Enhancing 2-Pyrone Synthase Efficiency by High-Throughput Mass-Spectrometric Quantification and In Vitro/In Vivo Catalytic Performance Correlation
Graphical Abstract
To rationalize how the enzymatic performance in vivo is correlated with the enzyme catalytic parameters measured in vitro, we have applied a mass-spectrometry based high-throughput quantitative method as well as synthetic biology tools to systematically improve the performance of 2-pyrone synthase for triacetic acid lactone biosynthesis in E. coli.
Abstract
Engineering efficient biocatalysts is essential for metabolic engineering to produce valuable bioproducts from renewable resources. However, due to the complexity of cellular metabolic networks, it is challenging to translate success in vitro into high performance in cells. To meet such a challenge, an accurate and efficient quantification method is necessary to screen a large set of mutants from complex cell culture and a careful correlation between the catalysis parameters in vitro and performance in cells is required. In this study, we employed a mass-spectrometry based high-throughput quantitative method to screen new mutants of 2-pyrone synthase (2PS) for triacetic acid lactone (TAL) biosynthesis through directed evolution in E. coli. From the process, we discovered two mutants with the highest improvement (46 fold) in titer and the fastest kcat (44 fold) over the wild type 2PS, respectively, among those reported in the literature. A careful examination of the correlation between intracellular substrate concentration, Michaelis-Menten parameters and TAL titer for these two mutants reveals that a fast reaction rate under limiting intracellular substrate concentrations is important for in-cell biocatalysis. Such properties can be tuned by protein engineering and synthetic biology to adopt these engineered proteins for the maximum activities in different intracellular environments.
Introduction
Metabolic engineering seeks to optimize intracellular enzymatic reactions to produce desired compounds from biomass without protein purification.1-4 However, most of the properties of enzymes used in the metabolic engineering, such as substrate binding affinity, turnover number, yield, and stability are measured in vitro using purified enzymes. Because the environment and conditions in cells are quite different from those in vitro, the properties measured in test tubes are often not reflected in the performance in vivo.5 Therefore, in order to realize the full potential of metabolic engineering, it is very important to understand how they correlate with each other to ultimately guide the next generation biocatalyst design.
As a primary example of metabolic engineering, triacetic acid lactone (4-hydroxy-6-methyl-2-pyrone, TAL) has been ranked as one of the top ten bioprivileged molecules to produce valuable bioproducts.6 TAL is one of the simplest polyketides in nature. It is an important precursor to not only commodity chemicals such as preservatives, additives, and fragrances, but also many pharmaceuticals, due to the pyrone scaffold with rich biological activities (Figure 1A).7 TAL is naturally synthesized by a type III polyketide synthase (PKS) called 2-pyrone synthase (2PS, Figure 1B). Through two rounds of Claisen condensation, TAL is produced from one molecule of acetyl coenzyme A (CoA) and two molecules of malonyl-CoA (Figure 1C and Supplementary Figure S1).8 Such prominent applications of TAL have led to many metabolic engineering studies to improve the titer of TAL in bioreactors by genome-wide engineered eukaryotic hosts to over 2.2 g/L in Saccharomyces cerevisiae,9 28 g/L in Rhodotorula toruloides10 and 36 g/L in Yarrowia lipolytica.11
TAL conversion, biosynthesis and mass-spectrometry based quantification methods. (A) TAL-derived molecules. Molecules used as food preservatives, additives and fragrances are shown in blue. Molecules used as bifunctional intermediates and building blocks are shown in red. (B) Crystal structure of the homodimer 2PS (PDB : 1EE0). The ligand acetoacetyl-CoA is shown in atoms on the left and in surface on the right. (C) 2PS catalyzed TAL biosynthetic pathway. (D) LESA-MS quantification workflow. An in-house fully 13C-labeled TAL sample was prepared by feeding the microorganism expressing WT-2PS with 13C-labeled glucose, which was utilized as an internal standard. Such liquid culture containing the internal standard was transferred and mixed equally with the liquid cultures of 2PS mutants onto a glass slide using an acoustic liquid handler. This mixture droplet was dried on the slide, automatically extracted with solvents by LESA and directly injected into MS for quantification.
