Tau-Centric Multitarget Approach for Alzheimer's Disease: Development of First-in-Class Dual Glycogen Synthase Kinase 3β and Tau-Aggregation Inhibitors

Article abstract of DOI:10.1021/acs.jmedchem.8b00610

Several findings propose the altered tau protein network as an important target for Alzheimer's disease (AD). Particularly, two points of pharmacological intervention can be envisaged: inhibition of phosphorylating tau kinase GSK-3β and tau aggregation process. On the basis of this consideration and on our interest in multitarget paradigms in AD, we report on the discovery of 2,4-thiazolidinedione derivatives endowed with such a profile. 28 and 30 displayed micromolar IC₅₀ values toward GSK-3β, together with the capacity of inhibiting AcPHF6 aggregation of 60% and 80% at 10 μM, respectively. In addition, they showed PAMPA-BBB permeability, together with a suitable cellular safety profile. 30 also displayed inhibition of both K18 and full-length tau aggregations. Finally, both compounds were able to improve cell viability in an okadaic acid-induced neurodegeneration cell model. To the best of our knowledge, 28 and 30 are the first balanced, nontoxic, dual-acting compounds hitting tau cascade at two different hubs.

Full text of DOI:10.1021/acs.jmedchem.8b00610

Subscriber access provided by Kaohsiung Medical University
Article
Tau-centric multi-target approach for Alzheimer’s disease: development of
first-in-class dual glycogen synthase kinase 3# and tau-aggregation inhibitors
Annachiara Gandini, Manuela Bartolini, Daniele Tedesco, Loreto Martínez-González, Carlos
Roca, Nuria E. Campillo, Josefa Zaldivar-Diez, Concepcion Perez, Giampaolo Zuccheri,
Andrea Miti, Alessandra Feoli, Sabrina Castellano, Sabrina Petralla, Barbara Monti,
Martina Rossi, Fabio Moda, Giuseppe Legname, Ana Martínez, and Maria Laura Bolognesi
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00610 • Publication Date (Web): 04 Aug 2018
Downloaded from http://pubs.acs.org on August 6, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Tauꢀcentric multiꢀtarget approach for Alzheimer’s disease: development of firstꢀinꢀclass dual
glycogen synthase kinase 3β and tauꢀaggregation inhibitors
(a,b)
(a)
(a)
(c)
Annachiara Gandini,
Manuela Bartolini, Daniele Tedesco, Loreto MartinezꢀGonzalez,
(c)
(c)
(c)
(d)
Carlos Roca, Nuria E. Campillo, Josefa ZaldivarꢀDiez, Concepción Perez, Giampaolo
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(a,e)
(a,e)
(f)
(f)
(a)
Zuccheri,
Andrea Miti,
Alessandra Feoli, Sabrina Castellano, Sabrina Petralla, Barbara
(a)
(b)
(g)
(b)
(c)
Monti, Martina Rossi, Fabio Moda, Giuseppe Legname, Ana Martinez, Maria Laura
,
(a)
Bolognesi*
(a) Department of Pharmacy and Biotechnology, Alma Mater Studiorum ꢀ University of Bologna,
Via Belmeloro 6, Iꢀ40126 Bologna, Italy; (b) Laboratory of Prion Biology, Department of
Neuroscience, Scuola Internazionale Superiore di Studi Avanzati (SISSA), Via Bonomea 265, Iꢀ
34136 Trieste, Italy; (c) Centro de Investigaciones Biologica, CSIC, Ramiro de Maeztu 9, 28040
Madrid, Spain; (d) Instituto de Quimica Medica, CSIC, Calle Juan de la Cierva 3, 28006 Madrid,
Spain; (e) S3 Center of the Institute of Nanosciences, Italian National Research Council (CNR); (f)
EpigeneticMedChemLab, Department of Pharmacy, University of Salerno, Via Giovanni Paolo II
132, Iꢀ84084 Fisciano, Italy; (g) IRCCS Foundation Carlo Besta Neurological Institute, via Celoria
11, Iꢀ20133 Milan, Italy.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
Several findings propose altered tau protein network as an important target for Alzheimer’s disease
AD). Particularly, two points of pharmacological intervention can be envisaged: inhibition of
(
phosphorylating tau kinase GSKꢀ3β, and tau aggregation process. Based on this consideration and
on our interest in multiꢀtarget paradigms in AD, we report on the discovery of 2,4ꢀ
thiazolidinedioneꢀderivatives endowed with such profile. 28 and 30 displayed micromolar IC50
values towards GSKꢀ3β, together with the capacity of inhibiting AcPHF6 aggregation of 60% and
80% at 10 µM, respectively. In addition, they showed PAMPAꢀBBB permeability, together with a
suitable cellular safety profile. 30 also displayed inhibition of both K18 and fullꢀlength tau
aggregations. Finally, both compounds were able to improve cell viability in an okadaic acidꢀ
induced neurodegeneration cell model. To the best of our knowledge, 28 and 30 are the first
balanced, nonꢀtoxic, dualꢀacting compounds hitting tau cascade at two different hubs.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
ACS Paragon Plus Environment

Page 3 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
. Introduction
Drug development for Alzheimer’s disease (AD) has proven to be very difficult. A 0.4% success
rate for approval is among the lowest for any therapeutic area and it has been 15 years since the last
1
drug, memantine, was launched. What is even more discouraging is that despite the enormous
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2,
social and economic rewards of a successful treatment, big pharma are pulling out of AD research.
3
Over the years, some of the complications of AD drug discovery have been related to its multiꢀ
factorial nature. Advances in systems biology have been revealing a phenotypic robustness and a
network structure for AD pathology strongly implying that singleꢀtarget compounds are inadequate
4
ꢀ6
in curing AD and preventing its progression. Conversely, a polypharmacological approach
7,
appears better suited to confront such a complex disease and to provide diseaseꢀmodifying effects.
8
In light of this, the development of multiꢀtarget directed ligands (MTDLs), i.e. small molecules
able to hit multiple targets responsible for AD pathogenesis, has been pursued as a new therapeutic
9
ꢀ12
opportunity in recent years, leading to the discovery of different promising compounds.
In spite
of the clear preꢀclinical success, no rationally designed multiꢀtarget compound has been approved
1
3
so far for the treatment of AD. Indeed, AD multiꢀtarget drug discovery combines the hurdles of an
extremely challenging area, with those of novel pharmaceutical tools.
Clearly, to develop clinicallyꢀeffective MTDLs, the selection of appropriate targets is of utmost
importance. They should be causally linked to the most relevant AD pathological processes and
1
3
their ‘networking’ is critical for determining the final properties of the MTDL.
14
The amyloid hypothesis has long been the central dogma in AD drug discovery, leading to many
1
5ꢀ17
(
either single or multiꢀtarget)
drug candidates. However, recently this hypothesis has been
18ꢀ22
questioned on the basis of the repeating failures of amyloidꢀβ (Aβ)ꢀbased drugs.
In parallel,
several findings are suggesting that the altered tau network is more significant than the Aβ one from
23
a pathogenic point of view. Strikingly, the very close spatial correlation between neuronal loss
and tau aggregates in patients’ brains, and the genetic evidence for the role of tau aggregation in a
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
wide range of neurodegenerative disorders due to mutations in the tau gene, provides a strong
24
foundation to a tauꢀtargeted therapy.
Tau is a microtubuleꢀassociated protein (MAP), predominantly expressed in the neuronal axons
where it promotes microtubule stabilization and facilitates axonal transport of proteins, trophic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
25, 26
factors, and other cellular constituents, including neurotransmitters.
Under physiological
conditions, tau is in a dynamic equilibrium between the microtubuleꢀbound and ꢀunbound states,
regulated by specific kinases and phosphatases. However, under pathological conditions,
hyperphosphorylated tau loses its microtubules stabilizing and axonal transport functions.
Particularly, hyperphosphorylated tau becomes prone to misfolding and aggregation, ultimately
resulting in the formation of neurofibrillary tangles (NFTs) (Figure 1). The tau aggregation cascade,
once initiated, is selfꢀreplicating, depleting the physiological tau pool in the neuron, and converting
2
5, 26
it to toxic NFTs, which continue to accumulate, leading to neuronal death.
In view of all these
evidences, tau cascade is being recognized as a promising target for the development of drug
2
7ꢀ29
candidates for the treatment of AD.
Building on these considerations, we aimed to identify novel MTDLs able to hit tau cascade at two
validated hubs, namely by inhibiting glycogen synthase kinase 3β (GSKꢀ3β) and tau aggregation
process (Figure 1).
GSKꢀ3β is a multifunctional serine/threonine kinase that phosphorylate tau mainly at the Ser199,
3
0
Ser396, and Ser413 sites. In AD, GSKꢀ3β overꢀexpression induces tau hyperphosphorylation, and
consequently, its disjunction from the microtubules and its aggregation. Moreover, GSKꢀ3β has an
3
1
important role in other AD key features such as inflammation and apoptosis.
Over the past decade, an increased interest in GSKꢀ3β led to the discovery of several inhibitors,
with different chemical scaffolds and mechanisms of action (i.e. ATP competition, allosteric
30, 31
modulation, and irreversible inhibition).
There was considerable rationale from preclinical
30, 31
models that GSKꢀ3β inhibitors could be diseaseꢀmodifying drugs.
Importantly, the ATP nonꢀ
competitive GSKꢀ3β inhibitor tideglusib (1 in Figure 2), is currently under clinical development for
4
ACS Paragon Plus Environment

