Discovery of potent pyrrolidone-based HIV-1 protease inhibitors with enhanced drug-like properties

Article abstract of DOI:10.1016/j.bmcl.2004.08.039

SAR around the HIV-1 protease inhibitor lead 4, which utilizes the (3S,5R)-3,5-bis(phenylmethyl)-2-pyrrolidinone as P1-P2 scaffold, led to discovery of meta-amino- and meta-hydroxy inhibitors 17b and 19b, which while equipotent to 4, in addition offered a

Full text of DOI:10.1016/j.bmcl.2004.08.039

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5689–5692
Discovery of potent pyrrolidone-based HIV-1 protease
inhibitors with enhanced drug-like properties
Wieslaw M. Kazmierski,* Webb Andrews, Eric Furfine,
Andrew Spaltenstein and Lois Wright
GlaxoSmithKline Research, 5 Moore Drive, Research Triangle Park, NC 27709, USA
Received 29 June 2004; revised 17 August 2004; accepted 17 August 2004
Available online 15 September 2004
Abstract—We have developed efficient syntheses of the HIV-1 protease inhibitor 4 and its analogues, which incorporate the pyrrol-
idone scaffold 2 as P1–P2 moiety. Evaluation of these analogues in the HIV-1 protease enzyme assay resulted in discovery of potent
and more water soluble meta-amino- and meta-hydroxy inhibitors 17b and 19b. The SAR observed in this class of PIs could be
rationalized with aid of the X-ray structure of inhibitor 28 co-crystallized with the HIV-1 protease, which suggested that the polar
meta- (but not para-) benzyl substituents in P2 could side-step the hydrophobic S2 enzyme active pocket by rotating the P2 moiety
around its Cb–Cc bond. Such reorientation allows to engage the unsubstituted, hydrophobic edge of benzyl moiety in P2 in the
requisite P2/S2 hydrophobic interaction, and projects polar meta-substituent into the bound water. It appears that the meta-position
can be chemically derivatized without potency loss of thus resulting inhibitors, as evidenced by potent 22–26. We thus identified
pyrrolidone 2-based inhibitors exemplified by 17b and 19b, which uniquely accommodate both high enzyme potency and which pro-
vide a platform for fine-tuning of drug-like properties in this class of PIs by additional chemical manipulations on the meta-position.
Ó 2004 Elsevier Ltd. All rights reserved.
Recent publications from several laboratories including
ours disclosed novel and potent HIV-1 protease inhibi-
tors incorporating various P1–P2 scaffolds suchas
1,2,5-thiadiazolidine 1,1-dioxide, 2-imidazolidinone, iso-
thiazolidine 1,1-dioxide, 5S-(phenylmethyl)-3-morpholi-
none, and 2-pyrrolidone.1–5 The key driver for that work
has been to discover potent inhibitors with improved
drug-like properties, especially aqueous solubility, which
could help to alleviate some of the pill burden issues
encountered with the standard HAART therapy.1
Our initial targets included the picolyl derivatives 10–12
(Fig. 1), which were predicted to be more potent than
analogous P2-picolyl derivatives in the 2-imidazolidi-
none scaffold 1 series.6 These derivatives were obtained
by the condensation of 3- and 4-pyridinecarboxalde-
hydes with pyrrolidone 5,5 followed by a spontaneous
dehydration and subsequent loss of the N-Boc-protect-
ing group, resulting in respective derivatives 6 and 7
(Fig. 2).
Hydrogenation of 6 and 7 withstoichiometric H 2 (1atm,
PdÆC/CaCO3 catalyst) exclusively yielded cis products 8
Our previous letter described the synthesis of PIs of the
type 3 incorporating the 2-imidazolinone scaffold 1 as
P1–P2 moiety.1 One of the conclusions reached in that
work was that the trans relationship of both the P1
and P2 substituents was required in this class of PIs
for high potency against the HIV-PR. We thus redi-
rected our medicinal chemistry efforts into PIs utilizing
2-pyrrolidone 2 as P1–P2 scaffold, exemplified by lead
inhibitor 4, which features trans P1 and P2.
