Application of CoMFA and CoMSIA 3D-QSAR and Docking Studies in Optimization of Mercaptobenzenesulfonamides as HIV-1 Integrase Inhibitors

Article abstract of DOI:10.1021/jm030378i

An essential step in the HIV life cycle is integration of the viral DNA into the host chromosome. This step is catalyzed by a 32-kDa viral enzyme HIV integrase (IN). HIV-1 IN is an important and validated target, and the drugs that selectively inhibit thi

Full text of DOI:10.1021/jm030378i

J . Med. Chem. 2004, 47, 385-399
385
Ap p lica tion of CoMF A a n d CoMSIA 3D-QSAR a n d Dock in g Stu d ies in
Op tim iza tion of Mer ca p toben zen esu lfon a m id es a s HIV-1 In tegr a se In h ibitor s
Chih-Ling Kuo,† Haregewein Assefa,‡ Shantaram Kamath,‡ Zdzialaw Brzozowski,§ J aroslaw Slawinski,§
Franciszek Saczewski,§ J ohn K. Buolamwini,‡ and Nouri Neamati*,†
Department of Pharmaceutical Sciences, School of Pharmacy, University of Southern California, 1985 Zonal Avenue,
PSC 304, Los Angeles, California 90089, Department of Pharmaceutical Sciences, College of Pharmacy, University of Tennessee
Health Sciences Center, 847 Monroe Avenue Suite 327, Memphis, Tennessee 38163, and Department of Chemical Technology of
Drugs, Medical University of Gdansk, 80-416 Gdansk, Poland
Received August 7, 2003
An essential step in the HIV life cycle is integration of the viral DNA into the host chromosome.
This step is catalyzed by a 32-kDa viral enzyme HIV integrase (IN). HIV-1 IN is an important
and validated target, and the drugs that selectively inhibit this enzyme, when used in
combination with reverse transcriptase (RT) and protease (PR) inhibitors, are believed to be
highly effective in suppressing the viral replication. IN catalyzes two discrete enzymatic
processes referred to as 3′ processing and DNA strand transfer. As a part of a study to optimize
new lead molecules we previously identified from a series of 2-mercaptobenzenesulfonamides
(MBSAs), we applied three-dimensional quantitative structure-activity relationship methods,
comparative molecular field analysis (CoMFA), and comparative molecular similarity indices
analysis (CoMSIA) to training sets of up to 66 compounds. Two different conformational
templates were used: Conf-d, obtained from docking into the HIV-1 IN active site and Conf-s
obtained by a systematic conformational search, using lead compounds 1 and 14, respectively.
Reliable models of good predictive power were obtained after removal of compounds with high
residuals. The Conf-s models tended to perform better than the Conf-d models. Cross-validated
coefficients (q2) of up to 0.719 (strand transfer CoMSIA, Conf-s) regression coefficients (r2) of
up to 0.932 (strand transfer CoMSIA, Conf-d) were obtained, with the number of partial least
squares (PLS) components varying from 3 to 6, and the number of outliers being 4 in most of
the models. Because all biological data were determined under exactly the same conditions
using the same enzyme preparation, our predictive models are promising for drug optimization.
Therefore, these results combined with docking studies were used to guide the rational design
of new inhibitors. Further synthesis of 12 new analogues was undertaken, and these were
used as a test set for validation of the quantitative structure-activity relationship (QSAR)
models. For compounds with closely related structures, binding energies given by the FlexX
scoring function correlated with HIV-1 IN inhibitory activity.
In tr od u ction
candidate (for recent reviews see refs 3 and 4). A lead
is defined as a compound that demonstrates a desired
biological activity on a validated molecular target. To
fulfill the criteria of what the industry considers a useful
lead, the compound must exceed a specific potency
threshold against the target (e.g., <10 µM inhibition).5
Many exciting new models for lead discovery have
recently emerged that facilitate more rapid identifica-
tion and result in compounds whose physical and
“druglike” properties could be initially optimized. Start-
ing with intuitive structural modification to the devel-
opment of structure-activity relationship (SAR) and
quantitative SAR (QSAR), one can gain tremendous
information about the requirement of certain functional
groups for activity. To be considered for further develop-
ment, lead structures should be amenable to chemistry
optimization, and have good ADMET (absorption, dis-
tribution, metabolism, excretion, and toxicity) proper-
ties.5
An obligatory requirement in the retroviral life cycle
is the integration of the viral double-stranded DNA into
the host chromosome.1,2 Integrase (IN) catalyzes two
separate but chemically similar reactions, known as 3′-
processing and strand transfer. In 3′-processing, IN
removes a dinucleotide next to a conserved cytosine-
adenine sequence from each 3′-end of the viral DNA.
IN then attaches the processed 3′-ends of the viral DNA
to the host cell DNA in the strand transfer reaction.
Because IN has no cellular homologues and is also an
essential enzyme for effective viral replication inhibitors
of this enzyme are of paramount importance for the
treatment of HIV infection.
Many of the identified lead inhibitors of IN were not
subsequently optimized to produce a clinically viable
* To whom correspondence should be addressed. (N.N.): Department
of Pharmaceutical Sciences, School of Pharmacy, University of South-
ern California, 1985 Zonal Avenue, PSC 304BA, Los Angeles, CA
90089. Phone: 323-442-2341. Fax: 323-442-1390. E-mail: neamati@
usc.edu. Questions regarding modeling studies should be addressed
to J .K.B. at jbuolamwini@utmem.edu.
Because the 2-mercaptobenzenesulfonamide (MBSA)
class of compounds were potent IN inhibitors and
exhibited antiviral activity in cell-based assays with
minimal cytotoxicity (see ref 6), we were prompted to
† University of Southern California.
‡ University of Tennessee Health Sciences Center.
§ Medical University of Gdansk.
10.1021/jm030378i CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/11/2003

386 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Kuo et al.
F igu r e 1. Structures of mercaptoaryl-containing inhibitors of HIV-1 integase.
Sch em e 1. Synthesis of the 4-Chloro-mercapto-5-
methyl-N-(3-arylquinoxalin-2-yl)benzenesulfonamides
60-62.
employ a detailed QSAR as well as docking studies to
better understand their inhibitory properties. Our inter-
est in optimization of the MBSA series of compounds
as leads was based on several previous observations. (i)
The first 3D-pharmacophore-based study of the NCI
database7 generated a series of sulfonated (e.g., I)
compounds as inhibitors of IN, and one of those com-
pounds was subsequently cocrystallized with ASV IN,
providing the first crystal structure of IN-bound drug.8
(ii) Among a variety of sulfones (e.g., II),7,9,10 sulfides
and sulfonates tested as IN inhibitors, the MBSA class
of compounds showed superior antiviral activity in cell-
culture (e.g., III).6 (iii) High throughput docking of the
NCI database using the cocrystal structure of ASV-IN
in complex with compound I yielded several sulfonates
(e.g., IV) and sulfonamides as leads.10 (iv) Among a
series of hydrazides reported by us earlier, the mercap-
tosalicyl hydrazides (e.g., V) were shown to be selective
against IN and exhibited antiviral activity.11 In fact,
among the recently reported dipyrimidine inhibitors,
only compounds containing a free mercapto group
showed antiviral activity and inhibited IN (e.g., VI,
Figure 1).12 (v) Availability of numerous analogues
tested under the same conditions proved useful in
establishing a QSAR model. (vi) In general, these types
of structures have shown reduced cytotoxicity and many
clinically used antibacterials contain sulfones and sul-
fonamides (for a recent review see ref 13), making them
indeed “druglike”.
A series of 4-chloro-5-carbamoyl-2-mercapto-N-(4H-
1,2,4-triazol-3-yl)benzenesulfonamides derivatives 63-
76 and C-1 were synthesized by the reaction of 3-amino-
benzodithiazines (B-1-B-15)14-16 with 99% hydrazine
hydrate in DMF at 100 °C, as shown in Scheme 2. The
reactions were carried out under nitrogen atmosphere
to avoid spontaneous air oxidation of the mercaptane
functioanlity formed. Then, the 2-mercapto-N-(4H-1,2,4-
triazol-3-yl)benzenesulfonamides C-1 and 67 were sub-
jected to the reaction with chloroacetic acid under
alkaline conditions to give the S-alkylated N-(4H-1,2,4-
triazol-3-yl)benzenesulfonamides 77 and 78, respec-
tively.
In h ibition of HIV-1 IN. The two most extensively
studied classes of IN inhibitors are the hydroxylated
aromatics and the â-diketoacids17 (for recent reviews see
refs 3, 4, and 18). Although the latter class proved to
be very promising the hydroxylated aromatics especially
of the catechol type have so far not yielded any potential
drug candidate. In an effort to develop a new class of
IN inhibitors lacking hydoxylated aromatic moieties,
which in most cases proved to be responsible for their
cytotoxicity and lack of selectivity, we relied on a series
of sulfonamides that previously were shown to inhibit
Resu lts a n d Discu ssion
Ch em istr y. The desired 4-chloro-2-mercapto-5-meth-
yl-N-(3-arylguanidin-2-yl)benzenesulfonamides 60-62
were obtained by reacting o-phenylenediamine with
corresponding 3-benzoyl-6-chloro-7-methyl-1,1-dioxo-
1,4,2-benzodithiazines A1-A3 in toluene under reflux
(Scheme 1).

