Evolution of anti-HIV drug candidates. Part 1: From α-Anilinophenylacetamide (α-APA) to imidoyl thiourea (ITU)

Article abstract of DOI:10.1016/S0960-894X(01)00410-3

Stemming from work on a previous clinical candidate, loviride, and other α-APA derivatives, a new series of potent non-nucleoside reverse transcriptase inhibitors (NNRTIs) has been synthesized. The ITU analogues, which contain a unique diarylated imidoyl thiourea, are very active in inhibiting both wild-type and clinically important mutant strains of HIV-1.

Full text of DOI:10.1016/S0960-894X(01)00410-3

Bioorganic & Medicinal Chemistry Letters 11 (2001) 2225–2228
Evolution of Anti-HIV Drug Candidates.
Part 1: From ꢀ-Anilinophenylacetamide (ꢀ-APA)
to Imidoyl Thiourea (ITU)
Donald W. Ludovici,a Michael J. Kukla,a,* Philip G. Grous,a Suma Krishnan,a
Koen Andries,b Marie-Pierre de Bethune,c,y Hilde Azijn,c,y Rudi Pauwels,c,y
Erik De Clercq,c Edward Arnoldd and Paul A. J. Janssene
aJanssen Research Foundation, Welsh and McKean Roads, Spring House, PA 19477, USA
bJanssen Research Foundation, Turnhoutseweg 30, B-2340 Beerse, Belgium
cRega Institute for Medical Research, K.U.Leuven, B-3000 Leuven, Belgium
dCenter for Advanced Biotechnology and Medicine and Rutgers University Chemistry Department,
679 Hoes Lane, Piscataway, NJ 08854, USA
eJanssen Research Foundation, Center for Molecular Design, Antwerpsesteenweg 37, B-2350 Vosselaar, Belgium
Received 5 September 2000; accepted 9 May 2001
Abstract—Stemming from work on a previous clinical candidate, loviride, and other a-APA derivatives, a new series of potent non-
nucleoside reverse transcriptase inhibitors (NNRTIs) has been synthesized. The ITU analogues, which contain a unique diarylated
imidoyl thiourea, are very active in inhibiting both wild-type and clinically important mutant strains of HIV-1. # 2001 Elsevier
Science Ltd. All rights reserved.
As part of an anti-AIDS program initiated in the late
1980s at Janssen Pharmaceutica, representative com-
pounds from the corporate library were screened for
their ability to inhibit HIV replication. As a con-
sequence, the Janssen group was the first to discover
unique inhibitors of the key multifunctional HIV-1
enzyme, reverse transcriptase. Compounds of the tetra-
hydroimidazo[4,5,1 - jk][1,4] - benzodiazepin - 2(1H) - one
(TIBO) series were found to be noncompetitive inhibi-
tors. Crystallographic studies later showed that they
inhibited the enzyme’s action by binding to a hydro-
phobic pocket close to, but distinct from, the active
site.1À3 Subsequently, many new structural classes of
spite of this shortcoming, nevirapine from Boehringer
Ingelheim and delavirdine from Upjohn were developed
and approved for sale as important adjuncts to the
accepted multidrug therapies used to treat HIV-positive
patients. Recently a third NNRTI, efavirenz from
DuPont-Merck, has been approved and can be con-
sidered a true second generation NNRTI because of its
ability to inhibit some of the mutant strains resistant to
its predecessors.
Nearly simultaneous with the development of the TIBO
series, a second structural class of NNRTIs was dis-
covered in our laboratories. The a-anilinophenylaceta-
mide (a-APA) series was potent versus the wild-type
HIV-1 (LAI strain) and had a high selectivity index
(inhibition of HIV-1 versus cell toxicity).6 Indeed, the
simplicity of the structures and the relative ease of
synthesis made them an extremely attractive target for
optimization. Ultimately, a clinical candidate, loviride
(1), was chosen and pursued through Phase II clinical
trials. Its development was discontinued when it became
apparent that it was not going to offer any significant
advantage over the NNRTI therapies already approved
at the time (vide supra). As described in this and the two
following communications, continued research in this
non-nucleoside
reverse
transcriptase
inhibitors
(NNRTIs) were found to specifically inhibit the reverse
transcriptase of HIV-1 by binding at this site.4 Although
less likely to cause deleterious side effects than nucleo-
side inhibitors, NNRTIs were found to be more vulner-
able to HIV’s high mutation rate, leading to rapid
selection of strains that are resistant to inhibition.5 In
*Corresponding author. Tel.: +1-215-542-9405; e-mail: m-rkukla@
erols.com
yPresent address: TIBOTEC, Generaal De Wittelaan L 11 B3, B-2800
Mechelen, Belgium.
