Chemical Biotechnology for the Specific

Oxyfunctionalization of Hydrocarbons on
a Technical Scale

Bruno Bühler,1 Irene Bollhalder,1 Bernhard Hauer,2 Bernard Witholt,1

Andreas Schmid1
1
Institute of Biotechnology, Swiss Federal Institute of Technology Zurich,
CH-8093 Zurich, Switzerland; telephone: +41 1 633 36 91; fax: +41 1 633 10 51;
e-mail: andreas.schmid@biotech.biol.ethz.ch.
2
BASF Corporation, Research Fine Chemicals and Biotechnology, D-67056
Ludwigshafen, Germany
Received 6 September 2002; accepted 25 November 2002

DOI: 10.1002/bit.10637

Abstract: Oxygenases catalyze, among other interesting These enzymes use molecular oxygen to introduce oxygen
reactions, highly selective hydrocarbon oxyfunctionaliza- in specific substrates providing an effective and sustainable
tions, which are important in industrial organic synthesis
but difficult to achieve by chemical means. Many enzy- approach to oxyfunctionalization, since highly reactive oxi-
matic oxygenations have been described, but few of dants are avoided.
these have been scaled up to industrial scales, due to the Oxygenases are usually cofactor dependent, often multi-
complexity of oxygenase based biocatalysts and de- component, and/or membrane-associated enzyme systems,
manding process implementation. We have combined
recombinant whole-cell catalysis in a two-liquid phase which constricts the use of isolated enzymes in practical
system with fed-batch cultivation in an optimized me- applications (Faber, 2000; Li et al., 2002). Instead, efforts
dium and developed an industrially feasible process for during the past two decades have focused on whole cell
the kinetically controlled and complex multistep oxida- biocatalysis. Various solutions to issues such as biocatalyst
tion of pseudocumene to 3,4-dimethylbenzaldehyde us-
stability, narrow substrate range, low volumetric productivi-
ing the xylene monooxygenase of Pseudomonas putida
mt-2 in Escherichia coli. Successful scale up to 30 L work- ties, and process setup have been developed based on im-
ing volume using downscaled industrial equipment al- proved screening strategies (Wahler and Reymond,
lowed a productivity of 31 g L−1 d−1 and a product con- 2001a,b), the use of recombinant strains (Lin et al., 1999;
centration of 37 g L−1. These performance characteristics Panke et al., 2000, 2002), biocatalyst engineering (Arnold,
meet present industry requirements. Product purification
resulted in the recovery of 469 g of 3,4-dimethyl- 2001; Zhao et al., 2002), regulated substrate addition (Hack
benzaldehyde at a purity of 97% and an overall yield of et al., 2000), and in situ product recovery (Lye and Wood-
65%. This process illustrates the general feasibility of in- ley, 1999). However, despite these promising approaches,
dustrial biocatalytic oxyfunctionalization. © 2003 Wiley Pe- very few industrial biooxidation processes have been devel-
riodicals, Inc. Biotechnol Bioeng 82: 833–842, 2003.
Keywords: biocatalysis; oxygenase; chemical synthesis;
oped thus far for the production of high value compounds
integrated bioprocess (Schmid et al., 2001).
In this paper, we describe the initial steps for the devel-
opment of such a process, in this case for the production of
INTRODUCTION aromatic aldehydes, which serve as ingredients of natural
flavors and fragrances and as synthons for a variety of poly-
Selective oxyfunctionalization of petrochemicals is a major
mers, pharmaceuticals, and fine chemicals. Chemical C–H
topic in industrial organic synthesis (Wittcoff and Reuben,
bond activation via carbonylation or oxygen addition usu-
1996), as recognized by the 2001 Nobel Prize in chemistry,
ally requires the use of expensive and hazardous reactants
awarded to K.B. Sharpless for his work on chirally cata-
and catalysts and often yields product mixtures complicat-
lyzed oxidation reactions (Katsuki and Sharpless, 1980).
ing product isolation (Carelli et al., 1999; Thomas et al.,
Nature has much to offer in this area, given the many oxi-
1999). Thus, we consider it worth exploring the biocatalytic
dation reactions catalyzed by oxygenases that occur in mi-
crobial metabolic pathways (Faber, 2000; Harayama et al., production of aromatic aldehydes from cheap substrates
1992; Leadbetter and Foster, 1959; Witholt et al., 1990). such as xylenes via selective C-H bond functionalization.
The enzyme system which we use for this oxidation is the
xylene monooxygenase (XMO) of Pseudomonas putida
Correspondence to: Andreas Schmid, Institute of Biotechnology, ETH mt-2 (Harayama et al., 1989; Worsey and Williams, 1975),
Zürich, Hönggerberg HPT, CH-8093 Zürich, Switzerland. which consists of a NADH:acceptor reductase component

© 2003 Wiley Periodicals, Inc.

