ANRV345-BI77-25

ARI

29 February 2008

19:15

V I E W

A

Review in Advance first posted online on April 14, 2008. (Minor changes may still occur before final publication online and in print.)

N

I N

C E

S

R

E

D V A

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

Glycosyltransferases: Structures, Functions, and Mechanisms L.L. Lairson,1 B. Henrissat,2 G.J. Davies,3 and S.G. Withers1 1

Department of Chemistry, University of British Columbia, Vancouver V6T 1Z3, Canada; email: [email protected], [email protected]

2

Architecture et Fonction des Macromolecules Biologique, CNRS, Universites Aix-Marseille I and II, Marseille 13288, France; email: [email protected]

3

Structural Biology Laboratory, Department of Chemistry, University of York, Heslington YO10 5YW, United Kingdom; email: [email protected]

Annu. Rev. Biochem. 2008. 77:25.1–25.35

Key Words

The Annual Review of Biochemistry is online at biochem.annualreviews.org

carbohydrate-modifying enzymes, glycobiology, glycosylation, ion pair mechanisms, nucleophilic substitution

This article’s doi: 10.1146/annurev.biochem.76.061005.092322 c 2008 by Annual Reviews. Copyright  All rights reserved 0066-4154/08/0707-0001$20.00

Abstract Glycosyltransferases catalyze glycosidic bond formation using sugar donors containing a nucleoside phosphate or a lipid phosphate leaving group. Only two structural folds, GT-A and GT-B, have been identified for the nucleotide sugar-dependent enzymes, but other folds are now appearing for the soluble domains of lipid phosphosugar-dependent glycosyl transferases. Structural and kinetic studies have provided new insights. Inverting glycosyltransferases utilize a direct displacement SN2-like mechanism involving an enzymatic base catalyst. Leaving group departure in GT-A fold enzymes is typically facilitated via a coordinated divalent cation, whereas GT-B fold enzymes instead use positively charged side chains and/or hydroxyls and helix dipoles. The mechanism of retaining glycosyltransferases is less clear. The expected two-step doubledisplacement mechanism is rendered less likely by the lack of conserved architecture in the region where a catalytic nucleophile would be expected. A mechanism involving a short-lived oxocarbenium ion intermediate now seems the most likely, with the leaving phosphate serving as the base.

25.1

ANRV345-BI77-25

ARI

29 February 2008

19:15

Contents

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

INTRODUCTION . . . . . . . . . . . . . . . . . 25.2 Glycosyltransferase Activities . . . . . 25.2 Glycosyltransferase Folds . . . . . . . . . 25.2 Glycosyltransferase Classification . . . . . . . . . . . . . . . . . . 25.5 Genomic Distributions of Glycosyltransferases . . . . . . . . . . . 25.6 Mechanism of Inverting Glycosyltransferases . . . . . . . . . . . 25.7 Inverting GT-A Glycosyltransferases . . . . . . . . . . . 25.7 Inverting GT-B Glycosyltransferases . . . . . . . . . . .25.10 Mechanism of Retaining Glycosidases . . . . . . . . . . . . . . . . . .25.12 Mechanism of Retaining Glycosyltransferases . . . . . . . . . . .25.13 Challenges in Studying the Mechanisms of Retaining Glycosyltransferases . . . . . . . . . . .25.13 Retaining GT-A Glycosyltransferases . . . . . . . . . . .25.15 Retaining GT-B Glycosyltransferases . . . . . . . . . . .25.20 An Alternative SN i-Like Mechanism . . . . . . . . . . . . . . . . . . .25.23 Evolutionary Constraints upon Retaining Glycosyltransferase Mechanisms . . . . . . . . . . . . . . . . . . .25.24 Concluding Remarks on the Mechanism of Retaining Glycosyltransferases . . . . . . . . . . .25.27

INTRODUCTION The enormous complexity of the various oligosaccharide structures found in nature (1) is derived from a rational orchestration of the enzymatic formation and the breakdown of glycosidic linkages achieved by glycosyltransferases, glycosidases, glycan phosphorylases, and polysaccharide lyases. In contrast to the well-characterized mechanistic strategies used by glycosidases to catalyze glyco25.2

Lairson et al.

sidic bond hydrolysis (2, 3), the mechanisms of the glycosyltransferases responsible for glycoside bond formation remain less clear. Despite a lack of evolutionary relatedness, glycosyltransferases had been thought to use mechanistic strategies that directly parallel those used by glycosidases (2). However, some distinct differences are becoming apparent, as discussed below.

Glycosyltransferase Activities Glycosyltransferases are most accurately defined as those enzymes that utilize an activated donor sugar substrate that contains a (substituted) phosphate leaving group. Donor sugar substrates are most commonly activated in the form of nucleoside diphosphate sugars (e.g., UDP Gal, GDP Man); however, nucleoside monophosphate sugars (e.g., CMP NeuAc), lipid phosphates (e.g., dolichol phosphate oligosaccharides), and unsubstituted phosphate are also used. Nucleotide sugar-dependent glycosyltransferases are often referred to as Leloir enzymes, in honor of Luis F. Leloir who discovered the first sugar nucleotide and was awarded the Nobel Prize in chemistry in 1970 for his enormous contributions to our understanding of glycoside biosynthesis and sugar metabolism. The acceptor substrates utilized by glycosyltransferases are most commonly other sugars but can also be a lipid, protein, nucleic acid, antibiotic, or other small molecules. In addition, although glycosyl transfer most frequently occurs to the nucleophilic oxygen of a hydroxyl substituent of the acceptor, it can also occur to nitrogen (e.g., the formation of N-linked glycoproteins), sulfur (e.g., the formation of thioglycosides in plants), and carbon (e.g., Cglycoside antibiotics) nucleophiles.

Glycosyltransferase Folds As has been done for other classes of carbohydrate-active enzymes, glycosyltransferases are classified into families based on amino acid sequence similarities (4, 5).

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

Indicative of the pace and progress of genome sequencing endeavors, in the 10 years since Campbell and colleagues first started such a classification of glycosyltransferases, the number of distinct families has grown from 27 to 90. There are marked contrasts between the diversity of three-dimensional (3-D) folds observed for hydrolytic glycoside hydrolases and the very limited folds used by the synthetic glycosyltransferases. Structural characterization of representatives from a large number of the 110 families of glycosidases (http://www.cazy.org) has revealed an extraordinary degree of diversity in overall fold, despite considerable commonality in activesite features. This would indicate a convergence during the evolution of catalytic mechanism. In contrast, the recent burst of reported glycosyltransferase structures has revealed a quite different situation as only two general folds, called GT-A and GT-B (nomenclature proposed in Reference 6), have been observed for all structures of nucleotide-sugardependent glycosyltransferases solved to date (5, 7, 8). Furthermore, threading analysis has revealed that many of the uncharacterized families are also predicted to adopt one of these two folds. This finding may indicate that the majority of glycosyltransferases have

a GT-A

evolved from a small number of progenitor sequences. However, it also reflects the requirement for at least one nucleotide-binding domain of the Rossmann fold type. Tantalizing support for the former notion is derived from the fact that only two glycosyltransferase families (GT2 having the GT-A fold and GT4 having the GT-B fold) are possessed by primitive Archae, which may reflect the origins from which the vast majority of glycosyltransferases have evolved (5). In addition, within the GT4 family, there exist members that utilize donor substrates activated with a nucleoside diphosphate and others that utilize donor substrates activated with a phosphate group, suggesting an evolutionary link between enzymes that utilize these two substrate forms. The GT-A fold was first described for the inverting enzyme SpsA from Bacillus subtilis, for which both apo- and UDP-bound 3-D Xray crystal structures were obtained (9). Consisting of an open twisted β-sheet surrounded by α-helices on both sides, the overall architecture of the GT-A fold is reminiscent of two abutting Rossmann-like folds, typical of nucleotide-binding proteins. Two tightly associated β/α/β domains, the sizes of which vary, abut closely, leading to the formation of a continuous central β-sheet (Figure 1a). For this reason, some describe the GT-A

GT-A fold: glycosyltransferase protein topology consisting of two closely abutting β/α/β Rossmann domains GT-B fold: glycosyltransferase protein topology consisting of two β/α/β Rossmann domains that face each other and are linked flexibly

b GT-B

Figure 1 Overall folds observed for glycosyltransferase enzymes. (a) The GT-A fold is represented by the inverting enzyme SpsA from Bacillus subtilus, Protein Data Bank (pdb) 1qgq, and (b) the GT-B fold, by bacteriophage T4 β-glucosyltransferase, pdb 1jg7. www.annualreviews.org • Glycosyltransferases

25.3

ANRV345-BI77-25

ARI

29 February 2008

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

GT-C fold: a predicted protein topology for transmembrane glycosyltransferases that is not experimentally verified

25.4

19:15

fold as a single domain fold. However, distinct nucleotide- and acceptor-binding domains are present (7). Eukaryotic members possessing the GT-A fold typically have a short N-terminal cytoplasmic domain that is followed by transmembrane and stem regions leading to the globular catalytic region (10). It should be noted that not all enzymes that possess a GT-A fold are glycosyltransferases. For example, and possibly indicating a divergence in mechanism during the course of evolution, the sugar-1-phosphate pyrophosphorylase/nucleotidyl transferase superfamily of enzymes, responsible for the synthesis of nucleoside diphosphate sugars, has been shown to possess the GT-A architecture (11). Potential confusion arises with enzymes that almost display the canonical GT-A topology, but with a different order of β-strands. This is exemplified by sialyltransferases from family GT-42 (12, 13) whose structure could either be considered a new fold type (on the basis of differing connectivity) or as a modified GT-A fold (a description the original authors favor) (12). Most GT-A enzymes possess an AspX-Asp (referred to as DXD) signature in which the carboxylates coordinate a divalent cation and/or a ribose (14, 15). These are by no means conserved motifs (none of the residues is invariant), and although frequently described as a determining characteristic of GT-A glycosyltransferases, examples do now exist of enzymes from this fold family that do not possess the DXD “signature” (16). Furthermore, many sequences possess a DXD signature, but are not glycosyltransferases. The first determined 3-D structure for a nucleoside diphosphate-utilizing glycosyltransferase was in fact reported in 1994 and was that of a DNA-modifying βglucosyltransferase from bacteriophage T4 (17). The overall fold of this protein was found to be homologous to that of glycogen phosphorylase, and now that it has been observed for other glycosyltransferases, it is termed the GT-B fold. Like the GT-A fold, the architecture of GT-B enzymes consists of two β/α/β Rossmann-like domains; however, in this case, Lairson et al.

the two domains are less tightly associated and face each other with the active-site lying within the resulting cleft (Figure 1b). As with the GT-A fold, these domains are associated with the donor and acceptor substratebinding sites. Non-glycosyltransferase enzymes are also known to adopt the GT-B fold, with UDP GlcNAc 2-epimerase serving as one such example (18). A third glycosyltransferase fold termed GT-C was recently predicted on the basis of iterative sequence searches, using programs such as BLAST, (19). This report predicted that eight families (GT22, GT39, GT48, GT50, GT53, GT57, GT58, GT59) would possess the GT-C fold. Following the publication of this report, 4 (GT66, GT83, GT85, GT86) of the 15 glycosyltransferase families subsequently created were also predicted to adopt this fold by the CAZY database (http://www.cazy.org). The predicted architecture of the GT-C fold is that of a large hydrophobic integral membrane protein located in the endoplasmic reticulum or on the plasma membrane having between 8 and 13 transmembrane helices and an active site located on a long-loop region (20–22). Consistent with this predicted alternative architecture is the fact that 10 of the 12 families (all but GT48 and GT53) predicted to adopt the GT-C fold utilize lipid phosphate-activated donor sugar substrates. A subsequent comparison of glycosyltransferase families using a “profile hidden Markov method” of sequence analysis resulted in classification of family GT48, which uses UDP Glc as donor substrate, within the GT-A superfamily (23). Members of family GT53, predicted to adopt the GT-C fold, utilize UDP L-arabinose as a donor substrate. Very recently the first 3D structure of an enzyme assigned to GT-C was determined: the soluble C-terminal domain of the Pyrococcus furiosius oligosaccharyltransferase STT3 (23a). This structure weakens the value of the GT-C classification because it reveals that the portion of the protein on which the alignment is based is the trans-membrane region, which does not include the predicted loop bearing

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

the active site. Rather, the active site seems to be located within this soluble C-terminal domain, though a note of caution is necessary as the authors were not able to demonstrate catalytic activity for this truncated protein. However, this inactivity could well be a consequence of the need for the transmembrane region to bind the lipid portion of the substrates. Consequently, even though the members of GT-C are likely related, it is probably through their trans-membrane region rather than through their catalytic domain. It is therefore quite possible that other folds could be found for these catalytic domains whose structures are not constrained by the need to bind a nucleotide; thus, it is unlikely they will bear Rossmann folds. The soluble Cterminal domain of the Pyrococcus OST- STT3 in fact adopts a novel architecture with a central, mainly α-helical domain surrounded by three β-sheet-rich domains. The structure of a peptidoglycan synthesizing glycosyl transferase (GT51), another enzyme that uses a lipid phosphosugar donor, was also recently determined, as discussed in greater detail below (25, 26). Fascinatingly, this enzyme was found to have a bacteriophage-lysozyme-like fold, and not a Rossmann fold, consistent with it not being a nucleotide phosphosugar-dependent enzyme. It was not a predicted GT-C member. Other sequence families, such as GT76, are currently orphan families that are not predicted to adopt the GT-A, GT-B, or proposed GT-C fold. Glycoside hydrolases, phosphorylases, and glycosyltransferases all catalyze glycosyl group transfer. This does lead to confusion and inconsistency in the nomenclature for these enzymes and also obscures the placement of some of these enzymes into CAZy families. For example, the structural and mechanistic study of chitobiose phosphorylase from Vibrio proteolyticus revealed that enzymes from family GT36 share more of a structural and evolutionary relationship with glycosidases of clan GH-L, which have an (α/α)6 fold (24). As a result, this family was

subsequently reclassified as family GH-94. Very recently, however, the much-anticipated 3-D X-ray crystal structure of a peptidoglycan glycosyltransferase from family GT51 was solved with a bound substrate analogue/ inhibitor (25, 26). As mentioned above, members of this inverting family utilize lipid II phosphate-activated donor substrates, and the structure of the glycosyltransferase domain was not found to show similarity to the GT-A, GT-B, or proposed GT-C folds. Instead, the α-helical fold and several activesite features are more akin to the lysozyme family of enzymes. Given that peptidoglycan glycosyltransferases are solely synthetic, it would be questionable to reclassify them as glycoside hydrolases, but given their strong similarities to some lysozymes, they certainly highlight the muddy waters of classification. Future results with the other orphan families currently classified as glycosyltransferases, but not predicted by sequence analysis to adopt either of the GT-A, GT-B, or proposed GT-C folds, will undoubtedly shed more light on the evolutionary origin of glycosyltransferase activities and their relationships to the other major classes of carbohydrate-active enzymes. Whether these families will be found experimentally to actually adopt one of the GT-A, GT-B, or lipid phosphosugar glycosyltransferase folds or possibly to form new glycosyltransferase superfamilies or simply succumb to reclassification, as described above, remains to be determined.

