Impact of Starch Binding Domain Fusion on Activities and Starch Product Structure of 4-α-Glucanotransferase

A broad range of enzymes are used to modify starch for various applications. Here, a thermophilic 4-α-glucanotransferase from Thermoproteus uzoniensis (TuαGT) is engineered by N-terminal fusion of the starch binding domains (SBDs) of carbohydrate binding module family 20 (CBM20) to enhance its affinity for granular starch. The SBDs are N-terminal tandem domains (SBDSt1 and SBDSt2) from Solanum tuberosum disproportionating enzyme 2 (StDPE2) and the C-terminal domain (SBDGA) of glucoamylase from Aspergillus niger (AnGA). In silico analysis of CBM20s revealed that SBDGA and copies one and two of GH77 DPE2s belong to well separated clusters in the evolutionary tree; the second copies being more closely related to non-CAZyme CBM20s. The activity of SBD-TuαGT fusions increased 1.2–2.4-fold on amylose and decreased 3–9 fold on maltotriose compared with TuαGT. The fusions showed similar disproportionation activity on gelatinised normal maize starch (NMS). Notably, hydrolytic activity was 1.3–1.7-fold elevated for the fusions leading to a reduced molecule weight and higher α-1,6/α-1,4-linkage ratio of the modified starch. Notably, SBDGA-TuαGT and-SBDSt2-TuαGT showed Kd of 0.7 and 1.5 mg/mL for waxy maize starch (WMS) granules, whereas TuαGT and SBDSt1-TuαGT had 3–5-fold lower affinity. SBDSt2 contributed more than SBDSt1 to activity, substrate binding, and the stability of TuαGT fusions.


Bioinformatics Analysis
In order to put the three above-mentioned experimentally fused SBDSt1, SBDSt2, and SBDGA into the overall context of the CBM20 family, 65 different starch hydrolases and related enzymes were selected for in silico analysis ( Table 1). The emphasis was mainly on GH77 DPE2s, both from Eukaryota (including the StDPE2) and Bacteria, known to contain two recognizable CBM20s [16]. The set to be analysed was completed by various wellknown CBM20s from amylolytic enzymes classified into several CAZy families (including AnGA) as well as several non-CAZymes, such as phosphoglucan, water dikinase (GWD3), laforin, genethonin-1, etc. [5,[16][17][18][19].     Figure 2 and Figure S1.
From the 65 selected enzymes, it was possible to sample 87 CBM20 sequences (see Table 1 for details). It is worth mentioning that although there was a stretch in almost each DPE2 sequence (regardless the bacterial or eukaryotic origin) for two CBM20 copies at the N-terminus, only those not lacking most of the known CBM20 functionally important binding site residues [8,9,12] were taken into the analysis. Interestingly-based on a detailed inspection of their amino acid sequences-the hypothetical DPE2s from Linum tenue (Gen-Bank Acc. No.: CAI0439830.1) and Ricinus communis (UniProt Acc. No.: B9SCF0) obviously contain only one CBM20 copy (data not shown). It is of note that of the two potential starch binding sites of CBM20, only starch binding site one, being formed by Trp543, Lys578, and Trp590 (GH15 A. niger glucoamylase numbering [8]), is well conserved ( Figure S1), whereas residues forming starch binding site two may vary [5], as evidenced by the structural complexes of CBM20s from GH15 A. niger glucoamylase with cyclodextrin (Tyr527, Tyr556 and Trp563) [8] and GH13_2 Bacillus circulans cyclodextrin glucanotransferase with maltose (Tyr633 and Trp636) [19]-having only the tryptophan (Trp563 vs Trp 636) conserved ( Figure S1). Of the SBD St1 , SBD St2 , and SBD GA used in the present study, only SBD GA from GH15 A. niger glucoamylase, that possesses all the key residues involved in binding ( Figure S1), was previously demonstrated to bind starch [8]. SBD St1 and SBD St2 each lack one of the conserved residues at starch binding site one-the SBD St1 lysine (Lys578; A. niger GH15 CBM20 numbering) and the SBD St2 tryptophan (Trp590)-and only the tryptophan (Trp563) of starch binding site two is conserved in both; however, SBD St1 might have a stronger ability to bind since it has a tryptophan corresponding to Tyr527 at binding site two ( Figure S1).
The evolutionary tree ( Figure 2), constructed from the sequence alignment, illustrated several facts: (i) each of the two CBM20 copies from GH77 DPE2s forms its own cluster; (ii) all CBM20s from other CAZymes cluster together (including SBD GA of AnGA; cluster B) and separately from both groups covering the two CBM20 copies of GH77; (iii) the second CBM20 copy of GH77 DPE2s (including SBD St2 of StDPE2; cluster D) exhibits a closer relatedness to CBM20s from non-CAZymes (such as GWD3, laforin, genethonin-1, etc.; cluster C) than to those from other CAZyme families (cluster B); and (iv) the clade of the first CBM20 copy of GH77 DPE2s (including SBD St1 of StDPE2, cyan in Figure 2) covers also the second and the third CBM20 copies from laforins from Cyanidioschyzon merolae and Chondrus crispus, respectively, [18] (brown clade in cluster A, Figure 2) as well as the CBM20 from the four-domain GH13_2 cyclodextrin glucanotransferase from Nostoc sp. PC9229 [20] (green in cluster A, Figure 2). The results from the bioinformatics analysis thus indicate that the three CBM20s studied here, i.e., SBD St1 , SBD St2 , and SBD GA , are positioned in three different clusters of the evolutionary tree ( Figure 2) and may confer the parental enzyme TuαGT distinctly different biochemical properties by the fusion.

