Synthesis, biological evaluation, and computational studies of Tri- and tetracyclic nitrogen-bridgehead compounds as potent dual-acting AChE inhibitors and h H3 receptor antagonists

Research Article

pubs.acs.org/chemneuro

Synthesis, Biological Evaluation, and Computational Studies of Tri-

and Tetracyclic Nitrogen-Bridgehead Compounds as Potent Dual-

Acting AChE Inhibitors and hH₃Receptor Antagonists

Fouad H. Darras,^†Steﬀen Pockes,^†Guozheng Huang,^§Sarah Wehle,^§Andrea Strasser,^‡

Hans-Joachim Wittmann,^‡Martin Nimczick,^§Christoph A. Sotriﬀer,^§and Michael Decker*^,§

^†Lehrstuhl fur Pharmazeutische und Medizinische Chemie I, Institut fur Pharmazie, ^‡Lehrstuhl fur Pharmazeutische und Medizinische

̈

Chemie II, Institut fur Pharmazie, Universitat Regensburg, Universitatsstraße 31, D-93053 Regensburg, Germany

̈

^§Pharmazeutische und Medizinische Chemie, Institut fur Pharmazie und Lebensmittelchemie, Julius-Maximilians-Universitat

Wurzburg, Am Hubland, D-97074 Wurzburg, Germany

̈

S

* Supporting Information

ABSTRACT: Combination of AChE inhibiting and histamine H₃receptor

antagonizing properties in a single molecule might show synergistic eﬀects

to improve cognitive deﬁcits in Alzheimer’s disease, since both

pharmacological actions are able to enhance cholinergic neurotransmission

in the cortex. However, whereas AChE inhibitors prevent hydrolysis of

acetylcholine also peripherally, histamine H₃antagonists will raise

acetylcholine levels mostly in the brain due to predominant occurrence of the receptor in the central nervous system. In this

work, we designed and synthesized two novel classes of tri- and tetracyclic nitrogen-bridgehead compounds acting as dual AChE

inhibitors and histamine H₃antagonists by combining the nitrogen-bridgehead moiety of novel AChE inhibitors with a second

N-basic fragment based on the piperidinylpropoxy pharmacophore with diﬀerent spacer lengths. Intensive structure−activity

relationships (SARs) with regard to both biological targets led to compound 41 which showed balanced aﬃnities as hAChE

inhibitor with IC₅₀= 33.9 nM, and hH₃R antagonism with K_i= 76.2 nM with greater than 200-fold selectivity over the other

histamine receptor subtypes. Molecular docking studies were performed to explain the potent AChE inhibition of the target

compounds and molecular dynamics studies to explain high aﬃnity at the hH₃R.

KEYWORDS: Acetylcholinesterase inhibitors, histamine H₃antagonists, Alzheimer’s disease, docking, MD simulation

The histamine H₃receptor is postulated as a G protein-

lzheimer’s disease (AD) is an irreversible, progressive

A_{neurodegenerative brain disorder and the most common}coupled receptor (GPCR) drug target for several cognitive

disorders including attention deﬁcit hyperactivity disorder

(ADHD), AD, and schizophrenia.⁶The H₃receptor had

originally been described as a presynaptic autoreceptor that

inhibits histamine release in the brain,⁷and was subsequently

shown also to regulate the release of other important

neurotransmitters such as acetylcholine, glutamate, serotonin,

dopamine, norepinephrine, gamma-aminobutyric acid (GABA),

and others via a parallel role as a heteroreceptor.^8,9Therefore,

blocking these receptors could be promising for potential

treatment of central nervous system (CNS) disorders such as

AD,⁹since blocking the central hH₃receptor can induce the

release of procognitive neurotransmitters like acetylcholine

(ACh). Due to their multiple therapeutic possibilities including

cognitive deﬁcits, there has been a very considerable and

remarkable progress in the development of hH₃R antagonists

both in academia and especially pharmaceutical industry.^8,10

Johnson & Johnson have reported on dual-acting H₃

antagonists and serotonin transporter inhibitors, and Hoﬀ-

form of dementia. It is estimated that about 6% of the

population worldwide aged over 65 currently suﬀer from AD.¹

Worldwide, AD aﬄicts more than 35 million people.^2,3The

occurrence of dementia increases exponentially with increasing

age, and it is expected to more than triple during the next 30

years.⁴This alarming prognosis demands further eﬀort to

elucidate the molecular mechanisms of AD and search for

eﬀective treatment methods.

The cholinergic hypothesis, proposed by Davies and

Maloney in 1976, is one of the earliest and most extensively

researched theories regarding AD pathogenesis.⁵It has led to

the introduction of tacrine as the ﬁrst drug to be approved for

the treatment of AD, and subsequently galantamine, donepezil,

and rivastigmine, respectively. All these compounds represent

inhibitors of acetylcholinesterase (AChE), the inhibition of

which increases the amount of the neurotransmitter acetylcho-

line (ACh) in the synaptic cleft. Clinically this leads to a

symptomatic, but signiﬁcant improvement of cognitive deﬁcits

in early stages of AD. Together with the symptomatically acting

N-methyl-^D-aspartate (NMDA) blocker memantine, these

compounds represent the only therapeutic treatment of AD

currently available.

Received: November 22, 2013

Revised: January 3, 2014

Published: January 14, 2014

225

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

mann-LaRoche as well as Hudkines et al. on H₃ligands with

piperidinylpropoxyphenyl pharmacophores (Chart 1).¹¹⁻¹⁴

compounds, is available. In some cases, it was also possible to

synthesize dual acting H₃/AChE compounds, but despite the

fact that the principle of combination seems to work, it has not

yet been possible neither to avoid large molecular weight

tacrine-containing compounds nor to describe compounds with

high aﬃnities in the same concentration range at both targets, a

core problem of hybrid and dual acting compounds albeit rarely

discussed and addressed in the literature.¹⁵Gaining balanced

aﬃnities is a very demanding task, since compounds can follow

completely diverse SARs at the diﬀerent targets to be

addressed.

Chart 1. Schematic Illustration of the Merging Design

Methodology for Synthesizing Dual Acting AChE inhibitors/

H₃Antagonist Hybrid Structures¹¹⁻¹³

Previous SARs on numerous H₃ligands described show that

the phenol ethers in meta position to tertiary benzyl amine

possess high H₃receptor binding aﬃnity and antagonistic

potency, for example, compounds 1, 2 (Chart 1); similarly, the

presence of an anilinic amine in para position to the phenolic

ether is favorable, e. g. compound 2 (Chart 1). Finally,

connecting the basic cyclic amine through an alkyl spacer of

variable length to a phenolic OH group of heterocyclic and/or

aromatic rings is a key feature to get H₃receptor binding

aﬃnity and antagonistic potency (Charts 1 and 2).

We have previously synthesized a series of novel tri- and

tetracyclic N-bridgehead AChE inhibiting lead structures with

moderate activity (i.e., two digit micromolar) characterized by a

phenolic hydroxyl group in the para position to a tertiary

anilinic and in the meta position to the basic N-bridgehead

atom and used them as starting points for the development of

novel butyrylcholinesterase selective inhibitors (compound 8)

(Chart 3).²⁴The cholinesterase inhibitory activities of these

The complexity and multiple etiologies of AD make single-

target strategies diﬃcult to achieve desirable therapeutic eﬀects

and make the development of multitarget-directed ligands

(MTDL) a more eﬀective strategy, as intensively applied in

recent years using very diverse biological targets.¹⁴⁻¹⁷

Recently, dual-acting AChE inhibitors and hH₃receptor

antagonists have been developed, since such multitarget-

directed ligands might be able to achieve a more speciﬁc

cognition-enhancing level of cholinergic neurotransmission

than either an AChE inhibitor or histamine H₃receptor

antagonist alone. For example, such dual-acting compounds

have been obtained from H₃receptor antagonists that

additionally contain a tacrine-based molecular structure.

These compounds act at hH₃, N-methyltransferase (HMT),

AChE, and butyrylcholinesterase (BChE), respectively.¹⁸The

most eﬀective compound FUB-833 4 (Chart 2) displayed good

Chart 3. Tri- and Tetracyclic Compounds As Starting Points

for H₃/AChE Dual Acting Compounds’ SARs²⁴

Chart 2. Previously Described Acetylcholinesterase

Inhibitors/Histamine H₃Receptor Antagonists^18,21−23

classes of compounds and their selectivity toward BChE were

dramatically improved by introducing a carbamate group at the

phenolic hydroxyl group (compound 9) (Chart 3).²⁴The

chemical features of these two moieties (tri- and tetracyclic)

served as the starting point for our SAR studies in order to

synthesize a novel series of dual acting compounds as AChE

inhibitors/H₃receptor antagonists due to their similarity to the

aromatic and/or heterocyclic parts of the above-described H₃

ligands (Chart 1) and then, combining the phenolic heterocycle

as an ether to tertiary amines.

With the aim to investigate SARs of this class of

cholinesterase inhibitors, we synthesized novel heterocyclic

templates consisting of basic tri- and tetracyclic N-bridgehead

molecules with phenolic ether connected to the basic cyclic

amine moieties by diﬀerent carbon spacer lengths. The ability

of these structural templates to antagonize selectively the H₃

receptor and the eﬀect of structural modiﬁcations on

cholinesterase inhibitory activity were intensively investigated

with the ultimate goal to ﬁnd structures with high aﬃnity/

activity at both targets (i.e., balanced activities).

