Synthesis of azobenzenealkylmaleimide probes to photocontrol the enzyme activity of a bacterial histone deacetylase-like amidohydrolase

Article abstract of DOI:10.1016/j.bioorg.2014.10.004

A series of azobenzenealkylmaleimides (AMDs) with different spacer length was synthesized and coupled via Michael-Addition to a specific mutant of a bacterial histone deacetylase-like amidohydrolase (HDAH). Michaelis-Menten parameters (V_max and K_m) were employed to characterize the effect of both, the spacer length and the configuration (cis vs. trans) of the attached azobenzene moiety, on the HDAH enzyme activity. The photoswitch behavior of the AMD/enzyme conjugate activity was clearly influenced by the AMD spacer length. This study highlights the importance of steric rearrangement of the photoswitch with respect to the active site and describes a strategy to optimize the photocontrol of HDAH.

Full text of DOI:10.1016/j.bioorg.2014.10.004

Bioorganic Chemistry 57 (2014) 155–161
Contents lists available at ScienceDirect
Bioorganic Chemistry
journal homepage: www.elsevier.com/locate/bioorg
Synthesis of azobenzenealkylmaleimide probes to photocontrol the
enzyme activity of a bacterial histone deacetylase-like amidohydrolase
Benjamin Horstmann , Michael Korbus a,1, Tatjana Friedmann , Christiane Wolff ,
a,1
a
b
b
a,
⇑
Christina Marie Thiele , Franz-Josef Meyer-Almes
a
Department of Chemical Engineering and Biotechnology, University of Applied Sciences Darmstadt, 64287 Darmstadt, Germany
Clemens-Schöpf Institute of Organic Chemistry and Biochemistry, Technical University of Darmstadt, 64287 Darmstadt, Germany
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 6 August 2014
Available online 24 October 2014
A series of azobenzenealkylmaleimides (AMDs) with different spacer length was synthesized and coupled
via Michael-Addition to a specific mutant of a bacterial histone deacetylase-like amidohydrolase (HDAH).
Michaelis–Menten parameters (Vmax and K ) were employed to characterize the effect of both, the spacer
m
length and the configuration (cis vs. trans) of the attached azobenzene moiety, on the HDAH enzyme
activity. The photoswitch behavior of the AMD/enzyme conjugate activity was clearly influenced by
the AMD spacer length. This study highlights the importance of steric rearrangement of the photoswitch
with respect to the active site and describes a strategy to optimize the photocontrol of HDAH.
Ó 2014 Elsevier Inc. All rights reserved.
Keywords:
Deacetylation
External stimulus
Photochromism
Protein engineering
Photoswitch
Azobenzenes
Maleimides
Methylenemaleimideamide
Hofmann rearrangement
1
. Introduction
area [15,16]; this elongation can be achieved by the use of different
diamines. These advantages were combined by the synthesis of a
Bioconjugation of light-switchable molecules is a widely used
method to introduce stimulus sensitive moieties to biomolecules
1–3]. The isomerization process of light-responsive molecules
series of new azobenzenealkylmaleimides (AMDs) 1a–e. Subse-
quently, the reactive probes were conjugated to a specific cys-
teine-variant (M30C) of a photoswitchable bacterial histone
[
offers diversified applications like photocontrol of the protein func-
tion [4–8] or the conformational change of protein structures
deacetylase-like amidohydrolase (HDAH) from Bordetella/Alcalige-
nes strain FB188 [7] to investigate the influence of AMD spacer
length variation as well as the cis- and trans-configuration on the
photocontrol behavior of the enzyme activity. Furthermore, spatio-
temporal controlled deacetylation reactions could be employed in
biotechnological and biomedical applications to regulate the
deacetylation of biomolecules or chemical compounds in a precise
and defined manner to meet desired needs.
[
9,10] as well as protein assembly [11]. Azobenzene is the most
studied light-switchable molecule. It undergoes a reversible isom-
erization process from the E- to the Z-isomer by irradiating with
UV-light. The back-isomerization to the lower energy E-isomer
(trans) can be realized by visible light (vis) or by thermal relaxation
[
12,13]. In addition to the configuration, the polarity of the azoben-
zene group is also switched. Both factors play an important role if
azobenzene containing molecules are covalently bound to a pro-
tein surface. Bioconjugation of small molecules requires mild con-
ditions and an orthogonal coupling strategy whereupon the
Michael-Addition towards the thiol group of cysteine residues ful-
fills these conditions [14]. The variation of linker length is a com-
mon motif to probe the influence of a specific group to a defined
2
. Materials and methods
2.1. General methods
All chemical reagents were of analytical grade, obtained from
commercial suppliers and used without further purification. THF
was freshly distilled under argon from sodium and stored over
sodium and under argon atmosphere. TEA was distilled and stored
over KOH. Column chromatography was carried out on silica gel 60
⇑
Corresponding author.
E-mail address: franz-josef.meyer-almes@h-da.de (F.-J. Meyer-Almes).
1
First author.
(40–60 lm), Macherey-Nagel. Reactions and chromatography
http://dx.doi.org/10.1016/j.bioorg.2014.10.004
045-2068/Ó 2014 Elsevier Inc. All rights reserved.
