Computer-aided engineering of a branching sucrase for the glucodiversification of a tetrasaccharide precursor of S. flexneri antigenic oligosaccharides | #itsecurity | #infosec


Computer-aided re-design of BRS-B Δ2 branching sucrase active site

With the aim of performing site-selective (mathrm{alpha })d-glucosylation of tetrasaccharide ABC’D’ and thus enlarging accessible S. flexneri pentasaccharide diversity, we focused our work on the redesign of the active site of a branching sucrase from Leuconostoc citreum NRRL B-742 named BRS-B. To facilitate recombinant enzyme expression in Escherichia coli, a truncated variant -called BRS-B Δ2- was constructed by removing 153 amino acids from the N-terminus of BRS-B Δ1 enzyme31. This variant showed the same specific activity as the parental BRS-B enzyme or BRS-B Δ124 and had better soluble expression (data not shown). As the three-dimensional structure of this enzyme was unknown, we decided to construct a 3D-model of BRS-B Δ2 using as template the branching sucrase ΔN123-GBD-CD2, for which an X-ray structure is available (PDB ID: 3TTQ). BRS-B Δ2 contains 1053 amino acid residues and shares 49% identity with ΔN123-GBD-CD2. When considering only the active site of the enzymes, the sequence identity increases to 60%, indicating a high conservation of active site residues. This allowed construction of a 3D-model of BRS-B Δ2 by comparative modelling and opened the way to computer-aided design approaches illustrated in the framework of Fig. 2A. Using the 3D model, pentasaccharide products characteristic of prevalent S. flexneri serotypes, ABC’[E(1 → 4)]D’ (S. flexneri 1a/1b), AB[E(1 → 4)]C’D’ (S. flexneri 2a) and [E(1 → 3)]ABC’D’ (S. flexneri 3a), were docked in the active site (Fig. 3). The crystallographic structure of homologous GTF180 glucansucrase (PDB ID: 3HZ3) in complex with sucrose bound in the active site32 was used as template to guide docking of the different pentasaccharides. More particularly, pentasaccharides were initially constructed using 3D coordinates of the sucrose glucosyl moiety from the crystallographic complex. Systems were subsequently subjected to simulated annealing (from 0 to 350 K in 100 ps and vice versa) in vacuum with harmonic positional restraints of 50.0 kcal/mol/Å2 on the enzyme, the glucosyl unit, and the sugar pucker rings of ABC’D’. The lowest energy systems from each simulated annealing were then selected as starting points for the computational enzyme design procedure. After excluding the catalytic residues (the nucleophilic D671, the acid/base E709 and the transition state stabilizer D1136) and other amino acid residues described as important for either catalysis or sucrose recognition (R669, H787, Y144 and D1183 identified by homology with GTF-180)32, we selected 27 mutable (or designable) positions in total: 24 in catalytic domain A and 3 in domain B—(G594, W595 and F596 residues) (Fig. 2B) on the basis of Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) calculation (Figure S1) and careful visual inspection.

Figure 2

(A) Computer-aided approach for BRS-B Δ2 active site re-design. After successive steps, the final designed library contained 49 sequences featuring between 9 and 15 mutations in the active site, each. (B) View of the active site of BRS-B Δ2 model. The 27 redesignable amino acids are shown in red, green, cyan and magenta according to their belonging to the first, second, third and fourth contact shell, respectively, surrounding the glucosyl residue (yellow sticks) in the -1 position as extracted from the structure of inactive GTF-180 in complex with sucrose (PDB ID: 3HZ3). The docked pentasaccharides are not shown for clarity purpose. Molecular graphics were prepared using PYMOL 1.7 (PyMOL Molecular Graphics System, Schrödinger, LLC).

Figure 3

View of the space occupied by the three gathered pentasaccharides targeted by the design bound in the active site of parental BRS-B Δ2 (A) as well as the targeted amino acids in each of the designs: ABC’[E(1 → 4)]D’ (S. flexneri 1a/1b) targeted by mutants M1-M16 from Group I is shown in (B); AB[E(1 → 4)]C’D’ (S. flexneri 2a) targeted by mutants M17-M34 from Group II is shown in (C); [E(1 → 3)]ABC’D’ (S. flexneri 3a) targeted by mutants M35-M49 from Group III is shown in (D). Molecular graphics were prepared using PYMOL 1.7 (PyMOL Molecular Graphics System, Schrödinger, LLC).

