The Stone Guest: How Does pH Affect Binding Properties of PD-1/PD-L1 Inhibitors?

Article abstract of DOI:10.1002/cmdc.202000760

The interaction between programmed cell death-1 (PD-1) and its ligand PD?L1 activates a coinhibitory signal that blocks T-cell activation, promoting the immune escape process in the tumor microenvironment. Development of monoclonal antibodies targeting and inhibiting PD-1/PD?L1 interaction as anticancer immunotherapies has proved successful in multiple clinical settings and for various types of cancer. Notwithstanding, limitations exist with the use of these biologics, including drug resistance and narrow therapeutic response rate in a majority of patients, that demand for the design of more efficacious small molecule-based immunotherapies. Alteration of pH in the tumor microenvironment is a key factor that is involved in promoting drug resistance, tumor survival and progression. In this study, we have investigated the effect of pH shifts on binding properties of distinct classes of PD?L1 inhibitors, including macrocyclic peptide and small molecules. Results expand structure-activity relationships of PD?L1 inhibitors, providing insights into structural features and physicochemical properties that are useful for the design of ligands that may escape a drug resistance mechanism associated to variable pH conditions of tumor microenvironment.

Full text of DOI:10.1002/cmdc.202000760

Accepted Article
Title: The Stone Guest: How Does pH Affect Binding Properties of
PD-1/PD-L1 Inhibitors ?
Authors: Alessandra Riccio, Alice Coletti, Daniela Dolciami, Andrea
Mammoli, Bruno Cerra, Sonia Moretti, Antimo Gioiello,
Simone Ferlin, Efisio Puxeddu, and Antonio Macchiarulo
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemMedChem 10.1002/cmdc.202000760
Link to VoR: https://doi.org/10.1002/cmdc.202000760
0
1/2020

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
The Stone Guest: How Does pH Affect Binding Properties of PD-
/PD-L1 Inhibitors ?
1
[
a]
[b]
[a]
[a]
Dr. Alessandra Riccio, Dr. Alice Coletti, Dr. Daniela Dolciami, Dr. Andrea Mammoli, Dr. Bruno
[
a]
[b]
[a]
[c]
[b]
Cerra, Dr. Sonia Moretti, Prof. Antimo Gioiello, Dr. Simone Ferlin, Prof. Efisio Puxeddu, Prof.
Antonio Macchiarulo.*[a]
[
a]
Dr. A. Riccio, Dr. D. Dolciami, Dr. A. Mammoli, Dr. B. Cerra, Prof. A. Gioiello, Prof. A. Macchiarulo, ORCID:0000-0001-9535-4183
Department of Pharmaceutical Sciences
University of Perugia,
via del liceo n.1, 06123 – Perugia (Italy)
E-mail: antonio.macchiarulo@unipg.it
Dr. A. Coletti, Dr. S. Moretti, Prof. E. Puxeddu
Department of Medicine,
University of Perugia,
via Gambuli, 1 – 06132, Perugia (Italy).
Dr. S. Ferlin
[
b]
c]
[
Sterling S.p.A.,
Via della Carboneria, 30 – 06073, Corciano (Perugia, Italy).
Supporting information for this article is given via a link at the end of the document.
Abstract: The interaction between programmed cell death-1 (PD-1)
and its ligand PD-L1 activates a coinhibitory signal that blocks T-cell
activation, promoting the immune escape process in the tumor
microenvironment. Development of monoclonal antibodies targeting
and inhibiting PD-1/PD-L1 interaction as anticancer immunotherapies
has proved successful in multiple clinical settings and for various
types of cancer. Notwithstanding, limitations exist with the use of
these biologics, including drug resistance and narrow therapeutic
response rate in a majority of patients, that demand for the design of
more efficacious small molecule-based immunotherapies. Alteration
of pH in the tumor microenvironment is a key factor that is involved in
promoting drug resistance, tumor survival and progression. In this
study, we have investigated the effect of pH shifts on binding
properties of distinct classes of PD-L1 inhibitors, including
macrocyclic peptide and small molecules. Results expand structure-
activity relationships of PD-L1 inhibitors, providing insights into
structural features and physicochemical properties that are useful for
the design of ligands that may escape a drug resistance mechanism
associated to variable pH conditions of tumor microenvironment.
proteins, PD-1 is mostly expressed on T and B cells, natural killer
(NK) and monocytes, suggesting a wider regulation of immune
responses than other family members.[6,7]
Figure 1. Chemical structures of peptide-57 (1, S8158), BMS-202 (2), S7911
(3) and VIS1059 (4).
PD-L1 and PD-L2 share 20% of sequence identity with other
members of the B-7 family.[8] They are type I transmembrane
glycoproteins composed of a membrane-proximal Ig-C and a
membrane-distal Ig-V-type ectodomain.[9] The Ig-V domain
consists of antiparallel β-strands which form two possible
interaction surfaces, termed GFC and BED surface. The
interaction between PD-L1/2 and PD-1 occurs at the GFC surface,
Introduction
which is different in the two ligand proteins accounting for a
Programmed cell death-1 protein (PD-1) and its ligands (PD-L1
and PD-L2) are members of the B7–CD28–CTLA-4 family of
distinct molecular mechanism of interaction.[
10-13]
In particular,
PD-L1 and PD-L2 compete for the same binding surface on PD-
1. However, upon binding, the larger GFC surface of PD-L1
induces a peculiar conformational change in PD-1 that is not
observed in the case of PD-L2. Such different molecular
mechanism of interaction may account for a 3-fold lower affinity of
PD-L1 than PD-L2 to PD-1.[14]
coregulatory proteins which play a pivotal role in T-cell activation
and tolerance.[1-3]
Structurally, PD-1 is
a monomeric type I transmembrane
glycoprotein containing an IgV extracellular domain, followed by
a stalk region of about 20 amino acids and a cytoplasmic
domain.[4] This latter bears tyrosine-based motifs called
Among PD-1 ligands, PD-L1 is highly present in tumor cells, with
its expression being induced by many cytokines and tumor-
derived exosomes. It is also constitutively expressed in antigen
presenting cells (APCs), T-cells and on a variety of non-
hematopoietic cell types.[15] On this basis, recent studies suggest
that PD-L1 may constitute a key factor for triggering the promotion
Immunoreceptor Tyrosine-based Inhibitory Motif (ITIM) and
Immunoreceptor Tyrosine-based Switch Motif (ITSM).[5] The
engagement of PD-1 by its ligand PD-L1 promotes the
phosphorylation of the ITSM motif and the formation of a docking
site which is able to recruit SHP-1 and SHP-2, leading to the
inhibition of T cells activity. At odds with other B7–CD28–CTLA-4
1
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
[
16,
of the immune escape process in the tumor microenvironment.
binds to the GFC surface of PD-L1, mostly engaging the G strand
of the target protein by hydrophobic interactions (Figure 2).
1
7]
Accordingly, inhibition of PD-1/PD-L1 interaction is being
pursued as viable strategy to develop anticancer
immunotherapies, engineering humanized monoclonal antibodies
or designing peptidomimetics and small molecule inhibitors.[18] In
this framework, the use of PD-1/PD-L1 humanized monoclonal
antibodies has yielded 20%-50% positive response rates in
multiple clinical trials and for various types of cancer, evidencing
that a significant portion of patients shows drug resistance and
poor therapeutic response.[19-21]
Beyond limited bioavailability and safety issues of biologic drugs,
the lack of a wider response to PD-1/PD-L1 monoclonal
antibodies has been correlated to low tumor-infiltrating immune
cells, absence of activated effector T cells, and poor PD-L1
expression in cancer cells, which collectively constitute the
dynamic cell content profile of a tumor microenvironment.[22]
Spatiotemporal alteration of pH in tumor microenvironment is
another factor that has been reported to promote tumor survival
and progression, and may contribute to establish drug resistance
towards anticancer therapies.[23, 24] Specifically, normal tissues
present an intracellular pH (pHi) of 7.2, and an extracellular pH
Figure 2. Interactions of peptide-57 (1) on the GFC surface of PD-L1 (5O4Y).
Yellow dashed lines show ligand/target hydrogen bond interactions; cyan
dashed lines show ligand/target -stacking contacts.
