Ethics statement
Samples of human origin and the associated data were obtained from the IPC/CRCM Tumour Bank that operates under authorization # AC-2013-1905 granted by the French Ministry of Research. Prior to scientific use of samples and data, patients were appropriately informed and filed a written consent, in compliance with French and European regulations. The experiments were conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. Animal studies were conducted in agreement with the French Guidelines for animal handling and approved by local ethics committee (Agreement no. #16487-2018082108541206 v3). Of note, mouse weight loss > 20%, tumor necrosis, tumor volume > 1500 mm3, ruffled coat + hunched back, weakness, and reduced motility were monitored daily and considered as endpoints.
Cell culture
HT29, SW620, MKN45 and PANC1 come from ATCC (https://www.atcc.org/). CRC1 cells were established from a CRC biopsy (CHU-Carémeau, Nîmes, France, ClinicalTrial.gov Identifier#NCT01577511) as previously reported72. SUM159 and SUM149 was given by Dr. S.Ethier (Karmanos Cancer Center, Detroit, MI, USA), S68 was given by Dr. V. Castros (Université de Rennes, France). All cell lines were grown in the standard medium as previously described27. HeLa (WT, FANCD2KO, and ERCC1KO) were given by C. Lachaud (CRCM, Marseille)44 and cell were maintained at 37°C in DMEM (Dulbecco’s modified Eagle’s medium; Gibco) supplemented with 10% fetal bovine serum (FBS) and 1% Antibiotic-Antimycotic (Gibco, 15240-062).
To silence STAT3, STAT1 and FANCD2, SUM159 and S68 cells were transduced with lentiviral constructs: shSTAT3: pSMART-hCMV/TurboRFP (5’-AGCTGGAACAGATGCTCAC) (Horizon #V3SH7590-225360815) shSTAT1: pSMART-hCMV/TurboRFP (5’-CGAACATGACCCTATCACA) (Horizon #V3SH7590-226568750) shFANCD2#1: pSMART-hCMV/TurboGFP (5’ TGGTCCATCAACTACACCG) (Horizon #V3SH7590-228098457) shFANCD2#2: pSMART-hCMV/TurboGFP (5’ ACAATGAACAATTTCTTGG) (Horizon #V3SH7590-225852447) shCTRL: SMARTvector Lentiviral controls hCMV (Horizon #S-005000-01).
CRISPRi
To silence STAT3, SUM159, S68 and SW620 cells were co-transduced with two lentiviral constructs: pHR-SFFV-KRAB-dCas9-mCherry (Addgene Plasmid #60954,32and pU6-sgRNA EF1Alpha-puro-T2A-BFP (Addgene Plasmid #60955,32, the latter encoding a control (5’-GCGCCAAACGTGCCCTGACGG) or STAT3-targeting sgSTAT3#1 (5’-GGTTCCGACGTCGCAGCCGA) and sgSTAT3#2 (5’-AACAAGCCCCAACCGGATCC). Lentiviral infection was conducted by plating 250.000 cells on 6-well plates and incubating them overnight (o/n) with 1mL of culture medium, polybrene (8 µg/mL), and 5–10 µL of lentivirus (MOI = 2). Cells were then washed twice with PBS and expanded in their usual culture medium. Cell sorting was performed with a FACS Aria III instrument (BD Biosciences) to enrich mCherry and BFP double-positive cells.
To silence ALDH1A1 SUM159/S68-KRAB cells were transduced with the lentiviral vector pLKO.1-blast-U6-sgRNA-BfuA1-stuffer encoding 2 sgRNAs per target gene (ALDH1A1 sgRNA1: 5’-TGATTCGGCTCCTGGAACAC and sgRNA2: 5’- AGGTAAGTCTGGCGTGCCTG) as previously described73.
