Computer-aided design, synthesis and validation of 2-phenylquinazolinone fragments as CDK9 inhibitors with anti-HIV-1 tat-mediated transcription activity

Article abstract of DOI:10.1002/cmdc.201300287

The activity of the cyclin-dependent kinase 9 (CDK9) is critical for HIV-1 Tat-mediated transcription and represents a promising target for antiviral therapy. Here we present computational studies that, along with preliminary synthetic efforts, allowed us

Full text of DOI:10.1002/cmdc.201300287

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300287
Computer-Aided Design, Synthesis and Validation of 2-
Phenylquinazolinone Fragments as CDK9 Inhibitors with
Anti-HIV-1 Tat-Mediated Transcription Activity
Luca Sancineto,[a] Nunzio Iraci,[a] Serena Massari,[a] Vanessa Attanasio,[a, b]
Gianmarco Corazza,[b] Maria Letizia Barreca,[a] Stefano Sabatini,[a] Giuseppe Manfroni,[a]
Nilla Roberta Avanzi,[c] Violetta Cecchetti,[a] Christophe Pannecouque,[d]
Alessandro Marcello,*[b] and Oriana Tabarrini*[a]
The activity of the cyclin-dependent kinase 9 (CDK9) is critical
for HIV-1 Tat-mediated transcription and represents a promising
target for antiviral therapy. Here we present computational
studies that, along with preliminary synthetic efforts, allowed
us to identify and characterize a new class of nontoxic anti-
CDK9 chemotypes based on the 2-phenylquinazolinone scaf-
fold. Inhibition of CDK9 translated into the ability to interfere
selectively with Tat-mediated transactivation of the viral pro-
moter and in the inhibition of HIV-1 reactivation from latently
infected cells, with the most potent derivative 37 (2-(4-amino-
phenyl)-7-chloroquinazolin-4(3H)-one) showing an IC50 value of
4.0 mm. Because the herein reported 2-phenylquinazolinones
are merely fragments, they are largely optimizable, paving the
way to derivatives with improved potency.
Introduction
The human immunodeficiency virus type 1 (HIV-1) is a retrovirus
that requires integration into host chromatin to transcribe its
genome for productive infection. HIV-1 infection can be effec-
tively controlled by combination antiretroviral therapy (cART),
which improves the quality of life in infected individuals but
unfortunately fails to completely eradicate the virus even after
decades of treatment.[1,2] Viral reservoirs persist during therapy
mostly in memory Tcells that harbor a transcriptionally silent
integrated provirus.[3] Reactivation from latency critically de-
pends on the viral transactivator protein Tat, that acts like
a molecular switch between silent and productive transcrip-
tion.[4] In the absence of Tat, basal transcription from the long
terminal repeat (LTR) ultimately causes RNA polymerase II
(RNAPII) to pause after synthesis of a short RNA that includes
the Tat responsive element (TAR). The negative elongation fac-
tors are recruited to induce pausing of RNAPII on the promot-
er.[5] Tat counteracts RNAPII pausing by binding TAR and re-
cruiting the positive transcription elongation factor b (P-TEFb)
complex. The core constituents of P-TEFb are cyclin T1 (CycT1)
and the cyclin-dependent kinase 9 (CDK9) that phosphorylates
the RNAPII carboxy terminal domain (CTD) and the negative
transcription elongation factors, licensing RNAPII for produc-
tive elongation.[6]
Tat-mediated HIV-1 transcription represents a valid target to
prevent HIV-1 replication and its re-emergence from latency.[7]
Developing chemicals that target P-TEFb, which is a key ele-
ment in Tat-mediated activation of the HIV-1 LTR promoter but
is not needed for cell survival, could be a compelling strat-
egy.[8] Such “indirect antiviral drugs” should not select for
mutant viruses resistant to treatment and should be active
even against mutant viruses eventually escaping current thera-
pies. In addition, they may also be active against other viruses,
since replication of unrelated viruses often requires the same
cellular factors.
[a] Dr. L. Sancineto,+ Dr. N. Iraci,+ Dr. S. Massari, Dr. V. Attanasio,
Dr. M. L. Barreca, Dr. S. Sabatini, Dr. G. Manfroni, Prof. V. Cecchetti,
Dr. O. Tabarrini
Department of Chemistry and Technology of Drugs
University of Perugia, 06123 Perugia (Italy)
E-mail: oriana.tabarrini@unipg.it
Several experimental evidences have already validated CDK9
as a druggable component of the P-TEFb complex and con-
firmed that host and viral transcription might be differently
sensitive to P-TEFb inhibition,[6,9–15] thus providing a rationale
for targeting P-TEFb as a potential strategy for developing new
anti-HIV-1 therapeutics. The majority of CDK inhibitors were
firstly identified by retrospectively screening anticancer agents
towards a panel of CDKs. Among them, flavopiridol (1;
Figure 1) emerged as one of the most potent, but showed
only a 10-fold preference for CDK9[12,16] and a certain degree of
cytotoxicity.[17] Recently, few example of more potent and se-
[b] Dr. V. Attanasio, Dr. G. Corazza, Prof. A. Marcello
Molecular Virology Group, International Centre for Genetic Engineering and
Biotechnology (ICGEB), 34012 Trieste (Italy)
E-mail: marcello@icgeb.org
[c] Dr. N. R. Avanzi
Nerviano Medical Sciences, 20014 Nerviano (Italy)
[d] Prof. C. Pannecouque
Rega Institute for Medical Research
Katholieke Universiteit Leuven, 3000 Leuven (Belgium)
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300287.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1941

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
lective CDK9 inhibitors were re-
fragment scaffolds were visually inspected, taking into account,
beside the docking scores and the desired pharmacophore fea-
tures, their synthetic accessibility. Quinazoline fragment 2 satis-
fied the wanted requirements (Figure 2; XP GScore=À8.3207,
Rank=8) and was then selected for the enzymatic assay. Frag-
ment 2 in its Glide-predicted binding pose within the ATP
binding site overlapped to the experimental position of the
4H-chromen-4-one scaffold of 1 (Figure 3), making hydropho-
bic interaction mainly with residues Val33, Ala46, Val79,
Phe103, Leu156 and Ala166. Notably, 2 was able to establish
two hydrogen bonds with the backbone of Cys106 by its
amide moiety, whereas the 4H-chromen-4-one moiety was en-
gaged in only one hydrogen bond between its carbonyl
oxygen and the backbone NH of Cys106.
ported for their anti-HIV-1 activi-
ty.[18–21]
To identify novel anti-CDK9 che-
motypes and as a part of our con-
tinuous search for Tat-mediated
transcription inhibitors,[22,23] we
have implemented
a structure-
Figure 1. Flavopiridol (1).
based drug discovery (SBDD) strat-
egy exploiting the structural infor-
mation of P-TEFb in complex with
1.[24] We herein report the computational studies that, along
with a first round of structural optimization (Figure 2), led to
the identification of the 2-phenylquinazolinone scaffold for the
development of CDK9 inhibitors as promising anti-HIV-
1 agents.
The anti-CDK9 kinase activity evaluation showed that 2 was
able to inhibit CDK9 activity to about 50% albeit at a high con-
centration (500 mm, see figure S1
in the Supporting Information).
With the aim to build an ad hoc
fragment-based
library,
the
active fragment 2 was chosen as
scaffold. In particular, it was used
as a query in a substructure
search performed on the mole-
cules website,[26] looking for
commercially available mole-
cules based on such a scaffold.
This search resulted in a 13102-
membered library that was
screened in silico along with the
Gold (205659 compounds) and
Platinum (118763 compounds)
collections from the vendor
Asinex[27] and our in-house data-
base (about 1000 compounds).
The compound collections were
Figure 2. Drug discovery workflow.
screened in silico using as target
the crystallographic structure of
Results and Discussion
human CDK9/CycT1 in complex with 1[24] by means of the
Glide software, and employing Glide high-throughput virtual
screening (HTVS), standard precision (SP) and extra precision
(XP) scoring functions in three serial steps to increase the
screening speed. Fourteen molecules were selected as “virtual
hits” (see table S1) and subsequently submitted to the anti-
CDK9 assay at 50 mm concentration. Four derivatives showed
the ability to inhibit the kinase with an inhibition percentage
ꢀ 50% (see table S1). Among them, 2-phenylquinazolinones 3
and 4 (Figure 2), mined from the fragment-based library, were
selected for further studies. Their successive anti-kinase evalua-
tion, carried out at lower concentrations, showed an IC50 value
of 51.5 and 50.7 mm for derivatives 3 and 4, respectively (see
figures S2–S4). These compounds were also tested for their cy-
totoxicity in HeLa cells, where they were nontoxic up to
300 mm.
Hit generation
First of all, test docking calculations using flavopiridol (1) were
carried out to validate the docking protocol. The ligand was
thus extracted from the corresponding P-TEFb complex and
then docked back into the active site of the enzyme crystal
structure, with the aim of comparing experimental and predict-
ed binding modes and evaluate the program performance.
The best docking pose of 1 (selected on the basis of the high-
est GlideScore: À13.17) agreed well with the experimental
binding mode of 1, with a root-mean square deviation (RMSD)
value of 0.6747. To increase the probability of finding syntheti-
cally feasible inhibitors, we started our SBDD building an ad
hoc fragment-based library to be screened in parallel with two
commercial libraries and our in house database. To this end,
a first docking-based virtual screening was carried out on the
rule-of-three-compliant[25] 441-membered Glide fragment li-
brary (Schrçdinger, MW range 32–226 Da). The best scoring
The tenfold increased anti-CDK9 activity with respect to frag-
ment 2, coupled with the lack of cytotoxicity, reinforced the
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1942

