We performed high-throughput sequencing in three pairs of fresh GC tissues to construct differential mRNA expression profiles in GC (Fig. 1A). The DESeq algorithm analysis revealed that 762 genes were upregulated. We used GO and KEGG enrichment analyses to screen 11 genes related to fatty acid metabolism/synthesis and lipid modification. FAAH showed the highest fold change among the upregulated genes; thus, we chose this for further research (Fig. 1B). We checked the expression of FAAH in the TCGA and Oncomine databases, and the findings showed that the expression trend was consistent with the sequencing results despite the limited number of sequencing samples (Fig. 1C, D). Based on integrated genomics analysis, TCGA divides GC into four subtypes. Through further analysis, we found that the expression of FAAH in tumours with chromosomal instability (CIN) was significantly different from that in normal tissues (Fig. 1E).

The FAAH gene sequence was easily amplified by its specific primers indicating high operability for clinical application (Supplementary Fig. S1a–c). We subsequently collected another 96 pairs of cancerous and adjacent non-cancerous tissues and detected FAAH expression by qRT-PCR; we found that FAAH was relatively upregulated in 77 GC tissues (Fig. 1F, p < 0.0001). In addition, we randomly selected 20 pairs of the above 96 tissues to verify the expression of the FAAH protein. Moreover, we discovered that FAAH was highly expressed in 70% (14/20) of GC tissues compared to adjacent tissues (Fig. 1G). Analysis of clinicopathological parameters showed that the high expression of FAAH was related to the patient’s sex, tumour size, invasion depth, WHO classification and lymph node metastasis, but was not related to Lauren classification, degree of differentiation and H. pylori status of patients (Table 1). We also examined the levels of FAAH in other common solid tumours, such as lung, breast, and colorectal cancer, and found that the expression of FAAH was specific in GC (Supplementary Fig. S1d). Furthermore, ROC analysis showed that FAAH had diagnostic power as a marker of GC, with an AUC of 0.7059 (Supplementary Fig. S1e).

Analysis showed that FAAH was highly expressed in three GC cell lines (except BGC-823) compared to the GES-1 cells (Fig. 2A, B). Only two of the four interfering plasmids we constructed had good interfering functions (Supplementary Fig. S2a). The two plasmids were then packaged into lentiviral vectors to obtain vectors with relatively high knockdown efficiency, namely LV-shFAAH-1416 and LV-shFAAH-1557 (hereinafter referred to as shFAAH1 and shFAAH2, respectively, Fig. 2C, I). Overexpression of FAAH was also carried out in HGC-27 and BGC-823 cells (Supplementary Fig. S2b). Compared with the control group, the proliferation rate of GC cells was reduced after FAAH interference and the colony-forming units were significantly reduced (Fig. 2D, E). In addition, increased proliferation rate and clone-forming units were observed after FAAH overexpression (Supplementary Fig. S2c, d). The EdU assay also showed that the proportion of cells in the proliferation phase decreased after interference with FAAH, but increased after overexpression (Fig. 2F and Supplementary Fig. S2e). Moreover, silencing of FAAH weakened the cell migration and invasion ability, thereby inhibiting the epithelial-mesenchymal transition (EMT), which was manifested by increased level of E-cadherin and reduced level of N-cadherin, Snail (Slug), and Vimentin (Fig. 2G–I). And migration and invasion were strengthened after the overexpression of FAAH (Supplementary Fig. S2f, g). In addition, the apoptotic rate of GC cells increased and the proportion of S-phase cells decreased after FAAH interference (Fig. 2J, K), while the opposite effects were observed after FAAH overexpression (Supplementary Fig. S2h, i). Western blot analysis showed that the expression of Bcl-2 and Cyclin D1 decreased and the expression of Bax and p27 increased after interference with FAAH, indicating that apoptosis was promoted and the cell cycle was blocked (Fig. 2L). In addition, overexpression of FAAH showed the opposite trend (Supplementary Fig. S2j). Therefore, FAAH exhibited a significant cancer-promoting effect, and interference with FAAH in vitro inhibited the malignant phenotype of GC cells.

