Bornyl (3,4,5-trihydroxy)-cinnamate - An optimized human neutrophil elastase inhibitor designed by free energy calculations

Article abstract of DOI:10.1016/j.bmc.2007.11.070

Human neutrophil elastase (HNE), a serine protease, is involved in the regulation of inflammatory processes and controlled by endogenous proteinase inhibitors. Abnormally high levels of HNE can cause degradation of healthy tissues contributing to inflammatory diseases such as rheumatoid arthritis, and also psoriasis and delayed wound healing. In continuation of our research on HNE inhibitors we have used the recently developed binding mode model for a group of cinnamic acid derivative elastase inhibitors and created bornyl (3,4,5-trihydroxy)-cinnamate. This ligand exhibited improved binding affinity predicted by means of free energy calculations. An organic synthesis scheme for the ligand was developed and its inhibitory activity was tested toward the isolated enzyme. Its IC₅₀ value was found to be three times lower than that of similar compounds, which is in line with the computational result showing the high potential of free energy calculations as a tool in drug development.

Full text of DOI:10.1016/j.bmc.2007.11.070

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry 16 (2008) 2385–2390
Bornyl (3,4,5-trihydroxy)-cinnamate - An optimized
human neutrophil elastase inhibitor designed
by free energy calculations
Thomas Steinbrecher,a,ꢀ,ꢁ Andrea Hrenn,b,ꢁ Korinna L. Dormann,c
Irmgard Merfortb,* and Andreas Labahna,*
aInstitut fu¨r Physikalische Chemie, Albert-Ludwigs-Universita¨t Freiburg, Albertstr. 23a, 79104 Freiburg, Germany
bInstitut fu¨r Pharmazeutische Wissenschaften, Lehrstuhl fu¨r Pharmazeutische Biologie und Biotechnologie,
Albert-Ludwigs-Universita¨t Freiburg, Stefan-Meier-Str. 19, 79104 Freiburg, Germany
cInstitut fu¨r Organische Chemie und Biochemie, Albert-Ludwigs-Universita¨t Freiburg, Albertstr. 21, 79104 Freiburg, Germany
Received 7 June 2007; revised 20 November 2007; accepted 23 November 2007
Available online 19 December 2007
Abstract—Human neutrophil elastase (HNE), a serine protease, is involved in the regulation of inflammatory processes and con-
trolled by endogenous proteinase inhibitors. Abnormally high levels of HNE can cause degradation of healthy tissues contributing
to inflammatory diseases such as rheumatoid arthritis, and also psoriasis and delayed wound healing. In continuation of our
research on HNE inhibitors we have used the recently developed binding mode model for a group of cinnamic acid derivative elas-
tase inhibitors and created bornyl (3,4,5-trihydroxy)-cinnamate. This ligand exhibited improved binding affinity predicted by means
of free energy calculations. An organic synthesis scheme for the ligand was developed and its inhibitory activity was tested toward
the isolated enzyme. Its IC50 value was found to be three times lower than that of similar compounds, which is in line with the com-
putational result showing the high potential of free energy calculations as a tool in drug development.
ꢀ 2007 Elsevier Ltd. All rights reserved.
1. Introduction
rels, and is stabilized by four disulfide bridges. Nineteen
arginine residues on the protein surface and only nine
acidic residues make it a highly basic protein. The activ-
ity of HNE is tightly regulated in vivo by several phys-
iological inhibitors. However, in inflamed tissue an
impaired balance between the enzyme and its natural
inhibitors exists, leading to the degradation of healthy
tissue.1–3 Interestingly, HNE also has an important reg-
ulatory role in the local inflammatory response by pro-
teolytic modification of chemokines and cytokines and
has been implicated in wound healing.4 Active HNE is
detected in psoriatic lesions and induces keratinocyte
hyperproliferation via the EGFR signaling pathway.5,6
Thus, the involvement of HNE in such pathological pro-
cesses makes it an interesting target for the development
of anti-inflammatory drugs. Considering local inflam-
mations, small molecules might be more advantageous
compared to peptidic inhibitors because of their better
penetration rate into the skin.