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 387
Sch em e 2. Synthesis of the 5-Carbamoyl-4-chloro-N-
(4H-1,2,4-triazyn-3-yl)benzenesulfonamides 63-78a
transfer inhibition, respectively, while the IC50 values
of compounds 65 and 66 were 80 and 37 µM, respec-
tively.
In using the 3D-QSAR and docking analyses reported
here to optimize these sets of molecules and to further
investigate how the substituents on the triazolo ring and
amide affect the activity of compounds, compounds 67-
78 were designed and synthesized. The QSAR models
tended to attribute high inhibitory activities to these
compounds, and the docking studies predicted them to
be among the high affinity ligands. Among these new
analogues, compound 76 with IC50 values of 4 µM
against both 3′-processing and strand transfer activities
was the most potent. Interestingly, the FlexX docking
study predicted it to have the highest binding affinity
among the test set compounds. A subtle shift in potency
was observed when comparing compounds with different
substitution at the R1 group. Interestingly, when the
IC50 value of compound 76 was compared with those of
the corresponding alkyl analogues (68 and 71), the
presence of the phenyl ring increased the activity by 10-
fold. The docking also suggested that the compounds
with the phenyl ring generally bound the enzyme with
higher affinity (see FlexX scores). Surprisingly, the
activities of the compounds (77 and 78), in which the
mercapto group was protected, were not abolished.
Compound 78 still retains the activity of compound 67.
Compound 77 has an IC50 value of 333 µM, although it
is 5 times higher than that of compound 69. This may
be due to the possibility that under the experimental
condition the carboxylic group will be ionized. On the
basis of these data, several pertinent features emerge
to be important for IN inhibition. Substituting a bulky
group at 5-position and at sulfonamide may enhance
potency. Furthermore, a negative charge at 2-position
is important for activity and the mercapto group at
2-position is important for high activity, but may not
be essential.
a
Reagents and conditions: (a) H2N-NH2‚xH2O 99% (6 molar
equiv), dry DMF (12 molar equiv), under a nitrogen atmosphere,
100 °C, 15 h (63-66, 75, 76), 10-11 h (67-74); (b) H2N-NH2‚H2O
99% (4 molar equiv), dry DMF (12 molar equiv), under a nitrogen
atmosphere, 100 °C, 4
ClCH2CO2H (1 molar equiv), Na2CO3 (1 molar equiv), H2O 40 °C,
h (C-1); (c) NaOH (1 molar equiv),
1 h, 90-92 °C, 20 h; (d) HCl/H2O.
IN and exhibit antiviral activity in cell culture.6 We
selected three sets of compounds from three separate
studies to build a unified model for further drug
optimization. The first series of compounds (1-30 and
44-47) were originally discovered as IN inhibitors upon
testing approximately 5000 compounds that were se-
lected by the National Cancer Institute (NCI) Drug
Synthesis and Chemistry Branch as compounds with
antiviral activity in cell-based assays.6 The second series
of compounds (31-43 and 48) were discovered using a
three-point pharmacophore searching of the NCI data-
base.7 The third series of compounds (49-54) were
selected from a high throughput docking of the NCI
database using the X-ray structure of ASV integrase in
complex with a naphthalene bisulfonate, Y3.10 More-
over, a series of new compounds (55-66) that was
initially synthesized to answer the question of the
nature of the free mercapto group and substitution on
the sulfonamide moiety were also included in our
training set for our QSAR models (see below). Com-
pounds 55-66 showed very promising activity against
purified IN.
3D-QSAR Mod elin g. Partial least squares (PLS), the
statistical method used in deriving the 3D QSAR
models, is an extension of multiple regression analysis
in which the original variables are replaced by a small
set of their linear combinations.19-21 These latent
variables (components) so generated are used for mul-
tivariate regression, maximizing the commonality of
explanatory and response variable blocks. PLS has been
useful in cases in which the number of descriptors is
greater than the number of samples (compounds) as is
the case with comparative molecular field analysis
(CoMFA) and comparative molecular similarity indices
analysis (CoMSIA) 3D QSAR analyses.
Leave-one-out (LOO) cross-validated PLS analyses
were used to check the predictive ability of the models
and to determine the optimal number of components to
be used in the final QSAR models. The training sets
with the 3′-processing and strand transfer inhibitory
activities were composed of 66 and 64 compounds,
respectively. To generate a reasonable model in most
cases, four compounds were removed as outliers. In one
case, only two outliers were eliminated, whereas in
another case six compounds were eliminated from the
training set. Compounds 36 and 48 were common
outliers in CoMFA and CoMSIA models based on 3′-
processing inhibitory activity data, while compounds 36
In the previous report, the data indicated that the
bulkier group (two-rings and three-rings) substituted on
the N of the sulfonamide moiety was responsible for
enhanced potency. As expected, compounds 60-66 all
inhibited IN at low micromolar levels. Moreover, we
earlier showed that some compounds with an acetate
group at R5-position were more potent than compounds
having a methyl substituent at this position.6 Thus,
compound 63-66 were synthesized and tested for their
ability to inhibit IN. Compound 63 and 64 showed IC50
values of 6 and 12 µM for 3′-processing and strand

388 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Kuo et al.
and 57 were common outliers in models based on strand
transfer inhibitory activities. Compound 36 tended to
be an outlier in all QSAR models. This compound is the
only molecule with an aliphatic carboxylic acid sub-
stituent at the 4-position. Compound 48, which was a
common outlier in the 3′-processing models, is also
structurally different in terms of the substituents at the
4-position and on the sulfonamide group. Structural
uniqueness underscores the outlier status of compounds
for the most part.
values below 4.0. Compounds 49-54, which have pIC50
values above 4.0, are exceptions. This might be due to
the H-bond donor groups on the substituent moieties,
the interactions of which probably offset the steric
effects. Both the green contours appearing near the
5-position of the aromatic ring in Figure 4A and the
green contour map in Figure 4B stretching toward the
5-position of the aromatic ring suggest that a bulky
substituent at this position will enhance inhibitory
potency. A comparison of compounds 2 and 44 shows
that a change from a methyl to an acetate group at the
5-position increases the potency, which may be due to
an increase in the steric bulk of the group. The relatively
lower potency of compounds 45 and 46 as compared to
compounds 17 and 18 may be due partly to the absence
of any substituent at position 5. Compounds 63 and 64,
which have an N-phenyl carbamoyl substituent, have
high potencies possibly due to the relatively larger
substituent at this position. However, further increase
in the size of the substituent appears to decrease the
potency as shown by compounds 65 and 66. A green
contour around the phenylamino group attached to the
triazole ring in both Figure 4A,B suggests that bulky
substituents at this location might enhance potency.
Compounds with bulky substituents at this site such
as 3-5, 8-9, 12, 13, 17-23, and 25-28 have retained
high potency. The yellow contour on the other side of
the triazole ring shows that a bulky substituent around
that region would decrease potency. Red contours sur-
rounding the triazole ring show that high electron
density (electronegative groups) favors inhibitory activ-
ity. The relatively low potency of compounds 55-59,
which have a positive charge around this area, shows
the preference of a negative charge over a positive
charge.
Results of the PLS analyses are summarized in Table
3. Group cross-validation (10 groups) and scrambling
of the biological data (randomization control) were
performed to test the stability and reliability of the
models. The average q2 values obtained in the group
cross-validation and randomization exercises are pre-
sented in Table 4. These results indicate stability in the
CoMFA and CoMSIA QSAR models. The predictive
ability of the models was further validated using the
external test set of 12 compounds (see Table 2) in both
Conf-d and Conf-s conformations. The prediction of
activities for training set and test set compounds by the
different QSAR models generated are presented in
graphical form in Figures 2 and 3. The predictions for
the training set are shown as filled circles, whereas open
circles indicate predictions for the test set. In Figures 2
and 3, the solid line represents the regression fitted line
of the training set whereas the dotted line shows the
error limit of (1 log unit about the corresponding fitted
curve. The residuals from the prediction of the test set
by the final QSAR models are shown in Table 5.
CoMF A Usin g t h e In h ib it ion of 3′-P r ocessin g
Rea ction . With the 3′-processing inhibitory activity as
a response variable, initial CoMFA PLS analyses of the
66 compounds gave low q2 values. Removal of outliers
based on residual values afforded models with q2 values
of 0.557 and 0.630 for Conf-d and Conf-s models,
respectively (Table 3). For the two conformational sets,
r2 values of 0.921 and 0.858 were obtained for Conf-d
and Conf-s models, respectively. The graph of the actual
pIC50 versus the predicted pIC50 values for the training
set and test set compounds by the CoMFA models based
on 3′-processing are shown in Figure 2a,b. CoMFA
predictions with both Conf-d and Conf-s show a similar
trend. However, the Conf-s models performed better.
Examination of the residuals from the prediction (Table
5) shows that the inhibitory activities of 10 of the 12
compounds were predicted within (1 unit of the actual
pIC50 values by the model based on Conf-d, whereas the
Conf-s models predicted all 12 compounds well within
the (1 unit. In this instance, the model with a better
q2, and in which fewer outliers had to be removed
(Conf-s model) afforded a better predictive model. The
Conf-s model predicted the test set better than the
Conf-s possibly because the Conf-s had more structural
representatives than the Conf-d as far as the test set is
concerned.
CoMF A Usin g th e In h ibition of Str a n d Tr a n sfer
Rea ction . Using the strand transfer inhibitory activity
as a response variable, CoMFA PLS analyses of all the
64 compounds resulted in q2 values of 0.596 and 0.679
for the Conf-d and Conf-s conformational sets, respec-
tively, with removal of four outlier compounds. The r2
values for the final CoMFA models are 0.797 and 0.892
for the Conf-d and Conf-s conformational sets, respec-
tively (see Table 3). The plots of the predicted against
actual pIC50 values for the training and test sets are
shown in Figure 3a.
The residuals from the prediction of the test set
compounds by the Conf-d and Conf-s CoMFA models are
shown in Table 5. The Conf-d models predicted 11 of
the 12 test compounds within (1 unit of the actual pIC50
values, whereas the Conf-s was able to predict all 12
test compounds within (1 pIC50 unit. Comparing the
two conformational sets, the Conf-s again performed
better in predicting the activity of the test set. This
superiority is also reflected in the q2 and r2 values. The
CoMFA PLS contour maps for strand transfer inhibitory
activity (data not shown) suggest structure-activity
relationships similar to those of 3′-processing CoMFA
contour maps.
The CoMFA PLS stdev*coefficient contour maps for
the 3′-processing inhibitory activity, are shown in Figure
4A,B. In both figures, a yellow contour near the 4-posi-
tion of the common aromatic ring indicates that bulky
substituents at this position decrease activity. This is
reflected in compounds 31, 35, 37, and 39-42, which
have bulky substituents at this position and have pIC50
CoMSIA Stu d ies Usin g th e In h ibition of 3′-
P r ocessin g Rea ction . CoMSIA PLS analysis of the 66
compounds using the 3′-processing inhibitory activity
as a response variable afforded models with q2 values
of 0.594 (Conf-d) and 0. 0.658 (Conf-s), and r2 values of

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 389
Ta ble 1. Structures, Activities, and Docking Energies of Compounds in the Training Set

390 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Ta ble 1 (Continued)
Kuo et al.

