A novel series of potent and selective small molecule inhibitors of the complement component C1s

Article abstract of DOI:10.1016/j.bmcl.2004.04.034

Activation of the classical pathway of complement has been implicated in disease states such as hereditary angioedema, ischemia-reperfusion injury and acute transplant rejection. The trypsin-like serine protease C1s represents a pivotal upstream point of

Full text of DOI:10.1016/j.bmcl.2004.04.034

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3043–3047
A novel series of potent and selective small molecule inhibitors of
the complement component C1s
Nalin L. Subasinghe,a,* Farah Ali,a Carl R. Illig,a M. Jonathan Rudolph,a Scott Klein,a
Ehab Khalil,a Richard M. Soll,a Roger F. Bone,a John C. Spurlino,a Renee L. DesJarlais,a
Carl S. Crysler,a Maxwell D. Cummings,a Philip E. Morris, Jr.,b John M. Kilpatrickb
and Y. Sudhakara Babub
a3-Dimensional Pharmaceuticals Inc., 665 Stockton Drive, Exton, PA 19341, USA
bBioCryst Pharmaceuticals, Inc., 2190 Parkway Lake Drive, Birmingham, AL 35244, USA
Received 26 March 2004; revised 13 April 2004; accepted 13 April 2004
Abstract—Activation of the classical pathway of complement has been implicated in disease states such as hereditary angioedema,
ischemia-reperfusion injury and acute transplant rejection. The trypsin-like serine protease C1s represents a pivotal upstream point
of control in the classical pathway of complement activation and is therefore likely to be a useful target in the therapeutic inter-
vention of these disease states. A series of thiopheneamidine-based inhibitors of C1s has been optimized to give a 70 nM inhibitor
that inhibits the classical pathway of complement activation in vitro.
Ó 2004 Elsevier Ltd. All rights reserved.
The complement cascade is a major component of the
innate immune system in mammals and other vertebrate
species.1 Complement activation plays a major role in
the destruction of invading microorganisms and the
clearance of immune complexes. Complement is acti-
vated by three pathways, the classical pathway, the
alternative pathway, and the mannan-binding lectin
(MBL) pathway. The classical pathway is primarily
activated by immunoglobulin (Ig) antigen complexes
binding to C1q, a component of the macromolecular C1
complex that also includes C1s and C1r. Antigen-Ig
complex binding to C1q results in the auto-activation of
C1r, which in turn activates C1s. Activated C1s cleaves
its natural substrates C2 and C4 to give fragments C4b
and C2a, which combine to form the classical pathway
C3 convertase (C4bC2a). The remaining steps in the
activation pathway are common to all three pathways
and are initiated by C3 convertase of the corresponding
pathway cleaving C3 to release the anaphylatoxin C3a.
The remaining C3 fragment (C3b) combines with C3
convertase to form C5 convertase, which cleaves C5 and
releases the potent anaphylatoxin C5a. The resulting
C5b fragment, while still attached to C5 convertase,
sequentially combines with C6 and C7 to form the C5b-
7 complex, which is released into the fluid phase. The
C5b-7 complex inserts into the lipid bilayer of the cell
membrane and combines with C8 and several copies of
C9, to form a transmembrane pore (Membrane Attack
Complex, MAC) resulting in cytolysis.
The alternative pathway is in a state of continuous low-
level activation due to spontaneous hydrolysis of the
thioester residue in native C3, which results in a form of
C3 that can bind to factor Bto produce a C3 convertase
(C3H2OBb).2 Low-level production of C3b is main-
tained by proteolysis of C3 by C3H2OBb. Binding of
C3b to surfaces of pathogens that do not contain com-
plement inhibitors results in the amplification of the
alternative pathway following a series of steps that leads
to the formation of the alternative pathway C3 con-
vertase (C3Bb). The mannan-binding lectin pathway is
activated by mannose-binding lectin (MBL) binding to
carbohydrate polymers on pathogenic surfaces.
Under normal physiological conditions, complement
activation is regulated by a series of membrane-bound
and soluble complement control proteins. Unregulated
complement activation leading to acute inflammation
Keywords: C1s; Complement; Inhibitor; Small; Molecule; Thiophene-
amidine; Amidine; Classical.
