 |
Printable
PDF Version
Bivalent Ligands Derived
from Huperzine A as
Acetylcholinesterase
Inhibitors
H. Haviv1, D.M. Wong2,
I. Silman3 and J.L.
Sussman4,*
1Department of
Biological Chemistry,
Weizmann Institute of
Science, Rehovot 76100,
Israel, 2Department of
Chemistry, Virginia
Tech, Blacksburg, VA
24061, USA, 3Department
of Structural Biology,
Weizmann Institute of
Science, Rehovot 76100,
Israel, 4Department of
Neurobiology, Weizmann
Institute of Science,
Rehovot 76100, Israel
Abstract: The naturally
occurring alkaloid
Huperzine A (HupA) is an
acetylcholinesterase (AChE)
inhibitor that has been
used for centuries as a
Chinese folk medicine in
the context of its
source plant Huperzia
Serrata. The potency and
relative safety of HupA
rendered it a promising
drug for the
ameliorative treatment
of Alzheimer's disease
(AD) vis-à-vis the
"cholinergic hypothesis"
that attributes the
cognitive decrements
associated with AD to
acetylcholine deficiency
in the brain. However,
recent evidence supports
a neuroprotective role
for HupA, suggesting
that it could act as
more than a mere
palliative. Biochemical
and crystallographic
studies of AChE revealed
two potential binding
sites in the a ative-site
gorge of AChE, one of
which, the “peripheral
anionic site” at the
mouth of the gorge, was
implicated in promoting
aggregation of the beta
amyloid (Ab) peptide
responsible for the
neurodegenerative
process in AD. This
feature of AChE
facilitated the
development of dual-site
binding HupA-based
bivalent ligands, in
hopes of concomitantly
increasing AChE
inhibition potency by
utilizing the "chelate
effect", and protecting
neurons from Ab
toxicity. Crystal
structures of AChE
allowed detailed
modeling and docking
studies that were
instrumental in
enhancing the
understanding of
underlying principles of
bivalent
inhibitor-enzyme
dynamics. This monograph
reviews two categories
of HupA-based bivalent
ligands, in which HupA
and HupA fragments serve
as building blocks, with
a focus on the recently
solved crystallographic
structures of Torpedo
californica AChE in
complex with such
bifunctional agents. The
advantages and drawbacks
of such structured-based
drug design, as well as
species differences, are
highlighted and
discussed.INTRODUCTION
(–)-Huperzine A [(–)-HupA]
is an enantiomeric
lycodine alkaloid
isolated from the club
moss Huperzia serrata of
the Lycopodium species (Huperziaceae)
[1]. The plant H.
serrata itself is a
centuries-old Chinese
folk medicine (known as
Qian Ceng Ta),
traditionally used for
the treatment of various
maladies. (–)-HupA has
been identified as a
potent, specific and
reversible inhibitor of
the enzyme
acetylcholinesterase (EC
3.1.1.7, AChE), relative
to butyrylcholinesterase
(EC 3.1.1.8, BChE). The
extensive studies on
(–)-HupA as a lead
compound for the
development of more
effective anti-AChE
drugs for the treatment
of Alzheimer’s
disease (AD) relative to
those approved by the
FDA, such as donepezil
(E2020, Aricept®) [2-4],
(–)-galanthamine (Reminyl®)
[5] and rivastigmine
(Exelon®) [6], have been
attributed to its better
penetration through the
blood brain barrier, its
higher oral
bioavailability and its
longer duration of AChE
inhibitory action [7,
8]. Phase IV clinical
trials conducted in
China demonstrated that
(–)-HupA induces
significant improvement
in the memory of elderly
people with AD and
vascular dementia,
without any noticeable
side
effects [8], and it is
currently in phase II
trials in the United
States [9]. It has also
been considered for use
as a protective agent
against organophosphate
nerve agent intoxication
[10]. It should be
noted, however, that
(–)-HupA has been shown
to serve as a
neuroprotective agent
against a variety of
insults and, at a
molecular level, has
been shown to serve as
an NMDA receptor
antagonist [8]. Thus,
