HNK-1 sulfotransferase modulates α-dystroglycan glycosylation by 3-O-sulfation of glucuronic acid on matriglycan

Abstract

Mutations in multiple genes required for proper O-mannosylation of α-dystroglycan are causal for congenital/limb-girdle muscular dystrophies and abnormal brain development in mammals. Previously, we and others further elucidated the functional O-mannose glycan structure that is terminated by matriglycan, [(-GlcA-β3-Xyl-α3-)_n]. This repeating disaccharide serves as a receptor for proteins in the extracellular matrix. Here, we demonstrate in vitro that HNK-1 sulfotransferase (HNK-1ST/carbohydrate sulfotransferase) sulfates terminal glucuronyl residues of matriglycan at the 3-hydroxyl and prevents further matriglycan polymerization by the LARGE1 glycosyltransferase. While α-dystroglycan isolated from mouse heart and kidney is susceptible to exoglycosidase digestion of matriglycan, the functional, lower molecular weight α-dystroglycan detected in brain, where HNK-1ST expression is elevated, is resistant. Removal of the sulfate cap by a sulfatase facilitated dual-glycosidase digestion. Our data strongly support a tissue specific mechanism in which HNK-1ST regulates polymer length by competing with LARGE for the 3-position on the nonreducing GlcA of matriglycan.

Introduction

It has been recently appreciated that proper glycosylation of the cell surface glycoprotein α-dystroglycan (α-DG) is required, as part of the dystrophin–glycoprotein complex, for linking the actin cytoskeleton to components of the extracellular matrix (ECM) in muscle and brain tissue (Wells 2013; Yoshida-Moriguchi and Campbell 2015; Manya and Endo 2017; Sheikh et al. 2017; Hohenester 2018; Kanagawa and Toda 2018). Disease subtypes of congenital muscular dystrophy (CMD) that result in hypoglycosylated α-DG are referred to as secondary or tertiary dystroglycanopathies (Brockington et al. 2010; Beedle et al. 2012; Roscioli et al. 2012; Willer et al. 2012; Riemersma et al. 2015). Thus, a comprehensive understanding of the composition, biosynthesis and regulation of the α-DG functional glycan is essential for identifying and studying the biochemical etiologies of various dystroglycanopathies. Our groups, along with others (Gerin et al. 2016; Kanagawa et al. 2016), have recently contributed to this by identifying a novel phospho-ribitol component and the primer region of the functionally relevant O-mannosyl-initiated α-DG glycan (Praissman and Wells 2014; Willer et al. 2014; Praissman et al. 2016). The phospho-ribitol moieties, recently identified in humans, contribute to bridging a phosphorylated trisaccharide on α-DG to a glycosaminoglycan (GAG)-like polymer, matriglycan [(-GlcA-β3-Xyl-α3-)_n] that binds to certain proteins in the ECM (Figure 1A). A culmination of work from multiple groups has since proposed this functionally relevant Core M3 O-mannose structure on α-DG to be (-GlcA-β3-Xyl-α3-)_n-GlcA-β4-Xyl-β2-ribitol-1-P-ribitol-1-P-3-GalNAc-β3-GlcNAc-β4-(6-P)-Man-O-(Ser/Thr) (Yoshida-Moriguchi et al. 2010; Inamori et al. 2012; Inamori et al. 2013; Yoshida-Moriguchi et al. 2013; Praissman et al. 2014; Willer et al. 2014; Gerin et al. 2016; Kanagawa et al. 2016; Manya et al. 2016; Praissman et al. 2016; Sheikh et al. 2017).

Fig. 1.

Exoglycosidase digestion of matriglycan from different tissues. (A) Fully elaborated M3 glycan showing susceptible sites for exoglycosidase digestion. Glycan symbols and substituents are adapted from the Symbol Nomenclature for Glycans (Varki et al. 2015) and (Sheikh et al. 2017). (B) WGA-enriched extracts from mouse brain, heart and kidney were treated with a β-glucuronidase (bGus) and an α-xylosidase (XylS) simultaneously, followed by immunoblot analysis using an anti-α-DG-core antibody (AF6868). (C) A laminin overlay assay was performed by incubating membranes with Natural Mouse Laminin from Engelbreth-Holm-Swarm (EHS) sarcoma. Myd: The Myodystrophy (Myd) mouse model (Large1^myd/myd) possesses a mutation in the LARGE1 gene, resulting in hypoglycosylated α-DG, and is used as a nondigested negative control (Grewal et al. 2001). The images shown are the representative images of N = 3. This figure is available in black and white in print and in color at Glycobiology online.

