Roles of glycosylation in virus biology
Glycans on viral envelope glycoproteins play important roles in virus biology, with specific functions identified in various stages of viral infection (Table I and Figure 2). As previously mentioned, glycosylation is generally required for progeny formation and infectivity for many viruses (Knowles and Person 1976; Leavitt et al. 1977; Olofsson and Lycke 1980; Herrler and Compans 1983; Lambert and Pons 1983; Mann et al. 1983; Kuhn et al. 1988; Li et al. 2006; Reszka et al. 2010; Mathys and Balzarini 2015; Shrivastava-Ranjan et al. 2016; Mossenta et al. 2017). In some cases, mature N-glycans in particular are necessary for proper viral particle formation and release (Datema et al. 1984). As in human proteins, N-linked glycosylation is critical for proper folding of viral proteins (Land and Braakman 2001; Zai et al. 2013; Wu et al. 2017), and, in some cases, viral protein secretion (Hu et al. 1995; Lu et al. 1997; Yoshii et al. 2013; Mossenta et al. 2017). Drugs targeting glycoprotein maturation and exhibiting a favorable therapeutic ratio can therefore be used to combat viral infections, where no other therapeutics are available (Miller et al. 2012). Most importantly, N- and O-linked glycans shield immunodominant epitopes from immune recognition (Francica et al. 2010; Helle et al. 2011; Sommerstein et al. 2015; Behrens et al. 2016; Gram et al. 2016; Walls et al. 2016). In addition to general functions, specific types of glycans on viral glycoproteins often affect receptor binding and cell fusion (Lin et al. 2003; Lozach et al. 2003; Gramberg et al. 2005; Davis et al. 2006; Miller et al. 2008; Chen et al. 2014; Phoenix et al. 2016; Wang et al. 2016). Moreover, it is becoming increasingly clear that distinct glycans play separate roles in various stages of the viral cycle (Goffard et al. 2005; Falkowska et al. 2007; Helle et al. 2010; Wang et al. 2013, 2017; Lennemann et al. 2014; Bradel-Tretheway et al. 2015; Luo et al. 2015; Orlova et al. 2015; Wu et al. 2017). Some site-specific functions of viral glycosylation have been assigned for both N- and O-linked glycans mostly by site-directed mutagenesis studies. Identified roles of glycans within various stages of the viral cycle will be discussed in the following sections.
Table I. Glycosylation of viral proteins of human eveloped viruses
Abbr. Virus name Family Genus Genome N-glycosylateda O-glycosylateda Binding/entry/haemagglutination/infectivitya Spreada Formation/release/yield/protein transporta Immune responsea Protein–protein interaction and other functionsa
LASV Lassa virus Arenaviridae Arenavirus (−)ssRNA Shrivastava-Ranjan et al. (2016) Sommerstein et al. (2015)
MACV Machupo virus Arenaviridae Arenavirus (−)ssRNA Bowden et al. (2009);
INKV Inkoo virus Bunyaviridae Orthobunyavirus (−)ssRNA Pesonen, Ronnholm et al. (1982)
RVFV Rift Valley fever virus Bunyaviridae Phlebovirus (−)ssRNA Phoenix et al. (2016) Phoenix et al. (2016)
UUKV Uukuniemi virus Bunyaviridae Phlebovirus (−)ssRNA Pesonen, Kuismanen et al. (1982)
HCoV Human coronavirus Coronaviridae Alphacoronavirus (+)ssRNA Walls et al. (2016)
MHV Mouse hepatitis virusb Coronaviridae Betacoronavirus (+)ssRNA Niemann et al. (1982) Niemann et al. (1982); Niemann et al. (1984); Locker et al. (1992) Niemann et al. (1982)
SARS-CoV SARS-related human coronavirus Coronaviridae Betacoronavirus (+)ssRNA Gramberg et al. (2005)
EBOV Ebola virus Filoviridae Ebolavirus (−)ssRNA Jeffers et al. (2002); Ritchie et al. (2010); Collar et al. (2016) Jeffers et al. (2002); Ritchie et al. (2010); Collar et al. (2016) Lin et al. (2003); Takada et al. (2004); Gramberg et al. (2005); Lennemann et al. (2014) Wang et al. (2017) Dowling et al. (2007); Francica et al. (2010); Noyori et al. (2013); Lennemann et al. (2014)
