Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 May;19(5):565-574.
doi: 10.1038/s41589-022-01220-2. Epub 2023 Jan 2.

Structural basis for heparan sulfate co-polymerase action by the EXT1-2 complex

Hua Li #  1 , Digantkumar Chapla #  2 , Robert A Amos  2 , Annapoorani Ramiah  2 , Kelley W Moremen  3 , Huilin Li  4
Affiliations

Structural basis for heparan sulfate co-polymerase action by the EXT1-2 complex

Hua Li et al. Nat Chem Biol. 2023 May.

Abstract

Heparan sulfate (HS) proteoglycans are extended (-GlcAβ1,4GlcNAcα1,4-)n co-polymers containing decorations of sulfation and epimerization that are linked to cell surface and extracellular matrix proteins. In mammals, HS repeat units are extended by an obligate heterocomplex of two exostosin family members, EXT1 and EXT2, where each protein monomer contains distinct GT47 (GT-B fold) and GT64 (GT-A fold) glycosyltransferase domains. In this study, we generated human EXT1-EXT2 (EXT1-2) as a functional heterocomplex and determined its structure in the presence of bound donor and acceptor substrates. Structural data and enzyme activity of catalytic site mutants demonstrate that only two of the four glycosyltransferase domains are major contributors to co-polymer syntheses: the EXT1 GT-B fold β1,4GlcA transferase domain and the EXT2 GT-A fold α1,4GlcNAc transferase domain. The two catalytic sites are over 90 Å apart, indicating that HS is synthesized by a dissociative process that involves a single catalytic site on each monomer.

PubMed Disclaimer

Conflict of interest statement

Competing interests

The authors declare no competing interests.