(XylA) (Shaw and Harayama, 1992) and a membrane- for the culture before. The pH in the reactor was kept at 7.1
bound hydroxylase component (XylM) (Shaw and Hara- or 7.4 by the addition of 25% (wt/vol) ammonium water or
yama, 1995). This non-heme iron enzyme has a broad sub- 30% phosphoric acid. One and a half liters of such cultures
strate range (Wubbolts et al., 1994) and initiates xylene were used as inocula for biotransformations on a 30-L scale.
degradation by the specific hydroxylation of one methyl
group (Abril et al., 1989). We have demonstrated that re-
combinant Escherichia coli expressing the xylMA genes un- Biotransformation Setup
der control of the alk regulatory system (Panke et al., 1999)
and thus containing the integral membrane enzyme XMO Thirty-liter biotransformations were performed in a 42-L
catalyze the monooxygenation not only of toluene and pseu- stirred tank reactor with regulated temperature, pH, stirrer
documene but also of the corresponding benzyl alcohols and speed, aeration, and internal pressure (New MBR, Zürich,
benzaldehydes (Bühler et al., 2000). Thus, one single re- Switzerland). Data collection occurred every 30 s with the
combinant enzyme system (XylMA) can specifically pro- Caroline II software (PCS, Wetzikon, Switzerland) on an
duce desired benzaldehydes from xylenes. OS/9 operating system. Mixing was achieved with two six-
We now report on specific and efficient biocatalytic hy- bladed Rushton turbine impellers. Two-liter experiments
drocarbon oxidation achieved by XMO based recombinant were carried out in a 3-L reactor (see above). Maintenance
whole-cell catalysis in a two-liquid phase system combined of pH was achieved as described above. Typical biotrans-
with fed-batch cultivation in an optimized medium. We formation procedures as developed and optimized earlier
used downscaled industrial equipment to assess the scalabil- (Bühler et al., 2003) started with batch cultivation (Fig. 1).
ity and were able to produce the toxic and unstable 3,4- Depending on the scale, the batch culture volumes
dimethylbenzaldehyde (DMB aldehyde) from equally toxic amounted to 1 or 15 L. Before sterilization, the reactors
but inexpensive pseudocumene via a sophisticated multi- contained salts equivalent to 0.9 or 13.5 L of RB medium
step catalysis process using the complex XMO system. To dissolved in 0.85 or 12.5 L water. After sterilization, the
our knowledge, this is the first report on a scalable inte- reactor contents were supplemented aseptically with, per
grated process for selective biooxidation of a non- liter culture volume, 2.5 mL of 1 M magnesium sulfate, 14
functionalized hydrocarbon allowing good productivity, mL of a 50% (wt/vol) glucose solution, 1 mL of a 1%
yield, and product concentration. (wt/vol) thiamine solution, and 5 mL of trace element so-
lution USFe. Omitting kanamycin had no impact on process
efficiency (Bühler et al., 2003). The pH was adjusted to 6.8
MATERIALS AND METHODS with 25% (wt/vol) ammonium water and to pH 7.1 or 7.4
with 10 M NaOH and the volume to 0.9 or 13.5 L with
Strain and Precultivation water. After inoculation, batch cultivation was performed at
a stirring speed of 1,500 and 600 rpm and an aeration of 1
E. coli JM101 (supE thi ⌬(lac-proAB) F⬘[traD36 proAB+ and 15 L min−1 on small and large scale, respectively. After
lacIq lacZ⌬M15]) (Messing, 1979), an E. coli K-12 deriva- initial glucose and metabolic byproducts (e.g., acetic acid)
tive, was used as recombinant host strain. As expression had completely been consumed, a feed of 73% (wt/vol)
vector, we used the pBR322 derived plasmid pSPZ3 con- glucose and 19.6 g L−1 magnesium sulfate was initiated.
taining the XMO genes xylM and xylA under the control of Furthermore, the culture was supplemented with 5 mL of
the alk regulatory system (Panke, et al., 1999). All growth trace element solution USFe and 4 mL of 1% (wt/vol) thia-
steps were carried out at 30°C. E. coli JM101 (pSPZ3) mine and xylMA expression was induced by the addition of
transformants were selected on Luria-Bertani broth (LB) 0.02% (vol/vol) dicyclopropyl ketone (DCPK, 95%, Al-
plates (Sambrook et al., 1989) containing 50 mg L−1 kana- drich, Buchs, Switzerland) or 0.1% (vol/vol) n-octane
mycin and 10 g L−1 glucose to avoid indigo formation (>98.5%, Acros, Geel, Belgium). After 1–1.5 h of fed-batch
(Panke et al., 1999). To increase the reproducibility, freshly growth, the biotransformation was started by the addition of
transformed cells served as inoculum for 3-mL LB cultures a volume of organic phase equivalent to the culture volume.
supplemented with glucose and kanamycin as before. These The organic phase consisted of bis(2-ethylhexyl)phthalate
cultures were diluted 100-fold with 100 mL of RB minimal (BEHP, 97%, Fluka, Buchs, Switzerland) as the carrier sol-
medium (Bühler et al., 2003) complemented with, per liter, vent, variable amounts of pseudocumene (99%, Fluka), and
2 mL of 1 M magnesium sulfate, 5 g glucose, 1 mL of 1% 1% (vol/vol) n-octane. Adjusting stirrer speed and aeration
(wt/vol) thiamine, 50 mg kanamycin, and 1 mL of trace and, if necessary, oxygen admixture prevented oxygen limi-
element solution USFe (Bühler et al., 2003). Such precul- tation. Sampling and determination of reactant concentra-
tures were transferred into a 3-L stirred tank reactor, in tions in both phases and glucose, acetic acid, and cell con-
which biotransformations on a 2-L scale or precultivations centrations in the aqueous phase were performed as de-
for large-scale experiments were performed. Details of the scribed (Bühler et al., 2003). Foam formation was limited
reactor system and its instrumentation have been described by the addition of Antifoam 289 (Sigma, Buchs, Switzer-
(Wubbolts et al., 1996). For precultivations the reactor con- land). Volumetric productivities and specific product for-
tained 1.9 L of mineral medium of identical composition as mation rates were calculated for intervals between