0

1
56
B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
ꢁ1
ꢁ
1
fractions were analyzed on precoated silica gel plates (silica gel 60,
F254, 20 ꢀ 20 cm, 0.25 mm thick, Merck thin-layer chromatogra-
phy). Amino and Boc protected amino groups were stained by dip-
ping the TLC plate in a ninhydrin solution (0.66% m/v in EtOH)
following by heating or by absorbance of UV light at 254 nm.
NMR-spectra were recorded with an Avance 300 (AC300)
l mol cm ; 330 nm) and the predicted e of HDAH wild type
ꢁ1
ꢁ1
(wt) (43,890 l mol cm ; 280 nm).
2.2.5. Determination of IC50 values
To determine the inhibitory effect of the cis- and trans-configu-
ration of 4-PAM and AMDs (1a–e) on HDAH variant C51S activity,
dose–response curves were performed and IC50 values analyzed.
The HDAH C51S activity was determined by employing a fluores-
cence-based activity assay in a similar way as described previously
1
13
1
(
300 MHz H, 75 MHz C) and a Bruker DRX 500 (500 MHz H,
1
3
1
26 MHz C)-spectrometer. The chemical shift data for each signal
is given in units of d (ppm) relative to tetramethylsilane (TMS),
assigned d (TMS) = 0. Coupling constants (J) are stated in Hz.
HPLC–MS was performed with an Agilent Technologies Series
[
17,18]. 5 mM 4-PAM or AMD (1a–e) DMSO solutions were isomer-
ized to their cis-configuration by UV light irradiation (364 nm,
1
200 instrument using an Agilent Eclipse XDB-C18 column.
ꢁ2
4
3 mJ cm ) using a hand lamp as described before and serially
Method: 5–95% B from 2 to 17 min, 95–5% B from 20 to 25 min.
Solvent A: water + 0.1% formic acid; solvent B: acetonitrile + 0.1%
formic acid. The cis- and trans photoisomerization of AMDs (1a–e)
conjugated to HDAH variant M30C was verified by UV/Vis spec-
troscopy measurements in MC buffer (40 mM sodium phosphate,
diluted in FB188 buffer (15 mM Tris, 50 mM KH
2 4 2 4
PO /K HPO ,
2
50 mM NaCl, 250 M EDTA, 0.001% Pluronic F 68 and 0.005%
l
bovine serum albumin; pH 8.0). The trans-configuration of men-
tioned compounds was achieved by thermal relaxation of the seri-
ally diluted samples due to incubation at 30 °C for 5 h in the dark.
The HDAH catalyzed deacetylation reaction, in dependence on
increasing 4-PAM or AMD concentrations, was performed as a
one-step assay in the dark in black 96-well half-area microplates
1
50 mM NaCl, 1 mM EDTA, pH 7.2) at 30 °C. A solution of 3 lM
of the respective AMD/HDAH M30C conjugate was irradiated by
UV light by use of a hand lamp (6 W, Heraeus) at 364 nm for
ꢁ
2
2
5 min at a distance of 6 cm (43 mJ cm ) or blue light (1 W,
(
flat bottom, polystyrene, medium binding; Greiner Bio-One) pro-
viding final concentrations of 100 nM HDAH variant C51S,
.5 mg/mL Trypsin from bovine pancreas (Serva) and 50 M BocLy-
ꢁ2
LED, Osram) at 464 nm for 25 min at a distance of 6 cm (1.5 J cm
)
to achieve the cis- or trans-configuration. UV/Vis spectra (280–
00 nm) of photoisomerized samples were subsequently recorded
with a UV/Vis spectrophotometer (Jasco V-630).
0
l
4
s(Ac)AMC (Bachem) as substrate. The fluorescence intensity of 7-
amino-4-methylcoumarin (AMC) was detected at 450 nm
(
350 nm excitation) over a period of 2 h at 30 °C and the HDAH
2
2
.2. Biochemical methods
C51S activity determined from the slope of the linear initial phase
of measured kinetics and expressed as relative fluorescence units
ꢁ1
.2.1. Employed HDAH variants
HDAH variant C51S was used to investigate the inhibitory effect
per second (RFU s ). The latter was correlated to the activity of
a sample of HDAH C51S where no compound was added (stated
as 100%), the activity expressed as percentages and plotted against
the log of applied compound concentration. Dose–response curves
were analyzed by the program GraphPad Prism (GraphPad Soft-
ware) using a three-parameter dose–response model and IC50-val-
ues determined in independent triplicates and specified as
means ± SD, n = 3. Dose–response curves of DMSO were equally
determined as described before using HDAH-wt to verify the real
inhibitory effect, induced by 4-PAM and AMDs (1a–e) on HDAH
activity. Additionally, a dose–response curve of a known histone
deacetylase inhibitor like suberoylanilide hydroxamic acid (SAHA)
was tested on HDAH-wt activity to compare the inhibitory potency
with 4-PAM and AMDs. The dose–response curve of SAHA was per-
formed as a single determination by use of the above described
assay conditions and data points comparably analyzed by a
three-parameter dose–response model.
of the respective cis- and trans-configuration of (E)-1-(4-(phen-
yldiazenyl)phenyl)-1H-pyrrole-2,5-dione (4-PAM) and AMDs (1a–e)
by determination of IC50 values. HDAH variant M30C was used as
conjugate and test model to investigate the AMD dependent influ-
ence on the photoswitch capability with respect to the spacer
length. Both HDAH variants were generated by site-directed muta-
genesis, expressed in Escherichia coli strain XL1-blue and purified
as previously reported [7].
