Structural and mechanistic characterization of bifunctional heparan sulfate N-deacetylase-N-sulfotransferase 1

Mycroft-West, Courtney J.; Abdelkarim, Sahar; Duyvesteyn, Helen M. E.; Gandhi, Neha S.; Skidmore, Mark A.; Owens, Raymond J.; Wu, Liang

doi:10.1038/s41467-024-45419-4

Download PDF

Article
Open access
Published: 13 February 2024

Structural and mechanistic characterization of bifunctional heparan sulfate N-deacetylase-N-sulfotransferase 1

Nature Communications volume 15, Article number: 1326 (2024) Cite this article

1652 Accesses
4 Altmetric
Metrics details

Subjects

Abstract

Heparan sulfate (HS) polysaccharides are major constituents of the extracellular matrix, which are involved in myriad structural and signaling processes. Mature HS polysaccharides contain complex, non-templated patterns of sulfation and epimerization, which mediate interactions with diverse protein partners. Complex HS modifications form around initial clusters of glucosamine-N-sulfate (GlcNS) on nascent polysaccharide chains, but the mechanistic basis underpinning incorporation of GlcNS itself into HS remains unclear. Here, we determine cryo-electron microscopy structures of human N-deacetylase-N-sulfotransferase (NDST)1, the bifunctional enzyme primarily responsible for initial GlcNS modification of HS. Our structures reveal the architecture of both NDST1 deacetylase and sulfotransferase catalytic domains, alongside a non-catalytic N-terminal domain. The two catalytic domains of NDST1 adopt a distinct back-to-back topology that limits direct cooperativity. Binding analyses, aided by activity-modulating nanobodies, suggest that anchoring of the substrate at the sulfotransferase domain initiates the NDST1 catalytic cycle, providing a plausible mechanism for cooperativity despite spatial domain separation. Our data shed light on key determinants of NDST1 activity, and describe tools to probe NDST1 function in vitro and in vivo.

Luciferase- and HaloTag-based reporter assays to measure small-molecule-induced degradation pathway in living cells

Article 18 April 2024

Local environment in biomolecular condensates modulates enzymatic activity across length scales

Article Open access 18 April 2024

Proteome-scale discovery of protein degradation and stabilization effectors

Article 20 March 2024

Introduction

Heparan sulfate (HS) is a ubiquitous and evolutionarily ancient class of polysaccharide produced throughout the metazoan lineage¹. HS polysaccharides are abundant within the extracellular matrix (ECM) of multicellular organisms in the form of HS proteoglycans (HSPGs), wherein one or more polysaccharide chains are covalently linked to a membrane bound or pericellular core protein². HSPGs in the ECM interact with myriad partners, and are essential regulators of physiological processes including growth factor signaling³, morphogen patterning⁴, adhesion⁵, endocytosis⁶, barrier filtration⁷ and host-pathogen binding^8,9,10.

HS is a member of the glycosaminoglycan (GAG) family—linear polysaccharides typically comprised of alternating hexosamine and uronic acid monosaccharides (heparin, HS, chondroitin sulfate, dermatan sulfate and hyaluronan), whilst keratan sulfate is comprised of alternating hexosamine and galactose units. The hexosamine sugars of HS are predominantly N-acetyl- or N-sulfo-glucosamine (GlcNAc, GlcNS respectively), and the uronic acids of HS are glucuronic or iduronic acid (GlcA, IdoA). Each monosaccharide unit of HS can be further elaborated by variable O-sulfation (Supplementary Fig. 1a)¹¹, giving rise to diverse polysaccharide compositions that enable binding to multiple partners¹². HS sequence complexity is crucial for its biological function, and is tightly regulated by cells in response to developmental cues, physiological status and external stimuli¹³.

HS biosynthesis begins with polymerization to produce unmodified [GlcNAc-GlcA]_n chains (hereafter heparosan) by the EXT family of glycosyltransferases^14,15,16. Complexity is introduced by a series of modification reactions upon the growing heparosan chain (Supplementary Fig. 1b), creating a modular polysaccharide structure in which NS domains, containing high proportions of IdoA and sulfated sugars, are flanked by NA/NS transition domains of intermediate modification, and unaltered NA domains that resemble unmodified heparosan. The biological functions of HS are dictated by the presentation of highly modified NS domains, or adjoining NS/NA regions, which predominantly participate in protein interactions².

Bifunctional N-deacetylase-N-sulfotransferase (NDST) enzymes are the first to act on nascent heparosan after its polymerization, converting GlcNAc sugars to GlcNS, via deacetylated glucosamine (GlcN) intermediates (Supplementary Fig. 1c). HS modification by NDSTs does not occur stochastically, but instead produces discrete GlcNS clusters of variable length^17,18. Notably, these GlcNS clusters are themselves recognized by downstream enzymes, which work to generate further HS modifications, ultimately producing the highly sulfated and epimerized sequences characteristic of NS domains. The activity of NDST enzymes thus plays a central role in determining the overall locations and extents of modification within HS polysaccharide chains (Fig. 1A)^13,19,20,21.

Humans express 4 members of the NDST family (NDST1–4). NDST1 and NDST2 are systemically distributed, with NDST1 dominant in most tissues, and NDST2 dominant in mast cells, which produce the highly modified HS analogue heparin^22,23. NDST3 and NDST4 show more limited distribution, with both isoforms abundant in the brain^24,25,26. Commensurate with their essential role in HS biosynthesis, dysregulation or mutations of NDSTs are associated with a spectrum of pathologies. Total loss of Ndst1 produces perinatal lethality in mice due to severe cerebral and craniofacial defects²⁷, with impaired lung development also noted^28,29. Partially inactivating NDST1 mutations are also linked to autosomal recessive intellectual disability in humans³⁰, whilst changes to NDST3 expression have been implicated in schizophrenia and bipolar disorder³¹, and loss of NDST4 has been suggested to be a prognostic marker for adverse colorectal cancers³².

Despite substantial interest, the molecular basis of NDST activity remains poorly understood. An X-ray crystal structure of the NDST1 sulfotransferase domain was reported in 1999 (PDB 1NST)³³, providing some insight into the basis of sulfate transfer from 5’-phosphoadenosine-3’-phosphosulfate (PAPS) to deacetylated GlcN substrates. However, little is known about NDST deacetylase activity, and consequently, about how the deacetylase and sulfotransferase activities of NDSTs are coordinated. Biochemical experiments have demonstrated reduced GlcNS formation and no GlcNS clustering when heparosan oligosaccharides are modified by a mixture of truncated NDST1 deacetylase and sulfotransferase domains, strongly implying functional cooperativity in the full-length enzyme¹⁸. However, a lack of structural information has precluded analysis of the basis for such functional coupling to date.

Here, we present high-resolution cryogenic-electron microscopy (cryo-EM) structures of the major human NDST isoform NDST1. Structural characterization, aided by activity modulating nanobodies, reveal important determinants of NDST1 catalytic function, including active site architectures, and key residues and loops involved in substrate processing. In contrast to the known bifunctionality of NDST1, our structures clearly reveal a 3-domain enzyme, with the sulfotransferase and deacetylase domains flanked by a non-catalytic N-terminal domain. Counterintuitively, we find that the deacetylase and sulfotransferase catalytic domains of NDST1 project in opposing directions, limiting possible mechanisms of coordination. Analysis of NDST1-HS binding suggests that initial anchoring interactions likely occur at the sulfotransferase domain, despite deacetylation taking place first during the NDST1 catalytic cycle. Based on these observations, we propose a model for NDST1 bifunctionality that can operate even in the absence of direct domain-to-domain transfer.

Results

Production and characterization of recombinant NDST1

NDST1 is a Golgi lumen resident type-II transmembrane protein, which contains a single pass N-terminal helix that anchors the catalytic luminal portions of the enzyme to the Golgi membrane. To facilitate structural and biochemical characterization, we generated solubilized NDST1 truncated at P78/S79, removing the N-terminal helix and a linker region (Fig. 1B). Solubilized recombinant NDST1 was expressed using a baculoviral expression system and purified to homogeneity via metal ion affinity, anion exchange, and size-exclusion chromatography.

We tested the in vitro enzymatic activity of solubilized NDST1 using a recently described coupled enzyme system, in which NDST1 activity is linked to the rat sulfotransferase Sult1A1, which catalytically regenerates sulfate donor PAPS from fluorogenic 4-methylumbelliferyl (4-MU) sulfate³⁴ (Supplementary Fig. 2a, b). As deacetylation by NDST1 is typically rate limiting²⁵, depending on substrate concentrations, this coupled enzyme assay primarily probes this step of turnover. Initial activity trials were carried out using Mg²⁺ supplemented in the assay buffer, to satisfy the requirement of NDST1 catalysis on divalent cations, which are necessary for deacetylation. Surprisingly, we also noted robust activity in the absence of divalent supplementation, which was inhibited by treatment with the chelators EDTA or dipicolinic acid (DPA) (Supplementary Fig. 2c). Inductively coupled plasma optical emission spectroscopy (ICP-OES) revealed that this was the result of residual NDST1 associated cations, primarily Ca²⁺ with trace Zn²⁺, carried over from protein purification. Further supplementation with Ca²⁺ did not improve activity, indicating that carryover was stoichiometric (Supplementary Fig. 2c, d). Prior analysis by Dou et al. suggests that NDST1 deacetylation is compatible with a range of metals, including Mg²⁺, Mn²⁺, Ca²⁺ and Co²⁺, with highest activity found in the presence of Ca²⁺¹⁸. We thus assigned the most likely identity of the metal cofactor in our NDST1 samples to be Ca²⁺.

Pseudo-first order activity constants were determined for NDST1 with respect to PAPS (k_cat = 0.09 min⁻¹; K_M = 3.4 µM), as well as sulfate acceptors HS (k_cat = 0.21 min⁻¹; K_M = 0.5 mg mL⁻¹) and E. Coli K5 – a capsular polysaccharide that shares the same repeating disaccharide motif as heparosan³⁵ (k_cat = 0.14 min⁻¹; K_M = 0.8 mg mL⁻¹; Fig. 1C). NDST1 has previously been proposed to act upon 8-mer stretches of heparosan (MW 1519.3 Da)¹⁸, thus these K_M constants for HS and K5 approximate to ~329 and ~527 µM respectively.

To further confirm the activity of solubilized NDST1, we followed the conversion of GlcNAc to GlcNS in K5 polysaccharide by disaccharide analysis, employing exhaustive digestion by heparin lyase II from P. heparinus. NDST1 treatment of K5 resulted in time-dependent loss of ΔUA-GlcNAc peaks, matched to increasing ΔUA-GlcNS, clearly demonstrating enzymatic activity in a bona fide polysaccharide context (Fig. 1D).

Activity modulating nanobodies for NDST1

To aid mechanistic investigations of NDST1, we generated de novo nanobody (nAb) binders, which can provide insight by trapping conformational states relevant to function. A panel of NDST1 binding nAbs were isolated by primary immunization of a llama host with recombinant solubilized protein, followed by phage display bio-panning of the resulting immunized V_HH library over two rounds³⁶.

Binding analysis using surface plasmon resonance (SPR) and biolayer interferometry (BLI) confirmed 5 NDST1 interacting nAbs. Binding affinities (K_Ds) spanned the high nanomolar to the low micromolar range, in the rank order nAb7 > nAb13 ≈ nAb5 ≈ nAb6 ≫ nAb1, with rapid on/off kinetics in all cases (Fig. 2A, Supplementary Fig. 3, Table 1). Additive SPR signals indicative of co-binding were detected for most pairwise combinations of nAbs, except nAb5 + nAb7, nAb5 + nAb13 and nAb7 + nAb13, which displayed mutually exclusive binding (Supplementary Fig. 4).

**Fig. 2: Biophysical characterization of anti-NDST1 nAbs.**

Table 1 NDST1 binding and activity modulation by nAbs, as measured by SPR, BLI and enzyme kinetics

Full size table

We also assessed nAbs for their ability to modulate NDST1 activity using the coupled enzyme assay. NDST1 inhibition was observed in the presence of nAb5 and nAb7, suggesting that these nAbs bind regions relevant for catalysis. Unexpectedly, we also observed an approximately 2-fold enhancement of NDST1 activity upon nAb13 binding, indicative of allosteric modulation that enhances catalytic turnover (Fig. 2B, Table 1). Differential scanning fluorimetry showed ~2.6 °C, ~4.4 °C and ~1.9 °C stabilization of NDST1 thermal denaturation in the presence of 10.8 µM nAb5, nAb7 and nAb13 respectively, broadly in line with their relative binding affinities for the enzyme (Supplementary Fig. 5).

