Synthesis, biological evaluation, and molecular modeling studies of a novel, peripherally selective inhibitor of catechol-O-methyltransferase

Article abstract of DOI:10.1021/jm040848o

A novel series of potent, peripherally selective, and long-acting inhibitors of catechol-O-methyltransferase (COMT) has been synthesized. The introduction and nature of heteroatom-containing substituents to the side-chain of the nitrocatechol pharmacophor

Full text of DOI:10.1021/jm040848o

J. Med. Chem. 2004, 47, 6207-6217
6207
Synthesis, Biological Evaluation, and Molecular Modeling Studies of a Novel,
Peripherally Selective Inhibitor of Catechol-O-methyltransferase
David A. Learmonth,† P. Nuno Palma,‡ Maria A. Vieira-Coelho,‡,§ and Patr´ıcio Soares-da-Silva*,‡
Laboratory of Chemistry and Laboratory of Pharmacology, Department of Research & Development, BIAL,
4745-457 S. Mamede do Coronado, Portugal
Received May 25, 2004
A novel series of potent, peripherally selective, and long-acting inhibitors of catechol-O-
methyltransferase (COMT) has been synthesized. The introduction and nature of heteroatom-
containing substituents to the side-chain of the nitrocatechol pharmacophore was found to have
a profound effect on both peripheral selectivity and duration of COMT inhibition in the mouse.
This approach led to the discovery of 1-(3,4-dihydroxy-5-nitrophenyl)-3-[4-[3-(trifluoromethyl)-
phenyl]-1-piperazinyl]-1-propanone hydrochloride 35 (BIA 3-335), which was found to possess
a superior inhibitory profile in vivo over both the nonselective inhibitor tolcapone 1 and the
peripherally selective but short-acting entacapone 2. In this model, 35 retained 75% inhibition
of peripheral COMT at 6 h after oral administration, yet significantly, only a minor reduction
of central (cerebral) COMT activity was observed. Molecular modeling techniques were applied
to review the analysis of the ternary enzyme-inhibitor complex previously determined by X-ray
crystallography and to provide a deeper understanding of the structure-activity relationships
within this novel series. Furthermore, a computational approach was applied in an effort to
elucidate the particular structural factors relevant to the poor blood-brain permeability of
35. In conclusion, the improved biological properties herein reported reveal 35 as a candidate
for clinical studies as an adjunct to L-DOPA therapy for Parkinson’s disease.
Introduction
tabolite 3-OMD has a comparatively long in vivo elimi-
nation half-life and tends to accumulate in tissues.8
Since this metabolite actually competes with L-DOPA
for transport across the blood-brain barrier (BBB), it
would be anticipated that a reduction in circulating
3-OMD levels should permit improved passage of
L-DOPA across the BBB into the brain.
Catechol-O-methyltransferase (COMT)1 is a magne-
sium-dependent enzyme found in both the periphery
and central nervous system (CNS) that catalyses the
transfer of a methyl group from its cofactor S-adenosyl-
L-methionine (SAM) to one hydroxyl group of a substrate
bearing the catechol motif via an SN2-type reaction,
giving rise to an O-methylated catechol and S-adeno-
sylhomocysteine. COMT is known to play an important
role in the inactivation of endogenous neuroactive
catecholamines2 such as dopamine and in the detoxifi-
cation of several xenobiotic catechols.3,4 In recent years,
our group and others have been greatly interested in
the development of tight-binding, reversible inhibitors
of COMT as adjuncts5 to L-3,4-dihydroxyphenylalanine
(L-DOPA) therapy for Parkinson’s disease6 (PD, a neu-
rodegenerative disorder characterized by a deficiency
of cerebral dopamine). This attraction has been based
on the premise that COMT inhibition could significantly
reduce the extent of O-methylation of L-DOPA to 3-O-
methyl-L-DOPA (3-OMD), which is the principal meta-
bolic pathway for L-DOPA in peripheral tissues when
L-DOPA is coadministered with a peripheral decarboxy-
lase inhibitor such as carbidopa.7 Inhibition of COMT
could thereby increase L-DOPA bioavailability, provid-
ing an improvement in the percentage of the adminis-
tered L-DOPA dose penetrating into the brain. Further
still, it has been observed that the O-methylated me-
So-called “second-generation” COMT inhibitors9,10
incorporate the nitrocatechol subunit and are typified
by tolcapone 111 and entacapone 212 (Figure 1). Although
both structures share the same pharmacophore, the
pharmacological profile of 1 differs from that of 2 in that
it is characterized as an indiscriminate inhibitor of both
central and peripheral COMT, whereas 2 is essentially
a peripheral inhibitor. The issue of selectivity in COMT
inhibition has been the subject of some debate, although
we, and others, have hypothesized that central inhibi-
tion may be of less consequence should the principal role
of COMT inhibition be to prevent extensive metabolism
of L-DOPA through O-methylation in the periphery.
While both 1 and 2 recently reached the marketplace,
1 was subsequently withdrawn due to hepatotoxicity
concerns,13 leaving 2 as the sole COMT inhibitor pres-
ently available in the clinic for use as an adjunct to
L-DOPA treatment of PD patients. Our concerns related
to the rather short in vivo half-life of 2, which implies
a multiple and elevated daily dosage regime, led us to
investigate the possibility of designing a potent, revers-
ible, peripherally selective and long-acting COMT in-
hibitor and culminated in the discovery of BIA 3-202
314 (Figure 1), which is currently under clinical evalu-
ation as an adjunct to L-DOPA treatment of PD.15,16 This
compound also bears the nitrocatechol motif, yet we
were intrigued to find that homologation and structural
* Corresponding author. Tel: 351-22-9866100. Fax: 351-22-9866192.
E-mail: psoares.silva@bial.com.
† Laboratory of Chemistry.
‡ Laboratory of Pharmacology.
§ Present Address: Institute of Pharmacology and Therapeutics,
Faculty of Medicine, University of Oporto, 4200-319 Oporto, Portugal.
10.1021/jm040848o CCC: $27.50 © 2004 American Chemical Society
Published on Web 11/02/2004

6208 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25
Learmonth et al.
Scheme 1. Synthesis of Oxygen Containing Compounds
I (Method A), Sulfur-Containing Compounds II
(Methods B and C), and Nitrogen-Containing
Compounds III (Methods D and E).
Figure 1. Chemical structures of tolcapone (1), entacapone
(2), BIA 3-202 (3), and building blocks 4-7.
variation of the hydrocarbon side-chain substituent of
3 resulted in highly significant effects on biological
parameters, such as potency, duration, and selectivity
of COMT inhibition. This prompted us to undertake the
present study, principally investigating the effects of
introducing homologous heteroatom-containing side-
chain substitution (oxygen, nitrogen, and sulfur) to C-1
of the nitrocatechol pharmacophore, with the aim of
ascertaining whether structural manipulations of the
side chain could provide more structurally diverse
COMT inhibitors possessing greater peripheral selectiv-
ity than 1 and increased duration of action compared
to 2.
From the oxygen series, aryl compounds were cer-
tainly more active than 2 but less so than 1. Incorpora-
tion of electron-donating (9, methoxyl) or electron-
withdrawing groups (11, fluorine) made little difference.
Interestingly, increasing the size of the aromatic ring
had a noticeably positive effect on activity, and the more
lipophilic analogues derived from either 1- or 2-naphthol
(12-14) generally exhibited increased inhibition com-
pared to their phenyl counterparts. In a similar vein to
the oxygen analogues, sulfur-containing compounds
showed slightly improved in vitro inhibition over 2, but
this activity was generally unaffected by homologation,
as typified by 15 and 16. Rather more interesting
observations were made, however, in the nitrogen series.
