LIPH contributes to glycolytic phenotype in pancreatic ductal adenocarcinoma by activating LPA/LPAR axis and maintaining ALDOA stability

Human PDAC cell lines AsPC-1, BxPC-3, CFPAC-1, PANC-1, PNAC-1-luc, MIA PaCa-2, 8988t, 8988s, mouse PDAC cell lines Panc02, and KPC1199, and human normal pancreatic duct cells (HPNE) were all preserved at the Pancreatic Disease Center, Ruijin Hospital, School of Medicine, Shanghai Jiao Tong University. Doxycycline-inducible KRAS^G12D BxPC-3 cells were gifted by Pro. Sun (Shanghai Cancer Institute, Renji Hospital, School of Medicine, Shanghai Jiao Tong University). All cells were grown in the suggested culture medium, supplemented with 10% fetal bovine serum (FBS, Gibco), 100 U/mL penicillin and 100 μg/mL streptomycin at 37 ℃ with 5% CO₂. A final concentration of 2.5% horse serum was added to MIA PaCa-2 cultured cells. The base medium for AsPC-1 and BxPC-3 cells is RPMI-1640 Medium (Meilunbio, China); for CFPAC-1, Iscove's Modified Dulbecco’s Medium (IMDM) (BIOAGRIO, USA); for PANC-1, PANC-1-luc, MIA PaCa-2, 8988t, 8988s, Panc02, and KPC1199; and for HPNE, Dulbecco's Modified Eagle's Medium (DMEM) (Meilunbio, China). All cells were checked routinely for mycoplasma contamination. U0126, Ki16425, aldometanib, gemcitabine, MG132 and cycloheximide were purchased from MCE (Shanghai, China). LPA and PA were purchased from Sigma-Aldrich. Doxycycline was purchased from TargetMol (Shanghai, China), and 1 μg/mL doxycycline was added to induce the expression of KRAS^G12D.

Plasmid transfections were performed using Lipofectamine^®2000 (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. Specific luciferase plasmids, shRNAs, vector controls, and full-length cDNA encoding human LIPH were synthesized by BioGene (Shanghai, China). For lentivirus packaging, 293 T cells were transfected using lipofectamine 2000 with 2 μg of each plasmid and 2 μg of helper plasmid (1 μg pMD2G and 1 μg of psPAX2). Viral Supplements were collected 48 h later and filtered with a 0.22 μm filter. PDAC cells, 3 × 10⁵, were infected with viral supernatants and 10 μg/mL polybrene (Yeasen, China) for 6 h. Transduced cells were selected with 10 μg/mL puromycin (Meilunbio; China) to produce stable cell lines. Further information on plasmids is described in the Extended Information (Additional file 8).

For the cell proliferation assay, cells were seeded in 96­well plates at a density of 1 × 10³ cells/well. Cell Counting Kit-8 (CCK-8) (Meilunbio, China) was used according to the manufacturer’s instructions, and the optical density at 450 nm was measured (Biotek, USA). To exclude the effect of LPA in fetal bovine serum, one day after seeding, cells were changed to DMEM, and viability was measured at 72 h. To evaluate the effects of PA and LPA, PDAC cells were seeded at 3 × 10³ cells/well. One day after seeding, cells were changed to DMEM containing 0.5% FAF-BSA (starvation medium) (Beyotime, China) and incubated for 24 h. After starvation, cells were incubated in 0.5% FAF-BSA, 0.5% FAF-BSA along with 10 µM PA or 0.5% FAF-BSA along with 10 µM LPA. Cell viability was assayed at the indicated time points. For the colony formation assay, cells were seeded in 6­well plates at a density of 1 × 10³ cells/well for 10 days, followed by staining of crystal violet (Beyotime, China) dye for 20 min.

For the apoptosis assay, PC cells were seeded in 6-well plates at a density of 1 × 10⁵. After starvation or incubation with gemcitabine (1 μg/mL) for three days, cells were collected, resuspended, and incubated with fluorescein isothiocyanate-conjugated annexin V and propidium iodide (Vazyme, China) for 30 min in the dark at room temperature, and analyzed by flow cytometry (CytoFLEX5, Beckman).

