Organoid culture, TNFα treatment and sample preparation. UM77-2 human embryonic stem cells (NIH registration #0278) were cultured in accordance with University of Michigan’s Human Pluripotent Stem Cell Research Oversight Committee and NIH regulations, and kidney organoids were generated as previously described 18. Organoid cultures (whole wells of cells, isolated spheroids and/or overlying culture medium) were collected at indicated time points: days 21–29 from seeding. For TNFα experiments, almost all samples were collected on D25 following treatment with 5 ng/ml recombinant human TNF-alpha (R&D Systems, cat#10291TA050) reconstituted in DPBS (Gibco, cat#14190144; vehicle control) for 24 or 48 h (medium replenished every 24 h) before collection. Samples collected for scRNA-seq were collected on D24 following 24h TNFa treatment. For proteomics studies, isolated organoids (20 organoids/sample in triplicate) were harvested and washed with ice-cold PBS, and organoid culture medium was collected and snap frozen. For ELISA studies, whole well lysates were prepared by washing the organoids in ice-cold PBS and scraping into Cell Lysis Buffer 2 (R&D Systems, cat#895347) supplemented with Halt Protease and Phosphatase inhibitor Cocktail (Thermo, cat#78440) and medium from corresponding wells was collected and snap frozen. For qRT-PCR/Bulk RNA-seq, whole wells or isolated spheroids were washed, scraped into ice-cold PBS, pelleted and lysed in 500 µL of TRIzol (Invitrogen, cat#15596026) and then total RNA was extracted using Directzol RNA Mini Prep Plus kit (Zymo Research, cat# R2072), as per manufacturer’s instructions.
Organoid immunofluorescence. Organoid spheroids were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, cat#15710), subjected to a sequential gradient of sucrose, and embedded in 20% sucrose/OCT at ratio 2:1 (Tissue-Plus, ThermoFisher Scientific, cat#4585). 5µm cryosections were rehydrated and blocked with 5% normal donkey serum (Jackson ImmunoResearch Laboratories, cat#017-000-121) in PBS supplemented with 0.1% Triton X-100 (IBI Scientific, cat#IB07100). Slides were immunostained with primary antibodies in 3% BSA (Fraction V, Gibco, cat#15260-037) followed by appropriate Alexa Fluor secondary antibodies (Invitrogen) and mounted using Prolong Gold with DAPI (Invitrogen, cat#P36935). Samples were imaged using a Nikon A1 High Sensitivity Confocal Microscope at the University of Michigan’s Microscopy Core and processed with Nikon Elements software. Primary antibodies: N-Cadherin (R&D, cat#AF6426, 1:1000); ACTA2 (R&D, cat#MAB1420-SP, 1:50); PDGFRA (BD Biosciences, cat#556001, 1:200); Synaptopodin (Progen, cat#690094S, 1:80); NPHS1 (R&D, cat#AF4269, 1:500); TNFΑRSF1A (R&D, cat#MAB225SP, 1:20); VCAM1 (Invitrogen, cat#MA5-11447, 1:50).
Organoid ELISA. Protein levels in kidney organoid culture media and lysates were measured using commercially available ELISA kits for IP-10 (R&D Systems, cat#DIP100), MCP-1 (R&D Systems, cat# DCP00), TIMP-1 (R&D Systems, cat# DTM100), C3 (Abcam, cat#ab108823) and VCAM1 (R&D, cat# DVC00). Samples were processed in duplicate following the manufacturer’s protocol. Values were normalized to total protein content using Pierce BCA protein assay (Thermo Scientific, cat# 23227). Three independent experiments were performed and plotted using GraphPad Prism software. Statistical significance was calculated using Student's t Test.
