Wo

Ca

b

a

ARR2A

KTBM

1

dsmdoced[doidg

(

0d

Enzyme and Microbial Technology 48 (2011) 480–486

Contents lists available at ScienceDirect

Enzyme and Microbial Technology

journa l homepage: www.e lsev ier .com/ locate /emt

hole-cell bio-oxidation of n-dodecane using the alkane hydroxylase systemf P. putida GPo1 expressed in E. coli

hris Granta,∗, John M. Woodleyb, Frank Baganza,∗

Department of Biochemical Engineering, University College London, Torrington Place, London WC1E 7JE, UKCenter for Process Engineering and Technology, Department of Chemical and Biochemical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark

r t i c l e i n f o

rticle history:eceived 3 September 2010eceived in revised form8 December 2010ccepted 31 January 2011

eywords:wo-liquid phaseio-oxidationono-oxygenase

a b s t r a c t

The alkane-1-monoxygenase (alkB) complex of Pseudomonas putida GPo1 has been extensively studiedin the past and shown to be capable of oxidising aliphatic C5–C12 alkanes to primary alcohols both in thewild-type organism by growth on C5–C12 alkanes as sole carbon source and in vitro. Despite this, suc-cessful n-dodecane oxidation for the production of 1-dodecanol or dodecanoic acid has proven elusive inthe past when using alkB-expressing recombinants. This article demonstrates, for the first time in vivo,by using the Escherichia coli GEC137 pGEc47�J strain, that n-dodecane oxidation using this enzyme forthe production of primary alcohols and carboxylic acids is feasible and in fact potentially more promisingthan n-octane oxidation due to lower product and substrate toxicity. Yields are reported of 1-dodecanolof up to 2 g/Lorganic and dodecanoic acid up to 19.7 g/Lorganic in a 2 L stirred tank reactor with 1 L aque-ous phase and 200 mL of n-dodecane as a second phase. The maximum volumetric rate of combinedalcohol and acid production achieved was 1.9 g/Lorganic/h (0.35 g/Ltotal/h). The maximum specific activ-

ity of combined alcohol and acid production was 7-fold lower on n-dodecane (3.5 �mol/min/gdcw) thanon n-octane (21 �mol/min/gdcw); similar to the 5-fold difference observed between wild-type growthrates using the two respective alkanes as sole carbon source. Despite this, both total volumetric rate andfinal yield exceeded n-octane oxidation by 3.5-fold under the same conditions, due to the lower toxic-ity of n-dodecane and its oxidation products to E. coli compared to the 8-carbon equivalents. Substrate

e ovebe ad

access limitations and thimportant limitations to

. Introduction

The alkane hydroxylase (Alk) system that is native to Pseu-omonas putida GPo1 [1] is of considerable interest for the highlypecific oxidation of medium-long chain aliphatic alkanes to pri-ary alcohols and carboxylic acids [2]. This reaction remains

ifficult to perform by conventional chemistry, and biocatalysisffers an attractive near-ambient temperature alternative. The bio-atalytic route offers potentially higher selectivity with reducednergy demand and clear safety benefits compared to chemical oxi-ation at elevated temperature and pressure. Both Escherichia coli3,4] and P. putida [5] hosts containing the necessary alk genes haveemonstrated production of 1-octanol and octanoic acid, yet little

r no activity with n-dodecane has been reported in vivo [5,6]. Thiss surprising since the wild-type P. putida GPo1 is able to grow on n-odecane as the sole carbon source at 20% of the maximum specificrowth rate on n-octane [7]. Furthermore, the isolated enzyme has

∗ Corresponding authors. Tel.: +44 20 7679 2968; fax: +44 20 7216 3943.E-mail addresses: christopher.grant@ucl.ac.uk (C. Grant), f.banganz@ucl.ac.uk

F. Baganz).

141-0229/$ – see front matter © 2011 Elsevier Inc. All rights reserved.oi:10.1016/j.enzmictec.2011.01.008

roxidation of 1-dodecanol to dodecanoic acid were identified as the mostdressed.

© 2011 Elsevier Inc. All rights reserved.

shown between 50 and 100% activity on n-dodecane compared ton-octane [6,8] and thus 1-dodecanol production by this means maybe able to match or even exceed the economic potential indicatedfor 1-octanol production [9]. There is a growing market for fattyalcohols for use in detergents, personal care products and industrialapplications; a market currently estimated to be worth D1billion inEurope alone [10].

Alkane-1-monooxygenase (alkB) is the catalytic component ofthe alkane hydroxylase complex which performs the first reactionof the alkane metabolism pathway encoded on the OCT plasmid ofP. putida GPo1 (Fig. 1). The complex consists of three components:an integral membrane protein (alkB) and two electron transfer pro-teins named rubredoxin (alkG) and rubredoxin reductase (alkT). Byknocking-out the alcohol dehydrogenase (alkJ) from the alk operonand expressing in a host with no endogenous alcohol dehydro-genase activity, the selective accumulation of primary alcohols ispossible and has been previously demonstrated [5,11]. Two-liquid

phase fermentation has been the preferred choice of operation andthere are several reasons why this is the case. A whole-cell formathas been chosen because the complexity of the three-component,cofactor dependent, integral membrane complex favours the choiceof a whole-cell biocatalyst. A Two-Liquid-phase system because,

C. Grant et al. / Enzyme and Microbial Technology 48 (2011) 480–486 481

F ompleT e alco

fiasiae

claeto

wmbmid

2

Gso

wNy(md1

(4(

ad2o9c1a5

2l

2

1awr

ai

ig. 1. Alkane oxidation in P.oleovorans; the alkane hydroxylase (AlkB, AlkG, AlkT) che pGEc47�J plasmid contains the complete alk pathway with the exception of th

rstly, the poor solubility of both substrate and product in thequeous phase [2] can be enhanced by two-liquid phase operation;econdly, the convenient extraction of the toxic primary alcoholnto an auxiliary phase is possible. Fermentation because substratend product toxicity has been shown to have a reduced effect onxponentially growing cells compared to resting cells [12].

