Remarkable evolutionary relatedness among the enzymes and proteins from the α-amylase family

Abstract

The α-amylase is a ubiquitous starch hydrolase catalyzing the cleavage of the α-1,4-glucosidic bonds in an endo-fashion. Various α-amylases originating from different taxonomic sources may differ from each other significantly in their exact substrate preference and product profile. Moreover, it also seems to be clear that at least two different amino acid sequences utilizing two different catalytic machineries have evolved to execute the same α-amylolytic specificity. The two have been classified in the Cabohydrate-Active enZyme database, the CAZy, in the glycoside hydrolase (GH) families GH13 and GH57. While the former and the larger α-amylase family GH13 evidently forms the clan GH-H with the families GH70 and GH77, the latter and the smaller α-amylase family GH57 has only been predicted to maybe define a future clan with the family GH119. Sequences and several tens of enzyme specificities found throughout all three kingdoms in many taxa provide an interesting material for evolutionarily oriented studies that have demonstrated remarkable observations. This review emphasizes just the three of them: (1) a close relatedness between the plant and archaeal α-amylases from the family GH13; (2) a common ancestry in the family GH13 of animal heavy chains of heteromeric amino acid transporter rBAT and 4F2 with the microbial α-glucosidases; and (3) the unique sequence features in the primary structures of amylomaltases from the genus Borrelia from the family GH77. Although the three examples cannot represent an exhaustive list of exceptional topics worth to be interested in, they may demonstrate the importance these enzymes possess in the overall scientific context.

Introduction

α-Amylase (EC 3.2.1.1) represents probably the best known and most deeply studied amylolytic enzyme [1–6]. It catalyzes the hydrolytic cleavage of the α-1,4-glucosidic linkages in starch and related α-glucans in an endo-fashion employing the retaining reaction mechanism. Its evolution started to attract the serious scientific interest approximately 25 years ago when it became clear that there is a group of starch hydrolases and related enzymes possessing closely related functions within the frame of homologous amino acid sequences [7–9]. It was, for example, the enzyme cyclodextrin glucanotransferase (CGTase; EC 2.4.1.19) sharing with the α-amylase the first step of the catalyzed reaction, which in the CGTase proceeds not with the molecule of water like in the α-amylase but with a molecule of a saccharide to be transferred [10]. For CGTases, it was thus revealed that, despite obvious differences between an α-amylase and a CGTase, they exhibit interesting sequence similarities with α-amylases [11] that previously might even complicate the correct assignment of the α-amylase/CGTase specificity [12–14] as well as the answer concerning their evolutionary history [15]. The other closely related enzyme was the so-called maltohexaohydrolase (EC 3.2.1.98), i.e., the exo-amylase producing maltohexaose [16]. The discovery of a completely novel enzyme, the neopullulanase (EC 3.2.1.135) in 1989 [17], able to perform both hydrolysis and transglycosylation of both the α-1,4- and α-1,6-glucosidic linkages [18–20], can be considered as a milestone in creating the overall view of the group of sequentially and functionally related amylolytic enzymes. Thus, also from the evolutionary point of view, a new enzyme family, named according to its leading member as the α-amylase family, has been established [20–24].

Interestingly, from the very beginning the α-amylase family has offered various surprises or at least the phenomena deserving a special attention. For example, even the α-amylases originating from various sources (roughly from Bacteria, Archaea and Eucarya) may differ from each other quite substantially in their substrate preference and/or product profile although they are still active towards and produce various α-glucans [1]. Some of them have been confirmed to be able to transglycosylate to a small, limited extent, e.g., the α-amylase from Pseudoalteromonas haloplanktis [25].

Historically, the establishment of the four conserved sequence regions (CSRs) among the primary structures of α-amylases in 1986 [26] has played an unique role, especially as one of the requirements for an enzyme to become a member of the family [20]. Soon after the four CSRs were observed in other enzymes (members) of the developing α-amylase family [8, 9, 20, 22] and completed by the three additional CSRs [27–29]. Currently, it has been recommended to use as many as possible of the seven defined CSRs [30, 31] to best characterize a protein as a member of the α-amylase family, mainly if there are any doubts concerning the exact enzyme specificity [1, 32].

Almost simultaneously with the gradual appearance and final definition of the α-amylase enzyme family in the literature early in 1990s [7–9, 20–24], a pioneering study was published [33] delivering the newly developed concept of classification of glycoside hydrolases (GHs) to sequence-based families, placing the group of enzymes known already at that time as the α-amylase family into the family GH13. After a few published updates [34, 35], the increasing system of GH families has been incorporated into the current web-server database CAZy (Carbohydrate-Active enzymes) [36], the α-amylase enzyme specificity being potentially found also in families GH57, GH119 and even GH126 [1, 37]. Nevertheless, the family GH13 represents the main α-amylase family [1] and with approximately 31,500 sequences and more than 30 different enzyme specificities (Table 1) it belongs—among 129 GH functional families created until now—to the largest GH families at all [37]. On a higher level of hierarchy, it forms the clan GH-H together with families GH70 and GH77 [31, 37, 38]; the former of these two families covers various glucansucrases with typically circularly permuted primary structure with respect to that seen in the main family GH13 [39–41], whereas the latter is the monospecific family of 4-α-glucanotransferases known also as amylomaltase and disproportionating enzyme (DPE) in prokaryotes and eukaryotes, respectively [42–44]. On the lower level of hierarchy, in 2006 the family GH13, reflecting also the previous efforts [45], was officially divided into 35 subfamilies by the CAZy curators [46]; currently the number of GH13 subfamilies has reached 41 [37, 47, 48] and it will very probably raise in the future [49].

