E2F transcription factor-1 modulates expression of glutamine metabolic genes in mouse embryonic fibroblasts and uterine sarcoma cells

Metabolic reprogramming is considered as a hallmark of cancer and is clinically exploited as a novel target for therapy. The E2F transcription factor-1 (E2F1) regulates various cellular processes, including proliferative and metabolic pathways


Introduction
Cancer cells reprogram cellular metabolism to satisfy the demands of unrestrained growth and proliferation.Thus, metabolic reprogramming is considered as a hallmark of cancer and is clinically exploited for development of effective cancer therapeutics.This process enables cancer cells to adapt to intrinsic or extrinsic signals from the microenvironment by increasing the plasticity and flexibility in nutrient uptake and utilization.Rewiring of cancer metabolism involves several pathways such as increased aerobic glycolysis, imbalanced lipid synthesis, a shift toward pentose phosphate pathway, or elevated glutaminolysis [1,2].
Accumulating evidence supports the notion that glutamine metabolism is crucial in oncogenesis.Glutamine plays a versatile role in major biosynthetic pathways including tricarboxylic acid (TCA) cycle anaplerosis, nitrogen donation for nucleotide and hexosamine biosynthesis, precursor provision for glutathione production, lipid synthesis, glycolytic intermediates and amino acid synthesis [3].
Cancer cells consume glutamine at a rate exceeding that of endogenous glutamine biosynthesis.Therefore, glutamine is currently considered essential in some cancer types [8] and identifying the molecular mechanisms implicated in the metabolic adaptation of cancer cells appears as a promising strategy for cancer therapy to overcome drug resistance [9].
Undoubtedly, cell cycle regulators are critical factors for cancer progression.Major components of the cell cycle machinery consist of cyclins, cyclin-dependent kinases (CDKs), CDK inhibitors of the CIP/KIP and INK 4 families, as well as the pocket proteins of the retinoblastoma family (pRB, p107, p130) and the E2 factor (E2F) family of transcription factors that comprise ten different members encoded by eight distinct genes (E2F1-E2F8) [10].Upon mitogen activation, CDK4/6 and CDK2, with D-type and E-type cyclins respectively, sequentially phosphorylate the retinoblastoma proteins, thereby releasing E2F transcription factors [11].The E2F transcription factors are the effectors of this pathway and regulate the expression of genes involved in various cellular processes such as cell cycle progression and cellular metabolism [12].
Alterations in one or more key components of the CDK-pRB-E2F axis occur frequently in most cancers (e.g.inactivation of pRB, overexpression of CDKs or inactivation of CDK inhibitors).Accordingly, an increase in E2F activity is commonly found in many human cancers [10].However, depending on the cellular context and environmental conditions, E2F1 can function as an oncogene or a tumor suppressor gene as evidenced by the observation that E2f1-knockout mice develop spontaneous tumors [13,14].In addition, copy number variation (CNV) analyses in human cancer types identified that loss of function mutations occur but at a low frequency (<10 % of cases affected) (https://portal.gdc.cancer.gov/genes/ENSG00000101412).
The CDK-pRB-E2F1 pathway has also been implicated in the regulation of the glutamine metabolism.Indeed, a link between pRB/E2F1 and redox metabolism, specifically an increase in the synthesis of the antioxidant glutathione from glutamine after pRB loss was demonstrated in advanced diseases [15].In addition, in the context of pRB loss, glutamine metabolism is enhanced, partially through the direct control of glutamine uptake via specific E2F3-mediated transcription of target genes [16].Moreover, Qui et al. demonstrated the contribution of F-Box Protein 4 loss and hyperactivation of cyclin D1-CDK4/6 kinases, which causes the phosphorylation-dependent inactivation of pRB, to glutamine addiction in human esophageal squamous cell carcinoma [17].In sharp contrast, some studies showed that suppression of pRB phosphorylation by CDK4/6 inhibition is selectively associated with an increase in mitochondrial mass and mitochondrial metabolism fueled by an elevated glutamine metabolic flux in pancreatic cancer cells [18].In line with this, CDK4/6 knockdown in HCT116 colorectal carcinoma cells increases mitochondrial metabolism through elevated utilization of glutamine and enhanced mitochondrial respiratory capacity [19].Another study also showed that CDK4/6 inhibition by Palbociclib in A549 cells enhances glutaminolysis to maintain mitochondrial respiration and sensitizes A549 cell to the glutaminase inhibitor CB-839 [20].In summary, whereas several studies reported that CDK4 inhibition led to metabolic rewiring including increased glutamine addiction [18][19][20], others clearly demonstrated that RB loss or increased CDK4 activity increased glutamine metabolism [16,17,21].This discrepancy warrants a detailed molecular investigation of the molecular mechanism by which the CDK4/6-pRB-E2F pathway controls glutamine metabolism.
Cells were routinely monitored and confirmed to be free of mycoplasma.