The complexity of the two-liquid phase whole cell bio-oxidationan make it difficult to isolate the cause of any productivityimitations (Fig. 1). With the n-octane substrate, the scientific liter-ture reports that 1-octanol toxicity [13], inactivation of Alk genexpression [14] and liquid–liquid mass transfer limitations dueo bio-emulsifier accumulation [15] to be possible causes of thebserved limited product concentration.

The aim of the work reported here was to establishhether n-dodecane oxidation was possible with the alkane-1-onooxygenase system in vivo and, if so, to evaluate the current

ottlenecks for n-dodecane oxidation in a two-liquid phase fer-entation with a view to assessing methods to overcome the

dentified bottlenecks and assess future prospects of industrial 1-odecanol production using this system.

. Materials and methods

Strains and plasmids: All experiments described were carried out with the E. coliEC137 pGEc47�J strain kindly donated by Bernard Witholt at ETH, Zurich [11]. Thetrain contains all components of the OCT plasmid with the exception of the knockedut alcohol dehydrogenase alkJ (see Fig. 1).

Medium composition: The composition of the aqueous phase in all experimentsas as described by Wubbolts et al. [4]. KH2PO4, 4 g/L; K2HPO4 (3H2O), 15.9 g/L;a2HPO4 (12H2O), 7 g/L; (NH4)2SO4, 1.2 g/L; NH4Cl, 0.2 g/L (all from Sigma–Aldrich);east extract (Oxoid), 5 g/L; l-leucine, 0.6 g/L; l-proline, 0.6 g/L; thiamine, 5 mg/LAll from Alfa Aesar). After autoclaving, MgSO4 (7H2O), 1 g/L (BDH); 1 mL of trace

inerals (composition below); 1 mL of 4% (w/v) CaCl2 (2H2O) (Alfa Aesar) and 10 g/L-glucose (Sigma–Aldrich) were added having all been heat sterilised separately.mL of 10 mg/mL filter sterilised tetracycline (Sigma–Aldrich) was also added.

The solution of trace minerals contained per liter of 5 M HCl: FeSO4 (7H2O)Sigma–Aldrich), 40 g; MnSO4 (H2O) (Sigma–Aldrich), 10 g; CoCl2 (6H2O) (Fluka),.75 g; ZnSO4 (7H2O) (VWR), 2 g; MoO4Na2 (2H2O) (Sigma–Aldrich), 2 g; CuCl2,2H2O) (Riedel-de Haen), 1 g; H3BO3 (Sigma–Aldrich), 0.50 g.

Growth and oxidative bioconversion: An overnight seed culture was grown ton optical density (OD600 nm) of 2 ± 0.2 in a shake flask containing the mediumescribed with 10 mg/L tetracycline. A 50 mL seed culture was used to inoculate aL Adaptive Biosystems bioreactor containing 950 mL aqueous medium and 200 mLf organic phase (16.67%, v/v phase ratio) containing either n-octane (Alfa Aesar,9% purity), n-dodecane (Alfa Aesar, 99% purity), n-dodecane with 0.25%, v/v dicy-lopropylketone (DCPK) (Merck), n-dodecane with 0.25%, v/v DCPK and 0.1%, v/v-dodecanol (Procter and Gamble). The media and feeds used are as described withstarting concentration of glucose of 10 g/L, pH set point of 7 and DOT set to above0% with a minimum stirrer speed of 1000 rpm and aeration rate of 0.5 L/min.

Shake flask experiments were carried out in 1 L baffled shake flasks agitated at50 rpm with 25 mm throw diameter at 37 ◦C. 95 mL of aqueous phase was inocu-

ated with 5 mL seed culture with 20 mL of organic phase where appropriate.

.1. Analysis

Cell density: Cell density measurements were made in triplicate by centrifuging–2 mL two-liquid phase samples at 13,000 rpm (19,000 × g) for 15 min, marking the

queous volume on the side of the graduated Eppendorf tube, washing the pelletsith Tris–HCl pH 7.4 and drying the tube in an 80 ◦C oven until a constant mass was

eached (24–96 h).Gas chromatography: After centrifugation for 15 min at 13,000 rpm (19,000 g) in

microfuge, the organic phase was removed, diluted to a concentration of 20–50%n ethyl acetate (Alfa Aesar, 99.9% purity) and analysed by FID-GC using a BPX5

x performs the first reaction of the pathway (modified from Van Beilen 1994 [16]).hol dehydrogenase alkJ.

(30 m long; 0.53 mm internal diameter, 1 �m film) capillary column with heliumas carrier gas flowing under a pressure of 4 PSI. For 1-dodecanol and dodecanoicacid determination, the samples were eluted at an initial temperature of 150 ◦C for2 min, followed by a linear increase of 10 ◦C per minute to reach a final temperatureof 240 ◦C. Injector and detector temperatures were both 280 ◦C. The concentrationswere determined by cross-referencing to a set of n-dodecane, dodecanoic acid and1-dodecanol standards analysed in the same run. For 1-octanol and octanoic aciddetermination the samples were eluted at 70 ◦C followed by a linear increase of5 ◦C min−1 to a final temperature of 145 ◦C.

