Lipophilic Permeability Efficiency Reconciles the Opposing Roles of Lipophilicity in Membrane Permeability and Aqueous Solubility

Article abstract of DOI:10.1021/acs.jmedchem.8b01259

As drug discovery moves increasingly toward previously "undruggable" targets such as protein-protein interactions, lead compounds are becoming larger and more lipophilic. Although increasing lipophilicity can improve membrane permeability, it can also incur serious liabilities, including poor water solubility, increased toxicity, and faster metabolic clearance. Here we introduce a new efficiency metric, especially relevant to "beyond rule of 5" molecules, that captures, in a simple, unitless value, these opposing effects of lipophilicity on molecular properties. Lipophilic permeability efficiency (LPE) is defined as log D^7.4_dec/w - m_lipocLogP + b_scaffold, where log D^7.4_dec/w is the experimental decadiene-water distribution coefficient (pH 7.4), cLogP is the calculated octanol-water partition coefficient, and m_lipo and b_scaffold are scaling factors to standardize LPE values across different cLogP metrics and scaffolds. Using a variety of peptidic and nonpeptidic macrocycle drugs, we show that LPE provides a functional assessment of the efficiency with which a compound achieves passive membrane permeability at a given lipophilicity.

Full text of DOI:10.1021/acs.jmedchem.8b01259

Subscriber access provided by Université de Strasbourg - Service Commun de la Documentation
Article
Lipophilic Permeability Efficiency (LPE) reconciles the opposing
roles of lipophilicity in membrane permeability and aqueous solubility
Matthew R Naylor, Andrew M Ly, Mason J Handford, Daniel P Ramos, Cameron R Pye, Akihiro
Furukawa, Victoria Klein, Ryan P Noland, Quinn Edmondson, Alexandra C Turmon, William M Hewitt,
Joshua Schwochert, Chad E. Townsend, Colin N Kelly, Maria-Jesus Blanco, and R. Scott Lokey
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01259 • Publication Date (Web): 05 Nov 2018
Downloaded from http://pubs.acs.org on November 5, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
TITLE PAGE
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Lipophilic Permeability Efficiency (LPE) reconciles the opposing roles of lipophilicity in membrane
permeability and aqueous solubility
a
a
a
a
b
Matthew R. Naylor , Andrew M. Ly , Mason J. Handford , Daniel P. Ramos , Cameron R. Pye , Akihiro
c
a
a
a
a
Furukawa , Victoria G. Klein , Ryan P. Noland , Quinn Edmondson , Alexandra C. Turmon , William M.
d
b
a
a
e
a,
Hewitt , Joshua Schwochert , Chad E. Townsend , Colin N. Kelly , Maria-Jesus Blanco , R. Scott Lokey *.
aDepartment of Chemistry and Biochemistry, University of California Santa Cruz, 1156 High St., Santa
b
c
Cruz, CA 95064, United States; Unnatural Products, 335 Shake Mill Rd, Santa Cruz, CA 95060; Modality
Research Laboratories, Daiichi Sankyo Co., Ltd., 1-2-58 Hiromachi, Shingawa-ku, Tokyo 140-8710, Japan;
d
e
Loxo Oncology, Inc. 3122 Sterling Cir. Suite 200 Boulder CO 80301 Sage Therapeutics, 215 First Street, Suite
220, Cambridge, MA 02142, United States;
KEYWORDS permeability, lipophilicity, bRo5, ADME, cyclic peptide, macrocycle, partition coefficient,
conformation
ABSTRACT. As drug discovery moves increasingly toward previously “undruggable” targets such as protein-protein
interactions, lead compounds are becoming larger and more lipophilic. Although increasing lipophilicity can
improve membrane permeability, it can also incur serious liabilities, including poor water solubility, increased
toxicity, and faster metabolic clearance. Here we introduce a new efficiency metric, especially relevant to “beyond-
Rule of 5” molecules, that captures, in a simple, unitless value, these opposing effects of lipophilicity on molecular
properties. Lipophilic permeability efficiency (LPE) is defined as logD7.4dec/w – mlipo*cLogP + bscaffold, where logD7.4dec/w
is the experimental decadiene-water distribution coefficient (pH 7.4), cLogP is the calculated octanol-water
partition coefficient, and mlipo and bscaffold are scaling factors to standardize LPE values across different cLogP metrics
and scaffolds. Using a variety of peptidic and non-peptidic macrocycle drugs, we show that LPE provides a
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
functional assessment of the efficiency with which a compound achieves passive membrane permeability at a given
lipophilicity.
INTRODUCTION
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The property space of most small molecule drugs is bounded by the physico-chemical requirements for oral
absorption and/or membrane permeability, giving rise to the concept of “drug-likeness” and a host of associated
metrics that are often used to guide decision making in medicinal chemistry. The most famous of these guidelines,
the “Rule of 5,” describes limits on molecular weight and polarity that correlate with oral bioavailability, which are
primarily determined by membrane permeability, aqueous solubility, and metabolic stability1, 2. In addition to
placing constraints on molecular size, embodied within the Rule of 5 is a fundamental tension between achieving
the minimum lipophilicity required for membrane permeability on the one hand, and the negative consequences
of lipophilicity on properties such as aqueous solubility, metabolic stability, target specificity, and toxicity on the
other.
Despite widespread efforts to adhere to guidelines such as the Rule of 5, drugs have become larger and more
lipophilic over time3-10. The push toward targets previously considered “undruggable” is motivating an exploration
of larger, more complex molecular scaffolds such as peptidomimetics, macrocycles and cyclic peptides for lead
material, although increasing MW often comes with an increase in lipophilicity and an associated erosion in drug-
like properties. As ligand size increases and binding surfaces become correspondingly larger, adding polar contacts
that favor binding becomes increasingly difficult due to the geometric requirements of these primarily enthalpic
interactions11-15. In contrast, binding can be improved more reliably by the addition of less spatially demanding
hydrophobic contacts, which are primarily entropically driven16, 17. Thus, in early lead discovery, the improvement
in potency that results from increasing molecular size also often comes at the expense of added lipophilic liability.
Furthermore, with increasing MW, a compensating increase in lipophilicity is required to maintain the same level
of cell permeability14, 18-25
.
In light of these challenges, various efficiency metrics have emerged to assess the “quality” of lead candidates by
normalizing their potencies against MW and/or lipophilicity8, 9, 17, 26. Ligand efficiency (LE), for example, is the
affinity of a ligand expressed in terms of ΔG divided by the ligand’s heavy (i.e., non-hydrogen) atom count (HAC)
2
ACS Paragon Plus Environment