2.2.2. Modification of HDAH variant M30C
The single solvent accessible cysteine of HDAH variant M30C
was chemically modified by 4-PAM or AMDs (1a–e) by a 4-fold
molar excess in MC buffer (supplemented with 8% DMSO) and
the modification procedure continued as described previously [7].
2.2.3. Determination of Michaelis–Menten kinetics
The determination and analysis of Michaelis–Menten kinetics of
3. Experimental procedure
unmodified and AMD (1a–e) modified HDAH M30C variants,
ꢁ
2
photoisomerized to their cis- (UV light, 364 nm, 43 mJ cm ) and
3.1. Synthesis of AMD 1b–e
ꢁ
2
trans-configuration (blue light, 464 nm, 1.5 J cm ), was performed
as reported previously [7]. Pluronic F 68 was purchased from BASF
and bovine serum albumin from Sigma–Aldrich. Measurements
were performed as independent triple measurements and Michae-
The procedure of the synthesis of compounds 3c–e, 4c–e, 5c–e
1
13
and 7 as well as the H NMR and C NMR spectra of AMDs 1a–e
can be found in supplementary material.
lis–Menten parameters (K
dard deviation (SD), n = 3. The photoswitch efficiency of Vmax- and
-values, between the cis- and trans-configuration, was calcu-
lated as percentages by definition of the more active isomer.
m
and Vmax) represented as means ± stan-
3
.1.1. (E)-N-(2-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)ethyl)-4-(phen-
K
m
yldiazenyl)benzamide (1b)
To a solution of 1-(2-aminoethyl)-1H-pyrrol-2,5-dione 2,2,2-
hydrochloride: (150 mg, 0.849 mmol) and TEA (350 lL, 2.50 mmol)
2
.2.4. DOL value determination
The DOL value (average number of conjugated AMD molecules
in THF (4 mL) was added (E)-perfluorophenyl 4-(phenyldiaze-
nyl)benzoat 7 (666 mg, 1.70 mmol) in THF (4 mL) dropwise. After
16 h at rt, the solvent was evaporated, the resulting residue was
per each HDAH variant M30C molecule) calculation of AMD/HDAH
M30C variant conjugates was performed as described previously
dissolved in CHCl
(30 mL) and brine (30 mL). The organic layer was dried over
Na SO , filtered and concentrated. The crude product was purified
by silica column (eluent EtOAc:hexane: 1:4 ? 1:1; R : 0.10 ? 0.50)
3 2
and washed with H O (30 mL), 1 M HCl
[
7] via Lambert–Beer equation by use of the determined molar
extinction coefficient of trans 4-PAM at its absorbance
maximum after conjugation to HDAH variant M30C (15,700
(e)
2
4
f

B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
157
Scheme 1. Reaction scheme for the synthesis of AMDs 1b–e: (i) Boc
2
O, DCM; (ii) 1. Maleic anhydride, DCM; 2. Ac
2
O, NaOAc, 65 °C, 3 h; (iii) DCM:TFA (2:1); (iv)
perfluorophenyl 2,2,2-trifluoroacetate, THF; (v) TEA, THF. Commercial available 1-(2-aminoethyl)-1H-pyrrole-2,5-dione hydrochloride (5b) was used for the synthesis of 1b.
Scheme 2. Reaction scheme for the synthesis of the geminal amidomaleimide 1a:
Fig. 1. Chemical modification of HDAH variant M30C by AMDs. HDAH variant
(
2
i) glycineamide hydrochloride, DMF, TEA; (ii) PIFA, DMSO:H O (1:1); (iii) 1. Maleic
M30C was employed as a conjugation model to investigate the effect of the alkyl
spacer length (n = 1, 2, 3, 4 and 6) on the photoswitch capability of enzyme activity.
AMDs (1a–e) were covalently conjugated to the mutated cysteine at amino acid
position 30, which was original a methionine before (M30C, orange). The HDAH
variant C51S, where a cysteine was substituted by serine, was used to test the
inhibitory effect of AMDs by determination of IC50 values. The latter mentioned
mutation was also present in HDAH variant M30C to ensure a precise modification
of AMDs at the mentioned cysteine. The catalytic zinc ion (Zn2 , red sphere) displays
the active site which is complexed by one histidine (H182) and two aspartate
residues (D180 and D268). The illustration of mutated cysteines is based on the X-
ray crystal of HDAH-wt (PDB ID 2VCG, chain A, [24]). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of
this article.)
anhydride, DIEA, DCM; 2. EDC, HOBt, DMF.
to yield the product as an orange solid (243 mg; 82%). 1H NMR
500 MHz, DMSO) d 8.73 (t, J = 6.0 Hz, 1H), 7.96–7.88 (m, 6H),
.67–7.56 (m, 3H), 7.01 (s, 2H), 3.62 (t, J = 5.7 Hz, 2H), 3.45 (dd,
(
7
+
1
3
J = 11.6, 5.9 Hz, 2H); C NMR (126 MHz, DMSO) d 171.04, 165.76,
1
1
53.24, 151.89, 136.61, 134.51, 131.92, 129.49, 128.35, 122.67,
22.27, 37.71, 37.10; HPLC/MS (t = 13.92 min; ESI) found: 349.1
r
+
(
M+H) , calc.: 349.1.