MARV Marburg virus Filoviridae Marburgvirus (−)ssRNA Feldmann et al. (1991) Feldmann et al. (1991) Becker et al. (1995); Gramberg et al. (2005) Noyori et al. (2013)
DENV Dengue virus Flaviviridae Flavivirus (+)ssRNA Miller et al. (2008); Dubayle et al. (2015); Lei et al. (2015) Davis et al. (2006); Miller et al. (2008) Rouvinski et al. (2015)
JEV Japanese encephalitis virus Flaviviridae Flavivirus (+)ssRNA Zai et al. (2013) Wang et al. (2016) Zai et al. (2013)
MVEV Murray Valley encephalitis virus Flaviviridae Flavivirus (+)ssRNA Blitvich et al. (2001)
TBEV Tick-borne encephalitis virus Flaviviridae Flavivirus (+)ssRNA Yoshii et al. (2013) Yoshii et al. (2013)
WNV West Nile virus Flaviviridae Flavivirus (+)ssRNA Davis et al. (2006) Whiteman et al. (2010) Li et al. (2006)
ZIKV Zika virus Flaviviridae Flavivirus (+)ssRNA Mossenta et al. (2017) Mossenta et al. (2017)
HCV Hepatitis C virus Flaviviridae Hepacivirus (+)ssRNA Iacob et al. (2008) Brautigam et al. (2013) Lozach et al. (2003); Goffard et al. (2005); Falkowska et al. (2007); Helle et al. (2010); Chen et al. (2014) Fournillier et al. (2001); Goffard et al. (2005); Helle et al. (2010); Reszka et al. (2010); Orlova et al. (2015) Fournillier et al. (2001); Falkowska et al. (2007); Liu et al. (2007); Brown et al. (2010); Helle et al. (2010); Anjum et al. (2013); Zhao et al. (2014); Li et al. (2016) Meunier et al. (1999); Goffard et al. (2005); Orlova et al. (2015)
HBV Hepatitis B virus Hepadnaviridae Orthohepadnavirus dsDNA-RT Schmitt et al. (1999) Schmitt et al. (1999) Lu et al. (1997); Ito et al. (2010); Julithe et al. (2014) Julithe et al. (2014); Hyakumura et al. (2015)
HCMV (HHV-5) Human cytomegalovirus (human herpesvirus 5) Herpesviridae Cytomegalovirus dsDNA Kim et al. (1976)c; Britt and Vugler (1989); Kaye et al. (1992); Al-Barazi and Colberg-Poley (1996); Jones et al. (1996); Margulies et al. (1996); Maidji et al. (1998); Huber and Compton (1999); Mullberg et al. (1999); Rehm et al. (2001); Gewurz et al. (2002); Theiler and Compton (2002); Griffin et al. (2005); Margulies and Gibson (2007); Lin et al. (2008); Stanton et al. (2010); Engel et al. (2011); Gabaev et al. (2014); Geyer et al. (2014); Cavaletto et al. (2015) Britt and Vugler (1989); Kari et al. (1992); Theiler and Compton (2002); Gabaev et al. (2014); Geyer et al. (2014); Bagdonaite et al. (2016) Kropff et al. (2012)
EBV (HHV-4) Epstein-Barr virus (human herpesvirus 4) Herpesviridae Lymphocryptovirus dsDNA Edson and Thorley-Lawson (1983); Gong et al. (1987); Serafini-Cessi et al. (1989); Gong and Kieff (1990); Nolan and Morgan (1995); Paulsen et al. (2005); de Turenne-Tessier and Ooka (2007); Gore and Hutt-Fletcher (2009) Gong et al. (1987)c; Serafini-Cessi et al. (1989); Nolan and Morgan (1995); Borza and Hutt-Fletcher (1998); de Turenne-Tessier and Ooka (2007); Xiao et al. (2007); Bagdonaite et al. (2016) D’Arrigo et al. (2013); Gram et al. (2016)
KSHV (HHV-8) Kaposi’s sarcoma-associated herpesvirus (human herpesvirus 8) Herpesviridae Rhadinovirus dsDNA Chandran et al. (1998)c; Pertel et al. (1998); Li et al. (1999); Zhu et al. (1999); Baghian et al. (2000); Wu et al. (2000); Chung et al. (2002); Koyano et al. (2003); Dela Cruz et al. (2004); Meads and Medveczky (2004); Dela Cruz et al. (2009); Wu et al. (2015) Zhu et al. (1999); Wu et al. (2000) Dela Cruz et al. (2004); Wu et al. (2015)d
HHV-6 Human herpesvirus 6 Herpesviridae Roseolovirus dsDNA Okuno et al. (1990); Foa-Tomasi et al. (1992); Okuno et al. (1992); Pfeiffer et al. (1995)c Cardinali et al. (1998); Torrisi et al. (1999)
HHV-7 Human herpesvirus 7 Herpesviridae Roseolovirus dsDNA Hata et al. (1996); Mukai et al. (1997)c; Skrincosky et al. (2000); Glosson et al. (2010) Glosson et al. (2010)d