While much progress has been made in the metabolic engineering of TAL production, the catalytic efficiency of 2PS toward malonyl-CoA can only reach ~102 M−1 s−1, which is far below the average performance of other enzymes operating in secondary metabolism.12, 13 Therefore, in addition to above metabolic engineering approaches, protein engineering has played key roles in improving biocatalytic performance and creating new-to-nature reactions in living organisms.14-17 For example, alternative enzymes such as 6-methylsalycilic acid synthase have been redesigned to produce 1.8 g/L TAL in Saccharomyces cerevisiae.18, 19 Additionally, a PKS-independent thiolase pathway was recently discovered to produce ~0.86 g/L TAL in engineered E. coli.20 Besides applying alternative enzymes, wild type 2PS (WT-2PS) from Gerbera hybrida has been engineered to improve its catalytic efficiency. Originally, 0.47 g/L TAL titer was achieved in an E. coli strain by WT-2PS in a controlled fermentation. A TAL-biosensor was designed to allow extensive screening of 2PS mutants through saturation mutagenesis and such a method found a L202G/L261N-2PS mutant that resulted in ~2 g/L TAL titer in an E. coli strain and 19-fold improvement of catalytic efficiency.12, 21, 22 In a parallel effort, structure-based site-directed mutagenesis was carried out to rationally generate a C35S/C372S-2PS mutant with 2-fold improvement in catalytic efficiency, resulting in 10 g/L TAL in Saccharomyces cerevisiae.23 In a recent report from our group, we have rationally engineered a 2PS mutant named I201V/L202T/L261G/I343S-2PS to accommodate aromatic-CoAs instead of acetyl-CoA for styrylpyrone biosynthesis with 70-fold improvement in catalytic efficiency over WT-2PS and 30-fold over native styrylpyrone synthase.24
Despite the progress of protein engineering mentioned above, there are many more residues around the active site of 2PS that may impact substrate specificity, chain elongation, and reactivity.24-27 More importantly, there are numerous combinations of the mutations that may be difficult to discover. Directed evolution of the 2PS can address these issues, but the progress has been hampered by the lack of efficient methods to detect and quantify TAL. Currently, TAL quantification is accomplished by time-consuming HPLC methods (11–30 minutes for each sample) that require sample pretreatments.10, 12 Methods for faster detection using a biosensor and plate reader have also been reported.12, 21, 22 However, TAL is generally produced and screened directly in cell cultures, and such cell medium environment can be drastically different for each mutant. Furthermore, the UV-Vis spectrum of TAL (λmax≅287 nm) overlaps with that of cell growth media (Supplementary Figure S2). These factors influence the accuracy and detection limit in different workflows, which only allow for semi-quantitative analysis. Therefore, a reliable and sensitive detection method for TAL is crucial for the efficient screening of 2PS mutants.
To address the above issue, we resort to mass spectrometry (MS), which is a universal and label-free approach for metabolites detection and has been applied to a variety of protein engineering applications such as the acoustic mist ionization MS.28-33 Recently, we have developed a quantification method using liquid extraction surface analysis mass spectrometry (LESA-MS) for bioproducts quantitative screening directly from liquid culture with a throughput of 1 minute per sample (Figure 1C).34 Taking advantage of this advancement, we herein report applying this LESA-MS as an automatic, high-throughput and quantitative screening method to track the directed evolution trajectory of 2PS and systematically investigate how structural changes influence 2PS reactivity. The most reactive 2PS mutant to date for TAL biosynthesis, with 25-fold improvement in catalytic efficiency and 46-fold improvement in titer, is discovered and characterized. Furthermore, by carefully examining the relationship of intracellular substrate concentration, Michaelis-Menten parameters and TAL titer with synthetic biology tools, we have found intriguing correlations that can guide future biocatalyst design for in vivo biosynthesis.
Results and Discussion
Directed evolution of 2PS screened by LESA-MS
To improve the reactivity of 2PS for better TAL production, we adopted a semi-rational protein engineering approach combined with the high-throughput LESA-MS quantification method to sample a large set of mutant libraries efficiently (Figure 2A). The first step is to adopt the LESA-MS method for reliable detections of TAL (Figure 1C).34 To accomplish this goal, we have generated a calibration curve for TAL from 0.05 mM to 100 mM without sample pretreatment (Supplementary Figure S3). This calibration range is much broader than that from the reported biosensor (around 0.5–10 mM) and readily to be used for the TAL quantification below (See Supplementary Figure S4 for LESA-MS raw spectrum examples).12
LESA-MS assisted directed evolution result summary. (A) LESA-MS screening result summary for 96 deep well culture and the active site residues in 2PS. The box range covers 25 %–75 % of the data and the whisker covers 10 %–90 %. The line in the box represents the mean value. (B) Titer comparison in controlled OD600 growth for key mutants. (C) Enzymatic kinetic study for key mutants over 30 minutes. (D) Michaelis-Menten plot for key mutants. Detailed workflow, more titer comparison data, kinetics as well as raw data for Michaelis-Menten plot can be found in Figures S5–S11. Three parallel experiments were conducted for controlled OD600 growth in Figure 2B and enzymatic assays in Figure 2C–D using biological replicates. Error bar represents standard deviation.