RNA isolation, reverse transcription, and gene expression analysis
To investigate gene expression, primary MEFs or SV40 immortalized MEFs were plated in triplicates in 60 mm dishes (430,166, Corning) in mL standard growth media.After 24 h, MEFs were washed once with mL D-PBS (14,190,169, Thermo Fisher Scientific) and RNA was isolated using the TRI reagent (T9424, Sigma-Aldrich, Thermo Fisher Scientific) according to the manufacturer's protocols.RNA concentrations were determined with a NanoDrop 8000 (Thermo Fisher Scientific).Reverse transcription for cDNA generation was performed using the Super-Script™ II Reverse Transcriptase (8064-014, Invitrogen, Thermo Fisher Scientific).mRNA expression was assessed using real-time PCR on the QuantStudio™ 6 real-time PCR system (Applied Biosystems, Thermo Fisher Scientific) and SYBR Green detection (04913914001, Roche).Gene expression was normalized to GeoMean of housekeeping genes Tbp, Rs9 and Actb.Relative mRNA expression levels were calculated using the 2^( Delta Ct) method [25].For primer sequence see Supplementary table 2.

Cell lysis for protein extraction, determination of protein concentration and Western blot
To extract proteins from SV40 immortalized E2f1 +/+ and E2f1 − /− MEFs, cells were cultured in 60 mm dishes, washed twice with 4 mL D-PBS and scraped down in M-PER protein lysis buffer (78,501, Thermo Fisher Scientific) supplemented with 1× Halt™ Phosphatase (78,426, Thermo Fisher Scientific) and 1× Halt™ Protease inhibitor cocktail (78,429, Thermo Fisher Scientific).Afterwards, cell lysates were incubated for 30 min on ice, followed by 3 times 5 cycles of sonication (30 s ON, 30 s OFF) using the Bioruptor® Plus (Diagenode).After centrifugation (10,000 rpm, at 4 • C for 5 min), protein lysate was transferred into a new 1.7 mL tube and stored at − 20 • C until further usage.Protein concentration was determined using the Pierce BCA Protein Assay Kit (23,225,Thermo Fisher) according to the manufacturer's protocol.

RNA-seq
RNA was extracted as described above from 4 SV40 immortalized E2f1 +/+ and E2f1 − /− MEFs from 4 independent experiments.RNA quality was assessed on a Fragment Analyzer (Agilent Technologies) and all RNAs had a RQN above 9.3.RNA-seq libraries were prepared from 500 ng of total RNA with the Illumina TruSeq Stranded mRNA reagents (Illumina) using a unique dual indexing strategy, and following the official protocol.Libraries were quantified by a fluorimetric method (QubIT, Life Technologies) and their quality assessed on a Fragment Analyzer (Agilent Technologies).Sequencing was performed on an Illumina NovaSeq 6000 for 100 cycles single read.Sequencing data were demultiplexed using the bcl2fastq2 Conversion Software (version 2.20, Illumina).

Data acquisition -LC-HRMS analysis
Cell extracts were analyzed by Hydrophilic Interaction Liquid Chromatography coupled to high resolution mass spectrometry (HILIC -HRMS) in negative ionization mode using a 6550 Quadrupole Time-of-Flight (Q-TOF) system interfaced with 1290 UHPLC system (Agilent Technologies) as previously described [34].A SeQuant ZIC-pHILIC (100 mm, 2.1 mm I.D. and 5 μm particle size, Merck) column was used for metabolite separation.The mobile phase was composed of A = 20 mM ammonium Acetate and 20 mM NH4OH in water at pH 9.7 and B = % ACN.The linear gradient elution was from 90 % (0-1.5 min) to 50 % B (8-1L min) down to 45 % B (12-15 min).Finally, the initial chromatographic conditions were established during a 9 min post-run for column re-equilibration.The flow rate was 300 μL/min, column temperature 30 • C and sample injection volume 2 μL.Mass spectrometry ESI source conditions were set as follows: dry gas temperature 290 • C and flow 14 L min − 1 , fragmentor voltage 380 V, sheath gas temperature 350 • C and flow 12 L min − 1 , nozzle voltage 0 V, and capillary voltage − 2000 V.The instrument was set to acquire over the full m/z range 50-1000 with the MS acquisition rate of 2 spectra/s.In addition, AIF (all ion fragmentation) MS/MS analysis were performed on pooled QC samples at a collision energy (CE) of 0, 10 and 30 eV.