Glycosyltransferase Classification Two stereochemical outcomes are possible for reactions that result in the formation of a new glycosidic bond: the anomeric configuration of the product can either be retained or inverted with respect to the donor substrate (Figure 2). As such, enzymes catalyzing glycosyl group transfer are classified as either inverting or retaining, depending on the outcome of the reaction. Logically, it follows that www.annualreviews.org • Glycosyltransferases

25.5

ANRV345-BI77-25

ARI

29 February 2008

19:15

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

Figure 2 Like glycosidases, glycosyltransferases catalyze glycosyl group transfer with either inversion or retention of the anomeric stereochemistry with respect to the donor sugar (5).

Inverting glycosyltransferase: a glycosyltransferase that catalyzes group transfer reactions with net inversion of stereochemistry at the anomeric reaction center of the donor substrate Retaining glycosyltransferase: a glycosyltransferase that catalyzes group transfer reactions with net retention of stereochemistry at the anomeric reaction center of the donor substrate

25.6

this stereochemical outcome must result from the utilization of different mechanisms by the two classes of enzymes. In the case of glycoside hydrolases, with only very few exceptions (family GH4 enzymes, goose lysozymes, and soluble lytic transglycosylases from family GH23), the enzymes within a given GH family catalyze hydrolysis with the same stereoselectivity. This should also be the case among glycosyltransferase families. However, given the higher sequence similarity within the nucleotidebinding domains, which risks dominating the sequence searches and is common to enzymes catalyzing both mechanisms, it is possible that some families may contain enzymes that carry out reactions with different stereochemical outcomes. Nevertheless, glycosyltransferase families can be classified into clans depending on their fold and the stereochemical outcome of the reactions that they catalyze (Figure 3). Among GT-A and GT-B superfamilies, the overall fold of the enzyme does not dictate the stereochemical outcome of the reaction that it catalyzes, as examples of both inverting and retaining glycosyltransferases have been identified within both the GT-A and GT-B fold classes (5). Indicative of this phenomenon are recent findings from studies of a mannosylglycerate synthetase, which transfers mannose to the 2-OH of D-lactate, D-glycerate or glycollate with net retention of configuLairson et al.

ration (27). Based on amino acid sequence, this enzyme was initially classified among the GT-A inverting family GT2, however, structural and mechanistic studies led to its reclassification among the GT-A retaining family GT78. To date, all enzymes predicted to adopt the GT-C fold belong to inverting glycosyltransferase families.

Genomic Distributions of Glycosyltransferases Examination of more than 500 completely sequenced organisms, listed in the CAZy database (see related resource) as of September 2007, reveals that by and large the number of GTs encoded by the genome of an organism correlates reasonably well with the total number of genes of the genome. GTs account for about 1% to 2% of the gene products of an organism, whether archaeal, bacterial, or eukaryotic. This proportion seems to hold true even for double-stranded DNA viruses. The populations of the various GT families also show wide variations in the CAZy database, with two large families, GT2 and GT4, accounting for about half of the total number of GTs. Thus organisms, such as plants, with very large genomes that synthesize a complex cell wall or use the glycosylation of small molecules to tune bioactivity have many GTs (Arabidopsis encodes approximately 450

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

glycosyltransferases and Populus more than 800). In contrast, organisms that have undergone massive gene loss during evolution to become obligate symbionts or obligate parasites appear to have very few or sometimes no detectable glycosyltransferase genes at all (e.g., several Mycoplasma species). The massive number of plant glycosyltransferases is due to the presence of several extremely populated glycosyltransferase families. For example, Arabidopsis, Oryza, and Populus have respectively ∼120, ∼200, and ∼300 family GT2 genes. One of the surprises that came with the completion of the first animal genomes was that humans have slightly less glycosyltransferase genes (∼230) than the nematode Caenorhabditis elegans (∼240).

Mechanism of Inverting Glycosyltransferases Like inverting glycoside hydrolases, the mechanistic strategy employed by inverting glycosyltransferases is that of a direct displacement SN 2-like reaction. An active-site side chain serves as a base catalyst that deprotonates the incoming nucleophile of the acceptor, facilitating direct SN 2-like displacement of the activated (substituted) phosphate leaving group (Figure 4a). The key questions in examining the catalytic mechanism of inverting glycosyltransferases are, therefore, the identity of the base catalyst and the method used to facilitate departure of the (substituted) phosphate leaving group.

Inverting GT-A Glycosyltransferases As mentioned above, the family GT2 enzyme SpsA from B. subtilis was the first glycosyltransferase experimentally shown to adopt the GT-A fold (9). This family represents the largest and evolutionarily most ancient of inverting glycosyltransferases. Unfortunately, the natural donor and acceptor sugars of SpsA are not known. However, fortuitously, a molecule of the cryoprotectant glycerol was found bound within the SpsA active site in

GT-A

Inverting - clan I

2, 7, 12, 13, 14, 16*, 21, 25*, 31, 40*, 42, 43, 49*, 82, 84

Retaining - clan III

6, 8, 15, 24, 27, 34*, 44, 45*, 55, 60*, 62*, 64, 78, 81

Inverting - clan II

1, 9, 10, 17*, 19, 23, 26*, 28, 30, 33, 41, 47*, 56*, 63, 80

Retaining - clan IV

3, 4, 5, 20, 32*, 35, 72

Non-GT

GT-B Non-GT

Figure 3 Glycosyltransferase (GT) classification system proposed by Coutinho et al. (5). Families are classified into clans on the basis of their fold and activity. GT family numbers belonging to each clan are indicated on the far right. Bona fide families having members with solved 3-D structures are indicated in red. The remaining families are those predicted to adopt either the GT-A or GT-B fold. Families identified in black with an asterisk are those with structures predicted to adopt either the GT-A or GT-B fold solely by Liu & Mushegian (19), and those in black without an asterisk have GT-A or GT-B structures as predicted by both Liu & Mushegian and the CAZY Web site. This classification system does not include 39 of the 90 glycosyltransferases. Members from 12 (GT22, GT39, GT48, GT50, GT53, GT57, GT58, GT59, GT66, GT83, GT85, GT86) of those families not included were predicted to adopt a proposed GT-C fold. On the basis of a determined 3-D structure, family GT36 was reclassified among the glycosidases as family GH94. Structural characterization of the remaining 26 orphan families (GT11, GT18, GT29, GT37, GT38, GT46, GT51, GT52, GT54, GT61, GT65, GT67, GT68, GT69, GT70, GT71, GT73, GT74, GT75, GT76, GT77, GT79, GT87, GT88, GT89, GT90) will provide insights into the strengths and limitations of predictive bioinformatic tools.

a suitable position to mimic the natural acceptor sugar and hydrogen bonded to the side chain carboxylate of Asp191, leading to the proposition that this residue played the role of base catalyst in the proposed direct displacement mechanism (Figure 5a) (9). This notion was later supported by superpositioning (28) the SpsA structure on the solved GT-A structures of the enzymes lactose synthase (Gal-T1) from family GT7 (29– 35), GnT-I from family GT13 (36, 37), and GlcAT-I from family GT43 (38, 39). These enzymes all have a conserved Asp or Glu residue within their active sites, occupying a position equivalent to that of Asp191 in SpsA (Figure 5). Similarly, an analogously positioned conserved Asp or Glu residue has www.annualreviews.org • Glycosyltransferases

25.7

ARI

29 February 2008

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

19:15

Figure 4 (a) Inverting glycosyltransferases utilize a direct displacement SN 2-like reaction that results in an inverted anomeric configuration via a single oxocarbenium ion-like transition state. (b) The proposed double-displacement mechanism for retaining glycosyltransferases involves the formation of a covalently bound glycosyl-enzyme intermediate. Abbreviations: R, a nucleoside, a nucleoside monophosphate, a lipid phosphate, or phosphate (phosphorylases classified as glycosyltransferases); and R’OH, an acceptor group (e.g., another sugar, a protein, or an antibiotic).

been observed in the subsequently obtained structures of the enzymes C2GnT-L from family GT14 (16), Mfng from family GT31 (40), and GlcAT-P from family GT43 (41, 42) (Figure 5). Structures with bound acceptor substrates indicate that these conserved carboxylates are within hydrogen-bonding distance of the nucleophilic hydroxyl under25.8

Lairson et al.

going reaction, consistent with their roles as base catalysts (Figure 5). Further support comes from site-directed mutagenesis studies of the family GT2 enzyme ExoM in which it was shown that mutation of Asp187 (structurally homologous to Asp191 in SpsA) abolished all in vitro glycosyltransferase activity (43).

ANRV345-BI77-25

ARI

29 February 2008

19:15

a SpsA

b Gal-Tl

(pdb 1qgq) (GT2)

(pdb 2fyd) (GT7)

D191

UDP

D318

D255

D98

Glycerol UDP Glucose

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

D99 D254

c GnTl

d Mfng

(pdb 1foa) (GT13)

(pdb 2j0b) (GT31)

D232

D291 E211

D142

UDP

UDP GlcNAc D213

D144

f GlcAT-P

e GlcAT-l D281 (pdb 1fgg) (GT43)

D284

(pdb 1v84) (GT43)

D194

Gal

D195

Gal UDP

UDP

Gal

GlcNAc D196

D197

Figure 5 Comparison of the active sites of several metal-dependent inverting GT-A fold glycosyltransferases. A conserved side chain carboxylate is located in a near identical relative position within all active sites on the β-face of the donor substrate and (when present) within hydrogen-bonding distance of the nucleophilic hydroxyl of the acceptor sugar and plays the role of the base catalyst in a direct displacement mechanism. Mn2+ cations are shown as purple spheres.

www.annualreviews.org • Glycosyltransferases

25.9

ARI

29 February 2008

19:15

The vast majority of enzymes from glycosyltransferase Clan I that have been subjected to biochemical analysis use an essential divalent cation (usually Mn2+ or Mg2+ ), coordinated by the so-called DXD motif, to facilitate departure of the nucleoside diphosphate leaving group by electrostatically stabilizing the developing negative charge. Exceptions to this strategy have been reported for the metal ion-independent sialyltransferases from family GT42 (12) and β-1,6-GlcNAc transferase C2GnT-L from family GT14 (16), which use tyrosyl hydroxyls or basic amino acids, respectively, to electrostatically stabilize substituted phosphate leaving groups. This strategy is reminiscent of that used by metal ionindependent glycosyltransferases possessing a GT-B fold (as discussed below), indicative of a convergence of mechanisms among these two superfamilies. A convergence in mechanisms between these two superfamilies is also illustrated by the finding that the most likely candidate for the base catalyst in family 42 sialyltransferases, as revealed by crystal structures with a bound CMP 3F NeuAc donor substrate analogue (12, 13), is a His residue, as is also the case for many GT-B fold-inverting enzymes (as discussed below). Theoretical support for a concerted SN 2like displacement mechanism for inverting glycosyltransferases from Clan I has been provided by a hybrid quantum mechanical/ molecular mechanical study of the β-1,2GlcNAc transferase GnT-I (44). The results supported a catalytic mechanism involving a concerted SN 2-type transition state involving a near simultaneous nucleophilic attack, facilitated by proton transfer to the catalytic base (Asp291), and leaving group dissociation steps. An activation energy of ∼19 kcal/mol was estimated for the proposed transition state. Inverting GT-A glycosyltransferases, notably those in family GT2, are involved in the formation of many β-linked polysaccharides, such as cellulose, chitin, and hyaluronan. These enzymes exemplify a continuing controversy in the polysaccharide field: Is the

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.10

Lairson et al.

