Role of sialylation of N-linked glycans in prion pathogenesis

Abstract

Mammalian prion or PrP^Sc is a proteinaceous infectious agent that consists of a misfolded, self-replicating state of the prion protein or PrP^C. PrP^C and PrP^Sc are posttranslationally modified with N-linked glycans, which are sialylated at the terminal positions. More than 30 years have passed since the first characterization of the composition and structural diversity of N-linked glycans associated with the prion protein, yet the role of carbohydrate groups that constitute N-glycans and, in particular, their terminal sialic acid residues in prion disease pathogenesis remains poorly understood. A number of recent studies shed a light on the role of sialylation in the biology of prion diseases. This review article discusses several mechanisms by which terminal sialylation dictates the spread of PrP^Sc across brain regions and the outcomes of prion infection in an organism. In particular, relationships between the sialylation status of PrP^Sc and important strain-specific features including lymphotropism, neurotropism, and neuroinflammation are discussed. Moreover, emerging evidence pointing out the roles of sialic acid residues in prion replication, species transmission, strain competition, and strain adaptation are reviewed. A hypothesis according to which selective, strain-specified recruitment of PrP^C sialoglycoforms dictate unique strain-specific disease phenotypes is examined. Finally, the current article proposes that prion strains evolve as a result of a delicate balance between recruiting highly sialylated glycoforms to avoid an “eat-me” response by glia and limiting heavily sialylated glycoforms for enabling rapid prion replication.

Introduction

Prion diseases or transmissible spongiform encephalopathies is a class of lethal, transmissible, neurodegenerative disorders of humans and animals (Prusiner, 1982; Legname et al., 2004). Prions, or PrP^Sc, spread between organisms or within a brain by recruiting host-encoded prion protein (PrP^C) and replicating their disease-specific misfolded structures via a template-assisted mechanism (Cohen and Prusiner, 1998). Prion diseases display multiple disease phenotypes characterized by diverse clinical symptoms, differences in brain regions and cell types affected by the disease, and diverse PrP^Sc deposition patterns (Collinge and Clarke, 2007). The diversity of disease phenotypes within the same host is attributed to the ability of PrP^C to acquire multiple, alternative, conformationally distinct, self-replicating PrP^Sc states referred to as prion strains or subtypes (Bessen and Marsh, 1992; Safar et al., 1998; Peretz et al., 2001; Ayers et al., 2011; Gonzalez-Montalban et al., 2011; Klimova et al., 2015). The structural diversity of PrP^Sc strains has been well documented (Caughey et al., 1998; Thomzig et al., 2004; Legname et al., 2005; Spassov et al., 2006; Morales et al., 2016), yet the question on how PrP^Sc structures elicit multiple disease phenotypes remains poorly understood (Baskakov, 2021a). This question is closely related to other puzzling problems in prion biology. What molecular features of PrP^Sc are responsible for chronic inflammation and neurodegeneration? What features dictate strain-specific neuro- or cell-tropism of prions and control their fate in a host? The current article summarizes mounting evidence suggesting that N-linked glycans, which are a constitutive part of PrP^Sc, are important players in defining prion disease pathology and disease phenotype.

Diversity of N-glycans in PrP^C and PrP^Sc

PrP^C is posttranslationally modified with a glycophosphatidylinositol (GPI) anchor and one, two, or no N-linked glycans at the residues Asn-181 and/or Asn-197 (mouse PrP sequence) (Bolton et al., 1985; Stahl et al., 1987; Turk et al., 1988; Stahl et al., 1993). N-glycans on PrP^C are extremely diverse with more than 400 different PrP^C sialoglycoforms identified (Endo et al., 1989; Rudd et al., 1999; Stimson et al., 1999; Ritchie et al., 2002; Nakić et al., 2021). Each of the two N-glycans accommodate a wide range of structures consisting of mono-, bi-, tri-, tetra-, or penta-antennary structures with a variable numbers (up to five) of negatively charged terminal sialic acid residues (Endo et al., 1989; Rudd et al., 1999; Stimson et al., 1999; Ritchie et al., 2002). As a result, PrP^C molecules display a dramatic range of pIs at physiological pH (Katorcha et al., 2014; Katorcha et al., 2015a). The N-gycans are highly fucosylated at the core N-acetylglucosamine (GlcNAc) residue, by which glycans are attached to Asn (Ritchie et al., 2002). The GPI-anchor and N-linked glycans are carried over upon conversion of PrP^C into PrP^Sc (Stahl et al., 1987; Turk et al., 1988; Stahl et al., 1992; Stahl et al., 1993; Rudd et al., 1999).

