Synthesis and evaluation of clioquinol-rolipram/roflumilast hybrids as multitarget-directed ligands for the treatment of Alzheimer's disease

Article abstract of DOI:10.1016/j.ejmech.2018.12.013

Considering the importance of PDE4D inhibition and the modulation of biometals in Alzheimer's disease (AD) therapeutics, we have designed, synthesized and evaluated a series of new clioquinol-rolipram/roflumilast hybrids as multitarget-directed ligands fo

Full text of DOI:10.1016/j.ejmech.2018.12.013

Accepted Manuscript
Synthesis and evaluation of clioquinol-rolipram/roflumilast hybrids as multitarget-
directed ligands for the treatment of Alzheimer's disease
Jinhui Hu, Tingting Pan, Baijiao An, Zhengcunxiao Li, Xingshu Li, Ling Huang
PII:
S0223-5234(18)31050-X
DOI:
https://doi.org/10.1016/j.ejmech.2018.12.013
EJMECH 10946
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 16 October 2018
Revised Date: 4 December 2018
Accepted Date: 6 December 2018
Please cite this article as: J. Hu, T. Pan, B. An, Z. Li, X. Li, L. Huang, Synthesis and evaluation
of clioquinol-rolipram/roflumilast hybrids as multitarget-directed ligands for the treatment of
Alzheimer's disease, European Journal of Medicinal Chemistry (2019), doi: https://doi.org/10.1016/
j.ejmech.2018.12.013.
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 Abstracts

ACCEPTED MANUSCRIPT
Synthesis and evaluation of clioquinol-rolipram/roflumilast hybrids
as multitarget-directed ligands for the treatment of Alzheimer’s
disease
Jinhui Hu, Tingting Pan, Baijiao An, Zhengcunxiao Li, Xingshu Li*, Ling Huang*
School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou 510006,
China
Abstract: Considering the importance of PDE4D inhibition and the modulation of
biometals in Alzheimer’s disease (AD) therapeutics, we have designed, synthesized
and evaluated a series of new clioquinol-rolipram/roflumilast hybrids as
multitarget-directed ligands for the treatment of AD. In vitro studies demonstrated that
some of the molecules processed remarkable inhibitory activity against
phosphodiesterase 4D (PDE4D), strong intracellular antioxidant capacity, potent
inhibition of metal-induced aggregation of Aβ, and potential blood-brain barrier
permeability. Compound 7a demonstrated significant improvement in cognitive and
spatial memory in an Aβ25-35-induce mouse model in Morris water-maze test (MWM).
These results indicate that compound 7a is a promising multifunctional candidate that
is worthy of further study.
Keywords: Alzheimer’s disease, multitarget-directed ligands, PDE4D inhibitors,
metal chelating agents.

ACCEPTED MANUSCRIPT
Highlights:
New compounds with multifunctional activities against AD were synthesized and
evaluated.
Compound 7a exhibited effective PDE4D inhibitory potency, biometal chelating
activities and BBB permeability.
Compound 7a enhanced Aβ25-35-induced cognitive and memory impairment in mice.
Introduction
Alzheimer’s disease (AD), a progressive neurodegenerative disorder that is
characterized by memory loss and a decline of language skills and recognition, occurs
most frequently in elderly people. The World Alzheimer Report reported that there are
4
1
7 million people living with dementia, and the number would increase to more than
-
31 million by 2050 because of an aging population.[1] [2] Due to the complexity of
the pathogenesis, there is currently no drug that can completely treat AD. Among the
many factors that cause AD, aberrant protein processing is a major pathological
hallmark in which β-amyloid (Aβ) can form oligomers, fibrils and insoluble
2
+
2+
plaques.[3] Studies have confirmed that the levels of biometals including Cu , Zn
2
+
and Fe are redundant in the brains of AD patients (for copper, about 400 µM; for
zinc and iron, about 1 mM),[4] which, along with β-amyloid peptides, are the main
ingredients of these plaques.[5-7] Complexes of Aβ and metal ions can promote Aβ
aggregation and trigger the generation of reactive oxygen species (ROS) which may
initiate neuronal death and synaptic damage.[8-10] In addition, copper metabolism

ACCEPTED MANUSCRIPT
proteins are related to AD.[11, 12] Therefore, as one approach for the treatment of AD,
the regulation of biometal-Aβ interactions and amelioration the distribution of metals
in the AD patients’ brain seem to be a promising strategy. Studies have demonstrated
that clioquinol (CQ) and its analogs can significantly improve cognition[13-15] and
that these compounds can eliminate copper(II) from β-amyloid to prevent Aβ
aggregates resulting in promoting degradation of extracellular Aβ peptides.[16, 17]
The multitarget-directed metal-chelating agents HLA-20A and IQM-622 have been
developed.[18, 19]
Neuroinflammation is closely associated with the pathogenesis of AD.[20]
Phosphodiesterase 4 (PDE4) belongs to the phosphodiesterase subfamily of enzymes
that can hydrolyze the intracellular second messenger cAMP.[21] PDE4D is a subtype
of PDE4s (PDE4A to PDE4D) that has relatively high expression in the frontal
cortex.[22] PDE4D is one of the key subtypes participating in the process of memory
consolidation and long-term potentiation (LTP),[23] and PDE4B is involved in
neuroinflammation.[24] Recently, several strategies have been proposed to enhance
the levels of cerebral cyclic adenosine monophosphate (cAMP) for the treatment of
neurodegenerative diseases by the inhibition of PDEs.[25-28] The PDE4 inhibitor
rolipram not only facilitates memory performance in both LTP and contextual
learning but also repairs spine density to normal levels in APP/PS1 transgenic mouse
model of AD.[29] A preclinical research demonstrated that roflumilast (another PDE4
inhibitor) could improve cognitive deficits in rodent behavior experiments such as
new object recognition tasks and the Y-maze test. Currently, one Phase II study of the

