Modification of the clozapine structure by parallel synthesis

Article abstract of DOI:10.1016/j.bmcl.2006.06.034

A structure-activity study based on the core structure of clozapine 1b was accomplished by utilizing high-throughput synthesis. Several focused libraries were designed and synthesized to quickly develop SAR. The results indicate that by varying different regions of clozapine, both D₁-selective and D₂-selective compounds can be obtained.

Full text of DOI:10.1016/j.bmcl.2006.06.034

Bioorganic & Medicinal Chemistry Letters 16 (2006) 4548–4553
Modification of the clozapine structure by parallel synthesis
Jing Su,a,* Haiqun Tang,a Brian A. McKittrick,a Duane A. Burnett,a Hongtao Zhang,b
April Smith-Torhan,b Ahmad Fawzib and Jean Lachowiczb
aDepartment of Chemical Research, Schering-Plough Research Institute K15 2545, 2015 Galloping Hill Road,
Kenilworth, NJ 07033, USA
bDepartment of Biological Research, Schering-Plough Research Institute K15 3600, 2015 Galloping Hill Road,
Kenilworth, NJ 07033, USA
Received 16 May 2006; revised 8 June 2006; accepted 9 June 2006
Available online 27 June 2006
Abstract—A structure–activity study based on the core structure of clozapine 1b was accomplished by utilizing high-throughput
synthesis. Several focused libraries were designed and synthesized to quickly develop SAR. The results indicate that by varying dif-
ferent regions of clozapine, both D1-selective and D2-selective compounds can be obtained.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Dopamine receptors can be broadly classified into two
subtypes: the D1-like subtype (D1 and D5) that upon
activation stimulates the production of cAMP and the
D2-like subtype (D2–D4) that inhibits cAMP production
upon activation. Many D2 antagonists are used to treat
schizophrenia,1 while D2 agonists such as daverium are
used to treat Parkinson’s disease.2 D1/D5 antagonists
such as ecopipam have been studied for drug and alco-
hol abuse,3 and D1 agonists such as fenoldopam are
used to treat hypertension.4 Clozapine 1b has been used
to treat schizophrenia since 1982.5,6 Unlike typical anti-
psychotics such as chloropromazine and haloperidol,
which are effective only against positive symptoms of
schizophrenia, clozapine is effective against both posi-
tive and negative symptoms. Furthermore, clozapine
does not induce severe extrapyramidal side effects
(EPS) at clinically effective doses.7,8 The unique profile
of clozapine, however, may not be attributed solely to
its blockade of the D2 receptor, as it also binds other
dopaminergic serotonergic, adrenergic, histaminergic
and muscarinic receptors.5 This non-selective profile is
believed to elicit the superior overall efficacy of cloza-
pine.5 Although some SAR investigations of clozapine
have been reported in the literature, to our knowledge,
a broader exploration of the clozapine SAR is not avail-
able.9–20 Herein, we wish to report our results of modi-
fications of the N-5 nitrogen region, the distal piperazine
nitrogen region, and the tricyclic skeleton of clozapine
using high-throughput parallel synthesis21 (Scheme 1).
Our initial derivatization of the core clozapine N-5
nitrogen included sulfonylation, acylation, reductive
alkylation, and urea formation. As expected, this hin-
dered nitrogen was less reactive and only a few reactions
provided the desired products. Some representative re-
sults are shown in Table 1. Significant among these re-
sults, the sulfonamides 2b and 2c retained D1
affinity.22 Introducing an alkylated nitrogen atom exo-
cyclic to the core tricyclic ring provided the hydrazine
2e which also retained the D1 activity. Interestingly,
hydrazide 3a further improved the affinity for D1
(Ki = 13 nM) and the compound possessed high selectiv-
ity against D2 receptor (D2/D1 = 87). Compound 3a was
an ideal lead for SAR exploration using parallel synthe-
sis. Both solution-phase and solid-phase synthesis could
be used to prepare analogous targets. Thus, compound
4b23,24 was coupled with Argopore-CHO resin followed
by acylation with acid chlorides.21 The final product
could be cleaved from the resin using 100% TFA
(Scheme 2).25 In addition, a solution-phase parallel syn-
thesis was also developed to further expand the SAR.26
Selected results from our first 60-membered library by
solid-phase synthesis are listed in Table 2. Overall,
substitution on the phenyl ring improved the D1 affinity,
regardless of the electronic/steric nature of the
Keywords: Dopamine receptors; Clozapine; D1-selective antagonist;
D2-selective antagonist; High-throughput synthesis.