Page 5 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
AD and other tauopathies and has demonstrated no safety concerns even after longꢀterm
31
treatments.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Schematic depiction of tau cascade, tau structure, and points of pharmacological
intervention.
On the other side, agents capable of preventing tau selfꢀassembly represent another attractive
32
strategy for the treatment of AD. Importantly, phenothiazine derivatives methylene blue (2 in
3
3, 34
Figure 2) and LMTX (3 in Figure 2) are in clinical phase of AD drug development.
Figure 2. Chemical structures of tauꢀdirected AD investigational drugs 1ꢀ3.
Herein, we report on the development of the firstꢀinꢀclass MTDLs 4ꢀ38 (general structure in Figure
) which, by simultaneously inhibiting GSKꢀ3β and the tau aggregation process, may pave the way
3
for a completely novel AD treatment. In contrast with the available singleꢀtargeted drug candidates,
their superior potential efficacy is based on the assumption that, by targeting two points of the tau
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
network, they could trigger a synergistic response, and eliminate compensatory reactions and
feedback controls, thereby overcoming the pathway robustness. Furthermore, in principle, such dual
inhibitors should possess a lower targetꢀrelated toxicity. Because most links in neuronal networks
are weak, it can be inferred that even lowꢀaffinity dual inhibitors should be sufficient to achieve a
significant antiꢀtau effect. If the concentration required to obtain the desired effect is lower with
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
5
respect to singleꢀtargeted drugs, the risk for toxicity is reduced. This greatly eases the constraints
31
of druggability, which is a key issue for GSKꢀ3β.
To note, although interesting examples of MTDLs that combine antiꢀtau activity with a second one
beneficial for AD treatment have been recently reported (e.g. antiꢀamyloid, antiꢀoxidant, antiꢀ
36ꢀ46
inflammatory activity),
4ꢀ38 are the first compounds able to hit tau cascade at two hubs.
2. Design of dual GSKꢀ3β/tau aggregation inhibitors
Designing a ligand able to simultaneously bind two targets sharing no similarity in the binding sites,
47
as it is the case of GSKꢀ3β and tau protein, could be very challenging. Likewise, combining the
molecular frameworks of two CNSꢀdirected singleꢀtarget ligands (or the corresponding underlying
pharmacophores) in one small molecule might not be an easy task, especially in terms of
13
pharmacokinetics optimization.
6
ACS Paragon Plus Environment

Page 7 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 3. General chemical structures of TDZD, 2ꢀiminothiazolidinꢀ4ꢀone, hydantoin, rhodanine,
thiohydantoin, and TZDꢀderivatives 4ꢀ38.
In this context, 5ꢀmembered heterocyclic fragments caught our attention. Several known GSKꢀ3β
inhibitors carry 5ꢀmembered heterocyclic rings (i.e. 2ꢀiminothiazolidinꢀ4ꢀone and hydantoin, Figure
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
48, 49
3) with hydrogen bond acceptor and donor functionalities.
Thiadiazolidinediones (TDZDs,
exemplified by 1) have been reported as the first class of nonꢀATP competitive inhibitors of GSKꢀ
5
0
3β. Intriguingly, rhodanine, thiohydantoin and hydantoin isosters (Figure 3) were discovered by
5
1
Mandelkow and Waldmann as effective tau antiꢀaggregation motifs. These derivatives inhibited
the formation of tau aggregates in a doseꢀdependent fashion, as well as destabilized preformed tau
aggregates. The ability of rhodanines and thiohydantoins to directly interact with tau aggregates has
been confirmed by the tau imaging properties of derivatives RH1 ((Z)ꢀethyl 2ꢀ(5ꢀ((5ꢀ(3ꢀ
iodophenyl)furanꢀ2ꢀyl)methylene)ꢀ4ꢀoxoꢀ2ꢀthioxothiazolidinꢀ3ꢀyl) acetate) and TH2 ((Z)ꢀethyl 2ꢀ
52
(4ꢀ((5ꢀ(3ꢀiodophenyl)furanꢀ2ꢀyl)methylene)ꢀ5ꢀoxoꢀ2ꢀthioxoimidazolidinꢀ1ꢀyl) acetate). Thus, we
reasoned that the crossꢀtalk between all these fragments could be favorably exploited to identify
MTDLs directed to the two selected targets (i.e. GSKꢀ3β and tau protein).
In principle, being prone to bind a large number of targets with weak or moderate affinity,
5
3
rhodanines might be an optimal starting point for ligandꢀbased multiꢀtarget endeavors.
Nevertheless, the validity of rhodanine scaffold is questionable because of its notorious Pan Assay
53
Interference Compounds (PAINS) behavior. Although it is hard to draw the line between true
5
4ꢀ56
multiꢀtarget activity and aggregation effects or potential reactivity, which may or may not
to assay interference, we considered safer to disregard the rhodanine scaffold.
lead
Conversely, among the 5ꢀmembered heterocycle isosters, we selected the 2,4ꢀthiazolidinedione
(TZD, Figure 3) scaffold, which has never been explored as either GSKꢀ3
β
or tau antiꢀaggregation
57ꢀ59
fragment. As highlighted by Baell and Pouliot,
the TZD core itself does not aggressively drive
assay interference in the same way as rhodanine does. Indeed, TZD is not identified as a PAIN by
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
computational filters (see “PAINS analysis” in Experimental Section). To support the drugꢀlikeness
of the TZD scaffold in AD, the TZDꢀbased antidiabetic drug pioglitazone (PIO) is currently under
evaluation in a Phase 3 clinical trial as an antiꢀinflammatory drug in mild cognitive impaired
60
patients.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Starting from the TZD, as a first step, we decided to decorate it at position 5 with different aromatic
and heteroaromatic substituents, affording a 35ꢀmembered library of 5ꢀarylideneꢀ2,4ꢀ
thiazolidinediones (4ꢀ38, Table 1). This was based on previous studies showing that (i) 5ꢀarylidene
substitution improved affinity in a series of 2ꢀiminothiazolidinꢀ4ꢀone competitive GSKꢀ
3β
inhibitors; (ii) due to their volume and size, they do not fit in similar regions of homologous
49
kinases, therefore showing selectivity for GSKꢀ3
β
. In addition, this would allow to maintain the
61
planarity and aromaticity features, critical for interaction with tau fibrils. Among the possible
aromatic and heteroaromatic substituents, we included those (e.g. indoles, benzofurans and
6
2
benzothiophenes) reported as effective antiꢀamyloid scaffolds.
3. Chemistry
The synthesis of 4ꢀ38 was easily achieved through a oneꢀpot, optimized version of the Knoevenagel
condensation. The conventional protocol based on the reaction of equimolar amounts of TZD and
63
an aromatic aldehyde, in the presence of catalytic amounts of pyridine and acetic acid, suffers
from several disadvantages, such as long reaction times, low to moderate yields, use of toxic
solvents. To overcome these drawbacks, a solventꢀfree procedure using ethylenediamine diacetate
64
(
EDDA) as an inexpensive and effective Brönsted acidꢀbase catalyst has been proposed. Clearly,
there are several advantages in using a solventꢀfree reaction, such as reduced pollution, lower costs,
simplicity in process and handling, shorter reaction times, and higher yields. With the aim of further
improving the green chemistry aspects, we performed the reaction under microwave (MW)
irradiation. Thus, a newly optimized procedure based on the solventꢀfree condensation of aromatic
aldehydes with TZD (1:1) in the presence of EDDA (0.5 eq), under MW irradiation at 80°C, for 30
8
ACS Paragon Plus Environment

