Regular Paper
Crystal Structure and Mutational Studies of Cyanobacterial Branching Enzymes Reveal the Structural Determinants of Reaction Product Specificity
2025 Volume 72 Issue 4 Article ID: 7204105
Details
2025 Volume 72 Issue 4 Article ID: 7204105
Branching enzymes (BEs) are essential for defining the branching patterns of glycogen and starch by catalyzing the formation of α-1,6-glucosidic linkages. While most cyanobacteria accumulate glycogen, some species, such as Crocosphaera subtropica ATCC 51142, produce an insoluble branched α-glucan known as cyanobacterial starch. This strain possesses three BE isozymes: cceBE1, cceBE2, and cceBE3. Our previous studies demonstrated that cceBE1 and cceBE2 share similar enzymatic properties and that a “stopper structure” contributes to their preferential production of short chains with a degree of polymerization (DP) of 6 and 7. In contrast, cceBE3 produces small amounts of short (DP5-12) and long (DP30-40) chains and lacks the amino acid sequence corresponding to the stopper structure. To investigate the role of the stopper structure, we constructed a deletion mutant of cceBE1 lacking the stopper structure and characterized its enzymatic properties. The mutant retained catalytic activity but lost the ability to selectively produce glucan chains with DP6 and 7 (transferred chains), providing direct evidence for the stopper structure's role in regulating product chain length. Furthermore, we determined the crystal structure of cceBE3, confirming the absence of the stopper structure. We also identified a unique structural feature in cceBE3, termed subdomain B, located within the predicted substrate-binding site. Deletion of subdomain B led to increased production of short chains (DP3-7), suggesting its involvement in substrate binding and the determination of product specificity. These findings reveal structural determinants of product specificity in cyanobacterial BEs and offer a strategy for engineering BEs to produce novel starch-based materials.
amylopectin, amylopectin from potato; amylose, synthetic amylose; BE, branching enzyme; cceBE1, BE isozyme 1 from strain 51142; cceBE2, BE isozyme 2 from strain 51142; cceBE3, BE isozyme 3 from strain 51142; DP, degree of polymerization; G6, maltohexaose; GH, glycoside hydrolase; SBS, surface-binding site
Starch and glycogen are complex branched glucose polymers found in Archaeplastida and across all three domains of life, respectively [1, 2, 3]. Starch mainly consists of amylopectin which has a highly ordered branching pattern, whereas glycogen has a random branching pattern. Both polymers have backbones made of α-1,4-linkages, with branches connected by α-1,6-linkages [1]. Cyanobacteria are oxygenic photosynthetic prokaryotes, and nearly all of them accumulate glycogen as a photosynthetic product. However, only a limited number (approximately 10) of diazotrophic unicellular species accumulate cyanobacterial starch instead of glycogen [4, 5], except for Cyanobacterium sp. CLg1, which coaccumulates cyanobacterial starch and glycogen via separate metabolic pathways [2]. Cyanobacterial starch closely resembles starch from terrestrial plants in terms of fine structure and physicochemical properties, such as X-ray diffraction and gelatinization behavior [5]. Cyanobacterial starch-producing strains are valuable research models for understanding starch metabolism, as no other prokaryotes are known to accumulae starch.
The enzyme suite responsible for starch and glycogen biosynthesis is highly conserved and includes ADP-glucose pyrophosphorylase (AGPase, EC 2.7.7.27), starch/glycogen synthase (SS/GS, EC 2.4.1.21), branching enzyme (BE, EC 2.4.1.18), debranching enzymes (isoamylase/GlgX, EC 3.2.1.68; pullulanase, EC 3.2.1.41) [6]. Among these, BEs play a crucial role in determining the branching patterns of starch and glycogen. BEs catalyze intra- and/or inter-transglycosylation reactions by cleaving α-1,4-lined glucan backbone to form a new branch point consisting of α-1,6-linkage [7, 8]. BEs are classified into glycoside hydrolase family 13 (GH13) in the CAZy database (https://www.cazy.org) based on primary structural similarities [9]. They are further subdivided into subfamilies GH13_8 (mainly eukaryotic) and GH13_9 (mainly prokaryotic), with some exceptions [8, 10, 11]. BEs belonging to GH family 57 (GH57) have also been identified in certain archaea and bacteria [12, 13, 14].
