Structural optimization of photoswitch ligands for surface attachment of α-chymotrypsin and regulation of its surface binding

Article abstract of DOI:10.1002/chem.200903369

A modified gold surface that allows photoregulated binding of a-chymotrypsin has previously been reported. Here the development of this surface is reported, through the synthesis of a series of trifluoromethyl ketones and a-keto esters containing the azobenzene group and a surface attachment group as photoswitch inhibitors of a-chymotrypsin. All of the compounds are inhibitors of the enzyme, with activity that can be modulated by photoisomerization. The best photoswitch shows a reversible change in IC ₅₀ inhibition constant of >5.3 times on photoisomerization. The trifluoromethyl ketone 1 exhibited excellent photoswitching and was attached to a gold surface in a two-step procedure involving an azide-alkyne cycloaddition. The resulting modified surface bound α-chymotrypsin to a degree that could be modulated by UV/Vis irradiation, showing "slow-tight" enzyme binding as observed for inhibitors in solution.

Full text of DOI:10.1002/chem.200903369

FULL PAPER
DOI: 10.1002/chem.200903369
Structural Optimization of Photoswitch Ligands for Surface Attachment of
a-Chymotrypsin and Regulation of Its Surface Binding
David Pearson[b] and Andrew D. Abell*[a]
Abstract: A modified gold surface that
allows photoregulated binding of a-
chymotrypsin has previously been re-
ported. Here the development of this
surface is reported, through the synthe-
sis of a series of trifluoromethyl ke-
tones and a-keto esters containing the
azobenzene group and a surface attach-
ment group as photoswitch inhibitors
of a-chymotrypsin. All of the com-
pounds are inhibitors of the enzyme,
with activity that can be modulated by
photoisomerization. The best photo-
switch shows a reversible change in
IC50 inhibition constant of >5.3 times
on photoisomerization. The trifluoro-
methyl ketone 1 exhibited excellent
photoswitching and was attached to a
gold surface in a two-step procedure
involving an azide–alkyne cycloaddi-
tion. The resulting modified surface
bound a-chymotrypsin to a degree that
could be modulated by UV/Vis irradia-
tion, showing “slow-tight” enzyme
binding as observed for inhibitors in
solution.
Keywords: azo compounds
zyme inhibitors · photoswitching ·
surface chemistry surface
plasmon resonance
· en-
·
Introduction
biomolecules to surfaces in predictable and controllable
ways. With this in mind we recently published a communica-
tion reporting the photoreversible binding of a protease (a
class of enzyme that accepts a range of related peptide-
based substrates) to a surface-bound inhibitor.[7] Reversible
surface photoregulation was demonstrated for a-chymotryp-
sin with compound 1 (Figure 1), which contains a protease
inhibition group (a phenylalanine-based trifluoromethyl
ketone for specific targeting of a-chymotrypsin), an azoben-
zene switch (to allow reversible control of shape), an oligo-
ethylene glycol (OEG) tether (to extend the inhibitor into
solution), and a terminal alkyne for attachment to a surface-
bound azide through a Huisgen cycloaddition reaction.[8,9]
Photochemical irradiation with UV (or visible) light reversi-
bly controls the geometry of the azobenzene constituent and
hence protease binding to the surface, as shown schematical-
ly in Figure 1.[7] This modular design is amenable to surface
attachment of a range of peptide-based structures to allow
targeted binding of a range of proteases and other biological
receptors: that is, it provides a general approach to a photo-
switchable surface. a-Chymotrypsin was chosen for study
because it is particularly well characterized and thus repre-
sents a prototypical protease.[10]
Photoisomerization of the azobenzene group provides a
platform for substantial control of biological activity, to give
a variety of potential applications.[1,2] The incorporation of
this functionality into a (photo)switchable surface, for exam-
ple, shows much promise for the development of “intelli-
gent” nanoscale biosystems.[3] Photoresponsive surfaces that
are able to modulate the binding and release of molecules
(an enzyme, for example) in reversible fashion offer poten-
tial for the development of new biosensors, advanced medi-
cal implants, biomolecular computers, and advanced protein
purification procedures based on photoaffinity chromatogra-
phy. However, most such systems reported to date[4–6] have
lacked the versatility to merit development beyond initial
feasibility studies. In order to make advances in this area we
require general approaches for photoswitchable binding of
[a] Prof. A. D. Abell
School of Chemistry and Physics, The University of Adelaide
Adelaide, SA 5005 (Australia)
Fax : (+61)8-8303-4380
E-mail: andrew.abell@adelaide.edu.au
Here we report the preparation and testing of a number
of trifluoromethyl ketone and a-keto ester photoswitch in-
hibitors (1–3, 5–8, Figures 2 and 3) of a-chymotrypsin.
These studies have begun to define, for the first time, the
optimum inhibitory group and C-terminal acetylene-based
[b] Dr. D. Pearson
Department of Chemistry and Biochemistry
Ludwig-Maximilian University, 80539 Munich (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200903369.
Chem. Eur. J. 2010, 16, 6983 – 6992
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6983

ble surfaces based on these
structures. The general design
of these compounds is based on
the affinity of a-chymotrypsin
for inhibitors/substrates con-
taining aromatic amino acids in
their P1 positions and large hy-
drophobic groups in their P2 po-
sitions.[11] As such, the phenyl-
A
trifluoro-
methyl ketone/a-keto ester and
the azobenzene group are ex-
pected to bind to the S1 and
S2 subsites of the enzyme, re-
spectively, whereas the OEG
group should extend from the
peptide recognition site for sur-
face attachment without affect-
ing enzyme binding. The azo-
benzene units employed are
similar to those previously used
in a range of photoswitch pep-
tides.[12–15] Previous studies[11]
have found that substitution of
an azobenzene (as required for
attachment of the OEG group)
can significantly affect photo-
switching, but little is known
with regard to suitable sub-
strates for surface studies. A
range of substituents were
therefore investigated here.
