Confirmation of GA performance in alkaline conditions
This study aimed to acquire the PobA mutants with high hydroxylation activity towards 3,4-DHBA. As a product GA is unstable in alkaline conditions and the degradation products can react with each other to form a phenolic mixture [33-37]. In this work, we firstly mixed GA and alkali NaHCO3. After 2 h, we found the mixture with a pH of 9.3 displayed a green color visible to naked eyes (Fig. 1A). Moreover, the green color deepened with the increase of GA concentration. Besides, UHPLC and MS were used to analyze the mixture. Fig. S1 shows 200 mg/L GA can be completely degraded in 2 hours. MS results in Fig. S2A show several new compounds were formed in the mixture. According to MS results, we speculated one of the new compounds might be ellagic acid (Fig. S2B), whose amount was highest in the mixture. Subsequently, the mixture was scanned at full wavelength (340-820 nm). Results in Fig. 1B show the mixture has a maximum absorption wavelength of 640 nm. We then confirmed the relationship between OD640 value and GA concentration of the mixture. Fig. 1C shows GA concentration exhibited linear relationship with OD640 value. These results demonstrate through adding NaHCO3, the change of GA concentration can be observed by naked eyes, and GA concentration can be confirmed through detection of OD640 value. These suggest addition of NaHCO3 in the finished reaction of PobA hydroxylating 3,4-DHBA can be used as an efficient strategy for screening high active PobA mutants.
Screening and in vitro enzyme assay of PobA mutants
According to the performance of GA in NaHCO3, we designed a complete process for screening high active PobA mutants (Fig. 1D). Firstly, error-prone PCR was conducted on gene pobA, generating a PobA mutagenesis library. The single colonies of the library were pre-incubated into 96-deep-well plates containing LB medium, and the pre-inoculum was then transferred into another 96-deep-well plates containing M9Y medium with 1000 mg/L substrate 3,4-DHBA. After 12 h, the culture samples were taken into 96-well plates which contained 0.1 M NaHCO3. After reaction for 2 hours, the samples with deepest green color were picked out, and were then re-screened through detection of OD640 value. Based on that, a high active PobA mutant (Y385F/T294A/V349A) was screened out from PobA mutagenesis library. Subsequently, this mutant was expressed and purified. SDS-PAGE in Fig. S3 shows the purity of purified Y385F/T294A/V349A PobA was greater than 95%. Enzyme assay of the purified mutant was then performed. The non-linear regression curves of PobA mutants towards 4-HBA and 3,4-DHBA through the Michaelis-Menten equation are shown in Fig. S4. The Km of Y385F/T294A/V349A PobA was 30.3 ± 10.4 μM towards 3,4-DHBA (Table 2), which was 4.22-fold lower than that of the reported mutant (Y385F/T294A PobA), suggesting that Y385F/T294A/V349A PobA has stronger affinity towards 3,4-DHBA when compared with Y385F/T294A PobA. Besides, Y385F/T294A/V349A PobA has a kcat/Km of 0.059 μM-1s-1 towards 3,4-DHBA, a 4.92-fold higher value when compared with that of Y385F/T294A PobA. Besides, the kcat/Km of Y385F/T294A/V349A PobA towards 4-HBA was 0.094 μM-1s-1, which was 5.22-fold higher than that of Y385F/T294A PobA. These indicate Y385F/T294A/V349A PobA possesses higher catalytic efficiency towards 4-HBA or 3,4-DHBA than Y385F/T294A PobA towards 4-HBA or 3,4-DHBA.
