Toward the Discovery of Novel Anti-HIV Drugs. Second-Generation Inhibitors of the Cellular ATPase DDX3 with Improved Anti-HIV Activity: Synthesis, Structure-Activity Relationship Analysis, Cytotoxicity Studies, and Target Validation

DOI: 10.1002/cmdc.201100166

Toward the Discovery of Novel Anti-HIV Drugs. Second-

Generation Inhibitors of the Cellular ATPase DDX3 with

Improved Anti-HIV Activity: Synthesis, Structure–Activity

Relationship Analysis, Cytotoxicity Studies, and Target

Validation

Giovanni Maga,^[b]Federico Falchi,^[a]Marco Radi,^[a]Lorenzo Botta,^[a]Gianni Casaluce,^[a]

Martina Bernardini,^[a]Hamid Irannejad,^[a]Fabrizio Manetti,^[a]Anna Garbelli,^[b]

Alberta Samuele,^[b]Samantha Zanoli,^[b]Josꢀ A. Estꢀ,^[c]Emmanuel Gonzalez,^[c]Elisa Zucca,^[b]

Stefania Paolucci,^[d]Fausto Baldanti,^[d]Jan De Rijck,^[e]Zeger Debyser,^[e]and

Maurizio Botta*^{[a, f]}

A hit optimization protocol applied to the first nonnucleoside

inhibitor of the ATPase activity of human DEAD-box RNA heli-

case DDX3 led to the design and synthesis of second-genera-

tion rhodanine derivatives with better inhibitory activity

toward cellular DDX3 and HIV-1 replication. Additional DDX3

inhibitors were identified among triazine compounds. Biologi-

cal data were rationalized in terms of structure–activity rela-

tionships and docking simulations. Antiviral activity and cyto-

toxicity of selected DDX3 inhibitors are reported and dis-

cussed. A thorough analysis confirmed human DDX3 as a valid

anti-HIV target. The compounds described herein represent a

significant advance in the pursuit of novel drugs that target

HIV-1 host cofactors.

Introduction

Viruses are obligate intracellular parasites that hijack the meta-

bolic machinery of their host cells in order to replicate and

propagate the infection. Functional genomic screenings have

recently identified about 300 human proteins important for

the replicative cycle of human immunodeficiency virus type 1

(HIV-1).^[1,2]Such previously unrecognized complexity of host–

pathogen relationships, coupled to the seemingly unavoidable

problem of drug resistance due to the high mutation rate of

HIV-1, has renewed the long-standing interest toward cellular

proteins as novel targets for the treatment of HIV-1 infec-

tion.^[3,4]The main argument in favor of targeting a host factor

instead of a viral protein is the predicted low drug resistance

level. On the other hand, interfering with a cellular pathway

might lead to serious complications, including crucial cell func-

tion loss or oncogenic drift. Thus, it is of paramount impor-

tance to identify host cell factors that, while absolutely essen-

tial for the virus, are dispensable for the host cell metabolism.

One protein that has recently attracted much attention is the

human DEAD-box RNA helicase DDX3.^[5–7]It has been reported

that four different viruses, hepatitis C virus (HCV), hepatitis B

virus (HBV), HIV, and poxviruses encode proteins that can inter-

act with DDX3 to modulate its functions. For this reason, DDX3

seems a prime target for viral manipulation. In addition, while

DDX3 has been implicated in many aspects of RNA metabo-

lism, such as mRNA export, transcriptional regulation and

translation, several studies have shown that DDX3 knockdown

did not result in cellular toxicity, at least in cell culture

models.^[8–12]

In the case of HIV-1, it has been shown that knockdown of

DDX3 suppressed viral replication due to a block of the Rev-

dependent export of unspliced or partially spliced viral

[a] Dr. F. Falchi, Dr. M. Radi, Dr. L. Botta, Dr. G. Casaluce, Dr. M. Bernardini,

Dr. H. Irannejad, Dr. F. Manetti, Prof. M. Botta

Dipartimento Farmaco Chimico Tecnologico

University of Siena, Via Alcide de Gasperi 2, 53100 Siena (Italy)

Fax: (+39)0577-234-306

E-mail: botta.maurizio@gmail.com

[b] Dr. G. Maga, Dr. A. Garbelli, Dr. A. Samuele, Dr. S. Zanoli, Dr. E. Zucca

Istituto di Genetica Molecolare, IGM-CNR

Via Abbiategrasso 207, 27100 Pavia (Italy)

[c] Prof. J. A. Estꢀ, Dr. E. Gonzalez

Retrovirology Laboratory IrsiCaixa

Hospital Universitari Germans Trias i Pujol

Universitat Autꢁnoma de Barcelona, 08916 Barcelona (Spain)

[d] Dr. S. Paolucci, Dr. F. Baldanti

Molecular Virology Unit, Foundation IRCCS Policlinico S. Matteo

piazzale Golgi, 27100 Pavia (Italy)

[e] Dr. J. De Rijck, Prof. Z. Debyser

Molecular Medicine, Katholieke Universiteit Leuven and IRC KULAK

Kapucijnenvoer 33, 3000 Leuven, Flanders (Belgium)

[f] Prof. M. Botta

Sbarro Institute for Cancer Research and Molecular Medicine

Temple University, BioLife Science Building

Suite 333, 1900 N. 12th St., Philadelphia, PA 19122 (USA)

Supporting information for this article is available on the WWW under

http://dx.doi.org/10.1002/cmdc.201100166.

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

1371

M. Botta et al.

MED

mRNAs.^[10]Based on these considerations, our group has un-

dertaken a rational drug design approach aimed at finding

small molecules able to suppress DDX3 activity. Because no in-

hibitors of any DEAD-box helicase are known, a structure-

based pharmacophore modeling and molecular docking proto-

col applied to the X-ray crystallographic structure of the

human helicase DDX3 in complex with AMP^[13]was set to

search, within the Asinex Gold collection,^[14a]small-molecule in-

hibitors of the ATPase activity of DDX3. As a result of such sim-

ulations, hit compounds belonging to the class of rhodanine

derivatives were found. One of these compounds (4a) showed

low micromolar potency of inhibition against the DDX3 ATPase

activity and was able to confer protection to human cells

when challenged with HIV-1, albeit at substantially higher

doses.^[11]This was the first sucessful identification of a nonnu-

cleoside analogue able to interfere with HIV-1 replication by

targeting a cellular enzyme. Almost simultaneously, another

study identified ring-expanded nucleoside inhibitors of the

DDX3 helicase activity as potential anti-HIV-1 agents.^[15]The in-

teraction pattern between the ATPase pocket and the rhoda-

tional anchor point was represented by the rhodanine carbon-

yl oxygen that interacted with the terminal NH₂group of

Gln207. Hydrophobic interactions between the terminal bro-

mophenyl moiety and a region (thereafter referred to as the

left pocket) mainly defined by the alkyl chain of Arg202 and

Glu523, as well as possible p–p contacts between the rhoda-

nine heterocycle and the Tyr200 phenyl ring further stabilized

the complexes. Electrostatic interactions between the bromo

substituent and the terminal guanidino group of Arg202 were

also found.

Here we present the results of a hit optimization process

leading to the design, synthesis, and biological characterization

of second-generation DDX3 inhibitors endowed with higher

potencies against both the target cellular enzyme and HIV-1

replication. In addition, we have carefully re-evaluated the ef-

fects of DDX3 knockdown in human cells, providing further

support for DDX3 as a promising target for anti-HIV chemo-

therapy.

nine-based hit compound 4a (Figure 1A) showed its 2-OH sub- Results and Discussion

stituent at R¹embedded within a polar pocket (thereafter re-

Chemistry

ferred to as the right pocket) and involved either as a hydro-

gen bond acceptor or donor group in interactions with the

backbone NH group of Gly229, Lys230, and Ser228, and with

the terminal carbonyl oxygen of Gln225, respectively. An addi-

Based on the proposed binding mode of the hit 4a, a series of

analogues were synthesized following the classical synthetic

approach^[16]reported in Scheme 1. This first set of derivatives is

Figure 1. Representation of the binding mode of rhodanine and triazine derivatives within the DDX3 ATP binding site. Hydrogen bond interactions are repre-

sented by yellow dashed lines. Regions of best interactions with chloride probes are green spheres. A) Best pose for 4a. The distance between the 2-OH and

the 3-Br is represented as a red dashed line. The right, upper, and left sites of the binding site are labeled. B) Oxygen atoms of 4 f (green) and 4e (blue) are

important anchor points to interact with the left site. Also the fluorine of 4b (pink) lyes in the same region of space where electrostatic interactions are al-

lowed with the guanidino group of Arg202 and where one important minimum energy point for the fluorine probe (ꢀ3.88 kcalmol^ꢀ1, see Figure 2C) was

found by GRID. C) Representation of the interaction pattern between 4a (CPK notation) and the left pocket. The bromo substituent is embedded between

the Arg202 and His727 side chain, suggesting profitable electrostatic interactions. Contacts with the polymethylenic portion of the Arg202 side chain are

also shown. D) Our first triazine derivative (15a) identified by high-throughput docking simulations. E) Inactive compound 15b only maintains the interaction

with Gln281 and a possible stacking with Tyr200. F) The chloride of 15 f restores a significant activity, accounted for the interaction with a region of best in-

teraction found by GRID for the chloride probe. Higher-resolution images are available in the Supporting Information.

1372

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

ed from commercially available 3-bromobenzaldehyde 8 which

was transformed into the anhydride 9 passing through the (E)-

benzylidenesuccinic acid intermediate (Scheme 3).^[19]Treatment

Scheme 1. Synthesis of rhodanine derivatives 4a–h. Reagents and conditions:

a) CS₂, BrCH₂COOH, 22% aq. KOH, RT, 6 h; b) pivaloyl chloride, DIPEA, 2-ami-

nophenol, 1,2-dichloroethane, RT, 18 h (for 3a), SOCl₂, 2-nitroaniline, CH₂Cl₂,

RT, 2 h (for 3b); c) R²PhCHO, EtOH, Et₃N, 1 h, reflux (Method A), R²PhCHO,

MW, 1308C, 5 min (3 cycles) (Method B).

Scheme 3. Synthesis of succinimide derivative 11. Reagents and conditions:

a) 1. Diethyl succinate, tBuOH/tBuOK, reflux, 2 h, 2. 10% aq. NaOH, reflux,

4 h, 3. Ac₂O, THF, RT, 5 h; b) b-alanine, DMF, MW, 1508C, 10 min; c) pivaloyl

chloride, DIPEA, 2-aminophenol, 1,2-dichloroethane, RT, 18 h.

characterized by a few focused modifications on the two aro-

matic rings of the hit. The synthesis of the rhodanine deriva-

tives 4a–h started from commercially available b-alanine 1

that was allowed to react with carbon disulfide and a-bromo-

acetic acid in basic medium to give the N-substituted rhoda-

nine scaffold 2. This intermediate was then coupled with two

substituted anilines to give compounds 3a,b which were final-

ly treated with various benzaldehydes, under classical or micro-

wave-assisted Knoevenagel conditions, to give the desired final

compounds 4a–h.

A second set of analogues of 4a, characterized by an invert-

ed amide moiety and by the replacement of the rhodanine

with a different scaffold, was also synthesized. Considering

that a preliminary enzymatic screening^[11]identified 4a and 4b

as the most interesting derivatives, the synthesis of their corre-

sponding inverted amide analogues 7a,b was attempted

(Scheme 2). After some initial failures with the classical proce-

dure reported above, a fast and efficient one-pot, two-step

protocol was developed, combining the “Holmberg method”

and the Knoevenagel condensation under microwave-assisted

conditions.^[17]As a result, the desired inverted amide deriva-

tives 7a,b were obtained in high yield and purity by reacting

bis(carboxymethyl)trithiocarbonate 5 with the amine 6, synthe-

sized following a reported procedure.^[18]

of the anhydride 9 with b-alanine under microwave-assisted

conditions, followed by a coupling reaction with 2-aminophe-

nol, afforded the desired final compound 11.

A triazine-based series of derivatives was then synthesized

after the identification of the hit compounds 15a and 16h

(see Table 2 below) by a high-throughput docking approach

(HTD; see the next section for details). These hits were charac-

terized by two important pharmacophoric groups (namely, the

indolyl-2-one and the 2-hydroxyphenyl moieties) connected

via a hydrazone spacer to the central triazine scaffold. In addi-

tion, considering the positive contribution of the morpholino

group of 16h to the solubility of such compounds, we decided

to synthesize two sets of triazine derivatives characterized by

the presence of both the morpholine substituent and one of

the two above mentioned pharmacophoric moieties. The syn-

thetic approach to obtain such 2,4,6-trisubstituted-1,3,5-tria-

zines was dependent on the type of amine to be introduced in

the last available position of the triazine core. The benzylamino

derivatives 15e–g and 16i,j,o,p were obtained by consecutive

nucleophilic substitution of commercially available cyanuric

chloride 12 (Scheme 4).

The first chlorine of 12 was displaced with the moderately

nucleophilic benzylamines at ꢀ308C for 3 h to give the mono-

substituted triazines 13a–c. Subsequent treatment with a

more nucleophilic amine provided the disubstituted triazines

14a–d. The remaining chlorine of disubstituted tria-

zines was replaced by treatment with excess hydra-

zine at reflux to afford the corresponding 2,4,6-trisub-

stituted triazine intermediates, which were finally al-

lowed to react with the two selected pharmacophor-

ic moieties (salicyl aldehyde or isatin) to give the final

compounds 15e–g and 16i,j,o,p, respectively.

Looking for a different scaffold to replace the rhodanine

moiety of 4a, we selected the succinimide moiety as a valua-

ble alternative. The synthesis of the desired derivative 11 start-

The above reported synthetic approach was then

applied to the synthesis of the aniline derivatives

Scheme 2. Synthesis rhodanine derivatives 7a,b. Reagents and conditions: a) 1. DME, Et₃N,

908C, MW, 10 min, 2. R²PhCHO, 1108C, MW, 5 min.

16m and 20b,c (Scheme 5). Unfortunately, reacting

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1373

M. Botta et al.

MED

previously described, the intermediates 17 and 18 were con-

verted into the final compounds 16n and 20b. A different se-

quence of nucleophilic substitutions on cyanuric chloride was

then followed in order to cleanly obtain the aniline derivatives

16m and 20b,c. In brief, an initial treatment of 12 with mor-

pholine followed by a nucleophilic substitution with 4-fluoroa-

niline and then with hydrazine gave the trisubstituted hydra-

zine intermediates, which were finally condensed with a small

set of aromatic aldehydes to give the final compounds 16m

and 20b,c as pure products.

Finally, we decided to synthesize analogues of the 2,4,6-tri-

substituted-1,3,5-triazines having the triazine scaffold replaced

by a pyrimidine nucleus. Based on our expertise on the func-

tionalization of the pyrimidine scaffold^[20]and taking into ac-

count the pharmacophoric requirements for the interaction

with the ATP binding site of DDX3, 30 was considered as an in-

teresting candidate to be synthesized (Scheme 6). Starting

from the tosyl derivative 24,^[20b]available as intermediate in

our lab, a microwave-assisted nucleophilic substitution with 4-

fluoroaniline afforded the C4-substituted derivative 25. The

latter compound was oxidized to sulfone 26, protected as the

TBDPS ether 27, and treated with hydrazine under microwave

irradiation to give the intermediate 28. Compound 28 was

then condensed with salicylaldehyde to give 29 which was fi-

nally deprotected with triethylamine trihydrofluoride to give

the final compound 30.

Scheme 4. Synthesis of triazine derivatives 15e–g, 16i–p. Reagents and con-

ditions: a) DME, ꢀ308C, R²BnNH₂, 3 h; b) CH₂Cl_2,R¹NH, RT, 12 h; c) 1. CH₂Cl₂,

NH₂NH₂·H₂O, reflux, 12 h, 2. toluene, isatin (for 15) or salicylaldehyde (for 16),

3 h, reflux, Dean–Stark.

Scheme 6. Synthesis of pyrimidine derivative 30. Reagents and conditions:

a) 4-fluoroaniline, DME, MW, 1508C, 5 min; b) Oxone, H₂O/MeOH/THF;

c) TBDPSCl, DMF, imidazole, MW, 1008C, 5 min; d) NH₂NH₂·H₂O, DME, MW,

1208C, 5 min; e) salicylaldehyde, toluene, molecular sieves, reflux, overnight;

f) Et₃N-3HF, THF, RT, overnight.

