Transglucosylases of the GH70 family: structure/function relationships and understanding of their catalytic mechanism
Glucansucrases (GSs) are large (140 to 313 kDa) bacterial-transglucosylases, also named Glucosyltransferases (GTFs), which are classified in the family 70 of the Glycoside-Hydrolases (GH 70) in the Cazy database (Lombard et al., 2014), a family in which three subgroups of-transglucosylases can be found:
– the glucansucrases, which synthesize polymers from sucrose,
– the branching sucrases, which catalyze branch formation onto linear dextran in the presence of dextran acceptor and sucrose
– the glucanotransferases, inactive on sucrose but able to disproportionate maltodextrins to form new types of carbohydrate polymers
The Family GH70 constitutes with the families GH13 and GH77 the GH-H clan, also called-amylase super family that comprises mechanistically, structurally and evolutionary related enzymes. They share the same-retaining mechanism involving an Asp and a Glu residues – acting, respectively, as a nucleophile and acid/base catalyst – and found in a catalytic ()8 barrel domain (Davies and Henrissat, 1995; Henrissat, 1991). While the number of novel putative GH70 enzymes increases rapidly thanks to sequencing progress, the percentage of fully biochemically characterized enzymes remains low. Indeed, among the 493 sequences indexed, only 66 enzymes were biochemically characterized and no more than five structures were solved so far (http://www.cazy.org/GH70_characterized.html, Sept 2017).
The GSs and branching sucrases of GH70 family are usually extracellular enzymes reported to date to be exclusively produced by lactic acid bacteria (LAB) from Leuconostoc, Streptococcus, Weissella, Lactobacillus, Pediococcus or Oenococcus genera (Leemhuis et al., 2013a; Meng et al., 2016a; Monsan et al., 2010; Moulis et al., 2016; Shukla and Goyal, 2013; Vuillemin et al., 2017). Genomic sequence analyses of GH70 producing LAB revealed that they generally contain several GH70 encoding genes. For example, the inventory of the GH70 enzymes encoded by Leuconostoc citreum NRRL B-1299 allowed the identification of six distinct GSs, namely DSRA, DSRB, DSR-E, DSR-DP, DSR-M and BRS-A which are all characterized (Passerini et al., 2015). The various gs genes found in a strain usually encode enzymes showing different product specificities. However, extensive physiological characterization of their role has not been undertaken at this time. Thus, whether they are all produced and how is their expression regulated remain unclear. In contrast to GSs, the- 4,6/-4,3 glucanotransferases are more widespread in nature and can be found in a wider range of LAB bacteria from Bacillus, Fructobacillus (fructophilic LAB) genera as well as in non LAB genera such as Geobacillus Exigobacterium and Azotobacter (gammaproteobacteria, gram negative)( http://www.cazy.org/GH70_all.html).
Although, this literature review will mainly focus on the sucrose-active polymerases, some details on the two other subgroups will also be presented. (cf II.3. Particularity of the branching-sucrases and -4,6/-4,3 –glucanotransferases, p46).
Reactions catalyzed by GH70 glucansucrases
GH70 glucansucrases represent the major part of characterized GH70 enzymes active on sucrose. They usually catalyze the synthesis of long-glucan chains (>106 g/mol) containing all the possible glucosidic linkages, namely-(1→2),-(1→3),-(1→4) or-(1→6). They are named according to the type of polymer they synthesize to which the suffix “ase” is added (cf I. α-glucans produced from sucrose by lactic acid bacteria). Among the variety of glucansucrases, dextransucrases represent more than 65% of the characterized GH70 to date (CAZy Database, Sept 2017).
