Hooked on α-d-galactosidases: from biomedicine to enzymatic synthesis.

Abstract α-d-Galactosidases (EC 3.2.1.22) are enzymes employed in a number of useful bio-based applications. We have depicted a comprehensive general survey of α-d-galactosidases from different origin with special emphasis on marine example(s). The structures of natural α-galactosyl containing compounds are described. In addition to 3D structures and mechanisms of action of α-d-galactosidases, different sources, natural function and genetic regulation are also covered. Finally, hydrolytic and synthetic exploitations as free or immobilized biocatalysts are reviewed. Interest in the synthetic aspects during the next years is anticipated for access to important small molecules by green technology with an emphasis on alternative selectivity of this class of enzymes from different sources.

Introduction a-D-Galactosidases (EC 3.2.1.22) free galactose from different galactosides in nature; melibiose, raffinose and similar oligosaccharides or aryl and alkyl a-galactosides and polymers containing a-galactose (e.g. galactomannan), are all substrates. These enzymes are interesting activities from different points of view, from biomedicine to biocatalyzed synthesis.
Searching the database of the United States Patent and Trademark Office for patents containing the word a-D-galactosidase in the title, relatively few ''hits'' resulted starting from the early 1970s when interest in enzymatic manipulation of raffinose-containing food began (Kawamura et al., 1971) US Patent 1971. Coeval is the interest for immobilization (Linden, 1982), however over time many a-D-galactosidases were identified from different sources starting with brewer's yeast at the end of 1800s. In 1990, a dietary supplement called Beano was developed based on the research regarding gas-causing vegetables; the product containing the enzyme from Aspergillus niger (Di Stefano et al., 2007) was used to introduce students to enzymes (Hardee et al., 2000). It could also be appreciated as a ready-made biocatalyst for synthesis. Interest for these enzymes continues currently for the reduction of anti-nutritional factors in animal feeds (Opazo et al., 2012).
A search in Pubmed at the end of 2013 for the word a-D-galactosidase accounted for ca. 300 review articles. Most of them, if not all, are related to the importance of a-D-galactosidases in enzyme replacement therapy in Anderson-Fabry's disease (deficiency of a-D-galactosidase A) (Pisani et al., 2012). In the biomedicine field, the use of a-D-galactosidases in the conversion of red blood cells of group B (removing of a-1,3-linked galactose), is known (Olsson et al., 2004) and for this application a marine enzyme has been interestingly proposed . a-D-Galactosidases also generated some interest in the selective hydrolysis of glycosphingolipids (Li & Li, 1999) in conjunction with NMR spectroscopy and mass spectrometry for the structural elucidation of complex lipid biomolecules.
a-D-Galactosyl linkage is present in so-called a-Gal epitope (Gal-a-1,3-Gal-b-1,(3)4-GlcNAc-R), carbohydrate structures that are present on glycolipids and glycoproteins. The natural absence in humans of this epitope and the fact that a natural antibody to it, the anti-Gal antibody, is produced induced interest for the synthesis of these carbohydrate linkages present in drugs used in xenotransplantation and in immunotherapy (Galili, 2005).
Noteworthy is the recent involvement of a-D-galactosidases in analytical techniques for forensic and highthroughput applications like DBS (dried blood spots). The ''lab-on-a-chip'' approach, with very small volumes used (nano-to pico-liter samples), minimized cost and material consumption and holds promise as the next generation technology. Multiplexed enzymatic assays of acid a-glucosidase (GAA) and acid a-D-galactosidase, to screen for Pompe and Fabry disorders, have been recently developed (Demirev, 2013).
The biosynthetic aspect, genetic regulation, biological functions and structural classification of a-D-galactosidases are all topics of current interest. Galactose containing oligo-and poly-saccharides are natural substrates and generally, a-D-galactosidases are subject to induction by galactose and its derivatives (Pardee, 1957). Based on the understanding of the expression regulatory mechanisms and function of the a-D-galactosidase, it may be possible to obtain more efficient and novel enzymes with new functionalities and applicability (Ademark et al., 2001). The usual transferring capability common to many glycosyl hydrolases (Trincone & Giordano, 2006) was evidenced for a-D-galactosidases very early (Anagnostopoulos et al., 1954;Courtois & Petek, 1957).
