大家好，今天推送的文章是2024年5月发表在Journal of Agricultural and Food Chemistry上的“Tailor-Made α‑Glucans by Engineering the Processivity of α‑Glucanotransferases via Tunnel-Cleft Active Center Interconversions”，通讯作者为江南大学的柏玉香研究员。
The function of a polysaccharide is closely related to its size, which largely depends on the continuous synthesis capacity of the transferases responsible for its synthesis. The tunneling active center structure has been identified as a key factor in controlling the persistence of the synthesis of multiple glycoside hydrolases (GHs, such as cellulase and chitinase). Similar tunneling structures have been observed in Lactobacillus reuteri citrate 121 GtfB (Lr121 GTfB) α-dextran transferases of the GH70 family. The molecular components that support the sustained synthesis of these transglucosylases remain unexplored.
The team of authors reported that a novel 4,6-α-glucan transferase from Lactobacillus reuteri N1 (LrN1 GtfB) synthesized a minimal (α1 →→ 4) α-glucan interspersed with linear and branched chain (α1 6) linkages, which has a cracking active center rather than a tunneling structure. Ring exchange engineering based on the crystal structures of LrN1 GtfB and Lr121 GtfB elucidated the effect of loop-mediated tunneling/rip structures at donor subsites -2 to -3 on the sustained synthesis capacity of these α-glucan transferases, enabling customization of product size and substrate preference.
1
Conformational comparison of GtfB α-glucotransferase active centers
Previous work by the authors identified a novel GtfB enzyme from Lactobacillus reuteri N1 (LrN1 GtfB) with high enzymatic activity and heterologous expression of soluble protein yield. LrN1 GtfB exhibits 4,6-α-glucan transferase II activity and is capable of synthesizing linear and branched (α1 → 6) bonds scattered in the (α1 → 4)-dextran chains. In addition, the crystal structure of LrN1 GtfB provides clearer information about its active center conformation. As shown in Figure 1, the AlphaFold model of Lactobacillus reuteri NCC 2613 GtfB (Lr2613 GtfB), LrN1 GtfB, and Lf2970 GtfB showed that the loops A1 and B located at the enzyme donor subsite were too short to form the tunnels observed at the active center of Lr121 GtfB. As a result, LrN1 GtfB exhibits a relatively open substrate binding groove that is between Lr2613 GtfB and Lf2970 GtfB. In contrast, the conformation of the active center in Lr121 GtfB differs from the other three enzymes due to the longer loops of A1 (17 residues) and B (20 residues). In Lr121 GtfB, ring A1 forms a large bulge towards domain B, while ring B folds over the binding groove in the opposite direction. These two rings act as "lids" that cover donor subsites -2 to -3 and form a tunnel-like structure, resulting in a less open active center.
Figure 1
2
Molecular distribution of LrN1 GtfB and different substrate products
As shown in Figure 2, the authors' team determined the mean molecular weight distributions of LrN1 GtfB products incubated with maltoheptose (G7)/amylose/amylopectin by HPGC at 1.9, 3.0, and 2.9 kDa, respectively. As a result, broad substrate specificity is observed in LrN1 GtfB, which is able to exhibit considerable modification effects on amylose and amylopectin, similar to Lr2613 GtfB due to the relatively open active center. In addition, the modification of LrN1 GtfB significantly reduced the molecular weight of amylose and amylopectin, indicating that LrN1 GtfB has strong endolytic activity against these high molecular weight substrates, resulting in a significant reduction in the molecular weight of the substrates. It then completes the (α1→6) transglucosylation process, increasing the proportion of (α1→6) bonds in the α-dextran product.
Figure 2
Thus, LrN1 GtfB yields the smallest reuytec-chain α-dextran from amylose, which has not been observed in the product synthesized to date from the characterized GtfB homolog (Figure 3). Notably, Lr121 GtfB yields a linear IMMP with molecular weights up to 15 kDa from amylose.
Figure 3
3
Construction of LrN1 GtfB loop-engineered mutants
with AlphaFold models
The authors' team highlighted the correlation between the loop on the donor side of the active center and the size of the α-glucan during the evolution of GH70 GtfB α-glucan transferase through comparative structural and functional analysis. To investigate the effect of these rings on the product profile in LrN1 GtfB, the authors' team made reasonable substitutions for the A1 (residues 384-393) and B (residues 169-176) rings of the donor subloci in LrN1 GtfB. Using LrN1 GtfB as the "parent" enzyme, three cyclically engineered mutants (LrN1 GtfB121loop, LrN1 GtfB2970loop, and LrN1 GtfB2613loop) were generated by swapping the sequence fragments of loops A1 and B in LrN1 GtfB with the sequence fragments of three representative GtfB enzymes. As shown in Figure 4, the length of rings A1 and B in the mutant LrN1 GtfB2613loop and LrN1 GtfB2970loop is very close to the length of the wild-type LrN1 GtfB. Therefore, the active center conformation of these two mutants is not significantly different from that of wild-type enzymes and maintains a relatively open structure. However, the long loops in the LrN1 GTfB121loop mutant in the AlphaFold model successfully contributed to the formation of tunneling structures similar to those observed in Lr121 GTfB (Figure 1).
