Novel HIV-1 Non-nucleoside Reverse Transcriptase Inhibitor Agents: Optimization of Diarylanilines with High Potency against Wild-Type and Rilpivirine-Resistant E138K Mutant Virus

Article

pubs.acs.org/jmc

Novel HIV‑1 Non-nucleoside Reverse Transcriptase Inhibitor Agents:

Optimization of Diarylanilines with High Potency against Wild-Type

and Rilpivirine-Resistant E138K Mutant Virus

Na Liu,^†Lei Wei,^†Li Huang,^‡Fei Yu,^§,∥Weifan Zheng,^⊥Bingjie Qin,^†Dong-Qin Zhu,^†

Susan L. Morris-Natschke,^#Shibo Jiang,^§,∥Chin-Ho Chen,^‡Kuo-Hsiung Lee,*^,#,@and Lan Xie*^,†,#

^†Beijing Institute of Pharmacology & Toxicology, 27 Tai-Ping Road, Beijing 100850, China

^‡Surgical Oncology Research Facility, Duke University Medical Center, Box 2926, Durham, North Carolina 27710, United States

^§Key Laboratory of Medical Molecular Virology of Ministries of Education and Health, Shanghai Medical College and Institute of

Medical Microbiology, Fudan University, Shanghai 200032, China

^∥Lindsley F. Kimball Research Institute, New York Blood Center, New York, New York 10065, United States

^⊥Department of Pharmaceutical Sciences & BRITE Institute, North Carolina Central University, Durham, North Carolina 27707,

United States

^#Natural Products Research Laboratories, UNC Eshelman School of Pharmacy, University of North Carolina, Chapel Hill,

North Carolina 27599-7568, United States

^@Chinese Medicine Research and Development Center, China Medical University and Hospital, Taichung, Taiwan

S

* Supporting Information

ABSTRACT: Three series (6, 13, and 14) of new diarylaniline

(DAAN) analogues were designed, synthesized, and evaluated

for anti-HIV potency, especially against the E138K viral strain

with a major mutation conferring resistance to the new-

generation non-nucleoside reverse transcriptase inhibitor drug

rilpivirine (1b). Promising new compounds were then assessed

for physicochemical and associated pharmaceutical properties,

including aqueous solubility, log P value, and metabolic stability,

as well as predicted lipophilic parameters of ligand eﬃciency,

ligand lipophilic eﬃciency, and ligand eﬃciency-dependent

lipophilicity indices, which are associated with ADME property

proﬁles. Compounds 6a, 14c, and 14d showed high potency

against the 1b-resistant E138K mutated viral strain as well as good balance between anti-HIV-1 activity and desirable druglike

properties. From the perspective of optimizing future NNRTI compounds as clinical trial candidates, computational modeling

results provided valuable information about how the R¹group might provide greater eﬃcacy against the E138K mutant.

INTRODUCTION

■

Because of their high potency and low toxicity, non-nucleoside

reverse transcriptase inhibitors (NNRTIs) are used as key drugs

in highly active antiretroviral therapy (HAART)¹for the

treatment of HIV infection and AIDS. However, resistance to

some drugs, especially to the early NNRTI drugs nevirapine,

delavirdine, and efavirenz, has occurred as a result of rapidly

developing mutations of the viral NNRTI binding site. To

address this major problem, two next-generation NNRTI drugs,

etravirine (TMC125, 1a)²and rilpivirine (TMC278, 1b)³

(Figure 1), were successfully developed and approved by the

U.S. Food and Drug Administration (FDA) in 2008 and 2011,

respectively. Compounds 1a and 1b belong to the same

diarylpyrimidine (DAPY) family, possess extremely high

potency against the wild type and a broad spectrum of mutated

NNRTI strains resistant to early NNRTI drugs,⁴and also

present a higher genetic barrier against HIV-1 resistance,^5,6

Figure 1. New-generation HIV-1 NNRTI drugs approved by the U.S.

FDA.

most likely resulting from their molecular ﬂexibility. However,

because of the high mutation rate of HIV-1 RT and the lack

of intrinsic exonucleolytic proofreading activity, new resistance

Received: November 24, 2015

A

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 2. Prior leads 2 and 3, known SARs, and the current lead optimization strategy.

proﬁles associated with 1a and 1b are also being observed in

patients.^7,8These viral mutations are distinct from each other⁹

as well as distinct from those seen with the early NNRTI drugs.

Consequently, novel NNRTI drugs with high potency and

diﬀerent molecular scaﬀolds are still needed to overcome new

drug resistance regimes associated with the new-generation

NNRTI anti-HIV drugs 1a and 1b and to provide more

eﬃcacious therapies and treatment options.

antiresistance potency and good balance between antiviral

activity and multiple druglike properties. Guided by the prior

SAR results, we designed and synthesized three new series

(6, 13, and 14) of DAAN derivatives and evaluated them

against HIV-1 wild-type virus and the E138K variant, which

contains the most frequent mutation conferring resistance to

1b,¹⁵as well as other resistant viral strains. Meanwhile, essential

druglike properties and parameters were also assessed and

predicted. Herein, we reported our new results and further dis-

cussion of SAR and SPR correlations for DAANs and DAPAs.

