Engineering of CYP106A2 for steroid 9α- and 6β-hydroxylation

Article abstract of DOI:10.1016/j.steroids.2017.01.005

CYP 106A2 from Bacillus megaterium ATCC 13368 has been described as a 15β-hydroxylase showing also minor 11α-, 9α- and 6β-hydroxylase activity for progesterone conversion. Previously, mutant proteins with a changed selectivity towards 11α-OH-progesterone have already been produced. The challenge of this work was to create mutant proteins with a higher regioselectivity towards hydroxylation at positions 9 and 6 of the steroid molecule. 9α-hydroxyprogesterone exhibits pharmaceutical importance, because it is a useful intermediate in the production of physiologically active substances which possess progestational activity. Sixteen mutant proteins were selected from a library containing mutated proteins created by a combination of site-directed and saturation mutagenesis of active site residues. Four mutant proteins out of these catalyzed the conversion of progesterone to 9α-OH-progesterone as a main product. For further optimization site-directed mutagenesis was performed. The introduction of seven mutations (D217V, A243V, A106T, F165L, T89N, T247V or T247W) into these four mutant proteins led to 28 new variants, which were also used for an in vivo conversion of progesterone. The best mutant protein, F165L/A395E/G397V, showed a ten-fold increase in the selectivity towards progesterone 9α-hydroxylation compared with the wild type CYP106A2. Also 6β-OH-progesterone is a pharmaceutically important compound, especially as intermediate for the production of drugs against breast cancer. For the rational design of mutant proteins with 6β-selectivity, docking of the 3D-structure of CYP106A2 with progesterone was performed. The introduction of three mutations (T247A, A243S, F173A) led to seven new mutant proteins. Clone A243S showed the greatest improvement in 6β-selectivity being more than ten-fold. Finally, an in vivo conversion of 11-deoxycorticosterone (DOC), testosterone and cortisol with the best five mutant proteins displaying 9α- or 6β-hydroxylation, respectively, of progesterone was performed to investigate whether the introduced mutations also effected the conversion of other substrates.

Full text of DOI:10.1016/j.steroids.2017.01.005

Accepted Manuscript
Engineering of CYP106A2 for steroid 9α- and 6β-hydroxylation
Julia Nikolaus, Kim Thoa Nguyen, Cornelia Virus, Jan L. Riehm, Michael
Hutter, Rita Bernhardt
PII:
S0039-128X(17)30015-6
http://dx.doi.org/10.1016/j.steroids.2017.01.005
STE 8076
DOI:
Reference:
Steroids
To appear in:
Received Date:
Revised Date:
Accepted Date:
23 September 2016
7 January 2017
17 January 2017
Please cite this article as: Nikolaus, J., Nguyen, K.T., Virus, C., Riehm, J.L., Hutter, M., Bernhardt, R., Engineering
of CYP106A2 for steroid 9α- and 6β-hydroxylation, Steroids (2017), doi: http://dx.doi.org/10.1016/j.steroids.
2017.01.005
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Engineering of CYP106A2 for steroid 9α- and 6β-hydroxylation
Julia Nikolaus1, Kim Thoa Nguyen1, Cornelia Virus1, Jan L. Riehm2, Michael Hutter2, Rita
Bernhardt1*
1Department of Biochemistry, Saarland University, Campus B2.2, 66123 Saarbrücken,
Germany
*corresponding author
E-mail: ritabern@mx.uni-saarland.de
Phone: (+49) 681-302-4241
Fax: (+49) 681-302-4739
2Center for Bioinformatics, Saarland University, Campus E2.1, 66123 Saarbrücken, Germany
Keywords:
CYP106A2, cytochrome P450, steroid hydroxylation, regioselectivity, progesterone,
adrenodoxin, steroid 9α-hydroxylation, steroid 6β-hydroxylation
1

Abstract:
CYP 106A2 from Bacillus megaterium ATCC 13368 has been described as a 15β-hydroxylase
showing also minor 11α-, 9α- and 6β-hydroxylase activity for progesterone conversion.
Previously, mutant proteins with a changed selectivity towards 11α-OH-progesterone have
already been produced. The challenge of this work was to create mutant proteins with a
higher regioselectivity towards hydroxylation at positions 9 and 6 of the steroid molecule.
9α-hydroxyprogesterone exhibits pharmaceutical importance, because it is a useful
intermediate in the production of physiologically active substances which possess
progestational activity. Sixteen mutant proteins were selected from a library containing
mutated proteins created by a combination of site-directed and saturation mutagenesis of
active site residues. Four mutant proteins out of these catalyzed the conversion of
progesterone to 9α-OH-progesterone as a main product. For further optimization site-
directed mutagenesis was performed. The introduction of seven mutations (D217V, A243V,
A106T, F165L, T89N, T247V or T247W) into these four mutant proteins led to 28 new
variants, which were also used for an in vivo conversion of progesterone. The best mutant
protein, F165L/A395E/G397V, showed a ten-fold increase in the selectivity towards
progesterone 9α-hydroxylation compared with the wild type CYP106A2. Also 6β-OH-
progesterone is a pharmaceutically important compound, especially as intermediate for the
production of drugs against breast cancer. For the rational design of mutant proteins with
6β-selectivity, docking of the 3D-structure of CYP106A2 with progesterone was performed.
The introduction of three mutations (T247A, A243S, F173A) led to seven new mutant
proteins. Clone A243S showed the greatest improvement in 6β-selectivity being more than
ten-fold. Finally, an in vivo conversion of 11-deoxycorticosterone (DOC), testosterone and
cortisol with the best five mutant proteins displaying 9α- or 6β-hydroxylation, respectively,
of progesterone was performed to investigate whether the introduced mutations also
effected the conversion of other substrates.
2

