Most epidemics caused by viral infections that are associated with a significant death toll were caused by enveloped viruses such as influenza A virus, human immunodeficiency virus (HIV), Zika virus, Yellow fever virus, Dengue virus and Ebolavirus. Often, the main target for neutralizing antibodies to evoke a strong immune response is a glycosylated envelope membrane protein. Thus, in the development of vaccines, glycoproteins are typically in the focus of interest. In general, the glycosylation of proteins plays a critical role regarding structure, function, solubility, stability, trafficking, and ligand-binding (Imperiali and O’connor, 1999; Dalziel et al., 2014; Varki, 2017). Furthermore, glycosylation plays a major role for pharmacokinetics and pharmacodynamics of biologics and for pathogen-host interaction (Bagdonaite and Wandall, 2018; Cymer et al., 2018; Watanabe et al., 2019). In viral pathogenesis, glycosylation affects the attachment and release of virus particles as well as immune evasion (Bagdonaite and Wandall, 2018; Watanabe et al., 2019; Schön et al., 2020). Especially the latter is a major hurdle for vaccine design. The mode of actions that are known to be employed to invade the immune system are secretion and shedding of glycoproteins that function as a decoy to the immune system, and the shielding of epitopes (Watanabe et al., 2019). The latter is facilitated by occluding antigenic epitopes with host-derived glycans that are obtained through hijacking the host’s cellular glycosylation machinery (Schwarzer et al., 2009; Francica et al., 2010; Helle et al., 2011; Rödig et al., 2011; Rödig et al., 2013; Sommerstein et al., 2015; Behrens et al., 2016; Gram et al., 2016; Walls et al., 2016; Pralow et al., 2021). Moreover, it has been shown that also the glycoform itself can have an impact on binding and transmission assay as well as on transmissibility, antigenicity, and immunogenicity in animal models (Lin et al., 2003; Hütter et al., 2013; Chen et al., 2014; Li et al., 2016; Go et al., 2017). While it is assumed that immunogenic antigens benefit from mimicking the glycosylation of host cell proteins, it has also been proposed that modification of specific terminal sugar residues could be used to amplify vaccine efficacy (Galili, 2020; Chen, 2021). However, due to the complexity of protein glycosylation and the prevailing lack of methods to introduce defined modifications in the glycan composition of the proteins of interest, the topic is under investigated (Watanabe et al., 2019; Grant et al., 2020; Schön et al., 2020).

The ongoing corona virus disease 2019 (COVID-19) pandemic is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)—a single-stranded, positive-sense RNA virus (Walls et al., 2020). Its membrane envelope consists of three membrane proteins: the surface spike (S) glycoprotein, an integral membrane protein and an envelope protein (Wan et al., 2020; Zhou et al., 2020). Virus entry into human host cells is mediated by the S glycoprotein that binds to angiotensin-converting enzyme 2 (Walls et al., 2020). The S protein has 22 N-linked glycosylation sites. Thus, it is significantly more glycosylated than, for instance, the influenza A hemagglutinin (Wrapp et al., 2020). For the SARS-CoV-1 spike protein it has been shown previously that N-glycans significantly impact antibody response and neutralizing antibody levels (Chen et al., 2014; Walls et al., 2020).

For the investigation of the impact of glycoforms on the immunogenicity, mainly animal cell lines such as HEK and CHO cells that produce differentially glycosylated proteins are employed (Lin et al., 2013; Schön et al., 2020). However, due to need to develop specific expression protocols for each cell line, this approach is highly work-intensive. Additionally, the inherent macro- and microheterogeneity of glycoproteins complicate the elucidation of the role of specific glycans in, for instance, regarding their immunogenicity in animal models.

Over the past years the establishment of protocols for expression of eukaryotic and bacterial glycosyltransferases has facilitated the processing of glycans in cell-free one-pot reactions. As a platform technology, the corresponding in-vitro glycoengineering approaches have the potential to tailor the glycoform of proteins independent of the expression systems used (Van Landuyt et al., 2019; Rexer et al., 2020a). In our study, recombinant human β-1,2 N-acetlyglucosamintransferases I and II (MGAT1ΔTM and MGAT2 ΔTM) and β-1,4-galactosyltransferase (GalTΔTM) expressed in E. coli were utilized to convert insect cell-derived paucimannose structures of recombinant SARS-CoV-2 spike glycoprotein to typical mammalian, complex-type galactosylated structures in a cell-free one-pot reaction (Fujiyama et al., 2001; Boeggeman et al., 2003; Bendiak, 2014; Ramakrishnan and Qasba, 2014; Stanley, 2014). Glycan structures were analyzed using multiplexed capillary gel electrophoresis with laser-induced fluorescence detection (xCGE-LIF) and Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS). Results obtained clearly demonstrate that a large fraction of fucosylated and afucosylated, Man3-glycans were transferred to biantennary G2 and G2F structures (also Table 1).

Due to its scalability, eukaryotic protein processing and high productivity, the baculovirus-insect cell expression system is well-suited for the production of subunit vaccines (Felberbaum, 2015; Palomares et al., 2018). In addition to subunit vaccines against SARS-CoV-2 infections in development, there are currently three licensed vaccines, Flublok®, Cervarix® and Provenge® produced using this expression system with several more in clinical trials (Felberbaum, 2015; Palomares et al., 2018).

