Triketoacid inhibitors of HIV-integrase: A new chemotype useful for probing the integrase pharmacophore

Article abstract of DOI:10.1016/j.bmcl.2006.03.010

Integrase is one of three enzymes expressed by HIV and represents a validated target for therapy. This study reports on the discovery of a new triketoacid-based chemotype that selectively inhibits the strand transfer reaction of HIV-integrase. SAR studies showed that the template binds to integrase in a manner similar to the diketoacid-based inhibitors. Moreover, comparison of the new chemotype to two different diketoacid templates led us to propose two aryl-binding domains in the inhibitor binding site. This information was used to design a new diketoacid template with improved activity against the enzyme.

Full text of DOI:10.1016/j.bmcl.2006.03.010

Bioorganic & Medicinal Chemistry Letters 16 (2006) 2920–2924
Triketoacid inhibitors of HIV-integrase: A new chemotype
useful for probing the integrase pharmacophore
Michael A. Walker,a,* Timothy Johnson,a Zhuping Ma,a Jacques Banville,c
Roger Remillard,c Oak Kim,a Yunhui Zhang,a Andrew Staab,a Henry Wong,a
Albert Torri,b Himadri Samanta,b Zeyu Lin,b Carol Deminie,b Brian Terry,b
Mark Krystalb and Nicholas Meanwella
aDepartment of Medicinal Chemistry, Pharmaceutical Research Institute, Bristol-Myers Squibb,
The Richard L Gelb Center for Pharmaceutical Research and Development, 5 Research Parkway Wallingford, CT 06492, USA
bDepartment of Virology, Pharmaceutical Research Institute, Bristol-Myers Squibb,
The Richard L Gelb Center for Pharmaceutical Research and Development, 5 Research Parkway Wallingford, CT 06492, USA
cDepartment of Medicinal Chemistry, Pharmaceutical Research Institute, 100 de l’Industrie Boulevard, Que., Canada J5R1J1
Received 12 January 2006; revised 28 February 2006; accepted 1 March 2006
Available online 20 March 2006
Abstract—Integrase is one of three enzymes expressed by HIV and represents a validated target for therapy. This study reports on
the discovery of a new triketoacid-based chemotype that selectively inhibits the strand transfer reaction of HIV-integrase. SAR stud-
ies showed that the template binds to integrase in a manner similar to the diketoacid-based inhibitors. Moreover, comparison of the
new chemotype to two different diketoacid templates led us to propose two aryl-binding domains in the inhibitor binding site. This
information was used to design a new diketoacid template with improved activity against the enzyme.
Ó 2006 Elsevier Ltd. All rights reserved.
HIV-integrase is the viral enzyme responsible for insert-
ing a DNA-copy of the virus into the host chromosome.
Without this occurring, the virus cannot replicate, mak-
ing integrase an attractive target for the development
of novel anti-HIV agents.1 However, the discovery of
suitable drug candidates targeting HIV-integrase has
proven to be a formidable task. A number of drug dis-
covery campaigns have been launched over the last 15
years, and it is only recently that compounds suitable
for development have been described.2
phosphate ester cleavage involving the 30-hydroxyl
group of the vDNA. According to this mechanism, the
30-hydroxyl group of the incoming vDNA attacks the
phosphate ester of the host DNA-strand to form a pen-
tavalent intermediate.3 Two Mg+2 ions facilitate the flow
of negative charge from the deprotonated 30-hydroxyl of
the viral strand to that of the leaving 30-hydroxyl of the
host strand. This mechanism is similar to that invoked
for structurally related enzymes such as RNAse H,
Escherichia coli exonuclease, MuA-transposase, Tn5
transposase, and the RuvC-Holliday junction resol-
vase.4 An analogous pentacovalent phosphorus interme-
diate has been observed in the X-ray structure of
b-phosphoglucomutase.5
HIV-integrase catalyzes two reactions. In the first, re-
ferred to as 30-processing, the enzyme cleaves two base
pairs from the 30-ends of the transcribed viral DNA
(vDNA), leaving 30-CA dinucleotides at each end. The
second reaction occurs in the nucleus, wherein the 30-
ends of the vDNA are inserted into the host DNA. This
step is referred to as strand transfer and is currently the
preferred target for the development of inhibitors. It is
believed that this reaction occurs via a Mg2+-mediated
Cl
O
O
O
O
OH
1
IC50 = 1 μM
EC50 = 120 μM
Recent efforts from our labs have uncovered an authen-
tic lead exemplified by compound 1 bearing a triketoacid
structure.6 Compound 1 is a moderately potent inhibitor
Keywords: AIDS; HIV-integrase; Diketoacid; Triketoacid.