Human neutrophil elastase is a serine protease playing
an important role in physiological functions of the im-
mune system. It is produced and stored in the neutrophil
granulocytes and has the ability to hydrolyze a wide
variety of protein substrates, notably elastin, a protein
that gives skin its elastic properties. HNE’s specificity
lies in the cleavage after small aliphatic amino acid res-
idues, especially valine. The functional enzyme contains
218 amino acids, organized into two domains of b-bar-
Abbreviations: MD, molecular dynamics; HNE, human neutrophil
elastase; TI, thermodynamic integration; vdW, van-der-Waals.
Keywords: Inflammation; Human neutrophilelastase; Free energy calcu-
lation; Drug design.
*
Corresponding authors. Tel.: +49 761 203 8373; fax: +49 761 203
8383 (I.M.); tel.: +49 761 203 6188; fax: +49 761 203 6189 (A.L.);
e-mail irmgard.merfort@pharmazie.uni-freiburg.de;
andreas.labahn@physchem.uni-freiburg.de
addresses:
ꢀ
ꢁ
Recently, various natural compounds have been studied
as inhibitors of HNE7–9; among them cinnamic acid es-
ter derivatives were found to have IC50 values in the lM
Present address: Department of Molecular Biology, The Scripps
Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA.
These authors contributed equally to this work.
0968-0896/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2007.11.070

2386
T. Steinbrecher et al. / Bioorg. Med. Chem. 16 (2008) 2385–2390
range. No X-ray crystal structure of HNE cocrystallized
with an esterified cinnamic acid derivative has been re-
ported up to now, but a putative binding mode for bor-
nyl caffeate and similar compounds has been proposed
and evaluated in our recent work.10 Based on this com-
plex structure, the binding modes of modified com-
pounds can be simulated and their affinities for HNE
can be predicted.
lated by performing MD simulations of the
equilibrated systems at seven k values chosen after a
Gaussian numerical integration scheme (Eq. 211). Each
‘k-window’ consisted of a 50 ps equilibration phase
and a 100 ps data collection phase, performed under
the NPT ensemble using a 2 fs time step. The simula-
tions yielded a DA0
of 47.1 kcal/mol for the un-
bound (48.1 kcal/mol from the partial charge and
Unbound
ꢀ1.0 kcal/mol from the atom type and vdW transforma-
In this work, we report a binding affinity prediction, or-
ganic synthesis, and activity measurement for bornyl
(3,4,5-trihydroxy)-cinnamate. Our combination of com-
puter simulations with organic synthesis and pharmaco-
logical activity measurements shows that the advances in
computer hardware and molecular simulation software
in the last years have made free energy calculations a ro-
bust and efficient tool in biochemical and pharmaceuti-
cal studies. While still too time-consuming for routine
employment as screening tools, they are very well suited
for structural improvements of known ligands and accu-
rate binding free energy predictions.
tion) and a DA0
of 50.8 kcal/mol for the bound
Bound
states (51.0 and ꢀ0.2 kcal/mol from partial charge and
atom type transformations, respectively). This results
of +3.7 kcal/mol, meaning that bornyl
in a DDA0
Bind
(3,4,5-trihydroxy)-cinnamate is predicted to inhibit
HNE with a binding constant about 500 times lower
than that of bornyl caffeate, placing it in the low nano-
molar inhibitor range.
From the two separate steps of the transformation
the contributions of different interaction types to the
relative binding strengths can be seen. The change of
electrostatic interactions (step 1) resulted in a DDA0
Bound
of 2.9 kcal/mol and the change of vdW types (step 2)
of 0.8 kcal/mol. From this it
Bound
2. Results
resulted in a DDA0
can be inferred that the new ligand’s improved bind-
ing strength is mostly, but not completely, owed
to stronger electrostatic interactions with the recep-
tor. The promising prediction of bornyl (3,4,5-trihy-
droxy)-cinnamate’s binding properties prompted the
organic synthesis and pharmacological testing of the
compound.
2.1. Binding affinity prediction
Predictions of binding free energies were performed on a
number of various derivatives of bornyl caffeate selected
after visual inspection of the proposed protein-ligand
complex. Among these, only bornyl (3,4,5-trihydroxy)-
cinnamate resulted in a predicted higher binding affinity
compared to bornyl caffeate. Therefore, only results for
this compound will be reported here. To judge the bind-
ing strength of the proposed ligand to HNE, a TI calcu-
lation was set up that compares the binding free energy
of bornyl (3,4,5-trihydroxy)-cinnamate with that of bor-
nyl caffeate.