Biochemical Properties of TuαGT and SBD-TuαGT Fusions
The produced TuαGT, SBD St1 -TuαGT, SBD St2 -TuαGT, and SBD GA -TuαGT migrated in SDS-PAGE as single protein bands estimated to 56, 68, 67, and 69 kDa ( Figure 3A), respectively, in agreement with the theoretical values (see Section 3.5). The optimal reaction temperature and pH for the maltotriose disproportionation activity were around 70 • C and 7.0 for the different forms of TuαGT ( Figure 3B,D). However, SBD GA -TuαGT had a lower temperature optimum of 60 • C ( Figure 3B). This is in good agreement with previously reported pH and temperature optima for the total activity on amylose and maltose of TuαGT at 6.0 and 75 • C [13]. TuαGT was nearly 100% active at 80 • C, indicating it is a thermophilic enzyme, which also showed significantly reduced activity at <60 • C. Notably, all three SBD-TuαGT fusions were relatively less active than TuαGT at >70 • C, but more active at <60 • C ( Figure 3B). The improved affinity to starch of the SBD-fusions (see Section 2.4) may contribute to their relatively higher activity than the parent enzyme TuαGT at <60 • C, whereas the lower relative activity of the fusions at >70 • C may stem from their poorer thermostability as illustrated by the time progress for the loss of activity at 50 • C ( Figure 3C). Notably, after 20 h at 50 • C, the parent TuαGT maintained~35% activity. However, all SBD-TuαGT fusions lost more activity than TuαGT during the first 5 h at 50 • C and SBD St1 -TuαGT and SBD GA -TuαGT retained only about 20% activity after 8 h, whereas SBD St2 -TuαGT kept remarkably~65% of its activity after 20 h ( Figure 3C). Improved thermostability was previously found by N-terminal fusion of a CBM1 to β-mannanase from Aspergillus usamii YL-01-78 (reAuMan5A-CBM), having a temperature optimum at 75 • C compared with 70 • C for wild-type (reAuMan5A), indicating a stabilizing effect of the CBM1 on the CD [21]. In another study, Wang et al. [22] fused five different CBMs (of families CBM2, 3, 11, and 30) to the C-terminus of cis-epoxysuccinic acid hydrolase (CESH) and found a 5-times higher half-life for the CBM30-CESH than of wild-type CESH at 30 • C.
Molecules 2023, 28, x FOR PEER REVIEW 7 of 19 merolae and Chondrus crispus, respectively, [18] (brown clade in cluster A, Figure 2) as well as the CBM20 from the four-domain GH13_2 cyclodextrin glucanotransferase from Nostoc sp. PC9229 [20] (green in cluster A, Figure 2). The results from the bioinformatics analysis thus indicate that the three CBM20s studied here, i.e., SBDSt1, SBDSt2, and SBDGA, are positioned in three different clusters of the evolutionary tree ( Figure 2) and may confer the parental enzyme TuαGT distinctly different biochemical properties by the fusion.