We have depicted a general survey on a-D-galactosidases from different origin with a special emphasis on marine example(s). Common structures and functions of natural a-galactosyl containing compounds are reported; different sources of a-D-galactosidases, genetic regulation and natural functions are covered first with subsequent emphasis on mechanisms of action, before discussing hydrolytic and synthetic exploitation as free or immobilized biocatalysts. In-depth aspects are covered in Supplementary Material.
The interest of synthetic chemists for the production of such structures was already active in 1970s and it has been traced in the literature as being generally prompted by discovery of biological functions of a-galactosides. The production of these compounds using conventional techniques or innovative bio-based pathways is still currently supported.
The crystal structures of free forms and their complexes with ligands (galactose and N-acetylgalactosamine) of eight GH-27 members have been solved by the X-ray analysis (Clark & Garman, 2009;Garman et al., 2002) as detailed in Table 1 and Supplementary Material Part 4. All known subfamilies from I to V of GH-27 family are covered, improving knowledge on the activity of this important family of enzymes and giving deeper insight into the structural features that rule modularity and protein-carbohydrate interactions (Fernandez-Leiro et al., 2010). Less structural information is available for the GH-36 and GH-97 a-galactosidases. Crystal structures have been reported for six GH-36 members (Bruel et al., 2011;Comfort et al., 2007;Fredslund et al., 2011;Merceron et al., 2012) without full information about the evolution of 3D structure within this family. Five from these are homotetrameric enzymes of symbiotic and probiotic bacteria from the human gut belonging to subfamily I. These are a-galactosidases AgaA and AgaB from Geobacillus stearothermophilus strain KVE39 (PDB 4fnp and 4fnq, respectively) (Merceron et al., 2012), Ruminococcus gnavus E1 (PDB 2yfo) (Bruel et al., 2011), Lactobacillus acidophilus strain NCFM (PDB 2xn2) (Fredslund et al., 2011), Lactobacillus brevis strain ARQQ 367 (PDB 3mi6) (Fredslund et al., 2011) and one is from thermopile bacterium Thermotoga maritima of subfamily IV (PDB 1zy9) (Comfort et al., 2007) (Table 1, Supplementary Material Part 5). Only one 3D structure of GH-97 a-galactosidases of Bacteroides thetaiotaomicron BtGH97b (PDB 3a24) is known (Okuyama et al., 2009) (Table 1, Supplementary Material Part 6).
In general, tertiary structures of GH-27 enzymes consist of two domains whereas with GH-36 and GH-97 a-galactosidases, three domains are found. The large N-terminal b-supersandwich domain found in GH-36 and GH-97 proteins, is absent in the enzymes of the GH-27 family. It is formed by 16-20 antiparallel b-strands and some a-helices in the GH-36 member. However, b-supersandwich of GH-97 a-galactosidase is disordered and consists of the 21 antiparallel b-strands and 2 short 3 10 -helices. The central classical (b/a) 8 -barrel domain in GH-36 a-galactosidases has high homology with GH-27 catalytic N-terminal domain. It was observed earlier in a triose phosphate isomerase (Banner et al., 1975) and now represents a common motif in many glycoside hydrolases of Clan-D, functioning as catalytic domain. It is deformed in GH-97 a-galactosidase containing several 3 10 -helices in the loops connecting the parallel main b-strands and a-helices, moreover b-strand 7 is exchanged with a loop. Active cleft and the fundamental two catalytic residues Asp (nucleophile) and Asp (acid/base) are at the bottom of the (b/a) 8 -barrel in GH-27 and GH-36 enzymes, while Asp and Glu act as nucleophile and acid/base catalysts, respectively, in GH-97 a-galactosidases. The C-terminal domain is a b-sandwich resembling the structure of the Greek key motif taking place in the GH-27 GH-36 and GH-97 families. GH-27 and GH-36 enzymes do not require calcium ions for action unlike GH-97 enzymes in which a calciumcoordinated water molecule in the active site is involved. From the analysis of crystallographic results, excluding monomeric GH-36 a-D-galactosidase from Thermotoga maritima, the GH-36 enzymes exhibit a tetrameric organization and GH-97 enzyme functions in solution as a monomer (Supplementary Material Part 5 and 6, respectively). The structural comparison of GH-27, -36 and -97 families revealed not only significant similarity but also fine differences, despite the similarity of catalytic mechanisms and allowed the conclusion that they appear to have diverged from a common ancestor (Okuyama et al., 2009;Rigden, 2002).
