of 56
S1
Supporting Information
Directed Evolution of a Cytochrome P450 Carbene Transferase
for Selective Functionalization of Cyclic Compounds
Oliver F. Brandenberg
1,†
, Kai Chen
1
, Frances H. Arnold
1*
1
Division of Chemistry and Chemical Engineering, Cali
fornia Ins
titute of Technology,
Pasadena, California 91125, USA.
Present address: Howard Hughes Medical Institute, Department of Human Genetics, University
of California Los Angeles, Los Angeles, California 90095, USA.
* Corresponding author: Frances H
. Arnold
frances@cheme.caltech.edu
Supporting Information content:
I
Experimental Procedures
p. 2
II
Supplementary Tables
p. 9
III
Supplementary Figures
p. 16
IV
Characterization of reaction products
p. 27
V
NMR spectra
p. 38
VI
Nucleotide and amino acid sequences of P411 variants
p. 53
VI
I
Supplementary References
p. 55
S2
I. Experimental Procedures
General
Unless otherwise noted, all chemicals and reagents were obtained from commercial
suppliers (Sigma
-Aldrich, VWR, Alfa Aesar) and used without further purification. Silica gel
chromatography was carried out using AMD Silica Gel 60, 230-
400 mesh.
Synthetic reactions
were monitored using thin layer chromatography (Merck 60 gel plates) using a
UV-
lamp for
visualization.
1
H and
13
C NMR spectra were recorded on a Varian Inova 300 MHz or 500 MHz,
or Bruker Prodigy 400 MHz instrument, in CDCl
3
and are internally referenced to the resi
dual
solvent peak.
Data for
1
H NMR are reported as follows: chemi
cal shift (
δ
ppm), multiplicity (s =
singlet, d = doublet, t = triplet, q = quartet, p = pentet, m = multiplet, dd = doublet of doublets, dt
= doublet of triplets, ddd = doublet of doublet of doublets), coupling constant (Hz), integration.
High
-resolution
mass spectra were obtained at the California Institute of Technology Mass
Spectr
ometry
Facility.
Chromatography.
Analytical
reversed
-phase
high-
performance liquid chromatography (HPLC)
was carried out using an Agilent 1200 series instrument with water an
d acetonitrile as the mobile
phases
using either a
Kromasil 100
-5- C18 column (4.6 x 50 mm, 5 μM)
or a Poroshell 120 C18
column (4.6 x 50 mm, 5 μM). For quantitative HPLC analys
es of alkylation or cyclopropanation
reaction
s, c
alibration curves using
either
ethyl benzoate or 1,3,5-
trimethoxybenzene
as internal
standard
s were generated (Supplementary Figure
s 9 to 14
). The identity of enzymatic reaction
products was confirmed by HPLC or GC-
MS
co-injections of enzymatic
reaction mixtures with
chemically synthesi
zed authentic products,
and/or
by NMR analysis of products isolated from
reactions performed on preparative scale.
Enantiomeric products
were separated by
normal
-
phase
chiral
HPLC
on an Agilent 1200 series instrument with
isopropanol and hexane
as the
mobi
le phase
, using Chiral
pak
IA or
IC column
s (4.6 x 2
50 mm, 5 μm).
Analytical gas
chromatography coupled to mass spectrometry (GC
-MS) was carried out on a Shimadzu GCMS
-
QP2010SE
instrument equipped with a J&W HP
-5m
s column
. For quantitative GC-
MS
analys
es
of cyclopropanation
reaction products, calibration curves using 1,3,5-
trimethoxybenzene
as
internal standard
s were generated (Supplementary Figure 15)
.
Cloning and site
-saturation
mutagenesis.
pET22b(+) (Novagen)
with the pelB leader sequence
removed
was us
ed as a cloning
and expression vector for all constructs
described in this study.
S3
Site
-satur
ation
mutagenesis
libraries were generated using
the
22c
-trick method;
1
random
mutagenesis libraries were cloned using Taq polyme
rase (NEB) and var
ious
concentrations of
MnCl
2
(typically, 100
-400 μM) to steer polymerase error rate. P
rimers were obtained from IDT
(primer sequences are available upon request)
. PCR
was performed using Phusion polymerase
(NEB) and the resulting PCR products were digested with DpnI
(NEB)
, gel purified
(Zymo
Research
), repaired using the method of Gibson,
2
and used to directly transform
E. coli
strain
BL21(DE3)
(Lucigen) by electroporation.
