DGAT2 Inhibition Potentiates Lipid Droplet Formation To Reduce Cytotoxicity in APOL1 Kidney Risk Variants

Significance Statement

Variants G1 and G2 of the gene encoding apolipoprotein L1 (APOL1) are associated with increased risk of kidney disease in certain populations. In previous work, the authors demonstrated that recruitment of these variants from the endoplasmic reticulum to lipid droplets is associated with reduced cytotoxicity in podocytes. In this study, they confirm differences in lipid droplet distribution between isogenic human kidney organoids expressing wild-type APOL1 (G0) or risk variant APOL1 (G2), which are affected by the levels of APOL1 expression. They also demonstrate that inhibition of diacylglycerol O-acyltransferase 2 (DGAT2), a key enzyme in triglyceride biosynthesis, upregulates genes involved in lipid droplet formation. DGAT2 inhibitors may thus represent a potential therapeutic approach to reduce the cytotoxic effects of APOL1 risk variants that contribute to APOL1 nephropathy.

Visual Abstract

Abstract

Background

Two variants in the gene encoding apolipoprotein L1 (APOL1) that are highly associated with African ancestry are major contributors to the large racial disparity in rates of human kidney disease. We previously demonstrated that recruitment of APOL1 risk variants G1 and G2 from the endoplasmic reticulum to lipid droplets leads to reduced APOL1-mediated cytotoxicity in human podocytes.

Methods

We used CRISPR-Cas9 gene editing of induced pluripotent stem cells to develop human-derived APOL1^G0/G0 and APOL1^G2/G2 kidney organoids on an isogenic background, and performed bulk RNA sequencing of organoids before and after treatment with IFN-γ. We examined the number and distribution of lipid droplets in response to treatment with inhibitors of diacylglycerol O-acyltransferases 1 and 2 (DGAT1 and DGAT2) in kidney cells and organoids.

Results

APOL1 was highly upregulated in response to IFN-γ in human kidney organoids, with greater increases in organoids of high-risk G1 and G2 genotypes compared with wild-type (G0) organoids. RNA sequencing of organoids revealed that high-risk APOL1^G2/G2 organoids exhibited downregulation of a number of genes involved in lipogenesis and lipid droplet biogenesis, as well as upregulation of genes involved in fatty acid oxidation. There were fewer lipid droplets in unstimulated high-risk APOL1^G2/G2 kidney organoids than in wild-type APOL1^G0/G0 organoids. Whereas DGAT1 inhibition reduced kidney organoid lipid droplet number, DGAT2 inhibition unexpectedly increased organoid lipid droplet number. DGAT2 inhibition promoted the recruitment of APOL1 to lipid droplets, with associated reduction in cytotoxicity.

Conclusions

Lipogenesis and lipid droplet formation are important modulators of APOL1-associated cytotoxicity. Inhibition of DGAT2 may offer a potential therapeutic strategy to attenuate cytotoxic effects of APOL1 risk variants.

Two common coding variants in the apolipoprotein L1 (APOL1) gene referred to as G1 and G2 increase the risk of several forms of kidney disease under a recessive model of inheritance.^1–3 We previously demonstrated that recruitment of APOL1 risk variants from the endoplasmic reticulum (ER) to lipid droplets (LDs) is associated with reduced cytotoxicity in podocytes.⁴ Primary podocytes transiently transfected with RFP-tagged APOL1 isoforms G1 and G2, supplemented with palmitic acid (C16:0), resulted in fewer podocyte LDs and increased cytotoxicity, whereas APOL1 isoform G0 and oleic acid (C18:1) increased podocyte LD numbers and promoted cell survival.⁴ Our previous study suggested that the conversion of free fatty acid to triglyceride in the form of LD is important to reduce the APOL1 risk variant–associated lipotoxicity.

LDs, historically considered inert storage compartments for lipids, are now known to enhance cellular defense against lipotoxicity, ER stress, and autophagy-associated mitochondrial damage.⁵ The ability of free fatty acid uptake and metabolism to increase biosynthesis and enlargement of LDs requires the concerted activities of diacylglycerol O-acyltransferases 1 and 2 (DGAT1 and DGAT2), among many other enzymes. DGAT1 and DGAT2, key enzymes in the final step of triglyceride biosynthesis, catalyze the conversion of diacylglycerol to triacylglycerol.⁶ Both DGAT1 and DGAT2 are involved in the final step in triglyceride formation but they differ in tissue expression and share no homology at the protein level.⁷ Moreover, although Dgat1 knockout mice have reduced diet-induced obesity and less insulin resistance,⁸ genetic knockout of Dgat2 resulted in mice dying shortly after birth.⁹

Study of the in vivo biology of APOL1 poses a particular challenge due to the gene’s absence from the genomes of all standard experimental organisms used to model human disease. Multiple disease mechanisms have been proposed based on findings in cell-based models.³ Animal models developed to interrogate APOL1 biology have to date modeled APOL1 nephropathy with varying degrees of fidelity.^10–16 Interpretation of APOL1 function in podocytes has been especially difficult, as mouse models do not precisely recapitulate the human disease. A major barrier to the study of endogenous APOL1 is its restricted expression, limited to a few higher primates including humans, baboons, and gorillas.^17,18 In this context, the recent development of human kidney organoids represents a major technical advance in human disease modeling.^19,20 Several groups have already used kidney organoids to model human genetic diseases including polycystic kidney disease, nephrotic diseases, and congenital diseases of the kidney.^21–27 In a recent study by Liu et al., single-cell RNA sequencing was used to examine differences between kidney organoids expressing APOL1 G0 and G1 risk variants.²⁸ Single-cell transcriptomics revealed cell-type–specific differences in G1 organoid response to APOL1 induction.

In this study, we use induced pluripotent stem cell (iPSC)-derived kidney organoids, primary podocytes, and HEK 293 cells stably expressing APOL1 to characterize APOL1 risk variant–associated changes in LD biosynthesis and enlargement. We demonstrate that perturbations in enzymes involved in LD formation and fatty acid metabolism can alter the cytotoxicity associated with the APOL1 risk variants.

Methods

Human Primary Podocytes and Cell Culture

Human primary podocytes were grown in the manufacturer’s specified medium on flasks precoated with human podocyte primary cell culture complete extracellular matrix (Celprogen) as described previously.^4,29 Stable tetracycline-inducible transgenic APOL1-expressing HEK T-REx-293 cell lines (APOL1-G0/G0-T-REx-293, APOL1-G1/G1-T-REx-293 and APOL1-G2/G2-T-REx-293) were grown in DMEM (Corning) supplemented with 10% tetracycline system-approved FBS (Atlanta Biologicals), 0.2 mg/ml zeocin, 2 µg/ml blasticidin, and 1% antibiotic-antimycotic (Corning) at 37°C and 5% CO₂, as previously described.³⁰

Derivation of iPSC

Human iPSC isogenic lines were derived from cryopreserved human dermal fibroblasts from a donor of African ancestry (Cell Applications Cat. #106-05n, lot #1750 (G0/G0)) using CytoTune-iPS 2.0 Sendai reprogramming kit (Life Technologies Cat. #A16517) at the Harvard Stem Cell Institute iPS Core Facility. Nonisogenic lines were derived from Cell Applications Cat. #106-05n, lots 1750-C (G0/G0), 1481-G (G0/G0), 2370-B (G1/G1), and 1907-D (G2/G2). The iPSC lines were characterized for pluripotency and trilineage differentiation as per standard protocols at the Harvard Stem Cell Institute iPS Core Facility. All lines were confirmed to be of normal karyotype and with expected trilineage differentiation. iPSCs were maintained with daily medium changes of mTeSR1 medium in six-well plates coated with Matrigel. Cells were passaged using gentle cell dissociation reagent and transferred to T25 flasks coated with Matrigel before differentiation. iPSCs were routinely tested and confirmed negative for mycoplasma. iPSC immunofluorescence (IF) studies were performed as described by Takasato et al.³¹

