Jintao Guan,Chaofei Han,Yixin Guan*,Songhong Zhang,Junxian Yun,*,Shanjing Yao

1College of Chemical and Biological Engineering,Zhejiang University,Hangzhou 310027,China

2State Key Laboratory Breeding Base of Green Chemistry Synthesis Technology,College of Chemical Engineering,Zhejiang University of Technology,Hangzhou 310032,China

Keywords:Phenyllactic Acid Biotransformation Lactobacillus Buchneri Uniform Design Overlap Sampling

A B S T R A C T A Lactobacillus buchneri GBS3 strain isolated from the traditional Chinese pickles was used for the production of 3-phenyllactic acid(PLA),an important compound with antimicrobial activities against a wide species of grampositive and gram-negative bacteria and some fungi.The growth performance of this strain in the de Man,Rogosa and Sharpe(MRS)medium,the production of metabolites of valuable organic acids,and the biosynthesis of PLA using this strain as the whole-cell biocatalyst and phenylpyruvic acid(PPA)as the precursor,were investigated experimentally.The uniform design method with overlay sampling was developed for the optimization of the biotransformation conditions.The results showed that although it produced naturally lactic acid with the maximum concentration of 1.84 g·L?1and PLA with the concentration of 0.015 g·L?1after 66 to 72 h cultivation in MRS broth by fermentation,the present strain displayed an effective utilization ability by transforming PPA to PLA.By the uniform design method with overlay sampling for the design and optimization of transformation conditions,a maximum yield of 10.93 g·L?1PLA with the mole conversion ratio of 83.07%from PPA to PLA was achieved under the optimized condition,i.e.,20 g·L?1glucose,270 g·L?1cells,13 g·L?1PPA,pH 8.0 and the reaction time of 15 h,indicating that Lactobacillus buchneri GBS3 was an interesting strain for the biosynthesis of PLA via the microbial transformation.The prediction of PLA yield under different conditions was achieved successfully based on thelimitedinformationofonlyasmallnumberofexperimentsbytheuniformdesignwithoverlaysampling.Therefore,the present methodology is effective and helpful for the optimization of the biosynthesis processes of PLA.

1.Introduction

Bio-based organic acids have attracted considerable interests in recentyearsduetotheincreasingdemandsofthesechemicalsinfood,cosmetic,pharmaceutical,and chemical industries[1-4].The production of organic acids by microbial fermentation or transformation has received intensive attention because of the sustainable and friendly advantages compared with the traditional chemical synthesis.Phenyllactic acid(PLA)is one of such organic acids,which can be generated naturally as metabolites by various microorganisms,such as lactic acid bacteria[5-14],propionibacteria[15,16],phototrophic bacteria[17],Geotrichum candidum[18]and Wickerhamia fluorescens[19].This interesting compound has antimicrobial activities against a wide species of grampositive and gram-negative bacteria,such as Staphylococcus aureus,Bacillus cereus,Escherichia coli,and Listeria[11,20,21],and also some fungi like Aspergillus,Penicillium,and Fusarium species[6,12,22],and thus can be used as food preservatives or feed additives.PLA can also be used as pharmaceutical agent to treat coronary diseases or skin wrinkles[11].Recently,PLA has been suggested to be used as the monomer andbuildingblockchemicalfortheproductionofnewbio-basedplastics ofpoly(phenyllacticacid)s,analternativeclassofbiocompatibleandbiodegradable polyesters for poly(lactic acid)s plastics[19,23].Unlike the hard and brittle inherent characteristics of poly(lactic acid)s,plastics of poly(phenyllactic acid)s have improved properties regarding the thermostability and ultraviolet-absorbing ability due to the bulky aromatic side chains[23-25].Therefore,PLA has potential applications in chemical,pharmaceutical,biotechnological and food industries.

