Inhibition of monoamine oxidase by 8-benzyloxycaffeine analogues

Article abstract of DOI:10.1016/j.bmc.2009.12.064

Based on recent reports that several (E)-8-styrylcaffeinyl analogues are potent reversible inhibitors of monoamine oxidase B (MAO-B), a series of 8-benzyloxycaffeinyl analogues were synthesized and evaluated as inhibitors of baboon liver MAO-B and recombinant human MAO-A and -B. The 8-benzyloxycaffeinyl analogues were found to inhibit reversibly both MAO isoforms with enzyme-inhibitor dissociation constants (K_i values) ranging from 0.14 to 1.30 μM for the inhibition of human MAO-A, and 0.023-0.59 μM for the inhibition of human MAO-B. The most potent MAO-A inhibitor was 8-(3-methylbenzyloxy)caffeine while 8-(3-bromobenzyloxy)caffeine was the most potent MAO-B inhibitor. The analogues inhibited human and baboon MAO-B with similar potencies. A quantitative structure-activity relationship (QSAR) study indicated that the MAO-B inhibition potencies of the 8-benzyloxycaffeinyl analogues are dependent on the Hansch lipophilicity (π) and Hammett electronic (σ) constants of the substituents at C-3 of the benzyloxy ring. Electron-withdrawing substituents with a high degree of lipophilicity enhance inhibition potency. These results are discussed with reference to possible binding orientations of the inhibitors within the active site cavities of MAO-A and -B.

Full text of DOI:10.1016/j.bmc.2009.12.064

Bioorganic & Medicinal Chemistry 18 (2010) 1018–1028
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Inhibition of monoamine oxidase by 8-benzyloxycaffeine analogues
Belinda Strydom a, Sarel F. Malan a, Neal Castagnoli Jr. b, Jacobus J. Bergh a, Jacobus P. Petzer a,
*
a Pharmaceutical Chemistry, School of Pharmacy, North-West University, Private Bag X6001, Potchefstroom 2520, South Africa
b Department of Chemistry, Virginia Tech and Edward Via College of Osteopathic Medicine, Blacksburg, VA 24061, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Based on recent reports that several (E)-8-styrylcaffeinyl analogues are potent reversible inhibitors of
monoamine oxidase B (MAO-B), a series of 8-benzyloxycaffeinyl analogues were synthesized and evalu-
ated as inhibitors of baboon liver MAO-B and recombinant human MAO-A and -B. The 8-benzyloxycaffei-
nyl analogues were found to inhibit reversibly both MAO isoforms with enzyme–inhibitor dissociation
Received 30 October 2009
Revised 18 December 2009
Accepted 28 December 2009
Available online 6 January 2010
constants (Ki values) ranging from 0.14 to 1.30
lM for the inhibition of human MAO-A, and 0.023–
0.59 M for the inhibition of human MAO-B. The most potent MAO-A inhibitor was 8-(3-methylbenzyl-
l
Keywords:
oxy)caffeine while 8-(3-bromobenzyloxy)caffeine was the most potent MAO-B inhibitor. The analogues
inhibited human and baboon MAO-B with similar potencies. A quantitative structure–activity relation-
ship (QSAR) study indicated that the MAO-B inhibition potencies of the 8-benzyloxycaffeinyl analogues
Monoamine oxidase
Reversible inhibition
8-Benzyloxycaffeine
Caffeine
are dependent on the Hansch lipophilicity (p) and Hammett electronic (r) constants of the substituents
at C-3 of the benzyloxy ring. Electron-withdrawing substituents with a high degree of lipophilicity
enhance inhibition potency. These results are discussed with reference to possible binding orientations
of the inhibitors within the active site cavities of MAO-A and -B.
Quantitative structure–activity relationship
Molecular docking
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
tors are employed as adjuvants to levodopa in the symptomatic
treatment of PD.3 MAO-B inhibitors also may exert a neuroprotec-
Monoamine oxidase A and B (MAO-A and -B) are flavin adenine
dinucleotide (FAD) containing enzymes which catalyze the oxida-
tion of a variety of endogenous and xenobiotic amines in the brain
and peripheral tissues.1 The two isoforms are attached to the outer
mitochondrial membrane and have different substrate and inhibi-
tor selectivities.2 MAO-A preferentially utilizes serotonin and
norepinephrine as substrates and is irreversibly inhibited by clo-
rgyline while MAO-B preferentially utilizes benzylamine as sub-
strate and is irreversibly inhibited by (R)-deprenyl. Both isoforms
catalyze the oxidative deamination of dopamine.2 Due to their
roles in the metabolism of neurotransmittor amines, inhibitors of
MAO-A and -B have been used in the treatment of neurological dis-
orders. MAO-A inhibitors are used to treat depressive illness2 while
MAO-B inhibitors are useful in the treatment of Parkinson’s disease
(PD).3 In the basal ganglia, oxidation by MAO-B appears to be the
major metabolic pathway of dopamine and inhibitors of this
enzyme may therefore slow the depletion of dopamine stores in
the PD brain and posssibly elevate the concentrations of endoge-
nous dopamine and dopamine produced from administered
levodopa.4–6 In primates, MAO-B inhibitors have been shown to
enhance the elevation of dopamine concentrations in the striatum
following levodopa treatment.7 As a consequence, MAO-B inhibi-
tive effect by reducing the formation of potentially toxic side-prod-
ucts associated with the metabolism of monoamines. These
include H2O2 and aldehydes that may be neurotoxic if not rapidly
metabolized to inactive compounds.1 Since MAO-B activity as well
as density increases in most brain regions with age, MAO-B inhibi-
tion may be especially relevant as a treatment strategy in the aged
parkinsonian brain.8,9
We have reported recently that several (E)-8-styrylcaffeinyl
analogues (1) (Fig. 1) are potent and reversible inhibitors of mono-
R
O
1a:R = Cl
1b: R = Br
N
N
(E)
N
O
N
(E)-8-styrylcaffeinyl analogues (1a-b)
O
N
N
N
O
N
Caffeine (2)
* Corresponding author. Tel.: +27 18 2992206; fax: +27 18 2994243.
E-mail address: jacques.petzer@nwu.ac.za (J.P. Petzer).
Figure 1. The structures of (E)-8-styrylcaffeinyl analogues (1a–b) and caffeine (2).
0968-0896/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.12.064

B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
1019
amine oxidase B (MAO-B).10–12 The styryl side chain of these ana-
2. Results
logues is an important structural feature for inhibition since caf-
feine (2) is a weak inhibitor of MAO-B.13 A possible explanation
for this effect is that (E)-8-styrylcaffeinyl analogues may exhibit
a dual binding mode within MAO-B with the caffeinyl ring located
in the substrate cavity of the enzyme while the styryl side chain
extends into the entrance cavity.10 This would allow for more pro-
ductive binding interactions with the enzyme and hence more po-
tent inhibition. In contrast, caffeine is predicted to bind to either
the substrate or entrance cavity. This view is supported by crystal
structures of MAO-B that show that several potent reversible
inhibitors span both active site cavities. For example, in the com-
plex between safinamide (3) (Fig. 2) and MAO-B, the 3-fluoroben-
zyloxy side chain of safinamide is located in the entrance cavity
while the propanamidyl moiety binds within the substrate cav-
ity.14 Similarly, the 3-chlorobenzyloxy side chain of 7-(3-chlorob-
enzyloxy)-4-formylcoumarin (4) binds in the entrance cavity of
the enzyme with the coumarin ring occupying the substrate cav-
ity.14 Based on these observations it may be concluded that the sty-
ryl side chain of (E)-8-styrylcaffeinyl analogues and the benzyloxy
side chains of 3 and 4 likely exhibit similar binding modes within
the entrance cavity of MAO-B and are essential in stabilizing the
complexes between the enzyme and the respective inhibitors.
Since the styryl and benzyloxy side chains appear to have similar
biological properties with respect to binding to MAO-B, a series
of 8-benzyloxycaffeinyl analogues (5a–g) were synthesized and
evaluated as inhibitors of baboon liver MAO-B and recombinant
human MAO-B in the present study. By comparing the inhibition
potencies of these analogues with those of the corresponding (E)-
8-styrylcaffeinyl analogues, the efficacy by which the benzyloxy
side chain enhances the affinity of inhibitors for the active site of
MAO-B may be compared to that of styryl substitution. The 8-ben-
zyloxycaffeinyl analogues were also evaluated as potential inhibi-
tors of recombinant human MAO-A. Previous studies have
suggested that structures with a relatively larger degree of confor-
mational freedom may be better suited for binding to MAO-A than
relatively rigid structures.13 Since the benzyloxy side chain is rela-
tively flexible and free to rotate about the carbon–oxygen ether
bond, 8-benzyloxycaffeinyl analogues also could bind to the
MAO-A active site. These studies may contribute to the discovery
of a new class of potent reversible MAO inhibitors.
