Amines that transport protons across bilayer membranes: Synthesis, lysosomal neutralization, and two-phase pKa values by NMR

Article abstract of DOI:10.1021/jo960367f

It is desirable to be able to control the pH of lysosomes. A collection of lipophilic, nitrogenous bases, designed to act as membrane-active, catalytic proton transfer agents, were prepared and their effective pK_as measured in a vigorously stirred, two-phase system. One phase was a phosphate buffer whose pH was varied over the range ca. 1-11. The other was an immiscible, deuterated organic solvent in which the compounds preferentially resided even when protonated. When chemical shift changes versus the pH of the buffer were plotted, clear pK_a curves were generated that are relevant to transmembrane proton transfer behavior. The two-phase pK_as increased with increasing counterion lipophilicity and with increasing organic solvent polarity. The compounds were also tested for their ability to neutralize the acidity of lysosomes, a model for other acidic vesicles involved in drug sorting. The most successful of these, imidazole 6a, has > 100 times the neutralizing power of ammonia, a standard lysosomotropic amine, causing a 1.7 unit rise in lysosomal pH of RAW cells at 0.1 mM, compared to a 0.2 and 1.4 unit rise for ammonium chloride at 0.1 and 10 mM, respectively.

Full text of DOI:10.1021/jo960367f

4676
J . Org. Chem. 1996, 61, 4676-4684
Am in es Th a t Tr a n sp or t P r oton s a cr oss Bila yer Mem br a n es:
Syn th esis, Lysosom a l Neu tr a liza tion , a n d Tw o-P h a se p Ka Va lu es
by NMR
Gene M. Dubowchik,* Linda Padilla, Kurt Edinger, and Raymond A. Firestone
Bristol-Myers Squibb Pharmaceutical Research Institute, P.O. Box 5100, Wallingford, Connecticut 06492
Received February 21, 1996X
It is desirable to be able to control the pH of lysosomes. A collection of lipophilic, nitrogenous
bases, designed to act as membrane-active, catalytic proton transfer agents, were prepared and
their effective pKas measured in a vigorously stirred, two-phase system. One phase was a phosphate
buffer whose pH was varied over the range ca. 1-11. The other was an immiscible, deuterated
organic solvent in which the compounds preferentially resided even when protonated. When
chemical shift changes versus the pH of the buffer were plotted, clear pKa curves were generated
that are relevant to transmembrane proton transfer behavior. The two-phase pKas increased with
increasing counterion lipophilicity and with increasing organic solvent polarity. The compounds
were also tested for their ability to neutralize the acidity of lysosomes, a model for other acidic
vesicles involved in drug sorting. The most successful of these, imidazole 6a , has >100 times the
neutralizing power of ammonia, a standard lysosomotropic amine, causing a 1.7 unit rise in
lysosomal pH of RAW cells at 0.1 mM, compared to a 0.2 and 1.4 unit rise for ammonium chloride
at 0.1 and 10 mM, respectively.
In tr od u ction
The ability of eukaryotic cells to internally segregate
aqueous compartments differing in pH using lipid bilayer
membranes and ATP-driven proton pumps has allowed
the evolutionary development of sophisticated mecha-
nisms for cell nutrition and defense (lysosomes and
endosomes)1 and protein synthesis, modification, and
sorting (Golgi and endoplasmic reticulum).2 However,
harmful organisms3 and disease states4 have also evolved
that depend upon this compartmentalization. There is
increasing evidence, for example, that the phenomenon
of multiple drug resistance (MDR), one of the greatest
problems in cancer chemotherapy, is linked to intracel-
lular pH and may involve the sorting of anticancer drugs
in low pH vesicles before expulsion from the cancer cell.5
In a wholly different therapeutic area, many viruses such
as influenza use the low pH of endosomes in an elaborate
mechanism that results in membrane fusion and ulti-
mately infection of the host cell.6
F igu r e 1. Schematic model of lysosomal neutralization by a
membrane-active proton transporter.
As part of a program in which we seek to modulate
these pH differences in order to treat disease, we have
prepared membrane-active agents capable of transport-
ing protons across lipid bilayers resulting in the dissipa-
tion of pH gradients. Unlike standard lysosomotropic
amines such as ammonia, which transport protons from
aqueous to lipid to aqueous phases, these compounds are
designed to carry protons across bilayer membranes
much more efficiently by picking them up on one side
and depositing them on the other while remaining all the
while within the bilayer (Figure 1). They consist of a
basic nitrogen, aliphatic or heterocyclic, to which are
attached three lipophilic groups, usually alkyl chains of
varying length. Having only one alkyl chain would make
the molecule a detergent, which would disrupt the
membrane. Two alkyl chains would make it a bilayer
former and thus very slow to move from one face of the
membrane to the other. An exception is the adamantyl
group which is probably too bulky to contribute to bilayer
formation. Our experience with lysosomotropic deter-
gents suggested that the basic portion of the molecule
should have a pKa between 5 and 7 so that it would
X Abstract published in Advance ACS Abstracts, J une 15, 1996.
(1) (a) Holtzmann, E. Lysosomes; Plenum Press: New York, 1989.
(b) Pisoni, R. L.; Thoene, J . G. Biochim. Biophys. Acta 1991, 1071, 351-
373. (c) Mellman, I. J . Exp. Biol. 1992, 172, 39-45. (d) Morton, P. A.;
Zacheis, M. L.; Giacoletto, K. S.; Manning, J . A.; Schwartz, B. D. J .
Immunol. 1995, 154, 137-150.
(2) (a) Seksek, O.; Biwersi, J .; Verkman, A. S. J . Biol. Chem. 1995,
270, 4967-4970. (b) Chidgey, M. A. J . Bioessays 1993, 15, 317-321.
(c) Geuze, H. Eur. J . Cell. Biol. 1994, 64, 3-6.
(3) Examples of organisms that reside within lysosomes. Bacteria:
(a) Maurin, M.; Benoliel, A. M.; Bongrand, P.; Raoult, D. J . Infect. Dis.
1992, 166, 1097-1102. Trypanosomes: (b) Andrews, N. W. Trends Cell.
Biol. 1995, 5, 133-137.
(4) Examples of disease states whose degenerative mechanisms
involve lysosomal processing. Alzheimer’s disease: (a) Refolo, L. M.;
Sambamurti, K.; Efthimiopoulos, S.; Pappolla, M. A.; Robakis, N. K.
J . Neurosci. Res. 1995, 40, 694-706. Prion disease: (b) Arnold, J . E.;
Tipler, C.; Laszlo, L.; Hope, J .; Landon, M.; Mayer, R. J . J . Pathol.
1995, 176, 403-411. (c) Prusiner, S. B. Sci. Am. 1995, 272, 48-57.
Oxidative stress: (d) Ollinger, K.; Brunk, U. T. Free Radical Biol. Med.
1995, 19, 565-574.
(5) (a) Dubowchik, G. M.; Padilla, L. M.; Edinger, K.; Firestone, R.
A. Biochim. Biophys. Acta 1994, 1191, 103-108. (b) Marquardt, D.;
Center, M. S. Cancer Res. 1992, 52, 3157-3163. (c) Simon, S.; Roy,
D.; Schindler, M. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 1128-1132.
(d) Boiocchi, M.; Toffoli, G. Eur. J . Cancer 1992, 28A, 1099-1105. (e)
Slapak, C. A.; Lecerf, J . M.; Daniel, J . C.; Levy, S. B. J . Biol. Chem.
1992, 267, 10638-10644.
(6) (a) Bullough, P. A.; Hughson, F. M.; Skehel, J . J .; Wiley, D. C.
Nature 1994, 371, 37-43. (b) Gaudin, Y.; Ruigrok, R. W. H.; Brunner,
J . J . Gen. Virol. 1995, 76, 1541-1556.
S0022-3263(96)00367-2 CCC: $12.00 © 1996 American Chemical Society

Bilayer Membrane-Active Proton Transporting Agents
J . Org. Chem., Vol. 61, No. 14, 1996 4677
Sch em e 1
Sch em e 3
Sch em e 2
Sch em e 4
selectively localize in lysosomal membranes.7 However,
since these new agents are intended to protonate and
deprotonate while inside the membrane, and not in
water, we decided to investigate their acidity constants
in various organic solvents, as models for the environ-
ment of the lipid bilayer. For this we designed a series
of NMR experiments in which the compounds, in water-
immiscible, deuterated solvents, were stirred with an
aqueous phase consisting of a buffer whose pH could be
varied. Clear pKa curves were produced by monitoring
the change in chemical shift of various protons in the
molecules.8 As might be expected, these pKas were
dependent both on organic solvent and on counterion.
Members of each class of compounds were then tested to
determine whether they caused lysosomal neutralization
in vitro at low concentrations.
