Cell culture
A549 (RRID:CVCL_0023), H460 (RRID:CVCL_0459), H1703 (RRID:CVCL_1490), H2030 (RRID:CVCL_1517) and HEK293T (RRID:CVCL_0063) cells were obtained from the American Type Culture Collection, LK2 (RRID:CVCL_1377) cells were obtained from AstraZeneca and all parental and clonally derived cell lines were further authenticated by short tandem repeat profiling. Cells were cultured at 37 °C and 5% CO2 in a humidified atmosphere in all instances. A549 cells were cultured in F12-K medium (Gibco, 21127-022) supplemented with 10% FBS (VWR Seradigm, 97068-085), 0.1 mM nonessential amino acids (Gibco, 11140-050), 1 mM sodium pyruvate (Gibco, 11360-070) and 100 U per ml penicillin plus 100 µg ml−1 streptomycin (Gibco, 15140-122). H460, H1703, H2030 and LK2 cells were cultured in RPMI medium (Gibco, 22400-071) supplemented with 10% FBS (VWR Seradigm, 97068-085) and 100 U per ml penicillin plus 100 µg ml−1 streptomycin (Gibco, 15140-122). HEK293T cells were cultured in DMEM (Gibco, 11965-084) supplemented with 10% FBS (VWR Seradigm, 97068-085), 1 mM sodium pyruvate (Gibco, 11360-070) and 100 U per ml penicillin plus 100 µg ml−1 streptomycin (Gibco, 15140-122). All cells were regularly verified to be free of Mycoplasma using MycoAlert (Lonza) or MycoStrip (InvivoGen) Mycoplasma detection kits.
Reagents
Unless otherwise specified, DDRi were always used at the following concentrations: 1 µM PARPi (olaparib), 1 µM DNAPKi (AZD7648), 10 nM ATMi (AZD0156), 0.5 µM ATRi (ceralasertib), 0.3 µM WEE1i (adavosertib), 1 µM PARPi + 1 µM DNAPKi, 1 µM PARPi + 10 nM ATMi, 1 µM PARPi + 0.5 µM ATRi or 1 µM PARPi + 0.3 µM WEE1i. All drugs were provided by AstraZeneca. Other drugs were used at the following concentrations: 0.156 nM docetaxel (Selleck Chemicals), 3.125 nM cobimetinib (Selleck Chemicals), 500 U per ml CAT (Sigma-Aldrich), 200 U per ml SOD1 (Sigma-Aldrich), 2 mM NAC (Sigma-Aldrich), 2 mM GSH ethyl ester (Cayman Chemicals), 100 µM Trolox (Selleck Chemicals), 2 µM ferrostatin 1 (MedChemExpress), 5 mM ascorbic acid (Thermo Scientific Chemicals) and 0.5 µM TCEP (Thermo Scientific Pierce). These concentrations were proven effective in biological assays here or elsewhere33,34,49. DFO (MedChemExpress), FAC (Sigma F5879), L-alanine (Thermo Scientific Chemicals, A15804.14) and D-alanine (Thermo Scientific Chemicals, A10231.06) were used as indicated.
Lentiviral packaging and cell line generation
Lentiviral packaging was performed in HEK293T cells using TransIT-LT1 (Mirus, 2300) or PEI MAX (Polysciences) transfection reagents and transfer plasmid, VSVG (Addgene, 8454) or pMD2.G (Addgene, 12259) and pCMV-dR8.2 dvpr (Addgene, 8455) or psPAX2 (Addgene, 12260). The medium was changed after 24 h and lentiviral supernatant was harvested 48–72 h after transfection and passed through a 0.45-µm PVDF filter (Millipore) before downstream use.
A549, H1703 and H2030 cells stably expressing dCas9–KRAB (CRISPRi) were generated by transduction using lentiviral particles generated from Addgene plasmid 102244. BFP-positive cells were single-cell sorted into 96-well plates using a Sony SH800 or BD FACS Aria II, expanded and selected on the basis of maximal uniform repression or activation of cell surface marker CD55 or maximal depletion of sgRNA targeting the essential gene PLK1.