Drugs
Cell lines were continuously treated in adherent conditions with Nifuroxazide (NIF) (stock concentration SC = [10 mM], Selleckchem, S4182), 5-Nitro-2-furaldehyde (M2) (SC = [20mM], TCI, N0387), 4-Hydroxydenzhydrazide (HBH) (SC = [20mM], Alfa Aesar, A12702), Methyl 5-amino-2-furoate (M7) (SC = [20mM], Alfa Aesar, L05958), Olaparib (PARPi) (SC = [10mM], Selleckchem, S1060), Melphalan (SC = [20mM] Merck,M2011), Napabucasin (STAT3i) (SC = [10mM] TargetMol, T3218), Ruxolitinib (JAK2i) (SC = [10mM], Selleckchem, S1378), NIF-C, M2-C and HBH-C (SC =[10mM], from R. Rodriguez’s laboratory, Institut Curie, France, see supplementary methods), Melphalan-C (SC=[10mM], from C. Lachaud laboratory, CRCM, France)44. All compounds were resuspended in dimethyl sulfoxide (DMSO, Sigma). IL-6 human (Sigma, SRP3096) resuspended according to the manufacturer's instruction. For the in vivo experiments, NIF [300mg/kg] (Euromedex, TA-T1563) and Olaparib [50mg/kg] was resuspended in a solution of 25% cremophore/ethanol.
Screening strategy
An automated screening routine was developed on a robotic workstation equipped with a 96-well head probe (Nimbus, Hamilton) to screen a repurposing drug library (1,280 FDA-approved drugs, Prestwick Chemicals). Briefly, 45µL of cellular suspension was layered with automation into the wells of collagen‐coated, clear bottom, black‐walled 384‐well culture plates (Greiner µClear plates, Cat# 781091). Starting cell culture conditions were as follows: 1,400 MKN45 cells/well; 1,800 PANC1 cells/well; 1,000 SUM159 cells/well and 3,500 T84 cells/well. Plates were then incubated for 4 hours at 37°C and 5% CO2 in a humidified incubator to allow for cell attachment. Plates were then returned to the robotic workstation and 5µL of sample or control drugs were layered on top of the cell cultures (1 drug/well, final drug concentration: 10µM). Each drug from the library was tested as a separate triplicate in different well positions of three independent culture plates to minimize positional errors. Each culture plate also received different positive and negative controls: eight wells received medium alone (“Untreated” well, negative controls), twelve received the DMSO vehicle at 0.1% (v/v) final (“DMSO” Wells, negative control, Sigma), four received Doxorubicin at 5µM final (“Doxo” wells, positive cytotoxic control, Sigma), and four received ST102 at 50nM final (« ST102 » wells, positive control). Additionally, four wells were left untreated to receive the DEAB control during the ALDEFLUOR assay (see below).Three days post‐treatment, cell amount and the %ALDHbr cell amount (=%CSC) were assessed as previously described23.
ALDEFLUOR assay
The analysis was processed on single-cell suspension from cell lines. The ALDEFLUOR Kit (Stem Cell Technologies, #01700) was used to isolate population with differential aldehyde-dehydrogenase enzymatic activity and analyzed using an LSR2 cytometer (Becton Dickinson Biosciences) as previously described4.
Immunoblot analysis
Cells were lysed in ice-cold lysis buffer containing Hepes 50 nM, pH 7.5, EDTA 1 mM, pH 7, NaCl 150 mM, NaF 100 mM, Na3VO4 1 mM, Triton X-100 1%, and complete Proteinase Inhibitor Cocktail (Roche, #04693159001). Cell lysates were migrated in 4–12% SDS-PAGE (Sodium Dodecyl Sulfate–PolyAcrylamide Gel Electrophoresis). The following primary antibodies were used: anti-ALDH1A1 (mAb, Clone 44, Becton Dickinson, 1/200) anti-CyclinD1 (rabbit mAb, Cell Signaling #55506, 1/1,000), anti- p-STAT3 (Tyr705) (mouse mAb, Cell Signaling #4113, 1/2,000), anti-STAT3 (mouse mAb, Cell Signaling #9139, 1/1,000), anti-FANCD2 (rabbit mAb, abcam, ab108928, 1/1,000), anti-FANCI (santa cruz sc-271316, 1/1,000) anti-ɣH2AX (rabbit mAb, Cell Signaling #9718, 1/1,000). Detection of GAPDH (Rabbit pAb, Cell Signaling, 1/5,000) or α-Actin (mouse mAb, Sigma Aldrich #A5441, 1/5,000) was used as loading control.
