Cell culture conditions

Panc0203 (ATCC, CRL-2553), Panc0327 (ATCC, CRL-2549), and pancreatic derivative cells (SP0926, SP1030 and SP09131A, 1B & 1C) were grown in RPMI 1640 supplemented with 10 units/ml human recombinant insulin (Merck, 91077 C), 15% fetal bovine serum (FBS), and 1% penicillin-streptomycin (PS) solution. Human pancreatic normal fibroblasts (n-fibroblast; VitroBioPharma, SC00A5) and pancreatic CAFs (Vitro BioPharma, CAF08) were cultured in MSC-GRO low serum medium (Neuromics, PC00B1) with 1% PS solution as per the procedure recommended by provider. Confluent culturing adherent cells at greater than 80% were split 1:2 into new 100-mm culture dishes. Human monocytic cell line U937 cells (ATCC Cat# CRL-1593) were maintained in RPMI 1640 that included 10% FBS and 1% PS solution. To induce differentiation of monocytic U937 cells into macrophage-like cells (MØ-U937), phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich) dissolved in DMSO was added to 5 ×105 cells/ml at a final concentration of 10 ng/ml for 72 h. All cells were maintained in a humidified atmosphere with 5% carbon dioxide at 37 °C as a standard culture condition.

TME mimic model system

The following is a description of the gemcitabine-treated TME (s-TME) model. To collect the first conditioned medium (CM) as an Initial Culture Phase comprising exposure to hypoxia, n-fibroblasts and CAFs of greater than 90% confluence were cultured separately on 100-mm culture dishes for 1.5 days without gemcitabine treatment in RPMI 1640 medium that included 15% exosome-free (exo-free) FBS (System Biosciences), 10 units/ml human recombinant insulin, and 1% PS and using a GasPak EZ Anaerobe Pouch System (BD Biosciences). Then, as a Co-culture Phase giving the opportunity for interaction between n-fibroblasts and CAFs, cells of the two types were mixed (1:1) following trypsinization and allowed to continue to grow under hypoxia for an additional 1.5 days on a 100-mm culture dish until collection of the first CM. The culture dish contained a total of 20 ml of mixed medium obtained from the Initial Culture Phase after adding exo-free FBS (2.5% v/v) to provide fresh nutrient sources. For collection of the second CM, independently prepared culture dishes were treated with 10 μM of gemcitabine (TCI) at the beginning of the Initial Culture Phase for 1.5 days. Other than the Gemcitabine treatment, there is no difference between first CM and second CM acquisition process. Then, at the end of the Co-culture Phase for both the first CM and second CM collection, the co-culturing fibroblasts were scraped out from the culture dishes with cell lifters (Corning), and then the cell-containing media were centrifuged at 2500 rpm at room temperature for 5 min following gentle agitation for 30 min at 37 °C. Every supernatant was transferred to a new conical tube, and fresh exo-free FBS was added to a final concentration of 10% (v/v). The first CM and second CM samples collected were immediately used in the following the Pre-Incubation Phase and Seeding Phase, sequentially (Fig. 1a).

The first CM was added into 100-mm culture dishes paired with maternal pancreatic cancer cells (Panc0203 or Panc0327) and MØ-U937 cells, and then the paired dishes were cultured under hypoxic (Panc0203 and MØ-U937) or normoxic (Panc0327 and MØ-U937) condition for 1.5 days as a Pre-incubation Phase mimicking exposure of cancer cells and macrophages to biogenic substances derived from stromal fibroblasts without chemotherapy. Then, the first CM-exposed Panc0203 (or Panc0327) and MØ-U937 cells in the paired tissue-culture dishes were pre-stained with 5 μM of CellTracker Orange CMTMR and Green CMFDA (Invitrogen), respectively, for 45 min in serum-free medium under the standard culture condition. After trypsinization, the first CM-exposed MØ-U937CMFDA cells were washed three times with phosphate buffered saline (PBS), re-suspended in the second CM, and transferred onto another dish with the first CM-exposed/PBS-washed (three times) Panc0203CMTMR (or Panc0203CMTMR) cells. The mixed cells were co-cultured under hypoxic (Panc0203CMTMR and MØ-U937CMFDA) or normoxic (Panc0327CMTMR and MØ-U937CMFDA) condition for an additional 1.5 days, termed a Seeding Phase, mimicking a gemcitabine-exposed stromal environment.