Figure 1. a-Chymotrypsin photoswitch. a) Inhibitor (E)-1. b) Schematic of surface photoswitching showing sur-
face-attached (E)-1 and a-chymotrypsin-bound surface-attached (Z)-1.
The trifluoromethyl ketones
1 and 2 each contain an extend-
ed OEG linker that attaches
the terminal alkyne and the
azobenzene, through a second
acetylene and an amide, respec-
tively. In comparison, analogue
3 has the terminal alkyne di-
rectly attached to the azoben-
zene core. The alkyne and
amide groups were investigated
because they present signifi-
cantly different structures and
because of their initial ease of
synthesis as linkers. The a-keto
esters 6 and 7, each with an al-
ternative inhibitory group to
the otherwise identical trifluo-
ACHUTNGRENrNUG oACTHNGUTRENmUNG ethyl ketones 1 and 3, re-
spectively, allow comparative
studies to begin to identify opti-
mum groups for surface attach-
Figure 2. Trifluoromethyl ketone photoswitch inhibitors.
linker for simultaneous optimization of enzyme binding,
photoswitching, and surface attachment. This is an impor-
tant step in the generation of more versatile photoswitcha-
ment and photoswitchable binding. The trifluoromethyl
ketone 5 provides a solution-based model for surface-bound
3, whereas the previously reported trifluoromethyl ketone
6984
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 6983 – 6992

Photoswitch Ligands
FULL PAPER
10[21] in the presence of base (DIEA) gave the amide 11 in
38% yield. The alcohol 12 was treated with Boc-Ala-OH in
the presence of 1-ethyl-3-(3’-dimethylaminopropyl)carbodi-
AHCTUNGTREGiNNUN mide (EDCI) to give the amide 13 in 94% yield. This was
then treated with TFA to give the primary amine 14, which
was treated with the acid chloride 11 in the presence of base
(DIEA) to give the amide 15 in 70% yield over two steps.
The alcohol 15 was finally oxidized to the ketone 2 with
Dess–Martin periodinane[22] in quantitative yield.
The trifluoromethyl ketone 3 and the a-keto ester 7 were
each prepared by means of a Sonogashira coupling reaction
as the key step (Scheme 2). The bromoazobenzene 16[23] was
coupled to 2-methylbut-3-yn-2-ol in the presence of [Pd-
AHCTUNGTRENNG(UN PPh3)2Cl2] to give the alkyne 17 in 93% yield. Both pro-
tecting groups were removed on hydrolysis with aqueous
NaOH to give 18, which was separately treated with 10 and
with the a-hydroxy ester 20[24] in the presence of EDCI as in
similar syntheses[25] to give the amides 19 and 21, respective-
ly. Oxidation with Dess–Martin periodinane gave the ke-
tones 3 and 7, both in 91% yield. The triazole 5 was ob-
tained by treatment of the azide 22 with the alkyne 3 in the
presence of CuI catalyst formed in situ from CuSO4 and
sodium ascorbate.[26]
The extended derivatives 1 and 6 were prepared in a simi-
lar manner (Schemes 3 and 4, below). The dialkyne 23[27]
was treated with TMSCl to give the monoprotected alkyne
24 in 26% yield. Treatment of 24 with the bromoazoben-
Figure 3. a-Keto ester photoswitch inhibitors.
4[16] and the a-keto ester analogue 8 provide further deriva-
tives for comparative solution-based studies (see below).
The extended C-terminal OEG linkers of 1, 2, and 6 were
not expected to affect enzyme binding (or photoswitching)
significantly because of the low specificity of a-chymotryp-
sin beyond the S3 subsite.[11,17] However, the binding of a-
chymotrypsin to compounds 3 and 7 after surface attach-
ment might be expected to be significantly different, due to
the likely participation of the
zene 16 in the presence of [PdACTHUNTRGNEUNG(PPh3)2Cl2] and CuI was at-
tempted, but the desired product 26 was not obtained, possi-
bly due to homocoupling of the alkyne 24.[28] However,
less extended terminal alkyne/
triazole group in enzyme bind-
ing. This paper thus begins to
define the optimum chemical
make-up of a photoswitchable
peptide for surface attachment,
which is critical if one is to
begin to develop the generality
of the approach. Additional
work on extending our prelimi-
nary surface findings with fur-
ther analysis and discussion on
the binding of a-chymotrypsin
to surface-attached 1 is also re-
ported.
Results and Discussion
The trifluoromethyl ketone 2
was prepared from the symmet-
rical
azobenzenedicarboxylic
acid chloride 9[18,19] and com-
pound 12[20] as shown in
Scheme 1. Treatment of the
acid chloride 9 with the amine Scheme 1. Preparation of the trifluoromethyl ketone 2.
Chem. Eur. J. 2010, 16, 6983 – 6992
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6985

A. D. Abell and D. Pearson
Photoisomerization and in-
hibition assays were initially
carried out in solution to assess
the abilities of the designed
compounds to photoswitch and
to inhibit the enzyme. The com-
pounds 1–8 were irradiated
with UV light (320–380 nm) to
obtain Z-enriched photostation-
ary states. The Z-enriched
states were then irradiated with
visible light (>360 nm) to
return the E-enriched photosta-
1
tionary states. H NMR of these
samples showed good photoiso-
merization (see Table 1) from
an initial state of <10% Z
isomer to 66–93% Z after UV
irradiation and back to 8–
21% Z after visible irradiation,
with no observed decomposi-
tion (see Table 1).