Construction of Homozygous APOL1^G0/G0 and APOL1^G2/G2 iPSC

Single-stranded donor oligonucleotide (ssODN)-mediated CRISPR-Cas9 was used to generate the G2 risk variant (rs71785313) with a 6 bp deletion resulting in biallelic deletion of p.N388/Y389 (Supplemental Figure 1). Homozygous APOL1^G0/G0 and APOL1^G2/G2 iPSCs were generated on an isogenic background for the 1750-C line starting at passage 8. The guide RNA (guide X) was TAATTATAAGATTCTGCAGG (encoding amino acids NYKILQ). The ssODN was CAGCTGAGGAGCTGAAGAAGGTGGCTCAGGAGCTGGAGGAGAAGCTAAACATTCTCAACAACAAGATTCTGCAGGCGGACCAAGAACTGTGACCACAGGGCAGGGCAGCCACCAGGAGAGATATGCCTAACAGGGGCCAGGACAAAATGCAAACTTTTTTTTTTTTCTGAGACAGAGTCT. The iPSC line 1750-C with APOL1 mixed haplotype (E/K) at residue 150 (Supplemental Figure 1A) was transfected at passage 12 and sorted at passage 13. Twenty-four colonies were picked at passage 14. The remaining cells were used for genomic DNA analysis by TIDE (Tracking of Indels by Decomposition) (Supplemental Figure 1C). Two plates each with 96 clones were subjected to guide X. Clones were subjected to PCR using 5 µl per 40 µl Phusion high-fidelity buffer 5×, 10 µl cell lysate, with an annealing temperature of 64°C (Thermo Fisher Scientific). Several clones of APOL1^G0/G0 and APOL1^G2/G2 iPSCs were isolated and expanded to passage 18–26 before differentiation. The iPSC clones using in this study were 1750-C 1F1 and 2D8 for APOL1^G0/G0 and APOL1^G2/G2, respectively. The isogenic lines were confirmed to have normal karyotype and expression of pluripotency markers (Supplemental Figure 2).

Kidney Organoid Differentiation

Kidney organoids were differentiated using the protocol developed by Takasato et al. with minor modifications.³¹ In brief, a feeder-free system was used by plating human iPSCs in mTeSR1 supplemented with Rho kinase inhibitor Y-27632 (Stemcell Technologies Cat. #72304) for 24 hours in Matrigel-coated plates (Corning Cat. #CB-40234A) for the first thaw. The cells were maintained in mTeSR1 and passaged using gentle cell dissociation reagent (Stemcell Technologies Cat. #7174). Modification of the protocol used by Takasato et al.³¹ was the use of APEL2 medium (Stemcell Technologies Cat. #06270) supplemented with 5% PFHM-II (protein-free hybridoma mix II) in place of APEL. Cells were treated with 8 µM CHIR99021 (R&D Systems Cat. #4423/10) for 4 days, followed by recombinant human FGF-9 (200 ng/ml; R&D Systems Cat. #273-F9-025/CF) and heparin (1 μg/ml; Sigma Cat. #4784) for an additional 3 days. At day 7 cells were dissociated into single cells by a 2-minute trypsin EDTA 0.05% (Thermo Fisher Scientific Cat. #25300-054) exposure, and one million cells were pelleted at 400 × g for 2 minutes. Two cell pellets were transferred to each well of a six-well transwell plate (Corning Cat. #07-200-170).

Pellets were incubated with a pulse of 5 μM CHIR99021 in APEL2 + PFHM-II medium for 1 hour at 37°C, 5% CO₂. After 1 hour, the medium was changed to APEL2 + PFHM-II supplemented with FGF-9 (200 ng/ml) and heparin (1 μg/ml) for an additional 5 days. Cells were subsequently maintained in APEL2 + PFHM-II medium for 14 days, with medium change every other day.

Transfections

Cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific) in Opti-MEM (Life Technologies) using 0.5 µg plasmid (0.25 µg per plasmid for cotransfections) cDNA and 1 µl each of P3000 and Lipofectamine 3000 reagent per 24-well, as per the manufacturer’s instructions. The construction of plasmids for APOL1-TagRFPT (G0, G1, and G2) was described previously.⁴

Immunoblot

Cell lysates prepared from cells washed with ice-cold PBS were lysed in 1% NP-40 lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA, 1% NP-40; Boston BioProducts Cat. #BP119500ML) supplemented with cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail (Millipore Sigma Cat. #11836170001). Lysates cleared by a 10-minute centrifugation at 8000 × g and 4°C were boiled for 5 minutes in SDS sample buffer with β-mercaptoethanol and separated by SDS-PAGE (Bio-Rad). Proteins transferred to polyvinylidene difluoride membranes were blocked in 5% (wt/vol) skim milk in Tris-buffered saline with 0.05% Tween (TBST) for 1 hour, then incubated overnight at 4°C with primary antibodies (1:1000 unless otherwise specified). Immunoblots were washed with TBST and incubated with appropriate horseradish peroxidase-conjugated secondary antibodies (1:2500; Santa Cruz Biotechnology), visualized by enhanced chemiluminescence (SuperSignal West Dura or Femto kit; Life Technologies or GE Healthcare) and imaged by ProteinSimple FluorChem E or R (Bio-Techne) or with a ChemiDoc MP device (Bio-Rad). Band intensity was quantitated by densitometric analysis using ImageJ (version 1.47). Antibodies were from the following sources: APOL1 (1:1000; Millipore Sigma Cat. #HPA018885), APOL1 (1:1000; Abcam Cat. #ab252218), DGAT1 (1:500; Millipore Sigma Cat. #SAB4301075), DGAT2 (1:500; Millipore Sigma Cat. #SAB2500308), E-cadherin (1:1000; BD Transduction Laboratories Cat. #610181), vinculin (1:2500; Millipore Sigma Cat. #V9131), WT1 (1:100; Santa Cruz Biotechnology Cat. #sc-192), podocin (1:1000; Millipore Sigma Cat. #P0732), and nephrin (1:500; R&D Systems Cat. #AF4269).

