Abstract
GST omega 1 (GSTO1) is an atypical GST isoform that is overexpressed in several cancers and has been implicated in drug resistance. Currently, no small-molecule drug targeting GSTO1 is under clinical development. Here we have validated GSTO1 as an impactful target in oncology. Transcriptional profiling coupled with proteomics uncovered novel pharmacodynamic markers and cellular pathways regulated by GSTO1. CRISPR/Cas9 GSTO1 knockout (KO) cell lines failed to form tumors or displayed growth delay in vivo; they also formed smaller 3D spheroids in vitro. Multiomics analysis in GSTO1 KO cells found a strong positive correlation with cell adhesion molecules and IFN response pathways and a strong negative correlation with Myc transcriptional signature. In addition, several clinically used drugs showed significant synthetic lethality with loss or inhibition of GSTO1. Transcription and protein expression of tissue factor (gene name, F3) were downregulated in response to GSTO1 KO. F3 is associated with poor patient survival and promotion of tumor progression in multiple cancers and is a known risk factor for metastasis. Transcription of F3 was regulated by IL1β, whose secretion decreased upon inhibition of GSTO1, suggesting that IL1β links GSTO1 expression and F3 transcription. In summary, our results implicate GSTO1 as a potential therapeutic target in cancer and offer new mechanistic insights into its significant role in cancer progression.
Significance: These findings validate GSTO1 as a therapeutic target in cancer and implicate GSTO1 in the modulation of tumor growth, immune responses, and expression of F3.
Introduction
GST omega 1 (GSTO1), a member of the cytosolic GST family, is widely expressed in human tissues (1). GSTs are important detoxifying enzymes that catalyze the conjugation of electrophilic substrates to glutathione, and play a critical role in cellular detoxification (2). With a cysteine residue present in the active site instead of a tyrosine commonly found in other GSTs, GSTO1 carries distinct enzymatic and metabolic functions. It plays important roles in biotransformation of xenobiotics, glutathionylation, deglutathionylation, and redox detoxification (3). GSTO1 overexpression and polymorphism correlate with several pathologic diseases including neurologic disorders, cancers, and inflammatory diseases (3).
GSTO1 is upregulated in several human cancers including esophageal squamous cell carcinoma (4, 5), colorectal (5, 6), and urinary bladder cancer (7). However, how GSTO1 is mechanistically linked with disease progression is largely unknown. GSTO1 overexpression is associated with drug resistance (8, 9), and GSTO1 inhibitors have shown to sensitize cancer cell to cisplatin (5, 10). GSTO1 expression is also induced by chemotherapy and its knockdown halts breast cancer stem cell enrichment (11). Knockdown of GSTO1 leads to cell apoptosis in colorectal (5) and erythroleukemia cell lines (12) and causes significant reduction in cell survival (13). The glutathionylation of intracellular protein serves as an essential posttranslational modification modulating signal transduction (14, 15) and exerts diverse impact on cellular processes. The glutathionylation/deglutathionylation function positions GSTO1 as a key regulator of critical proteins in cell growth pathways (16, 17). Moreover, GSTO1 is crucial in modulating the Toll-like receptor 4 (TLR4)-mediated proinflammatory pathway (18), promoting NLRP3 inflammasome activation (19) and attenuating the release of proinflammatory cytokines including IL1β, IL6, and TNF (20), which are important in the tumor immune environment. GSTO1 may significantly contribute to various pathologies; however, the roles of GSTO1 in cancer progression are largely unexplored.
To fully characterize the roles of GSTO1 in cancer progression, we employed CRISPR/Cas9 to knockout GSTO1 in multiple cancer cell lines and evaluated its impact using a variety of phenotypic and genomic profiling. Our results demonstrate that depletion of GSTO1 suppressed tumor growth in vivo and inhibited three-dimensional (3D) spheroid growth in vitro. Using multiomics genome-wide profiling [bromouridine nascent RNA sequencing (Bru-seq), RNA sequencing (RNA-seq), and MS proteomics], we have characterized the transcriptional and translational response to GSTO1 deletion. Loss of GSTO1 caused transcriptional suppression of genes involved in cell-cycle checkpoints, as well as target genes for Myc and E2F transcription factors. Furthermore, genes and proteins involved in cell adhesion, IFNα, and IFNγ response were enriched in GSTO1 KO cells. On the basis of transcriptome signature of GSTO1 KO cells, we tested select FDA-approved drugs in GSTO1 KO cells and expanded the utility of GSTO1 inhibitors in combination with standard-of-care drugs. Cumulatively, our study demonstrates the significance of GSTO1 in cancer progression, reveals extensive regulation of gene expression by GSTO1, and further validates GSTO1 as a therapeutic target in cancer.
Materials and Methods
Cell lines
HCT116 (ATCC, CCL-247) and GSTO1 KO cells were cultured in RPMI1640 supplemented with 10% FBS. U-87 MG (generously provided by Dr. Alan L. Epstein, University of Southern California, Los Angeles, CA), A172 (ATCC, CRL-1620), and their related GSTO1 KO cells were cultured in DMEM supplemented with 10% FBS. Cells were kept at 37°C in a humidified atmosphere of 5% CO2. All cell lines were maintained in culture under 30 passages and tested regularly for Mycoplasma contamination using PlasmoTest.
Reagents
CB-839, mitomycin C, cryptotanshinone, altretamine, GSK583, liproxstatin-1, spautin-1, PD-1/PD-L1 inhibitor 2, RVX-208, selinexor, ferrostatin-1, diacerein, LY2603618, and patupilone were purchased from Selleckchem. Cisplatin, carboplatin, and oxaliplatin were purchased from Sigma-Aldrich. Olaprib, niraparib, and veliparib were purchased from Cayman Chemical. As2O3 was purchased from BioTang. C1-27 is a GSTO1 inhibitor previously identified and synthesized by our laboratory (5).
CRISPR/Cas9-mediated genome editing
GSTO1 KO cancer cell lines were generated using the Edit-R CRISPR-Cas9 system from Dharmacon (GE Healthcare). Edit-R Lentiviral hEF1α-Blast-Cas9 lentiviral particles (generously provided by Phillip L. Palmbos, University of Michigan, Ann Arbor, MI) were transduced into U-87 MG, HCT116, and A172 cells and stable Cas9-expressing lines were generated by blasticidin selection. Cells were cotransfected with tracrRNA and GSTO1-targeting crRNA (CM-011950-01-0002) specific for exon 2 or nontargeting crRNA control (U-002000-05). Individual colonies were further selected and screened for loss of GSTO1 expression. Targeted mutations were confirmed using Sanger sequencing.
Animal studies
Female NOD/SCID mice, 6- to 10-week-old, were used for implantation of HCT116 survival experiments. Female ICRSC mice, 6- to 10-week-old, were used for implantation of HCT116 and U-87 MG tumor monitoring experiments. A total of 1 × 106 HCT116 or 2 × 106 U-87 MG tumor cells were implanted subcutaneously into right flank regions of mice. Tumor growth was monitored twice a week by digital caliper and tumor volumes were calculated by the (length × width2)/2 equation. Mice were sacrificed once tumors reached 1,500 mm3. The University of Michigan Institutional Animal Care and Use Committee approved all animal experiments.
