Fig 2: Predicting Organizational Changes in Trafficking Defective Cell Lines(A and B) Observed and simulated glycan profiles of Cog4KD HeLa cell lines (A) and Cog4KO HEK293T cell lines (B). Error bars for glycan profiling are SEM for n = 3.(C) Total effective enzymatic rate changes predicted by the model following fitting as a result of Cog4 perturbation. The x axis is on a log scale. Error bars for total effective enzymatic rates and distributions are SD for n = 20 (Cog4KD HeLa) and n = 21 (Cog4KO HEK293T) individual fitting procedures.(D) Western blot analysis of endogenous MGAT1, GalT, and FUT8 and quantification normalized to WT HEK293T. Error bars for western blot quantification are SD for n = 3. ns, not significant; ∗p < 0.05 for a Student’s t test.(E) Predicted relative distribution of MAN1 following the fitting of WT and Cog4KD HeLa glycan profiles.(F) Predicted relative distributions of selected enzymes in WT and Cog4KO HEK293T cells in each cisterna normalized to the total predicted effective enzymatic rate for each enzyme for each cell line.(G) Airyscan confocal microscopy of GM130 and exogenous GalT-YFP in nocodazole-treated WT and Cog4KO HEK293T cell lines.(H) Pearson’s correlation coefficient for WT and Cog4KO HEK293T cells. Pearson’s correlation coefficients were calculated for each Golgi stack, and error bars are SD for n = 4 cells with 151 (WT) and 131 (Cog4KO) stacks. ∗∗∗p < 0.001 for a Student’s t test.See also Figures S3 and S6 and Table S3.

Fig 3: Schematic summary of the model proposed in this study.Drug-sensitive cancer cells responding to kinase inhibitors induce a core fucosylated, complex secretome of mostly <60 kDa proteins stimulating the clonal expansion of DR cancer cells, potentially contributing to disease relapse. Secretome core fucosylation is driven by the fucose salvage-SLC35C1-FUT8 axis. Among such secreted proteins that are heavily core fucosylated is the antioxidant PON1. Prior to secretion, PON1 is differentially fucosylated at multiple sites with the sequon at N253 critical for protein folding and stability. PON1 core fucosylation is tightly regulated by PON3 in the golgi, resulting to a more stable, degradation-resistant PON1 with a rewired enzyme activity in the secretion. Minority DR clones respond to the globally fucosylated TIS or core fucosylated secretome PON1 via activation of TF and gene effectors important for UPR, cell communication regulation and inflammatory niche neutralization.

Fig 4: Core fucosylation impacts PON1 folding and stability prior to secretion in therapy-resistant cancer cells.(A) Hypothetical model of N-glycosylation control of PON1 stability. (B) PON1-WT N-glycosylation site prediction using NetNGlyc 1.0, folding prediction using FoldIndex, and charge prediction using EMBOSS.>0.5 threshold score means significant glycosylation potential. Unfolded regions are depicted in red, folded regions in green. Positive charged is marked in red shades, negative charge in blue, and neutral charge in white. (C) Conservation of indicated PON1 sequons throughout species. (D) Closed conformation surface structure of PON1 (PDB ID: 1V04) highlighting arylesterase domain and predicted N-glycosylation sites and sequons. The 3D surface view was visualized using PyMOL. (E) N-glycan structural analysis of PON1 from our previous tandem MS/MS dataset. The m/z 1647.62 [(M+Na)+corresponding to GlcNAc2Man3+HexNAc2Hex1Fuc1] is the base peak (not visualized). Putative structure visualization of indicated monosaccharides and FUT8 substrate specificity were based on CID data and known glycobiology. (F) Prediction of PON1 stability, structural and functional properties upon indicated in silico N?G substitution at specific sequons using MutPred 2.0 and I-Mutant 3.0. Two N?G substituted sequons (N253G and N324G) with statistically significant potential of loss of N-glycosylation were chosen for validation experiments. (G) AAL blot analysis of PON1 immunoprecipitates from H1993-GR upon transfection with indicated PON1-WT, PON1-N253G, or PON1-N324G constructs for 36 