Figure 2.
Phylogenetic tree of CBM20s with focus on GH77 DPE2s. The tree is based on the alignment of entire CBM20 sequences ( Figure S1). The labels of protein sources consist of the name of the organism, letter "A", "B", or "E" for the archaeal, bacterial, and eukaryotic origin, respectively, CAZy family affiliation (if any), enzyme abbreviated name (for details, see Table 1), and the UniProt accession number. If there are more CBM20 copies in a single protein, the copies in the order of their appearance in the sequence are also indicated by the relevant number "1", "2", and "3" (at the end of the protein label). The three CBM20 domains, two from GH77 Solanum tuberosum DPE2 and one from GH15 Aspergillus niger glucoamylase, studied in the present work, are marked by an asterisk.

Biochemical Properties of TuαGT and SBD-TuαGT Fusions
The produced TuαGT, SBDSt1-TuαGT, SBDSt2-TuαGT, and SBDGA-TuαGT migrated in SDS-PAGE as single protein bands estimated to 56, 68, 67, and 69 kDa ( Figure 3A), respectively, in agreement with the theoretical values (see Section 3.5). The optimal reaction temperature and pH for the maltotriose disproportionation activity were around 70 °C and  Figure S1). The labels of protein sources consist of the name of the organism, letter "A", "B", or "E" for the archaeal, bacterial, and eukaryotic origin, respectively, CAZy family affiliation (if any), enzyme abbreviated name (for details, see Table 1), and the UniProt accession number. If there are more CBM20 copies in a single protein, the copies in the order of their appearance in the sequence are also indicated by the relevant number "1", "2", and "3" (at the end of the protein label). The three CBM20 domains, two from GH77 Solanum tuberosum DPE2 and one from GH15 Aspergillus niger glucoamylase, studied in the present work, are marked by an asterisk.