Among carbohydrate groups that constitute N-glycans, sialic acid residues play a key role in defining glycan diversity and function (Fig. 1). Sialic acids are a family of 9-carbon containing acidic monosaccharides that are found in terminal positions of N- and O-linked glycans of glycoproteins or glycolipids (Varki, 1999). In humans, there is only one type of sialic acids, which is N-acetylneuraminic acid (Neu5Ac) (Varki, 2010), whereas the rest of mammalian species, with the exception of the ferret (Ng et al., 2014), produce two types of sialic acids. Neu5Ac is the predominant type that is synthesized in the non-human mammalian brain, though both Neu5Ac and N-glycolylneuraminic acid (Neu5Gc) are synthesized by peripheral organs (Varki, 1999). The deficiency in synthesis of Neu5Gc in humans is linked to an irreversible mutation in the gene encoding N-acetylneuraminic acid hydroxylase (an enzyme that synthesizes Neu5Gc from Neu5Ac) that occurred during evolution from primates to humans (Varki, 2010). Like humans, ferrets can produce only Neu5Ac (Ng et al., 2014). While humans lack the ability to synthesize Neu5Gc, this sialic acid can be incorporated into human tissues metabolically from diet (Samraj et al., 2015).

Figure 1.

Structures of the two most common types of sialic acid residues, Neu5Ac and Neu5Gc (a), and a diagram illustrating the differences in sialic acids synthesized in humans versus non-human mammals (b). Structural diversity of sialic acid residues is achieved via naturally occurring modifications at 1-, 4-, 5-, 7-, 8-, or 9-carbon positions (c). Diversity of carbohydrate epitopes is achieved due to variations in sulfation of galactose and N-acetylglucosamine that produce several Lewis glycoepitope families (d). Panel d shows only a small subset of possible sulfated variants

Sialic acid contributes to the diversity of N-glycan structure and function via several means (Fig. 1). First, sialic acids can be attached to galactose or N-acetylgalactosamine of glycans via ✔2–3, ✔2–6, ✔2–8, or ✔2–9 linkages (Varki, 1999). Second, various natural substitutes including O-acetyl, N-glycolyl, O-lactyl, O-sulfate, O-phosphate, tauryl, hydroxyl, and O-methyl can be synthesized on carbons of sialic acids at 1-, 4-, 5-, 7-, 8-, or 9-positions, where O-acetyl is the most common substitute. (Schauer et al., 2011). Third, in combination with sialic acid, other groups including sulfate and fucose are involved in forming functional glycan epitopes including Sialyl Lewis^x, Sialyl Lewis^a, and 6’Sulfo-Sialyl Lewis^x (Fukuda et al., 1999), among which Lewis^x and Sialyl Lewis^x were found in N-glycans of PrP^Sc (Stimson et al., 1999; Ritchie et al., 2002). Fourth, complex glycans can exhibit several branching patterns that contribute to variations in density of sialic acid residues, which is important for the binding of multivalent ligands.

In traditional mass-spectroscopy analyses, the levels of sialylation of N-glycans are largely underestimated due to spontaneous desialylation caused by the acidic pH of solvents employed for mass-spectroscopy. As judged from 2D-Western blotting, which provides a more direct way of assessing the sialylation status, up to 90% of N-glycan branches are sialylated in PrP^Sc (Katorcha and Baskakov, 2017; Katorcha and Baskakov, 2018). The precise sialylation level of PrP^Sc is dictated by prion strain along with a tissue and brain region from which PrP^Sc originates (Katorcha et al., 2015b; Srivastava et al., 2015; Katorcha et al., 2016b; Makarava et al., 2020a). In PrP^Sc N-glycans, sialic acid residues are linked via ✔2–3 or ✔2–6 linkages with the majority being linked by an ✔2–6-bond (Endo et al., 1989; Katorcha and Baskakov, 2017). As judged from analysis of three mouse-adapted prion strains, high percentages of ✔2–6-linked sialic acids were observed in PrP^Sc originating from brain, spleen, or cultured cells (Katorcha and Baskakov, 2017). The type of linkage seems to be independent of PrP^Sc source, as only a minor variation in the ratio of ✔2–3- versus ✔2–6-linked sialic residues was observed between brain-, spleen- or cell-derived PrP^Sc (Katorcha and Baskakov, 2017).

N-Glycans are exposed on a surface of PrP^Sc

Each of the two N-glycans adds up to 5 kDa to the molecular weight of PrP. The high density of PrP molecules within PrP^Sc raises questions about spatial positioning of N-glycans and, specifically, whether N-glycans are exposed on the PrP^Sc surface or buried in its interior. Staining of brain slices from prion-infected animals with Sambucus Nigra agglutinin (SNA) lectin demonstrated that N-glycans are exposed on the PrP^Sc surface at high density (Fig. 2) (Baskakov et al., 2018). SNA lectin is specific to terminal sialic acid residues attached via α2–6 linkages regardless whether they are present on PrP^Sc plaques or not. Much higher intensity of staining of PrP^Sc plaques in comparison to that of brain sialoglycocalyx suggests that the density of sialylation on PrP^Sc surfaces exceeds the density of sialoglycocalyx (Fig. 2). High density of sialylation on PrP^Sc surfaces is also supported by intense staining of PrP^Sc plaques by Alcian Blue, a polyvalent basic dye that binds to polyvalent acidic compounds and is generally used for staining of sialylated glycocalyx (Makarava et al., 2010).

Figure 2.