UDP monosaccharide the donor (in which case the growing chain extends by addition to its nonreducing end) or is a UDP-growing chain the donor with a UDP monosaccharide acting as the acceptor (which would result in a growing chain extending by addition at its reducing end)? Labeling experiments on cellulose synthase (45) support nonreducingend elongation, as does direct analysis of both the chitooligosaccharide synthase Nod factor C and zebra fish chitin synthase DG42 (46). Furthermore, yeast chitin synthases have been shown to not use UDP chitobiose as a donor (47). Family GT2 hyaluronan synthases of class II also clearly use UDP monosaccharides (48) as the donor, but class I hyaluronan synthases, closely related to the chitin synthases mentioned above, have conversely been proposed to use reducing-end addition (49). The area is clearly confusing as may be the case with any polysaccharide synthases [indeed such controversy also stalked the peptidoglycan synthase field until the recently reported GT51 structure determination (25, 26)].

Inverting GT-B Glycosyltransferases Despite sharing homologous 3-D architectures, members of this superfamily seem to display a greater degree of diversity in the selected modes of catalyzing glycosyl group transfer between, and in some cases even among families, compared to what has been observed for inverting GT-A enzymes. This greater diversity in mechanism is perhaps a reflection of the greater diversity of chemistries catalyzed by this superfamily of enzymes and perhaps also by the physical separation of the two domains. The prototype of this superfamily is the β-glucosyltransferase (BGT) from T4 bacteriophage, the first nucleoside diphosphateutilizing glycosyltransferase for which a 3-D X-ray crystal structure was obtained (17). This unique enzyme, the sole member of family GT63, transfers glucose from UDP Glc to the hydroxymethyl substituents of modified

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

cytosine bases. This process has evolved as a defense mechanism to prevent genome degradation by phage and host nucleases (50). In addition to the initially reported structures with bound metal ions and UDP present (17, 51), a subsequent ternary complex Xray crystal structure with bound UDP and a DNA acceptor was obtained (52). The ternary complex structure revealed the ability of this glycosyltransferase to facilitate selective glycosylation by inducing the DNA to “flip out” in a fashion reminiscent of DNA glycosylases, methyltransferases, and endonucleases. Analogous to what has been found with the GT-A superfamily, mutagenesis studies revealed Asp100 to be a likely candidate for a base catalyst. This is supported by the observation that cocrystallization of the wildtype enzyme with UDP Glc results in complete hydrolysis of the donor sugar, yielding UDP product complexes, whereas analogous cocrystallization procedures with the D100A mutant resulted in the observation of intact donor sugar bound within the active site (53). A crystal structure with bound intact donor substrate was also obtained by a brief soaking of wild-type BGT crystals with UDP Glc. Interestingly, this donor substrate complex had no metal cation bound within the active site despite being soaked with high concentrations of Mg2+ (53). Instead, positively charged side chains were found to neutralize the negative charges of the pyrophosphate group of the bound donor substrate. In contrast, in the UDP product complex, obtained by cocrystallization with Mn2+ and UDP, an Mn2+ cation was found coordinating the pyrophosphate group and was located in the region occupied by the glucose moiety in the UDP Glc-complexed structure. This led the authors to propose that the divalent cation plays a role in facilitating product release and not necessarily in the cleavage of the glycosidic linkage of the donor substrate. This alternative mode of leaving group activation, compared to the GT-A superfamily, has also been proposed for several other GT-B enzymes (described below).

Glycosyltransferases from family GT1 adopt the GT-B fold and are responsible for the glycosylation of various important organic structures, such as terpenes, anthocyanins, cofactors, steroids, peptide antibiotics, and macrolides, making this one of the most intensely studied families of glycosyltransferases. Three family GT1 enzymes [GtfA (54), GtfB (55), and GtfD (56)] involved in the biosynthesis of the peptide antibiotic vancomycin have been subjected to structural and biochemical analysis. The results of these studies suggest that these enzymes use an aspartate as the catalytic base and that leaving group departure is facilitated in a metal ionindependent fashion using a helix dipole and interactions with side chain hydroxyl and imidazole groups to stabilize the developing negative charge. A second group of related enzymes from family GT1 have also been characterized, revealing interesting differences from the Gtf enzymes. These include the multifunctional terpene/flavonoid glycosyltransferase UGT71G1 (57), the flavonoid glucosyltransferase VvGTI (58), the macrolide glycosyltransferases OleD and OleI (59), the human drug-metabolizing glucuronyltransferase UGT2B7 (60) and the bifunctional Nand O-glucosyltransferase from Arabidopsis thaliana (60a). Althoughthese enzymes were found to use a metal ion-independent mechanism analogous to that described above to facilitate leaving group departure, the catalytic base was determined to be a side chain imidazole that interacts with an adjacent conserved side chain carboxylate group. These conclusions were based on comparisons to a 3-D X-ray crystal structure with a bound acceptor substrate (58), whereby the putative side chain imidazole was found within hydrogenbonding distance of the reactive acceptor substrate hydroxyl and confirmed by kinetic analysis of site-directed mutants. Other inverting GT-B enzymes that have been subjected to structural and mechanistic analysis include the heptosyltransferase WaaC from family GT9 (61), the fucosyltransferases FucT from family GT10 (62) www.annualreviews.org • Glycosyltransferases

25.11

ARI

29 February 2008

19:15

and FUT8 from family GT23 (63), and the GlcNAc transferase essential for bacterial cell wall synthesis known as MurG of family GT28 (64, 65). These enzymes all appear to use metal ion-independent methods for stabilizing the departure of nucleoside diphosphate leaving groups analogous to that described above. In the cases of WaaC and FucT, structural and kinetic analyses of site-directed mutants support the roles of the side chain carboxylates of Asp13 and Glu95, respectively, as base catalysts. In the cases of FUT8 and MurG, the identity of the base remains less clear. A metal ion-independent sialyltransferase, PmST1 from family GT80, has been reported to possess four distinct enzymatic activities with differing pH optima consisting of an α-2,3-sialyltransferase activity with an optimal pH of 7.5–9.0, an α-2,6-sialyltransferase activity with an optimal pH of 4.5–6.0, an α-2,3-sialidase activity with an optimal pH of 5.0–5.5, and an α-2,3-trans-sialidase activity with an optimal pH of 5.5–6.5 (66). It should be noted that the adventitious presence of CMP could be the cause of the observed sialidase and trans-sialidase activities, as this enzyme, like most other glycosyltransferases, presumably catalyzes the hydrolysis of its donor sugar substrate. At the high concentrations of enzyme used in determining these activities, back reaction in the presence of CMP would lead to the formation of CMP NeuAc in situ, which could then be hydrolyzed (leading to the observed sialidase activity) or used as a substrate in the catalyzed transfer of sialic acid to an alternative acceptor (leading to the observed trans-sialidase activity). In fact, it has recently been elegantly shown that glycosyltransferase reactions can be run in reverse to generate desired rare nucleoside diphosphosugars in situ that can then be used to generate a range of glycosylated antibiotic products (67). The 3-D X-ray crystal structure of PmST1 has been solved in apo and product complex form with bound CMP under conditions (pH 7.5) that favor the α2,3-sialyltransferase activity (68). Subsequent

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.12

Lairson et al.

cocrystallization with CMP 3F NeuAc, followed by soaking with a lactose acceptor, led to the production of a ternary complex structure (69).

Mechanism of Retaining Glycosidases It is important to discuss the mechanism of retaining glycosyltransferases in the context of the vast body of work on retaining glycoside hydrolases. With the notable exception of members from families GH4 and GH109, which use a recently described elimination mechanism involving transient remote oxidation (70, 71) and retaining β-hexosaminidases that use an intramolecular nucleophile (reviewed in Reference 72), the typical mechanism of retaining glycosidases is that of a double-displacement reaction involving a covalently bound glycosyl-enzyme intermediate species (3). This mechanism was first put forward by Koshland (73), who realized that, because inversion of configuration is a fundamental and universal property of bimolecular displacement at saturated carbon centers, glycosidases that gave sugar products with the same anomeric configuration as the substrate must catalyze the reaction using two distinct nucleophilic displacement steps involving an enzymatic catalytic nucleophile. An alternative SN 1-like mechanism, involving the formation of a discrete enzyme-stabilized oxocarbenium ion intermediate species that is shielded on one face by the enzyme, thereby preventing nucleophilic attack from the opposite face of the reaction center and leading to complete retention of anomeric configuration in the product, was subsequently proposed for the retaining glycosidase hen egg white lysozyme by Phillips (74, 75). The free energy of the intermediate, and therefore the associated transition states, for the Koshland mechanism would be lower than for those associated with the Phillips mechanism. The glycosidic linkage between two sugars is extremely stable. For example, the half-lives for the spontaneous hydrolysis of starch and

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

cellulose at room temperature and neutral pH are in the range of five million years (76). As such, the fact that glycosidases are observed to catalyze the hydrolysis of these materials with rate constants of up to 1000 s−1 would lead one to believe that this class of enzyme has evolved a mechanism that would involve an intermediate species with the lowest free energy possible without making turnover of this species too slow. The attendant decreases in the activation barrier allow these enzymes to achieve the formidable task of glycoside cleavage with such proficiency. With some notable exceptions (77), this logic and a large array of mechanistic data (for example, References 78–83) has led most in the field to accept the Koshland mechanism as the mechanism utilized by virtually all retaining glycosidases.

Mechanism of Retaining Glycosyltransferases Again by direct comparison to retaining glycosidases, the mechanism of retaining glycosyltransferases has been proposed to be that of a double-displacement mechanism involving a covalently bound glycosyl-enzyme intermediate (Figure 4b), demanding the existence of an appropriately positioned nucleophile within the active site (2). A divalent cation or suitably positioned positively charged side chains or helix dipoles would presumably play the role of the Lewis acid as was described above for the inverting glycosyltransferases. The leaving diphosphate group itself probably plays the role of a base catalyst activating the incoming acceptor hydroxyl group for nucleophilic attack. The role of substrate/product phosphates acting as base catalysts has a rich history and, as an example, has recently been postulated in the mechanisms of the farnesyl diphosphate (FPP) synthases from Escherichia coli (84) and Trypanosoma cruzi (85). In contrast to glycoside hydrolases, however, there is little work that supports a double-displacement mechanism as a canonical glycosyltransferase mechanism.

Challenges in Studying the Mechanisms of Retaining Glycosyltransferases Mechanistic characterization of this class of enzymes has proven to be a challenging task. The conclusive identification of a catalytic nucleophile and observation of a true kinetically and catalytically competent covalent intermediate has yet to be reported for any retaining transferase despite years of exhaustive studies using techniques that have been successfully applied to the characterization of retaining glycosidases. This may well be interpreted as evidence against the double-displacement mechanism, but it could also be the result of the inapplicability of these techniques for the study of transferases owing to inherent differences in the nature of the substrates being studied. For example, the most successful approach used for the characterization of retaining glycosidases has involved the use of fluorinated substrate analogues (86). The introduction of an electronegative fluorine at either the C2 or C5 position of a pyranose ring inductively destabilizes the oxocarbenium ion-like transition states through which both steps of the double-displacement reaction proceed and in some cases also removes key hydrogen-bonding interactions, resulting in a significant decrease in the rate of the overall reaction. By introducing a good leaving group (e.g., dinitrophenol or fluoride), the first step is “rescued,” resulting in the accumulation of the intermediate species with a significant lifetime that allows mass spectrometric and X-ray crystallographic characterization. Alternatively, or used in combination with the fluoro-sugar approach, removal of the acid/base catalyst of a retaining glycosidase by mutagenesis also leads to a decrease in the rates of both steps of the reaction, and again, by using a substrate with a highly activated leaving group, the glycosylation step can be rescued, leading to the accumulation of the glycosyl-enzyme intermediate. However, because of the strict glycosyltransferase requirement for their nucleoside

www.annualreviews.org • Glycosyltransferases

25.13

ANRV345-BI77-25

ARI

29 February 2008

19:15

a LgtC

b Glycogenin

(pdb 1ll2) (GT8)

(pdb 1ga8) (GT8) D190

G165

R86

K86

D188 Q164

8.9

Y151

D163 S134

UDP Glc

6.1

UDP 2F Gal

Q189 3.5

D130

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

D125

D104

N133

K218 H212

4'-Deoxy lactose

c 3GalT

d ppGalNAcT10

(pdb 2vfz) (GT6)

Y278

S318

(pdb 2d7r) (GT27)

D316

E317

Y347

E345

R221

6.1

Q346

GalNAc

4.5

UDP

3.4

R202

D237 D225

UDP 2F Gal

H280

H239

H370

R373

Y372 D227

e ToxB

f MGS (pdb 2bvl) (GT44)

L163

R273

(side chain omitted)

(pdb 2bo8) (GT78)

T193

N384 D192 D286

Q385

D270

S469

E166

D100

4.3

K76

3.5 K215 Glc H217

E515 UDP

25.14

Lairson et al.

