Symmetry-based ligand design and evaluation of small molecule inhibitors of programmed cell death-1/programmed death-ligand 1 interaction

Article abstract of DOI:10.1016/j.bmcl.2019.07.027

The development of small molecule inhibitors of PD-1/PD-L1 is eagerly anticipated for treatment of cancer. We focused on the symmetry of the ternary complex structure of reported small molecule ligands and hPD-L1 homodimers, and designed partially- or fully-symmetric compounds for more potent inhibitors. The design of the new compounds was guided by our hypothesis that the designed symmetric compound would induce a flip of sidechain of _ATyr56 protein residue to form a new cavity. The designed compound 4 exhibited substantially increased binding affinity to hPD-L1, as well as PD-1/PD-L1 inhibitory activity in physiological conditions. Compound 4 also showed a dose-dependent increase in IFN-γ secretion levels in a mixed lymphocyte reaction assay. These results not only indicate the feasibility of targeting the PD-1/PD-L1 pathway with small molecules, but illustrate the applicability of the symmetry-based ligand design as an attractive methodology for targeting protein-protein interaction stabilizers.

Full text of DOI:10.1016/j.bmcl.2019.07.027

Accepted Manuscript
Symmetry-based ligand design and evaluation of small molecule inhibitors of
programmed cell death-1/programmed death-ligand 1 interaction
Seiji Kawashita, Koichi Aoyagi, Hiroshi Yamanaka, Rie Hantani, Shiori
Naruoka, Atsuo Tanimoto, Yuji Hori, Yukiyo Toyonaga, Kyoko Fukushima,
Susumu Miyazaki, Yoshiji Hantani
PII:
S0960-894X(19)30472-X
https://doi.org/10.1016/j.bmcl.2019.07.027
BMCL 26568
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
12 June 2019
13 July 2019
16 July 2019
Please cite this article as: Kawashita, S., Aoyagi, K., Yamanaka, H., Hantani, R., Naruoka, S., Tanimoto, A., Hori,
Y., Toyonaga, Y., Fukushima, K., Miyazaki, S., Hantani, Y., Symmetry-based ligand design and evaluation of small
molecule inhibitors of programmed cell death-1/programmed death-ligand 1 interaction, Bioorganic & Medicinal
Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.07.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Symmetry-based ligand design and evaluation of small molecule inhibitors of programmed cell
death-1/programmed death-ligand 1 interaction
a,
a
a
b
b
Seiji Kawashita *, Koichi Aoyagi , Hiroshi Yamanaka , Rie Hantani , Shiori Naruoka , Atsuo
Tanimoto b, Yuji Hori b, Yukiyo Toyonaga b, Kyoko, Fukushima a, Susumu Miyazaki a, and Yoshiji
Hantani b,*.
b
aChemical Research Laboratories, Biological Pharmacological Research Laboratories, Central
Pharmaceutical Research Institute, Japan Tobacco Inc., 1-1, Murasaki-cho, Takatsuki, Osaka 569-
1125, Japan
*Corresponding author:
Biological/Pharmacological Research Laboratories, Central Pharmaceutical Research Institute, Japan
Tobacco Inc., 1-1 Murasaki-cho, Takatsuki, Osaka 569-1125, Japan
Tel: +81-72-681-9700; Fax: +81-72-681-9725
E-mail: seiji.kawashita@jt.com
E-mail: yoshiji.hantani@jt.com
Abstract:
The development of small molecule inhibitors of PD-1/PD-L1 is eagerly anticipated for treatment of
cancer. We focused on the symmetry of the ternary complex structure of reported small molecule
ligands and hPD-L1 homodimers, and designed partially- or fully-symmetric compounds for more
potent inhibitors. The design of the new compounds was guided by our hypothesis that the designed
symmetric compound would induce a flip of sidechain of Tyr56 protein residue to form a new
A
cavity. The designed compound 4 exhibited substantially increased binding affinity to hPD-L1, as
well as PD-1/PD-L1 inhibitory activity in physiological conditions. Compound 4 also showed a
dose-dependent increase in IFN-γ secretion levels in a mixed lymphocyte reaction assay. These
results not only indicate the feasibility of targeting the PD-1/PD-L1 pathway with small molecules,
but illustrate the applicability of the symmetry-based ligand design as an attractive methodology for
targeting protein-protein interaction stabilizers.
Keywords: PD-1, PD-L1, Small-molecules, Immune checkpoint inhibitor, Surface plasmon
resonance, Drug design
1