Adsorption and Enzyme Kinetic Parameters
The binding capacity to WMS granules was increased for all three SBD-TuαGT fusions, revealing that the SBD domains were functional and fulfilling the purpose ( Figure 4). Overall, SBD GA -TuαGT had an almost 5 times higher binding capacity (B max , Figure 4) and 10 times stronger affinity (K d = 0.7 mg/mL) than TuαGT (K d = 7.2 mg/mL). While SBD St1 -TuαGT and SBD St2 -TuαGT both had an essentially 3 times higher binding capacity to WMS granules than TuαGT, their affinity was quite similar and 5-fold larger, respectively, than of TuαGT ( Figure 4). This agrees with SBD St1 lacking the lysine (Lys578, AnGA numbering) and SBD St2 missing one of the two tryptophans (Trp590, AnGA numbering) at starch binding site one, respectively, compared with SBD GA (see Section 2.2; Figure S1). Notably, the positive effect of SBD St2 on binding was larger than of SBD St1 even though SBD St2 misses a tryptophan at binding site one, indicating that other features of these SBDs contribute to their binding determinants for WMS granules. This may likely include differences at the larger and more flexible binding site two, which is claimed for SBD GA to be the tighter binding of the two sites [8,9]. Until now, there has been no report of different functions of the two SBDs arranged in tandem in StDPE2 or in other DPE2 enzymes.
Notably, the positive effect of SBDSt2 on binding was larger than of SBDSt1 even though SBDSt2 misses a tryptophan at binding site one, indicating that other features of these SBDs contribute to their binding determinants for WMS granules. This may likely include differences at the larger and more flexible binding site two, which is claimed for SBDGA to be the tighter binding of the two sites [8,9]. Until now, there has been no report of different functions of the two SBDs arranged in tandem in StDPE2 or in other DPE2 enzymes. The fusion of SBDs to TuαGT also influenced the enzymatic activity. Thus, the maltotriose disproportionation was reduced, SBDSt2-TuαGT and SBDGA-TuαGT having slightly lower Km than TuαGT, but 4-fold lower kcat, and yielding 3-fold lower catalytic efficiency (kcat/Km) for these two fusion enzymes. Notably, kcat/Km for SBDSt1-TuαGT was 15-times reduced compared with TuαGT, due to a doubled Km and an almost 9-fold lower kcat (Table  2). By contrast, using amylose as a substrate, the SBD-fusion improved activity and kinetic parameters somewhat (Table 2). Thus, the similar Km and higher kcat of SBDSt2-TuαGT more than doubled the catalytic efficiency compared with TuαGT, whereas the overall outcome for SBDSt1-TuαGT and SBDGA-TuαGT was essentially the same catalytic efficiency as of the parent enzyme. Overall, the kinetic analyses indicated that the SBD-fusion hampered the The fusion of SBDs to TuαGT also influenced the enzymatic activity. Thus, the maltotriose disproportionation was reduced, SBD St2 -TuαGT and SBD GA -TuαGT having slightly lower K m than TuαGT, but 4-fold lower k cat , and yielding 3-fold lower catalytic efficiency (k cat /K m ) for these two fusion enzymes. Notably, k cat /K m for SBD St1 -TuαGT was 15-times reduced compared with TuαGT, due to a doubled K m and an almost 9-fold lower k cat ( Table 2). By contrast, using amylose as a substrate, the SBD-fusion improved activity and kinetic parameters somewhat (Table 2). Thus, the similar K m and higher k cat of SBD St2 -TuαGT more than doubled the catalytic efficiency compared with TuαGT, whereas the overall outcome for SBD St1 -TuαGT and SBD GA -TuαGT was essentially the same catalytic efficiency as of the parent enzyme. Overall, the kinetic analyses indicated that the SBD-fusion hampered the action of TuαGT on the oligosaccharide (maltotriose), but could improve it on the polysaccharide (amylose). Similarly, fusion of the SBD GA to barley α-amylase, albeit via the much longer natural linker from A. niger glucoamylase (AnGA), showed no adverse effect of the SBD on the active site integrity, as it did not change activity for soluble starch [23]. The improved catalytic efficiency for SBD St2 -TuαGT towards amylose may be caused by favourable polysaccharide binding to SBD St2 , increasing the local substrate concentration and perhaps also directing the substrate chain to the active site on the CD.

Hydrolysis and Cyclization Activities on Different Substrates
To gain insight into the modes of action of the SBD-TuαGT fusions on starch, the hydrolysis and cyclization activities were determined using different substrates (Table 3). SBD St1 -TuαGT had 1.3-1.7-fold higher hydrolytic activity on amylose and gelatinised starch and 1.5-fold higher cyclization activity on amylose than the TuαGT parent enzyme. Similarly, SBD St2 -TuαGT showed 1.5-1.7-fold increased hydrolysis of gelatinised starch, but more moderate 1.3-fold and 1.2-fold increased hydrolytic and cyclization activities, respectively, on amylose. As a glucanotransferase, it is not expected to show increased hydrolysis by SBD-fusion. However, from an industrial viewpoint, a small increase in hydrolytic activity can help to decrease the viscosity of gelatinised starch, which will also facilitate the TuαGT disproportionation reaction. Notably, for SBD GA -TuαGT containing an SBD that originates from the family GH15 of glucoamylases and not from the family GH77 of 4-α-glucanotransferases, to which TuαGT belongs, the hydrolysis and cyclization activities were both essentially the same as for the parent enzyme, except for a slight increase in hydrolysis of gelatinised waxy maize starch (WMS) ( Table 3). We speculate that, perhaps, the domain architecture matters and the naturally N-terminally placed SBDs from the StDPE2 of the family GH77, which constitutes glycoside hydrolase clan H together with GH13 and GH70 [1], are able to provide support in the different GH77 4-α-glucanotransferase reactions as opposed to the naturally C-terminally placed SBD GA connected via a long O-glycosylated linker to the CD of glucoamylase of the family GH15 that acts in an exo-manner on non-reducing ends of malto-oligosaccharides and α-glucans catalysing release of glucose [24].

