Glycosides of hydroxyproline: Some recent, unusual discoveries

Abstract

Glycosides of hydroxyproline (Hyp) in the plant cell wall matrix were discovered by Lamport and co-workers in the 1960s. Since then, much has been learned about these Hyp-rich glycoproteins. The intent of this review was to compare and contrast some less common structural motifs, in nontraditional roles, to uncover themes. Arabinosylation of short-peptide plant hormones is essential for growth, cell differentiation and defense. In a very recent development, prolyl hydroxylase and arabinosyltransferase activity has been shown to have a direct impact on the growth of root hairs in Arabidopsis thaliana. Pollen allergens of mugwort and ragweed contain proline-rich domains that are hydroxylated and glycosylated and play a structural role. In the case of mugwort, this domain also presents a significant immunogenic epitope. Major crops, including tobacco and maize, have been used to express and produce recombinant proteins of mammalian origin. The risks of plant-imposed glycosylation are discussed. In unicellular eukaryotes, Skp1 (a subunit of the E3^SCF ubiquitin ligase complex) harbors a key Hyp residue that is modified by a linear pentasaccharide. These modifications may be involved in sensing oxygen levels. A few studies have probed the impact of glycosylation on the structure of Hyp-containing peptides. These have necessarily looked at small, synthetic molecules, since natural peptides and proteins are often isolable in only minuscule amounts and/or are heterogeneous in nature. The characterization of native structural motifs, together with the determination of glycopeptide conformation and properties, holds the key to rationalizing nature's architectural design.

Introduction

O-Glycopeptides are most often identified by the fact that their carbohydrate portion is expelled upon base-promoted β-elimination (Figure 1) (Taylor 1998). This results in the formation of an α,β-unsaturated amino acid and the epimerizable hemiacetal of the sugar. The stereochemistry of the glycosidic linkage (α or β) is destroyed during the degradation process. Another consequence of the degradation is the formation of a dehydro-amino acid and a corresponding reduction in the level of Ser and/or Thr upon amino acid analysis.

Fig. 1.

Base-induced elimination of a generic pyranoside of a β-hydroxyamino acid.

Glycosides of (2S,4R)-4-hydroxyproline (Hyp), tyrosine (Tyr) and hydroxylysine (Hyl) (Figure 2), however, do not undergo this elimination since they are not β-hydroxyacids. Glycosides of Tyr have long been known but are not so common (Zarschler et al. 2010). Glycosides of (2S,5R)-hydroxylysine are found predominantly in collagens and apparently only in mammals (Spiro et al. 1971; Butler 2008). Hydroxyproline glycoconjugates were first reported in the plant cell wall by Lamport and Clark (1969). These hydroxyproline-rich glycoproteins (HRGPs), which can be further divided into extensins, proline-rich proteins and arabinogalactan proteins, have been reviewed in detail (Kieliszewski and Lamport 1994; Sommer-Knudsen et al. 1997; Jose-Estanyol and Puigdomenech 2000; Lamport 2001; Xu et al. 2008; Lamport et al. 2011). The focus of this discussion is recent developments and unusual glycosides of Hyp in nature. By drawing these functionally disparate peptides and proteins together, on the basis of this common structural motif, we begin to ask: what are the common themes and design features?

Fig. 2.

Generic pyranosides of trans-4-hydroxyproline (Hyp), Tyr and Hyl.

Hydroxylation and subsequent glycosylation of proline represent two sequential post-translational modifications (Figure 3), and hence a significant investment of genetic and enzymatic machinery on the part of an organism. It should be noted at the outset that four isomers of hydroxyproline have been isolated from nature (Mauger 1996). The most common isomer, by far, is Hyp, herein after referred to as Hyp (aka trans-4-hydroxyproline in reference to the trans-relationship between the –COOH and –OH substituents on the pyrrolidine ring). The only O-glycoconjugates that have been characterized unequivocally to date are those of Hyp. The most common linkages, as found in HRGPs, are β-galactosides and α-arabinosides (Figure 3), although others are known. Glycosides can range from monosaccharides—through oligomers—to polysaccharides that account for the majority of a glycoprotein's mass.

Fig. 3.

Two sequential post-translational modifications: hydroxylation and glycosylation.

Proline is one of the most important coded amino acids vis-à-vis the conformation of peptides and proteins. The amino group is covalently linked to the side chain of the amino acid with two consequences upon amide bond formation: (a) the amide is secondary, so there is no amide proton to participate in hydrogen bonding and (b) the energy of the cis conformation about the prolyl amide bond is closer in energy to the trans conformation than for most peptide linkages (Figure 4). Both conformations are accessible, and the position of the cis–trans equilibrium is important for secondary structural elements and protein folding. For example, proline residues are known as “α-helix breakers” but are vital elements in β-turns. It is, therefore, important to evaluate how this structural control element is impacted by the post-translational modifications under discussion.