Most of these positions belong to loops with the exception of R731, D732, K735, D793 and H797, which are located in α-helices. Overall, these 27 residues are scattered over the catalytic site along four successive contact shells, due to the sliding of the ABC’D’ core with respect to the -1 subsite where the glucosyl residue is bound (Fig. 2B). Moreover, all selected residues are found highly variable among all known GH70 enzymes with the exception of amino acids L627 and A672 from the first shell which are strictly conserved (Table S1). It should be noticed that, if all 27 positions were considered to be mutable by one of the 20 possible amino acids, the theoretical combinatorial sequence space would have been as large as 1.34 × 1035, clearly out of reach of currently available experimental screening methods and computational approaches. To explore and reduce this combinatorial space, we thus considered an enzyme design protocol based on the sampling of mutations at each of the 27 positions by performing 60,000 independent runs (20,000 per system) of Rosetta Enzyme_Design (https://www.rosettacommons.org/), which considers the backbone and side chain flexibility of amino acids. The designed sequences were subsequently filtered using the Rosetta scores corresponding to enzyme:pentasaccharide binding interaction for each of the three systems, followed by a second round of sequence filtering based on the docking of sucrose donor in each mutant and estimation of its binding interaction (Figure S2). Finally, in order to limit sequence redundancy and enhance sequence variability in the final set of sequences, all designed sequences were clustered based on the percentage of sequence identity. The best Rosetta scores (with respect to pentasaccharides and sucrose) of each cluster (total of 49) were then selected for experimental evaluation. The set of the corresponding 49 sequences (Table S2) contained between 9 and 15 mutations located between positions 595 and 811. These sequences were classified in three groups depending on the targeted pentasaccharide: group I (M1 to M16) for ABC’[E(1 → 4)]D’ (S. flexneri 1a/1b), group II (M17 to M34) for AB[E(1 → 4)]C’D’ (S. flexneri 2a) and group III (M35 to M49) for [E(1 → 3)]ABC’D’ (S. flexneri 3a) (Fig. 3).

Recombinant production of mutants

When using conditions for recombinant expression in E. coli of parental BRS-B Δ2 enzyme, formation of inclusion bodies was observed for the 49 mutants. In order to enhance solubilization and prevent aggregation, we selected at random the clones expressing mutants M14 and M34, which contained 15 and 11 mutations, respectively, and attempted their production in the presence of different combinations of chaperone proteins33 (Table S3), as well as by growing the cells at 21 °C and using another optimized auto-inducible medium34. Using co-expression with plasmids pG-KJE8 (coding for dnaK, dnaJ, grpE, groES, groEL chaperones) or pTf16 (coding for tig), mutant proteins were partially recovered in the soluble fraction, showing the highest amount of soluble proteins after 24 h of culture when using plasmid pTf16, and 32 h when using plasmid pG-KJE8. The optimization of the soluble expression of mutant M21 is presented in Figure S3 as an example. Based on these results, the 49 mutants were successfully produced using the optimized culture conditions and chaperones encoded by either plasmid pTf16 or pG-KJE8.

Screening of the mutant library for the ability to use sucrose as donor substrate

In order to assess the capacity of the mutants to utilize sucrose, a colorimetric “ON/OFF” assay was set up on sucrose as sole substrate, based on the use of dinitrosalicylic acid (DNS) to measure the amount of reducing sugars (i.e. fructose, glucose) released after substrate hydrolysis. The soluble fractions were incubated for 70 h with substrate and reducing sugar production was determined at 540 nm. As shown Fig. 4-panels A, C, and E, all the mutants were strongly impacted for sucrose consumption compared to the parental enzyme.

Figure 4

(A, C, E): Absorbance at 540 nm determined after the reducing sugar “ON/OFF” assay using crude enzyme extracts of the 49 mutants co-produced with either tig (Tf16, light salmon) or dnaK, dnaJ, grpE, groES, groEL chaperone proteins (KJE8, dark cyan) in presence of 100 g L−1 sucrose during 70 h incubation. The parental enzyme BRS-B Δ2 is indicated as a reference. Dashed lines indicate absorbance level of the inactive mutant BRS-B Δ2 E709Q. (B, D, F) boxplot representing the absorbance data. Parental enzyme BRS-B Δ2 was excluded from the analysis. Data points (diamonds) pinpoint the 16 mutants selected for further evaluation.