2.1. Chemistry
The 8-benzyloxycaffeinyl analogues (5a–g) (Fig. 2) examined in
this study were prepared according to a procedure previously re-
ported for the synthesis of 5a (Scheme 1).15 The appropriate benzyl
alcohol derivatives (6) were heated with dry 8-chlorocaffeine (7) at
high temperatures (150–170 °C) in the presence of metallic so-
dium. The products so obtained were recrystallized from ethanol
to give 5a–g in low to fair yields (13–44%). 8-Chlorocaffeine, in
turn, was synthesized in high yield by reacting chlorine with caf-
feine in chloroform.16 The structures and purity of the target com-
pounds were verified by mass spectrometry, 1H NMR and 13C NMR.
For previously reported 5a, the melting point obtained corre-
sponded to the literature value as cited in Section 4.
2.2. General enzymology
The 8-benzyloxycaffeinyl analogues (5a–g) examined here were
evaluated initially as inhibitors of baboon liver mitochondrial
MAO-B and then as inhibitors of recombinant human MAO-B and
MAO-A. The IC50 values (concentration of the inhibitor that pro-
duces 50% inhibition) for the inhibition of MAO by each test com-
pound were determined. Since the test inhibitors exhibited a
competitive mode of inhibition (see below), the IC50 values were
converted to the corresponding Ki values (enzyme–inhibitor disso-
ciation constants) using the Cheng–Prusoff equation.12,17 These Ki
values enabled the calculation of MAO-A/B selectivity ratios. Since
a variety of caffeinyl analogues are reported to inhibit baboon liver
MAO-B,12 a comparison of the Ki values obtained with the baboon
and human enzymes would make it possible to determine if the
inhibition data generated with the baboon enzyme may be extrap-
olated to the human.
To evaluate 5a–g as potential inhibitors of baboon liver mito-
chondrial MAO-B, the enzyme activity measurements were based
on the extent to which the MAO-A/B mixed substrate, 1-methyl-
4-(1-methylpyrrol-2-yl)-1,2,3,6-tetrahydropyridine (MMTP), is
oxidized to the corresponding dihydropyridinium metabolite
(MMDP+).18 The production of MMDP+ was measured spectropho-
tometrically at a wavelength of 420 nm where neither the sub-
strate (MMTP) nor the test inhibitors absorb. Inactivation of the
MAO-A isoform was deemed unnecessary since baboon liver mito-
chondrial fractions are devoid of MAO-A activity.18 For the purpose
of converting the IC50 values of the test inhibitors to Ki values, a Km
H
CH3
O
NH2
NH
F
value of 68.3
lM for the oxidation of MMTP by baboon liver MAO-
O
B was used.12
To evaluate 5a–g as potential inhibitors of recombinant human
Safinamide (3)
MAO-A and MAO-B, the enzyme activity measurements were based
CHO
O
Cl
O
O
O
R
N
N
Cl
+
HO
N
O
N
7-(3-Chlorobenzyloxy)-4-formylcoumarin (4)
6
O
7
a: R = 3-H
O
R
b: R = 3-Cl
c: R = 3 -B r
d: R = 3-F
e: R = 3-CF3
f: R = 3 -CH 3
R
N
N
N
N
N
O
i
O
O
N
N
O
N
g: R = 3-OCH3
8-Benzyloxycaffeinyl analogues (5a-g )
5a-g
Figure 2. The structures of safinamide (3), 7-(3-chlorobenzyloxy)-4-formylcoum-
arin (4) and 8-benzyloxycaffeinyl analogues 5a–g.
Scheme 1. Synthetic pathway to the 8-benzyloxycaffeinyl analogues 5a–g. Reagent
and condition: (i) Na, 170 °C.

1020
B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
on the extent to which kynuramine is oxidized to 4-hydroxyquino-
line by the MAO isoforms.19 The formation of 4-hydroxyquinoline
was measured fluorometrically at excitation and emission wave-
lengths of 310 and 400 nm, respectively. None of the test inhibitors
fluoresced at this excitation/emission wavelength or quenched the
fluorescence of 4-hydroxyquinoline at the concentrations used for
the inhibition studies. This fluorometric protocol is considerably
more sensitive than the spectrophotometric method described
above and was thus more suitable for the measurement of recombi-
nant human MAO-A (0.0075 mg protein/mL) and MAO-B (0.015 mg
protein/mL) activities, which were relatively low compared to the
activity of baboon liver mitochondrial MAO-B (0.15 mg protein/
mL) at the enzyme concentrations used for this study. For the pur-
pose of converting the IC50 values of the test inhibitors to Ki values
(via the Cheng–Prusoff equation),17 Km values for the oxidation of
kynuramine by recombinant human MAO-A and MAO-B were deter-
6
5
4
3
2
1
-4
-3
-2
-1
0
1
2
Log [5c]
Figure 3. The sigmoidal dose–response curve of the initial rates of oxidation of
kynuramine by recombinant human MAO-B vs. the logarithm of concentration of
inhibitor 5c (expressed in nM). The rates are expressed as nmoles 4-hydroxyquin-
oline formed/min/mg protein. The concentration kynuramine used was 30 lM, the
determinations were carried out in duplicate and all values are the mean SD.
mined (see Section 4).12 These were found to be 16.1 0.21
22.7 0.72 M for MAO-A and -B, respectively (data not shown).
lM and
l
2.3. MAO-B inhibition studies
at C-3 of the benzyloxy ring enhances MAO-B inhibition potency.
Also, the inhibition potencies recorded with baboon and human
MAO-B were found to be relatively similar. Based on this observa-
tion it can be concluded that the interactions of reversible inhibi-
tors with human and baboon MAO-B are similar and that there
are not many significant differences between the geometries and
amino acid residues of the active sites of these two enzymes.
To determine the modes of inhibition, sets of Lineweaver–Burk
plots were constructed for the inhibition of baboon MAO-B by one
representative inhibitor, compound 5e. As illustrated by example
in Figure 4, the lines of the Lineweaver–Burk plots intersected
which indicates that the mode of inhibition is competitive and that
inhibition is therefore reversible. Essentially the same result was
obtained when using recombinant human MAO-B as enzyme
source (data not shown). These results are in accordance with
expectation since (E)-8-styrylcaffeinyl analogues are also reported
to inhibit MAO-B competitively.10–12 To further confirm that the
test inhibitors bind reversibly in the active site of MAO-B, one rep-
resentative inhibitor, compound 5b, was selected for reversibility
studies. Baboon liver mitochondria were preincubated with 5b
As shown in Table 1, all of the 8-benzyloxycaffeinyl analogues
(5a–g) evaluated were found to be inhibitors of baboon liver and
recombinant human MAO-B (Fig. 3). The most potent inhibitor
was 5c with an IC50 value of 0.068
binant human MAO-B. This corresponds with a calculated Ki value
of 0.023 M (Table 2). For this series, the Ki values ranged from
0.023 to 0.59 M, indicating that all the test compounds are
lM for the inhibition of recom-
l
l
relatively potent inhibitors. Since the unsubstituted homolog 5a
was the weakest inhibitor, it can be concluded that substitution
Table 1
The IC50 values for the inhibition of baboon liver MAO-B, human MAO-B and human
MAO-A by 8-benzyloxycaffeinyl analogues 5a–g
O
R
N
N
O
N
N
O
(0.28
60 min and the rates of the MAO-B catalyzed oxidation of MMTP
(50
M) to MMDP+ were measured.20,21 As shown in Figure 5, there
is no decrease of MAO-B catalytic rate with increased incubation
time. Similarly, when 5b (0.23
M, approximately 2 ꢀ IC50) was
preincubated with recombinant human MAO-B for periods of 0,
15, 30, and 60 min, the rate of kynuramine (30 M) oxidation did
not decrease with increased incubation time (data not shown).