Fig. 4.

cis–trans isomerization of (A) regular peptide bond and (B) prolyl peptide bond.

Hydroxylation and glycosylation of proline vis-à-vis plant growth and development (Matsubayashi 2011)

Intercellular signaling is vital to all organisms for survival. Ligands based on proline-containing peptides are an important class of signal molecules in the plant kingdom. Moreover, the proline residues are frequently hydroxylated and subsequently sulfated or glycosylated.

The first secreted plant glycopeptide hormone to be structurally defined was PSY1 (plant peptide containing sulfated Y) depicted in Figure 5 (Amano et al. 2007). These researchers were specifically searching for sulfated peptides in Arabidopsis thaliana cell culture. Using LC-MS-MS, an 18-mer was identified wherein Tyr is sulfated and the first of two consecutive Hyp residues is glycosylated by a trimer of Ara (Figure 5). The structural details of the triarabinoside have not been reported; they could well resemble those of CLAVATA 3 (CLV3, vide infra).

Fig. 5.

(A) Precursor protein At5g58650 and the (B) structure of mature PSY1.

Natural PSY1 promoted cell proliferation in a concentration-dependent manner. A synthetic peptide devoid of carbohydrate had only marginal activity. Interestingly, PSY1 was significantly upregulated by wounding (vide infra). A leucine-rich repeat receptor kinase (LRR-RK) was identified as the receptor for the PSY1 ligand. There are three of these paralogous LRR-RKs with two ligands—PSY1 (vide supra) and phytosulfokine (PSK), a pentapeptide with two sulfated Tyr residues but no glycosylation. These LRR-RKs are transmembrane receptors (Figure 6), with the receptor ligand-binding domain on the extracellular surface. While little is known about their molecular interactions in ligand binding, LRR domains have been shown to bind all manner of ligands including peptides such as PSY1. Ligand binding induces activation of the kinase domain. In the innate immune system of animals, toll-like receptors (TLRs), that detect molecules of viral and bacterial origin on the cell surface, also feature leucine-rich repeat sequences (Akira 2003; Kumar et al. 2011).

Fig. 6.

Conceptual representation of an LRR-RK.

A hormone that binds tightly to CLV1 (CLAVATA 1, a receptor-like kinase), and thereby controls the fate of stem cells in the shoot apical meristem of A. thaliana, has recently been characterized by nano-LC-MS-MS (Ohyama et al. 2009). The structure of the CLAVATA 3 ligand (CLV3) and that of the related peptide CLE2 are shown in Figure 7.

Fig. 7.

Peptide hormones of A. thaliana, highlighting the l-Ara-β-(1 → 2)-l-Ara-β-(1 → 2)-l-Ara-β-(1 → 4)-Hyp motif.

An earlier report had implicated the dodecapeptide CLE domain in the regulation of stem-cell differentiation (Ito et al. 2006; Kondo et al. 2006). However, when clv3 mutants were treated with the synthetic CLE motif (H.Arg-Thr-Val-Hyp-Ser-Gly-Hyp-Asp-Pro-Leu-His-His.OH), the peptide did not fully rescue the activity. While proline hydroxylation had been identified, Ohyama et al. reasoned that further maturation (viz. glycosylation) was required for full activity. Chemical derivatization and degradation of the purified CLV3 glycopeptide demonstrated that Hyp⁷ was modified by a trimer of l-arabinose, two of which were attached via 1 → 2 glycosidic linkages at the nonreducing end (Figure 7). The β-stereochemistry was inferred from the observation that the glycopeptide was stable to an α-arabinofuranosidase. Without the glycosylation motif, the ability of the peptide to displace radiolabeled peptide from the receptor was reduced by more than two orders of magnitude (Table I).

Table I.

Binding of peptide hormones to the ectodomain of CLV1 receptor kinase

Peptide	Kd (nM)
CLV3a	1
CLV3	280
CLE2a	1.2
CLE	330

^aThe peptide bears the arabinose trimer at Hyp⁷. Peptide structures are as defined in Figure 3.