*
Corresponding author. Tel.: +1 908 740 7489; fax: +1 908 740
7164; e-mail: jing.su@spcorp.com
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.06.034

J. Su et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4548–4553
4549
Table 2.
Compound
Ar
D1 Ki
(nM)
D2 Ki
(nM)
D2/D1
3b
3c
3d
3e
3f
Ph
4-CNPh
157
68
1150
>2000
>3000
3648
1362
792
7.3
>29
4-MeOPh
3-CNPh
3-CF3Ph
47
62
>63
59
148
16
9.2
3g
3h
3i
2-BrPh
2-ClPh
50
44
4.1
13
181
Scheme 1.
Table 1.
2-MeOPh
2,3-F2Ph
1126
533
87
33
3j
16
56
3k
3l
3,4-OCH2OPh
2-CF3, 4-FPh
3,4,5-(MeO)3Ph
2,3,4,5-F4Ph
2,6-Me2Ph
2,6-(MeO)2Ph
3-Pyridyl
4-Pyridyl
828
15
33
49
31
1616
5561
1280
287
R
N
3m
3n
3o
3p
3q
3r
180
40
32
3.0
1.6
85
96
Cl
N
340
12710
28425
210
150
330
N
86
N
2a-e, 3a
selectivity. Furthermore, introduction of the pyridyl
group also significantly improved D2/D1 selectivity
(compounds 3q–r).
Compound
R
H
D1 Ki
(nM)
D2 Ki
(nM)
D2/D1
1b
2a
2b
2c
2d
2e
3a
132
1232
65
208
4927
338
1.6
4.0
5.2
0.9
22
Since little was known about the SAR on the distal
piperazine nitrogen in this particular series, we synthe-
sized compound 6a and used reductive alkylation to
introduce various alkyl groups on this nitrogen in a
high-throughput fashion. At the time of our investiga-
tion, synthesis of such analogs in a clozapine or cloza-
pine-like series using this synthetic strategy had not
been reported.28 Subsequently, Capuano et al. reported
synthesis of N-arylmethyl clozapine analogs using this
strategy.19 In our experiment, the reaction was conduct-
ed in solution phase with 75 aldehydes and ketones, fol-
lowed by resin cleanup as shown in Scheme 3.29 Selected
binding data are shown in Table 3. The data strongly
suggested that small and unhindered alkyl groups
were preferred on this distal nitrogen. The N-methyl
analog 3h was the best compound from this series.27
Larger alkyl groups decreased D1 affinity, leading to
lower D2/D1 selectivity. Our efforts to modify the
o-TolCO
p-TolSO2
o-TolSO2
71
63
3,5-Cl2PhNHCO
2-MeOPhCH2NH
2-MeOPhCONH
175
60
3896
202
3.4
87
13
1126
substituents and their position on the phenyl ring. Poly-
substitution also improved the D1 affinity (compounds
3j–m). In general, compounds with 2-substitution dis-
played lower Ki values than 3- or 4-substituted ones
(compounds 3i vs 3d and 3f).27 This trend was in agree-
ment with our previous observation that 2-substitution,
especially 2,6-disubstitution, afforded both strong bind-
ing to the D1 receptor and high D2/D1 selectivity (com-
pounds 3o–p).27 Interestingly, 3,4,5-trimethoxyphenyl
amide (compound 3m) also provided significant D2/D1
NH2
N
H
N
Cl
Cl
N
N
Argopore-CHO resin
NaBH3CN, HOAc
1. isoamyl nitrite
N
N
2. Zn, HOAc
42%
N
N
1b
4b
Ar
O
HN
N
NH
N
O
100%TFA
60-70%
Ar
Et3N
Cl
Cl
Cl
N
N
N
N
N
N
3
4c
Scheme 2.