Molecular docking simulation of PobA mutants
Subsequently, molecular docking simulation was conducted to provide mechanism explanation for the high activity of Y385F/T294A/V349A PobA. Wild-type PobA (PDB code: 1IUV) was used as template for simulation of Y385F/T294A PobA and Y385F/T294A/V349A PobA. After that, molecular docking of the mutants with substrate 3,4-DHBA and cofactor FAD were conducted. As shown in Fig. 2A and B, Y385F/T294A PobA and Y385F/T294A/V349A PobA possess similar catalytic pocket. In the pocket, amino acid residues Y201, P293 and T294A of PobA mutants, and 4-OH of 3,4-DHBA composed a hydrogen-bond loop, which was same as the complex of wild-type PobA with native substrate 4-HBA [28]. Besides, 3-OH of 3,4-DHBA formed hydrogen bonds with P293 of PobA mutants and cofactor FAD. The hydrogen-bond loop is a stable binding. Based on that, we speculated the catalytic mechanism of PobA mutants towards 3,4-DHBA was similar to that of wild-type PobA towards 4-HBA [38, 39]. First, FAD cofactor in the complex is reduced by NADPH, which is responded to the binding of 3,4-DHBA to PobA mutants. Subsequently, the oxygen in environment oxidizes the reduced FAD to produce a hydroperoxide. The hydroperoxide then attacks the C-H bond at 5th position of 3,4-DHBA to generate a new hydroxyl group, forming product GA.
Compared to Y385F/T294A PobA (Fig. 2A), a new hydrogen bond was formed between S348 and V349A in Y385F/T294A/V349A PobA (Fig. 2B), which further influenced the binding of 3,4-DHBA to Y385F/T294A/V349A PobA. As a generated result, the hydrogen bond distance between P293 of Y385F/T294A/V349A PobA and 4-OH of 3,4-DHBA (1.9 Å) was 1.58-fold shorter than that of Y385F/T294A PobA and 4-OH of 3,4-DHBA, suggesting the hydrogen bond between Y385F/T294A/V349A PobA and 3,4-DHBA was stronger than that between Y385F/T294A PobA and 3,4-DHBA. Besides, the hydrogen bond distances between 3-OH of 3,4-DHBA and P293 of Y385F/T294A/V349A PobA and between 4-OH of 3,4-DHBA and Y201 of Y385F/T294A/V349A PobA was close to these in complex of Y385F/T294A PobA with 3,4-DHBA. These indicate the tight binding of 3,4-DHBA to Y385F/T294A/V349A PobA resulted in the high activity.
Bioconversion of 3,4-DHBA into GA
To test the in vivo conversion ability of PobA mutants towards 3,4-DHBA, Y385F/T294A PobA and Y385F/T294A/V349A PobA were individually expressed in E. coli BW25113 (F’), generating strains CTT1 and CTT2, respectively. 1000 mg/L 3,4-DHBA was added to the culture at 5.5 h. As shown in Fig. 3A, CTT1 accumulated 104 ± 18 mg/L GA at 6.5 h, representing an initial in vivo conversion rate of 27.9 ± 4.9 mg/L/h/OD. The OD600 value of CTT1 raised rapidly from 5.5 h to 24 h and reached 7.58 ± 0.04 at 24 h. After this time point, the growth of CTT1 stopped. Within 24 h, about half of 1000 mg/L 3,4-DHBA was consumed and 651 ± 5 mg/L GA was generated in the culture. In the next 24 hours, the conversion ability of CTT1 decreased and no more GA was produced. At 48 h, GA titer reduced to 575 ± 7 mg/L because of the degradation induced by oxidation. Meanwhile, 3,4-DHBA with a titer of 502 ± 32 mg/L was detected, indicating that about half of 1000 mg/L 3,4-DHBA cannot be converted into GA by CTT1.
Similar to CTT1, 1000 mg/L 3,4-DHBA was also fed to CTT2 at 5.5 h. The results in Fig. 3B show OD600 value of CTT2 increased steadily throughout the 48-h feeding experiment and reached 9.07 ± 0.14 at 48 h. At 6.5 h, CTT2 produced 149 ± 5 mg/L GA in the culture and displayed an initial in vivo conversion rate of 35.4 ± 1.2 mg/L/h/OD, which was 1.27-fold higher than that of CTT1. Within 36 h, 1000 mg/L 3,4-DHBA was completely consumed by CTT2 and 830 ± 33 mg/L GA was generated, representing a 75% molar conversion ratio. Significantly, the titer of GA was 1.27-fold higher than that of CTT1 at 24 h. These results suggest that E. coli BW25113 (F’) with Y385F/T294A/V349A PobA expression exhibits higher in vivo conversion ability towards 3,4-DHBA than E. coli BW25113 (F’) with Y385F/T294A PobA expression.
