Design of Thioether Cyclic Peptide Scaffolds with Passive Permeability and Oral Exposure

Article abstract of DOI:10.1021/acs.jmedchem.0c01505

Advances in the design of permeable peptides and in the synthesis of large arrays of macrocyclic peptides with diverse amino acids have evolved on parallel but independent tracks. Less precedent combines their respective attributes, thereby limiting the p

Full text of DOI:10.1021/acs.jmedchem.0c01505

pubs.acs.org/jmc
Article
Design of Thioether Cyclic Peptide Scaffolds with Passive
Permeability and Oral Exposure
Andrei A. Golosov,# Alec N. Flyer,# Jakal Amin, Charles Babu, Christian Gampe, Jingzhou Li,
Eugene Liu, Katsumasa Nakajima, David Nettleton, Tajesh J. Patel, Patrick C. Reid, Lihua Yang,
and Lauren G. Monovich*,#
Cite This: J. Med. Chem. 2021, 64, 2622−2633
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Advances in the design of permeable peptides and
in the synthesis of large arrays of macrocyclic peptides with diverse
amino acids have evolved on parallel but independent tracks. Less
precedent combines their respective attributes, thereby limiting the
potential to identify permeable peptide ligands for key targets.
Herein, we present novel 6-, 7-, and 8-mer cyclic peptides (MW
774−1076 g·mol−1) with passive permeability and oral exposure
that feature the amino acids and thioether ring-closing common to
large array formats, including DNA- and RNA-templated synthesis.
Each oral peptide herein, selected from virtual libraries of partially N-methylated peptides using in silico methods, reflects the subset
consistent with low energy conformations, low desolvation penalties, and passive permeability. We envision that, by retaining the
backbone N-methylation pattern and consequent bias toward permeability, one can generate large peptide arrays with sufficient side
chain diversity to identify permeability-biased ligands to a variety of protein targets.
Consistent with the hypothesis that permeability is a limiting
factor for the oral exposure of cyclic peptides, we present 6-, 7-,
and 8-mer thioether macrocyclic peptides that are both
permeable and orally exposed. Starting from a virtual library
of macrocycles with varied, partial backbone N-methylation,
each example macrocycle herein was selected for permeability
using in silico methods and then synthesized and validated by
in vitro and/or in vivo experiment. The specific N-methylation
sites within each macrocyclic peptide govern conformation and
intramolecular hydrogen-bond network and also bias them for
permeability. Therefore, each case described herein is a
foundational member of a family of analogues that share the
same macrocycle backbone, N-methylation pattern, and
permeability bias. We show herein, through a brief survey of
lipophilic side chain analogues with shared backbone and N-
methylation sites, that the permeability and oral exposure of
the parent example can be ported to its family members. More
broadly, we chose the specific amino acids and thioether ring-
closing to be compatible with a variety of library technologies,
including synthesis and screening by in vitro translation
INTRODUCTION
■
Interest in cyclic peptides as therapeutics1,2 is renewed by a
growing awareness of design principles governing their passive
cellular permeability3−6 and by exponential advances in the
construction and screening of cyclic peptide libraries, especially
those including noncanonical amino acids.7−9 However,
published cases wherein cyclic peptides combine passive
permeability and oral bioavailability with high affinity for a
target protein are dominated by natural products.10−15 The
present contribution provides an approach to identify
permeable peptide ligands to therapeutic targets. Traditionally,
permeable macrocyclic peptides are identified by discrete
synthesis and careful side chain variation of privileged, natural
product scaffolds.16−21 Macrocyclic peptides with passive
permeability sample conformations with low desolvation
penalties upon transfer from aqueous to membrane environ-
ments.4,22 More recently, the basic principles governing passive
permeability were applied to the prospective design of
macrocyclic peptides with oral exposure.23,24 However, the
limited diversity of permeable macrocyclic peptide scaffolds
presents a challenge for their identification as protein ligands.
By contrast, cell-free, in vitro translation of peptides on
modified ribosomes offers enormous, highly diverse libraries of
macrocyclic peptides containing both canonical and non-
canonical amino acids and entry to macrocyclic peptide ligands
with exquisite affinity and selectivity.25−28 However, in this
context, ligands are discovered in the absence of selection
pressure for cellular permeability.
Received: August 28, 2020
Published: February 25, 2021
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
© 2021 American Chemical Society
J. Med. Chem. 2021, 64, 2622−2633
2622

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
platforms, to better enable the identification of permeable
ligands to therapeutic targets.
(Pro) and, for synthetic ease on bead or in vitro, no more than
three total residues with N-methylation.
The unique conformational bias introduced by proline
segregated its four unique positions along the chain (AA2−5),
plus proline null, into five separate congeneric series. The
phenylalanine stereochemistry (L- or D-) bifurcated each of
the five proline series, to produce a total of 2 × 5, or 10,
congeneric series. For example, Series A and Series B feature L-
or D-phenylalanine, respectively, in the absence of proline at
AA2−5. In silico selections were performed within each
congeneric series. Computational models of all possible N-
methylated variants with 0−3 N-methylated residues were
generated for each of the ten congeneric series. Conformations
of the resulting 292 virtual peptides were sampled in low
dielectric medium as a model of the nonpolar portion of the
membrane. The transfer free energy from water to membrane
was computed as ΔGtransfer for each conformation.31 There-
after, we performed an analysis of conformations with the
lowest energy and the lowest ΔGtransfer values for each virtual
compound. The term ΔGt*ransfer reflects the transfer free energy
calculation combined with conformational analysis. We
anticipated ΔG*transfer to be useful for identifying compounds
with low ΔGtransfer values and conformations with the fewest
solvent-exposed backbone NHs. We considered both ΔGtransfer
and ΔG*transfer values during the selection of compounds for
synthesis to evaluate the utility of each to select for permeable
macrocycles.
RESULTS AND DISCUSSION
■
In Silico Optimization of the N-Methylation Pattern.
By analogy to published work on the de novo discovery of
passive cell membrane permeable peptides,29 we selected 6-
mer macrocycles to begin our investigations. Similar to prior
efforts, we hypothesized that an optimal N-methylation pattern
for permeability could stabilize a macrocyclic conformation
with a low desolvation penalty for transit from an aqueous to
membrane environment. However, we substituted the
empirical on-resin N-methylation described previously with
an in silico analysis of virtual compound libraries; this also
removes the conformational biases of the unmethylated on-
resin parent peptide. The subset of virtual compounds with N-
methylation patterns predicted to have low desolvation penalty
and, therefore, enhanced permeability (plus comparator
control compounds) was selected for chemical synthesis.
A common ring-closing strategy for macrocyclic peptides
generated by in vitro translation features the reaction of a C-
terminal cysteine onto an N-terminal electrophilic capping
group to yield a cysteine thioether.30 No examples of orally
bioavailable, thioether-containing macrocycles have been
described to date, so we decided to investigate a thioether
ring-closing strategy by analogy to those commonly used with
in vitro translated peptides.
Synthesis. Synthesis of reference 6-mers, plus a subset of
N-methylated 6-mer comparators selected in silico, was
conducted on resin using an automated peptide synthesizer
(Scheme 1). The C-terminal-protected cysteine was coupled to
The cysteine (Cys) used to effect the macrocyclic ring
closure was held constant as the C-terminal residue and
capped as a methyl amide, which we chose in order to limit the
number of H-bond donors at the C-terminus to one (Chart 1).
a
Scheme 1. General Synthetic Route for 6-mer Macrocycles
Chart 1. Workflow for Identification of Permeable 6-mer
Macrocycles
a
a
Workflow steps: (a) select amino acid residue AA1 from Phe, Phe,
phe, or phe; (b) select amino acid residues AA2−5 from Nle, Nle, or
Pro; (c) exclude cases with more than one Pro; (d) allow R = H or
CH3; (e) exclude combinations with more than 3 N-methylations; (f)
sample conformations for all virtual library members; and (g) triage
based on ΔG*transfer and conformational analysis.
a
Reagents and conditions: (a) automated peptide synthesis (details in
experimental); (b) 5 equiv 2,5-dioxopyrrolidin-1-yl-2-chloroacetate,
NMP, 23 °C; (c) TFA/H2O/TIS 92.5:2.5:5; and (d) Et3N, DMSO,
23 °C, 5−15% yield (over 4 steps).
We evaluated virtual compounds with both unmethylated and
backbone N-methylated Cys in silico, although N-methyl Cys
may not be optimal for in vitro translation systems. To provide
a UV chromophore, the N-terminal amino acid AA1 was
selected from phenylalanine variants, including either enan-
tiomer of phenylalanine (Phe or D-Phe (phe)) or N-
methylated L- or D-phenylalanine (Phe or phe). Residues
AA2‑5 were selected from Pro, norleucine (Nle), or norleucine
with backbone methylation (Nle). Norleucine, with and
without backbone methylation, was chosen as a model
hydrophobic residue for its compatibility with specialized in
vitro translation methods and structural similarity to the Leu
residues in oral natural products [e.g., cyclosporine A (CsA)]
and oral peptides (e.g., 1NMe3). We allowed up to one proline
the resin, followed by the series of five variable residues. The
first four variable residues were chosen from norleucine (Nle),
N-methylnorleucine (Nle), or proline (Pro). The N-terminal
residue, selected from phenylalanine (Phe), D-phenylalane
(phe), N-methylphenylalanine (Phe), or N-methyl-D-phenyl-
alanine (phe), was capped with a chloroacetyl electrophile.
Subsequent deprotection of the cysteine and release from the
resin, followed by ring-closing nucleophilic attack of the
cysteine side chain onto the chloroacetyl moiety, afforded the
desired macrocyclic thioethers, representing Series A−J.
