Synthesis, biological evaluation and quantitative structure-active relationships of 1,3-thiazolidin-4-one derivatives. A promising chemical scaffold endowed with high antifungal potency and low cytotoxicity

Article abstract of DOI:10.1016/j.ejmech.2017.09.026

With reference to recent studies reporting on the various biological properties of the thiazolidinone scaffold, we synthesized more than a hundred compounds characterized by a 1,3-thiazolidin-4-one nucleus derivatised at the C2 with a hydrazine bridge linked to (cyclo)aliphatic or hetero(aryl) moieties, and their N-benzylated derivatives. These molecules were assayed as potential anti-Candida agents and they were shown to possess comparable, and in some cases higher biological activity than well-established topical and systemic antimycotic drugs (i.e. clotrimazole, fluconazole, ketoconazole, miconazole, tioconazole, amphotericin B). Compounds endowed with the lowest MICs underwent further testing in order to assess their cytotoxic effect (CC₅₀) on Hep2 cells, which demonstrated their relative safety. Finally, QSAR and 3-D QSAR models were used to predict putative chemical modifications of the 1,3-thiazolidin-4-one scaffold in order to design new and potential more active compounds against Candida spp.

Full text of DOI:10.1016/j.ejmech.2017.09.026

Accepted Manuscript
Synthesis, biological evaluation and quantitative structure-active relationships of
1,3-thiazolidin-4-one derivatives. A promising chemical scaffold endowed with high
antifungal potency and low cytotoxicity
Simone Carradori, Bruna Bizzarri, Melissa D'Ascenzio, Celeste De Monte, Rossella
Grande, Daniela Rivanera, Alessanda Zicari, Emanuela Mari, Manuela Sabatino,
Alexandros Pasilinakos, Rino Ragno, Daniela Secci
PII:
S0223-5234(17)30740-7
10.1016/j.ejmech.2017.09.026
EJMECH 9741
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 18 July 2017
Revised Date: 15 September 2017
Accepted Date: 15 September 2017
Please cite this article as: S. Carradori, B. Bizzarri, M. D'Ascenzio, C. De Monte, R. Grande, D.
Rivanera, A. Zicari, E. Mari, M. Sabatino, A. Pasilinakos, R. Ragno, D. Secci, Synthesis, biological
evaluation and quantitative structure-active relationships of 1,3-thiazolidin-4-one derivatives. A promising
chemical scaffold endowed with high antifungal potency and low cytotoxicity, European Journal of
Medicinal Chemistry (2017), doi: 10.1016/j.ejmech.2017.09.026.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical Abstract

ACCEPTED MANUSCRIPT
Synthesis, Biological Evaluation and Quantitative Structure-Active
Relationships of 1,3-thiazolidin-4-one Derivatives. A Promising
Chemical Scaffold Endowed with High Antifungal Potency and Low
Cytotoxicity
Simone Carradoria*, Bruna Bizzarrib, Melissa D’Ascenziob,§, Celeste De Monteb, Rossella Grandea,c,
Daniela Rivanerad, Alessanda Zicarie, Emanuela Marie, Manuela Sabatinob,f, Alexandros
Pasilinakosb,f,g, Rino Ragnob,f,g, Daniela Seccib,*
aDepartment of Pharmacy, "G. d'Annunzio" University of Chieti-Pescara, Via dei Vestini 31, 66100
Chieti, Italy
bDipartimento di Chimica e Tecnologie del Farmaco, Sapienza University of Rome, P.le A. Moro 5,
00185 Rome, Italy
cCenter for Aging Science and Translational Medicine (CeSI-MeT), Via dei Vestini 31, 66100
Chieti, Italy
dDipartimento di Sanità Pubblica e Malattie Infettive, Sapienza University of Rome, P.le A. Moro 5,
00185 Rome, Italy
eDipartimento di Medicina Sperimentale, Sapienza University of Rome, P.le A. Moro 5, 00185
Rome, Italy
fRome Center for Molecular Design, Dipartimento di Chimica e Tecnologie del Farmaco, Sapienza
University of Rome, P.le A. Moro 5, 00185 Rome, Italy
gAlchemical Dynamics s.r.l., 00125 Rome, Italy
*Corresponding authors: Simone Carradori, tel/fax: +39 0871 3554583; e-mail:
simone.carradori@unich.it
Daniela Secci, tel/fax: +39 06 49913975; e-mail: daniela.secci@uniroma1.it
§current address: Life and Biomedical Sciences Education, School of Life Sciences, University of
Dundee, Dundee DD1 4HN, United Kingdom.
1

ACCEPTED MANUSCRIPT
Abstract
With reference to recent studies reporting on the various biological properties of the thiazolidinone
scaffold, we synthesized more than a hundred compounds characterised by a 1,3-thiazolidin-4-one
nucleus derivatised at the C2 with a hydrazine bridge linked to (cyclo)aliphatic or hetero(aryl)
moieties, and their N-benzylated derivatives. These molecules were assayed as potential anti-
Candida agents and they were shown to possess comparable, and in some cases higher biological
activity than well-established topical and systemic antimycotic drugs (i.e. clotrimazole, fluconazole,
ketoconazole, miconazole, tioconazole, amphotericin B). Compounds endowed with the lowest
MICs underwent further testing in order to assess their cytotoxic effect (CC50) on Hep2 cells, which
demonstrated their relative safety. Finally, QSAR and 3-D QSAR models were used to predict
putative chemical modifications of the 1,3-thiazolidin-4-one scaffold in order to design new and
potential more active compounds against Candida spp..
Keywords: Antifungal activity; Candida spp.; Cytotoxicity; Thiazolidinone; 3-D QSAR.
2

