Discovery and structure-activity relationships of small molecules that block the human immunoglobulin G-human neonatal Fc receptor (hIgG-hFcRn) protein-protein interaction

Article abstract of DOI:10.1016/j.bmcl.2013.01.014

The neonatal Fc receptor, FcRn, prolongs the half-life of IgG in the serum and represents a potential therapeutic target for the treatment of autoimmune disease. Small molecules that block the protein-protein interactions of human IgG-human FcRn may lower pathogenic autoantibodies and provide effective treatment. A novel class of quinoxalines has been discovered as antagonists of the IgG:FcRn protein-protein interaction through optimization of a hit derived from a virtual ligand-based screen.

Full text of DOI:10.1016/j.bmcl.2013.01.014

Bioorganic & Medicinal Chemistry Letters 23 (2013) 1253–1256
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and structure–activity relationships of small molecules
that block the human immunoglobulin G–human neonatal Fc
receptor (hIgG–hFcRn) protein–protein interaction
à
,§
⇑
Zhaolin Wang , Cara Fraley , Adam R. Mezo
Biogen Idec Hemophilia, 9 Fourth Avenue, Waltham, MA 02451, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The neonatal Fc receptor, FcRn, prolongs the half-life of IgG in the serum and represents a potential ther-
apeutic target for the treatment of autoimmune disease. Small molecules that block the protein–protein
interactions of human IgG–human FcRn may lower pathogenic autoantibodies and provide effective
treatment. A novel class of quinoxalines has been discovered as antagonists of the IgG:FcRn protein–pro-
tein interaction through optimization of a hit derived from a virtual ligand-based screen.
Ó 2013 Elsevier Ltd. All rights reserved.
Received 5 November 2012
Revised 26 December 2012
Accepted 2 January 2013
Available online 11 January 2013
Keywords:
Neonatal Fc receptor
Protein–protein interactions
FcRn
Small molecule
Inhibitor
The neonatal Fc receptor, FcRn, is a 52 kDa heterodimeric glyco-
protein widely expressed in many different tissues and cell types
and is the key regulatory protein for IgG homeostasis in animals
and humans.1 FcRn extends the half-life of IgG in the serum by
binding to IgG in endosomal compartments and shuttling the pro-
tein back to the cell surface for release and recirculation.2 In this
process, FcRn diverts IgG from lysosomal degradation.3 As a result,
IgG has a serum half-life of three weeks in humans while other
proteins of similar size may only have a half-life on the order of
days.
The fact that FcRn can regulate the half-life of IgG has led to the
investigation of this receptor as a potential therapeutic target in
autoimmune disease.4 Reduction of the serum levels of pathogenic
antibodies may alleviate symptoms of diseases in which IgG is the
pathogenic agent. Indeed, it has been shown that FcRn-deficient
mice are protected against models of arthritis and models of vari-
ous skin blistering diseases.5 A monoclonal antibody targeting
FcRn reduced symptoms of experimental autoimmune myasthenia
gravis in rats.6 Recently it was reported that chemically modified
3.4 kDa peptide can bind to human FcRn and block its interaction
with human IgG both in vitro and in vivo.7,8
The inhibition of protein–protein interactions is a major chal-
lenge for small molecular drug discovery because such interactions
generally cover large protein surfaces.9 For example, the rat
Fc:FcRn protein–protein interface is approximately 1870 Å2.10
However, we were encouraged by the fact that a small peptide
inhibitor of FcRn was capable of blocking IgG binding to FcRn using
only 360 Å2 of buried surface area.11 This suggested that only key
FcRn surface ‘hot spots’ need to be targeted for effective inhibition
of the IgG–FcRn interaction.
Here we report our discovery and optimization of small mole-
cules that block the protein–protein interaction between human
IgG (hIgG) and the soluble extracellular domain of human FcRn
(shFcRn). Using the coordinates from a crystal structure of shFcRn
and an antagonist peptide,11 a ligand-based virtual screen of 2.5
million compounds generated a ranked list of virtual hits. The
top 500 compounds were tested in an IgG–FcRn competition assay
(vide infra) to identify compounds that could interfere with this
protein–protein interaction. Using this methodology, we identified
quinoxaline hit compound 1 (Fig. 1) with an IC50 of more than
⇑
Corresponding author. Tel.: +1 317 433 6320.
150 lM.
E-mail address: mezo_adam_r@lilly.com (A.R. Mezo).
Current address: Ra Pharmaceuticals, One Kendall Square, Suite B14301,
Cambridge, MA 02139, USA.
The synthesis of various quinoxaline analogs (2) is depicted in
Scheme 1. In general, the condensation of R1-substituted 2,3-
dichloroquinoxaline 3 with a various cyanoacetamides 4 generated
5. Displacement of the chlorine atom of 5 with various amines pro-
vided the modified quinoxaline analogs 2. Cyanoacetamides 4 were
à
Current address: Broad Institute, 7 Cambridge Center, Cambridge, MA 02141,
USA.
§
Current address: Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN
46225, USA.
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.01.014

