Thanapon Charoenwongpaiboona,
Methus Klaewklaab,
Surasak Chunsrivirotab,
Karan Wangpaiboona,
Rath Pichyangkuraa,
Robert A. Fieldc and
Manchumas Hengsakul Prousoontorn*a
aDepartment of Biochemistry, Faculty of Science, Chulalongkorn University, Pathumwan, Bangkok, 10330, Thailand. E-mail: manchumas.h@chula.ac.th; prath@chula.ac.th; thanapon.charoenwongpaiboon@gmail.com
bStructural and Computational Biology Research Unit, Department of Biochemistry, Faculty of Science, Chulalongkorn University, Pathumwan, Bangkok, 10330, Thailand
cDepartment of Biological Chemistry, John Innes Centre, Norwich Research Park, Norwich, NR4 7UH, UK
First published on 14th May 2019
Fructooligosaccharides (FOSs) are well-known prebiotics that are widely used in the food, beverage and pharmaceutical industries. Inulosucrase (E.C. 2.4.1.9) can potentially be used to synthesise FOSs from sucrose. In this study, inulosucrase from Lactobacillus reuteri 121 was engineered by site-directed mutagenesis to change the FOS chain length. Three variants (R483F, R483Y and R483W) were designed, and their binding free energies with 1,1,1-kestopentaose (GF4) were calculated with the Rosetta software. R483F and R483Y were predicted to bind with GF4 better than the wild type, suggesting that these engineered enzymes should be able to effectively extend GF4 by one residue and produce a greater quantity of GF5 than the wild type. MALDI-TOF MS analysis showed that R483F, R483Y and R483W variants could synthesise shorter chain FOSs with a degree of polymerization (DP) up to 11, 10, and 10, respectively, while wild type produced longer FOSs and in polymeric form. Although the decrease in catalytic activity and the increase of hydrolysis/transglycosylation activity ratio was observed, the variants could effectively synthesise FOSs with the yield up to 73% of substrate. Quantitative analysis demonstrated that these variants produced a larger quantity of GF5 than wild type, which was in good agreement with the predicted binding free energy results. Our findings demonstrate the success of using aromatic amino acid residues, at position D418, to block the oligosaccharide binding track of inulosucrase in controlling product chain length.
Previously, we reported the amino acid residues of Lactobacillus reuteri 121 inulosucrase (LrInu) that play an essential role in FOS chain length determination.13 Some variants of this enzyme, such as D479A, S482A, and R483A, can produce a broad range of oligosaccharides with a small amount of accumulating polymer, while N543A mainly produces the short chain FOSs (DP3-6, with traces of DP ≥ 7). Since FOSs show optimal prebiotic activity in the DP 2–8 range,14 it is interesting to use the N543A variant of LrInu as a biocatalyst for the synthesis of such oligosaccharides. However, this variant produced low amounts of transglycosylated product, but liberates high amounts of fructose from sucrose hydrolysis. Therefore, in this study, we engineered inulosucrase to change its product chain length using computer-aided rational design. The position of LrInu selected for mutagenesis was based on the oligosaccharide binding track of LrInu reported in previous study.13 The Rosetta program was employed using the homology model of LrInu, built from crystal structure of Lactobacillus johnsonii inulosucrase (PDB ID: 2YFS),15 to assess variants of LrInu to F, Y or W, to predict the substrate binding conformations and to compute the binding free energies (ΔGbinding) of the substrate/enzyme complexes. The enzyme activities, biochemical properties and kinetic parameters of LrInu variants were determined and compared to that of the wild type. Finally, the FOSs produced by variant enzymes were analysed by TLC, HPLC and MALDI-TOF MS. As predicted by Rosetta program, this study demonstrated the effectiveness of using aromatic amino acids to block the oligosaccharide binding track and to control the size of oligosaccharides synthesised by inulosucrase.
Fig. 1 Schematic display of enzyme engineering for modulation of the size of oligosaccharide produced by glycosyltransferase. |
From previous study, D479, S482, R483 and N543 residues were predicted to be the carbohydrate binding residues of Lactobacillus reuteri 121 inulosucrase.13 Nevertheless, we noticed that the size of oligosaccharides synthesised by the variants did not correlate well with the distance between the mutated sites and catalytic sites. In the case of LrInu, for example, the N543A variant mainly produced FOSs with DP3-6, while the R483A variant, whose R483 was located close to N543, produced the FOSs up to DP12. Although these two residues were close, the sizes of their FOSs products were dramatically different. This finding suggested that there was more than one amino acid residue located in the same binding site that played an important role in substrate binding. Although one of the binding residues was mutated, other residues still could hold the substrate, allowing transglycosylation to occur.