3
.1.2. (E)-N-(3-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)propyl)-4-(phen-
was dried over Na
reduced pressure. The crude product was purified by column chro-
matography (EtOAc/hexane 1:1 ? 6:4; R : 0.42) to obtain 41 mg
5%) of the desired compound. H NMR (500 MHz, DMSO) d 8.65
t, J = 5.6 Hz, 1H), 8.15–7.89 (m, 6H), 7.68–7.59 (m, 3H), 7.04 (s,
2 4
SO , filtered and the solvent was removed under
yldiazenyl)benzamide (1c)
To a solution of 1-(4-aminopropyl)-1H-pyrrole-2,5-dione 2,2,2-
trifluoracetat salt (5c) (608 mg, 2.28 mmol) and TEA (950
.84 mmol) in THF (14 mL) was added dropwise (E)-perfluorophe-
nyl 4-(phenyldiazenyl)benzoate (1.79 g, 4.56 mmol) in THF
14 mL). After 16 h at rt, the solvent was evaporated, the resulting
residue was dissolved in CHCl (200 mL) and washed with H
100 mL), 1 M HCl (100 mL) and brine (100 mL). The organic layer
lL,
f
1
(
(
6
2
2
1
H), 3.51 (t, J = 7.2 Hz, 2H), 3.34–3.27 (m, 2H), 1.83 (p, J = 7.2 Hz,
(
1
3
H);
C NMR (126 MHz, DMSO) d 171.51, 165.91, 153.73,
3
2
O
52.43, 137.19, 135.00, 132.44, 130.02, 128.93, 123.19, 122.81,
(

1
58
B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
3
7.51, 35.72, 28.52; HPLC/MS (t
r
= 14.30 min; ESI) found: 363.2
in 1.4 mL THF were added dropwise and the reaction stirred for
16 h. The solvent was removed, the residue was dissolved in
5 mL CHCl , washed with H O, 1 M HCl and brine. The organics
were dried over Na SO , filtered and concentrated. The crude prod-
uct was purified by silica column (eluent EtOAc:hexane: 1:4
1:1; R : 0.10 ? 0.50) to yield the product as an orange solid
70 mg; 61%). H NMR (500 MHz, CDCl
+
(
M+H) , calc.: 363.1.
2
3
2
2
4
3.1.3. (E)-N-(4-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)butyl)-4-(phen-
yldiazenyl) benzamide (1d)
?
(
f
To a solution of 1-(4-aminobutyl)-1H-pyrrole-2,5-dione 2,2,2-
1
3
) d 7.99–7.87 (m, 6H),
.64–7.42 (m, 3H), 6.66 (s, 2H), 6.49 (br s, 1H), 3.51 (t, J = 7.1 Hz,
H), 3.44 (dd, J = 13.0, 6.9 Hz, 2H), 1.67–1.55 (m, 4H), 1.46–1.38
trifluoracetate (5d) (104 mg, 0.370 mmol) and TEA (155
l
L,
.10 mmol) in THF (4 mL) was added dropwise (E)-perfluorophenyl
-(phenyldiazenyl)benzoate (7) (289 mg, 0.740 mmol) in THF
7
2
1
4
1
3
(
1
1
m, 2H), 1.35–1.29 (m, 2H); C NMR (126 MHz, CDCl
3
) d 170.98,
66.92, 154.25, 152.66, 136.66, 134.15, 131.62, 129.24, 128.02,
23.16, 122.97, 40.02, 37.64, 29.52, 28.45, 26.30, 26.26; HPLC/MS
(
3 mL). After 16 h at rt, the solvent was evaporated, the resulting
residue was dissolved in CHCl (40 mL) and washed with H
20 mL), 1 M HCl (20 mL) and brine (20 mL). The organic layer
was dried over Na SO , filtered and concentrated. Purification by
silica column (eluent: EtOAc:hexane: 1:4 ? 1:1; R : 0.10 ? 0.40)
afforded the product as an orange solid (56 mg; 45%). H NMR
500 MHz, CDCl ) d 8.08–7.87 (m, 6H), 7.60–7.50 (m, 3H), 6.73 (s,
H), 6.39 (br s, 1H), 3.62 (t, J = 6.9 Hz, 2H), 3.55 (dd, J = 12.6,
.5 Hz, 2H), 1.81–1.62 (m, 4H); 13C NMR (126 MHz, CDCl
) d
3
2
O
(
+
r
(t = 15.55 min; ESI) found: 405.2 (M+H) , calc.: 405.2.
2
4
f
1
3
3
1
.2. Synthesis of AMD 1a
(
3
2
6
1
1
.2.1. (E)-N-(2-amino-2-oxoethyl)-4-(phenyldiazenyl)benzamide (8)
To a solution of 2-aminoacetamide hydrochloride (16.8 mmol,
.86 g) in 84 mL DMF and TEA (4.67 mL, 33.8 mmol) was added a
3
70.98, 166.97, 154.39, 152.73, 136.48, 134.28, 131.69, 129.30,
28.05, 123.22, 123.08, 39.78, 37.51, 26.84, 26.30; HPLC/MS
solution of the active ester 7 (2.20 g, 5.61 mmol) in 28 mL DMF.
After 3 h the reaction mixture was extracted with CHCl . The
organic layer was washed with sat. NH Cl solution and brine, dried
over Na SO , filtered and concentrated. 200 mg of the desired com-
+
(t
r
= 14.56 min; ESI) found: 377.1 (M+H) , calc.: 377.1.