Structure Analysis of Modified NMS
The modification of maize starch both by TuαGT and the SBD-TuαGT fusions significantly affected its structural properties. Chain length distribution (CLD) of NMS and modified NMS ( Figure 5A) and the percentage of A-chains as well as of B 1 -, B 2 -, and B 3 -chains (Table 4) showed that all NMS starches treated by TuαGT and its SBD-fusions, to different degrees, contained significantly fewer of the short A-chains and more of the longer B 1 -, B 2 -, and B 3 -chains. Still, only minor differences appeared for the CLD in starches modified by the TuαGT parent compared with SBD-TuαGT fusions ( Figure 5A). Previous studies on tapioca starch similarly indicated that exterior chains of amylopectin were elongated by TuαGT [13].

Bioinformatics Analysis of CBM20
In total, 87 CBM20 domains from 65 different amylolytic and related enzymes were collected (Table 1) based on previous studies focused on GH77 DPE2s and different starch-binding domain CBM families [5,[16][17][18][19]. All sequences were retrieved from Gen-Bank (https://www.ncbi.nlm.nih.gov/genbank/, accessed on 23 December 2022; [26]) and/or UniProt (https://www.uniprot.org/, accessed on 23 December 2022) [27]) sequence databases. For DPE2s selected from various bacteria and eukaryotes, the number of CBM20 copies and their borders in respective sequences were taken from UniProt [27] and complemented by data available from the literature [5,[16][17][18]; questionable cases were also verified in the InterPro database (https://www.ebi.ac.uk/interpro/, accessed on 23 December 2022 [28]). Although each studied DPE2 could eventually contain two CBM20 copies in tandem at their N-terminus, putative CBM20 copies that lacked most of the functionally important binding site residues were not considered (Table 1). For CAZymes, the appropriate CAZy classification has been checked against the CAZy database (http://www.cazy.org/, accessed on 23 December 2022; [1]) and published data [5,[16][17][18][19]. Sequences were aligned using the program Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/, accessed on 23 December 2022; [29]) and the alignment was confirmed by comparison of three-dimensional structures of selected CBM20s: (i) two experimentally determined structures from Aspergillus niger GH15 The molecular weight distribution of NMS before and after enzyme treatment was analysed by SEC-MALLS-RI ( Figure 5B). Before modification, typical amylopectin (peak 1) and amylose (peak 2) molecules were observed in NMS by SEC. However, after the enzyme modification, three peaks were observed, namely the peaks one and two as well as a distinct later eluting peak three of smaller polysaccharide chains. Furthermore, a later elution of peak one from all modified starch samples indicated that amylopectin has a reduced molecular weight and was less well resolved from peak two than found for unmodified NMS. The newly appearing prominent peak three of smaller molecules may contain large-ring cyclodextrins (LR-CDs) produced in cyclization reactions [25] as well as polysaccharide hydrolysis products.
To further understand the reaction of TuαGT and the SBD-TuαGT fusions, the α-1,6/α-1,4-linkage ratio that indicates the degree of branching, was determined for the modified starches by using 1 H-NMR ( Figure 5C). NMS modified by TuαGT and SBD GA -TuαGT showed a slight increase in the α-1,6/α-1,4-linkage ratio from 3.76 for unmodified to 3.84 and 3.88%, respectively, after modification, whereas treatment by SBD St1 -TuαGT and SBD St2 -TuαGT increased the ratio to 4.13 and 4.08%, respectively. As TuαGT can catalyze hydrolysis, disproportionation, cyclization, and coupling, which all involve α-1,4-linkages, the increase in the α-1,6/α-1,4-linkage ratio can reflect the level of hydrolysis, in which α-1,4 linkages are lost and not generated, in agreement with the two fusions with SBD St1 and SBD St2 , i.e., the SBDs from StDPE2 belonging to the family GH77, showing an increased degree of hydrolysis of gelatinised NMS compared with TuαGT ( Figure 5B; Table 3).