Page 3 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
LE = ΔG/HAC), while lipophilic ligand efficiency (LLE)4, 27-30 represents a compound’s biochemical potency
normalized by its octanol/water partition coefficient (logPoct/w), the most common metric for quantifying
lipophilicity (LLE = pIC50 – logPoct/w). However, despite the importance of membrane permeability in drug
discovery, there are no similar metrics for assessing membrane permeability in the context of a molecule’s lipophilic
liability.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Here we describe a new metric in which the two opposing roles of lipophilicity are parametrized into separate
terms, one representing membrane permeability-relevant lipophilicity (“logPmembrane”) and the other representing
solubility-relevant lipophilicity (“logPsol”). Based on an analysis of a set of cyclic peptide and “beyond-Rule of 5”
(bRo5) model systems, we conclude that these are indeed separable quantities. Our results give rise to a new
efficiency metric, lipophilic permeability efficiency (LPE), in which “logPmembrane” is quantified as the experimental
1
,9-decadiene/water distribution coefficient at pH 7.4 (logD7.4dec/w; hereafter referred to as logDdec/w), and “logPsol
”
3
1
is quantified by the calculated octanol/water partition coefficient ALogP , a fragment-based 2D octanol/water
partition coefficient calculator which shows improved accuracy relative to the more familiar cLogP for more diverse
chemotypes31-33 including cyclic peptides33, 34. Thus, LPE reflects the efficiency with which a compound’s membrane
permeability is achieved at a given lipophilicity. Using a variety of model systems in which we are able to
systematically vary both “logPmembrane” and “logPsol,” we demonstrate the utility of LPE to assess and compare
permeability “performance” in terms of the degree to which lipophilicity must be introduced or removed in order
to achieve optimal membrane permeability.
RESULTS
Aqueous solubility is inversely proportional to logPoct/w and ALogP in small molecules.
Lipophilicity is often quantified in terms of partition coefficients between aqueous and organic media. Studies
using small molecule training sets have shown a strong negative correlation between thermodynamic aqueous
solubility (logSol) and logPoct/w35, 36. By contrast, the correlation between logPoct/w and membrane permeability is
relatively weak among different chemotypes and scaffolds37-39. The most accurate predictors relating lipophilicity
to membrane permeability (including the Ro5) combine logPoct/w with other factors such has HBD and HBA
counts40 41. This is due to the ability of octanol, which not only has a hydrogen bonding –OH group but also contains
a significant fraction of dissolved water, to interact with polar molecules, yielding an underestimation of the
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
desolvation penalty associated with the partitioning of highly polar groups (i.e. with significant H-bond acidity or
basicity) into the hydrocarbon interior of the membrane42 43. In contrast, membrane permeability in a variety of
systems, including model liposome bilayers44, 45, Caco-2 cells , and epidermal tissue , has been shown to correlate
39
46
strongly with hydrocarbon-water partition coefficients, especially with 1,9-decadiene19, 45, 47
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In order to determine the extent to which these correlations can be generalized to larger-MW systems, we
determined the octanol-water (logDoct/w) and hydrocarbon-water distribution coefficients (logDdec/w, defined by
partitioning between organic solvent and PBS at pH 7.4) in a library of cyclic hexapeptide-peptoid hybrids
2
6
(peptomers) with diverse side chains (Library 1) (Figure 1). The peptoid sidechains of Library 1 (R , R ) incorporated
20 different amines, representing aliphatic, arene, and heterocyclic moieties sampling varying numbers of HB
acceptors (but no HB donors). Previously, we had found a parabolic relationship between membrane permeability
4
8
(
determined in a cell-free parallel artificial membrane permeability assay [PAMPA]), and ALogP (Figure 1B) . This
relationship has been observed in other systems49-51 and has been rationalized as the result of two opposing
phenomena: 1) a positive correlation between lipophilicity and permeability at the low-ALogP (polar) end of the
continuum, and 2) a decrease in permeability observed in the high-ALogP (nonpolar) regime where aqueous
solubility and/or membrane retention become(s) limiting. In contrast to the parabolic relationship between ALogP
and permeability, there was almost no correlation between experimental logDoct/w and permeability (Figure 1C). In
the polar regime (ALogP < 3, blue markers), the correlation between logDdec/w and permeability was quite high (rsq
=
0.90), whereas in the non-polar regime (ALogP > 3, green and orange markers), the (negative) correlation
between logDdec/w and permeability was significantly weaker, indicating that the phenomena underlying
permeability at the low and high ends of the polarity continuum are fundamentally distinct from one
another(Figure 1D). Similar results were observed for an analogous hexapeptide library in which, in addition to side
chains, backbone N-methylation and stereochemistry were also varied (Supplementary Information, Figure S1).
4
ACS Paragon Plus Environment

Page 5 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Lipophilicity analysis of cyclic hexapeptomer scaffolds. (A) Side chain-diverse hexapeptomer design (1
scaffold, 400 compounds). PAMPA permeability (log cm/s), binned by ALogP (< 3: blue; 3—4: green; > 4: orange),
linear regressions defined for all compounds ALogP < 3, compared to: (B) ALogP, (C) logDoct/w , (D) logDdec/w (1,9-
decadiene). (E) Sidechain composition of Library 1.
Figure 2. The permeability of 8 cyclic hexapeptide diastereomers is linearly correlated to experimental distribution
coefficients. (A) Cyclic hexapeptide diastereomers. (B) MDCK Papp (log cm/s) vs. logD7.4 measured in 4 different solvents:
cyclohexane, 1,9-decadiene, toluene and 1-octanol.
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To test the efficacy of different solvents to capture permeability differences among closely related compounds,
we measured logD (in 1,9-decadiene, 1-octanol, cyclohexane and toluene) for 8 previously identified diastereomeric
cyclic hexapeptides whose cell permeabilities in a low-efflux MDCK cell line52 varied by more than two orders of
5
3
magnitude (1-8, Figure 2) . Although all three hydrocarbon solvents (decadiene, cyclohexane and toluene) yielded
linear correlations between cell permeability and logD, the correlation with logDdec/w was superior (rsq = 0.92).
Although the correlation with toluene was also good (rsq = 0.86), compared to 1,9-decadiene, there was a 100-fold
increase in solvation of compound in toluene, suggesting solvent interactions intermediate between aliphatic
hydrocarbon solvents (decadiene, cyclohexane) and octanol. Although their low-dielectric conformations have not
been determined experimentally, the high correlation between logDdec/w and passive cell permeability in this
stereoisomeric series demonstrates the ability of a simple hydrocarbon/water partition coefficient to quantitatively
reflect the 3-dimensional conformational factors that underlie “logPmembrane”. In contrast to their permeabilities,
which span nearly 2 orders of magnitude, compounds 1-8 all have similar aqueous solubilities (9-54 µM), consistent
with the good correlation between the 2-dimensional ALogP and solubility observed for small molecules.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. (A) General structures of cyclic hexapeptide scaffolds based upon known structures, with variation in nonpolar
side chains (lipophilicity scan: Ala, Abu, Nva, Leu, Cha). Orange dashed lines indicate previously identified IMHB; red
dashed ovals indicate exposed HBD. Class A: 0 exposed HBD (○); Class B: 1 exposed side chain HBD (); Class C: 1 exposed
backbone HBD (); Class D: 2 exposed HBD (x); binned by ALogP (< 3: blue; 3—4: green; > 4: orange). Cell permeability
6
ACS Paragon Plus Environment