Table 1.

Members of the α-amylase GH families

Enzyme class	Enzyme	EC no.	GH family	GH13 subfamily
Hydrolases	α-Amylase	3.2.1.1	13, 57, 119, 126	1, 5, 6, 7, 15, 20, 24, 27, 28, 32, 36, 37
	Oligo-1,6-glucosidase	3.2.1.10	13	23, 31
	α-Glucosidase	3.2.1.20	13	17, 21, 23, 30, 31, 40
	α-Galactosidase	3.2.1.22	57
	Pullulanase	3.2.1.41	13	12, 13, 14
	Amylopullulanase	3.2.1.1/41	13, 57	12, 14, 39
	Cyclomaltodextrinase	3.2.1.54	13, 57	20, 36
	Maltotetraose-forming amylase	3.2.1.60	13	19
	Isoamylase	3.2.1.68	13	11, 14
	Isoamylase/4-α-glucanotransferase	3.2.1.68/2.4.1.25	13	11
	Dextran glucosidase	3.2.1.70	13	31
	Trehalose 6-phosphate hydrolase	3.2.1.93	13	29
	Maltohexaose-forming amylase	3.2.1.98	13	5, 19
	Maltotriose-forming amylase	3.2.1.116	13	2, 32
	Maltogenic amylase	3.2.1.133	13, 57	2, 20
	Neopullulanase	3.2.1.135	13	20
	Maltooligosyltrehalose threhalohydrolase	3.2.1.141	13	10
	Sucrose hydrolase	3.2.1.–	13	4
	Maltopentaose-forming amylase	3.2.1.–	13	5
	Glycogen degrading enzyme	3.2.1.–	13	12
	Cyclic α-maltosyl-1,6-maltose hydrolase	3.2.1.–	13	20
Transferases	Amylosucrase	2.4.1.4	13	4
	Dextransucrase	2.4.1.5	70
	Sucrose phosphorylase	2.4.1.7	13	18
	Glucan branching enzyme	2.4.1.18	13, 57	8, 9
	Cyclodextrin glucanotransferase	2.4.1.19	13	2
	4-α-Glucanotransferase	2.4.1.25	13, 57, 77
	Glucan debranching enzyme	2.4.1.25/3.2.1.33	13	11, 20, 25
	Alternansucrase	2.4.1.140	70
	α-1,3-Glucan synthase	2.4.1.183	13	22
	α-1,4-Glucan: phosphate α-maltosyltransferase	2.4.99.16	13	3
	Sucrose-6-phosphate phosphorylase	2.4.1.–	13	18
	α-Transglucosidase	2.4.1.–	13	23
	Isocyclomaltooligosaccharide glucanotransferase	2.4.1.–	13
	α-4,6-Glucanotransferase	2.4.1.–	70
	Reuteran sucrase	2.4.1.–	70
	α-1,6/α-1,2-Branching glucansucrase	2.4.1.–	70
Isomerases	Isomaltulose synthase	5.4.99.11	13	31
	Maltooligosyltrehalose synthase	5.4.99.15	13	26
	Trehalose synthase	5.4.99.16	13	16, 33
HAT proteins	hc-rBAT protein	–	13	35
	4F2hc antigen	–	13	34

As far as the family GH13 is concerned in its entirety, it has become evident that the same enzyme specificity (e.g., the α-amylase or pullulanase) may exist in several separated groups or GH13 subfamilies, but on the other hand some other specificities (e.g., oligo-1,6-glucosidase, α-glucosidase and dextran glucosidase) may exist altogether within a single group or GH13 subfamily. The attractiveness of the α-amylase clan GH-H for scientists has been strengthened also by classifying into the family the heavy-chains of heteromeric amino acid transporters, known as rBAT protein and 4F2 antigen (i.e., neither amylolytic enzymes, nor any enzymes at all) that, being typically of animal origin, are evolutionary related to bacterial α-glucosidases [50]. The present review aims to deliver the updated view of a few perhaps most remarkable evolutionary relatedness observed within the α-amylase family.