We previously packaged two shRNA fragments into lentiviruses and obtained GC cell lines with stable FAAH knockdown through puromycin screening. Based on the results shown in Fig. 2C, we selected the shFAAH-1557 fragment with better knockdown efficiency for in vivo experiments. Untreated MKN-1 cells (mock), lentiviral negative control (LV-nc), and MKN-1 cells with stable knockdown of FAAH (LV-shFAAH) were injected into nude mice to construct tumour xenograft models, followed by 5-fluorouracil (5-FU) to act as the control group of chemotherapeutic drugs. The results showed that the LV-shFAAH group had small tumour size, light tumour weight, and slow growth rate compared to the control group (Fig. 3A–C). In addition, compared with the control group, the LV-shFAAH group had more necrotic areas, manifested as soma lysis and nuclear condensation (Fig. 3D). IHC staining showed that ki67 staining was attenuated in the LV-shFAAH group, while cleaved caspase-3 staining was enhanced, suggesting that FAAH knockdown reduced tumour proliferation and promoted apoptosis in vivo (Fig. 3E, F). TUNEL immunofluorescence also showed that the apoptosis of tumour cells increased after interference with FAAH (Fig. 3G). In addition, qRT-PCR and western blot analysis of tumour tissues showed that silencing of FAAH inhibited the EMT of GC cells, indicating that tumour metastasis was inhibited (Fig. 3H, I).

We searched four public databases (miRDB, miRanda, TargetScan, and miRWalk) for miRNAs that might have binding sites to the 3’-UTR of FAAH, respectively. Two miRNAs (miR-1275 and miR-502-5p) were obtained by taking the intersection of all possible miRNAs (Fig. 4A). Mimics of miR-1275 and miR-502-5p were then transfected into MKN-1 cells, and we found that only miR-1275 could downregulate the expression level of FAAH (Fig. 4B, C). The dual-luciferase reporter gene assay confirmed that miR-1275 had a binding site with the 3’-UTR of FAAH (Fig. 4D). Moreover, the IF-FISH assay showed that FAAH and miR-1275 could be partially combined in the cytoplasm (Fig. 4E). Next, miR-1275 was initially observed to be expressed at low levels in four GC cell lines and 80 pairs of GC tissues (Fig. 4F and Supplementary Fig. S3a). Spearman correlation analysis disclosed a negative correlation between miR-1275 and FAAH in GC tissues (Fig. 4G).

Our previous results showed that the expression of FAAH was elevated in three GC cell lines relative to the normal controls, with the highest expression level in MKN-1 cells and a relatively low expression level in HGC-27 cells (Fig. 2A, B). Considering that miR-1275 is an upstream regulator of FAAH, to verify whether miR-1275 can reverse the effect of FAAH on cell phenotype, we inhibited miR-1275 in HGC-27 cells with a stable knockdown of FAAH, and a miR-1275 mimic was added to MKN-1 cells overexpressing FAAH (Supplementary Fig. S3b). As shown in Supplementary Fig. S3c, miR-1275 could indeed regulate the expression of FAAH. FAAH expression in MKN-1 cells overexpressing miR-1275 was decreased compared to that of the control group, while that of HGC-27 cells with downregulated miR-1275 was increased. Functional tests showed that overexpression of miR-1275 could slow down the proliferation rate, reduce clonal colonies formation and the proportion of proliferating cells, inhibit cell metastasis, and induce apoptosis and cycle arrest; while miR-1275 inhibitor exerted a promoting effect on the malignant phenotype of GC cells (Supplementary Fig. S3d–j).

By co-transfecting the miR-1275 mimic and FAAH overexpression vector (pcFAAH) into MKN-1 cells, we found that the miR-1275 overexpression eliminated the proliferation-promoting effect of pcFAAH (Fig. 5A, C). Similarly, miR-1275 inhibitor reversed the inhibitory effect of the FAAH interference vector (shFAAH) on cell proliferation (Fig. 5B, D). In addition, the promoting effects on cell migration and invasion mediated by pcFAAH were also reversed by miR-1275 mimic, and miR-1275 inhibitor reversed shFAAH effects (Fig. 5E, F). When cells were treated with the miR-1275 mimic and pcFAAH, cell apoptosis decreased and the cell cycle slowed down, while the miR-1275 inhibitor restored shFAAH-mediated apoptosis and cycle arrest (Fig. 5G, H). This phenomenon was further verified by western blotting. MiR-1275 inhibitor could reverse the changes of apoptosis- and cycle-related proteins caused by FAAH interference, upregulate cyclinD1 and Bcl-2 protein levels, and downregulate the levels of p27, cleaved caspase 3, cleaved PARP-1, and Bax; the effects of FAAH overexpression on the above proteins could also be reversed by miR-1275 mimics (Fig. 5I).