Aqueous phase concentrations were measured by removing the organic andinterface phases after centrifugation and incubating 500 �L of aqueous sample with500 �L of ethyl acetate in a thermomixer for 15 min at 50 ◦C and 1000 rpm, thenanalysing the organic phase directly by GC-FID as described above.

SDS–PAGE AlkB expression determination: Lysate samples were prepared byresuspending cell pellets in distilled water and sonicating for 10 × 10 s pulses with10 s between each pulse. The Bradford assay [17] was then used to determine proteinconcentration before loading standardised protein concentrations of 5 �g per lane.SDS–PAGE [18] was then carried out according to the Invitrogen Nupage method[19]. Enriched membrane fractions were generated by performing centrifugationat 13,000 rpm (19,000 g) in a microfuge for 10 min and resuspending the cell pelletin 100 �L of Laemmli buffer and 100 �L of deionised H2O. Densitometry was usedto compare band intensity as a % of the whole lane after background intensity wassubtracted using the rolling disk method.

Cell viability/plasmid stability: Cell viability was determined by performing serialdilutions of a fresh fermentation sample (within 15 min) in Dulbecco’s PhosphateBuffered Saline to 10−8 and plating dilutions 10−3–10−8 on to LB-AGAR plates intriplicate on tetracycline+ and tetracycline- plates. These were incubated overnightat 37 ◦C and the colonies counted at the appropriate dilution where colonies numberbetween 30 and 300. Plasmid stability was estimated by dividing the tetracycline+viability by the tetracycline− viability for a given sample and therefore calculat-ing the proportion of cells able to grow in the presence of the antibiotic resistanceconferred by the plasmid.

3. Results

3.1. Characterisation of octane and dodecane two-liquid phasebio-oxidations in a stirred tank reactor

The initial n-dodecane oxidation with the E. coli GEC137pGEC47�J strain in a stirred-tank fed-batch fermentation resultedin successful production of 1-dodecanol, with an overall yieldof 0.28 g/Ltotal (Fig. 2a) after 48 h. The equivalent n-octane bio-oxidation resulted in 0.68 g/Ltotal and 0.52 g/Ltotal of octanoic acidafter only 24 h. The maximum specific rate of oxidation of n-dodecane was 3% of the maximum specific rate of n-octaneoxidation under comparable conditions. The low specific activitywas partially offset by a cell density 5-fold higher (Fig. 2b) thanthat achieved in the presence of n-octane; thus, the resulting max-imum volumetric rate of 1-dodecanol production reached 25% ofthe rate of 1-octanol production. Moreover, the maximum rate offormation of 1-dodecanol was delayed substantially compared to1-octanol production and occurred between 24 and 30 h rather than4–8 h for n-octane (Table 2).

3.2. Induction of alkS expression system and overoxidation

Expression of the alk genes involved in alkane degradation isunder the control of the transcriptional activator alkS [20], whichis induced by the presence of alkanes. SDS–PAGE was used to deter-mine if the low specific activity was due to poor expression of

482 C. Grant et al. / Enzyme and Microbial Technology 48 (2011) 480–486

Table 1Table showing the effect of the addition of the gratuitous inducer DCPK on cell density, average 1-dodecanol production and dodecanoic acid production in parallel flaskcultures grown from the same inoculum in the presence and absence of DCPK (n = 2).

Time (h) No DCPK addition DCPK addition

DCW (g/L) 1-Dodecanol (g/Lapolar) Dodecanoic acid (g/Lapolar) DCW (g/L) 1-Dodecanol (g/Lapolar) Dodecanoic acid (g/Lapolar)

tcpsnlEawa

F(ni(atwc

6 0.35 0.03 024 2 0.03 048 1.85 0.04 0

he alkane-1-monooxygenase, by either inherently poor inductionapacity by n-dodecane or poor uptake of n-dodecane into the cyto-lasm to induce. Four parallel shake-flask cultures grown from theame inoculum were grown and induced with one of n-dodecane,-octane, the gratuitous inducer of the alkS system dicyclopropy-

ketone (DCPK) [11] and compared to a non-induced control of the

. coli pGEC47�J. The membrane fractions from these flasks at 24nd 48 h were isolated and relative expression of alkB comparedith reference to a further control of E. coli DH1 with no plasmid

s described in the material and methods section.

ig. 2. Comparison of (A) product concentrations and (B) dry cell weight vs time ofi) n-octane bio-oxidation to 1-octanol ( ) and octanoic acid ( ), (ii)-dodecane bio-oxidation to 1-dodecanol ( ) (no DCPK induction), (iii) DCPK-

nduced n-dodecane bio-oxidation to 1-dodecanol ( ) and dodecanoic acid). Each condition shows the raw data from 2 biological replicates and the

verage line from stirred tank two-phase fermentative bioconversions in a 2 L stirredank reactor containing 1 L aqueous phase and 0.2 L organic phase. The experimentsere performed with different inoculum cultures grown from the same working

ell bank under identical conditions.

0.6 0 0.11.97 0.034 1.181.95 0.041 1.47

The results show (Fig. 3) that when DCPK and n-octane wereused to induce the cultures an additional band matching the sizeof alkB (41 kDa) was visible compared to the DH1 membrane frac-tion used as a control; this band showed a relative intensity of 3.1%with DCPK as inducer and 3.8% with n-octane. The alkB band inn-dodecane induced cultures was very faint (0.6%) and was com-parable to the band intensity of a non-induced control (0.4%). Theevidence that this band is alkB is strong. Firstly, 1-alkanols weredetected in the corresponding samples which contained the 41 kDaband. Secondly, the pre-induction samples lacking this band indi-cate the protein is under the control of alkS. Thirdly, its presencein the membrane fraction also indicates the band is a membraneprotein.

