2023 Volume 20 Issue 2 Article ID: e200017
Cooking with fire produces foods containing carbohydrates that are not naturally occurring, such as α-d-fructofuranoside found in caramel. Each of the hundreds of compounds produced by caramelization reactions is considered to possess its own characteristics. Various studies from the viewpoints of biology and biochemistry have been conducted to elucidate some of the scientific characteristics. Here, we review the composition of caramelized sugars and then describe the enzymatic studies that have been conducted and the physiological functions of the caramelized sugar components that have been elucidated. In particular, we recently identified a glycoside hydrolase (GH), GH172 difructose dianhydride I synthase/hydrolase (αFFase1), from oral and intestinal bacteria, which is implicated in the degradation of oligosaccharides in caramel. The structural basis of αFFase1 and its ligands provided many insights. This discovery opened the door to several research fields, including the structural and phylogenetic relationship between the GH172 family enzymes and viral capsid proteins and the degradation of cell membrane glycans of acid-fast bacteria by some αFFase1 homologs. This review article is an extended version of the Japanese article, Identification and Structural Basis of an Enzyme Degrading Oligosaccharides in Caramel, published in SEIBUTSU BUTSURI Vol. 62, p. 184–186 (2022).
Naturally occurring fructose polymers are essentially β-linked. Enzymes that degrade these polysaccharides have been extensively studied. On the other hand, information on enzymes that act on α-linked fructose polymers found in caramelized sugars is very limited, except for enzymes that synthesize and degrade some cyclic fructose disaccharide called difructose dianhydride (DFA). Our recently identified DFA I synthase/hydrolase is completely different from known carbohydrate-active enzymes in sequence homology and structural similarity, indicating its high degree of novelty. This discovery may lead to new developments in the field of applied glycoscience and structural enzymology and health science.
The major aspect of human eating habits that differs from those of other animals is the use of fire in cooking. Heating food causes various reactions that alter the chemical structure of the ingredients in the food, resulting in changes in texture and taste. The caramelization reaction is one of the typical examples of such chemical reactions. Fructose (Fru) and glucose (Glc), which constitute sucrose, are highly reactive reducing carbohydrates. Thus, when heated in the absence of water, they melt, and dehydration-condensation reactions occur within or between monosaccharides to form polymerized products with various structures.
Caramel, or caramelized sugar, is one of our most familiar food ingredients and delicacies, with an estimated 50 Mt consumed annually worldwide [1]. The composition of carbohydrates, heating temperature, heating time, and presence of additives are important factors that influence the type of compounds produced in the caramelization reaction. Due to the high complexity of the reactions, it is difficult to determine the full composition of the constituents of caramelized sugars even with current analytical techniques, but several reports using HPLC, MS, and NMR are available. For example, the distinctive sweet aroma of heated sucrose is attributed to a combination of several hundred different volatiles, including diacetyl [2,3]. The maximum degree of polymerization of the oligosaccharide produced is 6, and dehydration products are present in both monosaccharide and oligosaccharide forms [4]. Fragmentation products resulting from redox disproportionation reactions are also present as minor components with overall reduced and increased oxidation levels. Furthermore, trace amounts of aromatic compounds resulting from excessive dehydration reactions are present. The number of compounds detected reaches the hundreds, including those of unknown structure [4].
Because carbohydrates interconvert their ring structure by ring-opening aldehyde or ketone forms, α-anomers, β-anomers, pyranose, and furanose coexist in a certain equilibrium. Since dehydration-condensation reactions can occur in each conformation, the oligosaccharide products of the caramelization reaction are highly diverse. β-linked fructofuranose (Fruf) is found in plant-produced oligo- and polysaccharides such as sucrose (α-d-Glcp-1,2-β-d-Fruf), inulin ([β2,1-d-Fruf]n), and levan ([β2,6-d-Fruf]n) [5]. Caramelized sugars contain α-linked oligomers of Fruf, which are not produced by plants, and thus humans have created oligosaccharides that are rare in nature through the use of fire.