834 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 7, JUNE 30, 2003
Figure 1. Biotransformation process. Batch growth is followed by fed-batch cultivation and the two-liquid phase biooxidation of pseudocumene.
Pseudocumene, DMB alcohol, and DMB aldehyde, but not DMB acid (at neutral pH), mainly partition into the organic phase (Bühler et al., 2003).

two sampling points. One Unit is defined as the activity that DMB alcohol content in the organic phase was as low as
produces 1 ␮mol of product per minute. possible because the separation of DMB alcohol and DMB
aldehyde was difficult during such a purification procedure.

Downstream Processing
RESULTS
After harvesting, the two phases were separated by centrifu-
gation for 30 min at 8,400g in a Haereus-Christ Cryofuge
Multistep Laboratory-Scale Oxidation of
8000 centrifuge (Haereus, Zurich, Switzerland). The or- Pseudocumene to DMB Aldehyde
ganic phase was supplemented with dry sodium sulfate and
stored at 4°C. After filtration, DMB aldehyde was purified The laboratory-scale process is outlined in Figure 1. Re-
by fractionated distillation at a pressure of 0.1 mbar. At this combinant E. coli JM101 expressing the XMO genes under
pressure and room temperature, octane (boiling point [bp]: control of an alkane responsive regulatory system (Panke et
126°C) and pseudocumene (bp: 169°C) condensed in the al., 1999) was used as biocatalyst. After precultivation of
dry ice containing cryo trap. Organic phase portions of 4.5 the cells, the bioreactor was inoculated. Following batch
L were heated to 110°C, when the distillate of a temperature growth in 1 L of minimal medium, a feed of glucose and
of 35°C condensed in the cooler. During distillation, the magnesium sulfate was initiated and XMO synthesis was
temperature at the still was raised to 190°C and the tem- induced with DCPK or n-octane (see Materials and Meth-
perature of the distillate increased to 55°C. This fraction ods). The biotransformation started with the addition of 1 L
contained DMB aldehyde (bp: 233°C) at a purity of 97% as of organic phase consisting of BEHP as the carrier solvent,
determined by GC and HPLC, whereas traces of DMB al- which contained the substrate pseudocumene and 1% (vol/
cohol (bp: 218–221°C; mp: 62–65°C) sublimed in the vol) n-octane to maintain continuous induction. This proce-
cooler and DMB acid (mp: 165–167°C) as well as BEHP dure resulted from characterization and optimization work
(bp: 384°C) remained in the still. It was essential that the described earlier (Bühler et al., 2003). In such a system,

BÜHLER ET AL.: BIOCATALYTIC HYDROCARBON FUNCTIONALIZATION 835

high overall concentrations of the toxic and poorly water- DMB alcohol and DMB aldehyde to the formation of DMB
soluble reactants could be added to the two-liquid phase aldehyde only (Bühler et al., 2003). As full induction is
system without inhibiting bacterial growth and DMB alde- reached only after about 3 h (Bühler et al., 2000), specific
hyde accumulated to high concentrations. Glucose was fed product and DMB aldehyde formation rates increased at the
continuously as the carbon and energy source to maximize beginning of the bioconversion (Fig. 2B). After the induc-
biocatalyst concentrations and volumetric productivities. tion period, the specific DMB aldehyde formation rate re-
The left column of Table I lists relevant parameters of such mained high at around 40 U (g of CDW)−1 until cell growth
a biotransformation (experiment 1). ceased, whereupon it decreased steadily in the stationary
We found that the medium pH influences the bioconver- phase (Fig. 2). However, the increase of the medium pH
sion efficiency. Figure 2 shows a biotransformation (experi- from 7.1 to 7.4 significantly enhanced the specific activities
ment 2) performed at the same conditions as experiment 1 but lowered growth rates and maximal cell concentrations
except for the pH, which was 7.4 and enabled better pro- (Table I). Since such a pH difference has no effect on the
ductivity than biotransformation at pH 7.1. At constant aera- growth of wild type E. coli JM101, the impaired growth of
tion (1 L min−1) and stirrer speed (2,000 rpm), the dissolved biocatalytically active E. coli JM101 (pSPZ3) points to a
oxygen tension (DOT) never decreased below 50% satura- metabolic burden of the high specific activities for the host
tion. The initial simultaneous accumulation of 3,4- strain as observed for alkane hydroxylation (Bosetti et al.,
dimethylbenzyl alcohol (DMB alcohol) and DMB aldehyde 1992). Overall, volumetric productivity and final product
was followed by the formation of DMB aldehyde only, after concentration were significantly improved by the pH in-
the pseudocumene concentration had decreased below 150 crease to 7.4.
mM. This and the lack of significant 3,4-dimethylbenzoic
acid (DMB acid) formation can be attributed to the complex Scale Up of the Bioconversion to Technical Scale
kinetics of the multistep oxidation of pseudocumene and the In order to evaluate the scalability of the two-liquid phase
favorable partitioning behavior of the reactants in the two- biooxidation system, we carried out the biotransformation
liquid phase system. We have previously shown that the in a 42-L bioreactor containing 15 L of minimal medium
presence of pseudocumene and DMB alcohol inhibits DMB and 15 L of organic phase (experiment 3, Table I). The
aldehyde oxidation (Bühler et al., 2003). Furthermore, sub- lower power input of maximally 5 W L−1 in the 42-L reac-
strate limitation in the first oxidation step, as indicated by tor, as compared to 60 W L−1 in previous experiments, and
the parallel decrease of the pseudocumene concentration an aeration rate of 15 L min−1 sufficed to maintain the DOT
and the specific pseudocumene oxidation rate (Fig. 2), above 15%. Product formation and growth behavior were
causes a switch from the simultaneous accumulation of similar to that seen at laboratory scale. However, lower cell