novel lead compounds,
based on the (–)-HupA
scaffold, may have a
broad spectrum of
pharmacological actions,
and not serve merely as
anticholinesterase
(anti-ChE) agents. AD is
a neurological disorder
characterized by a
significant decrease in
hippocampal and cortical
levels of the
neurotransmitter
acetylcholine (ACh),
leading to severe memory
and learning deficits
[11]. According to the
“cholinergic hypothesis”
[12], inhibition of AChE
in the central nervous
system can alleviate
these deficits; indeed,
all first-generation AD
drugs are cholinesterase
inhibitors (ChEIs),
including the synthetic
compounds tacrine
(Cognex®) [13, 14],
donepezil (Aricept®)
[2-4] and rivastigmine
(Exelon®) [6], and the
alkaloids (–)-galanthamine
(Reminyl®) [5] and (–)-HupA
(see above). It has long
been known that AChE has
an enhanced affinity for
bifunctional inhibitors
such as the bis-quaternary
inhibitors,
decamethonium (DECA)
[15], BW284c51 [16] and
ambenonium [17] (Fig. 1)
[18]. It was postulated
that this is due to the
existence of two binding
sites in the enzyme
[15]. These sites bind
to the bivalent ligands
simultaneously, giving
rise to what is known as
the “chelate effect”.
This is a well-known
thermodynamic phenomenon
that is manifested as
the enhanced stability
of a complex formed from
bi- or polyvalent
ligands and their
targets, compared to a
similar complex formed
from the monovalent
counterparts. This
enhanced stability is
the result of a decrease
in entropic penalty for
binding a single large
ligand versus two
smaller ones, or the
increase in the binding
probability of a
functional group in the
ligand after the first
one has bound [19]. The
chelate effect is
efficiently exploited by
scientists in pursuit of
stronger and more
specific enzyme
inhibitors that could
serve as lead compounds
for the development of
potent drugs. Recent
examples of such
endeavors are the
development of an HIV
integrase inhibitor [20]
and a potent
bifunctional
anticoagulant [21]. The
crystal structure of
AChE from the ray
Torpedo californica (TcAChE)
has provided invaluable
insights into the
catalytic mechanism of
this remarkable enzyme
[22]. A
deep narrow gorge lined
with aromatic residues
penetrates halfway
through the
quasi-globular structure
of the enzyme. At the
bottom of the gorge
resides the catalytic
triad responsible for
hydrolyzing ACh, and two
aromatic side chains,
tryptophan and
phenylalanine (W84 and
F330 in the case of
TcAChE,) that constitute
the so-called “catalytic
anionic subsite” (CAS)
and make cation-p
interactions with the
quaternary ammonium
group of ACh [23, 24].
The “peripheral anionic
site” (PAS) [25], that
is responsible for the
enhanced binding of
bisquaternary ligands,
was shown to be located
near the entrance to the
active-site gorge, and
also contains two
aromatic side chains,
those of Y70 and W279.
Subsequent solution of
the 3D structures of a
large repertoire of
complexes and conjugates
of AChE with a variety
of inhibitors [26], in
particular anti-AD drugs
[27], provided a wealth
of structural
information which could
be applied to the design
of novel lead compounds.
The crystal structure of
the DECA/TcAChE complex
clearly showed how the
bisquaternary ligand
orients along the
active-site gorge,
bridging the CAS and the
PAS (Fig. 2) [28]. Thus,
the structure of AChE
and of complexes with
gorge-spanning ligands,
such as DECA and
BW284c51 [29], provided
a clear explanation for
the chelate effect
exhibited by dualbinding
inhibitors. A large body
of kinetic studies,
utilizing site-directed
mutagenesis, supports
these structural
assignments [30].
Solution of the crystal
structure of the (–)-HupA/TcAChE
complex revealed the key
structural features
involved in their
interaction, and opened
the way for
structure-based drug
design based on the (–)-HupA
template (Fig. 3) [32].