The SDS–PAGE results indicated that it was likely that n-dodecane oxidation was being limited by expression of the alkgenes. This is consistent with previous work showing induction

of alkS is not as effective with n-dodecane [21] but also can beimproved by pseudosolubilisation with biosurfactants [22]; there-fore providing a possible explanation of why induction will beimproved as the fermentation progresses and as emulsifying mate-rial is released.

Fig. 3. Figure showing SDS–PAGE of membrane fractions following sonication of E.coli GEC137 (DH1 fadR) pGEc47�J in the presence of flasks grown in the presence ofn-octane, n-dodecane, the gratuitous inducer DCPK and no inducer. A control laneshowing the membrane fraction of a lysate of E. coli DH1 containing no plasmid isalso shown. 5 �g of protein was loaded per lane. The arrow indicates the position ofthe 41 kDa alkB band.

C. Grant et al. / Enzyme and Microbial Technology 48 (2011) 480–486 483

Table 2Summary of the important metrics from the two-phase bio-oxidations described in this article (PCL = maximum product concentration; max spec rate = maximum spe-cific activity; T = time in fermentation hours; vol rate = volumetric rate; biocat efficiency = biocatalyst efficiency). The data was obtained from two-phase fermentativebioconversions described in Figs. 2, 4 and 5. Each metric indicates the average of 2 biological replicates.

n-Octane n-Dodecane n-Dodecane + DCPK n-Dodecane (0.1%, v/v1-dodecanol) + DCPK

Dry cell weight (g/Ltotal) 2.4 10 10.7 8Alcohol/acid ratio at PCLalcohol 1.8 N/A 0.1 0.08

Alcohol PCLtotal (g/L) 0.68 0.28 0.33 0.23PCLorg (g/L) 3.7 1.7 2 1.2PCLaqueous (g/L) 0.05 0.003 0.0038 0.001Max specific rate (�mol/min/g) 15 0.5 0.58 0.46Tmax spec rate (h) 4–8 24–30 12–23 22–25Max vol rate (g/L/h) 0.1 0.024 0.023 0.032Time PCL (h) 11–24 29 48 25–28Biocat efficiency at Tpcl (gproduct/gdcw) 0.55 0.03 0.03 0.016

Acid PCLtotal (g/L) 0.52 N/A 3.4 3.25PCLorg (g/L) 0.15 N/A 19.7 19.1PCLaqueous (g/L) 0.59 N/A 0.1 0.084Max spec rate (�mol/min/g) 11.2 N/A 3.2 3.1Tmax spec rate (h) 8–10 N/A 31 20Max vol rate (g/L/h) 0.09 N/A 0.33 0.24

dvecHdatadarwflt(

3

nw1ccectrtb[yatr(

ttT

Fig. 5 shows the change in biocatalyst efficiency against timeof the three bioconversions, in the presence of (i) n-octane, (ii)n-dodecane and (iii) n-dodecane with 0.1% 1-dodecanol. The fig-ure shows the exponential decrease in g1-dodecanol/gdcw in thebioconversion where 1-dodecanol is added at inoculation. Simul-

Time PCL (h) 28Biocat efficiency at Tpcl (gproduct/gdcw) 0.51

Aldehyde PCLtotal (g/L) 0.08

Following this discovery, the stirred tank bioconversion of n-odecane was repeated using DCPK at a concentration of 0.05%,/vaqueous to ensure effective expression of the alk genes. Unexpect-dly, this still resulted in a similar 1-dodecanol yield of 2 g/Lorganicompared to bio-oxidation using only n-dodecane as inducer.owever, interestingly, despite the absence of the alcohol dehy-grogenase alkJ, a high concentration of dodecanoic acid was nowlso detected to a concentration of 16.9 g/Lorganic. The repeat ofhis fermentation resulted in similar rates of product formationnd overall yields of 1.4 g/Lorganic and 19 g/Lorganic 1-dodecanol andodecanoic acid respectively (Fig. 2a). This represented an aver-ge acid:alcohol ratio of 11:1. For the n-octane bioconversion theatio was 0.75:1, indicating that overoxidation is more prevalentith n-dodecane compared to n-octane. Further parallel shakeask experiments in the absence and presence of DCPK confirmedhat this phenomenon was directly associated with DCPK additionTable 1).

.3. The effect of adding 1-dodecanol at inoculation

Since the previous DCPK+ and DCPK− fermentations with-dodecane resulted in similar 1-dodecanol yields, to confirmhether the dodecanoic acid produced was by consumption of

-dodecanol rather than direct conversion of the dodecane to dode-anoic acid, the bioconversion was repeated using an organic phaseontaining 0.1%, v/vorganic 1-dodecanol present at inoculation. Thisxperiment showed that dodecanoic acid formation was indeedoupled to consumption of 1-dodecanol (Fig. 4) in other wordshe dodecanoic acid is being formed at the expense of 1-dodecanolather than directly from n-dodecane via a mechanism similar tohat observed in P450ALK of C. tropicalis which produces dicar-oxylic acids directly from alkane substrates in a single enzyme step23]. This was an important finding as it indicates that 1-dodecanolield is severely limited by overoxidation. This also implies thatny dodecanoic acid formed is as a result of the initial n-dodecaneo 1-dodecanol conversion and hence dodecanoic acid generationates can be used as an indicator of the upstream alkB conversion

n-dodecane → 1-dodecanol → 1-dodecanal).