Initial results from compounds derived from small
aliphatic secondary amines 18-21 were rather disap-
pointing from the point of view of the relatively low in
vitro inhibition but did serve to indicate that compounds
with longer chains (n ) 2) may be endowed with greater
inherent activities than their lower homologues (n ) 1).
Indeed, this early hypothesis was in fact borne out, and
similar general trends of increased in vitro inhibition
were associated with the higher homologues of other
classes of nitrogen-containing compounds, such as, for
example, those derived from cyclic amines such as
substituted piperidines 22 and 23, thiomorpholines 24
and 25, and substituted piperazines 26-42. This latter
class in particular displayed greatest early promise.
Whereas the simple N-alkylpiperazines such as 26-29
were again relatively uninteresting, replacement of the
substituent on the piperazine nitrogen with phenyl
(30-39) or benzoyl groups (40-41) provided significant
improvement, with the higher homologues (n ) 2) now
very significantly more active than 2 and essentially
equipotent to 1. Within this N-arylpiperazine series,
incorporation of small alkyl substituents (38) or electron-
withdrawing groups, such as halogen (32-33, 36-37)
or trifluoromethyl (34-35), were well-tolerated, but
electron-donating groups such as methoxy (39) had quite
the opposite effect.
Results and Discussion
The new COMT inhibitors herein presented were
prepared using the readily accessible and suitably
functionalized nitrocatechol building blocks 4-7 (Figure
1) as substrates for nucleophilic substitution, Michael
addition, or Mannich reactions. Target compounds of
general formula I with oxygen at the â-position relative
to the carbonyl group of the side chain were easily
prepared in generally moderate yields by reaction of the
halo ketone 4 with appropriately substituted phenols
or naphthols in the presence of weak base (K2CO3) in
warm DMF (method A, Scheme 1). Derivatives of
general formula II (n ) 1) containing sulfur were
synthesized by reaction either of 4 with substituted
thiophenols under similar conditions as above (method
B, Scheme 1), and the longer homologues (n ) 2) were
prepared smoothly via Michael addition to the R,â-
unsaturated ketone 6 (method C, Scheme 1) again in
reasonably good yields. Finally, reaction of 4 and 6 with
selected secondary amines provided homologous com-
pounds containing nitrogen in the side chain of general
formula III (methods D and E, respectively, Scheme 1)
in moderate to good yields, which were more easily
isolated and purified as their hydrochloride salts. The
Mannich reaction of the acetophenone 7 was exemplary
in this respect (method E, Scheme 1), as the target
compounds tended to precipitate toward the end of the
reaction with excellent purity, thereby greatly facilitat-
ing isolation of the desired products by simple filtration.
Initially, high-throughput in vitro screening (human
neuroblastoma SK-N-SH cells)17 was used to test the
ability of the new compounds to inhibit COMT, the
results of which allowed construction of basic structure-
activity relationships. Selected in vitro results of syn-
thesized compounds are shown in Table 1 and are
compared to 1 and 2.

A Novel Inhibitor of Catechol-O-methyltransferase
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25 6209
Table 1. Percent Inhibition of COMT Activity in SK-N-SH Cells by Compounds 1, 2, and 8-41 (100 nM)a
a Results are means ( SEM of four experiments per group. *Represents point of substituent attachment.
Having identified from this series two classes of
nitrocatecholic compounds with novel side-chain sub-
stitution possessing potent in vitro COMT inhibitory
activity, namely, the aryloxy and the N-arylpiperazine
series, subsequent experiments were designed to evalu-
ate both the duration of COMT inhibition and access to
the brain of selected compounds from Table 1 and to
compare them with the standards 1 and 2. Test com-
pounds were administered to overnight fasted mice, and
thereafter, the animals were killed by decapitation at
either one or 6 h after administration, whereupon the
livers and brains were removed and used to determine
COMT activity as previously described.18 The time-
course inhibition profiles are summarized in Table 2.
Of the oxygenated compounds, 8 and particularly 9
displayed long-acting inhibition of peripheral COMT at
6 h but relatively lower selectivity than 2 at the shorter
time point. However, 12 was significantly longer acting
than 8, 9, and 2 and more selective at 1 h over
regioisomer 14 and 2 itself. The sulfur compound 15
displayed an inhibition profile very similar to that of 2.
Of the nitrogen-containing compounds, the lower di-
alkylamines 20 and 21 again displayed relatively low
activity, even at 1 h, whereas the N-benzoylpiperazine
40 displayed very high peripheral selectivity but very
short duration of COMT inhibition. Of the N-arylpip-
erazine series, compound 35, the 3′-trifluoromethylphe-
nyl-substituted piperazine (Figure 2), clearly displayed
the longest duration of COMT inhibition, retaining
almost 75% inhibition, even at 6 h after oral adminis-

6210 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25
Learmonth et al.
Table 2. Percent Inhibition of COMT Activity of Selected
Compounds from Table 1 in Homogenates of Mouse Brain and
Liver at 1 and 6 h after Administration (all at 30 mg/kg) by
Gastric Tubea
the protein-ligand complex was reexamined. One of the
major difficulties in evaluating and scoring intermo-
lecular interactions is the adequate treatment of en-
tropic terms associated with the desolvation processes
that occur upon complex formation. Such desolvation
terms are especially important for interactions that are
dominated by hydrophobic contacts, as in the present
case of COMT and the long side chain of 35.
COMT activity at 1 h
liver brain
COMT activity at 6 h
liver brain
%
%
%
%
no. inhibn SEM inhibn SEM inhibn SEM inhibn SEM
1
2
99
76
78
71
83
86
70
50
62
70
50
82
46
84
1
4
4
1
3
2
2
2
3
4
3
7
7
4
99
23
39
41
17
37
20
22
26
7
1
10
4
94
26
50
65
61
69
23
33
43
42
53
74
28
18
1
8
86
15
-4
7
-19
8
-15
-11
-26
-7
0
8
8
The HINT algorithm utilizes a fragmentation process
to assign atom-based hydrophatic constants, based on
experimentally determined solvent partition log P val-
ues between water and octanol. The atom-atom inter-
action function is based on the relative hydrophatic
constants, and the strength of an intermolecular as-
sociation is computed over the sum of atom-atom
scores. Since the parameter log P is a thermodynamic
constant associated with the free energy of solvent
transfer, both enthalpic (van der Waals, hydrogen-
bonding, Coulombic interactions) and entropic (solva-
tion, “hydrophobic interactions”) terms are treated
implicitly by this empirically based function. A correla-
tion between HINT hydrophatic scores and the free
energy of binding of 53 protein-ligand complexes has
been recently suggested.21
8
9
6
9
6
6
9
12
14
15
20
21
25
34
35
36
40
5
5
10
16
12
19
9
14
3
8
5
13
11
7
6
15
9
12
8
3
10
11
18
4
5
9
24
14
6
3
4
14
4
-22
0
13
5
a Results are means ( SEM of four experiments per group
Of relevance to the intermolecular interactions are
the actual protonation states of the nitrogen atoms of
the piperazine ring of 35. Since experimentally deter-
mined ionization constants for 35 were unavailable, the
piperazinyl nitrogen atoms were instead computed
using the ACD/pKa fragmental method (Advanced Chem-
istry Development, Inc) and found to have pKa values
below 6.0. One piperazinyl nitrogen atom is linked to a
trifluoromethyl-substituted phenyl ring and is therefore
expected to be neutral. Although the second nitrogen is
indirectly connected to the carbonyl and nitrocatechol
moieties, these groups appear to have an inductive effect
through the aliphatic spacer, thereby lowering the pKa
of this nitrogen atom. Therefore, the piperazine was
considered to be unprotonated at physiological pH and
modeled as such in subsequent studies.