3
4
3.1.4. 6-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)hexyl-4-(phenyldiazenyl)-
2
4
benzamide (1e)
pound could be obtained. To raise the yield, the not completely dis-
solved emulsion (obtained by extraction) was precipitated in
hexane. The precipitates were filtered and dried. Overall 603 mg
N-(6-aminohexyl)maleimide (85 mg, 283
l
mol) was placed in a
L TEA (900 mol). (E)-
perfluorophenyl 4-(phenyldiazenyl)benzoate (222 mg, 570 mol)
flask and dissolved in 1.4 mL THF and 125
l
l
1
l
(38%) of the benzamide 8 could be obtained. H NMR (500 MHz,
Fig. 2. Michaelis–Menten kinetics of AMD/HDAH M30C conjugates. HDAH M30C was covalently modified by AMD 1a (A), 1b (B) and the respective cis- and trans-
configuration achieved by UV-(364 nm) or blue-light (464 nm) irradiation. (C) (for AMD-1a) and (D) (for AMD-1b) show the relative difference of the velocity value (V)
between both configurations. If no error bars are detectable, SD values are so small that they are indistinguishable from most data points. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this article.)

B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
159
DMSO) d 8.88 (t, J = 5.9 Hz, 1H), 8.15–8.10 (m, 2H), 8.01–7.93 (m,
H), 7.67–7.60 (m, 3H), 7.44 (s, 1H), 7.06 (s, 1H), 3.88 (d,
J = 5.9 Hz, 2H); C NMR (126 MHz, DMSO) d 171.33, 166.11,
amine is obtained as hydrochloride from the aqueous layer [21,22];
especially for unpolar compounds the reaction lead to small yields
[23]. The used cosolvent dissolved the azobenzeneglycineamide 8
and provided water which is required for the decarbonylation to
the amine. Finally, the maleimide group was introduced similar
to Scheme 1.
4
1
3
1
4
2
53.82, 152.43, 136.79, 132.46, 130.02, 129.16, 123.20, 122.77,
5.98; HPLC/MS (t = 14.01 min; ESI) found: 283.1 (M+H) , calc.:
r
83.1.
+
3
.2.2. (E)-N-(aminomethyl)-4-(phenyldiazenyl) benzamide
4
.2. Bioconjugation of AMD 1a–e to HDAH
hydrochloride (9)
Bis(trifluoroacetoxy) iodo)benzene (1.12 g, 2.73 mmol) was
dissolved in 14 mL DMSO and 3.8 mL H O. Afterwards 8 was added
and the mixture was stirred at rt for 16 h. The solution was subse-
quently poured in 1 M HCl and washed with Et O. The azobenzene
compounds did not completely dissolve in the aqueous layer. The
orange foam was taken and lyophilized to obtain 411 mg of a mix-
tures of 8 and 9 (44%:56%; n/n calculation via H NMR; 38% yield
(
The molar ratio of the average number of conjugated AMD mol-
2
ecules per each HDAH variant M30C molecule was determined by
degree of labeling (DOL) experiments as described in the Materials
and Methods section. The bioconjugation of AMDs to HDAH M30C
is shown graphically in Fig. 1 and was experimentally verified by
UV/Vis spectroscopy (see supplementary material Fig. A.1, cis/trans
configuration). AMDs 1a–d could be coupled to the HDAH variant
M30C by Michael addition in a 1:1 M ratio, which corresponded
to previous work [7] where the conjugation of 4-PAM was
2
1
1
for the 9). H NMR (300 MHz, DMSO) (selected data) d 9.83 (t,
J = 5.9 Hz, 1H), 8.42 (s, 3H), 8.19–8.07 (m, 2H), 8.06–7.88 (m, 4H),
7
.68–7.55 (m, 3H), 4.44 (d, J = 6.0 Hz, 2H).
reported. The more hydrophobic AMD 1e (n = 6) yielded
a
decreased DOL-value of 0.8 (see supplementary material
Table A.1), which was slightly lower than expected. However after
bioconjugation of 1e to HDAH variant M30C the enzyme activity
was almost completely abolished.
3
.2.3. (E)-N-((2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)methyl)-4-
(
phenyldiazenyl)benzamide (1a)
50 mg of a mixtures of 8 and 9 (0.685 mmol 9) was dissolved
in 16 mL DCM and 660 L DIPEA. Maleic anhydride (159 mg;
.62 mmol) was added and the mixture was stirred at rt for 4 h.
3
l
1
Afterwards 15 mL 1 M HCl was added, the aqueous layer was sep-
arated and the not completely dissolved orange layer was precipi-
tated in hexane and filtered to obtain 319 mg of an orange solid.
The unpurified (Z)-4-oxo-4-(((4-((E)-phenyldiazenyl)benzamido)
methyl)amino)but-2-enoic acid was dissolved in 15 mL DMF and
was added dropwise to a solution of HOBt ꢀ H
2
O (166 mg;
.09 mmol), EDC ꢀ HCl (208 mg; 1.09 mmol) and DIPEA (371 L;
.17 mmol) in 50 mL DMF. After 16 h the reaction was concen-
were added and the
O (3 times à 50 mL) and brine
, filtered and concentrated. The crude
product was purified by silica column (eluent EtOAc:hexane: 1:1;
1
2
l
trated to a volume of 10 mL, 50 mL CHCl
organic layer was washed with H
30 mL), dried over Na SO
3
2
(
2
4
1
R
f
: 0.38) to yield the product as an orange solid (62 mg; 27%).
