coli TOP10 pQR239

R. Melgarejo-Torres a, O. Castillo-G.J. Lye c, S. Huerta-Ochoa a,

utnoman Sistem

rsity ColXalcaltz

c modeowed thbition aobtaineivation

Lactones have wide applications in avorings, as precursors ofanticancer and antihypertension drugs and in the pharmaceuticalindustry [1]. Lactones have been obtained by BaeyerVilliger reac-tions using catalytic process [2] or whole cell bioconversion with acyclohexanone monooxygenase (CHMO) expressed in Escherichia

ws enzymerically pur

pounds [4]. However, it has been reported [5,6] thatbioconversion using CHMO is inhibited by the substraproduct at concentrations above 0.4 and 4 g L1, respectively. Addi-tionally, Bennett [7] reported that enzyme inactivation may occurdue to residue oxidation of two serines close to the active site. Anumber of strategies have been proposed to avoid these types ofinhibition, such as substrate feeding and in situ product removalusing Lewatit resin [8], biocatalyzer encapsulation to preventCHMO oxidation [9], the use of ionic liquids as an immiscible phasesubstrate reservoir and in situ product removal and maintaining

Corresponding author. Tel.: +52 58044999; fax: +52 5558044712.E-mail addresses: coca@xanum.uam.mx (O. Castillo-Araiza),

danieltm99@hotmail.com (D. Torres-Martnez), g.lye@ucl.ac.uk (G.J. Lye),

Chemical Engineering Journal 240 (2014) 19

Contents lists availab

Chemical Engine

w.sho@xanum.uam.mx (S. Huerta-Ochoa).1. Introduction coli TOP10 pQR239 [3]. The use of whole cells allotor regeneration for the production of enantiome1385-8947/$ - see front matter 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cej.2013.11.047cofac-e com-ketonete anda r t i c l e i n f o

Article history:Received 20 September 2013Received in revised form 18 November 2013Accepted 23 November 2013Available online 1 December 2013

Keywords:BioconversionCyclohexanone monooxygenaseModellingKinetic parametersOxygen inactivationEscherichia coli

a b s t r a c t

The aim of the current work was to develop a pseudo intrinsic kinetic model, based on elementary reac-tions, to describe the behavior of the bioconversion of ketones using Escherichia coli TOP10 pQR239. Sincethere are no reports of the oxygen inactivation constant in the literature, this study gave new insights tond optimal conditions of a suitable oxygen supply during the bioconversion. In this model the reactionmechanism proposed followed the formalism of LangmuirHinshelwood and considered both substrateinhibition and oxygen inactivation by the formation of intermediary complexes. Therefore, approxima-tions of the pseudo equilibrium of reaction rates or steady state intermediary species were not consid-ered, which allowed for identifying the role of each reaction step involved in the bioconversion. Thiskinetic model adequately described the observations with and without substrate inhibition and/or oxy-gen inactivation. And the regression and the estimated parameters were statistically signicant, makingthese analyses reliable regarding the kinetic behavior of CHMO. Then, substrate and oxygen afnity andinhibition constants were obtained from the kinetic parameters of the model. It was observed that oxy-gen and substrate presented similar afnity constant values. The substrate inhibition (KIS) and oxygeninactivation (KIO2) constants were determined to be 9.98 lM and 22.3 lM, respectively, showing thatthe CHMO enzyme was twice more sensitive to inhibition by an excess of substrate than oxygen.

2013 Elsevier B.V. All rights reserved.aDepartamento de Biotecnologa, Universidad AbGrupo de Procesos de Transporte y Reaccin eD.F., MexicocDepartment of Biochemical Engineering, UnivedUniversidad Politcnica de Tlaxcala, San Pedro

h i g h l i g h t s

A non-reported pseudo intrinsic kineti The reaction mechanism proposed foll Complexes formation of substrate inhi Afnity and inhibition constants were This is the rst report for oxygen inactAraiza b, P. Lpez-Ordaz a, D. Torres-Martnez d, M. Gutirrez-Rojas a,

Metropolitana, P.A. 55-535, 09340 Iztapalapa, Mxico D.F., Mexicoas Multifsicos, Departamento de Ingeniera de Procesos e Hidrulica, Universidad Autnoma Metropolitana, Mxico

lege London, London WC1E 7JE, United Kingdominco Tepeyanco, Tlaxcala, Mexico

l was developed on basic reactions.e formalism of LangmuirHinshelwood.nd oxygen inactivation are accounted.d from the kinetic model.constant (22.3 lM).Kinetic mathematical model for ketone bioconversion using Escherichiajournal homepage: wwle at ScienceDirect

ering Journal

elsevier .com/locate /cej

l Enthe biocatalyzer (whole cells) in the aqueous phase [4], a stirredtank partitioning bioreactor using ionic liquids as the dispersedphase [10] and CHMO molecular structure changes to fold the ser-ines susceptible to oxidation inside the enzyme [11].

Despite several experimental studies on the molecular structureof CHMO, its catalytic activity and reaction rates of the intermedi-ate steps in the overall reaction in order to propose a basic biocon-version mechanism [1214], there have been few studies onkinetic modeling considering simultaneous substrate, productand oxygen inhibition phenomena. Some pseudo empirical kineticmodels have been reported following MichaelisMenten approachto describe product formation and substrate consumption inmonooxygenase kinetics [15,16,17]. However, they do not takeinto account elementary reactions accounting for oxygen as a sec-ond substrate for bioconversion. The use of this kind of kineticmodels reduces the number of required kinetic parameters; never-theless, the estimated kinetic values depend on the catalyst con-centration and become independent of the reactor size and itsgeometrical conguration, providing uncertainties for scaling-upketone bioconversion. In this sense the development of a kineticmodel based on an elementary reaction mechanism describingBaeyerVilliger bioconversion will make it possible to describe,understand and nd optimal conditions for carrying out this kindof bioconversion but mainly for design and scale-up.