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00410-3

2226
D. W. Ludovici et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2225–2228
.
Scheme 1. (a) HNR2 HCl, Me3Al, toluene; (b) ArNCS, 3 N NaOH,
MeOH.
Scheme 3. (a) 1. Na2SO3; 2. POCl3; (b) NH4OH, MeOH, 54% (c) 1.
p-cyanophenyl isothiocyanate, NaH, DMF; 2. K2CO3; 3. HCl, 41%.
Scheme 2. (a) SnCl4, KSCN, 69%; (b) p-cyanoaniline, 50%.
area has led to the discovery of several new families of
potent RT inhibitors from which second and possibly third
generation development candidates are being chosen.
Scheme 4. (a) p-Cyanoaniline, DMF, 30%; (b) DMF.
A series of ITUs was prepared using the synthetic route
illustrated in Scheme 1 that allowed for variation of the
substitution pattern of both aromatic rings (10a–10p,
Table 2) and the functional group backbone of the lin-
ker (13–21, Table 3). Isothiocyanates were allowed to
react with benzyl amidines under basic conditions, in
order to free the latter, which were added as acid addi-
tion salts. Many of the amidines (9) used in this study
were generated by addition of trimethyl aluminum
amide reagents generated in situ to the corresponding
nitriles (8).8
In structure–activity relationship (SAR) studies in the a-
APA series, intermediate 2 was allowed to react with a
number of electrophilic reagents to make analogues
related to 3a and 3b. Table 1 shows that extending the
ortho substituted inhibitor 3a by a methylene (4), a car-
bonyl (5), an amide (6), or thioamide (7a) yielded only
inactive compounds. However, activity was enhanced
when the aromatic substituent was moved to the meta
(7b) or para (7c) position of the thiourea. This was
contrary to the SAR that had evolved in the a-APA
series in which ortho substituents (cf. 3a and 3b) were
highly favored over para (3c). Thus, we realized that we
were potentially exploring a SAR different from the a-
APA series.
A number of conclusions were reached from this group
of analogues (Table 2). Substitutions on the anilino ring
(Y, 10a–10g) demonstrated a positional preference for
para substitution (10gꢀ10fꢀ10e). Di- or tri-substi-
tution significantly decreased activity as compared to
the corresponding mono-para substitution (10dꢀ10b or
10c). The order of activity (IC50, mM) of the synthesized
para substituents (data not shown in Table 2) was: CN
(0.003), NO2 (0.003)>Br (0.013), Cl (0.013)>OBzl
(0.020)>CF3 (0.048), I (0.059), n-Bu (0.065), Ac (0.066),
Et (0.074), F (0.102), CO2Et (0.112), H (0.123)>SMe
(0.407), OMe (0.417). There was some ambiguity about
the relative importance of electronic effects and steric
bulk. In general, electron withdrawing substituents were
favored. The fact that compounds with alkyl sub-
stituents were as active as those with fluorine indicated
that some bulk was preferable and size could play as
important a role as electronic effects.
A variety of aromatic substitutions on 7, as well as var-
iations to the basic structure, showed no significant
enhancement of anti-HIV-1 activity (data not shown)
until the imidoyl thiourea (ITU) 10d was prepared (see
Scheme 1 and Table 2). This compound was extremely
active when compared to the a-APA leads. Thus, we
embarked on an optimization program that is the sub-
ject of this communication.