Disproportionation
The disproportionation activity of TuαGT and SBD-TuαGT fusions was determined as reported [13] by incubating 1% (19.8 mM) maltotriose in 900 µL assay buffer (see Section 3.6.1) with 100 µL enzyme (10 nM, final concentration) at 75 • C for 1 h. The reaction was terminated (99 • C, 15 min) and the released glucose was quantified using the GOPOD assay (D-Glucose Assay Kit, Megazyme) with glucose (0-1000 µM) as standard [37]. One unit of disproportionation activity was defined as the amount of enzyme releasing 1 µmol/min glucose under the above conditions.

Hydrolysis
The hydrolytic activity of TuαGT and the SBD-TuαGT fusions was determined by incubating 2 mg/mL amylose in 900 µL assay buffer (see Section 3.6.1) with 100 µL enzyme (20 µM, final concentration) at 70 • C for 1 h [38]. Hydrolytic activity towards 25 mg/mL NMS (gelatinised at 99 • C, 30 min, 1100 rpm, and cooled to 70 • C before the assay) was determined by addition of enzyme (2 µM, final concentration) and incubated (70 • C, 1 h). The reaction was stopped by the PAHBAH reagent (1:1, v:v), heating (95 • C, 10 min) [39] and the absorbance was measured at 405 nm after cooling. One unit of activity was defined as the amount of enzyme releasing 1 µmol/min reducing sugar under the above conditions. Glucose (0-1000 µM) was used for the standard curve.

Cyclization
The cyclization activity of TuαGT and SBD-TuαGT fusions was determined by incubating 2 mg/mL amylose in 900 µL assay buffer (see Section 3.6.1) with 100 µL enzyme (20 µM, final concentration) at 70 • C for 1 h [40]. The reaction was terminated (99 • C, 15 min), and 0.24 U β-amylase was added and incubated at 40 • C for 10 h to degrade remaining amylose. The reaction was stopped by adding the PAHBAH reagent (1:1, v:v) and the absorbance was measured at 405 nm (as in Section 3.6.3). The amount of formed cycloamylose was determined by the difference of maltose released by β-amylase from untreated amylose and from amylose treated with TuαGT and SBD-TuαGT fusions. One unit of cyclization activity was defined as the amount of enzyme leading to release of 1 µmol less maltose per min under the above conditions using maltose (0-1000 µM) for the standard curve.

Effect of pH and Temperature on Activity
The pH optimum was determined at the optimum temperature 70 • C of TuαGT using the disproportionation activity assay (see Section 3.6.2) in universal buffer (20 mM MES, 20 mM Hepes, 150 mM NaCl, pH 4.0-9.0) [41]. The temperature optimum in the range of 50-90 • C was determined at the optimum pH 7.0 of TuαGT in the above buffer. To assess thermostability, TuαGT and SBD-TuαGT fusions (100 nM) were incubated at 50 • C and pH 7.0 (50 mM Hepes buffer, 150 mM NaCl) and the residual enzyme activity was measured during 8 h with 1 h intervals. The activity before incubation defined 100% stability.

Adsorption to Starch Granules
The binding capacity of TuαGT and SBD-TuαGT fusions on WMS granules at 25 • C was determined under the same conditions as used for the activity assay (see Section 3.6.1) by adding enzyme (200 nM, final concentration) to different WMS concentrations from 0.5 to 75 mg/mL [42]. After 10 min the mixtures were centrifuged (10,000× g, 5 min) and 100 µL supernatant was added to 100 µL 2.5-fold diluted protein assay dye reagent (Bio-Rad). The enzyme concentration was determined from the ratio of absorbance values at 590 over 450 nm using TuαGT and SBD-TuαGT (0-1.0 µM) as standards. The Langmuir isotherm (Equation (1)) is a commonly used model for analysis of molecular binding and was fitted to the results using GraphPad Prism 6 (GraphPad Software Inc.), where K d is the dissociation constant, Γ is the bound protein concentration, and B max is the (apparent) saturation coverage.

Preparation of Modified Maize Starch (MMS)
Enzymatic modification of NMS was performed essentially as reported [13]. Starch (6%, w/v) was suspended in activity assay buffer (see Section 3.6.1) and gelatinised (99 • C, 30 min, 1100 rpm). The modification was carried out by 1 µmol TuαGT or SBD-TuαGT fusions per 1 g starch at 70 • C for 8 h, and terminated by heating (99 • C, 30 min). The modified starch was precipitated by three volumes of ethanol overnight and isolated by centrifugation (4000× g, 10 min). The precipitated starch was kept overnight at −80 • C and freeze-dried for further analysis.