For the subcutaneous xenograft model, PANC-1 (2 × 10⁶ cells per mouse), CFPAC-1 (2 × 10⁶ cells per mouse), or MIA PaCa-2 (2 × 10⁶ cells per mouse) were injected into the right or left dorsal flanks of BALB/c nude mice. The tumor volumes were measured three times per week and calculated according to the formula V(volume) = 0.5 × L(length) × W(width)². After two weeks, the mice were euthanized and the tumors were collected for histological and immunohistochemical (IHC) analysis. To perform 18F-fluorodeoxyglucose (18F-FDG) PET/CT, the mice were anesthetized with sodium pentobarbital (10 mg/kg) and intravenous 18F-FDG was administered. One hour later, PET/CT images were obtained using an Inveon Animal PET/CT scanner combined with three-dimensional ordered-subset expectation maximization (OSEM3D)/maximum algorithms. The SUV_max within the tumor was used for further analysis.

An orthotopic xenograft model was constructed by injecting human luciferase-tagged PANC-1 cells (PANC1-Luc; 1 × 10⁶ cells per mouse). After four weeks, the mice were intraperitoneally injected with d-luciferin potassium (150 mg/kg) and analyzed using an IVIS Spectrum Imaging System (Tanon, China). Imaging was performed weekly to closely monitor tumor proliferation and drug efficiency. Tumors and adjacent areas were collected at the end of the experiment.

To construct a patient-derived xenograft (PDX), PDAC tissues from two individuals, which had been checked and confirmed by two pathologists, were used. Briefly, the surgical specimen from pancreatectomy was washed with P/S solutions, carefully cut into 1 mm³, stored at − 80 °C with freezing medium (10% DMSO, 90% FBS) and buried in the right or left dorsal flank of NGS mice within one week to generate first generation (F1) tumor. When the tumor reached 800 mm³, the mice were sacrificed, and the tumor was cut into 1 mm³ for further transplantation. The relevant drugs were administered to the tumors as soon as their volume reached approximately 100–150 mm³ (6–8 weeks after transplantation). The control group received physiological saline solution via intraperitoneal injection (i.p.), whereas the other four groups were treated with gemcitabine (50 mg/kg i.p., twice a week), ki16425 (4 mg/kg i.p., qod), or aldometanib (5 mg/kg, i.g., qod). The mice were sacrificed after two weeks of treatment.

Metabolic assays for the extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) were performed using a Seahorse XF96 Flux Analyzer (Agilent, USA) according to the manufacturer’s instructions. Briefly, 2 × 10⁵ cells were seeded in XF 96-well plates. One day after seeding, cells were changed to DMEM containing 0.5% FAF-BSA and incubated for 24 h. After starvation, cells were incubated in 0.5% FAF-BSA, 0.5% FAF-BSA along with 10 µM PA or 0.5% FAF-BSA along with 10 µM LPA for 6 h. One hour before testing, the medium was replaced with the assay medium. For the ECAR test, 10 mmol/L glucose, 1 mmol/L oligomycin, and 50 mmol/L 2-deoxyglucose (2-DG) were injected into each well. For OCR test, 1 μmol/L oligomycin, 2 μmol/L FCCP, 0.5 μmol/L rotenone, and 0.5 mmol/L actinomycin A were added to the wells. For ATP production assays, cells were grown overnight in 6-well plate culture dishes, followed by treatment with the indicated reagents. The cells were lysed and clarified by centrifugation, and the supernatants were extracted for BCA and ATP measurements (Beyotime, China).

Hematoxylin and eosin (H&E) staining was performed routinely [21] and observed with GRUNDIUM OCUS^® (Grundium, Finland). For IHC, paraffin sections were dewaxed with an ethanol gradient, steam-heated for antigen retrieval in citrate-based buffer for 20 min, blocked with 3% BSA diluted in Tris-buffered saline plus 0.1% Tween 20, and stained with primary and appropriate secondary antibodies to detect the corresponding proteins. The tissue staining was scored negative when < 5% tumor cells showed expression. Positive scores (+–+++) were based on the percentage of tumor cells and staining intensity within the tumor sample. Information on the antibodies is provided in the Extended Information section.

To assess the distribution of LIPH, ALDOA, and YAP1, cells (CFPAC-1, PANC-1, and MIA PaCa-2) were seeded on no. 1.5 cover glass (Corning, USA). Cells were fixed with 4% paraformaldehyde (Servicebio, China) for 10 min, washed, and permeabilized with 0.25% Triton-X100 (Sigma, USA) for 5 min. Cells were washed with ice-cold PBS, blocked and incubated for 30 min in blocking solution (PBS, 1% BSA, 0.1% Tween 20, 22.52 mg/mL glycine), followed by incubation with primary antibodies in dilution solution (PBS, 1% BSA, 0.1% Tween 20) for 1 h at 37 °C. After three washes of 5 min in PBS, the cells were incubated with goat anti-rabbit antibody or goat anti-mouse antibody for 30 min at room temperature. Cells were washed three times for 5 min in PBS, with the first wash containing 0.1 μg/mL DAPI (Thermo Scientific; USA). IF staining was performed on PDAC tissues from archived formalin-fixed paraffin-embedded tumors according to standard protocols [22]. Samples were observed under a confocal laser scanning microscope (Zeiss LSM900, Germany). Information on the antibodies is provided in the “Extended information”.