The PD-1/PD-L1 (programmed death protein 1/programmed death-ligand 1) pathway has
drawn intense research interest since monoclonal-antibodies (mAbs) targeting PD-1 or PD-L1
1, 2, 3
achieved great successes in the clinic for the treatment of cancer.
Although mAbs targeting
PD-1 and PD-L1 gave robust clinical outcomes, there is still room for improvement. For example,
immunogenicity, immune-related adverse effects as a result of a long half-life, pharmacokinetics and
4
high target occupancy, limited routes for administration, and high cost are major shortcomings in
mAbs. On the other hand, small molecule inhibitors are capable of overcoming some of these
disadvantages such as immunogenicity, lack of oral bioavailability, and high cost. Therefore, the
4, 5,
6
advent of small molecule inhibitors of PD-1/PD-L1 continues to be eagerly anticipated.
Recently, a series of small molecule PD-1/PD-L1 inhibitors (1–2) were disclosed in a patent
application (Figure 1).7 Later Zak and coworkers revealed in their crystallographic and biochemical
studies that these biaryl compounds induce and stabilize dimerization of PD-L1s, and consequently
inhibit PD-1/PD-L1 interaction. 8
Small molecule protein-protein interaction (PPI) stabilizers have been recently highlighted in
the research field as an opposing approach for modulating PPI with respect to its inhibition. 9, 10, 11, 12
13, 14
According to some reviews,
the majority of stabilized PPI complexes have simple structural
features, and the main feature is their symmetry in the complex. Indeed, most stabilizers were
observed in homomers and quasi homomers, and two protein monomers were virtually stabilized by
single symmetric or pseudo-symmetric compounds. For example, a class of inhibitors of the
15, 16
bromodomain and extraterminal (BET) proteins,
positive allosteric modulators of the AMPA
17, 18
19
receptor,
and the Toll-like Receptor agonist Diprovocim were identified as symmetric or
pseudo-symmetric compounds. These examples exploit symmetry-based ligand design for
discoveries of potent multivalent ligands.
Motivated by reports of compounds (1–2) that form a pseudo-C2-symmetric complex with hPD-
L1s and stabilize PD-L1 homodimer, we undertook the study of symmetric ligands for more potent
PD-1/PD-L1 inhibitors.
HO
N
O
O
H
N
Br
N
N
H
O
O
O
1
2
(BMS-202)
IC50 = 18 nM
(BMS-8)
IC50 = 146 nM
Figure 1 Structures of the reported small molecule inhibitors of PD-1/PD-L1.
Names and IC50 values reported in the patent are shown.
Figure 2 illustrates the symmetry-based ligand design in this work. We selected compound 3
20
(BMS-1327, Figure 2) as a template compound to bring the concept of symmetry-based ligand
2

