PMCID: PMC6145689

 

    Legend: Gene, Sites, Suger

Section : Quantitative glycoproteomics enable exploration of the urinary N- and O-glycoproteome

Content :
  1. Urine samples (∼2 ml) from six PCa and six BPH patients were obtained and investigated in this study
  2. The PCa group included only patients with clinically validated adenocarcinoma with Gleason score 7 (GS 7), which denotes intermediate grade PCa relative to GS 6 representing low-risk PCa suitable for an active surveillance program, and the more advanced GS 8–10 representing high-risk PCa cases that generally are referred to immediate treatment
  3. The serum PSA levels of the PCa patient group were 7.17 ± 3.02 ng/ml (n = 6) and 8.02 ± 7.30 ng/ml in the BPH patient group (n = 6)
  4. As expected, no significant difference was observed in the serum PSA levels between these two patient groups (p ≥ 0.05, n = 6, unpaired two-tailed t-test, Figure 1A) highlighting the shortcoming of this biomarker to accurately stratify PCa and BHP patients
  5. Lower yields of total urinary protein were observed from two biological replicates from both the PCa and BPH groups; hence, these protein-poor samples were combined (1:1, w/w) forming five biological replicates (n = 5) from each condition for comparative glycoproteomics
  6. Given the reasonably large protein starting material (∼100 µg/replicate), we applied a multi-faceted quantitative LC-MS/MS-based glycoproteomics strategy to obtain a deep coverage of the urinary glycoproteome and investigate for differentially abundant glycoproteins in PCa and BPH urine
  7. In short, after protein extraction, concentration and tryptic digestion of the 10 samples, the resulting peptide mixtures were labeled with an isobaric amine-reactive TMT10plex – peptides from the five PCa samples were labeled with TMT 126, 127N, 127C, 128N and 128C and peptides from the five BPH samples were labeled with TMT 129N, 129C, 130N, 130C and 131
  8. All TMT-labelled peptide samples were then mixed in 1:1 (w/w) relationships prior to a multi-faceted glycopeptide enrichment and pre-fractionation sample processing (Figure 1B)
  9. The glycopeptide enrichment was performed using two complementary solid phase extraction (SPE) strategies i.e. titanium dioxide (TiO2) SPE and HILIC SPE enrichment, to reach the deepest possible coverage of the urinary glycoproteome
  10. The sialic acid-retaining TiO2 SPE [28, 29] was included since we previously showed that urinary N-glycoproteins are highly sialylated [27]
  11. HILIC SPE is widely recognized to capture all glycopeptides displaying a minimum degree of hydrophilicity regardless of the nature of their conjugated glycan structure [30, 31]; however, this enrichment method may come short of quantitatively capturing O- and N-glycopeptides of very low hydrophilicity (unpublished observation)
  12. The TiO2 and HILIC SPE enriched glycopeptides were 1) analyzed directly in their intact form by high resolution LC-MS/MS on a Orbitrap Q-Exactive HF with HCD fragmentation , which provided information of the peptide carrier and site-specific glycan heterogeneity, or 2) treated simultaneously with N-glycosidase F ( PNGase F) and sialidase A to generate de-N-glycosylated and desialo-O-glycopeptides that are less challenging to characterize by LC-MS/MS
  13. Combining the latter approach with a similar Orbitrap Q-Exactive HF HCD based acquisition, provided N-glycosylation site information and site-specific glycan information of the O-glycosylation
  14. In addition, non-modified peptides that did not bind to the TiO2 SPE column (“flow through”) and PNGase F/sialidase-treated glycopeptides were pre-fractionated using off-line HILIC HPLC and the resulting fractions were analyzed by Orbitrap Q-Exactive HF HCD-MS/MS (Figure 1B)
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Complementary TiO2 and HILIC SPE-based glycopeptide enrichment provide deep coverage of the urinary glycoproteome