2.2. Organic synthesis
The synthesis of bornyl (3,4,5-trihydroxy)-cinnamate
was performed as outlined in Scheme 1. The cinnamic
acid 2 was prepared by O-allylation of trihydroxybenz-
aldehyde (1) and subsequent Knoevenagel condensation
of the O-allyl-protected aldehyde with malonic acid.
Dicyclohexylcarbodiimide/4-dimethylaminopyridine-in-
duced condensation of the cinnamic acid 2 with (R)-(+)-
borneol afforded the allyl-protected bornylic ester which
was deallylated by means of morpholine/Pd(PPh3)4 to
yield the trihydroxyester 3.
Since the free energy difference between two thermody-
namic states is only a physically meaningful quantity if
both states have the same number and types of atoms,
a simple thermodynamic cycle was used to predict bind-
ing free energies: in this cycle the two states are equal to
the two ligands that are compared (k = 0 was chosen to
represent bornyl (3,4,5-trihydroxy)-cinnamate and k = 1
equals bornyl caffeate) and two systems are set up repre-
senting those ligands bound to the receptor or simply
solvated in water. Simulating the transition between
both states, a free energy change, as calculated by Eq.
1, is determined for both the solvated ligands and the li-
gand-protein complexes.
A colorless crystalline solid in about 10% overall yield
was obtained. The final product had a purity of 98%
as proved by NMR to be directly used in the HNE activ-
ity assay. The product was found to be very susceptible
to oxidation when exposed to air, so handling under in-
ert gas conditions was necessary (Fig. 1).
2.3. Elastase inhibition assay
The ligand transformations were simulated in two steps,
first all atomic partial charges were transformed from
their initial (k = 0) to their final (k = 1) values and in a
subsequent second step the MD atom types and vdW
parameters were transformed. This was done for reasons
of simulation stability, because changing partial charges
and vdW radii simultaneously can result in vanishing
atoms with non-zero charges getting very close to non-
bonded neighboring atoms, leading to very high electro-
static energies. Each of the transformations was simu-
Bornyl (3,4,5-trihydroxy)-cinnamate and bornyl caffeate
were studied for their inhibitory effect on HNE using the
isolated enzyme (see Table 1).12,13 The recently reported
IC50 value of 1.6 lM of the latter one could be con-
firmed,12 whereas an IC50 value of 0.54 lM was ob-
tained for the newly synthesized compound. As a1-
antitrypsin from human plasma is the physiological
inhibitor of the elastase14 and widely used15,16 this com-
pound should be very specific. We obtained an IC50 va-

T. Steinbrecher et al. / Bioorg. Med. Chem. 16 (2008) 2385–2390
2387
Scheme 1.
IC50 of 540 nM making bornyl (3,4,5-trihydroxy)-cinna-
mate the strongest binding cinnamic acid derivative
known up to date. Considering that a binding constant
was predicted about 500 times lower than that of bornyl
caffeate, an IC50 value of about 3.1 nM should be ob-
tained. This is not the case. The measured IC50 value
is higher, which means that the calculated DDA0 overes-
timated the binding strength by ca. 3.1 kcal/mol. A con-
formational bias resulting from an induced fit of the
binding site to the starting ligand structure can be ex-
cluded as a cause of this error, because this could only
have led to an underestimation of the new ligand’s
strength. Bornyl caffeate lacks the additional hydroxyl
group of bornyl (3,4,5-trihydroxy)-cinnamate and could
thus exhibit a more flexible bound state. It is possible
that sampling of the conformational space of bound
bornyl caffeate was therefore less thorough than that
of bornyl (3,4,5-trihydroxy)-cinnamate, possibly leading
to the observed deviation. However, while very accurate
free energies have been reported in TI applications17 in
very long simulations for test cases, the authors believe
an error of 1–2 kcal/mol to be typical of biochemical
TI applications, caused by insufficient sampling and
inaccuracies of the used non-polarizable force fields. Ad-
vances both in computer power and quality of MD
models will certainly improve the precision of such cal-
culations in the future, but applications as the one de-
scribed here are already sufficiently predictive to
provide at least qualitative and sometimes quantitative10
predictions of ligand binding strengths superior to those
that could be obtained by using docking scoring func-
tions or other simple empirical models.
