Materials
Subcutaneous abdominal white adipose tissue was obtained with informed consent from 25 patients undergoing reconstructive breast surgery at the University Health Network (Toronto, ON, Canada; institutional research ethics board approval no. 13-6437-CE). Tissue culture-treated polystyrene (TCPS) was sourced from Corning (Corning, NY, United States). Unless indicated otherwise, all other materials were from Sigma-Aldrich (St. Louis, MO, United States).
Isolation of the stromal vascular fraction
The stromal vascular fraction was isolated from adipose tissue as previously described29. Briefly, adipose tissue was minced and enzymatically digested for 1 hr at 37 °C using collagenase type II (2 mg/mL) in Kreb’s Ringer bicarbonate buffer supplemented with 3 mM glucose, 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, and 20 mg/mL bovine serum albumin (BSA). The buoyant adipocytes were discarded following the centrifugation of the digest, and the pelleted tissue was subjected to another 15 min of enzymatic digestion at 37 °C using 0.25% trypsin-ethylenediaminetetraacetic acid (EDTA) solution. The cells were then resuspended in sterile distilled and deionized water supplemented with 0.154 M ammonium chloride, 10 mM potassium bicarbonate, and 0.1 mM EDTA for 10 min to facilitate erythrocyte lysis, before being filtered through a 100 µm sieve. The resulting filtrate was defined as the stromal vascular fraction and was immediately prepared for MACS or immunophenotyping.
Magnet-assisted cell sorting
IMPs directed against CD45 (Dynabeads 45) and CD31 (Dynabeads CD31 Endothelial Cell) were obtained from Invitrogen (Carlsbad, CA, United States). Alternatively, cIMPs were generated using the CELLection Biotin Binder Kit (Invitrogen). Specifically, 5 µg of either mouse anti-human CD31 antibodies (Miltenyi Biotec, Bergisch Gladbach, Germany; catalogue no. 130-119-893) or mouse anti-human CD93 antibodies (BioLegend, San Diego, CA, United States; catalogue no. 336104) were conjugated to 107 Dynabeads via biotin-streptavidin interactions through DNA linkers, enabling their release from cells with the supplied DNase.
HAMVECs and ASCs were isolated from the stromal vascular fraction using MACS. Cells were resuspended in sorting buffer (phosphate-buffered saline without calcium chloride and magnesium chloride (PBS−/−) supplemented with 2 mM EDTA and 0.1% (w/v) BSA), and were incubated with IMPs for 20 min at 4 °C before being magnetically separated using the DynaMag-5 Magnet (Invitrogen). The stromal vascular fraction was depleted of CD45+ leucocytes prior to separating CD45−CD31+ HAMVECs from CD45−CD31− ASCs. The enzymatic exclusion of cIMPs from cultures was performed in select experiments using the supplied DNase as per the manufacturer’s instructions. Briefly, cIMP-labelled cells were incubated with DNase to dissociate the cells from the superparamagnetic microparticles prior to using the DynaMag-5 Magnet to separate the superparamagnetic microparticles from the cells of interest. The latter were then seeded onto TCPS in the absence of the superparamagnetic microparticles used to isolate them. Cultures of HAMVECs were enriched using IMPs or cIMPs directed against CD31 or CD93 using the same procedure.
Cell culture
HAMVECs and ASCs were plated onto TCPS and cultured at 37 °C, 5% CO2 under a relative humidity of 85% in Endothelial Cell Growth Medium (EGM)-2 (Lonza, Walkersville, MD, United States). HUVECs (Lonza; n = 3 biologically independent samples), HCAECs (Lonza; n = 3 biologically independent samples), and HDMVECs (Lonza and PromoCell, Heidelberg, Baden-Württemberg, Germany; n = 3 biologically independent samples) were obtained commercially and cultured under the same conditions. Media was exchanged three times a week, and cells were passaged at 75–90% confluence using TrypLE Express (Invitrogen). Phase-contrast transmission light microscopy was used to assess morphology and confluence (Leica DMIL, Leica Microsystems, Wetzlar, Hesse, Germany). Cells were counted using a hemocytometer, and dead cells were excluded based on trypan blue staining.
