Team:UESTC-China/Model
Overview
Multienzyme processes’ efficiency can be enhanced by optimization of the stoichiometry of the biocatalysts. Our project is a three-enzyme system catalyzing a five-step chemical conversion, we hope that these three enzymes (DhaA31, HheC, EchA) will be used to degrade TCP into glycerol [1]. We built a mathematical model, and the enzyme stoichiometry of each pathway was optimized. Mathematical modeling and one-pot multienzyme experiments provided detailed insights into pathway dynamics, enabled the selection of a suitable engineered enzyme, and optimal ratios of enzymes. We also got lots of experimental data， observing excellent agreement between predicted and experimental data. The validated model was used to quantitatively describe the kinetic limitations of currently available enzyme variants and predict improvements required for further pathway optimization. We used the model to guide the design, and found an optimal strategy to improve project conversion efficiency.
The picture shows the degradation pathway of TCP
Model Foundation
∞ Basic Model
For kinetics of the multi-enzyme system, not only each enzymatic reaction is reversible, the overall reaction could reach the equilibria, but also the inhibition of the various products on the enzyme, and the affinity of the enzyme for different substrates, it is difficult to study at the same time.
So we first look at the whole reaction as independent kinetics, we suppose that the reaction of the enzyme is consistent with the Michaelis-Menten equation, so chain reaction in the first step of the DhaA31 catalytic TCP into DCP reaction(Eq.(1)), obeying the Michaelis-Menten model (where complex for the enzyme and substrate intermediates). Its reaction rate equations for the changes of substrate is described in Eq. (2).
We also obtain the equation for the reaction rate of all the intermediates in terms of the Michaelis-Menten model. which both similar to the Eq.(2). Because our reactions are single-substrate reaction to get a single product, and their material is approximately equal to 1: 1, so we can get a relationship based on this in Eqs. (3) - (7).
We integrate the reponse rate to each step Can get the TCP and the various intermediates and glycerol concentration with respect to the time expression. Then we can throw these Eqs. (2)-(7) can get the relationship between the amount of glycerol (GLY) produced and the other parameters, where t is the reaction time (h).
∞ Model development
Although we obtained the GDL expression by the relationship between the Michaelis-Menten equation and the chain reaction, the model only considers the case of single enzyme reaction, and the affinity of the enzyme to different substrates is too complicated to analyze. But the reaction process has two enzymes were used twice, we hope that through this aspect to improve the model. During this process, we assumed. We assume that there is no inhibitory effect of the substrate during the reaction
( 8 )
As the Eq. (8) shows, we can use HheC and EchA as the same reaction process to establish the model, the two substrates together to share the same enzyme can be approximated as competitive inhibition of the kinetic equation. (E is the enzyme, S, I is the two substrates, ES, EI are the intermediate complex, P, Q are the product)
In Eqs. (9), E is Total enzyme concentration, after the reaction began, divided into [ES], [EI], and the free enzyme [E_f].
Eq. (11) is constructed based on the 'steady state' theory and the law of quality, combine with Eqs. (10), (11) we can get the Eqs. (12) – (15), then the changes in the concentrations of substrates (i.e DCP), product TCP, and intermediates (i.e ECH, CPD, GDL), can be calculated by using Eqs. (3) - (7) and Eqs. (12) - (15).
HheC have the high enantioselectivity (E≥100), which made DCP accumulate. (figure 1) Nonselective DhaA converts the prochiral TCP into both enantiomers of DCP in an almost equimolar ratio. Because HheC prefers (R)-DCP, (S)-DCP tends to accumulate. The conversion of the prochiral TCP into either (R)-DCP or (S)-DCP by DhaA31 was described using equation (16).
1,2,3-trichloropropane (TCP) is converted via (R)- (S)-2,3-dichloropropane-1-ol (DCP),
The rate constants kcat,TCP,(R)-DCP and kcat,TCP,(S)-DCP were determined from the overall kcat for the appropriate DhaA variant with TCP and the ratio of (R)- and (S)-DCP produced from the prochiral TCP by DhaA31 and DhaA. The formation of (R)-DCP and (S)-DCP and the subsequent conversion of both enantiomers into (R, S)-ECH was described using equations (17) and (18). The remaining reaction steps were assumed to be non-selective. The consecutive conversions of (R,S)-ECH by HheC and EchA was described using equations (19).