Since the concept of ''GH family'' allows predicting the protein fold, function and catalytic mechanism of a glycoside hydrolase, the enzymes belonging to the same family generally show conservation of catalytic domain, catalytic residues and mechanism of action. Based on homology with a-amylases, whose crystal structure was established, Glu 117 was shown to be the catalytic nucleophile in the a-D-galactosidase from P. furiosus, whereas the acid/base catalyst remains to be identified. Moreover, it was suggested that the structural fold of the catalytic domain is (b/a) 7 -barrel containing 3 10 -helices far rare variation of the classical (b/a) 8 barrel found in some proteins (Dickmanns et al., 2006).

Examples of marine enzymes
Previous studies have shown that a-D-galactosidases are widespread in marine proteobacteria and Bacteroidetes (Bakunina et al., 2012;Ivanova et al., 1998). The a-D-galactosidase isolated from the marine bacterium Pseudoalteromonas sp. KMM 701 (PsGalA), attributed to the GH-36 family in accordance with its amino acid sequence, was the first to be fully characterized. Pseudoalteromonas sp. KMM 701 was isolated from the cold water of the Sea of Okhotsk. PsGalA is, indeed, an extremely cold active enzyme unlike all known biochemically characterized counterparts. It has also been characterized by halophilic and halotolerant properties but has low resistance towards urea and guanidine chloride. It catalyses the hydrolysis of a-1,3-galactose residues from the non-reducing end of B-trisaccharide and is capable to reduce the serological activity of B-red blood cells at neutral pH . Furthermore, it is able to interrupt the adhesion of pathogens to the human buccal epithelium (Balabanova et al., 2010). This anti-adhesive activity has been attributed to the breaking of a-Gal containing cell-surface structures of both buccal epithelium and/or bacterial cells thus causing a negative effect on bacterial colonization (i.e. biofilm development). These properties show the great therapeutic potential and open up broad prospects for application of this enzyme in medicine.
Sequences of 24 predictive marine bacteria proteins with putative properties of GH-36 a-D-galactosidase were extracted from the database UniProt at the end of September 2013 (Galperin & Fernandez-Suarez, 2012). At present, predicted protein amino acid sequences are reproduced from nucleotide sequences of the genes found in the genomes of marine bacteria. Properties of putative a-D-galactosidases are not known. Results of Blast UniProt PsGalA and other marine predicted a-D-galactosidases are shown in Table 2 in the Supplementary Material.

Mechanisms of action
It is widely known that the stereochemical outcome of the enzymatic hydrolysis of O-glycosides can furnish products with inversion or retention of the anomeric configuration with respect to the starting substrates. Two mechanisms were considered initially. Retaining glycosidases maintain the anomeric configuration of the substrate in the products via a double displacement catalytic mechanism. Inverting glycosidases induced inversion in a one-step reaction. On this basis, a classification of glycosyl hydrolases has been adopted (Koshland, 1953), although other classifications are based on sequence similarities (Coutinho & Henrissat, 1999) and other mechanisms making use of different chemistry are today known (Jongkees & Withers, 2014). Of particular importance is the elimination-hydration mechanism found in the GH-4 family involving the presence of NAD + and mercaptoethanol. Most, if not all, a-D-galactosidases belonging to GH-27 and GH-36 were ascribed to retaining enzymes but in a recent article, inverting examples were reported among a-D-galactosidases useful for removal of the immunodominant galactose . The enzymes belonging to Bacillus fragilis and B. thetaiotaomicron were classified in two subfamilies of GH-110A and GH-110B and characterized by inverting mechanisms as determined by 1 H NMR spectroscopy Morley et al., 2009), a direct way for determining the stereochemical mechanism. Chemical shifts and coupling constants of the anomeric protons for substrates and products are distinct and readily observed. Retaining enzyme stereochemistry was assessed by this technique for GH-36 a-D-galactosidase from Thermotoga maritima (Comfort et al., 2007) and for three other GH-27 enzymes from Streptomyces griseoalbus (Anisha et al., 2011).
