The Structural Basis for a Coordinated Reaction Catalyzed by a Bifunctional Glycosyltransferase in Chondroitin Biosynthesis

Background: Bifunctional enzyme K4CP polymerizes GlcA and GalNAc into a chondroitin chain.

Results: The different stages of polymerization are detailed using K4CP mutants.

Conclusion: K4CP coordinates these stages during polymerization, and a structural element is essential for this coordination.

Significance: A new basis for investigating polymerase reactions provides new avenues of research into template-less polymerization in glycosaminylglycans and their biological ramifications.

Abstract

Bifunctional chondroitin synthase K4CP catalyzes glucuronic acid and N-acetylgalactosamine transfer activities and polymerizes a chondroitin chain. Here we have determined that an N-terminal region (residues 58–134) coordinates two transfer reactions and enables K4CP to catalyze polymerization. When residues 58–107 are deleted, K4CP loses polymerase activity while retaining both transfer activities. Peptide ¹¹³DWPSDL¹¹⁸ within this N-terminal region interacts with C-terminal peptide ⁶⁷⁷YTWEKI⁶⁸². The deletion of either sequence abolishes glucuronic acid but not N-acetylgalactosamine transfer activity in K4CP. Both donor bindings and transfer activities are lost by mutating ⁶⁷⁷YTWEKI⁶⁸² to ⁶⁷⁷DAWEDI⁶⁸². On the other hand, acceptor substrates retain their binding to K4CP mutants. The characteristics of these K4CP mutants highlight different states of the enzyme reaction, providing an underlying structural basis for how these peptides play essential roles in coordinating the two glycosyltransferase activities for K4CP to elongate the chondroitin chain.

Introduction

Research into polysaccharide chains and their roles in biology dates back to 1918 when the anti-coagulant heparin was first purified and characterized from the liver (1). Since that initial discovery, many essential roles for polysaccharides have been established. Polysaccharide chains comprise the core structure of glycosaminoglycans and are present as O- or N-glycans in proteoglycans as well as in free polymers such as chondroitin, hyaluronan, and heparin (2). Glycosaminoglycans have been credited with controlling a diverse array of biological processes such as blood coagulation, cell division, adhesion, and bacterial, and viral infections (3). In addition, sulfation confers glycosaminoglycans with divergent biological functions from cell differentiation and morphogenesis (4) to fibroblast growth, nervous system, and cartilage development (5). The biosynthetic pathways of glycosaminoglycans are frequently altered in cancer cells; these alterations manifest in an array of forms, providing biological markers for the transformation process and progression of tumor cells (6).

Given their biological importance, various glycosyltransferases that are involved in the biosynthesis of glycosaminoglycans have been characterized, and their reaction mechanisms have been determined (7, 8). The majority of mammalian glycosyltransferases belong to the structural subclass of glycosyltransferases within the GT-A-fold group of enzymes and utilize the catalytic mechanisms for S_n2-type inverting (the α-linkage of the C1-O1 bond in the donor sugar is retained in the reaction product) and S_n1-type retaining (conversion to a β-configuration) transfer reactions (7, 8). Among glycosyltransferases, there are bifunctional glycosyltransferases that polymerize two different sugar molecules into chondroitin, hyaluronan or heparin/heparan chains. Although understanding the reaction mechanism of bifunctional glycosyltransferases is critical to investigating the biological functions and implications of glycosaminolyglycans in diseases, it remains unknown at the present time. Here we have utilized bacterial chondroitin synthase as an enzyme model for glycosaminoglycan chain polymerase to investigate this mechanism of polymerization. Of particular interest is whether or not the two transfer reactions are coordinated in the synthesis of a glucosaminoglycan chain. And if they are coordinated, what is the mechanism?

The K4 strain of Escherichia coli-produced chondroitin synthase K4CP is one such bifunctional glycosyltransferase that catalyzes β1–3 glucuronyltransfer and β1–4-N-acetylgalactosaminlytransfer reactions to polymerize glucuronic acid (GlcA) and N-acteylgalactosamine (GalNAc) into a chondroitin chain [GlcA β(1–3)-GalNAc β(1–4)]_n (9). K4CP consists of 686 amino acid residues, from which a truncated form was constructed by deleting the first 57 residues from the N terminus to produce K4CPΔ57. This deletion mutant fully retained the enzyme activity of K4CP. The x-ray crystal structure of K4CPΔ57 was recently determined (10). The K4CPΔ57 structure revealed that K4CP is a single globular protein consisting of two glycosyltransferase GT-A domains that are consistent with possessing S_n2-type GalNAc and GlcA transfer reactions at the N- and C-terminal domains, respectively. The N- and C-terminal domains orient their open access sites for donor substrates in directions perpendicular to one another, and their active sites do not share the same space within the K4CP molecule. What this x-ray crystal structure revealed posed a critical question with regard to the mechanism by which K4CP catalyzes the polymerization reaction; is this a random reaction? If it is not, then how does K4CP coordinate these two active sites, which are not in the same space and are positioned perpendicularly, to propel the polymerization reaction? Conversely, the K4CP structure also revealed the intriguing structural feature of a peptide consisting of residues 58–134 that wraps around the C-terminal domain before extending back into the N-terminal domain. Here we focus on this N-terminal peptide and examine its role in the polymerization reaction as catalyzed by K4CP.

