Physicochemical determinants of passive membrane permeability: Role of solute hydrogen-bonding potential and volume

Article abstract of DOI:10.1021/jm010253i

The relationship of solute structure with cellular permeability was probed. Two series of dipeptide mimetics consisting of glycine, alanine, valine, leucine, phenylalanine, and cyclo-hexylalanine with amino acids in the D-configuration were prepared. Part

Full text of DOI:10.1021/jm010253i

J . Med. Chem. 2001, 44, 3721-3729
3721
P h ysicoch em ica l Deter m in a n ts of P a ssive Mem br a n e P er m ea bility: Role of
Solu te Hyd r ogen -Bon d in g P oten tia l a n d Volu m e
J ay T. Goodwin, Robert A. Conradi, Norman F. H. Ho, and Philip S. Burton*
Drug Absorption and Transport, Pharmaceutical Development, Pharmacia, Kalamazoo, Michigan 49007
Received J une 6, 2001
The relationship of solute structure with cellular permeability was probed. Two series of
dipeptide mimetics consisting of glycine, alanine, valine, leucine, phenylalanine, and cyclo-
hexylalanine with amino acids in the D-configuration were prepared. Partition coefficients for
the peptidemimetics were obtained in the octanol/water (log Poctanol/water), hydrocarbon/octanol
(∆ log P), and heptane/ethylene glycol (log Pheptane/glycol) systems in order to explore the
contributions of solute volume, or surface area, and hydrogen-bond potential to the permeability
of the solutes. Permeability coefficients were obtained in Caco-2 cell monolayers as a model of
the human intestinal mucosa. The results were interpreted in terms of a partition/diffusion
model for solute transport where membrane partitioning into the permeability-limiting
membrane microdomain is estimated from the solvent partition coefficients. Neither log
Poctanol/water nor ∆ log P alone correlated with cellular permeability for all the solutes. In contrast,
log Pheptane/glycol gave a qualitatively better correlation. With regard to solute properties, log
Poctanol/water is predominantly a measure of solute volume, or surface area, and hydrogen-bond
acceptor potential, while ∆ log P is principally a measure of hydrogen-bond donor strength.
Log Pheptane/glycol contains contributions from all these solute properties. The results demonstrate
that both hydrogen-bond potential and volume of the solutes contribute to permeability and
suggests that the nature of the permeability-limiting microenvironment within the cell depends
on the properties of a specific solute. Collectively, these findings support the conclusion that a
general model of permeability will require consideration of a number of different solute
structural properties.
In tr od u ction
In the absence of significant chemical degradation or
metabolism, the rate and extent of absorption of a drug
after oral administration is dependent upon the con-
centration of the drug in the intestinal lumen and
permeability of the intestinal mucosa to the drug.1-3 In
many cases, especially with high-dose drugs, the con-
centration effectively equates with the solubility of the
drug in the lumen. Both solubility and permeability are
properties dependent upon the structural characteristics
of the drug. Clearly, a better understanding of the
relationship of solute structure with these properties
will be useful in the design of drugs with improved
bioavailability.
The relationship of solute structure with permeability
depends on the mechanism of permeation. As shown in
Figure 1, permeability of a solute from the intestinal
lumen to the portal circulation necessitates movement
through a number of environments, each of which can
be considered a resistance.4-6 At a minimum, the solute
must cross the mucus coat and associated unstirred
water layer overlying the continuous epithelial cell
monolayer comprising the intestinal epithelium. Trans-
port across the epithelium can be by a transcellular
(through the cell) pathway and/or a paracellular (be-
tween adjacent epithelial cells) pathway.7 Furthermore,
the transcellular pathway involves crossing the apical
plasma membrane, movement through the cytosol, and
F igu r e 1. Schematic representation of barriers to and path-
ways of intestinal permeability: (A) intestinal lumen; (B)
mucus; (C) enterocytes; (D) basement membrane; (E) inter-
stitium; (F) capillary; (G) paracellular permeation; and (H)
transcellular permeation.
passage across the basolateral membrane. The para-
cellular pathway is an aqueous route restricted by the
presence of tight junctions between the cells.8,9 Finally,
to gain access to the mesenteric circulation draining the
epithelium, the solute must cross the basement mem-
brane underlying the epithelial cells and overlying the
lamina propria, through the interstitial space, and
across the capillary endothelium. Given that these
environments, or potential resistances, have different
properties, the influence of solute structure upon each
* Address correspondence to this author: Phone 616-833-1322; fax
616-833-2325; e-mail philip.s.burton@am.pnu.com.
10.1021/jm010253i CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/25/2001

3722 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22
Goodwin et al.
F igu r e 2. X-phenylalanine dipeptide and X-phenethylamide peptidomimetic series, where X ) Gly, D-Ala, D-Val, D-Leu, D-Phe,
and d-cyclohexylalanine.
Ta ble 1. Permeability Data for Compounds 1-12
resistance will be different, resulting overall in a highly
complicated modeling problem.
peff
peff,verapamil
com- AP f BL f
AP f BL f
log
c
pound BLa APa ratiob BLa
APa ratiob pmono ppara ptrans ptrans
c
c
To simplify this process for modeling purposes, a
common strategy has been to collapse all of these steps
into a single, homogeneous membrane transport pro-
cess. This assumes that, for a given series of solutes,
one of these steps is the overall permeability-limiting
process and the others can be ignored. In some cases,
the unstirred aqueous boundary layer resistance and
parallel transcellular and paracellular pathways are
considered explicitly in exploring structure-transport
relationships. In previous work, we have used such a
model to examine the role of solute structure in trans-
port of a series of model peptide mimetics across Caco-2
cell monolayers.10,11 Caco-2 cells are human colon ad-
enocarcinoma cells that form confluent monolayers in
culture once seeded onto filter membranes. These mono-
layers spontaneously differentiate into a polarized,
enterocytelike morphology which mimics that of the
small intestinal epithelium.12-16 Using this model we
found that the permeability of these peptide mimetics
was dependent upon the hydrogen-bonding potential of
the solutes. This was estimated from the total number
of hydrogen bonds possible or, experimentally, ∆ log P
(obtained from log Phydrocarbon/water - log Poctanol/water).
Permeability was significantly less correlated with
octanol-water partition coefficients, a commonly em-
ployed surrogate for predicting oral absorption.10,11,17 In
the present studies, these relationships were further
explored by designing two additional series of peptide
mimetics (Figure 2) in which hydrogen-bond potential
was held constant while octanol-water partition coef-
ficients varied in an incremental fashion.
1
2
0.17 0.24 1.4 0.14
0.24 0.38 1.6 0.25
0.12 0.9 0.14 0.60 0.00
0.25 1.0 0.25 0.56 0.00
0.78 0.9 0.90 0.48 0.42 6.38
1.42 0.9 1.56 0.44 1.12 5.95
2.64 0.9 2.95 0.37 2.58 5.59
9.02 1.0 10.7 0.36 10.3 4.99
20.8 0.9 38.4 0.77 37.6 4.42
23.5 0.9 46.5 0.71 45.8 4.34
33.0 0.9 97.0 0.60 96.4 4.02
47.7 1.0 343 0.56 342 3.47
59.5 1.1
3a 0.91 1.84 2.0 0.89
1.41 2.24 1.6 1.52
5a 2.75 5.19 1.9 2.80
4
6
7
8
7.20 14.5 2.0 8.93
21.9 20.6 0.9 22.6
24.6 22.9 0.9 25.2
9a 34.5 32.4 0.9 35.1
10
11a 52.0 56.8 1.1 55.1
12
42.6 47.2 1.1 47.4
58.3 58.4 1.0 55.5
55.6 1.0
a
AP f BL ) apical-to-basolateral permeability; BL f AP )
b
basolateral-to-apical permeability. Ratio of BL f AP to AP f
BL permeabilities. pmono ) monolayer permeability, ppara
c
)
paracellular permeability, ptrans ) transcellular permeability;
permeabilities are given in units of 10-6 centimeters per second.
Standard errors in permeabilities are e10%, n g 4, except for
compounds 7, 8, 9a , and 10, where the standard error is e30%.
The mass balances for all permeabilities determined are 100%
((10%).
Resu lts
The effective, monolayer, and transcellular perme-
abilities for the X-D-Phe and X-phenethylamide peptide
series are given in Table 1. The effective permeability
of a given compound is a complex function of aqueous
boundary layers and cell monolayer and filter perme-
abilities, as described in Figure 1. This is represented,
analogously to resistances in series, as
1
peff
1
1
1
1
)
+
+
+
(1)
ap
bl
pmono pfilter
pABL
pABL
where peff is the effective permeability coefficient for the

Determinants of Passive Membrane Permeability
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22 3723
Ta ble 2. Physicochemical Data for Compounds 1-12
ap
bl
solute, pABL and pABL are aqueous boundary layer
permeabilities for the apical and basolateral compart-
ments, respectively, pfilter is the polycarbonate filter
permeability, and pmono is the Caco-2 cell monolayer
permeability.18,19
com-
log
log
∆
log
pound MW tHBa dHBa aHBa Poctanol/water Pheptane/water log P Pheptane/glycol
1
2
277
291
6
6
6
6
6
6
4
4
4
4
4
4
3
3
3
3
3
3
2
2
2
2
2
2
3
3
3
3
3
3
2
2
2
2
2
2
0.30
0.06
0.66
1.24
1.44
2.40
0.48
0.76
1.59
2.03
2.32
3.13
6.30
6.26
5.33
4.61
4.15
3.41
3.73
3.53
2.64
2.14
2.00
0.90
6.00
6.20
5.99
5.85
5.59
5.81
4.21
4.29
4.23
4.17
4.32
4.03
6.17
5.83
5.79
5.43
5.34
5.03
5.00
4.40
3.77
3.69
3.70
3.41
3a 319
333
5a 367
In previous studies, a value of 55 × 10-6 cm/s has
been obtained for the collective permeability of the two
aqueous boundary layers and filter in the unstirred
situation (unpublished observation). Therefore, upon
rearrangement and substitution we obtain
4
6
7
8
373
220
234
9a 262
10
11a 310
12
276
1
pmono
)
(2)
316
1
peff
1
-
a
55 × 10-6
tHB ) total number of solute hydrogen bonds possible, dHB
) number of donor hydrogen bonds, aHB ) number of acceptor
hydrogen bonds. Standard error for log Poctanol/water e15%, for log
b
However, pmono is itself comprised of transcellular and
paracellular permeabilities in parallel, i.e.
