CRISPR-based chemogenomic profiling reveals redox vulnerabilities to epigallocatechin-3-gallate and green tea polyphenol extract

Abstract

Green tea polyphenols, particularly epigallocatechin-3-gallate (EGCG), are widely recognized for their beneficial preventive effects against chronic diseases including cancer and obesity. These effects are traditionally attributed to EGCG's antioxidant, anti-inflammatory, and metabolic regulatory properties. In conditions characterized by persistent oxidative stress, the disrupted redox signaling further creates a unique vulnerability that EGCG may exploit through a dual redox mechanism. Emerging evidence therefore suggests that EGCG not only mitigates oxidative damage but could also induce selective pro-oxidant stress in cancer cells, enhancing its therapeutic potential. To investigate this duality, we performed a genome-wide CRISPR/Cas9 knockout screen to identify genetic determinants of EGCG sensitivity and resistance. Our chemogenomic analysis revealed that loss of key antioxidant genes, including PRDX1, CAT, GSS, GCLM, and GCLC, significantly heightened cellular susceptibility to EGCG and green tea extract (GTE), underscoring the critical role of glutathione biosynthesis and redox homeostasis in mediating cytotoxicity. In contrast, knockouts of Kelch-like ECH-associated Protein 1 (KEAP1) and peroxisome-associated PEX genes conferred resistance, implicating in part NRF2 (also known as nuclear factor erythroid-derived 2-like 2; NFE2L2) activation and peroxisomal reactive oxygen species clearance in protective responses. Comparative profiling with gallic acid (GA), which lacks EGCG's catechin structure, further highlighted the gallate moiety's contribution to glutathione-dependent antioxidant mechanisms. Altogether, these findings illuminate the complex redox biology of EGCG and identify novel genetic vulnerabilities that may be leveraged to enhance its anticancer efficacy, particularly in obesity-associated cancers. Clinically, this work could support the development of EGCG-based interventions tailored to individual redox profiles, offering a precise chemopreventive strategy for patients at high risk of malignancies driven by metabolic and oxidative dysregulation. Furthermore, the identification of new genetic markers of EGCG sensitivity and resistance may inform future exploration of patient stratification.

Graphical abstract

Highlights

• •EGCG shows dual redox action: antioxidant and selective pro-oxidant effects.
• •CRISPR screen reveals genes driving EGCG sensitivity and resistance.
• •Loss of glutathione pathway genes heightens EGCG-induced cytotoxicity.
• •KEAP1 and PEX knockouts confer resistance via NRF2 and ROS clearance.
• •Gallate moiety critical for EGCG's glutathione-dependent antioxidant role.

1. Introduction

Green tea has long been recognized for its health-promoting properties, often attributed to its high content of antioxidant polyphenols, most notably epigallocatechin-3-gallate (EGCG) and epicatechin gallate [1,2]. These catechins are traditionally viewed as antioxidants, believed to neutralize harmful oxygen-derived free radicals and thereby protect cellular structures and DNA from oxidative damage [3]. However, emerging research challenges this conventional view, revealing a complex and paradoxical oxidative profile for EGCG. Accordingly, a recent study demonstrated that EGCG and related catechins do not merely suppress oxidative stress but, in contrast, can transiently promote it in the C. elegans model organism [4]. This initial short-term increase in oxidative stress was shown to activate endogenous defense mechanisms, including the upregulation of genes encoding antioxidant enzymes such as superoxide dismutase and catalase (CAT) resulting in enhanced cellular resilience, improved fitness, and extended lifespan [5].

This phenomenon suggests that EGCG may act as a pro-oxidant trigger, promoting in turn adaptive stress responses akin to a biological “vaccination”. Rather than passively scavenging free radicals, EGCG appears to actively engage oxidative signaling pathways, potentially offering greater long-term protection through hormesis, a process where low-level stress induces beneficial adaptive responses [4]. Such findings underscore the need to further investigate the duality of EGCG, especially where oxidative stress plays a pivotal role such as in the context of cancer and obesity [6,7]. In cancer cells, EGCG's pro-oxidant activity may contribute to selective cytotoxicity [7], while in obesity, it may modulate metabolic pathways and mitochondrial function [8]. Yet, the precise balance between its antioxidant and pro-oxidant effects, and how this balance shifts across different tissues and disease states, remains poorly understood.

To fully harness EGCG's chemopreventive potential, research must delve deeper into its context-dependent oxidative behavior, exploring the molecular switches that determine EGCG's antioxidant versus pro-oxidant activity. Furthermore, tissue-specific responses and implications for cancer cell apoptosis versus normal cell protection, the role of metabolic state and mitochondrial activity in modulating EGCG's effects, and the longitudinal impacts of EGCG-induced oxidative stress on gene expression and cellular aging should be explored [1]. Understanding this duality is not just a matter of refining our view of green tea's health benefits, it is essential for eventually developing targeted interventions that leverage EGCG's oxidative dynamics to prevent and treat chronic diseases like cancer and obesity.

Despite extensive research, the specific genetic determinants of cellular sensitivity and resistance to tea catechins, and impact of their gallic acid moiety in particular, remain inadequately characterized. To address this, we performed a genome-wide CRISPR-Cas9 knockout (KO) screen in human NALM-6 leukemia cells, aiming to identify key genes modulating the cellular response to these compounds [9]. Chemogenomics is intended to systematically screen bioactive compounds toward all gene functions to inform on their genetic fingerprint that either suppress (rescue) or enhance (synthetic lethality) compound effects on cell proliferation. This information can confirm compound mechanism of action in an unbiased way and potentially identify the genes that are involved in the cellular process affected by the tested compound. Here, this approach allowed us to uncover critical insights into how green tea extract (GTE), EGCG and gallic acid (GA) differentially affect redox-sensitive pathways, including glutathione metabolism and peroxisomal function, ultimately contributing to their health beneficial properties.

2. Materials and methods

2.1. Chemicals and reagents

Epigallocatechin-3-gallate (EGCG) was obtained from MP Biomedicals (#199165, OH, USA), green tea extract (GTE) was from LKT Laboratories (G6817, St Paul, MN, USA), gallic acid (GA) was from Sigma (Oakville, ON, Canada), dodecyl gallate (DG) and Galunisertib (LY2157299) were from MedChemExpress (Monmouth Junction, NJ, USA). All other chemicals and reagents were of analytical grade and obtained from standard commercial sources.