Previous study found that blocking the oligosaccharide binding track of levansucrase with aromatic residues (F, Y and W) was an effective strategy to block elongation of polysaccharide and increase the yields of oligosaccharides.18 Therefore, we employed this approach in redesigning LrInu so that it could produce high yields of short to medium chain length oligosaccharides. Because R483 of LrInu is located next to N543, we hypothesized that blocking at this position might increase the yield of short FOSs like that obtained from the N543A variant. To test this hypothesis, R483 residue of the catalytically competent binding conformation (GF4/Fru-WT) was changed in silico to A, F, Y and W to create GF4/Fru-R483A, GF4/Fru-R483F, GF4/Fru-R483Y and GF4/Fru-R483W complexes, respectively, using the Rosetta program. This catalytically competent binding conformation (GF4/Fru-WT) was defined as a binding conformation of 1,1,1-kestopentaose (GF4) in the active site of Lactobacillus reuteri 121 inulosucrase containing fru-D272 (also see Experimental section). Since our previous study found that the R483A variant synthesised more GF5 (DP6) than the wild type, the GF4/Fru-R483A complex was used as positive control in this study. Fifty independent runs of the FastDesign protocol were employed to resolve unfavorable interactions and find low energy binding conformations of all complexes (Fig. 2). ΔGbinding of each binding conformation was computed, and the average values are shown in Table 1. We hypothesized that there might be a possible correlation between the quantity of GF5 (DP6) product and the recognition of GF4 (DP5). If an enzyme binds well with GF4, it should be able to effectively extend GF4 by one fructosyl residue to form GF5 via transfructosylation. As shown in Table 1, the Fru-R483A was predicted to bind GF4 better than the Fru-WT, suggesting a possible correlation between the quantity of GF5 products and ΔGbinding of GF4 in the active site of Fru-R483A LrInu. Furthermore, our results show that the average values of ΔGbinding of other variants are better than or about the same as that of the wild type, suggesting that they may be able to synthesise more quantity of GF5 products than the wild type as well. Therefore, the R483F, R483Y and R483W variants were selected for further kinetic study and product characterisations.
Fig. 2 The predicted catalytically competent binding conformations of (A) wild type (B) R483A (C) R483F (D) R483Y and (E) R483W variants by Rosetta. The coordinates of all models are in ESI.† |
System | ΔGbinding (REU) | |
---|---|---|
Average | s.e.m. | |
a REU = Rosetta energy unit. | ||
GF4/Fru-WT | −6.3 | 0.1 |
GF4/Fru-R483A | −7.1 | 0.1 |
GF4/Fru-R483F | −7.0 | 0.1 |
GF4/Fru-R483Y | −7.0 | 0.2 |
GF4/Fru-R483W | −6.3 | 0.2 |
When this study was conducted, the crystal structure of Lactobacillus reuteri 121 inulosucrase was not available. Therefore, its homology model was created in the previous study13 based on the crystal structure of Lactobacillus johnsonii inulosucrase (PDB ID: 2YFS)15 and also used in this study. The sequence identity of these two enzymes are reasonable with the value of 74.17%. Moreover, the quality of this homology model was evaluated by Ramachandran plots in the previous study. The results show that this homology model has promising quality because the majority of amino acid residues are in favored region (93.7%) and allowed region (5.6%). Additionally, the catalytic residues (D272, D424, and E523) of the homology model of inulosucrase from Lactobacillus reuteri 121 are in similar positions to those of inulosucrase from Lactobacillus johnsonii, and they should be in appropriate positions for the catalysis of transfructosylation. Moreover, other studies also employed homology models in creating structures of variant proteins using the Rosetta program.19–21 In any case, if a crystal structure of Lactobacillus reuteri 121 inulosucrase is available in the future and used as an input for the Rosetta program, more accurate results might be obtained.