design into practice. Compound 3 bears the (5-cyanopyridin-3-yl)methoxy group which appeared to
give high potency in our preliminary results in a cell-free assay (data not shown). The binding mode
of 3 is conjectured to arise from the ternary complex of 1 with hPD-L1s (PDB ID: 5J89), which is a
pseudo-symmetrical structure where the axis of symmetry is positioned at the center of the biphenyl
substructure. The major structural difference between the two protein monomers is the positions of
the side chain phenyl rings of both Tyr56 residues. The phenyl ring of ATyr56 is oriented toward the
surface of the chain B, creating a hydrophobic wall close to one side of the cleft between the two
proteins. In contrast, phenyl ring of Tyr56 is lying in a direction along the protein surface while
B
forming a π-π interaction with the methoxypyridine ring of 1.
On applying the concept of symmetry-based ligand design to compound 3, the aforementioned
hydrophobic wall formed by Tyr56 should be considered as a potential obstacle. In several reports
A
21, 22, 23
in the literature,
crystal structures of hPD-L1, including apo-hPD-L1 (PDB ID: 5C3T) and
the hPD-1/hPD-L1 complex (PDB ID: 4ZQK), 23 have been reported. The position of ATyr56 in the
ternary complex (5J89) was different from that of the two structures apo-PD-L1 and in a complex
with PD-1. Based on these facts, we hypothesized that the side chain of Tyr56 is sufficiently flexible
and capable of taking both configurations. Thus, we designed dimeric ligand 4 with the aim of
inducing a shift of ATyr56 to create a new cavity inside a large interface, while the other half of the
compound would bury into the newly generated space in the same way as the original monomeric
ligand does. Indeed, during our research, this assumption was experimentally confirmed with the co-
20
crystal structure of analogue of 1 with PD-L1s (BMS-1001, PDB ID: 5NIU 24), in which the 2,3-
dihydro-1,4-benzodioxinyl group induced a shift in the position of ATyr56 reported by Guzik et al. 25
CN
N
Axis of
Symmetry
CN
Axis of
Symmetry
N
HO
ATyr56
O
HO
OH
HO
N
H
Dimerization
O
A2
B1
O
N
O
OH
O
O
H
N
N
H
A1
B2
O
A2
B1
O
HO
OH
HO
O
A1
ATyr56
OH
BTyr56
OH
3
4
BTyr56
NC
Figure 2 Schematic representation of symmetry-based ligand design
In addition to the biphenyl-core group (A1 and A2) which forms a T-shaped π-π stacking with
ATyr56 and CH-π interactions with AMet115 and Ala121, three other structural components of
B
compound 3 were taken into account in design of the compounds: 1) the methyl group on the A1
ring which is buried in the lipophilic cavity surrounded by lipophilic residues; 2) an aromatic ring B1
3

which forms π-π interaction with the phenyl ring of Tyr56; and 3) a terminal amino acid moiety
B
which forms multiple hydrophilic interactions such as water-mediated or direct hydrogen bonds and
electrostatic interactions with both chains. For the purpose of elucidating the contribution of each
substructure in compound 3, both the fully dimerized compound (4) and partially dimerized
compounds 5–9 (Table 1) were designed and synthesized.
4

Table 1 Binding affinities and PD-1/PD-L1 inhibitory activities of designed compounds
CN
N
Axis of
Symmetry
HO
O
OH
N
R
H
O
O
Entry
R
KD (nM)a EC50 (µM)b
3
1.1
3.0
> 30
NT
(BMS-1327)
5
6
O
H
O
3.7
> 100
N
NC
O
O
O
H
N
7
8
0.29
0.21
NT
9.5
HO
HO
OH
O
O
H
N
O
O
OH
O
H
N
HO
9
4
0.20
8.4
1.0
OH
N
O
NC
O
H
N
HO
0.019
O
OH
N
NC
a
b
Binding affinities (KD) were determined by Surface Plasmon Resonance analysis. PD-1/PD-L1
inhibitory activities were measured by blockade bioassay (for details see Supporting Information).
20
Compounds 3 and 5 were synthesized according to the method described in the patent
.
Scheme 1 shows the general synthetic route for designed compounds 4 and 6–9. 2,4-
5

