DNA-mediated assembly of cytochrome P450 BM3 subdomains

Article abstract of DOI:10.1021/ja204993s

Cytochrome P450 BM3 is a versatile enzyme, which holds great promise for applications in biocatalysis and biomedicine. We here report on the generation of a hybrid DNA-protein device based on the two subdomains of BM3, the reductase domain BMR and the por

Full text of DOI:10.1021/ja204993s

ARTICLE
pubs.acs.org/JACS
DNA-Mediated Assembly of Cytochrome P450 BM3 Subdomains
Michael Erkelenz, Chi-Hsien Kuo,† and Christof M. Niemeyer*
TU Dortmund, Fakult€at Chemie, Biologisch-Chemische Mikrostrukturtechnik, Otto-Hahn Strasse 6, D-44227 Dortmund, Germany
S Supporting Information
b
ABSTRACT: Cytochrome P450 BM3 is a versatile enzyme, which
holds great promise for applications in biocatalysis and biomedicine.
We here report on the generation of a hybrid DNAÀprotein device
based on the two subdomains of BM3, the reductase domain BMR
and the porphyrin domain BMP. Both subdomains were fused
genetically to the HaloTag protein, a self-labeling enzyme, allowing
for the bioconjugation with chloroalkane-modified oligonucleotides.
The subdomain-DNA-chimeras could be reassembled by comple-
mentary oligonucleotides, thus leading to reconstitution of the
monooxygenase activity of BM3 holoenzyme, as demonstrated by
conversion of the reporter substrate 12-pNCA. Arrangement of the two chimeras on a switchable DNA scaffold allowed one to
control the distance between both subdomains, as indicated by the DNA-dependent activity of the holoenzyme. Furthermore, a
switchable chimeric device was constructed, in which monooxygenase activity could be turned off by DNA strand displacement.
This study demonstrates that P450 BM3 engineering and strategies of DNA nanotechnology can be merged to open up novel ways
for the development of novel screening systems or responsive catalysts with potential applications in drug delivery.
’ INTRODUCTION
Native P450 BM3 is comprised of two domains, the reductase
domain BMR, bearing FAD and FMN prosthetic groups, and the
porphyrin domain BMP, containing the catalytic heme
(protoporphyrin IX) group. P450 BM3 usually acts as a mono-
oxygenase accepting electrons from NADPH and transferring
them via FAD and FMN prosthetic groups to its heme moiety
where they finally serve to reductively cleave molecular oxygen
and use one oxygen atom for substrate hydroxylation (Figure 1).1
Because the detailed mechanism of electron transfer on the
atomic level is still the subject of current research,16À18 crucial
questions include influences of the relative distances of the redox
centers as well as the contact interface between BMP and BMR
domains. It has been demonstrated that BMP and BMR domains
can be expressed separately and that the NADPH-dependent
fatty acid hydroxylase function of the holoenzyme can be
reconstituted by using one of the domains in large excess.19,20
While this observation indicated that intermolecular electron
transfer can occur between the BMP and BMR domains, the low
activity of only about 5% of that of the wild-type holoenzyme
suggested that physical contact plays an important role for
electron transfer. It is also of interest that both domains con-
stitute independent catalytic entities. Some variants of BMP, in
particular mutant F87A,21 can hydroxylate substrates via the
so-called peroxide shunt pathway, in which H2O2 is used instead
of NADPH, thereby rendering BMR dispensable.22 On the other
hand, BMR can catalyze the reduction of various compounds
with NADPH, such as cytochrome C, in a stand-alone reaction.4
Cytochrome P450 BM3 is one of the best studied soluble
cytochrome P450 enzymes. Because it is considered and widely
used as a model to study its human counterparts, many studies
have focused on the structure, electron pathways, and reaction
cycles catalyzed by this enzyme.1,2 Besides the similarity to
human cytochromes, the convenient heterologous expression
of P450 BM3 in E. coli and its ability to hydroxylize inactivated
CÀH bonds have attracted chemists and synthetic biologists
to exploit the biotechnological potential of P450 BM3. Especially
protein engineers are making remarkable progress in increasing
the substrate scope and tailoring the catalyzed reactions in a way
that the P450 BM3 scaffold has become a most promising
candidate for future applications in biotechnology.3
The earliest reported reactivity of P450 BM3 is the mono-
oxygenation of the ω-2 position of fatty acids,4 and meanwhile
protein engineering has succeeded to create numerous mutants,
which possess altered substrate specificity, including conversion
of indole,5 propranolol,6 polycyclic aromatic hydrocarbons,7
fluorogenic alkoxyresorufins,8 terpenes,9 as well as other com-
ponents.3 It has been demonstrated that P450 BM3 can be used
in semisynthetic approaches to produce compounds of phar-
maceutical interest like Artemisinin10 or fluorine containing
scaffolds.11 In addition to creating new biocatalysts, studies
aiming to simulate human metabolism with P450 BM312À14 or
to convert the enzyme into a magnetic resonance imaging
contrast agent15 demonstrate the versatility of this enzyme
system and might provide new benchmarks for P450 BM3
applications.