Following electroporation, cells were recovered for
45
min at 37º
C in Luria
-Bertani (LB) medium, aliquots were plated on LB agar plates
supplemented with 100 μg/mL ampicillin
(LB
-Amp plates)
, and plates were incubated at 37
º C
overnight.
High
-throughput spectrophotometric l
ibrary screening in 96
-well plate format.
Single
colonies of
E. coli
BL21(DE3)
cells transformed with site
-saturation
or random mutagenesis
libraries
(or
cells transformed with
plasmid encoding the corresponding parent variant) were
picked with sterile toothpicks and cultured in 96-
deep
-well plates in LB
-Amp
medium
(300 μ
L/well) at 37 ºC, 250 r
pm, overnight. In a fresh 96-
deep
-well plate, H
yperbroth medium
(AthenaES
, 950
μL/well,
supplemented with
100 μg/mL
ampicillin
(HB
-Amp)
) was inoculated
with the pre
-cultures (50 μ
L/well) and incubated at 37 ºC, 250 rpm, for 2.5 h. The plates were
cooled on
ice for 20 minutes and then induced with 0.5 mM i
sopropyl β
-
D
-1-
thiogalactopyranoside
(IPTG
) and 1.0 mM 5-
aminolevulinic acid (ALA)
(final concentrations).
Expression was conducted at 24
ºC, 200 rpm for 16-
20 h. The cells were pelleted (3,000 ×
g,
5 min),
the supernatant
was
discarded, and the 96-
well plates were transferred to an anaerobic
chamber
(generations 1a to 2, see Supplementary Table 2 below
; as well as all evolutions for
substrate scope shown in Figure 4
in the main text
) or reactions were set u
p at the bench under
standard atmosphere (generations 3 to 6)
. Cell pellets were resuspended in reaction buffer (M9
-N
with 20 mM
D
-glucose
, 370
μL/well
. M9-
N minimal medium is 47.7 mM Na
2
HPO
4
, 22.0 mM
KH
2
PO
4
, 8.6 mM NaCl, 2.0 mM MgSO
4
, and 0.1 mM CaCl
2
, adjusted to
pH
7.4 at room
temperature
), and stocks of olefin (15 μ
L/well, in
EtOH or DMSO) and diazo reagent
(15 μ
L/well, in EtOH) were added. Substrate concentration typically
varied
, depending on the
reaction tested, between
10 and 30 mM for both heterocycles
and diazo reagent
. Following
substrate addition the plates were sealed with aluminum foil
and shaken at 500
rpm
, room
S4
temperature, overnight
. After overnight incubation, the seal was removed and
reactions were
worked up for either HPLC or GC
-MS analysis. For HPLC analysis, acetonitrile (typically 400 to
600 μL/well) supplemented with
internal standard was added
, the plates were resealed, briefly
mixed by vortexing, and incubated for 30 min to 1 h at room temperature. The plates were then
centrifuged (5,000
× g, 10 min), and the cleared supernatant was filtered through an AcroPrep
96-
well filter plate (0.2
μm cutoff) into a shallow
-well plate for HPLC analysis.
For GC-
MS
analysis, 400 μL of a 1:1 mixture of ethyl acetate and cyclohexane were added per we
ll, the
plates were re-
sealed, and vortexed vigorously for 2 min. The plates were then centrifuged
(5,000
× g, 10 min),
and 200 μL of the upper organic phase were transferred to 2 mL vials fit
ted
with 300 μL glass inserts for GC
-MS analysis.
Small-
scale
alkylation or cyclopropanation bioconversions using whole cells.
Single colonies
of
E. coli
BL21(DE3) cells transformed with plasmid encoding P411 variants were grown
overnight in 3 mL LB
-Amp medium
at 3
7 ºC and 250 rpm. 2 mL of the pre
-cultures were
used t
o
inoculate 48
mL of HB
-Amp medium
in 125 mL Erlenmeyer flasks
. Cultures were incubated at
37 ºC, 230 rpm for 2 h, typically reaching an OD
600
= 1.5. Cultures were then cooled on ice
(20 min) and induced with 0.5 mM IPTG and 1.0 mM ALA
(final concentration
s). Expression
was conducted at 2
4 ºC, 130 rpm, for 16-
18 h. Cultures were then cooled on ice, centrifuged
(3000
× g,
3 min, 4 ºC)
, and the pellets were resuspended to OD
600
= 30 in M9-
N minimal
medium. To determine P411 expression levels, aliquots of the cell suspension were
subjected to
carbon monoxide binding spectroscopy assay specified below (page S6).
