Biosynthesis of the glycosaminoglycan precursor UDP-α-D-glucuronic acid occurs through a 2-fold oxidation of UDP-α-D-glucose that is catalysed by UGDH (UDP-α-D-glucose 6-dehydrogenase). Structure–function relationships for UGDH and proposals for the enzymatic reaction mechanism are reviewed in the present paper, and structure-based sequence comparison is used for subclassification of UGDH family members. The eukaryotic group of enzymes (UGDH-II) utilize an extended C-terminal domain for the formation of complex homohexameric assemblies. The comparably simpler oligomerization behaviour of the prokaryotic group of enzymes (UGDH-I), in which dimeric forms prevail, is traced back to the lack of relevant intersubunit contacts and trimmings within the C-terminal region. The active site of UGDH contains a highly conserved cysteine residue, which plays a key role in covalent catalysis. Elevated glycosaminoglycan formation is implicated in a variety of human diseases, including the progression of tumours. The inhibition of synthesis of UDP-α-D-glucuronic acid using UGDH antagonists might therefore be a useful strategy for therapy.
UDP-GlcA (UDP-α-D-glucuronic acid), the activated form of D-glucuronic acid in cellular metabolism, plays a central role in a variety of biosynthetic pathways as well as in detoxification processes. UDP-GlcA is the donor substate for different UDP-glucuronosyltransferases, promoting the incorporation of D-glucuronosyl residues into nascent glycosaminoglycans such as hyaluronan  or catalysing O-glucuronidation of small molecules in xenobiotic metabolism . In addition, UDP-GlcA functions as a precursor in the biosynthesis of different carbohydrates, for example UDP-α-D-xylose and L-ascorbate . UGDH (UDP-α-D-glucose 6-dehydrogenase) (EC 18.104.22.168) is the enzyme responsible for formation of UDP-GlcA in a multitude of organisms ranging from bacteria to mammals. UGDH catalyses a 2-fold oxidation of UDP-Glc (UDP-α-D-glucose) using an NAD(P), typically NAD+, as the oxidant. The overall reaction catalysed by UGDH is:
UGDH was first described [4,5] and biochemically characterized [6–23] from bovine liver. The enzyme has since been reported from a variety of organisms [24–34] and UGDH-encoding genes have been found in members of all three domains of life [24,31,34]. UGDH from mammals has recently attracted special interest, as it was established that elevated levels of the matrix glycosaminoglycan hyaluronan are directly implicated in the progression of various forms of epithelial cancer [35–41]. Reduced formation of UDP-GlcA restricts hyaluronan production and this in turn was shown to slow tumour growth [42,43]. UGDH was recently proposed as a biomarker for prostate cancer . Inhibition of UGDH therefore presents a potential target for novel therapeutic strategies. Well defined structure–function relationships for UGDH are instrumental in a rational development of enzyme antagonists. The present review provides an update on the properties of UGDH and describes proposals for the catalytic mechanism of the enzyme. Inhibition of UGDH is also discussed.
Sequence-based classification and structural analysis of UGDH
More than 1000 open reading frames are retrieved from the UniProt database using ‘EC 22.214.171.124’ as a search query. Supplementary Figure S1 (at http://www.biochemsoctrans.org/bst/038/bst0381378add.htm) shows a multiple sequence alignment for a representative selection of UGDH enzymes. The crystal structures of UGDH from Streptococcus pyogenes (SpUGDH; PDB codes 1DLI and 1DLJ)  and Caenorhabditis elegans (CeUGDH; PDB code 2O3J) were used to obtain relevant structure–function assignments. In Supplementary Figure S1, secondary-structural elements of CeUGDH and SpUGDH are mapped on the corresponding linear sequence of each enzyme. UGDH folds into two domains that both adopt a highly similar α/β-fold with a β-sheet core flanked by several α-helices. The two domains are connected by a central α-helix (CeUGDH α10 and SpUGDH α9), which is approx. 48 Å (1 Å=0.1 nm) long and bridges the last strand of the N-terminal region and the first helix of a three-helical bundle of the C-terminal domain (Figure 1A) . The N-terminal domain is responsible for binding NAD+, whereas the C-terminal domain primarily contributes residues for binding of the UDP moiety of UDP-Glc. Catalysis takes place at the interdomain cleft (see below), and the active site of the enzyme is situated on top of the central interdomain α-helix.