HSV-1 (HHV-1) Herpes simplex virus type 1 (human herpesvirus 1) Herpesviridae Simplexvirus dsDNA Keller et al. (1970)c; Spear and Roizman (1970)c; Honess and Roizman (1975)c; Eisenberg et al. (1979); Wenske et al. (1982); Cohen et al. (1983); Respess et al. (1984); Serafini-Cessi et al. (1984) Keller et al. (1970)c; Olofsson et al. (1981); Serafini-Cessi, Dall’Olio, Scannavini, Costanzo et al. (1983); Olofsson et al. (1983); Dall’Olio et al. (1985); Lundstrom et al. (1987); Serafini-Cessi et al. (1988); Peng et al. (1998); Norberg et al. (2007)c; Bagdonaite et al. (2015); Norden et al. (2015) Kuhn et al. (1988); Teuton and Brandt (2007); Wang et al. (2009); Altgarde et al. (2015) Knowles and Person (1976); Campadelli-Fiume et al. (1982); Serafini-Cessi, Dall’Olio, Scannavini, Campadelli-Fiume et al. (1983) Olofsson and Lycke (1980); Norrild and Pedersen (1982); Serafini-Cessi, Dall’Olio, Scannavini, Campadelli-Fiume et al. (1983); Altgarde et al. (2015) Sodora et al. (1989)
HSV-2 (HHV-2) Herpes simplex virus type 2 (human herpesvirus 2) Herpesviridae Simplexvirus dsDNA Cohen et al. (1983); Zezulak and Spear (1983); Serafini-Cessi et al. (1985) Zezulak and Spear (1983); Serafini-Cessi et al. (1985); Serafini-Cessi et al. (1988); Iversen et al. (2016) Luo et al. (2015) Luo et al. (2015) Luo et al. (2015) Clo et al. (2012); Iversen et al. (2016)
VZV (HHV-3) Varicella-zoster virus (human herpesvirus 3) Herpesviridae Varicellovirus dsDNA Friedrichs and Grose (1984); Montalvo et al. (1985); Montalvo and Grose (1986); Montalvo and Grose (1987); Yao et al. (1993); Maresova et al. (2000); Yamagishi et al. (2008) Montalvo et al. (1985); Montalvo and Grose (1987); Yao et al. (1993); Bagdonaite et al. (2016) Suenaga et al. (2015) Montalvo et al. (1985)
FLUAV Influenza A virus Orthomyxoviridae Influenzavirus A (−)ssRNA Schwarz et al. (1977)c; Edwardson (1984); Blake et al. (2009); An et al. (2015) Stansell et al. (2015) Park et al. (2016); Wu et al. (2017) Edwardson (1984); Wu et al. (2017) Skehel et al. (1984); Yang et al. (2011); Wu et al. (2017)
HeV Hendra virus Paramyxoviridae Henipavirus (−)ssRNA Bowden et al. (2010); Colgrave et al. (2012) Colgrave et al. (2012) Bradel-Tretheway et al. (2015); Stone et al. (2016) Bradel-Tretheway et al. (2015); Stone et al. (2016) Bradel-Tretheway et al. (2015); Stone et al. (2016) Bradel-Tretheway et al. (2015) Bradel-Tretheway et al. (2015); Stone et al. (2016)
NiV Nipah virus Paramyxoviridae Henipavirus (−)ssRNA Bowden et al. (2008) Stone et al. (2016) Bradel-Tretheway et al. (2015); Stone et al. (2016) Aguilar et al. (2006); Garner et al. (2015) Stone et al. (2016)
MeV Measles virus Paramyxoviridae Morbillivirus (−)ssRNA Hu et al. (1995) Hu et al. (1995)
HRSV Human respiratory syncitial virus Paramyxoviridae Pneumovirus (−)ssRNA Lambert and Pons (1983); Gruber and Levine (1985) Gruber and Levine (1985)c Lambert and Pons (1983) Lambert and Pons (1983)
MuV Mumps virus Paramyxoviridae Rubulavirus (−)ssRNA Herrler and Compans (1983) Herrler and Compans (1983)
VACV Vaccinia virus Poxviridae Orthopoxvirus dsDNA Shida and Dales (1981) Shida and Dales (1981) Shida and Dales (1981)
HIV-1 Human immunodeficiency virus 1 Retroviridae Lentivirus ssRNA-RT Bernstein et al. (1994); Doores, Bonomelli et al. (2010); Bonomelli et al. (2011); Pabst et al. (2012); Go et al. (2013); Yang W et al. (2014); AlSalmi et al. (2015); Pritchard, Harvey et al. (2015); Behrens et al. (2016); Panico et al. (2016); Cao et al. (2017); Go et al. (2017); Struwe et al. (2017) Bernstein et al. (1994); Go et al. (2013); Yang Z et al. (2014); Stansell et al. (2015) Nakayama et al. (1998); Pollakis et al. (2001); Hong et al. (2002); Lin et al. (2003); Francois and Balzarini (2011); Zou et al. (2011); Wang et al. (2013); Mathys and Balzarini (2014); Lombardi et al. (2015); Mathys and Balzarini (2015); Termini et al. (2017) Mathys and Balzarini (2014); Mathys and Balzarini (2015) Mathys and Balzarini (2014); Mathys and Balzarini (2015) Sanders et al. (2002); Wei et al. (2003); Wang et al. (2013); Doores and Burton (2010); St-Pierre et al. (2011); van Montfort et al. (2011); Pritchard, Spencer et al. (2015); Behrens et al. (2016); Lee et al. (2016)