From these analyses it can be concluded that 5b interacts revers-
ibly with the active sites of baboon and human MAO-B.
l
M, approximately 2 ꢀ IC50) for periods of 0, 15, 30, and
R
IC50 MAO-B
IC50 MAO-B
(human)
IC50 MAO-A
(human) lM
(baboon)
lM
lM
l
5a
5b
5c
5d
5e
5f
H
Cl
Br
F
CF3
CH3
OCH3
2.49 0.041
0.120 0.052
0.113 0.006
0.534 0.091
0.112 0.023
0.698 0.010
1.59 0.609
1.77 0.214
0.107 0.019
0.068 0.004
0.542 0.060
0.152 0.004
0.546 0.032
1.01 0.279
1.24 0.050
0.666 0.015
0.941 0.009
1.07 0.131
3.72 0.253
0.397 0.018
3.15 0.024
l
l
5g
All values are expressed as the mean SD of duplicate determinations.
Table 2
The calculated Ki values for the inhibition of baboon liver MAO-B, human MAO-B and human MAO-A by 8-benzyloxycaffeinyl analogues 5a–g
R
Ki MAO-B (baboon)a
lM
Ki MAO-B (human)a
lM
Ki MAO-A (human)a
l
M
SIb
5a
5b
5c
5d
5e
5f
H
Cl
Br
F
CF3
CH3
OCH3
1.44
0.59
0.43
0.23
0.33
0.37
1.30
0.14
1.10
1.37
0.16
0.069
0.49
0.039
1.29
0.31
0.069
0.065
0.31
0.065
0.40
0.036
0.023
0.18
0.051
0.18
5g
0.92
0.34
a
The Ki values were calculated from the experimental IC50 values according to the equation by Cheng and Prusoff: Ki = IC50/(1 + [S]/Km) with [S] = 50
lM and Km
(MMTP) = 68.3 lM for baboon liver MAO-B. For human MAO-B, [S] = 30 lM and Km (kynuramine) = 22.7 lM while [S] = 45 lM and Km (kynuramine) = 16.1 lM for human
MAO-A.17
b
The selectivity index is the selectivity for the A isoform and is given as the ratio of Ki(MAO-B)/Ki(MAO-A).

B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
1021
0.75
0.50
0.25
0.06
0.05
0.04
0.03
0.02
0.01
-0.050
-0.025
0.000
0.025
0.050
0.075
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
1/[S]
1/[S]
Figure 6. Lineweaver–Burk plots of the oxidation of kynuramine by recombinant
human MAO-A in the absence (open squares) and presence of various concentra-
Figure 4. Lineweaver–Burk plots of the oxidation of MMTP by baboon liver MAO-B
in the absence (open squares) and presence of various concentrations of 5e (filled
squares, 0.0325 lM; open circles, 0.065 lM and filled circles, 0.130 lM). The rate
tions of 5e (filled squares, 0.931
lM; open circles, 1.861 lM and filled circles,
3.722 M). The rate (V) is expressed as nmol product formed/min/mg protein.
l
(V) is expressed as nmol product formed/min/mg protein.
40
30
20
10
0
4
3
2
1
0
0
15
30
60
0
15
30
60
Incubation time (minutes)
Incubation time (minutes)
Figure 7. Time-dependant inhibition of the recombinant human MAO-A catalyzed
oxidation of kynuramine (45 M) by 5b. The enzyme preparation was preincubated
for various periods of time (0–60 min) with 5b (1.31 M). The rates are expressed as
nmoles 4-hydroxyquinoline formed/min/mg protein.
Figure 5. Time-dependant inhibition of the baboon liver mitochondrial MAO-B
catalyzed oxidation of MMTP (50 M) by 5b. The enzyme preparation was
l
l
l
preincubated for various periods of time (0–60 min) with 5b (0.28 lM). The rates
are expressed as nmoles MMDP+ formed/min/mg protein.
incubation time which indicates that 5b binds reversibly to the ac-
2.4. MAO-A inhibition studies
tive site of human MAO-A.
Interestingly, the 8-benzyloxycaffeinyl analogues (5a–g) were
2.5. Quantitative structure–activity relationship (QSAR) studies
also inhibitors of recombinant human MAO-A. The most potent
inhibitor was 5f with an IC50 value of 0.397
corresponds to a calculated Ki value of 0.14
values ranged from 0.14 to 1.30 M, indicating that 5a–g are mod-
erate to potent inhibitors of MAO-A. In general, however, these
analogues were better MAO-B inhibitors. Based on the selectivity
indexes (Table 2), compound 5e displayed the highest selectivity
for MAO-B (25-fold). The most potent MAO-B inhibitor of the ser-
ies, 5c, also exhibited a high degree of selectivity for MAO-B (14-
fold). Compounds 5a and 5f were essentially nonselective.
To determine the mode of MAO-A inhibition, sets of Linewe-
aver–Burk plots were constructed for the inhibition of human
MAO-A by compound 5e, which was selected as a representative
inhibitor. As illustrated by the example in Figure 6, the lines of
the Lineweaver–Burk plots intersected which indicates that, simi-
lar to the inhibition of MAO-B, the mode of inhibition of MAO-A
is competitive and reversible. The reversibility of inhibition was
further demonstrated by preincubating recombinant human
l
M (Table 1) which
The results of the MAO-B inhibition studies (Tables 1 and 2)
indicate that substitution at C-3 of the benzyloxy ring enhances
MAO-B inhibition potency. To further investigate the effect of the
substituents on the MAO-B inhibitory activity, a Hansch-type QSAR
study was carried out. Five parameters were used to describe the
physiochemical properties of the substituents. The Van der Waals
volume (Vw)22 and Taft steric parameter (Es)23 served as descrip-
tors of the bulkiness of the substituents, while the lipophilicities
l
M (Table 2). The Ki
l
were described by the Hansch constant (
p
).23 The electronic prop-
m) and
erties were described by the classical Hammett constant (
r
the Swain–Lupton constant (F).23 The values of these physicochem-
ical parameters were obtained from standard compilations.22,23
QSAR studies were carried out with the inhibition data obtained
with baboon liver MAO-B and recombinant human MAO-B as en-
zyme sources and the results are summarized in Tables 3 and 4,
respectively. Using the data generated with baboon liver MAO-B,
a significant correlation was observed between the lipophilicity
MAO-A with compound 5b (1.31
periods of 0, 15, 30, and 60 min, followed by measuring the rates
of MAO-A catalyzed oxidation of kynuramine (45 M). As shown
in Figure 7, MAO-A catalytic rate is not reduced with increased
lM, approximately 2 ꢀ IC50) for
constant, p
, and the inhibition potency (log IC50) with an R2 value
of 0.82. The statistical F value for this correlation was 27.8, which
is higher than Fmax value (25.32) for 95% significance.24 A higher F
value indicates a better fit. Improved correlations were obtained
l

1022
B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
Table 3
for the binding affinity of 8-benzyloxycaffeinyl analogues (5a–g) to
baboon liver MAO-B is therefore:
Correlations of the baboon liver MAO-B inhibition potencies (log IC50) of 8-benzyl-
oxycaffeinyl analogues 5a–g with steric, electronic and hydrophobic descriptors of
the substituents at C-3 of the benzyloxy ringa
log IC50 ¼ ꢁ1:23ðꢂ0:19Þrm ꢁ 0:93ðꢂ0:10Þ
ðR2 ¼ 0:98 and F ¼ 147Þ
p
þ 0:31ðꢂ0:05Þ
Parameter
Slope
y-Intercept
R2
F
Significanceb
ð1Þ
rm
F
Vw
p
ꢁ2.25 0.68
ꢁ1.96 0.89
ꢁ0.49 0.44
ꢁ1.29 0.24
0.53 0.22
ꢁ1.23 0.19
ꢁ0.93 0.10
ꢁ1.12 0.19
ꢁ1.07 0.10
0.12 0.20
0.18 0.30
0.045 0.46
0.19 0.14
0.13 0.27
0.31 0.05
0.62
0.39
0.02
0.82
0.44
0.98
10.8
4.87
1.13
27.8
5.61
0.022
0.079
0.337
0.003
0.065
0.003
0.0007
0.005
0.0004
A two-parameter fit with the Swain–Lupton constant (F) and
p
also yielded good correlations with the inhibition potency
(log IC50) and R2 and F values of 0.98 and 120 (Fmax = 30.18),
Es
r
+
p
147
respectively, were recorded. The negative signs of the
F
(ꢁ1.12 0.19) and rm (ꢁ1.23 0.19) parameter coefficients indi-
cate that the inhibition potencies of 5a–g towards baboon liver
MAO-B may be enhanced by substitution with electron withdraw-
ing groups at C-3 of the benzyloxy ring. The negative correlations
F +
p
0.41 0.064
0.98
120
a
The logarithm of the IC50 values (expressed in
regression analysis.
lM) was used in the linear
b
of the inhibition potencies with the
p constant indicate that sub-
The significance is the fractional probability that the coefficient of the added
variable is zero.
stituents at C-3 of the benzyloxy ring with enhanced lipophilicity
may result in more potent inhibitors.