Page 9 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
minutes (Scheme 1) was developed. It was successfully applied to the vast array of aldehydes 4aꢀ
38a (with electronꢀwithdrawing and electronꢀdonating groups, heteroaromatic, conjugated and
planar rings) to obtain target compounds 4ꢀ38 (see Table 1 for structures). To the best of our
knowledge, microwaveꢀmediated EDDAꢀcatalyzed Knoevenagel condensation has never been
exploited so far for the synthesis of 5ꢀarylideneꢀ2,4ꢀthiazolidinediones.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Optimized Knoevenagel reaction for the synthesis of TZDꢀderivatives 4ꢀ38.
All the aromatic aldehydes were purchased from commercial vendors, except 14a and 15a. These
6
5, 66
were synthesized by slightly modified reported procedures (Scheme 2).
In the first case, 2ꢀ
fluorobenzaldehyde (39) was reacted with dimethylamine in the presence of potassium carbonate to
provide aldehyde 14a. In the second case, the commercially available 3ꢀ(dimethylamino)benzyl
alcohol (40) was oxidized under Swern conditions, affording aldehyde 15a.
Scheme 2. Synthetic procedure for the synthesis of aldehydes 14a and 15a.
1
13
All compounds were characterized using analytical (HPLC) and spectroscopic data ( Hꢀ and Cꢀ
NMR, ESIꢀMS), which are collected in the Experimental Section and Supporting Information.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4. Results and discussion
To verify the capability of the synthesized molecules to simultaneously hit the two targets and to
make a judicious use of resources, we designed a screening pipeline, whose main goals were: (i) to
make a preliminary screening of all synthesized molecules for GSKꢀ3β activity; (ii) to predict
bloodꢀbrain barrier (BBB) permeability of the active GSKꢀ3β inhibitors; (iii) to filter out toxic
compounds; (iv) to assess the most promising compounds for their antiꢀtau activity; (iv) to confirm
the antiꢀtau activity in a cellular model.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
GSKꢀ3β inhibition assays
GSKꢀ3β biochemical inhibition for 4ꢀ38 was first assessed using the KinaseꢀGlo luminescent assay,
6
7
which quantifies the decrease in ATP levels following the kinase reaction. All compounds were
first tested at 10 ꢁM, and hits with percentage inhibition ≥50% were further tested to obtain IC50
values (determined by using the linear regression parameter). TDZD8 (4ꢀbenzylꢀ2ꢀmethylꢀ1,2,4ꢀ
5
0
thiadiazolidineꢀ3,5ꢀdione) was used as reference compound. Encouragingly, sixteen out of the
starting thirtyꢀfive TZDs inhibited GSKꢀ3β with IC50 values in the singleꢀdigit micromolar range
(
Table 1). Compound 30 ((Z)ꢀ5ꢀ((5ꢀmethoxyꢀ1ꢀmethylꢀ1Hꢀindolꢀ3ꢀyl)methylene)thiazolidineꢀ2,4ꢀ
dione) was the most active, with a submicromolar IC50 of 0.89 ± 0.21 µM.
Table 1. GSKꢀ3β inhibition evaluation of TZDꢀderivatives 4ꢀ38 (values are the mean of three
independent experiments).
1
0
ACS Paragon Plus Environment

Page 11 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
cmpd
no.
GSKꢀ3β IC50
cmpd
no.
GSKꢀ3β IC50
(µM)
R
R
(µM)
4
5
>10
22
23
7.99 ± 0.57
7.47 ± 0.85
>10
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
7
8
9
>10
24
25
26
27
>10
>10
>10
>10
8.99 ± 0.78
7.08 ± 0.51
6.01 ± 0.72
10
9.22 ± 0.90
>10
28
29
4.93 ± 0.66
>10
11
12
>10
>10
30
31
32
0.89 ± 0.21
>10
1
1
3
4
9.22 ± 0.89
6.03 ± 0.44
15
8.54 ± 0.81
33
>10
1
1
1
6
7
8
>10
>10
34
35
36
>10
7.01 ± 0.61
>10
8.75 ± 0.77
19
20
1.42 ± 0.12
>10
37
38
6.13 ± 0.43
3.55 ± 0.23
0.69 ± 0.09
21
>10
TDZD8
From the data reported in Table 1, preliminary structureꢀactivity relationships (SAR) can be drawn.
In comparison with unsubstituted 4 (IC50 > 10 µM), hydroxyl substituents in ortho and meta
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
positions of the phenyl ring did not increase enzyme inhibition, as evident for 5 and 6. On the other
side, ꢀOH substitution in para improved the inhibitory activity, as shown by 7 (8.99 ± 0.78 µM).
Diꢀsubstituted 8ꢀ10 did not show any significant enhancement with respect to 7. Addition of the
fluorine atom (11ꢀ13) seemed not to affect the activity. Regarding the dimethylamino substituent,
the favored positions are the ortho and meta, as shown by 14 (9.22 ± 0.89 µM) and 15 (8.54 ± 0.81
µM). Replacement of the phenyl of 4 with a 4ꢀpyridil ring, as in 19, led to an IC of 1.42 ± 0.12
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
µM. Thiophene and benzothiophene substitutions (22, 23) also resulted in a good inhibitory
activity. In contrast, substitution with bulky moieties such as naphthyl, anthracene, phenylpropylen
or biphenyl, as in 24ꢀ27, 33, and 34, resulted in compounds inactive at 10 µM. Interestingly,
bicyclic Nꢀheterocycles, such as 28ꢀ32, showed different behaviors. More hydrophilic 28 and 32 are
better GSKꢀ3β inhibitors than 29. Moreover, the insertion of a methyl group on the nitrogen atom of
the indole, as in 30, resulted in an important improvement in the inhibitory profile (0.89 ± 0.21
µM). This trend in activity had already been described for previously reported indolylꢀmaleimide
6
8
series of GSKꢀ3β inhibitors. A retrospective computational study was performed to rationalize
these results and predict compound binding mode (see Supporting Information).
The GSKꢀ3β inhibitory activity of selected inhibitors 28 and 30 (vide infra) was also confirmed in a
TimeꢀResolved Fluorescence Energy Transfer (TRꢀFRET) assay. 28 and 30 were tested at ten
different concentrations and the resulting doseꢀresponse curves are depicted in Figure 4. The
obtained IC50 values for both 28 (21.0 ± 2.4 ꢁM) and 30 (5.5 ± 0.8 ꢁM) were in agreement with the
ones obtained with the KinaseꢀGlo luminescent assay. Thus, we were able to prove the genuine
GSKꢀ3β inhibitory activity of 28 and 30. In addition, because of the presence of the alkylidene
moiety, prone to potential covalent reactions (see “PAINS analysis” for further discussion), we
aimed to assess the reversibility of GSKꢀ3β inhibition. Thus, we performed a jumpꢀdilution
94
experiment on 30, demonstrating that the enzymatic activity was restored after dilution, pointing
to a reversible inhibition by 30 (see Figure S7).
12
ACS Paragon Plus Environment