Resu lts a n d Discu ssion
prepared in situ and then reacted with 1,2-diaminocy-
clohexane. The resulting imidazoline 7 was oxidized to
the imidazole 8 using Swern conditions. N-Alkylation
was effected by treatment with sodium hydride followed
by iodobutane to give 9.
Syn th esis. The trialkylated pyridines 1 were pre-
pared in one step by the method of Bo¨nneman,9 using
cobalt-mediated reductive cyclization of 2 equiv of alkynes
with 1 equiv of nitrile (Scheme 1). Trialkylated benzenes
and the unsymmetrical pyridines 2 are also formed but
are easily separated from the desired products.
Branched imidazoles 6 were obtained from repetitive
lithiation and alkylation of SEM-protected imidazole 3
(Scheme 2).10 The first alkylation occurs exclusively at
C-2, while subsequent reactions occur at both C-5 and
at the new R-position with little apparent discrimination.
Removal of the SEM group from the crowded nitrogen
in 5 requires vigorous conditions (HCl/TFA, 90 °C, 40 h)
which, nevertheless, proceeds cleanly.
Scheme 4 shows the preparation of the 2,2-difluoro-
trialkylamines 16. Methyl valerate and methyl hep-
tanoate were acylated with dimethyl oxalate, and the
intermediates 10 were decarboxylated.11 Treatment with
DAST converted the 2-oxo esters 11 to 2,2-difluoro esters
12 in high yield,12 and basic hydrolysis afforded the free
acids 13. Formation of the tertiary amides 15 was found
to proceed most cleanly through addition of the appropri-
ate secondary amines to the symmetric anhydrides 14.
The extra equivalent of 13 was easily recovered by basic
extraction of the reaction mixture. Reduction of the
amides 15 with diborane, followed by methanolysis, gave
the 2,2-difluoro amines 16 in good yield.
Fused bicyclo imidazole 9 was synthesized as shown
in Scheme 3. The methyl imidate from valeronitrile was
A series of (adamantylmethyl)amines (18 and 20) was
synthesized as shown in Scheme 5 by straightforward
acylation with adamantanecarbonyl chloride and reduc-
tion of the resulting amides (17 and 19). (Adamantyl-
methyl)imidazole 22 was prepared by the very slow
(7) Miller, D. K.; Griffiths, E.; Lenard, J .; Firestone, R. A. J . Cell
Biol. 1983, 97, 1841-1851. (b) Firestone, R. A.; Pisano, J . M.; Bonney,
R. J . J . Med. Chem. 1979, 22, 1130-1133.
(8) A recent paper describes the determination of pKa curves of
pyrrole carboxylic acids by plotting 13C NMR chemical shifts versus
pH in a single phase (H2O-DMSO-d6): Holmes, D. L.; Lightner, D. A.
Tetrahedron 1995, 51, 1607-1622.
(9) Bo¨nneman, H.; Brinkmann, S.; Schenkluhn, H. Synthesis 1974,
575-577.
(10) Lipschutz, B. H.; Huff, B.; Hagen, W. Tetrahedron Lett. 1988,
29, 3411-3414.
(11) Erni, B.; Khorana, H. G. J . Am. Chem. Soc. 1980, 102, 3888-
3896.
(12) Middleton, W. J ; Bingham, E. M. J . Org. Chem. 1980, 45, 2883-
2887.

4678 J . Org. Chem., Vol. 61, No. 14, 1996
Dubowchik et al.
Sch em e 5
Sch em e 6
F igu r e 2. Two-phase pKas measured in CDCl3 stirred with
phosphate buffer containing sodium chloride (1 M).
significant amount of the compound might dissolve inside
the acidic vesicle where it would have only a bulk effect
and not the desired catalytic one.
Tw o-P h a se p Ka Mea su r em en ts. Our aim was to
determine the effective pKas of these agents in an
environment that would be as close as possible to the one
they would experience if they were acting as we designed
them, i.e., undergoing protonation and deprotonation
while residing within the lipid bilayer. In any case,
aqueous pKas measured in the usual ways would be gross
approximations because all of these compounds, in their
unprotonated form, are practically insoluble in water. We
also wished to avoid using standard pH electrodes in
mixed aqueous/organic media because of the uncertain
correction factors involved.
The solution that we found was a two-phase system:
one an aqueous buffer whose pH could be varied and then
measured using a standard glass electrode and the other
an immiscible, deuterated NMR solvent containing the
compound under study. Indeed, a two-phase system is
more realistic than the standard single-phase titration
because it mimics more closely the actual situation in
living cells, where the drug will be inside the bilayer,
influencing the pH in an adjacent aqueous compartment.
By vigorously stirring the two phases over 10 min, we
found that the protonation state of the compound reached
an equilibrium that depended on three factors: the
inherent basicity of the compound, the nature of the
organic solvent (our substitute for the interior of the lipid
bilayer), and the lipophilicity of the counterion. We
assumed that the most lipophilic anion in the aqueous
phase, as long as it was present in excess, would be
carried into the organic phase as the counterion. Por-
tions of the organic phase were removed and the proton
NMR spectra measured. The extent of protonation of
each compound was clearly mirrored by the downfield
shifts of proton signals close to the basic site. When these
shifts were plotted versus the pH of the buffer, clear
titration curves were generated (Figure 2). In order to
ensure that all of the protonated compound remained in
the deuterated solvent, the aqueous phase at the acidic
limit of the experiment (pH e 2) was removed, basified
reaction of imidazole with mesylate 21 under phase
transfer conditions (Scheme 6). Formation of the very
crowded 1,1-diadamantylamine (25), even in low yield,
required generation of adamantyl 1-triflate 24 in situ by
treatment of the bromide 23 with silver triflate, followed
by quenching with adamantylamine. Neither 23 itself,
nor the corresponding nitrate (from treatment with silver
nitrate), was reactive enough to couple with adamantyl-
amine.
Hyd r oca r bon Ma ss. In order for our compounds to
localize in the hydrocarbon region of the phospholipid
bilayer and remain there in the protonated state, they
must carry sufficient hydrocarbon mass. On the other
hand, for effective delivery of these agents through
aqueous media, lipophilicity must be minimized. To
determine the minimum effective hydrocarbon bulk, we
used chloroform as a surrogate for the interior of the
membrane and tried to extract two borderline compounds
(tributylamine and tripentylamine) with pH 4 buffer
(somewhat more acidic than lysosomes) or 10% HCl. In
the case of tributylamine (C12), 21% was extracted into
pH 4 buffer, while none of the tripentylamine (C15) was
detected there. In 10% HCl the acid-extractable fractions
were 31% and 12%, respectively. This demonstrated to
us that a hydrocarbon mass of C12 was too low since a

Bilayer Membrane-Active Proton Transporting Agents
J . Org. Chem., Vol. 61, No. 14, 1996 4679
Ta ble 1. Tw o-P h a se p Ka s Deter m in ed by P r oton NMR in
CDCl3 After Stir r in g w ith Ch lor id e-Con ta in in g
P h osp h a te Bu ffer a t Va r iou s p H Va lu es
Ta ble 2. Tw o-P h a se p Ka s Deter m in ed by P r oton NMR in
CDCl3 After Stir r in g w ith a P h osp h a te Bu ffer Con ta in in g
Differ en t Cou n ter ion s a t Va r iou s p H Va lu es
Asp2-
3.88
Cl-
Br-
I-
TsO-
8.42
a
2-
compound
two-phase pKa
parent pKa
deviation
compound
PO4
3.29
trihexylamine (26)
trioctylamine (27)
6.43
6.42
3.32
3.11
6.23
5.51
5.42
1.77
5.3
5.5
6.1
5.5
6.25
10.65b
10.65b
7.48c
7.48c
8.3d
4.2
4.2
4.2
4.4
2.1
2.8
2.9
5.7
5.3
5.1
4.4
4.6
4.7
27
18c
16b
6.42
6.11
1.77
7.32
7.02
2.64
8.21
8.10
4.02
1a
1b
6a
6b
8.3d
9
8.3d
16b
18a
18b
18c
18d
25
7.5e
10.65b
10.65b
10.46f
10.30g
10.91h
a
b
d
See ref 14. Triethylamine. c 2,4,6-Trimethylpyridine. 2,4-
Dimethylimidazole. e (2,2-Difluoroethyl)diethylamine. f N-Meth-
ylpyrrolidine. N-Methylazacycloheptane. Bis-sec-butylamine.
g
h
to >pH 12, and extracted with chloroform. Proton NMR
showed that only (adamantylmethyl)imidazole 22, the
least lipophilic compound in our study, was extracted into
the acid to any detectable extent.13
For most of our compounds a steady downfield shift,
with no change in the signal patterns, was observed with
increasing protonation. The exceptions were the ada-
mantylmethyl series of compounds (18a -d ), which dis-
played broadened signals for two distinct species in the
pH range near the apparent pKa. Fully protonated
18a -d showed two very complex multiplets for each of
the alkyl chain (or cyclic) methylene groups next to the
nitrogen. These were separated by as much as 1.1 ppm
(18c) or as little as 0.2 ppm (18a and b). The unproto-
nated species, on the other hand, showed the expected
single triplet signal for these protons. Conversion from
one state to the other in either direction did not result
in a significant shift in any of the signals for 18a -d , only
a broadening and a gradual replacement by the signals
for the new specie, suggesting hindered rotation around
the C-C and C-N bonds between the bulky adamantyl
group and the nitrogen. Two factors are most likely to
account for this increased barrier to rotation: the de-
creased rate of nitrogen lone pair inversion on protona-
tion and the fact that the proton on nitrogen is an added
steric element that might encounter bridgehead adaman-
tyl methylene groups. The two-phase pKas of these
compounds could be estimated by integration of the
interconverting species.