In our prior studies of novel HIV-1 NNRTI agents, we

discovered two series of extremely potent anti-HIV NNRTI

agents, diarylanilines (DAANs, 2)¹⁰and diarylpyridiamines

(DAPAs, 3),^11,12as shown in Figure 2. The new chemo-type

scaﬀolds have topologic shape and molecular ﬂexibility similar

to those of new-generation NNRTI drugs 1a and 1b. The initial

leads 2a and 3a exhibited high anti-HIV potency against wild-

type HIV-1 replication with low to subnanomolar EC₅₀values

of 3.0 and 0.71 nM, respectively, comparable to those of 1a

(EC₅₀= 1.5 nM) and 1b (EC₅₀= 0.51 nM) in the same

assays.^10,12However, very low bioavailability (F% < 3) in rats

impeded their further development. Subsequently, multiple

structural modiﬁcations were introduced to identify structure−

activity relationship (SAR) and structure−property relationship

(SPR) correlations for DAANs and DAPAs as new HIV-1

NNRTI agents. The subsequent leads 2b¹³(DAAN) and 3b¹⁴

(DAPA) exhibited not only extremely high potency against

both wild-type (EC₅₀values of 0.39 and 3.25 nM, respectively)

and drug-resistant (EC₅₀values of 1.8−3.4 and 3.5−6.9 nM,

respectively) viral replication but also improved druglike pro-

perties. Lead 2b had a desirable log P value of 3.68 and a high

water solubility of 90 μg/mL at pH 2.0, but its higher meta-

bolism and instability in various solvents prompted continued

structural optimization. On the other hand, lead 3b had high

antiresistance potency, with especially low resistance fold

changes (FC = 1.1−2.1) for K101E and E138K viral strains

resistant to new drugs 1a and 1b, respectively, compared with

that for wild-type NL4-3 virus. This interesting advantage

provided the opportunity to explore compounds capable of

overcoming the drug resistance associated with 1a and 1b.

Therefore, current optimization eﬀorts have been aimed at

discovering new DAANs as potential drug candidates with high

DESIGN

■

As demonstrated in Figure 2, while potent NNRTI DAANs

and DAPAs diﬀer by having a benzene and pyridine ring,

respectively, as the central ring (B-ring), they also have three of

the same known pharmacophores, a p-cyanoaniline moiety

(A-ring), the NH₂group on the central B-ring ortho to the

A-ring, and a 2,4,6-trisubstituted phenoxy ring (C-ring). With

good molecular ﬂexibility and a topologic shape similar to that

of 1b, these compounds can adjust their binding conformations

to ﬁt the binding site within both wild-type and mutated

NNRTI binding pockets. Previous SARs revealed that a linear

hydrophobic para-R²substituent on the C-ring is favorable for

potency with the following rank order: cyanoethyl > cyanovinyl

> cyano. On the other hand, the R¹substituent on the central

B-ring and X substituents on the C-ring can be modiﬁed

to improve molecular properties without a loss of potency.

Importantly, on the basis of prior molecular docking results,¹²

the R¹substituent on the B-ring is outside of the main NNRTI

binding pocket and close to susceptible amino acid E138 on the

p51 subdomain of the HIV-1 reverse transcriptase. Thus, modiﬁ-

cation of this R¹group could provide an interaction between it

and the mutated K138, or other preserved amino acid(s) nearby

or on the interface between p66 and p51 subdomains, to possibly

overcome mutated E138K resistance. Therefore, in the newly

designed DAAN analogues (6, 13, and 14 series), continued

optimizations addressed modiﬁcations of the R¹and X sub-

stituents, while the p-cyanoethyl substituent (R²) on the phenoxy

ring (C-ring)¹⁶and p-cyanoaniline (A-ring) were maintained

(see the formula in Figure 2). Because the R¹substituent is also

B

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

a

Scheme 1

a

(i) THF/MeOH, aq NaOH, rt, 10 h; (ii) (a) SOCl₂, reﬂux, 4 h; (b) amine, THF, 0 °C, 0.5 h; (iii) LiBH₄, THF/MeOH, 0 °C, 7 h; (iv) 2,4,6-

trichloro[1,3,5]triazine, DMF/CH₂Cl₂, rt, 4 h; (v) amine, THF, 0 °C, for 5e−g; (vi) Ac₂O, NaOH(s), 100 °C, microwave, 5 min for 5h;

(vii) acylchloride, CH₂Cl₂/Py, for 5i and 5j; (viii) MeOH, BiCl₃, in CH₂Cl₂, rt, 4 h, for 5k; (ix) H₂/Pd-C, in EtOH or EtOAc.

associated with molecular physicochemical properties, the

current optimization strategy should result in a balance between

antiviral potency and multiple drug-likeness proﬁles, which are

criteria for advancing a potential drug candidate to the clinic.

Additionally, a suitable R¹side chain oriented toward the

interface between the HIV-1 RT p66 and p51 subdomains could

overcome the drug resistance of E138 K. The series 6 DAAN

compounds incorporated various R¹groups with diﬀerent shapes,

volumes, and ﬂexibilities, including amides (6a−d), alkylamines

(6e−g), esters (6h,i), carbamates (6j), alkoxyethers (6k,m),

and halogens (6n,p), to investigate the eﬀects of these groups

on potency, antiresistance, and druglike properties. Hetero

atoms, such as O or N, in certain R¹substituents could provide

additional interactions with the amino acids on interfaces

between p66 and p51 subdomains of the RT enzyme via

H-bonds or other forces to increase potency against the E138K

mutation and improve molecular physicochemical properties,

as well as related druglike properties. In addition, the X groups

on the C-ring were modiﬁed for antiresistance and metabolic

stability,^17,18resulting in two small sets of new DAANs, the

dimethoxy-13 series and the methoxy/bromo-14 series, with

either an N-methyl carboxamide or a methyl carboxylate as the

R¹substituent. These limited structural optimizations should

provide a better understanding of how structure aﬀects potency

against HIV-1 wild-type and resistant viral strains, especially for

the E138K mutation, as well as other drug properties, with the

aim of discovering and developing potential new NNRTI drug

candidates.

compounds 5a−c. On the other hand, treating 4b with 2,4,6-

trichloro[1,3,5]triazine (TCT)¹⁹produced the chloromethyl

product 4c, which was reacted immediately with dimethylamine,

cyclopropanamine, and 1-methylpiperazine to yield the cor-

responding 4-alkylaminomethyl compounds 5e−g, respectively.