A549 cells stably expressing PRDX1, PRDX1-C52S, PRDX1-C173S, PRDX1-C52S;C173S, PRDX1-NLS, PRDX1-NES, PRDX2, PRDX3, cytoPRDX3, PRDX4, cytoPRDX4, PRDX5, cytoPRDX5, PRDX6, FLAG–CAT and SOD1–FLAG were generated by transduction with lentivirus generated from POI-IRES-BFP pLEGO plasmids cloned in this work.
CRISPR screening
For single-guide CRISPRi screens, A549 dCas9-KRAB CRISPRi cells were transduced at 500× coverage and low multiplicity of infection (MOI) of ~0.3 with pooled sgRNA libraries obtained from Addgene (83969 and 83978). Cells were selected with 2 µg ml−1 puromycin 48 h after transduction with sgRNA library and left for 5 days to complete selection. Following selection, 100 million cell aliquots were frozen down to represent the T0 population and remaining cells were split into treatment conditions. Treatment conditions were as follows: DMSO, 1 µM PARPi, 1 µM DNAPKi, 10 nM ATMi, 0.5 µM ATRi, 0.3 µM WEE1i, 1 µM PARPi + 1 µM DNAPKi, 1 µM PARPi + 10 nM ATMi, 1 µM PARPi + 0.5 µM ATRi and 1 µM PARPi + 0.3 µM WEE1i. Treatment conditions were carried out in duplicate, maintained at 1,000× coverage and passaged every 3–4 days for 2 weeks. At the end of 2 weeks, 100 million cell aliquots were frozen down for each treatment condition and prepared for next-generation sequencing (NGS).
For dual-guide CRISPRi screens, wild-type and PRDX1-KO A549 dCas9-KRAB CRISPRi cells were transduced at 500× coverage and MOI of ~0.3 with pooled dual-sgRNA libraries generously donated by J. Weissman35. Cells were selected with 3 µg ml−1 puromycin at 48 and 72 h after transduction and then recovered in fresh medium for an additional 3 days. On day 6, aliquots of 25 million cells were frozen to represent T0 populations and the remaining cells were split into treatment conditions comprising wild type + vehicle (DMSO), wild type + 0.625 µM DNAPKi, PRDX1 KO + vehicle (DMSO) and PRDX1 KO + 0.625 µM DNAPKi. Treatment conditions were carried out in triplicate, maintained at 1,000× coverage throughout and passaged every other day for 11 additional days. On day 11 after treatment (day 17 after transduction), aliquots of 25 million cells per condition were frozen for library preparation.
NGS library preparation
For single-guide CRISPRi screens, genomic DNA was isolated using NucleoSpin Blood XL gDNA extraction kits (Macherey-Nagel) according to the manufacturer’s protocol and quantified by nanodrop. Next, the sgRNA cassette was PCR-amplified using a forward primer within the mU6 promoter, which contains an Illumina TruSeq index for sample multiplexing and a common reverse primer within the sgRNA constant region. PCR was performed with NEB Next Ultra II Q5 polymerase (New England Biolabs, M0544), allowing 10 µg of input DNA per 100-µl reaction. All purified gDNA was subject to PCR amplification with the following cycling conditions: denaturation at 98 °C for 30 s, followed by 22 cycles of denaturation at 98 °C for 10 s then annealing and extension at 65 °C for 75 s, with a final extension at 65 °C for 5 min. All reactions from each sample were subsequently pooled and purified by a double-sided 0.65–1× solid-phase reversible immobilization (SPRI) selection. First, 195 μl of SPRI beads were added to 300 µl of pooled PCR sample, incubated briefly at room temperature and separated by DynaMag. The supernatant was transferred, mixed with an additional 300 μl of SPRI beads and separated by DynaMag. The beads were washed twice with 80% ethanol and air-dried; then, the amplicons were eluted in water. Sample purity and concentration were determined using an Agilent Bioanalyzer and used to calculate equimolar sample pooling. Libraries were sequenced on a HiSeq4000 (Illumina) with single-end 50-bp reads to achieve >50 million reads per sample (that is, >500 reads per library element).