To generate SP0926 and SP09131A, 1B & 1C cells following DFCs isolation, single pair of 100-mm culture dishes for each group was prepared by culturing MØ-U937 cells and maternal cancer cells (Panc0203 or Panc0327 cells) at the beginning the Pre-incubation Phase, and all sorted DFCs through FACS from each group were seeded on a 100-mm culture dish until grown to more than 60% confluency. To generate SP1030 cells, five pairs of 100-mm culture dishes were used in which Panc0203 and MØ-U937 cells are being cultured, and all isolated DFCs were seeded on a T-75 flask. It took 129, 92, 61, 68, and 72 days to generate SP0926, SP1030, SP09131A, SP09131B, and SP09131C, respectively. STR analysis was then performed using each of the first stocks stored in a liquid nitrogen tank.

The following is a description of the normal co-culture TME condition. Compared with the s-TME model, the only difference of the normal co-culture TME condition was a normoxic/gemcitabine free condition in the process of obtaining the first and second CMs, which were then sequentially used in the Pre-incubation and Seeding Phases of the paired Panc0203 cells with MØ-U937 cells under normoxia. Note, fresh exo-free FBS was added to a final concentration of 2.5% (v/v) in first CM and 10% (v/v) in the second CM.

To observe the blockade effect of exosome biogenesis/release and TNTs formation, 10 μM GW4869 (Sigma), 10 μM ML-141 (Calbiochem) or 100 nM Cytocalasin B (abcam) was pre-treated to each n-fibroblasts and CAFs 24 h prior to the Initial Culture Phase, then the inhibitors were also applied to them at the beginning the Initial Culture Phase for the first CM and second CM generation. Each Panc0203 and MØ-U937 cells were also pre-exposed for 24 h to the inhibitors, and then the prepared first CM was treated to each Panc0203 and MØ-U937 cells and the second CM were treated to the mixed cell plate of Panc0203CMTMR and MØ-U937CMFDA cells, by sequence (Fig. 1a). Note, for every trypsinization step, take the trypsinized cell plates out of the incubator every few minutes and tap it to help the cells fall off sufficiently, avoiding the use of a cell lifter and exposure to light.

Fluorescence-activated cell sorting

The whole cultured media, including ‘suspending (detached) cells’ considered as damaged cells in the s-TME model, from the culture plates at the end of the Seeding Phase (Fig. 1c, Supplementary fig. 1a) were collected into 50 ml conical tube(s) and the pelleting cells were washed 3 times by PBS following centrifugation at 1500 rpm for 5 min. The cells were resuspended in ice-cold FACS buffer (0.1% bovine serum albumin and 0.01% sodium azide in 1 x PBS). The ‘adherent cells’ considered as viable cells at the end of the Seeding Phase of the s-TME model (Fig. 1c and Supplementary fig. 1a) were trypsinized, and washed three times by PBS following centrifugation at 1500 rpm for 5 min. The pelleting cells were resuspended by ice-cold FACS buffer. Note, take the trypsinized cell plates out of the incubator every few minutes and tap it to help the cells fall off sufficiently, avoiding the use of a cell lifter and exposure to light. After passing the ‘suspending cells’ and ‘adherent cells’ through the strainer cap, stored on ice in a dark condition until the scheduled time for analysis.

To isolate DFC (CMFDA- and CMTMR-positive) populations, FACS was performed using a FACSAria IIu cell sorter (BD Biosciences) according to the manufacturer’s instructions. Briefly, FITC and PE channels were used for detecting CellTracker™ Green CMFDA (Ex/Em: 492/517 nm) and CellTracker™ Orange CMTMR (Ex/Em: 541/565 nm), respectively, following fluorescence compensation (Supplementary fig. 1a).

Flow cytometry for detecting CD24 and CD44 of Panc0203 and derivative cells was performed by the procedure for the ‘adherent cells’ described above. Briefly, after washing with PBS, the cells were incubated for 45 min with anti-human antibodies CD24 PE (BioLegend, 311106) and CD44 APC (BioLegend, 338806) diluted (1:50) in FACS buffer.

