Bisecting GlcNAc levels are reduced in CRC tissues. We previously reported the downregulation of MGAT3 expression and bisecting GlcNAc levels in breast cancer (16, 17). To further investigate its potential role in other cancer types, we analyzed The Cancer Genome Atlas (TCGA) database and found a consistent reduction in MGAT3 expression in CRC, lung adenocarcinoma, and endometrioid cancer tissues compared with normal tissues (Figure 1A). Immunohistochemical (IHC) staining of CRC tissue microarrays confirmed a marked decrease in both MGAT3 expression and bisecting GlcNAc levels in CRC tissue compared with normal tissue (Figure 1, B and C). Notably, higher MGAT3 expression and elevated bisecting GlcNAc levels were positively correlated with improved overall survival in patients with CRC (Figure 1, B and C).

Bisecting GlcNAc levels are reduced in CRC tissues. (A) MGAT3 expression in TCGA database. BLCA, bladder cancer; BRCA, breast cancer; ESCA, esophageal cancer; LUAD, lung adenocarcinoma; STAD, stomach adenocarcinoma; UCEC, uterine corpus endometrial carcinoma. (B) Immunohistochemical (IHC) staining of MGAT3 in CRC and matched adjacent tissues on TMAs (scale bars: 50 μm), with corresponding Kaplan-Meier survival curves stratified by MGAT3 levels. (C) IHC staining of bisecting GlcNAc in CRC and matched adjacent tissues on TMAs (scale bars: 50 μm), with corresponding Kaplan-Meier survival curves stratified by bisecting GlcNAc levels. Bisecting GlcNAc levels in normal colon cells and CRC cells with low or high metastatic potential were evaluated by lectin blotting (D), immunofluorescence with Phaseolus vulgaris erythroagglutinating (PHA-E) lectin staining (see Supplemental Methods) (E, scale bars: 10 μm), and flow cytometry (F). The results are presented as mean ± SEM. The statistical significance of 2 groups was determined using a paired t test (B and C) or 2-tailed Student’s t test (A). Analysis of multiple groups was performed by 1-way ANOVA followed by Tukey’s multiple comparisons test (D and E). The cell culture experiments were performed with at least 3 independent repeats. *P < 0.05; **P < 0.01; ***P < 0.001.

Quantitative analysis of Western blot data from 3 independent experiments confirmed that MGAT3 protein levels were markedly reduced in CRC cells relative to normal cells, consistent with the observed decrease in bisecting GlcNAc (Figure 1D). Highly metastatic CRC cells exhibited a downregulation of bisecting GlcNAc compared with cells with lower metastatic potential (Figure 1D). However, global N-glycosylation levels were not altered in CRC cells with different migration abilities (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.179533DS1). These findings were further substantiated through immunofluorescence and flow cytometry experiments (Figure 1, E and F). Bisecting GlcNAc was predominantly localized to the cell membrane, with a subset detected intracellularly, consistent with its role in modifying membrane and secretory proteins (6, 18). Quantification of fluorescence intensity confirmed a substantial reduction in both total and plasma membrane–associated bisecting GlcNAc levels in CRC cells, particularly in highly metastatic lines (HCT116 and SW620), relative to normal colon cells (NCM460) (Figure 1E).

Bisecting GlcNAc levels are reduced in the colonic tissue of CRC mice. We further generated AOM/DSS-induced CRC mouse models to assess Mgat3 expression and bisecting GlcNAc levels in colonic tissues from control and CRC mice (Figure 2, A and B). Proteomic analysis identified 99 differentially expressed proteins in the colonic tissues of CRC mice, including 31 upregulated and 68 downregulated proteins (Figure 2C and Supplemental Tables 1 and 2). Mgat3 expression was reduced in the colonic tissues of CRC mice (Figure 2D), and this reduction at the mRNA level was further validated by qRT-PCR (Figure 2E). IHC staining likewise demonstrated reduced levels of Mgat3 and bisecting GlcNAc in the colonic tissues of CRC mice, accompanied by elevated β-catenin expression (Figure 2F). Collectively, these findings indicate that downregulation of bisecting GlcNAc may contribute to CRC pathogenesis.