The derived pKa values of our compounds in deuterated
chloroform with chloride ion are collected in Table 1. Also
included are aqueous pKa values, from the literature,14
of similar compounds that contain significantly less
hydrocarbon mass. Most of these comparison compounds
substitute a methyl or ethyl group in place of the
hydrocarbon chain or the adamantylmethyl group in the
compound of interest. The two-phase pKas are all
significantly lower than those of the parent compounds.
In addition, the range of deviation is quite large, extend-
ing from 2 to 3 pKa units for the imidazole compounds (6
and 9) to almost 6 units for the â-difluoro amine 16b. It
is not surprising that the pKas measured in this way
F igu r e 3. Two-phase pKas measured in various solvents
stirred with phosphate buffer containing sodium chloride (1
M).
should be lower than would be expected in water because
the solvation energy of the ion pair is much greater in
water.15,16 This is shown dramatically in Table 2 which
represents experiments in which the anion available in
the buffer is varied to include a range of hydrophobicities.
The relative solvation energy that is sacrificed in going
from water to the organic solvent is reflected in the two-
phase pKas, i.e., the less lipophilic the anion (e.g., PO42-),
the lower the pH required to carry it into the organic
phase. In the case of trioctylamine 27, phosphate
depresses the two-phase pKa in comparison with that
measured with chloride by more than 3 pKa units while
tosylate increases it by 2 units. The magnitude of this
effect appears to be consistent as well since the two-phase
pKas of all three compounds in Table 2 increase by nearly
the same amounts on going from chloride to bromide and
then to iodide. The ordering of these anion-dependent,
two-phase pKas are also mirrored in the order of lysoso-
mal membrane anion permeability.17
Significant variations in two-phase pKas of 27 (with
chloride ion) were also seen in experiments in which the
organic phase was varied (Figure 3). A bulk organic
solvent can only be a rough approximation of the hydro-
phobic environment inside a lipid bilayer, especially since
it cannot model aspects of molecular packing and ease
of motion of “solute” molecules.18 Chloroform is most
widely used to mimic the lipid bilayer in bulk liquid
membrane transport experiments.19 Early studies were
carried out using toluene, though a more useful model
(15) Reichardt, C Solvents and Solvent Effects in Organic Chemistry;
VCH: New York, 1988.
(16) For a discussion of pKa measurement in nonaqueous solvents,
see: Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456-463 and refernces
therein.
(17) Casey, R. P.; Hollemans, M.; Tager, J . M. Biochim. Biophys.
Acta 1978, 508, 15-26.
(18) Datta, D. B. A Comprehensive Introduction to Membrane
Biochemistry; Floral Publishing: Madison, 1987.
(13) Another compound whose pKa could not be measured using our
method was N-(adamantylmethyl)adamantylamine (20), the proto-
nated form of which was insoluble in both the aqueous and organic
phases.
(14) Perrin, D. D. Dissociation Constants of Organic Bases in
Aqueous Solution: Supplement 1972; Butterworths: London, 1972.

4680 J . Org. Chem., Vol. 61, No. 14, 1996
Dubowchik et al.
Ta ble 3. RAW Cell Lysosom a l Neu tr a liza tion In d u ced by
Lysosom otr op ic Am in es Mea su r ed by F low Cytom etr y
concn
(mM)
compound
NH4Cl
NH4Cl
NH4Cl
pKa
vehicle
∆pHc
9.21a,b
9.21a,b
9.21a,b
10.62a,b
10
1
water
water
water
water
water
1.4
0.7
0.2
1.4
0
0.5
1.0
1.7
0.4
0.6
0.2
0.3
<0.1
<0.1
<0.1
0.2
0.1
10
10
0.1
0.1
0.1
1
1
0.1
1
1
1
1
0.1
1
CH3NH3Cl
tributylamine 10.89a,b
6a
6.23
6.23
6.23
lipid emulsion
water
6a ‚HCl
6a ‚HCl
8‚HCl
9‚HCl
18b‚HCl
18b
18d
18c
22
1a
Molecusol complexd
water
5.42
5.3
5.5
5.5
6.1
e
water
water
lipid emulsion
lipid emulsion
lipid emulsion
lipid emulsion
lipid emulsion
lipid emulsion
3.32
1.77
16a
<0.1
F igu r e 4. Change in pH of an unbuffered HCl source phase
starting at ca. pH 4 during U-tube experiment in which
trioctylamine (27) and tetradecylpyridine (1b) in chloroform
(0.002 mM) transported protons to a phosphate buffer receiving
phase at pH 7.4.
a
b
Measured in a single-phase aqueous system. Hall, H. K. J .
Am. Chem. Soc. 1957, 79, 5441-5444. c Where pH 4.8 is taken
d
for the starting value (see ref 20). Molecusol alone had no effect
on the pH of the cells and was nontoxic. e Extracted into aqueous
acid.
was later found by adding 30% 1-butanol.20 Perhaps not
coincidentally chloroform and 7:3 toluene/1-butanol yield
virtually the same two-phase pKa values (6.2-6.4). When
chloroform is replaced by carbon tetrachloride, the pKa
falls to 3.3. This may represent a significant stabilization
of the chloride ion by hydrogen bonding in chloroform.
The pKas shown in Figure 3 correlate roughly with the
dielectric constants of the organic solvents. Note the
especially low pKa in cyclohexane, demonstrating extreme
difficulty in creating ions in this highly nonpolar medium.
The case of the butanol/toluene mixture is of course more
complex since the dielectric constants of mixtures of
solvents whose properties are different are not additive.
In a case like this it is conceivable that an ion pair could
assemble around itself an environment of higher polarity
if the loss in entropy is compensated by the gain in charge
stabilization.15
We also carried out several U-tube experiments in
which trioctylamine (27) and tridecylpyridine 1b were
dissolved in chloroform (0.002 mM) between an unbuf-
fered HCl/NaCl source phase at ca. pH 4 and a phosphate
buffer (0.1 M) receiving phase at pH 7.4. All three phases
were gently stirred, and the pH of the unbuffered side
was monitored. These experiments were clumsy to run
so we did not carry them out for each compound.
However, the effort that we put into them gave us some
confirmation of our two-phase pKa data. As shown in
Figure 4 proton transport by 27 is relatively fast until
the pH of the unbuffered side approaches the two-phase
pKa of the transporting agent (ca. 6.4). From there,
progress to the buffered side pH of 7.4 is very slow.
Proton transport by 1b is slow from the outset since the
pH of the acid side (3.9) is already above the two-phase
pKa (3.1).
ylamine, which raise lysosomal pH, are well-known
through the work of Ohkuma and Poole.21 Most of these
agents cause vacuolization of lysosomes as a result of a
buildup of osmotic stress. This is because the unproto-
nated bases pass easily into lysosomes through their
membranes, but cannot readily depart because, after
protonation, they are trapped in the aqueous phase, so
that the phospholipid membrane now becomes a barrier.
An exception is tributylamine which, Poole and Ohkuma
noted,21,22 may be lipophilic enough to cross freely even
when protonated.
A selected number of our compounds were tested for
their ability to cause lysosomal neutralization of RAW
cells over time by measurement of the fluorescence
emission of fluorescein isothiocyanate (FITC)-dextran
using a flow cytometer (Table 3). RAW cells are a murine
monocytic macrophage line that are rich in lysosomes.
Upon incubation, the cells selectively localize FITC-
dextran in lysosomes by endocytosis. The degree of
neutralization over time differed greatly depending on
the compound. As observed by Ohkuma and Poole,21
ammonium chloride caused an almost instantaneous rise
in pH. We have used their standard of pH 6.2 at 10 mM
ammonium chloride as the yardstick by which we mea-
sured the pH differences induced by our new compounds.