Cell lysates were prepared using an ice-cold RIPA lysis buffer (Solarbio, China) supplemented with a phosphatase inhibitor cocktail (NCM; China). After total protein was normalized by BCA assay (Invitrogen, USA), protein samples were separated by 8–10% SDS-PAGE gel electrophoresis and transferred onto PVDF membranes. After blocking with 5% BSA (Proliant, New Zealand) diluted in Tris-buffered saline plus 0.1% Tween 20 for 1 h at room temperature, membranes were incubated overnight with primary antibody and 1 h with secondary antibody (Proteintech, 1:10,000). Enhanced chemiluminescence (ECL) was performed using an ECL kit (Meilunbio, China) and visualized using the Tanon system (Shanghai, China). Information on the antibodies is provided in the Additional files. Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, China) was used to extract the nuclear and cytoplasmic proteins.

For immunoprecipitation, protein lysates were prepared in ice-cold IGEPAL^® CA-630 buffer (Beyotime; China) supplemented with an inhibitor cocktail (NCM, China). The cells were mechanically ruptured at least five times using a cell scraper (Corning, USA). Lysates were then centrifuged at 12000 rpm/min, 4 °C for 15 min, and the supernatant was collected. The samples were incubated with protein A/G beads (Beyotime; China) for 1 h at 4 °C to reduce nonspecific binding. The samples were centrifuged, and the supernatant was collected for immunoprecipitation. The supernatant was incubated with FLAG antibody (14793, Cell Signaling Technology), ALDOA antibody (11217-1-AP, Proteintech), and protein A/G beads overnight at 4 °C with gentle rocking. After that, the samples were centrifuged at 4 °C and the beads were washed with lysis buffer three times. The beads were reconstituted in Non-Reducing Lane Marker Sample Buffer (39001, Thermo Scientific) and boiled for 5 min before immunoblotting.

Total RNA was extracted using an RNA Extraction Kit (Accurate Biology, China), and reverse transcription was performed using an Evo M-MLV RT-PCR Kit (Accurate Biology, China) according to the manufacturer’s instructions. For quantitative PCR analysis, double-stranded cDNA was amplified using a SYBR Green PCR Kit (Accurate Biology, China) and detected using qTOWER384G (Analytik Jena, Germany). The primers used in this study are listed in the Extended Information.

For RNA-seq, total RNAs were isolated using an RNA Extraction Kit (Accurate Biology, China) following the manufacturer’s instructions and sent to Xurangene Co., Ltd. Clustering and sequencing were then performed. The mapping results were processed using custom scripts, and all samples were prepared in triplicate. The detailed methodology for transcript assembly and statistical analysis has been described in a previous study [23].

Gel fragments were cut from SDS PAGE. The in-gel proteins were reduced, alkylated, and digested with 12.5 ng/μL trypsin in 25 mM NH₄HCO₃ overnight. The peptides were then collected with solution (60% ACN /0.1% TFA), pooled, and dried entirely in a vacuum centrifuge.

LC–MS analysis was carried out on a Q-Exactive mass spectrometer (Thermo Fisher Scientific, USA) in positive ion mode, coupled to an Easy nLC (Thermo Fisher Scientific, USA) for 60 min. Survey scans (300–1800 m/z) were taken at a resolution of 70,000 at m/z 100, the AGC target was 1e6, the maximum inject time was 50 ms, the number of scan ranges was 1, and the dynamic exclusion duration was 30 s. The peptide recognition mode was used to operate the instrument. MS data were obtained using the data-dependent Top20 method. The resolution of the HCD spectra was 17,500 at m/z 100, AGC target was 1e5, isolation width was 1.5 m/z, microscans were 1, and the maximum injection time was 50 ms.

For data analysis, MS/MS spectra were searched using Proteome Discoverer 1.4 against uniprot_Homo_sapiens_203800_20220104. Protein identification options: Enzyme = trypsin; maximum missed cleavage = 2; fixed modification = carbamidomethyl (C); dynamic modification = oxidation (M); filter by peptide confidence = high.