ACCEPTED MANUSCRIPT
1. Introduction
Fungi belonging to Candida species are usually commensal organisms of the human microbial
flora. However, under various circumstances they can become the cause of opportunistic infections,
whose severity depends on the state of health of the patient [1]. In particular, mucocutaneous
mycoses are usually due to a functional damage of the epithelium that becomes unable to protect the
underlying mucous tissue from the infection, while systemic mycoses develop when the immune
system of the patient is compromised. Therefore, factors like cancer therapy, broad spectrum
antibiotic therapy, transplant surgery, or AIDS, can lead to a higher susceptibility of the patient to
Candida infections [2].
Together with Candida albicans, which is often isolated from the blood of patients affected by
invasive candidiasis, recent studies have shown that infections associated to other Candida spp. like
C. glabrata, C. parapsilosis, and C. tropicalis, are on the rise [3]. Moreover, several epidemiologic
studies have lately demonstrated an increasing incidence of drug resistance, mostly due to the
extensive use of azole-based agents. For these reasons, the design and discovery of novel antifungal
agents that differ from the conventionally used therapeutics in both chemical structure and
mechanism of action is an open challenge in the management of Candida infections, especially with
regard to immunocompromised patients [4].
In previously reported studies within our research group, we introduced an isosteric modification
into the imidazole ring of common antifungal agents to generate a series of (thiazol-2-yl)hydrazines
differently substituted at C2 and C4 positions of the thiazole ring (Scheme 1). These newly
synthesized anti-Candida agents were characterized by low cell toxicity and a synergistic
mechanism of action when co-administered with clotrimazole [5-9]. Our studies were based at the
time on the increasing literature reporting on the biological activity of the selected scaffold [10,11].
As a consequence, the thiazole moiety was conserved in the current study, although modified in
3

ACCEPTED MANUSCRIPT
accordance to the emerging therapeutic importance of 4-thiazolidinone derivatives, especially with
regard to their antimycotic activity [12]. The newly proposed scaffold combined the 4-
thiazolidinone ring with a hydrazonic portion bearing a (cyclo)aliphatic or (hetero)aryl substituent
at N-hydrazine (compounds 1a-59a). The lactamic nitrogen of the thiazolidin-4-one ring was further
derivatised by the addition of a benzyl group (compounds 1b-59b) in order to evaluate the increased
steric hindrance and the enhanced lipophilicity of this pendant (Scheme 1) on the biological effect.
According to the established guidelines of Clinical and Laboratory Standards Institute (CLSI) and
the European Committee for Antimicrobial Susceptibility Testing (EUCAST), we assayed the
susceptibility of twenty-two clinical Candida spp. strains to our compounds by determining their
minimum inhibitory concentration (MIC) [13].
H, 4-CH3, 4-OCH3, 3-NO2, 4-NO2,
4-CN, 4-F, 4-Cl, 4-Br, 4-I, 4-C6H5,
2,4-Cl, 2,4-OCH3
R
Different carbonylic
compounds
N
HN
N
S
(Thiazol-2-yl)hydrazines
Aliphatic chains (C3-C8)
Unsubstituted or alkyl-cycloaliphatic rings
(C5-C8)
R
O
N
N
N
S
R = H (1a-59a),
1b 59b
)
(Hetero)aryl or cycloaliphatic/bicyclicrings
= cyclopropane, furan,
thiophene,cyclohexane, pyridine, (nitro-
substituted)benzene,indole,benzofuran,
benzodioxole,benzodioxane,chromene,
naphthalene,phenantrene, ferrocene
Bz (
-
1,3-Thiazolidin-4-ones
Scheme 1. Chemical modification of the (thiazol-2-yl)hydrazine scaffold for the design of
thiazolidin-4-one derivatives (1a-59a and 1b-59b).
Nowadays, no information is available in the literature with respect to the pharmacological target of
the thiazolidinone derivatives as antifungal agents, although it could be speculated that the
4

ACCEPTED MANUSCRIPT
thiazolidine-4-one ring is a potential surrogate of the diphosphate moiety, present in UDP-N-
acetylenol-pyruvylglucosamine [12]. To better clarify this aspect, a structure-activity relationship
(QSAR) and three-dimensional QSAR (3-D QSAR) were performed using C. albicans growth
inhibition data as dependent variable.
2. Chemistry
We designed and synthesized one hundred and eighteen derivatives structurally characterized by the
presence of a common 1,3-thiazolidin-4-one nucleus combined with a hydrazine linker bearing
different substituents. In compounds belonging to a series the lactamic nitrogen was not substituted,
whereas for derivatives of b series it was functionalized with a benzyl group to establish the effect
of steric hindrance and improved lipophilicity on the pharmacological activity of the corresponding
compounds. The substituents on the hydrazine consisted of aliphatic chains (C3-C8), which include
structural isomers and their incorporation into cyclic structures, aromatic, bicyclic and
(hetero)aromatic rings (furan, thiophene, benzene, substituted benzene, pyridine, indole,
benzodioxole, benzofuran, benzodioxane, naphthalene, chromene, phenantrene, ferrocene). In
particular, derivatives 59a and 59b were obtained as products of the molecular doubling of the
parent compounds 58a and 58b, respectively.
The general strategy for the synthesis of the target molecules is outlined in Scheme 2. Different
carbonyl compounds were dissolved or suspended in ethyl alcohol and directly reacted with
thiosemicarbazide in the presence of catalytic amounts of acetic acid. The resulting
thiosemicarbazone intermediates were subsequently reacted with ethyl-bromoacetate in methanol
and sodium acetate in order to obtain the N-unsubstituted 1,3-thiazolidin-4-one derivatives (1a-59a)
in high yields by a single cyclo-condensation process. Finally, the reaction between the obtained
products and benzyl bromide in anhydrous acetone and potassium carbonate gave their
corresponding N-benzyl derivatives (1b-59b). All the synthesized compounds were washed with n-
5