In terms of carbohydrate units, as described above, dehydration-condensation reactions can occur within a monosaccharide or between two monosaccharides. In particular, with fructose, the latter case can also occur between two disaccharides that have already formed a glycosidic bond. The formed cyclic disaccharides are called difructose dianhydrides (DFAs) or diheterolevulosans (DHLs) [6]. Depending on the conditions, DFAs and DHLs may account for 16 to 50% of the caramelized sugar, with 13 or 14 different disaccharide linkages, without counting all the modifications such as glycosyl-DFA [7–9].
Carbohydrate research is anticipated to develop in the form of new food products, materials, use as bioenergy, and applications in the medical field. Each of the constituents of caramelized sugars has potential for such research development and is of particular interest in the health sciences. In the following sections, we will discuss the reported effects of carbohydrates found in caramelized sugar in the biomedical field, in addition to some structural features of the enzyme used for the mass production of DFAs.
d-Fruf-α2,6-Glc (Figure 1A, top left) has approximately 25% of the sweetness of sucrose and is purified from caramelized sugar. Because Streptococcus mutans does not grow on this disaccharide as a carbon source, d-Fruf-α2,6-Glc is not cariogenic and differs significantly from sucrose in this respect. Moreover, it is not degraded by amylase in human saliva and is only slightly hydrolyzed by artificial gastric juice, porcine pancreatic amylase and rat intestinal enzymes. This result led to the conclusion that d-Fruf-α2,6-Glc is a nondigestible carbohydrate that can reach the large intestines if ingested orally. In colonization tests using representative bacteria from the human gastrointestinal tract, bifidobacteria and lactobacilli hydrolyzed this disaccharide, resulting in a marked decrease in the pH of the culture medium. On the other hand, Enterobacter cloacae, Escherichia coli, and Clostridium perfringens did not show hydrolysis or pH changes [10]. Because both bifidobacteria and lactobacilli are widely recognized as beneficial bacteria for human and animal health [11], the results suggest that d-Fruf-α2,6-Glc may be a prebiotic that is utilized by beneficial bacteria in the colon. The data also indicated that bifidobacteria and lactic acid bacteria may possess α-d-fructofuranosidase, but no genes involved in the degradation of d-Fruf-α2,6-Glc have been reported at this time.
(A) Chemical structure of d-Fruf-α2,6-Glc (top left), DFA III (top middle), DFA IV (top right) and DFA I (bottom left) and DFA V (bottom right). (B–C) Crystal structure of IFTase (DFA III-forming) from Bacillus sp. snu-7 (B, PDB ID: 2INV, [17]) and LFTase (DFA IV-forming) from Peanarthrobacter ureafaciens (C, PDB ID: 4FFI, [32]). In B, one of the subunits is colored. In C, the β-propeller domain is colored blue, and the β-sandwich domain is colored purple. Fructose residues in the active sites are shown as yellow sticks. The other fructose residues that are not related to the biological function of the enzyme are shown as black sticks.
Several DFAs can be prepared by simpler methods than purification from caramelized sugars containing large amounts of foreign substances. These methods include organic synthesis using pyridinium poly(hydrogen fluoride) and enzymatic synthesis [3,12,13]. In particular, for α-d-Fruf-1,2':2,3'-β-d-Fruf (DFA III, Figure 1A, top middle), β-d-Fruf-2,6':6,2'-β-d-Fruf (DFA IV, Figure 1A, top right), and α-d-Fruf-1,2':2,1'-β-d-Fruf (DFA I, Figure 1A, bottom left) enzymatic large-scale preparation systems have been developed, and characterization of the sugars for chemical and biological properties has been facilitated. In the case of α-d-Fruf-1,2':2,6'-β-d-Fruf (DFA V, Figure 1A, bottom right), a DFA V synthase was found in the culture medium of Aspergillus fumigatus [8]. DFA V synthase expression was induced when the fungus was cultured with inulin as the carbon source. However, the isolation of the enzyme has not been achieved.