Pan-disease gene expression analysis
Patient data described in this work were, in part, based upon data generated by the The Cancer Genome Atlas (TCGA) Research Network: http://cancergenome.nih.gov/. Patient sample RNA-seq RSEM normalized gene expression values and related survival metadata were sourced from the TCGA GDAC Firehose (21). When multiple samples were available for a given patient, barcodes were sorted alphabetically and the first was selected for analysis. A total of 9,718 samples were evaluated across 34 total disease groups.
TCGA gene expression data downloaded from GDAC Firehose is upper-quartile normalized for comparison within disease groups. To survey GSTO1 expression across all 34 TCGA disease groups, we normalized log2 RSEM expression values by calculating z-scores for each patient using the following steps for each patient: (i) calculate the patient mean and patient SD for all protein-coding gene expression values; and (ii) Calculate GSTO1 z-score by subtracting the patient mean and dividing by the patient SD. The resulting GSTO1 z-scores are shown in Fig. 1A.
GSTO1 is upregulated in select cancers and correlates with decreased patient survival. A, Above average expression of GSTO1 was detected across 34 TCGA diseases. B, Kaplan–Meier analysis of five cancer types identified lower patient survival was associated with higher expression of GSTO1. Patients stratified using 75% expression percentile. GBMLGG, glioblastoma and lower grade glioma; KIRC, kidney renal clear cell carcinoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; UVM, uveal melanoma.
TCGA disease patient samples were evaluated for reduced survivability by comparing survival outcomes for patients with high and low expression of GSTO1. Thresholding for high and low expression patient populations was evaluated using five different quantile cutoffs: 95%, 90%, 75%, 50%, and 25%. A log-rank test statistic was calculated for each cutoff to compare the survival distributions of high and low expression patient populations with the null hypothesis that there was no difference in survival curves. P values were FDR adjusted across all diseases and quantile cutoffs. To reduce over-fitting of a single expression cutoff per disease, associations with reduced survival were required to have FDR-adjusted P values ≤0.1 for at least two quantile cutoffs. Survival analysis statistics were calculated using the R statistical programming language (22).
Gene-set enrichment analysis (GSEA) was used to identify the enrichment of pathways with genes that are coexpressed with GSTO1. GSEAv2.2.3 was used with v6 gene sets sourced from MSigDB. 10,000 gene set permutations were performed using weighted mode scoring and Pearson metric (23). Only genes with evidence of expression in >50% of a disease patient population were considered. Five diseases were evaluated on the basis of association with reduced survivability (SKCM, GBMLGG, KIRC, STAD, UVM).
Bru-seq analysis of nascent RNA synthesis
Bru-seq analysis was performed as described previously (24). Briefly, when cells reached 80%–90% confluence, bromouridine (final concentration of 2 mmol/L) was added to the media to label newly synthesized nascent RNA for 30 minutes. Cells were collected in TRIzol (Invitrogen) and total RNA was isolated. Bru-labeled, nascent RNA was captured with anti-BrdU antibodies conjugated to magnetic beads, converted into cDNA libraries, and sequenced using an Illumina HiSeq 2000 sequencer (University of Michigan DNA Sequencing Core, Ann Arbor, MI). Sequencing reads were mapped to the hg38 reference sequence. Ensemble gene identifiers were mapped to HGNC symbols and Entrez identifiers using Gencode v27 annotations (25). Only measurements mapping to protein coding genes with entrez identifiers were considered and gene changes with absolute fold change >2 and mean RPKM >0.5 were considered significant.
RNA-seq profiling
Cells were lysed with TRIzol Reagent (Thermo Fisher Scientific) at room temperature. RNA was further purified with DirectZol kit (Zymo Research). RNA quality was assessed using the TapeStation (Agilent Technologies). Samples with RNA integrity numbers (RIN) of 8 or greater were prepared with TruSeq Stranded mRNA Library Prep (Illumina) as per the manufacturer's protocol with 1 μg of RNA and 12 cycles of PCR amplification. Libraries were checked for size on the TapeStation and quantified using the Kapa Biosystems library quantification kit (Illumina). The libraries were barcoded, pooled, and sequenced using paired-end 50 bp (U-87 MG, University of Michigan DNA Sequencing Core, Ann Arbor, MI) and single-end 50 bp (HCT116, University of Michigan DNA Sequencing Core, Ann Arbor, MI) sequencing. Reads were mapped to GRCh38 using STAR v2.5.2 (26) and gene quantifications were calculated using Cufflinks v2.2.1 (27) to quantify refGene annotations. Gene read counts calculated using featureCounts (28) v1.6.1 were used to evaluate differential expression using DESeq2 v1.18.1 (29). Protein-coding genes were considered significantly differentially expressed with a mean FPKM >0.5 and absolute fold change >1.5 and FDR-adjusted P < 0.05. All gene readouts where required to be mappable to both an HGNC and Entrez identifier to be considered for gene-set enrichment analyses. This data has been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE147626 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE147626; ref. 30).
TMT labeling and LC/MS analysis
HCT116 nt WT and GSTO1 KO cells (triplicate each) were lysed with RIPA buffer (Thermo Fisher Scientific, 89901), followed by sonication and centrifugation at 10,000 × g for 15 minutes at 4 °C. The supernatant was saved and protein concentration determined by the BCA assay (Thermo Fisher Scientific, 23225). Seventy-five microgram of protein was used to perform tandem Mass Tag (TMT) labeling with the TMT 6-plex Isobaric Labeling Kit (Thermo Fisher Scientific) essentially according to the manufacturer's protocol. Briefly, upon reduction and alkylation of cysteines, the proteins were precipitated by adding six volumes of ice-cold acetone followed by overnight incubation at −20°C. The precipitate was spun down and the pellet was allowed to air dry. The pellet was resuspended in 0.1 mol/L TEAB and overnight digestion with trypsin (1:50; enzyme:protein) at 37°C was performed with constant mixing using a thermomixer. The TMT 6-plex reagents were dissolved in 41 μL of anhydrous acetonitrile and labeling was performed by transferring the entire digest to TMT reagent vial and incubating at room temperature for 1 hour. Reaction was quenched by adding 8 μL of 5% hydroxyl amine and further 15-minute incubation. Labeled samples were mixed together, and dried using a vacufuge. An offline fractionation of the combined sample (∼200 μg) into 8 fractions was performed using high pH reverse-phase peptide fractionation kit according to the manufacturer's protocol (Pierce, 84868). Fractions were dried and reconstituted in 9 μL of 0.1% formic acid/2% acetonitrile in preparation for LC/MS-MS analysis.