ChIP experiments were conducted according to the standard protocols [21]. In brief, MIA PaCa-2 cells stably transfected with NC-vector or OE-LIPH plasmid were cross-linked with 1% formaldehyde for 10 min at room temperature and then quenched with glycine (Thermo Fisher Scientific, USA). Appropriate DNA fragments, which were then incubated with YAP1 antibodies, were obtained by carefully sonicating the nuclear lysates. Normal rabbit IgG was used as the negative control. The primers used for ChIP are listed in the Extended Information.

To conduct docking [24], proteins with full-length sequences were collected from UniProt (https://​www.​uniprot.​org/​). Input Receptor Molecule: ALDAO (P04075). Input ligands: LIPH (Q8WWY8) or UBE2O (Q9C0C9).

Protein–protein docking was based on a hybrid algorithm of template-based modeling and ab initio-free docking (using a hybrid docking strategy). The metrics of ALDAO + UBE2O complex (Docking Score: − 206.85; Confidence Score:0.7571; Ligand rmsd (Å):78.10). The metrics of ALDAO + LIPH complex (Docking Score: − 255.10; Confidence Score:0.8911; Ligand rmsd (Å):61.20).

Superposition of the structures was performed using CCP4. Visualizations of the atomic and cartoon/surface models were performed using PyMOL (PyMOL Molecular Graphics System, version 2.0; Schrödinger, LLC).

The Gene Expression Omnibus (GEO) datasets GSE15471, GSE16515, GSE62452, and GSE71729 were used to screen potential SPs. Microarray data were normalized and Log2-transformed, and Linear Models for Microarray Data (Limma, v3.17) were used for differential analyses. For TCGA and GTEx data, gene expression data were obtained from the data hubs at UCSC Xena (https://​xenabrowser.​net/​hub/​). The scRNA-seq data were downloaded from NGDC (CRA001160) and reanalyzed using Seurat2.0 (http://​satijalab.​org/​seurat/​) and Monocle2 (http://​cole-trapnell-lab.​github.​io/​monocle-release). Gene set enrichment analysis (GSEA, v4.3.2) was performed on the Broad Institute Platform and Gene set variation analysis (GSVA, v3.17) analysis was performed by “GSVA” R package. The association of LIPH mRNA with clinical prognosis was determined using GEPIA [25] and Kaplan–Meier (http://​www.​kmplot.​com/​). Survival Statistical significance (P value) was set at 0.05.

Studies using human tissues were reviewed and approved by the Committee for Ethical Review of Research Involving Human Subjects at the Ruijin Hospital. All patients provided written informed consent for the use of their clinical specimens for medical research. Animal experiments were approved by the review board on the use of living animals at Ruijin Hospital (Shanghai, China). The mice were housed in a pathogen-free laboratory animal unit at PNK (Shanghai, China).

Data are presented as mean ± standard deviation. Student’s t-test was used to compare differences between two groups. One-way ANOVA was performed when multiple conditions were compared for each variable. Two-way ANOVA was used when multiple groups were compared for more than one variable. Survival was analyzed using the log-rank test. All statistical analyses were performed using GraphPad Prism 8.0, Excel and SPSS v23. Differences were considered statistically significant at P < 0.05.