From a physiological point of view, the most studied DFA is DFA III (Figure 1A, top middle), which is produced from inulin by inulin-fructotransferase (IFTase, DFA III-forming, EC 4.2.2.18) belonging to glycoside hydrolase family 91 (GH91) [14] in the carbohydrate-active enzyme (CAZy) database [15,16]. The crystal structure of IFTase (DFA III-forming) from Bacillus sp. snu-7 has been elucidated previously (PDB ID: 2INV) [17]. The enzyme is a trimer arranged in a C3 cyclic symmetry. Each monomer forms a right-handed parallel β-helix. Unlike most β-helix enzymes, whose active sites are located in the elongated groove parallel to the helical axis, the active site of IFTase (DFA III-forming) is formed by loops found at the subunit interface (Figure 1B). A close homolog of IFTase (DFA III-forming) has been reported to have DFA III hydrolase activity, and it is suggested that these two enzymes work in concert to monopolize the carbon source from inulin in several soil bacteria [18].
In both human and animal studies, DFA III has been reported to promote the absorption of minerals such as calcium, iron and zinc and the growth of the intestinal bacterium Ruminococcus productus [19–21]. In Japan, DFA III has been commercialized as a dietary supplement.
DFA IVDFA IV (Figure 1A, top right) can also be synthesized from levan by GH32 levan-fructotransferase (LFTase, DFA IV-forming, EC 4.2.2.16) [22]. LFTase from Peanarthrobacter ureafaciens is a monomeric enzyme composed of a five-bladed β-propeller appended to a β-sandwich, consisting of two sheets of six β-strands, common to GH32 enzymes (PDB ID: 4FFI) [20]. The active site is located close to the middle of the β-propeller domain (Figure 1C). DFA IV has improved in vitro fertilization and embryo development efficiency in pigs and is expected to have applications in assisted reproductive medicine in mammals [23]. It was also reported that DFA IV improves intestinal calcium absorption, wound healing, and barrier in the intestine, and is gradually metabolized by microorganisms in the large intestine [24,25].
DFA IDFA I (Figure 1A, bottom left) can be synthesized from inulin using GH91 IFTase (DFA I-forming, EC 4.2.2.17) [26], which exhibits approximately 40–50% sequence identity with IFTase (DFA III-forming). However, the structure of this enzyme has not been elucidated, and the difference in reaction mechanism compared to IFTase (DFA III-forming) remains an object of discussion [27]. Furthermore, the physiological function of DFA I has not been reported, except that it exhibits approximately 50% of the sweetness of sucrose [28].
Although bifidobacteria and lactic acid bacteria possibly have enzymes capable of degrading the rare oligosaccharides contained in caramelized sugars, none of these bacteria grow on DFA III or DFA IV as a carbon source, and no such experiments with DFA I have been reported at this time [22]. However, recently, we discovered a novel glycoside hydrolase (GH) suggesting the utilization of DFA I by bifidobacteria [29].
Discovery of a Novel α-d-FructofuranosidaseWe recently identified a novel GH, namely, GH172 difructose dianhydride I synthase/hydrolase (αFFase1) [29]. In this study, we focused on a gene cluster of Bifidobacterium dentium, a bacterium found in the human oral cavity [30] and gastrointestinal tract [31]. In this gene cluster, GH32 β-d-Frufase (BBDE_2039) and a protein containing a domain of unknown function, DUF2961 (BBDE_2040), were adjacent to each other. Since BBDE_2040 was then hypothesized to also be an enzyme involved in the degradation of Fruf molecular polymers, we cloned the gene for functional analysis. The release of d-Fruf was confirmed by thin-layer chromatography when the alkylated glycoside d-Fruf-α-Me was mixed with BBDE_2040 (Figure 2A). Considering this activity, BBDE_2040 was named αFFase1. A similar phenomenon was observed for d-Araf-α-Me, which has a similar structure to d-Fruf-α-Me (Figure 2B).