Scheme 5. Synthesis of triazine derivatives 16m,n, 20b,c. Reagents and con-

ditions: a) DME, ꢀ308C, 4-fluoroaniline, 3 h; b) CH₂Cl₂, morpholine, RT, 12 h;

c) 1. CH₂Cl₂, NH₂NH₂·H₂O, reflux, 12 h, 2. toluene, R³CHO, 3 h, reflux, Dean–

Stark; d) DME, morpholine, 3 h, ꢀ308C; e) CH₂Cl₂, 4-fluoroaniline, 12 h,

reflux.

Structure–activity relationships for the second-generation

DDX3 inhibitors

In addition to the compounds synthesized as described in the

previous section, SciFinder Scholar^[21]was used to search com-

mercially available analogues of 4a. As a result, many rhoda-

nine derivatives were identified and obtained from commercial

cyanuric chloride with 4-fluoroaniline at ꢀ308C always resulted

in a mixture of the disubstituted and monosubstituted deriva-

tives 17 and 18. Nevertheless, following the synthetic steps

1374

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

sources (in particular, from Asinex and ChemBridge Corpora-

tion)^[14b]to have a more complete set of derivatives for SAR

considerations (Table 1). Results from docking calculations on

rhodanine analogues suggested that the major interactions in-

volved both molecular edges that are located within the left

and right pocket of the ATP binding site on DDX3 (Figure 1A).

Moreover, the distance of ~15 ꢂ between the bromine and

the hydroxy substituents of 4a, taken as a representative com-

pound, seemed to be optimal for the interaction with the pro-

tein. An ethyl spacer (n=2 as in 4a) had the optimal length to

allow the edges of ligands to make profitable hydrophobic

and hydrogen bond contacts with the amino acids of the ATP

binding site. In fact, variation of this length was detrimental

for the activity (4a vs. 4x as well as 4u vs. 4y and 4z, Table 1).

These results suggested that R¹should be located at the right

distance of ~15 ꢂ from electron-rich substituents at R². At the

right edge of the inhibitor structure, for certain substituents at

R², replacement of the 2-OH group at R¹with a 2-NO₂(4g and

4h) and 2-COOH (4s and 4t) groups gave compounds of simi-

lar or slightly improved activity (4g vs. 4a and 4s vs. 4 f,

Table 1). The configuration of the double bond at C5 also had

a significant impact on the affinity toward the enzyme as it

can be appreciated from the low activity of the Z-derivative 4r

bearing a 2-COOH at R¹. On the contrary, the methoxy deriva-

tive 4q was inactive at the test dose (500 mm) probably be-

cause the terminal methyl group caused a steric clash with the

protein, avoiding the oxygen atom of the methoxy group to

behave as a hydrogen bond acceptor, as shown by docking

simulations. In addition, the position of the OH group on the

phenyl ring had a significant influence on the activity. In fact,

while the 2-OH derivative 4a showed a good affinity toward

the enzyme, the 3- and 4-OH analogues (4u,w) showed a dra-

matic decrease in affinity. Moreover, replacement of the 2-OH

group of 4l with a chlorine led to 4m with a significantly de-

creased activity (~40-fold).

Table 1. Structures and biological data for the new rhodanine deriva-

tives.

At the left edge, the 3-substituent at R²is able to contact

either the guanidino moiety of Arg202 or the hydrophobic

polymethylenic chain of the same residue, as well as the imida-

zole ring of His527 (Figure 1C). On this basis, the good activity

of the 3-F derivatives 4b and 4c relative to 4a is not surprising

if one takes into account the better electrostatic interactions.

Moreover, similar interactions are possible between the guani-

dino group of Arg202 and the oxygen atoms of methoxy (4e,

4l, 4m, 4p, and 4t that showed activities similar to that of 4a)

and dioxole derivatives (4 f and 4s). In fact, either such oxygen

atoms or the fluorine substituent of 4b and 4c are accommo-

dated within the same pocket, as also shown in Figure 1B. A

minimum energy point for the fluorine probe was found by

GRID software^[22](Figure 2C) in correspondence of this pocket.

Docking simulations accounted for the importance of methoxy

substituents accommodated within the left pocket of the bind-

ing site, the oxygen atom of the methoxy groups being locat-

ed within the same region occupied by the fluoro group of

4b, where electron-rich substituents should lie. Accordingly,

the unsubstituted derivatives 4i and 4k lacked the contacts

with the side chain of Arg202, thus explaining its inactivity.

The methyl group of 4d restored some activity (97 mm) be-

cause of the hydrophobic contacts with the polymethylene

chain of Arg202, while changing the substitution pattern of

the methyl group from position 3 to 4 resulted in inactive

compounds such as 4n and 4o. This detrimental effect was

probably due to the fact that the 4-methyl substituent was

pointing toward the solvated interface between DDX3 mono-

mers and, in several poses, showed steric clashes with the pol-

ymethylenic chain of Arg202.

Compd

n

R¹

R²

K_i^[a][mm]

4a^[b]

4b

4c

4d

4e

4 f

4g

4h

4i^[c]

4j^[c]

4k^[b]

4l^[b]

4m^[b]

4n^[c]

4o^[c]

4p^[b]

4q^[b]

4r^[c,d]

4s^[b]

4t^[b]

4u^[b]

4v^[b]

4w^[b]

4x^[b]

4y^[c]

4z^[b]

7a

2

3

1

3

2

–

2-OH

3-Br

3-F

3,5-diF

3-Me

5.4

0.3

0.5

97

3-OMe

3-(O-CH₂-O)-4

3-Br

10

2-OH

1.0

8.7

7.6

NA

0.1

3.9

NA

11

NA

200

0.4

2.0

90

200

300

500

NA

150

4.2

4.3

28

2-NO₂

2-OH

3-OH

2-OH

2-Cl

2-OH

2-OH, 3-Cl

2-OH, 5-Cl

2-OMe

2-COOH

3-OH

4-OH

2-OH

3-OH

3-F

H

2-Cl

H

3,4,5-triOMe

4-Me

4-OMe

3-Br

3-(O-CH₂-O)-4

4-OMe

3-Br

2-OMe

3-Br

3-F

3-Br

–

3-OH

2-OH

7b

11

31

In a similar way, the 2-Cl substituent of 4j showed steric

clashes with the backbone of Arg202 or, upon rotation of the

phenyl ring at R², with the backbone of Tyr200, resulting in an

inactive compound. SAR considerations are unable to account

for the little effect on potency observed upon replacing a 3-Br

–

0.2

[a] Apparent equilibrium dissociation constant for the inhibitor to the

target enzyme DDX3. [b] Purchased from Asinex. [c] Purchased from

ChemBridge. [d] The double bond at C5 has an E configuration.

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1375

M. Botta et al.

MED

Figure 2. Representation of the binding mode of additional triazine derivatives within the DDX3 ATP binding site. Hydrogen bond interactions are represent-

ed by yellow dashed lines. Regions of best interactions with chloride and fluoride probes are green spheres, while those with hydrophobic probes (mimicking

a methyl group) are grey spheres. A) The acetamido group of 16e (brown) is very important for activity. Lack of hydrogen bond contacts within the right

pocked leads to a decrease in activity of at least one order of magnitude (16b, pale brown). B) A very similar orientation and interaction pattern accounts for

the similar activity of the benzylidene derivative 16h (red) and the indolinone analogue 15d (green). Their 3-Cl substituent is located within a minimum

energy point (ꢀ6.97 kcalmol^ꢀ1) found by GRID for the chlorine probe. C) The difference between the benzylamino chain of 16i (orange) and the anilino chain

of 16m (magenta) is not important for activity. Their 4-F substituent is located in a region of space where a minimum for the fluoride probe (ꢀ3.49 kcalmol^ꢀ1

)

was found. D) The 2-naphthyl moiety of 20c is located within a large hydrophobic region in part corresponding to the right pocket. E) Changing the substitu-

tion pattern of the nitro group from position 4 (21a, light green) to 3 (21b, green) induces a conformational rearrangement of the terminal phenyl ring with

a consequent decrease in activity. F) The phenylethyl rhodanine moiety of 31 is reminiscent of rhodanine derivatives 4, while the hydroxypropylamino chain

at the condensed heterocyclic ring mimics the morpholine group of several triazine derivatives. Higher-resolution images are available in the Supporting In-

formation.

with a 3-F in nitro or inverted amide derivatives (4g–h and

7a–b, respectively). Moreover, replacement of the rhodanine

scaffold of 4a with a succinic moiety resulted in 11 with lower

activity. To overcome the intrinsic limitations possibly affecting

the pharmacophoric model generated by the software Li-

gandScout^[23]used to identify 4a, a high-throughput docking

approach (HTD, as described in the Experimental Section

below) was applied to the Asinex Gold collection.^[14a]As a

result, the triazine derivative 15a, endowed with submicromo-

lar activity (0.4 mm, Table 2), was identified. The best ranked in-

teraction mode within the ATP binding site of DDX3 showed

the indolinone moiety within a network of hydrogen bonds

with amino acid residues constituting the left pocket (Fig-

ure 1D). In particular, the carbonyl group of the ligand interact-

ed with the terminal portion of the Gln207 side chain, mimick-

ing the carbonyl group of the rhodanine nucleus. Moreover,

the NH of the lactam group was a hydrogen bond donor for

the carbonyl moiety of Arg202. It is important to note that the

hydrogen bond acceptor-donor motif involving Arg202 and

Gln207 was conserved in the X-ray crystal structure of DDX3,

where the adenine moiety of the nucleotide made two hydro-

gen bonds with the same residues. The heterocyclic portion of

the indoline nucleus also interacted with Tyr200 (in a similar

way already found for the rhodanine ring), while the con-

densed phenyl ring was located within the large cavity (the

left pocket) where the bromophenyl moiety of 4a was accom-

modated. An additional hydrogen bond was found between

one of the anilino NH groups and the carbonyl moiety of

Gly227 within the upper pocket. Hydrophobic contacts be-

tween the second anilino moiety and Thr231, Ile282, Ala232,

Gln281, and the alkyl spacer of Glu285 were also found. A fo-

cused library of structural analogues of 15a was synthesized,

while additional derivatives were purchased from commercial

sources to find compounds with better activity with respect to

the parent hit (Table 2). Removing the hydrogen bond donor

group by replacement of the anilino moiety of 15a with a

morpholine led to the inactive derivative 15b. Computational

results suggested that 15b could be located within the

enzyme binding site in a different orientation from that of

15a. In fact, the indoline nucleus occupied the right pocket

and interacted by hydrogen bonding with Gln281, whereas

the morpholino ring was accommodated within a region ex-

posed to the solvent, thus lacking the hydrogen bond with

Gly227 found for 15a. Finally, the anilino moiety was found

within the left pocket previously occupied by the indolinone

nucleus of 15a (Figure 1E). The lack of the hydrogen bond

donor-acceptor motif similar to that exhibited by the adenine

portion of the natural ligand and the inability of the morpho-

1376

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

Opening the indolinone ring of 15a into a benzyli-

denehydrazine chain resulted in a significant de-

crease in activity (16a–d; 33, 18, and 7.5 mm or no ac-

tivity, respectively), independently from the substitu-

ents at position R³. The sole exception was represent-

ed by the acetamido derivatives 16e–g that restored

a micromolar or submicromolar activity. Such com-

pounds restored the hydrogen bond with Gly227 of

the upper pocket and maintained the hydrogen

bonds involving their NH amide group and Gln281.

This contact was disrupted by transformation of the

acetamido terminus into hydroxy or halogen sub-

stituents, as in 16a–d (Figure 2A). Compound 16g,

although being able to maintain contacts mediated

by its acetamido group, was less active than 16e be-

cause of its inability to interact with Gly227 of the

upper pocket by means of one of its anilino moieties.

The benzylidene derivative 16h showed an activity

similar to that of the corresponding indolinone ana-

logue 15d (1.6 vs. 3.2 mm, respectively), accounted

by a very similar interaction pattern with the binding

site (Figure 2B). The 3-Cl and 4-Me substituents of

these compounds were accdommodated within mini-

mum point regions of the left pocket found by GRID

for the chloride and methyl probes, respectively,

while the phenyl ring was at the right distance for a

possible p–p interaction with the aromatic portion of

Tyr200.

Table 2. Structures and biological data for the new triazine derivatives.

Compd

n

R¹

R²

R³

K_i^[a][mm]

15a^[b]

15b^[b]

15c^[b]

15d^[b]

15e

0

1

0

1

0

1

0

1

0

–

NH-Ph

morpholinyl

NEt₂

NH-Ph

H

–

0.4

NA

1.6

3.2

NA

2.9

0.1

33

3-Cl

3-Cl, 4-Me

H

3-Cl

15 f

15g

–

H

16a^[b]

16b^[b]

16c^[b]

16d^[b]

16e^[b]

16 f^[b]

16g^[b]

16h^[b]

16i

2,4-OH

3,4-diCl

3-NO₂, 4-OH

4-Br

4-NHCOCH₃

2-OH

2-OH, 5-Cl

2-OH, 3-NO₂

2-OH

18

7.5

NA

0.3

0.5

2.2

1.6

0.7

0.6

1.9

4.0

0.4

0.1

62

0.5

0.2

86

8.5

0.3

3.6

3.2

0.4

9.0

NH-Ph

morpholinyl

NH-Ph(4-F)

morpholinyl

NEt₂

morpholinyl

piperidinyl

–

3-Cl, 4-Me

Keeping fixed the morpholine and the 2-OH group

at R¹and R³, respectively, the activity was neither af-

fected by decreasing the size of the substituent at R²

(16i), by eliminating it (16j–l), nor by shortening of

the benzylamino chain of 16i to the anilino moiety

of 16m. Docking simulations showed that the 2-OH

group of 16m and 16i was located within the right

pocket and interacted with the protein by means of

the same hydrogen bonds already found for 4a,

while the fluorine atom on their anilino or benzylami-

no groups occupied a region where electron-rich

substituents are preferred and where an additional

minimum interaction point (ꢀ3.49 kcalmol^ꢀ1) was

found for the fluorine probe. The same region also

accommodated the fluoro substituent of 16n. These

results also explained the fact that a benzylamino

chain instead of an anilino group (16i and 16m) did

not affect activity. Transforming the morpholino

4-F

H

16j

16k^[b]

16l^[b]

16m

16n

H

4-F

3-Cl

H

4-F

H

2-OH

16o

16p

20a^[b]

20b

2-methyl-indol-3-yl

1-naphthyl

2-naphthyl

2-Cl, 4-NO₂

2-Cl, 5-NO₂

2-Cl

20c

21a^[b]

21b^[b]

21c^[b]

21d^[b]

30

H

–

4-Cl

–

[a] Apparent equilibrium dissociation constant for the inhibitor to the target enzyme

DDX3. [b] Purchased from Asinex.

line ring to make the additional hydrogen bond contact with

Gly227 seemed to be responsible for the dramatic decrease in

activity of unsubstituted morpholino derivatives 15b and 15e.

A significant activity was restored by introduction of a chloride

substituent at position 3 (15c, 15d, and 15 f, Figure 1F) that

was accommodated within a best interaction region of the left

pocket, as identified by GRID software for the chloride probe.

The same trend in activity was found for compounds bearing a

benzylamino side chain instead of the anilino substituent (15e

and 15 f).

moiety of 16j into the diethylamino side chain of 16p did not

influence the activity, contrarily to that found for 15e and

15g.

Finally, replacement of the phenolic moiety of 16m with an

indole nucleus (20a) retained a micromolar activity, while bulk-

ier susbtituents (1- and 2-naphthyl) led to 20b and 20c with

an activity of 86 and 8.5 mm, respectively. The 2-naphthyl ring

showed hydrophobic contacts with a lipophilic region defined

by Thr226, Lys230, Leu278, and Phe234, where GRID found

several minimum points for a C3 (methyl) probe, while its 4-F

substituent was located within the minimum energy point for

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1377

M. Botta et al.

MED

the fluoride at the left pocket (Figure 2D). However, changing

the substitution pattern of the naphthyl ring was detrimental

for activity because of steric clashes found for the 1-naphthyl

analogue 20b.