Small
-scale biocatalytic alkylation or cyclopropanation
reactions were set up in 2 mL crimp vials
with
400 μ
L reaction volume. Typically, 370 μL of P411-
expressing cells at OD
600
= 30 in M9-
N
buffer
were
added to the vials and
transferred into an anaerobic chamber
(no sparging of cell
suspensions with argon was conducted as reported previously; transfer of the cell solution to the
anaerobic chamber, inclu
ding several cycles of applying vacuum and purging with N
2
, proved
sufficient for degassing). Subsequently, the vials were put on a shaker (500 rpm), and 15 μL of
heterocycle or internal
olefin stock solution (in DMSO or EtOH) were added, followed by 15 μL
of diazo stock solution (in EtOH).
Final concentrations
, depending on the substrates tested,
typically
ranged from
5 to 40
mM olefin
and
10 to 60 mM EDA, with 7.5 % co
-solvent.
For
aerobic alkylation reactions, reactions were set up as specified above on the bench under a
S5
standard atmosphere. Following addition of substrates t
he vials were sealed
and shaken at room
temperature, 50
0 rpm for 6–16 h.
For sample work-
up for HPLC analysis t
he reactions were
quenched by adding acetonitrile supplemented with inte
rnal standard
s (ethyl benzoate
or 1,3,5-
trimethoxybenzene
). Th
e sample solutions were incubated for 30 min,
transferred to
1.5
mL
tube
s and centrifuged at 13,000
× g for 10 minutes. The cleared supernatant was transferred to
clean 2 mL
vial
s and analyzed by HPLC.
For sample work-
up for GC
-MS or chiral HPLC
analysis, the reactions were quenched by adding 400 μ
L of a 1:1 solution of ethyl acetate and
cyclohexane, internal standard was added, and the mixture was transferred to 1.5 mL tubes,
followed by vigorous vortexing for 2 min. Samples were then centrifuged (10 min, 13,000 ×
g)
and 200 μ
L of the upper organic phase were transferred to 2 mL vials fitted with 300 μL glass
inserts, followed by GC
-MS or chiral HPLC analysis.
Determination of reaction kinetics
and
initial rates.
Kinetics
of 1
-methylindole alkylation
were determined using
whole
E. coli
cells
expressing P411 variants.
Reactions were set up as
described above (
Small
-scale alkylation or cyclopropanation bio-
conversions using whole cells
).
Following
addition of substrates, r
eactions were quenched in defined time
intervals by addition
of 500 μL acetonitrile, followed by vigorous mixing. The vials were removed from the anaerobic
chamber and f
urther sample work
-up and HPLC analysis was performed as described above.
Determination of
in vivo
P411 protein stability by
urea titration
. P411-
expressing whole
E.
coli
cells were prepared as described above. Subsequently, 700 μL of cells at OD
600
= 30 in
M9
-N buffer were transferred to a 1.7 mL tube, and urea stock solution (10 M) was added to
varying final concentrations (0, 1, 2, 3 or 4 M); the total volume was adjusted to 1.2 ml with
M9
–N buffer. The resulting cell suspensions were incubated at 25°C for 30 min in a
thermomixer (Eppendorf) at 500 rpm shake rate. Subsequently, the tubes were centrifuged
(1000 ×
g, 1 min) to pellet the cells, the supernatant was discarded, and pellets were washed in
800 uL M9-
N buffer. The centrifugation step was repeated, the supernatant was discarded, and
cells were resuspended in 800 μL M9
-N buffer. Of this cell suspension, 2 ×
370 μL were
transferred to 2 mL crimp vials, and reactions for 1-
methylindole alkylation were set up and
analyzed as described above.
S6
Preparative scale reactions.
E. coli
BL21(DE3) cells freshly transformed with plasmid
encoding P411 variants were grown overnight in 25 mL LB
-Amp
(37 ºC, 250 rpm).