1254
Z. Wang et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1253–1256
more active than hit compound 1. In contrast, methoxy substitu-
tions on the benzyl ring exemplified by compounds 28 and 29 re-
sulted in a reduction in activity as compared to the phenyl
substituted compounds 21 and 22. The activity of disubstituted
methoxy phenyl compound 27 is much lower than the correspond-
ing mono-substituted compounds. Interestingly, the conversion of
a methoxy group in compound 22 to a hydroxyl group (23) abol-
ished its activity demonstrating selectivity for the methoxy
substituent.
N
N
N
CN
O
N
H
O
1
μΜ
IC50 >150
A preliminary SAR study of compound 21 using compounds in
Figure 2 indicated the important role of the amide NH and the
quinoxaline ring. Methylation of the amide NH resulted in no mea-
Figure 1. Chemical structure of quinoxaline hit compound 1.
generally obtained by coupling of commercially available cyano-
acetic acid with various amines using PyBOP and standard condi-
tions. In cases where the substituted 2,3-dichloroquinoxaline 3
was not commercially available, the preparation of the requisite
R1-substituted quinoxalines is described in Scheme 2. Condensa-
tion of R1-substituted benzene-1,2-diamine (6) with diethyl oxa-
late (7) gave the corresponding quinoxaline dione 8. Reflux of 8
in phosphorus oxychloride generated the various 2,3-dichloroqui-
noxaline analogs 3.
The ability of these analogs to block hIgG binding to shFcRn was
assessed using a FcRn–IgG competition ELISA assay as described
previously.7 Briefly, various concentrations of small molecules
were each mixed with 3 nM hIgG and incubated with shFcRn-
coated plates. The concentration required to inhibit 50% of the
surable activity up to 500
larly, inclusion of smaller pyrazine ring instead of the
quinoxaline ring in compound 33 also showed no activity up to
500 M.
lM as shown by compounds 32. Simi-
a
l
With the optimized methoxy phenyl compound 21, we next
turned our attention to examining the effect of substitution at
the 6 or 7 positions on the quinoxaline ring, as shown by com-
pounds 35–38 (Table 2). Both electron withdrawing and donating
groups were attached to the quinoxaline ring in compound 21
but both changes resulted in reductions in activity. Further exten-
sion from the 6 position of the quinoxaline ring by an amide link-
age, as shown in compounds 39–42, also resulted in a loss of
activity. 6,7-Dichloro-quinoxaline analog 38 showed only a small
threefold reduction in activity. This may suggest that only small
substitutions at the 6 and 7 positions on the quinoxaline ring are
somewhat tolerated.
We also investigated ring size and modifications at the azepane
moiety (Table 3). For example, the 7-membered ring analogs 43–
48 were aimed at modulating the activity and solubility but were
inactive. However, compound 50 with an 8-membered alkyl ring
showed similar activity as compound 21. Interestingly, 6-mem-
bered piperidine analogs 51–53 were also inactive. Connection to
the quinoxaline of various phenyl or alkyl groups by either a sec-
ondary amine or thioether (compounds 54–58) also abolished
the activity. Our SAR suggested that the unsubstituted 7 and 8
membered-alkyl rings azepane and azocane provide optimal
activity.
IgG–FcRn interaction was determined and reported as its IC50
.
Starting from the initial hit 1, we first explored the amide posi-
tion and the effect of various alkyl groups on activity (9–13, Table
1). The alkyl ether chain of compound 9 with one less methylene
had similar activity to 1 while compounds with other alkyl groups
(10–12) showed no activity which suggested a possible role for the
methoxy functional group of compound 1. Next we sought to re-
place the alkyl ether group with aryl groups as illustrated in Table
1. The unsubstituted phenyl and heteroaryl analogs (13–17)
showed no activity. However, introduction of ether a para or a
meta-substituent on the phenyl ring had significant effects on the
activity. For example, the 3-F-substituted phenyl substituent (18)
possessed an IC50 of 150
original hit compound 1. The 4-meoxyphenyl compound 22 in-
creased the activity to 50 M, which is threefold higher than com-
pound 1. Notably, the 3-methoxy phenyl compound 21 was 15-fold
lM and recapitulated the activity of the
Additional optimization efforts were directed towards
extension of the methoxy phenyl ring which proved active in
compounds 21 and 22 (Table 4). Substitutions extended from
l
H
N
N
N
Cl
Cl
N
N
Cl
a
b
R2
R1
R1
NC
+
CN
O
R2
3
4
O
N
H
N
R2
5
R1
CN
N
R2
O
N
H
2
Scheme 1. Reagents and conditions: (a) t-BuOK, THF, reflux, 3 h (60–71%); (b) amines, Et3N, iPrOH, 80 °C, 4 h (80–90%).
O
H
N
NH2
NH2
O
O
N
N
Cl
Cl
b
a
O
R1
R1
+
R1
O
O
N
H
8
7
6
3
Scheme 2. Reagents and conditions: (a) EtOH, reflux, 12 h (80–90%); (b) POCl3, reflux, 12 h (90–95%).