2.2. CRISPR-based chemogenomic screens

The Genome-wide pooled CRISPR-Cas9 KO screens were performed at the ChemoGenix platform (IRIC, Université de Montréal; https://chemogenix.iric.ca/) as previously described [9]. Briefly, a human NALM-6 (pre-B ALL lymphocytes) clone bearing an integrated inducible Cas9 expression cassette generated by lentiviruses made from pCW-Cas9 (Addgene #50661) was transduced with the genome-wide KO EKO sgRNA library (278,754 different sgRNAs). After thawing the library from liquid nitrogen and allowing cells to recover in 10 % FBS RPMI for 1 day, KOs were induced for 7 days of culture with 2 μg/mL doxycycline. The pooled library was then split into different T-75 flasks (28 × 10⁶ cells per flask; a representation of 100 cells/sgRNA) in 70 mL at 4 × 10⁵ cells/mL. Cells were treated with 100 μM EGCG, 80 μM GA, 0.7 μM DG, 170 μM Galunisertib, or 35 μg/mL GTE for 8 days. Cell growth was monitored every 2 days, diluting back to 4 × 10⁵ cells/mL and adding more compound to maintain the same final concentration whenever cells reached 8 × 10⁵ cells/mL. Over that period, treated cells had close population doubling values (GTE 3.6; EGCG 3.3, GA 2.1, DG 3.9, Galu 4.0) ensuring that selective pressure and overall cell proliferation were comparable across conditions for genetic drug screening comparison. Whereas DMSO-only treated negative controls had 7.65 cell doubling. Cells were collected, genomic DNA extracted using the Gentra Puregene kit according to manufacturer's instructions (QIAGEN), and sgRNA sequences PCR-amplified as described [9]. sgRNA frequencies were obtained by next-generation sequencing (Illumina NextSeq 500). Reads were aligned using Bowtie 2.4.4 in the forward direction only (–norc option) with otherwise default parameters and total read counts per sgRNA were tabulated. Context-dependent chemogenomic interaction scores were calculated using a Context-dependent Robust Analytics and Normalization for Knockout Screens (CRANKS), a modified version of the RANKS algorithm that uses guides targeting similarly essential genes as controls to distinguish condition-specific chemogenomic interactions from non-specific fitness/essentiality phenotypes [9].

2.3. Bioinformatics and network analysis

Gene Ontology (GO) [10] and KEGG pathway [11] analyses was conducted to identify enriched biological pathways and processes. Lists of significant genes were input into the tools, and enrichment scores and p-values were calculated with false discovery rate (FDR) correction. Networks were visualized using STRING or Cytoscape software (version 3.10.3). Gene set enrichment analysis (GSEA) was performed to evaluate the predefined GO/KEGG glutathione metabolism and peroxisome pathways-level fingerprinting across all five treatment conditions [12]. Normalized enrichment score (NES) and the nominal p-values were reported, as multiple hypothesis testing correction was not applied for this targeted follow-up. Network topology was analyzed to identify hub genes and clusters, and functional modules within the network. Pathways and GO terms with FDR-adjusted p-values <0.05 were considered significantly enriched. Protein-protein interaction (PPI) data were obtained from the STRING database (p < 0.05) using whole genome as a background (https://www.string-db.org/) [13]. Clustering via k‐means, with negative and positive CRANKS scores reflected sensitizer and rescuer gene KO, respectively.

2.4. Statistical analysis

All statistical analyses were conducted in GraphPad Prism at an α = 0.05 significance threshold, except for CRISPR screen computations performed using the CRANKS pipeline (CRANKS ≤ −2 or ≥ +2 at FDR< 0.05 defined significant top hits). Data normality was assessed via the Kolmogorov-Smirnov test before applying either Wilcoxon rank tests (nonparametric) or one‐/two‐way ANOVA with Bonferroni's or Newman-Keuls post hoc comparisons. GO and KEGG pathway enrichment analyses were performed using SubCellulaRVis or STRING with an FDR cut-off of 0.05. Heat map pairwise comparisons were generated using the online analysis tool Heatmapper [14]. Further details on sample sizes, data transformations, and exact p‐values are available in the figure captions.

3. Results

3.1. Identification of redox and peroxisomal genes critical for green tea extract cytotoxicity using a genome-wide CRISPR-Cas9 screen

To investigate the molecular basis of EGCG's anti-proliferative effects, we performed a genome-wide chemogenomic screen grounded on the hypothesis that green tea polyphenols, particularly EGCG, modulate cellular redox balance and stress response pathways. While EGCG is traditionally viewed as an antioxidant, accumulating evidence suggests it can also act as a pro-oxidant under specific conditions, especially in cancer cells. This duality led us to hypothesize that EGCG may exploit intrinsic vulnerabilities in redox regulation and metabolic stress, inducing oxidative damage that contributes to its therapeutic potential. To capture a broader spectrum of polyphenolic activity, we included green tea polyphenols extract (GTE) as a reference compound. By systematically disrupting gene function across the genome, our CRISPR-Cas9 KO screen in NALM-6 pre-B leukemia cells using a doxycycline-inducible Cas9 system (Fig. 1A), aimed to identify key genetic determinants that sensitize or protect cells from EGCG/GTE-induced oxidative stress, thereby illuminating the cellular pathways that mediate the cytotoxic effects.

Fig. 1.

Genome-wide CRISPR-Cas9 screen identifies genetic determinants of GTE-induced cytotoxicity. (A) Schematic overview of the experimental workflow for the genome-wide CRISPR knockout screen in NALM-6 cells treated with green tea extract (GTE), aimed at identifying genes modulating compound sensitivity. (B) Distribution of CRANKS scores for 19,034 genes following GTE treatment. Negative scores (red arrows, CRANKS ≤ −2) indicate sensitizer genes whose knockout enhances compound sensitivity, while positive scores (green arrows, CRANKS ≥2) represent rescuers whose knockout confers resistance. The 2D scatter plot is aligned with a one-dimensional frequency distribution, illustrating the thresholds used to define top hits. (C) Volcano plot showing CRANKS scores plotted against –log₁₀p-values. Genes above the horizontal dashed line marks the p-threshold (p < 0.05), while those beyond the vertical dashed lines exceed CRANKS thresholds of ±2, identifying 30 filtered genes (p ≤ 0.01) labeled by gene symbol. Red points represent sensitizers; green points represent rescuers. Listed starred gene (∗) indicate top hits. (D) KEGG pathway enrichment analysis of highest CRANKS-filtered candidates (|CRANKS| ≥ 2; p ≤ 0.01; set including all top hits). Bubble color reflects false discovery rate (FDR), ranging from light green (FDR = 9 × 10⁻⁷) to dark blue (FDR = 4 × 10⁻³), and bubble size corresponds to the number of genes in each pathway. Pathways are ranked by significance on the y-axis. Analysis was performed using the STRING database with whole-genome background, requiring a minimum of two genes per pathway and FDR ≤0.05.

Guided by the hypothesis that EGCG could exert anti-proliferative effects through oxidative stress and redox imbalance, cells were treated with an intermediate concentration of GTE that reduced proliferation by ∼50 % over eight days, allowing us to capture both sensitizing and resistance-conferring genetic effects. Next-generation sequencing of sgRNAs counts between treated vs untreated pooled KOs enabled calculation of CRANKS scores, an algorithm scoring changes of sgRNA frequencies to predict each gene KO's effect upon increased resistance to compound-induced growth inhibition (positive CRANKS score, rescues, resisters) or upon increased sensitivity to compound-induced growth inhibition (negative CRANKS score, synthetic lethal, sensitizers). Genes with CRANKS ≤ −2 were classified as sensitizers (depleted under treatment), while those with CRANKS ≥2 were designated as rescuers (enriched under treatment) (Fig. 1B). The ranked distribution of CRANKS values showed that most genes clustered around zero with a frequency distribution median = 0.012, while a small subset deviated clearly in the positive or negative ranges. Statistical analysis revealed a list of the 30 highest CRANKS scoring gene candidates (|CRANKS| ≥ 2; p ≤ 0.01), with corresponding FDR within CRISPR screen score ranging from 0.0063 to 1.0000 applied on a gene‐by‐gene basis. Values and statistical outputs for each gene are provided in Supplemental Dataset S1. Of these, 18 sensitizers and 12 rescuers were identified, many of which clustered in redox and peroxisome-related pathways. Notable significant top hits (FDRs <0.05) sensitizers included GCLC, GCLM, CAT, PRDX1, GSS, OSGEP, and APEX2, while KEAP1 and PEX1 emerged as key rescuers (Fig. 1C).