Nanoparticle tracking analysis

NanoSight NS300 (NanoSight) was used for nanoparticle tracking analysis (NTA) measurements according to the manufacturer’s instructions using the purified s-TME CMFDA/DMSO-Exo and s-TME CMFDA/GW4869-Exo from total 20 ml volume of second CM. The data were collected following 3-time repeated running for taking exosome captures per each sample. Concentration (particles/ml) average and mean diameter (nM) within the range of 0–400 nm were used for data presentation. NTA Version: NTA 3.2 Dev Build 3.2.16, camera type: sCMOS, number of frames: 749, time: 30 seconds, temperature: 25°C, dilution factor: 5.

Quantification of exosome fluorescence intensity

For exosome fluorescence quantification, Varioskan LUX Multimode Microplate Reader (Thermo Fisher) was used. CMFDA (Ex/Em: 492/517 nm) fluorescence signal quantification was performed according to the manufacturer’s instructions using the purified s-TME non-fluorescent(fluor)/DMSO-EXO, s-TME CMFDA/DMSO-Exo and s-TME CMFDA/GW4869-Exo from total 20 ml volume of second CM. Briefly, each diluted exosome samples (100 μl, dilution factor: 5) in PBS were loaded into a 96-well black plate (SPL) and fluorescence signals were read by operating Skanlt software 6.1.1 (Thermo Scientific) linked to Varioskan LUX Multimode Microplate Reader.

Confocal microscopy

Plated cells at about 60% confluence were grown on Lab-Tek 2-chamber glasses (Nunc) in the normal culture medium under standard culture conditions. The cells were fixed with 3.7% formaldehyde solution for 15 min, followed by washing with PBS three times and incubation in PBS with 0.1% BSA for 45 min. After washing with PBS, the cells were incubated for 45 min with anti-human antibodies CD24 PE (BioLegend, 311106) and CD44 APC (BioLegend, 338806) diluted (1:20) in PBS with 0.1% BSA. The CD24 and CD44 specimens were mounted on microscope slides, and expression was observed with an LSM 880 AiryScan (Zeiss).

For the exosome treatment experiment, s-TME CMFDA/DMSO-Exo and s-TME CMFDA/GW4869-Exo were treated onto Panc0203 cells and maintained in the standard culture condition for 20 h.

Short tandem repeat analysis

Purified genomic DNA of U937 cells, maternal PDAC cells (Panc0203, Panc0327), and pancreatic derivative cells (SP0926, SP1030, and SP09131A, 1B & 1C) were used for STR analyses using an AmpFLSTR™ Identifier™ PCR Amplification Kit (Applied Biosystems). Amplified PCR products were separated by capillary electrophoresis on a 3530xL DNA analyzer (Applied Biosystems) according to the manufacturer’s instructions and analyzed using GeneMapper v5 software (Applied Biosystems).

In vitro proliferation and viability assay

Five-thousand and 2.5 × 103 cells were seeded into 96-well plates for either a cell proliferation or viability assays. The numbers of live cells were estimated using an EzCytox WST assay kit (Daeillab) at 0, 24, 48, and 72 h and measurement was performed by microplate spectrophotometer (Agilent) at OD 450 nm. The relative cell proliferation and viabilities were estimated at each time point via comparison to, respectively, the number of viable cells at the initial plating time (0 h), and the number of viable cells in the drug-untreated control cultures at the corresponding times. For hypoxic (15% FBS) and LOS (2% FBS, in vitro ischemic) conditions, cells were cultured in a GasPak EZ Anaerobe Pouch System (BD Biosciences).

For the exosome treatment experiment, 2.5 × 103 Panc0203 cells were seeded into 96-well plates and s-TME CMFDA/DMSO-Exo and s-TME CMFDA/GW4869-Exo were treated for 24 h. Gemcitabine (250 μM) was applied for observing the effect on each exosome-treated group.

Aldefluor assay

An ALDEFLUOR assay kit (STEMCELL technologies) was used in accordance with the manufacturer’s instructions. Briefly, trypsin-treated Panc0203, SP0926 and SP1030 cells were collected followed by wash in FACS buffer. Total 3 × 106 cells per 1.5 ml Aldeflour buffer including 7.5 μl activated Aldefluor reagent. For the DEAB-treated control, total 1 × 106 cells per 0.5 ml were relocated from the original tubes to an e-tube including 20 μl DEAB, immediately. Then, both e-tubes were incubated at 37 °C for 40 min followed by ice incubation. The incubated cells were washed two times with Aldefluor buffer through centrifugation all tubes for 5 min at 250×g. Resuspended cells in 200 μl Aldefluor buffer were used for Flow cytometry which was performed using an LSRFortessa cell analyzer (BD Biosciences, 520–540 nm green fluorescence channel) to detect ALDH activity in the cells.