HB-
Amp
medium
(500 mL) in a 2.8 L flask was inoculated with 20 mL of the pre
-culture and incubated at
37 ºC, 230 rpm for 2 h (to OD
600
ca. 1.5). Cultures were then cooled in an ice
-water bath for
20
min and induced with 0.5 m
M IPTG and 1.0 mM ALA
(final concentrations). Expression was
conducted at 24
ºC, 140 rpm, for 16-
20 h. Cultures
were then centrifuged (5,000 ×
g, 5 min,
4 ºC) and the cell pellets were resuspended to OD
600
= 40 in M9-
N buffer supplemented with
20
mM
D
-gluc
ose (typically yielding ca. 110 mL of cell suspension). Aliquots of the cell
suspension (4 mL) were used to determine the P411 expression level after cell lysis by
sonication. Aliquots of the cell suspension (28
mL) were then transferred to 50 mL Erlenmeye
r
flasks and degassed by sparging with argon for at least 30 minutes. The reaction flasks were then
transferred into an anaerobic chamber and olefin stock solution (1 mL, in DMSO) and EDA
(1 mL, in EtOH) were added. Final concentrations were typically 10
to 20 mM
heterocycle or
internal
olefin and 15
to
30 mM EDA, with 6.7% co
-solvent. The flasks were sealed with
parafilm, removed from the anaerobic chamber, and shaken at room temperature, 100 rpm for
18 h. The reactions were quenched by adding acetonitrile
(25 mL) and then centrifuged
(4,000 ×
g, 10
min). The supernatant was concentrated
in vacuo
to remove acetonitrile and
extracted with EtOAc or 1:1 EtOAc:cyclohexane (3
× 25 mL). The organic layers were washed
with brine (20
mL), dried over MgSO
4
, filtered
, concentrated, and purified by silica gel
chromatography on a Biotage Isolera instrument
.
Determination of P411 concentration.
Aliquots (4 mL) of the same
E.
coli
cell suspensions
used for whole
-cell biocatalytic reactions were lysed by sonication (QSoni
ca, 1 min sonication
time, 1s on/off cycles, 40 % output). 1 mL aliquots of the cell lysate were cleared by
centrifugation (13,000 ×
g, 10 min, 4°C), and
the
concentration of P411 enzymes in
the lysate
was determined from ferrous carbon monoxide binding di
fference spectra using the previously
reported extinction coefficient for serine
-
ligated enzymes (ε = 103,000 M
–1
cm
–1
). The
concentration of purified P411 enzymes was determined by quantifying the amount of free
hemin using the pyridine/hemochrome assay using the corresponding extinction coefficient
= 191,500 M
–1
cm
–1
)
4
.
S7
Protein purification.
E. coli
BL21(DE3) cells freshly
transformed with plasmid encoding P411
variants were grown overnight in 25 mL LB
-Amp
(30 ºC, 250 rpm).
HB
-Amp
medium (
500
mL)
in a 2.8
L flask was inoculated with 2
0 mL of the pre
-culture and inc
ubated at 37 ºC, 230 rpm for
2 h (
to
OD
600
ca. 1.5). Cultures were then cooled in an
ice
-water bath for 2
0 min and induced
with 0.5 mM IPTG and 1.0 mM ALA
(final concentrations). Expression was conducted at 24
ºC,
140 rpm, for 16-
20 h. Cultur
es were then centrifuged (5,000 ×
g,
10
min, 4 ºC) and the cell
pellets
were
frozen at –20 ºC. For protein
purification, frozen cells from two such cultures were
resuspended in buffer A (25 mM Tris
-HCl
, 20 mM
imidazole, 100 mM NaCl, pH 7.5;
4 mL
buffer per g of cell wet weight), supplemented with one protease inhibitor tablet (Roche
),
1 mg/mL lysozyme, and 0.1 mg/mL DNAse I.
The cell suspensions were lysed by sonication
(1 min sonication time, 1s on/off cycles, 40 % output
). To pellet insoluble material, cel
l lysates
were centrifuged (20,000
× g, 20 min, 4 ºC)
, followed by filtering the cleared lysate through a
2 μ
M syringe
-driven filter unit.
P411 proteins
were purified from the lysate using a nickel NTA
column (1 mL HisTrap HP, GE Healthcare) using an AKTAxpress purifier FPLC system (GE
Healthcare). P411 enzymes were eluted on a linear gradient from 100% buffer A to 100% buffer
B
(25 mM
Tris
-HCl
, 300 mM imidazole, 100 mM NaCl, pH 7.5) over 10 column volumes.
Fractions containing eluted protein were pooled and subjected to three rounds of buffer exchange
to storage buffer (25 mM Tris
-HCl,
100
mM NaCl, pH 7.5) using centrifugal spin filters (10 kDa
molecular weight cut
-off, Amicon Ultra). Subsequently, the concentrate
d protein was aliquoted,
flash
-frozen on pow
dered
dry ice, and stored at –
80 ºC. Protein concentrations were determined
using the pyridine/hemochrome assay specified above.
Alkylation
reactions
using purified protein.
