Engineering of Cyclodextrin Glycosyltransferase Reveals pH-Regulated Mechanism of Enhanced Long-Chain Glycosylated Sophoricoside Specificity

The low water solubility of sophoricoside seriously limits its applications in the food and pharmaceutical industries. Long-chain glycosylated sophoricosides show greatly improved water solubility. Here, the product specificity of cyclodextrin glycosyltransferase (CGTase) for long-chain glycosylated sophoricosides was significantly affected by pH. Our results reveal the pH-regulated mechanism of the glycosylated product specificity of CGTase. This work adds to our understanding of the synthesis of long-chain glycosylated sophoricosides and provides guidance for exploring related product specificity of CGTase based on pH regulation.

ABSTRACT

Sophoricoside glycosylated derivatives, especially long-chain glycosylated sophoricosides (LCGS), have greatly improved water solubility compared with sophoricoside. Here, cyclodextrin glycosyltransferase from Paenibacillus macerans (PmCGTase) was employed for sophoricoside glycosylation. Saturation mutagenesis of alanine 156, alanine 166, glycine 173, and leucine 174 was performed due to their nonconservative properties among α-, β-, and γ-CGTases with different product specificities. Variants L174P, A156V/L174P, and A156V/L174P/A166Y greatly improved the product specificity for LCGS. pH significantly affected the extent of glycosylation catalyzed by the variants. Further investigations revealed that the pH-regulated mechanism for LCGS synthesis mainly depends on a disproportionation route at a lower pH (pH 4) and a cyclization-coupling route at a higher pH (pH 8) and equivalent effects of cyclization-coupling and disproportionation routes at pH 5. Whereas short-chain glycosylated sophoricosides (SCGS) are primarily produced via disproportionation of maltodextrin at pH 4 and secondary disproportionation of LCGS at pH 8. At pH 5, SCGS synthesis mainly depends on a hydrolysis route by the wild type (WT) and a secondary disproportionation route by variant A156V/L174P/A166Y. Kinetics analysis showed a decreased K_m value of variant A156V/L174P/A166Y. Dynamics simulation results demonstrated that the improved LCGS specificity of the variant is possibly attributed to the enhanced affinity to long-chain substrates, which may be caused by the changes of hydrogen bond interactions at the –5, –6, and –7 subsites. Our results reveal a pH-regulated mechanism for product specificity of CGTase and provide guidance for engineering CGTase toward products with different sugar chain lengths.

IMPORTANCE The low water solubility of sophoricoside seriously limits its applications in the food and pharmaceutical industries. Long-chain glycosylated sophoricosides show greatly improved water solubility. Here, the product specificity of cyclodextrin glycosyltransferase (CGTase) for long-chain glycosylated sophoricosides was significantly affected by pH. Our results reveal the pH-regulated mechanism of the glycosylated product specificity of CGTase. This work adds to our understanding of the synthesis of long-chain glycosylated sophoricosides and provides guidance for exploring related product specificity of CGTase based on pH regulation.

INTRODUCTION

Genistein (4′,5,7-trihydroxyisoflavone) and its monoglucoside derivatives, such as sophoricoside (genistein-4′-O-glucoside) and genistin (genistein-7-O-glucoside), are important soy isoflavones that have multiple nutritional potentials and biological functions associated with their endogenous estrogen actions and are widely applied in the pharmaceutical and food industries (1, 2). They have a wide variety of pharmacological properties such as anticancer (3–5), antioxidant (6), and antiosteoporosis (7), as well as cell-mediated immunity and vascular smooth muscle inhibition (8). In addition, sophoricosides can potentially be applied as anti-inflammatory (9, 10) and antifertility agents (11).

Although genistein and its monoglucoside derivatives exhibit various bioactivities, low water solubility is a nonnegligible bottleneck limiting their further applications. Transglycosylation is regarded as a green and powerful approach to improve water solubility for many flavonoids, such as naringin (12), puerarin (13), and genistin (14). A series of enzymes are typically used for flavonoid glycosylation, including α-glucanotransferase (14), cyclodextrin glycosyltransferase (CGTase) (15), UDP-glycosyltransferase (16), and sulfoquinovosyltransferase (17). Among them, CGTase is regarded as a versatile glycosyltransferase with multiple catalytic functions (e.g., cyclization, coupling, disproportionation, and hydrolysis) and application potentials (18, 19). CGTase can also be used to improve other physiochemical properties of various raw materials, such as enhancing l-ascorbic acid stability (20), alleviating stevioside bitterness (21), and decreasing steroidal saponin cytotoxicity (22). Various protein engineering strategies have been attempted to improve CGTases’ properties for better applications. In our previous work, site-saturation mutagenesis (23–25) and gene fusion strategies (26, 27) were employed to enhance the glycosylation efficiency of CGTase toward l-ascorbic acid using inexpensive glycosyl donors, such as soluble starch and maltodextrin.