Organoid qRT-PCR. Total RNA was collected from whole well organoid cultures. One µg of total RNA was reverse-transcribed into cDNA using SuperScript First-Strand Synthesis kit (Invitrogen, cat# 11904018) per manufacturer’s protocol. Quantitative real-time PCR (qRT-PCR) was performed using TaqMan Fast Universal PCR Master Mix (2X) (Applied Biosystems, cat# 4352042) in a QuantStudio 7 Flex Real-Time PCR System (ThermoFisher) using TaqMan Pre-Developed Assay Reagents (PDARs) as follows: C3, Hs00163811_m1; CXCL10, Hs00171042_m1; GAPDH, Hs03929097_g1; VCAM1, Hs01003372_m1. The ΔΔCq method 46was applied to calculate the relative quantity (RQ) of target gene after normalization to GAPDH. Samples were assayed in duplicate. Graphs of three independent experiments were plotted using GraphPad Prism software. Statistical significance was calculated using Student's t-test.
Organoid bulk RNA-seq and bioinformatic analysis.
Organoid bulk RNA-seq and bioinformatic analysis. Total RNA from isolated organoid spheroids was prepared as described above. Library preparation using NEBNext Ultra II kit and sequencing using paired end read length of 150 bases on a NovaSeq4000 were performed by the University of Michigan Advanced Genomics Core. Fastq read quality was determined using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and reads aligned to the reference (ENSEMBL GRCh38.104) using STAR 2.7.8a 47. Uniquely mapped reads were inspected for unusual distribution across known annotated features using Picard Tools (https://broadinstitute.github.io/picard/). Gene level read counts were generated using HTSeq (version0.12.4) 48 and normalized with voom 49. PCA and hierarchical clustering were used to identify and remove samples with abnormal expression profiles due to technical issues, and the mapping statistics obtained from STAR. The data are submitted to GEO (pending accession ID: GSExxxxx).
Organoid scRNA-seq and bioinformatic analysis. Whole well organoid cultures were harvested and dissociated as previously described 18. Single cell suspensions were submitted to the Advanced Genomics Core at the University of Michigan for library preparation and sequencing on a 10x Genomics Chromium System. Organoid scRNA-seq data processing was performed using Seurat 4.0 50. Cells expressing > 500 genes were included in the analysis. The processing steps include log transformation, scaling or linear transformation using default settings, highly variable gene identification, dimensionality reduction using principal component reduction (PCA) and Uniform Manifold Approximation and Projection (UMAP), batch correction using harmony function embedded in Seurat and unsupervised clustering at 0.25 resolution. Data deposited here: http://18.188.163.197/
Proteomics sample preparation, organoids trajectory. Cell pellets (20 isolated organoid spheroids/sample in triplicate) were lysed using urea buffer containing urea (8M) and ammonium bicarbonate (100 mM) supplemented with 1X PPI (Thermo Scientific Halt Protease and Phosphatase Inhibitor Cocktail). Protein lysates were sonicated for 30 s on 10% power. After centrifugation at 1,400 rpm for 30 min at 4°C, the supernatant was transferred to a new tube. Protein concentrations were measured using a commercial BCA kit (Thermo Scientific, Waltham, MA). The samples were incubated with DTT (10 mM) followed by iodacetamide (40 mM) for 1 h at room temperature for the reduction and alkylation of disulfide bonds. Protein from each sample was digested with trypsin using a 1:100 ratio (1 µg trypsin per 100 µg protein). Digestion was stopped the next day by acidifying to pH 2–3 using formic acid.