NMR (500 MHz, DMSO) d 9.16 (t, J = 5.1 Hz, 1H), 8.08–8.02 (m,
H), 7.97–7.90 (m, 4H), 7.65–7.56 (m, 3H), 7.08 (s, 2H), 5.03 (d,
H
2
1
3
J = 5.2 Hz, 2H); C NMR (126 MHz, DMSO) d 170.08, 165.28,
1
4
3
53.46, 151.89, 134.80, 131.99, 129.50, 128.72, 122.70, 122.29,
2.69; HPLC/MS (t = 14.01 min; ESI) found: 335.1 (M+H) , calc.:
r
35.1.
+
4
. Results and discussion
4.1. Synthesis of AMDs
The synthesis procedure to achieve the desired AMDs is shown
in Schemes 1 and 2. AMDs 1b–e could be synthesized by starting
with selective monoprotection of the corresponding diamine 2c–e,
followed by an introduction of the maleimide moiety over two
steps. Maleic anhydride reacts with the primary amine followed
by cyclisation. After removal of the Boc-group aminoalkylmalei-
mides 5b–e were coupled to preactivated azobenzoic acid 7 to
obtain AMD 1b–e.
Due to the fact that monoprotected diaminomethane is not eas-
ily accessible [19,20] the reaction strategy for AMD 1a had to be
changed. The reaction of glycineamide hydrochloride with penta-
fluorophenyl 4-(2-phenyldiazenyl)benzoate 7 yielded the azoben-
zeneglycineamide 8. The geminale aminoamide hydrochloride 9
was obtained by acetic Hofmann rearrangement with [I,I-bis(triflu-
oroacetoxy)iodo] benzene (PIFA) in a DMSO and water mixture.
The rearrangement is typically performed in water and the desired
Fig. 3. Alkyl spacer dependent influence on Michaelis–Menten parameters of AMD/
HDAH M30C conjugates concerning cis- and trans-configurations. Michaelis–
Menten kinetics were performed and analyzed as described in the Materials and
Methods section. Vmax (A) and Km (B) values of the cis- and trans-configuration for
each tested AMD/HDAH M30C conjugate were respectively represented as relative
ratios of the AMD modified and corresponding unmodified HDAH M30C variant.
Values are means ± SD, n = 3. Statistical analysis for Vmax and K
m
values were
a
performed by using one-way ANOVA with Bonferroni post hoc test. P < 0.05 (cis
⁄
AMD vs. trans AMD). = Vmax
– photoswitchable AMD/HDAH M30C variant
ꢂ
conjugate. = K
m
– photoswitchable AMD/HDAH M30C variant conjugate.

1
60
B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
4.3. Michaelis–Menten kinetics
the trans- to the cis-configuration of the azobenzene moiety
depends on [S] (Fig. 2C and D). This demonstrates that the rate of
the enzyme reaction, which can be calculated using Vmax, is the
decisive index to compare the effectiveness of two enzyme activi-
ties on the same substrate concentration. In our case the enzyme
conjugate with AMD 1a in trans-configuration catalyzed the reac-
tion more effectively than the corresponding cis-configuration at
all substrate concentrations. In contrast, the enzyme reaction of
the trans-AMD 1b enzyme conjugate occurred at lower rates com-
pared with its cis-analog at all substrate concentrations. The photo-
isomerization from the trans- to the cis-configuration of AMD 1a
decreased the Vmax-value significantly by 35% whereas Vmax of
the AMD 1b conjugate was decreased by 54% upon cis–trans
photoswitching (Fig. 3A). Furthermore it could be observed that
The influence of the AMD spacer length with respect to the cis-
and trans-configuration was investigated by Michaelis–Menten
kinetics. Vmax- and K -values of the respective cis- and trans-iso-
m
mer of AMD (1a–e) modified HDAH M30C variants were calculated
relative to the corresponding unmodified HDAH M30C sample
Fig. 3 A and B). The half-life of the cis-configuration was estimated
to be at least 30 h for HDAH M30C/4-PAM [7] and HDAH M30C/1e
conjugates. The required time to perform the corresponding irradi-
ation of the samples and to measure Michaelis–Menten kinetics is
accomplished within 80 min. Therefore, the relaxation did not
influence the enzyme kinetics. The bioconjugation of AMD 1a
and b to HDAH M30C enabled a photoswitching of the rate of prod-
uct formation over a broad range of substrate concentrations
(
m
the K -values of AMDs 1a (44%) and 1b (84%) were significantly
(
Fig. 2A and B). During the past three decades, the ratio kcat/K
m
increased by cis-to-trans isomerization (Fig. 3B). The attachment
of AMDs 1c–e with an alkyl spacer length of P3 methylene groups
led to an almost complete inhibition of the HDAH M30C conjugate
in both configurations (residual activity < 6%) (Fig. 3A). By this rea-
or Vmax/K was frequently used to compare the catalytic efficiency
of different enzymes, e.g. mutated or otherwise modified, working
on the same substrate. However, Eisenthal et al. [25] clearly argued
m
that, the use of kcat/K
tiveness of enzymes is not only flawed, but also misleading in
many cases. Even if an enzyme has a higher kcat/K ratio, it can cat-
alyze the same reaction at lower rates than another enzyme (at the
same concentration [E] ) with lower kcat/K ratio at certain sub-
strate concentrations, [S]. Generally, neither the use of kcat/K
nor simply Vmax or K enables a fair comparison of two enzymes
at all substrate concentrations. Therefore, in this study both
and K , and their effect
m
as a quantitative index for catalytic effec-
m
son its enzyme activity could not be photoswitched. The K -value
of HDAH M30C was, compared to AMD 1a–b, increased by AMD
1c–e to a higher extent. Regarding the photoswitch behavior of
m
the K
alteration due to cis-to-trans isomerization (43%). Determined
max- and K -values were summarized in Table A.2 (see supple-
m
-values of AMD 1c–e, only AMD 1c showed a significant
0
m
m
V
m
m
mentary material) and performed Michaelis–Menten kinetics rep-
resented in Fig. A.2.