Biological production of PLA can be achieved by either the microbial fermentation with natural microorganisms and recombinant genetic strains,or the transformation using biocatalysts of whole cells[11,26].Dieuleveux et al.[18]have found that G.candidum of the natural milkflora grown in Trypticase soy broth with yeast extract could produce 0.6 to 1 g·L?1PLA.Valerio et al.[27]measured the production yield of PLA by 29 lactic acid bacterial strains belonging to 12 species regarding most used in the production of fermented foods.They reported that the maximumproductionyieldwas0.57mmol·L?1(0.095g·L?1)byLactobacillus brevis ITM1D among these strains in MRS broth.Li et al.[7,8]screened Lactobacillus strains from Chinese traditional fermented pickles and the obtained strain Lactobacillus sp.SK007 produced about 0.55 mmol·L-1(0.091 g·L?1)PLA in MRS broth.By optimizing the fermentation conditions using the response surface methodology based on a uniform design and a central composite rotatable design,the production yield of 2.30 to 2.42 g·L?1PLA was obtained by the fermentation with Lactobacillus sp.SK007[9,10].However,mostofotherlacticacidbacterialike Lactobacillus plantarumandWeissellaconfusastrainsproducedPLAwiththeconcentration less than 2 g·L?1in regular MRS broth[11,26].

In general,PLA is a by-product of phenylalanine(Phe)metabolism in lactic acid bacteria,in which Phe is transaminated to phenylpyruvic acid(PPA)byaromaticaminotransferaseandfurtherreducedtoPLAbydehydrogenase and NADH[7,9-11].The transamination is a reversible reaction and also the limited step in this process.As the direct precursor of PLA,PPA plays a more important role than that of Phe in the production of PLA.With the fed-batch fermentation using PPA as the substrate,the production yield of PLA by Lactobacillus sp.SK007 was improved to 17.4 g·L?1,but the conversion ratio of PPA to PLA was low,i.e.,51.1%[10].Zheng et al.[28]reported the high concentration of PLA,i.e.,37.3 g·L?1,but together with by-product lactic acid of 66 g·L?1was achieved by the fed-batch whole-cell bioconversion using the thermophilic strain BacilluscoagulansSDMwithfeedingofPPAandglucosesubstrates and maintained pH by addition of NaOH.Zhu et al.[14]produced PLA by biotransformation of PPA using permeabilized Lactobacillus casei cells as the whole-cell catalysts and the PLA concentration of 5.3 g·L?1with the conversion ratio of PPA to PLA of 65.5%.

Recently,the biosynthesis of PLA with the recombinant strains by employing technologies of genetic engineering have received attentions[23,29-35].Fujita et al.[23]expressed the Wickerhamia fluorescens gene(pprA)encoding a phenylpyruvate reductase in E.coli for the production of D-PLA and the final concentration of 29 g·L?1was achieved.The recombinant strain was also effective in the production of D-PLA by simultaneous saccharification and fermentation of renewable cellulose,sorghum bagasse or sweet sorghum[29-31].Zhu et al.[32]coexpressed L-lactate dehydrogenase from Lactobacillus plantarum subsp.plantarum and glucose dehydrogenase from Bacillus megaterium in the recombinant E.coli for the production of L-PLA from PPA and the L-PLA concentration of 17.2 g·L?1(103.8 mmol·L?1)was reported.Xu et al.[33]identified an NADH-dependent phenylpyruvate reductase LaPPR from Lactobacillus sp.strain and constructed the co-expression system withitandglucosedehydrogenaseintherecombinantE.coli fortheproduction of PLA.They reported that 100 g·L?1PPA was asymmetrically reducedintoD-PLAwiththeisolationyieldof91.3%.Althoughsomesuccesseshavebeenachievedbyusingtherecombinantstrainsandtheconcentration of PLA in the fermentation broth has been improved,the production of PLA by these genetic strains has some drawbacks,such as the complexity of the preparation of recombined plasmids,the instability of the plasmids and the uncertain bio-safety of the strains.