Out of the 49 mutants, the boxplot analysis enabled us to retain 16 mutants found in the top tertile (and above) of the absorbance boxplot (Fig. 4 panels B, D and F). These 16 mutants were purified to homogeneity. Among them, four mutants quickly aggregated after purification (M8, M12 and M17 produced with Tf16 and M11 produced with KJE8), indicating that their stability was impacted by the mutations. Active mutants (12 out of 49 screened, 24.5% active mutants in total) belonged to all three groups, with 2 from group I, 7 from group II, and 3 from group III.

These 12 mutants (M6, M14, M18, M21, M23, M30 and M41 produced with Tf16 and M28, M31, M34, M35 and M40 produced with KJE8) were retained for further evaluation of their ability to glucosylate ABC’D’. Meanwhile, the specific activity toward sucrose was found to be only 1.9% of that of the parental enzyme for M21, 0.2% for M23 1.2% for M30, 0.1% for M34, and 1% for M35 (Table 1). The specific activity of the other mutants (M6, M14, M18, M28, M31 and M40) could not be determined, indicating a tremendous loss of activity toward sucrose donor substrate for all these mutants.

Table 1 Specific activity on sucrose (in U mg−1 of purified enzyme) was determined for mutants M21, M23, M30, M34 and M35. a data from ref 35. n.d.: not determined.

Transglucosylation of tetrasaccharide ABC’D’

All 12 selected mutants were tested for their ability to glucosylate tetrasaccharide ABC’D’ and their products were analyzed by LC–MS. Reaction pH was set to 5.75 despite the ABC’D’ stability decrease at this pH24, due to the quick aggregation of mutants when lowering the pH. With the exception of two mutants (M14 and M18), all selected mutants were able to transfer glucosyl moieties onto ABC’D’, yielding at least one and up to four mono-glucosylated products (Fig. 5). Overall, six different pentasaccharide products were detected, which were named P1, P2, P2’, P2’’, P3 and P3’ based on their RP-HPLC retention time (tr) and molecular mass (Figure S4). P2 and P2’ are co-eluted products (tR = 21.9 min) hardly distinguishable by LC–MS. However their chemical structures have been characterized by NMR spectroscopy as being distinct pentasaccharides in prior work24. Mutant M6 showed a profile similar to that of the parental enzyme BRS-B Δ2 and produced mainly P1 (tR = 21.6 min) together with a small amount of P2/P2’. Relatively, mutant M34 produced significantly lower amounts of P1 and increased amounts of P2/P2’. M28 and M31 were found to produce only P2/P2’.

Figure 5

(A) Distribution of the products formed by BRS-B Δ2 enzyme and its mutants, and determined from HPLC–UV analyses of 16 h reaction mixture with 50 mM ABC’D’, 1 M sucrose at pH 5.75 and variable amounts of purified enzymes in 50 µL reaction volume. Mass Spectrometry revealed a mass increase of 162 g mol−1 compared to ABC’D’, corresponding to mono-glucosylated tetrasaccharides. M14 and M18 did not produce any pentasaccharide product (not shown). (B) An enlarged view of the products (notably novel products P2″ and P3’) formed by M35 and M40 is shown on the right panel.

Another group of mutants composed of M21, M23, M30 and M41 shared a common product profile (Fig. 5). While they all produced various ratio of products P1 and P2/P2’, they also formed a novel pentasaccharide named P3, seemingly not at all synthesized by the parental enzyme. Mutant M21 was found to produce P3 in larger amount. Interestingly, based on its retention time (tR = 16.3 min), P3 was found to be also produced by mutant W2135S-F2136L of the enzyme ΔN123-GBD-CD2 in our earlier report21. In addition, two compounds with distinct retention times compared to already characterized pentasaccharides are observed leading us to assume the formation of novel compounds which were named P2″ (tR = 22.1 min) and P3’ (tR = 15.3 min). While P2″ is observed for M35, P3’ is produced in trace amounts by M40.