Immunophenotyping
Cells were stained with either Live/Dead Fixable Aqua or Live/Dead Fixable Near-IR (Invitrogen), and Fc receptors were blocked using Human TruStain FcX (BioLegend). Cells were then stained for 20 min at 4 °C with combinations of the following fluorophore-conjugated mouse anti-human monoclonal antibodies sourced from BioLegend: CD45-APC/Cy7 (catalogue no. 368516), CD45-Brilliant Violet 785 (catalogue no. 304048), CD31-Alexa Fluor 488 (catalogue no. 303110), CD31-Brilliant Violet 421 (catalogue no. 303124), and CD93-PE (catalogue no. 336108). The staining concentration for each of the antibodies was 5 µg/mL. For select experiments, the signal of the Alexa Fluor 488-conjugated mouse anti-human CD31 antibody (BioLegend; catalogue no. 303110; staining concentration, 5 µg/mL) was amplified using an Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Invitrogen; catalogue no. A11001; staining concentration, 5 µg/mL; Supplementary Fig. 8). Stained cells were fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C. Compensation was achieved using the AbC Anti-Mouse Bead Kit and the ArC Amine Reactive Compensation Bead Kit (Invitrogen). Gates were set using fluorescence minus one controls. Flow cytometry was performed using a BD LSR II or BD LSR Fortessa flow cytometer (Becton, Dickinson and Company, Franklin Lakes, NJ, United States) at the Temerty Faculty of Medicine Flow Cytometry Facility (University of Toronto, Toronto, ON, Canada). Data was acquired using BD FACSDiva software version 8.0.1, and analysed using FlowJo software version 10.7.1 (Becton, Dickinson and Company). The immunophenotypes of two populations of cells were compared using two-tailed t-tests, blocking for donors when appropriate; and, those of ≥ 3 populations were compared using a one-way analysis of variance (ANOVA), with Tukey’s post-hoc test for multiple comparisons.
Gene expression
Reverse transcription quantitative real-time polymerase chain reaction was performed as previously described29. Briefly, total ribonucleic acid isolated from cells using Trizol (Invitrogen) was immediately reverse transcribed into complementary DNA (cDNA) using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, United States). The quantitative real-time polymerase chain reaction was performed on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, United States), using the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad). Each reaction comprised 10 ng template cDNA and 450 nM of both forward and reverse primers in a total volume of 10 µL, with thermal cycling performed as previously described29. The primers employed were previously designed and validated to amplify GAPDH, PECAM1, CDH5, and VWF29. Their nucleotide sequences are delineated in Supplementary Table 1. All experiments utilized three technical replicates for each biological sample and included no reverse transcriptase controls. Data were analysed using Bio-Rad CFX Maestro 1.1 software version 4.1.2433.1219 (Bio-Rad).
The abundance of messenger ribonucleic acid encoding genes of interest in the putative HAMVECs was compared to that in HUVECs, HCAECs, and HDMVECs using a two one-sided test for equivalence45. Specifically, equivalence was established within the significance level α when the (1–2α) × 100% confidence interval of the difference in mean quantification cycles (Cq) between HAMVECs and the EC controls was contained within the equivalence margin (± ∂). The significance level α was set to 0.05, and the equivalence margin ∂ was set to three standard deviations of the Gaussian distribution of Cq values amongst the EC controls about their gene-normalized mean (Supplementary Fig. 3). The EC controls were evaluated separately but statistically presented as a single population in order to ascertain the variability of their expression of characteristic endothelial genes29. GAPDH was used as a loading control.
