A platform for designing HIV integrase inhibitors. Part 2: A two-metal binding model as a potential mechanism of HIV integrase inhibitors

Article abstract of DOI:10.1016/j.bmc.2006.08.043

We propose a two-metal binding model as a potential mechanism of chelating inhibitors against HIV integrase (HIV IN) represented by 2-hydroxy-3-heteroaryl acrylic acids (HHAAs). Potential inhibitors would bind to two metal ions in the active site of HIV I

Full text of DOI:10.1016/j.bmc.2006.08.043

Bioorganic & Medicinal Chemistry 14 (2006) 8420–8429
A platform for designing HIV integrase inhibitors. Part 2: A
two-metal binding model as a potential mechanism of HIV
integrase inhibitors
Takashi Kawasuji,a,* Masahiro Fuji,a Tomokazu Yoshinaga,b Akihiko Sato,b
Tamio Fujiwarab and Ryuichi Kiyamaa
aShionogi Research Laboratories, Shionogi & Company, Ltd, Sagisu, Fukushima-ku, Osaka 553-0002, Japan
bShionogi Research Laboratories, Shionogi & Company, Ltd, Mishima, Settsu-shi, Osaka 556-0022, Japan
Received 21 June 2006; revised 28 August 2006; accepted 29 August 2006
Available online 26 September 2006
Abstract—We propose a two-metal binding model as a potential mechanism of chelating inhibitors against HIV integrase (HIV IN)
represented by 2-hydroxy-3-heteroaryl acrylic acids (HHAAs). Potential inhibitors would bind to two metal ions in the active site of
HIV IN to prevent human DNA from undergoing the integration reaction. Correlation of the results of metal (Mg2+ and Mn2+
)
titration studies with HIV IN inhibition for a series of active and inactive compounds provides support for the model. Results sug-
gest Mg2+ is an essential cofactor for chelating inhibitors.
Ó 2006 Elsevier Ltd. All rights reserved.
1. Introduction
IN. The C-terminal domain shows an SH-3 like folding
and has a non-specific affinity to DNA. A unique active
site in the core domain has become an attractive target
of molecular biology to understand the mechanism of
the integration reaction, and many crystallographic
studies have been reported.
Integration of viral DNA into host DNA is an essential
step in viral replication of HIV. The integration reaction
consists of two distinct stepwise reactions.1 The first step
30-processing is a hydrolyzing reaction to remove a ter-
minal dinucleotide to recess a CA-30 terminal. The
CA-30 sequence is generally conserved on the third and
fourth position from both 30-ends of long terminal re-
peats of the viral DNA. The second step called strand
transfer is a trans-esterification reaction of the recessed
30-ends with the phosphodiester backbone of the host
cell DNA. Both ends of the viral DNA are joined to
the host DNA at the same time, and subsequent repair
of the nick by host cell repair machinery completes the
integration sequence.
HIV IN belongs to a broader class of nucleotide-related
enzymes2 including nucleases, polymerases, and polynu-
cleotidyl transferases. These enzymes commonly have an
acidic amino residue cluster in the active sites, which are
known to be essential for their native functions. The
critical residues are thought to hold metal cofactors,
which directly catalyze hydrolysis and/or transesterifica-
tion reactions of phosphoryl esters of nucleotides
according to a two-metal-ion catalysis mechanism.2
HIV IN is composed of three necessary domains:1 N-ter-
minal (1–50aa), core (51–210aa), and C-terminal (211–
288aa) domains. The N-terminal domain has an HHCC
motif and shows a zinc finger-like folding with a Zn2+
ion. It is thought to affect multimer formation of HIV
In the active site of HIV IN, three acidic residues, D64,
D116, and E152, the ‘catalytic triad,’ are conserved
among all viral strains. HIV IN is thought to catalyze
the two-step integration reaction with two metal ions
held by the catalytic triad as illustrated in Scheme 1.
The two metal cofactors of HIV IN are thought to be
Mg2+ under physiological condition.1
Keywords: HIV integrase inhibitor; Two-metal binding model; Inhibi-
tion mechanism; Docking model; Metal affinity.
*
6 6458 0987; e-mail: takashi.
