Use of Reductive-Elimination
Reductive amination exploits the action of a strong base to release the glycan, but because of an extensive concomitant degradation reaction, known as peeling, the reaction is performed in the presence of an excess of the reducing agent sodium cyanoborohydride . Although performed mainly in solution, the reaction has also been used to release glycans from within SDS-PAGE gels [79, 80]. p-Elimination is used extensively for the release of O-glycans but has the disadvantage of producing glycans without a reducing terminus, thus precluding derivatization at this site. In response to this, several investigators have attempted the use of milder release reagents in the hope that peeling can be prevented and reduction avoided. Ammonia  has been used extensively in this context. It leaves the protein intact but converts the serine and threonine residues that were linked to the glycans in their dehydro forms. These subsequently react with excess ammonia to add an NH2 label that can be detected by MS to provide linkage information. Ammonia in the presence of ammonium
carbonate has also been used [82, 83]. Glycan release appeared to be good, and cleanup of the product was minimal as all reagents were volatile. Because the hydroxide ion appears to cause unfavorable peeling reactions, Miura et al.  have investigated the use of the ammonium salt ammonium carbamate for gly- can release and have reported efficient release with little peeling. The release was performed by addition of powdered ammonium carbamate and incubation for 20 h at 60 °C.
Although the reaction with ammonia was reported to produce quantitative release of O-glycans, a recent study  with human IgA1 has found incomplete liberation of O-glycans. MALDI time-of-flight (MALDI-TOF) MS analysis revealed that only one of the six glycosylated sites was susceptible to P-elimination under the conditions used. It was proposed that resistance to P-elimination was due to very close proximity of proline to the glycosylated serine or threonine residues. The author commented that the findings may have implications for similarly O-glycosylated peptides and proteins and possibly for other chemical methods that are used to carry out p-eliminations of O-glycans.
The reaction has been investigated in detail by Yu et al.  for O-glycan chains with pi,3-linked cores. In contrast to p 1,4-linkages of the N-glycan type, which were shown to be stable under the ammonium-based alkaline conditions, the p 1,3-linkage was found to be labile and to give considerable peeling. The results indicated that complete prevention of peeling under nonreducing alkali-catalyzed hydrolysis conditions remains difficult.
Zheng et al.  have compared ammonia, methylamine, and dimethylamine at 55 ° C for 6 h for the release of GalNAc from a small glycopeptide. The O-glycosylated Thr residue was converted into a stable derivative with various amines. p-Elimination with dimethylamine and methylamine resulted in the conversion of the glycopeptide to 69.2% of the dimethylamine derivative and 61.5% of the methylamine derivative, respectively. However, the incubation of the glycopeptide with ammonia only resulted in 8% production of the product. The authors concluded that elimination with dimethylamine was the most efficient for the release of O-linked glycans. In spite of these developments, the classical p-elimination reaction with sodium hydroxide and sodium borohydride remains the most popular method for releasing O-glycans. For glycoproteins that contain both N- and O-linked glycans, the N-linked glycans are usually released first enzymatically, followed by O-linked glycan release by P-elimination.