Pathway enrichment analysis was next performed on the 30 highest candidates comprising top hits, to determine whether specific biological processes were overrepresented among sensitizer and resistance genes. This analysis revealed four significantly enriched pathways, with the peroxisome pathway showing the highest significance and gene count (PRDX1, CAT, PEX1, PEX6, PEX12, PEX14), followed by glutathione metabolism (GCLM, GCLC, GSS, GSR), and two additional pathways, namely ferroptosis and cysteine/methionine metabolism, sharing overlapping genes (GCLM, GCLC, GSS) (Fig. 1D). These findings establish a genetic fingerprint of GTE's cytotoxic mechanism, highlighting redox regulation and peroxisomal function as central vulnerabilities. This foundational screen now provides a framework for comparing GTE's effects to those of isolated compounds such as EGCG and validates the use of chemogenomics to dissect polyphenol-induced stress responses in cancer cells. To gain mechanistic insight into the cytotoxic effects of GTE, we further conducted a genome-wide CRISPR-Cas9 KO screen, now aiming to identify genetic determinants that modulate cellular sensitivity to specific polyphenol-induced oxidative stress.

3.2. Chemogenomic dissection of EGCG highlights overlapping redox and organelle stress networks with green tea extract

To determine the specific contribution of EGCG to the anti-proliferative effects of GTE, we leveraged a genome-wide CRISPR-Cas9 chemogenomic screen, comparing gene KO sensitivity profiles under EGCG and GTE treatments. Given EGCG's dual antioxidant/pro-oxidant property, this approach now allows us to systematically identify genes whose loss either sensitized cells to oxidative stress or conferred resistance. A scatter plot of CRANKS scores revealed significant overlap in genetic determinants between EGCG and GTE (r ∼0.28, p < 0.0001), with 22 genes showing highly correlated effects (r ∼0.95, p < 0.0001) (Fig. 2A). At this stage less-correlating specific differences also emerged between treatments, such as MCM9, TRAPPC3, IGSF21 and ABCC1. Focusing on the previous 30 GTE list, we identified a core set of 17 genes which included all previously identified key top hits also consistently modulated by EGCG such as CAT, GSS, GCLM, GCLC, PRDX1, OSGEP and APEX2 along other sensitizers such as TRAF2, SEPHS2, and LRP8, whose KO enhanced sensitivity to EGCG-induced oxidative stress (Fig. 2B). Conversely, resistance genes such as PEX6, PEX12, PEX14, including top hits PEX1, and KEAP1, central to peroxisomal function and NRF2-mediated antioxidant responses, indicating their loss promotes survival despite EGCG exposure.

Fig. 2.

Comparative chemogenomic profiling reveals shared and distinct genetic dependencies between EGCG and GTE treatments. (A) Two-dimensional scatter plot showing CRANKS scores for 19,034 genes under EGCG (y-axis) versus GTE (x-axis) treatment. Each point represents a gene, positioned according to its sensitivity or resistance profile under both compounds. The overall correlation between treatments was quantified using Pearson's correlation (r = 0.28, p < 0.0001; 95 % CI: 0.27–0.29). A subset of 22 highly correlated genes is color-coded circles to indicate sensitizers (red) and rescuers (green), with a strong pairwise correlation (r = 0.95, p < 0.0001; 95 % CI: 0.88–0.98). White squares highlight notable non-correlated genes to either treatment (B) Venn diagram comparing the highest 30 CRANKS genes from GTE with the uppermost 22 correlated EGCG genes, identifying 17 overlapping genes (GTE ∩ EGCG) and highlighting unique contributions from each compound (13 GTE-specific vs. 5 EGCG-specific). Top hits are starred bolded fonts. (C) UpSet plot generated using SubCellulaRVis, showing enriched subcellular compartments among the 17 overlapping CRISPR hits. Horizontal bars indicate total gene counts per compartment, while vertical bars represent intersection sizes across compartments. Gene annotations were mapped to Gene Ontology Cellular Components and the Human Protein Atlas, with enrichment assessed at FDR ≤0.05. (D) Protein-protein interaction network of the 17 overlapping hits constructed using the STRING database. Nodes represent genes/proteins, color-coded as red for synthetic lethals/sensitizers and green for rescues/suppressors. Edges represent known interactions between proteins denoting known or predicted functional associations. Clusters were identified using k-means analysis and grouped by biological function: glutathione biosynthesis and recycling (GCLC, GCLM, GSS, PRDX1, KEAP1; solid circle), peroxisomal matrix import (PEX1, PEX6, PEX12, PEX14; dotted circle), and central hub catalase (CAT). Node color indicates functional classification: red = sensitizers (synthetic lethals), green = rescuers (resistance-conferring knockouts).

Subcellular localization analysis using SubCellulaRVis showed these genes clustered in compartments tied to redox regulation, particularly the cytoplasm and peroxisomes (Fig. 2C). The presence of PRDX1 and CAT among the sensitizers indicated that the oxidative stress detoxification pathways were particularly vital in counteracting the pro-oxidant effects of EGCG. PPI mapping using STRING revealed two major clusters: one involving glutathione biosynthesis (GCLC, GCLM, GSS, PRDX1, KEAP1) and another centered on peroxisomal import machinery (PEX1, 6, 12, and 14), both converging towards CAT as a hub (Fig. 2D). Additional synthetic lethal genes formed smaller clusters (not shown), including SEPHS2, OSGEP, LRP8, XPO4, and BRE.

3.3. EGCG uncovers unique genetic vulnerabilities beyond green tea extract in redox-driven cancer cytotoxicity

While EGCG largely recapitulated the redox- and peroxisome-associated genetic dependencies observed under GTE treatment, several genes however exhibited sensitization under EGCG such as BAK1 compared to GTE. The observation of ABCC1 specific sensitization to EGCG led us to investigate further with differential CRANKS scores exceeding −2.5 for potential additional mechanisms beyond those captured by GTE (Table 1). Among these, other EGCG-specific sensitizers included CIDEB, NUP88, PHKA2, and ARPC1A. However, only ABCC1 (encoding multidrug resistance protein-1, MRP1) showed the most pronounced significant shift (GTE: +0.11; FDR = 1.0000; EGCG: −3.11; FDR = 0.0048), suggesting that loss of this drug transporter significantly enhances EGCG-induced cytotoxicity. Having established EGCG/GTE concordance, we next isolated galloyl-driven versus catechin-dependent effects by comparing EGCG with GA before pathways scale analysis. EGCG/GA dot plot comparisons revealed shared and distinct sensitizer/rescuer response (Supplementary Fig. S1A). Overlapping genes included PRDX1, GCLC, GCLM and TRAF2 as sensitizers and rescuers such as KEAP1 and PEX1, but clear compound-specific rescuing shifts were observed (Supplementary Fig. S1B), and sensitizer profile displaying preferential genes under EGCG including UBR1 and ABCC1 or under GA such as CYB5A and RB1CC1 (Supplementary Fig. S2).

Table 1.

Differential sensitivity CRANKS scores.