*
Corresponding author. Tel.: +1 203 677 6686; e-mail: walkerma@
bms.com
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.03.010

M. A. Walker et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2920–2924
2921
of the isolated enzyme (1 lM)7 and of HIV-1 in cell cul-
ture. Additionally, the compound is able to inhibit inte-
grase in the context of the pre-integration complex (PIC)
(IC50 = 0.6 lM) isolated from infected cells. In counter
screens, it did not inhibit other viral or DNA-processing
enzymes (HIV-RT, HIV-protease, RNase H, BVDV-
polymerase, T3-polymerase, and Klenow fragment)
and was devoid of activity against other viruses (BVDV,
RSV, Flu, and Rhinovirus). Time-of-addition experi-
ments showed that antiviral activity coincided with the
viral integration step.
also exhibited decreased sensitivity to 1 (not shown).
Thus, the results clearly show that the triketoacid inhib-
itor 1 targets integrase.
Published studies have concluded that the diketoacid
inhibitors bind to the two catalytic Mg2+ atoms located
in the active site, as depicted in Figure 1.9 In contrast to
L-731,988, compound 1 offers the potential for two dif-
ferent modes of binding to Mg2+, illustrated by model A
and model B in Figure 1. A brief SAR investigation was
conducted in an effort to resolve this ambiguity. L-
731,988 is very sensitive to modifications to the Mg2+
-
In order to confirm that integrase is the target, multiple
populations of the virus were maintained in the presence
of 1. Serial passage of virus in the presence of increasing
amounts of inhibitor yielded two isolates with reduced
susceptibility, R1 and R2, after 47 passages. As shown
in Table 1, the two resistant isolates exhibited a >3-fold
decrease in susceptibility to compound. Sequence analy-
sis indicated two unique sets of mutations for each resis-
tant strain (R1 = T66I and L172F, R2 = V151A and
V165I), located in the integrase enzyme. These residues
are found in the core domain of the enzyme and are
close to the active site catalytic DDE triad (Asp64,
Asp116, and Glu152). In subsequent experiments, the
T66I mutation was introduced into the HIV-NL4-3
(HIV-1) strain which resulted in resistance to 1 in cell
culture. Similar results have been reported for the dike-
toacid-based inhibitor, L-731,988, where the T66I,
S153Y, and M154L amino acid changes were observed
to confer resistance.8 Recombinant bacterially expressed
integrase containing the T66I and M154I substitutions
binding elements, as shown for compound 2, where
the C-terminal carboxylic acid has been replaced by a
tetrazole. Even though the tetrazole moiety is well estab-
lished as a carboxylic acid mimic, in this setting activity
is severely compromised. The triketoacid 1 behaves in a
similar fashion, since compound 3 is essentially inactive.
However, the activity of 1 is only moderately affected by
removal of the d-carbonyl, since compound 4 is only an
order of magnitude weaker. Therefore, since the SAR
for the Mg2+-binding domain of 1 matches that of L-
731988, it is likely that the triketoacid binds to integrase
in the manner depicted by model B.
Next, the SAR related to the phenyl ring was evaluated
and the results of that survey are shown in Figure 2.