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 391
Ta ble 1 (Continued)
a
These compounds are diaminopyridine salts. ST: Strand Transfer.

392 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Kuo et al.
Ta ble 2. Structures, Activities, and Docking Energies of Compounds in the Training Seta
a
ST: strand transfer.
0.917 and 0.909 for Conf-d and Conf-s sets, respectively.
Only two and three compounds were removed as outliers
in the respective conformational sets. Figure 2c,d depicts
the fitted prediction curves of the training set and
predictions of the test set by the Conf-d and Conf-s
CoMSIA models, respectively. The differences in q2
values of the models are reflected in their ability to
predict the activities of the test set compounds (Table
5). As shown in Table 5, the CoMSIA model of the
Conf-d set predicted nine of the 12 test compounds
within (1 pIC50 units of the actual pIC50 values,
whereas the CoMSIA model of the Conf-s set predicted
11 of the compounds with a similar precision. The
Conf-d CoMSIA model for 3′-processing is better than
the corresponding CoMFA model, but that is not the
case with the Conf-s models in which the CoMFA model
outperforms the corresponding CoMSIA model.
NH in Figure 4C,D show that a H-bond donor group in
this area enhances potency. There is a H-bond acceptor
magenta contour around the sulfonyl oxygens and
mercapto sulfur consistent with the docking. The other
salient feature about the CoMSIA contour maps is the
indication of a significant hydrophobic contribution
(white contour) at the R1 substituent and the edge of
the common substituted phenyl ring around the chlorine
substituent. This reflects the significant contribution of
the hydrophobic descriptor to the CoMSIA QSAR mod-
els, which is the single most important contributor (see
Table 3). The preponderance of the steric fields in the
CoMFA models may also reflect this hydrophobic con-
tribution, which may be implicit in steric bulk allowed
regions.
CoMSIA Usin g th e In h ibition of Str a n d Tr a n sfer
Rea ction . PLS analysis of the CoMSIA descriptors
using the strand transfer inhibitory activity gave q2
values of 0.637 and 0.719 for the Conf-d and for the
Conf-s sets, respectively. The r2 values are 0.923 for the
Conf-d and 0.913 for Conf-s. The residuals in Table 5
The CoMSIA hydrophobic and H-bonding descriptors
PLS contour maps for the Conf-d and Conf-s conforma-
tional sets are shown in Figure 4, panels C and D,
respectively. The cyan contours near the sulfonamide

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 393
F igu r e 2. CoMFA and CoMSIA predictions for the training (b) and test (O) sets for 3′ processing. The fitted curves for CoMFA
models using Conf-d and Conf-s sets are shown in panels a and b, respectively, whereas the corresponding curves for the CoMSIA
models are shown in panels c and d, respectively. The solid line is the regression line for the training set predictions, whereas the
dotted lines indicate the ( 1.0 log point error margins.
Ta ble 3. CoMFA and CoMSIA PLS Statistics
CoMFA
CoMSIA
PLS statistics
q2
3′ processing
Conf _d
0.557
0.372
0.921
0.157
102.61
6
strand transfer
3′ processing
strand transfer
Conf_s
0.630
0.331
0.858
0.205
67.66
4
Conf_d
Conf _s
0.679
0.257
0.892
0.149
72.69
4
Conf_d Conf_s
Conf_d
Conf_s
0.719
0.240
0.913
0.134
93.104
4
0.596
0.291
0.790
0.210
68.77
4
0.594
0.348
0.917
0.158
104.74
2
0.658
0.318
0.909
0.164
92.97
3
0.637
0.282
0.923
0.131
107.69
2
PRESS
r2
s
F
no. of outliers
PLS comp
contribution:
steric
electrostatic
hydrophobic
H-bond donor
H-bond acceptor
6
5
3
6
6
6
6
6
0.658
0.342
0.569
0.431
0.610
0.390
0.582
0.418
0.158
0.204
0.260
0.145
0.199
0.162
0.307
0.266
0.266
0.175
0.175
0.215
0.239
0.160
0.211
0.148
0.133
0.265
0.207
0.247
Ta ble 4. Group Validation and Randomization Exercises
set
exercise
average PLS statistics
3′-processing
strand transfer
CoMFA
CoMSIA
CoMFA
CoMSIA
Conf_d
Conf_s
group validation
randomization
group validation
randomization
q2 (std)
q2
0.557 (0.030)
-0.179
0.625 (0.016)
-0.223
0.569 (0.033)
-0.276
0.639 (0.028)
-0.216
0.595 (0.018)
-0.156
0.671 (0.033)
-0.173
0.628 (0.021)
-0.145
0.688 (0.031)
-0.173
q2 (std)
q2
show that the Conf-d set performed a little better at
predicting the activities of the external test set in which
11 of the 12 compounds were predicted within (1 pIC50
unit of the actual activity. The Conf-s CoMSIA model
predicted 10 compounds within the same boundaries
(Figure 3d). The CoMSIA contour maps for both the
Conf-s and Conf-d (data not shown) were similar to the
CoMSIA contour maps of the models generated using
3′-processing inhibitory activity as a dependent variable.
Dock in g Stu d ies. All the molecules in a database
containing both the training and test sets were docked
onto the monomeric unit A of the crystal structure of
HIV-1 IN catalytic core (PDB 1QS4). Among the 30
possible docking structures of compound 1 generated by

394 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Kuo et al.
F igu r e 3. CoMFA and CoMSIA predictions for the training (b) and test (O) sets for strand transfer activity. The fitted curves
for CoMFA models using Conf-d and Conf-s sets are shown in panels a and b, respectively, whereas the corresponding curves for
the CoMSIA models are shown in panels c and d, respectively. The solid line is the regression line for the training set predictions
whereas the dotted lines indicate the ( 1.0 log point error margins.
Ta ble 5. Residuals of the Prediction of the Test Set by CoMFA and CoMSIA Models
residuals
residuals
compd
3′-Proc. PIC50
CoMFA
Conf_d Conf_s
CoMSIA
Conf_d
-0.170
-0.116
-0.365
-0.553
0.203
strand transfer PIC50
CoMFA
Conf_d Conf_s
CoMSIA
Conf_d
-0.183
-0.230
-0.139
-0.520
0.093
Conf_s
-0.293
-0.285
-0.557
-0.733
-0.321
-0.672
-0.818
0.695
Conf_s
-0.050
-0.122
-0.049
-0.403
-0.076
-0.512
-0.346
1.007
67
68
69
70
71
72
73
74
75
76
77
78
4.39
4.44
4.22
4.05
4.42
4.12
3.70
5.05
4.34
5.40
3.38
4.36
-0.489
-0.477
-0.694
-0.595
-0.345
-0.648
-1.339
0.559
-0.084
-0.055
-0.316
-0.439
-0.200
-0.535
-0.739
0.584
4.40
4.36
4.47
4.15
4.40
4.00
3.95
5.10
4.30
5.40
3.38
4.36
-0.271
-0.321
-0.204
-0.489
-0.233
-0.627
-0.734
0.549
0.307
0.197
0.328
-0.141
0.089
-0.287
-0.180
0.986
-0.043
0.889
-0.125
-0.735
1.294
-0.136
0.810
-0.325
-0.653
0.854
-0.202
0.781
-0.544
0.499
-0.060
0.824
-0.174
0.780
-0.300
0.778
-0.091
0.885
-1.174
-0.462
-0.845
0.094
-1.065
-0.131
-1.32
-0.401
-1.232
-0.301
-0.813
0.355
-1.261
-0.214
-1.158
-0.158
the FlexX docking program, the conformation that was
indicated as the most tightly bound by FlexX scoring,
overlapped with the cocrystallized inhibitor. Figure 5a
shows the cocrystallized compound in magenta and the
highest ranked conformation of compound 1 in ball-and-
stick rendering. This conformation of compound 1 was
used to build the rest of the molecules in the Conf-d set.
Figure 5b shows the CoMFA steric contour plot of the
Conf-d set for the 3′-processing inhibitory activity
projected onto the Connolly surface of the active site of
HIV-1 IN. As shown in this figure, the steric plot agrees
well with the topology of the IN active site, showing
yellow contours in regions of the active site with
restriction against bulky substituents and green con-
tours in regions of the active site that will accommodate
further substitution on the molecule.
The FlexX docking energy of the compounds in the
training set did not show a general correlation with
their 3′-processing or strand transfer inhibitory activi-
ties (Tables 1 and 2). Examination of the FlexX docked
structures of the compounds in the training set also
showed that most of the molecules did not have a
common binding mode. The wide range of binding modes
adopted by the molecules, as well as the lack of general
correlation between the docking energy and inhibitory
activities, may be due to the large structural differences
existing among the compounds.
On the other hand, most of the compounds in the test
set bound in a similar mode due to their close structural
similarity. Compounds 63-66 from the training set,
which are structurally similar to the compounds in the
test set, also adopted a binding mode similar to that of