Gene	Full name	Role in cancer	GTE	EGCG	Score diff.
ABCC1	ATP Binding Cassette Subfamily C Member 1	Associated with multidrug resistance in cancer cells; contributes to chemotherapy failure.	0.11	−3.11	3.22
ABCG5	ATP Binding Cassette Subfamily G Member 5	Altered expression may affect cholesterol metabolism in cancer cells; role in tumor microenvironment under investigation.	1.34	−1.23	2.57
ARPC1A	Actin Related Protein 2/3 Complex Subunit 1A	Regulates actin cytoskeleton; contributes to cancer cell migration and invasion.	1.48	−1.04	2.51
C7orf50	Chromosome 7 Open Reading Frame 50	Uncharacterized; may be involved in RNA processing and cancer-related gene regulation.	1.46	−1.12	2.58
CIDEB	Cell Death Inducing DFFA Like Effector B	Downregulated in hepatocellular carcinoma; may influence apoptosis and lipid metabolism in cancer.	1.33	−1.25	2.58
DNAJB14	DnaJ Heat Shock Protein Family (Hsp40) Member B14	Chaperone protein; may influence protein folding and stress response in cancer cells.	1.47	−1.05	2.53
FAM194A	Family With Sequence Similarity 194 Member A	Limited data; potential epigenetic regulator with altered expression in cancer tissues.	1.28	−1.34	2.62
NUP88	Nucleoporin 88	Overexpressed in various cancers; promotes cell proliferation and metastasis via NF-κB signaling.	0.44	−2.21	2.65
PHKA2	Phosphorylase Kinase Regulatory Subunit Alpha 2	Mutations linked to metabolic disorders; potential role in cancer metabolism remains to be clarified.	1.16	−1.36	2.52
PSG9	Pregnancy Specific Beta-1-Glycoprotein 9	Promotes immune tolerance and angiogenesis; implicated in tumor progression and metastasis.	0.95	−1.61	2.56
PSMC2	Proteasome 26S Subunit, ATPase 2	Involved in proteasome-mediated degradation; dysregulation linked to tumor progression and immune evasion.	1.69	−1.03	2.72

3.4. Structural variants of gallate moieties converge on redox and peroxisomal stress pathways

Based on the observed genetic dependencies shared between EGCG and GTE, we next hypothesized that the anti-proliferative activity of EGCG is strongly linked to its galloyl moiety, which may overwhelm cellular antioxidant defenses and promote reactive oxygen species (ROS) generation. To test whether this effect was unique to EGCG or shared among structurally related compounds, we extended our chemogenomic analysis to include additional galloyl ester molecules. This comparative approach aimed to determine whether the redox vulnerabilities identified under EGCG treatment, particularly those involving glutathione metabolism, peroxisomal function, and stress-response signaling, involved galloyl-driven oxidative stress by including additional compounds: gallic acid (GA), dodecyl gallate (DG), and Galunisertib, a non-gallate transforming growth factor-beta receptor I (TβRI) inhibitor used as a mechanistic control (Fig. 3A). Two-dimensional scatter plots of genome-wide CRISPR scores revealed distinct distribution profiles across treatments, with DG showing the most divergent pattern compared to the others (Supplementary Fig. S3A). Summed CRANKS scores indicated a net bias toward resistance for GTE, EGCG, GA, and Galunisertib (172.4, 149.9, 172.5, and 213.1 respectively), while DG displayed the lowest total score (123.9) and a near-zero median, with uncommon top hits displaying a pronounced rescuing phenotype (Supplementary Fig. S3B and Supplementary Table S1). To assess to what extent the galloyl moiety solely contributes to EGCG's cytotoxicity, we focused on two KEGG-curated gene sets, glutathione metabolism (50 genes) and peroxisomal function (78 genes), previously identified as enriched in the GTE/EGCG screens. GSH metabolism targeted GSEA revealed similar negative enrichment scores between GTE and galloyl compounds (Fig. 3B), with shared directional sensitivity enriched core involving leading edge genes top hits GCLs and GSS. The NES results show significant negative amplitudes of GTE (−1.60, p = 0.0077) and GA (−1.42, p = 0.040), followed with EGCG (−1.31) and moderately DG (−1.1), whereas Galunisertib showed a divergent positive score. These results indicate that the higher GTE scores point to an additive sensitizing effect of galloyl compounds identifying glutathione metabolism as a key vulnerability. Correlation heatmaps revealed that EGCG and GA closely mirrored GTE's glutathione-related KO profiles, with DG also showing positive correlation, while Galunisertib exhibited negligible or inverse correlation, confirming its distinct mechanism of action (Fig. 3C). Linear regression analyses further supported these relationships for glutathione metabolism, with EGCG (r = 0.56, p < 0.0001), GA (r = 0.62, p < 0.0001), and DG (r = 0.60, p < 0.0001) moderately correlating with GTE, whereas Galunisertib diverged (r = −0.36, p = 0.0049) (Fig. 3D–G). Gene-level trends within the glutathione pathway highlighted EGCG-specific sensitization, particularly for GCLC and GCLM, which showed progressively decreasing CRANKS scores across GTE, EGCG, GA, DG, and Galunisertib (e.g., GCLC: −3.19, −3.40, −1.48, 1.14, 0.71; GCLM: −2.89, −3.20, −2.92, −2.32, −0.16). These findings suggest that galloyl esters may act synergistically within GTE to induce oxidative stress, and that their cytotoxicity is modulated by the glutathione metabolic status of the cell.

Fig. 3.

Pathway-level comparison of CRISPR screening results in glutathione- and peroxisome-related genes. (A) Chemical structures of the four tested compounds: EGCG, gallic acid (GA), dodecyl gallate (DG), and Galunisertib (Galu; a TGF-β pathway inhibitor which serves as a structurally distinct control lacking any galloyl ester). (B) Gene set enrichment analysis profiles for each compound based on whole genomic CRANKS scores within the queried gene set in the glutathione metabolism pathway (KEGG hsa00480). Plots representing the cumulative enrichment score (ES), the position of the pathway genes within the ranked list of whole genomic CRANKS scores (black bars), from most resistant (red bars) to most sensitizing genes (blue bars). Calculated normalized enrichment scores (NES) and Nominal p-value, determined by 10,000 permutation tests: GTE (−1.60; 0.0077) EGCG (−1.31; 0.087); GA (−1.42; 0.040), DG (−1.10; 0.289) Galu (1.20; 0.176). GSEA pathways were selected from previous top hits and pathway enrichment analyses (Fig. 1, Fig. 2). Gene-related subpanels (biological processes) were used for further evaluation. (C) Heatmap analysis of a pairwise correlation matrix between all pairs of each compound data sets of CRANKS scores for 50 genes in the glutathione metabolism pathway (KEGG hsa00480), each cell represents the correlation of CRISPR scores (CRANKS) between two treatments (row vs. column). Yellow squares signify higher positive correlations (closer to +1), whereas blue squares denote lower or negative correlations. (D-G) 2D scatter plots showing each compound's CRANKS values (y-axis) plotted against GTE (x-axis) for the same glutathione-related genes, with a best-fit linear regression trend line and Pearson's correlation coefficient (r). Higher positive r-values indicate closer similarity to GTE's CRISPR “fingerprint”, whereas negative or near-zero correlations suggest divergent mechanisms. (H-M) Similar analysis performed for 78 genes in the peroxisome pathway (KEGG hsa04146). GSEA's NES and Nominal p-value: GTE (1.64; 0.0038) EGCG (1.35; 0.055), GA (1.23; 0.128), DG (−2.18; p < 0.00001) Galunisertib (1.57; 0.0054). Significance of correlation is indicated at the 95 % confidence interval, with p values summarized as ∗∗p < 0.01, ∗∗∗p < 0.001, or non-significant (ns).