Bisecting GlcNAc levels are reduced in the colonic tissue of CRC mice. (A) Scheme of the animal experimental design. (B) Quantification of colon tumors in mice (n = 7–8 mice per group). (C) Heatmap of differentially expressed proteins in colonic tissues form CRC and control mice (n = 3 mice per group). (D) Mgat3 protein expression in colonic tissues form CRC and control mice. (E) Mgat3 mRNA expression detected by qRT-PCR (n = 7–8 mice per group). (F) H&E staining and IHC for Pcna, Mgat3, and bisecting GlcNAc in colonic tissues (n = 7–8 mice per group, scale bars: 50 μm). The results are presented as mean ± SEM. The statistical significance of 2 groups was determined using a 2-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Bisecting GlcNAc inhibits tumorigenesis of CRC cells. To investigate the potential influence of bisecting GlcNAc on CRC cell behavior, we overexpressed MGAT3 in HCT116 cells (Figure 3, A and B). Elevated bisecting GlcNAc levels inhibited cell proliferation, migration, and invasion, while promoting apoptosis (Figure 3, C–F). We then generated MGAT3-knockdown (MGAT3-KD) SW480 cells using CRISPR/Cas9-mediated gene editing (Figure 3, G and H). In contrast with MGAT3 overexpression, MGAT3 KD markedly enhanced cell proliferation, migration, and invasion, while reducing apoptosis (Figure 3, I–L). These findings highlight the inhibitory role of elevated bisecting GlcNAc in CRC carcinogenesis.

Bisecting GlcNAc inhibits tumorigenesis of CRC cells. MGAT3 overexpression in HCT116 cells was confirmed by qRT-PCR (A) and Western blotting (B). Cell proliferation of MGAT3-overexpressing HCT116 cells was assessed by CCK-8 (C) and EdU assays (D). (E) Apoptosis in MGAT3-overexpressing HCT116 cells. (F) Migratory and invasive capacities of MGAT3-overexpressing HCT116 cells. MGAT3 KD in SW480 cells was assessed by qRT-PCR (G) and Western blotting (H). Cell proliferation of MGAT3-KD SW480 cells was assessed by CCK-8 (I) and EdU assays (J). (K) Apoptosis in MGAT3-KD SW480 cells. (L) Migratory and invasive capacities of MGAT3-KD SW480 cells. The results are presented as mean ± SEM. The statistical significance of 2 groups was determined using a 2-tailed Student’s t test. The cell culture experiments were performed with at least 3 independent repeats. *P < 0.05, **P < 0.01, ***P < 0.001.

PA increases bisecting GlcNAc levels in vitro and in vivo. Previous studies indicated that PA downregulates Mgat4a expression, resulting in a deficiency of the GnT-4a glycosyltransferase in pancreatic β cells (15). To assess whether PA similarly affects bisecting GlcNAc levels, CRC cells were treated with different fatty acids: PA (a saturated fatty acid), linoleic acid (LA, a polyunsaturated fatty acid), and oleic acid (OA, a monounsaturated fatty acid). PA treatment markedly elevated bisecting GlcNAc levels in HCT116 cells, whereas OA and LA had no such effect (Figure 4A). The maximum increase was observed after 48 hours of PA treatment (Supplemental Figure 2A). Therefore, subsequent functional assays were performed using 200 μM PA for 48 hours. Flow cytometry and Transwell assays demonstrated that PA markedly promoted apoptosis and inhibited the migratory and invasive capacities of HCT116 cells, with effects exceeding those of OA or LA (Figure 4, B and C). Consistent results were obtained in SW480 cells (Supplemental Figure 2, B–D).