Dual-guide library screens were prepared in much the same way with minor modifications because of the number of library elements and the increased size of the dual-sgRNA cassette. Briefly, genomic DNA was purified with NucleoSpin Blood L gDNA extraction kits (Macherey-Nagel). PCR was conducted with a common forward primer within the mU6 promoter and reverse primers within the sgRNA constant region 3′ of the second sgRNA that contain a sample-specific index. PCR amplification was performed with the following cycling conditions: denaturation at 98 °C for 30 s, followed by 22 cycles of denaturation at 98 °C for 10 s then annealing and extension at 67 °C for 75 s, with a final extension at 72 °C for 5 min. Amplicon purification was performed with a 0.5–0.65× double-sided SPRI selection. Libraries were sequenced on a NovaSeq6000 (Illumina) with paired-end reads (19 × 5 × 0 × 19) spiking in custom read 1, read 2 and index read 1 primers with 10% PhiX to achieve >10 million reads per sample.
Screen analysis
ScreenPro2 package is a flexible analysis tool for high-content CRISPR screening including single-guide (V2) or dual-guide (V3) CRISPRa/i screens. This tool’s website (https://arcinstitute.org/tools/screenpro2) and open-source code (https://github.com/ArcInstitute/ScreenPro2) are available online.
In the ScreenPro2 pipeline, FASTQ files from deep sequencing of sgRNA constructs are first processed using the ‘GuideCounter‘ module. This module first counts unique sequence events across all reads and then maps them to reference guide sequences provided in a library table. The output count matrix with guide RNA counts per sample is then available for downstream analysis.
The phenotype calculation module in ScreenPro2 is mostly derived from our previous screen processing pipeline50 with some improvements and additional analytical features. Element frequencies are normalized using the ‘deseq2_norm‘ preprocessing function from PyDESeq2 (ref. 51) to adjust for differences in sequencing depth. To calculate phenotype scores in each pair of conditions, the log2 enrichment of the normalized counts for guide elements across replicates are compared between the pairs of conditions (for example, treated versus untreated) and phenotype scores are normalized to negative control elements using the ‘PhenoScore’ module in ScreenPro2. As an optional feature, the phenotype scores can use a growth factor to further normalize the phenotype score. For the single-guide (V2) screens, the counts for group of guides targeting same genes or transcripts across multiple replicates are compared between two screen arms and then aggregated to assign gene-level phenotypes. For the dual-guide (V3) screens, there is usually only one guide element per target gene or transcript; thus, only counts from multiple replicates are compared between screen arms. For each comparison, the counts across each screen arm were used in dependent paired-sample t-tests and then adjusted P values were calculated using the Benjamini–Hochberg method.
Specifically for drug screens, phenotype scores for vehicle-treated (γ) or drug-treated (τ) conditions were calculated by comparing endpoint samples from vehicle or drug conditions to T0 samples (Extended Data Fig. 1a). The drug sensitivity and resistance phenotypes (ρ) were calculated by comparing drug-treated relative to vehicle-treated conditions (Extended Data Fig. 1a). CRISPR screening data analyzed in this manner can be normalized to a growth factor equal to the population doubling difference between conditions if desired. In this manuscript, the data shown in Fig. 1c, Extended Data Figs. 1b and 3a and associated supplementary datasets are not growth normalized, while the data shown in Fig. 5a,b, Extended Data Fig. 5c,d and associated supplementary datasets are growth normalized because of the higher selective pressure. To assess pathway-level enrichment of gene phenotypes, we used blitzGSEA, a Python package for the computation of gene set enrichment analysis (GSEA; https://github.com/MaayanLab/blitzgsea)52. For each GSEA analysis, the drug phenotypes and GO gene sets were used. Positive normalized enrichment scores (NESs) correspond to gene sets enriched among positive ρ phenotypes (that is, resistance phenotypes) and negative NES corresponds to gene sets enriched among negative ρ phenotypes (that is, sensitivity phenotypes). Additional graph plotting was performed in R (version 4.3.1).