ACCEPTED MANUSCRIPT
enhancement of translational cognition by roflumilast, and one Phase I study have
been completed which was to evaluate the co-administration of roflumilast and
donepezil for ameliorating scopolamine-induced cognitive performance.[30, 31] The
newly developed PDE4D, GEBR-32a, can enhance spatial and object recognition
performance in AD transgenic mice (Figure 1).[28, 32]
Due to the complicated pathogenesis and progression of neurodegenerative diseases,
the development of multitarget-directed ligands (MTDLs) that possess two or more
complementary biological activities for the treatment of AD, has been of interest in
recent years.[15, 33-35] Taking into account the importance of PDE4D inhibition and
the modulation of biometals in AD therapeutics,[36] we fused the key binding site
fragments of rolipram/roflumilast and linking it to the metal ion chelator framework
of CQ to obtain a series of hybrids. Herein, we report on the design, synthesis and
2
+
evaluation of PDE4D inhibitory activity, the inhibition of Cu -induced Aβ
aggregation, potential blood-brain barrier (BBB) penetration, antioxidation, and an
assessment of the water maze model for AD.

ACCEPTED MANUSCRIPT
Figure 1. Structures of PDE4D inhibitors and multitarget-directed metal-chelating agents.
Results and discussion
Chemistry. We have synthesized eight hybrids (5a-b, 6a-c, 7a-c) by coupling the
fragment of PDE4 inhibitors roflumilast and rolipram at the 5-position of
8
-hydroxyquinoline. Their synthetic routes are described in Scheme 1. Intermediates
1
a-c were prepared in accordance with a literature procedure.[37] The reaction of
amines 1a-c with corresponding substituted benzoyl chlorides yielded amides 2a-b,
a-c and 4a-c. Deprotection of the t-butyloxy carbonyl group with hydrogen chloride
3
produced compounds 5a-b, 6a-c and 7a-c.
a
Scheme 1. Synthesis of Compounds 5a-b, 6a-c, 7a–c
a
Reagents and conditions: (a) i). benzoic acid, SOCl , reflux; ii) anhydrous CH Cl , TEA, room
2
2
2

ACCEPTED MANUSCRIPT
temperature; (b) CH Cl :CH OH = 5:1, HCl (gas).
2
2
3
The preparation of compounds 13a-b and 14a-b is summarized in Scheme 2. The
starting materials 2-methylquinolin-8-ol 8a and 5,7-dichloro-2-methylquinolin-8-ol
8
b were reacted with benzyl chloride, and the subsequent oxidation of the methyl
group by SeO provided aldehydes. By the Pinnick oxidation, aldehydes were easily
2
converted to carboxylic acids 10a and 10b. By reaction with SOCl , and then with
2
corresponding amine analogues, 10a and 10b were converted to 11a-b and 12a-b,
which provided the target compounds 13a-b and 14a-b by removal of the Bn group
under an H atmosphere.
2
a
Scheme 2. Synthesis of Compounds 13a-b and 14a-b
a
Reagents and conditions: (a) benzyl bromide, K CO , acetone; (b) SeO ,1,4-dioxane; (c) NaClO ,
2
3
2
2
NaH PO , CH CN/H O; (d) i) SOCl2; ii) 3-(cyclopentyloxy)-4-methoxyaniline or
2
4
3
2
3
,4-dimethoxyaniline, anhydrous CH Cl , TEA, room temperature; (e) H , Pd / C, CH OH.
2 2 2 3
Scheme 3 shows the synthesis of target compounds 18a-b. Quinolin-8-ol 15 was
converted to TBS-protected 2-aminoquinolin-8-ol 16 by using previous reported
methods.[38] Compound 16 was treated with the corresponding benzoyl chlorides to
provide compounds 17a-b. Finally, deprotection of 17a-b afforded target compounds
1
8a-b.

ACCEPTED MANUSCRIPT
a
Scheme 3 Synthesis of Compounds 18a-b
a
Reagents and conditions: (a) mCPBA, CH Cl ; (b) i) (CH O) SO , CH CN; ii) NH ·H O; (c)
2
2
3
2
2
3
3
2
TBSCl, DIPEA, THF; (d) substituent acyl chloride, DIPEA, THF; (e) TBAF, THF.
Biology. PDE4D inhibitory activity. Considering that PDE4D is one of the key
subtypes participating in LTP, we chose PDE4D for screening with rolipram as the
reference compound to evaluate the PDE4D inhibitory activities of the synthesized
hybrids by using a previously described method.[39] As the results show in Table 1,
6
a-c
and
7a-c,
possessing
3-cyclopentyloxy-4-methoxy
and
3
-cyclopropylmethoxy-4-difluoromethoxy groups at the benzamide moiety, exhibited
potent inhibitory activities with IC50 values ranging from 0.399 to 3.83 µM, in
contrast to their analogues 5a and 5b (IC : 7.63 and 7.36 µM) which possess two
5
0
methoxy groups at the same position. Further examination of the relationship between
structure and activity indicated that substituents at the 7-position of the
8
-hydroxyquinoline moiety also have a great influence on the activity. Compounds 6a,
c, 7a and 7c, possessing hydrogen or iodine at the 7-position, provided much better
6
activities (IC : 0.492, 0.597, 0.399 and 0.426 µM for 6a, 6c, 7a and 7c) than did
5
0
chlorine at the same position (IC : 3.13 and 3.83 µM for 6b and 6b), which might be
5
0
related to the poor solubility of 6b and 6b in the buffer. Compounds 14a-b and 18a-b,
in which 3-(cyclopentyloxy)-4-methoxybenzamide or