Using the inhibition data generated with recombinant human
MAO-B as enzyme source, the best correlation observed was be-
Table 4
Correlations of the recombinant human MAO-B inhibition potencies (log IC50) of 8-
benzyloxycaffeinyl analogues 5a–g with steric, electronic and hydrophobic descrip-
tors of the substituents at C-3 of the benzyloxy ringa
tween the lipophilicity constant,
p, and the inhibition potency
(log IC50). The R2 and statistical F values of this correlation were
0.75 and 18.8 (Fmax = 25.32), respectively. A slightly improved cor-
relation was obtained with the addition of a second substituent
Parameter
Slope
y-Intercept
R2
F
Significanceb
rm
F
Vw
p
ꢁ2.14 0.71
ꢁ1.97 0.92
ꢁ0.56 0.43
ꢁ1.27 0.29
0.45 0.27
ꢁ1.10 0.54
ꢁ0.95 0.28
ꢁ1.15 0.42
ꢁ1.04 0.21
0.01 0.23
0.071 0.31
0.024 0.44
0.075 0.17
0.061 0.32
0.18 0.14
0.51
0.37
0.11
0.75
0.23
0.84
7.31
4.58
1.72
18.8
2.83
0.043
0.085
0.247
0.008
0.153
0.112
0.027
0.052
0.008
parameter. A two-parameter fit of F and
p versus the inhibition po-
tency (log IC50) yielded a model with an R2 value of 0.89 and a sta-
tistical F value of 25.4 (Fmax = 30.18). Similarly, a two-parameter fit
of
Es
r
+
p
17.3
rm and
p
versus the inhibition potency yielded a model with an R2
value of 0.84 and a statistical F value of 17.3 (Fmax = 30.18).
Although these correlations are not statistically significant, the re-
sults are similar to those obtained with baboon liver MAO-B and
may be viewed as further support of the notion that the interac-
tions of reversible inhibitors with human and baboon MAO-B are
similar.
F +
p
0.30 0.14
0.89
25.4
a
The logarithm of the IC50 values (expressed in
regression analysis.
The significance is the fractional probability that the coefficient of the added
variable is zero.
lM) was used in the linear
b
A similar analysis using the inhibition data obtained with re-
combinant human MAO-A as enzyme source failed to yield mean-
ingful correlations between the inhibition potencies and the
substituent parameters.
with the addition of a second substituent parameter. A two-param-
eter fit with the Hammett electronic parameter (rm) and p versus
the inhibition potency (log IC50) yielded the best correlation with
2.6. Molecular modeling
an R2 of 0.98 and a statistical F value of 147 (Fmax = 30.18)
(Fig. 8). For this correlation the probability that rm and
is 0.3% and 0.07%, respectively. The best mathematical description
p are zero
In an attempt to obtain insights into the possible binding modes
of the 8-benzyloxycaffeinyl analogues (5a–g) within the active
sites of MAO-A and -B, molecular docking experiments were per-
formed. The X-ray crystal structures of recombinant human
MAO-A and MAO-B were obtained from the Brookhaven Protein
Data Bank. The structures of MAO-A co-crystallized with harmine
(PDB code: 2Z5X)25 and MAO-B co-crystallized with safinamide
(PDB code: 2V5Z)14 were selected. These models were selected
based on the relatively high resolution of the crystallographic
structures and the observation that, in the complex between
MAO-B and safinamide, the side chain of Ile-199 is rotated out of
the normal conformation to allow for fusion of the entrance and
substrate cavities. This is the preferred conformation when rela-
tively large inhibitors, which span both the entrance and substrate
cavities, bind to the active site of MAO-B.26 In contrast, the active
site of MAO-A consists of a single cavity.
For the purpose of the docking studies, the LigandFit application
of the modeling software Discovery Studio 1.727 was used follow-
ing a modification of a previously reported protocol.13,28 The struc-
tures of 5a–g were built and geometry optimized in Discovery
Studio and prepared for docking with the Prepare Ligands protocol.
Following the addition of hydrogen atoms and the correction of the
valences of the FAD co-factors and co-crystallized ligands, the pro-
tein structures were subjected to an energy minimization while
the protein backbone was constrained (see Experimental). Subse-
0.6
0.4
0.2
H
OCH3
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
CH3
F
CF3
Cl
Br
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
π - (-1.23σ)
Figure 8. Correlations of the log IC50 values for the inhibition of baboon liver MAO-
B by 5a–g with the Hansch constant ( ) of the substituents at C-3 of the benzyloxy
ring. The values were adjusted by the contribution of the Hammett constant ( ) as
p
p
r
indicated on the x-axis title. The linear regression line is a representation of
equation 1 with a correlation coefficient of 0.98.

B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
1023
quently, the energy-minimized structures were deprived of the co-
crystallized ligands and the backbone constraints. Following the
docking procedure, the docked ligand conformations were further
refined using the Smart Minimizer algorithm. Ten possible docking
poses were calculated for each inhibitor. To evaluate the accuracy
of this protocol, the co-crystallized ligands, harmine and safina-
mide, were redocked into the active sites of MAO-A and -B, respec-
tively. The best-ranked docking solutions exhibited relatively small
RMSD values from the co-crystallized ligands of 0.65 and 1.79 for
harmine and safinamide, respectively. It may therefore be
concluded that this protocol is suitable for docking reversible
inhibitors into active site models of MAO-A and -B.
The best-ranked docking solutions obtained with the structure
of MAO-B showed one prevailing pose for all of the 8-benzyloxy-
caffeinyl analogues (5a–g). The inhibitors traverse both cavities
of MAO-B with the caffeinyl moiety oriented towards the FAD
co-factor in the substrate cavity while the benzyloxy side chain
protrudes into the entrance cavity. As shown by example with
inhibitor 5a, the carbonyl oxygen at C-6 of the caffeinyl ring forms
hydrogen bonds with the phenolic hydrogen of Tyr-435 (Fig. 9a).
The amide functional group of the Gln-206 side chain possibly
interacts with the caffeinyl ring via
p
–
p
interactions with an inter-
plane distance of approximately 4.0 Å.25 The benzyloxy ring is
stabilized by Van der Waals interactions in the hydrophobic
entrance cavity defined by Phe-103, Trp-119, Leu-164, Leu-167,
Phe-168 and Ile-316.29 Of importance is the observation that the
benzyloxy ring is partly located in space that would have been
occupied by the side chain of Ile-199 if it were in the normal
‘closed’ conformation. In the MAO-A active site the residue corre-
sponding to Ile-199 in MAO-B is Phe-208. Since the increased size
of the Phe aromatic ring compared to the side chain of Ile would
prevent it from rotating into an alternative conformation, this
binding mode of 5a would not be possible in the MAO-A active
site.25,26 As shown in Figure 9b the benzyloxy ring would partially
overlap with Phe-208 in the MAO-A active site.
The best-ranked docking solutions obtained with the structure
of MAO-A showed that the inhibitors bind with the caffeinyl moi-
ety oriented towards the FAD co-factor (Fig. 10a). This binding
mode is similar to that observed in the MAO-B active site. As illus-
trated by example with compound 5a, one hydrogen bond is
observed between the carbonyl oxygen at C-2 of the caffeinyl ring
and the phenolic hydrogen of Tyr-444 in the active site. Interest-
Figure 9. Stereo view representation of compound 5a (cyan) docked within the active site of MAO-B. In panel A the active site residues (grey) of MAO-B are displayed while in
panel B the active site residues of both MAO-B (grey) and MAO-A (yellow) are displayed. The FAD co-factor is shown in yellow and hydrogen bonds are indicated by the
dashes. The residues are numbered based on the structure of MAO-B. The numbers given in parenthesis are the corresponding MAO-A residue identifiers.