Moreover, proline hydroxylation and subsequent arabinosylation impacts the growth of root hair cells in A. thaliana, an important cell type vis-à-vis nutrient absorption (Mohnen and Tierney 2011; Velasquez et al. 2011). The initial observations were that ethyl-3,4-dihydroxybenzene and α,α-dipyridyl resulted in decreased root hair growth and the accumulation of nonglycosylated protein. These compounds chelate to ferric (Fe²⁺) ions and thus inhibit nonheme dioxygenases, including prolyl 4-hydroxylases (P4H). Three prolyl hydroxylases (PHDs) were characterized from A. thaliana; knockouts confirmed that inadequate PHD activity reduced root hair growth. Over-expression of P4H increased the length and density of hairs. Mutants of RRA3 and XEG113 (arabinosyl transferases) also resulted in retarded root hair phenotypes.

Plant hormones associated with defense

In 1991, an 18-amino acid peptide called “systemin” (Table II) was isolated from tomato plants and shown to be produced in response to attacks by herbivores. The term “systemin” was subsequently coined by Ryan and Pearce (2003) to describe polypeptide defense signals produced by plants in response to physical damage. The receptor (SR160) that binds systemin is also an LRR-RK (Scheer and Ryan 1999).

Table II.

Peptide sequences of systemins isolated from plant sources (O = Hyp = trans-4-hydroxyproline); centrally located contiguous prolines followed by an A/T/S residue are underlined

Plant	Peptide	Amino acid sequence	No. of pentoses
Tomato	Systemin	AVQSKPPSKRDPPKMQTD	0
	LeHypSysI	RTOYKTOOOOTSSSOTHQ	8–17
	LeHypSysII	GRHDYVASOOOOKPQDEQRQ	12–16
	LeHypSysIII	GRHDSVLPOOSOKTD	10
Tobacco	TobHypI	RGANLPOOSOASSOOSKE	9
	TobHypII	NRKPLSOOSOKPADGQRP	6
Petunia	PhHypSysI	RSLHKSOOOTOKPSDEQGQ	10
	PhHypSysII	RHDYHLSOOOAOKPADHTGQ	10
	PhHypSysIII	RGKRLPOOAOEYDPOYHQ	3–6
Sweet potato	IbHypSysIV	REEKPOOOOAQETDDPNRP	6–12
Black nightshade	SnHypSysI	RNRPYITOSOOEASOSTKQ	6
	SnHypSysII	GRHDHVLPOOSOKHEPIIGQ	6
	SnHypSysIII	GRHDHVOAOOAOKPEDEQGQ	6, 9

The same group has described three hydroxyproline-containing systemins (HypSys) from tomato, Lycopersicon esculentum (details in Table II) (Pearce and Ryan 2003). LeHypSysI and II are composed of 18 and 20 amino acids, respectively, and each contains four contiguous hydroxyprolines. LeHypSysIII is composed of 15 amino acids with a total of three hydroxyprolines. All peptides have positively charged amino acids near the N- and C-termini and are synthesized via the secretory pathway where they are hydroxylated and glycosylated. As for CLV3, comparison with synthetic peptides lacking the carbohydrate moiety showed a dramatic decrease in activity for LeHypSys I (less than 1000 times the natural peptide) and nearly inactive for LeHypSysII and III. Once again, structural details of the carbohydrate component have not been determined.

The HypSys peptides are genetically distinct from systemin and are produced via a different pathway (Narváez-Vásquez et al. 2007). The genes are expressed, and the resulting proteins sequestered in different cellular locations (Figure 8). Narváez-Vásquez et al. (2005) have shown that LeHypSys is synthesized in the phloem parenchyma cells of the vascular bundles of tomato leaves and that the nascent protein is sequestered in the cell wall matrix. Prosystemin does not have a signal sequence; the HypSys precursor proteins of the peptides listed in Table II have a signal sequence at their N terminus. A single copy of systemin is encoded, whereas the precursor proteins for HypSys hormones carry more than one defense peptide. Systemin and the HypSys peptides work together to regulate systemic wound signaling in the tomato.

Fig. 8.

Expression of defense hormones in tomato.

Two 18-mer hydroxyproline-rich systemins were isolated from tobacco (Table II) (Pearce et al. 2001) with the characteristic -Hyp-Hyp-Ser- sequence. Mass spectrometric analysis indicated the number of pentose sugars present in each peptide. Although the linkages and identities of the sugars have not been determined for these HypSys peptides, it is likely that they are related to HRGPs or the shorter hormone peptides described above. Synthetic TobHypI and TobHypII (without the appended carbohydrates) were 10,000 times less active than the native glycopeptides. This was the first discovery of a plant polypeptide prohormone that produces multiple polypeptide signals, a feature that is common in animals and yeast. The preproprotein consists of 165 amino acids, including a signal sequence. TobSysI is derived from the N terminus and TobSysII near the C terminus.