After establishing a reliable LESA-MS method for fast TAL quantification, we first carefully analyzed the crystal structure of 2PS to find target residues for directed evolution (Figure 2A). Multiple sequence alignment indicated that, while many of the residues in the active site of 2PS, such as Thr137, Glu197, Thr199, Phe220, Leu268, Phe270 and Leu272, are highly conserved, several other residues, such as Ile201, Leu202, Met259, Leu261 and Ile343 (Supplementary Figure S5) vary for different purposes.8, 24 These residues would be prime targets to investigate the influences of all 20 amino acids at each site by iterative site-saturation mutagenesis (ISM).35 A general directed evolution workflow from mutant library construction to screening using 96 deep-well plates, and to LESA-MS quantification is described in Supplementary Figure S6 and Table S1. We first studied the ISM of Leu202 because it is known to control the substrate specificity in polyketide synthases.8, 26 L202G-2PS was screened out to display the highest reactivity over WT-2PS with around 9-fold improvement in TAL production (Figure 2B). This result is consistent with the reported best mutant in a previous study.12 From the screening results (Supplementary Figure S7A), the mutants with smaller residues at the 202 position such as alanine and serine were among the ones with the highest improvements. This observation suggests that the size of the active site greatly influenced the 2PS function, with a larger pocket being more beneficial. We then measured enzymatic kinetics and yields of L202G-2PS in vitro to investigate reasons for the improvement. Interestingly, although L202G-2PS displayed a lower yield (70 %) compared to WT-2PS (100 %) in test tubes under standard conditions in 30 minutes, the initial reaction rate was drastically improved by 2.4-fold (e. g., 39 % vs 16 % at 5 min., see Figure 2C). This faster kinetics accounted for the higher TAL titer in cells because both substrates for 2PS were continuously supplied intracellularly by E. coli, in contrast to in vitro enzymatic assays that use limited starting materials.
After identifying L202G-2PS as the best mutant at the 202 position for TAL production, we carried out additional ISM at Leu261 site, which is reported to control the chain elongation steps.8 Previously, L202G/L261N-2PS was reported to be the most effective mutant based on a screening at this position using a TAL-biosensor.12 To our surprise, L202G/L261C-2PS was screened to be the best mutant in our study, displaying 1.5-folder higher titer than L202G/L261N (Supplementary Figure S7B) under the same controlled cell growth experiment and a 28-fold improvement over WT-2PS (Figure 2B). We further monitored the reaction kinetics for these two mutants (Figure 2C). The reaction rates of both mutants were significantly improved over WT-2PS and L202G-2PS, with TAL production reaching >60 % yield after 2.5 min. However, the L202G/L261C-2PS achieved a slightly higher yield than L202G/L261N-2PS later in the catalysis process (e. g., 96 % vs 90 % at 7.5 min).
Using L202G/L261C-2PS as the template for the next generation of ISM, we explored Ile343, Ile201 and Met259 sites, which was previously shown to have both steric and electronic effects in the chain elongation steps of polyketide synthesis.12, 24 This led to a I201V/L202G/M259L/L261C-2PS showing 25 % titer improvement compared to L202G/L261C-2PS (Supplementary Figure S7C–S7E). It is noteworthy that Val201 is commonly found in other PKSs (Supplementary Figure S5) and M259L mutation was proved to be beneficial in a previous study,12 which accounted for the ISM results.
To further improve the catalytic performance of 2PS, we used I201V/L202G/M259L/L261C-2PS as the template for more rational mutagenesis (Supplementary Figure S8).36 Since the surface cysteines reduced the stability of 2PS enzyme, with previous reports demonstrating that mutations of the surface cysteines to serine such as C35S/C372S benefited the enzymatic stability,23 we introduced these two mutations and found a 15 % improvement in titer over the I201V/L202G/M259L/L261C-2PS template. In addition to the rational mutagenesis, we carried out random mutagenesis using error-prone PCR (epPCR), cultured ~600 colonies and then screened them by LESA-MS. Several beneficial mutations with 10 %–26 % higher titer than I201V/L202G/M259L/L261C-2PS were discovered (Supplementary Figure S8). Based on these beneficial mutations from rational mutagenesis and epPCR, we used staggered-extension PCR (StEP) to shuffle all the beneficial mutations at different sites, cultured and screened another ~400 colonies.37 Among them, the best mutant L202G/G205S/M259L/L261C-2PS showed 28 % increased titer over I201V/L202G/M259L/L261C-2PS, 60 % higher over L202G/L261C-2PS, 2.5-fold over the reported L202G/L261N-2PS mutant and 46-fold over WT-2PS (Figure 2B and Supplementary Figure S9).