Function and structure of αFFase1. (A–D) The different reactions catalyzed by αFFase1. (A) d-Fruf-α-Me hydrolysis, (B) d-Araf-α-Me hydrolysis, (C) d-Fruf-β2,1-d-Fruf intramolecular dehydrating condensation, (D) inulobiose intramolecular dehydrating condensation. In panels C and D, atoms related to the formation of water molecules are shown in red. (E) The hexameric structure of αFFase1 (PDB ID: 7V1X, [29]). One protomer is shown color-coded by subdomain. A schematic representation of the domain organization is shown at the bottom of the figure. The active site is indicated by a red box.
To search for natural substrates of αFFase1, caramelized sugars were prepared by heating Fru and a mixture of Fru and Glc. Both caramelized sugars were mixed with αFFase1 and subjected to high-performance anion-exchange chromatography/pulsed amperometric detection (HPAEC-PAD) analysis, and a decrease in two peaks and an increase in another two peaks were observed compared to the enzyme-untreated negative control. To identify the chemical structures of the substrates and products of αFFase1, the compound corresponding to one of the two substrate peaks was purified from caramelized sugar. The reaction product sample was prepared by treating it with αFFase1, and NMR analysis of the substrate and product was performed. The substrate sample consisted of an isomeric mixture of disaccharides with fructopyranose (Frup) at the reducing end. The product was identified as cyclic fructose disaccharide, diheterolevulosan II (DHL II, α-d-Fruf-1,2':2,1'-β-d-Frup). So, it was suggested that the substrate is d-Frup-β2,1-d-Fruf in the isomerization mixture and that αFFase1 catalyzed the dehydration–condensation reaction of this disaccharide (Figure 2C).
d-Frup-β2,1-d-Fruf and DHL II have structural similarity with inulobiose (d-Fruf-β2,1-d-Fru) and DFA I (α-d-Fruf-1,2':2,1'-β-d-Fruf), respectively. Inulobiose was mixed with αFFase1, and DFA I was synthesized as expected. Interestingly, when DFA I was mixed with αFFase1, inulobiose was produced at a ratio of 9 to 1. Therefore, the reaction catalyzed by αFFase1 was an equilibrium reaction biased to promote the intramolecular dehydrating condensation reaction (Figure 2D).
The catalytic reactions of glycoside hydrolases are generally thought to follow the mechanism first outlined by Koshland [33]. The mechanistic details are beyond the scope of this review. However, to understand what follows, it is necessary to know about two typical reaction mechanisms of carbohydrate hydrolytic enzymes, the anomer-inverting mechanism and the retaining mechanism. The anomer-inverting mechanism is a hydrolytic mechanism involving an inversion of the anomeric configuration of the reducing end of released glycone. This is generally achieved by a single-displacement mechanism involving an oxocarbenium ion-like transition state. On the other hand, the anomer-retaining mechanism is a double-displacement mechanism involving a covalent glycosyl-enzyme intermediate, which is generally carried out while retaining the anomeric configuration. In both mechanisms, two aspartic acid or glutamic acid residues near the glycosidic linkage serve as catalytic residues (https://www.cazypedia.org/index.php/Glycoside_hydrolases). Among the glycoside hydrolases classified by CAZy (generally classified as EC 3.2.1.-), it is rare for an activated carbohydrate to be involved in the reaction, as in the case of glycosyltransferase (generally classified as EC 2.4.1.-). Thus, from a thermodynamic point of view, it is rare for glycoside hydrolases to catalyze the formation of glycosidic bonds (except for transglycosylases, glycoside phosphorylases and glycosynthases). However, all of the enzymes described in this review catalyze the synthesis of cyclic disaccharides by the formation of new glycosidic linkages and have partially different reaction mechanisms. Besides, to describe the binding mode of oligosaccharides, a nomenclature for subsites of glycoside hydrolases was proposed [34]. Subsites are numbered with increasingly negative numbers (–1, –2, –3, etc.) away from the cleavage point towards the non-reducing terminus, and with increasingly positive numbers (+1, +2, +3, etc.) towards the reducing terminus (https://www.cazypedia.org/index.php/Sub-site_nomenclature).
