Turning a riboflavin-binding protein into a self-sufficient monooxygenase by cofactor redesign

Article abstract of DOI:10.1039/c1cc14039f

By cofactor redesign, self-sufficient monooxygenases could be prepared. Tight binding of N-alkylated flavins to riboflavin-binding protein results in the creation of artificial flavoenzymes capable of H₂O ₂-driven enantioselective sulfoxidations. By altering the flavin structure, opposite enantioselectivities could be achieved, in accordance with the binding mode predicted by in silico flavin-protein docking of the unnatural flavin cofactors. The study shows that cofactor redesign is a viable approach to create artificial flavoenzymes with unprecedented activities.

Full text of DOI:10.1039/c1cc14039f

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
ChemComm
Cite this: Chem. Commun., 2011, 47, 11050–11052
www.rsc.org/chemcomm
COMMUNICATION
Turning a riboflavin-binding protein into a self-sufficient monooxygenase
by cofactor redesignw
a
b
Gonzalo de Gonzalo,z Christian Smit,z Jianfeng Jin,a Adriaan J. Minnaardb and
Marco W. Fraaije*a
Received 6th July 2011, Accepted 25th August 2011
DOI: 10.1039/c1cc14039f
By cofactor redesign, self-sufficient monooxygenases could be
prepared. Tight binding of N-alkylated flavins to riboflavin-
binding protein results in the creation of artificial flavoenzymes
capable of H2O2-driven enantioselective sulfoxidations. By
altering the flavin structure, opposite enantioselectivities could
be achieved, in accordance with the binding mode predicted by in
silico flavin-protein docking of the unnatural flavin cofactors.
The study shows that cofactor redesign is a viable approach to
create artificial flavoenzymes with unprecedented activities.
monooxygenases. By flavin cofactor redesign several artificial
flavoenzymes were created that function as peroxide-driven
monooxygenases, capable of enantioselective sulfoxidations.
Flavoprotein monooxygenases represent a major group within
the superfamily of monooxygenases.1b These biocatalysts exploit
the chemical versatility of the flavin cofactor to perform selective
oxygenations. A key enzyme intermediate in the catalytic cycle is
the so-called 4a-hydroperoxyflavin (Scheme 1, bottom), formed
by reduction of the flavin, typically at the expense of NAD(P)H,
and subsequent reaction with dioxygen. While flavoprotein
monooxygenases are able to form and stabilize such inter-
mediates,6 other flavoproteins are unable to do so and it is
known that peroxyflavins quickly decay when free in solution.7
The details on how enzymes are able to tune the stabilisation of
these specific enzyme intermediates still remain obscure.8
Monooxygenases perform selective oxidations of organic substrates
under environmentally benign conditions.1 Due to this and their
chemo-, regio- and/or enantioselectivity, monooxygenases are
regarded as extremely valuable biocatalysts. Nonetheless, the
applicability of monooxygenases is often hampered by their
dependence on nicotinamide cofactors (NAD(P)H). These cofac-
tors are required to deliver electrons to the oxidizing cofactor, e.g.
heme in P450 monooxygenases and flavin in flavoprotein mono-
oxygenases, enabling the enzymes to utilize dioxygen for oxygena-
tions. Several strategies to circumvent the stoichiometric use of
nicotinamide cofactors have been reported.2 Of these approaches,
the use of additional biocatalysts in combination with a sacrificial
cosubstrate has been the most effective.3 Nevertheless, the use of
cofactor-independent monooxygenases is far less complicated, as
has been demonstrated for a few monooxygenases. Thus, light can
be used to reduce cytochrome P450 and flavoprotein monooxy-
genases.4 Cytochrome P450 monooxygenases have also been
shown to operate at the expense of peroxides, thereby bypassing
the cofactor-dependent heme reduction and subsequent oxygenation
of the heme iron center.5 Whereas these methodologies are
viable, they are not very efficient. We here report on a novel
concept for the design of artificial self-sufficient flavoprotein
Analogous to the so-called ‘shunt’ pathway in cytochrome P450
monooxygenases,5 we have studied several flavoproteins, a.o.
