The pivotal genes correlated with MAC progression were identified
Given the distinct differences in biological behaviors and clinicopathological features between MAC and AC, specific genes related to the clinicopathological features of MAC were identified. First, the differentially expressed genes between MAC and AC were explored via analysis of data from The Cancer Genome Atlas (TCGA), which revealed 1934 differentially expressed genes (Fig. 1A). Next, we further examined the differentially expressed genes (DEGs) between cancerous and adjacent noncancerous tissues in both MAC and AC via analysis of TCGA data, identifying 4,539 DEGs in MAC and 5,485 in AC (Fig. 1B–C). The overlapping DEGs identified in the TCGA database were then visualized in a Venn diagram, and a set of 20 genes were identified (Fig. 1D). In addition, single-cell RNA sequencing (scRNA-seq) was performed with three primary MAC tissues and paired adjacent normal tissues. Then, the UMAP algorithm was used to classify the sequencing cells, and 12 cell clusters were presented. The 12 cell clusters included T cells identified by the expression of CD3E, B cells marked by CD19 expression, Plasma cells expressed JCHAIN positively, Myeloid cells characterized by expression of CD86、CD14 and FCGR3A, Neutrophils cells identified by CSF3R and S100A8 expression, Mast cells expressed KIT positively, Endothelial cells marked by expression of PECAM1, Epithelial cells captured by expression of EPCAM, Fibroblasts identified by the expression of COL1A1 and DCN, Neurous marked by PLP1 and CDH19 expression, Prolifering cells expressed MKI67 and TOP2A remarkably. During the initial clustering analysis, we identified a cell population characterized by high expression of DACH1、FFAR2 and SYNJ1. However, this cluster could not be reliably annotated to any known cell type (Fig. 1E–F). Subsequent analyses revealed that the proportion of this population did not differ significantly between tumor and adjacent non-tumor tissues, and importantly, these cells lacked expression of mucin-related genes, suggesting a limited role in defining the mucinous phenotype of colorectal mucinous adenocarcinoma (MAC). Given that members of the mucin protein family are well known to be highly expressed in MAC, we assessed the expression of mucin genes—including MUC2, MUC3A, MUC4, MUC5AC, MUC12, MUC13, MUC17, and MUC20—within MAC cell clusters. The results showed that both transmembrane and secreted mucins were abundantly expressed in epithelial cell populations (Fig. 1G).
Fig. 1: Exploration of pivotal genes correlated with MAC progression.
A Transcriptomic analysis of MAC and AC tissues via the TCGA database, highlighting differentially expressed genes. B Comparative transcriptomic analysis of MAC and paracancerous tissues via the TCGA database, highlighting differentially expressed genes. C TCGA data-based analysis of differences in gene expression between AC and paracancerous tissues. D Venn diagram showing the intersection of the differentially expressed genes between MAC and AC tissues and between cancerous and adjacent noncancerous tissues in both the MAC and AC groups. E UMAP visualization showing the distribution of cell subtypes in adjacent normal tissues (left) and MAC tissues (right), color-coded by cell type. F Dot plot showing the expression profile of signature genes in each cell cluster. G Expressive distribution of mucin protein family genes in overall cells for each cell cluster in MAC. H Volcano plots with differentially expressed genes in Epithelial cells in MAC vs. adjacent normal tissues. I Venn diagram showing the overlapping differentially expressed genes between MAC and AC tissues, between cancerous and adjacent noncancerous tissues in both the MAC and AC groups, and in epithelial cells between MAC tissues and adjacent tissues. J Western blot analysis of L1TD1 expression in 10 pairs of MAC patient samples and protein band intensities were measured by ImageJ software. K Representative images of immunohistochemical staining and semiquantitative analysis of L1TD1 expression in AC and MAC tumor tissues and tumor-adjacent tissues. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
To further identify genes potentially involved in the progression of MAC, we focused on epithelial cells and performed differential expression analysis between MAC and adjacent non-tumorous epithelial cells. A total of 524 genes were significantly upregulated, and 349 genes were downregulated in MAC (Genes with a p < 0.05 and Log2FC > 1.2 were only considered) (Fig. 1H). These differentially expressed genes may play crucial roles in the pathogenesis and progression of MAC. We then cross-referenced these genes with those differentially expressed between MAC and AC tissues as well as those differentially expressed between cancerous and adjacent noncancerous tissues in both MAC and AC. A Venn diagram was generated, revealing L1TD1 as the sole overlapping differentially expressed gene with increased expression in the MAC group and decreased expression in the AC group (Fig. 1I). Additionally, Western blot analysis of 10 paired samples of MAC and adjacent noncancerous tissues revealed that L1TD1 expression was higher in the MAC tissues than in the corresponding adjacent noncancerous tissues (Fig.1J). Immunohistochemical (IHC) analysis further confirmed that L1TD1 expression was much higher in MAC tissues than in adjacent noncancerous tissues. In contrast to MAC tissues, AC tissues presented lower L1TD1 expression levels than adjacent noncancerous tissues did (Fig. 1K). The differential expression between MAC and AC suggests that L1TD1 may be associated with the biological characteristics specific to MAC.