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 395
F igu r e 4. Contour maps of CoMFA and CoMSIA 3D-quantitative structure-3′-end processing activity relationships. (A) Conf-d
CoMFA contour map; green contours indicate regions where a bulky substituent increases potency, yellow shows areas where
bulky substituents decreases potency. (B) Conf-s CoMFA contour maps; color designation are same as the Conf-s contour map.
(C) Conf-d CoMSIA hydrophobic, hydrogen bond donor, and hydrogen bond acceptor contour maps; white shows areas where
hydrophobic groups enhance potency, whereas yellow indicates that such groups decrease potency. Cyan indicates areas where
hydrogen bond donor groups increase potency, whereas orange indicates the opposite. Areas where a hydrogen bond acceptor
moiety favors activity are indicated by magenta, while areas where a hydrogen bond acceptor group decreases activity are shown
by red. (D) Conf-s CoMSIA hydrophobic, hydrogen bond donor, and hydrogen bond acceptor contour maps; color designations are
the same as in panel C.
q ) quartet, m ) multiplet. The results of elemental analyses
for C, H, and N were within (0.4% of the theoretical values.
the test set compounds. Interestingly, the FlexX energy
scores of these compounds from the training set and
those of the test set compounds correlated well with
their 3′-processing and strand transfer inhibitory activi-
ties (Figure 6). This is noteworthy since the general
experience has been that the available docking scoring
functions perform poorly in predicting experimental
binding affinities. It indicates that FlexX docking may
be useful in optimizing these compounds.
Gen er a l P r oced u r e for th e P r ep a r a tion of 4-Ch lor o-
2-m er ca p to-5-m eth yl-N-(3-a r ylqu in oxa lin -2-yl)ben zen e-
su lfon a m id es 60-62. A stirred solution of O-phenylenedi-
amine (0.45 g, 0.0042mol) and the appropriate arolylbenzodi-
thiazine A-1, A-2, or A-3 (0.004 mol) in anhydrous toluene (35
mL) was refluxed for 9 h. Then, toluene (15 mL) was evapo-
rated under reduced pressure and the solid that precipitated
upon cooling to room temperature was filtered off and washed
with toluene (3 × 2 mL).
Exp er im en ta l Section
In this manner the following sulfonamides were obtained:
4-Ch lor o-[3-(4-ch lor op h en yl)q u in oxa lin -2-yl]-2-m er -
ca p to-5-m eth ylben zen e-su lfon a m id e (60). Starting from
6-chloro-3-(4-chlorobenzoyl)-7-methyl-1,1-dioxo-1,4,2-benzodithi-
azine (1.55 g) the compound 60 was obtained (1.6 g, 84%): mp
183-185 °C dec; IR (KBr) 3235, 3165 (NH), 2520 (SH), 1615
(CdN), 1345, 1150 (SO2) cm-1; 1H NMR (CDCl3) δ 2.37 (s, 3H,
CH3), 4.64 (br. s, 1H, SH), 7.39-7.69 (m, 6H, arom), 7.94-
8.06 (m, 4H, arom), 12.18 (br.s, 1H, NH) ppm; 13C NMR
(CDCl3) δ 20.00, 117.20, 125.65, 127.09, 128.79, 130, 36,
131.41, 131.49, 131.66, 131.92, 132.14, 133.96, 134.22, 137.27,
139.35, 141.40 ppm. Anal. (C21H15Cl2N3O2S2) C, H, N.
Ch em istr y. The following substrates: 3-aroyl-1,1-dioxo-
1,4,2-benzodithiazines A-1-A-322 and 3-amino-7-carbamoyl-
6-chloro-1,1-dioxo-1,4,2-benzodithiazines B-1-B-623 and B-7-
B-15.23,24 as well as final products: 30,25 44,26 45,27 46,27 4727
and 55-5928 were obtained according to the described proce-
dures.
Melting points were determined using a Buchi SMP 20
apparatus and are uncorrected. IR spectra in KBr were
recorded on a Perkin-Elmer 1600 FTIR spectrophotometer. 1H
and 13C NMR spectra were recorded on a Varian Gemini 200
at 200 and 50 MHz, respectively; chemical shifts are reported
in parts per million (ppm) relative to TMS. The following
abbreviations are used to describe peaks patterns when
appropriate: b ) broad, s ) singlet, d ) doublet, t ) triplet,
N-[3-(4-Br om op h en yl)qu in oxa lin -2-yl]-4-ch lor o-2-m er -
ca p to-5-m eth ylben zen e-su lfon a m id e (61). Starting from
3-(bromobenzoyl)-6-chloro-7-methyl-1,1-dioxo-1,4,2-benzodithi-

396 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
Kuo et al.
6-chloro-7-methyl-3-(3-nitrobenzoyl)-1,1-dioxo-1,4,2-benzodithi-
azine (1.59 g), the compound 62 was obtained (1.7 g, 87%):
mp 211-212 °C dec: IR (KBr) 3220, 3155 (NH), 2525 (SH),
1610 (CdN), 1350, 1130 (SO2) cm-1; 1H NMR (DMSO-d6) δ 2.30
(s, 1H, SH), 2.33 (s, 3H, CH3), 7.55-7.63 (m, 1H, arom), 7.71-
7.79 (m, 3H, arom.), 7.96-8.06 (m, 3H, arom), 8.30-8.36 (m,
2H, arom), 8.72 (t, J ) 1.9 Hz, 1H, H-2, PhNO2), 11.70 (br.s,
1H, NH) ppm; 13C NMR (DMSO-d6) δ 19.12, 124.68, 124.83,
127.02, 129.28, 129.73, 130.85, 131.34, 131.92, 132.43, 133.03,
136.21, 136.44, 137.02, 137.36, 137.54, 144.53, 147.46 ppm.
Anal. (C21H15ClN4O4S2) C, H, N.
Gen er a l P r oced u r e for th e P r ep a r a tion of 5-Ca r ba m -
oyl-4-ch lor o-2-m er ca p t o-N-(4H -1,2,4-t r ia zol-3-yl)b en ze-
n esu lfon a m id es 63-66, 75, a n d 77. To a stirred solution of
appropriate B-1-B-6 (5 mmol) in dry DMF (4.6 mL) was added
99% hydrazine hydrate (2.25 g, 45 mmol) under dry nitrogen
and the solution was heated at 100 °C for 15 h. After the
sample was cooled to room temperature, the reaction mixture
was diluted with cold water (75 mL) and acidified to pH 2.5
with 2% hydrochloric acid, and then stirred at room temper-
ature for 0.5 h. The resulting precipitate was separated by
suction, washed with cold water, and dried. The crude product
thus obtained was suspended in water (20 mL) and treated
with 5% aqueous NaOH (1 molar equiv). The suspension was
stirred at room temperature for 20 min, and the insoluble
material was filtered off. The filtrate was immediately acidified
to pH 3.0 with 2% hydrochloric acid, and the solid product that
deposited was collected by filtration, washed with water (3 ×
1.5 mL), and dried at temperatures gradually increased to
90 °C.
In this manner, the following benzenesulfonamides were
obtained:
F igu r e 5. FlexX docked structures. (a) Representative docked
structures of the training and test sets: compound colored
magenta is the cocrystallized HIV-1 integrase inhibitor in the
active site; compound in ball-and-stick rendering is the FlexX
docked structure of compound 1, carbon atoms are colored
cyan, nitrogen atoms blue, sulfur atoms yellow, and oxygen
atoms red; capped stick structures are compounds in the
training set, and most of the test set compounds that showed
common binding mode from FlexX docking. (b) Conf-d CoMFA
steric contour map projected onto Connolly surface of HIV-1
integrase active site.
4-Ch lor o-N-[4-(4-ch lor op h en yl)-4H-1,2,4-tr ia zol-3-yl]-2-
m er ca p to-5-p h en yl-ca r ba m oylben zen esu lfon a m id e (63).
Starting from B-1 (2.39 g, 5 mmol) the compound 63 was
obtained (1.6 g, 61%): mp 145-147 °C dec; IR (KBr) 3295,
3195 (NH), 2549 (SH), 1654 (CdO), 1313, 1140 (SO2) cm-1
;
1H NMR (DMSO-d6) δ 7.05-7.52 (m, 5H), 7.67-7.78 (m, 5H),
7.82 (s, 1H), 8.22 (s, 1H), 8.70 (s, 1H), 10.57 (s, 1H) ppm; 13C
NMR (DMSO-d6) δ 119.53, 123.99, 126.78, 128.38, 128.74,
129.17, 130.92, 131.27, 132.17, 133.42, 134.34, 137.23, 137.52,
138.66, 139.16, 148.22, 162.80 ppm. Anal. (C21H15Cl2N5O3S2)
C, H, N.
4-Ch lor o-2-m er ca p t o-N-[4-(4-m et h ylp h en yl)-4H-1,2,4-
tr iazol-3-yl]-5-ph en yl-car bam oylben zen esu lfon am ide (64).
Starting from B-2 (2.29 g, 5 mmol) the compound 64 was
obtained (1.4 g, 56%): mp 136-138 °C dec; IR (KBr) 3454,
3289 (NH), 2554 (SH), 1663 (CdO), 1340, 1140 (SO2) cm-1
;
1H NMR (DMSO-d6) δ 2.32 (s, 3H), 7.12-7.50 (m, 8H), 7.58-
7.80 (m, 3H), 8.22 (s, 1H), 8.65 (s, 1H), 10.57 (s, 1H) ppm; 13
C
NMR (DMSO-d6) δ 20.52, 119.57, 123.96, 124.85, 128.78,
129.10, 129.63, 129.88, 130.85, 132.17, 132.88, 136.84, 137.26,
137.66, 138.80, 139.40, 148.33, 163.03 ppm. Anal. (C22H18
ClN5O3S2) C, H, N.
-
4-Ch lor o-5-(4-flu or op h e n ylca r b a m oyl)-N -[4-(4-flu o-
r op h e n yl)-4H -1,2,4-t r ia zol-3-yl]-2-m e r ca p t ob e n ze n e -
su lfon a m id e (65). Starting from B-3 (2.4 g, 5 mmol) the title
compound 65 was obtained (1.8 g, 69%): mp 137-141 °C dec;
IR (KBr) 3248, 3142 (NH), 2554 (SH), 1648 (CdO), 1354, 1143
F igu r e 6. Plots of pIC50 vs FlexX docking energy (kcal/mol)
for compounds from the training and the test sets with close
structural similarity; filled circles, solid lines and nonitalicized
coefficient of determination (R2) represent 3′-end processing
data, while open circles, dotted lines, and italicized coefficient
of determination (R2) represent strand transfer data.
1
(SO2) cm-1; H NMR (DMSO-d6) δ 7.13-7.49 (m, 5H), 7.60-
7.88 (m, 5H), 8.22 (s, 1H), 8.67 (s, 1H), 10.62 (s, 1H) ppm; 13
C
NMR (DMSO-d6) δ 113.94, 114.43, 115.81, 120.60, 126.26,
126.76, 128.19, 129.72, 130.76, 131.80, 133.92, 136.14, 136.58,
138.11, 143.0, 147.23, 162.38 ppm. Anal. (C21H14ClF2N5O3S2)
C, H, N.
azine (1.73 g), the compound 61 was obtained (1.8 g, 89%):
mp 172-174 °C dec; IR (KBr) 3230, 3170 (NH), 2545 (SH),
1610 (CdN), 1345, 1130 (SO2) cm-1; 1H NMR (CDCl3) 2.34 (s,
1H, SH), 2.37 (s, 3H, CH3), 7.39 (s, 1H, H-3, PhSO2), 7.42-
7.65 (m, 5H, arom), 7.92-7.98 (m, 4H, arom), 12.10 (s, 1H,
NH) ppm: 13C NMR (CDCl3) δ 20.01, 117.30, 125.77, 127.18,
128.85, 130.36, 131.47, 131.53, 131.63, 131.79, 132.07, 132.16,
134.23, 134.41, 139.42, 145.81 ppm. Anal. (C21H15BrClN3O2S2)
C, H, N.
4-Ch lor o-5-(4-ch lor op h en ylca r b a m oyl)-N-[4-(4-flu o-
r op h e n yl)-4H -1,2,4-t r ia zol-3-yl]-2-m e r ca p t ob e n ze n e -
su lfon a m id e (66). Starting from B-4 (2.48 g, 5 mmol) the
compound 66 was obtained (1.7 g, 63%): mp 197-199 °C dec;
IR (KBr) 3283 (NH), 2543 (SH), 1663 (CdO), 1346, 1307, 1140
1
(SO2) cm-1; H NMR (DMSO-d6) δ 7.25-7.50 (m, 4H), 7.58-
4-Ch lor o-2-m er ca p t o-5-m et h yl-N-[3-(3-n it r ob en zoyl)-
qu in oxa lin -2-yl]ben zen e-su lfon a m id e (62). Starting from
7.83 (m, 6H), 8.20 (s, 1H), 8.65 (s, 1H), 10.67 (s, 1H) ppm; 13
C
NMR (DMSO-d6) δ 114.88, 115.86, 121.07, 125.78, 126.37,