Michaelis parameters, Vmax = kcat ꢀ [E]
0
m
On the basis of both verified Michaelis–Menten parameters
on the rate of product formation where analyzed separately. The
change in Vmax turned out to have a more profound effect on the
(Vmax and K
effects on the HDAH variant M30C enzyme activity and the ability
to switch Vmax- and K -values by irradiation. Conjugation of AMDs
m
), AMDs with varying spacer length showed distinct
rate of the enzyme reaction than a change in K
m
. The relative
m
change in the product formation rates upon photoswitching from
(1d and e) led to an almost complete inhibition of the enzyme
Fig. 4. 600 MHz 1H NMR spectra of 4-PAM in DMSO-d
2
. 5 mM 4-PAM/DMSO-d
solutions were treated differently. (A) incubation at 80 °C for 1 h; (B) 25 min, 364 nm; (C)
6
6
5 min, 464 nm. 1H NMR spectroscopy was performed as described in the supplementary material. The annotation of 6 and 6 protons of cis- and trans-4-PAM for PSS
0
calculation is indicated by integral values. For analysis and quantification of PSS for 4-PAM, please see Table A.4.

B. Horstmann et al. / Bioorganic Chemistry 57 (2014) 155–161
161
whereas short spacer AMDs (1a and b) had a moderate influence
on the activity. Therefore, they are able to control the protein func-
tion due the variation of the configuration. Determined Vmax- and
K -values for the AMD 1b/HDAH M30C conjugate coincide very
m
well. The photoisomerization to cis (catalytic more active enzyme)
led to a decrease and the isomerization to trans (more inactive
azobenzene and the alkyl spacer length the enzyme activity
(Vmax) could be photocontrolled through two different mechanisms.
In one case the cis-conjugate was more active, whereas in the
other case the trans-conjugate showed a higher rate of product
formation. Three AMD variants led to a complete inhibition of
the enzyme in both azobenzene configurations. Even when a
suitable position for the introduction of a cysteine in close
proximity to the functional site of a protein could be identified
from a crystal structure, the exact chemical structure of a
conjugated photoswitch for optimal control of protein function is
very hard to predict. We suggest the newly synthesized series
of cysteine reactive AMDs as a general tool box to generate
photoswitchable conjugates and enable a systematic exploration
of the optimal intramolecular arrangement of the photoswitch
with respect to the functional site. This approach should be
applicable to other proteins, e.g. enzymes or ion channels, with
cysteine residues at specified and solvent exposed positions on
the protein surface.
m
enzyme) to an increase of the K -value (Fig. 3A and B), which cor-
responds also to previously published work [7]. In comparison to
AMD 1b, AMD 1a showed an inverse photoswitching behavior
where the trans-isomer enabled the more active form of the
enzyme conjugate. The observation that the trans-isomer of AMD
1
m
a showed a higher K -value but also a higher Vmax-value could
be explained by a scenario where cis 1a blocks the active site more
sterically than the trans-isomer but does not initiate large confor-
mational changes at the enzyme or a decrease of the substrate
affinity. In contrast, the more hydrophobic trans-isomer could
induce a more pronounced distortion of the active site. This sug-
gestion could explain the significant differences for the cis- and
trans-isomer in the observed photoswitching behavior of 1a and
b conjugates.
Acknowledgments
4.4. Determination of the photostationary state (PSS) of 4-PAM
This work was supported by a grant of the hessian
Landes-Offensive zur Entwicklung Wissenschaftlich-ökonomischer
Exzellenz (LOEWE Soft Control).
For the evaluation of the actual photoswitching efficiency of
AMD/HDAH M30C conjugates, we measured the PSS of 4-PAM,
1
the shortest employed maleimide azobenzene derivate, by
NMR spectroscopy in DMSO-d (Fig. 4). A blue light irradiation
464 nm) of 25 min resulted in a calculated cis-/trans-isomer ratio
H
Appendix A. Supplementary material
6
(
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.bioorg.2014.
of 24.7/75.3%. In case of UV light irradiation (364 nm, 25 min) a
cis-/trans-isomer ratio of 53.5/46.5% was calculated (Table A.4).
Under the assumption, that the PSS behavior in DMSO could be
transferred to the photoswitchable AMD 1a- or 1b-HDAH-M30C
conjugates in water, the following theoretical Vmax values were cal-
1
0.004. These data include MOL files and InChiKeys of the most
important compounds described in this article.
References
ꢁ
1
culated for the 100% cis- or trans-isomers: 4.59 nM s
for the
ꢁ1
trans- and 0.35 nM s for the cis-isomer of AMD 1a as well as
.02 nM s for the trans- and 6.47 nM s for the cis-isomer of
[1] I. Willner, S. Rubin, Angew. Chem., Int. Ed. Engl. 35 (1996) 367–385.