CHIKV Chikungunya virus Togaviridae Alphavirus (+)ssRNA Lancaster et al. (2016)
RRV Ross River virus Togaviridae Alphavirus (+)ssRNA Nelson et al. (2016)
SFV Semliki Forest virus Togaviridae Alphavirus (+)ssRNA Pesonen (1979); Rasilo and Renkonen (1979)
SINV Sindbis virus Togaviridae Alphavirus (+)ssRNA Sefton (1975)c; Burke and Keegstra (1979); Hsieh et al. (1983) Mann et al. (1983) Leavitt et al. (1977); Datema et al. (1984)
RUBV Rubella virus Togaviridae Rubivirus (+)ssRNA Ho-Terry and Cohen (1984); Hobman et al. (1991); Lundstrom et al. (1991) Ho-Terry and Cohen (1984)c; Lundstrom et al. (1991) Ho-Terry and Cohen (1984) Hobman et al. (1991)
aLiterature references are indicated as first author last name and year.
bNonhuman infecting virus historically used as model for investigating betacoronaviruses.
cIndirect evidence.
dOther functions.
Fig. 2.  Roles of glycosylation in the biology of enveloped viruses. The cartoon depicts functions of glycosylation described in the literature to affect various stages of the infectious cycle of enveloped viruses (Table I). The glycan chains presented on the generic viral particles are of illustrative manner and do not represent the glycoprofiles of individual viruses. Likewise, the glycan structures shown to interact with specific glycan binding molecules represent the most likely type of structure (high mannose or complex type) based on known carbohydrate binding specificities, unless specified otherwise in individual studies defining the interactions or glycoprofiling/glycoproteomic studies of the viruses in question. In the bottom right panel, the viruses are grouped according to the predominant mode of exit within the individual families. Virus abbreviations as in Table I. Monosaccharide symbols follow the SNFG (Symbol Nomenclature for Glycans) system (PMID 26,543,186, Glycobiology 25: 1323–1324, 2015).

Attachment and entry
A number of viruses, such as HCV, Dengue virus, Ebola virus, SARS coronavirus, West Nile virus, Rift Valley fever virus and Japanese encephalitis virus use N-linked glycans as attachment and entry receptors by interacting with cellular lectins DC-SIGN, L-SIGN, LSECtin, ASGP-R or mannose receptor (Becker et al. 1995; Lin et al. 2003; Lozach et al. 2003; Gramberg et al. 2005; Davis et al. 2006; Miller et al. 2008; Chen et al. 2014; Phoenix et al. 2016; Wang et al. 2016). Mature N-glycans have been suggested to influence Lassa virus binding to cell entry receptor α-dystroglycan (Shrivastava-Ranjan et al. 2016). In addition, specific N-glycans on HIV-1 have been shown to affect CD4 receptor binding on T cells or regulate coreceptor usage (Nakayama et al. 1998; Pollakis et al. 2001; Francois and Balzarini 2011; Lombardi et al. 2015). Systemic functional analysis of HIV-1 gp160 potential N-glycosylation sites has identified a number of additional glycosites affecting infectivity (Wang et al. 2013). In addition, a certain N-glycan on HIV-1 gp41 was shown to have a consistent effect in several viral strains (Mathys and Balzarini 2014). Specific sialylated N-glycans on VZV gB have been identified that are important for interaction with myelin-associated glycoprotein and crucial for cell–cell fusion (Suenaga et al. 2015), and sialylation of virions in general was shown to affect entry of HSV-1 (Teuton and Brandt 2007). Similarly, specific N-linked glycans on HSV-2 gB influenced viral entry (Luo et al. 2015). An N-glycosylation site on IAV HA has also been shown to affect viral entry indirectly by modulating the avidity and specificity for sialosides (Wu et al. 2017). Finally, distinct putative N-glycosylation site mutations on HCV E1 and E2 resulted in diminished entry and altered CD81 binding (Goffard et al. 2005; Falkowska et al. 2007; Helle et al. 2010).