The results of SARs at AChE were used for computational

docking studies of the inhibitors on AChE. Additionally, the

binding mode of one of the most potent target compounds at

the hH₃was investigated by molecular dynamics (MD) studies.

aﬃnity for hH₃receptors, AChE and HMT.¹⁸Despite these

exceptional activities, tacrine-based compounds always bear the

risk of severe hepatotoxicity, the reason why it was withdrawn

from the market.^14,19,20

Several groups have also reported on interesting dual AChE

inhibitors and H₃antagonists.²²⁻²⁴Unfortuantely, these

compounds act at much lower concentration (nM) on H₃

than on AChE (μM).

Signiﬁcant amount of knowledge with regard to structure−

activity relationships (SARs) of H₃antagonists, including the

group of piperidinylpropoxyphenyl pharmacophore containing

226

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

We intended to synthesize both tri- and tetracyclic

compounds, diﬀerent spacer lengths, diﬀerent basicities within

the heterocycle, various structures of the tertiary amines, and

related structural features to cover systematically a range of

chemical structures for investigating SARs at both targets.

by reduction of the amide group of compounds 17−21 using

LiAlH₄to give compounds 22−26.

Tricyclic target compounds were synthesized by a related

synthetic procedure as for the tetracyclic compounds as shown

in Scheme 2. Solventless fusion of pyrrolidin-2-one with

CHEMISTRY

Scheme 2. Synthesis of Tricyclic N-Bridgehead Compounds

(quinazolinones and diamines)

■

a

Scheme 1 shows the synthesis of the tetracyclic target

compounds. The starting material 6-(benzyloxy)-1H-benzo[d]-

Scheme 1. Synthesis of Tetracyclic N-Bridgehead

a

Compounds

a

Reagents: (i) MW, 200 W, 220 °C, 1 h; (ii) H₂, Pd/C, ethanol, rt, 24

h; (iii) 1,3-dibromopropane or 1,6-dibromohexane, K₂CO₃, KI,

acetonitrile, 90 °C, 3 h; (iv) pyrrolidine, piperidine, or azepane,

K₂CO₃, KI, acetonitrile, 90 °C, 3 h; (v) LiAlH₄, THF, 70 °C, 3 h.

compound 11 under microwave conditions with 200 W

irradiation power at 220 °C for 1 h gave compound 27,

which was then debenzylated by catalytic hydrogenation over

Pd/C in ethanol at room temperature to give compound 28.

Reaction of compound 28 with 1,3-dibromopropane or 1,6-

dibromohexane in a Williamson ether synthesis gave com-

pounds 29 or 30, respectively. Reaction of diﬀerent cyclic

amine with compounds 29 or 30 under similar reaction

conditions as for compounds 17−21 yielded the series of

tricyclic quinazolinone target compounds 31−34. Tricyclic

compounds 35−38 were synthesized using the same reduction

conditions used for the preparation of tetracyclic compounds

22−26 (LiAlH₄).

a

Reagents: (i) CH₃I, N,N-diisopropylethylamine, N,N-dimethylaceta-

mide, 40 °C, 24 h; (ii) toluene/reﬂux, 24 h; (iii) H₂, Pd/C, ethanol, rt,

24 h; (iv) 1,3-dibromopropane or 1,6-dibromohexane, K₂CO₃, KI,

acetonitrile, 90 °C, 3 h; (v) piperidine, morpholine, pyrrolidine, or

azepane, K₂CO₃, KI, acetonitrile, 90 °C, 3 h; (vi) LiAlH₄, THF, 70 °C,

3 h.

[1,3]oxazine-2,4-dione 11 was synthesized in four steps with

almost quantitative overall yield, starting from 2-amino-5-

hydroxybenzoic acid. The resulting isatoic anhydride represents

an activated form of anthranilic acid with the benzylated

phenolic−OH, the benzylation of the phenolic group in

compound 11 is necessary to avoid methylation of the phenolic

group, and to improve the solubility of the resulting isatoic

anhydride.²⁵

Scheme 3 shows the reaction conditions for reduction of the

imine group of compounds 32 and 34 to the corresponding

Scheme 3. Selective Reduction of Quinazolinone Imine

Bond to the Corresponding Amine

10-Hydroxy-13-methyl-13,13a-dihydro-5H-isoquinolino[1,2-

b]quinazolin-8(6H)-one 14 was synthesized as shown in

Scheme 1: N-Methylation of 6-(benzyloxy)-1H-benzo-[d][1,3]-

oxazine-2,4-dione 11 in the presence of N,N-diisopropylethyl-

amine as a base and iodomethane as an alkylating agent in N,N-

dimethylacetamide as a solvent aﬀorded compound 12.²⁶

Compound 12 was fused with 3,4-dihydroisoquinoline in

toluene under reﬂux overnight to aﬀord compound 13 in an

acceptable yield (40%). After benzyl deprotection by hydro-

genation over Pd/C in ethanol at room temperature,

compound 14 was obtained in quantitative yield. Reaction of

compound 14 with 1,3-dibromopropane or 1,6-dibromohexane

in the presence of potassium carbonate as a base with a catalytic

amount of potassium iodide in acetonitrile as solvent for 3 h

under reﬂux yielded compounds 15 or 16, respectively. The last

synthetic step was nucleophilic substitution by reaction of

bromo spacer compounds 15 or 16 with diﬀerent cyclic amines

using again potassium carbonate as a base with a catalytic

amount of potassium iodide in acetonitrile as a solvent to yield

the ﬁrst series of quinazolinone target compounds 17−21. The

second set of N-bridgehead target compounds was synthesized

amines using sodium cyanoborohydride in ethanol at room

temperature for 24 h to yield compounds 39 and 40,

respectively.²⁷

The carbonyl group of quinazolinones (compounds 32 and

34) could be selectively reduced using Zn metal powder in the

presence of acetic acid as a solvent to give compounds 41 and

42, respectively. The reaction mixture was then heated at 70 °C

for 2 h after a few drops of conc. hydrochloric acid had been

added as shown in Scheme 4.²⁸

RESULTS AND DISCUSSION

■

A novel and potent class of dual acting AChE inhibitor/H₃

antagonist hybrid compounds was developed by connecting the

tri- and tetracyclic N-bridgehead structural templates via a

227

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

molecular docking studies (see below) with regard to the

orientation of the molecule within the enzyme.

Scheme 4. Selective Reduction of Quinazolinones by

Clemmensen Reduction

As we had planned from the compound design, H₃binding

data for these quinazolinone compounds with an amide

carbonyl group (compounds 17−21) (Scheme 1), showed

binding aﬃnity in the micromolar range. The corresponding

reduced tetracyclic compounds (23, 25, and 26) (Scheme 1)

displayed a moderate inhibitory activity toward AChE in the

micromolar range, while this set of compounds showed high

binding aﬃnity at H₃in two digit nanomolar range (K_i(hH₃) =

57.7, 46.9, and 17.5 nM respectively) (Table 1).

Interestingly, the tetracyclic compounds connected with

morpholine as a basic center showed poor cholinesterase

inhibition aﬃnities, compound 19 with IC₅₀(EeAChE) = 29.81

μM and IC₅₀(BChE) = 8.5 μM, and its corresponding reduced

form compound 24 with IC₅₀(EeAChE) = 13.1 μM and IC₅₀

(BChE) = 27.5 μM. Both compounds showed also poor

binding aﬃnities toward the H₃receptor (Table 1).

We also evaluated the inﬂuence of the carbon spacer length

on the biological activity for the tetracyclic moiety by

connecting the tetracyclic moiety with piperidine spaced by

six carbon atoms (compound 17), and the corresponding

reduced form (compound 22). Compound 22 showed

inhibitory activity in submicromolar range (IC₅₀(EeAChE) =

0.57 μM) toward AChE, and binding toward hH₃R in the same

range (K_i(hH₃) = 0.64 μM). In contrast, the corresponding

three carbon atom spacer length compound 23 showed also

submicromolar inhibition activity range toward AChE (IC₅₀

(EeAChE) = 0.53 μM), but signiﬁcantly improved the binding

aﬃnity toward hH₃R (K_i(hH₃) = 57.7 nM). So, a three carbon

spacer length is better for H₃aﬃnity with regard to the

tetracyclic moiety (Table 1).

hydroxyl group, in para position to anilinic N, with diﬀerent

cyclic amines spaced by three or six carbon atom linkers. A

detailed look at the respective SARs reveals several interesting

insights, also of importance to both the development of AChE

inhibitors as well as H₃antagonists individually.

All target compounds were tested for their ability to inhibit

acetylcholinesterase (EeAChE, EC.3.1.1.7 from electric eel) and

butyrylcholinesterase (BChE, EC.3.1.1.8 from equine serum).