Page 7 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
MDCK-LE log cm/s) vs. (B) ALogP; (C) logDoct/w; (D) logDdec/w (linear regression defined for all compounds ALogP < 4).
Thermodynamic aqueous solubility (log M, pH 7.4) vs. (E) ALogP; (F) logDoct/w; (G) logDdec/w
Passive permeability and solubility for hexapeptides classified by HBD count.
To further extend and quantify the relationship between logDdec/w and permeability, and between ALogP and
solubility, four classes of compounds were synthesized classified by the number and type (backbone vs. side chain)
of exposed HBD (Figure 3; Classes A-D). Classes A and B are based on the well-established tri-N-methylated cyclic
hexapeptide scaffold 1-NMe3 (34)49, 54-58, which is known to adopt a stable, low-dielectric conformation with both
of its backbone amide NH groups involved in transannular IMHB. Classes C and D are based on a non-N-
methylated cyclic hexapeptide scaffold (52) whose low-dielectric conformation has one solvent-exposed backbone
amide NH, with the other amide NH groups either participating in IMHB or being sterically occluded from
solvent53, 59. Compounds in Classes A and C contain only simple hydrocarbon side chains with no hydrogen bonding
groups, whereas compounds in Classes B and D contain a Tyr residue and thus display a single solvent-exposed
phenolic –OH. Within each class, a broad range of lipophilicities (0 < ALogP < 7) was sampled by incorporating
different combinations of simple aliphatic and aromatic side chains (Ala, Abu, Nva, Leu, Phe, and Cha) with no
contributions to HB-acidity or basicity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The parabolic relationship between passive cell permeability and lipophilicity observed in Library 1 (Figure 1A)
was also observed in the Class A-D compounds (Figure 4 B-D). For compounds with ALogP < 4 (Figure 3D, blue
and green markers), a linear correlation between passive permeability and logDdec/w was observed for all 4 classes.
Class A compounds (no HBD) achieved higher permeabilities over a wider lipophilicity range than the other classes,
-6
with MDCK-LE transport rates at or above 10 x 10 cm/s within the ALogP range 2 < ALogP < 5. In contrast, Classes
B and C (1 HBD) had a smaller number of permeable compounds, which fell within a narrower ALogP window,
-6
while only a few Class D (2 HBD) compounds achieved permeabilities above 1 x 10 cm/s.
The decrease in permeability above ALogP ≈ 4 for each class corresponded to a drop in aqueous solubility (Figure
3
E). Consistent with a recent report describing the generality of 2D metrics such as topological polar surface area
60
(
TPSA) in predicting solubility in “beyond-Rule of 5” molecules , a very good negative correlation between
solubility and ALogP was observed for all four cyclic peptide classes (Figure 3E). In contrast, the correlation
between solubility and experimental logDoct/w was relatively weak (Figure 3F), possibly due to experimental
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
limitations above ALogP ≈ 4 (corresponding to logDoct/w > 3). The correlation between solubility and logDdec/w
(Figure 3G), while somewhat strong within each class, showed a relatively weak correlation among all classes.
Lipophilic Permeability Efficiency (LPE) quantifies the exposed H-bond acidity of diverse scaffolds.
The above data, combined with the extensive literature on small molecules, suggests logDdec/w as an operational
surrogate for membrane permeability that is less subject to the negative influence of solubility in the high-
lipophilicity (high-ALogP) regime. Similarly, ALogP provides an overall lipophilicity index that correlates with
aqueous solubility across diverse scaffolds. Plotting logDdec/w vs. ALogP for compounds 9 - 57 (Figure 4) reveals an
excellent linear relationship within each class, with a common slope of 1.06 (lipophilic correlation, mlipo, rmsd =
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.13, defined by all compounds with logDdec/w < 2 to ensure experimental data quality), in which each sidechain
methylene gives rise to ALogP and logDdec/w increments of +0.5 units. The averaged y-intercepts (bscaffold, defined for
each scaffold) represent the level of HBD exposure for each scaffold. Addition of a constant (+ 5.47 = - bscaffold,D
)
-6
scales the y-intercept of Class D to 0, since the most permeable Class D compounds achieve a Papp of 1 x 10 cm/s,
which is commonly considered the minimum acceptable permeability for achieving oral absorption. Using these
scaling factors, we define LPE as:
퐿푃퐸 = 푙표푔퐷푑푒푐/푤 ―1.06퐴퐿표푔푃 +5.47
(1)
Eq. 1 quantifies the efficiency with which any compound achieves passive membrane permeability at a given
3
1
“
logPsol,” which is conveniently captured as the calculated 2-dimensional metric ALogP . This equation thus
summarizes the scaffold-defined permeability outcomes for each class (A-D) in a single value that is independent
of a particular compound’s overall lipophilicity (Figure 4B). Although numerous LogPoct/w calculators are in
common use with varying accuracy33, 34, 61, the Supplementary Information tabulates calculated mlipo and bscaffold,D for
a selection of these metrics, transforming and scaling each calculated LogPoct/w metric to predict permeability
outcomes with comparable LPE values (Table S2).
8
ACS Paragon Plus Environment

Page 9 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. (A) Experimental logDdec/w vs ALogP. Class A: 0 exposed HBD (○); Class B: 1 exposed side chain HBD (); Class
C: 1 exposed backbone HBD (); Class D: 2 exposed HBD (x); binned by ALogP (< 3: blue; 3—4: green; > 4: orange). (B)
a
Linear fit data for each class and LPE assignments per eq. 1., for all compounds with logDdec/w < 2. y-intercept of the linear
b
c
d
fit for each class, ΔLPE relative to Class A. Sidechain HBD, backbone HBD.
By scaling to the least permeability-efficient Class D, LPE = 0 defines the minimum efficiency required for a
scaffold in this size range (MW = 700) to achieve a Papp of 1 x 10-6 cm/s at ALogP = 4, which is the maximum
lipophilicity allowed for achieving an aqueous solubility of 10 µM. The general lipophilicity properties required to
-6
reach a Papp of at least 1 x 10 are logDdec/w > - 1 and ALogP < 5, defining the permeable quadrant of Figure 4A.
LPE thus provides a measure of the maximum achievable permeability for a given scaffold within the limits of
aqueous solubility. Although these targets and bscaffold are defined on the convenient reference point of hexapeptide
scaffolds, the other parameters (logDdec/w, ALogP and mlipo) are fundamental biophysical descriptors that are
expected to be relevant to any compound of interest. Any deviation in LPE between related compounds identifies
a change in minimum total polar exposure (in the alkane-like dielectric of the membrane), whether due to
conformational effects or simply by the addition of solvent-exposed hydrogen bond donors. Exposure of a single
HBD, either at the side chain (Class B) or backbone (Class C) level, imposes a penalty of approximately -2 LPE
units, representing a 100-fold penalty in partition coefficients for both species in hydrocarbon solvent (ΔLPE
relative to Class A). Class D represents the additive consequence of both HBD types, yielding an LPE penalty of
4
approximately -4. Thus, exposure of two HBD groups represents a 10 -fold decrease in decadiene-water partition
coefficient and a corresponding decrease in permeability peak and range for Class D (Figure 3B).
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Chemical scaffolds and permeability performance
The curves relating passive permeability to ALogP in Library 1 (Figure 1) and Classes A-D (Figure 3B) demonstrate
the concept of “permeability performance” predicted by LPE, in which scaffold structure defines both the peak
permeability and the range of 2-dimensional lipophilicity (typically imparted by sidechains) within which
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
-6
acceptable permeability (Papp > 1 x 10 cm/s) can be achieved. For example, the theoretical compounds represented
by data points 1, 3, and 5 in Figure 5 have similarly low permeabilities (Papp), and a comparison of their Papp values
alone would not be informative. However, the LPE of compound 3 (scaffold B) is significantly higher than that of
compound 1 (scaffold A), suggesting that the permeability of 3 can be enhanced with a relatively modest increase
in ALogP (e.g., by adding aliphatic groups to generate 4), whereas 5 would require a decrease in ALogP to achieve
optimal permeability. On the other hand, efforts to improve the permeability of 1 by increasing ALogP will be
thwarted by the solubility cliff that looms above ALogP ~ 4.
The association of higher permeability performance with higher LPE (independent of measured Papp) defines an
analytical framework for identifying scaffold structure capable of accessing quadrant I of the Biopharmaceutical
Classification System (Figure 5A, high permeability and high aqueous solubility)62, 63. Thus, comparing LPE values
can be more informative than membrane transport rates alone, which are complicated not only by their parabolic
relationship to lipophilicity, but also the often disparate results from different cell types or assay methods (e.g. cell-
64
based permeability vs. PAMPA, see Figure S2) . The experimental quantification of the degree to which any change
on the scaffold or periphery) influences membrane-relevant lipophilicity provides an orthogonal measurement
that is less impacted by artifacts related to high lipophilicity (e.g., high membrane retention, aggregation, etc.).
(
10
ACS Paragon Plus Environment