Matriglycan (Yoshida-Moriguchi and Campbell 2015) is the laminin globular (LG)-domain binding epitope synthesized by the Golgi-resident bifunctional glycosyltransferase LARGE1 or its paralog LARGE2 (Inamori et al. 2012; Inamori et al. 2013). To date, a precise understanding of how matriglycan chain lengths are directly modulated is not entirely known. However, overexpression of LARGE1 or LARGE2 in cells increases the functional glycosylation and apparent molecular weight of α-DG (Barresi et al. 2004; Patnaik and Stanley 2005; Nakagawa et al. 2012, 2013) and α-DG isolated from various tissues has vastly different and apparent molecular weights (Kunz et al. 2005; Yoshida-Moriguchi and Campbell 2015) (Figure 1B). Conversely, the sulfation of the α-DG functional glycan by the human natural killer-1 sulfotransferase (HNK-1ST/carbohydrate sulfotransferase [CHST10]) has been suggested to play a role in inhibiting extension (Nakagawa et al. 2012, 2013). The Golgi resident HNK-1ST is best known for its involvement in synthesizing the canonical HNK-1 carbohydrate epitope (SO₃-3-GlcA-β3-Gal-β4-GlcNAc-R) on N-linked glycans. HNK-1ST catalyzes the transfer of a sulfo group from 3′-phosphoadenosine 5′-phosphosulfate (PAPS) to carbon-3 of nonreducing end, terminal glucuronic acid that is β3-linked to N-acetyllactosamine (LacNAc) on glycoprotein targets in the brain, such as neural cell adhesion molecule (NCAM), myelin-associated glycoprotein (MAG), tenascin-R and tenascin-C (McGarry et al. 1983; Kruse et al. 1984; Chou et al. 1986; Xiao et al. 1997; Ong et al. 2002; Yagi et al. 2010; Morise et al. 2017). In addition to these well-studied cell adhesion and ECM-binding target molecules, HNK-1ST is capable of generating the HNK-1 epitope on glycolipids (Chou et al. 1986), sulfating O-glycan structures on proteoglycans (Hashiguchi et al. 2011), as well as sulfating estrogen and other glucuronidated steroid hormones contributing to hormone regulation and fertility (Suzuki-Anekoji et al. 2013).

Notably, the canonical HNK-1 epitope that is β2-linked to O-linked mannose (Core M1 and M2-type) was detected on glycopeptides isolated from rabbit brain (Yuen et al. 1997), and more specifically, detected on O-mannose glycans derived from phosphacan, a secreted chondroitin sulfate proteoglycan in the brain (Morise et al. 2014; Dwyer et al. 2015). Interestingly, before complete elucidation of the functionally relevant Core M3 structure, it was previously demonstrated that HNK-1ST sulfates an unknown glycan structure on α-DG that is sensitive to treatment with hydrofluoric acid (Nakagawa et al. 2013), thus presumably after a phosphodiester-linked glycan moiety. Since the LARGE glycosyltransferases possess both xylosyltransferase and glucuronyltransferase activities, it was hypothesized that HNK-1ST may sulfate glucuronic acid (GlcA) residues synthesized by LARGE1 or LARGE2, which has yet to be directly validated (Yoshida-Moriguchi and Campbell 2015).

In this report, we extend our studies of the functional glycan by directly investigating HNK-1ST sulfotransferase activity on LARGE-dependent matriglycan polymerization using purified components, while also probing the tissue-dependent susceptibility of α-DG to exoglycosidase digestion. Our studies demonstrate that HNK-1ST catalyzes 3-O-sulfation of terminal GlcA on matriglycan and prevents further matriglycan extension by LARGE1. Furthermore, exoglycosidase digestion analyses of matriglycan derived from α-DG isolated from various tissues reveals that sulfated α-DG from brain, where HNK-1ST expression is high, is amenable to digestion only upon removal of the sulfate cap by pretreatment using a sulfatase. These results strongly support a tissue specific mechanism in which HNK-1ST and LARGE compete to define the length of matriglycan on α-DG.

Results and discussion

Exoglycosidase digestion of matriglycan from different tissues

Using a recently developed dual exoglycosidase digestion assay (Briggs et al. 2016), we sought to probe the glycosylation status of native α-DG isolated from different mouse tissues. Wheat germ agglutinin (WGA)-enriched extracts from mouse brain, heart and kidney were treated with a β-glucuronidase (bGus) and an α-xylosidase (XylS) simultaneously (Moracci et al. 2000; Salleh et al. 2006) to degrade matriglycan extensions. Western blot analysis of the digestion products using an anti-α-DG-core antibody demonstrates that α-DG from mouse heart and kidney are susceptible to bGus/XylS digestion of matriglycan (Figure 1B). Furthermore, the digestion of heart and kidney α-DG that results in loss of binding to the ECM ligand laminin in an overlay assay (Figure 1C). However, dual-glycosidase treatment of α-DG from brain fails to affect molecular weight or laminin binding (Figure 1), suggesting a possible cap on matriglycan.