Click-chem Fluorescence
After cell sorting, cells were cytospun and fixed with 4% paraformaldehyde for 10 minutes and permeabilized with 0.1% Triton X-100 for 5 minutes before blocking with protein block (Dako). Click reaction were performed with azide alexa fluor 594 (Click-iT EdU AlexaFluor594 Imaging Kit, C10339) according to the manufacturer’s instructions. After 10 minutes of wash with TBST, DNA was counterstained with DAPI 4′,6-diamidino-2-phenylindole (Invitrogen, ProLong Gold antifade reagent with DAPI, P36935). Images were acquired using epifluorescence microscope Leica.
Immunofluorescence
After cell sorting, cells were cytospun and fixed with 4% paraformaldehyde for 10 minutes and permeabilized with 0.1% Triton X-100 for 5 minutes before blocking with protein block (Dako). Cells were labeled 1 hour at room temperature with an anti-phospho-Histone H2AX (Ser139, clone JBW301, Merck Millipore, 1/1,000) or with an anti-RAD51 (gift from M. Modesti lab, CRCM, Marseille, 1/1,000). After 10 minutes of wash with TBST, cells were incubated for 30 minutes with anti-mouse (A-11029, ThermoFisher), 1/500). DNA was counterstained with DAPI 4′,6-diamidino-2-phenylindole (Invitrogen, ProLong Gold antifade reagent with DAPI, P36935). Images were acquired using Nikon AX confocal microscope equipped with a 63× objective. Cells with more than 8 foci for ɣH2AX or RAD51 were considered as positive cells. For each condition, immunofluorescence scoring was done on 100 cells in three independent experiments.
mRNA extraction and quantitative real-time RT-PCR
Total RNA was isolated using the Maxwell RSC simply RNA Tissue Kit (AS1340) according to the manufacturer’s instructions. cDNA was synthesized from 1 µg of RNA with the Transcriptase inverse SuperScriptIV kit (Invitrogen; 18090050). Real-time PCR amplification and analysis were conducted with the TaqMan Universal Master Mix II with UNG on a 7500 Real-Time PCR System (Applied Biosystems). RNA levels were normalized to ACTB expression using the DDCt method. Probe; STAT3 (ThermoFisher, 4331182, Hs00374280_m1) FANCD2 (ThermoFisher, 4331182, Hs00276992_m1) FANCI (ThermoFisher, 4331182, Hs01105308_m1) FANCM (ThermoFisher, 4331182, Hs00326216_m1) FANCA (ThermoFisher, 4331182, Hs01116668_m1) FANCF (ThermoFisher, 4331182, Hs00256030_s1)
Clonogenic survival analysis
300 SUM159 cells, 500 HeLa cells, 1,200 SUM149 cells and 1,500 S68 cells were plated in triplicate in 10 cm dishes in complete growth medium. After cells attached, they were treated with the indicated dose of compound. After 10–15 days, cells were washed, fixed and stained with acetic acid : methanol (1:7) and 1% of Coomassie blue 1hour at room temperature. The number of colonies with > 100 cells was counted. For each genotype, cell viability of untreated cells was defined as 100%. Data are represented as means ± SD from three independent experiments.