To catalyse glucan synthesis, glucansucrases first cleave the sucrose osidic linkage through the formation of a-D-glucosyl enzyme intermediate with a concomitant release of fructose (Figure 5). The D-glucosyl unit can be then transferred onto various acceptor molecules with retention of the-anomeric configuration (Koshland, 1953; Uitdehaag, 1999). Depending on the nature of the acceptor, different products are obtained. The main ones result from the iterative transfer of the glucosyl unit on growing-glucan chains, leading to oligosaccharide or-glucan synthesis. In parallel, transfer reaction can also occur onto water leading to glucose, or fructose resulting in the formation of sucrose isomers (turanose, trehalulose, isomaltulose, and mainly leucrose). In addition, if other hydroxylated molecules are introduced in the reaction medium, they can possibly be glucosylated through the “acceptor reaction” at the expense of the polymer synthesis, the efficiency of the acceptor reaction being directly linked to the acceptor recognition (Koepsell et al., 1953; Mayer, 1987). This reaction also makes-transglucosylases very attractive biocatalysts for the production of glucoderivatives through the glucosylation of non-natural acceptors such as various sugars (maltose, cellobiose, gentiobiose, lactose, thiooligosaccharides, …), flavonoids, salicin or, with high industrial application potential (Argüello Morales et al., 2001; Bertrand et al., 2006; Hellmuth et al., 2007; Malbert et al., 2014; Meulenbeld et al., 1999; Monsan et al., 2010; Nam et al., 2007).
On sucrose only, the polymerization reaction is predominant, but transferase efficiency can vary from an enzyme to another. For example, Leuconostoc GSs are excellent polymerases for which the transferase activity is above 85% (Moulis et al., 2006a). In comparison, the hydrolytic activity of GTF-O reuteransucrase from Lactobacillus reuteri ATCC 55730 corresponds to more than 50% of the transferred glucosyl units (reuteran synthesis represents only 36%) whereas GTF-A, another reuterasucrase from Lactobacillus reuteri, converts 73% of the glucosyl unit from sucrose into polymer (Kralj et al., 2005a; Meng et al., 2016b).
-transglucosylases also have the ability to catalyze disproportionation reactions (Binder et al., 1983) which consist of the following reaction: 2(glucose)n (glucose)n+1 + (glucose)n-1. A novel GH70 enzyme category catalyzing only reactions close to the disproportionation mechanism was recently identified and named-4,6-/-4,3-glucanotransferase. These enzymes, non-active on sucrose but on maltooligosaccharides constitute a GH70 family subgroup as well as the branching sucrase and will be presented in the following paragraph.
Particularity of the branching-sucrases and-4,6/-4,3 –glucanotransferases
Although glucansucrases are the main representative enzymes of the GH70 family, two new enzyme subgroups were recently identified in this family, namely the branching sucrases and the glucanotransferases. Members of these subgroups present unique catalytic properties compared to glucansucrases.
Brief overview of the branching-sucrases
For a long time, GBD-CD2 was the sole known branching-sucrase of the GH70 family. This is however a non-natural enzyme obtained after domain truncation of the peculiar DSR-E from L. citreum NRRL B-1299, the unique bi-functional GS characterized to date which displays two catalytic domains separated by a glucan-binding domain (GBD) (Figure 6). From sucrose, recombinant DSR-E synthesizes a polymer of-(1→6) linked glucopyranosyl units with-(1→2) branches (Bozonnet et al., 2002; Fabre et al., 2005).
Figure 6: Schematic structure of DSR-E from L. citreum B1299 and GBD-CD2 truncated variant. CD, Catalytic Domain; GBD, Glucan Binding Domain.
By generating and characterizing two truncated forms, namely CD1-GBD and GBD-CD2, the first catalytic domain was shown to be responsible for the synthesis of-(1→6) linked-glucan while GBD-CD2 is only able to catalyze-(1→2) glucosyl transfer onto linear dextran. From sucrose only, GBD-CD2 acts as a sucrose hydrolase and is incapable of polymerizing the glucosyl units. Indeed, 78% of the available glucosyl units are transferred onto water and 16% onto fructose to produce leucrose (-D-Glcp-(1→5)-D-fructopyranose), kojibiose (-D-Glcp-(1→2)-D-Glcp), and traces of maltulose (-D-Glcp-(1→4)-D-fructofuranose). Nevertheless, in the presence of an exogenous dextran acceptor, this enzyme almost exclusively catalyzes polymer glucosylation through the formation of-(1→2) linkages (Figure 7). Moreover, it has been shown that the percentage of-(1→2)-linked glucose can be controlled from 10 to 40% by playing on the initial [sucrose]/[dextran] ratio, thus conferring a valuable potential to this original catalyst for the production of a wide diversity of branched-glucans (Brison et al., 2010).