GDP man

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

diphosphate (NDP) leaving group, the relative leaving group ability cannot so easily be manipulated, thus the relative rates of the glycosylation and deglycosylation steps cannot be altered. In order to prevent hydrolysis of metabolically expensive (substituted) phosphodonor sugar substrates, it would be expected that glycosyltransferases would have evolved a mechanism in which the step involving cleavage of the bond between the sugar and its activated leaving group would be rate limiting thereby preventing the accumulation of an (observable) intermediate species, which could be easily hydrolyzed. In addition, because the leaving groups are themselves believed to play the role of base catalyst, the deglycosylation step cannot be slowed down for this class of enzyme by simple mutagenesis of the base catalyst. This inability to alter the relative rates of glycosylation versus deglycosylation has rendered the fluoro sugar approach ineffective in the trapping of intermediates on retaining transferases (87). Possible mechanistic proposals are considered, below, in light of 3-D structural characterization of retaining glycosyltransferases.

Retaining GT-A Glycosyltransferases Structural and mechanistic studies indicate generally utilized strategies among members of retaining GT-A fold glycosyltransferases facilitating leaving group departure and activating the nucleophilic group of the incoming acceptor. However, the observed structures have not revealed a conserved structural architecture on the β-face of the donor sugar substrate in the region that would be expected to be occupied by the obligate enzymatic nucleophile of a proposed double-displacement mechanism. This implies either a different re-

action mechanism or the need for substantial and different conformational changes for each enzyme. Using uridine 5 -diphospho-(2-deoxy-2fluoro)-α-D-galactopyranose (UDP 2F Gal) and 4 -deoxy lactose as the nonreactive donor and acceptor substrate analogues, respectively, a well-resolved ternary complex structure was obtained for the family GT8 galactosyltransferase LgtC from Neisseria meningitidis (88) (Figure 6a). As was seen with the inverting GT-A enzymes, this structure revealed the presence of a Mn2+ cation, coordinated within the active site by the carboxylate side chains of a DXD motif and by the donor substrate diphosphate leaving group, which presumably acts as a Lewis acid to facilitate leaving group departure. In the region that would be occupied by the reactive axial 4 -hydroxyl of the acceptor substrate analogue, the only functional group that is suitably positioned to activate the incoming nucleophile is the leaving group oxygen of the donor substrate. This suggests that the departing diphosphate moiety acts as the base catalyst. Surprisingly, the functional group most suitably positioned within the active site on the β-face of the donor substrate to play the role of the catalytic nucleophile is that of the side chain amide of Gln189. The amide carbonyl oxygen of this residue is perfectly positioned 3.5 A˚ away and has an ideal trajectory for nucleophilic attack on the anomeric reaction center. However, the corresponding alanine mutant retained 3% activity, inconsistent with an essential role for this residue as the catalytic nucleophile. The possibility of a double-displacement mechanism involving nonenzymatic nucleophiles has also been investigated with LgtC because one of the lactose hydroxyls was intriguingly positioned for such a role. In

UDP 2F Gal: uridine 5 -diphospho-(2deoxy-2-fluoro)-αD-galactopyranose

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Figure 6 Comparison of the active sites of several retaining GT-A fold glycosyltransferases. Descriptions of conserved structural features, and a lack thereof, are provided in the text. Mn2+ cations, which play the role of a Lewis acid catalyst that facilitates leaving group departure, are shown as purple spheres. Distances are indicated in angstroms. www.annualreviews.org • Glycosyltransferases

25.15

ARI

29 February 2008

19:15

order to achieve overall net retention of the anomeric configuration, a hydroxyl group from one of the substrates could act as a nucleophile that displaces the UDP leaving group in an initial step. Subsequent attack on the resulting β-linked galactosyl intermediate by the 4 -hydroxyl of the lactose acceptor from the α-face would yield the resulting Gal-(α1,4)-Gal linkage. A mechanism involving the lactose acceptor in this way would have the advantage of minimizing unwanted donor substrate hydrolysis, as it would obligate ternary complex formation prior to the production of a reactive intermediate species. Limiting the plausibility of such a mechanism is the fact that the resulting intermediate would be one of an inherently unreactive glycoside species. This mechanism has been discounted as a result of the finding that the chemically synthesized putative intermediates are not turned over by the enzyme (88). To investigate the degree of nucleophilic character contributed by Gln189 during catalysis, a Q189E variant of LgtC was created. This substitution would introduce a better nucleophile and a worse leaving group for the putative covalent glycosyl-enzyme intermediate (pKa of −0.5 to −1.0 for protonation of the oxygen of an amide versus a pKa of ∼4.0 for that of a carboxylate), which may lead to a rate-limiting deglycosylation step and the potential accumulation of the intermediate species. The results of this work indeed led to the first direct observation of a catalytically relevant covalent glycosyl-enzyme intermediate for a retaining glycosyltransferase (89). However, the site of labeling was found to be a residue (Asp190) that is relatively remote from the anomeric reaction center in the ground state crystal structure. Support for a critical catalytic role for this residue was provided by kinetic analysis of the D190N mutant, which revealed a 3000-fold decrease in the observed turnover rate compared to the wild-type enzyme. However, if this residue were to act as a catalytic nucleophile in a double-displacement mechanism, a significant conformational change from that

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.16

Lairson et al.

of the ternary complex crystal structure would have to occur during the course of catalysis. Furthermore, the lack of sequence or structural conservation implies that not only is a conformational change required but also that this change would have to be different for related enzymes, even those with substantial similarities to LgtC. Subsequently, the 3-D X-ray crystal structure of rabbit muscle glycogenin, another enzyme of the GT8 family, with an intact UDP Glc donor substrate bound within the active site was reported (90). This enzyme catalyzes self-glucosylation of a tyrosine residue in another monomer of the enzyme; plus successive glycosylations of the glucosyl tyrosine formed in a process that is the initial step of glycogen biosynthesis. As was the case with LgtC, a Mn2+ cation was observed within the active site of glycogenin coordinated by the side chain carboxylates of a DXD motif positioned to act as a Lewis acid that activates the departing diphosphate leaving group (Figure 6b). However, unlike LgtC, the side chain most suitably positioned on the β-face of the donor substrate to act as the catalytic nucleophile is that of the carboxylate of Asp163, albeit at a distance of 6.1 A˚ from the reaction center (Figure 6b). This conserved residue, which corresponds to Asp188 of LgtC, is in close proximity to the positively charged side chain of Lys86, and this residue pairing constitutes the only conserved structural motif that can be identified on the β-face of the donor substratebinding sites of retaining glycosyltransferases. In LgtC, the side chain carboxylate of Asp188 is within hydrogen-bonding distance of the C6 and C4 hydroxyls of the galactose moiety of the donor substrate and the positively charged side chain of Arg86 (Figure 6a). In the glycogenin structure, the side chain amide of Gln164, which corresponds to Gln189 in LgtC, is within hydrogen-bonding distance of the C4 hydroxyl of the glucose donor substrate (Figure 6b). These differences in relative positioning indicate that either different donor sugar-binding modes exist among

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

enzymes of the same family and/or that significantly different ground state structures exist, indicating a high degree of structural plasticity for this class of enzyme. As a note of caution, in the described glycogenin structure, the glucose (from UDP Glc) is proposed to bind in a totally different orientation with respect to the UDP sugar moiety compared to that seen on all other GT-A fold retaining glycosyltransferases. Furthermore, this proposal is based upon substantially weaker electron density for the glucose than for the surrounding protein environment and UDP. Bovine α3GalT from family GT6 is another retaining galactosyltransferase possessing a GT-A fold that has been the subject of significant structural and mechanistic investigations. Based on an initial 3-D X-ray crystal structure, Glu317 (structurally equivalent to Gln 189 of LgtC) was proposed to act as the catalytic nucleophile in a doubledisplacement mechanism (91). In fact, in this initial report, the authors claimed to observe a covalently bound intermediate species. This claim was made on the basis of a poor electron density map. This, in combination with a lack of order in the active-site region and the presence of two different crystal forms for the structure containing the proposed covalent intermediate, has led to a general rejection of this interpretation. Later, a detailed study of the binding affinities of wild-type and E317Q enzymes for donor and acceptor substrates led to the proposition that this residue is more likely required for proper acceptor substrate orientation (92). However, in contrast to the mutagenesis results with the equivalent Gln189 residue in LgtC, this study revealed that mutation of Glu317 leads to a 2400-fold decrease in the observed turnover rate, indicating the critical catalytic importance of this residue (92). A structure with the nonreactive donor substrate analogue UDP 2F Gal bound within the active site was recently obtained and revealed the suitable glycosidaselike positioning of the side chain carboxylate of Glu317 poised to play the role of a catalytic nucleophile (93) (Figure 6c). In addi-

tion, it has been shown that the activity of the E317A mutant of α3GalT can be rescued by the addition of small exogenous nucleophiles such as azide, and indeed, a β-glucosyl azide derivative was isolated and characterized (94). Chemical rescue has been successfully employed to identify the acid/base and nucleophile catalytic residues of retaining glycosidases. As such, the chemical rescue of Glu317 is suggestive that this side chain is suitably positioned to play such a catalytic role in the mechanism of α3GalT. However, it must be kept in mind that, although chemical rescue does indicate that a side chain is suitably positioned to play a catalytic role, it does not prove that it plays such a role in the natural enzyme mechanism! It simply indicates the likely positioning of that residue with respect to the reaction center. It is not perhaps surprising that when reactions of a mutant enzyme, having an introduced space into which a good nucleophile can bind within an enzyme active site in the vicinity of a reactive electrophilic center, are run in the presence of high concentrations of a small nucleophile that the enzyme is found to act as a scaffold that catalyzes an unnatural reaction simply by proximity effects. Chemical rescue results alone cannot be used to deduce the natural catalytic mechanism of an enzyme. The conserved structural motif, involving the side chain carboxylate of Asp316 within hydrogen-bonding distance of donor sugar hydroxyls and the positively charged side chain of residue Arg202, is also present in α3GalT (Figure 6c). The α3GalT structures also reveal the presence of a bound divalent cation within the active site suitably positioned to play the role of a Lewis acid that facilitates leaving group departure. In addition, a product complex with bound UDP contained a second Mn2+ bound within the active site in a way that suggests a role in facilitating product release (93). The glycosyltransferases responsible for addition of the specificity-determining Gal/GalNAc moieties to the cell surface glycolipids and glycoproteins that constitute www.annualreviews.org • Glycosyltransferases

25.17

ARI

29 February 2008

19:15

the ABO(H) blood groups are another group of family GT6 enzymes that have been subjected to intensive structural scrutiny (95, 96). Glycosyltransferase A (GTA) specifically uses UDP GalNAc as a donor substrate to modify the 3-OH of the terminal galactose residue of the H antigen [terminal Fucα(1,2)-Gal-β-R acceptors], thereby creating the A blood group antigen. Conversely, glycosyltransferase B (GTB) specifically uses UDP Gal to modify the same hydroxyl of H antigen to generate the B blood group antigen. Remarkably, these two enzymes differ from each other by only 4 out of 354 amino acids. The 3-D X-ray crystal structures of these two enzymes with both UDP and H antigen bound revealed that distinction between the two donor substrates is achieved by the substitutions of residues Leu266 and Gly268 in GTA with the more bulky side chains of Met266 and Ala268 in GTB (97). Similarly, a crystal structure of the inactive mutant of this enzyme, expressed by individuals of the O blood group wherein terminal H antigens are not modified, revealed how a single mutation that changes Gly/Ala268 with the bulky side chain of Arg blocks donor substrate binding and results in the ablation of catalytic activity (98). The role of intramolecular hydrogen bonding within acceptors (99) and the role of the two residues (Gly/Ser235 and Leu/Met266) (100) in the assembly of Type I and II H antigens by GTA and GTB have also been revealed by X-ray crystallographic analysis. Finally, the role of various missense mutations leading to the production of enzymes of weak dual specificity has been investigated (101, 102). The GT27 family contains a series of UDP GalNAc:polypeptide α-N-acetylgalactosaminyltransferases (ppGalNAcTs) that initiate the formation of mucin-type O-linked glycans by catalyzing the transfer of GalNAc to serine or threonine residues on protein surfaces, thereby generating the Tn antigen (GalNAc-α-O-Ser/Thr). These enzymes are particularly interesting in that they possess C-terminal lectin domains. The

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.18

Lairson et al.

majority of ppGalNAcTs, termed peptide transferases, are able to catalyze transfer to both unmodified peptides as well as to glycopeptides. Others, termed glycopeptide transferases are only able to transfer to peptides that have been modified by an initial GalNAc residue (103, 104). The 3-D X-ray crystal structures of murine ppGalNAcT-1 isozyme with only Mn2+ -bound (105), human ppGalNAcT-10 isozyme in complex with either a hydrolyzed donor or a GalNAc-serine acceptor (106) and a human ppGalNAcT-2 isozyme with UDP and a bound acceptor peptide (107) have been reported. These structures indicate that activation of the UDP leaving group is achieved in the usual fashion by the presence of a coordinated Mn2+ cation within the active site poised to act as a Lewis acid (Figure 6d ). In addition, the conserved motif, comprising a side chain carboxylate (Glu345 in ppGalNAcT-10 and Glu334 in ppGalNAcT-2) positioned within hydrogen-bonding distance of donor substrate hydroxyls and a positively charged side chain (Arg221 in ppGalNAcT-10 and Arg208 in ppGalNAcT-2), was again observed (Figure 6d ). The side chain amides of Gln346 of ppGalNAcT-10 and Asn335 of ppGalNAcT-2 are located in a structurally analogous position to that of Gln189 in LgtC and are therefore the most likely candidates to act as the catalytic nucleophiles on the basis of available structural information only (Figure 6d ). These structural studies have also revealed how dynamic association between the lectin and catalytic domains, facilitated by a flexible tether, permits the formation of a range of binding modes for various glycopeptide acceptor substrates thereby facilitating the production of the characteristic high-density glycosylation patterns associated with mucins. The catalytic domain of Clostridium difficile toxin B (ToxB) is a retaining family GT44 glucosyltransferase possessing a GT-A fold that modifies host Rho proteins by glycosylating key threonine residues (108). A 3-D X-ray crystal structure of the catalytic domain with