(
pHe) of 7.4. In tumor microenvironment, pHe decreases to acidic
values below 7.0, while pHi assumes higher basic values.[
25-27]
These changes can alter the ratio of protonated versus
deprotonated forms of charged residues on target proteins and
chemical groups on cognate ligands, depending on the relative
acid dissociation constants (pKa). Accordingly, shifts of pHi/pHe
out of the physiological range can affect both target protein
functionality as well as drug efficacy, weakening electrostatic
interactions which govern protein structure, folding and ligand
binding.[28]
In this work, we attempt to investigate the effect of pH variations
on binding properties of a selection of PD-L1 inhibitors belonging
to the different chemical categories of macrocyclic peptides and
small molecule inhibitors (1-4, Figure 1). Firstly, docking studies
and molecular dynamic (MD) simulations were carried out to
depict putative electrostatic interactions and hydrogen bonds
featuring the binding modes of these inhibitors to PD-L1. Next,
analysis of physicochemical properties was carried out to assess
pKa values of aminoacidic residues and small molecules. Finally,
microscale thermophoresis (MST) analysis was performed to
Table 1. Stabilities (occupancy, %) of ligand/protein interactions as resulting
from trajectories of MD simulations.
Resid.
Inter.
Type
2/PD-L1
(5J89 C/D)
3/PD-L1
(5J89 C/D)
4/PD-L1
(5J89 C/D)
1/PD-L1
(504Y)
H-bond[a]
32% C[e]
Phe19
Glu58
Asn63
Gln66
Asp73
Asp122
Lys124
Glu58
Asn63
Gln66
Asp73
Val76
-
-
-
-
H-bond
H-bond
H-bond
H-bond
H-bond
-
-
68%
-
-
-
100%
62% D[e]
-
-
252%
-
-
-
32%
65% C
35% C
-
-
H-bond
Water[b]
35% C
65% C
-
-
-
-
-
40%
Water
Water
Water
Water
Water
Water
Water
-
-
-
32%
-
-
61% D
-
-
-
-
81%
evaluate dissociation constants (K
different pH conditions.
d
) of inhibitors to PD-L1 at
-
-
-
30%
Ala121
Asp122
Phe19
Thr20
Tyr56
58% C
30% C
63% C
-
-
-
-
-
-
-
31% C
-
-
-
Water
37% C
47% D
36% D
-
-
48%
31%
-
Results and Discussion
phobic[c]
40% D
-
58% D
-
Val68
phobic
phobic
phobic
phobic
Binding modes of inhibitors to PD-L1.
Ile54
46% D
38% D
43% C
64% C
-
A selection of inhibitors of PD-L1 belonging to chemical classes
of macrocyclic peptides and small molecules was collected from
literature, including peptide-57 (1, S8158; pdb code 5O4Y), BMS-
Met115
Ala121
Tyr123
Tyr56
-
39% D
36% C
32% D
76% C/D
41%
-
31% C
45% C
66% C/D
phobic
-[d]
56%
-
202 (2, pdb code: 5J89) and S7911 (3). A further compound (4,
67%
C/D[
e]
VIS1059) was ad-hoc designed and synthetized as structural
analogue of S7911, but endowed with different protonation state
and conformational property.
[
a] H-bond: hydrogen bond; [b] Water: water-bridge mediate hydrogen bond; [c]
phobic: hydrophobic contacts; [d] -: -interaction; [e] protein chain C and/or D
Peptide-57 (1) is a macrocyclic compound that is able to block
PD-1/PD-L1 interaction with an EC50 of 0.566 ± 0.122 µM, as
determined in a luciferase reporter assay using T-cell-like Jurkat
cells.[29] Crystallographic studies have shown that peptide-57
of PD-L1 dimeric form.
Specifically, tryptophan residues (W8 and W10) of peptide-57
occupy two narrow clefts of PD-L1, making hydrophobic contacts
and -stacking interactions with Ile54, Tyr56 and Val68 in the first
2
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
cleft, and Tyr56, Glu58, Arg113, Met115 and Tyr123 in the second
cleft. Furthermore, two hydrogen bond interactions are observed
between the side chain of Gln66 and backbone atoms of L6 and
W10, while one additional hydrogen bond is formed between
Asn63 and the backbone carbonyl of W8.
MD simulation provides insights into relative stabilities of these
interactions (Table 1). In particular, hydrogen bond interactions
involving P4, L6, and W8 of peptide-57, and the backbone of
Gln66 and Asn63 of PD-L1 show a very high stability along the
trajectory (Table 1; occupancy values 252% and 100%,
respectively). These are followed by a further stable hydrogen
bond involving residues W10 and Glu58 (occupancy 68%), and
by both water-bridge mediated and direct hydrogen bond
interactions connecting H5 and Asp73 (occupancy 81%, 32%).
Stable hydrophobic contacts and -stacking interactions are
observed between Met115, Tyr123 and Tyr56, and F1, W8 and
W10 of peptide-57. Noteworthy, the basic residue R13 of peptide-
2-aminoethylacetamide chain of BMS-202 protrudes into the
solvent with its secondary amine and acetamido groups facing a
salt bridge formed by the side chains of Asp122 and Lys124 of
chain C.
The MD simulation shows strong stability of the above -stacking
interactions with Tyr56 (Table 1; chains C and D, occupancy 67%),
and hydrophobic contacts with Tyr123 (chain C, occupancy 64%).
Moreover, water-bridge mediated hydrogen bonds are observed
between the backbone of Phe19, Ala121 and the secondary
amine group of BMS-202 along the MD trajectory (chain C,
occupancy 63%, 58%). Noteworthy, the secondary amine group
and the nitrogen atom of the amide moiety make stable hydrogen
bond interactions with the side chain of Asp122 (chain C,
occupancy 65%). The carbonyl group of the acetamide moiety is
involved in hydrogen bond interaction with the side chain of
Lys124 (chain C, occupancy 35%).
57 is not directly engaged in any polar interaction with specific
residues of PD-L1, but it is exposed to solvent molecules.
Although MD simulation confirms hydrophobic interactions as
important in the binding of peptide-57 to the GFC surface of PD-
L1, it also pinpoints a significant contribution of polar interactions
to the stability of the complex.
Figure 4. Interactions of S7911 (3) into the binding cleft of the dimeric form of
PD-L1 (5J89). Yellow dashed lines show ligand/target hydrogen bond
interactions; cyan dashed lines show ligand/target -stacking contacts; purple
dashed lines show ligand/target electrostatic interactions. Torsional ether angle
(τ1 = 156±17°) and amine angle (τ2 = 162±29°) are labelled.
Figure 3. Interactions of BMS-202 (2) into the binding cleft of the dimeric form
of PD-L1 (5J89). Yellow dashed lines show ligand/target hydrogen bond
interactions; cyan dashed lines show ligand/target -stacking contacts; purple
dashed lines show ligand/target electrostatic interactions.
No crystallographic studies are available for S7911 (3) and its
analogue (4, VIS1059). Accordingly, docking studies were carried
out using the dimeric crystal structure of PD-L1 (pdb code: 5J89)
followed by MD simulations (Figure 4). Akin to BMS-202 (2),
S7911 (3) engages the GFC surface of two PD-L1 units with a
binding mode that is characterized by stable aromatic interactions
between its biphenyl moiety and Tyr56 (Table 1; chains C/D,
occupancy 66%), as well as hydrophobic contacts with Ala121
BMS-202 (2) belongs to
a
class of (2-methyl-3-
biphenylyl)methanol derivatives identified by Bristol-Myers
Squibb as small molecule inhibitors of PD-L1/PD-1 interaction,[30]
(
chain C; occupancy 31%). A salt bridge interaction is observed
with an IC50 of 0.018 µM in homogeneous time-resolved
fluorescence (HFRET) binding assay. The determination of the
crystal structure of BMS-202 in complex with PD-L1 has allowed
to get insights into the mechanism of interaction of this class of
small molecules.[31] Specifically, it was found that BMS-202
induces a dimeric form of PD-L1, binding into a cavity shaped by
the GFC surface of two PD-L1 units (Figure 3). The biphenyl
moiety of 2 is buried within the cavity engaging Met115 and
Ala121 of both PD-L1 chains (e.g. chains C/D) with hydrophobic
contacts, while its distal phenyl ring makes -stacking interactions
with Tyr56 of chain C. The pyridine ring forms a -stacking
interaction with Tyr56 of chain D, and its methoxy substituent
makes hydrophobic contacts with Tyr123 of chain C. The terminal
between the negatively charged group of Asp122 and the
protonated ternary amine group of 3 (chain C, occupancy 35%),
whereas the side chain of Lys124 makes both salt bridge and
hydrogen bond interactions with the carboxylate group of the
compound (chain C, occupancy 65%). Compound VIS1059 (4)
was designed as distant analogue of S7911 (3), replacing the
ternary amine and central ether core of the latter compound with
two amide moieties. These replacements allow restraining the
conformational freedom of two dihedral angles due to the
energetic favored planar geometry of the amide torsional angle
with respect to the ether (τ1) and amine (τ2) torsional angles.
