**6. A mosses idea on O-glycosylation**

The most often encountered type of protein O-glycosylation in mammals is the so-called mucin-type O-glycosylation, where Ser or Thr residues are at first decorated with *N*-acetylgalactosamine (GalNAc). Usually, this priming event is followed by the addition of other sugars, such as galactose, sialic acid, or GlcNAc [28]. In all plants, including mosses, this type of glycosylation is totally unknown. Instead, arabinans (chains of arabinoses) and arabinogalactans (AGPs, complex structures starting with a galactan that is substituted by arabinose chains and maybe other subtleties are found [29]. Arabinogalactans mostly occur as type II arabinogalactans [30, 31], but different architectures may also occur [32, 33]. Bryophytes do generate arabinogalactans, but with certain differences as compared to seed plant AGPs [34–36].

Notably, these "exotic" oligosaccharides are not linked to the codogenic amino acids Ser or Thr but to 4-hydroxyproline (Hyp) [29, 37]. Neighboring amino acids, in particular proline and hydroxyproline themselves, dictate if a given Hyp residue rather falls prey to galactosyl- or arabinosyl-transferase [29].

So, the initial step of O-glycosylation in plants is the oxidation of proline to hydroxyproline. The remarkable fact now is that apparently, the sites of mucin-type O-glycosylation of mammalian proteins are also the sites prone to be modified by prolyl-4-hydroxylase and then by arabinosyl-transferase as exemplified by human erythropoietin expressed in the moss *P. patens* [38] or human IgA1 [39]. Hardly surprising, the same holds true for *N. benthamiana* with the only, albeit technologically relevant difference, that several prolyl-4-hydroxylase are redundantly at work in vascular plants [39, 40], whereas knock-out of just one paralogous gene sufficed to suppress erythropoietin oxidation in *P. patens* [41].