Introduction:
Cytochrome P450 enzymes (CYPs or P450s) are a large super family of heme–thiolate
proteins that are present in all domains of life (animals, plants, fungi and bacteria). So far
more than 18 000 different P450 genes have been discovered
(http://www.ncbi.nlm.nih.gov/gene?term=cytochrome%20p450). They play a key role in the
transformation of endogenous and exogenous substrates,[1] e.g. the biosynthesis of steroids,
antibiotics and signaling lipids.[2,3] Furthermore they are involved in the metabolism of drugs
and xenobiotics,[4] by catalyzing a broad spectrum of reactions such as hydroxylations,
epoxidations, N-oxidation, sulfoxidation, deamination, and dehalogenation.[5] Due to this
variety of reactions they catalyze and the fact that P450s are able to convert a huge range of
substrates, e.g. terpenes, alkaloids and steroids, these enzymes have enormous potential for
biotechnological applications.[6] Additionally, P450s show a high regio- and stereoselectivity
for hydroxylations of different substrates by catalyzing a selective insertion of oxygen into
activated and non-activated C-H bonds.[7]
Cytochrome P450 106A2 (CYP106A2), also known as 15β-hydroxylase, is an in interesting
candidate for biotechnological application because of its ability to hydroxylate a broad range
of steroids, which are of pharmaceutical interest. CYP106A2 was isolated as a soluble
enzyme from Bacillus megaterium ATCC 13368, but its natural function is still unknown. It
has a molecular weight of 47.5 kDa and comprises 410 amino acids.[8] In general, CYP106A2
hydroxylates 3-oxo-Δ4-steroids, i.e. androstendione, corticosterone and testosterone, mainly
at the 15β-position. CYP106A2 is in-vitro active when it is reconstituted with the bovine
electron carriers adrenodoxin (Adx) and adrenodoxin reductase (AdR).[9,10,11]
CYP106A2 as well as many other P450s, for example CYP102 (P450 BM3) have been
successfully engineered by i.e. directed evolution or side-directed mutagenesis to improve
their properties for biotechnological application. The replacement of one or more amino
acids can lead to an improvement of the hydroxylation activity as well as a change of the
substrate orientation and thereby the regiospecificity or -selectivity.[12,13]
Recently, the engineering of CYP106A2 towards the regioselectivity of progesterone
conversion was described. Progesterone is one of the described substrates of CYP106A2. The
main product is 15β-OH-progesterone. Additionally, 6β-, 9α- and 11α-OH-progesterone were
3

found as minor products (scheme 1).[14,15] Moreover, as side products di- and
polyhydroxylated progesterone can be observed. The replacements T89N/A395I and
A160T/A395I/R409L, respectively, lead to a change in the product pattern towards
hydroxylation at position 11, located in the C ring of progesterone. The production ratio of
11α-hydroxyprogesterone increased from 27.7 to 80.9%, whereas that of 15β-
hydroxyprogesterone decreased from 50.4% (wild type) to 4.8% in the double mutant
protein T89N/A395I.[16] The relevance of these results is based on the fact that 11α-OH-
progesterone is an important pharmaceutical compound. It exhibits an anti-androgenic
activity with minimal estrogenic and progestational side effects.[17,18] In addition, 11α-
hydroxyprogesterone influences the blood pressure regulation by causing a selective
inhibition of 11β-hydroxysteroid dehydrogenase type 2. This enzyme metabolizes cortisol to
cortisone in humans and is therefore important in the regulation of the renal electrolyte
balance.[19,20] Furthermore, 11α-hydroxyprogesterone can be used as substrate for the
preparation of cortisone and hydrocortisone.[21,22]
However, 6β- and 9α-hydroxyprogesterone also possess pharmaceutical importance. 6β-
hydroxyprogesterone can be used as an intermediate for the synthesis of 6β,14α-
dihydroxyandrost-4-ene-3,17-dione, an inhibitor of the growth of breast cancer cells.[23] It
was also found that 6β-OH-progesterone inhibits the 5α-reductase activity in male rats. This
way it is an interesting lead for the development of drugs against prostate tumors.[24]
Likewise, 9α-hydroxyprogesterone is of pharmaceutical interest, because it is a useful
intermediate in the production of physiologically active substances that exhibit
glucocorticoid and progestational activity. It is possible to dehydrate 9α-OH-11-
unsubstituted steroids to 9,11-dehydro-steroids that are intermediates in the production of
9α-fluoro-11β-OH-androstenes and pregnenes, for instance fluoxymesterone (9α-fluoro-
11β,17β-dihydroxy-17α-methyl-4-androstene-3-one) and fludrocortisone (9α-fluoro-
11β,17,21-trihydroxy-4-pregnene-3,20-dione acetate).[15,25]
For the production of 6β-hydroxyprogesterone and 9α-hydroxyprogesterone, respectively,
biotechnological synthesis is advantageous compared to the chemical synthesis, since it
takes several steps under mostly harsh reaction conditions, and, in case of 9α-
hydroxyprogesterone, even the use of protection groups.[26,27,28] Therefore, changing the
selectivity of the wild type CYP106A2 from 15β-hydroxylation to 9α- or 6β-hydroxylation of
steroids by site-directed mutagenesis and/or directed evolution is of great value for the
4

biotechnological production of important pharmaceutical compounds and industrial
applications of this enzyme.
Here, we describe the selection of mutant proteins with increased 9α-hydroxylase activities
from a saturation mutagenesis library and the site-directed mutagenesis of these mutant
proteins to improve the 9α-hydroxylase activity of CYP106A2 towards progesterone. We
were able to show an 11-fold increase in the regioselectivity of progesterone hydroxylation
towards the 9α position compared to the wild type CYP106A2 using a whole-cell system.
While studying these mutant proteins, the crystal structure of CYP106A2 was solved.[29] It
was used to perform docking studies of progesterone to predict mutations that would lead
to a change in the selectivity from 15β-hydroxyprogesterone to 6β-hydroxyprogesterone.
The regio-selectivity of mutant protein A243S towards the 6β position was changed from 9%
to 86%.
5