Both enzyme isoforms exhibit very high amino acid sequence

homology with regard to the human ones (88% and 90%,

respectively).^29,30Inhibitory activity at BChE, a plasma-soluble

isoform of AChE, has recently gained attention for dementia-

related drug development since the amount of AChE decreases

during the course of the disease, whereas this seems not to be

the case for BChE. Therefore, BChE inhibition might

compensate for the lack of eﬃciency of administration of

AChE inhibitors in later stages of AD.²⁴

Additionally, selected compounds were tested for their

inhibitory activity on human acetylcholinesterase hAChE

(E.C.3.1.1.7, acetylcholinesterase from human erythrocytes)

to conﬁrm binding proﬁles and be able to compare the data

with the docking results on the human enzyme.

AChE Inhibitor/H₃Antagonist Tetracyclic Hybrid

Compounds. With regard to the tetracyclic template, we

had previously synthesized a series of potent and selective

carbamate-based (pseudo)irreversible BChE inhibitors. The

free phenolic tetracyclic compound 8 displayed a low inhibitory

activity at both cholinesterases (IC₅₀(EeAChE) = 110.3 μM,

IC₅₀(BuChE) = 44.3 μM) (Table 1). The low activity was

attributed to low basicity and poor solubility of this tetracycle

moiety. However, in this project, the same template

unexpectedly showed submicromolar inhibitory activity toward

AChE and one digit micromolar inhibitory activity toward

BChE, simply due to connection with diﬀerent cyclic amines

spaced by three carbon atoms (18, 20, and 21). For example,

compound 18 (IC₅₀(EeAChE) = 0.55 μM) showed 200-fold

higher inhibitory activities than those for the free phenolic

tetracyclic lead structure compound 8 (Chart 3) (IC₅₀

(EeAChE) = 110.3 μM). Interestingly, the morpholino

compound 19 has an IC₅₀(EeAChE) = 29.81 μM and IC₅₀

(BChE) = 8.5 μM, being the only compound with higher BChE

than AChE inhibitory activity in this set of compounds (Table

1), albeit with lower inhibitory activity at AChE.

According to our previously published work, we expected

that the inhibitory activity of this heterocycle will increase by

amide reduction through increasing the basicity of the core

ring, but surprisingly, the AChE inhibition activity of this set of

the tetracyclic compounds (22−26) is lower than that of the

same set of compounds before reduction (17−21), but still in

the micromolar range (from 0.53 μM for compound 23 to 19.8

μM for compound 25). Therefore, selectivity and activity

proﬁles for these sets of compounds depend more on the

structure of the alicyclic N-ring connected to the heterocycle

than the heterocycle itself. We investigated this feature in the

AChE Inhibitor/H₃Antagonist Tricyclic Hybrid Com-

pounds. The tricyclic quinazolinone compound 10 (Chart 3)

showed moderate inhibitory activities toward both enzymes

(IC₅₀(EeAChE) = 82.5 μM and IC₅₀(BChE) = 25.1 μM).³¹

Surprisingly, the same template showed two digit nanomolar

inhibitory activities toward AChE and one digit micromolar

toward BChE by connecting it with diﬀerent cyclic amines

spaced by three C atoms (compounds 31, 32, and 33 depicted

in Scheme 2). This fact and the fact that these compounds did

not show high binding aﬃnity toward the H₃receptor

(compound 32 with K_i(hH₃) = 1.56 μM) (Table 1) intensively

made us study the binding mode of this class of compounds

toward AChE in a separate project by synthesis of additional

compounds and computational studies. Results will be reported

separately.

We also evaluated the inﬂuence of the spacer length for this

moiety, and again three carbon spacer length is better for

achieving dual acting compound from the tricyclic moiety,

when compound 32 with three carbon spacer length showed

IC₅₀(EeAChE) = 69.2 nM and (hH₃) = 1.6 μM, while the same

moiety with six carbon spacer length compound 34 showed

IC₅₀(EeAChE) = 390 nM and (hH₃) = 0.92 μM (Table 1).

For this moiety, we evaluated the eﬀect and biological

consequences of the amide oxygen, the double bond, and the

resulting basicity diﬀerences of the core ring of tricyclic

quinazolinone compounds (31−34). First, total reduction of

the core carbonyl and imine groups using LiAlH₄aﬀorded the

diamine class of compounds 35−38; these amines showed

good inhibitory activity toward AChE (submicromolar to one

digit micromolar range) and showed a high binding aﬃnity at

the H₃receptor (two digit nanomolar range) for compounds

228

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

Table 1. Chemical Structures, ChE Inhibition Results, and hH₃R Binding Data for Tri- and Tetracyclic Synthesized Compounds

229

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

Table 1. continued

a

b

AChE, BChE: data are the means of three independent determinations. hH₃R binding aﬃnity: data are means standard deviation of triplicate

c

independent experiments. n.d.: Not determined.

35, 36, and 37 (K_i(hH₃) = 31.9, 37.0, 36.0 nM), respectively.

Second, reducing the imine group in the quinazolinone core

ring using NaBH₃CN yielded its corresponding amine (39, 40),

but these compounds lost their AChE inhibitory activity and

showed very weak binding at H₃receptor (compound 39 with

IC₅₀(EeAChE) = 6.92 μM and (hH₃) = 2.3 μM, and

compound 40 with IC₅₀(EeAChE) = 3.33 μM and (hH₃) = 1.4

μM). Third, selective reduction of the carbonyl group was

achieved by Zn/AcOH reduction (Clemmensen reaction) and

left the imine double bond as it is (41, 42). This set of

compounds showed a balanced activity toward our two targets;

they act as balanced acetylcholinesterase inhibitors/hH₃

antagonists, with compound 41 showing IC₅₀(hAChE) = 34

nM, IC₅₀(EeAChE) = 67 nM, and K_i(hH₃) = 76 nM. Similarly,

compound 42 showed IC₅₀(hAChE) = 39 nM, IC₅₀(EeAChE)

= 88 nM, and K_i(hH₃) = 113 nM (Table 1).

understand the mode of inhibition and binding of this inhibitor

at EeAChE. The mechanism of inhibition was analyzed by

recording substrate−velocity curves in the presence of diﬀerent

concentrations of compound 41 using 50−450 μM acetylth-

iocholine (ATC) substrate EeAChE. Figure 1 shows the

Balanced activity, that is, compound activity and aﬃnity,

respectively, at the relevant targets in the same or at least a

similar concentration range, is a therapeutically highly relevant

issue, that is very rarely properly addressed in research eﬀorts

on hybrid and multitarget compounds. This is not surprising,

since two, often distinct and independent, SARs have to be

investigated at each target, and even if a balanced activity is

reached for one compound, it is rarely the highest aﬃnity/

activity of the set of compounds described.

Figure 1. Lineweaver−Burk plot resulting from substrate−velocity

curves of EeAChE activity with diﬀerent substrate (ATC) concen-

trations (50−450 μM) in the presence of 1, 10, 50, 100, and 200 nM

compound 41.

In general, all synthesized target compounds represent

moderate or potent AChE inhibitors with inhibitory activities

ranging from 30 μM (compound 19) down to the two digit

nanomolar range (67 nM for compound 41) with roughly 500-

fold higher activity. This compound is in the range of the “gold-

standard” of AChE inhibition, tacrine.

Aﬃnity at the H₃receptor turned out to be very high as

planned and desired from the compounds’ design. Since none

of our structures had been tested for H₃aﬃnity before, not

even our ChE inhibiting lead structures, this fact proved our

assumptions concerning ligand design. With regard to the H₃

aﬃnities, they ranged from the (this time one digit) micromolar

range (K_i(hH₃) = 2.3 μM for compound 39) to the nanomolar

range (K_i(hH₃) = 18 nM for compound 26) covering a range

of 2 orders of magnitude. Not surprisingly, SARs follow

diﬀerent routes than for AChE inhibition as described in detail

above.

Lineweaver−Burk plot (reciprocal rates versus reciprocal

substrate concentrations for the diﬀerent inhibitor concen-

trations resulting from the substrate−velocity curves for

EeAChE). In this Lineweaver−Burk plot, the K_mvalues diﬀer

with the diﬀerent inhibitor concentrations, while the v_maxvalue

remains unchanged. The Lineweaver−Burk plot for EeAChE

shows reversible and competitive inhibition by compound 41.

Aﬃnity toward the human histamine H₃receptor was

determined by displacement of speciﬁc [³H]-histamine binding

to H₃receptor binding sites in Sf9 insect cell membranes³²

(Table 1).

Putative aﬃnity at other hH receptor subtypes, that is, H₁,

H₂, and H₄, respectively, was determined by testing high

concentration (10 μM) of the most potent compounds at hH₃R

for displacement of [³H]mepyramine (hH₁R), [³H]UR-DE257

(hH₂R),³³and [³H]-histamine (hH₄R) (Figure 2). Generally,

all of the tested compounds exhibited a very good selectivity

proﬁle, since none of the tested compounds have aﬃnity at

hH₄R and hH₂R even at a concentration of 10 μM. At 10 μM,

some of the tested compounds showed to slightly interact with

hH₁R (Figure 2).

