Alkyl hydroxybenzoic acid derivatives that inhibit HIV-1 protease dimerization

Article abstract of DOI:10.2174/092986712803251557

The therapeutic potential of gallic acid and its derivatives as anti-cancer, antimicrobial and antiviral agents is well known. We have examined the mechanism by which natural gallic acid and newly synthesized gallic acid alkyl esters and related protocatechuic acid alkyl esters inhibit HIV-1 protease to compare the influence of the aromatic ring substitutions on inhibition. We used Zhang-Poorman's kinetic analysis and fluorescent probe binding to demonstrate that several gallic and protecatechuic acid alkyl esters inhibited HIV-1 protease by preventing the dimerization of this obligate homodimeric aspartic protease rather than targeting the active site. The tri-hydroxy substituted benzoic moiety in gallates was more favorable than the di-substituted one in protocatechuates. In both series, the type of inhibition, its mechanism and the inhibitory efficiency dramatically depended on the length of the alkyl chain: no inhibition with alkyl chains less than 8 carbon atoms long. Molecular dynamics simulations corroborated the kinetic data and propose that gallic esters are intercalated between the two N- and C-monomer ends. They complete the β-sheet and disrupt the dimeric enzyme. The best gallic ester (14 carbon atoms, K_id of 320 nM) also inhibited the multi-mutated protease MDR-HM. These results will aid the rational design of future generations of non-peptide inhibitors of HIV-1 protease dimerization that inhibit multi-mutated proteases. Finally, our work suggests the wide use of gallic and protocatechuic alkyl esters to dissociate intermolecular β-sheets involved in protein-protein interactions.

Full text of DOI:10.2174/092986712803251557

Send Orders of Reprints at reprints@benthamscience.org
4534
Current Medicinal Chemistry, 2012, 19, 4534-4540
Alkyl Hydroxybenzoic Acid Derivatives that Inhibit HIV-1 Protease Dimerization
O.A. Flausino Jr.a,b, L. Dufaua, L.O. Regasinib, M.S. Petrôniob, D.H.S. Silvab, T. Rosec, V.S. Bolzanib,* and
M. Reboud-Ravauxa,*
aEnzymologie Moléculaire et Fonctionnelle, UR4, UPMC-Sorbonne Universités, 7 Quai St Bernard, 75252 Paris Cedex 05, France;
bChemistry Institute, Sao Paulo State University, CEP 14800-060, SP, Brazil; cInstitut Pasteur, Unité d'Immunogénétique Cellulaire,
25 rue du Dr Roux, 75724 Paris Cedex 15, France
Abstract: The therapeutic potential of gallic acid and its derivatives as anti-cancer, antimicrobial and antiviral agents is well known. We
have examined the mechanism by which natural gallic acid and newly synthesized gallic acid alkyl esters and related protocatechuic acid
alkyl esters inhibit HIV-1 protease to compare the influence of the aromatic ring substitutions on inhibition. We used Zhang-Poorman’s
kinetic analysis and fluorescent probe binding to demonstrate that several gallic and protecatechuic acid alkyl esters inhibited HIV-1 pro-
tease by preventing the dimerization of this obligate homodimeric aspartic protease rather than targeting the active site. The tri-hydroxy
substituted benzoic moiety in gallates was more favorable than the di-substituted one in protocatechuates. In both series, the type of inhi-
bition, its mechanism and the inhibitory efficiency dramatically depended on the length of the alkyl chain: no inhibition with alkyl chains
less than 8 carbon atoms long. Molecular dynamics simulations corroborated the kinetic data and propose that gallic esters are interca-
lated between the two N- and C-monomer ends. They complete the ꢀ-sheet and disrupt the dimeric enzyme. The best gallic ester (14 car-
bon atoms, Kid of 320 nM) also inhibited the multi-mutated protease MDR-HM. These results will aid the rational design of future gen-
erations of non-peptide inhibitors of HIV-1 protease dimerization that inhibit multi-mutated proteases. Finally, our work suggests the
wide use of gallic and protocatechuic alkyl esters to dissociate intermolecular ꢀ-sheets involved in protein-protein interactions.
Keywords: dimerization inhibitors; gallic acid alkyl esters; HIV-1 protease inhibition; intermolecular ꢀ-sheet inhibitors; protocatechuic acid
alkyl esters.
Recent reports of the antiviral activities of gallic acid deriva-
tives [18, 19] led us to look for new inhibitors of HIV-1 protease
dimerization among these molecules. Gallic acid and its derivatives
are among the most abundant phenolic antioxidants in wines and
green tea [20] and their antioxidant, antibacterial, anti-
inflammatory, antimutagenic, and chemopreventive properties have
been well documented [18, 21-25]. Lauryl gallate (dodecyl gallate)
inhibits the enzyme xanthine oxidase and the peroxidation of mito-
chondrial lipids induced by FeIII-NADPH with an IC50 of 10 ꢀg/mL
[26]. The anti-HIV-1 activity in vitro of 50 ꢀM gallic acid and alkyl
gallates was tested by exposing peripheral blood mononuclear cells
to a suspension of HIV-1 [19]. The inhibition of viral replication
depended on the length of the gallate alkyl chain. Similarly, the
inhibition of HIV-1 integrase by gallic acid and its alkyl derivatives
was also influenced by the alkyl chain length [27].