Measured Properties. Data for representative compounds
from congeneric Series A−G are presented as Table 1. Series
H−J did not produce N-methylated structures with promising
2623
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Table 1. Calculated and Measured Property Data for Congeneric Series A−G
series
A
Cmpd
AA1
AA2
AA3
AA4
AA5
R
ΔG*transfer (kcal·mol−1
)
exposed NHs (count)
PAMPA log Papp
a
d
1
Phe
Phe
Phe
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Pro
Pro
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Pro
Pro
Pro
Pro
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
H
CH3
H
H
H
H
H
CH3
H
H
H
H
H
H
H
H
H
H
−4.30
−6.25
−5.44
−4.45
−7.55
−5.78
−4.26
−2.65
−5.16
−2.24
−3.41
−2.29
−3.23
−0.27
−3.22
−0.63
−2.03
3.69
3
0
0
2
0
1
1
2
3
4
0
3
0
4
0
2
0
4
<−5.9
−4.3
−4.3
<−6.0
−4.6
−5.1
1a
1b
2
2a
2b
2c
2d
2e
3
3a
4
4a
5
b
B
phe
phe
phe
phe
phe
phe
Phe
Phe
phe
phe
Phe
Phe
phe
phe
Phe
e
−5.1
<−6.0
<−5.7
<−6.7
−5.5
<−6.7
<−5.6
−5.9
−4.5
−5.8
−4.6
−5.8
f
C
D
E
Pro
c
Pro
Pro
Pro
Nle
Nle
Nle
Nle
Nle
Nle
5a
6
6a
7
F
G
7a
Phe
b
CH3
−3.56
0
−4.5
a
c
d
e
f
Phe (phenylalanine). phe (D-phenylalanine). Phe (N-Me-phenylalanine). Nle (norleucine). Nle (N-Me-norleucine). Pro (proline).
ΔGt*ransfer values and so were not prepared. For the parent
compound 1 of congeneric Series A (AA1 = Phe, AA2−5 = Nle),
the calculated −4.30 kcal·mol−1 ΔGt*ransfer is associated with
three exposed backbone NHs and undetectable permeability in
the PAMPA assay (log Papp < −5.9). Judicious addition of two
backbone methylations in either compound 1a (AA4 and Cys)
or 1b (AA3 and AA5) lowered ΔGt*ransfer by 1.95 or 1.14 kcal·
mol−1, respectively, and eliminated solvent-exposed backbone
NHs. Accordingly, the PAMPA data for 1a and 1b (log Papp
−4.3 and −4.9) are consistent with permeable compounds. In
addition, the NMR solution structure of 1a (Figure 1) in
respectively) than parent 2 (ΔG*transfer −4.45 kcal·mol−1 with
two solvent-exposed backbone NHs). The higher measured
permeability for 2a and 2b (PAMPA log Papp −4.6 and −5.1,
respectively) than for parent 2 (PAMPA log Papp < −6.0)
follows directly from the expectation set by the calculated
values. Example 2c (ΔG*transfer −4.26 kcal·mol−1) highlights the
number of solvent-exposed backbone NHs as an important
predictor of permeability (PAMPA log Papp −5.1) relative to
control 2. Examples 2d and 2e lend further support to the
trend. For Series C−G, we selected for synthesis N-methylated
examples with both lower ΔG*transfer values and number of
solvent-exposed backbone NHs; each pair mirrored the
anticipated consequences on permeability.
A representative set of ΔG*transfer values for two of the ten
congeneric series, Series A and B, are plotted versus the
experimentally determined permeability in the PAMPA assay,
as shown in Figure 2. In general, the lowest (most negative)
Figure 1. Overlay of calculated (green) and a representative
experimentally determined (cyan) conformation of 1a in low
dielectric media.
chloroform (green) overlays well with the calculated
conformation (cyan). A network of four intramolecular
hydrogen bonds (two transannular and two extraannular)
engages all four hydrogen-bond donors of 1a, including the
secondary amide NH of the C-terminal Cys. Overall, the low
energy conformation of 1a presents no hydrogen-bond donors
to the solvent.
Likewise, for the remaining congeneric series in Table 1, the
unmethylated parent examples 2−7 exhibit higher ΔGt*ransfer
values and number of exposed backbone NHs, as well as lower
PAMPA log Papp values than the corresponding methylated
analogues. For example, methylated analogues 2a and 2b
exhibit lower ΔGt*ransfer values (−7.55 and −5.78 kcal·mol−1,
respectively) and exposed hydrogen-bonds (0 and 1,
Figure 2. Calculated ΔGt*ransfer values and number of exposed NHs in
lowest energy conformation for congeneric Series A and B.
ΔGt*ransfer values track with the predicted conformations with
the lowest number of solvent-exposed backbone NHs. For
example, low PAMPA permeability was observed for macro-
cycles with calculated conformations with ≥ two solvent-
exposed backbone NHs; conversely, high PAMPA permeability
is observed for those calculated conformations with no solvent-
exposed NHs.
2624
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Measured log D (pH 7.4)32 for the seven example
compounds in Table 2 lie below calculated values by an
access of the side chains to metabolic oxidation. In fact, the
clearance of 1b is only about 10% of the rate of hepatic blood
flow in rats.34 Pharmacokinetic parameters for 7 and 7a match
the expected trend, with higher systemic exposure and lower
clearance observed for backbone methylated 7a relative to
parent 7.
Encouraged by the 6-mer results, we applied the ΔGt*ransfer
and conformational analysis-based strategy to larger macro-
cycles, beginning with 7-mers containing 0−4 N-methylated
residues. Overall, following the methods and design for the 6-
mers, calculations were performed on a virtual library of 768 7-
mer macrocyclic peptides. Based on the calculated values and
conformations (plus reference controls), 10 peptides were
selected for synthesis and profiled in vitro and in vivo. Results
from a representative 7-mer congeneric series with Phe1 and
Pro3 are shown in Table 4. For the 7-mers in Table 4, a larger
Table 2. In Vitro ADME Data for 1, 1a, 1b, 2, 2a, 7, and 7a
1
1a
1b
2
2a
7
7a
MW (g·mol−1
clog P
)
774
7.3
3.6
105
0.6
816
9.3
4.1
69
802
8.6
4.6
74
774
7.3
3.6
107
0.4
802
8.6
4.2
75
758
6.2
3.1
96
800
8.2
3.6
66
a
b
log D
c
EPSA (Å)2
d
MDCK-LE
8.0
3.9
1.5
0.6
6.3
(×10−6 cm·s−1
)
e
solubility (μM)
<4
12
5
5
18
200
216
f
RLM and HLM Cl
350 to >700
(μL·min−1·mg−1
)
a
b
c
Calculated log P by the method available in ChemDraw 17.1.
Measured log D at pH 7.4 using the column-based method.
d
Experimental polar surface area (EPSA).
PappA−B permeability in
Table 4. Profiling Data for 7-mers 8 and 8a−c
low efflux transporter expressing Madin−Darby canine kidney cells
(MDCK-LE). Aqueous solubility at pH 6.8. Clearance in rat (RLM)
and human liver microsomes (HLM).
e
f
average of 4.1 units, highlighting the risks of utilizing additive,
calculated values to predict the properties of large compounds.
Within each congeneric series, N-methylation increased log D
(pH 7.4) by ∼0.5 units, decreased EPSA by an average of 32
Å2 (to below the proposed 80 Å2 guidance),33 and increased
permeability in the MDCK-LE assay relative to the
unmethylated comparator. Overall, examples in Table 2
showed high clearance in both human and rat liver microsomes
and, excepting 7 and 7a, low aqueous solubility.
8
8a
8b
8c
ΔG*transfer (kcal·mol−1
exposed NHs (count)
)
−2.37 −2.49 −4.12 −4.96
2
1
0
0
MW (g·mol−1
)
871
7.9
3.7
91
899
9.2
4.7
77
913
9.8
4.8
69
947
9.6
4.9
81
a
Rat pharmacokinetic parameters for 1, 1a, 1b, 7, and 7a are
presented in Table 3. The oral exposure of 1 fell below the
clog P
b
log D
c
EPSA (Å2)
Table 3. Rat Pharmacokinetic Parameters for 1, 1a, 1b, 7,
and 7a Dosed 5 mg/kg PO or 0.5 mg/kg IV
PAMPA log Papp
−6.2
0.2
−5.3
0.8
9−24
480 to >700
−4.0
6.4
−4.4
2.3
d
MDCK-LE Papp A−B (×10−6 cm·s−1
)
e
solubility (μM)
c
d
d
d
d
1
1a
1b
7
7a
f
RLM and HLM Cl (μL·min−1·mg−1
)
a
systemic Cmax (nM)
hepatic portal vein Cmax
(nM)
systemic AUC (nM·h)
hepatic portal vein AUC
(nM·h)
bql
nd
114
140
1028
819
1.7
nd
152
nd
a
b
c
Calculated log P by the method available in ChemDraw 17.1.
Measured log D at pH 7.4 using the column-based method.
Experimental polar surface area (EPSA).
b
b
b
d
a
PappA−B permeability in
bql
nd
258
173
2288
1859
0.6
92
nd
low efflux transporter expressing Madin−Darby canine kidney cells
b
b
b
nd
e
f
(MDCK-LE). Aqueous solubility at pH 6.8. Clearance in rat (RLM)
and human liver microsomes (HLM).
Cl (mL·min−1·kg−1
t1/2 (h)
)
702 239
0.3 0.2
86
8.9
3.8
1
222
0.2
2
≪1
d
72
0.3
1.3
6
0.9
2.1
20
b
Vss (L·kg−1
)
nd
disparity exists between the clog P and measured log D
(average 4.6 units) than for the 6-mers in Table 1 (average 4.1
units). The N-methylated analogues 8a−c, characterized by
both lower ΔG*transfer and fewer number of exposed backbone
NHs, reflect earlier observations that N-methylation decreases
EPSA below 80 Å2 and increases both log D and permeability
relative to the unmethylated comparator 8. However, the
doubly N-methylated example 8a (AA2, AA4 = Nle) minimally
impacts ΔGt*ransfer and leaves a residual solvent-exposed
backbone NH. Therefore, example 8a, with improved
PAMPA permeability versus 8, shows lower permeability
relative to the examples 8b, c (AA2, AA4, and AA6 = Nle).
Example 8c, a representative aromatic side chain Phe, replaces
the aliphatic Nle of 8b with minimal impact on the profile. For
both 6- and 7-mer series, we found that PAMPA and MDCK-
LE values were generally correlated (Supporting Information
Figure 1). The present data suggest EPSA ≤ ∼80 Å2 as an
independent predictor of permeability. However, an expanded
set of Novartis peptides show that EPSA ≤ ∼80 Å2 must be
% F
a
≪1
19
b
c
Below the limit of quantitation. Not determined. n = 3. n = 2.
limit of quantitation (bql). Likewise, high clearance of 1
precluded calculation of its pharmacokinetic parameters. By
contrast, the N-methylated congeners 1a and 1b achieve far
higher systemic exposures, Cmax (114 and 1028 nM,
respectively) and AUC (258 and 2288 nM·h, respectively).