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
then removed (13) as well as replaced with a chlorine atom
(14). The o-chlorine and p-amino derivatives 15 and 16 were
also prepared as suggested by free energy perturbations
(FEP)[28] calculations (see table S2). In a more incisive structural
modification, an imidazole ring replaced the 2-phenyl ring in
compound 17.
In a parallel approach, in silico combinatorial experiments
were carried out to thoroughly explore the binding pocket
area around the 2-phenyl moiety. According to the synthetic
pathway used to obtain the quinazolinone derivatives
(Scheme 1), a combinatorial library was built using the Combi-
Glide software.[29] In particular, a substructure search was per-
formed on the entire Sigma–Aldrich database, looking for all
the commercially available benzaldehydes. All the mined mole-
cules were then reacted in silico with isatoic anhydride, thus
enumerating a combinatorial library (324 compounds) that was
screened by automated molecular docking. The docking poses
were then visually inspected taking into account, besides the
docking score and the required pharmacophore features, the
costs of the reactants. This approach resulted in the identifica-
tion of potentially active, low cost, and easily synthesizable
compounds. Among them, quinazolinones 18–26 were select-
ed to be synthesized and tested (Table 1, see docking scores
and ranking in table S3).
The anti-CDK9 activity of compounds 5–26 (Table 1) inspired
the design of further analogues (27–41). In particular, starting
from 16, the p-amino group was moved to the meta and ortho
positions giving derivatives 27 and 28, respectively; the corre-
sponding nitro intermediates 29–31 were also tested. The p-
amino group was decorated with alkyl and cycloalkyl substitu-
ents as in compounds 32–34.
A few modifications involved the quinazolinone nucleus. In
particular, 7-chloro derivatives 35–38 were synthesized follow-
ing the predictions of FEP simulations (see table S2), whereas
N1ÀC2 saturated quinazolinones 39–41 were included in the
kinase assay to acquire preliminary structural information
about the importance of the pyrimidine ring.
Figure 3. A) Predicted binding mode of fragment 2 (cyan sticks) into the ATP
binding site (green sticks and cartoons). Experimental position of 1 is depict-
ed in magenta sticks as reference. Intermolecular hydrogen bonds are
shown as yellow dashed lines. B) Schematic representation of the interac-
tions between 2 and CDK9. CDK9 residues lying within a distance of 4 ꢁ
from the docked ligand are shown and color coded as follows: green, hydro-
phobic; red, acidic; purple, basic.
validity of the quinazolinone scaffold as a new chemotype
worthy of further optimization.
Chemistry
Various procedures for the synthesis of the quinazolinone scaf-
fold are known.[30] Most of the quinazolinones herein reported,
5,[31] 6,[32] 8,[33] 9–12, 13,[34] 14,[34] 15,[35] 16,[36] 17–19, 20,[37]
21,[38] 22, 23, 32,[35] 33,[39] 34,[38] 35,[40] and 36, were synthesized
according to the one-pot multicomponent reaction proposed
by Bhat and colleagues in 2004,[34] as outlined in Scheme 1.
Hit optimization
With the aim to improve potency and to collect preliminary
structure–activity relationship (SAR) information, a two-step
optimization study was performed, synthesizing an enlarged
series of quinazolinones (5–41, Table 1) by combining different
medicinal chemistry approaches.
Starting from hit compounds 3 and 4, the main ef-
forts were initially focused on the phenyl ring at the
C-2 position of the quinazolin-4(3H)-one scaffold. In
particular, the p-hydroxy and p-methoxy groups were
shifted to ortho (5 and 6) and meta (7 and 8) posi-
tions, respectively, whereas a series of different p-O-
alkylated compounds were realized replacing the
Scheme 1. Synthesis of quinazolinone derivatives 3–6, 8–25 and 32–36. For R1 and Ar,
see Table 1. Reagents and conditions: a) NH4OAc, I2, KOH, DMF, 858C; b) NH4OAc, Yb(OTf)3,
methyl with linear (9 and 10), as well as branched
(11 and 12) alkyl chains. The para substituent was DMSO, 908C, 4 h, 40–61%.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1943

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
The same procedure was also adopted to resynthesize
compounds 3 and 4.[34]
Table 1. Structure of quinazolinones synthesized in this study and CDK9 inhibitory ac-
tivity.[a]
The reaction of the appropriate isatoic anhydride,
selected aryl aldehydes and ammonium acetate in the
presence of iodine and potassium hydroxide in dry
N,N-dimethylformamide (DMF) directly gave the
target compounds. Among the employed aldehydes,
4-ethoxy-,[41] 4-propoxy-,[41] 4-isopropoxy-,[42] and 4-iso-
pentoxy- (42) benzaldehydes were prepared accord-
ing to the literature through the alkylation of 4-hy-
droxybenzaldehyde with the appropriate alkyl iodide
in the presence of potassium carbonate, in acetoni-
trile at reflux (not shown). When the 2-chlorobenzal-
dehyde was employed, besides the target compound
15 (47% yield), the N1ÀC2 dihydro intermediate 39[35]
was also obtained. The sole N1ÀC2 dihydro derivative
40 was instead obtained when 4-[(2-hydroxyethyl)-
(methyl)amino]benzaldehyde was used. Compound
7[33] was obtained by the demethylation of its me-
thoxy analogue 8 using 48% hydrogen bromide (not
shown).
Compd
Ar
Inhibition Compd
[%][a]
Ar
Inhibition
[%][a]
3
62
55
19
20
26
43
4
5
6
38
54
21
22
8
0
For the preparation of 24 and 25, the above de-
scribed procedure was slightly modified by using yt-
terbium trifluoromethansulfonate (Yb(OTf)3) as pro-
moting agent and replacing DMF with dimethyl sulf-
oxide (DMSO), which also works as mild oxidizing
agent.[43] By synthesizing compound 25, N1ÀC2 dihy-
dro derivative 41 was also obtained.
7
8
31
24
23
24
73
73
In an alternative procedure (Scheme 2), the anthra-
nilamide replaced the isatoic anhydride as starting
material. In particular, the reaction of appropriate an-
thranilamide with nitrobenzoyl chloride in the pres-
ence of triethylamine gave intermediates 43,[44] 44,
45,[45] and 46,[46] which were cyclized using potassium
tert-butoxide in tert-butanol giving nitro derivatives
29–31[46]and 38.[47] The amino derivatives 27,[46] 28,[46]
and 37 were obtained by catalytic reduction of corre-
sponding nitro intermediates 30, 31, and 38. Deriva-
tive 26 was directly prepared by reacting anthranila-
mide with trihydroxybenzaldehyde using p-toluene-
sulfonic acid (pTsOH) as catalyst under reflux of meth-
anol.
9
43
54
32
25
26
27
17
70
52
10
11
12
13
55
55
28
29
25
0
Biological results
All synthesized compounds were initially tested for
their ability to inhibit CDK9 activity at a concentration
of 50 mm. The resynthesized compounds 3 and 4, in-
cluded in the biological evaluation, showed the same
activity of the commercial counterparts. As shown in
Table 1, most of the compounds were able to inhibit
the enzyme although with different degrees of inhibi-
tion.
14
2
30
10
15
16
42
70
31
32
0
41
Although a robust SAR cannot be delineated on
the basis of the percentage of inhibition, some clues
may emerge by comparing these data with those of
the hit compounds 3 and 4. The substituent at the
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1944

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
para position of the 2-phenyl ring did not seem to
be essential, since compound 13 showed similar ac-
tivity as compared to the hit compounds. However,
when present, the amino group was beneficial (16),
in line with the FEP predictions made for the para
position (see table S2). The presence of the p-chlorine
atom was detrimental (14, 2% inhibitory activity), but
its FEP-guided (see table S2) shifting to the ortho po-
sition permitted some activity to be recovered (15,
42% inhibitory activity). The addition of a further aro-
matic ring (19) as well as the fusion of the phenyl
moiety between its ortho and meta position with an
additional aromatic ring (22 and 25) were clearly un-
fruitful. The lengthening of the methoxy group char-
acterizing hit 4 (as in 9–12) was tolerated with the
only exception of 11 (32% of inhibitory activity). The
nitro group was absolutely not allowed in any posi-
tion of the phenyl ring (see data for 29–31, and 38).
Concerning the hydroxy and methoxy regioisomers
of 3 and 4 (5–8), only o-methoxy derivative 6 main-
tained the same activity as the hit compounds. Con-
trasting results were obtained when assaying dihy-
droquinazolinone derivatives 39–41, with compound
39 lacking any ability to inhibit CDK9. Analogously,
the addition of a chlorine atom at the C-7 position of
active quinazolinones 4, 16 and 24 gave very differ-
ent results; high inhibition was maintained by 37, the
7-chloro analogue of amino derivative 16.
Table 1. (Continued)
Compd
Ar
Inhibition Compd
[%][a]
Ar
Inhibition
[%][a]
17
0
33
34
10
29
18
16
35
36
52
4
37
38
70
0
In summary, the assessment of the anti-CDK9 activ-
ity of the whole set of synthesized compounds result-
ed in the identification of new active derivatives, con-
firming the 2-phenylquinazolinone as a suitable scaf-
fold for CDK9 inhibitors. Setting a 70% inhibition of
kinase activity as hit selection criterion, compounds
16, 23, 24, 26, and 37, emerged as the most promis-
ing. With the exception of 26, all of the compounds
are structurally characterized by a nitrogen atom at
the para position of the 2-phenyl ring, as a primary
amino group (16 and 37), N,N-
39
40
1
41
41
57
[a] CDK9 kinase activity was measured in vitro on a recombinant GST-CTD substrate
as described previously,[74,75] using 50 mm concentration of the compounds.
dialkylated (24), or embedded in
a pyrazole ring (23).
For these derivatives, an in-
depth biological evaluation was
carried out to determine IC50
values, cytotoxicity, and the abili-
ty to inhibit Tat-mediated tran-
scription, using 1 as reference
compound.
The anti-kinase activity evalu-
ated in serial dilutions showed
that all five selected compounds
are able to fully inhibit CDK9 in
a dose-dependent fashion with
IC50 values in the low micromolar
range (Table 2): 9.91 mm (23),
10.12 mm (24), 12.18 mm (16),
15.60 mm (26), and 26.20 mm
Scheme 2. Synthesis of quinazolinone derivatives 26–31, 37, 38 and 43–46. Reagents and conditions: a) Et3N, THF,
RT, 6 h, 71%; b) tert-butanol, potassium tert-butoxide, RT, 24 h, 70%; c) H2, Raney Ni, DMF, RT, 2 h, 60%; d) pTsOH,
2,3,4-trihydroxybenzaldehyde, MeOH, reflux, 5 h, 43%.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1945

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
scription was observed only at high concen-
trations, compounds 16, 23, 24, 26, and 37,
were assayed for their anti-HIV-1 activity on
both acutely and latently infected cells
(Table 2), none of the tested compounds
showed antiviral activity in acutely infected
MT-4 cells, where no relevant cytotoxicity was
observed with the exception of compound
26, which displayed some cytotoxicity (CC50 =
9.2 mm) in agreement to what was observed
in HeLa cells. Conversely, in a model of latent-
ly HIV-1-infected promyelocytic cells (OM-
10.1), where the virus is transcriptionally silent
unless stimulated with phorbol myristate ace-
tate (PMA), compounds 16, 24 and 37
showed good antiviral activities with IC50
values in the low micromolar range at concen-
trations that were far below the observed cy-
totoxicity. The best inhibitory activity was ob-
Table 2. Anti-CDK9 activity and anti-HIV-1 activity on acutely (MT-4) and latently (OM-10.1) in-
fected cells of selected 2-phenylquinazolinones.
CDK9
MT4 cells
OM-10.1 cells
CC50 [mm][c,d] SI[e]
Compd IC50 [mm][a]
IC50 [mm][b,d] CC50 [mm][c,d]
SI[e] IC50 [mm][d,f]
<1
11.8Æ1.7
16
23
24
26
37
1
12.18 (0.67) >238.1
238.1Æ9.7
338.8Æ46.4 29
9.91 (0.18)
10.12 (0.36)
15.60 (0.04)
26.20 (0.13) >170
0.015 (0.18) >0.27
>42.6
>42.2
>9.2
42.6Æ4.2
42.2Æ2.2
9.2Æ1.5
<1 ꢀ19.1
76.1Æ19.8 ꢂ4
<1
<1 >51.0
14.0Æ3.4
154.4Æ9.9
51.0Æ6.8
11
<1
86
>170
<1
4.0Æ0.7
0.003Æ0.0007
344.9Æ47.4
0.27Æ0.02 <1
0.18Æ0.01 60
[a] The CDK9 kinase activity was measured in vitro on recombinant GST-CTD substrate as de-
scribed previously;[73,74] standard errors are shown in brackets. [b] IC50: concentration of com-
pound required to achieve 50% protection of MT-4 cells from HIV-1 induced cytopathogenici-
ty, as determined by the MTT method. [c] CC50: concentration of compound that reduces the
viability of mock-infected cells by 50%, as determined by the MTT method. [d] All data repre-
sent mean values Æ standard deviations for at least two separate experiments. [e] Selectivity
index (SI): ratio of CC50/IC50. [f] IC50: concentration of compound required to achieve 50% re-
duction of p24 production in HIV-1 infected cells.
(37). Thus, compared to those of the hit compounds 3 and 4
(IC50 ꢁ50 mm), the inhibitory activity of the new derivatives was
improved two- to fivefold, and notably, the ligand ef-
ficacy (LE) was fully retained and in some cases
slightly increased (4: LE=0.31; 16: LE=0.37). Howev-
er, the compounds were less active than 1, which, as-
sayed in parallel, inhibited CDK9 with an IC50 value of
0.015 mm.
tained by 37 with an IC50 value of 4.0 mm, which, coupled with
a CC50 value of 345 mm, led to a selectivity index (SI) of 86.
At this point, the cytotoxicity was evaluated in
HeLa cells, and only derivative 26 was toxic
>100 mm, whereas for all the other derivatives no ap-
parent cytotoxicity was observed even at the highest
assayed concentration of 500 mm. It is worth noting
that 1 showed a 50% reduction of cell viability at
1 mm (see figure S5).
The five selected quinazolinones were then tested
for their ability to inhibit Tat transactivation in order
to evaluate whether the in vitro anti-CDK9 activity is
preserved and still holds in a whole cell context. The
Tat-mediated transcription assay was performed on
HeLa cells carrying an LTR-luciferase reporter and
using 1 as the positive control. With the only excep-
tion of derivative 16, all of the tested compounds in-
hibited the transactivation at 50 mm with 37 being
the most active, also inhibiting the transactivation at
10 mm (Figure 4).
Figure 4. HIV-1 LTR transactivation assay. HeLa LTR-luciferase were activated with Tat at
suboptimal doses and treated with increasing concentrations of 37 as described previ-
ously.[73] The Firefly luciferase assay was performed on cell lysates 24 h after treatment
following standard procedures (Promega) and analyzed with a plate reader (Wallac, Perki-
nElmer). DMSO, vehicle control; 1, flavopiridol. Luminescence intensities represented as
percentages.
A constitutive transcription assay was then per-
formed in order to assess whether the anti-transacti-
vation activity could be ascribed to a genuine inhibi-
tion of P-TEFb. To this end, HeLa cells were transfect-
ed with a plasmid where expression of the reporter
Renilla gene is driven by the constitutive CMV pro-
moter. No significant activity on the basal transcrip-
tion was displayed by incubating the cells with our
best inhibitor 37, suggesting a promising target se-
lectivity (Figure 5).
Figure 5. Constitutive transcription assay. HeLa cells were transfected with plasmid
pCMV-Renilla and treated with increasing concentrations of compound 37. The Renilla lu-
ciferase assay was performed as described in Figure 4. DMSO, vehicle control. Lumines-
Finally, although the anti-CDK9 activity was not
very high and the inhibition of Tat-mediated tran- cence intensities represented as percentages.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1946