The family GH13 α-amylases from plants and archaea

While the first amino acid sequences of the α-amylases from Archaea have become available only at the end of 1990s [51–55], the primary structures of their first plant counterparts were determined at least 10–15 years earlier [56–60]. Moreover, the three-dimensional structure of the high pI isozyme of the barley α-amylase AMY-2 was solved a few years before [61] the first archaeal α-amylase sequence from Pyrococcus furiosus was announced [51, 52]. The barley α-amylase AMY-2 structure [61], solved later also as a complex with acarbose [62], was at that time only the fourth known α-amylase tertiary structure [30], in addition to those from Aspergillus oryzae [63, 64], Aspergillus niger [65, 66] and pig pancreas [67, 68]. At present, with regard to plant and archaeal α-amylases, tertiary structures have been solved and published also for the low pI isozyme of the barley α-amylase AMY-1 [69, 70], the α-amylase from rice [71, 72] as well as the α-amylase from Pyrococcus woesei [73].

From the evolutionary point of view, before the α-amylases from archaea were known, plant α-amylases formed a compact cluster neighboring with the liquefying α-amylases from bacilli [28]. As documented by various evolutionarily oriented studies [73–78] this has remained true until now with a significant upgrading of the original picture [28] illustrating that the cluster of archaeal α-amylases shares the branch with that of plant counterparts [53, 79]. According to the CAZy nomenclature [46], the α-amylases from plants and archaea have been assigned the subfamily number GH13_6 and GH13_7, respectively; the bacterial homologues representing mostly liquefying α-amylases from bacilli being classified within the subfamily GH13_5. It should be pointed out that all these three subfamilies of α-amylases are mutually very closely related and the exact branching pattern among them in the evolutionary tree may depend on the aligned segment of their amino acid sequences (i.e., just CSRs, the catalytic domain or the entire sequence, etc.) [76].

The pronounced relatedness between the plant and archaeal α-amylases is remarkable not only due to their long taxonomical distance (eukaryotic plants and prokaryotic archaea) but also when differences in their thermostability (or the temperature optimal for their enzymatic activity) are considered [1]. The close positions in the evolutionary tree obviously reflect the similarities throughout their amino acid sequences, especially within the CSRs (Fig. 1) described first in detail in 1999 [79]. While the sequence similarities shared by both plant and archaeal α-amylases represent something that distinguishes both subfamilies from remaining GH13 α-amylases, there have to be some additional sequence features that are unique for each group, i.e., that discriminate the two subfamilies from each other, just these unique sequence features should be used in the efforts aimed at identifying the factors that could be responsible for the hyperthermostability of archaeal α-amylases (or, on the other hand, the low thermostability of the α-amylases from plants). Thus, for example, the two residues deserving the attention in the α-amylase from Thermococcus hydrothermalis could be as follows (Fig. 1): (1) the first position of the CSR-V, i.e., Tyr184Ala (184_YPDIC in the T. hydrothermalis α-amylase versus 146_APDID in the barley high pI α-amylase isozyme), because among the archaea it is occupied only by either phenylalanine or tyrosine (i.e., an aromatic residue); and (2) the third position of the CSR-IV, i.e., Ala308Asp (306_FVANHD versus 284_FVDNHD), because there is invariantly conserved alanine among the archaea in comparison with invariant aspartic acid in plants.

Fig. 1.

Comparison of seven CSRs of the family GH13 α-amylases with focus on the subfamilies GH13_6 and GH13_7. These CSRs cover mostly individual β-strands of the catalytic TIM-barrel domain [32]. Each α-amylase in the list is characterized by the GH13 subfamily number [46], the source of origin (the organism) and the accession number from the UniProt database [221]. The α-amylases from the subfamilies GH13_6 and GH13_7 were collected from the actual CAZy database [37], whereas the set of representative α-amylases from the GH13 remaining subfamilies was prepared according to previous in silico studies [74–79]. The catalytic triad is signified by black-and-white inversion. Sequence features characteristic of the GH13_7 archaeal α-amylases are highlighted in yellow. The features typical specifically for each group of α-amylases within the two GH13 subfamilies, i.e., (1) archaea and (flavo)-bacteria within GH13_7; and (2) plants and bacteria within GH13_6, which may discriminate the individual groups from each other are emphasized by respective colors

Recently, some hypothetical bacterial (i.e., not archaeal) α-amylases from genome sequencing projects were assigned to the “archaeal” subfamily GH13_7 [1, 37]. Currently, in the CAZy database, all these belong to the phylum Flavobacteria [37]. It is, however, worth mentioning that even a simple BLAST search [80] using the flavobacterial GH13_7 α-amylase from Sinomicrobium sp. 5DNS001 [81] as a query retrieves hypothetical bacterial α-amylases with GH13_7 sequence features, but not belonging to Flavobacteria. Anyhow, the moderately thermostable Sinomicrobium sp. 5DNS001 α-amylase [81], clearly homologous to its archaeal hyperthermostable counterparts (Fig. 2), may add to our understanding of the rules that have governed the evolution of plant, (flavo)-bacterial and archaeal α-amylases and the factors that are responsible for their thermostability differences [76, 79, 81–83]. In analogy with the above-mentioned features in which the plant and archaeal α-amylases differ from each other, it makes sense to try to identify the features in the amino acid sequences of (flavo)-bacterial α-amylases that are well conserved but simultaneously well discriminating them from their archaeal counterparts (Fig. 1). One example is the second position of the CSR-VII, i.e., Tyr363Gln (362_GYPTVFYGD in the Sinomicrobium sp. 5DNS001 α-amylase versus 330_GQPAIFYRD in the T. hydrothermalis α-amylase isozyme), because both the tyrosine and the glutamine are invariantly conserved among the respective groups of bacterial and archaeal α-amylases. Thus, a silico analysis as detailed as possible focused on comparison of amino acid sequences is of special importance, especially if, at the tertiary structure level the individual representatives of both GH13_6 and GH13_7 subfamilies look very similar and obviously without any substantial differences (Fig. 2).