Pheptane/water e10%, for ∆ log P e15%, and for log Pheptane/ethylene glycol
e5%; n g 2 for all solvent partition experiments.
pmono ) ppara + pcell
(3)
In general, pcell represents the overall contributions of
passive and active transport processes available to a
solute. Previous studies with related small-peptide
mimetics have demonstrated the presence of a concen-
tration-dependent polarized efflux pathway in Caco-2
cells in addition to the concentration-independent, pas-
sive diffusion mechanism.20,21 As shown in Table 1, the
BL f AP effective permeabilities for the N-Ac-X-D-Phe-
NHMe peptides were greater than AP f BL perme-
abilities. These differences were eliminated in the
presence of 300 µM verapamil in both the donor and
receiver chambers, suggesting the involvement of P-
glycoprotein in the polarized transport mechanism,
similar to the case with a related phenylalanine oligo-
mer series.21 In contrast, the N-Ac-X-phenethylamide
peptides showed no polarity in transport or verapamil
influences. For the purposes of the present study, we
wished to correlate physicochemical properties of the
solutes with the passive-diffusional transcellular path-
way. Therefore, the effective permeabilities in the
presence of verapamil (peff,verapamil), in the case of N-Ac-
X-D-Phe-NHMe peptides, were employed for further
analysis.
Finally, the paracellular permeabilities for the pep-
tides were estimated from a previously established
relationship in which paracellular flux is modeled in
terms of molecular size-restricted diffusion within an
electrostatic field of force.18,22 Upon substitution of the
values into eq 3, pcell, the transcellular passive perme-
abilities of the peptides, were obtained and are included
in Table 1. After application of these corrections Ac-Gly-
D-Phe-NHMe and Ac-D-Ala-D-Phe-NHMe are excluded
due to their primarily paracellular transport, and Ac-
D-Phe-phenethylamide and Ac-D-Cha-phenethylamide
(Cha ) cyclohexylalanine) are excluded due to their
exclusively unstirred water layer-limited diffusion.
In Table 2 are summarized the results of the partition
coefficient determinations for the peptide mimetics.
Note that within each peptide series the hydrogen-
bonding potential, as represented by ∆ log P (log
Pn-octanol/water - log Pheptane/water), does not vary more than
10%. The average value of ∆ log P for the X-D-Phe series,
with a hydrogen-bond number of 6, is 5.91, whereas for
the X-phenethylamide series, with a hydrogen-bond
number of 4, the average ∆ log P value is 4.21.
F igu r e 3. Transcellular permeability for compounds 1-12 as
a function of (top panel) solute lipophilicity and (bottom panel)
solute hydrogen-bond potential; 2 ) X-phenylalanine series,
1 ) X-phenethylamide series.
Furthermore, the estimated sizes of the peptides, as
approximated by their molecular weights, do not vary
beyond 10% within each series.
The relationship of transcellular permeability coef-
ficients with octanol-water partition coefficients is
shown in Figure 3 (top). Two distinctly different cor-
relations are observed, with the phenethylamide series
exhibiting greater cellular permeation than the Phe-

3724 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22
Goodwin et al.
NHMe analogues. Within each series, an incremental
increase in permeability is seen with increasing parti-
tion coefficient. The corresponding relationship with ∆
log P is shown in Figure 3 (bottom). Again the solutes
form two distinct clusters, with the more permeable
solutes having the smaller ∆ log P. Within each group,
differences independent of ∆ log P are seen in the
cellular permeabilities of the individual solutes.
This simple model was elaborated in an interesting
series of molecular dynamics simulations exploring
solute permeation across phospholipid membranes.
Marrink and Berendsen27 proposed a four-compartment
model consisting of two interfacial regions differing in
hydrogen-bonding characteristics and two interior do-
mains with different free volume distributions. This
model was used to define the local resistances to
permeation for several solutes with differing hydrogen-
bond capacity. For solutes with significant hydrogen-
bond potential, association with the interfacial region
was favorable but movement into the apolar interior
regions presented the most significant barrier. Con-
versely, solutes with little or no hydrogen-bonding
functionality were excluded from the interfacial region
but were freely permeable within the membrane inte-
rior.27
Discu ssion
One of the most successful conceptual constructs
relating drug physicochemical properties with drug
absorption is the partition-diffusion model. This posits
that the rate of drug transport across an absorbing
surface is proportional to the solubility, i.e., partitioning,
into the membrane and its diffusion coefficient in the
membrane.23 Implicit in this model is the assumption
of homogeneity in the composition of the permeability-
limiting membrane. Since membrane partition coef-
ficients are generally difficult to obtain experimentally,
surrogate organic solvent partition coefficients are usu-
ally employed instead. The most frequently utilized
solvent is octanol. Recent work has shown that the
solute molecular features important in octanol parti-
tioning are volume and hydrogen-bond basicity.24 By
extrapolation, then, these are assumed to be critical
features in drug absorption. For many classes of solutes
this seems to be true, and good correlations have been
established between drug absorption rates and octanol-
buffer partition coefficients.
Our results are consistent with these representations
of solute membrane microenvironment interactions. We
had previously shown that, for a series of phenylalanine
oligomers with essentially constant octanol/water par-
titioning, Caco-2 cell permeability was directly related
to the hydrogen-bonding properties of the molecules.17,25
This supported a model in which such molecules have
high affinity for the membrane interfacial region but
experience resistance in the interface to interior transfer
in which the interfacial hydrogen bonds need to be
broken.25 The energy required for this process could be
approximated from the ∆ log P parameter.
These previous observations have been confirmed and
extended in the present study. Comparison of the
permeability data for the two series shown in Figure 3
(top) clearly show that cellular permeability is depend-
ent upon hydrogen-bond potential. Compounds 5a and
9a , for example, have comparable log Poctanol/water values
(1.44 and 1.59, respectively) with a 37-fold difference
in permeability. This difference in permeability is
consistent with the decrease in the ∆ log P of ap-
proximately 1.7 upon removing the terminal CONHCH3,
resulting in two fewer hydrogen-bonding sites in 9a .
Although this structural change is also accompanied by
a decrease in molecular size, which could have an
impact on the diffusion step, these size changes are
relatively modest and are not expected to have much of
an impact on the transport of these solutes. Thus these
changes are attributed to differences in the free energy
of partitioning. Further, given that log Poctanol/water is
primarily a measure of volume and hydrogen-bond
acceptor strength, the small difference in partition
coefficients for 5a and 9a suggest that hydrogen-bond
basicity contributes little to the observed permeability
differences. This indicates that removal of CONHCH3
principally reduces hydrogen-bond donor potential,28
which is the major factor differentiating permeability
for these two solutes.
However, for highly functionalized solutes, such as
peptides, this simple relationship does not appear to
hold. In this case, partitioning systems that are more
dependent upon solute hydrogen-bond donor and accep-
tor strength, with little or no volume contribution,
correlate better with intestinal cellular permeability.25
Such differences in solute structure-permeability cor-
relations challenge the original assumption of mem-
brane structural homogeneity and suggest the different
solutes are experiencing different diffusion-limiting
microdomains within the absorbing epithelium. These
observations are consistent with earlier studies by
Diamond and Katz exploring the validity of the parti-
tion/diffusion model for describing the mechanism of
solute permeability across phospholipid vesicles.26 In
this case, partition coefficients were determined in
dimyristoylphosphatidylcholine (DMPC) vesicles and
used to model permeation into those vesicles. In com-
paring the resulting permeability coefficients with those
calculated by assuming the partition diffusion model for
a homogeneous membrane, the experimental values
were only a small fraction of those expected. The
conclusion of these studies was that the rate-limiting
barrier to permeation of the solutes was not the region
where most of the solute was partitioned but rather a
region where the solute was excluded.26 The simplest
inhomogeneous, microdomain model that was consistent
with the known structure of phospholipid vesicles
consists of a polar, interfacial headgroup region separat-
ing the aqueous environment from the much more
apolar membrane interior environment where the phos-
pholipid acyl hydrocarbon chains are located. These
regions differ in their ability to accommodate hydrogen-
bonding groups and/or nonpolar domains on a solute
molecule.