HNK-1ST transfers a sulfo group to GlcA-β4-Xyl-β-pNP and prevents extension by LARGE1

Since HNK-1ST (gene symbol CHST10) has been implicated in sulfating a glycan structure on α-DG, we decided to test HNK-1ST’s sulfotransferase activity directly in an isolated system. To validate HNK-1ST’s capability to modify noncanonical HNK-1 epitopes, we synthesized GlcA-β4-Xyl-β-pNP as a putative acceptor substrate using a recombinant secreted form of the β-1,4-glucuronyltransferase B4GAT1 (Praissman et al. 2014) to transfer GlcA to Xyl-β-para-nitrophenyl. This resultant disaccharide, linked to pNP at the reducing terminus, was selected since it is the primer for LARGE1-dependent matriglycan synthesis (Praissman et al. 2014; Willer et al. 2014). As the enzyme source, a secreted form of HNK-1ST, also lacking its transmembrane region, was expressed and purified from HEK293F cells. In a reaction containing PAPS as a sulfo-donor, HNK-1ST catalyzed the transfer of a sulfo group from PAPS to GlcA-β4-Xyl-β-pNP, as detected by ion trap mass spectrometry (ITMS) in negative ion mode (Figure 2A and B). MSⁿ analyses of the reaction product (m/z 526) confirm the structure, SO₃-GlcA-β4-Xyl-β-pNP (Figure 2D). In an attempt to use Xyl-β-pNP as an acceptor, HNK-1ST was unable to generate any detectable sulfated Xyl species (data not shown).

Fig. 2.

HNK-1ST transfers a sulfo group to GlcA-β4-Xyl-β-pNP and prevents extension by LARGE1. GlcA-β4-Xyl-β-pNP acceptor substrate was incubated without (A) or with (B) HNK-1ST and PAPS donor, and the reaction was analyzed by ion trap mass spectrometry (ITMS) in negative ion mode. Annotated peaks were confirmed by MS/MS. (C) Reaction from Panel B was further incubated with UDP-GlcA and UDP-Xyl with LARGE1. (D) MSⁿ analysis of the HNK-1ST product confirms the sulfate being located on GlcA. Spectra shown are the representative of N = 3. This figure is available in black and white in print and in color at Glycobiology online.

To determine if LARGE1 can synthesize matriglycan on a structure terminating in sulfated GlcA, we took the reaction mixture from Figure 2B, which contained both the sulfated disaccharide and unreacted GlcA-β4-Xyl-β-pNP, and performed a subsequent reaction with LARGE1 in the presence of UDP-GlcA and UDP-Xyl as the sugar-donor molecules. Remarkably, LARGE1 had the ability to extend GlcA-β4-Xyl-β-pNP (m/z 446) to a higher order polymer (m/z 754, GlcA-β3-Xyl-α3-GlcA-β4-Xyl-β-pNP and presumably even higher order polymers that were undetected) (Figure 2C), while SO₃-GlcA-β4-Xyl-β-pNP remained unmodified as determined by the increase in its relative abundance with respect to the m/z 446 peak (Figure 2B vs. C). No branched, sulfated structures were detected, consistent with previous reports were consistent that LARGE1 does not possess branching activity (Praissman et al. 2014).

HNK-1ST transfers a sulfo group to nonreducing end GlcA of matriglycan and prevents extension by LARGE1

Additionally, we performed a similar assay using a mixture of matriglycan polymers of varying lengths generated by LARGE1 (Figure 3). As anticipated, HNK-1ST was capable of sulfating any matriglycan structure terminating in GlcA at the nonreducing end (note the reaction product ions at m/z 526, 834 and 1142; Figure 3B, Supplementary Figure S1). These results are consistent with previous reports of HNK-1ST having the ability to modify synthetic acceptor substrates terminating in β-GlcA in vitro (Bakker et al. 1997). MS2 fragmentation of these structures revealed only terminal GlcA sulfation, as no internal GlcA sulfation fragments (data not shown), or species terminating in sulfated Xyl in the MS1 could be detected. As demonstrated with sulfo-GlcA-β4-Xyl-β-pNP, while LARGE1 was capable of extending matriglycan structures terminating in Xyl or nonsulfated GlcA (m/z 446, 578 and 886), LARGE1 was unable to extend structures sulfated by HNK-1ST (m/z 526, 834 and 1142), as judged by their presence and increase in relative abundances after the LARGE1 reaction (Figure 3C). A summary of the expected and observed glycoforms are listed in Table I. Taken together, these results demonstrate direct inhibition of LARGE1-mediated matriglycan polymerization due to sulfation by HNK-1ST.

Fig. 3.

HNK-1ST transfers a sulfo group to nonreducing end GlcA of matriglycan and prevents extension by LARGE1. pNP-matriglycan (heterogeneous mixture) acceptor substrate was incubated without (A) or with (B) HNK-1ST and PAPS donor, and the reaction was analyzed by ITMS in negative ion mode. Annotated peaks were confirmed by MS/MS. (C) Reaction from Panel B was further incubated with UDP-GlcA and UDP-Xyl with LARGE1. Spectra shown are the representative of N = 3. This figure is available in black and white in print and in color at Glycobiology online.

Table I.