SmiLib v2.0.^[24]In particular, a series of aldehydes and primary

amines available in our stockroom, were combined to generate

~16000 virtual compounds. The library was then submitted to

a virtual screening protocol with the aim of identifying com-

pounds characterized by a predicted affinity for the ATP bind-

ing pocket and an aqueous solubility (QP LogS - conformation

independent) better than that of 4a. Compound 31 was classi-

fied as the best ranked entry and purchased from commercial

sources. Docking simulations showed that its pose within the

ATP-binding site (Figure 2F) was reminiscent, at the same time,

of the binding mode of both rhodanine and morpholino-tria-

zine derivatives, such as 4a and 15 f (Figure 1A and Figure 1F,

respectively). In fact, the phenylethyl-rhodanine moiety of 31

resembled the left molecular portion of rhodanine derivatives

4, and the hydroxypropylamino chain mimicked the morpholi-

no moiety and interacted with Lys288, as the morpholine

oxygen of triazine derivatives did. Finally, the carbonyl group

of the condensed heterocycle was able to make hydrogen

bond contacts already found for the 2-OH group of 4a (Fig-

ure 1A) and 16h (Figure 2B). Biological evaluation of 31 result-

ed in good antienzymatic (Table 1) and antiviral activity

(Table 3), with a selectivity index higher than 10.

When the condensed nucleus was replaced by a phenylfur-

anyl chain as in 21a, activity was restored at the micromolar

concentration. The chloro and nitro substituents of 21a were

found within minimum energy regions identified by GRID for

the corresponding probes (Figure 2E). Consequently, changing

the substitution pattern (21b) or eliminating the nitro group

(21c) caused a decrease in activity of about one order of mag-

nitude. In fact, the 2-chloro substituent of 21b occupied a

minimum region for the chloride different from and less profit-

able than that found for the 2-Cl group of 21a, thus explaining

the lower activity. Simplification of the morpholine into a pi-

peridine as in 21d resulted in a good activity (0.4 mm), al-

though it is not possible to assess the individual impact of

these two simultaneous changes (i.e., morpholino vs. piperidi-

no and 2-Cl vs. 4-Cl) on activity.

Taking into account the above described pharmacophoric in-

teractions with the ATP binding pocket of DDX3, we tried to

introduce such key features into a different scaffold. Among

the designed compounds, the pyrimidine derivative 30 was

synthesized and showed an activity (9.0 mm) similar to that of

4a (5.4 mm) and 16h (1.6 mm). It

had an interaction pattern simi-

lar to that of AMP, with hydro-

Table 3. Cytotoxicity and antiviral activity of DDX3 inhibitors.

gen bond interactions involving

the left and right pockets (resi-

dues Arg202, Lys230, and

Thr231, respectively), a possible

p–p interaction with Tyr200 and

additional hydrogen bond con-

tacts with Glu285 and Tyr200.

However, the loss of important

hydrophobic interactions with

the right pocket accounts for its

activity being lower than that of

similar triazine analogues, such

as 16m. The pyrimidine deriva-

tive 30 represents, however, a

promising hit for future optimi-

zation.

Compd

EC₅₀^[a][mm]

CC₅₀^[b][mm]

MOLT-4

SI^[c]

MOLT-4

HeLa

200

PBMCs

150

HeLa

PBMCs

4a

4b

4l

4t

83ꢁ7

200

2.4

6.7

–

2.0

–

1.0

10

2.4

6.7

–

1.5

–

2.0

5.0

1.8

6.7

–

4.0

–

4.0

7.5ꢁ0.7

50ꢁ5

7.5ꢁ0.5

10ꢁ1

7.5ꢁ0.7

7.5ꢁ0.5

20ꢁ3

5ꢁ1

50ꢁ10

50ꢁ5

–

15c

15d

15 f

15g

16b

16e

16g

16h

16i

16j

16k

16l

16m

16n

16p

20a

21b

21c

21d

4c

100ꢁ10

20ꢁ2

7.5ꢁ0.5

10ꢁ1

20ꢁ2

2.5ꢁ0.3

20ꢁ2

5ꢁ1

10ꢁ1

15ꢁ2

40ꢁ6

–

Finally, considering the lower

toxicity profile of the rhodanine

derivatives (4a and 4b, Table 3)

relative to the triazine deriva-

tives (16e, 16m,n, and 20a,

Table 3), we focused our atten-

tion on the first class of com-

pounds in the attempt to identi-

fy analogues endowed with an

2.5ꢁ0.1

5ꢁ1

10ꢁ1

8.0ꢁ0.5

2.0ꢁ0.2

–

4.0ꢁ0.5

25ꢁ3

7.5ꢁ0.7

20ꢁ3

>50

17.9

>50

20ꢁ2

–

15ꢁ2

6.3

–

5

–

3.7

–

ND^[d]

ND

>50

15.0

21.4

11.7

4s

15a

16 f

21a

31

4.7

65ꢁ5

improved

selectivity

index.

6.5ꢁ0.7

70ꢁ7

10

10.7

Based on the synthetic approach

reported in Scheme 2, a virtual li-

brary of substituted rhodanines

was designed using the software

[a] Toward HIV-1, the micromolar concentration required to suppress 50% of HIV-1 proliferation. [b] The micro-

molar concentration required to suppress 50% of cell viability. [c] Selectivity index: the ratio of cytotoxicity

(CC₅₀) to antiviral activity (EC₅₀). [d] ND: not determined.

1378

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

Antiviral activity of second-generation DDX3 inhibitors

Table 4. Selectivity of inhibitor 16n toward various ATP-hydrolysing en-

zymes.

The first-generation DDX3 inhibitor 4a displayed very low cy-

totoxicity on HeLa cells (CC₅₀> 200 mm) but only modest anti-

HIV activity (EC₅₀=83 mm, Table 3). Figure 3A shows the antivi-

ral activity of 4a on PBMCs infected with HIV-1 in comparison

with the most potent second-generation derivatives 4b, 16e,

16m, 16n and 20a. It can be appreciated that the last com-

pounds showed between 40 and 60% decrease in viral load al-

ready at 3 mm, while 4a had to be added at 50 mm to show

similar effects. Dose–response curves were generated for 4a

and 16n. As shown in Figure 3B, 16n showed a 41-fold higher

antiviral potency on PBMCs infected with HIV-1 (EC₅₀=2 mm)

than the protoype DDX3 inhibitor 4a. Similar curves yielded

EC₅₀values of 7.5, 10, 2.5, and 4.0 mm, respectively, for 4b, 16e,

16m, and 20a. To evaluate the specificity of action, the most

active compound 16n was tested against different ATP-hydro-

lysing enzymes (Table 4). Compound 16n showed no activity

toward T4 PNK, HCV NS3 ATPase/helicase, and c-Src human ty-

rosine kinase up to 100 mm. When tested against another

member of the DDX family of ATPase/helicases, DDX1, com-

pound 16n proved to be 500-fold less active than its activity

Enzyme^[a]

K_i^[b][mm]

Selectivity

T4 PNK

HCV NS3

c-Src

DDX1

DDX3

>100

49ꢁ4

0.1ꢁ1

>1000

490

1

[a] T4 PNK: phage T4 polynucleotide kinase, HCV NS3: human hepatitis C

virus NS3 ATPase/RNA helicase, c-Src: human Src tyrosine kinase, DDX1:

human ATPase/RNA helicase DDX1, DDX3: human ATPase/RNA helicase

DDX3. [b] Values are the mean of three independent measurements;

assays were carried out as described in the Experimental Section.

against DDX3, further confirming its high selectivity for the de-

sired target.

Cytotoxicity studies on second-generation DDX3 inhibitors

Selected compounds were also tested against the HeLa cell

line for their cytotoxic activity. Cell viability was measured

using the MTS-based CellTiter assay. CC₅₀values (dose required

to suppress 50% of cell viability) varied significantly among

the considered compounds (Table 3), with several molecules

(4a, 4b, and 16b) displaying very low cytoxicity (CC₅₀>

50 mm), while others were endowed with potent antiprolifera-

tive activity (CC₅₀ꢂ5 mm). There was no clear correlation be-

tween the cytotoxic and antienzymatic potencies: for example,

compounds 4b and 16m display almost equal potencies

against DDX3, but 16m is 20-fold more toxic than 4b.

These data suggest that cytoxicity and antienzymatic activi-

ties could, in principle, be uncoupled through chemical modifi-

cations of the pharmacophore. Cytotoxicity of compounds en-

dowed with the highest anti-HIV activity (4a, 4b, 16e, 16m,

16n, 20a, and 31) was also evaluated on two additional cell

types: the T-lymphocytic MOLT-4 cell line and the primary

PBMCs used for the anti-HIV assays (Table 3). Values of the se-

lectivity index (defined as the ratio PBMC CC₅₀/HIV EC₅₀) for

4b, 16e, 16m, 16n, and 20a on PBMCs were 6.7, 4.0, 4.0, 4.0,

and 3.7, respectively, clearly indicating that the relative cyto-

toxicity was still an important issue limiting the efficacy of

these compounds. The gain in antiviral activity obtained by

the synthesis of 4b, 16e, and 16n (HIV EC₅₀ranging from 8- to

41-fold better with respect to first-generation compound 4a)

was only about twofold higher than the corresponding in-

crease in cytotoxicity (CC₅₀from 5- to 20-fold). Thus, SAR stud-

ies were conducted in order to improve the selectivity of our

inhibitors: compound 31, identified out of a library of ~16000

virtual compounds, showed a good antienzymatic and antiviral

activity with a selectivity index higher than 10 (Table 1 and

Table 3).

Figure 3. Antiviral activity of DDX3 inhibitors. A) Comparison of the antiviral

activity (expressed as percent viral RNA relative to control) on infected

PBMCs of a fixed dose of selected inhibitors. Measurements were performed

at 72 h post-infection. Values are the means plus SD of three independent

replicates. White bar: control without drugs; dark-grey bar: postive control

with AZT (0.01 mm); black bar: first-generation inhibitor 4a (50 mm); light-

grey bars: second-generation DDX3 inhibitors 4a, 4b, 16e, 16m, 16n, and

20a (3 mm). B) Dose–response curves for the inhibiton of HIV-1 replication

(expressed as percent viral RNA relative to control) in infected PBMCs by 4a

Effect of DDX3 knockdown on cell growth

~

*

(

, EC₅₀=83ꢁ7 mm) and 16n ( , EC₅₀=2ꢁ0.2 mm). Measurements were

Because optimized compounds all showed some degree of cy-

totoxicity, we revisited the target. Validating a cellular enzyme

performed at 72 h post-infection. Values are the means ꢁSD of three inde-

pendent replicates.

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1379

M. Botta et al.

MED

as a potential antiviral target requires a precise knowledge of

the potentially negative effects of its inhibition at the host cell

level. Since the first report of DDX3 as a cofactor for HIV-1 rep-

lication, studies addressing the consequences of its knockdown

in human cells have led to conflicting results. To implement a

rigorous knowledge-based approach for the understanding

and improvement of the toxicity profile of our compounds, we

have set up a carefully controlled cellular system to examine

the effects of DDX3 knockdown in HeLaCCR5 cells on cell

growth and viral replication.

Stable DDX3 knockdown HeLaCCR5 cell lines were generat-

ed using a lentiviral vector expressing two copies of specific

microRNA targeting DDX3. As a control, a vector with four spe-

cific mismatch mutations in the hairpins was generated. As

shown in Figure 4, the polyclonal knockdown cell line showed

Figure 5. Effect of DDX3 knockdown on cell growth. For growth curve analy-

sis, 50000 cells of the polyclonal knockdown cell line (poly KD), the mis-

match control cell line (MM), and the two monoclonal cell lines (Cl54 and

Cl56) were seeded at day zero in a 24-well plate. Cell growth was monitored

by counting the cells at the indicated time points. Values are the mean of

triplicate measurements in a representative experiment. Error bars represent

the corresponding standard deviations. The experiment was performed

twice.

in HIV-1 replication, as part of our target validation process. In

the initial report on DDX3^[10]the effect on HIV-1 was analysed

in a reporter system measuring Rev-mediated viral RNA (vRNA)

export. In this assay, a CMV-driven HIV-1 gag–pol construct

containing an HIV-1 Rev response element (RRE) sequence (Fig-

ure 6A) was co-transfected with a HIV-1 Rev expression con-

struct. Upon expression, Rev binds to the RRE enabling export

of non-spliced gag–pol coding vRNA. This results in gag-de-

rived p24 expression which can be measured in the superna-

tant. The same construct with a CTE (Mason-Pfizer monkey

virus constitutive transport element) instead of an RRE was

used as a control. Although much less active, this CTE enables

Rev-independent mRNA export.

Figure 4. Western blot analysis of DDX3 knockdown cell lines. Stable DDX3

knockdown HeLaCCR5 cell lines were generated using lentiviral vector tech-

nology. DDX3 levels in the polyclonal knockdown cell lines (poly KD), the

mismatch control cell line (MM), and the two strongest monoclonal cell lines

(Cl54 and Cl56) were analysed by Western blot using an anti-DDX3 antibody.

Equal loading was controlled by a-tubulin detection.

at least 50% knockdown relative to the mismatch control. Mor-

evoer, monoclonal cell lines were derived to achieve a better

knockdown. Among the two best cell lines (clones 54 and 56),

clone 54 showed ~75% knockdown relative to the mismatch.

This assay was used to evaluate the effect of DDX3 knock-

down on the RRE-Rev axis in the DDX3 knockdown cell lines

(Figure 6). Although we could not observe an effect in the pol-

yclonal cell lines in comparison with the mismatch control, the

monoclonal cell lines showed a 3- to 10-fold defect in p24 pro-

duction. Expression of the gag–pol RRE construct in the ab-

sence of Rev did not result in detectable levels of p24. In addi-

tion, as a control, a plasmid encoding eGFP driven by the CMV

promoter was co-transfected. No significant difference in eGFP

expression was observed excluding effects on expression or

Rev-independent mRNA export (results not shown). These re-

sults confirm the previously published results^[10]and corrobo-

rate the involvement of DDX3 in the RRE-Rev axis.

Evaluation of the cellular phenotype in DDX3 knockdown

cells

Following stable suppression of DDX3, the cellular phenotype

was studied with respect to proliferation potential. Equal num-

bers of the mismatch, polyclonal, and monoclonal DDX3

knockdown cells were seeded and cell growth was monitored

over time. As shown in Figure 5, knockdown of DDX3 did not

affect cell growth. Because the knockdown was not complete,

these results at least indicated that low levels (25% of normal)

of DDX3 might be sufficient to sustain cell growth. While Lee

and co-workers^[25]reported a strong effect on cell viability

upon siRNA-mediated DDX3 knockdown in HeLa cells and Dro-

sophila cells, our results confirm the observations of Ishaq and

co-workers.^[8]

Effects of DDX3 knockdown on virus production

If DDX3 is indeed essential for the RRE-Rev axis, knockdown of

DDX3 should hamper virus production. To test this hypothesis,

HeLaCCR5 DDX3 knockdown cells were transfected with a HIV-

1_NL4-3molecular clone. The p24 content of the supernatant was

measured 48 h post transfection and compared with the mis-

match cell line (Figure 7). A twofold decrease in p24 produc-

tion could be observed in the DDX3 knockdown cell lines. This

effect was less significant than that obtained by Yedavalli and

Effect of DDX3 knockdown on the HIV-1 RRE-Rev viral RNA

export pathway

Having established a stable cell line effectively depleting

DDX3, we used it as a tool to re-investigate the role of DDX3

1380

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

Figure 7. Effect of DDX3 knockdown on virus production. HeLaCCR5 DDX3

polyclonal knockdown cells and the monoclonal derived cells (clone 54 and

56) were transfected with a HIV_NL4-3molecular clone. The p24 content of the

supernatant was measured 48 h post-transfection and compared with the

mismatch (MM) cell line. Values are the mean of two measurements in a rep-

resentative experiment. Error bars represent the corresponding standard de-

viations. The experiment was done twice.

we report a rational drug design and full characterization of

second-generation DDX3 inhibitors endowed with nanomolar

antienzymatic activity in vitro and low-micromolar anti-HIV ac-

tivity in infected cells. In addition, we have conducted a thor-

ough analysis that confirmed DDX3 as a valid anti-HIV target.

The compounds described here represent a significant ad-

vancement in the quest of potential novel drugs targeting HIV-

1 host cofactors. The observed toxicity is still significant, but

the results presented in this work support the possibility to

design more specific DDX3 inhibitors as effective anti-HIV

agents.