PHARMACOLOGY

In addition to determination of inhibitory activities, compound

41 was selected for kinetic measurements, in order to

■

230

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

the latter, cation−π interactions are formed with mostly Trp86

or Tyr337 (hAChE numbering), residues known for attracting

the tertiary amine of the natural substrate acetylcholine.²¹

The ligands were docked in their expected ionization state at

pH 8. This means protonation at the tertiary amine of all

compounds, consistent with their pK_avalues between 10.13 and

9.58 as calculated with MoKa.³⁷A pK_aof 7.66 was calculated

for the morpholine substituent. The aminal structure in

compounds 35−38 is protonated due to a calculated pK_aof

8.73. Similarly, the amidine moiety was protonated in the

inhibitors 41 and 42 with a calculated pK_aof 10.72 for both

scaﬀolds resulting in double protonation of these compounds.

A very clear preference for the inverted binding mode is

shown by the most potent compounds, for which good

clustering of the docking results in the top poses was obtained.

This includes the nonchiral tricyclic inhibitors 32, 34, 41, and

42, but also the chiral tricyclic compounds 36 and 38, which are

closely related to 41 and 42, respectively (Figure 4). The

quinazolinone compounds 31 and 33 also ﬁt excellently in

here; the binding as well as the docking results for a whole set

of quinazolines will be presented in a subsequent publication.

The piperidinyl ring of 41 and 42 is placed near the key residue

of the choline binding site, Trp86. The alkyl chains point away

from the water molecules in the binding site, so that the

tricyclic ring systems are placed near the peripheral anionic site

(PAS) interacting with the Tyr72 hydroxyl group or the

backbone carbonyl of Tyr341. All of the mentioned compounds

(except 34: 3.8) achieve a per-atom-score (GoldScore/no.

heavy atoms) of >4, indicating inﬂuential contributions of most

atoms to the overall interaction. In contrast, the weaker

inhibitors generally show lower per-atom-scores and/or

inconsistent binding-orientation preferences for the two

enantiomeric forms. Yet, for the most active tetracyclic

compounds with a carbonyl function (17 and 21) the inverted

binding mode appears to be preferred over the classical binding

mode for both enantiomers with comparable score. In this

binding mode, a 3.0−3.6 Å hydrogen distance bond can be

observed between the carbonyl oxygen and a conserved water

molecule (Figure 4).

Figure 2. Binding assay results toward hH₁R, hH₂R, and hH₄R,

applying a 10 μM concentration of the most potent hH₃R antagonist

compounds. Diphenhydramine (DPH) as hH₁R standard ligand and

histamine (HIS) as hH₂R, and hH₄R standard ligand.

Determination of whether the synthesized compounds

possess antagonist, agonist, or inverse agonist properties was

investigated by steady-state GTPase activity assay. The

functional assay (GTPase) was performed at 10 μM

concentration of all synthesized compounds. As a reference

hH₃R antagonist, 10 μM concentration of thioperamide was

used (Figure 3). Almost all the tri- and tetracyclic synthesized

The docking results suggest that the inverted binding mode

is favored for most of the inhibitors tested in this study.

Especially for strong inhibitors (IC₅₀< 0.1 μM), the scores

show a clear preference for this binding mode. The main

interactions consist of a cation−π interaction with Trp86 and

hydrogen bonds with conserved water molecules. Weaker

inhibitors show lower per-atom-scores and in most cases less

clear preference of the inverted binding mode versus the

classical binding mode for the two enantiomeric forms.

Figure 3. Functional assay (GTPase) results at the hH₃receptor

applying a 10 μM concentration of all synthesized compounds with

water as a baseline and thioperamide as control (antagonist).³⁴

compounds are hH₃R antagonists or inverse agonists; only

compound 24, which is connected with morpholine by three

carbon spacer length, showed slight agonist properties at hH₃R.

THE BINDING MODE OF COMPOUND 41 AT hH₃R

■

The molecular dynamics (MD) simulation revealed a stable

binding mode of the double protonated form of compound 41

(the most potent compound at both targets investigated) in the

binding pocket of the inactive hH₃R (Figures 5, 6A). During

the whole productive phase of the MD simulation, a stable

electrostatic interaction between the positively charged

piperidine moiety of compound 41 and the highly conserved

Asp^3.32was observed (Figure 5). The mean distances between

the hydrogen of the piperidine nitrogen of compound 41 and

the oxygens of the Asp^3.32side chain were fairly well maintained

over time at about 0.22 nm (Figure 6A). A second stable

interaction during the MD simulation was observed between

Glu^5.46and the positively charged amidine moiety of 41

COMPUTATIONAL DOCKING STUDIES ON HUMAN

AChE

■

Docking studies were carried out on AChE to investigate the

putative preferred binding mode of the compounds in the

inhibitor library. Docking studies with the structure of human

AChE (PDB 4EY7)³⁶and consideration of conserved water

molecules (cf. Methods) suggest two possible modes of

inhibition: a “classical” binding mode corresponding to the

placement of the tri- or tetracyclic scaﬀold near the catalytic

active site (CAS) at the bottom of the binding gorge; and an

“inverted” binding mode showing the protonated tertiary amine

instead of the aromatic scaﬀold deep in the binding gorge. In

231

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

Figure 4. (A) Docking pose showing the inverted binding mode of inhibitor 41 (cyan) with a 4.2 Å cation−π interaction to Trp86 and potential

hydrogen bonds to a water molecule (3.2 Å) and Tyr72 (3.4 Å). Inhibitor 42 (green) shows the same placement of the piperidine ring near the CAS

(4.2 Å to Trp86) and a distance of 4.3 Å to the Tyr341 backbone carbonyl. (B) Selected docking pose of the tetracyclic inhibitor 21 (pink, R-

enantiomeric form) showing the inverted binding mode. The inhibitor is stabilized by a 3.0 Å interaction of the scaﬀold carbonyl oxygen with a water

molecule and a 4.3 Å cation−π interaction of the piperidine with Trp86 near the CAS. Color code: red, conserved water molecules; gray, hAChE

binding site; catalytic active site, Ser203, His447 (Glu334 not shown here); peripheral anionic site (PAS), Tyr72, Trp286. Both ﬁgures were created

using Pymol.³⁵

Figure 5. Compound 41 embedded in the binding pocket of the inactive hH₃R, obtained as a snapshot of MD simulation. The translucent surface of

the ligand is colored according to the electrostatic potential.

(Figures 5, 6B). Here, the mean distance between the hydrogen

and the oxygens of Glu^5.46is about 0.20 nm.

The observed binding mode based on modeling studies of

compound 41 at hH₃R is similar to the binding mode of the

hH₃R ligands FUB836,³⁸imoproxifan^39,40(Chart 4), and

compound 6⁴¹(Chart 2). The binding mode described for

compound 6 at hH₃R is very similar to that we observed for

compound 41 within this study: both positively charged

moieties were observed to interact electrostatically with Asp^3.32

The tricyclic partial structure of compound 41 remained

stable in a pocket formed by the amino acids Leu^3.29, Tyr^3.33

,

Cys^3.36, Trp^6.48, Tyr^6.51, Met^6.55, and Phe¹⁹³(E2-loop) (Figure

5). Furthermore, the hydrophobic part of the piperidine moiety

of compound 41 is in close contact to Tyr^2.61and Leu^7.42

(Figure 5). Besides, the MD simulations revealed that water

molecules occupy parts of the binding pocket in presence of

compound 41. An aromatic interaction was observed between

and Glu^5.46

.

The positively charged imidazole moiety of imoproxifan was

found to interact electrostatically with Asp^3.32, and the

remaining part of imoproxifan, including the aromatic moiety,

was embedded in a pocket, similar as described for compound

33

Tyr^3.and the aromatic partial structure of compound 41.

232

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

Figure 7. Comparison of selected amino acids, located in the binding

pocket of the four human histamine receptor subtypes; green, highly

conserved amino acids within all subtypes or identical amino acids

between hH₃R and hH₄R, red: amino acids being diﬀerent between

hH₃R and hH₄R.

are known, for example, proxifan, thioperamide, UR-PI294, and

UR-AK381.⁴³Arylbenzimidazols were described to show

aﬃnity to hH₄R.⁴⁵The 1-methyl-4-(3-phenoxypropyl) piper-

azine partial structure of these compounds is similar to the 1-

(3-phenoxypropyl) piperidine partial structure of compound

41. However, a reason for selectivity of compound 41 toward

hH₃R (compared to hH₄R) may be the exchange of the N-

methylpiperazine moiety into a piperidine moiety. The N-

methylpiperazine moiety is typical for a large number of H₄R

ligands,⁴⁶its basic amine is relevant for interaction with Asp^3.32

,

Figure 6. (A) Time evolution of the distances between the hydrogen

of the protonated piperidine of 41 and the oxygens of the Asp^3.32side

chain during the 10 ns productive MD. (B) Time evolution of the

distances between the hydrogen of the protonated amidine of 41 and

the oxygens of the Glu^5.46side chain during the 10 ns productive MD.

and it is located at a diﬀerent position compared to the

piperidine. It might be suggested that the sequence diﬀerence at

position 7.42 (hH₃R, Leu; hH₄R, Gln) (Figure 7) would be

responsible, since this amino acid side chain is in direct

neighborhood to compound 41. For the hH₃R and rH₃R, it was

shown that species diﬀerences between hH₃R and rH₃R at

Chart 4. FUB836³⁸and Imoproxifan^39,40

positions 3.37 and 3.40 (hH₃R: Thr^3.37, Ala^3.40; rH₃R: Ala^3.37

,

Val^3.40) are responsible for pharmacological diﬀerences of

improxifan.³⁹Thus, it could also be speculated that the

diﬀerences in amino acids between hH₃R and hH₄R at position

3.40 (Figure 7) may be involved in subtype diﬀerences.