Received: June 6, 2011
Published: September 15, 2011
r
2011 American Chemical Society
16111
dx.doi.org/10.1021/ja204993s J. Am. Chem. Soc. 2011, 133, 16111–16118
|

Journal of the American Chemical Society
ARTICLE
The enormous relevance of P450 BM3 for biotechnological
applications and the system’s modularity led us to the idea that
BMP and BMR might represent ideal candidates for design and
construction of an artificial allosterically tunable hybrid enzyme,
in which variation of the relative spatial orientation of the two
functional domains would not only allow for fundamental
mechanistic studies of electron transfer processes in multi
domain proteins but also for applications in biotechnology.
A potential approach to generate conformationally switch-
able multifunctional protein entities is based on the combina-
tion of structural DNA nanotechnology23À25 with semisyn-
thetic DNAÀprotein conjugates, which enables complementa-
tion of protein function with the specific addressability of
synthetic DNA oligonucleotides.26 This approach has already
been used to trigger activity of single domain proteins, such
as allosteric enzyme activation,27,28 reconstitution of apo-
enzymes,29,30 complementation of split fluorescent proteins31
and enzymes,32 or the cascading of individual consecutive
operating enzymes.30,33À35 While these studies primarily aimed
toward establishment of biosensors, no work has yet been
reported on the use of DNA scaffolds for systematic variation
of spatial configurations of multi domain enzymes, such as
BM3. We here report on the assembly of BMP and BMR
domains and the allosteric regulation of functional holoenzyme
based on DNA hybridization.
’ EXPERIMENTAL SECTION
A detailed description of cloning and heterologous expression of
BMR and BMP fusion proteins, 12-pNCA synthesis, oligonucleotide
sequences, synthesis and characterization of DNAÀprotein conjugates,
enzyme assays and determination of hybridization kinetics, as well as
model estimations on the spatial configuration of DNA-linked BM3
subdomains is given in the Supporting Information.
Figure 1. P450 BM3-catalyzed reactions and proposed electron path-
ways in putative dimer of BM3 holoenzymes. In the NADPH-dependent
conversion of 12-pNCA, the electrons are transferred from NADPH via
FAD to FMN of the BMR domain (yellow spheres) of monomer 1 and
subsquently to the heme group of BMP (red sphere) of monomer 2.
Note that BMP can hydroxylate 12-pNCA in the absence of BMR and
NADPH using H2O2 in the peroxide shunt pathway. Note also that
electrons from NADPH can be transferred to cytochrome C, thus giving
rise to monitor activity of the isolated BMR domain photometrically.
’ RESULTS AND DISCUSSION
To investigate the possibility of linking the monooxygenase
reaction of P450 BM3 to a DNA-dependent control mechanism,
we used the well-known variant P450 BM3 F87A,5,36À38 which
catalyzes the hydroxylation of the reporter substrate p-nitrophe-
noxycarboxylic acid (12-pNCA).37 Variant P450 BM3 F87A
not only shows enhanced activity with 12-pNCA substrates as
Figure 2. Schematic representation (A) and SDS-PAGE analysis (B) of proteinÀDNA conjugation. These are a Coomassie-stained gels. Unreacted
BMP-F87A-Halo and BMR-Halo are in shown lanes 1 and 4, respectively. Crude reaction mixtures are shown obtained from conjugation of BMP-F87A-
Halo + chlorohexane-aF9 (lane 2), BMP-F87A-Halo + chlorohexane-acF9 (lane 3), BMR-Halo + chlorohexane-aF9 (lane 5), and BMR-Halo +
chlorohexane-acF9 (lane 6). Black triangles indicate the conjugate bands. The presence of DNA in the slower migrating bands of the conjugates was
confirmed by SybrGold staining (Figure S2, Supporting Information). (C) Schematic representation of fusion proteins and their DNA conjugates. (D)
SDS-PAGE analysis of conjugates purified by anion-exchange chromatography: lane 1, BMP-Halo; lane 2, BMP-Halo-aF9; lane 3, BMR-Halo; lane 4,
BMR-Halo-acF9.
16112
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
HaloTag domain (Figure S1, Supporting Information). After
expression and purification, the recombinant proteins BMP-
F87A-Halo and BMR-Halo were both conjugated with single-
stranded oligonucleotides bearing a chlorohexane substituent,
which is the suicide substrate for the HaloTag and establishes a
permanent covalent bond between the DNA oligonucleotide and
the protein (Figure 2A).
For an initial test of function, we chose two oligonucleotides,
which are complementary to each other, termed as aF9 and acF9,
for the conjugation with BMP-Halo and BMR-Halo, respectively.