Staining of prion plaques with SNA lectin. Images of plaques (indicated by arrows) in Syrian hamsters infected with S05 PrP^Sc strain (Makarava et al., 2012b) (a, b), or atypical PrP^Sc (Kovacs et al., 2013) (c, d). Brain slices were stained with anti-PrP SAF-84 antibody (a, c) or SNA lectin (b, d), as indicated

Sialylation of PrP^Sc and prion infectivity

High density of sialic acid residues on the surface of PrP^Sc particles raises the question whether the sialylation status of PrP^Sc is important in determining their fate in an organism and the outcomes of prion infection. On the surface of a mammalian cell, sialoglycocalyx acts as a part of a self-associated molecular pattern helping the innate immunity to recognize “self” from “altered self” or “non-self” (Varki, 2008; Brown and Neher, 2014). Stripping sialic acid residues from glycans exposes galactose residues, which generate “eat me” signals for professional and non-professional macrophages including microglia. As an example, exposed galactose triggers clearance of erythrocytes and platelets by Kupffer cells (Aminoff et al., 1977; Jansen et al., 2012) or phagocytosis of neurons by microglia (Linnartz et al., 2012b; Linnartz-Gerlach et al., 2014; Linnartz-Gerlach et al., 2016).

Several recent studies suggested that carbohydrate epitopes on PrP^Sc surfaces and, specifically, exposed galactose, dictate the fate of prions in an organism. Administration of partially desialylated PrP^Sc with an increased amount of exposed galactose failed to induce prion disease in animals after intracranial or intraperitoneal injections (Katorcha et al., 2014; Katorcha et al., 2016a; Srivastava et al., 2017). Remarkably, animals infected with partially desialylated PrP^Sc were free of prions in their brains, as judged from Western blot and analysis by calibrated serial protein misfolding cyclic amplification, capable of detecting single PrP^Sc particles (Katorcha et al., 2016a; Srivastava et al., 2017). Upon intraperitoneal injections, desialylated PrP^Sc was found in the liver, whereas sialylated PrP^Sc is transported to and sequestered by the spleen and lymph nodes (Srivastava et al., 2017). Furthermore, prion infectivity could be switched off and on in a reversible manner via partially removing and reinstalling the sialylation of PrP^Sc, respectively (Katorcha et al., 2016a). These studies suggested that exposing galactose residues designates PrP^Sc for clearance and, vice versa, hiding galactose residues by sialylation protects PrP^Sc against clearance. Maintaining sufficient sialylation levels is important for prions to survive in an organism.

Sialylation of PrP^Sc, lymphotropism and animal-to-human transmission

Upon prion transmission via peripheral routes and prior to invasion of the CNS, PrP^Sc is sequestered by secondary lymphoid organs (SLOs) including spleen and lymph nodes (Huang et al., 2002; Takakura et al., 2011; Castro-Seoane et al., 2012; Michel et al., 2012). In fact, PrP^Sc replicates in spleen and lymph nodes independently of its replication in the CNS (Brown et al., 1999; Montrasio et al., 2000; Kujala et al., 2011; McCulloch et al., 2011). Some but not all prion strains exhibit lymphotropism, i.e., the ability to invade and replicate in SLOs (Aguzzi et al., 2013). It is not known whether the lack of lymphotropism is attributed to deficient trafficking of PrP^Sc to SLOs, impaired replication in SLOs, fast clearance in SLOs, or a combination of these factors.

The two major types of human Creutzfeldt-Jakob disease (CJD), sporadic CJD (sCJD) and variant CJD (vCJD), are different with respect to their lymphotropism, with vCJD known to be significantly more lymphotropic than sCJD (Hill et al., 1999; Wadsworth et al., 2001; Halliez et al., 2014). Being predominantly diglycosylated, vCJD PrP^Sc is expected to have a significantly higher density of sialic acid residues relative to sCJD PrP^Sc, which is predominantly monoglycosylated (Zanusso et al., 2004; Pan et al., 2005). Does sialylation status control lymphotropism? Studies of hepatocarcinoma demonstrated that sialylation of N-linked glycans via α2–6 linkages is responsible for directing the traffic and selective adhesion of hepatocarcinoma cells to SLOs (Zhang et al., 2013; Wang et al., 2015). Another work that employed synthetic glycoclusters demonstrated that in circulation, the glycoclusters with α2–6 linked sialic acid residues were more stable and showed slower clearance rates in comparison to the glycoclusters with α2–3 linkages (Tanaka et al., 2010). If the trafficking of CJDs relies on sialylation status the same way as trafficking of hepatocarcinoma, vCJD is expected to be more lymphotropic than sCJD. The relative ranking of the two types of CJDs with respect to sialylation is consistent with the hypothesis that sialylation is an important factor that controls trafficking of PrP^Sc to SLOs.

To test whether sialylation indeed dictates prion lymphotropism, we monitored trafficking of partially desialylated PrP^Sc to SLOs upon intraperitoneal administration (Srivastava et al., 2017). By 6 h post-inoculation, PrP^Sc with normal sialylation status was found primarily in spleen and lymph nodes, whereas partially desialylated PrP^Sc was targeted predominantly to the liver (Srivastava et al., 2017). This study argues that PrP^Sc sialylation is indeed critical for trafficking of prions to SLOs.