The objective of this work was to develop a pseudo-intrinsickinetic model to describe the behavior of the bioconversionof ketones using whole cells. The mathematical model wasbased on an elementary reaction mechanism that followed theLangmuirHinshelwood formalism, considering inhibition andinactivation by the formation of substrate or oxygen enzymecomplexes in the active site of the CHMO. The mathematical modelwas adjusted and kinetic parameters were estimated for threepossible cases of the bioconversion of ketones: (1) without anyinhibition, (2) inactivation by oxygen excess and (3) inhibition bysubstrate excess. The mathematical model was validated throughthe comparison of experimental data obtained for simultaneoussubstrate inhibition and oxygen inactivation versus calculated val-ues using the estimated kinetic parameters. The kinetic parameters

Nomenclature

CHMO cyclohexanone monooxygenaseNT enzyme total concentration (lg g1 of biomass)hE free enzyme fractionhEO2 enzymeoxygen complex fractionhEO2S enzymeoxygen-substrate complex fractionhEO2SS substrate inhibition complex fractionhO2EO2 oxygen inactivation complex fraction[P] product concentration (g L1)

2 R. Melgarejo-Torres et al. / Chemicaobtained yielded valuable information to determine which may bethe limiting step in the bioconversion reaction.

2. Materials and methods

2.1. Microorganism and chemicals

The E. coli strain TOP10 pQR239 was kindly provided by Profes-sor JohnM.Ward (University College London, London, United King-dom) for research and academic purposes, and is referred tohereafter simply as E. coli. To prepare inocula for bioconversionexperiments, E. coli cells were cultured in Erlenmeyer asks of250 mL containing 70 mL of a complex media (in g L1): tryptone10.0, yeast extract 10, NaCl 10.0, in phosphate buffer 50 mM pH7.0, supplemented with 10 g L1 glycerol. Culture media wassterilized in an autoclave at 120 C for 15 min and supplementedwith 100 mg L1 ampicillin (previously lter sterilized using a0.25 lm lter). Erlenmeyer asks were incubated at 150 rpm for16 h at 37 C. After this 16 h growth period, cyclohexanone mono-oxygenase expression was induced by adding the necessaryamount of arabinose solution (100 g L1) to reach a nal concen-tration of 2 g L1. After 3 h of induction, cells were harvested bycentrifugation at 5000 rpm for 10 min.

Bicyclic ketone bicycle[3.2.0]hept-2-to-6-one (P 98%) andbicyclic lactone (1S,5R)-(-)-2-oxabiciclo[3.3.0]oct-6-en-3-ona(P 99:0%)) (Fluka, Switzerland) were used as the substrate andproduct standards, respectively. Tryptone, yeast extract, NaCl andglycerol were purchased from Sigma Aldrich (EUA).

2.2. Stirred tank bioreactor description

A module with two glass 100 mL stirred tank bioreactors(MMBR100, UAM-I, Mexico) was used for all bioconversion studies.The jacketed bioreactors had an internal diameter of 4.75 cm andan operating volume of 70 mL (HL/DT = 0.87). The bioreactors weretted with a single, six at blade Rushton turbine, Di = 1.9 cm (Di/DT = 0.40), located 1.9 cm from the at base of the vessel. The bio-reactor was equipped with four equidistant bafes, 0.5 cm inwidth, to enhance mixing.

2.3. Oxygen mass transfer coefcient (kLa) determination

Optical ber dissolved oxygen mini sensors (PreSens, GmbHGermany) were used for kLa determinations. The oxygen sensorswere coupled to an OXY-4 mini four-channel oxygen meter(PreSens, GmbH Germany). Oxygen mass transfer coefcients(kLa) were calculated according to the dynamic method andmathematical model proposed by Fuchs et al. [18], which takesinto account the electrode response time and the dimensionlessdissolved oxygen concentrations in the bioreactor (Eq. (1)). Theeffect of operating conditions, including agitation (750, 1350 and1950 rpm) and aeration (0.75, 1.0 and 1.4 vvm) rates on kLa was

[O2] oxygen concentration (g L1)[S] substrate concentration (g L1)rn reaction rate of each enzymatic complex (g L1 h1)kj kinetic constant of the reaction (s1)kLa oxygen mass transfer coefcient (h1)K afnity or inhibition constant of substrate and/or

oxygen (g L1)

gineering Journal 240 (2014) 19studied.

YP KP ekLat kLa eKPtKP kLa 1

where YP is the dimensionless dissolved oxygen concentrations inthe bioreactor dened by Eq. (2).