* Corresponding author. Tel.: +1-610-458-6066; fax: +1-610-458-8249;
e-mail: nsubasin@prdus.jnj.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.04.034

3044
N. L. Subasinghe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3043–3047
S
and tissue damage has been implicated in the pathology
of many disease states.3 Activation of the classical
pathway has been implicated in humorally mediated
graft rejection,4 ischemia-reperfusion injury (IRI),5
hereditary angioedema (HAE),6 vascular leak syn-
drome,7 and ARDS.8 Treatment with C1 inhibitor (C1-
INH) a plasma serpin inhibitor of C1s has proven
beneficial in animal models of IRI.9 Life-threatening
mucosal edema in HAE, a condition resulting from C1
deficiency, is effectively managed with C1-INH
replacement therapy.10 C1-INH has also proven to be
beneficial in IRI,11 vascular leak syndrome following
bone marrow transplantation12 and acute pancreatits13
in small groups of patients. However, it is difficult to
specifically ascribe the therapeutic effect observed with
C1-INH to classical pathway inhibition due to its ability
to inhibit kallikrein, plasmin, and the MBL pathway
enzyme MASP-2. Even though classical pathway-selec-
tive inhibitors have not entered into clinical use, a
peptidomimetic C1s-selective inhibitor has shown effi-
cacy in a rabbit model of IRI.14
MeS
S
N
HN
NH2
1
Modeling studies with the deposited18 C1s crystal
structure15 and compound 1 indicate that when the
amidine moiety is docked into the S119 pocket such that
it forms a salt bridge with Asp 611(189) of C1s (C1s
numbering, followed by chymotrypsinogen numbering
in parentheses, see also Fig. 1), the thiophene ring
fully occupies the binding pocket. Based on proprietary
structures of similar small molecules bound to
related serine proteases (not shown) this aspect of
our binding model seems unequivocal. The uncer-
tainty of the binding model increases with distance
from S1.
The 5-methylthio substituent on the thiophene ring
occupies a small pocket near the catalytic triad. Of the
substituents tested at this position, the 5-methylthio
group is optimal (Table 1). Removal results in a signif-
icant loss of activity, as does replacement with smaller or
larger hydrophobic groups. The methylthio group was
replaced with electron withdrawing substituents in an
attempt to reduce the pKa of the amidine, and these
compounds had significantly reduced activity as well.
The recently reported high-resolution crystal structure
of C1s15 established that the protease domain adopts the
classic chymotrypsin-like fold.16 C1s exhibits trypsin-
like substrate specificity, and recognition of physiologi-
cal substrates involves an extended recognition surface
comprising elements from multiple modules of the C1
complex. Unfortunately there is currently no C1s
structure available with a bound active-site ligand. Thus,
structural analyses and modeling studies based on the
structural information presently available may only
yield a partial picture of biologically relevant protein–
ligand binding interactions.
Subsequent structure–activity studies were directed to-
ward the thiazole ring. A series of substituted thiazoles
was synthesized and tested in the C1s enzyme assay
(Table 2). Replacing the methyl group at R2 of com-
pound 1 with hydrophobic aromatic groups provides a
5–8-fold improvement in activity (e.g., 11–13, 16, 19,
and 21). Similar substitution at R3 (20) did not change
activity. In general these results are consistent with our
favored binding model, which indicates that substituents
A broad screen of small molecule libraries containing
compounds with basic residues that could potentially
bind to the C1s specificity pocket resulted in the iden-
tification of thiopheneamidine 1 (3.5 lM), a weak
inhibitor of C1s.17
Figure 1. (a) Binding model of 49 in the active site of C1s. This model is derived from the crystal structure of an analog of 49 bound to a C1s
homolog (unpublished results). The partially disordered side chain of Lys614(192)15 has been adjusted to a different rotamer from that in the
deposited structure, to accommodate the modeled inhibitor. Figure shows the C1s crystal structure15 with a Connolly surface24 prepared with
InsightII (Accelrys). (b) Schematic of C1s binding model of 49. Residue names denote key interactions (e.g., Asp611(189)) or residues closest to
indicated moiety of bound inhibitor (e.g., Trp640(215), Glu506(970)). *Residue numbers are according to the deposited C1s crystal structure,15
followed by chymotrypsinogen numbering in parentheses.