Reverse comet assay
Reverse alkaline comet assay was done as described previously74. Briefly, cells were initially treated with DMSO vehicle or NIF. After 4 hours of treatment and cell sorting cell samples were treated with PBS as a control or IR (10Gy). Cells were subsequently resuspended in molten 1% Type VII low gelling temperature agarose and then allowed to set on home-made glass slides pre-coated with agarose74. Cells were then lysed by bathing slides in ice-cold lysis buffer (100 mM Na2EDTA, 2.5 M NaCl, 10 mM Tris–HCl (pH 10.5), 1% Triton X-100) for 60 minutes and then subjected to 4 × 15 minutes washes with ice cold MilliQ H2O. Each slide was then submerged in alkali electrophoresis buffer (300 mM NaOH, 1 mM Na2EDTA) for 60 minutes and then electrophoresed at 20 V for 20 minutes at 4°C. Samples were neutralized by the addition of neutralization buffer (500 mM Tris–HCl pH 7.5) for 10 minutes and then allowed to dry overnight at ambient temperature. Comets were stained with SYBR green for 10 minutes and then washed using 3 × MilliQ H2O washes. Samples were visualized using an Apotome microscope and the level of DNA damage assessed using OpenComet. At least 100 comets were scored per slide.
Quantitative image-based cytometry (QIBC)
The nucleus was extracted and stained as previously described38. Briefly, after treatment and 45 minutes of EdU incubation, the nucleus was extracted with CSK buffer (50mM NaCl, 25mM Hepes pH7.4, 3mM MgCl2, 300mM sucrose, 0.5% triton, 1mM EDTA and protease and phosphatase inhibitor) 5 minutes on ice. Buffer was removed and nucleus fixed with PFA 2% 10 minutes at room temperature. For click analysis, the nucleus was stained with azide alexaFluor 594 (Click-iT EdU AlexaFluor594 Imaging Kit, C10339) according to the manufacturer’s instructions, wash nucleus with Washing buffer (BD Perm/Wash, 51-2091KZ). Nucleus were stained with analysis buffer (DAPI 0.5µg/ml and RNase 250µg/ml). For cell cycle analysis nuclei were stained only with an analysis buffer. These nuclei are analyzed by cytometry on Aurora (Cytek).
Tumosphere forming assay
Cell lines were plated on 96-well plates (pre-coated with Poly(2-hydroxyethyl methacrylate) at 56°C overnight) in serum-free mammary epithelial basal medium (MEBM, Lonza, CC-3151) supplemented with B-27 (Gibco, 17504-004), 20 ng/mL EGF (Gibco, PMG8043), 1X Antibiotic-Antimycotic (Gibco, 15240-062), 1 ng/mL hydrocortisone (Sigma, H0888), 5 µg/mL insulin (Lily, VL7510, from IPC). Frequency of tumorigenic cells with tumorsphere-forming ability was determined following the guidelines of the Extreme Limiting Dilution Analysis (ELDA)75. Briefly, a range of cells (1 to 25 for SUM159 and 1 to 100 for S68 and SUM149) was plated in each well, and the number of wells containing at least one tumorsphere was computed after 10–15 days of culture. 30 to 80 wells were evaluated per condition.
Apoptosis assay
After cell sorting, SUM159 cell line was treated in adherent conditions for 72 hours. According to the manufacturer’s instructions, at the end of treatment, the cells was resuspended in binding buffer 1X (0.1M hepes pH = 7.4, 1.4M NaCl and 25mM CaCl2 for 10X buffer), with 100,000 cells in 100µl of buffer. Then, cells were stained with 5µl of Annexin V-FITC (ThermoFisher; A13201) and 10µl of Propidium Iodide (ThermoFisher; 556463). Cells were incubated 15minutes at RT in dark and analyzed by spectral flow cytometry.
RNAseq
For analysis of SUM159 and SW620, three independent biological replicates of ALDHneg and ALDHbr cells were isolated by cell sorting for control condition of treated by NIF during 30 hours. Total RNA was extracted as described above and its quality was assessed by Tapestation (only samples with RIN score > 8 were considered for sequencing). The sequencing and GSEA analysis were performed by MGX-Montpellier GenomiX core facility.