Page 13 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Doseꢀresponse curves of GSKꢀ3β activity in the presence of 28 and 30. Values are the
means ± SD determined for at least two independent experiments.
To further investigate the mechanism of action, 28 and 30 were subjected to kinetic analysis.
Particularly, we determined the competition with the ATP. We varied both ATP and 28 and 30
concentrations with a constant concentration of GSꢀ2, the substrate used in the enzymatic reaction.
The doubleꢀreciprocal plot of data (Figure 5) indicates that both 28 and 30 act as competitive
inhibitors of ATP binding. In light of this, TZD derivatives represent a new chemical class of ATPꢀ
competitive GSKꢀ3β inhibitors.
Figure 5. Kinetic data for the TZDꢀderivatives 28 and 30. ATP concentration in the reaction
mixture varied from 1 to 50 ꢁM. Compounds concentrations used are depicted in the plot, and the
concentration of GSꢀ2 was kept constant at 25 ꢁM. Each point is the mean of two different
experiments, both analyzed in triplicate.
Blood Brain Barrier permeation
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
One of the main obstacles for CNS drug discovery is the drug penetration into the BBB at
6
9
therapeutic concentrations. Parallel Artificial Membrane Permeability Assay (PAMPA)ꢀBBB is a
high throughput technique developed to predict BBB passive permeability. In order to preliminary
explore the capacity of GSKꢀ3β inhibitors 7, 8ꢀ10, 14, 15, 18, 19, 22, 23, 28, 30, 32, 35, 37, 38 to
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
70
penetrate into the brain, we used the PAMPAꢀBBB method, which employs a brain lipid porcine
membrane. The in vitro effective permeability (P ) of commercial drugs through lipid membrane
e
extract together with TZDꢀderivatives was determined. An assay validation was made comparing
the reported permeability values of commercial drugs with the experimental data. A good
correlation between experimentalꢀdescribed values was obtained (Figure S6). From this equation
71
and following the pattern established in the literature for BBB permeation prediction, we could
ꢀ6
classify compounds as CNS+ when they present a permeability > 3.07 x 10 cm/s. Based on these
results, only compounds 14, 15, 23, 28, and 30 are predicted to cross the BBB by passive
permeation (Table S2), and thus were progressed for toxicity and tau antiꢀaggregation analysis.
Neuroꢀ and hepatoꢀtoxicity assessment
The failure of several AD drug discovery approaches has been attributed not only to the lack of
1
efficacy, but also to the toxicity of the drug candidates after a chronic treatment. In consideration
of this, and motivated by the promising in vitro results, we first tested neurotoxicity on the five
GSKꢀ3β inhibitors predicted to cross the BBB, in comparison to PIO (Figure 6).
Primary cultures of cerebellar granule neurons (CGNs) are an useful in vitro models to study
7
2
neuronal death. MTT (3ꢀ(4,5ꢀdimethylthiazolꢀ2ꢀyl)ꢀ2,5ꢀdiphenyltetrazolium bromide) assessment
of CGNs viability after 24 h treatment with 14, 15, and 23 showed a clear toxic effect for all the
three compounds, as well as PIO, even at 5 ꢁM concentration. Notably, 28 and 30 showed a very
low neurotoxicity only at the highest tested concentrations (50 ꢁM).
In AD patients, aging, comorbidity, and subsequent polytherapy significantly contribute to increase
the risk of drugꢀdrug interactions and, consequently, hepatotoxicity. Moreover, although being
14
ACS Paragon Plus Environment

Page 15 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
7
3
effective therapeutic agents, some TZD derivatives have been reported to show hepatotoxicity. To
this end, parallel experiments were performed on 14, 15, 23, 28, and 30 in a human hepatoma cell
line (HepG2) in comparison with PIO (Figure 6). After 24 h incubation at 0ꢀ50 ꢁM, a
concentrationꢀdependent decrease in cell viability was observed for PIO, 14, and 15.
Encouragingly, compounds 23, 28, and 30 showed no hepatotoxicity, even at 50 ꢁM. On the basis
of the toxicity profiles, indole derivatives 28 and 30 emerged as the most drugꢀlike compounds.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. (A) Neurotoxicity on primary rat cerebellar granule neurons (CGNs) after 24 h treatment.
(
B) Hepatotoxicity on human hepatoma cells (HepG2) after 24 h treatment. Results are expressed as
percentage of controls and are the mean ± SE of at least 3 different experiments, each run in
triplicate.
AcPHF6 aggregation and inhibition studies by Thioflavin T (ThT) fluorescence, circular dichroism
spectroscopy and atomic force microscopy
In parallel, the ability of PAMPAꢀBBB permeable 14, 15, 23, 28, and 30 to interfere with tau
aggregation in vitro was investigated, using the AcPHF6 peptide (306VQIVYK311) as a model
system. Indeed, AcPHF6 is a short segment found in the third repeat of the microtubuleꢀbinding
region of tau protein (Figure 1). As recently highlighted by cryoꢀelectron microscopy of tau
filaments, 306VQIVYK311 spatially arranges into a betaꢀstrand conformation and contributes to the
74
formation of the Cꢀshaped crossꢀβ structure representing the filament core of PHFs. As AcPHF6
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
undergoes to spontaneous fibrillation, it has been proposed as a suitable model for screening small
7
5ꢀ80
molecule inhibitors of tau aggregation.
As for other βꢀsheet proteins, the aggregation process
can be monitored by fluorescence in the presence of ThT, which displays a characteristic change in
81
the emission spectrum in the presence of βꢀrich structures. The antiꢀaggregating activity of 14, 15,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
23, 28, and 30 was therefore monitored in phosphate buffer 50 mM pH 7.4 using ThT as detection
75ꢀ80
dye. In agreement with previous studies,
when incubated alone, AcPHF6 rapidly aggregated, as
confirmed by the increase in the fluorescence intensity over a period of 80 min (Figure 7). Coꢀ
incubation of AcPHF6 with the selected compounds led to a significant (>50%) decrease of the
fluorescence intensities, indicating that all compounds could significantly interfere with AcPHF6
aggregation, even if at different extents (from 54.7% to 79.8%). In particular, derivatives 15 and 30
showed the highest inhibitory potencies reducing the fluorescence signal at plateau by almost 80%
(79.5 and 79.8%, respectively) and resulting only slightly less potent than the known inhibitor
82
myricetin. Figure 7B shows the details of the kinetics of aggregation of AcPHF6 incubated alone
and in the presence of compounds 28 and 30, namely the derivatives endowed with the best safety
profile (see above).
Figure 7. (A) Inhibition of AcPHF6 peptide (50 ꢁM) aggregation in the presence of selected
derivatives (10 ꢁM) after 80 min incubation. The data are the mean of at least three repeats. (B)
1
6
ACS Paragon Plus Environment

