Glycoregulation of E3(SCF) ubiquitin ligases in unicellular eukaryotes

Abstract

Skp1 is an essential adaptor within the Skp1/Cul1/F-box (SCF) class of E3 polyubiquitin ligases that regulate protein degradation in all eukaryotes. Skp1 is also a target of a 5-enzyme glycosylation pathway in parasites and other unicellular eukaryotes. Glycosylation of Skp1 is contingent upon oxygen-dependent hydroxylation of a critical Pro residue by a homolog of the HIFα PHD2 oxygen sensor of animals. The resulting hydroxyproline is modified by a series of soluble, cytoplasmic, sugar nucleotide-dependent glycosyltransferases that vary among branches of protist evolution, and are evolutionarily related to counterparts in the Golgi and the cytoplasm of prokaryotes. Pair-wise gene fusions of the six enzymes occur in various protists, suggesting processing efficiency. The terminal glycosyltransferases exhibit a second site interaction with Skp1 that may modulate its function irrespective of glycosylation status. The pentasaccharide adopts a constrained fold that in turn promotes Skp1 conformations that inhibit sequestration by homodimerization and encourage binding to select F-box protein substrate receptors with varied effects on their expression levels. The occurrence of a second Skp1 copy in some protists that is resistant to modification indicates a mechanism to bypass glycoregulation. This review details evidence from the social amoeba Dictyostelium discoideum and the pathogens Toxoplasma gondii and Pythium ultimum for the specificity of the enzymes for Skp1 and their regulation, as support for a role in regulating protein turnover via E3(SCF) ubiquitin ligases, and in turn sensing oxygen at the cellular level.

Introduction to Skp1 Modifications

The origin of a novel form of regulation of the nucleocytoplasmic protein Skp1 traces back to the discovery of its glycosylation, when in 1992 glycosylation was thought to be the nearly sole provenance of the secretory pathway. The glycan is a linear pentasaccharide consisting of 3 or 4 different monosaccharide types in widely diverged protists. It is attached to a hydroxyproline generated by a homolog of a prolyl 4-hydroxylase involved in oxygen sensing in animals. As described herein, the Skp1 glycan is assembled by conventional but freely soluble sugar nucleotide dependent GTs related to those found in the Golgi or the cytoplasm of prokaryotes. In contrast to most known glycosylation pathways, the resulting glycan has been found on only a single, highly conserved protein, Skp1, an adaptor subunit of a major class of polyubiquitin ligases. Further evidence indicates that the glycan lacks microheterogeneity and functions autonomously to influence the conformation of Skp1 rather than serving as a ligand for another effector protein. This pathway is distinct from other forms of nucleocytoplasmic glycosylation (West et al. 2022), including the O-GlcNAcylation of thousands of animal and plant proteins by the CAZy GT41 family member OGT (O-GlcNAc-transferase) (Zachara et al. 2022), and the paralogous process of O-fucosylation in protists and plants (van der Wel et al. 2023; Aizezi et al. 2025). In comparison, complex O-glycosylation is rarely documented in this compartment (Funakoshi and Suzuki 2009), with the best characterized example being the Tyr-linked oligo-glucose generated by the self-glucosylating CAZy GT8 family glycogenin that serves as a primer for glycogen synthesis in yeast and animals (Roach et al. 2012). This article updates past summaries (West et al. 2010; West and Blader 2015) about the assembly and effects of Skp1 modification in vitro and in the social amoeba Dictyostelium discoideum and the protist parasite Toxoplasma gondii.

Skp1 and the SCF Complex

Skp1, the target of complex cytoplasmic glycosylation in unicellular eukaryotes, is an adaptor subunit in Skp1/Cullin-1/F-box protein/RING (SCF) complexes (Fig. 1A). Also known as CRL1, SCF complexes are a major class of E3 Cullin Ring Ubiquitin Ligases (CRLs) that primarily mark proteins for degradation by the 26S proteasome. These ligases typically generate linear K48-linked ubiquitin (Ub) chains that direct the degradation of thousands of proteins (Willems et al. 2004; Harper and Schulman, 2021). The SCF family was originally discovered in yeast and cancer cells and has been extensively characterized in yeast, animals and higher plants. Unicellular protists have been less studied but accumulating evidence indicates that SCF complexes are also ubiquitous throughout this kingdom (e.g., Huysman et al. 2014; Silmon de Monerri et al. 2015; Rojas et al. 2017; van der Wel et al. 2019; Kim et al. 2022; Rizvi et al. 2024). Target specificity is mediated by members of a family of F-box proteins (FBPs) that number from dozens to hundreds in different eukaryotes (Hua et al. 2011; Skaar et al. 2013; Yao et al. 2019; Boland et al. 2022; Mandalasi et al. 2024). FBPs are linked through the Skp1 adaptor protein to Cullin1 (Cul1) (Zheng et al. 2002). Cul1 in turn acts as a scaffold bringing the substrate into the vicinity of a Ub-charged E2 donor that is linked to Cul1 via Rbx1. The Ub-E2 is replaced during each cycle of Ub transfer, and subsequent Ubs are linked to the K48 of the underlying Ub though exceptions occur. Poly-Ub chains are recognized by receptors that direct the target to the 26S proteasome to complete the process. SCF mediated protein turnover has been implicated in a wide swath of biological processes (e.g., Skaar et al. 2013; Wang et al. 2014).

Figure 1.

SCF complex and regulation. A) The core of the SCF complex includes the F-box protein (FBP), Skp1, Cul1, and Rbx1. The Ub-charged E2 transfers Ub to a Lys residue on the substrate bound to the FBP, and is serially replaced by Ub-E2 to reiteratively build a polyUb chain that targets the substrate to the 26S-proteasome. Major general regulatory steps (in blue) include 1) priming the substrate for recognition by the FBP, 2) expression of an FBP that can recognize the substrate, 3) activation of transfer by neddylation using the E3 subunit Dcn1, 4a) deneddylation by the COP9 signalosome (8 subunits depicted), 4b) which enables binding of CAND1, 4c) which evicts the FBP/Skp1 subcomplex from Cul1, and 5) homodimerization of Skp1 which competes with FBP and Cul1 binding. Protist specific regulation (in red) includes O₂-triggered hydroxylation and glycosylation that 6a) inhibits homodimerization of Skp1 (in Toxoplasma) and 6b) modulates FBP/Skp1 interactions, and 7) sequestration of Skp1 by AgtA (in Dictyostelium) and possibly Gat1 (in Toxoplasma). Sugar symbols and enzymes of the Skp1 modification pathway are shown for both species.

B) Antibody specificity for Skp1 glycoforms. Synthetic glycopeptide and peptide conjugates corresponding to the region of the modified Pro of DdSkp1 (Table 1A) were linked through an unnatural N-terminal Cys residue to an immunogen carrier as shown at the right (panels a-e). Variations are contrasted with red font. These were used to induce mouse monoclonal (mAb) or rabbit polyclonal (pAb) antibodies. The Abs shown were used to probe a panel of mutant cell extracts that accumulated the indicated glycoforms. Ab specificity correlated with the glycoform used for its induction, with the exception of mAb 3A9 (panel a). The pan-specific mAb 4E1 was derived from an immunization with full-length recombinant Skp1 from E. coli (Kozarov et al. 1995) and serves as a loading control. Panels a, c-f are from West et al. (2015); panel b is from Chinoy et al. (2015).

C) TgSkp1 domain diagram annotated for contacts with other proteins. Adapted from Cantrell et al. (2025). Although not shown, interaction of human Skp1 with V₁-subunits of the V-ATPase does not compete with FBPs (Li et al. 2023).

D) The average structure over three molecular dynamics simulations for the fully glycosylated isoform of Dictyostelium Skp1 is displayed, rendered as secondary and space-filling models. α-helices 5–8 are labeled. The inset shows hydrogen bonds that are maintained for >50% of the simulations, and are colored according to the atom to which the hydrogen is connected. From Sheikh et al. (2017).

E) A representative frame from a molecular dynamics simulation of fully glycosylated Toxoplasma Skp1, zoomed in to reveal the interaction of the glycan with the loop between α-helix-7 and α-helix-8. Amino acids are numbered relative to the attachment at Hyp154. The glycan carbon atoms are in green and amino acid carbon atoms are in gray or orange, and dotted black lines depict hydrogen bonds contributing to polar energies. Left: ribbon representation of secondary structure elements. Middle and right: the lower half rotated to illustrate packing. Glycan and peptide are represented as sticks to show hydrogen bonds or by spheres to show van der Waals packing. From Mandalasi et al. (2020).

A variety of regulatory mechanisms support specific and accurate regulation of SCF activity (Harper and Schulman 2021; Baek et al. 2023) and provide precedents for protists. Most FBPs reside in stable complexes with Skp1, mediated by docking their F-box domains to the C-terminal half of Skp1. FBP/Skp1 pairs interact with Cul1 of the Cul1/Rbx1 pair, in a manner that is influenced by FBP substrate occupancy and is under continuous interference by competitive binding of Cand1 to Cul1 (Fig. 1A). Cand1 binding is inhibited by Neddylation, which is in turn removed by the COP9 signalosome. Other modes of SCF regulation have been described and, among these, regulation of Skp1/FBP interactions are a relatively unexplored area; this mode of regulation appears to be a primary target of Skp1 glycosylation in protists.

The profile of Skp1/FBP complexes in cells is altered by Skp1 glycosylation (Sheikh et al. 2015; Boland et al. 2022; Mandalasi et al. 2024). Glycosylation also inhibits homodimerization which, because of overlapping binding interfaces, competes with binding of Skp1 to FBPs (Cantrell et al. 2025). Evidence suggests that glycosylation acts directly on Skp1 to affect its conformational profile (Sheikh et al. 2014; Sheikh et al. 2017; Xu et al. 2018). Functional impact on SCF activities is indicated by compensatory effects of proteasome inhibitors (Boland et al. 2022; Mandalasi et al. 2024). These effects are in turn associated with regulation of O₂-dependent cellular processes owing to the dependence of glycosylation on prolyl hydroxylation (West et al. 2007; Xu et al. 2012a; Cordonnier et al. 2024). First discovered in the social amoeba Dictyostelium discoideum, similar processes and effects have been documented in the human pathogen Toxoplasma gondii and the plant pathogen Pythium ultimum (van der Wel et al. 2019). Genomics studies indicate that this unusual and currently novel modification occurs in multiple protists, algae, and select fungi, but not animals and higher plants. This article reviews characteristics of this unusual glycosylation pathway and critically examines the evidence that it forms the basis for SCF regulation and O₂-sensing at the cell and organismic levels.