1
3
Y=H
Y=A Ki =1.2 nM
A
P1'
N
O
N
Y
OH
OH
H
N
2
4
Y=H
Y=A Ki=0.067 nM
O
P2'
N
Y
Keywords: HIV-1 protease inhibitor; Aspartyl protease inhibitor;
AIDS; Peptide mimetic; Peptidomimetic.
O
*
Corresponding author. Tel.: +1 919 483 9462; fax: +1 919 483
6053; e-mail: wieslaw.m.kazmierski@gsk.com
Figure 1. 2-Imidazolidinone 1 and pyrrolidone 2 P1–P2 scaffolds and
corresponding PIs 3 and 4.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.08.039

5690
W. M. Kazmierski et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5689–5692
Bn
methyl)-3-morpholinone and the 5R-(phenylmethyl)-2-
pyrrolidinone (2) scaffold series and emphasizes a neces-
sity to optimize SAR in eachseries independently.
Bn
Bn
NH
a
b
NH
NH
R
O
R
O
O
We optimized P2-benzyl motif even further by alkylat-
ing 52,3 withvarious benzyl bromides X-Bn-Br, which
generally proceeded witha high trans diastereoselectiv-
ity (ratio of 3S,5R to 3R,5R diastereomers ꢀ4:1). Alkyl-
ating agents withmasked functionalities, suchas
m-methoxy-benzyl bromide, m- and p-nitro-benzyl bro-
mide were used towards syntheses of 19a, 17a, and
18a, respectively. The resulting intermediates were then
progressed as outlined in Figure 4, yielding pyrrolidones
15a–20a. These derivatives were then coupled to epoxide
13 under either equilibrating or non-equilibrating condi-
tions,5 followed by the acetonide deprotection.9 When
feasible, mixtures of diastereomers were then chromato-
graphically enriched to >95% optical purity, yielding PIs
15b, 17b, 19b, and 20b (Fig. 4, if diastereomers were sep-
arated, data is shown only for the more potent 3S,5R
(trans) diastereomer).5 Couplings of 16a and 18a to
epoxide 13 were performed under equilibrating condi-
tions, yielding equimolar mixtures of bothdiastereomers
16b and 18b (Fig. 4), which were evaluated as such in the
HIV-1 protease assay.10
R: 3-py 8
4-py 9
R: 3-py 6
4-py 7
5
c, d
Bn
13
O
O
N
O
N
A
Py
O
Ki [nM]
0.17
14.0
0.31
10 3-py diastereomer 1
11 3-py diastereomer 2
12 4-py diastereomeric mixture
Figure 2. Reagents and conditions: (a) LDA, THF ꢁ78°C, 3- and 4-
pyridinecarboxaldehydes; 73–87% (b) 5% PdÆC, CaCO3, H2 titration;
99%; (c) NaH/DMF 80°C, epoxide 13; (d) 4N HCl in dioxane/water
(95:5, v/v), silica purification, 31–69%.
and 9.7 Since our objective was to synthesize the trans
P1–P2 derivatives, cis intermediates were coupled to
epoxide 13 under conditions known to epimerize the 3-
position in 2-pyrrolidone (NaH/DMF, 80°C).5 Products
were then chromatographically separated into the indi-
vidual diastereomers 10 (Ki = 0.17nM) and 11
(Ki = 14nM) (Fig. 2). The 4-picolyl analogue 12 was
synthesized in a similar fashion yielding equimolar mix-
ture of diastereomers, which was assayed as such and
proved to also be potent in the HIV-1 protease assay
(Ki = 0.31nM).