ACCEPTED MANUSCRIPT
3
2
-cyclopropylmethoxy-4-difluoromethoxy benzamide moieties are joined at the
-position of 8-hydroxyquinoline, also provided moderate PDE4D inhibitory
activities with IC values ranging from 4.06 to 6.78 µM, suggesting that coupling
5
0
pharmacophores at this position might be not beneficial. Comparing the activities of
1
3a and 14b, we conclude that chlorine at the 5- and 7-positions of
8
-hydroxyquinoline moiety was unfavorable.

Table 1. PDE4D inhibitory activity, oxygen radical absorbance capacity (ORAC value), Pe value from the
ACCEPTED MANUSCRIPT
PAMPA-BBB assay for selected compounds and prediction of potential penetration into the CNS
a
PDE4D2
Pe
1
2
b
Comd.
X
R
R
ORAC
Pred.
a
-6
-1 c
IC (µM)
(×10 cm s )
NT
NT
5.90 ± 0.31
NT
5.86 ± 0.70
12.92 ± 0.82
NT
7.05 ± 0.96
NT
5
0
5a
H
I
H
Cl
I
CH
CH
CH
CH
CH
3
3
3
3
3
CH
CH
3
7.63 ± 0.21
7.36 ± 0.28
0.492 ± 0.007 0.63 ± 0.069
3.13 ± 0.11 0.56 ± 0.021
0.597 ± 0.024 1.01 ± 0.030
0.399 ± 0.021 0.60 ± 0.013
3.83 ± 0.14
1.34 ± 0.17
1.30 ± 0.045
NT
NT
CNS+
NT
CNS+
CNS+
NT
CNS+
NT
NT
NT
NT
CNS+
CNS+
CNS+
CNS+
5b
3
6a
6b
6c
7a
H
CF
2
2
2
H
H
H
7b
Cl CF
0.74 ± 0.065
7c
I
H
Cl
H
Cl
-
CF
0.426 ± 0.067 0.69 ± 0.023
1
1
1
1
1
1
3a
3b
4a
4b
8a
8b
CH
CH
3
-
-
-
-
4.32 ± 0.15
8.07 ± 0.26
4.15 ± 0.19
6.78 ± 0.27
4.64 ± 0.17
4.06 ± 0.32
1.03 ± 0.039
1.05 ± 0.058
1.98 ± 0.093
1.29 ± 0.011
0.70 ± 0.050
0.34 ± 0.045
NT
NT
NT
3
CH
CF H
2
16.41 ± 0.44
15.77 ± 0.42
18.87 ± 0.57
9.22 ± 0.62
5.20 ± 0.33
3
-
Rolipram
Roflumilast
CQ
0.621 ± 0.028 0.070 ± 0.010
0.480 ± 0.035 0.067 ± 0.015
>10
0.60 ± 0.047
CNS+
a
b
c
Values are expressed as mean ± SD from triplicate independent experiments. ORAC values are expressed as trolox equivalents. Pe Values
are expressed as the mean ± SD from at least triplicate independent experiments. Compounds could potential cross the BBB when Pe > 4.7 ×
-
6
-1
1
0 cm s . NT = not test.

ACCEPTED MANUSCRIPT
Free oxygen radical absorbance capacity. Oxidative stress is an important event in
the pathogenesis and progression of AD.[40] To evaluate the antioxidant activity of
these compounds, we performed the free oxygen radical absorbance capacity (ORAC)
assay with the vitamin E analogue Trolox as the reference. Compounds 6a-c and 7a-c,
which
possess
3-cyclopentyloxy-4-methoxy
or
3
-cyclopropylmethoxy-4-difluoromethoxy in the benzamide moiety, exhibited nearly
the same ORAC values as that of clioquinol (0.60 ± 0.047 ORAC value). In
comparison with the target compounds, rolipram and roflumilast provided very weak
antioxidative abilities, with ORAC values of 0.070 and 0.067, respectively. It appears
that the 3,4-dimethoxybenzamide moiety was slightly favorable for providing
antioxidant activity. Compound 5a-b and 13a-b produced ORAC values of 1.03-1.34,
better than those of 6a-c and 7a-c.
In vitro blood–brain barrier penetration assay. Whether or not CNS drugs could be
effective, one important issue is if they can pass through the blood-brain barrier
(
BBB). We have performed the parallel artificial membrane permeation assay
PAMPA) to evaluate the potential BBB permeability of the selected compounds. [41]
(
By comparing the permeability of thirteen commercial drugs with reported values to
validate the assay, a plot of experimental permeability (Pe) versus reported values was
constructed using this assay and produced a linear correlation of Pe (exp.) = 1.4574Pe
2
(
lit.) - 1.0773 (R = 0.9427). From this equation, and considering the limit established
6
-1
in the literature, Pe value of compounds was greater than 4.7 × 10 cm·s indicating
that the compound has potential BBB permeability. As the results in Table 1