Alternatively, compound 4b was reacted with acetic anhydride,

cyclopropanecarbonyl choloride, and methylcarbamic chloride

to yield the corresponding reversed ester compounds 5h and 5i

and carbamate compound 5j, respectively. Moreover, 4b was

converted to methoxymethyl compound 5k by reaction with

methanol in the presence of BiCl₃at room temperature. Next,

the nitro group and the double bond of the p-cyanovinyl in

intermediates 5a−c and 5e−k were reduced simultaneously by

catalytic hydrogenation with Pd-C (10%) in anhydrous ethanol

or EtOAc to produce the new DAAN compounds 6a−c and

6e−k, respectively.

In addition, a new convergent synthetic route was explored to

eﬃciently synthesize additional p-cyanoethylphenoxy-DAANs

(Schemes 2 and 3). The C-ring building block 4-cyanoethyl-2,6-

a

Scheme 2

a

(i) BnCl, K₂CO₃or Et₃N, in CH₃CN, reﬂux, 3 h; (ii) (EtO)₂PCO-

Chemistry. As shown in Scheme 1, new compounds 6a−c

and 6h−n were synthesized from 2m, a prior intermediate in

the synthesis of DAANs.¹³The 4-ester group on the central

phenyl ring in 2m was either hydrolyzed under basic conditions

or reduced with LiBH₄to the corresponding carboxylic acid

(4a) or hydroxymethyl (4b) derivative, respectively. Com-

pound 4a was then treated with SOCl₂followed by amidation

with diﬀerent amines in THF at 0 °C to produce the 4-amido

(CH₂CN), t-BuOK/THF, 0 °C, 1 h; (iii) H₂/Pd-C in EtOAc or EtOH.

dimethyl phenol (9a) was prepared from commercial reagent 7a

in three steps: benzoylation of the phenol, Wittig reaction of

the aldehyde group with diethyl cyanomethylphosphonate, and

deprotection by catalytic hydrogenation. Because an electron-

donating p-cyanoethyl group was present, the coupling of 9a with

C

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

a

Scheme 3

a

(i) Cs₂CO₃/DMF, 80 °C, microwave, 20 min; (ii) K₂CO₃/DMF, 110 °C, heating for 4 h or microwave, 20 min; (iii) aq NaOH, THF/MeOH, rt,

8 h; (iv) LiBH₄, THF/MeOH, 0 °C, 7 h; (v) (a) SOCl₂, reﬂux, 3 h; (b) amine/THF, 0 °C, 0.5−1 h; (vi) (a) DPPA/Et₃N/THF, rt, 3 h; (b) H₂O,

reﬂux, 3 h; (vii) (a) HBr/NaNO₃/H₂O, 5 °C, 1 h; (b) CuBr/HBr, 5 °C, 3 h; (viii) 2,4,6-trichloro[1,3,5]triazine, DMF, CH₂Cl₂, rt, 4 h; (ix) EtOH,

microwave at 100 °C, 20 min; (x) H₂/Pd-C in EtOAc, 2-3 h, rt; (xi) Fe/NH₄Cl, THF/MeOH/H₂O, reﬂux, 3 h.

10a, an intermediate synthesized and identiﬁed previously,^10,13

was performed easily to produce 2,4-diaryl nitrobenzene 11a

in a high yield of 94%. By conversions similar to those shown

in Scheme 1, 11a was converted to the carboxylic acid (11b)

and hydroxymethyl (11c) compounds. Compound 11b was

converted to the N-substituted carbamoyl product 11d, while

11c was modiﬁed successively to the chloromethyl (11e) and

ethoxymethyl (11m) nitrodiarylanilines. By Curtius rearrange-

ment,²⁰the 4-carboxylic acid group of 11b was converted to

a 4-amino moiety in 11f, which underwent diazonization fol-

lowed by reaction with cuprous bromide to produce the desired

4-bromo product 11n and the nonbrominated byproduct 11o.

Alternatively, coupling of 9a with 10b was performed under

microwave irradiation in DMF in the presence of K₂CO₃at

110 °C for 10 min to produce 4-chloro compound 11p

(R₁= Cl). Finally, nitro compounds 11d and 11m were reduced

conveniently by catalytic hydrogenation to corresponding amino

products 6d and 6m, respectively. However, halogenated nitro

compounds 11n−p were reduced by using the mild reductive

reagent Fe/NH₄Cl to obtain corresponding amino compounds

6n−p, respectively. This reaction showed a clear color change

from yellow to colorless to identify the reaction’s end.

(X = OMe/Br) were synthesized from compound 10a by

coupling with 9b or 9c to produce corresponding esters 13a

and 14a followed by amidation with methylamine to yield

13b and 14b, respectively. Similar to previous reductions, the

4-nitro in compounds 13a and 13b was reduced by catalytic

hydrogenation, whereas halogenated compounds 14a and 14b

were reduced with Fe/NH₄Cl to generate corresponding

DAANs 13c, 13d, 14c, and 14d, respectively. All new target

compounds in the 6, 13, and 14 series were identiﬁed by

¹H NMR and mass spectrometry data.

RESULTS AND DISCUSSION

■

Antiviral Activity and SAR Analysis. All new DAAN

compounds (6, 13, and 14 series) were evaluated against

wild-type HIV-1 (NL4-3 and IIIB) and resistant viral strain

(E138K, A17, and K101E) replication in TZM-bl (NL4-3 and

E138K) and MT-2 (IIIB) cell lines, respectively. Their anti-

HIV potency (EC₅₀, as measured by a luciferase gene expres-

sion assay²⁹) and cytotoxicity (CC₅₀) as well as selective index

(SI) data are summarized in Table 1. As expected, with the

exception of 6d and 6g, most of the 6 series compounds with

various R¹substituents exhibited extremely high potency against

wild-type HIV-1 NL4-3 replication in TZM-bl cell lines with

low nanomolar EC₅₀values ranging from 0.53 to 16.0 nM.