To obtain superior quantitation accuracy, we employed multinotch-MS3 (McAlister GC), which minimizes the reporter ion ratio distortion resulting from fragmentation of coisolated peptides during MS analysis. Orbitrap Fusion (Thermo Fisher Scientific) and RSLC Ultimate 3000 nano-UPLC (Dionex) was used to acquire the data. Two microliters of the sample was resolved on a PepMap RSLC C18 column (75 μm i.d. × 50 cm; Thermo Fisher Scientific) at the flow-rate of 300 nL/minute using 0.1% formic acid/acetonitrile gradient system (2%–22% acetonitrile in 150 minutes; 22%–32% acetonitrile in 40 minutes; 20-minute wash at 90%, followed by 50-minute reequilibration) and directly spray onto the mass spectrometer using EasySpray source (Thermo Fisher Scientific). Mass spectrometer was set to collect one MS1 scan (Orbitrap; 120K resolution; AGC target 2 × 105; max IT 100 ms) followed by data-dependent, “Top Speed” (3 seconds) MS2 scans (collision-induced dissociation; ion trap; NCE 35; AGC 5 × 103; max IT 100 ms). For multinotch-MS3, top 10 precursors from each MS2 were fragmented by HCD followed by Orbitrap analysis (NCE 55; 60K resolution; AGC 5 × 104; max IT 120 ms, 100–500 m/z scan range).
Proteome Discoverer (v2.4, Thermo Fisher Scientific) was used for data analysis. MS2 spectra were searched against SwissProt human protein database using the following search parameters: MS1 and MS2 tolerance were set to 10 ppm and 0.6 Da, respectively; carbamidomethylation of cysteines (57.02146 Da) and TMT labeling of lysine and N-termini of peptides (229.16293 Da) were considered static modifications; oxidation of methionine (15.9949 Da); and deamidation of asparagine and glutamine (0.98401 Da) were considered variable. Identified proteins and peptides were filtered to retain only those with FDR ≤1%. Quantitation was performed using high-quality MS3 spectra (Average signal-to-noise ratio of 10 and <30% isolation interference). Protein changes with absolute fold change >1.5 and qval <0.1 were considered significant.
Bru-seq and RNA-seq gene set enrichment
Gene-set enrichment was performed using DAVID and GSEA. The RDAVIDWeb services Bioconductor package was used to identify KEGG and GO Biological Process categories that were enriched for differentially expressed genes regardless of fold change direction (31). GSEAv2.2.3 was used with v6.1 gene sets sourced from MSigDB to identify gene sets significantly up- or downregulated. 10,000 gene set permutations were performed using weighted mode scoring of genes ranked using DESeq2 shrunken log fold change APEGLM method (32). Gene sets with FDR-adjusted P values <0.05 were considered significant.
3D spheroid culture
Indicated number of cells were plated in nonadherent U-bottom 96-well plate (Corning, 4520) with 5% collagen (StemCell Technologies) in original medium (100 μL). Additional 100 μL of medium with 5% collagen was gently added on day 4. Images of spheroids were taken using 10× objective on an Olympus IX83 motorized inverted microscope with cellSens Dimension software (Olympus Corporation). For viability assay, 100 μL CellTiter-Glo 3D (Promega, G9681) were added to each well. Luminescence was measured on Synergy H1Hybrid Multi-Mode Reader (BioTek).
Sphere formation assay
Cells were dissociated and seeded in ultralow attached plates (Corning, 3474 or 3473) and cultured in serum-free medium (X-VIVO 20, Lonza, 04-448Q) for 10 days. Floating spheres in three fields per each sample were counted and images of spheroids were taken using 10× objective on an Olympus IX83 motorized inverted microscope with cellSens Dimension software (Olympus Corporation).
Western blot analysis
Western blotting was performed to test GSTO1 expression of knockout cells and other related biomarkers. Cells were lysed in RIPA buffer (Sigma-Aldrich, R0278) containing protease and phosphatase inhibitors (Thermo Fisher Scientific, 78442). Thirty microgram extracts from each sample was electrophoresed on Bis-Tris gel and transferred to a PVDF membrane (Bio-Rad, 1704274). Primary antibodies (Anti-GSTO1 antibody, Gene Tex, GTX105655, 1:2,000; anti-TF antibody, Santa Cruz Biotechnology, sc-393657, 1:500; Anti-GAPDH antibody, Cell Signaling Technology, 2118S, 1:1,000) were applied overnight at 4°C in 5% BSA (Calbiochem, 2930) followed by secondary antibody. The membranes were imaged by the Odyssey Imaging Systems (LI-COR Biosciences).
Results
GSTO1 is upregulated in select cancers and correlates with decreased patient survival
An initial pan-cancer patient survey of GSTO1 expression analysis was performed using RNA-seq, survival, and disease progression read-outs from 9,718 patients across 34 tumor types. RNA-seq data from each patient was z-score normalized and we found that GSTO1 expression was well above average (z-score = 0) across all 34 tumor types (Fig. 1A). We identified five tumor types for which high GSTO1 expression was associated with reduced survival (Fig. 1B): glioblastoma and lower grade glioma (GBMLGG), kidney renal clear cell carcinoma (KIRC), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), and uveal melanoma (UVM). This association of GSTO1 expression and poor survival was also observed across other non-TCGA glioma cohorts (Supplementary Fig. S1A). In addition, we observed a clear increase in GSTO1 expression with cancer disease progression. Glioma patients with more aggressive disease (GBM) expressed significantly more GSTO1 than patients with low-grade glioma (LGG; Supplementary Fig. S1B). GSTO1 expression also increased with KIRC tumor stage and grade (Supplementary Fig. S1C). These results show that increased expression of GSTO1 is associated with poor patient survival and disease progression in multiple cancers, suggesting that GSTO1 plays an important role in tumor progression.
GSTO1 depletion suppresses in vivo tumor growth
To investigate the function of GSTO1 in cancer, we used CRISPR/Cas9 to knock out GSTO1 in three representative cell lines (A172, HCT116, U87-MG) with high expression of GSTO1 (Fig. 2A). Deletion of GSTO1 was verified at both the protein (Fig. 2B; Supplementary Fig. S2A) and mRNA level (Supplementary Fig. S2B). Sanger sequencing further validated that the genome editing was caused by deletions and frameshifts in exon 2 of the GSTO1 gene (Supplementary Fig. S2C, one clone for each cell line).