However, in order to obtain a more detailed insight into the

interactions with hH₃R, site-directed mutagenesis studies

combined with further molecular modeling studies have to be

performed.

In summary, merging the piperidinylpropoxy phenyl and

related hH₃pharmacophores into heterocyclic templates with

AChE inhibiting properties yielded reversible and competitive

AChE inhibitors with antagonizing properties at hH₃R. SARs

led to highly potent compounds with balanced two digit

nanomolar aﬃnities at both targets. The compounds show

excellent selectivity over all other hHR receptor subtypes.

Docking studies could explain the pronounced inﬂuence of the

tertiary amine on SARs onto AChE inhibitory activity by an

“inverted” binding mode locating the tertiary amine into the

CAS. MD simulation conﬁrmed the compound design strategy

for hH₃R aﬃnity by identifying a binding mode related to other

highly potent hH₃R antagonists.

41. Surprisingly, imoproxifan is described to act, in contrast to

compound 41, as an agonist at hH₃R.⁴⁰A reason for that

diﬀerence may be the imidazole moiety (as present in the full

agonist histamine) in imoproxifan, instead of the piperidine

partial structure in compound 41. The pharmacological data

suggest that the introduction of an amide moiety into the

tricycle leads to a decrease in aﬃnity to hH₃R, as shown by

direct comparison of compound 41 with compound 32 (Table

1). This may be explained by a sterical hindrance of the amide

with the hH₃R or with the diﬀerences in electronic distribution

between the two compounds. Phe¹⁹³, located in the E2-loop,

points into the binding pocket and shows an interaction with

the ligand. This is in good accordance to data obtained for the

histamine H₄receptor.⁴²For the histamine H₄receptor, it was

described that Phe¹⁶⁹, which corresponds to Phe¹⁹³in the

hH₃R, is responsible for diﬀerences in binding of agonists

between hH₄R and mH₄R.⁴²As presented in Figure 2,

compound 41 shows no aﬃnity to hH₁R, hH₂R, and hH₄R.

It is not surprising that compound 41 shows no aﬃnity to

hH₁R or hH₂R, because the structural proﬁle of this compound

METHODS

■

Construction of Inactive-State Model of hH₃R. The crystal

structure of hH₁R 3RZE⁴⁷representing the inactive conformation of

the hH₁R, was used as a template to construct the inactive state of

hH₃R. Lysozyme which was artiﬁcially cocrystallized in 3RZE was

deleted. The N-terminus, which is missing in the crystal structure, was

added using SYBYL 7.0 (Tripos Inc.). The “Loop-Search” module of

SYBYL 7.0 was used to complete the E2-loop. Due to the lack of

information concerning the conformation of the long I3-loop of the

hH₃R (more than 100 amino acids) the amino acids Ala²³⁹to Arg³⁴⁷

were not included in the model. To close the gap between the

intracellular parts of TM V and TM VI eight alanines were inserted

instead. Since internal water molecules were shown to play an

important role in stabilization or activation of aminergic GPCRs these

internal water molecules were included into the model of inactive

does not ﬁt to the typical hH₁R or hH₂R ligand structure.⁴³

A

comparison of the most important amino acids in the binding

pocket of the four human histamine receptor subtypes (Figure

7) shows that only the highly conserved Asp^3.32, Trp^6.48, and

Tyr^6.51are identical, which corresponds to an identity of only

20% (with respect to the 15 compared amino acids, Figure 7).

A complete comparison of the diﬀerent histamine receptor

subtypes and species can be found in literature.⁴⁴In contrast,

between hH₃R and hH₄R about 60% of the amino acids

(comparison of 15 amino acids, Figure 7) are identical. A large

number of ligands with aﬃnity or potency at hH₃R and hH₄R

233

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

hH₃R.^48,49Using SYBYL7.0 the compound 41 was docked manually

into the binding pocket of hH₃R in such manner that the positively

charged piperidine moiety interacts electrostatically with the highly

conserved Asp^3.32, similarly to previous studies addressing other H₃R

ligands.^38,40The tricyclic partial structure of compound 41 was

embedded in a pocket between TM III, TM V and TM VI. The

resulting complex was energetically minimized. Afterward, the

minimized ligand−receptor complex was embedded in a POPC lipid

bilayer. Intracellular and extracellular water, as well as an appropriate

number of sodium and chloride ions were added into the simulation

box and the molecular dynamics simulation, using GROMACS 4.0.2

(http://www.gromacs.org), was performed as described.^50,51The

parametrization for the ligand was obtained from the PRODRG server

(http://davapc1.bioch.dundee.ac.uk/prodrg/), but the partial charges

of 12 analyzed docking runs, all water molecules were accepted in the

“toggle”-mode and kept for generating ligand binding modes. In only

three cases, one water molecule, located at the entrance of the binding

site, was excluded from the docking process. Docking studies were

thus carried out with the selected conserved water molecules in the

water-“on” mode. For the donepezil ligand in 1EVE, the redocking

with water gave an rmsd of 0.52 Å for the pose most similar to the

crystal structure and 47 poses in the cluster. For the ligand in 4EY7, an

rmsd of 0.35 Å was found (49 poses in the cluster).

Enzyme Inhibition. Acetyl- and Butyrylcholinesterase Inhibition

Assay. The assay has been previously described in detail:^14,24AChE

(E.C.3.1.1.7, Type VI-S, from electric eel), BChE (E.C.3.1.1.8, from

equine serum), and hAChE (E.C.3.1.1.7, acetylcholinesterase from

human erythrocytes) were purchased from Sigma-Aldrich (Steinheim,

Germany); DTNB (Ellman’s reagent), ATC, and BTC iodides were

obtained from Fluka (Buchs, Switzerland).

were adopted based on the Gasteiger-Huckel partial charges, calculated

̈

by SYBYL 7.0. For the POPC lipid, the force ﬁeld parameters available

at an appropriate online source (http://moose.bio.ucalgary.ca/index.

php?page=Structures_and_Topologies) were used. Subsequent to a 5

ns equilibration phase (force constants (250 kJ/(mol nm²) for the ﬁrst

2.5 ns) and 100 kJ/(mol nm²) for the second 2.5 ns were put onto the

backbone atoms of the TM domains of hH₃R), a 10 ns productive

phase was performed.

The assay was performed as described in the following procedure:

stock solutions of the tested compounds were prepared in ethanol/

water (1:4 ratio), 100 μL of which gave a ﬁnal concentration of 10⁻³M

when diluted to the ﬁnal volume of 3.32 mL. The highest

concentration of the tested compounds applied in the assay was

10⁻⁴M (10% EtOH in the stock solution did not inﬂuence enzyme

activity). In order to obtain an inhibition curve, at least ﬁve diﬀerent

concentrations (normally 10⁻⁴−10⁻⁹M) of the test compound were

measured at 25 °C and 412 nm, each concentration in triplicate.

For buﬀer preparation, 1.36 g of potassium dihydrogen phosphate

(10 mmol) was dissolved in 100 mL of water and adjusted with NaOH

to pH = 8.0 0.1. Enzyme solutions were prepared to give 2.5 units

mL⁻¹in 1.4 mL aliquots. Furthermore, 0.01 M DTNB solution and

0.075 M ATC and BTC solutions were used. A cuvette containing 3.0

mL of phosphate buﬀer, 100 μL of the respective enzyme, and 100 μL

of the test compound solution was allowed to stand for 4.5 min, then

100 μL of DTNB was added, and the reaction was started by addition

of 20 μL of the substrate solution (ATC/BTC). The solution was

mixed immediately, and exactly 2.5 min after substrate addition the

absorption was measured. For the reference value, 100 μL of water

replaced the test compound solution. For determining the blank value,

additionally 100 μL of water replaced the enzyme solution. The

inhibition curve was obtained by plotting the percentage enzyme

activity (100% for the reference) versus logarithm of test compound

concentration.

Radioligand Binding Assays. For the binding experiments, three

diﬀerent batches of Sf9 insect cell membranes were generated and

prepared as described in literature.^33,56−59The accordant membranes

were thawed and sedimented by a 10 min centrifugation at 4 °C and

13 000g. Membranes were resuspended in binding buﬀer (12.5 mM

MgCl₂, 1 mM EDTA, 75 mM Tris/HCl, pH 7.4). Each well (total

volume 100 μL) contained 40 μg (hH₁R), 60 μg (hH₂R), 50 μg

(hH₃R), or 90 μg (hH₄R) of protein. Competition binding

experiments were performed in attendance of 5 nM [³H]mepyramine

(K_D= 4.49 nM)⁵⁶(hH₁R) (American Radiolabeled Chemicals, St.