Microscopy

Brightfield microscopy was performed using a CKX31 inverted microscope (Olympus) with a CAch N 10×/0.25 PhP[infinity]/1/FN22 or LCAch N 20×/0.40 PhP[infinity]/1/FN22 objective equipped with an Exmor RS IMX315 12 MP camera (Sony) for image acquisition. For immunofluorescence (IF) and confocal microscopy, cells were grown on glass coverslips, fixed for 20 minutes with 4% paraformaldehyde (Electron Microscopy Science Cat. #15710), quenched with 50 mM ammonium chloride and permeabilized with 0.3% Triton X-100. Fixed cells were blocked with 0.2% gelatin in PBS followed by incubation with primary antibodies for 1 hour or overnight. Kidney organoids were fixed in 2% paraformaldehyde for 20 minutes at 4°C, stored in PBS, and processed as described by Takasato et al.³¹ Primary antibodies used for IF included APOL1 (Abcam Cat. #ab252218) and APOL1 (Millipore Sigma Cat. #HPA018885) at 1:100, CD31 (BD Pharmingen Cat. #555444) at 1:100, nephrin (R&D Systems Cat. #AF4269) at 1:200, and ECAD (BD Biosciences Cat. #610181) at 1:300. GATA3 (Cell Signaling Technology Cat. #5852S), HOXD11 (Santa Cruz Biotechnology Cat. #sc-81969), Nanog (Cell Signaling Technology Cat. #4903) and Oct3/4 (Santa Cruz Biotechnology Cat. #sc-5279) were used at 1:300. Lotus tetragonolobus lectin (LTL) fluorescein (Vector Laboratories Cat. #FL-1321) was used at 1:200 and incubated during secondary antibody incubation. Alexa Fluor 647 Phalloidin (Thermo Fisher Scientific Cat. #A22287) was used at 1:500. Immunofluorescence for Supplemental Figure 2 was prepared at the Harvard Stem Cell Institute using antibodies for TRA-1-60 (Millipore Cat. #MAB4360), SSEA-3 (Millipore Cat. #MAB4303), Nanog ( Abcam Cat. #ab21624), and Oct4 (Cat. #ab19857) as previously described.³² Antibody-labeled organoids were then incubated with Alexa Fluor 405–conjugated anti-mouse IgG (Abcam Cat. #ab175658), Alexa Fluor 488–conjugated anti-mouse IgG (Thermo Fisher Scientific Cat. #A21202), Alexa Fluor 568–conjugated anti-rabbit IgG (Thermo Fisher Scientific Cat. #A10042), and/or Alexa Fluor 647–conjugated anti-sheep IgG (Thermo Fisher Scientific Cat. #A21448) for 1 hour at 1:500 for cells and overnight at 4°C at 1:250 for kidney organoids. Fixed, stained cells were mounted with ProLong Gold antifade reagent with or without DAPI (Thermo Fisher Scientific Cat. #P36931 or #P36930). To stain LDs, cells were incubated for 5 minutes with BODIPY 493/503 (1 µg/ml; Millipore Sigma Cat. #D3922) and kidney organoids were incubated with Nile Red (1 μg/ml; Millipore Sigma Cat. #72485) for 15 minutes. Experiments using DGAT inhibitors were performed using the DGAT1 inhibitor (T863; Millipore Sigma Cat. #SML0539) and DGAT2 inhibitor (PF-06424439; Millipore Sigma Cat. #PZ0233). Confocal images were acquired by LSM 880 laser scanning microscope (Zeiss) with a 63× oil lens, NA 1.4 or Olympus FluoView FV1000 (Figure 4A).

Figure 4.

Glomeruli of APOL1^G2/G2 organoids have lower baseline LD number than APOL1^G0/G0 organoid glomeruli, upregulated by treatment with oleic acid or IFN-γ. (A) Immunofluorescence images of representative isogenic APOL1^G0/G0 and APOL1^G2/G2 kidney organoids treated with IFN-γ for 24 hours. Scale bar, 20 μm. White outline illustrates nephrin-positive glomerular region. The experiment shown is representative of five independent kidney organoid preparations. (B) Quantitation of kidney organoid nephrin-positive glomerular area per organoid from one of five representative organoid preparations. Bars represent mean±SD; ns, nonsignificant. (C) Quantitation of LD per glomerulus per treatment group from two combined experiments in APOL1^G0/G0 and APOL1^G2/G2 kidney organoids treated with vehicle or IFN-γ for 72 hours. Bars represent mean±SD; **P<0.01, ***P<0.001, ns, nonsignificant. (D) Immunofluorescence images of representative (n=2) APOL1^G0/G0 and APOL1^G2/G2 kidney organoids treated with vehicle or 500 µM oleic acid (OA) for 24 hours. Scale bar, 50 μm. (E) Quantitation of LD per glomerulus in APOL1^G0/G0 and APOL1^G2/G2 kidney organoids treated with vehicle or oleic acid (OA) for 24 hours from one of three representative experiments of similar results. Bars represent mean±SD; ***P<0.001 and ****P<0.0001.

Bulk RNA Sequencing

Total RNA was extracted from kidney organoids using the RNeasy Micro kit (Qiagen). After RNA quality control and quantification, the samples were subjected to library preparation with polyA selection. RNA sequencing was performed with the following parameters: 150 bp, paired end reads, and 20 million reads per sample. FASTA files underwent quality control and adaptor trimming with Trim Galore (https://github.com/FelixKrueger/TrimGalore). Trimmed reads were aligned to the reference genome GRCh38 with STAR aligner (Version 2.7.0a).³³ The Subread package was then used to count reads with featureCounts.³⁴ The count matrix was then used as input for the edgeR package for differential gene expression analysis. A 0.05 false discovery rate (FDR) cutoff was used. STRING (https://string-db.org/) was used for protein-protein interaction network generation.³⁵ The GeneMANIA website (genemania.org) was used to generate a function-based association protein-protein interaction network with the input genes list provided.³⁶ The sequencing data are available for download at http://www.ncbi.nlm.nih.gov/bioproject/795435 (Accession Number PRJNA795435).

Cytotoxicity and Lactate Dehydrogenase Release Assays

HEK T-REx-293 cells were plated at 1 × 10⁵ per well in 96-well plates. Cells induced with 10 ng/ml tetracycline for 16–18 hours were subjected to testing by the Multi Tox-Fluor Multiplex cytotoxicity/viability assay (Promega) as per the manufacturer’s instructions using a SpectraMax M5 microplate reader (Molecular Devices) as previously described.³⁰ Lactate dehydrogenase (LDH) release into the supernatant by dying cells was measured using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega) as previously described.³⁷ Briefly, primary human podocytes (Celprogen) were plated in 12-well plates and transiently transfected with APOL1-RFP (-TagRFPT) for 24 hours. Media were exchanged 8 hours post-transfection to include vehicle, DGAT1 inhibitor, or DGAT2 inhibitor in 500 µl of media. At 24 hours post-transfection (and 12 hours post-treatment), 30 µl of supernatant per well was collected in triplicates in a 96-well plate and incubated with 30 µl of Assay Diluent/Substrate Mix and allowed to incubate for 5 minutes at room temperature in the dark and 15 minutes in the 37°C/5% CO₂ incubator. The 96-well plate was immediately analyzed by spectrophotometry at 490 nm using a SpectraMax Plus 384 microplate reader (Molecular Devices). Spectrophotometry readings had background subtraction and normalized to the vehicle control per APOL1 genotype group and presented as an LDH release fold change relative to control.

Statistical Analyses

The t test was used for Figure 4C. Analysis among three groups, each including at least three biologic replicates, was by one-way ANOVA followed by Tukey’s multiple comparison test. Data are presented as mean±SD with P value as indicated in the figures. GraphPad Prism 8.4.1 was used to calculate statistical significance.

Results

Higher Expression of APOL1 Risk Variants G1 and G2 than of APOL1 G0 Following Treatment of Human Kidney Organoids with IFN-γ

Numerous studies have examined the role of APOL1 in kidney disease using overexpression cell-based models or transgenic mice. Liu et al. recently demonstrated the use of kidney organoids to compare the wild-type APOL1 genotype (G0) with the risk variant genotype G1.²⁸ To better model the effect of APOL1 risk variants and to study differences between endogenous APOL1 wild-type compared with risk variants, we generated iPSCs from human dermal fibroblast samples with genotypes homozygous for G0 (lines 1750-C and 1481-G), G1 (line 2370-C), and G2 (line 1907-D) and confirmed pluripotency with expression of Oct3/4 and Nanog (Supplemental Figure 3A). iPSCs treated with 8 µM CHIR99021 for 4 days expressed the anterior intermediate mesoderm cell marker, GATA3, and the posterior intermediate mesoderm cell marker, HOXD11 (Supplemental Figure 3B). Next, we generated kidney organoids as described by Takasato et al. with minor modifications (Figure 1B).³¹ Kidney organoids differentiated from all three APOL1 genotype iPSC lines formed appropriate glomerular structures (labeled with nephrin for podocytes) and distal tubules (labeled by ECAD, E-cadherin) (Supplemental Figure 4C).