As shown in Figure 3, within a series of constant ∆
log P, there is an incremental increase in permeability
with octanol-water partition coefficient. Taken alto-
gether, these results further support the presence of
different solute-membrane interactions during the
permeation process, all of which can influence the
overall transport rate. These results can also put into
perspective the relative influences of hydrogen-bonding
groups and apolar hydrocarbon substituents on the

Determinants of Passive Membrane Permeability
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22 3725
F igu r e 5. Transcellular permeability as a combined function
of solute lipophilicity and hydrogen-bond potential; 2 )
X-phenylalanine series, 1 ) X-phenethylamide series. The
F igu r e 4. Transcellular permeability versus heptane/ethylene
glycol partitioning for compounds 1-12; 2 ) X-phenylalanine
series, 1 ) X-phenethylamide series. The relationship is
described by log pcell ) 0.57 log Pheptane/ethyleneglycol - 2.03, with
r2 ) 0.63 and s ) 0.49. Although this supports that both solute
volume and hydrogen-bond potential are fundamental deter-
minants of cellular permeability, it is clear that this solvent-
partition measure does not accurately represent the interplay
of these properties influencing permeability.
relationship is described by log pcell ) 0.61 log Poctanol/water
-
1.13 ∆ log P - 0.04, with r2 ) 0.99 and s ) 0.13. These results
suggest that a more specific relationship of solute volume and
hydrogen-bond potential, not represented by log Pheptane/ethyleneglycol
,
is required in order to describe cellular permeability for these
solutes.
general permeability model for the peptide mimetics
examined here will need to incorporate all the solute
structural properties important in both log Poctanol/water
and ∆ log P.
permeation process. In the N-Ac-X-D-Phe-NHMe series,
as X progresses from hydrogen to cyclohexyl, log
Poctanol/water increases from -0.3 to 2.4, almost 3 log units,
while cellular permeability increases from about 0 to
10.3 × 10-6 cm/s. In contrast, removing hydrogen-bond
donor sites, as already discussed in comparing 5a and
9a , results in an increase in permeability from 2.6 ×
10-6 to 96.4 × 10-6 cm/s. Clearly, removal of hydrogen-
bonding functionality is much more effective in improv-
ing permeability than introduction of additional hydro-
carbon. Such insights may be useful in guiding the
design of more bioavailable drug development candi-
dates.
Finally, given that structurally different solutes can
experience different permeation limiting interactions
within the cell membrane, what is the possibility of
establishing a single in vitro model that will predict
permeability for a diverse series of solutes? In Figure 4
is shown the relationship of transcellular permeability
with log Pheptane/glycol for the eight peptide mimetics as a
group. Solvatochromic analysis of the log Pheptane/glycol
system showed that this partition coefficient is reflective
of solute hydrogen-bond donor and acceptor potential,
similar to the case with ∆ log P. However, it contains a
In support of this proposition is the finding that a
weighted, linear combination of log Poctanol/water and ∆
log P gives a reasonably good model for permeability of
these solutes (Figure 5). Very similar results were
reported recently in modeling efforts with the present
data set in the context of a polar surface area (PSA)
model.31 In that report, PSA, which is a putative
measure of hydrogen-bond potential for a solute, did not
correlate well with the permeability of these solutes,
while it had worked well for other, more homologous
solutes.32 However, a reasonably good correlation was
obtained by inclusion of a nonpolar surface area term
to the model. Although the correlation reported utilized
effective permeability coefficients, which are mechanis-
tically complicated as shown in Figure 1, and not
transcellular permeability coefficients, the qualitative
similarity to our results further supports the conclusion
that a general model of permeability will require
consideration of a number of solute structural proper-
ties.
significant volume contribution, more similar to the
25,29
Exp er im en ta l Section
situation with log Poctanol/water
.
Thus it is essentially
a hybrid of the two systems. Consistent with these broad
property contributions to log Pheptane/glycol, a better cor-
relation with permeability is obtained than with either
log Poctanol/water or ∆ log P alone, systems that are more
specific with regard to solute structure. These results
contrast with a recent comparison of ∆ log P and log
Pheptane/glycol as predictors of cell permeation.30 In that
case, ∆ log P was found to be more highly correlated
with permeation of solutes into algae cells. A possible
explanation for this apparent discrepancy may be due
to differences in the cell types used in the two studies.
Alternatively, the differences in composition of the
solutes in the two studies may be sufficient to distin-
guish their permeability-limiting cell interactions as
discussed herein. The present results suggest that a
Ma ter ia ls a n d Meth od s. Caco-2 cells were obtained from
the American Type Culture Collections (ATCC), Rockville, MD,
at passage 19. All cell culture reagents were from Gibco-BRL,
Gaithersburg, MD. Transwells were from Corning Costar,
Cambridge, MA. tert-Butoxycarbonyl- (BOC-) protected amino
acids and verapamil hydrochloride were purchased from
Sigma, St. Louis, MO. [14C]Acetic anhydride (105-120 µCi/
mmol) was purchased from Amersham. Melting points were
obtained on a Hoover Unimelt apparatus and were uncor-
rected. 1H and 13C NMR spectra were obtained on a Varian
400 spectrometer. Mass spectra and elemental analysis were
performed by the Structural and Medicinal Chemistry Group.
Reversed-phase high-pressure liquid chromatography (RP-
HPLC) was performed on a Hypersil BDS/C18 analytical (15
cm × 4.6 mm) column, with a Waters 600 gradient solvent
delivery system, a 712 WISP autosampler, and 490 multi-
wavelength detector operating at 206 and 254 nm, interfaced

3726 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22
Goodwin et al.
with a 486-based PC running Waters Millenium software for
controlling instrumentation and data collection.
port studies has been described previously.14 Briefly, confluent
monolayers of between 14 and 21 days, and between passage
25 and 40, were used for transport studies. Prior to use the
medium was siphoned off the monolayers, which were then
rinsed 3× with phosphate-buffered saline (PBS) containing 15
mM HEPES (pH 7.2) and 0.1% (m/v) D-glucose, and then
incubated in PBS/HEPES, 1.5 mL in the apical compartment
and 2.5 mL in the basolateral compartment, at 37 °C. For
apical-to-basolateral transport experiments, the buffer was
removed from the apical (donor) compartment and replaced
with 1.5 mL of PBS/HEPES containing [14C]-N-acetylated
peptide at ca. 250 000 dpm/mL, supplemented with unlabeled
peptide to a total peptide concentration of ca. 20 µM. At a
specified time interval the monolayer cup was transferred to
an adjacent well (receiver compartment) containing fresh
buffer. For the basolateral-to-apical transport experiments, the
well (donor compartment) was charged with 2.5 mL of the
peptide solution and the insert containing the monolayer and
buffer (receiver compartment) was then placed in the well. At
specified time intervals the buffer in the receiver compartment
was removed and replaced with fresh buffer. Time intervals
were chosen to maintain sink conditions, such that no greater
than 5% of the peptide permeated into the receiver compart-
ment per interval. All experiments were run at 37 °C in a 5%
CO2 atmosphere with 95% relative humidity. Cumulative
transport of peptide was obtained by summing the percentage
amounts of radionuclide appearing in the receiver compart-
ment after each sampling interval.
P a r tition Coefficien t Deter m in a tion s. Solvent partition
coefficients (n-octanol/water, n-heptane/water, and heptane/
ethylene glycol) were determined as described in detail previ-
ously.17,25 Briefly, a stock solution of the peptide was prepared
in the polar phase (water or ethylene glycol) at ca. 5-20 mM
(ethylene glycol) or 20 µM (water). Upon dissolution, the stock
solution was filtered to remove any undissolved peptide, and
aliquots (in triplicate) were placed in clean, oven-dried screw-
cap glass vials with Teflon septa. For ethylene glycol/heptane
partition experiments, all manipulations of the ethylene glycol
stock were done in a low-humidity (<25% relative humidity
at 25 °C) room; the water content of the ethylene glycol phase
was determined by Karl Fisher titration to be less than 0.2
wt %. HPLC-grade heptane was used in a phase-volume ratio
of 100:1 relative to the ethylene glycol solution of peptide. The
phases were mixed initially in the sealed screw-cap vials by
sonication for 10-15 min and then placed on a wrist-action
shaker at 25 °C for at least 2 days to ensure that equilibrium
was obtained. The heptane phase (25 mL) was removed with
a clean dry glass volumetric pipet and then dried down in
vacuo, and the residual peptide was rinsed into an amber
silanized glass vial with ca. 1 mL of MeOH. MeOH was
removed under a nitrogen stream with warming and the
contents were transferred to an HPLC vial with 250 µL of 10%
MeOH/water, resulting in a concentration factor of 100. The
ethylene glycol phase was diluted with 10% MeOH/water to
30 mL, for a dilution factor of 100. Comparison of the peptide
concentrations in the two phases by RP-HPLC, accounting for
the concentration and dilution factors, provides the heptane/
ethylene glycol partition coefficient as the log of the ratio of
the concentrations. For octanol/water partition experiments,
the peptide stock solutions were prepared in octanol-saturated
water and filtered, and aliquots were placed in screw-cap glass
vials in triplicate. Water-saturated octanol was added in a
phase-volume ratio of 1:20, and the phases were initially mixed
by sonication, followed by agitation on an orbital shaker (200
rpm) for not less than 2 days at room temperature to ensure
equilibration. The phases were separated by centrifugation and
300 µL of the aqueous phase was transferred to an HPLC vial
for comparison with the initial aqueous peptide stock. The
octanol/water partition coefficients were obtained from the log
of the ratio of peptide concentrations in the octanol and water
phases. The heptane/water partition coefficients were deter-
mined in a similar fashion, with a heptane/water phase-volume
ratio of 50:1. After 2 days the heptane phase was removed and
concentrated and the peptide was taken up into 10% MeOH/
water for a concentration factor of 100. Aliquots of the aqueous
phase were analyzed, without dilution, by HPLC. The heptane/
water partition coefficient was obtained as the log of the ratio
of peptide concentrations in the heptane and aqueous phases.