Portions of
M9
-N buffer
(290
μL) and NADPH
(40
μL,
50 mM
in M9-
N) or sodium dithi
onite
(40
μL, 50 mM in M9-
N)
, or multiples thereof,
were combined in a 6 mL crimp vial and degassed by sparging with argon for at least
30
minutes.
Purified protein solutions were adjusted to 25
μM in M9
-N. After degassing was
complete, the M9
-N/reductant
solution and
purified protein stocks
were brought into the
anaerobic chamber. Reactions were set up on a 400 μ
L scale in 2 mL crimp vials: first, 3
40
μL
of the M9-
N/reductant
solution were added per vial, followed by 40 μ
L of purified protein
solution.
Nex
t, 10
μL of heterocycle or internal
olefin
stock solution were added, followed by
10
μL of EDA stock solution
. Final concentrations were typically 5 mM heterocycle or internal
S8
olefin
, 7.5 mM EDA
, 5 mM
reductant
, and
2.
5 μM P411
, with 5% co
-solvent
. The via
ls were
sealed, removed from the anaerobic chamber, and shaken at room temperature, 40 rpm for 1 to
6
h. The reactions were quenched by adding acetonitrile (400 μL)
supplemented with
internal
standard. This mixture was transferred to a
1.5 mL
tube and cent
rifuged at 13,000
× g for
10
minutes. The supernatant was transferred to 2 mL
vial
s and analyzed by HPLC.
Chemical synthesis of alkylation or cyclopropanation authentic product standards
. The
general protocol followed to synthesize authentic product standards is provided. In a fume hood,
a 25 mL round-
bottom flask with a stir bar was evacuated and back
-filled
with argon three times.
10 mL of anhydrous dichloromethane were added, followed by 0.02 mmol (
9 mg)
of rhodium
(II)
acetate dimer. While stirring, 1
.0
mmol of neat heterocycle or internal olefin were added, and
1.5
mmol of neat diazo compound were added over 2
h using a syringe pump. Following
addition of the diazo
compound, the reaction mixture was stirred at room temperature under
argon
over 16
h. The
reaction mixture was concentrated
in vacuo
, and the crude mixture
was
purified by silica flash chromatography
using ethyl acetate and hexanes as mobile phase
on a
Biotage Isolera instrument.
S9
II. Supplementary Tables
Supplementary Table 1
.
Hemoprotein variant
screening
for
1- methylindole alkylation
.
Hemoprotein
variant
a
Description
, m
utations
, references
Yield 3
[%]
TTN
Hemin
b
Free hemin in buffer, no protein
9
18
E. coli
BL21(DE3)
Whole
E. coli
cells, n
o protein
overexpression
4
n.d.
P450
BM3
Wild
-type P450
BM3
2
22
P450
BM3
T268A
C400S
P450
BM3
with mutations T268A and
C400S
3
7
183
P450
BM3
-CIS
P450
BM3
with mutations V78A F87V
P142S T175I A184V S226R H236Q
E252G T268A A290V L353V I366V
E4
42K
3
5
53
P411
BM3
-CIS
P450
BM3
with mutations V78A F87V
P142S T175I A184V S226R H236Q
E252G T268A A290V L353V I366V
C400S E442K
3
14
691
HStar H92N
H100N
P450
BM3
with mutations V78M H92N
H100N L181V T268A C400H L437W
4,5
3
570
P4
50-
Cam
P450cam from
Pseudomonas putida
6
3
273
Cyp119
Cyp119 from
Sulfolobus acidocaldarius
7
6
n.d.
S10
a
Reactions
were
performed
on 400 μL scale
with
whole
E. coli
cells
expressing the respective
hemo
protein variants
, at
10
mM substrate
loading
(1-methylindole; with 15 mM EDA)
. R
esults
are the average of duplicate reactions. TTN (total turnover numbers) are defined as HPLC
-
determined product concentrations divided by hemoprotein concentrations as measured by CO
binding difference spectroscopy. n.d
. = not
detected.
b
Reaction with hemin: in the anaerobic
chamber,
10
μL of hemin solution (
2 mM in DMSO) w
ere
added to 3
50
μL of argon-
sparged
M9
-N
buffer
, followed by
20
μL of sodium dithionite as reductant (
100 mM in M9-
N buffer),
1- methylindole
(1
0 μL, 400 mM in DMSO) and EDA (
1
0 μL,
600 mM in EtOH). Final
concentrations were 5
0 μM hemin,
5 mM sodium dithionite, 10 mM 1-
methylindole
and
15
mM
EDA with 5
% co
-solvent.
S11
Supplementary Table 2.