CUT&RUN qPCR
Cut&Run was performed as described previously76. Briefly, after 1 or 6 hours of IL-6 stimulation SUM159 cells were washed twice with Wash buffer (20 mM HEPES pH = 7.5, 150 mM NaCl, 0.5 mM spermidine (Sigma; S2626-5G) supplemented with protease inhibitors). Cells were then resuspended in Binding buffer (20 mM HEPES pH = 7.9, 10 mM KCl, 1 mM CaCl2, 1 mM MnCl2) containing 10 µL of blocked BioMag®Plus Concanavalin A-coated beads (CliniSciences, #86057-10). After 10 minutes at RT, cells:beads were transferred to 50 µL of antibody solution (1:100 dilution of primary antibody (STAT3; ab171360 and H3K27ac; active motif; #39133) in Wash buffer plus 0.1% digitonin) and incubated for 1 hours at RT on an end-to-end rotator. Cells:beads were washed three times with Wash buffer-0.1% digitonin and incubated for 10 minutes with pA-MNase (given by E. Pasquier, CRCM, France)76 diluted in Wash buffer-0.1% digitonin. After three washes in Wash buffer-0.1% digitonin, tubes were placed in an ice/water bath and equilibrated to 0°C for 10 minutes. To trigger digestion of the DNA, CaCl2 was added (2 mM final concentration) for 30 minutes. Digestion was stopped by addition of 2X STOP buffer (340 mM NaCl, 20 mM EDTA, 4 mM EGTA, 0.02% digitonin, 1:200 RNase A [50 µg/mL final concentration]). Digested DNA fragments were released from cells by incubating sample tubes on a heat block at 37°C for 10 minutes. The supernatant was then recovered by placing tubes on a magnetic stand and DNA was purified using the NucleoSpin Gel and PCR Cleanup kit (#28104). DNA fragment has been analysed by q-PCR in 384 wells with iQ SYBRE green supermix according to the manufacturer’s instructions. We used the probes for STAT3 promoter (Fw: ATGACCGGAATGTCCTGCTG and Rv: TCACGCACTGCCAGGAAC) FANCI (Fw: CTCCGACTGTGAGCTGGGA and Rv: ATGAAGACTGAAGGGGTGCC). Data were normalized with GAPDH probe (Fw: ACTCACCCTGCCCTCAATATC and Rv: AGACAGTGTGCCTTTCATTCCAT).
FANCI reporter
To study effect of STAT3 and M2 on FANCI expression we used a reporter assay (LightSwitch Luciferase vector) with luciferase expression under the control of FANCI or GAPDH promoter as control vector (Active Motif). 24 hours post-transfection, luciferase expression was revealed using a LightSwitch Dual assay kit (Active Motif; #32031) according to the manufacturer’s instructions. Data were normalized to the GAPDH expression.
Mutation and copy-number detection
For each PDX, we identified molecular alterations array-comparative genomic hybridization (aCGH) as previously described60. aCGH was done using high resolution 244 K CGH microarrays (Hu-244A, Agilent Technologies). To determine copy-number alterations in each PDX, we mapped all aCGH probes according to the hg19/NCBI human genome mapping database. The copy number was estimated for each gene by taking the value of the segment with the highest amplitude, then categorized into “Amplified” (Log2ratio > 1), “Gain” (0.5 < Log2ratio < = 1), “Loss” (− 1 < = Log2ratio < − 0.3), and “deletion” (Log2ratio < − 1). Focal events were defined as genomic alterations with a size less than 5 Mb and a copy number higher than the surrounding segments. The percentage of genome altered was calculated as the sum of altered probe divided by the total number of probes. To determine the mutation profile of each PDX, we combined two data analysis pipelines as described previously 60. Briefly, the first pipeline used FreeBayes version 0.9.9 for single-nucleotide variant (SNV) calling and insertions/deletion (indel) calling was done using GATK haplotype caller version 2.5-gf57256b with default parameters. For the second pipeline SNV calling was done with Mutect 1.7 and somatic indel calling with scalpel. All variants were then annotated for genes and function using ANNOVAR (version 2013 − 1112). In order to remove false positives, recurrent variants with no entry in public databases such as COSMIC or dbsnp were removed. Variants identified by both pipeline analyses were retained as somatic.