Multiple GH13-type BE isozymes, two (BE1 and BE2) or three (BE1, BE2, and BE3), are invariably found in cyanobacterial starch-producing strains [15]. BE1 and BE2 isozymes are phylogenetically related, sharing approximately 60 % amino acid sequence identity (Fig. 1), and predominantly produce glucan chains with degree of polymerization (DP) of 6 and 7 [15, 16, 17]. In contrast, BE3 isozymes exhibit much lower sequence identity (about 20 %; Fig. 1) and generate small amounts of both short (DP5-12) and long (DP30-40) glucan chains [15, 16, 17]. The cyanobacterial starch-producing Crocosphaera subtropica ATCC 51142 has three GH13_9-type BE isozymes (cceBE1, cceBE2, and cceBE3). The crystal structures of cceBE1 in complex with maltooligosaccharides have revealed its active site cleft and five surface-binding sites (SBSs) [18]. SBSs, located far from the active site, are thought to assist in substrate targeting, guiding the substrate to the active site, and enhancing processivity [19, 20, 21, 22]. Based on the structural features of cceBE1, we previously proposed a mechanistic model for the BE reaction [18] (Fig. 2). In this model, the two SBSs (A1 and A2) located within the catalytic domain play essential roles. C chain of donor glucan chain initially binds to SBS A1 at the substrate entrance, and its A chain subsequently enters the subsite + and active-site cleft (subsite −). According to Hizukuri's classification [23], glucan chains in amylopectin are defined as follows: A chains (DP6-12) are unbranched; B1 chains (DP13-24) contain at least one branch point; B2 chains (DP25-36) connect two clusters; B3 chains (DP ≥ 37) connect three clusters; and the C chain carries the reducing end. During the first catalytic step (glycosylation), the donor chain is cleaved into two segments: the chain to be transferred (at subsite −) and the residual chain (at subsite +). The chain to be transferred forms a covalent bond with the catalytic nucleophile, generating a covalent glycosyl-enzyme intermediate, while the residual chain is released from the enzyme. In the second catalytic step (deglycosylation), the acceptor glucan chain binds along a path extending from SBS A1 through the active site to SBS A2 at the substrate exit (subsites + to +ʹ). The chain to be transferred is then linked to the acceptor via an α-1,6-linkage. This model highlights the significance of structural elements near the active site in determining both cleavage and transfer sites, which in turn influence the chain length of the reaction product (Fig. 2).
Scientific names are shown with NCBI accession numbers in parentheses. The cceBE1 and cceBE3 used in this study are indicated by arrowheads. Escherichia coli GlgX, belonging to GH13_11, was used as an outgroup. The tree was constructed using the neighbor-joining method with ClustalW [31] and MEGA [32].
Domain A of wild-type cceBE1 is depicted as a gray ellipse. SBS A1 (substrate entrance) and SBS A2 (substrate exit) are labeled. The active site cleft (subsite −) and the donor/acceptor substrate-binding paths (subsites + and +ʹ) are indicated in dark gray and white, respectively. Glc units bound at the active site cleft (subsite −) and along the donor (subsite +) and acceptor (subsite +ʹ) substrate-binding paths are represented as black and gray hexagons, respectively. Glc units not accommodated in subsites are shown as white hexagons, with the reducing end is represented by a hexagon with a slash. The residual chain (subsite +) is numbered from the branch point (1 to 6), the chain to be transferred (subsite −) is numbered from the active site (1 to 6), and the acceptor chain, which forms a new branch point with the transferred chain, is numbered from the branch point (1 to 10). Glucan chains in the donor substrate, residual chain, and reaction product are labeled according to Hizukuri's classification (see Introduction section).