Dihydroxybenzaldehyde 10 was used as a starting material. After protection of the 4-hydroxy moiety
of 10 with THP, the 2-hydroxy group was converted to the (5-cyano-3-pyridine)methoxy group, and
subsequent deprotection gave phenol 11. Alkylation of phenol 11 with benzyl chloride 12 afforded
key intermediate 13. Similarly, intermediate 17, 21, 24, and 27 were prepared in one or two steps
from 4-hydroxybenzaldehyde 15, 4-hydroxy-2-methoxybenzaldehyde 19, phenol 11, or resorcinol 26,
respectively. Each halogenated intermediate (13, 17, 21, 24, and 27) was subjected to palladium-
catalyzed borylation to give aryl boranes (14, 18, 22, 25, and 28). Formation of dimeric biphenyl
core products was conducted with halogenated product 13 and aryl boranes (14, 18, 22, 25, and 28)
via palladium-catalyzed Suzuki coupling reaction to give 29a–d and 30. Reductive amination of
mono- or bis-aldehyde product 29a–d and 30 with D-serine finally afforded compounds 4 and 6–9.
CN
CN
CN
N
O
Br
N
N
Cl
O
O
12
O
OH
CHO
CHO
CHO
CHO
a, b, c
13 + 14
13 + 18
13 + 22
13 + 25
d
h
i
O
X
O
HO
HO
4, 7–9
Ra
OHC
Rb
10
11
13
14
: X = Br
: X = Bpin
e
29a
29b
29c
29d
(Rb = (5-cyanopyridin-3-yl)methoxy, Ra = Me)
(Rb = H, Ra = Me)
(Rb = methoxy, Ra = Me)
CHO
(Rb = (5-cyanopyridin-3-yl)methoxy, Ra = H)
12
CHO
X
O
d
CN
N
HO
17
18
: X = Br
: X = Bpin
15
e
O
CHO
i
h
O
6
O
Br
13 + 28
OH
O
CHO
O
O
20
X
CHO
O
f
N
30
HO
NC
21
22
: X = Br
: X = Bpin
19
e
CN
N
I
OH
O
23
CHO
X
f
O
11
24
25
: X = I
: X = Bpin
e
CN
N
12
g, b
OH
O
HO
X
O
26
27
28
: X = Br
: X = Bpin
e
Scheme 1 Synthetic route of dimeric compounds 4 and 6–9.
a) DHP, PPTS, CH2Cl2, rt. b) 5-Chloromethylnicotinonitrile, K2CO3, DMF, rt. c) p-TsOH, MeOH-
H2O, rt. d) K2CO3, DMF, rt. e) Bis(pinacolato)diboron, AcOK, Pd2(dba)3, Pcy3, Dioxane, 100 °C. f)
DIAD, PPh3, THF, rt. g) NaH, DMF, rt. h) method A: Pd(PPh3)4, Na2CO3, CPME/H2O, 100 °C.,
6

method B: Sphos Pd G1, K3PO4, DMF, 80 °C. i) D-Serine, AcOH, NaBH(OAc)3, DMF or DMSO, rt.
Binding affinities of 3–9 against hPD-L1 were evaluated with surface plasmon resonance (SPR).
As shown in Table 1， it is evident that there are marked increases in binding affinity with
compounds bearing two amino acid moieties (4 and 7–9), while partially-dimerized compounds such
as 5 or 6 did not show significant increases in KD from monomeric ligand 3. Additionally, compound
4 exhibited 50 times smaller KD as compared to 3. Compounds 7–8, bearing single replacements of
the (5-cyanopyridin-3-yl)methoxy group of 4 with hydrogen or methoxy group, also showed
subnanomolar binding affinity. However, 10-fold decreases in KD from 4 were observed in both
compounds. Similarly, compound 9, with the single displacement of the methyl substituent on the
phenyl ring A2 also resulted in 10-fold drop in KD from 4. These results revealed that terminal
amino acids groups have a great impact on binding affinity. The (5-cyano-3-pyridine)methoxy group
on ring B and the methyl substituent on the ring A also appeared to contribute significantly in
binding to hPD-L1. To investigate whether compound 4 induces the PD-L1 dimerization in solution,
a size-exclusion chromatography was conducted in the presence and absence of 4. The retention time
of PD-L1 was significantly shortened in the presence of 4 and the change in retention time was
indicative of dimer formation (Supplementary Figure 1).
PD-1/PD-L1 inhibitory activities were evaluated with a blockade bioassay (Table 1). Monomer
or partially-dimerized compound 3 and 6 did not show EC50 values of less than 10 μM, while anti-
PD-1 antibody as a positive control exhibited EC50 of 5.7-14 nM. In contrast, fully-dimerized
compounds 4, 8, and 9 showed significantly lower EC50 values. Especially, compound 4, EC50 = 1.0
μM, showed the highest potency among all compounds tested. We were able to show the efficacy of
dimerized compounds in the cell assay, but the EC50 was approximately 40,000 ~ 50,000-fold less
potent than KD values (Table 1). Since non-specific binding of compounds to serum protein was
considered as the reason of the discrepancy, cell-free PD-1/PD-L1 inhibitory activities of several
compounds were measured in the presence or absence of serum protein. However, no remarkable
effect of serum protein on inhibitory activity was observed (data not shown). Therefore, the reason
for the discrepancy is still unclear, but may result from non-specific binding to other factors such as
cell surface protein, cell membrane.
For further evaluation of the most potent compound 4, cellular function was evaluated with a
26
mixed lymphocyte reaction assay (MLR assay).
tested. Results are presented in Figure 3.
For comparison, anti-PD-1 antibody was also
7