Cryo-EM structure of bifunctional NDST1

To better understand the molecular basis for NDST1 activity, and how it is modulated by nAb binding, we sought to resolve three-dimensional structures of NDST1 and its complexes with nAb7 and nAb13 (Supplementary Fig. 6). A cryo-EM volume of NDST1 alone was calculated to a nominal resolution of 2.70 Å (Relion Au-FSC = 0.143), with the resulting map of sufficient quality to model nearly all residues along the protein chain, using the AlphaFold2 model as a starting point for refinement (Fig. 3A; Supplementary Fig. 7; Supplementary Table 1).

**Fig. 3: Three-dimensional structure of NDST1.**

NDST1 adopts an elbow-shaped monomeric structure comprised of 3 clearly resolved domains, in contrast to its widespread annotation as a bifunctional protein (Fig. 3B; compare Fig. 1B). Disulfide bonds between C586–C601 and C818–C828 were clearly visible in the cryo-EM density, with the former at lower occupancy, possibly due to its surface location being susceptible to reduction by 1 mM DTT, present in the sample buffer (Supplementary Fig. 8). Proteins produced by baculoviral expression can sometimes differ in glycosylation compared to mammalian expression³⁷. However, we were unable to resolve distinct glycoforms in our NDST1 sample by cryo-EM. Density consistent with the first GlcNAc of an N-glycan tree could be seen at the predicted N-glycosylation site N401, but this was not of sufficient quality to model (Supplementary Fig. 9).

The previously solved NDST1 sulfotransferase domain resides at the C-terminus of the protein (residues D602–R882), and our cryo-EM model agrees well with the crystal structure (PDB accession 1NST; RMSD 0.96 Å over 277 Cαs; Supplementary Fig. 10a)³³, displaying a well-defined binding cleft capable of accommodating HS polysaccharides (Fig. 3C). We observed unambiguous density within the sulfotransferase active site corresponding to a molecule of 5’-phosphoadenosine-3’-phosphate (PAP; present in the cryo-EM sample), which binds NDST1 via H-bonds and salt bridges to K614–T618, R782, K833, R835, Y837, R782 and W817, as well as a π-stacking interaction from adenine to F816 (Fig. 3D). The C828–D841 loop at the mouth of the PAP site partially occludes the ligand from solvent, suggesting a role as a stabilizing lid that encloses PAP(S) after its engagement with the sulfotransferase domain (Supplementary Fig. 10b). The positions of both PAP and the C828–D841 loop closely match the previously solved crystal structure.

Deacetylase activity was assigned to the central domain of NDST1 (L308–C601), based on close structural homology to the fungal galactosaminogalactan deacetylase Agd3 from Aspergillus fumigatus (PDB accession 6NWZ; 21.5% identity to NDST1 residues 1-601; Supplementary Fig. 11a, b)³⁸. The deacetylase domain of NDST1 adopts a (β/α)₇ barrel fold with a clear HS binding cleft (Fig. 3C), and contains a His-His-Asp triad characteristic of metal-dependent deacetylases³⁹. Clear density, distinct from surrounding protein sidechains, was visible at the center of this triad, consistent with Ca²⁺ coordinated to H389, H393, D320, although a full coordination sphere could not be resolved, possibly due to local disorder within the open catalytic cleft (Fig. 3E). Further density within the deacetylase active site was tentatively modeled as a second Ca²⁺, consistent with the presence of multiple divalent ions in the active site of Agd3 and other deacetylases^38,40 (Supplementary Fig. 11c). To confirm the role of the deacetylase domain active site residues, we tested the activity of H389A, H393A and D320A mutants, as well as H529A and D319A – two nearby residues that likely function as general acid/base during deacetylation (see below). In all cases, mutation to alanine abrogated NDST1 turnover with minimal disruption to protein structure, demonstrating key roles for these sidechains in enzymatic processing (Fig. 3F; Supplementary Fig. 12).

To our knowledge, the NDST1 N-terminal domain (NTD; residues 79–307) has not been previously reported. The NDST1 NTD adopts a Rossman-like fold, comprising a central layer of β-sheets flanked by α-helices on each side, although disorder prevented modeling of loops K171–S179, S191–D197, V215–W225 and K242–L262 (Fig. 3C). Interestingly, a similar non-catalytic domain in Agd3 was proposed to function as a carbohydrate-binding module, which extends the span of the substrate binding cleft³⁸. Whilst the analogous cleft in NDST1 is occluded by the Q95–S106 helix (Supplementary Fig. 11d), inspection of the NDST1 surface does reveal a channel formed at the deacetylase and N-terminal domain interface, which extends the deacetylase cleft via a sharp turn (Supplementary Fig. 11e). This channel may guide long HS polysaccharides out of the deacetylase active site, and implicates a role for the NDST1 NTD in controlling enzyme-substrate interactions.

Structures of NDST1-nAb complexes

Complexes of NDST1 with activity-modulating nanobodies nAb7 and nAb13 were isolated by size-exclusion chromatography, and the resulting purified complexes analyzed by cryo-EM (Supplementary Fig. 6).

Initial reconstructions of NDST1-nAb7 showed substantial disorder, consistent with unresolved heterogeneity, which was revealed by 3D variability analysis to correspond to a hinging motion around the NDST1 elbow (Supplementary Movie 1). Further local refinements were thus carried out, focussed around either side of the hinge, yielding 2.42 Å partial maps (Au-FSC = 0.143) that were combined for model building purposes (Supplementary Fig. 7; Supplementary Table 1). The resulting composite map was of sufficient quality to model both NDST1 and nAb7 in detail, including the amino acid sidechains present at their interface.

Examination of the NDST1-nAb7 complex revealed nAb binding at the NDST1 deacetylase domain, across its substrate cleft, providing a clear steric rationale for inhibition (Fig. 4A, B). nAb7 binding was predominantly mediated through its CDR3 loop, with H-bonds between D593_NDST1–Y102_nAb7, K465_NDST1–Q113_nAb7, R590_NDST1–R99_nAb7, K589_NDST1–Y58_nAb7 and salt bridges between D593_NDST1–K101_nAb7, E459_NDST1–R99_nAb7, K465_NDST1–D99_nAb7 and K465_NDST1–D115_nAb7 (Fig. 4C; Supplementary Fig. 13a). The overall structure of NDST1 complexed to nAb7 was similar to that of free NDST1, except for a slightly more closed conformation of the NDST1 elbow, consistent with conformational flexibility around its hinge (Supplementary Fig. 14; Supplementary Movie 1). We also observed improved ordering of the D319–V332, C486–G512 and T528–L538 helical loops, likely due to interactions with the proximal nAb7 (Supplementary Fig. 15).

**Fig. 4: Three-dimensional structures of NDST1 in complex with nAbs.**

The NDST1-nAb13 volume was reconstructed following the same local refinement strategy as NDST1-nAb7, owing to a similar hinging movement (Supplementary Movie 2; Supplementary Fig. 7; Supplementary Table 1). Consistent with its weaker binding, density for the nAb13 portion of the reconstructed map appeared disordered, and could only be modeled by rigid-body fitting of an all Cα nAb chain, derived from the nAb7 structure (Fig. 4D).

nAb13 binds toward the side of the deacetylase domain, in a region that does not obstruct the substrate cleft (Fig. 4E). The rigid-body fitted nAb13 chain was found to lie within 6 Å of NDST1 residues K331, E333, K336 and R537, indicating these amino acids may be involved in binding (Supplementary Fig. 13b). As with nAb7, the nAb13 complex displayed a more closed NDST1 conformation, as well as improved ordering of the D319–V332, C486–G512 and T528–L538 loops (Supplementary Fig. 14, 15). No overlap was observed between the nAb7 and nAb13 binding sites, suggesting their inability to mutually bind NDST1 arises from allosteric effects (Supplementary Fig. 4, 16).

We hypothesized that enhancement of NDST1 activity by nAb13 may be caused by ordering of the D319–V332, C486–G512 and T528–L538 loops, which together form one face of the deacetylase cleft, and likely contact HS during catalytic processing (Supplementary Figs. 15, 17). Pre-organization of this interface by nAb13 would remove a substantial entropic penalty to HS engagement at the deacetylase domain, thereby enhancing the typically rate-limiting NDST1 deacetylation step¹⁸. Accordingly, Michaelis-Menten kinetics for NDST1-nAb13 with respect to K5 showed ~86% increased k_cat compared to NDST1 alone (0.26 min⁻¹ vs 0.14 min⁻¹), indicating improved turnover efficiency. Conversely, NDST1-nAb7 displayed a ~3x reduced k_cat/K_M compared to NDST1 alone (0.06 min⁻¹ mg⁻¹ mL vs 0.18 min⁻¹ mg⁻¹ mL), primarily driven by increased K_M, consistent with competitive inhibition of the deacetylase domain (Fig. 4F).

NDST1-HS interactions analyzed by molecular docking

Despite extensive trials, we were unable to resolve an experimental NDST1-HS complex by cryo-EM. We thus employed molecular docking to gain insights into NDST1 interactions with HS substrates, using GlycoTorch Vina⁴¹ to model the binding of an unsulfated octasaccharide [GlcNAc-GlcA]₄ (hereafter NS0) or di-N-sulfated octasaccharide [GlcNAc-GlcA]₂-[GlcNS-GlcA]₂ (NS2) to each catalytic domain of the nAb free NDST1 model.

Both NS0 and NS2 docked well into the NDST1 deacetylase cleft, placing their nonreducing termini towards the NTD, close to the channel formed at the deacetylase/NTD interface (Fig. 5A; Supplementary Fig. 11e). The resulting contact surface incorporated residues from both the NDST1 deacetylase and N-terminal domains (Supplementary Fig. 17a). Notably, the second GlcNAc (from reducing end) of each octasaccharide was found to dock with its N-acetyl moiety close to a catalytic metal center, in a position well-placed for deacetylation (Fig. 5B). Whilst waters were not explicitly considered during docking, this ligand pose supports a mechanism (based on Agd3 homology) in which a metal coordinated water is deprotonated by the nearby sidechain of base D319, inducing nucleophilic attack upon the GlcNAc N-acetyl center. Breakdown of the resulting oxyanion intermediate would be aided by the nearby acid H529, leading to the formation of GlcN with concomitant release of acetate. This mechanism is consistent with the inactivity of NDST1 deacetylase mutants (Fig. 3F), as well as previously proposed mechanisms for metal-dependent deacetylases^38,39.

**Fig. 5: Computational docking of octasaccharides in active site clefts of NDST1.**

Docking in the NDST1 sulfotransferase domain showed both NS0 and NS2 closely tracking the active site cleft (Fig. 5C; Supplementary Fig. 17b). The trajectory of both octasaccharides was similar to that observed for HS complexed with O-sulfotransferases HS3ST1 and HS3ST3 (PDB accessions 3UAN and 6XL8 respectively; Supplementary Fig. 17c), highlighting the conserved binding modes of these homologous enzymes. Although we did not observe close approach of docked NS0 and NS2 to the bound PAP ligand (Fig. 5D; ~9.5 Å between the PAP 5’ phosphorous and the closest GlcNAc nitrogen), it is plausible that HS chains containing deacetylated GlcN would be able to approach closer to facilitate sulfate transfer from a sulfated PAPS donor.

Sulfotransferase domain interactions initiate HS substrate engagement

Coordination between the deacetylase and sulfotransferase activities of NDST1 is essential for efficient HS modification and GlcNS cluster formation^14,16. Unexpectedly, we found that the active sites of the NDST1 deacetylase and sulfotransferase domains project in opposing directions in the bifunctional enzyme, effectively hindering direct transfer of deacetylated HS intermediates from one domain to another (Fig. 6A; Supplementary Fig. 17d).