Hepatic portal vein concentrations of 1a or 1b following oral
administration reflect 67−100% of the systemic Cmax and AUC
values, suggesting that the ∼20% F for 1a and 1b follows from
low fraction absorbed rather than first pass metabolism.
Despite uniformly high intrinsic clearance across the series
(Table 2), the in vivo clearances of 1a and 1b differentiate from
that of 1 and from each other. The lower clearances observed
in the pharmacokinetic studies, which are lower than would be
predicted by the liver microsomal data, may stem from their
hydrophobicity, high plasma protein binding, and restricted
2625
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
combined with a minimum lipophilicity (log D) to achieve
passive membrane permeability (Supporting Information
Figure 2).
Table 6. Structures of 8-mers 9, 9a, 10, and 10a−c
Rat pharmacokinetic parameters for 7-mers 8 and 8a−c are
presented in Table 5. Like 1 and 7, unmethylated parent 8
Table 5. Rat Pharmacokinetic Parameters for 8 and 8a−c
AA1
AA2
AA3
AA4
AA5
AA6
AA7
Dosed 5 mg/kg PO or 0.5 mg/kg IV
9
9a
10
10a
10b
10c
Phe
Phe
phe
phe
phe
phe
Nle
Nle
Nle
Nle
Nle
Thr
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Nle
Phe
Phe
a
b
b
b
8
8a
8b
8c
systemic Cmax (nM)
systemic AUC (nM·h)
bql
bql
72 21
0.08 0.04
0.2 0.05
≪1
18
10
106
0.7
4.4
1
868
2130
35
0.6
1.2
86
1946
1308
19
1.5
1.1
Cl (mL·min−1·kg−1
t1/2 (h)
)
Vss (L·kg−1
)
(AA2 = Thr, AA7 = Phe) to briefly explore the impact of the
side chain on properties.35−37
% F
84
a
b
n = 3. n = 2.
Calculated, in vitro, and in vivo properties for 9a and 10a−c
shows no/weak systemic exposure following oral dosing and
short half-life (t1/2 0.08 0.04 h) following IV administration.
Example 8a shows detectable oral exposure, but both 8b and c
(MW 913 and 947 g·mol−1, respectively; clog P 9.8 and 9.6,
respectively) show ∼85% F in the context of low/moderate
clearance. Again, the in vivo rank order is best predicted by
ΔGt*ransfer, number of exposed backbone NHs, and consequent
relative in vitro permeability, rather than molecular weight, clog
P, log D, or liver microsome stability.
are presented in Table 7. For comparison, their unmethylated
Table 7. In Vitro and Rat in Vivo Pharmacokinetic Profile for
9a and 10a−c
9a
10a
10b
10c
ΔG*transfer (kcal·mol−1
exposed NHs (count)
)
−9.98
0
−9.69
0
−9.82
0
−7.27
0
MW (g·mol−1
clog P
)
1056
13
1042
13
1076
12
1064
10
To explore the oral exposure of 8b in nonrodent species, we
assessed its pharmacokinetic profile in dogs (Figure 3).
log D
EPSA (Å2)
6.1
nd
6.0
73
6.2
85
5.2
86
MDCK-LE Papp A−B (×10−6 cm·s−1
solubility (μM)
)
<0.35
<4
120 to >700
RLM/HLM Cl (μL·min−1·mg−1
)
a
Cmax (nM)
436
1077
12
411
1269
33
77
161
372
28.4
13
a
AUC (nM·h)
237
65.4
21
a
Cl (mL·min−1·kg−1
)
a
% F
17
54
a
n = 2.
Figure 3. Intravenous and oral exposure profiles for 8b in beagle dog
(n = 3).
parent scaffolds 9 and 10 had calculated ΔG*transfer values in the
range of −7 to −8 kcal·mol−1 and at least three solvent-
exposed backbone NHs. Each N-methylated example in Table
7 exceeds 1000 g·mol−1 molecular weight and calculated log P
values of 10, but is consistent with no exposed backbone NHs.
Not unexpectedly, measured log D values at pH 7.4 (all ≥5.2)
exceed the values reported above for the 6- and 7-mers and
reflect the largest gap from the calculated values (average 6.1
units). Compound 10a achieves EPSA below 80 Å2, in line
with guidance for peptides with passive permeability, with
values near 80 Å2 for 10b and 10c (85 and 86 Å2, respectively).
However, compounds 10a−c show undetectable Papp A−B
in the MDCK-LE assay (with ≥67% recovery) and consistently
poor stability in liver microsomes. All four examples in Table 7
are orally exposed at levels consistent with a subset of potential
therapeutic applications. Furthermore, despite their high
intrinsic clearance, three of the four examples (9a, 10a, and
10c) show low to moderate clearance in rats. Compounds
10a−c, which share identical backbone N-methylation patterns
and vary only by side chain identity, illustrate the potential for
expanding validated N-methylated backbones to larger families
of orally exposed analogues.
Macrocycle 8b was dosed 0.1 mg/kg IV or 0.5 mg/kg PO in
5% neat solutol and 95% PBS. Following the IV dose, the
clearance (24
6 mL·min−1·kg−1) was moderate relative to
liver blood flow34 and the half-life was 1.1 0.3 h. Following
the PO dose, a fraction of the compound was rapidly absorbed,
reaching a Cmax of 6 4 nM at a Tmax of 0.7 0.3 h. The oral
bioavailability was only 3 2%. This may be due to the strong
efflux observed in a Caco-2 permeability assay (efflux ratio
>25) reducing the amount absorbed in the dog at a low dose
while the 10× higher dose in rat could have saturated the efflux
transporter.
With successful identification of oral 7-mers, we further
expanded our studies to 8-mers by selecting from 1514 virtual
compounds with 0−4 N-methylated residues. In Table 6 are
shown two congeneric series, marked by either Phe (9) or phe
(10) at the N-terminal amino acid residue and no Pro residues.
For each series, an N-methylation pattern predicted by in silico
methods to have no exposed backbone NHs were prepared as
9a and 10a, respectively. Thereafter, additional side chain
variants of 10a were prepared as 10b (AA7 = Phe) and 10c
2626
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
translation methods, to effect the synthesis of families of
macrocycles with preset N-methylation patterns link property/
permeability-biased structure design with powerful ligand
identification technology, an important step toward identifying
passively permeable ligands to therapeutically relevant protein
targets.
CONCLUSIONS
■
We show that multiple series of orally exposed cyclic peptides
with MWs at and above 1 kDa can be reliably identified by 3D
physics-based in silico methods in advance of chemical
synthesis. By contrast, standard 2D in silico descriptors (e.g.,
MW, clog P, and TPSA) that neglect 3D macrocycle
conformation, used alone or in combination, poorly predict
macrocycle permeability and oral exposure. Like rule-
compliant, orally exposed small molecules,38 orally exposed
macrocycles balance lipophilicity and polar surface area,
properties that can be assessed using measured log D and
EPSA, respectively. However, due caution is warranted when
utilizing strict cutoffs. For the largest, most lipophilic oral
macrocycles examined, log D and EPSA exceed proposed
limits. In advance of chemical synthesis, ΔG*transfer and number
of solvent-exposed polar groups, respectively, can focus
synthetic efforts on compounds with balanced log D and
EPSA.
Despite the kinetic nature of permeability and complexity of
oral bioavailability, it is remarkable that the combination of 3D
physics-based predictors, such as ΔG*transfer and number of
solvent-exposed hydrogen-bond donors, can identify perme-
able and orally bioavailable macrocycles. The data support
combining ΔG*transfer calculations with conformational analysis
to select compounds with favorable ΔG*transfer values and
minimal number of solvent-exposed hydrogen-bond donors.
Further efforts are needed to enable quantitative comparison
across noncongeneric series. Recent advances in macrocyclic
sampling39,40 and structure-kinetic simulation studies41 of
passive membrane simulations should facilitate this and
provide insights into descriptors suitable for high-throughput
permeability predictions.
Adding to the body of existing knowledge, these data
illustrate further that privileged N-methylated peptide back-
bones, whether identified by Nature, by serendipity, or by
design, can seed families of oral peptides. Families of
permeable peptides share common peptide backbones and
N-methylation patterns, which reduce polar surface area, drive
macrocycle conformation, and set the intramolecular hydro-
gen-bond network. No doubt, permeability can be over-
whelmed by side chains with limited hydrophobicity, high
polarity, and/or charge. However, the available diversity of
lipophilic, noncanonical side chains alone is vast. Reports
describing permeable peptides bearing polar side chains35 add
further opportunity. In addition, as Nature shows, there is
utility at the intersection of passively permeable peptides and
challenging protein surfaces. While the rat oral exposures of
many of the examples contained herein do not meet the levels
commonly achieved by standard small molecules, a subset of
compounds, especially 1b, 8b, and 10a, offer impressive rat
exposure relative to most peptides. We do note that the
predictability of the rat oral data to higher species may be
variable. Nonetheless, the oral exposures achieved herein may
be suitable for a subset of therapeutic applications in humans,
including protein surfaces that are especially challenging for
traditional, rule-compliant small molecules. It should also be
noted that several of the examples presented herein, such as
8b, inhibit CYP3A4 (IC50 370 nM), a property shared by other
lipophilic macrocycles such as 1NMe3 (IC50 3640 nM) and
CsA (IC50 1500 nM), so potential for drug−drug interaction in
a therapeutic setting must also be considered. Finally, the
potential of large array library formats, including in vitro
EXPERIMENTAL SECTION
■
Modeling and Simulation Details. The sequences of all possible
N-methylated variants with 0−3, 4, or 4 N-methylated residues (not
counting Pro) were generated for each of 10, 12, or 14 scaffolds for 6-
mers, 7-mers, and 8-mers, respectively, using in-house Perl script. This
resulted in 292, 768, and 1514 sequences of virtual peptides,
respectively. The initial conformers for all-atom models of these
peptides were generated via LEaP program in AmberTools1242 based
on manually created poly Ala 6-mer, 7-mer, and 8-mer templates and
using the Amber99sb43 force field with the parm@Frosst44 small-
molecule extension and AM1-BCC45 charges. The resulting initial
conformers were minimized with Sander program in AMBER 1242 to
relax the side chains; restraints were applied to prevent chirality
inversions. The minimized structures were converted to Maestro
̈
format by Protein Preparation Wizard from Schrodinger software
package46 for further calculations with MacroModel.47
We adapted the published approach to passive permeability
predictions from Jacobson’s group31 and complemented it with
intramolecular hydrogen-bonding analysis. We focused on transfer
free energy from water to membrane (ΔGtransfer and ΔGt*ransfer) as a
dominant factor for permeability. This is because (a) the studied
compounds are charge neutral and exist as a single tautomer and,
therefore, the free energy penalty for our compounds to adopt their
neutral form for membrane permeation is zero; (b) our comparisons
were done within each of the congeneric series and, therefore, based
on their work,31 the permeant size effect on permeability was expected
to be small for that purpose. An approximate transfer free energy from
water to membrane was computed as follows. Conformations within 5
kcal/mol were generated in low dielectric media (LDM, the dielectric
constant of chloroform), mimicking the cell membrane inner
hydrophobic layer. Conformations in high dielectric media (HDM,
representing water) were generated for a few cases initially as well but
were found not necessary for our studies and not pursued further.
