Min Wu,Juanjuan Ding,Zhaofeng Zhang,Shengping You,4,*,Wei Qi,2,3,4,*,Rongxin Su,2,3,4,Zhimin He,2,*

1 Chemical Engineering Research Center,School of Chemical Engineering and Technology,Tianjin University,Tianjin 300350,China

2 State Key Laboratory of Chemical Engineering,Tianjin University,Tianjin 300350,China

3 Collaborative Innovation Center of Chemical Science and Engineering (Tianjin),Tianjin 300072,China

4 Tianjin Key Laboratory of Membrane Science and Desalination Technology,Tianjin University,Tianjin 300072,China

Keywords:Kinetic modeling Fermentation Gamma-aminobutyric Lactobacillus bre123vis pH control Rolling correction

ABSTRACT Gamma-aminobutyric acid(GABA)is a natural non-protein functional amino acid,which has potential for fermentation industrial production by Lactobacillus bre123vis.This work investigated the batch fermentation process and developed a kinetic model based on substrate restrictive model established by experimental data from L25(56)orthogonal experiments.In this study,the OD600 value of fermentation broth was fixed to constant after reaching its maximum because the microorganism death showed no effect on the enzyme activity of glutamate decarboxylase (GAD).As pH is one of the key parameters in fermentation process,a pH-dependent kinetic model based on radial basis function was developed to enhance the practicality of the model.Furthermore,as to decrease the deviations between the simulated curves and the experimental data,the rolling correction strategy with OD600 values that was measured in real-time was introduced into this work to modify the model.Finally,the accuracy of the rolling corrected and pH-dependent model was validated by good fitness between the simulated curves and data of the initial batch fermentation (pH 5.2).As a result,this pH-dependent kinetic model revealed that the optimal pH for biomass growth is 5.6-5.7 and for GABA production is about 5,respectively.Therefore,the developed model is practical and convenient for the instruction of GABA fermentation production,and it has instructive significance for the industrial scale.

1.Introduction

Gamma-aminobutyric acid (GABA) is a natural non-protein amino acid that is widespreadly existed in microorganisms,plants and animals [1,2].In the mammalian central nervous system,GABA plays an important role as a major inhibitory neurotransmitter[3].GABA is produced from L-glutamic acid by glutamate decarboxylase (GAD) and is reported as a bioactive component with various physiological functions[4].GABA has extensive application prospect since its functions in improving brain function [5],antianxiety and sedative effects [6],lowering blood ammonia concentration and blood pressure[7,8],diuretic effect[9]and fighting diabetes mellitus effect [10,11].

The production methods of GABA include chemical synthesis and biological synthesis[12].However,due to some disadvantages of chemical synthesis method such as expensive reactants,harsh reaction conditions,and toxic by-products[13],the GABA production cannot be applied to food industry [3].Microbial production method is considered to be the inevitable trend of GABA production due to easily available substrates,mild fermentation conditions,less environmental pollutants,and product safety [14].Lactic acid bacteria(LAB) are considered as significant GABA production microorganisms in industrial production [15] because the probiotics are generally regarded as safe (GRAS)food microorganisms andLABhave high glutamate decarboxylase (GAD,EC 4.1.1.15) activity [16].

However,there were only a few studies concentrating on kinetics of GABA production.Huanget al.[17] constructed a serials of simple kinetic models including Logistic model for microorganism growth,Luedeking-Piret model for product generation and substrates consumption model by batch fermentation ofLactobacillusbre123vis(L.bre123vis)CGMCC1306;Zhanget al.[18]used resting cells ofL.bre123visTCCC 13,007 isolated from Chinese kimchi to produce GABA and analyzed its enzymatic reaction by Michaelis-Menten equation;Liet al.[19] created a continuous cultivation method withL.bre123visNCL 912 and constructed a Monod model and a Luedeking-Piret model for microorganism growth and product generation,respectively;Gangarajet al.[20] evaluated a variety ofLAB,and performed kinetic and thermodynamic analysis onEnterococcus faeciumCFR 3003.Shiet al.[16]also used resting cells ofL.bre123visTCCC 13007 and analyzed the fermentation process by Monod model.The above-mentioned studies of GABA production process byLABfocused on optimization of media composition and fermentation strategy to get high production yields in static process.However,the fluctuation effect of input parameters in the industrial scale is inevitable and significant,which should pay attention to study in.The kinetic analysis of whole fermentation system can help to reveal significant parameters and figure out the influence of various factors in dynamic process,and it is benefit to make more detailed fed-batch strategies of fermentation process.