The nature of the azobenzene
substituent strongly influences
the makeup of the photosta-
tionary states; compare, for ex-
ample, compounds
2 and 4,
which have 93% and 66% of
the Z isomer, respectively, after
UV irradiation. These com-
pounds are identical except in
the nature of the attached azo-
benzene substituent. In con-
trast, the nature of the electro-
philic “warhead” (enzyme-bind-
ing group) has little effect in
this regard (compare compound
pairs 1/6, 3/7, 4/8 in Table 1,
with common azobenzene sub-
stituents, but different protease
Scheme 2. Preparation of the trifluoromethyl ketones 3 and 5 and the a-keto ester 7.
treatment of 24 with the corresponding iodide 25[29] in the
presence of [Pd(PPh3)2Cl2] gave the alkyne 26 in 49% yield.
binding groups at their C termini). This observation is as
would be expected, because photoisomerization is depen-
dent on the electronic transitions of the azobenzene group,
which are affected by the electron-withdrawing/donating ef-
fects of substituents.[31,32] Photoswitching is influenced both
by the overlap of the UV/Vis absorption bands with the
UV/Vis filters used (selected on the basis of good wave-
length overlap with the appropriate bands; data not shown)
and by the relative stabilities of the E and Z isomers. Each
of the compounds studied here contains an electron-with-
drawing amide substituent on one aromatic ring and a varia-
ble substituent on the other (alkyne, amide, CH2-amide, or
triazole). Whereas electron-donating substituents are known
to red-shift absorption bands, often leading to better irradia-
tion overlap, the electron-withdrawing or neutral substitu-
ents used here cannot be assumed to have a major effect on
photoswitching without analysis of UV/Vis spectra. The best
ACHTUNGTRENNUNG
CuI was not added because copper has been reported to
promote alkyne homocoupling.[30] NaOH hydrolysis of 26
gave the carboxylic acid 27 in 64% yield, and this was treat-
ed with the amine 10 in the presence of EDCI to give the
amide 28 in 80% yield. Oxidation of 28 with Dess–Martin
periodinane gave the desired trifluoromethyl ketone 1 in
90% yield.
The carboxylic acid 27 was also treated with the amine 20
(Scheme 4) to give the amide 29 in 58% yield. This was oxi-
dized with Dess–Martin periodinane to give the a-keto ester
6 in 51% yield.
Coupling of 30[11,16] with the amine 20 in the presence of
EDCI, followed by oxidation of the resulting alcohol 31
with Dess–Martin periodinane, gave the a-keto ester 8
(Scheme 5).
6986
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 6983 – 6992

Photoswitch Ligands
FULL PAPER
compound with similar substitu-
ents,[14] but poor Z!E photo-
switching.
The potential photoswitch in-
hibitors 1–8 were assayed
against a-chymotrypsin as their
initial E-enriched states, as
their Z-enriched PSSs resulting
from UV irradiation, and as
their E-enriched PSSs after visi-
ble irradiation. The correspond-
ing IC50 values were then calcu-
lated.
All the compounds were
found to inhibit a-chymotryp-
sin, with IC50 values as shown
in Table 2. Whereas the a-keto
esters (6–8) exhibited normal
inhibition characteristics, the in-
hibition by the trifluoromethyl
ketones (1–5) appeared to in-
crease over time. This slow in-
Scheme 3. Preparation of the trifluoromethyl ketone 1.
crease in binding strength is
consistent with the “slow-tight”
binding mode.[34–36] In this case, it is likely that the slow
binding is due to the active trifluoromethyl ketone species
existing in equilibrium with their hydrates.[34] These hydrates
were observed in the NMR spectra of selected trifluoro-
methyl ketones in (wet) acetone. To record steady-state hy-
drolysis rates, trifluoromethyl ketones were allowed to incu-
bate with enzyme and substrate for 8 min before rate data
were collected. Consequently, the IC50 values given for the
trifluoromethyl ketones reflect the final “tight” binding. In
all cases, the initial binding of trifluoromethyl ketones was
too weak for accurate IC50 values to be obtained.
A range of structure/activity/photoswitching comparisons
can be drawn from the data in Table 2. The a-keto esters
(6–8) are more potent inhibitors than the trifluoromethyl
ketones (1–5), with inhibition constants ranging from 0.23 to
2.2 mm for the a-keto esters and 12 to >85 mm for the tri-
fluoromethyl ketones. This suggests that the a-keto ester
group binds more rapidly and/or strongly to the Ser195 resi-
due[17] of the active site. Both a-keto esters and trifluoro-
methyl ketones are more active as the Z-enriched PSSs ob-
tained from UV irradiation, as previously reported for these
classes of photoswitch inhibitors[16,37] and for a related bo-
Scheme 4. Preparation of the a-keto ester 6.
photoswitching (as measured by the decrease in the amount
of Z isomer on visible irradiation) was exhibited by the al-
kynes 1 and 7. Compounds with alkyne groups in the 4’-po-
sition (1, 3, 6, and 7) generally gave better photoswitching
than compounds with methylene (4 and 8) or triazole (5)
groups in the 4’-position. Alkyne substituents have been
shown to have no significant effect on UV/Vis absorption.[33]
However, the conjugating effect of the triple bond may alter
the influence of the electron-withdrawing amide substituent.