Page 11 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Scaffold permeability performance, as predicted by Lipophilic Permeability Efficiency (LPE, eq. 1) (A)
Comparison of theoretical chemical scaffolds A (green, “poor”) and B (blue, “good”) via Papp and ALogP. Sampling
nonpolar side chain lipophilicity reveals greater “performance” for scaffold B: higher peak Papp, and satisfactory Papp over
a larger range of lipophilicity. Quadrants are labeled per the Biopharmaceutical Classification System (BCS). (B)
Quantification of LPE (as class-average or single-point measurements) ranks “poor” scaffold A and “good” scaffold B,
predicting overall permeability performance, and the opportunity to access BCS quadrant I (high permeability and
solubility) with appropriate lipophilic optimization. Suggested axis limits (ALogP < 5 and logDdec/w > -1) assigned from
hexapeptide data in Figs. 1, 4, although the logDdec/w limits will vary by molecular size (see Figures 7 and 12).
Sidechain ΔLPE penalties for monomers and libraries
Although measured permeability rates are useful for assigning the properties of large library collections, LPE
analysis enables the evaluation of scaffold permeability performance independent of the ALogP sampled by
individual sidechains composing each library member. This was demonstrated via analysis of capped monomer
species sampling various druglike polarities (listed fully in the Supporting Information, Tables S8, S9), including
the full chemical diversity of Library 1 (Figure 6A). Sidechains measuring high LPE as monomers (e.g. M4, LPE =
4
.5, Figure 6B) compose the library subset with highest permeability performance (Figure 6D). Conversely,
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
sidechains measuring low LPE as monomers (e.g. M15, LPE = 3.0, Figure 6B) compose the library subset with lowest
permeability performance (Figure 6D). ΔLPE measured between these sidechain monomers (e.g. monomer
ΔLPEM15-M4 = -1.5) also predicts the ΔLPE for all library members representing that sidechain comparison (e.g. library
ΔLPEA16-A11 = -1.6 based upon linear regressions to m = 1.07, Figure 6D,E, Table S7). Although the pyridyl moiety is
known to have a deleterious effect on membrane permeability53, 55, the monomer ΔLPE generates a quantitative
prediction for this effect (or any other moiety of interest) on any scaffold of interest. Various druglike polarities are
sampled via these monomers in Tables S8, S9, although highly-polar residues (e.g. Glu, Lys, etc.) could not be
quantified due to dynamic range limits at low logDdec/w, suggesting severe LPE penalties for greater HB acidity than
Tyr. Individual deviations from these predictions will identify opportunities or liabilities for optimizing
lipophilicity into passive permeability due to sidechain-sidechain or sidechain-backbone interactions not measured
by monomer LPE values. Although these monomer penalties can also be derived from Library 1 via matched-pair
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
7
analysis (Figure S7) or predicted via hydrocarbon-defined QSPR-based models , the simple LPE measurement
provides a framework for derivatives and libraries not compatible with matched-pair analysis.
Figure 6. LPE-directed prediction and analysis of Library 1. (A) Monomers for calculation of peptoid and amino
acid ΔLPE; full structural and analytical details listed in Tables S8, S9. (B) Measured ΔLPE for a representative
monomer pair. (C) Library 1 depicted as logDdec/w vs ALogP, colored by PAMPA Papp; the library subsets composed
by A11 and A16 at position R2 are selected in (D), and represented structurally in (E).
12
ACS Paragon Plus Environment

Page 13 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Case study: cyclosporins
1
LPE analysis of a collection of cyclosporins and two synthetic derivatives (Bmt -to-Leu) describes the
permeability landscape for this class of natural products (58-66, Table 1). All compounds that share the same
scaffold as CSA (62) (defined by stereochemistry and N-methylation) fall on the same LPE line (LPE = 3.3)(Figure
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
7B). These compounds maintain satisfactory MDCK and PAMPA (not shown) permeability within a narrow
lipophilicity window (3.2 < ALogP < 4.4), but their permeabilities fall precipitously beyond ALogP ≈ 4.5 (Figure 7C),
-6
giving Papp in MDCK cells of less than 1 x 10 cm/s for CSD (64, ALogP = 4.65) and valspodar (66, ALogP = 4.90)
and a Papp that was below the limit of detection for MeBmt-O-acetyl-CSA (65, ALogP = 4.71). Thus, the cyclosporins
show the same parabolic relationship between ALogP and permeability as seen in the smaller systems described
above.
11
The epimer of CSA at Val (CSH, 63) shows a very slight decrease in both LPE and Papp compared to CSA,
suggesting a minimal impact of stereochemistry at this position on the scaffold’s low-dielectric conformation. The
1
synthetic derivative “iso-CSA” (66), the product of an acid-catalyzed N -to-O acyl shift at the MeBmt residue, shows
a significant deviation from the LPE line defined by the other CSA derivatives. The x-ray crystal structure of iso-
CSA is entirely different from that of CSA, showing two solvent-exposed backbone NH groups in addition to the
65
NH of the secondary amine resulting from the acyl shift . Although the crystal structure does not necessarily
represent its membrane-associated conformation, the deviation of the iso-CSA structure from the canonical low-
dielectric structure of CSA, combined with the introduction of an ionizable amine, is consistent with the observed
1
decrease in 1.5 LPE units relative to CSA. Interestingly, despite the removal of the MeBmt β-hydroxyl group in
synthetic derivatives 60, 61,65 and 66, only a small LPE increase is observed for 60 and 61 relative to CSA, consistent
1
with NMR evidence showing the presence of an IMHB between the β-hydroxyl of the MeBmt residue and its own
backbone C=O66, 67. CSC (58) differs from CSA by an Abu-to-Thr mutation at R , and therefore contains an
2
2
additional side chain –OH. Nonetheless, CSC has the same LPE as CSA, suggesting that the Thr -OH of CSC may
also engage in side-chain-to-backbone IMHB. Indeed, this H-bond is observed in the x-ray crystal structure of CSC
bound to cyclophilin68, 69, suggesting that the same interaction is also accessible in low-dielectric media.
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. LPE analysis of cyclosporin derivatives, natural and synthetic. (A) Structure of cyclosporin A and derivatives.
(
B) logDdec/w (10x excess of aqueous layer) vs. ALogP assigns most cyclosporin derivatives along a common line. (C)
MDCK-LE cell permeability (log cm/s) defines a common curve for permeability on the cyclosporin scaffold. 65 could not
be detected in permeability assays.
The results for the CSA derivatives (MW ≈ 1200) show that the underlying biophysical phenomena measured by
LPE extend into larger, more complex molecules. The LPE values of CSA and its simple congeners (LPE = 4.3 - 4.9)
are higher than those of the high-performance Class A compounds (LPE = 3.7), and yet their permeabilities are
-6
significantly lower. In fact, for the CSA derivatives, the logDdec/w boundary for achieving a Papp of 1 x 10 cm/s is 3.5
orders of magnitude above the same threshold for the hexapeptides (hexapeptides: logDdec/w > - 1.0; cyclosporins:
logDdec/w > 2.5). By contrast, the ALogP threshold for solubility is similar between the cyclosporins and the
hexapeptides. Thus, in the size range above MW ~ 1000, even a single unsatisfied H-bond donor will decrease
permeability below the 1 x 10-6 cm/s threshold. These data support similar conclusions reached by Whitty63,
Kihlberg60 and Waring14 that for compounds of MW 1000 and beyond, chameleonicity is strongly favored as a
mechanism for simultaneously achieving the low-dielectric lipophilicity (e.g. high LPE) required for permeability
while also staying below the maximum ALogP threshold for maintaining solubility.
Table 1. Cyclosporins experimental data
c
ID
ALogP
Sola
logDdec/wb
LPE
P
app
14
ACS Paragon Plus Environment