1024
B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
Figure 10. Stereo view representation of compound 5a (cyan) docked within the active site of MAO-A. In panel A the active site residues (grey) of MAO-A are displayed while
in panel B the active site residues of both MAO-B (grey) and MAO-A (yellow) are displayed. The FAD co-factor is shown in yellow and hydrogen bonds are indicated by the
dashes. In panel B, the residues are numbered based on the structure of MAO-B. The numbers given in parenthesis are the corresponding MAO-A residue identifiers.
ingly, the caffeinyl ring is rotated through ꢃ180° compared to the
orientation in MAO-B. As expected, the benzyloxy side chains of
the inhibitors are rotated to avoid unfavourable interactions with
the side chain of Phe-208. For 5a, the benzyloxy side chain is bent
at the CH2–O ether bond by about a 34° angle from the plane of the
caffeinyl ring (Fig. 11). The relatively large degree of conforma-
tional freedom of the benzyloxy side chain may be an important
structural feature required for MAO-A inhibition. As mentioned
in Section 1, the benzyloxy side chain is relatively flexible and free
to rotate about the carbon–oxygen ether bond which would enable
Figure 11. The docked orientations of 5a within the active site of MAO-B (cyan) and
MAO-A (green) with the caffeine rings of the two respective orientations
superimposed.
these inhibitors to avoid structural overlap with MAO-A active site
amino acid residues.13 This bended orientation would not be pos-
sible in MAO-B since the benzyloxy ring would partially overlap
with Tyr-326 in the MAO-B active site (Fig. 10b). Other interactions
3. Discussion
between the inhibitors and the enzyme include
p–p interactions
between the caffeinyl ring and the amide functional group of the
Gln-215 side chain, with an interplane distance of approximately
3.5 Å.25 The benzyloxy side chain may be stabilized via Van der
Waals interactions with Val-210, Ile-325, Ile-335 and Leu-337.30
As mentioned in the Introduction, based on the observation that
several (E)-8-styrylcaffeinyl analogues are potent reversible inhib-
itors of MAO-B,10–12 a series of 8-benzyloxycaffeinyl analogues

B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
1025
(5a–g) were synthesized and evaluated as inhibitors of baboon li-
ver MAO-B and recombinant human MAO-B. The 8-benzyloxycaf-
feinyl analogues were found to inhibit reversibly and potently
side chain which may facilitate charge transfer or dipole interac-
tions in the entrance cavity. The absence of correlations between
the MAO-A inhibition potency and the substituent descriptor val-
ues used in the QSAR study suggest that the benzyloxy side chain
of the inhibitors interact differently with MAO-A than with MAO-B.
This may be the result of differing amino acid residues between the
active site cavies of MAO-A and -B as well as differing binding ori-
entations of the inhibitors in the active sites of the respective
enzymes.
MAO-B with Ki values ranging from 0.023 to 0.59 lM for the inhi-
bition of the human enzyme. Similar results were obtained using
baboon MAO-B as enzyme source. In general, the 8-benzyloxycaf-
feinyl analogues were found to be approximately equipotent to
the corresponding (E)-8-styrylcaffeinyl analogues as inhibitors of
MAO-B. For example, (E)-8-(3-chlorostyryl)caffeine (1a) (Fig. 1)
has a reported IC50 value of 0.146 lM for the inhibition of mito-
chondrial baboon liver MAO-B while the corresponding 8-benzyl-
4. Experimental
oxycaffeinyl analogue, 5b, inhibits baboon MAO-B with an IC50
Caution —MMTP is a structural analogue of the nigrostriatal
neurotoxin, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP),
and should be handled using disposable gloves and protective eye-
wear. Procedures for the safe handling of MPTP have been described
previously.31
value of 0.120
inhibits baboon MAO-B with an IC50 value of 0.107
corresponding 8-benzyloxycaffeinyl analogue 5c exhibits an IC50
value of 0.113
M.12 These results confirm that the styryl and ben-
l
M.12 Similarly, (E)-8-(3-bromostyryl)caffeine (1b)
lM while the
l
zyloxy side chains have similar properties with respect to binding
to MAO-B and probably exhibit similar binding modes within the
active site of MAO-B. The molecular docking studies suggest that
the 8-benzyloxycaffeinyl analogues bind with the caffeinyl ring lo-
cated within the substrate cavity of MAO-B while the benzyloxy
side chain is located within the entrance cavity. (E)-8-Styrylcaffei-
nyl analogues are also thought to exhibit this dual binding mode
with the styryl side chain extending towards the entrance cavity
of the enzyme.10 The proposal that the benzyloxy and styryl side
chains bind within the entrance cavity of MAO-B is supported by
the observation that, in the crystal structures of safinamide (3)
and 7-(3-chlorobenzyloxy)-4-formylcoumarin (4) in complex with
human MAO-B, the benzyloxy side chains of these inhibitors also
occupy the entrance cavity of the enzyme.14
4.1. Chemicals and instrumentation
All starting materials not described elsewhere were obtained
from Sigma–Aldrich and were used without purification. The oxa-
late salt of MMTP was prepared according to previously reported
procedures.32 Kynuramine.2HBr was obtained from Sigma–Aldrich.
Proton (1H) and carbon (13C) NMR spectra were recorded on a Var-
ian Gemini 300 or on a Bruker Avance III 600 spectrometer. All
NMR measurements were conducted in CDCl3. Chemical shifts
are reported in parts per million (d) downfield from the signal of
tetramethylsilane added to the deuterated solvent. Spin multiplic-
ities are given as s (singlet), d (doublet), t (triplet) or m (multiplet).
Direct insertion electron impact ionization (EIMS) and high resolu-
tion mass spectra (HRMS) were obtained on an Autospec ETOF
(Micromass) mass spectrometer. Melting points (mp) were deter-
mined on a Stuart SMP10 melting point apparatus and are uncor-
rected. UV–vis spectra were recorded on a Shimadzu Multispec-
1501 photodiode array spectrophotometer while fluorescence
spectrophotometry was conducted with a Varian Cary Eclipse fluo-
rescence spectrophotometer. Microsomes from insect cells con-
taining recombinant human MAO-A and -B (5 mg/mL) were
obtained from Sigma–Aldrich.
Interestingly, the 8-benzyloxycaffeinyl analogues were also
found to be reversible inhibitors of recombinant human MAO-A
with Ki values ranging from 0.14 to 1.30 lM. In general, the 8-ben-
zyloxycaffeinyl analogues, however, displayed higher binding
affinities for MAO-B than for MAO-A. The molecular docking stud-
ies indicate that the ability of the 8-benzyloxycaffeinyl analogues
to bind to the MAO-A active site may, to a large degree, depend
on the relatively large degree of rotation freedom of the benzyloxy
side chain at the carbon–oxygen ether bond. This is in accordance
with previous studies which suggested that relatively flexible
inhibitors may be better accommodated in the MAO-A active site,
by avoiding unfavourable interactions with active site amino acid
residues.13
4.2. Synthesis of 8-chlorocaffeine (7)
8-Chlorocaffeine was prepared according to a previously de-
scribed procedure by reaction of caffeine (2) with chlorine in chlo-
roform.16 The required chlorine gas was obtained from the reaction
between concentrated HCl and KMnO4 and was dried by passing
through concentrated H2SO4. High yields of 8-chlorocaffeine was
obtained (94–100%) and the melting point of 188 °C corresponds
to that reported in the literature (188 °C).33 NMR data also corre-
lated to the corresponding published values.34
In the present study, substitution at C-3 of the benzyloxy ring
has been shown to have a considerable effect on MAO-B inhibition
potencies of 8-benzyloxycaffeinyl analogues. Since the benzyloxy
side chain is predicted to bind within the entrance cavity of
MAO-B, different interactions between the aromatic moiety, facili-
tated by the different C-3 substituents, and the entrance cavity
residues may explain this apparent SAR. The entrance cavity of
MAO-B is reported to be composed of hydrophobic amino acid res-
idues where hydrophobic inhibitor side chains may be stabilized
by Van der Waals interactions.29 It can therefore be expected that
hydrophobic substituents at C-3 of the benzyloxy ring should en-
hance binding affinity of the 8-benzyloxycaffeinyl analogues to
MAO-B. Accordingly, the QSAR study indicates that the inhibition
potencies of the test inhibitors correlate with the lipophilicity con-
stant of the C-3 substituents, with a higher degree of lipophilicity
leading to enhanced inhibition potency. These results further sup-
port the findings of the molecular docking studies which suggest
that the benzyloxy side chain is bound to the entrance cavity of
MAO-B. An explanation for the observed correlations between
the MAO-B inhibition potency and the electronic parameters (rm
and F), and thus enhancement of inhibition potency by electron
withdrawing C-3 substituents, may be that electron withdrawing
groups induce an electron deficiency or dipole in the benzyloxy
4.3. General procedure for the synthesis of the 8-benzyloxycaffeinyl
analogues (5a–g)
The 8-benzyloxycaffeinyl analogues (5a–g) were prepared
according to a previously reported procedure.15 Metallic sodium
(1.76 mmol) was allowed to react with the appropriately substi-
tuted benzyl alcohol (6, 24.65 mmol). 8-Chlorocaffeine (7,
1.75 mmol) was added to the mixture and the reaction was heated
at 170 °C for a period of 6 h. After the reaction was cooled, the 8-
benzyloxycaffeine analogues were obtained by crystallization. In
most cases the yields obtained were improved by the addition of
10–20 mL ethanol and cooling the mixture to 4 °C during the crys-
tallization process. For previously reported 5a the melting point
was found to be 173 °C while the literature value is 172–173.5 °C.15

1026
B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
4.3.1. 8-Benzyloxycaffeine (5a)
159.80; EIMS m/z 330 (MÅ+); HRMS calcd 330.13281, found
330.13554.