Figure 1.

Generation of isogenic kidney organoids of genotypes APOL1^G2/G2 and APOL1^G0/G0 using CRISPR-Cas9. (A) Experimental design for the generation of isogenic iPSC. CRISPR-Cas9 and an ssODN encoding the NY deletion was used to edit cells and create a precise, biallelic deletion in an APOL1^G0/G0 iPSC line, generating an APOL1^G2/G2 iPSC line on an isogenic background verified by karyotypic analysis. (B) Kidney organoid differentiation protocol with minor modification from previous protocols of Takasato et al.³¹ (C) Brightfield phase-contrast micrographs of representative iPSC and kidney organoids for APOL1^G0/G0 and APOL1^G2/G2. Scale bars, 200 µm for iPSC and 1 mm for kidney organoids. (D) Heat map of the relative transcriptional identity based on scoring from 0 to 1 determined using the KeyGene algorithm³⁸ of APOL1^G0/G0 and APOL1^G2/G2 kidney organoids differentiated to day 28 compared with 13 human fetal tissues. Representative of RNA sequencing from three whole kidney organoids per genotype. (E) Confocal immunofluorescence micrographs of APOL1^G0/G0 and APOL1^G2/G2 human kidney organoids. Human kidney organoids were labeled for nephrin (podocytes; light blue), LTL (LTL fluorescein; green) and CD31 (endothelial cells; red). Scale bar, 50 μm.

To ensure that any differences observed as a function of APOL1 genotype reflected neither APOL1-independent heterogeneity among different iPSC lines nor heterogeneous responsiveness to the kidney organoid differentiation protocol, we used CRISPR-Cas9 genome editing to engineer an iPSC line homozygous for the APOL1 G2 risk allele (APOL1^G2/G2) using a G0/G0 control iPSC line (1750-C; APOL1^G0/G0) (Figure 1A). The G2 variant, defined by an in-frame NY two-codon deletion, was introduced into both APOL1 alleles (Supplemental Figure 1). This allowed us to compare the effects of G0 and G2 APOL1 on an isogenic background while maintaining their native genomic locus under the control of identical, endogenous regulatory elements. Characterization of APOL1^G0/G0 and APOL1^G2/G2 after CRISPR-Cas9 gene editing confirmed expression of all pluripotency markers tested (TRA-1-60, SSEA-3, Nanog, and Oct4) (Supplemental Figure 2). APOL1^G0/G0 and APOL1^G2/G2 kidney organoids differentiated to day 28 had comparable structure/size (Figure 1C), relative transcriptional identity determined using the KeyGene algorithm³⁸ (Figure 1D), and expression of comparable glomerular (nephrin), proximal tubular (LTL), and endothelial (CD31) structures (Figure 1E).

To determine the distribution of endogenous APOL1 protein in isogenic APOL1^G0/G0 and APOL1^G2/G2 kidney organoids, organoids were treated with IFN-γ to upregulate endogenous APOL1 for 24 hours. IFN-γ treated organoids showed markedly higher levels of APOL1 compared with vehicle controls with APOL1 localizing to nephrin-positive glomerular structures and to a lesser degree proximal tubules (LTL) but absent in endothelial cells (CD31) (Figure 2A). Following 24 hours of IFN-γ treatment, APOL1^G0/G0 and APOL1^G2/G2 protein increased 4-fold and 8-fold respectively (Figure 2, B and C). After 4 days of IFN-γ continuous treatment, APOL1^G0/G0 and APOL1^G2/G2 protein further increased 7-fold and 11-fold, respectively (Figure 2, B and C). Upregulation of APOL1^G0/G0 and APOL1^G2/G2 mRNA followed a similar trend to the protein levels of APOL1 (Figure 2D). Although the amounts of APOL1 protein in IFN-γ-treated nonisogenic organoids were not statistically significant as a function of genotype, there was a trend toward higher levels in G1 (1.5-fold) and G2 (2.1-fold) compared with G0 organoids (Supplemental Figure 4B). Taken together, these results demonstrate that APOL1 localizes to glomeruli and proximal tubules of kidney organoids. Moreover, there are APOL1 variant-specific expression differences between APOL1 genotypes in kidney organoid genotypes with higher APOL1 induction for the risk variants when subjected to IFN.

Figure 2.

IFN-γ upregulates APOL1 expression at glomeruli and proximal tubules of kidney organoids. (A) Confocal immunofluorescence micrographs of APOL1^G0/G0 and APOL1^G2/G2 human kidney organoids treated with 100 ng/ml IFN-γ for 24 hours. Human kidney organoids were labeled for APOL1 (red), nephrin (podocytes; light blue), LTL (proximal tubular cells; green), and CD31 (endothelial cells; yellow). Scale bar, 50 μm. (B) Immunoblot of lysates prepared from APOL1^G0/G0 and APOL1^G2/G2 human kidney organoids treated with vehicle or 100 ng/ml IFN-γ for 24 hours and 96 hours, representative of five independent experiments. (C) Quantitation of protein band intensities of lysates (n=3) prepared from isogenic APOL1^G0/G0 and APOL1^G2/G2 kidney organoids after vehicle and after treatment with 100 ng/ml IFN-γ for 24 hours (day 1) or 96 hours (day 4). Bars represent mean±SD; *P<0.05, **P<0.01, ***P<0.001, ns, nonsignificant using one-way ANOVA followed by Tukey’s multiple comparison test. (D) Quantitation of APOL1 RNA expression (n=3) prepared from isogenic APOL1^G0/G0 and APOL1^G2/G2 kidney organoids before (day 0) and after treatment with 100 ng/ml IFN-γ for 24 hours (day 1) or 96 hours (day 4). Bars represent mean±SD; **P<0.01, ****P<0.0001, ns, nonsignificant using one-way ANOVA followed by Tukey’s multiple comparison test.

Fatty Acid Biosynthesis and LD Biogenesis-Associated Genes Are Downregulated in APOL1^G2/G2

To investigate transcriptional differences between APOL1^G0/G0 and APOL1^G2/G2 kidney organoids, we performed bulk RNA sequencing on organoids treated with IFN-γ for 24 hours. Unsupervised principal component analysis showed genotype-specific clustering (Figure 3A). A total of 1683 genes were differentially expressed in APOL1^G2/G2 compared with APOL1^G0/G0 (log2 fold change ≥1; FDR <0.05). The top 40 differentially expressed genes (top 20 upregulated and top 20 downregulated) are shown in Figure 3B. Gene set enrichment analysis demonstrated upregulation of immune pathways, including “induction of antigen presentation by IFN-γ” and “JAK/STAT signaling,” and downregulation of “cholesterol and fatty acid biosynthesis” pathways (Figure 3C). Notable upregulated IFN-γ-stimulated genes included CXCL9, CXCL11, STAT1, and STAT2, which have been previously reported to be associated with APOL1-associated nephropathy (Figure 3D).³⁹

Figure 3.