To identify the potential oncogenes among SPs that play a pivotal role in supporting PDAC with limited vascularization (Fig. 1A), we first analyzed the expression of SPs that were expressed at significantly higher levels in PDAC tissues than adjacent areas in three GEO datasets [26]: GSE15471, GSE16515, and GSE62452 (Cross1, 108 genes) (Additional file 1: Fig. S1A). Among 108 genes, only 17 SPs were upregulated in PDAC compared to multiple non-tumor tissues in GSE71729 (Additional file 1: Fig. S1B). We then investigated the clinical prognosis of these SPs using the PAAD-TCGA database and found that 10 genes were positively correlated with poor prognosis of PDAC (Additional file 7: Table S1). LIPH was mainly expressed in ductal cells and was most prominently upregulated during the transition from abnormal to malignant according to the scRNA-seq data (Additional file 1: Fig. S1C, D, E; Additional file 7: Table S1). Pan-cancer analysis using the expression of 28 tumors and the Cancer Cell Line Encyclopedia (CCLE) revealed that LIPH was highly expressed in PDAC tissues and cell lines originating from pancreatic cancer compared to other tumor types (Fig. 1B; Additional file 1: Fig. S1F). LIPH expression was significantly higher in PDAC tissues than in the adjacent normal tissues in the Ruijin cohort (Fig. 1D), PAAD-TCGA (Additional file 1: Fig. S1G), GEO dataset (GSE15471, GSE16515 and GSE62452, Additional file 1: Fig. S1H). Kaplan–Meier analysis revealed that high LIPH expression in cancer tissues was associated with poor prognosis in patients with PDAC (Fig. 1C, D; Additional file 1: Fig. S1I). Immunofluorescence (IF) staining confirmed the presence of LIPH in the plasma membrane and cytoplasm of PANC-1 and CFPAC-1 cells (Fig. 1E). We validated our results at the protein level by performing immunohistochemistry (IHC) staining in Ruijin tissue microarray (TMA) containing 99 pathologist-verified PDAC specimens with paired corresponding adjacent pancreatic tissues. The staining score was highly correlated with advanced tumor stage (Fig. 1F, G) and survival (Fig. 1H). We also examined LIPH expression patterns in Kras^G12D(KI/+)/Trp-53^R172H(KI/+)/Pdx1-Cre (TG/+) (KPC) mice (Fig. 1I). LIPH staining positively correlated with PDAC progression. In addition, univariate Cox regression analysis showed that LIPH expression, smoking, metastasis, American Joint Committee on Cancer (AJCC) stage, and recurrence were significantly associated with overall survival (Fig. 1J). Multivariate Cox regression analysis identified LIPH expression and tumor recurrence as independent predictors of overall survival in patients with PDAC (Additional file 1: Fig. S1J). Taken together, these data indicate that LIPH is an unfavorable prognostic factor for pancreatic cancer and is highly correlated with disease progression.

To study the effect of LIPH on PDAC, we examined the expression level of LIPH in eight pancreatic cancer cell lines at the mRNA and protein levels, with the highest expression in CFPAC-1 cells and the lowest expression observed in MIA PaCa-2 cells (Fig. 2A). PANC-1 and CFPAC-1 cells were stably transfected with control short hairpin RNA (shRNA) or shRNA (sh1-LIPH and sh3-LIPH) targeting LIPH, and mRNA and protein levels were compared (Fig. 2B). Additionally, the full-length human LIPH cDNA was cloned into lentiviral vectors (OE-LIPH) and transfected into MIA PaCa-2 cells (Additional file 2: Fig. S2B). Gene set variation analysis (GSVA) of the Ruijin cohort showed a significant positive correlation between high LIPH expression and glycolysis, hypoxia, and Hippo signaling pathways (Additional file 2: Fig. S2A). Therefore, we speculate that LIPH is mainly involved in the regulation of pancreatic cancer (PC) cell proliferation and self-renewal. We used the CCK8 assay, colony formation assay, and flow cytometry to measure the effect of LIPH on PC cell proliferation and growth. LIPH knockdown inhibited colony formation and proliferation of PANC-1 and CFPAC-1 cells (Fig. 2C, D). Conversely, LIPH overexpression enhanced colony formation and proliferation of MIA PaCa-2 cells (Additional file 2: Fig. S2C). Knockdown of LIPH expression effectively promoted apoptosis of cancer cells as well as upregulation of the apoptosis-associated protein BAX (Fig. 2E), increasing chemosensitivity to high gemcitabine concentrations (Additional file 2: Fig. S2E), whereas overexpression of LIPH reversed this effect (Additional file 2: Fig. S2D). Serum deprivation during apoptosis suggests a potential role for LIPH in maintaining cancer cell survival by catalyzing PA in a PDAC microenvironment with extremely poor vascularization. Knockdown of LIPH significantly constrained PANC-1 and CFPAC-1 proliferation ability in the serum-deprived medium compared with 10% fetal bovine serum (FBS)-complete medium. Conversely, LIPH overexpression in MIA PaCa-2 cells promoted their proliferation in a serum-deprived medium (Fig. 2F). To evaluate the exact role of LIPH in PDAC without the interference of LPA in FBS, fatty acid-free (FAF)-BSA with an additional supplement was used to maintain cell cultures. The results showed that LIPH knockdown constrained the utilization of exogenous PA, thus alleviating cell proliferation. The addition of LPA to the medium efficiently reversed the effects of LIPH knockdown (Fig. 2G; Additional file 2: Fig. S2F, G).