INTRODUCTION
The mature human immunodeficiency virus type 1 protease
(HIV-PR) has proven to be a successful target for the development
antiviral agents. Synthetic PR inhibitors are now included in highly
active antiretroviral therapy (HAART). HIV-PR is responsible for
processing of polyproteins Gag and Gag-Pol into the mature viral
enzymes and structural proteins need to build the virus capsid. Sev-
eral protease active site inhibitors of HIV-PR (PIs) have been de-
veloped, but cross-resistance to these PIs seriously limits the long-
term treatment of AIDS patients. It is therefore essential to identify
new classes of antiretroviral drugs that act via another mechanism
of action and have few or no adverse effects.
HIV-PR is only active as a dimer since the oligomerization of
two identical protease monomers is required for catalytic activity
[1]. The dimer interface involves mainly the ꢀ-hairpin flaps, the
aspartic protease triad residues (Asp25-Gly27) and N (residues 1-4)-
terminal and C (residues 96-99)-terminal ꢀ-strands of the two
monomers [2]. The antiparallel ꢀ-sheet formed by interdigitation of
these N- and C-terminal monomer ends contributes close to three-
fourths of the total Gibbs free energy of dimerization [3]. Moreo-
ver, its amino acid residues are highly conserved in most HIV-1
isolates, in some HIV-2 isolates, and in drug-resistant HIV-1 vari-
ants [4]. Since mutations involved in resistance to one or more
clinical PIs lie within or outside the active site but rarely in the
antiparallel ꢀ-sheet formed by the N- and C-terminal monomer
ends, this region has been the target of efforts to develop new pro-
tease inhibitors, named as HIV-PR dimerization inhibitors (Fig. 1).
Several series of dimerization inhibitors targeting this ꢀ–sheet re-
gion have been developed by us or others, including C- and N-
terminal mimetic peptides [5-7], lipopeptides [8-10], bicyclic gua-
nidinium derivatives [11], and cross-linked peptides with flexible
spacers [12, 13] and semi-rigid ones [14-17].
A recent report on the ability of alkyl tripeptides to inhibit the
dimerization of HIV-PR showed a similar dependence on the length
of alkyl chain [10]. The alkyl chain length influenced both their
mechanism of action (active-site or dimerization or mixed inhibi-
tion) and their inhibitory potency.
We have now examined the mechanism by which two series of
hydroxybenzoic acid derivatives inhibit HIV-PR (Fig. 2): gallic
acid (1a) and 15 alkyl esters (1b-p), and protocatechuic acid (2a)
and 14 alkyl esters (2b-n). The protocatechuic series was chosen to
compare the influence of aromatic ring substitution on the inhibi-
tion. The length of the linear alkyl chain varied from 1 to 14 (1b-
1n) and 1 to 18 (2b-2o) carbon atoms. Two branched alkyl chains
were introduced into compounds 1o and 1p. We used kinetic analy-
sis, Zhang-Poorman’s method, and a complementary assay, fluores-
cent probe binding using 8-anilino-1-naphtalene-sulfonic acid
(ANS), to investigate the influence of the hydrophobic alkyl chain
on the mechanism by which these new hydroxybenzoic acid deriva-
tives inhibit HIV-PR. We demonstrated for the first time that sev-
eral gallic and protocatechuic alkyl esters acted as dimerization
inhibitors of HIV-PR. They also inhibited by the same mechanism
the multi-mutated protease MDR-HM whom mutations are identical
to that found in some resistant HIV strains. Finally, we checked out
our experimental results with molecular dynamics simulations of
the alkyl hydroxybenzoic acid derivatives complexed with wild-
type and mutated HIV-PR monomers.
*Address correspondence to these authors: M.R.R. at the Enzymologie
Moléculaire et Fonctionnelle, UR4, UPMC-Sorbonne Universités, 7 Quai St
Bernard 75252 Paris Cedex 05, France; Tel: +33(0)144275078; Fax:
+33(0)144275140; E-mail: michele.reboud@upmc.fr; V.S.B. V.S.B. at the
Institute of Chemistry, Rua Prof. Francisco Degni, 55, Sao Paulo State Uni-
versity, CEP 14800-900, SP Brazil; Tel: +55-16-331-6660; Fax: +55-16-
3322-2308; E-mail: bolzaniv@iq.unesp.br
1875-533X/12 $58.00+.00
© 2012 Bentham Science Publishers

New Alkyl Esters Inhibitors of HIV Protease
Current Medicinal Chemistry, 2012 Vol. 19, No. 26
4535
ꢀ
ꢂꢃꢄꢅꢆ ꢀꢇꢅꢄ ꢀ
ꢁꢀ
ꢌꢃꢄꢅꢆ ꢇꢍꢅꢈ ꢊꢇ
ꢀꢁꢂꢃꢄꢅꢆ ꢇꢈꢉꢁꢉꢈ ꢊꢋꢇ
ꢂꢈꢉꢊꢈꢋꢃꢌꢍꢅꢎꢀ
ꢏꢊꢐꢑꢒꢓ ꢎꢀ
ꢔꢅꢎꢐꢅꢎꢓꢀꢕꢒꢅ ꢄꢊꢀ
ꢀꢁꢂꢃꢄꢅꢆ ꢇꢈꢉꢁꢉꢈ ꢊꢎꢅꢁꢏꢅꢐꢅꢄꢉꢊꢇ
ꢇꢃꢉꢈꢑꢒ ꢓꢇ
ꢀ
Fig. (1). Mechanism by which an alkyl dimerization inhibitor may act as a homodimer disruptor to inhibit HIV-1 PR.