Comparisons with literature data are difficult, due to the
wide variation in approaches to data reporting, but it can be
noted that the CH2-amide compounds 4 and 8 gave similarly
good E!Z photoswitching to that previously reported for a
AHCTUNGTRENNUNG
ing of the trifluoromethyl ketones are very different from
those of the a-keto esters, despite the fact that both inhibit
serine proteases through their constituent electrophilic car-
bonyl groups. In particular, the Z-enriched forms of the tri-
fluoromethyl ketones exhibit a narrow range of inhibition
constants (12–17 mm) whereas the E-enriched forms vary
considerably (19–85 mm), Therefore, the 4’-substituents have
little impact on the potencies of the Z-trifluoromethyl ke-
tones, but a major effect on those of the E-trifluoromethyl
Chem. Eur. J. 2010, 16, 6983 – 6992
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6987

A. D. Abell and D. Pearson
ketones. In comparison, the in-
hibition constants of both the
E- and the Z-enriched states of
the a-keto esters vary signifi-
cantly between compounds,
showing that the natures of the
4’-substituents affect the poten-
cies of both the E- and the Z-a-
keto esters. As a result, the dif-
ferences in activity between the
E and Z isomers of the inhibi-
tors (i.e., the extent of enzyme
photoswitching) vary more be-
tween individual trifluorometh-
yl ketones than between indi-
vidual a-keto esters. For exam-
Scheme 5. Preparation of the a-keto ester 8.
ple, the best and the worst pho-
toswitching results are observed
for trifluoromethyl ketones (3/1
and 2, respectively), whereas all of the a-keto esters exhibit
intermediate photoswitching of 1.8- to 3.0-fold changes in
activity between isomers.
Table 1. E/Z compositions obtained by UV/Vis photoisomerization of azo-
benzenes 1–8.
Compound Inhibitor type Amount
Amount
of Z
Amount
of Z
Decrease
in Z
A comparison of the 4’-azobenzene substituents in the tri-
fluoromethyl ketones shows that the strongest binding is ex-
hibited by the triazole 5 and the weakest by the alkynes 1
and 3. Similarly, the alkynes 6 and 7 exhibit the weakest
binding of the a-keto esters. However, the best enzyme pho-
toswitching is exhibited by alkynes 1 and 3, which give
>3.6- and >5.3-fold changes in activity on photoisomeriza-
tion, respectively. Because the Z-trifluoromethyl ketones all
exhibit similar potencies, the excellent photoswitching of 1
and 3 is due to particularly weak binding of the E isomers
of these compounds, relative to other E-trifluoromethyl ke-
tones.
The trifluoromethyl ketones 1 and 3 exhibited the best
enzyme photoswitching results. However, surface attach-
ment of alkyne 3 would yield a triazole of identical local
structure to 5 in the region expected to bind to the enzyme.
Triazole 5 is a poor photoswitch, so it is likely that 3 would
give poor photoswitching results after surface attachment,
unless the poor photoswitching of 5 is due to an interaction
with the terminal OH group. Consequently, the trifluoro-
methyl ketone 1 was chosen for use in surface enzyme pho-
toswitching. The alkyne 1 has the additional advantage of
containing an extended OEG group, which should facilitate
enzyme binding by spacing the inhibitor from the surface
(see discussion earlier).
of Z
isomer
isomer
isomer
isomer
before ir- after UV after visi- after visi-
radiation
[%]
irradiation ble irradi- ble irradi-
[%]
ation [%] ation
1
trifluoromethyl
ketone
trifluoromethyl
ketone
trifluoromethyl
ketone
trifluoromethyl <5
ketone
trifluoromethyl
ketone
6
5
5
85
8
11-fold
2
66
87
93
76
9
7.3-fold
7.3-fold
5.8-fold
4.2-fold
3
12
16
18
4[16]
5
7
6
7
8
a-keto ester
a-keto ester
a-keto ester
6
<5
5
88
85
94
14
7
14
6.3-fold
12-fold
6.7-fold
Table 2. Inhibition constants (IC50) for assay against a-chymotrypsin.
Compound Inhibitor type IC50 IC50 after IC50 after Decrease
before ir- UV irradi- visible ir- in activity
radiation
[mm]
ation [mm] radiation
[mm]
after visi-
ble irradi-
ation
1
trifluoromethyl >51
ketone
14
17
16
13
12
>51
21
>3.6
The trifluoromethyl ketone 1 was attached to commercial-
ly available surface plasmon resonance (SPR) surfaces (de-
tailed in ref. [7]). The binding of a-chymotrypsin to the sur-
face with attached 1 was then monitored by SPR (also brief-
ly discussed in Ref. [7]). Solutions of enzyme (0, 2.0, 6.0, and
18 mm) were separately injected over the surface for periods
of 300 s to monitor enzyme binding to the surface, and after
each experiment, buffer was run over the surface for 300 s
to monitor detachment of enzyme (Figure 4, enzyme injec-
tion at t=70 s).
2
trifluoromethyl
ketone
20
1.2
3
trifluoromethyl >85
ketone
>85
29
>5.3
2.2
4[16]
5
trifluoromethyl
ketone
trifluoromethyl
ketone
32
19
21
1.8
6
7
8
a-keto ester
a-keto ester
a-keto ester
2.0
1.0
0.71
0.73
0.47
0.23
2.2
0.85
0.59
3.0
1.8
2.6
6988
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 6983 – 6992

Photoswitch Ligands
FULL PAPER
rate of association of a-chymotrypsin with surface-bound 1
is decreased at high enzyme concentrations. This trend of a
decrease in the apparent association constant with increas-
ing enzyme concentration has also been observed in solu-
tion-phase studies of the binding of trifluoromethyl ketones
to a-chymotrypsin.[34] The decrease in kA was attributed to
an equilibrium between the active ketone inhibitor and an
inactive hydrated form, as shown in the equation:
E
ƒƒ!
ƒƒ
!
Ih
Ik
E-I
ð1Þ
H2O
where Ih represents hydrated inhibitor, Ik represents ketone
inhibitor, E represents a-chymotrypsin, and E–I represents
the enzyme–inhibitor complex. In aqueous solutions, the
hydrated form Ih is known to be present in an approximately
4500-fold excess.[34] At a high concentration of enzyme, the
concentration of ketone will be significantly decreased by
reaction with the enzyme and it has previously been pro-
posed that a relatively slow rate of hydrate to ketone de-
hydration should lead to a decrease in the actual concen-
tration of ketone, dependent on the enzyme concentration.