L1TD1 promotes MAC progression
The expression of L1TD1 was analyzed in normal colorectal mucosal cells (FHC cells) and MAC cell lines (H498 cells, LS513 cells and LS174T cells) via qRT‒PCR and Western blot (Fig. 2A). To investigate the role of L1TD1 in regulating MAC cell metastasis and proliferation in vitro, L1TD1 was overexpressed in LS513 cells (LS513-Vector and LS513-L1TD1) and silenced in LS174T cells (LS174T-shNC, LS174-shL1TD1.1, and LS174T-shL1TD1.2) through stable transfection (Fig. 2B). L1TD1 silencing significantly inhibited the migration of MAC cells in the Scratch and Transwell assays, whereas L1TD1 overexpression significantly increased their migration (Fig. 2C–D). The results of the Matrigel invasion assay revealed that L1TD1 overexpression significantly increased the invasive ability but L1TD1 silencing significantly reduced the invasive ability of LS174T cells (Fig. 2D). The results of the Cell Counting Kit-8 (CCK8) and plate colony formation assays revealed that silencing of L1TD1 significantly inhibited but overexpression of L1TD1 expression significantly increased cell proliferation (Fig. 2E–F). The differential expression between MAC and AC suggests that L1TD1 may be associated with the biological characteristics specific to MAC. In addition, high L1TD1 expression was correlated with poor differentiation, a large tumor size, vascular infiltration, and advanced TNM stages in patients with MAC (Supplementary Table 1).
Fig. 2: L1TD1 promotes MAC progression.
A Western blot and qRT‒PCR analyses of L1TD1 expression in FHC cells, H498 cells, LS513 cells and LS174T cells. B Western blot and qRT‒PCR analyses of L1TD1 expression in LS513 cells and LS174T cells after stable transfection. C Representative images of the scratch assay at 0 h and 48 h; the data for the healing rate are shown (scale bar, 150 μm). D Representative images of the Transwell assay (upper panel) and Matrigel invasion assay (lower panel) (scale bar, 200 μm). The data for migrating and invading cells are shown. E Representative images of the colony formation assay; the colonies formed were counted. F The cell proliferation ability was evaluated via a CCK8 assay. G Western blot and qRT‒PCR analyses of L1TD1 expression in MAC cells (H498, LS174T, and LS513) treated with either DMSO or oxaliplatin (OX). H CCK8 assay results showing the effects of L1TD1 overexpression and knockdown (shL1TD1-1 and shL1TD1-2) on cell survival in response to increasing concentrations of oxaliplatin. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
Metastatic nude mouse model was established by tail vein injection of L1TD1 overexpressing LS531 cells or L1TD1 silencing LS174T cells. The in vivo results revealed that the mice in the L1TD1-overexpressing group had more metastatic cancer nodules and L1TD1 silencing group had less metastatic cancer nodules in the liver than those in the control group (Fig. 3A). Moreover, all six nude mice with L1TD1 overexpressing LS174T cells but only two in the control group had visible mesenteric metastases (Fig. 3B). Only one nude mouse with L1TD1 silencing LS174T cells developed mesenteric metastases, whereas five mice in the control group did (Fig. 3B). These results highlight the important role of L1TD1 in promoting metastasis in MAC. Subcutaneous tumor formation assays demonstrated that L1TD1 overexpression significantly increased but L1TD1 silencing significantly decreased the tumor growth rate and weight (Fig. 3C). Collectively, these findings confirmed the function of L1TD1 in promoting MAC cell metastasis and proliferation both in vivo and in vitro.