O
O
HO
OR
HO
HO
OR
HO
OH
2
1
R
Series 1
1a
R
Series 2
2a
H
H
n
0
1
2
3
4
5
6
7
8
9
n
0
1b
1c
1d
1e
1f
2b
2c
2d
2e
2f
1
2
3
4
1g
1h
1i
5
2g
2h
2i
-(CH2)n-CH3
R -(CH2)n-CH3
6
7
1j
8
2j
1k
1l
9
2k
2l
10
11
13
15
17
11
13
1m
1n
2m
2n
2o
n
0
1
-(CH2)n-CH-(CH3)2
1o
1p
Fig. (2). Chemical structures of semi-synthetic derivatives 1 (gallates) and 2 (protocatechuates).

4536 Current Medicinal Chemistry, 2012 Vol. 19, No. 26
Flausino Jr. et al.
MATERIALS AND METHODS
1. Gallic Acid Source
4. Enzyme and Inhibitor Assays
The fluorogenic substrate DABCYL-ꢀ-abu-SQNYPIVQ-
EDANS (DABCYL, 4-(4’-dimethylaminophenylazo)benzoyl; ꢀ-
Gallic acid was extracted from Alchornea glandulosa. The
leaves were collected in the Biological Reserve and Experimental
Station at Mogi Guaçu, SP, Brazil, in March 2005. A voucher
specimen (SP319257) has been deposited in the herbarium of the
Botanic Institute (São Paulo, SP, Brazil). The shade-dried plant
material (1.5 kg) was ground and defatted with n-hexane (3.5 L x 3,
at room temperature) and exhaustively extracted by maceration
with MeOH (4.2 L x 3). The crude extract was concentrated under
pressure to yield 3.8 g of a syrup residue. The concentrate was then
diluted with MeOH/H2O (4:1) and successively partitioned with
EtOAc and n-BuOH. After solvent removal under reduced pressure,
the partition phases yielded 2.5 and 0.8 g, respectively. The EtOAc
residue (2.0 g) was chromatographed by gel permeation over Se-
phadex LH-20, eluted with methanol, to afford 11 fractions (A1-
A11). Fraction A2 (730 mg) was further purified by reverse-phase
(RP)-HPLC (7.5:92:0.5 MeOH/H2O/HOAc, UV detection at 265
nm, and flow rate at 15 mM/min) to yield gallic acid (355 mg). The
identification was based on analysis of its 1H and 13C NMR data, as
well as by comparison to the chemical product obtained from
Sigma-Aldrich Chemical Co. (St. Louis, MO).
abu,
ꢀ-aminobutyric
acid;
EDANS,
5[(2-aminoethyl)
amino]naphthalene-1-sulfonic acid) was purchased from Bachem
(Voisins-le-Bretonneux, France). 1-anilino-8-naphthalene sulfonate
(ANS) and acetylpepstatin were from Sigma-Aldrich, saquinavir
was from the NIH (USA). Fluorescence intensities were measured
in a FluoStar Galaxy spectrophotometer (BMG LabTechnologies,
Offenburg, Germany) in black opaque microplates (Costar, Corning
Incorporated, Corning, NY, USA).
The proteolytic activities of mutated proteases were determined
fluorometrically (ꢁex = 340 nm; ꢁem = 490 nm) in 100 mM sodium
acetate, 1 mM EDTA, and 1 M NaCl at pH 4.7 and 30°C. The sub-
strate (5.2 ꢀM) and the compounds were first dissolved in DMSO.
The final DMSO concentration was kept at 3% (v/v). The mecha-
nism of inhibition and the corresponding kinetic constants Kic
(competitive inhibition) and/or Kid (dimerization inhibition) were
determined using [7]. Plots were fitted using KaleidaGraph 4.0
(Synergy Software). Six concentrations of PR and mutated HDR-
HM protease ([E]0 = 5.33 - 18.7 nM for PR, and 10-20 nM for
MDR-HM protease) were used for Zhang-Poorman kinetic analysis.
The inhibitor concentrations were 1.13 and 0.85 ꢀM for 1n, 1.42
and 1.13 ꢀM for 1m, 2.55 and 1.98 ꢀM for 1l, 5.67 and 3.54 ꢀM
for 1k, 11.33 and 8.5 ꢀM for 1j, 85 and 20 ꢀM for 2m, 5.67 and 1.7
ꢀM for 2l, 28.3 and 14.2 ꢀM for 2k, 17 and 8.5 ꢀM for 2j, 85 and
28 ꢀM for 2i.