Figure 7: Dextran glucosylation by the branching-sucrase GBD-CD2. Oxygen atoms of-(1→6) glucosidic bonds are in red while-(1→2) ones are in blue.
Recently, natural branching-sucrase activities were identified in Leuconostoc and Lactobacillus strains highlighting their more spread occurrence and diversity. Contrary to DSR-E, they display a “classic” primary structure similar to that of polymerases with only one catalytic domain and a GBD (see paragraphII.4.1. Global structure and domain organization of GH70 enzymes, p49) but they were proposed to constitute a new subgroup of the GH70 family as they used two substrates. This group houses now five characterized enzymes, namely, the above described GBD-CD2, BRS-A (Passerini et al., 2015), BRS-B, BRS-C, and BRS-D (Vuillemin et al., 2016). As DSR-E or the truncated version GBD-CD2, BRS-A is originated from L. citreum NRRL B-1299 and is responsible for the high content of a-(1→2) linkages (28%) observed for the dextran produced by this strain (Passerini et al., 2015). BRS-B encoding gene was isolated from the L. citreum NRRL B-742 genome and the recombinant protein was shown to display an-(1→3) branching specificity. As for GBD-CD2 or BRS-A, playing on the [sucrose]/[dextran] initial ratio allows a tight control of the polymer branching degree up to 50%, meaning that each glucose of the linear dextran is glucosylated. For that reason, BRS-B was proposed to be responsible of the comb-like hyper-branched dextran of L. citreum NRRL B-742 synthesis. BRS-C from Leuconostoc fallax KCTC 3537 and BRS-D from Lactobacillus kunkeei EFB6 catalyze α-(1→3) and α-(1→2) transglucosylation, respectively. Sequence analyses revealed that these branching-sucrases share common motifs with GH70 GSs but also display distinctive features. This is particularly the case in the +1 binding subsite, where Phe675 of motif II and Ile783, His785, Lys789, and Val795 of motif IV (BRS-B numbering) are only conserved in the branching sucrases (for subsites and motifs description, see paragraph: II.4.2.1. Structure of the catalytic core, p51). These particularities can be used to rapidly sort out branching sucrases from genomic data (Vuillemin et al., 2016).
The determinants responsible for their specificity to branch dextrans and incapacity to catalyze polymer formation have not clearly been identified yet, although the structure of GBD-CD2 has been solved. To this aim, further structure-function studies are necessary. However, as they are more specific for acceptor reaction and are able to catalyze a variety of glucosidic linkage type, branching-sucrases appear to be promising for new functional glucooligosaccharide or glucoconjugate synthesis.
The first GS-like enzyme of this subgroup was found in L. reuteri 121 and named GtfB. Being non active on sucrose but only on maltodextrins and starch substrates, GtfB catalyzes-(1→4)-linkage cleavage and the formation of new-(1→6)-linkages (see Figure 15, p58). This α-4,6-glucanotransferase activity results in the production of-glucans with linear chains of-(1→6) and-(1→4)-linkages (Kralj et al., 2011). Lately, two other GtfB homologues from Lactobacillus reuteri strains DSM 20016 (GtfW) and ML1 (GtfML4) displaying the same donor/acceptor specificity were also characterized (Leemhuis et al., 2013b). α-4,6-glucanotransferase activity can also be predicted by sequence analysis thanks to signature residues in acceptor binding subsites +1 and +2 differing from those found in GH70 GSs (Leemhuis et al., 2013b). Beside these three biochemically characterized enzymes, 46 putative GtfB-like enzymes are currently found in the GenBank database, and they are almost all from Lactobacillus genus (three are found in Pediococcus strains). The first α-4,6-glucanotransferase structure, that of GtfB-NV from Lactobacillus reuteri 121 (PDB: 5JBD), was solved last year (Bai et al., 2017). Thanks to structural and phylogenetic analyses, notably focusing on three loops shaping the active site, the enzymes of this subfamily are suggested to be evolutionary intermediates between GH13 amylosucrases and GH70 sucrose-active enzymes (see Figure 11, p54).