Figure 1. Comparison of the three compounds regarding their
inhibitory activity against HNE.
Table 1. IC50 values and 95% confidence intervals of the tested
compounds, calculated by GraphPad Prism 4
Compound
IC50 in lM
95% Confidence
interval
Bornyl (3,4,5-trihydroxy)-
cinnamate
0.54
0.33–0.89
Bornyl caffeate
a1-Antitrypsin
1.56
0.12
1.05–2.30
0.05–0.29
lue of 0.12 lM. Hernandez et al.15 reported that 90 lg/
mL of this physiological compound inhibited HNE to
84%.
The necessity to divide the TI transformation into two
separate steps of changing electrostatic and vdW inter-
actions led to stable simulations and easily integrateable
free energy curves. This however came with the cost of
doubling the necessary simulation time. A more efficient
mixing function for the two potential energies than Eq. 3
could be employed to improve sampling and simulation
3. Discussion
Free energy calculations predicted that the ligand would
inhibit HNE with an IC50 value in the nanomolar range.
This prediction turned out to be correct with a measured

2388
T. Steinbrecher et al. / Bioorg. Med. Chem. 16 (2008) 2385–2390
DDA0Bind ¼ DAB0 oundðA ! BÞ ꢀ DAU0 nboundðA ! BÞ
ꢁ DDG0Bind
is positive if A
Bind
convergence. In this respect, a ‘softcore’ potential simi-
lar to the one described in Refs. 17 and 18 seems like
one promising way to improve the efficiency of free en-
ergy calculations in the future.
ð4Þ
Note that in our definition DDA0
binds stronger to the receptor than B. Since all simula-
tions were performed under conditions of constant pres-
sure and temperature (the NPT ensemble), we interpret
the result as approximately equal to DDG0Bind, the bind-
ing free enthalpy difference of the ligands obtained by an
experiment.
Notably, bornyl (3,4,5-trihydroxy)-cinnamate lacks the
catechol group that was a feature of previously known
good binders to HNE like bornyl caffeate or fucinolic
acid. However, vicinal di-hydroxy functional groups are
present. Substances rich in hydroxyl groups have been
studied as promising candidates of HNE inhibition be-
fore7 and the structure of the newly synthesized ligand
gives rise to the assumption that more available hydroxyl
functionalities tend to improve the affinity of a compound
to HNE. Nevertheless, our results show that prediction of
binding constants is a valuable tool to improve the inhib-
itory activity of ligands. Finally, in vivo studies have to
demonstrate whether these small molecule inhibitors are
suitable for the treat-ment of inflammatory diseases, espe-
cially for local inflammations.
4.2. Molecular dynamics simulations
Molecular dynamics simulations were carried out using
the Amber8 molecular modeling suite.22 Two different
sets of force field parameters were used, namely ff0323
for the protein parts of the system and gaff24 for the li-
gands. The TIP3P water model25 was used for solvation.
Ligand partial charges were derived by the RESP meth-
odology,26 based on quantum mechanical calculations
on the HF/6-31G* level using Gaussian98.27 As the
starting geometry for the protein–ligand complex the re-
cently proposed binding model for cinnamic acid deriv-
ative bornyl esters was taken from,10 namely from the
equilibrated structure of the #2 placement group of bor-
nyl caffeate as determined by ligand docking and MM-
PBSA calculations. The bound bornyl caffeate was
changed to bornyl (3,4,5-trihydroxy)-cinnamate, the
4. Methods
4.1. Thermodynamic integration calculations
Free energies were computed by performing thermody-
namic integration calculations10,19,20 in which the free
energy difference between two states is computed by
coupling their potential functions V0 and V1 via a
parameter k and simulating a transition of k from zero
to one. Statistical mechanics allows for the calculation
of the free energy difference between the states, DA0:
˚
complex was solvated in a 12 A deep solvation layer
and neutralized using Clꢀ counterions.
