Identification of nonpeptidic small-molecule inhibitors of interleukin-2

Article abstract of DOI:10.1016/j.bmcl.2004.12.045

The identification, design, and synthesis of a series of novel sulfamide- and urea-based small-molecule antagonists of the protein-protein interaction IL-2/IL-2Rα are described. Installation of a furan carboxylic acid fragment onto a low-micromolar sulfamide resulted in a 23-fold improvement in activity, providing a sub-micromolar, nonpeptidic IL-2 inhibitor (IC ₅₀ = 0.60 μM).

Full text of DOI:10.1016/j.bmcl.2004.12.045

Bioorganic & Medicinal Chemistry Letters 15 (2005) 983–987
Identification of nonpeptidic small-molecule
inhibitors of interleukin-2
Nathan D. Waal,* Wenjin Yang, Johan D. Oslob, Michelle R. Arkin, Jennifer Hyde,
Wanli Lu, Robert S. McDowell, Chul H. Yu and Brian C. Raimundo
Sunesis Pharmaceuticals, Inc., 341 Oyster Point Boulevard, South San Francisco, CA 94080, USA
Received 28 September 2004; revised 18 November 2004; accepted 15 December 2004
Available online 18 January 2005
Abstract—The identification, design, and synthesis of a series of novel sulfamide- and urea-based small-molecule antagonists of the
protein–protein interaction IL-2/IL-2Ra are described. Installation of a furan carboxylic acid fragment onto a low-micromolar
sulfamide resulted in a 23-fold improvement in activity, providing a sub-micromolar, nonpeptidic IL-2 inhibitor (IC50 = 0.60 lM).
Ó 2005 Elsevier Ltd. All rights reserved.
Binding of the cytokine interleukin-2 (IL-2) to its recep-
tor (IL-2R) leads to the proliferation of activated T-lym-
phocytes and stimulation of the T-helper 1 (Th1)
immune response.1–4 Irregular Th1 immune responses
play a central role in the development of autoimmune
disorders such as rheumatoid arthritis, multiple sclero-
sis, psoriasis, and graft rejection.5,6 Two currently mar-
keted antibodies that specifically target the alpha
subunit of the IL-2 receptor (IL-2Ra) are known to de-
crease the incidence of graft rejection;7,8 however, these
agents lack oral bioavailability and induce side effects
such as hypersensitivity.9 A small-molecule inhibitor of
the protein–protein interaction IL-2/IL-2Ra could
potentially circumvent the disadvantages associated
with these immunosuppressive therapies.
A
previous report describes
a
compound (1,
IC50 = 6 lM) that binds to IL-2, preventing its associa-
tion with IL-2Ra.10 Two components critical for the
activity of 1—the polar guanidine and the hydrophobic
tricyclic ring system—were initially connected via a
dipeptide chain. Optimization of 1 included the installa-
tion of a furanoic acid fragment onto the dichlorophenyl
ring. This fragment, discovered through Tethering,
offered a 30-fold improvement in activity and led to the
discovery of a potent inhibitor (2, IC50 = 0.060 lM).11
Although the dipeptide provided a useful framework
for rapid analog synthesis and SAR development, its
potential in vivo liabilities sparked an effort to identify
alternate linkages. The co-crystal structures12,13 of 1
and
3 (a known small-molecule IL-2 inhibitor;
IC50 = 3 lM)14 bound to IL-2 offered some guidance
into the design of new linkers. The guanidine moiety
of 3 forms two hydrogen bonds with a neighboring glu-
tamic acid E62 (Fig. 1). In contrast, the guanidine moi-
ety of 1 forms only one hydrogen bond with E62, and
appears to adopt a high-energy conformation relative
to 3. The piperidine ring of 3 complements the protein
surface better than the dipeptide, while the cyclohexyl
group of 1 occupies a shallow, hydrophobic pocket
not accessed by 3 (Fig. 2). We reasoned that the ideal
nonpeptidic linker would combine the advantages of 1
and 3. It was thought that alkyl substituted sulfamide
or urea linkers could satisfy the above requirements
Keywords: Fragment assembly; Cytokine; Protein–protein interaction;
IL-2.
*
Corresponding author. Tel.: +1 650 266 3654; fax: +1 650 266
3501; e-mail: nwaal@sunesis.com
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.12.045