Fig. 2.

Evolutionary tree of the family GH13 α-amylases with focus on the subfamilies GH13_6 and GH13_7. The tree is based on the alignment of seven conserved sequence regions, shown in Fig. 1. The tree was calculated using the neighbor-joining clustering [222] implemented in the Clustal-W2 phylogeny package [223] available at the European Bioinformatics Institute’s web-site (http://www.ebi.ac.uk/), and then displayed with the program iTOL [224]. Tertiary structures of representatives of flavobacterial, archeal and plant family GH13 α-amylases are shown near their clusters in the tree. Sources of the α-amylases: Sinomicrobium sp. 5DNS001 [81] (subfamily GH13_7; flavobacteria); Pyrococcus woesei [72] (subfamily GH13_7; archaea); Hordeum vulgare—barley high pI isozyme AMY-2 [61] (subfamily GH13_6; plants). The archaeal and plant α-amylases are experimentally solved crystal structures retrieved from the Protein Data Bank (PDB) [225] under the PDB codes 1MWO and 1AMY, respectively. The flavobacterial α-amylase is a tertiary structure model obtained at the fold recognition server Phyre-2 [89] for its amino acid sequence (UniProt accession number: L7Y1I6; residues: Gly52-Gly477) based on the P. woesei α-amylase structure (1MWO) as template. The individual domains are colored as follows: catalytic (β/α)₈-barrel—red, domain B—blue, domain C—green. The structural models were displayed with the program PyMol [226]

It is worth mentioning that, in addition to archaeal α-amylases from the class Thermococci (namely the two genera Pyrococcus and Thermococcus) classified within the subfamily GH13_7, there are several potential α-amylases produced by halophilic archaea (the class Halobacteria) deserving the attention that, however, until now have not been assigned any GH13 subfamily [37]. Some of them simply may not be true α-amylases, e.g., those from Natronococcus amylolyticus [84, 85] and Haloarcula japonica [86] producing mainly maltotriose and maltose, respectively. For others, the ambiguity arises from the fact that the studied sequences originate from genome projects [87], i.e., they are still hypothetical amylolytic enzymes. Interestingly, for the sequence of the halophilic α-amylase from Haloarcula hispanica [88] the fold recognition server Phyre-2 [89] revealed the Bacillus stearothermophilus α-amylase from the subfamily GH13_5 [90] as the best structural template.

Animal transport proteins rBAT and 4F2 versus the microbial GH13 α-glucosidases

A sequence resemblance of heavy-chains of the heteromeric amino acid transporters (hcHATs), the rBAT protein and the 4F2 antigen, to α-glucosidases from the α-amylase family was originally revealed already in 1992 in two independent studies [91, 92]. However, for a few subsequent years it was, in fact, not seriously taken into account by researchers engaged in the enzymology of the α-amylase family. The main reason was that the hcHAT proteins obviously do not obey all the criteria defined for an enzyme to be a member of the α-amylase family [93, 94]. The eventual evolutionary relatedness between the hcHATs and the α-amylase family enzymes was first studied in a detail in 1997 [95] when the attention was paid mainly to domain B, protruding typically in the family GH13 out of the catalytic (β/α)₈-barrel (simply called the TIM-barrel [1, 30, 31]) between the strand β3 and the helix α3 [61, 63, 67, 72, 96–104], which in the rBAT protein is present, whereas the 4F2 antigen lacks it (Fig. 3a). Both the rBAT protein and 4F2 antigen belong to the so-called solute carrier families SLC7 (the light chain) and SLC3 (the heavy chain); the respective hetero-chains being connected by a disulphide bridge [105–109]. Both groups have been accepted as members of the CAZy α-amylase family GH13 members and received the subfamily numbers GH13_34 and GH13_35 for the heavy chains of 4F2 antigens and rBAT proteins, respectively [46]. HAT proteins are responsible for transport of various types of amino acids across the plasma membrane in animals, namely in mammals, and their defects may lead to a failure in amino acids re-absorption and digestion (e.g., cystinuria, lysinuric protein intolerance) [107–109]. While the light chain has nothing to do with the family GH13 (it is a hydrophobic transmembrane protein consisting of 12 α-helices), the heavy subunit consists of the intracellular N-terminus, a transmembrane region and a large extracellular C-terminal part that exhibits the unambiguous sequence-structural similarity with α-glucosidases from the α-amylase family GH13 [91, 92, 110, 111]. Thus, the transportation activity is operated by the light subunit, whereas the heavy chain functions as a chaperone to help to orient the light chain to a proper position in the plasma membrane [112–114]. Note that, only the heavy subunit in its C-terminal part exhibits clear sequence similarities with the members of the α-amylase family [50, 91, 92, 95, 103, 104].