2. Synthesis
2.1. Synthesis of Gallates and Protocatechuates
A 3.0 mL solution of N,N’-dicyclohexylcarbodiimide (DCC,
1.0 mmol) in dried p-dioxane was added to a cooled (5 C) solution
o
5. Fluorescent Binding Probe
of 0.2 mmol of protocatechuic acid or gallic acid and 20 mmol of n-
alkyl alcohol in 6.0 mL of dried p-dioxane. The solution was stirred
for 48 h, and the solvent was subsequently removed under reduced
pressure. The residue was partitioned with EtOAc (3 times) and
filtered. The filtrate was washed successively with saturated aque-
ous citric acid solution (3 times), saturated aqueous NaHCO3 (3
times), dried over anhydrous MgSO4, and evaporated under reduced
pressure. The crude products were purified over a silica gel column
eluted with CHCl3/MeOH (98:2) to afford alkyl esters. Structures of
the semi-synthetic esters were established by 1H and 13C NMR
spectral analysis.
The intensity of the 8-anilino-1-naphtalene-sulfonic acid (ANS)
fluorescence (ꢁexc= 340 nm; ꢁem= 490 nm) was measured (Perkin
Elmer LS50B spectrofluorometer) using six ANS concentrations
varying from 0 to 60 ꢀM (final volume 200 ꢀL) in the presence of
PR alone (0.27 nM) (blank), and PR plus inhibitors (1.0 ꢀM) as
described in [10]. The final DMSO concentration was always 0.2 %
(v/v). Each fluorescence measurement was made five times and
averaged. PR, acetylpepstatin, saquinavir and compounds 1n and
1m had negligible intrinsic fluorescence under our experimental
conditions.
2.2. Synthesis of Acylated Analogues
6. Molecular Modeling
For synthesis of acylated analogues, protocatechuic acid or gal-
lic acid (20 mmol) was dissolved in dried pyridine (5.0 mL) and
acetic anhydride (5.0 mL) under hydrogen atmosphere. The mixture
was stirred for 48 h at room temperature, dried under reduced pres-
sure, and purified by column chromatography with CHCl3/MeOH
(85:15). Structures of the acylated analogues were established by
1H and 13C NMR spectral analysis.
The PR/inhibitor complexes were modeled in five steps, as de-
tailed previously for PR/lipopeptides [10]. 1) We first built an
PR/compound 2m model from crystallographic data for the dimer
(1AXA, resolution 2.0 Å; [29]), replacing the monomer B C-
terminus with 2m, then with compounds 1 (n = 9-13) or 2 (n = 9-
16). 2) We then evaluated the flexibility of monomer A loops from
restrained molecular dynamics (Discover, Accelrys, San Diego,
CA) and selected normal modes [30] that might enhance compound
binding and set up starting models for PR. 3). The docking of com-
pounds was processed on the same starting models using the super-
imposition of the alkyl chain. 4) We ranked the models of the com-
plex according to stereochemical and energy criteria and optimized
the best model PR monomer/compound 1 and PR monomer
/compound 2 for each value of n from 9 to 16. 5) We compared the
binding energies of the two series of inhibitors to PR and multi-
mutated MDR-HM.
3. Mutated Proteases
We produced a pseudo-wild type protease (called here PR) and
the multi-mutated protease MDR-HM (L10I/M46I/ I54V/V82A/
I84V/L90M) as described in [10]. The monomer of PR had 5 muta-
tions: Q7K, L33I, L63I to minimize protease autolysis, and C67A
and C95A to prevent cysteine-thiol oxidation. The enzymatic activ-
ity of the pseudo-wild type protease containing these protective
mutations was identical to that of wild-type protease. The plasmid
pET9PRWT bearing the wild-type sequence of the protease gene
from HIV (BRU) DNA clone (subclone of ꢁJ19) was used as tem-
plate for PCR amplification cycles. The resulting mutated plasmids
were sequenced (Genome Express, Meylan, France) on both coding
and noncoding strands. The PR and MDR-HM protease [28] (Prof.
E. Freire, the Johns Hopkins University, Baltimore, MD) were ex-
pressed in Escherichia coli BL21 (DE3) rosetta pLysS strain (No-
vagen, Darmstadt, Germany), purified, and refolded as described
before [10].
RESULTS AND DISCUSSION
Initial screening assays showed that gallic acid (1a) and proto-
catechuic acid (2a) and their derivatives with linear chains shorter
than 9 carbons did not inhibit PR, even at the highest inhibitor con-
centration tested (50 ꢀM). We determined the mechanism of inhibi-
tion for the active compounds (1j-1n and 2j-2n) using Zhang-

New Alkyl Esters Inhibitors of HIV Protease
Current Medicinal Chemistry, 2012 Vol. 19, No. 26
4537
ꢀꢁ
ꢀꢁ
10
8
5
4
6
3
ꢀꢁ
ꢀꢁ
4
2
1
0
2
0
0
0,5
1
1,5
2
2,5
3
0
1
2
3
4
5
6
ꢂꢁ
ꢂꢁ
ꢁꢁ
ꢂꢁ
5
4
3
2
1
0
6
5
4
3
2
1
0
ꢀꢁ
ꢀꢁ
0
1
2
3
4
5
0
1
2
3
4
5
6
ꢂꢁ
ꢂꢁ
Fig. (3). Zhang-Poorman graphics for the inhibition of PR (A, C and D) and MDR-HM protease (B) by alkyl hydroxybenzoic acid derivatives at pH 4.7 and 30
°C. A. PR inhibition by 1n: 1.13 ꢀM (ꢀ), 0.85 ꢀM (ꢁ). B. MDR-HM protease inhibition by 1n: 5.55 ꢀM (ꢀ), 1.98 ꢀM (ꢁ). C. PR inhibition by 2m: 5.6 ꢀM
(ꢀ), 1.7 ꢀM (ꢁ). D) PR inhibition by 1l: 2.55 ꢀM 4 (ꢀ), 1.98 ꢀM (ꢀ). Enzyme activity was also always measured without inhibitor (ꢀ).