[2] C. Brieke, F. Rohrbach, A. Gottschalk, G. Mayer, A. Heckel, Angew. Chem., Int.
Ed. 51 (2012) 8446–8476.
ꢁ1
ꢁ1
0
AMD 1b, respectively. This means that the enzyme activity of both,
AMD 1a- and 1b-, conjugates would be theoretically switched
more substantially by 92% and 99% upon complete photoconver-
sion of the azobenzene group instead of the observed 35% and
[
[
[
[
[
3] W. Szymanski, J.M. Beierle, H.A. Kistemaker, W.A. Velema, B.L. Feringa, Chem.
Rev. 113 (2013) 6114–6178.
4] B. Schierling, A.-J. Noël, W. Wende, E. Volkov, E. Kubareva, T. Oretskaya, et al.,
Proc. Natl. Acad. Sci. 107 (2010) 1361–1366.
5] W.-C. Lin, C.M. Davenport, A. Mourot, D. Vytla, C.M. Smith, K.A. Medeiros, et al.,
ACS Chem. Biol. (2014).
6] T. Shimoboji, E. Larenas, T. Fowler, S. Kulkarni, A.S. Hoffman, P.S. Stayton, Proc.
Natl. Acad. Sci. 99 (2002) 16592–16596.
5
4%. The experimental determination of PSS of 4-PAM and the
AMDs conjugated to a HDAH variant dissolved in buffer, was not
possible by NMR due to solubility and concentration issues.
7] M. Korbus, G. Balasubramanian, F. Müller-Plathe, H. Kolmar, F.-J. Meyer-Almes,
Biol. Chem. (2014) 401–412.
[
[
8] J.H. Harvey, D. Trauner, ChemBioChem 9 (2008) 191–193.
9] L. Chi, O. Sadovski, G.A. Woolley, Bioconjug. Chem. 17 (2006) 670–676.
4.5. Inhibition of the HDAH variant C51S
[
[
[
10] S.-L. Dong, M. Löweneck, T.E. Schrader, W.J. Schreier, W. Zinth, L. Moroder,
et al., Chem.–Eur. J. 12 (2006) 1114–1120.
11] D. Hoersch, S.-H. Roh, W. Chiu, T. Kortemme, Nat. Nanotechnol. (2013) 928–
HDAH variant C51S was used to determine the IC50-values of 4-
PAM and AMDs in cis- and trans-configuration. No significant dif-
ferences with respect to the enzyme activity could be found
between the azobenzene derivatives (4-PAM and AMDs 1a–e)
and the respective configuration of the azobenzene group (see sup-
plementary material Table A.3). All tested azobenzene derivatives
932.
12] A.A. Beharry, G.A. Woolley, Chem. Soc. Rev. 40 (2011) 4422–4437.
[13] D. Bullock, C. Cumper, A. Vogel, J. Chem. Soc. (1965) 5316–5323.
[14] N. Stephanopoulos, M.B. Francis, Nat. Chem. Biol. 7 (2011) 876–884.
[
[
[
[
15] P. Gorostiza, M. Volgraf, R. Numano, S. Szobota, D. Trauner, E.Y. Isacoff, Proc.
Natl. Acad. Sci. 104 (2007) 10865–10870.
16] R. Numano, S. Szobota, A.Y. Lau, P. Gorostiza, M. Volgraf, B. Roux, et al., Proc.
Natl. Acad. Sci. 106 (2009) 6814–6819.
showed only a weak (<10 lM) inhibition on the HDAH C51S
enzyme activity and were at least 260 times less potent compared
to a typical pan-inhibitor like SAHA (see supplementary material
Fig. A.3 vs. Fig. A.3). The weak apparent inhibitory effect was
mainly attributed to DMSO (see supplementary material Fig. A.3)
which was used to prepare the stock solutions of the compound.
17] D. Wegener, C. Hildmann, D. Riester, A. Schwienhorst, Anal. Biochem. 321
(
2003) 202–208.
18] D. Wegener, F. Wirsching, D. Riester, A. Schwienhorst, Chem. Biol. 10 (2003)
61–68.
19] F.E. DeBons, G.M. Loudon, J. Org. Chem. 45 (1980) 1703–1704.
20] R.H. Boutin, G.M. Loudon, J. Org. Chem. 49 (1984) 4277–4284.
21] A.E. Sutton, J. Clardy, J. Am. Chem. Soc. 123 (2001) 9935–9946.
[
[
[
5
. Conclusion
[22] A.E. Sutton, J. Clardy, Org. Lett. 2 (2000) 319–321.
[
[
[
23] K.-S. Song, M.J. Kim, H.J. Seo, S.-H. Lee, M.E. Jung, S.-U. Kim, et al., Bioorg. Med.
Chem. 17 (2009) 3080–3092.
24] S. Schäfer, L. Saunders, E. Eliseeva, A. Velena, M. Jung, A. Schwienhorst, et al.,
Bioorg. Med. Chem. 16 (2008) 2011–2033.
A novel series of photoswitchable maleimide derivatives was
synthesized and covalently coupled to a specific variant of HDAH
close to the active site. Depending on the configuration of
25] R. Eisenthal, M.J. Danson, D.W. Hough, Trends Biotechnol. 25 (2007) 247–249.