In silico docking
experiments implied that
(–)-HupA might bind not
only at the bottom of
the gorge, but also to
the PAS [33]. Evidence
for the existence of two
potential binding sites,
and the availability of
high-resolution
structures, have
prompted the rational
design of blood-brain
barrier penetrable
dual-affinity ligands,
with the prospects of
augmented specificity
and potency of
inhibition. The first
such success was
reported in 1996 with
the design of bis(7)-
tacrine (Fig. 4) [18,
33], and its structure
in complex with TcAChE
has been recently
published [34]. There is
accumulating evidence
that bis(7)-tacrine has
neuroprotective
qualities that do not
involve AChE inhibition,
such as attenuation of
neuronal apoptosis
induced by the b-amyloid
(Ab) peptide via
regulation of L-type
Ca+2 channels [35, 36].
It is generally accepted
that Ab is the agent
responsible for the
neurodegenerative
process in AD [37]. In
this context, it
is important to note
that evidence has been
presented that AChE can
enhance the rate of
aggregation of Ab, and
that the PAS is the site
on the enzyme involved
in this action [38- 42].
Thus, in addition to the
palliative effect
achieved by increasing
ACh levels in the brain,
inhibition of AChE may
prove to have a
neuroprotective role and
to actually retard the
progress of AD. It
should also be noted
that there is evidence
that AChE plays
‘non-classical’ roles,
in addition to its role
in terminating
transmission at
cholinergic synapses
[43], e.g. serving as an
adhesion protein [44].
Thus, again, anti-ChE
agents may have
additional biological
actions. In the
following, two
categories of huperzine
dimers designed as lead
compounds will be
discussed: symmetric
dimers, in which two
HupA or HupA-derived
pharmacophores are
linked via an alkyl
chain, and asymmetric
ligands, in which the
linker connects between
HupA or a structurally
related entity at one
extremity and a
different pharmacophore,
e.g. tacrine, at the
other extremity. Some of
these ligands hold great
promise as lead
compounds; others have
failed to show
potential, but have
taught us that much is
yet to be explored to
fully comprehend
enzyme-inhibitor
interplay within the
active-site gorge of
AChE. SYMMETRIC
HUPERZINE LIGANDS The
most trivial species of
a bis-huperzine ligand
would be two (–)-HupA
moieties connected via a
tether (1), as shown in
Fig. 5. Docking
experiments suggested
that a tether length of
7-12 carbons would be
required for optimal
dualsite binding of a
bis-(–)-HupA dimer, 1.
Such dimers were
synthesized and tested
for their potency.
Despite the coincidence
of measured inhibition
potencies with trends
calculated from the
docking studies, none of
these bivalent ligands
surpassed the monomer in
their affinity [45].
Comparison of the
crystal structures of
(–)-HupA/TcAChE and (+)-HupA/TcAChE
suggests an explanation
for these findings [46].
Both (–)-HupA itself and
(+)-HupA bind similarly
at the bottom of the
active-site gorge (Fig.
6), suggesting that one
(–)-HupA subunit of the
dimer would bind in a
similar orientation to
that of the monomer at
the bottom of the gorge.
In this binding
orientation the
pseudoequatorial N5 that
is fused to the linker
in the (–)-HupA dimer
points towards the
bottom of the gorge,
i.e. away from the PAS.
One can envision that in
this position, the bound
(–)-HupA moiety will not
easily accommodate an
alkylene tether leading
upwards towards the PAS;
a longer linker may be
required for optimum
binding of both subunits
to the enzyme. However,
for the enantiomorph,
(+)-HupA, the amino N5
atom is more favorably
oriented; thus, a dimer
composed of two (+)-HupA
units might tolerate a
shorter linker for
optimal dual-site
binding. HupB] moieties
connected via a
carbon-nitrogen chain
(Fig. 7) [48]. (–)-HupB,
which is very similar in
structure to (–)- HupA,
was also isolated from
H. serrata [1]. The
crystal structure of the
(–)-HupB/TcAChE complex
showed that (–)-HupB
makes the same contacts
with the enzyme as (–)-HupA,
except that the
ethyledene-W84 p-p
interaction is
absent in the (–)-HupB
complex [46]. (–)-HupB
has a weaker affinity
for rat AChE (rAChE)
than (–)-HupA, but
dimerization enhanced
potency by up to almost
three orders of
magnitude. In the
optimal dimer (2)
studied, the IC50 was
improved from 19.3 μM
for the monomer to 4.9
nM for the dimer. Under
the experimental
conditions reported,
this value was even
lower than that for (–)-HupA
(72.4 nM). Docking
studies suggested that
favorable interactions
of the tether with
several residues in the
middle of the gorge
contribute to the high
potency of this dimer
(Fig. 8). It is
noteworthy that the
optimal tether length,
that of 2, is 18 atoms.