Our previous studies on BE isozymes, interpreted within this mechanistic model, suggest that the reaction product specificity of BE1 and BE2 isozymes arises from the presence of a stopper structure [15]. The amino acid residues forming the stopper structure (29 residues) are exclusively conserved in the cyanobacterial BE1 and BE2 isozymes (Fig. 3). It has been proposed that the stopper structure interacts with the Glc unit at the non-reducing end of the chain to be transferred, acting as a steric barrier that prevents longer donor chains from accessing the subsite −. Consequently, the stopper structure is thought to promote the selective production of chains with DP6 and 7 (transferred chains) [15, 18]. The distinct reaction product specificity of the BE3 isozymes (see above) may result from the absence of a stopper structure sequence (Fig. 3). However, direct experimental evidence supporting the role of stopper structures is lacking. To validate this hypothesis, structural studies of BE3 isozymes and biochemical characterization of stopper structure deletion mutants of BE1 or BE2 isozymes are required.
The stopper structure (29 residues) and subdomain B (32 residues) are indicated above the alignment. The deleted regions in the cceBE1ΔStopper mutant (20 residues) and the cceBE3Δα-helix mutant (14 residues) are indicated below the alignment. The alignment was generated using ClustalW [31] and ESPript 3.0 [33].
In this study, we determined the crystal structure of cceBE3 and characterized a stopper structure deletion mutant of cceBE1 (cceBE1ΔStopper). The cceBE1ΔStopper mutant retained enzymatic activity comparable to that of the wild-type cceBE1, but its ability to selectively produce DP6 and 7 (transferred chains) was abolished. The structure of cceBE3 lacks a stopper structure and appears capable of accommodating longer glucan chains compared to cceBE1, supporting the functional role of the stopper structure in determining reaction product chain length. Furthermore, we identified a unique subdomain, designated B, located along the path from SBS A1 to the active site (subsite +) in cceBE3. This subdomain is presumed to interfere with substrate-binding. Deletion of α-helix region within subdomain B resulted in increased production of glucan chains with DP3 and 4 (residual chains) and DP6 and 7 (transferred chains), indicating its involvement in substrate interactions. These findings provide a structural basis for understanding the determinants of product specificity in cyanobacterial BEs and offer strategies for engineering BEs to produce novel starch-based materials.
Construction of expression plasmids
The previously constructed expression plasmids pET15b/cceBE1 and pET15b/cceBE3, encoding the wild-type cceBE1 and cceBE3 genes, respectively, were used in this study [17]. Two deletion mutants, designated as ΔStopper structure and Δα-helix, were constructed by inverse PCR to eliminate the respective coding regions. To generate the expression plasmid pET15b/cceBE1ΔStopper, which encodes cceBE1 lacking the stopper structure region, inverse PCR was performed using pET15b/cceBE1 as the template. The following primers were used: sense primer 5ʹ-TCTTCAGCCGGGGCCCGGTTTCTCAGTTACT-3ʹ and antisense primer 5ʹ-CCGGGCCCCGGCTGAAGAACCGTGTAACCA-3ʹ. The expression plasmid pET15b/cceBE3Δα-helix, which carries a truncated version of the cceBE3 gene lacking the α-helix region within subdomain B, was constructed by inverse PCR using pET15b/cceBE3 as the template. The primers used were: sense primer 5ʹ-GAACAAGTTCGCTATTTTGAAGAGCGAGGCT-3ʹ and antisense primer 5ʹ-AAAATAGCGAACTTGTTCGCGATCATTTCT-3ʹ. All PCRs were performed using PrimeSTAR® HS DNA Polymerase (Takara Bio Inc., Shiga, Japan). The PCR products were directly transformed into Escherichia coli DH5α cells, where the plasmids were repaired through in vivo nick repair. For purification by immobilized metal-affinity chromatography, all plasmids were engineered to include an N-terminal His6 tag and a linker (20 amino acids). The nucleotide sequences of the constructed genes were verified using an ABI PRISM 3100xl Genetic Analyzer (Thermo Fisher Scientific Inc., Waltham, MA, USA).