In the event of trauma, HypSys peptides are released from the cell wall matrix. It was proposed that HypSys peptides are widespread throughout the plant kingdom and have a general role in plant responses to herbivores and/or pathogens. Indeed, Pearce et al. (2007) have now reported similar peptides from petunia (Petunia hybrida), the leaves of the common sweet potato (Ipomoea batatas) (Chen et al. 2008) and black nightshade leaves (Solanum nigrum) (Pearce et al. 2009). The sweet potato was the first example outside the Solanaceae family.

Plant defense genes are activated in an analogous fashion to the inflammatory response pathways found in animals. In plants, linoleic acid is produced at the site of injury from damaged membrane lipids (Figure 9). This fatty acid is converted to jasmonates that trigger the release of HypSys peptides which in turn promote jasmonate formation via a feedback amplification mechanism. In animals, arachidonic acid is released from damaged membranes, leading to the production of prostaglandins. In both classes of organisms, these triggers activate similar enzymatic processes (Figure 9).

Fig. 9.

Plant defense compared with the mammalian inflammatory response.

Hydroxyproline glycosides in pollen allergens

Art v 1, the major allergen of mugwort (Artemisia vulgaris), is a glycoprotein with defensin-like (head) and hydroxyproline-rich (tail) domains (Himly et al. 2003). The prolyl domain facilitates protein folding (Gadermaier et al. 2010) and influences the conformation of the globular domain (Razzera et al. 2010). The defensin domain (residues 1–55) is cysteine-rich and stabilized by disulfide bonds. The hydroxyproline-rich C-terminal domain comprises residues 57–108 (Table III).

Table III.

Proline-rich domains of Art v 1 and Amb a 4; Pro residues are bold for emphasis

	57	61	71	81	91	101
Art v 1	PPGA	TPAPPGAAPP	PAAGGSPSPP	ADGGSPPPPA	DGGSPPVDGG	SPPPPSTH
Amb a 4	NPGP	PPGAPKGKAP	APSPPSGGGA	PPPSGGEGGD	GPPPPEGGEG	GGDGGGE…

Two novel O-glycans in this section of Art v 1 have been described (Leonard et al. 2005). While the sugars are the plant-ubiquitous l-arabinose and d-galactose, the glycosyltransferases involved are atypical, producing unprecedented glycan structures. Characterized by chemical and enzymatic degradation, mass spectrometry and high-field nuclear magnetic resonance (NMR), the first new type of arabinogalactan was designated the hydroxyproline polysaccharide (Hyp PS). This O-glycan has a small β-(1 → 6)-linked galactan core, instead of the usual β-(1 → 3)-linked units (Figure 10A). Mass spectrometry indicated that Hyp PS exists as isoforms that contain 5–28 α-linked arabinofuranose residues in the positions indicated in Figure 10A. Leonard et al. proposed that this be termed a type III arabinogalactan and that this motif might be more widespread in allergenic pollens produced by the Asteraceae. Surprisingly, Hyp PS bound only very weakly to antibodies from the sera of mugwort-allergic patients. A similar glycoprotein from Phleum pratense also disappointed investigators by its insignificant binding to IgE (Haavik et al. 1987a, 1987b).

Fig. 10.

(A) Hyp PS of Art v 1; (B) the single β-l-Ara-Hyp motif found in both Art v 1 and Amb a 4; (C) Hyp PS of Amb a 4; arrows indicate potential sites for further α-arabinosylation.

The second novel O-glycoside of Art v 1 is a single β-arabinofuranoside linked to a hydroxyproline residue (Figure 10B). This O-glycan can exist in isolation (as a monomer), but up to four adjacent β-arabinosylated prolines may be present (Figure 11). This clustering of glycosylated hydroxyproline is well documented for HRGPs (Kieliszewski 2001). Unlike other well-known HRGPs having the Ser-(Hyp)₄ motif, no oligo-arabinosides were found in Art v 1. This second type of Hyp glycoside did react with antibodies from the sera of mugwort-allergic patients. In a study involving 100 subjects, only 39% reacted with recombinant Art v 1 (Oberhuber et al. 2008). The remaining 61% of patients responded only to natural Art v 1, signifying the importance of post-translational modifications (viz. proline hydroxylation and subsequent β-arabinosylation) in recognition of the allergen.

Fig. 11.

l-Ara-β-(1 → 4)-Hyp tetrad.

Altmann and coworkers recently characterized a new allergen, Amb a 4, from ragweed, Ambrosia artemisiifolia (Leonard et al. 2010). This protein also contained a defensin-like domain with 50% homology to Art v 1. The much shorter, C-terminal, Hyp-rich domain contained a different Hyp PS (Figure 10C) and significantly lower levels of the β–(1 → 4)-Ara-Hyp motif. Recombinant Amb a 4 showed reactivity with the sera of more than 30% of weed pollen allergic patients. The majority of mugwort- and ragweed-allergic patients nevertheless require glycosylation of the protein in order to show an allergic reaction.