Data (pre)processing
Raw LC/MS files was processed using Profinder B.08.00 software (Agilent Technologies) and the targeted data mining in isotopologue extraction mode for metabolite annotation based on an in-house database containing around 600 polar metabolites (analyzed in the same analytical conditions).Metabolites will be identified based on accurate mass and retention time using the parameters settings as follows: mass accuracy tolerance 10 ppm, retention time tolerance 0.2 min, height filter 1000 counts, peak spectrum obtained as an average of scans at % of the peak maximal height.The relative quantification of metabolites was based on EIC (Extracted Ion Chromatogram) areas.The correction for natural isotope abundance was integrated in ProFinder [35].The obtained tables, containing peak areas of detected metabolites across all samples, were exported to "R" software http://cran.r-project.org/ and signal intensity drift correction was done within the LOWESS/Spline normalization program [35] followed by noise filtering (CV (QC features) > 30 %).Finally, the metabolite abundance was reported to total protein content (measured in the cell pellet using BCA assay, Thermo Fisher Scientific).

Metabolite identification
Short listed ions of interest were matched against the in-house created Accurate Mass Retention Time (AMRT) database.The identity of these putatively identified metabolites was further confirmed using the fragmentation (MS/MS data) pattern matching against METLIN standard spectral library [36].

Exploratory statistical analysis of label incorporation
The label incorporation or 13 C ##enrichment was calculated based on relative isotopolgue abundances in two different conditions [37,38].The univariate analysis (on log 10 transformed data) was applied to test the significance of label enrichment and changes in metabolite levels in different conditions (p-value = 0.05 followed by the Benjamini-Hochberg correction for multiple testing).

Data visualization
GraphPad prism software (v9.4.1) has been used to create bar charts.

Cell proliferation: Cy-quant assay
SV40 immortalized E2f1 +/+ and E2f1 − /− MEFs were plated in replicates (n = 6 / dish) in 96-well dishes (3599, Corning), with an initial seeding density of 2000 cells per well in DMEM (41966-062, Thermo Fisher Scientific) supplemented with 10 % heat-inactivated FBS (SV30160.03,Lot No. RB35956, HyClone, GE Healthcare Life Sciences).After seeding, cells were allowed to equilibrate overnight (16 h) and one 96-well dish was washed once with PBS and immediately frozen at − 80 • C to quantify the starting cell number prior to treatment.The wells in the remaining dishes were washed twice with PBS and media was replaced by DMEM no glucose, no glutamine, no phenol red (A14430-01, Thermo Fisher Scientific) supplemented with 10 % dialyzed and filtered FBS (26400-044, Thermo Fisher Scientific), and either 25 mM glucose (G8769, Sigma-Aldrich) and 4 mM glutamine (25030-024, Thermo Fisher Scientific) or only 4 mM glutamine.The media was changed every 24 h.Final cell counts were determined 1-4 days after the initial treatment using the CyQuant assay (C7026, Thermo Fisher Scientific) according to manufacturer's instructions.

ChIP-Seq tracs
The genome-wide occupancy of E2F1 in the genomic sequences of genes involved in glutamine metabolism in human cervical cancer cell line (HeLa) and human breast cancer cell line (MCF7) were analyzed using custom tracks from published ChIP-Seq datasets available from the UCSC genome browser (http://genome.ucsc.edu).Data were illustrated using the integrative genomics viewer (version 2.5.3, https://software.broadinstitute.org/software/igv/).

MYC-silencing in MEFs
siRNA-mediated gene knockdown in SV40 MEFs was achieved using 30 nM of a mix of 2 control siRNAs (SIC001 and SIC002, Sigma-Aldrich), or 30 nM of esi-RNA targeting Myc (EMU075291, Sigma-Aldrich).One day before transfection, 50,000 cells were plated in 6 cm dish and transfection was performed using Lipofectamine RNAiMax transfection reagent (13,778,075, Thermo Fisher Scientific) according to manufacturer's protocols.Cells were harvested 24 h post transfection.RNA was extracted as described above.MYC silencing was confirmed by qPCR.