These results were consistent with our previous ﬁndings, which

indicated that the R¹group can be modiﬁed without loss of anti-

HIV potency. However, the presence of very bulky R¹groups,

With the same method that was used for the preparation of

9a, building blocks 4-cyanoethyl-2,6-dimethoxy phenol (9b)

and 2-bromo-4-cyanoethyl-6-methoxyphenol (9c) were synthe-

sized from commercially available 7b and 7c, respectively.

As described in Scheme 4, compounds 13 (X = OMe) and 14

D

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

a

Scheme 4

a

(i) K₂CO₃/DMF, 120 °C, 4 h; (ii) aq CH₃NH₂/THF, 40 °C, 4 h; (iii) H₂/Pd-C in EtOAc for 13; (iv) Fe/NH₄Cl, THF/MeOH/H₂O, reﬂux, for 14.

e.g., 4-(N-isopropyl)carbamoyl in 6d and 4-(N-methylpiperazine)-

carbamoyl in 6g, impeded molecular antiviral potency, resulting

in EC₅₀values of 72 and 115 nM, respectively. We supposed

that steric conﬂict between the bulky R¹groups and the viral

amino acid residues within a limited space between the p51 and

p66 domains might aﬀect intermolecular aﬃnity and binding,

leading to reduced potency. Interestingly, the three most potent

compounds, 6c (EC₅₀= 2.00 nM), 6f (EC₅₀= 0.53 nM), and 6i

(EC₅₀= 1.63 nM), all contain a cyclopropyl moiety at the end

of their R¹side chains; this moiety might have a suitable shape

and volume to insert into a cleft between the p66 and p51

subdomains, thereby increasing molecular aﬃnity. When the

two X groups on the C-ring were changed, the dimethoxy-

DAANs [13c (R¹= COOCH₃) and 13d (R¹= CONHCH₃)]

were at least 20-fold less potent than the corresponding bromo/

methoxy-DAANs (14c and 14d). Indeed, compounds 14c and

14d exhibited similar high potency (EC₅₀values of 5.19 and

5.58 nM, respectively) against wild-type NL4-3 virus replication

and were almost as potent as the prior corresponding dimethyl-

DAANs 2c (EC₅₀= 4.32 nM) and 2d (EC₅₀= 2.73 nM).¹³

Therefore, molecular potency could be maintained by replacing

the two methyl groups (X) on the C-ring with methoxy

and bromo groups, but not with two methoxy groups. Thus,

appropriate X groups are needed to achieve a favorable binding

conformation of the C-ring moiety.

The active DAANs described above (6, 13, and 14 series)

were also tested in the MT-2 cell line for inhibition of wild-type

HIV-1 IIIB, as well as the multiple-resistant A17 viral strain,

replication, and cytotoxicity. The EC₅₀values against two wild-

type viruses (NL4-3 and IIIB) in diﬀerent cellular assays showed

some ﬂuctuation, but similar potency and activity patterns were

observed as shown in Table 1. Meanwhile, the low cytotoxicity

(CC₅₀> 10 μM) of the compounds resulted in extremely high

selective index values ranging from 10³to 10⁵(Table 1).

To evaluate our hypothesis concerning a possible link

between the R¹substituent of a compound and its eﬀectiveness

against NNRTI-resistant virus, active DAAN compounds were

tested against the E138K mutated viral strain, which has a

major mutation conferring resistance to 1b, and the data are

summarized in Table 1. Comparison of the results in the wild-

type NL4-3 and drug-resistant E138K assays identiﬁed the most

promising compounds as 6a, 6b, 14c, and 14d, as indicated

by low resistance FCs of 1.9−2.2, which were much better

than that of 1b (FC = 10), lower than those of the other new

DAANs, and comparable with those of the prior 2c (FC = 1.7)

and 2d (FC = 3.7) in the same assay. Notably, 6a, 6b, 14c, and

14d, as well as 2c and 2d, all contain a carbonyl (CO) group

conjugated with the B-ring, regardless of whether the R¹group

is an amide or ester. Thus, this carbonyl group might be

important for inhibiting the E138K mutation, possibly because

it could interact with the mutated K138 or nearby amino

acid(s). Support for this postulate was found from the results

for 6e, 6f, 6h, 6i, and 6k, in which the carbonyl group was

changed to a methylene (-CH₂-). Despite their high potency

(EC₅₀= 0.53−3.3 nM) against wild-type NL4-3 virus, these

compounds were much less potent against the E138K viral

strain (EC₅₀> 13.3−39 nM), leading to poorer resistance FC

values of 9−48 compared with those of 6a, 6b, 14c, and 14d.

Thus, the presence of a conjugated carbonyl group in R¹on the

B-ring is favorable against the E138K drug 1b resistance. In

addition, representative compounds 6b, 6p, and 2d, which had

markedly improved drug-resistant proﬁles for the E138K mutant,

were tested against another 1b-resistant viral strain, K101E. Again,

improved drug-resistant proﬁles with low FC values of 3−4 were

seen (Table 1). Moreover, eight active compounds (EC₅₀< 5 nM

against IIIB) were further tested in the MT-2 cell line against the

multi-NRTI-resistant A17 viral strain with mutants K103N and

Y181C, the two major mutants from the ﬁrst-generation NNRTI

drugs. Except 6a and 6p, most of them showed high potency

against the drug-resistant A17 viral strain with low nanomolar

EC₅₀values and low FC values ranging from 0.9 to 14 [lower

than that of 1b (FC = 18)] compared with wild-type IIIB virus in

the same cellular assay, indicating that these compounds are also

eﬀective against multidrug resistance to the ﬁrst-generation

NNRTI drugs. Therefore, these results demonstrate that the

R¹substituent is the key modiﬁcation point for overcoming the

1b drug resistances and other multi-NRTI-resistant strains.