1
1
ꢁ
1 ꢄ
ꢁ
1 ꢄ
[E]0
vi
x:
; y:
.
vi ꢀ105 M2 .s2
ꢀ104 M2 .s2
ꢃ
ꢆ
ꢃ
ꢆ
ꢂ
ꢅ
ꢂ
ꢅ
Poorman kinetic analysis [7]. Plots of [E]0/ꢁvi vs ꢁvi were con-
structed to discriminate between inhibition of dimerization alone
(parallel lines), competitive inhibition (altered slopes and unaltered
y-axis intercepts), and mixed inhibition (altered slopes and altered
y-axis intercepts) (Fig. 3). The data for alkyl hydroxybenzoic de-
rivatives showed that inhibitors with longer alkyl chains of 12 to 14
C (series 1) and 12 to 16 C (series 2) acted as pure dimerization
inhibitors of PR. Gallic acid derivatives inhibited PR with inhibi-
tion constants (Kid) of 320 nM for 1n (14 C) and 650 nM for 1m
(12 C) (Fig. 3). The same mechanism also characterized the effects
of 2n (16 C), 2m (14 C) and 2l (12 C) with Kids of 9.9, 0.5, and 2.1
ꢀM, respectively (Fig. 3). Compounds with shorter alkyl chains (9-
11 carbons for series 1, 9-10 for series 2) acted as mixed inhibitors
with an antidimer component characterized by a Kid value and a
competitive component (binding within the active site) character-
ized by a Kic value (Table 1). The values of Kid and Kic were almost
identical in some cases, (1l, 1k, 2j) while the dimerization process
was preponderant in others (1j, 2k). Branched alkyl chains were
unfavorable (no inhibition at 50 ꢀM for 1o and 1p).
We used direct structural investigation to examine further how
alkyl hydroxybenzoic acid derivatives inhibit HIV proteases. The
fluorescence emission for the binding of the fluorescent probe ANS
was measured in the presence of two active site inhibitors (ace-
tylpepstatin and saquinavir) and compounds 1m and 1n both identi-
fied as dimerization inhibitors. ANS is reported to preferentially
attach to the hydrophobic zones of protein surfaces and binding
leads to an increase in fluorescence. Adding ANS to a mixture of
PR and compound 1n increased the emission fluorescence (factor ꢀ
2-times more than for PR plus ANS ([ANS] = 30 ꢀM) (Fig. 4). The
same ANS concentration produced a smaller fluorescence increase
with compound 1m (ꢀ 1.6-fold). In contrast, the active site inhibi-
tors saquinavir and acetylpepstatin decreased the fluorescence
emission. These results suggest that compounds 1m and 1n were
bound to unmasked hydrophobic surfaces of PR, in agreement with
an inhibition of PR dimerization as this process is expected to ex-
pose hydrophobic interface areas due to dimer destabilization or
dissociation [10, 15].
Gallic acid derivatives 1 had greater antidimer potential towards
PR than did protecatechuic acid derivatives 2 with an identical alkyl
chain. It was about 1.6-fold for C14 alkyl chains (1n and 2m), 3.2-
fold for C12 (1m and 2l), 1.7-fold for C10 (1k and 2k) and 11-fold
for C9 (1j and 2k). The same was true for mixed inhibitors (1j and
2j, and 1k and 2k). The difference could be because gallic acid
derivatives have three aromatic hydroxyl groups that can form hy-
drogen bonds, while protecatechuic acid derivatives have only two
as evidenced below in modeled complexes.
Bannwarth and collaborators reported that the length of the al-
kyl chains greatly influenced the antidimer activity of alkyl tripep-
tides [10]. They showed that the lipotripeptides CH3(CH2)nCO-YEL
and CH3(CH2)nCO-LEY with alkyl chain length of 12-16 carbons
inhibited the dimerization of HIV-1 PR, whereas compounds with
8-10-carbon alkyl chains were mixed inhibitors. Lipopeptides with
alkyl chains shorter than 6 carbons were pure competitive inhibi-
tors. Our present results indicate that hydroxybenzoic acid deriva-
tives with alkyl chains shorter than 7 carbons do not inhibit HIV-1
PR. The major difference between alkyl tripeptides and alkyl hy-
droxybenzoic acid derivatives resides is the nature of the moiety

4538 Current Medicinal Chemistry, 2012 Vol. 19, No. 26
Flausino Jr. et al.
a
Table 1.
Inhibition of PR and Mutated HIV-1 Protease MDR-HM by Compounds 1 and 2 (30 °C and pH 4.7). Standard Errors of Initial Rates
are Less than 5%. NI: No Inhibition at 50 ꢀM.