4550
J. Su et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4548–4553
NH2
N
H
N
1. TeocOC4H4-p-NO2
2. isoamyl nitrite
Cl
Cl
N
N
ArCOCl
Et3N
TBAF
75%
N
N
3. Zn, HOAc
35%
N
H
N
Teoc
1a
5
Cl
Cl
O
O
HN
HN
N
N
1. reductive alkylation
Cl
Cl
2. resin cleanup
85-90%
N
N
N
N
N
N
6b-i
H
6a
R
Scheme 3.
Table 3.
amisulpride have been reported as effective antipsychot-
ic drugs,32 to our knowledge, this is the first report of a
highly selective D2 antagonist with a clozapine-like
structure. This finding is of particular relevance to anti-
psychotic drug research.
Compound
R
D1 Ki
(nM)
D2 Ki
(nM)
D2/D1
3h
6b
6c
6d
6e
6f
Me
Et
4.0
54
181
709
44
13
n-Pr
i-Pr
26
409
998
16
191
81
5.2
Our exploration of the tricyclic skeleton SAR started
with relocating the 8-Cl group to the 2-position based
on literature reports that such a manipulation to
clozapine improved D1 and D2 affinity by 10-fold.10,11
The synthesis of the building block 12 is depicted in
Scheme 4.33 Using the solution-phase chemistry de-
scribed above, modification of the top part of the mol-
ecule produced 51 compounds. Some representative
examples from this library are shown in Table 5. The
anticipated improvement of D1 affinity did not occur in
this amide series, whereas improvement of D2 affinity
was observed in some cases, resulting in lower D2/D1
selectivity (compounds 13j vs 3p). In general, the SAR
of this series was consistent with that of compound 3 in
Table 2. The best compounds were still those bearing
2-substituted and 2,6-disubstitutedbenzoyl amides (com-
pounds 13i–j). Interestingly, 3,4,5-trimethoxyphenyl
amide 13g still held a D2 selectivity greater than 100-fold
(Scheme 5).
CH2Pr-i
CH2Bu-t
>3000
>3000
>3000
>3000
>2000
>37
>6.7
>5.9
>10
>4.0
447
505
294
503
6g
6h
6i
CH2CHEt2
CH2cyclohexyl
2-FBn
piperazine conformation by introducing a methyl group
alpha to the distal piperazine nitrogen provided similar
SAR trends.30 Other efforts to explore this region
included replacing the piperazine ring with homopiper-
azine or N,N,N0-trimethylethylenediamine and these
modifications resulted in at least 100-fold loss of D1
and D2 activity.31
For direct comparison of the hydrazine and clozapine
series, the reductive alkylation chemistry was applied
to N-desmethyl clozapine 1a and 43 compounds were
synthesized. A previous study by Capuano et al. focused
on reductive alkylation of 1a with a few substituted ben-
zaldehydes and it was concluded that introduction of an
N-arylmethyl group into the clozapine structure did not
have a significant effect on D2 binding.19 Our results, in
Table 4, were in sharp contrast to Capuano’s. We found
that bigger alkyl groups led to diminished D1 affinity for
compounds 1c–i as we had previously found for hydra-
zide 6b–i. However, it was quite interesting to see that
the D2 affinity of 1c–i was dramatically improved to
the extent that they became D2-selective antagonists.
The best compound 1i had a D2 Ki of 3 nM with greater
than 1000-fold selectivity over D1, whereas the corre-
sponding N-acylhydrazino analog 6p had D2 Ki of
>3000 nM. Although D2-selective antagonists such as
In order to further evaluate the combination effect of
introducing an additional chlorine atom and a hydra-
zide on dopamine receptor affinity, we used the
above chemistry to synthesize the dichloroclozapine
building block 16 and a 54-membered library synthe-
sis of compounds 17 was accomplished (Table 6).
Similar to their analogs in Table 5, the 2-substituted
and 2,6-disubstituted phenyl amides provided the best
D1 affinity (compounds 17h–j), with D2/D1 selectivity
generally being low except for 17h. A comparison of
the results in Tables 2 and 6 demonstrated that the
chlorine atom at the 2-position on the tricyclic
skeleton significantly improves the D2 activity but
not the D1 activity.

J. Su et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4548–4553
4551
Table 4.