The Modification Pathway of Skp1

The glycoprotein Skp1

The Skp1 modification, illustrated symbolically in Fig. 1A in the context of the full SCF complex, is currently the most complex known example of a cytoplasmic glycoprotein’s glycan with the exception of glycogen (West et al. 2022). Skp1 was initially discovered as a cytosolic glycoprotein, named FP21 or Fpa1, based on metabolic labeling with [³H]-fucose in growing cells of Dictyostelium (Gonzalez-Yanes et al. 1992). FP21 could also be fucosylated with GDP-[³H]-fucose in extracts of a mutant that cannot produce GDP-Fuc from GDP-mannose. The radioactive label served as a marker for its 20,000-fold purification to homogeneity using a series of conventional protein chromatography columns. Early work was confounded by its isolation from crude cytosolic extracts in two pools (West et al. 1997), which was later attributed to involvement in different complexes. Soon after, FP21 was found to be equivalent to the newly discovered Skp1 subunit of the SCF complex (Bai et al. 1996; Connelly and Hieter 1996). Fucose was identified as the middle sugar of a linear pentasaccharide attached to a specific hydroxylated proline residue based on mass spectrometric analysis of its tryptic peptides (Teng-umnuay et al. 1998). The hydroxylated Pro is highly conserved on at least one Skp1 copy in unicellular eukaryotes, plants and invertebrates, but was replaced by Glu in vertebrates (West et al. 2004). Early data on the radiolabeled glycosylated structure obtained after treatment using conditions of reductive alkaline degradation had been interpreted under the assumption that the glycan was O-linked to Ser or Thr (Gonzalez-Yanes et al. 1992), but was later revised because the Hyp-linkage is not susceptible to cleavage by β-elimination (Spiro 1970). As summarized here and detailed below, the identity of individual sugars and suggestions of linkages were indicated by sequential exoglycosidase studies. Further support for monosaccharide identities and linkages relied upon cloning the responsible enzymes and product analysis of in vitro reactions with radioactive sugar nucleotides and synthetic model acceptor substrates (Fig. 2A). NMR analysis of the fully glycosylated recombinant protein unequivocally established each glycosidic linkage, and methylation analysis of the glycopeptide confirmed the linear organization of the glycan (Sheikh et al. 2017). These findings in turn enabled molecular dynamics simulations of the glycan’s conformation and interactions with the Skp1 protein (Sheikh et al. 2017; Mandalasi et al. 2020). Much as occurs in Dictyostelium, glycosylation of Skp1 has been confirmed for the distantly related apicomplexan protist Toxoplasma (Rahman et al. 2016; Mandalasi et al. 2020), and the initial modification enzymes from P. ultimum are active in vitro (van der Wel et al. 2019). These organisms present a slightly divergent glycan as described below.

Figure 2.

Skp1 modification pathways. Enzyme proteins, substrates, domain diagrams including CAZy GT classifications, and evolutionary relatives of the protein domains are represented for (A) Dictyostelium discoideum, (B) Toxoplasma gondii, and (C) examples from Pythium ultimum and Bigelowiella natans.

The unique structural features of the Dictyostelium Skp1 glycoforms allowed for the generation of monospecific antibodies induced against synthetic peptides or glycopeptides is shown in Fig. 1B. However, attempts to synthesize the full-length pentasaccharide glycopeptide or isolate it from Skp1 were unsuccessful owing to its unexplained instability. Notably, greater success at generating monospecific Abs was achieved using outbred rabbits rather than inbred mice to recover specific antibodies (Wang et al. 2009; West et al. 2015). The interesting patterns of cross reactivity of some mouse mAbs (e.g., mAb 3A9) that react with glycoforms distinct from the immunogen suggest that the epitopes have conformational in addition to chemical determinants. This is supported by the observation that mAbs 1C9 and 3A9 were not specific at the peptide level in ELISA assays, only at the protein level. In addition, the epitope for UOK104 for the trisaccharide glycoform involves the peptide because UOK104 does not recognize the trisaccharide conjugated to other aglycons (Chinoy et al. 2015). The Abs have also been useful for differentiating glycoforms in Pythium (van der Wel et al. 2019) and Toxoplasma. The antibodies fill a void given that all tested lectins and commercial antibodies against type 1 blood group H antigen, whose structure is equivalent to Skp1’s core trisaccharide, have failed to recognize any of the glycoforms. This is consistent with peptides and glycopeptides favoring specific conformations (West et al. 2015), perhaps reliant on the Hyp-linkage and unique features of the surrounding peptide such as the high density of acidic residues (Table 1C). The Abs have been useful for in vitro enzyme assays and monitoring Skp1 glycoforms in vivo, because of their efficiency in whole-cell Western blots.

Table 1.

Comparison of C-terminal sequences of representative Skp1 proteins across phylogeny. To facilitate visualization of relatedness, acidic residues are in blue, basic in dark red, Gly and Pro in red, and hydrophobic in green, as previously described (West et al. 2002). Positions possessing a consensus chemical characteristic are highlighted in yellow (hydrophobic), gray (acidic), dark grey (basic), or teal (small). Positions of near perfect conservation are bolded. Red asterisk indicates the hydroxylated Pro residue. Bold black asterisks indicate conserved Asn residues that are modeled to be hydrogen bonded with the glycan in DdSkp1 and TgSkp1. Other black asterisks indicate stop codon.

Prolyl hydroxylation of Skp1

Prolyl hydroxylation of Skp1 by PhyA (originally named P4H1) primes Dictyostelium Skp1 for glycosylation at Pro143 (Wang et al. 2011), as represented in Fig. 2A. PhyA was first detected by incubating cytosolic extracts with recombinant Skp1 (from E. coli) and quantifying the reaction product by labeling with [³H]GlcNAc from UDP-[³H]GlcNAc with Gnt1 (van der Wel et al. 2005). However, activity was lost after partial purification from extracts. Bioinformatic analysis suggested that, among 5 Dictyostelium putative cytoplasmic prolyl hydroxylase, the sequence later named PhyA was most like the then recently discovered C. elegans Egl-9 and human PHD2 (West et al. 2004). This enzyme had just been recognized as the agent for O₂-control of the stability of HIFα, the transcriptional mediator of O₂-sensing in mammalian (including human) cells (Kaelin and Ratcliffe 2008). Disruption of phyA blocked Skp1 glycosylation, indicating that it was the sole Skp1 enzyme among the other predicted cytosolic prolyl hydroxylases (van der Wel et al. 2005). Highly purified recombinant PhyA was competent to hydroxylate Skp1 but not Skp1(P143A) in vitro, confirming its ability to perform the function directly. The 4-(trans)-Hyp (or 2S,4R) hydroxyproline configuration was established using proton NMR of the peptide isolated from in vitro hydroxylated Skp1, using synthetic dipeptides as reference standards (van der Wel et al. 2011; Liu et al. 2020). A similar function was described for the homolog from Toxoplasma, and its PhyA (TgPhyA) was capable of rescuing Skp1 hydroxylation in a phyA-KO strain of Dictyostelium (Xu et al. 2012a). Furthermore, a recombinant version of the homolog from a stramenopile protist, Pythium ultimum, was also active as a Skp1 prolyl hydroxylase in vitro (van der Wel et al. 2019).

A phylogenetic analysis of animal PHD2, protist PhyA, and related catalytic domain sequences in eukaryotes and prokaryotes indicates a deep separation between nucleocytoplasmic animal PHD and protist PhyA proteins, and prolyl hydroxylases that modify animal collagen and plant cell wall proteins in the rER (West et al 2004). Interestingly, hydroxyproline is glycosylated in arabinogalactan and extension proteins secreted from plant and algal cells (Showalter and Basu 2016; Bollig et al 2007), but the responsible GTs are independent of the Skp1 pathway and sequestered in the secretory system (West et al 2021). A more focused analysis suggests that animal PHDs descended from a protist PhyA (Florimond et al. 2019). Biochemical, structural and physiological studies suggest that, despite modifying distinct proteins, protist PhyA and animal PHD enzymes appear to be orthologs, at least at the sequence level. They catalyze the formation of (2S,4R) hydroxyproline (Hyp) on limited if not single targets (Fiorini and Schofield 2024). There is a striking degree of overlap of the X-ray crystal structures determined for Dictyostelium PhyA (DdPhyA) and PHDs from humans and the early metazoan Trichoplax adherens (Liu et al. 2020). The main structural difference is the absence of a C-terminal extension in DdPhyA. DdPhyA and human PHD2 (HsPHD2) belong to the large group of non-heme dioxygenases that depend on concentrations of its co-substrates O₂ and α-ketoglutarate (αKG or 2-oxo-glutarate) and an Fe(II) cofactor, and are supported by ascorbate (Herr and Hausinger 2018). Significantly, their in vitro K_m (O₂) values, measured on different acceptor constructs using different methods (van der Wel et al. 2005; Liu et al. 2020), are well above the 21% atmospheric O₂ concentration (at sea level), which when equilibrated with aqueous buffers corresponds to ~250 μM O₂. Thus their activity is rate-limited by O₂ consistent with a role in sensing O₂ in cells. Kinetic studies indicate similar K_m values for αKG and ascorbate, and both bind Fe(II) very tightly retaining it during purification. In a similar manner to PHD2 (Hewitson et al. 2007; Fiorini and Schofield 2024), DdPhyA is also inhibited by several Krebs cycle intermediates such as citrate and oxaloacetate and other general non-heme dioxygenase inhibitors (van der Wel et al. 2011). Considerable similarities are found between DdPhyA and TgPhyA but a dramatic difference is their K_m (O₂) values (van der Wel et al. 2005; Xu et al. 2012a; Liu et al. 2020). In contrast to DdPhyA and PHD2s, TgPhyA has its K_m below 1% O₂. While the structural basis for this difference is unclear, the evolutionary tuning of this kinetic parameter was likely important as earth’s atmosphere varied and fluctuated over the time of existence of eukaryotes, currently estimated to have ranged from 2 to 37% O₂ (Mills et al. 2023). It also addresses the distinct needs of Toxoplasma, which traverses anaerobic environments following oral ingestion, and Dictyostelium, which lives at higher O₂-levles in the soil. In comparison, their concentration dependences on αKG were indistinguishable (Xu et al. 2012a; Liu et al. 2020), as were their sensitivities to a panel of inhibitors (Liu T, Schofield CJ, van der Wel H, West CM, unpublished observations).

Skp1 is thus far the only known substrate of DdPhyA and TgPhyA, which conforms with HIFα being the only direct target confirmed for HsPHD2 (Cockman et al. 2019; Fiorini and Schofield 2024). However, identifying hydroxylation substrates can be challenging because of lack of facile detection methods, interference from spontaneous oxidation events, and indirect effects of genetic manipulations in cells. The interpretation that PhyA is specific for Skp1 in Dictyostelium and Toxoplasma is reinforced by related biochemical and physiological effects of mutating downstream GTs including Gnt1 compared to DdPhyA and TgPhyA (Zhang et al. 2012; Mandalasi et al. 2024). In contrast, evidence for GT specificity for Skp1 is strong owing to the availability of radioactive sugar precursors that afford high sensitivity of detection. The target Pro or Hyp is located in a partially disordered region of the C-terminus (Fig. 1C), but point mutations in the well folded N-terminal region interfere with in vitro (van der Wel et al. 2011) and in vivo (Teng-umnuay et al. 1998) activity. Peptides containing the target Pro or Hyp of up to 40 amino acids in length have far lower though measurable activity (van der Wel et al. 2011), consistent with distal determinants that are likely to be protein-specific. Indirect evidence indicates that alkylation of a specific internal Cys residue (C89) that is exposed after denaturation of full-length DdSkp1 followed by spontaneous refolding inhibits hydroxylation, and substitution of all six Cys residues with Ser is also inhibitory. However, these C89 is not conserved and none of the Cys residues is uniformly conserved in organisms with predicted PhyA enzymes. The presence of PhyA in co-IPs of Skp1 from Toxoplasma (Mandalasi et al. 2024), even when the great majority of Skp1 is glycosylated, is also consistent with PhyA’s unique specificity for Skp1. Furthermore, only one of the two Skp1s from P. ultimum is a PuPhyA-Gnt1 substrate in vitro, despite possessing the homologous Pro residue within an identical hexapeptide sequence (van der Wel et al. 2019). This again points to distal structural determinants that are more likely to be unique to Skp1 in the cytosolic environment.