For all peptides calibration curves were established to deter-
mine the region of linear response of the detector, at 206 nm,
to the peptide concentrations.
Qu a n tita tion . Concentrations of the labeled peptides, in
disintegrations per minute (dpm) per milliliter, in the receiver
solutions were determined by liquid scintillation counting in
a Beckman LS 3801 scintillation counter. Effective perme-
ability coefficients (in centimeters per second) were calculated
from these concentrations and the following relationship:
VR
peff
)
dC/dt
(4)
[
]
AC0
where VR is the volume of the receiver compartment, A is the
surface area of the monolayer, C0 is the initial donor concen-
tration, and dC/dt is the slope of the plot of the cumulative
receiver concentration with time.
P ep tid e Syn th esis An d P u r ifica tion . All peptides used
in this study were prepared by solution-phase methods with
diisopropylcarbodiimide (DIC) and hydroxybenzotriazole (HOBT)
as the coupling reagents.33,34
P r ep a r a tion of N-BOC-D-p h en yla la n in e N-Meth yla -
m id e. An oven-dried round-bottom flask was charged with
N-BOC-D-phenylalanine p-nitrophenyl ester (5.00 g, 12.9
mmol), to which was added MeCl2 (20 mL) under dry nitrogen
with stirring. To this solution methylamine (2.0 M in THF,
32 mL, 65.0 mmol, 5.0 equiv) was added via syringe with
continued stirring at 0 °C, and immediate generation of a
bright yellow color corresponding to the p-nitrophenylate anion
was observed. The reaction was allowed to warm to room
temperature upon removal from the ice bath and monitored
by TLC. After 3 h, precipitated methylammonium p-nitrophe-
nylate salt was removed by gravity filtration and the solvents
were removed in vacuo. The resulting pale yellow solid was
triturated with Et2O/petroleum ether (1:2), and the white
microcrystalline solid was collected by vacuum filtration; 3.43
g, 12.3 mmol, 95.3% yield. This material (2.78 g, 10.0 mmol)
was then taken up into 8 mL of MeCl2, followed by addition of
7.7 mL of trifluoroacetic acid (TFA, 100 mmol, 10 equiv) with
stirring at room temperature. Solvents were removed in vacuo
after 1 h and the residue was coevaporated twice with MeCl2,
after which the residue was triturated with Et2O, yielding 2.48
g of the TFA salt of d-phenylalanine N-methylamide as a white
microcrystalline solid, 8.49 mmol, 85% yield. A further recrys-
tallization of this salt from MeCN provided white microcrys-
talline needles and removed any residual excess TFA. 1H NMR
(400.1 MHz, DMSO-d6) δ 2.58 (3 H, d, J ) 4.4 Hz, NH-CH3),
2.98 (2 H, m, â-CH2 (Phe)), 3.90 (1 H, dd, J ) 7.2 Hz, R-CH
(Phe)), 7.19-7.34 (5 H, m, Ar-H (Phe)), 8.21 (3 H, broad s, CF3-
Cell Cu ltu r e. Caco-2 cells were cultured as previously
described.14 Briefly, monolayers used in this study were grown
at 37 °C with 5% CO2, in T-150 flasks with Dulbecco’s modified
Eagle medium (DMEM), pH 7.2, containing 4.5 g/L D-glucose
and 584 mg/L L-glutamine, supplemented with 1% (v/v)
nonessential amino acids, 1% (v/v) sodium pyruvate, and 10%
(v/v) fetal bovine serum. After 1 week the confluent cell
monolayers were washed 3× with Hank’s balanced salt solu-
tion, without calcium or magnesium (HBSS-CMF), and then
removed from the flask by incubation with 0.25% (v/v) trypsin
in 1 mM EDTA for 10 min at 37 °C. The trypsinized cells were
diluted with DMEM/FBS to a concentration of 63 000 cells/
cm2, as determined with a hemocytometer. The Caco-2 cells
were seeded into 24-mm-diameter Transwell inserts, with 0.4
µm pore polycarbonate filters, and maintained at 37 °C, 5%
CO2, and 95% relative humidity for at least 14 days prior to
use in transport studies. After 7 days the medium was
exchanged every second day.
Tr a n sp or t Stu d ies. The procedure for conducting trans-

Determinants of Passive Membrane Permeability
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22 3727
CO2- H3N+(Phe)), 8.31 (1 H, d, J ) 4.4 Hz, NH-CH3); 13C NMR
(100.6 MHz, DMSO-d6) δ; high-resolution FAB MS m/z 179.1185
(M+ + 1). Anal. Calc. for C12H15N2O3F3: C, 49.32; H, 5.17; N,
9.59. Found: C, 49.23; H, 5.06; N, 9.68.
peptide (relative to total radioactivity) indicated by RP-HPLC
provided the volume of crude product solution containing a
given quantity (in disintegrations per minute) of labeled
peptide. Labeled peptide was purified as needed by preparative
silica gel thin-layer chromatography (TLC), using unlabeled
peptide to locate the pure peptide band by UV, and recovered
from the silica gel by extraction with MeOH/MeCl2 (1:1) and
filtration.
Rep r esen ta tive P r ep a r a tion of N-Ac-D-Va lin e-D-p h en -
yla la n in e N-Meth yla m id e (3). An oven-dried round-bottom
flask was charged with N-BOC-D-valine (0.22 g, 1.0 mmol),
followed by 0.14 g of dry 1-hydroxybenzotriazole (HOBT, 1.0
mmol, 1.0 equiv). (Note: 1-hydroxybenzotriazole is com-
mercially available as a hydrate. Dissolution of the hydrate
in toluene removes the water as the azeotrope in vacuo, as
determined by a Karl Fisher test.) The amino acid and HOBT
were suspended in 5 mL of MeCl2, followed by addition of
diisopropylcarbodiimide (DIC, 0.16 mL, 1.0 mmol, 1.0 equiv)
with stirring under dry nitrogen at 0 °C. After 1 h the TFA
salt of D-phenylalanine N-methylamide (0.29 g, 1.0 mmol, 1.0
equiv) was added along with 0.14 mL of triethylamine (1.0
mmol, 1.0 equiv), and the reaction was allowed to warm to
room-temperature overnight. After 15 h solvents were removed
in vacuo and the residue was purified by silica gel flash
chromatography (MeCl2/MeOH, 96:4), yielding a white amor-
phous solid, 0.32 g, 0.84 mmol, 84% yield. The N-BOC-
protected dipeptide (0.32 g, 0.84 mmol) was dissolved in 4 mL
of MeCl2, and to this solution was added TFA (0.65 mL, 8.4
mmol, 10 equiv) with stirring. After 1 h the solvents were
removed in vacuo and the residue was triturated with Et2O,
yielding a white granular solid (0.29 g, 0.73 mmol, 87%)
recovered by vacuum filtration. The TFA salt of the dipeptide
was then resuspended in 4 mL of MeCl2, to which was added
acetic anhydride (0.69 mL, 7.3 mmol, 10 equiv) and triethy-
lamine (1.0 mL, 7.3 mmol, 10 equiv) with stirring. After ca. 1
h a precipitate began to form and the solution began to gel.
After 15 h the reaction was quenched with ca. 1 mL of MeOH,
solvents were removed in vacuo, and the residue was purified
by flash chromatography (95:5 MeCl2/MeOH), providing a
white solid with some acetic acid still detected by scent. The
N-acetylated dipeptide was recrystallized from MeOH/water,
yielding a white granular solid that was dried in a vacuum
desiccator.
An a lytica l Da ta . N-Acetylglycyl-D-p h en yla la n in e N-
Meth yla m id e (1). Mp ) 195.0-195.5 °C; RP-HPLC retention
time (10% MeOH/H2O:MeCN, 85:15) ) 3.0 min; 1H NMR
(400.1 MHz, CD3OD) δ 1.99 (3 H, s, Ac-CH3), 2.70 (3 H, s, NH-
CH3), 2.91 (1 H, dd, J ) 13.6, 8.4 Hz, â-CH2 (Phe)), 3.15 (1 H,
dd, J ) 13.6, 8.0 Hz, â-CH2 (Phe)), 3.78 (2 H, m, J ) 16.4 Hz,
CH2 (Gly)), 4.57 (1 H, dd, J ) 8.0 Hz, R-CH (Phe)), 7.22-7.29
(5 H, m, Ar-H (Phe)); 13C NMR (100.6 MHz, CD3OD) δ 21.5
(Ac-CH3), 25.3 (NH-CH3), 37.8 (â-CH2 (Phe)), 42.6 (CH2 (Gly)),
55.1 (R-CH (Phe)), 126.8, 128.5, 129.2, 156.3 (Ar-C (Phe)), 170.3
(Ac-CO), 171.8 (CONH (Gly)), 172.8 (CONH (Phe)); high-
resolution FAB MS m/z 278.1511 (M+ + 1). Anal. Calcd for
C
14H19N3O3: C, 60.65; H, 6.86; N, 15.16. Found: C, 60.66; H,
6.81; N, 15.09.
