Comparative site-specific N-glycoproteome analysis reveals aberrant N-glycosylation and gives insights into mannose-6-phosphate pathway in cancer

Abstract

N-glycosylation is implicated in cancers and aberrant N-glycosylation is recognized as a hallmark of cancer. Here, we mapped and compared the site-specific N-glycoproteomes of colon cancer HCT116 cells and isogenic non-tumorigenic DNMT1/3b double knockout (DKO1) cells using Fbs1-GYR N-glycopeptide enrichment technology and trapped ion mobility spectrometry. Many significant changes in site-specific N-glycosylation were revealed, providing a molecular basis for further elucidation of the role of N-glycosylation in protein function. HCT116 cells display hypersialylation especially in cell surface membrane proteins. Both HCT116 and DKO1 show an abundance of paucimannose and 80% of paucimannose-rich proteins are annotated to reside in exosomes. The most striking N-glycosylation alteration was the degree of mannose-6-phosphate (M6P) modification. N-glycoproteomic analyses revealed that HCT116 displays hyper-M6P modification, which was orthogonally validated by M6P immunodetection. Significant observed differences in N-glycosylation patterns of the major M6P receptor, CI-MPR in HCT116 and DKO1 may contribute to the hyper-M6P phenotype of HCT116 cells. This comparative site-specific N-glycoproteome analysis provides a pool of potential N-glycosylation-related cancer biomarkers, but also gives insights into the M6P pathway in cancer.

Fig. 1. Combination of Fbs1-GYR enrichment and timsTOF enables large-scale N-glycoproteome identification from HCT116 and DKO1 cell lysates.

a The workflow for N-glycoproteome identification. b Fbs1-GYR enables more than 100-fold N-glycopeptide enrichment. Pre, pre-enrichment, samples were not enriched by Fbs1-GYR; Fbs1, samples were enriched by Fbs1-GYR. The data shown is from MS/MS with 45 min LC. c Filtering with Byonic score ≥ 300 improves the MS accuracy. The data shown is a combination of six MS/MS runs with 45 min LC and 90 min LC. Detailed information can be found in Supplemental Data 1–4.

Fig. 2. Summary and visualization of the large-scale N-glycoproteomes of HCT116 and DKO1 cells.

a Comparison of overall numbers of unique intact N-glycopeptides, N-glycosites and N-glycoproteins identified from HCT116 and DKO1 cells. b Western blots of P15144|AMPN, P16422|EPCAM, Q9Y5Y6|ST14, P04066|FUCO, Q9Y4L1|HYOU1, and P10253|LYAG. “g” and “dg” denote “glycosylated protein” and “deglycosylated protein”, respectively. β-actin blot serves as a protein loading control. Representative blots are shown from n = 3 experiments. c Comparison of the protein relative levels between HCT116 (H) and DKO1 (D) cells determined by Western blots and N-glyco PSM. The Western blots of P11717|MPRI (CI-MPR) and P20645|MPRD (CD-MPR) are shown in Fig. 7c, d. Relative protein levels determined by Western blots were an average from three biological samples. N-glyco PSM of each protein were retrieved from the N-glycoproteomes (Supplemental Data 5). d, e P04066|FUCO is used as an example to demonstrate differences in site-specific N-glycosylation between HCT116 and DKO1 cells. d is a full summary with all N-glycan compositions, and e is a simplified summary with N-glycan classes. The green row indicates N-glycoprotein identity (Uniport # and protein name). N-glycosites in the protein are listed below N-glycoprotein identity in yellow rows. Under the N-glycosite, N-glycan compositions are listed (d) and (e) lists the N-glycan classes attached to each N-glycosite. N-glyco PSM from DKO1 or HCT116 are listed next to protein identities, N-glycosites, and N-glycans. Relative glycosylation abundance of a site-specific N-glycosylation is calculated by N-glyco PSM of a site-specific N-glycosylation divided by the total N-glyco PSM of the corresponding protein. HM, PauciM, and M6P indicate high mannose, paucimannose, and mannose-6-phosphate N-glycans, respectively. The entire site-specific N-glycoproteomes of HCT116 and DKO1 can be found in Supplemental Data 5.