Compared with the genetic engineering approaches,the production of PLA via fed-batch fermentation or bioconversion by the whole-cell biocatalysts with strains generally regarded as safe using PPA as the substrate is more attractive in some application areas,especially those in food industry,due to the cost and safety benefits.For the bioconversion,severalimportantvariablesliketheconcentrationsofthecells,glucose and PPA,the initial pH and the reaction time,could significantly influence the bioconversion process and thus the yield of PLA and the conversion ratio from PPA to PLA.However,some drawbacks existed inthisapproachsuchaseitherthelackofbacteriastrainswithhighsynthesis efficiency,or the lack of effective design methods for the optimization of these processes by limited experimental trials.Orthogonal design has been used to optimize the process of PLA production[14].But this method is useful only for the processes with a few variables and levels,and a large number of trails are generally needed.The uniform design based on number theory[36,37]has been used to optimize the medium components for the production of PLA by the combination of the canonical analysis and response surface approach[9].By this method,the multiple regression was employed to get the prediction model for the production yield of PLA and the number of experimental trials were reduced significantly.However,in the case that interactions ofthevariableswereconsidered,thenumberofexperimentaltrialscould be not enough to ensure the satisfied regression and thus,caused the uncertainty of the model prediction.In this case,the experimental numbers could be too small to ensure the establishment of an accurate prediction equation of the yields or the conversion ratios under various conditions of the variables.Therefore,we proposed a new optimization method by combining the overlap sampling method with the uniform design.In this method,it would take twice the number of level by alternating the order sequence of factors,then fitting the results of these experiments,and we could obtain a function that can describe the relationship between the independent variables and dependent variable.

Inthisstudy,astrainofLactobacillusbuchneriGBS3isolatedfromthe Chinese fermented pickle was employed to synthesize PLA.The growth performance and the natural metabolite of this strain in MRS medium wereinvestigatedexperimentally.TheproductionofPLAbytheconversion of PPA substrate using the cells of this strain as the whole-cell catalysts was achieved.In addition,a new method by combining the uniformdesignwiththeoverlapsamplingmethodwasusedtooptimize the PLAyield regarding the conversion conditions of buffer pH,reaction time,as well as the concentrations of cells,glucose and PPA in a reasonable set of experimental trials to enhance the certainty of the model prediction.

2.Materials and Methods

2.1.Materials and bacterial strains

PLA(98%)andPPA(98%)werepurchasedfromSigma-Aldrich(Saint Louis,Missouri,USA).Glucose,peptone,beef extract and yeast extract were obtained from Sangon Biotech(Shanghai,China).Other chemicals were of analytical grade from local resources.

L.buchneri GBS3 strainwasisolatedfromChinesetraditionalpicklesand depositedasthepatentstrainattheChinaCenterforTypeCultureCollection(CCTCC NO:M2016215)after being identified by 16S rRNA gene sequence.

2.2.Growth performance and fermentation production of PLA by L.buchneri GBS3

The growth performance of L.buchneri GBS3 was investigated in MRS medium(20 g·L?1glucose,10 g·L?1peptone,10 g·L?1beef extract,5 g·L?1yeast extract,5 g·L?1CH3COONa,2 g·L?1K2HPO4,2 g·L?1triammonium citrate,1 ml·L?1Tween-80,0.58 g·L?1MgSO4,0.25 g·L?1MnSO4).Typically,a pre-culture was prepared as the seed in a 50 ml flask containing 40 ml MRS medium without shaking at 30°C for 16 h.Then,24 ml of the obtained seed culture was inoculated into to 1200 ml MRS medium.The mixture was distributed into 50 ml flasks and each flask contained 30 ml broth.These flasks were then maintained underthesameconditioninanincubatorforculture.Ata giventimeof2,4,6,8,10,12,14,16,18,20,24,28,32,42,54,66,72or84h,culturesfrom two of these flasks were used as the samples for the detection of the cell growth and the metabolites of organic acids.

Bacterial cells of L.buchneri GBS3 were collected from the culture samples via centrifugation(8000 r·min?1,15 min)and used for the measurements of cell concentration and morphology.Cells at a given volume of the culture broth were dried at 105°C to constant mass for 12 h and the dry cell mass(DCM)was then determined by the cell mass and the culture volume.The supernatant of the culture samples after centrifuging was used for the investigation of the glucose consumption,the metabolite contents of lactic acid(LA)and PLA,as well as pH of the culture broth.