Molecular Weight Distribution
Size exclusion chromatography with multi-angle laser light scattering-refractive index detector (SEC-MALLS-RI) was used to analyse the molecular weight of starch samples [43]. Dry starch (5 mg/mL) was suspended in a mixture of DMSO and MilliQ water (9:1, v/v) and gelatinised on a boiling water bath (1 h, shaking every 10 min) until the solution was clear and free of floc. The gelatinised starch was incubated (30 • C, 250 rpm, 48 h) to disrupt remaining starch particles. The samples were re-boiled and filtrated through a 0.45 µm filter. Filtrate (100 µL) was injected on a tandem column (Ohpak SB-804 HQ, Ohpak SB-806 HQ) using 0.1 M NaNO 3 (in 0.02% NaN 3 ) as mobile phase at a flow rate of 0.6 mL/min with the column temperature set at 50 • C. Data obtained from the MALLS and RI detectors were analysed by ASTRA software version 5.3.4 (Wyatt Technologies).

Chain Length Distribution
High performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) was used to analyse the chain length distribution of NMS before and after enzyme modification. Starch (5 mg/mL, dry solid (w/v)) was suspended in 50 mM sodium acetate, pH 4.5, followed by gelatinisation (99 • C, 30 min). The gelatinised starch was debranched by incubation with 0.18 U pullulanase per 5 mg starch at 42 • C for 12 h and centrifuged (10,000× g, 10 min). The supernatant was analysed by HPAEC-PAD [44].

Conclusions
In the present work, three phylogenetically diverse SBDs, two from StDPE2 and one from AnGA, fused one by one via an 18-residues linker to the N-terminus of the thermophilic 4-α-glucotransferase (TuαGT), conferred the TuαGT with altered distinct substrate binding and activity characteristics. The bioinformatics analysis shows the distant relationship between SBD St1 , SBD St2 , and SBD GA each found in well-separated clusters of the evolutionary tree and sharing this position with close homologues, i.e., copies one and two of GH77 DPE2s and SBDs from various CAZymes. Relative to the parent enzyme TuαGT, the SBD St2 -fusion had improved thermostability after 5 h of thermal treatment and also doubled the disproportionation activity on amylose. By contrast, all three SBD-fusions decreased the disproportionation activity using maltotriose as substrate. The SBD GA -fusion resulted in the highest binding affinity and binding capacity on starch granules, presumably reflecting the superior function of the two binding sites in this SBD containing all of the canonical aromatic residues. The structural analysis of starch before and after modification by TuαGT and the three SBD-fusion enzymes indicated that the fusion with SBD St1 and SBD St2 enhanced hydrolysis the most, along with their highest cyclization activity, and a slightly higher loss of the short A chains and gain of B chains, which is caused by the disproportionation reaction, compared with fusion by SBD GA . As is known for TuαGT, the starch products may represent nutritional values reminiscent of resistant starch dietary fibres. According to the separation in the evolutionary tree and the different functional improvements, we conclude that SBD St1 and SBD St2 contribute different effects by fusion with TuαGT and that they probably play different, albeit not yet identified, functional roles in the StDPE2. In the longer perspective, the obtained results disclose the potential for utilising insight into the wide diversity of SBDs for enzyme engineering and also to connect individual properties of the two "in tandem" SBDs with structure/function relationships of disproportionating enzymes in plants and bacteria.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/molecules28031320/s1, Figure S1: Sequence alignment of CBM20s with focus on GH77 DPE2s.  Acknowledgments: Karina Jansen (Department of Biotechnology and Biomedicine, Technical University of Denmark, Denmark) is gratefully acknowledged for technical assistance. We are thankful to Cargill for providing waxy maize starch and to Archer Daniels Midland for providing normal maize starch.

Conflicts of Interest:
The authors declare that they have no competing financial interest or personal relationship influencing the work reported in this paper.