WTa | R483F | R483Y | R483W | |||
---|---|---|---|---|---|---|
a Charoenwongpaiboon et al. (2019).13 nd = the result cannot be determined. | ||||||
Specific activity | Total | U mg−1 | 530 ± 14 | 123 ± 3 | 157 ± 6 | 125 ± 7 |
Hydrolysis | U mg−1 | 282 ± 25 | 90.2 ± 2.4 | 117 ± 5 | 88.4 ± 2.4 | |
Transglycosylation | U mg−1 | 248 ± 19 | 32.7 ± 3.5 | 40.0 ± 3.9 | 36.3 ± 5.5 | |
Kinetic parameter | kGcat | s−1 | 2060 ± 530 | 876 ± 460 | 691 ± 389 | 608 ± 373 |
KG50/m | mM | nd | nd | nd | nd | |
kGcat × (KG50/m)−1 | mM−1 s−1 | nd | nd | nd | nd | |
Hill factor | 0.44 ± 0.04 | 0.36 ± 0.03 | 0.41 ± 0.07 | 0.41 ± 0.07 | ||
kFcat | s−1 | 523 ± 62 | 120 ± 3 | 153 ± 5 | 119 ± 3 | |
KF50/m | mM | 25.3 ± 7.3 | 26.5 ± 3.0 | 24.7 ± 3.9 | 27.6 ± 2.8 | |
kFcat × (KF50/m)−1 | mM−1 s−1 | 20.7 | 4.52 | 6.18 | 4.31 | |
kG–Fcat | s−1 | 1830 ± 300 | 154 ± 35 | 269 ± 76 | 176 ± 35 | |
KG–F50/m | mM | 1200 ± 280 | 638 ± 250 | 1110 ± 470 | 733 ± 241 | |
kG–Fcat × (KG–F50/m)−1 | mM−1 s−1 | 1.52 | 0.241 | 0.242 | 0.240 |
In comparison to the previous studies, the variants of fructosyltransferase that produced the higher yield of oligosaccharides usually have a higher hydrolysis activity, such as, Y246A, N251A, K372A, R369A, R369S and R369K variants of levansucrase from Bacillus licheniformis 8-37-0-1,12 N252A and K373R variant of levansucrase from Bacillus megaterium,11 and N543S variant of inulosucrase from Lactobacillus reuteri 121.17
The kinetic parameters of the wild-type and variant LrInu were determined based on activity versus sucrose concentration curves, fitted with Hill and Michaelis–Menten equations.13,22 In previous study, we reported the kinetic parameters of wild-type LrInu. The total activity (VG) of wild-type LrInu was fitted with Hill equation, while the transglycosylation activity (VG–F) was fitted with a Michaelis–Menten equation, and hydrolysis activity (VF) was fitted with a substrate inhibition model.13 In this study, the kinetic behaviour of R483F, R483Y and R483W was also determined. The results demonstrated that the total activity (VG), transglycosylation activity (VG–F) and hydrolysis activity (VF) of all variants were best fitted with Hill equation, Michaelis–Menten equation, and substrate inhibition model, respectively, indicating that the variant LrInu exhibited the same kinetic behaviour as the wild type. As shown in Table 2, the turnover rate of all activities (kGcat, kFcat and kG–Fcat) of R483F, R483Y and R483W inulosucrases were significantly decreased compared to that of wild type, while the K50/m values of these inulosucrase variants were increased. Thus, the reduction of catalytic efficiency (kcat/K50/m) of R483F, R483Y and R483W variants were observed. The reduction of catalytic activity of variants might be result from the conformational change of the enzyme cavity, which possibly influence the proximity between C2′ atom of fructosyl intermediate and O1′ atom of acceptors.
The effects of pH and temperature on enzyme activities were also investigated in the pH and temperature ranges of 3.6–8.0 and 10–70 °C, respectively. The results showed that the optimum pH of all of the variant enzymes was not significantly changed, while the optimum temperature of R483F, R483Y and R483W was significantly shifted from 50–60 °C to 40–50 °C (Fig. 4). The change in optimal temperature of inulosucrase might be resulted from the change of enzyme kinetics or stability after mutation. However, in practise, we usually synthesised the FOSs at sub-optimal temperature because of a higher transglycosylation activity and stability. At high temperature, inulosucrase usually synthesised high amount of fructose due to the increase in hydrolysis activity.