ACCEPTED MANUSCRIPT
6
-1
demonstrate, all the tested compounds with a Pe value greater than 4.7 × 10 cm·s
could potentially cross the BBB (6a, 5.90 ± 0.31; 6c, 5.86 ± 0.70; 7a, 12.92 ± 0.82; 7c,
.05 ± 0.96; 18a, 16.41 ± 0.44; 18b, 15.77 ± 0.42). The Pe values for rolipram,
7
roflumilast and CQ were determined to be 18.87 ± 0.57, 9.22 ± 0.62 and 5.20 ± 0.33,
respectively.
Metal-chelating properties of compounds 6a and 7a. After consideration of their
PDE4D inhibitory activity, antioxidant activity and ability to pass the BBB,
compounds 6a and 7a were selected for further study with CQ as a reference. An
ultraviolet–visible spectroscopy assay was performed to determine their ability to
2
+
2+
3+
2+
chelate biometals including Cu , Fe , Fe and Zn . As shown in Figure 2 (A, B and
C), compound 6a exhibits a maximum absorption at 243 nm with a secondary peak at
2
23 nm in HEPES buffer. When CuSO was added into the solution, the maximum
4
absorption shifted to 267 nm, indicating the formation of a 6a-Cu(II) complex.
2
+
Likewise, a maximum absorption at 263 nm for 6a + Zn also indicates the
coordination of compound 6a with this metal. Compound 7a in buffer exhibited a
maximum absorption at 243 nm and a secondary peak was at 223 nm. New
absorbance bands were at 285 nm and 277 nm when CuSO and ZnCl were added
4
2
into the solution of 7a. Even though there was no obvious absorbance change with the
addition of FeSO or FeCl to the solution of 6a and 7a, the disappearance of the
4
3
2
+
3+
secondary peak indicated that there was potential interaction between the Fe , Fe
and the compounds. To confirm the components of the complex, the binding
2
+
stoichiometry of 6a and 7a with Cu was determined by measuring the changes in

ACCEPTED MANUSCRIPT
absorption at 415 and 422 nm, respectively. As shown in Figure 2D and 2E, the results
2
+
of titration analysis indicated that the Cu / ligand molar ratio is 1 : 2, which is
consistent with the added amount of CuSO and compound 6a or 7a. These results are
4
in accordance with that of CQ (Figure 3F).
Figure 2. (A, B and C) UV-vis absorption spectra of compounds 6a, 7a and CQ (50 µM) with
2+
2+
2+
3+
Cu , Zn , Fe or Fe (50 µM) in HEPES buffer. (D, E and F) UV-vis titration of 6a, 7a and CQ
2
+
2+
in the present of Cu in HEPS buffer. The concentrations of Cu and 6a/7a/CQ were 0-144 µM

ACCEPTED MANUSCRIPT
2
+
and 80 µM, respectively. Breakpoints are observed at [Cu ] : [6a/7a/CQ] = 0.5 : 1.
2
+
Inhibition of ROS produced in the Cu -ascorbate system. Reactive oxygen
species (ROS) play a key role in the pathogenesis of AD, and accumulated evidence
suggests that the production of ROS is dependent on the redox of metal ions.
2
+
Redox-active metal ions, especially redox-active Cu , can catalyze the generation of
ROS which triggers neuronal cell death in AD patients. To evaluate the ability of
metal chelators 6a and 7a to prevent copper redox cycling under aerobic conditions,
the copper redox cycling model was used as described by Faller and coworkers
(
Figure 3A).[42] Hydroxyl radicals (OH·) were generated by Cu-ascorbate redox
system with ascorbate and were determined by their reaction with
coumarin-3-carboxylic acid (CCA) to generate fluorescent
-hydroxy-coumarin-3-carboxylic acid. The fluorescence of the copper-ascorbate
7
system increased linearly with time for approximately 400 s and reached a plateau at
about 700 s, demonstrating a marked production of OH· (Figure 3B). In contrast,
there was no significant production of HO· when 6a and 7a were co-incubated with
the Cu(II)-ascorbate system, suggesting these compounds could prevent the copper
redox cycling by metal chelation. The metal chelator CQ was used as a control.

ACCEPTED MANUSCRIPT
Figure 3. (A) Production of hydroxyl radicals in the copper-ascorbate system (B) Fluorescence
intensity of copper-ascorbate-6a, copper-ascorbate-7a, copper-ascorbate-CQ, copper-ascorbate
and ascorbate alone systems (excitation: 395 nm; emission: 450 nm). Coumarin-3-carboxylic acid
(50 µM), ascorbate (150 µM), Cu(II) = 5 µM, 6a, 7a and CQ =10 µM. PBS (pH = 7.4), 37 °C.
Intracellular antioxidant assay. To evaluate the in vivo antioxidant activity of 6a-b
and 7a-b, an intracellular antioxidant assay was determined using 2',7'-
dichlorofluorescein diacetate (DCFH-DA) in SH-SY5Y cells. First, the cytotoxicity of
the compounds was evaluated using an MTT assay by assessing the viability of
SH-SY5Y cells. The results are summarized in Figure 4A. Compounds 6a and 7a
showed no effect on the cell viability at a concentration of 10 µM, while 6b and 7b
exhibited some cytotoxic effect at this concentration. Therefore, we chose 6a and 7a
for use in the cellular antioxidant assay. The ROS level was remarkably increased
when treating with t-BuOOH on SH-SY5Y cells (Figure 4B). Compound 7a exhibited
a concentration-dependent protective effect against t-BuOOH-induced intracellular