ACCEPTED MANUSCRIPT
hexane, petroleum ether or diethyl ether and purified by column chromatography before
characterization by spectroscopic methods and elemental analysis. In general, in the IR spectrum
(neat) derivatives 1a-59a showed a strong band at about 3144-3070 cm-1 due to the stretching of the
lactam NH bond, at about 1710-1697 cm-1 due to the stretching of C=O bond and at about 1639-
1600 cm-1 for the stretching of C=N, thus confirming the formation of the thiazolidin-4-one ring;
for derivatives 1b-59b there were strong bands at about 1693 cm-1 due to the stretching of C=O
bond, at about 1622 cm-1 for the stretching of C=N bond and at about 1582 and 1447 cm-1 for C=C
stretching. These observations were further substantiated by characteristic signal resonances in the
1H (~3.85 ppm for methylene protons of thiazolidin-4-one nucleus and D2O exchangeable NH
proton at around 11.60 ppm) and 13C (~33.0 ppm for C5 of thiazolidin-4-one nucleus) NMR
spectra.
S
ii
i
S
H2N
N
HN
O
N
NH2
H
O
NH2
Br
OEt
H
N
O
N
N
Br
iii
O
N
N
S
N
S
1b 59b
1a 59a
-
-
Scheme 2. Synthesis of derivatives 1a-59a and 1b-59b. Reagents and conditions: (i) EtOH, acetic
acid (cat.), rt; (ii) MeOH, CH3COONa, rt; (iii) anhydrous acetone, anhydrous K2CO3, rt.
3. Antibacterial and anti-Candida spp. activity
6

ACCEPTED MANUSCRIPT
Derivatives 1a-59a and 1b-59b were dissolved in dimethylsulfoxide (DMSO) and evaluated for
their antibacterial activity (MIC values ≥256 µg/mL for the new compounds, MIC = 8 µg/mL for
ceftazidime as a reference drug). Organisms from routine clinical Gram-positive (Staphylococcus
aureus, Staphylococcus warneri, Streptococcus faecalis, Streptococcus α-hemolyticus) and Gram-
negative isolates (Escherichia coli, Proteus mirabilis, Enterobacter spp., Klebsiella oxytoca) from
the respiratory tract were collected from specimens of patients at the Hospital ‘Azienda Policlinico
Umberto I°’ (Sapienza University of Rome). The isolates were subcultured on selected media to
ensure purity and identified by conventional methodologies (evaluation of colony morphology and
biochemical identification). All isolates were subcultured and incubated at 35 °C under aerobic or
anaerobic conditions depending on the species to ensure optimal growth. The in vitro antibacterial
activities of the compounds reported herein were determined by the broth microdilution method, as
recommended by the National Committee for Clinical Laboratory Standards (NCCLS) [14] using
Mueller-Hinton II broth (BBL Microbiology Systems, Cockeysville, MD).
Antifungal activity was evaluated against twenty-two clinical fungal isolates of Candida spp. (C.
albicans, C. glabrata, C. tropicalis, C. krusei, C. parapsilosis, and C. sakè) and compared with
topical and systemic reference drugs such as clotrimazole, fluconazole, ketoconazole, miconazole,
tioconazole and amphotericin B (see Tables 1-3 according to chemical classification into aliphatic,
cycloaliphatic and (hetero)aromatic derivatives). These Candida spp. clinical isolates were recently
collected from specimens of patients at the ‘Azienda Policlinico Umberto I°’ (Sapienza University
of Rome) and were obtained from haematology/oncology and surgery departments, which also
included an intensive care unit. In particular, the samples were isolated from the upper and lower
respiratory tract, blood, and indwelling venous catheters; the isolates were identified by
conventional methodologies. Prior to testing, each isolate was subcultured on a qualified medium to
7

ACCEPTED MANUSCRIPT
ensure purity and optimal growth. All derivatives were dissolved in DMSO. The in vitro antifungal
activities were determined by the broth microdilution method with Sabouraud dextrose broth (BBL
Microbiology Systems, Cockeysville, MD) as recommended by the NCCLS [15].
Table 1. Geometric mean of minimum inhibitory concentrations (MICs) of aliphatic derivatives 1a-
12a, 1b-12b and reference compounds against twenty-two clinical strains of Candida species.
R
O
N
N
N
S
Tested fungi (MIC values in µg/mL)
C.
albicans glabrata tropicalis krusei
(8 (4 (3 (3
strains) strains) strains) strains)
C.
C.
C.
C.
C. sakè
Compound
R
MW
parapsilopis
(2 strains) strains)
(2
H
H
H
171.22
185.25
199.27
2
8
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1a
2a
3a
H
H
H
199.27
211.28
211.28
2
2
2
1
2
2
4
8
2
2
2
2
2
2
2
2
4
4
4a
5a
6a
H
H
H
H
213.30
213.30
213.30
227.33
227.33
241.35
261.34
275.37
289.40
289.40
301.41
2
2
2
2
2
32
16
4
2
2
2
4
2
2
7a
8a
H
2
2
2
4
2
9a
H
2
2
8
8
8
10a
11a
12a
1b
H
4
1
1
2
2
4
H
2
2
2
16
256
16
16
8
16
256
8
16
256
32
32
8
Bz
Bz
Bz
Bz
Bz
256
8
256
16
2
256
8
2b
8
8
8
3b
32
128
2
32
128
8
4b
2
8
8
8
5b
H
Bz
301.41
8
2
8
8
8
8
6b
H
Bz
Bz
Bz
Bz
Bz
Bz
303.42
303.42
303.42
317.45
317.45
8
2
8
2
16
2
16
2
16
2
16
4
7b
8b
4
4
4
4
8
8
9b
8
8
16
16
16
4
16
8
16
4
10b
11b
12b
128
32
331.48
344.83
8
8
16
16
16
16
2
2
2
2
2
2
Fluconazole
8

