今天推送的文章发表在J.Agric.Food Chem.上的“De Novo Biosynthesis of Difucosyllactose by Artificial Pathway Construction and α1,3/4-Fucosyltransferase Rational Design in Escherichia coli”，作者为江南大学的沐万孟教授。
Human milk oligosaccharides (HMOs) are the third major solid component of human milk after lactose and lipids, and are generally divided into fucosylated HMOs, non-fucosylated neutral HMOs, and sialylated HMOs. 2'-fucosyllactose (2'-FL), 3-fucosyllactose (3-FL), difucosyllactose (DFL), and lactofucose pentatose I (LNFPI) are representative fucosylated HMOs with three, four, and five monosaccharide units. DFL can reduce the risk of pathogenic bacteria adhering to the intestinal cell wall and can promote gut health by selectively promoting the growth of beneficial bacteria such as bifidobacteria. At present, few studies have reported microbial synthesis of DFL. In this study, the authors propose an alternative and economical DFL production strategy in E. coli through de novo synthesis of fucose from GDP. A combinatorial metabolic engineering strategy was adopted to achieve efficient synthesis of DFL by balancing and optimizing the four endogenous enzymes involved in the GDP-fucose pathway. In order to overcome the problem of catalytic activity in HP3/4FT, a computer-aided modification was designed and experimentally tested with the help of the Hot-Spot Wizard 3.1 server. Then, the authors artificially divided all the beneficial mutants into three regions for rational design. After three rounds of site-directed mutagenesis, a combination of five mutation sites was obtained, corresponding to the optimal strain, BLC09-58, which was able to significantly improve DFL biosynthesis. Both titers are the highest titers reported to date at shake flask and fed-batch levels.
De novo design and construction of the DFL biosynthetic pathway
Due to the narrow receptor selectivity of the two fucosyltransferases, WbgL and HP3/4FT were selected, and fucosylated lactose was sequentially performed in Escherichia coli for DFL biosynthesis by exogenous addition of fucose and lactose as substrates, glycerol and glucose as carbon sources. In the current study, DFL can be economically produced by supplementing with lactose only by adopting the de novo GDP-fucose synthesis pathway in E. coli as an alternative biosynthetic pathway that does not require the provision of fucose (Figure 1).
Specifically, an engineered strain E. coli BL21 (DE3) ΔwcaJΔlacZ (BWL) with the inactivating competition pathway genes wcaJ (encoding UDP-glucose-lipid carrier transferase) and lacZ (encoding β-galactosidase) was selected as the initial strain for subsequent DFL biosynthesis (Figure 2A). Based on the success of other fucosylated HMO biosynthesis previously, the native gene clusters involved in GDP-fucose biosynthesis, such as manC-manB and gmd-wcaG, were enhanced individually or synergistically using a strong PJ23119 promoter (Figure 2B). Engineered strains were transformed with two- and three-plasmid systems, both controlled by the T7 promoter (Figure 2C). The wbgL gene was cloned into pCDFDuet-1 and HP3/4FT was cloned into the vector pETDuet-1. The additional GDP-fucose biosynthesis genes manC-manB and gmd-wcaG were cloned into a third vector, pRSFDuet-1 (Table S1). All flask cultures were cultured in flasks containing 20 g/L glycerol and 5 g/L lactose. In this study, only extracellular DFL, FL (2'-FL plus 3-FL), lactose, and glycerol levels were measured.
As shown in Figure 2D, DFL was detected in the fermentation supernatants of 8 engineered strains, while the titers of DFL and FL differed greatly. Overall, the four strains expressing the two-plasmid system produced lower levels of DFL than the strains expressing the three plasmids. Although the strain BLC02 overexpressed the gene cluster manC-manB by replacing the strong promoter in situ, BLC01 and BLC02 produced similar amounts of DFL, corresponding to 0.06 and 0.07 g/L, respectively, and retained the same FL (0.04 g/L). This may indicate an inadequate supply of fucose in GDP, and simply improving the overexpression of the manC-manB gene cluster leads to very low levels of FL and DFL production. In contrast, strain BLC03 only overexpressed the gene cluster gmd-wcaG on the basis of strain BLC01 and produced a DFL of 1.01 g/L, which was significantly increased by 16.8 times compared with the wild-type strain BLC01. Compared with strain BLC03, strain BLC04 enhanced manC-manB and gmd-wcaG in the genome, and its DFL titer increased slightly to 1.05 g/L, while the residual FL decreased slightly (4.11 g/L). These results suggest that the overexpression of the gene cluster gmd-wcaG is more effective than simply enhancing manC-manB in terms of de novo biosynthesis of GDP-fucose. However, the corresponding BLC05 and BLC08 contained three plasmids and were additionally supplemented with plasmid pRSF-CBGW to produce DFLs of 1.02, 1.01, 1.25, and 1.31 g/L, respectively. Accordingly, FL titers were 3.53, 3.86, 3.29, and 3.17 g/L, respectively. DFL synthesis was confirmed by LC/MS analysis of the supernatant of the shake flask culture generated by BLC08 (Figure 2E).