Methylamine hydrochloride, also at 10 mM, raised the
pH to the same degree but over several minutes. We
were unable, however, to duplicate their results with
tributylamine (using no dispersing agent). Tributyl-
amine was not very soluble in the medium, and little may
have reached the cells. Our compounds caused a steady
rise in lysosomal pH over ca. 15 min, at which time it
leveled off.
All of the compounds exhibited some degree of activity
and were nontoxic to the cells at the concentrations
tested. Being very lipophilic, they would not simply
dissolve unaided in medium. At first we used a milky
emulsion formed by sonicating the compounds with egg
yolk phosphatidylcholine in medium. This was sufficient
In Vitr o Lysosom a l Neu tr a liza tion . The best mod-
els for intracellular acidic vesicles are lysosomes. Ni-
trogenous bases such as ammonium chloride and meth-
(19) Recent examples: (a) Lambert, E; Breinlinger, E. C.; Rotello,
V. M. J . Org. Chem. 1995, 60, 2646-2647. (b) Miyake, H.; Yamashita,
T.; Kojima, Y.; Tsukube, H. Tetrahedron Lett. 1995, 36, 7669-7672.
(c) Colton, I. J .; Kazlauska, R. J . J . Org. Chem. 1994, 59, 3626-3635.
(20) (a) Behr, J .-P.; Lehn, J .-M. J . Am. Chem. Soc. 1973, 95, 6108-
6110. (b) Lehn, J .-M. In Physical Chemistry of Transmembrane Ion
Motions; Elsevier: Amsterdam, 1983.
(21) Ohkuma, S; Poole, B. Proc. Natl. Acad. Sci. U.S.A. 1978, 75,
3327-3331.
(22) Poole, B; Ohkuma, S. J . Cell Biol. 1981, 90, 665-669.

Bilayer Membrane-Active Proton Transporting Agents
J . Org. Chem., Vol. 61, No. 14, 1996 4681
calcd for C17H30N 248.2378, found 248.2369. Hyd r och lo-
r id e: 1H NMR δ 0.88 (m, 9H), 1.36 (m, 6H), 1.59 (m, 2H),
1.73 (m, 4H), 2.69 (t, 2H), 3.17 (t, 4H), 7.12 (s, 2H); MS (FAB)
248 (MH+). Anal. Calcd for C17H29NCl: C, 71.93; H, 10.65;
N, 4.93. Found: C, 72.13; H, 10.51; N, 4.66.
2,4,6-Tr id ecylp yr id in e (1b) was prepared as described
above for 1a (38%): 1H NMR δ 0.83 (m, 9H), 1.31 (m, 42H),
1.58 (m, 2H), 1.64 (m, 4H), 2.48 (t, 2H), 2.68 (t, 4H), 6.73 (s,
2H); MS (FAB) 500 (MH+); HRMS calcd for C35H65N 500.5195,
found 500.5185. Anal. Calcd for C35H65N: C, 84.09, H, 13.11,
N, 2.80. Found: C, 83.78, H, 12.73, N, 2.45. Hyd r och lo-
r id e: 1H NMR δ 0.81 (t, 9H), 1.23 (m, 42H), 1.60 (m, 2H),
1.72 (t, 4H), 2.68 (t, 2H), 3.08 (t, 4H), 7.02 (s, 2H).
to determine which compounds looked promising but was
hardly satisfactory. Imidazole 6a was clearly superior
in these initial tests even when it was obvious that much
of the compound was not reaching the cells. At 1 mM
most of the other compounds showed detectable but very
slight neutralization. Exceptions were the tributylpyri-
dine 1a and (adamantylmethyl)dibutylamine 18b. Nei-
ther retained enough activity in subsequent tests to
warrant continued interest.
It became clear to us in the course of these studies that
these compounds displayed much less activity when
assays were done using plastic, presumably because of
binding of the lipophilic compounds to the walls of the
vessel. Therefore, whenever possible, biological testing
was done in glass vessels.
We found that delivery of 6a to the cells could be
improved greatly, first by forming the hydrochloride salt
and adding it to the cells as a water solution, thereby
doubling the lysosomal pH increase, and then by forming
the hydroxypropylated â-cyclodextrin complex of the
hydrochloride. This was completely water soluble to a
concentration of 6a of at least 10 mM. At 0.1 mM, 6a
raises lysosomal pH to ca. 6.5 within 15 min as its
cyclodextrin complex. This is superior to ammonium
chloride at 100 times the concentration, and taken
together with the fact that the hydrochloride salt of 6a
is not extracted out of chloroform into water, argues
strongly for its acting by the catalytic mechanism pro-
posed above. The fused cyclohexyl imidazoles 8 and 9
also exhibited activity that was significant, but still
greatly inferior to that of 6a .
1-[[2-(Tr im eth ylsilyl)eth oxy]m eth yl]im id a zole (SEM-
im id a zole) (3). A stirred suspension of sodium hydride (60%
dispersion in mineral oil, 0.68 g, 1.1 equiv) in THF (14 mL)
under argon was cooled to 0 °C and then treated with a
solution of imidazole (1.05 g, 15.4 mmol) in THF (6 mL). After
5 min the reaction was warmed to rt for 30 min, recooled to 0
°C, and treated with [2-(trimethylsilyl)ethoxy]methyl chloride
(2.86 mL, 1.05 equiv). After the solution was warmed gradu-
ally to rt over 4 h, the reaction was quenched with excess
saturated NH4Cl and the solution was concentrated in vacuo.
The aqueous residue was extracted with CH2Cl2 (4×), and the
combined extracts were dried over MgSO4. The crude product
was purified by short path distillation to give 2.56 g of 3 as a
colorless oil (83%) (bp 112-113 °C/0.1 mm Hg, lit.10 bp 115-
118 °C/0.15 mmHg): 1H NMR δ 0.02 (t, 9H), 0.93 (t, 2H), 3.48
(t, 2H), 5.29 (s, 2H), 7.07 (s, 1H), 7.14 (s, 1H), 7.63 (s, 1H); MS
(FAB) 199 (MH+); HRMS calcd for C9H19N2OSi 199.1267, found
199.1262.
1-[[2-(Tr im eth ylsilyl)eth oxy]m eth yl]-2-(1-pr opyl-1-pen -
tyl)-4-bu tylim id a zole (5a ). A stirred solution of SEM-
imidazole (3) (0.59 g, 3 mmol) in THF (6 mL) under argon at
-78 °C was treated with n-BuLi (1.15 M in hexanes, 3.1 mL,
1.2 equiv). After 15 min at -78 °C, 1-bromobutane (0.34 mL,
1.05 equiv) was added. After 10 min the solution was warmed
to rt for 1 h. The reaction was recooled to -78 °C, and the
procedure was repeated three times, each with n-BuLi (3.39
mL, 1.3 equiv) and bromobutane (0.35 mL, 1.1 equiv), after-
ward allowing the reaction to stir at rt for 16 h. The mixture
was then concentrated, diluted with CH2Cl2, washed with
saturated NH4Cl (2×) and water, and dried over MgSO4.
Flash chromatography on silica gel eluting with 15% ethyl
acetate/hexane gave 440 mg of 5a as an oil (40%). Mono- and
dialkylated imidazoles were obtained and recycled. 5a : 1H
NMR δ 0.04 (s, 9H), 0.84 (m, 11H), 1.07-1.31 (m, 6H), 1.31-
1.44 (m, 2H), 1.52-1.78 (m, 6H), 2.51 (t, 2H), 2.74 (m, 1H),
3.41 (t, 2H), 5.10 (s, 2H), 6.68 (s, 1H); MS (DCI) 367 (MH+);
HRMS calcd for C21H43N2OSi 367.3145, found 367.3151. Anal.
Calcd for C21H42N2OSi: C, 68.79; H, 11.54; N, 7.64. Found:
C, 68.73; H, 11.38; N, 7.58.
1-[[2-(Tr im eth ylsilyl)eth oxy]m eth yl]-2-(1-h eptyl-1-n on -
yl)-4-octylim id a zole (5b) was prepared as described for 5a
(19%): 1H NMR δ 0.02 (s, 9H), 0.91 (m, 11H), 1.05-1.49 (m,
30H), 1.71 (m, 6H), 2.58 (t, 2H), 2.78 (m, 1H), 3.49 (t, 2H),
5.18 (s, 2H), 6.74 (s, 1H); 13C NMR δ (selected) 0.00, 15.62,
19.50, 24.22, 25.99, 29.33, 30.88, 33.47, 36.90, 38.99, 67.07,
73.32, 126.24, 133.12, 154.48; MS (ESI) 535.4 (MH+); HRMS
calcd for C33H67N2OSi 535.5023, found 535.5021. Anal. Calcd
for C33H66N2OSi: C, 74.09; H, 12.43; N, 5.24. Found: C, 73.87;
H, 12.25; N, 5.19.