Figure 6. Role of DDX3 in the Rev-RRE mRNA export axis. A) Schematic rep-

resentation of the Rev reporter system. To measure Rev-dependent mRNA

export, the reporter plasmid was co-transfected with pcRev and a plasmid

encoding eGFP to control for transfection efficiency. The amount of p24 in

the supernatant was measured 48 h post-transfection. B),C) The different He-

LaCCR5-derived DDX3 knockdown cell lines (poly KD and monoclonal 54

and 56) were cotransfected with the gag–pol RRE reporter construct and a

Rev and eGFP expression plasmid. Panel B shows the p24 content in the su-

pernatant 48 h post-transfection. Panel C shows the eGFP transfection effi-

ciency as measured by FACS analysis. Values are the mean of triplicate meas-

urements in a representative experiment. The experiment was performed

twice. Error bars represent the corresponding standard deviations.

Experimental Section

Biology

Enzymatic assays: Recombinant His-tagged human DDX3 was ex-

pressed in E. coli and purified as described.^[12]ATPase activity was

tested with a luciferase-based luminescence assay (easylite-Kinase,

PerkinElmer) on 96-well microtiter plates as described.^[11]Briefly, re-

combinant purified human DDX3 (50–100 ng) was incubated in a

15 mL total reaction volume with reaction buffer (25 mm Tris-HCl,

pH 7.5, 5 mm MgCl₂), in the presence of increasing amounts of ATP

and/or different combinations of ATP and the inhibitor to be

tested. Reference curves in the absence of inhibitor (giving the

100% of enzymatic activity) and in the absence of enzyme (giving

the baseline), were also included in each experiment. Reading was

performed with a Microbeta Trilux (PerkinElmer) luminometer, ac-

cording to the manufacturer’s protocol. T4 PNK (OptiKinase) was

from USB Corp. (Cleveland, OH, USA). Activity was assayed accord-

ing to the manufacturer’s protocol in the absence or presence of

increasing amounts of 16n. HCV NS3 was purified and its ATPase

activity assayed as described^[26]in the absence or presence of in-

creasing amounts of 16n. Human c-Src was purchased from Up-

state (Lake Placid, NY, USA). Activity was assayed as described^[27]in

the absence or presence of increasing amounts of 16n. The cDNA

for human DDX1 was kindly provided by Dr. U. Dietrich (Georg-

Speyer-Haus, Institute of Biomedical Research, Frankfurt, Germany).

Human DDX1 was cloned from human PBMCs and purified as de-

co-workers.^[10]Possibly, the knockdown in our cell lines might

not be strong enough to detect a more pronounced replica-

tion defect. In this regard, Yedavalli^[10]obtained a stronger

knockdown and a stronger defect in the Rev reporter system.

However, our data collectively confirm a role of DDX3 in the

HIV-1 replication cycle, thus validating it as a novel anti-HIV

target. We cannot exclude that the toxicity of our compounds

might be, at least in part, also due to targeting other DEAD-

box helicases. For the future design of more specific DDX3 in-

hibitors, starting from the most active compounds identified

here, we are now conducting a counterscreening for lack of in-

hibition of other helicases, such as DDX1.

Conclusions

The problem of drug resistance development due to mutations

in HIV-1 proteins targeted by antiviral drugs could be over-

come by the development of novel anti-HIV agents targeting

host factors essential for viral replication. One of such host fac-

tors is the DEAD-box RNA helicase/ATPase DDX3.^{[5,8,10,11,15]}Here

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1381

M. Botta et al.

MED

scribed above for DDX3. DDX1 ATPase activity was tested under

the same conditions used for DDX3, in the absence or presence of

increasing amounts of 16n.

24 h. After 72 h of continuous exposure of the cells to the com-

pounds at 378C, cell viability was measured with the CellTiter 96

(Promega) viability assay, which contains MTS [3-(4,5-dimethylthia-

zol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazoli-

um, inner salt] and an electron coupling reagent (phenazine etho-

sulfate). Assays were performed by adding 20 mL of the CellTiter 96

reagent directly to culture wells, followed by incubation for 4 h

and subsequent absorbance measurement at 490 nm with a 96-

well plate reader. The quantity of formazan product as measured

by the amount of 490 nm absorbance is directly proportional to

the number of living cells in culture. Determinations for each com-

pound dose were done in quadruplicate. The CC₅₀values were cal-

culated from dose–response curves using the mean values of each

set of determinations.

Kinetic analysis: To determine the inhibitory constant (K_i), ATPase re-

actions were performed as described, in the presence of increasing

amounts of inhibitor and variable ATP concentrations. Variations of

the initial velocities of the reaction as a function of ATP concentra-

tions in the absence or presence of inhibitors were analysed with

Equation (1), in which k_catis the reaction rate, E₀is the input

enzyme concentration, [ATP] is the variable substrate concentra-

tion, and K_Sis the apparent affinity for the substrate. K_S=K_Min the

absence of the inhibitor, whereas K_S=K_M(app)in the presence of the

inhibitor.

K_i¼ k_catE₀=f1 þ ðK_S=½ATPꢃÞg

ð1Þ

Plasmids and vector production: pcRev was a kind gift from M.

Malim (London, UK). pCMV NL gag–pol RRE was a kind gift from E.

Freed (Frederick, MD). Lentiviral vectors encoding DDX3 microRNA

were essentially cloned as described before.^[29,30]The DDX3 target

sequence was 5’-GAG GAA CAT AAA TAT TAC TAA-3’. As a mismatch

control, the sequence 5’-GAG GAA CAT ATT ATT TAC TAA-3’ was

used. Lentiviral vectors were prepared as described previously.^[31]

Variations of the percent inhibition as a function of compound

concentrations at different ATP concentrations were fitted to the

equation for cooperative binding:

n

% inhibition ¼ V_max½Iꢃ =ðK_iþ ½Iꢃ Þ

ð2Þ

Cell culture and generation of DDX3 knockdown cells: HeLaP4-CCR5

cells (a gift from P. Charneau, Paris, France) were grown in Dulbec-

co’s modified Eagle’s medium (Gibco-BRL, Merelbeke, Belgium)

supplemented with 10% fetal calf serum (Sigma, Bornem, Belgium)

and 20 mgmL^ꢀ1gentamicin (Gibco-BRL) at 378C and 5% CO₂in a

humidified atmosphere. To generate the DDX3 knockdown and

control cells, cells were transduced with the DDX3 microRNA en-

coding lentiviral vector or mismatch control vector and selected

with zeocin (200 mgmL^ꢀ1; Invitrogen, Merelbeke, Belgium).

for which V_maxis the apparent maximal rate of the reaction in the

absence of the inhibitor, [I] is the variable inhibitor concentration,

n is the cooperativity index, and K_iis the apparent inhibition con-

stant.

Antiviral assays: The susceptibility of HIV-1 recombinant strains to

drugs was performed as previously reported.^[28]In brief, 0.5 mg of

each HIV-1 plasmid construct was transfected into CD4⁺HeLa cells

by using the Lipofectin Transfection Reagent (Invitrogen, Gronin-

gen, The Netherlands) according to the recommendations of the

manufacturer. After 3 days of incubation at 378C, the cell superna-

tants, which contained reconstituted viable recombinant viruses,

were collected. Quantification of the newly produced recombinant

strains was done by determination of the HIV RNA copy number in

the cell culture supernatants. In detail, transfected HeLa CD4⁺cell

culture supernatants containing 1ꢃ10⁹RNA copies mL^ꢀ1were

used to infect aliquots of 20ꢃ10⁶phytohemagglutinin-stimulated

peripheral blood mononuclear cells (PBMCs) obtained from HIV-se-

ronegative blood donors. After 4 h of incubation, supernatants

were removed and infected PBMCs were incubated at 378C in

10 mL of RPMI 1640 medium (Eurobio, Les Ulis Cedex B, France)

supplemented with 20% fetal calf serum (Life Technologies, Ltd.,

Paisley, UK), 2 mmL-glutamine, 100 UmL^ꢀ1penicillin, 100 mgmL^ꢀ1

streptomycin, 50 UmL^ꢀ1interleukin-2 (Roche Diagnostics, Mann-

heim, Germany), 5 mgmL^ꢀ1hydrocortisone (Sigma Chemical Co.,

St. Louis, MO, USA), and with fourfold dilutions of antiretroviral

drugs. A no-drug control for each drug dilution was included in

each assay. The HIV-1 RNA level in the cell culture supernatant was

quantified at 72 h post-infection. Recombinant HIV-1 strains from

treatment-naive patients and multidrug resistance-associated

changes were assayed in parallel. The degree of inhibition of viral

replication was measured by determining the HIV-1 RNA level in

the supernatants of cell cultures and was expressed as the fold in-

crease in the 50% inhibitory concentration (IC₅₀) for resistant re-

combinant HIV-1 variants, compared with the IC₅₀for the wild-type

recombinant variant. Each test was performed in triplicate.

Western blot analysis: SDS (1%) protein extracts were separated by

sodium dodecyl sulfate-polyacrylamide gel electrophoresis. DDX3

was detected using a DDX3 antibody (Imtec, Belgium). Equal load-

ing was controlled by a-tubulin detection using anti-a-tubulin anti-

body (Sigma, Belgium). Proteins were visualized by chemilumines-

cence (ECL Plus; Amersham Biosciences, Uppsala, Sweden).

Analysis of DDX3 knockdown cells: To analyse the effect of DDX3 on

the RRE-rev axis, 100000 cells per well of the respective cell lines

were plated in a 24-well plate. At 24 h post-seeding, cells were

transfected with 1 mg pCMV NL gag–pol RRE, 0.5 mg pcRev, and

0.5 mg peGFP-C1. Forty-eight hours later, p24 content in the super-

natant was measured using the Alliance HIV-1 P24 ANTIGEN ELISA

Kit (PerkinElmer, Zaventem, Belgium), and eGFP expression was an-

alysed by FACS. To analyse viral production, cells were transfected

with pNL4-3 (NIH AIDS Research and Reference Reagent Program)

and analysed for p24 content at 48 h post-transfection.

Computational methods

The structure-based pharmacophore modeling and molecular

docking protocol that was used to identify 4a was previously de-

scribed.^[11]In brief, the pharmacophore was generated by means of

LigandScout, a software able to codify the crucial interactions be-

tween DDX3 and the natural ligand AMP (in the pdb complex

coded as 2I4I) into pharmacophoric features.

Importantly, a kinetic characterization of 4a suggested it as an un-

competitive inhibitor. Considering that uncompetitive inhibition

occurs when the inhibitor binds only in the presence of ATP and

that DDX3 may act as a multimeric enzyme, 4a could bind in the

ATP binding site of one or more monomers of the multimeric

DDX3 complex, while ATP binds to the other monomers. As this

Cytotoxicity assays: Compounds were tested against the human

cervical cancer cell line HeLa, the human T-lymphocytic leukemia

cell line MOLT-4, and human primary PBMCs. Increasing doses of

the compounds to be tested were added to 2ꢃ10⁶cells. Culture

medium supplemented with fresh compound was replaced every

1382

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

model was consistent with the observed kinetic behavior of 4a,

we chose to dock the new compounds within the ATP binding site

of a DDX3 monomer. However, we are aware of the fact that an al-

ternative binding mode for the new compounds could be based

on DDX3 sites that are different from the ATP binding pocket.

assays. As a result, the triazine derivative 15a showed a submicro-

molar IC₅₀value (0.4 mm).

SciFinder Scholar was then applied to search for commercially

available derivatives of both 4a and 15a. Resulting compounds

were docked into the DDX3 binding site and best ranked entries

were purchased from Asinex and ChemBridge.^[14]

The model resulting from LigandScout was exported to Catalyst

software^[32]and then used as a three-dimensional query to perform

a search within a database of commercially available compounds.

In particular, the Asinex Gold and Synergy databases were down-

loaded from the web and converted into a multiconformer data-

base by means of the catDB command of Catalyst (FAST method,

250 as the maximum number of conformations for each entry).

Catalyst was also applied to filter the database with the model

(fast flexible search routine), keeping as putative hits all the data-

base entries with a fit value higher than 1.00. The resulting com-

pounds were exported to Cerius2 (version 4.10L)^[33]to calculate

their physicochemical properties and remove entries unable to sat-

isfy Lipinski’s rule of five, thus leading to 70 entries. They were sub-

mitted to docking simulations and scoring, allowing us to choose

10 compounds including 4a.

Details of high-throughput docking (HTD)

The pharmacophoric model used to identify 4a could suffer from

intrinsic limitations affecting all the structure-based pharmacophor-

ic models. In fact, it was built on the basis of the interactions be-

tween the natural ligand AMP and its binding site within the struc-

ture of DDX3 that were codified by the software LigandScout into

pharmacophoric features. The application of such a model as a

three-dimensional filter to screen in silico databases of compounds

results, by definition, in the identification of compounds able to

satisfy the geometrical constraints imposed by the pharmacophoric

features of the model. As a consequence, compounds able to par-

tially fit the model will be ranked with a low fit score and, proba-

bly, discarded. However, although these compounds are unable to

fully mimic the interactions involving AMP, they could be in princi-

ple able to occupy the AMP binding site by means of interactions

involving amino acid residues not interacting with AMP itself. On

the basis of these considerations, with the purpose of finding com-

pounds able to be accommodated within the AMP binding site of

DDX3 and thus to potentially interfere with its enzymatic activity, a

high-throughput docking approach (HTD) was also applied to the

Asinex Gold collection. The computational protocol was based on

a preliminary database optimization consisting in the application

of a filter based on the rule of five that allowed to prune ~20% en-

tries. Moreover, a rule-based approach was also applied to enumer-

ate all the possible tautomers and protonation forms of each com-

pound that were in turn individually docked by means of GOLD

software toward the binding site of DDX3. Compounds classified

as best scored by a consensus scoring approach based on both

GoldScore and ChemScore scoring functions were selected to be

further analysed. Visual inspection, chemical diversity, and commer-

cial availability of the best ranked entries led us to choose ten

compounds to be purchased and submitted directly to biological

assays. Among them, a hit compound belonging to the class of tri-

azine derivatives (15a), endowed with submicromolar activity

(0.4 mm, Table 2), was identified. Compounds’ binding poses result-

ing from docking simulations were also supported by GRID calcula-

tions performed to search minimum interaction points between

several chemical probes and the ADP binding pocket on DDX3.

GRID software was also applied with the aim of taking into consid-

eration, at least in part, DDX3 flexibility by setting to 1 the Direc-

tive MOVE of the program. In fact, when MOVE=1, flexible side

chains tend to move toward the probes to facilitate hydrogen

bond and electrostatic interactions. This setting was suggested by

the presence of several flexible amino acids within the DDX3 AMP

binding site, such as Arg202, Lys230, and Lys288.

In a different approach, to overcome the intrinsic limitations affect-

ing the structure-based pharmacophoric model, a virtual screening

protocol based on a high-throughput docking approach (HTD) was

also applied to the Asinex database in an attempt to find addition-

al small molecules able to bind the ATP binding site of DDX3. A

rule-based approach was applied to enumerate all the plausible al-

ternative tautomeric forms of each database entry, that were in

turn docked into the DDX3 binding site by means of the GOLD

software. Considering that no known small molecules able to bind

the ATP pocket of DDX3 are currently available, the reliability of

the docking approach and the performance of the scoring function

in finding active entries within a database cannot be checked a pri-

ori. The only way to validate the entire virtual screening protocol

was through the identification of putative hit compounds and

their submission to biological evaluation. On the basis of this con-

siderations and taking into account that the DDX3 ATP binding site

showed either hydrophobic clefts or residues with hydrogen bond

capabilities, both the scoring functions of GOLD were applied:

GoldScore usually recommended to score ligand binding mainly

based on hydrogen bond contacts, and ChemScore preferred for

evaluating interactions mainly attributable to hydrophobic groups

and residues. Default settings of the GOLD genetic algorithm were

applied.