Recombinant K4CPΔ57 and its mutants were subjected to assays to determine enzyme activity and to isothermal titration calorimetry (ITC)² analyses to characterize donor and acceptor substrate binding. Interaction between the ¹¹³DWPSDL¹¹⁸sequence within the N-terminal peptide with the peptide ⁶⁷⁷YTWKI⁶⁸² in the C-terminal region of the K4CP molecule was characterized as the regulatory motif that determines each of the two transfer reactions as well as coordinates the polymerization reaction. We have now generated K4CP mutants that represent different states of the polymerization reaction. These states are consistent with the hypothesis that interaction between specific N- and C-terminal peptides supports an underlying mechanism that coordinates the transfer reaction to induce polymerization.

EXPERIMENTAL PROCEDURES

Materials

Thrombin and trypsin were purchased from Sigma. Escherichia coli BL21 (DE3) and C41 (DE3) cells were produced by Agilent Technologies (Cary, NC) and Lucigen (Middleton, WI), respectively. pGEX plasmid was obtained from GE Healthcare. Primers were generated by Invitrogen. HEPES was procured from Sigma. CH polymer (a chemically desulfated derivative of CS-C from shark cartilage) was obtained from Seikagaku Corp. (Tokyo, Japan). UDP-GlcA, UDP-GalNAc, testicular hyaluronidase, and β-glucuronidase were purchased from Sigma. Ni-NTA^TM-agarose and anti-tetra His antibody came from Qiagen (Hilden, Germany). QuikChange^TM site-directed mutagenesis kit was obtained from Stratagene (La Jolla, CA). The Superdex^TM Peptide HR10/30 column, Superdex^TM 30 HiLoad 16/60 column, Q-Sepharose ion exchange resin, sulfo-N-hydroxysuccinimide-activated Sepharose beads, and the ECL detection system were from GE Healthcare. A semiquantitative SAX MAGNAM ion exchange column was purchased from Whatman (Clifton, NJ).

Preparation of CH6 and CH7

CH oligosaccharides were prepared from CH polymer as previously described (11). Briefly, for the preparation of even-numbered oligosaccharides such as CH6, CH polymer was digested with testicular hyaluronidase. For the preparation of odd-numbered oligosaccharides such as CH7, the hyaluronidase digests were further treated with β-glucuronidase at 37 °C. CH6 and CH7 were separated from these digests by chromatography on a Q-Sepharose ion exchange column and a Superdex 30 gel filtration column. The structures of the oligosaccharides were confirmed with MALDI-TOF Mass spectrometry (MS) spectrometer (AutoFlex, Bruker Daltonics, Bremen, Germany).

Site-directed Mutagenesis

Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene) following the protocols described in the accompanying instruction manual utilizing proper primers. The mutations were confirmed by sequencing with the Big Dye Terminator Cycle Sequencing Reaction kit (Applied Biosystems).

Purification of Recombinant Proteins

Escherichia coli BL21 (DE3) cells were transformed with a given pGEX plasmid in SOC medium, (Invitrogen) and the transformed cells were selected from a Luria-Bertani agar plate containing a 100 μg/ml concentration of ampicillin. Transformed cells grown in Luria-Bertani medium were inoculated into 2YT media containing 100 mg/ml ampicillin at 37 °C. When A₆₀₀ of the culture reached 0.6, the temperature was set to 23.5 °C, and isopropyl-1-thio-β-d-galactopyranoside (final concentration of 0.2 mm) was added 14 h before cells were harvested. Purification of protein and confirmation of protein purity was performed as previously reported (12).