Compounds
Gallic acid and derivatives
HIV-1 protease
Kida (ꢀM)
Kica (ꢀM)
R = H
1a
NI
n
0-7
8
1b-1i
1j
NI
2.5
7.9
3.9
0.65
0.32
NI
31.5
12.3
5.5
R = -(CH2)n-CH3
1k
1l
9
PR
10
11
13
0
1m
1n
1o
R = -(CH2)n-CH-(CH3)2
1p
1
NI
Protocatechuic acid and derivatives
R = H
2a
NI
n
1-7
8
2b-2i
2j
NI
27.8
13.4
2.1
46.8
73.9
PR
2k
2l
9
R = -(CH2)n-CH3
11
13
15
17
13
2m
2n
2o
0.5
9.9
NI
MDR-HM
1n
0.69
bearing the alkyl chain: it is a tripeptide in one case and a substi-
tuted benzoic acid in the other. This could lead to different interac-
tions with the monomer N- and C-ends as suggested in
PR/monomer inhibitor complexes.
Models of PR monomer/inhibitor complexes corroborated the
kinetics, with C14 alkyl chain providing an optimal occupancy of
the hydrophobic enzyme cleft. In the model of the PR monomer/2m
complex, the wide cleft in monomer A where the helix RB87-GB93 of
monomer B usually sits is narrowed by the closure of loops TA4-
PA9, LA24-DA30, IA66-KA70 and the ꢁ-hairpin flap MA46-KA55 over the
alkyl chain. The cleft grasps the ligand tightly, which buries several
hydrophobic side chains from the cleft floor (LA24, IA84), the adja-
cent loops (IA47, IA50, IA54, VA82) and covering the catalytic residue
DA25. The flap opening measured by the distance CꢀDA25-CꢀIA50 is
much more closed over the cleft filled by gallate 1n (10 Å) in this
“super-closed” model, than in the closed crystallographic dimer (13
Å) as shown previously [10]. This flap also buries hydrophobic
residues VA32, VA56, VA75 and LA76. The side chains of WA6 and FA99
are also closer to each other, partially burying the alkyl chains (C1-
C4) of 2m and 1n. The other side of the alkyl moiety (C1-C6) is
protected from solvent by LA97 and by aromatic rings themselves.
The protocatechuate and gallate aromatic rings sit in between the
two N-term and C-term ꢁ- strands. For protocatechuate derivatives:
the ester carbonyl forms an H-bond with the N98 amide, the H from
meta hydroxyl group makes an H-bond with the T96 carbonyl, the
para hydroxyl group O forms an H-bond with the T96 hydroxyl H
and amide. For gallate derivatives: the ester carbonyl forms an H-
bond with the N98 amide, O from a meta hydroxyl group makes an
H-bond with the T96 and the I3 H amide. The H from the meta
hydroxyl group is engaged in H-bond with P1 carbonyl. Gallate
derivatives form H bonds with both ends of the skeleton, while
protocatechuate derivatives form H bonds with the C-terminus only.
The tight fit of the ligand in the cleft excludes most water mole-
cules from around the hydrophobic residues of the PR active site,
thus optimizing the interface area. The rims of the cleft close up the
ligands. The interface surface areas were computed from differ-
ences between the water accessible surface area of free molecules
300
250
200
150
100
50
0
0
10
20
30
40
50
60
70
ꢃꢀꢀꢁꢂꢁꢂꢃꢁ
Fig. (4). ANS emission fluorescence (470 nm) measured at pH 4.7 and
25°C. The excitation wavelength was 370 nm. PR (275 nM) was preincu-
bated without inhibitor (ꢀ) or with acetylpepstatin (ꢂ), saquinavir (ꢃ) or
compounds 1n (ꢁ), 1m (ꢀ) prior to adding ANS. The inhibitor concentra-
tions was 1 ꢀM for all compounds. The final DMSO concentration was
always 0.2 % (v/v). Fluorescence intensity was corrected for the fluores-
cence of the same concentrations of ANS and inhibitor in the absence of
enzyme.

New Alkyl Esters Inhibitors of HIV Protease
Current Medicinal Chemistry, 2012 Vol. 19, No. 26
4539
A
chain A
I50
chain B
D25
antiparallel ꢀ-sheet
B
C
I50
I54V
M46I
I50
V82A
N
I84V
L10I
P1
Q2
T96
C14
D25
L97
I3
T4
N98
F99
L90M
N
C
C
Fig. (5). Models of uncomplexed HIV-1 protease and complexes of C14 gallate 1n, C14 catechuate 2m and C14-LEY associated to a protease monomer (wild-
type or MDR-HM). A. Open active site conformation of HIV-1 protease (ITW7, ligand-free dimer); chain A in red, chain B in blue. B. Models of monomer
(red sketch) /1n (green sticks) and monomer (red sketch) /2m (turquoise sticks) complexes. N- and C- terminal monomer ends are in blue (green sticks) (90°
rotation of the monomer from the front view A. The two alkyl chains of compounds 1n and 2m are perfectly superimposed. C. Models of compounds 1n
(green sticks) , 2m (turquoise sticks) and C14-LEY (white sticks) associated to MDR-HM monomer showing a ‘superclosed’ protease conformation. Loops
and flap motions improve the inhibitor binding. The closely superimposed alkyl chains do not interact with mutated side chains (mutations indicated in green).