Page 15 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
5
8
9
3.30
3.81
3.99
4.31
4.33
4.33
4.65
4.71
33
27
2.43 ± 0.18
3.18 ± 0.01
3.64
4.4
4.6
4.9
4.8
4.6
4.3
4.4
4.6
0.5 ± 0.1
1.4 ± 0.2
2.7 ± 0.6
1.8 ± 0.3
1.4 ± 0.2
1.1 ± 0.3
0.7 ± 0.1
--d
5
60
10
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
6
6
6
1
22
3.94 ± 0.31
3.70 ± 0.11
3.44 ± 0.12
3.86 ± 0.10
4.14 ± 0.63
2
3
4
10
4.3
0.12
0.005
65
6
6
6
7
4.90
4.92
b
7.8
2.2
4.11 ± 0.22
2.61 ± 0.46
4.4
2.9
0.2 ± 0.1
0.2 ± 0.1
c
6
a
-6
uM, pH 7.4. Performed with 10x excess of PBS buffer for increased dynamic range MDCK-LE, x10
d
cm/s. No signal detected.
Structure-permeability relationships via Ala- and Pro-scanning
Ala-scanning, which is often used to identify structure-activity relationships in peptides, can also be used in the
context of LPE analysis to study structure-permeability relationships. Substitution of nonpolar residues with Ala
should have no impact on LPE unless the side chains of those residues affect permeability-relevant lipophilicity,
for example, by steric occlusion of polar backbone atoms. Ala scans of Class A compounds (13, 15, 17, 18, 68-72)
showed no significant LPE penalty for any Ala mutation, including double substitutions (13, 71), indicating a
minimal impact of side chains on intrinsic permeability performance with these scaffolds.
In addition to Ala-scanning as a method to probe the effect of specific side chains on properties, “Pro-scanning”
has been used to evaluate the impact of flexibility on molecular properties. However, Xaa-to-Pro substitutions not
only affect scaffold rigidity, but depending on the nature of Xaa, they can also have a significant impact on overall
lipophilicity (i.e. ALogP). For example, a recent study designed to assess the impact of flexibility on the
5
pharmacokinetic properties of 1-NMe3 (34) showed that substituting D-Pro with D-Leu in the parent scaffold
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
caused a decrease in cell permeability, while addition of a second rigidifying Pro residue (by substituting D-Leu2
56
for D-Pro in 34) led to an increase in permeability . These observations were attributed to a beneficial effect of
backbone rigidity on permeability, although each Leu-to-Pro substitution also decreased ALogP by 1.12 units, thus
raising the question whether the observed effect was simply the result of improved solubility that accompanied the
decrease in ALogP.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To address this question, we made analogous Pro-to-(NMe)Ala substitutions on compound 21, the parent
6
compound in Class A which differs from 1-NMe3 by the replacement of Tyr for Phe. In contrast to the Pro-to-
(NMe)Leu substitutions described above, substitution of a Pro residue with (NMe)Ala decreases ALogP by only -
-6
0.11 units. The parent compound 21 has low PAMPA permeability (0.7 x 10 cm/s) due to its high ALogP of 4.68,
5
and substitution of the D-Pro with (NMe) D-Ala to furnish 70 (ALogP = 4.57) had virtually no effect on permeability.
Substituting one of the Leu side chains in 21 to Ala (17) decreased the ALogP from 4.68 to 3.45, resulting in a
-6
5
dramatic increase in PAMPA permeability from 0.7 to 9.3 x 10 cm/s. Further substitution of D-Pro with (NMe) D-
Ala in 17 to provide 73 also had very little effect on permeability, indicating that flexibility in this system has very
little influence on passive permeability.
Figure 8. Lipophilicity-controlled flexibility outcomes from Ala- and Pro-scan on Class A compound 21. All
compounds measure the same LPE, with consequently similar Papp (PAMPA or MDCK) for similar ALogP.
The LPE values corroborate these results; compounds containing zero (70 and 72; avg. LPE = 3.67), one (13, 17,
15, 18, 21, 68, 69, and 71, avg. LPE = 3.67) or two (73, LPE = 3.49) rigidifying Pro residues all had average LPE values
16
ACS Paragon Plus Environment

Page 17 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
that were within 0.18 units, indicating that the imparted flexibility plays a small role if any in the desolvation penalty
associated with membrane transport. In the case of these Pro-to-(NMe)Ala matched pairs, since their ALogP values
were very similar, their Papp values could be compared directly. However, in most cases a substitution designed to
explore conformational or steric effects will also affect ALogP-defined lipophilicity. By normalizing for 2-
dimensional lipophilicity, LPE provides direct insight into the effect of potential scaffold-altering substitutions on
permeability, independent of their effect on overall lipophilicity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Case study: cordyheptapeptides
Figure 9. (A) Chemical structure of cordyheptapeptides A—C (74—76). Non-H-bonding amide NH highlighted as
defined by (B) the crystal structure of 74. (C) LPE graph (logDdec/w vs. ALogP) of alanine scanning on 74. Black: natural
compounds; blue: Ala scan compounds. (D) PAMPA Papp (log cm/s) vs. ALogP.
Cordyheptapeptides A, B, and C (74-76)70-72 are fungally derived cyclic heptapeptide natural products whose
compositions, including extensive backbone N-methylation and the lack of charged or highly polar side chains, are
reminiscent of other passively permeable cyclic peptides. Indeed, 74-76 show measurable, albeit low, permeabilities
in both PAMPA and MDCK-LE cells (Table 2). Cordyheptapeptides A and C (74 and 76) each have a single Tyr and
therefore display an exposed side chain HBD, while cordyheptapeptide B contains a Phe at this position and
therefore lacks any exposed side chain HBD. Illustrating the consistency of LPE values between different systems,
the LPE of - 1.9 between cordyheptapeptides A and B (74 vs. 75) is very close to the LPE of - 2.1 between cyclic
hexapeptide Classes A and B (Figure 4B), which also differ by a Phe vs. Tyr substitution.
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
An X-ray crystal structure of cordyheptapeptide A (74) crystallized from MeOH shows a conformation in which
1
two of the backbone amide NH groups are involved in transannular IMHB, while one backbone NH (that of Ile ) is
70
3
6
free of any polar contacts, either intra- or intermolecular (Figure 9B) . Ala substitutions in 74 at Phe , Phe , and
7
Leu maintain the LPE of the parent scaffold (LPE = 0.9, 0.9, and 0.8, respectively), indicating that these side chains
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
have little impact on the low-dielectric conformation or degree of polar exposure in the cyclic peptide core.
2
Replacement of Tyr with Ala produces a compound (78) whose LPE, not surprisingly, is very similar to that of the
2
5
Tyr -to-Phe replacement (74) (LPE78 = 3.1 vs. LPE75 = 3.0). Replacing Pro with NMeAla (82) yields an increase in
LPE (ΔLPE82 - 74 = +1.0), with an associated increased Papp and aqueous solubility relative to parent 74. The contrast
of this result versus the negligible flexibility outcomes for the 1NMe3 scaffold highlights the scaffold-specific
consequences of flexibility, which are experimentally sampled with the LPE measurement.
Table 2. Cordyheptapeptides experimental data
ID
74
ALogP
Sola logDdec/w LPE
P
app
b
4.22
0.45
2.3
2.7
10
-0.13
1.1
3.0
1.1
1.2 ±
0.4
75
4.46
3.76
2.92
2.90
2.66
2.66
2.13
2.3 ±
1
.3
1.9 ±
.1
0.2 ±
.05
5.9 ±
.4
1.2 ±
.5
76
-0.43
-2.44
0.71
1
1
A 77
-0.1
3.1
0
2
A 78
16
1
3
A 79
17
-1.78
-1.71
0.9
0.9
0
6
n.d.c
A 80
1.3 ±
0.2
18
ACS Paragon Plus Environment

