Synthesis and characterization of 1,2,4-triazolo[1,5-a]pyrimidine-2-carboxamide-based compounds targeting the PA-PB1 interface of influenza A virus polymerase

Article abstract of DOI:10.1016/j.ejmech.2020.112944

Influenza viruses (Flu) are responsible for seasonal epidemics causing high rates of morbidity, which can dramatically increase during severe pandemic outbreaks. Antiviral drugs are an indispensable weapon to treat infected people and reduce the impact on

Full text of DOI:10.1016/j.ejmech.2020.112944

Journal Pre-proof
Synthesis and Characterization of 1,2,4-Triazolo[1,5-a]pyrimidine-2-carboxamide-
based Compounds Targeting the PA-PB1 Interface of Influenza A Virus Polymerase
Serena Massari, Chiara Bertagnin, Maria Chiara Pismataro, Anna Donnadio, Giulio
Nannetti, Tommaso Felicetti, Stefano Di Bona, Maria Giulia Nizi, Leonardo Tensi,
Giuseppe Manfroni, Maria Isabel Loza, Stefano Sabatini, Violetta Cecchetti, Jose
Brea, Laura Goracci, Arianna Loregian, Oriana Tabarrini
PII:
S0223-5234(20)30916-8
DOI:
https://doi.org/10.1016/j.ejmech.2020.112944
EJMECH 112944
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 13 September 2020
Revised Date: 13 October 2020
Accepted Date: 13 October 2020
Please cite this article as: S. Massari, C. Bertagnin, M.C. Pismataro, A. Donnadio, G. Nannetti, T.
Felicetti, S. Di Bona, M.G. Nizi, L. Tensi, G. Manfroni, M.I. Loza, S. Sabatini, V. Cecchetti, J. Brea, L.
Goracci, A. Loregian, O. Tabarrini, Synthesis and Characterization of 1,2,4-Triazolo[1,5-a]pyrimidine-2-
carboxamide-based Compounds Targeting the PA-PB1 Interface of Influenza A Virus Polymerase,
European Journal of Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112944.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
© 2020 Elsevier Masson SAS. All rights reserved.

Graphical Abstract

Synthesis and Characterization of 1,2,4-Triazolo[1,5-a]pyrimidine-2-
carboxamide-based Compounds Targeting the PA-PB1 Interface of Influenza A
Virus Polymerase
Serena Massari,a,* Chiara Bertagnin,b Maria Chiara Pismataro,a Anna Donnadio,a Giulio Nannetti,b
Tommaso Felicetti,a Stefano Di Bona,c Maria Giulia Nizi,a Leonardo Tensi,c Giuseppe Manfroni,a
Maria Isabel Loza,d Stefano Sabatini,a Violetta Cecchetti,a Jose Brea,d Laura Goracci,c Arianna
Loregian,b,# Oriana Tabarrini.a,#
a Department of Pharmaceutical Sciences, University of Perugia, 06123 Perugia, Italy
b Department of Molecular Medicine, University of Padua, 35121 Padua, Italy
c Department of Chemistry, Biology and Biotechnology, University of Perugia, 06123 Perugia, Italy
d CIMUS Research Center, University of Santiago de Compostela, 15782, Santiago de Compostela,
Spain
*Corresponding authors: Serena Massari, Department of Pharmaceutical Sciences, University of
Perugia, 06123 Perugia, Italy; Tel: +39 075-5855134, Fax: +39 075-5855115, E-mail:
serena.massari@unipg.it
# Co-last authors

Abstract
Influenza viruses (Flu) are responsible for seasonal epidemics causing high rates of morbidity,
which can dramatically increase during severe pandemic outbreaks. Antiviral drugs are an
indispensable weapon to treat infected people and reduce the impact on human health, nevertheless
anti-Flu armamentarium still remains inadequate.
In search for new anti-Flu drugs, our group has focused on viral RNA-dependent RNA polymerase
(RdRP) developing disruptors of PA-PB1 subunits interface with the best compounds characterized
by cycloheptathiophene-3-carboxamide and 1,2,4-triazolo[1,5-a]pyrimidine-2-carboxamide
scaffolds. By merging these moieties, two very interesting hybrid compounds were recently
identified, starting from which, in this paper, a series of analogues were designed and synthesized.
In particular, a thorough exploration of the cycloheptathiophene-3-carboxamide moiety led to
acquire important SAR insight and identify new active compounds showing both the ability to
inhibit PA-PB1 interaction and viral replication in the micromolar range and at non-toxic
concentrations. For few compounds, the ability to efficiently inhibit PA-PB1 subunits interaction
did not translate into anti-Flu activity. Chemical/physical properties were investigated for a couple
of compounds suggesting that the low solubility of compound 14, due to a strong crystal lattice,
may have impaired its antiviral activity. Finally, computational studies performed on compound 23,
in which the phenyl ring suitably replaced the cycloheptathiophene, suggested that, in addition to
hydrophobic interactions, H-bonds enhanced its binding within the PAC cavity.
Keywords: Influenza virus, PA-PB1 heterodimerization, RNA-dependent RNA polymerase,
protein-protein interaction.

1. Introduction
Influenza (Flu) viruses are responsible for seasonal epidemics resulting in about 3 to 5 million cases
of severe respiratory illness and 290,000 to 650,000 deaths each year all over the world [1].
Moreover, they are also able to generate pandemic outbreaks that occur when new highly virulent
FluA subtypes generated by antigenic shift are transmitted from animals to humans and sustainably
spread among people.
In 1918, humanity experienced the “Spanish Flu” caused by FluA(H1N1) [2], an influenza
pandemic that intensively and speedily struck world population infecting about 500 million people
and killing from 20 to 40 million people globally [3]. Other two pandemics occurred in 20th century:
the “Asian Flu” caused by Flu A(H2N2) virus [4], which started in China in 1957 and spread
globally causing about one to four million deaths, and the “1968 Flu pandemic” caused by Flu
A(H3N2) virus, which started in Hong Kong and spread to the United States causing one million
deaths. In 2003, the avian Flu A(H5N1) re-emerged passing the species barrier but fortunately not
spreading sustainably from person to person, while, in 2009, the “swine Flu” A(H1N1) virus was
responsible for the first pandemic of the 21st century [5]; it started in Mexico and spread rapidly
around the world causing from 150,000 to 600,000 deaths [6]. To date, the avian Flu A(H5N1) and
Flu A(H7N9) are of particular concern to public health due to their potential to cause an influenza
pandemic [7].
Previous pandemic events from Flu viruses highlighted that a vaccine is an essential mean to
rapidly prevent the spread of infection, but, in the absence of a vaccine, antiviral drugs could be a
valid weapon to treat infected people and reduce the impact on human health. Unfortunately, a
universal anti-Flu vaccine does not exist yet, so in case of a pandemic event, a new specific vaccine
should be developed on the basis of the emerging strain.
Regarding the anti-flu treatment, until recently, the only approved anti-Flu drugs were the NA
inhibitors oseltamivir and zanamivir (laninamivir octanoate and peravimivir were approved only in
some Asian countries), and adamantanes targeting the M2 ion channels protein, although the letter