YP CP CP

CP CP02

where CP is the saturated oxygen concentration (7.2 mg L1, which

was determined with the OXY-4 module software using theatmospheric pressure of Mexico City, 1016.9 hPa), CP is thedissolved oxygen concentration, CP0 is the initial dissolved oxygenconcentration the bioreactor. KP is the electrode constant denedas the inverse of the response time, kLa is the oxygen mass transfer

product formation. The reaction mechanism proposed in this paper(Fig. 1) takes into account the possible formation of two inactivecomplexes, one involving the over-oxidation of the active site(O2EO2) at a reaction rate r2 and a stoichiometric coefcient of 1,

EO2 O2 k2

k02O2EO2; r2 k2NThEO2 O2 k02NThO2EO2 6

and another that represents substrate inhibition (EO2SS) at a reac-tion rate r4 and a stoichiometric coefcient of 1

EO2S S k4

k04EO2SS; r4 k4NThEO2SS k04NThEO2SS 7

Product inhibition was not considered in the proposed reactionmechanism since the maximum product concentration, estimatedfrom the specic rate of 0.11 g lactone g biomass1 h1 [10], andthe substrate concentrations used in this study did not reach inhi-bition concentrations.

l Encoefcient, and t is the time. Fitting was performed using a non-lin-ear regression LevenbergMarquardt algorithm in Polymath.

2.4. Aerated power consumption (Pg) measurement

Aerated power consumption (Pg) was measured using the meth-odology recently discussed by Ascanio et al. [19]. Basically, themethod is based on electrical measurements performed directlyin the bioreactor stirrer shaft motor by watt meters and ammeters.To take into account losses occurring in the agitation system, ablank of the measurements was rst performed with an empty bio-reactor. All measurements were performed in triplicate.

2.5. Bioconversion experiments

Bioconversion experiments were carried out using 3.0 g of bio-mass L1. A buffer solution consisting of 50 mM phosphate pH 7.0supplemented with 10 g glycerol L1 was used as the aqueousphase for the bioconversion media. A full face-centered centralcomposite experimental design with three factors and elevenexperiments was used. Three levels of each factor were studied:agitation rate (750, 1350 and 1950 rpm), aeration rate (0.75, 1and 1.4 vvm) and substrate concentration (0.35, 0.80 and1.0 g L1). During bioconversion experiments, samples of 500 lLwere taken every 3 min during the rst 15 min and then everyhour for three hours. Samples were quickly frozen to stop the reac-tion. For substrate and product analysis, the samples were centri-fuged at 5000 rpm for 10 min to separate the biomass and thesupernatant was analyzed. The substrate and product were quanti-ed by gas chromatography.

2.6. Analysis

Gas chromatography (GC) was used to quantify the concentra-tions of bicycle[3.2.0]hept-2-en-6-one and its correspondingregioisomeric lactones. Samples (5 lL) were injected into an XLgas chromatograph (Perkin Elmer, Norwalk, CT) tted with aCYCLOSILB 113-6632 capillary column (30 m 530 lm) (J&WScientic), and concentrations were determined using anexternal calibration curve. The GC injector temperature was setat 250 C. The GC temperature program used was as follows: theinitial oven temperature was 100 C, held for 1 min and followedby a temperature increase at 10 C min1 up to 150 C, whichwas then held for 3 min. Retention times were 3.7 min and3.95 min for the substrate (mixture of ketone isomers) and8.5 min for the product.

3. Mathematical model

3.1. Kinetic model

To develop a kinetic model for ketone bioconversion, the reac-tion mechanism suggested by Sheng et al. [12] was modied. Thereaction mechanism proposed in this paper follows the Lang-muirHinshelwoodHougenWatson formalism [20], consideringthe following assumptions:

(a) Glycerol enters in the pentose phosphate pathway and citricacid cycle, generating a sufcient concentration of NADPH2+

to regenerate and hold constant the active site of the CHMOin the reduced form, FADPH2+ [21].

(b) The substrate is not consumed as an energy and carbonsource.

R. Melgarejo-Torres et al. / Chemica(c) All the steps with the exception of product formation areconsidered reversible.(d) The amount of CHMO in lg per gram of biomass (NT) wasapproximated using the specic rate obtained previously[10] and a kcat value of 6 s1. This kcat value was also usedas an initial value during the kinetic parameter estimation.

Sheng et al. reaction mechanism considers oxidation of theCHMO active site (E) to form the complex enzyme-oxygen (EO2)at a reaction rate r1 and a stoichiometric coefcient of 3

E O2 k1

k01EO2; r1 k1NThEO2 k01NThEO2 3

where h represents the free or intermediary complex enzyme frac-tions in the reaction mechanism. EO2 interacts with the substrate(S) to form the complex enzyme-oxygen-substrate (EO2S) at a reac-tion rate r3 and a stoichiometric coefcient of 2

EO2 S k3

k03EO2S; r3 k3NThEO2 S k03NThEO2S 4

EO2S is the intermediate to form the product of interest (P) at areaction rate r5 and a stoichiometric coefcient of 1, and nally theregeneration of the active site (E) to start again the catalytic cycle.

EO2S!k5 E P; r5 k5NThEO2S 5

However, this mechanism does not consider any type of inacti-vation by oxygen excess or substrate inhibition, and only proposesthe oxidationreduction dynamic of the active site of the CHMO for

Fig. 1. Reaction mechanism proposed for ketone bioconversion consideringsubstrate inhibition and oxygen inactivation.

gineering Journal 240 (2014) 19 3Fractions of the enzyme intermediary complex are described bythe differential equations

NTdhEdt

r5 r1 8

NTdhEO2dt

r1 r2 r3 9

NTdhO2EO2

dt r2 10

NTdhEO2SSdt

r4 11

NTdhEO2Sdt

r3 r4 r5 12

And the balance of enzyme fractions is given as follows:

hE hEO2 hO2EO2 hEO2SS hEO2S 1 13

3.2. Mathematical model in the bioreactor

Mass transfer and the bioconversion mechanism were consid-ered for the development of the mathematical model in the biore-actor. The following assumptions were made:

Substrate [S], product [P] and oxygen [O2] concentrations in thereactor during the bioconversion are described by differentialequations:

dSdt

r3 r4 14

dPdt

r5 15

dO2dt

kLaO2 O2 r1 r2 16

The proposed mathematical model was solved by integrating aset of differential equations (ODEs) with the Runge Kutta Fehlbergmethod. The estimation strategy of kinetic parameters is presentedin Fig. 2, which considers oxygen mass transfer and the reactionsthat take place in the cell during the bioconversion.