In contrast to the rather well documented roles of N-linked glycans, there are very few known examples, where specific O-glycans participate in virus–host interaction. Among these, two sialylated O-glycans on HSV-1 gB have been identified that determine binding to cellular receptor paired immunoglobulin-like type 2 receptor α residing on immune cells (Wang et al. 2009; Arii et al. 2010; Bagdonaite et al. 2015). In addition, HSV-1 O-glycans are involved in a few other aspects of viral binding to the host cell, as deletion of the densely O-glycosylated region of attachment factor gC affects both the binding affinity to the cell surface, and the release of progeny virus via modulation of interactions with cell surface glycosaminoglycans (Altgarde et al. 2015). Another example includes the carbohydrate-dependent binding of filoviruses to the macrophage galactose lectin that is known to recognize GalNAc-O-glycans (Takada et al. 2004). In hepatitis C virus, the mutation of several putative O-glycosites has also been shown to decrease HCV E2 affinity for CD81, suggesting that O-glycans might be of a more general importance in mediating interaction with host cells (Falkowska et al. 2007). O-glycoproteomic analysis of Hendra virus glycoprotein G has recently sprouted the first, to the best of our knowledge, systematic functional analysis of known O-glycosylation sites, revealing a multitude of functions including attachment and entry to host cells (Colgrave et al. 2012; Stone et al. 2016). Importantly, most of the functions could also be identified in analogous O-glycosites of a closely related Nipah virus (Stone et al. 2016). Interestingly, only in Nipah virus a single O-glycosite significantly affected Ephrin B2 receptor binding (Stone et al. 2016). This is an important example of conserved O-glycan function between closely related viruses, and presents intriguing possibilities in the light of recently identified consistent O-glycosylation patterns of several herpesviruses. Besides confirming the HSV-1 O-glycosites involved in binding PILRα in vitro, a few other O-glycosites residing in protein regions expected to influence HSV-1 attachment via cell entry receptors nectin-1 and HVEM (Carfi et al. 2001; Krummenacher et al. 2005; Heldwein et al. 2006; Di Giovine et al. 2011; Gallagher et al. 2014), or VZV attachment to receptor insulin degrading enzyme (Berarducci et al. 2010) were identified (Bagdonaite et al. 2015, 2016). This is another example how MS-based approaches combined with structural knowledge can help narrow down a list of O-glycosite candidates for focused studies. In conclusion, glycans on viral entry proteins are widely used for modulation of receptor binding and entry, with both N-linked and O-linked glycans having the capacity to affect the interaction (Figure 2, top left panel). In some cases, though, it seems to be an effect of conformational stability, rather than direct interaction (Stone et al. 2016).

Assembly and exit
Specific N-glycosites have been identified, that are important for glycoprotein secretion and viral particle formation in some viruses, including Zika virus, hepatitis viruses B and C (HBV and HCV), HSV-2 and HIV-1 (Fournillier et al. 2001; Falkowska et al. 2007; Helle et al. 2010; Ito et al. 2010; Wang et al. 2013; Luo et al. 2015; Mossenta et al. 2017). A few N-glycosites on influenza A virus neuraminidase have been shown to affect protein activity and viral particle release (Wu et al. 2017). Also, it is well documented that N-linked glycosylation of alphaherpesvirus envelope proteins is required for trafficking to the cell surface and viral particle egress (Norrild and Pedersen 1982; Montalvo et al. 1985; Luo et al. 2015; Suenaga et al. 2015). In particular, complex-type N-glycans are required for robust egress (Serafini-Cessi, Dall’Olio, Scannavini, Campadelli-Fiume et al. 1983). In varicella zoster virus the role of potential O-glycosylation sites has also been addressed by site-directed mutagenesis. Several glycosites were found to affect cell surface expression of the protein. However, such an approach is too laborious for analysis of all potential O-glycosylation sites, given the lack of conserved sequence motifs for O-glycosylation (Suenaga et al. 2015). As a more targeted approach, site-directed mutagenesis of some of the O-glycosites on Hendra virus glycoprotein G, identified via an O-glycoproteomic approach, affected the incorporation of glycoprotein F into pseudotyped virions. This suggests that the O-glycans might be required for protein–protein interaction (Stone et al. 2016). Similarly, in HIV gp120 O-glycosylation of a conserved threonine that is crucial for the association with gp41 has recently been shown to enhance infectivity and incorporation of gp120 into virions, though the virus maintains normal functionality in the absence of the O-glycan (Termini et al. 2017).
In conclusion, manipulation of glycans or glycosylation sites seems to affect both viruses that use direct budding and those using exocytosis. It is therefore likely that glycans affect many different aspects of viral particle formation (Figure 2, bottom right panel), possibly through an indirect consequence of disrupted trafficking of glycoproteins that are required for virus formation (Stone et al. 2016; Mossenta et al. 2017).