Table 1. Activity (IC50, mM) versus HIV-17
Keeping the p-CN anilino motif (Y) as the optimum
subunit, a number of benzyl aromatic substitutions (X,
10h–10p) were compared. Each substitution enhanced
activity. Indeed, unsubstituted compound 10k was the
least active analogue. There seemed to be little differ-
ence in electronic effects with monosubstitution. Thus,
the chloro, methyl, and methoxy substituted molecules
were basically equipotent when positional (ortho) iso-
mers were compared (10l, 10o, and 10p). In two of three
possible direct comparisons, we found ortho>meta>para
Compd
E
Y
IC50 (mM)
3a
3b
3c
4
5
6
7a
7b
7c
—
—
—
CH2
o-NO2
o-Cl
p-Cl
o-NO2
o-NO2
o-NO2
o-Cl
0.1
0.4
>160
>250
>79
>320
>60
5.2
CO
CONH
CSNH
CSNH
CSNH
m-Cl
p-Cl
0.16

D. W. Ludovici et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2225–2228
2227
to react sequentially with p-cyanoaniline and with
dimethylamidine 9b to yield the cyanoguanidine analo-
gue 22. When the reaction was carried out with the
unsubstituted amidine 9a, the analogous product was
not isolated. Instead a cyclization occurred with the
nitrile of the cyanoguanidine to yield the triazine pro-
duct 30. The biguanide analogue 23 was synthesized by
methylating N-p-cyanophenyl thiourea and displacing
methylmercaptan with 9a (Scheme 5). Methylated ana-
logue 29 was prepared by allowing N-methyl p-cyano-
aniline to react with diimidazolethiocarbonyl followed
by displacement of the second imidazole with amidine
9a (Scheme 6).
Scheme 5. (a) Mel, acetone, NH4OH, 84%; (b) 9a, CH3CN, 42%.
Scheme 6. (a) Im2CS, THF, 38%; (b) 9a, DMF, 63%.
substitution (10l>10m>10n). Clearly multiple substi-
tution, especially ortho disubstitution, led to the highest
potency (10g and 10j).
Table 3 summarizes the biological activity of a series of
analogues of 10g after optimization of the aromatic
substituents. Substitution (14 and 15), elongation (13
and 17), or shortening (16) of the methylene (A–B) from
the lead all led to significant decreases in activity. When
the imidoyl nitrogen was substituted with a methyl
group (18), potency was nearly the same as 10g. How-
ever, larger groups such as ethyl (20) and phenyl (21)
gave diminished potency. A dimethylated analogue (19)
was nearly 2000Â less active. Replacement of the imi-
doyl with carbonyl (24) or sulfonyl (25) led to complete
loss of activity. Conversion of the thiourea in 10g to a
urea gave a compound (11) that was 6-fold less active.
When this change was combined with methylation of
the imidoyl nitrogen (12), the resultant analogue was
nearly 100Â less active. Other replacements of the sul-
fur, such as the biguanide (23) or the bioisosteric cya-
noguanidine (22) were nearly inactive. Unfortunately,
the direct cyanoguanidine analogue of 10g could not be
prepared since, as mentioned previously, it spontaneously
cyclizes. Similarly, elongation at the cyanoaniline end
We then turned our attention to variations of the lead
structure 10g to explore the importance of each portion
of the imidoyl thiourea backbone. Schemes 2–6 illus-
trate the preparation of compounds 22–25, 29, and 30.
In general, the basic strategy was the same as was used
to make the lead structures; that is, the combination of
aromatic electrophilic and aromatic nucleophilic moi-
eties corresponding to the target analogues. Variations
in the imidoyl portion of the lead involved initial repla-
cement of the nitrogen by a carbonyl group. In Scheme
2, treatment of the acid chloride with potassium thio-
cyanate in the presence of tin chloride yielded the elec-
trophilic acyl thioisocyanide, which was allowed to react
with p-cyanoaniline to readily yield the acyl thiourea
analogue 24.9 The corresponding sulfone analogue 25
(Scheme 3) was generated by treatment of p-cyanophe-
nyl isothiocyanate with the anion of the dichlorobenyl
sulfonamide, itself obtained by a three step sequence.