Bioconversion of 3,4-DHBA into pyrogallol
Further, in vivo conversion of 3,4-DHBA into pyrogallol were achieved by expressing PobA mutants and decarboxylase PDC in E. coli BW25113 (F’). Plasmids pZE-PobA** and pZE-PobA*** were individually introduced into E. coli BW25113 (F’) harboring pCS-PDC, resulting in strains CTT3 and CTT4, respectively. 1000 mg/L 3,4-DHBA was fed to CTT3 or CTT4 at 5.5 h. As shown in Fig. 3C, CTT3 possessed a high growth rate in the first 12 hours and has an OD600 value of 6.62 ± 0.05 at 12 h. Within 12 h, 9.17 ± 5.20 mg/L GA, 171 ± 26 mg/L pyrogallol and 503 ± 3 mg/L byproduct catechol accumulated in the culture. During the next 36 hours, the titer of pyrogallol increased to 222 ± 16 mg/L at 48 h, while the titer of catechol decreased to 465 ± 9 mg/L. Though CTT3 could convert 3,4-DHBA into pyrogallol, large accumulation of byproduct catechol was accompanied.
As a comparison, the results in Fig. 3D show CTT4 grew rapidly in the first 12 hours and has an OD600 value of 6.66 ± 0.11 at 12 h. Meanwhile, pyrogallol with a titer of 237 ± 21 mg/L was detected in the culture, which was 1.39-fold higher than that of CTT3 at the same time point. In addition, the byproduct catechol has a titer of 367 ± 14 mg/L, a 1.37-fold lower value when compared with that of CTT3. After 12 h, OD600 value has no significant increase, which was similar to CTT3. Significantly, the titer of pyrogallol gradually increased to 323 ± 23 mg/L at 48 h, which was 1.45-fold higher than that of CTT3. These results indicate expressing PobA mutant and PDC in E. coli BW25113 (F’) could achieve the in vivo conversion of 3,4-DHBA into pyrogallol. Moreover, Y385F/T294A/V349A PobA coupling with PDC represents higher in vivo ability of converting 3,4-DHBA into pyrogallol when compared with Y385F/T294A PobA coupling with PDC.
Establishment of the biosynthetic pathway for 3,4-DHBA production
Construction of an efficient 3,4-DHBA biosynthetic pathway was significant for achieving the de novo production of GA and pyrogallol. In previous study, 3,4-DHBA was produced from 4-HBA through expression of heterogenous PobA in E. coli [28]. For GA and pyrogallol production, heterogenous PobA required to catalyze two reactions, hydroxylating 4-HBA into 3,4-DHBA and hydroxylating 3,4-DHBA into GA (Fig. 4). Generally, the efficiency of two reactions induced by one kind of enzyme was lower than that of one reaction induced by one kind of enzyme. In this work, to avoid the issue of PobA-catalyzing two reactions and achieve efficient GA and pyrogallol production, E. coli BW25113 (F’) was engineered to produce 3,4-DHBA from 3-dehydro-shikimate (DHS) (Fig. 4). Firstly, 4-HBA biosynthetic pathway in E. coli BW25113 (F’) was blocked through knockout of gene aroE (strain CTT5) or knockout of genes aroE and ydiB (strain CTT6). AroE and YdiB are isoenzymes that can catalyze DHS to produce shikimate. Fig. S5 shows CTT5 can grow in M9 medium, while CTT6 cannot grow in M9 medium because it cannot synthesize the essential amino acids phenylalanine, tyrosine and tryptophan. These results were consistent with the theoretical expectation. Subsequently, the growth curves of CTT5 and CTT6 were measured in LB medium. As shown in Fig. 5A, the OD600 values of CTT5 and CTT6 increased with the extension of culture time. At 15 h, CTT5 and CTT6 reached maximum OD600 values, 4.13 ± 0.10 and 4.15 ± 0.13, respectively. After this time point, the growth of CTT5 and CTT6 stopped. As a comparison, the original strain E. coli BW25113 (F’) has a maximum OD600 value of 4.61 ± 0.13 at 16 h, which was close to that of CTT5 and CTT6 at 15 h. After 16 h, E. coli BW25113 (F’) stopped growing. These results suggest knockout of aroE or knockout of aroE and ydiB has no significant influence on the cell growth of E. coli BW25113 (F’).