Isothermal Titration Calorimetry

Isothermal titration calorimetry measurements were carried out in HEPES buffer using an iTC₂₀₀ MicroCalorimeter (GE Healthcare) at 20 °C. Substrate solutions containing UDP, UDP-GlcNAc, UDP-GalNAc, C6, and C7 at 4 mm or UDP-GlcA at 1 mm were injected into a reaction cell containing ∼100–200 μm protein. Thirty injections of 7 μl at 120 s intervals were performed. Data acquisition and analysis were performed by the MicroCal Origin software package. Data analysis was performed by generating a binding isotherm and best fit using the following fitting parameters: N (number of sites), ΔH (cal/mol), ΔS (cal/mol/deg), and K (binding constant in M⁻¹) and the standard Levenberg-Marquardt methods (13). After data analysis, K (m⁻¹) was then converted to K_d (μm).

Partial Proteolysis

One microgram of protein in HEPES buffer was incubated with 50, 5, or 0.5 ng of trypsin for 30 min at room temperature. Digestion was halted by adding 1 μl of 100 mm phenylmethylsulfonyl fluoride (Active Motif) and then boiling for 1 min. Samples were then loaded onto a NuPage 4–12% Bis-Tris gel (Invitrogen) with 6 μl of NuPage 4× LDS sample buffer (Invitrogen) and subjected to electrophoresis. Gel was then stained with Coomassie Brilliant Blue G 250 (Fluka).

Mass Spectroscopy

Two major bands stained with the Colloidal Blue Staining kit (Invitrogen) were subjected to mass spectrometric analysis. Gel bands were excised manually and digested with trypsin (Promega) for 8 h in an automated fashion with a Progest In-gel Digester from Genomics Solutions. Samples were lyophilized to dryness and resuspended in 50:50 (v/v) 0.2% formic acid:acetonitrile. Samples (0.3 μl) were then spotted onto a 192-sample stainless steel MALDI plate and mixed on target with 0.3 μl of 33% saturated α-cyano-hydroxycinnamic acid. MS and tandem mass spectrometry (MS/MS) were then performed with the use of an Applied Biosystems 4700 Proteomics Analyzer in the positive ion and reflector modes, respectively. The MS was calibrated internally using autolytic tryptic peptides, and the MS/MS was calibrated externally using the fragment ions of the angiotensin I (M+H)⁺ ion (m/z 1296.68). A focus mass of m/z 2000 was used for the MS acquisition. For the MS/MS, 1000 V was used for the collision energy, and argon was used as the collision gas with a recharge threshold set at 1.0 × 10⁻⁷ torr. Protein identification was then performed by interrogating both MS and MS/MS using the MASCOT search engine and the entire NCBI non-redundant database. Search parameters included an allowance of two missed tryptic cleavages, a 0.06-Da mass tolerance for the MS data, a 0.1-Da mass tolerance for the MS/MS data, and an allowance for variable oxidation of methionine residues.

Enzyme Assays

GalNAc transfer, GlcA transfer, and chondroitin polymerase activities of the recombinant enzymes were measured using radioisotope donor substrates as described previously (13) with a slight modification; for the GalNAc transfer activity assay, a 50-μl mixture containing 50 mm Tris-HCl (pH 7.2), 20 mm MnCl₂, 0.15 m NaCl, UDP-[³H]GalNAc (3 nmol, 0.1 μCi) as the donor substrate and 1 nmol of chondroitin hexasaccharide as the acceptor substrate was incubated with the recombinant enzymes (2.0 μg) at 30 °C for 60 min and then heated in boiling water. For the GlcA transfer activity assay, UDP-[¹⁴C]GlcA (3 nmol, 0.1 μCi) and 1 nmol of chondroitin pentasaccharide were used as the donor and acceptor substrates, respectively. For the polymerase activity assay, UDP-[³H]GalNAc (3 nmol, 0.1 μCi) and UDP-GlcA (3 nmol) were used as the donor substrates, and 0.1 nmol of chondroitin pentasaccharide was used as the acceptor substrate. The radiolabeled saccharides were separated by a Superdex Peptide column and measured by a liquid scintillation counter. The enzyme activities were determined by calculating the amount of the incorporated radioactive sugars.

RESULTS

Effect of Deleting the N-terminal Peptide on K4CP Activity

The N-terminal peptide (residues 58–134), which wraps around the C-terminal domain of K4CPΔ57 in the x-ray crystal, comprises a linear structure that contains a random coil and three α-helices (see supplemental Fig. 1A for locations). This peptide was successively deleted to produce mutants K4CPΔ95, K4CPΔ101, K4CPΔ107, and K4CPΔ113 (Fig. 1). Subsequently, ITC was employed using these deletion mutants to determine their bindings to the donor substrates UDP-GalNAc and UDP-GlcA. The K4CPΔ95 mutant, which removed the first two α-helices, retained similar K_d values for binding to UDP-GalNAc and UDP-GlcA to those observed with K4CPΔ57 (Table 1). Therefore, K4CPΔ95 was further deleted by six amino acid residues at a time to produce K4CPΔ101, K4CPΔ107, and K4CPΔ113 (Fig. 1). The K4CPΔ101 mutant exhibited K_d values for binding to UDP-GlcA and UDP-GalNAc similar to those of K4CPΔ57 and K4CPΔ95 (Table 1). Mutant K4CPΔ107 maintained a K_d value for UDP-GlcA binding similar to that of the K4CPΔ95 mutant while exhibiting a significant decrease in that of UDP-GalNAc binding. With its further deletions, K4CPΔ113 lost binding to UDP-GalNAc while retaining UDP-GlcA binding. The donor substrate product UDP bound to K4CPΔ95 but not to K4CPΔ107 or K4CPΔ113 (Table 1).