This would lead to a decrease in the apparent kA value as
the enzyme concentration increases, and thus an increase
in the apparent KD value. We propose that the surface-
attached trifluoromethyl ketone in this study is exhibiting
the same behavior. Alternative explanations for slow bind-
ing kinetics, such as those proposed for a-keto acid inhibi-
tors of a hepatitis C protease,[35] do not explain the observed
decrease in association rate at high enzyme concentra-
tions.
Figure 4. SPR sensorgrams for binding of a-chymotrypsin (0, 2.0, 6.0,
18 mm) to surfaces modified with 1. SPR data simultaneously recorded for
a control surface were subtracted from these plots.
Remaining enzyme was washed from the surface by injec-
tions of guanidine (6m, 5 mL), followed by acetic acid (1m,
10 mL), to regenerate the surface rapidly after each measure-
ment. For each nonzero enzyme concentration, the SPR re-
sponse increased over the period during which the enzyme
was injected relative to the surface control lacking attached
1. Furthermore, the response was enhanced for each in-
crease in enzyme concentration (0!18 mm in Figure 4),
which is consistent with attachment of enzyme to the sur-
face. This represents binding of enzyme to approximately
10% of the theoretical maximum number of surface-at-
tached inhibitor molecules,[7] at the highest enzyme concen-
tration.
Calculation of the KD affinity constant for enzyme binding
to the surface-bound inhibitor from this data was attempted
by use of a curve-fitting model (with the aid of Biacore
Biaevaluation software). However, no good fit could be ob-
tained by simultaneous modeling of all three concentration
curves. Individual modeling of the association and dissocia-
tion regions for each curve gave calculated kA, kD, and KD
values as shown in Table 3.
Next, the enzyme-free surface was removed from the SPR
instrument and irradiated with UV light (carried out as
above for solution-phase irradiations) to photoisomerize sur-
face attached (E)-1 to (Z)-1. The surface was returned to
the instrument, and injection of enzyme (2.0, 6.0, 18 mm) was
repeated to assess the extent of enzyme binding after this
photoisomerization (see Figure 5, curve 2 for 2.0 mm injec-
tion, Table 4 for maximum SPR responses, and Table 5 for
apparent binding constants).
A significantly higher response (1200 RU for the 2.0 mm
injection, Figure 5, curve 2) was observed than before irradi-
ation (690 RU for the 2.0 mm injection, Figure 5, curve 1).
Therefore, significantly more enzyme binds to the surface
after UV irradiation. The apparent kA and kD constants
after UV irradiation are greater than those observed before
irradiation (compare Table 3 and Table 5), representing
more rapid attachment and detachment of a-chymotrypsin
from the surface. This is as would be expected for a larger
proportion of the more active Z isomer at the surface. How-
ever, the increase in enzyme binding is only approximately
twofold, much less than the 11-fold change in the proportion
of Z isomer observed in solution (see Table 1). This is at
least partially attributable to enzyme binding by the weaker
E isomer that is expected to be predominantly present at
the surface before UV irradiation, and possibly also to de-
creased E/Z photoswitching at the surface. These factors
could not be assessed in detail for the surface-bound inhibi-
Table 3. Apparent kA, kD, and KD constants for binding of a-chymotryp-
sin to a surface containing attached inhibitor 1, calculated by individual
curve fitting to SPR data.
Enzyme concentration
[mm]
kA
kD [sꢀ1
]
KD [mm]
[mꢀ1 sꢀ1
]
2.0
6.0
18
3260ꢁ70 3.24ꢁ10ꢀ4 ꢁ4ꢁ10ꢀ6 0.099ꢁ0.003
1400ꢁ20 3.25ꢁ10ꢀ4 ꢁ2ꢁ10ꢀ6 0.233ꢁ0.005
590ꢁ4
3.19ꢁ10ꢀ4 ꢁ2ꢁ10ꢀ6 0.541ꢁ0.007
The apparent affinity constant (KD) increases with in-
creasing enzyme concentration, corresponding to apparently
weaker binding at higher enzyme concentrations. This is sur-
prising, because a constant KD value would be expected.
The apparent kinetic dissociation rate constant (kD) is ap-
proximately constant at all enzyme concentrations, but the
apparent kinetic association rate constant (kA) decreases
with increasing enzyme concentration. This shows that the
Chem. Eur. J. 2010, 16, 6983 – 6992
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6989

A. D. Abell and D. Pearson
These results show that the surface can be reversibly pho-
toswitched between two states, one of which binds a signifi-
cantly larger amount of enzyme. Photochemical switching
can therefore modulate binding of a-chymotrypsin to a sur-
face.
Conclusion
A new series of azobenzene-containing electrophilic ketones
has been synthesized and assayed against a-chymotrypsin as
both their E- and Z-enriched photostationary states. All
compounds (1–8) inhibited the enzyme with activity that
could be modulated by irradiation. The a-keto esters exhib-
ited 1.8- to 3.0-fold changes in activity on photoisomeriza-
tion, whereas the trifluoromethyl ketones exhibited 1.2- to
>5.3-fold changes in activity. The best enzyme photoswitch-
ing was exhibited by trifluoromethyl ketones with alkyne
groups as 4’-substituents on their azobenzene groups, due to
the relatively weak inhibition exhibited by the E isomers of
these compounds. These results show that the photoswitch-
ing of azobenzene-based photoswitch inhibitors can poten-
tially be optimized through the use of a relatively bulky 4’-
azobenzene substituent (e.g., alkyne) that interferes with in-
hibitor binding by the E (or Z) form of the photoswitch.
The best candidate, the trifluoromethyl ketone 1, was at-
tached to an SPR chip in a two-step procedure involving an
amide coupling followed by an azide–alkyne cycloaddition.