984
N. D. Waal et al. / Bioorg. Med. Chem. Lett. 15 (2005) 983–987
Figure 1. Crystal structures of 1 (orange) and 3 (cyan) bound to IL-2.
Glutamic acid (E62) is shown in green.15
Scheme 1. Reaction conditions: (a) SO2Cl2 or triphosgene, Et3N,
CH2Cl2, ꢀ78 °C ! rt, 70–80%; (b) mono-Boc-protected diamine,
Et3N, CH2Cl2; (c) TFA, CH2Cl2; (d) N,N0-bis-Boc-1-guanylpyrazole,
Et3N, MeOH; (e) TFA, CH2Cl2, 23–40% after four steps.
5 and 6 were reacted with various mono-Boc-protected
diamines17 to produce sulfamides and ureas, which were
subjected to a rapid three-step deprotection, guanidinyl-
ation, deprotection sequence. Target compounds were
purified via reverse-phase preparatory HPLC.
In order to determine the preferred atom count between
the piperidine and guanidine moieties, we synthesized a
small set of analogs incorporating alkyl diamine linkers
of varying chain lengths. As shown in Table 1, analogs
bearing a propanediamine linkage (9, 10) are more
active than their ethyl and butyl counterparts (7, 8, 11,
12). This finding is consistent with the atom count
required for activity of the peptidic parent compounds.
In addition, corresponding sulfamides and ureas display
similar SAR and are essentially equipotent. Once the
appropriate chain length was established we focused
our attention on linker substitutions, anticipating that
substituents placed on either the internal guanidine
nitrogen or the adjacent carbon atom would access the
hydrophobic pocket. Results of linker N-alkylation are
shown in Table 2. Methylation of the internal guanidine
Figure 2. Crystal structures of 1 (orange) and 3 (cyan) bound to IL-2.
The piperidinyl ring of 3 better complements the protein surface than
the dipeptide of 1. The cyclohexyl group of 1 sits in a shallow
hydrophobic pocket not accessed by 3.
and be accessed quickly from readily obtainable diamine
precursors.
Table 1. Sulfamide and urea linker length dependence
Compds
X
n
IC50 (lM)a
7
CO
SO2
CO
SO2
CO
SO2
2
2
3
3
4
4
>100
>100
29
Sulfamides and ureas were synthesized in five steps from
intermediate 4 (Scheme 1).16 The requisite sulfamoyl
and carbamoyl chlorides (5, 6) were obtained by treating
4 with sulfuryl chloride or triphosgene, respectively. In
generating 5, slow addition of sulfuryl chloride at
reduced temperature was necessary to suppress chlorina-
tion at the 4-position of the pyrazole ring. Intermediates
8
9
10
11
12
23
>100
>100
a All reported results were obtained using either
Proximity Assay or ELISA.16
a
Scintillation