are no longer recommended due to the emergence of widespread resistance [8]. This limited
therapeutic armamentarium was recently enriched by a new class of anti-Flu drugs involved in the
inhibition of the RNA-dependent RNA polymerase (RdRP) [9–12]. In particular, the nucleoside
analog favipiravir was approved in 2014 in Japan for stockpiling against Flu pandemics [13]; to
date, the safety and efficacy of favipiravir are also being evaluated in seven clinical trials against
COVID-19. In 2018, the endonuclease inhibitor baloxavir marboxil was approved in both Japan and
the United States [14]; moreover, the cap-binding inhibitor pimodivir is currently in late phase
clinical trials [15].
The RdRP is thus emerging as an important drug-target [16–18]. It is a heterotrimer composed by
the polymerase basic protein 1 (PB1), polymerase basic protein 2 (PB2), and polymerase acidic
protein (PA, P3 in FluC) subunits. The three RdRP subunits are connected in head-to-tail fashion by
extensive interactions that occur between the PB1N and PAC termini and the PB1C and PB2N
termini. In the context of the viral ribonucleoprotein complex (vRNP), the RdRP plays a key role in
the viral life cycle performing both the transcription and replication of viral genome [19,20].
The publication of crystal structures of PAC-PB1N interface (pdb codes: 3CM8 [21] and 2ZNL [22])
prompted the search for small-molecule inhibitors of this protein-protein interaction (PPI) able to
interfere with RdRP functions [23–25]. We have been pioneers in this research field, identifying
several compounds characterized by different chemotypes. In particular, with the exception of a
compound AL18 that was found serendipitously [26], all the other PA-PB1 inhibitors have been
identified by an initial structure-based drug design [27,28] followed by hit-to-lead optimization
campaigns, such as cycloheptathiophene-3-carboxamide (cHTC) [29–31], 1,2,4-triazolo[1,5-
a]pyrimidine (TZP) [32], pyrazolo[1,5-a]pyrimidine [33] and 3-cyano-4,6-diphenylpyridine [34]
derivatives. During the optimization of cHTC derivative 1 [27] and TZP derivative 2 [27] (Figure
1), hybrid compounds 3 and 4 (Figure 1) were prepared by joining the cHTC and the TZP moieties,
exhibiting very interesting activities [32]. In particular, compound 3, with an IC50 = 1.1 µM,
emerged as one of the most potent among the small-molecule PA-PB1 inhibitors developed so far.

Figure 1. Structures and biological activities of compounds 1-4 previously reported [27,32].
Accordingly, it affected PA-PB1 binding in the cell cytoplasm and blocked the intranuclear
translocation of PA, which requires the formation of the PA-PB1 complex. Compound 3 also
showed good anti-polymerase activity (IC50 = 12 µM in minireplicon assay) and broad anti-FluA
and -FluB activity (EC50s ranging from 7 to 25 µM, in MDCK cells) without showing any
cytotoxicity up to concentrations of 250 µM. Compound 4 exhibited slightly better antiviral activity
(EC50 values ranging from 5 to 14 µM against FluA and FluB strains), even if it was endowed with
lower ability to inhibit the PA-PB1 interaction (IC50 = 28 µM) [32].
Computational studies performed on the positional isomers 3 and 4 within the PAC cavity suggested
that the molecules have a different orientation and extents of hydrophobic and H-bond interactions
within the cavity [32]. Thus, while the elongated shape of 4 allows it to recognize all the three
hydrophobic regions described by Liu and Yao [35] within the PB1 binding site, compound 3 is

shifted toward the opposite side of the cavity and matches only the first hydrophobic region,
generated by W706. Nevertheless, compound 3 establishes a very favorable H-bond between its C-2
amidic carbonyl group and the Q408 residue, which could be the reason for its efficient inhibition
of PA-PB1 heterodimerization.
In the present study, additional hybrid compounds were synthesized as analogues of compound 3
(compounds 5-18, Figure 2) and compound 4 (compounds 19-26, Figure 2). By mainly focusing on
the cHTC core, various structural modifications were undertaken investigating the cycloheptane, the
thiophene and the 2-carboxamide moieties. From the antiviral activity, SAR insights were obtained
with the main indication entailing the favorable replacement of the cHTC core by a simpler 2-
carbamoylphenyl moiety. In depth studies were pursued to determine the ability of the compounds
to interfere with RdRP functions, their metabolic stability as well as to predict their binding mode
within the PAC cavity. Further studies were also performed on a couple of regioisomers to
investigate the different behavior in inhibiting PA-PB1 interaction and viral growth.

