Design, synthesis and SAR analysis of potent BACE1 inhibitors: Possible lead drug candidates for Alzheimer's disease

Article abstract of DOI:10.1016/j.ejmech.2016.11.021

We have identified potent isophthalic acid derivatives armed with imidazol and indolyl groups as potent β-secretase inhibitors. The most effective analogs demonstrated low nano-molar potency for the BACE1 (β-secretase cleaving enzyme) as measured by FRET (Fluorescence Resonance Energy Transfer) and cell-based (ELISA) assays. Our design strategy followed a traditional SAR approach and was supported by molecular modeling studies based on previously reported hydroxyethylene transition state inhibitor derived from isophthalic acid I. In the FRET assay, the most potent compound, 10a, displayed an IC₅₀value for BACE1 of 75 nM, and exhibited cellular activity with an EC50 value of 0.81 μM. On the other hand, compound 11b was found to be the most potent compound in the cell-based assay with an EC₅₀value of 0.29 μM.

Full text of DOI:10.1016/j.ejmech.2016.11.021

Accepted Manuscript
Design, synthesis and SAR analysis of potent BACE1 inhibitors: Possible lead drug
candidates for Alzheimer's disease
Hamadeh Tarazi, Raed Abu Odeh, Raed Al-Qawasmeh, Imad Abu Yousef, Wolfgang
Voelter, Taleb H. Al-Tel
PII:
S0223-5234(16)30959-X
10.1016/j.ejmech.2016.11.021
EJMECH 9054
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 6 June 2016
Revised Date: 12 October 2016
Accepted Date: 10 November 2016
Please cite this article as: H. Tarazi, R.A. Odeh, R. Al-Qawasmeh, I.A. Yousef, W. Voelter, T.H.
Al-Tel, Design, synthesis and SAR analysis of potent BACE1 inhibitors: Possible lead drug
candidates for Alzheimer's disease, European Journal of Medicinal Chemistry (2016), doi: 10.1016/
j.ejmech.2016.11.021.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical Abstract
Design, Synthesis and SAR Analysis of Potent BACE1 Inhibitors: Possible Lead Drug Candidates
for Alzheimer’s Disease
Hamadeh Tarazi,a,b Raed Abu Odeh,b,c Raed Al-Qawasmeh,d Imad Abu Yousef,e Wolfgang Voelterf
and Taleb H. Al-Tel,*a,b
aCollege of Pharmacy, University of Sharjah, P.O. Box 27272, Sharjah, UAE
bSharjah Institute for Medical Research, University of Sharjah, P.O.Box 27272, Sharjah, UAE
cCollege of Health Sciences, Department of Medical Laboratory Sciences, University of Sharjah, P.O. Box
27272, Sharjah, United Arab Emirates
dDepartment of Chemistry, The University of Jordan, Amman 11942, Jordan
eCollege of Arts and Sciences, Department of Biology, Chemistry and Environmental Sciences,
American University of Sharjah, Sharjah, United Arab Emirates
fInterfakultäres Institut für Biochemie, Eberhard-Karls-Universität Tübingen, Hoppe-Seyler-Straße 4, D-
72076 Tübingen
New potent isophthalic acid derivatives armed with imidazolyl and indolyl groups as β-secretase
inhibitors have been synthesized. The most potent compound, 11b, displayed an EC50 value of 0.29 µM.

ACCEPTED MANUSCRIPT
Design, Synthesis and SAR Analysis of Potent BACE1 Inhibitors: Possible Lead Drug Candidates
for Alzheimer’s Disease
Hamadeh Tarazi,a,b Raed Abu Odeh,b,c Raed Al-Qawasmeh,d Imad Abu Yousef,e Wolfgang Voelterf
and Taleb H. Al-Tel,*a,b
aCollege of Pharmacy, University of Sharjah, P.O. Box 27272, Sharjah, UAE
bSharjah Institute for Medical Research, University of Sharjah, P.O.Box 27272, Sharjah, UAE
cCollege of Health Sciences, Department of Medical Laboratory Sciences, University of Sharjah, P.O. Box
27272, Sharjah, United Arab Emirates
dDepartment of Chemistry, The University of Jordan, Amman 11942, Jordan
eCollege of Arts and Sciences, Department of Biology, Chemistry and Environmental Sciences,
American University of Sharjah, Sharjah, United Arab Emirates
fInterfakultäres Institut für Biochemie, Eberhard-Karls-Universität Tübingen, Hoppe-Seyler-Straße 4, D-
72076 Tübingen
ABSTRACT: We have identified potent isophthalic acid derivatives armed with imidazol and indolyl
groups as potent β-secretase inhibitors. The most effective analogs demonstrated low nano-molar potency
for the BACE1 (β-secretase cleaving enzyme) as measured by FRET (Fluorescence Resonance Energy
Transfer) and cell-based (ELISA) assays. Our design strategy followed a traditional SAR approach and
was supported by molecular modeling studies based on previously reported hydroxyethylene transition
state inhibitor derived from isophthalic acid I. In the FRET assay, the most potent compound, 10a,
displayed an IC50 value for BACE1 of 75 nM, and exhibited cellular activity with an EC50 value of 0.81
µM. On the other hand, compound 11b was found to be the most potent compound in the cell-based assay
with an EC50 value of 0.29 µM.
Keywords: β-secretase inhibitors, Alzheimer’s Disease, Transition state mimics, Structure-based drug
design, Structure activity relationship, Isophthalic acid derivatives.
* Corresponding author. Tel.: +971-501732950;
E-mail address: taltal@sharjah.ac.ae ( T. H. Al-Tel).
1

ACCEPTED MANUSCRIPT
1. INTRODUCTION
Alzheimer’s disease (AD) is a progressive, neurodegenerative disorder and considered the most common
form of irreversible dementia. To date, there are no cures that stop or reverse the underlying progression
of the disease. With an aging population and increasing number of people with Alzheimer’s, the need to
develop ways to halt and treat the disease has become paramount [1]. According to many resources, more
than 25 million people are suffering from dementia and the worldwide annual socioeconomic burden has
been estimated to exceed US$200 billion [2]. As increased age is the leading risk factor of the disease, AD
frequency is expected to increase to an estimated 7.7 million cases in 2030 and 11–16 million cases in
2050 in the United States alone. These numbers do not encompass the large number of people with mild
cognitive impairment, a considerable proportion of whom will progress to AD [3]. Although, the cause of
AD remains unknown, a large bodies of evidence that highlights the central role of Aβ in the pathogenesis
of the disease [4-7]. Genetic and pathological evidences strongly support the amyloid cascade hypothesis
for AD: It states that Aβ (amyloid β), a proteolytic derivative of the large transmembrane amyloid
precursor protein (APP), particularly the least soluble 42 amino-acid fragment Aβ42 isoform, has an early
and imperative role in all forms of AD [3] (Figure 1). Aβ−peptides are produced as a result of the
sequential cleavage of APP (amyloid precursor protein), first at the N-terminus by β-secretase enzyme (β-
site APP cleaving enzyme, BACE1) [8-10], followed at the C-terminus by one or more γ-secretase
complexes (intramembrane aspartyl proteases) [9], as part of the β-amyloidogenic pathway. During this
process, two β-secretase cleavage products are produced; a secreted ectodomain fragment named APPsol,
and the membrane-bound C-terminal fragment C99 of APP [10]. Following β -secretase cleavage, a
second protease, γ-secretase, cleaves C99 to generate the toxic Aβ peptides (Aβ, 39–43 residues) which
are secreted from the cell. Thus, processes that limit Aβ production and deposition by preventing
formation, inhibiting aggregation, and/or enhancing clearance may offer effective treatments for AD.
Since β-secretase mediated cleavage of APP is the rate-limiting step of the amyloidogenic pathway,
BACE1 inhibition is considered a prominent therapeutic target for treating AD by diminishing Aβ peptide
formation in AD patients (Figure 1) [11-14].
2

