophagy in human podocytes. A: independent sets of differentiated podocytes were treated with either vehicle (C) or with 3-MA (3 mM) for 24 h under starved conditions (n = 5). RNAs were extracted and assayed for miR193a. Distribution data are shown in a dot plot. ***P < 0.001 vs. C. B: podocytes were transfected with empty vector or miR193a (25 nm) or treated with 3-MA for 24 h under starved conditions (n = 5). RNAs were extracted and cDNAs were amplified with a specific primer for PIK3R3 (n = 3). Distribution data are shown in a dot plot. ***P < 0.001 vs. other variables. C: RNAs were extracted from independent sets of differentiated vector (V-), G0-, G1-, and G2-podocytes and assayed for miR193a (n = 4). Distribution data are shown in a dot plot. *P < 0.05, **P < 0.01 vs. V. D: RNAs were extracted from independent sets of V-, G0-, G1-, and G2-podocytes, and cDNA were amplified with a specific primer for PIK3R3 (n = 5). Distribution data are shown in a dot plot. ***P < 0.001 vs. other variables. E: independent sets of podocytes were treated with buffer, empty vector (EV), or different concentrations of a miR193a plasmid (25 or 50 nm; n = 5). After 48 h, cells were harvested, and protein blots were probed for phospho (p)-mTOR and reprobed for GAPDH. Representative gels from two independent lysates are displayed. F: densitometric distribution data of p-mTOR/GAPDH ratios from E are shown in a dot plot. ***P < 0.001 compared vs. C and EV. 3-MA, 3-methyladenine; PIK3R3, PI3KC3 regulatory unit; mTOR, mammalian target of rapamycin.

To determine the effect of 3-MA and miR193a on the transcription of the regulatory unit of PI3KC3 (PIK3R3), independent sets of podocytes were transfected with either an empty vector or miR193a (25 nm) or treated with 3-MA (for 24 h, under starved conditions; n = 5). RNAs were extracted, and cDNAs were amplified with a specific primer for PIK3R3. Cumulative data are shown in dot plot format (Fig. 6B). Both 3-MA and miR193a inhibited transcription of PIK3R3. These findings suggest that 3-MA and miR193a use a similar mechanism to inhibit the expression of PI3KC3.

We (25, 26, 32) recently reported that APOL1G0 inversely regulates the expression of miR193a in human podocytes and parietal epithelial cells. On the other hand, APOL1 risk alleles upregulated the expression of miR193a in podocytes (25). To confirm the disruption of the APOL1-miR193a axis in podocytes expressing APOL1 risk alleles, RNAs were extracted from independent sets of differentiated V-, G0-, G1-, and G2-podocytes and assayed for miR193a (n = 4). Distribution data are shown in a dot plot (Fig. 6C). G0-podocytes displayed downregulation of miR193a, but both G1- and G2-podocytes showed enhanced expression of miR193a. These findings confirm disruption in the APOL1-mi193a axis in podocytes expressing APOL1 risk alleles.

If APOL1 risk alleles were to block autophagy through the generation of miR193a, then one would expect attenuation of the transcription of PIK3R3 in APOL1 risk milieu. To validate this hypothesis, RNAs were extracted from independent sets of V-, G0-, G1-, and G2-podocytes, and cDNAs were amplified with a specific primer for PIK3R3 (n = 5). Distribution data are shown in a dot plot (Fig. 6D). G0-podocytes displayed an 18-fold increase (P < 0.001) in PIK3R3 expression compared with V-, G1-, and G2-podocytes. However, G1- and G2-podocytes displayed a 30% decrease in PIK3R3 expression compared with V-podocytes.

miR193a has been known to inhibit the mTOR pathway (50). To determine the effect of miR193a on the expression of p-mTOR, independent sets of differentiated podocytes were treated with buffer or transfected with empty vector (EV) or different concentrations of miR193a plasmid (25 or 50 nm; n = 5). After 48 h, cells were harvested, and protein blots were probed for p-mTOR and reprobed for GAPDH. Gels from two independent lysates are displayed in Fig. 6E; raw data are displayed in Supplemental Fig. S6. Densitometric distribution is shown as a dot plot (Fig. 6F). Podocytes expressing miR193a displayed attenuated (P < 0.001) expression of p-mTOR. These findings indicate miR193a downregulates the activation of mTOR.

Autophagy Blockade Induces Dedifferentiation in Podocytes

Enhanced autophagy has been demonstrated to maintain molecular phenotype in podocytes (16). To confirm the role of autophagy in the maintenance of molecular phenotype, we evaluated the effect of autophagy blockade in normal podocytes and G0-podocytes.