Louis, MO), 30 nM [³H]UR-DE257 (K_D= 31.33 nM)³³(hH₂R), or

15 nM [³H]histamine (K_D= 39.53 nM (hH₃R); K_D= 15.88 nM

(hH₄R))⁶⁰(American Radiolabeled Chemicals, St. Louis, MO),

increasing concentrations of unlabeled ligands in hH₃R binding assay

and for hH₁R, hH₂R, and hH₄R experiments, a single concentration

(10 μM) of unlabeled ligand was used. The assays were conducted in

96-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany)

and incubated for 60 min at 25 °C under shaking at 250 rpm using a

Heidolph Titramax 101 instrument (Heidolph Instruments GmbH &

Co. KG, Schwabach, Germany). Bound radioligand was separated

from free radioligand by ﬁltration through GF/C ﬁlters, which were

pretreated with 0.3% polyethyleneimine solution. Subsequently, three

washing steps with cold binding buﬀer (4 °C) using a Brandel

Harvester instrument (Brandel, Gaithersburg, MD) were followed.

Filter-bound radioactivity was detected after an equilibration phase of

at least 12 h by liquid scintillation counting in 96-well plates

(PerkinElmer, Waltham, MA).

Computational Docking of Inhibitors to hAChE Binding Site.

Docking was carried out with GOLD v5.1 and GOLD v5.2

(GOLDSUITE 5.1 and 5.2, CCDC Software, http://www.ccdc.cam.

ac.uk)⁵²in the 4EY7 crystal structure of human AChE. The protein

was set up for docking by removing all water molecules and

nonprotein atoms and protonation of the amino acids was performed

with MOE 2011.10⁵³using Protonate 3D at standard conditions. The

search region was focused on the binding area comprising the CAS,

the binding gorge, and the PAS. The binding site was deﬁned by the

following residues: Tyr72, Asp74, Trp86, Asn87, Gly120, Gly121,

Tyr124, Ser125, Gly126, Ala127, Tyr133, Glu202, Ser203, Trp236,

Trp286, Phe295, Phe297, Glu334, Tyr337, Phe338, Tyr341, His447,

Gly448, And Tyr449. The ligand structures were either built in MOE

(compounds 17−26, 32 and 34−42 and the chiral compounds in their

R- and S-forms) or extracted from the crystal structure (ligands for

redocking experiments, cf. below). The protonation for the ligands was

set according to calculations performed with the program MoKa.³⁷

The inhibitor structures were energy minimized in MOE with the

MMFF94x force ﬁeld to an rms-gradient of 0.001 kcal/(mol·Å). Using

default settings in GOLD, the best-suited scoring function was selected

based on redocking experiments in the Torpedo californica AChE

binding site, with the galanthamine derivative of PDB structure 1W4L

and donepezil of 1EVE. Another redocking experiment was performed

using the donepezil ligand of 4EY7 and (−)-huperzine A of 4EY5,

both human AChE structures. With each of the four scoring functions

available in GOLD (ASP, CHEMPLP, ChemScore, GoldScore), 50

ligand poses were generated with the default number of operations of

100 000. In all cases, the top-ranked docking pose was deviating less

than 1.07 Å with GoldScore.^54,55By raising the number of operations

in the GOLD GA settings to 500 000, thereby prolonging the

optimization time, the top-pose showed a signiﬁcantly lower rmsd to

the crystal pose of 0.55 Å (compared to 1.04 Å with GA 100 000) for

redocking using 4EY7 and 0.63 Å for 1EVE (compared to 0.65 Å with

GA 100 000). In all cases, a top-ranked docking pose deviating less

than 0.91 Å from the crystal structure was obtained with GoldScore.

Based on a docking study for AChE selective compounds [manuscript

in preparation], the GoldScore function^54,55was best able to reﬂect the

aﬃnity and, hence, chosen for this project, too. All docking poses were

clustered with a 1.5 Å cluster-cutoﬀ by applying the complete linkage

method. Of the ﬁve best-scored poses, the pose associated with the

biggest cluster was selected for further pose analysis. The described

redocking experiments showed that the largest cluster always

contained the pose with the lowest rmsd to the crystal structure and

the top-scoring pose. Seven conserved water molecules (HOH722,

729, 731, 737, 881, 952, 954 form the structure 4EY7) were chosen

from an alignment of 1EVE, 4EY7, and the apoprotein structure 4EY4.

The usage of these selected water molecules was validated using the

water-“toggle” and water-“on” mode in the docking program. In 9 out

The stock solutions of all tested compounds were prepared by

dissolving the free base compound in equivalent ethanolic HCl and

234

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

water. The dilution series of all stock solutions were prepared by using

water (the highest concentration (10 μM) of all tested compounds

contains less than 0.01% ethanol (v/v)).

7.50−7.27 (m, 10H, arom.), 7.19 (dd, J = 9.2, 3.1 Hz, 1H, arom.), 5.35

(s, 2H, OCH₂Ph), 5.04 (s, 2H, OCH₂Ph), 4.22 (d, J = 7.1 Hz, 2H,

OCH₂CH₃), 1.37−1.25 (m, 3H, OCH₂CH₃). Lit. H NMR (CDCl₃)

1

Steady-State GTPase Activity Assay. GTPase activity assays were

operated as described for the hH₁R, hH₂R, hH₃R, and hH₄R^33,56−59

hH₁R assays: Sf9 insect cell membranes coexpressing the hH₁R and

RGS4 were employed; hH₂R assays: Sf9 insect cell membranes

expressing the hH₂R-GsαS fusion protein were employed; hH₃R

assays: Sf9 insect cell membranes coexpressing the hH₃R, mammalian

Giα2, and Gβ1γ2 were employed; hH₄R assays: Sf9 insect cell

membranes coexpressing the hH₄R-RGS19 fusion protein, mammalian

Giα2, and Gβ1γ2 were employed. The relative membranes were

thawed and then centrifuged for 10 min at 4 °C and 13 000g.

Membranes were resuspended in 10 mM Tris/HCl, pH 7.4. Each

assay tube contained Sf9 membranes expressing the respective

histamine-receptor subtype (10 μg protein/tube), MgCl₂(hH₁R,

hH₂R assays: 1.0 mM, hH₃R, hH₄R assays: 5.0 mM), 100 μM EDTA,

100 μM ATP, 100 nM GTP, 100 μM adenylyl imidophosphate, 5 mM

creatine phosphate, 40 μg creatine kinase, 0.2% (w/v) bovine serum

albumin in 50 mM Tris/HCl (pH 7.4), and the investigated ligands at

various concentrations. All hH₄R assays additionally included 100 mM

NaCl.

Reaction mixtures (80 μL) were incubated for 2 min at 27.5 °C.

After the addition of 20 μL of [γ-³³P]GTP (0.05 μCi/tube), reaction

mixtures were incubated for 20 min at 27.5 °C. The reactions were

stopped by the addition of 900 μL of slurry consisting of 5% (w/v)

activated charcoal and 50 mM NaH₂PO₄, pH 2.0. Charcoal absorbs

nucleotides but not ³³P_i. Charcoal-quenched reaction mixtures were

centrifuged for 7 min at 4 °C and 13 000g. A volume of 600 μL of the

supernatant was taken, and the activity of ³³P_iwas determined by

liquid scintillation counting.

δ: 10.2 (s, 1H, NH), 8.35 (d, 1H, J = 9 Hz, H3), 7.62 (d, 1H, J = 3 Hz,

H6), 7.40 (m, 10H, OCH₂Ph), 7.19 (dd, 1H, J = 3 and 9 Hz, H4),

5.34 (s, 2H, OCH₂Ph), 5.04 (s, 2H, OCH₂Ph), 4.21 (q, 2H, J = 7 Hz,

OCH₂CH₃), 1.31 (t, 3H, J = 7 Hz, OCH₂CH₃).

6-Amino-3-benzyloxy-N-ethoxycarbonylbenzoic Acid (II).²⁵The

benzyl ester I (0.405 g, 1.0 mmol) and KOH (0.112 g, 2.0 mmol) were

heated at reﬂux in MeOH/H₂O (10 mL, 3:1) for 2 h. The reaction was

cooled and acidiﬁed with 1 N HCl (2 mL). The precipitate was ﬁltered

and air-dried to aﬀord the title compound (0.281 g, 89%) as a white

1

solid, mp 157−160 °C (lit. mp 159−160 °C). H NMR (300 MHz,

CDCl₃) δ: 9.94 (s, 1H, CO₂H), 8.40 (d, J = 9.3 Hz, 1H, arom.), 7.69

(d, J = 3.1 Hz, 1H, arom.), 7.51−7.18 (m, 7H, arom.), 5.08 (s, 2H,

OCH₂Ph), 4.24 (q, J = 7.1 Hz, 2H, OCH₂CH₃), 1.34 (s, 3H,

1

OCH₂CH₃). Lit. H NMR (CDCl₃) δ: 9.94 (s, 1H, CO₂H), 8.39 (d,

1H, J = 9 Hz, H3), 7.69 (d, 1H, J = 3 Hz, H6), 7.37 (m, 5H,

OCH₂Ph), 7.24 (dd, 1H, J = 3 and 9 Hz, H4), 5.07 (s, 2H, OCH₂Ph),

4.24 (q, 2H, J = 7 Hz, OCH₂CH₃), 1.34 (t, 3H, J = 7 Hz, OCH₂CH₃).