The model contains nine kinetic parameters, kj, which wereestimated by the weighted least-squares of the residuals (RSS)between the calculated and experimental concentrations accord-ing to the following minimized weighted objective function (Eq.(17)):

XnexpXnresp

4 R. Melgarejo-Torres et al. / Chemical Engineering Journal 240 (2014) 19(a) The reaction system is isothermic and perfectly mixed, andthe agitated tank bioreactor was batch operated.

(b) Oxygen mass transfer (kLa) from the gas phase to aqueousphase was considered. The effective kLa that characterizesthis mechanism was determined with independent experi-ments in an abiotic system.

(c) Intraparticle and interparticle mass transfer resistanceswere considered negligible because of the cell size(6 lm) and insignicant shear stress between the celland the aqueous phase, respectively.

(d) Due to the aeration rates used, there was no loss of substrateby evaporation.

(e) There was no cell shear damage due to stirring according tostudies previously carried out on the same bioconversionsystem.Fig. 2. Estimation strategyRSSb j l

Wjlyij y^ijyil y^il !b1 ;b2 ;...;bp min 17

The responses used in the regression are the concentration ofdissolved oxygen, lactone and ketone trough time. In Eq. (17) yijdenotes the calculated value and yij denotes the observedconcentration in experiment j, bj is the kinetic parameter vector(ki) to be estimated, nexp is the number of independentexperiments, nresp is the number of the model response variablesand Wjl is the weighting factor that can be used to give greaterimportance to some portion of the response variables. The kineticparameters were estimated by a software program (ODRPACK2.01) using multi-response non-linear regression and theLevenbergMarquardt method with a 95% condence interval. Todetermine the statistical signicance of the parameters, the t-testof kinetic parameters.

l EnR. Melgarejo-Torres et al. / Chemicawas used, while the F test was used to obtain the regressionsignicance. In this sense, statistical condence in the develop-ment of the kinetic model will allow us to relate the kinetic

a

b

c

Fig. 3. Ketone bioconversion under various operating conditions of agitation and aeragineering Journal 240 (2014) 19 5parameters to the rate of formation of different intermediaries thatinduce inhibition and product formation in the catalytic cycle ofthe cell.

[P] g/L

0.45900.37900.29800.21800.13800.0571

[P] g/L

0.45900.37900.29800.21800.13800.0571

tion rates, and substrate concentrations: (a) 0.4 g L1; (b) 0.7 g L1; (c) 1.0 g L1.

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90

C

0.0

0.1

0.2

0.3 3k SEO2S

ntra

0.5 gL-

1

6

l EnTable 1Estimated kinetic parameters.

Kinetic parameters Estimated value

k1 1653.33 (Lg1 s1)k01 3.32 (s

1)k2 1.11 101 (Lg1 s1)k02 7.85 105 (s1)k3 83.43 (Lg1 s1)k03 6.35 10

1 (s1)k4 1.66 101 (Lg1 s1)k04 1.79 10

4 (s1)

k05 (kcat) 56.66 (s1)

F 2200Ftab = 3.98 for 1 a = 0.95

Table 2Afnity and inhibition constants obtained in this work and reported in the literature.

References KMO2 (lM) KMS (lM) KIO2 (lM) KIS (lM)

Trower et al. [23]a 0.5 Branchaud and Walsh [24]b 6 Sheng et al. [12]c 6.8 Torres-Pazmio et al. [16]e 10 80 Bucko et al. [9]d 1.4 20.5Shannon et al. [25]f 1530 This work 62.72 70.47 22.30 9.98

a Ciclohexanone monooxygenase from Xanterobacter sp.b,c Ciclohexanone monooxygenase from Acinetobacter NCIB9871.d Ciclohexanone monooxygenase from Escherichia coli.e Phenylacetone Monooxygenase from Thermobida fusca.f Methane monooxygenase Methylococcus capsulatus.

6 R. Melgarejo-Torres et al. / Chemica4. Results and discussion

4.1. Oxygen mass transfer studies in the absence of bioconversion

Mass transfer was studied in an abiotic system through thedetermination of oxygen mass transfer coefcients by the dynamicmethod under various operation conditions. The agitation rate hada greater effect than aeration rate on kLa. The kLa values achieved athigh aeration and agitation rates (1950 rpm and 1.4 vvm) were be-tween 220 and 290 h1, while at low aeration and agitation rates(750 rpm and 0.71 vvm), the kLa values were between 20 and32 h1. These mass transfer coefcients were used for MMBR-100bioreactor modeling under various reaction conditions.