Figure 2. Chemical structure of 35 (BIA 3-335).
tration, compared to only 25% for 2. Additionally, 35
exhibited far superior selectivity than 1 at both time
points for peripheral over central COMT inhibition.
In an effort to rationalize these results obtained from
both the in vitro and in vivo studies, we resorted to
application of a combination of molecular modeling and
computational methods with the 3-fold intention of
analyzing and reviewing the structure of the COMT in
complex with 35, explaining the structure-activity
relationships observed within this novel series of inhibi-
tors, and justifying the poor blood-brain permeability
of 35.
The hydrophatic interactions between 35 and COMT
within the complex were calculated using the HINT
scoring function, as detailed in the methods. Figure 3a
shows isocontour surfaces representing favorable inter-
molecular hydrophobic interactions. Polar interactions
are mostly confined to the catechol-binding region and
are not shown. Apart from the strong electrostatic and
hydrogen-bonding interactions of the catechol moiety,
the so-called “gatekeepers” Pro174, Trp38, and Leu198
are responsible for most of the hydrophobic stabilizing
interactions. Also Trp143 and Met40 establish impor-
tant van der Waals contacts with the nitrocatechol
group. Most notably, the terminal trifluoromethylphenyl
group points away from the protein surface (atoms
colored cyan) and makes only weak van der Waals
contacts (Figure 3a, green surface) with Trp38 and
Met201. Such a bound conformation is somewhat un-
expected in a solvated complex, since the hydrophobic
trifluoromethylphenyl group would remain extensively
solvated. Interestingly, however, if crystallographic
neighboring molecules are generated using the sym-
metry operations of the crystallographic space group
P3221, it can be seen that the piperazine and trifluo-
romethylphenyl moieties of the inhibitor, as well as few
The structure of the ternary complex between recom-
binant rat liver COMT, the cosubstrate SAM, and the
inhibitor 35 has been determined by X-ray crystal-
lography,19,20 and a discussion of the interactions at the
atomic level has been reported20 (indeed, this is the first
3D structure determination of COMT complexed with
a potentially therapeutically useful inhibitor). Briefly,
the nitrocatechol is responsible for “anchoring” the
inhibitor to the enzyme, via hydrogen-bonding to Glu199
and Lys144 and through coordination of the magnesium
ion in the catalytic site. In contrast, the propanone,
piperazine, and trifluoromethylphenyl moieties of 35
extend out from the narrow catalytic pocket, flanked by
Trp38, Pro174, and Leu198, and form hydrophobic
interactions along a shallow groove at the surface of the
protein. Due to the lack of hydrogen bonds or tight steric
restrictions, the bound conformation of the inhibitor’s
side chain groups appears to be loosely determined, as
indicated by the corresponding high average thermal
factors observed (Bavg of 45.1 and 55.7 A2 for the piper-
azine and trifluoromethylphenyl moieties, respectively).
To quantify and characterize the nature of the inter-
molecular interactions responsible for its high stability,

A Novel Inhibitor of Catechol-O-methyltransferase
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25 6211
Figure 3. HINT interaction maps showing favorable hydrophobic contacts between COMT and 35. (A) Structure of the
crystallographic complex 1H1D showing the inhibitor (color scheme according to element type: carbon, white; hydrogen, cyan;
oxygen, red; nitrogen, blue; fluorine, green) complexed with COMT (cyan). A symmetry-related, closely interacting molecular
complex that is present in the crystal framework is shown in red. Green and yellow isocontour surfaces define favorable hydrophobic
interactions between 35 and the single protein chain (intracomplex) or with the nearest complex structure within the crystal,
respectively. (B) Structure of the lowest energy binding mode of the inhibitor in complex with COMT, obtained by docking.
Interaction isocontour surfaces are plotted at the same energy level in both panels.
vicinal residues, form close van der Waals contacts with
a symmetrically positioned complex molecule (Figure 3a,
red atoms). In particular, the trifluoromethylphenyl
group interacts closely (Figure 3a, yellow surface) with
the side chains of a cluster of hydrophobic residues of
the second COMT monomer, comprised by Val173,
Leu198, Met201, and Val203. Such interactions would
be absent under physiological conditions, where COMT
functions exclusively as a monomeric protein. The above
observation raises the hypothesis that the terminal side
chain groups (piperazine and trifluoromethylphenyl) of
35 are stabilized in an alternative conformation, induced
by a vicinal complex structure present in the crystal
form and that the observed conformation may not be
representative of the actual complex formed in solution.
To further assess this hypothesis, we first explored
alternative binding conformations of 35 with the struc-
ture of one monomer of COMT using molecular docking
simulations. Docking was performed with the program
GOLD, which uses a genetic algorithm to optimize the
conformation of the bound inhibitor. Fifty independent
docking runs, followed by local energy minimization,
were performed (starting from random configurations)
to ensure a wide exploration of the configurational
landscape of interaction. During docking simulations,
the coordination bonds between the Mg2+ ion and the
catechol hydroxyl groups are implicitly treated as
special hydrogen bonds, where the metal behaves as a
hydrogen-bond donor. This enables adequate binding
geometries to be found without imposing any additional
constraints, although the energetics of the coordination
bonds are likely to be Hill-estimated. However, the
relative coordinates of the catechol moieties are virtually
identical in all of the docked structures, so the errors
are expected to be fairly constant, allowing direct
comparisons to be made.
energetics of the crystallographic complex, with and
without considering the presence of the second protein
monomer. It must be pointed out that HINT potential
functions also lack parameters for metal coordination
complexes, and therefore, the (important) contribution
of the coordination bonds to the overall interaction
energetics of the complexes is currently ignored. How-
ever, the same reasoning indicated above may apply and
comparison between different poses should still be valid.
Figure 4a plots the hydrophatic interaction scores of
inhibitor 35 in each of 50 putative binding modes
against the corresponding root-mean-square (rms) de-
viations from the crystallographic structure. Visual
inspection of the independently docked poses reveals
three main families of conformationally related struc-
tures, with rms deviations from the experimental struc-
ture ranging from 1.0 to 3.7 Å. In particular, the position
of the nitrocatechol and the propanone carbonyl groups
were accurately predicted, but the conformation of the
two side-chain rings of the inhibitor molecule could not
be equally well reproduced in any of the docked solu-
tions. Structurally, they differ from that of the crystal-
lographic complex by a dihedral torsion of the propanone
side chain as well as different rotations of the terminal
trifluoromethylphenyl group (Figure 4b). However, such
conformational adjustments predicted by docking enable
a better stacking of the piperazine motif and, addition-
ally, in the case of structure III, of the trifluorometh-
ylphenyl moiety over Trp38.
In terms of interaction energetics, any of the three
predicted binding modes could represent a more stable
complex than that observed in the crystallographic
complex between 35 and one monomer of COMT (HINT
score ×1 ) -1641.0). The additional stabilizing energy
is mostly attributed to an optimization of the hydro-
phobic contacts between the two molecules. In particu-
lar, the lowest energy binding mode (structure cluster
III in Figure 4) enables the stacking of the trifluorom-
Finally, the hydrophatic interactions were computed
for all alternative poses and compared to the binding

6212 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25
Learmonth et al.
Figure 4. Docking results between COMT and compound 35. (A) Hydrophatic interaction scores calculated for each of the 50
putative binding modes versus the corresponding rms deviations from the crystallographic structure of the complex. ×1, interaction
score between 35 and one single chain of COMT in the crystal structure; ×2, interaction score calculated including the coordinates
of the nearest symmetry-related molecular complex in the crystal (see the text). (B) Three alternative bound conformations of 35
(red) superimposed on the atomic coordinates (black) of the crystallographic complex (protein not shown).
ethylphenyl ring onto Trp38, consequently excluding
water molecules from hydrophobic surfaces. Figure 3b
shows the favorable hydrophobic interaction maps
between COMT (single chain) and 35 in the lowest
energy conformation predicted by docking. Comparison
of this interaction map to that of the crystallographic
complex (Figure 3a) clearly reveals additional hydro-
phobic contacts between 35 and Trp38 that contribute
to a more negative (more favorable) hydrophatic inter-
action score (HINT score III ) -1761.0), while the
interactions made by the nitrocatechol or the propanone
moieties are essentially unchanged.