ACCEPTED MANUSCRIPT
2
2
2
8
4
2
2
4
4
4
2
2
2
8
2
4
2
4
8
4
2
2
4
8
2
4
2
2
4
2
306.27
531.43
416.12
387.71
923.49
Ketoconazole
Clotrimazole
Miconazole
Tioconazole
Amphotericin B
Table 2. Geometric mean of minimum inhibitory concentrations (MICs) of cycloaliphatic
derivatives 13a-25a and 13b-25b against twenty-two clinical strains of Candida species.
R
O
N
N
N
S
Tested fungi (MIC values in µg/mL)
C.
albicans glabrata tropicalis krusei
(8 (4 (3 (3
strains) strains) strains) strains)
C.
C.
C.
C.
C. sakè
Compound
R
MW
parapsilopis
(2 strains) strains)
(2
H
H
H
197.26
197.26
211.28
2
2
2
2
2
1
32
2
2
2
2
4
2
2
2
2
2
13a
14a
15a
8
H
H
211.28
211.28
2
2
2
2
2
2
16a
17a
64
64
64
256
256
256
H
225.31
2
1
8
2
2
2
18a
H
H
H
225.31
225.31
239.34
8
16
2
8
16
2
8
2
16
2
2
2
16
2
19a
20a
21a
16
32
16
128
H
H
H
267.39
225.31
239.34
8
8
8
8
8
8
8
8
8
2
8
2
8
2
8
22a
23a
24a
32
32
32
H
239.34
8
8
16
16
16
16
25a
Bz
Bz
Bz
287.38
287.38
301.41
2
8
8
4
8
8
2
16
8
2
16
4
2
16
4
2
16
4
13b
14b
15b
Bz
Bz
301.41
301.41
256
2
256
2
256
2
256
2
256
2
256
2
16b
17b
Bz
Bz
315.43
315.43
16
8
2
8
32
8
16
4
4
4
16
4
18b
19b
9

ACCEPTED MANUSCRIPT
Bz
Bz
315.43
329.46
8
2
8
4
16
16
16
4
16
8
16
2
20b
21b
Bz
357.51
16
2
8
16
16
16
22b
Bz
Bz
315.43
329.46
256
8
256
8
256
8
256
8
256
8
256
8
23b
24b
Bz
329.46
8
8
8
8
8
8
25b
Table 3. Geometric mean of minimum inhibitory concentrations (MICs) of aromatic and
heteroaromatic derivatives 26a-59a and 26b-59b against twenty-two clinical strains of Candida
species.
R
O
N
N
N
S
Tested fungi (MIC values in µg/mL)
C.
C.
C.
C.
krusei
(3
C.
C. sakè
(2
strains)
albicans glabrata tropicalis
(8
strains)
Compound
R
MW
parapsilopis
(2 strains)
(4
strains)
(3
strains)
strains)
H
H
H
H
H
209.23
223.25
225.29
239.32
239.32
8
2
2
2
256
2
256
2
256
2
256
2
26a
27a
28a
29a
30a
256
2
256
1
256
2
256
8
256
16
8
256
16
8
8
8
16
16
H
273.76
8
8
8
16
8
16
31a
H
H
H
H
H
263.36
219.26
264.26
2
4
2
8
8
16
4
2
16
2
2
8
32a
33a
34a
16
16
128
128
16
4
10

ACCEPTED MANUSCRIPT
H
H
264.26
264.26
16
16
8
8
16
4
8
8
4
4
8
8
35a
36a
H
H
233.29
278.29
2
2
2
2
16
2
8
8
16
2
16
2
37a
38a
H
H
278.29
278.29
2
4
4
8
2
8
4
4
2
2
39a
40a
16
16
H
H
H
H
H
H
H
220.25
234.28
220.25
234.28
220.25
234.28
2
2
8
1
2
1
4
8
2
2
41a
42a
43a
44a
45a
46a
N
16
16
8
8
16
256
256
8
16
256
256
16
16
16
256
256
16
256
256
16
256
256
256
256
256
256
H
258.30
128
16
8
16
256
256
47a
H
H
H
H
272.33
273.31
263.27
269.32
4
8
8
2
4
16
2
48a
49a
50a
51a
2
8
2
2
2
256
256
256
256
256
256
256
256
256
256
256
256
11

ACCEPTED MANUSCRIPT
H
H
283.35
283.35
2
2
16
32
16
16
2
2
52a
53a
32
16
256
256
H
H
291.33
301.32
4
4
4
8
16
8
54a
55a
256
16
256
256
256
256
H
H
333.41
341.21
4
4
4
8
16
16
56a
57a
256
16
256
256
256
256
H
H
275.33
388.47
2
2
2
2
2
2
8
2
2
2
8
2
58a
59a
Bz
Bz
Bz
Bz
Bz
299.35
313.37
315.41
329.44
329.44
8
256
8
4
256
8
8
4
4
4
26b
27b
28b
29b
30b
256
8
256
32
256
32
8
256
32
2
4
16
128
4
8
256
4
128
128
128
Bz
363.88
256
8
256
128
128
128
31b
H
Bz
Bz
Bz
353.48
309.39
354.38
4
4
2
4
4
2
2
16
2
4
16
2
4
4
2
2
16
4
32b
33b
34b
H
12

ACCEPTED MANUSCRIPT
Bz
Bz
354.38
354.38
2
4
2
2
2
8
2
2
2
8
4
35b
36b
16
Bz
Bz
323.41
368.41
8
2
4
2
128
2
4
8
4
2
4
4
37b
38b
Bz
Bz
368.41
368.41
4
8
2
4
8
8
4
4
16
4
8
8
39b
40b
H
Bz
Bz
Bz
Bz
Bz
Bz
310.37
324.40
310.37
324.40
310.37
324.40
4
256
256
2
4
16
256
2
4
28
256
2
4
256
256
2
16
256
256
2
2
41b
42b
43b
44b
45b
46b
N
256
256
2
256
8
256
16
256
32
256
2
256
32
256
32
Bz
Bz
Bz
Bz
348.42
362.45
363.43
353.39
256
8
16
4
256
16
32
8
2
4
16
32
32
8
16
8
47b
48b
49b
50b
4
4
16
8
4
8
8
8
Bz
Bz
359.44
373.47
8
4
8
4
16
32
16
16
16
16
16
4
51b
52b
13