Viral spread and innate immunity
Different viruses employ various spread mechanisms both systemically and within the site of infection, including cell–cell fusion, transmigration and neuroinvasion. Viruses often take advantage of components of the innate immune system to target the cell type of interest at distant locations. Glycans play an important role in these mechanisms (Figure 2, bottom left panel). In the deadly Hendra virus the globular head region of glycoprotein G is important for binding to the entry receptors Ephrin B2/B3. Mutation of N-glycosylation sites in the globular head of glycoprotein G increased cell–cell fusion, whereas removal of a glycosite in the stalk region of Hendra glycoprotein G abolished cell–cell fusion (Bradel-Tretheway et al. 2015). In case of O-glycans, however, removal of distinct O-glycosites within a cluster in the stalk region resulted in both hypo- and hyperfusogenic phenotypes (Stone et al. 2016), and neither N- nor O-glycosite mutants exhibited impaired receptor binding (Bradel-Tretheway et al. 2015; Stone et al. 2016). This suggests that glycans influence other aspects of membrane fusion; possibly by affecting the thermodynamics of conformational changes. The impact of N-glycans in viral spread has also been demonstrated in in vivo models. Interestingly, the glycosylation status of West Nile virus glycoproteins has been associated with neuroinvasiveness, where abolishment of glycosites led to attenuation (Whiteman et al. 2010). Similarly, loss of a N-glycosite on Ross River virus glycoprotein E1 was associated with milder pathogenesis and increased virus clearance in a disease mouse model (Nelson et al. 2016). Finally, specific N-glycosylation patterns on both HA and NA on avian influenza A virus are required for efficient transmission in mammalian models (Park et al. 2016). A more direct role of glycans emerges from the viral exploitation of innate immune receptors as a helping hand during establishing the viral infection. Such innate immune receptors and other secreted glycan binding proteins present on infected host cells or antigen presenting cells are naturally the first line of defense against invading pathogens. For instance, secreted C-type lectins, ficolin-2 and mannose-binding lectin, inhibit HCV entry into hepatocytes (Brown et al. 2010; Zhao et al. 2014). However, these carbohydrate binding proteins can convey both protective and deleterious effects during viral infection, which are exploited by viruses to enter host cells. For example, secreted galectin-1 promotes entry of Nipah virus into endothelial cells by bridging viral and cellular glycans, when the glycan binding proteins are present during initial phases of infection. In contrast, galectin-1 inhibits viral spread, when added postinfection (Garner et al. 2015). Galectin-1 has also been shown to promote HIV-1 infection by crosslinking the viral particles and cell entry receptors, whereas influenza virus infection was inhibited (St-Pierre et al. 2011; Yang et al. 2011). As already mentioned, several viruses specifically use C-type lectin receptors for entry or transmission. Ebola virus utilizes an array of different C-type lectins, such as L-SIGN, DC-SIGN and LSECtin, for entry into various cell types (Lennemann et al. 2014). HIV-1 uses DC-SIGN on dendritic cells for migration to T cells (Hong et al. 2002). Although it is not entirely clear how the balance between antigen presentation and transmission is maintained, it has been proposed that homogeneous high-mannose type N-glycans enhance antigen presentation and virus degradation compared to intrinsically heterogeneous gp120 N-glycosylation (van Montfort et al. 2011). Interestingly, HIV-1 has also been shown to exploit sialic-acid binding lectin Siglec-1 for more efficient entry into macrophages, likewise promoting the existence of complex-type N-glycans on viral particles (Zou et al. 2011). In a similar context, we have demonstrated, that very early chemokine-mediated immune response to HSV-2 infection of the vaginal mucosa in mice relies on the presence of elongated O-linked glycans on the viral particles (Iversen et al. 2016). Although truncation of O-glycans might benefit the virus in escaping the early recognition by the host, elongated O-glycans are essential for viral replication (Bagdonaite et al. 2015; Iversen et al. 2016). Various carbohydrate structures from many bacteria and parastic worms are known to elicit robust innate immune responses via toll-like and other receptors (Rabinovich and Toscano 2009; Prasanphanich et al. 2013). It is only conceivable that analogous carbohydrate-mediated mechanisms could exist for early detection of viral infections, but have not been described in detail.