6-Benzyloxyisatoic Anhydride (11).²⁵The acid II (0.472 g, 1.5

mmol) was dissolved in dry THF (15 mL), oxalyl chloride (1 mL)

added, and the reaction heated at reﬂux temperature for 2 h. The

solvent was evaporated, ether (10 mL) added to the residue, and the

mixture heated under reﬂux for 10 min. The precipitate was ﬁltered

and washed with ether (10 mL) to aﬀord the title compound (0.350 g,

1

87%) as a white solid, mp 220−222 °C (lit. mp 218−220 °C). H

NMR (300 MHz, DMSO) δ: 11.62 (s, 1H), 7.39 (ddd, J = 19.4, 13.6,

1

7.6 Hz, 7H), 7.11 (d, J = 8.6 Hz, 1H), 5.17 (s, 2H). Lit. H NMR

(DMSO-d₆) δ: 11.60 (s, 1H, NH), 7.40 (m, 7H, CH₂Ph, H7 and H5),

7.12 (d, 1H, J = 8.5 Hz, H4), 5.16 (s, 2H, OCH₂Ph).

Spontaneous [γ-³³P]GTP degradation was determined in tubes

containing all components described above, plus a high concentration

of unlabeled GTP (1 mM) that, due to competition with [γ-³³P]GTP,

prevents [γ-³³P]GTP hydrolysis by enzymatic activities in the presence

of Sf9 membranes. Spontaneous [γ-³³P]GTP degradation was <1% of

the total amount of radioactivity added. The experimental conditions

chosen ensured that less than 10% of the total amount of [γ-³³P]GTP

added was converted to ³³P_i. For all the experiments, three diﬀerent

batches of Sf9 membranes were prepared and used. All experimental

data were analyzed by nonlinear regression using Prism 5 program

(GraphPad Software, San Diego, CA).

6-(Benzyloxy)-1-methyl-1H-benzo[d][1,3]oxazine-2,4-dione (12).

To a solution of 6-benzyloxyisatoic anhydride compound 11 (500 mg,

1.86 mmol) in 5 mL of DMAc (N,N-dimethyl acetamide),

diisopropylethylamine (370 μL, 2.23 mmol) was added. The solution

was stirred for 10 min prior to addition of methyl iodide (140 μL, 2.23

mmol). The solution was heated at 60 °C for 6 h. The solution was

allowed to cool to room temperature, and 15 mL of cooled water was

added. The resulting suspension was stirred vigorously for 30 min,

ﬁltered, and washed with 20 mL of water and 20 mL of n-hexane to

aﬀord the title compound (490 mg, 93%) as a yellow solid, mp 248−

1

251 °C. ESI-MS: 284.18 m/z [M]⁺. H NMR (300 MHz, DMSO) δ:

Chemistry. General Methods. Melting points are uncorrected and

7.61−7.29 (m, 8H, arom.), 5.22 (s, 2H, CCH₂C), 3.44 (s, 3H, NCH₃).

¹³C NMR (75 MHz, DMSO) δ: 158.73 (CO), 153.85 (CO), 147.68

were measured in open capillary tubes, using a Barnstead Electro-

1

thermal IA9100 melting point apparatus. H and ¹³C NMR spectral

(arom.), 136.42 (arom.), 136.39 (arom.), 128.39 (arom.), 127.87

(arom.), 127.58 (arom.), 125.61 (arom.), 116.52 (arom.), 112.08

(arom.), 112.02 (arom.), 69.68 (CCH₂C), 31.65 (NCH₃).

data were obtained from a Bruker Advance spectrometer (300 and 75

MHz, respectively). TLC was performed on silica gel on aluminum

foils with ﬂuorescent indicator 254 nm (Merck). For detection, iodine

vapor or UV light (254 nm) was used. ESI-MS samples were analyzed

using electrospray ionization ion-trap mass spectrometry in nanospray

mode using a Thermo Finnigan LCQ Deca instrument . The CHN

analyses were undertaken using Perkin-Elmer elemental analyzer

PE2400CHNS. In case purity was determined by elemental analysis,

small amounts of solvent had to be taken into account which had been

applied in previous column chromatography and appeared in the

corresponding NMR spectra. For column chromatography, silica gel

60, 230−400 mesh (Merck) was used.

10-(Benzyloxy)-13-methyl-13,13a-dihydro-5H-isoquinolino[1,2-

b]quinazolin-8(6H)-one (13). A suspension of 6-(benzyloxy)-1-

methyl-1H-benzo[d][1,3]-oxazine-2,4-dione 12 (300 mg, 1.45

mmol) and 3,4-dihydroisoquinoline (228 mg, 1.74 mmol) in 50 mL

of toluene was reﬂuxed for 24 h. After cooling, the solvent was

removed under reduced pressure and the residue was puriﬁed by

column chromatography using (100:1) CH₂Cl₂/MeOH as eluent

system to aﬀord the title compound (150 mg, 35%) as yellow thick oil.

HRESIMS (C₂₄H₂₂N₂O₂+H)⁺, m/z calcd: 371.1754; found: 371.1758.

¹H NMR (300 MHz, CDCl₃) δ: 7.68 (d, J = 2.7 Hz, 1H), 7.51−7.21

(m, 9H), 7.18−7.07 (m, 2H), 5.73 (s, 1H, NCHN), 5.19−5.05 (m,

2H, OCH₂C), 4.66 (ddd, J = 12.8, 4.8, 3.1 Hz, 1H), 3.26 (ddd, J =

12.8, 11.4, 3.9 Hz, 1H), 3.10−2.79 (m, 2H), 2.37 (d, J = 4.5 Hz, 3H,

NCH₃). ¹³C NMR (75 MHz, CDCl₃) δ: 164.12 (CO), 155.04

(arom.), 145.30 (arom.), 136.93 (arom.), 136.86 (arom.), 132.02

(arom.), 128.62 (arom.), 128.57 (arom.), 128.42 (arom.), 128.25

(arom.), 128.05 (arom.), 127.65 (arom.), 127.12 (arom.), 124.11

(arom.), 123.65 (arom.), 122.06 (arom.), 111.82 (arom.), 71.80

(NCHN), 70.46 (OCH₂C), 38.85 (NCH₂CH₂), 36.29 (NCH₃), 28.72

(CH₂CH₂C).

Benzyl 2-Amino-5-benzyloxy-N-ethoxycarbonylbenzoate (I).²⁵2-

Amino-5-hydroxybenzoic acid (0.765 g, 5.0 mmol) and ethyl

chloroformate (0.814 g, 0.72 mL, 7.5 mmol) were heated in dry

dioxane (25 mL) at reﬂux temperature for 2 h. The solvent was

evaporated, the residue dissolved in dry DMF (10 mL), K₂CO₃(2.07

g, 0.015 mol), and benzyl bromide (1.88 g, 1.31 mL, 0.011 mol) added

and the reaction stirred under N₂for 2 h at ambient temperature. The

reaction was diluted with water (50 mL) and extracted with ether (3 ×

20 mL). The ether extract was concentrated and puriﬁed by column

chromatography using 10:1 (EtOAc/hexanes) as eluent system to

aﬀord the title compound (1.63 g, 80%) as a white solid, mp 65−67

10-Hydroxy-13-methyl-13,13a-dihydro-5H-isoquinolino-[1,2-b]-

quinazolin-8(6H)-one (14). To a mixture of compound 13 (300 mg,

0.81 mmol) and 10% Pd−C (20 mg) was added 10 mL of ethanol.

1

°C (lit. mp 64 °C). H NMR (300 MHz, CDCl₃) δ: 10.17 (s, 1H,

NH), 8.37 (d, J = 9.3 Hz, 1H, arom.), 7.63 (d, J = 3.1 Hz, 1H, arom.),

235

dx.doi.org/10.1021/cn4002126 | ACS Chem. Neurosci. 2014, 5, 225−242

ACS Chemical Neuroscience

Research Article

The mixture was stirred at room temperature and under an

atmosphere of H₂for 4 h. The mixture was ﬁltered over celite to

aﬀord the title compound (190 mg, 84%) as greenish yellow solid, mp

189−192 °C. ESI-MS: 281.0 m/z [M]⁺. ¹H NMR (300 MHz, CDCl₃)

δ: 2.33 (s, 3H, NCH₃). 2.86 (m, 1H, CCH₂CH₂), 2.99 (m, 1H,

CCH₂CH₂), 3.37−3.23 (m, 1H, NCH₂CH₂), 4.66 (m, 1H,

NCH₂CH₂), 5.75 (s, 1H, NCHN), 7.13−7.02 (m, 2H, arom.),

7.25−7.19 (m, 1H, arom.), 7.39−7.27 (m, 2H, arom.), 7.55−7.41 (m,

1H, arom.), 7.97 (s, 1H, arom.) ppm. ¹³C NMR (75 MHz, CDCl₃) δ:

28.69 (CCH₂CH₂), 36.31 (NCH₃), 38.99 (NCH₂CH₂), 71.98

(NCHN), 114.37 (arom.), 121.72 (arom.), 123.85 (arom.), 123.88

(arom.), 127.19 (arom.), 128.27 (arom.), 128.51 (arom.), 128.51

(arom.), 131.78 (arom.), 136.75 (arom.), 144.44 (arom.), 153.43

(COH), 164.63 (CO) ppm.