**Fig. 6: NDST1 domain organization and substrate binding analysis.**

To determine possible mechanisms involved in domain cooperativity, we used BLI to measure the interaction of full-length NDST1 to surface immobilized K5 polysaccharide. The binding of NDST1 to K5 was well-modelled by a biphasic on/off model, with K_Ds of 234 nM and 679 nM respectively (Fig. 6B; Table 2). Whilst biphasic interactions are not necessarily unusual for a multifunctional enzyme, we were intrigued by the substantial difference between the nanomolar K_Ds measured for direct NDST1-K5 binding, compared to the high-micromolar K_Ms determined for enzymatic NDST1-K5 processing (Fig. 1C). Although K_D and K_M are not directly analogous, both constants quantitate interactions between proteins and ligands. A substantial difference between these values for NDST1-K5 suggested the presence of additional binding events that might take place alongside enzymatic processing.

Table 2 NDST1 binding affinities for K5 with full length NDST1 protein, in the absence and presence of deacetylase domain binders nAb7 or nAb13, and truncated NDST1 sulfotransferase domain. Binding constants calculated from on/off rates from global curve fitting across 6 protein concentrations

Full size table

HS-protein interfaces are typically characterized by charge-based interactions between anionic polysaccharides and cationic surfaces on proteins. Nascent heparosan polysaccharides lack sulfates, but still possess considerable negative charge due to their repetitive carboxylate moieties. We thus assessed the binding of NDST1 to K5 polysaccharide in the presence of increasing concentrations of NaCl, to disrupt charge-based interactions. NaCl concentrations above ~300 mM essentially abrogated all binding, consistent with an electrostatic interface between NDST1 and HS, and in line with functional reports that NaCl strongly inhibits NDST1 catalysis^42,43 (Fig. 6C).

To identify regions on NDST1 that may bind HS via charge, we computed an electrostatic surface for NDST1 using the nAb-free structure, which revealed substantial cationic potential around the sulfotransferase cleft, contrasting with neutral to net-anionic potential at the deacetylase cleft (Fig. 6D). Whilst such charge-based considerations imply initial HS binding at the sulfotransferase domain, deacetylation is known to occur first during the NDST1 catalytic cycle (Supplementary Fig. 1c). We thus reassessed NDST1-K5 binding in the presence of 10 µM nAb7 or nAb13 (≫10x K_D for both nAbs), which bind to different epitopes of the NDST1 deacetylase domain, with nAb7 also binding in a mode that occludes the polysaccharide substrate. As with full length NDST1, nanomolar K5 affinities were measured for both NDST1-nAb7 and NDST1-nAb13, thereby ruling out the NDST1 deacetylase cleft as a site of initial substrate binding (Fig. 6E; Table 2). Finally, we tested the binding of truncated NDST1 sulfotransferase domain alone to K5 polysaccharide. The NDST1 sulfotransferase domain displayed essentially equivalent nanomolar affinity for K5 as compared to full-length enzyme, further reinforcing the primacy of this domain in driving NDST1 polysaccharide interactions (Fig. 6F; Table 2).

Taken together, our data indicate that initial, high-affinity NDST1-HS binding likely takes place at the sulfotransferase domain, despite deacetylation being the first step of the enzymatic processing cycle.

Discussion

Mammalian HS biosynthesis is a complex multistep process, which involves the coordinated activity of multiple enzymes along the ER-Golgi pathway. NDSTs are key enzymes in the early stages of HS maturation, which convert GlcNAc sugars in nascent heparosan to GlcNS. Clusters of GlcNS laid down by NDSTs play a major role in directing further downstream HS processing enzymes, thus understanding NDST function fundamentally informs upon how cells create the complex and heterogenous HS sequences required by biology.

We have carried out a detailed structural and functional characterization of NDST1, the dominant human NDST isoform, revealing a 3-domain architecture consisting of a non-catalytic NTD, in addition to catalytic deacetylase and sulfotransferase domains. The previously unsolved deacetylase domain of NDST1 is characterized by a His-His-Asp triad typical of metal-dependent sugar deacetylases. Herein, we have assigned the presence of catalytic Ca²⁺ coordinated by this triad, based on its high affinity carry through from protein purification. However, NDST1 has also been shown to be active in the presence of other divalent ions, including Mn²+, Mg²+ and Co²⁺¹⁸. It is possible that native NDST1 within the Golgi apparatus, which is enriched in many divalent cations⁴⁴, may employ a variety of metal ions to facilitate deacetylation.

Computational docking with HS octasaccharides suggests that the hitherto undescribed NDST1 NTD may play a role in polysaccharide engagement, by contributing to the formation of a substrate binding cleft adjacent to the deacetylase domain. The presence of carbohydrate binding module (CBM) domains that assist substrate binding is common amongst polysaccharide processing enzymes⁴⁵. Indeed, the fungal deacetylase Agd3, which is homologous to the NDST1 deacetylase domain, possesses an N-terminal CBM that enhances substrate attachment³⁸. Whilst this N-terminal domain of Agd3 extends a linear substrate cleft projecting from its catalytic site, the cleft formed by the analogous NDST1 NTD is kinked with respect to the deacetylase domain (Supplementary Fig. 11e), potentially reducing its effectiveness in aiding HS binding. Accordingly, we find that initial NDST1 engagement of HS is driven by interactions at its sulfotransferase, rather than deacetylase domain, despite the latter operating first in the catalytic cycle. A further role of the NDST1 NTD may be in enzyme oligomerization. Vallet et al. have recently reported the intriguing observation of stable NDST1 homodimers⁴⁶, which interact via a C2 symmetric interface involving the NTD and sulfotransferase domains (Supplementary Fig. 18). NDST1 oligomerization may be mechanistically significant during catalytic turnover (see below), thus implicating the NTD in functions beyond simple substrate binding.

A distinctive feature of the NDST1 protein architecture is the opposing orientation of its catalytic deacetylase and sulfotransferase domains, which effectively precludes a simple substrate transfer mechanism for bifunctionality. Our data indicate that initial NDST1-HS binding occurs at the sulfotransferase domain via an electrostatic interface, which occurs with higher affinity than that required for enzymatic conversion. This high-affinity interaction may help NDST1 enzymes accumulate on heparosan polysaccharides prior to processing. Initial sulfotransferase binding also provides a ready mechanism for bifunctionality, which obviates the need for direct domain-to-domain substrate transfer. Putatively, deacetylation of HS polysaccharides anchored to the NDST1 sulfotransferase domain would produce GlcN intermediates that remain bound to the enzyme. Such intermediates, already located at the NDST1 sulfotransferase domain, would consequently be ideally positioned for immediate sulfation to produce GlcNS. Since initial NDST1-HS interactions are predominantly electrostatic, they may also bind newly formed GlcNS preferentially, further raising the intriguing possibility of a molecular ratcheting mechanism that can lead to the formation of GlcNS clusters (Supplementary Fig. 19). Consistent with this, we note that NDST1 processing of HS, which contains some N-sulfation, shows improved K_M and k_cat compared to its processing of K5, which entirely lacks N-sulfation (Fig. 1C).

The precise nature of HS coordination between the NDST1 deacetylase and sulfotransferase domains remains to be elucidated. In the absence of direct visualization of NDST1-HS binding, two models may be considered for how a polysaccharide chain can access both NDST1 catalytic sites in a coordinated fashion. A 1:1 model, in which a single HS chain loops around NDST1 to access both catalytic domains, appears somewhat implausible, given the entropic penalty that would arise from constraining extended HS polysaccharides (Supplementary Fig. 17, 19). Alternatively, a 2:1 (or >2:1) model, in which the catalytic cycle is shared between multiple NDST1 enzymes, might operate. Such a model accords well with recent observations of catalytically competent NDST1 homodimers by Vallet et al., in which an alternative arrangement of NDST1 catalytic domains is created by the dimeric interface, although the active sites remain some 60 Å apart (Supplementary Fig. 18)⁴⁶. Formation of NDST1 oligomers may provide a route to direct HS chains, initially anchored to the sulfotransferase domain of one protomer, to the deacetylase domain of a neighbor, thereby enabling functional coupling in a manner distinct from the 1:1 model (Supplementary Fig. 19). Whilst stable NDST1 oligomers were not observed in this work, possibly due to differences in construct design (6xHis-tagged vs phosphatase-tagged) or expression system (insect vs mammalian) compared to Vallet et al., transient oligomerization may still operate in our system to coordinate NDST1 activities during catalysis. Interestingly, low (~7.3) µM K_Ms were measured by Vallet et al. for processing of heparosan by partially dimeric NDST1, contrasting with the high µM values recorded here (Fig. 1C), and consistent with improved functionality for a dimer (albeit caution must be exercised when comparing values measured using different assays). We also do not discount the possibility of further oligomeric NDST1 states within intracellular milieu, formed by the preorganization of enzymes on Golgi membranes. Detailed analysis of NDST1-HS interactions across multiple polysaccharide sequences and NDST1 oligomerization states will be required to dissect the consequences of different substrate-binding modes and stoichiometries on catalysis.

One final aspect of NDST1 function that has not been addressed here is the formation of free GlcN residues, which typically comprise ~0.7–4% of HS sequences⁴⁷. These rare GlcN residues may arise from NDST1-mediated deacetylation of heparosan that occurs without initial sulfotransferase domain binding. Alternatively, stalling of the NDST1 catalytic cycle at the sulfotransferase step (e.g. due to lack of PAPS), may also cause premature release of deacetylated intermediate. Both possibilities are likely to be rare, given the weaker HS binding capacity of NDST1 deacetylase domains (Fig. 1C), and the abundance of sulfation reactions within cells requiring a stable pool of PAPS, respectively. Such mechanistic rarity is consistent with the paucity of GlcN in mature HS.

Important to our characterization of NDST1 was the de novo generation of a panel of nAb binders, which aided our analysis of NDST1-HS interactions, and shed light on allosteric factors that influence NDST1 catalytic activity. Beyond their clear utility for functional analysis, these nAbs provide a potential toolkit for functional studies of NDST1 biology e.g. as probes to track enzymes in situ⁴⁸. Notably, we have isolated nAbs that both inhibit (nAb7), or enhance (nAb13) NDST1 activity. Given the paucity of high-quality tools to perturbate HS biosynthesis in cells, nAbs may represent useful, genetically encodable modules to up- or downregulate HS modifications in a reversible manner. In situ use may require higher binding affinities than the high-nM potencies currently displayed by nAb7 and nAb13. Thus affinity maturation steps will likely form part of future efforts to develop HS modulating nAb binders⁴⁹.

With our characterization of bifunctional NDST1, all enzyme families involved in the mammalian HS biosynthesis pathway (linker biosynthesis, polymerization, N-sulfation, epimerization, 2O-, 3O- and 6O-sulfation) now have at least one experimentally resolved representative^{14,50,51,52,53,54,55,56}. The GAGosome hypothesis, first proposed over 20 years ago, posits the existence of a large macromolecular assembly of biosynthesis enzymes tasked with constructing HS in a coordinated fashion¹⁷. Although some HS biosynthesis enzyme interactions have been inferred by coimmunoprecipitation or immunofluorescence, the only multimers to have been directly visualized remain those between the obligate dimeric EXT enzymes^14,15,16. Our observation of initial high-affinity binding interactions between NDST1 and HS hints at an alternative model for the GAGosome, in which HS itself may induce the formation of complexes that transiently assemble around newly synthesized polysaccharide chains. Detailed analysis of individual HS biosynthesis enzymes such as NDST1 provides a vital basis for understanding mechanisms of cooperativity within complex multicomponent systems such as the GAGosome, and will underpin future studies of HS biosynthesis within in situ cellular contexts.

Methods

Cloning and gene expression—NDST1

The pDONR221-NDST1 plasmid was obtained from the DNASU repository (Arizona State University), from which a cDNA fragment encoding solubilized NDST1 (residues 79–882) was amplified using Q5 high fidelity DNA polymerase (New England Biolabs). The NDST1(79–882) fragment was subcloned using Infusion (Takara Bio) into the pOMNIBac transfer vector (Geneva Biotech), behind a honeybee melittin secretion peptide, 6xHis tag and TEV cleavage site. NDST1 mutants were derived from the wild-type NDST1(79–882) construct by site-directed mutagenesis using a PCR-based method⁵⁷. C-terminal Avitagged NDST1 was generated by PCR with tailed primers, followed by subcloning into pOMNIBac. All constructs were verified by Sanger sequencing before further use. DNA and protein sequences for NDST1 constructs are available in Supplementary Table 2. Primers used for cloning, subcloning and mutagenesis are outlined in Supplementary Table 3.