How does PrP^Sc survive clearance by the innate immune cells in SLOs? Upon sequestration by SLOs, PrP^Sc replicates in the germinal centers of the spleen and lymph nodes (Brown et al., 1999; Montrasio et al., 2000; Kujala et al., 2011; McCulloch et al., 2011). Surprisingly, in SLOs, PrP^Sc was found to be sialylated at considerably higher levels relative to PrP^Sc in a brain (Srivastava et al., 2015). For strains that manage to reach SLOs, enhanced sialylation in SLOs was observed regardless of prion strain, host species, or inoculation route (Srivastava et al., 2015) arguing that enhanced sialylation is a generic property of spleen-derived PrP^Sc. Notably, enhanced sialylation of PrP^Sc was attributed to post-conversion sialylation by extracellular sialyltransferases, but not to elevated levels of sialylation of PrP^C expressed in SLOs. While sialyltransferases are traditionally believed to localize within the trans-Golgi (Harduin-Lepers et al., 2001), a number of studies reported sialyltransferase activities in circulation or on surfaces of the cells of the immune system including polymorphonuclear leukocytes, monocyte-derived dendritic cells, lymphocytes, and T cells (Gross et al., 1996; Kaufmann et al., 1999; Schwartz-Albiez et al., 2004; Rifat et al., 2008; Cabral et al., 2010; Nasirikenari et al., 2014). Consistent with the hypothesis that extracellular sialyltransferases are involved in enhancing sialylation of PrP^Sc, the sialylation status of donor PrP^Sc acquired via peripheral administration was found to be elevated upon colonization of SLOs (Srivastava et al., 2015). Moreover, enhanced sialylation of PrP^Sc was recapitulated in vitro by incubating brain-derived PrP^Sc with primary splenocytes or cultured macrophage RAW 264.7 cells, whereas inhibitors of sialyltransferases suppressed enhanced sialylation of PrP^Sc (Srivastava et al., 2015). It is not known whether post-conversion sialylation camouflages PrP^Sc in SLOs, protecting it from clearance by innate immune cells. It is also not known if the ability of SLOs to enhance sialylation accounts for the high permissiveness of these tissues to prion infection upon cross-species transmission (Béringue et al., 2012).

Humans can synthesize only Neu5Ac, whereas non-human mammals synthesize both Neu5Ac and Neu5Gc (Varki, 2010). In non-human mammals, Neu5Gc is the predominant type of sialic acid residue expressed in the periphery (Fig. 1b) (Varki, 2010). The difference in the type of sialic acids synthesized by humans and other mammals raises several important topics for discussion. The first is whether this difference contributes to the animal-to-human prion transmission barrier. Notably, in humans with high consumption of red meat, Neu5Gc incorporates metabolically into cells and induces antibody responses against Neu5Gc (Samraj et al., 2015). While incorporation of Neu5Gc increases the likelihood of systemic inflammation (Samraj et al., 2015), antibodies against Neu5Gc might be beneficial for neutralizing prion infection of zoonic origin. The second important topic to consider is the functional consequences of enhanced sialylation of foreign PrP^Sc in SLOs (Srivastava et al., 2015). Enhanced sialylation in human SLOs could “humanize” prions of animal origin by decorating them with Neu5Ac and helping to deceive the human immune system (Srivastava et al., 2015). The third topic is related to the differences in the binding sites of Siglecs expressed in humans and non-human mammals with respect to selective recognition of Neu5Ac over of Neu5Gc (Varki, 2010). Siglecs are a family of sialic acid-binding proteins with a number of important functions (reviewed in (Rabinovich and Croci, 2012)). While interactions between prions and Siglecs have not been documented (Bradford et al., 2014), such a possibility should not be excluded considering the large number of Siglecs expressed in mammals. The differences between humans and non-human mammals with respect to the affinity of Siglecs to Neu5Ac and Neu5Gc are also important for the critical assessment of the results on prion transmission to transgenic mice expressing human PrP genes.

Role of sialylation of PrP^Sc in neuroinflammation and neurotropism

Activation of microglia and astrocytes have been recognized as obligatory features of prion diseases (Lu et al., 2004; Gomez-Nicola et al., 2013; Vincenti et al., 2016; Makarava et al., 2019; Makarava et al., 2020c; Baskakov, 2021b; Kushwaha et al., 2021; Makarava et al., 2021). Several protein families involved in innate immunity including Siglecs, selectins, galectins, complement system components, mannose receptors, and asialoglycoprotein receptors recognize carbohydrate groups (Linnartz et al., 2012a; Rabinovich and Croci, 2012). The majority of carbohydrate-binding molecules have multivalent binding sites, so the strength and selectivity of binding is not only dependent on the composition of functional carbohydrate epitopes but also on their density and spatial configuration of carbohydrate groups on PrP^Sc surfaces.

Is sialylation of PrP^Sc responsible for the pro-inflammatory phenotype of glia? What molecular features of PrP^Sc do glia sense and respond to? Exposure of primary microglia or BV2 cells to purified PrP^Sc was found to trigger pro-inflammatory responses characterized by an increase in the levels of TNFα, IL6, nitric oxide, and expression of inducible nitric oxide synthase (Srivastava et al., 2018). Notably, partial cleavage of sialic acid residues on PrP^Sc surfaces boosted the inflammatory response of microglia to PrP^Sc (Srivastava et al., 2018). Moreover, transient degradation of Iκβα observed upon treatment with partially desialylated PrP^Sc suggests that the canonical NFκB activation pathway is involved in inflammatory response (Srivastava et al., 2018). Thus, PrP^Sc can directly trigger a proinflammatory response in microglia, where the degree of response is dictated by the level of desialylation of PrP^Sc (Fig. 3a, b).