We next examined how structural variations in gallate esters influenced peroxisome-related gene responses. Peroxisome GSEA plots demonstrated similar cumulative enrichment scores patterns between GTE, EGCG and GA (Fig. 3H). The positive NES suggested resistance across GTE (1.63), EGCG (1.35) and GA (1.23), most significantly with GTE. Together sharing specific leading edge core resistance genes, involving multiple PEXs genes, as compared to Galunisertib also having a significant positive score (1.57), but rather with distinct core genes. In contrast, DG showed stark differences with exceptionally significant high negative score (−2.18), with clear opposing leading-edge gene enrichment core sensitizing effect, including PEXs genes. These results suggest that GTE mixture exhibits resistance as components like EGCG and GA can override other potential components carrying strong peroxisome sensitizing effects, exemplified here by DG. Further analysis showed EGCG, GA, and GTE clustered closely in their CRISPR KO profiles for peroxisomal genes (Fig. 3I), with EGCG showing the strongest correlation to GTE (r = 0.75, p < 0.0001), followed by GA (r = 0.68, p < 0.0001). In contrast, DG exhibited a marked negative correlation (r = −0.45, p < 0.0001), suggesting that its long alkyl chain significantly altered the interaction of the gallate moiety with peroxisome-dependent processes (Fig. 3J–M). This divergence was particularly evident in the behavior of key peroxisomal import genes, PEX1, PEX6, PEX12, and PEX14, which shifted from sensitizers (negative CRANKS scores) under GTE, EGCG, and GA to rescuers (positive scores) under DG, indicating a reversal in functional impact on peroxisomal matrix protein import.

Further analysis revealed compound-specific modulation of antioxidant defense genes. PRDX1, a key ROS-detoxifying enzyme, transitioned from a sensitizing hit under GTE (−3.53), EGCG (−3.36), and GA (−4.27) to a resistance-conferring gene under DG (4.31), with Galunisertib showing minimal impact (0.61). Similarly, CAT exhibited a progressive reduction in sensitization across treatments (GTE: −3.79; EGCG: −1.79; GA: −1.04; DG: −0.34; Galunisertib: −0.62), suggesting that H₂O₂ detoxification via CAT may be less critical or differently regulated under DG and Galunisertib. These patterns reinforce the role of EGCG as the principal mediator of GTE's antiproliferative effects, acting through glutathione-dependent redox homeostasis and peroxisomal integrity. Collectively, these findings demonstrate that structural modifications to the gallate moiety, particularly the addition of a long alkyl chain in DG, can significantly reshape the redox and peroxisomal gene response landscape. While EGCG closely mirrors GTE's cytotoxic signature and GA partially recapitulates it, DG diverges sharply, especially in its impact on antioxidant defense and peroxisomal import machinery. Lacking the galloyl ester moiety, Galunisertib consistently showed minimal engagement with redox-related genes, highlighting the mechanistic specificity of gallate-containing compounds in driving oxidative stress and cell death.

3.5. Gallate ester structure dictates selective engagement of glutathione and peroxisomal stress pathways

Given the relatively consistent shifts observed among gallate ester polyphenols in GTE, and the structural influence of alkyl chain addition on peroxisomal gene responses noted previously, we next sought to clarify the extent to which specific biological pathways mediate treatment-induced cytotoxicity. In particular, we aimed to determine whether glutathione biosynthesis genes were uniformly modulated across treatments or if distinct functional gene subsets were differentially affected. To refine this analysis, we examined CRISPR KO responses for genes involved in glutathione biosynthesis and peroxisomal import across EGCG, GA, DG, and the non-galloyl control compound Galunisertib. Within the glutathione biosynthesis gene set, two distinct functional clusters emerged: a sensitizing group composed of GCLM, GCLC, and GSS, which are key regulators of glutathione synthesis, and a rescuing group including GGTs and CHACs, which are involved in glutathione catabolism. These patterns were consistently observed across gallate-containing treatments, while Galunisertib remained largely neutral or showed a slight opposing trend in catabolic genes, reinforcing its mechanistic distinction (Fig. 4A). Although pooled statistical analysis of glutathione biosynthesis genes revealed significant differences (p < 0.0001), with GCLC and GSS contributing most to the sensitizing effect, at this level of analysis, treatment-specific differences were not detectable due to opposing regulatory influences within the gene set (Fig. 4B).

Fig. 4.

EGCG's structural specificity in glutathione biosynthesis and peroxisomal import processes. (A) Unclustered heatmap overview (relative to GTE-incremented scores) comparing CRISPR KO (CRANKS) scores for a 14-gene subset involved in the human glutathione biosynthetic process (GO:0006750). Each row represents a specific gene, and each column represents a treatment: Green tea extract (GTE), EGCG, gallic acid (GA), dodecyl gallate (DG), or Galunisertib (Galu). Red cells indicate positive CRANKS scores (sensitizers), whereas green cells indicate negative CRANKS scores (rescuers). Intermediate (near-zero) values appear in lighter shades, reflecting a weaker phenotype. (B–D) Box-whisker plots comparing CRANKS score distributions for treatments in (B) Glutathione biosynthesis process genes, which were stratified by (C) Glutathione catabolism genes (GGT1, GGT5, GGT7, CHAC1-2) and (D) Ferroptosis-related genes (GCLM, GCLC, GSS, SLC7A11). (E) Heatmap analysis for peroxisomal matrix protein import genes (GO:0016558, 13-gene subset). (F) Box-whisker plot analysis summarizing CRANKS scores for the peroxisomal import gene panel. Legend key: Positive CRANKS scores indicate resistance KO that reduce compound toxicity. Negative scores indicate sensitizing knockouts that heighten compound toxicity. Each box color represents the median CRANKS values as either sensitizing (red) or rescuing (green), with whiskers indicating the Min-Max range for each treatment group. Statistical analyses: (A, E) Two-way ANOVA followed with Bonferroni's post hoc test. (B, C, D, F). Data are presented with median indicated, p-values were calculated using a one-way ANOVA followed by Newman Keuls post hoc test. Significance is indicated at the 95 % confidence interval; Statistical significance is denoted as follows: ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; or non-significant (ns). (A) Treatment effect (F = 1.91, p = 0.12, not significant); Gene effect (F = 6.32, p < 0.0001). Post hoc: GTE vs. Galu (GSS, GCLC: p < 0.05); EGCG vs. Galu (GCLC: p < 0.05). (B) Glutathione biosynthesis process genes: n.s (C) Glutathione catabolism genes: (p = 0.0086) (D) Ferroptosis-related genes: (p = 0.010). (E) Peroxisomal matrix import genes: Treatment effect (F = 28.3, p < 0.0001); Gene effect (F = 1.35, p = 0.222, not significant). Post hoc: GTE vs. Galu (PEX1: p < 0.001, PEX12: p < 0.01); EGCG vs. DG (PEX1: p < 0.001, PEX2: p < 0.05, PEX12: p < 0.05, PEX13: p < 0.05); GA vs. DG (PEX1: p < 0.05, PEX2: p < 0.05, PEX12: p < 0.01). (F) Peroxisomal matrix protein import genes: (p = 0.04).