Content :
  1. Even without peptide pre-fractionation, the combined use of TiO2 and HILIC SPE facilitated the identification of 630 unique N-linked glycosylation sites belonging to 361 N-glycoproteins across the urine samples (Supplementary Figure 1A, and Supplementary Tables 1–2)
  2. Partial overlap between the TiO2 and HILIC SPE enrichment approaches were observed when measured by the proportion of unique N-glycosylation sites (48%, 304/630) and the corresponding glycoproteins (58%, 211/361) detected in both sample preparation techniques (Supplementary Figure 1B–1C)
  3. However, when assessed by the identified intact glycopeptides , the enrichment strategies displayed high complementarity: From a combined total of 954 N- and 210 O-glycopeptides identified with high peptide confidence (0% peptide FDR Pep-2D) [32] (Supplementary Figure 1D and Supplementary Tables 3–6), the degree of unique N- and O-glycopeptides identified from both the TiO2 and HILIC SPE preparations was 23% (216/954) and 10% (20/210), respectively (Supplementary Figure 1E–1F)
  4. Due to the unbiased enrichment of the hydrophilic glycopeptides , HILIC SPE showed, as expected, a higher capture efficiency of N-glycopeptides compared to TiO2 SPE
  5. In contrast, TiO2 retained a greater number of O-glycopeptides compared to HILIC SPE indicating a high degree of sialylation of the O-glycopeptides and an inability of HILIC SPE to retain such lowly hydrophilic glycopeptides
  6. We also investigated the glycosylation features of the TiO2 and HILIC SPE enriched intact N-glycopeptides (Supplementary Figure 1G, Supplementary Tables 3–4) and O-glycopeptides (Supplementary Figure 1H, Supplementary Tables 5–6)
  7. In agreement with our previous study [27], the urinary N-glycoproteins were predominantly carrying complex and hybrid type N-glycans with a high level of fucosylation and/or sialylation accounting for ∼85% of all identified intact N-glycopeptides (Supplementary Figure 1G)
  8. Sialylated O-glycopeptides were also abundant (∼70%) and were, as expected, efficiently enriched using TiO2 SPE (Supplementary Figure 1H)
  9. Collectively, these data clearly demonstrated that the parallel usage of HILIC and TiO2 SPE is highly beneficial to increase the coverage of the urinary N- and O-glycoproteome
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : HILIC HPLC pre-fractionation enhances the coverage of the urinary glycoproteome

Content :
  1. In attempts to reach an even greater coverage of the urinary glycoproteome, off-line HILIC HPLC was implemented in the workflow to pre-fractionate the TiO2 SPE retained glycopeptide fraction
  2. Endo- (PNGase F) and exo- (broad specificity sialidase ) glycosidase treatments were also introduced to render the HILIC peptide fractions more amendable to LC-MS/MS detection; it is commonly known that such de-N-glycosylated and desialo-O-glycopeptides are easier to identify than their native counterparts, albeit with less structural information obtained [33, 34]
  3. Interesting, the de-N-glycosylated peptides eluted early in the HILIC HPLC gradient and were efficiently separated from the highly retained desialo-O-glycopeptides (Figure 2A)
  4. When combined with Orbitrap Q-Exactive HF HCD-MS/MS detection, the HILIC HPLC pre-fractionation facilitated an unprecedented coverage of the urinary glycoproteome by the identification of a total of 1,217 N-glycosylation sites from 696 N-glycoproteins and the detection of 887 desialo-O-glycopeptides from 160 O-glycoproteins (Figure 2B and 2C and Supplementary Tables 7–8)
  5. In total, 1,310 N-glycosylation sites were identified by combining the de-N-glycosylated peptides enriched by HILIC SPE and TiO2 SPE and pre-fractionated by HILIC HPLC (Supplementary Tables 2, 7)
  6. We then compared our coverage of the urinary N-glycosylation sites in PCa and BPH urine to a SWATH based identification of the N-glycosylation sites reported in normal prostate, non-aggressive, aggressive and metastatic PCa tumor tissues [21]; a substantial overlap of 337 unique glycosylation sites (25%) and 321 N-gycoproteins (44%) was observed (Supplementary Figure 2A–2B)
  7. This result indicates that almost half of the glycoproteins expressed in prostate tissues appear to be mirrored in the excreted urine
  8. We then compared our urinary N-glycoproteome coverage to the N-glycoproteome of two PCa cell lines LNCap and PC3 reported by Shah et al. [20]
  9. In total, 134 N-glycoproteins (11%) were found to be common between our two data sets (Supplementary Figure 2B)
  10. Interestingly, the urinary glycoproteins identified in our study had a closer resemblance to the glycoproteome of PCa tissues than the glycoproteome of the PCa cell lines
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Quantitative comparison of PCa-associated glycan-com positions and lectin blotting analysis