Most notably, under the particular crystallization
conditions, the trifluoromethylphenyl group is not in fact
exposed to the solvent, as suggested by the isolated
complex structure. Instead, it forms extensive van der
Waals contacts with a cluster of hydrophobic residues
of a crystallographic neighboring complex, as discussed
above. The side chains of Val173, Leu198, Met201,
Val203 and a second inhibitor molecule belonging to this
symmetrically positioned complex form a hydrophobic
pocket that shields the nonpolar trifluoromethylphenyl
group from the solvent. Such intercomplex molecular
interactions, depicted in Figure 3a (yellow surface),
contribute largely to the stabilization of the crystal-
lographic conformation of the inhibitor. If the coordi-
nates of the neighboring molecule are also included in
the calculations, the hydrophatic interaction energy of
35 becomes significantly more negative (HINT score ×2
) -1873.0) than that estimated for the isolated complex
or for any of the docked models (Figure 4a).
actual complex in solution is herein proposed (Figure
3b), based on docking simulations and hydrophatic
interaction calculations. In this model, the piperazine
and trifluoromethylphenyl moieties of 35 adopt a stack-
ing conformation over Trp38, which enables better
solvent exclusion from hydrophobic regions and conse-
quently an additional intermolecular stabilizing energy.
At the opposite end, the nitrocatechol fits into the
catalytic pocket, exactly in the same manner as for the
crystallographic complexes of 3520 and 3,5-dinitrocat-
echol,22 forming coordination bonds to the magnesium
ion and two hydrogen bonds to Lys144 and Glu199.
Extensive van der Waals contacts with Trp38, Met40,
Trp143, Pro174, and Leu198 also contribute signifi-
cantly to the stabilization of the complex.
With regard to analysis of structure-activity within
this novel series, observed in vitro inhibition showed a
consistent dependency on side chain homologation, with
longer chains (n ) 2) endowing greater inherent activi-
ties than their lower homologues (n ) 1). To understand
the nature of this effect, the conformational spaces
accessible to each pair of homologues (n ) 1 or 2), within
the active site of COMT, were compared by flexible
docking simulations again using GOLD. Results are
exemplified with the case of homologues containing the
small aliphatic amine side chain, 18 and 19, which are
the simplest nitrogen-containing compounds of the
series but which also show the highest dependency on
homologation. Nevertheless, the same general conclu-
sions can be extended to the other series of homologues
tested.
In summary, the present results suggest that the
bound conformation of 35, as observed in the crystal
structure of the complex, may indeed represent a stable
configuration within the framework of the crystal.
However, it is proposed that this particular binding
conformation could be induced by the crystallization
conditions due to nonphysiological interactions between
symmetry-related molecules in the crystal and, conse-
quently, may not be representative of the actual complex
in solution. An alternative, hypothetical structure of the
As already discussed, the nitrocatechol pharmaco-
phore, as well as the propanone carbonyl function, fits
inside the narrow catalytic site of COMT, which is
flanked by Trp38, Pro174, and Leu198. On the other
hand, the rest of the side chain extends out of the
catalytic pocket and interacts loosely with the protein
surface. In the case of lower homologues (n ) 1), the
shorter (dimethylamino)ethanone chain is sterically
constrained by Trp38 and Pro174 residues (Figure 5a),
thereby conferring only limited conformational flex-

A Novel Inhibitor of Catechol-O-methyltransferase
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25 6213
and the larger hydrophobic side chain of the ligand.
Within the piperazine compounds, the same pattern of
higher activities observed for the higher homologues due
to optimization of binding configurations could again be
explained as discussed above.
Finally, we studied the blood-brain barrier perme-
ation of 35. The entry of a drug into the brain is a
complex process, which depends on multiple factors.
Binding to plasma proteins, passive diffusion across the
BBB, differential metabolism, and active efflux from the
CNS may all contribute to a drug’s partitioning from
blood into the brain. It is known from practice that
lipophilicity and H-bonding capacity influence to some
extent the passive diffusion of drugs across the BBB.
Molecular size, flexibility, polarity, and electrostatics are
other properties that are thought to influence mem-
brane partition, and since biological membranes are
anisotropic systems, the three-dimensional distribution
of such molecular determinants must also be considered
important.
We applied a computational approach to study the
structural factors determinant for the observed selectiv-
ity of 35 for peripheral COMT inhibition. A set of 72
molecular determinants relevant to the process of
membrane partitioning was computed with the software
VOLSURF as detailed in the methods and projected
onto a predictive model of BBB permeation. This model,
which was developed elsewhere,23 correlated the same
set of descriptors with the experimentally obtained
permeation of nearly 300 known drugs and was able to
correctly assign the BBB behavior for more than 90%
of the compounds. Compounds computed to have posi-
tive scores are predicted to partition into the membrane
(BBB+), whereas those with negative scores (BBB-)
will have reduced probability of crossing the BBB by
passive diffusion. Near-zero scores may indicate com-
pounds with borderline or dubious behavior. The aver-
age scores for nearly 300 drugs in the training and
validation sets are +0.5 for compounds showing BBB+
behavior and -0.6 for the BBB- drugs.
From the results herein obtained, 35 is predicted to
behave as a moderate BBB- compound (BBB score )
-0.7), which is in good agreement with the observed
selectivity for peripheral COMT inhibition. It must be
pointed out that the current model is based on a
discriminant PLS procedure that aims solely at clas-
sifying a given compound in either the BBB+ or BBB-
categories but does not necessarily permit prediction of
the diffusion rate across the membrane system.
It is informative to analyze how the partitioning
behavior is influenced by the individual molecular
determinants. Figure 6a plots the variable weights for
the first and most significant component of the BBB
model. The vertical bars represent the relative contribu-
tions of each individual molecular descriptor to the
global BBB score. A detailed discussion of such contri-
butions is presented elsewhere,23 but a brief summary
of the most relevant characteristics is given below to
ease the interpretation of our results. Positive bars
indicate descriptors whose values are directly propor-
tional to membrane permeability, while descriptors with
negative bars show an inverse correlation with perme-
ation. It can be seen from this plot that BBB perme-
ability decreases with an increase of the compound’s
Figure 5. Illustrative structures of docked configurations of
homologues 18 (panel A) and 19 (panel B). The partial shape
of the catalytic pocket and surrounding protein surface is also
shown.
ibility to the dimethylamine function. As a result, both
terminal methyl groups are forced into the solvent,
making only suboptimal contacts with the protein
surface. Conversely, the homologation of the ethanone
chain by a single methylene group confers a much wider
configurational landscape to the side chain in the bound
state. With the longer propanone spacer extending out
of the narrow pocket entrance, the dimethylamine group
becomes free to rotate about the external C-C bond of
the propanone. Our docking results (not shown) indicate
that, due to such conformational topology, the dimeth-
ylamine group is able to stack on top of the hydrophobic
Trp38 side chain, therefore minimizing the solvent
exposure of the two methyl groups, as illustrated in
Figure 5b. This more favorable interaction may also
compensate for the increased loss of entropy upon
association. Similar behavior was observed for homolo-
gous cyclic amines such as the substituted piperidines
22 and 23, thiomorpholines 24 and 25, and the substi-
tuted piperazines 26-35, where the van der Waals
interactions of the ring systems can be systematically
optimized in the case of the higher homologues. The
results herein discussed suggest that steric constraints
caused by Trp38 and Pro174 prevent optimization of
hydrophatic interactions of the shorter homologues with
the protein surface, thereby explaining the increased
activity for the higher homologues.