ACCEPTED MANUSCRIPT
oxidative stress, which suggested that 7a may be an efficient multifunctional agent for
further study.
Figure 4. (A) Cell viability of 6a-b and 7a-b on SH-SY5Y cells. (B) t-BuOOH-induced ROS
without or with the treatment of compounds by using DCFH-DA. The results are expressed as the
percent of vehicle control in three independent experiments. ***, p < 0.001 vs the vehicle group; #,
p < 0.05, ###, p < 0.001 vs the model group.
Inhibition of metal-induced Aβ aggregation. High levels of biometals (copper and
zinc) in AD patients may potentially be the cause of neurotoxicity. We have performed
the thioflavin T (ThT) fluorescence assay to evaluate the inhibitory activity of the
target compounds against metal-induced Aβ aggregation with clioquinol (CQ) as a
positive control. The fluorescence units increased from 100 % (Aβ alone) to 156 %
(
Aβ + Cu(II)) after the addition of CuSO (Figure 5A). The fluorescence units were
4
decreased significantly when CQ and compounds 6a, 6b, 7a and 7b were incubated
with Aβ and Cu(II). The results indicated that these compounds can strongly prevent
Cu(II)-induced Aβ aggregation. Furthermore, compounds 6a and 7a were more potent
2
+
than CQ (Aβ + Cu(II) + CQ, 66 %; Aβ + Cu(II) + 6a, 46 %; Aβ + Cu + 6a, 51 %).

ACCEPTED MANUSCRIPT
2
+
While 6b and 7b were slightly less potent than CQ (Aβ + Cu + 6b, 70.2 %; Aβ +
2
+
Cu + 7b, 82 %). Based on the ThT assay, morphological changes in the Aβ species
were studied by transmission electron microscopy (TEM). A marked production of Aβ
fibrils was detected in Cu(II)-treated samples of fresh Aβ (Figure 5B, sequence 2). In
contrast, smaller and amorphous conformations of Aβ aggregates were observed when
compounds 6a, 7a and CQ were incubated with Cu(II)-treated Aβ samples (Figure 5B,
sequences 3, 4 and 5), which was consistent with results of the ThT fluorescence
assay. Overall, ThT and TEM assay results jointly demonstrated that compounds 6a
and 7a have the potential inhibit metal-induced Aβ1−42 aggregation.
2
+
Figure 5. Inhibition by test compounds of Cu -induced Aβ
aggregation (A) Top: Scheme of
−42
1
the inhibition experiment. Bottom: Results of the thioflavin T fluorescence binding assay. Data are

ACCEPTED MANUSCRIPT
expressed as the mean ± SD in three independent experiments. *** p < 0.001, versus the Aβ +
Cu(II) group. (B) Visualization by TEM of the inhibition by compounds 6a and 7a on
2
+
Cu -induced Aβ aggregation.
1
-42
Molecular docking studies. Molecular docking studies were performed to investigate the
potential binding sites of compounds 6a and 7a with PDE4D. The molecular docking results of the
PDE4 crystal structure (PDB code: 1XOQ) with compounds 6a and 7a are shown in Figure 6 with
roflumilast as a reference. Compounds 6a and 7a were both sandwiched by the hydrophobic
residues Phe372 and Ile336. Cyclopentyl and methyl oxygen of 6a, and cyclopropylmethyl and
difluoromethyl oxygen of 7a formed hydrogen-bonds with the amide nitrogen of the Gln369 side
chain (Figures 6A and 6B). The hydrogen bond between Gln369 and stacking on Phe372 of
roflumilast (ROF) are shown in Figure 6C. It deserves mention that the binding mode of 7a was
very similar to that of roflumilast (Figure 6D), which may explain the relatively better activity in
the enzyme.

ACCEPTED MANUSCRIPT
Figure 6. Compounds 6a and 7a in the PDE4D (PDB: 1XOQ) binding sites with roflumilast
(ROF). (A) Binding mode for 6a (shown in cyan); (B) Binding mode for 7a (shown in cyan); (C)
Binding mode for roflumilast (ROF) (shown in green); (D) Comparison of the binding modes for
compounds 7a and roflumilast (ROF) (7a is colored in cyan and roflumilast is colored in green).
The backbone of PDE4D is shown as a ribbon representation. The dashed lines show
hydrogen-bonds as well as stacking between Phe372 and the aromatic ring of the inhibitors.
Acute toxicity of compounds 6a and 7a. To investigate the safety properties of
compounds 6a and 7a, acute toxicity experiments were performed. After intragastric
administration of compounds 6a and 7a with various dosages, none of the mice died
or were found to exhibit any abnormal behavior over the following 14 days.[43] The
organs (heart, liver, lungs and kidneys) of mice were examined after sacrifice on the
1
4th day . No apparent abnormal changes and damages in their organs were found in

ACCEPTED MANUSCRIPT
both compounds 6a and 7a groups. In general, compounds 6a and 7a exhibited safety
and well tolerated up to 2000 mg/kg which can use for further in vivo study.
Metabolic stability of compounds 6a and 7a. To assess drug-like properties of
compound 6a and 7a, the metabolic stability assay was performed by using rat liver
microsomes with CQ and donepezil as standards and testosterone as a positive control.
Compounds 6a and 7a exhibited T1/2 values of 51.7 and 90.3 min, respectively, in
liver microsomes (Table 2). The half-lives of testosterone (T1/2 = 2.3 min), donepezil
(T1/2 = 75.6 min) and CQ (T1/2 = 35.8 min) were consistent with previous reports.[44,
4
5] These results suggested that compound 6a and 7a were potentially metabolic
stable which could be carried out further studies in vivo.
Table 2. Metabolic stability of 6a and 7a in rat liver microsomes.
-
1
a
Compd
Testosterone
Donepezil
CQ
K (min )
T (min)
1/2
0.301 ± 0.033
2.3 ± 0.3
75.6 ± 4.4
35.8 ± 2.9
51.7 ± 3.3
90.3 ± 5.7
0.00917 ± 0.00044
0.0194 ± 0.0013
0.0134 ± 0.0080
0.00767 ± 0.00040
6
a
a
7
a
Results are expressed as the mean ± SD from three independent experiments.
Cognitive and memory strengthening in a mouse model by oral uptake of 6a and
a. To investigate the in vivo effects of candidate 6a and 7a, the Morris water-maze
7