WST-1 cell growth assay
To assess the impact of E2F1 on cellular growth, 2,5 × 10 3 LacZ pool, E2F1-ko clone 2 or E2F1-ko clone3 SK-UT1 cells were plated in a 96-well plate (8 replicates, one plate for each time point).Cells were cultivated for 24, 48, 72, and 96 h.At each time point, WST-1 proliferation reagent (Roche) was added to the wells in a 1:10 ratio according to the manufacturer's instructions.Plates were incubation at 37 • C for 120 min and colorimetric changes were measured using a SPECTROstar Omega (BMG Labtech, Ortenberg, Germany) at a wavelength of 450 nm with a reference wavelength of 620 nm.

Co-regulatory network and motif analyses
E2F1 co-regulatory network analysis was perform using ChIP-seq ENCODE data for HeLa-S3 cell line (67 transcription factors) downloaded from the ENCODE website (https://genome.ucsc.edu/)and analyzed as followed: Gene promoters (from EPDnew [42], 2 kb region upstream the TSS) with E2F1 binding sites were studied for the presence of other transcription factor (TF) binding sites (co-occurrence).First, the co-occurrence frequency of each TF with E2F1 was calculated (number of promoters with both E2F1 binding sites and target TF; versus the total number of promoters with E2F1 binding sites).Then, the co-occurrence frequency was compared to the inverse frequency (calculated with a bootstrap subsampling approach: frequency of E2F1 binding sites on a random selection of promoters with target TF binding sites, repeated 100 times).The difference between co-occurrence and inverse frequencies was then calculated and normalized by the inversed frequency.This analysis was restricted to Cell Cycle genes (KEGG id: hsa04110) and Alanine Aspartate Glutamate Metabolism genes (KEGG id: hsa00250).Finally, the presence of E2F family and MYC binding sites in genes promoters from these two pathways were clustered hierarchically (euclidean distance).

Computational analyses of the cancer genome atlas patient datasets
Datasets containing normalized bulk transcriptomic profiles of 29 tumor types generated with RNA-seq by The Cancer Genome Atlas (TCGA) consortium were downloaded from GDC data portal (https://portal.gdc.cancer.gov/)and cBioPortal (https://www.cbioportal.org/)[43], along with clinical information.The gene set constituting the glutamine signature was retrieved from the KEGG database (entry: 'hsa00250', name: 'Alanine, aspartate and glutamate metabolism -Homo sapiens (human)') [44].In a second analysis, the KEGG hsa00250 gene list was restricted to genes, which are direct E2F1 and MYC targets and were differentially expressed as indicated by the RNA-seq experiment (Supplementary table 1).The signature scores of each tumor transcriptome were computed with R package 'singscore' v1.12.0 [45].Survival analyses were performed with the R package 'survival' v3.2-13 [46], with p-values computed using log-rank test.All computational analyses were implemented in R version 4.1.1.

Statistical analysis
Results are illustrated as mean ± standard deviation (SD) or standard error of the mean (SEM) of at least three replicates if not indicated otherwise.Statistical significance was determined using the unpaired two-tailed student's t-test.Statistical analyses were carried out using the Prism software (GraphPad, v9.5.1).Normal distribution was assessed by Shapiro-Wilk normality test and further analyzed using a Student t-test (normal distribution) or a Mann-Whitney U test or an ordinary one-way ANOVA with Holm-šidák's multiple comparison (normal distribution) or Kurskal-Wallis test with Dunnett's multiple comparisons test and defined as *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001).