To test the tumorigenic ability of LIPH in vivo, PANC-1 and CFPAC-1 cells stably transfected with vector plasmid or sh1-LIPH, or MIA PaCa-2 cells stably transfected with vector and overexpression plasmids were inoculated into the right or left dorsal flank of BALB/c nude mice. Tumor size was carefully measured every three days after seven days injection. IHC was used to further validate transfection efficiency (Additional file 3: Fig. S3A). Tumors formed in the presence of sh1-LIPH were significantly smaller than those formed by the vector control (Fig. 3A). In contrast, LIPH overexpression enhanced tumorigenicity in vivo (Fig. 3A). To validate the ability of LIPH to induce cell proliferation and apoptosis, xenograft tumors were stained with H&E, ki67 and TUNEL. The expression of ki67 was decreased in sh1-LIPH pancreatic cancer tumors, and the apoptosis marker, TUNEL, was significantly increased (Fig. 3D, E). In addition to the downstream genes of the Hippo pathway, CTGF and glycolysis hub genes GLUT1 and HK2 were considerably downregulated in sh1-LIPH xenograft tumors, whereas LIPH overexpression reinstated their expression (Fig. 3D; Additional file 3: Fig. S3C). These results were highly consistent with our in vitro findings that LIPH knockdown inhibited the proliferation and promoted the apoptosis of pancreatic cancer cells in vitro.

To gain a comprehensive insight into the mechanism by which LIPH promoted pancreatic cancer cell growth, gene set enrichment analysis (GSEA) was performed in the PAAD-TCGA, GSE15471 and GSE28735. A highly positive correlation between LIPH expression and glycolysis was observed in high-LIPH PDAC tissue samples (Additional file 4: Fig. S4A). The global gene expression of sh1-LIPH PDAC cells compared with control cells after PA treatment was profiled and analyzed using GSEA and KEGG pathway analysis. The results indicated that the differentially expressed genes were mainly related to metabolic processes, including genes associated with glycolysis, hypoxia and MTORC1 signaling, suggesting that LIPH knockdown might restrain metabolic processes in PDAC (Fig. 4A). Expression of several critical genes in the glycolysis and hypoxia pathways, including GLUT1, HK2, IGFBP3, PDK3, EGLN3, and CTGF, was significantly downregulated in sh1-LIPH cells compared to the control cells in the presence of PA (Fig. 4B). Alterations in GLUT1 and HK2 expression were further confirmed using real-time qPCR or IHC (Figs. 4C, 3D). To further investigate glycolysis alternation with the manipulation of LIPH expression, glycolysis stress tests using extracellular acidification rate (ECAR) were performed to evaluate the glycolytic activities in PDAC cells. Exogenous PA effectively enhanced glycolytic capacity and glycolytic reserve in cells transfected with the control vector; however, it had limited influence on LIPH-knockdown cells, and the effect could be reversed by supplementing with exogenous LPA (Fig. 4D, Additional file 4: Fig. S4B). However, overexpression of LIPH in MIA PaCa-2 cells promoted the utilization of exogenous PA (Additional file 4: Fig. S4C). Measurement of basal and maximal respiration also suggested that LIPH overexpression induced a metabolic shift towards glycolysis (Additional file 4: Fig. S4D). LIPH knockdown led to lower ATP production, whereas LIPH overexpression reversed this effect (Fig. 4E).

18F-fluorodeoxyglucose (18F-FDG) PET/CT has been widely used for the visualization of the metabolic activity of tumor cells owing to their high glucose uptake ability. The maximum standardized uptake value (SUV_max) has been used as a reliable marker for the prognosis of PDAC [27]. To further examine the glycolysis uptake ability in vivo, 18F-FDG PET/CT was performed. As expected, LIPH knockdown significantly inhibited the glucose uptake ability of subcutaneous tumors (Fig. 4F). A systematic analysis of PET/CT images and corresponding paraffin sections from 24 patients at Ruijin Hospital revealed that patients with high LIPH (++/+++) levels exhibited higher SUV_max, which indicates higher glucose uptake ability, than patients with low LIPH (−/ +) levels (Fig. 4G). Taken together, these data indicated that LPA produced by LIPH promoted pancreatic cancer cell proliferation by enhancing glycolysis in vitro and in vivo.