Nearly anti-planar conformations of the ether angle (τ1 = 156±17°)
and amine angle (τ2 = 162±29°) are indeed observed along the
3
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
MD trajectory of S7911. Furthermore, the conversion of the amine
nitrogen to an amide nitrogen bestows a different protonation
state to the derived compound. Docking and MD simulation
reveals conserved aromatic interactions between the biphenyl
moiety of the analogue 4 and Tyr56 (Table 1, chains C/D;
occupancy 76%), as well as hydrophobic contacts with Ala121
Treatment of 7 with oxalyl chloride and Et3N in dry CH2Cl2/DMF
followed by reaction with a slight excess of pipecolic acid (1.05
equiv.) afforded VIS1059 (4) in 8% yield.
Physicochemical properties.
Calculated and experimental acidic constants (cpKa, pKa) were
assessed for the macrocyclic peptide (1), small molecule
inhibitors (2-4) and selected polar residues of PD-L1 (Table 2).
As a result, small molecules cover a different range of pKa values,
and may assume diverse protonation states depending on the
environmental pH condition. In particular, BMS-202 (2) bears a
secondary amine group (cpKa2= 7.67/7.74; pKa2= 8.39±0.05)
and a pyridine nitrogen (cpKa1= 0.64/1.48). While the latter group
assumes an uncharged state, the protonated/unprotonated ratio
of the amine group may be affected by changes of pHe and pHi
values in the tumor microenvironment.[
(
chain C, occupancy 36%). Stable hydrogen bonds are formed
between the side chain of Gln66 and the carbonyl group of the
terminal amide moiety (chain D, occupancy 62%). An additional
water-bridge mediated hydrogen bond is observed between the
side chain of Gln66 (chain D, occupancy 61%) and the carbonyl
group of analogue 4.
25-27]
S7911 (3) is an α-aminoacidic compound featuring a cyclic
ternary amine group. According to its experimental pKa values
(
pKa1= 2.78 ± 0.08; pKa2= 10.49 ± 0.19), this compound mostly
adopts a zwitterion form at physiological pHe values. VIS1059 (4)
is an acidic compound (cpKa1= 3.44/3.65; pKa1= 4.77 ± 0.06)
that assumes a deprotonated negatively charged state at
physiological pHe value, whereas it may increasingly adopt a
neutral uncharged state with decreasing pHe value in the tumor
microenvironment.
Table 2. Calculated and experimental pKa properties of compounds 1-4 and
selected residues of PD-L1 (Glu58, Asp73, Asp122 and Lys124).
Figure 5. Interactions of VIS1059 (4) into the binding cleft of the dimeric form of
PD-L1 (5J89). Yellow dashed lines show ligand/target hydrogen bond
interactions; cyan dashed lines show ligand/target -stacking contacts.
Cpd./
Res.
cpKa1
cpKa2
cpKa3
pKa1
pKa2
[
a]
[a]
[a]
6
7
0
1
1.04
1
3
.74
.18
.64
.48
11.80
11.49
7.67
7.74
6.41
8.87
>12.0
>12.0
1
2
3
4
[b]
a]
[b]
a]
[b]
[b]
-
-
[
[a]
[c]
[b]
-
-
-
n.d.
8.39±0.05
10.49±0.19
-
[
[a]
Synthesis of VIS1059.
The synthesis of compound VIS1059 (4) was realized according
to Scheme 1. In particular, methyl-[1,1'-biphenyl]-3-yl)methanol
[b]
[a]
[b]
[b]
2.78±0.08
4.77±0.06
.99
.44
-
3.65
[
d]
Glu58
Asp73
3.1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
(
5) was converted into the corresponding carboxylic acid
[
d]
2.3
derivative 6 by oxidation reaction with sodium bromate (3 equiv.)
and sodium hydrogen sulfate (1 equiv.) in refluxing MeCN.[32]
[
e]
Asp122
Lys124
<0.0
[e]
>12.0
The acid 6 obtained in 70% yield was reacted with oxalyl chloride
in dry CH2Cl2/DMF at room temperature and the acyl chloride
thus formed was readily treated with 4-amino-2-methoxybenzoic
acid in the presence of Et3N to form the desired 2-methoxy-4-(2-
methyl-[1,1'-biphenyl]-3-carboxamido)benzoic acid (7) in 8%
isolated yield.
[
a] Marvin pKa prediction; [b] Epik pKa prediction in water; [c] n.d. = not
determined (out of the range of values of Sirius T3, pH 2-12); [d] calculated
using 5O4Y structure; [e] calculated using 5J89 structure. Experimental pKa
values are reported as mean ± standard deviation from n.3 titrations.
The macrocyclic peptide (1) has one arginine residue (R13) with
a calculated cpKa of 11.80/11.49, and a histidine residue (H5)
with calculated pKa values of 6.74/7.18 and >12.0. Hence, H5
assumes a protonated state with decreasing value of pHe in the
tumor microenvironment, whereas the side chain of R13 stably
adopts a protonated form. Acidic residues of PD-L1 that are in
contact with the macrocyclic peptide (1) in the crystallographic
complex (pdb code 5O4Y) are Glu58 (cpKa1= 3.1) and Asp73
(
cpKa1= 2.3). Side chains of these residues are exposed to the
solvent and mostly adopt a deprotonated negatively charged state
at the pHe of the tumor microenvironment. In the crystallographic
complex of BMS-202 (2) bound to PD-L1 (pdb code 5J89), acidic
and basic residues in contact with the ligand include Asp122
Scheme 1. Synthesis of VIS1059 (4). Reagents and conditions: a) NaBrO
NaHSO , MeCN, reflux, 70%; b) (COCl) , CH Cl , DMF, r.t.; c) 4-amino-2-
methoxybenzoic acid, Et N, CH Cl , r.t., 8% from 6; d) (COCl) , dry CH Cl , DMF,
r.t.; e) pipecolic acid, Et N, CH Cl , r.t., 8% from 7.
3
,
(
cpKa< 0.0) and Lys124 (cpKa> 12.0) that engage each other in
4
2
2
2
a salt bridge interaction. Accordingly, the carboxylic group of
Asp122 assumes a deprotonated form whereas the primary
3
2
2
2
2
2
3
2
2
4
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
amine group of Lys122 adopts a positively charged state at any
pHe of the tumor microenvironment.
Moreover, it is possible that conformational restraining upon
amide group insertion may be detrimental for the binding activity
of 4. Peptide-57 (1) results the most potent PD-L1 ligand among
Effect of pH on dissociation constants of PD-L1 inhibitors.
d
tested compounds, generating a dissociation constant (K ) value
Since the selected inhibitors (1-4) have different binding modes
and physicochemical properties, in this part of the study we
investigated how and how much shifts of pH affect their binding
properties to PD-L1. Ligand binding assays were thus performed
using microscale thermophoresis (MST). Thermophoresis is the
movement of a biomolecular complex in a temperature gradient
depending on size, charge, and hydration shell that typically
change upon ligand/target interaction.[33] In combination with
cellular and other biochemical assays, this approach was
previously used to investigate the target engagement of a
selected set of PD-1/PD-L1 inhibitors which included peptide-57
of 0.80 ± 0.15 μM with a negative thermophoretic trend (Figure 6).
(
1) and other small-molecule inhibitors. In particular, authors
found that peptide-57 (1) and two analogues of BMS-202 (2) were
able to bind to PD-L1 in the nanomolar range of potency.[34]
Binding properties of PD-L1 inhibitors were assessed at three
different pH values (pH = 6.2, 7.2, 8.2), above the theoretical
isoelectric point of the protein (pI = 5.73). Stability of unlabelled
and labelled recombinant protein at the three different pH values
was checked using thermal shift assays (Figure S1 and Table S1,
supplementary material). Results show very similar unfolding
profiles and Ti values of the protein, suggesting that labelling and
different pH conditions do not affect the proper folding of PD-L1.
The binding of compounds 1-4 to fluorescently labelled PD-L1
Figure 7. MST binding curve of BMS-202 (2) to red labelled recombinant human
PD-L1 at pH 7.2 condition. The binding curve and K
from n.3 independent experiments.
d
± c.v value were generated
(
red-hPD-L1) was firstly evaluated at pH 7.2. As a result,
compounds 1-3 yield a productive binding interaction to PD-L1
Figures 6-8), whereas no binding activity is observed for VIS1059
at tested concentrations (4, Figure 9; K > 1mM).
