CHEMCATCHEM
COMMUNICATIONS
We therefore turned to a strategy whereby the separately
expressed proteins are mixed as cell-free extracts (lysates) to
perform the cascade in a one-pot fashion. This approach not
only retained the high-level expression, but also allows for the
regulation of each step depending on the specific activity of
the particular biocatalyst. Since our cascade was not redox bal-
anced, the mixing of cell-free extracts had the additional ad-
vantage that native E. coli proteins and endogenous NAD(P)H
were available for co-factor regeneration. Glucose and glycerol
were therefore included in the reaction mixture to drive co-
factor regeneration by these native enzymes. Cycloheptane
was used as a model substrate, since the CYP450 showed the
best activity with cycloheptane.[17] The concentration of cyclo-
heptane used (165 mm) far exceeded its aqueous solubility
(ꢀ0.3 mm[21]). As reported by us[17] and others,[22,23] CYP450 ac-
tivity greatly benefits if these hydrophobic substrates are
added as a second phase, thereby also serving as a co-solvent.
Gratifyingly, all four combinations of biocatalysts yielded
enantholactone from cycloheptane (Figure 1 and Figure S2 of
the Supporting Information). The highest concentration of
total product was obtained when the NADH-dependent
mutant of the CYP450 was used. This could be attributed to
the higher levels of endogenous NADH as well as E. coli’s abili-
ty to more efficiently regenerate NADH using glucose and
glycerol as substrates. The highest concentration of enantho-
lactone (3.2 mm, 0.41 gLÀ1) was produced in combination IV
(Scheme S1, Supporting Information), where the NADH-depen-
dent CYP450 was paired with TeSADH and CHMO, which are
both NADP(H) dependent. In this combination, E. coli regener-
ates NADH for the hydroxylation step, with the last two
NADP(H)-dependent steps being redox balanced (self-suffi-
cient). CYP450 BM3 and its mutants also often have low cou-
pling efficiencies whereby reducing equivalents are wasted by
unproductive catalytic cycles.[24] This wasting will result in the
requirement for additional co-factor regeneration by the
native-metabolism of E. coli. Nevertheless, a turnover frequency
(TOF, calculated over the first 2 h) of 11.2 minÀ1 was achieved
for the CYP450 with respect to total oxygenated products
formed with a total turnover number (TTN) of 2011 after 20 h.
In an attempt to construct a redox-balanced cascade, we in-
cluded an enzyme for co-factor regeneration. Glucose dehydro-
genase is an often employed choice as it efficiently regener-
ates both NADH and NADPH.[25] This however led to much
higher cycloheptanol concentrations, but without the concom-
itant production of cycloheptanone and enantholactone. Both
the alcohol dehydrogenase and glucose dehydrogenase re-
quires oxidized cofactor [NAD(P)+] for activity, resulting in the
glucose dehydrogenase competing with the alcohol dehydro-
genases for oxidized co-factor. We therefore decided to uncou-
ple the co-factor dependence of the different steps of the cas-
cade by changing the co-factor regenerating enzyme to for-
mate dehydrogenase (FDH).[26] Unlike glucose dehydrogenase
that can accept both NAD+ and NADP+,[27] FDH can only
accept NAD+. This resulted in much higher levels of product
formation, with combination IV again giving the highest level
of enantholactone (10.6 mm, 1.4 gLÀ1, Figure S3, Supporting
Information) with the TTN reaching 6915 and a TOF of
33.7 minÀ1 calculated over the first 2 h. Similar to when native
E. coli enzymes were regenerating the required co-factor for
the initial CYP450 step, the first step is uncoupled from the
second and third steps by means of co-factor dependence:
step 1 is self-sufficient in that FDH recycles the required NADH,
and steps 2 and 3 are self-sufficient in the oxidation and reduc-
tion of NADPH. Similar uncoupling of cofactor dependency be-
tween different biocatalytic steps have previously been shown
for the successful two-step amination[28] or deracemization[29]
of sec-alcohols.
To determine if the endogenous levels of co-factor is limiting
the cascade, we included additional NAD+ and NADP+
(0.1 mm each) in the bioreaction mixture. This resulted in an
approximately 20% increase in enantholactone production for
combination IV (Figure S4, Supporting Information) with the
TOF now reaching 49.7 minÀ1 (TTN of 6023). Also less than 1%
of the total products formed was of the alcohol and ketone
intermediates.
Finally, in an effort to improve the overall performance of
the cascade we increased the biocatalyst concentrations by
using highly concentrated cell suspensions for the preparation
of the cell-free extracts. Time-course analysis revealed maxi-
mum lactone formation after approximately 8–12 h (Figure 2).
This is followed by a decrease in the intermediate alcohol
levels, and an increase in the corresponding ketones (Fig-
ure S5, Supporting Information) implicating both the BVMO
and CYP450 as unstable under the prolonged reaction times.
Nonetheless, yields of more than 23 mm (~3 gLÀ1; Table 1)
enantholactone were achieved. Likewise more than 0.5 gLÀ1 of
caprolactone and caprylolactone (2-oxanone) was produced
from cyclohexane and cyclooctane respectively.
Figure 1. One-pot conversion of cycloheptane to enantholactone by four dif-
ferent combinations of biocatalysts. Reaction conditions: 1 mL reaction vol-
umes with 20 mL substrate added neat, containing 1.33 mm BM3M_NADPH
and 1.77 mm BM3M_NADH (CFE of 0.02 gWCWmlÀ1), CFE of 0.04 gWCWmlÀ1
TeSADH or TADH, CFE of 0.02 gWCWmlÀ1 CHMO_M16, 100 mm glucose,
glycerol, 200 mm Tris-HCl (pH 8), 308C, 200 rpm, 20 h.
In summary, we have developed a redox-balanced one-pot
synthetic cascade employing four oxidoreductases for the syn-
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 4
&
2
&
ÞÞ
These are not the final page numbers!
Pennec, Aliz
