Recent advances on glycosyltransferases involved in the biosynthesis of the proteoglycan linkage region

Proteoglycans are an essential family of glycoproteins consisting of a core protein with one or multiple glycosaminoglycan (GAG) chains, which are covalently attached to the protein through a common tetrasaccharide linkage that consists of glucuronic acid (GlcA)-β(1 → 3)-galactose (Gal)-β (1 → 3)-galactose (Gal)-β(1 → 4)-xylose (Xyl)-β(1 → 0)-serine (Ser) (Fig. 1). PGs are widely present on the cell surface and on the extracellular matrix. Their functions are critically important in numerous biological events, including cell adhesion, proliferation, cellular signaling, and interactions with growth factors.^1–4

Fig. 1.

Schematic demonstration of the structure of proteoglycans. The tetrasaccharide linkage is highlighted in the blue box.

The biosynthesis of the PG linkage tetrasaccharide involves the sequential deployment of four glycosyl transferases: xylosyltransferase-I/II [XT-I/II], β-1,4-galactosyltransferase [7 (β4GalT7)], β-1,3-galactosyltransferase [6 (β3GalT6)], and β-1,3-glucuronyltransferase [3 (β3GAT3)] (Fig. 2). The first successful expression and characterization of β3GalT6 were reported by the Furukawa and Esko groups two decades ago.^5,6 The Sugahara group reported the first molecular cloning and expression of β3GAT3, and subsequent characterization of this enzyme in 1990s.^7,8 The follow-up investigations on β3GalT6 and β3GAT3 have been rather limited.^9,10 Therefore, this current review will focus on the recent progress made on the expression, characterization and applications of the PG linkage glycosyltransferases XT-I/II and β4GalT7.

Fig. 2.

Biosynthetic assembly of the PG linkage region.

1. Xylosyltransferase-I/II

The Wilson group first comprehensively summarized the understanding toward UDP-α-d-xylose:proteoglycan core protein β-d-xylosyltransferases I and II (XT-I/II).¹¹ In 2007, Götting et al. published a review emphasizing the impacts of mammalian xylosyltransferases on PG-related diseases and human health.¹² Since then, significant advances have been achieved to gain insight on XT-I/II.

1.1. Expression and purification of XT-I/II

The discovery of peptide O-xylosyltransferase dates back to the 1960s.^13–17 Afterward, this GAG-synthesis-initiating enzyme has been isolated from multiple sources.^13–22 In 2000, Götting and co-workers reported the first molecular cloning and expression of XT-I and its isoform.²³ In their study, the recombinant XT-I proteins from humans, mice and rats were successfully expressed in Chinese Hamster Ovary (CHO-K1) cells.

In 2003, the Kleesiek group described high-level expression of a soluble histidine-tagged recombinant XT-I using the High Five/pCG255–1 insect cell expression system.²⁴ Stable clones that express XT-I-V5-His (rXT-I-His) were generated. The human XT-I was purified by heparin affinity chromatography using a POROS 20 HE2 column followed by a nickel affinity column. The purified protein was verified by Western blot using polyclonal anti-XT-I antibodies.

Shortly thereafter, Götting and co-workers prepared a series of XT-I enzymes with point mutations on the aspartate-any residue-aspartate (DXD) motifs by transient expression in High Five insect cells.²⁵ A stable clone of High Five/pCG255–1 that expresses the soluble form of histidine- and V5-tagged recombinant human XT-I with N-terminal 1–148 sequence truncated, rXT-I-(Δ1–148)-V5-His, was also made in this study.

Müller et al., in 2005, carried out individual site-directed mutagenesis of all 14 cysteine residues of human XT-I into alanine.²⁶ The recombinant wild-type enzyme and the single mutants were successfully expressed in High Five insect cells to assist the structure–activity study of XT-I. A year later, in work published by the same group, multiple N-terminal truncated human XT-I enzymes were smoothly produced in the same insect cell expression system.