(
d
Figure 8. MST binding curve of S7911 (3) to red labelled recombinant human
PD-L1 at pH 7.2 condition. The binding curve and K
from n.3 independent experiments.
d
± c.v value were generated
Figure 6. MST binding curve of peptide-57 (1) to red labelled recombinant
human PD-L1 at pH 7.2 condition. The binding curve and K
generated from n.3 independent experiments.
d
± c.v value were
This finding pinpoints that presence of a negative net charge on
the molecule (pKa1= 4.77 ± 0.06) is detrimental for the interaction
to PD-L1, and aromatic as well as hydrophobic contacts between
the biphenyl moiety of 4 and binding site residues are unable to
compensate for it. Noteworthy, in the MD simulation the carboxylic
group of VIS1059 (4) does not form any specific salt bridge
interaction with the nearby positively charged Lys124, as instead
observed for its α-aminoacidic analogue 3. This is also in
agreement with statistical studies suggesting that binding sites of
Figure 9. MST binding curve of VIS1059 (4) to red labelled recombinant human
PD-L1 at pH 7.2 condition. The binding curve was generated from n.3
independent experiments.
d
Previous MST studies reported a K value of 0.019 ± 0.002 μM for
this compound.[34] Discrepancy between the two results, however,
can be explained by the use of different experimental settings in
carboxylic groups have mostly two polar residues with two
hydrogen bond donors to make stable interactions.[35]
5
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
the two studies (e.g. labeling protocols, time of measurements,
type of capillaries). According to a rank order of binding potency,
piperidine ring of S7911 (3) which tends to be less protonated as
far as the pH shift approaches the pKa of the ternary amine group
(pKa2= 10.49 ± 0.19), leading to an increased fraction of the
negatively charged microspecies.
d
peptide-57 (1) is followed by BMS-202 (2; K = 8.13 ± 1.38 μM)
and S7911 (3; K = 38.6 ± 6.2 μM), with these compounds
d
showing a positive thermophoretic trend (Figures 7, 8). Hence, at
the opposite of VIS1059 (4), presence of a positive net charge in
BMS-202 (2) favors the binding activity to PD-L1, while the neutral
net charge of the zwitterionic group in S7911 (3) produces a
slightly decrease of potency. Remarkably, the positive
thermophoretic trend of small molecules (2, 3) and the negative
thermophoretic trend of macrocyclic peptide (1) may be ascribed
to a distinct mechanism of interaction of these classes of ligands
to PD-L1, with small molecules inducing a dimerization form of the
protein whereas macrocyclic peptide binding to a monomeric form
of PD-L1.[36]
Overall, MST studies suggest that presence of a net positive
charge in the macrocyclic peptide (1) and BMS-202 (2) promotes
binding activity to PD-L1 with decreasing pH values. This is
achieved through formation and/or enforcement of salt bridge
interactions with target acidic residues (Asp73 and Asp122).
Moreover, this is further sustained by the lack of activity of
VIS1059 (4), whose negative net charge and conformational
restraining are detrimental for the binding activity.
Conclusions
Next, binding activities of compounds 1-3 to PD-L1 were
evaluated at pH 6.2 and 8.2, with results being reported in table
3
. Inspection of the table reveals that the dissociation constant of
PD-1 and PD-L1 belong to a broad family of immune checkpoint
molecules which are involved in regulating T-cell activation and
immune tolerance. Accordingly, they are attractive drug targets in
cancer therapy because the inhibition of their interaction provides
a means to remove coinhibitory signals that block antitumor
immune responses. Although several humanized monoclonal
antibodies targeting these immune checkpoints have been
developed and are in clinical use, these biologics suffer from drug
resistance and narrow therapeutic response in a majority of
patients, high risk of immune-related adverse reactions, poor oral
bioavailability and high costs of production. Distinct chemical
classes of small molecule inhibitors of PD-1/PD-L1 interaction are
being designed to tackle these limitations, though their
development as novel anticancer immunotherapies still lags in
preclinical settings. Recent studies support the notion that
alteration of pHe and pHi in tumor microenvironment is a
mechanism driving drug resistance towards anticancer therapies.
In this framework, our study presented here has investigated the
effect of pH shifts on the binding properties of distinct chemical
classes of PD-L1 inhibitors. Results highlight the importance of
pH variation in strengthening/weakening the electrostatic
contribution to ligand/protein interactions of PD-L1 inhibitors. In
particular, both macrocyclic peptide and small molecule inhibitors
bearing a net positive charge show enhanced binding activity to
the target protein with decreasing pH values, making and
strengthening salt bridge interactions with acidic residues. As a
consequence, decrease of pHe value below 7.0 in tumor
microenvironment may enforce target engagement of PD-L1
inhibitors bearing amine groups, thus allowing escape of a tumor
resistance mechanism based on acidosis. Collectively, these
results provide further insights into structure-activity relationships
of PD-L1 inhibitors that are useful for the design of next
generation small molecule inhibitors with physicochemical
properties that are resilient to variable pH conditions of tumor
microenvironment.
peptide-57 (1) is marginally affected by the shift of pH conditions,
with only pH 6.2 producing a slight improvement of the binding
activity (K = 0.38 ± 0.07 μM).
d
Table 3. Dissociation constants (Kd) of compounds 1-3 against PD-L1 at
different pH conditions.
Cpd.
pH 6.2
± c.v.[a]
pH 7.2
± c.v.
pH 8.2
d
K ± c.v.
K
d
K
d
1
2
3
0
4
.38 ± 0.07 μM
.50 ± 0.56 μM
0.80 ± 0.15 μM
8.13 ± 1.38 μM
38.6 ± 6.2 μM
0.84 ± 0.13 μM
12.37 ± 1.91 μM
57.3 ± 11.8 μM
3
3.5 ± 4.6 μM
[
a] c.v.: confidence values. K
d
± c.v values were generated from n.3 independent
experiments.
This observation is in agreement with previous studies reporting
hydrophobic contacts of 1 rather than electrostatic interactions as
important for macrocyclic peptide binding to PD-L1. However, the
slight improvement of the binding activity of peptide-57 finds a
likely explanation in the presence of a histidine residue (H5;
cpKa= 6.74) whose fraction of protonated form becomes the
major microspecies at pH 6.2. As a consequence, the hydrogen
bond interaction observed in the MD simulation between H5 and
Asp73 (Table 1, Figure 2) would be enforced by an energetically
stronger salt bridge interaction.
The analysis shows that BMS-202 (2) is affected by the pH
condition, with lower and higher pH values strengthening and
weakening the interaction to PD-L1, respectively. This
observation reflects the protonation state of the secondary amine
group of 2, with its pKa value (pKa2= 8.39 ± 0.05) standing at the
edge of the pH range adopted for this study. Accordingly, the
major microspecies of the amine group at pH 6.2 is the protonated
form which makes a stronger salt bridge interaction with Asp122
(
Table 1). In contrast, levels of the protonated microspecies of the
amine group incrementally decrease at pH 7.2 and 8.2,
weakening such an electrostatic interaction. S7911 (3) is an α-
aminoacidic compound that, on the basis of its experimental pKa
values (pKa1= 2.78 ± 0.08; pKa2= 10.49 ± 0.19), adopts a
zwitterionic form as major microspecies at the investigated pH
conditions. In this case, a thin effect of pH on the binding activity
of 3 is only observed at pH 8.2, suggesting that raise of pH values
may weaken ligand/protein electrostatic interactions. This
observation can be explained by the protonation state of the
Experimental Section
Docking Study and Molecular Dynamics. The chemical structures of PD-
L1
inhibitors
1-3
were
retrieved
from
PubChem
(
https://pubchem.ncbi.nlm.nih.gov) and compound 4 was generated using
Maestro (Schrödinger Release 2019-3: Maestro, Schrödinger, LLC, New
York, NY, 2019). Next, chemical structures 1-4 were refined using LigPrep
6
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
(
2
Schrödinger Release 2019-3: LigPrep, Schrödinger, LLC, New York, NY,
019) to add hydrogens, generate protonation states at pH = 7.2, and
purifications were performed using Biotage Isolera™ Prime (Biotage AB,
Uppsala, Sweden). Mass spectroscopy was performed with a Dionex
UltiMate 3000 HPLC separations module combined with HCT ultra ion trap
(Bruker). The analytical column was a Waters Xselect CSH Phenyl-Hexyl
(5 μm, 2.1 x 150 mm), protected by a guard column 2.1 x 4 mm.