Fig. 3: L1TD1 promotes MAC cell metastasis and proliferation in vivo.
A Representative images of livers and HE is staining of liver tissues (scale bar, 200 μm) from nude mice after tail vein injection. B Representative images of mesenteric tumors harvested from nude mice after tail vein injection. C–D Gross images of subcutaneous tumors formed from LS174T and LS513 cells before and after dissection in mice (n = 6) treated with or without licoisoflavone A or oxaliplatin. The volume and weight of tumors in the subcutaneous tumor mouse models (n = 6) are shown. E Representative images of IHC staining of tumor tissues from the subcutaneous hormonal model (scale bar, 200 μm). F Representative images of PAS-stained tumor tissues from the subcutaneous tumor model (scale bar, 200 μm). *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
MAC is generally much more chemoresistant than AC, which makes the available treatment strategies quite limited in patients with advanced stages. The role of L1TD1 in chemoresistance in MAC was explored. Our in vitro results revealed cells with high L1TD1 expression presented increased resistance to oxaliplatin, consistengly, L1TD1 expression was significantly increased in oxaliplatin-resistant cells (Fig. 2G–H). We further investigated whether L1TD1 had an effect on the chemosensitivity of MAC cells in vivo. The results from subcutaneous nude mouse models indicated that, following oxaliplatin administration, LS513-L1TD1 cells exhibited faster growth and formed larger and heavier tumors than the control cells (Fig. 3D). In summary, our findings indicate that L1TD1 plays a crucial role in promoting chemoresistance in MAC.
L1TD1 promotes mucus production and tumor progression in MAC
Mucus accounts for more than 50% of MAC tissues, and secretion of mucus is the most notable feature of MAC, and our previous results have demonstrated that the percentage of mucus was notably correlated with the prognosis of patients [16]. Next, the role of L1TD1 in mucus production was then investigated. L1TD1 overexpression significantly increased but L1TD1 silencing significantly decreased the MUC2 and MUC5AC levels (Fig. 4A). Additionally, goblet cell and secretory progenitor cell markers (AGR2, ATOH1, and SPDEF) presented increased expression levels in L1TD1-overexpressing cells and decreased expression levels in L1TD1-silenced cells (Fig. 4B). The solid-phase ELISA results confirmed that MUC2 secretion from L1TD1-overexpressing cells was greater than that from L1TD1-silenced cells (Fig. 4C). Furthermore three-dimension (3D) cell culture technology detected significantly higher MUC2 expression in L1TD1-overexpressing cells and significantly lower expression in L1TD1-silenced cells (Fig. 4D). In addition, by IHC and Periodic acid–Schiff (PAS) staining found that the L1TD1 expression level was positively correlated with the percentage of mucus in MAC tissues (Fig. 4E). In addition, in mice, subcutaneous nude mouse models with high L1TD1 expression presented an increased percentage of mucus, and those with low L1TD1 expression exhibited a decreased percentage of mucus (Fig. 3F). These results demonstrated that L1TD1 could increase mucus production in MAC.
Fig. 4: L1TD1 promotes mucus production.
A Western blot and qRT‒PCR analyses of MUC2 and MUC5AC expression in LS513-overexpressing cells and L1TD1-silenced LS174T cells. B Analysis of AGR2, ATOH1, and SPDEF expression by Western blot and qRT‒PCR. C ELISA was used to measure the concentration of MUC2. D Representative images of Matrigel culture and of MUC2 (red) and DAPI (blue) staining in MAC cells (scale bar, 50 μm). E Correlation analysis of the L1TD1 immunohistochemical score and PAS staining score with the percentage of mucus in MAC tissues. *P < 0.05, **P < 0.01, ***P < 0.001, ns not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
At present, the role and potential mechanisms of mucus influencing tumor progression were rarely reported. To further explore the role of mucus in L1TD1-mediated MAC progression, MUC2, a key component of mucus in MAC tissues, was targeted in MAC cells. Interference with MUC2 expression reversed the increases in the proliferative, invasive, and migratory abilities of MAC cells induced by L1TD1 overexpression (Fig. S1A–E). Besides, the improved chemoresistant ability was also reduced after MUC2 silencing in L1TD1-overexpressed cells (Fig. S1F and S1G). Collectively, these results suggest that L1TD1 promotes MAC progression and chemoresistance by increasing mucus production.