phenylacetone monooxygenase, for oxygenation activity when
using hydrogen peroxide.9 However, we could never observe
any oxygenating activity using this approach. In order to develop
a flavoprotein monooxygenase that operates independent from
NAD(P)H, we then focused at flavin cofactor redesign.
a Laboratory of Biochemistry, Groningen Biomolecular Sciences and
Biotechnology Institute, University of Groningen, Nijenborgh 4,
9747 AG, Groningen, The Netherlands. E-mail: m.w.fraaije@rug.nl;
Fax: +31 503634165, +31 503634345
b Department of Bio-Organic Chemistry, Stratingh Institute for
Chemistry, University of Groningen, Nijenborgh 4, 9747 AG,
Groningen, The Netherlands
w Electronic supplementary information (ESI) available: All experi-
mental procedures as well as synthesis and characterization of flavins
2–6, UV-vis of free and bound compound 3 and NMR spectra are
included. See DOI: 10.1039/c1cc14039f
Scheme 1 Mechanism of flavin-based oxygenations. The solid arrows
indicate the mechanism operating in natural flavoprotein mono-
oxygenases, the dashed arrow indicates the ‘shunt’ pathway explored in
this study by employing alkylated flavins 2–6.
z Contributed equally.
c
11050 Chem. Commun., 2011, 47, 11050–11052
This journal is The Royal Society of Chemistry 2011

View Article Online
For the elucidating mechanistic details of flavoenzymes and to
explore flavin-based catalysis, numerous flavin derivatives have
been prepared and studied in the last few decades.10 From these
studies, it has become clear that, while natural flavin cofactors are
unstable in their peroxy form, N5-alkylated flavins can form
remarkably stable 4a-peroxyflavins upon reaction with hydrogen
peroxide. This was first shown by Bruice and co-workers by using
alkylated lumiflavins.11 Subsequent studies have shown that
similar N-alkylated flavins are powerful oxidizing catalysts.12
Inspired by these findings, we selected riboflavin-binding
protein (RfBP) from Gallus gallus (chicken), which has no catalytic
activity as such, but it has the capacity to bind avidly riboflavin
(1, vitamin B2).13 Rf BP serves as a riboflavin storage protein
and can be easily isolated from eggs while established protocols
are available to prepare the protein devoid of 1.14 Interest-
ingly, it has been shown that this apo form of Rf BP binds a
wide range of flavin derivatives in the nanomolar range.15 For
this study, a concise library of alkylated riboflavin and lumi-
flavin derivatives was prepared (Scheme 2). Alongside, apo
Rf BP was prepared and reconstituted with flavin derivatives.
Incorporation of the reduced forms of these flavins into Rf BP
was successful in all cases. This could be concluded from visual
inspection as, upon reconstituting the protein with an alky-
lated coloured flavin derivative and excessive dialysis, the
protein remained coloured. Additionally, the binding of flavin
2 was determined by fluorescence titrations which confirmed
tight binding: Kd = 53 nM. These findings are in line with the
reported high affinity of Rf BP for riboflavin and alkylated
derivatives which are all in the nM range.15
Table 1 Enantioselective peroxide-driven sulfoxidation of prochiral
sulfide by employing RfBP complexed with modified flavinsa
Entry
R1
R2
Flavin
Protein
Conv.b (%)
Eec (%)
1
2
3
4
5
6
7
8
Ph
p-Tol
Bz
Cy
Ph
p-Tol
Ph
Ph
Bz
Cy
Ph
p-Tol
Ph
Bz
Me
Me
Me
Me
Me
Me
Et
Me
Me
Me
Me
Me
Me
Me
Et
2
2
2
2
3
3
3
4
4
4
5
5
6
6
6
6
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
ꢀ
37
44
52
55
40
45
44
48
40
44
43
46
43
60
49
53
35
15
14
22 (R)
30 (R)
16 (R)
10 (R)
22 (S)
25 (S)
15 (S)
25 (S)
10 (S)
25 (S)
12 (S)
16 (S)
8 (R)
9
10
11
12
13
14
15
16
17
18
19
36 (R)
27 (R)
33 (R)
r3
Bz
Bu
Ph
Ph
Ph
Et
Me
Me
Me
3
None
1
ꢀ
r3
+
r3
a
5.0 mM substrate, 50 mM RfBP catalyst, reaction time 24 h.