Mg2+ was frequently found to be coordinated by the
carboxylic groups of D64 and D116 of the catalytic triad
Corresponding author. Fax: +81
kawasuji@shionogi.co.jp
0968-0896/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2006.08.043

T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
8421
Stage B
Stage A
Base
O
O
O
O
O
O
O
O
Virus DNA -5'
O
O
O
3' -Virus DNA
O
D116
D64
E152
P
O
H2O
O-
Base(A)
O
O
Virus DNA -5'
Stage C
O
O
O
O
O
O
HO
Base(A)
D116
D64
E152
O
O
O
O
O
O
D116
D64
E152
Stage X
O
Host DNA -3'
O
Virus DNA -5'
Stage D
Inhibitor
O
Base
HO
O
5' -Host DNA
O
O
O
Virus DNA -5'
Base(A)
O
O
O
O
O
O
O
O P
O
H
O
O
O-
Base
O
D116
D64
E152
Base(A)
Host DNA -3'
O
O
O
O
O
O
Stage E
Base
D116
D64
E152
O
O
Virus DNA-5'
O
P
O
HO
O
O
Base(A)
O
O
O
O
O
O
D116
D64
E152
Scheme 1. The two-metal-ion catalysis and inhibition mechanism. Stage A to Stage C indicates the 30-processing reaction and Stage C to Stage E the
strand transfer reaction. (Stage A to B) The enzyme recognizes the adenine base conserved in the third position from 30-end of viral DNA1, then
activates the next phosphoric ester with the two metals. (Stage B to C) The activated phosphoryl ester is hydrolyzed to excise the terminal
dinucleotide and the recognized adenosine is exposed as the new 30-end, giving a pre-integration complex.1 (Stage C to D) The pre-integration
complex non-specifically1 binds to host DNA to activate a phosphoryl ester by the two metals. (Stage D to E) The activated phosphoric ester is
attacked by the recessed 30-end in the manner of SN2-like nucleophilic reaction,1 then the viral DNA and the host DNA are joined with each other.
(Stage X) An inhibitor chelates to the two metal ions of Stage C to block the host DNA binding.
in crystal structures of the HIV IN core domain
(Fig. 1a). No another metal ions were reported in crystal
structures of HIV IN in the Protein Data Bank,3 howev-
er, two Zn2+ or Cd2+ were detected in a core domain of
ASV IN (avian sarcoma virus integrase).4 ASV IN core
domain exhibits high sequential and functional homolo-
gy with the HIV IN core domain (Fig. 1a). In ASV IN,
one metal is coordinated by the carboxylic groups of
D64 and D121 with a similar coordination geometry
to that observed in the HIV IN cocrystal structure with
a Mg2+ (1biu.pdb, 1qs4.pdb). The other Mg2+ is coordi-
nated by D64 and E157. Each of the carboxylic oxygens
of D64 is coordinating with a different metal ion. Two
˚
metal ions are 3.6 A apart and surrounded by water
molecules. Such a complex of zinc or cadmium would
be artificial only in the crystallization buffer condition
because it does not promote a complete enzymatic reac-
tion.4 The coordinates of the active site presumably rep-
resent the structure that catalyzes the integration
reaction, and possibly become a primary template to
construct a relevant active site model of HIV IN. Preli-
minary modeling work was performed starting from the
co-crystal structure5 of HIV IN with a Mg2+ (1biu.pdb).
Only rotating torsion angles of the E152 residue and an

8422
T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
(a)
(b)
D64
E152
D116
Figure 1. Modeling of HIV IN active site with two magnesium ions. (a) Superimposed model of the template and the starting material based on Ca
positions of each catalytic triad (ASV IN: D64, D121, E157. HIV IN: D64, D116, E152). Template: ASV IN active site with two Zn2+ (1vsh.pdb).
Zn2+ are represented as yellow spheres, and the catalytic triad by yellow sticks. White lines indicate coordination of the carboxylic oxygen to the
metal ion. Starting material: HIV IN active site with a Mg2+ (1biu.pdb). Mg2+ is represented as a magenta sphere, and the catalytic triad by stick
model colored by atom type. (b) A relevant model of HIV IN active site with two Mg2+, achieved by torsional rearrangement of E152 side chain and
the additional metal ion.
a
HO2C
extra magnesium ion were sufficient to be compared to
the active site of ASV IN as shown in Figure 1b. The
Mg2+ from the starting material is coordinated by D64
and D116, and the additional ion by D64 and E152.
Each of the carboxylic oxygens of D64 is coordinating
Cmpd.1
Cmpd.2
Cmpd.3
OH
N
O
Ph
Ph
HO2C
HO2C
˚
to different metal ions with a distance of 3.6 A. The ac-
OH
OH
N
N
N
tive site exposes two Mg2+ on its surface for access to
DNA.
We previously reported 2-hydroxy 3-heteroaryl acrylic
acid derivatives (HHAAs) as potential inhibitors against
HIV IN.6 The general structure is composed of a substi-
tuted heteroaromatic ring ‘Prime-Ring’ and a 2-hydroxy
acrylic acid substructure (HAA), represented by three
compounds with characteristic hydrophobic substitu-
ents (Fig. 2a). Modeling studies of the three compounds
identified ‘the most favorable benzene (MFB) space’
where the terminal phenyl ring should occupy as shown
in Figures 2b and c. Also the carboxylic acid, the enolic
hydroxyl group, and the common nitrogen atom in the
Prime-Ring were found to be essential for inhibitor
activity. Finally we produced the pharmacophore model
which is composed of the three hydrophilic functional
groups and the MFB space.