To achieve the de novo production of 3,4-DHBA, 3-dehydroshikimate dehydratase (AroZ) which can catalyze 3-dehydro-shikimate to produce 3,4-DHBA, was individually introduced into E. coli BW25113 (F’), CTT5 and CTT6, resulting in strains CTT7, CTT8 and CTT9, respectively. Results in Fig. 5B show 3,4-DHBA titers of CTT8 and CTT9 continued to increase during the 48-h fermentation. The growth curves of CTT8 and CTT9 were similar. The OD600 values of CTT8 and CTT9 raised rapidly in first 12 hours and have no significant improvement during the next 36 hours. At 48 h, CTT8 produced 752 ± 17 mg/L 3,4-DHBA. Meanwhile, the OD600 value was 2.03 ± 0.06. For CTT9, 420 ± 26 mg/L 3,4-DHBA accumulated in the culture at 48 h, which was 1.79-fold lower than that of CTT8. These indicate the ability of strain CTT8 to produce 3,4-DHBA was higher than that of CTT9. As a comparison, CTT7 has negligible 3,4-DHBA accumulation throughout the 48-h fermentation, suggesting without knockout of aroE or ydiB E. coli BW25113 (F’) could not synthesize 3,4-DHBA in large amount. These results suggest the engineered E. coli BW25113 (F’) (CTT8 or CTT9) has ability to de novo produce 3,4-DHBA and can be used as host for de novo GA and pyrogallol production.
De novo production of GA
To achieve the de novo production of GA, plasmid pZE-AroZ-PobA** was individually introduced into E. coli BW25113 (F’), CTT5 and CTT6, generating strains CTT10, CTT11 and CTT12, respectively. The fermentation results are displayed in Fig. 6. For strain CTT10, the production of GA lasted up to 36 h. At 36 h, only 14.1 ± 1.0 mg/L GA accumulated in the culture, meanwhile, the OD600 value was 3.08 ± 0.42 (Fig. 6A). Besides, negligible 3,4-DHBA was observed in the culture, suggesting the generated 3,4-DHBA could be immediately converted into GA by strain CTT10. For strain CTT11, 3,4-DHBA and GA titers, as well as the cell growth, kept increasing throughout the 48-h fermentation (Fig. 6B). Significantly, within 48 h, 3,4-DHBA with a titer of 400 ± 17 mg/L and GA with a titer of 180 ± 31 mg/L were detected in the culture. At the same time point, CTT11 has an OD600 value of 9.10 ± 0.51. Notably, CTT11 produced 12.8-fold higher amount of GA when compared with CTT10, indicating that knockout of aroE significantly increased the ability of E. coli BW25113 (F’) to de novo produce GA. For strain CTT12, GA titer and OD600 value continued to increase during the 48-h fermentation (Fig. 6C). CTT12 has a GA titer of 46.5 ± 8.0 mg/L and an OD600 value of 4.35 ± 0.84 at 48 h. Significantly, GA titer of CTT12 was 3.87-fold lower than that of CTT11, suggesting that E. coli BW25113 (F’)ΔaroEΔydiB has lower ability to de novo synthesize GA when compared with E. coli BW25113 (F’)ΔaroE.
As a comparison, plasmid pZE-AroZ-PobA*** was individually introduced into E. coli BW25113 (F’), CTT5 and CTT6, generating strains CTT13, CTT14 and CTT15, respectively. As shown in Fig. 6D, CTT13 produced trace amount of GA (10.9 ± 2.3 mg/L) as CTT10. For strain CTT14, the increasing trends of 3,4-DHBA and GA titers, and OD600 value were similar to that of CTT11 (Fig. 6E). Within 48 h, the accumulation of 3,4-DHBA reached 344 ± 11 mg/L, which were 1.16-fold lower than that of CTT11. Meanwhile, GA production reached 301 ± 15 mg/L, a 1.67-fold higher value when compared with that of CTT11, suggesting mutant Y385F/T294A/V349A PobA has stronger ability to de novo produce GA when compared with mutant Y385F/T294A PobA. Besides, results in Fig. 6F show strain CTT15 produced 208 ± 20 mg/L 3,4-DHBA and 204 ± 17 mg/L GA at 48 h, which were 1.25- and 4.39-fold higher than these of CTT12, respectively. Compared to that of CTT14, 3,4-DHBA and GA titers of CTT15 were 1.65- and 1.48-fold lower, respectively. Overall, introducing the designed artificial pathway into E. coli could achieve GA biosynthesis from simple carbon sources. E. coli BW25113 (F’)ΔaroE demonstrates stronger ability for de novo producing GA when compared with E. coli BW25113 (F’) or E. coli BW25113 (F’)ΔaroEΔydiB. Assembling mutant Y385F/T294A/V349A PobA into GA biosynthetic pathway enabled more GA production than that of assembling Y385F/T294A PobA into GA biosynthetic pathway, which were consistent with the results of in vitro enzyme assay and in vivo conversion experiments.