Figure 2. Structure of the compounds synthesized in this study. Compounds 5-18 are analogues of
compound 3 and compounds 19-26 are analogues of compound 4.
2. Results and Discussion
2.1. Chemistry
The synthesis of all the target compounds 5-13, 15-22 and 24-29 was accomplished, as reported in
Schemes 2-8, by coupling reaction of the appropriate reagent with 5-methyl-7-phenyl-
[1,2,4]triazolo[1,5-a]pyrimidine-2-carbonyl
chloride
30
[32]
or
7-methyl-5-phenyl-
[1,2,4]triazolo[1,5-a]pyrimidine-2-carbonyl chloride 31 [32]. As shown in Scheme 1, derivatives 30
and 31 were obtained by chlorination of the corresponding carboxylic acids [32], which were in turn
prepared by cyclocondensation of ethyl 5-amino-1,2,4-triazole-3-carboxylate [36] and 1-
phenylbutane-1,3-dione in acetic acid at reflux furnishing ethyl 5-methyl-7-phenyl-
[1,2,4]triazolo[1,5-a]pyrimidine-2-carboxylate
[37]
and
ethyl
7-methyl-5-phenyl-
[1,2,4]triazolo[1,5-a]pyrimidine-2-carboxylate [37], followed by basic hydrolysis.
Scheme 1. Synthetic route of the intermediates 30 and 31. Reagents and conditions: (a) glacial
acetic acid, reflux; (b) NaOH, MeOH, reflux; (c) oxalyl chloride, CH2Cl2, DMF, room temperature
(r.t.).
The synthesis of tetrahydrobenzothiophene- (5 and 19), cyclopentathiophene- (6 and 20), 4,5-
dimethylthiophene- (7), 6-ethyltetrahydrobenzothieno- (8), and benzothiophene- (9) 3-carboxamide
derivatives entailed the preparation of reagent 36 [38], 37 [38], 38 [39], 39 [40], and 40 [41],
respectively (Scheme 2). In particular, compounds 36-39 were obtained by applying the one-pot
Gewald reaction entailing the reaction of alkyl ketones 32-35 with cyanoacetamide in the presence

of sulphur and N,N-diethylamine or morpholine in EtOH. Then, by using conditions different from
those reported in literature [41], compound 40 was obtained by oxidization of 39 by using p-
chloranil in 1,4-dioxane. Coupling reaction of intermediates 36-39 with 30 and reagents 36 and 37
with 31 in CH2Cl2 in the presence of N,N-diisopropylethylamine (DIPEA) provided the target
compounds 5-8, 19 and 20.
Scheme 2. Synthetic route of the target compounds 5-8, 19 and 20. Reagents and conditions: (a)
sulphur, N,N-diethylamine, EtOH, rt or sulphur, morpholine, EtOH, reflux; (b) compound 30 or 31,
CH2Cl2, DIPEA, rt; (c) p-chloranil, 1,4-dioxane, 90 °C.
The synthesis of cHTC derivative 10 entailed the preparation of key intermediate 45, as reported in
Scheme 3. The preparation of thiophene-2-carboxamide derivatives was reported in literature by
straight reaction of 2-chloro-cyanoderivatives with 2-mercaptoacetamide [42]. However, difficulties
in the preparation of the latter as well as its commercial availability led us to undertake a different
synthetic route entiling the preparation of ethyl carboxylate intermediate 43 and its successive
transformation into carboxamide derivative 45. In particular, cycloheptanone was reacted with
DMF and POCl3 to give intermediate 41 [43] that was used for the preparation of compound 42 [44]
by reaction with hydroxylamine hydrochloride in N-methyl-2-pyrrolidinone (NMP). Intermediate

42 was then reacted with ethyl thioglycolate in presence of K2CO3 in a MeOH/THF mixture, to give
compound 43 [45]. In order to convert the ethyl carboxylate moiety of compound 43 into a
carboxamide group, it was reacted with hydrazine hydrate at 80 °C to give carbohydrazide
derivative 44, which was then treated with Ni-Raney in DMF at 90 °C providing compound 45.
Coupling reaction of 45 with 30 furnished the target derivative 10.
Scheme 3. Synthetic route of the target compound 10. Reagents and conditions: (a) DMF, POCl3,
from 0 °C to r.t.; (b) NMP, NH2OH hydrochloride, 115 °C; (c) K2CO3, MeOH/THF (5:1), ethyl
thioglycolate, reflux; (d) NH2NH2 hydrate, 80 °C; (e) Ni-Raney, DMF, 90 °C; (f) compound 30,
CH2Cl2, DIPEA, rt.
Through an analogous procedure, benzothiophene-2-carboxamide derivative 11 and thiophene-2-
carboxamide derivatives 13 and 22 were prepared as reported in Schemes 4 and 5, respectively.
Thus, ethyl carboxylate derivative 46 [46], prepared by reacting 2-chlorobenzonitrile and ethyl
thioglycolate in presence of KOH in DMF, and methyl 3-aminothiophene-2-carboxylate were
reacted with hydrazine hydrate providing carbohydrazide derivatives 47 [47] and 49 [47], which
were in turn treated with Ni-Raney to give carboxamide derivatives 48 [48] and 50 [49],
respectively. Coupling reaction of 48 with 30 and 50 with 30 and 31 provided derivatives 11, 13
and 22, respectively.

Scheme 4. Synthetic route of the target compound 11. Reagents and conditions: (a) ethyl
thioglycolate, DMF, KOH, from 0 °C to 80 °C; (b) NH2NH2 hydrate, 80 °C; (c) Ni-Raney, DMF,
90 °C; (d) compound 30, CH2Cl2, DIPEA, rt.
Scheme 5. Synthetic route of the target compounds 13 and 22. Reagents and conditions: (a)
NH2NH2 hydrate, 80 °C; (b) Ni-Raney, DMF, 90 °C; (c) compound 30 or 31, CH2Cl2, DIPEA, rt.
Thienopyridine-2-carboxamide derivatives 12 and 21 were prepared, as reported in Scheme 6, by
coupling reaction of 30 and 31, respectively, with carboxamide derivative 52 [50], which was
obtained by reaction of 2-chloroniconitrile and thiourea in EtOH at reflux to give intermediate 51,
followed by reaction with 2-bromoacetamide in DMF in the presence of DIPEA.