Fig. 3.

a Comparison of tertiary structures of the family GH13 hcHAT proteins from animals and α-glucosidase from bacteria. Sources of the proteins: Homo sapiens 4F2hc antigen [103] (subfamily GH13_34; left); Homo sapiens rBAT protein [111] (subfamily GH13_35; middle); Bacillus cereus oligo-1,6-glucosidase [96] (subfamily GH13_31; right). b Amino acid sequence alignment of the same three proteins: human 4F2hc antigen [103] (UniProt accession No.: P08195-1); human rBAT protein [111] (Q07837); Bacillus cereus oligo-1,6-glucosidase (OGLU) [96] (P21332). Color code for the selected residues: W, yellow; F, Y—blue; V, L, I—green; D, E—red; R, K—cyan; H—brown; C—magenta; G, P—black. The seven characteristic CSRs are boxed and marked above the alignment. The colored lane above the alignment blocks means the three domains shown in a. The catalytic triad is signified by asterisks under the alignment. c Structural overlay emphasizing the longer loop 4 connecting the strand β4 to the helix α4 in the oligo-1,6-glucosidase (green) and rBAT protein (blue) in comparison to a very short version present in the 4F2hc antigen (red). The structures were superimposed using the MultiProt web-server [227] (http://bioinfo3d.cs.tau.ac.il/MultiProt/); the overlap being characterized by 280 corresponding C_α-atoms and the RMSD value of 0.96 Å. The human 4F2hc antigen and bacterial oligo-1,6-glucosidase are experimentally solved crystal structures retrieved from the PDB [225] under the PDB codes 2DH3 and 1UOK, respectively. The human rBAT protein is a tertiary structure model obtained at the fold recognition server Phyre-2 [89] for its amino acid sequence (UniProt accession number: Q07837; residues: Asp116-Glu649) based on the B. cereus oligo-1,6-glucosidase structure (1UOK) as template. The individual domains are colored as follows: catalytic (β/α)₈-barrel—red, domain B—blue, domain C—green. The structures were visualized with the program PyMol [226]

Interestingly, however, the heavy chains of both the rBAT protein and the 4F2 antigen display the best similarity to the α-amylase family GH13 enzymes in the parts of their sequences that are not involved in the catalytic action (Fig. 3b), i.e., in CSR-VI (the strand β2 of the catalytic TIM-barrel) and CSR-VII (the strand β8) [50]. While the heavy chain of the 4F2 antigen does not possess the segment corresponding with the domain B [50, 95, 103], the heavy subunit of the rBAT protein contains the entire domain B that is, moreover, very closely related to its counterpart seen in the members of the so-called oligo-1,6-glucosidases [45, 48, 95], sharing even the sequence fingerprint QPDLN of the CRS-V [50, 95]. The evidently closer relationships between the enzymes from the oligo-1,6-glucosidase subfamily and the heavy-chains of the rBAT proteins is reflected also in the presence of catalytic residues in the transport proteins, although in all cases the entire GH13 catalytic triad is not conserved completely. This feature could even be traced among the potential sequences of the heavy subunits of the rBAT proteins originated from basal metazoa [50]. For the heavy chain of the 4F2 antigen, no α-glucosidase activity has been detected [103], which is in agreement with a lack of residues that would correspond to the catalytic triad of the members of the α-amylase family [50, 103, 104], i.e., the aspartic acid as the catalytic nucleophile at the strand β4 (the CSR-II), the glutamic acid as the proton donor at the strand β5 (the CSR-III) and the aspartic acid as the transition-state stabilizer at the strand β7 (the CSR-IV) [1, 31].

With regard to sequence-structural resemblance of heavy chains of both the rBAT protein and 4F2 antigen, it is worth mentioning that, for example, the human 4F2 heavy chain lacks not only domain B, but also a stretch of about 40 amino acid residues succeeding the strand β4, i.e., it possesses a very short loop 4 connecting the strand β4 to helix α4 [50], whereas in both the oligo-1,6-glucosidase and the human rBAT heavy chain (both having also the entire domain B [93, 102]) the loop connecting the strand β4 to the helix α4 is longer (Fig. 3c). It is of note that even the GH13 neopullulanase subfamily members [45, 115], possessing shorter domain B [95], also lack the longer excursion of the loop 4 segment, but currently it is still unknown whether or not the domain B in GH13 oligo-1,6-glucosidase subfamily (and also in rBATs) operates in conjunction with the prolonged loop 4 (the β4 to α4 connection) [50]. It seems, however, that the consecutive loss of domain B in the heavy subunits of the 4F2 antigens might be connected with adequate shortening of the loop4, since this observation can be generalized to all 4F2 heavy-chains proteins [50]. Noticeably, similar heteromeric amino acid transporter system known in mammals was observed in insect [116, 117], nematodes [118] and schistozomes [119]. According to the in silico analysis of hcHATs and their enzymatic counterparts from the α-amylase family GH13 [50], a protein that could be close to the common ancestor of hcHATs and GH13 α-glucosidases might be represented by the hypothetical GH13-like protein from the cnidarian starlet sea anemone Nematostella vectensis.