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Compound 1, assayed in parallel, did not show anti-
HIV activity in MT-4 cells at concentrations lower than
its CC50 value, whereas a marked activity was mea-
sured in OM-10.1 cells, where however the stronger
cytotoxicity led to a SI value of 60, lower than that of
37.
Intriguingly, the antiviral activity observed in la-
tently infected cells with compound 37 was higher
than what observed in the kinase assay and in the
LTR-luciferase assay, suggesting that an additive
mechanism might contribute to the antiviral effect.
Therefore, we investigated other kinases that are in-
volved in the Tat-mediated transcription process and
that share structural features with CDK9. We focused
our attention on CDK2, which is emerging as a valid
target[48] being not essential for mammalian develop-
ment[49,50] but required for Tat-dependent transcrip-
tion.[51,52] Specifically, CDK2 in complex with cyclin E
phosphorylates the RNAPII CTD once recruited to the
promoter by Tat, which is itself a target of CDK2-
mediated phosphorylation. CDK2 also phosphorylates
CDK9 on Ser90, as recently reported.[53] Indeed, when
compound 37 was tested on CDK2, an inhibition
comparable to that obtained on CDK9 was observed
with and IC50 value of 3.39 mm.
Figure 6. Predicted binding mode of 37 (yellow sticks) into the ATP pocket of CDK2
(PDBID: 3TNW; magenta sticks and cartoon) and CDK9 (PDBID: 3TNH; green sticks and
cartoon). Hydrogen bonds are shown as dashed grey lines. The loops comprising CDK9
residues 25–32 (10–17 on CDK2) are represented as cartoon. This Figure was prepared
using PyMOL.[80]
In its predicted bound conformation, 37 does not overlap
An insight into the double activity of 37 can be achieved
looking at the predicted binding mode into both enzymes
along with the comparison to several published crystallograph-
ic structures and literature data showing the chemical features
needed to obtain selectivity.[24,54–59] Compound 37, in its dock-
ing-predicted pose, binds to the ATP pocket engaging in hy-
drogen-bond interactions with the backbone of CDK9 Cys106/
CDK2 Leu83 and making contacts with several residues con-
served among the two enzymes. Although the two binding
sites share a high number of structural features, some peculiar
differences have been shown to be exploitable to impart selec-
tivity to ligands towards one of the two enzymes.[54,56,57,59] The
region around CDK2 Lys88 and Lys89 (cyan sphere) is more
closed in CDK2, since the two residues are bulkier than the cor-
responding CDK9 Ala111 and Gly112 (Figure 6). Indeed, rigid
and bulky substituents in the zone highlighted by the cyan
sphere would hamper ligand binding to CDK2 because of
steric clash with the Lys residue, whereas flexible substituents
capable of accepting a hydrogen bond from Lys89 would
probably result in a more efficient binding to CDK2. The loop
comprising CDK9 residues 25–32 (10–17 on CDK2) has been
observed to rearrange, moving towards the ATP binding site
but is next to the above described exploitable features that
could be used to optimize activity toward CDK9 or CDK2 or
design a single molecule endowed with both activities, thus
making our low-weight compounds highly optimizable in both
potency and selectivity.
Conclusions
Tat-mediated transactivation is an attractive target for anti-HIV-
1 therapy because it controls a key step during reactivation
from latency. The kinase subunit of the positive transcription
elongation factor b (P-TEFb) complex that is recruited by Tat to
the viral promoter is a potential target. In this work, we took
advantage of the crystallographic data of cyclin-dependent
kinase 9 (CDK9) in complex with 1 and used it for the design
and synthesis of novel inhibitors. We have identified quinazoli-
none fragment 2 endowed with a residual anti-CDK9 activity at
500 mm that was improved by just adding a phenyl substituent
at the C-2 position as in derivatives 3 and 4. These compounds
were then elaborated by a combination of different medicinal
chemistry approaches, leading to analogues 16, 23, 24, 26,
and 37, which inhibited CDK9 with IC50 values in the micromo-
lar range. With the exception of compound 16, these 2-phenyl-
quinazolinones were able to inhibit Tat-mediated transactiva-
tion in cell lines, with 7-chloro derivative 37 being the most
potent. Notably, the inhibitory activity was specific to Tat-medi-
ated transactivation. More interestingly, the inhibition of Tat-
mediated transactivation translated into the ability to inhibit
HIV-1 reactivation from latently infected cells, with compound
37 showing a potency that was superior to what could be pre-
dicted looking at its CDK9 activity, thus leading to the hypoth-
esis of an additional target. Indeed, CDK2, which plays a sup-
upon binding of several diverse competitive inhibitors on
[24,54]
CDK9
and thus offering the possibility to take advantage
of this structural difference to obtain CDK9 selectivity by using
substituents capable of engaging interaction with the loop res-
idues (green sphere, Figure 6). Another structural difference
between the two enzymes was observed in the region marked
by the pink sphere, where CDK9 Glu66 can move away from
the ATP binding site, thus making the binding pocket next to
Phe168 larger than the same pocket in CDK2.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1947

