A novel EGR1-driven GLUL/mTOR axis regulates macropinocytosis-mediated crosstalk in pancreatic stellate cell-cancer metastasis Authors: Ran Xue, Jun Zhou, Yawei Liu, Yuxi Liu, Jing Wu, Xinran Song Published: 2026 ABSTRACT AbstractBackgroundPancreatic cancer liver metastasis relies on crosstalk between pancreatic stellate cells (PSCs) and cancer cells, yet the role of transcriptional regulation in this interplay, particularly via macropinocytosis, remains unclear. Here, we aimed to explore a novel EGR1-driven pathway linking PSC function to cancer cell nutrient acquisition and liver metastasis.MethodsSingle-cell RNA sequencing (scRNA-seq) was performed on human pancreatic cancer tissues to identify key regulators of PSC-cancer cell crosstalk. ChIP, EMSA, luciferase reporter assays, and western blotting were used to validate transcriptional regulation of GLUL by EGR1. Functional assays included cell proliferation, migration, invasion, and macropinocytosis analyses in co-culture systems. In vivo studies utilized a murine model of pancreatic cancer liver metastasis to assess the impact of the EGR1-GLUL/mTOR axis on metastasis.ResultsscRNA-seq identified EGR1 as a transcription factor enriched in PSCs, with strong co-expression of GLUL and mTOR pathway genes. EGR1 directly bound the GLUL promoter, promoting its transcription and activating mTOR signaling. This axis suppressed PSC activation (reduced α-SMA expression). EGR1 over-expressed PSCs inhibited pancreatic cancer cell macropinocytosis, leading to impaired nutrient uptake, reduced ATP production, and suppressed malignant behaviors (proliferation, migration, invasion). Additionally, the EGR1-GLUL/mTOR axis reduced liver metastasis in vivo.ConclusionThe EGR1-driven GLUL/mTOR axis in PSCs suppresses pancreatic cancer progression by inhibiting PSC activation, reducing cancer cell macropinocytosis, and restraining metastasis. This axis represents a promising therapeutic target for disrupting PSC-cancer cell crosstalk in pancreatic cancer.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08479-2.Keywords: Pancreatic cancer, Metastasis, Pancreatic stellate cells, Macropinocytosis, Metabolic reprogramming Introduction Pancreatic cancer, notorious for its dismal prognosis, ranks among the deadliest malignancies worldwide, with a 5-year survival rate of less than 10% [1]. Metastasis, particularly the spread to distant organs, accounts for over 90% of pancreatic cancer-related mortalities [2], underscoring the urgent need to decipher the molecular mechanisms driving this process. Although extensive research has focused on genetic mutations and signaling pathways within cancer cells, emerging evidence highlights the critical role of the tumor microenvironment (TME) in promoting metastasis [3]. Among TME components, pancreatic stellate cells (PSCs) have emerged as pivotal regulators, interacting closely with cancer cells to facilitate invasion, angiogenesis, and immune evasion [4]. However, the precise molecular mechanisms underlying the crosstalk between PSCs and cancer cells, especially those involving unconventional cellular processes, remain incompletely understood. Macropinocytosis, a conserved endocytic pathway enabling cells to internalize large volumes of extracellular fluid and macromolecules, has recently emerged as a novel driver of cancer progression [5]. In pancreatic cancer, macropinocytosis has been implicated in nutrient acquisition, immune evasion, and drug resistance [6]. However, its role in mediating crosstalk between PSCs and cancer cells during metastasis remains unexplored. Understanding how macropinocytosis orchestrates communication within the TME could unveil new therapeutic strategy for metastatic pancreatic cancer. The early growth response 1 (EGR1) transcription factor, a key regulator of cellular stress and inflammation, has been linked to tumorigenesis and metastasis across multiple cancers [7]. In pancreatic cancer, EGR1 exhibits context-dependent functions, acting as both a tumor suppressor and promoter depending on the cellular microenvironment [8]. Notably, EGR1 has been shown to modulate metabolic reprogramming and signaling pathways [9], yet its potential role in regulating the crosstalk between PSCs and pancreatic cancer cells via macropinocytosis remains elusive. Glutamine synthetase (GLUL) and the mammalian target of rapamycin (mTOR) pathway represent critical nodes in cellular metabolism and survival [10, 11]. GLUL catalyzes the conversion of glutamate to glutamine, a key amino acid fueling cancer cell proliferation and survival under nutrient-deprived conditions [12]. The mTOR pathway, a central regulator of cell growth, metabolism, and autophagy, integrates signals from nutrients, growth factors, and energy status [13]. Recent studies have revealed a functional interplay between GLUL and mTOR in promoting tumor progression [14]; however, whether these molecules form an axis regulated by EGR1 to drive macropinocytosis-mediated crosstalk in pancreatic cancer remains unknown. In this study, we aim to demonstrate that an EGR1-driven GLUL/mTOR axis serves as a critical regulator of macropinocytosis-mediated crosstalk between PSCs and pancreatic cancer cells during metastasis. Our findings uncover a previously unrecognized mechanism underlying the TME-dependent progression of pancreatic cancer and highlight potential therapeutic targets for combating metastatic disease. Materials and methods RNA-seq and data analysis Single-cell sequencing data related to pancreatic cancer were downloaded from the NCBI GEO database (accession numbers: GSE212966). The dataset comprised a total of six PDAC and six adjacent noncancerous resection specimens. All patients with PDAC did not receive any treatments before collecting specimens. Preprocessing of the data was performed using the Seurat R package (version 4.0.2). The single-cell transcriptome data were normalized using the NormalizeData function to improve data comparability and analytical accuracy. Subsequently, Principal Component Analysis (PCA) was performed on the normalized dataset employing the RunPCA function to reduce its dimensionality, retaining the top 50 principal components for subsequent analysis. To integrate and correct for batch effects, we employed the Harmony method. Seurat was then utilized to perform graph-based clustering, analyzing each cell based on its gene expression profiles and visualizing the results through t-SNE non-linear dimensionality reduction. Known markers were used to define each cluster, referencing the CellMarker2.0 database. The parameters were set as follows: a minimum of 2,000 highly variable genes (nfeatures = 2000), dimensions ranging from 1 to 10 (dims = 1:10), and a resolution of 0.8. A preliminary gene filtering step was conducted to retain only those genes expressed in over 200 cells. Subsequently, criteria for high-quality cells were established after assessing the overall cellular composition, which include: (1) Detection of more than 200 genes per cell with an average cell count exceeding 20,000; (2) A mitochondrial gene proportion below 10%. Differential expression, differential and enrichment analysis and gene set enrichment analysis (GSEA) To identify the distinct genes associated with each subgroup, we performed a comparative analysis of gene expression levels within each group and across all other groups following the integration process. Bubble plots were generated, showcasing the top three marker genes from each group’s list. We conducted differential analysis to compare the PDAC group and versus the adjacent noncancerous resection specimens group for each subgroup, merging and clustering the results. The screening criteria were set as p-value < 0.05 and an absolute fold change (|FC|) > 1.5. Concurrently, we utilized the clusterProfiler software to perform GO (Gene Ontology) functional enrichment analysis and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis on the significantly different gene sets of each cell population. The GO enrichment results were categorized into three main classes: BP (Biological Process), CC (Cell Composition), and MF (Molecular Function), and KEGG pathway analysis was also carried out. GSEA was employed to conduct potential functional analysis on all genes and cells. The MSigDB (Molecular Signature Database) was utilized, incorporating Hallmark gene sets and polyamine metabolism-related gene sets. This analysis examined the expression levels of different genes and their relationships across various pathways. Bioinformatics analysis and prediction of potential binding sites for EGR1 in the promoter region of GLUL gene and the relationship between EGR1 and GLUL The 2000 bp upstream nucleotide sequence of the promoter region of GLUL was analyzed using NCBI and UCSC Genome Brower online software. Utilizing Jaspar online analysis system (http://jaspar.genereg.net/cgi-bin/jaspar_db.pl) Analyze the promoter region of the GLUL gene and screen potential binding sites for EGR1 through scoring. We used the Gene Expression Profiling Interactive Analysis platform to visually analyze pancreatic cancer cases in the TCGA database (http://gepia.cancer-pku.cn/). Cell culture and isolation Primary mouse PSCs were isolated from C57BL/6 mice. Paracancerous pancreatic tissues were cut into 0.2–0.5 mm3 tissue blocks with sterile surgical scissors and transferred to 6-well culture plates. PSCs were cultured in DMEM/F-12 containing 20% foetal bovine serum(FBS with 5% penicillin and streptomycin). Mouse pancreatic cancer cell line (Panc02 cells) was obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and cultured in Roswell Park Memorial Institute (RMPI) supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (0.1 mg/ml). All cell lines were cultured in 100% humidity at 37 °C and 5% CO2. Plasmid construction, lentivirus transfection, and RNA interference The EGR1 coding sequence was amplified by PCR and cloned into the pcDNA3.1 vector (Invitrogen) to generate the EGR1 overexpression plasmid. For stable overexpression, lentiviruses encoding EGR1 and GLUL (JI KAI Gene Technology Co. Ltd.) were transfected into PSCs at a multiplicity of infection (MOI) of 30, and stable cell lines were selected with puromycin (2 µg/mL) for 2 weeks. Empty lentiviral vectors were used as negative controls. Lentivirus-based RNA interference (RNAi) was performed to knock down EGR1 and GLUL expression. Lentiviral vectors encoding shRNAs targeting EGR1 and GLUL, as well as scrambled shRNA controls, were purchased from JI KAI Gene Technology Co. Ltd. PSCs were transfected with lentiviral shRNAs at MOI 30, and stable knockdown cell lines were established by puromycin selection (2 µg/mL) for 2 weeks. Transfection efficiency was verified by qRT-PCR and Western blot at 48 h post-transfection, with a knockdown efficiency > 70% considered acceptable for subsequent experiments. PSCs and Pan02 cell co-culture Pan02 cells were co-cultured with different group PSCs to verify the malignant behavior of Pan02 cells by PSCs using Transwell systems (0.4 μm membrane). Quantitative real-time PCR (qRT-PCR) Total RNA was extracted from cells using TRIzol reagent (Invitrogen) and reverse-transcribed into cDNA using the PrimeScript RT Reagent Kit (TaKaRa). qRT-PCR was performed using SYBR Green Master Mix (Roche) on a LightCycler 480 instrument (Roche). Relative gene expression was calculated using the 2^(-ΔΔCt) method, with GAPDH as the internal control. Here are the primer sequences as follow: GAPDH: Sense5’-TGAAGGGTGGAGCCAAAAG-3’Antisense5’-AGTCTTCTGGGTGGCAGTGAT-3’ EGR1:Sense 5’-CCTTCGCTCACTCCACTATCC-3’ Antisense 5’-GTTACGCATGCAGATTCGAC-3’GLUL: Sense5’-GTCCAAGCCATGTATATCTGGG-3’ Antisense 5’-TCCGGTTATACTTGAAAACTTCAC-3’. Western blot assay Cells were lysed on ice, and total protein was collected. The bicinchoninic acid (BCA) method was used to determine protein concentrations. Proteins were then separated via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (Merck Millipore, MA, USA). The membranes were blocked with 5% bovine serum albumin (BSA) in TBST for 1 h at room temperature and subsequently incubated with primary antibodies overnight at 4 °C. Following this, the membranes were incubated with secondary antibodies for 1 h at room temperature. Immunoreactive bands were detected using chemiluminescence (ECL Plus, Merck Millipore). The following antibodies were used: anti-α-SMA (Proteintech, Wuhan, China), anti-EGR1 (Cell Signaling Technology [CST], Beverly, MA, USA), anti-P62 (CST), anti-LC3 (CST), anti-GLUL (abcam, USA), anti-phospho-mTOR (CST), anti-VE-cadherin (CST), anti-E-cadherin (CST), anti-VEGF (CST), anti-MMP2 (CST), or anti-MMP9 (CST), all incubated overnight at 4 °C. Flow cytometry To measure the cell cycle change of Pan02 cells treated with different group PSCs, 2 × 105–1 × 106 cells were collected and centrifuged. After washing, centrifugation, and supernatant discard, 1 mL of DNA staining solution and 10µL of permeabilization solution were added. The mixture was vortexed for 5–10 s and incubated at room temperature in the dark for 30 min. Detection was performed on a flow cytometer. CCK-8 assay Cell proliferation was assessed using the cell counting kit-8 (CCK-8; Dojindo, Kumamoto, Japan) assay. Pan02 cells were seeded into 96-well plates in DMEM supplemented with 20% FBS and cultured for 12 h. Subsequently, the medium was replaced with fresh DMEM containing 2% FBS, and the cells were incubated for an additional 72 h. To evaluate the proliferation of Pan02 cells, the cells were cultured in conditioned medium (CM) derived from PSCs supplemented with 20% FBS for 72 h. CCK-8 solution (20 µL per well) was then added, and the plates were incubated for 4 h. Absorbance was measured using a microplate reader (Beckman Coulter, Brea, CA, USA) at a wavelength of 490 nm. Wound-healing assay To evaluate the migration capacity of Pan02 cells, the cells were seeded onto a 24-well plate. A scratch was created across the well using a 200-µL pipette tip mounted in a custom-made holder. The plates were then washed to remove any detached cells and incubated with fresh serum-free medium. The wound width was measured 36 h post-scratch induction. Transwell assay Transwell assay was performed to determine the cell migration and invasion ability. Pan02 cells were cultured in conditioned media from different group PSCs. Transwell permeable supports (8.0- µm pore size; Corning, Kennebunk, ME, USA) were used to determine the cell migration ability. For the invasion assay, chambers coated with Matrigel (BD CioCoat#354480) were used. After 24 h of incubation at 37 °C, the migratory or invading cells on the lower surface of the filter were stained with crystal violet and quantified using GraphPad Prism version 10.1.2 (GraphPad Software, San Diego, CA, USA). Dual luciferase reporter assay Upon reaching 90% confluence, PSCs were transfected for 4 h, followed by culture in medium containing 10% fetal bovine serum (FBS) for 24 h. The Dual luciferase reporter assay (DLR) was conducted using the DLR™ assay system per the manufacturer’s protocol. PGL-3 reporter and PRL-TK vectors were co-transfected at a 20:1 mass ratio. Luminescence measurements were obtained using a Mithras LB940 luminometer (Berthold Technologies). Experiments were performed in triplicate. Data validity required Renilla luciferase luminescence to be at least 1000-fold lower than firefly luciferase luminescence; failing this, results were invalid. Electrophoretic mobility shift assay (EMSA) We utilized ultracompetitive EMSA, a method derived from standard competitive EMSA. PSC nuclear extracts were isolated using a nuclear extraction kit (KeyGEN Biotech, China) following the manufacturer’s instructions. Double-stranded DNA probes were generated by annealing complementary oligonucleotides with a DNA annealing buffer (Beyotime, China). The wild-type (WT) probe sequence is as follows: 5’-CCCACTCCACCCACCCCTTGG-3’. 