Fig. 3. Comparative analysis of site-specific N-glycosylation between HCT116 and DKO1 cells.

a Scatter plot to divide all the detected N-glycoproteins into 4 groups based on the N-glyco PSM numbers in HCT116 cells (y-axis) and DKO1 cells (x-axis). The dotted line is a mark of N-glyco PSM = 5. Green dots represent a group of proteins with N-glyco PSM > 5 in both DKO1 and HCT116 cells. The rest of proteins with N-glyco PSM ≤ 5 in one of the cell lines are further divided into three groups. Red dots represent HCT116-specific proteins where N-glycosylation is at least five times greater in HCT116 cells than in DKO1 cells; purple dots represent DKO1-specific proteins where N-glycosylation in DKO1 is at least five times greater than in HCT116 cells; gray dots represent the proteins with N-glycosylation ratio between HCT116 and DKO1 < 5. b The N-glyco PSM in HCT116 (H) and DKO1 (D) of the proteins represented by red dots and purple dots (in a) are listed. Cancer driver score (CDS) of the corresponding genes is also evaluated. The proteins with CDS ≥ 1 are highlighted in blue. The site-specific N-glycosylation of P16422 | EPCAM is presented as an example in these two groups. c Volcano plot to show the comparative analysis of site-specific N-glycosylation of the proteins with N-glyco PSM > 5 in both HCT116 and DKO1 cells (green dots in a). A total of 1554 site-specific N-glycosylation data points are plotted according to log2 (the fold change of HCT116/DKO1) in x-axis and log10 (the difference of N-glyco PSM between DKO1 and HCT116) in y-axis. The orange and blue dots indicate the upregulated (≥3-fold increase and ≥10 N-glyco PSM difference) site-specific N-glycosylation in HCT116 and DKO1 cells, respectively. The identities of 8 orange or blue dots are indicated as examples. Detailed calculations can be found in Supplemental Data 6, Red tab. d Pie charts to categorize by N-glycan classes the upregulated site-specific N-glycosylation in HCT116 and DKO1 cells. The numbers within the pie charts indicate the numbers of upregulated site-specific N-glycosylation in the respective N-glycan class. e, f list the detailed information of the upregulated site-specific N-glycosylation in HCT116 cells (orange dots in c) and DKO1 cells (blue dots in c), respectively. Searchable tables can be found in Supplemental Data 6, Orange tab or Blue tab).

Fig. 4. Comparative analysis of N-glycosylation at the cell level and the protein level.

a Distribution of 8 N-glycan classes in HCT116 and DKO1 cells. Data were summarized from 6 MS/MS runs. The PSM of a specific N-glycan class in an MS/MS run was normalized to the total N-glyco PSM of the MS/MS run, and the values in percentage were plotted to the y-axis. ns: not significant; **, P ≤ 0.01; ****, P ≤ 0.0001, mean with SD, t-test, two-tailed, n = 6. b The N-glycoproteins with significant differences in high mannose N-glycans (HM), complex or hybrid N-glycans (C or Hyb), Sialylated N-glycans (Sia), Fucosylated N-glycans (Fuc), or Fucosylated and Sialylated N-glycans (Fuc/Sia) in HCT116 and DKO1 cells. The glycan ratio between HCT116 and DKO1 is ≥3. The relative glycosylation abundance (RGA) of N-glycans in DKO1 is presented with a negative symbol as a result of the plotting method. Ignore the negative sign and read the absolute value. A more detailed analysis can be found in Supplemental Data 7.

Fig. 5. Paucimannosidic N-glycosylation in HCT116 and DKO1.

a A list of the N-glycoproteins with high paucimannosidic N-glycosylation (≥ 10% of the protein total N-glyco PSM in either HCT116 or DKO1). RGA of paucimannose in DKO1 is presented with a negative symbol as a result of the plotting method. Ignore the negative sign and read the absolute value. b The gene ontology analysis of the paucimannose-rich N-glycoproteins with the cellular location indicated. 80% (45 out of 56 proteins tested), and 73% (41 out of 56 proteins tested) of the paucimannose-rich N-glycoproteins are annotated to extracellular exosome (highlighted in a red box) and lysosome (highlighted in a dark green box), respectively. These protein identities are also highlighted in the red box and green box, respectively in (a). c The N-glycoproteins with a significant difference in paucimannosidic N-glycosylation in HCT116 and DKO1 cells. Q9Y4L1|HYOU1 is highlighted in purple for an example due to its large total N-glyco PSM (2659 in DKO1 and 1269 in HCT116) and a 169.4-fold change in paucimannose glycosylation. A more detailed analysis can be found in Supplemental Data 8.