Figure 3.

A schematic representation of the hypothesis proposing that carbohydrate epitopes on PrP^Sc surfaces determine the outcomes of prion infection. a A high density of glycans with terminal sialylation leads to low grade chronic neuroinflammation. b A high density of asialo-glycans with exposed galactose triggers an “eat me” signal in glia and leads to profound neuroinflammation. c Atypical PrP^Sc has a low density of glycosylation and sialylation; it does not trigger an “eat me” signal, nor does it induce chronic neuroinflammation. d In mice expressing unglycosylated PrP^C, only partial attack rate along with significant delay in disease or complete lack of clinical disease were observed upon challenges with prions. Adapted and modified from Baskakov et al. (2018)

Consistent with the idea that asialo-branches on N-glycans dictate the degree of neuroinflammation are recent studies, in which the most profound and widespread neuroinflammation was observed in mouse-adapted SSLOW strain (Makarava et al., 2012a). SSLOW is characterized by the highest percentage of asialo N-glycans when compared to other mouse-adapted strains (Makarava et al., 2020b; Makarava et al., 2021). Colocalization of SSLOW PrP^Sc with microglia supports the idea that PrP^Sc with asialo N-glycans is subject to phagocytosis by microglia (Makarava et al., 2020b; Makarava et al., 2021) (Fig. 3b). In the reactive state associated with prion disease, the phagocytic activity in microglia is upregulated (Sinha et al., 2021). However, it is not known whether asialo N-glycans serve as the main phagocytic cues sensed by reactive microglia. Moreover, because reactive microglia phagocytose neurons along with PrP^Sc (Sinha et al., 2021), it is also not known whether upregulation of phagocytic clearance has net positive or negative impact on disease progression. Another question of great interest is whether phagocytic activity can be selectively upregulated or downregulated toward specific substrates.

Another example that sheds light on the relationship between molecular features of PrP^Sc and neuroinflammation deals with a unique, transmissible PrP^Sc-like state which has substantially lower levels of glycosylation and sialylation relative to those of PrP^Sc (this will be referred to as atypical PrP^Sc) (Makarava et al., 2011; Makarava et al., 2012b; Kovacs et al., 2013). Atypical PrP^Sc is a self-replicating, transmissible state that accumulates as small synaptic deposits and large plaques, but does not cause neuronal death, pathological lesions or clinical signs associated with prion diseases (Makarava et al., 2011; Makarava et al., 2012b; Kovacs et al., 2013; Makarava et al., 2015; Makarava et al., 2016). Remarkably, atypical PrP^Sc recruits primarily unglycosylated and monoglycosylated PrP^C at the expense of diglycosylated PrP^C molecules (Makarava et al., 2015). However, unlike PrP^Sc with asialo N-glycans, atypical PrP^Sc does not expose terminal galactose at a high density, which is considered to be an “eat me” signal. As a result, atypical PrP^Sc replicates and accumulates steadily in the CNS in the absence of considerable neuroinflammation (Fig. 3c) (Kovacs et al., 2013).

Transgenic mice that lacked both N-linked glycosylation motifs in PrP^C provide additional illustration in support of a relationship between glycosylation status of PrP^Sc and disease outcomes. In mice expressing unglycosylated PrP^C, only partial attack rate along with significant delay in disease progression were observed upon intracranial challenges with 79A strain, or complete lack of clinical disease were found upon challenge with ME7 strain (Cancellotti et al., 2010). Regardless of the clinical status, diffuse PrP^Sc deposits or large PrP^Sc plaques were found in brains of PrP^C glycan-deficient mice challenged with 79A or ME7 (Cancellotti et al., 2010). Moreover, brain material from mice with unglycosylated PrP^C transmitted infection to wild-type mice suggesting that host with unglycosylated PrP^C replicates PrP^Sc and sustains prion infection (Tuzi et al., 2008). Significant delay or lack of clinical disease in host with unglycosylated PrP^C support the hypothesis that N-glycans along with their sialylation status are important for dictating the disease outcomes (Fig. 3d). It would be interesting to learn about neuroinflammation status of mice expressing unglycosylated PrP^C and challenged with prions. In addition to receptors that recognize carbohydrate groups, microglia express a broad range of receptors that do not rely on carbohydrate groups for phagocytic uptake. For instance, phagocytic uptake of A peptides, A oligomers, or fibrils is driven by cues that do not rely on glycosylation (Baik et al., 2016; Huang et al., 2021). Similarly, carbohydrate-independent mechanisms of PrP^Sc sensing by glia should exist.