Because the product has an inhibitory effect on the reaction, The GDL concentration progress curve (Eq. 21) was fitted to equation (20), where P is the concentration of the reaction product and Ki is a product inhibition constant[2].
Model fitting with experimental data
One-pot biotransformation experiment was conducted in the 10mM Tris-SO4 (pH=8.5) containing 2 mM TCP, 2.4 mg DhaA31, 2.4 mg HheC, 2.4mg EchA at 37℃, The time dependent profiles of substrates (i.e DCP), production TCP. Three parameter settings for the model are listed in Table 1. the first setting was based on kinetic parameters of DhaA31, HheC, EchA from literature in this study. All the studied reactions exhibited Michaelis–Menten kinetics. The conversion of glycidol (GDL) into GLY was described by a Michaelis–Menten equation, with one inhibition constants defining the inhibitory effects of GLY (product inhibition constant K_i=1.00 mM).
The model simulation curves based on first setting data did not match well with the experimental data. especially for the time required for reaching the equilibrium. It was not surprising that experimental data and simulation curves based on the first set of parameters did not fit well because kinetic parameters of an individual enzyme may not be the same as its apparent catalytic activity in the mixture of multiple enzymes. The first set of parameters need be adjusted for the best fitting. Among a number of parameters tested, it was found out that adjusting three kinetic parameters. Simulation curves based on the second set of data exhibited the best fitting (Fig. 2).
( A )
( B )
Figure.2 (A) Simulation modeling and experimental results of the synthesis of GLY from TCP. Experimental conditions: 10mM Tris-SO4 buffer (pH 8.5) containing 2 mM TCP, 2.4mg DhaA31, 2.4mg HheC, 2.4mg EchA at 37℃(B) Under the best fitting, the curves of TCP and various intermediates, glycerol concentration over time.
Modeling prediction and experimental validation
∞ Optimal enzyme loading ratio
The important function of modeling is the prediction of the optimal reaction conditions (e.g. enzyme ratio and catalytic constant of the enzyme). Predictive simulation of the kinetic model was conducted based on the second set of parameters under conditions: in the 10mM Tris-SO4 (pH=8.5) containing 2 mM TCP, DhaA31, HheC, EchA at 37℃.
The enzymes have similar molecular weights (34.1, 29.3, and 36.5 kDa, respectively), so mass ratio roughly equals molar ratio. Degradation has relationship with these two factors (i) expression profile, (ii) viability. So our kinetic model was employed for rational selection of suitable plasmid combinations and balancing the stoichiometry of the enzyme DhaA:HheC:EchA. The pathway was optimized toward faster removal of toxic metabolites and higher production of GLY.
Significant efforts have been invested in the past few years in engineering the first enzyme of the TCP pathway: haloalkane dehalogenase (DhaA)[2]. Constructed mutants(DhaA31) could possibly help overcome the previously mentioned limitations of the TCP pathway in the last studies[2]. So we chose the DhaA31 to construct plasmid. We then evaluated the effects of enzyme stoichiometry on efficiency. An algorithm was used to minimize the total enzyme load without sacrificing productivity (we define the sum of the three enzymes in contrast to 1.0)[3]. We set the step to 0.01, through the algorithm traversal of all the ratio of the situation. The modeled time courses for individual reactions were similar in each case, but the optimized ratios and total enzyme loadings differed significantly between pathways (Figure 3). We find that the optimal enzyme loading ratio is DhaA31: HheC: EchA=0.45:0.45:0.1.
Experiment at this enzyme ratio was conducted to validate the model prediction. As shown in Fig.4, the simulation curves of TCP, intermediates and the glycerol fit well with experimental data, indicating the validity of the model.
From the figure 3 and optimal enzyme loading ratio we found that loading quality of DhaA31 and HheC had greater effect on glycerol.
Figure 3. The rate of glycerol production was observed over a fixed period of time. Isoproductivity color charts showing how biocatalyst concentration ratio affects productivity in the three-enzyme conversion of 1,2,3-trichloropropane into glycerol.