The formation of the expected DNAÀprotein conjugates BMP-
Halo-aF9 and BMR-Halo-acF9 was confirmed by SDS-PAGE
analysis (Figure 2B). Crude reaction mixtures were purified to
homogeneity by anion exchange chromatography (Figure 2D
and Figure S3, Supporting Information). It should be noted that
chromatographic separation of BMP-Halo- and BMR-Halo-
DNA conjugates from the unreacted proteins was hampered
by a relatively high tendency of aggregation, in particular for the
BMR-Halo protein, leading to partial coelution of the conjugated
and unconjugated protein. The tendency of dimer formation has
previously been reported for P450 BM3.42 The purified con-
jugates were quantified by the Bradford assay, and extinction
coefficients at 260 and 280 nm were determined (Table S4,
Supporting Information). A 260/280 ratio of greater than 1 was
observed, which is indicative for DNAÀprotein conjugates.43
The same synthetic method was used to generate BMP-Halo-
and BMR-Halo-DNA conjugates from a number of oligonucleo-
tides with different sequences (Table S2, Supporting In-
formation). Yields of purified conjugates were typically about
30% with respect to the amount of chlorohexane-derivatized
oligonucleotide used in the synthesis. The ready applicability and
efficiency of BMP- and BMR-DNA conjugate syntheses carried
out in this study underlined the versatility and power of the
recently developed HaloTag-based coupling strategy.40 It should
be noted, however, that the fusion with the HaloTag domain led
to reduced enzyme activity, as determined by MichaelisÀ
Menten kinetic analysis of BMP- and BMR-Halo fusion pro-
teins and their DNA conjugates. In particular, kcat values were
significantly lower, while KM values remained almost unchanged,
as compared to wild-type BM3 and native F87A BM3 (Table S1,
Supporting Information). To omit bulky connectors and to
elucidate the influence of sterical factors, further developments
may include more direct chemical coupling systems, such as
alkyneÀazide cycloaddition or other bioorthogonal linking
chemistries.26
To investigate whether the cloned domains can be reas-
sembled to the functional BM3 holoenzyme, we initially carried
out titration experiments (Figure 3). To this end, initial controls
were conducted in which the recombinant BMP-Halo was mixed
with increasing amounts of BMR-Halo. As indicated by the
circular data points in Figure 3, a large excess of BMR-Halo
(>10 mol equiv) was necessary to initiate assembly of the two
domains, as indicated by 12-pNCA conversion. In strong con-
trast, titration of BMP-Halo-aF9 with increasing amounts of
BMR-Halo-acF9 revealed that active holoenzyme is formed
even in the case of substoichiometric amounts of BMR-Halo-
acF9 (<1 mol equiv, crossed data points in Figure 3). Maximum
activity was obtained when the two DNA-conjugated domains
were present in equimolar amounts and further increase of
BMR-Halo-acF9 did not lead to significant changes in the rate
of 12-pNCA conversion. We also confirmed that complementar-
ity of the two protein tethered oligonucleotides is necessary for
Figure 3. Establishment of DNA-based reassembly of BM3 holoen-
zyme by titration of BMP-Halo-aF9 with complementary BMR-Halo-
acF9. (A) Schematic representation of the conducted experiments. (B)
Plot of the initial rate of 12-pNCA conversion (v0) versus the relative
molar ratio of BMP to BMR domains. Shown are two experiments in
which BMR-Halo was titrated with increasing amounts of BMP-Halo
(O) and DNAÀprotein conjugates BMP-Halo-aF9 with increasing
amounts of BMR-Halo-acF9 (+).
compared to the wild-type, but it also catalyzes hydroxylation of
fatty acids by employment of the so-called peroxide shunt
pathway, that is, that H2O2 is used as the cosubstrate instead
of NADPH.21 In the presence of the native cofactor NADPH,
electrons are transferred from NADPH to the FAD/FMN
prosthetic groups of BMR to the heme group of BMP where
they are used to reduce molecular oxygen, thereby forming
compound II, which finally enables substrate oxidation
(Figure 1). Although the exact electron pathway is still the
subject of current research, it has recently been suggested that
a dimer of holoenzymes constitutes the active form of P450
BM3.16À18 According to this hypothesis, the electrons are
transferred in this dimer from NADPH via FAD to the FMN
group of monomer 1 and subsquently to the heme group of the
BMP domain of monomer 2 (Figure 1). As described above, the
modular BMR/BMP system offers the advantage that enzymatic
activity of reconstituted BM3 holenenzyme can be assayed by
NADPH-dependent 12-pNCA conversion, while activity of the
BMR and BMP domain alone can be monitored by the reduction
of cytochrome C,4 which shows increased absorption at 550 nm
in its reduced state,39 or the H2O2-dependent 12-pNCA con-
version, respectively (Figure 1).
To facilitate site-selective conjugation of BMR and BMP with
DNA oligonucleotides under mild conditions, we used a pre-
viously reported method,40 which is based on self-labeling fusion
proteins, such as the HaloTag.41 The encoding DNA of the two
subdomains BMP-F87A and BMR was cloned into pEXP-n9
expression plasmids to enable for the C-terminal fusion with a
16113
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
Figure 4. Digestion and activity analysis of DNAÀprotein conjugates. The oligonucleotide moiety of BMP-Halo-aF9 or BMR-Halo-acF9 was digested
with T4-Polymerase. Subsequently, all possible combinations of undigested and digested DNAÀprotein conjugates were prepared, and the enzymatic
activity of these mixtures was tested, as schematically depicted in (A). (B) SDS-PAGE analysis of T4-polymerase digestion. M, broad range marker; lane
1, BMP-Halo-aF9; lane 2, digested BMP-Halo-aF9; lane 3, BMR-Halo-acF9; lane 4, digested BMR-Halo-acF9. (C) The activity of digested/undigested
DNAÀprotein conjugate combinations depicted in (A) was assayed with NADPH/12-pNCA to determine which combination is capable of
reassembling functional holoenzyme. Mixtures were also analyzed with NADPH/cytochrome C to independently determine BMR activity (D) or with
H2O2/12-pNCA to independently determine BMP activity (E). Error bars indicate deviations obtained in three independent experiments. NC denotes
activity measured in a negative control where no protein species were present.