Tumorsphere assay

In the standard culture condition, 5 × 103 cells were incubated in ultra-low attachment, 6-well plates (Corning) containing DMEM/F12, HEPES (Gibco), which included B-27 (Gibco), and 1% PS solution, without disturbing the plates for 6–11 days. For the second-round experiment, the first-round tumorspheres were trypsinized to generate single cells followed by re-suspension with a syringe with a 31-G needle (BD). For CD24 and/or CD44 neutralization, 1 μg of anti-human CD24 antibody (BioLegend, 311102) and 0.15 μg anti-human CD44 antibody (BioLegend, 338802) were added as treatment into each well. To observe the effect of the MCT1 inhibitor, 20 μM of AZD3965 was added as treatment to each cell line. Spheroids were observed microscopically, and the number and diameter of tumorspheres (per field or total) were analyzed on ImageJ.

Invasion assay

Cells in serum-free RPMI 1640 (1 × 105 cells / 200 μl) were added to the upper compartment of Millicell inserts of 8.0-μm pores and 12-mm diameter (Merck Millipore) that were pre-coated with 200 μl (0.4 mg) of Matrigel (BD Biosciences). The basal compartment were filled with 750 μl of RPMI 1640 including 15% FBS, 1% PS solution, and 10 units/ml of human recombinant insulin. In order to observe the effect of the MCT1 inhibitor, 20 μM of AZD3965 (MedChemExpress) was added to each cell line at the seeding step in the prepared inserts and the basal compartment of the 24-well plate, respectively. To observe the effect of CD24 overexpression, pRP-hCD24 plasmid (Vectorbuilder) was transiently transfected into SP0926 cells. In order to observe the CD24 and CD44 inhibition effect, anti-human CD24 antibody (BioLegend, 311102) was added into the insert (0.4 μg/200 μl) and basal compartment (0.6 μg/750 μl), and anti-human CD44 antibody (BioLegend, 338802) was added into the insert (60 ng/200 μl) and lower compartment (90 ng/750 μl) for CD44 inhibition. Mouse IgG was used as a control (Invitrogen, 31903). After overnight or 24 h of incubation, the remaining cells on the apical side of the insert membrane were removed by cotton swabs, fixed in 3.7% formaldehyde for 2 min, and then exposed to 100% methanol permeabilization for 10 min followed by trypan blue staining for 20 min. Cells were washed with PBS three times between every step.

RNA sequencing

MØ-U937, Panc0203, SP0926, and SP1030 cells were prepared for RNA sequencing. A TruSeq Stranded mRNA LT Sample Prep Kit (Illumina) was used according to the manufacturer’s library protocol. Sequencing of the QC-passed libraries was performed on the NovaSeq 6000 sequencing system (Illumina) in paired-end with 101-nt read length. The pre-processed trimmed reads were mapped to the genomic DNA reference UCSC hg 19 (original GRCH3 7 from NCBI, Feb. 2009) using the HISAT2 program, which is capable of splice junction processing. After reading the mapping, transcript assembly was performed using the StringTie program. As a result, the expression profile values for each sample were obtained for the known transcripts, and the read count and FPKM (Fragment per Kilobase of transcript per Million mapped reads) were acquired based on the transcript/gene. From the original raw data (27,685 genes, 4 samples), processed data (14,657 genes, 4 samples) were obtained following exclusion of genes with an FPKM value of 0 in at least one of the four samples and by performing DEG (differentially expressed genes) analysis containing 7,360 significant genes that satisfy conditions of ≥2-fold change and <0.05 P-value conditions in at least one of the total comparison pairs, followed by hierarchical clustering (Euclidean distance, complete linkage). Fusion genes were analyzed by SOAPfuse, Defuse, FusionCatcher and STAR-Fusion based on RNAseq data, and compared with the results of DNA structural variants analyzed by Manta and Breakdancer following whole genome sequencing.