Recently, several studies that focused on genistein glycosylation with CGTase have been reported. The water solubility of genistein was increased by 5-, 3.7 × 10³-, and 4.4 × 10⁴-fold for its monoglucoside, diglucoside, and triglucoside derivatives, respectively (14, 28), suggesting that longer sugar chains can lead to higher water solubility. In addition, similar bioactivities and biofunctionalities have been identified between genistein and its glycosylated derivatives (2). Therefore, long-chain glycosylated derivatives could have great potential in food and pharmaceutical applications.

In our recent study, genistein transglycosylation was performed using CGTase from Paenibacillus macerans (PmCGTase) for producing sophoricoside and its glycosylated derivatives (29). However, this biocatalytic process was hampered by the low yield of long-chain glycosylated sophoricosides (LCGS). Here, PmCGTase was engineered to improve its specificity in long-chain glycosylated products, and sites 156, 166, 173, and 174 were identified to be critical to product specificity due to their nonconservative properties among α-, β-, and γ-CGTases. The influence of pH on the product specificity of PmCGTase and its variants were also investigated. Furthermore, a pH-regulated mechanism for LCGS specificity by PmCGTase and its variants was elucidated. To the best of our knowledge, the pH-regulated mechanism of CGTase for transglycosylated product specificity has not been reported. This study reveals a pH-regulated mechanism of CGTase for product specificity of sugar chain length, which provides useful information for modulating CGTase product specificity by protein engineering.

RESULTS

Glycosyl acceptor selection.

Genistein glycosylation by PmCGTase resulted in the synthesis of sophoricoside and multiglucoside derivatives (29). Here, genistein and sophoricoside were tested as glycosyl acceptors. As shown in Fig. S1 in the supplemental material, there were clearly more products with sophoricoside than with genistein as an acceptor. Furthermore, the conversion ratio of sophoricoside (more than 40%) was much higher than that of genistein (less than 20%). It is speculated that PmCGTase prefers α-glucosidic bonds to β-glucosidic bonds in transglycosylation, since the glycosylation of genistein and sophoricoside depend on a β-1,4-glucosidic bond and an α-1,4-glucosidic bond, respectively (Fig. S2).

To further identify each product of sophoricoside glycosylation, LC-MS analysis was performed. As shown in Fig. S3, in addition to sophoricoside (corresponding to a molecular weight of 432), six products were identified as Glc-sophoricoside (GS), (Glc)₂-sophoricoside (G₂S), (Glc)₃-sophoricoside (G₃S), (Glc)₄-sophoricoside (G₄S), (Glc)₅-sophoricoside (G₅S), and (Glc)₆-sophoricoside (G₆S), corresponding to the molecular weights of 594, 756, 918, 1,080, 1,242, and 1,404, respectively (Fig. S3). Therefore, sophoricoside was used as the acceptor for further study due to its higher conversion ratio and multiglucoside products. Based on product characterization, six products were divided into two parts, short-chain glycosylated sophoricoside (SCGS) [(Glc)_n-sophoricoside (G_nS, n = 1, 2, 3)] and long-chain glycosylated sophoricoside (LCGS) [(Glc)_n-sophoricoside (G_nS, n = 4, 5, 6)].

Identification and saturation mutagenesis of key sites.

CGTases are usually classified as α-, β-, or γ-CGTase according to their different product specificities (30). To investigate the correlation between CGTase structure and the product specificity, amino acid sequence BLAST search in the NCBI database was performed with PmCGTase (GenBank accession number AFS50075.1) as a template, and then two γ-CGTases (GenBank accession numbers BAE87038.1 and BAH14968.1) and four β-CGTases (GenBank accession numbers ABG02281.1, AGG09664.1, 8CGT_A, and CDGT1_BACCI) were selected due to their high homology (>60%) with PmCGTase. As shown in Fig. S4, the region of domain B (sites 151 to 177) was investigated due to its critical role in substrate binding (19). After amino acid sequence alignments of region 151 to 177, amino acids were strictly or partly conservative except at four sites, 156, 166, 173, and 174. Thus, these four sites may play key roles in product specificity due to their flexibility in the substrate binding structure of CGTase.

Saturation mutagenesis at sites 156, 166, 173, and 174 of PmCGTase was performed. Mutants at sites 156, 166, and 174 had significant improvement in the LCGS product specificity. As shown in Fig. S5, the best variants, A156V, L174P, and A166Y, were chosen, with the highest LCGS product ratios of 30.3%, 32.0%, and 28.6%, respectively, which were far higher than that of wild-type PmCGTase (13.2%). Based on these three best single-site mutants, combination mutants of A156V/L174P, A156V/A166Y, A166Y/L174P, and A156V/L174P/A166Y were constructed to evaluate the synergistic effect of multiple mutations, and their LCGS ratios were 33.4%, 31.1%, 31.6%, and 34.9%, respectively (Fig. S5). Subsequently, the best single-site mutant, L174P (M1), double-site mutant, A156V/L174P (M2), and triple-site mutant, A156V/L174P/A166Y (M3), were purified to over 90% purity with molecular weights of about 75 kDa (Fig. 1) and subjected to further investigation.