N. D. Waal et al. / Bioorg. Med. Chem. Lett. 15 (2005) 983–987
Table 4. Representative diaminobutyric acid derivatives
985
Table 2. Linker SAR: N-alkylation
Compds
X
R
IC50 (lM)
Compds
R
IC50 (lM)
9
CO
SO2
CO
SO2
CO
SO2
CO
SO2
H
29
23
27
28
OH
OMe
>100
>100
10
13
14
15
16
17
18
H
Me
Me
Pr
16
14
29
30
31
32
>100
7
>100
32
Pr
i-Pr
i-Pr
>100
48
11
nitrogen (13, 14) leads to a slight improvement in activ-
ity compared with unsubstituted derivatives 9 and 10.
Compounds 15–18 demonstrate that guanidine alkyl
substituents larger than a methyl group are generally
not favored, although this trend is less pronounced with
sulfamides (16, 18). Additional substitutions were made
on the carbon atom adjacent to the guanidine (Table 3).
Corresponding sulfamides and ureas are generally equi-
potent. Increasing the alkyl group size has little effect on
activity (19–24), and the presence of isobutyl groups (23,
24) results in compounds that are roughly equipotent to
the unsubstituted derivatives 9 and 10. Furthermore,
compounds 25 and 26 confirm a strict dependence on
20
33
34
35
60
60
>100
the stereochemistry of linker substituents for activity.
The alkyl groups of 25 and 26 are predicted to point into
the protein surface, likely causing the lack of activity ob-
served for these compounds. Although no increase in
activity is observed relative to 9 and 10, the tolerance
of bulky groups and the stereochemical preference in
this region of the protein is consistent with the dipeptide
SAR. A diaminobutyric acid-derived linker provided a
functional handle for further exploration of the hydro-
phobic pocket (Table 4). Because corresponding sulf-
amides and ureas are equipotent and display similar
SAR trends, we pursued only one series. Polar substitu-
ents lead to inactive compounds (27–29), while more
lipophilic groups result in increased activity (30–32).
Activity diminishes, however, as amide substituents be-
come larger (33–35). With 30, the goal of replacing the
peptidic linker while retaining potency relative to 1
was achieved.
Table 3. Linker SAR: C-alkylation
Compds
X
Linker
IC50 (lM)
9
CO
SO2
29
23
10
19
20
21
22
23
24
25
26
CO
SO2
CO
SO2
CO
SO2
CO
SO2
57
90
55
Finally we wanted to confirm that addition of the fura-
noic acid fragment to this new series would result in a
similar boost in affinity as observed with 2. Sulfamide
14 was chosen as a low-micromolar starting point.
Scheme 2 outlines the synthetic route employed, starting
from intermediates 36 and 38.16 Alkyne 36 was treated
with hydrochloric acid, followed by sulfuryl chloride
and the appropriate Boc-protected diamine to provide
37. Acid chloride 38 was reacted with 37 under palla-
dium-mediated conditions to give an alkynyl ketone,
which was cyclized with methylhydrazine to provide
pyrazole 39 as one regioisomer. Removal of the silyl
protecting group with tetrabutylammonium fluoride
76
23
30
>100
>100