RNA sequencing of APOL1^G0/G0 and APOL1^G2/G2 kidney organoids reveals differences in cholesterol and fatty acid biosynthesis pathways. (A) Principal component analysis of RNA sequencing samples. (B) The 20 most upregulated and 20 most downregulated coding genes differentially expressed in IFN-γ treated APOL1^G0/G0 and APOL1^G2/G2 organoids. The heatmap was sorted by logFC with gene expression scale normalized by Z-score. (C) Gene set enrichment analysis (Metacore) generated with differentially expressed genes with FDR <0.05. The pathways were considered upregulated or downregulated based on the gene expression patterns within the pathway. The differentially regulated pathways included both upregulated and downregulated genes. (D) Volcano plots highlighting selected genes grouped by biologic process. The horizontal dotted line represents the FDR cutoff of <0.05. The two vertical lines represent log2 fold change ±1. (E) Bioinformatic prediction of protein-protein interactions between APOL1 and differentially expressed lipid metabolism genes using the STRING algorithm. The different colored nodes represent first shell interactions clustered by K-means; the dashed lines connect the three different clusters (green, red, blue). The network edges (solid lines) represent functional associations (not necessarily indicative of binding). The network enrichment P value was <0.001. The top three gene ontology terms were “malonyl-CoA biosynthetic process,” “negative regulation of receptor biosynthetic process,” and “chylomicron remodeling.” (F) GeneMANIA function-specific association network via a binary classification algorithm. Input genes: APOL1, FASN, MGLL, FABP4, FITM1, ACLY, ACSS2, ACSL3, ACSL4, and CPT1. The molecular function-based weighting method predicted the genes represented in the outer circle. The edge colors represent type of association: physical interaction, coexpression, predicted, pathway, or colocalization. The colored nodes indicate predicted biologic functions of each gene node.

We then analyzed lipid metabolism pathways based on the enrichment analysis findings. Genes associated with lipogenesis and lipid metabolism were globally downregulated in APOL1^G2/G2 organoids. The observed downregulation of lipid biosynthetic genes such as ACSL3 and ACSL4 suggest a reduction of LD biogenesis and suppression of glycerolipid metabolism (Figure 3D). Genes associated with LD biogenesis including SCD (encoding stearoyl-CoA desaturase, a mediator of fatty acid desaturation) and FITM1 (encoding fat storage-inducing transmembrane protein 1, a protein involved in LD formation), both functionally related to DGAT1 and DGAT2, were found to be downregulated. LD formation pathway genes DGAT1, DGAT2, and BSCL2 (encoding seipin) were not found to be differentially regulated (Figure 3D). FASN (encoding fatty acid synthase) important for lipogenesis and lipid metabolism was downregulated whereas CPT1 (encoding carnitine palmitoyltransferase 1) was upregulated, indicative of an increase in fatty acid oxidation in APOL1^G2/G2 organoids (Figure 3D). Also notably downregulated was LPL (encoding lipoprotein lipase), associated with receptor-mediated lipoprotein uptake and lipoprotein-associated triglyceride hydrolysis generating free fatty acids, whereas ANGPTL4 (encoding angiopoietin-like protein 4), an endogenous inhibitor of lipoprotein lipase, was upregulated. LMF1 which encodes lipase maturation factor 1, a closely associated protein responsible for maturation and transport of LPL, was also downregulated in the APOL1^G2/G2 organoids (Figure 3D). With respect to cholesterol biosynthesis, the levels of SREBP1 (sterol regulatory element binding protein 1) and SREBP2 (sterol regulatory element binding protein 2) were downregulated in APOL1^G2/G2 organoids compared with APOL1^G0/G0. APOL1^G0/G0 and APOL1^G2/G2 organoids exhibited no differences in expression of cholesterol efflux pathway genes NCEH1, SOAT, and ABCG1, except for upregulation of ABCA1 (a gene encoding ATP-binding cassette transporter A1), which is involved with transcellular cholesterol trafficking.¹¹ Overall, these results indicate a complex network of lipid biosynthesis, metabolism, and transport pathways that are altered in APOL1^G2/G2 kidney organoids.

Next, we analyzed the possible association of APOL1 with the differentially expressed lipid genes by computational analysis with the STRING algorithm, in which a protein-protein interaction network is generated based on an input gene list.³⁵ STRING showed interactions between APOL1 and other lipid metabolism proteins (enrichment P value <1.0e−16) (Figure 3E). K-means clustering identified three distinct protein-protein interaction subnetworks. First-degree interaction of APOL1 was predicted with LPL and APOA1, and second degree with MGLL, LIPE, FABP4, PPARG, and ADIPOQ (Figure 3E, green cluster). We then applied GeneMANIA, a molecular function-based weighting method to generate protein-protein interactions with functional associations related to fatty acid biosynthetic process, glycerolipid metabolic process, acyl-CoA synthesis, lipid particle, and lipid transport. APOL1 was found to be coexpressed with APOL2, APOL3, APOL6, ACSL5, MGLL, and APOA1 (Figure 3F). APOL1 physical interaction was predicted only with APOA1 (Supplemental Figure 5). The overall predicted effects on lipid metabolism pathways indicate decreased fatty acid biosynthesis, increased fatty acid oxidation, decreased LD formation, and potential decrease in lipoprotein uptake.

LD Numbers in APOL1^G2/G2 Organoids Can Be Potentiated by Exposure to Oleic Acid or IFN-γ

We hypothesized that expression of APOL1^G2/G2 could attenuate LD formation, as we previously observed with overexpressed APOL1 risk variants in cultured podocytes.⁴ To understand the relationship between APOL1 and LD distribution, APOL1^G0/G0 and APOL1^G2/G2 kidney organoids were treated with IFN-γ to upregulate APOL1 protein expression and then stained with Nile Red to label LDs (Figure 4A). To ensure that organoids for APOL1^G0/G0 and APOL1^G2/G2 were comparable, we measured the size (area) of glomeruli at the largest diameter for APOL1^G0/G0 and APOL1^G2/G2 organoids. APOL1^G0/G0 and APOL1^G2/G2 organoids had mean sizes of 1927.5 µm² and 1830.5 µm², respectively, which was not statistically different (Figure 4B). Next, comparison of Nile Red-stained LDs in nephrin-labeled glomeruli of IFN-γ-treated APOL1^G0/G0 and APOL1^G2/G2 organoids revealed significantly higher numbers of LDs in the former than in the latter (Figure 4, A and C). Quantitation of Nile Red-positive structures per glomerulus trended approximately 2-fold higher in APOL1^G0/G0 compared with APOL1^G2/G2 with a statistically significant 2.8-fold higher number of LDs after 72 hours of IFN-γ treatment (Figure 4C). As expected, treatment of the organoids with oleic acid for 24 hours also increased glomerular LD numbers (Figure 4, D and E). Taken together, these results show that baseline LD numbers are lower in APOL1^G2/G2 than in APOL1^G0/G0 glomeruli, but LD numbers can be upregulated in both types of isogenic glomeruli in response to oleic acid or to IFN-γ.