Prion strains invade brain regions in a strain-specific manner, a phenomenon referred to as selective neurotropism (Collinge and Clarke, 2007; Piro et al., 2009). Closely related to this phenomenon is selective vulnerability of brain regions to prion infection (Makarava et al., 2020a). The thalamus displays prion deposition and chronic neuroinflammation prior to cortex and hippocampus and is regarded as one of the most vulnerable regions. By the terminal stage of the disease, the thalamus is affected more severely than any other brain regions (Karapetyan et al., 2009; Sandberg et al., 2014; Carroll et al., 2016; Makarava et al., 2019; Makarava et al., 2020c; Makarava et al., 2021). Recent studies revealed that in addition to strain-specific differences, PrP^Sc is sialylated in a brain region-specific manner (Makarava et al., 2020a). In thalamus and brain stem, PrP^Sc was sialylated less than in cortex and hippocampus (Makarava et al., 2020a). Lower levels of sialylation promotes faster prion replication and triggers profound neuroinflammation, so this study suggested that region-specific differences in the sialylation status of PrP^Sc is an important parameter that dictates the timing of the spread and the severity of prion pathogenesis in different brain regions. In support of this hypothesis, mouse-adapted SSLOW, which displays the lowest sialylation levels among mouse-adapted strains, was found to induce the most profound neuroinflammation of PrP^Sc across brain regions (Makarava et al., 2020b; Makarava et al., 2021).

Selective, strain-specified recruitment of PrP^C sialoglycoforms

Do N-glycans play a role in defining strain-specific PrP^Sc structures? For examining the relationship between glycosylation status and the strain-specific structure of PrP^Sc, we proposed two alternative views (Baskakov et al., 2018). According to one view, prion strains can partially overcome spatial and electrostatic constraints imposed by N-glycans by selectively recruiting those PrP^C sialoglycoforms that fit into a strain-specific PrP^Sc structure. The alternative view proposed that recruitment is not selective. Instead, PrP^C glycoforms are recruited proportionally to their relative representations in a pool of molecules expressed by a cell. However, under these circumstances, the spectrum of PrP^Sc structures is expected to be limited to those that can accommodate a high percentage of diglycosylated glycoforms, which account for approximately 80% of all PrP^C molecules.

Analysis of sialoglycoform composition using 2D Western blots revealed that prion strains exhibit a broad selectivity range in recruiting PrP^C sialoglycoforms (Fig. 4) (Katorcha et al., 2015b). Consistent with the first hypothesis, some strains do show preferences for monoglycosylated and unglycosylated PrP^C while excluding highly sialylated, diglycosylated molecules (Fig. 4). However, in support of the second view, there are strains that do not display strong preferences in recruiting PrP^C with respect to their glycosylation or sialylation status (Fig. 4). Because 2D Western blotting provides limited information about glycan heterogeneity, “non-selective” strains might still show preferences based on features that are not detectible by 2D Western blots. In summary, prion strains display a broad range of selectivity of recruitment ranging from non-selective to highly selective. Consistent with the hypothesis on selective recruitment, unglycosylated PrP^C molecules were found to be required for propagating the RML strain, which belongs to the group of selective strains (Nishina et al., 2006).

Figure 4.

A schematic diagram illustrating strain-specified selective recruitment of PrP^C sialoglycoforms. The left panel shows distribution of PrP^C molecules according to their glycosylation status (in the horizontal dimension) and sialylation status (in the vertical dimension) ranging from non-sialylated to highly sialylated molecules. PrP^C molecules are shown as blue circles and sialic acid residues — as red diamonds. The panels on the right show 2D Western blots of three prion strains with different recruitment selectivity. Symbols on the top of the 2D blots indicate positions of hypersialylated (left) and hyposialylated (right) PrP molecules in 2D blots. 263K (strain #1) recruits PrP^C sialoglycoforms without strong preferences. Hypersialylated PrP^C molecules are excluded from RML (strain #2). Diglycosylated and hypersialylated PrP^C molecules are excluded even stronger from atypical PrP^Sc (strain #3). As a result of strain-specific exclusion of highly sialylated PrP^C, ratios of glycoforms within PrP^Sc shift toward monoglycosylated and unglycosylated glycoforms, as illustrated by corresponding 1D Western blots. Adapted from Baskakov and Katorcha (2016).

Strain-specific ratios of glycoforms within PrP^Sc have been used for decades for identification and differentiation between strains and subtypes of CJD (Kuczius et al., 1998; Lawson et al., 2005; Hill et al., 2006; Kang et al., 2020). The model on strain-selective recruitment proposes that the strain-specific differences in glycoform ratios arise as a result of negative selection of heavily sialylated, diglycosylated PrP^C molecules carrying bulky, highly charged glycans (Katorcha et al., 2015b). Remarkably, upon replication in vitro using desialylated PrP^C as a substrate, prion strains were found to lose their strain-specific glycoform ratios and acquire uniform glycoform ratios, which represent the proportion of unglycosylated, monoglycosylated, and diglycosylated isoforms of PrP^C (Katorcha et al., 2015b). Removal of sialic acid residues from the terminal position of N-glycans was shown to be sufficient to restore uniform glycoform ratios across prion strains suggesting that sialylation status is more important than the size of glycans for defining strain-specific glycoform ratios (Katorcha et al., 2015b).