Enzyme binding to the modified surface was detected, but
no KD affinity constant could be calculated, due to “slow-
tight” binding of enzyme to the surface-bound inhibitor. Ir-
radiation of the surface with UV/Vis light modulated the
amount of enzyme that attached to the surface. As in the
cases of previously reported solution studies, a decrease in
the apparent association constant is observed for the surface
binding of trifluoromethyl ketones to a-chymotrypsin. This
is attributable to an equilibrium between the active ketone
inhibitor and an inactive hydrated form.
Figure 5. SPR sensorgrams for photoswitching of a-chymotrypsin
(2.0 mm) binding to a surface modified with 1. SPR data simultaneously
recorded for a control surface were subtracted from these plots. 1) sur-
face before irradiation, 2) surface after UV irradiation, and 3) surface
after UV and then visible irradiation.
Table 4. Maximum SPR response (RU) for binding of a-chymotrypsin
binding to surface attached inhibitor 1, before and after UV/Vis irradia-
tions.
Enzyme con- Maximum SPR
Maximum SPR
Maximum SPR re-
sponse after visi-
ble irradiation
(RU)
centration
response before response after
irradiation (RU) UV irradiation
(RU)
[mm]
2.0
6.0
18
690
1100
1600
1200
1700
2100
720
1200
1700
Table 5. Apparent kA, kD, and KD constants for binding of a-chymotryp-
sin to a surface containing attached inhibitor 1 after UV irradiation, cal-
culated by individual curve fitting to SPR data.
Enzyme concentration
[mm]
kA
kD [sꢀ1
]
KD [mm]
[mꢀ1 sꢀ1
]
2.0
6.0
18
4220ꢁ70 5.02ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.119ꢁ0.002
1890ꢁ20 4.82ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.255ꢁ0.003
748ꢁ6
4.43ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.593ꢁ0.006
tor. After removal of attached enzyme as above, the surface
was irradiated with visible light to “switch back” to the less
active E-enriched state [(E)-1], and injection of enzyme was
repeated. The response (720 RU for the 2.0 mm injection,
Figure 5, curve 3, Table 6) was close to the initial response
before UV irradiation, corresponding to a similar amount of
enzyme binding.
Experimental Section
General: NMR spectra were obtained with a Varian INOVA spectrome-
1
ter, operating at 500 MHz for H NMR and at 126 MHz for 13C NMR, or
with
a Varian UNITY 300 spectrometer, operating at 300 MHz for
1H NMR and at 75 MHz for 13C NMR. Two-dimensional NMR experi-
ments including COSY and HSQC were used to assign spectra, and were
obtained with the Varian INOVA spectrometer operating at 500 MHz.
Electrospray ionization mass spectra were detected with a Micromass
LCT TOF mass spectrometer operating in electrospray mode with aceto-
nitrile/H2O (50%) as solvent. Dry THF was distilled from sodium or po-
tassium, and dry DCM was distilled from CaH2. EtOAc, NMM, AcOH,
and EtOH were distilled before use.
Table 6. Apparent kA, kD, and KD constants for binding of a-chymotryp-
sin to a surface containing attached inhibitor 1 after visible irradiation,
calculated by individual curve fitting to SPR data.
Photoisomerization: Compounds were photoisomerized by irradiation
with UV or visible light from a Mercury arc lamp (500 W). The light
beam was filtered through water to reduce heat, and through either an
Edmund optics 340 nm interference filter for UV, or a Corning 0–51
visible filter for visible light. Samples were dissolved in CD3CN
(ꢂ1.5 mgmLꢀ1), and irradiated for 70 min with UV light or for 30 min
with visible light. The ratios of E/Z isomers before and after irradiation
Enzyme concentration
[mm]
kA
kD [sꢀ1
]
KD [mm]
[mꢀ1 sꢀ1
]
2.0
6.0
18
3060ꢁ50 3.31ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.108ꢁ0.002
1330ꢁ10 3.22ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.242ꢁ0.003
533ꢁ3
2.98ꢁ10ꢀ4 ꢁ1ꢁ10ꢀ6 0.559ꢁ0.004
6990
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 6983 – 6992

Photoswitch Ligands
FULL PAPER
1
were determined by H NMR analysis by use of the integrals of well-sep-
Enzyme binding to this modified surface was monitored by running a
sensorgram in Tris buffer at 30 mLminꢀ1. A range of a-chymotrypsin solu-
tion concentrations were injected over the surface and monitored as fol-
lows:
arated peaks. The major isomer was assigned the thermodynamically
more stable E configuration on the basis of literature precedence.[37,38]
The minor Z isomer characteristically gave rise to upfield signals for the
aryl protons, which were enhanced on irradiation. 2D NMR was used to
confirm the presence of minor isomers in some representative cases. All
experiments involving photoisomerization were carried out in dim light-
ing conditions (that is, glassware wrapped in foil and with the lights
turned off).
An a-chymotrypsin solution (0, 2.0, 6.0, or 18 mm in Tris buffer, 150 mL)
was injected over a period of 5 min, flow rate 30 mLminꢀ1. After the in-
jection was complete, buffer was run for 5 min to monitor dissociation of
enzyme from the surface, and then the surface was regenerated. Regener-
ation was carried out at 5 mLminꢀ1 in Tris buffer by the injection of gua-
nidine hydrochloride (6m, 5 mL) followed by acetic acid (1m, 10 mL), and
buffer was then run for 5 min.
Enzyme assays: Buffer solution (Tris): tris(hydroxymethyl)aminomethane
(1.21 g), CaCl2·6H2O (0.44 g) and Triton X-100 (0.05 g) were dissolved in
Milli-Q deionized water (75 mL), adjusted to pH 7.8 with HCl solution
(1m), and made up to 100 mL with Milli-Q water.