HsPHD2, on the other hand, is reportedly highly active toward peptides containing either of the two target Pro sites in HIFα, which also occur in disordered regions (Koivunen et al. 2006). Though HIFα-like sequences appear to be absent from protists, Skp1 is highly conserved in all eukaryotes, notwithstanding that vertebrate Skp1 sequences lack the critical Pro (West et al. 2004). Though HsPHD2 does not hydroxylate Dictyostelium Skp1 (DdSkp1) or TgSkp1, DdPhyA and TgPhyA are quite active toward HsSkp1 in which the critical Pro is restored with an E147P replacement (van der Wel et al. 2011; Liu et al. 2020). Furthermore, HsSkp1 inhibited spontaneous αKG turnover by TgPhyA in the absence of substrate, confirming conservation of recognition determinants despite loss of the Pro target (Liu et al. 2020). These findings attest to a high degree of Skp1’s structural refinement that had already occurred during protist evolution, shaped at least in part by interactions with other subunits and regulatory proteins of the SCF complex, and consistent with the high sequence conservation of the FBP-combining region of Skp1 throughout phylogeny (Table 1C,D). The evolutionary mechanism of hypothesized target protein transition from Skp1 to HIFα remains unexplained. It is noted that hydroxylated HIFα is recognized by the VHL substrate receptor of CRL2, which contains the Skp1 homolog elongin C. Changing the substrate of the prolyl hydroxylase from a Ub ligase subunit to a substrate of a related Ub ligase would have changed the impact on the proteome from degradation of the FBP substrates to the biosynthesis of HIFα targets. In some respects, activating the SCF to degrade proteins is functionally analogous to inhibiting transcription of HIFα-dependent genes, but would manifest in different time frames.

The actual oligomeric state of Skp1 that is the native substrate for PhyA in vivo is unclear. Evidence shows that Skp1 is not a substrate for PhyA when bound to an FBP (van der Wel et al. 2011). Skp1 self-associates as a homodimer with a K_d of ~2.5 μM in vitro (Kim et al. 2020), similar to the value estimated for human Skp1 (Henzl et al. 1998). This is close to the estimated Skp1 concentrations in cultured human cells (Reitsma et al. 2017) and Dictyostelium (Gonzalez-Yanes et al. 1992). The in vivo concentration of free Skp1 is likely much less, however, as most will be incorporated into higher affinity complexes with FBPs (Reitsma et al. 2017). Interestingly, some assays indicate that the K_m of PhyA toward Skp1 is in the low to sub-μM range (van der Wel et al. 2011; Xu et al. 2012a), which is consistent with a generalization that an enzyme affinity converges to the range of its substrate concentrations. However, a separate study found a K_m for PhyA against Skp1 that was two orders of magnitude higher (Liu et al. 2020). The reason for the discrepancy is unknown but might be due to differences in the Skp1 isoforms, assay designs, or temperatures used. Apparent μM or lower K_m values for Skp1 have also been observed for most of the GTs (see below). The correlation between the homodimer K_d for DdSkp1 (2.5 μM) and the DdPhyA K_m (~2 μM) invites speculation that only the monomer is a substrate, which deserves investigation.

A comparison between the in vitro and in vivo substrate activities of Skp1 mutants suggests that other factors are important in cells. For example, ectopically expressed C-terminally tagged Skp1A1-myc , which possesses point mutations (I34T/D71G) far from the hydroxylation site that were adventitiously generated during PCR generation of the cDNA, is poorly hydroxylated in vivo (Teng-umnuay et al. 1998). It was originally interpreted that its expression level exceeded the capacity of prolyl hydroxylation but, subsequently, overexpressed native Skp1A-myc was found to be nearly fully Pro-modified (Sassi et al. 2001). Similar low modifications were found for Skp1A2-myc (K53A/V54S/L55S), which had distinct intentional point mutations (Sassi et al. 2001). In addition, appending an N-terminal FLAG-tag to ectopically expressed Skp1 renders it inert to hydroxylation and prevents formation of complexes with FBPs and other proteins (Wang et al. 2011). However, these mutant forms appear to have substrate activity similar to normal Skp1 in vitro (van der Wel et al. 2011), suggesting that N-terminal regions of Skp1 are important for efficient modification in vivo for reasons other than PhyA recognition. Thus unknown factors besides O₂, αKG, and other carboxylate metabolites may be important for regulating Skp1 hydroxylation in cells.

Generating the core trisaccharide

DdGnt1 (UDP-GlcNAc: Skp1-hydroxyproline α-N-Acetyl-D-GlcNAc transferase 1) is responsible for linking the first sugar of DdSkp1’s pentasaccharide to the hydroxyproline (Hyp) generated by PhyA (van der Wel et al. 2002b). Previously named Gnt51 based on its size, Gnt1 is a sugar nucleotide dependent GT that resides in the cytoplasm requiring a reducing environment for functionality. Activity appeared to be Mg²⁺-dependent owing to inhibition by MnCl₂ in the presence of MgCl₂ (Teng-umnuay et al. 1999). DdGnt1 was purified 130,000-fold from cytosolic extracts of Dictyostelium based on its activity toward the previously mentioned accidental mutant Skp1A1-myc ectopically expressed in and purified from Dictyostelium. This mutant was invaluable for assaying several pathway enzymes because it also accumulated hydroxylated-only and trisaccharide and tetrasaccharide isoforms in cells (Teng-umnuay et al. 1998). Purification of DdGnt1 was benefited by use of a tailor made UDP-GlcNAc affinity column and photoaffinity labeling by an iodinated GDP-GlcNAc derivative to tag it during purification (Teng-umnuay et al. 1999). Sequencing of peptides from the purified protein identified a gene (van der Wel et al. 2002b) whose subsequent disruption resulted in accumulation of HO-Skp1 in cells, confirming its hypothesized role. Similar findings were observed for the homolog in Toxoplasma (Rahman et al. 2016).

Gnt1 is highly homologous to Gnt2, a Golgi resident αGlcNAcT that initiates mucin-like O-glycosylation of proteins destined to the cell surface or for secretion in Dictyostelium (Wang et al. 2003; West et al. 2004). By analogy, Gnt1 was inferred to also be a retaining GT, and the α-linkage was later confirmed by NMR of the isolated glycopeptide (Sheikh et al. 2017). Phylogenetic analysis of Gnt1 indicates a close evolutionary relationship to several cytoplasmic, prokaryotic GT-A superfamily enzymes (van der Wel et al. 2002b), except for the occurrence of a conserved C-terminal domain required for activity in eukaryotes. The presence of both Gnt1 and Gnt2 or Gnt2-like GTs in other protists indicates a likely gene duplication event before the divergence of the two enzymes (West et al. 2004), and together the data suggest that the cytoplasmic version is ancestral (van der Wel et al. 2002b). Gnt1 and Gnt2 were the founding members of the CAZy GT60 family, and Gnt2 is likely ancestral to the relatively closely related Golgi-associated animal CAZy GT27 animal mucin type αGalNAc-transferases (Teng-umnuay et al. 1999; van der Wel et al. 2002b). GT27 GTs likely diversified when the UDP-Gal epimerase evolved to generate UDP-GalNAc from UDP-GlcNAc (Roper and Ferguson 2003).

The specificity of DdGnt1 was tested by supplementing gnt1-KO extracts with recombinant Gnt1 and UDP-[³H]GlcNAc. Skp1 was the only radiolabeled protein detected suggesting that it is the sole target (Zhang et al. 2012). DdGnt1 modification of HO-Skp1 relies on determinants in the folded region of Skp1, similar to those required by PhyA, well beyond the target Hyp (van der Wel et al. 2011). A 23-mer peptide including the target Hyp (Table 1A) is a specific substrate but has a 3000-fold higher K_m and 3-fold reduced V_max, consistent with the specificity determinants for PhyA. Intriguingly, PhyA and Gnt1 are expressed as a natural fusion protein in an unrelated protist from the stramenopile group, Pythium ultimum (van der Wel et al. 2019)(Fig. 2C), suggesting coordinate action that furthermore implies specificity of PhyA toward Skp1. Unfortunately, sequence comparisons of the 2 enzymes and the fusion enzyme have not revealed a basis for the shared specificities of the two enzymes. The apparent micromolar K_m values described for recombinant Gnt1 isolated from E. coli for both UDP-GlcNAc and HO-Skp1 (Teng-umnuay et al. 1999) suggest sufficient affinity to permit efficient GlcNAcylation of Skp1 at cytoplasmic Skp1 concentrations.

Additions of the second (βGal) and third (αFuc) monosaccharides are catalyzed by PgtA, a bifunctional di-GT with separate domains for each activity. PgtA was first identified as an α-fucosyltransferase (originally FTase or FT85) toward Skp1 isolated from a Dictyostelium mutant unable to synthesize GDP-Fuc from GDP-Man, causing accumulation of a disaccharide isoform of Skp1 that was a substrate for the fucosyltransferase activity (West et al. 1996). A survey of synthetic mono-, di- and trisaccharides used to compare mammalian fucosyltransferases associated with ABO and Lewis blood group antigens pointed to lacto-N-biose-1 (Galβ1,3GlcNAcβ-) as an optimal acceptor, while independent labs identified the fucose to be α2-linked to the Gal (Trinchera and Bozzaro 1996; West et al. 1996). Synthetic substrates in conjunction with GDP-[³H]Fuc were used to purify the activity as FT85 over 3 million-fold from cytosolic extracts of Dictyostelium. Although not pure by SDS-PAGE, selection of a candidate band by photoaffinity labeling with a radioiodinated probe provided sufficient information for sequencing PgtA and cloning by PCR. When PgtA was disrupted, Skp1 surprisingly accumulated as the monoglycosylated Gn-Skp1 glycoform rather than the expected GGn-Skp1 glycoform. This led to the realization that PgtA, with an M_r of 85,000, is a di-GT with separate domains that mediate its β3-GalT and α2-FucT activities (van der Wel et al. 2002a). Domain activities were assigned using site specific mutagenesis and independent expression. The α2FucT domain became the founding member of the CAZy GT74 family – unrelated to the CAZy GT10 family of α2FucTs known at the time. In comparison, the N-terminal CAZy GT2 domain, which commonly occurs in the cytoplasm of bacteria for assembly of polysaccharides slated for export to the cell surface, encodes the β3GalT activity. The result is formation of a glycan similar to the type 1 blood group H trisaccharide antigen structure (Fucα1,2Galβ1,3GlcNAcβ-), by convergent evolution with the mammalian glycan. As described below, this structure appears to influence the conformational profile of Skp1. Phylogenetic analyses are consistent with both enzyme domains descending evolutionarily from the cytoplasm of bacteria (West et al. 2004).