The subunit of DDX3 complexed with the ligand AMP was extract-

ed from the entire crystallographic structure 2I4I and submitted to

a preliminary structural optimization to avoid any possible steric

clash between the ligand and the protein (MacroModel software,^[34]

OPLS_2005 force field,^[35]Polak–Ribiere conjugate gradient, and a

100 kJmol^ꢀ1ꢂ^ꢀ2as a constraint constant applied to backbone

atom; 0.01 kJꢂ^ꢀ1mol^ꢀ1as the energy gradient root mean square

to converge). The region included in a 10 ꢂ shell from AMP was in-

cluded in docking calculations. For each ligand, 100 independent

genetic algorithm (GA) runs were performed, with a maximum of

100000 GA operations on a set of five islands with a population

size of 200 individuals. The remaining GA parameters were kept to

their default values. For each compound, a cluster analysis was

then performed and the best scored conformation of the most

populater cluster was kept as the binding conformation. Chem-

Score and GoldScore scoring functions were independently used

to score interactions between ligand and target. Upon application

of consensus scoring, a visual inspection of remaining entries led

to ten selected compounds that were submitted to biological

Chemistry

General: All commercially available chemicals were used as pur-

chased (From Aldrich-Italia, Milan). CH₃CN was dried over CaH₂,

tBuOH was dried over Mg/I₂, CH₂Cl₂was dried over NaH, and THF

and dioxane were dried over Na/benzophenone prior to use, while

DMF was bought already anhydrous. Anhydrous reactions were

run under a positive pressure of dry N₂or argon. IR spectra were

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1383

M. Botta et al.

MED

recorded on a PerkinElmer BX FTIR system, using KBr pellets. TLC

was carried out using Merck TLC plates silica gel 60 F254. Chroma-

tographic purifications were performed on columns packed with

(ESI) m/z: 295.4 [MꢀH]^ꢀ; Anal. calcd for C₆H₇NO₃S₂: C 35.11, H 3.44,

N 6.82, found: C 35.10, H 3.48, N 6.88.

N-(2-Nitrophenyl)-3-(4-oxo-2-thioxothiazolidin-3-yl)propanamide

(3b): To a suspension of 2 (100 mg, 0.48 mmol) in 4 mL of dry

CH₂Cl₂, thionyl chloride (40 mL, 0.58 mmol) was added. The reaction

was monitored by TLC and after the disappearance of the starting

material, 2-nitroaniline (80 mg, 0.58 mmol) was added. The result-

ing mixture was stirred at RT for 2 h. Subsequently, H₂O was added

to the reaction mixture. The organic phase was separated and the

aqueous phase was extracted with EtOAc. The combined organic

phases were then dried over Na₂SO₄, concentrated under reduced

pressure, and purified by flash chromatography using a mixture of

hexane/EtOAc (1:1) to give the desired product 3b as a yellow

solid (130 mg, 84%); mp: 170–1718C; ¹H NMR (400 MHz, CDCl₃):

d=10.27 (s, 1H), 8.64–8.62 (m, 1H), 8.15–8.13 (m, 1H), 7.60–7.56

(m, 1H), 7.15–7.11 (m, 1H), 4.45 (t, J=8, 2H), 3.94 (s, 2H), 2.22 ppm

(t, J=8, 2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=206.2, 174.0,

168.8, 136.0, 134.2, 128.0, 125.8, 123.6, 122.4, 40.3, 35.4, 34.9 ppm;

MS (ESI) m/z: 324.4 [MꢀH]^ꢀ; Anal. calcd for C₁₂H₁₁N₃O₄S₂: C 44.30,

H 3.41, N 12.91, found: C 44.36, H 3.38, N 12.90.

1

Merk 60 silica gel, 23–400 mesh, for flash technique. H NMR and

¹³C NMR spectra were recorded at 400 MHz on a Brucker Avance

DPX400, at 300 MHz on a Varian VXR-300, and at 200 MHz on a

Bruker AC200F spectrometer. Chemical shifts are reported relative

to (CH₃)₄Si at 0.00 ppm. Elemental analyses (C, H, N) were per-

formed in-house using a PerkinElmer 240C Elemental Analyzer, and

the data for C, H, and N are within 0.4% of the theoretical values.

Melting points were taken using a Gallenkamp melting point appa-

ratus and are uncorrected. MS data were obtained using an Agilent

1100 LC/MSD VL system (G1946C) with a 0.4 mLmin^ꢀ1flow rate

using a binary solvent system of MeOH/H₂O (95:5). UV detection

was monitored at 254 nm. MS were acquired in positive and nega-

tive mode scanning over the mass range 50–1500. The following

ion source parameters were used: drying gas flow, 9 mLmin^ꢀ1

;

nebulizer pressure, 40 psig; drying gas temperature, 3508C. All

target compounds possessed a purity of ꢄ95% as verified by ele-

mental analyses by comparison with the theoretical values.

Microwave irradiation experiments were conducted using a CEM

Discover Synthesis Unit (CEM Corp., Matthews, NC, USA). The appa-

ratus consists of a continuous focused microwave power delivery

system with operator-selectable power output from 0 to 300 W.

The temperature of the contents of the vessels was monitored

using a calibrated IR temperature control mounted under the reac-

tion vessel. All experiments were performed using a stirring option

whereby the contents of the vessel are stirred by means of a rotat-

ing magnetic plate located below the floor of the microwave

cavity and a Teflon-coated magnetic stir bar in the vessel.

General procedure for the synthesis of 4a–h (Method A). To a

solution of intermediate 3a or 3b (0.33 mmol) in 4 mL of EtOH,

the appropriate benzaldehyde (0.33 mmol) and Et₃N (0.33 mmol)

were added. The reaction mixture was held at reflux for 1 h, and

the formed precipitate was filtered off and washed with EtOH and

hexane to give the desired pure product.

General procedure for microwave-assisted synthesis of 4a–h

(Method B). A mixture of intermediate 3a or 3b (0.16 mmol) and

the appropriate benzaldehyde (0.16 mmol) was microwave irradiat-

ed for 5 min at 1308C (3 cycles). The resulting residue was washed

with EtOH and hexane to give the desired pure compound.

3-(4-Oxo-2-thioxothiazolidin-3-yl)propanoic acid (2): To a solution

of b-alanine 1 (3.4 g, 38.1 mmol) in 17 mL of 22% KOH solution,

CS₂(2.5 mL, 42 mmol) was added dropwise in such a manner that

the temperature of the reaction did not exceed 258C. The mixture

was allowed to stir at room temperature (RT) for ~3 h and then

bromoacetic acid (5.3 g, 38.1 mmol) was added portionwise over

20 min. The reaction mixture was stirred at RT for additional 3 h

during which a precipitate formed. The pH of the reaction mixture

was adjusted to 3–4 using conc. H₂SO₄, and the resulting solution

was stirred overnight at RT. The precipitate was filtered off and

washed with H₂O to afford the desired compound 2 as a yellow

solid (3.52 g, 45%); mp: 155–1568C; ¹H NMR (400 MHz, CD₃OD):

d=5.00 (brs, 1H), 4.22 (t, J=7.5, 2H), 4.11 (s, 1H), 2.64 ppm (t, J=

7.6, 2H); ¹³C NMR (100 MHz, CH₃OD): d=202.6, 174.4, 172.9, 39.6,

35.1, 30.3 ppm; MS (ESI): m/z: 204.1 [MꢀH]^ꢀ; Anal. calcd for

C₆H₇NO₃S₂: C 35.11, H 3.44, N 6.82, found: C 35.10, H 3.48, N 6.88.

3-[(Z)-5-(3-Bromobenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-N-

(2-hydroxyphenyl)propanamide (4a): Yellow solid; mp: 222–

1

2248C; H NMR (400 MHz, [D₆]DMSO): d=9.56 (s, 1H), 9.38 (s, 1H),

7.89 (s, 1H), 7.81 (s, 1H), 7.71 (d, J=7.9 1H), 7.65–7.60 (m, 2H),

7.53–7.49 (m, 1H), 6.92–6.90 (m, 1H) 6.83 (d, J=7.8, 1H) 6.75–6.72

(m, 1H), 4.32 (t, J=7.1, 2H), 2.79 ppm (t, J=7.1, 2H); ¹³C NMR

(100 MHz, [D₆]DMSO): d=193.4, 169.1, 167.1, 148.7, 135.9, 133.9,

133.9, 132.0, 131.5, 129.0, 126.4, 125.3, 124.7, 123.4, 123.1, 119.3,

116.1, 41.4, 33.4 ppm; MS (ESI): m/z: 462.9 [MꢀH]⁺; Anal. calcd for

C₁₉H₁₅BrN₂O₃S₂: C 49.25, H 3.26, N 6.05, found: C 49.31, H 3.29, N

6.06.

3-[(Z)-5-(3-Fluorobenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-N-

(2-hydroxyphenyl)propanamide (4b): Yellow solid (45%); mp:

1

210–2118C; H NMR (400 MHz, [D₆]acetone): d=9.17 (s, 1H), 7.69

(s, 1H), 7.59–7.54 (m, 1H), 7.44–7.41 (m, 1H), 7.37 (d, J=8, 1H),

7.26–7.21 (m, 1H), 6.97–6.92 (m, 1H), 6.82 (d, J=8 1H), 6.76–6.72

(m, 1H), 4.45 (t, J=7.1, 2H), 2.92 ppm (t, J=7.1, 2H); ¹³C NMR

(75 MHz, [D₆]DMSO): d=193.7, 169.3, 167.4, 164.7, 161.4, 148.9,

136.1, 135.9, 132.4, 132.2, 132.0, 126.8, 126.7, 126.6, 126.5, 124.9,

123.6, 119.6, 118.5, 118.2, 118.2, 117.9, 116.3, 40.9, 33.3 ppm. MS

(ESI): m/z: 401.5 [MꢀH]^ꢀ; Anal. calcd for C₁₉H₁₅FN₂O₃S₂: C 56.70, H

N-(2-Hydroxyphenyl)-3-(4-oxo-2-thioxothiazolidin-3-yl)propana-

mide (3a): Compound 2 (1.0 g, 4.87 mmol) and N,N-diisopropyle-

thylamine (932 mL, 5.36 mmol) were dissolved in 20 mL of dry 1,2-

dichloroethane. Subsequently, pivaloyl chloride (660 mL, 5.36 mmol)

was added dropwise. The reaction mixture was stirred at RT for 3 h

and then 2-aminophenol (531 mg, 4.87 mmol) was added. The mix-

ture was allowed to stir at RT for additional 18 h. The solvent was

removed under reduced pressure and the mixture was purified by

flash chromatography using a mixture of hexane/EtOAc (1:1) to

give the desired product 3a as a yellow solid (1.23 g, 85%); mp:

3.76,

N

6.96, found:

C

56.76,

H

3.75,

N

6.98.

3-[(Z)-5-(3,5-Difluorobenzylidene)-4-oxo-2-thioxothiazolidin-3-

yl]-N-(2-hydroxyphenyl)propanamide (4c): Yellow solid (77%);

mp: 235–2378C; ¹H NMR (400 MHz, [D₆]DMSO): d=9.56 (s, 1H),

9.30 (s, 1H), 7.72 (s, 1H), 7.58–7.56 (m, 1H), 7.37–7.33 (m, 1H),

7.28–7.27 (m, 2H), 6.87–6.83 (m, 1H), 6.77–6.75 (m, 1H), 6.68–6.64

(m, 1H), 4.23 (t, J=7.16, 2H), 2.70 ppm (t, J=7.16, 2H); ¹³C NMR

(100 MHz, [D₆]DMSO): d=193.1, 169.0, 167.0, 164.4, 164.2, 161.9,

161.8, 148.6, 136.7, 130.3, 126.3, 126.1, 125.2, 123.3, 119.3, 116.0,

1

168–1708C; H NMR (400 MHz, [D₆]DMSO): d=9.65 (s, 1H), 9.36 (s,

1H), 7.63 (d, J=7.7 1H), 6.95–6.91 (m, 1H) 6.83 (d, J=7.8, 1H)

6.75–6.72 (m, 1H), 4.24 (s, 2H), 4.13 (t, J=7.4, 2H), 2.68 ppm (t, J=

7.5, 2H); ¹³C NMR (50 MHz, [D₆]DMSO): d=202.8, 174.0, 168.5,

148.1, 125.8, 124.7, 122.9, 118.8, 115.6, 38.2, 35.7, 32.6 ppm; MS

1384

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

113.7, 113.4, 106.8, 106.5, 106.3, 41.4, 33.3 ppm. MS (ESI): m/z:

419.4 [MꢀH]^ꢀ; Anal. calcd for C₁₉H₁₄F₂N₂O₃S₂: C 54.28, H 3.36, N

6.66, found: C 54.30, H 3.30, N 6.68.

(1 equiv, synthesized following a reported procedure)^[17]were

added. The reaction mixture was heated at 908C under microwave

irradiation for 10 min. Then the appropriate aldehyde (1 equiv) was

added and the mixture was heated at 1108C under microwave irra-

diation for 5 min. The reaction mixture was evaporated to dryness.

MeOH was added to the residue and a precipitate was formed

upon standing. The solid was isolated by filtration, washed with

hexane, and dried under high vacuum to give the final compounds

7a,b.

3-[(Z)-5-(3-Methylbenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-N-

(2-hydroxyphenyl)propanamide (4d): Yellow solid (61%); mp:

1

208–2098C; H NMR (400 MHz, [D₆]DMSO): d=9.57 (s, 1H), 9.30 (s,

1H), 7.69 (s, 1H), 7.57 (d, J=8, 1H), 7.39–7.36 (m, 3H), 7.27–7.25

(m, 1H), 6.87–6.83 (m, 1H), 6.76 (d, J=8, 1H), 6.69–6.64 (m, 1H),

4.25 (t, J=7.2, 2H), 2.72 ppm (t, J=7.2, 2H); ¹³C NMR (100 MHz,

[D₆]DMSO): d=193.7, 169.0, 167.2, 148.6, 139.3, 133.4, 132.1, 131.4,

129.8, 128.2, 126.3, 125.2, 123.3, 122.6, 119.3, 116.1, 41.2, 33.3,

21.3 ppm; MS (ESI): m/z: 397.5 [MꢀH]^ꢀ; Anal. calcd for

C₂₀H₁₈N₂O₃S₂: C 60.28, H 4.55, N 7.03, found: C 60.21, H 4.59, N

7.12.

N-{2-{(Z)-5-(3-Bromobenzylidene)-4-oxo-2-thioxothiazolidin-3-

yl]ethyl}-2-hydroxybenzamide (7a): Yellow solid (69%); ¹H NMR

(400 MHz, [D₆]DMSO): d=12.32 (s, 1H), 8.86 (t, J=5.2 Hz, 1H), 7.80

(s, 1H), 7.71 (s, 1H), 7.63 (d, J=7.9 Hz, 1H), 7.57–7.54 (m, 2H), 7.44

(t, J=7.9 Hz, 1H), 7.30 (t, J=7.7 Hz, 1H), 6.78 (t, J=8.6 Hz, 2H),

4.20 (t, J=5.2 Hz, 1H), 3.57 ppm (q, J=5.2 Hz, 1H); ¹³C NMR

(100 MHz, [D₆]DMSO): d=194.2, 170.2, 167.6, 160.7, 135.9, 133.9,

131.1, 129.0, 128.1, 124.8, 123.1, 115.4, 44.4, 36.7 ppm; MS (ESI) m/

z: 462.9 [MꢀH]^ꢀ; Anal. calcd for C₁₉H₁₅BrN₂O₃S₂: C 49.25, H 3.26, N

6.05, found: C 49.19, H 3.34, N 6.12.

3-[(Z)-5-(3-Methoxybenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-

N-(2-hydroxyphenyl)propanamide (4e): Yellow solid (67%); mp:

1

200–2028C; H NMR (400 MHz, [D₆]DMSO): d=9.31 (s, 1H), 7.73 (s,

1H), 7.57 (d, J=8, 1H), 7.42–7.38 (m, 1H), 7.14–7.12 (m, 2H), 7.03–

7.01 (m, 1H), 6.87–6.83 (m, 1H), 6.76 (d, J=8, 1H), 6.68–6.64 (m,

1H), 4.25 (t, J=7.2, 2H), 2.72 ppm (t, J=7.2, 2H); ¹³C NMR

(100 MHz, [D₆]DMSO): d=193.7, 169.1, 160.2, 148.7, 134.8, 133.3,

131.1, 126.4, 125.3, 123.4, 122.9, 119.3, 117.5, 116.3, 116.2, 55.8,

41.4, 33.4 ppm; MS (ESI): m/z: 413.5 [MꢀH]^ꢀ; Anal. calcd for

C₂₀H₁₈N₂O₄S₂: C 57.95, H 4.38, N 6.76, found: C 57.97, H 4.44, N

6.70.

N-{2-[(Z)-5-(3-Fluorobenzylidene)-4-oxo-2-thioxothiazolidin-3-

yl]ethyl}-2-hydroxybenzamide (7b): Yellow solid (62%); ¹H NMR

(400 MHz, [D₆]DMSO): d=12.31 (1H, s), 8.85 (t, J=4.6 Hz, 1H), 7.72

(1H, s), 7.58–7.50 (m, 2H), 7.45–7.37 ( m, 2H), 7.31–7.29 (m, 2H),

6.80–6.76 (m, 2H), 4.20 (t, J=4.5 Hz, 2H), 3.56 ppm (q, J=4.6 Hz,

2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=193.8, 169.8, 167.2, 164.8,

160.2, 135.4, 133.7, 131.7, 131.6, 130.8, 127.6, 126.0, 125.9, 124.3,

118.6, 117.4, 114.9, 43.9, 36.2 ppm; MS (ESI): m/z: 401.5 [MꢀH]^ꢀ;

Anal. calcd for C₁₉H₁₅FN₂O₃S₂: C 56.70, H 3.76, N 6.96, found: C

56.59, H 3.83, N 6.84.