Proteomics analysis, organoids trajectory. For time-course analysis of organoid cell pellets, peptides were purified using in-house made stage-tips. For nLC-MS/MS analysis of proteomic data, we used a Q Exactive Plus (Thermo) instrument coupled to a nLC, with a 2.5-h gradient. A binary buffer system with buffer A: 0.1% formic acid (FA) and Buffer B 80% Acetonitrile (ACN) was used. The flow rate was 250 nl/min. The gradient settings were as follows t = 0 min; 4% (Buffer B), 05 min, 6%; 125 min, 23%; 132 min, 54%; 138 min, 85%; 143 min, 85% and 145 min 5%. The flow rate was constant with 250 nL/min. The Q Exactive Plus was operated in positive ion mode. One survey scan (resolution = 70000, m/z 300–1750) was followed by up to 10 MS2 scans (resolution = 17500, m/z 200–2000). Dynamic exclusion was enabled (20 s). AGC target was 3e6 for MS1 scans, and 5e5 for MS2 scans. MS data was processed using MaxQuant as described below. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 40 partner repository with the dataset identifiers. Project Name: LC MS/MS of kidney organoids during differentiation from day 21 to 29 Project accession: PXD029716 Project DOI: Not applicable Reviewer account details: Username: [email protected], Password: y2wNauaO
Integration of organoid trajectory proteome and transcriptome. The proteomic data analysis was performed using the Perseus software suite 51 and R 52 for differential expression analysis, GO-term analysis and plotting heatmaps. For trajectory analysis, raw LFQ data were log2 transformed. Missing data was accepted at a threshold of 33% across samples, with subsequent imputation of missing data. Imputation was carried out sample wise with a width of 0.3 SD and a downshift of 1.8 SD as implemented in the Perseus software. Groups were compared using t-tests adjusted for multiple comparisons (permutation-based FDR 0.05), as implemented in the Perseus software. Heatmaps and clustering were carried out in Perseus or in R software via the complexHeatmap package32 using mean-subtracted data and maximum distance clustering. GO-term annotation of proteins and GO-term enrichment of the heatmap clusters was carried out in Perseus and enrichment assessed by Fisher’s exact test. Volcano plots for relevant comparisons were also generated in Perseus. Correlation analysis between protein and bulk RNA transcript was carried out in Perseus. 2D GO-enrichment was also carried out using bulk RNA-seq and bulk proteomics data, with enrichment terms plotted against each other33. For integration with scRNA-seq, cell type marker genes from organoid scRNA-seq were used to map protein dynamics indicative of cell type.
Proteomic analysis of TNFa-treated organoids. Cell pellets (20 isolated organoid spheroids /sample in triplicate) were lysed using 1:1 4% SDS/0.1M HEPES pH 7.4/5mM EDTA, complemented with protease inhibitor cocktail (Roche) and denaturation at 95°C for 5min. 10 mM TCEP and 50 mM CAA were used for reduction/alkylation of the samples. 50µg aliquots were purified with paramagnetic, mixed 1:1 hydrophobic:hydrophilic SP3 beads 53. Purified proteins were resuspended in 50 mM HEPES, pH 7.4 and digested over night at 37°C with trypsin (Serva) in a 1:100 (w/w) ratio. Samples were acidified with 2% formic acid and peptides were purified using in-house made stage-tips. Peptides were separated on an Ultimate3000 RSLC nanoHPLC coupled on-line to an Exploris480 orbitrap tandem mass spectrometer (Thermo). The HPLC was operated in a two-column setup with an Acclaim 5mm C18 cartridge pre-column (Thermo) and an ionopticks aurora 25cm column with integrated emitter tip. Separation was performed at 400 nL/min in a heated column oven at 50°C (Sonation) with the following gradient of solvents A (H2O + 0.1% FA) and B (ACN + 0.1% FA): 120 min from 2–30% B and a high-organic washout at 90% B for 9 min followed by a re-equilibration to the starting conditions (2% B). The mass spectrometer was operated with the FAIMS device at standard resolution with a total carrier glas flow of 3.8 L/min at three CVs: -40, -55 and − 75V. The Orbitrap resolution for the MS1 full scan was set to 120k, whereas the MS2 scans were recorded with 1.5s cycle time for − 40V CV and 0.75s cycle time for − 55/-70V FAIMS CVs at an orbitrap resolution of 15k. Dynamic exclusion mode was set to custom with a 40s exclusion window and a mass tolerance of 10 ppm each.