reassembly of functional BM3 holoenzyme. As discussed below
in the context of Figure 6, equimolar mixtures of the two non-
complementary DNAÀprotein conjugates BMP-Halo-aF9 and
BMR-Halo-aF1 did not show any conversion of 12-pNCA. These
results clearly indicate that reassembly of functional BM3
holoenzyme is indeed mediated by and dependent on hybridiza-
tion of the complementary DNA oligonucleotides.
To analyze whether and how conjugation of DNA oligonu-
cleotides affects the activity of the two isolated domains, we
digested the oligonucleotide moiety of BMP-Halo-aF9 and
BMR-Halo-acF9 with T4-Polymerase. This enzyme has an
exonuclease activity in the absence of dNTPs.44 Successful
removal of the conjugated oligonucleotide was confirmed by
SDS-PAGE analysis (Figure 4B). The digested samples were
mixed with equimolar amounts of undigested conjugates to
prepare all possible combinations of binary mixtures of digested/
undigested BMR and BMP domains. These mixtures were
then analyzed with NADPH/12-pNCA to prove that reassemby
of functional holoenzyme is possible only in the case of two
undigested domains (entry 4 in Figure 4C). However, velocity
rates (v0 ≈ 1.5 μM/min) were slightly higher than those observed
in the analogous experiment in Figure 3 (v0 ≈ 0.9 μM/min).
We attribute this increase in activity to changed buffer condi-
tions, likely DTT and/or Mg2+, which are components of the
T4-polymerase digestion buffer (see experimental section
in the Supporting Information). Moreover, activity of BMR in
the mixtures was confirmed by cytochrome C reduction in the
presence of NADPH cofactor (Figure 4D). As expected, all
samples revealed almost identical activity in cytochrome C
reduction, thereby indicating that neither DNA conjugation
nor T4-polymerase digestion interfered with the activity of the
reductase domain. Taking advantage of the peroxide shunt
pathway, the mixtures were also assayed with H2O2/12-pNCA
to independently determine BMP activity (Figure 4E). All
mixture showed conversion of 12-pNCA, thus indicating intact-
ness of the BMP domains. However, notable differences in the
reaction rates were observed. The samples 2 and 3 in which BMP
conjugates were digested showed a higher activity than
those containing intact BMP conjugates (samples 1 and 4,
in Figure 4E). This suggests that the conjugated oligonucleotide
slightly reduces the activity of BMP. Interestingly, the activity of
BMP conjugates seemed to be dependent on the state of the
BMR conjugates. We observed that both free and DNA-con-
jugated BMP-Halo showed slightly higher activity when the
BMR conjugates were digested: Comparison of either samples
3 and 2, or else samples 1 and 4, containing digested and
undigested BMR, respectively, reveal a slightly higher activity
in samples containing digested BMR. Although the increase is
16114
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
Figure 5. Looped DNA scaffold as the template for BM3 reassembly. (A) Schematic drawing of the scaffold containing two addresses (black and red
portions, complementary to BMP-Halo-aF9 and BMR-Halo-aF1) connected by a stem loop (dark blue). The stem loop can be destabilized by
hybridization with complementary effector oligonucleotides (light blue) of different numbers of nucleotides (n). (B) Various scaffolds, preformed from
the stem loop and effectors, were used for assembly of BM3 holoenzyme by hybridization with BMP-Halo-aF9 and BMR-Halo-aF1. The height of the
bars represents the initial rate of 12-pNCA conversion measured. Data in nc denotes a negative control lacking a DNA scaffold; pc1 is a positive control
using the stem loop in the absence of effector oligomers; and pc2 is a positive control as in pc1, where a noncomplementary 51 nt effector was added.
(C) Schematic representation of putative configurations of selected samples.
close to average deviations, it might suggest a mechanistic effect,
such as a conformational change of BMP induced by BMR, which
can occur more effectively in the absence of conjugated oligonucleotide.