With the successes from the CHO mammalian cell and High Five insect cell expression systems, the expressions of xylosyltransferases were extended to the human embryonic kidney 293 (HEK-293), human osteosarcoma (SaOS-2) mammalian system, and Pichia pastoris yeast system.^27,28 In 2006, the Götting group reported the first recombinant expressions of GFP-fused human XT-I and multiple GFP-tagged XT-I/II mutants using mammalian HEK-293 and SaOS-2 cells.²⁷ In the same year, Brunner et al. expressed two invertebrate and two vertebrate xylosyltransferases, Drosophila peptide O-xylosyltransferase (OXT), Caenorhabditis peptide O-xylosyltransferase (SQV-6), and human xylosyltransferase I/II (XT-I/II), with the Pichia pastoris expression system.²⁸ Two years later, another successful story with the Pichia pastoris expression system was reported by the Götting group.²⁹

1.2. Acceptor specificity of XT-I/II

The first description of the acceptors for XT-I dates back roughly five decades.^{13,18,19,30–32} In the pioneering studies, various exogenous or endogenous proteins were validated to be acceptors of xylosyltransferases. Since then, our understanding of the acceptor specificity of XT-I/II has been significantly expanded.

In addition to acceptor proteins, diverse peptide acceptors have been derived from the amino acid sequence around the GAG attachment sites of different proteoglycans.^{11,16,17,28,30,32–37} Among the reported acceptors of XT-I/II, bikunin protein is known to be one of the best acceptors based on the Michaelis–Menten constants (K_m). The bikunin peptide sequence derived from the bikunin GAG-attachment site has later on been extensively used to study the acceptor recognition properties of XT-I/II.^{16,23,28,29,33,35,38–40}

As the acceptor scope of XT-I expands, considerable effort has been expended to determine its minimal binding motif, Gly-Ser-Gly or Ser-Gly-x-Gly, where x = any amino acid.^{12,36,41–44} Meanwhile, some evidence indicates that the presence of a serine residue may not be absolutely required as threonine-mutated core protein could also be glycosylated, albeit with lower degrees of glycosylation.^33,45 Beyond the minimal motif of acceptor binding, a consensus-favored acceptor sequence for XT-I, a-a-a-a-Gly-Ser-Gly-a-b-a, where “a” is Glu or Asp and “b” is Gly, Glu or Asp, was deduced by Brinkmann and co-workers in 1997, based on the peptide sequence of reported acceptors of xylosyltransferases.¹⁶ Shortly thereafter, the common sequence was refined by the same research group to a-a-a-x-Ser-Gly-x-Gly, where a=Glu or Asp and x = any amino acid.¹⁷

With the successful expression of XT-1, research focus was subsequently extended to XT-II. Roch and co-workers discovered that XT-II possesses a consensus sequence analogous to that for XT-I, i.e., a-a-a-a-Gly-Ser-Gly-a-a/Gly-a, where a = Asp or Glu.⁴⁰

Lately, to investigate the acceptor recognition properties of XT-I, Briggs and Hohenester performed detailed analyses using a comprehensive bikunin-derived 12-amino-acid peptide acceptor library in which the amino acid residue at each position was mutated to 1 of the 20 common natural amino acids.³⁹ Consistent with the prior reports,^33,45 although a serine residue is highly preferred at the xylosylation site, peptides with a threonine residue at position 0 also show noticeable activity levels. The −1 position, originally a glycine, can accept a wide variety of uncharged amino acids. While the −2, −3 and −4 sites generally favor acidic amino acids, individual replacement of the glutamic acid residues does not exert a strong influence on the enzymatic activity. The preference for the acidic amino acids at positions preceding the xylosylation site has been attributed to non-specific charge–charge interactions with the positively charged residues around the binding pocket. For the +1 position, small amino acids including glycine, alanine, serine and threonine are strongly favored. Surprisingly, a valine residue at the +2 site enhances the activity level considerably, as opposed to the native glycine. Overall, XT-I does not strictly require a certain acceptor peptide sequence for the activity and exhibits a greater structural tolerance than previously described (Fig. 3). This recent discovery furthers contemporary understanding toward XT-I acceptor recognition properties and implies vast application potentials attributing to the relaxed acceptor requirements.

Fig. 3.

XT-I acceptor specificity. Eight peptides complexed with XT-I are superimposed.³⁹

1.3. Donor specificity of XT-I/II

Unlike the extensive study of acceptor promiscuity, investigations on the donor specificity of XT-I/II are rather limited, and until recently, both xylosyltransferases were considered monofunctional to UDP-xylose. In a study done by the Götting group, various non-native UDP-sugars, including UDP-glucose (UDP-Glc), UDP-galactose (UDP-Gal), UDP-glucuronic acid (UDP-GlcA), and UDP-N-acetyl-glucosamine (UDP-GlcNAc) were examined with a soluble XT-II to test its donor promiscuity.²⁹ However, there were no observable transfers of the non-native sugar to the selected peptide acceptors under testing. It suggests that the donor substrate scope of human XT-II is rather limited and may be restricted to UDP-xylose.