2.3.Whole-cell bioconversion of PPA to PLA by L.buchneri GBS3

The production of PLA by whole-cell bioconversion using L.buchneri GBS3 as the biocatalysts and PPA as the precursor were carried out according to the method reported by Zhu et al.[14].Typically,cultures was incubated in 1000 ml flask containing 800 ml in MRS medium for 72 h at 30°C without shaking and the bacterial cells were collected by centrifugation.After washed twice with phosphate buffer,the cells were frozen at?20°C for 4 h and then thawed at room temperature.Then,a given amount of cells were re-suspended in 5 ml phosphate buffer by adding glucose and PPA in a 50 ml flask for the production of PLA via bioconversion.The bioconversion was achieved for a certain time at 35°C.The obtained broth after bioconversion was centrifuged(8000 r·min?1,15 min)and the supernatant was used for analysis.The concentrations of PPA and PLA were measured and the mole conversion ratio from PPA to PLA was calculated by the final concentration of PLA produced and the initial concentration of PPA in the broth.

2.4.Morphology of cells by scanning electron microscope

Morphology of cells during the growth process in the static culture was investigated by scanning electron microscope(SEM,S-4700,Hitachi,Japan).The fresh cells collected from the samples at different growth stages were washed twice with 0.9%NaCl and then fixed in 2.5%glutaraldehyde for 8 h at 4°C.The obtained cells were dehydrated stepwise in ethanol(10%-30%-50%-70%-90%-95%-100%),freezing dried and coated with platinum for SEM.

2.5.Analysis

The concentration of glucose during the culture procedure was analyzed by 3,5-dinitrosalicylic acid(DNS)method[38],and the changing of pH was measured by a pH meter(FE20,Mettler-Toledo Ltd.,Switzerland).Concentrations of PPA and PLA in the bioconversion broth as well as concentrations of these compounds and lactic acid(LA)in the fermentation broth were analyzed by high performance liquid chromatography(HPLC)using Agilent 1260 infinity system equipped with DAD detector at the wavelength of 210 nm[27].Agilent Zorbax SB-C18 column(4.6 mm × 150 mm,5 μm)was used and the temperature was maintained at 30°C.Mobile phase A was 0.05%aqueous trifluoroacetic acid and the phase B was methanol containing 0.05%trifluoroacetic acid.The elution was performed by changing the ratio of A/Bas follows,90/10,0/100,0/100and 90/10 withtheruntimeof0,20,23 and 25 min,respectively.LA,PPA and PLA was eluted separately at 2.1,9.8 and 10.7 min.The stereo-selective assay of PLA was measured using HPLC with a Chiralcel OD-H column(2.1 mm × 150 mm,5 μm)at210nm.Themobilephasewasahexane/isopropanolsolventmixture(98:2,v/v)containing 0.05%trifluoroacetic acid with a flow rate of 1 ml·min?1at 30 °C.The enantiomeric excess(ee)value of L-PLA was defined as(CL-PLACD-PLA)/(CL-PLA+CD-PLA)×100%,where CL-PLAand CD-PLArepresented the concentrations of L-PLA and D-PLA,respectively.

3.Uniform Design with Overlay Sampling Methodology

In this work,the uniform design with overlay sampling methodology is proposed to optimize the bioconversion conditions.This method is aimed to optimize the bioconversion conditions by considering both the reasonable number of experimental trials and the improved model prediction between the objective parameters and the variables.The uniform design is used to reduce the experimental numbers,while an overlay sampling methodology is combined together to ensure the reasonable set of experimental trails for the satisfied regression and thus the improved model prediction.The overlay sampling methodology is carried out by the arrangement of the experiments twice at the same set of uniform design levels and variables.The only difference between these two sets was that the sequences of the partial variables and their levels were exchanged each other.This strategy was based on the fact that in the uniform design,each of the variables and their levels were not fixed but randomly.