FIG 1.

SDS-PAGE of purified PmCGTase WT and its variants. M, protein marker; 1, WT; 2, M1; 3, M2; 4, M3.

Effect of pH on sophoricoside glycosylation by PmCGTase WT and its variants.

As shown in Fig. 2, the conversion ratio and product specificity of the WT were consistent at different pHs, while those of the variants were significantly different over a pH range of 4 to 8. The LCGS ratio reached the highest level at pH 4, whereas it was reduced to the lowest level at pH 5 and then gradually increased from pH 5 to 8. For instance, the LCGS ratio of M3 at pH 4 was 3.5-fold of that of the WT. However, the conversion ratios of all variants were lower than that of the WT. In particular, conversion ratios were usually in inverse ratio to the LCGS ratios (Fig. 2).

FIG 2.

Influence of reaction pH on the synthesis of glycosylated products by PmCGTase WT and its variants. (a) WT; (b) M1; (c) M2; (d) M3. pH was adjusted with disodium hydrogen phosphate-citrate buffers.

Effect of pH on four catalytic activities of PmCGTase WT and variants.

To reveal the effect of the pH-regulated mechanism on product specificity, four activities (cyclization, coupling, disproportionation, and hydrolysis) of the WT and variants (M1, M2, and M3) were determined at different pHs (4, 8). As shown in Fig. 3, except for the transition from pH 4 to pH 5, the cyclization activities of the WT and variants gradually decreased with increasing pH, and the highest activities were obtained at pH 5. At pH 4 to 8, the WT, M2, and M3 showed higher cyclization activities than variant M1. A similar variation trend appeared in both the coupling and disproportionation activity charts (Fig. 3). The hydrolysis activities of variants M2 and M3 were lower than those of the WT and L174P at pH 4 to 8 (Fig. 3).

FIG 3.

Influence of reaction pH on the reaction activities of PmCGTase WT and its variants. (a) Cyclization activity; (b) coupling activity; (c) disproportionation activity; (d) hydrolysis activity.

Effect of pH on α-cyclodextrin (α-CD) synthesis from maltodextrin with/without sophoricoside substrate.

Figure 4 shows α-CD synthesis by PmCGTase WT and M3 from maltodextrin with/without sophoricoside substrate. For both the WT and M3, the α-CD yield from double substrates (maltodextrin and sophoricoside) was lower than that with a single maltodextrin substrate, and the consumed α-CD may be reacted with sophoricoside. At pH 4, the amounts of both α-CD synthesis and consumption by the WT were more than those by M3 (Fig. 4a). Compared with the WT, variant M3 resulted in less α-CD synthesis but more α-CD consumption at pH 5 (Fig. 4b). At pH 8, the amount of α-CD synthesis by the WT was less than that by M3, whereas their amounts of α-CD consumption were consistent (Fig. 4c).

FIG 4.

α-CD synthesis from maltodextrin with/without sophoricoside acceptor by PmCGTase WT and variant M3 at different pHs. (a) pH 4; (b) pH 5; (c) pH 8. *, α-CD consumption = amount of α-CD without sophoricoside − amount of α-CD with sophoricoside.

Effect of pH on G₆S synthesis with α-CD or maltodextrin as a glycosyl donor.

Figure 5 showed the G₆S synthesis by PmCGTase WT and M3 with sophoricoside as a glycosyl acceptor and α-CD or maltodextrin as a glycosyl donor at different pHs. At both pH 4 and 8, the WT contributed to higher G₆S yields than M3, whereas M3 exhibited great superiority compared with the WT at pH 5. Compared with α-CD, using a maltodextrin donor resulted in superior G₆S synthesis at pH 4 and 5. In contrast, the amount of G₆S synthesized with an α-CD donor was less than that with maltodextrin donor at pH 8.

FIG 5.

(Glc)₆-sophoricoside synthesis by PmCGTase WT and variant M3 with α-CD and maltodextrin as glycosyl donors. (a) pH 4; (b) pH 5; (c) pH 8. Equal weights of α-CD and maltodextrin were used for the reactions.

Effect of pH on G₆S degradation by PmCGTase WT and M3.

Figure 6 shows the trend of the relative amount of different glycosylated products by the WT and M3 within 60 min at pH 4, 5, and 8. At pH 4, both the WT (Fig. 6a) and M3 (Fig. 6g) only produced G₆S product with an α-CD donor within 60 min. However, mixed-product G_nS (n = 1, 2, …, 6) was produced by both the WT (Fig. 6d) and M3 (Fig. 6j) with a maltodextrin donor, although the ratio of G_nS (n < 6) was much lower than that of G₆S (Fig. S7).