ACCEPTED MANUSCRIPT
test was performed on a Aβ25-35-induced cognitive dysfunction in mice with rolipram,
roflumilast and CQ as positive controls. Mice were randomly allocated into 7 groups:
sham, model, rolipram, CQ, roflumilast, 6a and 7a groups.[46, 47] The mice in each
group (except the sham group) received intracerebroventricular injections (ICV) of a
solution of Aβ25-35 (3 nmol). For the sham group, mice were received ICV with saline
as the same volume. Then, the mice from each group were treated with corresponding
drug dosages by intragastric administration for 26 consecutive days (the sham and
model groups: 0.5 % CMC-Na solution; the rolipram group: 4 mg/kg; the roflumilast
group: 4 mg/kg; the CQ group: 30 mg/kg; the 6a group: 30 mg/kg, and the 7a group:
3
0 mg/kg). As shown in Figure 7A, there was no significant loss of body weight or
adverse or abnormal events (diarrhea, emesis-like behavior) during the drug
administration period, indicating that compounds 6a and 7a were safe in mice at a
dose of 30 mg/kg/day.
Twenty-one days later, training trials were performed by using the Morris water maze
(MWM) task. The model group exhibited a longer escape latency and required more
paths to find the hidden platform in the water pool compared to sham group.
Compared with the model group, compounds 6a and 7a could reduce the escape
latency and path for searching the platform. Escape latency for compound 7a was
shorter than that for rolipram, roflumilast and CQ, indicating that 7a could strengthen
the memory capability and cognitive functions in Aβ25-35-induced AD mice (Figure
7
B and 7C).

ACCEPTED MANUSCRIPT
Figure 7. (A) Daily weight for each group (sham, model, rolipram, CQ, roflumilast, 6a and 7a
groups) for 26 consecutive days (n = 11). (B) The escape latency for each group in training trials
(mean ± SD, n = 11). (C) The representative tracks for each group in training trials.
A spatial probe trial in the MWM test was performed by removal of the platform after
the training trials. The original platform location in the training trials was set as the
virtual platform, and its 2-fold diameter area was set as the effective region. The
number of virtual platform crossings, times in the effective region, as well as the paths
in the virtual platform were recorded as neurobehavioral parameters. The results
demonstrate that 6a and 7a provided approximately equivalent swimming speeds
relative to the sham group, indicating that administration of compounds 6a and 7a
have no effect on the motility and exploratory activities of the mice (Figure 8A, sham
=
=
17.5 ± 2.3, model = 15.6 ± 2.4, rolipram = 15.1 ± 3.8, CQ = 16.0 ± 2.2, roflumilast
14.5 ± 2.2, 6a = 15.3 ± 2.1 and 7a = 15.0 ± 2.8). The number of crossing the virtual
platform in the model group (1.5 ± 1.3) was remarkably (p < 0.01) less than that of the
sham group (3.63 ± 1.6), indicating that intrahippocampal injections of Aβ25-35 lead to
spatial learning and memory deficiency in mice (Figure 8B). The numbers of virtual

ACCEPTED MANUSCRIPT
platforms in the roflumilast (3.8 ± 1.5, p < 0.01), 6a (3.4 ± 1.7, p < 0.05) and 7a (4.3 ±
1
.9, p < 0.001) treatment groups were significantly increased compared with that of
the model group. Furthermore, paths in the virtual platform and times in the effective
region for the 6a and 7a groups (18.0 ± 7.0, p < 0.05 and 19.0 ± 8.0, p < 0.5; 6.1 ±
2
.37 and 6.8 ± 2.2, p < 0.1) were remarkably longer than that for the model group
10.2 ± 3.2; 4.3 ± 1.3), indicating that compounds 6a and 7a could alleviate the
deficits of learning and memory in mice induced by Aβ25-35 with rolipram (16.5 ± 9.3;
.7 ± 2.5), roflumilast (16.0 ± 9.0; 5.4 ± 1.6) and CQ (18.0 ± 12.7, p < 0.5; 5.8 ± 1.8)
(
5
at the tested dosage (Figures 8C and D). Taken together, these results demonstrate that
compound 7a could significantly improve spatial learning and cognition functions.
Compound 7a is the most promising potential candidate against AD.

ACCEPTED MANUSCRIPT
Figure 8. Effect of compounds 6a and 7a on the spatial probe trial in Aβ25-35-induced learning and
recognition dysfunction mice. Rolipram (4 mg/kg), roflumilast (4 mg/kg) and CQ (30 mg/kg)
were used as references. The results are expressed as the mean ± SD. Student’s t test: ∗ p < 0.05,
∗
∗ p < 0.01, ∗∗∗ p < 0.001 versus the model group. Mean ± SD. (A) Swimming speed. (B)
Number of virtual platform crossings. (C) Path in the virtual platform (D) Time in the effective
region. (E) Representative tracks of the mice in the spatial probe trials. Student’s t test: ∗ p < 0.05,
∗
∗ p < 0.01, ∗∗∗ p < 0.001 versus the model group.
Histopathological studies in the hippocampus. Histopathological studies in the
hippocampus were performed using hematoxylin and eosin (H&E) stain (Figure 9).
The hippocampus neurons in the CA1, CA3 and dentate gyrus (DG) regions in the
sham group were intact and properly aligned. Conversely, the neurons in the mice in
the model group exhibited robust nuclear pyknosis and a reduction in the number of