Animal models
In this study, we utilized four primary human breast cancer xenografts generated from four different patients (CRCM434, CRCM494, Pandora21 and Pandora7). These patient-derived xenografts (PDXs) were generated triple-negative breast tumors59. For each PDX, we determine HRD status using the SOPHiA DDM™ GIInger Genomic Integrity Solution that is based on Low-pass whole genome sequencing. We utilized these PDXs to perform preclinical assay in vivo. Cells from these PDXs were transplanted orthotopically into fat pads of NSG female mice without cultivation in vitro. We injected 100,000 cells per fat pad of NSG mice (with one injected fat pads per mouse) and monitored tumor growth. When tumors reached an average size of 10–150 mm3, mice were randomized (n = 8, i.e., 8 tumors for each PDX and for each group) and used to determine the response to the treatment. We initiated treatment with NIF (i.p., 300 mg/kg, 5 out of 7 days, 3 weeks), alone, PARPi alone (i.p., 50 mg/kg, 5 out of 7 days, 3–4 weeks), NIF/PARPi combination, or placebo injected with a solution of 12.5% ethanol/12.5% cremophore/75% water. After 3–4 weeks of treatment, mice from each group were sacrificed according to ethic statements. Tumors was dissociated into single cells. These cells were reimplanted into secondary NSG mice. We performed serial dilution to functionally evaluate the proportion of residual CSCs in each group of treatment (CTRL, NIF, PARPi, and NIF + PARPi) from the 4 different PDXs. Each mouse that presents a tumor reaching a size of 100 mm3 was considered as a tumor bearing mouse.
Patient-derived organoids (PDXOs)
To grow organoids from PDX models (CRCM494, CRCM434, Pandora21, and Pandora7), 250,000 cells were resuspended in 28µL of culturex (Biotechne), seeded on a 48-well plate, and cultured in 400µL of medium supplemented with 10µM of L-Y27632 (Sigma, G9145) as previously described77. After 7–10 culture days, PDXOs were passed and only PDXOs > 40µm were replated (400 organoids per well) in 400µL of medium without L-Y27632. After 3–7 days of culture, PDXOs were treated with 25µM of M2-C, or Melphalan-C for 4 hours. After treatment all wells was wash twice with PBS, and 400µL of fresh medium was added. After 0,24,72 and 144 hours of treatment, PDXOs were dissociated with TrypLE Express 1X (Gibco, #12605-010) during 15 minutes at 37°C under agitation (155 RPM), fixed in 70% ethanol, and immediately stored at -20°C. For ICLick analysis, cells were stained with azide Alexa Fluor 594 (Click-iT EdU AlexaFluor594 Imaging Kit, C10339) according to the manufacturer’s instructions. Cells were stained with analysis buffer (DAPI 1/5,000). These cells were analyzed by cytometry on LSRII (BD).
Statistical
Graphpad Prism 5.0 was used for data analysis and imaging. The results are presented as mean ± SD for at least three repeated independent experiments. To investigate associations among variables, using nonparametric Wilcoxon rank-sum test, ANOVA and Sidak or Dunn’s test, chi-squared test or Fisher’s exact test when appropriate. Extreme limiting-dilution analysis (http://bioinf.wehi.edu.au/software/elda/) was used to evaluate breast CSC frequency. In all cases, a p-value < 0.05 was considered as statistically significant.