It is possible that the
position of N5, oriented
towards the bottom of
the gorge in the crystal
structure of the (–)-
HupB/TcAChE complex, as
was the case for the
corresponding (–)-HupA
complex, is the reason
why a longer tether is
required. This supports
the explanation for the
apparent failure to
increase affinity in the
case of the (–)- HupA
dimers studied (see
above). Another
successful
implementation of the
chelate effect
was the synthesis and
evaluation of a (–)-HupA
fragment, hupyridone
(3a,b) (Fig. 9), and of
corresponding bivalent
derivatives. The dimers,
(S,S)-(–)-4a-c, were
designed on the basis of
cumulative evidence that
dimerization of weak or
inactive fragments
derived from (–)-HupA or
from tacrine can produce
potent ligands [47, 50,
51]. Compounds 3a,b were
initially synthesized
with the rationale of
preserving the key
functionalities in (–)-HupA
that had been observed
to interact with
residues at the
active-site of the
enzyme in the crystal
structure. These are,
specifically, the
pyridone oxygen and the
ring nitrogen, that are
hydrogen-bonded to Y130
and to G117/Q199 (via a
water molecule), and the
5-amino group which
associates with W84 and
F330 via cation-p
interactions (see Figs.
3 and 9) [32, 51]. The
3a,bfragments, like
their parent molecule,
have a chiral center
atposition 5. The
N-butyl-hupyridone
control (3c) shows very
weak inhibitory activity
(IC50 of ~0.5 mM), but
the enantiomeric dimers
of the parent monomer
(3a) are surprisingly
potent. For example, the
dimers (S,S)-(–)-4b,c,
with corresponding 12-
and 13-carbon tethers,
both display IC50 values
of 52 nM for rAChE. This
four orders of magnitude
increase in potency can
be attributed in part to
hydrophobic interactions
of the carbon tether
with the multiple
aromatic residues lining
the active-site gorge.
In addition, docking
studies performed for
(R)- and (S)-3b showed
that the binding
orientation in the (R)
configuration forces the
5-amino group to point
towards the bottom of
the
gorge, which would not
easily support a tether
leading up the gorge to
the PAS, thus
rationalizing a 60-fold
difference in affinity
between the (R,R)- and (S,S)-enantiomers
of 4b [51]. In an
attempt to understand
the molecular
determinants of the
affinity of the 4 dimer
series for AChE, and to
compare them with those
for the (–)-HupA/TcAChE
complex, the crystal
structures of both the (S,S)-(–)-4a/TcAChE
and (S,S)-(–)-4b/TcAChE
complexes were
determined (Fig. 10)
[52]. In these
structures, both dimers
bind in the expected
manner, one 3a monomer
being positioned at the
active-site, and the
other at the PAS. At the
active-site, the 5-amino
moiety makes a cation-p
interactionwith the
aromatic rings of W84
and F330, and the
aliphatic part of the 3a
entity makes hydrophobic
interactions with W84
and F330. Similar
interactions are made
with W279 at the PAS.
The pyridone oxygen and
nitrogen make the same
interactions at the
bottom of the gorge as
made by (–)-HupA.
Obviously, interactions
with the absent (–)-HupA
ethylidene
function are lacking,
such as a p-p
interaction with the p
system of W84/F330, and
the H-bond between the
terminal methyl group of
the ethylidene moiety
and H440. A striking
feature common to the
structures of the TcAChE
complexeswith (–)-HupA,
(+)-HupA, (–)-HupB and
the (–)-4a and (-)-4b
dimers, is a peptide
bond-flip between G117
and G118 in the oxyanion
hole at the active-site.