Figures

Extended Data Fig. 1 |
Extended Data Fig. 1 |. Expression and purification of the EXT1–2 complex for structural studies.
a, Single and co-expression of EXT1 and EXT2 as small N-terminal fusion proteins (pGEn1 vector) and larger N-terminal fusion proteins (pGEn2 vector). SDS-PAGE of the expressed and purified construct combinations is shown indicating only co-expression of EXT1 and EXT2 in either fusion vector format leads to appreciable secretion of the respective fusion proteins. b, The purified EXT1–2 complex prior to TEV cleavage was characterized by size exclusion-multiangle light scattering (SEC-MALS). A280 is shown by the green line, refractive index in blue, light scattering in red and calculated molar mass in black. The molecular mass derived from SEC-MALS analysis (~223 kDa) is in close agreement with the predicted size of the heterodimeric EXT1–2 complex with the respective fusion tags (EXT1: 88 kDa + EXT2: 109 kDa + 4 N-glycans). b, The EXT1-pGEn1 and EXT2-pGEn2 expression constructs were co-transfected into HEK293F cells and the secreted heterocomplex in the media (Crude media) was purified by Ni2+-NTA purification (IMAC run-through, Wash 1, Wash 2) to yield a highly-enriched enzyme preparation (IMAC1 elution). The enzyme was concentrated (IMAC1 elution conc) and cleaved with TEV protease to remove the fusion tag sequences (+TEV). Ni2+-NTA chromatography separated the unbound EXT1–2 complex from the bound tag sequences and TEV protease (IMAC2 runthru), as the latter were all His-tagged. Panels a and b are each representative of data from two biological replicate experiments. Original uncropped images are provided in the Source Data.
Extended Data Fig. 2 |
Extended Data Fig. 2 |. Structure of EXT1–2.
a, Superimposition of GT-B fold of EXT1 and EXT2. The intervening loop in red dash cycle. b, Superimposition of GT-A fold from EXT1, EXT2, and mEXTL2 (PDB ID: 1ON6). c, Top view of GT-B fold in EXT1–2, suggesting a pseudo-twofold symmetry. d, Superimposition of GT-A fold from EXT1–2 and mEXTL2 (PDB ID: 1ON6) in top view, suggesting a pseudo-twofold symmetry for GT-A fold in EXT1–2. EXT1–2 is colored the same as Fig. 1, and mEXTL2 in cyan.
Extended Data Fig. 3 |
Extended Data Fig. 3 |. Comparison of 4-mer complex with EXT1–2.
a, Superimposition of EXT1–2 and 4-mer complex. b, Superimposition of EXT1 GT-B fold from EXT1–2 and 4-mer complex. 4-mer complex colored the same as in Fig. 1, and EXT1–2 in light gray. Residues involved in active pocket are shown as sticks. The disordered loop H300-C302 in EXT1–2 marked in red dashed oval. The density of NAG at Asn330 is shown as 50% transparency.
Extended Data Fig. 4 |
Extended Data Fig. 4 |. Comparison of 7-mer complex with EXT1–2, 4-mer complex, and EXT1–2 complexed with UDP-GlcA.
a, Superimposition of EXT1–2, 4-mer and 7-mer complex. b, Orthogonal view highlighting the GT-A fold comparison. 7-mer complex colored the same as in Fig. 1, EXT1–2 in light gray, and 4-mer complex in wheat. c, Superimposition of EXT1 GT-B fold from EXT1–2 complexed with UDP-GlcA and 7-mer complex. 7-mer complex are colored the same as in Fig. 1, and EXT1–2 complexed with UDP-GlcA in cyan. Residues involved in active pocket are shown as sticks.
Extended Data Fig. 5 |
Extended Data Fig. 5 |. Development of an HS co-polymerase assay for EXT1–2.
A co-polymerase assay was developed using the HS proteoglycan, glypican-1 (a and b), or heparosan primers (4-mer-pNP or 5-mer-pNP) (c) as acceptors and UDP-GlcA and UDP-GlcNAc (200 µM each) as donors. Individual assay components were tested in the reaction mixture and HS polymerization was detected using the UDP-Glo assay format (a and c). Recombinant human glypican-1 was expressed in HEK293 cells and purified for use as an acceptor substrate for HS extension of the proteoglycan. a, Low, but detectable activity was revealed when single UDP sugars were added to the reaction mix reflecting single sugar additions to the HS chains. When both UDP-GlcA and UDP-GlcNAc were added to the reaction, the enzyme activity increased ~40 fold indicating an iterative use of the sugar donors during the HS extension reaction. b, Reaction products were resolved by SDS-PAGE and Coomassie staining. Extended HS polymer product on glypican-1 appearing as a high molecular weight smear (red brackets) was detected only when enzyme and both sugar nucleotide donors were present in the co-polymerase reaction. The data are representative of n = 2 biologically independent samples. Original uncropped images are provided in the Source Data. c, UDP-Glo reactions using 4-mer-pNP and 5-mer-pNP as acceptors. Extension of the 4-mer primer with single sugar nucleotide donors only occurred when UDP-GlcA, but not UDP-GlcNAc, was used as donor as expected for an acceptor containing a non-reducing terminal GlcNAc residue. In contrast, extension of the 5-mer primer with single sugar nucleotide donors only occurred with UDP-GlcNAc, but not UDP-GlcA, was used as donor consistent with the presence of a non-reducing terminal GlcA residue on the acceptor. Addition of both donors in extension reactions with either the 4-mer or 5-mer as acceptor led to enhanced activity indicating an iterative use of the sugar donors during the HS extension reaction. Plots show the mean values (bar) ± s.d. (error bars) for n = 2 technical replicates (red circles).