Materials and Methods:
Chemicals:
Progesterone, testosterone, cortisol and 11-deoxycorticosterone (DOC) were purchased
from Sigma-Aldrich Inc. All other chemicals were from standard sources and of the highest
purity available. Enzymes for DNA digestion, PCR and ligation were from New England Biolabs.
Site-directed mutagenesis:
The site-directed mutagenesis was performed by PCR using plasmid pACYC_FHH2.8
containing the genes for CYP106A2 with primers containing the new mutation (Table 1).
They were synthesized by Eurofins MWG Operon. The reaction mixture contained 1x HF
buffer, 0.2 mM dNTPs, 10 pmol primers forward and reverse respectively, 10 ng of template
DNA and 2 U Phusion polymerase to a total volume of 20 µl. The original DNA was digested
with the restriction enzyme DpnI for 3 h.
In vivo conversion:
E. coli cells JM109 were cotransformed with pACYC_FHH2.8 containing the genes for
CYP106A2 wild type or its mutant proteins and pBar_Twin[30], encoding for AdR and Adx (4-
108). The cells were used for inoculation of an overnight culture (50 mL LB-medium)
containing 50 µg/mL chloramphenicol and 100 µg/mL ampicillin. The main culture (250 mL
Terrific broth medium) with 50 µg/mL chloramphenicol and 100 µg/mL ampicillin was
inoculated with 1/100 of the pre-culture and incubated at 37 °C. At an OD600 of 0.6 to 0.8,
isopropyl-β-D-thiogalactopyranoside (1 mM) and 5-aminolevulinic acid (0.5 mM) were added
for induction. After 48 h protein expression at 30°C, an aliquot (50 mL) was centrifuged
(4500 g, 10 min), washed with 50 mM potassium phosphate buffer (pH 7.4) and suspended
in 50 mL buffer containing 100 mM glucose. The conversion was started by adding
progesterone, testosterone, cortisol, or DOC, respectively (200 µM), and polymyxin B
(32µg/ml). After 24 h incubation at 30 °C the enzyme reaction was quenched with
chloroform. The extraction of the steroids was performed twice with 500 mL chloroform.
The resulting organic phase was dried and the extract dissolved in acetonitrile for HPLC
analysis. All conversions were performed three times to determine the standard deviation.
6

HPLC analysis of products:
Samples were analyzed in water/acetonitrile (60:40) for DOC, testosterone andcortisoland in
water/acetonitrile (90:10) for progesterone, respectively,with a CC 125/4 nucleodur 100-5
C18 ec column (Macherey & Nagel) and detected at 240 nm wavelength.
Computational Methods:
Chain B of the preliminary X-ray structure of CYP106A2 was used as receptor within
AUTODOCK (version 4.2).[31,32] Hydrogen atoms and Kollman charges were added using the
Windows version 1.5.6r3 of AutoDock Tools.[33] The retangular grid box (58 x 56 x 54 points
with a grid spacing of 0.375 Å) was centered about 5 Å above the heme iron. The atomic
charge of the porphyrin iron a was set to +0.400 e, corresponding to Fe(II), and the partial
charges of its ligating nitrogen atoms were set to –0.348 e, respectively, to compensate this.
The structure of progesterone was prepared manually and energetically minimized applying
the MM+ force field as implemented in HYPERCHEM (HYPERCHEM, Version 6, Gainesville, FL,
Hypercube Inc. 1999). Corresponding Gasteiger-Marsili charges and atom types for use in
AUTODOCK were generated with an in-house PERL script. A total of 250 docking runs were
carried out applying the Lamarckian genetic algorithm using default parameter settings. The
resulting docked conformations of progesterone inside the active site were evaluated by
manual determination of the corresponding distances between eligible hydroxylation sites
and the porphyrin iron.
Molecular dynamics simulations were carried out on the docked conformation of
progesterone in the binding pocket of CYP106A2 with the mutation A243S applying
GROMACS (version 4.4.5).[34] The corresponding side chain of serine was introduced and
energetically minimized using HYPERCHEM. The cytochrome was centered in a cubic box
with a minimum distance of 10 Å from the box edges and surrounded by TIP3P waters.[35]
Prior to equilibration, steepest descent energy minimization was started until a gradient
norm below 1000 kJ mol-1 nm-2 was reached. Equilibration consisted of two steps: The first
step comprised a NVT (constant volume, temperature, and number of particles) ensemble
where the temperature was kept at 300K using a modified Berendsen thermostat for 100
ps.[36] The second step involved a NPT (constant pressure, temperature, and number of
7

particles) ensemble to stabilize the molecular system at 1 bar applying a Parinello-Rahman
barostat for 100 ps.[37] All bonds in both steps were constrained by the LINCS algorithm.[38]
As a last step the production run of 20 ns was carried out. Hereby Newton’s equations of
motion were integrated by the leapfrog algorithm at time steps of 2 fs. Trajectory snapshots
were taken at intervals of 1 ps.
8

Results and Discussion:
The bacterial enzyme CYP106A2 was regarded as suitable platform to change the steroid
hydroxylation regioselectivities and to improve the catalytic efficiencies of the obtained
mutant proteins. Therefore, a library of CYP106A2 with different mutations, containing more
than 13 000 clones was created based on the replacements at positions A395 and G397 by a
combination of site-directed and saturation mutagenesis of active site residues, followed by
several rounds of site-directed mutagenesis based on a low-homology model of
CYP106A2.[16,39]
The library was screened using 200 µM progesterone as a substrate followed by HPLC to
identify mutant proteins with increased 9α-hydroxylase activity. Sixteen mutant proteins
with improved activity were selected from this library. HPLC analysis identified the yields of
15β-, 11α-, 9α- and 6β-hydroxyprogesterone for each mutant protein (Table 2). In addition
to the single-hydroxylated products in dependence on the reaction time also small amounts
of di- and polyhydroxylated products were formed (data not shown). Ten mutant proteins
formed 9α-hydroxyprogesterone as a main product. It can be seen from Table 2 that
CYP106A2 A395W/G397K showed the highest (55.9%) and A395H/G397Y, respectively, the
lowest (38.6%) 9α-hydroxyprogesterone formation. This comprises an at least 8-fold
improvement compared to the wild type. In contrast, CYP106A2 A395R/G397R possessed
the greatest decrease of 15β-hydroxylase activity, from 67.9% to 8.8%. The other six mutant
proteins (A395I, A395M/G397S, A395R/G397P, G397W, A395F and A395V) exhibited a
smaller 9α-hydroxylase activity compared to A395W/G397K and A395H/G397Y, whereby
two mutant proteins (A395R/G397P and A243V) showed almost no 9α-hydroxyprogesterone
formation at all. In addition, A243V showed a higher selectivity towards 15β-
hydroxyprogesterone formation compared to the wild type enzyme.
The four best mutant proteins A395E/G397V, A395R/G397K, A395W/G397K, and
A395R/G397R with the highest 9α-hydroxyprogesterone formation were selected for further
mutation via site-directed mutagenesis to improve their regioselectivity. Therefore, seven
single replacements (D217V, A243V, A106T, F165L, T89N, T247V and T247W) were inserted
into the four mutant proteins. These mutations were shown before to influence the
regioselectivity and catalytic efficiency in a positive way. Rauschenbach and Virus could
show that they exhibit higher turnover rates towards progesterone conversion compared to
the wild type.[40,41] In addition, it was reported that the highly conserved threonine residue in
9