A number of point mutants associated with inherited prion diseases including T183A, V180I, D178N, F198S, and E200K (human PrP sequence) are located in close proximity to the glycosylaton sites. Interestingly, glycoform ratios in PrP^Sc associated with these and others, even more distant mutations, were found to be different from those observed in PrP^Sc in sCJD or vCJD (Hill et al., 2006). Unique glycoform profiles of familial PrP^Sc is consistent with the hypothesis that prion strains associated with inherited prion diseases are conformationally different from sCJD or vCJD strains and, as such, select unique subsets of PrP^C sialoglycoforms. Alternatively, unique glycoform profiles could be attributed to altered cellular trafficking along with changes in composition of N-glycans of familial PrP^C mutants. Indeed, a loss of glycosylation at N181 site and altered glycosylation at N197 site was found for T183A PrP^C mutant (Grasbon-Frodl et al., 2004; Zou et al., 2011). F198S mutations altered glycosylated at both sites (Zou et al., 2011). Analysis of sialylation status of PrP^C mutants associated with familial prion disease along with corresponding PrP^Sc will help to answer the question whether altered cellular trafficking/glycosylation of PrP^C mutants, selection of unique subsets of PrP^C syaloglycoform, or both contribute to phenotypic heterogeneity associated with inherited prion diseases.

Role of N-glycans and sialic acid residues in defining the fidelity and rate of prion replication

According to a well-accepted model, prions replicate via a template-assisted mechanism, which postulates that the folding pattern of a newly recruited PrP polypeptide chain accurately reproduces that of PrP^Sc template (Cohen and Prusiner, 1998). The traditional template-assisted mechanism postulates that the high fidelity of prion replication is attributed to compatibility between amino acid sequences of PrP^C and PrP^Sc, whereas potential role of N-glycans has been ignored. We proposed that N-glycans are important for maintaining high fidelity of prion strain replication (Baskakov et al., 2018). Specifically, loss of structural constraints imposed by PrP^C glycans might impair fidelity to the extent that in some cases, it leads to changes of strain-specific properties and disease phenotype. In support of this view, transmission of prions to hosts expressing PrP^C deficient in N-glycans was found to change strain-specific characteristics of the 79A strain; however, the transmission did not significantly affect strain-specific properties of ME7 or 301C strains (Cancellotti et al., 2013). Another support of the idea that N-glycans are important for maintaining high fidelity is offered by quaking or amyloid seeding assays, in which recombinant PrPs that lack N-glycan are used as substrates for formation of amyloid fibrils upon seeding by PrP^Sc (Atarashi et al., 2007; Colby et al., 2007). While amyloid seeding assays are extremely sensitive for detecting minute amounts of PrP^Sc (Atarashi et al., 2007), loss of prion infectivity and PrP^Sc-specific structure upon fibrillation of recombinant PrP in vitro argues that N-glycans are important for maintaining high fidelity of replication.

For probing a relationship between strain-specific characteristics and incubation time to disease, previous studies interrogated a number of physical features of PrP^Sc. Among them are conformational stability, intrinsic fragmentation and elongation rates, resilience to proteolytic clearance, and the size of aggregates (Sun et al., 2008; Colby et al., 2009; Tixador et al., 2010; Ayers et al., 2011; Gonzalez-Montalban et al., 2011; Gonzalez-Montalban and Baskakov, 2012; Morales et al., 2016), whereas possible role of N-glycans in defining prion replication rates has been overlooked (Baskakov, 2017). Due to electrostatic repulsion, the high density of sialic acid residues on PrP^Sc surfaces is expected to slow down the rate of prion replication. Indeed, a significant increase in replication rates was observed when partially desialylated PrP^C was used as a substrate instead of PrP^C in protein misfolding cyclic amplification reactions (Katorcha et al., 2014; Katorcha et al., 2015b). A rise in replication rates was found to be strain-specific ranging from a 10-fold to ~ a 10⁶-fold increase (Katorcha et al., 2014; Katorcha et al., 2015b). Consistent with the idea that negative charges of sialic acid residues dictate replication rates, a significant increase in efficiency of amplification along with sensitivity of PrP^Sc detection was described upon conducting reactions using unglycosylated PrP^C as a substrate for several subtypes of sCJD (Camacho et al., 2019). The fact that the replication rates are increased in a strain-specific manner upon removal of sialic acid residues suggests that sialic acid decorates surfaces of PrP^Sc in a strain-specific fashion, giving rise to a strain-specific pattern of negative charges.

Role of sialylation in strain competition and adaptation

Strain-specified selection of PrP^C sialoglycoforms is an important parameter to be considered for examining strain adaptation, competition, and evolution. As a general notion, highly selective strains rely on a narrow range of PrP^C sialoglycoforms as substrates and are unlikely to be strong competitors against non-selective strains. Vice versa, strains capable of recruiting a broad range of PrP^C sialoglycoforms have a greater chance of succeeding in competition against selective strains.