FIGURE 1.

Schematic representation of the K4CP molecule and its mutants. The full-length K4CP consists of 686 amino acid residues and K4CPΔ57 deleted in first 57 residues from the N terminus. K4CPΔ57 was further deleted to generate K4CPΔ95, K4CPΔ101, K4CPΔ107, and K4CPΔ113.

TABLE 1.

Donor substrate binding of N-terminal truncated K4CP enzymes

The results obtained from ITC analysis are presented: thermodynamic parameters ΔS (cal/mol/degree), n (number of binding sites), and the binding constant, K_d (μm) for UDP, UDP-GlcA, and UDP-GalNAc at 20 °C. ND signifies a reaction in which no binding was detected.

Enzyme	UDP-GalNAc			UDP-GlcA			UDP
	ΔS	Kd	n	ΔS	Kds	n	ΔS	Kd	n
	cal/mol/degree	μm		cal/mol/degree	μm		cal/mol/degree	μm
K4CPΔ57	−22.61	358.17 ± 26.64	1.25 ± 0.32	−5.29	2.75 ± 0.19	1.00 ± 0.03	−10.96	89.53 ± 22.77	0.91 ± 0.09
K4CPΔ95	−1.84	494.56 ± 26.24	1.14 ± 0.33	−16.70	2.99 ± 0.06	1.02 ± 0.05	−18.70	143.88 ± 35.65	0.94 ± 0.14
K4CPΔ101	−6.20	429.18 ± 77.66	0.94 ± 0.04	−16.30	5.92 ± 1.31	0.89 ± 0.11	10.80	89.29 ± 14.55	0.92 ± 0.12
K4CPΔ107	0.29	61.73 ± 20.61	0.81 ± 0.06	2.34	8.55 ± 3.27	0.96 ± 0.15	ND	ND	ND
K4CPΔ113	ND	ND	ND	17.80	3.45 ± 1.65	0.90 ± 0.04	ND	ND	ND

Given these donor substrate interactions, the K4CP deletion mutants were then subjected to enzyme assays to determine GalNAc and GlcA transfer and polymerase activities (Table 2). K4CPΔ95 and K4CPΔ101 catalyzed these three activities as effectively as K4CPΔ57. K4CPΔ107 abrogated polymerase activity while fully retaining both GalNAc and GlcA transfer activities. K4CPΔ113 retained levels of UDP-GalNAc transfer activity that were decreased by 50% while virtually abrogating UDP-GlcA transfer activity; as expected, K4CPΔ113 did not catalyze the polymerization reaction. Thus, the deletions of the N-terminal peptide resulted in generating K4CP mutants with diverse enzymatic features. Among them, K4CPΔ107 provided the most critical insight into the nature of K4CP; K4CP needs to coordinate its two transfer activities to catalyze the polymerase reaction and residues 101–113 are critical for this coordination to occur.

TABLE 2.

GalNAc and GlcA transfer and polymerase activities

Transfer and chondroitin polymerase activities of the recombinant enzymes were measured using radioisotope donor substrates as described under “Experimental Procedures.” K4CPΔ57 activity is presented as having full (100%) activity, and the percentage of activity possessed by the mutant constructs is described relative to that of K4CPΔ57.

Enzyme	GalNAc-T		GlcA-T		Polymerase
	pmol/min/μg	%	pmol/min/μg	%	pmol/min/μg	%
K4CPΔ57	1.54 ± 0.27	100.0	6.12 ± 0.76	100.0	0.527 ± 0.031	100.0
K4CPΔ95	1.56 ± 0.30	101.0	5.62 ± 0.39	92.8	0.453 ± 0.071	85.9
K4CPΔ101	1.54 ± 0.27	99.9	6.00 ± 0.67	97.9	0.515 ± 0.066	97.7
K4CPΔ107	1.27 ± 0.10	82.3	5.43 ± 0.28	87.2	0.095 ± 0.033	18.1
K4CPΔ113	0.84 ± 0.09	54.8	0.48 ± 0.01	8.0	0.007 ± 0.002	1.4