Content :
  1. We compared the relative abundance of all intact N-glycopeptides grouped by glycan class and terminal features from PCa and BPH urine
  2. In doing this, no significant differences were observed between PCa and BPH with respect to the overall relative abundance of high mannose , afucosylated/asialylated complex/hybrid (C/H), fucosylated complex/hybrid (C/H\+Fuc) sialylated complex/hybrid (C/H\+NeuAc) and fucosylated/sialylated complex/hybrid (C/H\+Fuc/Sia)
  3. However, increased intensity of Fuc-containing intact desialo-O-glycopeptides was observed (T-test, p < 0.05) (Supplementary Figure 3)
  4. Comparative lectin blotting was also performed using concanavalin A (Con A), Maackia amurensis lectin ( MAL ), wheat germ agglutinin (WGA), Aleuria aurantia lectin (AAL) and Ricinus communis agglutinin (RCA) to assess the level of a range of glycoepitopes in PCa and BPH urine in an orthogonal manner
  5. After protein normalization, significantly increased levels of AAL reactivity were observed for the PCa urinary glycoproteome indicating a higher degree of core fucosylation (Figure 3)
  6. Slightly higher Con A reactivity was also observed in PCa relative to BPH urine, a feature not reflected in the quantitative LC-MS/MS data of high mannose glycopeptides
  7. However, Con A may also cross-react with paucimannosidic and biantennary complex glycoproteins possibly explaining this discrepancy [35]
  8. Based on these observations, we therefore sought to determine whether site-specific glycosylation instead would display a higher potential of discriminating between urine from PCa and BPH patients
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Comparative quantitative glycoproteomics to identify unique site-specific glycosylation signatures in PCa urine

Content :
  1. The datasets of de-N-glycosylated peptides and desialo-O-glycopeptides generated after TiO2 SPE-based glycopeptide enrichment and HILIC HPLC pre-fractionation as well as the intact N-glycopeptides identified after HILIC SPE and TIO2 SPE-based glycopeptide enrichments was used to quantitatively compare the glycosylation profile of PCa and BPH urine using TMT reporter ion intensities
  2. In addition, to assess for altered protein levels, the intensities from the non-modified peptides (TiO2 SPE flow-through fraction) were also compared between PCa and BPH
  3. The ion intensities of all glycopeptides or non-modified peptides were summed, log2 transformed, and median normalization was applied
  4. One BPH sample showed different outlier behavior when comparing all samples using PCA and was therefore removed
  5. To reveal statistical differences in the abundance of glycopeptides from PCa (n = 5) and BPH groups (n = 4), the Limma R package was used [36]
  6. The total number of differentially abundant intact N-glycopeptides and desialo-O-glycopeptides , de-N-glycosylated peptides and non-modified peptides , are shown in Figure 2D and Supplementary Table 9
  7. We then performed unsupervised PCA to assess if these glycopeptides and non-modified peptides were able to discriminate PCa from BPH
  8. Excitingly, all differentially abundant peptides , including the intact N- and O-glycopeptides , de-N-glycosylated peptides and non-modified peptides , were able to almost completely separate the PCa and BPH groups (Supplementary Figure 4A)