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
2-(4-Ethoxyphenyl)quinazolin-4(3H)-one (10): The synthetic pro-
cedure for 9 was used, replacing 4-propoxybenzaldehyde with 4-
ethoxybenzaldehyde, to give 10 as white crystals (0.50 g; 55%):
mp: 254–2558C; 1H NMR (400 MHz, [D6]DMSO): d=12.50 (bs, 1H,
CONH), 8.10–8.20 (m, 3H, H-5, H-3’, and H-5’), 7.60–7.80 (m, 2H, H-
6 and H-8), 7.45 (dt, J=7.0 and 1.0 Hz, 1H, H-7), 7.10 (d, J=8.0 Hz,
2H, H-2’ and H-6’), 4.10 (q, J=7.5 Hz, 2H, OCH2), 1.40 ppm (t, J=
7.5 Hz, 3H, OCH2CH3); 13C NMR (100 MHz, [D6]DMSO): d=11.2, 65.6,
115.0, 121.1, 125.3, 126.7, 126.9, 127.8, 130.1, 134.6, 149.8, 152.8,
161.5, 162.0 ppm; Anal. calcd for C16H14N2O2: C 72.16, H 5.30, N
10.52, found: C 72.00, H 5.20, N 10.60.
portive role within the transcriptional machinery and shares
structural features with CDK9, was also inhibited by compound
37 at comparable concentrations.
The appeal of the reported 2-phenylquinazolinones are
strengthened by the lack of toxicity in several cell lines, which,
coupled with the inability to influence the basal transcription,
suggests a selective action towards the Tat-hijacked CDKs. In
addition, as these compounds merely represent fragments,[60]
there is plenty of room for further optimization, which will of
course require an overall kinase inhibition profiling, once more
potent compounds are obtained.
2-(4-Isopropoxyphenyl)quinazolin-4(3H)-one (11): The synthetic
procedure for 9 was used, replacing 4-propoxybenzaldehyde with
4-isopropoxybenzaldehyde, to give 11 as white crystals (0.55 g;
1
58%): mp: 236–2388C; H NMR (400 MHz, [D6]DMSO): d=12.50 (bs,
1H, CONH), 8.10–8.20 (m, 3H, H-5, H-3’, and H-5’), 7.70–7.80 (m,
2H, H-6 and H-8), 7.45 (dt, J=7.0 and 1.0 Hz, 1H, H-7), 7.10 (d, J=
8.0 Hz, 2H, H-2’ and H-6’), 4.55–4.65 (m, 1H, OCH), 1.20 ppm (d, J=
6.0 Hz, 6H, CH(CH3)2); 13C NMR (100 MHz, [D6]DMSO): d=22.0, 66.8,
114.9, 121.3, 124.8, 125.2, 126.8, 126.9, 127.3, 129.9, 131.0, 134.6,
149.8, 152.1, 161.5, 162.0 ppm; Anal. calcd for C17H16N2O2: C 72.84,
H 5.75, N 9.99, found: C 73.10, H 5.75 N 10.10.
Experimental Section
Chemistry
All reactions were routinely checked by thin-layer chromatography
(TLC) on silica gel 60 F254 (Merck) and visualized using UV light or
iodine. Flash column chromatography separations were carried out
on silica gel 60 (mesh 230–400, Merck). Melting points (mp) were
determined in capillary tubes using an Electrothermal 9100 (Bꢂchi,
Essen, Germany) and are uncorrected. Elemental analyses were per-
formed on a EA1108CHN elemental analyzer (Fisons, Ipswich, UK),
and the data for C, H and N are within Æ0.4% of the theoretical
values. 1H NMR and 13C NMR spectra were recorded at 200 MHz
(Bruker Avance DPX-200) and 400 MHz (Bruker Avance DRX-400)
using residual solvents such as CHCl3 (d=7.26 ppm) or dimethyl
sulfoxide (DMSO, d=2.48 ppm) as internal standard. Chemical
shifts (d) are given in ppm, and the spectral data are consistent
with the assigned structures. Reagents and solvents were pur-
chased from common commercial suppliers and were used as
such. After extraction, organic solutions were dried over anhydrous
Na2SO4, filtered, and concentrated with a Bꢂchi rotary evaporator
at reduced pressure. Yields are of purified product and were not
optimized. All starting materials were commercially available unless
otherwise indicated.
4-(3-Methylbutoxy)benzaldehyde (42): 1-Iodo-3-methylbutane
(0.81 mL, 6.15 mmol) was added to a suspension of 4-hydroxyben-
zaldehyde (0.50 g, 4.10 mmol) and K2CO3 (0.85 g, 6.15 mmol) in
CH3CN. The suspension was allowed to stir at reflux for 5 h. After
cooling to RT, the suspension was filtered, and the solution was
evaporated in vacuo to give 42 as a yellow oil (0.78 g; 99%):
1H NMR (400 MHz, [D6]DMSO): d=9.95 (s, 1H, COH), 7.80–7.85 (m,
2H, H-3’, and H-5’), 7.10–7.15 (m, 2H, H-2’ and H-6’), 4.05 (t, J=
6.0 Hz, 2H, OCH2), 1.50–1.80 (m, 3H, CHCH2), 0.95 ppm (d, J=
7.0 Hz, 6H, CH(CH3)2).
2-[4-(3-Methylbutoxy)phenyl]quinazolin-4(3H)-one (12): The syn-
thetic procedure for 9 was used, replacing 4-propoxybenzaldehyde
with 42, to give 12 as a white powder (0.49 g; 47%): mp: 181–
1848C; 1H NMR (400 MHz, [D6]DMSO): d=12.50 (bs, 1H, CONH),
8.05–8.15 (m, 3H, H-5, H-3’, and H-5’), 7.60–7.80 (m, 2H, H-6 and H-
8), 7.45 (dt, J=7.0 and 1.0 Hz, 1H, H-7), 7.10 (d, J=8.0 Hz, 2H, H-2’
and H-6’), 4.10 (t, J=6.0 Hz, 2H, OCH2), 1.60–1.80 (m, 3H, CH2CH),
0.9 ppm (d, J=6.0 Hz, 6H, CH(CH3)2); 13C NMR (100 MHz,
[D6]DMSO): d=22.83, 25.03, 37.76, 66.67, 114.83, 121.12, 125.06,
126.25, 126.48, 127.70, 129.87, 134.91, 149.39, 152.30, 161.75,
162.74 ppm; Anal. calcd for C19H20N2O2: C 74.00, H 6.54, N 9.08,
found: C 74.25, H 6.54, N 9.20.
Compound 2 was purchased from Sigma–Aldrich (code H57807),
compound 3 was purchased from Enamine (code T6147622; NJ,
USA) and 4 was purchased from Princeton Biomolecular Research
(code OSSK540864; NJ, USA). Their identity was independently as-
sessed by 1H NMR (see Supporting Information).
2-(4-Propoxyphenyl)quinazolin-4(3H)-one (9): NH4OAc (0.3 g,
4.12 mmol), 4-propoxybenzaldehyde (0.6 g, 4.12 mmol), I2 (2.7 g,
10.97 mmol), and powdered KOH (0.2 g, 4.12 mmol) were added to
a solution of isatoic anhydride (0.5 g, 3.4 mmol) in N,N-dimethylfor-
mamide (DMF; 10 mL), and the solution was stirred at 858C for
4 h. After cooling, the reaction mixture was poured into ice water
and extracted with EtOAc. The organic layers were evaporated to
dryness to give a residue that was purified by flash chromatogra-
phy eluting with MeOH/CHCl3 (5% v/v), and then crystallized from
EtOH to give 9 as white crystals (0.57 g, 60%): mp: 212–2158C;
1H NMR (400 MHz, [D6]DMSO): d=12.50 (bs, 1H, CONH), 8.15–8.25
(m, 3H, H-5, H-3’, and H-5’), 7.80–7.85 (dt, J=7.0 and 2.0 Hz, 1H, H-
6), 7.35–7.45 (dt, J=8.0 Hz and 1.0 Hz, 1H, H-7), 7.65 (d, J=8.0 Hz,
1H, H-8), 7.00 (d, J=8.0 Hz, 2H, H-2’ and H-6’), 4.00 (t, J=7.0 Hz,
2H, OCH2), 1.65–1.75 (m, 2H, OCH2CH2), 1.00 ppm (t, J=7.0 Hz, 3H,
CH2CH3); 13C NMR (100 MHz, [D6]DMSO): d=10.7, 22.4, 69.6, 114.8,
121.1, 125.0, 126.2, 126.5, 127.7, 130.0, 134.9, 149.3, 152.3, 161.4,
162.7 ppm; Anal. calcd for C17H16N2O2: C 78.84, H 5.75, N 9.99,
found: C 78.99, H 5.85, N 9.95.
2-(4-Aminophenyl)quinazolin-4(3H)-one (16):[36] The synthetic
procedure for 9 was used, replacing 4-propoxybenzaldehyde with
4-aminobenzaldehyde, to give 16 as a light-yellow powder (0.47 g;
1
58%): mp: 280–2828C; H NMR (400 MHz, [D6]DMSO): d=12.10 (bs,
1H, CONH), 8.05–8.15 (m, 1H, H-5), 7.90 (d, J=8.0 Hz, 2H, H-2’ and
H-6’), 7.75 (t, J=7.0 Hz, 1H, H-7), 7.65 (d, J=8.0 Hz, 1H, H-8), 7.40
(t, J=7.0 Hz, 1H, H-6), 6.60 (d, J=8.0 Hz, 2H, H-3’ and H-5’),
5.85 ppm (bs, 2H, NH2); 13C NMR (100 MHz, [D6]DMSO): d=114.2,
121.6, 125.7, 126.2, 126.4, 127.6, 130.1, 135.1, 149.6, 152.0, 156.4,
162.0 ppm; Anal. calcd for C14H11N3O: C 70.87, H 4.67, N 17.71,
found: C 71.00, H 4.80, N 17.67.
2-(1H-Imidazol-5-yl)quinazolin-4(3H)-one (17): The synthetic pro-
cedure for 9 was used, replacing 4-propoxybenzaldehyde with 1H-
imidazole-5-carbaldehyde, to give 17 as a white powder (0.28 g;
1
40%): mp: 256–2588C; H NMR (400 MHz, [D6]DMSO): d=12.00 (bs,
1H, CONH), 10.10 (bs, 1H, NH), 8.20 (d, J=8 Hz, 1H, H-5), 8.00 (s,
1H, H-5’), 7.85 (s, 1H, H-3’), 7.75–7.80 (m, 1H, H7), 7.65 (d, J=8 Hz,
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1948