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

bound Mn2+ , UDP, and glucose has been reported (109). This structure revealed the conserved GT-A retaining architectural features of a bound Mn2+ that acts as a Lewis acid activator as well as a side chain carboxylate (Asp270) within hydrogen-bonding distance of the donor sugar hydroxyls and a positively charged side chain (Arg273) (Figure 6e). The side chain amide of Asn384 is the most suitably positioned candidate for a putative catalytic nucleophile. Mannosylglycerate synthase (MGS) from Rhodothermus marinus is a family GT78 mannosyltransferase that is responsible for the synthesis of the stress protectant 2-O-α-Dmannosylglycerate. An X-ray crystal structure of MGS with Mn2+ and intact GDP Man bound within the active site also revealed the presence of a coordinated divalent cation Lewis acid activator and a side chain carboxylate (Asp192) within hydrogen-bonding distance of donor sugar hydroxyls and a positively charged side chain (Lys72) (Figure 6f ) (27). The main chain amide of Leu163 is the only functional group suitably positioned on the β-face of the anomeric reaction center of the mannosyl donor substrate to act as a catalytic nucleophile (Figure 6f ). Three-dimensional X-ray crystal structures have been reported for the retaining enzymes Kre2 (110), a yeast mannosyltransferase from family GT15, and EXTL2 (111), a mouse N-acetylhexosaminyltransferase involved in heparan biosynthesis from family GT64. Both of these enzymes were shown to possess a GT-A fold; however, the absence of bound donor sugar substrates limits the utility of these structures in identifying candidate catalytic nucleophiles. Both reported structures had bound divalent cations coordinating the nucleoside diphosphate product. A ternary complex structure of EXTL2 with bound UDP and an acceptor substrate supports a role for the phosphate leaving group in acting as the base catalyst that activates the incoming nucleophile. A comparison of representative GT-A fold retaining glycosyltransferases from various

families illuminates conserved structural features within some regions of the active site and a lack thereof in other regions for this class of enzyme. The mode of leaving group activation is similar to that used by the majority of inverting GT-A enzymes. A divalent cation, coordinated by the side chains of a conserved DXD motif, acts as a Lewis acid that facilitates leaving group departure. In contrast to the conserved features in the region surrounding the leaving group and incoming acceptor (the α-face of the donor substrate), there appear to be very few conserved architectural features within the active-site region that accommodates the β-face of the donor. The only strictly conserved feature observed on this face of the reaction center is a side chain carboxylate (from an Asp/Glu residue typically situated on helix 6) that is within hydrogenbonding distance of the donor sugar hydroxyls and a positively charged side chain. The side chain of this Arg/Lys residue is in turn hydrogen bonded to a side chain carboxylate, derived from a residue of the DXD motif, which is in turn coordinated to the bound divalent cation. Interestingly, in the case of glycogenin, the side chain from this β-face residue (Asp163) is the most suitably situated group to act as the catalytic nucleophile in a double-displacement mechanism, indicating how subtle differences in the mode of donor sugar binding can influence our interpretation of a mechanism when based solely on structural information. In most cases, a side chain or main chain amide is most suitably positioned in an appropriate trajectory and at a reasonable distance from the anomeric reaction center to play the role of the catalytic nucleophile in a putative double-displacement mechanism. An exception to this observation is found with family GT6 enzymes that have a side chain carboxylate at this position. This apparent lack of conserved activesite architecture among retaining GT-A glycosyltransferases, all of which appear to have diverged from a common ancestor, is in stark contrast to what has been observed for the analogous retaining glycosidases. In www.annualreviews.org • Glycosyltransferases

25.19

ARI

29 February 2008

19:15

those cases, despite having a multitude of observed 3-D folds, highly conserved pairs of carboxylates are situated within the active site, with one clearly positioned to play the role of catalytic nucleophile. This would indicate a convergence in mechanisms among retaining glycosidases. If all retaining glycosyltransferases utilize the analogous doubledisplacement mechanism, it would seem likely that during the course of divergent evolution the most stringently conserved active-site feature would have been the relative positioning of the most important component of the catalytic machinery. On the basis of the available crystal structures, this feature appears to be absent. This observation may very well suggest that retaining glycosyltransferases utilize a mechanism that differs from that of retaining glycosidases, as discussed below.

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

Retaining GT-B Glycosyltransferases As is the case for the inverting glycosyltransferases possessing a GT-B fold, retaining enzymes with the GT-B fold also use a metal ionindependent method for facilitating leaving group departure. Crystal structures with substrates or inhibitors bound within the donor sugar site reveal the conserved presence of two active-site functional groups on the β-face of the donor substrate, as noted below. Members of family GT35 catalyze the phosphorolysis of α(1,4)-linked glucans leading to the production of α-D-glucose-1phosphate (Glc1P), which is in turn isomerized to glucose-6-phosphate and fed into the glycolytic pathway. The enzymes of this family play an essential role in energy storage/ mobilization in organisms ranging from bacteria to mammals and have a noteworthy history of investigation, with a Nobel Prize awarded to Carl and Gerty Cori in 1947 for their studies of rabbit muscle glycogen phosphorylase (rmGP) (112) and another to Edmond Fischer and Edwin Krebs in 1992 for the first discovery of protein phosphorylation as a control mechanism on eukaryotic members of this family (113, 114). By 25.20

Lairson et al.

contrast, the activities of the bacterial enzymes are controlled at the level of expression (115). The most intensively studied member of family GT35 is that of rmGP, with over 30 structural investigations having been reported to date (for examples, see References 116 and 117). However, despite a long history of exhaustive and laborious research with rmGP, a detailed understanding of the catalytic mechanism remains elusive. Crystal structures are also available for bacterial maltodextrin phosphorylase (MalP) (118–121), yeast glycogen phosphorylase (122), human muscle glycogen phosphorylase (123), and human liver glycogen phosphorylase (124–127). Members of this family are unique among the glycosyltransferases in that they contain a pyridoxal phosphate group covalently bound within their active site via a Schiff base to a lysine residue. The current view is that pyridoxal phosphate acts as a surrogate for the nucleoside monophosphate portion of a nucleoside diphosphate lost during the evolution from an NDP sugar-dependent GT-B glycosyltransferase to a GT-B glycosyltransferaselike phosphorylase (128–131). A ternary complex 3-D X-ray crystal structure of rmGP, with the putative transition state analogue nojirimycin-tetrazole and with phosphate also bound within the active site, revealed the main chain amide of His377 to be the functional group most suitably positioned to play the role of a catalytic nucleophile (117), should one invoke such a mechanism. Soaking (Glc)4 or (Glc)5 maltooligosaccharides into crystals of E. coli MalP with bound GlcP resulted in the production of ternary complex structures with both transfer product and phosphate bound within the active site (120). Like rmGP, the main chain amide of His345 (corresponding to His377 in rmGP) was found to be the most suitably positioned candidate to act as the catalytic nucleophile (Figure 7a). The only other suitable functional group positioned on the β-face of the donor substrate to act as a catalytic nucleophile is the side chain carboxylate of Asp307

ANRV345-BI77-25

ARI

29 February 2008

19:15

a MalP

b WaaG

(pdb 1l6i) (GT35)

(pdb 2iw1) (GT4)

H345

D307

2.8

D19

3.0

5.5

A99

2.8

2.7

3.3 D100

Glc

4.1

Glc

6.3

2.9

UDP 2F Glc

2.6

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

E281

2.7

2.8 3.1

2.8 2.8

K539 R534

2.9

R208

PLP

K209

c OtsA

d AGT

D130

(pdb 1uqu) (GT20)

H140

(pdb 1y6f) (GT72) D115

H154

H114 3.2

2.8

4.9

4.6

5.6

H116 5.8 UDP Glc E306

D361

UDP Glc 3.0

K267

2.7 2.3

K209

3.4

R262 R204

Figure 7 Comparison of the active sites of several retaining GT-B fold glycosyltransferases. Distances are indicated in angstroms.

(Figure 7a). However, this carboxylate is situated within hydrogen-bonding distance of the C2- and C3-hydroxyls of the acceptor glucose residue and therefore most likely serves to recognize and orient the incoming acceptor. From the MalP structure, it is apparent that formation of a hydrogen bond between

the side chain imidazole of His 345 and the C6-hydroxyl of what would have been the glucose donor sugar results in the closure of a flexible loop region that makes up a significant portion of the maltose acceptor recognition site (Figure 7a). Analogous to the electrostatic strategy of inverting GT-B enzymes, the bound phosphate product is located within www.annualreviews.org • Glycosyltransferases

25.21

ARI

29 February 2008

19:15

hydrogen-bonding distances of the positively charged MalP side chains of Arg534 and Lys539, indicating a role for these residues in stabilizing the departing phosphate leaving group (Figure 7a). Three-dimensional X-ray crystal structures with bound intact donor substrates/ analogues are also available for retaining GT-B fold glycosyltransferases from families GT4, GT20, and GT72. The structure of E. coli α-(1,3) glucosyltransferase WaaG from family GT4, involved in lipopolysaccharide biosynthesis, was solved with bound nonreactive donor sugar substrate analogue uridine 5 -diphospho-(2-deoxy-2-fluoro)-αD-glucopyranose (UDP 2F Glc) (Figure 7b) (132). Such 2-deoxy-2-fluorosugar analogues of NDP-sugars have proved valuable donor substrate analogues that do not undergo reaction at the glycosyltransferase active site because the fluorine inductively destabilizes transition states for glycosyl transfer, effectively prohibiting reaction. Only in the case of GT80 sialyl transferases has any turnover been observed with such fluorosugar analogues (69). Similarly, structures of E. coli OtsA from family GT20, responsible for the biosynthesis of the very interesting stress response molecule α,α-1,1 trehalose-6-phosphate, were obtained with either UDP 2F Glc or UDP Glc (Figure 7c) bound (133). Finally, ternary complex structures of AGT from family GT72, the retaining DNA-modifying glucosyltransferase counterpart from the T4 bacteriophage of the inverting BGT, were obtained with UDP Glc and various DNA acceptor substrates bound (134) (Figure 7d ). Like BGT, AGT was shown to use a “base-flipping” mechanism to activate DNA acceptors. Crystal structures of bacterial (135) and archael (136) glycogen phosphorylases from family GT5 are also available. However, the absence of the sugar moiety of the donor substrate limits their utility in identifying candidate nucleophilic residues. A comparison of these structures, along with those of the rmGP and MalP ternary

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.22

Lairson et al.

complexes, facilitates an initial identification of several conserved active-site structural features among GT-B fold glycosyltranferases. On the β-face of the bound donor sugar substrate, a main chain amide is most suitably positioned, at an appropriate distance and trajectory, to act as the catalytic nucleophile in a double-displacement mechanism. The conserved presence of hydrogen bonding partners, donated from main chain carbonyl or side chain amide groups, in proximity to the main chain NH group could help facilitate this catalytic role. The only other candidate for the role of nucleophile is that of the side chain carboxylate of a structurally conserved aspartate (Asp307, Asp100, Asp130, Asp115 in MalP, WaaG, OtsA, and AGT, respectively, Figure 7) situated ∼6 A˚ from the anomeric reaction center. This residue is more likely involved in acceptor substrate recognition/orientation, and the plausibility of a nucleophilic role for this residue is severely limited by mutagenesis investigations with AGT in which it was found that the D115A mutant displayed ∼10% residual transferase activity (134). Another conserved feature on the β-face of the donor is the hydrogen bond formed between the donor sugar C6-hydroxyl and, with the exception of WaaG where Asp19 is the bonding partner, the side chain imidazole of the His residue whose main chain amide is positioned to act as a nucleophile (His354/377, His154, and His114 in MalP/rmGP, OtsA, and AGT, respectively, Figure 7). As mentioned above, the formation of this interaction is believed to cause the closure of a flexible loop leading to the complete formation of the acceptor recognition site. On the αface of the donor substrate, a structurally conserved Arg/Lys pair of positively charged side chains is present within hydrogen-bonding distance of the phosphate leaving group. These are poised to replace the role of a divalent cation in GT-A fold enzymes, electrostatically stabilizing the leaving group departure (Figure 7). The ternary complex structure of MalP with bound products suggests

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

Figure 8 (a) The SN i mechanism for the decomposition of alkyl chlorosulfites. (b) A comparison of the intermediates involved in the SN i mechanism for the decomposition of alkyl chlorosulfites and in the proposed mechanisms for the solvolysis of glycosyl fluorides and of retaining glycosyltransferases.

that the substrate/product phosphate group is most suitably positioned to play the role of the base catalyst that activates the functional group of the incoming acceptor (Figure 7a).