FIG 6.

Trend of different glycosylated products along with the time spent at different pHs. Shown are the WT with α-CD as the glycosyl donor at pH 4 (a), pH 5 (b), and pH 8 (c); WT with maltodextrin as the glycosyl donor at pH 4 (d), pH 5 (e), and pH 8 (f); M3 with α-CD as the glycosyl donor at pH 4 (g), pH 5 (h), and pH 8 (i); M3 with maltodextrin as the glycosyl donor at pH 4 (j), pH 5 (k), and pH 8 (l). Relative amount = each product amount at different time/each product amount at 1 min.

At pH 5, the amounts of SCGS increased along with a decrease in G₆S amount (Fig. 6). Furthermore, the WT exhibited the highest synthesis rate for GS and G₂S (Fig. 6b and e). For M3, the production of G₄S and G₅S were faster than that of GS and G₂S, especially with a maltodextrin donor (Fig. 6k). The results corresponded to the trend of different product ratios (Fig. S7).

At pH 8, all the products were increased along with prolonged time (Fig. 6), whereas, for the WT, the synthesis rates of G_nS (n < 6) from α-CD were much faster than that of G₆S (Fig. 6), and the ratio of G₆S was decreased along with prolonged time (Fig. S7a).

Kinetics analysis of PmCGTase WT and variant M3.

Based on the Lineweaver-Burk plots, both the WT and M3 displayed a normal ping-pong mechanism (Fig. S6). The detailed kinetics parameters are listed in Table 1. Compared with the WT, the V_max and K_cat values of M3 were increased by 20%, and the K_m values for maltodextrin and sophoricoside were decreased by 32% and 17%, respectively. In addition, the K_cat/K_m value for maltodextrin of M3 was slightly higher than that of the WT.

TABLE 1.

Kinetics parameters of WT and M3 (variants A156V/L174P/A166Y)

Enzymea	Kcat (min−1)	Kmb (g/liter)	Kcat/Kmb [liter/(g · min)]	Kmc (g/liter)	Kcat/Kmc [liter/(g · min)]
WT	4.76	1.03	4.62	0.30	15.87
M3	3.80	0.70	5.42	0.25	15.20

^aThe concentration of the enzymes (WT and M3) was 0.01 g/liter.

^bMaltodextrin.

^cSophoricoside.

DISCUSSION

In this study, variants M1, M2, and M3 showed enhanced LCGS specificity, which has greater water solubility than SCGS. Furthermore, their product specificities were significantly affected by pH (Fig. 2). Especially at pH 4 and 8, variants showed much higher LCGS specificity than at pH 5 to 7. Therefore, revealing the pH-regulated mechanism of CGTase would be instructive for further engineering of its LCGS product specificity.

Synergistic effect of different catalytic reactions of CGTases on product specificity.

To study the pH-regulated mechanism, the influence of pH on the catalytic activities of CGTase was investigated (Fig. 3), and the results suggest that the synergistic effect of different catalytic reactions may have contributed to the product specificity of glycosylated sophoricoside.

Glycosylated sophoricoside may usually be synthesized by transglycosylation reactions of CGTases (e.g., cyclization, coupling, and disproportionation), and hydrolysis may result in SCGS synthesis by hydrolyzing long oligosaccharide chains. Fig. 3 shows that the WT, M2, and M3 had higher transglycosylation activities (cyclization, coupling, and disproportionation) than M1 over the pH range of 4 to 8, which may explain the higher conversion ratio of the WT, M2, and M3 compared to M1 (Fig. 2). Variants M1, M2, and M3 showed significantly lower hydrolysis activities than the WT (Fig. 3d), explaining the higher LCGS ratios of variants than WT. Therefore, the product specificities may be closely related to the synergistic effect of four reactions at different pHs. The following proposed scheme possibly further reveals the pH-regulated mechanism of product specificity.

Proposed synthesis routes for LCGS and SCGS.

Figure 7 shows the proposed synthesis routes of LCGS and SCGS. To understand the synthesis mechanism of LCGS products, G₆S was mainly investigated because it had the longest oligosaccharide chain among LCGS products in this study. As shown in Fig. 7a, two possible routes (route 1 and 2) may have contributed to G₆S synthesis. Route 1 mainly comprises cyclization and coupling reactions, in which α-CD could be synthesized by a cyclization reaction and then utilized by a coupling reaction to produce G₆S. In route 2, G₆S could be directly produced from maltodextrin by a disproportionation reaction of CGTase.

FIG 7.

Proposed pH-regulated mechanism on product specificity. (a) Possible routes of G₆S and SCGS synthesis at different pHs; (b) proposed mechanism of G₆S and SCGS synthesis at different pHs by PmCGTase WT and variants.