Recombinant bacmids were produced via the Tn7 transposition method in DH10EMBacY cells (Geneva Biotech), and purified using the PureLink miniprep kit (Invitrogen) according to manufacturers’ protocols. V1 baculovirus stocks were produced by transfection of bacmid into low-passage ExpiSf9 cells (Invitrogen; Spodoptera frugiperda) in adherent format using FuGENE HD (Promega) at a ratio of ~2 mg bacmid DNA to ~5 µL FuGENE. V1 to V2 virus amplification was carried out in ExpiSf9 cells in suspension culture, at a density of ~1–2 × 10⁶mL⁻¹. Viral propagation was tracked using the YFP marker encoded by the EMBacY virus. Optimum baculovirus amplification was typically achieved 72 h after inoculation, when cells appeared 60 + % fluorescent. V2 virus stocks were harvested by centrifugation at 200g for 5 min at 4 °C to remove cells, and supplemented with 2% v/v FBS for storage.

Large scale gene expressions were carried out in High-Five (Invitrogen; Trichoplusia ni) or ExpiSf9 cells in log-phase growth. V2 baculovirus was added to cells at a multiplicity of infection > 1, and the resulting infection tracked using the EMBacY YFP marker. Cultures were typically harvested at ~72 h, or when cells showed ~80% or greater fluorescence.

Protein purification—NDST1

Solubilized recombinant NDST1 was purified from conditioned insect cell media. Typically, ~2–3 L of insect cell culture was cleared of cells by centrifugation at 200g for 15 min at 4 °C, followed by further clearing of debris by centrifugation at 5000g for 60 min at 4 °C. Clarified media was supplemented with DL-dithiothreitol (DTT; 1 mM) and phenylmethylsulfonyl fluoride (PMSF; 0.2 mM), before loading onto a 5 mL Histrap Excel column (Cytiva), pre-equilibrated with Histrap buffer A (20 mM Tris pH 8.0, 500 mM NaCl, 20 mM Imidazole, 1 mM DTT), at a rate of 5 mL min⁻¹. The loaded Histrap column was washed with 10 column volumes (CV) Histrap buffer A, before eluting with a linear gradient of Histrap buffer A to Histrap buffer B (20 mM Tris pH 8.0, 500 mM NaCl, 500 mM Imidazole, 1 mM DTT) over 10 CV. NDST1 containing fractions, as determined by SDS-PAGE, were pooled, diluted 10-fold in milli-Q water, and adjusted to pH 6.5 by the addition of 1 M MES acid, before loading onto a Hitrap SP XL column (Cytiva) pre-equilibrated with ion exchange (IEX) buffer A (20 mM MES pH 6.5, 1 mM DTT). Loaded Hitrap SP XL columns were washed with 10 column volumes (CV) IEX buffer A, before eluting with a linear gradient of IEX buffer A to IEX buffer B (20 mM MES pH 6.5, 1500 mM NaCl, 1 mM DTT) over 10 CV. NDST1-containing fractions were pooled again and digested overnight at room temperature with TEV protease at 1:100 mass ratio TEV:NDST1 to cleave the 6xHis tag. TEV digested NDST1 was rerun over a 5 mL Histrap Excel column pre-equilibrated with Histrap buffer A, which was then washed with 4 CV Histrap buffer A. Flowthrough and wash fractions were pooled and concentrated to 2 mL volume using a 30 kDa molecular weight cut-off (MWCO) Vivaspin concentrator (Cytiva). Final purification was carried out by size-exclusion chromatography (SEC), using a HiLoad 16/600 Superdex 200 pg column (Cytiva), pre-equilibrated and run in SEC buffer (20 mM MES pH 6.5, 200 mM NaCl, 1 mM DTT). Pure NDST1 containing fractions were pooled, adjusted to <100 mM NaCl using IEX buffer A, and concentrated to ~10 mg/mL using a 30 kDa MWCO Vivaspin concentrator. Small aliquots were flash frozen using liquid nitrogen, and stored at −80 °C for further use.

Mutant NDST1 constructs were purified in the same way as for wild type protein. Biotinylated NDST1 was prepared using the C-terminal Avitagged protein, which was treated with the BirA biotin-protein ligase reaction kit (Avidity, LLC) following the manufacturers protocol. Biotinylated NDST1 was purified by a final round of SEC following the BirA reaction.

Cloning and gene expression – Sult1a1

A plasmid encoding cDNA for Sult1a1 from Rattus norvegicus was obtained from GenScript Biotech. A cDNA fragment encoding for wild type Sult1a1 was subcloned into the pET15b backbone, behind a thrombin cleavage site, using Ndel and BamHI restriction sites. Mutations K65E and R68G, which prevent the formation of an inhibitory Sult1a1-PAP-4-methylumbelliferone complex⁵⁸, were introduced into the plasmid by site-directed mutagenesis. Primers used for cloning and mutagenesis are outlined in Supplementary Table 3. The final construct was verified by Sanger sequencing before further use.

For expression, Sult1a1-K65E-R68G was used to transform BL21 Gold (DE3) chemically competent cells (Agilent) by heat shock. Transformed cells were grown in TB media containing 100 µg mL⁻¹ ampicillin at 37 °C with shaking, until reaching an OD₆₀₀ of 0.8–1.0. Cultures were induced by the addition of isopropyl-b-thiogalactoside (IPTG; Generon) to a final concentration of 0.5 mM, then grown overnight at 16 °C with shaking. Cultures were harvested by centrifugation at 5000g for 15 min at 4 °C, and pellets stored at –80 °C until further processing.

Protein purification—SULT1A1-K65E-R68G

Sult1a1-K65E-R68G pellets were thawed and resuspended in 100 mM Tris pH 7.6, 1 mM DTT containing 120 kU DNase I (Sigma) and cOmplete™ EDTA-free Protease Inhibitor Cocktail (Roche). Resuspended cell pellets were lysed via two rounds of cell disruption at 30 kpsi (Constant Systems) at 4 °C. Cell debris was removed by centrifugation at 14,000g at 4 °C for 30 min, and supernatant filtered prior to loading onto a 5 mL Histrap FF column, pre-equilibrated in Histrap buffer A (100 mM Tris pH 7.6, 300 mM NaCl, 20 mM imidazole, 1 mM DTT). The loaded column was washed with 5 CV of buffer A prior to elution of Sult1a1-K65E-R68G using a linear gradient of 0–100% Histrap buffer B (100 mM Tris pH 7.6, 300 mM NaCl, 250 mM imidazole, 1 mM DTT) over 10 CV. Sult1a1-K65E-R68G containing fractions were pooled and digested overnight at room temperature with thrombin at 1:100 mass ratio thrombin:Sult1a1-K65E-R68G. Thrombin-cleaved Sult1a1-K65E-R68G was passed over a 5 mL Histrap FF column pre-equilibrated in buffer A, and the column was washed with 5 CV of the same buffer. The flow-through and wash fractions, containing de-tagged Sult1a1-K65E-R68G, were collected and pooled before being concentrated using a 10 kDa MWCO spin filter. Sult1a1-K65E-R68G was further purified by SEC using a HiLoad 16/600 Superdex 75 pg column pre-equilibrated in SEC buffer (100 mM Tris pH 7.6, 1 mM DTT). Fractions containing Sult1a1-K65E-R68G were assessed by SDS-PAGE and Coomassie staining before being pooled and concentrated to 6 mg mL⁻¹ using a 10 kDa MWCO spin filter. Aliquots of concentrated protein were flash frozen in LN₂ and stored at −80 °C until further use.

Expression—heparin lyase II

The gene encoding heparin lyase II from Pedobacterium heparinus, inserted into pRSET-A (gifted from Marcelo Lima, University of Keele), was transformed into chemically competent C41(DE3) pLysS cells by heat shock. 1 L cultures of TB media, containing 100 µg mL⁻¹ ampicillin and 50 µg mL⁻¹ chloramphenicol were inoculated and grown at 37 °C, 250 rpm with shaking until reaching an OD₆₀₀ of 0.6–0.8, whereupon they were induced with 0.5 mM IPTG before further incubation at 22 °C for 16 h, 250 rpm with shaking. Cultures were harvested at 6000 g for 15 min at 4 °C. Pellets were stored at –80 °C until further use.

Protein purification—heparin lyase II

Thawed cell pellets were resuspended in 20 mM sodium phosphate pH 7.9, 500 mM NaCl, 5 mM imidazole supplemented with DNAse I (Sigma) and cOmplete™ EDTA-free protease inhibitor (Roche). Cells were lysed via two rounds of cell disruption at 30 kpsi, then centrifuged at 14,000g, 4 °C, for 30 min to remove cell debris. The supernatant was filtered and loaded onto a 5 mL HisTrap FF column pre-equilibrated in Histrap buffer A (20 mM sodium phosphate pH 7.9, 500 mM NaCl, 5 mM imidazole), which was washed with 5 CV of the same buffer prior to elution with a linear gradient of 0–100% Histrap buffer B (20 mM sodium phosphate pH 7.9, 500 mM NaCl, 250 mM imidazole) over 10 CV. Fractions containing heparinase lyase II were pooled, diluted with 10 volumes of IEX buffer A (20 mM MES pH 6.0) and then loaded onto a 5 mL Hitrap SP XL column (Cytiva) pre-equilibrated in the same buffer. Elution was conducted with a linear gradient of 0–100% IEX buffer B (20 mM MES pH 6.0, 1.5 M NaCl) over 20 CV. Fractions containing heparin lyase II were pooled and concentrated using a 30 KDa MWCO spin filter, then applied to a HiLoad 16/600 Superdex 200 pg column equilibrated and run with SEC buffer (20 mM sodium phosphate pH 6.8, 150 mM NaCl). Heparin lyase II containing fractions were confirmed by SDS-PAGE, before being pooled and concentrated to ~2 mg/mL using a 30 kDa MWCO spin filter. Purified protein was stored at 4 °C until required.

Circular dichroism

The CD spectrum of 250 μg mL⁻¹ NDST1, and mutants thereof, were acquired in 20 mM phosphate buffer pH 7.4 using a Chirascan V100 (Applied Photophysics), equipped with a 0.2-mm path length quartz cuvette (Hellma, United States). All spectra were recorded with a scanning speed of 1 nm s⁻¹ with 0.5 nm resolution, between 190 and 260 nm using Chirascan control software version 2.02. Spectra are presented as the mean of five independent scans smoothed with second-order polynomial smoothing through 11 neighbors, using GraphPad Prism 10.1.0.

Inductively coupled plasma optical emission spectrometry (ICP-OES)

150 µg mL⁻¹ NDST1 was digested by the addition of 0.25 volumes of HNO₃ in a total volume of 4 mL. Transferrin or apotransferrin (440 µg mL⁻¹) controls were digested with the addition of 0.25 volumes of HNO₃ and diluted two-fold across 5 concentrations in 1.25% HNO₃. Following digestion, each sample was filtered using a 0.25 mm PVDF filter prior to measurement of Ca (λ_em = 422.673 nm), Cu (λ_em = 324.754 nm), Fe (λ_em = 260.709 nm), Mg (λ_em = 279.553 nm), Mn (λ_em = 279.482 nm) and Zn (λ_em = 206.2 nm) using an Agilent Technologies 700 Series ICP-OES. ARISTAR® multi-element quality control standard (VWR), containing metal ions at a concentration of 1 mg mL⁻¹ each, was serially diluted 2-fold across 5 concentrations covering the range of 0.0625 − 1 ppm to allow the construction of a calibration curve for each element. Suitable wavelengths were chosen where interferences were not observed, and a sufficient signal:noise ratio was observed above baseline. All samples were obtained as a mean of 3 replicates under a plasma argon flow rate of 15 L min⁻¹, auxiliary argon flow rate of 1.5 L min⁻¹ and a nebulizer argon flow rate of 0.75 L min⁻¹.