Gene set enrichment analysis

To identify gene sets enriched in derivative cells compared to mother Panc0203 cells, gene set enrichment analysis (GSEA) was performed using the javaGSEA desktop application (GSEA v4.0.3)69. For this analysis, the hallmark gene sets were applied using pre-ranked options derived from the rank of fold changes of derivative cells compared to Panc0203 cells. P-values were calculated by permuting the data 1000 times to find enriched gene sets. The GSEA software produced enrichment score (ES), normalized ES (NES), nominal P-value, and false discovery rate (FDR; Q-value). Gene sets that were up- or down-regulated with a Q-value of <0.05 were considered significant. For RNA sequencing, GSEA was performed using data from the TCGA Pancreatic Cancer (PAAD) database (n = 182, http://xena.ucsc.edu/), using the same parameters by dividing samples into two groups based on the median expression level of CD24, CD44, or MCT1.

Whole genome sequencing

For whole genome sequencing, libraries of Panc0203, SP0926 and SP1030 genomic DNAs were prepared according to the Illumina TruSeq DNA PCR-free library preparation guide and sequenced using Illumina HiSeq X sequencer following quality control and quantification with 151nt read length, which was performed by Macrogen. After generation of raw data through an integrated primary analysis software called RTA 2 (Real Time Analysis 2), next-generation sequencing data was aligned by Isaac Aligner 01.15.02.08., the mapping reference was the hg19 from UCSC, and analyzed by IVC (Isaac Variant Caller 2.0.13), SnpEff 3.3, Control-FREEC 6.4, Manta 0.20.2 and Breakdancer.

Quantitative real-time PCR

Total RNA was purified using a RNeasy Plus Mini kit according to the manufacturer’s instructions (Qiagen). Total RNA (1 μg) was transcribed into complementary DNA using a Maxime RT PreMix (Intron Biotechnology) for 1 h at 45 °C. Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control for normalization. The primer sequences for the total form hCD44, hCD24 and hSLC16A1 are listed below. Total form hCD44 (forward; 5′- CCGCTATGTCCAGAAAGGA -3′, reverse; 5′- CTGTCTGTGCTGTCGGTGAT-3′), hCD24 (forward; 5′-GACTCAGGCCAAGAAACGTC-3′, reverse; 5’-CCTGTTTTTCCTTGCCACAT-3′), and hSLC16A1 (forward; 5′- GTGGCTCAGCTCCGTATTGT-3′, reverse; 5′- GAGCCGACCTAAAAGTGGTG-3′).

Fusion gene PCR

Fusion genes were validated at both DNA (Supplementary Fig. 2d) and mRNA (data now shown) levels through fusion junction PCR and Sanger sequencing. PCR amplification of each fusion gene of genomic DNA was performed using primer sets listed below. BCL2L1/GLB1L2 fusion gene (forward; 5′- GGTAGTACCCACTGACAGAGTG-3′, reverse; 5′- CTCTTTCCCAGTCCCACCTC-3′), CABIN1/CSNK1E fusion gene (forward; 5′-CCTCCCCTAGCCCTGTGG-3′, reverse; 5′-CAGGGGAGCCTGCGTCATC-3′), TPTEP2-CSNK1E readthrough/ KIAA1671 fusion gene (forward; 5′-CTGGCTTGTGTCTGAACTG-3′, reverse; 5′-GATGGAGTGAGACTCCATCTC-3′).

Mass spectrometry for peptide analysis

Exosome of CMs collected from in vitro TME models (s-TME model and normal co-culture condition using fibroblasts, CAFs, MØ-U937 and Panc0203 cells) by total exosome isolation reagent (Invitrogen). Whole cell lysates of Panc0203, SP0926, and SP1030 cells, which were prepared in sample buffer (7 M Urea, 2 M Thiourea, 100 mM DTT, 4.5% CHAPS, 40 mM Tris). Peptides were analyzed following two-dimensional gel electrophoresis (2-DE) and liquid chromatography-tandem mass spectrometry (LC-MS/MS) methodologies. Nano LC-MS/MS analysis was performed with an Easy n-LC (Thermo Fisher Scientific) and an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) equipped with a nano-electrospray source. Samples were separated on a C18 nanobore column (150 mm × 0.1 mm, 3 μm pore size; Agilent). The mobile phase A for LC separation was 0.1% formic acid and 3% acetonitrile in deionized water, and the mobile phase B was 0.1% formic acid in acetonitrile. The chromatography gradient was designed for a linear increase from 0% B to 60% B in 9 min, 60% B to 90% B in 1 min, and 3% B in 5 min. The flow rate was maintained at 1,800 nl/min. Mass spectra were acquired using data-dependent acquisition with a full mass scan (380–1,700 m/z). For database searching, the Mascot algorithm (Matrixscience) was used to identify peptide sequences present in a protein sequence database. The peptides were filtered with a significance threshold P-value of <0.05. Database: nr_Human_20180410_curated (1249071 sequences; 414470512 residues).