Z. Wang et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1253–1256
1255
Table 1
Table 3
SAR of the amide substituent of compound 1
SAR of ring size
N
N
R
N
N
N
CN
CN
O
O
N
H
O
N
H
R
R
IC50
M)
R
IC50
M)
(
l
(
l
Entry
R
IC50 (lM)
1
9
Methoxyethyl
Methoxymethyl
Allyl
n-Propyl
Cyclohexanyl
Ph
2-Pyridine
3-Pyridine
2-Furan
2-Thiophen
3-F-Ph
>150
120
N
N
N
N
43
51
52
>500
>500
>500
>500
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
>250
>500
>500
>500
>1000
>1000
>1000
>1000
150
100
80
10
50
>500
40
>500
150
250
100
120
>500
>500
O
N
N
44
45
46
O
N
NH
53
54
>500
>500
>500
>500
N
3-Cl-Ph
N
HN
3-CF3O-Ph
3-MeO-Ph
4-MeO-Ph
4-OH-Ph
N
O
NH
4-CN-Ph
N
47
48
55
56
4-N(Me)2-Ph
2-MeO-Ph
3,4-(MeO)2-Ph
3-MeO-Bn
4-MeO-Bn
4-Amino-Bn
4-Acetamide-Bn
>500
>500
>500
>500
N
OH
S
N
N
O
O
H
N
>500
6
>500
>500
49
50
57
58
N
N
N
N
N
N
N
N
CN
H
CN
N
N
O
O
O
N
H
O
33
IC50 > 500 μM
32
IC50 > 500 μM
two isomers, 64 and 65, with the same activity indicating that the
stereochemistry at the benzyl position had little effect on activity.
Addition of a methyl group in this position also resulted in a minor
improvement in activity as compared to compound 22. Encouraged
by these findings, we re-positioned the solubilizing morpholine
group to the benzyl carbon to generate compound 66 which im-
Figure 2. Chemical structures of compounds 32 and 33.
meta-position (59–62) of the phenyl ring resulted in a reduction in
activity. Addition of a morpholine group at the para-methoxy
phenyl (63) also reduced the activity. From our limited data set,
it appears that only a small methoxy group is tolerated at the meta
and para-position of the phenyl ring in compounds 21 and 22.
Introduction of a methyl group to the benzyl group in 22 produced
proved the IC50 to 2 lM (Fig. 3). This represents a 25-fold increase
in activity as compared to compound 22, and at least a 75-fold in-
crease in activity from the original hit compound 1. It should be
noted from Figure 3 that the Hill slopes of compounds 1, 21 and
Table 2
SAR of quinoxaline 6 and 7 positions
O
R3
N
N
N
6
R2
7
N
N
N
N
H
CN
CN
O
O
O
N
H
O
N
H
Entry
R2
IC50
100
500
>500
30
(lM)
Entry
R3
IC50 (lM)
35
36
37
38
6-CN
39
40
41
42
Ph
>500
>500
>500
400
6-NO2
7-OMe
6,7-Cl, Cl
Pyridin-3ylmethyl
Pyridin-3ylethyl
Dimethylaminopropyl