Table I. Process characteristics at varying pH and scale.*

Experiment no.
a
Parameter Unit 1 2 3 4

Working volume L 2 2 30 30
pH 7.1 7.4 7.4 7.1 → 7.4
Growth rateb h−1 0.35 0.24 0.25 0.34
Maximal CDW reached g Laq−1 30 18 15.5 20
Biotransformation timec h 14.5 13 14.5 14

Initial [pseudocumene]org mM 315 500 400 355

Final [DMB aldehyde]org mM 220 330 254 274
DMB aldehyde produced g Lorg−1 29.5 44.3 34.1 36.8
Molar yield % 70 66 64 77
Product share in all reactantsd % 92 80 76 97

Maximal specific activitye U (g of CDW)−1 27 61 65 51

Maximal specific activityald U (g of CDW)−1 16 43 36 30
Maximal volumetric activityald U Laq−1 390 710 440 500
Maximal productivityald g Ltot−1 h−1 1.6 2.9 1.8 2
Average productivityald g Ltot−1 h−1 1 1.7 1.2 1.3

*All experiments were carried out under the same conditions at varying pH and scale (organic carrier solvent, BEHP; phase ratio, 0.5; for details see
Materials and Methods). CDW: cell dry weight. Subscripts: org, organic phase; aq, aqueous phase; tot, total volume; ald, referring to aldehyde formation.
a
Adapted from an earlier study (Bühler et al., 2003).
b
In the exponential growth phase.
c
Second phase addition is defined as the start point.
d
In the organic phase at the end of the biotransformation.
e
Based on the sum of all products formed (corresponds to the specific pseudocumene oxidation rate).

836 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 7, JUNE 30, 2003
Figure 2. Fed-batch based two-liquid phase biotransformation on a 2-L scale (pH 7.4) with E. coli JM101 (pSPZ3) as biocatalyst and BEHP as organic
carrier solvent (experiment 2). After batch growth of a 1-L culture, cells were induced and a glucose feed was initiated (time 0) and, 1.25 h later, the
biotransformation was started by the addition of 1 L of organic phase (arrow). For details see Materials and Methods. (A) Reactant and octane concentrations
represent the sum of the respective concentrations in the organic and the aqueous phase. (B) Specific DMB aldehyde and product formation rates and DMB
aldehyde volumetric productivities are calculated for intervals between two sampling points. The specific product formation rate is based on the sum of
all products formed in a time interval and is synonymous to the specific pseudocumene oxidation rate. (C) Course of glucose and acetic acid concentrations,
biomass formation, and glucose feed rate. CDW: cell dry weight.

concentrations and slightly lower specific DMB aldehyde lyst concentrations in combination with high biocatalyst ac-
formation rates as compared to laboratory scale reduced the tivities, we introduced a pH shift from 7.1 via 7.2 to 7.4. As
average productivity from 1.7 to 1.2 g L−1 h−1. intended, DMBaldehyde was the predominant product at the
To ensure complete conversion and thus facilitate product end of the biotransformation and high cell concentrations
isolation, the initial pseudocumene concentration was re- were combined with high biocatalyst activities (Fig. 3,
duced (experiment 4). Furthermore, to reach higher biocata- Table I). After induction and pH shift, the specific DMB