Purification of K5 polysaccharide

K5 polysaccharide was purified from E. coli strain Bi 8337/41(O10:K5:H4). Typically, 1 L of LB media was inoculated with 10 mL of E. coli strain Bi 8337/41(O10:K5:H4), grown for 16 h at 37 °C with shaking, before being incubated for a further 55 h at 37 °C. Cells were removed via centrifugation at 8000g for 15 min at 4 °C and the supernatant collected. The supernatant was filtered and diluted with 1 volume of IEX buffer A (20 mM sodium acetate pH 4.0, 50 mM NaCl), further adjusted to pH 4.0 with acetic acid as required, before being loaded onto a C10/20 column (Cytiva) packed with DEAE Sepharose resin (Biorad), pre-equilibrated with the same buffer. Following loading, the column was washed with IEX buffer A and bound K5 polysaccharide eluted in a single step with IEX buffer B (20 mM sodium acetate pH 4.0, 1 M NaCl). The eluant was mixed with 3 volumes of ice-cold absolute ethanol and stored overnight at –20 °C. The resulting precipitate containing K5 polysaccharide was collected by centrifugation at 13,000g, 4 °C for 30 min and resuspended in dH₂O before being lyophilized. Lyophilized K5 was dissolved in 1 M NaCl to 15 mg mL⁻¹ and pH adjusted to 9.5 with 1 M NaOH. 30% hydrogen peroxide was added to a final concentration of 1.5%, before incubation overnight at room temperature. Bleached K5 polysaccharide was recovered by the addition of 3 volumes of ice-cold ethanol followed by storage at −20 °C overnight. The precipitate was collected by centrifugation at 13,000g at 4 °C for 30 min and dialyzed against dH₂O, exchanged a minimum of three times using a 5 kDa MWCO dialysis membrane (Biodesign). Dialyzed K5 polysaccharide was filtered and lyophilized before being stored at −20 °C for further use. The purity of K5 was assessed using nuclear magnetic resonance (NMR). Polysaccharide was exchanged into D₂O (700 µL) and one-dimensional (¹H) spectra were recorded using a Bruker Avance NEO 600 MHz spectrometer fitted with a BBOH&F cryoprobe at 298 K.

Fluorometric assay for determination of Sult1a1-K65E-R68G kinetics

An 8-point 2-fold serial dilution of PAP from 250 µM was incubated with 350 nM Sult1a1-K65E-R68G, 4 mM 4-methylumbellferyl (4MU)-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT, in black half area 96 well plates (Greiner) in a total volume of 20 µL. Controls omitting PAP were included and subtracted from each reaction prior to further analysis. Reactions were initiated by the addition of Sult1a1-K65E-R68G and the change in fluorescence recorded using a Clariostar plate reader (BMG Labtech; software version 5.70 R2) using the 4MU fluorescence preset (Ex. 360 [b.p. 20] nm; Em. 450 [b.p. 30] nm). An 8 point 2-fold serial dilution standard curve of 4MU (from 10 µM) in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂ was used to calculate the evolution of 4MU during the reaction course. Initial reaction velocities at each substrate concentration were fitted to the Michaelis-Menten equation to derive K_M and V_max. k_cat was derived from V_max using the relationship V_max=k_cat*[E_t]. All graphs and curve fittings were processed using GraphPad Prism 10.1.0.

Fluorometric coupled enzyme activity assay for NDST1 and mutants

For the determination of pseudo-first order NDST1 kinetics with respect to polysaccharide substrate, varying concentrations of K5 or HS polysaccharide (low sulfated fraction; Iduron, GAG HS-I) were incubated with 250 nM NDST1, 350 nM Sult1a1-K65E-R68G, 20 µM PAPS and 4 mM 4MU-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT. Controls omitting polysaccharide were subtracted from reactions prior to further analysis. For determination of pseudo-first order kinetics with respect to PAPS, a 2-fold serial dilution of PAPS from 50 µM was incubated with 0.5 mg mL⁻¹ HS, 250 nM NDST1, 350 nM Sult1a1-K65E-R68G and 4 mM 4MU-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT. Controls omitting PAPS were subtracted from reactions prior to further analysis.

Evaluation of NDST1 mutants was performed using 1 µM enzyme in the presence of 350 nM Sult1a1-K65E-R68G, 20 µM PAPS, 0.5 mg mL⁻¹ K5 and 4 mM 4MU-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT. For evaluation of NDST1 activity with nanobodies, reactions were performed with an 8 point, 3-fold dilution of each nanobody from 90 µM in the presence of 250 nM NDST1, 350 nM Sult1a1-K65E-R68G, 20 µM PAPS, 0.5 mg mL⁻¹ K5 and 4 mM 4MU-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT.

The effect of nAb7 or nAb13 on NDST1 kinetics was evaluated by the addition of 30 µM nanobody to pseudo-first order NDST1 kinetics reactions with respect to K5 (see above). Reactions were incubated with 250 nM NDST1, 350 nM Sult1a1-K65E-R68G, 20 µM PAPS and 4 mM 4MU-sulfate in 50 mM Tris-HCl pH 7.5, 15 mM MgCl₂, 1 mM DTT. Controls omitting polysaccharide were subtracted from reactions prior to further analysis.

The effect of different metal ions on NDST1 activity was evaluated in the presence of 1 µM NDST1, 350 nM Sult1a1-K65E-R68G, 20 µM PAPS, 0.5 mg mL⁻¹ K5 and 4 mM 4MU-sulfate in 50 mM Tris, pH 7.4, where the reaction buffer had been supplemented with 50 mM of MgCl₂, CaCl₂ or H₂O (metal free control), or 20 mM EDTA or dipicolinic acid.

In all cases reactions were performed in black half-area 96 well plates (Greiner) with a total reaction volume of 20 µL. All reactions were initiated by the addition of polysaccharide before recording the change in fluorescence with a Clariostar plate reader (BMG Labtech; software version 5.70 R2) using the 4MU preset. An 8 point 2-fold serial dilution standard curve of 4MU (from 10 µM) in 50 mM Tris-HCl pH 7, 15 mM MgCl₂, was used to calculate the evolution of 4MU during the reaction timecourse.

For most datasets, initial reaction velocities at each substrate concentration were fitted to the Michaelis-Menten equation to derive K_M and V_max values. k_cat was derived from V_max using the relationship V_max=k_cat*[E_t]. For kinetics with the NDST1-nAb7 complex (Fig. 4F) where reaction velocities did not plateau (i.e. [S] ≪ K_M), a combined k_cat/K_M value was derived from the slope of a linear fit to the data. K_i values for NDST1 with nAb5 and nAb7 were calculated using the Morrison quadratic equation, which applies when K_I and [E_t] are in similar ranges: Y = V₀*(1-((((E_t + X + Q)-(((E_t + X + Q)^2)−4*E_t*X)^0.5))/(2*E_t))); Q = (K_I*(1 + ([S]/K_M))). Graphs were processed and visualized using GraphPad Prism 10.1.0.

Δ-Disaccharide analysis of time course-NDST1 treated K5

600 µg of K5 was incubated with, 250 µM PAPS and 250 nM NDST1 in 50 mM Tris-HCl pH 7.4, 15 mM MgCl₂ for 18 h at 37 °C with shaking. Aliquots of the digest reaction were taken at various timepoints and heat inactivated at 95 °C for 10 min, before being stored at −20 °C until analysis.

Heparin lyase II (2 ug) was added to each sample and incubated at 30 °C for a total of 24 h, with the addition a further 2 ug of Heparin lyase II after 8 h. Samples were heat denatured at 95 °C for 5 min and stored at −20 °C before analysis. Chromatographic separation of Heparin lyase II digested samples was performed using high performance anion exchange chromatography (HPAEC). Samples were made up to 1 mL in HPLC-grade H₂O (Fisher) prior to being injected onto a ProPac PA-1 analytical column (4×250 mm), pre-equilibrated in HPLC-grade H₂O, at a flow rate of 1 mL min⁻¹. The column was held under isocratic flow for 10 minutes, followed by elution of Δ-disaccharides using a linear gradient of NaCl from 0 to 2 M NaCl in HPLC-grade H₂O over 60 minutes. Elution was monitored by in-line UV detection of A₂₃₂ via the C = C unsaturated bond introduced by Heparin lyase digestion. Retention times were compared to Δ-disaccharide reference standards (Iduron). The column was washed extensively with 2 M NaCl and HPLC-grade H₂O in between runs.

Biotinylation of K5 polysaccharide

K5 polysaccharide (4 mM) was biotinylated at the reducing end through a reaction with 1:1 molar ratio of N-(aminooxyacetyl)-N’-(D-Biotinoyl) hydrazine in the presence of 100 mM aniline, in 100 mM NaOAc pH 4.6^59,60. The reaction was incubated for 48 h at 37 °C before free biotin was removed by desalting on a C 10/40 column packed with Sephadex G25 (Cytiva), pre-equilibrated in dH₂O. Separation was conducted over 1.2 CV with inline UV monitoring at 232 and 210 nm. Fractions eluting in the void volume corresponding to K5 were pooled and lyophilized. Biotinylation was assessed by dot blot; 2 µL of 1 mg mL⁻¹ K5 or biotinylated K5 was spotted onto a nitrocellulose membrane and allowed to dry, before being blocked with 5% (w/v) BSA in PBS-T (PBS, 0.05% Tween-20) for 1 h. The blocked nitrocellulose membrane was washed 3x with PBS-T before being incubated with 1:2000 HRP-streptavidin (Thermo 21124) in 1% (w/v) BSA in PBS-T for 30 min. Following three washes with PBS-T, the presence of biotin was detected by the addition of SuperSignal West Femto chemiluminescence substrate (Thermo), and imaged using a ChemiDoc system (BioRad).

Immunization and nanobody library generation

Antibodies to NDST1 were raised in a llama by intra-muscular immunization with purified protein using Gerbu LQ#3000 as the adjuvant. Immunizations and handling of the llama were performed under the authority of the project license PA1FB163A (University of Reading, UK). Total RNA was extracted from peripheral blood mononuclear cells, and V_HH complementary DNAs were generated by RT-PCR. The pool of V_HH-encoding sequences was amplified by two rounds of nested PCR and cloned into the SfiI sites of the phagemid vector pADL-23c⁶¹. Electrocompetent E. coli TG1 cells were transformed with the recombinant pADL-23c vector, and the resulting TG1 phagemid library stock stored at −80 °C.

Nanobody panning

The VHH-displaying phage library was recovered by inoculation into 2x TY media supplemented with 100 µg mL⁻¹ ampicillin, followed by incubation at 37 °C, 250 rpm until OD₆₀₀ 0.4–0.6. The culture was then infected with M13 helper phage and incubated for an hour at 37 °C, before centrifugation at 2800g for 10 min at ambient temperature. The pellet was resuspended in 2x TY media supplemented with 100 µg mL⁻¹ ampicillin and 50 μg mL⁻¹ kanamycin and grown overnight at 25 °C, 250 rpm with shaking to amplify the library. The amplified VHH-phage library was recovered by centrifugation at 3200g for 15 min at 4 °C, to remove bacterial cells, and the supernatant collected. VHH-presenting phage were subsequently precipitated from the supernatant by the addition of 0.2 volumes of 20% (w/v) PEG6000, 2.5 M NaCl and incubated on ice for 1 h. The precipitated VHH-phage library was collected by centrifugation at 2300g for 10 min at 4 °C, then resuspended in 1 mL ice-cold PBS. A further centrifugation step was performed at 20,000g, 4 °C for one minute to remove bacterial contaminants. The VHH-phage library was reprecipitated by the addition of 0.2 volumes of 20% PEG6000 (w/v), 2.5 M NaCl and incubated on ice for 30 min, harvested by centrifugation at 20,000g, 4 °C for 15 min and resuspended in ice-cold PBS, before being stored at 4 °C.

The VHH-phage library was enriched for NDST1 binders by two rounds of bio-panning. The library was first blocked with StartingBlock (Thermo) for 30 min, followed by an incubation with 50 nM biotinylated NDST1 for an hour, rotating at room temperature. Pre-blocked Dynabeads (Thermo) were mixed with the VHH-phage-biotinylated NDST1 mixture, and incubated for 15 minutes at room temperature. Dynabeads were captured and washed six times with PBS-T, followed by one wash with PBS, before bound phage were released by digestion with 0.25 mg mL⁻¹ trypsin (Sigma-Aldrich) in 10 mM Tris, 137 mM NaCl, 1 mM CaCl₂, for 30 min at room temperature. The eluted enriched VHH-phage library was collected and amplified by inoculation into exponentially growing TG1 E. coli cells, which were grown for 30 min at 37 °C with shaking, then harvested by centrifugation at 2800g for 10 min before being resuspended in 1 mL of 2xTY media and plated onto LB agar containing 100 µg mL⁻¹ ampicillin. Following incubation overnight at 37 °C, TG1 E. coli cells were collected in 2xTY media containing 25% glycerol and stored at −80 °C for further use. The recovered amplified and enriched phage library was used for a further round of bio-panning, performed with 5 nM biotinylated NDST1. Enrichment after bio-panning was determined by plating a 10-fold serial dilution of the recovered cell culture on LB agar plates containing 100 µg mL⁻¹ ampicillin.