To determine the role of autophagy blockade in dedifferentiation of podocytes, independent sets of differentiated podocytes were treated with either vehicle (C) or 3-MA (3 mM) for 48 h (n = 6). Protein blots from independent cellular lysates were probed for p-PI3KC3, podocalyxin, and podocin and reprobed for GAPDH. The same cellular lysates were also probed for Wilms tumor type (WT)1 and reprobed for GAPDH (n = 6). Gels from two independent lysates are displayed in Fig. 7, A and B; raw data are shown in Supplemental Fig. S7. Densitometric distribution data are shown as dot plots (Fig. 7, A, i–iii, and Biv). Podocytes treated with 3-MA showed a decrease in expression of p-PI3KC3, podocalyxin, podocin, and WT1. These findings suggest that 3-MA-induced podocyte dedifferentiation is associated with decreased phosphorylation of PI3KC3.

Fig. 7.

Autophagy blockade causes dedifferentiation in podocytes. A: independent sets of differentiated podocytes were treated with vehicle (C) or 3-MA (3 mM) for 48 h under starved conditions (n = 6). Protein blots were probed for phospho (p)-PI3KC3, podocalyxin, and podocin and reprobed for GAPDH. Gels from two independent lysates are displayed. B: cellular lysates from A were probed for WT1 and reprobed for GAPDH (n = 6). Gels from two independent lysates are displayed. i: Densitometric distribution of p-PI3CK3/GAPDH ratios from A are shown in a dot plot. ***P < 0.001 vs. C. ii: Densitometric distribution of podocalyxin/GAPDH ratios from A are shown in a dot plot. ***P < 0.001 vs. C. iii: Densitometric distribution of podocin/GAPDH ratios from A are shown in a dot plot. ***P < 0.001 vs. C. iv: Densitometric distribution of WT1/GAPDH ratios from B are shown in a dot plot. ***P < 0.001 vs. C. C: independent sets of differentiated G0-podocytes were treated with vehicle or 3-MA for 48 h under starved conditions (n = 6). Protein blots of vector (V-), G0- and G0 + 3MA podocytes were probed with APOL1, nephrin, CD2AP, and podocalyxin and reprobed for actin. Gels from three independent lysates are displayed. i: Densitometric distribution of APOL1/GAPDH ratios from C are shown in a dot plot. ii: Densitometric distribution of nephrin/actin ratios from C are shown in a dot plot. *P < 0.05 vs. Vs and G0 + 3MA. iii: Densitometric distribution of CD2AP/actin ratios from C are shown in a dot plot. iii: Densitometric distribution of podocalyxin/actin ratios from C are shown in a dot plot. *P < 0.05 vs. Vs and G0 + 3MA. PI3KC3, class III phosphatidylinositol 3-kinase; 3-MA, 3-methyladenine; mTOR, mammalian target of rapamycin; WT1, Wilms tumor type 1; CD2AP, CD2-associated protein.

To determine the effect of autophagy blockade in podocytes overtly expressing APOL1G0, independent sets of differentiated G0-podocytes were treated with either vehicle or 3-MA for 48 h (n = 6). Protein blots of V- and G0-podocytes and G0 + 3-MA podocytes were probed for APOL1, nephrin, CD2AP, and podocalyxin and reprobed for actin. Gels from three independent lysates are displayed in Fig. 7C; raw data are shown in Supplemental Fig. S7. Densitometric distribution is shown in dot plots (Fig. 7C, i–iv, and Civ) G0-podocytes showed enhanced expression of nephrin and podocalyxin. However, 3-MA inhibited the expression of podocyte markers in G0-podocytes. These findings confirm that autophagy blockade contributes to podocyte dedifferentiation.

APOL1 Risk Alleles Are Associated with Dedifferentiation of Molecular Phenotype in Podocytes

If autophagy blockade is contributing to dedifferentiation of podocytes, then podocytes expressing APOL1 risk alleles should also display loss of podocyte molecular phenotype. To test this concept, protein blots from independent sets of V-, G0-, G1-, and G2-podocytes were probed for APOL1 and reprobed for CD2AP, nephrin (podocyte markers), and GAPDH (n = 6). Gels from three independent lysates are displayed in Fig. 8A; raw data are shown in Supplemental Fig. S8. Densitometric distribution is shown in dot plots (Fig. 8A, i–iii). G0-podocytes displayed enhanced expression of nephrin and CD2AP, but G1- and G2-podocytes showed a decrease in CD2AP and nephrin expression.