Similarly to the enzymes acting before it, PgtA is a cytoplasmic protein with μM apparent K_m values for Gn-Skp1 and GDP-Fuc (Trinchera and Bozzaro 1996; West et al. 1996; Wang et al. 2009). Specificity of the β3-GalT activity is likely mediated in large part by the folded Skp1 protein, as alkylated Skp1 and glycoconjugates are relatively poor substrates when compared to full length Skp1 (Wang et al. 2009). PgtA then appears to retain GGn-Skp1 to processively modify the βGal using the C-terminal αFucT domain, based on in vitro studies with UDP-[³H]Gal and GDP-[¹⁴C]Fuc (van der Wel et al. 2002a). A role for the whole Skp1 protein was reinforced by the lack of processivity for a synthetic glycopeptide substrate. In contrast to earlier acting enzyme activities, the FucT activity of PgtA depends primarily on local features of its substrate, in this case the specific disaccharide (West et al. 1996; Wang et al. 2009). This is consistent with processivity and affirms that the Galβ1–3GlcNAcα- moiety is unique to Skp1 in the cell. Interestingly, the β3-GalT appears to be strongly inhibited by concentrations of UDP-Gal above 10 μM in vitro (Wang et al. 2009).

PgtA is conserved in Toxoplasma and its parallel role in modifying Skp1 was confirmed by mass spectrometry analysis of Skp1 in a pgtA-KO strain (Rahman et al. 2016). Ability of TgPgtA to glycosylate Skp1 was shown by radioactive labeling of only exogenous Dictyostelium Gn-Skp1 added to wild-type extracts in the presence of UDP-[³H]Gal, or GDP-[³H]Fuc plus unlabeled UDP-Gal, that did not occur in pgtA-KO extracts (Rahman et al. 2016). Surprisingly, the order of the two GT domains is reversed compared to DdPgtA (Fig. 2B), but again comparisons of the two PgtA sequences has not revealed the basis for Skp1 recognition. Genomics studies predict that some Rhizarian genomes express the domains as separate proteins (Table 2) which, if correct, could explain how the domains are in a different order after refusion of domains in different organisms.

Table 2.

Homologs of Skp1-glycosylation pathway enzymes. Databases were searched using pBLAST (E≤1) with Dictyostelium and Toxoplasma sequences as bait. Protein models were edited from genomic data using protein sequence alignments as a guide. tBLASTn searches were implemented as needed. TSA and EST data were consulted where available. Sequences were evaluated for presence of key amino acid sequence motifs from master alignments, and absence of signal peptide- or signal anchor-like motifs. Instances of protein fusions are represented by dashed lines. All genes are single copy with the exception of Skp1, which in some organisms is a multigene family. y=presence of a Skp1-Pro gene; y/n=presence of second Skp1 gene lacking the Pro or predicted to be unmodified (see text). nd= not detected. See Table S1 for gene identifiers.

Kingdom	Phylum	Organism	Skp1-Pro	PhyA P4H	Gnt1 αGlcNAcT	PgtA-β3GalT	PgtA-α2FucT	AgtA α3GalT x 2	Glt1 α3GlcT	Gat1 α3GalT
Amoebozoa	cellular slime mold	Dictyostelium discoideum	y	+	+	+-------	-------+	+	nd	nd
	cellular slime mold	D. purpureum	y	+	+	+-------	-------+	+	nd	nd
	cellular slime mold	Cavenderia fasciculata	y/n	+	+	+-------	-------+	+	nd	nd
	cellular slime mold	Heterostelium album	y/n	+	+	+-------	-------+	+	nd	nd
	cellular slime mold	Acytostelium subglobosum	y/n	+	+	+-------	-------+	+	nd	nd
	cellular slime mold	Tieghemostelium lacteum	y	+	+	+-------	-------+	+	nd	nd
	acellular sl. mold	Physarum polycephalum	y	+	+	+-------	-------+	+	nd	nd
	amoebae	Acanthamoeba castellani	y/n	+	+	nd	nd	nd	nd	nd
	amoebae	Balamuthia mandrillaris	y	+	+	nd	nd	nd	nd	nd
Fungi	Chytridiomycota	Spizellomyces punctatus	y	+-------	-------+	nd	nd	nd	nd	nd
	Chytridiomycota	Rhizoclosmatium globosum	y	+-------	-------+	nd	nd	nd	nd	nd
	Chytridiomycota	Gonapodya proliferia	y	+-------	-------+	nd	nd	nd	nd	nd
Rhizaria	Foraminifera	Reticulomyxa filosa	y	+	+	+	+	nd	+-------	-------+
	Cercozoa	Spongospora subterranea	y	+	+	+	+	nd	+	+
	Cercozoa	Plasmodiophora brassicae	y	+	+	+	+	nd	+-------	-------+
	Cercozoa	Bigelowiella natans	y	+	+	+	+	nd	+-------	-------+
Alveolates	Apicomplexa	Toxoplasma gondii	y	+	+	+-------	-------+R	nd	+	+
	Apicomplexa	Hammondia hammondi	y	+	+	+-------	-------+R	nd	+	+
	Apicomplexa	Neospora caninum	y	+	+	+-------	-------+R	nd	+	+
	Apicomplexa	Sarcocystis neurona	y	+	+	+-------	-------+R	nd	+	+
	Chromerida	Vitrella brassicaformis	y	+	+	+-------	-------+R	nd	+	+
	Chromerida	Chromeria velia	y	+	+	+-------	-------+R	nd	+	+
	Ciliophora	Stylonychia lemnae	y	+-------	-------+	+-------	-------+R	nd	+	+
	Ciliophora	Oxytricha trifallax	y	+-------	-------+	+-------	-------+R	nd	+	+
	Dinoflagellate	Karlodinium micrum	y	+	+	+-------	-------+R	nd	+	+
	Dinoflagellata	Karenia brevis	y	+-------	-------+R	+-------	-------+R	nd	+	+
	Dinoflagellata	Symbiodinium microadriaticum	y	+-------	-------+	+-------	-------+R	nd	+	+
Stramenopiles	oomycete	Albugo laibachii	y	+-------	-------+	+-------	-------+R	nd	+	+
	oomycete	Pythium insidiosum	y/n	+-------	-------+	+-------	-------+R	nd	+	+
	oomycete	Aphanomyces invadans	y	+-------	-------+	+-------	-------+R	nd	nd	nd
	oomycete	Aphanomyces euteiches	y	+-------	-------+	+-------	-------+R	nd	+	+
	oomycete	Aphanomyces stellatus	y	+-------	-------+	+-------	-------+R	nd	+	+
	oomycete	Saprolegnia parasitica	y	+-------	-------+	nd	nd	nd	nd	nd
	oomycete	Phytophtora infestans	y/n	+-------	-------+	nd	nd	nd	nd	nd
	oomycete	P. ramorum	y/n	+-------	-------+	nd	nd	nd	nd	nd
	oomycete	P. sojae	y/n	+-------	-------+	nd	nd	nd	nd	nd
	diatom	Phaeodactylum tricornutum	y	+	+	nd	nd	nd	nd	nd
	diatom	Fragilariopsis cylindrus	y/n	+-------	-------+	nd	nd	nd	nd	nd
	diatoma	Thalassiosira pseudonana	y/n	+-------	-------+	+-------	-------+R	nd	+	+
	microalga	Nannochloropsis gaditana	y	+-------	-------+	+-------	-------+R	nd	+	+
	pelagophyte	Aureococcus anophagefferens	y	+-------	-------+	+-------	-------+R	nd	+	+
	brown alga	Ectocarpus siliculosis	y/n	+-------	-------+	+-------	-------+R	nd	+	+
Cryptista	Cryptophyta	Guillardia theta CCMP 2712	y	+-------	-------+	+-------	-------+R	nd	+	+
Haptista	Haptophyta	Emiliania huxleyi	y	+-------	-------+	nd	nd	nd	nd	nd
Viridiplantae	Charophyta	Klebsormidium nitens	y	+-------	-------+	nd	nd	nd	nd	nd
	Chlorophyta	Chlamydomonas reinhardtii	y	+-------	-------+	nd	nd	nd	nd	nd
	Chlorophyta	Volvox carteri	y	+-------	-------+	nd	nd	nd	nd	nd
	Chlorophyta	Coccomyxa subellipsoidea	y	+-------	-------+	nd	nd	nd	nd	nd
	Chlorophyta	Chlorella variabilis	y	+-------	-------+	nd	nd	nd	nd	nd
Excavates	Loukozoa	Malawimonas californiana	y	+	+	nd	nd	nd	nd	nd
	Percolozoa	Naegleria gruberi	y	+-------	-------+	nd	nd	nd	nd	nd
Bacteria	Cyanobacteria	Trichodesmium erythraeum	nd	+	+	nd	nd	nd	nd	nd
	Protobacteria	Desulvofibrio desulfuricans	nd	nd	nd	+-------	-------+R	nd	nd	nd
	Firmicutes	Pelosinus sp UFO1	nd	nd	nd	+-------	-------+R	nd	nd	nd

Sources: NCBI, dictyBase, VeuPathDB R=domain order reversed relative to P. ultimum R=domain order reversed relative to D. discoideum bifunctional order: Gat1-Glt1

Completing the pentasaccharide in Dictyostelium

Addition of the final two sugars (both αGal) in Dictyostelium is catalyzed by AgtA (formerly GT78) (Ercan et al. 2006; Schafer et al. 2014). This bifunctional enzyme contains a single catalytic domain and a series of C-terminal WD40 repeats expected to fold into a 7-bladed β-propeller domain. Both sugars are αGal residues, α3-linked to their underlying sugars (Sheikh et al. 2017). AgtA was identified after purification to homogeneity from cytosolic extracts of Dictyostelium based on its ability to modify Fuc terminated synthetic glycans using UDP-[³H]Gal (Ketcham et al. 2004). Sequences of tryptic peptides from the purified protein led to the identification of the encoding gene (Ercan et al. 2006). Its catalytic domain was the first experimentally documented GT in the newly formed CAZy GT77 family that initially included GTs associated with pectin synthesis in the plant secretory pathway (Ercan et al. 2006; West et al. 2010). Its role in applying the fourth sugar on Skp1 was confirmed by the finding that the trisaccharide glycoform of Skp1 accumulated when its gene was disrupted (Wang et al. 2009). Its surprising role in also catalyzing the addition of the fifth and final sugar resulted from mass spectrometry analysis of the product of its in vitro reaction with FGGn-Skp1 (Schafer et al. 2014). Interestingly, AgtA can add a single αGal residue to Fuc-terminated sugars attached to a non-protein aglycon but adds the second αGal only when the acceptor glycan is attached to Skp1. Initially, the first αGal was shown to be linked to the 3-position of Fuc based on chromatographic co-elution with standards (Ketcham et al. 2004). Other studies showed specificity for fucose that was α-linked and in the L-configuration (Schafer et al. 2014). Determination that the 5^th and final Gal is also α3-linked awaited analysis of the full length Skp1 modified by AgtA with UDP-[¹³C]Gal by NMR (Sheikh et al. 2017). This corrected the prior assignment as an α1,6-linkage that had been based on recalcitrance to an α3-galactosidase (Teng-umnuay et al. 1998), likely because it was sensitive to the nature of the underlying sugar.