Summary of direc
ted evolution for 1-
methylindole alkylation
.
Round
Parent variant
Libraries evaluated
a
Screening
substrates
Mutation
s
identified
1a
P411
BM3
-CIS
Single site
-saturation at
positions V87, L181,
I263, L437 and T438
1- methylindole,
EDA, anaerobic
V87T,
L181G/I
/L/R,
I263M/G/W
,
L437G/Q,
T438C/T
1b
P411
BM3
-CIS
Random mutagenesis of
reductase domain
(residues 460-
1048)
1- methylindole,
EDA, anaerobic
Various stop
codons in
region 620-
720,
e.g. Q674*
2
P411
BM3
-CIS
Recombination of
mutations
identified in
rounds 1a and 1b
b
: V87T,
L181G/I/L/R,
I263M/G/W
, L437G/Q,
T438C/T, Q674*
1- methylindole,
EDA, anaerobic
V87T, L181G,
I263M,
V281L
c
,
T438C, Q674*
3
P411
BM3
-CIS
V87T L181G
I263M
V281L
T438C
Q674*
Random mutagenesis of
heme domain
, followed by
recombination of
improve
d variants
d
1- methylindole,
EDA, aerobic
N201S, L215Q
4
P411
BM3
-CIS
V87T L181G
N201S L215Q
I263M V281L
T438C Q674*
Random mutagenesis of
FMN domain (residues
460-
674)
1- methylindole,
EDA, aerobic
K472T
, N573D
,
F646S
5a
P411
BM3
-CIS
Random mutagenesis of
1- methylindole,
No further hits
S12
V87T L181G
N201S L215Q
I263M V281L
T438C K472T
N573D F646S
Q674*
entire heme
-
FMN domain
open reading frame
(residues 1
-674)
EDA
, aerobic
identified
5b
P411
BM3
-CIS
V87T L181G
N201S L215Q
I263M V281L
T438C K472T
N573D F646S
Q6
74*
Double site
-saturation
at
positions H92_H100 and
F391_Q403; single site
-
saturation
at positions
L75, A82 and L437
1- methylindole,
EDA, aerobic
H92F
6
P411
BM3
-CIS
V87T H92F
L181G N201S
L215Q I263M
V281L T438C
K472T N573D
F646S
Q674*
Double site
-saturat
ion
at
positions T327
and A328
1- methylindole,
EDA, aerobic
T327P A328S
Final variant:
P411
-
H
F
= P411
BM3
-
CIS
V87T H92F L181G N201S L215Q I263M V281L
T438C K472T N573D F646S
Q674*
Best variant for
aerobic
conditions:
P411
BM3
-CIS V87T H92F L181G N201S L215Q
I263M
V281L T327P A328S T438C K472T N573D F646S
Q674* (= P411-
HF T327P A328S)
a
Some residues were saturated twice, in different parent variants.
b
The corresponding wild-
type
residues at each position were included in the recombination library.
c
Mutat
ion
V281L
arose
as
an additional random
mutation in the best variant identified from the recombination library; this
position was subsequently subjected to site
-saturation mutagenesis, confirming 281L as the
optimal residue.
d
Library screening identified an improved va
riant with point mutations N201T
S13
and L215V
. For recombining these mutations, a library was cloned and tested with an expanded
codon set at both positions, consisting of a WNT codon at position 201 (encoding C, F, I, N, T,
Y, S) and a VWM codon at position 215 (encoding D, E, H, K, N, Q, I, L, V). This resulted in an
improved variant with mutations N201S and L215Q.
Supplementary Table 3
.
Summary of 1
-methy
lpyrrole
alkylation with EDA (Figure 4A
).
Variant
a
Substrate
Loading (mM)
Yield (%)
P411 (μM)
TTN
Ratio
4b
:4a
P411-
HF
10
3.8 ± 0.6
2 ± 0.15
184 ± 23.3
74:26
P411-
HF M263T
10
4.8 ± 0.4
2.1 ± 0.1
230 ± 13.4
49:51
P411-
HF M263T
G181E
10
12.4 ± 0.6
2.7 ± 0.1
458 ± 9.9
17:83
P411-
HF M263T
G181E A82T
10
20.7 ± 0.9
2.9 ± 0.12
705 ± 6.9
9:91
P411-
HF M263T
T87W
10
7.2 ± 0.3
3.0 ± 0.1
242 ± 4.7
93:7
P411-
HF M263T
T87W G181K
10
19.3 ± 0.8
3.9 ± 0.05
485 ± 18.1
98:2
a
Reactions were performed with P411-
expressing
E. coli
at OD
600
=30, at RT, for 16 h.