Compounds purity was assessed by HPLC-DAD at 254 and 280 nm.
optimize their geometry using OPLS3e force field with an energy gradient
convergence criterion of 0.05 kJ Å-1mol-1. Crystal structures of PD-L1
bound to peptide-57 (1, S8158; pdb code 5O4Y)[26] and BMS-202 (2, pdb
code
(
5J89)[36]
were
downloaded
from
RCSB
PDB
https://www.rcsb.org).[37] The crystal structure 5O4Y was split into single
chain structures, keeping chain A with compound 1 bound to the GFC
surface of PD-L1. In the case of crystal structure 5J89, chains C/D were
kept with the co-crystallized ligand 2 bound into the cavity shaped by the
GFC surface of the two PD-L1 units. The Protein Preparation Wizard
-Methyl-[1,1'-biphenyl]-3-carboxylic acid (6).[32] To a suspension of (2-
methyl-[1,1'-biphenyl]-3-yl)methanol (5) (1 g, 5.04 mmol) in MeCN (25 mL),
NaBrO (2.28 g, 15.13 mmol) and NaHSO (5.04 mmol) were added at
room temperature and the resulting suspension was refluxed for 1.5 h. The
mixture was allowed to cool to room temperature and filtered on a pad of
celite. The filtrate was concentrated under reduced pressure and the
residue was redissolved in 1 M aqueous solution of NaOH (50 mL) and
washed with EtOAc (3 x 20 mL). The aqueous phase was acidified with 3
N aqueous solution of HCl (pH = 3) and extracted with EtOAc (3 x 50 mL).
2
3
4
(
PPW)[38] tool implemented in Maestro was then employed to process and
refine the selected chain structures, removing water and detergent
molecules, adding hydrogen atoms and optimizing internal geometries.
During geometry optimization, the OPLS3e force field was used with an
atomic coordinate displacement restrain on heavy atoms set to 0.30 Å.
Docking studies of compounds 3 and 4 into the binding site shaped by
chains C and D of 5J89 were performed employing Glide (Schrödinger
Release 2019-3: Glide, Schrödinger, LLC, New York, NY, 2019).[39]
Specifically, a grid box was defined with its centre located on the centre of
mass of BMS-202 (2) bound to chains C/D of 5J89 (x = 3.23, y = 6.18, z =
The combined organic extracts were washed with H
2
O (30 mL), brine (30
mL), dried over anhydrous Na SO and concentrated under reduced
2
4
pressure. The desired product (6, 750 mg, 3.53 mmol, 70% yield) was
obtained as off-white solid and used for the next step without further
1
purification. H-NMR (400 MHz, CDCl
3
): δ 2.47 (s, 3H), 7.22-7.31 (m, 3H),
4.56). The inner grid box was sized 10 x 10 x 10 Å. Docking studies were
7.34-7.36 (m, 1H), 7.39-7.44 (m, 3H), 7.99 (dd, J1= 1.36 Hz, J2= 7.79 Hz,
carried out using the standard precision (SP) method and the Gscore
scoring function (kcal/mol). The top scored docking pose for each molecule
was selected (3 pose #1; Gscore = -8.65 kcal/mol; 4 pose #1; Gscore = -
1
H).
2-Methoxy-4-(2-methyl-[1,1'-biphenyl]-3-carboxamido)benzoic acid (7). To
10.04 kcal/mol), since the other remaining docking poses showed
a solution of 2-methyl-[1,1'-biphenyl]-3-carboxylic acid (6) (750 mg, 3.53
mmol) in a mixture of dry CH Cl /DMF (1.5:1 v/v, 60 mL), (COCl) (318 μL,
.71 mmol) was slowly added at room temperature and under argon
differences from that one only at level of the conformations of the solvent
exposed polar group (Figures S1-S2 and Tables S1-S2, supplementary
material), while presenting a conserved binding mode of the aromatic core
structure. Then, the selected binding poses of compounds 3 and 4 were
submitted to MD simulations, along with the refined crystallographic
complexes of peptide-57 (1) bound to chain A of 5O4Y, and 2 bound to
chains C/D of 5J89. Briefly, each PD-L1 ligand bound complex was
solvated in an orthorhombic box using TIP3P water molecules, extended
2
2
2
3
atmosphere. The resulting mixture was stirred at room temperature for 2
h. Volatiles were evaporated under reduced pressure and the crude acyl
chloride was dissolved in dry CH
solution of 4-amino-2-methoxybenzoic acid (562 mg, 3.36 mmol) in CH
30 mL) and Et N (703 μL, 5.05 mmol). The resulting mixture was stirred
2
Cl
2
(10 mL) and added dropwise to a
2
Cl
2
(
3
for 1 h at room temperature and then quenched by adding 3 N aqueous
solution of HCl (20 mL). The phases were separated and the aqueous
phase was extracted with CH
extracts were washed with H
anhydrous Na SO and concentrated under reduced pressure. The crude
was purified by automated flash chromatography on silica gel (eluent:
petroleum ether/EtOAc from 100:0 to 50:50, v/v) affording the desired 2-
methoxy-4-(2-methyl-[1,1'-biphenyl]-3-carboxamido)benzoic acid (7, 100
10 Å away from any protein atom. The resulting system was neutralized
by adding sodium and chlorine ions at a concentration of 0.15 M. Periodic
boundary conditions were applied to avoid finite-size effects. Atomic partial
charges of ligand were maintained as obtained from OPLS3e calculation.
MD simulations were performed using Desmond (Schrödinger Release
2
Cl
2
(2 x 25 mL). The combined organic
O (50 mL), brine (50 mL), dried over
2
2
4
2019-3: Desmond Molecular Dynamics System, D. E. Shaw Research,
New York, NY, 2019. Maestro-Desmond Interoperability Tools,
Schrödinger, New York, NY, 2019) and the OPLS3e force field.[40] The
simulation protocol included starting relaxation steps and a final production
phase of 50 ns. The occupancy of intermolecular and water-bridge
mediated hydrogen bonds, hydrophobic contacts, aromatic (- stacking)
and electrostatic interactions was calculated along the last 45 ns of the
production phase of each MD simulation, using a cut-off value of 30% and
the Simulation Interaction Diagram Tools.
1
mg, 0.28 mmol) as off-white solid. H-NMR (400 MHz, CDCl3): δ 2.34 (s,
3
7
H), 4.13 (s, 3H), 6.93 (dd, J1= 1.59 Hz, J2= 8.52 Hz, 1H), 7.26-7.47 (m,
H), 8.07-8.09 (m, 2H), 8.13 (brs, 1H).
1-(2-Methoxy-4-(2-methyl-[1,1'-biphenyl]-3-
carboxamido)benzoyl)piperidine-2-carboxylic acid (VIS1059, 4). To a
solution of 2-methoxy-4-(2-methyl-[1,1'-biphenyl]-3-carboxamido)benzoic
acid (7, 80 mg, 0.22 mmol) in a mixture of dry CH
2 2
Cl /DMF (1.5:1 v/v, 5
Chemical compounds. Peptide-57 (S8158, 1), BMS-202 (S7912, 2) and
S7911 (3) were purchased from Selleckchem (Selleckchem, Houston,
mL), (COCl) (25 μL, 0.29 mmol) was slowly added at room temperature
2
and under argon atmosphere. The resulting mixture was stirred at room
temperature for 2 h. Volatiles were evaporated under reduced pressure
USA).