The presence of 1-dodecanol at inoculation, even to a concentra-ion of just 0.1% of n-dodecane, caused dodecanoic acid formationo occur at a higher initial volumetric and specific rate (Figs. 4 and 5,able 2) and for the maximum rate of 3 �mol/min/gdcw to be

N/A 48 31N/A 0.35 0.410.1 0.04 0.06

reached 15 h earlier. This indicates that either (i) the 1-dodecanolis a preferred substrate, i.e., with a lower km than n-dodecane,(ii) that substrate access of the enzyme to 1-dodecanol is betterthan n-dodecane (due to either aqueous solubility, preferred pas-sage across the membrane or permeabilising the membrane) or (iii)the 1-dodecanol has a surfactant/permeabilising effect improvingdelivery of n-dodecane to the cytoplasmic membrane enzyme com-plex. The more likely explanation seems to be substrate access sincethe observed delay in reaching maximum specific activity in thedodecane fermentations compared to the n-octane fermentationhas been reduced by the presence of 1-dodecanol at inoculation.Once the initial 1-dodecanol has been reduced to a concentration of0.03%, v/vorganic the specific activity equals the rate of the previousDCPK-induced fermentation.

Fig. 4. Comparison of product concentrations vs time of (i) DCPK-inducedn-dodecane bio-oxidation to 1-dodecanol ( ) and dodecanoic acid( ) and (ii) 1-dodecanol-spiked bio-oxidation to 1-dodecanol () and dodecanoic acid ( ). Each condition shows the raw data from 2biological replicates and the average trendline.

484 C. Grant et al. / Enzyme and Microbial

Fig. 5. Comparison of biocatalyst efficiency vs time of (i) n-octane bio-oxidation to1-octanol ( ) and octanoic acid ( ), (ii) DCPK-induced n-dodecaneb1ai

tigstIgnta8l

4

stwatt

4

tadsha(p(sooin

io-oxidation to 1-dodecanol ( ) and dodecanoic acid ( ) and (iii)-dodecanol-spiked bio-oxidation to 1-dodecanol ( ) and dodecanoiccid ( ). Results from the two-phase fermentative bioconversinos shownn Figs. 2 and 3 (n = 2).

aneously, the addition of the 1-dodecanol at inoculation resultsn a higher initial rate of increase in biocatalyst efficiency ofdodecanoic acid/gdcw compared to n-dodecane alone. After 8 h, thepecific rate of formation of acid starts to decrease to a rate similaro the slope of acid formation in the n-dodecane bioconversion.t can also be seen from this figure that the rate of increase in1-dodecanol/gdcw is negligible compared to acid formation in the-dodecane bioconversion until after 20 h. This contrasts substan-ially with the n-octane fermentative bioconversion. With n-octanes the substrate, a rapid linear increase in g1-octanol/gdcw from 3 h toh is observed. After this, from 6 to 9 h a linear increase of equiva-

ent rate is observed in goctanoic acid/gdcw.

. Discussion

Using the E. coli GEC137 strain pGEc47�J, we have observedpecific n-dodecane oxidation rates of 14–16% of n-octane oxida-ion; indicating that the same bottleneck is in effect as with theild-type strain. Mutagenesis studies have indicated this to be the

ctivity of the enzyme itself [7] although our own studies indicatehat, for the less soluble n-dodecane, substrate access is likely to behe major bottleneck.

.1. Carboxylic acid formation

It has previously been reported that recombinant E. coli con-aining pGEC47�J do not require alkJ for growth on alkanes [24]nd therefore the presence of the carboxylic acid could be expectedespite the absence of the alcohol dehydrogenase (alkJ). There wereeveral possible causes of acid formation: (i) an endogenous alco-ol dehydrogenase present in E. coli, (ii) direct overoxidation by thelkB, (iii) utilisation of the primary alkanol as a substrate by alkH,iv) oxidation by another component under the control of the alkromoter. As the overoxidation seems to be linked to DCPK additionTable 1), the presence of an endogenous alcohol dehydrogenase

eems an unlikely explanation, especially as literature has previ-usly shown this strain to be incapable of growth on n-octane or 1-ctanol [11]. This was further shown to be an unlikely option exper-mentally after a control fermentation containing 1-dodecanol buto DCPK inducer resulted in no detectable dodecanoic acid produc-

Technology 48 (2011) 480–486

tion. Literature reports indicate that the purified alkB is known toconvert alkanes to both the primary alkanol and the aldehyde in thecase of C8 alkanes [8,25]. It therefore seems likely that 1-dodecanolis either overoxidised directly to the aldehyde and carboxylic acidby alkB or overoxidised to the aldehyde directly by alkB and thenthe aldehyde is rapidly converted to the acid by alkH. Either of thesescenarios would therefore minimise any accumulation of aldehydeand reverse reaction back to the alkanol.

Alkanals were detected in both n-octane and n-dodecane fer-mentations but in both cases to lower levels than either alkanolsor carboxylic acids (Table 2). In the n-dodecane fermentation with-out DCPK the intermediate aldehyde was also detected. It seemsthat in the absence of DCPK alkH was not expressed as effectivelyas alkB so either no carboxylic acid was formed or any acid pro-duced may have been metabolised without detection. Thus withlow alkH levels the consumption of carboxylic acids may be greaterthan production and hence no carboxylic acid is detected. Anotherexplanation could be that when dodecane is the sole inducerthe quantity of enzyme expressed is sufficiently low that the n-dodecane substrate remains in excess leaving less free enzymeavailable for any competing 1-dodecanol produced to preferentiallybind. Conversely, when DCPK is used for induction, the quantity ofn-dodecane remains similar but the quantity of enzyme is now inexcess of the substrate and therefore any 1-dodecanol producedwill have a higher probability of interacting with alkB and becomingoveroxidised before it leaves the cell.