To understand the mechanism of SCGS synthesis, three possible routes (routes 3, 4, and 5) are proposed in Fig. 7a. In route 3, G₆S can be degraded into SCGS by secondary disproportionation. “Secondary disproportionation” was proposed in a previous report (31) and means that oligoglucosylated products serve as donor substrates for another transglycosylation. Thus, the sophoricoside moiety of G₆S may move directly to the +2 subsite or +3 subsite and serve as a glycosl donor for secondary disproportionation. In route 4, G₆S may be mainly degraded into SCGS by hydrolysis. GS and G₂S are produced by hydrolysis when the sophoricoside moiety of G₆S moves into the +2 subsite or +3 subsite. In route 5, short maltodextrin may be used as a glycosyl donor for SCGS synthesis by disproportionation.

Proposed pH-regulated mechanism of LCGS synthesis.

The amount of α-CD synthesis with double substrates (maltodextrin and sophoricoside) was lower than that with a single maltodextrin substrate (Fig. 4), suggesting that partial α-CD may be utilized as the glycosyl donor for sophoricoside glycosylation. To confirm this conclusion, G₆S was synthesized with α-CD and sophoricoside as the glycosyl donor and acceptor, respectively (Fig. 5). These results further verified the proposed route 1 (Fig. 7a). pH has a critical effect on the LCGS synthesis routes. Lower α-CD consumption at pH 4 than at pH 8 suggests that route 1 may prefer pH 8 rather than pH 4. Furthermore, G₆S synthesized from maltodextrin may mainly depend on route 2 at pH 4 due to the higher G₆S yield with a maltodextrin donor than a α-CD donor (Fig. 5a).

At pH 5, the amount of G₆S synthesized by the WT with either an α-CD or maltodextrin donor are similar (Fig. 5b), suggesting that both routes 1 and 2 play important roles in G₆S synthesis. However, for M3, the amount of G₆S synthesized from α-CD was much more than that from maltodextrin from 0 to 20 min, while after 20 min, G₆S synthesized from α-CD began to decrease and was lower than that from maltodextrin (Fig. 5b). One possible reason for this may be that the coupling activity of M3 remained high in the beginning (0 to 20 min), whereas it decreased with prolonged time. This result is also in accordance with the results shown in Fig. 4b, which shows that α-CD consumption of M3 remained high from 0 to 20 min, while it decreased after 20 min. However, both route 1 (cyclization-coupling) and route 2 (disproportionation) (Fig. 7a) may contribute to G₆S synthesis with a maltodextrin donor, leading to the larger amount of G₆S than that with an α-CD donor after 20 min. In Fig. 5b, the amount of G₆S increases in the beginning and then decreases, which may be attributed to its higher degradation rate than synthesis rate with prolonged time.

Proposed pH-regulated mechanism of SCGS synthesis.

At pH 4, both the WT and M3 produced only G₆S from α-CD but produced mixed-product G_nS (n = 1, 2, …, 6) from maltodextrin (Fig. 6; Fig. S7), suggesting that the mixed products may be produced by disproportionation of maltodextrin (route 5 in Fig. 7a), rather than G₆S degradation.

At pH 5, the amount of G₆S was decreased, while SCGS was dramatically increased by the WT (Fig. 6b and e; Fig. S7a), indicating that SCGS were produced by G₆S degradation. Furthermore, the synthesis rates of GS and G₂S were much higher than those of G₅S and G₄S, demonstrating that hydrolysis may play a more important role than secondary disproportionation by WT during G₆S degradation (route 4 in Fig. 7a). For M3, secondary disproportionation may outperform hydrolysis during G₆S degradation, because the synthesis rates of GS and G₂S were similar to those of G₅S and G₄S from α-CD (Fig. 6h). However, with a maltodextrin donor, much higher G₅S and G₄S were achieved than GS and G₂S (Fig. 6k; Fig. S7d), owing to the disproportionation of maltodextrin (route 2 in Fig. 7a). Furthermore, compared with the WT (Fig. S7a and b), M3 (Fig. S7c and d) maintained the higher G₆S amount over a prolonged time. It is speculated that G₆S degradation of M3 was less than that of the WT, which corresponds to the lower disproportionation and hydrolysis activities observed with M3 than with the WT (Fig. 3c and d).

At pH 8, the WT showed similar production rates for SCGS (GS and G₂S) and LCGS (G₄S and G₅S) with both α-CD and maltodextrin donors (Fig. 6c and f), suggesting that secondary disproportionation prevailed over hydrolysis during G₆S degradation (route 3 in Fig. 7a). M3 also showed higher secondary disproportionation than hydrolysis during G₆S degradation at pH 8 due to the similar production rates of all products with an α-CD donor (Fig. 6i). However, M3 exhibited higher production rates for G₅S and G₄S than for GS and G₂S with a maltodextrin donor (Fig. 6l), owing to higher synthesis rates of LCGS than their degradation rates.