ACCEPTED MANUSCRIPT
Bz
Bz
Bz
373.47
381.45
391.44
8
2
16
8
256
2
16
4
16
2
16
8
53b
54b
55b
256
32
256
128
128
128
Bz
Bz
423.53
431.33
2
2
8
2
8
4
2
8
56b
57b
16
32
128
128
Bz
Bz
365.45
568.71
8
8
2
2
32
32
8
8
32
32
8
8
58b
59b
4. Cell toxicity
Compounds endowed with the strongest anti-Candida activity (1a, 3a, 4a, 10a, 11a, 14a, 15a, 16a,
18a, 29a, 42a, 17b and 44b) were also tested at five increasing concentrations ranging from 0.05 to
100 µg/mL (Table 4) to assess their cytotoxic activity against Hep2, a cultured cell line derived
from human laryngeal epidermoid carcinoma. Cell viability was analysed using Trypan Blue
exclusion test, as previously reported in the literature [16]. Clotrimazole was used as a reference
drug.
14

ACCEPTED MANUSCRIPT
Table 4. Cytotoxic effect (CC50) and selectivity index (SI) of the most active compounds and
clotrimazole tested on Hep2 cells after 24 h of incubation at 37 °C using Trypan Blue exclusion test
expressed as cell survival fraction (%).
Trypan Blue exclusion testa
SIc
Compound
CC50 (µg/mL)
68.4 ± 5.8
93.4 ± 4.8
49.2 ± 4.2
71.8 ± 6.8
46.4 ± 5.2
22.2 ± 3.6
52.8 ± 3.6
50.8 ± 6.4
52.6 ± 6.8
>100.0
34
46.5
24.5
35.5
11.5
11.1
26
25
26
>50
28
11
1a
3a
4a
10a
11a
14a
15a
16a
18a
29a
42a
17b
44b
56.8 ± 4.8
22.8 ± 4.0
37.4 ± 4.0
38.8 ± 4.4
18.5
19
Clotrimazole
aCells incubated with culture medium alone represented the controls (negative control, CC50= 98 ±
1.0 µg/mL). Viability with 2 µL of DMSO = 94.2%, viability with 20 µL of DMSO = 87.6%.
Positive control: cells incubated with 1 mM sodium nitroprusside (CC50= 10 ± 0.6 µg/mL)
bData represent arithmetic means of at least three independent experiments ± SD.
cSI: selectivity index defined as the ratio between CC50 and MIC value for C. albicans.
5. Results and discussion
A large series of 1,3-thiazolidin-4-one derivatives were synthesized in high yield due to this
optimized procedure allowing the introduction of several structural modifications, e.g. by a
variation of the starting carbonyl compound and the addition of a benzyl group at the lactamic
nitrogen. A robust evaluation of the structure-activity and structure-selectivity relationships (SARs
and SSRs) could be extrapolated from the data in Tables 1-3 after the evaluation of the in vitro
growth inhibition of several bacterial and fungal clinical isolates. None of the tested compounds
showed appreciable antibacterial activity, thus suggesting the products to be active selectively
towards fungal pathogens.
15

ACCEPTED MANUSCRIPT
According to the in vitro antifungal activity assay, most of a series compounds (non-benzylated
thiazolidinones) characterized by aliphatic and cycloaliphatic chains, aromatic, heteroaromatic and
polycyclic rings, exhibited good antifungal activities, which were comparable or better than
reference drugs, against C. albicans and non-albicans spp.. In particular, the aliphatic compounds
1a, 3a and 7a, the cycloaliphatic 14a and 16a, the furan derivative 27a and the product of molecular
doubling 59a, displayed MIC values of 2 µg/mL against both C. albicans and non-albicans spp.
strains similar to fluconazole and clotrimazole and could be identified as best-in-class antifungal
agents. Moreover, the aliphatic 4a, the cyclic derivatives 15a and 18a, and the thiophene derivative
29a were selective toward C. glabrata possessing MIC values of 1 µg/mL, which is lower than the
MIC reported for reference drugs. In the case of compounds 11a and 42a, a promising MIC value of
1 µg/mL was also obtained against both C. glabrata and C. tropicalis. With respect to b series (N-
benzylated thiazolidinones), which is characterized by the functionalization of the lactamic NH with
a benzyl group, a loss in antifungal activity was observed for some aliphatic and cycloaliphatic
derivatives (MIC values of 256 µg/mL), while the MICs of the heterocyclic and aromatic products
were the same of those belonging to a series. In particular, the cyclohexane derivative 17b and the
pyridine derivative 44b showed a potent inhibitory activity (2 µg/mL) against all the Candida spp.
when compared to their corresponding non-benzylated compounds.
In consideration of these results, the most active derivatives against all Candida spp. (1a, 3a, 4a,
10a, 11a, 14a, 15a, 16a, 18a, 29a, 42a, 17b, 44b) were investigated for their cytotoxic effects
(EC50) by Trypan Blue exclusion test at five different concentrations (Table 4). Except for 11a, 14a
and 17b, which were not selective between fungal and Hep2 cells in terms of cytotoxicity, and for
44b that possessed a CC50 (37.4 ± 4.0) similar to that of the reference drug clotrimazole (38.8 ±
4.4), most of the selected compounds proved to be less cytotoxic than clotrimazole and
16