ACCEPTED MANUSCRIPT
Figure 1. The amyloid hypothesis of Alzheimer’s disease and proteolytic processing of the amyloid-b
precursor protein (APP) leading to dementia.
The normal biological role of β-secretase is still unclear. As expected, BACE1-knockout mice are
deficient in Aβ production, indicating no compensatory mechanisms for β-secretase cleavage in mice.
More surprisingly, knockout mice did not exhibit serious problems due to β-secretase deletion, they were
healthy, fertile, and clinical chemistry parameters were normal in both young and aged animals [15].
The absence of Aβ production and the distinct pathology in BACE1-knockout mice is encouraging for β-
secretase drug development. However, inhibitor development has proven to be highly challenging, and so
far, only one company has reported clinical data on a β-secretase inhibitor [16]. Most potent aspartic
protease inhibitors are large hydrophilic peptides and the need for blood–brain barrier penetration adds an
additional hurdle on the path towards development of a β-secretase inhibitor [17-18].
3

ACCEPTED MANUSCRIPT
Figure 2. De novo design strategies utilized to find optimal fragments based on isophthalic acid motif I.
Recently, we have reported on the discovery of small molecules, isophthalic acid and imidazopyridine
derivatives as potent BACE1 inhibitors [19-20]. From our early SAR investigations we quickly learned
that the truncation of the benzamide portion on isophthalic acid I (Figure 2), leads to a compact scaffold II
and proved to be a conceptually valid strategy. Soon afterwards we carried out a detailed structure-activity
relationship studies around compound I that ultimately led to compounds of type III which resulted in
improvement of ligand affinity for the BACE1 enzyme (IC50 = 20 nM; Figure 2). Furthermore, in these
reports we proposed a pharmacophoric model based on isophthalic acid scaffolds that ultimately guided us
in this work to efficiently design small collections of potent BACE1 inhibitors (Tables 1 and 2).
2. RESULTS AND DISCUSSION
2.1. Chemistry
2.1.1. Inhibitors Design Concept
In our previous reports we found that an imidazopyridine ring is an optimal bioisoster of the left amide
appendage of compound I, however, an H-bond donor attached to this moiety may lead to strong
interactions with the catalytic amino acid residues in the enzyme active site. At this point in time, we
decided to replace the left amide portion of compound I by an indolyl group anchored on an imidazole
4

ACCEPTED MANUSCRIPT
ring leading to compounds of type IV, hypothesizing this might improve the affinity of such motifs
against BACE1. Accordingly, the IC50 of such motifs (5-9 and 10-11) ranged from 0.29 µM to 1.33 µM in
the FRET assay (Tables 1 and 2). As a result, we report herein the design and synthesis of new motifs
derived from isophthalic acid as potential BACE1 inhibitors. As can be concluded from Figure 2, we
followed traditional medicinal chemistry strategies, among others, extension/contraction, rigidification,
and bioisostere replacement to arrive at our projected motifs.
As we previously described, to improve pharmacodynamic properties for this challenging drug target, we
decided to eliminate the amide functionalities of the peptide backbone present in I (Figure 2). To reduce
the rather high flexibility of acyclic hydroxyethylene (HE) TS mimetics, the HE TS mimetic was
embedded into a rigid TS mimetic (imidazole moiety cHE), retaining the essential H-bonding interactions
with the aspartyl protease catalytic dyad and the Flap amino acid residues. In addition, the cyclic scaffold
should provide suitable vectors for direct extensions into the corresponding S2′ and S1-S3 subpockets
(Figure 4 and 5). We envisioned that this concept should provide inhibitors with enhanced potency and
improved BBB permeability, due to their plausible PSA/LogP balance.
2.1.2. Identification of Initial Candidates by Molecular Modeling
In an iterative process, a concise set of heterocyclic scaffolds with an embedded HEA moiety and
attached fragments were docked using induced-fit like protocol with FITTED modeling package.
Conformational energy, S1, S2, S3 and S2′ site occupancy, and the overall binding potential were
assessed, including key H bonding interactions to the catalytic residues Asp32 and Asp228, to the Flap
residues. Lead differentiation and subpocket fragments were modified and scored in BACE1 active site
until inhibitor candidates within the desired property limits and binding profiles were optimized.
The significant binding features exhibited by I, II and III in BACE-1 active site, and the poor blood brain-
barrier penetrating characteristics of compound I prompted us to design new peptido-mimetic analogs as
potent BACE-1 inhibitors applying structure-based drug design strategies: 1) Isophthalic acid from
compound I was taken as core skeleton responsible for placing the ligands in most favorable binding
mode; 2) Aryl amides, benzimidazole and aryl-anchored imidazole moieties were introduced on the
isophthalic based core to form exquisite bonding with the corresponding S1-S3 pockets; 3) An indolyl
group was incorporated on the left arm of the isophthalic acid scaffold to fit into the S1′ or S2′ binding
pockets of BACE-1.
5

ACCEPTED MANUSCRIPT
To confirm our hypothesis and predict accurate binding modes of designed compounds within the active
pockets of BACE-1, molecular docking studies were performed as the first step. Docking BACE-1 could
be a challenging task, since the enzyme reported to undergo considerable conformational changes upon
ligand binding. Two regions of BACE-1 structure showed notable rearrangement upon ligand binding:
The Flap region (residues 68−74) and the 10s-loop (residues 9−14) [21]. The Flap region and the 10s loop
may adapt many conformations upon ligand binding: The Flap region may adapt closed or open
conformations, whereas, the 10s region may adapt up or down conformations. Accordingly, selection of
appropriate docking protocols would greatly affect predictions; hence we decided to tackle this by
employing an induced-fit like protocol as implemented within the FITTED package.
Four BACE-1 crystal structures were selected from the protein data bank as representative structures of
the enzyme conformations (Figure 3). These crystal structures were later used as basis for our fully
flexible approach.
Figure 3. Overlay of four BACE-1 crystal structures (PDB-codes; 1TQF, 2B8L, 2QZL and 3EXO)
illustrating the Flap, 10s region and the catalytic aspartate Dyad.
6

ACCEPTED MANUSCRIPT
An analysis of docking results revealed that the indole arm was projected towards the amphiphilic S2′
pocket and it was engaged in direct hydrogen bonding with the catalytic residue, Asp32. This finding was
observed in all of our derivatives. Based on that, we decided to use the indole group as part of the fixed
scaffold and only vary the S1-S3 appendages. Therefore, four series of 3-indolyl isophathalic acid
derivatives were synthesized (Schemes 1-4) and evaluated for their BACE-1 inhibitory activities.
2.1.3. Synthesis
We commenced our synthetic plan as contemplated in Scheme 1, we first assembled disubstituted
isophthalic acid scaffolds using indole 3-carboxylic acid, 3 and compound 2. Thus, the α-bromoketone 2
was reacted with the indolyl carboxylic acid followed by condensation with ammonia including multiple
rearrangements to generate the core scaffold I (Figure 4). With the latter in hand, it was taken into three
directional synthetic en routes to produce three series of scaffolds, IV, V and VI (Figure 4), thus allowing
diverse prime and nonprime side structure-activity relationship (SAR) explorations needed to fit into the
enzyme active site.
Figure 4. Retrosynthetic modifications of indolyl imidazolyl isophthalic acids scaffold
The preparation of compounds needed to delineate the SAR for this study was carried out according to
synthetic Schemes 1-4. Four types of scaffolds were synthesized: In the first, indolyl-imidazoles 4a-b,
were synthesized using well established procedures [22]. The latter were coupled with the desired aniline
derivative to deliver compounds 5a-d (Scheme 1). Thus, treatment of 4a-b with TBTU and DIPEA in
DMF at 0 °C followed by addition of desired aniline, furnished compounds 5a-b in fairly good yields.
7

ACCEPTED MANUSCRIPT
In the second approach, compound 3 (Scheme 2) was coupled with the nitro α-bromo ketone 7 using
°
DIPEA in DMF at 0 C. Subsequently, the product from this step was subjected to cyclization reaction
with the amidic carbonyl group via reaction with NH4OAc in acetic acid followed by silica gel
chromatography to deliver compounds 8a-b. Subsequently, traditional reduction of the nitro group using
Pd/C followed by coupling with the desired benzoic acid derivative furnished compounds 9a-b.
O
OH
O
O
O
O
X
i
O
O
3a-b
N
Br
H
2
ii, iii, iv
1
Y
O
O
N
N
X
X
v
N
OH
N
H
N
H
H
4a-b
N
H
N
5a-d
H
Scheme 1. Reaction conditions and reagents: (i) Br2, DCM-MeOH; (ii) DIPEA-DMF; (iii) HCO2NH4,
AcOH; (iv) NaOH-MeOH then H+; (v) DIPEA, TBTU, DMF, amine.
Scheme 2. Reaction conditions and reagents: (i) Br2, DCM-MeOH; (ii) DIPEA-DMF; (iii)-HCO2NH4,
AcOH, reflux; (iv) NaOH-MeOH then H+; (v) Pd/C, HCO2NH4, MeOH, reflux.
Scheme 3. Reaction conditions and reagents: (i) α-bromo ketone, DIPEA-DMF; (ii) HCO2NH4, AcOH,
reflux; (iii) NaOH-MeOH then H+;
8