E2F1 regulates the expression of the glutamine transporter Slc1a5 and genes in the glutamine metabolic pathway
Previously, it was found that loss of E2F1 in mice results in tumorigenesis, demonstrating that E2F1 can also function as a tumor suppressor [13,14].If E2F1's function in metabolic rewiring contributes to tumor formation in mice has not been investigated in detail.
Immortalized embryonic fibroblasts (MEFs) were isolated from E2f1wild-type (E2f1 +/+ ) and E2f1-knock-out (E2f1 − /− ) mice (Fig. 1 A-C) and used as a genetic tool to determine the gene expression profiles of metabolic pathways underlying E2F1 regulation by RNA-seq analysis.2143 genes were identified differentially expressed in E2f1 − /− versus E2f1 +/+ MEFs (561 upregulated, 1582 downregulated) by gene expression profiling.Among them was the glutamine transporter Slc1a5, which was one of the most significantly upregulated genes in E2f1 − /− MEFs (Fig. 1D).KEGG pathway analysis indicated changes in several metabolic pathways including oxidative phosphorylation, which has been previously related to E2F1 activity [47].Interestingly, alterations in specific metabolic pathways comprising amino acid metabolism were detected (Supplementary Fig. 1A).A subset of these altered genes belongs to the glutamine family of amino acid metabolism-related genes, which were significantly upregulated in E2f1 knockout MEFs identified by RNA-Seq (Fig. 1E).Together, these data reveal the transcriptional networks that are disturbed by E2F1 depletion and highlight the upregulation of metabolic pathways that are known to be involved in tumor development and progression, such as glutamine metabolism.

Reprogramming of metabolism upon E2f1 deletion
Next, the metabolic consequences of the increased expression of genes that participate in the glutamine metabolic pathway in MEFs were determined.Steady-state metabolomic analysis was performed in E2f1 +/ + and E2f1 − /− MEFs and identified significantly upregulated metabolites related to glutamine metabolism in the absence of E2F1.Moreover, glutamine levels were significantly decreased in E2f1 − /− MEFs.These data indicated a role of E2F1 in regulating several amino acid metabolic pathways and suggested an increase in the glutamine catabolism, which is required for the biosynthesis of amino acids such as creatine, proline, aspartate, and glutathione in E2f1 − /− MEFs (Fig. 2A).Moreover, the analysis of the glutamine metabolic flux using [U -13 C 5 ]-glutamine as a tracer (Fig. 2B -C), showed a direct contribution of glutamine to the biosynthesis of glutamate m5, α-ketoglutarate (αKG) m5, succinate m4, fumarate m4, and malate m4, demonstrating both enhanced glutamine oxidation and the consequent carbon contribution to the oxidative TCA cycle (Fig. 2D).Interestingly, E2f1 − /− MEFs presented elevated levels of aspartate m3, and malate m3 (Fig. 2E), revealing a more active reductive carboxylation.Nevertheless, the relative proportion of aspartate m3 and malate m3 versus aspartate m4, and malate m4-labeled species indicated that oxidative TCA cycle was the main pathway of the glutamine metabolism, shown by the similar labeling pattern of total carbon contribution (Fig. 2F).In contrast, E2f1 +/+ MEFs demonstrated an increase in label enrichment of αKG m3, succinate m2 and malate m2, suggesting higher incorporation of glucose-derived pyruvate into the TCA intermediates (Supplementary Fig. 2D).Besides the elevated glutamine turnover and incorporation into oxidative TCA cycle in E2f1 − /− MEFs, we also found a higher contribution of glutamine to glutathione, proline, and asparagine levels (Fig. 2G) pointing to the enhanced incorporation of glutamine into additional glutamine consuming pathways.This increased glutamine metabolic fluxes were accompanied by increased ATP levels detected in E2f1 − /− MEFs (Supplementary Fig. 2A-C).Together, these data identified metabolic control and reprogramming of glutamine metabolism upon E2F1 depletion and indicate that E2f1 − /− MEFs are more efficient in metabolizing glutamine, thereby producing important intermediates for cellular proliferation.