4.2. Bioconversion of ketones in an aqueous-gas system

The bioconversion experimental results obtained through theexperimental design under various operation conditions in termsof agitation rates (750, 1350 and 1950 rpm), aeration rates (0.75,1.0 and 1.4 vvm) and substrate concentrations (0.35, 0.7 and1 g L1) are shown in Fig. 3. One hundred percent ketone biocon-version was observed at 1 vvm and 1350 rpm (kLa = 180 h1), withsubstrate concentrations less than 0.4 g L1 (Fig. 3a). However,when the aeration and agitation rates were increased to 1.4 vvmand 1950 rpm (kLa = 290 h1) at substrate concentrations less than0.4 g L1, the bioconversion decreased to 42%. Fig. 3b shows that atsubstrate concentrations greater than 0.4 g L1, and at 1 vvm and1350 rpm (kLa = 180 h1), only 20% bioconversion was obtained.Once again, when the aeration and agitation rates were increasedto 1.4 vvm and 1950 rpm (kLa = 290 h1) at substrate concentra-tions over 0.4 g L1, the bioconversion decreased to 17%. Statisticalanalysis using the Pareto box indicated that the substrate had thegreatest effect on the ketone bioconversion. The second mostimportant factor was the interaction between agitation andompl

ex fr

actio

n0.4

0.5

0.6

0.7

0.8

0.9

1.0

[EO2] [EO2S] [E]

b

k1k

k5 k1

E

k3

O2

PEO2

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90S

ubst

rate

and

pro

duct

con

ce

0.0

0.1

0.2

0.3

0.4

Dis

olve

d O

xyge

n (m

0

1

2

3

4

5Substrate ProductOxygena

tion

(gL-

1 )

0.6

0.7

)

7

8

gineering Journal 240 (2014) 19aeration rates, which relate the oxygen mass transfer from thegas phase to the aqueous phase. The decrease in the percentageof bioconversion when the substrate concentration was increasedfrom 0.4 to 1 g L1 was mainly due to substrate inhibition. Forthe same biocatalyzer, Doig et al. [22], working under varioussubstrate concentrations (0.26 g L1), observed that greaterketone bioconversion was achieved in the range of 0.20.4 g L1

in a 1.0 L ask with a volume of 20 mL at 250 rpm. Modicationand control of the kLa value (180290 h1) could increase oxygentransfer from the gas phase to the aqueous phase by maintaininga constant concentration of dissolved oxygen; however, thismechanism causes oxidation in the biocatalyzer and a decreasein bioconversion. Bennett [7] reported that oxygen excess in thereaction medium caused oxidation of two peripheral serine resi-dues to form sulfonic acid, causing a change and permanent inac-tivation of CHMO. These results were also observed by Oppermanand Reetz [11] who designed a CHMO by identifying the surfaceamino acids susceptible to oxidation and folding back them insidethe enzyme. The mutated enzyme retained 40% activity in H2O2(0.2 M), while the wild-type enzyme lost all activity in 5 mM H2O2.

4.3. Mathematical model for ketone bioconversion

The proposed kinetic model based on elementary reactionsincluding substrate inhibition and oxygen inactivation was cou-pled to the bioreactor model that accounts for an oxygen transfermechanism from the gas to the aqueous phase. The resulting

Fig. 4. Case study without substrate inhibition or oxygen inactivation. Theoperating conditions were 1350 rpm, 1 vvm (kLa = 180 h1) and 0.35 g substrateL1. (a) Substrate, product and dissolved oxygen concentrations of the experimentaldata and calculated values versus bioconversion time. (b) Reaction mechanism andestimated intermediary complex fractions during the bioconversion.

l En

fr

actio

n

0.6

0.8

1.0

[EO2] [EO2O2] [EO2S] [E]

k1E

O2

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90S

ubst

rate

and

pro

duct

con

cent

ratio

n (g

L-1 )

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Dis

olve

d O

xyge

n (m

gL-1

)

2

3

4

5

6

7

8

Substrate ProductOxygen

a

b

R. Melgarejo-Torres et al. / Chemicamodel was fed with bioconversion experimental data and oxygenmass transfer coefcients obtained from abiotic experiments in or-der to estimate the kinetic parameters.

In order to obtain a global minimum, to reduce the statisticalcorrelation between kinetic parameters to be estimated and,hence, to describe the intrinsic enzymatic behavior inside the cellunder various reaction conditions, the estimation of kinetic param-eters considered the following strategy: (a) kinetic parameter esti-mation for bioconversion without either substrate inhibition oroxygen inactivation were carried out. This allowed for obtainingreliable values of the parameters k1, k

01, k3, k

03 and k5 related to

the reaction steps where there is no substrate inhibition or oxygeninactivation; (b) the kinetic parameters k2 and k

02 were estimated

from the experimental data when oxygen inactivation was present,relating to the reaction step that considers oxygen inactivation; (c)the kinetic parameters k4 and k

04 were estimated from the experi-

mental data when substrate inhibition was present, relating tothe reaction step that causes substrate inhibition; (d) nally, theproposed kinetic model was validated by predicting observationsunder reaction conditions in the presence of substrate inhibitionand oxygen inactivation simultaneously. According to t-tests andF-tests, the regression and parameters showed statistical signi-cance within the 95% condence interval. There was no statisticalcorrelation between the parameters estimated according to thevariancecovariance matrix. The kinetic parameters estimated bythe model are shown in Table 1.

Bioconversion time (min)

Com

plex

0.0

0.2

0.4k2

k2

O2

O2EO2

3

kk5

k

k1

k3P

EO2

SEO2S

0 10 20 30 40 50 60 70 80 90

Fig. 5. Case study with oxygen inactivation. The operating conditions were1950 rpm and 1.4 vvm (kLa = 290 h1) and 0.35 g substrate L1. (a) Substrate,product and dissolved oxygen concentrations of the experimental data andcalculated values versus bioconversion time. (b) Reaction mechanism and esti-mated intermediary complex fractions during the bioconversion. frac

tion

0.6

0.8

1.0

k1kk5 k

1

E

k3

O2

PEO2

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90S

ubst

rate

and

pro

duct

con

cent

ratio

n (g

L-1 )

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Dis

olve

d O

xyge

n (m

gL-1

)

2

3

4

5

6

7

8

Substrate ProductOxygen

a

b

gineering Journal 240 (2014) 19 7Table 2 shows the afnity and inhibition constants of the sub-strate and oxygen obtained from the kinetic parameters of themodel (Table 1), expressed in lM for comparison with thosereported experimentally by other authors for monooxygenases en-zymes. Afnity and inhibition constants (Eq. (18)) were dened asthe inverse of the equilibrium constants of the individualreactions:

K k0j

kj18

where K is the constant of afnity or inhibition, and k0j and kj are thedissociation and association kinetic parameters of the enzyme-sub-strate complex.