Cl
O
HN
N
H
N
Cl
N
Cl
N
N
N
N
N
X
X
6j-p
1c-i
X
Compound
D1 Ki (nM)
D2 Ki (nM)
Compound
D1 Ki (nM)
D2 Ki (nM)
2,4-F2
2,3-(OCH2O)
2,4-(Dime)2
1c
1d
1e
1f
835
637
177
80
65
29
11
8
6j
886
571
>3000
>2000
>3000
>2000
>3000
>3000
>3000
6k
6l
482
583
1097
238
3,4-(OC2H4O)
4-t-Bu-Cyclohexyl
2,3-(OCF2O)
3-CF3O
6m
6n
6o
6p
1g
1h
1i
1734
3106
3799
2273
4749
2568
3.4
COOH
H
N
Br
H2N
Cu, K2CO3
+
amyl alcohol, 150 0
100 %
C
NO2
HOOC
Cl
NO2
Cl
9
8
7
H
N
H
N
HN
Cl
N
Cl
o-xylene
53 %
Na2S2O4
TiCl4
80%
N
N
N
H
O
11
N
10
R
O
HN
N
NH2
N
1. isoamyl nitrite
2. Zn, HOAc
40%
Cl
Cl
RCOCl
Et3N
80-95%
N
N
N
N
N
N
13
12
Scheme 4.
Table 5.
In summary, we have utilized a parallel synthesis strate-
gy to develop SAR of three different regions of clozapine
1b. Modification of the aryl group on the top region of
the molecule resulted in high affinity D1 antagonists
such as 3h, which was also selective against D2. Modifi-
cation of the tricyclic skeleton provided D1 antagonists
such as 13j and 17h with similar affinity but lower D2
selectivity. While installation of bigger alkyl groups on
the distal piperazine nitrogen resulted in loss of D1
and D2 activity in hydrazide series 6e–6p, highly D2-se-
lective compounds such as 1i were discovered when the
same chemistry was applied to N-desmethyl clozapine
1a.
Compound
R
D1 Ki
(nM)
D2 Ki
(nM)
D2/D1
13a
13b
13c
13d
13e
13f
13g
13h
13i
4-CNPh
3-CNPh
64
99
1969
1186
1379
421
31
12
3-OMePh
2-CF3,5-FPh
2,3,4,5-F4Ph
86
22
16
19
156
759
4.9
15
3,4-OCH2OPh
3,4,5-(MeO)3Ph
2-MePh
83
20
48
1261
3230
1571
231
160
33
2-IPh
2,6-(MeO)2Ph
7.0
6.0
33
33
13j
197

4552
J. Su et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4548–4553
COOH
H
Br
H2N
N
Cu, K2CO3
amyl alcohol, 150 0
+
C
Cl
NO2
HOOC
Cl
Cl
Cl
NO2
Cl
95%
14
8
1
5
R
HN
O
NH2
N
N
RCOCl
Cl
see Scheme 4
Cl
Cl
Et3N
80-95%
N
N
N
N
N
N
17
16
Scheme 5.
Table 6.
10. Liao, J.; Venhuis, B. J.; Rodenhuis, N.; Timmerman, W.;
Wikstrom, H.; Meier, E.; Bartoszyk, G. D.; Boettcher, H.;
Seyfried, C. A.; Sundell, S. J. Med. Chem. 1999, 42, 2235.
11. Liao, Y.; DeBoer, P.; Meier, E.; Wikstrom, H. J. Med.
Chem. 1997, 40, 4146.
12. Chernov, G. S.; Khlienko, Z. N.; Mekhova, G. M.;
Susekova, T. V.; Kolla, V. E.; Sul’din, A. V.; Shelenkova,
S. A.; Tregubov, A. L.; Pilipchuk, T. V. WO 9110661 A1,
1991.
Compound
R
D1 Ki
(nM)
D2 Ki
(nM)
D2/D1
17a
17b
17c
17d
17e
17f
17g
17h
17i
4-CNPh
3-CNPh
3-OMePh
51
47
194
175
139
69
3.8
3.7
3.2
2.9
2.8
2.2
17
44
24
2-CF3, 5-FPh
2,3,4,5-F4Ph
3,4-OCH2OPh
3,4,5-(MeO)3Ph
2-MePh
98
270
286
267
175
36
130
16
13. Horrom, B. W.; Minard, F. N.; Zaugg, H. E. US4097597,
1978.