Regulation of glutamine metabolism by E2F1 affects proliferation capacity
Previously, it was found that loss of E2F1 in mice results in tumorigenesis, demonstrating that E2F1 can also function as a tumor suppressor [13,14].Our results of increased cellular proliferation are consistent with the previous finding, showing that loss of E2f1 increased the proliferation of the E2f1 − /− compared to E2f1 +/+ SV40immortalized MEFs [48] (Fig. 3A).Glucose and glutamine are two main sources for carbon fuel in proliferating cells [49].To assess if the efficient metabolization of glutamine in E2f1 − /− MEFs can provide them a growth advantage, cells were cultured in media withdrawn of glucose and in the presence of high glutamine concentration.Indeed, E2f1 − /− MEFs continued to proliferate (albeit much more slowly compared to standard growth media), whereas E2f1 +/+ MEFs lost viability without glucose over a time course of 72 h (Fig. 3B-E).To investigate how E2f1 − / − MEFs can adapt to glucose starvation conditions by increasing glutamine metabolism, we carried out gene expression profiles in cells derived from media without glucose for 16 h and identified that glutamine metabolic genes were significantly upregulated in E2f1 − /− MEFs compared to E2f1 +/+ MEFs (Fig. 3F).
Together, these results confirm increased proliferation upon E2F1 loss, which partly dependents on glutamine metabolism.
(A-E) Proliferation rate of SV40-immortalized E2f1 +/+ and E2f1 − /− MEFs was determined in the presence or absence of glucose.(A-B) Cell counts, normalized to cell number at t = 0 when treatment media were applied, were assessed over a time course of 72 h and (Fig. 3C-E) used to calculate proliferation rate.Data are shown as mean ± SD of n = 8 technical replicates illustrating one representative experiment out of three.Statistical significance was determined using ordinary one-way ANOVA with multiple comparisons and defined as p < 0.05 (*), p < 0.01 (**), and p < 0.0001 (****).(F) Gene expression in SV40immortalized E2f1 +/+ and E2f1 − /− MEFs from RNA-seq data shown as z-score (light green to dark green).* padj ≤0.05, ** padj ≤0.01, **** padj ≤0.0001 were derived from differential expression analysis carried out in R using DESeq2 and FDR corrected using Benjamini Hochberg correction.

E2F1 binds to the promoter of the glutamine transporter Slc1a5 and of glutamine metabolic genes
To investigate if E2F1 could directly regulate the expression of these metabolic pathways, gene ontology (GO) analysis on biological processes using published ChIP-Seq datasets (http://genome.ucsc.edu)was performed, which revealed that pathways related to glutamine metabolism were enriched in genes that contain an E2F1 binding site in the promoter region (data not shown) and confirmed a direct binding of E2F1 to the promoter of glutamine metabolic genes, including the glutamine transporter SLC1A5 (Fig. 4A).In order to elucidate the molecular mechanisms underlying the observed increase in the expression of glutamine metabolic genes in E2f1 − /− MEFs, the hypothesis was tested if in the absence of E2F1, other transcription factors (TF) regulate the expression of glutamine metabolic genes in the proximity of the E2f1 binding site in the promoter of these genes.Therefore, to uncover E2F1 co-regulatory networks specifically on glutamine metabolic gene promoters compared to cell cycle gene promoters, ChIP-seq experiments including 61 unique TF in HeLa cells (ENCODE project) were used.Using these processed data, the co-occurrences of E2F1 with other TFs was uncovered.These analyses identified MYC as a factor that shows a high co-occurrence with E2F1 specifically on the promoter of glutamine metabolic genes compared to the promoters of cell cycle genes (Fig. 4B).To further investigate if these changes in glutamine metabolic gene expression were likely to be regulated by increased MYC binding after E2F1 depletion, MYC binding to the promoter of the glutamine transporter Slc1a5 was validated by ChIP-qPCR analysis, which showed an increased binding of MYC upon E2F1 loss compared to MYC binding in control MEFs (Fig. 4C).Finally, knockdown of Myc in E2f1 +/+ and E2f1 − /− MEFs (Fig. 4D-E) led to a significant reduction of Slc1a5 expression in E2f1 − /− MEFs (Fig. 4F).