Due to the obviously closer similarity of the heavy chains of rBAT proteins to the family GH13 α-glucosidases than that of those of 4F2 antigens, it is not easy to draw an unambiguous scenario of their evolution [95]. Based on a detailed in silico analysis [50], two different scenarios should be considered (Fig. 4). One of the possibilities means that the division of hcHATs from the enzymes of the α-amylase family might occur in one single event in basal Metazoa and subsequent split to rBAT proteins and 4F2 antigens in chordates. The above-mentioned higher similarity of the rBAT proteins to the GH13 enzymes than to the 4F2hc antigens could be explained by a selection pressure acting against change upon rBAT (and not upon 4F2hc), to preserve its eventual enzymatic capabilities, although until now no evidence has been delivered concerning the enzymatic activity of any rBAT protein [50]. Alternatively, the 4F2hc protein could be experiencing accelerated evolution toward some new function, accompanied by changes in both sequence and structure, while rBAT has retained its original function, remained mostly unchanged and thus similar to GH13 enzymes. The other eventuality assumes two independent branching events: one of the 4F2 antigens in the basal Metazoa and the other one of the rBAT proteins directly from enzymes of oligo-1,6-glucosidase subfamily of the α-amylase family in chordates. Since chordates obviously do not possess enzymes of the oligo-1,6-glucosidase subfamily [45, 48, 50], their α-glucosidases counterparts could have been transformed into the heavy chains of the rBAT proteins [120]. Both scenarios (Fig. 4) reflect the ancestry of both rBAT proteins and 4F2 antigens anchored within the GH13 α-amylase family, the difference being only in the way leading from the GH13 enzymes either to rBAT and 4F2 together or to rBAT and 4F2 separately [50]. At present, it seems that the hcHATs are found in animals starting from basal Metazoa, but it is not possible to exclude that in the future some new sequences of hcHAT-like proteins of non-metazoan origin become available [121].

Fig. 4.

Evolutionary scenarios of hcHAT proteins with respect to the α-amylase family GH13. The left eventuality considers a single-event division of both rBAT proteins and 4F2hc antigens from the enzymes of the α-amylase family in basal Metazoa with a subsequent split to rBAT proteins and 4F2 antigens in higher animals like chordates. The right possibility counts with two independent evolutionary events leading first to separation of the 4F2hc antigens from the α-amylase family enzymes in the basal Metazoa and second to recruitment of the rBAT proteins from enzymatic members of the oligo-1,6-glucosidase subfamily in higher animals

The family GH77 amylomaltases from borreliae

The family GH77, together with families GH13 and GH70, is a member of the α-amylase clan GH-H [31, 37]. As already mentioned above, it is, in contrast to the main α-amylase family GH13, a monospecific family, i.e., it contains only one enzyme specificity of 4-α-glucanotransferase (EC 2.4.1.25) [122]. The trivial name of the 4-α-glucanotransferase within the family GH77 has been distinguished with regard to taxonomy—while in prokaryotes (both Bacteria and Archaea) the name amylomaltase has been used, the name disproportionating enzyme (DPE) has been established in eukaryotes (mainly in plants and green algae) [42–44, 123–135]. In general, the 4-α-glucanotransferase catalyzes, employing the retaining reaction mechanism, the intermolecular transglycosylation of α-1,4-glucans, i.e., it transfers a glucan chain from one α-glucan to another one or within a single linear glucan molecule to produce a cyclic α-1,4-glucan [42–44, 136–138]. Currently, the family counts almost 3000 sequenced members with absolute domination of Bacteria accompanied approximately equally by a few dozens each from Archaea and Eucarya [37].

From the structural point of view, since the family GH77 is a member of the α-amylase clan GH-H, the GH77 4-α-glucanotransferase possesses a typical catalytic (β/α)₈-barrel (TIM-barrel). It contains just more insertions in comparison with the family GH13 TIM-barrel having, in fact, only domain B protruding out of the barrel in the place of the loop 3 connecting the strand β3 to the helix α3 (Fig. 2). The family GH77 4-α-glucanotransferase TIM-barrel has thus three subdomains called B1, B2 and B3 (Fig. 5), of which B1 and B3 correspond with domains B and C in the family GH13, while the subdomain B2 is unique for the family GH77 [44, 139–143]. It is, however, worth mentioning that although the subdomain B3 may play the role of the C-terminal family GH13 domain C succeeding the catalytic TIM-barrel, it is not an antiparallel β-sandwich (a Greek key motif) seen typically in the GH13 (see Figs. 2, 5).

Fig. 5.