The title compound was prepared from 8-chlorocaffeine (7) and
benzyl alcohol in a yield of 29%: mp 173 °C, lit mp 172–173.5 °C;15
1H NMR (CDCl3) d 3.36 (s, 3H), 3.52 (s, 3H), 3.69 (s, 3H), 5.47 (s, 2H),
7.32–7.46 (m, 5H); 13C NMR (CDCl3) d 29.79 (CH3), 29.71 (CH3),
27.68 (CH3), 72.52 (CH2), 103.53 (C), 135.00 (CH), 128.84 (CH),
128.66 (CH), 128.41 (CH), 155.51 (C), 154.79 (C), 151.66 (C),
146.16 (C); EIMS m/z 300 (MÅ+); HRMS calcd 300.12224, found
300.12351.
4.4. Mitochondrial baboon liver MAO-B inhibition studies
Baboon liver mitochondrial fractions were isolated as described
previously and stored at ꢁ70 °C.35 Following addition of an equal
volume of sodium phosphate buffer (100 mM, pH 7.4) containing
glycerol (50%, w/v) to the mitochondrial isolate, the protein concen-
tration was determined by the method of Bradford using bovine
serum albumin as reference standard.36 MMTP (Km = 68.3
4.3.2. 8-(3-Chlorobenzyloxy)caffeine (5b)
The title compound was prepared from 8-chlorocaffeine (7) and
3-chlorobenzyl alcohol in a yield of 22%: mp 174 °C; 1H NMR
(CDCl3) d 3.40 (s, 3H), 3.54 (s, 3H), 3.74 (s, 3H), 5.42 (s, 2H), 7.45
(s, 1H), 7.35 (m, 3H); 13C NMR (CDCl3) d 29.85 (CH3), 29.70 (CH3),
27.69 (CH3), 71.49 (CH2), 103.63, 126.34, 128.44, 128.96, 129.96,
134.58, 136.96, 146.03, 151.62, 154.79, 155.17; EIMS m/z 334
(MÅ+); HRMS calcd 334.08327, found 334.08352.
1.60
enzymatic reactions were conducted in sodium phosphate buffer
(100 mM, pH 7.4) and contained MMTP (50 M), the mitochondrial
isolate (0.15 mg protein/mL) and various concentrations of the test
inhibitors (0–100 M). The stock solutions of the inhibitors were
l
M),12,18 served as substrate for the inhibition studies. The
l
l
prepared in DMSO and were added to the incubation mixtures to
yield a final DMSO concentration of 4% (v/v). DMSO concentrations
higher than 4% are reported to inhibit MAO-B.37 The final volume
of the incubations was 500
10 min, the enzyme reactions were terminated by the addition of
lL. Following incubation at 37 °C for
4.3.3. 8-(3-Bromobenzyloxy)caffeine (5c)
The title compound was prepared from 8-chlorocaffeine (7) and
3-bromobenzyl alcohol in a yield of 33%: mp 169 °C; 1H NMR
(CDCl3) d 3.34 (s, 3H), 3.49 (s, 3H), 3.68 (s, 3H), 5.43 (s, 2H), 7.58
(s, 1H), 7.48 (d, 1H), 7.36 (d, 1H), 7.22 –7.29 (t, 1H); 13C NMR
(CDCl3) d 29.86 (CH3), 29.71 (CH3), 27.69 (CH3), 71.43 (CH2),
103.64, 126.85, 130.23, 131.41, 131.90, 122.66, 137.18, 146.03,
151.62, 154.79, 155.15; EIMS m/z 378 (MÅ+); HRMS calcd
380.03070, found 380.02580.
10 lL perchloric acid (70%). The MAO-B catalyzed production of
MMDP+ is reported to be linear for the first 10 min of incubation un-
der these conditions.12 The samples were centrifuged at 16,000g for
10 min, and the concentrations of the MAO-B generated product,
MMDP+, were measured spectrophotometrically at 420 nm
(e
= 25,000 Mꢁ1) in the supernatant fractions.18 The IC50 values were
determined by plotting the initial rates of oxidation versus the loga-
rithm of the inhibitor concentrations to obtain a sigmoidal dose–re-
sponse curve. This kinetic data were fitted to the one site
competition model incorporated into the Prism software package
(GraphPad Software Inc.). The IC50 values were determined in dupli-
cate and are expressed as mean standard deviation (SD). The Ki val-
ues were calculated from the experimental IC50 values according to
the equation by Cheng and Prusoff: Ki = IC50/(1 + [S]/Km) with
4.3.4. 8-(3-Fluorobenzyloxy)caffeine (5d)
The title compound was prepared from 8-chlorocaffeine (7) and
3-fluorobenzyl alcohol in a yield of 44%: mp 185 °C; 1H NMR
(CDCl3) d 3.35 (s, 3H), 3.5 (s, 3H), 3.69 (s, 3H), 5.43 (s, 2H), 6.98–
7.08 (m, 1H), 7.11–7.22 (m, 2H), 7.30–7.38 (m, 1H); 13C NMR
(CDCl3) d 29.83 (CH3), 29.69 (CH3), 27.69 (CH3), 71.51 (CH2),
103.62, 115.29, 115.86, 123.74, 130.33, 137.45, 155.19, 146.05,
151.62, 161.18, 164.46; EIMS m/z 318 (MÅ+); HRMS calcd.
318.11282, found 318.11033.
[S] = 50 lM and Km (MMTP) = 68.3 l
M.17
4.5. Recombinant human MAO-A and -B inhibition studies
Microsomes from insect cells containing recombinant human
MAO-A and -B (5 mg/mL) were obtained from Sigma–Aldrich, pre-
aliquoted and stored at ꢁ70 °C. All enzymatic reactions were carried
out in potassium phosphate buffer (100 mM, pH 7.4, made isotonic
with KCl) containing MAO-A (0.0075 mg/mL) or MAO-B (0.015 mg/
4.3.5. 8-[(3-Trifluoromethyl)benzyloxy]caffeine (5e)
The title compound was prepared from 8-chlorocaffeine (7) and
3-(trifluoromethyl)benzyl alcohol in a yield of 22%: mp 146 °C; 1H
NMR (CDCl3) d 3.34 (s, 3H), 3.56 (s, 3H), 3.69 (s, 3H), 5.50 (s, 2H),
7.27–7.51 (m, 4H); 13C NMR (CDCl3) d 29.85 (CH3), 29.66 (CH3),
27.69 (CH3), 71.52 (CH2), 103.67, 129.23, 131.26, 335.96, 125.65,
131.26, 122.93, 125.65, 146.01, 151.62, 154.81, 155.09; EIMS m/z
368 (MÅ+); HRMS calcd 368.10963, found 368.10714.
mL), various concentrations of the test inhibitor (0–100
kynuramine. Thefinalconcentrations ofkynuramineinthereactions
were 45 M and 30 M where MAO-A and -B, respectively, served as
substrates. The final volume of the reactions was 500 L. Stock solu-
lM) and
l
l
l
tions of the test inhibitors were prepared in DMSO and added to the
reactions to yield a final concentration of 4% (v/v) DMSO. The reac-
tions were incubated for 20 min at 37 °C and terminated with the
4.3.6. 8-(3-Methylbenzyloxy)caffeine (5f)
The title compound was prepared from 8-chlorocaffeine (7) and
3-methylbenzyl alcohol in a yield of 13%: mp 145 °C; 1H NMR
(CDCl3) d 3.42 (s, 3H), 3.58 (s, 3H), 3.74 (s, 3H), 2.43 (s, 3H), 5.49
(s, 2H), 7.24–7.34 (m, 4H); 13C NMR (CDCl3) d 21.30 (CH3), 27.64
(CH3), 29.66 (CH3), 29.77 (CH3), 72.59 (CH2), 103.47, 125.48,
128.54, 129.13, 129.56, 134.87, 138.37, 146.15, 151.63, 154.75,
155.54; EIMS m/z 314 (MÅ+); HRMS calcd 314.13789, found
314.13983.