To resolve this complexity, we stratified the genes into two subsets: glutathione catabolism (Fig. 4C) and glutathione synthesis/ferroptosis-related genes (Fig. 4D). This refined analysis revealed significant treatment-dependent effects (p < 0.01), particularly in the glutathione catabolism group, where all gallate-containing compounds showed a consistent rescuing phenotype relative to Galunisertib (Fig. 4C). The glutathione synthesis-related subset, which includes ferroptosis regulators such as SLC7A1, GSS, and the rate-limiting enzyme GCLs [[15], [16], [17], [18]] showed strong treatment-driven modulation, with significant differences across compounds (p < 0.01) (Fig. 4D). Post hoc comparisons confirmed that all gallate-containing treatments significantly differed from Galunisertib, particularly in their impact on GCLC and GSS, reinforcing the link between glutathione homeostasis and ferroptosis sensitivity.

3.6. Peroxisomal import genes uncover mechanistic specificity in gallate ester-induced redox stress

To further investigate the role of peroxisomal function in compound-induced cytotoxicity, we analyzed the CRISPR KO profiles of peroxisomal matrix protein import genes across treatments with GTE, EGCG, GA, DG, and the non-galloyl control Galunisertib. Peroxisomal import genes, particularly PEX1 and PEX12, displayed significant differential sensitivity to gallate esters, with GTE, EGCG, and GA inducing strong sensitization, while Galunisertib remained largely neutral (Fig. 4E). These findings suggest that oxidative stress adaptation mechanisms extend beyond glutathione regulation and involve peroxisomal integrity as a key vulnerability. Notably, cells lacking proper peroxisomal import machinery exhibited increased viability under GTE, EGCG, and GA, whereas DG treatment led to reduced survival, indicating a reversal in phenotype. Post hoc comparisons confirmed the significance of these differences, especially between GTE/EGCG and DG (p < 0.0001), highlighting DG's distinct regulatory impact on peroxisomal stress responses (Fig. 4F). Genome-wide CRISPR profiling across structurally related gallate esters revealed consistent patterns of genotype-specific sensitivity and resistance. EGCG, the dominant catechin in GTE, closely mirrored the broader GTE phenotype, sensitizing cells to KO of glutathione biosynthesis genes (GCLM, GCLC, GSS) and oxidative stress regulators (PRDX1), while resistance was conferred by loss of KEAP1 and PEX genes. Comparative analysis showed that these effects are dynamically shaped by the structural configuration of the galloyl group. In particular, the long-chain derivative DG inverted key CRANKS scores, suggesting a distinct mechanism of action that alters redox and peroxisome-linked stress responses. These results underscore the mechanistic specificity of gallate esters and demonstrate how subtle structural variations can profoundly influence cellular sensitivity to oxidative stress and anti-proliferative effects.

4. Discussion

Genome-wide screens have identified key genes that regulate cellular sensitivity to EGCG and GTE toxicity, revealing a critical redox balance between glutathione metabolism and peroxisomal function as essential for cell survival. Among approximately 19,000 genes analyzed, those involved in glutathione synthesis, ferroptosis, ROS detoxification, and peroxisomal import emerged as particularly relevant, especially in the context of structural variations in galloyl esters that shift genetic dependencies.

Glutathione synthesis enzymes, notably GCLs, are central to antioxidant defense. Galloyl esters such as EGCG were found to deplete glutathione in GCL-deficient cells, leading to oxidative stress [19]. While glutathione turnover via gamma-glutamyl transferases modulates sensitivity, cells lacking GCLs become highly susceptible to ferroptosis [18,20]. This vulnerability is linked to EGCG's pyrogallol structure, which enhances redox cycling and lipid peroxidation through Fenton chemistry [7,[21], [22], [23]]. EGCG exhibits concentration-dependent effects as low concentrations activate survival pathways, while higher concentrations elevate p53 and pro-apoptotic genes such as BAK1 [24,25]. Its pyrogallol rings undergo auto-oxidation, generating ROS and hydroxyl radicals, a process accelerated by metal ions [26,27]. At low concentrations (1–10 μM), EGCG promotes cytoprotection, whereas higher concentrations (>50 μM) overwhelm antioxidant defenses, triggering cytotoxicity. EGCG oxidation leads to o-quinone (cyclohexa-3,5-diene-1,2-dione) formation, which depletes glutathione and activates stress kinases (JNK, PI3K, ERK), inducing oxidative stress response genes [28,29].

Catechins, including EGCG, act as pro-oxidants that induce transient oxidative stress and hormesis, selectively enhancing the resilience of surviving cells [4]. EGCG engages the KEAP1-NRF2 oxidative stress sensor, activating genes involved in antioxidant defense and glutathione synthesis (SLC7A11, GSS, GCLC, GCLM), along with detoxifying enzymes such as CAT and PRDX1 [22,24,[30], [31], [32]]. Studies have shown that EGCG binds KEAP1-NRF2 and forms glutathione adducts that target KEAP1 cysteines [24,33]. Its effects are both dose- and context-dependent as EGCG activates NRF2 in normal cells, promoting antioxidant defenses, but suppresses NRF2 in cancer cells, thereby enhancing oxidative stress-induced apoptosis [34,35]. This selective modulation underlines EGCG's potential in cancer prevention. Complementing these mechanisms, the dependency on selenocysteine machinery has also been described as a liability for cancer cells. LRP8 is required for selenium uptake in cells and promotes resistance to ferroptosis in cancer cells and SEPHS2 is responsible for the synthesis of selenophosphate from selenide and ATP [36]. Together, these findings demonstrate that glutathione catabolism and ferroptosis regulation are distinctly responsive to gallate esters, whereas non-galloyl Galunisertib exerted minimal influence. This supports the hypothesis that gallate-containing compounds selectively engage redox and peroxisomal stress pathways and further justifies the investigation of peroxisomal dynamics as a central component of their anti-proliferative mechanism.

PRDX1 plays a key role in detoxifying intracellular H₂O₂, suppressing ferroptosis, and maintaining redox balance, functions that are crucial for countering EGCG-induced stress [30,37,38]. Recent screens identified EGCG as a potent activator of PRDX1's H₂O₂-scavenging activity at low concentrations [39]. However, under sustained oxidative stress, PRDX1 becomes hyperoxidized, losing its peroxidase function and shifting to a chaperone role that increases intracellular H₂O₂ and triggers KEAP1 inhibition [25,40,41]. EGCG may exhaust PRDX1 activity, and its deletion removes a critical buffer against oxidative stress. Interestingly, PRDX1 loss increased resistance to DG suggesting that alkylation may disrupt PRDX1 oligomerization or membrane interactions. PRDXs, including PRDX1, also exhibit tumor-promoting roles, reflecting the context-specific nature of oxidative signaling [32]. Given PRDX1's broad subcellular distribution [[42], [43], [44]], this again prompts further investigation into peroxisomal contributions.

Peroxisomes are vital for oxygen metabolism, generating approximately 35 % of cellular H₂O₂. They rely on PRDXs, CAT, and glutathione for redox homeostasis [43,45]. Although peroxisomes lack their own genome, they import proteins primarily via PEX genes. Supporting our findings, loss of import factors PEX1 and PEX12 reduces peroxisome abundance and confers resistance to ferroptosis [46]. PEX12 mediates membrane protein translocation, while PEX1, a component of the exportomer complex, is essential for matrix protein import and overall peroxisomal function. Deficiencies in PEX1 or PEX6 trigger pexophagy, disrupting redox and apoptotic signaling [[47], [48], [49]], thereby weakening EGCG-induced oxidative stress and ferroptosis. Moreover, functional peroxisomal membranes are permeable to glutathione, which protects matrix proteins from H₂O₂ oxidation even in the absence of antioxidant enzymes [50]. This protective mechanism is likely compromised by EGCG's glutathione conjugation and depletion, intensifying peroxisomal oxidative load.