Calculated matriglycan polymers. Species shaded in gray were observed in this study

		Calculated mass (Da)	[M–H]− (m/z)	+Sulfate (+80) (m/z)
Repeats	Starting GlcA-β4-Xyl-β-pNP mass →	447	446	526
1	+Xyl (+132)	579	578	658
	+GlcA (+176)	755	754	834
2	+Xyl (+132)	887	886	966
	+GlcA (+176)	1063	1062	1142

Nuclear magnetic resonance analysis of HNK-1ST reaction product demonstrates sulfation at position 3 of GlcA

As mentioned previously, the classical HNK-1 structure terminates in a sulfo group that is 3-O-linked to GlcA attached to LacNAc. To determine if the sulfation position of GlcA on matriglycan catalyzed by HNK-1ST is as generated on the classical HNK-1 structure, we analyzed the HNK-1ST reaction product, SO₃-GlcA-β4-Xyl-β-pNP by nuclear magnetic resonance (NMR). The one-dimensional ¹H-NMR spectrum of unmodified GlcA-β4-Xyl-β-pNP revealed the presence (~4-fold excess) of Xyl-β-pNP starting material, presumably from an incomplete B4GAT1 reaction since the reaction product was not purified away from Xyl-β-pNP starting material (Figure 4A). Comparison of the untreated and sulfated 1D proton and COSY spectra clearly indicate the downfield GlcA-H3 shift due to the sulfate on the 3′hydroxyl group, defining the complete linkage as SO₃-3-GlcA-β4-Xyl-β-pNP (Figure 4, Table II). Thus, HNK-1ST is capable of sulfating matriglycan, a noncanonical substrate, at the 3-position of GlcA, in vitro, which is the same position that LARGE adds a nonreducing Xyl residue to during extension.

Fig. 4.

Nuclear magnetic resonance (NMR) analysis of HNK-1ST reaction product demonstrates sulfation at position 3 of GlcA. (A) GlcA-β4-Xyl-β-pNP starting material was analyzed by COSY ¹H-¹H-NMR and compared to the HNK-1ST-sulfated reaction product (B). This figure is available in black and white in print and in color at Glycobiology online.

Table II.

Summary of observed and literature GlcA/Xyl NMR chemical shifts

	GlcA-Xyl-pNPa		SO3-3-GlcA-Xyl-pNPa		Inamori et al.b		Voshol et al.c
Residue	GlcA	Xyl	SO3-3-GlcA-	Xyl	GlcA	Xyl	SO3-3-GlcA-
1	4.52	5.18	4.63	5.19	4.72	5.38	4.77
2	3.27	3.60	3.47	3.60	3.38	3.74
3	3.45	3.67	4.27	3.68	3.54	3.88	4.33
4	3.48	3.87	3.70	3.88	3.51	3.70
5	3.79	3.55, 4.12	3.93	3.56, 4.14	3.75	3.69, 3.95

^a(SO₃-3) ± GlcA-β1,4-Xyl-β-pNP chemical shifts referenced to residual HDO at 4.7 ppm.

^b Inamori et al. (2012), for terminal disaccharide in GlcA-β1,3-Xyl-α1,3-GlcA-MU, where MU = 4-methylumbelliferyl.

^c Voshol et al. (1996), SO₄-3-GlcA-β1,3-Gal-β1,4-GlcNAc, taken at different temperature.

Sulfation serves as a cap on matriglycan

Since α-DG isolated from mouse brain was resistant to dual glycosidase treatment, we speculated that terminal GlcA sulfation may be inhibiting the glucuronidase activity of bGus, thus blocking subsequent digestion by XylS (Figure 1). To test the effect of sulfation on bGus activity, we digested unsulfated and sulfated GlcA-β4-Xyl-β-pNP with bGus and detected the reaction products by ITMS (Figure 5A). Unsulfated GlcA-β4-Xyl-β-pNP (m/z 446) was efficiently digested as determined by the disappearance of the m/z 446 peak and the increase in relative abundance of the Xyl-β-pNP (m/z 316) peak. However, in a parallel reaction, sulfo-GlcA-β4-Xyl-β-pNP (m/z 526) was resistant to bGus digestion.

Fig. 5.

Sulfation serves as a cap on matriglycan. (A) GlcA-β4-Xyl-β-pNP is amenable to bGus digestion by bGus, while sulfated GlcA-β4-Xyl-β-pNP is not, as detected by ITMS. Spectra shown are representative of N = 3. (B) Sulfatase treatment of brain α-DG facilitates dual glycosidase digestion (bGus + XylS) as detected by immunoblot (IB) analysis using the antibody IIH6 that recognizes matriglycan on α-DG, or anti-α-DG-core (AF6868). (C) IIH6 and anti-α-DG-core IBs of exoglycosidase treated skeletal muscle. Asterisk (*) indicates β-DG. Results shown are the representative of N = 3.

Next, we attempted to remove the suspected sulfate cap on native brain α-DG by use of a commercially available arylsulfatase in conjunction with the dual glycosidase assay used above. This was followed by western blot analysis using the antiglyco-α-DG antibody (IIH6) that recognizes matriglycan (Figure 5B) (Ervasti and Campbell 1993; Yoshida-Moriguchi and Campbell 2015). Sulfatase treatment facilitated the digestion of matriglycan on brain α-DG, owing to the time-dependent reduction of the IIH6 signal (Figure 5B). Sulfatase treatment was not required for the exoglycosidase digestion of matriglycan on skeletal muscle α-DG (Figure 5C). Additionally, the observed band shifts were not caused by sulfatase treatment alone (Supplementary Figure S2). Taken together, sulfated GlcA is resistant to bGus digestion and removal of the sulfate cap facilitates dual glycosidase digestion of brain α-DG.