  9. Using only the non-modified peptides that were significantly changing in abundance, an overlap between PCa and BPH groups was observed
  10. Interestingly, only the differentially abundant intact N-glycopeptides and desialo-O-glycopeptides separated completely the PCa and BPH groups in the PCA
  11. As a complementary data representation style, we performed clustering analysis using Euclidean distance and a heat map to visualize the altered levels of intact N-glycopeptides and desialo-O-glycopeptides
  12. Two large clusters of glycopeptides were observed containing 120 over-represented glycopeptides and 106 under-represented glycopeptides in PCa compared to BPH (Supplementary Figure 4B)
  13. These clusters perfectly separated the PCa samples from the BPH samples
  14. These results showed that the urinary glycoproteome provided better discrimination between the PCa and BPH groups than the non-modified peptides
  15. Parallel reaction monitoring (PRM), a targeted proteomics strategy [37], was performed to: 1) confirm the levels of some intact glycopeptides in PCa urine; 2) increase the number of peptide-spectrum matches ( PSM ) for the individual intact glycopeptides to improve the identification confidence and 3) improve the quantification accuracy and ion statistics of intact glycopeptides with aberrant levels in the PCa urinary glycoproteome
  16. The intact N-glycopeptide SVVAPATDG GLNLTSTFLR displaying the triantennary and core-fucosylated sialoglycoform HexNAc(5)Hex(6)Fuc(1)NeuAc(3) from prostaglandin-H2 D-isomerase ( PTGDS ) was present in 29 PRM-MS/MS spectra within the retention time 31.2–32.9 min
  17. The intact O-glycopeptide DLCNFNEQLENGGTSLSEK carrying the glycoform HexNAc(3)Hex(1)Fuc(1) from CD59 glycoprotein CD59 glycoprotein ( CD59 ) was present in 24 PRM-MS/MS spectra within the retention time 22.8–24.9 min
  18. In comparison, each of these two glycopeptides were only identified from two acquired MS/MS spectra in a data-dependent acquisition (DDA) mode (Supplementary Tables 4 and 6)
  19. The TMT reporter ion intensities from the identified spectra of each glycopeptide were summed and normalized to the sum of all TMT reporter ion intensities from each TMT10plex channel during the entire LC-MS/MS run
  20. Statistical differences were assessed by conventional t-test analyses
  21. The increased levels of the HexNAc(5)Hex(6)Fuc(1)NeuAc(3) N-glycopeptide from PTGDS and decreased levels of the HexNAc(3)Hex(1)Fuc(1) O-glycopeptide from CD59 in PCa compared to BPH urine was confirmed (Supplementary Figure 5)
  22. Example of annotated PRM-MS/MS spectra of these two intact N- and O-glycopeptides are shown in Supplementary Figure 6
  23. Importantly, seven non-modified peptides from CD59 and eight non-modified peptides from PTGDS were identified
  24. Most of these non-modified peptides were unaltered in abundance when comparing between the PCa and BPH urine (Supplementary Table 10)
  25. This indicates that the expression levels of these two glycoproteins are likely not altered in PCa, but that the glycosylation of them instead is the molecular feature undergoing regulation
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Intact N-glycopeptide marker panel for PCa diagnosis