Potential for glycosylation of Hyp during recombinant protein expression in plant hosts

The growth in biopharmaceuticals, including a significant number of glycoproteins, has prompted investigation of their expression in nonmammalian cells or host organisms. In particular, crops such as tobacco and maize offer the potential for mass production without the liabilities associated with mammalian cell lines (e.g. costs, facility validation and scalability). The protein end products need to be an adequate replica of the natural, approved drug and the accuracy of glycosylation motifs is of critical importance. Desired glycosylation must be articulated and adventitious host-induced post-translational modifications need to be monitored. N-Glycosylation in plants is different to that in mammals but has been engineered to produce acceptable, sometimes even improved, constructs in recombinant protein (Karg and Kallio 2009).

Human immunoglobulin A1 (hIgA1) is used for topical disease treatments, including herpes simplex virus. Enzymatic degradation of the native human antibody, in combination with amino acid analysis, established that O-glycans are located at Thr²²⁸, Ser²³⁰ and Ser²³² of the hinge region, with Thr²²⁵ and Thr²³⁶ modified to a lesser degree (Figure 12B). The most common glycan motif was found to be a sialylated T-antigen trisaccharide (Figure 12C) (Mattu et al. 1998). During efforts to produce hIgA1 in transgenic maize, glycosylation of the hinge region of the heavy chain was significantly altered (Karnoup et al. 2005).

Fig. 12.

(A) Schematic representation of an antibody, modified by glycosylation in the exaggerated hinge region of the heavy chain [V = variable, C = constant, H = heavy, L = light]; (B) Amino acid sequence of the hinge region with a comparison of the sites of glycosylation [solid arrow = highly likely; dashed arrow = possible]; and (C) Comparison of glycan structures.

Detailed analysis of the recombinant protein from transgenic maize involved enzymatic degradation, in combination with amino acid analysis and matrix-assisted laser desorption mass spectrometry. The studies revealed that proline residues were substrates for post-translational hydroxylation and subsequent glycosylation. These modifications were attributed to the sequence similarity of the heavy chain hinge region of hIgA1 to the maize extensin family of HRGPs. The study estimated that up to six Pro residues are hydroxylated and that 6–10 arabinose units are associated with each hIgA1 molecule. Two residues (Pro²²⁷ and Pro²³³) were identified as sites for hydroxylation, but other sites of modification were not identified unequivocally (Figure 12C). Approximately 90% of the recombinant protein was modified.

The proline-rich hinge region of native hIgA1 was proposed to adopt a rod-like conformation, and heavy glycosylation was suggested to protect against degradation (Mattu et al. 1998). While the maize-produced protein has a very different glycosylation pattern, it may serve the same net function. There has been no experimental verification of the impact of proline hydroxylation and glycosylation on the stability, biological activity and efficacy of the maize-recombinant IgA1 antibodies.

While Karnoup et al. predicted that their “findings may be of great importance to the field of plant biotechnology”, the literature of the intervening 6 years has not revealed numerous other examples of inappropriate protein glycosylation in plants. It may be that the hinge region of hIgA1 displays an uncommon resemblance to the substrates of PHDs and arabinosyltransferases. A recent report advocates inclusion of α,α′-dipyridyl in hydroponic growth media to suppress plant-specific O-glycosylation (Moriguchi et al. 2011). As discussed earlier, this compound is a PHD inhibitor. Without proline hydroxylation, the subsequent attachment of arabinosides is moot. The broader implications of inhibiting other nonheme dioxygenases were not addressed.

A 44-kDa fragment of human collagen I α1 (CIα1) was accurately expressed in corn grain to render protein that is molecularly equivalent to that produced in recombinant yeast (Zhang et al. 2009). This collagen-related protein is rich in proline residues and thus presents the potential for plant-induced hydroxylation and subsequent glycosylation. The recombinant CIα1 was extracted from early generation plants having low levels of recombinant protein accumulation. The low level of hydroxyproline is suggestive of an endogenous PHD in corn seed, although activity is limited compared with that in human collagen (Table IV). Co-expression of mammalian prolyl 4-hydroxylase (P4H) would be necessary to obtain levels of hydroxylation characteristic of human collagen I. A commercial kit for staining glycoproteins gave a negative result for corn-derived CIα1. Thus, it would seem that the collagenous sequence is a poor substrate for hydroxylation in corn. Moreover, sequences do not sufficiently mimic those of HRGPs to induce glycosylation.

Table IV.