ACCEPTED MANUSCRIPT
characterized by higher values of in vitro Selectivity Index (SI), defined as the ratio between CC50
and MIC value for C. albicans (≥24.5 vs 19 of clotrimazole).
Finally, in order to exclude nonspecific and promiscuous compounds (PAINS), an accurate analysis
was performed by means of computational filters. The 118 molecules were screened with a PAINS
filter implemented using the RDKit library and 480 PAINS SMARTS patterns. None of the
reported molecules were potentially showing artificial activities.
6. Computational Studies
6.1. Overall procedure
To gather three-dimensional structure-activity relationships for the titled derivatives, a Python
implementation of 3-D QSAutogrid/R procedure (Py-3-D_QSAR) [17] was utilized to develop
partial least square (PLS) field-based 3-D QSAR models. At first trial, when all the assayed
compounds were submitted to the previously described algorithm, no 3-D QSAR models could be
obtained. Therefore, the procedure depicted in Figure 1 was followed: firstly, the anti-Candida
activity was chosen as dependent variable (biological activity) among the six different biological
evaluations listed in Tables 1-3. Secondly, the list of 118 assayed molecules was pruned of the two
ferrocene-containing compounds. Then, QSAR models were built to select as much as possible
structural information for the construction of 3-D QSAR models. Finally, 3-D QSAR models were
built, internally validated, and graphically described (Figure 1).
17

ACCEPTED MANUSCRIPT
Figure 1. QSAR and 3-D QSAR models’ building flowchart.
6.2. Initial Training Set Composition and Preparation
As reported in the experimental section, the training set compounds activities against the six
different Candida species showed high level of intercorrelation (not shown). Therefore, all the
models herein reported were developed on the basis of MIC values obtained against C. albicans
only, since this was the strain against whom the compounds showed a wider range of activities. For
the 3-D QSAR studies, the activities reported in Tables 1-3 in µg/mL were converted in pMIC
values.
6.3. QSAR studies
Similarly, as for 3-D QSAR, initial trials to get a QSAR model were not successful when using all
the investigated compounds (TrSet116, 116 C. albicans activities and 196 molecular descriptors).
Assuming the correctness of either the biological assays or the statistical application, it could be
postulated that likely not all the compounds were acting with similar mechanism of action or
18

ACCEPTED MANUSCRIPT
binding. Hence, in order to seek for an optimal subset of compounds, Backward Datapoint
Elimination (BDE) algorithm [18] was adopted in conjunction with Partial Least Square (PLS) and
cross-validation using either Leave-One-Out (LOO) and Leave-Some-Out (LSO) with 5 groups.
During BDE process, compounds were deleted from the training set and new models were
developed with the new training set. The best BDE sub-model was obtained with 62 derivatives
displaying r2, q2
and q2
values of 0.95, 0.91 and 0.77, respectively (Figure 2). To retain as
LSO
LOO
much as possible structural information, selection of sub-models was driven by r2 and q2LOO values.
It was arbitrarily chosen to build QSAR models at r2/q2LOO values levels of 0.70/0.58 and 0.80/0.69
characterized by 92 (TrSet92) and 88 (TrSet88) compounds, respectively.
Number of compounds
Figure 2. QSAR q2 values as function of the number of compounds retained
19

ACCEPTED MANUSCRIPT
in the training set. The blue line indicates r2, the green line indicates q2
,
LOO
and the red line indicates q2
.
LSO
6.4 3-D QSAR steps
Without any information on the biological target, the procedure to build reliable 3-D QSAR models
was based on an extensive study that involved trying different combinations of several
conformational analyses, superimposition techniques, and alignment rules based on the use of
different molecules as templates for the molecular alignment (Tables 5 and 6).
6.4.1 Training Set Conformational Analysis
Conformational analysis was performed on the training set molecules in order to prepare
conformational ensemble from which conformation could be selected as template for the molecular
superimposition. To explore the conformational space of the molecules under study as completely
as possible, different methods were used. In particular, six different conformational analyses were
performed for each compound: one using the Balloon software [19] and the remaining five by
means of in-house Python programs using the Open Babel (OB) library [20] and 4 different force
fields.
Table 5. List of conformational searches and molecular superimposition used to prepare aligned
datasets for the subsequent 3-D QSAR model building.
Conformational Search Engine
Force Field
MMFF94
GAFF
Balloon
OB
OB
OB
OB
Ghemical
MMFF94s
UFF
It could be argued that similar conformational ensembles should be obtained from the five
conformational searches. Analysis of the different ensemble revealed that, due to different
20

ACCEPTED MANUSCRIPT
conformational pruning and force fields, each method allowed the inspection of different
conformational spaces (Figure 3).
Balloon
OB/GAFF
OB/Ghemical
OB/MMFF94s
OB/UFF
Figure 3. Example of the five conformational ensembles for the compound 54b. Conformations
are disposed in similar orientation to highlight different inspected conformational space.
6.4.2 Template selection
In 3-D QSAR studies, selection of molecules’ conformation and their alignment (alignment rules)
are the critical steps to obtain reproducible models [21]. To analyze the different alignment rules, it
was necessary to take into account a series of features, which then allowed selecting the molecular
conformations to be used as a template for the molecular alignments (Supplementary material
Tables S1-S3). In particular, nine different calculated features were used: biological activity (Act),
conformation length (L), molecular weight (MW), total polar surface area (TPSA), number of
rotatable bonds (Rot), number of hydrogen bond acceptors (HA), number of hydrogen bond
donators (HD), logarithm of n-octanol-water partition coefficient (LogP), and molecular refractivity
(MR). The highest and lowest values for Act, TPSA, Rot, HA, HD, LogP and MR for a total of
virtually 16 templates were identified. For all training sets described above (TrSet116, TrSet92 and
TrSet88), except for the length feature, the global minima conformation was used for each template.
With regards to the templates related to conformation length feature, the longest conformation was
selected (Tables S1-S3). In case of compounds displaying the same feature’s values (i.e. same
values of min_MR for two or more molecules), the heaviest one was selected.
21