It is also interesting to note that, within the series of
equivalent homologues containing a piperazine motif,
the in vitro data indicates that replacement of a small
alkyl substituent (26-29) on the piperazine nitrogen
with a phenyl or a nonpolar substituted phenyl group
(30-38) results in a significant increase in inhibition.
Given the predominantly hydrophobic nature of the
protein surface surrounding the catalytic site and in
light of the discussion above, it is herein suggested that
this increased potency results from more favorable and
extended hydrophatic interactions between the enzyme

6214 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25
Learmonth et al.
BBB. On the other hand, descriptors of hydrophobic
interactions (D) are directly correlated with BBB per-
meability, especially if evenly distributed along the
molecular surface, as indicated by the negative weights
associated with the hydrophobic integy moments (ID
descriptors).
Figure 6b shows the profile of molecular descriptors
computed for 35. The original values were scaled to unit
variance and centered on the average value of each
descriptor. Positive or negative values consequently
indicate molecular determinants with values above or
below the population’s average, respectively. The low
BBB permeability of 35 can be generally explained by
the combined effects of four major classes of molecular
determinants. The presence of two catechol hydroxyl
groups, along with the solvent accessible carbonyl
function and two piperazinyl nitrogen atoms, provides
an extended hydrophilic surface (W, Figure 6b,c), which
contributes to limiting membrane partitioning. The
detrimental contribution of such hydrophilic potentials
to BBB permeability is only slightly mitigated by the
fact that they are diffusely distributed along large areas
of the molecular surface, as indicated by the net nega-
tive capacity factors (CW). Moreover, hydrogen-bonding
potential, namely the high content of hydrogen-bond
acceptor functions, also contributes negatively to the
diffusion of 35 through the membrane system.
The anisotropy of distribution of hydrophilic regions
along the molecular axis (hydrophilic integy moments,
IW) generally correlates directly with BBB permeability.
However, in the case of 35, the hydrophilic regions
distribute evenly along most of the molecular axis, with
the exception of the terminal trifluoromethylphenyl
group. In particular, the presence of the hydrophilic
piperazinyl, which is “separated” from the polar nitro-
catechol moiety by the 1-propanone chain, contributes
to a reduction in the hydrophilic integy moments,
therefore limiting the ability to partition into the BBB.
Other molecular determinants of 35 that contribute to
poor permeability are the large molecular volume (V)
and surface (S), polarizability (P), and elongated shape
(G), which is also related to molecular flexibility. On the
other hand, the descriptors related to hydrophobic
regions are all above average, which correlates posi-
tively, though weakly, with a BBB+ behavior. The
inclusion of a polar group at the terminal (trifluorom-
ethyl)phenyl moiety of the molecule could have the effect
of decreasing the hydrophobicity as well as the hydro-
philic integy moments, potentially contributing to low-
ering the partitioning of the compound between aqueous
and membrane phases.
Figure 6. Volsurf descriptors of BBB model. (A) PLS weights
plot for the first and most significant component of the blood-
brain permeation model. Vertical bars represent the relative
contributions of each of the 72 individual molecular descriptors
to the global BBB score. Positive bars indicate descriptors
whose values are directly proportional to membrane perme-
ability while descriptors with negative bars show an inverse
correlation with permeation. Different shades help differen-
tiating groups of descriptors. V, volume; S, surface; G, globu-
larity; W, hydrophilic regions; IW, hydrophilic integy moments;
CW, hydrophilic capacity factors; HB, hydrogen-bonding; HL,
hydrophilic-lipophilic balance; D, hydrophobic regions; ID,
hydrophobic integy moments; A, amphiphilic moment; P,
polarizability; MW: molecular weight. (B) Profile of molecular
descriptors computed for 35. The original values were scaled
to unit variance and centered on the average value of each
descriptor. (C) Hydrophilic molecular interaction fields of 35
calculated with a water probe. The surfaces shown represent
regions contoured at -3 kcal/mol.
H-bonding (HB descriptors) potential, the extent of
hydrophilic regions (W descriptors), and the hydrophilic
capacity factors (CW descriptors). The three-dimensional
distribution, as well as the number, of polar or hydrogen-
bonding groups in the molecule is very important. The
negative weights associated with the capacity factors,
which relate to hydrophilic interactions per surface unit,
indicate that dense and localized polar regions tend to
hinder partitioning into the membrane, whereas weak
or diffuse polar surfaces are better tolerated. Moreover,
the greater the off-set between the barycenter of the
hydrophilic regions and the center of mass of the
molecule, i.e., the hydrophilic integy moment (IW de-
scriptors), the more likely a compound is to cross the
Conclusions
A novel series of COMT inhibitors containing the
nitrocatechol pharmacophore has been prepared. The
introduction and nature of heteroatom-containing sub-
stituents were found to have a most profound effect on
both the peripheral selectivity and duration of COMT
inhibition. Compounds containing oxygen in the side
chain such as 12 and 14 were identified as highly potent
in vitro inhibitors and subsequently as long-acting and
reasonably selective inhibitors of peripheral COMT in
vivo. However, from the series of highly in vitro active
N-arylpiperazine inhibitors, 1-(3,4-dihydroxy-5-nitro-

A Novel Inhibitor of Catechol-O-methyltransferase
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25 6215
Method B: 1-(3,4-Dihydroxy-5-nitrophenyl)-2-[(4-meth-
ylphenyl)thio]ethanone 15. To a stirred solution of 4-thio-
cresol (0.19 g, 1.49 mmol) in DMF (4 mL) at room temperature
was added powdered potassium carbonate in one portion
followed by R-bromo ketone 4 (0.41 g, 1.49 mmol) in three
portions. The resulting suspension was stirred for 20 min and
then filtered, and the filter cake was washed by DMF (1 mL).
The combined filtrate was evaporated (Labconco Centrivap
Concentrator, 75 °C) and water (5 mL) was added to the
residue. The mixture was extracted by ethyl acetate (2 × 15
mL), and the organic extracts were washed by dilute hydro-
chloric acid (10 mL), water (10 mL), and brine (10 mL); dried
over anhydrous sodium sulfate; filtered; and evaporated (40
°C, water aspirator pressure) to leave a pale yellow oil that
solidified on standing. Recrystallization from acetic acid af-
forded yellow crystals (0.304 g, 64%): νmax (KBr disk)/cm-1
3369 (OH), 1684 (CO), and 1539 (NO2); δH (400 MHz, DMSO-
d6) 10.90 (2 H, br, 3-OH, 4-OH), 8.01 (1 H, d, J 2.1, H-6), 7.62
(1 H, d, J 2.1, H-2), 7.22 (2 H, d, J 7.9, Ph), 7.10 (2 H, d, J 8.2,
Ph), 4.51 (2 H, s, CH2), and 2.30 (3 H, s, OCH3); δC (100 MHz,
DMSO-d6) 192.91, 148.64, 147.07, 138.21, 137.07, 132.30,
130.74, 130.52, 126.51, 118.47, 117.90, 40.91, and 21.60. Anal.
(C15H13NO5S) C, H, N.
phenyl)-3-[4-[3-(trifluoromethyl)phenyl]-1-piperazinyl]-
1-propanone hydrochloride (35, BIA 3-335) was found
to exhibit potent and almost completely peripherally
selective COMT inhibition in the mouse, unlike the
nonselective inhibitor tolcapone 1, and is endowed with
significantly improved duration of action compared to
entacapone 2. Molecular modeling was utilized to gain
an understanding of the basic structure-activity rela-
tionships within this novel series and has also provided
fresh insight into the structure of the ternary complex
formed between COMT and 35. Furthermore, compu-
tational analysis was successfully applied to the iden-
tification of the molecular determinants responsible for
the low BBB permeability of 35, which in effect propor-
tions excellent peripheral selectivity on this compound.