For screening for NDST1 binding clones, 93 individual clones were selected and grown in 2x TY media (100 µg mL⁻¹ ampicillin) overnight at 37 °C, 250 rpm. Overnight cultures were inoculated into 2x TY media containing 100 µg mL⁻¹ ampicillin and incubated for 3 h at 37 °C, 250 rpm shaking, before being infected with M13 helper phage. The infected cultures were grown for a further hour at 37 °C, 250 rpm before being centrifuged at 2800g for 10 min at 4 °C. The cell pellet was resuspended in 2x TY media containing 100 µg mL⁻¹ ampicillin and 50 µg mL⁻¹ kanamycin, before being grown at 25 °C, 250 rpm overnight. Clonal VHH-presenting phage were harvested from the culture supernatant by centrifugation at 4000g for 15 min at 4 °C to remove bacterial cells, and the supernatant stored at 4 °C for further analysis.

Enzyme-linked immunosorbent assays to identify NDST1 binding phage

A 96-well high binding microtiter plate (Greiner) was coated with 5 µg mL⁻¹ neutravidin in 100 µL PBS per well, and stored overnight at 4 °C. The plate was then washed 5x with PBS-T before the addition of 100 µL of biotinylated NDST1 (50 nM in 20 mM MES, 100 mM NaCl, pH 6.5, 0.1% fat free skimmed milk) per well, and then incubated for 1 h at room temperature, with agitation. The plate was washed 5x with 20 mM MES, 100 mM NaCl, pH 6.5 before being blocked with PBS-T, 2% non-fat skimmed milk (250 µL per well) for 1 h at room temperature, with agitation. Following a further 5 washes with PBS-T, 100 µL of each clonal VHH-presenting phage stock (diluted with 1 volume of PBS-T, 2% non-fat skimmed milk) was added to each well and incubated for 1 h at room temperature, with agitation. The plate was washed five times with PBS-T, before being incubated for 1 h at room temperate with 100 µL anti-M13-HRP (Cytiva 27-9421-01; 1:5000 dilution in StartingBlock; Thermo) per well, with agitation. The plate was washed 5x with PBS-T before the addition of 100 µL per well TMB substrate (SeraCare). Absorbance was read at 405 nM using a ClarioStar plate reader (software version 5.70 R2).

Nanobody expression and purification

Identified VHH hits were transformed into WK6 E. coli cells using electroporation. 1 L of TB media containing 100 µg mL⁻¹ ampicillin, 0.1% glucose (w/v) and 1 mM MgCl₂ was inoculated and grown at 37 °C, 250 rpm until OD600 1–1.2. Expression was induced with 1 mM IPTG and the culture incubated at 28 °C overnight, 250 rpm shaking, before harvesting by centrifugation at 4000g for 15 min at 4 °C. nAbs were extracted from the periplasm by resuspending the cell pellet in TES buffer (200 mM Tris pH 8.0, 0.5 mM EDTA, 500 mM sucrose) overnight with constant agitation. Two volumes of 50 mM Tris pH 8.0, 125 mM sucrose, supplemented with 120 kU DNase I was added, and the resulting suspension incubated for a further 2 h to lyse cells. Cell debris was removed by centrifugation at 40,000g for 30 minutes at 4 °C and the supernatant collected. The supernatant was diluted with 5 volumes of PBS and filtered before being loaded onto a 5 mL HisTrap FF column (Cytiva) pre-equilibrated in Histrap buffer A (PBS, 30 mM imidazole). The column was washed with 5 CV of the same buffer, before elution of the nanobody with a linear gradient of 0-100% Histrap buffer B (PBS, 300 mM imidazole) over 10 CV. Fractions containing nAbs, as judged by SDS-PAGE, were pooled and concentrated using a 10 kDa MWCO Vivaspin filter. nAbs were purified further by SEC using a HiLoad 16/600 Superdex 75 pg column pre-equilibrated in PBS. nAb containing fractions were pooled, concentrated to >2 mg mL⁻¹ and flash frozen in LN₂. nAbs were stored at −80 °C until required. DNA and protein sequences for nAbs are available in Supplementary Table 2.

Binding analyses—SPR

Surface plasmon resonance experiments were performed using a Biacore T200 system (GE Healthcare; control software version 2.0.2) primed with 20 mM MES pH 6.5, 100 mM NaCl, 0.05% Tween-20. All assays were performed using a BiotinCAP Sensor chip (GE Healthcare) at 25 °C. To determine the binding affinity of nanobody clones for NDST1, biotinylated NDST1 (10 µM) was immobilized onto the sample channel of the sensor chip at a flow rate of 2 µL min⁻¹ for 600 s. The reference channel was left blank. A titration of each nanobody (from 10 µM) was injected over both sensor channels at a flow rate of 30 µL min⁻¹, for 100 s. Dissociation was monitored for 100 s. Steady-state binding responses were fitted to a 1:1 binding model (Data evaluation software version 3.1; One-site specific binding) using GraphPad Prism 10.1.0 to fit K_D values. In combinatorial assays, nanobodies were either pre-mixed (1:1, 10 µM) or injected alone at a flow rate of 30 µL min⁻¹ for 100 s, with monitoring of the dissociation for 100 s, followed by comparison of the maximal response during the association phase.

Binding analyses—BLI

Biolayer interferometry (BLI) experiments were performed with an Octet R8 system (Sartorius; software Octet BLI discovery version 12.2.2.20) using Octet SA biosensors. For determination of the binding affinity of nanobodies to NDST1, biosensors were hydrated with two washes of 20 mM MES pH 6.5, 100 mM NaCl, before immobilization of biotinylated NDST1 (75 nM) for 300 s in the same buffer. The loaded sensors were washed twice for 60 s in PBS-T, and the baseline recorded. The association of nanobodies in PBS-T to immobilized NDST1 or unloaded reference sensors were recorded for 100–600 s, followed by measurement of the dissociation in PBS-T for 100–300 s. All steps were performed with shaking at 1000 rpm at 25 °C. Control reactions containing no nanobody and reference sensors were subtracted prior to alignment and smoothing of the sensorgrams with Savitzky-Golay filtering using Octet Analysis studio version 12.2.2.26 (Sartorius). Steady-state binding responses were fitted to a 1:1 binding model (One-site specific binding) using GraphPad Prism 10.1.0 to calculate K_D values.

For determination of the binding affinity of NDST1 to K5, Octet SA biosensors were hydrated with three washes in PBS-T for 300 s before immobilization of biotinylated-K5 (25 µg mL⁻¹ or 50 µg mL⁻¹) in the same buffer for 300 s. The loaded sensors were washed twice, for 60 s, in PBS-T and the baseline recorded. Association of NDST1 to immobilized biotinylated-K5 or unloaded reference sensors was recorded for 300 s followed by measurement of dissociation in PBS-T for 300 s. All steps were performed with shaking at 1000 rpm at 25 °C. Control reactions containing no NDST1 and reference sensors were subtracted prior to smoothing of the sensorgrams with Savitzky-Golay filtering using Octet Analysis studio version 12.2.2.26. Data were fitted using a 2:1 heterogenous binding model with global fitting in Octet Analysis studio to determine k_assoc, k_dis and K_D values. The effect of nanobodies nAb7 and nAb13 on NDST1 binding to K5 was assessed using the same conditions, with the inclusion of 10 µM nanobody. To evaluate the effect of NaCl on NDST1 binding to K5, an 8 point 1.5-fold dilution of 1.5 M NaCl into 10 mM phosphate buffer pH 7.4, 0.05% Tween-20 was performed using the same assay conditions.

Differential scanning fluorimetry

Differential scanning fluorimetry experiments were conducted using a QuantStudio5 real-time PCR system running QuantStudio Design & Analysis Software v1.5.2 (Applied Biosystems), employing the TAMRA filter setting, with a step increase of 0.5°C per 30 s from 25 to 95 °C. The thermal stability of NDST1 (1 µg) in the presence of different nAb concentrations was measured in a total volume of 20 µL in PBS pH 7.4 containing 20X Sypro Orange (Invitrogen). Control reactions containing equivalent amounts of nAb or NDST1 were run alongside on the same plate. Denaturation temperatures (T_M) were quantitated from the peak of the first differential of the resulting fluorescence curves, as calculated using Protein Thermal Shift Software v1.4 (Applied Biosystems). Raw data and first differential plots were visualized using GraphPad Prism 10.1.0.

Generating NDST1 nanobody complexes

NDST1 and nanobodies were incubated in 20 mM MES pH 6.5, 200 mM NaCl, at 1:1 mass ratios for 1 h at room temperature. Complexes were isolated by size exclusion chromatography on a Superdex 200 Increase 10/300 GL column (Cytiva), pre-equilibrated in the same buffer. NDST1-nAb7 and NDST1-nAb13 containing fractions were pooled and concentrated to >5 mg mL⁻¹, using a 10 kDa MWCO vivaspin concentrator. Small aliquots were flash frozen using liquid nitrogen and stored at –80 °C for further use, or used immediately.

Cryo-EM sample preparation

Frozen aliquots of NDST1-nAb7 or NDST1-nAb13 were thawed rapidly and protein concentration adjusted to 0.4–0.6 mg mL⁻¹ using 50 mM Tris-HCl pH 7.5, supplemented with PAP (5 mM) and MgCl₂ (15 mM). 2.5 µL of samples were applied to Quantifoil R1.2/1.3 300 Cu mesh grids, glow discharged for 60 s at 30 mA negative current in a GloQube Plus glow discharger (Quorum Technologies). Grids were blotted for 1–3 s using a blot force of 3 before plunging into liquid ethane using a Vitrobot Mark IV (Thermo) set to 4 °C, 100% humidity. Grids were clipped into Autogrid rings (Thermo) before loading into microscopes for screening and data collection.

Cryo-EM data collection and processing

Th NDST1 alone and NDST1-nAb13 datasets were collected from the same cryo-EM grid, in which a large proportion of NDST1-nAb13 complexes had dissociated, enabling the classification of both nAb13 bound and unbound particles. Data was collected on a Krios G4 microscope using an accelerating voltage of 300 kV, Falcon 4i direct electron detector and Selectris X energy filter with a slit width of 10 eV. Movies were collected using EPU (version 3.2.0) in EER format at 165,000x magnification, using a total dose of 50 electrons per Å², a calibrated pixel size of 0.73 Å, and target defocus values of –1.2 µm to –2.6 µm in 0.2 µm intervals.

The NDST1-nAb7 dataset was collected on a Krios G3i microscope, using an accelerating voltage of 300 kV, Falcon 4 direct electron detector and Selectris X energy filter with a silt width of 10 eV. Movies were collected using EPU (version 3.2.0) in EER format at 165,000x magnification, using a total dose of 50 electrons per Å², a calibrated pixel size of 0.73 Å, and target defocus values of –1.2 µm to –2.6 µm in 0.2 µm intervals.

EM processing workflows for each dataset are outlined in Supplementary Fig. 7. In brief, micrograph pre-processing was performed in cryoSPARC version 2.14.1^62,63,64 live interface using path-motion CTF correction. Initial particles were picked using unbiased blob-picking, and good particles isolated via two rounds of 2D classification. 2D class averages with secondary structure features were used to guide template-based particle picking, followed by further rounds of 2D classification. Ab initio models generated in cryoSPARC were used as non-biased references for 1 or more rounds of heterogenous refinement.

The best classes from heterogenous refinement, as judged by resolution, were pooled and refined further using cryoSPARC non-uniform refinement, followed by rounds of local and global CTF refinement where appropriate, followed again by non-uniform refinement. Final postprocessing and local resolution estimates were carried out in Relion 3.1.1⁶⁵. 3DFSC calculations were carried out using the Remote 3DFSC Processing Server (3dfsc.salk.edu)⁶⁶.