Considering that the essence of FAAH is a hydrolase, we found that the activity of FAAH was significantly enhanced in three GC cell lines (excluding BGC-823, Supplementary Fig. S4a). PF-3845 is a potent, selective, irreversible, and orally active inhibitor that inhibits FAAH activity by carbamylating FAAH’s serine nucleophile. First, the IC₅₀ of MKN-1, HGC-27, and BGC-823 cells were measured by CCK-8, respectively (Supplementary Fig. S4b). We then treated cells with different concentrations of PF-3845 according to the measured IC₅₀, and found that PF-3845 inhibited the proliferation of GC cells in a time- and concentration-dependent manner. Specifically, the longer the administration time, the lower the PF-3845 concentration required to achieve 50% proliferation inhibition (Supplementary Fig. S4c). In addition, the effects of PF-3845 on the metastasis, apoptosis, and cell cycle of MKN-1 cells were also concentration-dependent (Supplementary Fig. S4d–f).

To clarify whether this oncogenic effects of FAAH is due to its increased enzyme activity, different concentrations of PF-3845 were used in BGC-823 and HGC-27 cells that stably overexpress FAAH. The results showed that both 1/2 IC₅₀ and IC₅₀ concentrations of PF-3845 eliminated the proliferation effect of FAAH overexpression (Supplementary Fig. S4g), and the combined use of shFAAH and PF-3845 resulted in a higher inhibition rate (Supplementary Fig. S4h). In addition, PF-3845 reduced the strengthened ability of cell migration and invasion caused by FAAH overexpression (Supplementary Fig. S4i). Flow cytometry also suggested that PF-3845 reversed the effects of pcFAAH on apoptosis and cell cycle progression (Supplementary Fig. S4j, k). These results suggest that FAAH mediates cancer-promoting effects mainly due to its increased activity, and its activity inhibitor can weaken the malignant phenotype of GC.

The GSEA analysis showed that FAAH significantly affected fatty acid metabolism (Fig. 6A). We next explored whether a series of lipid mediator changes involved in the process of FAAH-hydrolysing substrates had a certain impact on its biological behaviour. We performed lipidomics detection using MKN-1 cells with stable FAAH knockdown, and found ~255 lipid molecules changed significantly in the negative ion mode after FAAH interference (Fig. 6B, C). The content of LPA and FFA (especially FFA 20:4, also known as AA) decreased, while the content of AEA and phosphatidylethanolamine (PE) components increased (Fig. 6D). Then the content of different types of PGs in the cell supernatant were detected, and we discovered that the content of PGE2 decreased after interference with FAAH, but increased after FAAH overexpression (Fig. 6E). Moreover, after interference with FAAH, the expression of AEA-related enzymes increased, such as NAT-1 and NAPE-PLD, as well as the receptor of AEA (CB1/2). However, the enzymes involved in PG synthesis, COX-2, and LPA-induced hypoxia-inducible factor-1α (HIF-1α) decreased, suggesting that PGE2 synthesis was inhibited (Fig. 6F).

To verify whether miR-1275 was indirectly involved in the regulation of AEA/LPA signals, we treated MKN-1 cells with miR-1275 mimic, and observed that the content of PGE2 in the cell supernatant decreased, while the miR-1275 inhibitor increased PGE2 in HGC-27 cells (Fig. 6G). Further analysis showed that the miR-1275 mimic could reverse the changes in the FAAH-mediated AEA/LPA signal axis, manifested as a decrease in the content of HIF-1α and COX-2, and an increase in the expression of CB1/2, NAT-1, and NAPE-PLD, while the miR-1275 inhibitor showed the opposite trend in regulating the above signals (Fig. 6H).

Three pairs of fresh GC tissues and adjacent tissues were selected for RNA-seq (not included in the 96 pairs of tissues that were subsequently collected). Total RNA was isolated using a HiPure Total RNA Mini Kit (Magen, Germany). Concentration and integrity were measured using a Qubit 3.0 fluorometer (Invitrogen, Carlsbad, USA) and Agilent 2100 Bioanalyzer (Applied Biosystems, Carlsbad, USA), respectively. The RNA-seq library was prepared with ~2 μg of total RNA using the KAPA RNA HyperPrep Kit with RiboErase (HMR) for Illumina® (Kapa Biosystems, Inc., Woburn, MA) according to the manufacturer’s instructions. The Bioanalyzer 2100 system (Agilent Technologies, USA) was used to detect the size range, and qRT-PCR was used to quantify the effective concentration of the library. After the library was qualified, 10 nM was pooled equimolar prior to clustering. paired-end (PE150) sequencing was performed on all samples.

The two shRNA fragments with high knockdown efficiency (shFAAH-1416 and shFAAH-1557) were packed into pGV112-GFP vector (GeneChem, Shanghai, China) and co-transfected 293T cells with pHelper1.0 and pHelper2.0 plasmids to produce lentivirus. An appropriate amount of lentivirus was added to the cells according to the multiplicity of infection (MOI = 50) until 20% confluence was reached. After 72 h of infection, GFP-tagged gene expression was observed under a fluorescence microscope, followed by selection with 2 mg/ml puromycin (MCE, New Jersey, USA) for 2 weeks to obtain stable FAAH knockdown cells.