Molecular Modeling. A computational study was per-

formed to demonstrate how the introduction or modiﬁcation of

a R¹substituent could contribute to overcoming E138K-

induced drug resistance to 1b. If DAANs bind to both wild-type

and mutant viruses such that the R¹group is oriented toward

virus residue 138, then modiﬁcation of its size, shape, and other

properties could aﬀord compounds that would interact with the

mutant virus better than drug 1b, which could be measured

analytically in terms of lowered FC values. Because of their

high wild-type potency (single-digit nanomolar levels) with

good FC values (2-fold), compounds 6a and 14d were selected

as representatives for the computational study. Again, the latter

result is highly desirable for combating resistance in the E138K

mutant virus. The compounds were docked computationally to

both wild-type RT/1b [Protein Data Bank (PDB) entry 3MEC]²¹

and mutated E138K RT/nevirapine (PDB entry 2HNY)²²

complex crystal structures. For comparison, drug 1b was also

docked to both the wild-type and mutant structures. The docking

protocol is described in the computational section.

E

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Table 1. Anti-HIV Activity Data of 6, 13, and 14 Series against Wild-Type HIV-1 and Mutated Strains

a

Concentration of compound that causes 50% inhibition of viral infection, presented as the mean standard deviation (SD) and determined in at

b

c

least triplicate against HIV-1 virus in TZM-bl cell lines. Against HIV-1 virus in MT-2 cells. NL4-3 and IIIB are wild-type HIV-1 viral strains.

d

Resistant viral strains E138K and K101E mutated in the HIV-1 RT non-nucleoside inhibitor binding pocket (NNBP). A17 from NIH is the multi-

e

NRTI-resistant strain with mutations K103N and Y181C and is highly resistant to NRTIs. Resistance fold change (FC) is the EC_50(E138K)

/

f

EC_50(NL4‑3), EC_50(K101E)/EC_50(NL4‑3), or EC_50(A17)/EC_50(IIIB)ratio. The FC of an NL4-3 mutant with the K101E mutation in HIV-1 RT.

g

Concentrations that cause cytotoxicity to 50% of cells; CytoTox-Glo cytotoxicity assays (Promega) were used, and values were averaged from two

h

i

j

independent tests. The selective index (SI) is the CC₅₀/EC_50(IIIB)ratio. Not determined. Prior compounds published. See ref 13.

F

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Figure 4. Predicted and overlapped binding modes of 14d (green

sticks) and 13d (red sticks) with the HIV-1 RT wild-type crystal

structure (3MEC). The key amino acid side chains surrounding the

C-ring are shown in gray stick format and labeled. The distance

between ligands and protein is <3 Å.

B-ring in both molecules superimposed well, but the C-ring was

tilted upward in 13d because of the steric eﬀect of the bulkier

methoxy group (outside) on this ring. We suppose that the

C-ring in 13d is thus forced away from the Y181 or Y188

residue, reducing the molecular aﬃnity for the viral target and

then decreasing the potency. However, the C-ring in 14d

superimposed completely with that of active compound 6f

[X = Me/Me (data not shown)], as the bromine atom in 14d is

similar in size to the methyl (X) in series 6. Furthermore, less

active 6g and active 6f were docked in the wild-type RT crystal

structure (3MEC) following the same docking protocol. The

bulkier group on R¹in 6g pushed more against E138, resulting

in a minor shift in 6g’s binding mode compared with that of 6f

(Figure in the Supporting Information). This shift may aﬀect

the critical interaction between the C-ring and residues W229,

Y181, and Y188, thus providing support for our previous

supposition for why 6g was less potent than 6f.

Druglike Properties and Parameters. Active DAANs

6a−c, 6e, 6f, 6h, 6i, 6k, 6n, 14c, and 14d (EC₅₀< 10 nM) were

further assessed for certain physicochemical properties re-

quired for potential drug candidates to attain a critical balance

between potency and druglike properties. According to methods

described previously,^13,23aqueous solubility at pH 7.4 and 2.0,

log P values, and in vitro metabolic stability were measured, and

the results are summarized in Table 2. Compounds 6e and 6f

with an alkylamine-R¹side chain displayed good molecular

solubility at both pH 2.0 (260 and 276 μg/mL, respectively)

and pH 7.4 (23.4 and 13.6 μg/mL, respectively), typical

physiological environments encountered by a drug in stomach

and plasma, respectively. However, the carbamoyl-DAANs 6a,

6b, 6c, and 14d showed reasonable aqueous solubility (25.8−

141 μg/mL) only at pH 2.0, regardless of whether X is Me/Me

or MeO/Br. These results indicated that the increased mole-

cular alkalescence aﬀorded by the N atom in the R¹substituent

of 6e and 6f led to improved molecular aqueous solubility,

especially in an acidic medium. In contrast, the compounds with

oxygen (O) or halogen atoms in the R¹side chain [see 6h and

6i (R¹= CH₂OCOR′), 6k (R¹= CH₂OMe), 6n (R¹= Br), and

14c (R¹= COOMe, and X = Br/OMe)] exhibited lower

aqueous solubility ranging from 0.12 to 2.63 μg/mL at both pH

2.0 and 7.4. In total, the current results indicated the follow-

ing rank order of aqueous solubility based on R₁side chain:

alkylamine > amide > ester and ether groups. Improved

molecular aqueous solubility under acidic conditions should

enhance oral drug absorbability in the stomach. Along with

aqueous solubility, lipophilicity is another important druglike

Figure 3. Predicted binding modes of 6a (green sticks) and ligand 1b

(gray sticks) with the HIV-1 RT wild-type crystal structure (3MEC)

(top) and with the E138K mutant crystal structure (2HNY) (bottom),

respectively, and overlapped. Both molecules 1b and 6a form two

characteristic hydrogen bonds with K101 in both situations.