Fig. 6. M6P modification is more than 5-fold higher in HCT116 cells than that in DKO1 cells.

Comparison of overall (a) and individual (b) M6P N-glycosylation in HCT116 and DKO1 cells detected by mass spectrometry. Please note (a) in this figure is derived from Fig. 4a and is presented to show the correlation between M6P modification levels detected by mass spectrometry and those observed by Western blot analyses shown in (c) and (d) in this figure. RGA of M6P in DKO1 is presented with a negative symbol as a result of the plotting method. Ignore the negative sign and read the absolute value. #, O60911|CATL2 was not detected in HCT116 cells. c, d Western blots show M6P modification is significantly higher in HCT116 cells (H) than in DKO1 cells (D). Representative Western blots are shown in the c), and the quantification summary of M6P modification from three independent Western blots is shown in (d). In the Western blot, soluble cation-independent M6P receptor (sCI-MPR) conjugated with HRP (illustrated) was used to probe M6P level in HCT116 and DKO1 cell lysates. The M6P binding domains (Domains 3, 5, and 9) are highlighted in dark red. To demonstrate the specificity of sCI-MPR-HRP, 10 mM free M6P was added to a parallel Western blot to compete off sCI-MPR binding to M6P-modified N-glycoproteins. PNGase F deglycosylated cell lysates were also used as negative controls. β-actin blot serves as a protein loading control. H, HCT116; D, DKO1; M, protein standard marker; **, P < 0.01; ****, P < 0.0001, mean with SD, t-test, two-tailed, n = 6 in (a), n = 3 in (d).

Fig. 7. Insight into the mechanism of hyper-M6P modification in HCT116 cells relative to DKO1 cells.

Comparison of the site-specific N-glycosylation of M6P receptors, CI-MPR (a) and CD-MPR (b) in HCT116 and DKO1 cells. 15 extracellular domains, transmembrane domain (TM), and cytosolic tail (CT) are illustrated in CI-MPR. CD-MPR contains one extracellular domain, TM and CT. The domains that interact with M6P moiety are in purple. The N-glycan that modifies the N-glycosite is listed on the top of the bar. c Western blots to determine the protein level of CI-MPR and CD-MPR in HCT116 and DKO1 cells. d Quantification of the Western blots from three experiments. e qRT-PCR to determine the relative mRNA level of CI-MPR and CD-MPR in HCT116 and DKO1 cells. f A proposed model to explain hyper-M6P modification in HCT116 cells. In DKO1 cells, the normal function of M6P receptors is to sort M6P-modified lysosomal hydrolases to lysosomes, where M6P gets quickly dephosphorylated by phosphatases (ACP2/5). In contrast, aberrant N-glycosylation of CI-MPR in HCT116 cells may result in lower receptor stability as well as lower affinity to M6P. The lack of CI-MPR function in HCT116 may reroute the M6P-modified lysosomal hydrolases to the secretory pathway, where M6P is not dephosphorylated. Yellow color signifies an acidic environment. g, h Western blots show that HCT116 cells secrete more M6P-modified proteins to the medium than DKO1 cells. A representative Western blot is shown in (g). A Ponceau S-stained Western blot membrane is to show the protein level in the conditioned media. The quantification summary of M6P modification from three independent Western blots is shown in (h). 10× concentrated conditioned medium was probed by sCI-MPR-HRP. Same as in Fig. 6c, Western blots in the presence of 10 mM free M6P and PNGase F deglycosylated conditioned media were used as negative controls. H, HCT116; D, DKO1; M, protein standard marker; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001: ****, P ≤ 0.0001 mean with SD, t-test, two-tailed, n = 6 in (d), n = 24 in (e), n = 3 in (h).