Expression and purification of recombinant BEs in E. coli
The BE genes were expressed in E. coli BL21 (DE3) cells (Merck KGaA, Darmstadt, Germany), and recombinant proteins were purified as described previously [15, 16, 17, 18]. Briefly, E. coli cells were cultured at 37 °C with shaking, and gene expression was induced by adding 0.1 mM isopropyl β-D-thiogalactopyranoside (IPTG), followed by incubation at 15 °C. Cells were harvested, disrupted by sonication, and centrifuged to obtain the crude supernatant. Recombinant BEs were purified from the crude extract using Ni2+-affinity column chromatography. The resulting protein solutions were desalted and concentrated using Amicon Ultra-15 centrifugal filter units (30,000 MWCO; Merck KGaA), and further purified by gel filtration chromatography using a HiLoad 16/600 Superdex 200 pg column (Cytiva, Tokyo, Japan). Protein concentrations were determined using a Bradford Assay Kit (Bio-Rad Laboratories, Inc., Hercules, CA, USA), with bovine serum albumin (Bio-Rad Laboratories, Inc.) as the standard. Protein purity and homogeneity were evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).
Characterization of recombinant BEs
The enzymatic activities of the recombinant BEs were measured using an iodine-staining assay with commercially available substrates, potato amylopectin (Merck KGaA) and synthetic amylose (Ezaki Glico Co., Ltd., Osaka, Japan), as described previously [15, 16]. The chain-length distribution of the reaction products was analyzed using the P/ACE MDQ Carbohydrate System (Beckman Coulter, Inc., Brea, CA, USA), following our previously published method [16].
Crystallographic studies of wild-type cceBE3
Purified wild-type cceBE3 was concentrated to 15 mg/mL, and initial crystallization conditions were manually screened at 20 °C using commercial kits, Crystal Screen and Crystal Screen 2 (Hampton Research Corp., Aliso Viejo, CA, USA) and Wizard Classics 1-4 (Rigaku Reagents, Inc., Woodlands, TX, USA), using the sitting-drop vapor-diffusion method. Initial crystals were obtained under condition No. 29 of the Crystal Screen kit [0.8 M potassium sodium tartrate, 0.1 M HEPES-NaOH (pH 7.5)]. The crystallization conditions were subsequently optimized using the hanging-drop vapor-diffusion method. Crystals suitable for X-ray diffraction were obtained within 1 week under optimized conditions: 0.3 M potassium sodium tartrate, 0.1 M HEPES-NaOH (pH 7.2), and 0.2 M MgCl2. X-ray diffraction data were collected at a wavelength of 1.00000 Å on beamline AR-NW12A at the Photon Factory Advanced Ring (PF-AR), High Energy Accelerator Research Organization (KEK), Tsukuba, Japan. The data were processed with iMOSFLM [24] and scaled using Scala [25], both of which are included in the CCP4 suite [26]. The structure was solved by molecular replacement using MOLREP [27], with the cceBE1 structure (PDB ID: 5GQU) [18] as the search model. Structural refinement was performed using REFMAC5 [28] and manual model building was conducted using Coot [29]. Data collection and refinement statistics are summarized in Table 1. Structural figures were prepared using PyMOL (Schrödinger, LLC, New York, NY, USA).
Table 1. Summary of data collection and refinement statistics.
| Data collection statistics | |
| X-ray source | KEK PF AR-NW12A |
| Detector | PILATUS3 S 2M |
| Wavelength (Å) | 1.00000 |
| Space group | P212121 |
| Unit-cell parameters (Å) | a = 148.05, b = 149.91, c = 162.01 |
| Resolution range (Å) | 88.31-2.50 (2.64-2.50) |
| R merge | 0.075 (0.351) |
| Completeness (%) | 100.0 (100.0) |
| Multiplicity | 12.3 (12.1) |
| Average I/σ(I) | 22.2 (6.7) |
| Unique reflections | 124,975 (18,092) |
| Total reflections | 1,541,305 |
| Refinement statistics | |
| PDB code | 7XSY |
| Resolution | 48.06-2.50 (2.57-2.50) |
| R-factor | 0.200 (0.249) |
| R free-factor | 0.261 (0.328) |
| RMSD from ideal value | |
| Bond length (Å) | 0.007 |
| Bond angle (degrees) | 1.523 |
| Average B factors (Å2) | |
| Protein (Chain A/B/C/D) | 46.7/51.3/50.4/52.4 |
| Water molecules | 39.5 |
| Ramachandran plot (%) | |
| Favored regions (Chain A/B/C/D) | 96.5/96.8/96.8/95.0 |
| Allowed regions (Chain A/B/C/D) | 3.5/3.2/3.2/4.4 |
| Outlier regions (Chain A/B/C/D) | 0.0/0.0/0.0/0.6 |
Values for the highest resolution shell are given in parentheses.