Levels of proline hydroxylation in CIα1 as a percentage of the total amino acids

Source	Pro (%)	Hyp (%)
human	12.4	10.8
Pichia (yeast)	23.71	0
corn grain	16.35	2.01

Skp1: A glycoprotein associated with ubiquitination

Skp1 is a cytoplasmic and nuclear protein that constitutes a subunit of the E3^SCF ubiquitin ligase complex that is responsible for ubiquitination of cell cycle and other regulatory proteins and transcriptional factors. In Dictyostelium, the proline residue at position 143 of Skp1 undergoes hydroxylation and glycosylation (Teng-umnuay et al. 1998). These post-translational modifications involve six enzyme activities coded by four genes (West et al. 2004, 2010).

Specifically, Pro¹⁴³ is the target of a cytoplasmic prolyl hydroxylase (P4H1). The Dictyostelium P4H1 operates via a similar mechanism to that whereby HIFα is hydroxylated by animal PHDs involving a nonheme Fe(II)-dependent dioxygenase (van der Wel et al. 2005). P4H1-catalyzed hydroxylation of Pro¹⁴³ is stereospecific and regiospecific. Recent ¹H NMR studies showed that the oxidation delivers (2S,4R)-4-hydroxy-L-proline (Hyp) (van der Wel et al. 2011) by analogy to Hyp⁴⁰² and Hyp⁵⁶⁴ in HIFα and plant Hyp. These enzymes belong to a clade of the same superfamily that modifies the plant proteins described elsewhere in this review and animal collagens, but are distinctive in their lower affinity for O₂. In addition, their substrate spectrum is relatively narrow, and Skp1 appears to be the only substrate for P4H1 (West et al. 2007). Genetic evidence suggests that P4H1 hydroxylation mediates O₂-dependent development of Skp1 of Dictyostelium, enabling the organism to differentiate into a fruiting body at the soil surface. P4H1 may serve as a sensor for O₂ and/or 2-oxoglutarate, a cosubstrate that may also be limiting in certain metabolic regimes (van der Wel et al. 2011; Wang et al. 2011).

The Skp1 protein is further modified by a pentasaccharide chain (Figure 13). The sequence of the linear pentasaccharide was determined by mass spectrometry, exoglycosidase digestions and in vitro enzyme specificities to be d-Gal-(1 → 6)-d-Gal-α(1 → 3)–l-Fuc-α(1 → 2)-d-Gal-β-(1 → 3)-GlcNAc with all residues in the pyranose form, with uncertainty remaining about the position of the last galactose residue. A sequence of papers by West and co-workers described the isolation and identification of the glycosyltransferases from the organism's cytoplasm. Analysis of mutant strains, used to demonstrate the biosynthetic roles of the glycosyl transferase (GT) genes, revealed that glycosylation has modulatory effects on O₂-dependence, as if intermediate stages of glycosylation permit a reversal of the effect of hydroxylation, and completion of the chain restores the original effect.

Fig. 13.

Linear oligosaccharide linked to Pro¹⁴³ in Skp1. The position of the last Gal residue is tentative.

The GlcNAc residue at the reducing end of the pentasaccharide is transferred from UDP-GlcNAc to hydroxyproline by Gnt1 (Figure 14) (Teng-umnuay et al. 1999). This GT60 glycosyltransferase is related to the family of GT27 enzymes that form α-GalNAc-Ser/Thr linkages in the Golgi. This makes it highly likely that the GlcNAc-Hyp linkage is an α-glycoside (van der Wel, Fisher, et al. 2002). Further evidence for the site of glycosylation was afforded by the observation that the Skp1 isoform with unmodified Pro at position 143 was not a substrate for Gnt1, but could inhibit the enzyme-catalyzed reaction (Figure 14). Gnt1 exhibits unusually low K_m values in the submicromolar range, consistent with the enzyme's occurrence in the cytoplasm, and contrasting with K_m values for GTs that operate in the Golgi. Short peptides based on the Skp1 sequence are not good substrates for Gnt1, signaling that the enzyme is recognizing more than the primary sequence of the Skp1 protein.

Fig. 14.

Gnt1-catalyzed transfer of GlcNAc.

Conformational changes in this region of the protein, following glycosylation, likely contribute to its recognition by a PgtA, a 768-residue bifunctional protein that introduces the next two monosaccharides. The N-terminal domain of the protein has galactosyl transferase activity, and the fucosyl transferase activity is associated with the C-terminal region (van der Wel, Morris, et al. 2002; Wang et al. 2009). The two activities have a processive relationship in in vitro assays. In common with previous enzymes in the pathway, the β3GalT activity depends heavily on recognition of full-length Skp1. In contrast, the α2FucT activity exhibits substantial activity toward the Gal-(β-1,3)-GlcNAc disaccharide.