Predicted pH-regulated mechanism on product specificity of the WT and M3.

We can summarize the pH-regulated mechanisms of product specificity based on the above-proposed routes for LCGS and SCGS synthesis at different pHs. For both the WT and M3, G₆S synthesis mainly depends on route 2 at pH 4 and route 1 at pH 8, and routes 1 and 2 play equal roles at pH 5 (Fig. 7b).

The synthesis of SCGS mainly depends on route 5 at pH 4 and route 3 at pH 8 by both the WT and M3 (Fig. 7b). However, at pH 5, SCGS were produced by the WT and M3 mainly via route 4 and route 3, respectively (Fig. 7b). High LCGS specificity by M3 may be attributed to its higher LCGS synthesis rate and lower SCGS synthesis rate than the WT.

Kinetics and molecular dynamics simulation analysis for enhanced LCGS specificity.

Most CGTases follow ping-pong kinetics in the disproportionation reaction, such as Bacillus circulans 251 CGTase (19), P. macerans CGTase (23, 32), and Bacillus stearothermophilus NO₂ CGTase (33). In addition, a random ternary complex mechanism was usually observed with CGTases in the coupling reaction, such as B. circulans 251 CGTase (19) and Bacillus clarkii 7364 CGTase (34). In this study, the glycosylation of sophoricoside with maltodextrin as a glycosyl donor exhibited a ping-pong mechanism at pH 5, which further confirms our prediction that disproportionation plays a key role at pH 5.

To understand the kinetics mechanism of PmCGTase and its variants, the substrate maltononaose was docked into the WT and M3. Then, a 20-ns molecular dynamics (MD) simulation was performed on protein-ligand complexes (Fig. S8). The average conformations of WT-maltononaose and M3-maltononaose under a catalytic state were extracted (Fig. 8a). Mutagenesis of residues 156, 166, and 174 rendered appreciable changes on the flexibility of the “145 to 174” loop region, although they are far from the substrate (Fig. 8a).

FIG 8.

MD simulation analysis of interactions between maltononaose and CGTases (WT and M3).

Based on the interaction analysis (Fig. 8b), 9 hydrogen bonds and 13 van der Waals interactions existed between WT and maltononaose, whereas in M3, 15 hydrogen bonds and 14 van der Waals interactions were observed. More interactions may lead to stronger affinity between maltononaose and M3, corresponding to the K_m values in Table 1.

Hydrogen bond interactions were obviously different around –7, –6, and –5 sugars between the WT and M3 (Fig. 8b). In the WT, 0 and 1 hydrogen bond were formed with –7 and –6 sugars, respectively, whereas 1 and 2 hydrogen bonds with –7 and –6 sugars were observed in M3. Sugars –6 and –7 are usually regarded as entrance positions of the binding pocket that are crucial for the product specificity of CGTase (35). Therefore, more hydrogen bonds formed with –6 and –7 sugars in M3 could lead to easier binding of long linear oligosaccharide for LCGS synthesis compared to the WT. For –5 sugar, 2 hydrogen bonds were present in the WT, while there were no hydrogen bonds in M3 (Fig. 8b). Low substrate affinity at the –5 subsite has been regarded as a selective strategy by preventing the binding of short oligosaccharides (35). Consequently, variant M3 may prefer longer oligosaccharide substrates due to its lower affinity at the –5 subsite compared with the WT. Different interactions at the –5, –6, and –7 subsites between the WT and M3 may be attributed to the flexibility of the 145 to 174 loop (Fig. 8) (35).

Conclusions.

PmCGTase was engineered to improve its product specificity toward LCGS in sophoricoside glycosylation. pH exhibited an important role in product specificity, and the pH-regulated mechanism of PmCGTase and its variants was revealed by investigating their four activities (cyclization, coupling, disproportionation, and hydrolysis). The results demonstrate that several routes could contribute to the synthesis of G_nS products (n = 1, 2, …, 6) at different pHs. Structure simulation and molecular dynamics analysis further revealed that mutations of A156V, A166Y, and L174P may affect the flexibility of the 145 to 174 loop, leading to enhanced binding ability of long oligosaccharide substrates in variant M3. Based on the proposed pH-regulated mechanism, enhancing LCGS synthesis routes (e.g., cyclization, coupling, and disproportionation) and decreasing SCGS synthesis routes (e.g., hydrolysis and secondary disproportionation) by protein engineering may be effective approaches for improving LCGS production.

MATERIALS AND METHODS

Bacterial strains, plasmids, and materials.