1256
Z. Wang et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1253–1256
Table 4
SAR of phenyl ring substitutions
N
N
N
CN
R
O
N
H
R
IC50
40
(l
M)
R
IC50 (lM)
O
O
N
59
63
N
>500
40
O
S
60
64
65
O
200
400
N
N
O
61
O
40
N
O
O
62
66
O
200
2
N
O
N
protein–protein interaction. Chemical optimization of the initial
hit compound 1 to compound 66 resulted in a 75-fold improve-
ment in in vitro activity. This family of compounds may serve as
useful tools in the study of FcRn biology, as well as starting points
for the further development of orally available small molecule
inhibitors of FcRn for therapeutic use.
100
80
60
40
20
0
Acknowledgment
We thank Alan Bitonti for helpful discussions and Holly Palmi-
eri for technical assistance.
SYN1327
Compound 1
Compound 21
Compound 66
References and notes
1. Roopenian, D. C.; Akilesh, S. Nat. Rev. Immunol. 2007, 7, 715.
2. Ober, R. J.; Martinez,C.;Vaccaro, C.;Zhou,J.;Ward, E. S.J. Immunol. 2004, 172, 2021.
3. Ober, R. J.; Martinez, C.; Lai, X.; Zhou, J.; Ward, E. S. Proc. Natl. Acad. Sci. U.S.A.
2004, 101, 11076.
0.01
0.1
1
10
100
Concentration of Compound, μM
4. Yu, Z.; Lennon, V. A. N. Eng. J. Med. 1999, 340, 227.
5. (a) Akilesh, S.; Petkova, S.; Sproule, T. J.; Shaffer, D. J.; Christianson, G. J.;
Roopinean, D. C. J. Clin. Invest. 2004, 113, 1328; (b) Li, N.; Zhao, M.; Hilario-
Vargas, J.; Prisayanh, P.; Warren, S.; Diaz, L. A.; Roopinean, D. C. J. Clin. Invest.
2005, 115, 3440.
Figure 3. Representative inhibition curves for compounds 1, 21, 66 and control
peptide SYN13277 using the FcRn–IgG competition ELISA assay.
6. Lui, L.; Garcia, A. M.; Santoro, H.; Zhang, Y.; McDonnell, K.; Dumont, J.; Bitonti,
A. J. Immunol. 2007, 178, 5390.
7. Mezo, A. R.; McDonnell, K. A.; Tan Hehir, C. A.; Low, S. C.; Palombella, V. J.;
Stattel, J. M.; Kamphaus, G. D.; Fraley, C.; Zhang, Y.; Dumont, J. A.; Bitonti, A. J.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 2337.
66 vary among themselves despite their similar chemical struc-
tures. In addition, the slopes vary in comparison to control peptide
SYN1327 when assayed in the same experiment. This phenomenon
suggests some form of negative cooperativity for the small mole-
cule binding, or heterogeneous binding and is not fully understood.
Further work will be required to elucidate the cause of this effect
and their mode of binding and inhibition.
8. McDonnell, K. A.; Low, S. C.; Hoehn, T.; Donnelly, R.; Palmieri, H.; Fraley, C.;
Sakorafas, P.; Mezo, A. R. J. Med. Chem. 2010, 53, 1587.
9. For example: (a) Garner, A. L.; Janda, K. D. Curr. Top. Med. Chem. 2011, 11, 258;
(b) Whitby, L. R.; Boger, D. L. Acc. Chem. Res. 2012, 45, 1698.
10. Martin, W. L.; West, A. P., Jr.; Gan, L.; Bjorkman, P. J. Mol. Cell. 2001, 7, 867.
11. Mezo, A. R.; Sridhar, V.; Badger, J.; Sakorafas, P.; Nienaber, V. J. Biol. Chem. 2010,
285, 27694.
In summary,
a series of novel quinoxaline analogs were
prepared and found to inhibit the human IgG:human FcRn