This flip had been
ascribed to repulsion
between G117 O and the
pyridone oxygen [46].
However, molecular
dynamics studies
simulating the binding
and release of (–)-HupA
from TcAChE indicate
that the flip occurs
even when the inhibitor
is too distant to induce
it directly, implying
that a tendency to flip
is an intrinsic property
of this particular bond
[53]. These
simulations attribute a
key role to D72 in
drawing in and guiding
(–)-HupA out of the
active site, and stress
the importance of
“lubricating” water
molecules in the
activesite gorge. The
tether in the complexes
of both the (–)-4a dimer
and the (–)-4b dimer
winds its way up the
gorge, making what
appear to be favorable
hydrophobic interactions
with the aromatic rings
lining the gorge. At the
PAS, the 3a moiety is
seen to fold back into
the gorge, stabilized by
hydrogen bonds of the
pyridone oxygen and
nitrogen with F288 and,
via a water molecule,
with R289 and S286. The
structures of the
(–)-4a,b/TcAChE
complexes make an
important contribution
to enhancing our
understanding of the
elements responsible for
species-dependent
differences in
specificity and affinity
of inhibitors. For
TcAChE, the shorter
10-carbon tethered
inhibitor binds better
than its longer
12-carbon homologue, the
IC50 values for (–)-4a
and (–)-4b being 2.4 and
16 nM, respectively.
However, for rAChE, the
dimers reverse their
affinities. The longer
inhibitor is preferred,
with the IC50 values
being 151 and 52 nM for
(–)-4a and (–)-4b,
respectively. It is
difficult to reduced
entropy loss, as it
should favor the shorter
linker in both cases, or
likewise in terms of a
reduced
liganddesolvation
penalty, which should be
consistent for the
longer tether. The most
plausible explanation
is, therefore, that the
preferences are inherent
in different
enzymeinhibitor
complementarities.
Indeed, the mouse AChE
(mAChE) structure [54],
which is most likely
very similar to that of
rAChE, displays a
narrowing of the gorge
at its bottom relative
to that of TcAChE. In
addition, the flexible
F330 in TcAChE is
homologous to Y337 in
mAChE, which appears to
be restricted in motion
by H-bonding to Y341 via
its hydroxyl, and is
proposed by molecular
modeling [55] to make a
H-bond with the amino
group of (–)-HupA. This
may explain the enhanced
affinity of (–)-HupA for
mammalian AChE relative
to TcAChE [52]. The
rigidity of Y337 in
mammalian AChE, relative
to the flexibility of
F330 in TcAChE, may
force the linker in
4a/4b to take a path in
the mammalian enzymes
different from the one
observed in the TcAChE
crystal structure, as is
the case for DECA in
mAChE vs TcAChE [54],
resulting in the
observed switch in dimer
potencies between
species (Fig. 11). The
contributions of these
subtle differences in
structure to the
differences observed in
affinity demand more
investigation. A clear
understanding of their
influence on, and
interplay with, the
thermodynamics of
inhibitor binding is
crucial for successful
drug design.
ASYMMETRIC HUPERZINE
LIGANDS
Several bivalent ligands
containing (–)-HupA or
related pharmacophores
linked to other groups
have been reported. One
such set of inhibitors
are hybrids of a (–)-HupA-based
pharmacophore and a
fragment of E2020 [56].
E2020 (donepezil,
Aricept®) is a potent,
enantiomeric,
benzylpiperidine- based
AChE inhibitor approved
for clinical treatment
of AD in 1996, which is
administered as a
racemate, and
extensively used for
this purpose [57] (Fig.
12). The crystal
structure of the E2020/TcAChE
complex clearly shows
its mode of binding [3].
(R)-E2020 spans the
length of the gorge,
with its proximal benzyl
moiety stacked against
W84, and its distal
dimethoxyindanone group
bound at the PAS (Fig.