Structural comparison of UGDH-I and UGDH-II enzymes
UGDH enzymes can be categorized according to molecular size into a group (UGDH-I) consisting of proteins whose length of ~380–460 amino acids is substantially shorter than that of proteins of the second group (UGDH-II; ~480–500 amino acids). Subclassification of UGDHs into two groups appears to reflect the source of the enzymes. UGDH-I members are typically from lower organisms of the domains Prokaryota and Archaea, whereas enzymes belonging to UGDH-II are from higher and lower Eukaryota. Membership of UGDH-I is also indicated by a conserved internal peptide whose consensus sequence is AEXXK(Y/L)-(F/A)XNX(F/Y)LAX(K/R)(I/V)(S/A)(F/Y)(I/F)N(E/D), where X is any amino acid and the amino acids in parentheses represent alternatives at the same position (Supplementary Figure S1). The characteristic signature for the corresponding internal region of UGDH-II is WS(S/A)ELSKLXANA(F/M)LAQRISS(I/V)N(S/A)XSA-(I/L/V)CEATGA, as shown in Supplementary Figure S1. In both UGDH-I and UGDH-II, the conserved internal peptide is responsible for formation of the interdomain α helix (CeUGDH α10 and SpUGDH α9) that represents a key secondary-structural component of the enzyme subunit and serves as the core of the dimer interface in SpUGDH (Figure 1A). The α9 helix was shown to contribute 37% of the interface solvent-inaccessible surface area in the crystallographic dimer of SpUGDH. The largest portion (52%) of this surface area is provided by the C-terminal domain (α10–α12) and, interestingly, differences between enzymes of UGDH-I and UGDH-II are most pronounced in the C-terminal region (Supplementary Figure S1). Residues involved in making dimer contacts in SpUGDH are relatively well conserved among UGDH-I members, but differ compared with the UGDH-II group. Supplementary Figure S1 therefore supports a distinction between UGDH-I and UGDH-II on the basis of the interactions responsible for dimer and higher oligomer contact formation. In SpUGDH, a network of 31 hydrogen bonds stabilize the dimer interface. Within UGDH-I, aromatic amino acids dominate the dimer interface, mainly residues of α9 (Phe206, Tyr210, Tyr217 and Tyr224) and α11 (Tyr272)  as indicated in Supplementary Figure S1. Corresponding residues involved in dimer formation in UGDH-II diverge from those seen in UGDH-I, for CeUGDH on helices α10 (Val229, Phe233, Ser240 and Val247) and α13 (Cys295). Furthermore, in CeUGDH, a network of 34 hydrogen bonds stabilize the interface. Interestingly, amino acids playing a role in forming the hexameric arrangement in CeUGDH are only conserved within UGDH-II (13 out of 36 are strictly conserved), whereas in UGDH-I, corresponding amino acids cannot be found (Supplementary Figure S1).
Biochemical data for different members of the UGDH-I group corroborate the notion from the crystal structure of SpUGDH that these enzymes have a homodimeric quarternary structural organization . Figure 1(A) shows that orientation of the SpUGDH subunits is ‘back-to-back’ (with the position of the active site considered to be the front side) and ‘head-to-tail’. However, liver UGDH and other members of the UGDH-II group are reported to exist predominantly as homohexamers [21,46]. ‘Half-the-sites’ reactivity was proposed for the liver enzyme, arguably indicating that the enzyme could function as a trimer of dimers . An unpublished crystal structure of CeUGDH (PDB code 2O3J), a member of the UGDH-II group, shows a disc-shaped homohexamer, which is assembled from three dimers whose overall structural organization is highly homologous with that of the SpUGDH dimer (Figure 1B). More detailed structural comparison of SpUGDH and CeUGDH (Figure 1C) reveals that the N-terminal domains of the two proteins share a large amount of similarity, whereas structural divergence is observed for the C-terminal domain. The extended C-terminus of CeUGDH results in an increased number of β-strands in the core of the domain, compared with SpUGDH. The C-terminal region of the bacterial enzyme, in contrast, is more flexible than that of CeUGDH and packs on top of the uridine moiety of bound substrate. The extended and structured C-terminus of CeUGDH (Figure 1C) and, by extension (Supplementary Figure S1), that of other UGDH-II enzymes, seem to assist in dimer–dimer interaction within the hexameric arrangement. These higher oligomeric contacts leading to ‘trimerization’ of the dimers are not available to SpUGDH and other UGDH-I enzymes. Despite recent investigations , the role of dimer oligomerization for UGDH enzyme function is currently not clear. Although the assembly of SpUGDH and CeUGDH subunits into dimers is highly similar, the intersubunit contacts utilized in formation of this ‘core’ UGDH oligomer are different in the two enzymes.