In 2018, Briggs and Hohenester provided an in-depth structural investigation of XT-I with high-resolution crystal structures.³⁹ In the crystal structure of the ternary complex of XT-I with both UDP-xylose and a peptide substrate, the presence of residue W392 in the UDP-xylose binding site restricts the available space around the C-5 of xylose, providing a potential explanation for the donor specificity of XT-I (Fig. 4). This finding further supports the belief that XT-I/II could be monofunctional to UDP-xylose.

Fig. 4.

UDP-xylose binding pocket of XT-I. Residue W392 is found in close proximity to the C-5 of xylose providing a potential rationale for the donor specificity.³⁹

An interesting observation on XT-I donor specificity was reported by the Hendig group in 2015.³⁸ They discovered that XT-I was able to recognize the UDP-4-azido-4-deoxyxylose (UDP-XylAz) and transferred the 4-azido-4-deoxy-xylose to the bikunin-like peptide QEEEGSGGGQKK. In comparison, the glycosylation activity from XT-II using UDP-XylAz was not observed. This is the first reported differentiation of XT-I/II activity, and to the best of our knowledge, the first example showing that XT-I could accept a non-native UDP-sugar as a donor substrate.

Since XT-I could tolerate the azido-modification at the C-4 position, other small alterations on xylose may potentially be accepted by the enzyme. To better understand the donor profile of XT-I, follow-up investigations were carried out by the Huang group recently.⁴⁶ Based on the results, noticeable human XT-I glycosylation activities were observed with non-native UDP-Glc and UDP-6AzGlc sugar donors, using a bikunin-derived peptide as the acceptor. Further studies would be of value to have a more comprehensive understanding on XT-I donor specificity.

1.4. Determinations of XT-I/II activity and product characterizations

Over the past decades, a variety of tools has been developed or applied to determine the XT-I/II activity. Dating back to the 1960s, the Neufeld and Dorfman groups documented the first measurements of the XT-I activity with ¹⁴C radioactive-labeled UDP-xylose sugar donor substrate.^13,18,19 In 2006, Brunner and co-workers applied matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF MS) and reversed-phase high-performance liquid chromatography (RP-HPLC) to analyze products of xylosyltransferase reactions.²⁸ To obtain detailed structural information, electrospray-ionization (ESI) tandem mass spectrometry was applied for the first time to pinpoint the location of the xylose unit.²⁸

To confirm the β-glycosyl linkage between the glycan and the peptide, Götting and co-workers examined the XT-I glycosylated products with linkage-specific cleavage by α- and β-xylosidase and base-promoted release of the glycan from the glycopeptide.²³ The results clearly indicated a β linkage between xylose and serine. This method was later extended to XT-II-catalyzed reactions by Casanova and co-workers.²⁹ In their study, the linkage-specific digestion of the reaction products reveals that XT-II is also a β-xylosyltransferase.

Recently, Briggs and Hohenester utilized a commercial glycosyltransferase kit to quantify the XT-I activity by monitoring the release of UDP from the sugar donor. The luminescence was then measured to correlate the readout with the enzymatic activity.³⁹

Until recently, the involvement of modern nuclear magnetic resonance (NMR) techniques to characterize the product structures had not yet been reported. Earlier this year, with improvements on reaction scale and sample preparation, the stereochemistry of the glycosyl-peptide linkage was determined to be β by NMR experiments by the Huang group.⁴⁶ The application of NMR spectroscopy provides direct evidence on the stereoselectivity of the enzymes.

1.5. Structure-activity relationships

With advances on efficient expression and purification of XT-I, substantial progress on the structure–activity relationships of this important enzyme has been achieved during the past two decades. The high-quality crystal structures of XT-I and its ternary complex with UDP-xylose and peptide acceptors have drastically enhanced the current understanding of how XT-I interacts with the substrates and offered valuable insights on the catalytic mechanism.³⁹

In 2004, Götting et al. first investigated the functions of XT-I DXD motifs with mutants that carried point mutations on the two short segments, ³¹⁴DED³¹⁶ and ⁷⁴⁵DWD⁷⁴⁷.²⁵ Mutations on the first ³¹⁴DED³¹⁶ motif do not affect the XT-I function. In contrast, the D745G mutation abolishes the catalytic function of XT-I, even though the alterations on ⁷⁴⁵DWD⁷⁴⁷ do not strongly affect the donor substrate bindings.