2-(1-P r op yl-1-p en tyl)-4-bu tylim id a zole (6a ). The SEM-
protected imidazole 5a (2.16 g, 5.9 mmol) under argon at rt
was treated with a mixture of TFA (20 mL) and concd HCl
(15 mL). The reaction was heated at 91 °C for 40 h. After
concentration on the rotary evaporator, CH2Cl2 was added and
the aqueous layer was basified with 1 M NaOH, extracted with
CH2Cl2 (4×), and dried over Na2SO4. The crude product was
purified by chromatography on alumina, eluting with 10%
ethyl acetate/hexane, giving 1.09 g of 6a as an oil (79%): 1H
NMR δ 0.76-0.89 (m, 9H), 1.08-1.40 (m, 8H), 1.47-1.71 (m,
6H), 2.50 (t, 2H), 2.72 (m, 1H), 6.60 (s, 1H), 8.8-9.2 (broad,
1H); MS (DCI) 237 (MH+). Hyd r och lor id e: mp 82-84 °C;
1H NMR (D2O) δ 0.72, 0.76, and 0.82 (3 × t, 9H), 0.98 (m, 2H),
1.16-1.37 (m, 6H), 1.59-1.80 (m, 2H), 1.80-1.99 (m, 2H), 2.58
Imidazole 6a exhibited potent reversal of doxorubicin
(DOX) resistance in the HCT116-VM46 cell line by a
factor of 14 over the DOX-sensitive strain and superior
to that of widely used verapamil by a factor of 1.75 at
0.02 mM. In addition, replication of influenza virus was
inhibited by 6a (ED50 ) 0.048 mM (13.3 µg/mL)).5a
Exp er im en ta l Section
Molecusol (hydroxypropylated â-cyclodextrin) was pur-
chased from Pharmatec Inc. (Alachua, FL). Chloroquine and
FITC-dextran were purchased from Sigma Chemical Co. (St.
Louis, MO). All other regents were purchased from Aldrich
Chemical Co. (Milwaukee, WI) and were used as received. THF
was distilled from sodium benzophenone ketyl, and chloroform
was stored over basic alumina. All other dry solvents were
Aldrich anhydrous grade and were used as received. Culture
medium was purchased from GIBCO Laboratories (Grand
Island, NY). Cell lines (RAW and HCT116-VM46) were
obtained from the American Type Culture Collection. NMR
spectra were measured on a Varian Gemini 300 spectrometer
in CDCl3, unless indicated. Microanalyses were carried out
at Oneida Research Services.
P r ep a r a tion of Hyd r och lor id e Sa lts. The compound in
dry ether at 0 °C was treated with 2 equiv of 1.0 M HCl in
ether. After 1 h of stirring, excess HCl and solvent were
removed on the rotary evaporator. The residue was flushed
several times with CH2Cl2 and then dried in vacuo (ca. 0.05
mmHg).
2,4,6-Tr ibu t ylp yr id in e (1a ). Valeronitrile (2.39 g, 28.7
mmol) was added to cobalt(II) chloride hexahydrate (68 mg,
0.01 equiv) under argon at rt. The reaction was cooled to -40
°C and treated with 1-hexyne (1.18, 0.5 equiv), followed by
NaBH4 (22 mg, 0.04 equiv). The mixture was stirred at rt for
24 h and then diluted with hexane. The solution was washed
with water, dried over Na2SO4, and concentrated in vacuo. The
residue was purified by flash chromatography, eluting with
5% ethyl acetate/hexane, to give 713 mg (40%) of 1a as an oil:
1H NMR δ 0.89 (m, 9H), 1.42 (m, 6H), 1.59 (m, 6H), 2.49 (t,
2H), 2.68 (t, 4H), 6.82 (s, 2H); MS (DCI) 248 (MH+): HRMS

4682 J . Org. Chem., Vol. 61, No. 14, 1996
Dubowchik et al.
(t, 2H), 2.97 (m, 1H), 6.98 (s, 1H). Anal. Calcd for C15H29N2-
Cl: C, 66.03; H, 10.71; N, 10.27. Found: C, 65.66; H, 10.43;
N, 10.00.
CH2Cl2/methanol, to give the product as an oil (1.94 g, 60%):
1H NMR δ 0.92 (q, 6H), 1.38 (m, 4H), 1.60 (m, 2H), 1.75 (m,
6H), 2.44 (m, 2H), 2.52 (m, 2H), 2.59 (t, 2H), 3.66 (t, 2H); MS
(DCI) 235 (MH+); HRMS calcd for C15H27N2 235.2174, found
235.2166. Hyd r och lor id e: 1H NMR δ 0.94 (q, 6H), 1.40 (m,
4H), 1.68 (m, 2H), 1.84 (m, 6H), 2.51 (brt, 2H), 2.74 (brt, 2H),
2.95 (t, 2H), 3.85 (t, 2H). Anal. Calcd for C15H27N2Cl: C,
66.52; H, 10.05; N, 10.34. Found: C, 66.84; H, 10.08; N, 10.19.
Meth yl 2-Oxoocta n oa te (11a ). A stirred suspension of
sodium hydride (278.7 mg, 11.61 mmol) in THF (5 mL) under
argon at 0 °C was treated with methanol (0.11 mL, 0.25 equiv)
and gradually allowed to warm to rt. To this was added
dimethyl oxalate (1.247 g, 0.9 equiv) followed by methyl
heptanoate (1.523 g, 0.9 equiv). The mixture was heated at
reflux for 3 h and cooled to rt, and the reaction was carefully
quenched with water (5 mL). THF was removed in vacuo, and
the residue was dissolved in acetic acid (15 mL) and treated
with concd HCl (5 mL). The solution was heated at reflux for
6 h and then poured into ice water. The oily residue was
extracted with ethyl acetate (3×), and the organic phase was
dried over MgSO4 and concentrated in vacuo. The residue in
methanol (20 mL) was treated with a catalytic amount of
p-toluenesulfonic acid (15 mg) and the solution heated at reflux
for 2 h. The mixture was concentrated in vacuo, diluted with
ethyl acetate, washed with 20% K2CO3 (3×), and dried over
Na2SO4. Evaporation of the solvent and chromatography on
silica gel, eluting with 5% ethyl acetate/hexane, gave 0.60 g
(33%) of an oil: 1H NMR δ 0.87 (t, 3H), 1.27 (m, 6H), 1.58 (m,
2H), 2.79 (t, 2H), 3.85 (s, 3H); MS (DCI) 173 (MH+); HRMS
calcd for C9H17O3 173.1178, found 173.1176. Anal. Calcd for
C9H16O3: C, 62.77; H, 9.36. Found: C, 62.52; H, 9.31.
Meth yl 2-oxoh exa n oa te (11b) was prepared as described
for 11a : 1H NMR δ 0.89 (t, 3H), 1.33 (m, 2H), 1.58 (m, 2H),
2.81 (t, 2H), 3.83 (s, 3H); MS (DCI) 145 (MH+). Anal. Calcd
for C7H12O3: C, 58.32; H, 8.39. Found: C, 58.28; H, 8.32.
Meth yl 2,2-Diflu or oocta n oa te (12a ). To a stirred solu-
tion of methyl 2-oxooctanoate (11a ) (107.7 mg, 0.63 mmol) in
CH2Cl2 (1.5 mL) under nitrogen at 0 °C was added (diethyl-
amino)sulfur trifluoride (DAST) (0.091 mL, 1.1 equiv), drop-
wise over 30 s. The reaction mixture was allowed to warm to
rt, and was stirred overnight. Water (5 mL) was then added,
and the pH was adjusted to ca. 7 by the addition of small
amounts of solid NaHCO3. The mixture was extracted with
CH2Cl2 and the organic phase dried over Na2SO4 and concen-
trated in vacuo. The residue was chromatographed on silica
gel, eluting with 10% ethyl acetate/hexane to give the product
as a thick oil (83.7 mg, 69%): 1H NMR δ 0.85 (t, 3H), 1.25 (m,
6H), 1.42 (m, 2H), 2.03 (m, 2H), 3.84 (s, 3H); MS (DCI) 195
(MH+); HRMS calcd for C9H17O2F2 195.1197, found 195.1194.
Meth yl 2,2-d iflu or oh exa n oa te (12b) was prepared as
described for 12a : 1H NMR δ 0.89 (t, 3H), 1.39 (m, 6H), 2.04
(m, 2H), 3.85 (s, 3H); MS (DCI) 167 (MH+); HRMS calcd for
C7H13O2F2 167.0884, found 167.0883.
2,2-Diflu or oocta n oic Acid (13a ). A stirred solution of
methyl 2,2-difluorooctanoate (12a ) (352.6 mg, 1.82 mmol) in
ethanol (5 mL) at 0 °C was treated with KOH (87%, 1.17 g, 10
equiv). The reaction mixture was stirred at rt for 2 days,
concentrated, and diluted with water. After acidification to
pH 2 with 10% HCl, the mixture was extracted with hexane.