the I-helix of the active side near the heme of P450s directly affects the catalytic efficiency
due to its participation in the proton shuttle which is important for proton delivery and
molecular oxygen activation.[14] Therefore, the mutations T247V and T247W were also
carried out. The insertion of these seven amino acid replacements into the four “parent
mutant proteins” led to 28 new mutant proteins that were also tested for in vivo conversion
of 200 µM progesterone, respectively.
Five mutant proteins showed almost no conversion. Due to their low activity, these clones
were not followed up in further experiments. Sixteen mutant proteins possessed a 9α-
hydroxylase activity similar to that of the parent mutant proteins. There were also two
mutant proteins (A243V/A395E/G397V and T89N/A395W/G397K) with a strong shift in
regioselectivity, but they also showed an increase in di- and polyhydroxylated progesterone.
These products were observed in the HPLC chromatograms (between 3.5 and 7.5 min) with
the peak size depending on the conversion time as well as the corresponding mutant
protein. By shortening the reaction time, the amounts of side product can be reduced.[16]
Interestingly, five mutant proteins displayed a low amount of di- and polyhydroxylated
progesterone and conversely a high amount of 9α-hydroxyprogesterone
(D217V/A395E/G397V, F165L/A395E/G397V, A106T/A395R/G397K, A243V/A395R/G397R
and A106T/A395R/G397R) (Figure 1). The relative amounts of hydroxylated metabolites
after conversion of progesterone in the presence of the best five of the 28 mutant proteins
of CYP106A2 are shown in Table 3.
CYP106A2 F165L/A395E/G397V had the strongest shift in the regioselectivity showing a
formation of 59.7% 9α-hydroxyprogesterone. This represents an improvement of the yield of
9α-hydroxyprogesterone by 15% compared to the initial CYP106A2 A395E/G397V, but a 12-
fold improvement compared to the wild type CYP106A2. In turn, the amount of 15β-
hydroxyprogesterone decreased from 67.9% to 11.3%. Mutant proteins
D217V/A395E/G397V, A106T/A395R/G397K, and A106T/A395R/G397R showed a shift in
regioselectivity of about 10% compared to the parent mutant proteins A395E/G397V,
A395R/G397K, and A395R/G397R, respectively, and of about 51% compared to the wild
type. CYP106A2 A243V/A395R/G397R possessed the smallest enhancement of 9α-
hydroxyprogesterone formation (49.7%), but a large decrease of 15β-hydroxyprogesterone
from 67.9% to 11.8% compared to the CYP106A2 wild type. Taken together, this data clearly
demonstrate that using a combination of site-directed mutagenesis/saturation mutagenesis
10

not only changes the selectivity from the 15β- to the 11α-position of progesterone as
described before,[16] but this approach can also be applied to change the regioselectivity of
progesterone hydroxylation from position 15 in the D ring to position 9 in the B ring.
For creating mutant proteins leading to 6β-hydroxyprogesterone production, the recently
solved crystal structure of CYP106A2 was used.[29] After performing docking of progesterone
into this structure, several amino acids in the enzyme were identified, whose mutation
should lead to a change in the selectivity from 15β to the 6β position leading to the
production of 6β-hydroxyprogesterone. It was suggested that the change from threonine to
alanine at position 247 should interrupt a hydrogen bond, which should then prefer an
alternative conformation of the substrate within the active site. Moreover, serine instead of
alanine in position 243 was expected to stabilize a corresponding conformation in which 6β
hydroxylation is favored by introducing a new hydrogen bond with the carbonyl oxygen in
position 21 (Figure 2). To support this assumption, we carried out molecular dynamics
simulations of progesterone in a homology model of this CYP106A2 A243S. Figure 3 and 4
show the time course of the distances of the hydrogen atoms in 15β and 6β position from
the iron atom of the heme group, respectively. Whereas both curves show a decrease of the
distance from the beginning of the simulation to about 5 ns, the 6β position remains closer
to the heme iron within the subsequent 5 ns. The shortest distance for the 15β position was
observed after 19.2 ns (2.6 Å), whereas the minimum distance of the 6β position was found
already after 6.8 ns (3.0 Å). Likewise, the average distances are 5.7 and 5.5 Å, respectively,
over the total simulation time of 20 ns. The maximum distances were seen rather early
between 0.5 and 3.0 ns, both being beyond 9.2 Å. Obviously this short time span of 5 to 10
ns in which the 6β position is about 1 Å closer to the heme iron than the 15β position seems
to determine the product spectrum. Finally, changing phenylalanine against the smaller
alanine at position 173 should create more space over the heme for an optimal
conformation of progesterone for 6β-hydroxylation.
Single mutations as well as combinations of these three changes were inserted into
CYP106A2. The obtained mutant proteins were used for the conversion of progesterone.
CYP106A2 F173A/T247A did not show any significant conversion, whereas mutant proteins
F173A, T247A, A243S/T247A and F173A/A243S converted progesterone to the main product
9α-hydroxyprogesterone. However, CYP106A2 with the replacements A243S and
F173A/A243S/T247A, respectively, showed 6β-hydroxyprogesterone as main product with a
11