Page 17 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Kinetics of the AcPHF6 aggregation in the absence and in the presence of compounds 28 and 30
monitored by ThT fluorescence.
Circular dichroism (CD) studies were then carried out to gain further insights into the mechanism of
inhibition of 28 and 30 and to exclude false positive outcomes due to the possible displacement of
bound ThT by the tested compounds. The CD spectrum of AcPHF6 (Figure 8) shows a negative
band at 221 nm followed by a positive band at shorter wavelengths, both of which are typical
features of βꢀsheet structures. A decrease in intensity was observed for the CD spectrum of AcPHF6
in the presence of 28, suggesting that the abundance of βꢀsheet structures is lower. On the other
hand, the CD profile of AcPHF6 in the presence of 30 lacks for the typical bands of βꢀsheet
structures, with the peptide displaying an unordered conformation. Both behaviors are consistent
with the inhibitory trends highlighted by the fluorescenceꢀbased assay; therefore, CD data confirm
the inhibitory capacity of derivatives 28 and 30 and suggest that their antiꢀaggregating activity is
related to their ability to stabilize the AcPHF6 peptide in a less fibrillogenic conformation, with
derivative 30 being much more effective in doing so.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. CD spectra of AcPHF6 (100 µM) in the absence (solid) and in the presence of 28 (20 µM,
dashed) or 30 (20 µM, dotted), measured after 2 h incubation. Solvent: phosphate buffer (50 mM
pH 7.4)/water/methanol/DMSO 89:10:0.95:0.05 (v/v/v/v). Pathlength: 1 mm.
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Atomic force microscopy (AFM) studies were also performed to examine the morphology of the
fibrils formed by a 50 µM phosphate buffer solution of AcPHF6 peptide alone and in the presence
of 28 and 30, when prepared in the same conditions employed for the aboveꢀdescribed ThT studies
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(but in the absence of ThT). After incubation, aliquots of the solutions were subjected to
ultrafiltration to separate the low molecular weight unꢀpolymerized hexapeptide fraction from the
highꢀmolecular weight (HMW) fiber fraction, and such HMW fraction was adsorbed on the surface
of mica for AFM imaging. We believe that filtering out unꢀpolymerized peptides at the time of
AFM specimen preparation could avoid some confusing results sometimes found in AFM data. As
evident in the micrographs in Figure 9, long and thick amyloid fibrils were imaged after adsorbing
on mica the solution of AcPHF6 incubated in the absence of inhibitors (Figure 9A). Here, fibers
with individual diameters of 10 nm or more (and lengths of several hundreds of nanometers) were
found in bundles and higherꢀorder aggregates. On the other hand, only much smaller structures
(possibly oligomers or protofibrils) were observed from the adsorption of the hexapeptide incubated
in the presence of the inhibitors 28 and 30 (Figure 9B and C). We can hypothesize that such lower
molecular weight aggregated structures could be responsible for the small but nonꢀzero
fluorescence signal recorded in the ThTꢀbased assay when the hexapeptide was incubated in the
presence of 28 and 30.
Figure 9. AFM micrographs of amyloid fibers obtained from the aggregation of 50 µM AcPHF6
peptide in phosphate buffer in the absence (A), and in the presence of 10 µM compound 28 (B) and
1
8
ACS Paragon Plus Environment

Page 19 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compound 30 (C). The height of fibers on the mica substrate is coded according to the common
included color table (ranging from 0 to 50 nm of heights) as to allow visual comparison of the
different size of the adsorbed structures. Insets of panels B and C: the rectangular portions of panels
B and C situated at the inset location in the micrograph (and of the size of the inset) are displayed
according to the reported magnified 0ꢀ5 nm color lookꢀup table in order to better show the smaller
and thinner adsorbed structures.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
K18 and fullꢀlength tau aggregation and inhibition studies
Encouraged by the results obtained with the AcPHF6 peptide, we decided to confirm the antiꢀ
aggregating properties of the most active 30 in more clinically relevant models of tau aggregation.
The tau fibrillization process can be replicated in vitro, however, recombinant tau shows very little
intrinsic tendency to aggregate due to the lack of a series of postꢀtranslational modifications
required for its aggregation. Indeed, during the last years, recombinant truncated tau isoforms are
more frequently used for in vitro tau aggregation, where the addition of anionic coꢀfactors (e.g.
heparin, arachidonic acid) results in the formation of fibrils that resemble those observed in the
8
3, 84
brains of AD patients.
Thus, 30 was firstly evaluated in a heparinꢀinduced tau assembly assay,
in which the fibrillization of the truncated K18 tau fragment (comprising four microtubulesꢀbinding
repeats, Figure 1) was monitored by ThT. The aggregation of the tau fragment K18 was achieved
85
through the RealꢀTime QuakingꢀInduced Conversion (RTꢀQuIC) technique. As shown in Figure
10, K18 alone was completely aggregated after 11ꢀ12 h and the fluorescence intensity reached the
highest value, while the treatment with 30 (1, 5, 10 ꢁM) resulted in a delay of K18 selfꢀaggregation.
Moreover, the analysis of the averaged fluorescence intensities at the plateau of the kinetic showed
a 35% inhibition of the aggregation process.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 10. RTꢀQuIC analysis of 30 activity on tau K18 selfꢀaggregation monitored by ThT
fluorescence.
Motivated by the promising results, we also tested the inhibitory activity of 30 towards the heparinꢀ
induced selfꢀaggregation of fullꢀlength 2N4R tau, through RTꢀQuIC technique. Treatment of 2N4R
tau with 30 (10 ꢁM) did not show a delay in the aggregation process (Figure 11). However, coꢀ
incubation with 30 led to a lower amount of formed aggregates, as indicated by a 30% lower
plateau signal. Importantly, this result is in agreement with that obtained with K18 fragment,
confirming the potential of 30 as tau antiꢀaggregating compound.
2
0
ACS Paragon Plus Environment