De novo production of pyrogallol
E. coli BW25113 (F’) harboring pZE-AroZ-PobA** and pCS-PDC (CTT16), CTT5 harboring pZE-AroZ-PobA** and pCS-PDC (CTT17) and CTT6 harboring pZE-AroZ-PobA** and pCS-PDC (CTT18) were constructed to de novo produce pyrogallol. The fermentation results of CTT16 in Fig. 7A show trace amount of 3,4-DHBA, GA, pyrogallol and byproduct catechol accumulated in the culture throughout the 48-h fermentation. For CTT17, the OD600 value increased during the first 24 hours and has no remarkable raise during the next 24 hours (Fig. 7B). Pyrogallol and byproduct catechol titers increased throughout the 48-h fermentation, and reached 48.6 ± 12.0 and 121 ± 12 mg/L at 48 h, respectively. At the same time point, the OD600 value was 3.02 ± 0.21. Besides, the accumulation of GA cannot be significantly detected in the culture, indicating that the generated GA was immediately converted into pyrogallol. For CTT18, pyrogallol and byproduct catechol titers raised continuously in 48 hours, and reached 19.2 ± 5.4 and 43.7 ± 4.8 mg/L at 48 h, respectively (Fig. 7C). Significantly, pyrogallol and byproduct catechol titers were 2.53- and 2.77-fold lower than these of CTT17, which suggest E. coli BW25113 (F’)ΔaroE performed better than E. coli BW25113 (F’)ΔaroEΔydiB for de novo production of pyrogallol.
Subsequently, plasmids pZE-AroZ-PobA*** and pCS-PDC were co-transferred into E. coli BW25113 (F’), CTT5 and CTT6, resulting in strains CTT19, CTT20 and CTT21, respectively. As shown Fig. 7D, CTT19 hardly produced 3,4-DHBA, GA, pyrogallol and catechol as CTT16. In Fig. 7E, CTT20 continued to grow in the first 24 hours and stopped growing in the subsequent 24 hours. CTT20 yielded 67.4 ± 9.7 mg/L pyrogallol at 48 h, which was 1.39-fold higher than that of CTT17. Meanwhile, 99.7 ± 20.3 mg/L catechol accumulated in the culture, which was 1.21-fold lower than that of CTT17. These indicate the efficiency of mutant Y385F/T294A/V349A PobA was higher than that of mutant Y385F/T294A PobA for de novo biosynthesis of pyrogallol. For CTT21, pyrogallol was continuously synthesized in the first 36 hours (Fig. 7F) and has a titer of 129 ± 15 mg/L at 36 h, a 1.91-fold higher value when compared with that of CTT20. Meanwhile, only 6.12 ± 0.46 mg/L catechol were detected, which was 12.0-fold lower than that of CTT20 at 36 h. Within 48 h, pyrogallol titer decreased to 68.5 ± 5.0 mg/L and catechol increased to 15.8 ± 2.6 mg/L. These suggest CTT21 could achieve efficient de novo pyrogallol production, meanwhile, the accumulation of byproduct catechol was trace. In all, E. coli containing the designed artificial pathway could achieve the de novo biosynthesis of pyrogallol. Among the engineered strains, E. coli BW25113 (F’)ΔaroEΔydiB with overexpression of Y385F/T294A/V349A PobA and PDC demonstrates strongest ability for de novo production of pyrogallol.