ACCEPTED MANUSCRIPT
Scheme 4. Reaction conditions and reagents: (i) Phenylene diamine, TBTU, DIPEA, DMF; (ii) ACOH,
reflux.
For the synthesis of the third class of compounds, namely aryl-imidazole analogs of type 10 (Scheme 3),
compounds 4a-b were reacted with the desired α-bromo ketone. Subsequently, without further
purification, the products from this step were subjected to cyclization reaction with the amidic carbonyl
group using NH4OAc in refluxing AcOH followed by silica gel chromatography, to deliver compounds
10a-b.
Having demonstrated the successful synthesis of the previous three unique collections, we next turned our
attention towards the conversion of 4a-b to more compact scaffolds. In this regard, the fourth series of
compounds (Scheme 4) were synthesized using the union of phenylene diamines with the carboxylic acid
derivatives 4a-b. The products from this step were subjected to cyclization reaction through reflux in
AcOH to furnish benzimidazole motifs 11a-b.
2.2. Biological activities and SAR analysis
Having established robust en routes to the aforementioned compound collections, we initially profiled
those for their activities against BACE1. The goal of the current study is to develop selective and potent
BACE1 inhibitors. Toward this objective, we have first replaced the P1-P1' portions in compound I by an
indolyl arm anchored on imidazole to project directly toward the S1'/S2' sub-pockets of the BACE1 active
site. This was initiated based on our earlier extensive SAR analysis of compounds II and III (Figure 2).
To prove that the HE transition state mimic in compound I could be replaced by imidazole group, we have
subjected compound I to traditional medicinal chemistry strategies leading to scaffolds such as IV, V and
VI (Figure 4). With the necessary tools needed to fully investigate the SAR in hand, the synthesized
compounds were profiled for their potency at the target enzyme BACE1. The BACE1 primary screening
assay utilized for this program was the fluorescence resonance energy transfer (FRET) protocol. We used
9

ACCEPTED MANUSCRIPT
an APP-based peptide substrate (rhodamine-EVNLDAEFK-quencher, KM value of 20 µM) carrying the
Swedish mutation and containing rhodamine as a fluorescence donor and a quencher acceptor at each end.
IC50 values were calculated by plotting the obtained relative fluorescence unit per hour (RFU/h) against
the logarithmic of inhibitor concentration. The measured inhibition data were analyzed in GraphPad Prism
4 for Windows (GraphPad Software Inc., La Jolla, USA) by nonlinear regression (curve fitting). Cellular
potency of advancing compounds was done via cell-based Aβ inhibition (Aβ40 or Aβ42) in an enzyme-
linked immune sandwich assay (ELISA) in H4 cells (human neuroglioma cell line), expressing the double
Swedish mutation (K595N/M596L) of human APP (APPsw). The concentration at which the cellular
production of Aβ40 or Aβ42 was reduced by 50% (EC50) was determined and reported in Tables 1 and 2.
Compared to compounds I and II (Figure 2), the initial simple amides as baseline analogs (e.g. 5a-d,
Table 1) showed about 20-fold less potency, when compared to compound I (Figure 2) and a 5-fold
decrease in potency when compared to compounds of type II and III. As we were not satisfied with these
results, we decided to reverse the amide appendage, as we thought this might be engaged in additional
network of interactions in S1-S3 sub-pockets leading to higher affinity.
Thus, compounds 9a-b were synthesized according to Scheme 2 and found to possess approximately
similar potencies as those of 5a-d. Interestingly, compounds 9a-b, were found to be more potent than 5a-b
in the cell-based assay and possessed EC50 values of 480 and 410 nM, respectively. To understand the
biological sensitivity to which compounds of types 5 and 9 were subject and to find a plausible
explanation for the difference in the activity profiles between these and compound I, one needed to
consider their docking profile in the enzyme active site. Thus, we analyzed several X-ray crystal structures
of BACE1 (Figure 3) available in the Protein Data Bank and used them as a basis to calculate molecular
modeling data for our compounds using the FITTED program [23] (for details see experimental section).
Based on docking modes, compounds 5a-5d were found to share similar binding modes with respect to
BACE-1. The aryl amide group located in the S3 pocket and the indole arm fit within the critical S2’
pocket. Furthermore, we observed that the amidic and imidazole nitrogens formed additional hydrogen
bonding with the corresponding near 10s loop residue, Gly230 (Figure 5a). However, for compounds with
reversed amide appendage 9a-b, two consequences were observed; firstly, the aryl amide projected toward
10

ACCEPTED MANUSCRIPT
(A)
(B)
Figure 5. Best docked poses of compounds 5b (A) and 9b (B) within the active site of BACE-1 enzyme
based on ensemble docking. Ligands are in stick rendering and enzyme subsites are marked for clarity.
Hydrogen bond interactions are shown as green dashed lines.
Table 1. IC50 and EC50 values of BACE1 inhibitors.a
R
R
O
N
R1
N
N
H
H
N
N
H
HN
R1
N
H
HN
O
5a: R = H, R1 = Phenyl
5b: R = H, R1 = 4-flouroPhenyl
5c: R = H, R1 = 4-pyridyl
5d: R = H, R1 = 3-pyridyl
9a: R = H, R1 = 3-pyridyl
9b: R = F, R1 = 3-pyridyl
ELISA, WT
EC50
(µM, %inh)
BACE1
IC50 (nM)
Binding
PSA (Å2)
Score
Compound
ClogP LogBB
110±12
98 ±13
660±21
310±18
220±15
130±16
1.21 (64%)
-41.26
-41.29
-39.93
-38.68
-41.01
-41.84
73.6
73.6
86.5
86.5
86.5
86.5
4.62
4.78
3.54
3.54
3.54
3.70
-0.197
-0.171
-0.561
-0.561
-0.561
-0.535
5a
5b
5c
5d
9a
9b
1.33 (69%)
0.59 (70%)
0.88 (42%)
0.48 (75%)
0.41 (82%)
aIC50 and EC50 values are the mean values of at least three experiments ± SD.
11

ACCEPTED MANUSCRIPT
the S1 pocket and secondly, the amidic nitrogen is engaged in different interaction with the Thr232 amino
acid residue (Figure 5b). With these results in hand, we turned our attention toward finding groups that fit
more properly with S1-S3 sub-pockets. Based on our earlier extensive experience in these systems, we
envisioned that a more rigid imidazole ring would allow direct projection towards the unoccupied S2 sub-
pocket. Therefore, we have introduced a phenyl group at 4-position of the imidazole moiety leading to
compounds 10a-b (Scheme 3), and tested those for their activity profile against BACE1 (Table 2). As
expected, compounds 10a-b were found to be more potent against BACE1 when compared to the 5 and 9
series. The activity was decreased when the fluorine atom was replaced by the OCF3 group, however,
cellular potency increased (Table 2). Upon docking using FITTED program under induced-fit mode, it
was found that in the aryl-anchored imidazole derivatives 10a-b, the isophthalic core favors its orientation
towards the S2 pocket and appeared to be buried deeper into the S2 pocket (Figure 6a) with profound S2
shape complementarity. This may explain the improved potency of these rigid systems.
With these promising results in hand, we shifted our efforts towards further rigidification, as we felt that
this strategy was more likely to lead to the identification of drug-like BACE1 inhibitors with the potential
for good oral bioavailability and CNS penetration. Accordingly, benzimidazole-anchored derivatives 11a-
b were synthesized according to Scheme 4 and subsequently tested for their activities against BACE1
(Table 2). Fortunately, this brought about a similar potency toward BACE1; however, cellular potency
was increased four-folds. As shown in Figure 6b, these derivatives showed shallower penetration into the
S2 pocket with ability to form hydrogen bonding with the Lys321 residue and water molecules. In both
groups, the fluorine atoms appeared to influence their appropriate projection toward the S2 pocket
compared to OCF3 group.
This analysis could explain the enhanced potency of compounds 10a-b and 11a-b compared to the more
flexible amide series, which could be due to differences in the dynamic motion of the “Flap” loop of
BACE1 upon ligand engagement. This apparently may have influenced the ligand/protein contacts and
impeded the ligand’s affinity for BACE1 site, thus resulting in increased activity.
12