Generalized Born (GB)48−54 implicit solvation models for chloroform
and water were used for LDM and HDM, respectively. ΔGtransfer,i was
approximated by its energy difference between LDM and HDM for
ith conformation; ΔGtransfer was for the lowest energy conformer (see
also ensemble-based ΔGt*ransfer below). The conformation generation
was performed with macrocycle conformational sampling script in
MacroModel, which combines simulated annealing with large-scale
low-mode conformational search; we used optimal parameters
previously described47 (10,000 search cycles, 10 kcal/mol energy
window, recalculating the eigenvectors when a new global energy
minimum is found). The resulting conformations were clustered with
the cutoff of 0.75 Å.
The 6-mers selection was done as follows. The top 2−4 peptides
from each scaffold with the lowest ΔGtransfer were selected for
synthesis and the conformations of these peptides were examined to
make sure that all backbone NHs were paired with backbone
carbonyls in at least some of them. ΔGtransfer was calculated for all
generated conformations and the lowest ΔGtransfer was recorded from
an ensemble (ΔG*transfer) and the corresponding conformation was
analyzed. The conformations for selected peptides were sampled with
both OPLS200555 and MMFF94s56 force fields. In addition, mutants
that were trimmed to Ala at each residue were simulated as well to
improve overall sampling.
The 7-mers and 8-mers selections were done similarly to 6-mers,
but in addition, ΔGtransfer for each peptide was calculated for all
generated conformations and the lowest ΔGtransfer was recorded from
an ensemble (ΔGt*ransfer). For the 7-mers, the corresponding
conformations were examined to make sure that backbone NHs
were paired with backbone carbonyls as much as possible, and these
2627
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
compounds were selected for synthesis. For the 8-mers, we restricted
synthesis to only compounds that had all backbone NHs paired with
backbone carbonyls.
cleavage cocktail consisting of TFA/H2O/TIS 92.5:2.5:5 (3−5 mL).
The resin was filtered and treated again with cleavage cocktail (2−3
mL) for 30 min. The resin was filtered and the combined filtrates
were concentrated to yield the crude linear peptide.
NMR Experimental Details. Compound 1a was dissolved in 50
μL of CDCl3 and placed in a 1.7 mm NMR tube. The tube was
purged with nitrogen gas before and after the sample delivery and
sealed with Parafilm. All data were recorded on a Bruker 600 MHz
(for 1H) AVANCE III spectrometer equipped with a 1.7 mm TCI
cryoprobe. Data analysis was done using Mnova software. Resonance
assignments, ROESY cross peak assignments, and integrations were
carried out manually. The mixing time used for the ROESY spectrum
was 400 ms. ROESY peak integrals were qualitatively categorized as
weak, medium, or strong and are listed in Supporting Information
Table 1, with the corresponding atom numbers in Supporting
Information Figure 3. This table was used for NMR structural
refinements. Two conformers were observed in the NMR spectra in a
∼60:40 ratio at 300 K and only the peaks from the major conformer
were used in the modeling.
NMR Structural Refinements. An all-atom model of 1a was built
as previously described above. The respective topology and
coordinate files were generated via LEaP program in AmberTools12
using Amber99sb force field with the parm@Frosst small-molecule
extension and conformationally and topologically averaged AM1-BCC
charges.
N-(Chloroacetoxy)succinimide Preparation. To a stirred
suspension of 1-hydroxypyrrolidine-2,5-dione (30 g, 261 mmol) in
dichloromethane (300 mL) cooled to 0 °C was added solid sodium
bicarbonate (32.8 g, 391 mmol), followed by 2-chloroacetyl chloride
(22.81 mL, 287 mmol) slowly at 0 °C. The ice bath was removed and
the reaction mixture was then stirred at room temperature for 2 h.
Sodium sulfate (20 g) was added to the reaction mixture and the
resulting mixture was filtered. The filtrate was evaporated in vacuo.
The residue was triturated with ethyl acetate/heptane 1:1 (120 mL)
to provide N-(chloroacetoxy)succinimide (36.4 g, 190 mmol, 72.9%
yield), whose analytical data matched with the literature.58
Peptide Cyclization. The crude peptide was dissolved in
dimethylsulfoxide (5−8 mL) and treated with triethylamine until a
pH of 9 was obtained by pH paper (typically ca. 200−500 μL of
triethylamine). The reaction mixture was then shaken overnight at
room temperature. The following morning, the product mixture was
concentrated on a GeneVac to a few milliliters of solution and was
directly purified in 1−1.5 mL injection aliquots onto a reverse-phase
preparative HPLC Waters Autopure with UV-MS detection, a flow
rate of 75 mL/min, outfitted with either 30 × 50 Waters Sunfire C18
5 μm column (ACN and water with a 0.1% TFA modifier) or 30 × 50
Waters XBridge C18 5 μm column (ACN and water with 5 mM
ammonium hydroxide). Desired fractions were collected, triggering
either by absorption at 210 nm or by mass spectrometry, and
lyophilized to yield the final products.
Molecular dynamics simulations in implicit solvent (Generalized
Born model)57 were run with the Sander program in AMBER 12. The
dielectric constant of the solvent was 4.8, corresponding to
chloroform. The temperature was controlled using a Langevin
thermostat with a collision frequency of 5 ps−1.
The system was first subjected to a 5000 step minimization (300
steps by the steepest descent followed by the conjugate gradient
method) followed by 50 independent 700 ps simulated annealing runs
with restraints based on ROE data (Supporting Information Table 1).
Each simulated annealing run consisted of heating from 0 to 1200 K
for 200 ps, followed by equilibration at 1200 K for 200 ps and then
cooling to 600 K and then to 300 K for 100 ps, followed by further
cooling from 300 to 0 K for 200 ps; the restraints were gradually
turned on for the first 400 ps and stayed on till the end of the
simulation. The resulting ensemble of structures was clustered and
selected based on the smallest number of ROE violations.
Cyclic Peptide Synthesis. A general synthetic scheme for 6-mer
compounds is shown in Scheme 1, and 7-mer and 8-mer compounds
were prepared in an analogous manner. All chemicals and resins were
commercially available and used without further purification or
prepared as indicated. All amino acid building blocks used were
Fmoc-protected. Anhydrous solvents were obtained from Sigma-
Aldrich and used as received. Reactions were monitored using LCMS,
and products were purified using reverse-phase preparative HPLC,
followed by lyophilization.
Automated Peptide Synthesis. All peptides were synthesized
on the Liberty peptide synthesizer from CEM, Inc. Peptides were
synthesized on a 0.1 mmol or 0.25 mmol scale, starting from Methyl
Indole AM resin obtained from EMD Millipore. Either Fmoc-
Cys(Trt)-OH or Fmoc-Cys(Mtt)-OH was used to introduce the
initial cysteine residue. Fmoc deprotection was achieved in two cycles
of treatment with 4-methylpiperidine, the first for 30 s, and the second
for 3 min. All amino acids were activated with HATU and Hunig’s
base. At a 0.1 mmol scale, 5 equiv of the amino acid were used; at a
0.25 mmol scale, 4 equiv were used. All amino acids were coupled for
10 min at 75 °C, except for cysteine, which was coupled for 10 min at
50 °C. After the peptide synthesis was complete, the resins were
filtered and washed twice each with dimethylformamide and
dichloromethane and dried under house vacuum.
Peptide Acylation and Deprotection. The resin was suspended
in N-methylpyrrolidine (3−8 mL), and 5 equivalents of N-
(chloroacetoxy)succinimide (prepared as described below) were
added. The resulting resin mixture was shaken at room temperature
overnight. The following morning, the resin was filtered, washed three
times each with dimethylformamide and dichloromethane, and then
dried under house vacuum. The resin was then treated for 1 h with a
Compound Characterization Data. All peptides were charac-
terized as ≥95% purity by either UPLC or HPLC (traces provided in
Supporting Information Figure 4). Table 8 indicates for each product
Table 8. Characterization Data for Compounds 1−10c
compound
no
retention
time
analytical
method
HRMS
calculated
HRMS
found
c
1
3.12
2.94
3.03
1.34
2.99
4.48
3.18
2.82
2.94
2.48
2.66
2.35
2.43
2.38
2.73
2.92
2.85
2.50
2.74
2.66
3.11
3.13
1.47
2.81
2.81
1.61
1.51
A
G
C
F
NA
774.4
1a
1b
2
2a
2b
2c
2d
2e
3
3a
4
4a
5
5a
6
6a
7
7a
8
816.5057
802.4900
774.4615
800.4744
802.4900
816.5057
802.4901
788.4744
758.4274
800.4744
758.4276
772.4431
NA
816.5020
802.4854
774.4623
800.4742
802.4908
816.5040
802.4904
788.4805
758.4288
800.4788
758.4275
772.4491
C
B
C
a
C
C
D
G
D
C
C
G
A
c
758.9
800.4744
NA
800.4742
c
759.0
G
C
C
D
800.4744
758.4275
800.4744
871.5116
899.5428
NA
947.5428
1055.6817
1041.6660
1076.6582
1064.6218
800.4815
758.4315
800.4748
871.5119
899.5476
b
8a
8b
8c
9a
10a
10b
10c
C
c
C
F
E
E
F
F
913.9
947.5507
1055.6816
1041.6660
1075.6504
1064.6285
a
b
Poor UV chromophore, purity assessed by ELSD. Column
c
contained a UV-active contaminant, purity assessed by CAD. LRMS.