The GH91 and GH32 enzyme groups possessed by soil bacteria and some enteric bacteria have been studied in detail for DFA synthesis and degradation enzymes [17,18,32]. These enzymes utilize inulin or levan polysaccharides, which have a high degree of polymerization, as substrates. GH91 IFTase (DFA-III forming) catalyzes the reaction by a lyase-like mechanism, that is, an intramolecular hydroxide attack that does not involve any water molecule (Figure 3A and 3B, [17]). GH32 LFTase requires cleavage of the glycosidic bond before intramolecular fructosylation occurs (Figure 3C and 3D, [32]). This explains the attribution of EC 4.2.2.- for these enzymatic reactions. In contrast to these features, the substrate of αFFase1 (e.g., inulobiose) has the same oligomerization degree as the product (e.g., DFA I). Thus, it was speculated that the reaction mechanism of αFFase1 differs from that of the previously reported DFA synthases.
Molecular mechanism of the different DFA-synthesizing enzymes. The active site in the crystal structure (A, C, E) and the proposed catalysis mechanism (B, D, F) are represented for GH91 IFTase (DFA III-forming, A–B), GH32 LFTase (DFA IV-forming, C–D) and GH172 αFFase1 (E–F). In panels A, C and E, residues constituting the –2 subsites are indicated in orange characters, the –1 subsite in blue, the +1 subsite in red, the +2 subsite in green, and the residues placed in an ambiguous position in purple. In panel C, the active site of the D54N mutant, used to obtain the complex structure of GH32 LFTase with levanbiose, is shown. In panels B, D and F, subsites are indicated with blue numbers, and the bond rotations required for the ligand to fit in the active site are indicated with blue arrows.
The chemical structure of the initial product of the reaction catalyzed by αFFase1 was analyzed by 1H NMR. After the addition of the enzyme, the hydrolysis reaction of the synthetic substrate pNP-α-d-Araf was monitored, and the initial products showed the same α-anomeric form, indicating that αFFase1 catalyzes hydrolysis by an anomer-retaining reaction mechanism. However, since αFFase1 did not show any sequence homology with known carbohydrate-active enzymes, the catalytic residues of the enzymes implicated in such reactions were hard to estimate.
The closest structure to αFFase1 in the Protein Data Bank (PDB) was BACUNI_00161 from Bacteroides uniformis with 36% sequence identity (PDB ID: 4KQ7). However, this structure was solved as part of a structural genomics project, and its molecular function was unknown. Since there were no other similar structures in the PDB, it was difficult to determine the molecular function of αFFase1 from the previously reported protein structures. The structure of αFFase1 was then determined by X-ray crystallography at a resolution of 1.9 Å (PDB ID: 7V1V). αFFase1 has a hexameric structure with D3 bilateral symmetry. Each protomer consisted of two double β-jelly roll domains (DJR), common to DUF2961, and a long α-helix at the C-terminal end (Figure 2E).
In the complex structure of αFFase1 with d-Fruf (PDB ID: 7V1X) and d-Araf (PDB ID: 7V1W) obtained by cocrystallization, β-d-Fruf or β-d-Araf was visible at a specific site between protomers, determined as the active site. Based on the positional relationship with the anomeric carbon, E291 was identified as the catalytic nucleophile, and E270 was identified as the general acid-base catalyst. As both ligands interact stably with αFFase1 in the same form, we hypothesized that inulobiose and DFA I would also enter the active site in the same form (Figure 3E). We also identified the amino acid residues of αFFase1 that form the –1 subsite where the α-d-Fruf of inulobiose enters and the +1 subsite where the β-d-Fruf enters. Based on these findings, we proposed an anomer-retaining dehydration reaction mechanism by αFFase1 (Figure 3F). Although the anomer-retaining mechanism is major among glycoside hydrolases, to our knowledge, αFFase1 is the only enzyme that utilizes this mechanism in intramolecular dehydrating condensation reactions [35,36].