Surrounding key amino acid side chains are shown in gray ball-and-

stick format and labeled. Hydrogen bonds are shown as yellow dashed

lines, and the distance between ligands and protein is <3 Å.

As shown in Figure 3, the docking poses of 6a (green) and

1b (gray) in both 3MEC (wild-type) and 2HNY (E138K

mutant) structures, respectively, displayed similar binding

orientations with “U”-like conformational shapes and two

characteristic hydrogen bonds formed with the key amino acid

K101, consistent with prior results and literature reports on 1b.

In both cases, the R¹side chain of 6a pointed toward the

domain space containing viral residue 138. The greater ﬂexibility

of K138 could make the mutant virus more tolerant than the

wild-type virus toward various R¹side chain substitutions, as

reﬂected by better FC values. Ten of the 14 docking cases were

consistent with this observation. Moreover, as seen in the

bottom half of Figure 3, the carbonyl group in the R¹substituent

of 6a was oriented toward the amino group in the side chain

of K138. Even though a clear interaction between the carbonyl

in R¹and the amino group of K138 was not observed in the

current models, additional H-bond(s) with a bridging water

molecule could be formed to reach high aﬃnity with the mutant

target. The second representative compound 14d provided

similar results, and its docking ﬁgures can be found in the

Supporting Information. Therefore, the current modeling results

supported our hypothesis that the R¹substituent is associated

with the potency of the NNRTIs against viral mutants with

resistance to NNRTIs and should, therefore, help in the further

optimization of new drug candidates that can overcome the drug

resistance to current NNRTI drugs.

On the other hand, to understand why a small structural

change made a signiﬁcant reduction in potency, we compared

the docked models of both less active 13d (X = MeO/MeO)

and the corresponding active 14d (X = MeO/Br) into the wild-

type structure (3MEC). As shown in Figure 4, the A-ring and

G

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

Table 2. Physicochemical Parameters of New Active DAAN Compounds

aqueous solubility

a

c

(μg/mL)

HLM

a

b

d

e

f

h

pH 2.0

pH 7.4

log P at pH 7.4

t-PSA (Å²)

LE

LLE

LELP

8.67

10.7

maint. % (30 min)

t_1/2(min)

6a

65.1

69.4

141

1.08

1.94

23.4

13.6

1.86

1.64

0.66

0.12

0.20

3.38

0.24

2.86

3.63

2.61

2.52

3.69

2.84

2.51

3.25

>5

115

124

98.1

107

121

104

94.9

140

133

97.4

0.33

0.34

0.36

0.37

0.34

0.33

0.38

0.39

0.33

0.46

5.35

4.70

6.09

6.04

5.59

5.64

6.28

5.59

<3.49

5.43

5.09

5.69

72.2

85.2

63.2

93.3

78.4

3.00

4.18

90.8

58.3

81.5

75.7

80.2

182

99

173

63

6b

6c

7.68

7.00

9.97

8.35

7.61

8.55

6e

260

231

99

6f

276

6h

2.63

1.24

1.82

0.25

1.73

25.8

74.1

10.2

11.7

231

50

6i

6k

6n

>12.8

14c

2.85

3.16

>5

8.64

9.57

7.87

139

116

14d

h

1b

g

propranolol

89.1

a

b

c

d

Measured by HPLC in triplicate. Topological polar surface area predicted by ChemDraw Ultra 12.0. Human liver microsome assay. Calculated

by the formula −ΔG/HA_{(non‑hydrogen atom)}, in which normalizing binding energy ΔG = −RT ln K_d, presuming K_d≈ EC₅₀; negative logarithm values of

e

f

potency (pEC₅₀) converted from experimental data against wild-type virus IIIB listed in Table 1. Calculated by the formula pEC₅₀− log P. Deﬁned

g

h

as a ratio of measured log P and LE. A reference compound with moderate metabolic stability with a t_1/2of 3−5 h in vivo. Calculated with ﬁve time

points as 0, 5, 15, 30, and 45 min.

property related to molecular biological potency, as well as

absorption, distribution, metabolism, and excretion (ADME)

proﬁles and toxicity risk. As shown in Table 2, except for 6n

(log P > 5), all of the active compounds tested had desirable

log P values within a range of 2.51−3.69 at pH 7.4. Therefore,

insertion of hetero atoms into the R¹substituent served as an

eﬀective strategy for decreasing lipophilicity, while maintaining

high potency, improving aqueous solubility, and decreasing the

high risk of unwanted pharmacology.

incubation for 30 min, with short t_1/2values of 10.2 and

11.7 min, respectively. It is likely that the benzyl ester in the R¹

side chain of both compounds is easily metabolized by enzymatic

catalysis. In contrast, the remaining tested compounds

maintained >50% of the original compound after incubation

for 30 min. Among them, compounds 6e and 6k showed greater

metabolic stability than propranolol, a common reference

compound with moderate metabolic stability. Compounds 6b,

14c, and 14d were less stable than propranolol but more stable

than 1b (t_1/2= 116 min). Compounds 6a and 6f (t_1/2= 99 min)

had stability similar to that of 1b. Compounds 6c and 6n (R¹=

Br) (t_1/2values of 63 and 50 min, respectively) were less stable

than 1b. Notably, both bromo compounds 14c and 14d (X =

Br/OMe) showed reasonable metabolic stability (t_1/2= 139 min)

in the HLM assay, indicating that the halogen Br is acceptable

at the X position on the C-ring instead of a methyl group.