2628
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
the analytical method used to determine retention time and purity,
which was assessed by UV absorption at 214 nm unless otherwise
indicated. HRMS data were obtained on a Waters AcQuity UPLC
using either a Waters LCT Premier or Waters Xevo G2 Qtof detector.
For a few peptides where indicated, HRMS data were not obtained,
and so LRMS data are provided. The analytical method descriptions
are listed below.
Method AInstrument: Agilent 1100/1200 HPLC; column:
XBridge C18 3.5 μm 3.0 × 30 mm; 5.10 min run time, 5 → 80%
solvent B: solvent A from 0 to 4.30 min, 80 → 95% solvent B: solvent
A for 0.4 min, 95% solvent B for 0.3 min, 95 → 5% solvent B: solvent
A for 0.1 min. Solvents: solvent A = 5 mM ammonium hydroxide in
water, solvent B = acetonitrile. UV detection array 210−400; mass
detection 150−1600; column at 40 °C; flow rate 2.0 mL/min; pH
10.2.
Method BInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC BEH C18 1.7 μm, 2.1 × 50 mm; 8.05 min run time, 2 → 98%
solvent B: solvent A from 0 to 7.5 min, 98% solvent B for 0.4 min, 98
→ 2% solvent B: solvent A for 0.15 min. Solvents: solvent A = 5 mM
ammonium hydroxide in water, solvent B = 5 mM ammonium
hydroxide in acetonitrile. UV detection array 210−400; mass
detection 120−1250; column at 50 °C; flow rate 1.0 mL/min; pH
10.2.
Method CInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC BEH C18 1.7 μm, 2.1 × 50 mm; 5.19 min run time, 2 → 98%
solvent B: solvent A from 0 to 4.40 min, 98% solvent B for 0.75 min,
98 → 2% solvent B: solvent A for 0.04 min. Solvents: solvent A =
0.1% formic acid in water (v/v), solvent B = 0.1% formic acid in
acetonitrile (v/v). UV detection array 210−400; mass detection 120−
1250; column at 50 °C; flow rate 1.0 mL/min; pH 2.6.
triplicate in 96-well polypropylene 2 mL plates. The DMSO was
removed under temperature and vacuum.
Dried samples were incubated with 300 μL of octanol and 300 μL
of buffer and shaken for at least 4 h. Plates were centrifuged for 15
min at 4000 rpm to separate the octanol and buffer layers. Sample
preparation and phase separation were automated using liquid
handling workstations. Both octanol and buffer phases were quantified
using liquid chromatography with tandem mass spectrometry. Log D
was derived from the ratio of compound peak area responses in each
phase, adjusted to the internal standard peak areas.
EPSA Method. Analysis for the determination of the exper-
imentally derived polar surface analysis value (EPSA) was performed
using a modified set of conditions from those previously
described.33,59 The sample was dissolved to an approximate
concentration of 5 mM in DMSO. Analysis was performed on the
Waters Acquity UPC2 supercritical fluid chromatography system with
a photodiode array and a Waters QDa mass spectrometer with a
single quadrupole (Milford, MA). The mobile phase consisted of two
parts. Mobile phase A consisted of carbon dioxide pressurized to 850
psi using a Waters Bulk Delivery System (BDS). Mobile phase B was
ammonium formate diluted to 20 mM in HPLC-grade methanol.
Analysis was performed using a 4.6 mm × 250 mm Chirex 3014
column (Phenomenex, Torrance, CA) with 5 μm particle size and 100
Å pore size. The flow rate was 3 mL/min with an automated back
pressure regulator set point of 2100 psi. Each injection was performed
using an injection volume of 3 μL. The column temperature set point
during the analysis was 40 °C. The composition of the mobile phase
was varied linearly from 5 to 65% mobile phase B over 4.6%/min,
holding at 65% for 2 min, and then reverting to the original 5% until
the end of the run with a total data acquisition time of 19 min.
MDCK-LE Method.60 MDCK-LE (low efflux) cells were cultured
at 37 °C under a 5% CO2 atmosphere, at 95% relative humidity in
DMEM containing 10% FBS, penicillin−streptomycin (100 μg/mL),
and 2 mM Ala−Gln. Cells were passaged every 3−4 days. For assay
purposes, cells were seeded at a density of approximately 265,000
cells/cm2 of a 96-well Transwell plate (Corning Life Sciences, Acton,
MA) and cultured in the same media noted above for a period of 4
days.
Method DInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC BEH C18 1.7 μm, 2.1 × 50 mm; 5.19 min run time, 2 → 98%
solvent B: solvent A from 0 to 4.40 min, 98% solvent B for 0.75 min,
98 → 2% solvent B: solvent A for 0.04 min. Solvents: solvent A = 5
mM ammonium hydroxide in water, solvent B = 5 mM ammonium
hydroxide in acetonitrile. UV detection array 210−400; mass
detection 120−1250; column at 50 °C; flow rate 1.0 mL/min; pH
10.2.
Assay. The determination of the apparent permeability (Papp) was
performed in the A → B (apical to basal) direction where each
compound was assayed in triplicate. The zwitterion bestatin, a poorly
permeably compound, was used as a marker of monolayer integrity.
To initiate the assay, media was aspirated, and the cells and basal
chambers were washed three times with Hank’s Balanced Salt
Solution (HBSS) containing 10 mM HEPES (pH 7.4). Compound
test solutions were prepared in triplicate in HBSS containing 10 mM
HEPES (pH 7.4) and 0.02% bovine serum albumin (BSA) to a final
concentration of 10 μM and centrifuged for 2 min at 4000g and then
applied to the donor compartment at time zero. Additionally, at time
zero, a 37 °C solution without test articles [HBSS + 10 mM HEPES
(pH 7.4) plus 0.02% BSA] was added to the receiver chamber of the
Transwell plate. A time zero sample of the donor solution was also
sampled for further analysis. The assay was conducted for a period of
120 min at 37 °C without shaking. At the time of assay termination,
samples were taken from each donor compartment and each acceptor
compartment of the Transwell plate. To each of the 0 and 120 min
samples was added an internal standard solution containing glyburide
in water: acetonitrile, 50:50 (v/v). Concentration curves were
prepared using a Labcyte Echo in the same matrix noted above.
Samples and concentration curve samples were centrifuged for 10 min
at 4000g and subsequently analyzed by mass spectroscopy.
Method EInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC BEH C18 1.7 μm, 2.1 × 50 mm; 8.05 min run time, 40→98%
solvent B: solvent A from 0 to 5.00 min, 98% solvent B for 2.09 min,
98 → 40% solvent B: solvent A for 0.06 min. Solvents: solvent A =
0.1% formic acid in water (v/v), solvent B = 0.1% formic acid in
acetonitrile (v/v). UV detection array 210−400; mass detection 125−
1200; column at 50 °C; flow rate 1.0 mL/min; pH 2.6.
Method FInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC CSH 1.7 μm, 2.1 × 50 mm; 2.2 min run time, 2% solvent B,
from 0 to 0.6 min, 2 → 98% solvent B/solvent A for 1.7 min, 98%
solvent B for 0.24 min, 98 → 2% solvent B/solvent A for 0.16 min.
Solvents: solvent A = 0.05% trifluoroacetic acid in water (v/v), solvent
B = 0.05% trifluoroacetic acid in acetonitrile (v/v). UV detection
array 210−400; mass detection 100−2050; column at 50 °C; flow rate
1.0 mL/min; pH 2.6.
Method GInstrument: Waters AcQuity UPLC; column: AcQuity
UPLC BEH C18 1.7 μm, 2.1 × 50 mm; 5.19 min run time, 2 → 98%
solvent B: solvent A from 0 to 4.40 min, 98% solvent B for 0.75 min,
98 → 2% solvent B: solvent A for 0.04 min. Solvents: solvent A =
0.1% formic acid in water (v/v), solvent B = 0.1% formic acid in
acetonitrile (v/v). UV detection array 210−400; mass detection 200−
2000; column at 50 °C; flow rate 1.0 mL/min; pH 2.6.
Profiling and Pharmacokinetic Experiments. In vitro ADME,
physicochemical, and pharmacokinetic profiling was conducted by
published methods or modifications to published methods. All animal
experiments performed in the manuscript were conducted in
compliance with institutional guidelines.
Mass Spectrometry. Assay samples were loaded onto a RapidFire
C4 cartridge by means of a RapidFire autosampler (Agilent, Santa
Clara, CA). Chromatography was performed at a flow rate of 1.25
mL/min, loading with 0.1% formic acid in water, and eluting in 0.1%
formic acid in methanol. Mass spectroscopy was performed using an
AB Sciex API5500 (Sciex, Framingham, MA) equipped with a turbo
ion spray source. The analyte concentration was calculated from the
chromatographic peak area ratio of the analyte to internal standard
Log D Method. Phosphate-buffered saline (pH 7.4) saturated
with 1-octanol and 1-octanol saturated with buffer was prepared prior
to the start of the log D assay. Aliquots of 10 mM DMSO compound
solution along with an internal reference compound were dispensed in
2629
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
(glibenclamide, m/z 494 → 169), using Multiquant software V3.0
(Sciex, Framingham, MA).
Calculations. Papp values were determined as
Sprague−Dawley rats (age approximately 9−11 weeks, weight
approximately 300−325 g, n = 2 or 3). The compounds were
formulated in solution in PEG300: 20% Solutol/PBS (10:25:65) and
dosed intravenously (IV, via injection into the jugular vein cannula) at
a dose of 0.5 mg/kg (compounds 1, 1a, 1b, 7, 7a, 8a, 8b, 9a, and 10a)
or 1 mg/kg (compounds 8, 8c, 10b, and 10c) with a dosing volume of
1 mL/kg. Compounds were formulated in solution at 0.5 mg/mL in
the same formulation for PO dosing (PO, via oral gavage; at 5 mg/
kg). Approximately, 100 μL of whole blood was collected from each
animal via a jugular vein cannula at 5 min (IV dose only), 15 min, 0.5,
1, 2, 4, 7, and 24 h post-dose and transferred to EDTA tubes. Blood
samples were centrifuged at 3000 rpm and the resultant plasma was
transferred to a capped PCR 96-well plate and frozen at −20 °C until
subsequent preparation and analysis by HPLC-MS/MS.