L1TD1 specifically binds to the GUGU motif in ABCC3 and thereby enhances its mRNA stability
As L1TD1 is an RBP, to identify the mRNAs that potentially bind to L1TD1, RNA immunoprecipitation–sequencing (RIP-seq) was conducted. The comparison results were categorized into comparisons with the positive strand and comparisons with the negative strand via SAMtools. Subsequently, peak calibration was performed via MACS2, which identified 460 mRNAs enriched in the precipitate compared to the input in the RIP assay (according to the peak position). Subsequently, RNA sequencing (RNA-seq) analysis was conducted to compare the mRNA expression levels in LS513-L1TD1 cells and wild-type LS513 cells, and the results revealed 5815 differentially expressed genes, among which 3355 were significantly upregulated and 2460 were significantly downregulated (Fig. 5A). Moreover, 24 overlapping genes between the RIP-seq and RNA-seq datasets were identified (Fig.5B). L1TD1 is located in the cytoplasm, and regulating mRNA stability is a potential function of this RBP. Among the 24 overlapping mRNAs, the binding sites in the 3’-UTR were associated with three specific genes (ABCC3, SPPL2B and ADGRB), as indicated by the annotations determined based on the RIP-seq data. Further verification via qRT‒PCR analysis of the RIP products confirmed significant enrichment of ABCC3, SPPL2B and ADGRB mRNA in the IP group compared with the negative control group (Fig. 5C). ABCC3 was the most enriched among these three genes. Therefore, ABCC3 was selected for further investigation. The results of RNA pulldown followed by agarose gel electrophoresis, with OCT4 and SOX2 functioning as positive controls [11], confirmed that L1TD1 bound to ABCC3 mRNA (Fig. 5D–E). Further analyses revealed that ABCC3 expression was significantly reduced in L1TD1-silenced cells and significantly increased in L1TD1-overexpressing cells (Fig. 5F). IHC analysis confirmed a positive correlation between L1TD1 and ABCC3 expression in both human and mouse subcutaneous tumor tissues (Fig. 3E and Fig. 5G). And we performed loss-of-function experiments using siRNA-mediated knockdown of ABCC3 in LS513 and LS174T wild-type colorectal cancer cell lines. Interestingly, we observed that L1TD1 mRNA and protein levels remained largely unchanged following ABCC3 silencing, as verified by both qRT-PCR and Western blot analysis (Fig. 5H). When treated with actinomycin D, an RNA polymerase inhibitor, LS513-L1TD1 cells presented a significant increase in ABCC3 mRNA stability (Fig. 5I). In summary, the binding of the L1TD1 to the 3’-UTR of ABCC3 increases ABCC3 mRNA stability.
Fig. 5: The L1TD1 binds to ABCC3 mRNA and increases its stability.
A Heatmap showing the expression levels of the upregulated and downregulated (fold change >1.5) genes in L1TD1-overexpressing compared with wild-type MAC cells. B Venn diagram showing the overlapping genes identified by RNA-seq (L1TD1-overexpressing and wild-type MAC cells) and RIP-seq (L1TD1-overexpressing cells), with the expression of the overlapping genes visualized in a heatmap and a volcano plot. C Quantitative enrichment of ABCC3, SPPL2B, and ADGRB in the RIP products. D Analysis of ABCC3, OCT4, SOX2, and GAPDH transcript abundances in immunocomplexes via agarose gel electrophoresis. E Analysis of proteins interacting with ABCC3 mRNA via RNA pulldown. F Western blot and qRT‒PCR analyses of ABCC3 expression in LS513-L1TD1 cells and LS174T-L1TD1 cells. G Representative images of IHC staining for L1TD1 and MRP3 and PAS staining in MAC tissues. The correlation between L1TD1 and ABCC3 expression was analyzed (scale bar, 200 μm). H Western blot and qRT‒PCR analyses of ABCC3 and L1TD1 expression in LS513 cells and LS174T cells transfected with ABCC3 siRNAs. I Analysis of ABCC3 mRNA levels at 0, 2, 4, 6, and 8 h after actinomycin D treatment (10 μg/ml) in L1TD1-overexpressing cells and L1TD1-silenced cells. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