The organic phase was dried over MgSO4 and concentrated
under reduced pressure to give 272 mg (83%) of crude product
which was used without further purification: 1H NMR δ 0.86
(t, 3H), 1.27 (m, 6H), 1.47 (m, 2H), 2.06 (m, 2H); MS (DCI)
181 (MH+); HRMS calcd for C8H15O2F2 181.1040, found
181.1037.
2,2-Diflu or oh exa n oic a cid (13b) was prepared as de-
scribed for 13a : 1H NMR δ 0.90 (t, 3H), 1.39 (m, 6H), 2.07 (m,
2H); MS (DCI) 153 (MH+); HRMS calcd for C6H11O2F2 153.0727,
found 153.0725.
N,N-Dioctyl-2,2-d iflu or oocta n a m id e (15a ). A stirred
solution of 2,2-difluorooctanoic acid (13a ) (118 mg, 0.66 mmol)
in CH2Cl2 (2 mL) at 0 °C was treated with P2O5 (140 mg, 1.5
equiv). The reaction mixture was stored at 5 °C for 2 days,
then recooled to 0 °C, and treated with dioctylamine (0.1 mL,
0.5 equiv) and triethylamine (0.05 mL, 0.55 equiv). After 2
days at rt, the mixture was washed with 10% KOH to remove
2-(1-Hep tyl-1-n on yl)-4-octylim id a zole (6b) was prepared
as described for 6a : 1H NMR δ 0.83 (m, 9H), 1.17 (brs, 14H),
1.26 (m, 18H), 1.59 (m, 6H), 2.51 (t, 2H), 2.75 (m, 1H), 6.59 (s,
1H); MS (DCI) 406 (MH+). Anal. Calcd for C27H52N2: C,
80.13; H, 12.95; N, 6.92. Found: C, 79.87; H, 12.88; N, 6.86.
Cyclod extr in Com p lex of 6a . A solution of the hydro-
chloride salt of 6a (19.6 mg, 71.8 mmol) in absolute ethanol
(0.2 mL) was treated with 0.40 mL (ca. 1.25 equiv) of a 37%
w/v solution of Molecusol (hydroxypropylated â-cyclodextrin)
in water. The solvents were evaporated by passing a stream
of N2 over the stirred mixture at rt. When only a glassy solid
remained, residual solvent was removed in vacuo (ca. 0.05
mmHg) over 24 h. The solid was dissolved in water (7.2 mL),
giving a final concentration of 10 mM of 6a . The proton NMR
of 6a in D2O showed the expected changes in chemical shift
upon complexation: δ 2.58 and 2.97 (free) to 2.62 and 3.02
(complexed).23
2-Bu tyl-3a ,4,5,6,7,7a -h exa h yd r o-1H-ben zim id a zole (7).
A stirred solution of valeronitrile (10.0 mL, 95.6 mmol) in ether
(50 mL) and methanol (5.8 mL, 1.5 equiv) and cooled to 0 °C.
HCl gas was bubbled through the solution at 0 °C over which
was passed a continuous stream of N2. After 3 h the solvents
were removed on the rotary evaporator and the residue was
flushed with methanol several times, giving a white foam.
Methanol (30 mL) was added, and the mixture was cooled to
0 °C and treated with 1,2-diaminocyclohexane (11.5 mL, 1
equiv) in methanol (13 mL), dropwise over 45 min. The
reaction was allowed to warm to rt overnight. The solvent
was then evaporated and the residue basified with 1 M NaOH
(30 mL) followed by extraction with CH2Cl2. The organic phase
was washed with brine (1×), dried over Na2SO4, filtered, and
concentrated. The crude product was purified by flash chro-
matography on silica gel, eluting with 13:1 CH2Cl2/methanol,
to yield 13.79 g (80%) of a white solid: 1H NMR δ 0.89 (t, 3H),
1.33 (m, 6H), 1.55 (m, 2H), 1.74 (d, 2H), 2.15 (d, 2H), 2.11 (t,
2H), 2.88 (m, 2H); MS (DCI) 181 (MH+); HRMS calcd for
C11H21N2 181.1705, found 181.1700. Anal. Calcd for C11H20N2‚
0.5H2O: C, 69.79; H, 11.18; N, 14.80. Found: C, 69.86; H,
10.97; N, 14.69.
2-Bu t yl-4,5,6,7-t et r a h yd r o-1H -b en zim id a zole (8).
A
stirred solution of oxalyl chloride (3.82 mL, 1.1 equiv) in CH2-
Cl2 (45 mL) under argon at -78 °C was treated with DMSO
(6.22 mL, 2,2 equiv) in CH2Cl2 (10 mL), dropwise over 5 min.
After 10 min, imidazoline 7 (7.18 g, 39.8 mmol) in CH2Cl2 (11
mL) was added over 10 min, followed by triethylamine (27.8
mL, 5.0 equiv), dropwise over 10 min. The mixture was stirred
with gradual warming overnight, and then the reaction was
quenched with water. The organic phase was washed with
brine, dried over MgSO4, filtered, and concentrated to give a
yellow solid. The crude product was purified by flash chro-
matography on silica gel using 20:1 CH2Cl2/methanol to give
5.82 g (82%) of a waxy solid: 1H NMR δ 0.88 (t, 3H), 1.46
(m, 2H), 1.67 (m, 2H), 1.78 (s, 4H), 2.53 (s, 4H), 2.67 (t, 2H).
Hyd r och lor id e: 1H NMR δ 0.84 (t, 3H), 1.31 (m, 2H), 1.78
(m, 6H), 2.60 (s, 4H), 2.98 (t, 2H); MS (DCI) 179 (MH+); HRMS
calcd for C11H19N2 179.1548, found 179.1544. Anal. Calcd for
C11H20N2Cl: C, 61.53; H, 8.92; N, 13.05. Found: C, 61.24; H,
8.84; N, 12.85.
1,2-Dibu tyl-4,5,6,7-tetr a h yd r o-1H-ben zim id a zole (9). A
stirred suspension of NaH (60% dispersion in mineral oil, 0.61
g, 1.1 equiv) in THF (50 mL) under argon at 0 °C was treated
with 2-butyl bicyclo imidazole 8 (2.46 g, 13.8 mmol) in THF
(20 mL). The stirred mixture was warmed to rt for 30 min
and then heated to 56 °C. After 1 h iodobutane (1.65 mL, 1.05
equiv) was added and heating was continued overnight. The
reaction was quenched with saturated NH4Cl, and the solution
was concentrated in vacuo. The aqueous phase was extracted
with CH2Cl2 (4×). The combined organic phases were dried
over Na2SO4 and concentrated. The crude product was puri-
fied by flash chromatography on silica gel, eluting with 27:1
(23) Cabral Marques, H. M.; Hadgraft, J .; Kellaway, I. W.; Pugh,
W. J . Int. J . Pharm. 1990, 63, 267-274.

Bilayer Membrane-Active Proton Transporting Agents
J . Org. Chem., Vol. 61, No. 14, 1996 4683
starting material (recycled) and then with 10% HCl. The
organic phase was dried over Na2SO4 and then concentrated
to give 69 mg (quant.) of crude product which was used without
further purification: 1H NMR δ 0.85 (m, 9H), 1.25 (br, 26H),
1.51 (m, 6H), 2.08 (m, 2H), 3.26 (t, 2H), 3.37 (t, 2H); MS (DCI)
404 (MH+); HRMS calcd for C24H48NOF2 404.3704, found
404.3703.
N,N-Dih exyl-2,2-d iflu or oh exa n a m id e (15b) was pre-
pared as described for 15a : 1H NMR δ 0.90 (m, 9H), 1.29 (m,
16H), 1.57 (m, 4H), 2.06 (m, 2H), 3.28 (t, 2H), 3.39 (t, 2H); MS
(DCI) 320 (MH+); HRMS calcd for C18H36NOF2 320.2765, found
320.2758.
15.84 mmol) in THF (10 mL) under argon at 0 °C was treated
with diborane (1.0 M in THF, 32 mL, 2.0 equiv). Over 2 h the
reaction mixture was allowed to warm to rt. After 16 h the
mixture was cooled to 0 °C and the reaction was quenched
with methanol (50 mL). The solvent was evaporated and the
residue dissolved in methanol (150 mL) and heated at 70 °C
for 45 min. The solvent was removed in vacuo and the crude
product purified by flash chromatography on silica gel, eluting
with 2% ethyl acetate/hexane, to give 4.00 g of 18a as a thick
oil (76%):
1H NMR δ 0.86 (t, 6H), 1.22 (m, 12H), 1.31 (m, 4H),
1.43 (s, 6H), 1.63 (ABq, 6H), 1.89 (brs, 3H), 1.98 (s, 2H), 2.32
(ABq, 4H); MS (ESI) 334.3 (MH+). Anal. Calcd for C23H43N:
C, 82.81; H, 12.99; N, 4.20. Found: C, 83.01; H, 12.77; N, 4.03.