Previous studies and our databases have confirmed that LPA-LPAR interaction is involved in PI3K/AKT signaling, hypoxia, and self-regeneration pathways, such as the Hippo pathway (Figs. 4A, 5A; Additional file 2: Fig. S2A) [17, 28]. Therefore, we first evaluated the mRNA expression of genes from the Hippo pathway and its downstream components. The expression of YAP/TAZ was relatively stable in each group, whereas a significant change of CTGF, a downstream gene of the Hippo pathway, was detected, suggesting higher utilization of exogenous PA cells in LIPH^high cells (Fig. 5B, C), which hinted a change in YAP/TAZ at the protein level, as well as an alteration in the localization of YAP/TAZ. To further understand the molecular effects arising from LIPH knockdown, two crucial transcription factors, YAP1 and HIF1A, were measured to explain the high ECAR levels and survival ability, even in a limited nutrient environment. The results showed that LIPH knockdown significantly promoted the phosphorylation of YAP1 at Ser¹²⁷, a key phosphorylation site that mediates the degradation of YAP1 by ubiquitin [29] and inhibits the transcription of its downstream gene CTGF (Fig. 5D). In contrast, in PANC-1 and CFPAC-1 cells, LIPH knockdown inhibited the utilization of exogenous PA, limited the phosphorylation of AKT, reduced the expression of HIF1A, and further downregulated key downstream glycolysis genes such as HK2 and GLUT1 (Fig. 5D). Notably, treatment with exogenous LPA reversed the effects of LIPH knockdown (Fig. 5D). The subcellular localization of YAP1 determines its biological function [30]. LIPH overexpression promoted the utilization of PA and further enhanced the nuclear localization of YAP1 even in serum-deprived cultures (Fig. 5E). Patients with high LIPH expression also exhibited significant YAP1 nuclear localization, whereas YAP1 tended to localize in the cytoplasm in patients with low LIPH expression (Fig. 5F). Overexpression of LIPH in MIA PaCa-2 cells also reduced the phosphorylation of YAP1 at Ser¹²⁷, promoted YAP1 nuclear localization, and conferred a high glycolysis phenotype to PDAC (Fig. 5G). These data indicate that LIPH enhances PDAC glycolysis and viability via the AKT/HIF1A pathway and inhibits the phosphorylation of YAP1.

To test whether the LIPH-LPA-LPAR axis affects the stability of YAP1, MIA PaCa-2 cells stably transfected with the control vector or overexpressing LIPH were treated with the protein synthesis inhibitor cycloheximide and YAP1 expression was monitored over time. Overexpression of LIPH stabilized YAP1 (Fig. 5H) and reduced proteasomal degradation of YAP1, which might be due to LPA-LPAR-mediated inhibition of LATS1/2 phosphorylation [17]. The subcellular localization of YAP1 is critical for its function, and phosphorylation of YAP1 promotes cytoplasmic retention of YAP1 and further degradation by ubiquitin. Cellular fractionation assays confirmed that the nuclear translocation of YAP1 was induced by LIPH overexpression (Fig. 5I). As an essential transcription factor, YAP1 sustains cellular metabolic reprogramming and self-regeneration by manipulating gene transcription [30]. GSEA showed a positive correlation between YAP1 and hypoxia and glycolysis pathways (Additional file 5: Fig. S5C). Given the significant enrichment in glycolysis and hypoxia pathway proteins based on our results and previous research [31‐33], we focused on HIF1A, a vital transcriptional factor mediating hypoxia endurance, and abnormally activating glycolysis in tumors to explain the highly activated aerobic metabolism in LIPH^high tumors. LIPH overexpression effectively promoted YAP1 nuclear localization as well as its combination with the HIF1A transcription start site (Fig. 5I). Finally, we used ki16425, an LPAR1/3 inhibitor, to explore the role of LIPH-LPA-LPAR axis in PDAC. Using ki16425 effectively promoted the phosphorylation of YAP1, inhibited the expression of its downstream gene, CTGF (Fig. 5J), and restrained the growth of subcutaneous tumors (Fig. 5K; Additional file 5: Fig. S5D). Together, these data indicate that nuclear localization of HIF1A and YAP1 is responsible for LIPH-mediated tumorigenesis and abnormal glycolysis.