Immunofluorescence
Cells were rinsed with phosphate-buffered saline (with calcium chloride and magnesium chloride; PBS+/+), fixed with ice-cold methanol for 10 min at − 20 °C, and blocked with 3% (w/v) BSA in PBS−/− for 30 min prior to being stained overnight at 4 °C with 5 µg/mL of primary antibodies in the same blocking solution. The primary antibodies were sourced from Abcam, and included a mouse anti-human CD31 antibody (catalogue no. ab24590), a rabbit anti-human VE-cadherin antibody (catalogue no. ab33168), and a mouse anti-human vWF antibody (catalogue no. ab194405). Cells were then blocked with Normal Serum Block (BioLegend) for 30 min at 25 °C and stained for 1 hr with 2.5 µg/mL of either a goat anti-mouse IgG-Alexa Fluor 594 secondary antibody (BioLegend; catalogue no. 405326) or a goat anti-rabbit IgG-Alexa Fluor 488 secondary antibody (Abcam; catalogue no. ab150077) diluted in 3% (w/v) BSA in PBS−/−. Cells were then stained for 5 min with 3 µM 4′,6-diamidino-2-phenylindole (Abcam) in PBS−/−. Wide-field immunofluorescence microscopy was performed using a Leica DMi8, operated using the Leica Application Suite X software version 3.5.5.19976 (Leica Microsystems). Image processing was performed using Fiji software version 2.1.0/1.53c46.
Acetylated low-density lipoprotein uptake
AcLDL uptake was assessed as previously described29. Cells were incubated with 10 µg/mL Alexa Fluor 488-conjugated AcLDL (Invitrogen) in EGM2 for 4 hr, after which they were fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C. Cells were analysed by flow cytometry.
Angiogenic capacity
The angiogenic capacity of cells was assessed as previously described47. Briefly, TCPS was coated with 150 µL/cm2 of Cultrex PathClear Basement Membrane Extract (Bio-Techne, Minneapolis, MN, United States). Cells were then seeded at a density of 45,000 cells/cm2 and incubated in EGM2 for 6 hr before their formation of capillary-like tubules was assessed using phase-contrast transmission light microscopy (Leica DMIL).
Adipogenic plasticity
The accumulation of lipids by ASCs and HAMVECs was assessed using a protocol adapted from Kraus et al.48. Cells were seeded onto TCPS at a density of 4,000 cells/cm2 and cultured in EGM2 for 24 hr before the media was replaced with StemPro Adipogenesis Differentiation Kit (Invitrogen). The latter was exchanged every other day for 10 days before cells were rinsed with PBS−/− and fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at room temperature. Fixed cells were rinsed with 40% (v/v) isopropanol in distilled water before being stained with 0.2% (w/v) Oil Red O in 40% (v/v) isopropanol in distilled water for 30 min at room temperature. Stained cells were rinsed with distilled water before being counterstained with Mayer’s hematoxylin (Abcam; Cambridge, United Kingdom) for 5 min at room temperature. Cells were rinsed with distilled water, and representative photomicrographs were acquired using brightfield transmission light microscopy (Leica DMIL). Oil Red O was then eluted from the cells by incubating them in 100% isopropanol for 10 min at room temperature, and the absorbance of the eluent at 515 nm was quantified by spectrophotometry (EnVision 2104 Multilabel Reader operated using EnVision Manager software version 1.14.3049.528; PerkinElmer, Waltham, MA, United States). The absorbance of the eluent was compared using a two-way ANOVA with Tukey’s post-hoc test for multiple comparisons, blocking for donors.
Liquid chromatography–tandem mass spectrometry
Reversed-phase LC–MS/MS was performed as previously described29. Briefly, cells were lysed and proteins extracted in 50% (v/v) 2,2,2-trifluoroethanol in PBS–/– supplemented with 100 mM ammonium bicarbonate. Proteins were reduced with 5 mM dithiothreitol, alkylated with 15 mM iodoacetamide, and digested with mass spectrometry-grade Trypsin-Lys C mix (Promega, Madison, WI, United States) following their dilution with 100 mM ammonium bicarbonate and supplementation with 2 mM calcium chloride. Formic acid was used to quench the digestion, after which the tryptic peptides were de-salted using OMIX C18 solid-phase extraction tips (Agilent, Santa Clara, CA, United States). Samples were dried by vacuum centrifugation and reconstituted in 5% (v/v) formic acid in high-performance liquid chromatography-grade water.