The expression of glutamine metabolic genes is negatively-correlated with E2F1 and is associated with poor prognosis in human soft tissue sarcoma
To validate the association between E2F1 and glutamine metabolism and its relevance in human tumors, we performed a correlation analysis using RNA-seq data from TCGA patients across 29 different cancer types.The signature of glutamine metabolic genes (KEGG id: hsa00250; restricted to direct target genes of E2F1 and MYC, which were differentially expressed in RNA-seq experiment) was examined and its enrichment score showed a negative correlation in certain cancer types including a significantly negative correlation in soft tissue sarcoma (STS) when compared with expression of E2F1 (Fig. 5A).In STS, the glutamine signature score was also positively correlated with MYC expression, in agreement with its hypothesized role in regulating glutamine gene expression.In addition, a survival analysis comparing patients with high vs. low glutamine scores confirmed [50] that high expression of glutamine metabolic genes was indicative of poorer survival outcomes in STS (Fig. 5B).As previously described, one third of all spontaneously formed tumors in E2F1 knockout mice were found in the reproductive tract, including uterine sarcoma.Therefore, the uterine sarcoma cell lines SK-UT-1 was selected to silence E2F1 expression and a significant increase in the expression of glutamine metabolic genes such as SLC1A5, GLS, GOT1 and PPAT1 was observed by qPCR (Fig. 5C-D).
To further explore the role of E2F1 in SK-UT-1 cells, a E2F1-deficient cell line was generated by CRISPR/Cas9 technology (E2F1-ko and, as control, LacZ SK-UT-1 cells, Fig. 5E).We confirm that E2F1 knockout leads to elevated expression of glutamine transporter SLC1A5 and GLS levels (Fig. 5F).In addition, an increased cellular proliferation (Fig. 5G).and increased MYC binding to the promoter of SLC1A5, CAD, and GOT1 by CUT&RUN-qPCR were observed upon loss of E2F1 (Fig. 5H).
Together, these analyses confirm our findings in a human cancer model and indicate that E2F1 loss drives an increase in glutamine metabolism in uterine sarcoma cells.
(A) Correlation analyses illustrating the Spearman correlation coefficients and p-values between the expression of E2F1/MYC (in log2 (TPM + 1) unit) across patients of each tumor type and the single-sample scores of the selected glutamine signature (genes of KEGG id: hsa00250, with E2F1 and MYC binding peak and which were differentially expressed in RNA-seq, see Supplementary table 1 computed using the R package 'singscore'.These scores quantify how much the genes in the