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 397
127.75, 129.82, 130.71, 132.92, 133.18, 135.18, 136.63, 138.11,
138.36, 146.98, 147.48, 162.28 ppm. Anal. (C21H14Cl2FN5O3S2)
C, H, N.
Hz, 2H), 7.65 (s, 1H), 8.30 (s, 1H), 8.65 (s, 1H) ppm; 13C NMR
(CDCl3) δ 13.42, 19.59, 20.91, 30.62, 43.89, 120.36, 129.42,
129.74, 131.60, 131.62, 134.45, 135.06, 135.55, 136.88, 137.63,
138.45, 149.40, 163.42 ppm. Anal. (C20H22ClN5O3S2) C, H, N.
N-(4-Bu tyl-4H-1,2,4-tr ia zol-3-yl)-4-ch lor o-2-m er ca p to-
5-(p h en ylm et h yl-ca r b a m oyl)b en zen esu lfon a m id e (70).
Starting from B-10 (2.19 g, 5 mmol) the compound 70 was
obtained (1.2 g, 50%): mp 103-105 °C dec; IR (KBr) 3325,
4-Ch lor o-N-[4-(4-flu or op h en yl)-4H-1,2,4-tr ia zol-3-yl]-2-
m er ca p to-5-p h en yl-ca r ba m oylben zen esu lfon a m id e (75).
Starting from B-5 (2.3 g, 5 mmol) the compound 75 was
obtained (1.5 g, 59%): mp 220-223 °C dec; IR (KBr) 3283,
3148 (NH), 2543 (SH), 1683 (CdO), 1346, 1140 (SO2) cm-1
;
1H NMR (DMSO-d6) δ 7.13-7.50 (m, 6H), 7.60-7.83 (m, 5H),
8.20 (s, 1H), 8.68 (s, 1H), 10.55 (s, 1H) ppm; 13C NMR (DMSO-
d6) δ 115.47, 116.64, 119.53, 124.10, 127.74, 128.71, 128.95,
130.85, 132.88, 134.20, 134.45, 137.19, 138.66, 139.34, 139.69,
148.36, 162.99 ppm. Anal. (C21H15ClFN5O3S2) C, H, N.
4-Ch lor o-5-(4-ch lor op h en ylca r ba m oyl)-2-m er ca p to-N-
[4-(4-m eth ylph en yl)-4H-1,2,4-tr iazol-3-yl]ben zen esu lfon a-
m id e (76). Starting from B-6 (2.46 g, 5 mmol) the compound
76 was obtained (1.9 g, 71%): mp 138-140 °C dec; IR (KBr)
3301, 3189 (NH), 2537 (SH), 1672 (CdO), 1340, 1143 (SO2)
3148 (NH), 2549 (SH), 1655 (CdO), 1331, 1140 (SO2) cm-1
;
1H NMR (CDCl3) δ 0.87 (t, J ) 7.5 Hz, 3H), 1.08-1.75 (m,
4H), 3.45 (s, 1H), 3.72 (t, J ) 7.5 Hz, 2H), 4.55 (d, J ) 8 Hz,
2H), 6.90 (t, J ) 8 Hz, 1H), 7.27 (s, 5H), 7.35 (s, 1H), 7.57 (s,
1H), 8.35 (s, 1H) ppm; 13C NMR (DMSO-d6) δ 13.45, 19.62,
30.95, 44.06, 44.10, 127.46, 127.78, 128.59, 130.09, 131.06,
131.84, 134.24, 137.30, 137.73, 138.05, 138.55, 149.54, 165.21
ppm. Anal. (C20H22ClN5O3S2) C, H, N.
4-Ch lor o-5-(4-ch lor op h en ylca r ba m oyl)-2-m er ca p to-N-
[4-(2-m eth ylp r op yl)-4H-1,2,4-tr ia zol-3-yl]ben zen esu lfon a -
m id e (71). Starting from B-11 (2.29 g, 5 mmol) the compound
71 was obtained (1.8 g, 72%): mp 130-132 °C dec; IR (KBr)
3372, 3301, 3154 (NH), 2555 (SH), 1663 (CdO), 1307, 1149
(SO2) cm-1; 1H NMR (CDCl3) δ 0.85 (d, J ) 7.5 Hz, 6H), 1.85-
2.20 (m, 1H), 3.55 (d, J ) 7.5 Hz, 2H), 7.20-7.32 (m, 3H),
7.55-7.70 (m, 3H), 8.25 (s, 1H), 8.98 (s, 1H) ppm; 13C NMR
(CDCl3) δ 19.66, 27.94, 51.20, 121.64, 128.85, 129.81, 131.42,
131.85, 134.74, 136.45, 137.02, 137.48, 138.91, 139.20, 148.58,
163.49 ppm. Anal. (C19H19Cl2N5O3S2) C, H, N.
1
cm-1; H NMR (DMSO-d6) δ 2,32 (s, 3H), 7.25-7.57 (m, 6H),
7.68-7.83 (m, 4H), 8.22 (s, 1H), 8.64 (s, 1H), 10.72 (s, 1H) ppm;
13C NMR (DMSO-d6) δ 20.52, 120.30, 121.07, 124.25, 126.76,
127.85, 128.24, 128.74, 129.13, 130.12, 130.83, 131.77, 136.50,
136.65, 137.62, 138.55, 147.43, 162.38 ppm. Anal. (C22H17
Cl2N5O3S2) C, H, N.
-
Gen er a l P r oced u r e for th e P r ep a r a tion of 5-Ca r ba m -
oyl-4-ch lor o-2-m e r ca p t o-N -(4H -1,2,4-t r ia zol-3-yl)b e n -
zen esu lfon a m id es 67-74. To a stirred solution of appro-
priate B-7-B-14 (5 mmol) in dry DMF (4.6 mL) was added
99% hydrazine hydrate (2.25 g, 45 mmol) under dry nitrogen
and the solution was heated at 100 °C for 10-11 h. After the
sample was cooled to room temperature, the reaction mixture
was diluted with cold water (75 mL), acidified to pH 2.5 with
2% hydrochloric acid, and then stirred for 0.5 h at room
temperature. The resulting precipitate was filtered off, washed
with cold water, and dried. The crude reaction product thus
obtained was suspended in water (20 mL), treated with 5%
NaOH (1 molar equiv), and stirred vigorously for 20 min at
room temperature. The insoluble side product was filtered off,
and the filtrate was immediately acidified with 2% hydrochlo-
ric acid to pH 3.0. The pure product thus obtained was collected
by filtration, washed with water (3 × 1.5 mL), and dried at
temperature gradually increasing to 80 °C.
4-Ch lor o-2-m er ca p to-5-(4-m eth oxyp h en ylca r ba m oyl)-
N -[4-(2-m e t h ylp r op yl)-4H -1,2,4-t r ia zol-3-yl]b e n ze n e -
su lfon a m id e (72). Starting from B-12 (2.27 g, 5 mmol) the
compound 72 was obtained (1.5 g, 61%): mp 138-140 °C dec;
IR (KBr) 3365, 3265, 3148 (NH), 2560 (SH), 1681 (CdO), 1299,
1
1138 (SO2) cm-1; H NMR (DMSO-d6) δ 0.82 (d, J ) 7.5 Hz,