Using the unsubstituted parent 5 as a reference point,
it is clear that the introduction of an electron-withdraw-
ing group at the ortho- (e.g., 6–9) or meta- (e.g., 10) po-
sition enhances in vitro activity. The phenyl ring can be
replaced by pyridine, but its activity is sensitive to regio-
chemistry with the more exposed 3-isomer 12 weaker
than the 2-isomer 11. The improved activity observed
for 1 and 6–9, compared to 11 and 12, suggests that a
hydrophobic binding site exists in the region between
the ortho- and meta-sites. Compound 13, wherein this
region is occupied by a fused phenyl ring, was designed
to test this concept. In accordance with the hypothesis,
13 showed a noticeable improvement in inhibitory activ-
ity. In contrast, compounds 14 and 15, which also have
Table 1. Activity of 1 against wild-type (HIV-1) and resistant virus
strain
EC50 (lM)
HIV-NL3-4
HIV-R1
HIV-R2
55
>200
>200
>200
HIV-NL3-4 (T66I)
Mg+2
F
Mg+2
O-
O
O
O
O
O
O
H
N
O
N
N
OH
O
2
N
N
binding model
IC50 > 125 μM
L-731,988
IC50 = 0.3 μM
Mg+2
O-
Mg+2
O
Mg+2
O
Cl
O
O
O
Mg+2
O-
O
O
O
O
O
OH
OH
O
1
model A
model B
Cl
O
O
Cl
O
O
O
H
OH
N
N
O
N
N
3
4
IC50 = 125 μM
IC50 = 9 μM
Figure 1. Binding models for L-731,988 and compound 1. In vitro inhibitory activities (reference 7) for each compound are indicated.

2922
M. A. Walker et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2920–2924
O
O
O
O
OH
5
IC50 = 80 μM
X
O
O
O
Cl
N
O
N
OH
10
11
12
IC50 = 8 μM
IC50 = 100 μM IC50 > 300 μM
IC50 (μM)
X
1
6
7
8
9
Cl
1
3
6
4
7
F
CF3
OCH3
OCH2Ph
13
14
15
IC50 = 0.6 μM
IC50 = 6 μM
IC50 = 2 μM
Figure 2. SAR around the aryl group of the triketoacid chemotype.
an additional phenyl ring, do not demonstrate increased
potency, indicating that the improved activity seen with
13 is not simply due to non-specific, hydrophobic bind-
ing. This suggests a specific aryl-binding domain exists
within the integrase enzyme that is capable of distin-
guishing subtle differences in structure.
The SAR survey was further expanded in order to map
the binding site occupied by the aryl portion of the
triketoacid template. It was assumed that the fully eno-
lized form of the triketoacid was more consistent with
the SAR than the keto form. As such, there are two pos-
sible configurations for the enol bond, Z and E. In order
to help delineate this issue, diketoacids L-731,988 and 16
were used as model templates. Since L-731,988 was
specifically designed to take full advantage of the inter-
action of the aryl group with integrase, it represents a
compelling template for the purpose of comparison.10
Compound 16 is also very potent and, as such, should
be useful to compare to the triketoacid template
Figure 3.
Figure 4. Overlay of L-731,988 and 16 (A). Overlay of compound Z-51
and 16 (B). Overlay of E-13 and L-731,988 (C). Halogen atoms are not
shown.
two yields a misalignment of the aryl groups.11 This
result can be interpreted as that there are two aryl-bind-
ing domains available in the active site of integrase.
When the triketoacid template is compared to com-
pound 16, the Z-enol configuration was found to over-
lay better than the E. As seen in Figure 4, although
the alignment of the aryl groups of 16 and 1 is not opti-
mal, the ortho chloro group of the phenyl ring can pro-
ject toward the fused phenyl ring of the indole, allowing
it to occupy this region. This provides an explanation
for the activity enhancing effect of the ortho-substituent.
In contrast, when L-731,988 is used as a model, the E-
enol provides a better overlay. Figure 4 shows an over-
lay of 13, the most active triketoacid, and L-731,998,
and here the aryl groups exhibit very good alignment.