Our previous ITC analysis of donor substrate binding demonstrated that UDP-GalNAc does not bind to the N-terminal active site where GalNAc transfer occurs unless the C-terminal binding motif DSD is inactivated by mutation to ASA (12). Therefore, the mutant constructs K4CPΔ101 ASA, K4CPΔ107 ASA, and K4CPΔ113 ASA were generated to examine UDP-GalNAc binding to their N-terminal active sites. ITC analysis on these ASA mutants confirmed that all of these ASA mutants bind UDP-GalNAc to their N-terminal active sites (Table 3), supporting the fact that K4CPΔ101, K4CPΔ107, and K4CPΔ113 catalyzed GalNAc transfer activity (Table 1). Noticeably, these ASA mutants exhibited K_d values that were 6–20-fold lower for UDP binding as compared with those for UDP-GalNAc binding. However, these higher UDP bindings did not prevent them from catalyzing GalNAc transfer at the N-terminal active site.

TABLE 3.

UDP and UDP-GalNAc binding to the N-terminal active site of truncated K4CP enzymes utilizing their ASA mutant

The results obtained from ITC analysis are presented: thermodynamic parameters ΔS (cal/mol/degree), n (number of binding sites), and the binding constant, K_d (μm) for UDP and UDP-GalNAc at 20 °C.

Enzyme	UDP			UDP-GalNAc
	ΔS	Kd	n	ΔS	Kd	n
	cal/mol/degree	μm		cal/mol/degree	μm
K4CPΔ101 ASA	−0.04	20.44 ± 7.26	1.06 ± 0.05	−5.93	362.32 ± 46.06	1.00 ± 0.09
K4CPΔ107 ASA	3.53	18.97 ± 6.52	0.89 ± 0.13	12.30	103.89 ± 7.95	1.27 ± 0.14
K4CPΔ113 ASA	18.90	1.48 ± 0.74	0.89 ± 0.02	18.80	22.52 ± 7.25	1.15 ± 0.08

The Peptides That Determine K4CP Activity

Given that K4CPΔ107 altered enzymatic activity, a partial proteolysis experiment was employed to test the hypothesis that deletion of residues 58–107 affected the K4CP structure in such a manner that resulted in altered enzyme activity. Peptide fragments, which were generated from digested K4CPΔ57 and K4CPΔ107, were then separated by SDS-PAGE and subjected to mass spectroscopy (supplemental Fig. 2). A peptide fragment consisting of nine amino acids found in the C terminus beginning with Tyr-677 and ending with Leu-686 was uniquely generated from the K4CPΔ107 protein (supplemental Table 3). Analysis of the x-ray structure of K4CPΔ57 revealed that the peptide ⁶⁷⁷YTWEKI⁶⁸² within this fragment forms an interface with the N-terminal peptide ¹¹³DWPSDL¹¹⁸ in the K4CP molecule: Tyr-677, Trp-679, and Lys-681 form hydrogen bonds with Asp-113, Pro-115, Asp-117, and Leu-118 within ¹¹³DWPSDL¹¹⁸. This ¹¹³DWPSDL¹¹⁸ peptide appeared to be involved in determining K4CP enzyme activity.

The determining role of the interaction between ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸² was further investigated by internally deleting these peptides from K4CPΔ57. K4CPΔDWPSDL was capable of binding to both UDP-GlcA and UDP-GalNAc (Table 4). Despite binding to UDP-GlcA, K4CPΔDWPSDL nearly abrogated all GlcA transfer and polymerase activities (Table 5). K4CPΔDWPSDL ASA confirmed that K4CPΔDWPSDL retained GalNAc transfer activity at the N-terminal active site, although ITC analysis did not detect UDP-GalNAc binding to this deletion mutant. As expected, K4CPΔDWPSDL ASA completely eliminated the ∼1% residual GlcA transfer activity that remained in the K4CPΔDWPSDL mutant. UDP did not bind to either K4CPΔDWPSDL or K4CPΔDWPSDL ASA. Similar to K4CPΔDWPSDL, K4CPΔYTWEKI, which internally deleted ⁶⁷⁷YTWEKI⁶⁸², bound to both UDP-GalNAc and UDP-GlcA and retained GalNAc transfer activity; however, GlcA transfer and polymerase activities were completely abrogated (Tables 4 and 5). Thus, K4CPΔDWPSDL and K4CPΔYTWEKI decoupled UDP-GlcA binding from GlcA transfer activity in K4CPΔ57. In an alternate to internal deletions, Tyr-677, Trp-679, and Lys-681 within ⁶⁷⁷YTWEKI⁶⁸² were simultaneously substituted with Asp, Ala, and Asp, respectively, to disrupt the interactions between these two peptides. The triple mutants K4CP YWKpm and K4CP YWKpm ASA abolished all function of K4CP, and no donor substrate binding and no enzyme activities were detected (Tables 4 and 5). On the other hand, both K4CP YWKpm and K4CP YWKpm ASA bound UDP at K_d values of around 1 μm at both N- and C-terminal active sites (Table 4).