To confirm that L1TD1 binds to the 3’-UTR of ABCC3 mRNA, plasmids containing the ABCC3 3’-UTR sequence were constructed and cotransfected with L1TD1 expression plasmids. The dual-luciferase reporter assays demonstrated that L1TD1 bound to the ABCC3 3’-UTR. Further assays with L1TD1 wild-type and RNA recognition motif (RRM) mutant plasmids revealed that the RRM1 structural domain mediates this binding (Fig. 6A). Based on RIP-seq analysis, the RNA recognition motif (RRM) domain of L1TD1 was found to potentially interact with a 157-nucleotide segment of the ABCC3 mRNA 3′ untranslated region (3′UTR), though the precise binding site remained undefined. To resolve this, the 3′UTR fragment was subdivided into five partially overlapping ~50-nucleotide RNA probes (F1–F5). Electrophoretic mobility shift assay (EMSA) revealed a distinct RNA–protein complex formation with probe F4 (nt 90–140), while no binding was observed with probes F1, F2, F3, or F5 (Fig. 6B). These findings indicated that the RRM domain of L1TD1 binds specifically to the 90–140 nt region of ABCC3 3′UTR. To further narrow down the interaction site, probe F4 was segmented into four overlapping ~15-nucleotide fragments: P1 (nt 90–105), P2 (nt 100–115), P3 (nt 110–125), and P4 (nt 120–140). Among these, EMSA showed a specific shift only with probe P3, indicating that nucleotides 110–125 constitute the core binding region for L1TD1 (Fig. 6C). To explore the molecular interaction in greater detail, the three-dimensional structure of the ABCC3 mRNA 3′UTR fragment was modeled using RNAComposer (https://rnacomposer.cs.put.poznan.pl/), while the L1TD1 protein structure was retrieved from the UniProt database (PDB ID: AF-Q5T7N2-F1-model_v4.pdb). The RRM domain was isolated using PyMOL for docking analysis. Visualization of the protein–RNA complex revealed multiple polar interactions between L1TD1 and the RNA, as indicated by yellow dashed lines (Fig. 6D). Structural alignment highlighted a conserved GU dinucleotide at positions 115–116, which was further extended to a GUGU motif (nt 115–118) upon sequence analysis—a known canonical element recognized by RRM-containing RNA-binding proteins.
Fig. 6: L1TD1 RRM Domain specifically binds a GUGU motif within the ABCC3 3′UTR and enhances mRNA stability.
A Dual-luciferase reporter assay results indicating the binding site between the L1TD1 protein and ABCC3 mRNA. B Electrophoretic mobility shift assay (EMSA) using biotin-labeled RNA probes F1–F5 and recombinant L1TD1 protein. A clear RNA–protein complex is observed only with probe F4 (nt 90–140), indicating specific binding within this region. C F4 was further divided into four overlapping probes (P1–P4). Only probe P3 (nt 110–125) shows a shifted band. D Molecular docking of the L1TD1 RRM domain with the predicted 3D structure of the ABCC3 3′UTR reveals polar interactions at the RNA–protein interface. E A schematic diagram shows the structure of the ABCC3 mRNA 3’UTR. F EMSA using wild-type P3 and GUGU-mutated probe (P3-mut). G EMSA showing dose-dependent binding of L1TD1 to the biotin-labeled wild-type P3 RNA probe. H EMSA showing increasing amounts of L1TD1 protein incubated with fixed concentration of biotin-labeled P3. I Dual-luciferase reporter assay in cells co-transfected with L1TD1 expression plasmid and either wild-type or GUGU-mutated ABCC3 3′UTR reporter. J ABCC3 mRNA stability assay following actinomycin D treatmen. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
To determine whether the GUGU sequence is essential for this interaction, a mutant version of probe P3 was generated in which the GUGU motif was replaced with ACAC (P3-mut) (Fig. 6E). EMSA using this mutant probe showed complete loss of the shift band (Fig. 6F), confirming that the GUGU motif is critical for complex formation between L1TD1 and ABCC3 mRNA. Next, we examined the affinity and specificity of this interaction using concentration gradient EMSA. Biotin-labeled wild-type P3 probes at increasing concentrations were incubated with a fixed amount of L1TD1 protein. The shift band intensity increased in a dose-dependent manner (Fig. 6G). In a competition assay, increasing amounts of unlabeled (cold) wild-type probe efficiently competed with the biotin-labeled probe, reducing complex formation, whereas the mutant probe did not (Fig. 6H). Similarly, increasing L1TD1 protein concentrations with fixed P3 probe also enhanced the binding signal in a concentration-dependent fashion (Fig. 6G), indicating strong affinity and specificity. Luciferase reporter constructs harboring either wild-type or GUGU-mutated ABCC3 3′UTR-P3 fragments were co-transfected with an L1TD1 expression vector. Dual-luciferase assays revealed that L1TD1 significantly enhanced reporter activity through the wild-type ABCC3 3′UTR, while the effect was abolished in the mutant construct (Fig. 6I). Furthermore, mRNA stability assays using actinomycin D demonstrated that mutation of the GUGU motif led to a pronounced reduction in ABCC3 mRNA half-life (Fig. 6J). Overall, these findings establish that the RRM domain of L1TD1 specifically recognizes the GUGU motif within the ABCC3 3′UTR, thereby enhancing mRNA stability through a sequence-specific interaction.