Although EGCG's peroxisomal impact in animals remains underexplored, GTE has been shown to trigger peroxisome proliferation [51] and stimulate fatty acid β-oxidation [52], both of which are beneficial in combating obesity and metabolic disorders. Catechins exhibit dose-dependent hormesis versus cytotoxicity, as demonstrated in plant and lower animal studies [4,53]. Notably, GA and methyl gallate accumulate significantly (∼70 %) in the peroxisomes of Camellia sinensis [54], suggesting that peroxisomes serve as both ROS sources and polyphenol reservoirs, amplifying local ROS and lipid peroxidation. Conversely, impairing peroxisome function may mitigate galloyl-induced oxidative stress. Small structural changes, such as alkyl chain length, may influence membrane partitioning and organelle targeting, thereby modulating cytotoxicity and therapeutic potential.

While EGCG mirrors the cytotoxic signature of whole GTE, additional rescuers, such as BAK1, and sensitizers have also emerged. A notable finding was the specific vulnerability of cells lacking ABCC1 under EGCG treatment. ABCC1 is further known to mediate ATP-dependent efflux of glutathione conjugates and xenobiotics [55], and its KO may facilitate intracellular accumulation of EGCG or its oxidative derivatives, amplifying stress responses. ABCC1 exports glutathione conjugates and xenobiotics, and its loss likely impairs EGCG or its oxidative byproduct efflux, increasing intracellular accumulation and redox stress. ABCC1's known role in polyphenol export and chemoresistance supports this model [55,56]. These differences refine our understanding of EGCG accumulation, oxidative stress, and therapeutic response.

Collectively, these findings demonstrate that EGCG engages a conserved redox-peroxisomal stress response signature similar to GTE, while also revealing unique genetic vulnerabilities, such as ABCC1, that may underlie EGCG's enhanced cytotoxicity. These findings underline the power of chemogenomic screening to uncover coordinated, organelle-spanning networks that govern cellular sensitivity to polyphenol-induced oxidative stress. By dissecting EGCG's specific contribution within the broader GTE context, this approach provides a mechanistic framework for understanding how redox vulnerabilities may be exploited for therapeutic purposes.

DG appears to kill cells via a mechanism distinct from EGCG and GA, relying less on H₂O₂ buildup and more on membrane-targeted oxidative damage. Its C12 alkyl chain enhances cellular uptake, lipid bilayer insertion, and organelle membrane destabilization. Similar to propyl gallate, which disrupts peroxisomal integrity, DG likely damages peroxisomes and mitochondria, causing focal oxidative stress by blocking enzyme diffusion and triggering autolysis [57,58]. DG's dependency on BAK1 suggests pore formation in peroxisomal membranes [59], enabling ROS and enzyme leakage and uncoupling membrane stress from H₂O₂ detoxification. DG-treated cells showed reduced reliance on CAT and PRDX1, potentially relying on other mechanisms for superoxide clearance, diverging from EGCG's H₂O₂-centric action. This aligns with reports that alkyl gallates with chains ≥ C8 induce apoptosis more efficiently via ROS generation, calcium influx, and membrane permeability [60,61]. Overall, DG likely acts through BAK1-driven membrane permeabilization, lipid peroxidation, and alternate detox pathways, illustrating how subtle structural changes can shift galloyl ester cytotoxicity and localization.

Our data support a model where EGCG undergoes autoxidation, generating sustained intracellular H₂O₂. Cells deficient in peroxisomal import or with constitutive NRF2 activation show resistance to EGCG, highlighting the role of peroxisomes and glutathione-dependent antioxidant pathways in modulating its toxicity. Loss of PEX genes appears protective, suggesting that functional peroxisomes may amplify EGCG's cytotoxicity by concentrating oxidative reactions or sequestering the compound. In contrast, PEX deficiency reduces oxidative burden, enhancing resistance. Intact peroxisomes, however, may allow redox imbalance to exceed cancer cell tolerance, revealing a therapeutic vulnerability.

EGCG's pro-oxidant properties selectively target cancer cells reliant on glutathione biosynthesis and peroxisomal import, pathways often dysregulated in obesity-related malignancies. These tumors typically exhibit elevated ROS and compromised antioxidant defenses, which EGCG exploits by promoting H₂O₂ accumulation that overwhelms detox systems [62,63]. This selective cytotoxicity spares healthy tissue while reducing adiposity and inflammation, thereby lowering obesity-driven cancer risk [64,65]. Our screens identified exploitable antioxidant vulnerabilities, suggesting that combining EGCG with ferroptosis inducers, glutathione synthesis inhibitors, or peroxisomal modulators could enhance its anticancer efficacy, particularly in ROS-dysregulated tumors. While our in vitro findings offer mechanistic insight, additional cancer cell screen models will be essential to assess the broader applicability of these genetic vulnerabilities. In vivo validation will also be essential to confirm whether dependencies on PEX and glutathione genes predict therapeutic response. Optimizing EGCG dosing and delivery is critical to balance efficacy and safety. Further research is needed to understand how obesity reshapes redox and metabolic pathways, enabling precision nutrition or combination therapies that sensitize tumors to EGCG.

5. Conclusion

In conclusion, genome-wide CRISPR screens reveal that EGCG, and related green tea polyphenols disrupt redox balance and peroxisomal function, inducing cytotoxicity in cancer cells dependent on glutathione synthesis and peroxisome biogenesis. The galloyl core drives oxidative stress, while alkyl chain length modulates organelle targeting and redox vulnerabilities. Key sensitivity genes, including KEAP1, PRDX1, and PEXs, emerge as potential therapeutic targets. These pro-oxidant mechanisms provide novel insights from the current genome-wide CRISPR screening (Fig. 5), as findings further lay the groundwork for developing EGCG-based strategies to exploit oxidative stress in cancers with redox imbalances, particularly those linked to obesity.

Fig. 5.

EGCG-induced pro-oxidant mechanisms: Insights from genome-wide CRISPR screening. (1) EGCG undergoes auto-oxidation generating reactive oxygen species (ROS) such as hydrogen peroxide (H₂O₂) and quinone intermediates. A genome-wide CRISPR-Cas9 screen identified key genes modulating cellular response to EGCG-induced oxidative stress. (2) Knockouts of glutathione biosynthesis genes (GCLC, GCLM, GSS) impair H₂O₂ detoxification, sensitizing cells to ferroptosis. (3) Peroxisomal genes (PEX1, PEX6, PEX12, PEX14) regulate ROS metabolism; their disruption compromises H₂O₂ clearance via catalase (CAT) and peroxiredoxin-1 (PRDX1). (4) The KEAP1-NRF2 axis controls antioxidant gene expression, inducing enzymes (e.g., CAT, PRDX1, GCLs) that mitigate ROS toxicity. KEAP1 knockout enhances NRF2 signaling and confers resistance. (5) Additional modulators include ABCC1 (drug efflux), SLC7A11 (cysteine transporter) and BAK1 (apoptosis). Sensitizer hits (red labels) indicate knockouts that heighten EGCG toxicity, while resistance hits (green labels) protect against cell death. Color intensity reflects CRANKS scores relative to the highest scoring genes. Together, these findings illustrate how EGCG's pro-oxidant activity can overwhelm cancer cell defenses when redox-regulating pathways are genetically compromised.