N-Acetyl-D-a la n yl-D-p h en yla la n in e N-Meth yla m id e (2).
Mp ) 246.0-247.0 °C; RP-HPLC retention time (10% MeOH/
1
H2O:MeCN, 85:15) ) 3.5 min; H NMR (400.1 MHz, DMSO-
d6/10% TFA-d1) δ 1.00 (3 H, d, J ) 6.8 Hz, â-CH3 (Ala)), 1.81
(3 H, s, Ac-CH3), 2.55 (3 H, s, NH-CH3), 2.80 (1 H, dd, J )
13.6, 9.6 Hz, â-CH2 (Phe)), 3.00 (1 H, dd, J ) 13.6, 5.2 Hz,
â-CH2 (Phe)), 4.14 (1 H, q, J ) 6.8 Hz, R-CH (Ala)), 4.36 (1 H,
dd, J ) 5.2, 9.6 Hz, R-CH (Phe)), 7.15-7.24 (5 H, m, Ar-H
(Phe)); 13C NMR (100.6 MHz, DMSO-d6/10% TFA-d1) δ 18.3
(â-CH3 (Ala)), 23.1 (Ac-CH3), 26.1 (NH-CH3), 38.2 (â-CH2
(Phe)), 49.2 (R-CH (Ala)), 54.6 (R-CH (Phe)), 127.0, 128.8, 129.9,
138.7 (Ar-C (Phe)), 170.3 (Ac-CO), 171.8 (CONH (Ala)), 172.8
(CONH (Phe)); high-resolution FAB MS m/z 292.1659 (M+
1). Anal. Calcd for C15H21N3O3: C, 61.86; H, 7.22; N, 14.43.
Found: C, 61.82; H, 7.32; N, 13.86.
+
N-Acetyl-D-va lyl-D-p h en yla la n in e N-Meth yla m id e (3a ).
Mp ) 274.5-275.0 °C; RP-HPLC retention time (10% MeOH/
H2O:MeCN, 85:15) ) 4.2 min; 1H NMR (400.1 MHz, DMF-d7)
δ 0.96 (6 H, 2d, J ) 6.8 Hz, γ-CH3 (Val)), 2.18 (1 H, m, J ) 6.8
Hz, â-CH (Val)), 2.13 (3 H, s, Ac-CH3), 2.83 (3 H, d, J ) 4.8
Hz, NH-CH3), 3.10 (1 H, dd, J ) 14.0, 9.6 Hz, â-CH2 (Phe)),
3.31 (1 H, dd, J ) 14.0, 5.2 Hz, â-CH2 (Phe)), 4.29 (1 H, dd, J
) 7.6 Hz, R-CH (Val)), 4.78 (1 H, ddd, J ) 5.2, 8.4, 9.6 Hz,
R-CH (Phe)), 7.40 (5 H, m, Ar-H (Phe)), 7.95 (1 H, m, NH-
CH3), 8.21 (1 H, d, J ) 8.2 Hz, CONH (Val)), 8.26 (1 H, d, J )
8.4 Hz, CONH (Phe)); 13C NMR (100.6 MHz, DMF-d7) δ 18.1
(γ-CH3 (Val)), 19.4 (â-CH(Val)), 22.7 (Ac-CH3), 25.9 (NH-CH3),
38.2 (â-CH2 (Phe)), 55.0 (R-CH (Val)), 59.8 (R-CH (Phe)), 126.8,
128.7, 129.8, 138.9 (Ar-C (Phe)), 165.7 (Ac-CO), 171.8 (CONH
(Val)), 172.1 (CONH (Phe)); high-resolution FAB MS m/z
320.1979 (M+ + 1). Anal. Calcd for C17H25N3O3: C, 63.93; H,
7.89; N, 13.16. Found: C, 64.03; H, 7.99; N, 12.89.
The stereomeric purity of the peptides was assessed by
comparison of retention times on a chiral HPLC column for
the diastereomer pairs N-Ac-D-Val/L-Val-D-Phe-NHMe and
N-Ac-D-Phe/L-Phe-D-Phe-NHMe and for the enantiomeric pairs
N-Ac-D-Val/L-Val-phenethylamide and N-Ac-D-Phe/L-Phe-phen-
ethylamide.
Syn th esis a n d P u r ifica tion of 14C-La beled P ep tid es.
The N-BOC-protected peptide (50 µmol) to be 14C-labeled was
dissolved in MeCl2 (0.5 mL), and trifluoroacetic acid (0.5 mL)
was added with stirring at room temperature under an
atmosphere of dry nitrogen. After 30 min the solvents were
removed in vacuo, and the residue was coevaporated with
MeCl2 followed by coevaporation with Et2O. The free amine
was obtained by aqueous workup with NaHCO3 and saturated
brine and extracted with CHCl3/MeOH (7:3). After the organic
layer was dried over anhydrous Na2SO4 and filtered, the
solvents were removed in vacuo and the residue was coevapo-
rated twice with MeCl2 to remove residual MeOH. The free
amine was dissolved in MeCl2 (2 mL) followed by addition of
triethylamine (70 µL, 10 equiv) and bis(2-oxo-3-oxazolidinyl)-
phosphinic chloride (BOP-Cl, 2 mg, 0.2 equiv) under dry
nitrogen. [1-14C]Acetic anhydride (5% w/w in toluene, 500 µCi)
was diluted with 200 µL of MeCl2 under nitrogen, and ca. 250
µCi of the acetic anhydride was transferred to the reaction
(the remaining [1-14C]acetic anhydride was used in a separate
labeling reaction), which was allowed to proceed overnight at
room temperature. Aliquots (ca. 2-5 µL) of the reaction were
diluted into 1 mL of 10% aqueous MeOH/MeCN for reversed-
phase HPLC analysis with radiolytic detection. In some cases
the [14C]acetylated peptide precipitated from solution; redis-
solution was achieved with addition of 0.5 mL of MeOH. The
resultant labeled peptides were stored as the crude product
solutions at -40 °C in sealed screw-cap vials. Determination
of the specific activity of the solution [in disintegrations per
minute (dpm) per microliter] combined with the yield of labeled
N-Acetyl-D-leu cyl-D-p h en yla la n in e N-Meth yla m id e (4).
Mp ) 223.0-224.0 °C; RP-HPLC retention time (10% MeOH/
1
H2O:MeCN, 75:25) ) 3.1 min; H NMR (400.1 MHz, DMSO-
d6/10% TFA-d1) δ 0.76, 0.81 (6 H, 2d, J ) 6.4 Hz, δ-CH3 (Leu)),
1.30 (2 H, m, â-CH2 (Leu)), 1.50 (1 H, m, γ-CH (Leu)), 1.81 (3
H, s, Ac-CH3), 2.54 (3 H, s, NH-CH3), 2.80 (1 H, dd, J ) 13.6,
9.2 Hz, â-CH2 (Phe)), 2.97 (1 H, dd, J ) 13.6, 5.2 Hz, â-CH2
(Phe)), 4.16 (1 H, dd, J ) 6.0, 9.6 Hz, R-CH (Leu)), 4.39 (1 H,
dd, J ) 5.2, 9.2 Hz, R-CH (Phe)), 7.14-7.23 (5 H, m, Ar-H
(Phe)); 13C NMR (100.6 MHz, DMSO-d6/10% TFA-d1) δ 22.2,
23.1 (δ-CH3 (Leu)), 23.5 (γ-CH(Leu)), 24.8 (Ac-CH3), 26.1 (NH-
CH3), 38.2 (â-CH2 (Phe)), 41.1 (â-CH2 (Leu)), 52.1 (R-CH (Leu)),
54.5 (R-CH (Phe)), 126.9, 128.7, 129.9, 138.6 (Ar-C (Phe)), 170.4
(Ac-CO), 171.8 (CONH (Leu)), 172.5 (CONH (Phe)); high-
resolution FAB MS m/z 334.2129 (M+ + 1). Anal. Calcd for
C
18H27N3O3: C, 64.86; H, 8.11; N, 12.61. Found: C, 64.30; H,
8.22; N, 12.58.
N-Acetyl-D-p h en yla la n yl-D-p h en yla la n in e N-Meth yla -
m id e (5a ). Mp ) 264.0-265.0 °C; RP-HPLC retention time
(10% MeOH/H2O:MeCN, 75:25) ) 3.5 min; 1H NMR (400.1

3728 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22
Goodwin et al.
MHz, DMSO-d6/30% TFA-d1) δ 1.85 (3 H, s, Ac-CH3), 2.62 (3
H, s, NH-CH3), 2.81 (1 H, dd, J ) 14.0, 9.6 Hz, â-CH2 (Phe)),
2.90 (1 H, dd, J ) 13.6, 8.8 Hz, â-CH2 (Phe)), 3.02 (1 H, dd, J
) 14.0, 5.2 Hz, â-CH2 (Phe)), 3.09 (1 H, dd, J ) 13.6, 5.2 Hz,
â-CH2 (Phe)), 4.52 (1 H, dd, J ) 5.2, 8.4 Hz, R-CH (Phe)), 4.57
(1 H, dd, J ) 5.2, 9.2 Hz, R-CH (Phe)), 7.08-7.17 (5 H, m,
Ar-H (Phe)); 13C NMR (100.6 MHz, DMSO-d6/30% TFA-d1) δ
22.3 (Ac-CH3), 25.7 (NH-CH3), 37.8 (â-CH2 (Phe)), 38.2 (â-CH2
(Phe)), 54.7 (R-CH (Phe)), 54.8 (R-CH (Phe)), 126.7, 126.8,
128.5, 128.6, 129.7, 129.8, 138.3, 138.4 (Ar-C (Phe)), 170.9 (Ac-
CO), 171.5 (CONH (Phe)), 171.8 (CONH (Phe)); high-resolution
FAB MS m/z 368.1973 (M+ + 1). Anal. Calcd for C21H25N3O3:
C, 68.66; H, 6.81; N, 11.44. Found: C, 68.14; H, 6.99; N, 11.32.