The o,o-dichlorobenzyl chloride was treated with
sodium sulfite, followed by phosphorous oxychloride, to
give the sulfonyl chloride that, after treatment with
ammonium hydroxide, gave the sulfonamide.10
Table 3. Activity (IC50, mM) versus HIV-17
To modify the thiourea portion of the imidoyl thiourea
(Scheme 4), diphenyl cyanocarbonimidate was allowed
Compda
A
B
W
Z
5I0C(mM)
Table 2. Activity (IC50, mM) versus HIV-17
10g
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26b
27b
28b
29
CH2
C¼NH
S
NH
0.0025
0.015
0.214
2.00
1.48
0.158
22.9
0.525
0.0039
5.13
0.158
1.32
12.9
7.94
631
100
0.013
0.692
1.62
O
O
O
C¼N–Me
CH¼CH (E)
1,1-CycloPr
CHCH3
Nothing
CH2CH2
Compd
X
Y
IC50 (mM)
10a
10b
10c
10d
10e
10f
10g
10h
10i
10j
10k
10l
10m
10n
10o
10p
o,o-diCl
o,o-diCl
o,o-diCl
o,o-diCl
o,o-diCl
o,o-diCl
o,o-diCl
o,p-diCl
o,o,m-triCl
o,o-diF
H
o-Cl
m-Cl
p-Cl
o-Me
o-OMe
H
o,p-diCl
o,o,p-triCl
p-Cl
0.123
0.347
42.7
0.013
631
0.126
0.003
0.007
0.005
0.006
0.380
0.014
0.068
0.148
0.026
0.030
C¼N–Me
C–N–Me2
C¼N–Et
C¼N–Ph
o-CN
m-CN
p-CN
p-CN
p-CN
p-CN
p-CN
p-CN
p-CN
p-CN
p-CN
p-CN
C–NMe2 N–CN
NH
C¼O
SO2
NH
NHCH2
NHCH2CH2
N–CH3
58.9
aUnless noted, all variables (A, B, W, Z) for listed compounds (11–29)
are the same as 10g.
bIn this analogue, the p-cyano was replaced by p-Cl.

2228
D. W. Ludovici et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2225–2228
Table 4. Activity (IC50, mM) of 10g and reference NNRTI versus
CABM and Rutgers University for X-ray analysis of
key compounds with the reverse transcriptase;12 to
Diane Gautier and Greg Leo for help with NMR spec-
tral analyses; and to Theo Thielemans for continuously
updating extremely well organized data presentations.
mutant strains of HIV-17
LAI 100I 101E 103N 106A 181C 188L 190A
Delavirdine 0.063 2.51 0.158 2.51 1.59 2.00
Loviride 0.050 0.050 0.063 1.26 1.00 15.8
Nevirapine 0.032 0.316 0.316 6.31 5.01 10.0 >100.0 7.94
10g 0.003 0.513 0.019 0.589 0.382 0.511 0.318 0.002
1.26 0.063
50.1 2.51
References and Notes
1. Pauwels, R.; Andries, K.; Desmyter, J.; Schols, D.; Kukla,
M. J.; Breslin, H. J.; Raeymaeckers, A.; Van Gelder, J.;
Woestenborghs, R.; Heykants, J.; Schellenkens, K.; Janssen,
M. A. C.; De Clercq, E.; Janssen, P. A. J. Nature 1990, 470.
2. Kukla, M. J.; Breslin, H. J.; Pauwels, R.; Fedde, C.; Mir-
anda, M.; Scott, M. K.; Sherrill, R. G.; Raeymaekers, A.; Van
Gelder, J.; Andries, K.; Janssen, M. A. C.; De Clercq, E.;
Janssen, P. A. J. J. Med. Chem. 1991, 34, 746.
3. (a) Kohlstaedt, L. A.; Wang, J.; Friedman, J. M.; Rice,
P. A.; Steitz, T. A. Science 1992, 256, 1783. (b) Ding, J.; Das,
K.; Moereels, H.; Koymans, L.; Andries, K.; Janssen, P. A. J.;
Hughes, S. H.; Arnold, E. Nat. Struct. Biol. 1995, 2, 407. (c)
Mager, P. P. Med. Res. Rev. 1997, 17, 235.
by one (27) or two (28) carbons decreased activity sig-
nificantly. Thus, the length of the molecule was critical
to good potency. From the options explored, there
seems to be little tolerance for changes to 10g. Virtually
all modifications cause a significant decrease in potency
except the simple methylation of the imidoyl nitrogen
(18); even extension to the homologous ethyl substituent
(20) causes a drastic drop in activity. Replacement of
the thiourea in 10g by urea (11) was tolerated but led to
6-fold loss of activity.
4. (a) Swindells, S. Emerging Drugs 1997, 2, 155. (b) Pedersen,
O. S.; Pedersen, E. B. Antiviral Chem. Chemother. 1999, 10,
285. (c) De Clercq, E. Il Farmaco 1999, 54, 26.