The recombinant plasmid pET-20b(+)/co-α-cgt was constructed in our previous study (29). Escherichia coli JM109 and E. coli BL 21(DE3) were purchased from TaKaRa (Dalian, China). PrimeSTAR HS DNA polymerase, restriction endonucleases, PCR reagents, and DpnI restriction enzyme were purchased from TaKaRa (Dalian, China). DNA sequence identification was performed by Synbio Technologies (Suzhou, China). Genistein, sophoricoside, and maltodextrin were purchased from Sangon Biotech (Shanghai, China). High-pressure liquid chromatography (HPLC)-grade acetonitrile was purchased from Titan Scientific Co. Ltd. (Shanghai, China), and deionized water was purchased from Watson Group Ltd. (Hong Kong, China). All other chemicals used were of reagent grade from Chinese reagent companies.

Site-saturation mutagenesis.

PrimeSTAR HS DNA polymerase was employed for site-saturation mutagenesis with the plasmid pET-20b(+)/co-α-cgt as the template DNA. Sequences of complementary primers are shown in Table S1. DpnI was used to eliminate the template, and the PCR products were transformed into E. coli JM109. Then, clones were further confirmed by DNA sequencing, and the correct clones were transformed into E. coli BL21(DE3) for expression.

Preparation and purification of PmCGTase and its variants.

The PmCGTase and variants were prepared according to our previous report (29). Purification of the crude enzyme was carried out by salting with 30% (wt/vol) ammonium sulfate and then dialyzing with 14-kDa dialysis membranes (Sangon Biotech, Shanghai, China). Subsequently, a Ni-NTA agarose column (Qiagen, Chatsworth, CA, USA) was used for further purification as described previously (29), and the purity of CGTases was determined using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Activities assay of PmCGTase and its variants.

The cyclization activity of CGTase was analyzed as previously described with a slight modification (36). Briefly, 20 μl of appropriately diluted purified enzyme and 80 μl of 1% (wt/vol) maltodextrin were mixed together and incubated at 40°C for 10 min. Then, 100 μl HCl (1.0 M) and methyl orange (0.1 mM) were added immediately. After incubation for 20 min at 16°C, the absorbance of the mixture was measured at a wavelength of 505 nm. One unit of cyclization activity was defined as the amount of enzyme capable of producing 1 μM α-cyclodextrin per minute.

The coupling activity of CGTase was determined as described previously (19) with slight modification. Briefly, the mixture comprised 500 μl of β-cyclodextrin (10 mM), 500 μl of methyl α-d-glucopyranoside (MαDG) (400 mM), and 100 μl of appropriately diluted CGTase. The reaction was performed at 50°C for 10 min and was terminated by the addition of 200 μl of HCl (3 M), and then 200 μl of NaOH (3 M) was added for neutralization. Subsequently, α-glucosidase was added to liberate glucose. Lastly, the glycose concentration was measured using a Biosensor analyzer (BISD, Shandong, China). One unit of coupling activity is defined as the amount of enzyme coupling 1 μM cyclodextrin to MαDG.

The hydrolyzing activity of CGTase was determined using the starch-degrading method as described previously (19). Appropriately diluted purified enzyme (100 μl) and 1% (wt/vol) soluble starch (dissolved in 50 mM pH 6.0 phosphate buffer) (900 μl) were mixed together and incubated at 50°C for 10 min. One unit of hydrolyzing activity was defined as the amount of enzyme producing 1 μM reducing sugar per minute.

The disproportionation activity of CGTase was determined as previously described (19). Briefly, 4-nitrophenyl-α-d-maltoheptaoside-4-6-O-ethylidene (EPS; Megazyme, County Wicklow, Ireland) (4 mM) and maltose (20 mM) were used as donor and acceptor substrates, respectively. Then, appropriately diluted purified CGTase was added and incubated at 50°C for 10 min, and the reaction was terminated by the addition of HCl (3.0 M). Then, appropriate NaOH (3.0 M) was used for neutralization. Subsequently, α-glucosidase was added for the liberation of para-nitrophenol from the product of nonblocked linear oligosaccharide. Lastly, the absorbance of the mixture at 401 nm was measured after the addition of 1 ml sodium carbonate (1 M). One unit of activity was defined as the amount of enzyme converting 1 μM EPS per minute.

α-CD synthesis and analysis.

α-CD formation was performed as previously described (37) with slight modification. The whole reaction solution (1 ml) contained 200 μl of 0.01 mg/ml purified wild-type or mutant CGTases and 600 μl of 20 g/liter maltodextrin (diluted with pH 4, 5, and 8 phosphate-buffered saline [PBS] buffer). The remaining 200 μl was supplemented with 2 g/liter sophoricoside or the corresponding buffer. The whole reaction solution was mixed in a 2-ml closed tube and shaken for the appropriate time on a 120-rpm rotary shaker at 40°C. The concentrations of α-CD were determined by HPLC analysis with a LiChrosorb NH₂ column (Merck, Darmstadt, Germany).

Transglycosylation of genistein and sophoricoside by PmCGTase and its variants.