In vitro/in vivo catalytic performance correlation
With the best L202G/G205S/M259L/L261C-2PS mutant that produced the highest titer in hand, we measured its reaction kinetics and compared it with other mutants. To our surprise, its initial reaction rate was not further improved over that of L202G/L261C-2PS (Figure 2C and Supplementary S10). We then wondered if it is due to a stronger CoA substrate binding affinity, which can contribute to the overall catalytic efficiency. To test this hypothesis, we measured the Michaelis-Menten parameters of L202G/G205S/M259L/L261C-2PS in the presence of different concentrations of malonyl-CoA and compare it with those of WT-2PS, L202G/L261C-2PS and the best reported mutant L202G/L261N-2PS before this work (Figure 2D and S11, Table 1).12 While the kcat of L202G/G205S/M259L/L261C-2PS (0.093±0.003) is slightly less than that of L202G/L261C-2PS (0.12±0.01 s−1), the substrate-binding affinity (KM=41±4 μM) is almost 2-fold stronger than that of L202G/L261C-2PS (KM=78±4 μM). As a result, the catalytic efficiency of L202G/G205S/M259L/L261C-2PS ((2.3±0.2)×103 s−1 M−1) is 1.5-fold more than that of L202G/L261C-2PS ((1.5±0.1)×103 s−1 M−1).
Mutant |
KM (μM) |
kcat (s−1) |
kcat/KM (s−1 M−1) |
|---|---|---|---|
WT-2PS |
30±3 |
(2.7±0.1)×10−3 |
91±9 |
L202G/L261N-2PS |
34±6 |
(3.5±0.2)×10−2 |
(1.0±0.2)×103 |
L202G/L261C-2PS |
78±4 |
(1.2±0.1)×10−1 |
(1.5±0.1)×103 |
L202G/G205S/M259L/L261C-2PS |
41±4 |
(9.3±0.3)×10−2 |
(2.3±0.2)×103 |
The above analysis indicates that the substrate-binding affinity plays a key role for enzymes, especially towards endogenous molecules in cells such as CoAs, because their concentrations are highly regulated and difficult to change in vivo. Intracellular concentration of malonyl-CoA in exponentially growing E. coli in the presence of glucose was measured to be around 35 μM.38 Since KM of L202G/G205S/M259L/L261C-2PS (41±4 μM) is much closer to the cellular malonyl-CoA concentration than KM of L202G/L261C-2PS (78±4 μM), L202G/G205S/M259L/L261C-2PS may be able to take the full advantage of the intracellular malonyl-CoA concentration, leading to the largest titer improvement (46-fold over WT-2PS) than any 2PS mutants investigated here and reported in the literature.
To further probe the relationship between intracellular substrate concentration, Michaelis-Menten parameters and TAL titer, we co-transformed a plasmid encoding malonyl-CoA synthetase (MatB) and malonate carrier protein (MatC) with the 2PS gene into E. coli BL21(DE3) to increase the intracellular malonyl-CoA concentration.39-41 We measured the titer of cell cultures in 20 mL shaking flasks based on a reported protocol (Table 2).12 In the cell culture containing only 2PS gene without MatB or MatC, L202G/G205S/M259L/L261C-2PS produced 690 mg/L titer, the highest in comparison with those of L202G/L261C-2PS and L202G/L261N-2PS. Interestingly, when MatB and MatC were co-transformed with 2PS to supply more intracellular malonyl-CoA, the TAL titers of all three mutants were enhanced, but L202G/L261C-2PS instead of L202G/G205S/M259L/L261C-2PS achieved the highest titer with 880 mg/L. It is reported that the intracellular malonyl-CoA can be enhanced around 3-fold by adding a malonyl-CoA biosynthesis pathway from malonate in E. coli.20 From the Michaelis-Menten plot, L202G/L261C-2PS has 1.3-fold higher kcat than that of L202G/G205S/M259L/L261C-2PS, but the catalytic efficiency was hampered by a weakened KM. However, a higher intracellular concentration of malonyl-CoA from MatB/MatC pathway provided high enough substrate concentration for L202G/L261C-2PS and enabled a higher reaction rate at the increased level of malonyl-CoA concentration, which accounted for the overall higher titer than L202G/G205S/M259L/L261C-2PS (Figure 2D).