Inhibition of DGAT2 Increases LD Number in Kidney Organoids and Podocytes

We previously demonstrated the protective effect of promoting LD formation in the context of APOL1 risk variant overexpression.⁴ DGAT1 and DGAT2 are key enzymes important for the final step in LD formation. DGAT1 inhibition is known to inhibit LD formation thus leading to a reduction in the number of LDs in cells, but some studies have shown that DGAT2 inhibition can lead to a paradoxical increase of LDs.^40–42 Therefore, we hypothesized that DGAT2 inhibition could increase both LD size and number, leading to both recruitment of APOL1^G2/G2 and improved cell survival. To determine the effect of DGAT1 and DGAT2 inhibitors (DGAT1i and DGAT2i) on APOL1 kidney organoids, organoids were treated with IFN to upregulate the LD number with or without DGAT1i or DGAT2i. We found that upregulation of the LD number after 24 hours of IFN-γ treatment of APOL1^G0/G0 organoids was blocked by inhibitors of both DGAT1 (DGAT1i; T863) and DGAT2 (DGAT2i; PF-06424439) (Figure 5, A and B). Following treatment with DGAT1i, there was no statistically significant reduction of LD number for APOL1^G2/G2 organoids (Figure 5, A and B). In the organoids treated with DGAT2i and IFN-γ, APOL1^G2/G2 but not APOL1^G0/G0 organoids led to increased LD number. Neither DGAT1i nor DGAT2i had an effect on APOL1 protein levels (Figure 5C). To determine if DGAT2i could increase LD number in podocytes, we treated primary human podocytes transiently transfected with APOL1-TagRFPT G0, G1, or G2 (hereafter referred to as G0-RFP, G1-RFP, and G2-RFP) with DGAT1i or DGAT2i. Podocytes transfected with G0-RFP and treated with DGAT1i showed a 3.5-fold reduction in LD number (from 38 LD per cell to 11 LD per cell) with accompanying relocalization of G0-RFP to a reticular (ER) pattern (Figure 5, D and E). Podocytes transfected with G1-RFP or G2-RFP and treated with DGAT1i maintained their ER-like distribution, with a reduction of LD number for G1-RFP and no significant change in mean LD number for G2-RFP (Figure 5E). After treatment with DGAT2i, the LD number per cell increased for G0-RFP from 38 to 52, for G1-RFP from 13 to 32, and for G2-RFP from 17 to 33 (Figure 5E). The increased LD number associated with exposure to DGAT2i was accompanied by translocation of APOL1 risk variant polypeptides from ER to LDs. These results demonstrate that APOL1^G2/G2 kidney organoids treated with a DGAT2i can upregulate the production of LDs and enrichment of APOL1 risk variants to LDs.

Figure 5.

DGAT2 inhibition potentiates LD formation and promotes the redistribution of APOL1 risk variants from ER to LDs. (A) Immunofluorescence images and (B) quantitation of three representative experiments for APOL1^G0/G0 and APOL1^G2/G2 kidney organoids treated with vehicle or 100 ng/ml IFN-γ for 24 hours in the presence of DGAT1 inhibitor T863 (20 µM) or DGAT2 inhibitor PF-06424439 (50 µM). Scale bar, 10 μm. Quantitation of LDs per unit glomerular area (10,000 µm²) in sections prepared from APOL1^G0/G0 and APOL1^G2/G2 kidney organoids. Bars represent mean±SD; *P<0.05, ***P<0.001, ****P<0.0001, ns, nonsignificant. (C) Immunoblot of lysates prepared from APOL1^G0/G0 and APOL1^G2/G2 kidney organoid after treatment with vehicle or IFN-γ for 24 hours at the indicated concentrations in the absence or presence of DGAT1 inhibitor T863 (20 µM) or DGAT2 inhibitor PF-06424439 (50 µM). Representative of three independent experiments. (D) Fluorescence micrographs of human primary podocytes transiently transfected with APOL1-RFP and treated for 16 hours with vehicle (DMSO), DGAT1 inhibitor T863 (DGATi; 10 µM), or DGAT2 inhibitor PF-06424439 (DGAT2i; 20 µM). LDs were labeled with BODIPY 493/503 (green) and nuclei with DAPI. Scale bar, 10 µm. (E) Number of LDs per podocyte in APOL1-RFP-expressing human primary podocytes. Representative data from three independent experiments, quantification of at least 25 cells per group, are presented as mean±SD; *P<0.05, ***P<0.001, ****P<0.0001, ns, nonsignificant.

Redistribution of APOL1 Risk Variants to Lipid Droplets Is Associated with Reduced Cell Toxicity

Based on our previous observation demonstrating a cytoprotective effect of APOL1 recruitment to LDs, we evaluated the effects of DGAT1i and DGAT2i on APOL1-mediated cytotoxicity. We hypothesized that inhibition of LD formation by DGAT1i would increase APOL1 toxicity, whereas the increased LD number after treatment with the DGAT2i would be associated with lower APOL1 toxicity. HEK T-REx-293 cell lines stably expressing tetracycline-inducible APOL1 were subjected to treatment with either of the DGAT inhibitors. DGAT1i treatment (10 µM) of APOL1-expressing HEK T-REx-293 cells decreased the cell number for all genotypes and rendered the cells with spindle-like morphology, with effects more pronounced at 20 µM (Figure 6A). By contrast, DGAT2i produced concentration-dependent rescue of the reduced cell number seen with the APOL1 risk variants (Figure 6, A and B). DGAT2i treatment at 50 µM resulted in approximately 2-fold higher cell number per visual field for the G1 and G2 lines (Figure 6B). Next, we used a cytotoxicity assay to determine if the DGAT2i treatment-induced increase in LD number is associated with reduced cell death. Consistent with the observed DGAT2i treatment-associated increase in cell number, DGAT2i exhibited dose-dependent reduction of cytotoxicity for cells expressing G1 and G2 (Figure 6C). In contrast, DGAT1i increased cytotoxicity for cells expressing either G1 or G2 (Figure 6C). To confirm DGAT2i-associated reduction of cytotoxicity in podocytes, primary podocytes were transiently transfected with RFP or APOL1-RFP. Expression of DGAT1 and DGAT2 protein levels were not altered in podocytes transfected with RFP or APOL1-RFP (G0 and G2) (Figure 6D). Next, podocytes were transfected with G0-RFP, G1-RFP, and G2-RFP for 24 hours followed by treatment with vehicle, DGAT1i or DGAT2i (Figure 6E). Supernatant collected from G0-, G1- and G2-RFP transfected podocytes quantified for cell death using an LDH release assay showed that cytotoxicity was higher for G1-RFP (1.4-fold) and G2-RFP (2-fold) compared with G0-RFP (Figure 6F). Similar to the HEK 293 cells, DGAT2i reduced cytotoxicity for both G1-RFP and G2-RFP transfected podocytes, consistent with the pharmacological inhibition of DGAT2 for G1 and G2 in HEK cells (Figure 6G). Interestingly, treatment with DGAT1i also yielded a statistically significant reduction of cell death in G2-RFP transfected cells, suggesting either cell-type differences in LD regulation and/or sensitivity to the level of APOL1 expression (Figure 6F). Taken together, we provide evidence that DGAT2i can reduce cytotoxicity, perhaps by leading to a greater quantity of LDs per cell and/or sequestration of APOL1 to the LDs.

Figure 6.