Page 21 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 11. RTꢀQuIC analysis of 30 activity on fullꢀlength tau selfꢀaggregation monitored by ThT
fluorescence.
Okadaic acidꢀinduced tau hyperphosphorylation cell model
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Increasing evidence has suggested that inhibition of phosphatases by okadaic acid (OA) represents
the most robust way to induce tau hyperphosphorylation. Thus, OAꢀtreated cell lines and primary
neuronal cultures have been used as established cellular models of hyperphosphorylated tauꢀ
8
6, 87
induced neurodegeneration.
In addition, OAꢀtreated rodents recapitulate the morphological
8
8
hallmarks of AD.
To confirm the antiꢀtau profile of 28 and 30 in a cellular context, we further explored their rescue
50
potential in the OAꢀinduced neurodegeneration cell model, in comparison with TDZD8.
We studied the effect of 28 and 30 and TDZD8, at two different concentrations (10 and 20 ꢁM), in
a human neuroblastoma cell line (SHꢀSY5Y) treated with OA (Figure 12). As expected, OA
induced a decrease in cell viability higher than 50%. Importantly, both 28 and 30 were able to
increase the cell viability probably due to the decrease on tau phosphorylation by GSKꢀ3β
inhibition. Particularly, 28 was only slightly less effective than TDZD8. To note, other indole
43
derivatives have been reported as effective in counteracting OAꢀinduced neurotoxicity.
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 12. OAꢀinduced neurodegeneration cell model. Effects of treatment with 28 and 30. Results
are mean value ± SEM, represented as relatives to control value (=100), of 6 replicates from three
independent experiments.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Inhibitory activity profile towards casein kinase 1 (CK1) δ and ε, and cell division cycle 7 (Cdc7)
89
Finally, to preliminarily address kinase promiscuity issues of ATP competitive inhibitors, we
tested 28 and 30 on three selected kinases, i.e. casein kinase 1 (CK1) δ and ε, and cell division cycle
7 (Cdc7) kinase, available in our laboratories. Cdc7 is another representative cell cycle regulation
90
kinase, which should show low binding affinity toward new GSKꢀ3β hits to avoid possible
toxicities. CK1δ and CK1ε represent GSKꢀ3β phylogenetically diverse kinases, able to
91
phosphorylate tau.
Therefore, the inhibition of 28 and 30 toward CK1δ, CK1ε, Cdc7 and GSKꢀ3β were assessed at 10
ꢁM. Encouragingly, 28 and 30 only poorly, or not inhibited at all, CK1δ, CK1ε, and Cdc7 (Figure
13), while they inhibited GSKꢀ3β by 95% and 100%, respectively. This reveals a preliminary
selectivity of both 28 and 30 towards GSKꢀ3β.
Figure 13. Inhibitory activity profile of 28 (blue) and 30 (purple) towards human CK1δ, CK1ε,
Cdc7 and GSKꢀ3β kinases, at a fixed concentration of 10 ꢁM.
5. Conclusions
2
2
ACS Paragon Plus Environment

Page 23 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We have been interested in MTDLs that can confront the multifactorial nature of AD for years.
Recently, in accordance with the drug discovery trends, we argued that a single molecule able to act
at two tau validated targets belonging to tau cascade might have clinical benefit and show higher
therapeutic potential than those addressing amyloid cascade.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Although in principle very promising, this approach must deal with the problem of targeting two
proteins with no binding site similarity.
On the basis of structural resemblances, as well as the available SAR data for GSKꢀ3β and amyloid
aggregation inhibitors, we were able to select the thiazolidinedione as a starting scaffold and to
design a novel, firstꢀinꢀclass chemical series of dual inhibitors.
A library of thirtyꢀfive derivatives was synthesized and tested in a pipeline tailored to our academic
setting. This has delivered two hit molecules (28 and 30) that display features we consider relevant
in terms of MTDL drug discovery principles: (i) they are low molecular weight fragment hits with
evident advantages in terms of intended “promiscuity” and pharmacokinetic properties; (ii) their
activity is balanced; (iii) they show no significant toxicity, even towards mammalian primary
neuronal cells.
Thus, although very preliminary, the collected results show that it is possible to design small
molecules that act specifically at two structurally different tau targets, like a kinase and a fibrillar
protein, with a good potency and in the same concentration range to yield compounds effective at a
cellular level.
These hits, targeting two independent but interrelated activities, might serve as tools to validate a
completely new tauꢀcentric network approach as a potential new diseaseꢀmodifying strategy to treat
AD.
6. Experimental section
Chemistry
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
All the commercial available reagents and solvents were purchased from SigmaꢀAldrich, Alpha
Aesar, VWR, and TCI, and used without further purification. Reactions were followed by analytical
thin layer chromatography (TLC), on preꢀcoated TLC plates (layer 0.20 mm silica gel60 with a
fluorescent indicator UV254, from SigmaꢀAldrich). Developed plates were airꢀdried and analyzed
under a UV lamp (UV 254/365 nm). CEM Discover SP focused microwave reactor was used for
microwave mediated reactions. Nuclear magnetic resonance (NMR) experiments were run on
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
13
1
13
Varian VXR 400 (400 MHz for H, 100 MHz for C). H and C NMR spectra were acquired at
300 K using deuterated dimethyl sulfoxide ((CD SO) and chloroform (CDCl ) as solvents.
3
)
2
3
Chemical shifts (δ) are reported in parts per million (ppm) relative to tetramethylsilane (TMS) as
internal reference, and coupling constants (J) are reported in hertz (Hz). The spin multiplicities are
reported as s (singlet), br s (broad singlet), d (doublet), t (triplet), q (quartet), and m (multiplet).
Mass spectra were recorded on a VG707EHꢀF apparatus, and electrospray ionization (ESI) both in
positive and negative mode was applied. Compounds were named following IUPAC rules as
applied by ChemBioDraw Ultra (version 14.0). All of the final compounds showed ≥95% purity by
92
analytical HPLC (see Table S3). As reported in literature, the TZDꢀderivatives 4ꢀ38 were obtained
1
as single Z isomers. Indeed, their HꢀNMR spectra shows only one signal attributable to the
13
resonance of the 5ꢀmethylidene proton in the range 7.50ꢀ8.70 ppm; while in their CꢀNMR spectra
the 5ꢀmethylidene carbon and the C5 of the TZD ring resonated in the ranges 130.5ꢀ140.7 and
117.5ꢀ128.0 ppm, respectively.
PAINS analysis
Aware of the liabilities that might arise from the alkylideneꢀTZD scaffold, we have analyzed 4ꢀ38
9
3
for known classes of assay interference compounds. First, 4ꢀ38 were not recognized as PAINS
according to the Free ADMEꢀTox Filtering Tool (FAFꢀDrugs4) program (http://fafdrugs4.mti.univꢀ
parisꢀdiderot.fr/), and to the “False Positive Remover” software (http://www.cbligand.org/PAINS/),
neither as aggregators according to the software “Aggregator Advisor” (http://advisor.bkslab.org/).
However, FAFꢀDrugs4 labeled them as covalent inhibitors due to their electrophilic nature. Thus,
24
ACS Paragon Plus Environment