In the case study without substrate inhibition or oxygen inacti-vation, the operating conditions were 1350 rpm and 1.0 vvm (kL-a = 180 h1), with 0.35 g substrate L1. The comparison betweenthe experimental and calculated data obtained from the t of themodel versus bioconversion time is shown in Fig. 4a. It was ob-served that the kinetic model adequately described the experimen-tal data. The substrate was completely consumed in 70 min;assuming a substrate-product yield (YS/P) of 0.871 and 3.0 g bio-mass L1, a specic rate of 0.11 g lactone g biomass1 h1 was ob-tained. This value is lower but on the same order of magnitudethan values previously reported for the same strain; Melgarejo-Torres et al. [10] observed a specic rate of 0.45 g lactone g bio-mass1 h1 at 1.0 g biomass L1 and 0.5 g ketone L1. Also, Baldwin

Bioconversion time (min)

Com

plex

0.0

0.2

0.4[EO2] [EO2SS] [EO2S] [E]

4k4 kk

EO2SS

S

3k SEO2S

0 10 20 30 40 50 60 70 80 90

Fig. 6. Case study with substrate inhibition. The operating conditions were1350 rpm and 1.0 vvm (kLa = 180 h1) and 0.70 g substrate L1. (a) Substrate,product and dissolved oxygen concentrations of the experimental data andcalculated values versus bioconversion time. (b) Reaction mechanism and esti-mated intermediary complex concentrations during the bioconversion.

substrate inhibitory complex [EO2SS] formation. It is known thata reduction in the substrate concentration can reverse inhibitionsince the enzyme remains active, as most substrates do not bindto enzymes by covalent binding [21]. Fig. 6b shows the estimatedconcentrations of the [EO2] complex, which disappeared to formthe [EO2S] complex; this was transformed into the substrate inhi-bition [EO2SS] complex, leading to reduced product [P] formation.

Finally, the model was validated for the case study with bothsubstrate inhibition and oxygen inactivation; the operating condi-tions were 1950 rpm and 1.4 vvm (kLa = 290 h1), and 0.7 g L1 ofsubstrate. The comparison between the experimental and calcu-lated data versus bioconversion time is shown in Fig. 7a. Biocon-version performance under substrate inhibition and oxygeninactivation operational conditions was calculated by the modelusing the estimated kinetic constant obtained for the other casestudies. Under these operational conditions, the percentage ofexperimental bioconversion was 17%. The low bioconversion per-centage may be explained analyzing the estimated kinetic param-eters values of complex formation. The estimated intermediarycomplexes are shown in Fig. 7b; k4 was two orders of magnitudegreater than k2, which indicates that the formation rate of the [EO2-SS] complex by substrate inhibition was higher than that of the[O2EO2] complex by oxygen inactivation. It was observed that the

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90

Com

ple

0.0

0.2

0.4[EO2] [EO2O2] [EO2SS] [EO2S] [E]

EO2SS

Fig. 7. Case study with both substrate inhibition and oxygen inactivation. Theoperating conditions were 1950 rpm and 1.4 vvm (kLa = 290 h1) and 0.70 gsubstrate L1. (a) Substrate, product and dissolved oxygen concentrations of theexperimental data and calculated values versus bioconversion time. (b) Reactionmechanism and estimated intermediary complex fractions during thebioconversion.

l Enand Woodley [6] observed a specic rate of 0.65 g lactone g bio-mass1 h1 at 2.0 g biomass L1 and 0.5 g ketone L1. The concen-tration proles of the estimated intermediates present inside thecell during the bioconversion versus bioconversion time are shownin Fig. 4b. CHMO oxidation occurs in the rst few minutes, gener-ating the [EO2] complex that is essential in the formation of the[EO2S] complex which generates the product [P]. Analyzing theafnity and kinetic parameters estimated for this case study, inthe rst reversible reaction, it was observed that k01 was smallerthan k1, favoring the reaction equilibrium toward the formationof the [EO2] complex. With these kinetic parameters, substrateafnity constant (KMO2) value of 62.72 lM was obtained (Table 2),which is on the same order of magnitude as the values reported byShannon et al. [25] and Torres-Pazmio et al. [16], i.e. 1530 and10 lM, respectively. In the second reversible reaction, k03 was smal-ler than k3, favoring the reaction equilibrium toward the formationof the [EO2S] complex. With these kinetic parameters, an substrateafnity constant (KMS) value of 70.47 lM was obtained (Table 2),which is reasonably similar to that reported by Torres-Pazmioet al. [16] (Table 2). In the last reaction, the k5 parameter was56.66 s1 (Table 1), which is the reaction rate constant (kcat) forproduct [P] formation. Kamerbeek et al. [26] reported kcat valuesof 14 and 3.7 s1 for a wild type CHMO (E. coli TOP10 pQR230)and its mutant, respectively.