13). The (S)-enantiomer
of E2020 was not
seen in the crystal
structure, an
observation suggestive
of an inherent
enantioselectivity of
TcAChE for the (R)-enantiomer.
Three HupA-E2020 hybrid
compounds consisting of
the benzylpiperidine
moiety of E2020 linked
to derivatives of (–)-HupA
(5a-c) were reported
(Fig. 12) [56]. These
ligands were designed on
the basis of docking
studies with E2020 that
resulted in an opposite
orientation of the bound
E2020 to the one seen in
the solved crystal
structure of the E2020/TcAChE
complex, with the benzyl
group being bound
at the PAS, and the
dimethoxyindanone moiety
at the bottom of the
gorge. Compound 5c was a
weak inhibitor (IC50 of
17.2 μM) of rAChE, while
the other two hybrids
showed no significant
inhibition. No reference
to the stereochemistry
of the compounds, nor
any rationale for the
selected modifications
of the (–)-HupA-derived
pharmacophores were
provided, and it is
possible that they were
not tolerated by the
enzyme. (–)-Huperzine-tacrine
hybrid ligands were
designed with the
objective of inhibiting
both human AChE and BChE
(hAChE and hBChE,
respectively) [58]. BChE
is widely distributed in
vertebrate tissues, and
in the brain it appears
to be mostly of glial
origin, while AChE is
mostly of neuronal
origin [59]. Despite its
wide distribution, the
physiological function
of BChE is still
uncertain [60]. An
experiment conducted
with knockout mice
lacking the two
functional alleles of
AChE provided a truly
striking observation. In
spite of the complete
absence of AChE from
their bodies, and
certain physiological
features that bear
resemblance to symptoms
of inhibition of AChE,
such as gastrointestinal
hypomotility and a fine
tremor, such mice are
very much alive, can
actually live up to
several months, and have
developed principal
anatomical components of
functional cholinergic
pathways without any
evident compensatory
increase in the
distribution of BChE
[59, 61]. These
remarkable results argue
that BChE plays a
constitutive role in
hydrolyzing ACh in the
mouse brain, since
otherwise, the mice
would have died from
excessive cholinergic
stimulation. Thus, by
projection onto humans,
BChE should be
considered an important
target, together with
AChE, in the treatment
of AD [60]. Molecular
modeling studies based
on the crystallographic
structure of hAChE (PDB
code 1B41) and a
homology model of hBChE
(PDB code 1EHO, based on
TcAChE) assisted in the
identification of a
putative mid-gorge
binding site in hAChE
consisting of Y72/D74,
and in hBChE consisting
of N68/D70 [62]. These
binding sites were
targeted in the design
of three bivalent
inhibitors in which
tacrine was connected,
via a 7-carbon tether,
to a (–)-HupA-derived
pharmacophore (6a-c)
(Fig. 14) [58]. Docking
experiments done with
the three inhibitors on
hAChE suggest that the
favorable binding
orientations of both 6a
and 6b position the
tacrine group for a p-p
stacking interaction
against W286 at the PAS,
while the bicyclic group
binds at the bottom of
the gorge close to W86
(homologous to W84 in
TcAChE). However, for
the more potent
inhibitor 6c, the
proposed binding
orientation to hAChE is
the opposite to that
predicted for 6a and 6b;
the tacrine moiety of 6c
is stacked against W86
at the CAS, as seen in a
number of crystal
structures of complexes
of tacrine and tacrine-based
inhibitors with TcAChE
and mAChE (see for
example Lit. [34,
63-66]), and the
huperzine-fragment
moiety interacts
favorably with W286 at
the PAS. In hBChE, the
tacrine is stacked
against W82 (homologous
to W84 in TcAChE) for
all three inhibitors,
while the bicyclic
moiety binds at the
midpoint of the
activesite gorge. The
reported dissociation
constants are of the
same order of magnitude
for all three inhibitors
with respect to both
enzymes. For hAChE, the
inhibitors have
significantly lower
dissociation constants
than tacrine, and with
regard to hBChE they all
display higher binding
affinity than (±)-HupA.