Page 19 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
7
A 81
2.98
4.10
9
-1.52
0.92
0.8
2.0
2.2 ±
.1
7.8 ±
.6
0
5
A 82
37
0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
b
-6
c
uM, pH 7.4. PAMPA, x10 cm/s. Not determined
1
Whereas substituting Ile with Val (76) maintains the LPE of the parent compound 74 (LPE76 = 1.1 vs. LPE74 = 0.9),
1
the Ile -to-Ala substitution (77) causes a significant drop in LPE (ΔLPE77 - 74 = -0.9), suggesting that -branching
may sterically shield the exposed NH at this position (Figure 9B). The LPE of 77 is almost identical to that of the
Class D hexapeptides, both of which have one Tyr OH and one fully exposed backbone NH, although the effect of
1

-branching at R (ΔLPE74 - 77 = 1.0) is roughly half that of removing an exposed backbone HBD altogether (e.g.,
ΔLPEClass A – Class C = 1.8). Analyzing the LPE of amino acid monomers presented above (Table S9), N-methylation of
an exposed backbone amide (e.g. ΔLPEAla - NMeAla = +1.2 or ΔLPEPhe - NMePhe = +0.8) gives rise to a ΔLPE of ≈1.0,
suggesting that the effect of steric shielding (e.g., by way of -branching in a side chain) is quantitatively similar to
the effect of backbone N-methylation on permeability efficiency. However, based on previous studies on the effect
of -branching on macrocycle properties73, 74, this effect is clearly scaffold-dependent. On the other hand, based on
4
4
previously reported logDdec/w values for a different series of di- and tripeptides , the ΔLPE of sequestering a single
amide in an IMHB (avg. ΔLPE(Class A – Class C) and (Class B – Class D) = 1.7) is similar to that of removing a residue entirely
(ΔLPE of p-tolyl-Ala-Ala-OH vs. p-tolyl-Ala-OH = 1.8). This similarity is not unexpected given that both the
complete removal of a residue and a single IMHB serve to eliminate the solvent exposure of both an NH and C=O
group.
LPE and other drug scaffolds
To ascertain whether the LPE framework is applicable to non-peptidic b-Ro5 drugs, we determined LPEs
for a series of rifamycin and erythronolide antibiotics. The rifamycins are large macrocycles (MW ~823-877,
-6
>
5 HBD) with moderate-to-high Caco-2 permeabilities (5-10 x 10 cm/s) and high oral absorption (50-70
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
%
F). As previously noted by Kihlberg and colleagues, this outlier behavior may be due in large part to the
75
extensive network of IMHB in these compounds , which correlates to their relatively modest minimum 3D
60
polar surface areas . The rifamycins have LPE values ranging from 2.1 to 3.1, equivalent to 0.5 to 1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
exposed HBD based on the Class A-D cyclic peptides described earlier. Although there is a good deal of
conformational heterogeneity in the crystal structures of the rifamycins, 5 of the 6 HBD in rifampicin (83),
and 4 out of the 5 HBD in rifabutin (84), are involved in IMHB. Thus, the LPE values of the rifamycins are
consistent with expectations based on their crystal structures and comparison with the Class A-D
compounds. The ~1.0 ΔLPE between rifampicin and rifabutin may be due to differences in HBA exposure
76
governed by the IMBH between the aminoquinonimine of rifabutin and the lactam carbonyl (Figure 11A) .
Similarly, LPE analysis of the erythronolides is also consistent with their reported ADME and PK properties.
The parent compound, erythromycin (85), has an LPE of approximately -0.3 (corresponding to logDdec/w
=
-3.9, at the dynamic range limit of the experiment with 10x excess of 1,9-decadiene), similar to our Class D
cyclic peptides with the equivalent of two exposed HBD. Erythromycin has 5 -OH groups, two of which are
involved in IMHB (Fig. xb, blue arrows) based on a crystal structure of erythromycin bound in a low-3D PSA
conformation to a protein sensor60, 77. This suggests that the three exposed -OH groups are not completely
exposed, or that other polar elements (in particular the dimethylamino group) contribute to lowering the
1
LPE. Methylation of one of the non-hydrogen bonded -OH groups (R ) to generate clarithromycin (86)
increases the LPE by 2.7 units, slightly higher than the LPE of ~2.0 that we observed upon removal an
20
ACS Paragon Plus Environment

Page 21 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
exposed phenolic -OH in Classes B vs. A, and D vs. C, and in the cordyheptapeptides 74 vs. 73.
2
Esterification of the R -OH in the aminoglycoside unit of erythromycin, producing the prodrug erythromycin
ethylsuccinate (87), increased the LPE by 2.0 units, to 1.7. This value is similar to the LPE of Classes B
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(1.6) and C (1.9) which contain one exposed HBD. In contrast, roxithromycin (88), in which the ketone is
replaced with an oxime ether, has an LPE of 3.0. Roxithromycin formally has the same level of HBD
exposure as clarithromycin (LPE = 2.4), suggesting that the 0.6-unit increase could be due to further polar
shielding in the macrocyle, possibly mediated by a new IMHB between a highly structured water molecule
and the lactone carbonyl (Figure 11B). The low LPE of both erythromycin and azithromycin is consistent
-6
with their poor cell permeabilities (Caco-2 Papp = 1-3 x 10 cm/s), while the high LPE of clarithromycin and
-6
roxithromycin reflects the high permeability of these derivatives (Caco-2 Papp > 10 x 10 cm/s).
Figure 10. Structural representations of the discussed (A) rifamycins and (B) erythronolides, with noted LPE
values, and red-highlighted HBD.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 11. Crystal structures of the discussed (A) rifamycins and (B) erythronolides. IMHB are noted with green
dashes, and exposed HBD are noted with arrows. A crystallographic water in roxithromycin (88) is shown in
magenta.
DISCUSSION AND CONCLUSIONS
LPE and aqueous solubility
The excellent correlation between ALogP and solubility in small molecules, as well as the various cyclic peptide
scaffolds studied here, supports the use of ALogP as a surrogate for solubility in the LPE equation. However, since
ALogP does not take 3-dimensional structure into account, it ignores the potential impact of conformational states
on solubility. Indeed, the importance of “chameleonicity” in determining both permeability and solubility in bRo5
compounds is well documented. For example, although CSA has only 1 major conformation in low dielectric
solvent, its dynamic conformational equilibrium in water is presumed to facilitate solvation78-80. We have also
observed a strong conformational effect on solubility in analogs of the natural product sanguinamide A, in which
a single Ile-to-Leu substitution led to a nearly 2-order of magnitude increase in solubility. In such scenarios,
experimental logDoct/w may be an appropriate substitute for ALogP, providing that both values lie within the
experimental dynamic range.
22
ACS Paragon Plus Environment