Knockout of GSTO1 impairs tumor growth in vivo. A, GSTO1 mRNA expression in Cancer Cell Line Encyclopedia cell lines among 20 types of cancers. Dashed line shows the average expression of GSTO1 across all cell types. HCT116, U-87 MG, and A172 all showed above average GSTO1 expression. B, GSTO1 protein expression in single clones of cell lines HCT116, U-87 MG, and A172 transfected with nontargeting (nt WT) or GSTO1 targeting crRNA/tracrRNA (GSTO1 KO). C, Tumor growth of HCT116 nt WT and GSTO1 KO cells implanted subcutaneously in ICRSC mice. Statistical difference was determined by two-way ANOVA. Changes in body weight of mice are shown in the bottom panel. D, Tumor growth of U-87 MG nt WT and GSTO1 KO cells implanted subcutaneously in ICRSC mice. nt WT tumor group was terminated when the tumors reached 2,000 mm3. GSTO1 KO group was terminated 10 weeks after tumor implantation. Changes in body weight of mice are shown in the bottom panel. E, Kaplan–Meier curve of an independent xenograft study with HCT116 nt WT and GSTO1 KO cells implanted subcutaneously in NOD/SCID mice. F, Tumor weights of mice with nt WT and GSTO1 KO cell implantation after necropsy. Statistical difference was determined by Student t test. ***, P < 0.001.
To directly explore the role of GSTO1 in tumorigenesis, we tested the tumor growth of GSTO1 KO cells and their nontargeted wild-type counterparts (nt WT) in xenograft models in ICRSC or NOD/SCID mice (n = 5). Strikingly, GSTO1 KO HCT116 cells showed remarkably reduced tumor burden as compared with nt WT cells (Fig. 2C and F). Furthermore, in an independent experiment, the HCT116 GSTO1 KO group had significantly prolonged survival (log-rank P = 0.032) for up to four weeks as compared with the nt WT group (Fig. 2E). Loss of GSTO1 in U-87 MG cells inhibited tumor formation (Fig. 2D and F), and no tumors were detected even after reinjection of U-87 MG GSTO1 KO cells and extension of tumor monitoring for an additional four weeks. Despite numerous attempts to implant A172 cells in vivo, multiple strategies failed to generate tumor in flanks of mice.
In conclusion, these results demonstrate that the loss of GSTO1 significantly suppresses tumor growth in vivo for both HCT116 and U-87 cells, suggesting that GSTO1 is an impactful therapeutic target in cancer.
Deletion of GSTO1 hampers U-87 MG cell proliferation
To understand underlying mechanism of the growth inhibition in vivo, we first determined whether CRISPR/Cas9-mediated loss of GSTO1 affected cell proliferation. We measured the doubling time of nt WT and three single clones in all three GSTO1 KO cell lines (Fig. 3A) to avoid characterization of single clone behaviors. Interestingly, depletion of GSTO1 in U-87 MG cells slowed cell growth, but had no significant effect on the growth of HCT116 and A172 cells. Consistent with this finding, loss of GSTO1 in U-87 MG cells significantly decreased the numbers and sizes of cell colonies, but had little effect on HCT116 and A172 cells (Fig. 3B). It should be noted that the three HCT116 GSTO1 KO clones grew at different rates, possibly due to variations in single clone behavior.
Deletion of GSTO1 reduces U-87 MG cell proliferation, impedes the growth of 3D tumor spheroids and neurosphere formation of U-87 MG and HCT116 cells. A, Doubling time of HCT116, U-87 MG, and A172 nt WT and GSTO1 KO single clones. Same number of nt WT and GSTO1 KO cells were seeded. After 24, 48, 72, and 96 hours, cells were harvested and counted manually. B, Representative images of colony-forming abilities of HCT116, U-87 MG, and A172 nt WT and GSTO1 KO cell lines. C, Representative images of HCT116, A172, and U-87 MG nt WT and GSTO1 KO tumor spheroid growth. D and E, CellTiter-Glo 3D cell viability assay was used for quantification seven days after seeding. F, Representative images of HCT116, A172, and U-87 MG nt WT and GSTO1 KO spheroids. Spheroids were grown in ultralow attachment wells in medium without FBS. G, Diameters were calculated with at least 20 spheroids using ImageJ. Statistical difference was determined by Student t test. ***, P < 0.001.
In summary, our results demonstrate that deletion of GSTO1 significantly reduced U-87 MG in vitro cell proliferation, indicating that GSTO1 is essential for in vitro cell growth in certain cell types. Interestingly, GSTO1 KO affected HCT116 tumor growth in vivo but not in vitro 2D culture.
GSTO1 KO impedes the growth of 3D tumor spheroids
Because of the difference in access to nutrients, oxygen, and multi-cellular 3D contacts, cells grown in 3D culture are of phenotypic heterogeneity and in that sense resemble in vivo tumors. In particular, 3D culturing incorporates some aspects of the tumor microenvironment, including cell adhesion and cross-talk with extracellular matrix (ECM; ref. 33).
As described above, HCT116 GSTO1 KO cells showed significant tumor growth delay while no difference in growth rate was observed when grown in 2D culture, suggesting that deletion of GSTO1 may affect the tumor microenvironment. To test this hypothesis, we generated 3D tumor spheroid models for each cell line (Fig. 3C; Supplementary Fig. S3A and S3B for other single clones). Interestingly, we found that HCT116 KO cells showed significant retardation of cell proliferation in 3D culture despite no observed difference in growth rate in 2D culture. The growth rate is measured by the diameter of spheroids and was further assessed by CellTiter-Glo 3D cell viability assay, which measures the presence of metabolically active cells (Fig. 3D and E). We also observed that spheroids of HCT116 nt WT cells were more invasive than those of KO cells. U-87 MG GSTO1 KO 3D tumor spheroids grew slower than those of the nt WT, which was also reflected in 2D culture. A172 cells formed a monolayer and failed to form proper spheroid structures over collagen (Fig. 3D and E). Although GSTO1 did not affect HCT116 cell growth in 2D culture, it significantly inhibited 3D cell growth, further supporting the hypothesis that GSTO1 influences vital pathways involved in cell adhesion and ECM cross-talk.
Depletion of GSTO1 inhibits neurosphere formation in U-87 MG
In agreement with the observed slow growth rate of U-87 MG GSTO1 KO cells in monolayer and 3D cultures, the growth of these cells as xenografts in vivo was also blocked. One hypothesis for this behavior is that GSTO1 affects cancer stem cell survival. Studies have shown that HIF-dependent induction of GSTO1 expression promotes chemotherapy-induced breast cancer stem cell enrichment (11). Therefore, we next sought to test whether GSTO1 depletion affects the survival of cancer stem cells (CSC). Growing cells as neurospheres enriches for CSCs or cells with progenitor characteristics (34). Interestingly, deletion of GSTO1 in U-87 MG cells inhibited neurosphere formation (Fig. 3F and G). This phenotype was consistent for all U-87 MG GSTO1 clones tested (Supplementary Fig. S4). We further assessed the neurosphere formation in A172 cells and found that deletion of GSTO1 did not impact the size or number of neurospheres. HCT116 GSTO1 KO cells display small tumor spheroids as compared with the nt WT (P < 0.05, Fig. 3F and G). In conclusion, GSTO1 expression is required for spheroid formation in two of the three cell lines examined.