Enzyme inhibition data for 15b–20b strongly suggests
that meta-substituents in 15b, 17b, and 19b yielded
inhibitors, which were practically equipotent to lead
molecule 4, with the exception of methyl ester derivative
20b. On the other hand, para-substituents in 16b and 18b
caused a significant decrease in potency against HIV-1
protease (Fig. 4), far beyond when their Ki values are
adjusted for 50% diastereomeric contents. We then at-
tempted to expand the SAR by directly derivatizing
these PIs. Thus, the oxidation of 15b withurea–hydro-
gen peroxide (UHP) provided potent carboxamide 21
(Ki = 0.1nM), while the alkylation of 19b withbromo-
acetonitrile resulted in an approximately equipotent
inhibitor 22 (Ki = 0.08nM), oxidation of which also
yielded potent carboxamide 23 (Ki = 0.09nM) (Fig. 5).
We also explored aliphatic P2-substituents in the context
of pyrrolidone P1–P2 scaffold 2. The rationale for this
was based on SAR established in another scaffold series;
5S-(phenylmethyl)-3-morpholinone, in which the P2-al-
lyl substituted analogue was found to be more potent
than the P2-benzyl-substituted analogue.8 Following
this lead, the synthesis of the P2-allyl pyrrolidone-based
inhibitor 14 was then accomplished by a diastereoselec-
tive alkylation of lactam 5 withallyl bromide and Boc-
deprotection, followed by coupling of epoxide 13 and
acetonide deprotection (Fig. 3).5
R
a, b
Bn
15a m-CN
16a p-CN
Bn
NH
a, b, f
17a m-NH2
18a p-NH2
R
O
Perhaps surprising in light of prior SAR, the allyl-P2
inhibitor 14 (Fig. 3, Ki = 0.45nM) was found less potent
than the benzyl-P2 lead 4 (Ki = 0.067nM).5 This finding
underscores different SAR in bothteh 5 S-(phenyl-
N
a, b, c
a, b
19a m-OH
Boc
4:1 trans/cis
O
20a m-COOMe
d, e, g
5
R
Ki [nM]
Bn
A
15b m-CN
16b p-CN
17b m-NH2
diastereomer 1
0.05
14
1:1 diast. mixture 0.34
N
diastereomer 1
1:1 diast. mixture
diastereomer 1
0.10
2.30
0.05
0.42
a, b, c, d
R
5
18b p-NH2
19b m-OH
O
OH
OH
H
N
N
20b m-COOMe diastereomer 1
O
O
Figure 4. Reagents and conditions: (a) LHMDS ꢁ78°C, R-Bn-Br, 79–
93%; (b) TFA; (c) BBr3; (d) P4-phosphazine 1equiv, THF, epoxide 13,
ꢁ78 to 0°C or NaH, DMF, 80°C; (e) 4N HCl in dioxane/water (95:5,
v/v), 56%–87% for (b)–(e); (f) PdÆC/H2; (g) silica gel chromatography,
51–73% for (f)–(g).
>95:5 trans/cis
Figure 3. Reagents and conditions: (a) LHMDS ꢁ78°C, allyl bromide,
78%; (b) TFA; (c) P4-phosphazine base, 1equiv, THF, epoxide 13, ꢁ78
to 0°C; (d) 4N HCl in dioxane/water (95:5, v/v), 81% for (b)–(d).

W. M. Kazmierski et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5689–5692
5691
Capitalizing on the finding that even quite large meta-
substituents were apparently compatible withbinding
to the S2 enzyme pocket, we then attempted to incorpo-
rate additional solubility modifiers into P2 by trans-
forming ester 20b to glycol- and morpholine-amides 24
and 25, bothof whichwere found to be subnanomolar
inhibitors. In addition, amine 17b could be converted
to urea-based inhibitors 26 and 27, bothessentially equi-
potent to 17b (Fig. 5).
Importantly, several potent analogues appeared to offer
the potential for improved aqueous solubility. This was
confirmed by the use of well-established computational
methods, which indeed predicted marked improvements
in aqueous solubilities of 10, 17b, 19b, and 21 as com-
pared to lead 4.11 Moreover, increased solubility of these
compounds was not only compatible withtheir potent
inhibition of the HIV–PR enzyme, but also the inhibi-
tion of live HIV-1 virus in MT4 cells (0.32, 0.21, 0.32,
0.20, and 0.14lM for 4, 10, 17b, 19b, and 21,
respectively).