6H), 1.88-2.25 (m, 1H), 3.62 (d, J ) 7.5 Hz, 2H), 3.75 (s, 3H),
6.95 (d, J ) 11 Hz, 2H), 7.62 (d, J ) 11 Hz, 2H), 7.80 (s, 1H),
8.20 (s, 1H), 8.40 (s, 1H) ppm; 13C NMR (DMSO-d6) δ 19.23,
27.18, 49.74, 55.13, 113.86, 121.07, 128.76, 130.50, 131.74,
132.16, 132.74, 136.98, 137.77, 139.98, 148.65, 155.64, 163.64
ppm. Anal. (C20H22ClN5O4S2) C, H, N.
4-Ch lor o-2-m er ca p to-5-(4-m eth ylp h en ylca r ba m oyl)-N-
[4-(1-m eth ylp r op yl)-4H-1,2,4-tr ia zol-3-yl]ben zen esu lfon a -
m id e (73). Starting from B-13 (2.19 g, 5 mmol) the compound
73 was obtained (1.9 g, 79%): mp 120-122 °C dec; IR (KBr)
3295, 3195, 3136 (NH), 2543 (SH), 1654 (CdO), 1313, 1137
(SO2) cm-1; 1H NMR (CDCl3) δ 0.80 (t, J ) 7 Hz, 3H), 1.32 (d,
J ) 7 Hz, 3H), 1.50-1.82 (m, 2H), 2.31 (s, 3H), 4.08-4.38 (m,
1H), 7.10 (d, J ) 9 Hz, 2H), 7.45-7.70 (m, 4H), 8.32 (s, 1H),
8.68 (s, 1H) ppm; 13C NMR (CDCl3) δ 10.24, 19.59, 20.91, 28.72,
52.70, 120.32, 129.81, 129.42, 131.66, 131.77, 134.45, 135.09,
In this manner, the following benzenesulfonamides were
obtained:
4-Ch lor o-2-m er ca p to-5-(4-m eth ylp h en ylca r ba m oyl)-N-
(4-p r op yl-4H-1,2,4-tr ia zol-3-yl)ben zen esu lfon a m id e (67).
Starting from B-7 (2.12 g, 5 mmol) the compound 67 was
obtained (1.7 g, 72%): mp 122-124 °C dec; IR (KBr) 3277,
3227, 3142 (NH), 2543 (SH), 1654 (CdO), 1313, 1140 (SO2)
1
cm-1; H NMR (CDCl3) δ 0.85 (t, J ) 7.5 Hz, 3H), 1.45-1.80
136.23, 136.80, 137.73, 149.19, 163.35 ppm. Anal. (C20H22
ClN5O3S2) C, H, N.
-
(m, 2H), 2.30 (s, 3H), 3.50-3.75 (t, J ) 7.5 Hz, 2H), 6.50 (b.s,
1H), 7.07 (d, J ) 8.5 Hz, 2H), 7.45 (d, J ) 8.5 Hz, 2H), 7.65 (s,
1H), 8.28 (s, 1H), 8.68 (s, 1H) ppm; 13C NMR (CDCl3) δ 10.84,
20.91, 22.05, 45.63, 120.35, 129.42, 129.78, 131.77, 131.88,
134.45, 134.70, 135.08, 136.84, 137.59, 138.88, 149.40, 163.42
ppm. Anal. (C19H20ClN5O3S2) C, H, N.
N-(4-Ben zyl-4H-1,2,4-tr ia zol-3-yl)-4-ch lor o-2-m er ca p to-
5-(4-m eth ylp h en yl-ca r ba m oyl)ben zen esu lfon a m id e (74).
Starting from B-14 (2.36 g, 5 mmol) the compound 74 was
obtained (1.9 g, 74%): mp 170-172 °C dec; IR (KBr) 3283,
3142 (NH), 2531 (SH), 1663 (CdO), 1313, 1139 (SO2) cm-1
;
1H NMR (DMSO-d6) δ 2.32 (s, 3H), 5.05 (s, 2H), 7.20 (d, J ) 8
Hz, 2H), 7.40 (s, 6H), 7.65 (d, J ) 8 Hz, 2H), 7.85 (s, 1H), 8.22
(s, 1H), 8.55 (s, 1H), 10.30 (s, 1H) ppm; 13C NMR (DMSO-d6)
δ 20.41, 46.10, 119.57, 127.88, 128.08, 128.63, 129.13, 130.77,
132.24, 132.95, 133.15, 135.19, 137.12, 136.16, 137.73, 139.67,
148.76, 163.28 ppm. Anal. (C23H20ClN5O3S2) C, H, N.
Syn t h esis of N-(4-Bu t yl-4H -1,2,4-t r ia zol-3-yl)-5-ca r -
b a m oyl-4-ch lor o-2-m er ca p t o-b en zn esu lfon a m id e (C-1).
To a stirred solution of B-15 (2.29 g, 5 mmol) in dry DMF (4.6
mL) 99% hydrazine hydrate (1.5 g, 30 mmol) was added under
dry nitrogen and the reaction mixture was heated at 100 °C
for 4 h. After the sample was cooled to room temperature, the
solution was diluted with cold water (75 mL), acidified to pH
2.5 with 2% hydrochloric acid, and then stirred for 0.5 h at
room temperature. The solid that precipitated was collected
by filtration, washed with cold water, and dried. The crude
product thus obtained was purified by crystallization from
glacial acetic acid and dried at 100 °C to afford C-1 (1.7 g,
89%): mp 224-225 °C dec; IR (KBr) 3477, 3454, 3312, 3148
N-(4-Bu tyl-4H-1,2,4-tr iazol-3-yl)-4-ch lor o-5-(4-ch lor oph e-
n ylcar bam oyl)-2-m er captoben zen esu lfon am ide (68). Start-
ing from B-8 (2.29 g, 5 mmol) the title compound 68 was
obtained (1.5 g, 60%): mp 129-132 °C dec; IR (KBr) 3406,
307, 3160 (NH), 2549 (SH), 1657 (CdO), 1310, 1143 (SO2) cm-1
;
1H NMR (CDCl3) δ 0.87 (t, J ) 7.5 Hz, 3H), 1.05-1.45 (m,
2H), 1.50-1.78 (m, 2H), 3.70 (t, J ) 7.5 Hz, 2H), 7.20-7.30
(m, 3H), 7.55-7.68 (m, 3H), 8.27 (s, 1H), 8.90 (s, 1H) ppm; 13
C
NMR (CDCl3) δ 13.41, 19.66, 30.68, 44.10, 121.64, 128.88,
129.70, 131.49, 131.95, 134.77, 136.40, 137.02, 137.50, 138.51,
138.94, 149.43, 163.74 ppm. Anal. (C19H19Cl2N5O3S2) C, H, N.
N-(4-Bu tyl-4H-1,2,4-tr ia zol-3-yl)-4-ch lor o-2-m er ca p to-
5-(4-m eth ylp h en yl-ca r ba m oyl)ben zen esu lfon a m id e (69).
Starting from B-9 (2.19 g, 5 mmol) the compound 69 was
obtained (1.3 g, 55%): mp 107-110 °C dec; IR (KBr) 3277,
3248, 3195 (NH), 2543 (SH), 1648 (CdO), 1313, 1140 (SO2)
1
cm-1; H NMR (CDCl3) δ 0.88 (t, J ) 7.5 Hz, 3H), 1.05-1.40
(m, 2H), 1.48-1.75 (m, 2H), 2.30 (s, 3H), 3.72 (t, J ) 7.5 Hz,
2H), 6.63 (b.s, 1H), 7.08 (d, J ) 8.5 Hz, 2H), 7.50 (d, J ) 8.5