Page 25 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
we provided evidences that the activities reported herein are highly likely not caused by pan assay
interference.
First, the genuine GSKꢀ3β inhibitory activity of 28 and 30 has been confirmed in two assays (i.e.
KinaseꢀGlo and TRꢀFRET). Possible promiscuous kinase inhibition was preliminary ruled out by
the fact that 28 and 30 do not inhibit similar kinases (CK1δ, CK1ε) and Cdc7.
As regards tau aggregation inhibition, activities of 28 and 30 has been evaluated in three different
assays (i.e. ThTꢀbased fluorometric assay, CD, and AFM), which show consistent results.
Concerning the potential unspecific covalent reactions with proteins of alkylideneꢀTZD, we first
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9
5
investigated the thiolꢀtrapping reactivity of 28 and 30 in a simple NMRꢀbased assay. Reaction of
Michael acceptorꢀbearing compounds with two equivalents of 2ꢀaminoethanethiol (cysteamine), a
96
biologically relevant model thiol, in DMSO should lead to the instantaneous formation of the
corresponding Michael adduct and the simultaneous disappearance of the olefin signals. In contrast
97
to what is reported for 5ꢀbenzylidene barbiturates, reaction of 28 and 30 (0.01 mmol) with
cysteamine (0.02 mmol), after 30 minutes, did not lead to the formation of the Michael adduct
(Figure 14). The same was observed after 24 h. Thus, we can expect that the double bond of 28 and
30 did not react covalently with sulphur nucleophiles, with indiscriminate reactivity. In addition,
based on the results of the jumpꢀdilution and the ATPꢀcompetition assays, we are confident that
they did not react covalently with GSKꢀ3β kinase.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 14. Thiol trapping assay. HꢀNMR spectrum of 28 (A) and 30 (B) in DMSOꢀd and in
6
6
DMSOꢀd with cysteamine (1:2). In the squares, the olefin signal is highlighted. No reaction occurs,
even after 30 minutes.
General procedure for the synthesis of compounds 4ꢀ38.
The corresponding aldehydes (1 mmol) were reacted with 2,4ꢀthiazolidinedione (1 mmol), using
EDDA (0.5 mmol) under microwave irradiation at 80°C for 30 minutes. The reaction mixture was
diluted with water and the solid was collected by filtration. After washing it with water all the final
compounds were purified through crystallization from ethanol/water.
(
Z)ꢀ5ꢀbenzylidenethiazolidineꢀ2,4ꢀdione (4). The title compound 4 was obtained as a light yellow
1
solid, according to general procedure using benzaldehyde (4a). Yield 87%. HꢀNMR (401 MHz, d6
13
ꢀ
DMSO): δ 12.61 (br s, 1H); δ 7.79 (s, 1H); δ 7.61ꢀ7.46 (m, 5H). CꢀNMR (101 MHz, d
6
ꢀDMSO):
ꢀ
δ 168.37; δ 167.89; δ 133.49; δ 132.10; δ 130.81; δ 130.41; δ 129.74; δ 124.11. MS (ESI ) m/z: 204
ꢀ
[MꢀH] .
2
6
ACS Paragon Plus Environment

Page 27 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
Z)ꢀ5ꢀ(2ꢀhydroxybenzylidene)thiazolidineꢀ2,4ꢀdione (5). The title compound 5 was obtained as a
1
yellow solid, according to general procedure using 2ꢀhydroxybenzaldehyde (5a). Yield 56%. Hꢀ
NMR (401 MHz, d
6
ꢀDMSO): δ 12.48 (br s, exch, 1H); δ 10.48 (s, exch, 1H); δ 8.02 (s, 1H); δ 7.33ꢀ
13
7
.28 (m, 2H); δ 6.96ꢀ6.92 (m, 2H). CꢀNMR (101 MHz, d
6
ꢀDMSO): δ 168.60; δ 167.97; δ 157.69;
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ꢀ
ꢀ
δ 132.66; δ 128.73; δ 127.41; δ 122.36; δ 120.37; δ 120.11; δ 116.55. MS (ESI ) m/z: 220 [MꢀH] .
(Z)ꢀ5ꢀ(3ꢀhydroxybenzylidene)thiazolidineꢀ2,4ꢀdione (6). The title compound 6 was obtained as a
1
white solid, according to general procedure using 3ꢀhydroxybenzaldehyde (6a). Yield 43%. Hꢀ
NMR (401 MHz, d
t, J = 7.9, 1H); δ 7.03 (d, J = 7.6, 1H); δ 6.97(s, 1H); δ 6.87 (d, J = 8, 1H). CꢀNMR (101 MHz, d
ꢀDMSO): δ 168.38; δ 167.81; δ 158.29; δ 134.63; δ 132.37; δ 130.78; δ 123.78; δ 121.73; δ 118.14;
6
ꢀDMSO): δ 12.58 (br s, exch, 1H); δ 9.82 (s, exch, 1H); δ 7.68 (s, 1H); δ 7.32
1
3
(
6
ꢀ
ꢀ
δ 116.32. MS (ESI ) m/z: 220 [MꢀH] .
(Z)ꢀ5ꢀ(4ꢀhydroxybenzylidene)thiazolidineꢀ2,4ꢀdione (7). The title compound 7 was obtained as a
1
yellow solid, according to general procedure using 4ꢀhydroxybenzaldehyde (7a). Yield 69%. Hꢀ
NMR (401 MHz, d
6
ꢀDMSO): δ 12.42 (br s, exch, 1H); δ 10.28 (s, exch, 1H); δ 7.69 (s, 1H); δ 7.45
13
(s, 1H); δ 6.91 (s, 1H). CꢀNMR (101 MHz, d
6
ꢀDMSO): δ 168.48; δ 167.94; δ 160.30; δ 132.79; δ
ꢀ
ꢀ
1
32.70; δ 124.36; δ 119.43; δ 116.73. MS (ESI ) m/z: 220 [MꢀH] .
(Z)ꢀ5ꢀ(2ꢀhydroxyꢀ4ꢀmethoxybenzylidene)thiazolidineꢀ2,4ꢀdione (8). The title compound 8 was
obtained as
a
yellow solid, according to general procedure using 2ꢀhydroxyꢀ4ꢀ
1
methoxybenzaldehyde (8a). Yield 75%. HꢀNMR (401 MHz, d
6
ꢀDMSO): δ 12.39 (br s, exch, 1H);
δ 10.64 (s, exch, 1H); δ 7.98 (s, 1H); δ 7.27 (d, J = 8.8, 1H); δ 6.59 (dd, J = 8.8, 2.5, 1H); δ 6.51 (d,
13
J = 2.5, 1H); δ 3.77 (s, 3H). CꢀNMR (101 MHz, d ꢀDMSO): δ 168.24; δ 167.63; δ 162.66; δ
6
+
159.13; δ 129.72; δ 126.96; δ 118.38; δ 113.07; δ 106.56; δ 101.13; δ 55.32. MS (ESI ) m/z: 274
+
[
M+Na] .
(Z)ꢀ5ꢀ(4ꢀhydroxyꢀ3ꢀmethoxybenzylidene)thiazolidineꢀ2,4ꢀdione (9). The title compound 9 was
obtained as a light yellow solid, according to general procedure using 4ꢀhydroxyꢀ3ꢀ
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
methoxybenzaldehyde (9a). Yield 62%. HꢀNMR (401 MHz, d ꢀDMSO): δ 12.43 (br s, exch, 1H);
6
δ 9.90 (s, exch, 1H); δ 7.71 (s, 1H); δ 7.17 (s, 1H); δ 7.08ꢀ7.05 (m, 1H); δ 6.94ꢀ6.90 (m, 1H); δ 3.83
1
3
(
s, 3H). CꢀNMR (101 MHz, d
6
ꢀDMSO): δ 168.51; δ 167.99; δ 149.84; δ 148.40; δ 132.96; δ
ꢀ
ꢀ
1
24.83; δ 124.54; δ 119.73; δ 116.60; δ 114.58; δ 56.07. MS (ESI ) m/z: 250 [MꢀH] .
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(Z)ꢀ5ꢀ(4ꢀhydroxyꢀ3ꢀnitrobenzylidene)thiazolidineꢀ2,4ꢀdione (10). The title compound 10 was
obtained as a light yellow solid, according to general procedure using 4ꢀhydroxyꢀ3ꢀ
1
nitrobenzaldehyde (10a). Yield 79%. HꢀNMR (401 MHz, d ꢀDMSO): δ 8.14 (d, J = 2.3 Hz, 1H); δ
6
13
7
.76 (s, 1H); δ 7.72 (dd, J = 8.8, 2.3 Hz, 1H); δ 7.22 (d, J = 8.8 Hz, 1H). CꢀNMR (101 MHz, d
6
ꢀ
DMSO): δ 168.88; δ 168.61; δ 155.86; δ 137.81; δ 135.78; δ 130.09; δ 128.19; δ 123.16; δ 122.55;
ꢀ
ꢀ
δ 121.57. MS (ESI ) m/z: 265 [MꢀH] .
(
Z)ꢀ5ꢀ(2ꢀfluorobenzylidene)thiazolidineꢀ2,4ꢀdione (11). The title compound 11 was obtained as a
pale yellow solid, according to general procedure using 2ꢀfluorobenzaldehyde (11a). Yield 49%.
1
HꢀNMR (401 MHz, d
6
ꢀDMSO): δ 12.73 (br s, 1H); δ 7.79 (s, 1H); δ 7.56 (t, J = 6.9, 2H); δ 7.39 (t,
ꢀDMSO): δ 168.05; δ 167.51; δ 160.93; δ 129.24; δ 126.81; δ
13
J = 8.3, 2H). CꢀNMR (101 MHz, d
25.85; δ 123.10; δ 121.36; δ 116.66. MS (ESI ) m/z: 246 [M+Na] .
6
+
+
1
(
Z)ꢀ5ꢀ(3ꢀfluorobenzylidene)thiazolidineꢀ2,4ꢀdione (12). The title compound 12 was obtained as a
1
yellow solid, according to general procedure using 3ꢀfluorobenzaldehyde (12a). Yield 60%. Hꢀ
6
NMR (401 MHz, d ꢀDMSO): δ 12.73 (br s, 1H); δ 7.83 (s, 1H); δ 7.62 (dd, J = 14.5, 7.5, 1H); δ
13
7
1
.48 (t, J = 9.9, 2H); δ 7.37 (t, J = 8.5, 1H). CꢀNMR (101 MHz, d
6
ꢀDMSO): δ 167.67; δ 167.27; δ
ꢀ
62.28; δ 135.46; δ 131.41; δ 130.32; δ 125.56; δ 125.37; δ 117.19; δ 116.74. MS (ESI ) m/z: 222
ꢀ
[MꢀH] .
(
Z)ꢀ5ꢀ(4ꢀfluorobenzylidene)thiazolidineꢀ2,4ꢀdione (13). The title compound 13 was obtained as a
1
yellow solid, according to general procedure using 4ꢀfluorobenzaldehyde (13a). Yield 76%. Hꢀ
6
NMR (401 MHz, d ꢀDMSO): δ 12.63 (br s, exch, 1H); δ 7.81 (s, 1H); δ 7.67 (m, 2H); δ 7.39 (t, J =
28
ACS Paragon Plus Environment