BÜHLER ET AL.: BIOCATALYTIC HYDROCARBON FUNCTIONALIZATION 837

Figure 3. Biocatalytic synthesis of DMB aldehyde from pseudocumene by E. coli JM101 (pSPZ3) on a 30-L scale (experiment 4). Time 0 indicates
induction and initiation of a glucose feed to a 15-L culture grown in batch mode. After 1 h, the biotransformation was started by the addition of 15 L of
organic phase (arrow). For details see Materials and Methods. (A–C) As described for Figure 2. In panel B, additionally, the course of the pH in the reactor
is shown. (D) Course of stirrer speed, dissolved oxygen tension (DOT), and aeration rate. Due to the high oxygen demand, the reactor was aerated with
an air–oxygen mixture, which was readjusted several times. Open and closed arrowheads indicate start and stop of oxygen admixture, respectively. CDW:
cell dry weight.

aldehyde formation rate and thus productivity remained Downstream Processing

high until cell growth ceased, illustrating the necessity of In situ product extraction in the two-liquid phase system
metabolically active cells for optimal process performance enabled a straightforward downstream processing consist-
as detailed in an earlier study (Bühler et al., 2003). ing of phase separation, drying the organic phase over so-
At the beginning of the biotransformation at pH 7.1, the dium sulfate, and vacuum distillation. Phase separation was
oxygen requirement apparently exceeded the available oxy- achieved by centrifugation and resulted in the recovery of
gen. Increasing the power input to the maximum of 5 W L−1 13.5 L of dry organic phase, corresponding to 90% of the
at a stirrer speed of 1,000 rpm and the aeration rate to a apolar phase originally added to experiment 4 (Fig. 4). Dur-
maximum of 50 L min−1 did not satisfy the oxygen require- ing harvesting and phase separation, acetic acid production
ment of the cells. Thus, the reactor was aerated with an by the cells caused a pH decrease. Thus, DMB acid increas-
air–oxygen mixture. A molar yield of 100% was not reached ingly partitioned into the organic phase, due to its proton-
because aeration of the reactor caused substrate evaporation. ation. However, this had no impact on further product pu-
Overall, the up scaled two-liquid phase process allowed the rification. A single distillation of the organic phase at a
production of 550 g of DMB aldehyde, a final product share pressure of 0.1 mbar yielded 469 g of DMB aldehyde at a
in all reactants of 97% in the organic phase, and a maximal purity of 97% (GC, HPLC). 1H-NMR analysis confirmed
productivity of 2 g Ltot−1 h−1. the identity of the product (results not shown).

838 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 7, JUNE 30, 2003
 Figure 3. Continued.

DISCUSSION 1999, 2000), enabled high volumetric productivities on a

technical scale.
Recombinant Whole-Cell Biocatalysis
Fed-Batch Cultivation in an Optimized Medium
Complex cofactor dependent enzymes such as oxygenases
are normally used in vivo to enable inexpensive cofactor In order to maintain high oxygenase activities over time, we
regeneration (Faber, 2000; Li et al., 2002). This approach employed growing cells with high metabolic activity, which
has been useful when dead end products are produced, as in were expected to provide enough energy for efficient co-
the oxidation of aromatic heterocycle methyl groups to the factor regeneration and oxygenase synthesis (Doig et al.,
corresponding carboxylic acids by wild-type P. putida mt-2, 1999; Faber, 2000). Cells were cultivated in fed-batch mode
which harbors the xylene degradation pathway (Kiener, in a medium optimized for E. coli and XMO expression
1992). Aromatic aldehydes such as DMB aldehyde are, (Bühler et al., 2003; Riesenberg et al., 1991), and high and
however, further metabolized in Pseudomonas strains, stable productivities were indeed reached during growth. In
which reduces yields. Such product degradation was effi- the stationary phase, when metabolic activity and viability
ciently prevented by the use of recombinant E. coli, which of the cells decreased, biocatalyst activities declined inde-
could be grown to high cell densities on glucose as a cheap pendently from the reactant concentrations (Bühler et al.,
source of carbon and energy. Expression of the XMO genes 2003), for an optimal biotransformation time of about 14 h
under alk regulatory control, which has been shown to sup- for the chosen process setup. Also with respect to genetic
port high levels of inducible monooxygenase activities in E. stability, E. coli JM101 (pSPZ3) proved to be a suitable
coli (Bühler et al., 2000; Nieboer et al., 1993; Panke et al., biocatalyst, since plasmid as well as biocatalyst activity

BÜHLER ET AL.: BIOCATALYTIC HYDROCARBON FUNCTIONALIZATION 839

Figure 4. Efficiency of the downstream processing. Phase separation via centrifugation is followed by drying over sodium sulfate and filtration yielding
90% of the organic phase volume initially added in experiment 4 (Fig. 3). Fractionated distillation at a pressure of 0.1 mbar yielded 97% pure DMB
aldehyde. The overall purification yield amounted to 85%.