The establishment of fermentation kinetic models can be divided into three types: mechanism model,data-driven model and hybrid model of data-driven and mechanism.The mechanism model,aka white box model,is established with reactions and metabolic pathways involved in fermentation process,such as metabolic flux analysis (MFA) [21] and flow balance analysis(FBA) [22].The mechanism model has good interpretability but many complex equations of states and parameters are required to achieve its accuracy,making the model too complicated to be applied.The data-driven model,aka black box model,relies on large amounts of fermentation data to recognize patterns,as Warneset al.[23] applied feedforward artificial neural networks and radial basis function toE.colifermentation process.The datadriven model avoids complex mechanism but it neither has definite physical meaning nor good universality.The hybrid model of data-driven and mechanism,aka gray box model,avoids the model being too complicated and has certain physical meaning by fitting part of parameters with experimental data.Liuet al.[24] developed a temperature-dependent model of Chinese wine fermentation with hybrid model,and provided a variable temperature fermentation strategy.On the other hand,the pH is a very important parameter in dynamic fermentation process,as the pH would influence the whole metabolic process ofL.bre123visfor the GABA production.Thus,the hybrid model was chosen to establish the model for instructing the practical fermentation process,which can measure pH parameters with data-driven methods and measure kinetic parameters with mechanism model.

This research aims to analyzing and modeling the batch process ofL.bre123visfor GABA production,pointing on pH effect on static kinetic parameters.The L25(56) orthogonal experiments were carried out in media with different pH,substrate concentration and glucose concentration.A rolling correction model was obtained that was corrected in real-time based on the OD600value of fermentation broth.Thus,the production of GABA products from the consumption of glucose as carbon source and L-monosodium glutamate (MSG) as substrate were well predicted.This research would be benefit to the effective process design of enhancing GABA production.

2.Materials and Methods

2.1.Microorganism and media

L.bre123visstrain was obtained by screening from fermented food in our laboratory.The strain was preserved in 80% (volume) glycerol at -80 °C as stock.The MRS liquid medium [25] consisted of(per liter): 10.0 g peptone,5.0 g beef extract,4.0 g yeast extract,1.0 g tween-80,20 g glucose,2.0 g K2HPO4,5.0 g sodium acetate,2.0 g tri-ammonium citrate,0.2 g MgSO4·7H2O and 0.05 g MnSO4-·4H2O.The initial pH was adjusted to 6.2.The GYP seed medium[26] consisted of (per liter): 10.0 g glucose,10.0 g yeast extract,5.0 g peptone,2.0 g sodium acetate,0.02 g MgSO4·7H2O,trace MnSO4·4H2O,trace NaCl,trace FeSO4·7H2O and 20.0 g MSG.The initial pH was adjusted to 6.8.The normal GYP fermentation medium [26] consisted of (per liter): 20.0 g glucose,15.0 g yeast extract,5.0 g peptone,3.0 g sodium acetate,0.03 g MgSO4·7H2O,0.02 g MnSO4·4H2O,trace NaCl,trace FeSO4·7H2O and 75.0 g MSG.The initial pH was adjusted to 5.2.

2.2.Cultivations conditions

The strain was preserved in 80%(volume)glycerol at -80°C as stock.The primary activation preculture was prepared by inoculating 10 μl of glycerol stock into 5 ml MRS liquid medium.The test tube was incubated in an air rotary shaker (HZQ-F160,Peiying,Jiangsu,China) at 35 °C,200 r·min-1for 20 h.The secondary activation preculture was prepared by inoculating 0.5% (volume) of primary activation preculture into GYP seed medium and the shake flask was statically incubated at 35°C for 24 h.Batch fermentation experiments were carried out by modified GYP fermentation medium from the normal one.The batch fermentation experiments aimed at investigating the effect of OD600value decrease on enzyme activity of MSG conversion after 26 h.When the fermentation came to 26 h,the fermentation medium was divided into three parts for control group and two experimental groups.75 g·L-1MSG solid was added into the group F1 at 26 h and the group F2 at 32 h,respectively.The pH and concentrations of MSG and GABA were designed based on orthogonal experimental design,the L25(56)orthogonal table provided in Table S1.All the experiments were performed in 1 L four-necked flask containing 500 ml of the medium.The pH was measured by pH electrode (Hamilton,Bonaduz,Switzerland) and automatically regulated by fermentation system(Biolun,Shanghai,China),using 3 mol·L-1NaOH and 3 mol·L-1H2SO4.Temperature maintenance and stirring were provided by magnetic stirring water bath.All the experiments were carried out at 35 °C and inoculating 10% (volume) of secondary activation preculture.