ACCEPTED MANUSCRIPT
Considering that the same conformation template could occur for a given feature, a different
number of templates were selected depending on either each training set or conformational analysis
method. In particular, 8, 14 and 14 different templates were selected for TrSet116, TrSet92 and
TrSet88, respectively.
6.4.3 Molecular superimposition
Once the template conformation was selected, the five conformational ensembles were
superimposed in turn by means of three different softwares: fkcombu [22], ShaEp [23] and Surflex-
Sim (herein just Surflex) [24]. In the case of ShaEp, the program was run either considering the
field-graph matching or based on shape-density only, leading to 4 different alignment trials and
hence to 160, 280 and 280 aligned data sets for TrSet116, TrSet92 and TrSet88, respectively.
6.4.4 3-D QSAR models’ definition
For each of the 720 aligned data sets, standard 3-D QSAR models (see Experimental section) were
derived for A, OA and HD probes leading to a total of 2160 initial models. Models showing r2
values greater than 0.6 and q2
coefficients higher than 0.1 were subjected to an optimization
LOO
procedure (Supplementary material Table S4). Firstly, for the selected 21 initial models that are
listed in Supplementary material Table S4, grid spacing was varied to seek for values giving higher
q2
coefficients; secondly, pretreatment parameters were selected through VPO procedure (see
LOO
Experimental section); lastly, each GRID-VPO model was subjected to variable selection procedure
by a fast Simulate Annealing (SA) algorithm. Models showing q2
values greater than 0.4 were
LOO
finally subjected to slow SA variable selection using LSO cross-validation (Table 6).
Table 6. Final 3-D QSAR models obtained by SA variable selection.
Fast SA – LOO
r2e q2f
0.755 0.495
Slow SA LSO
PCg r2e q2f
0.769 0.526
Probea Methodb
OB/UFF
Alignerc
Templated
#
N
PCg
5
92
A
fkcombu
rigid
5
1SA
22

ACCEPTED MANUSCRIPT
88
88
92
92 OA
92 OA
92 OA
88 OA
88 OA
92 HD
A
A
A
OB/Ghemical
OB/MMFF94s
OB/GAFF
OB/Ghemical
OB/UFF
fkcombu
fkcombu
surflex
fkcombu
fkcombu
surflex
min_HA
max_HD
rigid
min_HA
rigid
rigid
max_HD
min_HA
0.728 0.539
0.749 0.551
0.802 0.619
0.720 0.505
0.761 0.516
0.744 0.533
0.739 0.552
0.718 0.559
0.753 0.542
5
4
5
5
5
3
4
3
4
0.724 0.486
0.740 0.554
0.783 0.617
0.710 0.472
0.729 0.506
0.820 0.553
0.743 0.563
0.737 0.519
5
4
5
5
4
5
4
5
5
2SA
3SA
4SA
5SA
6SA
7SA
8SA
9SA
10SA
OB/GAFF
OB/MMFF94s
OB/Ghemical
OB/UFF
fkcombu
fkcombu
fkcombu
rigid
0.765 0.527
a
b
c
#: model id; N: number of compounds; Probe used; Conformational analysis method; Alignment
program; dTemplate feature; eSquared correlation coefficient; Cross-validation coefficient;
f
gNumber of optimal PC.
6.4.4 3-D QSAR models’ interpretation
As any 3-D QSAR methodology, the herein Python scripts were also implemented to save the
contour maps in a ready format, so that they could be read by the common molecular graphical
interfaces (i.e. UCSF Chimera). To avoid redundancy, only models obtained on TrSet92 with
alignment rules derived by conformational analyses performed with OB/GAFF combination and
aligned with Surflex on the most rigid template (4SA, and 7SA) were described herein, as they were
the ones showing the highest r2 and q2LSO values, Figure 4).
A
B
Figure 4. Recalculated and predicted versus experimental pMICs as obtained by A (A) and OA
(B) probes based 3-D QSAR models.
23

ACCEPTED MANUSCRIPT
The 3-D QSAR Average Activity Contribution (AAC) contour plots were calculated by the scalar
product among each grid node PLS coefficients and the average molecular interaction field values.
The AAC plots (Figure 5) usually give an important indication on the space around molecules, in
particular polyhedron from A probe (positive values in green and negative values in yellow), which
are almost exclusively related to sterical information. Positive values describe the portion of
molecule correlating with positive contribution to the biological activity and, for a given non-
overlapping molecule, an increment in bulkiness in those areas would lead to an increase in
potency. On the contrary, decreasing contribution is indicated by negative AAC values. Therefore,
the positive green polyhedrons in the low part of Figure 5 (circled in dark green) propose that, for
molecules of b series, an increment of activity could be likely obtained by inserting substituents in
the ortho position of the phenylmethyl moiety linked to the hydrazine bridge. Positive increments
are also suggested when a larger group is placed to overlap the green polyhedron situated between
the ortho position of 54b benzyl group and the methyl group on the hydrazine sp2 carbon atom
(blue circle in Figure 5A). On the other hand, a yellow polyhedron indicates that reducing the size
of the substituent on the para position of the phenyl(methyl) group would benefit to the potency of
the resulting compound (orange circle in Figure 5A). A slightly different scenario is instead
observable for molecules of a series. In particular, for comparison purposes, in Figure 5 is also
depicted derivative 3a as the one endowed with the best activity profile against all fungal strains
(Tables 1-3). Derivative 3a is deprived of the benzyl substituent on the thiazolidinone nitrogen
overlapping a negative yellow polyhedron (red circle in Figure 5A) which could explain the little
potency improvement of a series over b series. Furthermore, the smaller and more flexible sec-
pentyl substituent is not overlapping the negative yellow polyhedron circled in orange (Figure 5A).
24