For benchmarking our screens against the ‘gold-standard’ DDRi screen, we selected sensitizing genes from the published datasets for PARPi19, ATRi18, ATMi17 and WEE1i16. First, we defined individual gene sets for each of these conditions. To evaluate the quality of defined inhibitor gene sets, we used gget enrichr to report the top enriched GO terms for each initial gene set53. Lastly, we used each of the gene sets for performed GSEA analysis on all single-agent phenotype scores (note that genes overlapping with the PARPi gold-standard list were excluded for the other three conditions to focus on more drug-specific gene sets). The final results are reported as NES bar graphs and in Supplementary Dataset 2.
For the graph analysis, gene-level phenotypes for all single-agent and combination treatment CRISPRi screens were merged and significant genes were defined using an empirically derived threshold of <−6. These gene lists were then used to create a single-drug–gene knowledge graph dataset using the Knowledge Graph Mastery module from TDC (https://github.com/mims-harvard/TDC/)54. These data were used to perform downstream graph analysis such as node clustering or betweenness centrality analysis using the igraph package in Python. A drug sensitivity and resistance subgraph was constructed by filtering drug–gene interactions only exhibiting drug sensitivity or resistance phenotypes, respectively. To evaluate enrichment of consistent pathways across treatments, we performed GSEA analysis of GO gene sets using the gene connectivity (vertex degree) values as numeric inputs for full, resistance and sensitivity networks in separate analysis. Results were shown as rank plots and genes in one selected GO term visualized as a heat map. To highlight common sensitizing genes across all drug treatments, we performed a betweenness analysis on sensitizing subnetwork in which nodes with a single edge were filtered.
All code for preprocessing primary and secondary screens, assessing screen quality and conducting integrative graph analyses is publicly available on Zenodo (https://doi.org/10.5281/zenodo.21302988)55.
Individual sgRNA plasmid cloning
Individual sgRNA plasmids were cloned into the CRISPRi/a-V2 library parental plasmid (Addgene, 84832) as previously described14. Briefly, the forward and reverse complement primers of each sgRNA were annealed by preincubation at 37 °C for 30 min in the presence of T4 polynucleotide kinase (New England Biolabs), followed by incubation at 95 °C for 5 min and then ramp down to 25 °C at 5 °C min−1. The annealed sgRNA inserts were then ligated into BstXI and BlpI-digested CRISPRia-v2 plasmid using Quick Ligase (New England Biolabs). For dual-sgRNA plasmids, cassettes encompassing sgRNA1–constant region–U6 promoter–sgRNA2 plus 20-nt overhangs on the 5′ and 3′ end were inserted into BstXI and BlpI-digested CRISPRia-v2 plasmid by Gibson assembly with HiFi assembly mix (New England Biolabs). Dual-sgRNA cassettes were synthesized as gene fragments (Twist) or produced by PCR. For the PCR method, oligos comprising 20-nt Gibson overhangs, the desired protospacer sequence and a region complementary to the dual-guide backbone were used to PCR-amplify the entire dual-sgRNA cassette, which was then gel-purified and used in Gibson reactions as above. All sgRNA constructs used in this study are detailed in Supplementary Dataset 5. It should be noted that, in some instances, sgRNAs used for validation share bidirectional promoters with other genes (for example, ANKRD49 with MRE11 and DZIP3 with CIP2A).
cDNA cloning
To generate PRDX1–PRDX6–IRES–BFP pLEGO, FLAG–CAT–IRES–BFP pLEGO and SOD1–FLAG–IRES–BFP pLEGO constructs, pLEGO-iG2 (Addgene, 27341) was first engineered to replace GFP with tagBFP (pLEGO-iB2). Briefly, tagBFP was PCR-amplified with 5′ and 3′ overhangs from dCas9–KRAB (Addgene, 102244) and inserted into pLEGO-iG2 digested with MscI (New England Biolabs) and BsrGI (New England Biolabs) by Gibson assembly using HiFi assembly mix (New England Biolabs) according to the manufacturer’s instructions. Subsequently, the coding sequences of PRDX1–PRDX6, CAT plus an N-terminal FLAG tag and SOD1 plus a C-terminal FLAG tag, which were synthesized as gene fragments (Twist Bioscience) with 5′ and 3′ overhangs, were inserted into the pLEGO-iB2 backbone digested by EcoRI-HF (New England Biolabs) and NotI-HF (New England Biolabs) using Gibson assembly with HiFi assembly mix (New England Biolabs). All PRDX family relocalization cDNAs were synthesized as gene fragments (Integrated DNA Technologies) with 5′ and 3′ overhangs and inserted into EcoRI-HF (New England Biolabs) and NotI-HF (New England Biolabs) digested pLEGO-iB2 backbone by Gibson assembly using HiFi assembly mix (New England Biolabs). Cytosolic variants of PRDX3, PRDX4 and PRDX5 comprised amino acids 62–256, 38–271 and 54–214 of the full-length variants, respectively.