Accordingly, 35 is presented as a promising candidate
for clinical evaluation as an adjunct to L-DOPA therapy
for the treatment of Parkinson’s disease.
Experimental Section
Method C: 1-(3,4-Dihydroxy-5-nitrophenyl)-3-[(4-meth-
ylphenyl)thio]-1-propanone 16. To a stirred solution of
4-thiocresol (0.071 g, 0.57 mmol) in DMF (4 mL) at room
temperature was added powdered potassium carbonate in one
portion followed by 1-(3,4-dihydroxy-5-nitrophenyl)-2-propen-
1-one 6 (0.1 g, 0.48 mmol) in one portion. The resulting
suspension was stirred for 20 min and then filtered, and the
filter cake was washed by DMF (1 mL). The combined filtrate
was evaporated (Labconco Centrivap Concentrator, 75 °C) and
water (5 mL) was added to the residue. The mixture was
extracted by ethyl acetate (2 × 15 mL), and the organic
extracts were washed by dilute hydrochloric acid (10 mL),
water (10 mL) and brine (10 mL); dried over anhydrous sodium
sulfate; filtered; and evaporated (40 °C, water aspirator
pressure) to leave an orange-brown solid. Recrystallization
from acetic acid afforded orange crystals (0.11 g, 67%): νmax
(KBr disk)/cm-1 3377 (OH), 1689 (CdO), and 1542 (NO2); δH
(400 MHz, DMSO-d6) 10.86 (2 H, br, 3-OH, 4-OH), 7.97 (1 H,
d, J 2.1, H-6), 7.52 (1 H, d, J 2.1, H-2), 7.21 (2 H, d, J 8.2, Ph),
7.10 (2 H, d, J 8.2, Ph), 3.31 (2 H, t, CH2), 3.26 (2 H, t, CH2),
and 2.34 (3 H, s, CH3); δC (100 MHz, DMSO-d6) 196.51, 148.65,
146.89, 138.10, 136.51, 133.24, 130.81, 129.96, 127.71, 117.93,
117.01, 38.30, 28.50, and 21.51. Anal. (C16H15NO6S) C, H, N.
Method D: 1-(3,4-Dihydroxy-5-nitrophenyl)-2-(dimeth-
ylamino)ethanone Hydrochloride 18. To a stirred solution
of R-bromo ketone 4 (0.1 g, 0.36 mmol) in DMF (4 mL) at room
temperature was added a 40% aqueous solution of diethyl-
amine (0.43 mL, 3.46 mmol), and the resulting deep red
solution was stirred for 1 h, causing formation of a dark orange
precipitate. The solvent was removed by evaporation (Labconco
Centrivap Concentrator, 75 °C) and ethanol (2 mL) was added
to the residue, which was acidified to pH 1 by the addition of
a few drops of concentrated hydrochloric acid. On standing, a
precipitate formed which was removed by filtration and
washed with ethanol (1 mL) to give, after drying, a yellow solid
(0.096 g, 96%): νmax (KBr disk)/cm-1 3265 (OH), 1688 (CdO),
and 1548 (NO2); δH (400 MHz, DMSO-d6) 11.22 (1 H, br, 4-OH),
10.12 (1 H, br, 3-OH), 8.04 (1 H, d, J 2.1, H-6), 7.76 (1 H, d, J
2.1, H-2), 5.02 (2 H, s, CH2), and 2.91 (6 H, s, 2 × CH3); δC
(100 MHz, DMSO-d6) 190.22, 149.33, 148.37, 138.15, 124.73,
117.32, 62.34, and 44.72. Anal. (C10H12N2O5.HCl‚0.5H2O) C,
H, N.
Chemistry. Melting points were measured in open capillary
tubes on an Electrothermal Model 9100 hot stage apparatus
and are uncorrected. NMR spectra were recorded on a Bruker
Avance DPX (400 MHz) spectrometer with solvent used as
internal standard, and data are reported in the order: chemi-
cal shift (ppm), number of protons, multiplicity (s, singlet; d,
doublet; t, triplet; q, quartet; m, multiplet; br, broad), ap-
proximate coupling constant (J) in hertz, and assignment of a
signal. IR spectra were measured with a Bomem Hartmann
& Braun MB Series FTIR spectrometer using KBr disks.
Analytical TLC was performed on precoated silica gel plates
(Merck 60 Kieselgel F 254) and visualized with UV light.
Elemental analyses were performed on a Fisons EA 1110
CHNS instrument, and all analyses are consistent with
theoretical values to within (0.4%, unless otherwise indicated.
Solvents and reagents were purchased from Aldrich, Merck,
Fluka and used as received unless otherwise noted.
Building blocks 4 and 7 are known compounds,24 and 5 can
be prepared in three steps by standard procedures from
2-methoxyphenol via Friedel-Crafts acylation (chloropropionyl
chloride, AlCl3), nitration (60% HNO3 in AcOH), and dem-
ethylation (AlCl3, Cl(CH2)2Cl, ∆). The unsaturated ketone 6
was isolated in 87% yield from the reaction of 5 with potassium
phenolate in warm DMF (1 h at 100 °C). The following details
are representative procedures for synthesis of compounds
8-41.
Method A: 1-(3,4-Dihydroxy-5-nitrophenyl)-2-(1-naph-
thalenyloxy)ethanone 14. To a stirred solution of 1-naphthol
(1.54 g, 10.69 mmol) in DMF (10 mL) at room temperature
was added powdered potassium carbonate (1.48 g, 10.69 mmol)
in one portion followed by 2′-bromo-1-(3,4-dihydroxy-5-nitro-
phenyl)ethanone 4, (0.75 g, 2.72 mmol). The resulting deep
red suspension was then stirred at 100 °C for 1 h and then
allowed to cool to room temperature. The inorganic material
was removed by filtration and the filter cake was washed with
DMF (5 mL). The combined filtrate was then evaporated
(Labconco Centrivap Concentrator, 75 °C), and water (15 mL)
was added to the dark residue. The mixture was extracted by
ethyl acetate (2 × 20 mL), and the organic extracts were
washed by water (10 mL) and brine (10 mL) and then dried
over anhydrous sodium sulfate, filtered, and evaporated (40
°C, water aspirator pressure) to leave an orange-brown solid.
Recrystallization from acetic acid afforded yellow-orange
crystals (0.62 g, 67%): νmax (KBr disk)/cm-1 3376 (OH), 1699
(CO), and 1545 (NO2); δH (400 MHz, DMSO-d6) 11.05 (2 H, br,
3-OH, 4-OH), 8.31-8.25 (2 H, m, Ph), 8.22 (1 H, d, J 2.1, H-6),
8.15-7.85 (2 H, m, Ph), 7.60 (1 H, d, J 2.1, H-2), 7.52-6.97 (3
H, m, Ph), and 5.70 (2 H, s, CH2); δC (100 MHz, DMSO-d6)
193.01, 154.29, 148.82, 147.11, 138.34, 135.14, 128.51, 127.56,
127.02, 126.41, 125.91, 125.77, 122.73, 121.42, 117.90, 117.21,
106.83, and 71.21. Anal. (C18H13NO6) C, H, N.