N. L. Subasinghe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3043–3047
Table 1. SAR of thiophene C-5 substitution Table 3. SAR of phenyl substitution for compound 19
3045
S
R4
S
R1
S
MeS
S
N
N
R5
HN
R1
HN
NH2
NH2
Compound
Ki, lM
Compound
R4
R5
Ki, lM
2
SCH3
H
0.56
>20
>20
>20
3
22
23
24
25
26
27
28
29
30
31
32
33
4-CH3
4-Cl
H
H
H
H
H
H
H
H
H
0.42
0.436
15.4
1.0
3
4
SCH2CH3
SBnz
4-Phenyl
4-NO2
5
6
CH3CH2
CH3
4-CH3O
2-CH3O
2-CH3
0.9
7
6
0.85
0.85
0.47
0.96
0.64
3.0
8
SO2CH3
CF3
OCH3
>28
12.5
14.4
9
10
3-CH3O
3-CH3
2-CH3O
2-Cl
3-NO2
4-CH3O
4-Cl
4-Cl
0.86
Table 2. SAR of thiazole substitution
R3
substituted ketoester 37 was condensed with N,N-di-
methylaminoformaldehyde dimethylacetal to give the
corresponding enamine 38. Treatment of the enamine 38
with appropriately substituted hydrazine in refluxing
ethanol provided the pyrazole ester 39.21 The pyrazole
ester was hydrolyzed to the acid 40, which was then
converted to the corresponding acid chloride 41. The
acid chloride 41 was treated with trimethylsilyldi-
azomethane to give the corresponding silyldiazoketone,
which was treated with hydrogen bromide in acetic acid
to give the corresponding bromomethylketone 42.17
Several of the bromomethylketones were prepared from
commercially available pyrazole carboxylic acids. The
bromomethylketones required to accesses 3-O-alkyl-
containing pyrazoles (50–52) were synthesized via
Mitsunobu chemistry according to Scheme 3.22 The
commercially available pyrazole 43 was alkylated under
Mitsunobu conditions to give the ether 44. Compound
44 was hydrolyzed to the acid 45 and converted to the
bromomethylketone according to Scheme 2.
S
MeS
R2
N
S
HN
NH2
Compound
R2
R3
Ki, lM
11
12
13
14
15
16
17
18
19
20
21
Benzothiophen-3-yl
2-Naphthyl
Thiophene-2-yl
2-Cl-3-pyridyl
4-Cl-3-pyridyl
PhOCH2
H
H
H
H
H
H
H
H
H
0.45
0.49
0.70
1.02
0.87
0.52
4.7
Cyclohexyl
PhCH2
Ph
1.06
0.56
3.25
0.46
H
N-Methylpyrazol-4-yl
Ph
H
at R3 are likely to be more fully solvent exposed, while
substituents at R2 are more likely to yield binding con-
tacts with the protein (Fig. 1).
In comparison with compound 21, when R6 or R8 is
propyl there is approximately 2-fold enhancement in
activity (Table 4, 46 and 47). Replacement of Me with Ph
at R7 (48) results in a 3-fold increase in activity. Phenyl at
R7 and propyl at R6 (49) gave the most potent compound
in this series. Substituents larger than propyl at R6
attenuate activity (51–52). Substitution of phenyl at R7
(49) with a pyridyl residue (53) provides a compound
with similar potency and marginally improved solubility.
Substitution around the phenyl ring of compound 19 did
not provide a useful SAR (Table 3). 4-Ph substitution
(24) results in a 28-fold loss in activity suggesting a
steric-limit to 4-substitution. Modeling studies were less
useful in understanding the SAR derived from these
analogs, due to the possibility of multiple potential
binding modes for these more distal regions of the
inhibitors.
The C1s selectivity of compound 53 was assessed by
screening against a panel of related serine proteases
(Table 5). Against related complement cascade proteases
Fd and C1r, compound 53 has greater than 53-fold and
700-fold selectivity, respectively. Compound 53 has 70-
fold or greater selectivity over thrombin, plasmin, and
FXa, and also has 3-fold selectivity over trypsin.