S14
Supplementary Table 4
.
Summary of 1
-methy
lindole alkylation with Me
-EDA (Figure 4B
).
Variant
a
Substrate
Loading (mM)
Yield (%)
P411 (μM)
TTN
P411-
HF M263Y
10
< 0.1
1.7 ± 0.02
2 ± 0.8
P411-
HF M263Y
T87C
10
0.1 ± 0.01
1.5 ± 0.06
5 ± 0.7
P411-
HF T87C
Y263E
10
0.3 ± 0.05
1.7 ± 0.04
15 ± 3.0
P411-
HF T87C
Y263E A328Y
10
0.7 ± 0.08
2.4 ± 0.06
31 ± 3.6
P411-
HF T87C
Y263E A328Y
L437M
10
1.0 ± 0.15
2.1 ± 0.09
47 ± 7.3
P411-
HF T87C
Y263E A328Y
L437M A268G
10
3.1 ± 0.24
1.7 ± 0.06
181 ± 12.6
P411-
HF T87C
Y263E A328Y
L437M A268G
T327P
10
6.0 ± 0.26
3.1 ± 0.09
193 ± 5.4
P411-
HF Y263E
A328Y L437M
A268G T327P
C87A
10
22.9 ± 2.43
3.0 ± 0.05
776 ± 79.4
a
Reactions were performed with P411-
expressing
E. coli
at OD
600
=30, at RT, for 16 h.
S15
Supplementary Table 5
.
Summary of
indene cyclopropanation with EDA (
Figure 4C
).
Variant
a
Substrate
Loading (mM)
Yield
(%)
P411
(μM)
TTN
dr
ee
(%)
P411-
CIS
10
5 ± 0.5
1.7 ± 0.12
307 ± 23
69:31
73
P411-
HF
10
39 ± 3.0
2.1 ± 0.15
1881 ± 59
95:5
94
P411-
HF M263Y
10
49 ± 4.9
1.3 ± 0.13
3996 ± 43
97:3
96
a
Reactions were per
formed with P411-
expressing
E. coli
at OD
600
=30, at RT, for 16 h.
Supplementary Table 6
.
Summary of 1,3-
cyclohexadiene cyclopropanation with EDA (Figure
4C
).
Variant
a
Substrate
Loading (mM)
Yield (%)
P411
(μM)
TTN
dr
ee
(%)
P411-
CIS
10
0.2 ± 0.02
1.8 ± 0.29
11 ± 1.4
90:10
n.d.
P411-
HF
10
0.3 ± 0.04
2.0 ± 0.22
14 ± 1.1
89:11
n.d.
P411-
HF
M263
W G181L
10
3.2 ± 1.4
3.2 ± 0.54
98 ± 27
91:9
73
a
Reactions were performed with P411-
expressing
E. coli
at OD
600
=30, at RT, for 16 h.
S16
III Supplementar
y Figure
s
Supplementary Figure 1
:
Development of a UV
-Vis spectrophotometric high
-throughput
screening assay for 1
-methylindole alkylation.
(A) Directed evolution screening reaction using 1
-methylindole
1
and EDA
2
to yield the
C
3
-
alkylated product
3a
. (B) Abso
rbance of
1
and
3a
, both
at 1 mM concentration,
in 1:1
H
2
O:MeCN,
showing a pronounced red-
shift for the alkylation product
3a
. (C) Exemplary data
for the UV
-Vis high-
throughput alkylation screen. Shown are 96 individual whole
-cell reactions
of either rando
m mutagenesis library variants (grey dots), P411-
CIS (parent variant, red dots, in
quadruplicate) and P450
-BM3 (negative control, black dots, in quadruplicate). Absorbance at
305 nm indicates enzymatic activity for catalytic production of
3a
.
Potentially improved variants
(showing higher absorption than the corresponding parent variants) were subsequently analyzed
by HPLC to verify and quantify higher biocatalytic activity for the production of
3a
.
Availability
of a rapid UV-
Vis
screening system is notewort
hy: optimization of hemoproteins for most
carbene transfer reactions reported to date typically required measurement of enzymatic activity
and/or selectivity by liquid chromatography (LC) or gas chromatography (GC). As these
techniques are time-
consuming (
several minutes per run, or several hours per 96-
well plate),
evolutionary strategies to optimize carbene transferases typically employed iterative, single site-
saturation mutagenesis of active-
site residues, thereby keeping library sizes sufficiently smal
l for
the limited throughput of LC or GC. However, for P450-
BM3, a protein of 1048 residues, this
neglects a considerable amount of protein sequence space.