In addition, it is expected that the ortho or meta substit-
uents of 1 and 6–10 would occupy this region. Although
model compounds 16 and L-731,988 predict different
binding geometries for the enol bond of the triketoacid,
both allow reasonable interpretation of the SAR pre-
sented in Figure 2. More significantly, assuming two dif-
ferent aryl-binding domains exist in the enzyme active
site, it is apparent that the triketoacid is able to access
either one with a simple change of enol configuration.
Despite the fact that L-731,988 and 16 are equipotent,
they differ in the positioning of their aryl groups, with
respect to the diketoacid template. This is illustrated in
Figure 4, where it is observed that an overlay of the
OH
OH
O
OH
OH
O
OH
HO
O
5
5
O
E
Z
Cl
O
OH
F
O
OH
OH
OH
N
O
HN
O
16
IC50 = 0.4 μM
L-731.988
Figure 3. Possible enol configurations for 5.

M. A. Walker et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2920–2924
Table 2. In vitro activity of diketoacid analogues
OH
2923
different assay than that used to assess the potency of
compounds 4, 17, and 18, preventing a direct compari-
son.12 However, under the modified assay conditions,
compounds 1, L-731,988, and 16 had IC50s = 0.2, 0.02,
and 0.08 lM, respectively, indicating that 19 is indeed
potent when compared to other diketoacid-based
inhibitors.
O
OH
R
O
Compound
R
IC50 (lM)
17
30
In summary, we disclose the basic SAR associated with
a new class of triketoacid-based inhibitors of HIV-inte-
grase. Preliminary SAR studies indicate that this chem-
otype binds to integrase in a manner similar to the
diketoacid-based inhibitors. Preferred substitution on
the phenyl group was also studied and it was discov-
ered that an appropriately positioned naphthyl group
yielded optimal activity. Comparison of the triketoac-
ids with two different diketoacid-based inhibitors pro-
vided some insight into the SAR and helped to
establish a model for the binding conformation. The
data are consistent with dual aryl-binding domains
within the integrase protein. Using this information,
we were able to modify the diketoacid template by
removing the carbonyl group adjacent to the aryl
group, which led to the discovery of 19 as a potent
inhibitor of integrase.
4
26
Cl
18
19
5
*
Br
Cl
Cl
0.01a
*
a Assay protocol different to that utilized for 4, 17, and 18 – see
reference 12.
The final phase of the survey was to re-examine the effect
of removing the enol group connecting the phenyl ring
to the triketoacid moiety. Since the saturated derivative
4 is an effective inhibitor of integrase, compounds 17–19,
were synthesized and evaluated in vitro, as summarized
in Table 2. As anticipated, this modification was well
tolerated and aryl ring substitution resulted in increased
potency. While the SAR trends are similar to that seen
with the enol-containing series, there are some subtle
differences. For example, an ortho Cl substituent is less
effective than an ortho Br. The 3,4-dichloro analogue
19 is highly active, but it should be noted that the data
reported for this compound were determined using a
Diketoacid-based inhibitors have proven to be very use-
ful as a starting point for the design of clinically verified
agents, thus the discovery of 1 represents a useful lead
toward the goal of developing compounds for the treat-
ment of HIV infection. The objective of the current
investigation was to gain a better understanding of the
tri/diketoacid inhibitor pharmacophore, in order to
develop a more ‘drug like’ template. Toward that end,
the current study provides information regarding the
binding of the aryl domain of the diketoacid-based
3 eq. tBuOK,
10 eq. CH3CO2CH2CH3
THF, 24 h
4 eq. tBuOK,
1 eq. CH3O2CCO2CH3
THF
O
O
O
O
O
OH
O
Ar
Ar
Ar
R
O
22
23
24 (R = CH3)
1, 6-16 (R = H)
O
O
O
O
NaCl
DMSO 170 oC
O
K2CO3, PhCH3
reflux
Ar
Ar
Ar
Br
25
27
LDA
CO2Et
26
CH3O2CCO2CH3
THF
O
OH
O
R
Ar
O
28 (R = CH3)
4, 19-21 (R = H)
Figure 5. Synthesis of tri- and diketoacids.