Because the above results clearly indicated successful DNA
conjugation of BMR and BMP domains as well as DNA-based
reassembly of functional BM3 holoenzyme, we attempted to
build more complex DNAÀprotein architecture. To this end, a
DNA scaffold was designed in which a stem-loop structure was
inserted between two address sequences, cF1 and cF9, comple-
mentary to conjugates BMR-Halo-aF1 and BMP-Halo-aF9,
respectively (Figure 5). The stem-loop substructure contained
a 11 bp double-stranded stem and an open 23 nucleotide (nt)
loop. We reasoned that the stem loop should be destabilized
upon hybridization with complementary effector oligonucleo-
tides of different length, thereby leading to altered distances of
hybridized protein conjugates, and consequently to a change in
activity of reconstituted holoenzyme. To test this hypothesis,
various scaffolds were preformed from the stem loop and
effectors (PME oligomers in Table S2), and the resulting duplex
structures were used for assembly of BM3 holoenzyme by
hybridization with BMP-Halo-aF9 and BMR-Halo-aF1. Catalytic
activity of reconstituted holoenzyme was measured by determin-
ing the initial rate of 12-pNCA conversion in the presence of
NADPH (Figure 5B). A negative control lacking the DNA
scaffold revealed no activity, as expected. Positive controls, in
which the stem-loop oligomer was used in the absence or the
presence of a noncomplementary effector (pc1 or pc2, respec-
tively, in Figure 5B), showed a high activity. Because initial
velocity rates (v0 ≈ 0.8 μM/min) were similar to that obtained
for the binary superstructures described above (v0 ≈ 0.9 μM/min,
Figure 3), one may conclude that the undisturbed stem-loop
scaffold enables similar effective contacts between the BMR and
BMP domain.
In contrast, scaffolds containing partial duplex regions led
to decreased activity of the holoenzyme superstructures. Short
effectors (n = 21 or 27 nt) had only little influence on holo-
enzyme activity, while longer effectors (n = 33, 39, or 45 nt) led to
significant reduction in enzyme activity. As determined by
F€orster resonance energy transfer (FRET) measurements
(Figure S4, Supporting Information), hybridization with the 39
or 45 nt effectors opens the stem-loop, while shorter effectors (n =
21, 27, or 33 nt) are not capable of stem-loop opening in the
absence of proteins. In conjunction with geometric model
calculations to estimate spatial dimensions of DNA-assembled
constructs (Figure S5, Supporting Information), the observation
that the 39 nt and 45 nt effectors almost completely shut down
holoenzyme activity (Figure 5B) suggests that the linearized stem
loopÀeffector duplex displays sufficient rigidity to effectively
separate BMR and BMP domains. Even though the duplex can
bend and rotate at sites where the dsDNA backbone is nicked or
incomplete, a linear double helix seems favored, presumably due
to stacking interactions and electrostatic repulsion. Interestingly,
the 33 nt effector clearly induced a decrease in holoenzyme
activity (Figure 5B), while it was not able to open the stem-loop
in the absence of proteins (Figure S4B). This observation sug-
gests that protein conjugation weakens the stability of the
adjacent stem loop. To sum, the results obtained with this initial
system clearly demonstrate that designed DNA scaffolds can be
used to adjust physical contact of two interacting protein
domains and thus tune the rate of electron transfer.
We also tested whether the BMR and BMP domains in active
stem-loop DNAÀprotein superstructures can be pulled apart
from each other by hybridization with effector oligomers.
For example, the undisturbed stem-loop superstructure pc1
(Figure 5C) was incubated with the 45 nt effector during 12-
pNCA conversion. Because no clear decrease in activity could be
16115
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
to complete strand displacement and thus release of BMP-Halo-
aF9 and BMP-Halo-aF1 from the template. Because of the
absence of secondary structures within the th-cF1-cF9 and F9-
F1-ct oligomers, the system was expected to react faster than the
stem-loop system. Second-order rate constants of 105À106 MÀ1 sÀ1
have been reported for strand displacement reactions with
linear sequences.46 On the basis of these constants, the strand
displacement system should exhibit reaction time half-lives of
2À20 s, at initial reactant concentrations of 0.5 μM (Figure S7,
Supporting Information). Thus, strand displacement should lead
to detectable signals within the time scale of less than 30 min.
This estimation was experimentally confirmed by analysis of the
strand displacement of the th-cF1-cF9/F9-F1-ct oligomers using
a FRET approach (Figure S6). Indeed, displacement showed a
half life of only a few seconds. We then applied this oligonucelo-
tide system for assembly and disassembly of the BMP- and BMR-
DNA conjugates. It is clearly evident from the green and red
curves in Figure 6 that addition of template th-cF1-cF9 to a
mixture of BMR-Halo-aF1 and BMP-Halo-aF9 led to immediate
formation of functional holoenzyme. After about 10 min, stopper
oligomer F9-F1-cth was added. This led to a marked effect on the
reaction rate, which was not observed in a control reaction where
the noncomplementary oligomer PME 38 was added (green vs
red curve in Figure 6). Addition of the stopper immediately
decreased the slope of reaction rate (green curve), while the
noncomplementary control oligonucleotide did not affect ho-
loenzyme activity of reassembled BMR-BMP (red curve). How-
ever, no abrupt stopping occurred upon addition of the
complementary stopper (green curve), presumably due to a
comparably slow dissociation kinetics and/or attractive interac-
tions of BMP and BMR domains. Nonetheless, these results
clearly indicate the functionality of assembly and disassembly of
BM3 holoenzyme by means of designed DNA templates.