5’-CCCACTCCACCCACCCCTTGG-3’. The double-stranded WT probe was biotinylated at the 5’ end, and the mutation (MUT) probe sequence is: 5’-CCCACCTACGACGCCCCTTGG-3’. 5’-CCCACCTACGACGCCCCTTGG-3’. EMSA binding reactions were performed using a chemiluminescent EMSA kit (Thermo Scientific, China). For competitive EMSA, unlabeled probes were added to the reaction mixture to confirm binding specificity. DNA-protein complexes were then resolved by 4% native polyacrylamide gel electrophoresis. Chromatin immunoprecipitation (ChIP) assays ChIP assays were performed using the SimpleChIP® Enzymatic Chromatin Immunoprecipitation Kit (Cell Signaling Technology, USA) according to the manufacturer’s instructions. Nuclei were first isolated from PSCs. Chromatin was then sonicated using a Bioruptor (Belgium) to generate DNA fragments primarily ranging from 100 to 500 bp. Immunoprecipitation was carried out with an anti-EGR1 antibody. Bound DNA was subsequently purified and amplified by PCR using the following primers:5′-TAGAGCAAAATCCTAATACAAAGCT-3′;5′-TCTGTTTCTTGGTACACACCATATC-3′. Amplified PCR products were separated via 1% agarose gel electrophoresis using Tris-Acetate-EDTA (TAE) buffer (Beyotime, China). Purified DNA was amplified by quantitative PCR (qPCR). The relative enrichment of target DNA in IP samples versus IgG control samples, expressed as a percentage of total input DNA, was statistically analyzed. Immunofluorescence staining For immunofluorescence, PSCs were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, blocked with 5% BSA, and incubated with primary antibodies against EGR1 (Cell Signaling Technology) overnight at 4 °C. After washing, cells were incubated with Alexa Fluor-conjugated secondary antibodies (Thermo Fisher Scientific) and DAPI (Sigma-Aldrich) for nuclear staining. Images were captured using a confocal microscope (Zeiss LSM 880). Macropinocytosis assay Cells were incubated with 1 mg/mL fluorescently labeled dextran (Alexa Fluor 488, 70 kDa) in serum-free medium for 1 h at 37 °C. For Pan02 cells, laser confocal microscopy (Zeiss LSM 880) was used for high-resolution imaging to assess macropinocytosis. Briefly, after incubation, cells were washed with ice-cold PBS, fixed with 4% paraformaldehyde, and stained with DAPI for nuclear visualization. To assess macropinocytic activity, the proportion of positive cells was quantified using manual counting. A positive cell was defined as one with clear DAPI-stained nuclei and distinct, bright green fluorescent puncta tightly surrounding the perinuclear region. And the cells located at the image borders were excluded. After separately counting the number of positive cells and the total number of cells, the percentage of positive cells was calculated. The data are presented as the mean ± SD of three independent biological replicates. Additionally, to minimize subjective bias, all images were coded and counted separately by two independent observers under double-blind conditions. Metabolic analysis Targeted metabolic assays Amino acid was measured using the Amino Acid Assay Kit according to the manufacturer’s instructions. ATP levels were determined using the ATP Assay Kit (Sigma-Aldrich). Glutamine and glutamate levels were measured using the Glutamine Assay Kit (Cayman Chemical). The levels of glutamine and glutamate were determined using the Glutamate Assay Kit (Catalog No.: A074-1) and the Glutamine Assay Kit (Catalog No.: A073-1) (Nanjing Jiancheng Bioengineering Institute, China), respectively, according to the manufacturer’s instructions. Untargeted metabolomics Pan02 cells were co-cultured with different PSCs for 48 h. Pan02 cells were harvested, and metabolites were extracted using 80% methanol (pre-cooled to -80 °C). Metabolomic analysis was performed using an ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) system (Thermo Fisher Scientific, Vanquish UHPLC and Q Exactive HF-X mass spectrometer). Chromatographic separation was achieved on an ACQUITY UPLC HSS T3 column (100 × 2.1 mm, 1.8 μm) with a gradient elution of 0.1% formic acid in water and 0.1% formic acid in acetonitrile. MS data were acquired in both positive and negative ion modes. Data processing was performed using Compound Discoverer 3.1 software (Thermo Fisher Scientific), including peak alignment, annotation, and quantification. Differential metabolites were identified using multivariate statistical analysis (OPLS-DA) with VIP > 1 and P < 0.05 (Student’s t-test). Pathway enrichment analysis was conducted using MetaboAnalyst 5.0. Glutamine synthetase (GS) via ELISA GS protein in mouse cell lysates was quantified using a Mouse GS ELISA Kit (Aviva Systems Biology). PSCs were lysed in RIPA buffer (with 1% protease inhibitor) on ice for 30 min; supernatants were collected post-centrifugation (12,000×g, 15 min, 4 °C), and protein concentrations were measured by BCA assay. 100 µL of GS standard (0.312–20 ng/mL) or diluted sample (1:50 in assay buffer) was added to pre-coated wells and incubated at 37 °C for 2 h. After 3 washes, 100µL biotin-conjugated anti-mouse GS antibody was incubated (37 °C, 1 h), followed by 3 washes and 100 µL streptavidin-HRP incubation (37 °C, 30 min). 90 µL TMB substrate was added (10–20 min, dark, RT); reaction was stopped with 50 µL stop solution, and absorbance was read at 450 nm (Bio-Rad Model 680). GS concentrations were calculated from the standard curve and expressed as pg/mgprot protein (mean ± SD, n = 3). Vasculogenic mimicry (VM) assay Matrigel (Corning) was added to 96-well plates and allowed to polymerize. Cells were seeded on Matrigel and cultured for 24 h. VM structures were imaged under a microscope, and the number of branch points (junctions) was counted using ImageJ software. Animal All animal experiments were approved by the Institutional Animal Care and Use Committee of Beijing Cancer hospital (Approval No.: 2020KT120) and conducted in accordance with the Guide for the Care and Use of Laboratory Animals. Six-week-old male C57BL/6 mice (20–22 g) housed under specific pathogen-free (SPF) conditions with a 12 h light/dark cycle, controlled temperature (22 ± 1 °C) and humidity (50 ± 5%), and free access to food and water. Pancreatic cancer liver metastasis model Murine pancreatic cancer Pan02 cells were stably transfected with a firefly luciferase-expressing plasmid (pGL4.51, Promega) using Lipofectamine 3000, and positive clones were selected with G418 (400 µg/mL) for 2 weeks. Luciferase activity was verified using the Luciferase Assay System (Promega) to ensure stable expression. To evaluate the role of EGR1 in pancreatic cancer liver metastasis through the regulation of GLUL and to verify the functional relationship between them, the following overexpression, knock-down, control, and rescue experimental groups were established, mice were randomly divided into 11 groups (n = 8, per group): Group A: LV-luc marked Pan02 cells Group A: LV-luc marked Pan02 cells Group B: LV-luc marked Pan02 cells + LV-GLUL(overexpressed) PSCs Group B: LV-luc marked Pan02 cells + LV-GLUL(overexpressed) PSCs Group C: LV-luc marked Pan02 cells + si-GLUL(knocked down) PSCs Group C: LV-luc marked Pan02 cells + si-GLUL(knocked down) PSCs Group D: LV-luc marked Pan02 cells + LV-GLUL-NC(control) PSCs Group D: LV-luc marked Pan02 cells + LV-GLUL-NC(control) PSCs Group E: LV-luc marked Pan02 cells + si-GLUL-NC(control) PSCs Group E: LV-luc marked Pan02 cells + si-GLUL-NC(control) PSCs Group F: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs Group F: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs Group G: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs Group G: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs Group H: LV-luc marked Pan02 cells + LV-EGR1-NC(control) PSCs Group H: LV-luc marked Pan02 cells + LV-EGR1-NC(control) PSCs Group I: LV-luc marked Pan02 cells + si-EGR1-NC(control) PSCs Group I: LV-luc marked Pan02 cells + si-EGR1-NC(control) PSCs Group J: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs + si-GLUL(knocked down) PSCs Group J: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs + si-GLUL(knocked down) PSCs Group K: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs + LV-GLUL(overexpressed) PSCs Group K: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs + LV-GLUL(overexpressed) PSCs Mice were anesthetized with isoflurane (induction: 3–5%, maintenance: 1.5-2%) and placed in a supine position. The abdomen was disinfected with 75% ethanol, and a 0.5 cm midline incision was made to expose the spleen. A 30-gauge needle was used to inject 100 µL of cell suspension (containing 1 × 10⁶ Pan02-luc cells and 1 × 10⁶ different PSCs in PBS) into the splenic parenchyma. After injection, the injection site was compressed with sterile gauze for 30s to prevent cell leakage. The spleen was returned to the abdominal cavity, and the incision was closed with 4 − 0 absorbable sutures. The randomization sequence was generated using a computerized random number generator (R software version 4.2.1), and group allocation was concealed via sequentially numbered, opaque sealed envelopes opened only by personnel not involved in animal handling. Outcome assessors responsible for histopathological analysis and biochemical testing were blinded to group allocation throughout the experiment and data analysis process. Small animal optical molecular imaging In vivo bioluminescent imaging was performed using an IVIS Spectrum imaging system (PerkinElmer) at 28 days post-inoculation. Mice were intraperitoneally injected with D-luciferin potassium salt (150 mg/kg body weight, PerkinElmer) and anesthetized with isoflurane. After 10 min of luciferin metabolism, bioluminescent images were acquired with an exposure time of 1–5 min, binning factor of 4, and f/stop of 1. The total flux (photons/s) of the abdominal region (for primary tumors) and liver region (for metastatic lesions) was quantified using Living Image software (PerkinElmer) to assess tumor growth and metastasis progression. Tumor size measurement and metastasis counting At 28 days post-inoculation, mice were euthanized by cervical dislocation under deep anesthesia. The pancreas was carefully dissected to expose the primary tumor, and tumor dimensions were measured using a digital caliper. Tumor volume was calculated using the formula: Volume = (length × width²) / 2. The liver was excised, rinsed with cold PBS, and observed under a stereomicroscope (Olympus). The number of visible metastatic nodules on the liver surface was counted and recorded. For each liver, the left, middle, and right lobes were separated, and representative images were captured using a digital camera. Histopathological analysis Tissues were fixed in 4% paraformaldehyde for 24 h, dehydrated through a graded ethanol series, embedded in paraffin, and sectioned into 4 μm slices. Hematoxylin and eosin (HE) staining was performed as follows: sections were deparaffinized in xylene, rehydrated in graded ethanol, stained with hematoxylin for 5 min, differentiated in 1% hydrochloric acid-ethanol for 30 s, blued in 1% ammonia water for 1 min, and stained with eosin for 3 min. After dehydration and clearing, sections were mounted with neutral balsam. Histopathological changes were observed under a light microscope (Olympus BX53), and images were captured at 100× and 400× magnifications. Tumor cell infiltration, stromal reaction, and metastatic foci were evaluated by a pathologist blinded to the experimental groups. Biochemical analysis of tumor tissues Fresh tissues were collected, weighed, and homogenized in ice-cold PBS (1:10 w/v) using a tissue grinder. The homogenate was centrifuged at 12,000 × g for 15 min at 4 °C, and the supernatant was collected for biochemical assays. ATP level measurement: ATP concentration was determined using the ATP Assay Kit (Sigma-Aldrich) according to the manufacturer’s instructions. Briefly, 50 µL of tissue supernatant was mixed with 50 µL of luciferin-luciferase reagent, and luminescence was measured using a microplate reader (BioTek). ATP concentration was calculated from a standard curve and normalized to tissue weight (nmol/mg tissue). Statistical analysis Data are presented as the mean ± SD of at least three independent biological replicates. Statistical significance was determined using Student’s t-test, one-way ANOVA, or two-way ANOVA, followed by Tukey’s post-hoc test. Correlation analysis was performed using Pearson’s correlation coefficient. P < 0.05 was considered statistically significant. All analyses were performed using GraphPad Prism 8 software. Results Human pancreatic tissue single-cell sequencing analysis reveals the potential role of EGR1 - GLUL/mTOR axis in PSC-cancer cell crosstalk To identify key molecular players in the crosstalk between PSCs and pancreatic cancer cells, we performed single-cell RNA sequencing (scRNA-seq) analysis on human pancreatic cancer tissues and adjacent normal tissues using GSE212966 data. Unbiased clustering analysis revealed distinct cell populations, including PSCs and pancreatic cancer cells (Fig. 1). The bubble plot using the top 3 genes of each group in the Markgene table was drawn (Fig. 1). Differential expression analysis identified a set of genes with significantly differences in PSCs from cancer tissues compared to normal tissues. The violin plot was used to display the expression levels of genes. It is worth noting that the results showed significant down-regulation of EGR1 and GLUL of PSCs in the pancreatic cancer tissues compared to the adjacent normal tissues (Fig. 1). GSEA further indicated that the mTOR signaling pathway was significantly activated in EGR1 + PSCs in pancreatic cancer (Fig. 1). Meanwhile, we used the Gene Expression Profiling Interactive Analysis platform (http://gepia.cancer-pku.cn/) to visually analyze pancreatic cancer cases in the TCGA database. As a result, there was a significant correlation between the expression of EGR1 and GLUL in pancreatic cancer (P = 9.6E-10, R = 0.32) (Fig. 1). By NCBI and UCSC Genome Brower online software, we analyzed the upstream 2000 bp nucleotide sequence of the promoter region of GLUL. The Jaspar online analysis system Analysis (http://jaspar.genereg.net/cgi-bin/jaspar_db.pl) was utilized to screen potential binding sites for EGR1 through scoring (Fig. 1). These bioinformatics findings suggested that the EGR1-GLUL/mTOR axis might play a crucial role in the crosstalk between PSCs and cancer cells during pancreatic cancer progression. A: The tSNE diagram of cell distribution between PDAC and ADJ. B: Bubble plot displaying the top 3 marker genes for each cell type in the Markgene table. C: Violin plot showing GLUL and EGR1 expression levels in PSCs. Wilcoxon Test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. D: GSEA enrichment plots showing EGR1’s correlation with mTOR pathway. E: Scatter plot showing the correlation between EGR1 and GLUL expression in pancreatic cancer. Pearson correlation test. (P = 9.6E-10, R = 0.32). F: NCBI, UCSC Genome Brower online software and The Jaspar online analysis system Analysis (http://jaspar.genereg.net/cgi-bin/jaspar_db.pl) showing potential binding sites for EGR1 EGR1 directly bind the GLUL promoter and drove its transcription in activated PSCs To uncover the fundamental molecular mechanism driving the EGR1-mediated regulation in the pancreatic tumor microenvironment, we first focused on identifying the direct downstream targets of EGR1 in PSCs. PSCs were isolated from C57/BL6 mice and phenotypically characterized by Oil Red O staining and immunofluorescence staining. The results showed that freshly isolated PSCs exhibited a quiescent phenotype, characterized by abundant lipid droplets and positive staining for Desmin. In contrast, cultured PSCs displayed an activated phenotype, with high expression of α-SMA. These findings confirm the successful establishment of a PSC cell model suitable for further research (Fig. 2). A: Lipid droplet content in PSCs. Oil Red O staining images were representative of three independent experiments, showing abundant intracellular lipid droplets in quiescent PSCs at different magnifications. (1: original magnification, × 100; 2: original magnification, × 200; 3: original magnification, × 400). B: Expression of characteristic markers in PSCs. Immunofluorescence staining were representative of three independent experiments and identifies markers of different activation states. a: Activated PSCs positive for α-SMA, stained by IF (1: Hoechst 33,258 staining, original magnification, × 400; 2: Alexa Fluor 488 staining for α-SMA, original magnification, × 400; 3: merged image, original magnification, × 400). b: Quiescent PSCs positive for desmin, stained by IF (1: Hoechst 33,258 staining, original magnification, × 400; 2: Alexa Fluor 594 staining for desmin, original magnification, × 400; 3: merged picture, original magnification, × 400). C: Western blot analysis and quantitative analysis of EGR1, GLUL, α‑SMA, and p‑mTOR in PSCs. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. D: qRT-PCR analysis of EGR1 and GLUL mRNA levels in PSCs (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. E: Western blot analysis and quantitative analysis of EGR1 and GLUL in PSCs in rescue experiment. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. F: Western blot analysis and quantitative analysis of α-SMA in PSCs in rescue experiment. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 qRT-PCR analysis revealed that EGR1 overexpression in PSCs led to a 1.63-fold increase in GLUL mRNA levels. While si-EGR1 in PSCs led to a 0.47-fold decrease in GLUL mRNA levels, accompanied by a corresponding downregulation of GLUL protein expression. The rescue experiments also deeply demonstrated that EGR1, as an upstream regulator, positively mediated the expression of GLUL (Fig. 2). Subsequently, we performed immunofluorescence staining and laser confocal scanning and observed the phenomenon of translocation and nucleation of EGR1 (Fig. 3). To detect GLUL promoter transcription core region, we searched out the CDS sequence of GLUL gene provided by NCBI. A promoter plasmid was truncated every 500 bp and connected to luciferase reporter vector pGL3-basic. The results of the double luciferase reporter gene detection system showed that after truncation of the region from − 3000 to -2500 bp (the GLUL-1 truncation), the promoter activity plummeted to a low level. The activities of the remaining truncations (GLUL-2 to GLUL-5) all remained near this low basal level, indicating that the region truncated in GLUL-1 (-3000 to -2500) is the core region of the promoter (Fig. 3). We further investigated the transcriptional activation of GLUL by EGR1 through dual luciferase reporter system in promoter region. The results showed that transfection of EGR1 and GLUL plasmids alone did not change luciferase activity, while luciferase activity increased significantly after transfection of EGR1 and GLUL, indicating that EGR1 may activate and promote the transcription of GLUL (Fig. 3). Through ChIP-qPCR assays conducted on primary cultured PSCs, we observed significant enrichment of EGR1 on the GLUL promoter region (Fig. 3), suggesting a potential transcriptional regulatory relationship. EMSA experiment also showed that EGR1 binds directly to the GLUL promoter (Fig. 3). A: Immunofluorescence staining were representative of three independent experiments, detecting the expression of EGR1 in PSCs. B: Promoter truncation assay of GLUL promoter regions (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. C: Dual-luciferase assays of EGR1 binding to GLUL promoter (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. D: EMSA image was representative of three independent experiments and showed that EGR1 binds directly to the GLUL promoter. E: ChIP analysis of EGR1 binding to GLUL promoter in PSCs using anti-EGR1 antibody or negative control IgG (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. F: Elisa analysis of glutamine synthetase in PSCs (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. G: Metabolic analysis of glutamate (Glu) and glutamine (Gln) in pre-treatment (before transduction) and post-treatment (after transduction) PSCs (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 GLUL-derived glutamine metabolic reprogramming inhibits the activation of PSCs through mTOR signaling pathway To further determine the role of glutamine and glutamate in the process, ELISA was performed and cells were harvested before transduction and at 48 h post-infection, showed as pre-treatment groups (before transduction) and post-treatment groups (after transduction). The results revealed that LV-GLUL in PSCs showed an increase in glutamine synthetase and glutamine in supernatant and intracellular concentration, together with a significant decrease in glutamate; On the contrary, the results could be reversed in si-GLUL in PSCs (Fig. 3). These data suggested that GLUL overexpression induce enhanced Gln anabolism in PSCs. Given that GLUL catalyzes the synthesis of glutamine, a key amino acid fueling cellular metabolism, we then investigated its impact on the mTOR pathway. Phosphorylation level of mTOR, crucial effector of the mTOR pathway, were significantly elevated in LV-GLUL PSCs, while si-GLUL in PSCs completely abrogated mTOR activation (Fig. 2). Next, we examined the function of GLUL in the activation of PSC. The expression of α-SMA, a marker of PSC activation, significantly increased in si-GLUL PSCs (Fig. 2). Rapamycin (100 nM, 24 h) fully reversed GLUL-mediated α-SMA suppression (p < 0.05), confirming mTOR-dependency (Fig. 2). To conclude, GLUL can increase the Gln synthesis metabolism and secretion and inhibit the activation of PSC through mTOR signaling pathway. LV-EGR1/GLUL PSCs suppress proliferation, migration, and invasion of pancreatic cancer cells Building upon the discovery of the EGR1-GLUL/mTOR regulatory axis in PSCs, we next sought to determine its functional significance in the context of PSC cancer cell crosstalk. To investigate intercellular crosstalk, we cultured Pan02 cells in conditioned media from different group PSCs. Cell proliferation assays revealed that Pan02 cells co-cultured with LV-EGR1/GLUL in PSCs exhibited a significantly blunted growth curve compared to those co-cultured with si-EGR1 PSCs, as measured by CCK-8 assay (Fig. 4). Flow cytometric analysis of cell cycle distribution further showed an increase in the proportion of Pan02 cells arrested in the G0/G1 phase when co-cultured with LV-EGR1/GLUL PSCs, accompanied by a corresponding decrease in S-phase cells. On the contrary, the results could be reversed in si-EGR1/GLUL PSCs (Fig. 4). These data collectively indicate that LV-EGR1/GLUL PSCs inhibit pancreatic cancer cell proliferation by inducing cell cycle arrest. A: CCK-8 assay of cell viability (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. B: CCK-8 assay of cell survival rate (n = 5). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. C: Flow cytometric analysis of cell cycle distribution. Data were presented as mean ± SD (n = 3). Student’s t-test. ****P < 0.0001 si-EGR1vs. LV-EGR1. D: Wound healing assay of PSC migratory ability (original magnification, ×40) (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. E: Transwell migration assay of PSCs (original magnification, ×100) (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. F: Matrigel invasion assay of PSCs (original magnification, ×100) (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 Subsequent functional assays evaluating cell motility demonstrated that LV-EGR1/GLUL PSCs significantly impaired the migratory and invasive capacities of Pan02 cells. Specifically, the scratch wound closure rate of Pan02 cells was reduced in the presence of LV-EGR1/GLUL PSCs compared to si-EGR1/GLUL PSCs (Fig. 4). Transwell invasion assays further confirmed a decrease in the number of invasive Pan02 cells when co-cultured with LV-EGR1/GLUL PSCs. Conversely, the outcomes can be inverted in si-EGR1/GLUL PSCs (Fig. 4). To investigate the key mediator of this inhibitory effect, we directly overexpressed GLUL in the culture medium and supplemented it with a corresponding concentration of glutamine. It was found that, compared to GLUL knockdown and low glutamine concentration, this treatment significantly inhibited the proliferation of Pan02 cells (Fig. 4). Taken together, these results demonstrate that LV-EGR1/GLUL PSCs suppress the proliferation, migration, and invasion of pancreatic cancer cells via paracrine glutamine, supporting a functional role of EGR1/GLUL in modulating PSC-mediated regulation of tumor cell behavior. LV-EGR1/GLUL PSCs inhibits pancreatic cancer cell macropinocytosis and triggers metabolic adaptation Given the role of macropinocytosis in cancer cell nutrient uptake, we assessed macropinocytic activity in pancreatic cancer cells exposed to PSC CM. Live-cell imaging with fluorescent dextran showed that CM from LV-EGR1/GLUL PSCs significantly reduced macropinocytic activity in Pan02 cells compared to control CM (Fig. 5). As a key pathway for cancer cells to uptake extracellular macromolecules in nutrient deficient microenvironments, the inhibition of its activity directly reduces the ability of cancer cells to obtain nutrients through this method. Concurrently, LV-EGR1/GLUL PSCs significantly suppressed autophagic flux in Pan02 cells, evidenced by reduced LC3-II/I conversion and accumulated P62 protein (p < 0.05) (Fig. 5). Autophagy is another important nutritional emergency mechanism of cancer cells, which can provide energy and raw materials by degrading their own organelles and proteins. Both macropinocytosis and autophagy, the two main nutrient acquisition pathways, were simultaneously inhibited, indicating that cancer cells are facing more severe nutritional pressure. A: Untargeted metabolomics (LC-QTOF-MS) analysis of co-cultured Pan02 cells. (A Pan02; B PSC+Pan02; A3 LV-GLUL NC PSC+Pan02; B3 LV-GLUL PSC+Pan02; C3 si-GLUL NC PSC+Pan02; D3 siGLUL PSC+Pan02; E3 LV-EGR1 NC PSC+Pan02; F3 LV-EGR1 PSC+Pan02; G3 si-EGR1 NC PSC+Pan02; H3 si-EGR1 PSC+Pan02). B: Macropinocytosis assay and quantitive analysis of macropinocytosis-positive cell in pancreatic cancer cells. Immunofluorescence images were representative of three independent experiments (original magnification, ×600) (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. C: Metabolic analysis of ATP in Pan02 cells (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. D: Western blot analysis and quantitative analysis of p62 and LC3Ⅱ/LC3Ⅰin Pan02. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 Untargeted metabolomics (LC-QTOF-MS) of co-cultured Pan02 cells further exposed multi-pathway metabolic reprogramming induced by LV-EGR1/GLUL PSCs. KEGG pathway enrichment analysis revealed that metabolic pathways was the most significantly enriched category in Pan02 cells after co-culture, and the downregulated metabolites significantly outnumbered upregulated ones. Specifically, the most notably enrichment in Pan02 cells co-cultured with LV-GLUL PSCs was identified as dysregulation of purine metabolism, alanine, aspartate and glutamate metabolism, the core pathway of energy metabolism in pan02 cells. This metabolic rewiring and metabolite depletion culminated in bioenergetic crisis, as Pan02 cells co-cultured with LV-EGR1/GLUL PSCs exhibited lower intracellular ATP levels, while si-EGR1/GLUL group showed opposite trend. However, when the rescue experiment was conducted using LV-EGR1/GLUL, the ATP level in the si-EGR1/GLUL group also decreased (Fig. 5). Collectively, these data demonstrate that EGR1-driven GLUL activation in PSCs orchestrates a nutrient-starvation paradigm-simultaneously impairing extracellular nutrient scavenging (macropinocytosis) and intracellular recycling (autophagy)-thereby depleting energy reserves and attenuating metastatic fitness in pancreatic cancer cells. LV-EGR1/GLUL PSCs modulates pancreatic cancer vasculogenic mimicry (VM) and liver metastasis Finally, we investigated the impact of the EGR1-GLUL-regulated PSCs on pancreatic cancer progression, namely vasculogenic mimicry and hematogenous metastasis. In co-culture systems, conditioned media from LV-GLUL PSCs significantly enhanced the number of branch junction compared to controls in pancreatic cancer cells, and upregulated the expression of vasculogenic mimicry markers VE-cadherin (Fig. 6), speculating a potential tendency of pancreatic cancer cells co-cultured with LV-EGR1/GLUL PSCs to form VM. Although VM structures appeared more prominent by LV-EGR1/GLUL PSCs, the results of western blot showed that LV-EGR1/GLUL PSCs could increase the expression of E-Cadherin, with the decreased expression of angiogenesis-related key factors MMP2, MMP9, VEGF expression in Pan02 cells (Fig. 6). However, those may suggest that Pan02 cells co-cultured with LV-EGR1/GLUL PSCs have a tendency to inhibit angiogenesis, which is contrary to the above results related to VM. A: Vasculogenic mimicry (VM) assay was conducted in mice liver metastasis model (Pan02; PSC+Pan02; LV-GLUL NC PSC+Pan02; LV-GLUL PSC+Pan02; si-GLUL NC PSC+Pan02; siGLUL PSC+Pan02; LV-EGR1 NC PSC+Pan02; LV-EGR1 PSC+Pan02; si-EGR1 NC PSC+Pan02; si-EGR1 PSC+Pan02). VM structures were imaged under a microscope (original magnification, ×40), and the number of branch points was counted using ImageJ software (original magnification, ×100) (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. B: Western blot analysis and quantitative analysis of VE cadherin, in Pan02. Data were normalized to GAPDH. Data were presented as mean ± SD (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. C: Western blot analysis and quantitative analysis of MMP2 in Pan02. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. D: Western blot analysis and quantitative analysis of E-Cadherin, MMP2, MMP9 and VEGF in Pan02. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. E: Western blot analysis and quantitative analysis of E-Cadherin, MMP9 and VEGF in Pan02. Data were normalized to GAPDH (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. F: Metabolic analysis of ATP. Data were presented as mean ± SD (n = 3). Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 To further determine whether LV-EGR1/GLUL PSCs influenced liver metastasis in pancreatic cancer, we labeled Pan02 cells with luciferase and respectively mixed them with several groups of si/LV-EGR1 or si/LV-GLUL PSCs in the ratio 1:1 to maintain the cell suspensions. The liver metastasis model of C57 / BL6 mice was constructed by splenic injection. In vivo imaging systems, fluorescence microscope, and histological dyed technique were applied to detect the volumetric and malignant change of metastatic tumor. The results showed that EGR1-overexpressing or GLUL-overexpressing PSCs significantly reduced the hepatic metastatic tumor burden, intratumoral ATP levels, and malignancy grade compared to their respective empty vector control groups. Microvessel density (MVD) was also significantly decreased (Figs. 6 and 7). Conversely, EGR1 knockdown or GLUL knockdown PSCs showed a trend toward promoting hepatic metastasis compared to their respective empty vector control groups (Figs. 6 and 7). Key rescue experiments further confirmed that GLUL knockdown on the basis of EGR1 overexpression and GLUL overexpression under EGR1 knockdown conditions reversed the tumor-suppressive effects observed in LV-EGR1 and LV-GLUL PSCs group respectively. This demonstrates that EGR1 indeed functions through the regulation of GLUL (Figs. 6 and 7). This figure presents the data of the experimental groups vertically (n = 3). From top to bottom, they are groups A to K (definitions are provided below). For each experimental