Characterization of cceBE1 stopper structure deletion mutant (cceBE1ΔStopper)
Our previous studies suggested that the stopper structure of cceBE1 acts as a steric barrier and contributes to the selective production of glucan chains with DP6 and 7 (transferred chains) (Fig. 4A). To test this hypothesis, we constructed a cceBE1ΔStopper mutant by deleting the E268-P287 region (Fig. 3). A purified preparation of the recombinant cceBE1ΔStopper mutant was successfully obtained (Fig. S1A; see J. Appl. Glycosci. Web site). Surprisingly, the deletion mutant retained specific activities toward both amylopectin and amylose that were comparable to those of the wild-type cceBE1 (Fig. 5). To investigate the impact of the stopper structure deletion on reaction product specificity, we analyzed the chain-length distribution of the reaction products formed by both wild-type and mutant enzymes. The wild-type cceBE1 utilized glucan chains with DP ≥ 12 as donor substrates, predominantly producing chains with DP6 and 7 (Fig. 6A), consistent with previous findings and the sugar-bound structure (Fig. 4A) [15, 17]. In contrast, the cceBE1ΔStopper mutant produced substantially lower amounts of DP6, 7, and 10, while generating increased amounts of longer chains (DP ≥ 12) (Figs. 6B and C). This shift likely reflects improved accessibility of the longer donor chains to the active site cleft (subsite −) in the absence of the stopper structure (Fig. 6D). Notably, the mutant enzyme still produced a considerable amount of DP6 (Fig. 6B), which may be due to the residual chains at subsite + remaining after cleavage of the donor substrate [15] (Fig. 6D). These findings suggest that deletion of the stopper structure does not impair enzymatic activity or donor substrate binding, but rather alters the length of the chains to be transferred (at subsite −), thereby modifying the reaction product specificity. Taken together, these results support our proposed mechanistic model of the BE reaction and reinforce the hypothesis that the stopper structure plays a key role in the selective production of DP6 and 7 (transferred chains) [15, 18].
(A) Sugar-bound structure of wild-type cceBE1 (PDB: 5GQV). The subsites (−6 to −1), two carbohydrate-binding sites in CBM48, and five surface-binding sites (SBSs) are labeled. The reducing end of the bound sugars is marked with a φ. The presence of the stopper structure and the absence of subdomain B are indicated with dashed circles. (B) Overall structure of the ligand-free form of wild-type cceBE3 (PDB: 7XSY). The active site is highlighted in magenta. The absence of the stopper structure and the presence of subdomain B are indicated with dashed circles. The α-helix and associated loops, including two 310-helices and two β-strands, in subdomain B are colored orange and yellow, respectively. (C) Surface representation of wild-type cceBE3. The structure was superimposed onto the sugar-bound structure of wild-type cceBE1 (PDB: 5GQV), with the bound sugars shown as white ball-and-stick models. The route of longer glucan chain is indicated with dashed arrow. (D) Substrate-binding paths in wild-type cceBE3. The sugar-binding model was generated as in (C). Bound sugars at the subsites (−6 to −1), SBS A1 (substrate entrance), and SBS A2 (substrate exit) are shown. The donor and acceptor substrate-binding paths are indicated by white and black arrows, respectively. Subsites −, +, and +ʹ are labeled. (E) Substrate-binding paths in wild-type cceBE1. All structural figures were generated using PyMOL (Schrödinger, LLC).