Integrated transcriptional profiles across cell lines reveals GSTO1 dysregulates critical pathways in cancer
To obtain a comprehensive molecular understanding of GSTO1 function in cancer, we performed in-depth nascent RNA Bru-seq on nontargeted wild-type parental cells and GSTO1 KO U-87 MG (clone 1, n = 1), HCT116 (clone 1, n = 1), and A172 (clone 1, n = 1) cells (24). Depletion of GSTO1 caused a strong perturbation of transcription in U-87 MG with less effect observed in HCT116 and A172 cells (Fig. 4A, top 30 up- or downregulated genes in Supplementary Tables S1–S6). A number of genes were similarly transcriptionally affected by GSTO1 loss in the HCT116 and U-87 MG cell lines (upregulated: P = 9.37e-15, downregulated: P = 0.00205; Fig. 4B, commonly up- or downregulated gene sets in Supplementary Table S7). However, there was little overlap between the changes in these two cell lines and the A172 cell line (Supplementary Fig. S5A). Thus, we focused our subsequent analyses on HCT116 and U-87 MG cells. We used gene set enrichment analysis (GSEA) to identify gene sets significantly enriched for genes changing in response to loss of GSTO1 (Supplementary Fig. 5B; Supplementary Tables S8–S13). DNA repair, p53 pathway, and negative regulation of protein-binding gene sets were significantly upregulated in both HCT116 and U-87 MG cells after loss of GSTO1 (Fig. 4C; Supplementary Fig. S6A). In U-87 MG cells, loss of GSTO1 resulted in the upregulation of UV response gene sets (Supplementary Fig. S6B) along with DNA repair gene sets. In addition, genes involved in epithelial–mesenchymal transition (EMT) were downregulated in response to loss of GSTO1 in U-87 MG cells (Supplementary Fig. S6C). In HCT116 cells, deletion of GSTO1 caused significant upregulation of gene sets related to IFN production, IFN responses, and IFN-related signaling pathways (Supplementary Fig. S6D). Gene sets related to tight junction and apical junction were also upregulated, while Myc-target gene sets were downregulated in response to GSTO1 KO in HCT116 (Supplementary Fig. S6E).
Integrated transcriptional profiles across cell lines reveal that GSTO1 dysregulates critical pathways in cancer. A, Number of differentially expressed genes upon GSTO1 KO in Bru-seq transcription profile in U-87 MG, HCT116, and A172 cells. B, Significantly changed genes overlapped in nascent (Bru-seq) transcriptomic profiles of HCT116, U-87 MG, and A172 cells in response to GSTO1 KO. C, Common significantly changed gene sets (GSEA) overlapped in nascent (Bru-seq) transcriptomic profiles between HCT116 and U-87 MG cells in response to GSTO1 KO. D, Volcano plots of RNA-seq genes differentially expressed upon GSTO1 KO in U-87 MG and HCT116 cells. Top 30 differentially expressed genes are highlighted. E, Significantly changed genes overlapped in RNA-seq between HCT116 and U-87 MG cells in response to GSTO1 KO. F, Common significantly changed gene sets (GSEA) overlapped in transcriptomic profiles (RNA-seq) between HCT116 and U-87 MG cells in response to GSTO1 KO. G, Gene sets commonly enriched with upregulated GSTO1 expression across GBMLGG, KIRC, STAD, UVM, and SKCM. Shared gene sets are highlighted along with the disease type. Differential expression thresholds are described in Materials and Methods. Gene sets with FDR < 0.1 were considered as significant. H, Gene sets enriched for up- and downregulated genes using DAVID in HCT116 and U-87 MG RNA-seq. The enrichment of gene sets involved in the tumor microenvironment, EMT, proliferation, and differentiation and immune responses are highlighted.
We also performed RNA-seq on U-87 MG (n = 3, clone 1) and HCT116 (n = 3, clone 1) GSTO1 KO cells (Fig. 4D, top 30 dysregulated genes in Supplementary Tables S14–S17). Similar to the nascent RNA transcription profile, U-87 MG had a larger number of differentially expressed genes as compared with HCT116. The two cell lines shared 77 downregulated genes and 191 upregulated genes (Fig. 4E; P = 1.06e-59 and P = 3.42e-14, respectively). Significantly downregulated gene sets shared by HCT116 and U-87 MG included E2F targets, MYC targets, and G2–M checkpoint (Fig. 4F; Supplementary Tables S18 and S19). Although G2–M checkpoint genes were upregulated in both cell lines, a difference in growth rate was only observed in U-87 MG cells. One potential explanation is that additional compensatory pathways were inhibited or activated to maintain cell growth regardless of G2–M checkpoint activation in HCT116 cells. Interestingly, compared with 11 upregulated gene sets in U-87 MG, we observed 182 uniquely upregulated gene sets in HCT116 including Notch and VEGF signaling pathway, which may provide insight to these compensatory effects. Similar to Bru-seq findings for the HCT116 cells, the upregulated gene sets following GSTO1 loss included complement and coagulation cascades, detection of biotic stimulus, and positive regulation of IFNα production (Fig. 4F; Supplementary Tables S20 and S21), suggesting that GSTO1 deletion augments the transcription of innate immune system–related pathways.
We further analyzed gene sets coexpressed with GSTO1 in patient samples (Fig. 4G; Supplementary Tables S22-S26). Thirty-seven gene sets were commonly enriched in GBMLGG, KIRC, SKCM, STAD, and UVM, including Parkinson diseases, Huntington disease, Alzheimer disease, antigen processing and presentation, and reactive oxygen species. As mentioned in the introduction, GSTO1 has been shown to be involved in the reactive oxygen species pathway; this finding, while not novel, shows that GSTO1 coexpression in TCGA patient cohorts is able to recapitulate known canonical biological function of GSTO1. In addition, several reports indicate that GSTO1 plays important roles in neurodegenerative diseases (35, 36) and inflammation (17, 19). This validates our coexpression analysis, and highlights the novelty of our findings. Similar to our Bru-seq and RNA-seq results, gene sets involved in DNA repair, UV response, and Myc target gene sets were also enriched in multiple disease types. IFNα response gene set was enriched in four disease types including GBMLGG, SKCM, STAD, and UVM. These findings using GSTO1 coexpression in patient gene expression across multiple disease cohorts further support the clinical relevance of our in vivo and in vitro GSTO1 KO results and further shed light on the molecular pathways modulated by GSTO1.