Transcript abundance of HNK-1ST is greatest in the brain

To further explore why the capping seemed to be prevalent in α-DG isolated from brain, we analyzed the transcripts of major genes involved with the synthesis of the functional Core M3 structure in the various mouse tissues used in this study (Figure 6). While there is some variability in the expression profiles of the selected genes across the different tissues, a striking difference is revealed among the analysis of LARGE1, LARGE2 and HNK-1ST (Figure 6A). As demonstrated previously, LARGE1 is expressed in all tissues analyzed, with greater expression in heart and brain, whereas LARGE2 is preferentially expressed in kidney (Peyrard et al. 1999; Grewal et al. 2005). The relative transcript abundance of HNK-1ST is greatest in the brain, and comparisons of HNK-1ST expression relative to either LARGE2 or LARGE1 reveal a 5- to 10-fold increase in expression in brain, respectively (Figure 6B) (Kleene and Schachner 2004; Kizuka and Oka 2012). These results are consistent with our observation that brain α-DG is of smaller molecular weight and capped with sulfate to a greater degree than in other analyzed tissues.

Fig. 6.

Transcript analysis of genes involved with Core M3 glycan synthesis. (A) qRT-PCR analysis of transcripts involved with Core M3 glycan synthesis (error bars represent SEM). (B) Comparison of HNK-1ST expression levels from C, relative to LARGE1 or LARGE2, normalized to that of skeletal muscle. (C) Proposed model comparing HNK-1ST activities and effects on matriglycan synthesis. Major enzymes involved with the synthesis of the α-DG Core M3 glycan are indicated. This figure is available in black and white in print and in color at Glycobiology online.

Conclusion

In summary, we demonstrated that HNK-1ST possesses in vitro activity toward nonreducing end GlcA-terminated matriglycan and that sulfation of GlcA blocks LARGE-mediated matriglycan extension. Sulfation of matriglycan by HNK-1ST is at the 3-position of GlcA, the same site of αXyl linkage. Thus, the mechanism by which sulfation of GlcA inhibits LARGE-mediated matriglycan extension is by a sulfate group occupying the site for Xyl attachment (Figure 6C). Reminiscent of our model, HNK-1ST sulfation of terminal GlcA in a GAG-linkage tetrasaccharide has also been demonstrated to inhibit chondroitin sulfate synthesis on the proteoglycan thrombomodulin (Hashiguchi et al. 2011; Nakagawa et al. 2011). Outstanding questions include the significance of matriglycan chain length and sulfation on recruitment of specific binding partners as well as any avidity enhancements longer chains may induce due to multivalent interactions. While it is appreciated that longer matriglycan polymers may assist in the structural cohesion of muscle and heart tissue, the function of shorter, sulfated matriglycan in the brain, where HNK-1ST expression is the highest, is unclear. Of the five carboxy-terminal LG-domains in laminin-α chains residing in the ECM, matriglycan interactions have been mapped to the LG4/5 domains (Talts et al. 1999; Tisi et al. 2000; Smirnov et al. 2002; Briggs et al. 2016), whereas direct interactions with the classical HNK-1 epitope occur in the LG2 domain (Hall et al. 1993; Hall et al. 1997; Bhunia et al. 2010). Interestingly, while both skeletal muscle and brain α-DG bind laminin-1, -2 and -8 isoforms, brain α-DG preferentially binds to Laminin 511/521 (previously referred to as Laminin-10/11) (Aumailley et al. 2005; McDearmon et al. 2006). Whether this preferential binding is regulated in part by matriglycan sulfation remains to be investigated. Targeted, HNK-1ST-mediated GlcA sulfation of matriglycan may constitute an additional, yet poorly understood, mechanism of regulating α-DG’s glycans binding to ECM components.

Materials and methods

Protein expression and purification

The Thermotoga maritima bGus was expressed and purified as previously described (Salleh et al. 2006; Briggs et al. 2016). The enzyme’s bGus activity was determined by the hydrolysis of pNP-β-d-GlcA in 100 μL reactions containing various amounts of bGus in 100 mM NaOAc pH 5.5 at 65°C for 30 min. Assays were quenched by the addition of 100 μL of 50 mM NaOH, and the release of pNP was determined spectrophotometrically by measuring absorbance at 405 nm (A₄₀₅).