ACCEPTED MANUSCRIPT
(A)
(B)
Figure 6. Best docked poses of compounds 10a (A) and 11b (B) within the active site of BACE-1 enzyme
based on ensemble docking. Ligands are in stick rendering and enzyme subsites are marked for clarity.
Hydrogen bond interactions are shown as green dashed lines.
Table 2. IC50 and EC50 values of BACE1 inhibitors.a
R
N
F
N
HN
N
H
N
H
11a: R = F
11b: R = OCF3
ELISA, WT
EC50
(uM, %inh)
0.81 (86%)
BACE1
IC50 (nM)
Binding
PSA (Å2)
Compound
ClogP LogBB
Score
-39.60
-42.42
-37.29
-43.34
75±9
180±8
85±10
120±9
73.2
82.4
73.2
82.4
5.46
6.67
5.22
6.42
-0.054
0.005
10a
10b
11a
11b
0.67 (77%)
0.35(90%)
0.29 (87%)
-0.093
-0.035
aIC50 and EC50 values are the mean values of at least three experiments ± SD.
With these promising results in hand, we next turned our focus toward assessing the ability of our
derivatives in crossing the blood-brain barrier. It has been noted that molecules possessing a large PSA
may encounter difficulty in transiting biological membranes. This inability to cross membranes may result
13

ACCEPTED MANUSCRIPT
in poor absorption or lack of blood-brain barrier (BBB) penetration. Some of the first works correlating
PSA and oral absorption were published by Kelder and Palm [24] who found that highly absorbed (>90%)
drugs had a polar surface area around 70 Å2 while drugs that were poorly absorbed (<10%) had a polar
surface area larger than 140 Å2. Furthermore, the contribution of molecular LogP to the overall BB-
permeation capacity was found to be well correlated, and numerous models were reported describing such
influence including that proposed by Clark [25]. For this purpose, we regenerated the previously reported
Clark's equation (See supporting information) to serve as the basis for evaluating our compounds. As a
result, the predicted LogBB values (Tables 1 and 2) ranged from modest -0.5 to very encouraging 0.005
value. The derivatives, 10a-b and 11a-b, with less polar character as indicated by their calculated LogP
values, showed superior predicted penetration profile compared with the other derivatives. This indicates a
plausible PSA/LogP balance characteristics favoring blood-brain penetration compared to the basic amide
analogs.
These promising biological results of the aforementioned four chemically diverse collections encouraged
us to find the hidden relationship that encodes the chemical features of our derivatives with respect to the
observed BACE1 IC50. Towards this objective, we decided to conduct a classical QSAR study by
considering the logarithmic IC50 values of our compounds as dependent variable, and calculating ca. 1875
different descriptors to serve as independent variables. We found that the best developed model capable of
describing the inhibitory pattern against BACE1 is illustrated by the following equation.
pIC50 = 14.0703-1.4585×ATSC6c+24.5008×BCUTc1L
n = 10, F-Statistic = 61.15, R2 = 0.946, Q2LOO = 0.882, Q2LMO = 0.869, R2Y-scr = 0.215, s = 0.076
Where, the first descriptor, ATSC6c, is the centered Broto-Moreau autocorrelation of lag 6 weighted by
atomic charges. It is basically the spatial autocorrelation over the 2D molecular graph and describes how
atomic charges are distributed along the topological structure. On the other hand, the BCUTc-1l descriptor
is the number of highest eigen-values with respect to lowest partial charge weighted BCUTS. This
descriptor is based on the earlier modified burden matrix taking into account molecular connectivity as
well as the Gasteiger-Marsilli atomic partial charges.
The illustrated model was found to fit the experimental IC50 values (Figure 7) with considerable level of
significance as indicated by its high Fischer’s value (F), squared correlation coefficients (R2) and the
14

ACCEPTED MANUSCRIPT
minor standard errors of estimate. The model internal predictive power was judged based on leave-one-out
(LOO) and leave-many-out (LMO) procedures. The probability of chance correlation was examined by Y-
scrambling, in which lower (R2Y-Scr) value indicates lesser probability of chance correlation. The internal
validation parameters fulfilled the predicted model and demonstrated the reliability and robustness in
predicting new derivatives within our series and can be utilized for future development.
Figure 7. Scatter plot of Experimental versus predicted bioactivities (expressed by pIC50) against BACE1
enzyme derived from the best QSAR.
3. CONCLUSION
In this work we have optimized the synthesis of novel, low molecular weight and potent transition state
inhibitors of BACE1. Furthermore, we have developed peptido-mimetic groups to replace P1-P1' and P3
moieties resident on compound I. The synthetic methodologies followed were simple, efficient and
economic. Several of the developed low molecular weight scaffolds possessed high affinity, promising
ligand efficiency and acceptable PSA/LogP balance. Many of these motifs exhibited comparable cellular
potency. Moreover, since the peptidic nature and the number of rotatable bonds of these motifs are low,
the designed scaffolds represent a novel class of drug-like leads. The results described here merit further
investigations and developments in our laboratories.
15

ACCEPTED MANUSCRIPT
4. EXPERIMENTAL SECTION
4.1. General Methods
All reagents were used as purchased from commercial suppliers without further purification. The
reactions were carried out in oven-dried or flamed graduated vessels. Solvents were dried and purified by
conventional methods prior use. Flash column chromatography was performed with silica gel 60, 0.040-
0.063 mm (230-400 mesh). Aluminum-backed plates pre-coated with silica gel 60 (UV254) were used for
thin layer chromatography. 1H and 13C NMR spectra were recorded on a 400 MHz/100 MHz spectrometer.
Splitting patterns are designated as s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad.
Chemical shifts (δ) are given in ppm relative to the resonance of their respective residual solvent peak,
CD3OD. High and low resolution mass spectroscopic analyses were conducted using positive ion mode by
Electrospray Ionization (ESI). The samples were dissolved in acetonitrile, diluted in spray solution
(methanol/water 1:1 v/v + 0.1% formic acid) and infused using a syringe pump with a flow rate of 2
µL/min. External calibration was conducted using the arginine cluster in a mass range m/z 175-871.
4.2. Chemistry
4.2.1. General procedures for the synthesis of compounds 5a-d and 11a-b: To a solution of 3-(2-(1H-
o
indol-3-yl)-1H-imidazol-5-yl) benzoic acid (4) (303 mg, 1 equiv) in DMF (8 mL) at 0 C was added
DIPEA (0.21 mL, 1.2 mmol, 1.2 equiv). After 10 min, TBTU (551 mg, 1.2 mmol, 1.2 equiv) was added
and the resulting mixture stirred at the same temperature for 30 min. Then the desired aniline or
o
diaminobenzene (1.1 mmol, 1.1 equiv) was added. The resulting mixture was stirred at 0 C for 4 h, and
then quenched with ice-water. The precipitated solid was filtered, washed with water and dissolved in
EtOAc. The organic phase was washed with a 1 N HCl aqueous solution, then with a saturated NaHCO3
aqueous solution, and finally H2O, dried over MgSO4, and concentrated in vacuo, which was used in the
next stage without further purification. For 5a-b, the crude product was purified by silica gel flash
chromatography using increasing concentrations of DCM/AcOEt. For the synthesis of compounds 11a-
b, to the amides from the previous step, was added AcOH (30 mL) and the resulting suspension was
refluxed for 5 h, cooled to room temperature, concentrated in vacuo and diluted with crushed ice. The
brown solid was filtered, washed thoroughly with water. The crude was dissolved in EtOAc washed with a
saturated NaHCO3 aqueous solution and with H2O, dried over MgSO4, and concentrated in vacuo. The
16