2.3.Analytical methods

The samples were drawn from flask for estimation of biomass,glucose,MSG and GABA concentrations.Biomass was measured at 600 nm using a UV-Vis spectrophotometer (TU-1810PC;PERSEE,Beijing,China).The glucose concentration was measured using DNS method (the DNS reagent was from Solarbio,Beijing,China).The method to measure concentrations of MSG and GABA[27]were using HPLC (Agilent 1200 series;Hewlett-Packard,Palo Alto,CA,USA) methodology using UV detector.The samples were centrifuged at 10,000 r·min-1for 5 min and the supernatant obtained from centrifugation was diluted × 500 times.The diluent (200 μl)was derivatized with 0.8% (mass/volume) dimethylaminonaphthalene sulfonyl chloride[DNS-Cl(200 μl)]and NaHCO3-NaOH buffer (0.5 mol·L-1,pH 9.8,400 μl).Derivatization was carried out at 40°C for 1 h,and the reaction mixture was filtered by microfilters(pore size 0.22 μm).The aliquots (10 μl) were analyzed at 40 °C using a Hypersil ODS2 analytical column (250 × 4.6 mm;5 μm;Thermo Fisher Science,Waltham,MA,USA).The gradient elution program and the composition of mobile phases are provided in Table S2.The flow rate of mobile phases was 1 ml·min-1and the wavelength for detection was 254 nm.All the experimental data are the arithmetic mean values of triple repeated experiments.

2.4.Kinetic model development

Firstly,the data of microorganism death should be corrected since the glutamate decarboxylase is an enzyme that does not require ATP for energy to function.So,assumption that the enzyme activity in fermentation broth would not decrease with the death ofL.bre123visshould be considered,and the enzyme activity should be correlated with microbial concentration.Hence,the microbial concentrationXwould be assumed to only increase,not decrease.

The raw data was noisy and the data points were discrete,smoothing splines method was used to process the data.This step was to remove the noise in the raw data and to make the discrete data consecutive,which was convenient for solving the differentiation.Smoothing splines method is described as Eq.(1):

the specified weightswiwere assumed to be 1 for all data points,and the specified smoothing parameterpwas defined to be 0.025.

The rate of increase in biomass would be explained by Eq.(2):

where μXis the biomass specific growth rate (h-1),Xis the absorbance in OD600(dimensionless).

Generally,Monod equation is a simple unstructured nonsegregated kinetic model to describe biomass growth.As the most famous gray box kinetic model [28],Monod equation can accurately describe the growth of microorganisms under the action of restrictive substrate.The equation is given in Eq.(3):

where μX0is the maximum biomass specific growth rate (h-1),Sis the concentration of glucose(g·L-1)andKSis the Monod saturation constant (g·L-1).

In general,the substrate glucose consumed is used for biomass growth and product formation and cell maintenance activities in microbial fermentation.The substrate consumption in a batch fermentation process equation is given in Eq.(4):

whereYP/Sis product yield coefficient on glucose consumed(g·g-1),YX/Sis biomass yield coefficient on glucose consumed (g-1·L),andmsis the maintenance coefficient (g·h-1).

In this research,the GABA biosynthesis is carried out by the glutamate decarboxylase system containing GAD enzyme and glutamate/GABA antiporter GadC [29-33].There is little of GABA product synthesized from glucose,which synthesize αketoglutarate as the precursor of L-glutamate by the glycolysis pathway and part of the TCA cycle.And it was also assumed that the consumption of glucose needed to maintain biomass is very little.Eq.(4) could be simplified to Eq.(5):