Page 29 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
3
8
.7, 2H). CꢀNMR (101 MHz, d ꢀDMSO): δ 167.79; δ 167.32; δ 162.86; δ 132.48; δ 130.70; δ
6
ꢀ
ꢀ
129.74; δ 123.31; δ 116.51. MS (ESI ) m/z: 222 [MꢀH] .
(
Z)ꢀ5ꢀ(2ꢀ(dimethylamino)benzylidene)thiazolidineꢀ2,4ꢀdione (14). The title compound 14 was
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
obtained as
a
yellow solid, according to general procedure using 2ꢀ(N,Nꢀ
1
dimethylamino)benzaldehyde (14a). Yield 70%. HꢀNMR (401 MHz, d
6
ꢀDMSO): δ 12.45 (br s,
exch, 1H); δ 7.86 (s, 1H); δ 7.43 (dd, J = 12.9, 7.3, 2H); δ 7.18 (d, J =8.1, 1H); δ 7.10 (t, J = 7.5,
13
1
1
H); δ 2.68 (s, 6H). CꢀNMR (101 MHz, d ꢀDMSO): δ 168.65; δ 167.88; δ 153.28; δ 131.38; δ
6
ꢀ
ꢀ
30.53; δ 129.83; δ 125.65; δ 122.82; δ 122.08; δ 118.81; δ 44.15. MS (ESI ) m/z: 247 [MꢀH] .
(
Z)ꢀ5ꢀ(3ꢀ(dimethylamino)benzylidene)thiazolidineꢀ2,4ꢀdione (15). The title compound 15 was
obtained as
a
yellow solid, according to general procedure using 3ꢀ(N,Nꢀ
1
dimethylamino)benzaldehyde (15a). Yield 78%. HꢀNMR (401 MHz, d
6
ꢀDMSO): δ 12.55 (br s,
13
exch, 1H); δ 7.73 (s, 1H); δ 7.31 (t, J = 7.9, 1H); δ 6.87ꢀ6.81 (m, 3H); δ 2.93 (s, 6H). CꢀNMR
(101 MHz, d ꢀDMSO): δ 168.45; δ 167.74; δ 151.06; δ 134.00; δ 133.51; δ 130.16; δ 123.16; δ
17.73; δ 114.79; δ 113.99; δ 40.33. MS (ESI ) m/z: 247 [MꢀH] .
6
ꢀ
ꢀ
1
(Z)ꢀ5ꢀ(4ꢀ(dimethylamino)benzylidene)thiazolidineꢀ2,4ꢀdione (16). The title compound 16 was
obtained as an orange solid, according to general procedure using 4ꢀ(N,Nꢀ
1
dimethylamino)benzaldehyde (16a). Yield 88%. HꢀNMR (401 MHz, d ꢀDMSO): δ 12.29 (br s,
6
13
exch, 1H); δ 7.66 (s, 1H); δ 7.42 (d, J = 8.9, 2H); δ 6.81 (d, J = 9, 2H); δ 3.01 (s, 6H). CꢀNMR
(
101 MHz, d
6
ꢀDMSO): δ 168.10; δ 167.52; δ 151.42; δ 132.83; δ 132.06; δ 119.80; δ 115.71; δ
ꢀ
ꢀ
1
12.01; δ 30.63. MS (ESI ) m/z: 247 [MꢀH] .
(
Z)ꢀ5ꢀ(pyridinꢀ2ꢀylmethylene)thiazolidineꢀ2,4ꢀdione (17). The title compound 17 was obtained as
1
a yellow solid, according to general procedure using 2ꢀpyridinecarbaldehyde (17a). Yield 55%. Hꢀ
6
NMR (401 MHz, d ꢀDMSO): δ 12.43 (br s, 1H); δ 8.75 (d, J = 4.7, 1H); δ 7.93 (t, J = 7.6, 1H); δ
1
3
6
7.86ꢀ7.82 (m, 2H); δ 7.42 (dd, J = 6.7, 5.6, 1H). CꢀNMR (101 MHz, d ꢀDMSO): δ 172.13; δ
29
ACS Paragon Plus Environment