ACCEPTED MANUSCRIPT
crude product was purified by flash chromatography on silica gel (DCM/AcOEt, 8/2 to 7/3) to give
benzimidazole derivatives 11a-b.
4.2.2. General procedures for the synthesis of compounds 4a-4b and 10a-10b: To a solution of 1H-indole-
3-carboxylic acid (3) (161 mg, 1.0 mmol, 1 equiv) in DMF (10 mL) or 3-(2-(1H-indol-3-yl)-1H-imidazol-
o
5-yl)benzoic acid (4) (303 mg, 1 equiv) at 0 C was added DIPEA (0.21 mL, 1.2 mmol, 1.2 equiv). The
o
resulting mixture was stirred at 0 C for 30 min followed by dropwise addition of the appropriate α-
o
bromoketone (1.1 mmol, 1.1 equiv) in DMF. The resulting mixture was stirred at 0 C for 4 h, and then
quenched with ice-water. The precipitated solid was filtered, washed with water and dissolved in EtOAc.
The organic phase was washed with a 1 N HCl aqueous solution, and then a saturated NaHCO3 aqueous
solution, then H2O, dried over MgSO4, and concentrated in vacuo, which was used in the next stage
without further purification. To this product, AcOH (25 mL) and AcONH4 (924 mg, 12 mmol, 12 equiv)
was added and the resulting suspension was refluxed for 5 h, cooled to room temperature, concentrated in
vacuo and diluted with crushed ice. The brown solid was filtered, washed thoroughly with water. The
crude cake was dissolved in EtOAc washed with a saturated NaHCO3 aqueous solution and with H2O,
dried over MgSO4, and concentrated in vacuo. The crude product was purified by flash chromatography
on silica gel (DCM/EtOAc, 8/2 to 7/3) to give imidazole derivatives 4a-4b or 10a-10b.
4.2.3. General procedures for the synthesis of compounds 9a-b: To a solution of of 1H-indole-3-
carboxylic acid (3) (161 mg, 1.0 mmol, 1 equiv) in DMF (10 mL) or 3-(2-(1H-indol-3-yl)-1H-imidazol-5-
o
yl)benzoic acid (4) (303 mg, 1 equiv) at 0 C was added DIPEA (0.21 mL, 1.2 mmol, 1.2 equiv). The
o
resulting mixture was stirred at 0 C for 30 min followed by dropwise addition of the appropriate α-
o
bromoketone (1.1 mmol, 1.1 equiv) in DMF. The resulting mixture was stirred at 0 C for 4 h, and then
quenched with ice-water. The precipitated solid was filtered, washed with water and dissolved in EtOAc.
The organic phase was washed with a 1 N HCl aqueous solution, and then a saturated NaHCO3 aqueous
solution, then H2O, dried over MgSO4, and concentrated in-vacuo, which was used in the next stage
without further purification. To this product, AcOH (25 mL) and AcONH4 (924 mg, 12 mmol, 12 equiv)
was added and the resulting suspension was refluxed for 5 h, cooled to room temperature, concentrated in
vacuo and diluted with crushed ice. The brown solid was filtered, washed thoroughly with water. The
crude cake was dissolved in EtOAc washed with a saturated NaHCO3 aqueous solution and with H2O,
dried over MgSO4, and concentrated in vacuo. The crude product was purified by flash chromatography
on silica gel (DCM/EtOAc, 8/2 to 7/3) to give nitro imidazole derivatives 8a-b. To this product (1 mmol)
17

ACCEPTED MANUSCRIPT
in MeOH (10 ml) was added HCO2NH4 (4 mmol) and Pd/C (10%wt/wt) and DMF (0.1 ml) and the
o
mixture was heated to 70 C for 1h, filtered over celite, evaporated to dryness. The crude solid was
dissolved in EtOAc and washed with a saturated NaHCO3 aqueous solution, then H2O, dried over MgSO4,
and concentrated in vacuo, which was used in the next stage without further purification. The amine
product (1 mmol) was dissolved in DMF (8 mL) at 0 oC followed by the adition of DIPEA (0.21 mL, 1.2
mmol, 1.2 equiv). After 10 min, TBTU (551 mg, 1.2 mmol, 1.2 equiv) was added and the resulting
mixture stirred at the same temperature for 30 min. Then the desired benzoic acid (1.1 mmol, 1.1 equiv)
o
was added. The resulting mixture was stirred at 0 C for 4 h, and then quenched with ice-water. The
precipitated solid was filtered, washed with water and dissolved in EtOAc. The organic phase was washed
with a 1 N HCl aqueous solution, then with a saturated NaHCO3 aqueous solution, and finally H2O, dried
over MgSO4, and concentrated in vacuo, which was purified by flash chromatography on silica gel
(DCM/EtOAc) to give amide derivatives 9a-b.
4.2.4. 3-(2-(1H-Indol-3-yl)-1H-imidazol-5-yl)-N-phenylbenzamide (5a)
This derivative was synthesized according to the general procedure C. Yield 55%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.18 (1H, m), 7.32 (5H, m) 7.19 (3H, m), 7.04 (3H, m), 6.94(2H, m),
6.63(2H, m). 13C NMR (100 MHz, CD3OD, in ppm): δ 161.2, 150.3, 137.1, 133.8, 132.7, 127.9, 125.3,
125.0, 124.7, 123.3, 122.2, 121.2, 120.7, 119.6, 117.4, 115.6, 112.5, 112.2. HRMS (ESI): calcd. for
C24H18N4O [M]+: 378.1481; found 378.1451.
4.2.5. 3-(2-(1H-Indol-3-yl)-1H-imidazol-5-yl)-N-(4-fluorophenyl)benzamide (5b)
This derivative was synthesized according to the general procedure C. Yield 63%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.18 (1H, s), 7.63 (3H, m) 7.26 (m, 3H), 7.19 (m, 4H), 7.10 (1H, t, J = 6.7
Hz), 7.02 (2H, m). 13C NMR (100 MHz, CD3OD, in ppm): δ 161.2, 157.9, 150.3, , 139.1, 136.3, 132.7,
129.2, 128.8,127.7, 126.9, 126.3, 125.3, 125.0, 124.1, 123.3, 122.3, 121.0, 121.2, 120.9, 119.5, 119.3,
117.4, 116.3, 115.9, 112.5, 108.2. HRMS (ESI): calcd. for C24H17FN4O [M]+: 396.1386; found 396.1381.
4.2.6. 3-(2-(1H-Indol-3-yl)-1H-imidazol-5-yl)-N-(pyridin-4-yl)benzamide (5c)
This derivative was synthesized according to the general procedure C. Yield 45%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.64 (2H, d, J = 8.1 Hz) , 8.11(4H, m), 8.02(1H, s) 7.82 (5H, m), 7.64(2H,
t, J = 7.1 Hz), 7.14(1H, t, J = 7.4 Hz). 13C NMR (100 MHz, CD3OD, in ppm): δ 166.6, 148.4, 148.2,
18