Ph
O
b
c
Hydrophilic Domain
We propose a two-metal binding model7 as a potential
inhibition mechanism of HHAAs against HIV IN. An
inhibitor chelates to the two metal ions at Stage C to
block access of a host DNA to the integrase enzyme
as shown in Stage X in Scheme 1. Inhibition of HIV
IN can also occur through Stage A however, the
HHAAs being strand transfer selective inhibitors as
has been shown for the related DKAs.8 A potential rea-
son that only one Mg2+ has been detected in the active
site in crystallographic studies is that binding of the viral
DNA might be critical for stability of the second metal
binding and thus a more realistic representation of actu-
al physiological conditions. The inhibitors would need
both metal ions to bind strongly. A series of modeling
studies were performed using the typical HHAAs as test
ligands to construct a binding model based on this
hypothesis. This inhibition model was compared with
the structure–activity relationships (SARs) and metal
titration studies of HHAAs.
MFB
Hydrophobic Domain
Y
X
O
Figure 2. Pharmacophore model for HIV IN inhibitors. (a) Potent
inhibitors representing different types of HHAAs. (b) The inhibitors
are superimposed based on the hydrophilic region, and every terminal
phenyl rings occupy an identical space. (c) The pharmacophore model
consists of the hydrophilic and the hydrophobic features. The atoms
‘X’ and ‘Y’ represent possible hetero-atoms that serve a lone-pair
indicated by white arrows. The yellow hexagon labeled MFB (the most
favorable benzene) indicates the average space of all terminal phenyl
rings in the superimposed model. The arrow passing through the MFB
indicates a favorable direction of the ring plane. The semicircles
written in broken line indicate heteroaromatic rings optionally
including the ‘C@X’ or ‘C@Y’ bond. Two bonds connecting three
hydrophilic functional groups are possible to rotate. The winding line
indicates a possible linker to connect the hydrophilic domain and a
terminal phenyl ring.

T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
8423
The original concept of the two-metal binding model
was disclosed in a patent application by our group in
2001.7 Subsequently, others have published similar ideas
to present not only an inhibition mechanism against
HIV IN but also a potential strategy9 to design novel
inhibitors of several nucleotide-related enzymes working
with two metal cofactors. In brief, new inhibitors can be
designed as a possible two-metal chelating scaffold with
appropriate substituents for each target molecule.
Recently a general inhibition model of DKAs and some
chelatable agents against influenza endonuclease was
reported by Parkes.10 Herein we present a modeling
study of the two-metal inhibition model of the HHAAs,
with additional SAR information around the HHAAs.
X
X
X
O
Y
Y
Substr. A
X
OH
O-
O
Y
Substr. B
Metal Ion
O
Y
X
2. Chemistry
O
Y
A selective benzyl extension on the methyl group of
compound 46 was achieved by deprotonation and alkyl-
ation with benzyl bromide resulting in formation of
compound 5 (Scheme 2). The condensation reaction
was performed with excess di-tert-butyl oxalate under
anhydrous conditions to give the corresponding ester
6. Hydrolysis using trifluoroacetic acid served to provide
compound 7. The corresponding ethyl ester of com-
pound 6 derived from compound 5 and diethyl oxalate
was not useful since attempts at basic hydrolysis resulted
in complete cleavage back to compound 5. Compounds
9 and 10 were prepared from compound 86 by aminoly-
sis under refluxing conditions. Compound 12 was ob-
tained from compound 116 by amide coupling.
Figure 3. The hydrophilic pharmacophore with two-metal ions. The
equilibrium indicates translations from/to each localized form of keto,
enol, enolate anion, and two metal chelating state. Substructures A
and B pictorially represent each bidentate chelating manner.
dle hydroxyl group should be easily ionizable resulting
in a formal negative charge enhancing its ability to coor-
dinate to both metal ions at the same time. This is in
agreement with the fact that a reduction of the enolic
group to the corresponding secondary alcohol gave a
complete loss of activity as reported.6 Although an alter-
nate keto form can serve two lone-pairs to coordinate to
the two metals, the neutral molecule is unlikely to che-
late strongly. It should be noted that the proposed coor-
dinating system is represented in the cocrystal structure
of 4-(10-carbonyl-20-oxopropylidene)-2,2,5,5-tetrameth-
yl-3- imidazolidine-1-oxyl with divalent copper ion
(Fig. 4).11 In brief, two 5,6-parallel chelating systems
are constructed with two ligands and two copper ions
3. Results and discussion
The hydrophilic pharmacophore region with two metal
ions shows a 5,6-bicyclic chelating system, as shown in
Figure 3. The hydrophilic region is composed of sub-
structure A and B, known to be bidentate chelatable
fragments making a five-membered and a six-membered
coordinating system, respectively. In this case, the mid-
˚
4 A apart from each other.
Scheme 2. Preparations of HHAA and amide derivatives. Reagents and conditions: (i) n-BuLi, benzylbromide/THF ꢀ78 °C; (ii) n-BuLi, (CO2-tert-
Bu)2/THF ꢀ78 to 0 °C; (iii) TFA/CH2Cl2 rt; (iv) methylamine or ammonia, AcOH/EtOH, reflux; (v) PyBOP, morpholine, HOBt/THF, rt.