398 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2
(NH2, NH), 2537 (SH), 1666 (CdO), 1316, 1143 (SO2) cm-1
Kuo et al.
;
added, and incubation was continued for an additional 1 h.
Reactions were quenched by an addition of (8 µL) of loading
dye (98% deionized formamide, 10 mM EDTA, 0.025% xylene
cyanol, and 0.025% bromophenol blue). An aliquot (5 µL) was
electrophoresed on a denaturing 20% polyacrylamide gel (0.09
M tris-borate pH 8.3, 2 mM EDTA, 20% acrylamide, 8 M urea).
Gels were dried, exposed on a PhosphorImager cassette, read
on a Typhoon 8610 Variable Mode imager (Amersham Bio-
sciences), and quantitated using ImageQuant 5.2 (Amersham
Biosciences).
1H NMR (DMSO-d6) δ 0.85 (t, J ) 7.5 Hz, 3H), 1.05-1.37 (m,
2H), 1.45-1.85 (m, 2H), 3.77 (t, J ) 7.5 Hz), 7.72 (s, 1H), 7.95
(b.s, 2H), 8.12 (s, 1H), 8.84 (s, 1H), 13.15 (b.s, 1H) ppm; 13C
NMR (DMSO-d6) δ 13.27, 18.94, 30.00, 42.81, 128.77, 132.16,
130.77, 134.73, 136.48, 137.83, 139.87, 148.65, 166.70 ppm.
Anal. (C13H16ClN5O3S2) C, H, N.
Gen er a l P r oced u r e for th e P r ep a r a tion of 4-Ca r ba m -
oyl-5-ch lor o-2-[(4H-1,2,4-tr ia zol-3-yl)a m in osu lfon yl]p h e-
n ylth ioa cetic Acid s 77, 78. A vigorously stirred suspension
of appropriate compound C-1 or 67 (5 mmol) and NaOH (0.2
g, 5 mmol) in water (10 mL) was treated dropwise with a
solution of chloroacetic acid (0.47 g, 5 mmol) and Na2CO3 (0.26
g, 2.5 mmol) in water (12 mL). The reaction mixture was
heated at 40 °C for 1 h, and then the temperature was raised
to 90 °C at the rate of 10 °C/h. The solution was further stirred
at 90 °C for 20 h, and then was left to stand overnight at
ambient temperature. The reaction mixture (pH 6.6) was
neutralized to pH 7.2-7.8 with aqueous 10% Na2CO3. The side-
product that precipitated was filtered off, and the filtrate was
acidified to pH 2.5 with 2% hydrochloric acid. The crude
product that deposited was collected by filtration, washed with
water, dried, and purified by crystallization from 2-propanol.
In this manner, the following acetic acid derivatives were
obtained:
Molecu la r Mod elin g a n d Align m en t. The structures
were built in SYBYL 6.729 on a Silicon Graphics Octane
(R12000) workstation. Two different conformational templates
were used in the QSAR studies. These were derived by (i)
FlexX docking of compound 1 or (ii) performing a systematic
conformational search in SYBYL. Compound 1 was docked
onto the active site of HIV-1 IN using the FlexX docking
program interfaced with SYBYL. Among the 30 conformations
and binding modes generated, the highest scored conformation
by FlexX scoring function was used as a template to build all
the molecules in one conformational set, which is referred to
as “Conf-d”. The other conformational template was obtained
by systematic conformational search of compound 14. In the
conformational search, all the rotatable bonds (a-d, Figure
1) in compound 14 were varied by 10°. The lowest energy
conformation identified in this conformational search was used
as a template to build the molecules in the second conforma-
tional set, which is referred to as “Conf-s”. The molecules in
each conformational set constructed using the respective
templates were superimposed using the multifit function in
SYBYL and energy minimized using the Powell conjugate
gradient methods.29 The common structure highlighted in
compound 1 (Figure 1) that includes the six aromatic carbon,
sulfur, two oxygen, and nitrogen atoms were used for aligning
the molecules. Charges were calculated using the semiem-
perical method AM1 in MOPAC 6.0,30 interfaced with SYBYL.
CoMF A a n d CoMSIA Stu d ies. The CoMFA descriptors,
steric (Lennard-J ones 6-12 potential) and electrostatic (Cou-
lombic potential) field energies, were calculated using the
SYBYL default parameters: 2 Å grid points spacing, an sp3
carbon probe atom with +1 charge and a van der Waals radius
of 1.52 Å, and energy cutoff of 30 kcal/mol.
2-[(4-Bu tyl-4H-1,2,4-tr ia zol-3-yl)a m in osu lfon yl]-4-ca r -
ba m oyl-5-ch lor op h en yl-th ioa cetic a cid (77). Starting from
C-1 (1.95 g, 5 mmol) the compound 77 was obtained (1.4 g,
63%): mp 114-116 °C dec; IR (KBr) 3430-2850 (OH), 3412,
3307, 3237 (NH2, NH), 1733 (CdO), 1639 (CdO), 1319, 1137
(SO2) cm-1 1H NMR (DMSO-d6) δ 0.83 (t, J ) 7.5 Hz, 3H),
;
1.0-1.35 (m, 2H), 1.45-1,72 (m, 2H), 3.82 (t, J ) 7.5 Hz, 2H),
3.95 (s, 2H), 7.45 (s, 1H), 7.68 (b.s, 1H), 7.99 (b.s, 1H), 8.1 (s,
1H), 8.37 (s, 1H) ppm; 13C NMR (DMSO-d6) δ 13.27, 18.95,
29.97, 34.15, 42.53, 127.13, 128.71, 131.99, 133.16, 138.16,
139,50, 139.91, 148.72, 166.78, 169.88 ppm. Anal. (C15H18
ClN5O5S2) C, H, N.
-
5-Ch lor o-4-(4-m et h ylp h en ylca r b a m oyl)-2-(4-p r op yl-
4H-1,2,4-tr iazol-3-yl)am in osu lfon yl)ph en ylth ioacetic acid
(78). Starting from 67 (2.33 g, 5 mmol) the compound 78 was
obtained (1.7 g, 66%): mp 232-234 °C dec; IR (KBr) 3315-
2830 (OH), 3307, 3184, 3154 (NH), 1710 (CdO), 1669 (CdO),
1310, 1137 (SO2) cm-1; 1H NMR (DMSO-d6) δ 0.80 (t, J ) 7.5
Hz, 3H), 1.51-1-85 (m, 2H), 2.28 (s, 3H), 3.70 (t, J ) 7.5 Hz,
2H), 4.05 (s, 2H), 7.16 (d, J ) 9.5 Hz, 2H), 7.55 (s, 1H), 7,60
(d, J ) 9.5 Hz, 2H), 8.20 (s, 1H), 8.38 (s, 1H), 10.53 (s, 1H)
ppm; 13C NMR (DMSO-d6) δ 10.49, 20.41, 21.34, 34.08, 44.60,
119,57, 127.03, 129.09, 131.90, 132.06, 132.95, 133.34, 136.16,
The five similarity indices in CoMSIA, i.e., steric, electro-
static, hydrophobic, H-bond donor, and H-bond acceptor de-
scriptors, were calculated31 using a C1+ probe atom with a
radius of 1.0 Å placed at regular grid spacing of 2 Å. CoMSIA
similarity indices (AF) for a molecule j with atoms i at a grid
point q are calculated by eq 1:
138.23, 139.77, 148.79, 163.35, 169.92 ppm. Anal. (C21H22
ClN5O5S2) C, H, N.
-
2
AF,kq(j) ) -Σωprobe,kωike-Rr
(1)
iq
Ch em ica ls. For IN assays, all compounds were dissolved
in DMSO, and all aliquots were also made in DMSO prior to
each experiment. The stock solutions were kept at -20 °C.
P r ep a r a tion of Oligon u cleotid e Su bstr a tes. The oligo-
nucleotides 21top, 5′-GTGTGGAAAATCTCTAGCAGT-3′ and
21bot, 5′-ACTGCTAGAGATTTTCCACAC-3′ were purchased
from Norris Cancer Center Core Facility (University of South-
ern California) and purified by UV shadowing on polyacryla-
mide gel as described. To analyze the extent of 3′-processing
and strand transfer using 5′-end labeled substrates, 21top was
5′-end labeled using T4 polynucleotide kinase (Epicenter,
Madison, WI) and γ[32P]-ATP (Amersham Biosciences). The
kinase was heat-inactivated and 21bot was added in 1.5-molar
excess. The mixture was heated at 95 °C, allowed to cool slowly
to room temperature, and run through a spin 25 mini-column
(USA Scientific) to separate annealed double-stranded oligo-
nucleotide from unincorporated material.
In tegr a se Assa y. To determine the extent of 3′-processing
and strand transfer, HIV-1 IN was preincubated at a final
concentration of 200 nM with the inhibitor in reaction buffer
(50 mM NaCl, 1 mM HEPES, pH 7.5, 50 µM EDTA, 50 µM
dithiothreitol, 10% glycerol (w/v), 7.5 mM MnCl2, 0.1 mg/mL
bovine serum albumin, 10 mM 2-mercaptoethanol, 10% DMSO,
and 25 mM MOPS, pH 7.2) at 30 °C for 30 min. Then, 20 nM
of the 5′-end 32P-labeled linear oligonucleotide substrate was
where k represents the following physicochemical properties:
steric, electrostatic, hydrophobic, H-bond donor, and H-bond
acceptor. A Gaussian type distance dependence was used
between the grid point q and each atom i of the molecule. The
default value of 0.3 was used as the attenuation factor (R).
Here, steric indices are related to the third power of the atomic
radii, electrostatic descriptors are derived from atomic partial
charges, hydrophobic fields are derived from atom-based
parameters,32 and H-bond donor and acceptor indices are
obtained by a rule-based method based on experimental
results.33
The CoMFA and CoMSIA descriptors derived above were
used as explanatory variables, and pIC50 (-log IC50) values
were used as the target variable in PLS regression analyses
to derive 3D QSAR models using the implementation in the
SYBYL package. The predictive value of the models was
evaluated by leave-one-out (LOO) cross-validation. The cross-
validated coefficient, q2, was calculated using eq 2:
2
(Ypredicted - Yobserved
)
∑
∑
q2 ) 1 -
(2)
2
(Yobserved - Ymean
)
where Ypred, Yactual, and Ymean are predicted, actual, and mean