3-{(Z)-5-[(Benzo[d][1,3]dioxol-5-yl)methylene]-4-oxo-2-thioxothia-

zolidin-3-yl}-N-(2-hydroxyphenyl)propanamide (4 f). Yellow solid

1

(53%); mp: 240–2418C; H NMR (400 MHz, [D₆]DMSO): d=9.56 (s,

1H), 9.30 (s, 1H), 7.66 (s, 1H), 7.58–7.56 (m, 1H), 7.14–7–12 (m,

1H), 7.09 (s, 1H), 7.04–7.01 (m, 1H), 6.87–6.84 (m, 1H), 6.77–6.75

(m, 1H), 6.68–6.65 (m, 1H), 6.07 (s, 2H), 4.23 (t, J=7.2, 2H),

2.70 ppm (t, J=7.2, 2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=193.3,

169.0, 167.2, 150.3, 148.8, 148.6, 133.5, 127.6, 127.4, 126.3, 125.2,

123.3, 120.2, 119.3, 116.1, 110.0, 109.7, 102.6, 41.2, 33.3 ppm. MS

(ESI): m/z: 427.0 [MꢀH]^ꢀ; Anal. calcd for C₂₀H₁₆N₂O₅S₂: C 56.06, H

3.76, N 6.54, found: C 56.03, H 3.84, N 6.51.

(E)-3-(3-bromobenzylidene)dihydrofuran-2,5-dione (9): To a solu-

tion of tBuOK, from 1.13 g (29 mmol, 1.2 equiv) of potassium in

15 mL of tert-butyl alcohol, 4.86 mL of diethyl succinate (29 mmol,

1.2 equiv) in 2.8 mL of tert-butyl alcohol were added dropwise. A

solution of 3-bromobenzaldehyde 8 (2.82 mL, 1 equiv) in 2.6 mL of

tert-butyl alcohol was added dropwise, and the resulting heteroge-

neous mixture was held at reflux for 2 h and then stirred at RT

overnight. After cooling, 30 mL of H₂O were added and tert-butyl

alcohol was removed by distillation. To this mixture, a solution of

3.12 g (55.6 mmol, 2.3 equiv) of KOH in 10 mL of H₂O was added.

After holding at reflux for 4 h, the mixture was washed with Et₂O.

The aqueous layer was treated with conc. HCl until pH 1 was

reached. The solution was stirred for 5 min and then it was extract-

ed with EtOAc. The organic phase was dried over Na₂SO₄and final-

ly evaporated under reduced pressure. The crude material was re-

crystallized from cyclohexane/hexane. The latter intermediate

(1.67 g, 5.86 mmol, 1 equiv) was dissolved in THF, then acetic anhy-

dride (1.22 mL, 2.2 equiv) was added and the resulting mixture

stirred at RT for 5 h. Next THF was removed and methyl tert-butyl

ether (MTBE) was added to induce a more complete crystallization.

The crystals were filtered, washed with MTBE and dried in vacuo to

yield the desired compound 9 as a white solid (2.26 g, 35%); mp:

151–1528C; ¹H NMR (400 MHz, [D₆]DMSO): d=7.80 (s, 1H), 7.62–

7.53 (m, 3H), 3.94–3.95 ppm (m, 2H); MS (ESI) m/z: 266 [MꢀH]^ꢀ;

Anal. calcd for C₁₁H₇BrO₃: C 49.47, H 2.64, found: C 49.56, H 2.66.

3-[(Z)-5-(3-Bromobenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-N-

(2-nitrophenyl)propanamide (4g): Yellow solid (85%); mp: 204–

1

2068C; H NMR (400 MHz, [D₆]DMSO): d=10.32 (s, 1H), 7.87–7.84

(m, 1H), 7.79 (s, 1H), 7.72 (s, 1H), 7.63–7.60 (m, 2H), 7.54–7.52 (m,

2H), 7.45–7.41 (m, 1H), 7.31–7.27 (m, 1H), 4.23 (t, J=7.4, 2H),

2.70 ppm (t, J=7.4, 2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=193.4,

168.9, 167.0, 143.0, 135.8, 134.3, 133.8, 131.9, 131.4, 131.2, 129.0,

126.0, 125.8, 125.2, 124.7, 123.0, 40.9, 33.2 ppm. MS (ESI): m/z:

491.4 [MꢀH]^ꢀ; Anal. calcd for C₁₉H₁₄BrN₃O₄S₂: C 46.35, H 2.87, N

8.53, found: C 46.42, H 2.85, N 8.56.

3-[(Z)-5-(3-Fluorobenzylidene)-4-oxo-2-thioxothiazolidin-3-yl]-N-

(2-nitrophenyl)propanamide (4h): Yellow solid (77%); mp: 207–

1

2088C; H NMR (400 MHz, [D₆]DMSO): d=10.32 (s, 1H), 7.87–7.85

(m, 1H), 7.74 (s, 1H), 7.64–7.60 (m, 1H), 7.54–7.52 (m, 2H), 7.45–

7.42 (m, 1H), 7.39–7.37 (m, 1H), 7.31–7.27 (m, 2H), 4.23 (t, J=8,

2H), 2.70 ppm (t, J=8, 2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=

193.6, 169.0, 167.1, 164.0, 161.6, 143.1, 135.8, 135.7, 134.4, 132.1,

132.0, 131.8, 131.3, 126.5, 126.1, 125.9, 125.3, 124.7, 118.3, 118.1,

117.9, 117.7, 40.9, 33.3 ppm. MS (ESI): m/z: 430.5 [MꢀH]^ꢀ; Anal.

calcd for C₁₉H₁₄FN₃O₄S₂: C 52.89, H 3.27, N 9.74, found: C 52.80, H

3.26, N 9.81.

3-[(E)-3-(3-bromobenzylidene)-2,5-dioxopyrrolidin-1-yl]propanoic

acid (10): Compound 9 (0.564 mmol) and b-alanine (150 mg,

0.564 mmol) were dissolved in dry DMF and the solution was

heated under microwave irradiation (200W) at 1508C for 10 min.

H₂O was added and the resulting mixture was extracted with

CH₂Cl₂. The organic phases were collected, washed with brine, and

dried over Na₂SO₄. The solvent was removed under reduced pres-

General procedure for the synthesis of the inverted amide de-

rivatives 7a,b: To a solution of bis(carboxymethyl)trithiocarbonate

5 (1 equiv) in dimethoxyethane (2 mL), Et₃N (1 equiv) and amine 6

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1385

M. Botta et al.

MED

sure, and the residue was crystallized in EtOH to obtain the desired

product 10 as a white solid (78 mg, 42%). ¹H NMR (400 MHz,

[D₆]DMSO): d=12.31 (brs, 1H), 7.76 (s, 1H), 7.62–7.52 (m, 2H),

7.39–7.31 (m, 2H), 3.66 (t, J=7.5, 2H,), 2.47 ppm (t, J=7.5 Hz, 2H);

MS (ESI): m/z: 337.1 [MꢀH]^ꢀ; Anal. calcd for C₁₄H₁₂BrNO₄: C 49.73,

H 3.58, N 4.14, found: C 49.70, H 3.63, N 4.08.

N-Benzyl-4-chloro-6-morpholino-1,3,5-triazin-2-amine

(14a):

1

Yield 67%; H NMR (200 MHz, CDCl₃): d =7.28–7.25 (m, 5H), 7.07

(brs, 1H), 4.58 (d, J=6 Hz, 2H), 3.74–3.65 ppm (m, 8H). MS (ESI):

m/z: 306.8 [M+H]⁺; Anal. calcd for C₁₄H₁₆ClN₅O: C 54.99, H 5.27, N

22.90, found: C 55.05, H 5.25, N 22.87.

N2-Benzyl-6-chloro-N4,N4-diethyl-1,3,5-triazine-2,4-diamine

1

3-[(E)-3-(3-bromobenzylidene)-2,5-dioxopyrrolidin-1-yl]-N-(2-hy-

droxyphenyl)propanamide (11): Compounds 10 (72 mg,

0.259 mmol, 1 equiv) and DIPEA (50 mL, 0.285 mmol, 1.1 equiv)

were dissolved in DCE (3 mL). Subsequently, pivaloyl chloride

(35 mL, 0.285 mmol, 1.1 equiv) was added dropwise and the result-

ing mixture was stirred at RT for 1 h. Then aminophenol (27 mg,

0.247 mmol, 0.95 equiv) was added and the mixture was stirred at

RT overnight. The solvent was removed and the resulting residue

was crystallized in EtOH to obtain the final product 11 as a white

solid (72 mg, 62%); mp: 226–2288C; ¹H NMR (400 MHz,

[D₆]DMSO): d=9.54 (s, 1H), 9.28 (s, 1H), 7.77 (s, 1H), 7.60–7.53 (m,

3H), 7.41–7.31 (m, 2H), 6.86 (t, J=7.4 Hz, 1H), 6.76 (d, J=7.6 Hz,

1H), 6.67 (t, J=7.4 Hz, 1H), 3.71 (t, J=7.1 Hz, 2H), 3.66 (s, 2H),

2.61 ppm (t, J=7.1 Hz, 2H); ¹³C NMR (100 MHz, [D₆]DMSO): d=

173.8, 170.1, 168.2, 148.2, 136.4, 132.5, 132.3, 130.9, 130.3, 128.7,

126.9, 125.3, 124.6, 123.9, 122.2, 118.8, 115.7, 34.8, 33.6, 33.5 ppm;

MS (ESI): m/z: 428.3 [MꢀH]^ꢀ; Anal. calcd for C₂₀H₁₇BrN₂O₄: C 55.96,

H 3.99, N 6.53, found: C 55.88, H 4.06, N 6.46.

(14b): Yield 75%; H NMR (200 MHz, CDCl₃): d=7.29–7.24 (m, 5H),

6.06 (brs, 1H), 4.57 (s, 2H), 3.55–3.48 (m, 4H), 1.16–1.07 (m, 6H);

MS (ESI): m/z: 292.8 [M+H]⁺; Anal. calcd for C₁₄H₁₅Cl₂N₅: C 57.63, H

6.22, N 24.00, found: C 57.60, H 6.16, N 24.08.

4-Chloro-N-(3-chlorobenzyl)-6-morpholino-1,3,5-triazin-2-amine

(14c): Yield 85%; mp: 168–168.58C; ¹H NMR (400 MHz, CD₃OD):

d=7.72 (s, 1H), 7.19–7.16 (m, 2H), 6.67 (brs, 1H), 4.54–4.49 (m,

2H), 3.72–3.61 ppm (m, 8H); ¹³C NMR (100 MHz, CDCl₃): d=165.46,

164.34, 163.10, 144.97, 134.07, 129.34, 128.90, 128.82, 115.42,

115.21, 66.54, 66.37, 44.03, 43.79 ppm; MS (ESI): m/z: 340.1 [M+H]⁺

; Anal. calcd for C₁₄H₁₅Cl₂N₅: C 49.43, H 4.44, N 20.59, found: C

49.40, H 4.46, N 20.51.

4-Chloro-N-(4-fluorobenzyl)-6-morpholino-1,3,5-triazin-2-amine

(14d): Yield 95%; mp: 1928C; ¹H NMR (400 MHz, CD₃OD): d=

7.23–7.17(m, 2H), 6.95 (t, J=8.43, 2H), 6.00 (brs, 1H), 4.52–4.47 (m,

2H), 3.72 (s, 4H), 3.62 ppm (s, 4H); MS (ESI) m/z: 324.7 [M+H]⁺.

Anal. calcd for C₁₄H₁₅ClFN₅O: C 51.94, H 4.67, N 21.63, found: C

51.88, H 4.74, N 21.60.

General procedure for the synthesis of compounds 13a–c: To a

stirred solution of cyanuric chloride 12 (1 g, 5.42 mmol) in dime-

thoxyethane (35 mL) at ꢀ308C, the appropriate benzylamine

(1 equiv) was added dropwise. The reaction mixture was vigorously

stirred for 3 h at ꢀ308C, then warmed to RT and washed succes-

sively with 3N HCl, H₂O, and brine. The organic phase was dried

over anhydrous Na₂SO₄and then evaporated to dryness. The re-

sulting white residue was dissolved in a minimum amount of

CH₂Cl₂. Finally, by addition of petroleum ether, the final compound

was precipitated and collected by filtration.

General procedure for the synthesis of compounds 15e–g and

16i,j,o,p. To a solution of the appropriate intermediate 14a–d

(1 equiv) in CH₂Cl₂, hydrazine (4 equiv) was added and the result-

ing mixture was held at reflux for 12 h. After cooling down to RT,

the mixture was washed with H₂O and brine. The organic phase

was dried over anhydrous Na₂SO₄and evaporated to dryness. The

resulting residue was dissolved in toluene and reacted with the ap-

propriate aldehyde (2 equiv). The reaction mixture was held at

reflux for 3 h using a Dean–Stark apparatus for azeotropical remov-

al of H₂O. The reaction mixture was evaporated to dryness and the

resulting residue was dissolved in a minimum amount of CH₂Cl₂.

Upon subsequent addition of petroleum ether, the desired com-

pounds 15e–g and 16i,j,o,p precipitated and were collected by fil-

tration.

N-Benzyl-4,6-dichloro-1,3,5-triazin-2-amine (13a): Yield 85%;

¹H NMR (200 MHz, CDCl₃): d=7.30–7.26 (m, 5H), 4.48 ppm (m,

2H); MS (ESI): m/z: 256.1 [M+H]⁺; Anal. calcd for C₁₀H₉Cl₂N₄: C

47.08, H 3.16, N 21.96, found: C 47.02, H 3.21, N 22.04.

4,6-Dichloro-N-(4-fluorobenzyl)-1,3,5-triazin-2-amine (13b): Yield

75%; mp: 151.0–151.58C; ¹H NMR (400 MHz, CD₃OD): d=7.23–

7.19 (m, 2H), 6.98 (t, J=8.48 Hz, 2H), 6.19 (brs,1H); MS (ESI) m/z:

274.1 [M+H]⁺; Anal. calcd for C₁₀H₇Cl₂FN₄: C 43.98, H 2.58, N 20.52,

found: C 43.88, H 2.62, N 20.59.

(E)-2-({2-[4-(Benzylamino)-6-morpholino-1,3,5-triazin-2-yl]hydra-

zono}methyl)phenol (15e): Yield 80%; H NMR (400 MHz, CDCl₃):

1

d=12.39 (s, 1H), 8.87 (s, 1H), 7.60–7.58 (m, 1H), 7.34–7–30 (m,

6H), 6.96–6.95 (m, 1H), 6.83–6.81 (m, 1H), 6.15 (brs, 1H), 4.64 (s,

2H), 3.60–3.57 (m, 4H), 1.26–1.12 ppm (m, 6H); ¹³C NMR (CDCl₃):

d=164.4, 163.3, 141.2, 139.8, 138.2, 137.8, 129.9, 127.9, 126.8,

126.4, 122.2, 120.8, 120.2, 110.3, 44.0, 40.9, 12.0 ppm. MS (ESI): m/z:

431.5 [M+H]⁺; Anal. calcd for C₂₂H₂₂N₈O₂: C 61.38, H 5.15, N 26.03,

found: C 61.34, H 5.21, N 26.09.

4,6-Dichloro-N-(3-chlorobenzyl)-1,3,5-triazin-2-amine (13c): Yield

78%; mp: 1498C; ¹H NMR (400 MHz, CD₃OD): d=7.23–7.19 (m,

3H), 7.13–7.10 (m, 1H), 6.58 (brs, 1H), 4.60 ppm (d, J=6.29 Hz, 2H);

¹³C NMR (100 MHz, CDCl₃): d=170.0, 166.0, 138.5, 134.8, 130.2,

128.2, 127.7, 125.7, 44.7 ppm. MS (ESI): m/z: 289.5 [M+H]⁺; 311.5

[M+Na]⁺; Anal. calcd for C₁₀H₇Cl₃N₄: C 41.48, H 2.44, N 19.35,

found: C 41.45, H 2.38, N 19.36.