Tryptic peptides were analysed on an Easy-nLC 1200 (Thermo Fisher Scientific, Waltham, MA, United States) coupled to a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) through a Nanospray Flex Ion Source (Thermo Fisher Scientific). Tryptic peptides were loaded onto an in-house packed reversed-phase 10-cm, 75 µm internal diameter column (Reprosil-Pur Basic C18, 3 µm, 100 Å; Dr. Maisch HPLC, Ammerbuch, Baden-Württemberg, Germany), and separated using a 3-hr acetonitrile linear gradient (2–35% (v/v) in 0.1% (v/v) formic acid in high-performance liquid chromatography-grade water) at 250 nL/min. All experiments utilized two technical replicates for each biological sample. Spectra were collected using a top ten data-dependent acquisition method as previously described29, using Tune software version 2.8 (Thermo Fisher Scientific) and Xcalibur software version 4.0.27.19 (Thermo Fisher Scientific). Raw files were searched against the UniProt human proteome database (updated to 2017-07-24) using MaxQuant software version 1.6.0.1 (Max Planck Institute of Biochemistry, Planegg, Bavaria, Germany)49, with ‘match between runs’ enabled. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation, N-terminal acetylation, and asparagine or glutamine deamidation were selected as variable modifications. The false discovery rate (FDR) was set to 1% using a reversed-target decoy database.
Data visualization and statistical analyses were completed using the Perseus 1.6.1.2 software package (Max Planck Institute of Biochemistry)50. Label-free quantification values were log2-transformed51, and missing values were imputed from a normal distribution using a downshift of 1.8 and width of 0.3 standard deviations (Supplementary Data 1 and 2). Unsupervised hierarchical clustering and associated heat maps of proteins that were identified in ≥2 biological replicates in at least one group were generated from normalized values across all samples for each protein. Differentially expressed proteins between groups of cells were defined as being either uniquely detected (i.e. detected vs. not detected in ≥2 biological replicates per group) or quantified in statistically different abundances (i.e. detected in ≥2 biological replicates in each group, but p < 0.05 or FDR < 0.05). Statistically overrepresented (p < 0.05) biological pathways (Reactome version 65; released 2019-12-22) among the differentially expressed proteins were identified using the Fisher’s Exact test with Bonferroni’s correction for multiple testing within PANTHER version 15.0 released 2020-02-1452, and the Gene Ontological Term Mapper was used to map differentially expressed proteins to their cellular components.
Population doubling time
Cells were seeded at a density of 4,000 cells/cm2 into seven TCPS flasks, and every 24 hr cells from one flask were counted using a hemocytometer. Population doubling times were calculated from the exponential growth phase using the formula PDT = [∆t × log10(2)] ÷ [log10(nf ÷ ni)], where PDT represents population doubling time, ∆t represents the duration of exponential growth, and ni and nf represent the number of cells at the initiation and termination of the exponential growth phase, respectively53. The population doubling times of two populations of cells were compared using two-tailed t-tests, blocking for donors where appropriate, and those of ≥3 populations were compared using a one-way ANOVA, with Tukey’s post-hoc test for multiple comparisons.
Population growth rates in co-culture
The effect of seeding purity on the population growth rates of HAMVECs and ASCs was assessed using an assay adapted from Gerashchenko (Supplementary Fig. 4)54. HAMVECs and ASCs were seeded onto TCPS in several pre-defined proportions amounting to a total density of 4,000 cells/cm2 or separately at their respective constituent densities. Cells were cultured for 4 days, after which the co-cultured cells were stained with 4 µM CellTrace Far Red (Invitrogen) and combined with the mono-cultured controls. The mixture was then blocked with Human TruStain FcX (BioLegend), stained with 5 µg/mL of mouse anti-human CD31-Alexa Fluor 488 (BioLegend; catalogue no. 303110) for 20 min at 4 °C, and fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C. Cells were analysed by flow cytometry. Population growth rates in co-culture were calculated using the formulas, R1 = [Dye+CD31+ ÷ Dye−CD31+] × [Dye ÷ Dye−] × 100 and R2 = [Dye+CD31− ÷ Dye−CD31−] x [Dye+ ÷ Dye−] × 100, where R1 and R2 represent the population growth rates of HAMVECs and ASCs relative to their mono-cultured controls, respectively. Two-tailed one-sample t-tests were used to compare the population growth rates in co-culture to that in mono-culture (i.e. 100%), and a two-way ANOVA with Bonferroni’s post hoc test for multiple comparisons was used to evaluate the relationship between the seeding purity and the resultant population growth rates, blocking for donors.