8424
T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
(a)
Figure 4. The known crystal structure of two-metal chelating state. It
represents a five-membered and a six-membered chelating system
partially overlapping with each other. Two Cu2+, two water molecules,
and two main molecules are observed in an asymmetric unit.
(b)
Figure 5a represents a structural model of the 5,6-paral-
lel chelating system consisting of a chelating region of
HHAAs and two divalent magnesium ions. Both of
the five- and six-membered coordinating systems were
structurally optimized to be completely planar and flush
with each other. The distance between the two magne-
˚
sium ions represents 3.6 A.
(c)
The distances between two metal ions in the ligand mod-
el are almost identical to that in the HIV IN model de-
scribed above. This suggested to us that active inhibitors
might bind to the two metals in the active site. Our
docking approach was simple, docking the ligand model
into the protein model so as both metal pairs overlap
with each other. Figures 5c and d show a binding model
that best achieves tetrahedral-like coordination in both
chelating systems. An alternative reversed direction
mode, which is not shown in the figure, can be modeled
with similar coordinating systems.
3.1. Metal titration studies
(d)
The two-metal inhibition model can directly elucidate
why compounds 13–156 were inactive (Fig. 6). Every
heteroatom in the pharmacophore would be indispens-
able to chelate to the two metal ions. Mutated com-
pounds 13 and 15 have ability to chelate to only one
metal ion whereas compound 14 is unable to chelate.
These results are supported by following metal titration
studies using Mg2+ ranging 0 to 3.6 mol/L in the aque-
ous condition. Figures 6c and e show simple change of
UV absorption curves by metal chelation for each of
compounds 13 and 15 as indicated by the arrow. No
shift was observed for compound 14. On the contrary,
the original compound 2 showed an interesting two-step
shift. As indicated by arrows in Figure 6a raising the
metal concentration, first induces a change in absorption
around 380 nm, followed by the top of the curve shifting
toward shorter a wavelength around 340 nm. It could
not strictly demonstrate the generation of the 5,6-paral-
lel chelating system, but might indicate participation of
the three functional groups in metal coordination in the
final status observed at higher metal concentration. It
was not observed for one-metal chelating agents 13
and 15 even at high metal concentration.
Figure 5. A docking mode of the inhibitor, and the formation of
coordinating systems. (a) The graphic shows an optimized model of a
HHAA’s hydrophilic domain with two Mg2+. (b) Side chains of the
catalytic triad and two Mg2+ extracted from the protein model. A view
from the bottom of Figure 1b. (c) A model of the ligand binding to the
two metal ions. (d) A view from the top of (c). Magenta spheres and
white lines indicate Mg2+ and coordination bonds, respectively.
curves were found using monovalent ions Na+ and
K+, divalent metal ions Ca2+ and Mn2+ showed two-
step shifts. Ca2+ gave a similar shift to that of Mg2+
whereas Mn2+ did not. It should be emphasized that
50 mM Mn2+ is sufficient to find the second shift indicat-
ing generation of the last component, even 500 mM is
not enough as yet in the case of Mg2+. This indicates
that Mg2+ needs higher concentration than Mn2+ to
Other metal ions were investigated to see if they show
the same results. Though no shifts of UV absorption

T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
8425
b
a
O
O
O
HO
O
N
N
OH
N
N
Cmpd. 2
c
Cmpd. 13
O
N
N
d
O
Cmpd. 14
HO
N
N
e
O
O
Cmpd. 15
O
N
Figure 6. Relationship between potential chelating states and the observation from metal titration studies. The gray spheres indicate metal ions. The
pictorial structures of compounds 13–15 indicate each potential chelating state. The bending arrows emphasize the mutated position in comparison
with model compound 2. (a and b) The UV spectrum changes of compound 2 by the titration of Mg2+ and Mn2+, respectively. The bold curve in (a)
indicates the spectrum at 500 mM Mg2+. The bold curve in (b) at 50 mM of Mg2+. (c–e) The UV spectrum changes of each mutated compound by the
titration of Mg2+. Arrows in the chart indicate directions of the absorption curve’s shift depend on concentration of the metal ion.
generate the final state, or compound 2 is likely to che-
late to Mg2+ less than Mn2+. Compound 2 showed high
potential activity in the MWPA-Mn enzymatic assay
using Mn2+ as a cofactor, but shows 2 orders of magni-
tude lower activity in the MWPA-Mg using Mg2+ for
the replacement of Mn2+ as shown in Table 1. The
correlation between the inhibitory activity reduction
and the metal affinity reduction indicates that the inhib-
itor directly interacts with metal ions in the active site of
HIV IN. Compounds 1 and 3 are representing the differ-
ent types of HHAAs, showed a similar reduction of
activity in the MWPA-Mg, and must show a similar
reduction of metal affinity. Among the monovalent
and divalent metal cations, only Mn2+ and Mg2+ are
known to promote total enzymatic function in vitro. It
should be noticed that antiviral activities in Table 1
are slightly correlated with MWPA-Mg but not
MWPA-Mn. Compounds 2, 3, 9, and 10 show moderate
activities in both MWPA-Mg and anti-HIV assays,
however these are ranging over 3 orders of magnitude
in MWPA-Mn. The correlation with MWPA-Mg may
provide an experimental support of the generally accept-
ed opinion that Mg2+ is the physiological cofactor be-
cause of its dominant concentration in cells.5 Finally it
can be suggested that the deficient affinity to Mg2+ rep-
resents a possible reason of insufficient antiviral activi-
ties of HHAAs.