Tertiary structures of the family GH77 amylomaltases from bacteria and their comparison. Sources of the amylomaltases: a Thermus aquaticus [139]; b Borrelia burgdorferi [129]. c Structural overlay focusing on the active-site residues with complexed acarbose in the Thermus amylomaltase (red) compared with the naturally mutated corresponding residues in the Borrelia amylomaltase (blue). The structures were superimposed using the MultiProt web-server [227] (http://bioinfo3d.cs.tau.ac.il/MultiProt/); the overlap being characterized by 488 corresponding C_α-atoms and the RMSD value of 0.18 Å. The Thermus amylomaltase is experimentally solved crystal structure retrieved from the PDB [225] under the PDB code 1ESW, whereas the Borrelia amylomaltase is a tertiary structure model obtained at the homology modeling server SwissModel [228] for its amino acid sequence (UniProt accession number: A6YM39; residues: Asn12-Ala507) based on the T. aquaticus amylomaltase structure (1CWY [44]) as template. The individual domains are colored as follows: catalytic (β/α)₈-barrel—red, subdomain B1—cyan, subdomain B2—magenta, subdomain B3—green. The exact positions of displayed active-site residues in the amino acid sequence can be identified in the alignment shown in Fig. 6. Based on the accepted nomenclature [229] the acarbose occupies the subsites from −3 to +1. The structures were visualized with the program PyMol [226]

Within the family GH77, the amylomaltases from the genus Borrelia obviously play a special (evolutionary) role [122]. Originally, based on the complete genome sequence of the Lyme disease spirochaete Borrelia burgdorferi [144], an in silico analysis published in 2003 [145] delivered a remarkable observation of mutations in several conserved and functionally important positions in the B. burgdorferi hypothetical amylomaltase encoded by the gene malQ. The importance of the eventual mutations in the hypothetical GH77 amylomaltase was strengthened by the fact that, one of the residues was the otherwise throughout the entire clan GH-H invariantly conserved arginine in the position i-2 with respect to catalytic nucleophile (aspartic acid) located at the strand β4 of the catalytic TIM-barrel, i.e., in the CSR-II [145]. The possible implications of remarkable in silico observations thus evoked a serious interest for experimental confirmation. The gene malQ from B. burgdorferi was cloned and expressed in Escherichia coli and the recombinant amylomaltase was shown to exhibit not only all the unique sequence features seen in the hypothetical MalQ but also the amylomaltase activity [129].

According to the most recent study focused on analysis of all available amylomaltase sequences from borreliae within the context of the entire family GH77 [122], 32 such sequences can be, in fact, divided into a few groups. Basically, there are only two major groups with respect to the presence/absence of the above-mentioned functional arginine, which—if naturally substituted—only lysine has been observed to replace it (Fig. 5c). It is the position of the Arg291 in the amylomaltase from Thermus aquaticus [43] mutated to Lys306 in the B. burgdorferi counterpart [129]. It should be pointed out that with regard to domain composition and length, all 32 amylomaltases from borreliae resemble typical bacterial GH77 amylomaltases represented by the enzyme from T. aquaticus. Nevertheless, some of those with the arginine to lysine mutation contain additional substitutions in several other well-conserved positions. On the other hand, some of those, possessing the original arginine, keep also those well-conserved positions (Fig. 6) characteristic for typical prokaryotic amylomaltases of non-borrelian origin [145]. Besides there are also some that exhibit an intermediary character. This phenomenon as a whole makes the amylomaltases from borreliae a unique evolutionary lineage in the family GH77 [122]. However, it should also be taken into account that until now of the 32 available GH77 amylomaltase sequences from borreliae, the MalQ from B. burgdorferi is the only one that has already been biochemically characterized and found enzymatically active [129, 146].

Fig. 6.

Amino acid sequence comparison of family GH77 amylomaltases from borreliae and Thermus aquaticus. The four amylomaltases from borrelian origin represent different subgroups within the genus Borrelia [122] exhibiting unique mutations in several important active-site positions with respect to a typical bacterial amylomaltase represented by the one from Thermus [129, 145]. The sequences were retrieved from the UniProt database [221] according to their accession numbers succeeding the name of the organism. The alignment was done using the program Clustal-Omega [230] available at the European Bioinformatics Institute’s web-site (http://www.ebi.ac.uk/). The unique borreliae-like positions (Asn228, Lys306 and Gly407) and the catalytic triad (Asp308, Glu355 and Asp408; B. burgdorferi amylomaltase numbering) are signified by yellow/red and blue highlighting, respectively. The positions in amylomaltases from borreliae identical to that from Thermus are shown as dots. The black highlighting in the T. aquaticus amylomaltase means that all four borrelian counterparts have the same residue in those positions. The positions that are signified by black highlighting in amylomaltases from borreliae are identical among them but different from those in the enzyme from Thermus. The seven CSRs known for the entire α-amylase clan GH-H are boxed and marked as CSR-I to CSR-VII with indicated well-accepted secondary structure elements [32]. The individual domains are indicated as a colored lane above the alignment blocks as follows: catalytic (β/α)₈-barrel—red, subdomain B1—cyan, subdomain B2—magenta, subdomain B3—green

The family GH77 offers also other examples that are of interest from the evolutionary point of view, e.g., the DPE2 version of a typical plant DPE1 [42, 134, 135] possessing an insert of ~140 residues between catalytic nucleophile and proton donor [147–152] that have been found also among some bacteria [122], or a large group of additional bacterial amylomaltases represented by the enzyme from Escherichia coli [125] including also the well-characterized amylomaltase from Clostridium glutamicum [131, 153–155] that have at the N-terminus a separate carbohydrate-binding module-(CBM)-like domain [122, 142]. In any case it is evident that with regard to the family GH77 a gradual evolutionary transition has occurred among borreliae that can be characterized from the version of a typical bacterial Thermus-like amylomaltase to the version with progressively mutated functionally important conserved residues [122].