A year later, with 14 mutants carrying individual point mutations of cysteine into alanine, Müller and co-workers investigated the importance of available cysteine residues to XT-I functions.²⁶ In terms of enzymatic activities, mutations on 5 of the 14 cysteine residues resulted in over 90% loss of XT-I function. These findings imply the importance of the five Cys residues to the XT-I activity. Interestingly, alanine replacement of the cysteine residues close to the C-terminus did not exhibit any considerable effects on XT-I catalysis. The treatment of the cysteine-targeting N-phenylmaleimide reagent induced concentration-dependent inhibitions on all enzymatically active cysteine-to-alanine mutants but not the wild-type XT-I. These results indicate that all the 14 cysteine residues may exist in form of cystine, and there are no free thiol groups available in wild-type XT-I. In addition, the enzymatic activity of wildtype XT-I and its single mutants could also be effectively reduced under the treatment of high-dose UDP or GAG. Meanwhile, all the mutants demonstrated comparable binding to immobilized UDP and heparin as the wild-type XT-I. Taken together, it is likely that the cysteine residues present in XT-I do not directly participate in UDP or GAG bindings and mutations on them triggered no drastic conformational changes in the corresponding binding sites.

Shortly after, Müller and co-workers furthered their investigations with a series of N-terminal truncated forms of human XT-I.⁴⁷ According to their results, the first 260 amino acids at the N-terminus of the wild type are not required for the enzymatic activity. However, the XT-I catalytic function would be abolished with an additional deletion of 12 amino acids, G²⁶¹KEAISALSRAK²⁷², from the N-terminus. Since the individual replacement of each non-aliphatic residue in the 12 amino-acid sequence by alanine did not exert substantial influence on the enzyme activity in their study, it was suggested that this motif could be crucial to maintain the proper conformation of the enzyme. Interestingly, the truncation of P⁷²¹KKVFKI⁷²⁷ motif, which is similar to the heparin-binding consensus sequence identified by Cardin and Weintraub,⁴⁸ does not affect the heparin binding of XT-I but dramatically impairs the proper enzymatic function, implying the necessity of this motif to the protein conformation.⁴⁹

Over a decade later, in 2018, Briggs and Hohenester provided an in-depth structural investigation of XT-I with high-resolution crystal structures.³⁹ The structures in complex with UDP-xylose and peptide acceptors offer valuable insights on how the enzyme recognizes and interacts with the substrates. To obtain the ternary complex of XT-I with both UDP-xylose and a peptide substrate, the serine residue originally in the acceptor peptide sequence was replaced by alanine to abolish its acceptor function. The UDP diphosphate moiety of the donor binds with positively charged amino acid residues R598 and K599, instead of a divalent metal ion. The presence of residue W392 in the UDP-xylose binding site restricts the available space around the C-5 of xylose, providing an explanation for the limited donor scope of XT-I (Fig. 5).

Fig. 5.

Active site of XT-I in complex with UDP-xylose donor and a peptide acceptor.³⁹

The crystal structure around the peptide-binding site suggests that the network of hydrogen bonds is not sequence specific. Ten out of the eleven hydrogen bonds between the acceptor peptide and the catalytic domain occur on the carbonyl and amide groups along the peptide backbone. To gain insights into the characteristic C-terminal domain of XT-I (Xylo_C domain), a variety of single mutants was expressed. Results demonstrated that point mutations on the Xylo_C structure in contact with the catalytic GT_A domain did not impede the XT-I enzymatic functions. Briggs and Hohenester suggest that the presence of the Xylo_C domain, instead of being directly required for xylosylation activity, likely facilitates the recruitment of enzymes involved in subsequent GAG biosynthesis.

1.6. Synthetic applications

In a research study recently reported by the Huang group,⁴⁶ human XT-I enzyme was applied to efficiently synthesize structurally diverse glycopeptides for the first time (Fig. 6). These native glycopeptides are derived from heparan sulfate proteoglycan (HSPG) linkage region. Excellent synthetic yields were achieved with the enzyme at the milligram scale, showing a great synthetic potential of XT-I in glycopeptide preparations.

Fig. 6.

Structures of representative peptide acceptors (1–4) transformed by human XT-I to glycopeptide products (5–8) with the serine xylosylation sites highlighted in red.