Gray and white bars represent the specific activities of wild-type and mutant enzymes, respectively. The inset shows a close-up view of the values for wild-type cceBE3 and cceBE3Δα-helix. Values are expressed as the mean ± SD from three independent experiments.
Structure of wild-type cceBE3
The crystal structure of the ligand-free form of wild-type cceBE3 was successfully determined at 2.50 Å resolution (Table 1 and Fig. 4B); however, the structure of cceBE3 in complex with a maltooligosaccharide could not be solved. The crystal structure of cceBE3 contained four molecules (chains A, B, C, and D) in the asymmetric unit, whereas gel-filtration chromatography indicated that the purified wild-type cceBE3 exists and functions as a monomer in solution [17]. In cereals, BEs exert their function by associating with other starch biosynthetic enzymes (e.g., SSs); however, whether BEs undergo self-assembly into oligomeric complexes remains unclear [7]. The presence of four molecules in the asymmetric unit does not necessarily reflect the biologically relevant oligomeric state. The root-mean-square deviations among the four molecules were less than 0.298 Å, indicating that they adopt nearly identical conformations. Therefore, the following structural descriptions are based primarily on molecule A. The ligand-free cceBE3 structure consisted of three domains: a carbohydrate-binding module family 48 (CBM48) at the N-terminus, domain A (the catalytic domain), and domain C at the C-terminus (Fig. 4B). These three domains were conserved among the BE2 and BE3 isozymes, whereas BE1 isozymes consistently contained an additional N-terminal domain, referred to as domain N (Fig. 4A and Fig. S1B; see J. Appl. Glycosci. Web site). Within domain A, three catalytically important residues characteristic of GH13 enzymes are conserved, namely Asp294 (nucleophile), Glu374 (acid/base catalyst), and Asp442 (second aspartate/transition-state stabilizer). As suggested by amino acid sequence analysis and the predicted structural model in our previous study [15], the crystal structure of cceBE3 confirmed the absence of the stopper structure (Fig. 4B). To investigate glucan chain recognition during the transfer reaction, we superimposed a maltohexaose (G6) molecule originally bound to the active-site cleft of wild-type cceBE1 (PDB ID: 5GQV) (Fig. 4A) [18], onto the cceBE3 structure (Fig. 4C). The G6 molecule is mostly accommodated within the active-site cleft (subsite −) of cceBE3. However, no specific interaction was observed between the 4-OH group of the non-reducing end Glc residue and the enzyme. This structural observation provides a mechanistic explanation for the distinct product specificity of the BE3 isozymes. Specifically, the absence of a stopper structure likely accounted for the inability of cceBE3 to selectively produce short glucan chains with DP6 and 7, instead enabling the production of longer glucan chains (transferred chains) (Figs. 4C and D).
Intriguingly, a unique structural element consisting of 32 residues, including one α-helix, two 310 helices, two β-strands, and associated loops, was observed at the bottom of the cceBE3 structure. This element is inserted between strand β4 and helix α4 of domain A (Figs. 4B and C). GH13 α-amylases typically contain a small domain B inserted between strand β3 and helix α3 of the domain A [30]. In analogy to this structural feature, we hereafter refer to the inserted region in cceBE3 as subdomain B, although its position differs from that in α-amylases. This subdomain is located along the substrate-binding path (subsite +), positioned between SBS A1 and the active site (Fig. 4D). This path was proposed in our previous study as part of a mechanistic model of the BE reaction based on cceBE1 structures (Figs. 2 and 4E) [18]. Subdomain B is highly conserved among BE3 isozymes, but absent in all other BEs (Fig. 3). Its presence appears to sterically hinder the binding of both donor and acceptor substrates to subsite +, which may explain the significantly lower specific activity of BE3 isozymes compared to BE1 and BE2 isozymes (Fig. 5) [15, 16, 17]. Furthermore, this structural feature may facilitate the accommodation of longer B1, B2, and B3 chains (Hizukuri's classification of glucan chains, see Introduction section) within the active site cleft (subsite −) through interactions with a branch point via the α-helix portion (residues E320-D333) of subdomain B (subsite +) (Fig. 7D). The most abundant product in the chain-length profile of wild-type cceBE3 was DP8 (Fig. 7A). Assuming that the major residual chain at subsite + of wild-type cceBE3 is DP8 (Fig. 7D), A chains may be too short to be properly accommodated in the active site cleft (subsite −). This hypothesis provides a plausible explanation for the ability of BE3 isozymes to produce long glucan chains (transferred chains) (Fig. 7A). To test this hypothesis, further mutational studies using a subdomain B deletion mutant are needed.