Annotation of transcriptional profiles reveals the role of GSTO1 in tumor microenvironment, epithelial-to-mesenchymal transition, proliferation, and cancer immune responses
A feedback effect at the transcriptional or translational level is common when cells lose the function of a gene. Numerous studies have highlighted that compensatory mechanisms induced by genetic mutation events contributed largely to the phenotypes (37). To further study the transcriptional changes induced upon GSTO1 loss, we used DAVID to perform an over-representation gene set enrichment analysis by combining the significantly upregulated and downregulated genes (FC > 1.5, FDR q-value <0.1; Fig. 4H; Supplementary Tables S27 and S28). HCT116 and U-87 MG cell lines shared 24 enriched gene sets (Supplementary Fig. S7). Deletion of GSTO1 had a significant impact on gene sets related to angiogenesis (GO:0001525, GO:114814, and GO:1901342), cell adhesion (GO:0022409), and ECM interaction (GO:0007160 and hsa04512), further supporting the role of GSTO1 in modulating the tumor microenvironment. Several gene sets involved in cell migration are significantly enriched (GO:0010631, GO:0043542, GO:0090132, and GO:0051272). Gene sets involved in cell proliferation (GO:0050678 and GO:0050673), differentiation (GO:0042692), and apoptosis (GO:2001236) are also enriched for genes changing in response to GSTO1 KO. Interestingly, we also found 13 gene sets related to immune response, which is consistent with previous studies showing GSTO1's relationship with inflammation and immune response (17, 19, 20). Deletion of GSTO1 in tumor cells significantly enriched the transcription of gene sets in cytokine production, signaling pathways, and cellular responses, including type I IFN (GO:0035455, GO:0034340, and GO:0060337) and TNF (GO:0032640, GO:0032680, and GO:1903555). T-cell activation gene sets (GO:0042110) were also enriched in both GSTO1 cell lines, suggesting that GSTO1 may be involved in the activation of T cells.
These results highlight the multifaceted nature of GSTO1 function in tumor progression through modulation of tumor microenvironment, cell proliferation and differentiation, EMT and immune response.
In-depth multiomics analysis in HCT116 cells reinforces impact of GSTO1 in cell adhesion and IFN responses
Nascent mRNA expression, mRNA expression, and steady-state protein levels may not always correlate due to posttranscriptional, translational, and protein degradation regulation (38). Thus, a joint analysis of the transcriptomic and proteomic profiles provides more detailed insights that may not be deciphered from individual analysis. In addition to transcriptome profiling, we performed mass spectrometry–based proteomic profiling to provide further insight into the regulatory programs impacted by GSTO1 KO in HCT116 cells (Fig. 5A, top 30 significantly changed proteins in Supplementary Tables S29 and S30). Differential expression was used to quantify significant changes across all three platforms (Materials and Methods) and used to identify four genes and proteins that were commonly downregulated and 12 genes and proteins that were commonly upregulated in the datasets generated from Bru-seq, RNA-seq, and mass spectrometry proteomic profiling (Fig. 5B). SLC2A1, which was downregulated across all three platforms, encodes for glucose transporter type 1 (GLUT1), an important glucose transporter in the glycolytic pathway. GLUT1 has been associated with tumor progression and poor prognosis in many human solid tumors (39). GLUT1 gene transcription is promoted by c-Myc (40) while suppressed by wild-type p53 (41). Our observation that GSTO1 deletion downregulates MYC target genes and the DNA repair pathway is consistent with such regulation. We found that the IFN-stimulated gene IFIT1 was transcriptionally upregulated following loss of GSTO1. Interestingly, IFIT1 expression is commonly downregulated in patients with cancer (42) and its higher expression correlated with improved clinical outcome (43). Thus, targeting of GSTO1 may improve clinical outcome by causing upregulation of IFIT1.
Multiomics analysis of HCT116 in response to GSTO1 deletion leads to the enrichment of cell adhesion, IFNα, and IFNγ response gene sets. A, Number of differentially expressed genes as a result of GSTO1 KO across the three platforms. Differential expression thresholds used per platform are described in Materials and Methods. B, Sixteen genes were commonly up- (4) or down- (12) regulated across all three profiling platforms. C, Significantly changed gene sets (GSEA) overlapped among the three omics platforms. Upregulated gene sets include cell adhesion molecules, IFNα response, IFNγ response, and viral myocarditis. Gene sets with FDR < 0.1 were considered significant. D, GSEA plots for gene sets enriched in multiomics. NES, normalized enrichment score. E and F, Heatmap showing log2-fold changes of 107 expressed genes belonging to the cell adhesion molecules gene set and IFNα response gene set. Differentially expressed genes have black border.
We further conducted gene set enrichment analysis (GSEA) to identify GSTO1 direction–specific regulatory effects across all three platforms (Supplementary Tables S31 and S32). Consistent with the DAVID enrichment results, cell adhesion molecules, IFNα response, and IFNγ response gene sets were upregulated (Fig. 5C and D). Myc target gene sets were downregulated in RNA-seq and Bru-seq datasets, but not at the protein level. Limited by detection sensitivity of mass spectrometry, proteomics results covered a smaller proportion of corresponding genes, which complicated a direct comparison between the different platforms. Despite differences in coverage, high consistency among the three platforms were observed in cell adhesion molecules gene set covering 107 differentially expressed genes (Fig. 5E). Strikingly, the upregulation of IFNα response was concordant not only among the three platforms in HCT116 cells, but also in U-87 MG (RNA-seq/Bru-seq) and in A172 (Bru-seq; Fig. 5F). This strong agreement between platforms and cell lines reinforces the relevance of GSTO1 to cell adhesion and IFN response as potential mechanism by which GSTO1 contributes to tumor progression.
Tissue factor expression is regulated by GSTO1 and shows significant correlation in patient samples
Nascent RNA (Bru-seq), steady-state RNA (RNA-seq), and protein profiling (MS) of control and GSTO1 KO HCT116 and U-87 MG cells revealed that Tissue factor gene (F3) was significantly downregulated as compared with the corresponding Cas9 nontargeted control cells across all platforms in both cell lines (Figs. 5B and 6A). Consistent with F3 downregulation in GSTO1 KO cell lines, nascent transcription of F3 was also downregulated in siGSTO1-treated and GSTO1 inhibitor C1-27–treated HCT116 (Fig. 6B). We further confirmed the downregulation of tissue factor protein (TF) following the loss of GSTO1 in cells and tumor samples using Western blot analysis (Fig. 6C). These results suggest that TF transcription is regulated by GSTO1.
Deletion of GSTO1 suppresses F3 transcription and TF expression. A, Venn diagram illustrating commonly downregulated genes among Bru-seq, RNA-seq, and proteomic in HCT116, and Bru-seq, RNA-seq in U-87 MG. F3 is shared by five platforms and cell lines. B, Bru-Seq gene trace of F3 after GSTO1 inhibitor C1-27 treatment or siGSTO1 treatment in HCT116 cells. C, Expression of TF and GSTO1 in HCT116 nt WT and GSTO1 KO cell lines and in tumors harvested from xenograft mice model. D and E, Correlation between F3 and GSTO1 in colorectal cancer (Marra colorectal cancer database; D) and glioblastoma stem cells (Pollard glioblastoma stemcells database; E) from the R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl).
TF is a crucial mediator of tumor cell proliferation, metastasis, and the tumor microenvironment (44). High expression of F3 in patient samples significantly correlates with patient survival, suggesting that F3 is essential for tumor progression. We further explored the correlation of GSTO1 and F3 expression in patients. Strikingly, in colorectal cancer (n = 64), expression of GSTO1 was significantly correlated to F3 expression (Fig. 6D), further validating the molecular link between GSTO1 and F3. Interestingly, we observed that in glioblastoma stem cells (n = 20), F3 RNA expression also showed remarkable correlation with GSTO1 expression (Fig. 6E). In addition, tissue factor promotes cancer stem cell activity and serves as a therapeutic oncotarget for the eradication of cancer stem cells (45).