Table 1. Inhibitory activities of each compound in enzymatic and
cellular assay
c
Compound
MWPA-Mna
MWPA-Mgb
MTT EC50
(lM)
(lM)
(lM)
1
2
0.059
0.037
0.37
2.2
9.6
4.8
34
3
12
78
7
>277
71
39
277
6.7
11
>55
49
9
10
12
24
>59
262
236
a,b
Inhibitory activities against the strand transfer. Different metal
cofactors were used for each method.
c
Anti-HIV activities.

8426
T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
3.2. Other SARs of HHAAs
Ph
N
Ph
O
O
Compounds 7, 9, 10, and 12 listed in Figure 7 complete-
HO
O
ly agree with the hydrophilic and hydrophobic pharma-
cophore summarized in Figure 2, but these were
nevertheless inactive in enzymatic assay (MWPA-Mn,
Mg) as shown in Table 1. Those contradictions can be
clearly explained by the two-metal inhibition model.
The model gives additional rules as follows: (1) Every
coordinating hetero-atom to serve a lone-pair should
be arranged to face the same way to chelate the adjoin-
ing two metal ions. (2) A chelating region of an inhibitor
should be planar to point each lone-pair toward the met-
al ions. The extra benzyl group of compound 7 induces
serious intramolecular repulsion and consequently
warps the chelating region. A ‘N–H’ group of the car-
bamoyl and the methyl carbamoyl derivatives 9 and 10
makes unfavorable intramolecular hydrogen bonding
with the enolic oxygen and consequently let the amide
oxygen face away from the metal ion. Also, the ‘N–H’
group is not acidic enough to serve as a Lewis base li-
gand. Though the morpholyl amide derivative 12 does
not have such a hydrogen bonding donor, a methylene
next to the amide nitrogen makes an unfavorable intra-
molecular repulsion, giving a contortion of the plane.
The 2-pyridyl derivative 16 shows some reduction of
inhibitory activity in comparison to 2-pyrimidyl one.6
The carbon–hydrogen unit for the replacement of the
nitrogen atom of the 2-pyrimidyl moiety makes a small
intramolecular repulsion as depicted in Figure 7. This
discussion does not contradict a potential reason for
the activity loss due to extra volume of the benzyl or
the morpholyl group.
N
<<
O
N
OH
N
Cmpd.7
N
N
O
N
N
<<
O
N
N
Cmpd.16
R
O
NH
O
R
N
O
N
N
H
O
N
N
<<
Cmpd.9: R = H, 10: R = Me
O
N
O
N
O
O
O
N
N
O
N
N
<<
Cmpd.12
Figure 7. Explanations of why the compounds do not show potential
activity. This is to illustrate observations in molecular modeling for the
compounds. Sign of inequality indicates likelihood of the conforma-
tional change. The red star is attached around the site where the
unfavorable intramolecular repulsion can be observed. The blue
broken line means coordination to a metal ion represented as a gray
sphere, the red broken line is an intramolecular hydrogen bonding.
The red curved arrow indicates the bond which would be rotated by an
unfavorable intramolecular repulsion or hydrogen bonding.
4. Conclusions
Figure 8 shows one possible docking model of three
compounds 1–3 representing different types of HHAAs
a
b
K159
K156
N155
D64
E152
D116
N-terminal
C-terminal
Figure 8. A docking model of HHAAs with IN core domain. (a) IN core domain is represented by a green tube. Side chains of the catalytic triad are
shown by stick representation as labeled. Magenta spheres indicate Mg2+. The molecules represented by cyan sticks indicate the active conformers of
three compounds 1–3 overlapped with each other. The white dotted line indicates a flexible loop (140–149aa) blank in coordinate data. (b) With a
Connolly surface around the active site.

T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
8427
in the active site. Those compounds are reproducing
each active conformer (a mirror image of Fig. 2b) and
chelate to the two Mg2+ held by the catalytic triad of
HIV IN in the manner of pseudo-tetrahedral coordina-
tion (Figs. 5c and d). The terminal phenyl rings in the
docking model are placed close to the hydrophobic re-
gion of N155, K156, and K159, and no specific interac-
tion or repulsion of the linkers with the protein can be
observed. In brief, there is no contradiction between
the MFB space model and the observation from the
docking model. It should be emphasized that N155S
was previously12 reported to be a resistant mutant of
S-1360, whose terminal phenyl ring can be placed6 at
the MFB space to interact with the hydrophobic region.