PA increases bisecting GlcNAc levels in vitro and in vivo. HCT116 cells were treated with PA, OA, LA, or BSA for 48 hours. (A) Bisecting GlcNAc levels assessed by lectin blotting. (B) Apoptosis detected by flow cytometry. (C) Migratory and invasive capacities assessed by Transwell assays. (D) Scheme of the animal experimental design. (E) Measurement of mouse body weight. (F) Quantification of colon tumors in mice (n = 9–12 mice per group). (G) H&E and IHC staining for Pcna, Mgat3, and bisecting GlcNAc in colonic tissues (n = 9–12 mice per group, scale bars: 50 μm). (H) mRNA expression of Mgat3, Pcna, and Mki67 detected by qRT-PCR (n = 8–12 mice per group). (I) Bisecting GlcNAc levels in colonic tissues assessed by lectin blotting. The results are presented as mean ± SEM. Analysis of multiple groups was performed by 1-way ANOVA followed by Tukey’s multiple-comparison test. The cell culture experiments were performed with at least 3 independent repeats. NS, not significant. *P < 0.05; **P < 0.01; ***P < 0.001.

Consistent with in vitro findings, in vivo results demonstrated a substantial reduction in total tumor number of mice fed a PA-rich diet compared with the control group, whereas no significant changes occurred in the PA-medium group (Figure 4, D–F). Furthermore, a PA-rich diet substantially downregulated the expression of protumorigenic genes (Pcna and Mki67) in the colon (Figure 4, G and H), accompanied by upregulation of bisecting GlcNAc level and Mgat3 expression (Figure 4, G–I). No significant changes occurred in the PA-medium group. Collectively, these results demonstrate that PA specifically increases the levels of bisecting GlcNAc and attenuates CRC carcinogenesis both in vitro and in vivo.

PA inhibits CRC carcinogenesis through the regulation of bisecting GlcNAc. We further treated MGAT3-KD SW480 cells with PA. Remarkably, MGAT3 KD (PA + MGAT3-KD) reversed the PA-induced (PA + negative control transfection [NC]) upregulation of MGAT3 and bisecting GlcNAc (Figure 5A). Moreover, MGAT3 KD (PA + MGAT3-KD) abolished the PA-induced (PA + NC) suppression of migration (Figure 5B). We generated Mgat3fl/fl Villin-Cre mice to further assess the impact of bisecting GlcNAc on the inhibitory effect of PA on CRC in vivo. First, both Mgat3fl/fl Villin-Cre mice and Mgat3fl/fl control mice were fed a normal diet and subjected to AOM/DSS to establish the CRC model. Mgat3fl/fl Villin-Cre mice on a normal diet developed markedly more tumors than Mgat3fl/fl control mice (Supplemental Figure 3), consistent with the tumor-suppressive role of Mgat3. Then, both Mgat3fl/fl Villin-Cre mice and Mgat3fl/fl mice were fed a PA-rich diet and subjected to the AOM/DSS treatment (Figure 5, C and D). Remarkably, compared with the control group (Mgat3fl/fl), the Mgat3fl/fl Villin-Cre group exhibited a marked increase in both total tumor number and size (Figure 5E). IHC, qRT-PCR, and Western blot analyses confirmed the KD of Mgat3 expression and bisecting GlcNAc levels in the colonic tissues of Mgat3fl/fl Villin-Cre mice (Figure 5, F–H). Additionally, upregulation of protumorigenic genes (Pcna and Mki67) occurred in the colonic tissues of Mgat3fl/fl Villin-Cre mice (Figure 5, F and G). Moreover, Mgat3fl/fl mice fed a PA-rich diet showed a marked reduction in tumor number compared with those fed a normal diet, with no significant difference in tumor number between Mgat3fl/fl Villin-Cre mice fed a PA-rich diet and those maintained on a normal diet (Supplemental Figure 3). These findings underscore the key role of bisecting GlcNAc in mediating the inhibitory effects of PA on CRC.