Method E: 1-(3,4-Dihydroxy-5-nitrophenyl)-3-[4-[3-(tri-
fluoromethyl)phenyl]-1-piperazinyl]-1-propanone Hy-
drochloride 35. To a stirred suspension of 3,4-dihydroxy-5-
nitroacetophenone 7 (0.2 g, 1.01 mmol) in 2-propanol (5 mL)
at room temperature was added 1-(R,R,R-trifluoro-m-tolyl)-
piperazine (0.29 g, 1.28 mmol) followed by a 35% aqueous
solution of formaldehyde (0.5 mL) and concentrated hydro-
chloric acid (0.42 mL, 5.07 mmol). The mixture was then
stirred and heated at reflux for 7 h, during which time a

6216 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25
Learmonth et al.
copious yellow precipitate formed. The mixture was allowed
to cool to room temperature and the solid was filtered off and
washed with 2-propanol (1 mL). After drying, there was
obtained a yellow solid (0.27 g, 61%): νmax (KBr disk)/cm-1 3415
(OH), 1688 (CdO), and 1548 (NO2); δH (400 MHz, DMSO-d6)
10.90 (2 H, br, 4-OH, 3-OH), 8.13 (1 H, d, J 2.1, H-6), 7.73 (1
H, d, J 2.1, H-2), 7.51 (2 H, t, J 7.9, Ph), 7.33 (2 H, m, Ph),
3.64 (2 H, t, COCH2), 3.51 (2 H, t, CH2), and 3.35-3.22 (8 H,
m, 4 × CH2); δC (100 MHz, DMSO-d6) 195.11, 150.96, 148.98,
147.34, 138.22, 131.27, 130.99, 127.22, 125.01, 120.35, 117.86,
117.41, 116.83 (m), 112.81 (m), 51.86, 51.72, 45.91, and 33.45.
Anal. (C20H20F3N3O5.HCl‚0.75H2O) C, H, N.
Pharmacology. Complete experimental protocols for the
evaluation of in vitro and in vivo COMT inhibitory activity of
new compounds have been previously reported.14,17-18 All
animal interventions were performed in accordance with
European Directive number 86/609 and the rules of the
National Institute of Health’s Guide for the Care and Use
of Laboratory Animals (http://oacu.od.nih.gov/regs/guide/
guidex.htm).
teractions and hydrogen bonds, while counteracting contribu-
tions (positive scores) include acid-acid, base-base, and
hydrophobic-polar interactions. Note that the sign notation
used in this work is opposite to that of the original HINT
scores.
Membrane Permeation. Prediction of blood-brain barrier
permeation of 35 was performed with the VOLSURF algo-
rithm,32-34 as implemented in SYBYL version 6.9.1. The three-
dimensional structure of 35 was generated from its connection
table with CONCORD (Tripos Inc., St. Louis, MO) and the
final structure optimized by energy minimization using
MMFF94 force field. Three-dimensional interaction fields were
first calculated around the target molecule, using the energy
(Lennard-Jones, hydrogen bonding, and Coulombic electrostat-
ics) of interaction with a molecular probe placed at each node
of a regular 3D lattice. A grid resolution of 0.5 Å was employed,
and three distinct molecular probes (water, DRY and carbonyl
probes) were used to evaluate both polar and nonpolar mo-
lecular regions. In a subsequent step, VOLSURF transforms
the rich information content of the 3D molecular fields into a
set of scalar descriptors and projects their values onto a
predictive model of blood-brain barrier permeation. The
predictive model used in the present work was developed
elsewhere,23 by correlating the same set of descriptors with
the experimental permeation of nearly 300 known drugs, using
a cross-validated discriminant partial least squares procedure.
Compound 35 was modeled in the neutral form, with catechol
hydroxyls protonated and piperazine nitrogens unprotonated.
Computational Methods. Preparation of Molecular
Structures. Atomic coordinates of the ternary complex be-
tween COMT, the cosubstrate SAM, and the inhibitor 35 (BIA
3-335) were obtained from the PDB file 1H1D.19,20 Nearest
symmetry-related molecules were generated with the sym-
metry operations of the crystallographic space group P321
using the program Swiss-Pdb Viewer.25 For all subsequent
molecular modeling operations, the crystallographic water
molecules were deleted (except HOH53 coordinated to Mg2+
)
and all hydrogen atoms were added to the ligands and protein
atoms, using standard procedures. Due to the vicinity of the
Mg2+ ion and the positively charged SAM moiety, the catalytic
Lys144 residue is suggested to have an exceptionally lowered
pKa.26,27 Therefore, the side chain NH2 function of Lys144 was
modeled in the neutral form. Hydrogen atoms in the crystal-
lographic structure were finally optimized by energy minimi-
zation and intermolecular hydrogen bonds were checked for
optimal geometry.
Molecular Docking. Unrestrained flexible docking be-
tween COMT and selected inhibitors was performed with
the program GOLD version 1.228-30 using a genetic optimi-
zation algorithm. Default parameters were employed, except
for torsion constants of rotation about the C.ar-N.pl3 and
C.ar-C.2 bond types. These were increased to 12.0 kcal/mol
to hinder unrealistic free rotation of the catechol-nitro and
catechol-carbonyl bonds. For the genetic algorithm used in
the conformation space exploration, a population of 100
individuals (conformations) was subjected to 105 mutational
generations with a selection pressure of 1.1. All atoms at
COMT molecular surface within a radius of 14.0 Å from the
Mg2+ ion were used as the target binding site. The inhibitor’s
nitrocatechol moiety was docked with both hydroxyls proto-
nated and Lys144 was considered unprotonated. The coordina-
tion of the catechol hydroxyl groups to Mg2+ is implicitly
treated by GOLD as a special hydrogen bond, where the metal
behaves as a hydrogen bond donor. Therefore, no constraints
were used to force the catechol into the catalytic position.
Finally, all docked structures (protein and inhibitor) were
minimized with the molecular modeling package SYBYL
(Tripos Inc., St. Louis, MO), using the MMFF94 force field.
Hydrophatic Interactions. Intermolecular hydrophatic
interactions between COMT and the inhibitor (in either the
crystallographic or docked configurations) were evaluated with
the HINT algorithm,31 as implemented in SYBYL version 6.9.1.
Prior to HINT calculations, the Mg2+ ion and the catalytic
water molecule HOH53 were deleted. Partitioning of the
protein molecule was done using the dictionary method and
inferring the ionization states of protein residues from their
explicit hydrogen count. For the inhibitor, partitioning was
calculated explicitly by applying corrective factors to polar
proximity effects using the via-bond method. Polar interactions
were directed along vectors coinciding with lone pairs or π
orbitals. Stabilizing contributions to the HINT scores (negative
values) include all acid-base, hydrophobic-hydrophobic in-
Supporting Information Available: Spectroscopic and
microanalytical data for all synthesized compounds not in-
cluded in the Experimental Section. This material is available
free of charge via the Internet at http://pubs.acs.org.
References
(1) Axelrod, J.; Senoh, S.; Witkop, B. O-Methylation of catechol
amines in vivo. J. Biol. Chem. 1958, 233, 697-701.
(2) Guldberg, H.; Marsden, C. Catechol-O-methyltransferase: Phar-
macological aspects and physiological role. Pharmacol. Rev.
1975, 27, 135-206.
(3) Zhu, B. T.; Ezell, E. L.; Liehr, J. G. Catechol-O-methyltrans-
ferase-catalysed rapid O-methylation of mutagenic flavanoids.
Metabolic inactivation as a possible reason for their lack of
carcinogenicity in vitro. J. Biol. Chem. 1994, 269, 292-299.
(4) Burba, J. V.; Becking, G. C. Effect of the antioxidant norhy-
droguaiaretic acid on the in vitro activity of catechol-O-meth-
yltransferase. Arch. Int. Pharmacodyn. Ther. 1969, 180, 323-
329.
(5) Mannisto, P. T.; Kaakkola, S. New selective COMT inhibitors:
Useful adjuncts for Parkinson’s disease? Trends Pharmacol. Sci.
1989, 10, 54-56.
(6) Calne, D. B. Treatment of Parkinson’s disease. N. Engl. J. Med.