An amino-terminal His₆-tagged version of the Sulfolobus solfataricus XylS was generated as follows. The full-length coding region for XylS was amplified from pT7SCII-XylS (Kind gift from Dr. Marco Moracci, University of Naples Federico II) (Moracci et al. 2000) by polymerase chain reaction (PCR) using the following primers (start codon indicated in bold and stop codon (reverse complement) underlined):

XylS-Donor-Forward: 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGAGAATAGGGAATTTAAATG and XylS-Donor-Reverse: 5′- GGGGACCACTTTGTACAAGAAAGCTGGGTTCTAACCC. The resultant XylS PCR fragment was cloned into pDONR221 (Invitrogen, Carlsbad, CA) using a Gateway BP Clonase II reaction to generate the entry vector, pDONR221-XylS, which was subsequently cloned into the pDEST17 destination vector (Invitrogen, Carlsbad, CA) using a Gateway LR Clonase II reaction to generate the final vector, pDEST17-HisXylS, which was verified by restriction digest analysis and DNA sequencing. Chemically competent Escherichia coli BL21(DE3) cells were transformed with the pDEST17-HisXylS plasmid via heat shock at 42°C using standard protocols. Transformed cells were grown in LB medium supplemented with 100 μg/mL ampicillin at 37°C shaking at 170 rpm, and cell density was monitored by absorbance at 600 nm (A₆₀₀). Log phase cultures were grown until A₆₀₀ was between 0.4 and 0.6, and protein expression was induced by the addition of isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.2 mM at 16°C and allowed to incubate for 16 h shaking at 170 rpm. Cells were harvested by centrifugation and stored at −80°C until ready for lysis and purification. Cells were then resuspended in lysis buffer (25 mM HEPES-NaOH pH 7.2, 400 mM NaCl, 20 mM Imidazole, 0.3% Triton-X100, 1× Protease Inhibitor Cocktail Set V [EDTA-Free, Millipore Sigma, St. Louis, MO]), probe sonicated for 4 min on ice (cycles of 10 s on and 10 s off), and Pierce™ Universal Nuclease was added to degrade released DNA and RNA to reduce viscosity of the lysate. The lysate was clarified by centrifugation at 17,000 × g for 30 min at 4°C, followed by Heat Fractionation by incubation in a 70°C water bath for 10 min. The lysate was again clarified by centrifugation at 17,000 × g for 30 min at 4°C. The enzyme (His₆-XylS or simply, XylS) was purified by Ni-NTA chromatography at 4°C, eluted with 300 mM imidazole and buffer exchanged into phosphate buffered saline (PBS), pH 7.2. Protein concentration was determined by bicinchoninic acid assay. The average yield of purified His₆-XylS was 1.6 mg per liter of E. coli culture. The enzyme’s XylS activity was determined by the hydrolysis of pNP-α-d-Xyl in 100 μL reactions containing various amounts of His₆-XylS in 100 mM NaOAc pH 5.5 at 65°C for 30 min. Assays were quenched by the addition of 100 μL of 50 mM NaOH, and the release of pNP was determined spectrophotometrically by measuring absorbance at 405 nm (A₄₀₅).

Recombinant expression of soluble, secreted versions of green fluorescent protein (GFP)-B4GAT1 and LARGE1 were expressed and purified as previously described (Praissman et al. 2014). The catalytic domain of human HNK-1ST (Gene symbol CHST10, amino acid residues 28-356, UniProt O43529) (Moremen et al. 2018) was expressed as a soluble, secreted fusion protein (amino-terminal signal sequence, 8 × His-tag, AviTag and ‘superfolder’ GFP followed by a TEV-protease cleavage site) by transient transfection of HEK293F suspension cultures (Moremen et al. 2018). Suspension culture HEK293F cells (Life Technologies, Grand Island, NY) were transfected as previously described (Meng et al. 2013), and the cell culture media was subjected to Ni-NTA chromatography (Millipore Sigma, St. Louis, MO). Enzyme preparations eluted with 300 mM imidazole were concentrated to ~1 mg/mL using an Amicon centrifugal concentrator (Millipore Sigma) with a 10 kDa molecular weight cutoff and buffer exchanged into PBS pH 7.2.

WGA-enrichment from tissues

Wildtype and Large1^myd/myd mouse brain, heart and kidney tissues were dissected and snap frozen in liquid nitrogen. Enrichment of α-DG from these tissues was formed as previously described (Combs and Ervasti 2005; Briggs et al. 2016). Briefly, each tissue sample was minced and homogenized in 50 mM Tris (pH 7.5), 300 mM NaCl, 1% Triton X-100 with protease inhibitors. After clarification by centrifugation, the lysate was applied to WGA agarose beads. The WGA beads were then washed, and bound proteins were eluted with 50 mM Tris (pH 7.5), 300 mM NaCl, 0.1% Triton X-100 and 300 mM GlcNAc. Elution fractions were incubated at 99°C for 5 min in the presence of 10 mM 2-mercaptoethanol, followed by buffer-exchanging into 150 mM sodium acetate pH 5.5, and concentration to a final volume of 200 μL.

Exoglycosidase and sulfatase assays

Exoglycosidase reactions with bGus and XylS were performed as previously described (Briggs et al. 2016). Lyophilized arylsulfatase from Helix pomatia (Product #S9626) was purchased from Sigma-Aldrich (St. Louis, MO) and resuspended in 0.2% (w/v) NaCl to a stock concentration of 0.2 units/μL of sulfatase. For each sample, bGus and XylS were added and allowed to incubate at 65°C for 6 h. The pH was neutralized with Tris buffer before loading onto 3–15% polyacrylamide gels for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting.