10-(3-Bromopropoxy)-13-methyl-13,13a-dihydro-5H-

isoquinolino[1,2-b]quinazolin-8(6H)-one (15). 1,3-Dibromopropane

(505 mg, 2.5 mmol), compound 14 (350 mg, 1.25 mmol), potassium

carbonate (260 mg, 1.9 mmol), and potassium iodide (21 mg, 0.125

mmol) were reﬂuxed in absolute acetonitrile (50 mL) for 2 h. After

cooling, the inorganic salts were ﬁltered oﬀ and the ﬁltrate was

concentrated under reduced pressure. The remaining residue was

taken up in aq. solution of NaCl, and extracted with methylene

chloride. The combined organic extracts were washed with brine and

dried over magnesium sulfate. The organic solvent was removed under

reduced pressure. The resulting crude product was puriﬁed by column

chromatography giving the title compound (400 mg, 80%) as a yellow

thick oil. HRESIMS (C₂₀H₂₁BrN₂O₂+H)⁺, m/z calcd: 401.0859;

found: 401.0855. ¹H NMR (300 MHz, CDCl₃) δ: 7.60−7.53 (m, 1H),

7.47 (dd, J = 7.9, 5.1 Hz, 1H), 7.36−7.26 (m, 2H), 7.24−7.18 (m,

1H), 7.12−7.01 (m, 2H), 5.72 (s, 1H), 4.75−4.56 (m, 1H), 4.14−4.06

(m, 2H), 3.67−3.54 (m, 2H), 3.33−3.16 (m, 1H), 3.07−2.77 (m, 2H),

2.36 (s, 3H), 1.31−1.20 (m, 2H). ¹³C NMR (75 MHz, CDCl₃) δ:

164.10 (CO), 154.92 (arom.), 145.27 (arom.), 136.90 (arom.), 132.00

(arom.), 128.55 (arom.), 128.37 (arom.), 128.24 (arom.), 127.10

(arom.), 124.09 (arom.), 123.62 (arom.), 121.67 (arom.), 111.58

(arom.), 71.82 (NCHN), 65.83, 38.86, 36.29 (NCH₃), 32.34, 29.96,

28.69.

10-((6-Bromohexyl)oxy)-13-methyl-13,13a-dihydro-5H-

isoquinolino[1,2-b]quinazolin-8(6H)-one (16). 1,6-Dibromohexane

(610 mg, 2.5 mmol), compound 14 (350 mg, 1.25 mmol), potassium

carbonate (260 mg, 1.9 mmol) and potassium iodide (21 mg, 0.125

mmol) were reﬂuxed in absolute acetonitrile (50 mL) for 2 h. After

cooling, the inorganic salts were ﬁltered oﬀ and the ﬁltrate was

concentrated under reduced pressure. The remaining residue was

taken up in aq. solution of NaCl, and extracted with methylene

chloride. The combined organic extracts were washed with brine and

dried over sodium sulfate. The organic solvent was removed under

reduced pressure. The resulting crude product was puriﬁed by column

chromatography giving the title compound (390 mg, 70%) as a yellow

thick oil. HRESIMS (C₂₃H₂₇BrN₂O₂+H)⁺, m/z calcd: 443.12829;

found: 443.1283. ¹H NMR (300 MHz, CDCl₃) δ: 7.54 (d, J = 2.5 Hz,

1H), 7.48 (dd, J = 7.9, 5.3 Hz, 1H), 7.37−7.26 (m, 2H), 7.21 (dd, J =

7.8, 5.1 Hz, 1H), 7.12−7.00 (m, 2H), 5.72 (s, 1H), 4.65 (ddd, J = 12.8,

4.8, 3.1 Hz, 1H), 4.11−3.90 (m, 2H), 3.42 (t, J = 6.8 Hz, 2H), 3.34−

3.16 (m, 1H), 3.07−2.77 (m, 2H), 2.35 (s, 3H), 2.01−1.73 (m, 4H),

1.57−1.43 (m, 4H). ¹³C NMR (75 MHz, CDCl₃) δ: 164.17 (CO),

155.36 (arom.), 145.01 (arom.), 136.91 (arom.), 132.05 (arom.),

128.53 (arom.), 128.4 (arom.), 128.20 (arom.), 127.10 (arom.),

124.13 (arom.), 123.65 (arom.), 121.78 (arom.), 111.32 (arom.),

71.82 (NCHN), 68.25, 38.81, 36.29 (NCH₃), 33.84, 32.72, 29.07,

28.73, 27.95, 25.30.

dried over magnesium sulfate. The organic solvent was removed under

reduced pressure. The resulting crude product was puriﬁed by column

chromatography giving the title compound (160 mg, 71%) as a yellow

thick oil. HRESIMS (C₂₈H₃₇N₃O₂+H)⁺, m/z calcd: 448.2959; found:

1

448.2960. H NMR (300 MHz, CDCl₃) δ: 7.53 (d, J = 2.4 Hz, 1H),

7.49−7.43 (m, 1H), 7.36−7.27 (m, 2H), 7.24−7.18 (m, 1H), 7.10−

7.00 (m, 2H), 5.71 (s, 1H), 4.64 (ddd, J = 12.8, 4.8, 3.1 Hz, 1H),

4.18−3.82 (m, 2H), 3.24 (ddd, J = 12.8, 11.4, 3.9 Hz, 2H), 3.12−2.68

(m, 2H), 2.47−2.23 (m, 9H), 1.88−1.70 (m, 2H), 1.64−1.30 (m,

12H). ¹³C NMR (75 MHz, CDCl₃) δ: 164.19 (CO), 155.45 (arom.),

144.95 (arom.), 136.91 (arom.), 132.05 (arom.), 128.51 (arom.),

128.41 (arom.), 128.18 (arom.), 127.09 (arom.), 124.13 (arom.),

123.66 (arom.), 121.78 (arom.), 111.31 (arom.), 71.81 (NCHN),

68.42, 59.54, 54.65, 38.78 (NCH₃), 36.27, 29.18, 28.73, 27.52, 26.84,

26.03, 25.95, 24.48. Elem. Anal. (C₂₈H₃₇N₃O₂+ 0.13 CH₂Cl₂) Calcd.:

C, 73.66; H, 8.19; N, 9.16; O, 6.98. Found: C, 73.85; H, 8.12; N, 9.14.

13-Methyl-10-(3-(piperidin-1-yl)propoxy)-13,13a-dihydro-5H-

isoquinolino[1,2-b]quinazolin-8(6H)-one (18). Piperidine (64 mg,

0.75 mmol), compound 15 (200 mg, 0.50 mmol), potassium

carbonate (105 mg, 0.75 mmol), and potassium iodide (8 mg, 0.05

mmol) were reﬂuxed in absolute acetonitrile (50 mL) for 2 h. After

cooling, the inorganic salts were ﬁltered oﬀ and the ﬁltrate was

concentrated under reduced pressure. The remaining residue was

taken up in aq. solution of NaCl and extracted with methylene

chloride. The combined organic extracts were washed with brine and

dried over magnesium sulfate. The organic solvent was removed under

reduced pressure. The resulting crude product was puriﬁed by column

chromatography giving the title compound (136 mg, 67%) as a

colorless thick oil. HRESIMS (C₂₅H₃₁N₃O₂+H)⁺, m/z calcd:

1

406.2495; found: 406.2501. H NMR (300 MHz, MeOD) δ: 7.50−

7.36 (m, 2H), 7.36−7.18 (m, 3H), 7.17−7.08 (m, 2H), 5.73 (s, 1H),

4.66−4.47 (m, 1H), 4.18−3.96 (m, 2H), 3.32−3.17 (m, 2H), 3.08−

2.83 (m, 2H), 2.77−2.55 (m, 6H), 2.40 (s, 3H), 2.06 (qd, J = 11.8, 6.8

Hz, 2H), 1.70 (ddd, J = 16.2, 11.8, 5.9 Hz, 4H), 1.53 (d, J = 5.0 Hz,

2H). ¹³C NMR (75 MHz, MeOD) δ: 166.00 (CO), 156.25 (arom.),

146.48 (arom.), 138.26 (arom.), 133.47 (arom.), 129.73 (arom.),

129.57 (arom.), 129.21 (arom.), 128.17 (arom.), 124.26 (arom.),

124.23 (arom.), 122.88 (arom.), 112.65 (arom.), 73.38 (NCHN),

67.68, 57.00, 55.40, 49.92, 49.64, 49.36, 49.07, 48.79, 48.51, 48.22,

40.44, 36.71 (NCH₃), 29.37, 27.07, 26.18, 24.82. Elem. Anal.

(C₂₅H₃₁N₃O₂+ 0.3 CHCl₃) Calcd.: C, 68.85; H, 7.15; N, 9.52; O,

7.25. Found: C, 68.74; H, 7.54; N, 9.29.

13-Methyl-10-(3-morpholinopropoxy)-13,13a-dihydro-5H-

isoquinolino[1,2-b]quinazolin-8(6H)-one (19). Morpholine (65 mg,