Other examples of interest

The main goal of this review article was to demonstrate some selected examples of remarkable evolutionary phenomena within the α-amylase clan GH-H. There are many other cases that would deserve to be mentioned, including also those seen in the second and smaller α-amylase family GH57. To mention at least a few, the non-exhaustive list could be as follows: (1) close similarity and relatedness between the family GH13 α-amylases from animals and actinomycetes [28, 74, 79] constituting a well-known group of chloride-dependent α-amylases [156, 157]; (2) existence of an intermediary group of α-amylases, classified in the subfamily GH13_36 [48], with a mixed enzyme specificity exhibiting simultaneous relatedness to various α-glucosidases and debranching enzymes [158–163] from the so-called subfamilies of oligo-1,6-glucosidase and neopullulanase [45, 46]; (3) presence of the so-called α-amylase-like homologues in the family GH57 [164–166] strikingly similar to α-amylases, but having incomplete GH57 catalytic machinery—the real α-amylases being more frequent among Archaea, whereas the α-amylase-like homologues are being found rather among Bacteria [164]; (4) the pronounced similarity between the α-amylase families GH57 and GH119 revealed by an in silico study suggesting that the members of the family GH119 share with those of the GH57 catalytic machinery, CSRs and fold of the catalytic domain [167]; (5) observation of a totally novel lineage or subfamily of glucansucrases within the circularly-permuted family GH70 [39–41, 168–170], exhibiting a close homology to the rest of the family but without any circular permutation of the catalytic TIM-barrel [171, 172]; (6) existence of various and evolutionarily independent starch/glycogen-binding domains classified as different CBM families [173–185] in, e.g., not typical amylolytic enzymes and proteins, such as glucan phosphatases—animal laforin and plant SEX-4 [186–197], AMP-activated protein kinase [198–201], genethonin-1 [202, 203], and starch synthases [204–206], glucan water dikinases [207–211] and lytic polysaccharide monooxygenases [212–214]; and (7) the so-called carbohydrate surface binding sites different from the distinct CBMs mentioned above and present in various carbohydrate-active enzymes, well represented just in the α-amylase family GH13 [69, 72, 215–219]—in the particular case of the barley α-amylase low pI isozyme AMY1 named as “a pair of sugar tongs” and a starch granule binding site, respectively [69, 218]. All these particular details make the α-amylases and the remaining enzyme specificities an attractive subject not only for evolutionary studies but also from the practical point of view for the approaches focused on their protein design. Even if known that some living organisms can survive without any sugars [220].

Conclusions

Due to a huge amount of accumulated sequence data, the questions concerning the structure, function and evolution of amylolytic enzymes have really become complicated, especially when they are ambitiously treated in an effort to reach answers to complex questions. This obvious trend can simply be illustrated by a few dozens of the α-amylase family GH13 members when it was created in 1991 versus more than 30 thousand sequences classified in the family nowadays. There is, however, not only the prevalent number of sequences available complicating the situation, i.e., our knowledge, but there is also the necessity to accept the changes in minds of scientists reflecting the continuously widening scope of the α-amylase family. Thus, for example, this family was originally established as a family of starch hydrolases and related enzymes active towards the α-1,4- and α-1,6-glucosidic linkages catalyzing either the hydrolysis (EC 3) or transglycosylation (EC 2). Currently, also other bonds are attacked, e.g., those in trehalose (α-1,1-) and sucrose (α-1,2-), the enzymatic repertoire of the family members has been expanded including the isomerases (EC 5). What might originally be unbelievable, even the non-enzymatic transport proteins (rBAT and 4F2hc), lacking in most cases the GH13 catalytic machinery, have been classified constituting their respective GH13 subfamilies. One of the ways that could lead to a consolidation of the knowledge may be to continue in looking for and revealing the features in amino acid sequences and structures that would reflect the exclusivity of smaller family groups (i.e., subfamilies) of either particular substrate specificity or from the point of view of taxonomy, or both, and confirming their roles experimentally.

Abbreviations

CAZy

Carbohydrate-Active enZymes

CBM

Carbohydrate-binding module

CGTase

Cyclodextrin glucanotransferase

CSR

Conserved sequence region

DPE

Disproportionating enzyme

GH

Glycoside hydrolase

hcHATs

Heavy chains of the heteromeric amino acid transporters