The structure of CeUGDH was obtained from protein crystallized in the absence of ligands (PDB code 2O3J). However, structural superimposition of CeUGDH (apo-form) and SpUGDH (ternary complex with UDP-α-D-xylose and NAD+) reveals that residues involved in the binding of substrate and coenzyme are largely conserved in the two enzymes. Residue conservation in functional sites spans both UGDH-I and UGDH-II groups. Amino acid variations in the binding pockets for UDP-Glc and NAD+ are restricted primarily to non-polar residues, which are unlikely to be essential for UGDH enzymatic function. The structure of the enzyme active site appears to be highly conserved among members of UGDH-I and UGDH-II. First- and second-shell residues that are essentially invariant across the entire UGDH family are Cys260, Asp264, Lys204, Asn208, Thr118, Glu141 and Glu145 (Supplementary Figure S1). Of these, Cys260 plays a central role in the enzymatic mechanism, participating as catalytic nucleophile of the reaction. Unless indicated otherwise, amino acid numbering is according to the sequence of SpUGDH.
Mode of action of UGDH
UGDH from bovine liver and SpUGDH were subject to detailed mechanistic examination using elegant biochemical experiments. Mutagenesis studies were used further to interrogate the catalytic mechanism of SpUGDH and human UGDH. It was established early that the stereochemical course of the reaction of UGDH involves transfer of the pro-R-C6 hydride of the substrate alcohol to the si face (B face) of NAD+ [10,48]. Two-step oxidation of UDP-Glc into UDP-GlcA was proposed to proceed in the absence of release into solution of an UDP-α-D-gluco-hexodialdose (UDP-aldehyde) intermediate [13,14,49]. The conclusion was based on the finding that aldehyde-trapping reagents, such as semicarbazide or hydroxylamine, did not affect the distribution of products formed in the reaction catalysed by liver UGDH [4,6,50]. The requirement for a catalytic cysteine residue in the reaction of UGDH was supported by a number of studies, initially using chemical modification, and the formation of a covalent enzyme intermediate was common to the catalytic mechanisms considered [13,24]. The mechanism of Ordman and Kirkwood  (Figure 2A) suggested initial trapping of the UDP-aldehyde by the catalytic side chain of an active-site lysine residue. The Schiff base (aldimine) intermediate is thought to be hydrolysed while being attacked concomitantly from a nearby cysteine side chain. The resulting thiohemiacetal would be oxidized further by a second NAD+ to thioester, yielding UDP-GlcA upon hydrolysis. Evidence supporting the mechanism in Figure 2(A) came from experiments employing tritium reduction of the aldimine formed from 14C-labelled UDP-Glc substrate . Figure 2(A) is consistent with a Bi-Uni-Uni-Bi Ping Pong steady-state kinetic mechanism as proposed for SpUGDH and bovine liver UGDH [15,24]. The absence of a primary kinetic isotope effect accompanying the reaction of SpUGDH with UDP-[6″,6″-di-2H]D-glucose under Vmax (substrate-saturating) conditions implies that hydride transfer to NAD+ is not rate-determining overall. Generally, the final hydrolysis step was considered to be the slowest step of the reaction catalysed by UGDH enzymes.
Catalytic mechanism proposed for bovine liver UGDH (A) and SpUGDH (B and C)
Tanner and co-workers re-examined participation of an aldimine intermediate in the reaction catalysed by UGDH . They performed oxidation of UDP-Glc in H218O solvent and measured 18O incorporation into the UDP-GlcA product using SpUGDH. If, as proposed in Figure 2(A), reaction of UGDH involved an extra hydrolysis step at the level of the aldimine, one would expect that both oxygens of the carboxy group in UDP-GlcA become isotopically labelled from solvent. Observation that only a single 18O atom was incorporated into product is inconsistent with the mechanism in Figure 2(A). A scenario was therefore proposed in which UDP-aldehyde immediately reacts with an active-site cysteine residue to generate the thiohemiacetal (Figures 2B and 2C).
The UDP-aldehyde was chemically synthesized and examined as substrate of wild-type and mutated forms of SpUGDH . In assays carried out with wild-type enzyme, the aldehyde displayed similar reactivity and apparent binding affinity as did the natural substrate UDP-Glc, supporting the idea that UDP-aldehyde is a kinetically competent intermediate of the catalytic reaction of UGDH. Muteins of SpUGDH having Cys260 replaced by alanine or serine were strongly impaired in catalytic function for oxidation of UDP-Glc . Reaction of the C260S mutein with UDP-Glc or UDP-aldehyde led to the slow accumulation of an ester intermediate, in which the side chain of Ser260 had become acylated. Thus, surprisingly, UDP-aldehyde was converted by the C260A mutein with catalytic constants that were within one order of magnitude of the corresponding constants for the wild-type enzyme reacting with UDP-Glc. However, UDP-Glc was not a substrate of this mutein. Tanner and co-workers concluded that the C260A mutein presumably recognizes the hydrated form of the aldehyde which resembles the ‘normal’ thiohemiacetal intermediate of the reaction and converts it into the acid product . Earlier studies from Kirkwood and co-workers agree with these findings, showing that a variant of bovine liver UGDH in which the essential (active-site) cysteine residue had been derivatized with cyanide also retained limited ability to catalyse the second oxidation step .