A comparison of the activity of recombinant versions of AgtA isolated from either Dictyostelium or E. coli toward synthetic and natural substrates illuminated important features of the reaction (Ketcham et al. 2004; Ercan et al. 2006; Schafer et al. 2014). AgtA prefers Mn⁺⁺ compared to Mg⁺⁺ for optimal activity in vitro. FGGn-Skp1 was a superior substrate, as expected, with a low μM K_m as typical for the other GTs. However, prior denaturation of Skp1 dramatically reduced its acceptor activity. The FGGn-pNP analog was, by comparison, a very weak acceptor equivalent to Fuc-pNP alone, despite its identity with the native glycan sequence, and received only the first of the 2 Gal residues. In contrast, FG-pNP was a substantially better acceptor though still weaker than the protein bound glycan and also accepted only a single Gal. The recombinantly expressed catalytic domain alone (AgtA-CAT) showed the same pattern of reactivity with the free glycans but reacted with FGGn-DdSkp1 no better than with free FGGn-pNP (Ketcham et al. 2004; Schafer et al. 2014). Remarkably, addition of free recombinant C-terminal β-propeller domain strongly activated activity of AgtA-CAT toward the free glycans, as did addition of unmodified Skp1. These findings were supported by studies indicating stable binding of AgtA to DdSkp1 mediated by its β-propeller domain in vitro and during purification of AgtA from cell extracts (Ketcham et al. 2004). The AgtA/Skp1 interaction is relatively tight with an apparent affinity of about 60 nM in the presence of 50 mM NaCl (Schafer et al. 2014). Similar AgtA affinities were observed for the earlier stages (mono-to-trisaccharide) of DdSkp1 modification, but were 2-fold weaker upon completion of the pentasaccharide (Schafer CM, West CM, unpublished data). The low K_d and K_m of AgtA toward FGGn-DdSkp1, together with cross-linking studies, are most consistent with AgtA binding the monomeric form of DdSkp1 (Schafer et al. 2014). Together these findings support a model in which recognition of the folded region of the DdSkp1 protein by the β-propeller domain activates the catalytic domain to unfold the trisaccharide to expose the terminal fucose for modification, and to retain DdSkp1 for a second addition of αGal with the same 3-linkage to the new αGal. Unlike the other enzymes, which are estimated to be present at low levels commensurate with catalytic roles, AgtA has been estimated to be expressed equimolar with Skp1 in cells (Ercan et al. 2006). As described below, the intrinsic binding of AgtA to all isoforms of Skp1 is likely to impact Skp1 activity in the cell.

Completing the pentasaccharide in Toxoplasma

Sequence-based bioinformatics searches failed to detect a homolog of Dictyostelium AgtA in the Toxoplasma genome (Rahman et al. 2016). However, a broad search for GTs that lacked organelle targeting sequences and were phylogenetically co-distributed with TgGnt1 and TgPgtA yielded TgGlt1 and TgGat1 as candidates (Rahman et al. 2017). TgGlt1 is a CAZy family GT32 GT and disrupting its gene showed it to be required for addition of the fourth sugar based on mass spectrometry detection of the trisaccharide form of TgSkp1. Radioactive sugar incorporation using the recombinant enzyme showed the sugar to be a Glc, and the assignment to the GT32 family, whose members are typically retaining, indicated an α-linkage. NMR of the product of the in vitro reaction with FGGn-pNP showed α-Glc to be linked to the 3-position of Fuc, as occurs in DdSkp1 except that it is an αGlc rather than an αGal. Glt1 is highly specific for the UDP-Glc substrate with minimal activity for all other tested sugar donors (Rahman et al. 2017). The μM K_m of Glt1 for FGGn-TgSkp1 was over 3 orders of magnitude lower than the free trisaccharide conjugated to pNP, indicating a substantial role of the TgSkp1 protein for recognition by Glt1. Interestingly, FGGn-pNP was a better receptor than FG-pNP, opposite the preference of AgtA for the same substrates.

TgGat1 then functions to extend the resulting tetrasaccharide, as suggested by genetic deletion of gat1 resulting in accumulation of the tetrasaccharide glycoform of TgSkp1 (Mandalasi et al. 2020). Recombinant Gat1 isolated from E. coli was dependent on Mn⁺⁺ and highly active in labeling GlFGGn-TgSkp1 from UDP-[³H]Gal, confirming a direct role in vivo. The apparent K_m values of TgGat1 for UDP-Gal and GlFGGn-Skp1 of around 5 μM (Mandalasi et al. 2020) are similar to the values measured for TgGlt1 against UDP-Glc and FGGn-TgSkp1.

Phylogenetic analysis of Gat1 evolution showed it to be closely related to the catalytic domain of glycogenin (Mandalasi et al. 2020). Glycogenin serves as an initiating α4GlcT for priming glycogen synthesis in yeast and metazoa (Roach et al. 2012), but differs from Gat1 by its possession of a C-terminal glycogen synthase binding domain (Zeqiraj et al. 2014). Like glycogenin, Gat1 modifies αGlc but in a 3- rather than 4-linkage. Interestingly, Gat1 has weak GlcT activity and glycogenin has weak GalT activity (Bilyard et al. 2018), and both prefer Mn⁺⁺ over Mg⁺⁺, consistent with their close evolutionary relationship. Gat1 strongly prefers GlGFGGn-Skp1 over synthetic Glc acceptors and, while it can modify Glcα1,4Glcα1-pNP, longer homooligosaccharides are poorer substates consistent with no apparent effect on starch synthesis in parasites. An ortholog of TgGat1 is expressed in Pythium ultimum, which is in the stramenopile clade of the larger SAR group in which Toxoplasma exists as an apicomplexan (Fig. 3A). The substrate activity profiles of its recombinantly expressed version are similar to TgGat1 including absence of the self-modification that occurs in glycogenin (Mandalasi et al. 2020). Both Gat1 proteins form high affinity homodimers with submicromolar K_d values like glycogenin and with an identical dimer interface as determined by X-ray crystallography, even though they do not self trans-glycosylate between subunits as suggested for glycogenin (Roach et al. 2012). Modeling the active site confirms a preference for the TgSkp1 tetrasaccharide, and AlphaFold-3 modeling suggests a mechanism for TgSkp1 binding compatible with enzymatic processing (Cantrell DA, West CM, unpublished data). Furthermore, Gat1 is found in the Skp1 interactome of Toxoplasma as assessed by co-immunoprecipitation, regardless of Skp1’s modification status. This was not observed for the other enzymes except for PhyA (Mandalasi et al. 2024). These observations invite speculation that, like the terminal GT of the Dictyostelium pathway, Gat1 has a second non-enzymatic role that influences TgSkp1.

Figure 3.

A) Evolutionary tree of eukaryotic life. Locations of Dictyostelium, Toxoplasma, and Pythium are indicated. Adapted from Al Jewari and Baldauf (2023). See Table 2 for list of species and predicted modification genes. Four species where potential Skp1 modification genes occur, Naegleria gruberi, Malawimonas californiana, Guillardia theta (Cryptophyte), Emiliania huxleyi (Haptophyte), lacked sufficient genomic information to be confidently placed. B) List of phylogenetic groups harboring modification genes.

Summary of the glycosylation pathway

Together the findings document the occurrence of 5 distinct GT activities in the cytoplasm of Dictyostelium, 3 of which are shared in Toxoplasma with the other 2 having different evolutionary origins. Bioinformatics studies indicate a broad yet selective evolutionary distribution across a variety of protists, algae and fungi, but not higher plants and metazoa (Fig. 3, Table 2). PhyA and all GTs except for the α2FucT are evolutionarily related to homologs in the secretory pathway but lack sequence motifs for targeting to the rER. While there is no biochemical evidence for co-associating as complexes, they appear to act efficiently as substantial accumulation of glycoform intermediates has not been observed using glycoform-specific mAbs (Fig. 1B) and mass spectrometry. This suggests that once the attachment Pro is hydroxylated, glycosylation occurs rapidly to generate the full glycan. The order of addition is controlled by the intrinsic specificity of the GTs, without the benefit of a spatial gradient of GTs as occurs in the Golgi stack. This is likely expedited by the frequent fusions of β3GalT and α2FucT in PgtA and the dual activities of AgtA, and other instances of gene fusions between PhyA and Gnt1 or Glt1 and Gat1 emphasize this possibility (Fig. 2C, Table 2).

The substitution of AgtA for Glt1 and Gat1 to complete the glycan is a remarkable example of convergent evolution that may be driven by a beneficial attribute of the full pentasaccharide, as suggested by structural studies described below. The only difference between the two glycans is the conservative replacement of an α3-linked Glc by an α3-linked Gal. An interesting hypothesis is that by this transition, Gat1 was liberated to evolve into glycogenin to perform a distinct role to facilitate glycogen synthesis in yeast and animals, also in the cytoplasm (Mandalasi et al. 2020). At present, Glt1- and Gat1-like sequences are found in the Diaphoretickes and AgtA-like sequences in the separate Amorpheae branch (Fig. 3, Table 2). While it is unclear which came first, we theorize that Glt1 and Gat1 are ancestral owing to presence in currently unassigned and potentially basal groups and the hypothesized repurposing of Gat1 as glycogenin. Interestingly, the genomes of other unicellular organisms including Acanthamoeba castellanii, Chlamydomonas reinhardtii and Naegleria guberi possess likely orthologs of PhyA and Gnt1, but lack evidence for PgtA, Glt1, Gat1, and AgtA (Table 2). These organisms are of high interest because of the potential to reveal the significance of a single GlcNAc, which could be integrated into a model for the stepwise evolution of the pathway. Alternatively, these organisms might be a source of discovery of new GTs should their Skp1 proteins possess more elaborate glycans.