Papp = VA/(S[D0]) × A120/t
Percent recovery values were determined as
% recovery = 100 × ((A120 + D120)/D0)
where VA is the volume of the acceptor (mL), S is the surface area of
the membrane, D0 is the donor solution concentration at t = 0, D120 is
the donor solution concentration at t = 120, A120 is the acceptor
solution concentration at t = 120, and t = time (seconds).
High-Throughput Equilibrium Solubility. The high-through-
put (HT) equilibrium solubility assay was performed as previously
described.61 Briefly, aliquots of 10 mM DMSO stock solution were
plated and DMSO was removed under temperature and vacuum.
Media (pH 4.0 acetate buffer, pH 6.8 phosphate buffer, or FaSSIF-v2:
Fasted State Simulated Intestinal Fluid) was added to the 96-well
plate for a target concentration of 1 mM. The plate was sealed,
incubated on a shaker at 1350 rpm and ambient temperature for 16−
24 h, and then centrifuged for 20 min at 3750 rpm to pellet the
precipitate. The supernatant was transferred to another plate and
centrifuged a second time. The supernatant was diluted 200-fold with
50:50 acetonitrile/water and a 4-point calibration curve was
constructed using 50:50 acetonitrile/water. Analysis of the super-
natant concentration was performed based on the calibration curve,
using an Agilent RapidFire-MS/MS mass spectrometer system.
Human Microsomal Incubations. The experiments were
performed in 96-well format with shaking incubation at 37 °C on
an automated platform. Test articles, at a concentration of 10 mM in
DMSO, were diluted 1:1000 into a 100 mM potassium phosphate
solution to a concentration of 10 μM. This solution was added to
human liver microsomal protein (1.25 mg/mL) suspended in
phosphate buffer (pH 7.4). The assay reactions were initiated by
the addition of a cofactor solution (2 mM NADPH, 4 mM MgCl2 in
100 mM potassium phosphate). At specific reaction time points (0, 5,
15, and 30 min), reaction aliquots were removed and reactions were
terminated by the addition of three volumes of acetonitrile containing
the analytical internal standard (glibenclamide). The samples were
then centrifuged at 4000g at 4 °C for 10 min, and the supernatants
were analyzed by LC/MS/MS for quantitation of the remaining test
article. The percentage of test article remaining, relative to time zero
minute incubation, was used to estimate the in vitro elimination-rate
constant (kmic), which was subsequently used to calculate the in vitro
metabolic clearance rates.
LC/MS/MS Analysis. Analysis of samples was performed on a high-
performance liquid chromatography (HPLC)-tandem mass spectrom-
etry (LC/MS/MS) system consisting of a Shimadzu 30 series
autosampler and HPLC pump coupled to an AB Sciex API6500.
Compound-specific parameters (precursor ion, product ion, decluster-
ing potential, and collision energy for single reaction monitoring)
were obtained by automatic tuning using the Multiquant software
V3.0. Samples were injected onto a 3 μm ACE 3 C18, 2.1 mm × 30
mm column, by means of the Shimadzu 30 series autosampler. The
components were eluted with a gradient of 0.1% formic acid (mobile
phase A) and 0.1% formic acid in acetonitrile (mobile phase B) at a
flow of 700 μL/min using the following gradient: 0 min 2% B; 0.25
min 2% B; 1.00 min 98% B; 1.55 min 98% B; 1.95 min 2% B; 2.00
min 2% B. The analyte concentration was calculated from the
chromatographic peak area ratio of the analyte to an internal standard
(glibenclamide, m/z 494 → 169), using Multiquant software V3.0
(Sciex, Framingham, MA).
Pharmacokinetics Study in Dogs. The dog PK study was
conducted at Agilux (One Innovation Drive, Worchester, MA 01605).
The pharmacokinetics of compound 8b was determined in male
beagle dogs (age ≥2 yr, wt approximately 10 kg, n = 3). Compound
8b was formulated in 5% Solutol (neat): 95% PBS at 0.5 mg/mL for
IV dosing. This formulation was administered by slow bolus iv
injection into the cephalic vein at 0.2 mL/kg (0.1 mg/kg dose; n = 3).
Compound 8b was formulated in 5% Solutol (neat): 95% PBS at 0.25
mg/mL and dosed orally (PO; 2 mL/kg; 0.5 mg/kg; n = 3 animals/
compound) by gavage. Blood (approximately 0.3 mL/sample) was
collected from the saphenous vein of each animal at 5 min (IV dose
only), 15 min, 0.5, 1, 2, 4, 7, and 24 h post-dose. Blood samples were
centrifuged at 3000 rpm and the resultant plasma was transferred to a
matrix plate and stored frozen until analysis.
PK Sample Preparation and Analysis. Standard curve solutions
of each compound were prepared from 10 mg/mL DMSO stock
solution diluted into plasma to final concentrations of 1−5000 ng/
mL. These standard curve samples were prepared for LC/MS/MS like
the PK plasma samples below.
PK plasma samples were thawed and 25 μL aliquots of each sample
(or standard curve solution) were transferred to a fresh plate. A 150
μL aliquot of extraction solution (100% acetonitrile containing 100
ng/mL of glyburide as an internal standard) was added to each well.
The plate was covered and mixed for approximately 5 min on a pulse-
vortex mixer. The plate was centrifuged at 4000 rpm for 10 min at 4
°C. A 125 μL volume of the resulting supernatant was transferred into
the corresponding well of a clean 1 mL 96-well assay plate and mixed
with 100 μL of water. A 10 μL aliquot of the extract was injected onto
an LC/MS/MS system for analysis. HPLC conditions: Agilent 1290,
LC column: ACE C18, 30 × 2.0 mm, 3 μm; Acquity C18 50 × 2.1
mm, 3 μm, or equivalent. Flow rate 600−800 μL/min. MS/MS
system (i.e. ABSciexQ5500, Agilux6500, or similar) and analyzed in
the positive mode (each compound was tuned separately from neat
methanol stock solution).
PK parameters were derived from plasma concentration values by
noncompartmental analysis using Excel. Terminal half-life (t1/2) =
−0.693/kel, where kel is the slope of the line formed from the times of
the last 3 measured concentrations versus the natural log of the last 3
measured concentrations. Initial concentration (C0) was set to 0 for
PO dosing and to e(ln(C1)+(ln(C1)−ln(C2))/2) for IV dosing. The area under
the curve (AUC) was calculated by the linear trapezoidal rule: AUC =
n−1
∑
(ti+1 − ti) × (Ci + Ci+1)/2. Extrapolated AUC (AUCext) =
i=0
AUC + Clast × t1/2/0.693, where Clast is the last quantifiable
concentration. The area under the moment curve (AUMC) was
calculated by: AUMC = i=0∑
n−1 (ti+1 − ti) × (Ci × ti + Ci+1 × ti+1)/2.
Mean residence time (MRT) was calculated by: MRT = AUMC/
AUC. Clearance (Cl) was calculated by: CL = (IV Dose)/AUCext.
Volume of distribution (Vdss) was calculated by: Vdss = CL × MRT.
Bioavailability (% F) was calculated by: % F = 100% × (IV dose × PO
AUC)/(PO dose × IV AUC).
Ethics Statement. All in vivo research was reviewed and approved
by the Novartis Institutes of Biomedical Research Institutional Animal
Care and Use Committee in accordance with applicable local, state,
and federal regulations.
Pharmacokinetic Studies in Rat. PK studies in rat were
conducted either at Charles River laboratories (Worchester, MA
01605) or internally at Novartis. Studies were conducted in male
ASSOCIATED CONTENT
■
sı
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jmedchem.0c01505.
2630
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
molecules. We also thank Robert Pearlstein, Callum Dickson,
and Viktor Hornak for fruitful discussions on the kinetic nature
of permeability.
Supplemental Table 1, Supplemental Figures S1−S4,
and LCMS traces for compounds (PDF)
Molecular formula strings (CSV)
ABBREVIATIONS
■
AUTHOR INFORMATION
Corresponding Author
Lauren G. Monovich − Novartis Institutes for BioMedical
Research, Cambridge, Massachusetts 02139, United States;
orcid.org/0000-0001-8455-4171;
■
AUC, area under the curve; bql, below the limit of
quantitation; clog P, calculated octanol−water partition
coefficient; CsA, Cyclosporine A; CYP3A4, cytochrome P450
3A4; DMSO, dimethylsulfoxide; EPSA, experimental polar
surface area; HPLC, high-performance liquid chromatography;
IV, intravenous; LCMS, liquid chromatography mass spec-
trometry; log Papp, log of apparent permeability in cm·s−1;
MDCK-LE, Madin−Darby canine kidney−low efflux cells;
MW, molecular weight; nd, not determined; Nle, norleucine;
Nle, N-methylnorleucine; NMR, nuclear magnetic resonance;
PO, per os; PAMPA, parallel artificial membrane permeability
assay; Phe, N-methylphenylalanine; phe, D-phenylalanine; phe,
N-methyl-D-phenylalanine; TEA, triethylamine; TFA, trifluoro-
acetic acid; TIS, triisopropylsilane; TPSA, topological polar
surface area
Email: lauren.monovich@novartis.com
Authors
Andrei A. Golosov − Novartis Institutes for BioMedical
Research, Cambridge, Massachusetts 02139, United States
Alec N. Flyer − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Jakal Amin − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Charles Babu − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Christian Gampe − Novartis Institutes for BioMedical
Research, Cambridge, Massachusetts 02139, United States
Jingzhou Li − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Eugene Liu − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Katsumasa Nakajima − Novartis Institutes for BioMedical
Research, Cambridge, Massachusetts 02139, United States;
orcid.org/0000-0001-8295-5086
REFERENCES
■
(1) Vinogradov, A. A.; Yin, Y.; Suga, H. Macrocyclic Peptides as
Drug Candidates: Recent Progress and Remaining Challenges. J. Am.
Chem. Soc. 2019, 141, 4167−4181.
(2) Zorzi, A.; Deyle, K.; Heinis, C. Cyclic Peptide Therapeutics:
Past, Present and Future. Curr. Opin. Chem. Biol. 2017, 38, 24−29.