hr. Representative of two independent experiments. Beside shows HLE analysis of secretome PON1 fucosylation and N-glycan release assay in AAL-enriched PON1 immunoprecipitates from H1993-GR upon similar transfection. Values are relative to PON1-WT (means ± SD of three biological replicates). ***p<0.001, Student’s t-test. (H) GDP-Fuc activity analysis of FUT8 in cross-linked FUT8 and PON1 co-immunoprecipitates from H1993-GR upon transfection with constructs as in G. Values indicate luminescence units and are relative to control reaction (means ± SD of three biological replicates). **p<0.01, ***p<0.001, two-tailed Mann–Whitney U test. (I) Immunoblot analysis of PON1 expression in H1993-GR upon transfection with constructs as in G. Lysates were exogenously treated with or without indicated trypsin concentration. Representative of two independent experiments. (J) ELISA analysis of PON1 expression in H1993-GR upon transfection with constructs as in G. Golgi/ER fractionated cell lysates were exogenously treated with or without indicated trypsin concentrations. Values are relative to no treatment (means ± SD of three biological replicates). *p<0.05, **p<0.01, ***p<0.001, Student’s t-test. NS, not significant. (K) EZClick labeling analysis of polypeptide synthesis in H1993-GR upon transfection with constructs as in G and treated with or without 25 µg/mL CHX concentrations for indicated times. Values indicate raw fluorescence units (means ± SD of two biological replicates). For statistical analysis, Student’s t-test was used. NS, not significant. (L) Immunoblot analysis of PON1 expression in H1993-GR upon transfection with constructs as in G and treated with or without 25 µg/mL CHX for indicated times. GAPDH was used as a loading control. Blot intensity quantification of the lower PON1 kDa isoform is shown. Representative of two independent experiments. (M) ELISA analysis of secretome PON1 expression in H1993-GR upon transfection with constructs as in G. Values are relative to WT (means ± SD of three biological replicates). *p<0.05, ***p<0.001, Student’s t-test. Figure 4—source data 1.Uncropped blots and gels (labeled and unlabeled) for Figure 4G, I and L. Figure 4—source data 2.Uncropped gels (labeled and unlabeled) for Figure 4—figure supplement 1B and D, 1 G, and 1 H.

Fig 5: Secretome fucosylation is a post-translational mechanism associated with targeted therapy resistance in cancer.(A) Dot plot visualization of correlation between indicated FUT gene expression and drug response per cancer type screened in GDSC and CCLE. Size of circle refers to mean log2 gene expression while color corresponds to Spearman’s rank coefficients. Per-sample estimates of area under the fitted dose-response curve were used as a metric of drug response per cell line. Only statistically significant correlations are shown (p<0.05). Beside is a relative mean proportion of mutational signatures of all FUT genes per cancer type queried in GDSC and CCLE. FUT mutations were classified as “GDP-Fuc binding site mutations” if any mutations (amino acid change) occurred near (±5 amino acid position) or at the annotated GDP-Fuc binding sites. Domain information was queried in UniProt. Spearman’s rank coefficients (correlation between FUT expression and drug response) were calculated in cell lines carrying these mutations as opposed to those that do not (‘others’). Note that FUT4 data is not available in the GDSC dataset. (B) Positive correlation between FUT8 gene expression and resistance to drugs grouped accordingly per target process in GDSC. Data from both GDSC and CCLE are summarized. Color represents Spearman’s rank coefficients per target process. Only statistically significant correlations are shown (p<0.01). Bars indicate the number of drugs per class while the size of the circle corresponds to relative Spearman’s rank coefficients per drug. Beside is a proportion of drug categories (GDSC classification) from all drugs with resistance profiles positively correlated with FUT8 expression. (C) Dot plot visualization of differential (TCGA primary