Cysteine mutant PRDX1 constructs were generated by site-directed mutagenesis of the wild-type open reading frame in PRDX1-IRES-BFP pLEGO. Primers were designed to introduce the relevant nucleotide alterations by PCR before DpnI (New England Biolabs) digestion of template DNA and transformation of amplified modified plasmid. Consecutive rounds of mutagenesis were used to generate the PRDX1-C52S;C173S double mutant.
The mito–DAO construct was generated by PCR of mito–DAO–7×His from pC1-CMV-mito-DAO (Addgene 141132) and inserted into the pLEGO-iB2 backbone digested by EcoRI-HF (New England Biolabs) and NotI-HF (New England Biolabs) using Gibson assembly with HiFi assembly mix (New England Biolabs).
Generation of PRDX1-KO cells
Parental A549 cells were nucleofected using a Lonza 4D-Nucleofector system and recombinant Cas9 ribonucleoprotein complexed with sgRNA targeting exon 2 of PRDX1 according to previously described approaches56 (Supplementary Dataset 5). Nucleofected cells were subsequently sorted to single cells by fluorescence-activated cell sorting (FACS) and screened for protein ablation using Sanger sequencing and immunoblot.
Mixed-population growth competition assays
A549 were transduced with sgRNAs to repress individual genes or cDNAs to overexpress relevant proteins in combination with a fluorescent protein (BFP or GFP). Cell populations expressing a mixture of both sgRNA-positive and sgRNA-negative or cDNA-positive and cDNA-negative cells were seeded at 25,000 cells per well in a 24-well format and exposed to the relevant treatment conditions. sgRNA-positive or cDNA-positive fractions were tracked over time by measuring the coexpressed fluorescent marker every 3–4 days on an Attune NxT flow cytometer. Data are plotted as RI scores calculated using the formula RI = (G2 − (G1 × G2))/(G1 − (G1 × G2)), where G1 is the fraction of sgRNA-positive or cDNA-positive cells in the control population and G2 is the fraction of sgRNA-positive or cDNA-positive cells in the perturbed population. As a result, vehicle control conditions have an RI score of 1 and deviations above or below 1 indicate resistance or sensitivity to the perturbation, respectively.
Clonogenic assays
Six-well plates were seeded with 112.5 cells per well and treated with DDRi at previously described concentrations. The medium was replenished every 4 days with fresh dilutions of DDRi. After 2 weeks, plates were gently washed with PBS, stained with 0.5% crystal violet in 6% glutaraldehyde for 30 min at room temperature and then washed with water. Plates were scanned using a flatbed scanner and colonies were scored using ImageJ.