In the case study where oxygen inactivation was present sinceoxygen mass transfer in the reaction medium was increased, theoperation conditions were 1950 rpm and 1.4 vvm (kLa = 290 h1),and 0.35 g substrate L1. The comparison between the experimen-tal and calculated data versus bioconversion time is shown inFig. 5a. It was observed that the model was able to predict the de-crease in lactone production due to CHMO oxidation. The relevantkinetic parameters estimated for this case study were k2 and k

02. It

was observed that k2ok02 by four orders of magnitude; this high k2value indicates that the CHMO oxidation reaction was virtuallyirreversible and coincides with that reported by Bennett [7], whomentioned that oxidation causes permanent enzyme inactivation.With these kinetic parameters, an oxygen inactivation constant(KIO2) value of 22.3 lMwas obtained (Table 2). There are no reportsof the oxygen inactivation constant in the literature. Under theseoperational conditions, the percentage of experimental bioconver-sion was 42%. The concentration proles of the estimated interme-diates present inside the cell during the bioconversion versusbioconversion time are presented in Fig. 5b. It was observed thatthe formation of the rst [EO2] complex was decreased, rapidlyforming the [O2EO2] and [EO2S] complexes, oxygen inactivationcomplex and a complex essential for the formation of the product[P], respectively. The linear trend in [O2EO2] complex formation inrelation to CHMO oxidation shows that exposure to these condi-tions will cause total enzyme inactivation.

For the case study where substrate inhibition was present dueto a greater substrate concentration in the reaction medium, theoperation conditions were 1350 rpm and 1.0 vvm (kLa = 180 h1),and 0.70 g of substrate L1. The comparison between the experi-mental and calculated data obtained from the t of the model ver-sus bioconversion time is shown in Fig. 6a. Under these operationalconditions, the experimental bioconversion was 20%. It was ob-served that the model was also able to predict the decrease in lac-tone production due to substrate inhibition. The relevant kineticparameters estimated for this case study were k4 and k

04. With

these kinetic parameters, a substrate inhibition constant (KIS) valueof 9.98 lMwas obtained (Table 2), which is half of that reported byBucko et al. [9] (Table 2). It was found that k4 was three orders ofmagnitude greater than k04; this means that for substrate concen-

1

8 R. Melgarejo-Torres et al. / Chemicatrations above 0.4 g L , the global reaction will tend to form moreof the substrate inhibitory [EO2SS] complex than the product [P]. k

04

is the kinetic dissociation constant of the reversible reaction ofx fra

ctio

n

0.6

0.8

1.0

k4

k4

S

k2

k2

O2

O2EO2

3

k1k

k5

k

k1

E

k3

O2

PEO2

SEO2S

Bioconversion time (min)0 10 20 30 40 50 60 70 80 90S

ubst

rate

and

pro

duct

con

cent

ratio

n (g

L-1 )

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Dis

olve

d O

xyge

n (m

gL-1

)

2

3

4

5

6

7

8

Substrate ProductOxygen

a

b

gineering Journal 240 (2014) 19[EO2] complex disappeared to form the inhibition [EO2SS] andinactivation [O2EO2] complexes, while the rest was transformedinto the product [P].

5. Conclusions

In this study, we developed a non-reported pseudo intrinsic ki-netic model based on elementary reactions accounting for sub-strate inhibition and oxygen inactivation for bicyclic ketonebioconversion. The kinetic model was coupled to the reactor modeltaking into account interfacial oxygen mass transport, adequatelytted observations and was validated in predictive mode describ-ing observations that were not used in the parameter estimation.The regression and these estimated parameters were statisticallysignicant, making their analysis reliable regarding the kineticbehavior of CHMO, particularly allowing identify/understand thekinetic role of each substrate, intermediate and reaction producton the CHMO enzyme. In further studies, this kinetic model willbe essential to model and scale-up a partitioning bioreactor usingthis CHMO enzyme expressed in whole cells of E. coli TOP10pQR239.

References

[1] M.J. Fink, F. Rudroff, M.D. Mihovilovic, BaeyerVilliger monooxygenases inaroma compound synthesis, Bioorg. Med. Chem. Lett. 21 (2011) 61356138.

[2] Z. Kang, X. Zhang, H. Liu, J. Qiu, K. Lun Yeung, A rapid synthesis route for Sn-beta zeolites by steam-assisted conversion and their catalytic performance inBaeyerVilliger oxidation, Chem. Eng. J. 218 (2013) 425432.

[10] R. Melgarejo-Torres, D. Torres-Martnez, M. Gutirrez-Rojas, A. Gmez deJess, G.J. Lye, S. Huerta-Ochoa, Regime analysis of a BaeyerVilligerbioconversion in a three-phase (airwaterionic liquid) stirred tankbioreactor, Biochem. Eng. J. 5859 (2011) 8795.

[11] D.J. Opperman, M.T. Reetz, Towards practical BaeyerVilliger-monooxygenases: design of cyclohexanone monooxygenase mutants withenhanced oxidative stability, Chem. Bio. Chem. 11 (18) (2010) 25892596.

[12] D. Sheng, D.P. Ballou, V. Massey, Mechanistic studies of cyclohexanonemonooxygenase: chemical properties of intermediates involved in catalysis,Biochemistry-US. 40 (2001) 1115611167.

[13] E. Malito, A. Aleri, M.W. Fraaije, A. Mattevi, Crystal structure of a BaeyerVilliger monooxygenase, Proc. Natl. Acad. Sci. U.S.A. 101 (36) (2004) 1315713162.