We further postulate that series 14 might have improved drug

absorption leading to better oral bioavailability, because the high

electronegativity of the introduced bromo atom (Br) could the

balance molecular aqueous solubility and lipophilicity.

Most recently, concepts such as ligand eﬃciency (LE),²⁴

ligand lipophilic eﬃciency (LLE),²⁵and ligand eﬃciency-

dependent lipophilicity (LELP)²⁶have been proposed and

are being applied increasingly in many medicinal chemistry

programs to usefully direct optimization away from highly

lipophilic compounds. Deﬁned as the diﬀerence between the

negative logarithm of the measured potency (pEC₅₀) and log P,

the LLE value is a simple but important index combining

in vitro potency and lipophilicity, whereas LELP correlates with

ADME properties. Accordingly, lipophilic parameters of the

active DAANs were calculated by using experimental EC₅₀and

log P values with the formulas cited at the bottom of Table 2.

Fortunately, most compounds, including 6a, 6c, 6e, 6f, 6h, 6i,

6k, 14c, and 4d, met acceptable levels for all three ligand

lipophilic eﬃciency indices (LE > 0.3, LLE > 5, LELP < 10),²⁷

while the remaining compounds in Table 2 did not. The com-

pounds with a low LELP value would most likely have the

greatest chance of passing all ADME and safety criteria.

Additionally, topological polar surface area (t-PSA) parameters

of active compounds were calculated with ChemDraw

Ultra 12.0 and met the criterion²⁸of <140 Å². These property

assessment results indicated the postulate that optimization of

optimized R¹and X substituents could satisfy the dual goals

of maintaining high antiviral potency and improving druglike

properties at the same time by adjusting molecular lipophilicity.

Furthermore, the metabolic stability of active compounds

was evaluated in a human liver microsome (HLM) assay; 1b

and propranolol were also tested for comparison. As shown in

Table 2, 6h and 6i were metabolized quite quickly, maintaining

only 3.0 and 4.18%, respectively, of the parent compound after

CONCLUSIONS

■

Current structural modiﬁcation studies of DAANs showed

that optimization of the R¹and X substituent on the central

phenyl ring (B-ring) and phenoxy ring (C-ring), respectively,

could enhance, or maintain, high potency against wild-type

and E138K mutant HIV-1 replication and improve molecular

aqueous solubility, lipophilicity, and metabolic stability, thus

reaching a good balance between high antiviral potency and

multiple druglike properties. On this basis, compounds 6a,

14c, and 14d might be potential drug candidates for further

development to address new drug resistance deﬁciencies.

Current SAR study further revealed that (1) a carbonyl group

in the R¹substituent conjugated with the central phenyl ring

(B-ring) is necessary to combat drug resistance from the E138K

mutant, (2) hetero atoms (O, N, or Br) in the R¹substituent

favor improved druglike ADME properties, (3) bulky R¹groups

should be avoided, and (4) Br and OMe could replace the two

original methyl groups in DAANs, but two methoxy groups (X)

H

DOI: 10.1021/acs.jmedchem.5b01827

J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

0.50 mmol) in THF (5 mL) was added LiBH₄(36 mg, 0.16 mmol) at

0 °C while the mixture was being stirred over 30 min. After MeOH

(13 drops) was added dropwise, the mixture was being stirred at room

temperature for 1 h. The mixture was poured into ice−water, and the

pH was adjusted to 2−3 with aq HCl (5%). The solid crude product

was ﬁltered out and puriﬁed by CombiFlash column chromatography

(gradient elution, EtOAc/CH₂Cl₂, 0 to 3%) to produce 200 mg of 4b:

on the C-ring led to reduced potency, because a steric eﬀect

altered the favorable conformation of the C-ring. Most

interestingly, we found that the introduced R¹side chain

provided a possibility for overcoming the new drug resistance

from E138K, a major HIV-1 viral mutant with resistance to the

NNRTI drug 1b. For instance, compounds 6a, 6b, 14c, and

14d exhibited high potency against the wild type as well as

E138K mutant (low nanomolar EC₅₀values) with low resistance

FC values of 1.9−2.2, which are much lower than that of 1b

(FC = 10). The modeling study with representative compounds

6a and 14d provided a rationale supporting the role of an

optimized R¹group promoting greater eﬃcacy against the

E138K mutant, as well as demonstrating possible modiﬁcation

space between subdomains p66 and p51.

The structural design of these compounds was aimed at

overcoming the E138K drug-resistant variants, and the current

results should help in the development of new NNRTI candidates

that can overcome the major drug resistance conferred to the

new-generation HIV-1 NNRTI drug 1b, as well as other RT-

mutated drug resistances. It is likely that additional structural

optimizations may be needed to obtain compounds eﬀective

against other less prevalent 1b-resistant variants. Nevertheless,

the approach and the results of this study are expected to

provide a useful blueprint for further structural optimization

against the drug-resistant mutants.

1

91% yield; yellow solid; mp 232−234 °C; H NMR δ 2.15 (6H, s,

2 × CH₃), 4.89 (2H, s, CH₂O), 5.85 (1H, d, J = 16.4 Hz, CH),

6.27 (1H, s, ArH-6), 7.05 (2H, d, J = 8.8 Hz, ArH-2′,6′), 7.21 (2H, s,

ArH-3″,5″), 7.28 (1H, d, J = 16.4 Hz, CH), 7.47 (2H, d, J = 8.8 Hz,

ArH-3′,5′), 8.41 (1H, s, ArH-3), 9.70 (1H, s, NH); MS m/z (%) 439.3

(M − 1, 100).