Determined by GC. Measured by GC or HPLC.
b
c
Interestingly, the use of the three N-alkylated lumiflavin deriva-
tives, flavins 3–5, yielded similar yields and optical purities of
sulfoxides but, with in most cases, an excess of the (S)-enantio-
mers. When using the demethylated form of flavin 4 (1,3,5-
triethylalloxazine, flavin 6), again a competent peroxide-driven
monooxygenase is formed which forms preferentially (R)-sulfoxides.
In all reactions the amount of sulfone formed was very limited
(B5%). For all artificial flavoenzymes, the best enantioselec-
tivities were obtained for the aromatic sulfides. Furthermore,
the results indicate that by increasing the size of the alkyl
moieties of the modified lumiflavins the enantioselectivity is
decreased.
After reconstituting apo RfBP with these flavins and verifica-
tion of tight binding, the derived synthetic flavoproteins were
tested for their ability to perform peroxide-driven enantioselective
sulfoxidations (Table 1). We were pleased to observe that the
created artificial flavoproteins were active as catalysts and even
displayed enantioselectivity. Using N5-ethylated riboflavin (flavin
2) bound to Rf BP, all tested sulfides could be converted (37–55%
conversion), yielding the (R)-sulfoxides in excess (10–30% ee).
As a control, the reactions were performed with each flavin
derivative in the absence of apo RfBP. Without exception, this
led to the formation of racemic sulfoxide (see ESIw and entry 18).
Progress analysis of the oxidation reaction by RfBP complexed
with flavin 3 revealed that the enantioselective outcome was best
in the early phase of conversion. This appears to be due to the
chemical oxidation of sulfides by hydrogen peroxide while the
flavin-mediated oxidation gradually decreases in time. Finally,
the oxygenating activity of 1-RfBP with hydrogen peroxide was
studied, as shown in entry 19, but expectedly the results were
similar to the background reaction using only hydrogen peroxide
(entry 18). Concluding, the results clearly demonstrate that RfBP
reconstituted with artificial flavins acts as a hydrogen-peroxide
driven enantioselective monooxygenase.
Intriguingly, the observed enantioselectivity of the created
flavo-enzymes is strongly dependent on the type of flavin
cofactor employed. This is a very attractive feature as it allows
the selective preparation of both enantiomers of the product
just by changing the type of flavin cofactor. In order to
provide a rationale for this, all N5-alkylated flavin derivatives
studied were docked in the crystal structure of Rf BP (Fig. 1).
This in silico study provides insight in the origin of the
enantioselectivity and cofactor modifications that steer this.
Flavin 2 was found to be bound very similar to 1. The binding
is for a large part achieved by stacking of the isoalloxazine moiety
Scheme 2 Selective sulfoxidation of prochiral sulfides by employing
RfBP complexed with modified flavins 2–6.
c
This journal is The Royal Society of Chemistry 2011
Chem. Commun., 2011, 47, 11050–11052 11051

View Article Online
in a recombinant manner.18 Because there are hardly any other
proteins known that bind effectively riboflavin,19 we are
currently exploring proteins that bind the phosphorylated
form of riboflavin, FMN. A plethora of FMN-containing
enzymes is known, providing a good starting point for design
of new flavoenzymes by cofactor redesign.
Notes and references
1 (a) F. Hollman, I. W. C. E. Arends, K. Buhler, A. Schallmey and
¨
¨
B. Buhler, Green Chem., 2011, 13, 226; (b) D. E. Torres Pazmino,
M. Winkler, A. Glieder and M. W. Fraaije, J. Biotechnol., 2010,
146, 9; (c) V. B. Urlacher and R. D. Schmid, Curr. Opin. Chem.
Biol., 2006, 10, 156.