For the combined sulfatase/exoglycosidase treatment, α-DG was isolated from the rabbit brain and buffer exchanged into 150 mM sodium acetate pH 5.5 as described above. To ~1 mL of buffer exchanged eluate, 8 units of sulfatase was added and incubated for 24 h at 65°C. Approximately 750 μL of the α-DG sample pretreated with sulfatase was repurified using 100 μL of WGA-agarose beads at 4°C for 1 h, followed by extensive washing with 150 mM sodium acetate pH 5.5 and elution with 150 mM sodium acetate pH 5.5 and 300 mM GlcNAc. The bGus and XylS were added and allowed to incubate at 65°C for 16 h, taking aliquots for analysis at 0, 1, 2, 4 and 16 h. Samples were analyzed by SDS-PAGE and immunoblotting using mAb IIH6 or anticore α-DG (AF6868) as described below. Similarly processed samples from rabbit skeletal muscle were digested with bGus and XylS for 23 h, taking aliquots for analysis at 0, 4, 8, 11, 14 and 23 h.

Immunoblotting and laminin overlay assay

Following SDS-PAGE, proteins were transferred to PVDF-FL (Millipore), and probed with various antibodies as follows: The anti-α-DG core primary antibody (Goat 20 AP, 1:100 dilution; Jung et al. 1996) was detected by secondary antibody donkey anti-goat IgG IR800CW (1:4000, Li-Cor). The anti-glyco α-DG mAb IIH6 (1:100 Dilution, Developmental Studies Hybridoma Bank, University of Iowa; Ervasti and Campbell 1991) was detected by secondary antibody goat anti-mouse IgM IR800CW (1:4000, Li-Cor). Laminin overlay assays were performed as previously described (Michele et al. 2002). Natural mouse laminin from the Engelbreth-Holm-Swarm (EHS) sarcoma was purchased from Invitrogen (23017-015). The rabbit polyclonal anti-laminin antibody (Sigma L9393, 1:1000 dilution) was detected by secondary antibody donkey anti-rabbit IgG IR800CW (1:4000, Li-Cor). All immunoblots were imaged using a Li-Cor Odyssey scanner.

Synthesis of GlcA-β4-Xyl-β-pNP and pNP-matriglycan

The HNK-1ST acceptor substrate, GlcA-β4-Xyl-β-pNP, was generated in a 500 μL reaction containing 16 mM Xyl-β-pNP, 100 mM MOPS pH 7.0, 10 mM MnCl₂, 10 mM MgCl₂, 20 mM UDP-GlcA, 0.02 U/μL CIAP and 6 μM GFP-B4GAT1. The reaction was allowed to incubate at 37°C for 24 h. Reaction products were purified using C18 MicroSpin Columns (The Nest Group, Southborough, MA) by applying and washing in Buffer A (0.1% formic acid) and elution with Buffer B (0.1% formic acid, 80% acetonitrile). Samples were dried down using a SpeedVac and resuspended in MilliQ water. Aliquots of reaction products, diluted in 50% methanol were analyzed by Direct Infusion into an ITMS (Velos Pro, ThermoFisher Scientific, San Jose, CA) using nanospray ionization in negative ion mode.

pNP-matriglycan was generated in a 300 μL reaction containing 500 μM GlcA-β4-Xyl-β-pNP, 100 mM MES pH 6.0, 10 mM MnCl₂, 10 mM MgCl₂, 1 mM UDP-GlcA, 1 mM UDP-Xyl, 0.01 U/μL CIAP and 1 μM GFP-LARGE1. The reaction was allowed to incubate at 37°C for 66 h. Reaction products were purified using C18 MicroSpin Columns (The Nest Group) by applying and washing in Buffer A (0.1% formic acid) and elution with Buffer B (0.1% formic acid, 80% acetonitrile). Samples were dried down using a SpeedVac and resuspended in MilliQ water. Aliquots of reaction products, diluted in 50% methanol were analyzed by Direct Infusion into an ITMS (Velos Pro, ThermoFisher Scientific, San Jose, CA) using nanospray ionization in negative ion mode.

HNK-1ST assays

HNK-1ST reactions were performed in 50 mM Tris-HCl pH 7.4, 50 μM GlcA-β4-Xyl-β-pNP (or pNP-matriglycan), 100 μM PAPS (Santa Cruz, Dallas, TX) and 0.2 μM GFP-HNK-1ST. The reaction was allowed to incubate at 37°C for 17 h. Reaction products were purified using C18 MicroSpin Columns (The Nest Group) by applying and washing in Buffer A (0.1% formic acid) and elution with Buffer B (0.1% formic acid, 80% acetonitrile). Samples were dried down using a SpeedVac and resuspended in 50% methanol. Reactions were analyzed by Direct Infusion into an ITMS (Velos Pro, ThermoFisher Scientific, San Jose, CA) using nanospray ionization in negative ion mode. For extension of matriglycan by GFP-LARGE1, After sulfation with HNK-1ST, reaction products were dried down using a SpeedVac and resuspended in a cocktail containing 100 mM MES pH 6.0, 10 mM MnCl₂, 10 mM MgCl₂, 1 mM UDP-GlcA, 1 mM UDP-Xyl and 1 μM GFP-LARGE1. The reaction was allowed to incubate at 37°C for 16 h. Reaction products were purified using C18 MicroSpin Columns as described above and analyzed by ITMS as described above.