2PS mutant |
Titer (×102 mg/L)[a] |
Titer (×102 mg/L)[b] |
|---|---|---|
L202G/L261N-2PS |
5.3±0.2 |
6.9±0.1 |
L202G/L261C-2PS |
5.8±0.1 |
8.8±0.2 |
L202G/G205S/M259L/L261C-2PS |
6.9±0.3 |
7.9±0.3 |
- [a] E. coli transformed with pET16b-2PS mutant plasmid. [b] E. coli transformed with pET16b-2PS mutant and pCK-matBC plasmid.
The above examination of the correlations among intracellular substrate concentration, enzymatic catalytic parameters and TAL titer has revealed the difference between in vitro biocatalysis and in vivo biocatalysis under the specific metabolism of cells. Firstly, comparing L202G-2PS with WT-2PS, L202G-2PS showed 8-fold higher titer in in vivo assays (Figure 2B), faster reaction rate but lower final yield in in vitro assays (Figure 2C) than those of WT-2PS. Such important mutant could be readily missed if 2PS was engineered or screened based on the yield of in vitro enzymatic assays with limited amount of substrates, which was often the criteria for in vitro directed evolution studies.42 This observation also showed the importance of reaction kinetics especially for intracellular reactions to convert substrates into the products in high speed, because the reactants can be continuously generated. Secondly, comparing L202G/G205S/M259L/L261C-2PS with L202G/L261C-2PS mutant, the latter mutant showed higher initial reaction rate, faster kcat in in vitro assays (Figure 2D) but lower titer in standard E. coli BL21(DE3) in in vivo assays (Figure 2B). The best mutant L202G/G205S/M259L/L261C-2PS cannot be screened out solely by in vitro assays without the intracellular selection pressure from limited substrate concentration, even if the 2PS was engineered for both faster reaction kinetics and higher yield. Furthermore, we applied a malonyl-CoA biosynthetic pathway which increased the intracellular substrate concentration and made L202G/L261C-2PS the highest titer. This result illustrated that the 2PS mutant should be screened and chosen correspondingly to match the working substrate concentration, since different microorganisms have their unique metabolism and malonyl-CoA concentration.9-11 These comparisons overall led to an important lesson we learnt here that a fast reaction rate under a specific substrate concentration intracellularly is the key for the high performance of in vivo biocatalysis, which should be considered seriously if the target biocatalyst is engineered in vitro or in a different microorganism.
Conclusions
In summary, by employing a high-throughput LESA-MS quantitation method, we are able to screen a large mutant library of 2PS and discovered two new mutants (L202G/L261C-2PS and L202G/G205S/M259L/L261C-2PS) that exhibit over 44-fold faster kcat, 25-fold higher catalytic efficiency and 46-fold higher titer over WT-2PS for TAL production. The correlation between intracellular substrate concentration, Michaelis-Menten parameters and TAL titer for these two mutants is carefully studied by enzymatic kinetic assays and synthetic biology tools to explicitly show how 2PS was evolved in the directed evolution process. The study reveals that a fast reaction rate under a specific substrate concentration intracellularly is important for in-cell biocatalysis. Such properties can be tuned by protein engineering and synthetic biology to be adopted for different intracellular environments of life. Overall, the demonstrated workflow that combines protein engineering with high-throughput screening and the insights gained from the correlation between in-vitro and in-cell catalytic properties would be highly applicable to many other key enzymes for secondary metabolites bioproduction besides polyketides.
Acknowledgments
We wish to thank Prof. Markus Jeschek for providing pCK-matBC, Prof. Emily Que and Sky Price for HPLC analysis, Dr. Yunling Deng, Mandira Banik, Danielle Lawson and Whitney Lewis for their help with revising the manuscript. The work described in this report is supported by the DOE Center for Advanced Bioenergy and Bioproducts Innovation (U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research under Award Number DE-SC0018420). We also thank the Robert A. Welch Foundation (Grant F-0020) for support of the Lu group research program at the University of Texas at Austin.
Conflict of interests
The authors declare no conflict of interest.
Open Research
Data Availability Statement
The data that support the findings of this study are available in the supplementary material of this article.