remained stable for 11 generations without selection pres- organic phase to the cells were reached. As outlined earlier,
sure during two successive biotransformations performed in DMB alcohol oxidation is the limiting step in the overall
a repetitive fed-batch mode of cultivation (Bühler et al., biotransformation (Bühler et al., 2003). By optimizing the
2003). Similarly, Favre-Bulle et al. reported segregational medium pH the specific activities including the DMB alco-
and structural stability for as long as 29 generations in E. hol oxidation rate were significantly enhanced. The BEHP
coli HB101 (pGEc47) without selection pressure, but highly based two-liquid phase system allowed exploitation of the
varying genetic stabilities in different E. coli host strains complex kinetics of the multistep oxidation of pseudoc-
(Favre-Bulle et al., 1993). This genetic stability permits the umene to produce high concentrations of DMB aldehyde as
elimination of antibiotics from these industrial biopro- the predominant product in the organic phase. Furthermore,
cesses, which is useful from an economical and ecological this concept facilitated product isolation.
point of view. The feasibility of a repetitive fed-batch mode
of cultivation also emphasizes the possibility to reuse the
biocatalyst. Such reuse, however, still necessitates cell Industrial Feasibility
growth.
Two-liquid phase processes have thus far mostly operated at
laboratory scale. In order to assess the industrial feasibility
Two-Liquid Phase Concept of biocatalytic hydrocarbon oxyfunctionalization, we used a
reactor with an estimated power input (5 W L−1) close to
Hydrocarbons and their oxidized products are often poorly values typically reached on industrial scales (Manfredini
water-soluble and toxic or inhibitory for the biocatalyst. In and Manfredini, 1996; Woodley, 1990). A pilot scale of
situ product removal, including the use of membrane reac- 20–30 L often serves as a platform for scalability evaluation
tors, solid phase extraction, and two-phase systems, is com- of industrial processes and for scale up to production (50–
monly used to overcome these hurdles (Leon et al., 1998; 100 m3). Requirements for industrial oxidation processes
Lye and Woodley, 1999). In two-liquid phase processes, the include productivities between 10 and 60 g L−1 d−1, high
properties of the chosen recyclable organic carrier solvent biocatalyst stability, high yields (60–80%), product concen-
are of special importance (Bruce and Daugulis, 1991; Salter trations of 10–100 g L−1, and simple product recovery
and Kell, 1995; Schmid et al., 1998). The use of BEHP with (Straathof et al., 2002). Schmid et al. envisioned volumetric
its low flammability and low toxicity to E. coli (Bühler et activities in the range of 200–1,000 U L−1 for large-scale
al., 2003; Panke et al., 2000) and humans (Hasmall et al., two-liquid phase processes considering future developments
2000) allowed the safe operation of a 42-L stirred tank in biocatalyst and reactor engineering (Schmid et al., 1998).
reactor with a 50% volume fraction of this solvent. Further- Taking an average volumetric activity of 100 U L−1 for
more, substrate and product toxicity were efficiently pre- alkane hydroxylation, Mathys et al. demonstrated the eco-
vented, and high pseudocumene mass transfer rates from the nomic feasibility of such processes on the basis of detailed