group, from left to right, the in situ and ex vivo liver observation images, representative in vivo fluorescence imaging, and H&E staining of liver metastases are displayed. The lower figure shows the quantitative statistics of the number of visible metastatic nodules on the liver surface and MVD of liver metastases. The experimental groups are defined as follows: A: LV-luc marked Pan02 cells. B: LV-luc marked Pan02 cells + LV-GLUL(overexpressed) PSCs. C: LV-luc marked Pan02 cells + si-GLUL(knocked down) PSCs. D: LV-luc marked Pan02 cells + LV-GLUL-NC (control) PSCs. E: LV-luc marked Pan02 cells + si-GLUL-NC (control) PSCs. F: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs G: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs. H: LV-luc marked Pan02 cells + LV-EGR1-NC (control) PSCs. I: LV-luc marked Pan02 cells + si-EGR1-NC (control) PSCs. J: LV-luc marked Pan02 cells + LV-EGR1(overexpressed) PSCs + si-GLUL(knocked down) PSCs. K: LV-luc marked Pan02 cells + si-EGR1(knocked down) PSCs + LV-GLUL(overexpressed) PSCs. Student’s t-test. All Data were presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 The results released that compared with si-EGR1/GLUL group, LV-EGR1/GLUL group showed significantly-reduced liver metastasis, low ATP level and malignant degree of liver metastases, indicating that EGR1-GLUL-mediated PSCs suppresses the liver metastasis in pancreatic cancer. Discussion Pancreatic cancer remains one of the most lethal malignancies, with metastasis being the primary driver of mortality [15]. The tumor microenvironment (TME), particularly the crosstalk between PSCs and cancer cells, plays a pivotal role in this process [2, 16]. In this study, we identified a novel regulatory axis—EGR1-driven GLUL/mTOR—that suppresses PSC activation, inhibits macropinocytosis-mediated nutrient acquisition in pancreatic cancer cells, and may modulate vasculogenic mimicry and restrains liver metastasis. These findings unravel a previously unrecognized mechanism governing stromal-tumor interactions and provide new insights into therapeutic targeting of pancreatic cancer progression. Our results establish EGR1 as a key transcriptional regulator of PSC phenotype. Single-cell sequencing analysis first highlighted EGR1 as a top enriched transcription factor in cancer-associated PSCs, with strong co-expression patterns with GLUL and mTOR pathway genes. Mechanistic studies confirmed that EGR1 directly binds to the GLUL promoter, driving its transcription and subsequent activation of mTOR signaling in PSCs. This axis potently suppresses PSC activation, as evidenced by reduced expression of α-SMA. This finding contradicts some reports indicating that EGR1 promotes tumorigenesis in other contexts [17, 18], but aligns with emerging evidence that the role of EGR1 is highly dependent on cell type and pathological context. Its functional diversity extends far beyond the field of oncology, often resulting in seemingly contradictory effects [19]. For instance, in liver cancer, it promotes proliferation via the Wnt/β-catenin pathway [20], whereas in drug-induced liver injury, it exerts a protective effect by upregulating fatty acid oxidation genes [21]. However, in hepatic stellate cells, TGFβ1-induced EGR1 can promote fibrosis [22]. In this context, the inhibitory effect on activation mediated by the EGR1–GLUL/mTOR axis in PSCs identified in this study is significant, which demonstrates that even among stellate cells, the functional output of EGR1 can vary depending on the organ microenvironment and upstream signaling stimuli. The unique cytokine or metabolic stress in pancreatic cancer leads to an EGR1‑driven downstream program in PSCs characterized by increased GLUL transcription, which subsequently exerts its effects via the mTOR pathway, rather than being linked to EMT or classical fibrogenic genes. In contrast, the downstream program in pancreatic cancer cells involves pro‑invasive factors such as SNAI2 or ZEB2 [8, 17], while in hepatic stellate cells it involves pro‑fibrotic targets. Therefore, the tumor‑suppressive function exhibited by EGR1 in PSCs does not contradict other studies, but serves as another example of its context‑dependent functionality, emphasizing that targeting EGR1 or its downstream specific programs must fully consider the cellular signaling and environmental context. Notably, identifying the upstream regulators of EGR1 expression in PSCs is of critical importance. This regulation may be associated with the SPHK1/S1P pathway modulated by damaged pancreatic cells [23], as well as the hypoxic microenvironment [24], which involved require further investigation and validation. Furthermore, following co-culture with LV-EGR1/GLUL PSCs, pancreatic cancer cells exhibited phenotypes including macropinocytosis inhibition, autophagy blockade, and ATP depletion. Among these, autophagy, as a key downstream biological process of the mTOR signaling pathway, can be directly regulated by the mTOR pathway through proteins such as LC3-II/I and P62 [25]. This study observed a reduction in LC3-II/I conversion and an accumulation of P62 protein, providing profound experimental evidence for the perturbation of mTOR signaling activity in cancer cells within the co-culture system. Moreover, this study has confirmed that LV-EGR1/GLUL PSCs can synthesize and secrete more glutamine into the microenvironment. According to previous research, glutamine can exert a preventive effect on colon cancer by regulating the mTOR signaling pathway [26]. This also suggests the possibility that LV-EGR1/GLUL PSCs may regulate the mTOR signaling activity of co-cultured pancreatic cancer cells, thereby inhibiting their malignant phenotypes, through the secreted glutamine. However, this study did not directly measure the phosphorylation level of mTOR in cancer cells, and further research is required to confirm this hypothesis. A key insight from our study is the role of macropinocytosis as a downstream effector of the EGR1-GLUL/mTOR axis in mediating PSC-cancer crosstalk. EGR1-overexpressing PSCs, via their suppressed activation, reduced macropinocytic activity in pancreatic cancer cells. Macropinocytosis has recently emerged as a vital nutrient acquisition pathway in nutrient-deprived TMEs, enabling cancer cells to scavenge extracellular macromolecules [27, 28]. Our data extend this understanding by linking stromal signaling (via PSCs) to cancer cell macropinocytosis. The observation that EGR1-GLUL/mTOR-mediated suppression of PSCs reduces cancer cell macropinocytosis identifies a novel “stromal check” on this adaptive metabolic process. This mechanism may explain why desmoplastic stroma, rich in activated PSCs, often correlates with increased cancer cell aggressiveness: activated PSCs may enhance macropinocytosis to fuel cancer cell growth, while EGR1-activated PSCs counteract this effect. Additionally, regarding the mechanism by which LV-EGR1/GLUL PSCs inhibit macropinocytosis in cancer cells, the Transwell and conditioned medium experiments in this study confirmed the dominant role of soluble paracrine factors by avoiding direct contact between cancer cells and PSCs. Functional experiments further identified glutamine as the core mediator. Specifically, the elevation of its secretion level mimics the inhibitory effect of PSCs on cancer cell behavior. This untangles a novel, metabolism-dependent mode of PSCs-cancer cell interaction, where PSCs with high EGR1 expression can inhibit micropinocytosis and malignant progression in cancer cells by secreting glutamine. Although our study clarified the central role of glutamine in this pivotal axis, the potential involvement of other paracrine factors closely linked to cancer cell metabolism (such as TGF-β or IL-6) in fine-tuning this process cannot be excluded [29, 30], which warrants further clarification in future research. Furthermore, whether direct cell-cell contact within the intact tumor microenvironment dynamically