Taken together, we identified that TF was suppressed upon GSTO1 knockdown and its expression correlated with GSTO1 expression in multiple cancer datasets. These results suggest that GSTO1 is important in regulating F3 transcription, further supporting an important role of GSTO1 in tumor progression.
Platinum drugs, PARP inhibitors, and glutaminase inhibitor CB-839 show synthetic lethality with the loss of GSTO1
An attractive strategy to overcome drug resistance is to target enzymes that exhibit synthetic lethality. To maximize the efficacy of GSTO1 inhibitors, we aimed to identify compounds that exhibit selective cytotoxicity to GSTO1 KO cells, potentially targeting the compensatory pathways caused by the loss of functional GSTO1. We used HCT116 GSTO1 KO cells as our screening platform and further tested synergy using GSTO1 inhibitor C1-27 (Fig. 7A). Elevated GSTO1 in cancer cells correlates to drug resistance (8, 46), and GSTO1 inhibitors sensitizes cells toward cisplatin and other platinum drugs (5, 10). We first tested the hypothesis that platinum drugs would be synthetically lethal with the loss of GSTO1 by treating GSTO1 nt WT and KO cell lines with cisplatin, carboplatin, and oxaliplatin (Fig. 7B; Supplementary Fig. S8). Results showed that these platinum drugs all showed stronger cytotoxicity in HCT116 GSTO1 KO cells as compared with the nt WT, suggesting that the platinum drugs synergized with the loss of GSTO1. Nascent transcriptome signatures in response to GSTO1 KO show upregulation of DNA repair, p53 pathway, and G2–M checkpoint gene sets. On the basis of this, we tested 38 clinical or preclinical drugs including DNA damage inducers, PARP inhibitors, p53 inhibitors, and CDK inhibitors. Out of the 38 agents, 13 were significantly more potent in cells depleted of GSTO1, including glutaminase inhibitor CB-839, mitomycin C, and PARP inhibitors (Fig. 7B; Supplementary Fig. S9). We next combined these drugs with C1-27, and found that mitomycin C, CB-839, and olaparib all showed synthetic lethality with C1-27 (Fig. 7C). Our results demonstrate that GSTO1 inhibitors may be used in combination with standard-of-care drugs to synergistically kill cancer cells and highlight the strong potential of GSTO1 inhibitors as a therapeutic treatment for cancer.
PARP inhibitors and glutaminase inhibitor CB-839 are synthetically lethal with the loss of GSTO1. A, Workflow for the lead identification in the HCT116 GSTO1 KO cell line. B, Thirty-eight clinical and preclinical drugs were tested on HCT116 nt WT and GSTO1 KO cell lines. IC50 values in each cell line were calculated by the intensity of crystal violet staining in colony formation assay wells (ImageJ). Differential index is illustrated by log2 (IC50 of HCT116 nt WT/IC50 of GSTO1 KO) and determined by three independent experiments. C, HCT116 cells were treated with indicated concentrations of CB-839, cisplatin, carboplatin, or olaparib, in combination with C1-27 for 7 days, and stained with crystal violet. The intensity of each well was calculated using ImageJ. The synergistic effect of C1-27 with CB-839 is illustrated with HSA model.
Discussion
Although GSTO1 is overexpressed in various cancers, its function in cancer is not fully understood and its relationship to cancer signaling pathways is largely unexplored. In this study, we extensively characterized the function of GSTO1 in diverse genetic backgrounds. Our data demonstrate that deletion of GSTO1 significantly inhibits tumor growth in vitro and in vivo. Multiomics (Bru-seq, RNA-seq, and MS proteomics) analysis showed that loss of GSTO1, promoted the downregulation of gene sets pertaining to tumor progression, microenvironment, metastasis, and immune response. Thus, the therapeutic targeting of GSTO1 can block tumor growth.
Targeting the tumor microenvironment is an important strategy to overcome acquired resistance and metastasis. It is widely accepted that in 3D cultures, adhesive interactions with the surrounding ECM and neighboring cells define cell shape and organization, and influence fundamental cellular behaviors. Phenotypically, GSTO1 KO cells showed significant growth delay in 3D spheroid and tumor growth in vivo as compared with the 2D monolayer culture, suggesting that deletion of GSTO1 causes a tumor-suppressing microenvironment. Consistent with that, pathways essential for tumor microenvironment, such as angiogenesis and ECM–receptor interactions, were also dysregulated in response to GSTO1 KO. A recent study also showed that GSTO1 inhibition resulted in decreased migration and invasion in sorted breast cancer stem cells (47). In conclusion, our results demonstrate that GSTO1 inhibit 3D cell growth, influence gene sets involved in angiogenesis, cell adhesion, and interaction with the surrounding ECM, and play an important role in modifying tumor microenvironment.
As a proinflammatory protein, GSTO1 governs the release of proinflammatory cytokines such as IL6, IL1β, and TNFα (20). GSTO1 is also a signature gene in both human and mouse IL4-treated macrophages (48). By deglutathionylating NIMA related kinase 7 (NEK7), GSTO1 promotes activation of NLRP3, which is important for IL1β and IL18 maturation (19). There is a broad cross-talk between GSTO1 and diverse cytokines; however, the function of GSTO1 in tumor immune responses is still unclear. In this work, we report for the first time that interferon alpha and gamma responses are induced in response to GSTO1 KO. IFN have important roles in protecting the host against cancer through effects in tumor cell–intrinsic property and in immunoregulatory functions. IFNs have an intrinsic impact on tumor cells by inhibiting proliferation, and modulating apoptosis, differentiation, migration, and cell surface antigen expression (49). In addition, IFNs affect different biological processes involved in tumor progression, such as angiogenesis and immunity (49, 50). Importantly, IFNs play major roles in activating anticancer immunity and inhibiting the activity of immune-suppressive cells and the conversion of tumor-associated macrophages (49). Through the perturbation of IFN responses, GSTO1 might serve as a promising therapeutic target to influence biological pathways in cancer, affect the tumor environment, and alter tumor-infiltrating immune cell properties. Results from our study suggest that deletion of GSTO1 promotes downstream responses of IFN in tumors, which may affect tumor progression and elicit an immune response, but the mechanisms behind this need to be further explored.
The transcription of G2–M checkpoint gene sets, MYC targets gene sets, and E2F targets gene sets were downregulated following GSTO1 depletion in all three GSTO1 KO cell lines. Upregulation of MYC and E2F transcriptional targets are known for promoting cell growth and proliferation (51, 52). In addition, gene-set enrichment analysis using DAVID demonstrated that GSTO1 KO caused transcriptional changes of genes regulating cell proliferation, differentiation, and apoptotic signaling pathways. Consistent with our findings, studies have shown that several critical proteins involved in cell proliferation and metastasis are deglutathionylated by GSTO1, which might influence their biological function (16, 17, 53). Collectively, our findings demonstrate that GSTO1 deletion inhibits tumor growth, downregulates gene sets involved in the cell cycle, and impacts MYC and E2F pathways in cancer.