Content :
  1. To provide a panel of confidently characterized PCa-associated intact glycopeptides , we selected only intact N-glycopeptides with significant changes in abundance, whose glycosylation sites were also identified in the de-N-glycosylated peptide fractions and whose glycan com positions were commonly known human serum glycoforms reported in UniCarbKB [38]
  2. Besides, we considered only intact glycopeptides displaying fold changes greater than 1.3 or less than 0.77 when comparing PCa to BPH urine
  3. A total of 56 intact N-glycopeptides matched these criteria (Table 1)
  4. Information of the relative levels of their corresponding glycoproteins was obtained using quantitative LC-MS/MS data of the non-modified peptide fraction (Supplementary Table 10)
  5. Table 1 shows site-specific glycoforms that are more/less abundant in PCa independently of the protein level
  6. For example, protein AMBP AMBP ( AMBP ) was clearly more abundant at the protein level (as detected by ten non-modified peptides with increased levels, Supplementary Table 9) and site-specific glycan regulation was observed by the increased relative abundance of three intact glycopeptides covering two N-glycosylation sites in PCa urine (Table 1)
  7. On the other hand, apolipoprotein D ( APOD ) was less abundant in PCa (two non-modified peptides with decreased levels, Supplementary Table 9), but increased levels of site-specific glycosylation (i.e. HexNAc(7)Hex(6)) were observed
  8. In contrast, LTF was found to be both less abundant at the protein level (six under-represented non-modified peptides in PCa) and with respect to the N-glycosylation site occupancy as evaluated by the under-representation of the de-N-glycosylated peptide TAGWNIPMGLLFDQTGSCK and the site-specific hybrid glycoform HexNAc(4)Hex(7)NeuAc(1) in PCa
  9. Most of the glycoproteins showed no changes at the protein level, but were aberrantly changing with respect to their site-specific glycosylation levels e.g. CLU , LOX , SERPINA4 , WFDC2 , LAIR1 , PIK3IP1 (Table 1)
  10. This indicates that the glycosylation machinery, and not the protein translation and secretion, of the cells producing these cancer-associated glycoforms is significantly impacted by the malignant processes associated with PCa
  11. Some glycopeptides of the urinary glycoproteome were putatively identified with terminal NeuGc
  12. Although being a non-human type of sialic acid, NeuGc may, however, be a component of human glycoproteins arising from exogenous building blocks in particular in glycoproteins of cancer cell origin [39]
  13. These putative NeuGc-containing glycopeptides were manually checked for the presence of the corresponding oxonium ions i.e. m/z 308/290 in the corresponding HCD-MS/MS spectra
  14. All of the glycopeptides showed diagnostic ions for NeuAc (m/z 274/290) instead of NeuGc
  15. As verified by manual annotation, the NeuAc-containing glycoforms were found to be the correct glycan com position in all of these cases, Table 1
  16. Besides, the com position HexNAc(4)Hex(5)NeuGc(1) was found to be incorrectly identified due to an concomitant oxidation of the peptide
  17. In fact, the correct com position was HexNAc(4)Hex(5)NeuAc(1)
  18. PCA was performed using a panel of 56 intact N-glycopeptides identified with high confidence
  19. Complete segregation was observed between PCa and BPH (Figure 4A)
  20. Interestingly, both the abundance of the glycosylation sites (as measured by the de-N-glycosylated peptides ) and proteins (measured by the non-modified peptides ) from the glycoproteins that were present at different levels in PCa vs BHP were not able to discriminate between the two diseases (Figure 4B–4C)
  21. These results indicate that most PCa-specific glyco-features are not due to quantitative changes in the protein or site occupancy level, but arise from altered abundance of specific glycoforms of the individual urinary glycoproteins from the PCa patients
  22. The PCa-specific panel of 56 intact N-glycopeptide was also visualized by a heat map
  23. Unsupervised clustering was able to accurately separate the PCa and BPH donors groups (Figure 4D)
  24. The N-glycopeptide panel was also evaluated by ROC curve analysis
  25. The AUC was calculated using combinations of 3, 5, 10, 20, 28 and 56 selected N-glycopeptides within the entire glycopeptide panel (56 N-glycopeptides )
  26. Using 20, 28 or 56 intact N-glycopeptides , an AUC of 1 was obtained demonstrating high specificity and sensitivity of these urinary glycopeptides in the discrimination between the PCa patients from the BHP patients (Figure 4E)
  27. The relationship between the glycoproteins that carried aberrant N-glycosylation in PCa relative to BHP was explored in a protein-protein network analysis to search for over-represented biological functions and pathways
  28. Significant enrichment of pathways pertaining to the complement and coagulation cascade (KEGG pathway) and a regulation of immune system response (Gene Ontology Biological process) were observed in the network of glycoproteins carrying aberrant glycosylation in PCa (Figure 4F)
  29. In support of our finding, many of the glycoproteins in our panel were already shown in previous studies to be altered in PCa or involved in tumor growth and development
  30. For example: AMBP for PCa diagnosis [40], CD59 associated with PCa progression [41–43], CLU as a therapeutic target against PCa [44] and predictor for PCA recurrence [45], and PTGDS was found in increased concentration in urine from PCa patients [40]
  31. The association of these glycoproteins to PCa in these previous studies was established based on the protein expression level
  32. HexNAc(5)Hex(6)Fuc(1)NeuAc(2)NeuGc(1), HexNAc(5)Hex(6)Fuc(1)NeuAc(1)NeuGc(1), HexNAc(5)Hex(6)Fuc(1)NeuAc(2)NeuGc(1), HexNAc(4)Hex(5)Fuc(1)NeuAc(1)NeuGc(1) HexNAc(5)Hex(7)NeuAc(3), HexNAc(5)Hex(7)NeuAc(2), HexNAc(5)Hex(7)NeuAc(3), HexNAc(4)Hex(6)NeuAc(2), In contrast, we show here for the first time, that site-specific glycan com positions are also altered in this set of glycoproteins in PCa
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : TMT label-assisted quantitative glycoproteomics of urinary glycoprotein from PCa and BPH patients (A) The serum PSA levels of PCa (n = 6) and BPH (n = 6) patients showed no statistical difference (p > 0.05, unpaired two tailed t-test)