It is known that many GH-27 and GH-36 retaining enzymes are capable of catalyzing transglycosylation. Transglycosylation activities remain unexplored for retaining GH-97 a-D-galactosidases. GH-4 belonging a-D-galactosidases from Bacillus halodurans and Citrobacter freundii catalyze galactoside hydrolysis via an NAD-dependent redox reaction that is coupled to an a,b-elimination process involving the formation of a glycal intermediate (Anggraeni et al., 2008;Chakladar et al., 2011). Addition of water to the anomeric center, reprotonation of C2-C3 double bond and reduction of 3-keto group by NADH previously formed, regenerate the pyranose ring with net retention of configuration. The overall result is the hydrolysis of the substrate. For optimal activity, the GH-4 a-D-galactosidase, found in the species Citrobacter freundii, requires two cofactors, NAD + and Mn 2+ , and the addition of a reducing agent, such as mercaptoethanol. It seems, not reported in literature, to be a possible transglycosylation event for this a-D-galactosidase. In addition, the proposed mechanism of the GH-4 galatosidases could be considered not prone to transglycosylation as for GH-109 a-N-acetylgalactosaminidases (Liu et al., 2007). Indeed, the authors reported that when the enzyme from Citrobacter freundii catalyzed hydrolysis of the substrate in the presence of methanol (5 M), no trapping by methanol was observed. However, in the case of another enzyme (6-phospho-b-glucosidase) from Thermotoga maritima, it was proved that the biocatalyst had transglycosylation activity, as in the presence of methanol, methyl 6-phospo-b-glucoside was formed (Yip et al., 2004).

Practical applications of free and immobilized a-D-galactosidases
The a-D-galactosidase from coffee beans was one of the first to be purified and characterized and it is not a surprise that it is one of the most enzymes used in hydrolysis in the food industry and in biomedicine and for synthetic reactions described as follows. Hydrolytic properties of other enzymes from fungi or bacteria, especially regarding hydrolysis of antigens or highly improved performances by immobilization techniques, are detailed. Immobilization of enzymes has been one of the major activities in the field of biotechnology for any biocatalyst used in industry. The great draw that a-D-galactosidases have had in the food industry has made them biological materials for development of important immobilization techniques. A dedicated paragraph (Supplementary Material Part 8) is justified by the amount of studies starting with a didactic article (Mulimani & Dhananjay, 2007) and specialized ones (Bakunina et al., 2006;Corchero et al., 2012;Filho et al., 2008;Hernaiz & Crout, 2000;Kuo & Goldstein, 1983;Ohtakara & Mitsutomi, 1987;Okutucu et al., 2010;Pessela et al., 2008;Prashanth & Mulimani, 2005;Shankar et al., 2011;Singh & Kayastha, 2012;Tippeswamy & Mulimani, 2003) such as the study of human a-D-galactosidase A immobilization. The use of the synthetic ability of a-D-galactosidases by transglycosylation and reverse hydrolysis is described in Table 3.