Temporal stability of population purities
The effect of seeding purity on the temporal composition of cultures was assessed. HAMVECs and ASCs were seeded onto TCPS in pre-defined proportions (100 or 90% HAMVECs) amounting to a total density of 4,000 cells/cm2. The purity (% CD31+) of the cultures was assessed by flow cytometry after 1, 7, 14, and 28 days. The effect of seeding purity on the purity of the cultures over time was assessed using a two-way ANOVA with Tukey’s post-hoc test for multiple comparisons, blocking for donors.
Immunomagnetic microparticle size distribution
The size distribution of IMPs was evaluated with a Multisizer 4e Coulter Counter (Beckman Coulter Life Sciences, Indianapolis, IL, United States), using a 100 µm aperture calibrated with 10 µm beads (Coulter CC Size Standard L10; Beckman Coulter Life Sciences) or a 20 µm aperture calibrated with 2 µm beads (Coulter CC Size Standard L2; Beckman Coulter Life Sciences)55. Data was acquired for a modal count of 5,000 using the Multisizer 4e software version 4.03 (Beckman Coulter Life Sciences). The polydispersity index (PDI) of the IMPs was calculated using the formula PDI = (standard deviation/mean)2, as previously described55. The size distributions of the different IMPs were compared using a one-way ANOVA with Tukey’s post-hoc test for multiple comparisons.
Immunomagnetic microparticle uptake and localization
IMPs were detected on the basis of their autofluorescence, and in select experiments, a membrane-impermeable secondary antibody directed against their binding moiety was used to discriminate between their extracellular and intracellular localization (Supplementary Fig. 5). The autofluorescence of the IMPs and the effect of conjugating the goat anti-mouse IgG-Alexa Fluor 647 secondary antibody (Cell Signalling Technology, Danvers, MA, United States; catalogue no. 4410S; staining concentration, 5 µg/mL) on their fluorescence was characterized by spectral scanning confocal microscopy (Leica SP8 microscope operated by Leica Application Suite X software version 3.5.5.19976, Leica Microsystems; Advanced Optical Microscopy Facility, University Health Network), with 20 nm detection bandwidths employed to assess the emission spectra for each of the 405 nm, 488 nm, 552 nm, and 638 nm excitation wavelengths. 30 IMPs were selected as regions of interest in generating representative emission spectra. Candidate excitation-emission features of the IMPs were validated by flow cytometry.
The localization of IMPs in culture was assessed by confocal microscopy. ASCs were seeded at a concentration of 4,000 cells/cm2 onto 35 mm µ-Dishes (ibidi GmbH, Gräfelfing, Germany) with 400,000 anti-CD31 IMPs/cm2 (Dynabeads; Invitrogen). After 24 hr, cells were rinsed with PBS+/+ to remove non-adherent IMPs, fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C, and permeabilized with 0.1% (v/v) Triton X-100 in PBS−/− for 15 min. F-actin was stained with Alexa Fluor Plus 647 Phalloidin (Invitrogen), and nuclei were stained with 3 µM 4′,6-diamidino-2-phenylindole in PBS−/− for 5 min (Abcam). Anti-CD31 IMPs were detected on the basis of their autofluorescence, and their localization was assessed by z-stacking confocal microscopy. Images were processed using Fiji software version 2.1.0/1.53c46.
The prevalence of IMP uptake and their localization in ASCs was assessed by flow cytometry (Supplementary Fig. 5). Specifically, 105 ASCs were exposed to 107 IMPs in suspension for 20 min at 4 °C, or in culture in 25 cm2 TCPS flasks for pre-defined durations. Cells were then stained with 5 µg/mL of goat anti-mouse IgG-Alexa Fluor 647 secondary antibody (Cell Signalling Technology; catalogue no. 4410S) for 20 min at 4 °C, and fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C. Cells were analysed by flow cytometry. The prevalence of IMP localization over time was compared by a two-way ANOVA with Bonferroni’s post-hoc test for multiple comparisons, blocking for donors; and, the total prevalence of IMP uptake by ASCs after 20 min in suspension vs. 48 hr in culture was compared using a two-tailed t-test, blocking for donors.