1993, 329, 1021-1027.
(7) Messiha, F.; Hsu, T.; Bianchine, J. Peripheral aromatic L-amino
acid decarboxylase inhibitor in parkinsonism. I. Effect on
O-methylated metabolites of L-Dopa-2-14C. J. Clin. Invest. 1972,
51, 452-455.
(8) Tohgi, H.; Abe, T.; Kikuchi, T.; Takahashi, S.; Nozaki, Y. The
significance of 3-O-methyldopa concentrations in the cerebrospi-
nal fluid in the pathogenesis of wearing-off phenomenon in
Parkinson’s disease. Neurosci. Lett. 1991, 132, 19-22.
(9) Borgulya, J.; Bruderer, H.; Bernauer, K.; Zurcher, G.; Da Prada,
M. Catechol-O-methyltransferase-inhibiting pyrocatechol deriva-
tives: Synthesis and structure-activity studies. Helv. Chim.
Acta. 1989, 72, 952-968.
(10) Backstrom, R.; Honkanen, E.; Pippuri, A.; Kairisalo, P.; Py-
stynen, J.; Heinola, K.; Nissinen, E.; Linde´n, I.-B.; Mannisto, P.
T.; Kaakkola, S.; Pohto, P. Synthesis of some novel potent and
selective catechol-O-methyltransferase inhibitors. J. Med. Chem.
1989, 32, 841-846.
(11) Borgulya, J.; Da Prada, M.; Dingemanse, J.; Scherschlicht, R.;
Schlappi, B.; Zurcher, G. Ro 40-7592. Catechol-O-methyltrans-
ferase (COMT) inhibitor. Drugs Future 1991, 16, 719-721.
(12) Nissinen, E.; Linde´n, I.-B.; Schultz, E.; Pohto, P. Biochemical
and pharmacological properties of a peripherally acting catechol-
O-methyltransferase inhibitor Entacapone. Nauyn Schmiede-
bergs Arch. Pharmacol. 1992, 346, 262-266.
(13) Assal, F.; Spahr, L.; Hadengue, A.; Rubbici-Brandt, L.; Burkhard,
P. R. Tolcapone and fulminant hepaptitis. Lancet 1998, 352, 958.

A Novel Inhibitor of Catechol-O-methyltransferase
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 25 6217
(14) Learmonth, D. A.; Vieira-Coelho, M. A.; Benes, J.; Alves, P. C.;
Borges, N.; Freitas, A. P.; Soares-da-Silva, P. Synthesis of 1-(3,4-
dihydroxy-5-nitrophenyl)-2-phenylethanone and derivatives as
potent and long-acting peripheral inhibitors of catechol-O-
methyltransferase. J. Med. Chem. 2002, 45, 685-695.
(15) Almeida, L.; Soares-da-Silva, P. Pharmacokinetics and pharma-
codynamics of BIA 3-202, a novel COMT inhibitor, during first
administration to humans. Drugs R&D 2003, 4(4), 207-217.
(16) Almeida, L.; Soares-da-Silva, P. Pharmacokinetics and pharma-
codynamic profiles of BIA 3-202, a novel catechol-O-methyl-
transferase (COMT) inhibitor, during multiple-dose adminis-
tration to health subjects. J. Clin. Pharmacol. 2003, 43(12),
1350-1360.
(24) Bernauer, K.; Borgulya, J.; Bruderer, H.; Da Prada, M.; Zurcher,
G. US Patent No. 5,705,703, 1998.
(25) Guex, N.; Peitsch, M. C. SWISS-MODEL and the Swiss-
PdbViewer: An environment for comparative protein modeling.
Electrophoresis 1997, 18, 2714-2723.
(26) Zheng, Y. J.; Bruice, T. C. A theoretical examination of the
factors controlling the catalytic efficiency of a transmethylation
enzymesCatechol-O-Methyltransferase. J. Am. Chem. Soc. 1997,
119, 8137-8145.
(27) Kuhn, B.; Kollman, P. A. QM-FE and molecular dynamics
calculations on catechol-O-methyltransferase: Free energy of
activation in the enzyme and in aqueous solution and regiose-
lectivity of the enzyme-catalyzed reaction. J. Am Chem. Soc.
2000, 122, 2586-2596.
(17) Costa, J. L.; Vieira-Coelho, M. A.; Soares-da-Silva, P. Catechol-
O-methyltransferase activity in SK-N-SH cells. Eur. J. Neu-
rosci. 2000, 12(suppl. 11), 280.
(28) Jones, G.; Willett, P.; Glen, R. C. A genetic algorithm for flexible
molecular overlay and pharmacophore elucidation. J. Comput.-
Aided Mol. Des. 1995, 9, 532-549.
(18) Borges, N.; Vieira-Coelho, M. A.; Parada, A.; Soares-da-Silva,
P. Studies on the tight-binding nature of tolcapone inhibition of
soluble and membrane-bound rat brain catechol-O-methyltrans-
ferase. J. Pharmacol. Exp. Ther. 1997, 282, 812-817.
(19) Rodrigues, M. L.; Archer, M.; Bonifacio, M. J.; Soares-da-Silva,
P.; Carrondo, M. A. Crystallization and preliminary crystal-
lographic characterization of catechol-O-methyltransferase in
complex with its cosubstrate and an inhibitor. Acta Crystallogr.
2001, D57, 906-908.
(20) Bonifacio, M. J.; Archer, M.; Rodrigues, M. L.; Matias, P. M.;
Learmonth, D. A.; Carrondo, M. A.; Soares-da-Silva, P. Kinetics
and crystal structure of catechol-O-methyltransferase complex
with co-substrate and a novel inhibitor with potential thera-
peutic application. Mol. Pharmacol. 2002, 62, 795-805.
(21) Cozzini, P.; Fornabaio, M.; Marabotti, A.; Abraham, D. J.;
Kellogg, G. E.; Mozzarelli, A. Simple, intuitive calculations of
free energy of binding for protein-ligand complexes. 1. Models
without explicit constrained water. J. Med. Chem. 2002, 45,
2469-2483.
(29) Jones, G.; Willett, P.; Glen, R. C. Molecular recognition of
receptor sites using a genetic algorithm with a description of
desolvation. J. Mol. Biol. 1995, 245, 43-53.
(30) Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R.
Development and validation of a genetic algorithm for flexible
docking. J. Mol. Biol. 1997, 267, 727-748.
(31) Kellogg, G. E.; Joshi, G. S.; Abraham, D. J. New tools for
modelling and understanding hydrophobicity and hydrophobic
interactions. Med. Chem. Rev. 1992, 1, 444-453.
(32) Cruciani, G.; Pastor, M.; Clementi, S. Handling information from
3D grid maps for QSAR studies. Molecular Modelling and
Prediction of Bioactivity; Kluwer Academic/Plenum Publishers:
New York, 2000; pp 73-82.
(33) Cruciani, G.; Crivoni, P.; Carrupt, P.-A.; Testa, B. Molecular
fields in quantitative structure-permeation relationships: The
VolSurf approach. J. Mol. Struct. (Theochem) 2000, 503, 17-
30.
(34) Cruciani, G.; Pastor, M.; Guba, W. VolSurf: A new tool for the
pharmacokinetic optimization of lead compounds. Eur. J. Pharm.
Sci. 2000, 11(Suppl 2), S29-39.
(22) Vidgren, J.; Svensson, L. A.; Liljas, A. Crystal structure of
catechol-O-methyltransferase. Nature 1994, 368, 354-358.
(23) Crivori, P.; Cruciani, G.; Carrupt, P. A.; Testa, B. Predicting
blood-brain barrier permeation from three-dimensional molec-
ular structure. J. Med. Chem. 2000, 43, 2204-2216.
JM040848O