Extended Data Fig. 6 |
Extended Data Fig. 6 |. Expression and purification of EXT1 and EXT2 mutants.
a. Diagrammatic representation of the positions of the EXT1 and EXT2 active site mutations that were generated. The respective GT-B and GT-A domains are indicated relative to the N-terminal GFP encoded by the pGEn2 vector. Residues for mutagenesis were chosen based on proximity to the respective GT-A or GT-B domain active sites. Truncation to form expression constructs encoding the respective single domains are also indicated. Termination codons were introduced in the linker regions between the GT-A and GT-B domains of EXT1 and EXT2 to generate the respective single GT-B domains (EXT1-GT-B and EXT2 GT-B). Each of the GT-A domains were isolated by PCR and transferred into the pGEn2 vector (EXT1 GT-A and EXT2 GT-A) to generate the respective single GT-A domains. Each of the mutants in the pGEn2 vector were co-expressed along with wild type versions of the corresponding partner EXT (in pGEn1 vector) to generate a secreted heterocomplex (for example EXT1 K269A mutant co-expressed with wild type EXT2) that was subsequently purified by Ni2+-NTA chromatography and resolved by SDS PAGE c, Initial expression level of the proteins in the culture media was monitored by GFP fluorescence b, since the mutant EXT form harbored a GFP fusion as a result of expression in the pGEn2 vector. The upper band on the SDS-PAGE gel (~115 kDa) corresponds to the mutant EXT form expressed in the pGEn2 vector, while the lower band (~90 kDa) corresponds to the wild-type EXT form (pGEn1 vector) that was co-expressed. Individual EXT1 and EXT2 isoforms and individual domains were also expressed singly or in combination. Of these latter expression tests, only EXT1 alone or the individual EXT2 GT-A or GT-B domains resulted in any appreciable secreted products. Each co-transfection experiment was performed once and two different SDS PAGE gels of the samples were generated. Data presented are representative of the respective experiments. Original uncropped images are provided in the Source Data.
Extended Data Fig. 7 |
Extended Data Fig. 7 |. Enzymatic extension of 4-mer and 5-mer acceptor primers with single sugar residues.
To complement the UDP-Glo assays performed in Fig. 4 and Extended Data Fig. 5, we analyzed the reaction products for glycan extension by MALDI-MS. Individual spectra were obtained for the indicated enzyme, sugar nucleotide donors (a and b), and the 4-mer-pNP (c) and 5-mer-pNP (e) acceptors. Electrospray mass spectra of each of the synthetic acceptors indicated a predicted parent mass of 897.75 and 1073.87 for the 4-mer-pNP and 5-mer-pNP acceptors, respectively (c and f). However, MALDI-MS of the compounds produced a broad pair of mass peaks where the smaller, higher mass peak matched the predicted mass (singly charged masses of 898 and 1074 for the 4-mer-pNP and 5-mer-pNP acceptor, respectively, insets for panels c and f at the top). For each compound, the more abundant second species was 15 mass units smaller than the predicted mass, consistent with an in-source rearrangement resulting in neutral mass loss during MALDI analysis. The cause of this 15 mass unit neutral loss is not clear at the present time, but it did not impact the ability of the 4-mer and 5-mer to act as acceptors for extension by EXT1–2. Reactions containing the 4-mer acceptor, which harbored a reducing terminal GlcNAc residue, could be extended by the mass of a GlcA unit in the presence of the UDP-GlcA donor (e), while reactions containing the UDP-GlcNAc donor led to no extension (d). By comparison, reactions containing the 5-mer acceptor (harboring a reducing terminal GlcA residue) could be extended by the mass of a GlcNAc residue in the presence of a UDP-GlcNAc donor (g), but reactions containing UDP-GlcA as donor led to no extension (h). Data presented are representative of n > 3 independent reactions.
Extended Data Fig. 8 |
Extended Data Fig. 8 |. Time course of co-polymer extension of 4-mer and 5-mer primers.