Content :
  1. (B) Overview of the quantitative glycoproteomics workflow employed in this study
  2. Peptides were generated from extracted urinary proteins from each patient, labelled with isobaric TMT tags and mixed 1:1 (w:w)
  3. The glycopeptides were selectively enriched using either TiO2 or HILIC SPE and the resulting fractions were either analyzed directly or after simultaneous endo- and exoglycosidase treatment by LC-MS/MS
  4. Former N-glycosylated peptides generated by PNGase F (referred to as de-N-glycosylated peptides ) and desialo-O-glycosylated peptides generated by sialidase and non-modified peptides were pre-fractionated using off-line HILIC HPLC; the resulting fractions were analyzed by LC-MS/MS
  5. Different information was obtained from the analysis of intact glycopeptides (red), de-N-glycopeptides (green) and non-modified peptides (blue)
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Deep coverage of the urinary N- and O-glycoproteome using HILIC HPLC pre-fractionation reveals quantitative glycoproteome differences in PCa and BPH (A) Distribution of de-N-glycopeptides and intact desialo-O-glycopeptides identified across the fractions arising from the HILIC HPLC pre-fractionation of TiO2 SPE enriched peptides of urinary proteins

Content :
  1. (B) Total and unique de-N-glycopeptides/N-glycoproteins and (C) intact desialo-O-glycopeptides/O-glycoproteins after HILIC HPLC pre-fractionation
  2. (D) Overview of the differentially abundant intact N-glycopeptides (from TiO2 and HILIC SPE enrichment), desialo-O-glycopeptides , de-N-glycosylated peptides and non-modified peptides identified in the glycoproteome of PCa relative to BHP urine
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Lectin blotting analysis of PCa and BPH urine Proteins from PCa and BPH urine (20 ug) were separated on SDS-PAGE gels, transferred to PVDF membranes and blotted against Con A (A), MALII (B), WGA (C), RCA (D), AAL (E) lectins

Content :
  1. The intensity of each lane after lectin blotting was normalized by the intensity of the respective sample stained by Coomassie (F)
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : Intact N-glycopeptides as candidate biomarkers for PCa detection (A) Principal component analysis (PCA) using a panel of 56 N-glycopeptides with aberrant levels in PCa relative to BHP (Table 1)

Content :
  1. (B) PCA using a set of 27 deamidated peptides covering the same glycosylation sites as identified with intact N-glycopeptide analysis (Table 1)
  2. (C) PCA using a set of non-modified peptides from the glycoproteins that carried the panel of differentially abundant intact N-glycopeptides (Table 1)
  3. (D) Clustering of intact N-glycopeptides displaying altered expression in PCa is shown as a heat map after applying Euclidean distance
  4. (E) ROC curve analysis was performed based on partial least squares discriminant analysis (PLS-DA) using the panel of PCa-associated intact N-glycopeptides (Table 1)
  5. (F) Network analysis using the list of glycoproteins which carried the differentially abundant intact N-glycopeptides (Table 1)
  6. Significant enriched KEGG pathway complement and coagulation cascades (FDR < 0.05) and regulation of immune system response (Gene Ontology, FDR < 0.05) are colored in blue and red, respectively
*Output_Site_Fusion* (sent_index, protein, sugar, site):
Section : In this study, 1,310 de-N-glycosylated peptides , 954 intact N-glycopeptides and 887 desialylated but otherwise intact O-glycopeptides belonging to a total of 788 glycoproteins were identified