The tyrosine residue of C14-LEY occupies the” F99 pocket”.
and the complex (probe radius = 1.4 Å). The PR interface with 2m,
ꢀASAPR/2m = 616 Å2, is composed of 14 % polar and 86 % nonpolar
contributions. The PR interface with 1n ꢀASAPR/1n = 719 Å2 is
composed of 29 % polar and 71 % apolar contributions. The lipo-
philic alkyl chain is an important contributor to the interface (453
Å2 for 1n and 380 Å2 for 2m) and drive entropy of the free energy
of inhibitor binding.
690 nM). The ratio Kid(MDR-HM)/Kid(PR) was 2. The model of the
complex formed between the PR monomer and 1n suggests that the
mutations in the monomer do not interfere with the inhibitor bind-
ing (Fig. 5C).
All the above data indicate that the hydroxybenzoic acid deriva-
tives are suitable new scaffolds on which to develop inhibitors of
HIV-1 PR dimerization. They have the advantage to be non-peptide
molecules that should be more stable and bioavailable in vivo than
peptides. A weak toxicity of the gallic acid core and its alkyl de-
rivatives has been reported [25,31]. The small size of the benzoic
acid moiety indicates that they can be rationally optimized to in-
crease the energy of binding to the C- and N-ends of the protease
monomer by targeting the hydrophobic F99 pocket. Their unusual
mechanism of action should enable them to circumvent the resis-
tances due to treatments with active-site inhibitors PIs. They may
also be good candidates for developing multi-target therapeutical
agents in AIDS, since they may react with both the HIV-1 protease
and its integrase [27], another pertinent target for AIDS treatment.
We also built protease-ligand models for alkyl chain lengths (n
= 0-13) for the gallate series, n = 0-17 for the protocatechuate series
and compared them with the previously reported alkyl tripep-
tide/monomer complexes [10]. The longest alkyl chains that could
fit into the “super-closed” models were obtained for 1n and 2m
displaying 14 carbon atoms whereas the best fit was observed with
the palmitoyl moiety (16 carbon atoms) for lipopeptides. This
agrees with the experimental results obtained for alkyl hydroxyben-
zoic acid derivatives, since compound 2n (C16) acted as an antidi-
mer whereas 2o (C18) was not an inhibitor. The catalytic side chain
D25 in the cleft is accessible to alkyl chains of C10 and longer ones
(Fig. 5B).
Finally, we measured the capacity of compound 1n, which had
the greatest inhibitory potential of all the compounds tested, to in-
hibit the multi-mutated protease MDR-HM. The MDR-HM variant
has six mutations, two in the active site (V82A and I84V), two in
the flap region (M46I and I54V) and two in the nearest dimerization
region (L10I and L90M) [28]. Compound 1n impaired the activity
of MDR-HM by inhibiting the dimerization of the protease (Kid of
CONFLICT OF INTEREST
The author(s) confirm that this article content has no conflicts
of interest.
ACKNOWLEDGEMENTS
We thank Prof. Ernesto Freire of the Department of Biology
and Biocalorimetry Center, Johns Hopkins University, Baltimore,

4540 Current Medicinal Chemistry, 2012 Vol. 19, No. 26
Flausino Jr. et al.
Ravaux, M., J Med Chem, 1999, 42, (6), 957-962.
MD, for the generous gift of the plasmid encoding the MDR-HM
mutant. Saquinavir was obtained through the NIH AIDS Research
and Reference Reagent Program, Division of AIDS, NIAID, NIH.
The English text was edited by Dr Owen Parkes. We thank the
French National Agency for Resaerch on AIDS and Viral Hepatitis
for financial support (ANRS n°11359), CAPES (5329-09-4) and
Sao Paulo State Research Funding Agency FAPESP for financial
support (#03/02176-7 and #04/07932-7).
[15]
[16]
Merabet, N.; Dumond, J.; Collinet, B.; Van Baelinghem, L.; Boggetto, N.;
Ongeri, S.; Ressad, F.; Reboud-Ravaux, M.; Sicsic, S., J Med Chem, 2004,
47, (25), 6392-6400.
Bannwarth, L.; Kessler, A.; Pethe, S.; Collinet, B.; Merabet, N.; Boggetto,
N.; Sicsic, S.; Reboud-Ravaux, M.; Ongeri, S., J Med Chem, 2006, 49, (15),
4657-4664.
[17]
[18]
[19]
Vidu, A.; Dufau, L.; Bannwarth, L.; Soulier, J.L.; Sicsic, S.; Piarulli, U.;
Reboud-Ravaux, M.; Ongeri, S., ChemMedChem, 2010, 5, (11), 1899-1906.
Ximenes, V.F.; Lopes, M.G.; Petronio, M.S.; Regasini, L.O.; Silva, D.H.; da
Fonseca, L.M., J Agric Food Chem, 2010, 58, (9), 5355-5362.
Kratz, J.M.; Andrighetti-Frohner, C.R.; Kolling, D.J.; Leal, P.C.; Cirne-
Santos, C.C.; Yunes, R.A.; Nunes, R.J.; Trybala, E.; Bergstrom, T.; Frugul-
hetti, I.C.; Barardi, C.R.; Simoes, C.M., Mem Inst Oswaldo Cruz, 2008, 103,
(5), 437-442.