The DGAT2 inhibitor PF-06424439 can reduce APOL1 risk variant-associated cytotoxicity. (A) Brightfield phase-contrast micrographs of representative APOL1 T-REx-293 cells stably expressing empty vector (EV), G0, G1, and G2 treated with vehicle (methanol), DGAT1 inhibitor T863 (20 µM), or DGAT2 inhibitor PF-06424439 (20 µM). Cells were induced 18 hours with tetracycline (10 ng/ml) to express APOL1. Scale bar, 50 µm. (B) Quantitation of T-REx-293 cells per visual field stably expressing an empty vector, or APOL1 variants G0, G1, or G2 (top to bottom), and treated either with vehicle (DMSO), DGAT1 inhibitor T863 (DGAT1i; 10 and 20 µM), or DGAT2 inhibitor PF-06424439 (DGAT2i; 20 µM and 50 µM). APOL1 expression was induced with tetracycline (10 ng/ml) for 18 hours. Bars represent mean±SD; **P<0.01, ****P<0.0001, ns, nonsignificant. (C) Cell cytotoxicity/viability assay after 18-hour treatment with tetracycline (10 ng/ml) for cells expressing EV, G0, G1, and G2 (top to bottom) in the presence either of vehicle (Veh; DMSO), DGAT1 inhibitor T863 (DGAT1i; 10 µM and 50 µM), or DGAT2 inhibitor PF-06424439 (DGAT2i; 10 µM and 50 µM). Data from two independent experiments, each with eight experimental replicates. Bars represent mean±SD; ****P<0.0001, ns, nonsignificant. (D) Lysates prepared from podocytes (Celprogen) transiently transfected with RFP or APOL1-RFP (G0 or G2) and immunoblotted for DGAT1 and DGAT2. (E) Immunoblot of podocyte lysates representative of three separate experiments transiently transfected with APOL1-RFP (G0, G1, or G2) and treated with vehicle, DGAT1 inhibitor T863 (DGAT1i; 20 µM), or DGAT2 inhibitor PF-06424439 (DGAT2i; 50 µM). (F) LDH assay of media collected from podocytes transiently transfected with APOL1-RFP (G0, G1, or G2). Quantitation of three separate experiments. **P<0.01, ***P<0.001, ****P<0.0001, ns, nonsignificant. (G) LDH assay of media collected from podocytes transiently transfected with APOL1-RFP (G0, G1, or G2) treated with vehicle, DGAT1 inhibitor, or DGAT2 inhibitor. Representative of three independent experiments performed in duplicate. **P<0.01, ns, nonsignificant.

Discussion

Our study highlights a relationship between APOL1 and lipid regulation, with results suggesting bidirectional effects of APOL1 on lipid transport and biosynthesis: APOL1 risk variants downregulate LD biogenesis and lipid metabolism by downregulating genes involved in lipogenesis and upregulating genes associate with fatty acid oxidation (Figure 7). Targeting fatty acid metabolism and LD biosynthesis by DGAT2 inhibition may reduce APOL1-mediated cell toxicity by shunting the APOL1 protein away from ER to LDs.

Figure 7.

Regulation of genes involved in lipid biosynthesis and metabolism for APOL1^G2/G2 after IFN-γ treatment. Downregulated genes are shown in blue and upregulated genes in red. (1) Lipogenesis: fatty acid synthesis-related genes are downregulated in APOL1^G2/G2 kidney organoids (IFN-γ treated) compared with APOL1^G0/G0 (IFN-γ treated), including the stearoyl-CoA desaturase gene SCD, an important regulator of lipid synthesis. (2) Fatty acid oxidation: carnitine palmitoyl transferase 1 (CPT1) catalyzes the regulated, rate-limiting step in mitochondrial fatty acid oxidation. CPT1 is upregulated in APOL1^G2/G2 kidney organoids, suggesting increased fatty acid oxidation.⁴⁶ (3) LD biogenesis: in the ER (purple lipid bilayer) SCD interacts with DGAT2, playing an important role in triglyceride synthesis and LD formation. Storage-inducing transmembrane (FIT) proteins including FITM1 contribute to LD biogenesis by facilitating LD budding; FITM1 was downregulated in APOL1^G2/G2 organoids. Inhibition of DGAT1 reduces LD number whereas DGAT2 inhibition potentiated LD formation and promoted APOL1^G2/G2 binding to LD. At the LD surface level, the fatty acid binding protein FABP4, an important protein in lipid trafficking, can form a complex with hormone sensitive lipase/lipase E (encoded by LIPE), is upregulated in APOL1^G2/G2 organoids, and may facilitate lipolysis.

APOL1 risk variants expressed in experimental systems can lead to cytotoxicity when concentrated in the ER by various possible pathways, including lipotoxicity, autophagy, ER stress, inflammasome activation, and mitochondrial damage.^{4,14,28,43,44} We speculate that ER accumulation of APOL1 risk variants contributes to multiple cytotoxic cellular pathways, including ER stress, autophagy, and downregulation of lipid metabolism and trafficking, and that transferring APOL1 risk variants to LDs attenuates these cytotoxic pathways. Here we have shown that DGAT1 and 2, key enzymes in the final step of triglyceride formation, regulate the localization and cytotoxic effects of APOL1 risk variants. Our work highlights a key difference between DGAT1 and DGAT2 inhibition that was observed in prior studies: unlike inhibition of DGAT1, chemical inhibition of DGAT2 using PF-06424439 (used in this study) or JNJ-DGAT2-A results in increased LD size and number.^40,41 We observe that DGAT2 inhibition increased the number of LDs in kidney cells which prompted us to consider DGAT2 inhibitor as a therapy to reduce APOL1 risk variant-associated cytotoxicity. Consistent with our prior study, the increased LD number associated with DGAT2 inhibition, but not DGAT1 inhibition, improved survival in cells expressing APOL1 risk variants. DGAT2 inhibitors may thus be of therapeutic benefit in slowing progression of kidney disease in patients with APOL1 renal risk variants. Mouse Dgat2 knockout display a lethal phenotype, so total inhibition of DGAT2 would not be a viable therapy. We look forward to exploring opportunities to develop new therapies to inhibitor or knock down DGAT2 or repurpose DGAT2 inhibitors (e.g., PF-06865571 and NCT04044053) currently being used in clinical trials for nonalcoholic steatohepatitis. We note that we cannot determine if protective effect of DGAT2 inhibition is mediated by increased numbers of LDs, increased trafficking of APOL1 to LDs, or both.

Our data demonstrated the downregulation of numerous genes involved in lipid metabolism and biosynthesis APOL1^G2/G2 kidney organoids (Figure 3 and summarized in Figure 7). Fatty acid synthesis-related genes are downregulated in APOL1^G2/G2 organoids compared with APOL1^G0/G0 treated with IFN-γ for 24 hours, including the gene SCD, an important regulator of lipid synthesis.⁴⁵ SCD also interacts with DGAT2 in the ER, playing an important role in triglyceride synthesis and LD formation.⁴⁶ FITM1 (encoding fat storage-inducing transmembrane protein 1) was downregulated in APOL1^G2/G2 organoids and is a gene associated with storage-inducing transmembrane (FIT) proteins which also contribute to LD biogenesis by facilitating LD budding.⁵ The fatty acid binding protein FABP4, an important protein in lipid trafficking, forms a complex with hormone sensitive lipase/lipase E (encoded by LIPE) on the LD surface to facilitate lipolysis.⁴⁷ By contrast, carnitine palmitoyl transferase 1 (CPT1), upregulated in APOL1^G2/G2 kidney organoids, would further reduce the LD number by catalyzing the regulated, rate-limiting step in mitochondrial fatty acid oxidation.⁴⁸ Overall, lipid biosynthesis and metabolism appears to be an important area for further exploration for APOL1-associated kidney disease and likely other chronic kidney diseases. LMF1, among the top 20 downregulated genes in APOL1^G2/G2 kidney organoids, is important for the maturation of lipoprotein lipase, but the intracellular role of LMF1 remains poorly characterized. ANGPTL4 RNA expression was upregulated in APOL1^G2/G2 compared with APOL1^G0/G0 kidney organoids, suggesting a pathway that may be tied to APOL1^G2/G2 expression. Interestingly, angiopoietin-like protein 4 (encoded by ANGPTL4) induces hyperlipidemia and hepatic steatosis in mice.⁴⁹ Choi et al. showed that antisense oligonucleotides to DGAT2 in a rat model of diet-induced nonalcoholic fatty liver disease may help to prevent hepatic steatosis.⁵⁰ They reported that knocking down DGAT2 could paradoxically lower diacylglycerol content by reducing SREBP1c-mediated lipogenesis and increasing hepatic fatty oxidation. Similarly, we propose that APOL1^G2/G2 kidney organoids which had reduced lipogenic gene expression (e.g., SREBP1, SCD) and increased expression of oxidative/thermogenic genes (e.g., CPT1) can be important contributors to the reduced LDs observed in APOL1^G2/G2 organoids. The downregulation of genes involved in lipogenesis and monounsaturated fatty acid biosynthesis (ACSS2, FASN, ACSL3, ACSL4, and SCD), the upregulation of CPT1 (a gene involved in fatty acid oxidation), and LD biogenesis (FITM1) may contribute to the low fatty acid substrates required for the synthesis of triglyceride or LD formation in the APOL1 risk variants (Figure 7). It remains to be determined how expression of APOL1 risk variants leads to a lower LD number, but certainly careful assessment of a complex interplay of pathways that can affect LD biogenesis and metabolism will be required. Future investigations will examine perturbation of lipase activity resulting in accumulation of free fatty acids that cannot be converted to neutral cholesterol esters and triglycerides stored in LDs.