(3) Price, D. A.; Mathiowetz, A. M.; Liras, S. Designing Orally
Bioavailable Peptide and Peptoid Macrocycles. In Practical Medicinal
Chemistry with Macrocycles; Marsault, E., Peterson, M., Eds.; John
Wiley & Sons, Inc.: Hoboken, NJ, 2017; pp 59−76.
(4) Rezai, T.; Bock, J. E.; Zhou, M. V.; Kalyanaraman, C.; Lokey, R.
S.; Jacobson, M. P. Conformational Flexibility, Internal Hydrogen
Bonding, and Passive Membrane Permeability: Successful in Silico
Prediction of the Relative Permeabilities of Cyclic Peptides. J. Am.
Chem. Soc. 2006, 128, 14073−14080.
David Nettleton − Novartis Institutes for BioMedical
Research, Cambridge, Massachusetts 02139, United States
Tajesh J. Patel − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Patrick C. Reid − PeptiDream, Inc., Kawasaki, Kanagawa
210-0821, Japan
(5) Over, B.; Matsson, P.; Tyrchan, C.; Artursson, P.; Doak, B. C.;
Foley, M. A.; Hilgendorf, C.; Johnston, S. E.; Lee, M. D.; Lewis, R. J.;
McCarren, P.; Muncipinto, G.; Norinder, U.; Perry, M. W. D.; Duvall,
J. R.; Kihlberg, J. Structural and Conformational Determinants of
Macrocycle Cell Permeability. Nat. Chem. Biol. 2016, 12, 1065−1074.
(6) Blanco, M.-J. Building upon Nature’s Framework: Overview of
Key Strategies Toward Increasing Drug-Like Properties of Natural
Product Cyclopeptides and Macrocycles. In Methods in Molecular
Biology; Goetz, G., Ed.; Humana Press: New York, NY, 2019; Vol.
2001, pp 203−233.
Lihua Yang − Novartis Institutes for BioMedical Research,
Cambridge, Massachusetts 02139, United States
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.jmedchem.0c01505
Author Contributions
A.A.G., A.N.F., and L.G.M. conceptualized the study, wrote the
manuscript, and contributed equally; A.A.G. performed the
modeling and simulation experiments; A.N.F., C.G., J.L., E.L.,
K.N., T.J.P., and L.Y. prepared the compounds; J.A. and D.N.
performed and analyzed the pharmacokinetic studies; C.B.
performed the NMR studies to determine the 1a solution
structure; P.C.R. advised on amino acid selection and technical
capabilities of in vitro translation technology. All authors have
given approval to the final version of the manuscript.
(7) Foster, A. D.; Ingram, J. D.; Leitch, E. K.; Lennard, K. R.; Osher,
E. L.; Tavassoli, A. Methods for the Creation of Cyclic Peptide
Libraries for Use in Lead Discovery. J. Biomol. Screen 2015, 20, 563−
576.
(8) Sohrabi, C.; Foster, A.; Tavassoli, A. Methods for Generating
and Screening Libraries of Genetically Encoded Cyclic Peptides in
Drug Discovery. Nat. Rev. Chem. 2020, 4, 90−101.
̌
̌
(9) Bozovicar, K.; Bratkovic, T. Evolving a Peptide: Library
Platforms and Diversification Strategies. Int. J. Mol. Sci. 2020, 21,
215−246.
Author Contributions
#A.A.G., A.N.F., and L.G.M. contributed equally to the
manuscript.
(10) Naylor, M. R.; Bockus, A. T.; Blanco, M.-J.; Lokey, R. S. Cyclic
Peptide Natural Products Chart the Frontier of Oral Bioavailability in
the Pursuit of Undruggable Targets. Curr. Opin. Chem. Biol. 2017, 38,
141−147.
Notes
The authors declare no competing financial interest.
(11) Nielsen, D. S.; Shepherd, N. E.; Xu, W.; Lucke, A. J.; Stoermer,
M. J.; Fairlie, D. P. Orally Absorbed Cyclic Peptides. Chem. Rev. 2017,
117, 8094−8128.
(12) Kling, A.; Lukat, P.; Almeida, D. V.; Bauer, A.; Fontaine, E.;
Sordello, S.; Zaburannyi, N.; Herrmann, J.; Wenzel, S. C.; König, C.;
Ammerman, N. C.; Barrio, M. B.; Borchers, K.; Bordon-Pallier, F.;
Brönstrup, M.; Courtemanche, G.; Gerlitz, M.; Geslin, M.; Hammann,
ACKNOWLEDGMENTS
■
We acknowledge John Reilly for the column-based profiling
data (log D and EPSA); Jennifer Poirier for peptide
purification methods; David Ayres, Adey Kefelegn, and
Xuena Lin for PAMPA and MDCK-LE data; and Daniel
McKay for enabling AMBER parametrization of custom
2631
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
P.; Heinz, D. W.; Hoffmann, H.; Klieber, S.; Kohlmann, M.; Kurz, M.;
Lair, C.; Matter, H.; Nuermberger, E.; Tyagi, S.; Fraisse, L.; Grosset, J.
̈
H.; Lagrange, S.; Muller, R. Targeting DnaN for Tuberculosis
Therapy Using Novel Griselimycins. Science 2015, 348, 1106−1112.
(13) Nielsen, D. S.; Hoang, H. N.; Lohman, R.-J.; Diness, F.; Fairlie,
D. P. Total Synthesis, Structure, and Oral Absorption of a Thiazole
Cyclic Peptide, Sanguinamide A. Org. Lett. 2012, 14, 5720−5723.
(14) Borel, J. F.; Kis, Z. L.; Beveridge, T. The History of the
Discovery and Development of Cyclosporine (Sandimmune). In The
Search for Anti-inflammatory Drugs; Merluzzi, V. J., Adams, J., Eds.;
(29) White, T. R.; Renzelman, C. M.; Rand, A. C.; Rezai, T.;
McEwen, C. M.; Gelev, V. M.; Turner, R. A.; Linington, R. G.; Leung,
S. S. F.; Kalgutkar, A. S.; Bauman, J. N.; Zhang, Y.; Liras, S.; Price, D.
A.; Mathiowetz, A. M.; Jacobson, M. P.; Lokey, R. S. On-Resin N-
Methylation of Cyclic Peptides for Discovery of Orally Bioavailable
Scaffolds. Nat. Chem. Biol. 2011, 7, 810−817.
(30) Shinbara, K.; Liu, W.; van Neer, R. H. P.; Katoh, T.; Suga, H.
Methodologies for Backbone Macrocyclic Peptide Synthesis Compat-
ible With Screening Technologies. Front. Chem. 2020, 8, 447.
(31) Leung, S. S. F.; Mijalkovic, J.; Borrelli, K.; Jacobson, M. P.
Testing Physical Models of Passive Membrane Permeation. J. Chem.
Inf. Model. 2012, 52, 1621−1636.
(32) Kerns, E. H.; Di, L.; Petusky, S.; Kleintop, T.; Huryn, D.;
McConnell, O.; Carter, G. Pharmaceutical Profiling Method for
Lipophilicity and Integrity Using Liquid Chromatography-Mass
Spectrometry. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci.
2003, 791, 381−388.
(33) Goetz, G. H.; Philippe, L.; Shapiro, M. J. EPSA: A Novel
Supercritical Fluid Chromatography Technique Enabling the Design
of Permeable Cyclic Peptides. ACS Med. Chem. Lett. 2014, 5, 1167−
1172.
(34) Peters, S. A. Physiologically-Based Pharmacokinetic (PBPK)
Modeling and Simulations: Principles, Methods, and Applications in the
Pharmaceutical Industry, 1st ed.; John Wiley & Sons, Inc.: Hoboken,
New Jersey, 2012.
(35) Rand, A. C.; Leung, S. S. F.; Eng, H.; Rotter, C. J.; Sharma, R.;
Kalgutkar, A. S.; Zhang, Y.; Varma, M. V.; Farley, K. A.; Khunte, B.;
Limberakis, C.; Price, D. A.; Liras, S.; Mathiowetz, A. M.; Jacobson,
M. P.; Lokey, R. S. Optimizing PK Properties of Cyclic Peptides: The
Effect of Side Chain Substitutions on Permeability and Clearance.
MedChemComm 2012, 3, 1282−1289.
̈
Birkhauser Boston: Cambridge, MA, 1995; pp 27−63.
(15) Ermert, P.; Luther, A.; Zbinden, P.; Obrecht, D. Frontier
Between Cyclic Peptides and Macrocycles. In Methods in Molecular
Biology; Goetz, G., Ed.; Humana Press: New York, NY, 2019; Vol.
2001, pp 147−202.
(16) Brayden, D. J.; Hill, T. A.; Fairlie, D. P.; Maher, S.; Mrsny, R. J.
Systemic Delivery of Peptides by the Oral Route: Formulation and
Medicinal Chemistry Approaches. Adv. Drug Deliv. Rev. 2020, 157, 2−
36.
(17) Boesch, D.; Gavériaux, C.; Jachez, B.; Pourtier-Manzanedo, A.;
Bollinger, P.; Loor, F. In Vivo Circumvention of P-Glycoprotein-
Mediated Multidrug Resistance of Tumor Cells with SDZ PSC 833.
Cancer Res. 1991, 51, 4226−4233.
(18) Paeshuyse, J.; Kaul, A.; De Clercq, E.; Rosenwirth, B.; Dumont,
J.-M.; Scalfaro, P.; Bartenschlager, R.; Neyts, J. The non-
immunosuppressive cyclosporin DEBIO-025 is a potent inhibitor of
hepatitis C virus replicationin vitro. Hepatology 2006, 43, 761−770.
(19) Riss, B.; Grandeury, A.; Gut, T.; Seeger-Weibel, M.; Zuercher,
C.; Li, J.; Gallou, F. Development of a Practical Process for the
Opening of Macrocyclic Cyclosporin A and Amino Acid Deletion.
Org. Process Res. Dev. 2014, 18, 1763−1770.
(20) Tuvia, S.; Atsmon, J.; Teichman, S. L.; Katz, S.; Salama, P.;
Pelled, D.; Landau, I.; Karmeli, I.; Bidlingmaier, M.; Strasburger, C. J.;
Kleinberg, D. L.; Melmed, S.; Mamluk, R. Oral Octreotide Absorption
in Human Subjects: Comparable Pharmacokinetics to Parenteral
Octreotide and Effective Growth Hormone Suppression. J. Clin.
Endocrinol. Metab. 2012, 97, 2362−2369.