Figure 4
The authors' team performed biochemical characterization of three cyclic engineered mutants to investigate the effect of loop A1 and B-mediated tunneling structures on GtfB α-dextrantransferase synthesis capacity. The enzymatic activity of these cyclically engineered variants is similar to or even slightly higher than that of wild-type LrN1 GtfB, providing evidence for the structural integrity of these mutants. Among them, the hydrolytic activity of LrN1 GtfB2970loop is about five times that of the wild type, and its transglucosylation efficiency is significantly reduced compared with LrN1 GtfB. As expected, LrN1 GtfB121loop mutants with tunneling structure exhibited lower hydrolytic activity and higher transglucosylation efficiency compared to wild-type LrN1 GtfB. This can be attributed to the tunneling structure, which is formed by rings A1 and B located at the -2 to -3 subsites, which help shield the solvent and reduce the hydrolysis effect, which is more conducive to transglucosylation reactions.
4
Characterization of LrN1 GtfB-loop-engineered mutant products
LrN1 GtfB121loop, LrN1 GtfB2970loop, and LrN1 GtfB2613loop incubated with amylose accounted for 29%, 30%, and 28% of the (α1 → 6) bond ratio, respectively, which is almost identical to that observed in LrN1 GtfB (28%). Therefore, the 1H NMR results preliminarily ruled out the critical role of loops A1 and B in enzyme bond specificity. In addition, HPGC analysis showed that there was no significant difference in the molecular weight of the amylose product after exchanging the LrN1 GtfB ring with Lf2970 GtfB and Lr2613 GtfB (Figure 5), possibly reflecting the absence of an intrinsic change in the active center of LrN1 GtfB for these ring substitutions. Strikingly, the introduction of a longer cyclic mimetic of Lr121 GtfB into LrN1 GtfB resulted in an average nearly doubling of the product size of amylose to 5.9 kDa, while the shorter cyclic variant retained a similar product size to the wild-type enzyme (Figure 5). These findings innovatively establish a major contribution of the donor site loop to the synthesis capacity of LrN1 GtfB and possibly other GtfB-like α-glucan transferases. Therefore, the synthesis capacity of LrN1 GtfB is successfully regulated by ring engineering of donor-bound subsite tunneling structures without losing the original linkage specificity.
Figure 5
5
Reverse loop engineering in tunnel structure Lr121 GtfB
To further confirm the effect of the tunneling structure on the synthesis capacity of these GtfB α-dextran transferases, and to verify the effect of loop-mediated tunneling on its broad substrate specificity, the authors' team performed reverse loop engineering to form loops A1 (residues 1139-1151, Lr121 GtfB number) and B (residues 905-924, Lr121 GtfB number) (Fig. 6) by exchanging Lr121 GtfB with the corresponding LrN1 GtfB shorter loops. to destroy the tunnel architecture.
As shown in Figure 6, in wild-type Lr121 GtfB, rings A1 and B synergistically form a tunnel structure at −2 to −3 of the donor subsite. However, the AlphaFold model of Lr121 GtfBN1loop exhibits a relatively open fracture conformation in the active center, which is quite different from the tunneling observed in wild-type Lr121 GtfB. The hydrolytic activity of Lr121 GtfBN1loop without tunneling structure was enhanced relative to that of the wild type, and the transglycosylation/hydrolysis (T/H) ratio was reduced. The opposite change was observed in Lr121 GtfB. Thus, the tunneling structure does promote sustained transglucosylation and inhibit hydrolysis.
Figure 6
The product mixture synthesized from amylose or amylopectin by Lr121 GtfBN1loop showed a 3- and 8-fold reduction in molecular weight distribution, respectively, compared to wild-type Lr121 GtfB (Figure 7). Therefore, combined with the results of the ring engineering of LrN1 GtfB and Lr121 GtfB, it can be inferred that the tunneling structure formed by the long A1 and B loops is the main determinant of the synthesis capacity of GtfB α-dextran glycosyltransferase.
Figure 7
6
Donor site loops along GtfB α-glucotransferase
Evolutionary pathways confer diversification of substrates and products
To gain insight into the catalytic changes associated with the donor site loop, the authors' team measured the kinetic parameters of LrN1 GTfB, Lr121 GTfB, and their ring-engineered variants against amylose and amylopectin, and found that the abolition of the tunnel structure makes the enzyme once again broadly substrate-specific. Analysis of the phylogenetic tree of 503 putative GtfB proteins showed that the majority (82.7%) of the GtfB sequences were expected to have tunneling structures formed by longer A1 and B loops of 17 and 20 residues that were highly conserved (Figure 8). Notably, only 17.3% of sequences exhibit unique shorter loops A1 and B. In the rest of the sequences, the shortening of these rings hints at a cracked structure. The evolutionary relationship between GtfB enzymes with tunneling structures and GtfB enzymes with cracking structures cannot be conclusively determined from phylogenetic analysis. However, it can be inferred that the sustained processivity is due to the evolution of the loops in these GtfB α-dextrantransferases. The presence of a tunnel structure contributes to the efficiency of transglucosylation and the ability to synthesize. However, this enhancement comes at the cost of reduced substrate diversity.
Figure 8
This study provides unprecedented insight into the determinants and evolutionary diversification of the persistence of GH70 α-dextran transferase, and provides a simple pathway for engineered starch to convert α-dextran transferases to produce multiple α-dextrans for different biotechnology applications.
Article Information:
https://doi.org/10.1021/acs.jafc.4c01842