9- and 6-fold improvement, respectively, compared to the wild type. The amount of other
products, 15β-, 11α- and 9α-hydroxyprogesterone, of CYP106A2 A243S is much lower than
that of the triple mutant protein (Table 4). Figure 5 shows the chromatogram of CYP106A2
A243S compared to the wild type. It can be seen that site-directed mutagenesis of CYP106A2
based on the results of docking of progesterone into the crystal structure of this protein, led
to a nearly complete change of the regioselectivity of hydroxylation from the 15β- to the 6β-
position, i.e. from the D-ring to the B-ring. More than 80% 6β-hydroxylation was obtained.
To find out whether mutant proteins, which are changing the regioselectivity of
progesterone hydroxylation from the 15 to the 11, 9, or 6 position, show another substrate
space than the wild type, a natural product library was screened with CYP106A2
T89N/A395I[42], F165L/A395E/G397V, and A243S, respectively. Screening of the library with
these three mutant proteins showed the same twelve compounds as already published by
Schmitz et al[43] for wild type CYP106A2 so that obviously no change in the substrate space
occurred.
To investigate a possible change in the regioselectivity of other known CYP106A2 substrates,
dipterocarpol was converted with the three mutant proteins and the results were compared
to those of the wild type, which formed two products, 7β-hydroxydipterocarpol and 7β,11α-
dihydroxydipterocarpol. Interestingly, the three mutant proteins converted dipterocarpol
only into one product, 7β-hydroxydipterocarpol, indicating an increased specificity of
hydroxylation towards the 7β-position. Figure 6 shows exemplarily the thin layer analysis of
CYP106A2 T89N/A395I. Finally, 11-deoxycorticosterone, was converted in vivo in the
presence of wild type CYP106A2 and the six best mutant proteins created in this work
(D217V/A395E/G397V, F165L/A395E/G397V, A106T/A395R/G397K, A243V/A395R/G397R or
A106T/A395R/G397R, A243S). Figure 7 shows the HPLC chromatogram of the conversion of
DOC with wild type CYP106A2 and with CYP106A2 A106T/A395R/G397R. The wild type
produced 15β-hydroxy-DOC as the main product as well as 6β-hydroxy-DOC and 7β,15β-
dihydroxy-DOC as minor products.[44] In case of the mutant proteins the peaks 2 (7β,15β-
dihydroxy-DOC) and 3 (15β-OH-DOC) drastically decreased whereas peak 4 was enhanced,
demonstrating an increased production of 6β-OH-DOC. To date product 1 has not been
identified because the amount is not sufficient for NMR analysis. We assume that this peak
comprises either a mixture of several monohydroxylated DOC products or di- and/or
polyhydroxylated DOC.
12

Taken together, it can be concluded that the mutations introduced by directed evolution and
site-directed mutagenesis led also to a change in the regioselectivity of other steroidal
substrates besides progesterone resulting in an increase in the selectivity of dipterocarpol
hydroxylation towards the 7β position and in the new main product 6β-hydroxy-DOC instead
of 15β-hydroxy-DOC. The stongest shift in selectivity towards DOC hydroxylation was shown
by CYP106A2 D217V/A395E/G397V displaying an improvement of 6β-OH-DOC by 44% and,
on the other hand, showing a decrease of 15β-OH-DOC by 36% compared to CYP106A2 wild
type.
13

Conclusion:
In this work we were able to demonstrate that a combination of site-directed mutagenesis
and saturation mutagenesis of CYP106A2 is able to influence the regioselectivity of
progesterone hydroxylation, shifting it from C15 in the D ring to C9 and to C6 in the B ring,
respectively. Besides changes in the regioselectivity, also a change in the stereo-selectivity
from the 15β to the 9α position was obtained. It was shown that the hydroxylation position
in the steroid depends on the mutation introduced into the protein. Whereas mutant
proteins with a change in the selectivity from 15β- to 11α-hydroxyprogesterone as described
previously[16] showed the largest shift in mutant proteins A106T/A395I, A106T/A395I/R409L,
and T89N/A395I, CYP106A2 F165L/A395E/G397V demonstrated the strongest shift from the
15β towards 9α position. Interestingly, a nearly complete shift in the regioselectivity of
progesterone hydroxylation from the 15β to the 6β position was obtained with CYP106A2
A243S, which was constructed by rational protein design/site-directed mutagenesis based
on the recently available X-ray structure of the wild type.[29]
In addition, we were able demonstrate that also other steroids similar to progesterone, such
as DOC and dipterocarpol, are hydroxylated with changed selectivity when using the above
mentioned mutant proteins.
Finally, the approach described here, especially when using the 3D structure of CYP106A2 as
a basis for the rational design of mutant proteins, leads to desired changes in the regio- and
stereo-selectivity of progesterone hydroxylation. The obtained products, 9α- and 6β-
hydroxyprogesterone, are of biotechnological importance for a sustainable production of
pharmaceutically valuable compounds.
Acknowledgments:
This work was supported by a grant of the Bundesministerium für Forschung und
Technologie (BMBF) to R.B. and M.H. (grant 031A166A).
14