Western blotting
Cells were lysed in RIPA buffer (0.5 M Tris-HCl pH 7.4, 1.5 M NaCl, 2.5% deoxycholic acid, 10% NP-40 and 10 mM EDTA) supplemented with protease and phosphatase inhibitor cocktail (Thermo Scientific, 1861281). Cells were lysed on ice for 10 min and spun at 21,000g for 10 min to pellet insoluble material. Next, 4× LDS sample buffer (Invitrogen) was added to lysates and samples boiled at 98 °C for 5 min. Before gel loading, 10× Novex sample reducing agent (Invitrogen) was added to each sample. Samples were run on NuPAGE Bis–Tris 4–12% gels (Invitrogen) in MES or MOPS buffer (Invitrogen) depending on protein size. Samples were transferred to 0.45-µm LF-PVDF membranes (Bio-Rad) using a TransBlot Turbo semidry transfer system (Bio-Rad) at 2.5 A and 25 V for 7 min. After transfer, membranes were blocked with 5% (w/v) milk in Tris-buffered saline containing 0.1% Tween-20 for 30 min at room temperature. Primary antibodies incubations were performed for 1–2 h at room temperature or overnight at 4 °C using antibodies to PRDX1 (Abcam, ab41906 or ab70666, clone 3G5, 1:1,000), PRDX2 (Abcam, ab109367, clone EPR5154, 1:1,000), PRDX3 (Abcam, ab128953, clone EPR8115, 1:1,000), PRDX4 (Abcam, ab184167, clone EPR15458(B), 1:1,000), PRDX5 (Abcam, ab180587, clone EPR14529(B), 1:1,000), PRDX6 (Proteintech, 13585-1-AP, 1:1,000), FLAG (Proteintech, 20543-1-AP, 1:1,000), GFP (Cell Signaling, 2955T, clone 4B10, 1:1,000) and tubulin (Sino Biological, 100109-MM05T, clone 05, 1:10,000). After washing, blots were incubated with anti-mouse (Cell Signaling Technology, 5470P, 1:4,000) and/or anti-rabbit 800 (Cell Signaling Technology, 5151P, 1:4,000) for 1 h at room temperature. Blots were imaged using a LI-COR Odyssey CLx imaging system.
Immunofluorescence
Cells were seeded on poly(L-lysine)-coated coverslips, allowed to adhere and exposed to the relevant treatment conditions. Cells were then fixed in 4% (w/v) paraformaldehyde for 30 min, permeabilized with 0.1% Triton X-100 in PBS for 10 min and blocked with 1% BSA plus 0.1% Triton X-100 in PBS (blocking buffer) for 1 h. Coverslips were incubated with γH2AX (Millipore, 05-636, clone JBW301, 1:500), PRDX1 (Abcam, ab70666, clone 3G5, 1:100), PRDX3 (Abcam, ab128953, clone EPR8115, 1:100), PRDX4 (Abcam, ab184167, clone EPR15458(B), 1:100), PRDX5 (Abcam, ab180587, clone EPR14529(B), 1:100) or His (Abcam, ab18184, clone HIS.H8, 1:100) antibodies in blocking buffer for 2 h at room temperature or overnight at 4 °C. After washing, coverslips were incubated with Alexa Fluor 488 secondary antibody (Invitrogen, A21206, 1:500) in blocking buffer for 1 h at room temperature. Coverslips were washed and mounted onto glass slides using ProLong Gold Antifade reagent with DAPI (Invitrogen, P36935) and dried. Slides were imaged at the Microscopy and Advanced Bioimaging CoRE at the Icahn School of Medicine at Mount Sinai using an inverted Zeiss LSM780 confocal microscope using a ×40 oil-immersion Zeiss Plan-Neofluar ×40 (NA: 0.9) objective and diode (405 nm), argon (488 nm) and DPSS (561 nm) lasers.
γH2AX flow cytometry
After exposure to the relevant treatment conditions, A549 cells were dissociated, washed and fixed with 2% paraformaldehyde for 20 min, followed by permeabilization with 90% ice-cold methanol for 10 min at 4 °C. Cells were washed in PBS supplemented with 10% FBS (FACS buffer) before resuspending in FACS buffer containing γH2AX antibody (Millipore, 05-636 clone, JBW301) at a final concentration of 0.5 µg ml−1 for 1 h at room temperature. Cells were washed with FACS buffer and labeled with α-mouse Alexa Fluor 647 (Life Technologies) at a final concentration 2 µg ml−1 in FACS buffer for 30 min at room temperature with protection from light. After washing, cells were analyzed on an Attune NxT Flow Cytometer and quantified using FCS Express 7.
RNA-seq library preparation
Total RNA was extracted using the Qiagen RNeasy Plus 96 Kit on the Integra Viaflo 96/384 liquid-handling platform to enable high-throughput sample processing. Briefly, cells or tissues were lysed in RLT Plus buffer with genomic DNA removal using gDNA Eliminator plates. RNA was captured on silica membranes, washed and eluted in RNase-free water. The extracted RNA was quantified using a fluorescence-based RNA assay and Agilent Fragment Analyzer to ensure sufficient yield and quality for downstream applications.