This assists the plausibility of this docking mode, but
does not exclude the other modes and should be investi-
gated further. We will need to consider folding of a flex-
ible loop (140–149aa) close to the active site, folding of
other domains, and multimer formation of INs and
DNAs.
4H), 3.79 (s, 2H), 6.69 (d, J = 1.2 Hz, 1H), 7.05 (d,
J = 3.0 Hz, 2H), 7.17–7.27 (m, 8H), 8.77 (d,
J = 1.2 Hz, 1H).
5.1.3. 2-Hydroxy-3-(6-phenethyl-pyrimidin-4-yl)-4-phen-
yl-but-2-enoic acid (7). A solution of compound 6
(100 mg, 0.24 mmol) in CH2Cl2 (2 mL) and TFA
(1 mL) was stirred for 5 h at room temperature. Then
the solvent was removed in vacuo, and the precipitate
was washed with EtOAc to give the product as pale yel-
low crystals (80 mg, 92% yield). Mp 188–190 °C, 1H
NMR (DMSO-d6) d 2.88 (s, 4H), 3.78 (s, 2H), 6.95 (s,
1H), 7.05–7.28 (m, 10H), 8.81 (s, 1H). Anal. Calcd for
C22H20N2O3Æ0.3H2O: C, 72.23; H, 5.68; N, 7.66. Found:
C, 72.41; H, 5.47; N, 7.46.
5.1.4. 2-Hydroxy-3-(6-phenethyl-pyrimidin-4-yl)-acryl-
amide (9). This compound was prepared by a similar
method to that described for compound 10. Yellow crys-
tal, 52% yield. Mp 155–156 °C, 1H NMR (CDCl3) d
3.00–3.05 (m, 4H), 5.70 (br s, 1H), 6.41 (s, 1H), 6.80
(d, J = 1.2 Hz, 1H), 6.94 (br s, 1H), 7.15–7.35 (m, 5H),
8.83 (d, J = 1.2 Hz, 1H). Anal. Calcd for C15H15N3O2:
C, 66.90; H, 5.61; N, 15.60. Found: C, 66.90; H, 5.62;
N, 15.69.
5. Experimental
5.1. Chemistry
1H NMR spectra were determined at 300 MHz. All reac-
tions were carried out under a nitrogen atmosphere with
˚
anhydrous solvents that had been dried over type 4 A
molecular sieves.
5.1.5. 2-Hydroxy-N-methyl-3-(6-phenethyl-pyrimidin-4-
yl)-acrylamide (10).
(1 mL, 30 w/v% in EtOH) was added to a solution
of 2-hydroxy-3-(6-phenethyl-pyrimidin-4-yl)-acrylic
A
solution of methylamine
acid ethyl ester 86 (200 mg, 0.67 mmol) and acetic
acid (40 mg, 0.67 mmol) in EtOH (5 mL). The mix-
ture was refluxed for 1 h then the solvent was re-
moved in vacuo. The residue was purified by flash
column chromatography using n-hexane/EtOAc = 1:1
as eluent to give the product as yellow crystals
(50 mg, 26% yield). Mp 155–157 °C, 1H NMR
(CDCl3) d 2.95 (d, J = 5.1 Hz, 3H), 3.05 (s, 4H),
6.41 (s, 1H), 6.79 (s, 1H), 7.06 (s, 1H), 7.15–7.32
(m, 5H), 8.88 (s, 1H). Anal. Calcd for
C16H17N3O2Æ0.9H2O: C, 64.16; H, 6.33; N, 14.03.
Found: C, 64.13; H, 6.08; N, 14.18.
5.1.1. 4,6-Diphenethyl-pyrimidine (5). A solution of
n-BuLi (6.7 mL, 1.50 M in n-hexane, 10 mmol) was
added to a solution of 4-methyl-6-phenethyl-pyrimi-
dine 46 (1.98 g, 10 mmol) in THF (10 mL) at
ꢀ78 °C. Then a solution of benzyl bromide (1.71 g,
10 mmol) in THF (2 mL) was added to the mixture
at ꢀ78 °C. The mixture was diluted with a saturated
aqueous solution of NH4Cl and extracted with
EtOAc. The organic layer was washed with water
and brine, and dried over MgSO4. The solvent was
removed in vacuo, and the precipitate was washed
with n-hexane to give the product as colorless crystals
1
(1.61 g, 56% yield). H NMR (CDCl3) d 3.01 (s, 8H),
6.84 (d, J = 1.5 Hz, 1H), 7.15–7.30 (m, 10H), 9.08 (d,
J = 1.5 Hz, 1H).