An Alternative SN i-Like Mechanism Albeit rather rare, reactions involving a single nucleophilic displacement step can lead to the formation of a product in which the stereochemistry of the reaction center is completely retained. This is a special case of the SN 1 mechanism that involves the formation of discrete ion pair intermediates, with lifetimes longer than a bond vibration, which can collapse to give back the starting material or a product (Figure 8a). This unique form of SN 1 reaction, termed SN i wherein the “i” indicates internal return, was first described to account for the observed products and the nature of the reaction involved with the decomposition of alkyl chlorosulfites (137–139). The unique feature of the SN i mechanism that accounts

for the exclusive retention of stereochemistry is that the leaving group undergoes decomposition leading to the production of a nucleophile that is held as an ion pair on the same face as the leaving group. Decomposition of the initial intermediate species and attack by the formed nucleophile occur at a rate that exceeds that of solvent attack and the ion pair reorganization that would be required for nucleophilic attack from the back face. Many authors now propose SN i-like mechanisms for retaining glycosyltransferases; this potentially confusing nomenclature demands some historical explanation. In 1980, Sinnott & Jencks (140) reported that the solvolysis of glucose derivatives in mixtures of ethanol and trifluoroethanol could, under certain conditions, proceed with retention of anomeric configuration via a mechanism that they considered “a type of internal return.” The rationale for this description was derived from the proposed requirement for the formation of a www.annualreviews.org • Glycosyltransferases

SN i: a form of SN 1 reaction in which the nucleophile is derived by decomposition of the leaving group and attacks from the same face

25.23

ARI

29 February 2008

19:15

hydrogen bond between the departing fluoride ion and the attacking nucleophile. The stereoelectronic context of this species is extremely similar to that displayed in the second intermediate species of alkyl chlorosulfite decomposition (Figure 8a), although in the latter case the nucleophile was previously a covalently attached part of the substrate, hence its “internal” nature. Sinnott & Jencks thus drew the analogy; if one considers F-. . .OR as a single species, then the solvolysis of glycosyl fluorides is a type of internal return. They further suggested that a sufficiently “open” transition state can allow for a nucleophilic push on the reaction center by the entering solvent molecule on either the opposite or the same face as the leaving group, thereby directly challenging the classical Ingold view of SN 2 displacement reactions involving Walden inversion. A mechanism termed SN i-like was subsequently proposed for glycogen phosphorylase (141), perhaps unfortunately given that the acceptor is external and not internal. In light of structural, mutagenesis, and mechanistic results with LgtC, the SN i-like mechanism was proposed (88) and has since been suggested for the structurally defined retaining transferases Kre2 (110), ToxB (109), Extl2 (111), Mgs (27), WaaG (132), OtsA (142), and AGT (143) primarily because of the lack of an appropriately positioned nucleophile within their active sites and also on account of the likely interaction between the departing phosphate and the incoming acceptor nucleophile. One should heed Sinnott’s words on this topic; “the evidence in favor of the internal return mechanism. . .is essentially negative. . . .” (144) This mechanism is often advocated simply because of the lack of supporting evidence for other mechanisms. It is useful to draw parallels from the decomposition of alkyl chlorosulfites through the solvolysis of glycosyl fluorides to the retaining glycosyltransferase reaction (Figure 8b). But such a mechanism should, in the absence of evidence to the contrary, follow the rules of Woodward and Hoffman

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.24

Lairson et al.

(145, 146) and should not be drawn as a single exploded four-center transition state as it frequently is. An enzyme-catalyzed SN ilike mechanism would in all likelihood proceed with the required formation of a shortlived ion pair intermediate with a lifetime longer than that of a bond vibration (discussed below).

Evolutionary Constraints upon Retaining Glycosyltransferase Mechanisms By contemplating the inherent differences in the reactivities of the substrates utilized by glycosidases and glycosyltransferases, a rationale can be developed as to why these two classes of enzymes may have evolved differing mechanistic strategies for catalyzing glycosyl group transfer reactions with net retention of anomeric configuration. Glycosidases catalyze the hydrolysis of one of the most stable linkages found in nature, the glycosidic bond between two sugars, with the half-lives for spontaneous hydrolysis being on the order of ∼5 million years (76). The stability of this linkage, under neutral conditions, is imparted by the relatively poor leaving group ability (pKa ∼14) of natural glycosidase substrates. This stability leads to a large energy barrier to bond cleavage in the first step of the glycosidase reaction (Figure 9). It would seem logical that retaining glycosidases would have evolved a catalytic mechanism involving an intermediate and, in accordance with the Hammond postulate, also a transition state of the lowest possible free energy. In other words, by selecting a mechanism with a lower energy intermediate species, catalytic efficiency is easier to achieve because the ratedetermining transition state being stabilized will also have a lower free energy. By this logic, the Koshland SN 2-like double-displacement mechanism, involving the formation of a covalently bound glycosyl-enzyme intermediate, would have been preferred over the Phillips SN 1-like mechanism, involving the formation of a higher energy oxocarbenium ion

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

intermediate, during the evolution of the retaining glycosidase mechanism (Figure 9). In contrast, glycosyltransferases utilize high-energy (substituted) phosphosugar donor substrates. The leaving group ability for the donor substrate is much greater to start with, compared to glycosidase substrates, with a pKa of ∼7 for the second ionization of phosphate. The higher free energy of the starting material and lower free energy of the first transition state make the barrier for bond cleavage in the first step of the glycosyltransferase reaction less than that of the glycosidase reaction (Figure 9). Because there is not such a large barrier for the initial bond cleavage step compared to that for the analogous glycosidase reaction, an SN 1-like pathway involving the formation of an intermediate oxocarbenium ion may have been accessible, and sufficient to have been selected for, during the course of the evolution of the mechanisms of retaining glycosyltransferases (Figure 9). Precedence for enzymatic mechanisms involving SN 1-like pathways, and for the intermediate formation of cationic species, can be derived from enzymes involved in terpenoid biosynthesis: isopentenyl diphosphate isomerases, FPP synthetases, and the terpene cyclases. These enzymes use SN 1-like mechanisms involving the formation of carbocation intermediates stabilized by either aromatic side chains (pi-cationic interactions), the carbonyl oxygens of main or side chain amides, and/or side chain hydroxyls (147–153). The chemistry involved in FPP synthetase-catalyzed reactions is rather analogous to that involved in glycosyltransferase catalysis. In both cases, a nucleophilic substitution reaction occurs at a carbon center with a (substituted) phosphate acting as the leaving group. Stereochemically, they are more like inverting enzymes in that they preorganize the nucleophilic alkene in a position to directly trap the carbocation and do not suffer the same stereochemical impediment as do the retaining enzymes. However, analogies between the overall chemistries

being catalyzed provide a foundation for a comparison of the transition states chosen to be stabilized and the possibility of a convergence in mechanism during the course of evolution. As is observed for glycosyltransferases, FPP synthetases use a divalent cation and the positively charged side chains donated from a conserved Arg/Lys pair to stabilize charge development and facilitate departure of the phosphate leaving group (84, 85). Intriguingly, conserved electrostatic interactions with neutral carbonyl groups, derived from the amides of a main chain (Lys202) and a side chain (Gln241), plus a side chain hydroxyl (Thr203), are found within the FPP synthetase active site and are thought to stabilize the positive charge distributed over the C1, C2, and C3 atoms of the cationic intermediate (84, 85). This is similar to the observed positioning of the carbonyl oxygen of a main chain or a side chain amide in close proximity (on the β-face) to the anomeric reaction center within the active sites of the vast majority of structurally characterized retaining glycosyltransferases. In both cases, formation of a cationic intermediate is also stabilized by inherent properties of the substrates. In the case of glycosyltransferases, positive charge formation at the anomeric center of an oxocarbenium ion is stabilized by a lone pair of nonbonding electrons donated by the endocyclic oxygen of the sugar ring. In the case of FPP synthetases, an allylic tertiary center stabilizes positive charge development at the carbon center that undergoes bond fission. Also, as is proposed for retaining glycosyltransferases, FPP synthetases appear to position the phosphate leaving group in such a manner that it is suitably positioned to play the role of the base catalyst required for proton abstraction in the elimination step of the reaction. These described similarities in the chemistries catalyzed and the observed conservation of structural architecture might well suggest that retaining glycosyltransferases utilize an SN 1-like mechanism, analogous to that of FPP synthetase, which given the need for interaction between leaving group and www.annualreviews.org • Glycosyltransferases

25.25

25.26

Lairson et al. δ+ O

–

O

Nuc

Covalent glycosylenzyme intermediate (SN2-like pathway)

HO

O

GH product complex

b

GH substrate complex

δ+

O

δ+

NH2

δ+

O

Nuc

Covalent glycosylenzyme intermediate (SN2-like pathway)

HO

Oxocarbenium ion intermediate (SN1-like pathway)

HO

GH product complex

Comparison of the potential differences in the free energy barriers for SN 1-like and SN 2-like pathways that could be utilized by (a) retaining glycosidases (GHs) and (b) retaining glycosyltransferases (GTs). Because glycosyltransferases utilize high-energy (substituted) phosphosugar donor substrates containing a very good leaving group (pKa of ∼2 for the coordinated species), the free energy barrier for the first step is much less than that for the glycosidases that utilize a very stable substrate containing a poor leaving group (pKa of ∼14). This smaller free energy barrier for the first step of the reaction may have led to the SN 1-like pathway being sufficiently effective to be selected for during the course of the evolution of retaining glycosyltransferase mechanisms. In addition, for glycosyltransferases, because of the near equality in leaving group ability for both steps, utilization of a side chain carboxylate as a catalytic nucleophile (as is the case for the vast majority of retaining glycosidases) in a double-displacement mechanism might lead to a rate-limiting deglycosylation step. This would increase the probability of unwanted donor substrate hydrolysis.

Figure 9

GH substrate complex

δ+

O

29 February 2008

Oxocarbenium ion intermediate (SN1-like pathway)

HO

ARI

a O

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25 19:15

ANRV345-BI77-25

ARI

29 February 2008

19:15

attacking nucleophile could be considered a type of internal return.

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

Concluding Remarks on the Mechanism of Retaining Glycosyltransferases Presently, the catalytic mechanism of retaining glycosyltransferases is a topic without clear or definitive answers. By analogy to the well-characterized glycosidases, a doubledisplacement mechanism involving discrete nucleophilic catalysis and the formation of a covalently bound glycosyl-enzyme intermediate would seem possible. Such a mechanism necessitates an appropriately positioned enzymatic nucleophile within the active site on the β-face of the donor substrate in close proximity to the anomeric reaction center. However, candidates for this role are not clear because in the majority of cases the most suitably positioned functional group is a main chain or side chain amide group. The plausibility of a nucleophilic role for these groups is compromised by what is observed for the family GT8 retaining enzyme LgtC, in which mutation of the glutamine (Gln189) whose side chain is so positioned results in an unexpectedly modest decrease in observed rates. Furthermore, the finding that a competent glycosyl-enzyme intermediate involving Asp190 is trapped by substrate in the Q189E variant of LgtC confused issues somewhat but suggests a highly plastic active site. Alternatively, LgtC may well use a SN 1-like pathway, and the altered electrostatic environment in the active site of the Q189E mutant results in an electrophilic oxocarbenium ion being quenched by a nearby nucleophilic side chain. A mechanism involving the formation of an ionic intermediate might be more reasonable for glycosyltransferases than for glycoside hydrolases, given the greater reactivity of their substrates. However, it is also possible that some retaining glycosyltransferases use a distinctively SN 2-like mechanism involving formation of a covalently bound glycosylenzyme intermediate as suggested for the fam-

ily GT6 retaining enzyme α3GalT. Indeed, the presence of a correctly disposed carboxylate side chain and the catalytic rescue afforded to mutants at that position by azide supports such a notion. Overall, it is most probable that a mechanistic continuum exists for this class of enzyme, analogous to that of nonenzymatic nucleophilic substitution reactions for which the SN 2 direct displacement and SN 1 ionic mechanisms are simply the extremes. Between these two extremes, a continuum of nucleophilic substitution mechanisms exists that is best defined using the International Union for Pure and Applied Chemistry (IUPAC)-recommended designations for reaction mechanisms (154, 155) (see the sidebar IUPAC-Recommended Nomenclature for Reaction Mechanisms). These involve ion pair formation and can have the kinetic properties of an SN 2 process yet produce

IUPAC-RECOMMENDED NOMENCLATURE FOR REACTION MECHANISMS In 1989, following several years of vigorous debate, the IUPAC commission on physical organic chemistry described a system for the systematic description of reaction mechanisms (154, 159). This system was proposed to replace the Ingold system that served to define the type of transformation and the empirically observed molecularity (e.g., SN 2) but not the details of the mechanism (e.g., concerted versus step wise). Although the use of this system may appear to be a purely semantic exercise, its benefit is clear in light of the frequent unnecessary controversies that arise during written discussions of reaction mechanisms, resulting from ambiguities of the meaning of the Ingold definitions. This argument is particularly applicable to the discussion of the mechanism of retaining glycosyltransferases. Using the IUPAC recommended nomenclature, the SN i-like mechanism described for retaining glycosyltransferases would be defined as DN ∗ ANss , where the asterisk represents the formation of a short-lived intermediate and “ss ” represents the formation of a solvent-separated ion pair (Figure 10). Although the term “solvent separated” seems unusual in the context of an enzymatic mechanism, it is required to differentiate it from an “intimate” ion pair that would not allow for attack by an external acceptor from the front face.

www.annualreviews.org • Glycosyltransferases

25.27

ARI

29 February 2008

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

19:15

Figure 10 Proposed DN ∗ ANss ion pair mechanism for retaining glycosyltransferases.

stereochemical outcomes consistent with an SN 1 reaction (and vice versa) (156–158). The mechanism that seems likely for most retaining glycosyltransferases, including LgtC, is one involving a short-lived ion pair intermediate that requires a back-side nucleophilic “push” (without discrete covalent interaction) for formation (Figure 10). This ion pair could collapse to give back starting material or, following a slight shift in the activesite positioning of the cation, could undergo 25.28

Lairson et al.

front-side attack by an incoming nucleophile, leading to a product with retained stereochemistry. Indeed, consideration of the reactivities of the substrates in question, the lack of conserved structural architecture among retaining glycosyltransferase β-face active-site regions, and the biological precedents for such intermediates in related cationic mechanisms suggest that this is the most probable mechanism for the majority of retaining glycosyltransferases.