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
1H, H-8), 7.45 ppm (t, J=8.0 Hz, 1H, H-6); 13C NMR (100 MHz,
[D6]DMSO): d=105.9, 123.2, 124.2, 126.3, 129.4, 131.9, 138.6, 142.9,
148.0, 163.1 ppm; Anal. calcd for C11H8N4O: C 62.26, H 3.80, N
26.40, found: C 62.30, H 3.92, N 26.43.
J=7.0 Hz, 0.9H, CH3); 13C NMR (100 MHz, [D6]DMSO): d=18.9, 56.4,
107.5, 121.3, 122.6, 122.7, 124.9, 125.4, 125.5, 126.2, 126.8, 127.6,
127.7, 129.7, 132.2, 134.8, 149.3, 154.3, 155.8, 162.5 ppm; Anal.
calcd for C18H12N2O2·1/3 C2H5OH): C 71.84, H 5.43, N 8.38, found: C
72.00, H 5.55, N 8.25.
2-(2-Chloro-3-hydroxy-4-methoxyphenyl)quinazolin-4(3H)-one
(18): The synthetic procedure for 9 was used, replacing 4-propoxy-
benzaldehyde with 2-chloro-3-hydroxy-4-methoxybenzaldehyde, to
2-(1H-Indazol-5-yl)quinazolin-4(3H)-one (23): The synthetic proce-
dure for 9 was used, replacing 4-propoxybenzaldehyde with 1H-in-
dazole-5-carbaldehyde, to give 23 as a white powder (0.37 g;
1
give 18 as colorless crystals (0.60 g; 53%): mp: 276–2798C; H NMR
1
(400 MHz, [D6]DMSO): d=12.50 (bs, 1H, CONH), 9.65 (bs, 1H, OH),
8.25 (dd, J=1.0 and 8.0 Hz, 1H, H-5), 7.80 (dt, J=7.0 and 2.0 Hz,
1H, H-6), 7.70 (d, J=8.0 Hz, 1H, H-8), 7.55 (t, J=8.0 Hz, 1H, H-7),
6.95–7.05 (m, 2H, H-5’ and H-6’), 3.90 ppm (s, 3H, OCH3); 13C NMR
(100 MHz, [D6]DMSO): d=56.6, 112.7, 114.8, 121.5, 125.1, 126.2,
127.4, 127.9, 132.9, 135.0, 149.2, 151.3, 156.9, 161.7 ppm; Anal.
calcd for C15H11ClN2O3: C 59.52, H 3.66, N 9.25, found: C 59.20, H
3.47, N 9.41.
42%): mp: 340–3438C; H NMR (400 MHz, [D6]DMSO): d=13.35 (bs,
1H, NH), 12.50 (bs, 1H, CONH), 8.70 (s, 1H, H-3’), 8.20–8.25 (m, 2H,
H-2’, and H-7’), 8.15 (d, J=8.0 Hz, 1H, H-5), 7.85 (t, J=7.0 Hz, 1H,
H-7), 7.75 (d, J=8.0 Hz, 1H, H-8), 7.70 (d, J=9.0 Hz, 1H, H-6’),
7.50 ppm (t, J=7 Hz, 1H, H-6); 13C NMR (100 MHz, [D6]DMSO): d=
108.1, 116.3, 120.9, 121.8, 124.5, 125.2, 127.3, 134.2, 141.8, 150.0,
157.8, 161.5 ppm; Anal. calcd for C15H10N4O: C 68.69, H 3.84, N
21.36, found: C 68.88, H 4.95, 21.20.
2-(4-Pyridin-2-ylphenyl)quinazolin-4(3H)-one (19): The synthetic
procedure for 9 was used, replacing 4-propoxybenzaldehyde with
4-pyridin-2-ylbenzaldehyde, to give 19 as a white powder (0.46 g;
2-{4-[(2-Hydroxyethyl)(methyl)amino]phenyl}quinazolin-4(3H)-
one (24): NH4OAc (0.46 g, 6.00 mmol), 4-[(2-hydroxyethyl)-
(methyl)amino]benzaldehyde (0.89 g, 5.0 mmol), and Yb(OTf)3
(0.03 g, 0.05 mmol) were added to a solution of isatoic anhydride
(0.89 g, 5.50 mmol) in DMSO (5 mL), and the mixture was main-
tained at 908C for 15 h. After cooling, the reaction mixture was
poured into ice water, obtaining a precipitate, which was filtered
and crystallized from EtOH, to give 24 as a light-yellow powder
1
46%): mp: 279–2828C; H NMR (400 MHz, [D6]DMSO): d=12.5 (bs,
1H, CONH), 8.70 (dd, J=8.0 and 1.0 Hz, 1H, pyridine CH), 8.30 (d,
J=8.0 Hz, 2H, H-2’ and H-6’), 8.25 (d, J=8.0 Hz, 2H, H-3’ and H-5’),
8.15–8.20 (m, 1H, pyridine CH), 8.10 (d, J=8 Hz, 1H, H-5), 7.90–
7.95 (m, 1H, pyridine CH), 7.85 (dt, J=7.0 and 2.0 Hz, 1H, H-6),
7.75 (d, J=8.0 Hz, 1H, H-8), 7.55 (dt, J=7.0 and 1.0 Hz, 1H, H-7),
7.35–7.40 ppm (m, 1H, pyridine CH); 13C NMR (100 MHz,
[D6]DMSO): d=121.2, 121.4, 123.7, 126.3, 127.0, 127.1, 127.9, 128.6,
133.4, 135.0, 141.5, 149.0. 150.1, 152.3, 155.2, 161.8 ppm; Anal.
calcd for C19H13N3O: C 76.24, H 4.38, N 14.04, found: C 76.33, H
4.37, N 14.00.
1
(0.09 g, 40%): mp: 220–2238C; H NMR (400 MHz, [D6]DMSO): d=
12.00 (bs, 1H, CONH), 7.95–8.05 (m, 3H, H-5, H-3’, and H-5’), 7.75 (t,
J=7.0 Hz, 1H, H-7), 7.65 (d, J=8.0 Hz, 1H, H-8), 7.45 (t, J=7.0 Hz,
1H, H-6), 6.75 (d, J=8.0 Hz, 2H, H-2’ and H-6’), 4.70–4.80 (m, 1H,
OH), 3.50–3.60 (m, 4H, NCH2CH2O), 3.00 ppm (s, 3H, NCH3);
13C NMR (100 MHz, [D6]DMSO): d=39.1, 54.2, 58.2, 111.3, 118.7,
120.8, 125.7, 126.2, 127.4, 129.3, 134.8, 149.8, 151.8, 152.6,
162.8 ppm; Anal. calcd for C17H17N3O2: C 69.14, H 5.80, N 14.23,
found: C 69.00, H 5.98, N 14.21.
2-(5-Bromo-2-methoxyphenyl)quinazolin-4(3H)-one (20):[37] The
synthetic procedure for 9 was used, replacing 4-propoxybenzalde-
hyde with 5-bromo-2-methoxybenzaldehyde, to give 20 as a white
powder (0.55 g; 49%): mp: 178–1828C; 1H NMR (400 MHz,
[D6]DMSO): d=12.20 (bs, 1H, CONH), 8.10 (d, J=8.0 Hz, 1H, H-5),
7.80–7.85 (m, 2H, H-7 and H-6’), 7.60–7.75 (m, 2H, H-8 and H-4’),
7.55 (t, J=7 Hz, 1H, H-6), 7.20 (d, J=9.0 Hz, 1H, H-3’), 3.80 ppm (s,
3H, OCH3); 13C NMR (100 MHz, [D6]DMSO): d=56.6, 112.7, 114.8,
121.5, 125.1, 126.2, 127.4, 127.9, 132.9, 135.0, 149.2, 151.3, 156.9,
161.7 ppm; Anal. calcd for C15H11BrN2O2: C 54.40, H 3.35, N 8.46,
found: C 54.70, H 3.21, N 8.60.
2-Isoquinolin-5-ylquinazolin-4(3H)-one (25) and 2-isoquinolin-5-
yl-2,3-dihydroquinazolin-4(1H)-one (41): The synthetic procedure
for
24
was
used,
replacing
4-[(2-hydroxyethyl)-
(methyl)amino]benzaldehyde with isoquinoline-5-carbaldehyde, to
give 25 and 41, respectively, after purification by flash chromatog-
raphy eluting with EtOH/CHCl3 (5% v/v).
25 (0.38 g; 41%): white powder; mp: 318–3208C; 1H NMR
(400 MHz, [D6]DMSO): d=12.75 (bs, 1H, CONH), 9.45 (s, 1H, H-5’),
8.60 (d, J=6.0 Hz, 1H, H-3’), 8.30 (d, J=8.0 Hz, 1H, H-5), 8.10–8.25
(m, 3H, H-8’, H-6’, and H-2’), 7.80–7.90 (m, 2H, H-7, and H-7’), 7.75
(d, J=8.0 Hz, 1H, H-8), 7.50 ppm (t, J=7.0 Hz, 1H, H-6); 13C NMR
(100 MHz, [D6]DMSO): d=118.5, 121.7, 126.3, 127.1, 127.4, 127.9,
128.6, 130.7, 130.8, 132.5, 133.2, 135.0, 144.3, 149.0, 152.8. 153.3,
162.4 ppm; Anal. calcd for C17H11N3O: C 74.71, H 4.06, N 15.38,
found: C 74.78, H 4.19, N 15.25.
2-(4-Piperidin-1-ylphenyl)quinazolin-4(3H)-one (21):[38] The syn-
thetic procedure for 9 was used, replacing 4-propoxybenzaldehyde
with 4-piperidin-1-ylbenzaldehyde, to give 21 as a grey powder
1
(0.63 g; 61%): mp: 274–2778C; H NMR (400 MHz, [D6]DMSO): d=
12.20 (bs, 1H, CONH), 8.10 (d, J=9.0 Hz, 3H, H-5, H-3’, and H-5’),
7.80 (dt, J=7.0 and 1.0 Hz, 1H, H-7), 7.65 (d, J=8.0 Hz, 1H, H-8),
7.45 (t, J=7.0 Hz, 1H, H-6), 7.00 (d, J=9.0 Hz, 2H, H-6’ and H-2’),
3.30 ppm (s, 10H, piperidine CH2); 13C NMR (100 MHz, [D6]DMSO):
d=24.4, 25.3, 48.5, 114.2, 120.8, 125.8, 126.0, 126.2, 127.5, 129.4,
134.9, 149.7, 152.5, 153.4, 162.8 ppm; Anal. calcd for C19H22N2O2: C
73.52, H 7.14, N 9.03, found: C 73.62, H 7.30, N 9.00.
41 (0.33 g; 36%): white crystals; mp: 227–2308C; 1H NMR
(400 MHz, [D6]DMSO): d=9.40 (s, 1H, H-5’), 8.55 (d, J=6.0 Hz, 1H,
H-3’), 8.35 (d, J=6.0 Hz, 1H, H-2’), 8.30 (bs, 1H, CONH), 8.20 (d, J=
8.0 Hz, 1H, H-5), 7.90 (d, J=8.0 Hz, 1H, H-8), 7.60–7.70 (m, 2H, H-7
and H-7’), 7.25 (t, J=7.0 Hz, 1H, H-6), 7.10 (bs, 1H, CNH), 6.65–6.70
(m, 2H, H-8’ and H-6’), 6.50 (bs, 1H, CH); 13C NMR (100 MHz,
[D6]DMSO): d=68.2, 112.5, 114.6, 115.8, 118.3, 123.2, 126.2, 127.2,
128.0, 128.7, 129.2, 129.6, 133.5, 145.7, 146.6, 155.2, 162.4 ppm;
Anal. calcd for C17H13N3O: C 74.17, H 4.76, N 15.26, found: C 74.00,
H 4.95, N 15.18.
2-(4-Hydroxy-1-naphthyl)quinazolin-4(3H)-one (22): The synthetic
procedure for 9 was used, replacing 4-propoxybenzaldehyde with
4-hydroxy-1-naphthaldehyde, to give 22 as a light-pink powder
1
(0.56 g; 57%): mp: 263–2678C; H NMR (400 MHz, [D6]DMSO): d=
12.45 (bs, 1H, CONH), 10.75 (bs, 1H, OH), 8.20–8.30 (m, 2H, H-8’
and H-5’), 8.10–8.15 (m, 1H, H-5), 7.85 (dt, J=7.0 and 1.0 Hz, 1H,
H-7), 7.70 (d, J=8.0 Hz, 1H, H-8), 7.65 (d, J=8.0 Hz, 1H, H-3’), 7.50–
7.55 (m, 3H, H-6, H-6’, and H-7’), 7.00 (d, J=8.0 Hz, 1H, H-2’), 4.30
(bt, J=5.0 Hz, 0.3H, CH2OH), 3.40–3.50 (m, 6H, CH2), 1.10 ppm (t,
2-(2,3,4-Trihydroxyphenyl)quinazolin-4(3H)-one (26): p-Toluene-
sulfonic acid (pTsOH; 0.03 g, 0.22 mmol) was added to a suspension
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1949