L1TD1 promotes mucin production and tumor progression in MAC via the ABCC3/AMPK/MAPK axis
KEGG enrichment analysis of RNA-Seq data comparing L1TD1-overexpressing cells with wild-type controls indicated that L1TD1 may regulate the MAPK signaling pathway (Fig. 7A). To validate this, we assessed the phosphorylation levels of ERK1/2 and MEK as indicators of MAPK activity using Western blot analysis. We observed that L1TD1 overexpression led to increased phosphorylation of both ERK1/2 and MEK. Conversely, L1TD1 silencing resulted in a marked reduction in their phosphorylation levels (Fig. S2A). To investigate the role of ABCC3 in L1TD1-mediated tumor progression in MAC, we interfered with ABCC3 expression in LS513-L1TD1 cells (Fig. 7B). We observed that knockdown of ABCC3 also diminished the phosphorylation of ERK1/2 and MEK, suggesting that L1TD1 activates the MAPK pathway via upregulation of ABCC3 (Fig. S2B). But MRP3 (the protein encoded by ABCC3) is an ATP-dependent membrane export pump rather than a kinase and is therefore unlikely to act directly on the MAPK cascade. In other words, it cannot activate MAPK in the manner of a receptor tyrosine kinase. Although it lacks intrinsic kinase activity, it may nonetheless influence MAPK signaling indirectly by altering cellular metabolism. To probe this mechanism, we standardized the metabolic baseline with Sodium pyruvate pre-buffering and defined the medium switch as time zero. In LS513-Vector and LS513-L1TD1 cells, and in LS513-L1TD1 cells transfected with siRNAs against ABCC3, we measured ATP and kinase phosphorylation at 0, 1, 2, and 6 h. L1TD1 overexpression (with higher ABCC3) produced a marked decline in intracellular ATP at ~1 h, followed by a lower yet relatively stable level thereafter (Fig. 7C). Subsequently, Western blot analysis revealed that AMPK phosphorylation rose first and most strongly, andphosphorylation of CRAF, MEK, and ERK occurred after AMPK activation. Silencing ABCC3 attenuated the ATP decline and proportionally blunted phosphorylation of AMPK, CRAF, MEK, and ERK (Fig. 7D). Taken together, these time-ordered effects indicate that MAPK pathway engagement is activity-mediated and expression-dependent: L1TD1-driven ABCC3 upregulation increases pump abundance and transport-coupled ATP use, which is sensed by AMPK and then propagates to CRAF/MEK/ERK. After ABCC3 silencing, the expression of MUC5AC, MUC2, AGR2, ATOH1, and SPDEF was downregulated (Fig. 7E, G). The solid-phase ELISA revealed that the volume of MUC2 secretion induced by L1TD1 was reversed by ABCC3 silencing. (Fig. 7F). Immunofluorescence analysis after 3D cell culture revealed reduced MUC2 expression following ABCC3 interference (Fig. 7H). PAS staining revealed that tumors with high L1TD1 expression had a greater percentage of mucus when ABCC3 was also highly expressed (Fig. 5G). Subsequently, to determine whether the association between MAPK activation and mucin production is causal rather than merely correlative, we began by exposing LS513-Vector, LS513-L1TD1, and ABCC3-silenced LS513-L1TD1 cells to the MEK inhibitor JTP-74057 (trametinib; MCE, USA). As expected, MEK inhibition markedly reduced the phosphorylation of MEK and ERK in all groups. Under these conditions, MUC2 and MUC5AC expression was significantly downregulated in vector control, L1TD1-overexpressing, and ABCC3-silenced cells (Fig. S2C). Importantly, the differences in mucin expression between the groups were largely abolished. These findings indicate that MEK/ERK signaling is required for