(21) Hewitt, W. M.; Leung, S. S. F.; Pye, C. R.; Ponkey, A. R.;
Bednarek, M.; Jacobson, M. P.; Lokey, R. S. Cell-Permeable Cyclic
Peptides from Synthetic Libraries Inspired by Natural Products. J. Am.
Chem. Soc. 2015, 137, 715−721.
(22) Rezai, T.; Yu, B.; Millhauser, G. L.; Jacobson, M. P.; Lokey, R.
S. Testing the Conformational Hypothesis of Passive Membrane
Permeability Using Synthetic Cyclic Peptide Diastereomers. J. Am.
Chem. Soc. 2006, 128, 2510−2511.
(23) Wang, C. K.; Northfield, S. E.; Colless, B.; Chaousis, S.;
Hamernig, I.; Lohman, R.-J.; Nielsen, D. S.; Schroeder, C. I.; Liras, S.;
Price, D. A.; Fairlie, D. P.; Craik, D. J. Rational Design and Synthesis
of an Orally Bioavailable Peptide Guided by NMR Amide
Temperature Coefficients. Proc. Natl. Acad. Sci. U.S.A. 2014, 111,
17504−17509.
(24) Fouché, M.; Schäfer, M.; Berghausen, J.; Desrayaud, S.; Blatter,
M.; Piéchon, P.; Dix, I.; Martin Garcia, A.; Roth, H.-J. Design and
Development of a Cyclic Decapeptide Scaffold with Suitable
Properties for Bioavailability and Oral Exposure. ChemMedChem
2016, 11, 1048−1059.
(25) Passioura, T.; Suga, H. Flexizyme-Mediated Genetic
Reprogramming as a Tool for Noncanonical Peptide Synthesis and
Drug Discovery. Chem.−Eur. J. 2013, 19, 6530−6536.
(26) Passioura, T.; Suga, H. A RaPID Way to Discover Nonstandard
Macrocyclic Peptide Modulators of Drug Targets. Chem. Commun.
2017, 53, 1931−1940.
(27) Otero-Ramirez, M.; Passioura, T.; Suga, H. Structural Features
and Binding Modes of Thioether-Cyclized Peptide Ligands.
Biomedicines 2018, 6, 116−128.
(36) Furukawa, A.; Townsend, C. E.; Schwochert, J.; Pye, C. R.;
Bednarek, M. A.; Lokey, R. S. Passive Membrane Permeability in
Cyclic Peptomer Scaffolds Is Robust to Extensive Variation in Side
Chain Functionality and Backbone Geometry. J. Med. Chem. 2016, 59,
9503−9512.
(37) Fouché, M.; Schäfer, M.; Blatter, M.; Berghausen, J.;
Desrayaud, S.; Roth, H.-J. Pharmacokinetic Studies around the
Mono- and Difunctionalization of a Bioavailable Cyclic Decapeptide
Scaffold. ChemMedChem 2016, 11, 1060−1068.
(38) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and Computational Approaches to Estimate Solubility
and Permeability in Drug Discovery and Development Settings. Adv.
Drug Deliv. Rev. 2012, 64, 4−17.
(39) Poongavanam, V.; Danelius, E.; Peintner, S.; Alcaraz, L.; Caron,
G.; Cummings, M. D.; Wlodek, S.; Erdelyi, M.; Hawkins, P. C. D.;
Ermondi, G.; Kihlberg, J. Conformational Sampling of Macrocyclic
Drugs in Different Environments: Can We Find the Relevant
Conformations? ACS Omega 2018, 3, 11742−11757.
(40) Sindhikara, D.; Spronk, S. A.; Day, T.; Borrelli, K.; Cheney, D.
L.; Posy, S. L. Improving Accuracy, Diversity, and Speed with Prime
Macrocycle Conformational Sampling. J. Chem. Inf. Model. 2017, 57,
1881−1894.
(41) Dickson, C. J.; Hornak, V.; Pearlstein, R. A.; Duca, J. S.
Structure-Kinetic Relationships of Passive Membrane Permeation
from Multiscale Modeling. J. Am. Chem. Soc. 2017, 139, 442−452.
(42) Case, D. A.; Darden, T. A.; Cheatham, T. E., III; Simmerling,
C. L.; Wang, J.; Duke, R. E.; Luo, R.; Walker, R. C.; Zhang, W.; Merz,
K. M.; Roberts, B.; Hayik, S.; Roitberg, A.; Seabra, G.; Swails, J.; Götz,
A. W.; Kolossváry, I.; Wong, K. F.; Paesani, F.; Vanicek, J.; Wolf, R.
M.; Liu, J.; Wu, X.; Brozell, S. R.; Steinbrecher, T.; Gohlke, H.; Cai,
Q.; Ye, X.; Wang, J.; Hsieh, M.-J.; Cui, G.; Roe, D. R.; Mathews, D.
H.; Seetin, M. G.; Salomon-Ferrer, R.; Sagui, C.; Babin, V.; Luchko,
T.; Gusarov, S.; Kovalenko, A.; Kollman, P. A. AMBER 12; University
of California: San Francisco, 2012.
(28) Guillen Schlippe, Y. V.; Hartman, M. C. T.; Josephson, K.;
Szostak, J. W. In Vitro Selection of Highly Modified Cyclic Peptides
That Act as Tight Binding Inhibitors. J. Am. Chem. Soc. 2012, 134,
10469−10477.
(43) Wang, J.; Cieplak, P.; Kollman, P. A. How Well Does a
Restrained Electrostatic Potential (RESP) Model Perform in
Calculating Conformational Energies of Organic and Biological
Molecules? J. Comput. Chem. 2000, 21, 1049−1074.
2632
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
(44) Bayly, C. I.; McKay, D.; Truchon, J.-F. An Informal AMBER
Small Molecule Force Field: parm@Frosst. http://www.ccl.net/cca/
data/parm_at_Frosst/parm_Frosst_note.pdf (accessed Nov 1, 2010).
(45) Jakalian, A.; Jack, D. B.; Bayly, C. I. Fast, efficient generation of
high-quality atomic charges. AM1-BCC model: II. Parameterization
and validation. J. Comput. Chem. 2002, 23, 1623−1641.
̈
(46) Schrodinger Release 2012: Protein Preparation Wizard;
̈
Schrodinger, LLC: New York, NY, 2012.
(47) Watts, K. S.; Dalal, P.; Tebben, A. J.; Cheney, D. L.; Shelley, J.
C. Macrocycle Conformational Sampling with MacroModel. J. Chem.
Inf. Model. 2014, 54, 2680−2696.
(48) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical Treatment of Solvation for Molecular Mechanics and
Dynamics. J. Am. Chem. Soc. 1990, 112, 6127−6129.
(49) Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. The GB/
SA Continuum Model for Solvation. A Fast Analytical Method for the
Calculation of Approximate Born Radii. J. Phys. Chem. A 1997, 101,
3005−3014.
(50) Reddy, M. R.; Erion, M. D.; Agarwal, A.; Viswanadhan, V. N.;
McDonald, D. Q.; Still, W. C. Solvation Free Energies Calculated
Using the GB/SA model: Sensitivity of Results on Charge Sets,
Protocols, and Force Fields. J. Comput. Chem. 1998, 19, 769−780.
(51) Weiser, J.; Shenkin, P. S.; Still, W. C. Approximate Solvent-
Accessible Surface Areas from Tetrahedrally Directed Neighbor
Densities. Biopolymers 1999, 50, 373−380.
(52) Weiser, J.; Shenkin, P. S.; Still, W. C. Approximate Atomic
Surfaces from Linear Combinations of Pairwise Overlaps (LCPO). J.
Comput. Chem. 1999, 20, 217−230.
(53) Weiser, J.; Shenkin, P. S.; Still, W. C. Fast, Approximate
Algorithm for Detection of Solvent-Inaccessible Atoms. J. Comput.
Chem. 1999, 20, 586−596.
(54) Weiser, J.; Shenkin, P. S.; Still, W. C. Optimization of Gaussian
Surface Calculations and Extension to Solvent-Accessible Surface
Areas. J. Comput. Chem. 1999, 20, 688−703.
(55) Banks, J. L.; Beard, H. S.; Cao, Y.; Cho, A. E.; Damm, W.;
Farid, R.; Felts, A. K.; Halgren, T. A.; Mainz, D. T.; Maple, J. R.;
Murphy, R.; Philipp, D. M.; Repasky, M. P.; Zhang, L. Y.; Berne, B. J.;
Friesner, R. A.; Gallicchio, E.; Levy, R. M. Integrated Modeling
Program, Applied Chemical Theory (IMPACT). J. Comput. Chem.
2005, 26, 1752−1780.
(56) Halgren, T. A. MMFF VII. Characterization of MMFF94,
MMFF94s, and other widely available force fields for conformational
energies and for intermolecular-interaction energies and geometries. J.
Comput. Chem. 1999, 20, 730−748.
(57) Onufriev, A.; Bashford, D.; Case, D. A. Exploring Protein
Native States and Large-Scale Conformational Changes with a
Modified Generalized Born Model. Proteins 2004, 55, 383−394.
(58) Guzaev, A.; Manoharan, M. Conjugation of Oligonucleotides
via an Electrophilic Tether: N-Chloroacetamidohexyl Phosphorami-
dite Reagent. Bioorg. Med. Chem. Lett. 1998, 8, 3671−3676.
(59) Goetz, G. H.; Farrell, W.; Shalaeva, M.; Sciabola, S.; Anderson,
D.; Yan, J.; Philippe, L.; Shapiro, M. J. High Throughput Method for
the Indirect Detection of Intramolecular Hydrogen Bonding. J. Med.
Chem. 2014, 57, 2920−2929.
(60) Dickson, C. J.; Hornak, V.; Bednarczyk, D.; Duca, J. S. Using
Membrane Partitioning Simulations to Predict Permeability of Forty-
Nine Drug-Like Molecules. J. Chem. Inf. Model. 2019, 59, 236−244.
(61) Zhou, L.; Yang, L.; Tilton, S.; Wang, J. Development of a high
throughput equilibrium solubility assay using miniaturized shake-flask
method in early drug discovery. J. Pharm. Sci. 2007, 96, 3052−3071.
2633
https://dx.doi.org/10.1021/acs.jmedchem.0c01505
J. Med. Chem. 2021, 64, 2622−2633