Quantitative analysis of FOSs was performed by HPLC. The result showed that the total transglycosylation products (total FOSs) of variant LrInu were slightly decreased when compared to that of wild type (Fig. 6A). The transglycosylation products of R483A, R483F, R483Y and R483W were approximately 73%, 73%, 68% and 71% of total carbohydrate, respectively, while that of wild type was 76%. This finding indicated that the increase in hydrolysis activity of variant enzymes slightly affected the yield of total FOSs. Furthermore, although the catalytic activity of variant enzymes decreased, it can be compensated by adding more biocatalyst to reach sufficient enzyme activity. Despite the fact that this strategy might increase the cost of FOS synthesis, these variant LrInu are still useful since they produce higher valued product (bioactive FOSs). In addition, the FOS products are easier to purify, hence, the cost of purification can be reduced.
The amounts of some identified FOSs of variant and wild-type LrInu were also determined. It was found that R483F, R483Y and R483W produced a higher amount of some FOS species when compared to that of wild type (Fig. 6B). R483F produced a higher amount of DP5-8, whereas R483Y synthesised higher yield of DP4-8. Moreover, R483W also increased the yield of FOSs with DP3-7 when compared to that of the wild type. These findings supported our conclusions from computational studies which suggested that substitution by aromatic side chain would alter the product chain length specificity of LrInu.
According to the predicted ΔGbinding in Table 1, Fru-R483F and Fru-R483Y have better binding affinity to substrate GF4 (DP5) than the wild type, which may promote transfructosylation between the fru-D272 intermediate and acceptor GF4 (DP5). This would result in the accumulation of product GF5 (DP6) and support the synthesis of longer oligosaccharides. However, although the average value of ΔGbinding of the R483W variant is about the same as that of the wild type, its ability to produce DP6 was unexpectedly higher than that of the wild type. This might be the limitation of the computational protein design since no protein design software guarantees 100% successful results. Therefore, it should be confirmed by the in vitro studies. However, in the author's opinion, this methodology is still useful for protein engineering and might have a potential application for improving product specificity of other enzymes.
The recombinant plasmids were transformed into E. coli BL21 (DE3). The E. coli carrying plasmid were cultured in LB broth supplemented with 100 μg mL−1 ampicillin, 10 mM CaCl2 and 0.5% glucose, at 37 °C, shaking at 250 rpm. After the cell density reached (OD600) 0.4–0.6, IPTG was added to the final concentration of 0.1 mM. The cells were further cultured at 37 °C, shaking at 200 rpm for 18–20 h. The cells were then separated from media by centrifugation at 5000 × g for 20 min and lysed by ultra-sonication. The cell debris was removed from crude enzyme by centrifugation at 12000 × g for 20 min.
Crude enzymes were purified on a TOYOPEARL™ AF-Chelate-650M column pre-equilibrated with 25 mM potassium phosphate buffer (pH 7.4). The column was washed with the same buffer containing 20 mM imidazole and 500 mM NaCl. Finally, the enzyme was eluted with 500 mM imidazole in the previous buffer. Protein concentrations were determined by Bradford assay using a BSA as standard.
The optimal pH of wild-type and variant LrInu was measured in the pH range of 3.6–8.0 at 50 °C using the DNS assay. The buffer systems used were 50 mM sodium acetate buffer (3.6–6.0) and 50 mM bis–Tris buffer (6.0–8.0). The optimal temperature of enzymes was determined in the temperature range of 30–70 °C in 50 mM acetate buffer pH 5.5.
The TLC system used in this study consisted of 1-butanol:glacial acetic acid:water, 3:3:2 (v/v/v). The separation was performed by using TLC silica gel 60 F254 (Merck). The TLC plates were dried and stained with a solution comprising 8 mL of water, 10 mL of concentrated H2SO4, 27 mL of ethanol and 0.1 g of orcinol. The TLC were visualized by heating.
HPLC was performed on a Shimadzu™ (Prominence UFLC) instrument equipped with a refractive index detector. FOSs were separated with an Asahipak NH2P-50 4E column (Shodex™) using isocratic elution with 70% acetonitrile at a flow rate of 1 mL min−1. As described in our previous study, the amount of identified FOSs was determined using the standard curve of glucose and fructose to determine monosaccharide, sucrose to determine disaccharide, 1-kestose to determine trisaccharide, and nystose to determine longer oligosaccharides.13,29
The mass of FOSs products was evaluated by MALDI-TOF mass spectrometry (JEOL™ SpiralTOF MALDI Imaging-TOF/TOF Mass Spectrometer (JMS-S3000)). 2,5-Dihydroxybenzoic acid (DHB) was used as the matrix.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9ra02137j |
This journal is © The Royal Society of Chemistry 2019 |