3D-variability analyses on the NDST1-nAb7 and NDST1-nAb13 complexes were each carried out using the cryoSPARC pipeline, incorporating 3 variability components. Focussed local refinements for both complexes were carried out in cryoSPARC, with masking to subtract either the NTD or the sulfotransferase domain, generating subtracted particles for the deacetylase-sulfotransferase-nAb or NTD-deacetylase-nAb regions respectively. Locally refined volumes were postprocessed using Relion 3.1.1, before being combined using UCSF ChimeraX, by taking the maximum value from either locally refined map at each voxel. The composite map was used to guide model bulding. All other analyses and depositions were carried out with the individual locally refined maps.

For all structures, cryo-EM model building was carried out in COOT⁶⁷, using starting models initially derived from AlphaFold2 (NDST1) or the PDB (for nAb7, from accession 7TGF). Models were improved by manual building using COOT, iterated with rounds of real-space refinement using PHENIX.REFINE⁶⁸. The nAb13 nanobody structure was derived from the nAb7 structure by conversion to an all Cα chain in COOT, and removal of the long nAb7 unique CDR3 loop (R98–F116), before fitting the Cα model into the NDST1-nAb13 volume by 5 rounds of rigid-body refinement in PHENIX.REFINE. Structural figures were generated using ChimeraX⁶⁹. Electrostatic surfaces were calculated using ChimeraX. Binding interactions were determined using PISA analysis⁷⁰ with default settings. RMSD values were calculated using CCP4MG⁷¹. Sequence alignment values were calculated using ClustalΩ⁷². All cryo-EM processing and model building statistics are shown in Supplementary Table 1.

Molecular docking

Interactions between NDST1 and HS were analyzed by molecular docking using GlycoTorch Tools and GlycoTorch Vina (GTV)⁴¹. The protonation of residues of both protein and ligands were set to reflect physiological pH 7.4. Glycam carbohydrate builder (Glycam.org) was used to build the two octasaccharides, [GlcNAc-GlcA]₄ and [GlcNAc-GlcA]₂-[GlcNS-GlcA]₂. GTV was used to convert the octasaccharides obtained from Glycam to pdbqt format. A centroid selected from the PAP binding site and metal binding sites from the cryo-EM structure was used to define the coordinates for the docking of heparosan oligosaccharides. Docking of oligosaccharides was performed using the grid box sizes of 40 × 50 × 40 Å and 55 × 40 × 25 Å from the centroids in the deacetylase and sulfotransferase domains respectively. The octasaccharides were docked using GTV with energy range, exhaustiveness and num_modes values set to be 12, 80, and 100, respectively. The conformations chosen for further analysis were based on the best-estimated affinities and their ranking in the clustering process. Docking visualization and figures were generated using ChimeraX⁶⁹.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The cryo-EM data generated in this study have been deposited in the PDB and EMDB under accession codes 8CCY (NDST1 alone model), EMD-16564 (NDST1 alone map), 8CD0 (NDST1-nAb7 model), EMD-16627 (NDST1-nAb7 NTD and deacetylase domain local map), EMD-16629 (NDST1-nAb7 deacetylase and sulfotransferase domain local map), EMD-16565 (NDST1-nAb7 composite map), EMD-16626 (NDST1-nAb7 original map), 8CHS (NDST1-nAb13 model), EMD-16662 (NDST1-nAb13 deacetylase and sulfotransferase domain local map), EMD-16663 (NDST1-nAb13 deacetylase and sulfotransferase domain local map), EMD-16664 (NDST1-nAb13 composite map), EMD-16661 (NDST1-nAb13 original map). We have also referenced in this work PDB accessions 6NWZ (Agd3 deacetylase), 3UAN (HS3ST1), 6XL8 (HS3ST3), and EMDB accession EMD-17349 (NDST1 dimer). DNA and protein sequences for NDST1 and nAb constructs are available in Supplementary Table 2. Source data are provided with this paper.

References

Esko, J. D. & Lindahl, U. Molecular diversity of heparan sulfate. J. Clin. Invest 108, 169–173 (2001).
Article CAS PubMed PubMed Central Google Scholar
Sarrazin, S., Lamanna, W. C. & Esko, J. D. Heparan sulfate proteoglycans. Cold Spring Harb. Perspect. Biol. 3, a004952 (2011).
Article PubMed PubMed Central Google Scholar
Pellegrini, L. Role of heparan sulfate in fibroblast growth factor signalling: a structural view. Curr. Opin. Struct. Biol. 11, 629–634 (2001).
Article CAS PubMed Google Scholar
Yan, D. & Lin, X. Shaping morphogen gradients by proteoglycans. Cold Spring Harb. Perspect. Biol. 1, a002493 (2009).
Article PubMed PubMed Central Google Scholar
Lim, H. C., Multhaupt, H. A. & Couchman, J. R. Cell surface heparan sulfate proteoglycans control adhesion and invasion of breast carcinoma cells. Mol. Cancer 14, 15 (2015).
Article CAS PubMed PubMed Central Google Scholar
Christianson, H. C. & Belting, M. Heparan sulfate proteoglycan as a cell-surface endocytosis receptor. Matrix Biol. 35, 51–55 (2014).
Article CAS PubMed Google Scholar
Bode, L. et al. Heparan sulfate and syndecan-1 are essential in maintaining murine and human intestinal epithelial barrier function. J. Clin. Invest. 118, 229–238 (2008).
Article CAS PubMed Google Scholar
Clausen, T. M. et al. SARS-CoV-2 infection depends on cellular heparan sulfate and ACE2. Cell 183, 1043–1057.e15 (2020).
Article CAS PubMed PubMed Central Google Scholar
Mycroft-West, C. J. et al. Heparin inhibits cellular invasion by SARS-CoV-2: structural dependence of the interaction of the spike S1 receptor-binding domain with heparin. Thromb. Haemost. 120, 1700–1715 (2020).
Article PubMed PubMed Central Google Scholar
Shukla, D. & Spear, P. G. Herpesviruses and heparan sulfate: an intimate relationship in aid of viral entry. J. Clin. Invest. 108, 503–510 (2001).
Article CAS PubMed PubMed Central Google Scholar
Rabenstein, D. L. Heparin and heparan sulfate: structure and function. Nat. Prod. Rep. 19, 312–331 (2002).
Article CAS PubMed Google Scholar
Meneghetti, M. C. et al. Heparan sulfate and heparin interactions with proteins. J. R. Soc. Interface 12, 0589 (2015).
Article PubMed Google Scholar
Kreuger, J. & Kjellen, L. Heparan sulfate biosynthesis: regulation and variability. J. Histochem. Cytochem. 60, 898–907 (2012).
Article PubMed PubMed Central Google Scholar
Leisico, F. et al. Structure of the human heparan sulfate polymerase complex EXT1-EXT2. Nat. Commun. 13, 7110 (2022).
Article CAS PubMed PubMed Central ADS Google Scholar
Wilson, L. F. L. et al. The structure of EXTL3 helps to explain the different roles of bi-domain exostosins in heparan sulfate synthesis. Nat. Commun. 13, 3314 (2022).
Article CAS PubMed PubMed Central ADS Google Scholar
Li, H. et al. Structural basis for heparan sulfate co-polymerase action by the EXT1-2 complex. Nat. Chem. Biol. 19, 565–574 (2023).
Article PubMed PubMed Central ADS Google Scholar
Esko, J. D. & Selleck, S. B. Order out of chaos: assembly of ligand binding sites in heparan sulfate. Annu. Rev. Biochem 71, 435–471 (2002).
Article CAS PubMed Google Scholar
Dou, W., Xu, Y., Pagadala, V., Pedersen, L. C. & Liu, J. Role of deacetylase activity of N-deacetylase/N-sulfotransferase 1 in forming N-sulfated domain in heparan sulfate. J. Biol. Chem. 290, 20427–20437 (2015).
Article CAS PubMed PubMed Central Google Scholar
Annaval, T. et al. Heparan sulfate proteoglycans biosynthesis and post synthesis mechanisms combine few enzymes and few core proteins to generate extensive structural and functional diversity. Molecules 25, 4215 (2020).
Article CAS PubMed PubMed Central Google Scholar
Sheng, J., Liu, R., Xu, Y. & Liu, J. The dominating role of N-deacetylase/N-sulfotransferase 1 in forming domain structures in heparan sulfate. J. Biol. Chem. 286, 19768–19776 (2011).
Article CAS PubMed PubMed Central Google Scholar
Carlsson, P., Presto, J., Spillmann, D., Lindahl, U. & Kjellen, L. Heparin/heparan sulfate biosynthesis: processive formation of N-sulfated domains. J. Biol. Chem. 283, 20008–20014 (2008).
Article CAS PubMed Google Scholar
Kjellen, L. Glucosaminyl N-deacetylase/N-sulphotransferases in heparan sulphate biosynthesis and biology. Biochem. Soc. Trans. 31, 340–342 (2003).
Article CAS PubMed Google Scholar
Pikas, D. S., Eriksson, I. & Kjellen, L. Overexpression of different isoforms of glucosaminyl N-deacetylase/N-sulfotransferase results in distinct heparan sulfate N-sulfation patterns. Biochemistry 39, 4552–4558 (2000).
Article CAS PubMed Google Scholar
Aikawa, J. & Esko, J. D. Molecular cloning and expression of a third member of the heparan sulfate/heparin GlcNAc N-deacetylase/ N-sulfotransferase family. J. Biol. Chem. 274, 2690–2695 (1999).
Article CAS PubMed Google Scholar
Aikawa, J., Grobe, K., Tsujimoto, M. & Esko, J. D. Multiple isozymes of heparan sulfate/heparin GlcNAc N-deacetylase/GlcN N-sulfotransferase. Structure and activity of the fourth member, NDST4. J. Biol. Chem. 276, 5876–5882 (2001).
Article CAS PubMed Google Scholar
Pallerla, S. R. et al. Altered heparan sulfate structure in mice with deleted NDST3 gene function. J. Biol. Chem. 283, 16885–16894 (2008).
Article CAS PubMed PubMed Central Google Scholar
Grobe, K. et al. Cerebral hypoplasia and craniofacial defects in mice lacking heparan sulfate Ndst1 gene function. Development 132, 3777–3786 (2005).
Article CAS PubMed Google Scholar
Ringvall, M. et al. Defective heparan sulfate biosynthesis and neonatal lethality in mice lacking N-deacetylase/N-sulfotransferase-1. J. Biol. Chem. 275, 25926–25930 (2000).
Article CAS PubMed Google Scholar
Fan, G. et al. Targeted disruption of NDST-1 gene leads to pulmonary hypoplasia and neonatal respiratory distress in mice. FEBS Lett. 467, 7–11 (2000).
Article CAS PubMed ADS Google Scholar
Reuter, M. S. et al. NDST1 missense mutations in autosomal recessive intellectual disability. Am. J. Med. Genet. A 164A, 2753–2763 (2014).
Article PubMed Google Scholar
Lencz, T. et al. Genome-wide association study implicates NDST3 in schizophrenia and bipolar disorder. Nat. Commun. 4, 2739 (2013).
Article PubMed ADS Google Scholar
Tzeng, S. T. et al. NDST4 is a novel candidate tumor suppressor gene at chromosome 4q26 and its genetic loss predicts adverse prognosis in colorectal cancer. PLoS ONE 8, e67040 (2013).
Article CAS PubMed PubMed Central ADS Google Scholar
Kakuta, Y., Sueyoshi, T., Negishi, M. & Pedersen, L. C. Crystal structure of the sulfotransferase domain of human heparan sulfate N-deacetylase/ N-sulfotransferase 1. J. Biol. Chem. 274, 10673–10676 (1999).
Article CAS PubMed Google Scholar
Atienza, J., Tkachyova, I., Tropak, M., Fan, X. & Schulze, A. Fluorometric coupled enzyme assay for N-sulfotransferase activity of N-deacetylase/N-sulfotransferase (NDST). Glycobiology 31, 1093–1101 (2021).
Article CAS PubMed Google Scholar
Ly, M. et al. Analysis of E. coli K5 capsular polysaccharide heparosan. Anal. Bioanal. Chem. 399, 737–745 (2011).
Article CAS PubMed Google Scholar
Pardon, E. et al. A general protocol for the generation of Nanobodies for structural biology. Nat. Protoc. 9, 674–693 (2014).
Article CAS PubMed PubMed Central Google Scholar
Shi, X. & Jarvis, D. L. Protein N-glycosylation in the baculovirus-insect cell system. Curr. Drug Targets 8, 1116–1125 (2007).
Article CAS PubMed PubMed Central Google Scholar
Bamford, N. C. et al. Structural and biochemical characterization of the exopolysaccharide deacetylase Agd3 required for Aspergillus fumigatus biofilm formation. Nat. Commun. 11, 2450 (2020).
Article CAS PubMed PubMed Central ADS Google Scholar
Burger, M. & Chory, J. Structural and chemical biology of deacetylases for carbohydrates, proteins, small molecules and histones. Commun. Biol. 1, 217 (2018).
Article PubMed PubMed Central Google Scholar
Ahangar, M. S. et al. Structural and functional determination of homologs of the Mycobacterium tuberculosis N-acetylglucosamine-6-phosphate deacetylase (NagA). J. Biol. Chem. 293, 9770–9783 (2018).
Article CAS PubMed PubMed Central Google Scholar
Boittier, E. D., Burns, J. M., Gandhi, N. S. & Ferro, V. GlycoTorch vina: docking designed and tested for glycosaminoglycans. J. Chem. Inf. Model 60, 6328–6343 (2020).
Article CAS PubMed Google Scholar
Wei, Z., Swiedler, S. J., Ishihara, M., Orellana, A. & Hirschberg, C. B. A single protein catalyzes both N-deacetylation and N-sulfation during the biosynthesis of heparan sulfate. Proc. Natl Acad. Sci. USA 90, 3885–3888 (1993).
Article CAS PubMed PubMed Central ADS Google Scholar
Riesenfeld, J., Hook, M. & Lindahl, U. Biosynthesis of heparin. Assay and properties of the microsomal N-acetyl-D-glucosaminyl N-deacetylase. J. Biol. Chem. 255, 922–928 (1980).
Article CAS PubMed Google Scholar
Kellokumpu, S. Golgi pH, ion and redox homeostasis: how much do they really matter? Front. Cell Dev. Biol. 7, 93 (2019).
Article PubMed PubMed Central Google Scholar
Boraston, A. B., Bolam, D. N., Gilbert, H. J. & Davies, G. J. Carbohydrate-binding modules: fine-tuning polysaccharide recognition. Biochem. J. 382, 769–781 (2004).
Article CAS PubMed PubMed Central Google Scholar
Vallet, S. D. et al. Functional and structural insights into human N-deacetylase/N-sulfotransferase activities. Proteoglycan Res. 1, e8 (2023).
Article Google Scholar
Westling, C. & Lindahl, U. Location of N-unsubstituted glucosamine residues in heparan sulfate. J. Biol. Chem. 277, 49247–49255 (2002).
Article CAS PubMed Google Scholar
Traenkle, B. & Rothbauer, U. Under the microscope: single-domain antibodies for live-cell imaging and super-resolution microscopy. Front. Immunol. 8, 1030 (2017).
Article PubMed PubMed Central Google Scholar
Doria-Rose, N. A. & Joyce, M. G. Strategies to guide the antibody affinity maturation process. Curr. Opin. Virol. 11, 137–47 (2015).
Article CAS PubMed PubMed Central Google Scholar
Xu, Y. et al. Structure based substrate specificity analysis of heparan sulfate 6-o-sulfotransferases. ACS Chem. Biol. 12, 73–82 (2017).
Article CAS PubMed ADS Google Scholar
Wander, R. et al. Structural and substrate specificity analysis of 3-O-sulfotransferase isoform 5 to synthesize heparan sulfate. ACS Catal. 11, 14956–14966 (2021).
Article CAS PubMed PubMed Central Google Scholar
Tsutsui, Y., Ramakrishnan, B. & Qasba, P. K. Crystal structures of beta-1,4-galactosyltransferase 7 enzyme reveal conformational changes and substrate binding. J. Biol. Chem. 288, 31963–31970 (2013).
Article CAS PubMed PubMed Central Google Scholar
Pedersen, L. C. et al. Heparan/chondroitin sulfate biosynthesis. Structure and mechanism of human glucuronyltransferase I. J. Biol. Chem. 275, 34580–34585 (2000).
Article CAS PubMed Google Scholar
Liu, C. et al. Molecular mechanism of substrate specificity for heparan sulfate 2-O-sulfotransferase. J. Biol. Chem. 289, 13407–13418 (2014).
Article CAS PubMed PubMed Central Google Scholar
Debarnot, C. et al. Substrate binding mode and catalytic mechanism of human heparan sulfate d-glucuronyl C5 epimerase. Proc. Natl Acad. Sci. USA 116, 6760–6765 (2019).
Article CAS PubMed PubMed Central ADS Google Scholar
Briggs, D. C. & Hohenester, E. Structural basis for the initiation of glycosaminoglycan biosynthesis by human xylosyltransferase 1. Structure 26, 801–809.e3 (2018).
Article CAS PubMed PubMed Central Google Scholar
Liu, H. & Naismith, J. H. An efficient one-step site-directed deletion, insertion, single and multiple-site plasmid mutagenesis protocol. BMC Biotechnol. 8, 91 (2008).
Article PubMed PubMed Central Google Scholar
Lu, L. Y., Hsu, Y. C. & Yang, Y. S. Spectrofluorometric assay for monoamine-preferring phenol sulfotransferase (SULT1A3). Anal. Biochem 404, 241–243 (2010).
Article CAS PubMed Google Scholar
Thakar, D. et al. A quartz crystal microbalance method to study the terminal functionalization of glycosaminoglycans. Chem. Commun. (Camb.) 50, 15148–15151 (2014).
Article CAS PubMed Google Scholar
Su, D. et al. Analysis of protein-heparin interactions using a portable SPR instrument. PeerJ Anal. Chem. 4, e15 (2022).
Article CAS Google Scholar
Huo, J. et al. A potent SARS-CoV-2 neutralising nanobody shows therapeutic efficacy in the Syrian golden hamster model of COVID-19. Nat. Commun. 12, 5469 (2021).
Article CAS PubMed PubMed Central ADS Google Scholar
Punjani, A., Zhang, H. & Fleet, D. J. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat. Methods 17, 1214–1221 (2020).
Article CAS PubMed Google Scholar
Rubinstein, J. L. & Brubaker, M. A. Alignment of cryo-EM movies of individual particles by optimization of image translations. J. Struct. Biol. 192, 188–195 (2015).
Article PubMed Google Scholar
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).
Article CAS PubMed Google Scholar
Zivanov, J. et al. New tools for automated high-resolution cryo-EM structure determination in RELION-3. Elife 7, e42166 (2018).
Article PubMed PubMed Central Google Scholar
Tan, Y. Z. et al. Addressing preferred specimen orientation in single-particle cryo-EM through tilting. Nat. Methods 14, 793–796 (2017).
Article CAS PubMed PubMed Central Google Scholar
Casanal, A., Lohkamp, B. & Emsley, P. Current developments in Coot for macromolecular model building of electron cryo-microscopy and crystallographic data. Protein Sci. 29, 1069–1078 (2020).
Article CAS PubMed PubMed Central Google Scholar
Liebschner, D. et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr. D: Struct. Biol. 75, 861–877 (2019).
Article CAS PubMed ADS Google Scholar
Pettersen, E. F. et al. UCSF ChimeraX: Structure visualization for researchers, educators, and developers. Protein Sci. 30, 70–82 (2021).
Article CAS PubMed Google Scholar
Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372, 774–797 (2007).
Article CAS PubMed Google Scholar
McNicholas, S., Potterton, E., Wilson, K. S. & Noble, M. E. Presenting your structures: the CCP4mg molecular-graphics software. Acta Crystallogr. D: Biol. Crystallogr 67, 386–394 (2011).
Article CAS PubMed ADS Google Scholar
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011).
Article PubMed PubMed Central Google Scholar
Mistry, J. et al. Pfam: The protein families database in 2021. Nucleic Acids Res. 49, D412–D419 (2021).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