Here,five variables(the concentration of cells X1,the concentration of PPAX2,theconcentrationof glucoseX3,theinitialpH X4andthereactiontimeX5)withfifteenlevelseachareconsideredandtheexperiment value of experimental trials are shown in Table 1.We used a(157)design table to arrange the experiments,where U represents the uniform design,subscript 15 is the test number,superscript 7 is the maximum factor number and normal 15 is the level number,respectively.From uniform design theory,this table has the least deviation as seen in Table 2.Following the overlay sampling strategy,15 experiments in the first group were carried out according to the arrangement variables andlevelsinafixedsequence,andthenpartialvariablesandtheirlevels were exchanged each other(the concentration of PPA switched to the position of X4as X2′,the concentration of glucose switched to the position of X2,as X3′,and the initial pH switched to the position of X3as X4′,respectively)and another 15 experiments in the second group were performed,as shown in Table 2.In order to decrease the measurement errors of data,each trial was conducted in duplicate.

Table 2 Uniform design table (157)for the whole-cell bioconversion of PPA to PLA by L.buchneri GBS3

Table 2 Uniform design table (157)for the whole-cell bioconversion of PPA to PLA by L.buchneri GBS3

No. X1,X1′ X2,X3′ X3,X4′ X4,X2′ X5,X5′1 5 7 9 11 15 2 10 14 2 6 14 3 15 5 11 1 13 4 4 12 4 12 12 5 9 3 13 7 11 6 14 10 6 2 10 7 3 1 15 13 9 8 8 8 8 8 8 9 13 15 1 3 7 10 2 6 10 14 6 11 7 13 3 9 5 12 12 4 12 4 4 13 1 11 5 15 3 14 6 2 14 10 2 15 11 9 7 5 1

Table 1 Variables and their levels for the whole-cell bioconversion of PPA to PLA

The relationship between the yield of PLA and the variables was obtained by multipleregressionanalysis ofthenormalizeddata ofthevariables and the objective under the consideration of interactions of variables.The nonlinear equation is shown as follows:whereYistheyieldofPLA,andxi=(Xi?Xi,min)/(Xi,max?Xi,min)(i=1,2,3,4,5)are the normalized concentration of cells,the normalized concentration of PPA,the normalized concentration of glucose,the normalized initial pH,and the normalized reaction time,respectively,and a0,bi,ciand dijare coefficients of each terms.

4.Results and Discussion

4.1.Growth characteristics andcell morphologyofL.buchneri GBS3inMRS medium

The growth curve of L.buchneri GBS3 cells,the change of pH in the broth and the glucose consumption during the static fermentation process are shown in Fig.1.After about 4 h culture time,the glucose concentration and pH of the culture decreased slightly with the increase of time,indicating that the metabolism was active within the cells and metabolites like organic acids were produced into the broth.At the logarithmic phase(18 to 28 h),vigorous growth of cells occurred and the dried concentration of cells increased from 0.9 to 2.3 g·L?1,while pH of the brothdecreasedrapidlyfrom5.4to4.8andtheglucoseconcentrationdecreasedfrom12.2g·L?1to1.5g·L?1.After32h,glucoseinthebrothwas exhausted(＜0.2 g·L?1),the culture pH was about 4.4,the dried concentrationofcellswasabout2.3g·L?1andthestationaryphaseoccurred.No obvious decline phase was observed even the culture was maintained for 50 h more,indicating the stable cells and enzymes within the broth.

Fig.1.The curvesof cell growth,pHchange andglucose consumption for L.buchneri GBS3 in MRS medium.

The morphology of L.buchneri GBS3 was investigated by SEM.Illustrations of the obtained 3 images were selected in the lag,logarithmic and stationary phases at the culture time of 12 h,20 h,or 52 h,respectively,are shown in Fig.2.Compared with the morphology at the lag or logarithmic phase,there are no obvious changes of morphology of the cells in the stationary phase(up to 84 h,data not shown),indicating that the cells were stable and thus could be suitable as stable wholecell catalysts for bioconversion.