Figure 6. (a) The structure of inhibitor 28; (b) close-up of the X-ray
structure of 28 co-crystallized with HIV-PR, which shows P2 in 28
interacting withS2 and bound water (yellow).
the Cb–Cc bond. This positions the meta-substituent
toward bound water at the mouth of the S2 site
(Fig. 6b). Importantly, this model explains the relative
lack of influence of the substituent size on inhibitorꢀs
Ki value. Thus, inhibitors with small polar meta-substit-
uents, suchas 17b and 19b, as well as withlarger substit-
uents suchas 26, are essentially equipotent (Figs. 4 and
5). Somewhat more hydrophobic meta-ester 20b is too
big to snugly fit the S2 subsite, but the solvation forces
driving polar substituents (as in 19b) away from the S2
by rotating P2 around the Cb–Cc bond, are absent in
20b. Therefore, we further speculate that it is the steric
clashbetween bulky m-COOCH3 in 20b and the S2 site,
which results in approximately 8-fold loss of potency
(Ki = 0.42nM) compared to 19b. The X-ray structure
also suggests that para-substituents would be positioned
in a way which results in their head-on collision with the
S2 subsite (Fig. 6b), which cannot be avoided by rotat-
ing P2 around the Cb–Cc axis. Indeed, para-16b and 18b
are less potent than their meta-analogues 15b and 17b,
respectively.
The key finding of this work is that of a dual role as-
sumed by meta-substituents in scaffold 2, which can
simultaneously support high potency of inhibitors and
accommodate polar, solubilizing groups. This outcome
is unique and in contrast to our experience withthe
other scaffolds synthesized and examined in the course
of this work,1–5 for which attempts to introduce solubi-
lizing groups into variety of position of the scaffold
invariably resulted in a decrease of potency in the
HIV-1 protease assay.10 Of all the scaffolds exam-
ined,1–5 only pyrrolidone 2 eventually allowed us to dis-
cover a specific meta-site, which supported both high
enzyme potency and improved solubility, when polar
substituents are used.
We next attempted to rationalize these unexpected find-
ings with the aid of the X-ray structure of a co-crystal of
HIV-1 protease and of inhibitor 28 (Ki = 0.05nM) (Fig.
6a). This structure revealed that the hydrophobic edge
of the P2-benzyl makes the critical hydrophobic interac-
tion with the S2 enzyme pocket. To stay solvated, the
relatively hydrophilic P2-meta-carboxamide benzyl in
28 sidesteps the hydrophobic S2 by rotating 180° around
OH
O
P2
O
H2N
OH
OH
H
N
N
O
O
28
Bn
A
Bn
A
a
O
N
b
N
15b
O
R
19b
In summary, polar meta-substituents in the (3S,5R)-3,5-
bis(phenylmethyl)-2-pyrrolidinone 2, exemplified by
inhibitors 17b and 19b, are unique in their ability to sup-
port bothhighpotency against HIV-1 protease and an
improved aqueous solubility. The relative lack of sensi-
tivity of Ki to various polar meta-substituents in the
(3S,5R)-3,5-bis(phenylmethyl)-2-pyrrolidinone P1–P2
moiety should allow further fine-tuning of drug-like
properties in this series by chemical modifications of
the meta-position. This opens the opportunity to dis-
cover other potent and soluble inhibitors, potentially
with other improved drug-like properties, which could
lower the pill burden in HAART. Work from our labo-
ratories to that effect will be disclosed in due course.