Page 23 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
LPE and molecular size
Permeability is inversely proportional to molecular volume, independent of alkane partition coefficients18, 19, 23, 25
.
However, LPE is not dependent on molecular size, since the molecular volume contributions for partitioning into
42
both 1,9-decadiene and octanol are very similar . Correspondingly, matched libraries of permethylated octa-, nona-
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
,
and decapeptides yielded identical LPE values despite dramatic reductions in Papp for their increasing molecular
25
volume . Thus, care should be taken when expecting permeability performance outcomes from ΔLPE between
compounds that vary significantly in size. However, increasing molecular size simply elevates the minimum LPE
and logDdec/w required for achieving acceptable permeability; the Class D hexapeptides, cordyheptapeptides,
-6
permethylated octapeptides, and cyclosporin scaffolds all barely achieve permeabilities above Papp = 1 x 10 cm/s,
even for members with optimized lipophilicity (Figure 12). Yet for the larger cyclosporins, a much higher LPE (~4.5)
is required to attain the same peremeability as the smaller Class D hexapeptides (LPE = 0.0). Combined, these
compounds define the minimum LPE required to achieve specified permeability goals at a given MW (green
-6
boundary in Figure 12). Higher goals (e.g. Papp = 10 x 10 cm/s and above, blue boundary in Figure 12) will require
even higher LPE, with severe penalties expected for increased molecular volume.
Figure 12. Approximate LPE and MW limits to achieve permeability goals. Papp boundaries are coarsely
approximated by the peak-Papp members of each scaffold class of the peptide classes discussed above (○), including
25
previously reported permethylated cyclic octa-, nona-, and decapeptides () . The bRo5 drugs are included (red
x) to demonstrate LPE optimization of non-peptidic scaffolds.
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Applying LPE as a general analytical framework in a drug discovery setting
Although we propose LPE as a general efficiency metric that reflects a compound’s permeability performance,
LPE does not necessarily correlate with permeability per se. For example, high-LPE compounds can have very low
permeabilities if they happen to fall either in the low- (highly polar) or high- (highly lipophilic) ALogP regimes.
Rather, LPE reflects the potential for a given scaffold to achieve high permeability at an optimal ALogP-defined
lipophilicity.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Several recent efforts have demonstrated the importance of framing the challenge of permeability (especially for
diverse bRo5 molecules) considering conformationally-defined lipophilicity. An analysis of structurally diverse
bRo5 drugs with multiple crystal structure conformations identified a strong correlation between the minimum 3D
60
PSA for these conformations and Caco-2 cell permeability . LPE is also conceptually similar to “∆푙표푔푃”, defined as
the difference beween octanol-water and cyclohexane-water partition coefficients, which has been used to quantify
“
chameleonicity” in small molecules and peptides81, 82. Hydrocarbon/water partition coefficients and
chromatographic models thereof have been presented as methods for evaluating small molecules by their
propensity to form IMHB83-86. LPE builds on these approaches by employing calculated rather than experimental
logDoct/w, thus reducing both the cost, error, and limited dynamic range of experimental logDoct/w measurements.
The scaling factors mlipo and bscaffold normalize LPE across 15 widely used calculated logPoct/w descriptors; however,
for standard LPE calculations we recommend calculators based on the simple ALogP octanol-water calculator
provided by Ghose & Crippen31, 87, which yielded the best correlation with aqueous solubility (Figure S4) and the
lowest RMSD in the relationships with logDdec/w (determined for Classes A-D)(Figure S5, Table S2). Operationally,
the logDdec/w assay is inexpensive, simple, and, with the use of modern LCMS, highly amenable to multiplexing.
Therefore, we propose the adoption of LPE as a routine part of the experimental characterization of compounds in
a drug discovery setting. We envision a variety of scenarios in which LPE may be useful in guiding medicinal
chemistry efforts in a drug discovery campaign:

Optimizing properties within a series of related compounds. Comparing experimentally derived
LPE values within a discrete compound series can inform the process of property optimization.
Depending on the LPE of the compound (placed in the context of its MW), maximizing permeability
may take different paths. For high-LPE compounds, permeability can be maximized simply by
24
ACS Paragon Plus Environment

Page 25 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
modulating ALogP (e.g., by the addition or removal of aliphatic groups). For lower-LPE compounds,
increasing permeability may not be achievable without incurring solubility penalties. In these cases,
optimization may first require increasing LPE, for example, by placing a sterically bulky group near an
exposed HBD, altering the scaffold to elicit IMHB, or by removing an exposed HBD altogether.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0

Prioritizing scaffolds in a lead development campaign. Comparison of LPE values between
different chemotypes or related scaffolds with different geometries (e.g., stereochemical or N-methyl
variants in cyclic peptides), can provide a means for prioritizing leads in a medicinal chemistry
campaign, even with poor Papp before lipophilic optimization for these leads. High-LPE scaffolds provide
a better starting point for potency optimization than low-LPE scaffolds, since higher-LPE leads will be
more tolerant of polar substituents (especially those containing HBD).

Defining structure-property relationships (SPR). The same synthetic series used to explore SAR
toward bioactivity optimization (e.g., Ala scanning for peptides) may now also define structure-property
relationships toward physicochemical optimization. Each ΔLPE effect yields hypotheses for the
influence of a particular structural change on a scaffold’s hydrocarbon-defined 3D structure: ΔLPE ~ 0
identifies handles for exploring SAR without influencing scaffold structure; ΔLPE < 0 identifies sites
essential for maintaining polarity-hiding structure; and LPE > 0 identifies structural changes that
improve overall permeability performance.