5.1.6. 2-Hydroxy-1-morpholin-4-yl-3-(6-phenethyl-pyrim-
idin-4-yl)-propenone (12). A mixture of 2-hydroxy-3-(6-
phenethyl-pyrimidin-4-yl)-acrylic acid 116 (100 mg,
0.37 mmol), PyBOP (420 mg, 0.8 mmol), N-meth-
ylmorpholine (80 mg, 0.8 mmol), HOBT (110 mg,
0.8 mmol) and morpholine (70 mg, 0.8 mmol) in
THF (10 mL) was stirred for 4 h at room tempera-
ture. The reaction mixture was diluted with EtOAc,
washed with water and a saturated aqueous solution
of NaHCO3, then dried over Na2SO4. The solvent
was removed in vacuo, then the residue was purified
by flash column chromatography using EtOAc/
MeOH = 20:1 as eluent. Crystallization with n-hexane
and EtOAc gave the desired compound as yellow
crystals (60 mg, 48% yield). Mp 130–132 °C, 1H
NMR (CDCl3) d 2.95–3.10 (m, 4H), 3.65–3.80 (m,
8H), 5.74 (s, 1H), 6.62 (d, J = 1.2 Hz, 1H), 7.18–
7.35 (m, 5H), 8.62 (s, 1H). Anal. Calcd for
C19H21N3O3Æ0.8H2O: C, 64.50; H, 6.44; N, 11.88.
Found: C, 64.33; H, 6.30; N, 11.88.
5.1.2. 2-Hydroxy-3-(6-phenethyl-pyrimidin-4-yl)-4-phen-
yl-but-2-enoic acid tert-butyl ester (6). A solution of
n-BuLi (2.0 mL, 1.50 M in n-hexane, 3 mmol) was added
to a solution of compound 5 (865 mg, 3 mmol) in THF
(10 mL) at ꢀ78 °C. The reaction mixture was stirred for
30 min and then oxalic acid di-tert-butyl ester (3.0 g,
15 mmol) was added in one portion. The mixture was
then warmed to 0 °C and stirred for 30 min. After dilu-
tion with a saturated aqueous solution of NH4Cl, the
mixture was extracted with EtOAc. The organic layer
was washed with water and brine, and dried over
Na2SO4. The solvent was removed in vacuo, then the
residue was purified by flash column chromatography
using n-hexane/EtOAc = 3:1 as eluent to give the prod-
uct as pale yellow crystals (780 mg, 62% yield). Mp
1
116–118 °C, H NMR (CDCl3) d 1.47 (s, 9H), 2.91 (s,

8428
T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
5.2. Metal titration studies
of the ligands was superimposed into the coordination
model based on the hydrophilic region using the Fit
Atoms command of SYBYL, giving the final docking
model.
UV spectrum data ranging 250–450 nm as wavelength
were collected using 1 cm cell. The solution conditions
were unified for every measurement as MOPS–NaOH
buffer (50 mmol/L, pH 7.4) with 10% DMSO. NaCl
was added in exchange for the titrated divalent metal
ion to keep the ion strength constant. Compound 2
(50 lmol/L) was titrated by MgCl2 of 0.0, 0.005, 0.01,
0.03, 0.05, 0.5, 1.0, 1.5, 2.0, 3.6 mol/L, or by MnCl2 of
0.0, 0.0005, 0.002, 0.005, 0.05, 0.1 mol/L. Compound
13 (50 lmol/L) was titrated by MgCl2 of 0.0, 0.5, 1.0,
2.0, 3.6 mol/L. Compound 14 (50 lmol/L) was titrated
by MgCl2 of 0.0, 0.5, 1.0, 3.6 mol/L. Compound 15
(50 lmol/L) was titrated by MgCl2 of 0.0, 0.005, 0.05,
0.5, 2.0 mol/L.
5.4. Assay
Evaluation of inhibitory activities was done by the same
method as described in the companion paper.6 MWPA-
Mg enzymatic assay was done with a similar manner to
that of MWPA-Mn with a minor modification as the fol-
lowing. Divalent manganese ion was used as a cofactor
through all the steps of MWPA-Mn assay. On the con-
trary, the manganese ion was replaced with divalent
magnesium ion after making IN-DNA complex in
MWPA-Mg assay, then incubation of a compound and
a strand transfer reaction were performed with the same
manner of MWPA-Mn assay.
5.3. Molecular modeling
Modeling studies were performed using the SYBYL 6.9
software package13 and the WinMOPAC v 2.0 pro-
gram.14 Schematic representations of the models were
prepared with SYBYL.
Acknowledgments
The authors thank Prof. Shu Kobayashi for valuable
discussions and Dr. Brian A. Johns for assisting with
the manuscript preparation.
5.3.1. Modeling of the ligand model. The 3-D molecular
structure of the hydrophilic domain was generated
using the Sketch Molecule command of SYBYL.
After a relaxation of the hydrophilic domain using
the Minimize command, two magnesium atoms were
References and notes
˚
attached to be 2.1 A from each coordinating atom.
1. For recent reviews on structure and biology of HIV
integrase, see: Chiu, T. K.; Davies, D. R. Curr. Top. Med.
Chem. 2004, 4, 965–979.
2. (a) Yang, W.; Steitz, T. A. Structure 1995, 3, 131–134; (b)
Haren, L.; Ton-Hoang, B.; Chandler, M. Annu. Rev.