ANRV345-BI77-25

ARI

29 February 2008

19:15

SUMMARY POINTS

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

1. Glycosyltransferases catalyze group transfer reactions with either net inversion or retention of stereochemistry at the anomeric reaction with respect to the (substituted) phosphosugar donor substrate, thereby necessitating differing catalytic mechanisms for these two classes of enzymes. 2. Thus far, the 3-D structures of glycosyltransferases have revealed two major structural folds, with others just now being uncovered. The GT-A and GT-B folds, which incorporate a pair of Rossmann folds in each case, are found for nucleoside phosphosugar-dependent glycosyltransferases. Lipid phosphosugar-dependent glycosyltransferases appear to adopt different folds, as exemplified by the lysozyme-like fold reported for a peptidoglycan synthase and the novel fold recently found for an oligosaccharyltransferase. 3. The mechanism of inverting glycosyltransferases is that of a straightforward SN 2-like reaction facilitated by an enzymatic base catalyst and by Lewis acid activation of the departing (substituted) phosphate leaving group. 4. The mechanism of retaining glycosyltransferases remains less clear. Although some members of this class may utilize Koshland’s double-displacement mechanism involving the formation of a covalently bound glycosyl-enzyme intermediate, a mechanism involving the formation of a short-lived ion pair intermediate seems likely for the majority of the members of this class of enzyme. 5. There is no correlation of overall fold with catalytic mechanism. Inverting and retaining enzymes are known with both GT-A and GT-B topologies.

FUTURE ISSUES 1. Does the predicted GT-C fold have any predictive relevance beyond indicating the presence of a large trans-membrane component? First indications from the very recently solved structure of the oligosaccharyltransferase STT3 suggest it may not. 2. Will a whole new series of folds for nonnucleoside phosphosugar-dependent glycosyltransferases be discovered? In the absence of the constraints of (nucleotide-binding) Rossmann folds, the structural diversity may well rival that of the glycosidases. 3. When is a transferase a class of hydrolase? Both glycoside hydrolases and transferases catalyze glycosyl group transfer, and whether an enzyme is classified as one or the other follows few clear rules. Classification is sometimes based upon thermodynamics, often upon sequence, and occasionally upon structural homologies, and it is frequently influenced by historical perspective. For example, family GT51 is demonstrably a glycosyltransferase but could, arguably, be reclassified to a glycoside hydrolase family given its structural similarities to hydrolyzing lysozymes. This can occasionally be a confusing area for the nonspecialist. 4. The overall folds of multiple “orphan” families of glycosyltransferases remain to be determined.

www.annualreviews.org • Glycosyltransferases

25.29

ANRV345-BI77-25

ARI

29 February 2008

19:15

5. The catalytic mechanism of retaining glycosyltransferases remains ill defined. Whether some members (e.g., family GT6) utilize a double-displacement mechanism and others an ion pair mechanism remains to be determined. 6. The structure and mechanism of the major polymerizing glycosyltransferases, such as those involved in chitin, cellulose, and hyaluronan biosynthesis, need to be determined as well as a resolution of the continuing controversies as to whether the UDP-monosaccharide is the glycosyl donor (nonreducing end addition) or the UDPgrowing chain (reducing end addition). 7. Many, probably all, glycosyltransferases display conformational changes upon ligand binding and catalysis. For most enzymes the nature and extent of these changes has not been well-defined. Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

8. Perhaps the greatest challenge in the field is the functional characterization of glycosyltransferases. There are over 33,000 open reading frames known that encode this class of enzyme (as of January 2008), yet the donor and acceptor specificity for the vast majority (>95%) is not known.

DISCLOSURE STATEMENT The authors are not aware of any biases that might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS L.L.L is the receipient of a Natural Science and Engineering Reseach Council (NSERC) doctoral postgraduate scholarship and a Michael Smith Foundation for Health Research (MSFHR) senior graduate fellowship. G.J.D. is a Royal Society-Wolfson Research Merit Award recipient. B.H. thanks the Centre National de la Recherche Scientifique for supporting the visit of S.G.W. to France. S.G.W. thanks the Natural Sciences and Engineering Research Council of Canada and the Canadian Institutes for Health Research for continuing support.

LITERATURE CITED 1. Rademacher TW, Parekh RB, Dwek RA. 1988. Annu. Rev. Biochem. 57:785–838 2. Davies GS, Withers SG. 1998. In Comprehensive Biological Catalysis, ed. ML Sinnott, pp. 119–208. London: Academic 3. Zechel DL, Withers SG. 2000. Acc. Chem. Res. 33:11–18 4. Campbell JA, Davies GJ, Bulone V, Henrissat B. 1997. Biochem. J. 326:929–39 5. Coutinho PM, Deleury E, Davies GJ, Henrissat B. 2003. J. Mol. Biol. 328:307–17 6. Bourne Y, Henrissat B. 2001. Curr. Opin. Struc. Biol. 11:593–600 7. Unligil UM, Rini JM. 2000. Curr. Opin. Struc. Biol. 10:510–17 8. Hu YN, Walker S. 2002. Chem. Biol. 9:1287–96 9. Charnock SJ, Davies GJ. 1999. Biochemistry 38:6380–85 10. Breton C, Imberty A. 1999. Curr. Opin. Struc. Biol. 9:563–71 11. Brown K, Pompeo F, Dixon S, Mengin-Lecreulx D, Cambillau C, Bourne Y. 1999. EMBO J. 18:4096–107 25.30

Lairson et al.

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

12. Chiu CPC, Watts AG, Lairson LL, Gilbert M, Lim D, et al. 2004. Nat. Struct. Mol. Biol. 11:163–70 13. Chiu CPC, Lairson LL, Gilbert M, Wakarchuk WW, Withers SG, Strynadka NCJ. 2007. Biochemistry 46:7196–204 14. Wiggins CAR, Munro S. 1998. Proc. Natl. Acad. Sci. USA 95:7945–50 15. Breton C, Bettler E, Joziasse DH, Geremia RA, Imberty A. 1998. J. Biochem. 123:1000–9 16. Pak JE, Arnoux P, Zhou SH, Sivarajah P, Satkunarajah M, et al. 2006. J. Biol. Chem. 281:26693–701 17. Vrielink A, Ruger W, Driessen HPC, Freemont PS. 1994. EMBO J. 13:3413–22 18. Campbell RE, Mosimann SC, Tanner ME, Strynadka NCJ. 2000. Biochemistry 39:14993– 5001 19. Liu J, Mushegian A. 2003. Protein Sci. 12:1418–31 20. Strahlbolsinger S, Immervoll T, Deutzmann R, Tanner W. 1993. Proc. Natl. Acad. Sci. USA 90:8164–68 21. Takahashi M, Inoue N, Ohishi K, Maeda Y, Nakamura N, et al. 1996. EMBO J. 15:4254– 61 22. Maeda Y, Watanabe R, Harris CL, Hong JJ, Ohishi K, et al. 2001. EMBO J. 20:250–61 23. Kikuchi N, Kwon YD, Gotoh M, Narimatsu H. 2003. Biochem. Biophys. Res. Commun. 310:574–79 23a. Igura M, Maita N, Kamishikiryo J, Yamada M, Obita T, et al. 2007. EMBO J. 27:234–43 24. Hidaka M, Honda Y, Kitaoka M, Nirasawa S, Hayashi K, et al. 2004. Structure 12:937–47 25. Lovering AL, de Castro LH, Lim D, Strynadka NCJ. 2007. Science 315:1402–5 26. Yuan YQ, Barrett D, Zhang Y, Kahne D, Sliz P, Walker S. 2007. Proc. Natl. Acad. Sci. USA 104:5348–53 27. Flint J, Taylor E, Yang M, Bolam DN, Tailford LE, et al. 2005. Nat. Struct. Mol. Biol. 12:608–14 28. Tarbouriech N, Charnock SJ, Davies GJ. 2001. J. Mol. Biol. 314:655–61 29. Gastinel LN, Cambillau C, Bourne Y. 1999. EMBO J. 18:3546–57 30. Ramakrishnan B, Qasba PK. 2001. J. Mol. Biol. 310:205–18 31. Ramakrishnan B, Shah PS, Qasba PK. 2001. J. Biol. Chem. 276:37665–71 32. Ramakrishnan B, Balaji PV, Qasba PK. 2002. J. Mol. Biol. 318:491–502 33. Ramasamy V, Ramakrishnan B, Boeggeman E, Ratner DM, Seeberger PH, Qasba PK. 2005. J. Mol. Biol. 353:53–67 34. Ramakrishnan B, Ramasamy V, Qasba PK. 2006. J. Mol. Biol. 357:1619–33 35. Ramakrishnan B, Qasba PK. 2002. J. Biol. Chem. 277:20833–39 36. Unligil UM, Zhou SH, Yuwaraj S, Sarkar M, Schachter H, Rini JM. 2000. EMBO J. 19:5269–80 37. Gordon RD, Sivarajah P, Satkunarajah M, Ma D, Tarling CA, et al. 2006. J. Mol. Biol. 360:67–79 38. Pedersen LC, Tsuchida K, Kitagawa H, Sugahara K, Darden TA, Negishi M. 2000. J. Biol. Chem. 275:34580–85 39. Pedersen LC, Darden TA, Negishi M. 2002. J. Biol. Chem. 277:21869–73 40. Jinek M, Chen YW, Clausen H, Cohen SM, Conti E. 2006. Nat. Struct. Mol. Biol. 13:945– 46 41. Ohtsubo K, Imajo S, Ishiguro M, Nakatani T, Oka S, Kawasaki T. 2000. J. Biochem. 128:283–91 42. Kakuda S, Shiba T, Ishiguro M, Tagawa H, Oka S, et al. 2004. J. Biol. Chem. 279:22693– 703 www.annualreviews.org • Glycosyltransferases

25.31

ARI

29 February 2008

19:15

43. Garinot-Schneider C, Lellouch AC, Geremia RA. 2000. J. Biol. Chem. 275:31407–13 44. Kozmon S, Tvaroska I. 2006. J. Am. Chem. Soc. 128:16921–27 45. Koyama M, Helbert W, Imai T, Sugiyama J, Henrissat B. 1997. Proc. Natl. Acad. Sci. USA 94:9091–95 46. Kamst E, Bakkers J, Quaedvlieg NEM, Pilling J, Kijne JW, et al. 1999. Biochemistry 38:4045–52 47. Chang R, Yeager AR, Finney NS. 2003. Org. Biomol. Chem. 1:39–41 48. DeAngelis PL. 1999. J. Biol. Chem. 274:26557–62 49. Tlapak-Simmons VL, Baron CA, Gotschall R, Haque D, Canfield WM, Weigel PH. 2005. J. Biol. Chem. 280:13012–18 50. Kornberg SR, Zimmerman SB, Kornberg A. 1961. J. Biol. Chem. 236:1487–93 51. Morera S, Lariviere L, Kurzeck J, Aschke-Sonnenborn U, Freemont PS, et al. 2001. J. Mol. Biol. 311:569–77 52. Lariviere L, Morera S. 2002. J. Mol. Biol. 324:483–90 53. Lariviere L, Gueguen-Chaignon V, Morera S. 2003. J. Mol. Biol. 330:1077–86 54. Mulichak AM, Losey HC, Lu W, Wawrzak Z, Walsh CT, Garavito RM. 2003. Proc. Natl. Acad. Sci. USA 100:9238–43 55. Mulichak AM, Losey HC, Walsh CT, Garavito RM. 2001. Structure 9:547–57 56. Mulichak AM, Lu W, Losey HC, Walsh CT, Garavito RM. 2004. Biochemistry 43:5170– 80 57. Shao H, He XZ, Achnine L, Blount JW, Dixon RA, Wang XQ. 2005. Plant Cell 17:3141– 54 58. Offen W, Martinez-Fleites C, Yang M, Kiat-Lim E, Davis BG, et al. 2006. EMBO J. 25:1396–405 59. Bolam DN, Roberts S, Proctor MR, Turkenburg JP, Dodson EJ, et al. 2007. Proc. Natl. Acad. Sci. USA 104:5336–41 60. Miley MJ, Zielinska AK, Keenan JE, Bratton SM, Radominska-Pandya A, Redinbo MR. 2007. J. Mol. Biol. 369:498–511 60a. Brazier-Hicks M, Offen WA, Gershater MC, Revett TJ, Lim EK, et al. 2007. Proc. Natl. Acad. Sci. USA 104:20238–43 61. Grizot S, Salem M, Vongsouthi V, Durand L, Moreau F, et al. 2006. J. Mol. Biol. 363:383– 94 62. Sun HY, Lin SW, Ko TP, Pan JF, Liu CL, et al. 2007. J. Biol. Chem. 282:9973–82 63. Ihara H, Ikeda Y, Toma S, Wang XC, Suzuki T, et al. 2007. Glycobiology 17:455–66 64. Ha S, Walker D, Shi YG, Walker S. 2000. Protein Sci. 9:1045–52 65. Hu YN, Chen L, Ha S, Gross B, Falcone B, et al. 2003. Proc. Natl. Acad. Sci. USA 100:845–49 66. Yu H, Chokhawala H, Karpel R, Yu H, Wu BY, et al. 2005. J. Am. Chem. Soc. 127:17618– 19 67. Zhang CS, Griffith BR, Fu Q, Albermann C, Fu X, et al. 2006. Science 313:1291–94 68. Ni LS, Sung MC, Yu H, Chokhawala H, Chen X, Fisher AJ. 2006. Biochemistry 45:2139– 48 69. Ni L, Chokhawala HA, Cao H, Henning R, Ng L, et al. 2007. Biochemistry 46:6288–98 70. Yip VLY, Varrot A, Davies GJ, Rajan SS, Yang XJ, et al. 2004. J. Am. Chem. Soc. 126:8354– 55 71. Liu QYP, Sulzenbacher G, Yuan HP, Bennett EP, Pietz G, et al. 2007. Nat. Biotechnol. 25:454–64 72. Mark BL, James MNG. 2002. Can. J. Chem. 80:1064–74

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.32

Lairson et al.