2. β-1,4-Galactosyltransferase 7

2.1. Expression and purification of β4GalT7

The β4GalT7 enzyme represents the seventh member of human β-1,4-galactosyltransferase family. Its molecular cloning and expression were first achieved by the Clausen group in 1999.⁴⁹ The full-length β-1,4-galactosyltransferase and a truncated version containing amino acid residues 63–327 were prepared using the Sf9 and High Five insect cell expression systems. The purification of β4GalT7 was then accomplished by sequential DEAE/Amberlite and S-Sepharose chromatography.⁵⁰

The Lattard group, in 2009, successfully expressed the membrane form of β4GalT7 in HeLa cells and a soluble maltose-binding protein (MBP)-β4GalT7 fusion protein with an N-terminal truncation in E. coli BL21 cells.⁵¹ The MBP-fused β4GalT7 was purified by an amylose column. The desired protein was eluted out with 20 mM maltose in buffer (20 mM MOPS containing 150mM NaCl at pH 7.0), and it was further dialyzed against the same buffer.

In a research work published by Ramakrishnan and Qasba in 2010, the catalytic domain of Drosophila melanogaster β4GalT7, in its native form or with a variety of modifications, was individually prepared for crystallization studies.⁵² The variants included an enzyme with an 11-amino acid truncation from the C-terminus (Cd7ΔC) and ones carrying additional bovine β4GalT1 peptide fragments at the N-terminus (P-Cd7ΔC and P1-Cd7ΔC).

Since the MBP-β4GalT7 fusion protein produced previously only exhibited modest solubility and was prone to aggregation after the release of MBP fusion partner by protease, in 2010, the Qasba group designed a soluble form of human β4GalT7 using galectin-1 as the fusion partner to facilitate the folding and improve its stability and solubility.⁵³ This fusion form of β4GalT7 was expressed in an E. coli expression system. The initial purification was achieved with an α-lactose column, and the target protein, galectin-1-human-β4GalT7, was eluted with 100 mM lactose. Subsequently, the galectin-1 was cleaved off the protein with the Tobacco Etch Virus (TEV) protease. In this study, another MBP-fusion form of human β4GalT7 plasmid, pmal-2x-hum-β4GalT7, was constructed, and the enzyme, MBP-human-β4GalT7, was expressed effectively in E. coli. The MBP-tag assisted the purification with an amylose column as previously reported.⁵⁴ Factor Xa protease cleaved off the MBP tag. The soluble form of human β4GalT7 was eventually purified with UDP-agarose columns. In a direct comparison of the two MBP-fusion forms, the galectin-1-human-β4GalT7 created exhibited great solubility and was less prone to aggregation, displaying its superior stability. This was the first documented success of galectin-1 as a fusion partner acting as a chaperone for the preparation of human β4GalT7 in E. coli cells.

Meanwhile, in a study reported by Talhaoui and co-workers, HeLa cells or CHO pgsB-618 cells were transfected with either wild-type human β4GalT7 plasmid or single-mutant plasmids, individually, to aid the determination of catalytically active residues. In addition, E. coli BL21(DE3) cells were also used to prepare a soluble GST-fusion form of β4GalT7. Its purification was attained via the GST tag with glutathione-Sepharose 4B packed affinity column.

In 2013, the Qasba group unveiled the crystal structures of Drosophila β4GalT7 and a single mutant D211N β4GalT7 in complex with UDP-galactose as the donor and xylobiose as the acceptor, respectively.⁵⁶ In this study, the plasmid of an N-terminally truncated human β4GalT7 (β4GalT7Δ81) was constructed, and the preparation of this truncated protein was carried out following previously reported conditions.⁵³

The Foumel-Gigleux group, in 2015, constructed multiple vectors for different forms of human β4GalT7 and successfully expressed N-terminus truncated GST-tagged human β4GalT7 (β4GalT7ΔNt60) using E. coli BL21 (DE3) cells.⁵⁷ This is the most recent reports of unique expression of human β4GalT7.