2,2-Diflu or otr ioctyla m in e (16a ). A stirred solution of
N,N-dioctyl-2,2-difluorooctanamide (15a ) (112 mg, 0.28 mmol)
in THF (3.5 mL) under argon at 0 °C was treated with diborane
(1.0 M in THF, 2.8 mL, 10 equiv). The reaction mixture was
allowed to warm to rt and was stirred overnight. The mixture
was then cooled in an ice bath, the reaction was carefully
quenched with methanol (15 mL), and the mixture was
concentrated in vacuo, diluted with more methanol (30 mL),
and heated at reflux overnight. After evaporation of the
solvent, the residue was purified by flash chromatography on
silica gel, eluting with 2% ethyl acetate/hexane, giving the
product (105 mg, 97%) as an oil: 1H NMR δ 0.85 (t, 9H), 1.24
(br, 26H), 1.36 (m, 6H), 1.85 (m, 2H), 2.46 (t, 4H), 2.67 (t, 2H);
MS (DCI) 390 (MH+); HRMS calcd for C24H50NF2 390.3911,
found 390.3910. Anal. Calcd for C24H49NF2: C, 73.98; H,
12.67; N, 3.59; F, 9.75. Found: C, 74.02; H, 12.67; N, 3.58; F,
9.81.
2,2-Diflu or otr ih exyla m in e (16b) was prepared as de-
scribed for 16a : 1H NMR δ 0.85 (m, 9H), 1.28 (brs, 14H), 1.39
(m, 6H), 1.88 (m, 2H), 2.47 (t, 4H), 2.69 (t, 2H); MS (DCI) 306
(MH+); HRMS calcd for C18H38NF2 306.2972, found 306.2960.
Anal. Calcd for C18H37NF2: C, 70.77; H, 12.21; N, 4.58; F,
12.44. Found: C, 70.63; H, 12.35; N, 4.46; F, 12.54.
N-(Ad a m a n tylca r bon yl)d ih exyla m in e (17a ). To a solu-
tion of 1-adamantanecarbonyl chloride (5.15 g, 25.9 mmol) in
CH2Cl2 (25 mL) at 0 °C were added dihexylamine (6.1 mL, 1
equiv) and pyridine (6.3 mL, 3 equiv), dropwise over 5 min.
The reaction mixture was allowed to warm to rt. After 2 h
the mixture was washed with 10% HCl and brine. The organic
phase was dried over Na2SO4 and concentrated to give an oil
which was purified by flash chromatography on silica gel,
eluting with 4% ethyl acetate/hexane, to give 6.96 g of 17a as
a thick oil (77%): 1H NMR δ 0.83 (t, 6H), 1.22 (m, 12H), 1.48
(m, 4H), 1.67 (brs, 6H), 1.94 (s, 6H), 1.99 (s, 3H), 3.28 (brt,
4H); MS (ESI) 348.2 (MH+). Anal. Calcd for C23H41NO: C,
79.48; H, 11.89; N, 4.03. Found: C, 79.26; H, 11.71; N, 4.01.
N-(Ad a m a n tylca r bon yl)d ibu tyla m in e (17b) was pre-
pared as described above for 17a (88%): 1H NMR δ 0.87 (t,
6H), 1.30 (m, 4H), 1.48 (m, 4H), 1.65 (brs, 6H), 1.92 (s, 6H),
1.98 (s, 3H), 3.31 (brt, 4H); MS (ESI) 292.2 (MH+). Anal.
Calcd for C19H33NO: C, 78.29; H, 11.41; N, 4.81. Found: C,
77.98; H, 11.29; N, 4.79.
N-(Ad a m a n tylm eth yl)d ibu tyla m in e (18b) was prepared
as described above for 18a (82%): 1H NMR δ 0.88 (t, 6H), 1.30
(m, 8H), 1.44 (s, 6H), 1.62 (ABq, 6H), 1.90 (brs, 3H), 1.97 (s,
2H), 2.33 (ABq, 4H); 13C NMR δ 14.39, 20.89, 28.90, 30.19,
35.06, 37.62, 41.64, 57.21, 69.25; MS (ESI) 278.2 (MH+); HRMS
calcd for C19H35N 277.2769, found. Anal. Calcd for C19H35N:
C, 82.24; H, 12.71; N, 5.05. Found: C, 82.40; H, 12.94; N, 5.09.
Hyd r och lor id e. Anal. Calcd for C19H36NCl: C, 72.69; H,
11.56; N, 4.46; Cl, 11.29. Found: C, 72.68; H, 11.60; N, 4.40;
Cl, 11.34.
N-(Ad a m a n tylm eth yl)p yr r olid in e (18c) was prepared as
described above for 18a (92%): 1H NMR δ 1.48 (d, 6H), 1.61
(m, 6H), 1.67 (m, 4H), 1.89 (brs, 3H), 2.08 (s, 2H), 2.50 (m,
4H); MS (ESI) 220.2 (MH+). Anal. Calcd for C15H25N: C,
82.13; H, 11.49; N, 6.38. Found: C, 82.40; H, 11.63; N, 6.40.
Hyd r och lor id e. Anal. Calcd for C15H26NCl: C, 70.42; H,
10.24; N, 5.48; Cl, 13.86. Found: C, 70.38; H, 10.17; N, 5.21;
Cl, 13.62.
N-(Ad a m a n tylm eth yl)a za cycloh ep ta n e (18d ) was pre-
pared as described above for 18a (94%): 1H NMR δ 1.30 (m,
2H), 1.45 (m, 10H), 1.62 (m, 6H), 1.83 (s, 2H), 1.90 (brs, 3H),
2.34 (m, 4H); 13C NMR δ 27.45, 28.83, 29.10, 35.83, 37.63,
41.19, 59.23, 71.85; MS (ESI) 248.2 (MH+). Anal. Calcd for
C17H29N: C, 82.53; H, 11.81; N, 5.66. Found: C, 82.58; H,
11.79; N, 5.70. Hyd r och lor id e. Anal. Calcd for C17H30NCl:
C, 71.93; H, 10.65; N, 4.93; Cl, 12.49. Found: C, 72.02; H,
10.85; N, 4.90; Cl, 12.49.
[(Ad a m a n tylm eth yl)a m in o]a d a m a n ta n e (20) was pre-
pared as described for 18a and purified by chromatography
on silica, eluting with 20% ethyl acetate/hexane (519.2 mg,
91%): 1H NMR δ 1.43 (d, 6H), 1.54 (d, 6H), 1.60 (m, 12H),
1.91 (brs, 3H), 2.00 (brs, 3H), 2.17 (s, 2H); 13C NMR δ 28.39,
28.77, 29.88, 37.12, 37.58, 39.30, 41.17, 43.05, 53.20, 74.21;
MS (ESI) 300.2 (MH+). Anal. Calcd for C21H33N: C, 84.22;
H, 11.11; N, 4.68. Found: C, 84.04; H, 11.08; N, 4.65.
Hyd r och lor id e. Anal. Calcd for C21H34NCl: C, 75.08; H,
10.20; N, 4.17; Cl, 10.55. Found: C, 75.11; H, 10.38; N, 4.14;
Cl, 10.85.
Ad a m a n tylm eth yl Meth a n esu lfon a te (21). A stirred
solution of adamantylmethanol (784.0 mg, 4.72 mmol) in CH2-
Cl2 (15 mL) at 0 °C was treated with methanesulfonyl chloride
(0.55 mL, 1.5 equiv) and triethylamine (1 mL, 1.5 equiv). After
being warmed to rt, the mixture was stirred for 16 h and then
diluted with ethyl acetate (80 mL). The solution was washed
with 10% citric acid, water, and brine, dried over Na2SO4, and
concentrated in vacuo. The crude product was carried on
without further purification (1.073 g, 93%): 1H NMR δ 1.41
(brs, 6H), 1.55 (m, 6H), 1.88 (brs, 3H), 2.88 (s, 3H), 3.66 (s,
2H); MS (DCI) 243 (MH+). Anal. Calcd for C12H20O3S: C,
58.99; H, 8.25; S, 13.12. Found: C, 59.22; H, 8.40; S, 12.92.