TABLE 4.

Regulation of enzyme functions by the interaction between ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸²

ITC analysis for these mutants were performed as described under “Experimental Procedures,” and the results are presented as designated in the legend of Table 1. ND signifies a reaction in which no binding was detected.

Enzyme	UDP-GalNAc			UDP-GlcA			UDP
	ΔS	Kd	n	ΔS	Kd	n	ΔS	Kd	n
	cal/mol/degree	μm		cal/mol/degree	μm		cal/mol/degree	μm
K4CPΔDWPSDL	14.90	101.42 ± 26.46	1.11 ± 0.11	22.80	0.96 ± 0.35	1.36 ± 0.19	ND	ND	ND
K4CPΔDWPSDL ASA	ND	ND	ND	ND	ND	ND	ND	ND	ND
K4CPΔYTWEKI	6.53	19.72 ± 7.14	1.17 ± 0.14	−15.60	3.52 ± 0.98	0.97 ± 0.03	15.10	14.77 ± 5.09	0.71 ± 0.12
K4CP YWKpm	ND	ND	ND	ND	ND	ND	19.70	0.91 ± 0.34	1.31 ± 0.05
K4CP YWKpm ASA	ND	ND	ND	ND	ND	ND	22.60	1.21 ± 0.14	1.00 ± 0.13

TABLE 5.

Regulation of enzyme functions by the interaction between ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸²

GalNAc and GlcA transfer and polymerase activities. Enzyme assays were performed as described under “Experimental Procedures,” and the results are presented as those in Table 1.

Enzyme	GalNAc-T		GlcA-T		Polymerase
	pmol/min/μg	%	pmol/min/μg	%	pmol/min/μg	%
K4CPΔDWPSDL	1.00 ± 0.16	65.0	0.06 ± 0.01	1.0	0.01 ± 0.00	1.1
K4CPΔDWPSDL ASA	1.02 ± 0.16	65.6	0.00 ± 0.00	0.0	0.00 ± 0.00	0.0
K4CPΔYTWEKI	1.49 ± 0.02	100.0	0.08 ± 0.01	1.5	0.00 ± 0.00	0.0
K4CP YWKpm	0.00 ± 0.00	0.0	0.00 ± 0.00	0.0	0.00 ± 0.00	0.0

Binding of Acceptor Substrates CH6 and CH7

ITC analysis for acceptor substrates with K4CPΔ57, K4CPΔ57 ASA, and K4CPΔ57 ACA determined that CH6 (GlcA at the non-reducing end) and CH7 (GalNAc at the non-reducing end) bind to the N- and C-terminal active sites, respectively (Table 4). Acceptor bindings remain constant in K4CPΔ107, which abrogates polymerase activity while retaining transfer activities. In addition, the binding of CH6 to K4CPΔ57 ASA, but not to the K4CPΔ57, indicates that CH6 binding to the N-terminal active site is regulated by the C-terminal active site. On the other hand, CH7 binding to the C-terminal active site is not controlled by the N-terminal active site. The characteristics of these acceptor bindings are reminiscent of the donor substrates. Although K4CP YWKpm ASA neither binds to UDP-GalNAc nor catalyzes GalNAc transfer activity, this mutant retains binding to donor substrate CH6 (Table 6). Likewise, CH7 binding is retained by K4CPΔYTWEKI (Table 7), which does bind to the donor substrate UDP-GlcA but does not catalyze GlcA transfer activity. Thus, K4CPΔYTWEKI appears to alter the substrate binding conformation so that this mutant loses transfer activity.

TABLE 6.

C6 acceptor substrate binding to the N-terminal active site of truncated K4CP enzymes utilizing their ASA mutants

Non-ASA mutants are presented as controls. ND signifies a reaction in which no binding was detected.

Enzyme	C6 acceptor substrate
	ΔS	Kd	n
	cal/mol/degree	μm
K4CPΔ57	ND	ND	ND
K4CPΔ57 ASA	28.40	0.25 ± 0.01	0.86 ± 0.06
K4CPΔ107	ND	ND	ND
K4CPΔ107 ASA	10.80	8.78 ± 0.88	0.92 ± 0.05
K4CP YWKpm	ND	ND	ND
K4CP YWKpm ASA	12.90	8.47 ± 3.43	1.23 ± 0.21

TABLE 7.