NMR analysis

For NMR analysis, ~0.4 mg of either GlcA-β4-Xyl-β-pNP or SO₃-GlcA-β4-Xyl-β-pNP was prepared as described above. Each sample was resuspended in 99% D₂O (Cambridge Isotope Laboratories, Inc., Tewksbury, MA) and analyzed on an Agilent 900 MHz DD2 spectrometer equipped with a 5-mm cryogenically cooled probe. After stabilization and shimming, NMR experiments were performed at 25°C using standard pulse sequences (PRESAT, gCOSY) from the Agilent library. Two-dimensional data were collected with default values. Spectra were processed with MestReNova software (Mestrelab Research SL). A summary of observed and literature monosaccharide NMR chemical shifts is presented in Table II.

RNA isolation and transcript analysis

Total RNA was isolated from flash frozen skeletal muscle, kidney, heart and brain tissues using the RNeasy kit (Qiagen) following the tissue extraction protocol except that the tissues were homogenized in RLT buffer containing 3% β-mercaptoethanol. Methods for cDNA synthesis and qRT-PCR were described previously (Nairn et al. 2010). Briefly, validated gene specific primer pairs (Table III) were used for technical triplicate reactions with diluted cDNA and iQ™ SYBR® Green Supermix in a RealPlex² MasterCycler (Eppendorf). Relative transcript abundance values for each gene were calculated by normalizing to the housekeeping gene, β-Actin and the SEM from two biological replicates of each tissue were calculated.

Table III.

Gene specific primer pairs used in transcript analysis

Gene name	Forward primer (5′–3′)	Reverse primer (5′–3′)	GenBank accession #
POMT1	CTACCTGCCCTTCTTCCTGA	AGCAGGATCTGGAAGGTGAG	NM_145145
POMT2	CTTAAACGGCCTGCTTGGAGC	AAGGACAGCAGCGTCACCAC	NM_153415
POMGNT1	ACAGCTGAGGATCCAGCACT	TTCCTGAGCACCCATCCTAG	NM_026651
POMGNT2	TGACTGAAAGGCTGAACGTG	AGTCTGTTCTGGGTGCGA	NM_001289558
POMK	GGTGGGCTACTGTGAGGAAG	CTTCCAGGTTGCTCAAGGAA	NM_029037
FKTN	AGCTGACCACTCATGCCATCC	CGTAGGTGACCATGCCAGAGG	NM_139309
FKRP	CTAACAACTACCGCCGCTTC	GTACTCCGGGTTCTCGATGA	NM_173430
RXYLT1 (TMEM5)	TTGCTTGGAAATGAGCACTG	CAGATCCACAAAGCCTCCAT	NM_153059
B4GAT1	GGAGAGCTTGCTGAGACCTG	CCTTTCCTCCAGCCACATAA	NM_175383
LARGE1	GCTCACCACCGTGGGAACC	TGCCAGCTCGCAGATTCTCAC	NM_010687
LARGE2	GAGCCTGGCCTTGTACCTGAC	AGAGAGCACTGGCGACGTTTC	NM_172670
HNK-1ST (CHST10)	GATTCGAGCCTTGGTACAGG	GTCCGGTTCTTCCGGTACTT	NM_145142
β-ACTIN	GATCGGTGGCTCCATCCTGG	GCCGGACTCATCGTACTCCTG	NM_007393

Abbreviations

B3GAT1

β-1,3-glucuronyltransferase 1 (GlcAT-P)

B3GAT2

β-1,3-glucuronyltransferase 2(GlcAT-S)

bGus

Thermotoga maritima β-glucuronidase

CMD

congenital muscular dystrophy

COSY

homonuclear correlation spectroscopy

ECM

extracellular matrix

GAG

glycosaminoglycan

GlcA

glucuronic acid

HNK-1ST

HNK-1 sulfotransferase (CHST10)

ITMS

ion trap mass spectrometry

LacNAc

N-acetyl-d-lactosamine

LARGE1

like-acetylglucosaminyltransferase 1

LARGE2

like-acetylglucosaminyltransferase 2

LG

laminin globular

MAG

myelin-associated glycoprotein

NCAM

neural cell adhesion molecule

PAPS

3′-phosphoadenosine 5′-phosphosulfate

pNP

para-nitrophenyl

UDP

uridine diphosphate

WGA

wheat germ agglutinin

Xyl

xylose

XylS

sulfolobus solfataricus α-xylosidase

α-DG

α-dystroglycan