Effect of lactose concentration on DFL biosynthesis
To test the hypothesis, the effect of lactose concentration on the DFL and FL titers and yield produced by strain BLC08 was investigated by varying lactose concentrations in the range of 1 to 5 g/L. As shown in Figure 3A, lactose is almost completely converted to DFL with only trace amounts of FL remaining, and when 1 and 2 g/L lactose are added, the yield of DFL is 0.94 mol/mol lactose consumed. However, as lactose concentration increased, the DFL production gradually decreased, while the FL production gradually increased. As shown in Figure 3B, the extracellular concentrations of DFL and FL were measured throughout the fermentation process to study the association between their concentration changes at 3 and 5 g/L lactose conditions. The concentration of FL is consistently higher than that of DFL throughout the fermentation process, regardless of whether lactose is below 3 or 5 g/L. Strain BLC08 can produce 1.75 and 3.51 g/L DFL from 1.0 and 2.0 g/L lactose, respectively, within 72 hours without the addition of fucose.
The authors replaced the plasmid pCD-wbgL with pCD-fucT2 to construct the corresponding strains BLC08 and BLC09. As shown in Figure 3C, strain BLC09 cultured in medium with different lactose concentrations (1−5 g/L) was effective in producing DFL. The concentration of DFL produced by BLC09 gradually increased from 1.81 g/L with the addition of 1 g/L lactose to 5.28 g/L with the addition of 5 g/L lactose. When the lactose concentration was less than 2 g/L, the results of BLC09 were similar to that of BLC08, whereas when the lactose concentration was greater than 2 g/L, the results were completely different between the two strains. When the lactose concentrations were 3, 4 and 5 g/L, the titers of DFL were 4.63, 4.83 and 5.28 g/L, respectively, and the yield of lactose consumed per mole was 0.83, 0.78 and 0.75 mol DFL, respectively, to 0.81, 1.03 and 1.31 g/L FL, respectively, and the lactose conversion rate gradually increased. Similar to the case of BLC08, extracellular concentrations of DFL and FL were detected throughout the fermentation process. Contrary to the product distribution presented in Figure 3B, strain BLC09 produced higher levels of DFL than intermediate FL throughout fermentation regardless of the addition of 3 g/L or 5 g/L lactose (Figure 3D).
To compare the substrate affinities of WbgL and FucT2 for 3-FL, the binding free energies of the WbgL+3-FL and FucT2+3-FL complexes were calculated and analyzed using MD simulations. As shown in Figure 4A, the binding free energy of each system is stable, and the binding free energy of FucT2+3-FL is always significantly lower than that of WbgL+3-FL. This can lead to differences in ligand binding, which can lead to a preference for 3-FL. The FucT2-expressing strain BLC09 preferentially produces DFL using 3-FL as a substrate. FucT2 exhibits excellent receptor substrate flexibility and has been shown to biosynthesize 2'-FL, DFL, and LNFPI in vivo using lactose, FL, and lactotetraose (LNT), respectively. The binding free energy of LNT is slightly lower, corresponding to a higher LNT preference, with an accumulation of 0.41 g/L for 2'-FL and 0.63 g/L for LNFPI. Based on the above, the proposed DFL biosynthesis route is shown in Figure 4B. FucT2 fucosylated lactose to generate 2′-FL, and further fucosylation catalyzed by HP3/4FT to produce DFL. At the same time, HP3/4FT competed with FucT2 for lactose to form 3-FL, and FucT2 catalyzed further fucosylation to form DFL. Thus, 2'-FL and 3-FL can be used as receptors, catalyzed by HP3/4FT and FucT2, respectively (Figure 4B).
Computer-aided screening and site-directed mutagenesis of HP3/4FT
Based on computer-aided screening and structure-based rational design, the authors made molecular modifications to HP3/4FT. Using 5ZUI as a template, the SWISS-MODEL server was used for homologous modeling of HP3/4FT (Figure S2), and the first round of mutation was designed using the HotSpot Wizard server. Location prediction and selection based on mutagenesis and amino acid frequency. Eventually, 12 mutations were identified and experimentally verified. All of these mutations are located on two helices: in particular, R43K, D44K, I46A, H48I, F49K, K52R, and Q53E are located at helical α2 near the N-terminus, while Y128R, H130D, Y131D, A133D, and M134E are located at helical α4, which belongs to the N-terminal domain (Figure 5A). Each mutation is inserted into pETDuet-1 and introduced into the host BWLCG to generate a series of BLC09-derived strains. As shown in Figure 5B, 5 of the 12 mutations produced higher levels of DFL than wild-type, consumed more lactose, and slightly lower FL residues than wild-type. Strains BLC09-05 and BLC09-10 express HP3/4FT-F49K and HP3/4FT-Y131D, respectively, yielding 6.05 and 6.35 g/L DFL, corresponding to the remaining 1.04 and 1.26 g/L FL. Taking BLC09-10 as an example, the yield of DFL was 1.2 times higher than that of BLC09, indicating that Y131D of HP3/4FT was more conducive to the catalytic synthesis of DFL.