840 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 7, JUNE 30, 2003
modeling analysis (Mathys et al., 1999). We now achieved corresponding alcohols, aldehydes, and acids in Escherichia coli
for the first time an average volumetric activity of 330 U JM101. J Biol Chem 275:10085–10092.
Bühler B, Witholt B, Hauer B, Schmid A. 2002. Characterization and
Laq−1 with a maximum at 500 U Laq−1 for a technical-scale application of xylene monooxygenase for multistep biocatalysis. Appl
two-step oxygenation process. This corresponds to an over- Environ Microbiol 68:560–568.
all oxygenation activity of 1000 U Laq−1 or 500 U Ltot−1. Burton SG, Cowan DA, Woodley JM. 2002. The search for the ideal
The resulting productivity of 31 g Ltot−1 d−1, a final DMB biocatalyst. Nat Biotechnol 20:37–45.
aldehyde concentration of 37 g L−1 in the organic phase, and Carelli I, Chiarotto I, Cacchi S, Pace P, Amatore C, Jutand A, Meyer G.
1999. Electrosynthesis of aromatic aldehydes by palladium-catalyzed
an isolated product yield of 65% (conversion yield: 77%) carbonylation of aryl iodides in the presence of formic acid. Eur J Org
clearly illustrate the practical feasibility of large-scale two- Chem 1999:1471–1473.
liquid phase processes and point to the large potential of Doig SD, Boam AT, Livingston AG, Stuckey DC. 1999. Epoxidation of
biocatalytic hydrocarbon oxidation in organic synthesis. 1,7-octadiene by Pseudomonas oleovorans in a membrane reactor.
Another recent example is the use of recombinant E. coli Biotechnol Bioeng 63:601–611.
Ellis LBM, Hershberger CD, Bryan EM, Wackett LP. 2001. The University
containing styrene monooxygenase of Pseudomonas sp. of Minnesota Biocatalysis/Biodegradation Database: emphasizing en-
VLB120 for the enantioselective single step oxidation of the zymes. Nucleic Acids Res 29:340–343.
styrene vinyl function to produce (S)-styrene oxide with an Faber K. 2000. Biotransformations in organic chemistry. Berlin: Springer-
enantiomeric excess of >99% at a purity of >97% (Panke et Verlag.
al., 2002). Favre-Bulle O, Weenink E, Vos T, Preusting H, Witholt B. 1993. Con-
tinuous bioconversion of n-octane to octanoic acid by recombinant
Escherichia coli (alk+) growing in a two-liquid-phase chemostat. Bio-
technol Bioeng 41:263–272.
Perspectives
Hack CJ, Woodley JM, Lilly MD, Liddell JM. 2000. Design of a control
system for biotransformation of toxic substrates: toluene hydroxylation by
Enzymes are optimized by nature for the survival of a spe-
Pseudomonas putida UV4. Enzyme Microb Technol 26:530–536.
cies in a specific ecological environment and not to meet the Harayama S, Kok M, Neidle EL. 1992. Functional and evolutionary relation-
requirements of industrial biocatalytic processes. Neverthe- ships among diverse oxygenases. Annu Rev Microbiol 46:565–601.
less, biochemical engineering and modern DNA technology Harayama S, Rekik M, Wubbolts M, Rose K, Leppik RA, Timmis KN.
provide a variety of tools to develop and optimize a bio- 1989. Characterization of five genes in the upper-pathway operon of
TOL plasmid pWW0 from Pseudomonas putida and identification of
catalyst and a process setup (Burton et al., 2002; Panke and
the gene products. J Bacteriol 171:5048–5055.
Wubbolts, 2002; Patnaik et al., 2002; Schmid et al., 2002). Hasmall SC, James NH, Macdonald N, Soames AR, Roberts RA. 2000. Spe-
The present example shows that the combined use of dif- cies differences in response to diethylhexylphthalate: suppression of ap-
ferent concepts allows the application of complex enzyme optosis, induction of DNA synthesis and peroxisome proliferator acti-
systems such as oxygenases for complex reactions such as vated receptor alpha-mediated gene expression. Arch Toxicol 74:85–91.
kinetically controlled multistep reactions. The clear feasi- Katsuki T, Sharpless KB. 1980. The first practical method for asymmetric
epoxidation. J Am Chem Soc 102:5974–5976.
bility of biocatalytic oxyfunctionalization on a technical Kiener A. 1992. Enzymatic oxidation of methyl groups on aromatic het-
scale and the large pool of oxygenases in nature (Ellis et al., erocycles: a versatile method for the preparation of heteroaromatic
2001) augur well for the development of many and varied carboxylic acids. Angew Chem Int Ed Engl 31:774–775.
industrial scale hydrocarbon oxidations in the coming de- Leadbetter ER, Foster JW. 1959. Incorporation of molecular oxygen in bac-
cade. terial cells utilizing hydrocarbons for growth. Nature 184:1428–1429.
Leon R, Fernandes P, Pinheiro HM, Cabral JMS. 1998. Whole-cell bioca-
talysis in organic media. Enzyme Microb Technol 23:483–500.
The authors thank H.-J. Feiten and D. Meyer for their excellent Li Z, van Beilen JB, Duetz WA, Schmid A, de Raadt A, Griengl H, Witholt
technical assistance and F. Hollmann for NMR analyses. B. 2002. Oxidative biotransformations using oxygenases. Curr Opin
Chem Biol 6:136–144.
Lin Z, Thorsen T, Arnold FH. 1999. Functional expression of horseradish
References peroxidase in E. coli by directed evolution. Biotechnol Prog 15:467–471.
Lye GJ, Woodley JM. 1999. Application of in situ product-removal tech-
Abril M-A, Michan C, Timmis KN, Ramos JL. 1989. Regulator and en- niques to biocatalytic processes. Trends Biotechnol 17:395–402.
zyme specificities of the TOL plasmid-encoded upper pathway for Manfredini R, Manfredini A. 1996. Design of industrial bioreactors. In:
degradation of aromatic hydrocarbons and expansion of the substrate Larsson G, Förberg C, editors. Bioreactor engineering. Saltsjöbaden,
range of the pathway. J Bacteriol 171:6782–6790. Sweden: Working Parties Bioreactor Performance, Measurement, and
Arnold FH. 2001. Combinatorial and computational challenges for bio- Control (EFB). p 49–92.
catalyst design. Nature 409:253–257. Mathys RG, Schmid A, Witholt B. 1999. Integrated two-liquid phase biocon-
Bosetti A, van Beilen JB, Preusting H, Lageveen RG, Witholt B. 1992. version and product-recovery processes for the oxidation of alkanes: pro-
Production of primary aliphatic alcohols with a recombinant Pseudo- cess design and economic evaluation. Biotechnol Bioeng 64:459–477.
monas strain, encoding the alkane hydroxylase system. Enzyme Mi- Messing J. 1979. A multipurpose cloning system based on single-stranded
crob Technol 14:702–708. DNA bacteriophage M13. Recomb DNA Tech Bull 2:43–49.
Bruce LJ, Daugulis A. 1991. Solvent selection strategies for extractive Nieboer M, Kingma J, Witholt B. 1993. The alkane oxidation system of
biocatalysis. Biotechnol Prog 7:116–124. Pseudomonas oleovorans: induction of the alk genes in Escherichia
Bühler B, Bollhalder I, Hauer B, Witholt B, Schmid A. 2003. Use of the coli W3110(pGEc47) affects membrane biogenesis and results in over-
two-liquid phase concept to exploit kinetically controlled multistep expression of alkane hydroxylase in a distinct cytoplasmic membrane
biocatalysis. Biotechnol Bioeng 81:683–694. subfraction. Mol Microbiol 8:1039–1051.
Bühler B, Schmid A, Hauer B, Witholt B. 2000. Xylene monooxygenase Panke S, Held M, Wubbolts MG, Witholt B, Schmid A. 2002. Pilot-scale
catalyzes the multistep oxygenation of toluene and pseudocumene to production of (S)-styrene oxide from styrene by recombinant Esch-