Scheme 6. Synthetic route of the target compound 12 and 21. Reagents and conditions: (a) thiourea,
EtOH, reflux; (b) 2-bromoacetamide, DIPEA, DMF, rt; (c) compound 30 or 31, CH2Cl2, DIPEA, rt.
A one-pot Gewald reaction was initially applied for the synthesis of derivatives 16 bearing a C-3
pyridin-2-yl-carbamoyl moiety, 15 and 24 characterized by a C-3 carboxylic acid, and
cycloheptathienooxazinones 18 and 26 (Scheme 7). Thus, cycloheptanone was reacted with 2-
cyano-N-pyridin-2-ylacetamide [51] and ethyl 2-cyanoacetate, providing intermediates 53 [29] and
54 [52], respectively. Reagent 53 was then coupled with 30 to give target compound 16, while
reagent 54 was reacted with 30 and 31 yielding ethyl ester intermediates 55 and 56, respectively.
Hydrolysis under basic conditions of 55 and 56 provided acid derivatives 15 and 24, which were in
turn cyclized to give tricyclic derivatives 18 and 26. Cyclization of acid 15 was performed in acetic
anhydride at 100 °C providing 18 in very low yield (19%). As a consequence, cyclization of 24 was
performed through an alternative procedure entailing the use of EDC, DIPEA, and 1-
hydroxybenzotriazole in CH2Cl2, which permitted to obtain derivative 26 in 52% yield.
Finally, benzothiazole derivatives 17 and 25 were synthesized as reported in Scheme 8, by coupling
reaction of 2-aminobenzothiazole with 30 and 31, respectively.

Scheme 7. Synthetic route of the target compounds 15, 24, 18 and 26. Reagents and conditions: (a)
sulphur, N,N-diethylamine, EtOH, rt; (b) compound 30 or 31, CH2Cl2, DIPEA, rt; (c) LiOH,
H2O/THF (1:1), 50 °C; (d) Ac2O, 100 °C; (e) DIPEA, EDC, HOBt, CH2Cl2, from 0 °C to rt.
Scheme 8. Synthetic route of the target compounds 17 and 25. Reagents and conditions: (a)
compound 30 or 31, CH2Cl2, DIPEA, rt.
2.2. Inhibition of PA-PB1 heterodimerization and evaluation of anti-Flu activity
The study started from hit compound 3, considering its better ability to interfere with PA-PB1
interaction. In particular, compounds 5-18 (Table 1) were synthesized by structurally modifying the
cycloheptane, the thiophene, and the 2-carboxamide moieties of compound 3, while keeping the 5-
methyl-7-phenyl-[1,2,4]triazolo[1,5-a]pyrimidine-2-carboxamide portion.
All the compounds were evaluated for their ability to inhibit the PA-PB1 subunits interaction in an
ELISA-based assay, including the Tat-PB11-15 peptide as a positive control, and for the antiviral

activity in FluA virus-infected MDCK cells by plaque reduction assays (PRA) with the A/PR/8/34
(PR8) strain, using ribavirin (RBV), a known inhibitor of RNA viruses polymerase, as a positive
control. To exclude that the observed antiviral activities could be due to toxic effects in the target
cells, all the compounds were tested in parallel by MTT assays in MDCK cells.
Regarding the modifications made on the cycloheptane ring, size reduction from
cycloheptathiophene to tetrahydrobenzothiophene (compound 5) and cyclopentathiophene
(compound 6) provided compounds showing a comparable anti-Flu activity (EC50 = 32 and 22 μM,
respectively, vs 21 μM of 3) but a slightly decreased ability to inhibit PA-PB1 interaction (IC50 = 15
and 6.2 μM, respectively) with respect to 3 (IC50 = 1.1 μM). On the other hand, the deletion of the
cycloheptane ring gave 4,5-dimethylthiophene 7 devoid of both the activities. The alkylation and
aromatization of the tetrahydrobenzene of compound 5 were explored in 6-ethyl-
tetrahydrobenzothiophene 8 and benzothiophene 9, respectively, which were both endowed with
anti-Flu activity (EC50 = 15 and 49 μM, respectively) but devoid of the ability to disrupt PA-PB1
interaction. Of note, derivative 8 emerged as the most active anti-Flu compound herein reported,
even more than hit compound 3. These results suggested that the presence of a cycloalkyl moiety
fused to the thiophene ring is important to obtain anti-Flu activity, although its aromatization is still
tolerated but detrimental for anti-PA-PB1 activity.
Then, a set of four derivatives was synthesized in which the thiophene-based core was linked to the
TZP moiety by the C-3 position. The thienopyridine derivative 12 was the most potent PA-PB1
inhibitor among the compounds herein reported (IC50 of 3.3 μM), but at the expense of the antiviral
activity (EC50 > 100 μM), while derivative 13 exhibited a balanced biological profile (IC50 of 31
μM and EC50 of 43 μM). These results indicated that, in this series of compounds, the lack of a
cycloalkyl ring fused to the thiophene does not impair the activity.
No anti-PA-PB1 activity was shown by cycloheptathiophene compound 10 and benzothiophene
compound 11, the strict analogs of compounds 3 and 9, respectively, although the latter showed
anti-Flu activity (EC50 = 22 μM) comparable to that of hit compound 3 (EC50 = 21 μM). Of note,

the whole cHTC core can be replaced by a simpler benzene ring to achieve anti-PA-PB1 activity, as
shown by the 2-carbamoylphenyl derivative 14 [53], characterized by a good IC50 of 11 μM but
devoid of anti-Flu activity. Finally, in order to explore the role of the C-3 carboxamide moiety, it
was replaced by a bioisosteric carboxylic group or by an N-(2-pyridyl)-carboxamide moiety, which
also characterized hit compound 1, as in derivatives 15 and 16, respectively. Moreover, a
benzothiazole replaced the cycloheptathiophene ring in derivative 17, while the two amide linkages
at the C-2 and C-3 positions of the cycloheptathiophene were constrained into a 1,3-oxazinone ring
in the tricyclic derivative 18. The four compounds showed a comparable anti-PA-PB1 activity, with
IC50s ranging from 19 to 23 μM, and compounds 15 and 16 also shared a similar anti-Flu activity
(EC50 = 50 and 48 μM, respectively). The benzothiazole ring (compound 17) was not suitable for
achieving anti-Flu activity, while the cycloheptathienooxazinone derivative 18 was the most active
in inhibiting the viral growth with an EC50 of 26 μM.
All the compounds were non-toxic up to 250 μM concentration, with the exception of derivatives 10
and 18, which, however, showed a mild cytotoxic effect (CC50 = 90 and 101 μM, respectively).
Table 1. Structure and Biological Activity of Compound 3 Analogues.
ELISA
PRA
Cytotoxicity in
MDCK cells
CC50, µMc
PA-PB1
Interaction Assay
IC50, µMa
Compd
3
R
in MDCK cells
EC50, µMb
1.1 ±0.3
21 ±4
>250