A series of analogs around compound 21 was synthe-
sized according to Scheme 1 (Table 4).17 Pyrazolylbro-
momethyl ketone 34 was treated with the thioamide 35
in refluxing acetone to give the corresponding thiazole,
which was converted to the corresponding amidine 36
by treating with chlorodimethylaluminum ammonia
complex.20 Pyrazolylbromomethyl ketones were syn-
thesized according to Scheme 2. An appropriately
Compound 53 was tested in an erythrocyte hemolytic
assay23 and, as expected, proved to be a potent inhibitor

3046
N. L. Subasinghe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3043–3047
O
R8
S
Br
S
S
N
N
S
R8
S
1.
acetone, reflux, 5 h
NH2
R6
R7
S
N
HN
N
34
N
NH2
R6
MeO2C
R7
2. 2M AlMe3, NH4Cl, toluene, reflux 2h
30%
35
36
Scheme 1.
Table 4. SAR of the pyrazole series
R8
S
N
N
MeS
S
N
R7
R6
HN
NH2
Compound
R6
H
R7
R8
Ki, lM
21
46
47
48
49
50
51
52
53
CH3
CH3
CH3
Ph
H
H
0.46
0.33
0.19
0.15
0.06
0.09
1.5
CH3CH2CH2
H
CH3CH2CH2
H
H
H
H
H
H
H
CH3CH2CH2
CH3CH2O
Ph
Ph
(CH3)2CHCH2CH2O
Cyclobutyl-CH2O
CH3CH2CH2
Ph
Ph
0.53
0.07
2-Pyr
O
O
O
O
O
EtO
Me2NCH(OMe)2
R7NHNH2
R6
O
R4
O
anhyd. EtOH
reflux 2.5 h
R6
N
anhyd. EtOH
reflux 1.5 h
50%
N
R7
N
37
75%
38
39
Br
O
O
O
HO
R6
1. TMSCH2N2
4N NaOH
Oxalyl chloride
Cl
R6
MeOH
2. 30% HBr/AcOH
00C to rt, 2h
cat. DMF, CH2Cl2
00C, 2h
R6
N
N
N
N
R7
reflux 2h
N
R7
N
R7
quantitative
80%
quantitative
41
42
40
Scheme 2.
O
O
O
EtO
HO
HO
R9O
EtO
R9O
PPh3, DEAD,
R9OH
67%83%
4N NaOH, MeOH
reflux 4h
N
N
N
N
N
N
Ph
Ph
Ph
43
44
45
Scheme 3.
of the classical pathway of complement activation
(IC50 ¼ 300 nM).
Table 5. Protease selectivity for compound 53
Ki, lM
In conclusion, starting from a weakly active P1 scaffold,
we have constructed a novel series of potent and selec-
tive inhibitors of human C1s. These compounds inhibit
the classical pathway of complement activation in vitro.
C1r
Fd
3.7
Fxa
5.0
Thrombin
>50a
Plasmin
5.0
Trypsin
0.18
>50a
a No observable inhibition at this screening concentration.

N. L. Subasinghe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3043–3047
3047
13. Schneider, D. T.; Nurnberger, W.; Stannigel, H.; Bonig,
H.; Gobel, U. Gut 1999, 45, 733.
14. Buerke, M.; Schwertz, H.; Seitz, W.; Meyer, J.; Darius, H.
J. Immunol. 2001, 167, 5375.
Further results from structure-based optimization and
in vivo biological results will be reported in due course.
15. Gaboriaud, C.; Rossi, V.; Bally, I.; Arlaud, G. J.;
Fontecilla-Camps, J. C. EMBO J. 2000, 19, 1755.
16. Perona, J. J.; Craik, C. S. J. Biol. Chem. 1997, 272,
29987.
References and notes
1. (a) Walport, M. J. N. Engl. J. Med. 2001, 344, 1058; (b)
Walport, M. J. N. Engl. J. Med. 2001, 344, 1140.
2. Pangburn, M. K.; Muller-Eberhard, H. J. Biochem. J.
1986, 235, 723.