ACCEPTED MANUSCRIPT
neurons with noteworthy nuclear shrinkage, irregular array, much more darkly stained
cells and neuronal loss, proving the success of model. In 7a-treated mice,
hippocampal neurons were markedly protected exhibited a nearly normal state. The
hippocampal neurons in the rolipram, clioquinol, roflumilast and compound 6a groups
possessed a more normal morphology than the model group, however, 7a produced a
better protective effect on the hippocampal neurons than did the former compounds.
The denatured cell index (DCI) in CA1, CA3 and DG region of the hippocampus is
shown in Figure 10. It can be seen that 7a remarkably reduced the ratio of apoptotic
neurons in the CA1 (12.5 % ± 5.1), CA3 (7.9 % ± 5.4) and DG (18.7 % ± 4.8) regions,
comparing to the model (40.6 % ± 12.5, 80.2 % ± 8.6, 55.8 % ± 7.2, respectively),
rolipram (33.5 % ± 4.8, 41.1 % ± 9.3, 25.7 % ± 8.7, respectively), CQ (24.2 % ± 6.8,
3
2
7.8 % ± 8.6, 19.7 % ± 5.5, respectively), roflumilast (16.5 % ± 6.9, 38.7 % ± 4.3,
5.1 % ± 5.9) and compound 6a (13.8 % ±5.1, 37.1 % ± 7.0, 26.8 % ± 6.9) groups.
Figure 9. Histopathological examination in CA1, CA3 and DG regions of hippocampal (H&E
stain ×200, n = 5, 400× magnification) for sham, model, rolipram, CQ, roflumilast, compounds 6a
and 7a group.

ACCEPTED MANUSCRIPT
Figure 10. Denatured cell index (DCI) in CA1, CA3 and DG regions of mouse hippocampal
neurons (DCI = denatured cells/total cells H&E stain, n = 5). These results are expressed as the
means ± SD. Two-way ANOVA: *** p < 0.001 versus the model.
Conclusion.
Multitarget-directed ligands possess two or more complementary biological activities
and may represent an important advance in treating the multifaceted conditions
present in AD. Considering the importance of PDE4 inhibition and the modulation of
biometals in AD therapeutics, we developed a series of novel compounds that possess
the fragments of PDE4D (rolipram or roflumilast) and a modulator of metal-Aβ
interactions (CQ). In vitro assays demonstrate that these hybrids have good PDE4D
inhibitory activity, antioxidation properties, and the ability to modulate biometal ions.
The structure-activity-relationship (SAR) studies revealed that substituents at
5
-position of CQ are very important for PDE4D inhibitory activity. Compounds 6a
and 7a exhibited effective PDE4D inhibitory potency, biometal chelating activities
and modulation of metal-induced Aβ aggregation with improved BBB penetrability

ACCEPTED MANUSCRIPT
and metabolic stability in rat liver microsomes. Finally, an oral intake of the optimal
compounds 6a and 7a by AD mice induced by Aβ25-35 demonstrated significant
enhancements in cognitive and spatial memory performance in the Morris water-maze
test. Remarkably, 7a had better behavioral performance and could protect
hippocampal neurons against Aβ toxicity. Other studies on the pharmaceutical
properties of 7a are in progress.
Experiment section
1
13
General chemistry. H NMR and C NMR was recorded on a Bruker Avance III
spectrometer (400 or 500 MHz) with trimethylsilane as the internal standard. Melting
points were measured in a SRS-Opti melting point apparatus. Reactions progress were
monitored by thin-layer chromatography (TLC) with UV light (254 nm) for
visualization. High-resolution mass spectra (HRMS) measurements were generated on
a Shimadzu LCMS-IT-TOF mass instrument. HPLC analyses were carried out on an
Agilent 1200 series using C18 column (Agilent ZORBAX Eclipse XDB-C18 5 µm.
4
0
.6 mm × 250 mm column). The eluent was CH CN / H O = 90 / 10, at a flow rate of
3 2
.5 mL/min and the purity was record at 254 nm. The purity of all new compounds
was at least 95 %.
General procedures for the preparation of 2a-c, 3a-c and 4a-c. To a solution of
3
3
,4-dimethoxybenzoic acid or 3-(cyclopentyloxy)-4-methoxybenzoic acid, or
-(cyclopropylmethoxy)-4-(difluoromethoxy)benzoic acid in 10 mL of
dichloromethane, thionyl chloride (15 mL) and N,N-dimethylformamide (0.5 mL)