Paixao, N.; Pereira, V.; Marques, J.C.; Camara, J.S., J Sep Sci, 2008, 31,
(12), 2189-2198.
Kang, M.S.; Oh, J.S.; Kang, I.C.; Hong, S.J.; Choi, C.H., J Microbiol, 2008,
46, (6), 744-750.
Kim, S.H.; Jun, C.D.; Suk, K.; Choi, B.J.; Lim, H.; Park, S.; Lee, S.H.; Shin,
H.Y.; Kim, D.K.; Shin, T.Y., Toxicol Sci, 2006, 91, (1), 123-131.
Birosova, L.; Mikulasova, M.; Vaverkova, S., Biomed Pap Med Fac Univ
Palacky Olomouc Czech Repub, 2005, 149, (2), 489-491.
Giftson, J.S.; Jayanthi, S.; Nalini, N., Invest New Drugs, 2010, 28, (3), 251-
259.
Morais, M.C.; Luqman, S.; Kondratyuk, T.P.; Petronio, M.S.; Regasini, L.O.;
Silva, D.H.; Bolzani, V.S.; Soares, C.P.; Pezzuto, J.M., Nat Prod Res, 2010,
24, (18), 1758-1765.
Kubo, I.; Masuoka, N.; Xiao, P.; Haraguchi, H., J Agric Food Chem, 2002,
50, (12), 3533-3539.
REFERENCES
[1]
Wlodawer, A.; Miller, M.; Jaskolski, M.; Sathyanarayana, B.K.; Baldwin, E.;
Weber, I.T.; Selk, L.M.; Clawson, L.; Schneider, J.; Kent, S.B., Science,
1989, 245, (4918), 616-621.
[20]
[21]
[22]
[23]
[24]
[25]
[2]
Louis, J.M.; Weber, I.T.; Tozser, J.; Clore, G.M.; Gronenborn, A.M., Adv
Pharmacol, 2000, 49, 111-146.
[3]
[4]
[5]
Todd, M.J.; Semo, N.; Freire, E., J Mol Biol, 1998, 283, (2), 475-488.
Hamacher, K., Gene, 2008, 422, (1-2), 30-36.
Franciskovich, J., Houseman, K., Mueller, R., Chmielewski, J., Bioorganic
& Medicinal Chemistry Letters, 1993, 3, (4), 765-768.
[6]
[7]
[8]
[9]
[10]
Schramm, H.J.; Billich, A.; Jaeger, E.; Rucknagel, K.P.; Arnold, G.;
Schramm, W., Biochem Biophys Res Commun, 1993, 194, (2), 595-600.
Zhang, Z.Y.; Poorman, R.A.; Maggiora, L.L.; Heinrikson, R.L.; Kezdy, F.J.,
J Biol Chem, 1991, 266, (24), 15591-15594.
Schramm, E.; Nowicky, J.W.; Godysh, Y., Drugs under experimental and
clinical research, 1996, 22, (3-5), 247-254.
[26]
[27]
[28]
[29]
[30]
[31]
Dumond, J.; Boggetto, N.; Schramm, H.J.; Schramm, W.; Takahashi, M.;
Reboud-Ravaux, M., Biochem Pharmacol, 2003, 65, (7), 1097-1102.
Bannwarth, L.; Rose, T.; Dufau, L.; Vanderesse, R.; Dumond, J.; Jamart-
Gregoire, B.; Pannecouque, C.; De Clercq, E.; Reboud-Ravaux, M., Bio-
chemistry, 2009, 48, (2), 379-387.
Breccia, P.; Boggetto, N.; Perez-Fernandez, R.; Van Gool, M.; Takahashi,
M.; Rene, L.; Prados, P.; Badet, B.; Reboud-Ravaux, M.; de Mendoza, J.,
Journal of medicinal chemistry, 2003, 46, (24), 5196-5207.
Zutshi, R.; Chmielewski, J., Bioorg Med Chem Lett, 2000, 10, (17), 1901-
1903.
Ahn, M.J.; Kim, C.Y.; Lee, J.S.; Kim, T.G.; Kim, S.H.; Lee, C.K.; Lee, B.B.;
Shin, C.G.; Huh, H.; Kim, J., Planta Med, 2002, 68, (5), 457-459.
Ohtaka, H.; Schon, A.; Freire, E., Biochemistry, 2003, 42, (46), 13659-
13666.
[11]
Spinelli, S.; Liu, Q.Z.; Alzari, P.M.; Hirel, P.H.; Poljak, R.J., Biochimie,
1991, 73, (11), 1391-1396.
Lindahl, E.; Azuara, C.; Koehl, P.; Delarue, M., Nucleic Acids Res, 2006, 34,
(Web Server issue), W52-56.
[12]
[13]
[14]
Niho, N.; Tamura, T.; Toyoda, K., Uneyama, C., Takahashi, N.; Hirose, M.,
Food and Chemical Toxicology, 2001, 39, 1063-1070.
Ulysse, L.G.; Chmielewski, J., Bioorg Med Chem Lett, 1998, 8, (22), 3281-
3286.
Bouras, A.; Boggetto, N.; Benatalah, Z.; de Rosny, E.; Sicsic, S.; Reboud-
Received: July 19, 2012
Revised: August 14, 2012
Accepted: August 26, 2012