The design of a
hupyridone-tacrine
hybrid inhibitor, 6d
(Fig. 14), was based on
the observation that
p-p, cation-p and
hydrophobic interactions
contribute to ligand
affinity for the PAS
[67]. The optimal tether
length was found to be
10 carbons, yielding an
IC50 of 8.8 nM for rAChE,
i.e. 10- and 20-fold
more potent than (–)-HupA
and tacrine,
respectively [68]. In an
attempt to understand
the mode of binding of
this hybrid dimer, the
structure of the 6d/TcAChE
complex was recently
solved in both trigonal
and orthorhombic crystal
forms, after soaking the
enzyme crystals with a
racemic
mixture of 6d [64]. In
both crystal forms, the
tacrine moiety was seen
to be bound at the
active-site, sandwiched
between W84 and F330,
with its acridine amine
hydrogen bound to H440,
and the linker-fused
nitrogen bound to a
system of three
conserved water
molecules. This
active-site binding mode
is practically identical
to the binding mode of
monomeric tacrine
reported previously in
the tacrine/TcAChE
complex [63], and
similar to the
orientation in complexes
of other tacrine-based
bivalent ligands with
TcAChE and
mAChE [34, 65, 66]. The
10-carbon linker spans
the gorge up to the PAS,
where the secondary
amino group linking it
to the 3a moiety makes a
cation-p interaction
with W279. Surprisingly,
in the trigonal
structure solved
initially, only the
(R)-3a moiety of 6d was
visible, protruding
outside the gorge, and
aligned parallel to the
gorge axis, with the
pyridone oxygen and
nitrogen bound to K11
and Q185 of a proximal,
symmetry-related AChE
molecule (Fig. 15). It
was not clear from this
structure whether the
visible isomer in the
gorge was selected out
of the racemate due to
an inherent preference
of the enzyme arising
out of steric
constraints, or because
of the crystal symmetry
contacts of the 3a
subunit. This phenomenon
merited further
investigation, since
both explanations bore
interesting
consequences: inherent
preference of the enzyme
for one enantiomer would
imply potentially
superior inhibition,
whereas enantiomeric
selection of
an inhibitor as an
artifact arising as a
consequence of crystal
packing had never been
reported. The issue was
resolved bysoaking
orthorhombic TcAChE
crystals with the
racemate. For the
orthorhombic crystals it
had been established
that the entrance to the
gorge was remote from a
symmetry-related
molecule, a priori
precluding
symmetry-related
interactions of the
possibly protruding 3a
moiety of 6d [69]. The
orthorhombic 6d/ TcAChE
crystal structure
revealed both nantiomers
bound at the
active-site, i.e.
approximately half of
the molecules in the
crystal bound the (S)-enantiomer
(Fig. 16), while the
other half bound the
(R)-enantiomer. The
(S)-3a moiety of 6d at
the PAS adopts a folded
conformation essentially
identical to the
conformation of the
(S)-3a moiety in the
(–)-4a/TcAChE structure
(see above), while the
(R)-3a moiety of 6d
assumes a novel
conformation,
perpendicular to the
gorge axis. The
unequivocal
interpretation of these
results is of an
artifactual enantiomeric
selectivity, due to
crystal packing, which
occurs in the trigonal
crystals. Comparison of
the orthorhombic (S)-3a
vs. the trigonal (R)-3a
mode of binding supplies
a plausible explanation:
the (S)- 3a subunit
makes only two H-bonds
with the enzyme, while
the (R) moiety makes
three: two with the
symmetry-related side
chains of K11 and Q185,
and one with S286 via a
water molecule. Thus it
is likely that in a
competition between (R)
and (S)–6d for binding
to the enzyme in the
trigonal crystal, the
(R)-enantiomer has the
upper hand due to
favorable thermodynamics
at the PAS. It is
interesting to compare
this phenomenon to the
putative
enantioselectivity
reported for E2020 (see
above).