GtfC of Exiguobacterium sibiricum 255–15 (Gangoiti et al., 2016a) and GtfD of Azotobacter chroococcum NCIMB 8003 (Gangoiti et al., 2016b) are also both inactive on sucrose but display α-4,6-glucanotransferase activity that resulted in the synthesis of isomalto/malto oligosaccharides (IMMO) and of a reuteran-like α-glucan, respectively. The domain organization of GtfC and GtfD resembles that of GH13 family enzymes with a conserved order of motifs I-II-III and IV (contrary to GH70s) and an absence of domain V (see paragraph: II.4.2.1. Structure of the catalytic core, p51, for description of these motifs). Thus, GtfC and D are also proposed to be representative of two other GH70 subfamilies.
Most recently, genome sequencing of Lactobacillus fermentum NCC 2970 resulted in the discovery of a novel GtfB-like enzyme presenting divergence in the conserved motifs II and IV located in donor/acceptor binding subsites. This enzyme was further shown to possess an α-4,3-glucanotransferase activity meaning that it cleaves-(1→4)-linkages and synthesizes new-(1→3)-linkages (Gangoiti et al., 2017). All these recent findings – due to progress of sequencing- highlight the diversity existing in GH70 family in terms of substrate and linkage specificity.
Structures of GH70 family enzymes and understanding of their mechanism
Global structure and domain organization of GH70 enzymes
Until the first 3D-structure characterization, the primary structures of GH70 enzymes were traditionally represented in three different domains linearly organized along the polypeptidic chain and preceded by a signal peptide (Figure 8). The sequence was proposed to be composed of i) a poorly conserved N-terminal variable region, ii) a catalytic domain of around a thousand of very conserved residues and iii) a C-terminal domain comprising repeated units proposed to be likely involved in glucan binding (see paragraph, II.4.3.1. Structural organization of domain V, p58).
Figure 8: General primary structure of GS from LAB as depicted before the first X-ray 3D structure resolution of a GH70 glucansucrase. SP signal peptide, VR variable region, CD catalytic domain, GBD glucan binding domain.
Since 2010, attempts to solve the tridimensional structure of GH70 enzymes remained unsuccessful probably due to the large molar mass of these enzymes added to an important flexibility rendering protein purification and crystallization very challenging. When the first 3D-structure was solved, the previously proposed linear organization had to be corrected. Indeed, the structure of GTF180-N from L. reuteri (PDB: 3KLK) revealed an atypical U-shape fold resulting in an organization in five domains, namely the domains A, B, C, IV and V (Vujicic-Zagar et al., 2010a). Domain C is the sole domain composed of a continuous polypeptide fragment – forming an eight-stranded-sheet with a Greek key motif – and constitutes the base of the “U” but its precise role remains unknown. Domains A, B, IV and V consist in a tangle of both N- and C-terminal segments (Figure 9). While domains IV and V are unique to GH70 enzymes, domains A, B and C are common with GH13 family members. The active site is localized in a cavity of domain A at the interface with domain B (Figure 9). Domain A adopts a (/)8 barrel fold circularly permuted compared to that of the GH13 enzymes (cf. next paragraph). Domain B is folded in five-stranded-sheets and some of its loops participate in the catalytic cleft shaping. A calcium binding site is found at its interface with domain A at around 10 Å of the nucleophilic aspartate. Domain IV connects domains B and V and displays a particular fold showing no similarity to any other characterized protein structures. As the connection between domains IV and V is made of two relatively long unstructured polypeptide chains, Ito et al. (Ito et al., 2011) proposed that domain IV may serve as a “hinge” allowing domain V movement near or away from the catalytic core. This was further supported by domain V flexibility analyses (Pijning et al., 2014)(see paragraph, II.4.3.1. Structural organization of domain V, p58). The secondary structure elements organization of domain V is very similar from an enzyme to another and their role in glucan binding was recently structurally confirmed (Brison et al., 2016) (cf section “II.4.3. Focus on domain V”, p58 for detailed description).