For photoswitching experiments, the above enzyme binding assay was
carried out, and the chip was then regenerated, removed from the instru-
ment, rinsed with water, and irradiated with UV light (as detailed above
for solution-phase photoisomerization) for 10 min. The chip was then re-
placed in the SPR instrument, and the enzyme binding assay was repeat-
ed. The photoisomerization process was repeated with visible light irradi-
ation and this assay was repeated.
Substrate solution: N-Succinyl-(Ala)2-Pro-Phe-4-nitroanilide (21 mg) was
dissolved in Tris buffer solution (10 mL) with the aid of ultrasonication.
The solution was stored at ꢀ188C for up to two weeks. The concentration
of the solution was determined at the start of each day from its UV spec-
trum (e315 =14000 Lmolꢀ1 cmꢀ1).
Enzyme solution: A stock solution was prepared from a-chymotrypsin
(15 mg) in HCl solution (10 mL, pH 3, made up by dilution of conc. HCl
with Milli-Q water). The stock solution was stored at ꢀ188C for up to 1
month. Each day an enzyme solution was prepared. Stock solution
(200 mL) and Triton X-100 (25 mg) were made up to 50 mL with Milli-Q
water.
Details of synthesis and NMR spectra are given in the Supporting Infor-
mation.
Acknowledgements
Inhibition of a-chymotrypsin was determined with Suc-Ala-Ala-Pro-Phe-
4-nitroanilide as the substrate by an assay procedure developed from the
technique described by Geiger,[39] except that the order of addition of
enzyme and substrate was inverted, and only one substrate concentration
was used, in order to obtain inhibition constants as IC50 values rather
than Ki values. Briefly, inhibitors were dissolved in CD3CN at a series of
dilutions ranging from 0.5–500 mm as appropriate for each compound. For
each rate measurement, inhibitor solution (or acetonitrile blank, 50 mL),
substrate solution (60 mL), and buffer solution (910 mL) were mixed in a
cuvette and incubated for 5 min at 258C. Enzyme solution (30 mL) was
added, and the absorbance at 405 nm was monitored for 5 min for a-keto
esters or 10 min for trifluoromethyl ketones. Absorbance vs. time plots
were obtained, and the slopes were used to find the initial rates for a-
keto esters or the final rates for trifluoromethyl ketones. From the differ-
ences between these rates and the rates for the acetonitrile blanks, the
percent inhibitions were calculated. Each experiment was repeated over
as many inhibitor concentrations as required to obtain a good straight-
line plot of percent inhibition vs. log (inhibitor concentration), from
which the IC50 value was interpolated.
The authors thank Andrew Muscroft-Taylor for synthesis of compound
26, Nathan Alexander for synthesis of compound 10, and Fiona Clow for
help with SPR. Financial support from the Royal Society of NZ, the
Marsden Fund and the ARC (DP0771901) is also gratefully acknowl-
edged. D.P. was funded by a TEC Bright Future Top Achiever Doctoral
Scholarship.
[1] M. Bose, D. Groff, J. Xie, E. Brustad, P. G. Schultz, J. Am. Chem.
Soc. 2006, 128, 388.
[2] R. Numano, S. Szobota, A. Y. Lau, P. Gorostiza, M. Volgraf, B.
Roux, D. Trauner, E. Y. Isacoff, Proc. Natl. Acad. Sci. USA 2009,
106, 6814.
[3] P. M. Mendes, Chem. Soc. Rev. 2008, 37, 2512.
[4] J. Auernheimer, C. Dahmen, U. Hersel, A. Bausch, H. Kessler, J.
Am. Chem. Soc. 2005, 127, 16107.
[5] R. Blonder, S. Levi, G. Tao, I. Ben-Dov, I. Willner, J. Am. Chem.
Soc. 1997, 119, 10467.
[6] G. Hayashi, M. Hagihara, C. Dohno, K. Nakatani, J. Am. Chem.
Soc. 2007, 129, 8678.
[7] D. Pearson, A. J. Downard, A. Muscroft-Taylor, A. D. Abell, J. Am.
Chem. Soc. 2007, 129, 14862.
Surface plasmon resonance experiments: SPR was performed with a Bia-
core 2000 instrument. All solutions were filtered through a 0.22 mmfilter
and centrifuged (1300 rpm, 10 s) before use.
Buffers used: HEPES [0.01m, pH 7.4 containing NaCl (0.15m) and
EDTA (3 mm)], Tris [0.1m, pH 7.8, containing CaCl2 (0.02m) and P20 sur-
factant (12.5 mL)].
[8] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. 2001, 113,
2056; Angew. Chem. Int. Ed. 2001, 40, 2004.
[9] The use of amine chemistry for surface attachment was deemed in-
appropriate since it is incompatible with the constituent trifluoro-
methyl ketone group.
[10] a) L. Borissenko, M. Groll, Chem. Rev. 2007, 107, 687; b) D. C.
Martyn, A. J. Vernall, B. M. Clark, A. D. Abell, Org. Biomol. Chem.
2003, 1, 2103; c) A. D. Abell, B. K. Nabbs, Bioorg. Med. Chem.
2001, 9, 621; d) A. D. Abell, M. D. Oldham, J. Org. Chem. 1997, 62,
15093; e) A. D. Abell, J. M. Taylor, J. Org. Chem. 1993, 58, 14.
[11] D. Pearson, A. D. Abell, Org. Biomol. Chem. 2006, 4, 3618.
[12] F. O. Koller, R. Reho, T. E. Schrader, L. Moroder, J. Wachtveitl, W.
Zinth, J. Phys. Chem. B 2007, 111, 10481.
[13] L. Ulysse, J. Chmielewski, Bioorg. Med. Chem. Lett. 1994, 4, 2145.
[14] L. Ulysse, J. Cubillos, J. Chmielewski, J. Am. Chem. Soc. 1995, 117,
8466.