C7 acceptor substrate binding to the C-terminal active site of K4CP enzymes

The results obtained from ITC analysis are presented: the thermodynamic parameters ΔS (cal/mol/degree), n (number of binding sites), and the binding constant, K_d (μm) for C6 and C7 at 20 °C.

Enzyme	C7 acceptor substrate
	ΔS	Kd (μm)	n
	cal/mol/degree
K4CPΔ57	16.30	29.59 ± 7.12	1.00 ± 0.09
K4CPΔ57 ACA	8.45	35.71 ± 4.31	1.47 ± 0.08
K4CPΔ107	15.80	71.94 ± 6.19	1.11 ± 0.09
K4CPΔYTWEKI	12.80	59.88 ± 11.41	1.36 ± 0.18

DISCUSSION

The bifunctional glycosyltransferase chondroitin synthase K4CP alternatively transfers GalNAc and GlcA at the N- and C-terminal active sites, respectively, to polymerize them into the chondroitin chain. A characteristic of this polymerization reaction is the fact that there is no template to assist K4CP with the reaction, as compared with the polymerization reactions catalyzed by DNA and RNA polymerases and peptide synthesis. Our study utilized K4CP mutants and defined the different states that occur during the polymerization reaction. Moreover, the peptide interaction between ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸² has been characterized as an essential factor that determines transfer activities and enables the polymerization reaction. These findings are consistent with the hypothesis that K4CP possess an intrinsic mechanism within itself to coordinate the two transfer reactions, enabling K4CP to extend the chondroitin chain.

Based on enzyme activities, K4CPΔ57 mutants can be organized into three different groups that possess structural features that could represent distinct stages of the enzyme reaction: I, II, and III (Fig. 2A). Stage I enzyme possesses both GlcA and GalNAc transfer activities but no polymerase activity; stage II possesses GalNAc transfer activity but no GlcA transfer or polymerase activities; stage III comprises an inert enzyme with no transfer or polymerase activities. Based on these stages the reaction cycle of the proposed polymerization cycle is depicted in Fig. 2B. In rejecting the notion that polymerization is a random reaction, the first compelling evidence in support of the concept of a coordinated reaction mechanism came when K4CPΔ107 was found to fully retain both donor and acceptor substrate binding as well as transfer activities, but polymerization activity was abolished (stage I). K4CPΔ107 does not remove ¹¹³DWPSDL¹¹⁸ but still appears to destabilize the interaction of this signature peptide with ⁶⁷⁷YTWEKI⁶⁸² as indicated by our present partial proteolysis of this deletion mutant and mass spectroscopic identification of the digested peptide. Therefore, acting as an interdomain mechanism, this instability causes a structural disconnect between the two glycosyltransfer activities, disabling the enzyme ability to coordinate these activities and elongate the chondroitin chain.

FIGURE 2.

Schematic representation of the proposed different stages of the polymerization reaction. A, boxes represent domains, and circles indicate their active sites. The red circles with DSD and DCD indicate an active site with donor binding (stage I). A blue circle with DCD signifies an inactive site with donor binding (stage II). The green circles with UDP denote the inactive sites that exhibited strong UDP binding (stage III). The observed transfer and polymerase activities are shown below for each stage. B, the proposed sequence of the enzyme reaction: red and blue circles show sites that are enzymatically active and inactive, respectively. The reaction starts at the C-terminal active site transferring GlcA to GalNAc at the non-reducing end of the oligosaccharide (n = 1), during which the N-terminal active site is inactive and could be occupied by UDP; after this first transfer reaction, the product moves into the N-terminal active site from which GalNAc is transferred, during which the C-terminal active site remains occupied by UDP-GlcA (because of the UDP-GlcA high binding affinity) and is inactive; the second transferred product then moves back to the C-terminal active site. Once a single reaction cycle is completed, then chondroitin chain is elongated n1 to n2. The dots between the N-terminal peptide and the C-terminal domain indicate their interactions. Open circles with N indicate the N terminus of K4CPΔ57 molecule, and the N-terminal peptide interacts with the C-terminal domain.