Since that time, four other 3D-structures of recombinant truncated GH70 enzymes sharing the same features were solved: the reuteransucrase GTFA-ΔN from L. reuteri 121 (Pijning et al., 2012), the mutansucrase GTF-SI from Streptococcus mutans (Ito et al., 2011), the branching sucrase ΔN123-GBD-CD2 engineered from L. citreum NRRLB-1299 DSR-E (Brison et al., 2012) and most recently, the -4,6-glucanotransferase, Gtf-B-NV from L. reuteri 121 (Bai et al., 2017).
Catalytic core structure and reactional mechanism
Structure of the catalytic core
The catalytic core comprising domains A, B and C of GH70 members is very similar to that of GH13 family. However, except for GtfC and GtfD (see p48), the GH70 (/)8 barrel fold of domain A is generally circularly permuted compared to that of GH13 family (MacGregor et al., 1996; Mooser and Wong, 1988). This (/)8 barrel consists of eight parallel β-strands (1-8) forming the inner β-barrel and alternating with eight α-helices (1-8) surrounding the-strands. Due to the circular permutation, from N- to C-terminus, GH70 ()8 barrel starts with helix3 of family GH13 followed by Nter-4-4-5-5-6-6-7-7-8-8-1-1-2-2-3-Cter (Figure 10).
Figure 10: Topology diagrams models of family GH70 GSs with a circularly permutated (β/α)8 barrel (a) and the family GH13 α-amylase (β/α)8 barrel (b). Cylinders represent α-helices and arrows represent β-strands. The equivalent α-helices and β-strands in GH70 GSs and GH13 α-amylases are numbered the same. The different domains in GH70 and GH13 enzymes are indicated. The four conserved sequence motifs (I–IV) which are located in β-strands 3, 4, 5, and 7, respectively, are indicated within the β-strand. The structure of the catalytic domain in the GH70 GSs representative GTF180-ΔN (c, PDB: 3KLK) of L. reuteri 180 and in the GH13 representative α-amylase of Bacillus licheniformis (d, PDB: 1BPL). From (Meng et al., 2016a).
Accordingly the conserved motifs characteristic of GH13 enzymes are also permuted in GH70 proteins, with motif I located at C-terminal to motifs II, III and IV. All are located in domain A. Residues of the catalytic triad, namely, the nucleophile Asp1025, the acid/base catalyst Glu2063 and the transition state stabilizer Asp1136 (GTF-180 numbering) are located in the conserved motifs II, III and IV, respectively (Figure 9 and Figure 10). Thanks to previous works on structure-function relationships, GH70 specificity can be predicted by analysis of these “signature” motifs (Table 3). Indeed, it has been shown that amino acids upstream and downstream the catalytic residues are involved in the linkage type specificity (see paragraph, II.4.4.2. Linkage specificity, p67).
Table of contents :
CHAPTER I: GH70 -TRANSGLUCOSYLASES: ATTRACTIVE TOOLS FOR THE PRODUCTION OF BIO-SOURCED TAILORED POLYSACCHARIDES, A LITERATURE REVIEW
I. α-glucans produced from sucrose by lactic acid bacteria
I.1. Structural diversity
I.2. Dextran applications
I.2.1. Pharmaceutical and medical sectors
I.2.2. Prebiotics sector
I.2.2.1. Definition and market
I.2.2.2. Dextrans and isomaltooligosaccharides as prebiotics
I.2.3. Agri-food sector
I.2.4. Analytical chemistry
I.2.5. Other applications
I.3. Dextran production processes from sucrose
II. -transglucosylases of the GH70 family: structure/function relationships and understanding of their catalytic mechanism
II.2. Reactions catalyzed by GH70 glucansucrases
II.3. Particularity of the branching-sucrases and -4,6/-4,3 –glucanotransferases
II.3.1. Brief overview of the branching-sucrases
II.3.2. -4,6- and -4,3-glucanotransferases
II.4. Structures of GH70 family enzymes and understanding of their mechanism
II.4.1. Global structure and domain organization of GH70 enzymes
II.4.2. Catalytic core structure and reactional mechanism
II.4.2.1. Structure of the catalytic core
II.4.2.2. Catalytic mechanism