ACCEPTED MANUSCRIPT
Regarding OA derived 3-D QSAR model, contour plots indicate that most of the information are
overlapped to those of A probe derived model. Due to the dual nature of the OA probe (sterical and
hydrogen bonding acceptor) only one big red polyhedron (dark red circle in Figure 5B) clearly
suggests that negatively charged groups in that area are important for the potency (Figure 6). In
fact, all compounds showing strong hydrogen acceptor atoms tend to display them within this area,
thus indicating that highly electronegative atoms in this position are correlated with biological
activity.
A
B
Figure 5. AAC contour plots for A) probe A (positive in green; negative in yellow) and B) probe
OA (positive in blue; negative in red). The most potent derivative of a series (3a, cyan colored
carbon atoms) and one among the most potent molecules of b series (54a, yellow colored carbon
atoms) are displayed.
25

ACCEPTED MANUSCRIPT
Figure 6. Graphical depiction of OA derived 3-D QSAR model (7SA). Dark red circle indicates
hydrogen acceptor atoms overlapping the red polyhedron.
7. Conclusions
A large number of 1,3-thiazolidin-4-ones were designed, synthesized, and assayed as anti-Candida
spp. agents. The obtained biological results showed the intrinsic potential of the newly proposed
compounds whether they were benzyl substituted or not, especially when compared to a set of
reference drugs. The most active compounds were also shown to display lower cytotoxicity than
clotrimazole against Hep2, a cell line derived from human epidermoid carcinoma and selected as
representative of the main cellular components of the skin, as topic administration of these
26

ACCEPTED MANUSCRIPT
pharmacological agents could be envisaged. Finally, despite the high structural redundancy of the
training set molecules, 3-D QSAR model driven by QSAR compound selection provided some
insights into potential further modifications of the 1,3-thiazolidin-4-one core that would lead to the
design of new and more potent antifungal derivatives.
8. Experimental section
Chemicals, solvents for the synthesis, and spectral grade solvents were purchased from Sigma-
Aldrich® (Italy) and used without further purification. Melting points (uncorrected) were
1
13
determined automatically on an FP62 apparatus (Mettler-Toledo). H and C NMR spectra were
recorded at 400 (and 101) MHz on a Bruker spectrometer using CDCl3 or DMSO-d6 as solvent.
Chemical shifts are expressed in ppm (parts per millions) relative to the solvent peak. Splitting
patterns are described as s, singlet; d, doublet; t, triplet; m, multiplet; bs, broad singlet. Coupling
constants J are valued in Hertz (Hz). The assignment of exchangeable protons (NH) was confirmed
by the addition of D2O. Infra-red spectra (neat) were recorded on a Bruker Tensor 27 FTIR
spectrometer equipped with an attenuated total reflectance attachment with internal calibration.
Absorption frequency maxima (νmax) are reported in wavenumbers (cm-1). Elemental analyses for C,
H, and N were recorded on a Perkin-Elmer 240 B microanalyzer and the analytical results were
within ± 0.4% of the theoretical values for all compounds. All reactions were monitored by
analytical TLC performed on precoated 0.2 mm thick silica gel plates (60 F254 Merck) and the spots
were detected under UV lamp (254 nm). Yields shown are not optimized. All reactions involving
air- or moisture-sensitive compounds were performed under N2 atmosphere using dried glassware
and syringes techniques to transfer solutions.
8.1 General procedure for the synthesis of compounds 1a-59a and 1a-59a.
The initial carbonyl compound (50 mmol) was dissolved/suspended in ethanol (50 mL) and
magnetically stirred with thiosemicarbazide (50 mmol) and catalytic amounts of acetic acid for 8-24
27

ACCEPTED MANUSCRIPT
h at room temperature. The obtained thiosemicarbazone was filtered, washed with the appropriate
solvent (n-hexane, petroleum ether or diethyl ether) and dried under vacuum. Ethyl-bromoacetate
(50 mmol) was added to a suspension of the intermediate thiosemicarbazone (50 mmol) and sodium
acetate (50 mmol) in methanol (50 mL) and the resulting mixture stirred at room temperature for
24-48 h. The obtained crude 4-thiazolidinone was poured on ice, filtered or extracted with
chloroform (3 x 30 mL) and purified by chromatography (SiO2, ethyl acetate/n-hexane 1/2) to give
compounds 1a-58a in high yields. The product (50 mmol) was dissolved/suspended in 50 mL of
anhydrous acetone in the presence of anhydrous potassium carbonate (50 mmol), and reacted with
equimolar amounts of benzyl bromide for 24-48 h. The mixture was poured on ice, filtered or
extracted with chloroform (3 x 50 mL) and purified by column chromatography (SiO2, ethyl
acetate/n-hexane 1/3) in order to obtain compounds 1b-58b in high yield as previously reported [25,
26]
With respect to the synthetic approach towards compound 59a, two equivalents of
thiosemicarbazide were used in the first step, while two equivalents of ethyl-bromoacetate and
sodium acetate were added to the bis-thiosemicarbazone in the second step. Similarly, the addition
of two equivalents of anhydrous potassium carbonate and benzyl bromide allowed the synthesis of
59b starting from 59a.
Compounds 1a, 14a [27], 32a, 32b [28], 2a-7a, 1b-7b, 10a-12a, 10b-12b, 15a-20a, 14b-20b, 22a-
29a, 22b-29b, 37a, 37b, 42a-47a, 42b-47b, 50a, 50b, 51a, 51b, 53a, 53b, 55a, 55b, 57a and 57b
[29] were already fully characterized in our previous works, whereas compounds 35a, 36a [30], 48a
1
[31] and 56a [32] were reported in the literature by other research groups. Some representative H
and 13C NMR spectra are reported in the Supporting information.
8.1.1 2-(2-(Hexan-2-ylidene)hydrazono)thiazolidin-4-one (8a). Light orange powder, mp 98-100
°C, 91% yield; 1H NMR (400 MHz, DMSO-d6): δ 0.87-0.91 (m, 3H, CH3), 1.28-1.34 (m, 2H, CH2),
28