A very interesting property of the C260A mutein was the ability to reduce UDP-aldehyde back to UDP-Glc when NADH was present. The kinetic constants for this reaction (kcat 1.9 s−1; Km 58 μM) were comparable with the kinetic constants of the wild-type enzyme for NAD+-dependent oxidation of UDP-Glc (kcat 1.2 s−1; Km 14 μM). The observation of a reasonably rapid reductive conversion of UDP-aldehyde was interpreted to imply that Cys260 is not a key catalytic residue in the first step of UDP-Glc oxidation catalysed by the wild-type enzyme. The absence of measurable oxidation of UDP-Glc by the C260A mutein was explained by an unfavourable equilibrium for conversion of alcohol substrate into aldehyde product. Owing to an enzymatic dismutation, in which aldehyde is transformed into both UDP-Glc and UDP-GlcA, conversion of UDP-aldehyde by wild-type SpUGDH could not be measured as consumption of NADH. A deuterium ‘wash-out’ experiment was performed in which the C260A mutein was incubated in the presence of UDP-[6″,6″-di-2H]D-glucose, and 20 mM NADH and 4 mM NAD+ were added as coenzymes. If substrate was oxidized to UDP-aldehyde in this reaction, deuterated NADH would be formed. Now, if the NAD2H was exchanged by unlabelled NADH at the level of the ternary complex, UDP-aldehyde could be reduced back to starting material having the original pro-R deuterium replaced by hydrogen. Therefore, overall, the deuterium content of substrate would decrease. However, no significant loss of deuterium was observed upon prolonged incubation under the conditions just described. This finding might imply the absence of coenzyme exchange in the complex of enzyme, UDP-aldehyde and NADH, as proposed by Ge et al. . However, it could also indicate the absence of reaction catalysed by the C260A mutein to give the aldehyde product. In any case, it seems probable that wild-type enzyme utilizes a mechanism by which escape of UDP-aldehyde is prevented during reaction. One possibility is that the first-formed NADH is not released before the (ionized) side chain of Cys260 has added to the aldehyde, generating the thiohemiacetal intermediate, as shown in Figures 2(B) and 2(C).
Cys276 of human UGDH, which is homologous with Cys260 of SpUGDH, was replaced by serine, and the resulting mutein showed no activity in a conventional steady-state assay . However, it did perform a single step of oxidation on UDP-Glc, producing NADH in a concentration corresponding to the molar equivalent of enzyme present in the reaction. These findings suggest that nucleophilic catalysis by cysteine is utilized by human UGDH and, by extension (Supplementary Figure S1), members of UGDH-II.
Role of active-site residues other than cysteine
The crystal structure of SpUGDH has revealed residues potentially involved in catalysis to 2-fold oxidation of UDP-Glc, as shown in Figure 3. Lys204 and water bonded to Asp264 were previously considered to have a catalytic base function in the first step of oxidation. Thr118 (T118A), Glu141 (E141Q) and Glu145 (E145Q) were replaced in SpUGDH. The T118A mutein showed a 164-fold lowered kcat as compared with the kcat of the wild-type enzyme, suggesting an auxiliary function of Thr118 in the enzymatic reaction, perhaps as stabilizer (via active-site water; Figure 3) of negative charge developing on the reactive oxygen atom of substrate in the course of the reaction. E141Q and E145Q muteins showed only modestly decreased kcat values that were within about one order of magnitude of the wild-type value. Glu141 was proposed as a candidate catalytic base facilitating attack of water during thioester hydrolysis. Muteins of SpUGDH at position Lys204 (K204A) and Asp264 (D264N) were not expressed as soluble proteins in Escherichia coli. However, replacing the corresponding Lys220 (by alanine, histidine or arginine) and Asp280 (by asparagine) in human UGDH very strongly impaired the enzyme activity . Although different functions were considered, the exact role of these two residues in UGDH catalysis remains currently elusive. In addition to the abovementioned catalytic base facilitation of the ‘alcohol dehydrogenase’ step of the reaction, deprotonation of cysteine (by the couple of Asp264 and water) as well as oxyanion stabilization (together with Asn208) during formation and breakdown of thiohemiacetal and thioester might be roles accomplished by the two (highly conserved) active-site residues, as shown for SpUGDH.