To assess the organoid secretome, proteins within the overlying cull culture medium of organoid spheroids were denatured using 1:1 4% SDS/0.1M HEPES pH 7.4/5mM EDTA, complemented with protease inhibitor cocktail (Roche) and heating at 95°C for 5 minutes. 10 mM TCEP and 50 mM CAA were used for reduction/alkylation of the samples. 50 µg aliquots were purified with paramagnetic, mixed 1:1 hydrophobic:hydrophilic SP3 beads 35. Purified proteins were resuspended in 50 mM HEPES, pH 7.4 and digested over night at 37°C with trypsin (Serva) in a 1:100 (w/w) ratio. Samples were acidified with 2% formic acid. All peptides were purified using in-house made stage-tips. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 40 partner repository with the dataset identifiers. Project Name 1 (organoid spheroid cell lysate): Proteome analysis of kidney organoid cells during TNFα stimulation Project accession: PXD029718 Project DOI: 10.6019/PXD029718 Reviewer account details: Username: [email protected] Password: tdei7cSf. Project Name 2 (secretome): Proteomic analysis of kidney organoid supernatant during TNFα stimulation Project accession: PXD029696 Project DOI: 10.6019/PXD029696 Reviewer account details: Username: [email protected] Password: jKHFtF9T
Organoid proteome data processing. For organoid trajectory samples, raw files were searched, quantified and normalized using the MaxQuant 29 version 1.5.3.8 (FDR = 1%). Label-free quantification 29, intensity based absolute quantification (iBAQ) with log fit, and the match-between-runs feature were enabled. We used the UniProt human reference proteome as database (downloaded in January 2017) and default settings for orbitraps. Enzyme specificity was set to Trypsin/P, cysteine carbamidomethylation was set as a fixed modification (+ 57.021464) and methionine oxidation (+ 15.994914) as well as protein N-terminal acetylation (+ 42.010565) were set as variable modifications. Data analysis was performed using Perseus software suite (V1.5.1.6). For organoid TNFα experimental samples, raw FAIMS data were converted into MzXML files with the FAIMS_MzXML_Generator tool (v1.0.7639, 36) and queried with MaxQuant v 1.6.7.0 (FDR = 1%, match between runs = on) using the UniProt reference proteome database for human (May 2020, canonical only, 20600 entries) and default settings for orbitrap instruments. Enzyme specificity was set to Trypsin/P, cysteine carbamidomethylation was set as a fixed modification (+ 57.021464) and methionine oxidation (+ 15.994914) as well as protein N-terminal acetylation (+ 42.010565) were set as variable modifications. The match-between-runs feature was activated with default settings.
Organoid proteome data analysis. The data analysis was performed using the Perseus software suite and R for GO-term analysis. For organoid trajectory sample analysis and for GO-term annotation, raw LFQ data were log2 transformed, triplicates were averaged, and proteins detected on all four days of differentiation were used. For volcano plotting and t-tests individual samples were used with missing data accepted at a threshold of 33% across samples, with subsequent imputation of missing data. Imputation was carried out sample wise with a width of 0.3 SD and a downshift of 1.8 SD as implemented in the Perseus software. GO terms were annotated and between group testing was carried out using Student’s t-test and Volcano plots for relevant comparisons. For organoid TNFα experimental samples, data were log2 transformed and GO-terms annotated, data were filtered to only include proteins for which we had complete data in 3 of 3 replicates within at least one group with subsequent imputation of missing data. Imputation was carried out sample wise with a width of 0.3 SD and a downshift of 1.8 SD. The FDR for Volcano plots was held at 0.05 for the trajectory analysis, at 0.1 for TNFα treated organoids. For GO term enrichment we used Fisher Exact testing implemented in Perseus (FDR 0.05) or the Enrichr R package (adjusted p-value < 0.05 as significant). Top GO terms per annotation term category were then used for plotting. Heatmaps and clustering were carried out in Perseus using mean-subtracted data and maximum distance clustering or using the R package complexHeatmap 32. Venn diagrams were generated using gene symbols of the relevant data sets and produced using the Rvennr package. GO-term annotation of intersecting and unique Venn sections was performed using EnrichR. 2D scattering of RNA count vs protein IBAQ and keyword analysis was carried out in Perseus software with visualization of 2xSD in R. UMAP and marker lists of scRNA-seq were produced using the Seurat R package. Matrisome networks were generated from proteins in our dataset annotated with matrisome as a uniprot keyword. A list of matrisome associated identifiers was uploaded to string-db.org (V.11.0) with standard settings and ’highest confidence’ selected for the minimum required interaction score. The STRING database54 was used to generate the network which was then imported into Cytoscape (V 3.8.2)55 and log2 fold changes mapped to the network indicating magnitude of the fold change through size and directionality through color. To assess to which degree organoids express disease relevant proteins, we mapped proteins between OMIM disease genes and organoids.