ACCEPTED MANUSCRIPT
147.9, 146.5, 146.1, 137.4, 131.2, 130.7, 130.3, 128.2, 127.8, 126.2, 125.7, 123.2, 121.3, 119.8, 118.9,
115.9, 114.7, 114.4, 112.5, 112.3. HRMS (ESI): calcd. for C23H17N5O [M]+: 379.1433; found 379.1435.
4.2.7. 3-(2-(1H-Indol-3-yl)-1H-imidazol-5-yl)-N-(pyridin-3-yl)benzamide (5d)
This derivative was synthesized according to the general procedure C. Yield 71%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 9.09 (1H, s),8.45 (1H, d, J = 7.4 Hz), 8.37 (1H, dd, J =1.2, 6.7 Hz),
8.07(3H, m),8.10(1H, s),7.93 (2H, m),7.84(1H,t, J =7.2 Hz), 7.72 (2H, m),7.57(1H, dd, J = 6.7, 8.3 Hz),
7.21(1H, t, J = 8.2 Hz). 13C NMR (100 MHz, CD3OD, in ppm): δ 167.1, 147.2, 144.1, 141.4, 140.1, 137.4,
134.2, 133.2, 130.9, 130.1, 129.4, 128.3, 127.6, 126.5, 125.9, 125.2, 124.4, 122.7, 120.4, 119.2, 116.3,
112.7, 112.5. HRMS (ESI): calcd. for C23H17N5O [M]+: 379.1433; found 379.1421.
4.2.8. N-(3-(2-(1H-Indol-3-yl)-1H-imidazol-5-yl)phenyl)isonicotinamide (9a)
This derivative was synthesized according to the general procedure C. Yield 69%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.92 (1H, s) 8.70 (1H, t, J = 3.8 Hz), 8.61(1H, d, J = 7.1 Hz), 8.34 (1H, s),
8.13 (3H, m), 8.07 (1H, s), 7.79 (1H, d, J = 7.2 Hz), 7.67 (2H, m), 7.53 (1H, t, J = 3.5 Hz), 7.41(1H, dd, J
13
= 3.8, 7.4 Hz) , 7.19 (1H, t, J = 7.3 Hz). C NMR (100 MHz, CD3OD, in ppm): δ 163.7, 154.1, 147.4,
147.1, 143.5, 138.7, 137.5, 137.2, 133.4, 132.2, 131.1, 127.5, 125.1, 123.2, 121.4, 120.3, 119.2, 118.9,
118.2, 115.7, 113.1, 112.7. HRMS (ESI): calcd. for C23H17N5O [M]+: 379.1433; found 379.1430.
4.2.9. N-(3-(2-(6-Fluoro-1H-indol-3-yl)-1H-imidazol-5-yl)phenyl)nicotinamide (9b)
This derivative was synthesized according to the general procedure C. Yield 58%. 1HNMR (CD3OD, 400
MHz, in ppm): δ = 9.00 (1H, s), 8.71(1H, t, J = 3.7 Hz), 861(1H, d, J = 7.1 Hz), 8.39 (1H , s), 8.13 (2H,
m), 8.10 (1H, s),7.99 (1H, d, J = 7.2 Hz), 7.84(1H, d, J = 7.3 Hz), 7.54 (2H, m) 7.43(1H, dd, J = 6.9, 7.3
13
Hz), 7.14 (1H, dd, J = 3.2,6.7 Hz). C NMR (100 MHz, CD3OD, in ppm): δ 164.1, 157.9, 154.3, 153.8,
149.7, 148.4, 147.9, 139.5, 139.2, 138.8, 137.7, 134.5, 133.2, 131.3, 124.7, 123.4, 122.9, 119.1, 118.4,
114.1, 113.2, 109.1, 104.5. HRMS (ESI): calcd. for C23H16FN5O [M]+: 397.1339; found 397.1341.
4.2.10. 6-Fluoro-3-(5-(3-(5-(4-fluorophenyl)-1H-imidazol-2-yl)phenyl)-1H-imidazol-2-yl)-1H-indole
(10a)
This derivative was synthesized according to the general procedure C. Yield 67%. 1HNMR (CD3OD, 400
MHz, in ppm): δ = 8.52(1H, d, J = 8.1 Hz), 8.16 (1H, t, J = 7.9 Hz), 8.07 (2H, d, J = 3.4 Hz), 7.41 (1H, s),
19

ACCEPTED MANUSCRIPT
7.39 (2H, m),7.18 (1H, d, J = 7.4 Hz), 7.16 (1H, d, J = 7.1 Hz), 6.93 (1H, dd, J = 4.74 Hz), 6.90 ( 3H, m),
13
6.67 (1H, dd, J = 1.3, 3.2, 6.5 Hz). C NMR (100 MHz, CD3OD, in ppm): δ 160.2, 159.4, 155.9, 150.3,
150.0, 140.2, 136.9, 132.5, 132.4, 129.8, 129.7, 128.7, 126.2, 125.5, 124.0, 123.8, 123.7, 118.2, 117.4,
114.3, 111.7, 108.2, 102.5, 102.7. HRMS (ESI): calcd. for C26H17F2N5 [M]+: 437.1452; found 437.14492.
4.2.11. 3-(5-(3-(5-(4-Fluorophenyl)-1H-imidazol-2-yl)phenyl)-1H-imidazol-2-yl)-6-(trifluoromethoxy)-
1H-indole (10b)
This derivative was synthesized according to the general procedure C. Yield 60%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.47 (1H, d, J = 8.1 Hz), 8.31(1H, t, J = 7.3 Hz), 8.03 (2H, m), 8.11(1H,
13
s), 8.06 (1H, s), 7.73 (1H, d, J = 7.4 Hz), 7.70 (2H, m), 7.16 (1H, m), 6.90 (1H, bs), 6.80 (1H, bs). C
NMR (100 MHz, CD3OD, in ppm): δ 159.4, 156.8, 155.9, 153.2, 150.3, 147.4, 146.7, 145.9, 139.9, 136.7,
134.7, 129.7, 128.2, 127.0, 126.0, 125.7, 125.0, 123.8, 121.5, 117.4, 116.2, 116.1, 112.0, 111.8, 108.2,
102.5, 102.2. HRMS (ESI): calcd. for C27H17F4N5O [M]+: 503.1369; found 503.1372.
4.2.12. 6-Fluoro-2-(3-(2-(6-fluoro-1H-indol-3-yl)-1H-imidazol-5-yl)phenyl)-1H-benzo[d]imidazole (11a)
This derivative was synthesized according to the general procedure C. Yield 51%, white solid. 1HNMR
(CD3OD, 400 MHz, in ppm): δ = 8.36 (1H, s), 8.01(1H, d, J = 7.1 Hz) , 7.91 (1H, d, J = 7.7 Hz), 7.69 (3H,
m), 7.61 (2H, bs), 7.51 (4H, m), 7.12 (3H, m). 13C NMR (100 MHz, CD3OD, in ppm): δ 159.4, 156.8,
159.9, 153.3, 150.3, 139.9, 135.7, 134.7, 128.2, 127.0, 126.4, 125.6, 125.0, 123.8, 121.5, 121.5, 117.4,
116.2, 112.0, 111.8, 108.2, 102.5, 102.2. HRMS (ESI): calcd. for C24H15F2N5 [M]+: 411.1295; found
411.1299.
4.2.13. 6-Fluoro-2-(3-(2-(6-(trifluoromethoxy)-1H-indol-3-yl)-1H-imidazol-5-yl)phenyl)-1H-
benzo[d]imidazole (11b)
This derivative was synthesized according to the general procedure C. Yield 73%, white solid. 1HNMR
(CD3OD, 400 MHz, , in ppm): δ = 9.121 (1H, bs), 8.74(1H, s), 8.34 (1H, d, J = 7.4 Hz), 8.12 (2H, m), 8.27
(1H, s), 8.02 (1H, d, J = 3.8 Hz), 7.91 (1H, dd, J = 1.4, 7.4 Hz), 7.70 (2H, m), 7.50 (2H, m), 7.34 (1H, dd,
J = 3.4, 7.3 Hz), 7.12 (2H, m). 13C NMR (100 MHz, CD3OD, in ppm): δ 160.0, 156.6, 153.2, 150.3,
149.0, 142.8, 138.1, 135.7, 127.0, 126.4, 125.7, 125.0, 123.0, 122.9, 122.4, 121.5, 117.4, 112.9, 112.5,
109.8, 109.7, 103.9, 103.5, 98.8. HRMS (ESI): calcd. for C25H15F4N5O [M]+: 477.1213; found 477.1220.
20

ACCEPTED MANUSCRIPT
4.3. Biology
4.3.1. BACE1 Enzymatic Assay
BACE1 assay was carried out according to the manufacturer described protocol available from Invitrogen,
USA (http://tools.invitrogen.com/content/sfs/manuals/L0724.pdf). Briefly, BACE1 in-vitro assay was
carried out by fluorescence resonance energy transfer (FRET). An APP-based peptide substrate
(rhodamine-EVNLDAEFK-quencher, KM value of 20 µM) carrying the Swedish mutation and containing
a rhodamine as a fluorescence donor and a quencher acceptor at each end was used. The intact substrate is
weakly fluorescent and becomes highly fluorescent upon enzymatic cleavage. The assays were conducted
for both enzymes in 50 mM sodium acetate buffer pH 4.5, in a final enzyme concentration (1 U/mL),
inhibitor (first line screening: 30 µM, 10 µM, 3 µM, 1 µM, and 0.3 µM) compounds that showed high
activity at 0.3 µM were validated at low concentration and both substrates were used at a concentration of
750 nM. Inhibitor compounds were diluted from stock solutions to result in 3.3% DMSO final
concentration. The reaction was incubated for 60 min at 25 °C under dark conditions and then stopped
with 2.5 M sodium acetate. Fluorescence measured with a Victor3 1420 (Wallac) microplate reader at 545 nm
excitation and 585 nm emissions. The assay kit was validated by manufacturer. The obtained values are
the mean values of 3 different experiments. IC50 values were calculated by plotting the obtained relative
fluorescence unit per hour (RFU/h) against the logarithmic of inhibitor concentration. The measured
inhibition data were analyzed in GraphPad Prism 4 for Windows (GraphPad Software Inc., La Jolla, USA)
by nonlinear regression (curve fitting).
4.3.2. Cell based assay
Aβ42 and Aβ40 were measured in culture medium of H4 cells (human neuroglioma cell line), expressing
the double Swedish mutation (K595N/M596L), of human APP (APPsw). Cells were seeded onto 24-well
plates (2x105 cell well-1) and allowed to grow for 24h, in 5% CO2/95% air in humidified atmosphere.
Increasing concentrations of compounds were added to the cells for overnight in a final volume of 0.5 ml.
R-flurbiprofen was used as positive control. DMSO-d6 (1%) was used as negative control. At the end of
the incubation, 100 µl of supernatants were removed and treated with a biotinylated mouse monoclonal
antibody (4G8, Signet Laboratories Inc., Dedham, MA, USA), specifically recognizing the 17-24 amino
acid region of Aβ and two rabbit polyclonal antibodies (C-term 42 and C-term 40, BioSource
International, Camarillo, CA, USA), specifically recognizing the C-terminus of Aβ42and Aβ40,
respectively. Antigen–antibodies 5 complexes were recognized by TAG-donkey anti-rabbit IgG (Jackson
21