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
of 2-aminobenzamide (0.3 g, 2.20 mmol) and 2,3,4-trihydroxyben-
zaldehyde (0.40 mg, 2.64 mmol) in MeOH (6.5 mL), and the mixture
was heated at reflux for 5 h. After cooling, the obtained precipitate
was filtered and washed several times with cold MeOH and crystal-
lized from a mixture of EtOH/DMF to give 26 as a white powder
111.6, 114.8, 115.4, 117.3, 127.7, 128.8, 133.5, 148.6, 149.8,
164.2 ppm; Anal. calcd for C17H19N3O2: C 68.67, H 6.44, N 14.13,
found: C 68.70, H 6.10, N 14.21.
4-Chloro-2-[(4-nitrobenzoyl)amino]benzamide (44): 4-Nitrobenzo-
yl chloride (1.2 g, 6.44 mmol) was added to a mixture of 2-amino-
4-chlorobenzamide (1.1 g, 6.44 mmol) and Et3N (0.9 mL, 6.44 mmol)
in tetrahydrofuran (THF; 9 mL). After stirring 6 h at RT, the reaction
mixture was poured into ice water, obtaining a precipitate, which
was filtered to give 44 (1.9 g, 71%): mp: 350–3518C; 1H NMR
(400 MHz [D6]DMSO): d=13.40 (bs, 1H, NH), 8.80 (d, J=2.0 Hz, 1H,
H-3), 8.60 (bs, 1H, NH2), 8.45 (d, J=9.0 Hz, 2H, H-3’ and H-5’), 8.00
(d, J=9.0 Hz, 2H, H-2’ and H-6’), 7.95–8.05 (m, 2H, H-6 and 1H
NH2), 7.35 ppm (dd, J=2.0 and 9.0 Hz, 1H, H-5).
1
(0.14 g, 43%): mp: 332–3358C; H NMR (400 MHz, [D6]DMSO): d=
12.50 (bs, 1H, CONH) 8.15, 9.75, and 14.50 (bs, each 1H, OH), 8.15
(d, J=8.0 Hz, 1H, H-5), 7.90 (t, J=8.0 Hz, 1H, H-7), 7.60–7.70 (m,
2H, H-8 and H-2’), 7.50 (t, J=8.0 Hz, 1H, H-6), 6.40 ppm (d, J=
9.0 Hz, 1H, H-3’); 13C NMR (100 MHz, [D6]DMSO): d=105.6, 110.0,
117.1, 120.6, 121.6, 122.7, 123.0, 133.5, 135.4, 146.4, 150.5, 151.4,
154.9, 161.8 ppm; Anal. calcd for C14H10N2O4: C 62.22, H 3.73, N
10.37, found: C 62.32, H 3.99, N 10.37.
2-[4-(Diethylamino)phenyl]quinazolin-4(3H)-one (33):[39] The syn-
thetic procedure for 9 was used, replacing 4-propoxybenzaldehyde
with 4-(diethylamino)benzaldehyde, to give 33 as a white powder
(0.53 g; 53%): mp: 207–2098C; 1H NMR (400 MHz [D6]DMSO): d=
12.15 (s, 1H, CONH), 8.10 (m, 3H, H-5, H-2’, and H-6’), 7.80 (t, J=
8.0 Hz, 1H, H-7), 7.70 (d, J=8.0 Hz, 1H, H-8), 7.40 (t, J=7.0 Hz, 1H,
H-6), 6.80 (d, J=9.0 Hz, 2H, H-3’ and H-5’), 3.45–3.50 (q, J=7.0 Hz,
4H, NCH2), 3.85 ppm (t, J=7.0 Hz, 6H, CH3); 13C NMR (100 MHz,
[D6]DMSO): d=12.85, 42.1, 111.0, 118.2, 120.7, 125.7, 126.2, 127.3,
129.6, 134.9, 149.8, 150.1, 152.7, 162.9 ppm; Anal. calcd for
C18H19N3O: C 73.69, H 6.53, N 14.32, found: C 73.69, H 6.55, N
14.26.
7-Chloro-2-(4-nitrophenyl)quinazolin-4(3H)-one (38):[47] Potassium
tert-butoxide (0.4 g, 3.55 mmol) was added portion wise to a sus-
pension of 44 (0.2 g, 0.71 mmol) in tert-butanol (15 mL). After stir-
ring 24 h at RT, the reaction mixture was evaporated to dryness to
give a residue, which was treated with 2n HCl. The obtained solid
was filtered and crystallized from EtOH to give 38[47] (0.15 g, 70%):
mp: 358–3598C; 1H NMR (400 MHz [D6]DMSO): d=13.40 (bs, 1H,
CONH), 8.35–8.40 (m, 4H, H-2’, H-3’, H-5’, and H-6’), 8.20 (d, J=
9.0 Hz, 1H, H-5), 7.90 (d, J=2.0 Hz, 1H, H-8), 7.60 ppm (dd, J=2.0
and 9.0 Hz, 1H, H-6); 13C NMR (100 MHz, [D6]DMSO): d=119.9,
122.9124.2, 124.3, 127.6, 129.0, 135.6, 140.1, 146.7, 148.9, 156.7,
162.2 ppm; Anal. calcd for C14H8ClN3O3: C 55.74, H 2.67, N 13.96,
found: C 56.04, H 2.81, N 13.85.
7-Chloro-2-(4-methoxyphenyl)quinazolin-4(3H)-one (35):[40] The
synthetic procedure for 9 was used, replacing isatoic anhydride
and 4-propoxybenzaldehyde with 7-chloro-isatoic anhydride and 4-
methoxybenzaldehyde, respectively, to give 35 as a light-green
powder (0.52 g; 54%): mp: 277–2808C; 1H NMR (400 MHz
[D6]DMSO): d=3.85 (s, 3H, OCH3), 7.10 (d, J=8.0 Hz, 2H, H-3’ and
H-5’), 7.50 (dd, J=2.0 and 9.0 Hz, 1H, H-6), 7.75 (d, J=2.0 Hz, 1H,
H-8), 8.10 (d, J=9.0 Hz, 1H, H-5), 8.20 (d, J=8.0 Hz, 2H, H-2’ and
H-6’), 12.50 ppm (s, 1H, CONH); 13C NMR (100 MHz, [D6]DMSO): d=
55.9, 114.5, 119.8, 124.7, 126.7, 126.8, 128.3, 130.1, 139.6, 150.4,
153.7, 162.1, 162.5 ppm; Anal. calcd for C15H11ClN2O2: C 62.84, H
3.87, N 9.77, found: C 62.48, H 4.05, N 9.59.
2-(4-Aminophenyl)-7-chloroquinazolin-4(3H)-one (37): A solution
of 38[47] (0.1 g, 0.3 mmol) in DMF (10 mL), was hydrogenated over
a catalytic amount of Raney nickel at RT and atmospheric pressure
for 2 h. The mixture was then filtered over Celite, and the filtrate
was evaporated to dryness to afford a residue, which was crystal-
1
lized from EtOH to give 37 (0.05 g, 60%): mp: 350–3518C; H NMR
(400 MHz [D6]DMSO): d=12.20 (bs, 1H, CONH), 8.10 (d, J=9.0 Hz,
1H, H-5), 7.95 (d, J=8.0 Hz, 2H, H-2’ and H-6’), 7.70 (d, J=2.0 Hz,
1H, H-8), 7.40 (dd, J=2.0 and 9.0 Hz, 1H, H-6), 6.60 (d, J=8.0 Hz,
2H, H-3’ and H-5’), 6.00 ppm (bs, 2H, NH2); 13C NMR (100 MHz,
[D6]DMSO): d=113.5, 118.6, 125.8, 126.3, 128.3, 129.7, 139.3, 151.0,
152.8, 154.2 162.3 ppm; Anal. calcd for C14H10ClN3O: C 61.89, H
3.71, N 15.47, found: C 61.96, H 3.71, N 15.47.
7-Chloro-2-{4-[(2-hydroxyethyl)-
(methyl)amino]phenyl}quinazolin-4(3H)-one (36): The synthetic
procedure for 9 was used, replacing isatoic anhydride and 4-pro-
poxybenzaldehyde with 7-chloro-isatoic anhydride and 4-[(2-
hydroxyethyl)(methyl)amino]benzaldehyde, respectively, to give 36
Computational methods
1
as white crystals (0.51 g; 46%): mp: 250–2518C; H NMR (400 MHz
Docking target preparation: The crystal structure of CDK9 in com-
plex with cyclin T1 (CycT1) and flavopiridol was retrieved from the
RCSB Protein Data Bank (PDBID: 3BLR)[24] and used as target for
our modeling studies. First, Schrçdinger’s Protein Preparation
Wizard[61–63] was used to obtain satisfactory starting structures for
docking studies. This facility is designed to ensure chemical cor-
rectness and to optimize a protein structure for further analysis. In
particular, all water molecules were deleted, hydrogen atoms were
added, bond orders and charges were then assigned, the orienta-
tion of hydroxy groups on Ser, Thr and Tyr, the side chains of Asn
and Gln residues, and the protonation state of His residues were
optimized. Moreover, the process relieves steric clashes by per-
forming a series of restrained, partial minimizations on the co-crys-
tallized structure, each of which employs a limited number of mini-
mization steps. It was not intended to minimize the system com-
pletely. In our study, the minimization (OPLS 2005 force field)[64]
was stopped when the root-mean square deviation (RMSD) values
of the non-hydrogen atoms reached 0.30 ꢁ, the specified limit by
default. An enclosing box was used as docking space, centered on
the active site using the corresponding ligand crystallographic po-
[D6]DMSO): d=12.50 (bs, 1H, CONH), 8.05–8.15 (m, 3H, H-5, H-2’,
and H-6’), 7.65 (d, J=2.0 Hz,1H, H-8), 7.45 (dd, J=2.0 and 9.0 Hz,
1H, H-6), 6.75 (d, J=9.0 Hz, 2H, H-3’ and H-5’), 4.50 (bs, 1H, OH),
3.50 and 3.55 (t, J=6.0 Hz, each 2H, NCH2 and CH2OH), 3.00 ppm
(s, 3H, NCH3); 13C NMR (100 MHz, [D6]DMSO): d=39.1, 54.2, 58.5,
111.3, 118.1, 118.5, 125.9, 126.2, 128.3, 129.7, 139.4, 150.8, 152.1,
154.1, 162.3 ppm; Anal. calcd for C17H16ClN3O2: C 61.91, H 4.89, N
12.74, found: C 61.99, H 4.98, N 12.49.
2-{4-[(2-Hydroxyethyl)(methyl)amino]phenyl}-2,3-dihydroquina-
zolin-4(1H)-one (40): The synthetic procedure for 9 was used, re-
placing
4-propoxybenzaldehyde
with
4-[(2-hydroxyethyl)-
white powder
(methyl)amino]benzaldehyde, to give 40 as
a
(0.43 g; 43%): mp: 177–1808C; 1H NMR (400 MHz [D6]DMSO): d=
8.00 (bs, 1H, CONH), 7.65 (d, J=8.0 Hz, 1H, H-5), 7.20–7.30 (m, 3H,
H-7, H-3’, and H-5’), 6.90 (bs, 1H, NH), 6.60–6.70 (m, 4H, H-6, H-8,
H-2’ and H-6’), 5.60 (bs, 1H, CH), 4.60 (bt, J=5.0 Hz, 1H, OH), 3.45–
3.55 (m, 2H, CH2O), 3.30 (t, J=5.0 Hz, 2H, NCH2), 2.90 ppm (s, 3H,
NCH3); 13C NMR (100 MHz, [D6]DMSO): d=39.1, 54.5, 58.3, 67.1,
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1950