Since the GABA biosynthesis is mainly derived from the conversion of MSG through the glutamate decarboxylase system,it was assumed that the kinetic of GABA production would fit to the restrictive substrate model like Monod equation,which means the GABA production restricted by the concentration of MSG.The equation of the GABA production is given in Eq.(6) and the equation of MSG consumption is given in Eq.(7):

where μPis the product formation rate(g·L-1·h-1),Pis the concentration of GABA product(g·L-1),kPis the maximum specific formation rate of product (g·L-1·h-1),Mis the concentration of MSG(g·L-1),KMis the restrictive coefficient of MSG (g·L-1),YP/Mis product yield coefficient on MSG consumed(g·g-1)andYX/Mis biomass yield coefficient on MSG consumed g·L-1.Since sufficient glucose has added to fermentation medium which acts as a fast-acting carbon source for biomass growth,it was assumed that the consumption of MSG on biomass growth could be disregarded.Eq.(7) could simplify to the Eq.(8):

In the process of enzymatic catalysis,the effect of pH is a very important factor.Since the effect of pH on enzyme activity is similar with Gaussian function,it was assumed that the effect of pH on kinetic parameters corresponds to Gaussian function distribution,which is described as Eq.(9):

wheref(x)is the value of the kinetic parameter,ais the value of the kinetic parameter under optimal pH,bis the value of optimal pH andcis defined to sensitivity of the biotransformation process after pH deviates from the optimal value,which is inversely proportional to the sensitivity.

For improving the accuracy of the kinetic model,the strategy of rolling correction on OD600replaced the predicted data points of OD600with the actual ones.By substituting the actual OD600values into Eqs.(5),(6),and(8),the predicted data of the rolling corrected model can be gotten.

2.5.Kinetic parameters estimation

Data using to estimate the kinetic parameters carried out from the L25(56) orthogonal experiments with varying initial carbon source concentration (0-40 g·L-1) and varying MSG concentration(0-150 g·L-1).Non-linear least square method was used to minimize the sum of square errors (SSE) between model simulation data experimental data.All the kinetic parameters estimation and fitting process were performed in Microsoft Excel:

wherexsimis the model simulatedxvalue andxexpis the experimentalxvalue.

3.Results and Discussion

3.1.Correction of microorganism death

In the fermentation for the GABA production byL.bre123visstrain,the OD600value reached its maximum in 22 h and was decreased after 24 h.As shown in Fig.1(a),glucose and MSG were depleted in 20 h and 24 h,respectively.And the production of GABA had basically reached the highest value.As shown in Fig.1(a),only the OD600value decrease and GABA concentration maintenance could be spotted,but the bioconversion activity was unable to be known as MSG has been depleted.

In order to investigate the effect of OD600value decrease on enzyme activity of MSG conversion,the fed-batch fermentations were carried out in the GYP medium (pH 5.2) and the result was shown in Fig.1.The OD600value of fermentation broth is 6.39 at 26 h and 4.57 at 32 h.After the MSG solid was added to the fermentation broth,the average consumption rate of fermentation processes in the group F1 and F2 is 6.411 g·L-1·h-1and 6.249 g·L-1-·h-1in 12 h before MSG was depleted,respectively.And for the product,the average production rate is 3.252 g·L-1·h-1and 3.395 g·L-1·h-1,respectively.In the two groups,the consumption rate of MSG and the production rate of GABA is nearly identical,respectively,which indicates that the enzyme activity for GABA production in the medium is not affected by microorganism death,namely the decrease in OD600value.Furthermore,the OD600value of group F1 and F2 at 48 h was 2.41 and 2.32,respectively,slightly larger than 2.21 of the control group,which indicated the influence of MSG to biomass maintenance is slight.Thus,a reasonable assumption could be validated that the microbial concentrationXwould maintain its maximum after reaching the peak value,indicating that the part of kinetic model for microbial death could be omitted.Based on this assumption,the model contributes to kinetic model establishment and its modification,which may be convenient and practical for application.

Fig.1.The fed-batch fermentations for investigating the effect of microorganism death on enzyme activity: (a) control group;(b) group F1;(c) group F2.

3.2.Development of the pH-dependent model

The pH of media could significantly influence the biological process of microorganism.It is an important parameter for enzyme activity,so the entire metabolism of microorganism would be affected by pH parameter of fermentation environment.The lactic acid bacteria would produce large amounts of organic acids such as lactic acid and acetic acid in the process of metabolism,which would reduce the pH of the fermentation broth and enhance the survival advantage against other bacteria[34].However,most lactic acid bacteria are neutrophilic bacteria,and the accumulation of organic acids in the fermentation process will also affect their own metabolism.On the other hand,because GABA is an alkaline substance,the continuously converted GABA will increase the pH of the fermentation broth.Thus,the research of pH effect on GABA production is significant to its industrial production.