14. Hunziker, F. Ger. Offen. DE 76-2641378, 1977.
2.0
12
88
2-IPh
2,6-(MeO)2Ph
3.0
8.8
15. Burki, H.; Fischer, R.; Hunziker, F.; Kunzle, F.; Petcher,
¨
¨
17j
11
97
T.; Schmutz, J.; Weber, H. P.; White, T. Eur. J. Med.
Chem.-Chim. Ther. 1978, 13, 479.
16. Horrom, B. W.; Barta, W. D. US4096261, 1978.
17. Tehim, A.; Fu, J. M.; Rakhit, S. WO 9517400 A1, 1995.
18. Tehim, A.; Fu, J. M.; Rakhit, S. WO 9618629 A1, 1996.
19. Capuano, B.; Crosby, I. T.; LLoyd, E. J.; Taylor, D. A.
Aust. J. Chem. 2002, 55, 565.
20. Capuano, B.; Crosby, I. T.; LLoyd, E. J.; Podloucka, A.;
Taylor, D. A. Aust. J. Chem. 2003, 56, 875.
21. For an example of SAR development using high-through-
put synthesis, see: Su, J.; McKittrick, B. A.; Tang, H.;
Czarniecki, M.; Greenlee, W. J.; Hawes, B. E.; O’Neill, K.
Bioorg. Med. Chem. 2005, 13, 1829.
Acknowledgments
We thank Drs. Michael Czarniecki, William Greenlee
for their support and suggestions. We also thank Dr.
T. K. Sasikumar and Ms. Li Qiang for carrying out
early SAR exploration in the clozapine series.
References and notes
22. For experimentals: Ltk-cells stably expressing D1 and D2
receptors at a density of 4–7 pmol/mg protein were lysed in
hypotonic buffer and centrifuged at 48,000g. Membrane
pellets were frozen and stored at ꢀ80 ꢁC for use in binding
assays. Receptor affinities were determined by equilibrium
binding experiments in which bound and free radioligands
were separated by rapid filtration, and bound counts were
quantified by liquid scintillation counting. For D1 binding,
the radioligand was [3H] SCH 23390 (0.3 nM), and
non-specific binding was defined by addition of 10 lM
unlabeled SCH 23390. For D2 binding, the radioligand
was [3H]methylspiperone (0.5 nM) and non-specific bind-
ing was defined using 10 lM (ꢀ)-sulpride. Test com-
pounds, radioligand, and membrane homogenates
prepared from CHO cells expressing each receptor subtype
were incubated in a 200 lL volume for 1 h at room
temperature prior to filtration on GF-C plates. Competi-
tion binding data were analyzed using Graphpad Prism, in
which curves fit a one-site competition model with a Hill
Slope equal to or approximately 1. Mean Ki values from
four separate determinations are reported. The SEM was
below 15% in each case. LCMS analysis was performed on
1. Schaus, J. M.; Bymaster, F. P. Annu. Rep. Med. Chem.
1998, 33, 1.
2. Grognet, J. M.; Istin, M.; Zanotti, A.; Mailland, F.;
Coppi, G. Drugs Exp. Clin. Res. 1991, 17, 309.
3. Hany, M.; Ward, A. S.; Foltin, R. W.; Fishman, M. W.
Psychopharmacology 2001, 155, 330.
4. Lefevre-Borg, F.; Lorrain, J.; Lechaire, J.; Thiry, C.;
Hicks, P. E.; Cavero, I. J. Cardiovasc. Pharmacol. 1988,
11, 444.
5. Burns, M. J. Clin. Toxicol. 2001, 39, 1.
6. Beaumont, G. Curr. Med. Res. Opin. 2000, 16, 37.
7. Kane, J. Drugs 1993, 46, 585.
8. Lieberman, J. A.; Johns, C. A.; Kane, J. M.; Rai, K.;
Pisciotta, A. V.; Saltz, B. L.; Howard, A. J. Clin.
Psychiatry 1988, 49, 271.