15 ± 3
6.2 ± 0.3
>200
32 ± 8
22 ± 7
>50
>250
>250
>250
5
6
7
>200
>200
>200
15 ± 2
49 ± 1
>50
>250
>250
8
9
90 ± 7
10
>200
22 ± 4
>250
11
3.3 ± 0.7
31 ± 5
11 ± 3
19 ± 1
>100
43 ± 2
>100
>250
250
12
13
14
15
>250
>250
50 ± 3

23 ± 2
48 ± 2
>250
16
17
21 ± 1
19 ± 1
>50
>250
26 ± 1
101 ± 2
18
10 ± 2
41 ± 5
>250
>100
RBV
Tat-PB11-15
peptide
35 ± 4
a
Activity of the compounds in ELISA PA-PB1 interaction assays. The IC50 value represents the compound
concentration that reduces by 50% the interaction between PA and PB1. b Activity of the compounds in plaque
reduction assays with the Flu A A/PR/8/34 strain. The EC50 value represents the compound concentration that
inhibits 50% of plaque formation. c Cytotoxicity of the compounds in MTT assays. The CC50 value represents the
compound concentration that causes a decrease of cell viability of 50%. All the reported values represent the
means ± SD of data derived from at least three independent experiments in duplicate.
In the next step, on the basis of the biological results obtained, for all the compounds showing the
ability to inhibit PA-PB1 interaction or both the PA-PB1 interaction and Flu replication
(compounds 5, 6 and 12-18), the corresponding analogues functionalized with a 7-methyl-5-phenyl-
[1,2,4]-triazolo[1,5-a]pyrimidine-2-carboxamide moiety were prepared. As shown in Table 2, the
thienopyridine derivative 21 and benzothiazole derivative 25 showed an opposite profile to that
exhibited by their isomers 12 and 17. Indeed, both the compounds lost the ability to interfere with
PA-PB1 heterodimerization but acquired some anti-Flu activity (EC50 = 43 and 40 μM,
respectively), which however in compound 21 was coupled with cytotoxicity (CC50 = 51 μM). On
the other hand, acid derivative 24 and tricyclic compound 26 showed good ability to inhibit PA-
PB1 interaction (IC50 = 18 and 15 μM, respectively), with 26 exhibiting slightly higher anti-Flu
activity (EC50 = 40 μM) than 24 (EC50 = 70 μM). A decreased anti-PA-PB1 activity was instead
shown by thiophene derivative 22 (IC50 = 50 μM), which was also devoid of antiviral activity. Also

in this series, the cHTC could be favorably replaced by a phenyl ring, as shown by compound 23
[53] that, besides possessing very good anti-PA-PB1 activity (IC50 = 7 μM), also acquired
interesting anti-Flu activity (EC50 = 31 μM) compared to its analogue 14 (IC50 = 11 μM and EC50
>100 μM).
In summary, the synthesis of novel hybrid compounds, aimed at clarifying the best substitution
pattern for the C-2 position of the TZP ring, has led to the identification of some interesting
compounds able to inhibit the PA-PB1 interaction and the viral growth. Interesting SAR insights
have been outlined. In particular, the cHTC portion can be replaced by a benzene ring (as in
compounds 14 and 23), leading to a huge simplification also in terms of a chemical synthesis. The
cycloheptathiophene moiety can be also favorably replaced by the tricyclic
cycloheptathienooxazinone core (as in compounds 18 and 26), which thus emerged as new scaffold
able to ensure both anti-PA-PB1 and anti-Flu activity.
Among the synthesized derivatives, compounds 8, 9, 11, 21 and 25 were able to interfere only with
viral replication (EC50s ranging from 15 to 49 µM). For these compounds, a different mechanism of
action could be hypothesized, although none of them showed potent anti-Flu activity deserving
mechanistic studies. On the other hand, compounds 12, 14 and 17 showed the ability to disrupt PA-
PB1 heterodimerization (IC50s ranging from 3.3 to 50 µM) but no antiviral activity, leading to
hypothesize an inappropriate pharmacokinetic profile. Thus, preliminary chemical/physical studies
were performed for compound 14 and its isomer 23, which was instead endowed with both anti-PA-
PB1 and anti-Flu activities.
Table 2. Structure and Biological Activity of Compound 4 Analogues.

ELISA
PA-PB1
PRA
Cytotoxicity in
MDCK cells
CC50, µMc
Compd
4
R
in MDCK cells
Interaction Assay
EC50, µMb
IC50, µMa
28 ±1
N.D.d
N.D.
8 ±2
N.D.
N.D.
>250
N.D.
N.D.
19
20
>200
50 ± 28
7 ± 1
43 ± 7
>100
51 ± 1
240 ± 8
>250
21
22
23
31 ± 10
18 ± 11
70 ± 15
40 ± 13
>250
24
25
>200
233 ± 5
15 ± 5
40 ± 6
10 ± 2
>50
26
>250
RBV

Tat-PB11-15
peptide
35 ± 4
41 ± 5
>100
For the definition of a IC50, bEC50, and cCC50, see Table 1. d N.D. = not determined due to solubility issues.
2.3. Chemical-physical properties determination for compounds 14 and 23
Trying to investigate as to why some compounds were able to interfere with PA-PB1 interface in
the micromolar range while not showing anti-Flu activity, some preliminary chemical-physical
properties were determined. To this aim, we selected the couple of regioisomer 14 and 23 both
showing very good and comparable ability to inhibit PA-PB1 interaction (IC50 = 11 and 7 µM,
respectively) but only 23 showing anti-Flu activity (EC50 = 31 μM).
First, we wished to exclude the hypothesis that the activity shown by compound 14 could be related
to an artefact. Various mechanisms of assay interference or promiscuous behaviour have been
described as responsible for Pan-Assay Interference Compounds (PAINS) activity, including
chemical aggregation [54]. To test this hypothesis, compounds 14 and 23 were examined by using
the ZINC15 remover filter (http://zinc15.docking.org/patterns/home/) [55] and none of them were
found as potential aggregators or PAINS.
We next hypothesized that, in the case of compound 14, solubility issues could have compromised
the antiviral evaluation. Indeed, similarly to PPI inhibitors reported in the literature that are
generally planar and hydrophobic molecules, the PA-PB1 inhibitors developed by us often suffered
from poor solubility, as already highlighted for compounds 3 and 4 [32]. Thus, we evaluated the
equilibrium solubility in phosphate buffer at pH 7.4 for the two isomers. Briefly, stock solutions
(10-2 M) of the compounds in DMSO were diluted to decreased molarity, from 300 μM to 0.1 μM,
in 384 well transparent plate (Greiner 781801) with 1% DMSO: 99% PBS buffer, incubated at 37
ºC and read after 2 hours in a NEPHELOstar Plus (BMG LABTECH). The results were adjusted to
a segmented regression to obtain the maximum concentration in which compounds are soluble.