[15] L. G. Ulysse, J. Chmielewski, Chem. Biol. Drug Des. 2006, 67, 127.
[16] D. Pearson, N. Alexander, A. Abell, Chem. Eur. J. 2008, 14, 7358.
[17] L. Hedstrom, Chem. Rev. 2002, 102, 4501.
[18] S. Ameerunisha, P. S. Zacharias, J. Chem. Soc. Perkin Trans. 2 1995,
1679.
[19] M. L. Tomlinson, J. Chem. Soc. 1946, 756.
Compound 1 was immobilized to an SPR chip as follows. A Biacore CM5
chip was docked in the SPR instrument and prepared by priming and
running a sensorgram in HEPES buffer at 100 mLminꢀ1 flow rate, with
injection of NaOH (50 mm, 20 mL ꢁ 2), HCl (10 mm, 20 mL ꢁ 2), and SDS
(0.1%, 20 mL ꢁ 2). Surface modification was then carried out at
5 mLminꢀ1 flow rate by injection of EDCI/N-hydroxysuccinimide
(35 mL), followed by 11-azido-3,6,9-trioxaundecan-1-amine (32,
11 mgmLꢀ1 in HEPES buffer, 3 ꢁ 5 mL) and then ethanolamine (1m,
pH 8.5, 35 mL). During attachment of 32, the SPR resonance increased
by 147 RU. The chip was then undocked and removed from the instru-
ment and rinsed with water. A solution of 1 (0.3 mg, 0.44 mmol), CuBr
(0.063 mg, 0.44 mmol), 2,6-lutidine (0.094 mg, 0.88 mmol), 2,2’-bipyridine
(0.14 mg, 0.88 mmol), and sodium ascorbate (0.17 mg, 0.88 mmol) in DMF/
H2O (1:1, 150 mL) was pipetted onto the surface and left for 1 h at room
temperature. The surface was then rinsed with DMF/H2O (1:1), H2O,
EDTA (0.1m), DMF/H2O (1:1), and H2O and was then redocked in the
SPR instrument. A sensorgram was run in Tris buffer for several hours in
order to obtain a stable baseline. A reference cell was prepared by the
same method except with omission of the injection of 32.
Chem. Eur. J. 2010, 16, 6983 – 6992
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6991

A. D. Abell and D. Pearson
[20] K. Fujiwara, Y. Murata, T. S. M. Wan, K. Komatsu, Tetrahedron
1998, 54, 2049.
[32] O. Sadovski, A. A. Beharry, F. Zhang, G. A. Woolley, Angew. Chem.
2009, 121, 1512; Angew. Chem. Int. Ed. 2009, 48, 1484.
[33] J. Zeitouny, C. Aurisicchio, D. Bonifazi, R. D. Zorzi, S. Geremia, M.
Bonini, C.-A. Palma, P. Samori, A. Listorti, A. Belbakrab, N. Ar-
maroli, J. Mater. Chem. 2009, 19, 4715.
[34] K. Brady, R. H. Abeles, Biochemistry 1990, 29, 7608.
[35] F. Narjes, M. Brunetti, S. Colarusso, B. Gerlach, U. Koch, G. Biasiol,
D. Fattori, R. De Francesco, V. G. Matassa, C. Steinkuhler, Biochem-
istry 2000, 39, 1849.
[36] B. Perdicakis, H. J. Montgomery, J. G. Guillemette, E. Jervis, Anal.
Biochem. 2005, 337, 211.
[37] a) A. J. Harvey, A. D. Abell, Tetrahedron 2000, 56, 9763; b) A. J.
Harvey, A. D. Abell, Bioorg. Med. Chem. Lett. 2001, 11, 2441.
[38] P. R. Westmark, J. P. Kelly, B. D. Smith, J. Am. Chem. Soc. 1993,
115, 3416.
[21] N. P. Peet, J. P. Burkhart, M. R. Angelastro, E. L. Giroux, S. Mehdi,
P. Bey, M. Kolb, B. Neises, D. Schirlin, J. Med. Chem. 1990, 33, 394.
[22] R. J. Linderman, D. M. Graves, J. Org. Chem. 1989, 54, 661.
[23] B. Priewisch, K. Rꢂck-Braun, J. Org. Chem. 2005, 70, 2350.
[24] W. Yuan, B. Munoz, C.-H. Wong, J. Z. Haeggstriim, A. Wetterholm,
B. Samuelsson, J. Med. Chem. 1993, 36, 211.
[25] A. R. Katritzky, Q. Y. Chen, S. R. Tala, Chem. Biol. Drug Des. 2009,
73, 611.
[26] V. V. Rostovtsev, L. G. Green, V. V. Fokin, K. B. Sharpless, Angew.
Chem. 2002, 114, 2708; Angew. Chem. Int. Ed. 2002, 41, 2596.
[27] M. M. McPhee, S. M. Kerwin, Bioorg. Med. Chem. 2001, 9, 2809.
[28] Y. Nishihara, K. Ikegashira, K. Hirabayashi, J.-i. Ando, A. Mori, T.
Hiyama, J. Org. Chem. 2000, 65, 1780.
[29] G. H. Coleman, C. M. McCloskey, J. Am. Chem. Soc. 1943, 65, 1588.
[30] J.-H. Li, Y. Liang, Y.-X. Xie, J. Org. Chem. 2005, 70, 4393.
[31] H. Rau in Photochromism: Molecules and Systems (Eds.: H. Dꢂrr,
H. Bouas-Laurent), Elsevier, Amsterdam, 1990, p. 165.
[39] R. Geiger, Methods of Enzymatic Analysis, Vol. 5, 3rd ed., VCH,
Weinheim, 1984.
Received: December 9, 2009
Published online: May 7, 2010
6992
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 6983 – 6992