An unusual feature of the Skp1 glycosylation pathway is its apparent dedication to the glycosylation of a single protein in both Dictyostelium and Toxoplasma. Confirmation of this conclusion is essential to ensure that the phenotypic effects of the modification gene knockouts can be ascribed to Skp1 activities. Dependence of PhyA and the GT enzymes on structural features of Skp1 far from the site of glycosylation suggest occurrence of second site interactions beyond the active site that likely contribute to this specificity (Teng-umnuay et al. 1998; Sassi et al. 2001; van der Wel et al. 2011). Failure to detect alternative substrates in GT-KO extracts using biochemical complementation with recombinant GTs and highly sensitive radioactive sugars adds further support for GT specificity. The PhyA and GT K_m values for Skp1 in the sub-μM to low μM range are consistent with estimated concentrations of Skp1 in cells. The fusion of PhyA and Gnt1 into a single protein in Pythium (van der Wel et al. 2019) and in genomes of other protists (Table 2) further indicates that PhyA is specific for Skp1, owing to the more substantial biochemical evidence that Gnt1 is dedicated to Skp1. In turn, the specificity of PhyA for Skp1 is reinforced by related though not identical biochemical and physiological effects of disrupting phyA and gnt1 in both organisms. The phenotypic differences that do occur might be due to distinct structural effects of hydroxylation and subsequent αGlcNAcylation (see below). Using developmental assays for the role of PhyA in Dictyostelium O₂-sensing, genetic manipulation of Skp1 levels in cells reveals an opposing effect on PhyA (Wang et al. 2011; Xu et al. 2012b), as if PhyA suppresses hyperactivity of Skp1. Another unusual feature is that the terminal GTs in both Dictyostelium and Toxoplasma exhibit unexpected affinity toward Skp1 regardless of its modification status. Evidence from Dictyostelium suggests the AgtA modulates initial modification of Skp1 and its availability to associate with FBPs and other proteins (Schafer et al. 2014). Together, these biochemical and phenotypic findings strongly support the interpretation that the modification enzymes are dedicated to Skp1 in the cell.

It is likely that factors beyond Skp1 and the enzymes themselves influence the rates of its modifications in cells. The significance of O₂ and metabolites related to αKG as rate limiting factors for PhyA activity is summarized below. Though accumulation of glycosylation intermediates has not been noted during normal growth and development of either Dictyostelium or Toxoplasma, regulation by availability of sugar nucleotides under stress conditions is an unexplored area. An involvement of chaperones or other factors is suggested by the weak substrate activity of ectopically expressed Skp1 point mutants that are otherwise efficient substrates in vitro. In contrast, a negative effect is suggested by in vitro evidence that complexes with FBPs interfere with modifications (van der Wel et al. 2011). Finally, there is as yet no evidence for glycosidases that reverse the glycan modifications, though this possibility has not been comprehensively explored.

Effects of Modifications on Skp1 Structure

Determining how glycosylation affects Skp1 structure will be important for understanding how it affects its interactome and ultimately SCF function and O₂-sensing. Skp1 is highly conserved throughout eukaryotes, with 160-180-amino acids that includes an N-terminal BTB domain and a C-terminal region (CTR) that folds into a canonical structure when complexed with FBPs as assessed by X-ray crystallography and cryo-EM (Fig. 4B) (Schulman et al. 2000; Horn-Ghetko et al. 2021; Hopf et al. 2022). The interface with the F-box domain consists of a core interaction involving α-helices-5, -6 and -7 (Subsite 1) and a variable interface involving α-helix-8 (Subsite 2) (Fig. 1C). A comparison of select sequences from different organisms is shown in Table 1C. Subsite-2 is of particular interest as prolyl hydroxylation/glycosylation occurs at the start of helix-8. Skp1 maintains a similar bound conformation to FBPs across phylogeny (e.g., Sheard et al. 2010; Kumanomidou et al. 2015; Wong et al. 2017; Gorelik et al. 2018; Kuchay et al. 2019), except that Subsite 2 takes on a strikingly different conformation when bound to Fbs1 (Mizushima et al. 2007) which does not appear to be a crystal packing artifact. Conformational flexibility and backbone contacts likely underlie the ability of Skp1 to bind the many various and often degenerate F-box domain sequences (Chandra Dantu et al. 2016; Bhattacharya et al. 2022).

Figure 4.

Schematic diagrams depicting proposed regulation of TgSkp1 by glycosylation. A) The disorder to order transition, which flows to two independent modes of regulation. B) Alteration of direct contacts with the FBPs (conformation ensemble). C) Disruption of the fuzzy interface leading to a weaker homodimer. D) A proposed FBP affinity hierarchy showing the predicted effect TgSkp1 glycosylation on selective FBP binding. Different potential Skp1/FBP complexes (varied in color) are vertically stacked according to their hypothetical dissociation constants, as represented by the central scale. On the left, lower affinity FBP-3 and FBP-4 will have less opportunity to compete with high affinity TgSkp1 homodimerization, whereas relaxation of homodimer affinity by glycosylation will enable greater access of these FBPs to the SCF complexes. Structures are derived from PDB: 6V88 for the Skp1 homodimer and from PDB: 2P1O for a representative Skp1/FBP complex.

In comparison, free Skp1 does not crystallize even after deletion of two internal, poorly conserved disordered regions. Small angle X-ray scattering and gel filtration analyses initially indicated that DdSkp1 is, like mammalian Skp1, a homodimer at concentrations required for structural studies (Sheikh et al. 2014). NMR analysis of a perdeuterated form Skp1 showed that the C-terminal ~40 aa were relatively disordered (Xu et al. 2018). This was consistent with the significant entropic penalty described upon binding the soluble mammalian FBP Fbs1 (Tan et al. 2008), which likely originates from ordering Skp1’s C-terminal region. The solution structure of the dimer was solved on a ¹H/¹⁵N/¹³C-labeled form deleted of the C-terminal region (CTR) and an internal disordered loop (Fig. 1C) to achieve a smaller size suitable for NMR analysis. The truncated dimer exhibits the same BTB domain fold as seen in complexes with FBPs, but not a canonical BTB/BTB interface (Kim et al. 2020). Referred to as Subsite-A (Fig. 4C), the homodimer interface overlaps and is therefore competitive with part of Subsite-1 of the F-box interface (Fig. 4B). Subsequent studies on TgSkp1 using analytical ultracentrifugation showed that the partially disordered CTRs contribute a charge-based fuzzy self-interaction to the homodimer, labeled as Subsite-B (Cantrell et al. 2025). This improves homodimer affinity by over an order of magnitude to an estimated K_d of 50 nM for TgSkp1, substantially stronger than for DdSkp1 (Cantrell DA, West CM, unpublished data). The semiparallel orientation corrects the originally proposed antiparallel orientation for DdSkp1 (Sheikh et al. 2014) and allows for a direct interaction between the CTRs. The role of the CTR is salt-sensitive and computational analyses indicate that the CTR/CTR contribution, or Subsite-B, is analogous to other described charge-based fuzzy interactions (Williamson 2023) but is novel because it occurs between identical sequences. Glycosylation of TgSkp1, which almost certainly homodimerizes like DdSkp1, weakens the homodimer by over an order of magnitude similar to removing the entire CTR (Cantrell et al. 2025). As described below, the glycan is structurally ordered in a way that is predicted to sterically interfere with the dynamic fuzzy interaction. Thus Skp1’s CTR has dual roles in facilitating both homodimerization and heterodimerization with FBPs, using independent mechanisms.

Human Skp1 was the first to be reported to form a homodimer, with a K_d of 1-2 μM (Henzl et al. 1998). This value is in the range estimated for Skp1 in cultured cells (Gonzalez-Yanes et al. 1992; Reitsma et al. 2017), though Skp1 concentrations are much higher in specialized cells (Henzl et al. 1998). Dictyostelium Skp1 also homodimerizes with a similar K_d (Kim et al. 2020). In comparison, the ~50 nM K_d of Toxoplasma Skp1 homodimerization may be competitive with the only known K_d of a Skp1/FBP interaction: HsSkp1 with guinea pig Fbs1, 25 μM (Tan et al. 2008). The dynamic nature of Skp1’s fuzzy interface (Subsite-B) may facilitate exchange from the homodimer (Sottini et al. 2020), while the well-ordered nature of the Skp1/FBP interface (Schulman et al. 2000) will minimize exchange in the opposite direction. Subsite-B’s dynamic nature may also facilitate hydroxylation and glycosylation in the dimer state, though the BTB domain is required for recognition by most of the enzymes. These roles, in addition to acceptance of post-translational modifications, presumably impose unique constraints on the sequence of the CTR, which is highly conserved throughout eukaryotes (Table 1) (e.g., Baptista et al. 2019, van der Wel et al. 2019). Thus, point mutations of Skp1 that have previously been inferred to selectively affect FBP or Cul1 interactions (Bai et al. 1996; Connelly and Hieter 1996; Lehmann et al. 2004; Brace et al. 2006; Li et al. 2023) may also influence competing homodimerization.

The impact of the glycan was assessed in more detail by NMR experiments using fully glycosylated Dictyostelium GGFGGn-Skp1 labeled with [U-¹³C]-sugars (Sheikh et al. 2017). The first and fifth (last) sugars of the glycan exhibited rapid but correlated motions relative to the BTB domain, suggesting that their motions are due to the disordered CTR where the glycan is attached. Detected NOEs and inter-sugar interactions suggested by energy minimization and molecular dynamics (MD) methods led to the model that the pentasaccharide has a relatively ordered structure (Sheikh et al. 2017). This is consistent in part with prior NMR studies on the organization of the reducing end blood group H type 1 trisaccharide (Martin-Pastor and Bush, 2000; Almond et al. 2004). NMR studies of the perdeuterated DdSkp1 dimer showed changes in Skp1’s peptide ¹H:¹⁵N cross peaks for Gn-Skp1 and further changes for GGFGGn-Skp1 relative to unmodified Skp1, indicating significant reorganization of Skp1’s peptide backbone (Xu et al. 2018). One consequence of αGlcNAcylation was partial induction of α-helix-8 as seen in FBP complexes, which is apparently predisposed to fold in this way owing to its E/R/K composition (Swanson and Sivaramakrishnan 2014). The target Pro also favors induction of the future helix-8 (Serrano and Fersht 1989), and its modification may further satisfy hydrogen bonds to promote downstream helicity (Elbaum and Zondlo 2014). 4-(trans)-prolyl hydroxylation and αGlcNAcylation strongly favor the trans isomer of the Thr-Pro bond in model dipeptide studies analyzed by NMR (Karunaratne et al. 2014). This is the isomer observed in structures of unmodified Skp1 in Skp1/FBP complexes, but the preference might influence folding of this region on a longer time scale not currently accessible by MD studies. Proline puckering also changes depending on hydroxylation or glycosylation state, with the Cγ-exo conformation being favored upon addition of the GlcNAc (Karunaratne et al. 2014).

Overall, glycosylation promotes a disorder to order transition that is thought to inhibit homodimerization and favor conformations compatible with FBP binding (Fig. 4A). In support of this interpretation, MD modeling suggests how the relatively rigid glycan organizes the upstream sequences that form helix-7 and the intervening loop between helices-7 and -8 in FBP complexes (Sheikh et al. 2017). Interactions here are stabilized by hydrogen bonds to helix-7 and loop side chains and backbone atoms of DdSkp1, including the side chains of highly conserved N139, E129, and N136 (lower frequency) (Fig. 1D, Table 1C). MD simulations of the fully glycosylated and unmodified versions of DdSkp1 showed that the presence of the glycan increased the average width of the pocket that docks the F-box domain of the FBP, potentially facilitating the interaction (Sheikh et al. 2017). Analysis of the TgSkp1 glycan, which differs by substitution of the first αGal (fourth sugar) with αGlc, yielded a similar interpretation that also highlighted hydrogen bonding with N147 and N150 (homologous to N136 and N139 in DdSkp1), though specific hydrogen bonds with helix-7 were of lower frequency (Fig. 1E) (Mandalasi et al. 2020). Significance of these hydrogen bonds is suggested by amino acid substitutions at the equivalent positions of N136 and N139 in duplicate Skp1 isoforms that are posited to not be modifiable (see below). Furthermore, the modeling of TgSkp1 emphasized packing of the fifth sugar (αGal) in the turn from helix-7 to the loop, which could only be reached by a pentasaccharide. In comparing DdSkp1 and TgSkp1, the 4-OH of the fourth sugar of each is projected toward the solvent, with minimal consequence on the interface with the polypeptide. These interpretations correlate with other examples where glycosylation influences a protein’s folding or conformational ensemble (e.g., Hanson et al. 2009; Jayaprakash and Surolia 2017; Woods 2018), though an effect of such a long glycan is unusual (West and Kim, 2019).