While single sugar residue extensions were observed on 4-mer and 5-mer primers in the presence of UDP-GlcA and UDP-GlcNAc, respectively, a time course of co-polymer extension was examined by the addition of both donors to reactions containing the 4-mer or 5-mer primers. Blank reactions containing boiled enzyme (red asterisk) and both sugar nucleotides (a), 4-mer-pNP (b) or 5-mer-pNP (f) were examined for comparison to reactions containing 4-mer-pNP, enzyme and both donors (c-e) or 5-mer-pNP, enzyme and both donors (g-i) in a time course extension. Both reactions showed a similar set of extension products with the 4-mer being extended by the addition of a GlcA residue to form a 5-mer product followed by a GlcNAc addition and subsequent extension of a disaccharide unit terminating in a GlcNAc sugar residue indicating that the GlcA transferase activity was rate limiting (c-e). A similar extension of the 5-mer-pNP was observed with the extension to an 18-mer over the course of the 1 h reaction (g-i). Data presented are representative of n > 3 independent reactions.
Extended Data Fig. 9 |
Extended Data Fig. 9 |. Comparison of the catalytic pockets of GT-A domains and GT-B domains explain why some of these domains are inactive.
a, b, Electrostatic potential of GT-B cleft in EXT1 and EXT2. Positive charged surface colored in blue, and negative charged surface in red. Superimposition of GT-B fold of EXT1 and EXT2. c, d, Sequence alignment and close-up view of UDP binding pocket of EXT1 and EXT2 GT-B. The key residues are marked and shown in sticks. e, Superimposition of active pocket in hEXT1 GT-A, hEXT2 GT-A, and mEXTL2 (PDB ID: 1ON6). The residues involved in substrate interaction shown in stick, and residues in hEXT2 are labeled, except for V487, P489, and K684 those not conserved in hEXT1. hEXT1 and hEXT2 colored the same as in Fig. 1, mEXTL2 colored in cyan. f, Sequence alignment of hEXT1, hEXT2, and mEXTL2. The residues involved in substrate interaction marked in star, the same one in hEXT1 and hEXT2 in magenta, while different one in black. The first (loop487–495) and last (loop-C) loop are marked in olive dashed frame.
Extended Data Fig. 10 |
Extended Data Fig. 10 |. Mapping of disease related missense mutations onto the structure of EXT1 and EXT2.
Mutations in the EXT1 and EXT2 coding regions that lead to disease were identified in the Human Gene Mutation Database and are shown on the linear representations of the respective protein sequences as colored circles (c and d) and as cyan spheres on the structural representation of EXT1 (a) and EXT2 (b). EXT1–2 are colored the same as Fig. 1. Disease characteristics shown in the legend are based on the annotations for the respective mutations in the Human Gene Mutation Database.
Fig. 1 |
Fig. 1 |. Overall architecture of the EXT1–2 complex.
a, Sketch of HS GAG biosynthesis. The GAG is initiated by synthesis of a penta-saccharide primer linked to a serine residue in the core protein. The GAG is elongated by the alternating addition of GlcA and GlcNAc catalyzed by EXT1–2. b, Two orthogonal views of cryo-EM 3D map of EXT1–2. c, The atomic model of EXT1–2 in the cartoon is shown in its likely orientation with respect to the Golgi membrane. The truncated N-terminal transmembrane α-helices are shown as cylinders. The orange arrow indicates a pseudo two-fold symmetry axis relating EXT1 GT-B and EXT2 GT-B. d, Domain organization of EXT1 and EXT2. The intervening loop connects the Rossmann-1 and Rossmann-2 subdomains of the GT-B domain. The linker loop connects the N-terminal GT-B and the C-terminal GT-A domains. e,f, Cartoon view of the atomic models of EXT1 and EXT2 shown separately. The view is the same as in the right panel of b. Domains shown in color are the same as in d; N-glycan on EXT2 Asn637 is in sticks. D1–4 refer to the four disulfide bonds in each domain. TM, transmembrane domain.
Fig. 2 |
Fig. 2 |. Structures of EXT1–2 in the presence of donor substrates.
a, Structure of EXT1–2 in the presence of the donor UDP-GlcA. The enzyme is active and has hydrolyzed the donor into UDP and free sugar GlcA. The freed GlcA has likely diffused into solution. Only EXT1 GT-B catalytic pocket is occupied by UDP that is shown in orange spheres, whereas the potential pockets of EXT1 GT-A, EXT2 GT-B and EXT2 GT-A are unoccupied. b, Close-up view of the EXT1 GT-B catalytic pocket superimposed with the UDP density shown as transparent light gray surface. c, Structure of EXT1–2 in the presence of the donor UDP-GlcNAc and MnCl2. Both EXT2 GT-A and EXT1 GT-B pockets are occupied by UDP (orange and purple spheres). EXT1 GT-A and EXT2 GT-B are unoccupied. d, Close-up view of EXT2 GT-A pocket with the UDP and Mn2+ density shown in transparent light gray surface. Key residues lining the catalytic pocket are shown in sticks and labeled in b and d.
Fig. 3 |