RNA-seq libraries were prepared using a modular automated workflow on three liquid-handling platforms:
-
a.
Hamilton Prep for RNA input normalization and reagent aliquoting.
-
b.
Integra Viaflo for poly(A) RNA enrichment followed by Hamilton prep for cDNA synthesis setup.
-
c.
Dynamic Devices Lynx for bead cleanups and library pooling.
The libraries were constructed using the KAPA mRNA HyperPrep Kit (Roche), which includes mRNA enrichment through oligo(dT) capture, fragmentation, reverse transcription, second-strand synthesis, end repair, A-tailing, adaptor ligation and amplification. The workflow was optimized to minimize hands-on time and ensure consistency across plates. All libraries were quantified and checked using a Fragment Analyzer and fluorescence-based assay.
Prepared libraries were pooled and sequenced on the Illumina NovaSeq X Plus platform using a paired-end configuration of 2× 150-bp reads. Each sample was sequenced to a target depth of 30 million reads. Quality control metrics for the sequencing run, including Q30 scores and adaptor content, were monitored using MultiQC reports to ensure high-quality data output.
RNA-seq data processing and differential expression analysis
RNA-seq reads were processed using the nf-core/rnaseq pipeline57 (version 3.18.0) with default parameters. Briefly, raw reads were quality-checked and trimmed before alignment to the Homo sapiens (hg38) reference genome. Gene-level quantification was performed and count data were generated as part of the pipeline’s default outputs. Subsequent data exploration and differential expression analyses were conducted in R (version 4.4.2) using the limma/voom and edgeR packages. Normalization was carried out to account for library size and compositional biases. Differentially expressed genes were identified with limma-voom using moderated two-sided t-tests and empirical Bayes shrinkage (eBayes); P values were adjusted for multiple testing using the Benjamini–Hochberg procedure and genes with adjusted P < 0.05 were considered significant. All plots were generated using ggplot2 and gene annotations were retrieved from the appropriate annotation database (for example, org.Hs.eg.db). All codes and results related to the RNA-seq data are publicly available from Zenodo (https://doi.org/10.5281/zenodo.21285432)58.
FerroOrange
A total of 50,000 cells were seeded onto black 96-well plates overnight. The following day, cells were pretreated with vehicle or 10 mM FAC for 2 h. Cells were washed three times with HBSS and then incubated with FerroOrange probe (Dojindo, F374) diluted in HBSS, with or without 100 µM DFO, for 30 min at 37 °C. FerroOrange fluorescence was measured on a Cytation5 plate reader using a monochromator at an excitation wavelength of 543 nm and emission wavelength of 580 nm. For data normalization, equivalent cell numbers were contemporaneously seeded onto white 96-well plates (Corning, 3917) and used for CellTiter Glo. CellTiter Glo (Promega, G7572) diluted 1:5 in PBS was added in a 1:1 ratio directly to the culture medium and incubated for 30 min before measuring luciferase luminescence on a Cytation5 plate reader. Background from an empty well containing FerroOrange solution was subtracted from experimental wells and data were then normalized to ATP measurements derived from CellTiter Glo.
DCFH-DA
A total of 20,000 cells were seeded onto black 96-well plates overnight. The following day, cells were washed three times with HBSS, incubated with Hoechst33342 (Invitrogen, H3570, 1:1,000) in HBSS for 5 min, then incubated with high sensitivity DCFH-DA probe (Dojindo, R252), diluted 1,000× in manufacturer supplied loading buffer and combined with 10 mM L-alanine or D-alanine as indicated, for 30 min at 37 °C. DCFH-DA fluorescence was measured on a Cytation5 plate reader using a monochromator at an excitation wavelength of 495 nm and emission wavelength of 525 nm (DCFH-DA) or excitation wavelength of 350 nm and emission wavelength of 460 nm (Hoechst). Background from an empty well containing DCFH-DA solution was subtracted from experimental wells and data were then normalized to DNA/Hoechst fluorescence.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