High immunogenicity of a recombinant insect-cell produced spike protein ectodomain variant, very similar to the one used here, has been confirmed in non-human primates [40]. Moreover, the spike protein is the antigen target of virtually all COVID-19 vaccines and advanced vaccine candidates (Krammer, 2020). At the time of writing this article, there was one licensed COVID-19 protein subunit vaccine (RBD-Dimer from Anhui Zhifei Longcom Biopharmaceutical, China) in China, while for two more candidates (Covovax from Novavax, United States; VAT00002 from Sanofi Pasteur and GSK, France/United Kingdom) emergency authorization was pending in the US and Europe (Yang et al., 2020; Dai and Gao, 2021; Shrotri et al., 2021). All three are recombinant SARS-CoV-2 spike protein variants produced using the baculovirus-insect cell expression system (Kyriakidis et al., 2021).

Glycoforms of recombinant proteins produced using baculovirus-insect cell expression systems are profoundly different from those produced using mammalian expression systems. An extensive review on the glycosylation processing of insect cells is given by Geisler et al. (2015). Typically, these proteins display mainly paucimannose and hybrid-type N-glycans with, at most, minor fractions of complex-type and oligomannose-type N-glycan (Geisler et al., 2015). Moreover, in comparison to Spodoptera frugiperda Sf9 cells, High Five® Trichoplusia ni cells can also produce core α-1,3-fucose-linked glycans (Palomares et al., 2018). The presence of the latter on biologics may cause hypersensitivity reactions when applied to patients with allergy and, thus, should be avoided, for instance by cell line engineering (Palmberger et al., 2014). According to the manufacturer’s information the recombinant SARS-CoV-2 spike protein used here was produced in the High Five® cell line. However, we excluded core α-1,3-fucose-linked glycans from the examination by using PNGase F that does not release this type of glycans from the protein backbone (Tretter et al., 1991).

For vaccine development, it has been proposed that immunogen candidates benefit from closely mimicking the macro- and microheterogeneity of the live virus glycosylation (Grant et al., 2020; Watanabe et al., 2020). This is as eliciting antibodies against shielded or non-native epitopes could cause an inefficient immune response. To overcome such obstacles, novel strategies utilizing distinct non-human glycans containing N-glycolylneuraminic acid or α,1-3 linked galactose residues, have been proposed to alleviate immune responses (Hütter et al., 2013; Galili, 2020; Schön et al., 2020; Chen, 2021). However, such approaches still need to be investigated in detail experimentally as, for example, both compounds are also suspected to cause allergenic reactions in humans.

To convert the glycoform from primarily paucimannose-type to typical mammalian complex-type N-glycans, the recombinant human glycosyltransferases, MGAT1ΔTM, MGAT2ΔTM and GalTΔTM, were effectively combined in a cell-free, one-pot glycosylation reaction. The gene expression of these glycosyltransferases in E. coli and the activity of the His-tag purified, soluble recombinant proteins in one-pot reactions using free glycans as substrates has been shown before (Fujiyama et al., 2001).

The site-specific glycan analysis of recombinant SARS-CoV-2 spike protein ectodomain expressed in human-derived cell line FreeStyle™ 293-F showed that of the 22 N-glycosylation sites only eight contained substantial fractions of oligomannose-type N-glycans (Watanabe et al., 2020). It is assumed that the occurrence of oligomannose-type fractions is caused by the steric inaccessibility of these glycans to the glycan processing enzymes in the Golgi, i.e., the occurrence of oligomannose-type N-glycans at distinct sites has shown to be independent of the producer cell line for the HIV viral glycoprotein gp120 (Pritchard et al., 2015). In accordance with the human cell-derived spike protein, our engineered spike protein abundantly exhibited complex-type G2F N-glycans. To a minor extend, a range of hybrid- and oligomannose-type N-glycans were also detected on the engineered spike protein. In contrast to the engineered spike protein, human cell-derived spike proteins also exhibit complex-type multi-antennary and sialylated structures (Watanabe et al., 2020). Taken together, a significant overlap of the glycoform has been generated. Whether the overlap is also site-specific remains to be investigated in future.

Over the past years, many efforts have been made to engineer insect cell lines to express complex-type N-glycans. A comprehensive summary of the attempts is given by Palomares et al. (2018). Briefly, complex-type N-glycans can be produced by the co-expression of glycosyltransferases or by generating transient insect cell lines. While the former generates an additional metabolic burden and affects growth properties, the stability of the latter has not been examined for commercial scale use. The advantage of in-vitro glycoengineering lies in its independence of producer cell lines as well as its flexibility towards the option to readily generate different glycoforms that are close to homogeneity. However, expensive nucleotides sugars are required as substrates and, thus, it is so far not feasible to apply in-vitro glycoengineering at larger scales (Mahour et al., 2018; Rexer et al., 2018; Rexer et al., 2020b).

SARS-CoV-2 spike glycoprotein variants produced in a baculovirus-insect cell expression system were in-vitro glycoengineered using recombinant glycosyltransferases to mimic the glycoform observed on the human cell-derived protein. In-vitro glycoengineering reactions as conducted here, can be used to generate immunogen candidates for pre-clinical testing to investigate the role of glycosylation on the antigenicity and immunogenicity in animal models. In general, in-vitro glycoengineering approaches can virtually be used to tailor the glycoform of all prominent vaccine candidates such as activated and attenuated viruses and virus like particles. The application of the technology to larger scales depends on the bulk availability of sugar nucleotides at moderate costs.