ACCEPTED MANUSCRIPT
were added. After the reaction mixture was heated at 85 °C for 3 h, the solvents were
removed under vacuum to afford the corresponding acyl chloride which was used
without any further purification. Acyl chloride (1.5 equiv.) in 2 mL of CH Cl was
2
2
added dropwise to a solution amino 1a, 1b or 1c (1.0 equiv.), and triethylamine (10.0
equiv.) in 20 mL of CH Cl . The reaction mixture was stirred at room temperature for
2
2
1
6 h. The mixture was extracted with dichloromethane, washed with water, dried over
Na SO , and concentrated under reduced pressure to afford the crude residue which
2
4
was purified using column chromatography. (petroleum ether/ dichloromethane = 3 :
1
)
Tert-butyl (5-(3,4-dimethoxybenzamido)quinolin-8-yl) carbonate (2a) Yellow solid,
1
5
6 % yield. H NMR (400 MHz, CDCl ) δ 8.95 – 8.87 (m, 1H), 8.23 (s, 1H), 8.15 (dd,
3
J = 9.0, 7.6 Hz, 1H), 7.97 (s, 1H), 7.76 (s, 1H), 7.70 (d, J = 7.7 Hz, 1H), 7.43 (dd, J =
.5, 4.1 Hz, 1H), 7.33 – 7.29 (q, J - 8.9 Hz, 1H), 2.39 (s, 6H), 1.65 (s, 9H).
8
Tert-butyl (5-(3,4-dimethoxybenzamido)-7-iodoquinolin-8-yl) carbonate (2b) Yellow
1
solid, 58 % yield. H NMR (400 MHz, CDCl ) δ 8.97 (dd, J = 4.1, 1.5 Hz, 1H), 8.23
3
(
dd, J = 8.6, 1.5 Hz, 1H), 8.09 (s, 1H), 7.87 (d, J = 8.2 Hz, 1H), 7.78 (s, 1H), 7.71 (d,
J = 7.8 Hz, 1H), 7.56 (d, J = 8.2 Hz, 1H), 7.47 (dd, J = 8.6, 4.1 Hz, 1H), 7.30 (q, J =
.9 Hz, 1H), 2.39 (s, 3H), 2.39 (s, 3H), 1.62 (s, 9H).
7
Tert-Butyl (5-(3-(cyclopentyloxy)-4-methoxybenzamido)quinolin-8-yl) carbonate (3a)
1
Yellow solid, 64 % yield. H NMR (400 MHz, DMSO-d ) δ 10.58 (s, 1H), 9.11 (d, J =
6
4
.1 Hz, 1H), 8.92 (d, J = 8.4 Hz, 1H), 8.01 (dd, J = 7.9, 5.1 Hz, 1H), 7.76 (dd, J =
3.6, 8.5 Hz, 2H), 7.69 (s, 1H), 7.57 (d, J = 8.1 Hz, 1H), 7.11 (d, J = 8.4 Hz, 1H), 4.93
1

ACCEPTED MANUSCRIPT
(s, 1H), 1.92 (m, 2H), 1.75 (m, 4H), 1.59 (m, 2H). 1.66 (s, 9H).
Tert-Butyl
carbonate (3b) Yellow solid, 68 % yield. 1H NMR (400 MHz, CDCl ) δ 8.92 (d, J =
(7-chloro-5-(3-(cyclopentyloxy)-4-methoxybenzamido)quinolin-8-yl)
3
3
.9 Hz, 1H), 8.10 (d, J = 8.5 Hz, 1H), 8.06 (s, 1H), 7.81 (s, 1H), 7.54 (s, 1H), 7.48 (d,
J = 8.4 Hz, 1H), 7.35 (dd, J = 8.5, 4.0 Hz, 1H), 6.92 (d, J = 8.3 Hz, 1H), 4.86 (s, 1H),
.93 (s, 3H), 2.14 – 1.68 (m, 8H), 1.55 (s, 9H).
Tert-Butyl (5-(3-(cyclopentyloxy)-4-methoxybenzamido)-7-iodoquinolin-8-yl)
carbonate (3c) Yellow solid, 65 % yield. 1H NMR (400 MHz, CDCl ) δ 8.84 (d, J =
3
3
3
.9 Hz, 1H), 8.09 (d, J = 13.2 Hz, 1H), 8.05 (d, J = 5.5 Hz, 2H), 7.53 (s, 1H), 7.48 (d,
J = 8.3 Hz, 1H), 7.33 (dd, J = 8.4, 4.0 Hz, 1H), 6.90 (d, J = 8.3 Hz, 1H), 4.84 (s, 1H),
.92 (s, 3H), 2.04 – 1.70 (m, 8H), 1.60 (s, 10H).
3
Tert-Butyl (5-(3-(cyclopropylmethoxy)-4-(difluoromethoxy)benzamido)quinolin-8-yl)
carbonate (4a) Yellow solid, 67 % yield. 1H NMR (400 MHz, CDCl ) δ 8.89 (dd, J =
3
4
.1, 1.5 Hz, 1H), 8.30 (s, 1H), 8.04 (dd, J = 8.6, 1.5 Hz, 1H), 7.58 (dd, J = 4.9, 2.9 Hz,
H), 7.47 – 7.40 (m, 2H), 7.35 (dd, J = 8.6, 4.1 Hz, 1H), 7.24 (d, J = 8.2 Hz, 1H), 6.75
2
(
t, J = 75.0 Hz, 1H), 3.95 (d, J = 6.9 Hz, 2H), 1.61 (s, 9H), 0.71 – 0.64 (m, 2H), 0.39
q, J = 4.8 Hz, 2H).
(
Tert-Butyl
(7-chloro-5-(3-(cyclopropylmethoxy)-4-(difluoromethoxy)benzamido)quinolin-8-yl)
1
carbonate (4b) Yellow solid, 70 % yield. H NMR (400 MHz, CDCl ) δ 8.84 (dd, J =
3
4
.1, 1.6 Hz, 1H), 8.29 (s, 1H), 7.95 (dd, J = 8.5, 1.6 Hz, 1H), 7.55 (s, 1H), 7.52 (d, J =
.9 Hz, 1H), 7.44 (dd, J = 8.3, 1.9 Hz, 1H), 7.26 (dd, J = 8.6, 4.1 Hz, 1H), 7.17 (d, J =
1