II.4.3. Focus on domain V
II.4.3.1. Structural organization of domain V
II.4.3.2. Functional role of domain V
II.4.4. Polymerization mode: processive / non-processive and product specificity determinants
II.4.4.1. Mode of polymer elongation by glucansucrases
II.4.4.2. Linkage specificity and polymer size
III. DSR-M and DSR-OK, the two templates of this study
CHAPTER II: INVESTIGATIONS ON THE DETERMINANTS RESPONSIBLE FOR LOW MOLAR MASS DEXTRAN FORMATION BY DSR-M DEXTRANSUCRASE
II.1. Design and characterization of truncated variants DSR-M1 and DSR-M2
II.2. Kinetics of polymer formation reveal that DSR-M 2 can accept different chain initiators
II.3. 3D structures of DSR-M2 and DSR-M2 E715Q in complex with sucrose or isomaltotetraose
II.4. Functional implications of domain V
V. Material and methods
V.1. Construction of DSR-M1, DSR-M2 and DSR-MV deletion mutants
V.2. Protein expression and purification
V.3. Activity assays
V.4. Enzymatic reaction and product characterization
V.5. Crystallization and Data collection
V.6. Structure determination
V.7. SAXS measurements and processing
V.8. Mutagenesis studies
VII. Supplementary information
CHAPTER III : FUTILE CYCLE ENGINEERING OF THE DSR-M DEXTRANSUCRASE MODIFIES THE RESULTING POLYMER LENGTH
II. Results and discussion
II.1. The crystal structure of DSR-MV in complex with an isomaltotetraose defines novel anchoring points
II.2. The W624A mutation changes the reaction rate and final length distribution.
II.3. Beyond a simple stacking platform: the W624A mutation equally influences the dynamics of the catalytic site.
II.4. Monte Carlo model of the chain elongation.
IV. Complementary work (not part of the publication)
IV.I. Further NMR analysis
IV.2. The DSR-M W624A mutant produces short oligosaccharides in good yield
IV. Complementary conclusion
V. Material and methods
V.1. Protein expression and purification
V.2. Activity assays
V.3. Enzymatic reaction
V.4. Kinetic analysis of acceptor reactions
V.5. Acceptor reactions on glucose with DSR-M W624A mutants
V.6. Product characterization
V.7 Crystallization and Data collection
V.8. Structure determination
V.9. Mutagenesis studies
V.10. 15N protein expression
V.11. 15NTrp protein expression
V.12. NMR analysis of DSR-M variants
V.13. Monte Carlo simulation.
VI. Supplementary information
VI.1. Protein purification procedures
VI.2. Recombinant DSR-MDV protein sequence
VI.3. Electron density map and structural data statistics
VI.4. Analysis of reaction medium after 15min reaction before and after invertase digestion.
VI.5. Monte Carlo model of the chain elongation.
VII. NMR analyses
CHAPTER IV: HIGH MOLAR MASS DEXTRAN SYNTHESIS BY DSR-OK DEXTRANSUCRASE FROM OENOCOCCUS KITAHARAE DSM 17330, ROLE OF THE DOMAIN V?
II. Results & Discussion
II.1. Design and characterization of DSR-OK 1, a model of study
II.2. Monitoring of dextran synthesis by DSR-OK1
II.3. DSR-OK 1 structural insights
II.4. Functional implications of domain V
II.5. Effect of sugar binding pocket deletions
II.6. Effect of mutations targeting the aromatic residues of the sugar binding pockets
II.7. Construction of chimeric enzymes with domain V swapping
IV. Material and Methods
IV.1. Construction of DSR-OK1
IV.2. Protein expression and purification
IV.3. Activity assays
IV.4. Enzymatic reaction
IV.5. Product characterization
IV.6. SAXS measurements and processing
IV.7. Building the DSR-OK core models
IV.8. Circular dichroism analyses
IV.9. Chimera construction
IV.10. Mutagenesis study
IV.10.1. Construction of deletion mutants
IV.10.2. Site-directed mutagenesis
V. Supplementary information
CONCLUSION & PROSPECTS
DSR-M: short-chain polymerase from Leuconostoc citreum NRRL B-1299
DSR-OK: very long-chain polymerase from Oenococcus kitaharae DSM 17330
Discussion and perspectives
ARTWORKS & TABLE CONTENTS
Supplementary information content