An enzyme, AgtA, isolated and purified from the Dictyostelium cytoplasm has been shown to transfer galactose from UDP-Gal to the 3-position of the fucose at the nonreducing end of the free trisaccharide (Ketcham et al. 2004). Comparative studies showed that the recombinant enzyme is much more active when the trisaccharide is attached to Skp1, under which circumstances the enzyme is also capable of addition of the final αGal residue (West et al., unpublished data). AgtA is required for terminal glycosylation in cells and for normal sensitivity to O₂ during fruiting body formation (Ercan et al. 2006; Wang et al. 2009). Interestingly, AgtA was the first documented member of the Carbohydrate-Active EnZYme (CAZy) GT77 family (Cantarel et al. 2009), which appears to contain arabinosyltransferases that modify Hyp in the Arabidopsis Golgi (Egelund et al. 2004).

West et al. (2002, 2003, 2004, 2010) have hypothesized that these post-translational modifications are likely widespread in lower eukaryotes. Indeed, experiments have demonstrated cytoplasmic glycosylation of Skp1 in Toxoplasma gondii (West et al. 2006), the causative agent of toxoplasmosis.

Bulgecin and related peptides

The Skp1 of Dictyostelium, described above, is unusual for a number of reasons, including the attachment of a GlcNAc residue to Hyp. In this regard, bulgecin and its peptidyl derivatives are introduced for comparison (Figure 15). The bulgecins were isolated from Pseudomonas species (Imada et al. 1982; Shinagawa et al. 1984). Longer peptides were reported from Chromobacterium biolaceum (Cooper and Unger 1986). These compounds potentiate the activity of β-lactam antibiotics and subsequently were found to bind at the active site of the soluble lytic transglycosylase of E. coli (Thunnissen et al. 1995). Notable features are: hydroxymethylation (CH₂OH) at Cγ of the proline, a cis-4-hydroxyproline, a β-glycosidic linkage and sulfation of both the sugar and the peptide. With the exception of sulfation, these structural features are significantly different to others reported above and reflect the bacterial origins of the peptides.

Fig. 15.

Bulgecins and related peptides.

Structural studies on glycosylated hydroxyproline motifs

When the structure of a glycopeptide or glycoprotein is reported, varying levels of characterization are provided. Amino acid and carbohydrate analyses render an amino acid sequence and the identity and ratios of monosaccharides present. More in-depth analysis, employing mass spectrometry and enzymatic analysis, yields information on the position of the carbohydrates and the stereochemistry of their glycosidic linkages. Collectively, this information instructs us on the primary sequence of the molecule. The manner in which the molecule folds into secondary structural elements and is incorporated into larger domains or complexes is of obvious relevance to the biological function of the molecule.

NMR was used to study the plant hormone systemin (the eponymous peptide without the post-translational modifications). While the structure of the 18-mer could be assigned, there were no defined secondary structural elements (Russell et al. 1992). Subsequent investigations using circular dichroism (CD) advocated a 3₁ helix (Toumadje and Johnson 1995) or a mixture of random coil, beta turn and beta sheet (Mucha et al. 1999).

Schweizer's group has taken an empirical approach to assessing the structural impact of glycosylation on Hyp-containing peptides. They used NMR to study the family of compounds depicted in Table V (Owens et al. 2007). Within experimental error, there were no differences in the thermodynamics and kinetics as a consequence of alkylation or glycosylation of the alcohol. However, nuclear Overhauser experiments revealed close contacts between the proline and galactose rings of the α-galactoside (Figure 16), suggesting that glycosylation could impact backbone conformation in HRGPs.

Table V.

Cis–trans isomerism of glycosylated (2S,4R)-4-hydroxy-l-proline

R	Kt/c (37 °C)	kct (s−1)a	ktc (s−1)a
H	3.52 ± 0.05	0.73 ± 0.01	0.25 ± 0.01
tBu	3.34 ± 0.15	0.77 ± 0.02	0.27 ± 0.01
α-d-Gal	3.41 ± 0.30	0.83 ± 0.05	0.27 ± 0.02
β-d-Gal	3.37 ± 0.28	0.61 ± 0.04	0.21 ± 0.02

^aPhosphate buffer, pH 7.4 at 67°C.

Fig. 16.

Conformation of α- and β-galactosides of Hyp.

A more recent report indicates that the glycosylation of (4S)-hydroxyprolines (aka allo-hydroxyproline, hyp) has a greater influence on conformation (Owens et al. 2009). Specifically, both α- and β-galactosides of Ac-hyp-OMe showed increased rates of cis–trans amide bond isomerization relative to the unglycosylated residue (Table VI). allo-Hydroxyproline has been found in small peptides (Wieland 1986; Mauger 1996), specifically the Amanita toxins that have recently been shown to arise via ribosomal peptide synthesis (Hallen et al. 2007).