VIS1059
(1-(2-Methoxy-4-(2-methyl-[1,1'-biphenyl]-3-
carboxamido)benzoyl)piperidine-2-carboxylic acid, 4) was synthetized
according to the following procedure.
and the crude acyl chloride was dissolved in dry CH
dropwise to a solution pipecolic acid (30 mg, 0.23 mmol) in CH
and Et N (46 μL, 0.33 mmol). The resulting mixture was stirred for 1 h at
2
Cl
2
(1 mL) and added
2
Cl (3 mL)
2
3
General Methods. All the chemicals were purchased from Sigma-Aldrich
room temperature and then quenched by adding 3 N aqueous solution of
HCl (5 mL). The phases were separated and the aqueous phase was
(now Merck Life Science) and used without further purification. NMR
spectra were recorded on a Bruker AC 400-MHz spectrometer (Bruker,
Madison, WI, USA) in the indicated solvent. Chemical shifts are reported
extracted with CH
washed with H O (15 mL), brine (15 mL), dried over anhydrous Na
and concentrated under reduced pressure. The crude was purified by
automated flash chromatography on silica (eluent: CH Cl /MeOH from
00:0 to 95:5, v/v) affording the desired 1-(2-methoxy-4-(2-methyl-[1,1'-
biphenyl]-3-carboxamido)benzoyl)piperidine-2-carboxylic acid (VIS1059,
, 8 mg, 0.017 mmol) as pale yellow amorphous solid. HPLC purity (UV-
DAD, 254 nm/280 nm): 94.3-94.7%. HRMS (M + H)+: m/z 473.20872
requires 472.20234], ppm: 5.33. 1H-NMR (400 MHz,
CD3OD): δ 1.48-1.54 (m, 2H), 1.64-1.68 (m, 2H), 1.89 (s, 3H), 2.27-2.29
m, 2H), 3.51-3.65 (m, 2H), 3.88 (s, 3H), 4.56-4.58 (m, 1H), 7.26-7.46 (m,
2
Cl
2
(2 x 10 mL). The combined organic extracts were
2
2
SO
4
in parts per million (ppm) and are relative to CDCl
3
(7.26 ppm and 77.0
ppm) or CD OD (3.31 ppm and 49.2 ppm). The abbreviations used are as
3
2
2
follows: s, singlet; brs, broad singlet; d, doublet; dd, double of doublets; dt,
doublet of triplets; t, triplet; q, quartet; qui, quintet; m, multiplet; and brm,
broad multiplet. Coupling constants (J) are reported in Hertz (Hz). Melting
points were determined using a Buchi 535 electrothermal apparatus. Thin-
layer chromatography (TLC) was performed on aluminium backed silica
plates (silica gel 60 F254, Merck, Darmstadt, Germany). Spots were
visualised by UV detector (λ: 254 nm) and/or by staining and warming with
potassium permanganate. When required, flash chromatographic
1
4
28 28 2 5
[C H N O
(
7
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
1
3
1
1
1H), 7.68 (s, 1H). 13C-NMR (100.6 MHz, CD
3
OD): δ 16.2, 21.2, 24.9, 29.3,
running buffer, such as PB at the selected pH containing 4% DMSO in
PCR tubes supplied by NanoTemper Technologies to yield final volumes
of 10 µL. In particular, compounds were tested in triplicate using different
concentrations for each compound, starting from 0.1 mM for peptide-57
9.4, 54.7, 62.9, 102.7, 111.6, 121.7, 127.9, 125.4, 125.6, 126.8, 127.8,
27.9 (2C), 128.8 (2C), 130.9, 132.2, 138.2, 141.3, 143.2, 155.8, 168.7,
69.1, 170.2.
(
1), 0.5 mM for BMS-202 (2), 1 mM for S7911 (3), 1 mM for VIS1059 (4).
An amount of 10 µl solution of 100 nM red-hPD-L1 was added to each
compound dilution and mixed to halve the concentration of each
component, including 2% DMSO and a reaction volume of 20 µL. These
samples were loaded into 16 premium-coated capillary tubes (MO-K025;
NanoTemper Technologies, Munich, Germany) and inserted in the chip
tray of Monolith NT.115 for thermophoresis analysis. Compounds 1-4 were
tested with labeled protein at pH 7.2, whereas only active compounds (1-
Physicochemical Properties. Values of cpKa for macrocyclic peptide (1)
and small molecules (2-4) were calculated using Marvin v20.11
(ChemAxon; http://www.chemaxon.com) and Epik prediction in water
solvent (Schrödinger Release 2019-3: Epik, Schrödinger, LLC, New York,
NY, 2019), whereas cpKa for selected PD-L1 residues were calculated
using H++ (v.3.2) server.[
41-43]
The pKa plugin of Marvin calculates pKa
values of compounds based on the partial charge distribution of the small
3
) were also tested with labeled protein at pH 6.2 and pH 8.2. Capillary
molecule, whereas Epik uses Hammett and Taft methods for pKa
prediction in water or DMSO solvent.[44] It is worth noting that, in both cases,
significant differences were observed between predicted and experimental
tubes were illuminated with an infrared laser to generate a temperature
gradient. The diffusion of ligand/protein complex was monitored along the
temperature gradient through the measurement of fluorescence intensity
in each capillary tube. MST signals were recorded at medium laser power
for small molecules 2-4, whereas high laser power was used for the
macrocyclic peptide (1). A led power of 40% was used for all compounds
pKa values of compounds 2-4 (Table 2). Experimental acidic constant
(pKa) of compounds 2-4 were determined using Pion Sirius T3 instrument.
Each experiment was performed in triplicate, specifying sample weight,
number of expected pKa values, predicted pKa values, and titration mode.
In order to measure the pKa of a compound, a sample concentration of 1
mM was prepared. Poor soluble compounds (2, 4) were dissolved adding
methanol as co-solvent. In this case, three titrations were performed at
different methanol/water ratios (from 60:40 to 40:60, v/v)) in order to
determine aqueous pKa. The Yasuda-Shedlovsky method was chosen to
d
1-4. Dissociation constants (K ) were extrapolated by ligand concentration-
dependent changes of red-hPD-L1 normalized fluorescence (Fnorm). Raw
data were analyzed with the MO.Affinity Analysis software v2.3 in manual
mode setting the hot region at time 19/20s for small molecules (2-4) and
9/10s for macrocycle peptide (1). Confidence values (c.v., ±) are reported
2
next to K
certainty.
d d
values, indicating the range where the K falls with a 68% of
extrapolate aqueous pKa values. The R value was used to evaluate the
validity of the assay with co-solvent, and results were considered
acceptable with R2 value ≥ 0.9. Conversely, root mean square deviation
(
RMSD) value from the theoretical curve was used to assess the quality of
Label-free thermal shift analysis. The stability of red-hPD-L1 protein at
different pH after labelling procedure was checked using label-free thermal
shift analysis with Tycho instrument (NT.6; NanoTemper Technologies,
Munich, Germany). Specifically, label-free thermal shift analysis monitors
changes in emission intensity and wavelength maximum of intrinsic
fluorescence properties of buried Trp and Tyr residues in protein that
become exposed in the unfolded state when the protein is exposed to an
increasing temperature range from 35°C to 95°C. wt-hPD-L1and red-hPD-
L1 from labelling processes at pH 6.2, 7.2 and 8.2 were diluted in the
relative PB (PB6.2, PB7.2, PB8.2) at concentration of 2.7 μM. Samples were
loaded into Tycho NT.6 capillaries (TY-C001; NanoTemper Technologies,
Munich, Germany) and the thermal unfolding profiles for wt-hPD-L1 and
red-hPD-L1 were recorded. Experiments were carried out in triplicate and
inflection temperature (Ti) values are reported as mean ± standard
deviation values (Figure S3 and Table S3, supplementary material). As a
result, no shift of unfolding temperature (Ti < 1.0) was observed between
unlabeled wt-hPD-L1 and red-hPD-L1 at pH 6.2, pH 7.2 and pH 8.2. This
observation suggests that labelling process and pH values do not affect
the stability of the protein.
experiments performed for compound 3 dissolved without co-solvent.
Results were analyzed using Sirius T3 software v1.1.0.5.
Microscale Thermophoresis (MST). Recombinant human extracellular
domain of PD-L1 (wt-hPD-L1; cat. n. ab167713, 26 kDa) was purchased
from Abcam (Abcam, Cambridge, UK). wt-hPD-L1 was reconstituted
following Abcam’s protocol, adding sterile water to reach the final
theoretical concentration of 0.1 mg/0.32 ml. The protein amount was
assessed on the basis of its absorbance at 205 nm ( = 31 M-1 cm ) as
determined with NanoDrop-One spectrophotometer (v.1.3), yielding a
concentration of 12 μM. The protein was first labelled at pH 7.2 using the
protein labelling kit NT650 red-NHS 2nd generation dye supplied by
NanoTemper (MO-L011; NanoTemper Technologies, Munich, Germany),
using a 1:4 protein to dye ratio. Specifically, a volume of 100 µL of 12 µM
-1 [45]
solution of wt-hPD-L1 protein in phosphate buffer (PB7.2: 50 mM KH
.05% Tween 20, pH 7.2) was mixed with 100µL of 48µM NT650-NHS
fluorophore in PB7.2, and incubated for 30 minutes at room temperature
RT) in the dark. Then, unbounded fluorophore was removed by size-
2 4
PO ,
0
(
exclusion chromatography column using PB7.2 buffer as storage buffer. We
determined the labelled protein concentration by UV-measurement.