The systems were equilibrated under periodic boundary
conditions according to the following procedure: after
1000 steps of preliminary minimization to remove close
vdW contacts, the system temperature was brought to
300 K during 50 ps of constant volume dynamics by a
Berendsen-type28 temperature coupling algorithm. In
doing so, all solute atoms had 10 kcal/mol2 harmonic
positional restraints applied. Finally, a volume equili-
bration was performed during 150 ps of constant pres-
sure dynamics. A time step of 2 fs in combination with
the SHAKE29 algorithm to constrain bond lengths
involving hydrogen atoms was used for all simulations.
All equilibration was performed using the starting state
hamiltonian (k = 0).
ꢀ
ꢁ
Z
1
oV ðkÞ
ok
DA0 ¼ A01 ꢀ A00 ¼
dk
ð1Þ
k¼0
k
where A01 and A00 refer to the free energies of the perturbed
and unperturbed states, respectively. The angular brack-
ets denote
a
statistically weighted sampling of
conformational space that can (at least approximately)
be accomplished by performing molecular dynamics sim-
ulations on the system at varying values of k. Then, the
integral can be solved numerically by summing up the
weighted average contributions of different ‘k-
windows’.11
ꢀ
ꢁ
n
X
oV
ok
DA0 ꢁ
wi
ð2Þ
4.3. Materials and test compounds
ki
i¼1
The mixed potential function V(k) was set to:
Bornyl caffeate was isolated from Verbesina turbacensis
Kunth.12 Purity was evaluated by TLC to be 97%. a1-
Antitrypsin from human plasma was bought from Sigma.
Enzyme substrate N-methoxysuccinyl-L-Ala-L-Ala-L-
Pro-L-Val-p-nitroanilide (MeO-Suc-Ala-Ala-Pro-Val-
pNA), trypsin inhibitor from soybean, and DMSO as well
as elastase from human leukocytes (EC 3.4.21.37) were
purchased from Sigma. TRIS (tris-(hydroxymethyl)-ami-
no methane) and HCl 25% were from Merck.
4
4
V ðkÞ ¼ ð1 ꢀ kÞ V 0 þ ½1 ꢀ ð1 ꢀ kÞ ꢂV 1
ð3Þ
All transformations were set up so that atoms disap-
peared when changing k from zero to one. In combina-
tion with the potential mixing function in Eq. 3, this
prevents the ‘origin singularity’ effect21 from occurring,
by making the divergence of the potential energy differ-
ence at large k values integrateable. For predictions of
binding free energy changes, the transformation of a li-
gand into another is simulated both in the receptor
bound and unbound state. The difference between these
transformation free energies is equal to the difference in
binding free energy of the compared ligands A and B:
Ten millimolar of stock solutions were prepared in
DMSO and then diluted with DMSO and 50 lM Tris–
HCl buffer, pH 7.5, to the final concentration. DMSO
was restricted to 1% for the HNE assay.

T. Steinbrecher et al. / Bioorg. Med. Chem. 16 (2008) 2385–2390
2389
The absorbance was measured using a Uvikon 933 UV–
vis double-beam spectrometer.
5:1) to give pure
5
(92 mg, 80%). 1H NMR
(300.1 MHz, CDCl3): d = 0.88 (s, 3H), 0.90 (s, 3H),
0.95 (s, 3H), 1.05 (dd, J = 13.8, J = 3.7, 1H), 1.24–1.49
(m, 2H), 1.71 (mc, 1H), 1.79 (mc, 1H), 2.05 (mc, 1H),
2.41 (mc, 1H), 4.59 (m, 6 H), 5.02 (d, J = 8.6, 1H),
5.19 (dq, J = 10.4, J = 1.6, 1H), 5.29 (dq, J = 10.5,
J = 1.4, 2H), 5.33 (dq, J = 15.1, J = 1.5, 1H), 5.43 (dq,
J = 17.3, J = 1.6, 2H), 6.01–6.12 (m, 3H), 6.31 (d,
J = 16.0, 1H), 6.76 (s, 2H), 7.53 (d, J = 16.0, 1H). IR
(film) m = 2980, 2935, 2870, 1710, 1445, 1385, 1275,
4.4. Synthesis of bornyl (3,4,5-trihydroxy)-cinnamate
The synthesis of bornyl (3,4,5-trihydroxy)-cinnamate
was performed by modifying the synthetic strategy for
the similar natural product (+)-rosmarinic acid worked
out by Eicher et al.30 (Scheme 1). Educts were obtained
in highest commercially available purity. The enantio-
meric excess of 3 can be assumed to be the unchanged
ee of the (R)-(+)-borneol used (ee > 98% after GC).