Our study was limited to one set of isogenic iPSC lines for G0 and G2 and would benefit from validation using additional iPSC lines expressing APOL1 G0 and APOL1 risk variants on isogenic backgrounds. Because our template iPSC line for APOL1^G0/G0 was of mixed haplotype at residue 150 (E/K), it would be prudent to generate an isogenic background starting from the G1 and/or G2 haplotypes homozygous for glutamate at position E150 which has been reported to have higher toxicity.⁵¹ Human kidney organoids of G0 and G1 genotypes generated from iPSC by CRISPR-Cas9 genetic editing were recently demonstrated to model some APOL1-mediated transcriptional effects.²⁸ In our kidney organoids, APOL1 was detectable by immunoblotting in the unstimulated state with the Sigma anti-APOL1 antibody (Cat. #HPA018885) but not with the Abcam anti-APOL1 antibody (Cat. #ab252218). By contrast, APOL1 was consistently detectable by immunofluorescence using the Abcam anti-APOL1 (Supplemental Figure 6) and certain lots of Sigma anti-APOL1 antibody (HPA018885 lot #G115599; Supplemental Figure 4C). Furthermore, in contrast to Liu et al., who observed no difference in levels of APOL1 G1 and G0, APOL1 expression in our kidney organoid preparations was higher for APOL1 risk variant protein compared with wild-type (Figure 2 and Supplemental Figure 4).²⁸ Our analysis of kidney organoids using certain lots of anti-APOL1 antibody obtained from Sigma from 2017 to 2018 showed specificity of APOL1 localization by immunofluorescence to nephrin-positive glomerular-like and tubular structures (Supplemental Figure 4C). Subsequent lots of this anti-APOL1 antibody yielded nonspecific nuclear staining patterns in our hands. We confirmed APOL1 localization to podocyte and LTL-positive proximal tubular structures, although this was undetectable in CD31-positive regions using an anti-APOL1 antibody from Abcam (Figure 2A). The APOL1 staining pattern in kidney organoids are consistent with the detection of APOL1 protein and RNA in podocytes and proximal tubules in human kidney tissue reported by Madhavan et al.⁵² However, recent work by Blessing et al. could not detect APOL1 at proximal tubules of healthy kidney tissues.⁵³ It remains unclear if kidney tissue from an APOL1 patient will have higher expression that may be detectable by immunohistochemistry. Moreover, a fully developed kidney organoid with vascularization and mature cells may have differences in APOL1 protein expression. Future studies will require higher resolution microscopy to identify intracellular APOL1 expression and localization using other anti-APOL1 antibodies. We anticipate an improved characterization of endogenous, intracellular APOL1 as newer APOL1-specific antibodies become more widely available.^54,55 Future work to understand the mechanistic details of APOL1-LD interactions and their bidirectional regulation should benefit from the kidney organoid model.

Disclosures

S. Alper reports consultancy agreements with the Medical University of Vienna and the Swiss National Science Foundation; research funding from Quest Diagnostics; honoraria from the Swiss National Science Foundation; and scientific advisor or membership with the Swiss National Science Foundation. J. Chun reports scientific advisor or membership with the Canadian Journal of Kidney Health and Disease; and support from an Alberta Innovates Health Solutions Clinician Fellowship and the KRESCENT program as a postdoctoral fellow and new investigator. D. Friedman reports consultancy agreements with Vertex Pharmaceuticals; ownership interest with Apolo1bio; research funding from Vertex Pharmaceuticals; and co-inventor on patents related to APOL1 diagnostics and therapeutics. J. Magraner reports ownership interest with ADT, Apple, Cronos Group, GameStop, Triterras, and United Airlines. M. Pollak reports consultancy with Vertex Pharmaceuticals; ownership interest with Apolo1bio; research funding from Vertex Pharmaceuticals; co-inventor on patents related to APOL1 diagnostics and therapeutics; and honoraria from academic talks; and member of NephCure Foundation scientific advisory board. S. Shah reports ownership interest with Analog Devices and Chroma Medicine; and research support from Vertex Pharmaceuticals. J. Zhang reports current employment with Frontera Therapeutics. Because M. Pollak is an editor of JASN, he was not involved in the peer review process for this manuscript. A guest editor oversaw the peer review and decision-making process for this manuscript. All remaining authors have nothing to disclose.

Funding

This work was supported by the National Institutes of Health (grant MD014726), Vertex Pharmaceuticals, and the Lawrence Ellison Foundation.

Author Contributions

S. Alper, J. Chun, D. Friedman, and M. Pollak conceptualized the study; J. Chun and C. Riella were responsible for data curation; J. Chun, H. Chung, J. Magraner, S. Shah, and J.-Y. Zhang were responsible for investigation; J. Chun was responsible for methodology and wrote the original draft; J. Chun, G. Ribas, H. Ribas, and C. Riella were responsible for visualization; G. Ribas, H. Ribas, C. Riella, and M. Wang were responsible for formal analysis; D. Friedman and M. Pollak were responsible funding acquisition and project administration; S. Alper, D. Friedman, and M. Pollak were responsible for supervision; and S. Alper, J. Chun, D. Friedman, M. Pollak, and C. Riella reviewed and edited the manuscript.

Data Sharing Statement

The sequencing data are available for download at http://www.ncbi.nlm.nih.gov/bioproject/795435 (Accession Number PRJ NA795435).

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021050723/-/DCSupplemental.

Supplemental Figure 1. Analysis of isogenic APOL1^G0/G0 and APOL1^G2/G2 iPSC lines derived from CRISPR-Cas9 genome editing.

Supplemental Figure 2. Characterization of APOL1^G0/G0 and APOL1^G2/G2 iPSC pluripotency in iPSC lines derived from the 1750-C isogenic background.

Supplemental Figure 3. Characterization of human iPSC lines expressing markers of pluripotency and the intermediate mesoderm for APOL1 genotype lines.

Supplemental Figure 4. Induced expression of APOL1 in human kidney organoids treated with IFN-γ.

Supplemental Figure 5. Predicted interactions with APOL1.

Supplemental Figure 6. Validation of APOL1 staining in kidney organoids.