This work was supported by the Rosalind Franklin Institute, with funding delivery partner the Engineering and Physical Sciences Research Council (EPSRC) UK. We acknowledge support from the Biotechnology and Biological Sciences Research Council, UK (BB/L023717/1 & BB/X011739/1 to MAS; BB/V018523/1 to RJO), the Engineering and Physical Sciences Research Council, UK (EP/X019179/1 to MAS), and the Wellcome Trust (Sir Henry Dale Fellowship 218579/Z/19/Z to LW). Electron microscopy provision was provided via the Rosalind Franklin institute, and through the Oxford particle Imaging Centre (OPIC) facility, a UK Instruct-ERIC Centre founded by a Wellcome JIF award (060208/Z/00/Z) and supported by a Wellcome equipment grant (093305/Z/10/Z). Financial support for HMED was provided by the Wellcome Trust Core Award Grant (203141/Z/16/Z). Computation for cryo-EM analysis was performed using the Oxford Biomedical Research Computing (BMRC) facility, a joint development between the Wellcome Centre for Human Genetics and the Big Data Institute supported by Health Data Research UK and the NIHR Oxford Biomedical Research Centre. NSG is supported by QUT Accelerate Fellowship. Computational (and/or data visualization) resources and services used in this work were provided by the eResearch Office, Queensland University of Technology, Brisbane, Australia. NSG would also like to acknowledge the National Computational Infrastructure (NCI), Australia for providing the high-performance computing facility to carry out molecular docking simulations. We thank Jiandong Huo for assistance with phage display library construction, and Mark Arrowsmith at the University of Keele School of Pharmacy and Bioengineering for assistance with ICP-OES.

Author information

Authors and Affiliations

The Rosalind Franklin Institute, Harwell Science & Innovation Campus, OX11 0QX, Didcot, UK
Courtney J. Mycroft-West, Sahar Abdelkarim, Raymond J. Owens & Liang Wu
Division of Structural Biology, Nuffield Department of Medicine, University of Oxford, The Wellcome Centre for Human Genetics, OX3 7BN, Oxford, UK
Helen M. E. Duyvesteyn, Raymond J. Owens & Liang Wu
Department of Computer Science and Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, 576104, Karnataka, India
Neha S. Gandhi
School of Chemistry and Physics, Queensland University of Technology, QLD 4000, Brisbane, Australia
Neha S. Gandhi
Centre for Genomics and Personalised Health, Queensland University of Technology, Brisbane, QLD 4059, Australia
Neha S. Gandhi
Centre for Glycoscience Research and Training, Keele University, ST5 5BG, Newcastle-Under-Lyme, UK
Mark A. Skidmore

Authors

Courtney J. Mycroft-West
View author publications
You can also search for this author in PubMed Google Scholar
Sahar Abdelkarim
View author publications
You can also search for this author in PubMed Google Scholar
Helen M. E. Duyvesteyn
View author publications
You can also search for this author in PubMed Google Scholar
Neha S. Gandhi
View author publications
You can also search for this author in PubMed Google Scholar
Mark A. Skidmore
View author publications
You can also search for this author in PubMed Google Scholar
Raymond J. Owens
View author publications
You can also search for this author in PubMed Google Scholar
Liang Wu
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

L.W. and C.J.M.W. conceived and designed experiments. L.W. and C.J.M.W. carried out protein expression and purification. C.J.M.W. carried out polysaccharide production and purification, with assistance from M.A.S. C.J.M.W. and S.A. generated nanobodies, under supervision from R.J.O. C.J.M.W. carried out biochemical and biophysical analyses. C.J.M.W., H.M.E.D., and L.W. carried out cryo-EM sample preparation, data collection and data analysis. N.S.G. carried out molecular docking simulations. L.W. wrote the manuscript with input from all authors.

Corresponding author

Correspondence to Liang Wu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Rebekka Wild and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Movie 1

Supplementary Movie 2

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Mycroft-West, C.J., Abdelkarim, S., Duyvesteyn, H.M.E. et al. Structural and mechanistic characterization of bifunctional heparan sulfate N-deacetylase-N-sulfotransferase 1. Nat Commun 15, 1326 (2024). https://doi.org/10.1038/s41467-024-45419-4

Download citation

Received: 07 September 2023
Accepted: 22 January 2024
Published: 13 February 2024
DOI: https://doi.org/10.1038/s41467-024-45419-4

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.