1150, 1115, 1075, 920, 735 cmꢀ1
.
4.4.4. Bornyl (3,4,5-trihydroxy)-cinnamate (3). Pd(PPh3)4
(11 mg, 9.5 lmol, 5 mol-%) and morpholine (45 lL,
45 mg, 0.51 mmol, 2.9 equiv) were added to a solution
of 5(80 mg, 0.18 mmol) in THF (4 mL). The reaction mix-
ture was stirred under Argon at room temperature for 6 h.
Et2O (10 mL) and HCl (1 M, 10 mL) were added. After
phase separation the obtained phase was extracted with
Et2O (2· 10 mL), washed with HCl (1 M, 2· 5 mL), dried
(MgSO4), and evaporated. Flash chromatography (cyclo-
hexane/acetone 2:1) gave pure 3 (39 mg, 66%). 1H NMR
(300.1 MHz, CDCl3): d = 0.88 (CH3-100, s), 0.90 (CH3-
90, s), 0.95 (CH3-80, s), 1.06 (H-30a, d, J = 14.9), 1.26 (H-
50a, m), 1.32 (H-60b, m), 1.71 (H-40, m), 1.77 (H-50b, m),
2.02 (H-60a, m), 2.41 (H-30b, m), 5.00 (H-20b, d, J = 8.1),
6.29 (H-8, d, J = 15.6), 6.74 (H-2 and 6, s), 7.50 (H-7, d,
J = 15.6); 13C NMR (100.6 MHz, CDCl3): d = 13.7 (C-
100), 19.0 (C-80), 19.8 (C-90), 27.3 (C-60), 28.1 (C-50), 36.9
(C-30), 45.0 (C-40), 48.0 (C-70), 49.1 (C-10), 80.8 (C-20),
108.4 (C-2 and 6), 116.2 (C-8), 126.6 (C-1), 134.5 (C-4),
144.4 (C-7), 145.4 (C-3 and 5), 168.9 (C-9); MS (EI): m/z
(%) = 332 (M+), 179. HRMS (EI) Calcd for C19H24O5
332.1624, found: 332.1625; IR (film): t = 2980, 2935,
2870, 1445, 1385, 1115, 920, 735 cmꢀ1; assignment of the
NMR data according to Refs. 12 and 31.
(a) Allyl bromide (3.3 equiv), K2CO3 (3.3 equiv), EtOH,
2 h, 90 ꢁC, 36%; (b) malonic acid (2.0 equiv), piperidine
(0.3 equiv), pyridine, 4 h, 80 ꢁC, 46%; (c) dicyclohexyl-
carbodiimide (1.1 equiv), 4-(dimethylamino)-pyridine
(0.2 equiv), (R)-(+)-borneol (2.3 equiv, ee > 98%), tetra-
hydrofurane, 16 h, room temperature, 80%; (d)
Pd(PPh3)4 (5 mol-%), morpholine (2.9 equiv), tetrahy-
drofurane, room temperature, 6 h, 66%.
4.4.1. (3,4,5-Triallyloxy)benzylaldehyde (4). Allyl bro-
mide (810 lL, 1.2 g, 9.6 mmol, 3.3 equiv) and K2CO3
(1.33 g, 9.59 mmol, 3.3 equiv) were added to a solution
of trihydroxybenzaldehydeÆmonohydrate (1, 500 mg,
2.9 mmol) in EtOH (10 mL) at room temperature. The
reaction mixture was stirred at 90 ꢁC for 2 h. Buffer,
pH 7 (15 mL) and tert-butyl-methylether (TBME)
(15 mL) were added after cooling to room temperature.
After extraction with TBME (2· 15 mL) the combined
organics were dried (MgSO4), evaporated, and chro-
matographed (EtOAc/cyclohexane 10:1) to give pure 4
(295 mg, 36%). 1H NMR (300.1 MHz, CDCl3/TMS):
d = 4.63–4.67 (m, 6H), 5.20 (d, J = 10.3, 1H), 5.30 (d,
J = 9.2, 2H), 5.35 (d, J = 15.8, 1H), 5.44 (d, J = 17.0,
2H), 6.01–6.14 (m, 3H), 7.11 (s, 2H), 9.82 (s, 1H). IR
(film) m = 2980, 2935, 2870, 1695, 1585, 1440, 1385,
4.5. Human neutrophil elastase activity assay
1325, 1110, 920, 735 cmꢀ1
.