Microbiol. 1999, 53, 245–281.
The complex was optimized using a semiempirical
molecular orbital method with MNDOD Ham-
iltonian.
5.3.2. Modeling of the preliminary protein model. The
atomic coordinates of crystal structures of HIV IN
and ASV IN were retrieved from the Protein Data
Bank3 (entry: 1biu.pdb and 1vsh.pdb, respectively).
Water molecules were removed from the coordinate sets,
giving a primary model of HIV IN and a template model
of active site, respectively. The HIV IN model was
superimposed over the template model using a rigid-
body least-squares fit based on coordinates of Ca of
the three acidic amino acid (D64, D116, E152 in HIV
IN and D64, D121, E157 in ASV IN). A side chain of
the E152 in HIV IN model was fixed by manually rotat-
ing to overlap along that of the E157 in the template
model. Then an additional magnesium ion was generat-
ed in the same position of the zinc ion coordinated by
the D64 and E157. Hydrogen atoms were added for
whole molecule with standard geometry.
3. Bernstein, F. C.; Koetzle, T. F.; Williams, G. J. B.; Meyer,
E. F., Jr.; Brice, M. D.; Rodgers, J. R.; Kennard, O.;
Shimanouchi, T.; Tasumi, M. J. Mol. Biol. 1977, 112, 535–
542.
4. Bujacz, G.; Alexandratos, J.; Wlodawer, A. J. Biol. Chem.
1997, 272, 18161–18168.
5. Goldgur, Y.; Dyda, F.; Hickman, A. B.; Jenkins, T. M.;
Craigie, R.; Davies, D. R. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 9150–9154.
6. Kawasuji, T.; Yoshinaga, T.; Sato, A.; Yodo, M.; Fujiw-
ara, T.; Kiyama, R. Bioorg. Med. Chem., the companion
paper, doi:10.1016/j.bmc.2006.08.044.
7. Kiyama, R.; Kawasuji, T. PCT Int. Appl. 2001, WO-01/
95905.
8. (a) Hazuda, D. J.; Felock, P.; Witmer, M.; Wolfe, A.;
Stillmock, K.; Grobler, J. A.; Espeseth, A.; Gabryelski, L.;
Schleif, W.; Blau, C.; Miller, M. D. Science 2000, 287, 646–
650; (b) Espeseth, A. S.; Felock, P.; Wolfe, A.; Witmer, M.;
Grobler, J.; Anthony, N.; Egbertson, M.; Melamed, J. Y.;
Young, S.; Hamill, T.; Cole, J. L.; Hazuda, D. J. Proc.
Natl. Acad. Sci. U.S.A. 2000, 97, 11244–11249.
9. (a) Grobler, J. A.; Stillmock, K.; Hu, B.; Witmer, M.;
Felock, P.; Espeseth, A. S.; Wolfe, A.; Egbertson, M.;
Bourgeois, M.; Melamed, J.; Wai, J. S.; Young, S.; Vacca,
J.; Hazuda, D. J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99,
6661–6666; (b) Hazuda, D. J.; Anthony, N. J.; Gomez, R.
P.; Jolly, S. M.; Wai, J. S.; Zhuang, L., et al. Proc. Natl.
Acad. Sci. U.S.A. 2004, 101, 11233–11238.
5.3.3. Modeling of the docking model. The ligand model
was superimposed into the protein model as both metal
pairs overlap with each other by rigid-body manual
orientation. The ligand model was oriented to con-
struct a coordinating system that best achieves a tetra-
hedral-like coordination for both systems. Then the
two metals from the ligand model were removed, giv-
ing the coordination model shown in Figures 5c and
d. Each hydrophobic substituent of active compounds
1, 2, and 3 was attached to the ligand with special tor-
sion angles constructing active conformers. Then each
10. Parkes, K. E. B.; Ermert, P.; Fassler, J.; Ives, J.; Martin, J.
A.; Merrett, J. H.; Obrecht, D.; Williams, G.; Klumpp, K.
J. Med. Chem. 2003, 46, 1153–1164.

T. Kawasuji et al. / Bioorg. Med. Chem. 14 (2006) 8420–8429
8429
11. Ovcharenko, V. I.; Vostrikova, K. E.; Podoplelov, A. V.;
Romanenko, G. V.; Ikorskii, V. N.; Reznikov, V. A.
Polyhedron 1997, 16, 1279–1289.
on Retroviruses and Opportunistic Infections. Seattle,
Washington, USA, 24–28 February 2002. Abstracts,
55p, No. 8.
12. Yoshinaga, T.; Sato, A.; Fujishita, T.; Fujiwara, T. S-
1360: In Vitro Activity of a New HIV-1 Integrase
Inhibitor in Clinical Development. Ninth Conference
13. SYBYL 6.9; S. Tripos Inc.: St. Louis MO.
14. WinMOPAC v 2.0 program; Fujitsu Ltd: http://
www.fujitsu.com.