Fig. 8.

APOL1 risk alleles are associated with dedifferentiation of molecular phenotype in podocytes. A: protein blots of independent sets of vector (V-), G0-, G1-, and G2-podocytes were probed with APOL1 and reprobed for CD2AP, nephrin, and GAPDH (n = 6). Gels from three independent lysates are displayed. i: Densitometric distribution of CD2AP/GAPDH ratios from A are shown in a dot plot. **P < 0.01, ***P < 0.001 vs. V. ii: Densitometric distribution of nephrin/GAPDH ratios from A are shown in a dot plot. *P < 0.05, **P < 0.01 vs. V. iii: Densitometric distribution of APOL1/GAPDH ratios from from A are shown in a dot plot. *P < 0.05, **P < 0.01 vs. V. APOL1, apolipoprotein L1; CD2AP, CD2-associated protein.

Rubicon-Silencing Prevents Dedifferentiation of Podocytes in APOL1 Risk Milieu

We asked, if Rubicon was contributing to autophagy blockade in APOL1 risk milieu, then silencing of Rubicon in G1- and G2-podocytes should protect against APOL1 risk milieu-induced dedifferentiation. To test this hypothesis, independent sets of V-, G0-, G1-, and G2-podocytes were transfected with scrambled (25 nM) or Rubicon-siRNA (n = 6). After 48 h, protein blots were probed for WT1 and podocalyxin. The same blots were reprobed for GAPDH. Representative gels are displayed in Fig. 9A; raw data are shown in Supplemental Fig. S9. Densitometric distribution is shown in dot plots (Fig. 9A, i and ii). G1- and G2-podocytes silenced for Rubicon showed enhanced expression of WT1 and podocalyxin compared with scrambled (SCR)G1- and SCRG2-podocytes. These findings suggested the role of Rubicon in the induction of podocyte dedifferentiation in APOL1 risk milieu.

Fig. 9.

Effect of Rubicon-silencing on dedifferentiation and autophagy blockade in APOL1 risk milieu. A: independent sets of vector (V-), G0-, G1-, and G2-podocytes were transfected with scrambled (SCR, 25 nM) or Rubicon-small interfering (si)RNA (n = 5). After 48 h, protein blots were probed for WT1 and podocalyxin; the same blots were reprobed for GAPDH. Representative gels are displayed. i: Densitometric distribution of podocalyxin/GAPDH ratios from A are shown in a dot plot. **P < 0.01 vs. V only; #P < 0.05 vs. G2 only; ###P < 0.001 vs. G1 only. ii: Densitometric distribution of WT1/GAPDH ratios from A are shown in a dot plot. **P < 0.01 vs. V only; ***P < 0.001 vs. G1 and G2 only; #P < 0.05 vs. respective V and G0 only; ###P < 0.001 vs. G1 and G2 only. B: independent sets of HEK cells were transfected with vector (VHEK), APOL1G0 (G0HEK), APOL1G1 (G1HEK), and APOL1G2 (G2HEK) (n = 6). After 48 h, protein blots from independent cellular lysates of VHEKs, G0HEKs, G1HEKs, and G2HEKs were probed for Rubicon and p62 and reprobed for GAPDH. Gels from three independent cellular lysates are displayed. i: Densitometric distribution data of Rubicon/GAPDH ratios from B are shown in a dot plot. **P < 0.01 vs. VHEKs and G0HEKs. ii: Densitometric distribution data of p62/GAPDH ratios from B are shown in a dot plot. ***P < 0.001 vs. VHEKs and G0HEKs. Rubicon, run domain beclin-1-interacting and cysteine-rich domain-containing protein; APOL1, apolipoprotein L1; WT1, Wilms tumor type 1.

We further asked whether APOL1 risk alleles carry the potential to enhance the expression of Rubicon in other cells such as HEKs; if so, then the enhanced expression of Rubicon would be associated with the blockade of autophagy in these cells too. To validate our hypothesis, independent sets of HEKs were transfected with vector (VHEK), APOL1G0 (G0HEK), APOL1G1 (G1HEK), and APOL1G2 (G2HEK) (n = 6). After 48 h, protein blots from independent cellular lysates of VHEKs, G0HEKs, G1HEKs, and G2HEKs were probed for Rubicon and p62 and reprobed for GAPDH. Gels from three independent cellular lysates are displayed in Fig. 9B; raw data are displayed in Supplmental Fig. S9. Densitometric distribution is shown in dot plots (Fig. 9B, i and ii). G1HEKs and G2HEK showed enhanced (P < 0.01) expression of Rubicon and p62. These findings suggested that enhanced expression of Rubicon in HEKs could have contributed to the blockade of autophagy as indicated by the abundance of p62.