Enzyme-linked immunosorbent assay

Lysates from Panc0203, SP0926, and SP1030 cells were used for ELISA, which was performed using kits for PGK1 (BIOMATIK) according to the manufacturer’s instructions. Conditioned media and exosomes, which were collected from both the normal co-culture TME condition and the s-TME models and conducted ELISA for detecting stratifin (LifeSpan BioSciences) and LGALS3BP (LifeSpan BioSciences). Briefly, prepared samples (100 μl) were added into each well followed by incubation for 2 h at 37 °C. After aspiration, detection reagent A (first detection antibody) was added to each well followed by incubation for 1 h at 37 °C. Following 3 times wash, detection reagent B (horseradish peroxidase-linked second detection antibody) was added and incubated for 1 h at 37 °C. TMB substrate solution was treated following 5 times wash, and observation was conducted until the color change of the standard and sample solutions is observed. After adding stop solution, the colorimetric absorbance was read at 450 nm.

Stable cell-line generation

Maternal pancreatic cells and pancreatic derivative cells were transfected with GFP shRNA, human CD24 shRNA plasmids (Santa Cruz) using Lipofectamine 2000 (Invitrogen) and were then selected by 0.4 μg/ml of puromycin (Gibco) until stable cell-line generation. CD24 and CD44 levels on the cells were analyzed by LSRFortessa cell analyzer using the anti-human antibodies CD24 PE and CD44 APC, following cell harvest and PBS rinse.

l-lactate assay

The specific measurement and analysis of l-lactate in collected media and cell lysates were performed by using L-lactate assay kit (Colorimetric) (Abcam) according to the manufacturer’s instructions. Panc0203, SP0926 and SP1030 cells were cultured for 72 h in standard culture condition following 3.5 × 105 cells seeding per six-well of plates. l-lactate concentration and amount were measured using the collected media and the lysates of 1 × 106 collected cells. Particularly, cells were harvested in 200 μl of lactate assay buffer (Abcam) generating cell lysates after wash with cold PBS, then each 50 µL sample (of the collected media and the lysates) per well was used for assay. Plates were incubated at room temperature for 30 min and measurement was conducted by a microplate spectrophotometer (Agilent) at OD 450 nm.

Western blotting

Panc0203, SP0926, and SP1030 cells were lysed in RIPA buffer (Thermo Scientific) containing protease inhibitor cocktail (Roche), and were centrifuged at 13,000 × g for 10 min at 4 °C. After protein quantification by the BCA method (Thermo Scientific), 20 μg of protein were resolved by SDS-PAGE and transferred to a nitrocellulose membrane followed by blocking with 5% skim milk in Tris-buffered saline for 1 h, and were incubated with anti-MCT1 Polyclonal Antibody (1:1000 dilution in 10 ml of Tris-buffered saline with 0.1% Tween® 20 detergent (TBST) with 5% BSA) (Thermo Fisher, A304–357A). The membranes were washed and incubated with horseradish peroxidase-conjugated secondary rabbit antibody, followed by enhanced chemiluminescence development according to the manufacturer’s instructions (Pierce). Western blot quantification was performed by ImageJ software. Note, MCT1 was detected as monomer forms (43–48 kDa) and a dimer form (95–100 kDa)70,71,72, when compared to ladder sizes of the used protein marker (PageRuler™ prestained protein ladder, Thermo Fisher, 26616). All blots were derived from the same experiment and were processed in parallel.

Statistics

All statistical analysis was performed using Prism 8 (GraphPad). Error bars on graphs display the mean and SD. One-way ANOVA followed by Tukey’s or Dunnett’s multiple comparison test or unpaired t test were used; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.