Determination of the HNE activity was performed
according to Refs. 12 and 13 with slight modifications.
Briefly, 250 lL of substrate solution (700 lM MeO-
Suc-Ala-Ala-Pro-Val-pNA in Tris–HCl buffer, 50 lM,
pH 7.5) was mixed with 100 lL of test solution (test sub-
stance solubilized in Tris–HCl buffer, pH 7.5) and vor-
texed. The solution was prewarmed to 37 ꢁC. After
addition of 250 lL enzyme solution (approximately
25 mU) and vortexing, the samples were incubated for
1 h at 37 ꢁC in a shaker water bath. The positive control
consisted of buffer, substrate, and enzyme. The reaction
was stopped by addition of 500 lL soybean trypsin
inhibitor solution (0.2 mg/mL in Tris–HCl buffer, pH
7.5). The samples were vortexed and placed in an ice
bath. The absorbance of the released p-nitroaniline
was measured at 405 nm.
4.4.2. (3,4,5-Triallyloxy)-cinnamic acid (2). Piperidine
(30 lL, 26 mg, 0.30 mmol, 0.3 equiv) and malonic acid
(220 mg, 2.1 mmol, 2.0 equiv) were added to a solution
of aldehyde 4 (290 mg, 1.1 mmol) in pyridine (3 mL) at
room temperature. The reaction mixture was stirred at
88 ꢁC for 4.5 h and then acidified with HCl (3 M). After
extraction with EtOAc (2· 15 mL) the combined organ-
ics were dried (MgSO4) and evaporated to give pure 2
(150 mg, 46%). 1H NMR (300.1 MHz, CDCl3):
d = 4.60–4.62 (m, 6 H), 5.19 (d, J = 10.4, 1H), 5.30 (d,
J = 9.2, 2H), 5.34 (d, J = 14.4, 1H), 5.43 (d, J = 17.1,
2H), 6.02–6.11 (m, 3H), 6.30 (d, J = 16.0, 1H), 6.77 (s,
2H), 7.65 (d, J = 16.0, 1H). IR (film) m = 2980, 2930,
2865, 1670, 1625, 1495, 1385, 1285, 1120, 930, 735 cmꢀ1
.
4.4.3. Bornyl (3,4,5-triallyloxy)-cinnamate (5). To a solu-
tion of acid 2 (80 mg, 0.25 mmol) in THF (3 mL) dicy-
clohexylcarbodiimide (57 mg, 0.28 mmol, 1.1 equiv),
dimethylamino pyridine (5 mg, 0.04 mmol, 0.2 equiv),
were added, and (+)-borneol (90 mg, 0.58 mmol,
2.3 equiv), and the solution was stirred at room temper-
ature for 16 h. The solvents were evaporated, and the
crude mixture chromatographed (EtOAc/cyclohexane
4.6. Statistical analysis
All assays were performed at least in three separated
experiments. Inhibition rates were calculated relative
to the enzyme activity without inhibitor in percent. Sta-
tistical analysis and non-linear regression were per-
formed using GraphPad Prism 4 software. The results

2390
T. Steinbrecher et al. / Bioorg. Med. Chem. 16 (2008) 2385–2390
13. Siedle, B.; Cisielski, S.; Murillo, R.; Loser, B.;
Castro, V.; Klaas, C. A.; Hucke, O.; Labahn, A.;
Melzig, M. F.; Merfort, I. Bioorg. Med. Chem. 2002,
10, 2855.
are expressed as IC50 values with 95% confidence
interval.
14. Bernstein, P. R.; Edwards, P. D.; Williams, J. C. Prog.
Med. Chem. 1994, 31, 59.
Acknowledgments
15. Hernandez, V.; Manez, S.; Recio, M. C.; Giner, R. M.;
Rios, J. L. Eur. J. Pharm. Sci. 2005, 26, 162.