Content :
  1. This deep coverage should be evaluated in the light of past efforts in this space including a study by Halim et al. [46], which showed the characterization of 58 N- and 63 O-glycopeptides from 53 urinary glycoproteins , Saraswat et al. [47] who reported 51 N-glycosylation sites belonging to 37 glycoproteins from the urinary exosomes and our own past study of urine i.e. Kawahara et al. [27] reporting 472 unique N-glycosylation sites covering 256 urinary glycoproteins
  2. Thus, this present study clearly represents the deepest coverage of the N- and O-glycoproteome of human urine to date
  3. One of the technical issues associated with urine analysis is the high inter- and intra-individual variability [6], which was previously shown to account for 48% and 66% of CV variation, respectively [48]
  4. In addition, the multi-step sample preparation and LC-MS/MS acquisition are likely sources of technical variation [49]
  5. We took advantage of the available TMT labeling technology, which not only allowed multiplexing of samples to reduce the LC-MS/MS acquisition time, but also minimized the technical variation from the sample handling [49], especially the peptide enrichment and pre-fractionation, while simultaneously facilitating more accurate quantification of glycopeptides and non-modified peptides [50]
  6. Due to the relative low number of biological replicates and high variance between the individuals, our study was limited to a non-stringent threshold of q < 0.25 for the peptide FDRs to determine differentially abundant glycoproteins in PCa and BPH urine
  7. However, by validating the levels of specific intact glycopeptides by PRM-MS/MS we significantly enhanced the confidence in these observations
  8. An important limitation of the chosen workflow is that, in complex mixtures, co-isolation of multiple precursor ions is the most acknowledged limitation of quantitative proteomics using isobaric labels
  9. This shortcoming reduces the precision and accuracy of the quantification [51, 52]
  10. It was demonstrated that due to this interference problem, the actual abundance ratios are typically underestimated [49]
  11. Using precursor intensity fraction (PIF) tool available in MaxQuant [53], we filtered peptides that clearly suffer from co-isolation of other peptide precursor ions
  12. Glycans facilitate and contribute to many different aspects of tumor progression, including proliferation, invasion, angiogenesis and metastasis [54]
  13. In cancer, protein glycosylation is dynamically regulated due to altered expression of glycan-modifying enzymes [55, 56]
  14. Wang et al. [57] showed high expression of α1,6-fucosyltransferase (FUT8 ) in tumor tissue from patients with metastatic and aggressive primary PCa and was positively correlated with PCa with high Gleason scores
  15. The N-glycosylation of proteins was demonstrated previously to be altered in urine [58], tissues [21] and serum [59] from PCa patients as well as in PCa cell lines [20]
  16. Interestingly, using meta-analysis of publicly available transcriptomic data from different PCa studies, Barfeld et al. [60] reported a concise 33 gene signature with biological enrichment for protein glycosylation, which discriminates between PCa and BPH across multiple transcript detection platforms and sample types
  17. Meany et al. [59] reported that direct analysis of PSA N-glycosylation in sera may be able to improve the sub-optimal specificity of PSA as a PCa marker
  18. Altered glycosylation of PSA isolated from PCa serum and/or seminal plasma relative to PSA glycosylation from controls was repeatedly observed by Tabares et al. [61], Llop et al. [62] and Ohyama et al. [63]
  19. Collectively, these studies clearly support the association of altered of protein glycosylation with PCa
  20. Although glycoproteomics studies have previously been used to explore PCa [20–22], the quantitative information relating to site-specific N- and O-glycosylation and the glycan com positions were not reported in those studies
  21. We demonstrated that differentially abundant glycopeptides discriminate more accurately between PCa and BPH groups than differentially abundant non-modified peptides
  22. Thus, we argue that a selected panel of glycopeptides , but not their corresponding non-glycosylated peptides from the same protein , may serve as candidate biomarkers for PCa detection
  23. We demonstrated that the malignant processes associated with PCa are more frequently reflected directly in the glycosylation of proteins as oppose to the aberrations in the expression level or the glycosylation site occupancy of those proteins
  24. Although we were limited to investigating a relative small patient cohort, the potential of intact glycopeptides to stratify patient groups can be explored in follow-up studies using larger cohort of patients once the glycoproteomics technology matures further allowing more streamlined data collection and interpretation of a larger patient cohort
  25. Among the PCa-associated glycopeptides , we targeted the N-glycosylation of prostaglandin-H2 D-isomerase ( PTGDS ) and the O-glycosylation of CD59 using PRM analysis
  26. Interestingly, Davalieva et al. [40] and Ahmad et al. [7] showed that these two glycoproteins are over-expressed in the urine of PCa patients
  27. Herein, we complement their observations by reporting that PTGDS and CD59 carry aberrant glycosylation at defined positions in PCa urine
  28. CD59 is a glycosyl-phosphatidylinositol ( GPI )-anchored cell membrane glycoprotein that inhibits complement-mediated cell lysis by preventing full assembly of the membrane attack complex (MAC) on host cells, and thus might confer immune resistance to tumor cells [42]
  29. High expression of CD59 protein was associated with higher Gleason scores and higher pT stages in PCa [41]