N-Acet yl-D-cycloh exyla la n yl-D-p h en yla la n in e N-Me-
t h yla m id e (6). Mp ) 239.0-240.0 °C; RP-HPLC retention
time (10% MeOH/H2O:MeCN, 65:35) ) 3.4 min; 1H NMR
(400.1 MHz, DMSO-d6/30% TFA-d1) δ 0.68, 0.94, 1.26 (8 H,
m, CH2 (Cha)), 1.43 (5 H, m, γ-CH, CH2 (Cha)), 1.84 (3 H, s,
Ac-CH3), 2.52 (3 H, s, NH-CH3), 2.76 (1 H, dd, J ) 14.0, 8.8
Hz, â-CH2 (Phe)), 2.95 (1 H, dd, J ) 14.0, 5.6 Hz, â-CH2 (Phe)),
4.20 (1 H, dd, J ) 7.2 Hz, R-CH (Cha)), 4.51 (1 H, dd, J ) 5.6,
8.8 Hz, R-CH (Phe)), 6.98-7.08 (5 H, m, Ar-H (Phe)); 13C NMR
(100.6 MHz, DMSO-d6/30% TFA-d1) δ 20.9 (Ac-CH3), 25.7 (ꢀ-
CH2 (Cha)), 25.8 (ꢀ + 1 CH2 (Cha)), 26.0 (NH-CH3), 32.1 (γ-
CH2 (Cha)), 33.2 (δ-CH2 (Cha)), 33.8 (â-CH2 (Cha)), 38.7 (â-
CH2 (Phe)), 52.2 (R-CH (Cha)), 54.7 (R-CH (Phe)), 127.1, 128.6,
129.3, 136.5 (Ar-C (Phe)), 173.2 (Ac-CO), 173.3 (CONH (Cha)),
174.4 (CONH (Phe)); high-resolution FAB MS m/z 374.2448
(M+ + 1). Anal. Calcd for C21H31N3O3: C, 67.53; H, 8.36; N,
11.25. Found: C, 65.95; H, 8.11; N, 10.91.
7.32 (5 H, m, Ar-H (phen)), 7.86 (1 H, d, J ) 8.0 Hz, CONH
(Leu)), 7.95 (1 H, m, CONH (phen));
13C NMR (100.6 MHz,
DMF-d7) δ 21.6 (γ-CH3 (Leu)), 22.5 (Ac-CH3), 25.0 (γ-CH (Leu)),
35.9 (â-CH2 (phen)), 41.0 (R-CH2 (phen)), 41.8 (â-CH2 (Leu)),
52.0 (R-CH (Leu)), 126.5, 128.7, 129.2, 140.2 (Ar-C (phen)),
169.8 (Ac-CO), 172.8 (CONH (Leu)); high-resolution FAB MS
m/z 277.1918 (M+ + 1). Anal. Calcd for C16H24N2O2: C, 69.57;
H, 8.70; N, 10.14. Found: C, 68.30; H, 8.32; N, 9.82.
N-Acetyl-D-p h en yla la n in e P h en eth yla m id e (11a ). Mp
) 171.5-172.5 °C; RP-HPLC retention time (10% MeOH/H2O:
MeCN, 60:40) ) 6.8 min; 1H NMR (400.1 MHz, DMSO-d6/10%
TFA-d1) δ 1.85 (3 H, s, Ac-CH3), 2.73 (2 H, dd, J ) 6.0 Hz,
â-CH2 (phen)), 2.77 (1 H, dd, J ) 9.2, 14.0 Hz, â-CH2 (Phe)),
2.98 (1 H, dd, J ) 5.2, 14.0 Hz, â-CH2 (Phe)), 3.35 (2 H, m,
R-CH2 (phen)), 4.51 (1 H, dd, J ) 5.2, 9.2 Hz, R-CH (Phe)),
7.11-7.20 (5 H, m, Ar-H (phen)); 13C NMR (100.6 MHz, DMSO-
d6/10% TFA-d1) δ 22.5 (Ac-CH3), 35.7 (â-CH2 (phen)), 38.4 (â-
CH2 (Phe)), 40.7 (R-CH2 (phen)), 54.9 (R-CH (Phe)), 126.6,
126.8, 128.6, 128.8, 129.2, 129.8, 138.6, 140.1 (Ar-C (Phe,
phen)), 170.5 (Ac-CO), 171.7 (CONH (Phe)); high-resolution
FAB MS m/z (M+ + 1). Anal. Calcd for C16H24N2O2: C, 73.79;
H, 6.80; N, 9.06. Found: C, 71.16; H, 7.01; N, 8.77.
N-Acet yl-D-cycloh exyla la n in e P h en et h yla m id e (12).
Mp ) 131.0-132.0 °C; RP-HPLC retention time (10% MeOH/
1
H2O:MeCN, 60:40) ) 5.4 min; H NMR (400.1 MHz, DMSO-
d6) δ 0.84 (2 H, m, ꢀ-CH2 (Cha)), 1.11 (4 H, m, δ-CH2 (Cha)),
1.35 (2 H, m, â-CH2 (Cha)), 1.61 (5 H, m, γ-CH2, R-CH (Cha)),
1.81 (3 H, s, Ac-CH3), 2.68 (2 H, dd, J ) 7.2 Hz, â-CH2 (phen)),
3.24 (2 H, m, J ) 7.6 Hz, R-CH2 (phen)), 4.22 (1 H, m, J ) 5.2
Hz, R-CH (Cha)), 7.18-7.28 (5 H, m, Ar-H (phen)), 7.88 (1 H,
N-Acetylglycin e P h en eth yla m id e (7). Mp ) 147.5-148.5
°C; RP-HPLC retention time (10% MeOH/H2O:MeCN, 80:20)
) 3.0 min; 1H NMR (400.1 MHz, DMSO-d6) δ 1.85 (3 H, s,
Ac-CH3), 2.69 (2 H, d, J ) 8.0 Hz, â-CH2 (phen)), 3.27 (2 H,
dt, J ) 13.6, 6.0 Hz, R-CH2 (phen)), 3.61 (2 H, d, J ) 5.6, CH2
(Gly)), 7.19-7.30 (5 H, m, Ar-H (phen)), 7.88 (1 H, m, CONH
(phen)), 8.04 (1 H, m, CONH (Gly)); 13C NMR (100.6 MHz,
DMSO-d6) δ 24.2 (Ac-CH3), 31.8 (â-CH2 (phen)), 36.8 (R-CH2
(phen)), 43.8 (CH2 (Gly)), 127.7, 130.0, 130.3, 141.1 (Ar-C
(phen)), 170.5 (Ac-CO), 171.2 (CONH (Gly)); high-resolution
FAB MS m/z 221.1288 (M+ + 1). Anal. Calcd for C12H16N2O2:
C, 65.45; H, 7.27; N, 12.73. Found: C, 65.51; H, 7.38; N, 12.73.
d, J ) 8.4 Hz, CONH (Cha)), 7.92 (1 H, m, CONH (phen)); 13
C
NMR (100.6 MHz, DMSO-d6) δ 23.4(Ac-CH3), 26.4, 26.5, 26.9,
32.8, 33.9, 34.3 (ring CH, CH2 (Cha)), 35.9 (â-CH2 (phen)), 41
(â-CH2 (Cha)), 41 (R-CH2 (phen)), 51.1 (R-CH (Cha)), 126.9,
129.1, 129.5, 140.2 (Ar-C (Phe, phen)), 169.8 (Ac-CO), 172.9
(CONH (Cha)); high-resolution FAB MS m/z 317.2235 (M+
+
1). Anal. Calcd for C19H28N2O2: C, 72.15; H, 8.86; N, 8.86.
Found: C, 72.11; H, 8.81; N, 9.00.
N-Acetyl-L-va lyl-D-p h en yla la n in e N-Meth yla m id e (3b).
Mp ) 265.5-266.5 °C; RP HPLC retention time (Dynamax
C-18, 0.46 × 25 cm; 1.0 mL/min; H2O:MeCN, 75:25; detected
at 206 nm) ) 9.1 min (compared with the D,D-diastereomer
U-98001, retention time ) 6.7 min); 1H NMR (400.1 MHz,
DMSO-d6) δ 0.47, 0.61 (6 H, 2d, J ) 6.4, 6.8 Hz, γ-CH3 (Val)),
1.66 (1 H, m, J ) 6.8 Hz, â-CH (Val)), 1.84 (3 H, s, Ac-CH3),
2.58 (3 H, d, J ) 4.8 Hz, NH-CH3), 2.69 (1 H, dd, J ) 11.2,
13.6 Hz, â-CH2 (Phe)), 3.08 (1 H, dd, J ) 4.0, 13.6 Hz, â-CH2
(Phe)), 3.91 (1 H, dd, J ) 7.6 Hz, R-CH (Val)), 4.41 (1 H, m,
R-CH (Phe)), 7.14-7.22 (5 H, m, Ar-H (Phe)), 7.86 (1 H, m,
NH-CH3), 7.89 (1 H, d, J ) 7.2 Hz, CONH (Val)), 8.32 (1 H, d,
J ) 8.4 Hz, CONH (Phe)); 13C NMR (100.6 MHz, DMSO-d6) δ
19.2, 19.7 (γ-CH3 (Val)), 23.2 (â-CH(Val)), 26.5 (Ac-CH3), 30.5
(NH-CH3), 38.1 (â-CH2 (Phe)), 55.1 (R-CH (Val)), 60.0 (R-CH
(Phe)), 127.0, 128.8, 129.9, 139.0 (Ar-C (Phe)), 170.7 (Ac-CO),
172.1 (CONH (Val)), 172.2 (CONH (Phe)); high-resolution FAB
MS m/z 320.1982 (M+ + 1). Anal. Calcd for C17H25N3O3: C,
63.93; H, 7.89; N, 13.16. Found: C, 63.49; H, 7.85; N, 12.99.