PA inhibits CRC carcinogenesis through the regulation of bisecting GlcNAc. (A) Bisecting GlcNAc levels and MGAT3 expression in MGAT3-KD SW480 cells treated with PA. (B) Migratory capacities assessed by Transwell assays. (C) Scheme of the animal experimental design. (D) Measurement of mouse body weight. (E) Quantification of colon tumors in mice (n = 10 mice per group). (F) H&E and IHC staining for Pcna, Mgat3, and bisecting GlcNAc in colonic tissues (n = 10 mice per group, scale bars: 50 μm). (G) mRNA expression of Pcna, Mki67, and Mgat3 detected by qRT-PCR (n = 10 mice per group). (H) Mgat3 expression and bisecting GlcNAc levels in colonic tissues assessed by Western blotting and lectin blotting. The results are presented as mean ± SEM. The statistical significance of 2 groups was determined using a 2-tailed Student’s t test. The cell culture experiments were performed with at least 3 independent repeats. NS, not significant. *P < 0.05; **P < 0.01; ***P < 0.001.

PA stabilizes DSG2 expression through bisecting GlcNAc. To identify specific glycoproteins regulated by PA-mediated bisecting GlcNAc modification, we performed intact glycopeptide analysis using liquid chromatography–tandem mass spectrometry (LC-MS/MS) (Figure 6A). Glycopeptide spectra were interpreted using Glyco-Decipher, identifying an average of over 200 glycopeptides and more than 130 glycoproteins per group (Supplemental Figure 4 and Supplemental Table 3). Characteristic ions diagnostic of bisecting GlcNAc–bearing glycopeptides ([pep+N3H] and [pep+N3HF]) (19) revealed that desmoglein 2 (DSG2) and lysosome-associated membrane protein 1 (LAMP1) were the only glycoproteins exhibiting this modification in PA-treated groups (Supplemental Table 3 and Figure 6B). Given DSG2’s established role in maintaining epithelial integrity (20), we focused on its functional relevance to bisecting GlcNAc–mediated effects. Immunoprecipitation assays confirmed the presence of bisecting GlcNAc on DSG2 and a marked increase in total DSG2 expression following both MGAT3 overexpression and PA treatment (Figure 6C). Furthermore, quantitative fluorescence imaging confirmed the enhanced DSG2 expression induced by these interventions (Figure 6D).

PA stabilizes DSG2 expression through bisecting GlcNAc. (A) Scheme of intact glycopeptide analysis. (B) Representative MS/MS spectrum of a DSG2-derived glycopeptide bearing bisecting GlcNAc. (C) Levels of bisecting GlcNAc–modified DSG2 in MGAT3-overexpressing or PA-treated HCT116 cells, detected by immunoprecipitation. (D) DSG2 immunofluorescence in MGAT3-overexpressing or PA-treated HCT116 cells (scale bars: 10 μm). (E) DSG2 half-life in BSA- or PA-treated HCT116 cells evaluated following cycloheximide (CHX) treatment. (F) DSG2 expression in HCT116 cells treated with chloroquine (Chlo) or MG132. DSG2 expression in MGAT3-overexpressing (G) or PA-treated (H) HCT116 cells following treatment with DMSO (8 hours), CHX (8 hours), CHX + Chlo (8 hours), or CHX + MG132 (8 hours). The results are presented mean ± SEM. The statistical significance of 2 groups was determined using a 2-tailed Student’s t test (C–F). Analysis of multiple groups was performed by 1-way ANOVA followed by Tukey’s multiple-comparison test (G and H). The cell culture experiments were performed with at least 3 independent repeats. NS, not significant. *P < 0.05; **P < 0.01; ***P < 0.001.

