PMID: PMC6428843-1-1

 

    Legend: Gene, Sites

Title : AI-ETD performance for intact glycopeptides

Abstract :
  1. AI-ETD provides information about both peptide and glycan components of intact N-glycopeptides by concomitantly capitalizing on two complementary modes of fragmentation in a single MS/MS event (Fig. 1a)
  2. The combination of vibrational activation and electron-driven dissociation is concurrent in both space and time when performing AI-ETD, which also reduces overhead time in MS/MS scans compared to other supplemental activation techniques (e.g., ETcaD and EThcD)
  3. This enables slightly more scans per unit time and, ultimately, more identifications (Supplementary Fig. 1), although other supplemental activation methods can still be quite valuable
  4. Indeed, EThcD has proven suitable for glycoproteome characterization in a number of recent studies, and future studies will likely focus on more systematic comparisons of multiple supplemental activation strategies that include AI-ETD
  5. AI-ETD generates extensive fragmentation along the peptide backbone, including mainly c- and z●-type products with some y-type fragments (100% sequence coverage in this example of the glycopeptide TN*SSFIQGFVDHVKEDCDR, where N* is the glycosite modified with a high mannose HexNAc(2)Hex(9) glycan)
  6. Importantly, product ions from peptide backbone cleavage largely retain the entire intact glycan, as is seen here in a series of doubly charged c-type fragments
  7. Minimal b-type product generation indicates that the majority of peptide backbone fragmentation comes from electron-driven dissociation via ETD rather than vibrational activation, matching observations from non-modified peptides and proteins
  8. That said, vibrational activation from infrared photons does impart enough energy to dissociate more labile glycosidic bonds, producing extensive series of Y-ion fragments (i.e., ions that have lost a portion of the non-reducing end of the glycan but retain the intact peptide sequence ) that provide details about glycan com position
  9. Furthermore, the infrared photoactivation of AI-ETD also generates complementary B-type fragments and other oxonium ions to indicate the presence of various sugar moieties
  10. Thus, the vibrational and electron-driven dissociation modes together provide information rich spectra for high quality glycopeptide identifications
  11. We leveraged AI-ETD for glycoproteomic data collection by triggering scans based on the presence of oxonium ions in HCD spectra ( HCD-pd-AI-ETD), which allowed straightforward comparisons of AI-ETD and HCD spectra
  12. AI-ETD produced more peptide backbone fragments and more Y-type fragments (mainly glycan fragments from the charge reduced precursors ) than HCD , while HCD produced more oxonium ions (Supplementary Fig. 2)
  13. Supplementary Figure 3 displays the percent of AI-ETD and HCD identifications that contain a number of common glycopeptide Y-ion fragments and oxonium ions
  14. Only a small fraction of spectra from both AI-ETD and HCD contained the Y1-ion (i.e., the intact peptide plus one HexNAc) that carries the same charge of the precursor, while the Y1-ion with one less charge than the precursor was observed in 59.2% and 69.2% of AI-ETD and HCD spectra, respectively
  15. Some database search strategies for intact glycopeptides utilize the presence of Y1-ions in HCD spectra, and this data shows that AI-ETD may be a reasonable candidate for such an approach in future work
  16. Also, AI-ETD more often produced larger Y-type fragments , including the intact peptide with two HexNAc moieties and the intact peptide with the full HexNAc(2)Hex(3) common N-linked glycan core, and these fragments could also be used to improve glycopeptide searching strategies with AI-ETD spectra
  17. Both AI-ETD and HCD produced at least one of these Y-ions in ~72% identified spectra
  18. All HCD spectra contained the HexNAc oxonium ion (m/z 204.0867), which was also present in 99.97% of AI-ETD spectra (all but four spectra)
  19. Conversely, effectively no AI-ETD spectra (0.25%) contained the Hex oxonium ion at m/z 163.06, yet it was observed in 97.27% of HCD spectra
  20. Similar to the HexNAc oxonium ion, the m/z 366.14 oxonium ion (HexNAcHex) was present in nearly all HCD and AI-ETD spectra, but three common larger oxonium ions were more often observed in HCD spectra (Supplementary Fig. 3)
  21. That said, oxonium ions from sialylated glycans were the exception to this trend, which is discussed further below
  22. Others have reported the ability to distinguish glycan isomers using ratios of oxonium ion intensities in higher-energy collisional dissociation ( HCD ) spectra, namely to distinguish the presence of N-acetylglucosamine (GlcNAc , present in both N- and O-linked glycans) and N-acetylgalactosamine (GalNAc , only in O-linked glycans)
  23. In a second dataset, we extended the low mass range of AI-ETD spectra to 115 Th and calculated the GlcNAc/GalNAc ratio for AI-ETD and HCD spectra of intact glycopeptides (Supplementary Fig. 4a)