The parameters values of GABA fermentation process were estimated by the L25(56)orthogonal experiments with variable pH,glucose concentrations and MSG concentrations.Experimental data was obtained to be fitted to Eqs.(3),(5),(6),and (8) to estimate kinetic parameter values.The kinetic parameter values were estimated from experimental data under the varying fermentation status of pH level 1-5 (pH 4.5,5.0,5.5,6.0,6.5).The results were present in Table 1.The values of parameters present in Table 1 were used to estimate the pH effect on the kinetic parameters of the GABA production process.The effect of pH on the biomass growth,glucose consumption,MSG consumption and GABA production are quantitatively described by Eq.(9).In Fig.2,the data points ofYX/S,YP/M,μX0,KS,kPandKMwere from Table 1 and the curves were simulated profiles from Eq.(9).The estimated parameter values of the Eq.(9)are presented in Table 2.As shown in Fig.2,both the experimental data and the simulated curves show a trend of increasing first and then decreasing.The trend of the experimental data and the simulated curves both fit the trend of radial basis function,which denotes that the equation form about the effect of pH on the kinetic parameters of GABA fermentation production is suitable.The simulation results ofR-square in Table 2 denote that the simulated curves fit well with experimental data.In Fig.2(a),(b),(e),and(f),the peaks of simulated profiles were located at about pH 5,and in Fig.2(c) and (d),the peaks were located at pH 5.6 and 5.7,respectively.The peak location values are shown as parameterbin Table 2.Since parameter μX0andKSare determined by biomass growth,so the optimal pH for biomass growth is 5.6-5.7,which means that in this condition the microorganism could grow better in the lag period and the log phase.The same can be obtained that the optimal pH for GABA formation is about 5,which means that in this condition the GABA formation by GAD would be more efficient.Thus,a fermentation strategy could be developed that the initial pH is adjusted to 5.6 when it is at the period of bacteria growth,then pH would be adjusted to 5.0 at the period of GABA formation.The variable pH fermentation strategy would help to get better fermentation process and higher yield of GABA.

Table 1 Estimated kinetic parameters of biomass growth,glucose consumption,MSG consumption and GABA formation under varying pH

Fig.2.Effect of pH on the kinetic parameters: (a) YX/S;(b) YP/M;(c)μX0;(d) KS;(e) kP;(f) KM of GABA fermentation production.

Table 2 Estimated parameter values of pH-dependent equation

Fig.3.Experimental data and model simulations for GABA production orthogonal experiments with the developed pH-dependent kinetic model under pH level 2 (pH 5.0).Experiment group (a) 251,(b) 212,(c) 223,(d) 234,(e) 245.

Fig.4.Experimental data and model simulations for GABA production orthogonal experiments with the rolling corrected and pH-dependent kinetic model under pH level 2(pH 5.0).Experiment group (a) 251,(b) 212,(c) 223,(d) 234,(e) 245.

3.3.Analysis of problems in the model

As the pH-dependent kinetic model has been developed,pH 5.0 was chosen for validating the pH-dependent model established in Table 2 and the validation results were shown in Fig.3.When glucose concentration is 10-40 g·L-1and MSG concentration is 37.5-150 g·L-1,the trends of simulated curves and experimental data are both sigmoid profiles,and the form of established model is suitable.As the too low concentrations of glucose and MSG are not instructive for the actual fermentation process,the established model is practical and suitable for industrial production.As shown in Fig.3(b)-(e),the simulated curves of glucose consumption and GABA production fit well with data points.The trends of biomass growth and MSG consumption simulated curves are similar with data points,but there are a few of deviations between the simulated curves and experimental data.

Table 3 Simulation results evaluation of rolling corrected and non-corrected model

Because of the biomass yield coefficient on MSG consumed(YX/M) and the product yield coefficient on glucose consumed(YP/S) being omitted,the deviations between the simulated curves and the experimental data existed in biomass growth and MSG consumption.In the fermentation process,L.bre123viscould utilize MSG for biomass growth and could utilize glucose for GABA formation.These two parameters contribute little in fermentation process,but the accumulation of their effects lead to the deviations,especially in biomass growth.For the convenience and simpleness,the model should be kept in a simplified form,and the accuracy of the model can be improved by using correction method.