Discussion
Previous studies have demonstrated a dual role of E2F1 transcription factor as an oncogene and tumor suppressor.While an increase in E2F activity is commonly found in many human cancers [10], E2F1's tumor suppressive properties have become evident by the observation that E2f1-deficient mice develop spontaneous tumors between 8 and 18 months of age.The fact that mice lacking E2f1 show an increase in tumor incidence suggests that E2F1 can also function as a tumor suppressor [13].
Mechanistically, a failure of the protective apoptotic pathway was suggested, eventually resulting in elevated proliferation and increased tumorigenesis as cells that might otherwise go on to form a tumor are not eliminated and therefore accumulate additional oncogenic alterations.However, the observation that inactivation of E2f1 accelerates tumorigenesis in mice expressing MYC under the control of an epithelialspecific keratin 5 promoter (K5 Myc mice) while at the same time increasing apoptosis in transgenic epidermis and tumors does not support the model in which E2F1 suppresses tumor development by inducing apoptosis [51].This suggests that E2F1 might have additional tumor suppressive activities.
The ability of E2F1 to affect the expression of diverse collections of genes including metabolic processes, rather than just cellular proliferation and apoptosis, may also contribute to its roles as a tumor suppressor.However, the mechanisms of how E2F1 affects metabolism in these tumors and how it contributes to tumor formation has not been determined yet and warrants a detailed investigation of the molecular mechanism how signaling through CDK4/6-RB-E2F pathway controls glutamine metabolism.
Our data identified metabolic control and reprogramming of glutamine metabolism that follows E2F1 depletion.The glutamine transporter, SLC1A5, was one of the most up-regulated genes upon E2F1 knockout in MEFs.In addition, E2f1 − /− MEFs are more efficient in metabolizing glutamine, thereby producing important intermediates for cellular proliferation.
Furthermore, ChIP-seq analysis showed a direct binding of E2F1 to the promoters of glutamine metabolic genes, implying a direct or indirect E2F1-mediated transcriptional repression of the cancer hallmark glutamine metabolism.Several levels of regulation of E2F1 activity have been described.The dual nature of E2F sites to positively or negatively regulate transcription may allow E2F1 to perform vastly different tasks depending on the promoter context in which E2F sites are located.E2F1 can function as a transcriptional repressor either through interaction with negative regulators including pRB [52][53][54][55], p53 [56], as well as alone [57][58][59] or through association with specific co-factors, such as HDAC1 [60], DNMT1 [59], KDMA4 [61] and BIN1 [62].The loss of E2F1 and the subsequent loss of the E2F1 repressor complexes may allow other transcription factors to bind to those binding sites on the promoter.Therefore, we studied which transcription factor showed a cooccupancy with E2F1 on the promoter of glutamine metabolic genes and compared them to the promoter of classical cell cycle targets.The transcription factor that showed the highest co-occupancy on glutamine metabolic gene promoter and not on cell cycle promoters was the oncogene MYC.
The transcription factor MYC in conjunction with the heterodimer partner protein MAX binds to E-box sequence elements and regulates several biological activities that contribute to tumorigenesis including glutaminolytic metabolism [63].The functional relationship between MYC and E2F1 is context dependent and appears to be complex.A cooperation of E2F1 inactivation and MYC overexpression has been identified as a specific oncogenic alteration to promote tumorigenesis in K5 Myc mice crossed with E2f1-null mice [51].Additionally, it has been reported that modulating upstream pathways of E2F1 such as inhibition of CDK4/6 in HCT116 cells prevents the phosphorylation of MYC at Thr58 and Ser62 and ensuing its ubiquitin-proteasome-dependent turnover.This leads to accumulation of MYC protein and subsequent upregulation of the gene networks orchestrated by MYC [19].Together with our data, there is increasing evidence for a complex link between MYC and the CDK4/6-Rb-E2F pathway that needs further investigation in certain human types of cancer.
Our observations in MEFs have been cross validated with analyses of transcriptomic profiles in 29 different human cancer types and highlighted STSs as a specific cancer type that showed a negative correlation between E2F1 and certain glutamine metabolic genes.
STSs are a rare group of heterogeneous mesenchymal cancers originating from connective tissue, now classified into >100 distinct subtypes based on their tissue of origin, genetic alterations, and age of occurrence.STSs account for 1 % of all adult cancers and due to their rarity and heterogeneity, these diseases remain relatively understudiedmaking new therapeutic targets necessary and challenging [64,65].
In accordance with the literature, we demonstrated that high expression of glutamine metabolic genes is closely associated with poor prognoses of patients [50,66].Moreover, tumor development was found to be accelerated in the absence of E2F1 in reproductive tract sarcomas, including uterine sarcoma, which represent one-third of all the tumors isolated from E2F1 − /− animals.Although, a broad spectrum of tumors developed in mice lacking E2F1 including lung tumors, lymphomas, und other tumor types found at lower frequency [13], our restricted analysis focusing on glutamine gene expression negatively correlated with E2F1 expression identified a specific importance in STS.Thus, we selected a cell line of leiomyosarcoma, which represents the most common type of uterine sarcoma [67], to validate our results in a human cancer model and confirm a significant increase in the expression of glutamine metabolic genes, increased proliferation and increased MYC binding to promoters of SLC1A5, GOT1 and CAD upon E2F1 knockout.
In conclusion, these data reveal a novel function of E2F1 in rewiring cellular metabolism through a cooperation with MYC to regulate glutamine metabolism-related genes at the transcriptional level.These findings seem to be relevant specifically for uterine sarcoma, encouraging testing novel pre-clinical strategies to tackle these metabolic pathways with pharmacological interventions.
signature are enriched among the most highly expressed genes.(B) KM plot showing the differences in overall survival between patients in low (bottom 25th percentiles of the signature score value) and high (top 25th percentiles) glutamine signature group for soft tissue sarcoma.Significance indicated by log-rank tests.(C) Analysis of E2F1 mRNA expression (D) and mRNA expression of glutamine metabolic genes in human uterine sarcoma SK-UT-1 cell line n = 8 replicates performed in 3 independent experiments.(E) Western blot illustrating CRISPR/Cas9 generated E2F1 knockout (ko) in SK-UT-1 cell line.n = 3 replicates generated on independent days.(F) Analysis of SLC1A5 and GLS mRNA expression in LacZ and E2F1-ko SK-UT-1 cells.n = 3 replicates generated on independent days.(G) Cell growth of LacZ control (ctrl) and E2F1-ko SK-UT-1 cells was assessed in WST-1 assay.n = 8 technical replicates.Graph illustrates one representative experiment out of three independent experiments.(H) CUT&RUN qPCR analysis of MYC enrichment at SLC1A5, CAD and GOT1 promoter in LacZ ctrl and E2F1-ko SK-UT-1 cells.CUT&RUN data show means of 2^( Delta Ct ) method, normalized to HPRT1 and VCAM1 control loci, as well as to the corresponding IgG

Fig. 4 .
Fig. 4. Co-occupancy of E2F1 and MYC on the promoter of glutamine metabolic genes.

Fig. 5 .
Fig. 5.The expression of glutamine metabolic genes is negatively-correlated with E2F1 and is associated with poor prognosis in human soft tissue sarcoma.