Predicting properties in an SAR series by combining calculated and experimental LPE values.
For virtual libraries based on a relatively complex core scaffold (e.g., derivatives based on a macrolide
core), LPE can be estimated using an experimentally determined value for the scaffold in combination
with calculated LPE values for individual R-group appendages, using QSPR-based models to predict
4
7
logDdec/w values for the R-groups . For high-LPE scaffolds, inclusion of one or two side chains with
exposed hydrogen bond donors may be tolerated, whereas for low-LPE scaffolds, side chain
combinations should be restricted to those with minimal (or positive) LPE consequences.
CONCLUSION
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Efficiency metrics have established their role for navigating the balance of target binding and physicochemical
properties in modern medicinal chemistry. Built upon the simple premise of modulating a desired property (e.g.
target binding energy or low clearance rates) by an undesired property (e.g. molecular size or lipophilicity) these
metrics have proven their utility on diverse data sets and have found clinical success in a field marked by rapidly
increasing challenges for bringing drugs to market26, 30. As medicinal chemists venture beyond Rule-of-5 chemical
space in pursuit of heretofore “undruggable” targets such as many protein-protein interactions, the biophysical
demands of achieving favorable physicochemical properties are only expected to steepen. LPE is a simple guide
for these efforts, measuring the structural factors for maximum passive cell permeability against lipophilic
liabilities, and conceptualizing their contribution to the permeability performance of chemical scaffolds.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EXPERIMENTAL SECTION
Detailed methods for synthesis, assays (permeability, solubility, and partition coefficients), and derivation are
described in the Supporting Information. LogD and solubility assays were measured as single points, permeability
assays were measured in quadruplicate; all assays were quantified via selected-mass quantification in UPLC-MS.
All individual compounds were purchased or synthesized as pure compounds with purity ≥95% purity via UV
absorbance. Library 1 was synthesized as purified mixtures of 32 compounds whereas the monomer compounds
were synthesized without purification; both were quantified via UPLC-MS and selected-mass monitoring.
ANCILLARY INFORMATION
Supporting Information. Experimental procedures, additional discussion for practitioners, and
characterization data (HPLC and NMR spectra) are provided. All experimental data is provided in the
attached spreadsheet, including SMILES molecular formula strings.
Abbreviations Used. LPE, Lipophilic Permeability Efficiency; logD, shake flask distribution
coefficient at pH 7.4; LogP, partition coefficient; ALogP, atomistic calculated octanol/water partition
coefficient; Papp, measured permeation rate (A-to-B, x10-6 cm/s); PAMPA; cell-free parallel artificial
membrane permeability assay; MDCK, Madin-Darby Canine Kidney cells; LE, low efflux; Caco-2, human
epithelial colorectal adenocarcinoma cells; PBS, phosphate-buffered saline; Sol, aqueous solubility in PBS
26
ACS Paragon Plus Environment

Page 27 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pH 7.4; ADME, absorption, distribution, metabolism, excretion; Ro5, Rule of 5; bRo5, beyond-Rule of 5;
QSPR, quantitative structure-activity relationship.
AUTHOR INFORMATION
Corresponding Author
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
*
Email: slokey@ucsc.edu
ORCID
Matthew R Naylor: 0000-0001-7043-5972
Victoria M Klein: 0000-0002-3438-2399
R. Scott Lokey: 0000-0001-9891-1248
ACKNOWLEDGMENTS.
This research was supported by Eli Lilly and Company through the Lilly Innovation Fellowship Award program.
References
1.
Lipinski, C.; Lombardo, F.; Dominy, B.; Feeney, P. Experimental and
computational approaches to estimate solubility and permeability in drug discovery and
development settings. Adv Drug Deliv Rev 1997, 23, 3-25.
2.
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 2001, 46, 3-26.
3.
Leeson, P. D.; St-Gallay, S. A. The influence of the 'organizational factor' on
compound quality in drug discovery. Nat Rev Drug Discov 2011, 10, 749-765.
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
.
Leeson, P. D.; Springthorpe, B. The influence of drug-like concepts on decision-
making in medicinal chemistry. Nat Rev Drug Discov 2007, 6, 881-890.
5.
Leeson, P. D.; Davis, A. M. Time-related differences in the physical property
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
profiles of oral drugs. J Med Chem 2004, 47, 6338-6348.
6.
Proudfoot, J. R. Drugs, leads, and drug-likeness: An analysis of some recently
launched drugs. Bioorg Med Chem Lett 2002, 12, 1647-1650.
7.
Vieth, M.; Siegel, M. G.; Higgs, R. E.; Watson, I. A.; Robertson, D. H.; Savin, K.
A.; Durst, G. L.; Hipskind, P. A. Characteristic physical properties and structural
fragments of marketed oral drugs. J Med Chem 2004, 47, 224-232.
8.
Wenlock, M. C.; Austin, R. P.; Barton, P.; Davis, A. M.; Leeson, P. D. A
comparison of physiochemical property profiles of development and marketed oral
drugs. J Med Chem 2003, 46, 1250-1256.
9.
Oprea, T. I. Current trends in lead discovery: Are we looking for the appropriate
properties? J Comput Aided Mol Des 2002, 16, 325-334.
10.
Proudfoot, J. R. The evolution of synthetic oral drug properties. Bioorg Med
Chem Lett 2005, 15, 1087-1090.
11.
Hann, M. M. Molecular obesity, potency and other addictions in drug discovery.
MedChemComm 2011, 2, 349-355.
12.
Freire, E. Do enthalpy and entropy distinguish first in class from best in class?
Drug Discov Today 2008, 13, 869-874.
13.
Ladbury, J. E.; Klebe, G.; Freire, E. Adding calorimetric data to decision making
in lead discovery: A hot tip. Nat Rev Drug Discov 2010, 9, 23-27.
14.
Waring, M. J. Defining optimum lipophilicity and molecular weight ranges for drug
candidates-molecular weight dependent lower logd limits based on permeability. Bioorg
Med Chem Lett 2009, 19, 2844-2851.
2
8
ACS Paragon Plus Environment

Page 29 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
5.
Gleeson, M. P.; Hersey, A.; Montanari, D.; Overington, J. Probing the links
between in vitro potency, admet and physicochemical parameters. Nat Rev Drug Discov
011, 10, 197-208.
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
16.
Hann, M. M. In Molecular obesity, potency and other addictions in drug
discovery, Dordrecht, 2015; Springer Netherlands: Dordrecht, 2015; pp 183-196.
17.
Meanwell, N. A. Improving drug design: An update on recent applications of
efficiency metrics, strategies for replacing problematic elements, and compounds in
nontraditional drug space. Chem Res Toxicol 2016, 29, 564-616.
18.
Xiang, T. X.; Anderson, B. D. The relationship between permeant size and
permeability in lipid bilayer membranes. J Membr Biol 1994, 140, 111-122.
19.
Nitsche, J. M.; Kasting, G. B. Permeability of fluid-phase phospholipid bilayers:
Assessment and useful correlations for permeability screening and other applications. J
Pharm Sci 2013, 102, 2005-2032.
20.
Xiang, T. X.; Anderson, B. D. Permeability of acetic acid across gel and liquid-
crystalline lipid bilayers conforms to free-surface-area theory. Biophys J 1997, 72, 223-
2
37.
1.
2
Xiang, T. X.; Anderson, B. D. Influence of chain ordering on the selectivity of
dipalmitoylphosphatidylcholine bilayer membranes for permeant size and shape.
Biophys J 1998, 75, 2658-2671.
22.
Lieb, W. R.; Stein, W. D. Implications of two different types of diffusion for
biological membranes. Nat New Biol 1971, 234, 220-222.
23.
Lieb, W. R.; Stein, W. D. Non-stokesian nature of transverse diffusion within
human red cell membranes. J Membr Biol 1986, 92, 111-119.
24.
Wolosin, J. M.; Ginsburg, H.; Lieb, W. R.; Stein, W. D. Diffusion within egg
lecithin bilayers resembles that within soft polymers. J Gen Physiol 1978, 71, 93-100.
29
ACS Paragon Plus Environment