Studies on the evolution of prions of synthetic origin provide insightful illustrations on how differences in selectivity of recruitment between alternative self-replicating states dictate the outcomes of competition. Despite the fact that amyloid fibrils generated in vitro using recombinant PrP are structurally different from bona fide PrP^Sc (Bocharova et al., 2005; Sun et al., 2007; Tycko et al., 2010; Baskakov et al., 2019; Spagnolli et al., 2019), recombinant PrP fibrils induced transmissible prion diseases along with authentic PrP^Sc upon serial passaging in non-transgenic animals (Makarava et al., 2010; Makarava et al., 2011; Makarava et al., 2012b). Detail analysis of brain tissues revealed that the first product of PrP^C misfolding triggered by PrP fibrils in vivo was atypical PrP^Sc, a self-replicating state with a strong preference for recruiting unglycosylated and hyposialylated monoglycosylated PrP^C (Makarava et al., 2011; Makarava et al., 2012b; Makarava et al., 2015; Makarava et al., 2016). Replication of atypical PrP^Sc in animals gave rise to authentic PrP^Sc via the mechanism referred to as deformed templating (Makarava and Baskakov, 2013). Only a small fraction of PrP^C could be recruited as a substrate by atypical PrP^Sc, so its replication rate was much slower relative to that of PrP^Sc and, upon serial transmission, atypical PrP^Sc was gradually replaced by PrP^Sc (Makarava et al., 2012b; Makarava et al., 2015; Makarava et al., 2016). Remarkably, the outcomes of competition between atypical PrP^Sc and PrP^Sc can be shifted in favor of atypical PrP^Sc in vitro by using deglycosyated PrP^C as a substrate for protein misfolding cyclic amplification reactions (Makarava et al., 2013). These studies illustrated that knowledge on selective recruitment could be useful for designing experimental approaches for selective amplification of self-replicating states from a mixture as well as improving sensitivity of detection of individual self-replicating states.

Cross-species transmission of prions is controlled by a species barrier that manifests itself in a low attack rate, prolonged incubation time to clinical disease, or lack of disease (Katorcha et al., 2018). When cross-species transmission is followed by serial passaging in a new host, strains can gradually adapt to a new species, a phenomenon known as prion strain adaptation (Baskakov, 2014). Recent studies demonstrated that in addition to amino acid sequences of host PrP^C and donor PrP^Sc, constraints generated by N-linked glycans dictate a parallel set of rules that govern strain adaptation (Makarava et al., 2020b). The majority of hamster prion strains display minimal constraints and can accommodate diglycosylated and highly sialylated PrP^C molecules, whereas mouse prion strains exhibit much stronger constraints and preferentially exclude hypersialylated and diglycosylated PrP^C isoforms (Katorcha et al., 2015b). In recent studies, the adaptation of a hamster strain to mice was accompanied by dramatic changes in selective recruitment of PrP^C sialoglycoforms, giving rise to a new strain with a unique sialoglycoform composition and disease phenotype (Makarava et al., 2020b). A unique disease phenotype along with a highly distinctive sialoglycoform signature suggested that a causative link between PrP^Sc sialoglycoform composition and disease phenotype exists. Notably, lengthy adaptation to mice over five serial passages was accompanied by an equally lengthy process of transformation in the selectivity of recruitment of sialoglycoforms along with changes in conformational stability, arguing that adjustment in the composition of glycans occurs in parallel to tuning of PrP^Sc structures. To summarize, in addition to congruency between the amino acid sequences of host PrP^C and the structure of donor PrP^Sc, constraints generated by N-linked glycans dictate a parallel set of rules that govern strain adaptation (Makarava et al., 2020b).

Concluding remarks

Mounting evidence suggests that N-linked glycans of the prion protein play an important role in defining strain-specific structures and disease phenotypes. On PrP^Sc surfaces, N-glycans expose negatively charged sialic acid residues. For overcoming constraints imposed by negative charges of sialic acids, prion strains exhibit a wide range of selectivity in recruiting PrP^C sialoglycoforms. While some strains recruit sialoglycoforms proportionally to their representation in PrP^C, others avoid diglycosylated and highly sialylated PrP^C. As a result of selective recruitment, individual strain-specific patterns of functional carbohydrate epitopes are formed on PrP^Sc surfaces. Among terminal carbohydrate groups, the innate immunity primarily senses galactose and sialic acid residues, which serve as molecular cues for triggering responses by glia. In agreement with the general concept that terminal galactose generates an “eat me” signal, desialylation of PrP^Sc ablates its ability to infect a host. Sialylation status of PrP^Sc is primarily dictated by strain, yet also depends on tissues and brain region. Sialylation affects the spread of PrP^Sc across the brain via several mechanisms. On one hand, the rate of PrP^Sc replication is accelerated by PrP^C with low sialylation levels. As such, brain regions with mild sialylation of PrP^C are more susceptible to the spread of prions than regions expressing PrP^C with heavy sialylation. On the other hand, highly selective strains that prefer PrP^C with low sialylation levels rely on less substrate in comparison to non-selective strains. Moreover, selective strains are likely to expose more terminal galactose residues than non-selective strains, and are subject to enhanced phagocytosis by microglia which use asialo-glycans for identification of phagocytic targets. All aforementioned factors come into play when defining outcomes of prion strain evolution, which is a result of a delicate balance between recruiting highly sialylated glycoforms, generating sufficient density of sialylation for avoiding an “eat-me” response by glia, and limiting heavily sialylated glycoforms for enabling efficient prion replication.