Thus, we were pleased to have the
S17
high-
throughput spectrophotometric indole alkylation assay for
fac
ile screening of several
thousand enzyme variants per generation. This screening system enabled us t
o search for
improved carbene transfer activity across the entire P450
-BM3
protein sequence space
using
random mutagenesis
of both the
heme and reductase domains, complemented by site
-saturation
mutagenesis of selected heme domain active site residues.
Supplementary Figure 2
:
Position of mutations in P411-
HF.
Structural m
odel of wild
-type
P450
-BM3
heme and FMN domains (
pdb
: 1BVY
8
). The heme
domain is shown on the left in light gray, while the FMN domain is on the right in da
rk gray;
both heme and FMN co-
factor
s are shown in red. The heme domain active site cavity is on the
left
side of the heme co
-factor. The FAD domain is missing in both P411-
HF and in this crystal
structure (included here are heme domain residues 20
-458 and
FMN domain residues 479-
630).
Residues mutated
in P411-
HF are shown in blue; note that P411-
CIS, the evolutionary starting
point, carries a F87V mutation, however, since the
structure
shown here
is derived from P450-
BM3, F is shown at position 87. P411-
HF mutations K472T and F646S are not shown, as these
residues are not included in the crystal structure.
S18
Supplementary Figure
3:
Sequential substrate addition and substrate pre
-incubation experiments
.
(A) Sequential substrate addition experiment: Equivalents
of 1
-methylindole and EDA
(10 mM
each)
were added to P411-
HF whole
-cell reactions every 30 min, up t
o a substrate loading of 60
mM. For reactions receiving less substrate, solvent was added at the indicated time points to
ensure equal volume and solvent concentration across all samples. Samples were worked
-up and
analyzed 2 h after the last substrate add
ition. The data indicate that substrate availability is not
limiting enzymatic turnover numbers. (B) Substrate pre
-incubation experiment: P411-
HF
expressing whole cells were pre
-incubated with either 1
-methylindole (30 mM) or EDA (45 mM)
for 15 min, before
adding the corresponding second substrate. Activity was compared to whole
-
cell reactions receiving both substrate simultaneously. 1-
Methylin
dole pre
-incubation caused an
8% reduction in activity, while EDA pre
-incubation resulted in a 26% activity loss, i
ndicati
ve of
enzyme inactivation by alkylation. Data shown in (A) and (B) were derived from reactions
performed in duplicate; error bars indicated standard deviation.
S19
Supplementary Figure
4:
Investigating the effects of FAD domain truncation.
(A)
Com
parison of P411-
CIS and P411-
CIS Q674* yield and TTN for the alkylation of
1- methylindole (10 mM) with EDA (15 mM) to yield product
3a
. (B) Comparison of P411-
VAC
cis
and P411-
VAC
cis
Q674* for the cyclopropanation of
N
-vinylphthalimide (30 mM) with
EDA (30
mM
; product analysis by HPLC was performed as previously described
10
), conducted
with cells at OD
600
=10. (C) Urea titration assay to assess
in vivo
P411 variant stability.
11,12
E. coli
cells expressing the indicated P41
1 variants were treated with different concentrations of
the protein denaturant urea, followed by assessing the remaining 1-
methylindole alkylation
activity.
Data shown in (A) to (C) were derived from two independent assays with reactions
performed in dupl
icate; error bars indicated standard deviation.
S20
Supplementary Figure 5:
In vitro
1 -methylindole alkylation activity of purified P411 variants
.
In vitro
1-methylindole alkylation activity of purified P411 variants
was tested using either
NADPH or sodiu
m dithionite as reductant, or no externally added reductant. A
ssays were
conducted with 5
mM 1
-methylindole, 7.5 mM EDA, 5
mM reductant and 2.5 μM enzyme
in
M9
-N buffer
. The highest catalytic
activity, observed with P411-
HF and NADPH as reductant,
correspo
nds to 94%
product yield.
S21
Supplementary Figure 6:
Evolution of P411-
HF for 1
-methylpyrrole alkylation with EDA.
Evolutionary lineage from P411-
HF over three generations to yield variants selective for
C
2
or
C
3
1- methylpyrrole
alkylation
with EDA. Dat
a were derived from two independent experiments
performed in duplicate; assays were performed with 10 mM substrate loading (1
-methylpyrrole
and EDA).
See Supplementary Table 3 for experimental details.