2.2. Acceptor specificity of β4GalT7

The early report on β4GalT7 acceptor specificity studies dates back to 1994.⁵⁸ Esko and co-workers examined the priming of heparan sulfate using a variety of xylosides carrying non-native aglycones. This is the first demonstration that a certain galactosyltransferase accepts xylosides as its substrates to enable heparan sulfate biosynthesis. In the following years, an increasing number of chemically modified xylosides were tested, and the β4GalT7 acceptor scope expanded as investigations continued.^49,59,60

In 2007, a library of thio-xylosides was prepared by the Ellervik group to examine the effect on GAG chain priming.⁶¹ In the study, for the first time, they demonstrated that thio-xylosides could be tolerated by the enzymes for GAG biosynthesis. Shortly after, Abrahamsson et al. assessed the GAG priming capability of various xylosylated naphthoic acid–amino acid conjugates.⁶² Only the most nonpolar analog initiated the GAG biosynthesis in T24 cells. Two years later, Victor and co-workers built a library of metabolically stable click-xylosides with hydrophobic groups attached. Priming activities were observed with this novel group of xylosides using the CHO cell line.⁶³ The in vitro studies unveiled that aglycone moieties of xylosides affected sulfation, GAG chain composition, and length. These results demonstrated that multiple O-, S-, and C-xylosides could be processed by β4GalT7 in vitro.

In a research work published by the Fernandez-Mayoralas and Garcia-Junceda groups in 2011, a collection of decoy xyloside acceptors was chemically synthesized and tested with a recombinant, soluble β4GalT7. This was the first demonstration that recombinandy expressed β4GalT7 is promiscuous in the aglycon moieties of the xylose acceptor.⁶⁴

Three years later, the Ellervik group further explored the substrate promiscuity of the enzyme with a truncated GST-β4GalT7 and chemically modified xyloside analogs.⁶⁵ In contrast to the great tolerance on aglycones, the truncated GST-β4GalT7 failed to process most of the xyloside analogs to any significant extent. Only a few xyloside analogs carrying modifications on the C-2 or C-5 positions were galactosylated. Subsequent molecular modeling revealed that the binding pocket of β4GalT7 is narrow. Xylose, as the optimal substrate, is required to match with the precise set of hydrogen-bond acceptors in the pocket.

In 2015, more in-depth investigations were carried out to gain understandings on acceptor structure requirements.⁶⁶ In this study, xylosides with varied aglycon size, anomeric configuration, linker length and electronic properties were carefully examined and compared. In general, only xylosides with the β-anomeric configuration would be smoothly converted by β4GalT7. The galactosylation capability of substrate can be enhanced by replacing the anomeric oxygen with sulfur. Substituting it with carbon reduces the enzymatic activity. In line with prior findings, bulky aglycons could be accepted.

Recently, a variety of xylosides and xyloside analogs carrying a 2-naphthyl (Nap) or a 4-methylumbelliferyl (MU) aglycone was synthesized by the Ellervik group and the Wagner group, respectively.^67–71 From the assay results, xyloside analog 2-naphthyl β-d-GlcNAc functioned as an acceptor substrate.⁶⁸ And analogs having an endocyclic sulfur atom proved to be great substrates for the enzyme.⁷⁰

2.3. Donor specificity of β4GalT7

In comparison with acceptor specificity, investigations on β4GalT7 donors are limited.^51,55 The first detailed examination on β4GalT7 donor scope was reported in 2009 by the Lattard group. Several non-native UDP-sugars, including UDP-Xyl, UDP-Glc, UDP-Man, UDP-GlcA, UDP-GalNAc, and UDP-GlcNAc, were individually incubated with the purified MBP-β4GalT7. Among them, UDP-Xyl and UDP-Glc were accepted by the enzyme, although with much lower activities with 27- and 11-fold decreases, respectively, as opposed to UDP-Gal.⁵¹

The Fournel-Gigleux group reported similar results a year later.⁵⁵ Using 4-MU xyloside as acceptor, wild-type β4GalT7 was able to process UDP-Xyl and UDP-Glc, even though the observed activity levels were low. The W224H mutant failed to retain the donor promiscuity.

2.4. Determinations of β4GalT7 activity and product characterizations

Back to the 1990s, in cellulo GAG priming with β-d-xylosides was probed using radioactive [³⁵S]SO₄²⁻ and [6-³H] d-glucosamine.⁵⁸ Later, UDP-[¹⁴C]-Gal was used to track the activity of secreted β4GalT7 enzyme.^49,72,73 Almeida and co-workers performed one-dimensional ¹H NMR, two-dimensional ¹H-¹H TOCSY, and ¹³C-decoupled ¹H-¹³C HSQC and HMBC experiments to analyze the product structure in detail. The NMR data confirmed the newly formed Galβ1 → 4Xylβ linkage.⁴⁹ In 2009, the Lattard group applied NMR techniques, including ¹H, ¹³C, HSQC, TOCSY, COSY and NOESY, to thoroughly characterize the reaction products.⁵¹ The significant chemical shift changes on H-4 and C-4, together with a large ³J_H1′,H2′ value, supported the desired β1 → 4 Gal → Xyl linkage.