Funding

This work was funded by the Institutional Research Chair in Cancer Prevention and Treatment held by Dr Borhane Annabi at UQAM, and by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC, RGPIN-2024-04541) to BA.

CRediT authorship contribution statement

Naoufal Akla: Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Anes Boudah: Data curation, Formal analysis, Writing – review & editing. Thierry Bertomeu: Conceptualization, Data curation, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing. Andrew Chatr-aryamontri: Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review & editing. Michel Desjarlais: Formal analysis, Writing – original draft, Writing – review & editing. Borhane Annabi: Conceptualization, Formal analysis, Funding acquisition, Project administration, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Borhane Annabi reports financial support was provided by Natural Sciences and Engineering Research Council of Canada. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Abbreviations

CAT

Catalase

CRANKS

Context-dependent Robust Analytics and Normalization for Knockout Screens

DG

Dodecyl Gallate

EGCG

Epigallocatechin Gallate

FDR

False Discovery Rate

GA

Gallic Acid

GO

Gene Ontology

GSEA

Gene Set Enrichment Analysis

GTE

Green Tea Extract

KEAP1

Kelch-like ECH-associated Protein 1

KO

Knockout

NES

Normalized Enrichment Scores

PPI

Protein-Protein Interaction

PRDX1

Peroxiredoxin 1

ROS

Reactive Oxygen Species

TβR1

Transforming Growth Factor Beta Receptor 1

Appendix A. Supplementary data

The following are the supplementary data to this article.

Supplementary Fig. S1.

Comparative analysis of CRANKS scores for EGCG and GA. Panel (A) presents a scatter plot comparing CRANKS scores from genome-wide CRISPR-Cas9 KO screens for EGCG (y-axis) and GA (x-axis). Each dot represents a gene, positioned according to its CRANKS scores for both compounds. Genes are color-coded: red dots indicate negative CRANKS scores (Synthetic Lethals/Sensitizers), and green dots indicate positive CRANKS scores (Rescues/Suppressors). Key genes are labeled for reference. Panel (B) shows a bar plot of CRANKS scores for genes identified as Rescues/Suppressors in both EGCG and GA screens. Genes are listed on the y-axis, with horizontal bars representing CRANKS scores for EGCG (right panel) and GA (left panel). Reverse-drawn lines across bars illustrate score differentials. Note: In both panels, CRANKS scores reflect the impact of gene KO on cell proliferation in the presence of each compound, calculated as log-fold changes in sgRNA abundance relative to untreated controls. Positive scores indicate sgRNA enrichment (resistance), while negative scores indicate depletion (sensitivity). Consistent color-coding, green for positive and red for negative scores, enhances visual interpretation. Discussion: To determine the specific contribution of the gallate moiety in green tea extract (GTE) and EGCG's biological activity, we compared gene knockout effects on cellular responses to EGCG and gallic acid (GA). GA contains the gallate group without the catechin structure, allowing independent assessment of the gallate moiety's influence. While EGCG and GA share redox-modulating properties, structural differences lead to distinct genetic dependencies. CRANKS score analysis revealed both overlapping and unique gene-level responses (Supplementary Fig. S1A). Knockouts of GCLC and GCLM, key enzymes in glutathione synthesis, increased sensitivity to both compounds, indicating that the gallate moiety induces oxidative stress mitigated by glutathione-dependent pathways. Conversely, KEAP1 knockout conferred resistance to both EGCG and GA, consistent with NRF2 pathway activation and enhanced antioxidant defenses. Some genes showed compound-specific effects. For example, knockouts of APEX2 and TRAF2 increased sensitivity only to EGCG, suggesting that EGCG's catechin structure engages additional pathways related to DNA repair and apoptosis. Bar plots of CRANKS scores for shared resistance genes (Supplementary Fig. S1B) showed consistent positive scores under both treatments, highlighting the gallate moiety's role in mediating resistance via oxidative stress response and peroxisomal function. These findings underscore the gallate group as a key determinant of cellular sensitivity and resistance, while EGCG's catechin structure contributes additional biological effects. Understanding these distinct molecular contributions can inform strategies to enhance the anticancer efficacy of EGCG and related compounds.

Supplementary Fig. S2.

Bar plot of CRANKS scores for synthetic lethals/sensitizers. This figure presents a bar plot of CRANKS scores for genes whose KO increases cellular sensitivity to EGCG and GA. Genes are listed along the y-axis, with horizontal bars representing CRANKS scores for each compound. Negative CRANKS scores indicate sgRNA depletion, reflecting enhanced sensitivity and suggesting that these genes play protective roles in cancer cell survival. This visualization highlights key genetic vulnerabilities that may be exploited to enhance the therapeutic efficacy of EGCG and related galloyl compounds. Discussion: The CRISPR/Cas9 chemogenomic screen identified genes whose knockout significantly increased cellular sensitivity to EGCG and GA, with CRANKS scores of −2.0 or lower. Some genes, such as TRAF2, LRP8, and UBE2N, showed pronounced sensitivity to EGCG, while also displaying moderate sensitivity to GA. Conversely, genes like AHCY, DDX60L, RB1CC1, and MSL2 exhibited stronger synthetic lethality with GA, suggesting compound-specific roles in mediating polyphenol-induced cytotoxicity. Moderate sensitivity differences were observed for genes such as ATP8B2, ZFX, LRPAP1, and PFN1, which were more sensitive to GA than EGCG. In contrast, APEX2, ATP6V0B, VIPR2, DHH, UBE4B, C12orf66, PDIA5, NUP88, SPRYD3, and MIS18BP1 showed higher sensitivity to EGCG, indicating differential involvement in each compound's mechanism of action. Distinct compound-specific responses were noted for genes like MYO19, GSPT2, NFKBIZ, and CCDC91, which were highly sensitive to EGCG but resistant to GA. Similarly, FRYL, NADK, and CYB5A were sensitive to GA with minimal response to EGCG. These patterns suggest that certain genes play unique roles in mediating cellular survival depending on the compound. The structural differences between EGCG and GA may explain these variations. EGCG contains a gallate moiety, contributing to shared sensitivities, while its catechin structure may drive EGCG-specific responses. GA lacks this catechin component, which may account for its distinct gene interactions. These insights highlight the importance of molecular structure in determining compound-specific genetic vulnerabilities and inform strategies to enhance the therapeutic potential of polyphenols.

Supplementary Fig. S3.

CRISPR screening distributions across compounds. This figure presents the overall distribution of CRISPR screening results for each tested compound using 2D scatter plots. Panel (A) shows raw CRANKS scores (y-axis) plotted by gene index (x-axis) for GTE, EGCG, GA, DG, and Galu. Panel (B) displays CRANKS scores ranked from lowest to highest (x-axis, log₁₀ scale), with each dot representing a gene's average score across biological replicates. Normality was assessed using the D'Agostino & Pearson omnibus test, which indicated non-Gaussian distributions for all compounds (∗∗∗p < 0.0001). Total CRANKS score sums were GTE = 172.4, EGCG = 149.9, GA = 172.5, DG = 123.9, and Galu = 213.1. Median scores were compared to a hypothetical zero using the Wilcoxon signed-rank test, yielding: GTE: 0.012, p = 0.01; EGCG: 0.012, p = 0.02; GA: 0.014, p = 0.025; DG: −0.00003, p = 0.75 (n.s.); Galu: 0.016, p = 0.009. These results highlight distinct gene-level responses and compound-specific screening profiles.

Data availability

Data will be made available on request.