4.2. Cause of the product plateau

Formation of 1-dodecanol and dodecanoic acid both plateau at48 h, indicating that the same limitation is in effect for both prod-ucts. The cause of the plateau remains unclear but we have evidencethat cell viability, plasmid stability and alkB expression remainhigh after the plateau. For one of the n-dodecane fermentations,as long as glucose is fed cell viability remained equivalent before(2.9 × 1011 cfu/L) and 14 h after (1.3 × 1011 cfu/L) the plateau isreached in the presence of n-dodecane. Plasmid stability remainedabove 90% and alkB expression is observed up to 72 h, 16 h after theplateau was reached in that case.

There are several other possible causes. The most likely beingeither (i) the product concentration is such that the rate of pro-duction and the metabolism of the product are equal, (ii) toxicityor product inhibition of 1-dodecanol or dodecanoic acid or (iii)mass transfer rate limitations due to the presence of high levelsof bio-emulsifier released due to cell lysis causing the formationof stable emulsions [15]. Although the presence of bio-emulsifiercauses the organic phase to become gel-like (as the fermentationreached 56 h); which resulted in a lower aqueous concentration ofn-dodecane at this point. This is not the primary cause of the plateausince the maximum specific activity and volumetric rate droppedabruptly over 10 h prior to this.

One interesting observation is that the product plateau in eachcase also coincided with the maximum cell density being reached(Fig. 2) which indicates that conversion is directly associated withgrowth; though it is unclear whether the termination of cell growthis causative or consequential with regard to the product concentra-tion limit.

It seems that the most likely cause of the plateau is producttoxicity. Product/organic phase inhibition has been reported pre-viously [3,13,26] and both literature and our own work show thatby prolonging the exponential growth phase, cell lysis caused by

organic phase toxicity can be delayed. It is well known that organiccompounds can have a detrimental effect on bacteria, and alcoholsin particular are known to cause membranes to leak ions [27]. Itis generally accepted that this is caused by accumulation of theorganic compound in the membrane bilayer [28–30]. This has been

robial

scrotwottp

ruae

4i

tbhfaaiplPa

4

pfaalsdf

atsdatocib

omboloct

4d

C. Grant et al. / Enzyme and Mic

hown to occur in the presence of n-octane and 1-octanol causinghanges in membrane physiology [3,31] changes in morphology,educed viability and hole formation in the membrane having beenbserved in both P. putida and E. coli [13]. It is known to occur par-icularly in the log Poct/water range of 1–4 [32] and so 1-dodecanol,ith a log Poct/water of 5.1 and dodecanoic acid, with a log Poct/water

f 4.2 are outside of this range and, as confirmed here experimen-ally, may be expected to be able to reach higher concentrationshan the equivalent C8 oxidation products before toxicity becomesroblematic.

There is also evidence that product consumption could beesponsible: metabolism of the carboxylic acid has been shown bys to occur by consumption of dodecanoic acid when it is addedt inoculation in shake flasks. This is not surprising since the alkKnzyme is still present in this strain (Figs. 1 and 6).

.3. Why has in vivo n-dodecane oxidation not previously beendentified?

The lack of in vivo oxidation of n-dodecane reported in litera-ure by the P. putida Ps81 pGEc41 strain [5,6], even when inducedy DCPK, could be due to intrinsic host factors with the P. putidaost, the lack of the outer membrane protein alkL which is still

ound on the pGEc47�J plasmid or mass transfer limitations in thessays previously used. The function of alkL has not been reporteds confirmed in literature but it has been hypothesised that it isnvolved in substrate uptake [33]. Other possibilities are that the P.utida membrane may be a more effective barrier to transport ofong-chain alkanes than the equivalent E. coli membrane or that the. putida host has a greater tendency to metabolise these long-chainlkanes rendering the oxidation products undetectable.

.4. Substrate access limitations and overoxidation

An interesting finding is that dodecanoic acid is the favouredroduct of the n-dodecane oxidation whereas this is not the caseor n-octane oxidation. A 1-dodecanol concentration of 2 g/Lorganicnd dodecanoic acid concentration of 19.7 g/Lorganic was achievedfter 48 h compared to 3.7 g/Lorganic 1-octanol after only 24 h and aiterature maximum of 5 g/Lorganic 1-octanol using a P. putida hosttrain containing the pGEc41 plasmid which also lacked the alcoholehydrogenase alkJ [5]. Octanoic acid was observed in the n-octaneermentation to 0.52 g/Ltotal compared with 0.68 g/Ltotal 1-octanol.

It may be that overoxidation is exacerbated by limited substrateccess to the enzyme complex as very low n-dodecane concentra-ions in the cytoplasm are outcompeted for position in the activeite by any 1-dodecanol produced; therefore encouraging overoxi-ation to the aldehyde. The equilibrium driving the reaction towardldehyde production is maintained by consumption of the aldehydeo the carboxylic acid. If substrate access is an important factor inveroxidation, the mechanism of the n-dodecane cytoplasmic con-entration limit could either be due to low solubility of n-dodecanen the aqueous phase or exclusion from passage across the mem-rane due to a higher hydrophobicity than the membrane.