Human podocyte culture. Human podocytes56 were cultured in dishes as previously described and regularly tested for mycoplasma using a commercial kit (LookOut, Sigma). 50,000 cells were seeded in 6-well dishes (Thermo Scientific) and cultured at 32°C with 5% CO2 in RPMI 1640 (Gibco, Thermo Scientific) containing 10% FBS (Gibco, Thermo Scientific), 1% Penicillin-Streptomycin (Gibco, Thermo Scientific), 1% insulin-transferrin-sodium selenite (Thermo Scientific), 1% MEM (Gibco, Thermo Scientific), 1 mM Sodiumpyruvate (Gibco, Thermo Scientific) and 20 mM HEPES (Gibco, Thermo Scientific). 24 h after seeding, cells were washed once with PBS and medium was replaced with FBS-free medium containing 5 ng/mL TNFα (R&D Systems, same vendor as for organoid treatment) or vehicle control. After 24 or 48 h, culture medium was removed and snap frozen. Cells were washed twice with PBS and scraped into ice-cold urea buffer (8M urea, 100 mM ammonium bicarbonate, 1X PPI), snap frozen and stored until analysis.
Analysis of the human cultured podocyte proteome. Cells and supernatants were analyzed using MaxQuant and Perseus. Raw files were searched, quantified, and normalized using MaxQuant version 1.6.17.0 with default settings for orbitraps. The match between runs (MBR), LFQ, IBAQ and classical normalization features were enabled. We used the UniProt human reference proteome as database (UP000005640_9606, downloaded in April 2021 with 20612 entries with enzyme specificity set to Trypsin/P, cysteine carbamidomethylation as a fixed modification (+ 57.021464) and methionine oxidation (+ 15.994914) as well as protein N-terminal acetylation (+ 42.010565) were as variable modifications. Data analysis was performed using Perseus software suite (V1.6.2.3). TNFα-treated podocyte data were log2 transformed and filtered to only include proteins that were measured in at least 4 out of 6 replicates in at least one group. Missing data was imputed sample wise from a normal distribution with a width of 0.3 SD and a downshift of 1.8 SD. Volcano plots were created according to a two-sided t-test with an FDR of 0.2 for supernatants and an FDR of 0.1 for cells. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 40 partner repository with the dataset identifiers. Project Name 1 : Proteomic analysis of cultured human podocytes stimulated with TNFα - cell pellet, Project accession: PXD032107 Project DOI: Not applicable, Reviewer account details: Username: [email protected] Password: TCWTgTMa. Project Name 2: Proteomic analysis of cultured human podocytes stimulated with TNFα – supernatant, Project accession: PXD032130, Project DOI: Not applicable,Reviewer account details: Username: [email protected], Password: j2wELsGJ
Human glomeruli and tubule proteomics data. Datasets were retrieved for human microdissected proximal tubules and single human glomeruli from adult human kidneys 19 to compare with our organoid data.
Deep proteomic analysis of human glomeruli. For deep mass spectrometry analysis sieved human glomeruli were used as previously described 34. Proteins from glomeruli were extracted, solubilized and trypsinized as described above. The tryptic digests were fractionated using in-house made stage-tips, applying high pH reverse phase fractionation with fresh 100 mM ammonium formate, pH 10 and stepwise increase of ACN from 0–50% (n = 8 fractions). The eight fractions were analyzed on the same Q Exactive Plus instrument as indicated above with analogous settings.