2 (a) F. Hollman, I. W. C. E. Arends and K. Buhler, ChemCatChem,
¨
2010, 2, 762; (b) S. H. Lee, D. H. Nam and C. B. Park, Adv. Synth.
Catal., 2009, 351, 2589; (c) F. Hollman, K. Hofstetter and A. Schmid,
Trends Biotechnol., 2006, 24, 163; (d) G. de Gonzalo, G. Ottolina,
M. W. Fraaije and G. Carrea, Chem. Commun., 2005, 3724.
3 (a) D. E. Torres Pazmino, R. Snajdrova, B.-J. Braas, M. Ghrobial,
M. D. Mihovilovic and M. W. Fraaije, Angew. Chem., Int. Ed.,
2008, 47, 2275; (b) V. I. Tishkov and V. O. Popov, Biomol. Eng.,
2006, 23, 89; (c) C.-H. Wong and G. M. Whitesides, J. Am. Chem.
Soc., 1981, 103, 4890.
Fig. 1 (a) The electrostatic surface of RfBP is shown with ethylated
flavin 2 in the riboflavin binding pocket. (b) Close-up view of the bound
flavins 3–6 as obtained by docking in RfBP. The comparative binding
energy scores for docking flavins 2–6 (ꢀ115.8, ꢀ87.9, ꢀ90.3, ꢀ100.3, and
ꢀ98.0, resp.) confirm that all flavins display a good affinity for RfBP.
4 See for example: F. Hollman, A. Taglieber, F. Schulz and
M. T. Reetz, Angew. Chem., Int. Ed., 2007, 46, 1.
5 (a) K. Matsuura, T. Tosha, S. Yoshioka, S. Takahashi, K. Ishimori
and I. Morishima, Biochem. Biophys. Res. Commun., 2004, 323, 1209;
(b) H. Joo, Z. Lin and F. H. Arnold, Nat. Biotechnol., 1999, 17, 670.
6 (a) J. R. Cashman, Biochem. Biophys. Res. Commun., 2005,
338, 599; (b) D. M. Ziegler, Drug Metab. Rev., 2002, 34, 503.
7 V. Massey, J. Biol. Chem., 1994, 269, 22459.
8 (a) A. Mattevi, Trends Biochem. Sci., 2006, 31, 276; (b) V. Massey,
Biochem. Soc. Trans., 2000, 28, 283.
9 Unpublished results.
between tyrosine and tryptophan residues. This p–p stacking
interaction is also present in the other docked flavins. However,
surprisingly, flavins 3–5 bind in such a way that the alloxazine
moiety is rotated by 1801. In this way, these flavins are positioned
with the N5-alkyl group in a pocket that is normally occupied by
the ribityl chain of riboflavin. Such an inverse binding mode may
well explain the observed opposite enantioselectivity. Yet another
mode of binding is predicted for flavin 6. In this case, the
pyrimidine moiety of the alloxazine ring is pointing inwards while
the N5-ethyl group still occupies the same binding pocket as
observed for flavin 2. This is in line with the observed similarity in
enantioselectivities for flavins 2 and 6 (Table 1).
10 (a) Y. Imada, T. Kitagawa, T. Ohno, H. Iida and T. Naota, Org.
Lett., 2010, 12, 32; (b) V. Mojr, V. Herzig, M. Budesı
R. Cibulka and T. Kraus, Chem. Commun., 2010, 46, 7599;
´
nsky´ ,
(c) H. Schamaderer, P. Hilgers, R. Lechner and B. Konig, Adv.
¨
Synth. Catal., 2009, 351, 163; (d) J. Svoboda, H. Schamaderer and
B. Konig, Chem.–Eur. J., 2008, 14, 1854; (e) F. G. Gelalcha, Chem.
¨
Rev., 2007, 107, 3338.
11 (a) C. Kemal, T. W. Chan and T. C. Bruice, Proc. Natl. Acad. Sci.
U. S. A., 1977, 74, 405; (b) C. Kemal and T. C. Bruice, Proc. Natl.
Acad. Sci. U. S. A., 1976, 73, 995.