DNA synthesis
The effect of binding and internalizing IMPs on the proliferative capacity of ASCs was assessed using the Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit (Invitrogen; Supplementary Fig. 6). ASCs were seeded at a concentration of 4,000 cells/cm2 onto TCPS with 400,000 anti-CD31 IMPs/cm2 (Dynabeads; Invitrogen). After 48 hr in culture, cells were rinsed with PBS+/+ to remove free IMPs and pulsed with 10 µM 5-ethynyl-2´-deoxyuridine (EdU) in EGM2 for 6 hr. Cells were then resuspended in cell sorting buffer (2 mM EDTA and 0.1% (w/v) BSA in PBS−/−), and IMP-laden ASCs were magnetically separated from IMP-free ASCs. Fixation, permeabilization, and staining were performed using the kit. Cells were analysed by flow cytometry. A two-tailed t-test was used to compare EdU incorporation between IMP-laden and IMP-free ASCs, blocking for donors.
Effect of size on immunomagnetic microparticle uptake
The effect of IMP size on their uptake by ASCs was assessed. Superparamagnetic microparticles ranging in diameter from 1 to 8 µm (Spherotech, Lake Forest, IL, United States; catalogue no. SVM-10-10, SVM-40-10, and SVM-80-5) were conjugated to mouse anti-human CD31 antibodies (Miltenyi Biotec; catalogue no. 130-119-893) via biotin-streptavidin interactions (5 µg of antibody per 107 particles). Their uptake by ASCs was compared to that of the commercial anti-CD31 IMPs (Dynabeads CD31 Endothelial Cell; Invitrogen) and the anti-CD31 cIMPs. The size distributions of these five distinct anti-CD31 IMPs was assessed using a Coulter counter.
The binding and internalization of the anti-CD31 IMPs was assessed by flow cytometry (Supplementary Fig. 9b and 9c). Specifically, 105 ASCs were exposed to 107 IMPs in suspension for 20 min at 4 °C, or in culture in 25 cm2 TCPS flasks for pre-defined durations. Cells were then resuspended in cell sorting buffer (2 mM EDTA and 0.1% (w/v) BSA in PBS−/−), and IMP-laden ASCs were magnetically separated from IMP-free ASCs. IMP+ ASCs were stained with 4 µM CellTrace Far Red (Invitrogen) before being re-combined with the IMP− ASCs. The mixture was then stained with 5 µg/mL of goat anti-mouse IgG-Alexa Fluor 488 secondary antibody (Invitrogen; catalogue no. A11001) for 20 min at 4 °C, and fixed with 4% (w/v) paraformaldehyde in PBS−/− for 15 min at 4 °C. Cells were analysed by flow cytometry. The effect of IMP size on their uptake by ASCs over time was assessed using a two-way ANOVA with Tukey’s post-hoc test for multiple comparisons, blocking for donors.
Enrichment efficacies of alternative MACS strategies
The effects of alternative target antigens, sizes, and exposures of IMPs on the enrichment efficacy of the MACS procedure were investigated using an in vitro model of contaminated primary cultures. HAMVECs and ASCs were seeded onto TCPS at a density of 4,000 cells/cm2 in a 9:1 proportion with or without 400,000 IMPs/cm2 to recapitulate their potential exclusion from primary cultures. Their enrichment was performed using MACS after 4 days, and their purities (% CD31+) were evaluated after a total of 7 days by flow cytometry. A two-way ANOVA with Tukey’s post-hoc test for multiple comparisons was used to compare the culture purities of the different MACS strategies, blocking for donors. The effects of target antigen and IMP size/exposure were assessed using a one-way ANOVA with Tukey’s post-hoc test for multiple comparisons; pooling replicates where applicable.
Statistics and reproducibility
Experiments were performed three times using cells derived from three different donors (n = 3 biologically independent samples), unless indicated otherwise. Other than the proteomics data that was analysed using the Perseus 1.6.1.2 software package (Max Planck Institute of Biochemistry)50, statistical analyses were performed using Prism 8 software version 8.4.3 (GraphPad Software, San Diego, CA, United States). Where applicable, normality of data was assessed using quantile-quantile plots and the Shapiro–Wilk test. Unless stated otherwise, p < 0.05 was accepted as statistically significant and values are represented as mean ± standard deviation.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