Figure 6. Reversible assembly of BM3 domains. (A) Schematic drawing
of the DNA-mediated assembly and disassembly of BM3 holoenzyme by
strand displacement. Template th-cF1-cF9 can bind the two BM3
domains BMP-Halo-aF9 and BMR-Halo-aF1 with its red and green
portions, respectively. The blue colored single-stranded 10 nt toehold is
used to initiate strand displacement by hybridization with oligomer F9-
F1-cth, which is fully complementary to template th-cF1-cF9. The
formation of the duplex leads to release of BMP-Halo-aF9 and BMP-
Halo-aF1. (B) Plot of 12-pNCA conversion measured by absorbance at
405 nm versus time. To three identical samples containing BMP-Halo-
aF9, BMP-Halo-aF1, NADPH, and 12-pNCA was added either template
th-cF1-cF9 (red and green curves) or a noncomplementary oligonucleo-
tide (black curve) at time point (+). Note that addition of template leads
to immediate formation of active enzyme. At time point (#), either
stopper oligomer F9-F1-cth (green curve) or a noncomplementary
control oligonucleotide (red and black curves) was added. Note that
stopper F9-F1-cth led to a decrease in 12-pNCA conversion (green
curve), while the noncomplementary control strand did not affect
holoenzyme activity (red curve).
’ CONCLUSIONS
We here demonstrated that DNA-based assembly of the two
subdomains of BMP and BMR can be harnessed to create a
tunable monooxygenase system. BMP and BMR domains con-
jugated with synthetic oligonucleotides can be reversibly as-
sembled and disassembled by aid of appropriate DNA scaffolds.
DNA-based protein assembly has already been used to create
biocatalytic cascades of consecutively working enzymes, such as
glucose oxidase (GOx) and horseradish peroxidase (HRP).30,33À35
This approach to functional coupling of enzymes aims at improving
the reaction efficiency by limiting the diffusion distance between
catalytic centers for the intermediate reactant, such as H2O2 in the
case of the GOx/HRP system. In contrast to this approach, we here
describe a system that enables control of enzyme activity by a DNA-
dependent mechanism. To the best of our knowledge, this is the first
study that demonstrates the directional incorporation of two
delicate protein domains in functional DNA scaffolds to enable
control of enzymatic activity by tunable approximation of two
subdomains. Resulting changes in distance between the redox
centers embedded in the BMP and BMR domains in turn regulate
efficiency of electron transfer, and thus enzymatic activity of the
reconstituted holoenzyme. It should be noted that more rigid
scaffolds, such as DNA double-crossover tiles or origami super-
structures, may serve even better as molecular pegboard and ruler to
study distance dependency of transport processes in reassembled
multidomain proteins. However, methods to synthesize such
observed, we analyzed the stem-loop system in the absence of
proteins by measuring the kinetics of hybridization with the
longest effector using a FRET approach (Figure S6, Supporting
Information). Although this model reaction indicated switch-
ability of our stem-loop system, we observed very slow hybridiza-
tion kinetics, which is in good agreement with a previous study
on DNA hairpins.45 It was therefore evident that the loop
opening did not proceed sufficiently fast to enable measurement
of changes in the enzymatic conversion of pNCA.
To circumvent the obstacles of the above-described stem-loop
system, we explored another DNA design for demonstration of
reversible domain assembly. To this end, the linear single-
stranded template th-cF1-cF9 was chosen (Figure 6), which
contains the two binding sites for BMR-Halo-aF1 and BMP-
Halo-aF9 in addition to a 10 nt toehold sequence. The toehold
initiates hybridization with stopper oligomer F9-F1-cth, which is
fully complementary to template th-cF1-cF9. This process leads
16116
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
DNAÀprotein superstructures on a preparative scale are not yet
available.
founded by the European Comission, and the Zentrum f€ur
Angewandte Chemische Genomik (ZACG). C.-H.K. acknowl-
edges support through the International Max-Planck Research
School in Chemical Biology, Dortmund, and a student fellow-
ship from Deutscher Akademischer Austauschdienst (DAAD).
One important technical feature of the here reported system
stems from the fact that both domains not only display a specific
activity when brought together but that the domains can be
interrogated independently by convenient analytical assays. This
aspect clearly exceeds the concept of split enzymes, where the
separated components have no stand-alone activity,32 and it will
be of great importance for future development of advanced BM3-
based systems. It has already been emphasized that the BMP
domain can be adopted to the specific conversion of a broad
range of substrates by directed evolution,3 and, likewise, engi-
neering of BMR enables control over the cofactor, which, for
instance, can be switched from NADPH to NADH.47 Hence,
immediate perspectives of the modular BMR/BMR assembly
system arise from the fact that it can be used for ready exploration
of combinatorial assemblies of different P450 and reductase
domains to find suitable combinations. This approach will
significantly reduce the amount of cloning. Moreover, the control
of spatial configuration of the two domains by DNA scaffolds
adds an additional degree of freedom, because nucleic acid
hybridization is highly specific and its kinetics can be easily
controlled.46 One may thus foresee long-term applications
in which switchable DNA-BMP/BMR constructs are used for
drug delivery. For instance, constructs could be designed to be
triggered by micro RNAs, which are endogeneously expressed
during development and disease.48 Binding to nucleic acid
targets could switch on the construct’s activity, thereby enabling
localized enzymatic activation of prodrugs. In view of these
perspectives, we anticipate that the concept demonstrated here
could open up ways for the development of novel screening
systems or responsive biocatalysts for therapeutic applications.