addition of 200 lL NaOH (2 M). Distilled water (1200 lL) was added
to each reaction before it was centrifuged for 10 min at 16,000g. The
concentrations of the MAO generated 4-hydroxyquinoline in the
reactions were determined by measuring the fluorescence of the
supernatant at an excitation wavelength of 310 nm and an emission
wavelength of 400 nm.19 Quantitative estimations of 4-hydroxy-
quinoline were made by means of a linear calibration curve ranging
from 0.188 to 6.25
final volume of 500
7.4) and contained 4% DMSO. To each standard was added 200
NaOH (2 M) and 1200 L distilled water. IC50 values were deter-
mined by plotting the initial rate of oxidation versus the logarithm
of the inhibitor concentration to obtain a sigmoidal dose–response
curve. This kinetic data were fitted to the one site competition model
incorporated into the Prism software package and the IC50 values
l
l
M. Each calibration standard was prepared to a
L in potassium phosphate buffer (100 mM, pH
4.3.7. 8-(3-Methoxybenzyloxy)caffeine (5g)
The title compound was prepared from 8-chlorocaffeine (7) and
3-methoxybenzyl alcohol in a yield of 30%: mp 144 °C; 1H NMR
(CDCl3) d 3.42 (s, 3H), 3.56 (s, 3H), 3.73 (s, 3H), 3.85 (s, 3H), 5.51
(s, 2H), 6.92 –7.35 (m, 4H); 13C NMR (CDCl3) d 29.85 (CH3), 29.75
(CH3), 27.72 (CH3), 55.27 (CH3), 72.38 (CH2), 103.56, 114.14,
114.07, 120.56, 129.79, 136.48, 146.16, 151.68, 154.80, 155.49,
lL
l

B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
1027
were determinedin duplicateandare expressedas mean SD. The Ki
values were calculated from the experimental IC50 values according
to the equation by Cheng and Prusoff: Ki = IC50/(1 + [S]/Km). For hu-
enzyme concentration in the incubations was 0.0075 mg/mL while
the final concentration of MAO-B was 0.015 mg/mL. All the incuba-
tions were prepared in potassium phosphate buffer (100 mM, pH
7.4, made isotonic with KCl) and were carried out for 20 min at
man MAO-B, [S] = 30
l
M and Km (kynuramine) = 22.7
lM while
[S] = 45 M and Km (kynuramine) = 16.1
l
l
M for human MAO-A.17
37 °C. The reactions were terminated with 200
lL NaOH (2 M)
and 1200 L distilled water was added. The rates of the MAO cat-
l
4.6. Lineweaver–Burk plots
alyzed formation of 4-hydroxyquinoline were measured as de-
scribed above. The kinetic data (initial rates as a function of
substrate concentration) were fitted to a Michaelis–Menten equa-
tion using the one site binding model incorporated into the Prism
software package. All measurements were carried out in duplicate
and the Km values are expressed as mean SEM.
For selected inhibitors, Lineweaver–Burk plots were con-
structed in order to determine the modes of inhibition. For this
purpose the initial rates of oxidation at four different substrate
concentrations in the absence and presence of three different con-
centrations of the inhibitors were used to construct Lineweaver–
Burke plots. Where baboon liver mitochondrial MAO-B was used
4.9. QSAR studies
as enzyme source, compound 5e (0.0325–0.130
as inhibitor and MMTP (30–120 M) served as substrate. Where
recombinant human MAO-A or -B was used as enzyme source,
compound 5e at concentrations of 0.931–3.722 M and 0.0325–
0.130 M, respectively, was selected as inhibitor and kynuramine
(15–90 M) served as substrate. All enzymatic reactions and mea-
surements were carried out as described above. Linear regression
analysis was performed using the SigmaPlot software package (Sy-
stat Software Inc.).
lM) was selected
l
The values of the substituent descriptors rm, F, p, Es, Vw were
obtained from standard compilations.22,23 Stepwise multiple linear
regression analysis of the inhibition potencies (log IC50 values) as a
function of the substituent descriptor values was carried out with
the Statistica software package (StatSoft Inc.). In order to estimate
the significance of the regression equations, the F statistic was em-
ployed. An F value higher than the critical F value (Fmax) was judged
to be significant. The Fmax value for 95% significance for models
constructed from seven log IC50 values (Tables 3 and 4) and which
l
l
l
4.7. Time-dependant inhibition studies
contains one descriptor (out of a possible five: rm, F, p, Es, Vw) was
calculated to be 25.32, while the Fmax value for models containing
Time-dependant inhibition studies were carried out in order to
determine whether a selected inhibitor, 5b, acts as a reversible
inhibitor or as a time-dependant inactivator of baboon liver
MAO-B, human MAO-A and human MAO-B. The respective MAO
preparations were preincubated for periods of 0, 15, 30, 60 min
two descriptors was calculated to be 30.18.24
4.10. Molecular docking studies
The molecular docking studies were carried out in the Windows
based Discovery Studio 1.7 molecular modeling software.27 All the
inhibitors (5a–g) were constructed and the hydrogen atoms were
added in Discovery Studio at pH 7. The geometries are optimized
by Discovery Studio using a fast Dreiding-like forcefield (1000 iter-
ations). Atom potential types and partial charges were subse-
quently automatically assigned with the Momany and Rone
CHARMm forcefield and the inhibitors were prepared for docking
with the Prepare Ligands protocol. The crystallographic structures
of MAO-A co-crystallized with harmine (PDB code: 2Z5X)25 and
MAO-B co-crystallized with safinamide (PDB code: 2V5Z)14 were
retrieved from the Brookhaven Protein Data Bank (www.rcsb.org/
pdb). Following the addition of hydrogen atoms according to the
appropriate protonation states of the ionizable amino acids at pH
7, and the correction of the valences of the FAD co-factors (oxidized
state) and co-crystallized ligands, the receptor models were auto-
matically typed with the Momany and Rone CHARMm forcefield
and subjected to a three step energy minimization cascade while
the protein backbone was constrained. Minimisation of the recep-
tor models were carried out since the X-ray crystal structures may
contain residual energetic tensions as a result of the crystallisation
process. The first step was a steepest descent minimization with
the termination criteria set to a maximum of 2500 steps or a min-
imum value of 0.1 for the root mean square of the energy gradient.
The second step was conjugate gradient minimization with the
same termination criteria. The third step was an adopted basis
Newton–Rapheson minimization with the termination criteria set
to a maximum of 5000 steps or a minimum value for the root mean
square of the energy gradient of 0.01. For this minimization cas-
cade the implicit generalized Born solvation model with simple
switching was used with the dielectric constant set to 4. The co-
crystallized ligands, crystal waters and the backbone constraints
were removed and the binding site was identified by a flood-filling
algorithm. X-ray crystal structures of MAO-B have shown that only
three active site water molecules are conserved, all in the vicinity
of the FAD co-factor.14 Preliminary docking studies in our labora-
at 37 °C with 5b at concentrations of 0.28
lM, 1.31 lM and
0.23 M for baboon liver MAO-B, human MAO-A and human
l
MAO-B, respectively.21,28 For this purpose the concentrations of
the enzyme preparations were 0.3 mg/mL baboon liver MAO-B,
0.015 mg/mL human MAO-A and 0.03 mg/mL human MAO-B. The
incubations were carried out in sodium phosphate buffer
(100 mM, pH 7.4) for the studies with baboon liver MAO-B and
in potassium phosphate buffer (100 mM, pH 7.4, made isotonic
with KCl) for studies with the recombinant human enzymes. A final
concentration of 50
lM MMTP for baboon liver MAO-B, 45
lM
kynuramine for human MAO-A and 30
lM kynuramine for human
MAO-B were then incubated with the preincubated enzyme prep-
arations at 37 °C for 15 min. The final volumes of these incubations
were 500
l
L and the final concentrations of 5b were 0.14
lM,
0.65 M and 0.115
l
lM for baboon liver MAO-B, human MAO-A
and human MAO-B, respectively. These concentrations of the
inhibitor are approximately equal to the IC50 values for the inhibi-
tion of the respective enzyme preparations by 5b. The final concen-
trations of the enzyme preparations were 0.15 mg/mL baboon liver
MAO-B, 0.0075 mg/mL human MAO-A and 0.015 mg/mL human
MAO-B. The reactions with baboon liver MAO-B were terminated
with 10
combinant human enzymes were terminated with 200
(2 M). A volume of 1200 L distilled water was added to the incuba-
lL perchloric acid (70%) while the reactions with the re-
lL NaOH
l
tions containing the recombinant human MAO preparations. The
rates of formation of MMDP+ and 4-hydroxyquinoline were mea-
sured as described above. All measurements were carried out in trip-
licate and are expressed as mean standarderror of themean (SEM).