References:
[1] R. Bernhardt (1996), Cytochrome P450: structure, function, and generation of reactive
oxygen species, Rev. Physiol. Biochem. Pharmacol. 127, 137–221.
[2] R. Bernhardt, V. B. Urlacher (2014), Cytochromes P450 as promising catalysts for
biotechnological application: chances and limitations, Appl. Microbiol. Biotechnol. 98, 6185-
6203.
[3] A. W. Munro, H. M. Girvan, K. J. McLean (2007), Variations on a (t)heme–novel
mechanisms, redox partners, and catalytic functions in the cytochrome P450 superfamily,
Nat. Prod. Rep. 24, 585–609.
[4] K. Ruckpaul, H. Rein (Eds.) (1991) in Frontiers in Biotransformation, Vol. 6. Akademie
Verlag, Berlin.
[5] M. Sono, M. P. Roach, E. D. Coulter, J. H. Dawson (1996), Heam-containing oxygenases,
Chem. Rev. 96, 2841–2888.
[6] R. Bernhardt (2006), Cytochromes P450 as versatile biocatalysts, J. Biotechnol. 124, 128 -
145.
[7] H. Joo, Z. Lin, F. H. Arnold (1999), Laboratory evolution of peroxide-mediated cytochrome
P450 hydroxylation, Nature 399, 670-673
[8] R. Rauschenbach, M. Isernhagen, C. Noeske-Jungblut, W. Boidol, G. Siewert (1993),
Cloning sequencing and expression of the gene for cytochrome P450meg, the steroid-15β-
monooxygenase from Bacillus megaterium ATCC 13368, Mol. Gen. Genet. 241, 170-176.
[9] A. Berg, J. A. Gustafsson, M. Ingelman-Sundberg (1976), Characterization of a cytochrome
P-450-dependent steroid hydroxylase system present in Bacillus megaterium, J. Biol. Chem.
251, 2831–2838.
[10] A. Berg, M. Ingelman-Sundberg, J. A. Gustafsson (1979), Isolation and characterization
of cytochrome P-450meg, Acta Biol. Med. Ger. 38, 333-344.
[11] A. Berg, M. Ingelman-Sundberg, J. A. Gustafsson (1979), Purification and
characterization of cytochrome P-450meg, J. Biol. Chem. 254, 5264–5271.
[12] C. F. Oliver, S. Modi, M. J. Sutcliffe, W. U. Primose, L.-Y. Lian, G. C. K. Roberts (1997), A
Single Mutation in Cytochrome P450 BM3 Changes Substrate Orientation in a Catalytic
Intermediate and the Regiospecificity of Hydroxylation, Biochem. 36, 1567-1572.
15

[13] Q.-S. Li, U. Schwaneberg, M. Fischer, J. Schmitt, J. Pleiss, S. Lutz-Wahl, R. D. Schmid
(2001), Rational evolution of a medium chain-specific cytochrome P-450 BM-3 variant,
Biochim. Biophys. Acta 1545, 114-121.
[14] M. Lisurek, B. Simgen, I. Antes, R. Bernhardt (2008), Theoretical and Experimental
Evaluation of a CYP106A2 Low Homology Model and Production of Mutants with Changed
Activity and Selectivity of Hydroxylation, ChemBioChem 9, 1439 – 1449.
[15] M. Lisurek, M. J. Kang, R. W. Hartmann, R. Bernhardt (2004), Identification of
monohydroxy progesterones produced by CYP106A2 using comparative HPLC and
electrospray ionisation collision-induced dissociation mass spectrometry, Biochem. Biophys.
Res. Commun. 319, 677-682.
[16] K. T. Nguyen, C. Virus, N. Günnewich, F. Hannemann, R. Bernhardt (2012), Changing the
Regioselectivity of a P450 from C15 to C11 Hydroxylation of Progesterone, ChemBioChem
13, 1161-1166.
[17] J. Tamm, M. Seckelmann, U. Volkwein, E. Ludwig (1982), The effect of the antiandrogen
IIα-hydroxyprogesterone on sebum production and cholesterol concentration of sebum, Br.
J. Dermatol. 107, 63-70.
[18] A. H. van der Willigen, J. D. Peereboom-Wynia, T. van Joost, E. Stolz (1987), A
preliminary study of the effect of 11a-hydroxyprogesterone on the hair growth in men
suffering from androgenetic alopecia, Acta Derm.-Venereol. 67, 82-85.
[19] H. Morita, M. Zhou, M. F. Foecking, E. P. Gomez-Sanchez, E. N. Cozza, C. E. Gomez-
Sanchez (1996), 11β-Hydroxysteroid dehydrogenase type 2 complementary deoxyribonucleic
acid stably transfected into Chinese hamster ovary cells: specific inhibition by 11α-
hydroxyprogesterone, Endocrin. 137, 2308-2314.
[20] G. W. Souness, D. J. Morris (1996), 11α- and 11β-Hydroxyprogesterone, Potent
Inhibitors of 11β-Hydroxysteroid Dehydrogenase, Possess Hypertensinogenic Activity in the
Rat, Hypertension 27, 421-425.
[21] J. A. Hogg, P. F. Beal, A. H. Nathan, F. H. Lincoln, W. P. Schneider, B. J. Magerlein, A. R.
Hanze, R. W. Jackson (1955), The adrenal hormones and related compounds. I. A “direct”
synthesis of hydrocortisone acetate and cortisone acetate from 11α-hydroxyprogesterone, J.
Am. Chem. Soc 77 (16), 4436-4438.
16

[22] J. Fried, R. W. Thoma, J. R. Gerke, J. E. Herz, M. N. Donin, D. Perlman (1952), Oxidation
of steroids by microorganisms. II. hydroxylations in position 11 and synthesis of cortisone
from Reichenstein’s compound S, J. Am. Chem. Soc. 74 (15), 3962–3963.
[23] H. Yoshioka, S. Asada, S. Fujita (1994), Process for production of 6β, 14α-
hydroxyandrost-4-ene-3,17-dione, European Patent 5: 99, 658.
[24] A. Yamada, M. Yamada, Y. Fujita, T. Nishigami, K. Nakasho, K. Uematsu (2001), Self-
augmentation Effect of Male-specific Products on Sexually Differentiated Progesterone
Metabolism in Adult Male Rat Liver Microsomes, J. Biol. Chem. 276, 4604-4610.
[25] D. Perlman, J. D. Dutcher, J. Fried, E. O. Titus (1958), 9alpha-hydroxyprogesterone, U.S.
patent no. 2,840,578.
[26] J. Fried (1964), 9alpha-hydroxypregnanes and process for production there of, U.S.
patent no 3,141,877.
[27] L. S. Costanzo (2006), Endocrine physiology. Physiology 3rd ed., 377-439.
[28] P. H. Yu, L. TAN (1977), Synthesis and properties of two new steroid hydroperoxides: an
alternative pathway for the formation of 6β-hydroxyprogesterone, J. Steroid Biochem. 8,
825-834.
[29] S. Janocha, Y. Carius, M. Hutter, C. R. D. Lancaster, R. Bernhardt (2016), Crystal Structure
of CYP106A2 in Substrate-Free and Substrate-Bound Form, ChemBiochem 17, 852-860.
[30] F. Hannemann, C. Virus, R. Bernhardt (2006), Design of an Escherichia coli system for
whole cell mediated steroid synthesis and molecular evolution of steroid hydroxylases, J.
Biotechnol., 124 (1), 172-181.
[31] R. Huey, G. M. Morris, A. J. Olson, D. S. Goodsell (2007), A semiempirical free energy
force field with charge-based desolvation, J. Comput. Chem. 28 (6), 1145-1152.
[32] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K. Belew, A. J. Olson
(1998), Automated Docking Using a Lamarckian Genetic Algorithm and Empirical Binding
Free Energy Function, J. Comput. Chem. 19, 1639-1662.
[33] M. F. Sanner (1999), Python: a programming language for software integration and
development, J. Mol. Graph. Model. 17 (1), 57–61.
[34] S. Pronk , S. Páll, R. Schulz, P. Larsson, P. Bjelkmar, R. Apostolov, M. Shirts, J. C. Smith, P.
M. Kasson, D. van der Spoel, B. Hess, E. Lindahl (2013), GROMACS 4.5: a high-throughput and
highly parallel open source molecular simulation toolkit, Bioinformatics 29, 845-854.
17