Fig. 3 |. Structures of EXT1–2 in the presence of acceptor substrates and UDP.
a–d, Close-up view of EXT1 GT-B (a) and GT-A (b) and EXT2 GT-B (c) and GT-A (d) in the 4-mer acceptor-bound EXT1–2 complex structure. The 4-mer binds only to the EXT1 GT-B domain, with three of its four sugars stabilized. e–h, Close-up view of EXT1 GT-B (e) and GT-A (f) and EXT2 GT-B (g) and GT-A (f) in the 7-mer acceptor-bound EXT1–2 complex structure. The 7-mer binds only to the EXT2 GT-A domain with four of the seven sugars stabilized. EXT1 and EXT2 are colored as in Fig. 1e,f. UDP, the 4-mer and 7-mer acceptors and key residues lining the potential active pockets are in sticks; H-bonds and salt bridges are shown as black dashed line.
Fig. 4 |
Fig. 4 |. Enzyme activity of wild-type and mutant forms of EXT1–2.
Wild-type, mutant or truncated forms of EXT1 or EXT2 were co-expressed or expressed as single proteins as indicated, and the recombinant products were purified and assayed for enzyme activity using recombinant glypican-1 (a) or HS oligosaccharides (b–e) as acceptors. a, HS co-polymerase assays were performed using glypican-1 as acceptor substrate and UDP-GlcA and UDP-GlcNAc as donors. Enzyme assays were also performed with 4-mer (b,c) or 5-mer (d,e) heparosan primer acceptors for the EXT1 mutants (b,d) or EXT2 mutants (c,e) using either UDP-GlcA (b,c) or UDP-GlcNAc (d,e) as donors. Mutation of the EXT1 GT-B active site led to complete loss of GlcA transferase activity (b), whereas mutations in EXT2 GT-A active site generally led to a loss of GlcNAc transferase activity (e), similarly to the loss on co-polymerase activity (a) for the same mutations. EXT1 expressed alone (EXT1 (GTA + GTB)) led to low but detectable co-polymerase, GlcA and GlcNAc transferase activities, whereas the individual EXT2 GT-A or GT-B domains expressed alone had minimal levels of enzyme activity. In each of the latter cases, the resulting purified proteins exhibited extensive proteolysis (Extended Data Fig. 6) that compromised the ability to quantitate the intact proteins (red asterisks), and the resulting enzyme activities are considered qualitative. Plots show the mean values (bar) ± s.d. (error bars) for n = 2 technical replicates (red circles). WT, wild-type.
Fig. 5 |
Fig. 5 |. Catalytic reaction mechanisms of EXT1 GT-B and EXT2-GTA.
a, Modeled Michaelis complex of EXT1 GT-B in the presence of donor UDP-GlcA and the three-sugar acceptor (*GlcNAc-GlcA-GlcNAc). b, Modeled Michaelis complex of EXT2 GT-A in the presence of the donor UDP-GlcNAc and the four-sugar acceptor (*GlcA-GlcNAc-GlcA-GlcNAc). In a and b, the acceptors are from our experimental structures. The binding modes and poses of the donors UDP-GlcA (a) and UDP-GlcNAc (b) were computed by Rosetta and are consistent with UDP binding in our structures and the UDP-sugar binding models in previously determined glycosyltransferase structures. EXT1 GT-B and EXT2 GT-A are colored as in Fig. 1. Substrates are in sticks. Key residues involved in donor and substrate binding are shown in sticks and labeled. c, Sketch for the potential SN1-type inverting reaction mechanism for EXT1 GT-B. d, Sketch for the established SNi-type retaining reaction mechanism of EXT2 GT-A. The residues marked in red are confirmed in our study to have significant impact on EXT1–2 activity.
Fig. 6 |
Fig. 6 |. Proposed model for GAG polymerization.
a, Two orthogonal surface views of EXT1–2. The UDP and acceptor substrates in the active pockets of EXT1 GT-B and EXT2 GT-A are shown as spheres. EXT1 and EXT2 are colored as in Fig. 1d. Left panel is in the same view as in Fig. 1c. b, EXT1 and EXT2 form a heterodimer that docks on the Golgi membrane. HSPG, usually a membrane-attached protein, moves close to EXT1–2 complex. The stem region of EXT1 and EXT2 may adopt extended and condensed conformation to accommodate the alternating addition of GlcA and GlcNAc for HS GAG chain elongation.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Carlsson P & Kjellen L in: Heparin—A Century of Progress (eds Lever R, Mulloy B & Page CP) 23–40 (Springer, 2012).
    1. Esko JD & Selleck SB Order out of chaos: assembly of ligand binding sites in heparan sulfate. Annu. Rev. Biochem 71, 435–471 (2002). - PubMed
    1. Sarrazin S, Lamanna WC & Esko JD Heparan sulfate proteoglycans. Cold Spring Harb. Perspect. Biol 3, a004952 (2011). - PMC - PubMed
    1. Bishop JR, Schuksz M & Esko JD Heparan sulphate proteoglycans fine-tune mammalian physiology. Nature 446, 1030–1037 (2007). - PubMed
    1. Kim SH, Turnbull J & Guimond S Extracellular matrix and cell signalling: the dynamic cooperation of integrin, proteoglycan and growth factor receptor. J. Endocrinol 209, 139–151 (2011). - PubMed

Publication types

  NODES
twitter 2