’ REFERENCES
(1) Munro, A. W.; Daff, S.; Coggins, J. R.; Lindsay, J. G.; Chapman,
S. K. Eur. J. Biochem. 1996, 239, 403–409.
(2) Warman, A. J.; Roitel, O.; Neeli, R.; Girvan, H. M.; Seward, H. E.;
Murray, S. A.; McLean, K. J.; Joyce, M. G.; Toogood, H.; Holt, R. A.;
Leys, D.; Scrutton, N. S.; Munro, A. W. Biochem. Soc. Trans. 2005,
33, 747–753.
(3) Jung, S. T.; Lauchli, R.; Arnold, F. H. Curr. Opin. Biotechnol.
2011, 22, in press; DOI 10.1016/j.copbio.2011.02.008.
(4) Narhi, L. O.; Fulco, A. J. J. Biol. Chem. 1986, 261, 7160–7169.
(5) Li, Q. S.; Schwaneberg, U.; Fischer, P.; Schmid, R. D. Chemistry
2000, 6, 1531–1536.
(6) Otey, C. R.; Bandara, G.; Lalonde, J.; Takahashi, K.; Arnold, F. H.
Biotechnol. Bioeng. 2006, 93, 494–499.
(7) Carmichael, A. B.; Wong, L. L. Eur. J. Biochem. 2001, 268,
3117–3125.
(8) Lussenburg, B. M.; Babel, L. C.; Vermeulen, N. P.; Commandeur,
J. N. Anal. Biochem. 2005, 341, 148–155.
(9) Seifert, A.; Vomund, S.; Grohmann, K.; Kriening, S.; Urlacher,
V. B.; Laschat, S.; Pleiss, J. ChemBioChem 2009, 10, 853–861.
(10) Dietrich, J. A.; Yoshikuni, Y.; Fisher, K. J.; Woolard, F. X.;
Ockey, D.; McPhee, D. J.; Renninger, N. S.; Chang, M. C.; Baker, D.;
Keasling, J. D. ACS Chem. Biol. 2009, 4, 261–267.
(11) Rentmeister, A.; Arnold, F. H.; Fasan, R. Nat. Chem. Biol. 2009,
5, 26–28.
(12) Reinen, J.; Kalma, L. L.; Begheijn, S.; Heus, F.; Commandeur,
J. N.; Vermeulen, N. P. Xenobiotica 2011, 41, 59–70.
(13) Sawayama, A. M.; Chen, M. M.; Kulanthaivel, P.; Kuo, M. S.;
Hemmerle, H.; Arnold, F. H. Chemistry 2009, 15, 11723–11729.
(14) van Leeuwen, J. S.; Vredenburg, G.; Dragovic, S.; Tjong, T. F.;
Vos, J. C.; Vermeulen, N. P. Toxicol. Lett. 2011, 200, 162–168.
(15) Shapiro, M. G.; Westmeyer, G. G.; Romero, P. A.; Szablowski,
J. O.; Kuster, B.; Shah, A.; Otey, C. R.; Langer, R.; Arnold, F. H.; Jasanoff,
A. Nat. Biotechnol. 2010, 28, 264–270.
(16) Neeli, R.; Girvan, H. M.; Lawrence, A.; Warren, M. J.; Leys, D.;
Scrutton, N. S.; Munro, A. W. FEBS Lett. 2005, 579, 5582–5588.
(17) Kitazume, T.; Haines, D. C.; Estabrook, R. W.; Chen, B.;
Peterson, J. A. Biochemistry 2007, 46, 11892–11901.
(18) Girvan, H. M.; Dunford, A. J.; Neeli, R.; Ekanem, I. S.;
Waltham, T. N.; Joyce, M. G.; Leys, D.; Curtis, R. A.; Williams, P.;
Fisher, K.; Voice, M. W.; Munro, A. W. Arch. Biochem. Biophys. 2010,
507, 75–85.
’ ASSOCIATED CONTENT
S
Supporting Information. A detailed description of clon-
b
ing and heterologous expression of BMR and BMP fusion
proteins, 12-pNCA synthesis, oligonucleotide sequences, synth-
esis and characterization of DNAÀprotein conjugates, enzyme
assays and determination of hybridization kinetics, as well as
model estimations on the spatial configuration of DNA-linked
BM3 subdomains. This material is available free of charge via the
Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
(19) Boddupalli, S. S.; Oster, T.; Estabrook, R. W.; Peterson, J. A.
J. Biol. Chem. 1992, 267, 10375–10380.
(20) Sevrioukova, I.; Truan, G.; Peterson, J. A. Arch. Biochem.
Biophys. 1997, 340, 231–238.
(21) Li, Q. S.; Ogawa, J.; Shimizu, S. Biochem. Biophys. Res. Commun.