Compound 23 exhibited a higher equilibrium solubility (82.2 µM, 30.60 μg/mL) than 14 (17.6 µM,
6.55 μg/mL), suggesting that the low solubility of 14 could be at the basis of its inactivity in PRA
assay.
Since the solubility of a molecule is a result of the dissociation from its crystal unit cell and the
hydration once in solution, we investigated both these properties for the two isomers. In particular,
in order to understand whether the difference in solubility shown by the compounds could originate
by a different hydration, we calculated the Log Po/w value (average of five predictions calculated
by using http://www.swissadme.ch/) [56] for 14 and 23. Compounds showing poor solubility by
poor hydration are referred to as ‘greaseball’ molecules and LogP values of 2-3 are considered as
the cut-off point for hydration becoming a significant limitation for solubility [57]. Compounds 14
and 23 showed Log Po/w values of 2.15 and 2.07, respectively, suggesting that their different
solubility was not originating from the degree of polarity and related solvation.
Then, we sought to investigate whether their different solubility in aqueous solvent could be
dependent from the crystal lattice. Indeed, compounds with a strong crystal lattice, commonly
referred to as ‘brick dust’ molecules, often show a limited capacity to dissociate from the solid form
[58]. To this aim, we performed X-Ray analysis for the two isomers. Single crystals of 14 and 23
suitable for X-Ray analysis were obtained by crystallization by EtOH/DMF. The Ortep drawings of
the asymmetric unit of the crystal structures are shown in Figure 3. Compound 14 crystallized in the
Triclinic space group P-1 with unit cell parameters: a = 6.412 (2) Å, b = 8.167 (2) Å, c = 16.243(5)
Å, b ¼ 90.63(3), α = 88.374(11)°, β = 87.191(11)°, γ = 85.292(12)°, V = 846.5(4) Å3, Z = 2, while
compound 23 crystallized Monoclinic space group P 2(1)/n with unit cell parameters: a = 7.9084(4)
Å, b = 16.8578(13) Å, c = 12.4907(10) Å, b ¼ 90.63(3), α = γ = 90°, β = 91.014(3)°, γ =
85.292(12)°, V = 1665.0(2) Å3 and Z = 4. The other structural details are given in the supporting
information (Tables S1 and S2). The solution and the refinement of the crystal structures of the two
compounds allowed us to highlight similarities and differences of the two species in the solid state.
In both cases, the configuration of the molecule is the same with the terminal amide group pointing

in the same direction with respect to the NH moiety of the secondary amide due to the presence of
an intramolecular hydrogen bond between O1 and H2N. On the contrary, the dihedral angle
involving the atoms C6-C7-N2-C8 (2.64° for 14, 27.10° for 23) and the torsion angle of the phenyl
substituent with respect to the plane of the pyrimidine ring (35.46° for 14, 11.88° for 23) are
relevantly different due to the presence of selective weak interactions that stabilize the two crystal
structures. Particularly, the distortion of the structure of 23 is caused by the presence of a relatively
strong intermolecular hydrogen bond between H1Nb and N3 (2.275 Å), while the phenyl
substituent on 14 is involved in multiple intermolecular π interactions with the phenyl moieties of
other molecules of 14 causing a parallel displaced π - π stacking (planes distance 3.710 Å,
centroids distance 4.072 Å) [59], which provides a stabilization of the three-dimensional crystal
lattice. In the latter case thus, the molecular packing can be attributed also to the formation of
favorable interactions between neighboring aromatic side chains. The established interactions and
consequently the different molecular packing may account for a substantial amount of the crystal
stabilization. Accordingly, as they account for many characteristics of compounds, such as
solubility, these features may justify the differences into the solubility values for compounds 14 and
23, which could have compromised the antiviral evaluation.
Figure 3. Asymmetric unit of 14 and 23, showing a partial atom-numbering scheme. Displacement
ellipsoids are drawn at 50% probability level.

2.4. In depth characterization of compound 23
2.4.1. Inhibition of Flu RdRP activity in a minireplicon assay
Then we investigated whether the compound ability to disrupt the PA−PB1 interaction in vitro
correlated with the ability to interfere with the catalytic activity of FluA RdRP in a cellular context,
by disrupting the correct assembly of the polymerase complex. To this aim, a minireplicon assay
was performed for compound 23. 293T cells were co-transfected with plasmids for the expression
of FluA nucleoprotein (NP), PA, PB1, and PB2 proteins and the firefly luciferase RNA, flanked by
the noncoding regions of Flu A/WSN/33 segment 8, in the presence of the test compound or DMSO
as a control. The expression of the firefly reporter gene indicates that a negative-sense RNA is
synthesized and is reconstituted intracellularly into functional vRNP in which all four NP, PA, PB1,
and PB2 proteins are co-expressed and interact with each other. Compound 23 exhibited a potent
inhibitory effect on FluA polymerase activity, showing an EC50 of 5.8 ± 2.0 μM, and thus being
more effective than RBV (24 ± 4 μM). These data confirmed the ability of compound 23 to inhibit
RdRP functions by interfering with PA-PB1 heterodimerization.
2.4.2. Metabolic stability in human liver microsomes
Metabolic stability in human liver microsomes (HLM) was studied by monitoring the percentage of
unchanged substrate and the formation of the metabolite(s) at five time points (0, 10, 20, 40 and 60
min). LC-MS raw data were analyzed by using Mass-Metasite (version 3.3.6; Molecular Discovery
Ltd., Middlesex, UK) [60,61] and WebMetabase (version 4.0.6; Molecular Discovery Ltd.,
Middlesex, UK) [62,63] software for automatic metabolite identification and structure elucidation
as well as for kinetic analysis. After 60 min incubation, the 76% of compound 23 was in its
unchanged form, indicating a good metabolic stability. In addition, the kinetic analysis performed in
WebMetabase allowed to estimate the half-life value and the intrinsic clearance for compound 23,
as 166 min and 4.18 µL/min* mgprot, respectively (Figure S1, supporting information). The analysis
of formed metabolites allowed for the detection of one metabolite only, obtained by an aromatic
hydroxylation reaction at the 2-carboxybenzamide moiety (Figure S2, supporting information).