Characterization of α-helix in subdomain B deletion mutant of cceBE3 (cceBE3Δα-helix)
The α-helix in subdomain B deletion mutant (cceBE3Δα-helix), which lacks the α-helix region (E320-D333; 14 residues) (Fig. 3), was successfully constructed. Although partial misfolding cannot be completely ruled out, the recombinant protein was expressed as a soluble protein (Fig. S1A; see J. Appl. Glycosci. Web site) and retained detectable enzymatic activity. The specific activity of the cceBE3Δα-helix mutant was reduced by 66.7- to 86.5-fold compared to wild-type cceBE3 (Fig. 5), indicating that the α-helix region within subdomain B contributes to substrate binding. This result supports our model in which the path between SBS A1 and the active site serves as a substrate-binding region (Figs. 4D and 7D). Unexpectedly, the reaction product specificity of the cceBE3Δα-helix mutant was drastically altered. The production of glucan chains with DP3-7 and 10 increased, whereas that of DP8, 9, and ≥ 12 decreased (Figs. 7B and C). Notably, a substantial increase was observed in the production of DP3, 4, 6, and 7. Based on these data, we hypothesized that glucan chains with DP3 and 4 represent the major residual chains, whereas those with DP6 and 7 are predominantly transferred (Fig. 7D). The deletion of the α-helix region likely disrupts the enzyme's ability to recognize the branch point of the donor substrate, resulting in a shift in substrate-binding mode that favors the utilization of shorter glucan chains (A chains) as donor substrates (Fig. 7D). As a consequence, longer glucan chains (B chains) could no longer serve as donor substrates in the mutant. This model explains the altered chain-length profile observed in the mutant (Figs. 7B-D). Collectively, these findings suggest that the unique reaction product specificity of cceBE3 is governed by α-helix region within subdomain B, which plays a crucial role in the selection and binding of longer donor substrates (B chains).
In this study, we determined the crystal structure of wild-type cceBE3 and elucidated the functional roles of the stopper structure in cceBE1 and the α-helix region within subdomain B of cceBE3. Characterization of the cceBE1ΔStopper mutant revealed that the stopper structure at subsite − is a critical structural determinant responsible for the selective production of glucan chains with DP6 and 7 (transferred chains). Moreover, analysis of the cceBE3Δα-helix mutant demonstrated that the α-helix region within subdomain B plays a crucial role in recognizing the lengths of donor substrates and residual chains, with the region between SBS A1 and the active site functioning as a substrate-binding path (subsite +). These findings strongly support our previously proposed mechanistic model of the BE reaction (Fig. 2) [15, 18] and provide deeper insights into the distinct functional roles of BE isozymes in cyanobacterial starch biosynthesis. Engineering the stopper structure may allow for fine-tuning of the reaction product specificity without significantly compromising enzymatic activity, offering a promising strategy for the development of novel starch-based materials.
The authors declare no conflicts of interest.
The authors would like to thank Drs. Naoko Fujita, Naoko Crofts, and Satoko Miura for their valuable discussions. We are also grateful to Nagase Viita Co., Ltd. for providing Pseudomonas isoamylase. The authors thank Editage (www.editage.jp) for the English language editing. This work was supported by JSPS KAKENHI grants 15K18685 (to R. S.), 16K07467 (to E. S.), and 18K06135 (to R. S.).