We also reported that F3 transcription is downregulated in response to GSTO1 loss or inhibition. Associated with poor patient survival and promotion of tumor progression in multiple types of cancers, TF is also a known risk factor of metastasis (44). Previous studies have revealed that TF is associated with sustaining proliferation signaling, preventing cell death, maintaining cancer stem cell behavior, and promoting EMT (54). Decreased TF expression upon GSTO1 deletion may reduce invasion, impair formation of neuro-spheroids, and more importantly, prevent tumor-promoting processes in the tumor microenvironment in vivo. F3 transcription is modulated by several transcription factors. Early growth response gene-1 (EGR-1) regulates TF expression under hypoxic conditions in glioblastoma (55). In the human endometrium, TF expression could be regulated through Sp1 and Sp3 binding on its promoter (56). TF expression is also activated by the Stat3 transcription factor (57). Importantly, F3 is under the regulation of IL1β (58), and we demonstrate that secretion of IL1β is decreased upon GSTO1 inhibition (59), suggesting that IL1β is a potential regulating link between GSTO1 and F3 transcription. Besides that, ILlα and TNFα also induce tissue factor in endothelium (60). Previous studies indicated that GSTO1 KO mice produced significantly lower levels of TNFα, suggesting that TNFα may also play a role in regulating F3 transcription. In addition, F3 expression is upregulated downstream of mutant K-Ras oncogene signaling and inactivation of p53 (61). We also observed that p53 pathway is significantly upregulated in both HCT116 and U-87 MG cells after loss of GSTO1, which might serve as another mechanism in suppressing F3 transcription. In conclusion, GSTO1 might affect the transcription of F3 through multiple mechanisms.
In colorectal biopsies, the increased γH2AX expression is correlated to GSTO1 nuclear translocation during disease progression, suggesting that GSTO1 plays a role in genome instability (6). Here, we report for the first time that GSTO1 deletion up-regulates DNA repair gene sets along with UV response gene sets. Importantly, PARP inhibitors are more toxic in GSTO1 KO cells than in GSTO1-expressing cells and show significant synergy with the GSTO1 inhibitor C1-27. Because of the accumulation of unrepaired single-strand breaks, PARP inhibition leads to stalling of replication forks, resulting in genomic instability and cell death (62). Similarly, we found that DNA alkylating agent mitomycin-C, showed stronger potency in GSTO1 KO cells. Although temozolomide did not show differential activity in U-87 MG and A172 GSTO1 nt WT and KO cell line (Supplementary Fig. S10), U-87 MG cells were more sensitive toward radiation after GSTO1 deletion (Supplementary Fig. S11). These results suggest that despite the upregulation of DNA repair genes in response to GSTO1 deletion, GSTO1 KO cells are more vulnerable to DNA damage. It is well accepted that chemotherapeutics such as cisplatin and mitomycin-C increase ROS levels, which contributes to their genotoxicity (63). Increased oxidative stress is also reported to augment the antitumor effect of PARP inhibition (64). Interestingly, GSTO1 knockdown and inhibition increases accumulation of cellular ROS (13), and GSTO1 protects Drosophila to ROS damage (65). The effect of GSTO1 on ROS might be the reason for the transcriptional upregulation of DNA repair or UV response gene sets and sensitization of GSTO1 KO cells toward DNA-damaging agents. Although further studies are needed to elucidate the specific mechanism of mediating DNA damage and repair, our study for the first time, sheds light on the idea that GSTO1 deletion show synthetic lethality with select DNA-damaging drugs.
In this work, we employed CRISPR/Cas9 technologies coupling with extensive bioinformatics studies to validate GSTO1 as an impactful target in oncology. RNA transcription of all GSTO1 isoforms is downregulated and protein production is attenuated due to the CRISPR/CAS9 genome editing (Supplementary Fig. S12). To address the concern that CRISPR/Cas9 might produce off-target effect, we compared genetic results across multiple platforms with those of pharmacologic inhibition of GSTO1. Mitomycin C, CB-839, and olaparib show differential activity in GSTO1 KO versus ntWT cells and are also synergistic with C1-27. In addition to that, gene enrichment analysis in GSTO1 KO, siGSTO1-treated and C1-27- -treated HCT116 cells reveals high consistency in transcriptional profiles including MYC and IFN pathways (Supplementary Fig. S13). We further compared the phenotypes of GSTO1 knockdown and GSTO1 KO in terms of cell growth (Supplementary Fig. S14) and proliferation upon glutamine starvation (Supplementary Fig. S15), which showed strong agreement between the two gene silencing methods. Thus, the phenotypes in GSTO1 KO cell lines are due to on target effect.
In conclusion, using an integrated genomics and proteomics platform we have uncovered GSTO1's involvement in multiple biological functions and pathways involved in cancer progression. We confirmed that GSTO1 expression is associated with reduced survival and tumor progression in multiple TCGA cohorts. Our study provides the first evidence that GSTO1 interferes with type I IFN response in cancer cells. We discovered F3 transcription is downregulated in response to GSTO1 KO and further implicating GSTO1 in tumor microenvironment. Clinical implication of this study is that PARP inhibitors, mitomycin-C, and glutaminase inhibitor CB-839 have significant synergy with the loss or inhibition of GSTO1. Together, these findings demonstrate the essential role of GSTO1 in tumor progression, shed light on its molecular signaling pathways, and further provide new insights on its potential use as a therapeutic target.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Y. Xu: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. A. Bankhead: Formal analysis, validation, investigation, visualization, methodology, writing-review and editing. X. Tian: Formal analysis, validation, investigation. J. Tang: Investigation, methodology. M. Ljungman: Data curation, formal analysis, investigation, writing-review and editing. N. Neamati: Conceptualization, resources, data curation, supervision, funding acquisition, project administration, writing-review and editing.
Acknowledgments
This work was supported by a grant from the NIH (CA193690). We thank Phillip L. Palmbos and Yin Wang for their expertise on CRISPR/Cas9 genome editing. We thank Michelle Paulsen and Karan Bedi for conducting and helping with Bru-seq and data analysis. We thank Dr. Venkatesha Basrur (Mass Spectrometry-Based Proteomics Resource Facility, Department of Pathology, University of Michigan, Ann Arbor, MI) for running the proteomics experiments. We thank Andrea Shergalis, Christine Cuthbertson, Ding Xue, and Maha Zidan for reading early versions of this manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Cancer Res 2020;80:3692–705
- Received February 15, 2020.
- Revision received April 16, 2020.
- Accepted June 17, 2020.
- Published first June 22, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 80, Issue 17