Figure 4. The curves of TCP and GLY in different enzyme loading ratio
∞ Improvement of project
According to the optimal enzyme loading ratio, we found that the ratio of DhaA31:HheC≈1:1. So we design the same promoter to drive the enzymes. 2A Peptides were used to control the expression of the polygene on the same vector is consistent with its protein activity and expression efficiency.
There was epoxide hydrolase in the plant. Besides, we considered the introduction of more foreign genes had exogenous inhibition of gene expression. Based on the optimum enzyme ratio previously obtained, loading quality of EchA had little effect on glycerol yield. We constructed a new plasmid which only had the first two enzymes to improve the productivity of glycerol. We hope that by this method, as far as possible to reduce the damage to plants without affecting the yield of glycerol.
We want to improve the conversion efficiency by dealing with the first two enzymes. Because the rate of degradation is related to expression profile and viability. However, overexpression of exogenous genes is harmful to the plant. So we paid attention to the viability of enzyme. According to our team in iGEM2013, the activity of HheC-W249P is wild type ten times. So we constructed our plasmid by this mutant.
∞ Future work
In our work, we found that the viability of HheC-W249P not high enough, so we need to obtain a reasonable and ideal structure by molecular docking. Through the screening of PDB database, we found X-ray crystal structure of HheC with serial number of 1pwz. (Fig. 5)
Figure 5. HheC structure diagram (PDB serial number 1pwz)
First of all we have to do some simple processing, the use of Warren Lyford DeLano developed by the open source visualization of the PyMOL software 1pwz selected as the monomer WT, and the system of water molecules removed. We used alanine scanning technology, the active pocket around the 4A within the amino acid ALA replacement. The predicted structure of the mutant was obtained by SWISSMODEL homologous modeling. The WT and mutant models were subjected to Autodock removal of water, plus polar hydrogen, plus Kollman charge treatment. When setting the docking simulation parameters, it is reasonable to consider whether the substrate and the enzyme S132 and Y145 can form hydrogen bonds. Therefore, W139A, F186A, Y187A may be a potential mutation hot spots. (table 2)
| NAME | s132 | y145 |
|---|---|---|
| WT | 9.2 | 7.6 |
| F82A | 8.9 | 7.5 |
| E85A | 8.9 | 7.4 |
| F86A | 8.9 | 7.5 |
| W139A | 3.0 | 3.0 |
| P184A | 9.1 | 7.5 |
| Y185A | 8.8 | 7.4 |
| F186A | 3.0 | 2.9 |
| Y187A | 2.9 | 2.9 |
| P188A | 9.9 | 7.4 |
| W192A | 8.9 | 7.4 |
Table 2. The docking results of the distance analysis results
Figure 6. (A) We got Mutant W139A obtained from SWISSMODEL (B) Docking substrate 2,3 DCP
In the figure (7), red is the active center of the enzyme, and the yellow circle shows the tryptophan at position 139 in WT. As the volume of the 139W side chain is too large, the substrate is prevented from approaching the active site. On the right, the substrate can be docked smoothly when W is mutated to a smaller side chain.
Figure (7). (A) WT docking result (B) W139A docking result
Next, we will target these mutations and saturation mutations, screening for 2,3 - DCP has a highly effective activity of mutants. We constructed plasmids through different mutants, and our kinetic model was employed to improve the TCP degradation pathway using engineered enzyme variants and balanced enzyme ratios. And model will show excellent correspondence with experimental data. An approximate estimation of expression profiles obtained from sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of cell free extract (CFE) will show the relative ratio of enzymes which will be constructed by different mutants and promoter. Combining viability of mutants with optimal ratios of enzymes for the maximal production of glycerol, we predict improvements required for further pathway optimization.
- Dvorak, P., et al., Immobilized synthetic pathway for biodegradation of toxic recalcitrant pollutant 1,2,3-trichloropropane. Environ Sci Technol, 2014. 48(12): p. 6859-66.
- Dvorak, P., et al., Maximizing the efficiency of multienzyme process by stoichiometry optimization. Chembiochem, 2014. 15(13): p. 1891-5.
- Kurumbang, N.P., et al., Computer-assisted engineering of the synthetic pathway for biodegradation of a toxic persistent pollutant. ACS Synth Biol, 2014. 3(3): p. 172-81.