2924
M. A. Walker et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2920–2924
formed, compounds at desired concentrations were added
to the wells followed by a 10 min incubation at 37 °C.
Biotinylated target DNA was then added and the reaction
was carried out for an additional 1 h at 37 °C. Wells were
washed thoroughly to remove any free DNA and integra-
tion activity was measured by using a commercial kit to
quantitate the amount of biotinylated target DNA inte-
grated into the substrate.
inhibitors. Future publications will demonstrate how
this information can be used to introduce structural
modifications to this portion of the template.
The syntheses of the compounds described in this letter
are illustrated in Figure 5. The triketoacids were pre-
pared using a modified version of the method reported
by Stiles, which is simply a series of two Claisen conden-
sation reactions starting from the methyl-arylketone,
ethylacetate, and dimethyl oxalate.13 Compounds 4,
and 17–19, were also synthesized using Claisen method-
ology as demonstrated.
8. 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.
9. 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.
10. Wai, J. S.; Egbertson, M. S.; Payne, L. S., et al. J. Med.
Chem. 2000, 43, 4923.
11. The overlaid structures were generated in the following
manner: the equilibrium conformations of compounds
were determined in Spartan (Wavefunction Inc.) using
molecular mechanics (MMFF94 forcefield). The structures
were then manually overlaid using the alignment feature in
DSViewerPro (Accelrys Inc.).
12. To evaluate in vitro activity against HIV-integrase, 5 pmol
of biotin labeled substrate DNA was bound to 100 lg of
Streptavidin coated PVT SPA beads (Amersham Pharma-
cia Biotech). Recombinant integrase (0.26 ng) was incu-
bated with the beads for 90 min at 37 °C. Unbound
enzyme was removed by washing the complex followed by
addition of inhibitors and 0.1 fmol of P33 labeled target
DNA. The reaction was stopped by adding EDTA to a
final concentration of 40 mM. Samples were counted in
TopCountNXT (Packard) and the CPM was used as a
measure of integration. The reaction conditions were as
described in A. Engelman and R. Craigie, J. Virol. 1995,
69, 5908–5911. The sequences of substrate and target
DNA were described in Nucleic Acid Research 1994, 22,
1121–1122.
References and notes
1. LaFemina, R. L.; Schneider, C. L.; Robbins, H. L.;
Callahan, P. L.; LeGrow, K.; Roth, E.; Schleif, W. A.;
Emini, E. A. J. Virol. 1992, 66, 7414.
2. Pommier, Y.; Johnson, A. A.; Marchand, C. Nat. Rev.
Drug Disc. 2005, 4, 236.
3. Engelman, A.; Mizuuchi, K.; Craigie, R. Cell 1991, 67, 1211.
4. (a) Dyda, F.; Hickman, A. B.; Jenkins, T. M.; Engelman,
A.; Craigie, R.; Davies, D. R. Science 1994, 266, 1981; (b)
Lovell, S.; Goryshin, I. Y.; Reznikoff, W. R.; Rayment, I.
Nat. Struct. Biol. 2002, 9, 278.
5. Lahiri, S. D.; Zhang, G.; Dunaway-Mariano, D.; Allen,
K. N. Science 2003, 299, 2067.
6. (a) Walker, M. A.; Johnson, T. D.; Kim, O. K.; Zhang, Y.
US-6548546, April 15, 2003; Triketoacid integrase inhib-
itors have recently been described by; (b) Sechi, M.;
Sannia, L.; Carta, F.; Palomba, M.; Dallocchio, R.; Dessi,
A.; Derudas, M.; Zawahir, Z.; Neamati, N. Antiviral
Chem. Chemother. 2005, 16, 41.
7. In vitro activity was measured in the following manner
which is similar to previously disclosed methods. Purified
recombinant HIV-1 integrase was incubated with immo-
bilized precleaved substrate DNA in a 96-well plate for
20 min at 37 °C. After the integration complex was
13. Stiles, M.; Selegue, J. P. J. Org. Chem. 1991, 56, 4067.