5. (a) Richman, D.; Shih, C. K.; Lowy, I.; Rose, J.; Prodano-
vich, P.; Goff, S.; Griffin, J. Proc. Natl. Acad. Sci. U.S.A.
1991, 88, 11241. (b) Richman, D. Annu. Rev. Pharmacol.
Toxicol. 1993, 132, 149.
6. Pauwels, R.; Andries, K.; Debyser, Z.; Van Daele, P.;
Schols, D.; Stoffels, P.; De Vreese, K.; Woestenborghs, R.;
Vandamme, A.-M.; Janssen, C. G. M.; Anne, J.; Cau-
wenbergh, G.; Desmyter, J.; Heykants, J.; Janssen, M. A. C.;
De Clercq, E.; Janssen, P. A. J. Proc. Natl. Acad. Sci. U.S.A.
1993, 90, 1711.
One of the biggest challenges in finding improved gen-
erations of NNRTIs has been to find compounds that
maintain activity against the common mutations. We
tested the lead structure against a number of the most
commonly reported drug-resistant mutant strains of
HIV-1. As indicated in Table 4, 10g has superior activ-
ity as compared to our original clinical candidate,
loviride, as well as the FDA-approved agents dela-
virdine and nevirapine. Of the 24 possible comparisons
(3 agentsÂ8 HIV strains), 10g is the more potent in 22
cases.
7. All compounds were tested for potency (IC50, mM) to
achieve 50% protection of MT-4 cells from HIV-1 cytopathi-
city as determined by the MTT method (Pauwels, R.; Balzar-
ini, J.; Baba, M.; Snoek, R.; Schols, D.; Herdewijn, P.;
Desmyter, J.; De Clercq, E. J. Virol. Methods 1988, 20, 309).
Unless noted otherwise, the LAI strain of HIV-1 was the
infecting virus. Other infecting viral strains with mutations in
reverse transcriptase are characterized in the tables by the
mutated amino acid position and the one letter codes. For
instance, 181C refers to replacement of tyrosine at position
181 with cysteine. Each determination is the result of multiple
tests. Although the data are not reported, in the same experi-
ment mock-infected cells were tested with compound to deter-
mine the dose to reduce cells to 50% viability (CC50). Thus a
selectivity index (CC50/IC50) could be determined.
In summary, through a SAR program beginning with
the clinical candidate loviride, 10g was discovered to be
a potent, noncytotoxic compound with activity against
a wide variety of HIV-1 mutant strains. Furthermore, it
could be synthesized in one step from commercially
available starting materials and it contained no optical
centers! Consequently, 10g was considered for clinical
development. Unfortunately, the hydrolytic instability
of the imidoyl thiourea functionality became an issue
during formulation studies. An obvious solution to this
problem was to synthesize the well known cyanoguani-
dine bioisostere. However, as mentioned earlier in the
SAR discussion, this compound was not available since
it immediately cyclized to triazine 30. Evaluation of this
unexpected product indicated that it was nearly as active
as 10g; this led to a renewed optimization program that
is the subject of the accompanying communication.11
8. Garigipati, R. S. Tetrahedron Lett. 1990, 31, 1972.
9. Cambie, R. C.; Davis, P. F.; Rutledge, P. S.; Woodgate,
P. D. Aust. J. Chem. 1984, 37, 2073.
10. Hamer, M.; Batzer, O. F.; Moats, M. J.; Wu, C. C.; Lira,
E. P. J. Pharm. Sci. 1975, 64, 1961.
11. Ludovici, D. W.; Kavash, R. W.; Kukla, M. J.; Ho, C. Y.;
Ye, H.; De Corte, B. L.; Andries, K.; de Bethune, M.-P.;
Azijn, H.; Pauwels, R.; Moereels, H. E. L.; Heeres, J.; Koy-
mans, L. M. H.; de Jonge, M. R.; Van Aken, K. J. A.;
Daeyaert, F. F. D.; Lewi, P. J.; Das, K.; Arnold, E.; Janssen,
P. A. J. Bioorg. Med. Chem. Lett. 2001, 11, 2229 (following
paper in series).
Acknowledgements
Many people have made important contributions to this
work. Thanks go to Paul Lewi and our colleagues in the
JRF-CMD for many stimulating discussions and sug-
gestions;12 to co-workers in Eddy Arnold’s group at
12. Detailed results will be part of future publications.