ACCEPTED MANUSCRIPT
Immuno Research Laboratories, Soham, UK). Streptavidin coated magnetic beads captured the complexes
and the signals were read by an electrochemiluminescence instrument (Origen M8 Analyzer, BioVeris
Corporation, Gaithersburg, MD, USA). The cytotoxicity potential of test compound was assessed in the
same cells of the Aβ assay using the 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazol-iumbromide (MTT)
assay. After medium removal for of Aβ42 and Aβ40 determination, cells were incubated for 3h with 500µl
culture medium containing 0.5 mg ml-1 MTT, at 37 ºC, 5% CO2 and saturated humidity. After removal of
the medium, 25-500µl of 100% DMSO-d6 were added to each well. The amount of formed formazan was
determined reading the samples at 570 nm (background 630 nm) using a microplater reader (model 450,
Bio-Rad, Hercules, CA, USA).
4.4. Molecular Modeling
4.4.1. Molecular Docking
The binding modes for the synthesized compounds were explored by means of fully flexible molecular
docking computations employing the FITTED Suite, version 3.6 [23]. A well resolved X-ray crystal
structures for human beta secretase complexed with potent inhibitors were retrieved from the protein data
bank (http://www.rcsb.org/pdb/) under the entry codes (1TQF, 2B8L, 2QZL and 3EXO). Our compounds
were drawn using ChemBioDraw Ultra, version 11 (http://www.cambridgesoft.com) and were optimised
for energy and geometry using MMFF94s force field. Next, the compounds were prepared for docking by
assigning the proper bond orders, atom types, hybridization, partial atomic charges using the SMART
module within FITTED Suite.
The β-secretase crystal structures were subjected to a number of automated preparation steps prior
docking employing the different FITTED Suite modules. The PREPARE module was utilized to perform
sequence alignment, deletion and mutation to ensure that all of the protein ensembles are similar and differ
only by their Cartesian coordinates. Subsequently, MOL2 files for the different proteins ensembles
including water molecules apart from their complexed inhibitors were optimised for energy and
orientation by adding hydrogens, identifying alternative conformations and protonation states. The
catalytic Asp 32 residue was kept protonated during this procedure [26]. Later, the different protein
conformations were processed using ProCESS module to generate a truncated cavity and interaction site
within 10 Å from the center of the binding site and to identify the flexible residues within such
dimensions. Each of our compounds was docked individually using FITTED program under a fully
22

ACCEPTED MANUSCRIPT
flexible mode using the default parameters. The program accounts for the flexibility of protein/ligand
complexes via an hybrid matching/genetic algorithm search engine (i.e., evolution, mutation and
crossover) allowing water molecules to be exchanged independently from the different protein backbones.
Such computations provided an Induced-Fit like approach that simulates reality in biological environment.
The optimal binding mode for each compound was identified according to the lowest energy complex
emerged after each run. Finally, the results were presented using OpenEye VIDA [27] specialized 3D
molecular docking results viewer.
4.4.2. Quantitative-Structure Activity Relationship
The studied compounds were drawn using ChemBioDraw Ultra 11 (http://www.cambridgesoft.com) and
for each compound, the lowest energy conformation was obtained by means of MOPAC2012 program
[28] embedded within VEGA-ZZ [29-30]. The accurate energy minimization was achieved by Austin
Model-1 (AM1) semiemperical force-field within MOPAC. The AM1 force-field is characterized by its
high accuracy and coverage for wide range of atoms. Molecular descriptors calculations were carried-out
using PaDEL-Descriptor software [31]. The software calculates a large set of 1D-3D molecular
descriptors (ca. 1875) of different classes including constitutional, topological, information indices,
eigenvalue-based indices, radial distribution function (RDF), 2D/3D autocorrelation..etc. The calculated
descriptors were served as independent variables during QSAR model development. The logarithmic
values of experimentally observed IC50 values were served as the dependent variable. The QSAR models
were developed and validated employing the QSARINS software [32-33]. Model development process
initiated by reducing co-linear variables (corr. > 0.98) and excluding the descriptor showing higher pair-
wise correlation with others. All subsets procedure was adapted for the variables selection. Multiple linear
regression (GA-MLR) method was used for the final model building. The final models robustness were
validated employing internal predictive measures based on Q2LOO (leave one-out), Q2LMO (leave many-out)
and R2Y-scr (Y-scrambling).
4.4.3. Evaluation of Blood-Brain Permeation
The ability of the synthesized compounds for targeting the central nervous system and permeating the
blood-brain barrier was assessed computationally employing the previously reported Clark's equation [25].
For the original compound's training set found in Clark's paper, both the polar surface area (PSA) and
23

ACCEPTED MANUSCRIPT
ClogP were recalculated using Marvin, version 5.8.2 (http://www.chemaxon.com). Subsequently, the
QSAR equation was regenerated using QSARins software, where such regeneration was carried out to
account for the difference between the PSA and ClogP used by Clark and those generated by Mervin as
such difference would affect the corresponding descriptor's coefficients. The resulted QSAR equation (see
supporting information) was applied to predict the (LogBB) of our compounds.
5. Acknowledgement
The authors would like to thank Molecular Forecaster/McGill University as well as to Prof. Paola
Gramatica for providing the licensed software. This work was supported by generous grants from AlJalila
Foundation, UAE (grant number AJF201531); the Vice Chancellor’s office of Research and Graduate
Studies-University of Sharjah-UAE (grant number 15011101007), and the KEF Holding Charity
Foundation (Faizal and Shabanah Foundation)-UAE.
6. References
[1] M. Citron, Strategies for disease modification in Alzheimer's disease, Nat. Rev. Neurosci. 5 (9)
(2004) 677-685.
[2] A. Wimo, L. Jönsson, J. Bond, M. Prince, B. Winblad B, The worldwide economic impact of
dementia 2010, Alzheimers Dement. 9 (1) (2013) 1–11.
[3] Alzheimer’s Association, 2009 Alzheimer's disease facts and figures, Alzheimers Dement. 5 (3)
(2009) 234–270.
[4] D. J. Selkoe, Toward a comprehensive theory for Alzheimer's disease. Hypothesis: Alzheimer's
disease is caused by the cerebral accumulation and cytotoxicity of amyloid beta-protein, Ann.
N.Y. Acad. Sci. 924 (2000) 17-25.
[5] C. M. Couglan, K. C. Breen, Factors influencing the processing and function of the amyloid beta
precursor protein—a potential therapeutic target in Alzheimer's disease?, Pharmacol. Ther. 86 (2)
(2000) 111-144.
[6] M. Racchi, S. Govone, Rationalizing a pharmacological intervention on the amyloid precursor
protein metabolism, Trends Pharmacol. Sci. 20 (1999) 418-423.
[7] F. Checler, Processing of the β-Amyloid Precursor Protein and Its Regulation in Alzheimer's
Disease, J. Neurochem. 65 (4) (1995) 1431-1444.
[8] D. J. Selkoe, D. Schenk, Alzheimer's disease: molecular understanding predicts amyloid-based
therapeutics, Annu. Rev. Pharmacol. Toxicol. 43 (2003) 545–584.
[9] J. Hardy, D. J. Selkoe, The amyloid hypothesis of Alzheimer's disease: progress and problems on
the road to therapeutics, Science 297 (2002) 353–356.
24