Hydrolysis
Due to the presence of non-digestible raffinose-based sugars (Di Stefano et al., 2007) Figure 1) in soy products, the enzymatic hydrolysis by a-galactosidases has been of interest for years. The filamentous actinobacterium, Streptomyces griseoloalbus is a source of a-D-galactosidase that was used for the hydrolysis of soymilk and is also applied to reduce the  (Kitahata et al., 1992;Koizumi et al., 1995) Coffee bean a-galactosidase/ melibiose

6-O-a-D-galactosyl a-cyclodextrin
and minor 2-O-a-D-galactosyl a-cyclodextrin 2:1 w/w ratio donor/acceptor, 28% molar yield, 600 mg/ml donor initial concentration Transgalactosylated derivative of the donor was also present 2 1996 (Vic et al., 1996) Aspergillus oryzae a-galactosidase/galactose  content of raffinose oligosaccharides in horse gram and green gram flours. In comparison to traditional techniques, such as soaking and cooking, the enzymatic treatment was most effective and the raffinose content was reduced by 97.5% while stachyose was lowered by 93.2% (Anisha & Prema, 2008). The same reaction using Phaseolus vulgaris (family GH-27) and different crude enzymes from fungi was studied with essentially the same positive results (Song & Chang, 2006). An interesting application of a-D-galactosidase in the biomedicine domain for treatment of red blood cells is known from the early 1980s when the idea of converting blood group A and B antigens to H using specific exo-glycosidases capable of removing the immune-determinant sugar residues, appeared. The monosaccharide determining type A specificity is the terminal a-1,3 linked N-acetylgalactosamine, while the corresponding monosaccharide for B type specificity is an a-1,3 linked galactose. In group O, cells lack either of these terminal monosaccharides with presence of fucose a-1,2 linked. Very expensive acid enzyme extracted from coffee beans (GH-27) was initially used for conversion of B red blood cells but lack of appropriate biocatalyst for A antigens has been the reason of a lag in the progress of this specific aspect. Therefore, novel glycosidases with improved kinetic properties and specificities for this kind of reaction are always of interest (Olsson et al., 2004). Recently, one of the novel isoforms of a-D-galactosidases derived from B. fragilis was applied for the enzymatic removal of the major a-3-Gal xenotransplantation antigen using porcine and rabbit red blood cells . However, a-D-galactosidases inactivating serological activity of human B red blood cells are relatively rare enzymes. Such an effect on the B red blood cells was shown for the CAZy unclassified enzymes from the eukaryotes Colocacia esculenta (Chien & Lin-Chu, 1991), Trichomonas foetus (Yates et al., 1975) and Streptomyces sp. (9917S2) (Oishi & Aida, 1972) as well as Cephalosporium acremonium (Zaprometova et al., 1990) and Penicillium sp. 23 a-galactosidase (Varbanets et al., 2001). Complete conversion of B red blood cells to O red blood cells, was achieved with wild and recombinant GH-27 a-D-galactosidases from green coffee beans Coffea canephora (Harpaz et al., 1974;Harpaz et al., 1975;Goldstein, 1989) and the action of recombinant GH-27 a-D-galactosidase of soybean Glycine max (Davis et al., 1996;Hobbs et al., 1996). It was reported that recombinant GH-27 a-D-galactosidase from the cell culture of rice Oryza sativa, expressed by cells Pichia pastoris, has acquired the ability to completely transform B red blood cells to O red blood cells (Chien et al., 2008). Clinical tests on volunteers showed that modified red blood cells are viable and can function practically as native ones (Goldstein, 1984;Gong et al., 2005;Kruskall et al., 2000;Zhang et al., 2007;Zhu et al., 1996). Data on clinical trials of red blood cells transformed by GH-110 enzymes have not been found in the literature.
There is another interesting application of hydrolytic capabilities of a-D-galactosidases applied in chemical synthesis. Developing a ''natural strategy'' of these biocatalysts Fessner and collaborators used these enzymes to resolve anomeric product mixtures obtained from a simple acidcatalyzed Fischer galactosylation, providing single diastereomers easy to isolate (Ruiz et al., 2001).