H2N
O
O
O
21
22 R= -CH2-CN
a
23 R= -CH2-CO-NH2
H
N
Bn
A
R
N
H
N
HO
d
c
O
O
O
R
N
20b
17b
Ki=0.15 nM
24 Ki =0.24 nM
25
O
O
Bn
A
O
Et
R
Et
N
H
N
N
R
NH
H
O
O
27 Ki=0.2 nM
26 Ki=0.1 nM
Figure 5. Reagents and conditions: (a) UHP, pH = 10, 2h, 87–93%; (b)
Br-CH2CN, 79%; (c) isocyanate, 87–92%; (d) neat amine, 67–79%.

5692
W. M. Kazmierski et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5689–5692
Similarly synthesized benzyl morpholinone-based diaster-
Acknowledgements
eomeric mixture B was muchless potent ( Ki = 30nM),
suggesting preference of allyl-P2 over benzyl-P2 in the
morpholino series.
We thank Richard Hazen for generating the MT4 data.
9. Deprotection of N-, O-acetonide C withTFA gives
substantial amounts of the cyclization cleavage products,
lacton D and indanolamine E. Additional experimentation
allowed us to dramatically suppress the formation of D
and E by the use of 4N HCl in dioxane (95%) and water
References and notes
1. Previous manuscript in this series: Kazmierski, W. M.;
Furfine, E.; Gray-Nunez, Y.; Spaltenstein, A.; Wright, L.
Bioorg. Med. Chem. Lett. 2004, 14, in press. doi:10.1016/
j.bmcl.2004.08.038.
2. Salituro, F. G.; Baker, C. T.; Court, J. J.; Deininger, D.
D.; Kim, E. E.; Li, B.; Novak, P. M.; Rao, B. G.;
Pazhanisamy, S.; Porter, M. D.; Schairer, W. C.; Tung, R.
D. Bioorg. Med. Chem. Lett. 1998, 8, 3637.
3. Kazmierski, W. M.; Furfine, E.; Spaltenstein, A.; Wright,
L. L. Bioorg. Med. Chem. Lett. 2002, 12, 3431.
4. Fritch, P. C.; Kazmierski, W. M. Synthesis 1999, 1, 112.
5. Spaltenstein, A.; Cleary, D.; Bock, B.; Kazmierski, W.;
Furfine, E.; Wright, L.; Hazen, R.; Andrews, W.; Almond,
M.; Tung, R.; Salituro, F. Bioorg. Med. Chem. Lett. 2000,
10, 1159.
6. Several pairs of analogues in both 1 and 2 series suggested
that the trans relationship of P1 and P2 enabled by the
pyrrolidone scaffold 2 improves potency approximately by
a factor of 20.
(5%)
O
Bn
Bn
Bn
O
OH
O
acid
N
Bn
N
N
O
O
O
+
OH
D
H2N
C
E
.
10. Kazmierski, W.; Andrews, W.; Baker, C.; Bock, B.;
Cleary, D.; Furfine, E.; Gray-Nunez, Y.; Hale, M.; Hazen,
R.; Salituro, F.; Schairer, W.; Spaltenstein, A.; Tung, R.;
Wright, L. Book of Abstracts, 219thACS National
Meeting, San Francisco, CA, March26–30, 2000, poster
MEDI-126.
11. We utilized methods described in Klopman, G.; Wang, S.;
Balthasar, D. M. Chem. Inf. Comput. Sci. 1992, 32, 474.
Comparison of relative predicted solubilities for 10, 17b,
19b, and 21 (concentration at saturation 0.526, 0.126,
0.079, and 0.063lM) indicate that these compounds are
predicted to be up to 10-fold more soluble than the
7. PdÆC catalyst and/or higher pressure of H2 result in
overreduction of pyridines to piperidines.
8. The coupling of P1–P2 allyl-morpholinone scaffold to
Amprenavir-like P10–P20 resulted in 1:1 mixture of
diastereomers A (Ki = 2nM), Ref. 10.
reference inhibitor
0.0525lM).
4
(concentration at saturation
OH
O
O
N
N
S
R
O
O
O
A
B
R= Allyl
IC50=2 nM
Benzyl IC50= 30 nM