Among carbohydrate-active enzymes, the proteins belonging to the GH172 family have novel structures. Structural comparison revealed that except for BACUNI_00161, αFFase1 is somewhat similar to the virus-derived capsid-constitutive protein [37]. Indeed, detailed observation of the region corresponding to the GH172 family enzymes reveals a short α-helix downstream of the fifth β-strand of each β-jelly roll, formed by eight β-strands. Moreover, a short β-hairpin is inserted upstream of the sixth β-strand. As shown in figure 4A, although there is little sequence homology, these minor features are structurally similar to the double β-jelly roll major capsid protein (DJR-MCP) of viruses belonging to the Bamfordiviridae kingdom (PDB ID: 1HX6 [38], 2VVF [39], 3J31 [40], 5OAC [41], etc.). Furthermore, proteins with the GH172 family enzymes domain interact closely by extensive contacts at the interface between three subunits, assembling into a C3 cyclic group to form the basic structural unit (six supersecondary structures forming a C6 cyclic group if we count the β-jelly rolls). Depending on the enzymes, the basic structural units assemble in various manners to form trimers, hexamers, dodecamers, and others [42]. For example, in the case of αFFase1, two basic structural units form a quaternary structure with a D3 dihedral group, i.e., a hexamer. The tertiary and quaternary structures of the GH172 family enzymes and DJR-MCP are similar and further confirm the evolutionary relationship between these proteins (Figure 4B).
Structural comparison of different proteins with a DJR domain. A, Monomeric structure of the DJR sub-domain of GH172 αFFase1 (Top left, PDB ID: 7V1V), viral DJR-MCP (Top right, PDB ID: 2VVF), PNGase F (Bottom left, PDB ID: 1PNF) and PHM (Bottom right, PDB ID: 1SDW). Each protomer is colored in blue at the N-terminus to red at the C-terminus. B, Trimeric assembly of the GH172 αFFase1 basic structural unit (Top) and biological assembly of DJR-MCP. For each protein, protomers are colored by chain. Adapted from reference [37].
In addition to the GH172 family enzymes, some bacteria-originated proteins, such as peptide-N4-(N-acetyl-β-d-glucosaminyl) asparagine amidase F (PNGase F, PDB ID: 1PNF [43]) and peptidylglycine α-hydroxylating monooxygenase (PHM, PDB ID: 1SDW [44]), have a DJR fold (Figure 4A). However, the above-described short α-helix and β-hairpin are absent in PNGase F. The DALI Z score, a measure of structural similarity, is above 8.0 between DJR-MCP and the GH172 family enzymes but even lower between DJR-MCP and PNGase F or PHM. The same results were obtained when the two β-jelly rolls were analyzed separately. Single β-jelly rolls (SJR) are more ubiquitous than DJR folds and are found as SJR-MCP, GH68, and carbohydrate-binding modules (CBM). However, none of them exhibit higher structural similarity to DJR-MCP than the GH172 family enzymes domain (Z score).
Although a large number of theories are discussed as to the origin of viruses, many virus-derived proteins are thought to have diverged phylogenetically from cell-derived ones. Similarly, it is believed that cell-derived DJR fold proteins evolved from those of the SJR fold. In light of the above, one can envision a scenario in which DJR-MCP and the GH172 family enzymes diverged from the cell-derived DJR fold protein, respectively, or the GH172 family enzymes domain evolved from the cell-derived DJR fold protein, from which the virus acquired DJR-MCP.