BÜHLER ET AL.: BIOCATALYTIC HYDROCARBON FUNCTIONALIZATION 841

 erichia coli synthesizing styrene monooxygenase. Biotechnol Bioeng component of xylene monooxygenase, the first enzyme of the TOL-
80:33–41. plasmid-encoded pathway for the mineralization of toluenes and xy-
Panke S, Meyer A, Huber CM, Witholt B, Wubbolts MG. 1999. An alkane- lenes. J Ferm Bioeng 79:195–199.
responsive expression system for the production of fine chemicals. Straathof AJJ, Panke S, Schmid A. 2002. The production of fine chemicals
Appl Environ Microbiol 65:2324–2332. by biotransformations. Curr Opin Biotechnol 13:548–556.
Panke S, Wubbolts MG. 2002. Enzyme technology and bioprocess engi- Thomas JM, Raja R, Sankar G, Bell RG. 1999. Molecular-sieve catalysts
neering. Curr Opin Biotechnol 13:111–116. for the selective oxidation of linear alkanes by molecular oxygen.
Panke S, Wubbolts MG, Schmid A, Witholt B. 2000. Production of enan- Nature 398:227–230.
tiopure styrene oxide by recombinant Escherichia coli synthesizing a Wahler D, Reymond JL. 2001a. High-throughput screening for biocata-
two-component styrene monooxygenase. Biotechnol Bioeng 69: lysts. Curr Opin Biotechnol 12:535–544.
91–100. Wahler D, Reymond JL. 2001b. Novel methods for biocatalyst screening.
Patnaik R, Louie S, Gavrilovic V, Perry K, Stemmer WPC, Ryan CM, Curr Opin Chem Biol 5:152–158.
Cardayré D. 2002. Genome shuffling of Lactobacillus for improved Witholt B, de Smet M-J, Kingma J, van Beilen JB, Lageveen RG, Eggink
acid tolerance. Nat Biotechnol 20:707–712. G. 1990. Bioconversions of aliphatic compounds by Pseudomonas
Riesenberg D, Schulz V, Knorre WA, Pohl H-D, Korz D, Sanders EA, oleovorans in multiphase bioreactors: background and economic po-
Ross A, Deckwer W-D. 1991. High cell density cultivation of Esch-
tential. Trends Biotechnol 8:46–52.
erichia coli at controlled specific growth rate. J Biotechnol 20:17–28.
Wittcoff HA, Reuben BG. 1996. Industrial organic chemicals. New York:
Salter GJ, Kell DB. 1995. Solvent selection for whole cell biotransforma-
John Wiley & Sons, Inc.
tions in organic media. Crit Rev Biotechnol 15:139–177.
Woodley JM. 1990. Stirred tank power input data for the scale-up of
Sambrook J, Fritsch EF, Maniatis T. 1989. Molecular cloning: a laboratory
two-liquid phase biotransformations. In: Copping LG, editor. Oppor-
manual. Plainview, NY: Cold Spring Harbor Laboratory Press.
tunities in biotransformations. London: Elsevier. p 63–66.
Schmid A, Dordick JS, Hauer B, Kiener A, Wubbolts M, Witholt B. 2001.
Industrial biocatalysis today and tomorrow. Nature 409:258–268. Worsey MJ, Williams PA. 1975. Metabolism of toluene and xylenes by
Schmid A, Hollmann F, Park JB, Bühler B. 2002. The use of enzymes in Pseudomonas putida (avarilla) mt-2: evidence for a new function of
the chemical industry in Europe. Curr Opin Biotechnol 13:359–366. the TOL plasmid. J Bacteriol 124:7–13.
Schmid A, Kollmer A, Mathys RG, Witholt B. 1998. Development toward Wubbolts MG, Favre-Bulle O, Witholt B. 1996. Biosynthesis of synthons
large-scale bacterial bioprocesses in the presence of bulk amounts of in two-liquid-phase media. Biotechnol Bioeng 52:301–308.
organic solvents. Extremophiles 2:249–256. Wubbolts MG, Reuvekamp P, Witholt B. 1994. TOL plasmid-specified
Shaw JP, Harayama S. 1992. Purification and characterisation of the xylene oxygenase is a wide substrate range monooxygenase capable of
NADH:acceptor reductase component of xylene monooxygenase en- olefin epoxidation. Enzyme Microb Technol 16:608–615.
coded by the TOL plasmid pWW0 of Pseudomonas putida mt-2. Eur Zhao H, Chockalingam K, Chen Z. 2002. Directed evolution of enzymes
J Biochem 209:51–61. and pathways for industrial biocatalysis. Curr Opin Biotechnol 13:
Shaw JP, Harayama S. 1995. Characterization in vitro of the hydroxylase 104–110.

842 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 7, JUNE 30, 2003