DISCUSSION

In the present study, G0-, G1-, and G2-podocytes displayed enhanced expression of LC3-II and beclin-1, consistent with enhanced autophagy; however, only G0-podocytes displayed downregulation of the expression of p62 (an indicator of the completion of autophagy flux). On the other hand, G1- and G2-podocytes displayed an increase in the expression of p62, suggesting a partial blockade of the latter phase of the autophagy. Acridine orange staining in podocytes and immunolabeling for LC3 in HEKs showed an increase in accumulation of autophagosomes in both APOL1 nonrisk and risk milieus. Docking and protein-protein interaction studies suggested the formation of ATG14L complex I and UVRAG complex II in G0-podcytes, whereas they suggested Rubicon complex III formation in G1- and G2-podocytes. IP studies with anti-ATG14L and anti-UVRAG antibodies supported the formation of two different complexes, one containing ATG14L and the other having UVRAG in G0-podocytes; however, IP studies in cellular lysates of G1- and G2-podocytes indicated the formation of Rubicon III complexes. Interestingly, IP studies on cellular lysates of G0-podocytes with an anti-APOL1 antibody suggested that APOL1 fraction contained ATG14L, beclin-1, and PI3KC3, indicating that APOL1 is one of the constituents of the complex. G0-podocytes showed enhanced phosphorylation of PI3KC3, but podocytes-expressing APOL1 risk alleles displayed attenuated phosphorylation of PI3KC3. As expected, G0-podocytes showed attenuated expression of miR193a, thus confirming the presence of a functional APOL1-miR193a axis. In contrast, G1- and G2-podocytes showed a threefold increase in miR193a expression, indicating the occurrence of disruption in the APOL1-miR193a axis. Treatment of podocytes with 3-MA (a specific inhibitor of PI3KC3) induced miR193a expression in addition to dedifferentiation. G0-podocytes displayed a higher expression of PIK3R3 compared with podocytes expressing APOL1 risk alleles; moreover, podocytes treated with 3-MA as well as overexpressing miR193a showed attenuated transcription of PIK3R3. Podocytes expressing APOL1 risk alleles also displayed dedifferentiation. Nonetheless, silencing of Rubicon in podocytes prevented the dedifferentiation in the APOL1 risk milieu. These findings suggest that miR193a could be contributing to podocyte dedifferentiation via blockade of autophagy, specifically in the APOL1 risk milieu.

The role of p62, an adaptor protein in the progression of autophagy has been elucidated during the past decade (18a). It binds to LC3 along with the defective and surplus mitochondria tagged by ubiquitin, contributing to their entrapment in the autophagosome (17, 18a). This confirms the importance of p62 in the molecular mechanisms involved, by which cargos are identified and consequently sequestered within autophagosomes. In the present study, APOL1G0 expression in podocytes displayed a decreased expression of p62. Since p62 is considered the substrate for autophagy, its decrease is consistent with increased autophagy in a high APOL1G0 milieu. Since podocytes expressing APOL1 risk alleles displayed enhanced expression of p62, it indicated the accumulation of undegraded cargo as a consequence of partial blockade of autophagy. These findings are consistent with the observations in autophagy flux in podocytes expressing APOL1 risk and nonrisk alleles by other investigators (3, 23, 24).

Kruzel-Davila et al. (23) demonstrated that the expression of APOL1 risk alleles but not the nonrisk alleles in pericardial nephrocytes of adult flies contributed to the accumulation of constitutively secreted atrial natriuretic factor-red fluorescent protein (ANF-RFP) in vesicles, indicating a defect in endolysosomal traffic. These investigators also demonstrated that APOL1 kidney risk variants impaired the endosomal trafficking, an effect attributed to impaired acidification in a yeast experimental platform. Beckerman et al. (3) also demonstrated a lower degree of autophagy flux in APOL1 risk allele-transfected HEK293 cells. Similarly, the acidification of the late endosomes was defective in risk allele-transfected HeLa cells compared with APOL1G0-transfected HeLa cells. These investigators speculated that impaired acidification and diminished lysosomal fusion contributed to a low level of autophagy flux in the APOL1 risk milieu. In the present study, HEKs and podocytes expressing APOL1 risk alleles displayed enhanced expression of Rubicon, causing inhibition in the formation of autolysosomes. Moreover, disruption of the APOL1-miR193a axis in podocytes expressing APOL1 risk alleles was associated with an enhanced expression of miR193a; the latter could inhibit autophagy flux by its direct binding on mTOR and PIK3R3 genes (50).