(Z)-3-{2-[4-(3-Chlorobenzylamino)-6-morpholino-1,3,5-triazin-2-

yl]hydrazono}indolin-2-one (15 f): Yield 71%; mp: 144–144.58C;

¹H NMR (400 MHz, CD₃OD): d=12.39 (s, 1H), 8.15 (s, 1H), 7.62 (d,

J=6.80 Hz, 1H), 7.26 (s, 1H), 7.16 (m, 4H), 7.00–6.90 (m, 1H), 6.77

(d, J=7.82 Hz), 4.53 (s, 2H), 3.75 (s, 4H), 3.65 (s, 4H); ¹³C NMR

(100 MHz, CDCl₃): d=165.1, 164.3, 163.1, 141.1, 139.8, 134.3, 132.4,

130.2, 129.8, 127.7, 127.3, 125.5, 122.9, 121.3, 120.9, 110.5, 66.8,

44.4, 43.8 ppm; MS (ESI) m/z: 465.9 [M+H]⁺; Anal. calcd for

C₂₂H₂₁ClN₈O₂: C 56.84, H 4.55, N 24.10, found: C 56.88, H 4.47, N

24.11.

General procedure for the synthesis of compounds 14a–d: To a

suspension of the appropriate intermediate 13a–c (1 equiv) in

CH₂Cl₂, the appropriate amine (2 equiv) was added dropwise. The

reaction mixture was stirred at RT for 12 h and then washed suc-

cessively with 3N HCl, H₂O, and brine. The organic phase was

dried over anhydrous Na₂SO₄and evaporated to dryness. The re-

sulting white residue was dissolved in a minimum amount of

CH₂Cl₂. Finally, by addition of petroleum ether, the final compound

precipitated and was collected by filtration.

(E)-3-{2-[4-(benzylamino)-6-(diethylamino)-1,3,5-triazin-2-yl]hy-

drazono}indolin-2-one (15g): Yield 78%; ¹H NMR (400 MHz,

CDCl₃): d=11.52 (s, 1H), 7.92 (s, 2H), 7.33 (s, 3H), 7.18–7.17 (m,

1386

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

1H), 7.01–6.99 (m, 1H), 6.91–6.87 (m, 1H), 4.63 (s, 2H), 3.82–3.73

(m, 8H); MS (ESI) m/z: 417.5 [M+H]⁺; Anal. calcd for C₂₂H₂₄N₈O: C

63.45, H 5.81, N 26.90, found: C 63.53, H 5.88, N 26.81.

stirred at RT for 12 h and then washed consecutively with 3N HCl,

H₂O, and brine. The organic phase was then dried over anhydrous

Na₂SO₄and evaporated to dryness. The obtained white residue

was dissolved in a minimum amount of CH₂Cl₂. Upon addition of

petroleum ether, the final compound precipitated and was collect-

ed by filtration. Yield 82%; mp: 170–1728C; ¹H NMR (400 MHz,

CDCl₃): d=7.37 (q, J=8.60 Hz, 2H); 6.97 (t, J=8.63 Hz, 2H); 3.78

(brs, 2H); 3.72 (brs, 2H); 3.67–3.66 ppm (m, 4H). MS (ESI): m/z:

310.7 [M+H]⁺; 332.6 [M+Na]⁺; Anal. calcd for C₁₃H₁₃ClFN₅O: C

50.41, H 4.23, N 22.61, found: C 50.44, H 4.28, N 22.58.

(E)-3-({2-[4-(Benzylamino)-6-morpholino]-1,3,5-triazin-2-yl}hydra-

zono)indolin-2-one (16j): Yield 74%; ¹H NMR (400 MHz, CDCl₃):

d=12.47 (s, 1H), 9.27 (s, 1H), 7.51–7.49 (m, 1H), 7.32–7.23 (m, 5H),

7.15–7.13 (m, 1H), 7.05–7.01 (m, 1H), 6.90–6.79 (m, 1H), 6.48 (brs,

1H), 4.64 (s, 2H), 3.84–3.72 ppm (m, 8H); ¹³C NMR (CDCl₃): d=

166.1, 165.1, 164.1, 163.3, 134.0, 138.8, 132.5, 123.0, 128.4, 127.4,

127.1, 122.7, 121.2, 120.7, 110.5, 66.7, 44.9, 43.7; MS (ESI): m/z:

406.5 [M+H]⁺; Anal. calcd for C₂₁H₂₃N₇O₂: C 62.21, H 5.72, N 24.18,

found: C 62.26, H 5.64, N 24.15.

4-(4,6-Dichloro-1,3,5-triazin-2-yl)morpholine (22): To a stirred so-

lution of cyanuric chloride 12 (1 g, 5.42 mmol) in dimethoxyethane

(50 mL) at ꢀ308C, morpholine (470.1 mL, 5.42 mmol) was added

dropwise. The reaction mixture was vigorously stirred for 3 h at

ꢀ308C, then warmed to RT and washed consecutively with 3N

HCl, H₂O, and brine. The organic phase was dried over anhydrous

Na₂SO₄and evaporated to dryness to yield the pure product 22 as

a white solid. Yield 79%; mp: 156.7–156.98C; ¹H NMR (400 MHz,

CD₃OD): d=3.80 (t, J=4.59 Hz, 4H); 3.66 ppm (t, J=4.59 Hz, 4H);

¹³C NMR (100 MHz, CDCl₃): d=170.4, 164.1, 66.4, 44.5 ppm. MS

(ESI): m/z: 235.1 [M+H]⁺, 257.2 [M+Na]⁺; Anal. calcd for

C₇H₈Cl₂N₄O: C 35.77, H 3.43, N 23.83, found: C 35.74, H 3.38, N

23.92.

(E)-2-({2-[4-(4-fluorobenzylamino)-6-morpholino-1,3,5-triazin-2-

yl]hydrazono}methyl)phenol (16i). Yield 87%; mp: 114–114.58C.

¹H NMR (400 MHz, CD₃OD): d=11.48 (s, 1H), 7.77 (s, 1H), 7.19 (s,

3H), 7.02 (s, 1H), 6.91 (s, 3H), 6.80 (t, J=6.63 Hz, 1H), 4.47 (s, 2H),

3.74 (s, 4H), 3.65 ppm (s, 4H); ¹³C NMR (100 MHz, CDCl₃): d=

165.1, 163.9, 163.3, 160.8, 158.1, 143.9, 134.8, 130.8, 129.8, 129.2,

119.2, 118.2, 117.0, 115.5, 115.3, 66.8, 44.1, 43.7 ppm; MS (ESI): m/z:

424.4 [M+H]⁺; 446.4 [M+Na]⁺; Anal. calcd for C₂₁H₂₂FN₇O₂: C

59.57, H 5.24, N 23.15, found: C 59.56, H 5.06, N 23.14.

(E)-2-({2-[4-(3-chlorobenzylamino)-6-morpholino-1,3,5-triazin-2-

yl]hydrazono}methyl)phenol (16o): Yield 82%; mp: 163.5–1648C;

¹H NMR (400 MHz, CD₃OD): d=11.51 (s, 1H), 7.83 (s, 1H), 7.29–7.07

(m, 3H), 6.96 (d, J=8.2, 1H), 6.84 (t, J=7.4, 3H), 4.54 (d, J=4.9 Hz,

2H), 3.77 (s, 4H), 3.71 (s, 4H); ¹³C NMR (100 Hz, CDCl₃): d=164.8,

158.0, 143.9, 141.1, 134.3, 130.8, 129.8, 127.6, 127.3, 125.5, 119.1,

118.0, 116.9, 66.7, 44.2, 43.7; MS (ESI): m/z: 440.9 [M+H]⁺; 462.9

[M+Na]⁺; Anal. calcd for C₂₁H₂₂ClN₇O₂: C 57.34, H 5.04, N 22.29,

found: C 57.37, H 5.12, N 22.21.

4-Chloro-N-(4-fluorophenyl)-6-morpholino-1,3,5-triazin-2-amine

(23). To a stirred solution of compound 22 (650 mg, 2.76 mmol) in

CH₂Cl₂, 4-fluoro aniline (310.81 mL, 2.71 mmol) was added. The re-

action mixture was held at reflux for 12 h and then washed with

3N HCl, H₂O, brine and finally dried over anhydrous Na₂SO_4.The or-

ganic phase was evaporated to dryness and the white residue was

dissolved in a minimum amount of CH₂Cl₂. Upon addition of petro-

leum ether the final compound precipitated as a white solid and

was collected by filtration. Yield 90%. mp:170–1728C; ¹H NMR

(400 MHz, CDCl₃): d=7.38 (q, J=8.60 Hz, 2H); 6.97 (t, J=8.63 Hz,

2H); 3.78 (brs, 2H); 3.72 (brs, 2H); 3.67–3.66 ppm (m, 4H) MS (ESI):

m/z: 310.7 [M+H]⁺; 332.6 [M+Na]⁺; Anal. calcd for C₁₃H₁₃ClFN₅O: C

50.41, H 4.23, N 22.61, found: C 50.35, H 4.19, N 22.69.

(E)-2-({2-[4-(benzylamino)-6-(diethylamino)-1,3,5-triazin-2-yl]hy-

drazono}methyl)phenol (16p). Yield 85%; ¹H NMR (400 MHz,

CDCl₃): d=11.54 (brs, 1H), 7.86 (s, 1H), 7.31–7.19 (m, 6H), 7.10–

7.08 (m, 1H), 6.97–6.95 (m, 1H), 6.85–6.82 (m, 1H), 5.24 (brs, 1H),

4.59–4.57 (m, 2H), 3.56 (m, 4H), 1.18–1.11 ppm (m, 6H); MS (ESI)

m/z: 392.5 [M+H]⁺; Anal. calcd for C₂₁H₂₅N₇O: C 64.43, H 6.44, N

25.05, found: C 64.38, H 6.48, N 25.00.

General procedure for the synthesis of compounds 16m,n and

20b,c. To a solution of the appropriate intermediate 17, 19, or 23

(1 equiv) in CH₂Cl₂, hydrazine (4 equiv) was added and the result-

ing mixture was held at reflux for 12 h. After cooling to RT, the mix-

ture was washed with H₂O and brine. The organic phase was dried

over anhydrous Na₂SO₄and evaporated to dryness. The obtained

residue was dissolved in toluene and reacted with the appropriate

aldehyde (2 equiv). The reaction mixture was held at reflux for 3 h

using a Dean–Stark apparatus for azeotropical removal of H₂O and

then evaporated to dryness. The resulting residue was dissolved in

a minimum amount of CH₂Cl₂. Upon addition of petroleum ether

the desired final compound precipitated and was collected by fil-

tration.

4,6-Dichloro-N-(4-fluorophenyl)-1,3,5-triazin-2-amine (18): To a

stirred solution of cyanuric chloride 12 (2 g, 10.86 mmol) in dime-

thoxyethane (20 mL) at ꢀ308C, 4-fluoroaniline (1 equiv) was added

dropwise. The reaction mixture was vigorously stirred for 3 h at

ꢀ308C, then warmed to RT, and washed with 3N HCl, H₂O, and

brine. The organic phase was then dried over anhydrous Na₂SO₄

and evaporated to dryness. The obtained white residue was puri-

fied by flash chromatography using a mixture of petroleum ether/

Et₂O (3:1) to give the desired product 18 and the side product 17.

Yield 35%; mp: 1728C; ¹H NMR (400 MHz, CDCl₃): d=7.79 (brs,

1H), 7.41 (q, J=8.60 Hz, 2H), 7.03 ppm (t, J=8.63 Hz, 2H); ¹³C NMR

(100 MHz, CDCl₃): d=171.3, 164.3, 161.7, 159.2, 131.6, 123.8,

116.3 ppm; MS (ESI) m/z: 260.1 [M+H]⁺; 282.1 [M+Na]⁺; Anal.

calcd for C₉H₅Cl₂FN₄: C 41.73, H 1.95, N 21.63, found: C 41.77, H

1.93, N 21.64.

Example: (E)-N-(4-Fluorophenyl)-4-morpholino-6-[2-(naphthalen-

1-ylmethylene)hydrazinyl]-1,3,5-triazin-2-amine

(20b):

Yield

80%; mp: 220.0–220.48C; ¹H NMR (400 MHz, CDCl₃): d=8.62 (m,

1H); 8.47 (m, 1H); 7.91–7.81 (m, 3H); 7.53–7.49 (m, 5H); 6.99 (t, J=

8.4 Hz, 3H); 3.82 (s, 4H); 3.76 ppm (s, 4H); ¹³C NMR (100 MHz,

CDCl₃): d=165.9, 161.1, 156.3, 142.6, 133.8, 128.8, 127.6, 127.0,

126.8, 126.3, 121.8, 115.1, 77.6, 77.0, 76.4, 66.8, 43.8 ppm; MS (ESI):

m/z: 444.2 [M+H]⁺; Anal. calcd for C₂₄H₂₂FN₇O: C 65.00, H 5.00, N

22.11, found: C 65.06, H 5.10, N 22.02.

6-Chloro-N2,N4-bis(4-fluorophenyl)-1,3,5-triazine-2,4-diamine

1

(17): Yield 50%; H NMR (400 MHz, CDCl₃): d=7.79 (brs, 2H), 7.39–

7.36 (m, 4H), 7.04–7.02 (m, 4H); MS (ESI): m/z: 334.7 [M+H]⁺; Anal.

calcd for C₁₅H₁₀ClF₂N₅: C 53.99, H 3.02, N 20.99, found: C 53.94, H

3.05, N 20.95.

(E)-N-(4-fluorophenyl)-4-morpholino-6-[2-(naphthalen-2-ylmethy-

lene)hydrazinyl]-1,3,5-triazin-2-amine (20c): Yield 83%; mp: 225–

2268C; ¹H NMR (400 MHz, CDCl₃): d=10.95 (brs, 1H), 8.22(s,1H),

4-Chloro-N-(4-fluorophenyl)-6-morpholino-1,3,5-triazin-2-amine

(19). To a suspension of compound 18 (1 equiv) in CH₂Cl₂, morpho-

line (2 equiv) was added dropwise. The reaction mixture was

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1387

M. Botta et al.

MED

7.93–7.83 (m, 6H), 7.46–7.44 (m,2H), 7.05 (t, J=8.5; 2H), 3.68 (s,

4H), 3.58 ppm (s, 4H); ¹³C NMR (100, MHz, CDCl₃): d=165.3, 164.6,

158.9, 156.6, 142.9, 137.2, 133.8, 133.5, 133.4, 133.3, 128.8, 128.6,

128.2, 128.1, 127.1, 122.9, 121.8, 115.4, 115.2, 66.5, 43.9 ppm; MS

(ESI): m/z: 444.5 [M+H]⁺; Anal. calcd for C₂₄H₂₂FN₇O: C 65.00, H

5.00, N 22.11, found: C 65.05, H 5.09, N 22.14.

d=162.0, 161.0, 158.6, 135.4, 134.8, 134.5, 133.7, 129.6, 127.8,

127.6, 127.5, 124.8, 124.7, 116.1, 115.9, 95.3, 62.5, 40.8, 29.7,

26.8 ppm; MS (ESI): m/z: 502.7 [M+H]⁺; Anal. calcd for

C₂₈H₃₂FN₅OSi: C 67.04, H 6.43, N 13.96, found: C 67.12, H 6.40, N

13.94.

2-[(2-{4-[2-(tert-Butyldiphenylsilyloxy)ethyl]-6-(4-fluorophenyla-

mino)pyrimidin-2-yl}hydrazono)methyl]phenol (29): Compound

28 (0.046 mmol) and salicylaldehyde (0.046 mmol) were dissolved

in dry toluene containing molecular sieves. The mixture was held

at reflux for 2 h, then cooled to RT, concentrated under reduced

pressure, and purified by flash chromatography using a mixture of

CH₂Cl₂/EtOAc (9:1) to give the desired product 29. Yield 54%;

¹H NMR (200 MHz, CDCl₃): d =8.05 (s, 1H), 7.55 (d, J=4 Hz, 4H),

6.7–7.4 (m, 14H), 5.9 (s, 1H), 3.9 (t, J=8 Hz, 2H), 2.7 (t, J=8 Hz,

2H), 0.95 ppm (s, 9H); ¹³C NMR (50 MHz, CDCl₃): d=166.8, 162.4,

162.2, 158.3, 158.0, 143.8, 135.5, 133.9, 133.9, 133.5, 130.5, 129.8,

129.6, 127.6, 124.6, 124.5, 119.1, 118.4, 116.9, 116.4, 116.0, 95.6,

62.2, 39.9, 29.6, 26.8 ppm; MS (ESI): m/z: 606.8 [M+H]⁺; Anal. calcd

for C₃₅H₃₆FN₅O₂Si: C 69.39, H 5.99, N 11.56, found: C 69.38, H 6.05,

N 11.50.