[35] W. L. Jorgensen, J. D. Madura (1983), Quantum and Statistical Studies of Liquids. 25.
Solvation and Conformation of Methanol in Water, J. Am. Chem. Soc. 105, 1407-1413.
[36] H. J. C. Berendsen, J. P. M. Postma, W. F. Vangunsteren, A. Dinola, J. R. Haak (1984),
Molecular-dynamics with coupling to an external bath, J. Chem. Phys. 81, 3684-3690.
[37] M. Parrinello, A. Rahman (1981), Polymorphic transitions in single-crystals. A new
molecular-dynamics method, J. Appl. Phys. 52, 7182-7190.
[38] B. Hess (2008), P-LINCS: A parallel linear constraint solver for molecular simulation, J.
Chem. Theory. Comput. 4, 116-122.
[39] C. Virus, R. Bernhardt (2008), Molecular Evolution of a Steroid Hydroxylating
Cytochrome P450 Using a Versatile Steroid Detection System for Screening, Lipids 43, 1133-
1141.
[40] R. Rauschenbach (1993), PhD thesis: Klonierung, Expression und Mutagenese des Gens
für die Steroid-15β-Hydroxylase aus Bacillus megaterium ATCC 13368, Freie Universität
Berlin (Germany).
[41] C. Virus (2006), PhD thesis: Etablierung und Evaluierung eines Screeningsystems für die
molekulare Evolution der Steroid-15b-Hydroxylase (CYP106A2) aus Bacillus megaterium
ATCC 13368, Universität des Saarlandes (Germany).
[42] K. T. Nguyen, C. Virus, N. Günnewich, F. Hannemann, R. Bernhardt (2012), Changing the
regioselectivity of a P450 from C15 to C11 hydroxylation of progesterone, ChemBioChem 13
(8), 1161-1166.
[43] D. Schmitz, J. Zapp, R. Bernhardt (2012), Hydroxylation of the triterpenoid dipterocarpol
with CYP106A2 from Bacillus megaterium, FEBS Journal 279, 1663–1674.
[44] F. M. Kiss, D. Schmitz, J. Zapp, T. K. Dier, D. A. Volmer, R. Bernhardt (2015), Comparison
of CYP106A1 and CYP106A2 from Bacillus megaterium - identification of a novel 11-oxidase
activity, Appl. Microbiol. Biotechnol. 99 (20), 8495-8514.
18

Table 1. Oligonucleotides used for mutagenesis.
Mutation
Primer sequence
GCGGATGATATCATCTCAGTACTATTGAAGTCGG
CCGACTTCAATAGTACTGAGATGATATCATCCGC
ATTTTAGGTGTAGGAGTCGAG
D217Vfor
D217Vrev
A243Vfor
A243Vrev
CTCGACTCCTACACCTAAAAT
CCTGAAAAGATCCAAATTAATGAATCGGATCCACCT
AGGTGGATCCGATTCATTAATTTGGATCTTTTCAGG
T89Nfor
T89Nrev
A106Tfor
CTGGCAGCAACGTTCACACCT
AGGTGTGAACGTTGCTGCCAG
A106Trev
F165Lfor
F165Lrev
T247Vfor
T247Vrev
T247Wfor
T247Wrev
CGAAAGATCGTCTCTTGTTGAAGAAATGGGTGG
CCACCCATTTCTTCAACAAGAGACGATCTTTCG
GGTGCAGGAGTCGAGGTAACTAGTCATTTATTGGCC
GGCCAATAAATGACTAGTTACCTCGACTCCTGCACC
GGTGCAGGAGTCGAGTGGACTAGTCATTTATTGGCC
GGCCAATAAATGACTAGTCCACTCGACTCCTGCACC
GCAGGAGTCGAGGCAACCAGTCATTTATTGG
T247A_for
T247A_rev
CCAATAAATGACTGGTTGCCTCGACTCCTGC
GCTGATTTTAGGTTCAGGAGTCGAGAC
A243S_for
A243S_rev
GTCTCGACTCCTGAACCTAAAATCAGC
GGATACCTTAGCTCTTCCTTTTG
F173A_for
F173A_rev
CAAAAGGAAGAGCTAAGGTATCC
19