In support of the concept that the peptides ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸² are essential for the polymerization reaction to occur, the deletion of either ¹¹³DWPSDL¹¹⁸ (K4CPΔDWPSDL) or ⁶⁷⁷YTWEKI⁶⁸² (K4CPΔYTWEKI) abolished the enzyme ability to catalyze polymerization. However, unlike K4CPΔ107, which retained both GalNAc and GlcA transfer activities, K4CPΔDWPSDL and K4CPΔYTWEKI lost GlcA transfer activity while retaining GalNAc transfer activity, thereby suggesting that deletion of one of these two peptides abolishes GlcA, but not GalNAc, transfer activity, which may represent a distinct step during polymerization reaction (stage II). K4CPΔDWPSDL ASA in fact confirmed that the GalNAc transfer activity is retained in K4CPΔDWPSDL where it could be catalyzed at the N-terminal active site. Although it remains a question as to why ITC analysis did not detect UDP-GalNAc binding to the N-terminal active site (Table 4), this donor binding could have occurred in a manner that did not allow for detection by ITC. Despite possessing binding ability to both UDP-GlcA and C7 substrates at the C-terminal active site, K4CPΔYTWEKI was unable to catalyze GlcA transfer activity (stage II). Noticeably, K4CPΔYTWEKI strengthened its UDP binding affinity (K_d values 6-fold lower than those observed with K4CPΔ57) to be equivalent to UDP-GlcA binding constant at the C terminus. These changes in UDP binding indicate that the deletion mutants alter a portion of the active site structure to where the UDP moiety of UDP-sugar molecule binds; this alteration may have repressed GlcA transfer activity in K4CPΔYTWEKI.

A simultaneous triple mutation of the peptide YTWEKI (K4CP YWKpm) appears to alter K4CP structure differently from complete deletion of the peptide (K4CPΔYTWEKI). K4CP YWKpm and K4CP YWKpm ASA have provided experimental evidence indicating that K4CP can adopt structural features that force the enzyme to be totally free from donor substrate binding as well as catalytic activities while retaining CH6 and CH7 acceptor substrate binding to their respective sites (stage III). Because K4CP YWKpm ASA retains C6 acceptor substrate binding, the acceptor substrate cannot be the direct cause of repression of transfer activity. It is intriguing that we did not encounter a K4CP mutant that represses GalNAc transfer at the N-terminal active site while proceeding with GlcA transfer activity at the C-terminal active site, which should exist during the polymerization reaction. K4CP YWKpm and K4CP YWKpm ASA exhibit high affinity UDP binding (K_d values around 1 μm) to each active site, possibly resetting K4CP for the next round of the catalytic cycle to elongate the chondroitin chain. Because the binding affinity of UDP-GlcA at the C-terminal active site in K4CPΔ57, but not UDP-GalNAc, is equivalent to the UDP binding in K4CP YWKpm, stage III may represent an enzymatic state that precedes a subsequent structural alteration allowing UDP-GlcA binding to initiate the new-round of the reaction. Therefore, these structural features may enable K4CP to coordinate transfer reactions to elongate the chondroitin chain. The possibility of these structural features being conserved in other bi-functional glycosyltransferases is intriguing. Given that K4CP is the only such transferase whose structure has been solved, future investigations will have to determine whether other bifunctional glycosyltransferases possess peptides that interact in a manner similar to that of ¹¹³DWPSDL¹¹⁸ and ⁶⁷⁷YTWEKI⁶⁸² and the role this interaction in elongating carbohydrate chains.

In conclusion, these K4CP mutants exhibit at least three different states of the polymerization reaction that can be integrated into a hypothetical scheme for the overall reaction cycle used by K4CP to elongate chondroitin chains (Fig. 2). Because acceptor substrates remain bound to their respective active sites, donor substrate binding appears to be the determinant for K4CP's ability to coordinate the two transfer activities and polymerize chondroitin chains. The reaction may start by UDP-GlcA binding to the C-terminal active site of Stage III, transferring GlcA to the non-reducing end of the oligosaccharide. Then the produced oligosaccharide moves into the N-terminal active site of Stage II by virtue of the regulation imposed by the C-terminal active site and GalNAc transfer follows. Then the second transferred product may move back to the C-terminal active site. K4CP is endowed with an intrinsic molecular mechanism that may utilize the interaction of the N-terminal ¹¹³DWPSDL¹¹⁸ with the ⁶⁷⁷YTWEKI⁶⁸² peptide of the C-terminal domain to coordinate GalNAc and GlcA transfers and elongate the chondroitin chain. In this scheme, understanding the structural basis for why K4CP mutants possess both transfer activities but not polymerase activity will be most critical for us to determine the molecular mechanism of the polymerization reaction. With this in mind, solving the structural features that connects the ¹¹³DWPSDL¹¹⁸ peptide with the C-terminal domain, an area for which no electron density was detected in the current K4CP structure, may be critical to unifying these observed snapshots at stages during the polymerization process to fully decipher the molecular-based regulatory machinery that confers K4CP the ability to coordinate its polymerization reaction.