3. Kirschfink, M. Immunol. Rev. 2001, 180, 177; Caliezi, C.;
Wuillemin, W. A.; Zeerleder, S.; Redondo, M.; Eisele, B.;
Hack, C. E. Pharmacol. Rev. 2000, 52, 91.
4. Pascual, M. A.; Crespo, M.; Tolkoff-Rubin, N. Nefrologia
2001, 21, 327.
5. D’Ambrosio, A. L.; Pinsky, D. J.; Connolly, E. S. Mol.
Med. 2001, 7, 367.
17. Syntheses of thiazolylthiopheneamidines are described in:
Illig, C. R.; Subasinghe, N. L.; Hoffman, J. B.; Wilson, K.
J.; Rudolph, M. J. Preparation of heteroarylamidines and
analogs as protease inhibitors. PCT Int. Appl. WO0047578,
2000; 326 pp; Chem. Abstr. 2000, 133, 177162.
18. Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.;
Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E.
Nucl. Acids Res. 2000, 28, 235.
19. Schechter, I.; Berger, A. Biochem. Biophys. Res. Commun.
1968, 27, 157.
20. (a) Garigipati, R. S. Tetrahedron Lett. 1990, 31, 1969; (b)
Sidler, D. R.; Lovelace, T. C.; McNamara, J. M.; Reider,
P. J. J. Org. Chem. 1994, 59, 1231.
6. Carugati, A.; Pappalardo, E.; Zingale, L. C.; Cicardi, M.
Mol. Immunol. 2001, 38, 161.
7. Lissoni, P.; Barni, S.; Cattaneo, G.; Archili, C.; Crispino,
S.; Tancini, G.; D’Angelo, L.; Magni, S.; Fiorelli, G. Int.
J. Biol. Markers 1990, 5, 195.
8. Langlois, P. F.; Gawryl, M. S. Clin. Immunol. Immuno-
pathol. 1988, 47, 152.
9. (a) Scherer, M.; Demertzis, S.; Langer, F.; Moritz, A.;
Schafers, H. J. Ann. Thorac. Surg. 2002, 73, 233; (b)
Buerke, M.; Prufer, D.; Dahm, M.; Oelert, H.; Meyer, J.;
Darius, H. J. Pharmacol. Exp. Ther. 1998, 286, 429.
10. Nzeako, U. C.; Frigas, E.; Tremaine, W. J.; Nzeako,
U. C.; Frigas, E.; Tremaine, W. J. Arch. Int. Med. 2001,
161, 2417.
11. de Zwaan, C.; Kleine, A. H.; Diris, J. H.; Glatz, J. F.;
Wellens, H. J.; Strengers, P. F.; Tissing, M.; Hack, C. E.;
van Dieijen-Visser, M. P.; Hermens, W. T. Eur. Heart J.
2002, 23, 1670.
21. Menozzi, G.; Mosti, L.; Schenone, P. J. Heterocycl. Chem.
1987, 27, 1169.
22. Representative procedure: Diethyl azodicarboxylate (Al-
drich, 620 mL, 3.94 mmol) was added dropwise to a 0 °C
solution of 3-hydroxy-1-phenyl-1H-pyrazole-4-carboxylic
acid ethyl ester (Maybridge, 832 mg, 3.58 mmol), triphen-
ylphosphine (Aldrich, 1.03 g, 3.93 mmol), and ethanol
(Aldrich, 200 proof, 230 mL, 3.94 mmol) in THF (anhy-
drous, 14 mL). The reaction mixture was stirred at room
temperature overnight. The solvents were removed in
vacuo to afford a red-colored oil. The crude was purified
by MPLC (SiO2, 10% ethyl acetate in hexanes) to afford
635 mg of 5-ethoxy-1-phenyl-1H-pyrazole-4-carboxylic
acid ethyl ester as a red oil (68% yield).
23. Szalai, A. J.; Digerness, S. B.; Agrawal, A.; Kearney, J. F.;
Bucy, R. P.; Niwas, S.; Kilpatrick, J. M.; Babu, Y. S.;
Volanakis, J. E. J. Immunol. 2000, 164, 463.
12. Nurnberger, W.; Heying, R.; Burdach, S.; Gobel, U. Ann.
Hematol. 1997, 75, 95.
24. Connolly, M. L. J. Mol. Graph. 1993, 11, 139.