The 3’-UTR sequences of FAAH were inserted into a pmiR-RB-ReportTM vector (RiboBio, Guangzhou, China), and then transformed and amplified using DH5α. The extracted plasmid was identified using double-restriction enzyme digestion. After site-directed mutagenesis, wild-type and mutant-type recombinant plasmids were constructed, and either of which carrying mimic-NC or miR-1275-mimic were co-transfected into MKN-1 cells. After 48 h of transfection, firefly luciferase/renilla luciferase were detected using the Dual-Luciferase Reporter Assay System (Promega, Madison, USA).

Cells were seeded on glass coverslips at a density of 3 × 10⁵/mL and cultured for 24 h. Then 2 mL of 4% paraformaldehyde was added to fix the cells for 30 min. After washing with DEPC water twice, 1 mL of permeabilization solution was added and incubated at room temperature for 10 min. After washing three times with PBS, 1% paraformaldehyde was used to fix for 10 min. Then 70, 80, 95, and 100% ethanol were used for gradient dehydration. After 30 min of pre-hybridization, the hybridization system containing the miR-1275 probe (BersinBio, Guangzhou, China) was added, and then the glass coverslips were transferred to 73 °C for co-denaturation, and quickly transferred to 37 °C for overnight hybridization. After washing, cells were blocked with 5% BSA for 30 min and incubated with anti-FAAH (Affinity Biosciences, Cincinnati, USA) at 4 °C overnight. On the second day, after incubating with the fluorescent secondary antibody for 30 min, DAPI was used to stain the nucleus.

A total of 20 male BALB/c nude mice (aged 5–6 weeks) were purchased from the Laboratory Animal Center of Nantong University and randomly divided into five groups. The experimental procedures and animal care were approved by the Animal Experimental Ethics Committee of Nantong University. Approximately 2 × 10⁷ MKN-1 cells were resuspended in 200 μL of RPMI-1640 medium and inoculated subcutaneously on the dorsal side of nude mice. Since the day of inoculation, the long (a) and short diameters (b) of the tumour were measured every 2 days to calculate the volume (V = ab²/2) and draw the growth curve. After 10 days of inoculation, fluorouracil was injected intraperitoneally every 2 days for a total of 10 injections to construct a clinical chemotherapy drug treatment model. The nude mice were killed by cervical dislocation ~30 days after inoculation, and the tumours were completely removed. The size and weight of the tumours were recorded, and qRT-PCR was used to detect the expression of related genes.

About 1 mL of pre-cooled 60% methanol aqueous solution (chromatographic grade) was used to resuspend the cells and centrifuged at 1000 × g for 1 min to retain the cell pellet. To extract lipids, 100 μL of water, 200 μL of pre-cooled methanol, 400 μL of dichloromethane, and 120 μL of water were added to the cell pellet (50 mg) in sequence. Vortex mixing was performed for each step. After incubation at 26 °C for 20 min, the suspension was centrifuged at 8000 × g at 10 °C for 15 min to remove the lower organic phase and then dried with nitrogen. For mass spectrometry analysis, 100 µL of isopropanol solution was added for reconstitution, and the supernatant was collected after 15 min of centrifugation at 8000 × g. The samples were separated using an ACQUITY UPLC® I-Class system (Waters, Milford, USA). Mass spectrometry was performed using electrospray ionisation positive and negative ion modes. After separation by high-performance liquid chromatography, the samples were analysed using Xevo G2-XS QTof (Waters, Milford, USA). Progenesis QI software (Waters, Milford, USA) was used for raw data analysis.

Statistical analyses were performed using SPSS (Version 23.0; IBM, Armonk, USA) and GraphPad Prism software (Version 7; La Jolla, USA). Each experiment was repeated three times, and the data were represented as the mean ± SD. Student’s t test or one-way analysis of variance was used for statistical analysis when appropriate. The paired t-test was used to compare paired samples. The Mann–Whitney U test and one-way ANOVA were applied to determine the correlations between FAAH expression and clinicopathological characteristics of patients with GC. Spearman’s correlation analysis was used to evaluate the correlation between FAAH and miR-1275 expression. The grey value of protein bands and the proportion of positive cells in experiments such as EdU, immunohistochemistry, and immunofluorescence assays were quantified using ImageJ software (NIH, Bethesda, USA). Statistical significance was set at p < 0.05.

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.