Structural characterization of the formed products

Mutant M21 forming products P1, P2/P2’ and P3 was selected to produce sufficient amounts of pentasaccharides for structural characterization using high field NMR spectroscopy. M21 was produced and purified to homogeneity and used to carry out a 1 mL-scale reaction in presence of ABC’D’ and sucrose. The structure of P1 was previously determined without ambiguity24, therefore only products P2/P2’ (5.0 mg) and P3 (1.6 mg) were isolated. The 1D 1H NMR spectrum obtained for P2/P2’ isolated using M21 was perfectly superimposed to the one previously obtained using mutant F2163G of ΔN123-GBD-CD221 (Figure S5). This confirmed that glucosylation occurred on OH-4A, corresponding thus to P2’. The analysis of the HSQC spectra of P3 showed shifted resonances of the glucosylated positions and adjacent atoms (Figure S6); the glucosylated 4B carbon was high frequency shifted from 72.2 ppm to 80.9 ppm, while the adjacent 3B and 5B were shifted to lower frequency, from 69.8 ppm and 69.6 ppm to 68.6 ppm and 68.4 ppm, respectively. The cross-correlation peaks of the shifted B and E carbon resonances were assigned using Double Quantum Filtered COrrelation (QDF COSY) (Figure S7). P3 was thus confirmed as being glucosylated on OH-4B.

Given the very low amounts of P2″ and P3’ pentasaccharides produced by mutant M35 and M40, respectively, we turned to a mass spectrometry (MS) based approach for their identification. This analytical method is more sensitive (requires a few tens of ng of compounds), avoids laborious production and purification steps by enabling the analyses of complex mixtures coupled with appropriate UHPLC method, and allows a structural characterization by using tandem MS (MS/MS). For reference purpose, the M21 reaction mixture was analyzed in the same conditions. UHPLC-MS analysis of M21, M35 and M40 reaction mixtures using a porous graphitized carbon column (Figure S8) revealed the presence of three pentasaccharide isomers produced by M21 (tR = 21.12 min, tR = 22.70 min, tR = 22.93 min) and four isomers independently produced by M35 and M40 (tR = 21.16 min, tR = 21.96 min, tR = 22.07 min and tR = 22.71 min for M35 ; tR = 21.16 min, tR = 21.90 min, tR = 22.71 min and tR = 22.86 min for M40). Each of these species was characterized by UHPLC-MS/MS using collision induced dissociation (CID). By this approach, we faced the high lability of the chloroacetyl moiety. The collision energy had to be adjusted compared to classical MS/MS CID based approach to allow the production of interesting fragments. This explained that we had to magnify the low mass area on the figures below m/z 825, blue part) on the spectra for each species.

First, the structure of the produced molecules shared between M35 and M40 was studied in parallel with the HPLC–UV, NMR and UHPLC-MS data. Indeed, tandem mass spectrometry data using CID fragmentation is not sufficient on its own to characterize all the structural details. For example, tandem MS spectrum of the product P1 (M35-4 and M40-3, glucosylated on OH-6D’) at tR = 22.71 min is shown Figure S9. If it remains impossible to discriminate the branching of the hexose between the 4 or 6 hydroxyl groups of the glucosamine D′, the parallel with the NMR data confirmed the glucosylation on OH-6D’. The MS/MS spectrum of the product at tR = 21.96 for M35-2 and tR = 21.90 for M40-2 (the slight shift in retention time between the two samples for P2/P2’ can be explained by the presence of the partially co-eluted species is presented Figure S10). As shown by the HPLC–UV analysis, the OH-4A is absent in the samples so P2 was confirmed as being glucosylated on OH-3A. The MS/MS analysis of product M35-1 and M40-1 at tR = 21.16 min (Figure S8) is presented Figure S11. This analysis confirms a glucosylation on rhamnose B but it was impossible to decipher between glucosylation on OH-3B (P3’) or OH-4B (P3) from the CID experiments. However, from the HPLC–UV, NMR and UHPLC-MS data of the M21 reaction mixture, with M21-1 eluting at tR = 22.12 min, we assigned the P3 product at tR = 21.16 min (Figure S8). Interestingly, P3 was not detected in the M35 samples using HPLC–UV analysis. This reveals that the UHPLC-MS approach is more sensitive and can reveal a highest diversity in biological medium.

Concerning the molecules specific of each sample, the MS/MS spectrum analysis of the last eluted molecule M40-4 at tR = 22.86 min is shown in Fig. 6. As discussed previously, CID experiments did not allow to differentiate glucosylation on OH-3B (P3’) or OH-4B (P3). However, by deduction from the identification of P3 at tR = 21.16 min, the positioning of the branched hexose on rhamnose B of the tetrasaccharide was validated as being OH-3B. This new product corresponds to P3′.