Mercaptosalicylhydrazide Inhibitors of HIV-1 Integrase
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 2 399
(14) Brzozowski, Z. 2-Mercapto-N-(azolyl)benzenesulphonamides. I.
Synthesis of N-(1,1-dioxo-1,4,2-benzodithiazin-3-yl)guanidines
and their transformations into 2-mercapto-N-(5-amino-1,2,4-
triazol-3-yl) benzenesulphonamide derivatives with potential
anti-HIV or anticancer activity. Acta Pol. Pharm. 1995, 52, 91-
101.
(15) Brzozowski, Z. 2-Mercapto-N-(azolyl)benzenesulphonamides. II.
Syntheses, anti-HIV-1 and anticancer activity of some S-sub-
stituted 4-chloro-2-mercapto-5-methyl-N-(5-amino-1,2,4-triazol-
3-yl) benzenesulphonamides. Acta Pol Pharm. 1995, 52, 287-
292.
(16) Brzozowski, Z. 2-mercapto N-(azolyl)benzenesulfonamides. VI.
Synthesis and anti-HIV activity of some new 2-mercapto-N-
(1,2,4-triazol-3-yl)benzenesulfonamide derivatives containing the
1,2,4-triazole moiety fused with a variety of heteroaromatic
rings. Acta Pol. Pharm. 1998, 55, 473-480.
(17) 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. Inhibitors of strand transfer that prevent
integration and inhibit HI-1 replication in cells. Science. 2000,
287, 646-650.
(18) Pais, G. C. G.; Burke, T. R. Novel aryl diketo-containing
inhibitors of HIV-1 integrase. Drugs Future 2002, 27, 1101-
1111.
(19) Wold, S.; Rhue, A.; Wold, H.; Dunn, W. J . I. The covariance
problem in linear regression. The partial least squares (PLS)
approach to generalized inverses. SAIM J . Sci. Stat. Comput.
1994, 5, 735-743.
(20) Wold, S.; Albano, C.; Dunn, W. J ., III; Edlund, U.; Esbensen,
K.; Geladi, P.; Hellberg, S.; J ohanson, E.; Lindberg, W.; Sjostrom,
M. Multivariate data analysis in chemistry. NATO ASI Ser., Ser.
C 1984, 138, 17-95.
(21) Clark, M.; Cramer, I., R. D. The probability of chance correlation
using partial least squares (PLS). Quant. Struct-Act Relat. 1993,
12, 137-145.
(22) Brzozowski, Z.; Saczewski, F.; Kuo, C.-L.; Sanchez, T.; Gdaniec,
M.; Neamati, N., 2003, unpublished data.
(23) Brzozowski, Z.; Slawinski, J .; Angielski, S.; Szczepanska-Konkiel,
M. [Derivatives of 1,1-dioxo-1,4,2-benzodithiazine. III. Synthesis
and diuretic properties of 3-(R,R1-phenyl)amino-6-chloro-7-
methyl-1,1-dioxo-1,4,2-benzodithiazines]. Acta Pol. Pharm. 1985,
42, 313-318.
(24) Brzozowski, Z. Syntheses of some new 4-chloro-2-mercapto-N-
(4H-1,2,4-triazol-3-yl)benzenesulfonamides with potential anti-
cancer or anti-HIV activity. Acta Pol. Pharm. 1997, 54, 461-
466.
(25) Slawinski, J . Synthesis, anti-HIV and anticancer activity of some
S-substituted 4-chloro-2-mercapto-5-methyl-N-(5-amino-1,2,4-
triazol-3-yl)benzenesulfonamides. Acta Pol. Pharm. 1995, 52,
287-292.
(26) Brzozowski, Z. Synthesis of N-(1,1-dioxo-1,4,2-benzodithiazin-
3-yl)guanidines and their transformation into 2-mercapto-N-(5-
amino-1,2,4-triazol-3-yl)benzenesulfonamide derivatives with
potential anti-HIV or anticancer activity. Acta Pol. Pharm. 1998,
55, 91-101.
(27) Brzozowski, Z. Synthesis and anti-HIV activity of some new
2-mercato-N-(1,2,4-triazol-3-yl)benzenesulfonamide derivatives
containing the 1,2,4-triazole moiety fused with a variety of
heteroaromatic rings. Acta Pol. Pharm. 1998, 55, 473-480.
(28) Brzozowski, Z.; Saczewski, F.; Gdaniec, M. Synthesis, structural
characterization and in vitro anticancer activity of 4- dimethyl-
aminopyridinium (6-chloro-1,1-dioxo-1,4,2-benzodithiazin-3-yl)-
methanides. Bioorg Med Chem. 2003, in press.
values of the target property (pIC50), respectively. ∑(Ypred
Y
-
actual)2 is the predictive sum of squares (PRESS). To maintain
the optimum number of PLS components and minimize over
determination, the number of components giving the lowest
PRESS value was used for deriving the final PLS regression
models. Conventional correlation coefficient r2 and its standard
error, s, were also computed for the final PLS models. CoMFA
and CoMSIA coefficient maps were generated by interpolation
of the pairwise products between the PLS coefficients and the
standard deviations of the corresponding CoMFA or CoMSIA
descriptor values.
F lexX Dock in g. The FlexX program34 version 1.9 interfaced
with SYBYL 6.7 was used to dock the compounds onto the
active site of HIV-1 IN. FlexX is a fast automated docking
program that considers ligand conformational flexibility by an
incremental fragment placing technique.34,35 We used the
program to dock the training set as well as the test set
molecules into the active site of monomeric unit “A” of the
crystal structure of the HIV-1 IN catalytic core. The 3D
coordinates of the HIV-1 IN catalytic core in complex with an
inhibitor were taken from the Brookhaven Protein Databank
(PDB code: 1QS4).9 The missing residues at positions 141-
144 were inserted by means of the loop search algorithm in
the SYBYL BIOPOLYMER module. The active site for docking
was defined as all atoms within 6.5 Å radius of the cocrystal-
lized ligand. The default SYBYL FlexX parameters were used.
Ack n ow led gm en t. We would like to thank the Drug
Synthesis and Chemistry Branch, Developmental Thera-
peutics of the National Cancer Institute, for the anti-
viral testing. The work in N.N.’s lab was supported by
funds from the GlaxoSmithKline Drug Discovery Award.
Refer en ces
(1) Brown, P. O. Integration; Cold Spring Harbor Press: Cold Spring
Harbor, NY, 1999.
(2) Asante-Appiah, E.; Skalka, A. M. HIV-1 integrase: structural
organization, conformational changes, and catalysis. Adv. Virus
Res. 1999, 52, 351-369.
(3) Neamati, N. Patented small molecule inhibitors of HIV-1 inte-
grase: a ten-year saga. Expert Opin. Ther. Pat. 2002, 12, 709-
724.
(4) Dayam, R.; Neamati, N. Small-Molecule HIV-1 Integrase Inhibi-
tors: the 2001-2002 Update. Curr. Pharm. Des. 2003, 9, 1789-
1802.
(5) Neamati, N.; Barchi, J . J ., J r. New paradigms in drug design
and discovery. Curr. Top. Med. Chem. 2002, 2, 211-227.
(6) Neamati, N.; Mazumder, A.; Sunder, S.; Owen, J . M.; Schultz,
R. J .; Pommier, Y. 2-Mercaptobenzenesulfonamides as novel
class of human immunodeficiency type 1 (HIV-1) integrase and
HIV-1 replication. Antimicrob. Agents Chemother. 1997, 8, 485-
495.
(7) Nicklaus, M. C.; Neamati, N.; Hong, H.; Mazumder, A.; Sunder,
S.; Chen, J .; Milne, G. W.; Pommier, Y. HIV-1 integrase
pharmacophore: discovery of inhibitors through three- dimen-
sional database searching. J . Med. Chem. 1997, 40, 920-929.
(8) Lubkowski, J .; Yang, F.; Alexandratos, J .; Wlodawer, A.; Zhao,
H.; Burke, T. R., J r.; Neamati, N.; Pommier, Y.; Merkel, G.;
Skalka, A. M. Structure of the catalytic domain of avian sarcoma
virus integrase with a bound HIV-1 integrase-targeted inhibitor.
Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 4831-4836.
(9) Neamati, N.; Mazumder, A.; Zhao, H.; Sunder, S.; Burke, T. R.,
J r.; Schultz, R. J .; Pommier, Y. Diarylsulfones, a novel class of
human immunodeficiency virus type 1 integrase inhibitors.
Antimicrob. Agents Chemother. 1997, 41, 385-393.
(10) Chen, I. J .; Neamati, N.; Nicklaus, M. C.; Orr, A.; Anderson, L.;
Barchi, J . J ., J r.; Kelley, J . A.; Pommier, Y.; MacKerell, A. D.,
J r. Identification of HIV-1 integrase inhibitors via three-
dimensional database searching using ASV and HIV-1 inte-
grases as targets. Bioorg. Med. Chem. 2000, 8, 2385-2398.
(11) Neamati, N.; Lin, Z.; Karki, R. G.; Orr, A.; Cowansage, K.;
Strumberg, D.; Pais, G. C.; Voigt, J . H.; Nicklaus, M. C.;
Winslow, H. E.; Zhao, H.; Turpin, J . A.; Yi, J .; Skalka, A. M.;
Burke, T. R., J r.; Pommier, Y. Metal-dependent inhibition of
HIV-1 integrase. J . Med. Chem. 2002, 45, 5661-5670.
(12) Pannecouque, C.; Pluymers, W.; Van Maele, B.; Tetz, V.;
Cherepanov, P.; De Clercq, E.; Witvrouw, M.; Debyser, Z. New
class of HIV integrase inhibitors that block viral replication in
cell culture. Curr. Biol. 2002, 12, 1169-1177.
(29) SYBYL Version 6.7; Tripos Associates: St. Louis, MO.
(30) Stewart, J . J . MOPAC:
a semiempirical molecular orbital
program. J . Comput.-Aided Mol. Des. 1990, 4, 1-105.
(31) Klebe, G.; Abraham, U.; Mietzner, T. Molecular similarity indices
in
a comparative analysis (CoMSIA) of drug molecules to
correlate and predict their biological activity. J . Med. Chem.
1994, 37, 4130-4146.
(32) Viswanadhan, V. N.; Ghose, A. K.; Revenkar, G. R.; Robins, R.
Atomic and physichochemical parameters for three-dimensional
structure-directed quantitative structure-activity relationships.
4. additional parameters for hydrophobic and dispersive interac-
tions and their application for an automated superposition of
certain naturally occurring antibiotics. J . Chem. Inf. Comput.
Sci. 1989, 29, 163-172.
(33) Klebe, G. The use of composite crystal-field environments in
molecular recognition and the de novo design of protein ligands.
J . Mol. Biol. 1994, 237, 212-235.
(34) Kramer, B.; Rarey, M.; Lengauer, T. Evaluation of the FLEXX
incremental construction algorithm for protein-ligand docking.
Proteins 1999, 37, 228-241.
(35) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A fast flexible
docking method using an incremental construction algorithm.
J . Mol. Biol. 1996, 261, 470-489.
(13) Scozzafava, A.; Owa, T.; Mastrolorenzo, A.; Supuran, C. T.
Anticancer and antiviral sulfonamides. Curr. Med. Chem. 2003,
10, 925-953.
J M030378I