The solubility of n-dodecane in water is 500 fold less than n-ctane and 1000 fold less than 1-dodecanol [34]. In the case of theore water-soluble n-octane, overoxidation is less of a problem

ecause there is sufficient octane in the vicinity of the enzyme tout compete the 1-octanol produced and limit overoxidation untilater when the 1-octanol concentration increases. This critical 1-ctanol concentration is reached when there is enough present to

ompete with the n-octane present and become overoxidised tohe aldehyde.

The literature water solubility of 1-dodecanol is reported as× 10−3 g/Lwater compared to 3.7 × 10−6 g/Lwater of n-dodecane inistilled water at 25 ◦C [34]. Hence, the relative concentration in the

Technology 48 (2011) 480–486 485

aqueous phase of n-dodecane and 1-dodecanol may be expectedto be equivalent due to the 1000 fold difference in concentrationequalling the approximate 1000 fold difference in water solubility.Substrate uptake may therefore be more of a crucial issue than sub-strate solubility. However, the comparative solubilities are likely tobe different in the fermentation broth containing cells compared tothe model water data.

4.5. Specific activity

The delay in reaching maximum specific activity in the case ofn-dodecane (Table 2) was thought to be caused by poor mass trans-port of n-dodecane from the organic phase into the cell: as thefermentation progresses the limitation becoming partially offsetas surface-active material such as membrane components and the1-dodecanol product itself are released and facilitate liquid–liquidmass transport. Such a limitation could either be associated witha delay in induction of the alk genes due to poor transport ofn-dodecane into the cell or a more direct mass transfer limita-tion reducing dodecane availability to the enzyme. Previously,bio-emulsifier release at low concentrations has been shown tofacilitate mass-transport of decane into the aqueous phase withP. putida GPo1 [15], although the relationship proved complicatedsince in that particular study higher amounts of bioemulsifier ledto a return to a reduced mass transfer rate as the resulting emulsionbecomes stabilized.

The maximum specific rate of n-dodecane oxidation reached3–3.5 �mol/min/gdcw, which is 6–7 fold lower than the maximumspecific rate on n-octane (21 �mol/min/g) using the same strainunder the same conditions (Table 2). The difference in maximumspecific growth rate of the wild-type P. putida GP01 on the tworespective alkanes as sole carbon source is 5 fold [7]. The samestudy showed that mutagenesis increased the growth rate on n-dodecane which indicated that the rate of the enzyme itself was thelimiting factor in this conversion. This could also be the case herewith the E. coli GEC137 pGEc47�J strain: although the maximumspecific activity was reached earlier when 1-dodecanol was addedat inoculation the maxima are approximately equivalent. Interest-ingly, the solubility of n-octane in water is approximately 5–7 foldhigher than 1-dodecanol solubility [34] so it would be interesting tosee if further investigation reveals a correlation between solubilityand activity in vivo. Two in vitro studies reported previously [6,8]have already indicated that the purified alkB is capable of convert-ing n-dodecane at a rate 50–100% of n-octane oxidation rate andso improving substrate uptake could overcome the rate differenceobserved in wild-type substrate preference.

5. Conclusions

This study demonstrates that despite current specific rate limi-tations with n-dodecane compared to n-octane, the lower toxicityof n-dodecane to the host strain enables higher cell densities tobe achieved which translated to a maximum volumetric rate of0.35 g/L/h; exceeding the volumetric rate achieved with n-octaneas the substrate. Moreover, lower toxicity of the product coupledwith lower water solubility enabled a final yield of dodecanoicacid over 4 fold higher than 1-octanol yield. It seems that the cru-cial issues at present are associated with availability of the highlyhydrophobic n-dodecane substrate to the enzyme. Solubility and/orsubstrate uptake seems to be limiting the maximum rate beingreached and that this problem is also likely to be contributing to the

overoxidation of the primary alkanol to the carboxylic acid due toavailability of the more readily available alkanol. Work to overcomethe overoxidation to dodecanoic and the mass transfer limitation isongoing and if this can be successfully overcome then commercialn-dodecane bio-oxidation remains a possibility.

4 robial

A

oEP

R

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

86 C. Grant et al. / Enzyme and Mic

cknowledgements

The authors would like to acknowledge Professor Bernard With-lt (ETH, Zürich) for providing the strain used in this study and thengineering and Physical Sciences Research Council (EPSRC) androcter and Gamble Inc. for funding.

eferences

[1] Baptist JN, Gholson RK, Coon MJ. Hydrocarbon oxidation by a bacterial enzymesystem. I. Products of octane oxidation. Biochim Biophys Acta 1963;69:40–7.

[2] Witholt B. Bioconversions of aliphatic compounds by Pseudomonas oleovo-rans in multiphase bioreactors: background and economic potential. TrendsBiotechnol 1990;8(2):46–52.

[3] Favre-Bulle O. Bioconversion of n-octane to octanoic acid by a recombinantEscherichia coli cultured in a two-liquid phase bioreactor. Biotechnology (N Y)1991;9(4):367–71.

[4] Wubbolts M, Favre-Bulle O, Witholt B. Biosynthesis of synthons in two-liquid-phase media. Biotechnol Bioeng 1996;52:301–8.

[5] Bosetti A. Production of primary aliphatic alcohols with a recombinant Pseu-domonas strain, encoding the alkane hydroxylase enzyme system. EnzymeMicrob Technol 1992.