Adaptive immunity
During evolution, the immune system has developed redundant and fine-tuned mechanisms to identify and eliminate invading pathogens, including viruses. In turn, viruses continuously adapted to counteract or take advantage of the host’s defense mechanisms, which is an ongoing battle. Glycans on viral glycoproteins have dual roles in virus–host interplay (Figure 2, top right panel). On one hand, glycans can be part of the immune determinants themselves, but on the other hand glycans can mask the antigenic protein epitopes from recognition (Alexander and Elder 1984). Accumulating structural evidence suggests that glycans cover large surface areas of viral envelope proteins providing steric hindrance and physical shielding of vulnerable sites (Lee et al. 2016; Walls et al. 2016; Beniac and Booth 2017). Both N- and O-linked glycans can mask immunogenic B-cell epitopes, and, in some cases also determine or influence immunogenicity (Sodora et al. 1989; Hobman et al. 1991; Fournillier et al. 2001; Falkowska et al. 2007; Helle et al. 2010; Wang et al. 2013; Bradel-Tretheway et al. 2015). N-glycans covering structured domains and functional regions often confer protection from neutralization (Helle et al. 2010; Wang et al. 2013; Bradel-Tretheway et al. 2015; Wu et al. 2017). As an example, systemic functional analysis of N-glycosites in HCV revealed that protective N-linked glycans were mainly situated around the CD81 binding site on the HCV envelope glycoprotein E2 (Helle et al. 2010). In other cases, removal of specific N-linked glycans can alter the local antigenic landscape and broaden the immune response rather than simply uncovering underlying epitopes (Fournillier et al. 2001; Wang et al. 2013). In contrast, mutation of seemingly unrelated N-glycosites can reduce the sensitivity to neutralization at distant epitopes (Wang et al. 2013), highlighting the role of N-glycans on maintaining protein structure. In conclusion, mutation of potential N-glycosylation sites can result in unpredicted effects on protein conformation and molecular dynamics, which differentially affect neutralization by antibodies, and are not explained solely by epitope recovery. Site-specific N-glycosylation has also been reported to affect cellular immunity to HCV E1, where deletion of specific glycans enhanced recognition by cytotoxic T cells (Liu et al. 2007).
Dense O-glycosylation of viral proteins has been widely described to provide “bulk” shielding from host immunity, and protect from antibody neutralization (Machiels et al. 2011; Kropff et al. 2012). Besides shielding select immunodominant epitopes (Machiels et al. 2011), highly O-glycosylated mucin-like domains, such as those found in Ebola and Marburg viruses, have been suggested to physically hinder the interaction between virus-infected cells and immune cells (Francica et al. 2010; Noyori et al. 2013). Conversely, the deletion of the Ebola virus mucin-like domain decreased the immune response in mice, although it did not have an effect on immunogenicity in vitro (Dowling et al. 2007). Finally, O-glycans not only shield, but in some cases can be presented as part of neo-epitopes when elongation of the glycans is prevented. As an example, several O-glycosylated immunodominant B-cell epitopes have been reported for HSV-2 and EBV, by probing O-glycopeptides with sera of infected patients, suggesting that the glycan moiety can be recognized by B-cells (Clo et al. 2012; D’Arrigo et al. 2013).
In the context of epitope shielding and immune recognition of viruses, it might be of importance that RNA viruses, such as influenza virus, HIV-1 and HCV, intrinsically have a very high mutation rate, thus minimizing the effect of adaptive immune response (Sanjuan et al. 2010; Anjum et al. 2013; Lauring et al. 2013; Lynch et al. 2015). Early on it was observed that the highest amino acid substitution rates occurred in antigenic regions of envelope glycoproteins, including such mutations that changed the number and position of glycosylation sites (Krystal et al. 1983; Skehel et al. 1984; Wei et al. 2003). Due to high density of glycans in such regions, some of the carbohydrate chains are not fully processed resulting in exposure of immature structures as seen in the mannose patch of HIV gp120 (Doores, Bonomelli et al. 2010). Paradoxically, such underprocessed structures are recognized by the immune system, giving rise to broad neutralizing antibodies (bnAbs), with the classical example being 2G12 (Sanders et al. 2002; Doores and Burton 2010). In addition, highly conserved glycosylation sites that are directly involved in binding to cell entry receptors or viral interaction partners often become targeted by bnAbs (Rouvinski et al. 2015). Interestingly, it has been demonstrated, that the recognition by such bnAbs is largely unaffected by deletion of individual N-glycans within the gp120 mannose patch, mimicking the consequences of intrinsic mutagenesis (Pritchard, Spencer et al. 2015). In some cases, however, mutations at distant glycosites may have subtle effects on epitope recognition (Behrens et al. 2016). Similarly, antigenicity of the HBV surface antigen isoform S did not suffer from introduction of additional N-glycosylation sites, and even resulted in stronger and more persistent immune responses (Hyakumura et al. 2015). In contrast, arenaviruses represent very well adapted pathogens where glycan shield density inversely correlates with potency of neutralizing antibodies (Sommerstein et al. 2015). However, in other instances, over-glycosylation of important functional regions of viral proteins result in better shielding from the immune system at a cost of binding affinity and virus production (Aguilar et al. 2006; Julithe et al. 2014; Lennemann et al. 2014; Lynch et al. 2015). As a compromise, variable glycosite occupancy can both fulfill the requirement for receptor binding via the nonglycosylated protein, and provide protection via the glycosylated counterpart (Julithe et al. 2014).