modifies or enhances this regulatory process remains an open question worthy of further investigation. Future studies should employ co-culture models and conduct further screening of differentially expressed factors to elucidate the detailed mechanisms involved in this regulatory pathway. And glutamine, in turn, triggers downstream effects. As macropinocytosis is particularly vital within the nutrient-scarce tumor microenvironment [31], its inhibition forces cancer cells into a state of metabolic stress, compelling extensive reprogramming of their internal metabolic network. In this study, this reprogramming manifests as significant dysregulation of core pathways, including purine metabolism and the metabolism of alanine, aspartate, and glutamate. These pathways are closely linked to energy synthesis and amino acid interconversion; their disruption inevitably compromises cellular energy homeostasis [32]. Consequently, when exogenous nutrient uptake is blocked concurrently with endogenous metabolic imbalance, a marked depletion of intracellular ATP levels ensues. This depletion provides a mechanistic explanation for the subsequently observed phenomena: proliferation arrest, diminished migratory and invasive capabilities, and restricted metastasis. This insight offers a novel theoretical framework for understanding the complex stromal-tumor interactions in pancreatic cancer (Fig. 8). Schematic diagram illustrating the interaction between key metabolites and macropinocytosis in Pan02 cells As the ability of cancer cells to form vessel-like structures independent of endothelial cells, VM is a critical driver of hematogenous metastasis in pancreatic cancer [33, 34]. Our ex vivo and in vivo experimental results present a seemingly paradoxical phenomenon. On the one hand, in the Matrigel tube formation assay, conditioned medium from LV-EGR1/GLUL PSCs significantly increased the number of branch connections in pancreatic cancer cells and upregulated the expression of the VM marker VE-cadherin, suggesting that EGR1-positive PSCs may induce tumor cells to initiate a VM program. On the other hand, LV-EGR1/GLUL cancer-associated fibroblasts significantly upregulated the expression of E-cadherin in pancreatic cancer cells. Although the role of E-cadherin in tumors is context-dependent, studies have shown that its upregulation can exert anti-angiogenic effects [35]. Concurrently, this axis significantly downregulated the expression of pro-angiogenic factors MMP2, MMP9, and VEGF, which is consistent with the reduction in MVD observed in liver metastases, indicating suppressed tumor angiogenesis. This coexistence of ex vivo VM-promoting effects and in vivo metastasis-inhibiting effects suggests that EGR1-positive PSCs may induce a structural rather than functional form of VM. Specifically, we hypothesize that the EGR1/GLUL axis impairs cancer cell energy metabolism by reducing macropinocytosis, which may lead tumor cells under stress conditions to attempt initiating a VM program to maintain nutrient supply and counteract the effects of reduced microangiogenesis [36]. This is reflected in the increased morphogenesis of VM-like structures in the results. However, potentially due to factors such as metabolic disorders and energy insufficiency, truly functional VM capable of conducting blood flow and effectively transporting tumor cells may not be established, ultimately resulting in metastasis inhibition [37]. This is consistent with the observed downregulation of angiogenic factors, decreased MVD in liver metastases, and reduced ATP levels in the study. Furthermore, this outcome might also be attributed to the functional differences of PSCs between in vivo and in vitro conditions. In ex vivo experiments, the simplified environment lacking hemodynamic shear stress, immune cell infiltration, and extracellular matrix tension may allow PSCs alone to drive VM morphogenesis. In contrast, in vivo, metastasis inhibition likely arises from the combined actions of the complex in vivo microenvironment. In future studies, we will supplement in vivo functional vascular perfusion assays [38] and stromal component analyses to clarify whether VM structures with actual blood perfusion function are truly formed and to further dissect the multiple mechanisms by which PSCs regulate the tumor microenvironment in vivo. Nevertheless, the reduction in in vivo liver metastatic burden is consistent with E-cadherin upregulation, downregulation of pro-angiogenic factors, and attenuated metabolism, which already demonstrates the potential of this axis as a metastasis-specific therapeutic target. While our study identifies a novel regulatory axis, several questions remain. The specific paracrine factors secreted by EGR1-GLUL/mTOR-activated PSCs that inhibit macropinocytosis require further characterization. Cytokine profiling and receptor mapping in cancer cells could pinpoint key mediators. And the key factors in the microenvironment as well as the upstream signaling pathways that determine the expression of EGR1 in PSCs need further investigation. Moreover, translating these findings to clinical settings will require validation in patient-derived xenografts and assessment of EGR1-targeting strategies (e.g., small molecules or gene therapy) for safety and efficacy. Conclusions Our study uncovers a novel EGR1-driven GLUL/mTOR axis in PSCs that suppresses pancreatic cancer progression through multiple interconnected mechanisms: restraining PSC activation, inhibiting cancer cell macropinocytosis and energy metabolism, and inhibiting liver metastasis. These findings highlight the complexity of stromal-tumor crosstalk and identify new therapeutic targets for combating pancreatic cancer metastasis. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Supplementary Material 6 Supplementary Material 7 Supplementary Material 8 Supplementary Material 9 Supplementary Material 10 Supplementary Material 11 Acknowledgements All authors have no conficts of interest to disclose. Abbreviations Chromatin immunoprecipitation Extracellular matrix Early growth response 1 Electrophoretic mobility shift assay Gene set enrichment analysis Glutamine synthetase Pancreatic stellate cell Tumor microenvironment Author contributions RX, JZ conceived and designed the experiments; JW, XRS, YW L and YX Lperformed the experiments, interpreted the data and prepared the figures; RX, YWL and YX Lwrote the manuscript and analyzed the data. All authors read and approved the final manuscript. Funding This work was supported by grants from National Natural Science Foundation of China (No. 82002461), the Beijing Hospitals Authority Youth Programme (code: QML20231104), Fund for Fostering Young Scholars of Peking University Health Science Center (BMU2021PY010), Medjaden Academy and Research Foundation for Young Scientists (MJR20211110) and Science Foundation of Peking University Cancer Hospital (No. BJCH2026GG03). Data availability The single-cell sequencing data (accession number: GSE212966) related with pancreatic cancer can be obtained from NCBI GEO database. Declarations Ethics approval and consent to participate Not applicable. Consent for publication We have reviewed the final version of the manuscript and approve it for publication. To the best of our knowledge and belief, this manuscript has not been published in whole or in part nor is it being considered for publication elsewhere. Competing interests All authors have no conficts of interest to disclose. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Ran Xue, Jun Zhou, Yawei Liu and Yuxi Liu contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Supplementary Material 6 Supplementary Material 7 Supplementary Material 8 Supplementary Material 9 Supplementary Material 10 Supplementary Material 11 Data Availability Statement The single-cell sequencing data (accession number: GSE212966) related with pancreatic cancer can be obtained from NCBI GEO database.