Close-up structure of the active site of SpUGDH (PDB code 1DLI) bound with UDP-α-D-xylose and NAD+ (indicated in blue) and water (sphere).
Work by Feingold and co-workers focused on the occurrence of ‘half-of-the-sites’ reactivity in the hexameric UGDH from bovine liver [14,19,20]. An assortment of biochemical methods were employed to measure substrate binding and determine protein conformational changes associated with the binding event. The liver enzyme was proposed to bind only three molecules of UDP-Glc and NAD+ per protein hexamer. The structure of SpUGDH, which is a dimeric UGDH, however, shows only weak ‘communication’ between the active sites of adjacent protein subunits. The side chain of Arg244 contributes to the active site of the neighbouring subunit where it provides hydrogen bonds to the hydroxy groups at C-2 and C-3 of the glucosyl moiety of bound UDP-Glc. Arg244 is conserved in members of both UGDH-I and UGDH-II. SpUGDH exhibits only a weak allosterism with respect to UDP-Glc. It is currently not clear whether the ‘half-of-the-sites’ reactivity of the liver enzyme is a common feature of UGDH enzymes.
Possible role of UGDH in human pathophysiology
Hyaluronan is a linear polysaccharide composed of a [GlcAβ-1-3GlcNAcβ-1-4]n disaccharide unit. Unlike other members of the glycosaminoglycan class, hyaluronan is synthesized in the absence of a protein core at the inner face of the plasma membrane and it is not sulfated. The molecular size of hyaluronan can vary between 5×103 and 107 Da. Both the actual amount and the mass of the hyaluronan polymer determine its physiological function. In mammals, hyaluronan was implicated in numerous roles including development, tissue organization, cell proliferation and signalling processes [55,56]. Interactions of hyaluronan with protein receptors such as CD44 and RHAMM (receptor for hyaluronan-mediated motility) have been linked to cell survival and motility under normal conditions, but also under the pathophysiological conditions of cancer. It was shown that elevated concentrations of hyaluronan promote cancer progression and cell aging, both in animal models as well as in human patients. Work on tumours of breast , ovaries , colon  and prostate  collectively supports a strong correlation between the hyaluronan content in the connective tissue matrix (stroma) of the tissue and (negative) prediction of patient survival. Inhibition of hyaluronan synthesis in general [43,58] and by restriction of the building block UDP-GlcA in particular  has been shown to slow down tumour growth. Recently, human UGDH has been proposed as a novel field-specific candidate biomarker for prostate cancer . Limitation of UDP-GlcA in tumour cells by UGDH antagonism is a novel therapeutic target in combating cancer. Group-selective inhibition of UGDH is of an even more general interest in therapy because, in many pathogenic bacteria (e.g. group A streptococci), UGDH-I provides UDP-GlcA for synthesis of antiphagocytotic capsular polysaccharides, which are known to constitute an important virulence factor. UGDH-I is therefore a target for new antibacterial drugs .
Development of UGDH antagonists is still in its infancy. UDP-α-D-xylose is a strong feedback inhibitor of UGDH enzymes from different sources (UDP-α-D-xylose is produced from UDP-GlcA by oxidoreductive decarboxylation). It exhibits a Ki of ~2.7 μM against SpUGDH . Previously described irreversible inhibitors, such as UDP-chloroacetol, that are directed against the active-site cysteine residue, are probably not useful. However, Campbell and Tanner  synthesized UDP-7-deoxy-α-D-gluco-hept-6-ulopyranose as a UDP-Glc mimic and showed reversible inhibition of SpUGDH by this substrate analogue with a Ki of ~6.7 μM. The polyphenols gallic acid and quercetin were recently shown to inhibit human UGDH . Both compounds were also demonstrated to reduce proliferation of MCF-7 human breast cancer cells in vitro . In addition to binding efficacy and inhibition potency, future design of UGDH inhibitors will need to focus on selectivity (UGDH-I compared with UGDH-II) and cell targeting. Mechanistic details of the human enzyme would be useful to advance in this goal.
Structural Glycobiology and Human Health: A Biochemical Society Focused Meeting held at Royal Holloway, University of London, 30–31 March 2010. Organized and Edited by Tony Corfield (Bristol, U.K.) and Barbara Mulloy (National Institute for Biological Standards and Control, U.K.).
Funding by the Austrian Science Fund (DK Molecular Enzymology) [grant number W901-B05] is gratefully acknowledged.