N-Acetyl-D-a la n in e P h en eth yla m id e (8). Mp ) 151.5-
152.5 °C; RP-HPLC retention time (10% MeOH/H2O:MeCN,
1
80:20) ) 6.1 min; H NMR (400.1 MHz, CD3OD) δ 1.28 (3 H,
d, J ) 7.2 Hz, â-CH3 (Ala)), 1.85 (3 H, s, Ac-CH3), 2.80 (2 H, d,
J ) 7.2 Hz, â-CH2 (phen)), 3.39 (2 H, dd, J ) 7.2, 8.8 Hz, R-CH2
(phen)), 4.27 (1 H, q, J ) 7.2, CH2 (Ala)), 7.18-7.31 (5 H, m,
Ar-H (phen)); 13C NMR (100.6 MHz, CD3OD) δ 17.1 (â-CH3
(Ala)), 21.4 (Ac-CH3), 31.8 (â-CH2 (phen)), 40.9 (R-CH2 (phen)),
49.5 (R-CH2 (Ala)), 126.3, 128.4, 128.9, 139.4 (Ar-C (phen)),
172.0 (Ac-CO), 174.1 (CONH (Ala)); high-resolution FAB MS
m/z 235.1450 (M+ + 1). Anal. Calcd for C13H18N2O2: C, 66.67;
H, 7.69; N, 11.97. Found: C, 66.42; H, 7.61; N, 11.73.
N-Acetyl-D-va lin e P h en eth yla m id e (9a ). Mp ) 219.5-
220.5 °C; RP-HPLC retention time (10% MeOH/H2O:MeCN,
1
80:20) ) 9.0 min; H NMR (400.1 MHz, DMSO-d6/10% TFA-
d1) δ 0.83 (6 H, d, J ) 6.8 Hz, γ-CH3 (Val)), 1.95 (3 H, s, Ac-
CH3), 1.97 (1 H, m, â-CH (Val)), 2.75 (2 H, t, J ) 6.8 Hz, â-CH2
(phen)), 3.30, 3.40 (2 H, m, R-CH2 (phen)), 4.12 (1 H, d, J )
6.8 Hz, CH2 (Val)), 7.11-7.24 (5 H, m, Ar-H (phen)); 13C NMR
(100.6 MHz, DMSO-d6/10% TFA-d1) δ 18.1, 19.6 (γ-CH3 (Val)),
22.7 (Ac-CH3), 31.2 (â-CH (Val)), 36.1(â-CH2 (phen)), 41.1 (R-
CH2 (phen)), 59.0 (R-CH (Val)), 126.7, 128.9, 129.4, 140.3 (Ar-C
(phen)), 170.1 (Ac-CO), 171.9 (CONH (Val)); high-resolution
FAB MS m/z 263.1758 (M+ + 1). Anal. Calcd for C15H22N2O2:
C, 68.70; H, 8.40; N, 10.69. Found: C, 67.97; H, 8.55; N, 10.54.
N-Acetyl-L-va lin e P h en eth yla m id e (9b). Mp ) 218.0-
219.0 °C; RP HPLC retention time (Chiralcel OD-H, 0.46 ×
25 cm; 0.5 mL/min; 2-propanol:hexane, 5:95; detected at 206
nm) ) 16.1 min (compared with the D-enantiomer U-145144,
1
retention time ) 12.9 min); H NMR (400.1 MHz, DMSO-d6)
δ 0.76, 0.78 (6 H, 2 d, J ) 2.0 Hz, γ-CH3 (Val)), 1.85 (3 H, s,
Ac-CH3), 1.87 (1 H, m, J ) 6.8 Hz, â-CH (Val)), 2.70 (2 H, dd,
J ) 7.2 Hz, â-CH2 (phen)), 3.22, 3.30 (2 H, m, R-CH2 (phen)),
4.05 (1 H, d, J ) 7.2 Hz, CH2 (Val)), 7.16-7.28 (5 H, m, Ar-H
(phen)); 13C NMR (100.6 MHz, DMSO-d6) δ 19.0, 20.0 (γ-CH3
(Val)), 23.3 (Ac-CH3), 31.1 (â-CH (Val)), 35.9 (â-CH2 (phen)),
40.2 (buried under DMSO peak; R-CH2 (phen)), 58.7 (R-CH
(Val)), 126.9, 129.1, 129.5, 140.2 (Ar-C (phen)), 170.0 (Ac-CO),
171.8 (CONH (Val)); high-resolution FAB MS m/z 263.1754
(M+ + 1). Anal. Calcd for C15H22N2O2: C, 68.67; H, 8.45; N,
10.68. Found: C, 68.06; H, 8.42; N, 10.51.
N-Acetyl-D-leu cin e P h en eth yla m id e (10). Mp ) 133.0-
134.0 °C; RP-HPLC retention time (10% MeOH/H2O:MeCN,
1
70:30) ) 5.2 min; H NMR (400.1 MHz, DMF-d7) δ 0.85, 0.88
(6 H, 2d, J ) 6.8, 6.4 Hz, δ-CH3 (Val)), 1.50 (2 H, m, â-CH2
(Leu)), 1.62 (1 H, m, J ) 6.4 Hz, γ-CH (Leu)), 1.93 (3 H, s,
Ac-CH3), 2.76 (2 H, dd, J ) 7.2 Hz, â-CH2 (phen)), 3.39 (2 H,
m, R-CH2 (phen)), 4.38 (1 H, m, J ) 8.8 Hz, R-CH (Leu)), 7.20-

Determinants of Passive Membrane Permeability
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 22 3729
(13) Hidalgo, I. J .; Raub, T. J .; Borchardt, R. T. Characterization of
the human colon carcinoma cell line (Caco-2) as a model system
for intestinal epithelial permeability. Gastroenterology 1989, 96,
736-749.
N-Acetyl-L-p h en yla la n yl-D-p h en yla la n in e N-Meth yla -
m id e (5b). Mp ) 247.0-248.0 °C; RP HPLC retention time
(Dynamax C-18, 0.46 × 25 cm; 1.0 mL/min; H2O:MeCN, 75:
25; detected at 206 nm) ) 17.4 min (compared with the D,D-
(14) Hilgers, A. R.; Conradi, R. A.; Burton, P. S. Caco-2 cell mono-
1
diastereomer U-97995, retention time ) 12.9 min); H NMR
layers as
a model for drug transport across the intestinal
(400.1 MHz, DMSO-d6) δ 1.72 (3 H, s, Ac-CH3), 2.43 (1 H, dd,
J ) 13.6 Hz, â-CH2 (Phe)), 2.58 (3 H, s, NH-CH3), 2.60 (1 H,
dd, â-CH2 (Phe)), 2.71 (1 H, dd, J ) 13.6, 10.0 Hz, â-CH2 (Phe)),
2.97 (1 H, dd, J ) 13.6 Hz, â-CH2 (Phe)), 4.43 (2 H, m, R-CH
(Phe)), 7.06-7.23 (10 H, m, Ar-H (Phe)), 7.91 (1 H, d, J ) 3.0
Hz, NHCH3), 7.97 (1 H, d, J ) 8.0 Hz, CONH (Phe)), 8.38 (1
H, d, J ) 8.4 Hz, CONH (Phe)); 13C NMR (100.6 MHz, DMSO-
d6) δ 23.2 (Ac-CH3), 26.4 (NH-CH3), 38.3 (â-CH2 (Phe)), 38.6
(â-CH2 (Phe)), 54.8 (R-CH (Phe)), 55.0 (R-CH (Phe)), 127.0,
127.1, 128.7, 128.8, 129.9, 130.0, 138.7, 138.8 (Ar-C (Phe)),
170.1 (Ac-CO), 172.0 (CONH (Phe)), 172.1 (CONH (Phe)); high-
resolution FAB MS m/z 368.1973 (M+ + 1). Anal. Calcd for
C21H25N3O3: C, 68.64; H, 6.86; N, 11.44. Found: C, 67.44; H,
6.66; N, 11.16.
mucosa. Pharm. Res. 1990, 7, 902-910.
(15) Artursson, P. Epithelial transport of drugs in cell culture. I: a
model for studying the passive diffusion of drugs over intestinal
absorptive (Caco-2) cells. J . Pharm. Sci. 1990, 79, 476-482.
(16) Wilson, G.; Hassan, I. F.; Dix, C. J .; Williamson, I.; Shah, R.;
Mackay, M.; Artursson, P. Transport and permeability properties
of human Caco-2 cells: an in vitro model of the intestinal
epithelial cell barrier. J . Controlled Release 1990, 11, 25-40.