NEPTUNE bulk RNA-seq and snRNA-seq. Data from an existing study of individuals with proteinuric kidney disease were utilized for analysis 24. In brief, the NEPTUNE (NCT01209000) study objectives, design and procedures were described in previous publications 57,58. Consent was obtained from individuals or parents/guardians at enrollment, and the study was approved (HUM00158219) by University of Michigan, Medical School Institutional Review Board. Renal biopsies were microdissected into glomeruli and tubulointerstitial compartments. Bulk tubulointerstitium kidney biopsy transcriptional data from 220 NEPTUNE participants with FSGS/MCD are available through via GEO accession number GSE182380. Single nuclear transcriptional data for ten NEPTUNE participants within this cohort, five with high intrarenal TNF activity and five with low TNF activity, are available via GEO accession number GSE213030.
Organoid derived proteomic TNF signature applied to NEPTUNE kidney tissue transcriptome data. The proteomic signatures of TNFα treated organoids were derived from the total of 320 differentially expressed proteins (DEPs) under TNFα treatment at 24 and 48 h (supplemental table 10) by combining cellular proteins (n = 300 DEPs) plus those proteins secreted into the culture medium (n = 22 DEPs) (supplemental tables 3 and 4), 2 DEPs were represented in both sets. Peptide to ENSEMBL gene id conversions were performed using Biomart. In cases where peptides mapped to more than one protein, each gene encoding the given protein was used for the gene set. Each of the three gene sets (cellular, secreted, combined) was used to compute an eigengene (first principal component) from each patient-derived tubulointerstitial transcriptome profile in patients with MCD or FSGS from NEPTUNE (GSE140989).
TNF network and pathway analysis. For each of the two gene signatures, i) from the previously characterized TNF gene set in human kidney tissue (n = 272 genes) 24, ii) from the organoid TNFα proteome (n = 320 genes) and the combination of these two gene signatures (n = 582 genes), biologic literature-based networks were generated using Genomatix Pathway System (GePS) software (http://genomatix.de). In these networks, the 100 best connected genes co-cited in PubMed abstracts in the same sentence linked to a function word (most relevant genes/interactions) were represented. The TNF-centric network was generated from the 582 combined genes asking the software to include in the network the 10 common genes listed in Fig. 5B in the 100 best connected genes. The top 20 pathway-based and signal transduction networks were generated from the individual and combined gene signatures using GePS (Suppl. Tables 11 and 12).
Transcriptomic analysis of the European Renal cDNA Bank (ERCB) tissue. Previously generated microarray data from microdissected human glomeruli and tubulointerstitium sourced from individuals with kidney disease (n = 184) and healthy donors (n = 50) were used (GEO accession # GSE104948 and GSE104954) 59.Microarray assays captured a subset of 267 genes of the 320-gene product kidney organoid signature in the ERCB cohort. Gene expression was Z-transformed, and Z-scores were calculated as previously described 24.
Kidney Disease Explorer Shiny application. Seventeen different data sets from the following prior publications were combined with data from this study for this application. Rinschen et al. 37 defined the cellular effects of puromycin aminonucleoside (PAN) under different circumstances (in vitro differentiated and non-differentiated, in vivo after two and four days). Höhne et al. 19 created proteomic datasets for various kidney diseases (FSGS, congenital nephrotic syndrome) based on individual kidney segments. Bartram et al. analyzed FSGS (podocyte cell line and primary renal epithelial cells from urine) due to G195D mutation in the ACTN4 gene38. Koehler et al. analzyed the effects of doxorubicin and LPS on podocytes39. The app is available at https://kidneyapp.shinyapps.io/kidneyorganoids/.
Western blot for complement factor C3. Proteins in culture media were separated on 4–15% gradient SDS-PAGE gels (Bio-Rad, Criterion TGX gels #567–1083) and transferred to nitrocellulose membranes (Bio-Rad #170–4159) which was blocked and developed with polyclonal rabbit anti-human-C3c (DAKO #0368) in Tris-buffered saline, 1 mM EDTA, pH 7.4, with 1 mg human serum albumin (CSL Behring #109697) and 100 µg human IgG (CSL Behring #007815) per milliliter. The membrane was then washed, incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (DAKO #P0448) and developed with SuperSignal West Dura extended-duration substrate (Pierce). Emission was recorded by a charge-coupled device camera.