4-Chloromethyl-N¹-(4′-cyanophenyl)-5-(4″-cyanovinyl-2″,6″-di-

methylphenoxy)-2-nitroaniline (4c). 2,4,6-Trichloro[1,3,5]triazine

(TCT, 627 mg, 3.41 mmol), soluble in DMF (4 mL), was stirred

at room temperature for a few minutes, and a white solid appeared.

The mixture was stirred for 1−2 h and added to a solution of 4b (1 g,

2.27 mmol) in CH₂Cl₂(15 mL) while the mixture was being stirred

at room temperature for 4 h by TLC and was monitored until the

reaction reached completion. The organic phase was washed with

water, a saturated NaHCO₃solution, water, and brine, successively,

and dried over anhydrous Na₂SO₄overnight. After removal of solvent

in vacuo, the crude product was puriﬁed by ﬂash column chromato-

graphy (elution, CH₂Cl₂) to aﬀord 800 mg of 4c: 77% yield; yellow

solid; mp 237−239 °C; ¹H NMR δ 2.17 (6H, s, 2 × CH₃), 4.77 (2H, s,

CH₂Cl), 5.85 (2H, d, J = 16.4 Hz, CH), 6.26 (1H, s, ArH-6), 7.06

(2H, d, J = 8.4 Hz, ArH-2′,6′), 7.22 (2H, s, ArH-3″,5″), 7.33 (1H, d,

J = 16.4 Hz, CH), 7.49 (2H, d, J = 8.4 Hz, ArH-3′,5′), 8.41 (1H, s,

ArH-3), 9.77 (1H, s, NH); MS m/z (%) 481.3 (M + Na, 100).

General Amidation Procedure for the Preparation of 5a−c.

A solution of carboxylic acid compound 4a in SOCl₂(10 mL) was

heated to reﬂux for 4 h and then poured into petroleum ether while

the mixture was being stirred. The precipitated solid was collected by

ﬁlter. The crude aryl chloride was redissolved in THF and an amine

reagent (excess) added at 0 °C while the mixture was being stirred

over 30 min. The solvent was removed under reduced pressure, and

the residue was puriﬁed by ﬂash column chromatography (gradient

elution, CH₂Cl₂/petroleum ether, 70 to 100%) to obtain the amide

product.

EXPERIMENTAL SECTION

■

Chemistry. Melting points were measured with an RY-1 melting

apparatus without correction. Proton nuclear magnetic resonance

(¹H NMR) spectra were measured on a JNM-ECA-400 (400 MHz)

spectrometer using tetramethylsilane (TMS) as the internal standard.

The solvent used was CDCl₃unless otherwise indicated. Mass spectra

were recorded on an ABI PerkinElmer Sciex API-150 mass spectro-

meter with electrospray ionization, and the relative intensity of each

ion peak is presented as percent. The purities of target compounds

were ≥95%, as measured by HPLC analyses, which were performed on

an Agilent 1200 HPLC system with a UV detector and an Agilent

Eclipse XDB-C18 column (150 mm × 4.6 mm, 5 μm), eluting with a

mixture of solvents A and B under two conditions (acetonitrile/water

or MeOH/water). The microwave reactions were performed on a

microwave reactor from Biotage, Inc. Thin-layer chromatography (TLC)

and preparative TLC plates used silica gel GF254 (200−300 mesh)

purchased from Qingdao Haiyang Chemical Co. Medium-pressure

column chromatography was performed using a CombiFlash

companion puriﬁcation system. All chemical reagents, NADPH, and

solvents were obtained from Beijing Chemical Works or Sigma-

Aldrich, Inc. Pooled human liver microsomes (lot no. 20150323H)

(protein content of 20 mg/mL, 0.5 mL) were purchased from iPhase

Biosciences Ltd. (Beijing, China). HLM assay data analysis was

conducted on an LC−MS system with an ABI PerkinElmer Sciex

API-3000 mass spectrometer with electrospray ionization.

4-Carboxy-N¹-(4′-cyanophenyl)-5-(4″-cyanovinyl-2″,6″-dimethyl-

phenoxy)-2-nitroaniline (4a). To a solution of N¹-(4′-cyanophenyl)-

5-(4″-cyanovinyl-2″,6″-dimethylphenoxy)-4-methoxycarbonyl-2-nitro-

aniline (2m) (234 mg, 0.50 mmol) in a THF/MeOH solvent (6:1,

v/v) was added a few drops of aq NaOH (1 N), and the mixture was

stirred at room temperature for 10 h. The mixture was poured into

ice−water, and the pH was adjusted to 2 with aq HCl (2 N). The

precipitated solid was collected, washed with water, and dried to aﬀord

1

204 mg of 4a: 90% yield; yellow solid; mp 260−262 °C; H NMR

(DMSO-d₆) δ 2.06 (6H, s, 2 × CH₃), 5.95 (1H, s, ArH-6), 6.44 (1H,

d, J = 16.4 Hz, CH), 7.22 (2H, d, J = 8.8 Hz, ArH-2′,6′), 7.47 (2H,

s, ArH-3″,5″), 7.58 (1H, d, J = 16.4 Hz, CH), 7.65 (2H, d, J =

8.8 Hz, ArH-3′,5′), 8.72 (1H, s, ArH-3), 9.79 (1H, s, NH), 13.21 (1H, s,

COOH); MS m/z (%) 453.3 (M − 1, 45), 409.0 (M − 44, 100).

N¹-(4′-Cyanophenyl)-5-(4″-cyanovinyl-2″,6″-dimethylphenoxy)-

4-hydroxymethyl-2-nitroaniline (4b). To a solution of 3 (234 mg,

0.50 mmol) in THF (5 mL) was added LiBH₄(36 mg, 0.16 mmol) at

0 °C while the mixture was being stirred over 30 min. After MeOH

(13 drops) was added dropwise, the mixture was stirred at room