tumor versus paired normal) CCS and overlapped glycosylation gene set expressions (including O-/N-linked glycosylation) per cancer type. Size of circle refers to adjusted -log10 p-value while color corresponds to log2 fold change in expression. Statistically significant (p<0.05) Spearman’s correlation between drug sensitivity and CCS or glycosylation expression derived from GDSC are shown as a heatmap. In total, 169 drug profiles were queried; 33 are targeted, and 10 are cytotoxic drugs. (D) Dot plot visualization of mean promoter methylation fraction 1 kb upstream of the TSS of indicated FUT genes per cancer type from CCLE RRBS dataset. Size of circle refers to the number of screened cell lines while color corresponds to FUT promoter methylation. Only statistically significant changes are shown (p<0.05). Correlation between drug sensitivity and methylation is shown as heatmap as in C. (E) Schematic of secretome N-glycoprotein core fucosylation in the context of cancer TIS. (F) Preparation of patient- and cell culture-derived samples for capture and enrichment of fucosylated proteins and downstream fucosylation assays. (G) AAL blot analysis of total fucosylation in indicated crude patient sera prepared as in F. Representative of two independent experiments. Samples were originally performed in a single midi-SDS-PAGE format and blots were incised per sample group prior to incubations (samples #1–10, #11–16, and #17–22). Re-run of select samples in a single SDS-PAGE format, equal loading controls, and AAL specificity are presented in Figure 1—figure supplement 4. (H and I) AAL blot analysis of total fucosylation in indicated secretomes from sensitive cells (H) following treatment with or without indicated drugs for 48 hr or DR clones (I). Samples were prepared as in F. Representative of two independent experiments. Blot incisions per cell line or pair are shown. Equal loading controls and AAL specificity are presented in Figure 1—figure supplement 4. (J) Dot plot visualization of fucosylation characterization by sandwich ELLA in indicated secretomes from sensitive cells or DR clones prepared as in F following treatment with or without indicated drugs for 48 hr. Color indicates fold change values relative to DMSO or parental (means ± SD of three biological replicates) while size indicates p values; Student’s t-test. NS, not significant. Schematic of in-house developed sandwich ELLA and representative colorimetric output are displayed on the left panel. (K) Fucosylation characterization by sandwich ELLA in indicated crude patient sera (top) or tissues (bottom) prepared as in F. Each point indicates mean absorbance at 450 nm from two to three replicates. Representative of two independent experiments. ROC curves are shown. For statistical analysis, the nonparametric Kruskal-Wallis test was used. (L) Representative confocal images of indicated sensitive cells or DR clones stained for: core fucosylation (fluorescein-conjugated AAL; yellow) and nuclei (DAPI; blue) or golgi (RCAS1; green), core fucosylation(fluorescein-conjugated AAL; red), and nuclei (DAPI; blue).Cells were treated with and without 1 µM gefitinib or 0.1 µM vemurafenib for 48 h. The co-localization histogram plot of the indicated line is shown. Representative of two independent experiments. (M) Fucosylation characterization by sandwich ELLA in indicated golgi-fractionated cell lysates from sensitive cells or DR clones prepared as in schematic following treatment with or without gefitinib (for H1993) or vemurafenib (for A375) IC50 for 48 hr and filtered according to their indicated nominal molecular weight limit (NMWL). Values are relative to untreated parental (means ± SD of two biological replicates). For statistical analysis, a Student’s t-test was used. Figure 1—source data 1.Uncropped blots (labeled and unlabeled) for Figure 1G, H and M. Figure 1—source data 2.Uncropped blots and gels (labeled and unlabeled) for Figure 1—figure supplement 4A,B and C. Figure 1—source data 3.Uncropped blots (labeled and unlabeled) for Figure 1—figure supplement 11C.