Fig.5.Validation of the non-rolling corrected and pH-dependent model (a) and rolling corrected and pH-dependent model (b) for GABA production with the data in initial batch fermentation (Control group).

3.4.Development of the rolling correction model

The kinetic equations of glucose consumption (Eq.(5)),MSG consumption(Eq.(8))and GABA formation(Eq.(6))are all coupling with the equation of biomass growth (Eq.(3)).Furthermore,the OD600values are easier to measure than the concentrations of glucose,MSG and GABA,even they can be measured in real-time.So,the rolling correction method on OD600values could be considered to modify the classic mathematical semi-empirical model.The OD600values measured in real-time could help to fix the biomass concentrations and simulate glucose consumption,MSG consumption and GABA formation better.The Fig.4 shows the simulation results of pH 5.0 on pH-dependent model and corrected by OD600value rolling correction.In Fig.4,the trends of simulated curves are not only similar with experimental data points,but also the simulated curves are closer to experimental data than those without the correction term in Fig.3.It is meaning that the kinetic model with the rolling correction term fits better than the model without it.

After the kinetic parameters coupled with pH and the rolling corrected strategy of OD600values introduced into the model,the pH-dependent rolling correction model for describing the GABA production by fermentation was established.To investigate the accuracy of the developed model for instructing the actual fermentation process,the data of the control group in initial batch fermentation were applied to validate the rolling corrected and pHdependent model and the results were present in Fig.5.As shown in Fig.5,the rolling corrected simulated curves are obviously closer to experimental data points than the non-corrected simulated curves,which means that the rolling correction method can significantly improve the accuracy of the model.Correlation analysis of glucose,MSG and GABA concentrations throughout the fermentation process with their respective simulation data was carried out,and the results were listed in Table 3.The correlation coefficientrwith rolling correction was higher than the non-corrected one in three models.And the standard deviation(SD)of error with rolling correction was much lower than the non-corrected one in three models.These two parameters of model evaluation indicate the improvement of accuracy and practicality on model with the strategy of rolling correction.In summary,a pH-dependent model for GABA production of fermentation based on radial basis function was established,and a strategy of rolling correction on OD600values was introduced in to improve the accuracy of the model,which had been validated feasible.

4.Conclusions

In this work,the effect of the OD600decrease on GABA bioconversion activity was investigated firstly,and drew a conclusion that the GAD enzyme activity is not affected by microorganism death ofL.bre123vis.The pH-dependent model was established by estimating values of kinetic parameters from the L25(56) orthogonal experiments.The peak values of proposed pH-dependent model reveals that the optimal pH for biomass growth is 5.6-5.7 and for GABA formation is about 5,respectively.To improve the accuracy of the kinetic model and maintain its simplicity,a rolling corrected strategy with the measured OD600value in real-time was introduced to modify the model.Therefore,the rolling corrected pHdependent kinetic model for GABA production is accurate and practical for instructing the production of GABA which might be facilitated to the industrial scale production.

Nomenclature

KMrestrictive coefficient of MSG,g·L-1

KSMonod saturation constant,g·L-1

kPmaximum specific formation rate of product,g·L-1·h-1

Mconcentration of MSG,g·L-1

msmaintenance coefficient,g·h-1

Pconcentration of GABA product,g·L-1

rcorrelation coefficient,dimensionless

Sconcentration of glucose,g·L-1

Xabsorbance in OD600,dimensionless

YP/Mproduct yield coefficient on MSG consumed,g·g-1

YP/Sproduct yield coefficient on glucose consumed,g·g-1

YX/Mbiomass yield coefficient on MSG consumed,L·g-1

YX/Sbiomass yield coefficient on glucose consumed,L·g-1

μPproduct formation rate,g·L-1·h-1

μXbiomass specific growth rate,h-1

μX0maximum biomass specific growth rate,h-1

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (21621004,22078239),the Beijing-Tianjin-Hebei Basic Research Cooperation Project (B2021210008),Tianjin Synthetic Biotechnology Innovation Capacity Improvement Project(TSBICIP-KJGG-004),and the Tianjin Development Program for Innovation and Entrepreneurship (2018).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2022.05.021.