ANRV345-BI77-25

73. 74. 75. 76. 77.

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104.

ARI

29 February 2008

19:15

Koshland DE. 1953. Biol. Rev. Camb. Phil. Soc. 28:416–36 Phillips DC. 1966. Sci. Am. 215:78–90 Phillips DC. 1967. Proc. Natl. Acad. Sci. USA 57:484–95 Wolfenden R, Lu XD, Young G. 1998. J. Am. Chem. Soc. 120:6814–15 Fersht A. 1999. Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein Folding. New York: Freeman Sinnott ML, Souchard IJ. 1973. Biochem. J. 133:89–98 Kempton JB, Withers SG. 1992. Biochemistry 31:9961–69 Vocadlo DJ, Wicki J, Rupitz K, Withers SG. 2002. Biochemistry 41:9727–35 Huang X, Tanaka KSE, Bennet AJ. 1997. J. Am. Chem. Soc. 119:11147–54 Berti PJ, Tanaka KSE. 2002. In Advances in Physical Organic Chemistry, Vol 37. pp. 239– 314, ed. T Tidwell, J Richard. Oxford, UK: Elsevier Vocadlo DJ, Davies GJ, Laine R, Withers SG. 2001. Nature 412:835–38 Hosfield DJ, Zhang YM, Dougan DR, Broun A, Tari LW, et al. 2004. J. Biol. Chem. 279:8526–29 Gabelli SB, McLellan JS, Montalvetti A, Oldfield E, Docampo R, Amzel LM. 2006. Proteins 62:80–8 Withers SG, Aebersold R. 1995. Protein Sci. 4:361–72 Ly HD, Lougheed B, Wakarchuk WW, Withers SG. 2002. Biochemistry 41:5075–85 Persson K, Ly HD, Dieckelmann M, Wakarchuk WW, Withers SG, Strynadka NCJ. 2001. Nat. Struct. Biol. 8:166–75 Lairson LL, Chiu CPC, Ly HD, He SM, Wakarchuk WW, et al. 2004. J. Biol. Chem. 279:28339–44 Gibbons BJ, Roach PJ, Hurley TD. 2002. J. Mol. Biol. 319:463–77 Gastinel LN, Bignon C, Misra AK, Hindsgaul O, Shaper JH, Joziasse DH. 2001. EMBO J. 20:638–49 Zhang YN, Swaminathan GJ, Deshpande A, Boix E, Natesh R, et al. 2003. Biochemistry 42:13512–21 Jamaluddin H, Tumbale P, Withers SG, Acharya KR, Brew K. 2007. J. Mol. Biol. 369:1270–81 Monegal A, Planas A. 2006. J. Am. Chem. Soc. 128:16030–31 Watkins WM, Morgan WTJ. 1957. Nature 180:1038–41 Yamamoto F, Clausen H, White T, Marken J, Hakomori SI. 1990. Nature 345:229–33 Patenaude SI, Seto NOL, Borisova SN, Szpacenko A, Marcus SL, et al. 2002. Nat. Struct. Biol. 9:685–90 Lee HJ, Barry CH, Borisova SN, Seto NOL, Zheng RXB, et al. 2005. J. Biol. Chem. 280:525–29 Nguyen HP, Seto NOL, Cai Y, Leinala EK, Borisova SN, et al. 2003. J. Biol. Chem. 278:49191–95 Letts JA, Rose NL, Fang YR, Barry CH, Borisova SN, et al. 2006. J. Biol. Chem. 281:3625– 32 Persson M, Letts JA, Hosseini-Maaf B, Borisova SN, Palcic MM, et al. 2007. J. Biol. Chem. 282:9564–70 Hosseini-Maaf B, Letts JA, Persson M, Smart E, LePennec PY, et al. 2007. Transfusion 47:864–75 Ten Hagen KG, Tetaert D, Hagen FK, Richet C, Beres TM, et al. 1999. J. Biol. Chem. 274:27867–74 Ten Hagen KG, Bedi GS, Tetaert D, Kingsley PD, Hagen FK, et al. 2001. J. Biol. Chem. 276:17395–404 www.annualreviews.org • Glycosyltransferases

25.33

ARI

29 February 2008

19:15

105. Fritz TA, Hurley JH, Trinh LB, Shiloach J, Tabak LA. 2004. Proc. Natl. Acad. Sci. USA 101:15307–12 106. Kubota T, Shiba T, Sugioka S, Furukawa S, Sawaki H, et al. 2006. J. Mol. Biol. 359:708–27 107. Fritz TA, Raman J, Tabak LA. 2006. J. Biol. Chem. 281:8613–19 108. Just I, Selzer J, Wilm M, Voneichelstreiber C, Mann M, Aktories K. 1995. Nature 375:500– 3 109. Reinert DJ, Jank T, Aktories K, Schulz GE. 2005. J. Mol. Biol. 351:973–81 110. Lobsanov YD, Romero PA, Sleno B, Yu BM, Yip P, et al. 2004. J. Biol. Chem. 279:17921– 31 111. Pedersen LC, Dong J, Taniguchi F, Kitagawa H, Krahn JM, et al. 2003. J. Biol. Chem. 278:14420–28 112. Green AA, Cori GT, Cori CF. 1942. J. Biol. Chem. 142:447–48 113. Fisher EH, Graves DJ, Snyder Crittended ER, Krebs EG. 1959. J. Biol. Chem. 1959:1698– 704 114. Fletterick RJ, Sprang SR. 1982. Acc. Chem. Res. 15:361–69 115. Raibaud O, Schwartz M. 1984. Annu. Rev. Genet. 18:173–206 116. Watson KA, Mitchell EP, Johnson LN, Son JC, Bichard CJF, et al. 1994. Biochemistry 33:5745–58 117. Mitchell EP, Withers SG, Ermert P, Vasella AT, Garman EF, et al. 1996. Biochemistry 35:7341–55 118. O’Reilly M, Watson KA, Schinzel R, Palm D, Johnson LN. 1997. Nat. Struct. Biol. 4:405– 12 119. O’Reilly M, Watson KA, Johnson LN. 1999. Biochemistry 38:5337–45 120. Geremia S, Campagnolo M, Schinzel R, Johnson LN. 2002. J. Mol. Biol. 322:413–23 121. Watson KA, McCleverty C, Geremia S, Cottaz S, Driguez H, Johnson LN. 1999. EMBO J. 18:4619–32 122. Lin K, Rath VL, Dai SC, Fletterick RJ, Hwang PK. 1996. Science 273:1539–41 123. Lukacs CM, Oikonomakos NG, Crowther RL, Hong LN, Kammlott RU, et al. 2006. Proteins 63:1123–26 124. Rath VL, Ammirati M, LeMotte PK, Fennell KF, Mansour MN, et al. 2000. Mol. Cell 6:139–48 125. Ekstrom JL, Pauly TA, Carty MD, Soeller WC, Culp J, et al. 2002. Chem. Biol. 9:915–24 126. Wright SW, Rath VL, Genereux PE, Hageman DL, Levy CB, et al. 2005. Bioorg. Med. Chem. Lett. 15:459–65 127. Klabunde T, Wendt KU, Kadereit D, Brachvogel V, Burger HJ, et al. 2005. J. Med. Chem. 48:6178–93 128. Withers SG, Shechosky S, Madsen NB. 1982. Biochem. Biophys. Res. Commun. 108:322–28 129. Stirtan WG, Withers SG. 1996. Biochemistry 35:15057–64 130. Withers SG, Madsen NB, Sykes BD, Takagi M, Shimomura S, Fukui T. 1981. J. Biol. Chem. 256:10759–62 131. Withers SG, Wakarchuk WW, Strynadka NCJ. 2002. Chem. Biol. 9:1270–73 132. Martinez-Fleites C, Proctor M, Roberts S, Bolam DN, Gilbert HJ, Davies GJ. 2006. Chem. Biol. 13:1143–52 133. Gibson RP, Tarling CA, Roberts S, Withers SG, Davies GJ. 2004. J. Biol. Chem. 279:1950– 55 134. Lariviere L, Sommer N, Morera S. 2005. J. Mol. Biol. 352:139–50 135. Buschiazzo A, Ugalde JE, Guerin ME, Shepard W, Ugalde RA, Alzari PM. 2004. EMBO J. 23:3196–205

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

25.34

Lairson et al.

Annu. Rev. Biochem. 2008.77. Downloaded from arjournals.annualreviews.org by UNIWERSYTET WROCLAWSKI on 04/18/08. For personal use only.

ANRV345-BI77-25

ARI

29 February 2008

19:15

136. Horcajada C, Guinovart JJ, Fita I, Ferrer JC. 2006. J. Biol. Chem. 281:2923–31 137. Cowdrey WA, Hughes ED, Ingold CK, Masterman S, Scott AD. 1937. J. Chem. Soc. 1253–71 138. Hughes ED, Ingold CK, Whitfield IC. 1941. Nature 147:206–7 139. Lewis ES, Boozer CE. 1952. J. Am. Chem. Soc. 74:308–11 140. Sinnott ML, Jencks WP. 1980. J. Am. Chem. Soc. 102:2026–32 141. Klein HW, Im MJ, Palm D. 1986. Eur. J. Biochem. 157:107–14 142. Gibson RP, Turkenburg JP, Charnock SJ, Lloyd R, Davies GJ. 2002. Chem. Biol. 9:1337– 46 143. Sommer N, Depping R, Piotrowski M, Ruger W. 2004. Biochem. Biophys. Res. Commun. 323:809–15 144. Sinnott ML. 1990. Chem. Rev. 90:1171–202 145. Schreiner PR, Schleyer PvR, Hill RK. 1994. J. Org. Chem. 59:1849–54 146. Moss RA, Fu XL, Sauers RR. 2007. J. Phys. Org. Chem. 20:1–10 147. Christianson DW. 2006. Chem. Rev. 106:3412–42 148. Muehlbacher M, Poulter CD. 1985. J. Am. Chem. Soc. 107:8307–8 149. Muehlbacher M, Poulter CD. 1988. Biochemistry 27:7315–28 150. Reardon JE, Abeles RH. 1986. Biochemistry 25:5609–16 151. Durbecq V, Sainz G, Oudjama Y, Clantin B, Bompard-Gilles C, et al. 2001. EMBO J. 20:1530–37 152. Starks CM, Back KW, Chappell J, Noel JP. 1997. Science 277:1815–20 153. Shishova EY, Di Costanzo L, Cane DE, Christianson DW. 2007. Biochemistry 46:1941–51 154. Guthrie RD, Jencks WP. 1989. Acc. Chem. Res. 22:343–49 155. Jencks WP. 1996. J. Phys. Org. Chem. 9:337–40 156. Winstein S, Clippinger E, Fainberg AH, Heck R, Robinson GC. 1956. J. Am. Chem. Soc. 78:328–35 157. Jencks WP. 1980. Acc. Chem. Res. 13:161–69 158. Carey FA, Sundberg RJ. 1990. Advanced Organic Chemistry. Part A: Structure and Mechanisms. New York: Plenum 159. Comm. Phys. Org. Chem. I. 1989. Pure Appl. Chem. 61:23–56

RELATED RESOURCES Univ. Aix-Marseille I & II. 2007. CAZy database. http://www.cazy.org Consortium for Functional Glycomics/Nature. 2007. http://www.functionalglycomics.org/

www.annualreviews.org • Glycosyltransferases

25.35