L1TD1–ABCC3–driven mucin upregulation. We next used EGF (MCE, USA) to activate the MEK/ERK cascade. Treatment with the EGF strongly increased MEK and ERK phosphorylation in LS513-Vector, LS513-L1TD1, and ABCC3-silenced LS513-L1TD1 cells. Under these conditions, MUC2 and MUC5AC expression was significantly upregulated in all groups (Fig. S2D). Finally, when the ERK agonist was combined with MEK inhibition, phosphorylation of MEK and ERK was again markedly suppressed in LS513-Vector, LS513-L1TD1, and ABCC3-silenced LS513-L1TD1 cells. Under these combined conditions, MUC2 and MUC5AC expression was significantly reduced in all groups (Fig. S2E). Further experiments revealed that interfering ABCC3 expression reversed the proliferative, invasive, and migratory abilities of MAC cells induced by L1TD1 overexpression (Fig. S3A–D). Besides, the improved chemoresistant ability was also reduced after ABCC3 silencing in L1TD1-overexpressed cells (Fig. S3E–F).
Fig. 7: L1TD1 upregulate ABCC3 expression activate AMPK/MAPK signaling promote mucin production and accelerate tumor progression in MAC.
A KEGG pathway enrichment analysis showed the significantly affected signaling pathways. B Western blot and qRT‒PCR analyses of ABCC3 expression in L1TD1-overexpressing LS513 cells transfected with ABCC3 siRNAs. C Intracellular ATP was quantified at 0, 1, 2, and 6 h using a colorimetric ATP Content Assay Kit. D Western blot analyses of AMPK、P-AMPK、CRAF、P- CRAF、ERK、p-ERK、MEK and p-MEK expression in L1TD1-overexpressing LS513 cells transfected with ABCC3 siRNAs. E Western blot and qRT‒PCR analyses of MUC2 and MUC5AC expression in LS513-L1TD1 cells transfected with ABCC3 siRNAs. F ELISA was performed to measure the MUC2 concentration in LS513-L1TD1 cells transfected with ABCC3 siRNAs. G Western blot and qRT‒PCR analyses of AGR2, SPDEF, and ATOH1 expression in LS513-L1TD1 cells transfected with ABCC3 siRNAs. H Representative images of Matrigel culture and of MUC2 (red) and DAPI (blue) staining in LS513-L1TD1 cells transfected with ABCC3 siRNAs (scale bar, 50 μm). *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant (p ≥ 0.05), based on Student’s t test. All the results are from three independent experiments. The data are expressed as the means ± SDs, with at least one representative image acquired.
To further verify the role of ABCC3 in MAC progression, we treated LS513-L1TD1 cells with licoisoflavone A, an MRP inhibitor, which significantly decreased the expression levels of ABCC3, MUC2, MUC5AC, AGR2, ATOH1 and SPDEF (Fig. S4A). In addition, treatment with licoisoflavone A reversed the increase in the proliferative ability of MAC cells induced by L1TD1 overexpression (Fig. S4B). The solid-phase ELISA revealed that MUC2 secretion was inhibited after treatment with licoisoflavone A (Fig. S4C). We subsequently used immunofluorescence staining to visualize MUC2 expression in 3D cell culture and found that treatment with licoisoflavone A markedly reduced MUC2 expression (Fig. S4D). In vivo assays demonstrated that, compared with the PBS-treated group, the LS513-L1TD1 group presented significant reductions in tumor volume and growth rate. Moreover, tumor growth was notably inhibited in the LS513-L1TD1 group treated with both licoisoflavone A and oxaliplatin, indicating the synergistic inhibitory effect of these drugs on tumor proliferation (Fig. 3D). In conclusion, L1TD1-driven ABCC3 upregulation increases pump abundance and transport-coupled ATP use, which is sensed by AMPK and then propagates to CRAF/MEK/ERK.