2001, 280, 1258–1261.
(22) Ortiz de Montellano, P. R.; De Voss, J. J. Nat. Prod. Rep. 2002,
19, 477–493.
Corresponding Author
christof.niemeyer@tu-dortmund.de
Present Addresses
†Institute of Chemistry, Academia Sinica, 128 Academia Road
Sec. 2, Nankang, Taipei 11529, Taiwan.
(23) Seeman, N. C. Nature 2003, 421, 427–431.
(24) Simmel, F. C.; Dittmer, W. U. Small 2005, 1, 284–299.
(25) Lin, C.; Liu, Y.; Yan, H. Biochemistry 2009, 48, 1663–1674.
(26) Niemeyer, C. M. Angew. Chem., Int. Ed. 2010, 49, 1200–1216.
(27) Gianneschi, N. C.; Ghadiri, M. R. Angew. Chem., Int. Ed. 2007,
46, 3955–3958.
(28) Saghatelian, A.; Guckian, K. M.; Thayer, D. A.; Ghadiri, M. R.
J. Am. Chem. Soc. 2003, 125, 344–345.
(29) Fruk, L.; Niemeyer, C. M. Angew. Chem., Int. Ed. 2005, 44,
2603–2606.
’ ACKNOWLEDGMENT
We thank Andreas Arndt for assistance in protein expression,
Barbara Sacca for help in the design of DNA nanostructures,
Rebecca Meyer for assistance in the purification and modification
of oligonucleotides, and Marc Skoupi and Katja Straub for contribu-
tions in cloning. This work was supported in part by the Deutsche
Forschungsgemeinschaft (DFG) through project NI 399/10,
the project SMD in the course of FP7-NMP-2008-SMALL-2,
16117
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118

Journal of the American Chemical Society
ARTICLE
(30) Wilner, O. I.; Weizmann, Y.; Gill, R.; Lioubashevski, O.; Free-
man, R.; Willner, I. Nat. Nanotechnol. 2009, 4, 249–254.
(31) Takeda, S.; Tsukiji, S.; Ueda, H.; Nagamune, T. Org. Biomol.
Chem. 2008, 6, 2187–2194.
(32) Oltra, N. S.; Bos, J.; Roelfes, G. ChemBioChem 2010,
11, 2255–2258.
(33) Niemeyer, C. M.; Koehler, J.; Wuerdemann, C. ChemBioChem
2002, 3, 242–245.
(34) M€uller, J.; Niemeyer, C. M. Biochem. Biophys. Res. Commun.
2008, 377, 62–67.
(35) Wilner, O. I.; Shimron, S.; Weizmann, Y.; Wang, Z. G.; Willner,
I. Nano Lett. 2009, 9, 2040–2043.
(36) Oliver, C. F.; Modi, S.; Sutcliffe, M. J.; Primrose, W. U.; Lian,
L. Y.; Roberts, G. C. Biochemistry 1997, 36, 1567–1572.
(37) Schwaneberg, U.; Schmidt-Dannert, C.; Schmitt, J.; Schmid,
R. D. Anal. Biochem. 1999, 269, 359–366.
(38) Li, Q. S.; Schwaneberg, U.; Fischer, M.; Schmitt, J.; Pleiss, J.;
Lutz-Wahl, S.; Schmid, R. D. Biochim. Biophys. Acta 2001,
1545, 114–121.
(39) Mahler, H. R. DPNH cytochrome c reductase (animal).
Methods in Enzymology, 2nd ed.; Academic Press: New York, 1955; pp
688À693.
(40) Sacca, B.; Meyer, R.; Erkelenz, M.; Kiko, K.; Arndt, A.;
Schroeder, H.; Rabe, K. S.; Niemeyer, C. M. Angew. Chem., Int. Ed.
2010, 49, 9378–9383.
(41) Los, G. V.; Wood, K. Methods Mol. Biol. 2007, 356, 195–208.
(42) Black, S. D.; Martin, S. T. Biochemistry 1994, 33, 12056–12062.
(43) Niemeyer, C. M.; Sano, T.; Smith, C. L.; Cantor, C. R. Nucleic
Acids Res. 1994, 22, 5530–5539.
(44) Huang, W. M.; Lehman, I. R. J. Biol. Chem. 1972, 247, 3139–
3146.
(45) Green, S. J.; Lubrich, D.; Turberfield, A. J. Biophys. J. 2006,
91, 2966–2975.
(46) Turberfield, A. J.; Mitchell, J. C.; Yurke, B.; Mills, A. P., Jr.;
Blakey, M. I.; Simmel, F. C. Phys. Rev. Lett. 2003, 90, 118102.
(47) Neeli, R.; Roitel, O.; Scrutton, N. S.; Munro, A. W. J. Biol. Chem.
2005, 280, 17634–17644.
(48) Shruti, K.; Shrey, K.; Vibha, R. Biochem. Biophys. Res. Commun.
2011, 407, 445–449.
16118
dx.doi.org/10.1021/ja204993s |J. Am. Chem. Soc. 2011, 133, 16111–16118