Specifically, the degree of labeling of protein (DoL) was 0.27. This value
was determined using the protein extinction coefficient 205nm = 31 M-1 cm-
Acknowledgements
1
-1
-1
,
dye 650nm = 195,000 M cm , a correcting factor at 205nm (cf205nm) =
This work was supported by Sterling S.p.A and PO FSE Umbria
0.19, and employing equations 1-3:
2
014-2020, D.D. n. 3806 20/04/2018. AC acknowledges a
research grant fellowship (PO FSE Umbria 2014-2020, D.D. n.
806 20/04/2018).
eq.1 Cprot = [A205 – (A650 x cf) / 31 M-1 cm-1 x MW],
3
eq.2 Cdye = [A650/ 195,000 M-1 cm-1 x l],
eq.3 DoL = Cdye/Cprot
Supporting Information
In order to investigate binding properties of compounds to PD-L1 at pH 6.2
and 8.2, the removal of unbounded fluorophore after labelling reaction was
performed using size-exclusion chromatography column conditioned with
Figure S1: solutions (#1-4, Table S1) of the docking pose of
compound 3 (S7911) into the binding cleft of the dimeric form of
PD-L1 (5J89). Figure S2: solutions (#1-4, Table S2) of the docking
pose of compound 4 (VIS1059) into the binding cleft of the dimeric
form of PD-L1 (5J89). Figure S3: unfolding profile of wt-hPD-L1
(wild type) and red-hPD-L1 (RED pH 6.2; RED pH 7.2; RED pH
2 4
two different phosphate buffers (PB6.2: 50 mM KH PO , 0.05% Tween 20,
pH 6.2; PB8.2: 50 mM KH PO , 0.05% Tween 20, pH 8.2). Specifically, the
2
4
labelled protein solution was split into two columns and then eluted using
PB6.2 and PB8.2, respectively. The DoL was determined for each collected
solution, yielding in both cases a value of 0.34. Pre-dilutions of compounds
8.2) at different pH conditions. Table S1: energetic scores
(
1-4) were prepared for MST experiments by 16-fold 1:1 serial dilution in
8
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
(
Gscore, kcal/mol) and root mean square deviation (RMSD, Å) of
[18] T. Chen, Q. Li, Z. Liu, Y. Chen, F. Feng, H. Sun, Eur. J. Med. Chem.
2019, 161, 378–398. https://doi.org/10.1016/j.ejmech.2018.10.044.
docking poses of compound 3 (S7911) into the binding cleft of the
dimeric form of PD-L1 (5J89). Table S2: energetic scores (Gscore,
kcal/mol) and root mean square deviation (RMSD, Å) of docking
poses of compound 4 (VIS1059) into the binding cleft of the
dimeric form of PD-L1 (5J89). Table S3: inflection temperature
[
19] J.M. Taube, A.J. Korman, S.N. Gettinger, J.D. Powderly, G.D. Kollia, R.A.
Anders, S.J. Antonia, M. Jure-Kunkel, R.D. Carvajal, P.D. Leming, L.
Chen, J.M. Wigginton, J.R. Brahmer, H. Xu, T.L. McMiller, S. Agrawal,
D.M. Pardoll, D.C. Smith, S.L. Topalian, F.S. Hodi, W.H. Sharfman, C.G.
Drake, J.A. Sosman, D. McDonald, L. Horn, M.B. Atkins, M. Sznol, A.
Gupta, D.R. Spigel, D.F. McDermott, N. Engl. J. Med. 2012, 366, 2443–
(
Ti), initial ratio and ∆ratio of wt-hPD-L1 and red-hPD-L1 at
2
454. https://doi.org/10.1056/nejmoa1200690.
20] Y. Chen, Y. Pei, J. Luo, Z. Huang, J. Yu, X. Meng, Front. Immunol. 2011
1, 1088. https://doi.org/10.3389/fimmu.2020.01088.
different pH conditions.
[
[
[
1
Keywords: Immunotherapy
•
Check-point inhibitors
•
21] H. Zhu, C. Du, M. Yuan, P. Fu, Q. he, B. Yang, J. Cao, Drug Discov.
Today 2020, 20, 30265-8. https://doi.org/10.1016/j.drudis.2020.07.006.
22] R.S. Herbst, J.-C. Soria, M. Kowanetz, G.D. Fine, O. Hamid, M.S.
Gordon, J.A. Sosman, D.F. Mcdermott, J.D. Powderly, S.N. Gettinger,
H.E.K. Kohrt, L. Horn, D.P. Lawrence, S. Rost, M. Leabman, Y. Xiao, A.
Mokatrin, H. Koeppen, P.S. Hegde, I. Mellman, D.S. Chen, F.S. Hodi,
Nature 2014, 515, 563–567. https://doi.org/10.1038/nature14011.
23] C. Corbet, O. Feron, Nat. Rev. Cancer. 2017, 17, 577–593.
https://doi.org/10.1038/nrc.2017.77.
Thermophoresis • Molecular dynamics • Cancer
References:
[
1]
G.J. Freeman, A.J. Long, Y. Iwai, K. Bourque, T. Chernova, H. Nishimura,
L.J. Fitz, N. Malenkovich, T. Okazaki, M.C. Byrne, H.F. Horton, L. Fouser,
L. Carter, V. Ling, M.R. Bowman, B.M. Carreno, M. Collins, C.R. Wood,
[
[
[
[
T.
Honjo,
J.
Exp.
Med.
2000,
192,
1027–34.
24] B.A. Webb, M. Chimenti, M.P. Jacobson, D.L. Barber, Nat. Rev. Cancer
https://doi.org/10.1084/jem.192.7.1027.
2011, 11, 671–677. https://doi.org/10.1038/nrc3110.
[
2]
Y. Latchman, C.R. Wood, T. Chernova, D. Chaudhary, M. Borde, I.
Chernova, Y. Iwai, A.J. Long, J.A. Brown, R. Nunes, E.A. Greenfield, K.
Bourque, V.A. Boussiotis, L.L. Carter, B.M. Carreno, N. Malenkovich, H.
Nishimura, T. Okazaki, T. Honjo, A.H. Sharpe, G.J. Freeman, Nat.
Immunol. 2001, 2, 261–268. https://doi.org/10.1038/85330.
25] D.E. Korenchan, R.R. Flavell, Cancers (Basel). 2019, 11, 1026.
https://doi.org/10.3390/cancers11071026.
26] J.E. Ippolito, M.W. Brandenburg, X. Ge, J.R. Crowley, K.M. Kirmess, A.
Som, D.A. D'Avignon, J.M. Arbeit, S. Achilefu, K.E. Yarasheski, J.
Milbrandt,
PLoS
ONE,
2016,
11,
e0159675.
[
[
[
[
3]
4]
5]
6]
V.A. Boussiotis, N. Engl. J. Med. 2016, 375, 1767–1778.
https://doi.org/10.1056/NEJMra1514296
https://doi.org/10.1371/journal.pone.0159675.
[
27] N. Raghunand, X. He, R. Van Sluis, B. Mahoney, B. Baggett, C.W. Taylor,
B.M. Carreno, M. Collins, Annu. Rev. Immunol. 2002 20, 29–53.
https://doi.org/10.1146/annurev.immunol.20.091101.091806.
G. Paine-Murrieta, D. Roe, Z.M. Bhujwalla, R.J. Gillies, Br. J. Cancer
1999, 80, 1005–1011. https://doi.org/10.1038/sj.bjc.6690455.
E.O. Long,
https://doi.org/10.1146/annurev.immunol.17.1.875.
Y. Agata, A. Kawasaki, H. Nishimura, Y. Ishida, T. Tsubata, H. Yagita, T.
Honjo, Int. Immunol. 1996, 8, 765–72.
http://www.ncbi.nlm.nih.gov/pubmed/8671665.
Annu. Rev. Immunol. 2002,
17, 875–904.
[
28] H.X. Zhou, X. Pang, Chem. Rev. 2018, 118, 1691–1741.
https://doi.org/10.1021/acs.chemrev.7b00305.
[
29] K. Magiera-Mularz, L. Skalniak, K.M. Zak, B. Musielak, E. Rudzinska-
Szostak, Ł. Berlicki, J. Kocik, P. Grudnik, D. Sala, T. Zarganes-Tzitzikas,
S. Shaabani, A. Dömling, G. Dubin, T.A. Holak, Angew. Chem. Int. Ed.
Engl. 2017, 56, 13732–13735. https://doi.org/10.1002/anie.201707707.