Figure 3 MLR assay
CD4+ T cells were cultured for 7 days alone or in the presence of monocyte-derived dendritic cells
(DCs) with or without anti-PD-1 or compound 4. IFN-γ secretion in culture supernatants was
determined as an indicator of cell activation. The graphs present the mean ± SD from experiments
performed in triplicate.
As shown in Figure 3, compound 4 exhibited a dose-dependent increase in IFN-γ secretion
levels. At 100 nM, 4 showed similar levels of IFN-γ as compared to the maximum efficacy of anti-
PD-1 antibody exhibited. From these results, compound 4 appeared to function as a promising small
molecule PD-1/PD-L1 inhibitor to restore T-cell function. However, there is a large gap in potency
between 4 and anti-PD-1 antibody. Thus, further optimization and exploration remains for future
work.
In summary, we designed novel PD-1/PD-L1 inhibitors bearing a C2-symmetric scaffold. Our
ligand design is unique in terms of the consideration of both symmetricity of homodimer and a
potential shift of a protein residue which was projected from co-crystal structures. Designed
compound 4 induced dimerization of PD-L1, and showed significantly increased binding affinity to
hPD-L1, as well as inhibitory activity of PD-1/PD-L1 interaction under physiological conditions
compared to the original monomeric ligand. The most potent compound 4 also exhibited a dose-
dependent increase in IFN-γ production in the MLR assay. To our knowledge, this is the first
example of small molecule PD-1/PD-L1 inhibitors that verify cellular function with the MLR assay.
Therefore, we believe our results will lead to the development of small-molecule modulators based
8

on dimeric scaffold targeting the PD-1/PD-L1 pathway. Moreover, this work illustrates the
applicability of the symmetry-based ligand design as an attractive methodology for targeting protein-
protein interaction stabilizers.
Acknowledgments:
We thank Dr. Makoto Shiozaki for helpful discussions and Ms. Aoi Horiike for SPR assays. We
also appreciate Dr. Hiromasa Hashimoto for suggestions on writing this manuscript.
References and notes:
1.
Alsaab HO, Sau S, Alzhrani R, et al. PD-1 and PD-L1 Checkpoint Signaling Inhibition for
Cancer Immunotherapy: Mechanism, Combinations, and Clinical Outcome. Frontiers in
pharmacology. 2017;8: 561.
2.
Targeted Immunotherapy in Tumor. Journal of Cancer. 2017;8(3): 410-416.
3. Herbst RS, Soria JC, Kowanetz M, et al. Predictive correlates of response to the anti-PD-
L1 antibody MPDL3280A in cancer patients. Nature. 2014;515(7528): 563-567.
4. Konstantinidou M, Zarganes-Tzitzikas T, Magiera-Mularz K, Holak TA, Domling A.
Guo L, Zhang H, Chen B. Nivolumab as Programmed Death-1 (PD-1) Inhibitor for
Immune Checkpoint PD-1/PD-L1: Is There Life Beyond Antibodies? Angewandte Chemie
(International ed in English). 2018;57(18): 4840-4848.
5.
Zhan MM, Hu XQ, Liu XX, Ruan BF, Xu J, Liao C. From monoclonal antibodies to small
molecules: the development of inhibitors targeting the PD-1/PD-L1 pathway. Drug discovery today.
2016;21(6): 1027-1036.
6.
Wang T, Wu X, Guo C, et al. Development of Inhibitors of the Programmed Cell Death-
1/Programmed Cell Death-Ligand 1 Signaling Pathway. Journal of medicinal chemistry. 2019;62(4):
1715-1730.
7.
Organization. 2015;WO Patent 2015/034820 A1.
8. Zak KM, Grudnik P, Guzik K, et al. Structural basis for small molecule targeting of the
programmed death ligand 1 (PD-L1). Oncotarget. 2016;7(21): 30323-30335.
9. Thiel P, Kaiser M, Ottmann C. Small-molecule stabilization of protein-protein
Chupak LS ZX. Compounds useful as immunomodulators. World Intellectual Property
interactions: an underestimated concept in drug discovery? Angewandte Chemie (International ed in
English). 2012;51(9): 2012-2018.
10.
Interactions. Angewandte Chemie (International ed in English). 2015;54(52): 15720-15724.
11. Petta I, Lievens S, Libert C, Tavernier J, De Bosscher K. Modulation of Protein-Protein
Milroy LG, Bartel M, Henen MA, et al. Stabilizer-Guided Inhibition of Protein-Protein
Interactions for the Development of Novel Therapeutics. Molecular therapy : the journal of the
9