16. Sartor, L.; Pezzato, E.; Garbisa, S. J. Leukoc. Biol. 2002,
71, 73.
17. Beutler, T. C.; Mark, A. E.; Van Schaik, R. C.; Gerber, P.
R.; Van Gunsteren, W. F. Chem. Phys. Lett. 1994, 222,
529.
18. Shirts, M. R.; Pande, V. S. J. Chem. Phys. 2005, 122.
19. Gouda, H.; Kuntz, I. D.; Case, D. A.; Kollman, P. A.
Biopolymers 2003, 68, 16.
The authors thank Prof. Reinhard Bruckner for sup-
¨
porting and funding the organic synthesis part of this
work. Birgit Schneider greatly assisted in synthesizing
the compounds. We thank Prof. David A. Case for his
help with the preparation of the manuscript and the
Amber user community for countless helpful discussions
and support.
20. Miyamoto, S.; Kollman, P. A. Proteins 1993, 16, 226.
21. Simonson, T. Mol. Phys. 1993, 80, 441.
References and notes
22. Case, D. A.; Darden, T. A.; Cheatham, T. E., III;
Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Merz,
K. M.; Wang, B.; Pearlman, D. A.; Crowley, M.; Brozell,
S.; Tsui, V.; Gohlke, H.; Mongan, J.; Hornak, V.; Cui, G.;
Beroza, P.; Schafmeister, C.; Caldwell, J. W.; Ross, W. S.;
Kollman, P. A. AMBER 8; University of California: San
Francisco, 2004.
23. Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.
M.; Zhang, W.; Yang, R.; Cieplak, P.; Luo, R.; Lee, T.;
Caldwell, J.; Wang, J. M.; Kollman, P. J. Comput. Chem.
2003, 24, 1999.
24. Wang, J. M.; Wolf, R. M.; Caldwell, J. W.; Kollman, P.
A.; Case, D. A. J. Comput. Chem. 2004, 25, 1157.
25. Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.;
Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926.
26. Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A.
J. Phys. Chem. 1993, 97, 10269.
27. Frisch, M. J., et al. Gaussian Inc., Gaussian98, 1998.
28. Berendsen, H. J. C.; Postma, J. P. M.; Van Gunsteren, W.
F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684.
29. Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C.
J. Comput. Phys. 1977, 23, 327.
1. Bode, W.; Meyer, E., Jr.; Powers, J. C. Biochemistry 1989,
28, 1951.
2. Doring, G. Pediatr. Pulmonol. 1999, 28, 363.
3. Fitch, P. M.; Roghanian, A.; Howie, S. E. M.; Sallenave,
J. M. Biochem. Soc. Trans. 2006, 34, 279.
4. Pham, C. T. N. Nat. Rev. Immunol. 2006, 6, 541.
5. Meyer-Hoffert, U.; Wingertszahn, J.; Wiedow, O.
J. Invest. Dermatol. 2004, 123, 338.
6. Wiedow, O.; Meyer-Hoffert, U. J. Intern. Med. 2005, 257,
319.
7. Hrenn, A.; Steinbrecher, T.; Labahn, A.; Schwager, J.;
Schempp, C. M.; Merfort, I. Planta Med. 2006, 72, 1127.
8. Melzig, M. F.; Lo¨ser, B.; Lobitz, G. O.; Tamayo-Castillo,
G.; Merfort, I. Pharmazie 1999, 54, 712.
9. Siedle, B.; Hrenn, A.; Merfort, I. Planta Med. 2007, 73,
401.
10. Steinbrecher, T.; Case, D. A.; Labahn, A. J. Med. Chem.
2006, 49, 1837.
11. Chosen lambda values were 0.02544, 0.29707, 0.50000,
0.70292, 0.87076 and 0.97455, with weights selected as
described in the Amber9 manual p155.
12. Lobitz, G. O.; Heilmann, J.; Zschocke, S.; Tamayo-
Castillo, G.; Bauer, R.; Merfort, I. Pharm. Pharmacol.
Lett. 1998, 8, 115.
30. Eicher, T.; Ott, M.; Speicher, A. Synthesis 1996, 755.
31. Fiuza, S. M.; van Besien, E.; Milhazes, N.; Borges, F.;
Marques, M. P. M. J. Mol. Struct. 2004, 693, 103.