The location of the glycan in the CTR is consistent with a direct role in increased FBP binding and absence of effects on Cul1 binding. Related but lesser effects of partial glycosylation are also consistent. Understanding the structural basis for the partial effects, which were likely important for the stepwise evolution of the pathway, deserves further investigation.

Skp1 Interactions

To begin to assess the biochemical significance of Skp1 modification in cells, DdSkp1 complexes from different modification mutants were compared by size exclusion chromatography of cytosolic fractions from Dictyostelium (Sheikh et al. 2015). DdSkp1 from wild-type cells at ambient O₂, where it is mostly glycosylated, eluted primarily as a single peak. In contrast, unglycosylated DdSkp1 from phyA-KO cells eluted primarily at a later position, corresponding to a minor peak for fully glycosylated Skp1. DdSkp1 from the GT mutants eluted as a combination of the two peaks whose relative heights transitioned according to the position of GT action in the pathway. Thus, each step of the pathway appeared to contribute to altered complex formation. The composition of these complexes was examined by proteomics analyses of anti-Skp1 and anti-Cul1 pull-downs from the wild-type and mutant strains (Sheikh et al. 2015; Boland et al. 2022). Distinct profiles of binding partners provided direct evidence that glycosylation impacts Skp1 function. The abundances of CulE (Cul1 ortholog) and Rbx1, and the fraction of neddylated CulE, were unaffected suggesting that global SCF activity might not be affected. Complete glycosylation increased the amount of about half of the putative FBPs in the Dictyostelium Skp1 interactome, while not detectably reducing the amount of any FBPs. For example, the abundant FBP FbxwD was enriched 5-fold in the interactome of fully glycosylated Skp1 relative to unmodified Skp1, without affecting the level of FbxwD in the cell. An intermediate enrichment was observed for Skp1 in a mutant expressing only the trisaccharide form, and these effects were confirmed in reciprocal pull-downs. Similarly, the FBP JcdI also had increased abundance in the Skp1 interactome of wild-type cells, but the effect was obscured by reduced levels of JcdI in the cell. Notably, the effect on total cellular level of JcdI was only observed in cells at the slug stage and not in growing cells (Boland et al. 2022), indicating developmental regulation. The effect of the glycan on Skp1 binding to FBPs might be direct. This hypothesis is based on Dictyostelium Gn-Skp1 and GGFGGn-Skp1 exhibiting modestly preferential binding relative to unmodified Skp1 to the uniquely soluble mammalian Fbs1 (Sheikh et al. 2014), and effects on Skp1 structure (reviewed below). However, a caveat was Skp1 is not modified in mammals so its interaction with Fbs1 is not naturally influenced by Skp1 hydroxylation/glycosylation. While these findings do not fully explain the different complex sizes from gel filtration, increased association of proteasomes and the COP9 signalosome with CulE when Skp1 is glycosylated may also contribute (Sheikh et al. 2015).

Native Skp1 with its full glycan is normally enriched in Dictyostelium nuclei relative to the cytoplasm based on immunofluorescence localization using anti-Skp1 Abs or detection of ectopic epitope-tagged Skp1 (Sassi et al. 2001), similar to mammalian Skp1. Preventing glycosylation either by inhibiting prolyl hydroxylation or mutating the Pro to another amino acid slightly reduces nuclear enrichment, while presence of the disaccharide is sufficient to restore normal localization. The molecular interactions that are involved are unknown, nor is how compartmentalization may affect the Skp1 interactome.

The interactome of Skp1 in Toxoplasma is similarly influenced by its glycosylation. As found in Dictyostelium, about half of the 14 detected FBPs were enriched in Skp1 pulldowns from wild-type compared to phyA-KO cells (Baptista et al. 2019; Mandalasi et al. 2024). Cul1 and Rbx1 levels were not affected as observed for DdSkp1. Closer examination of three FBPs, after their genomic loci were epitope tagged, showed that Fbxo13 (homologous to DdJcdI) and Fbxo14 were enriched because their total amounts were reduced in phyA-KO parasites. In contrast, the level of the non-enriched FBP FbxO1 was unaffected. Significantly, treatment with proteasome inhibitors selectively increased the abundance of FbxO13 and FbxO14 in phyA-KO but not wild-type cells. Similar effects were observed for gnt1-KO parasites, demonstrating that glycosylation and not hydroxylation per se was responsible. It is currently unclear whether lower total protein levels were due to autoubiquitination in an SCF complex or greater susceptibility to other Ub ligases when not associated with Skp1. Interestingly, the comparison between species revealed opposite effects on abundance of the homologs DdJcdI and TgFbxo13. This indicates that other factors interact with the PhyA-dependence of FBP fate, as also inferred from the developmental dependence of DdJcdI levels (Boland et al. 2022).

Given that FBPs are generally insoluble in vitro unless complexed with Skp1 (van der Wel et al. 2011; Boland et al. 2022; Deng B, West CM, unpublished observations), non-Skp1 bound FbxwD and JcdI are expected to associate with chaperones (Melville et al. 2003; Wang et al. 2016) or other factors (Nelson and Laman 2011; Yu et al. 2015; Reiterer et al. 2017). However, clear candidates did not emerge from the co-pulldown approaches (Sheikh et al. 2015; Mandalasi et al. 2024). Sgt1, an Hsp90 co-chaperone found throughout eukaryotes, associates with Skp1 in a manner that is compatible with binding an FBP but not Cul1 (Willhoft et al. 2017), suggesting it might be an FBP/Skp1 assembly factor. However it has not been detected in Skp1 co-IPs. Given evidence that Skp1 has a long half-life (Wang et al. 2011; Zhang et al. 2012), an exchange mechanism for FBPs is plausible. Here we propose that exchange may occur at the FBP/Skp1 interface, as well as exchange at the Skp1/Cul1 interface that is influenced by Cand1.

There are many examples of glycans serving as ligands for specific carbohydrate binding domains (Taylor et al 2022), and the organized structure of the Skp1 glycan invites speculation for the existence of such factors in cells. Comparison of the interactomes of Skp1 from wild-type and Skp1 glycosylation mutant Dictyostelium cells identified proteins enriched in the wild-type sample that did not appear to be FBPs or other SCF-related proteins (Boland et al. 2022). Though none had sequences that could be related to known carbohydrate binding domains, it is impossible to exclude the existence of novel carbohydrate binding proteins. In addition, none had sequences that could be related to proteins that were differentially enriched in the Skp1 interactomes from wild-type and glycosylation mutant strains of Toxoplasma (Mandalasi et al. 2024), though their different glycan sequences might involve different carbohydrate binding proteins. Furthermore, hypothetical carbohydrate binding proteins might have been lost during capture of Skp1 for interactome studies. Therefore, the existence of additional factors that affect Skp1 by serving as glycan readers remain a possibility.

Implications and Generality of Regulation of Skp1/FBP Interactions

The interaction between Skp1 and its various FBP binding partners is a little recognized point of regulation, but there are precedents. The heavy metal stress response of the budding yeast Saccharomyces cerevisiae specifically targets this interaction (Hwang et al. 2025). In the presence of excess Cd²⁺ or other metals, the FBP Met30 dissociates from Skp1 and is ubiquitylated via interactions with a specific chaperone for unbound Met30. The mechanism involves metal binding to the F-box domain and action of the p97/cdc48 AAA-ATPase. Furthermore, phosphorylation of a Ser residue at the start of future helix-7 of Skp1 (Fig. 1C) has also been implicated in controlling interaction with Met30 in cells, using a phosphomimetic amino acid substitution (Beltrao et al. 2012). The same substitution, however, reportedly did not affect SCF assembly in human cells, but did promote Skp1’s interaction with V₁ subunits of the vacuolar ATPase (Li et al. 2023). Phosphorylation was also noted when Skp1 was expressed in insect cells (Kaplan et al. 1997), but was not detected on DdSkp1 based on 2D gel analysis after metabolic labeling with ³²PO₄ (West et al. 1997) or proteomic analyses of DdSkp1 isolated from cells. A recent alanine scanning mutagenesis study in human cells revealed that point mutations in the loop between future α-helices-7 and -8 exert selective effects on binding to different FBPs (Li et al. 2023). In plants, a highly conserved Cys residue in α-helix-6 was susceptible to in vitro S-nitrosylation and evidence suggested that this promoted assembly with two FBPs in vivo (Iglesias et al. 2018). These findings emphasize the importance of local sequence and structure on differential binding to FBPs, in a region whose organization is also affected by glycosylation (Fig. 1D, E). In another example, the interaction of the mammalian FbxL3 with Skp1 appears to require binding to its circadian rhythm protein Cry1, an interaction that is inhibited by an unknown factor (Yumimoto et al. 2013). Significantly, the F-box/Skp1 interface has been the target of small molecule inhibitors (Aghajan et al. 2010; Chan et al. 2013; Liu et al. 2015) and Ub variants (Gorelik et al. 2018) that are FBP-specific, further highlighting the potential for selectivity of this interface among FBPs. The point mutations and prolyl hydroxylation, glycosylation and phosphorylation modifications described above occur in the CTR. Evidence indicates that these modifications promote a disorder to order transition which would therefore selectively affect FBP and self-interactions over the other interactions (Fig. 1C), because interactions with Cul1, Syp proteins, Sgt1 and Cep3 are primarily mediated by other regions of the BTB domain.

The high affinity of the TgSkp1 homodimer, with an estimated K_d of 50 nM (Cantrell et al. 2025), suggests a mechanism by which PhyA and the GTs could tune the range of interacting FBPs. The availability of only a single Skp1 to interact with the dozens of FBPs that each have a distinct F-box sequence strongly implies the existence of an FBP affinity hierarchy (Fig. 4D). This is, however, challenging to demonstrate directly owing to the intrinsic insolubility of FBPs in the absence of Skp1. Higher affinity FBPs can be expected to be competitive with Skp1 homodimerization but, when the homodimer is weakened by glycosylation, weaker affinity FBPs should have access to Skp1 as well. Other non-SCF proteins may enter into this competition, as suggested by the example of AgtA in Dictyostelium, and CRM1, STYX and chaperones in mammals (Nelson and Laman 2011; Yu et al. 2015; Reiterer et al. 2017) and plants (Wang et al. 2016). DdSkp1 also homodimerizes, but its substantially weaker affinity of 2.4 μM (Kim et al. 2020) appears less likely to influence the FBP binding hierarchy unless there exist FBPs that bind Skp1 with lower affinity. The effects of Skp1 glycosylation in protists is contributing to a growing awareness of the significance of homodimerization for Skp1 activity in all cells (Kaiser and Flick. 2025).