4.2.Production of LA and PLA of L.buchneri GBS3 in MRS medium

The metabolic pathway of lactic acid bacteria often linked to accumulation of several organic acids.Lavermicocca et al.firstly reported that PLA was produced by L.plantarum 21B[6].Since then,various LAB strains have been discovered in the fermentation production of PLA with the relative low product concentration(0.01-0.53 g·L?1),[11,26].Herein we tested the metabolites by L.buchneri GBS3 during the static fermentation in MRS medium and several different main metabolites were observed.As an example,the chromatographic diagram of the cell-free supernatant sample obtained via centrifugation at the culture time of 42 h at the stationary phase is displayed in Fig.3(a).Organic acids like LA,PPA and PLA were observed among these metabolites.The concentrations of these three important metabolites in the culture broth during the fermentation procedure are shown in Fig.3(b).It was seen that the production of PPA started from the lag phase,however,LA and PLA mainly started from the logarithmic phase until to the early stage of the stationary phase with the culture time up to about 40 h.After that,the production of LA was almost stopped,but theproduction of PLAcontinuously increased up to about 66 h.The reasoncouldbeduetothedifferentmetabolismpathwaysofLAandPLAby L.buchneri GBS3.But the production of PPA always fluctuated within a narrow range of 0.0011 to 0.0014 g·L?1,the reason could be that PPA was an intermediate in multiple metabolic pathways.The maximum levelofLAwasobservedas1.84g·L?1attheculturetime66h,themaximum level of PPA was 0.0014 g·L?1at the time of 42 h and the maximum level of PLA was 0.015 g·L?1at the time of 72 h.These values indicated that this strain has the ability to produce PLA,albeit the concentrationof PLAwaslower thanthose intheculturesbyotherLactobacillus strains[7,8,11,26].

4.3.Uniform design with overlay sampling methodology for the whole-cell bioconversion process from PPA to PLA by L.buchneri GBS3

Theuniformdesignwithoverlaysamplingmethodwasemployedto investigate the bioconversion process regarding five variables,i.e.,the concentration of cells,the concentration of PPA,the concentration of glucose,the initial pH and the reaction time,and 15 levels of each of thesevariables.The yield ofPLAwasconsideredastheobjectiveparameter.A total of 30 experimental trials were carried out according to Tables 1 and 2 and the obtained results are listed in Table 3.In order to ensure the measurement accuracy,duplicates of each point were performed.

Fig.2.Cellmorphologyof L.buchneriGBS3 inMRS mediumat(a)thelag phase,12h,(b)thelogarithmicphase,20h,and(c)thestationaryphase,52h.Theculture broth ineachflask was 30 ml broth and the temperature was maintained at 30°C.

Fig.3.HPLC chromatogram of the cell-free supernatant at the time of 42 h(a)and variations of metabolites of LA,PPA and PLA with the culture time(b)during the static fermentation process of L.buchneri GBS3 in MRS medium.All the assays were carried out in duplicate.

The functional relationship between the objective and five variables in the bioconversion process of PLA with the whole-cell catalysts of L.buchneri GBS3 was assumed to following a quadratic equation considering the interactions of these variables.In order to eliminate the unit effects of different variables,the absolute values of each variable were normalized according to Eq.(1)and the following prediction model was then obtained by the multiple regression.

where Y is the yield of PLA,x1=(X1?30)/280 the normalized concentration of cells,x2=(X2?5)/28 the normalized concentration of PPA,x3=(X3?6)/14 the normalized concentration of glucose,x4=(X4?6.2)/2.8 the normalized initial pH,and x5=(X5?2)/14 the normalized reaction time,respectively.

In Eq.(2),there are 21 coefficients needed to be determined by the regression.Ifthetraditionaluniformdesignwasemployedtoarrangetheexperimental trials,a set with a total of 15 trials could be carried out.In that case,the data numbers were not enough to fix the five-variable quadratic equation.However,by the uniform design with overlay sampling methodology,the set with a total of 30 trials were achieved,which was enough tofit equation.The comparison of the yields of PLA in the model calculation Ycaland experimental data Yexpwas also summarized in Table 3.The relevancybetweencalculatedandexperimentalvalues was0.7323andtheaverage relative error was 34.9%.The deviation is slightly high.The reason is that the bioconversion using the whole-cell catalysts of L.buchneri GBS3 is acomplexprocessinfluencedbynumerousfactorssuchasenzymeactivity,inhibition of substrates,by-products or products,as well as cell properties.However,the agreement between the calculation and experimental data was acceptable and thus expected to be valid in the prediction of PLA yield under different conversion conditions.