986
N. D. Waal et al. / Bioorg. Med. Chem. Lett. 15 (2005) 983–987
gests that the sulfamide linkage facilitates the desired
bidentate interaction between the guanidine moiety
and E62.
In conclusion, we have identified a nonpeptidic series of
inhibitors of the protein–protein interaction IL-2/IL-
2Ra. The structure–activity relationship was developed
through the evaluation of a variety of linker substitu-
ents, which yielded several low-micromolar hits. Merg-
ing
a low-micromolar sulfamide with a binding
element identified through Tethering resulted in the dis-
covery of a sub-micromolar inhibitor. Further investiga-
tion of this series is currently in progress.
Acknowledgements
The authors wish to thank Stuart Lam, Charles Hatch-
ette, Tom Webb, and Alex Hsi for their support with
RP-HPLC purifications.
References and notes
1. Rubin, L. A.; Nelson, D. L. Ann. Intern. Med. 1990, 113,
619.
2. Nelson, B. H.; Willerford, D. M. Adv. Immunol. 1998, 70,
1.
3. Waldmann, T. A.; Dubois, S.; Tagaya, Y. Immunity 2001,
14, 105.
4. Nelson, B. H. Curr. Dir. Autoimmun. 2002, 5, 92.
5. LeMoine, A.; Goldman, M.; Abramowicz, D. Transplan-
tation 2002, 73, 1373.
Scheme 2. Reaction conditions: (a) 4 N HCl/dioxane, quant.; (b)
SO2Cl2, Et3N, CH2Cl2; (c) N-Boc-N-methyl-1,3-diaminopropane,
Et3N, CH2Cl2, 27% after two steps; (d) CuI, Pd(PPh3)2Cl2, Et3N,
toluene; (e) H2NNHMe, EtOH, 30% after two steps; (f) TBAF, THF,
0 °C; (g) ethyl 5-chloromethyl-2-furancarboxylate, K2CO3, DMF,
55 °C, 22% after two steps; (h) LiOH, THF/H2O, quant.; (i) TFA,
CH2Cl2; (j) N,N0-bis-Boc-1-guanylpyrazole, Et3N, MeOH; (k) TFA,
CH2Cl2, 25% after three steps.
6. Singh, V. K.; Mehrotra, S.; Agarwal, S. S. Immunol. Res.
1999, 20, 147.
7. Berard, J. L.; Velez, R. L.; Freeman, R. B.; Tsunoda,
S. M. Pharmacotherapy 1999, 19, 1127.
8. Waldmann, T. A.; OÕShea, J. Curr. Opin. Immunol. 1998,
10, 507.
9. For information regarding the side effects of basiliximab
and daclizumab, see Letters to Healthcare Providers:
http://www.fda.gov/medwatch/SAFETY/2000/simule.htm
http://www.fda.gov/medwatch/SAFETY/2003/zenapax.htm.
10. Braisted, A. C.; Oslob, J. D.; Delano, W. L.; Hyde, J.;
McDowell, R. S.; Waal, N.; Yu, C.; Arkin, M. R.;
Raimundo, B. C. J. Am. Chem. Soc. 2003, 125,
3714.
gave the phenol, which was alkylated using potassium
carbonate and ethyl 5-chloromethyl-2-furan carboxylate
to afford 40. Saponification of 40 followed by the three-
step guanidinylation sequence mentioned above sup-
plied 41. Addition of the furanoic acid to 14 imparts a
23-fold improvement in activity, resulting in an IC50 of
0.60 lM. A computational model of 41 with IL-2 (Fig.
3) indicates a binding mode in the adaptive region sim-
ilar to that observed for 2. Furthermore, the model sug-
11. Tethering is a registered trademark of Sunesis Pharma-
ceuticals for its fragment-based drug discovery.
12. Thanos, C. D.; Randal, M.; Wells, J. A. J. Am. Chem. Soc.
2003, 125, 15280 (PDB ID 1PW6).
13. Arkin, M. R.; Randal, M.; Delano, W. L.; Hyde, J.;
Luong, T. N.; Oslob, J. D.; Raphael, D. R.; Taylor, L.;
Wang, J.; McDowell, R. S.; Wells, J. A.; Braisted, A. C.
Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 1603 (PDB ID
1M48).
14. Tilley, J. W.; Chen, L.; Fry, D. C.; Emerson, S. D.;
Powers, G. D.; Biondi, D.; Varnell, T.; Trilles, R.;
Guthrie, R.; Mennona, F.; Kaplan, G.; LeMahieu, R.
A.; Carson, M.; Hen, R.-J.; Liu, C.-M.; Palermo, R.; Ju,
G. J. Am. Chem. Soc. 1997, 119, 7589.
15. DeLano, W. L. The PyMOL Molecular Graphics System.Õ
DeLano Scientific LLC: San Carlos, CA, USA. http://
www.pymol.org.
16. Raimundo, B. C.; Oslob, J. D.; Braisted, A. C.; Hyde, J.;
McDowell, R. S.; Randal, M.; Waal, N. D.; Wilkinson, J.;
Yu, C. H.; Arkin, M. A. J. Med. Chem. 2004, 47, 3111.
Figure 3. Computational model of 41 with IL-2.

N. D. Waal et al. / Bioorg. Med. Chem. Lett. 15 (2005) 983–987
987
17. Diamines were prepared according to common litera-
ture methods. Typically, N-protected amino acids were
reduced to amino alcohols, which were mesylated and
displaced using NaCN. Reduction of the cyano
group provided the desired mono-protected propane-
diamines.