Table VI.

Cis–trans isomerism of glycosylated (2S,4S)-4-hydroxy-l-proline

R	Kt/c (24.8 °C)	kct (s−1)a	ktc (s−1)a
H	2.4 ± 0.1	0.44 ± 0.04	0.20 ± 0.01
α-d-Gal	2.9 ± 0.3	0.59 ± 0.06	0.25 ± 0.03
β-d-Gal	2.9 ± 0.1	0.71 ± 0.04	0.30 ± 0.02

^aPhosphate buffer, pH 7.2, 0.1 M at 67.3°C.

More recently, Owens et al. (2010) reported on a contiguous O-galactosylated hydroxyproline sequence. Specifically, they compared Ac-(Pro)₉-NH₂, Ac-(Hyp)₉-NH₂ and Ac-[β-d-Gal(Hyp)]₉-NH₂. The CD spectra of all three peptides are typical of a polyproline II helix. Results indicate that the carbohydrate units provide conformational and thermal stability to the protein. The GROningen Molecular Dynamics Simulations (GROMOS) force field was used to depict a model that showed the galactose units interacting with both the peptide backbone and one another.

In their recent studies on root hair growth, Velasquez et al. (2011) retrieved the structure of SPPPP from the protein databank and added trans-4-hydroxy groups to the proline residues to generate SOOOO. Arabinosides were then appended according to the wild-type and the weg113 mutant that has a deficient arabinosyltransferase activity. They also used the GROMOS force field parameters and GROningen MAchine for Chemical Simulations (GROMACS) simulation suite to generate minimized structures. They concluded, in agreement with Owens et al. (vide supra), that the carbohydrate appendages stabilize the helical structure.

Summary

The developments described above make it clear that hydroxylation and glycosylation of proline—two sequential post-translational modifications—lead to peptides and proteins with important roles in nonmammalian systems. Our knowledge of the associated enzymes and their substrate selectivity is in its infancy. The same is true of the structural ramifications of these molecular embellishments: we know that they affect biological activity but much is yet to be learned about the impact on peptide and protein conformation. While there are no examples to date of hydroxyproline glycosides in mammals, theses motifs impact humans in terms of plant-produced, therapeutic glycoproteins, pollen allergens and some pathogenic organisms that rely on these post-translational modifications in their life cycle. Advances in mass spectrometry and NMR make the discovery of further examples, like those described herein, inevitable.

Funding

Research in the Taylor Group, relating to the glycosylation motif in Skp1 was supported by the NIH-NIGMS (R01GM84383-S).

Abbreviations

Standard three letter codes are employed for monosaccharides (e.g. Gal = galactose). Standard one letter and three letter codes are employed for amino acids (e.g. O = Hyp = trans-4-hydroxyproline). Ac, acetyl; Amb a 4, A. artemisiifolia 4; Art v 1, Artemisia vulgaris 1; CAZy, Carbohydrate-Active EnZYme (a database); CD, circular dichroism; CIα1, collagen I α1; CLE2, CLV3/ESR-related 2; CLV1, CLAVATA 1; CLV3, CLAVATA 3; E3^SCF, E3 ubiquitin ligase Skp1-Cullin-F-box protein complex; GROMACS, GROningen MAchine for Chemical Simulations; GROMOS, GROningen Molecular Dynamics Simulations; GT, glycosyl transferase; HIFα, hypoxia-inducible factor α; hIgA1, human immunoglobulin A1; HRGP, hydroxyproline-rich glycoprotein; Hyp, (2S,4R)-4-hydroxy-l-proline [aka trans-hydroxyproline]; Hyp, (2S,4S)-4-hydroxy-l-proline [aka cis-hydroxyproline]; Hyp PS, hydroxyproline polysaccharide; HypSys, hydroxyproline-rich systemin; K_d, dissociation constant; LC-MS-MS, liquid chromatography/mass spectrometry/mass spectrometry; LeHypSys, Lycopersicon esculentum hydroxyproline-containing systemin; LRR, leucine-rich repeat; Me, methyl; NMR, nuclear magnetic resonance; P4H, prolyl 4-hydroxylase; PHD, prolyl hydroxylase; PSK, phytosulfokine; PSY1, plant peptide containing sulfated tyrosine 1; RK, receptor kinase; Skp1, S-phase kinase-associated protein 1; TLR, toll-like receptor; TMD, transmembrane domain; Tyr, tyrosine.