(17) Burton, P. S.; Conradi, R. A.; Hilgers, A. R.; Ho, N. F. H.;
Maggiora, L. L. The relationship between peptide structure and
transport across epithelial cell monolayers. J . Controlled Release
1992, 19, 87-98.
(18) Adson, A.; Burton, P. S.; Raub, T. J .; Barsuhn, C. L.; Hilgers, A.
R.; Audus, K. L.; Ho, N. F. H. Passive diffusion of weak organic
electrolytes across Caco-2 cell monolayers: uncoupling the
contributions of hydrodynamics, transcellular and paracellular
barriers. J . Pharm. Sci. 1995, 84, 1197-1204.
(19) Ho, N. F. H.; Raub, T. J .; Burton, P. S.; Barsuhn, C. L.; Adson,
A.; Audus, K. L.; Borchardt, R. T. Quantitative Approaches to
Delineate Transport Mechanisms in Cell Culture Monolayers.
In Transport Processes in Pharmaceutical Systems; Himmelstein,
K. J ., Amidon, G. L., Lee, P. I., Eds.; Marcel Dekker: New York,
1999; pp 219-316.
(20) Burton, P. S.; Conradi, R. A.; Hilgers, A. R.; Ho, N. F. H.
Evidence for a polarized efflux system for peptides in the apical
membrane of Caco-2 cells. Biochem. Biophys. Acta 1993, 190,
760-766.
(21) Burton, P. S.; Goodwin, J . T.; Conradi, R. A.; Ho, N. F. H.;
Hilgers, A. R. In vitro permeability of peptidomimetic drugs: the
role of polarized efflux pathways as additional barriers to
absorption. Adv. Drug Deliv. Rev. 1997, 23, 143-156.
(22) Adson, A.; Raub, T. J .; Burton, P. S.; Barsuhn, C. L.; Hilgers, A.
R.; Audus, K. L.; Ho, N. F. H. Quantitative approaches to
delineate paracellular diffusion in cultured epithelial cell mono-
layers. J . Pharm. Sci. 1994, 83, 1529-1536.
N-Acetyl-L-p h en yla la n in e P h en eth yla m id e (11b). Mp
) 170.5-171.5 °C; RP HPLC retention time (Chirobiotic, 0.5
mL/min; MeOH:H2O, 30:70; detected at 206 nm) ) 18.0 min
(compared with the D-enantiomer U-145147, retention time
) 21.2 min).
1H NMR (400.1 MHz, DMSO-d6) δ 1.75 (3 H, s, Ac-CH3),
2.65 (2 H, dd, J ) 7.2 Hz, â-CH2 (phen)), 2.68 (1 H, dd, J )
9.6, 13.6 Hz, â-CH2 (Phe)), 2.88 (1 H, dd, J ) 4.8, 13.6 Hz,
â-CH2 (Phe)), 3.23 (2 H, m, R-CH2 (phen)), 4.41 (1 H, dd, J )
4.8 Hz, R-CH (Phe)), 7.16-7.29 (5 H, m, Ar-H (phen)), 8.06 (2
H, m, CONH (Phe, phen); 13C NMR (100.6 MHz, DMSO-d6) δ
23.3 (Ac-CH3), 35.9 (â-CH2 (phen)), 38.7 (â-CH2 (Phe)), 40.7
(R-CH2 (phen), buried under DMSO peak), 54.9 (R-CH (Phe)),
126.9, 127.0, 128.8, 129.1, 129.5, 129.9, 138.9, 140.2 (Ar-C (Phe,
phen)), 169.8 (Ac-CO), 171.9 (CONH (Phe)); high-resolution
FAB MS m/z 311.1761 (M+ + 1). Anal. Calcd for C16H24N2O2:
C, 73.52; H, 7.14; N, 9.02. Found: C, 72.47; H, 7.02; N, 8.90.
Ack n ow led gm en t. We thank Allen R. Hilgers and
Alan Clarke for assisting with performance of the
permeability studies and partition coefficient determi-
nations. We also acknowledge Dr. Boryeu Mao from
Computer Aided Drug Discovery and Dr. Thomas J .
Vidmar (Biostatistics) for helpful discussions.
(23) Schanker, L. S. On the mechanism of absorption of drugs from
the gastrointestinal tract. J . Med. Pharm. Chem. 1960, 2, 343-
359.
(24) El Tayer, N.; Tsai, R.-S.; Testa, B.; Carrupt, P.-A.; Leo, A.
Partitioning of solutes in different solvent systems: the contri-
bution of hydrogen-bonding capacity and polarity. J . Pharm. Sci.
1991, 80, 590-598.
(25) Paterson, D. W.; Conradi, R. A.; Hilgers, A. R.; Vidmar, T. J .;
Burton, P. S. A nonaqueous system for predicting the oral
absorption potential of peptides. Quant. Struct.-Act. Relat. 1994,
13, 4-10.
(26) Diamond, J . M.; Katz, Y. Interpretation of nonelectrolyte parti-
tion coefficients between dimyristoyl lecithin and water. J .
Membr. Biol. 1975, 17, 121-154.
(27) Marrink, S. J .; Berendsen, H. J . C. Permeation process of small
molecules across lipid membranes studied by molecular dynam-
ics simulations. J . Phys. Chem. 1996, 100, 16729-16738.
(28) Leo, A. J . Evaluating hydrogen-bond donor strength. J . Pharm.
Sci. 2000, 89, 1567-1578.
(29) Abraham, M. H.; Martins, F.; Mitchell, R. C.; Salter, C. J .
Hydrogen bonding. 47. Characterization of the ethylene glycol-
heptane partition system: hydrogen bond acidity and basicity
of peptides. J . Pharm. Sci. 1999, 88, 241-247.
(30) Platts, J . A.,; Abraham, M. J .; Hersey, A.; Butina, D. Estimation
of molecular linear free energy relationship descriptors. 4.
Correlation and predicition of cell permeation. Pharm. Res. 2000,
17, 1013-1018.
(31) Stenberg, P.; Luthman, K.; Artursson, P. Prediction of membrane
permeability to peptides from calculated molecular surface
properties. Pharm. Res. 1999, 16, 205-212.
Refer en ces
(1) Pond, S. M.; Tozer, T. N. First-pass elimination. Basic concepts
and clinical consequences. Clin. Pharmacokinet. 1984, 9, 1-25.
(2) Navia, M. A.; Chaturvedi, P. R. Design principles for orally
bioavailable drugs. Drug Discovery Today 1996, 1, 179-189.
(3) Mayerson, M. Principles of drug absorption. In Modern Phar-
maceutics, 2nd ed.; Banker, G. S., Rhodes, C. T., Eds.; Marcel
Dekker: New York, 1990; pp 23-90.
(4) Trier, J . S. Structure of the mucosa of the small intestine as it
relates to intestinal function. Fed. Proc. 1967, 26, 1391-1404.
(5) Parson, D. S.; Boyd, C. A. R. Transport across the intestinal
mucosal cell: hierarchies of function. Int. Rev. Cytol. 1972, 32,
209-255.
(6) Crissinger, K. D.; Kvietys, P. R.; Granger, D. N. Pathophysiology
of gastrointestinal mucosal permeability. J . Int. Med. 1990, 228
(Suppl. 1), 145-154.
(7) Wright, E. M. Solute and water transport across epithelia. Am.
Rev. Respir. Dis. 1983, 127, S3-S8.
(8) Diamond, J . M. The epithelial junction: bridge, gate and fence.
Physiologist 1977, 20, 10-18.
(9) Gumbiner, B. Structure, biochemistry and assembly of epithelial
tight junctions. Am. J . Physiol. 1987, 253, C749-58.
(10) Conradi, R. A.; Hilgers, A. R.; Ho, N. F. H.; Burton, P. S. The
influence of peptide structure on transport across Caco-2 cells.
Pharm. Res. 1991, 8, 1453-1460.
(32) Palm, K.; Luthman, K.; Ungell, A.-L.; Strandlund, G.; Artursson,
P. Correlation of drug absorption with molecular surface proper-
ties. J . Pharm. Sci. 1996, 85, 32-39.
(33) Tartar, A.; Gesquiere, J . C. Formation of CH2Cl2-soluble urea
derivatives during solid-phase peptide synthesis with unsym-
metrical carbodiimides. J . Org. Chem. 1979, 44, 5000-5002.
(34) Beyermann, M.; Henklein, P.; Klose, A.; Sohr, R.; Bienert, M.
Effect of tertiary amine on the carbodiimide-mediated peptide
synthesis. Int. J . Pept. Protein Res. 1991, 37, 252-256.
(11) Conradi, R. A.; Hilgers, A. R.; Ho, N. F. H.; Burton, P. S. The
influence of peptide structure on transport across Caco-2 cells.
II. Peptide bond modification which results in improved perme-
ability. Pharm. Res. 1992, 9, 435-439.
(12) Pinto, M.; Robine-Leon, S.; Appay, M. D.; Kedinger, M.; Triadou,
N.; Dussaulx, E.; LaCroix, B.; Simon-Assman, P.; Haffen, K.;
Fogh, J .; Zweibaum, A. Enterocyte-like differentiation and
polarization of the human colon carcinoma cell line Caco-2 in
culture. Biol. Cell 1983, 47, 323-330.
J M010253I