In 2005, reversed-phase high-performance liquid chromatography (RP-HPLC) equipped with a C₁₈ column was applied for the first time by Gulberti and co-workers to monitor the β4GalT7 reactions.⁷⁴ This analytical method was then optimized and more routinely used to assess β4GalT7 enzymatic activities.^51,64–66 A phosphatase-coupled glycosyltransferase assay, in which a phosphatase is used to convert the released UDP into inorganic phosphate for subsequent colorimetric quantification, was later developed and applied to kinetic studies of β4GalT7.^68,75

Recendy, the electron-transfer/higher energy collision dissociation (EThcD) hybrid fragmentation technique was successfully applied to examine the PG glycopeptide product formed under human β4GalT7 biocatalysis.⁷⁶

2.5. Structure-activity relationships

Pioneering investigations into the β4GalT7 catalytic domain can be traced back to 2010.^52,55 With the first high-resolution crystal structure of Drosophila β4GalT7 catalytic domain resolved, Ramakrishnan and Qasba discovered a new Mn²⁺-binding motif (²⁴¹HXH²⁴³), in addition to the DXD motif common in the β4GalT family.⁵² The molecular docking results show that the O-4 hydroxy group in xylose is expected to form a strong hydrogen bond with the Asp²¹¹ side-chain carboxylate oxygen atom for acceptor activation. The presence of Tyr¹⁷⁷ greatly limits the space in the binding pocket (Fig. 7). The steric hindrance imposed by this bulky residue may explain why β4GalT7 rejects most of the chemically modified xyloside analogs as acceptors.

Fig. 7.

Molecular docking of glucose into the binding pocket of Drosophila β4GalT7.O-2, O-3 and O-4 hydroxy groups of the docked glucose molecule are in close proximity to catalytic residues D211/D212. Residue Y177 imposes steric hindrance on the C-6/O-6 atom of the glucose molecule, implying only xylose would be accommodated by the enzyme.⁵²

In the same year, the Fournel-Gigleux group reported the first detailed SAR investigation of the active site of human β4GalT7.⁵⁵ Canonical motifs ¹⁶³DVD¹⁶⁵ and ²²¹FWGWGEDDE²³⁰ were identified in this enzyme (Fig. 8). D163A or D165A point mutation completely abolished the enzyme activity. In comparison, replacement of D165 with glutamic acid retained, albeit reduced, the human β4GalT7 activity. For the N-terminus of the conserved ²²¹FWGWGEDDE²³⁰ region, F221A mutation may affect the conformation of the acceptor-binding site, as reflected by a 13-fold increase in the K_m value of 4-MU-xylose. Meanwhile, W222F mutation did not show apparent effects on the affinity of either the donor or the acceptor. W224F and G225A mutants failed to demonstrate any observable enzyme activities, while the G223A mutant maintained roughly 40% of the enzyme function. Further investigations suggested that residue W224 plays a critical role in the donor and acceptor substrate binding. For the C-terminus of the peptide region, E227D/E230A did not impact the donor or acceptor binding. In contrast, E227A/D228A/D229A/D229E mutants abolished the catalytic activity.

Fig. 8.

Stereoview of the molecular modeling of human β4GalT7 in complex with UDP-Gal. (A) Predicted complex formed with UDP-Gal, Mn²⁺, and ¹⁶³DVD¹⁶⁵/²⁵⁷HLH²⁵⁹; (B) predicted interaction between β-phosphate of UDP-Gal and residue W224. The protein α-carbon backbone is colored green. Key residues in the active sites, UDP-Gal, and Mn²⁺, are highlighted.⁵⁵

In 2013, the co-crystal structure of Drosophila D211N β4GalT7 mutant in the closed conformation with donor UDP-Gal and acceptor xylobiose was published by Tsutsui and co-workers.⁵⁶ In their study, an additional hydrogen bond is observed between Tyr177 side-chain −OH group and the β-phosphate oxygen atom of the UDP-Gal donor. The catalytic base Asp211 interacts with the O-3 and O-4 atoms of the bound xylose acceptor via hydrogen bonds (Fig. 9). Although the acceptor binding site is hydrophobic due to the presence of Tyr194, Tyr196, Tyr199 and Trp224, its neighboring region is highly positively charged to provide a high affinity to the acidic-residue-rich xylose attachment sites of native proteoglycans.