Being the components of viral particles exposed to the extracellular environment, viral envelope glycoproteins serve as major targets for vaccine development; however, protein sequence variability within and across strains, high mutation rate, as well as enormous heterogeneity of glycan structures makes it very difficult to identify immunogens evoking universal reactivity. Various expression systems are used for production of subunit vaccine candidates, such as bacteria, yeast, plant, insect and mammalian cell lines (Cox 2012; Kushnir et al. 2012; Redkiewicz et al. 2014). While some glycosylation types can be replicated in insect and mammalian cells, yeast cells will not carry out mucin type O-glycosylation and proteins produced in bacterial cells will also lack N-glycans. Even mammalian expression systems may lack the required glycosyltransferase repertoire to glycosylate relevant sites and build up relevant structures. Although there are many successful examples of vaccines produced in aforementioned systems, there is quite a number of infectious diseases that still lack vaccine coverage due to failure of vaccine candidates to induce adequate and lasting immune responses (Grimm and Ackerman 2013). Therefore, cell lines closer to the natural host cell type should be explored to more accurately reproduce the overall protein structure, modifications, and exposed antigenic sites for the primed immune system to be able to neutralize the naturally encountered antigen. Recently, efforts are being made to identify consensus glycosylation patterns, as well as production platforms leading to elicitation of desired immune responses and pathogen neutralization (Li et al. 2016; Go et al. 2017). In contrast to fast mutating RNA viruses, DNA viruses, including herpesviruses, have relatively stable genomes and rely on other means for counteracting the host’s immune system (Sanjuan et al. 2010). It is therefore conceivable, that vaccine development should be less challenging for DNA viruses. However, most of herpesvirus-targeted subunit vaccines have failed so far. A recently developed HSV-2 vaccine lacking the main neutralizing antibody target is, however, showing great promise in mice. This vaccine evokes production of non-neutralizing antibodies to other envelope glycoproteins on infected cells, stimulating cellular NK cell immunity through engagement of Fcγ receptors (Petro et al. 2015). In conclusion, glycans on viral envelope glycoproteins have a tremendous impact on recognition by the host, and glycosylation heterogeneity makes it very difficult to identify universal vaccine candidates. While experimental evidence is key, development of accurate bioinformatic tools to predict glycosylation and likely mutation patterns would be of big value for vaccine research. This should become possible once a substantial number of viral strains are sequenced and analyzed for glycan modifications in native contexts.

Roles of glycosylation on viral protein complex formation
While N-linked glycans situated on structural domains often shield functionally important protein regions from antibody neutralization, stem region N-glycans can affect glycoprotein stability and protein–protein interactions. For example, the mutations in the N-glycosite localized to the stem region of the Hendra virus glycoprotein G affected the conformation of the molecule, leading to increased oligomerization and interaction with its viral binding partner (Bradel-Tretheway et al. 2015). Moreover, the stalk mutation had similar effects in a closely related Nipah virus (Bradel-Tretheway et al. 2015). In Hepatitis C Virus, disruption of specific N-glycosites in the two highly N-glycosylated envelope glycoproteins E1 and E2 results in formation of unproductive E1–E2 heterodimers (Meunier et al. 1999; Goffard et al. 2005; Orlova et al. 2015), whereas specific O-glycosites on Hendra virus protein G have been shown to affect association with protein F (Stone et al. 2016), suggesting both types of glycans can carry out similar functions. Similarly, O-glycosites have been found in regions of varicella zoster virus gE, important for interaction with partner gI (Bagdonaite et al. 2016), but the glycan specific functions are yet to be uncovered.

Induction of host glycosylation machinery
In addition to hijacking host glycosylation machinery for modification of viral proteins, certain viruses induce changes in the expression levels of host glycosyltransferases. For example, it is documented that herpesviruses induce expression of host fucosyltransferases leading to expression of sLex or Ley antigens (Nystrom et al. 2007; Norden et al. 2013, 2017). In addition, HSV-1 and HSV-2 infection lead to changes in gene expression related to glycosphingolipid synthesis (Miyaji et al. 2016). Also, in the case of HCV, it has been demonstrated that the total glycoprofile of HCV-infected hepatoma cells was shifted towards more fucosylated, sialylated, and in general complex N-glycans (Xiang et al. 2017). Another example includes HBV that has been shown to upregulate mannosidases, which supports increased glycan processing (Hu et al. 2016). In the case of influenza A virus, induction of the isoform, GalNAc-T3, has been detected in infected respiratory epithelial cells, possibly increasing O-glycosylation of airway mucins (Nakamura et al. 2015). The changes in expression levels of host carbohydrate active enzymes can either hint towards requirements for specific structures for functionality, or highlight the demand for increased glycosylation capacity to support viral glycoprotein glycosylation.