Although the MS/MS fragmentation is not sufficient to identify the exact position of the site of
metabolism in the aromatic ring, the most probable site for hydroxylation appears to be at C-4
position, according to MetaSite [64] predictions within the WebMetabase analysis tool (Table S3
and Figure S3, supporting information).
2.4.3. Human plasma protein binding
Regarding the human plasma protein binding, samples were analyzed by employing a rapid
equilibrium system device and adding them in buffer and measuring the compound presence in the
human plasma compartment after 4 h of incubation by means of UPLC/MS/MS. It was observed
that compound 23 showed an unbound fraction of 22.45%, which together with the good metabolic
stability predicts a good bioavailability of the compound in human plasma.
2.4.4. Computational studies
Finally, with the aim to gain information on how two different moieties, such as the cHTC and the
2-carbamoylphenyl, showed a favorable ability to disrupt PA-PB1 interaction, computational
studies were performed to predict the binding mode of benzamide derivative 23 within the PA
cavity with respect to hit compound 4. For comparative purpose, also its isomer 14, which was
active in ELISA-based assay but devoid of any antiviral activity, was studied in comparison to hit
compound 3.
The study was conducted using the same method applied for the evaluation of the binding mode of
compounds 3 and 4 using the FLAP software [65], and the most probable binding poses for the two
isomers are illustrated in Figure 4. In agreement with the biological results, both compounds
showed an efficient binding within the PAC, although through different orientations and
interactions. In particular, both compounds were found to efficiently interact with W706 through
hydrophobic interactions, as observed for all the inhibitors of the PA-PB1 interaction we have
studied so far [24].
Similarly to 4, compound 23 displayed a favorable hydrophobic interaction with all the three
hydrophobic regions previously described by Liu and Yao [35]. Indeed, according to the proposed

description of the protein cavity, the first hydrophobic region is centered on residues W706 and
F411, the second one is defined by F710 and L666, and the third one includes L640, V636, M595,
and W619. However, differently from 4, 23 seems to be involved in the formation of a favorable H-
bond between the 2-carboxamide NH group and the hydroxyl group of T639. This amino acid was
not found to be involved in such kind of interaction in other compounds studied so far by us.
Concerning compound 14, the hydrophobic interaction occurs only in the first hydrophobic pocket,
in a similar fashion to 3. Differently from 3, a favored H-bonding with Q408 was established with
the 2-carboxamide carbonyl group instead of the carbonyl group of the amide linking the TZP and
the phenyl ring. However, according to the GRID fields analysis in FLAP, for 14 the replacement of
the cycloheptathiophene moiety with the phenyl ring led to a slightly different orientation that
facilitates the formation of a second H-bond with I621, one of the key residues of PA in binding of
PB1 (data not shown). Nevertheless, this slight change in the pose results in a reduced π- π
interaction. This data seems to confirm the importance of an efficient interaction with W706 for
inhibition of PA-PB1 heterodimerization.
In conclusion, the computational study confirmed the central interaction between the inhibitors and
W706, and the 2-carboxyamide group of 23 and 14 was pivotal in the interaction with PA by
establishing favorable H-bonds. This is also experimentally confirmed, since the analogues lacking
the 2-carboxamide substituent showed a very weak activity in the ELISA PA-PB1 interaction
assays [32].

Figure 4. FLAP binding poses for compounds 23 (A,B) and 14 (C,D). Two orientations of the same
pose in the PA cavity are illustrated to better visualize the predicted interactions. Compounds 14
and 23 are shown in sticks mode and in green color, while reference compounds 3 and 4 are shown
in lines style and purple color.
3. Conclusions
The Flu RdRP is emerging as a privileged drug-target as demonstrated by the most recently
approved compounds or in the pipeline such as favipiravir, baloxavir marboxil and pimodivir,
which act by inhibiting each of the three RdRP subunits.
By applying an alternative approach to inhibit Flu RdRP functions, since many years we have been
working on the development of compounds able to interfere with RdRP subunits interaction and in

particular with PA-PB1 heterodimerization. cHTC and TZP emerged among the best scaffolds able
to disrupt this interaction, and their merging led us to identify interesting hybrid compounds (3 and
4). Starting from these hits, in this work, two sets of analogues were synthesized by maintaining the
TZP core while deeply modifying the cHTC moiety. Some promising compounds showed the
ability to interfere with the PA-PB1 interaction, such as cyclopentathiophene derivative 6 (IC50 =
6.2 μM), phenyl derivative 23 (IC50 = 7 μM) and cycloheptathienooxazinones 18 and 26 (IC50 = 19
and 15 μM, respectively), coupled with good anti-Flu activity (EC50 ranging from 22 to 40 μM). On
the other hand, some compounds showed only anti-Flu activity (EC50 = 15 – 50 μM) leading to
hypothesize a different mechanism of action, or only anti-PA-PB1 activity (IC50 = 3.3 – 50 μM). To
investigate this latter behavior, chemical/physical properties were determined for the couple of
isomers 14 and 23 both showing anti-PA-PB1 activity but only 23 having anti-Flu properties.
Compound 14 showed a lower equilibrium solubility with respect to 23, which could arise from a
major molecular packing, as shown by X-Ray studies, which may account for a substantial amount
of the crystal stabilization. In particular, the phenyl ring at the C-7 position of the TZP nucleus is
involved in multiple intermolecular π- π interactions with the phenyl moieties of other molecules
causing stabilization of the three-dimensional crystal lattice. These data confirmed one of the main
drawbacks of PPI inhibitors that, interacting with interfaces generally flat and highly hydrophobic,
often suffer from low solubility. In agreement, the PAC cavity is characterized by hydrophobic
regions and hydrophobic interactions are indispensable to achieve an efficient small molecule-PA
binding. Computational studies performed on compounds 14 and 23 confirmed the crucial
hydrophobic interaction between the TZP core and W706 for inhibition of PA-PB1
heterodimerization. Nevertheless, favorable H-bonds seem to contribute to the interaction of the
compounds with the PAC.
This work furnished important structure insights that could exploited for the design of further
analogues, also endowed with better drug-like properties. Indeed, the good biological and
chemical/physical properties shown by compound 23, in which a simpler phenyl ring replaced the