Figure 6

UHPLC-ESI–MS/MS spectrum of the pentasacharide M40-4 isolated as [M-H] at m/z 1038.21 at tR = 22.86 min in samples from M40 validated as being substituted at OH-3B (P3’). The blue area of the spectrum is enlarged by a factor of 4 in the intensity axis. Annotations in red correspond to intact product ions. Annotation in blue correspond to product ions with a loss of CHCl3. Annotations in purple correspond to product ions with a loss of COCHCl. Annotation in orange correspond to product ions with a loss of CHCl3 + COHCl3. All these labile function losses are in agreement with the structure. Red * indicate consecutive fragmentations, indicate H2O loss, indicate HCl loss. For clarity, only one fragment per pair (B, C and Y, Z) was reported on the annotated structure.

The MS/MS spectrum analysis of the partially coeluted species M35-3 at tR = 22.07 min is shown Fig. 7A. We integrated only the end of the chromatographic peak in order to exclude a cross contamination with the fragments acquired for the P2 species at tR = 21.96 min. The fragmentation spectrum obtained is really close from the MS/MS data obtained for the oligosaccharide P2 (glucosylation on OH-3A, Figure S10). However, as illustrated in Fig. 7B and Fig. 7C, there is a significant difference in the low mass range between the spectra acquired for P2 and the new product P2’’, respectively. On Fig. 7B, for the P2 structure, intracyclic fragments 1,5A1 at m/z 265.1, the corresponding water loss at m/z 247.1, and the fragment at m/z 221.1 which can be attributed to 1,3A1 or 2,4A1 (isobaric fragments) were observed. These fragments are missing for the new product. These indirect proofs lead to the structure P2″ (glucosylation on OH-2A).

Figure 7

(A) UHPLC-ESI–MS/MS spectrum of the pentasacharide isolated as [M-H] at m/z 1038.21 at tR = 22.07 min in samples M35 validated as being glucosylated on OH-2A. (B) Zoom in the mass range m/z 200–270 of the P2 product (glucosylation on OH-3A). (C) Zoom in the mass range m/z 200–270 of the P2″ product (glucosylation on OH-2A). The blue area of the spectrum is enlarged by a factor of 2 in the intensity axis. Annotations in red correspond to intact product ions. Annotation in blue correspond to product ions with a loss of CHCl3. Annotations in purple correspond to product ions with a loss of COCHCl. Annotation in orange correspond to product ions with a loss of CHCl3 + COHCl3. All these labile function losses are in agreement with the structure. Red * indicate consecutive fragmentations, indicate H2O loss, indicate HCl loss. For clarity, only one fragment per pair (B, C and Y, Z) was reported on the annotated structure.

Finally, all the structures deciphered by the use of HPLC–UV, NMR and UHPLC-MS/MS are summarized in Fig. 8.

Figure 8

Diversity of pentasaccharides obtained with the BRS-B Δ2 mutants. The various products are shown on the basis of the glucosylated subunit (A, B or D’). P1 ABC’[E(1 → 6)]D’, characterized in ref 24. P2 [E(1 → 3)]ABC’D’ and P2’ [E(1 → 4)]ABC’D’ characterized in ref 21. P3, P3’ and P2″ were characterized in this study to be A[E(1 → 4)]BC’D’ and A[E(1 → 3)]BC’D’, and E[(1 → 2)]ABC’D’, respectively. P3 was characterized by NMR spectroscopy and P3’ and P2″ by MS/MS, respectively. ClAc: chloroacetyl. Drawing of chemical structures done using CHEMDRAW (PerkinElmer).

Insight on the impact of mutations on product profiles

BRS-B Δ2 naturally favors glucosylation of the primary hydroxyl group OH-6D’ from ABC’D’. Group I comprising 16 mutants (M1 to M16) aimed at introducing beneficial mutations in order to obtain ABC’[E(1 → 4)]D’ (Fig. 4A). Redesigning D’ binding site required the introduction of 12–15 mutations to unclutter the active site and rend the hindered secondary OH-4D’ accessible to the glucosyl moiety. Such mutations turned out to have a highly detrimental impact on sucrose binding and cleavage, which could be explained by enzyme misfolding or stability loss. In the presence of ABC’D’, mutants from Group I kept the specificity of their parental enzyme, producing only pentasaccharides P1 and P2/P2’ but in different proportions. In spite of its 13 mutations and similarly to the parental enzyme, mutant M6 revealed a preference for D’ subunit glucosylation to produce P1.