9. For modification of the 5H-dibenzo-[b,e][1,4]diazepine
skeleton of clozapine, see Refs. 10,11. For modification
of the top NH region of clozapine, see Refs. 12–14.
For modification of the distal nitrogen of piperazine
region of clozapine, see Refs. 15–20.

J. Su et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4548–4553
4553
an Applied Biosystems API-100 mass spectrometer and
Shimadzu SCL-10A LC column: Altech platinum C18, 3
micron, 33 mm · 7 mm ID; gradient flow: 0 min, 10%
CH3CN; 5 min, 95% CH3CN; 7 min, 95% CH3CN;
7.5 min, 10% CH3CN; 9 min, stop. Chromatography was
performed with Selecto Scientific flash silica gel, 32–63 lM.
23. Clozapine was synthesized according to: Glamkowski, E.
J.; Chiang, Y. J. Heterocycl. Chem. 1987, 24, 1599.
24. The conversion of clozapine 1b to 4b was based on Ref.
12 with modifications. Two grams of clozapine 1b was
dissolved in 80 mL DCM and 40 mL isoamylnitrite at rt
for 3 h. The solvent was removed and the crude was
dissolved in 40 mL HOAc. This solution was added
dropwise to Zn (10 g)/HOAc (150 mL) over 1 h at 10–
15 ꢁC. Additional Zn (1 g) was periodically added to
keep the green color of the solution. After 3 h, the
solution was filtered and the solvent evaporated. Four
hundred milliliters of DCM was added along with 50 mL
of water. The pH was adjusted to 11 and extraction was
done with 3·100 mL DCM. The crude was recrystallized
with DCM and hexane to give 0.84 g of the hydrazine 4b
(42% yield). 1H NMR (CDCl3): d 2.40 (s, 3H), 2.70 (br m, 4
H), 3.60 (br m, 4H), 4.40 (br s, 2H) 6.77 (d, 1H, J = 8.2 Hz),
6.83 (s, 1H), 7.10–7.40 (m, 4H) 7.79 (d, 1H, J = 8.2 Hz).
25. 100% TFA for cleavage was necessary likely due to the
presence of the basic functional group. Novabiochem-
CHO resin gave similar results.
amine (6 equiv, ArgonautTM), resin-bound isocyanate
(3 equiv, ArgonautTM) were added to absorb the excess
acid chloride and unreacted hydrazine for 2 h. After
filtration, the sample was dried with no further
purification.29
27. Sasikumar, T. K. et al., preceding paper. In general,
all compounds in the library showed similar profiles
(single digit to a few hundred nanomolar Ki for D1
and similar D2/D1 selectivity to those in the tables).
28. All previously reported syntheses of such analogs utilized
a coupling of the N-alkylpiperazine with a tricyclic lactam
such as 10 going to 11. Our modified route provides ready
access to a wider range of products.
29. Siegel, M. G.; Hahn, P. J.; Dressman, B. A.; Fritz, J. E.;
Grunwell, J. R.; Kaldor, S. W. Tetrahedron Lett. 1997, 38,
3357.
30. For example, the analog of 6b shows D1 Ki 68 nM, D2 Ki
553 nM.
31. For example, the N,N,N0-trimethylethylenediamine ana-
log of 3h shows D1 Ki 435 nM, D2 Ki 1800 nM.
32. Schoemaker, H.; Claustre, Y.; Fage, D.; Rouquier, L.;
Chergui, K.; Curet, O.; Oblin, A.; Gonon, F.; Carter, C.;
Benavides, J.; Scatton, B. J. Pharmacol. Exp. Ther. 1997,
280, 83.
33. The key step is the Ullman coupling reaction of 2-bromo-
5-chloronitrobenzene with 4-chloroanthranilic acid in
amyl alcohol at 140 ꢁC according to Ref. 34. Purification
of lactam 10 was done by simply washing the crude
product with 0.5 N NaOH.
26. The general procedure for the solution-phase library
synthesis is the following: To 10 mg of 4b (0.029 mmol)
in 0.9 mL DCE were added acid chloride (1.2 equiv) and
Et3N (2 equiv). After stirring for 6 h, resin-bound tris-
34. Monro, A. M.; Quinton, R. M.; Wrigley, T. I. J. Med.
Chem. 1963, 6, 255.