12 Some examples: (a) C. Smit, M. W. Fraaije and A. J. Minnaard,
J. Org. Chem., 2008, 73, 9482; (b) Y. Imada, H. Iida,
S.-I. Murahashi and T. Naota, Angew. Chem., Int. Ed., 2005,
This study demonstrates that by cofactor redesign, novel
flavin-containing biocatalysts can be generated. By introdu-
cing peroxide-reactive flavins into a protein, it becomes cata-
lytically active while the environment of the flavin binding
cavity renders the protein selective in substrate binding, thus
creating a fully self-sufficient artificial flavoprotein mono-
oxygenase. It is worth noting that this approach of enzyme
engineering does not depend on the availability of an active
enzyme as a starting point. RfBP, with riboflavin bound, does
not display any catalytic activity. Replacing the natural ribo-
flavin cofactor by a mildly modified riboflavin (N5-ethylribo-
flavin) was found to be enough to create a peroxide-driven
oxidative catalyst. Both components of such artificial flavoen-
zymes are relatively easy to obtain: Rf BP can be obtained
from chicken eggs in huge quantities while riboflavin is
relatively cheap and its alkylation is easy. We could also show
that altering the enantioselective behaviour of the biocatalyst
is dependent on the employed flavin derivative. By this it was
possible to obtain both enantiomers of a set of sulfoxides,
albeit with moderate optical purities values. As has been
shown for other artificial enzymes, e.g. artificial metallo-
enzymes,16 tailoring by protein engineering is typically re-
quired for achieving highly (enantio)selective biocatalysts.17
Unfortunately, this is currently not an option when using
Rf BP as host protein because this protein cannot be produced
´
44, 1704; (c) A. A. Linden, N. Hermanns, S. Otts, L. Kruger and
¨
J.-E. Backvall, Chem.–Eur. J., 2005, 11, 112.
¨
13 (a) H. L. Monaco, EMBO J., 1997, 16, 1475; (b) F. Muller and
¨
W. J. H. van Berkel, in Chemistry and Biochemistry of Flavoenzymes,
ed. F. Muller, CRC Press, Boca Raton, 1991, vol. 1, p. 261.
¨
14 (a) M. H. Hefti, J. Vervoort and W. J. H. van Berkel, Eur. J.
Biochem., 2003, 270, 4227; (b) H. M. Farrell, Jr., M. F. Mallette,
E. G. Buss and C. O. Clagett, Biochim. Biophys. Acta, 1969, 194, 433.
15 (a) A. Wessiak, L. M. Schopfer, L. Yuan, T. C. Bruice and V. Massey,
Proc. Natl. Acad. Sci. U. S. A., 1984, 81, 4246; (b) J. Becvar and
G. Palmer, J. Biol. Chem., 1982, 157, 5607; (c) C. Walsh, J. Fisher,
R. Spencer, D. W. Graham, W. T. Ashton, J. E. Brown, R. D. Brown
and E. F. Rogers, Biochemistry, 1978, 17, 1942.
16 (a) P. J. Deuss, R. den Heeten, W. Laan and P. C. J. Kamer,
Chem.–Eur. J., 2011, 17, 4680; (b) M. Creus and T. R. Ward, Org.
Biomol. Chem., 2007, 5, 1835; (c) M. Ohashi, T. Koshiyama,
T. Ueno, M. Yanase, H. Fujii and Y. Watanabe, Angew. Chem.,
Int. Ed., 2003, 42, 1005.
17 C. M. Thomas and T. R. Ward, Chem. Soc. Rev., 2005, 34, 337.
18 P. Pattanaik, P. Sooryanarayana, P. R. Adiga and
S. S. Visweswariah, Eur. J. Biochem., 1998, 258, 411.
19 In order to vary the RfBP properties, we have explored RfBP from
Coturnix japonica (Japanese quail) because the protein sequence of
the respective Rf BP suggests several amino acid substitutions close
to the flavin binding pocket. Unfortunately, this did not yield
better results (see ESIw).
c
11052 Chem. Commun., 2011, 47, 11050–11052
This journal is The Royal Society of Chemistry 2011