Rational design based on mutant M32 and site-directed mutagenesis of HP3/4FT
The authors found that four of the seven residues in M32 (S45F, D127N, R128E, H131I) were located in the two helices mentioned above. Therefore, 7 residues in M32 were further analyzed, focusing on 6 parallel residues (A127/Y128/Y131/Y197/E338/R369) in HP3/4FT. Six selected residues in HP3/4FT were mutated to A127N, Y128E, Y131I, Y197N, E338D, and R369A. Subsequently, the beneficial mutants (F49K and Y131D) obtained in the first round of selection were combined with the 6 M32-based mutants mentioned above, and the 9 mutants were divided into three regions: region 1 (residue 38−54), region 2 (residue 124−136), and region 3 (residue 197-terminal) (Figure S3). Specific residues located in a single region are selected for the second round of mutations, in which a combination of single- and multi-site is used. As shown in Figure 6A, the beneficial mutants, namely F49K, Y131D, and Y197N/E338D/R369A, located in regions 1, 2, and 3, respectively, exhibited 1.15, 1.20, and 1.25-fold increases compared to wild-type HP3/4FT, respectively. In the case of regions 1 and 2, all single mutants of 7 residues showed positive performance of DFL biosynthesis in vivo, with the exception of Y128E, which showed a 10.5% reduction in DFL titers. Different combinations of mutants result in different DFL synthesis capabilities, but all are lower than the corresponding F49K and Y131D. In the case of region 3, all single- and multi-point combinations significantly increased the DFL titer in the range of 11.2−25.7%, especially the superposition of all mutation points.
A second round of mutations is performed by selecting beneficial mutants identified within the indicated region for overlapping mutations across regions. After overlapping the two regions, 17 mutants were obtained, most of which exhibited high DFL yields. For example, strains expressing the multi-point mutants F49K/A127N/Y131D (region 1 plus 2) and A127N/Y131D/Y197N/E338D/R369A (region 2 plus 3) produced DFL of 6.13 and 7.12 g/L, respectively. Finally, the three regions were overlapped to find the optimal mutant, all of which contributed to a significant improvement in DFL yield. Strain BLC09-58 containing the best mutant named MH5 (F49K/Y131D/Y197N/E338D/R369A) with residues in regions 1, 2, and 3 was identified as the best-performing mutant strain with a 1.37-fold increase in DFL biosynthesis (Figure 6B).
Structural analysis of mutant MH5
As a typical member of the GT-B family, HP3/4FT contains an N-terminal domain (NTD) containing residues 1-152 and a C-terminal domain (CTD) characterized by a Rothman-like fold, spanning residues 153 to the terminal. The best mutant MH5 has residues 49 and 131 in NTD, while positions 197, 338 and 369 are in CTD. Mutant F49K and Y131D exhibit side-chain characteristics similar to those of the wild type, and changes in the electrostatic potential environment may cause changes in structural flexibility, resulting in changes in biosynthetic production (Figure 7A). To further analyze the reasons for the difference in the yield of wild-type and mutant MH5, MD simulations were performed to determine the structural characteristics. As shown in Figure 7B, the mutant exhibits a similar tendency to flexibility as the wild-type enzyme. However, there was a significant difference in RMSF values between the NTD groups. Although the mutant F49K is located in region 1 and the mutant Y131D is located in region 2, the RMSF values in both regions are significantly higher than those of the wild type.
Demonstration of DFL biosynthesis by fed-batch fermentation
In order to comprehensively evaluate the fermentation amplification performance of the optimal variant BLC09-58, fed-batch fermentation was carried out in 5 L fermentation tank and 2 L working volume. The time course for DFL and FL generation during fed-batch fermentation is shown in Figure 8. Strain BLC09-58 continued to produce a large amount of DFL throughout the fermentation process, reaching a maximum titer of 45.81 g/L at the end of fermentation. In contrast, the residue of intermediate FL presents a very different picture. In the early stages of fermentation, no significant FL was observed, even within 44 hours. The titer of the FL residue reached less than 2.0 g/L at 77 hours and then increased at a relatively high rate to 3.04 g/L at 83 hours. This may be due to the reduced activity of the two fucosyltransferases on the FL substrate, or it may be due to the accumulation of FL due to the cleavage of some strains in the late fermentation period.
DOI:10.1021/acs.jafc.4c01691
Link to the article: https://doi.org/10.1021/acs.jafc.4c01691