Fig. 9.

(A) Xylobiose binding to Drosophila β4GalT7 in a closed conformation. The active site is colored green.⁵⁶ (B) Overview of proposed key interactions of xylosides and UDP-Gal in the β4GalT7 binding pocket (reprinted with copyright permission from Elsevier).⁶⁹

The Ellervik group later studied the enzyme–substrate interactions of their synthesized xyloside analogs with Drosophila β4GalT7.⁶⁶ Despite the steric effect imposed by the chemical modifications of the aglycon, O-2, O-3, and O-4 from the xylosides form a hydrogen-bonding network with the catalytic residues N211 and D212 (Fig. 10).

Fig. 10.

D211N β4GalT7 in complex with UDP-Gal, Mn²⁺ and a xyloside analog. The protein is colored blue. UDP-Gal and the xyloside analog are highlighted gray.⁶⁶

Recently, the Fournel-Gigleux group extended the computational analysis to human β4GalT7.⁵⁷ Their docking simulation results identified a hydrophobic region, formed by Tyr194, Tyr196 and Tyr199, that provides stacking interactions with the aglycone and the xylopyranoside sugar ring. The acceptor xyloside is oriented and activated through a hydrogen-bond network with Asp228, Asp229 and Arg226 (Fig. 11).

Fig. 11.

The active site of human β4GalT7 in complex with UDP-Gal, Mn²⁺ and 4-MUX. The protein α-carbon backbone is colored gray. Key residues in the active site and substrates are highlighted.⁵⁷

2.6. Synthetic application

Human β4GalT7 has been applied to glycopeptide synthesis by the Huang group.⁷⁶ In the study, human β4GalT7 efficiently galactosylated multiple xylosylated PG glycopeptides to the desired Gal-Xyl glycopeptide products at the milligram scale (Fig. 12A). Furthermore, the combination of human XT-I and β4GalT7 in one reaction flask enabled one-pot, two-enzyme (OP2E) glycosylations, which gave higher overall yields of the desired glycopeptide bearing Gal-Xyl disaccharides as compared to the more traditional stepwise reactions with XT-I and then β4GalT7 (Fig. 12B). This further improved the synthetic efficiency, and suggested that XT-I and β4GalT7 are compatible with each other in the same reaction flask.^46,76

Fig. 12.

(A) Structures of representative peptide acceptors (6–11) transformed by β4GalT7 to glycopeptide products (12–17) with the serine glycosylation sites highlighted in red. (B) One-pot, two-enzyme reactions with XT-I and β4GalT7 could convert the peptide to a glycopeptide bearing a Gal-Xyl disaccharide in higher yields than those of the stepwise reactions.

3. Future outlook

While significant progress has been made on the key glycosyltransferases involved in proteoglycan linkage region synthesis, application of these biocatalysts is in its infancy. From the perspective of synthesis, deploying the four glycosyl transferases, i.e., XT-I, β4GalT7, β3GalT6, and β3GAT3, may lead to a highly efficient preparation of the PG glycopeptides bearing the full linkage region. Together with well-developed GAG synthesis enzymes,^77,78 it may pave the road toward native homogeneous PG glycopeptides and glycoproteins. A library as such would be highly valuable for in-depth structure–activity relationship investigations. As traditional chemical synthesis can be highly tedious and labor intensive, PG enzymatic synthesis would serve as a disruptive approach to dramatically reduce the time, effort, and materials required to prepare PG compounds, making the process faster, easier, and “greener.”

In addition, enabled by advanced computational technology, biocatalytic enzymes could be re-designed or re-purposed and tailored for specific research needs. Among the four enzymes needed to make the PG linkage, XT-I is a particularly promising target. With its ability to recognize certain binding motifs, a properly engineered XT-I variant could potentially transfer non-native sugars, for instance, an azido-sugar, to a wide range of proteins. The labeled proteins may then be functionalized with a variety of fluorescent probes or affinity tags to support diverse research aspirations. If other enzymes involved in PG linkage assembly could tolerate the chemically modified glycoproteins as their substrates, they would become a highly valuable biocatalytic toolkit to facilitate investigation of the multifaceted biological functions of PGs.