[6] van Beilen JB, Kingma J, Witholt B. Substrate specificity of the alkanehydroxylase system of Pseudomonas oleovorans GPo1. Enzyme Microb Technol1994;16(10):904–11.

[7] van Beilen JB. Identification of an amino acid position that determinesthe substrate range of integral membrane alkane hydroxylases. J Bacteriol2005;187(1):85–91.

[8] McKenna EJ, Coon MJ. Enzymatic omega-oxidation. IV. Purification and prop-erties of the omega-hydroxylase of Pseudomonas oleovorans. J Biol Chem1970;245(15):3882–9.

[9] Mathys RG, Schmid A, Witholt B. Integrated two-liquid phase bioconversionand product-recovery processes for the oxidation of alkanes: process designand economic evaluation. Biotechnol Bioeng 1999;64(4):459–77.

10] Frost, Sullivan. Eur Fatty Alcohols Markets 2008.11] Eggink G. Controlled and functional expression of the Pseudomonas oleovorans

alkane utilizing system in Pseudomonas putida and Escherichia coli. J Biol Chem1987;262(36):17712–8.

12] Favre-Bulle O, Witholt B. Biooxidation of n-octane by a recombinant Escheri-cia coli in a two-piquid-phase system: effect of medium components oncell growth and alkane oxidation activity. Enzyme Microb Technol 1992;14:931–7.

13] Witholt B. Synthesis of aplolar organic compounds by Pseudomonas spp.

and Eschericia coli in two-liquid-phase fermentations, in Pseudomonas:molecular biology and biotechnology. Washington, DC: ASM Press.; 1992.p. 301–13.

14] Chen Q, Janssen DB, Witholt B. Physiological changes and alk gene instabil-ity in Pseudomonas oleovorans during induction and expression of alk genes. JBacteriol 1996;178(18):5508–12.

[

[

Technology 48 (2011) 480–486

15] Schmid A, Kollmer A, Witholt B. Effects of biosurfactant and emulsificationon two-liquid phase Pseudomonas oleovorans cultures and cell-free emulsionscontaining n-decane. Enzyme Microb Technol 1998;22:487–93.

16] Van Beilen JB, Wubbolts M, Witholt B. Genetics of alkane oxidation by Pseu-domonas oleovorans. Biodegradation 1994;5:161–74.

17] Bradford MM. A rapid and sensitive method for the quantitation of micro-gram quantities of protein utilizing the principle of protein–dye binding. AnalBiochem 1976;72(1–2):248–54.

18] Laemmli UK. Cleavage of structural proteins during the assembly of the headof bacteriophage T4. Nature 1970;227(5259):680–5.

19] Hachmann JP, Amshey JW. Models of protein modification in Tris–glycine andneutral pH Bis–Tris gels during electrophoresis: effect of gel pH. Anal Biochem2005;342(2):237–45.

20] Eggink G. Alkane utilization in Pseudomonas oleovorans. Structure and functionof the regulatory locus alkR. J Biol Chem 1988;263(26):13400–5.

21] Wubbolts MG. Xylene and alkane mono-oxygenases from Pseudomonasputida: genetics, regulated expression and utilization in the synthesis of opti-cally active synthons. Groningen; 1994.

22] Smits T. Cloning and functional analysis of bacterial genes involved in alkaneoxidation. ETH Zurich; 2001.

23] Picataggio S. Metabolic engineering of candida tropicalis for the production oflong-chain dicarboxylic acids. Biotechnology (N Y) 1992;10:894–8.

24] van Beilen JB. DNA sequence determination and functional characterization ofthe OCT-plasmid-encoded alkJKL genes of Pseudomonas oleovorans. Mol Micro-biol 1992;6(21):3121–36.

25] Katopodis AG. Mechanistic studies on non-heme iron monooxygenasecatalysis: epoxidation, aldehyde formation and demethylation by theomega-hydroxylation system of Pseudomonas oleovorans. J Am Chem Soc1984;106(25):7928–35.

26] Rothen SA. Biotransformation of octane by E. coli HB101[pGEc47] on definedmedium: octanoate production and product inhibition. Biotechnol Bioeng1998;58(4):356–65.

27] Bangham AD, Standish MM, Miller N. Cation permeability of phospholipidmodel membranes: effect of narcotics. Nature 1965;208(5017):1295–7.

28] De Smet MJ, Kingma J, Witholt B. The effect of toluene on the structure and per-meability of the outer and cytoplasmic membranes of Escherichia coli. BiochimBiophys Acta (BBA)—Biomembr 1978;506(1):64–80.

29] Weber FJ, de Bont JAM. Adaptation mechanisms of microorganisms to the toxiceffects of organic solvents on membranes. Biochim Biophys Acta (BBA)—RevBiomembr 1996;1286(3):225–45.

30] Sikkema J, De Bont JAM, Poolman B. Mechanisms of membrane toxicity ofhydrocarbons. Microbiol Rev 1995;59(2):201–22.

31] Chen Q, Janssen DB, Witholt B. Growth on octane alters the membrane lipidfatty acids of Pseudomonas oleovorans due to the induction of alkB and synthesisof octanol. J Bacteriol 1995;177(23):6894–901.

32] de Bont JAM. Solvent-tolerant bacteria in biocatalysis. Trends Biotechnol

1999;16(12):493–9.

33] van Beilen JB. Analysis of Pseudomonas putida alkane-degradation gene clustersand flanking insertion sequences: evolution and regulation of the alk genes.Microbiology 2001;147(Pt 6):1621–30.

34] Shaw DG. Hydrocarbons with water and seawater. Solubility data series.Oxford: Pergamon Press; 1989.