Group II encompassing 18 mutants (M17 to M34) aimed originally at glucosylating OH-4C’, the sole available hydroxyl function of C’ moiety (Fig. 4C). Accessibility of this secondary hydroxyl group is highly hindered due to the constrained β-1,3 linkage between rhamnosyl residues B and C’, and the presence of the three protecting groups at C’ and D’ subunits. Introduction of 10–13 mutations were proposed by the design. These mutations affect sucrose recognition less severely as seven mutants were found relatively active toward sucrose and successfully glucosylated ABC’D’, producing up to four distinct pentasaccharides (P1, P2, P2’ and P3) based on retention time. The versatility of these mutants to accommodate ABC’D’ in different manners led to broader product specificity yielding novel molecular diversity (Fig. 6) but also resulted in a loss of selectivity (Fig. 5). Unlike the parental enzyme, most mutants were shown to glucosylate preferentially rhamnose A, forming P2/P2’ glucosylated at OH-3A and OH-4A. M28 and M31 even showed exclusive glucosylation of the A residue. Mutants M21, M23 and M30 revealed their ability for inner chain glucosylation at the B moiety, producing P3, which was never reported to be synthesized by native branching sucrases.

Group III gathers 15 mutants (M35 to M49) containing between 9 and 12 mutations aiming at favoring end chain glucosylation, targeting OH-3A or OH-4A to improve the production of P2/P2’, only weakly achieved by parental enzyme (Fig. 4D). Here again, the introduced mutations turned out to drastically affect sucrose recognition. However, all three mutants (M35, M40 and M41) revealed their ability to form, in addition to P1 and P2/P2’, a new product not synthesized by the parental enzyme. Glucosylation of OH-2A was observed for the first time with mutant M35, yielding pentasaccharide P2″ and enabling glucosylation of the third hydroxyl group from the A moiety, although of less interest in the S. flexneri context as this position is involved in chain elongation. Mutants M40 and M41 enabled glucosylation of the inner chain B rhamnose, providing respectively access to P3’ (glucosylated at OH-3B) and P3 (glucosylated at OH-4B). Here again, with these mutants, we gained access to all possible glucosylation patterns of the B moiety.

Computational protein design undertaken here was highly challenging as four subsites had to be re-designed to improve tetrasaccharide ABC’D’ accommodation for each targeted pentasaccharide without losing the affinity for sucrose. The huge combinatorial sequence corresponding to the 27 selected positions of mutations was drastically reduced thanks to the design. By screening a very limited set of 49 sequences, containing a high number of mutations (between 9 and 15 depending on the design), several mutants were successfully isolated. Impressively, after a challenging structural analysis of the products, we found out that these mutants enabled the glucosylation of six out of the eight hydroxyl groups of the lightly protected ABC’D’ acceptor substrate. It is noteworthy that three of the resulting pentasaccharides were characteristic of S. flexneri type-specific O-Ags (4a/4b, 5a and 3a) for which no enzymatic route has been proposed yet. Remarkably, two mutants (M40 and M41) showed a product profile in line with the design expectations. Experimental deconvolution of the mutations in these mutants could help to better understand the contribution of each mutation and their combinatorial effect.

These results highlight the difficulties of redesigning an active site as large and exposed as that of branching sucrases for the selective glucosylation of a structurally complex and chemically modified molecule, presenting no similarity with the natural substrate. This required introduction of a large charge of amino acid mutations to target glucosylation at various hydroxyl positions but also made it difficult to control the enzyme selectivity to produce a single pentasaccharide. This effect was further pronounced due to the exposure and flexibility of the active site. Furthermore, re-engineering enzymes catalyzing multi-step reactions from multiple substrates considerably enhanced complexity of the design, requiring multi-objective optimization i.e. in our case the ability to utilize sucrose donor and novel specificity toward an unnatural acceptor.

Another limitation in the computational redesign was the lack of crystallographic structure that led us to assume that reliable 3D-modelling could be performed given the high sequence identity between BRS-B Δ2 and the sole GH70 branching sucrase of known structure to date (ΔN123-GBD-CD2). Nonetheless, the suggested flexibility of several loops surrounding the active site36 could drastically alter topology of the active site and recognition of acceptor ABC’D’. Undoubtedly, accuracy of the design would have benefited from the determination of crystallographic structures of the enzyme in complex with the acceptor or with the products. However, a successful outcome of the design would still not be warranted given the many limitations still faced by computational protein design methods such as the poor integration of molecular flexibility and conformational rearrangements21, the under consideration in energy functions of entropy penalty and solvent effect, etc.



Original Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

− one = 1