4.8. Km Determination of kynuramine for recombinant human
MAO-A and -B
The steady-state oxidation rates of kynuramine by recombinant
human MAO-A and -B were measured at six different substrate
concentrations ranging from 2.5 lM to 80 lM. For MAO-A the final

1028
B. Strydom et al. / Bioorg. Med. Chem. 18 (2010) 1018–1028
9. Fowler, J. S.; Volkow, N. D.; Wang, G. J.; Logan, J.; Pappas, N.; Shea, C.;
MacGregor, R. Neurobiol. Aging 1997, 18, 431.
10. Vlok, N.; Malan, S. F.; Castagnoli, N., Jr.; Bergh, J. J.; Petzer, J. P. Bioorg. Med.
Chem. 2006, 14, 3512.
11. Van den Berg, D.; Zoellner, K. R.; Ogunrombi, M. O.; Malan, S. F.;
Terre’Blanche, G.; Castagnoli, N., Jr.; Bergh, J. J.; Petzer, J. P. Bioorg. Med.
Chem. 2007, 15, 3692.
12. Pretorius, J.; Malan, S. F.; Castagnoli, N., Jr.; Bergh, J. J.; Petzer, J. P. Bioorg. Med.
Chem. 2008, 16, 8676.
13. Van der Walt, E. M.; Milczek, E. M.; Malan, S. F.; Edmondson, D. E.; Castagnoli,
N., Jr.; Bergh, J. J.; Petzer, J. P. Bioorg. Med. Chem. Lett. 2009, 19, 2509.
14. Binda, C.; Wang, J.; Pisani, L.; Caccia, C.; Carotti, A.; Salvati, P.; Edmondson, D.
E.; Mattevi, A. J. Med. Chem. 2007, 50, 5848.
15. Huston, R. R.; Allen, W. F. J. Am. Chem. Soc. 1934, 56, 1356.
16. Fischer, E.; Reese, L. Justus Liebigs Ann. Chem. 1883, 221, 336.
17. Cheng, Y. C.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22, 3099.
18. Inoue, H.; Castagnoli, K.; Van der Schyff, C. J.; Mabic, S.; Igarashi, K.; Castagnoli,
N., Jr. J. Pharmacol. Exp. Ther. 1999, 291, 856.
19. Novaroli, L.; Reist, M.; Favre, E.; Carotti, A.; Catto, M.; Carrupt, P. A. Bioorg. Med.
Chem. 2005, 13, 6212.
20. Khalil, A. A.; Steyn, S.; Castagnoli, N., Jr. Chem. Res. Toxicol. 2000, 13, 31.
21. Manley-King, C. I.; Terre’Blanche, G.; Castagnoli, N., Jr.; Bergh, J. J.; Petzer, J. P.
Bioorg. Med. Chem. 2009, 17, 3104.
tory suggests that caffeinyl analogues do not form hydrogen bonds
with these waters, but is rather stabilized by hydrogen bond inter-
action between the carbonyl oxygen at C-6 of the caffeinyl ring and
the phenolic hydrogen of Tyr-435. Automated docking was subse-
quently carried out with the LigandFit application of Discovery Stu-
dio. This docking protocol employed total ligand flexibility
whereby the final ligand conformations are determined by the
Monte Carlo conformation search method set to a variable number
of trial runs. The docked ligands were further refined using in situ
ligand minimization with the Smart Minimizer algorithm. Unless
otherwise specified (see above), all the application modules within
Discovery Studio were set to their default values and 10 docking
solutions were allowed for each ligand.13,28 The illustrations were
generated in PyMOL.38
Acknowledgements
The NMR and MS spectra were recorded by André Joubert and
Johan Jordaan of the SASOL Centre for Chemistry, North-West
University. This work was supported by grants from the National
Research Foundation and the Medical Research Council, South
Africa.
22. Van de Waterbeemb, H.; Testa, B. In Advances in Drug Research; Testa, B., Ed.;
Academic Press: London, 1987; pp 85–225.
23. Hansch, C.; Leo, A. Exploring QSAR. Fundamentals and Applications in Chemistry
and Biology; American Chemical Society: Washington, DC, 1995. pp 1–124.
24. Livingstone, D. J.; Salt, D. W. J. Med. Chem. 2005, 48, 661.
25. Son, S.-Y.; Ma, J.; Kondou, Y.; Yoshimura, M.; Yamashita, E.; Tsukihara, T. Proc.
Natl. Acad. Sci. U.S.A. 2008, 105, 5739.
26. Hubálek, F.; Binda, C.; Khalil, A.; Li, M.; Mattevi, A.; Castagnoli, N., Jr.;
Edmondson, D. E. J. Biol. Chem. 2005, 280, 15761.
Supplementary data
27. Accelrys Discovery Studio 1.7, Accelrys Software Inc., San Diego, CA, USA. 2006,
http://www.accelrys.com.
28. Ogunrombi, M. O.; Malan, S. F.; Terre’Blanche, G.; Castagnoli, N., Jr.; Bergh, J. J.;
Petzer, J. P. Bioorg. Med. Chem. 2008, 16, 2463.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.12.064.
29. Novaroli, L.; Daina, A.; Favre, E.; Bravo, J.; Carotti, A.; Leonetti, F.; Catto, M.;
Carrupt, P. A.; Reist, M. J. Med. Chem. 2006, 49, 6264.
30. La Regina, G.; Silvestri, R.; Gatti, V.; Lavecchia, A.; Novellino, E.; Befani, O.;
Turini, P.; Agostinelli, E. Bioorg. Med. Chem. 2008, 16, 9729.
31. Przedborski, S.; Jackson-Lewis, V.; Naini, A. B.; Jakowec, M.; Petzinger, G.;
Miller, R.; Akram, M. J. Neurochem. 2001, 76, 1265.
32. Bissel, P.; Bigley, M. C.; Castagnoli, K.; Castagnoli, N., Jr. Bioorg. Med. Chem.
2002, 10, 3031.
33. Fischer, E. Justus Liebigs Ann. Chem. 1882, 215, 267.
References
1. Youdim, M. B. H.; Bakhle, Y. S. Br. J. Pharmacol. 2006, 147, S287.
2. Youdim, M. B. H.; Edmondson, D.; Tipton, K. F. Nat. Rev. Neurosci. 2006, 7, 295.
3. Fernandez, H. H.; Chen, J. J. Pharmacotherapy 2007, 27, 174S.
4. Collins, G. G. S.; Sandler, M.; Williams, E. D.; Youdim, M. B. H. Nature 1970, 225,
817.
5. Youdim, M. B. H.; Collins, G. G. S.; Sandler, M.; Bevan-Jones, A. B.; Pare, C. M.;
Nicholson, W. J. Nature 1972, 236, 225.
34. Sono, M.; Toyoda, N.; Shimizu, K.; Noda, E.; Shizuri, Y.; Tori, M. Chem. Pharm.
Bull. 1996, 44, 1141.
6. Di Monte, D. A.; DeLanney, L. E.; Irwin, I.; Royland, J. E.; Chan, P.; Jakowec, M.
W.; Langston, J. W. Brain Res. 1996, 738, 53.
35. Salach, J. I.; Weyler, W. Methods Enzymol. 1987, 142, 627.
36. Bradford, M. M. Anal. Biochem. 1976, 72, 248.
7. Finberg, J. P.; Wang, J.; Bankiewich, K.; Harvey-White, J.; Kopin, I. J.; Goldstein,
D. S. J. Neural Transm. Suppl. 1998, 52, 279.
37. Gnerre, C.; Catto, M.; Leonetti, F.; Weber, P.; Carrupt, P.-A.; Altomare, C.;
Carotti, A.; Testa, B. J. Med. Chem. 2000, 43, 4747.
8. Nicotra, A.; Pierucci, F.; Parvez, H.; Senatori, O. Neurotoxicology 2004, 25, 155.
38. DeLano, W. L. Palo Alto, CA, USA. 2002.