In addition to localization on the membrane, a large amount of endogenous LIPH was found to be universally distributed in the cytosol (Fig. 1E), which has not been discussed in previous publications, and neither has its potential role in glycolysis. To explore this, LC–MS was performed for the products of Co-IP with the FLAG antibody. Various glycolytic enzymes, including ENO1, ALDOA, LDHB, and G6PD showed combination with LIPH (Additional file 6: Fig. S6A; Fig. 6B). ALDOA, a hub enzyme directly connecting lipid and glucose metabolism, which has a high overlap with LIPH in biological functions, was chosen for further analysis [19]. Multi-IF showed that ALDOA and LIPH were co-expressed in the cytosol and that the expression of ALDOA was higher in cells transfected with the OE-LIPH plasmid (Fig. 6A). More ALDOA protein was immunoprecipitated with anti-FLAG in LIPH-overexpressing cells in the co-IP analysis (Fig. 6B), and overexpression of LIPH also elevated ALDOA levels in the cytosol (Fig. 6E). In addition to a slight change in ALDOA mRNA levels (Fig. 3B), which might contribute to its expression in the cytosol, LIPH overexpression significantly enhanced the stability of ALDOA (Fig. 6C), with less degradation through the proteasome pathway (Fig. 6D). The potential ubiquitin sites were predicted using iUUCD (Additional file 6: Fig. S6C). Screening of proteins that interacted with LIPH in MIA PaCa-2 cells transfected with the NC-vector and OE-LIPH revealed that UBE2O, an E2/E3 hybrid ubiquitin-protein ligase that displays both E2 and E3 ligase activities and mediates mono-ubiquitination of target proteins, may be responsible for the elevated expression of ALDOA (Fig. 6E; Additional file 6: Fig. S6B). We hypothesized that the combination of LIPH and ALDOA would reduce the access of UBE2O to potential ubiquitin sites in ALDOA, such as K13, thereby maintaining a high rate of glycolysis in LIPH^high tumors (Fig. 6F). The protein levels of LIPH and ALDOA in the tumor tissues were highly positively correlated (Fig. 6G). Abnormally high ALDOA expression predicted an unfavorable prognosis (Additional file 6: Fig. S6D). Collectively, by competing with UBE2O, LIPH maintained the stability of ALDOA, thereby maintaining glycolysis.

Given the lack of direct and effective LIPH inhibitors, a three-drug combination therapy (ki16425/aldometanib/gemcitabine, KAG) was administered to restrict tumor glycolysis and progression. The bioluminescence data revealed that KAG-treated mice showed a slower rate of tumor growth than the control mice and the single-drug group (Fig. 6H). To further test this drug combination, we generated a patient-derived murine xenograft tumor model. IHC staining of PDX xenograft tumor samples confirmed the pathological diagnosis and showed high LIPH expression in both (Additional file 6: Fig. S6F). KAG contributed to a remarkable decrease in tumor growth (Fig. 6I) with limited side effects (Additional file 6: Fig. S6E). The tumors were resected, and histological sections were prepared for further investigation. As expected, significant inhibition of Ki67, p-AKT, and GLUT1 expression was observed in the KAG group (Fig. 6J; Additional file 6: Fig. S6G). The results show KAG effectively restrained tumor progression in vivo and combination therapy with gemcitabine may provide additional benefits.

Since mutant KRAS is a critical mediator of pancreatic cancer metabolism [5], we evaluated the association between KRAS mutations and LIPH expression. KRAS^mut tumors were highly positive for LIPH expression in the TCGA, MTAB-6134, and Ruijin cohorts (Additional file 6: Fig. S6H). A strong positive correlation was observed between LIPH and KRAS expression (Additional file 6: Fig. S6I). Cell line analysis showed that both non-neoplastic HPNE harbored wild-type KRAS and showed lower levels of LIPH expression than PANC-1, CFPAC-1, ASPC-1, 8988s, and 8988t cells, which have an activating mutation in KRAS (Fig. 2A) [34]. Interestingly, LIPH levels were not simply associated with KRAS mutation status, as Mia PaCa-2 cells harboring KRAS mutations expressed lower levels of LIPH than the wild-type KRAS cell line HPNE, which may be due to its semi-adherent phenotype. Mouse-derived cell lines [35] with KRAS^mut exhibit higher LIPH expression (Additional file 6: Fig. S6J). We also examined LIPH expression in pancreatic tissues of mice expressing KRAS^G12D. We confirmed that LIPH expression was higher than that in wild-type mice (Additional file 6: Fig. S6K). We investigated the correlation between LIPH and oncogenic KRAS using a doxycycline-inducible cell culture system. BxPC-3^{wild type KRAS} were stably transfected with the doxycycline-inducible lentiviral KRAS^G12D construct. Following the removal of doxycycline, the cells reverted to wild-type KRAS, as confirmed by the downstream ERK pathway. Reversion to wild-type KRAS gradually decreased the LIPH expression (Additional file 5: Fig. S5L). KRAS signaling plays a pivotal role in biological processes through several downstream pathways. To further confirm which signaling pathway mediates the high expression of LIPH, the MEK inhibitor U0126 was utilized. MEK inhibition significantly downregulated the expression of LIPH in the PANC-1 and CFPAC-1 cell lines (Additional file 6: Fig. S6M), indicating that KRAS-mediated LIPH upregulation occurred through the MEK/ERK pathway. Taken together, KRAS^mut elevates LIPH expression and promotes metabolic reprogramming in PDAC cells.