The transglycosylation of genistein by CGTases was performed as previously described (29). Briefly, the whole reaction solution (1 ml) contained 200 μl of 0.05 mg/ml purified wild-type or mutant CGTases (diluted with pH 6 PBS buffer), 200 μl of genistein or sophoricoside (2 g/liter) dissolved in dimethyl sulfoxide (DMSO) as the glycosyl acceptor, and 600 μl of maltodextrin (10 g/liter) dissolved in 50 mM of PBS buffer (pH 6) as the glycosyl donor. The whole reaction solution was mixed in a 2-ml closed tube and shaken for the appropriate time on a 120-rpm rotary shaker at 40°C for 12 h. Then the reaction products were determined by HPLC. Similarly, the transglycosylation of sophoricoside was performed by replacing genistein with sophoricoside as the glycosyl acceptor.

Analysis of products by HPLC.

Genistein or sophoricoside and their glycosylated products were quantified with an Agilent 1260 Infinity HPLC system equipped with a 250 × 4.6-mm Diamonisil C₁₈ column and an Agilent 1260 variable wavelength scanning ultraviolet detector (VWD) set at 260 nm. A mobile phase containing solution A (water/phosphoric acid, 100:0.1 [vol/vol]) and solution B (acetonitrile) was used for gradient elution as follows: solution B was gradually increased from 15% to 85% during a 15-min interval at 30°C. Then 10 μl of sample was injected into the analytical HPLC by an automatic sampler, and the flow rate was 0.8 ml/min.

Liquid chromatography-mass spectrometry analysis.

The molecular weight of each product was analyzed with a MALDI SYNAPT Q-TOF Premier mass spectrometer (Waters, USA) equipped with an electrospray ion source in the V-optics positive mode. An ethylene-bridged hybrid (BEH) C₁₈ column (2.1 × 100 mm) was used for liquid chromatogram on an ACQUITY UPLC instrument (Waters, USA). A mobile phase containing solution A (water/phosphoric acid, 100:0.1 [vol/vol]) and solution B (acetonitrile) was used for gradient elution as follows: solution B was increased gradually from 10% to 60% during an 8-min period and then reached 100% during a 4-min period at 45°C with a flow rate of 0.3 ml/min. The injection volume of the samples was 0.2 μl, and the mass spectrometer scanned from m/z 0 to 1,600.

Influence of reaction pH on sophoricoside glycosylation.

On the basis of the initial transglycosylation conditions (40°C, pH 6), the influence of the reaction pH (4–8) on sophoricoside glycosylation by purified WT and mutant CGTases was investigated. The glycosylated products were analyzed by HPLC.

Kinetics analysis of PmCGTase and variants for sophoricoside glycosylation.

The kinetics analysis of PmCGTase and its variants for sophoricoside glycosylation was performed as previously described (19) with slight modification. Briefly, maltodextrin and sophoricoside were used as the glycosyl donor and acceptor, respectively. The glycosylation rate was detected with various concentrations of maltodextrin (0.6, 1.5, 3, 6, and 12 g/liter) while fixing the concentration of sophoricoside (0.05, 0.1, 0.2, and 0.4 g/liter). The reaction time was 1 h. The obtained results were subjected to kinetics analysis using Origin 8.0 software.

Structure modeling, docking, and molecular dynamics simulation of PmCGTase and its variants.

The homology models of the PmCGTase and its variants were constructed using the crystallographic structure of CGTase from P. macerans (PDB accession code 4JCL) as the template (with 99% identity) in the program suite EasyModeller 4.0. Sequence alignment showed only two different amino acid residues at positions 156 and 157. All graphs were generated using Discovery Studio Client 4.5 (Accelrys, USA). Structural alignment was performed with the combinatorial extension method using the jCE/jFATCAT Structure Alignment Server 2.6 (http://cl.sdsc.edu/). The stereochemical quality of the model was examined with Procheck, Verify3D, and ProQ, with 100% of the residues in the most favored regions.

Molecular docking calculations with maltononaose as the inhibitor were accomplished using Discovery Studio Client 4.5 with the CDOCKER docking algorithm based on CHARMM force field. Docking was performed using interactive growing and subsequent scoring with standard parameters.

GROMACS 5.1.4 (38) was employed for molecular dynamics simulation of CGTase. PROPKA was employed for determining the protonation states of residues at pH 7. A rectangular box with the simple point charge (SPC) water model was used for solvation. The all-atom optimized potentials for liquid simulations (OPLS-AA) force field was used for protein topology generation and ligand parameterization. Na⁺ or Cl^– ions were used for neutralization. Steric clashes in the system were removed using steepest energy minimization. The particle-mesh-Ewald (PME) method was used to calculate long-range electrostatic interactions. A 2-fs time step was kept during the simulation. Subsequently, the equilibration phase was carried out with 100 ps. Lastly, molecular dynamics simulation was conducted at 313.15 K (40°C) for 20 ns.