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
sition; the ligand diameter midpoint box was set to the default
value (10 ꢁ). No van der Waals (VdW) radii scaling was used. The
same target preparation protocol was applied to crystal structures
3TNH and 3TNW.[57]
scribed previously.[72,73] Briefly, 293T cells were transfected with an
expression plasmid for flag-tagged CDK9.The cells were collected
36 h post-transfection in NHEN buffer (20 mm HEPES pH 7.5,
300 mm NaCl, 0.5% NP-40, 20% glycerol, 1 mm EDTA, protease in-
hibitors) and sonicated. The lysate was clarified by centrifugation
and incubated with anti-flag beads (M2; Sigma). The beads were
then washed several times with NHEN buffer and kept for storage
at À208C until use. The recombinant GST-CTD fusion protein was
produced and purified from E. coli BL21.[74] Glutathione agarose
beads were then washed extensively and kept frozen at À208C
until use. For the kinase assay, the beads with GST-CTD and CDK9
were combined at 5:1 ratio and washed (3ꢃ) with fresh kinase
buffer (50 mm Tris-HCl pH 7.4, 5 mm MgCl2, 5 mm MnCl2, 20 mm
NaF, protease inhibitors; Roche). The beads were then supplement-
ed with dithiothreitol (DTT; 2.5 mm), ATP (10 mm), [g32P]ATP (3 mCi)
and inhibitor (1 mL) diluted in DMSO (vehicle). This reaction mixture
was then incubated at 308C for 30 min. The reaction was stopped
by adding SDS sample buffer [100 mm Tris·HCl (pH 6.8), 4%
sodium dodecyl sulfate (SDS), 0.2% bromophenol blue, 20% glyc-
erol, 200 mm DTT] and boiling at 958C, and samples were loaded
into an 8% SDS-PAGE gel. After the run, the gels were stained with
Coomassie blue, dried on 3M paper, and radioactivity was detected
on a phosphor imaging system (Cyclone, Canberra Packard) and
analyzed with OptiQuant (Packard).
Docking of fragments: The Schrçdinger fragment library was
docked using Glide[65] in a stepwise manner. In particular, the li-
brary was first docked using Glide SP, using 0.80 to scale the VdW
radii of the nonpolar ligand atoms with a charge cutoff of 0.15.
Poses were discarded as duplicates if both RMSD values in the
ligand-all atom was less than 1.5 ꢁ and maximum atomic displace-
ment was less than 2.3 ꢁ. At most, two poses per ligand were re-
tained. The obtained docking poses were then refined, rescored
and minimized using Glide XP.
Ligand preparation: The structures of the molecules to screen
were processed using the Schrçdinger LigPrep utility.[66] This gener-
ates a number of low-energy 3D structures from each input mole-
cule, with various ionization states, tautomers, stereochemistries,
and ring conformations. For our study, a range of pH 6–8 was set,
specified chiralities were retained, desalt was applied, tautomeric
forms were obtained, and only one low-energy ring conformation
per ligand was generated.
Virtual screening procedure: The compounds were screened
using the virtual screening workflow of the Schrçdinger Suite.[64]
The databases were first docked with Glide HTVS and the 5% of
the top-scoring molecules was submitted to Glide SP, keeping the
10% of the top-ranked molecules for the final docking and scoring
by mean of Glide XP. All parameters of the three docking steps
were kept as default.
CDK2 kinase assay: The biochemical activity of compounds was de-
termined by incubation with cyclin-dependent kinase 2 (CDK2)/cy-
clin A and substrates, followed by quantification of the adenosine
diphosphate (ADP) product by using the ADPGlo assay format
(Promega). The recombinant GST-CDK2/cyclin A fusion protein was
produced and purified as described previously;[75] Histone H1 and
adenosine triphosphate (ATP) were commercially available (Sigma).
Enzyme, Histone H1 and ATP, tested at 5 nm, 5 mm and 1 mm, re-
spectively, were diluted in kinase buffer (50 mm Tris/HCl pH 7.5,
10 mm MgCl2, 3 mm Na3VO4, 1 mm DTT, 0.2 mgmLÀ1 BSA). The com-
pounds were threefold serially diluted in 100% DMSO followed by
an intermediate dilution step in assay buffer in order to have a con-
centration curve from 139 mm to 0.007 mm in 1% DMSO. Kinase re-
actions were performed in a 96-well plate at RT for 30 min; flavo-
piridol and staurosporine were used as internal standards while
positive controls (100% of inhibition, BOTTOM) in association with
negative controls (no inhibition, TOP) were used to normalize data
results. At the end of the incubation time, 25 mL of the kinase reac-
tion was transferred in 96-well Optiplate (Greiner) in order stop the
reaction and quantify the ADP generated by the ADPGlo assay. The
luminescence signals were measured using ViewLux Reader (Perki-
nElmer, Boston, MA, USA). The raw data were analyzed using
GraphPad Prism (version 5.2) by a four-parameter logistic equation,
which provides sigmoidal fittings of the ten-dilution curves for IC50
Docking of 37 into 3TNW and 3TNH: Compound 37 was docked
into both structures (PDBID: 3TNW and 3TNH) using Glide[65] in
a stepwise manner. In particular, the compound was first docked
using Glide SP, using 0.80 to scale the VdW radii of the nonpolar
ligand atoms with a charge cutoff of 0.15. Poses were discarded as
duplicates if both RMSD values in the ligand-all atom was less than
1.5 ꢁ and the maximum atomic displacement was less than 2.3 ꢁ.
At most, three poses per ligand were retained. The obtained dock-
ing poses were then refined and rescored using Glide XP, retaining
only the best scoring for each target. The resulting complexes
were finally minimized through 500 steps of conjugate gradient
minimization, using MM/GBSA[67] as solvation treatment and the
optimized potentials for liquid simulations (OPLS) force field.[64]
MC/FEP calculations: Starting from the PDBID 3BLR, a reduced
model of the protein was prepared using the residues comprised
in a 15 ꢁ distance from the ligand. The reduced protein model in
complex with 2 was relaxed by 100 steps of conjugate gradient
minimization. Monte Carlo (MC)/free energy perturbation (FEP) cal-
culations were performed by means of MCPRO[68] following a stan-
dard 11 windows simple overlap sampling[69] protocol. TIP4P water
model,[70] OPLS force-field[64] and CM1 A[71] ligand charges were em-
ployed. During the MC simulations, all degrees of freedom were
sampled for the ligand; side chains of residues comprised in a dis-
tance of 10 ꢁ from the ligand were sampled as well, whereas the
protein backbone was kept fixed. For the two ortho and meta posi-
tions, the most energetically favorable substitution was selected,
and an entropic penalty (0.6 kcalmolÀ1) was applied for the loss of
a rotameric state upon ligand binding.
determination:
Y=BOTTOM+(TOPÀBOTTOM)/(1+10((
logIC50ÀX)*SLOPE)), where X is the logarithm of the inhibitor con-
centration, Y is the response, SLOPE (Hill slope) is related to the
stoichiometry of inhibitor enzyme interaction, TOP is the signal
generated by the reaction without inhibitor and BOTTOM is the
signal completely inhibited.
Cytotoxicity tests: Cytotoxicity in HeLa cells was assayed with
Alamar blue (Invitrogen) as described in the product’s specifica-
tions. Briefly, cells were seeded in a 96-well plate and incubated
with the compounds at the indicated concentrations for 24 h
before addition of the dye. After 24 h, the fluorescent signal was
detected using a fluorimeter (Wallac plate reader, PerkinElmer).
Biology
Transactivation assay: The assay was performed as previously de-
scribed.[73]
CDK9 kinase assay: The cyclin-dependent kinase 9 (CDK9) activity
was measured in vitro on a recombinant GST-CTD substrate as de-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1951

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Anti-HIV-1 and cytotoxic assays in acutely infected MT-4 cells: Evalua-
tion of the antiviral activity of the compounds against HIV-1 strain
IIIB in MT-4 cells was performed using the MTT assay as previously
described.[76,77] Briefly, stock solutions (10ꢃ final concentration) of
test compounds were added in 25 mL volumes to two series of trip-
licate wells so as to allow simultaneous evaluation of their effects
on mock-and HIV-1-infected cells at the beginning of each experi-
ment. Serial fivefold dilutions of test compounds were made di-
rectly in flat-bottomed 96-well microtiter trays using a Biomekꢄ
3000 workstation (Beckman Coulter). Untreated control HIV-1- and
mock-infected cell samples were included for each sample. HIV-
1(IIIB)[78] stock (50 mL) at 100–300 50% cell culture infectious dose
(CCID50) or culture medium was added to either the infected or
mock-infected wells of the microtiter tray. Mock-infected cells were
used to evaluate the effect of test compound on uninfected cells
in order to assess the cytotoxicity of the test compound. Exponen-
tially growing MT-4 cells[79] were centrifuged for 5 min at 1000 rpm
(220 g), and the supernatant was discarded. The MT-4 cells were re-
suspended at 6ꢃ105 cellsmLÀ1, and 50 mL volumes were trans-
ferred to the microtiter tray wells. Five days after infection, the via-
bility of mock-and HIV-1-infected cells was examined spectrophoto-
metrically using the MTT assay as described previously.[76] All data
were calculated using the median optical density (OD) value of
three wells. The CC50 was defined as the concentration of the test
compound that reduced the absorbance (OD540) of the mock-in-
fected control sample by 50%. The concentration achieving 50%
protection from the cytopathic effect of the virus in infected cells
was defined as the EC50.
HTVS, high-throughput virtual screening; LTR, long terminal
repeat; OPLS, optimized potentials for liquid simulations; PMA,
phorbol myristate acetate; P-TEFb, positive transcription elon-
gation factor b; RNAPII, RNA polymerase II; SP, standard preci-
sion; TAR, Tat responsive element; XP, extra precision.
Acknowledgements
This work, coordinated by O.T., was financially supported by the
Italian Ministry of Health (AIDS, UPR-2009-1301355), PRIN 2008
(grant no. 2008CE75SA), PRIN 2011 (grant no. 2010W2KM5L_004),
and KU Leuven (GOA 10/014). We are grateful to Roberto Bianco-
ni (Department of Chemistry and Technology of Drugs, University
of Perugia) and Kristien Erven, Cindy Heens, and Kris Uyttersprot
(Rega Institute for Medical Research, Katholieke Universiteit
Leuven) for excellent technical assistance.
Keywords: antiviral agents · CDK9 · computational chemistry ·
inhibitors · kinases · P-TEFb · Tat-mediated transcription
[1] L. Colin, C. Van Lint, Retrovirology 2009, 6, 111.
[2] A. Marcello, Retrovirology 2006, 3, 7.
[3] L. Shen, R. F. Siliciano, J. Allergy Clin. Immunol. 2008, 122, 22–28.
[4] J. Karn, Curr. Opin. HIV AIDS 2011, 6, 4–11.
[5] K. Chiba, J. Yamamoto, Y. Yamaguchi, H. Handa, Exp. Cell Res. 2010, 316,
2723–2730.
Anti-HIV-1 and cytotoxic assays in latently infected OM-10.1 cells: The
activity values of the compounds against latent HIV-1 infection
were based on the inhibition of p24 antigen production in OM-
10.1 cells after stimulation with phorbol 12-myristate 13-acetate
(PMA; Sigma–Aldrich). Briefly, OM-10.1 cells (5ꢃ105 cellsmLÀ1) were
incubated in the presence or absence of the compounds for 2 h in
48-well plates. After this short incubation period, the cells were
stimulated with PMA (0.02 mm) followed by transfer of 2ꢃ 200 mL
into a 96-well plate for toxicity evaluation. After a 2 day incubation
at 378C, the cell-culture supernatants were collected from the 48-
well plates and examined for their p24 antigen levels with the HIV-
1 p24 ELISA kit (PerkinElmer). Cytotoxicity of the compounds for la-
tently HIV-1-infected OM-10.1 cell line in the 96-well plates were
based on the MTT cell viability staining as described previously.[76]
[6] H. S. Mancebo, G. Lee, J. Flygare, J. Tomassini, P. Luu, Y. Zhu, J. Peng, C.
Blau, D. Hazuda, D. Price, O. Flores, Genes Dev. 1997, 11, 2633–2644.
[7] S. Massari, S. Sabatini, O. Tabarrini, Curr. Pharm. Des. 2013, 19, 1860–
1879.
[8] W. Coley, K. Kehn-Hall, R. Van Duyne, F. Kashanchi, Expert Opin. Biol.
Ther. 2009, 9, 1369–1382.
[9] O. Flores, G. Lee, J. Kessler, M. Miller, W. Schlief, J. Tomassini, D. Hazuda,
Proc. Natl. Acad. Sci. USA 1999, 96, 7208–7213.
[10] L. M. Schang, J. Antimicrob. Chemother. 2002, 50, 779–792.
[11] Y. L. Chiu, H. Cao, J. M. Jacque, M. Stevenson, T. M. Rana, J. Virol. 2004,
78, 2517–2529.
[12] S. Wang, P. M. Fischer, Trends Pharmacol. Sci. 2008, 29, 302–313.
[13] Z. Yang, Q. Zhu, K. Luo, Q. Zhou, Nature 2001, 414, 317–322.
[14] L. Y. Liou, C. H. Herrmann, A. P. Rice, J. Virol. 2004, 78, 8114–8119.
[15] J. H. Yik, R. Chen, A. C. Pezda, C. S. Samford, Q. Zhou, Mol. Cell. Biol.
2004, 24, 5094–5105.
[16] A. Pumfery, C. de La Fuente, R. Berro, S. Nekhai, F. Kashanchi, S. H. Chao,
Curr. Pharm. Des. 2006, 12, 1949–1961.
[17] S. Biglione, S. A. Byers, J. P. Price, V. T. Nguyen, O. Bensaude, D. H. Price,
W. Maury, Retrovirology 2007, 4, 47.
Supporting Information
Dose–response curve and histogram of virtual hit 2 (figure S1),
CDK9 kinase assay for flavopiridol (1) (figure S2), dose–response
curve of virtual hit 3 (figure S3), dose–response curve of virtual hit
4 (figure S4), cytotoxicity assay for compound 37 and 1 (figure S5),
docking ranking, scores and anti-CDK9 activity of 14 selected virtu-
al screening hits (table S1), FEP calculation results (table S2), dock-
ing ranking and scores of the molecules selected from the combi-
natorial library (table S3), elemental analysis data for resynthesized
[18] A. Ali, A. Ghosh, R. S. Nathans, N. Sharova, S. O’Brien, H. Cao, M. Steven-
son, T. M. Rana, ChemBioChem 2009, 10, 2072–2080.
[19] G. Nꢅmeth, Z. Varga, Z. Greff, G. Bencze, A. Sipos, C. Szꢆntai-Kis, F.
Baska, A. Gyuris, K. Kelemenics, Z. Szathmꢆry, J. Minꢆrovits, G. Kꢅri, L.
Orfi, Curr. Med. Chem. 2011, 18, 342–358.
[20] V. Krystof, S. Baumli, R. Fꢂrst, Curr. Pharm. Des. 2012, 18, 2883–2890.
[21] A. Narayanan, G. Sampey, R. Van Duyne, I. Guendel, K. Kehn-Hall, J.
Roman, R. Currer, H. Galons, N. Oumata, B. Joseph, L. Meijer, M. Caputi,
S. Nekhai, F. Kashanchi, Virology 2012, 432, 219–231.
[22] O. Tabarrini, S. Massari, V. Cecchetti, Future Med. Chem. 2010, 2, 1161–
1180.
1
compounds 3 and 4, (table S4), and H NMR data for commercial
compounds 2, 3 and 4.
[23] O. Tabarrini, S. Massari, L. Sancineto, D. Daelemans, S. Sabatini, G. Man-
froni, V. Cecchetti, C. Pannecouque, ChemMedChem 2011, 6, 1249–
1257.
Abbreviations
[24] S. Baumli, G. Lolli, E. D. Lowe, S. Troiani, L. Rusconi, A. N. Bullock, J. E.
Debreczeni, S. Knapp, L. N. Johnson, EMBO J. 2008, 27, 1907–1918.
[25] M. Congreve, R. Carr, C. Murray, H. Jhoti, Drug Discovery Today 2003, 8,
876–877.
bs, broad singlet; cART, combination antiretroviral therapy;
CCID50, culture infectious dose; CDK2, cyclin dependent
kinase 2; CDK9, cyclin dependent kinase 9; CTD, carboxy termi-
nal domain; CycT1, cyclin T1; FEP, free energy perturbation;
[26] eMolecules. http://www.emolecules.com (accessed February 2010).
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1952