Among important members of the human gastrointestinal microflora Bifidobacteria possess high a-D-galactosidase activity. Only in rare cases, enzymes have been shown to be able to synthesize galactooligosaccharides with melibiose, stachyose and raffinose as starting substrates. The products have the potential to boost the growth of bifidobacteria in the human gastrointestinal tract. Under optimum pH conditions for activity (pH 6.0) and high melibiose concentration (40% w/v), the enzyme from Bifidobacterium bifidum was able to form oligosaccharides with degree of polymerization (DP) !3 with a total yield of 20.5% (w/w) (Goulas et al., 2009). The enzyme from Bifidobacterium breve 203 (Aga2) synthesized a trisaccharide (Gal-a-1,4-Gal-a-1,6-Glc) using melibiose as a substrate. It was a new oligosaccharide containing Gal-a-1,4 linkage, a novel galactosidic link formed by microbial a-D-galactosidase. In a reaction using 100 mM melibiose, ca. 11% of the trisaccharide was formed which was isolated by a Biogel P2 column and characterized by 2D NMR spectroscopy (Zhao et al., 2008). Although the yield was modest, the authors conducted an analysis on acceptor specificity using different structures. In the presence of the aryl galactoside donor, Aga2 was able to catalyze glycosyl transfer to various acceptors including monosaccharides, disaccharides and sugar alcohols. Among them lactose was of interest being the disaccharidic fragment of globotriose.
a-D-Galactosidases Aga1 and Aga2 from Ruminococcus gnavus E1 were both able to perform transglycosylation reactions with a-(1,6) regioselectivity, leading to the formation of product structures up to [Hex] 12 and [Hex] 8 , respectively, in the presence of melibiose. Aga1 and Aga2 also catalyzed transglycosylation reactions with PNP-Gal as the donor and various sugar acceptors such as Man, Gal, Glc, Mel, Suc, Lac and Raf. Aga1 could transglycosylate all hexoses tested and a-(1,6) linked oligosaccharides (Raf and Mel) but could not transglycosylate Xyl or the b-linked sugars. It was suggested that Aga1 and Aga2 play essential roles in the metabolism of dietary oligosaccharides and could be used for the design of galacto-oligosaccharide (GOS) prebiotics (Cervera-Tison et al., 2012).
Enzyme engineering has been used to enhance the transglycosylation activity of glycosidases and to modify other features of interest in biocatalysis (Trincone, 2013). It is important in this paragraph to discuss a-galactosynthase using b-galactosyl-azide as a possible donor. The a-D-galactosidase (TM1192) from the hyperthermophilic bacterium Thermotoga maritima (TmGalA), belonging to family GH-36, was chosen as a model system. The mutant Asp327Gly is an efficient a-galactosynthase producing different galactosylated disaccharides from a b-galactosyl-azide donor and 4-nitrophenylaand b-glycosides as acceptors. This is the first a-galactosynthase produced so far and the authors concluded that the instability of fluoride derivatives as substrates could have hampered the development of a-glycosynthases, thus b-azide derivatives can serve as an attractive alternative for the future production of novel a-glycosynthases (Cobucci-Ponzano et al., 2011).

Conclusion
From the general analysis of the literature reported in this review, the result is clear that a-D-galactosidases are known enzymes that have found a number of useful applications. The deficiency of a-D-galactosidase A in the Anderson-Fabry's disease has made the topic to be almost exclusively covered in biomedicine. Further interest in the same domain has been added when the possible conversion of group B of red blood cells, removing enzymatically a-1,3-linked galactose, was reported. However, applications of a-D-galactosidases in food technology have not been neglected since concern in food science is still active in particular for new immobilizations and analytical techniques or in applications for design of galacto-oligosaccharide prebiotics using the synthetic capabilities of these enzymes. Improvements in enzyme production and properties, with regard to industrial applications, are being attempted by genetic approaches. Genomic data from marine microorganisms that play a crucial role in the global carbon cycle, suggest that the importance of the variable genome in tailoring individual strains to their specific lifestyles and functional repertoire. In addition, in this context, these studies are of great interest to turn wild types in optimal synthetic catalysts by site directed mutagenesis.
It has to be said, that the synthetic abilities appear to be the less exploited aspect for a-D-galactosidases and in future years interest for this specific aspect could increase for access to important small molecules by green technology. The emphasis on alternative sources of this class of enzymes, reported in this review, focuses on this need, as new enzymes could possess characteristics suitable for synthesis (selectivity, resistance to reaction conditions, etc.) more pronounced than the ones found in known examples.