CONCLUSIONS AND FUTURE
PROSPECTS
The application of
quantitative
structure-activity
relationships (QSAR),
molecular dynamics and
X-ray crystallography in
rational drug design,
whether individually or
in combination, has
resulted in the
discovery of numerous
lead compounds that have
served as starting
points for obtaining
drugs for the treatment
of a broad repertoire of
major diseases. The fact
that, for AD, these
approaches have already
not only resulted in
several approved drugs,
but also in novel leads,
is of especial
importance, since the
severe side effects
often caused by ChEIs
render the availability
of several treatment
regimens an imperative.
Several fundamental
features of the drug,
the target enzyme, and
the underlying
principles of their
interaction must be
taken into consideration
if better inhibitors are
to be obtained: -
Crystal structures are
snapshots of stable
structures, and afford
an excellent starting
point for drug design;
but thermodynamic
considerations such as
desolvation energies of
the ligand, entropic
penalties, and
conformational
destabilization of the
protein, the ligand, or
both, must be included
in any simulation
attempting to accurately
predict the binding of a
ligand. - The solvent is
ubiquitously involved in
every aspect of the
catalytic process and of
inhibitor dynamics. Its
role and contribution
should, therefore, never
be underestimated. - The
importance of the tether
length in designing
bivalent ligands which
bind along the
active-site gorge of
AChE is threefold: i. It
reduces entropy loss on
binding; ii. It may
enhance or diminish
affinity by having its
own beneficial or
detrimental interactions
with residues in the
gorge; iii. It decreases
the desolvation penalty
of hydrophilic monomers
by making them more
hydrophobic. - In
certain cases, very high
affinity of a monomer
suggests strong
interactions in a
certain conformation or
orientation. The
introduction of a tether
might impose
conformational
constraints that would
interfere with optimal
binding and, as a
consequence, reduce
affinity despite the
chelate effect. - The
stereochemical
configuration of a
pharmacophore is
critical not only for
the monomeric inhibitor
but also with regards to
fusion position with the
linker in bivalent AChE
ligands. - The
crystalline state of the
enzyme may generate
artifacts with
significant
repercussions. Every
structure should thus be
scrutinized and analyzed
meticulously to avoid
erroneous
interpretations.- Subtle
differences in the
dimensions of the
active-site gorge,
existence or absence of
H-bond donors/acceptors,
and conformational
restriction of side
chains leading to
rigidity are all
parameters that may
explain variations in
affinity for the same
ligand from one species
to another. Overall, use
of bivalent ligands has
proven to be an
extremely useful
approach for improving
the potency and
specificity of AChE
inhibitors, and crystal
structures of AChE with
various bivalent ligands
have highlighted many of
the underlying
principles governing the
mechanism of enzymatic
catalysis and inhibition
of AChE. A striking
example for the
effectiveness of this
approach is the recent
use of click chemistry,
utilizing the
active-site gorge of
Electrophorus electricus
AChE (EeAChE) as the
‘reaction vessel’ for
generating a
tacrine-phenylphenanthridinium
bivalent inhibitor that
exhibited 77 fM affinity
for TcAChE (410 fM
affinity for mAChE), and
forced the enzyme into a
unique conformation [65,
70]. Bifunctional
derivatives have also
been developed which
successfully fulfill the
dual tasks of inhibiting
the catalytic activity
of AChE and slowing the
rate of Ab aggregation
by the PAS [71, 72].
Other bifunctional
gorge-spanning ligands
have been developed
which serve as dual
inhibitors of AChE and
monoamine oxidase [73],
as dual anti-ChE and
anti-inflammatory agents
[74, 75], or as dual
anti-ChE and protectants
against reactive oxygen
species [76]. As
mentioned in the
Introduction, (–)- HupA
acts as a
neuroprotectant against
a variety of agents,
whether as an
N-methyl-D-aspartate (NMDA)
receptor antagonist or
by other mechanisms [8].
Thus bifunctional
agents, in which (–)-HupA
and (–)-HupA fragments
serve as building
blocks, may not only
yield novel AChE
inhibitors
with improved
pharmacological profiles
for treatment of AD and
other dementias, but may
also find use as potent
and specific drugs in a
much broader context.
Printable
PDF Version
|
