The baculovirus expression vector system (BEVS) is a powerful tool for the production of recombinant foreign proteins in insect cells (Jarvis, 2009), allowing to perform biochemical analyses and structure-function studies. Many post-translational modifications such as protein glycosylation (N- and O-glycosylation) occur in lepidopteran cells but noticeable differences exist with those found in mammalian cells (Altmann et al., 1999). N-glycans are mostly complex-type and sialylated in mammalian cells, whereas paucimannose-type N-glycans with a pentasaccharide structure (Man3GlcNAc2) are often found in glycoproteins produced by insect cells such as in Spodoptera frugiperda (Sf)- derived Sf9 cells (Jarvis and Finn, 1995, Legardinier et al., 2005b, Tomiya et al., 2004).

Different O-linked monosaccharides and corresponding extended O-glycans are also found to be attached to glycoproteins produced by insect cells. Mucin-type O-N-acetylgalactosamine (O-GalNAc), which is the most abundant type of O-glycosylation found in mammalian glycoproteins (Brockhausen et al., 2022), is also found in natural and recombinant glycoproteins produced by insect cells (Lopez et al., 1999) such as Sf9 cells (Legardinier et al., 2005b) and Drosophila melanogaster S2 cells (Schwientek et al., 2007). In Drosophila melanogaster, many other O-linked monosaccharides are also found attached to their glycoproteins such as O-Mannose (O-Man) (Zhang and Ten Hagen, 2019) but also O-glucose (O-Glc), O-fucose (O-Fuc) and O-N-acetylglucosamine (O-GlcNAc). These three O-linked monosaccharides (Glc, Fuc and GlcNAc) are often bound to Epidermal Growth Factor (EGF)-like domains (shortly thereafter named EGFs) of membrane glycoproteins such as NOTCH receptors (Harvey et al., 2016, Pennarubia et al., 2021). EGFs are small protein domains comprising 35-45 amino acid residues and possessing six conserved cysteines, connected by three disulfide bonds. They form four loops or sub-domains (C1-C2, C2-C3, C3-C4 and C5-C6), which can be modified by N-glycans and/or different O-linked monosaccharides or extended O-glycans.

In Drosophila melanogaster and mammals, many studies showed that the homologous Notch receptors (one in fly and four in mammals), which contain 29-36 EGF repeats in their extracellular part, were highly modified with O-fucose such as EGF26, following the action of the protein O-fucosyltransferase 1 (POFUT1) which recognizes the consensus sequence C2XXXX(S/T)C3 (Shao et al., 2003). POFUT1-mediated O-fucosylation is known to modulate interactions of NOTCH receptors with the DSL (Delta, Serrate, LAG-2) family ligands (Luther and Haltiwanger, 2009, Okajima et al., 2003). There are about 100 target proteins of POFUT1 in human and mouse, including NOTCH1-4. We recently showed that WIF1, an inhibitor of the Wnt signaling pathway, exhibited on its EGF3 an O-fucose required for its optimal secretion (Pennarubia et al., 2020). PAMR1, another secreted multidomain protein considered as tumor suppressor in breast cancer (Lo et al., 2015), was also modified with O-fucose on its unique EGF (Pennarubia et al., 2021).

Very few studies reported the characterization of mammal EGF-containing glycoproteins modified by O-linked monosaccharides, with the BEVS using lepidopteran cells. The rat NOTCH1 EGF11-13, expressed with the BEVS using High-FiveTM cells from Trichoplusia ni, exhibited O-glucose in the loops C1-C2 or C3-C4 of different EGFs (EGF11, 12 or 13) and O-fucose in the loop C2-C3 of EGF12. The presence of these O-linked monosaccharides revealed that such insect cells can express functional POFUT1, POGLUT1 (protein O-glucosyltransferase 1) (Luca et al., 2015), as well as POGLUT2/3 (Takeuchi et al., 2018).

In this work, we focused on abilities of the BEVS using Sf9 insect cells to produce mammalian recombinant glycoproteins with O-fucose specifically bound to their EGFs. At first, we produced and purified a secreted form of Spodoptera frugiperda POFUT1 with N-terminal polyhistidine and V5 tags (named SfHisV5Po) from the supernatant of baculovirus-infected Sf9 cells. For the purpose of comparison with an enzyme already characterized, we also produced its Mus musculus recombinant counterpart (named MmHisV5Po) from previously obtained stable CHO cells (Pennarubia et al., 2018). After comparison of the primary sequences and 3D structures of POFUT1 for different species, the N-glycosylation of both recombinant enzymes was characterized by using PNGase F treatment, lectin blot and mass spectrometry. Recombinant SfHisV5Po mostly exhibited an N-glycan at the highly evolutionary conserved across Metazoa NRT site, whereas recombinant mouse POFUT1 (MmHisV5Po) from stable CHO cells exhibited an additional second N-glycan at N165 position.

Recombinant SfHisV5Po and MmHisV5Po were then compared for their ability to in vitro add O-fucose to EGFs of mammalian NOTCH1, WIF1 or to the unique EGF of PAMR1. Their O-fucose transfer ability was determined by using copper-catalyzed azide-alkyne cycloaddition (CuAAC) referred to as click chemistry. The purified SfHisV5Po was shown to be able to specifically add in vitro O-fucose to isolated EGFs of NOTCH1 (EGF26) and WIF1 (EGF3), but with less efficiency compared to MmHisV5Po. However, SfHisV5Po was unable to add in vitro O-fucose to the single EGF domain of PAMR1 contrary to MmHisV5Po. Finally, using multiple reaction monitoring-mass spectrometry (MRM-MS), we showed that recombinant full-length human WIF1 (HisWIF1), purified from the secretome of baculovirus-infected Sf9 insect cells, was as expected modified by O-fucose on its EGF3.