2-[6-(4-Fluorophenylamino)-2-(methylthio)pyrimidin-4-yl]ethanol

(25): A mixture of 24 (0.28 mmol) and 4-fluoroaniline (1.12 mmol)

in dimethoxyethane was microwave irradiated at 1508C for 5 min.

The reaction mixture was then concentrated under reduced pres-

sure, diluted with EtOAc, and washed with H₂O. The organic phase

was dried over anhydrous Na₂SO₄and purified by flash chromatog-

raphy using a mixture of CH₂Cl₂/EtOAc (1:1) to afford the desired

1

product 25. Yield 80%; H NMR (400 MHz, CDCl₃): d=7.3 (m, 2H),

7.1 (t, 2H), 6.62 (s, 1H), 6.1 (s, 1H), 3.95 (t, J=8 Hz, 2H), 2.75 (t, J=

8 Hz, 2H), 2.5 ppm (s, 3H); ¹³C NMR (50 MHz, CD₃OD): d=172.8,

167.1, 162.7, 162.1, 157.9, 136.9, 123.8, 123.7, 116.4, 116.0, 101.8,

61.6, 41.0, 14.2 ppm. MS (ESI): m/z: 280.3 [M+H]⁺; Anal. calcd for

C₁₃H₁₄FN₃OS: C 55.90, H 5.05, N 15.04, found: C 55.95, H 5.12, N

15.08.

2-[6-(4-Fluorophenylamino)-2-(methylsulfonyl)pyrimidin-4-yl]e-

thanol (26): Compound 25 (0.448 mmol) was dissolved in a 1:1

mixture of MeOH and H₂O (6 mL). Then oxone (0.896 mmol) in

3 mL of H₂O was added portionwise, and the mixture was stirred

overnight at RT. The crude mixture was concentrated under re-

duced pressure, neutralized with aq. NaHCO₃, and washed with

EtOAc. The organic extracts were collected, dried over Na₂SO₄, and

then evaporated under reduced pressure. The crude residue was

purified by flash chromatography using a mixture of CH₂Cl₂/EtOAc

2-({2-[4-(4-Fluorophenylamino)-6-(2-hydroxyethyl)pyrimidin-2-

yl]hydrazono}methyl)phenol (30): Compound 29 (0.025 mmol)

was dissolved in dry THF (2 mL). Then Et₃N·3HF (24 mL,

0.150 mmol) was added, and the resulting mixture was stirred

overnight at RT. After addition of aq. NaHCO₃to the reaction mix-

ture, it was extracted with EtOAc, dried over Na₂SO₄, and evaporat-

ed. The crude material was dissolved in CH₂Cl₂and precipitated by

petroleum ether to give the pure product 30 without further purifi-

1

cation. Yield 73%; H NMR (400 MHz, CD₃OD): d=8.27 (s, 1H), 7.73

1

(1:1) to give the desired product 26. Yield 71%; H NMR (200 MHz,

(m, 2H), 7.43 (m, 1H), 7.26 (t, J=8 Hz, 1H), 7.1 (t, J=8 Hz, 2H),

6.92 (q, J=8 Hz, 2H), 6.17 (s, 1H), 3.94 (t, J=6 Hz, 2H), 2.8 ppm (t,

J=6 Hz, 2H); ¹³C NMR (50 MHz, CDCl₃): d=163.2, 158.8, 145.5,

131.7, 130.4, 123.5, 123.5, 120.4, 117.3, 116.6, 116.1, 99.3, 61.4, 39.5,

30.8 ppm; MS (ESI) m/z: 368.4 [M+H]⁺; Anal. calcd for

C₁₉H₁₈FN5O₂: C 62.12, H 4.94, N 19.06, found: C 62.10, H 4.98, N

19.00.

CDCl₃): d=7.55 (br, 1H), 7.35 (m, 2H), 7.1 (m, 2H), 6.55 (s, 1H), 3.95

(t, J=8 Hz, 2H), 3.3 (s, 3H), 2.85 ppm (t, J=8 Hz, 3H); ¹³C NMR

(50 MHz, CD₃OD): d=168.7, 166.4, 163.2, 162.8, 158.8, 136.1, 124.1,

116.8, 116.3, 108.6, 61.1, 41.1, 39.3 ppm; MS (ESI): m/z: 312.3

[M+H]⁺; Anal. calcd for C₁₃H₁₄FN₃O₃S: C 50.15, H 4.53, N 13.50,

found: C 50.18, H 4.47, N 13.59.

6-[2-(tert-Butyldiphenylsilyloxy)ethyl]-N-(4-fluorophenyl)-2(meth-

ylsulfonyl)pyrimidin-4-amine (27): Compound 26 (0.16 mmol) and

tert-butyldiphenylsilyl chloride (0.16 mmol) were dissolved in dry

DMF and the mixture was microwave irradiated at 1008C for 5 min.

The reaction mixture was diluted with EtOAc, washed several times

with H₂O and brine, and dried over Na₂SO₄. The crude material was

purified by flash chromatography using a mixture of CH₂Cl₂/EtOAc

Acknowledgements

This study was supported in part by the 6th FP Excellent-Hit Con-

sortium (LSHP-CT-2006-037257) and by the Italian National Pro-

gram of AIDS Research Grant 40H26 to G.M. This work was also

supported by the European Union collaborative Projects “THINC”

(grant no. HEALTH-2007-2.3.2-1), “CHAARM” (grant no. HEALTH-

F3-2009-242135) and by the Italian Ministero dell’Istruzione, del-

l’Universitꢂ e della Ricerca, Prin 2008 research project (grant no.

2008CE75SA 004). A.S. is supported by a fellowship from the

“Franca Rame and Dario Fo” Nobel Foundation. Asinex is also ac-

knowledged for partial financial support. F.B. thanks Ministero

della Sanitꢂ, Fondazione IRCCS Policlinico San Matteo, and Ricer-

ca Corrente (grants no. 80622) for financial support.

1

(9:1) to give the desired product 27. Yield 80%; H NMR (200 MHz,

CDCl₃): d=7.55 (m, 4H), 7.40 (m, 6H), 7.25 (m, 2H), 7.05 (m, 2H),

6.60 (s, 1H), 4.00 (t, J=8 Hz, 2H), 3.20 (s, 3H), 2.75 (t, J=8 Hz, 2H),

0.95 ppm (s, 9H); ¹³C NMR (50 MHz, CDCl₃): d =169.0, 165.2, 163.1,

162.2, 158.1, 135.4, 133.2, 129.7, 127.7, 125.6, 125.5, 116.8, 116.3,

105.2, 62.0, 40.7, 38.8, 29.6, 26.7 ppm; MS (ESI): m/z: 550.7 [M+H]⁺;

Anal. calcd for C₂₉H₃₂FN₃O₃SSi: C 63.36, H 5.87, N 7.64, found: C

63.43, H 5.80, N 7.60.

6-[2-(tert-Butyldiphenylsilyloxy)ethyl]-N-(4-fluorophenyl)-2-hy-

drazinylpyrimidin-4-amine (28): Compound 27 (0.2 mmol) was

dissolved in dimethoxyethane (2 mL) and the resulting mixture

was microwave irradiated at 1208C for 5 min. The solvent was

evaporated under reduced pressure and the remaining crude ma-

terial was purified by flash chromatography using a mixture of

CH₂Cl₂/MeOH (19:1) to give the desired product 28. Yield 90%;

¹H NMR (400 MHz, CDCl₃): d=7.50 (d, J=4 Hz, 4H), 7.27 (m, 6H),

7.2 (m, 2H), 6.95 (t, J=8 Hz, 2H), 5.87 (s, 1H); 3.89 (t, J=8 Hz, 2H),

2.61 (t, J=8 Hz, 2H), 0.91 ppm (s, 9H); ¹³C NMR (100 MHz, CDCl₃):

Keywords: antiviral agents · DDX3 · helicase · HIV-1 · host

cofactors · inhibitors

[1] R. Kçnig, Y. Zhou, D. Elleder, T. L. Diamond, G. M. Bonamy, J. T. Irelan,

C. Y. Chiang, B. P. Tu, P. D. De Jesus, C. E. Lilley, S. Seidel, A. M. Opaluch,

J. S. Caldwell, M. D. Weitzman, K. L. Kuhen, S. Bandyopadhyay, T. Ideker,

A. P. Orth, L. J. Miraglia, F. D. Bushman, J. A. Young, S. K. Chanda, Cell

2008 , 135 , 4 9 –6 0.

1388

www.chemmedchem.org

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemMedChem 2011, 6, 1371 – 1389

Second-Generation Inhibitors of DDX3

[2] H. Zhou, M. Xu, Q. Huang, A. T. Gates, X. D. Zhang, J. C. Castle, E. Stec,

M. Ferrer, B. Strulovici, D. J. Hazuda, A. S. Espeseth, Cell Host Microbe

2008 , 4 , 49 5 –5 04.

c) M. Radi, L. Contemori, D. Castagnolo, R. Spinosa, J. A. Estꢀ, M. Botta,

Org. Lett. 2007 , 9 , 315 7 –3 16 0; d) M. Radi, E. Petricci, G. Maga, F. Corelli,

M. Botta, J. Comb. Chem. 2005 , 7 , 11 7 –1 22.

[3] C. D. Pauza, D. N. Streblow, Curr. Top. Microbiol. Immunol. 1995, 202,

117–132.

[21] SciFinder Scholar, American Chemical Society: http://www.cas.org/prod-

ucts/sfacad/ (accessed May 13, 2011).

[4] C. M. Hogan, S. M. Hammer, Ann. Intern. Med. 2001, 134, 978–996.

[5] A. D. Kwong, B. G. Rao, K.-T. Jeang, Nat. Rev. Drug Discov. 2005 , 4 , 84 5 –

853.

[6] A. Rosner, B. Rinkevich, Curr. Med. Chem. 2007, 14, 2517–2525.

[7] M. Schrçder, Biochem. Pharmacol. 2010 , 79 , 29 7 –3 06.

[8] M. Ishaq, J. Hu, X. Wu, Q. Fu, Y. Yang, Q. Liu, D. Guo, Mol. Biotechnol.

2008 , 39 , 23 1 –2 38.

[22] GRID ver. 22, Molecular Discovery Ltd., Pinner, Middlesex, UK: http://

www.moldiscovery.com/soft_grid.php (accessed May 13, 2011).

[23] G. Wolber, T. Langer, J. Chem. Inf. Model. 2005 , 45 , 16 0 –1 69.

[24] A. Schꢅller, V. Hꢆhnke, G. Schneider, QSAR Comb. Sci. 2007 , 26 , 40 7 –

410.

[25] C. S. Lee, A. P. Dias, M. Jedrychowski, A. H. Patel, J. L. Hsu, R. Reed, Nu-

cleic Acids Res. 2008 , 36 , 470 8 –4 718.

[9] Y. Ariumi, M. Kuroki, K. Abe, H. Dansako, M. Ikeda, T. Wakita, N. Kato, J.

Virol. 2007 , 81 , 1392 2 –1 3926.

[10] V. S. Yedavalli, C. Neuveut, Y. H. Chi, L. Kleiman, K. T. Jeang, Cell 2004 ,

119 , 38 1 –3 92.

[11] G. Maga, F. Falchi, A. Garbelli, A. Belfiore, M. Witvrouw, F. Manetti, M.

Botta, J. Med. Chem. 2008 , 51 , 663 5 –6 638.

[26] G. A. Locatelli, S Spadari, G. Maga, Biochemistry 2002 , 41 , 1033 2 –1 0342.

[27] E. Crespan, M. Radi, S. Zanoli, S. Schenone, M. Botta, G. Maga, Bioorg.

Med. Chem. 2010 , 18 , 399 9 –4 008.

[28] S. Paolucci, F. Baldanti, G. Maga, R. Cancio, M. Zazzi, M. Zavattoni, A.

Chiesa, S. Spadari, G. Gerna, Antimicrob. Agents Chemother. 2004 , 48 ,

461 1 –4 617.

[12] R. Franca, A. Belfiore, S. Spadari, G. Maga, Proteins Struct. Funct. Bioinf.

2007 , 67 , 112 8 –1 137.

[29] D. Sun, M. Melegari, S. Sridhar, C. E. Rogler, L. Zhu, Biotechniques 2006 ,

41 , 5 9 –6 3.

[13] M. Hçgbom, R. Collins, S. van den Berg, R. M. Jenvert, T. Karlberg, T. Ko-

tenyova, A. Flores, G. B. Karlsson Hedestam, L. H. Schiavone, J. Mol. Biol.

2007 , 372 , 15 0 –1 59.

[30] R. Gijsbers, K. Ronen, S. Vets, N. Malani, J. De Rijck, M. McNeely, F. D.

Bushman, Z. Debyser, Mol. Ther. 2010 , 18 , 55 2 –6 60.

[31] M. Geraerts, M. Michiels, V. Baekelandt, Z. Debyser, R. Gijsbers, J. Gene

Med. 2005 , 7 , 129 9 –1 310.

[14] a) Asinex Ltd.,

5 Gabrichevskogo St., Building 8, Moscow 125367

(Russia): http://www.asinex.com (accessed May 13, 2011); b) Chem-

Bridge Corporation, 16981 Via Tazon, Suite G, San Diego, CA 92127

(USA): http://www.chembridge.com (accessed May 13, 2011).

[15] V. S. Yedavalli, N. Zhang, H. Cai, P. Zhang, M. F. Starost, R. S. Hosmane,

K. T. Jeang, J. Med. Chem. 2008 , 51 , 504 3 –5 051.

[16] L. T. Bogolyubskaya, V. A. Bogolyubskii, Z. P. Sytnik, J. Org. Chem. USSR

(Engl. Transl.) 1966, 2, 1311–1313.

[17] M. Radi, L. Botta, G. Casaluce, M. Bernardini, M. Botta, J. Comb. Chem.

2010 , 12 , 20 0 –2 05.

[32] Catalyst ver. 4.10 was formerly distributed by Accelrys Inc., and was re-

cently re-engineered for improved usability in Discovery Studio: http://

accelrys.com/products/discovery-studio/pharmacophores/

May 13, 2011).

(accessed

[33] Cerius2 ver. 4.9 was formerly distributed by Accelrys Inc., and was re-

cently re-engineered for improved usability in Discovery Studio: http://

accelrys.com/products/discovery-studio/QSAR/ (accessed May 13, 2011).

[34] MacroModel ver. 8.5, Schrçdinger LLC, New York, NY (USA), 2007:

http://www.schrodinger.com/ (accessed May 13, 2011).

[18] G. A. Rogers, T. C. Bruice, J. Am. Chem. Soc. 1974 , 96 , 246 3 –2 472.

[19] H. Jendralla, R. Henning, B. Seuring, J. Herchen, B. Kulitzscher, J.

Wunner, Synlett 1993 , 15 5 –1 57.

[35] W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc. 1996 ,

118 , 1122 5 –1 1235.

[20] a) M. Radi, S. Schenone, M. Botta, Org. Biomol. Chem. 2009 , 7 , 284 1 –

2847; b) M. Radi, C. Falciani, L. Contemori, E. Petricci, G. Maga, A. Sa-

muele, S. Zanoli, M. Terrazas, M. Castria, A. Togninelli, J. A. Estꢀ, I. Clotet-

Codina, M. Armand-Ugꢄn, M. Botta, ChemMedChem 2008 , 3 , 57 3 –5 93;

Received: March 30, 2011

Revised: April 28, 2011

Published online on June 22, 2011

ChemMedChem 2011, 6, 1371 – 1389

ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemmedchem.org

1389

Article Doi

Toward the Discovery of Novel Anti-HIV Drugs. Second-Generation Inhibitors of the Cellular ATPase DDX3 with Improved Anti-HIV Activity: Synthesis, Structure-Activity Relationship Analysis, Cytotoxicity Studies, and Target Validation

DOI: 10.1002/cmdc.201100166

Source and publish data:

Authors:

Article abstract of DOI:10.1002/cmdc.201100166

Full text of DOI:10.1002/cmdc.201100166

Products guided by the article

R&D Labs maybe for 942045-32-1

Relevant to this article

Hot Product