[14] D.E. Torres Pazmio, H.M. Dudek, M.W. Fraaije, BaeyerVilligermonooxygenases: recent advances and future challenges, Curr. Opin. Chem.Biol. 14 (2) (2010) 138144.

[15] C.M. Hogan, J.M. Woodley, Modelling of two enzyme reactions in a linkedcofactor recycle system for chiral lactone synthesis, Chem. Eng. Sci. 55 (2000)20012008.

[16] D.E. Torres Pazmio, B.J. Baas, D.B. Janssen, M.W. Fraaije, Kinetic mechanism ofphenylacetone monooxygenase from Thermobida fusca, Biochemistry-US 47(2008) 40824093.

[17] J. Marugn, R. van Grieken, A.E. Cassano, O.M. Alfano, Kinetic modelling of thephotocatalytic inactivation of bacteria, Water Sci. Technol. 61 (6) (2010) 15471553.

[18] R. Fuchs, D. Dewey, A. Humphrey, Effect of surface aeration on scale-upproducers for fermentation processes, Ind. Eng. Chem. Proc. D. D. 10 (2) (1971)19901996.

[19] G. Ascanio, B. Castro, E. Galindo, Measurement of power consumption instirred vessels a review. Trans IChemE, Part A, Chem. Eng. Res. Des. 82 (2004)

R. Melgarejo-Torres et al. / Chemical Engineering Journal 240 (2014) 19 9[3] S.D. Doig, P.J. Avenell, P.A. Bird, P. Gallati, K.S. Lander, G.J. Lye, R. Wohlgemuth,M.J. Woodley, Reactor operation and scale-up of whole cell BaeyerVilligercatalyzed lactone synthesis, Biotechnol. Prog. 18 (2002) 10391046.

[4] H. Pfruender, R. Jones, V. Weuster-Botz, Water immiscible ionic liquids assolvents for whole cell biocatalysis, J. Biotechnol. 124 (2006) 182190.

[5] S.D. Doig, L.M. OSullivan, S. Patel, J.M. Ward, J.M. Woodley, Large scaleproduction of cyclohexanone monooxygenase from Escherichia coli TOP10pQR239, Enzyme Microb. Technol. 28 (2001) 265274.

[6] C. Baldwin, M.J. Woodley, On oxygen limitation in a whole cell biocatalyticBaeyerVilliger oxidation process, Biotechnol. Bioeng. 95 (2006) 362369.

[7] A. Bennett, Mechanism of oxidative inactivation of acinetobacter sp. NCIMB9871. Cyclohexanone monooxygenase, J. Undergrad. Res. 6(1) (2004) pp. 19.

[8] K. Geitner, J. Rehdorf, R. Snajdrova, U.T. Bornscheuer, Scale-up of BaeyerVilliger monooxygenase-catalyzed synthesis of enantiopure compounds, Appl.Microbiol. Biotechnol. 88 (5) (2010) 10871093.

[9] M. Bucko, A. Schenkmayerov, P. Gemeinera, A. Vikartovsk, M. Mihovilovibc, I.Lackc, Continuous testing system for BaeyerVilliger biooxidation usingrecombinant Escherichia coli expressing cyclohexanone monooxygenaseencapsulated in polyelectrolyte complex capsules, Enzyme Microb. Technol.49 (2011) 284288.12821290.[20] N. Staelens, M.F. Reyniers, G.B. Marin, LangmuirHinshelwoodHougen

Watson rate equations for the transalkylation of methylamines, Chem. Eng.J. 90 (2002) 185193.

[21] D. Voet, J.G. Voet, Biochemistry, fourth ed., John Wiley & Sons Ltd, 2011.[22] S.D. Doig, H. Simpson, V. Alphand, R. Furstoss, J.M. Woodley, Characterization

of a recombinant Escherichia coli TOP10 [pQR239] whole-cell biocatalyst forstereoselective BaeyerVilliger oxidations, Enzyme Microb. Technol. 32 (2003)347355.

[23] M.K. Trower, M.R. Buckland, M. Grifn, Characterization of an FMN-containingcyclohexanone monooxygenase-grown Xanthobacter sp, Eur. J. Biochem. 181(1989) 199206.

[24] B.P. Branchaud, C.T. Walsh, Functional group diversity in enzymaticoxygenation reactions catalyzed by bacterial avin-containingcyclohexanone oxygenase, J. Am. Chem. Soc. 10 (1985) 21532161.

[25] S.S. Shannon, A.F. Wilson, M. Maarten, J.P. Klinman, S.P. Lippard, Oxygenkinetic isotope effects in soluble methane monooxygenase, J. Biol. Chem. 276(7) (2001) 45494553.

[26] N.M. Kamerbeek, M.W. Fraaije, D.B. Janssen, Identifying determinants ofNADPH specicity in BaeyerVilliger monooxygenases, Eur. J. Biochem. 271(2004) 21072116.

Kinetic mathematical model for ketone bioconversion using Escherichia coli TOP10 pQR2391 Introduction2 Materials and methods2.1 Microorganism and chemicals2.2 Stirred tank bioreactor description2.3 Oxygen mass transfer coefficient (kLa) determination2.4 Aerated power consumption (Pg) measurement2.5 Bioconversion experiments2.6 Analysis

3 Mathematical model3.1 Kinetic model3.2 Mathematical model in the bioreactor

4 Results and discussion4.1 Oxygen mass transfer studies in the absence of bioconversion4.2 Bioconversion of ketones in an aqueous-gas system4.3 Mathematical model for ketone bioconversion

5 ConclusionsReferences