American Society of Gene Therapy. 2016;24(4): 707-718.
12. Ballone A, Centorrino F, Ottmann C. 14-3-3: A Case Study in PPI Modulation. Molecules
(Basel, Switzerland). 2018;23(6): 1386.
13.
Giordanetto F, Schafer A, Ottmann C. Stabilization of protein-protein interactions by small
molecules. Drug discovery today. 2014;19(11): 1812-1821.
14.
discovery. Expert opinion on drug discovery. 2017;12(9): 925-940.
15. Waring MJ, Chen H, Rabow AA, et al. Potent and selective bivalent inhibitors of BET
bromodomains. Nat Chem Biol. 2016;12(12): 1097-1104.
16. Tanaka M, Roberts JM, Seo HS, et al. Design and characterization of bivalent BET
inhibitors. Nat Chem Biol. 2016;12(12): 1089-1096.
17. Kaae BH, Harpsoe K, Kastrup JS, et al. Structural proof of a dimeric positive modulator
bridging two identical AMPA receptor-binding sites. Chemistry & biology. 2007;14(11): 1294-1303.
18. Drapier T, Geubelle P, Bouckaert C, et al. Enhancing Action of Positive Allosteric
Andrei SA, Sijbesma E, Hann M, et al. Stabilization of protein-protein interactions in drug
Modulators through the Design of Dimeric Compounds. Journal of medicinal chemistry.
2018;61(12): 5279-5291.
19.
Morin MD, Wang Y, Jones BT, et al. Diprovocims: A New and Exceptionally Potent Class
of Toll-like Receptor Agonists. Journal of the American Chemical Society. 2018;140(43): 14440-
14454.
20.
Chupak LS DM, Martin SW, Zheng X, Hewawasam P, Connoly TP, Xu N, Yeung KS,
Zhu J, Langley DR, Tenney DJ, Scola PM. Compounds useful as immunomodulators. World
Intellectual Property Organization. 2015;WO Patent 2015/160641 A2.
21.
Lin DY, Tanaka Y, Iwasaki M, et al. The PD-1/PD-L1 complex resembles the antigen-
binding Fv domains of antibodies and T cell receptors. Proceedings of the National Academy of
Sciences of the United States of America. 2008;105(8): 3011-3016.
22.
units or evolutionary relics? Protein & cell. 2010;1(2): 153-160.
23. Zak KM, Kitel R, Przetocka S, et al. Structure of the Complex of Human Programmed
Death 1, PD-1, and Its Ligand PD-L1. Structure (London, England : 1993). 2015;23(12): 2341-2348.
24. Skalniak L, Zak KM, Guzik K, et al. Small-molecule inhibitors of PD-1/PD-L1 immune
Chen Y, Liu P, Gao F, Cheng H, Qi J, Gao GF. A dimeric structure of PD-L1: functional
checkpoint alleviate the PD-L1-induced exhaustion of T-cells. Oncotarget. 2017;8(42): 72167-
72181.
25.
Guzik K, Zak KM, Grudnik P, et al. Small-Molecule Inhibitors of the Programmed Cell
Death-1/Programmed Death-Ligand 1 (PD-1/PD-L1) Interaction via Transiently Induced Protein
States and Dimerization of PD-L1. Journal of medicinal chemistry. 2017;60(13): 5857-5867.
26.
Wang C, Thudium KB, Han M, et al. In vitro characterization of the anti-PD-1 antibody
10

nivolumab, BMS-936558, and in vivo toxicology in non-human primates. Cancer immunology
research. 2014;2(9): 846-856.
11