Table 3 Uniform design and the obtained yield values for the bioconversion process with five variables and 15 levels

4.4.Optimization of bioconversion from PPA to PLA

The yields of PLA from PPA by conversion using the whole-cell catalysts of L.buchneri GBS3 under various conditions were predicted using Eq.(2).Six conditions(Nos.A to F)with the predicted yields higher than 10 g·L?1were employed as the possible candidates for the optimization,and four conditions(Nos.G to J)with the yields randomly from 0.81to7g·L?1werealsousedtoverifytheprediction.Theexperimental yields under these conditions were carried out and the corresponding mole conversion ratios from PPA to PLA(αexp)were also calculated.The HPLC diagrams of the conversion broth under these conditions are shown in Fig.4.The comparison between the experimental and predictedresultsissummarizedinTable4.Ascanbeseen,agoodagreement between the experimental results and predicted values was observed,indicating that the correlation was successful and Eq.(2)was effective in the prediction and optimization of the bioconversion process by L.buchneri GBS3.

In fact,those conditions for both the high yield of PLA and the high molar conversion from PPA to PLA could be considered as the optimization conditions for the bioconversion.From Table 4,both high yield and highconversionratiowereachievedincasesofNo.CandNo.F.Intheformer case,i.e.,the concentration of cells of 310 g·L?1,the PPA concentration of 11 g·L?1,theglucose concentrationof 20 g·L?1,thereactiontime of 10 h,and the buffer pH of 8.2,a yield of 10.03 g·L?1with a molar conversion of 90.08%was achieved.However,in the latter case No.F,the maximum yield of PLA,i.e.,10.93 g·L?1,which was about 9%higher than that in the case of No.C,and the molar conversion ratio from PPA to PLA,i.e.,83.07%,was also satisfied.Therefore,by the overall consideration of both the yield of PLA and the conversion ratio from PPA to PLA,the conditions with the concentration of cells of 270 g·L?1,the PPA concentration of 13 g·L?1,the glucose of 20 g·L?1,the reaction time of 15 h,and the buffer pH of 8.0 were considered as the optimization case for the bioconversion.In the study of PLA enantioselectivity,PPA could be asymmetrically reduced into D-PLA with ee value of 94.8%by using the resting wet cells of L.buchneri GBS3.

5.Conclusions

Fig.4.HPLC diagrams under conditions of A to J in the verification tests.

Table 4 The predicted and experimental yields and the mole conversion ratios in the verification tests

L.buchneri GBS3 is an interesting strain for the production of PLA by using PPA as the key substrate by whole-cell bioconversion.This strain grew rapidly in MRS and the logarithmic phase was observed after cultured about 18 h and the maximum concentration of dry cells was 2.3 g·L?1for the culture time of 28 h.It has naturally produced 1.84 g·L?1lactic acid and 0.015 g·L?1PLA after 66 to 72 h cultivation in MRS broth under the present culture conditions.No obvious changes ofcellmorphologyofL.buchneriGBS3wereobservedduringthelag,the logarithmicandthestationaryphasesandthecellswerestableformore than 50 h during the whole fermentation process.This strain displayed aneffectiveutilizationabilitybytransformingPPAtoPLAviathewholecell bioconversion.

The uniform design with overlay sampling methodology proposed in this work was an effective approach to achieve both the reasonable number of experimental trials in the design of the bioconversion processandthesatisfiedpredictionoftheyieldsofPLAunderdifferentconditions.The obtained model was then used successfully to optimize the transformation conditions.By this method,a maximum yield of 10.93 g·L?1PLA with the mole conversion ratio of 83.07%from PPA to PLA was achieved under the optimized condition,i.e.,20 g·L?1glucose,270 g·L?1cells,13g·L?1PPA,pH8.0 and thereactiontimeof 15h.This is a relatively high yield compared to previously reported results[7,14].This method required thelimited information of only a small number of experiments but reasonable prediction of the objectives under various conditions.The regression for the model is simple and easy to be conducted,and thus,the present methodology is interesting in the optimization of bioconversion processes.