  24. No GalNAc residues are expected to be present in this dataset because of the focus on N-glycopeptides , so ratios for each dissociation method should only indicate the presence of GlcNAc
  25. As noted by Nilsson and co-workers, a GlcNAc/GalNAc ratio below 1 indicates the presence of GalNAc , while a ratio above 2 is significant for the presence of GlcNAc
  26. Nearly the entire distribution (99.9%) of calculated GlcNAc/GalNAc ratios for AI-ETD spectra is >2 (median of 6.52), providing a strong indication for the sole presence of GlcNAc as the primary isomer for all HexNAc residues
  27. HCD spectra also provide ratios with a median value >2 (median of 3.41), but 13% of HCD spectra provide a ratio below 2 despite the collision energy being within the previously investigated range
  28. We also examined oxonium ions (m/z 292.1027 and m/z 274.0921) from the sialic acid residue N-acetylneuraminic acid (Neu5Ac)
  29. Both AI-ETD and HCD generated the m/z 274.0921 ion with high frequency for spectra from sialylated glycans (96% and 95%, respectively), and the m/z 292 was also present in both, although slightly less frequently (87% and 93%, respectively)
  30. We also observed these ions to some degree in both AI-ETD and HCD spectra assigned to glycopeptides without a Neu5Ac moiety
  31. This false indication of a Neu5Ac moiety can be controlled for by calculating a ratio of intensity of the m/z 274.0921 ion to the HexNAc oxonium (m/z 204.0867)
  32. Setting a threshold of >0.1 for this Neu5Ac/HexNAc oxonium ion ratio eliminated 97% and 99% of AI-ETD and HCD spectra, respectively, that were assigned an identification without a Neu5Ac residue while retaining 83 and 88% of AI-ETD and HCD spectra that were assigned an identification with a sialylated glycan
  33. Such a calculation could be considered in future glycopeptide-centric search algorithms that are capable of handling both AI-ETD and HCD spectra
  34. Remarkably, Pap et al. observed that EThcD fragmentation preserves larger sialylated oxonium ions than HCD for O-linked glycans (namely m/z 657.2349, HexNAcHexNeuAc) , and we observed a similar trend with AI-ETD for sialylated N-glycans (Supplementary Fig. 4b)
  35. The m/z 657.2349 ion was present in 87% of AI-ETD spectra from identifications containing a Neu5Ac residue , but only 44% of the analogous HCD spectra
  36. We also calculated Ln/Nn ratios for AI-ETD and HCD spectra to investigate the presence of isomeric glycoforms of Neu5Ac with either α2,3 and α2,6 linkages (Supplementary Fig. 4c)
  37. Both AI-ETD and HCD generate a wide range of Ln/Nn ratios, but distributions within the low values (from 0 to 3) of Ln/Nn ratios in spectra from both dissociation methods are the most interesting
  38. AI-ETD ratios show a distribution with a median close the previously reported value for α2,3 linkages but lack a distinct distribution for the higher values that would indicate α2,6 linkages
  39. HCD has two distinct distributions but they are much closer to each other than previously reported, and the lower distribution has a median with a greater value than expected
  40. Even with these observations, it is difficult to comment on the accuracy of these calculations without predefined glycopeptide standards with known linkage information
  41. Furthermore, others have used the presence of specific oxonium and neutral loss ions to discriminate between structure isomers (see Wu et al. for an example), and observation of both ion types in AI-ETD spectra indicates that AI-ETD could prove useful toward this goal
  42. The ability of AI-ETD to distinguish glycan isomers needs to be further investigated and validated with dedicated future studies, but these data indicate that AI-ETD may be as valuable as HCD for generating oxonium ion distributions to distinguish GlcNAc and GalNAc isomers of HexNAc residues and that the method may also be able to provide insight on NeuAc linkage information
Output (sent_index, trigger, protein, sugar, site):
  • 0. glycopeptides, , -, -, glycopeptides
  • 1. N-glycopeptides, , -, -, N-glycopeptides
  • 10. glycopeptide, , -, -, glycopeptide
  • 12. fragments, , -, -, fragments
  • 13. glycopeptide, , -, -, glycopeptide
  • 15. glycopeptides, , -, -, glycopeptides
  • 16. glycopeptide, , -, -, glycopeptide
  • 16. used, , -, -, fragments
  • 20. HCD, , HCD, the m/z 366.14 oxonium ion (HexNAcHex), -
  • 20. present, , HCD, the m/z 366.14 oxonium ion (HexNAcHex), -
  • 23. glycopeptides, , -, -, glycopeptides
  • 24. N-glycopeptides, , -, -, N-glycopeptides
  • 30. glycopeptides, , -, -, glycopeptides
  • 36. glycoforms, , -, Neu5Ac, -
  • 40. glycopeptide, , -, -, glycopeptide
  • 5. glycopeptide, , -, -, glycopeptide
  • 5. glycosite, , -, -, glycosite
Output(Part-Of) (sent_index, protein, site):
  • 12. Y-type, fragments
  • 12. precursors, fragments
  • 13. Y-ion, fragments
  • 16. Y-type, fragments
  • 8. Y-ion, fragments
  • 9. B-type, fragments
*Output_Site_Fusion* (sent_index, protein, sugar, site):

 

 

Protein NCBI ID SENTENCE INDEX