cycloheptathiophene, highlighted the possibility to reduce unnecessary hydrophobic moieties. A
further increase of the hydrophilicity of the compounds could be also achieved by a structure-based
drug design aimed at introducing polar groups establishing favorable H-bonds with the PAC.
4. Experimental section
4.1. Chemistry
Commercially available starting materials, reagents, and solvents were used as supplied. All
reactions were routinely monitored by TLC on silica gel 60F254 (Merck) and visualized by using
UV or iodine. Flash column chromatography was performed on Merck silica gel 60 (mesh 230-
400). After extraction, organic solutions were dried over anhydrous Na2SO4, filtered, and
concentrated with a Büchi rotary evaporator at reduced pressure. Yields are of purified product and
were not optimized. HRMS spectra were registered on Agilent Technologies 6540 UHD Accurate
Mass Q-TOF LC/MS, HPLC 1290 Infinity. Purities of target compounds were determined by HPLC
Waters LC Module I Plus and evaluated to be higher than 95% (at both 254 (reported below in the
description of the compounds) and 296 nm). In order to evaluate their stability, the purity was also
determined by HPLC analysis monitored at 296 nm after eleven days, and it was over 95%. HPLC
conditions to assess the purity of final compounds were as follows: column: XTerra MS C18
column reversed-phase (3.5 μ spherical hybrid, 4.6 mm x 150 mm, 3.5 μm particle size); flow rate,
0.8 mL/min; acquisition time, 10 min; isocratic elution, mobile phase: acetonitrile 80%, water 20%,
formic acid 0.1%. 1H NMR and 13C NMR spectra were recorded on Bruker Avance DRX-400MHz
using residual solvents such as dimethylsulfoxide (δ = 2.48) or chloroform (δ = 7.26) as an internal
standard. Chemical shifts were recorded in ppm (δ) and the spectral data are consistent with the
assigned structures. The spin multiplicities are indicated by the: symbols s (singolet), d (doublet), t
(triplet), m (multiplet), and bs (broad singolet).
4.1.1. 2-Amino-6-ethyl-4,5,6,7-tetrahydrobenzo[b]thiophene-3-carboxamide (39). [40] A
mixture of 4-ethylcyclohexanone (0.5 g, 3.96 mmol) 2-cyanoacetamide (0.35 g, 4.16 mmol) and

sulfur (0.13 g, 4.16 mmol) in ethanol (5 mL) was heated to reflux. Morpholine (0.36 g, 4.16 mmol)
was added dropwise to the heating solution and the mixture was allowed to heat to reflux 4h. After
cooling, the reaction mixture was poured into ice/water obtaining a precipitate which was filtered
and treated with cyclohexane, to give 39 (0.7 g, 79%); 1H NMR (DMSO-d6, 400 MHz): δ 0.86 (t, J
= 7.4 Hz, 3H, CH2CH3), 1.16-1.35 (m, 4H, CH2CH3 and cyclohexane CH2), 1.45-1.55, 1.77-1.80
and 2.00-2.06 (m, each 1H, cyclohexane CH), 2.51-2.57 (m, 2H, cyclohexane CH2), 6.48 (bs, 2H,
CONH2), 6.86 (s, 2H, NH2).
4.1.2. 2-Aminobenzo[b]thiophene-3-carboxamide (40). [41] A mixture of 36 [38] (0.5 g, 2.55
mmol) and p-chloroanil (1.25 g, 5.10 mmol) in 1,4-dioxane (30 mL) was heated at 90 °C for 4h.
After cooling, the reaction mixture was filtered over celite and the filtrate was evaporated to
dryness, to give a residue that was stirred with saturated NaHCO3 solution for 15 min. The solution
was then extracted with CH2Cl2 and the organic layers were evaporated to dryness to give a grey
solid, which was purified by flash chromatography eluting with MeOH/CH2Cl2 (2%), to give 9
(0.29 g, 59%); 1H NMR (DMSO-d6, 200 MHz): δ 6.80-6.95 (m, 3H, benzothiophene CH and NH2),
7.15-7.20 (t, J = 7.15 Hz, 1H, benzothiophene CH), 7.50-7.60 (m, 4H, aromatic CH and CONH2).
4.1.3. General procedure for carbodiimide formation (Method A). To a solution of 5-methyl-7-
phenyl-[1,2,4]triazolo[1,5-a]pyrimidine-2-carboxylic
acid
[37]
or
5-methyl-7-phenyl-
[1,2,4]triazolo[1,5-a]pyrimidine-2-carboxylic acid [37] (1.0 equiv) in well dry CH2Cl2, oxalyl
chloride (3 equiv) was added and after 30 min dry DMF (2 drops) was added. After 2h, the reaction
mixture was evaporated to dryness to give 30 [37] or 31 [37] that was dissolved in well dry CH2Cl2
and added of the appropriate amine (1.0 equiv) and DIPEA (1.0 equiv). The reaction mixture was
maintained at r.t. until no starting material was detected by TLC. Then, it was worked up through
two procedures: (procedure 1) the reaction mixture was evaporated to dryness to give a residue that
was poured into ice/water providing a precipitate which was filtered and purified as reported in the
description of the compounds; or (procedure 2) the precipitate formed in the reaction mixture was
filtered and purified as reported in the description of the compounds.