αFFase1 shows DFA I synthase/hydrolase activity and forms a gene cluster with GH32 β-d-Frufase in the B. dentium genome, and B. dentium grows on inulobiose and DFA I as a carbon source (unpublished results). These facts suggest that αFFase1 works in concert with β-d-Frufase to degrade DFA I. On the other hand, it is interesting to note that αFFase1 possesses α-d-Arafase activity. The arabinofuranosyl units of plant glycans are only l-isomers, and α-d-arabinofuranosides are rarely found. For example, cell wall polysaccharides of Mycobacteria [45] and Corynebacteria [46], pilins of Pseudomonas aeruginosa [47], and lipopolysaccharide O-antigen of Stenotrophomonas maltophilia [48] were reported to contain α-d-arabinofuranosides. Some homologs of αFFase1 may be the enzyme primarily responsible for the degradation of such α-d-arabinan. BACUNI_00161, the protein of unknown function used to solve the αFFase1 structure by the molecular replacement method, belongs to DUF2961 (GH172 family enzymes domain), as does αFFase1 (Figure 5A). The putative –1 subsite residue of BACUNI_00161 contains the two catalytic glutamate residues and contains the same conserved residues as αFFase1. However, Cys304 appears to collide with the O2 atom of Fruf in the –1 subsite, and the presumptive +1 subsite residue is not conserved (Figure 5B and 5C). These facts suggest that BACUNI_00161 does not synthesize or degrade DFA I and that the GH172 family contains a biologically distinct group of enzymes that differ from αFFase1 in their function.
Structural comparison of the protein of unknown function BACUNI_00161 and αFFase1. A, Hexameric structure of BACUNI_00161. One of the subunits is color-coded to distinguish the two β-jelly roll domains and the C-terminal α-helix. Bound polyethylene glycol molecules and sodium ions are shown as white sticks and purple spheres, respectively. B, Putative active site of BACUNI_00161. C, The active site of αFFase1 for comparison. In B and C, residues that are not conserved between the two enzymes are indicated with red characters.
We have discovered an enzyme corresponding to the above in a distantly related homolog of αFFase1 of the soil bacterium Microbacterium arabinogalactanolyticum (unpublished results). This enzyme does not have α-d-fructofuranosidase activity but functions as an α-d-arabinofuranosidase. Analysis of the gene products of peripheral genes forming a gene cluster with this α-d-arabinofuranosidase revealed that this enzyme is involved in the degradation system of lipoarabinomannan (LAM) and arabinogalactan (AG), which are frequently found in the cell walls of acid-fast bacteria. In parallel with this discovery, HPAEC-PAD analysis has revealed that AG and LAM are degraded by the GH172 family enzymes from Nocardia brasiliensis, Mycosynbacter amalyticus, and the gut microbe Dysgomonas gadei [42].
Within the carbohydrate-active enzyme database CAZy, it is common to find enzymes with different physiological roles within the same family, and the newly discovered α-d-Arafase is now expected to be applied in the development of reagents for tuberculosis research.
We recently discovered a novel DFA I synthase/hydrolase, αFFase1. This enzyme catalyzes the dehydration and condensation of inulobiose through a unique reaction mechanism. The novelty of this enzyme is also indicated by the absence of sequence homology or structural similarity to previously reported carbohydrate-active enzymes. This discovery has resulted in various considerations, and new research has been initiated around the world. This review clearly shows that many challenges remain in the biological study of caramelized sugars and that each of them needs to be addressed in the truth. This may lead to the discovery of new functional carbohydrates and useful carbohydrate-active enzymes. The discovery of αFFase1 in particular is a good example of the high level of excitement experienced in the discovery of novel genes and proteins.
The authors declare no conflict of interest.
T.K. drafted the manuscript. All authors edited and completed the manuscript.
The evidence data generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
The authors thank Prof. Motomitsu Kitaoka (Faculty of Agriculture, Niigata Univ.) for valuable discussion. Research work related to αFFase1 was supported by grant-in-aid for scientific research JSPS-KAKENHI (15H02443 to S. F., K. F., and A. I.; 26660083 to S. F.; and 24380053 to S. F. and K. F.) and partially supported by grant-in-aid for scientific research JSPS-KAKENHI (19H00929 to S. F. and A. I.; 19K05789 to S. F.; 18K05345 to A. I.; and 19K05816 to K. F.).