  30. It was demonstrated that prostasomes, which are secretory granules produced, stored, and released by the glandular epithelial cells of the prostate, had higher expression of CD59 than those of normal cells [64]
  31. CD59 glycosylation was previously characterized and several roles for the glycans, including spacing and orienting CD59 on the cell surface and protecting the molecule from proteases were proposed based on structural data [65]
  32. Prostaglandin-H2 D-isomerase was identified in human seminal plasma [66]
  33. Prostaglandin D2 (PGD2) and prostaglandin D2 synthase ( PTGDS ) is involved in the regulation of testis tissue differentiation [67, 68]
  34. PGD2, which is synthesized by PTGDS in many organs, has been implicated as a signaling molecule in the mediation or regulation of various biological processes, including tumorigenic process in PCa [69]
  35. It was shown that PTGDS activity is defined by post-translational modifications, and point mutation in a glycosylation site ( Asn51 ) inhibited L-PGDS-induced apoptosis and caspase 3 activity [70]
  36. Interestingly, we identified a significant lower abundance in the Asn51 glycoform HexNAc(4)Hex(5)Fuc(1)NeuAc(2) and HexNAc(5)Hex(6)Fuc(2) in PCa compared to BPH urine (Table 1)
  37. The quantitative analysis of intact glycoproteins opens novel avenues for assessing the dysregulation of glycoproteins in PCa and tie such information with aberrant glycoprotein function as drivers or passive by-products of tumour development and progression
  38. The quantification of specific peptides glycoforms is challenging due to the low ionization efficiency of glycan-modified peptides , as well as the lack of glycopeptides reference spectra
  39. Moreover, multiple glycoforms (micro-heterogeneity) are detected as separate glycopeptides , thus diminishing the signal intensity for each glycopeptide form
  40. Selected reaction monitoring ( SRM ) and parallel reaction monitoring (PRM) are currently the leading methods for targeted MS-based quantification of proteins
  41. Recently, SRM was applied to quantify candidate non-modified peptides biomarkers in expressed prostatic secretions (EPSs) from PCa and control patients [71]
  42. Song et al. [72] applied MRM to quantify intact glycopeptides in depleted human blood serum using the glycans oxonium ions as transitions
  43. However, to this date quantification of intact glycopeptides using targeted MS approaches is still poorly explored
  44. We showed PRM is a useful approach to quantify intact glycopeptides
  45. The number of MS/MS spectra recorder increased beyond a 20 fold by using PRM for targeted glycopeptide analysis, improving the accuracy of the identification and quantitation
  46. Integrating this approach with synthetic glycopeptides for normalization and absolute quantification would enhance value and utility of the PRM-MS/MS workflow even further
  47. We found the complement and coagulation cascades to be over-represented pathways in the set of urinary glycoproteins carrying differentially abundant site-specific glycoforms
  48. Interestingly, the same pathway was observed to be enriched in the meta-analysis from Barfeld et al. [60] which contained differentially expressed genes between localized PCa and benign prostate tissue, thus demonstrating that biological processes occurring in prostatic tissues may be reflected in urine
  49. Different coagulation disorders were reported during prostatic carcinoma evolution [73]
  50. The most frequent coagulation complication is the disseminated intravascular coagulation ( DIC ), which is a result of the release of pro-coagulant substances, such as tissue factors , into the bloodstream [74]
  51. This coagulation disorder was reported in many patients with PCa [75–77]
  52. We identified seven glycoproteins displaying aberrant site-specific glycosylation that are related to complement and coagulation cascade pathway
  53. The set of differentially abundant site-specific glycoforms belonging to these glycoproteins can be explored as candidate biomarkers for coagulation disorders that may be present during PCa progression
  54. Additionally, recent reports suggest that the complement elements can promote tumor growth in the context of chronic inflammation, immunosuppression, angiogenesis, and cancer cell signaling [78–80]
  55. Manning et al. also showed that human PSA , via its chymotrypsin-like serine protease activity, can modulate the complement system through degradation of iC3b to produce new C3 degradation fragments and through degradation of the complement protein C5 , thereby inactivating the complement cascade [81]
  56. In summary, this study reports several innovative approaches advancing our pursuit of reliable and sensitive urinary biomarker for PCa discovery and stratification from BPH by 1) providing the largest coverage of the urinary N- and O-glycoproteome to date, 2) providing a panel of 56 N-glycopeptides derived from aberrantly expressed glycoproteins in PCa urine representing an exciting collection of potential candidate biomarkers for discrimination of PCa from BHP and by 3) achieving confident structural characterization including the peptide carrier identity, site annotation, and glycan com position as well as quantitative validation of the panel of glycopeptide candidate markers by the application of innovative LC-MS/MS and PRM approaches of intact glycopetides
  57. This study opens new promising avenues for using urinary glycoproteins , a hitherto largely untapped resource, as candidate biomarkers for PCa detection
*Output_Site_Fusion* (sent_index, protein, sugar, site):
  • 35. 788, -, Asn51
  • 36. 788, the Asn51 glycoform HexNAc(4)Hex(5)Fuc(1)NeuAc(2), Asn51

 

 

Protein NCBI ID SENTENCE INDEX