Table 2. Progesterone hydroxylation catalyzed by CYP106A2. Relative values of the product
pattern (15β-, 11α-, 9α- and 6β-hydroxyprogesterone) after 24 h in vivo conversion of 200
µM progesterone catalyzed by wild type CYP106A2 and its 16 mutant proteins, which were
selected from a library. Values show percentages with the sum of all monohydroxylated
products set as 100%.
Number Mutation
15β-OH-P
[%]
11α-OH-P
[%]
9α-OH-P
[%]
5.4
6β-OH-P
[%]
8.7
WT
A395T/G397K
67.9
11.2
22.9
28.7
9.4
18.0
20.0
15.2
50.9
15.9
12.9
65.6
12.7
57.0
37.2
36.4
38.3
8.2
1
45.1
38.6
10.2
48.6
44.8
14.4
47.0
2.4
23.7
23.3
10.2
26.1
33.2
6.9
2
A395H/G397Y
A395I
3
4
A395E/G397V
A395R/G397K
A395M/G397S
A395R/G397I
A395R/G397P
A395K
5
9.1
6
13.1
13.2
38.4
16.5
27.3
32.5
78.0
10.5
7
8
27.2
2.2
9
42.4
27.3
11.5
3.6
4.0
9.1
10
11
12
13
14
15
16
G397W
A395F
A243V
A395W/G397K
17.7
10.2
14.3
25.8
26.8
30.6
19.2
15.5
13.7
14.9
55.9
45.4
41.1
45.8
A395E/G397Stop 13.4
A395S/G397Y
A395R/G397R
18.5
8.8
20

Table 3. Selection of progesterone hydroxylation catalyzed by improved mutated CYP106A2.
Relative values in % of progesterone and the monohydroxylated products 15β-, 11α-, 9α-
and 6β-hydroxyprogesterone after 24 h in vivo conversion of 200 µM progesterone catalyzed
by wild type CYP106A2 and the mutant proteins D217V/A395E/G397V, F165L/A395E/G397V,
A106T/A395R/G397K, A243V/A395R/G397R and A106T/A395R/G397R, respectively. Values
show percentages with the sum of the substrate progesterone and all monohydroxylated
products set as 100%.
Mutation
15β-OH-P
[%]
11α-OH-P
[%]
9α-OH-P
[%]
6β-OH-P
[%]
P
[%]
0.0
5.4
0.9
0.4
12.7
2.7
WT
D217V/A395E/G397V 7.5
67.9
18.0
23.2
5.4
8.7
7.9
55.9
59.7
57.4
49.7
56.2
F165L/A395E/G397V
11.3
17.2
22.4
11.0
6.8
A106T/A395R/G397K 13.0
A243V/A395R/G397R 11.8
A106T/A395R/G397R 12.9
17.6
14.9
8.3
13.3
21

Table 4. Progesterone hydroxylation catalyzed by improved mutated CYP106A2. Relative
values in % of progesterone as well as the monohydroxylated products 15β-, 11α-, 9α- and
6β-hydroxyprogesterone after 24 h in vivo conversion of 200 µM progesterone by wild type
CYP106A2 and the mutant proteins F173A, A243S, T247A, F173A/A243S, A234S/T247A and
F173A/A243S/T247A,F165L/A395E/G397V, A106T/A395R/G397K, A243V/A395R/G397R and
A106T/A395R/G397R, respectively. Values show percentages with the sum of the substrate
progesterone and all monohydroxylated products set as 100%.
Mutation
15β-OH-P
[%]
11α-OH-P
[%]
9α-OH-P
[%]
6β-OH-P
[%]
P
[%]
0.0
8.6
6.9
5.3
4.3
3.4
19.1
WT
F173A
67.9
9.7
18.0
32.5
4.6
13.1
15.5
13.2
6.5
5.4
44.1
2.4
45.7
51.8
48.0
8.7
8.7
5.0
A243S
T247A
3.4
82.7
24.3
15.9
20.7
57.2
11.7
12.5
14.6
F173A/A243S
A243S/T247A
F173A/A243S/T247A 13.6
22

Scheme 1. CYP106A2-catalyzed conversion of progesterone to 15β-, 11α-, 9α- and 6β-
hydroxyprogesterone.
23

Figure 1. HPLC analysis of the in vivo conversion of 200 µM progesterone by wild type CYP106A2 (A) and the mutant proteins
D217V/A395E/G397V(B), F165L/A395E/G397V (C), A106T/A395R/G397K (D), A243V/A395R/G397R (E), A106T/A395R/G397R (F). The figure shows
di- and polyhydroxylated(6) and monohydroxylated products (15β- (1), 11α- (2), 9α- (3) and 6β-hydroxyprogesterone (4)), as well as the substrate
progesterone (5). HPLC analysis was performed with H2O:ACN (90:10) as a mobile phase and a CC 125/4 nucleodur 100-5 C18 ec column at 40°C.
24

Figure 2. Docking of progesterone into the X-ray structure of wild-type CYP106A2 showed
position 15β (orange) being closest to the heme iron (3.9 Å). However, hydroxylation was
experimentally also observed in 6β position (purple) (5.1 Å). The replacement of alanine by
serine in position 243 would enable a new hydrogen-bond (blue dotted line) to the carbonyl
oxygen in position 21. As a further consequence the progesterone is also shifted towards this
residue thereby bringing position 6β in closer contact to the heme iron. The corresponding
mutant protein shifted the main product from 15β to 6β-hydroxy-progesterone.
25

Figure 3. Time course of the distance (given in Å) between the hydrogen atom in position
15β of progesterone and the iron atom of heme of the heme group during a 20 ns molecular
dynamic simulation of CYP106A2 A243S.
26

Figure 4. Corresponding time course of the distance for the hydrogen atom in position 6β of
progesterone. Between 5 and 10 ns this position is about 1 Å closer to the heme iron atom
and was experimentally found to be the main hydroxylation product of CYP106A2 A243S.
27

Figure 5. HPLC analysis of the in vivo conversion of 200 µM progesterone by wild type
CYP106A2 and with the mutation A243S. The figure shows di- and polyhydroxylated (6) and
monohydroxylated products (15β- (1), 11α- (2), 9α- (3) and 6β-hydroxyprogesterone (4)).
Product 5 only appears in the mutant protein and was not identified yet. HPLC analysis was
performed with H2O:ACN (90:10) as a mobile phase and a CC 125/4 nucleodur 100-5 C18 ec
column at 40°C.
28

Figure 6. Thin layer analysis of the in vivo conversion of 200 µM dipterocarpol by wild type
CYP106A2, the CYP106A2 T89N/A395I and a control without any enzyme. The figure shows
the two products, 7β-OH-dipterocarpol and 7β,11α-OH-dipterocarpol. Thin layer analysis
was performed with hexane/EE 1:1 as a mobile phase.
29