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[27] Asinex. http://www.asinex.com/ (accessed February 2010).
[28] R. W. Zwanzig, J. Chem. Phys. 1954, 22, 1420–1427
[29] CombiGlide, version 2.7, Schrçdinger, LLC, New York, NY, 2011.
[30] P. S. Reddy, P. P. Reddy, T. Vasantha, Heterocycles 2003, 60, 183–226.
[31] M. Baghbanzadeh, M. Dabiri, P. Salehi, Heterocycles 2008, 75, 2809–
2815.
[32] L. Gao, H. Ji, L. Rong, D. Tang, Y. Zha, Y. Shi, S. Tu, J. Heterocycl. Chem.
2011, 48, 957–960.
[33] M. J. Hour, L. J. Huang, S. C. Kuo, Y. Xia, K. Bastow, Y. Nakanishi, E.
Hamel, K. H. Lee, J. Med. Chem. 2000, 43, 4479–4487.
[34] A. B. Bhat, D. P. Sahu, Synth. Commun. 2004, 34, 2169–2176.
[35] L. Y. Zeng, C. Cai, J. Heterocycl. Chem. 2010, 47, 1035–1039.
[36] P. R. Bhalla, B. L. Walworth, (American Cyanamid Co., Stanford, CT, USA)
EP 0058822 A1, 1982. [Chem. Abstr. 1983, 98, 1669].
[37] J. J. Caldwell, E. J. Welsh, C. Matijssen, V. E. Anderson, L. Antoni, K.
Boxall, F. Urban, A. Hayes, F. I. Raynaud, L. J. Rigoreau, T. Raynham, G. W.
Aherne, L. H. Pearl, A. W. Oliver, M. D. Garrett, I. Collins, J. Med. Chem.
2011, 54, 580–590.
Lane, R. C. Jackson, D. M. Glover, D. G. Blake, P. M. Fischer, Chem. Biol.
2010, 17, 1111–1121.
[57] S. Baumli, A. J. Hole, M. E. Noble, J. A. Endicott, ACS Chem. Biol. 2012, 7,
811–816.
[58] H. Shao, S. Shi, S. Huang, A. J. Hole, A. Y. Abbas, S. Baumli, X. Liu, F. Lam,
D. W. Foley, P. M. Fischer, M. Noble, J. A. Endicott, C. Pepper, S. Wang, J.
Med. Chem. 2013, 56, 640–659.
[59] A. J. Hole, S. Baumli, H. Shao, S. Shi, S. Huang, C. Pepper, P. M. Fischer, S.
Wang, J. A. Endicott, M. E. Noble, J. Med. Chem. 2013, 56, 660–670.
[60] L. Sancineto, N. Iraci, S. Massari, O. Tabarrini, Curr. Med. Chem. 2013, 20,
1355–1381.
[61] Schrçdinger Suite 2011, Schrçdinger, LLC, New York, NY, 2011.
[62] Impact version 5.7, Schrçdinger, LLC, New York, NY, 2011.
[63] Prime version 2.3, Schrçdinger, LLC, New York, NY, 2011.
[64] W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc. 1996,
118, 11225–11236.
[65] Glide, version 5.7, Schrçdinger, LLC, New York, NY, 2011.
[66] LigPrep, version 2.5, Schrçdinger, LLC, New York, NY, 2011.
[67] W. C. Still, A. Tempczyk, R. C. Hawley, T. Hendrickson, J. Am. Chem. Soc.
1990, 112, 6127–6129.
[38] T. G. Deligeorgiev, S. Kaloyanova, A. Vasilev, J. J. Vaquero, J. Alvarez-Buil-
lab, A. M. Cuadro, Color. Technol. 2010, 126, 24–30.
[39] R. Zink, I. J. Fletcher, (Ciba-Geigy AG, Basel, Switzerland) EP 0109930 A1,
1984 [Chem. Abstr. 1984, 101, 181249].
[40] R. L. McKee, M. K. McKee, R. W. Bost, J. Am. Chem. Soc. 1947, 69, 940–
942.
[41] S. Lꢂhr, M. Vilches-Herrera, A. Fierro, R. R. Ramsay, D. E. Edmondson, M.
Reyes-Parada, B. K. Cassels, P. Iturriaga-Vꢆsquez, Bioorg. Med. Chem.
2010, 18, 1388–1395.
[68] W. L. Jorgensen, J. Tirado-Rives, J. Comput. Chem. 2005, 26, 1689–1700.
[69] W. L. Jorgensen, L. L. Thomas, J. Chem. Theory Comput. 2008, 4, 869–
876.
[70] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, M. L. J.
Klein, J. Chem. Phys. 1983, 79, 926–935.
[71] W. L. Jorgensen, J. Tirado-Rives, Proc. Natl. Acad. Sci. USA 2005, 102,
6665–6670.
[42] G. Solladiꢅ, Y. Pasturel-Jacopꢅ, J. Maignan, Tetrahedron 2003, 59, 3315–
3321.
[72] A. Sabꢉ, M. Lusic, A. Cereseto, M. Giacca, Mol. Cell. Biol. 2008, 28, 2201–
2212.
[43] E. Schipper, M. Cinnamon, L. Rascher, Y. H. Chiang, W. Oroshnik, Tetrahe-
dron Lett. 1968, 9, 6201–6204.
[44] J. Hanusek, L. Hejtmꢆnkovꢆ, L. Kubicovꢆ, M. Sedlꢆk, Molecules 2001, 6,
323–337.
[45] A. Bartolozzi, S. Campbell, B. Cole, J. Ellis, H. Foudoulakis, B. Kirk, S.
Ram, P. Sweetnam, M. Hauer-Jensen, M. Boerma, (Surface Logix, Inc.,
Brighton, MA, USA) Int. PCT Pub. No. WO 2008/054599 A3, 2008;
[Chem. Abstr. 2008, 148, 538292].
[46] A. D. Roy, A. Subramanian, R. Roy, J. Org. Chem. 2006, 71, 382–385.
[47] V. S. Patel, S. R. Patel, J. Indian Chem. Soc. 1968, 45, 167–170.
[48] F. Kashanchi, K. Kehn-Hall, Curr. Pharm. Des. 2009, 15, 2520–2532.
[49] J. Mꢅndez, Cell 2003, 114, 398–399.
[50] C. Barriꢇre, D. Santamarꢈa, A. Cerqueira, J. Galꢆn, A. Martꢈn, S. Ortega, M.
Malumbres, P. Dubus, M. Barbacid, Mol. Oncol. 2007, 1, 72–83.
[51] S. Nekhai, M. Zhou, A. Fernandez, W. S. Lane, N. J. Lamb, J. Brady, A.
Kumar, Biochem. J. 2002, 364, 649–657.
[52] T. Ammosova, R. Berro, M. Jerebtsova, A. Jackson, S. Charles, Z. Klase, W.
Southerland, V. R. Gordeuk, F. Kashanchi, S. Nekhai, Retrovirology 2006,
3, 78.
[73] S. Massari, D. Daelemans, M. L. Barreca, A. Knezevich, S. Sabatini, V. Cec-
chetti, A. Marcello, C. Pannecouque, O. Tabarrini, J. Med. Chem. 2010,
53, 641–648.
[74] A. Marcello, P. Massimi, L. Banks, M. Giacca, J. Virol. 2000, 74, 9090–
9098.
[75] P. Pevarello, M. G. Brasca, R. Amici, P. Orsini, G. Traquandi, L. Corti, C.
Piutti, P. Sansonna, M. Villa, B. S. Pierce, M. Pulici, P. Giordano, K. Martina,
E. L. Fritzen, R. A. Nugent, E. Casale, A. Cameron, M. Ciomei, F. Roletto,
A. Isacchi, G. Fogliatto, E.; Pesenti, W. Pastori, A. Marsiglio, K. L. Leach,
P. M. Clare, F. Fiorentini, M. Varasi, A. Vulpetti, M. A. Warpehoski, J. Med.
Chem. 2004, 47, 3367–3380; E. Pesenti, W. Pastori, A. Marsiglio, K. L.
Leach, P. M. Clare, F. Fiorentini, M. Varasi, A. Vulpetti, M. A. Warpehoski,
J. Med. Chem. 2004, 47, 3367–3380.
[76] R. Pauwels, J. Balzarini, M. Baba, R. Snoeck, D. Schols, P. Herdewijn, J.
Desmyter, E. De Clercq, J. Virol. Methods 1988, 20, 309–321.
[77] C. Pannecouque, D. Daelemans, E. De Clercq, Nat. Protoc. 2008, 3, 427–
434.
[78] M. Popovic, M. G. Sarngadharan, E. Read, R. C. Gallo, Science 1984, 224,
497–500.
[53] D. Breuer, A. Kotelkin, T. Ammosova, N. Kumari, A. Ivanov, A. V. Ilatovskiy,
M. Beullens, P. R. Roane, M. Bollen, M. G. Petukhov, F. Kashanchi, S.
Nekhai, Retrovirology 2012, 9, 94.
[54] S. Baumli, J. A. Endicott, L. N. Johnson, Chem. Biol. 2010, 17, 931–936.
[55] T. H. Tahirov, N. D. Babayeva, K. Varzavand, J. J. Cooper, S. C. Sedore,
D. H. Price, Nature 2010, 465, 747–751.
[79] H. Miyoshi, I. Taguchi, S. Kobonishi, Y. Yoshimoto, Y. Ohtsuki, T. Shiraishi,
T. Akagi, Gann Monogr. 1982, 28, 219–228.
[80] The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrçdinger,
LLC.
[56] S. Wang, G. Griffiths, C. A. Midgley, A. L. Barnett, M. Cooper, J. Grabarek,
L. Ingram, W. Jackson, G. Kontopidis, S. J. McClue, C. McInnes, J. McLa-
chlan, C. Meades, M. Mezna, I. Stuart, M. P. Thomas, D. I. Zheleva, D. P.
Received: June 28, 2013
Published online on October 21, 2013
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1941 – 1953 1953