30] X. Chupak, L. S.; Zheng, Compounds Useful as Immunomodulators,
WO2015034820, Bristol-Myers Squibb Company, US, 2015.
[
7]
H. Nishimura, Y. Agata, A. Kawasaki, M. Sato, S. Imamura, N. Minato,
H. Yagita, T. Nakano, T. Honjo, Int. Immunol. 1996, 8, 773–80.
https://doi.org/10.1093/intimm/8.5.773.
[
[
[
[
8]
9]
M. Collins, V. Ling, B.M. Carreno, Genome Biol. 2005, 6, 1–7.
https://doi.org/10.1186/gb-2005-6-6-223.
31] K. M. Zak, P. Grudnik, K. Guzik, B.J. Zieba, B. Musielak, A. Dömling, G.
Dubin,
T.A.
Holak,
Oncotarget.
2016,
7,
30323–30335.
K.M. Zak, P. Grudnik, K. Magiera, A. Dömling, G. Dubin, T.A. Holak,
https://doi.org/10.18632/oncotarget.8730.
Structure.
2017,
25,
1163–1174.
[
32] S. Ueno, M. Ohtsubo, R. Kuwano, [4+2] Cycloaddition of o-xylylenes with
https://doi.org/10.1016/j.str.2017.06.011.
imines using palladium catalyst. J. Am. Chem. Soc. 2009, 131, 12904–
[
[
10] S. Tang, P.S. Kim, Proc. Natl. Acad. Sci. USA. 2019, 116, 24500-24506.
https://doi.org/10.1073/pnas.1916916116.
12905. https://doi.org/10.1021/ja905988e.
[
33] S. Duhr, D. Braun, Proc. Natl. Acad. Sci. USA. 2006, 103, 19678–19682.
https://doi.org/10.1073/pnas.0603873103.
11] D.Y. Lin, Y. Tanaka, M. Iwasaki, A.G. Gittis, H.-P. Su, B. Mikami, T.
Okazaki, T. Honjo, N. Minato, D.N. Garboczi, Proc. Natl. Acad. Sci. USA.
[
34] A. Ganesan, M. Ahmed, I. Okoye, E. Arutyunova, D. Babu, W.L. Turnbull,
J.K. Kundu, J. Shields, K.C. Agopsowicz, L. Xu, Y. Tabana, N. Srivastava,
G. Zhang, T.C. Moon, A. Belovodskiy, M. Hena, A.S. Kandadai, S.N.
Hosseini, M. Hitt, J. Walker, M. Smylie, F.G. West, A.G. Siraki, M.J.
Lemieux, S. Elahi, J.A. Nieman, D.L. Tyrrell, M. Houghton, K. Barakat,
Sci. Rep. 2019, 9, 12392. https://doi.org/10.1038/s41598-019-48826-6.
35] A. Macchiarulo, R. Pellicciari, J. Mol. Graph. Model. 2007, 26, 728–739.
https://doi.org/10.1016/j.jmgm.2007.04.010.
2008, 105, 3011-3016. https://doi.org/10.1073/pnas.0712278105.
[
12] X. Cheng, V. Veverka, A. Radhakrishnan, L.C. Waters, F.W. Muskett, S.
H. Morgan, J. Huo, C. Yu, E.J. Evans, A. J. Leslie, M. Griffiths, C.
Stubberfield, R. Griffin, A.J. Henry, A. Jansson, J.E. Ladbury, S. Ikemizu,
M.D.
Carr,
J.
Biol.
Chem.
2013,
288,
11771-11785.
https://doi.org/10.1074/jbc.M112.448126.
[
[
[
[
[
[
[
13] K. M. Zak, R. Kitel, S. Przetocka, P. Golik, K. Guzik, B. Musielak, A.
Dömling, G. Dubin, T.A. Holak, Structure. 2015, 23, 2341-2348.
https://doi.org/10.1016/j.str.2015.09.010.
36] K. M. Zak, P. Grudnik, K. Guzik, B. J. Zieba, B. Musielak, A. Dömling, G.
Dubin, T. A. Holak, Oncotarget. 2016, 7, 30323–30335.
https://doi.org/10.18632/oncotarget.8730.
14] M. Ghiotto, L. Gauthier, N. Serriari, S. Pastor, A. Truneh, J.A. Nunè, D.
Olive,
Int.
Immunol.
2010,
22,
651–660.
[
37] H. M. Berman, T. Battistuz, T. N. Bhat, W. F. Bluhm, P. E. Bourne, K.
Burkhardt, Z. Feng, G. L. Gilliland, L. Iype, S. Jain, P. Fagan, J. Marvin,
D. Padilla, V. Ravichandran, B. Schneider, N. Thanki, H. Weissig, J. D.
Westbrook, C. Zardecki, Acta Crystallogr. D Biol. Crystallogr. 2002, 58,
https://doi.org/10.1093/intimm/dxq049.
15] T. Yamazaki, H. Akiba, H. Iwai, H. Matsuda, M. Aoki, Y. Tanno, T. Shin,
H. Tsuchiya, D.M. Pardoll, K. Okumura, M. Azuma, H. Yagita, J. Immunol.
2002 169, 5538–45. https://doi.org/10.4049/jimmunol.169.10.5538.
8
99–907. https://doi.org/10.1107/S0907444902003451.
38] G. Madhavi Sastry, M. Adzhigirey, T. Day, R. Annabhimoju, W. Sherman,
J. Comput. Aided. Mol. Des. 2013, 27, 221–234.
https://doi.org/10.1007/s10822-013-9644-8.
16] X. Jiang, J. Wang, X. Deng, F. Xiong, J. Ge, B. Xiang, X. Wu, J. Ma, M.
[
Zhou, X. Li, Y. Li, G. Li, W. Xiong, C. Guo, Z. Zeng, Mol. Cancer. 2019,
1
8, 1–17. https://doi.org/10.1186/s12943-018-0928-4.
17] D. F. Quail, J. A. Joyce, Nat. Med. 2013, 19, 1423-1437.
0.1038/nm.3394. https://doi.org/10.1038/nm.3394.
[
39] R.A. Friesner, J.L. Banks, R.B. Murphy, T.A. Halgren, J.J. Klicic, D.T.
Mainz, M.P. Repasky, E.H. Knoll, M. Shelley, J.K. Perry, D. E. Shaw,
1
9
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
Perry Francis, P. S. Shenkin, J. Med. Chem. 2004, 47, 1739-1749.
https://doi.org/10.1021/JM0306430.
[
40] K. J. Bowers, E. Chow, H. Xu, R. O. Dror, M. P. Eastwood, B. A.
Gregersen, J. L. Klepeis, I. Kolossvary, M. A. Moraes, F. D. Sacerdoti, J.
K. Salmon, Y. Shan, D. E. Shaw, ACM/IEEE Conf. Supercomput, ACM
Press,
https://doi.org/10.1145/1188455.1188544.
41] R. Anandakrishnan, B. Aguilar, A. V Onufriev, Nucleic Acids Res. 2012,
0, W537-W541. https://doi.org/10.1093/nar/gks375.
42] J.C. Gordon, J.B. Myers, T. Folta, V. Shoja, L.S. Heath, A. Onufriev,
Nucleic Acids Res., 2005, 33, W368-W371.
https://doi.org/10.1093/nar/gki464.
43] J. Myers, G. Grothaus, S. Narayanan, A. Onufriev, Proteins. 2006, 63,
28–938. https://doi.org/10.1002/prot.20922.
New
York,
USA,
2006,
84.
[
[
4
[
[
9
44] J.C. Shelley, A. Cholleti, L. Frye, J.R. Greenwood, M.R. Timlin, M.
Uchimaya, J. Comp. Aided Mol. Des. 2007, 21, 681-691.
https://doi.org/10.1007/s10822-007-9133-z.
[45] R.
K.
Scopes,
Anal.
Biochem.
1974,
59,
277–282.
https://doi.org/10.1016/0003-2697(74)90034-7.
1
0
This article is protected by copyright. All rights reserved.

ChemMedChem
10.1002/cmdc.202000760
FULL PAPER
Entry for the Table of Contents
While inhibitors of PD-1/PD-L1 interaction are being developed as anticancer immunotherapies, alteration of pH in the tumor
microenvironment is thought to promote drug resistance. In this study, we show that pH variation affects dissociation constants of PD-
L1 inhibitors, influencing polar interactions. Results suggest that a decrease of pH may enforce target engagement of PD-L1 inhibitors
bearing amine groups, thereby allowing to escape acidosis as tumor resistance mechanism.
1
1
This article is protected by copyright. All rights reserved.