To investigate whether PA stabilizes DSG2 expression by inhibiting its degradation, we treated HCT116 cells with the protein synthesis inhibitor cycloheximide (CHX). PA treatment robustly prolonged the half-life of DSG2 (Figure 6E). Conversely, MGAT3 KD substantially reduced the half-life of DSG2 (Supplemental Figure 5). Additionally, the expression of DSG2 was enhanced by the lysosome pathway inhibitor chloroquine (Chlo) but was unaffected by the proteasome pathway inhibitor MG132 (Figure 6F). Similarly, treatment of HCT116 cells with CHX in combination with either Chlo or MG132 showed that the degradation of DSG2 in control samples (NC transfection and BSA treatment) was substantially reversed by Chlo, but not by MG132. In contrast, the degradation of DSG2 was similar after overexpression of MGAT3 or PA treatment (Figure 6, G and H). To investigate the subcellular sorting mechanism, we examined DSG2 localization relative to the lysosomal marker LAMP2, the early endosome marker EEA1, and the recycling endosome marker TFR (21). MGAT3 overexpression markedly decreased DSG2 colocalization with lysosomes, while increasing its colocalization with early endosomes and recycling endosomes (Supplemental Figure 6). Furthermore, both MGAT3 overexpression and PA treatment markedly reduced DSG2 accumulation in lysosome-enriched fractions (Supplemental Figure 7). Collectively, these findings indicate that PA stabilizes DSG2 expression by enhancing its bisecting GlcNAc modification, which promotes sorting into recycling endosomes and transport to the plasma membrane, thereby reducing lysosomal degradation of DSG2.

PA inhibits CRC carcinogenesis through bisecting GlcNAc–modified DSG2. Downregulation of Dsg2 was observed in the colonic tissues of Mgat3fl/fl Villin-Cre mice (Figure 7A). Furthermore, reduced expression of DSG2, which correlated with levels of bisecting GlcNAc, was observed in human CRC tissues (Figure 7B). Overexpression of DSG2 in HCT116 cells substantially increased apoptosis and decreased migration compared with the NC group (Figure 7, C and D). In contrast, downregulating DSG2 expression in both HCT116 and SW480 cells increased the phosphorylation of EGFR and AKT, but not STAT3 or ERK phosphorylation (Figure 7E). These results indicated that high DSG2 expression inhibits CRC carcinogenesis through the EGFR/AKT signaling pathway. We then examined the impact of PA on the DSG2 function in HCT116 cells and found that KD of DSG2 (PA + shDSG2) abolished the PA-induced (PA + shNC) enhancement of apoptosis (Figure 7F) and the suppression of migration (Figure 7G). Notably, PA treatment (PA + shNC) decreased the activation of the EGFR/AKT signaling pathway, which was reversed by DSG2 inhibition (PA + shDSG2) (Figure 7H). These findings confirm that PA inhibits CRC carcinogenesis through the regulation of DSG2 via bisecting GlcNAc.

PA inhibits CRC carcinogenesis through bisecting GlcNAc–modified DSG2. (A) IHC staining of Dsg2 in mouse colonic tissues (n = 10 mice per group, scale bars: 50 μm). (B) IHC staining of DSG2 and bisecting GlcNAc in human CRC and adjacent normal tissues (n = 16 patients; 4× scale bars: 200 μm, 20× scale bars: 50 μm), with Pearson’s correlation analysis between DSG2 expression and bisecting GlcNAc levels. Apoptosis (C) and migratory abilities (D) of DSG2-overexpressing HCT116 cells. (E) Phosphorylation status of EGFR signaling pathway components in DSG2-KD HCT116 and SW480 cells. Apoptosis (F) and migratory abilities (G) of HCT116 cells treated with PA and shDSG2. (H) Phosphorylation status of EGFR signaling pathway in HCT116 cells treated with PA and shDSG2. The results are presented as mean ± SEM. The statistical significance of 2 groups was determined using a 2-tailed Student’s t test. The cell culture experiments were performed with at least 3 independent repeats. NS, not significant. *P < 0.05; **P < 0.01; ***P < 0.001.