Inhibition of mTOR in podocytes enhances early autophagy flux, but it slows down the latter phase of autophagy as a consequence of inhibition of the reformation of lysosomes; the latter is required in large numbers to generate autolysosomes (fusion of lysosomes and autophagosomes) during the accelerated phase of autophagy in cells such as podocytes (6). In the present study, podocytes expressing APOL1 risk alleles not only showed enhanced expression of miR193a but also displayed downregulation of mTOR; the latter could result in the partial blockade of autophagosome-lysosomal fusion in podocytes expressing APOL1 risk alleles. Of note, miR193a binds to the regulatory unit of PI3KC3 (PI3KR3), resulting in its downregulation (50); the latter would compromise the formation of ATG14L and UVRAG complexes and would result in the downstream slowing of PI3P generation and manifest in the form of inhibition of autophagosome maturation. Thus, it appears that enhanced expression of miR193a contributes to partial blockade of autophagy flux through multiple ways in APOL1 risk milieu.

Enhanced autophagy has been reported to play a role in the preservation of the differentiated podocyte molecular phenotype (16). On that account, inhibition of autophagy by 3-MA has been associated with podocyte dedifferentiation (12). In the present study, 3-MA enhanced miR193a expression in podocytes. Interestingly, both miR193a and 3-MA attenuated the expression of PIK3R3. It is likely that 3-MA may also be inhibiting autophagy through the generation of miR193a in podocytes. Since podocytes expressing APOL1 risk alleles also showed attenuated transcription of PIK3R3, it appears that the APOL1 risk milieu may be using mechanisms similar to 3-MA in slowing autophagy flux. Additionally, miR193a negatively regulates the expression of WT1 (a master transcriptional regulator of nephrin and podocalyxin) in podocytes (15); therefore, it is likely that 3-MA-induced podocyte dedifferentiation (besides its inhibitory effect on autophagy) may be partially mediated through miR193a-mediated modulation of WT1 expression.

In conclusion, the APOL1 risk milieu is associated with an attenuated completion of autophagy flux in podocytes at multiple steps, including mTOR inhibition, attenuation of the transcription PIK3R3, and enhanced expression of Rubicon, as shown by schematic representation in Fig. 10. Moreover, disruption of the APOL1-miR193a axis contributes to autophagy blockade and associated podocyte dedifferentiation in APOL1 risk milieu.

Fig. 10.

Schematic diagram showing the role of APOL1 risk alleles in blockade autophagy in podocytes. APOL1G1 and APOL1G2 enhance expression of Rubicon, which forms a Rubicon complex inhibiting phosphorylation of UVRAG and attenuates generation of PI3P, resulting in inhibition of fusion of autophagosomes with lysosomes. Disruption of APOL1-miR193a axis in APOL1 risk milieu enhances expression of miR193a. miR193a attenuates transcription of PIK3C3′s regulatory unit (PIK3R3) and compromises maturation of autophagosomes. miR193a-induced attenuated transcription of mTOR is inhibiting the reformation of lysosomes required for fusion with autophagosomes. APOL1, apolipoprotein L1; PI3P, phosphatidyinositol-3-phosphate; PI3K, phosphatidylinositol 3-kinase; Pi3KC3, class III phosphatidylinositol 3-kinase; PIK3E3, PI3KC3′s regulatory unit; mTOR, mammalian target of rapamycin; Rubicon, run domain beclin-1-interacting and cysteine-rich domain-containing protein; UVRAG, UV radiation resistance-associated gene.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants RO1 DK-098074 and RO1 DK-118017 to P. C. Singhal and by grants to K. Skorecki from the Israel Science Foundation (ISF 182/15) and Rambam Medical Center, Kaylie Kidney Health Center of Excellence, and the Beutler Foundation for Genomic Medicine Research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

P.C.S. conceived and designed research; V.K., K.A., A.J., A. Mishra, X.L., M.Q., S.C., R.P., and J.M. performed experiments; V.K., A.J., H.V., J.M., and A. Malhotra analyzed data; M.A.S. and K.S. interpreted results of experiments; V.K., A.J., and A. Mishra prepared figures; P.C.S. drafted manuscript; P.C.S. edited and revised manuscript; V.K., K.A., A.J., A. Mishra, H.V., X.L., M.Q., S.C., R.P., J.M., A. Malhotra, K.S., and P.C.S. approved final version of manuscript.