Many organisms, including humans, yeast, and protists including Toxoplasma, have only a single Skp1 protein. Thus a single Skp1 must bind up to 100s of FBPs through their highly varied F-box domains that are expected to exhibit different innate affinities. Given that Skp1 availability is thought to be limiting in cells, this situation makes it likely that variations in Skp1 abundance and conformation can also control the pool of Skp1/FBP subcomplexes in cells. Indeed, there are examples where Skp1 availability controls FBP levels in yeast (William et al. 2024), and cell differentiation and cancer in animals (Liu et al. 2015; Lu et al. 2025). Overexpression of Skp1 in Dictyostelium inhibits certain developmental processes (Wang et al. 2011), and amplification of identical Skp1 genes has been observed in other social amoebae. These observations suggest that availability of Skp1 may be an important aspect of SCF regulation in addition to post-translational modifications. Nevertheless, under standard conditions of Dictyostelium growth and development (Wang et al. 2011), and comparisons of intracellular and lysed out Toxoplasma (Mandalasi et al. 2024), Skp1 appears to be constitutively expressed at the protein level.

In addition to post-translational modifications, Skp1 is also diversified in some protists by virtue of multiple coding genes. Various protists in the stramenopile, cryptomonad (possibly related to algae), heteroloboseae (Discoba), and amoebozoan lineages (see Fig. 3 for phylogeny) have a second Skp1 gene (Table 2, Skp1-Pro column) whose sequence is highly similar to the primary Skp1. Within the stramenopile lineage, a group of oomycetes replaced the target Pro with a Glu residue, rendering them inert to the modification. If different Skp1s preferentially associate with different FBPs, as has been shown for animals and plants that have multiple Skp1s (but not PhyA) (e.g., Risseeuw et al. 2003), then glycoregulation will be partitioned. In the related oomycete Pythium ultimum, the second Skp1 (PuSkp1B) retained the target Pro but is nevertheless not modified by its own PhyA in vitro, in contrast to its primary PuSkp1A (van der Wel et al. 2019). Interestingly, PuSkp1B shares characteristic nearby sequence variations with the other oomycete Skp1B variants, which are non-modifiable owing to replacement of their target Pro and an example of which is shown in Table 1C for Pythium ramorum. In addition, several cellular slime molds also possess a second Skp1, an example of which is Cavenderia fasciculata (a.k.a. D. fasciculatum), that also shares these key point substitutions, including at the equivalent of N136 and N139 in D. discoideum (Table 1C). Notably, other differences between the second and first Skp1s of oomycetes and amoebozoa are not conserved. This is potentially significant because side chains of these two residues are primary hydrogen bonding partners for interaction with the glycan (Figs. 1D,E) as discussed above. These correlations suggest that if the second isoforms are not modified, as implied by the example of PuSkp1B, these amino acids may not then have been evolutionarily constrained to retain their identity thus reinforcing the structural model. In contrast, the D. discoideum also has a second Skp1 gene, but its sequence differs at only a single unconserved position (West et al. 1997), and no modification (Sassi et al. 2001) or functional differences were detected (Wang et al. 2011). They are similarly expressed at the mRNA and protein levels throughout the life cycle (Wang et al. 2011; West et al. 2007; Sassi et al. 2001; West et al. 1997), suggesting the duplication may occur for a dosage effect. Parsimony suggests that unmodified Skp1 is the basal state and the modifications modulate its activity, and unmodifiable paralogs evolved independently in protists to escape regulation.

Skp1 has been found in contexts other than the SCF in other organisms. These include with Syp proteins in a synaptonemal complex in C. elegans (Blundon et al. 2024), Cep3 in the CBF3 yeast kinetochore complex (Leber et al. 2018), the essential cochaperone Sgt1 involved in kinetochore assembly in yeast (Willhoft et al. 2017), Siah-interacting protein (SIP) (Bhattacharya et al. 2005; Santelli et al. 2005), the Snrk protein kinase in plants (Farrás et al. 2001), the vacuolar ATPase in yeast (Seol et al. 2001), and recently with the V₁-subunit of the vacuolar ATPase and Sec22B in human cells associated with autophagy and secretion (Li et al. 2023). If these interactions beyond Skp1’s canonical role linking FBPs to Cul1 occur in protists, they represent future roles to explore for its modifications.

Roles of Skp1 Modification in Protist O₂-Sensing

The evolutionary relationship of PhyA to animal PHD2 was the initial motivation to explore its role in O₂-dependent behaviors in Dictyostelium and Toxoplasma (West and Blader 2015). Early findings of similar outcomes from disrupting phyA and exposure to low O₂ for a specific developmental transition referred to as culmination in Dictyostelium suggested PhyA is a mediator of O₂-sensing in cells (West et al. 2007). It is beyond the scope of this review, but extensive biochemical and genetic data indicate that the modification enzymes (West et al. 2010; Zhang et al. 2012; West and Blader 2015), Skp1 itself (Wang et al. 2011; Xu et al. 2012b), select FBPs and proteasomes (Boland et al. 2022) contribute to this O₂-dependent transition. Similarly, substantial evidence indicates similar roles in support of O₂-dependent homing and virulence of Toxoplasma (Rahman et al. 2016; Rahman et al. 2017; Mandalasi et al. 2020; Mandalasi et al. 2024; Cordonnier et al. 2024). Other effects of GT-KOs on amoeba size and texture (van der Wel et al. 2001), or Dictyostelium morphogenesis (Zhang et al. 2012), suggest the existence of additional but more subtle O₂-dependent roles.

These findings predict that the degree of Skp1 modification correlates with increased O₂. This has been assessed by enzyme probing for substrate levels in extracts (Wang et al. 2011), using glycoform specific Abs (Wang et al. 2011; Xu et al. 2012b), and quantitative mass spectrometry (Mandalasi et al. 2024). Indeed, pulse labeling studies show that modification lags behind protein synthesis at lower O₂ levels indicating that it is posttranslational (Sheikh et al. 2015), and glycosylation of epitope-tagged DdSkp1 that is ectopically expressed in slug cells is slow to be glycosylated (Sassi et al., 2001). In addition, exposure of both Dictyostelium and Toxoplasma cells to low O₂ modestly raises the steady-state fraction of unmodified Skp1 (Wang et al. 2011; Xu et al. 2012b; Baptista et al. 2019; Mandalasi et al. 2024). Nevertheless, the great majority of Skp1 is normally glycosylated in Dictyostelium and Toxoplasma cells maintained at ambient O₂, and glycoform specific Abs have not revealed accumulation of glycosylation intermediates during Dictyostelium development. Since the fraction of the total Skp1 pool that is unmodified does not strongly correlate with O₂-dependent phenotypes, kinetically transient subpools of Skp1 are likely to be involved. The accumulating evidence indicates that glycoregulation of E3(SCF) Ub ligases represents a novel O₂-sensing mechanism across the algal, protist and fungal kingdoms.

Summary

The Skp1 modification related genes are found across unicellular eukaryotes from algae through a range of protists to certain fungi (Fig. 3A, Table 2) and, notably, the GT candidate genes are always associated with a PhyA-like gene. Toxoplasma, Dictyostelium and Pythium, protists for which enzyme activities have been documented, represent divergent clades of the eukaryotic tree of life. Current genomic information suggests the genes are absent from strictly anaerobic protists and obligate multicellular eukaryotes, including vascular plants and animals. In the case of animals, the first enzyme PhyA is thought to have evolved into PHD2 to modify HIFα. The effect of protist PhyA is nevertheless functionally comparable to animal PHD2, in the sense that both mediate reduced levels of specific proteins either through increased degradation or decreased transcription, respectively. Across evolution, the GT candidate genes are always associated with a PhyA-like gene. In protists, prolyl hydroxylation of Skp1 is likely to be the proximal O₂-sensing event, with glycosylation mediating the major biochemical outcomes. The confinement of Skp1 glycosylation to organisms with unicellular lifestyles correlates with apparent absence of a mechanism to reverse the modification. Given Skp1’s long half-life, reversal may simply be mediated by dilution via proliferation. Thus newly synthesized Skp1 may be most functional in response to O₂ and metabolic regulation for co-expressed FBPs.

The requirement for 6 enzymes to complete the modification of Skp1 in Dictyostelium and Toxoplasma raises questions about how such a complex pathway evolved, given that each enzyme appears to be specific for Skp1. Unique properties for each processing intermediate have been described, which might have supported their sequential selection during evolution. Nevertheless, there is no evidence thus far for accumulation of processing intermediates and, indeed, pairwise fusions of the processing enzymes in different protists support efficient completion once glycosylation is primed by prolyl hydroxylation. Thus Skp1s from protists whose genomes lack detectable homologs of PgtA, Glt1, Gat1 and AgtA, are of great interest. This makes them candidates for having a minimal glycan modification, or as sources of new GTs or other modification enzymes.

Altogether, the data suggest that optimal O₂ and metabolic conditions enhance the activity of select E3 SCF Ub ligases potentially to degrade target substrates. This occurs first by the glycan formed on the resulting Hyp inducing dissociation of the Skp1 dimer, making the Skp1 monomer more available for interacting with FBPs. Secondly this occurs by an effect on the ensemble of Skp1 conformations making Skp1 more available to FBP binding. There is as yet no evidence that the glycan acts by recruiting an independent reader. These effects are selective perhaps according to relative affinities of different FBPs. However, the model is provisional until target substrates and additional FBPs are identified and examined, and further structural data on glycosylated vs. unmodified Skp1 in complexes with FBPs are achieved. Obstacles to overcome will include i) the inherent insolubility of FBPs in the absence of Skp1, ii) their minimal conservation across different organisms making identification by homology challenging, and iii) the multiple roles of the Skp1 CTR in mediating homodimerization and FBP binding which confounds selective mutagenesis to test in cells. In addition, while a role in proteasomal degradation is strongly indicated, additional participation of Skp1 in non-SCF functions cannot be excluded. Overall, the findings highlight the value of the Dictyostelium model organism for originally exposing the basic tenets of this new mechanism, and the power of genomic comparisons to apply this precedent to other protists and parasites like Toxoplasma and Pythium where elements are conserved but differences yield novel insights.

Supplementary Material

This article contains supporting information, with one table.

Abbreviations:

Ab

antibody

α-KG

α-ketoglutarate

CTR

C-terminal region of Skp1

FBP

F-box protein

SCF

complex consisting of Skp1, cullin-1, F-box protein, and Rbx1

GGFGGn-

symbol for a pentasaccharide with the sequence Gal-Gal/Glc-Fuc-Gal-GlcNAc-

GT

glycosyltransferase

Hyp

(2S,4R) hydroxyproline

jmjC

jumonji-C domain containing

KO

a gene disruption resulting in no detectable function

MD

molecular dynamics

Ub

ubiquitin