ACCEPTED MANUSCRIPT
[10] T. Arendt, Alzheimer's disease as a disorder of mechanisms underlying structural brain self-
organization, Neuroscience 102 (4) (2001) 723-765.
[11] D.W. Dickson, The pathogenesis of senile plaques, J. Neuropathol. Exp. Neurol. 56 (4) (1997)
321-39.
[12] (a) J. Yuan, S. Venkatraman, Y. Zheng, B. M. McKeever, L. W. Dillard, S. B. Singh,
Structure-based design of β-site APP cleaving enzyme 1 (BACE1) inhibitors for the treatment of
Alzheimer's disease, J. Med. Chem. 56 (11) (2013) 4156-80; (b) S. Shangguan, F. Wang, Y.
Liao, H. Yu, J. Li, W. Huang, H. Hu, L. Yu, Y. Hu, R. Sheng, Design, Synthesis and Evaluation
of 3-(2-Aminoheterocycle)-4-benzyloxyphenylbenzamide Derivatives as BACE-1 Inhibitors,
Molecules 18 (3) (2013) 3577-3594.
[13] S. Butini, E. Gabellieri, M. Brindisi, S. Giovani, S. Maramai, G. Kshirsagar, E. Guarino, S.
Brogi, V. La Pietra, M. Giustiniano, L. Marinelli, E. Novellino, G. Campiani, A. Cappelli, S.
Gemma, A stereoselective approach to peptidomimetic BACE1 inhibitors, Eur. J. Med. Chem.
70 (2013) 233-247.
[14] M. Faraz Khan, M. M. Alam, G. Verma, W. Akhtar, M. Akhter, M. Shaquiquzzaman, The
therapeutic voyage of pyrazole and its analogs: A review, Eur. J. Med. Chem. 120 (2016) 170-
201.
[15] D. Dominguez, J. Tournoy, D. Hartmann, T. Huth,K. Cryns, S. Deforce, L. Serneels, I. E.
Camacho, E. Marjaux, K. Craessaerts, A. J. M. Roebroek, M. Schwake, R. D’Hooge, P. Bach, U.
Kalinke, D. Moechars, C. Alzheimer, K. Reiss, P. Paul Saftig, B. De Strooper, Phenotypic and
biochemical analyses of BACE1- and BACE2-deficient mice, J. Biol. Chem. 280 (35) (2005)
30797–30806.
[16] CoMentis. Press release 28 Jul 2008: CoMentis and Astellas to present Alzheimer’s disease
research at International Conference on Alzheimer’s Disease (ICAD). CoMentis website
[online], http://www.athenagen.com/index.php?/athenagen/press_releases/52/ (2008).
[17] D. Leung, G. Abbenante, D. P. Fairlie, Protease inhibitors: current status and future prospects, J.
Med. Chem. 43 (3) (2000) 305–341.
[18] T.B. Durham, T. A. Shepherd, Progress toward the discovery and development of efficacious
BACE inhibitors, Curr. Opin. Drug Discov. Develop. 9 (6) (2006) 776–791.
[19] T. H. Al-Tel, R. A. Al-Qawasmeh, M. F. Schmidt, A. Al-Aboudi, S. N. Rao, S. S. Sabri, W.
Voelter, Rational design and synthesis of potent dibenzazepine motifs as β-secretase inhibitors, J.
Med. Chem. 52 (20) (2009) 6484–6488.
[20] T. H. Al-Tel, M. H. Semreen, R. A. Al-Qawasmeh, M. F. Schmidt, R. El-Awadi, M. Ardah, R.
Zaarour, S. N. Rao, O. El-Agnaf, Design, Synthesis, and Qualitative Structure–Activity
Evaluations of Novel β-Secretase Inhibitors as Potential Alzheimer’s Drug Leads, J. Med. Chem.
54 (24) (2011) 8373–8385.
[21] Y. Xu, M. Li, H. Greenblatt, W. Chen, A. Paz, O. Dym, Y. Peleg, T. Chen, X. Shen, J. He, H.
Jiang, I. Silman, J. L. Sussman, Flexibility of the flap in the active site of BACE1 as revealed by
crystal structures and molecular dynamics simulations, Acta Crystallogr D Biol Crystallogr. 68
(2012) 13-25.
25

ACCEPTED MANUSCRIPT
[22] S.J. Stachel, C.A. Coburn, S. Sankaranarayanan, E. A. Price, B. L. Pietrak, Q. Huang, J.
Lineberger, A. S. Espeseth, L. Jin, J. Ellis, M. K. Holloway, S. Munshi, T. Allison, D. Hazuda,
A. J. Simon, S. L. Graham, J. P. Vacca, Macrocyclic inhibitors of beta-secretase: functional
activity in an animal model, J. Med. Chem. 49 (21) (2006) 6147−6150.
[23] (a) C.R. Corbeil, P. Englebienne, N. Moitessier, Docking Ligands into Flexible and Solvated
Macromolecules. 1. Development and Validation of FITTED 1.0, J. Chem. Inf. Model. 47
(2007) 435–449. (b) C. R. Corbeil, P. Englebienne, C. G. Yannopoulos, L. Chan, S. K. Das, D.
Bilimoria, L. L'Heureux, N. Moitessier, Docking Ligands into Flexible and Solvated
Macromolecules. 2. Development and Application of Fitted 1.5 to the Virtual Screening of
Potential HCV Polymerase Inhibitors, J. Chem. Inf. Model. 48 (2008) 902–909. (c) http://fitted.ca
[24] (a) J. Kelder, P. D. G. Grootenhuis, D. M. Bayada, L. P. C. Delbressine, J. P. Ploemen, Polar
molecular surface as a dominating determinant for oral absorption and brain penetration of drugs,
Pharm. Res. 16 (10) (1999) 1514−1519. (b) K. Palm, P. Stenberg, K. Luthman, P. Artursson,
Polar Molecular Surface Properties Predict the Intestinal Absorption of Drugs in Humans,
Pharm. Res. 14 (5) (1997) 568−571.
[25] D. E. Clarck, Rapid calculation of polar molecular surface area and its application to the
prediction of transport phenomena. 2. Prediction of blood-brain barrier penetration, J. Phar. Sci.
88 (8) (1999) 815-821.
[26] T. Polgẚr, G. M. Keserü, Virtual screening for β-secretase (BACE1) inhibitors reveals the
importance of protonation states at Asp32 and Asp228, J. Med. Chem. 48 (11) (2005) 3749-
3755.
[27] VIDA, version 4.2.1; Open Eye Scientific software, Inc.: Santa Fe, NM.
[28] MOPAC2012, J. J. P. Stewart, Stewart Computational Chemistry, version 15.321 web:
http://OpenMOPAC.net
[29] A. Pedretti , L. Villa , G. Vistoli, VEGA--an open platform to develop chemo-bio-informatics
applications, using plug-in architecture and script programming, J. of Comput. Aid. Mol. Des. 18
(3) (2004) 167-73.
[30] A. Pedretti , L. Villa , G. Vistoli, Atom-type description language: a universal language to
recognize atom types implemented in the VEGA program, Theor. Chem. Acc. 109 (4) (2003)
229-232.
[31] C. W. Yap, PaDEL-descriptor: an open source software to calculate molecular descriptors and
fingerprints, J. of Comput. Chem. 32 (7) (2011) 1466-1474.
[32] P. Gramatica, N. Chirico, N. Papa, S. Cassani, S. Kovarich, QSARINS: A new software for the
development, analysis, and validation of QSAR MLR models, J. of Comput. Chem. 34 (24)
(2013) 2121-2132.
[33] P. Gramatica, S. Cassani, N. Chirico, QSARINS-chem: Insubria datasets and new QSAR/QSPR
models for environmental pollutants in QSARINS, J. of Comput. Chem. 35 (13) (2014) 1036-
1044.
26

ACCEPTED MANUSCRIPT
Research Highlights
ꢀ
Four new series of isophthalic acid derivatives clubbed with Indolyl-Imidazole derivatives
were synthesized.
ꢀ Some of the synthesized compounds exhibited potent activity against BACE1.
A QSAR model of the synthesized motifs was developed using QSARINS software.
ꢀ