N-(Ad a m a n tylm eth yl)im id a zole (22). A two-phase sys-
tem consisting of adamantylmethyl methanesulfonate (21)
(532.9 mg, 2.18 mmol), imidazole (297.0 mg, 2 equiv), and
Aliquat 336 (25 mg) in toluene (5 mL) and 1 M NaOH (5 mL)
was heated at reflux for 1 week. The organic phase was
separated, washed with water and brine, dried over Na2SO4,
and concentrated. The residue was chromatographed on
alumina, eluting with 25% ethyl acetate/hexane, to give 141.5
mg of 22 (30%) as a waxy solid: 1H NMR δ 1.38 (s, 6H), 1.53
(ABq, 6H), 1.89 (brs, 3H), 3.48 (s, 2H), 6.70 (s, 1H), 6.91 (s,
1H), 7.27 (s, 1H); 13C NMR δ 28.25, 34.26, 36.79, 40.49, 59.87,
121.40, 128.96, 138.85; MS (ESI) 217.1 (MH+). Hyd r och lo-
N-(Ad a m a n tylca r bon yl)p yr r olid in e (17c) was prepared
as described above for 17a (93%): 1H NMR δ 1.67 (s, 6H), 1.76
(m, 4H), 1.93 (s, 6H), 1.99 (brs, 3H), 3.52 (m, 4H); 13C NMR δ
27.84, 28.60, 36.52, 36.93, 38.42, 41.95, 176.77; MS (ESI) 234.1
(MH+). Anal. Calcd for C15H23NO: C, 77.21; H, 9.93; N, 6.00.
Found: C, 77.07; H, 9.95; N, 6.01.
N-(Ada m a n tylca r bon yl)a za cycloh ep ta n e (17d ) was pre-
pared as described above for 17a (89%): 1H NMR δ 1.50 (m,
4H), 1.68 (m, 10H), 1.99 (brs, 9H), 3.51 (brt, 4H); MS (ESI)
262.1 (MH+). Anal. Calcd for C17H27NO: C, 78.11; H, 10.41;
N, 5.36. Found: C, 77.95; H, 10.24; N, 5.29.
[(Ad a m a n tylca r bon yl)a m in o]a d a m a n ta n e (19) was pre-
pared as described for 17a and purified by column chroma-
tography on silica, eluting with 30% ethyl acetate/hexane (1.46
g, 89%): 1H NMR δ 1.62 (m, 3H), 1.70 (m, 9H), 1.77 (s, 3H),
1.90 (s, 6H), 1.94 (m, 3H), 2.03 (brs, 6H), 5.20 (brs, 1H); 13C
NMR δ 29.15, 29.66, 35.59, 36.64, 36.76, 39.61, 40.72, 41.86,
53.22, 194.45; MS (ESI) 314.1 (MH+). Anal. Calcd for C21H31
-
NO: C, 80.46; H, 9.97; N, 4.47. Found: C, 80.30; H, 9.86; N,
4.49.
N-(Ad a m a n tylm eth yl)d ih exyla m in e (18a ). A stirred
solution of N-(adamantylcarbonyl)dihexylamine (17a ) (5.51 g,

4684 J . Org. Chem., Vol. 61, No. 14, 1996
Dubowchik et al.
r id e. Anal. Calcd for C14H21NCl: C, 66.52; H, 8.37; N, 11.08;
Cl, 14.02. Found: C, 66.50; H, 8.59; N, 10.93; Cl, 14.07.
1-Br om oa d a m a n ta n e (23). 1-Adamantanol (5.05 g, 33.2
mmol) was weighed into a 500 mL separatory funnel. HBr
(48% aqueous, 170 mL) was added, and the mixture was
shaken vigorously for 10 min. The solid was collected by
filtration and washed several times with water. The crude
material (6.42 g, 90%) was dried in vacuo and used without
further purification: 1H NMR δ 1.70 (t, 6H), 2.07 (s, 3H), 2.35
(d, 6H); MS (DCI) 215 (MH+). Anal. Calcd for C10H15Br: C,
55.83; H, 7.03; Br, 37.14. Found: C, 55.71; H, 7.05; Br, 36.93.
1,1-Dia d a m a n tyla m in e (25). A solution of 1-bromoada-
mantane (23) (1.39 g, 1.05 equiv) in ether (16 mL), under argon
at rt, was added to silver triflate (1.66 g, 1.05 equiv) in ether
(10 mL). The mixture was protected from light. After being
stirred at rt for 1 h, the reaction mixture was filtered through
Celite in the dark and rinsed with ether. The filtrate was
immediately added to a solution of 1-adamantanamine (0.93
g, 6.14 mmol) and triethylamine (1.5 mL, 1.16 equiv) in CH2-
Cl2 (10 mL). The mixture was stirred at rt for 2 days. The
solution was washed with 10% aqueous KOH (2×) and brine
(2×). The organic phase was dried over MgSO4. After
evaporation of the solvent the crude product was purified by
flash chromatography on silica gel, eluting with 10:1 CH2Cl2/
methanol, to yield 39.9 mg (2.3%) of a solid foam: 1H NMR δ
1.61 (b, 6H), 2.00 (b, 9H); MS (DCI) 286 (MH+); HRMS calcd
(0.002 mM) in CHCl3 (65 mL) was continuously stirred
between two aqueous solutions: a pH 7.4 phosphate buffer
(0.1 M, 35 mL) receiving phase and an unbuffered solution of
NaCl (1 M) in HCl at ca. pH 4 (the source phase). The two
aqueous solutions were continuously stirred using microstir-
rers (VWR Scientific). The pH of the unbuffered source phase
was monitored continuously. When the organic phase con-
tained no proton transport agent, the pH of the source phase
was stable to within (0.05 pH unit over the period of study.
Aqu eou s Acid Extr a ction s. Tr ibu tyla m in e. Tributyl-
amine (366.4 mg) in CHCl3 (30 mL) was shaken vigorously
with 10% HCl (2 × 50 mL). The aqueous layer was separated,
washed with CHCl3, basified with 15% NaOH to pH 11, and
then extracted with CHCl3 (3 × 30 mL). The combined organic
phases were dried over MgSO4 and concentrated in vacuo to
give 113.3 mg (31%) of tributylamine. When this procedure
was repeated using a pH 4 buffer (0.05 M biphthalate) in place
of 10% HCl, 21% of the tributylamine was recovered from the
acid washings.
Tr ip en tyla m in e. Tripentylamine (315.8 mg) in CHCl3 (30
mL) was shaken vigorously with 10% HCl (2 × 50 mL). The
aqueous layer was separated, washed with CHCl3, basified
with 15% NaOH to pH 11, and extracted with CHCl3 (3 × 30
mL). The combined organic phases were dried over MgSO4
and concentrated in vacuo to give 38.0 mg (12%) of tripentyl-
amine. When this procedure was repeated using a pH 4 buffer
(0.05 M biphthalate) in place of 10% HCl, no detectable
tripentylamine was recovered from the acid washings.
In Vitr o Lysosom a l Neu tr a liza tion . RAW cells were
incubated with fluorescein isothiocyanate-dextran (FITC-
dextran) (final concentration 5 mg/mL), which concentrates
selectively in lysosomes, for 24 h at 37 °C. They were then
spun down, washed four times with RPMI medium (without
serum or glutamine), and diluted to 107 cells/mL in RPMI; 106
cells were removed, diluted to 1 mL with RPMI, and loaded
on the flow cytometer (Epics V, Coulter Electronics). The
sample cell was excited at 488 nm and emission measured at
525 nm with a band pass of (5 nm. The amplitude of emission
of fluorescein increases proportionally with pH over the range
of interest. A base line pH (ca. 4.8) was measured, and the
compound of interest was added. The change in lysosomal pH
was measured over time. Ammonium chloride at 10 mM,
which has been shown to raise lysosomal pH to ca. 6.2, was
used as a standard. The viability of treated cells was evaluated
by monitoring the forward angle light scatter and the 90° angle
light scatter in the flow cytometer.
for C20H32
N 286.2535, found 286.2534. Anal. Calcd for
C20H31N: C, 84.15; H, 10.95; N, 4.91. Found: C, 84.04; H,
11.04; N, 4.80.
Mea su r em en t of p Ka s by P r oton NMR. In a 100 mL
pear-shaped flask equipped with
a magnetic stirrer, the
compound under study (ca. 50 mg) in deuterated solvent (3
mL) was vigorously stirred with a solution of KH2PO4 (1 M)
in 1 M HCl (pH 0.4-1) for 10 min. A portion of the organic
phase was removed with a pipet, and the proton NMR was
measured. The sample was returned to the flask, and the
aqueous phase was basified with portions of a solution of NaCl
(1 M) in 1 M NaOH. The procedure was repeated at least eight
times along the pH scale until the aqueous phase had reached
ca. pH 10. The aqueous pH was monitored before and after
stirring so that the difference was never greater than 0.05 unit.
To ensure that the fully protonated substrates did not extract
into the aqueous solution, at the end of the experiment the
mixture was acidified to pH <1 and the aqueous layer was
separated, basified, and extracted with CHCl3. Only in the
case of 22 was any aqueous solubility detected by NMR.
Depending on the nature of the compound, the change in
chemical shift of R, â, and ring protons was plotted against
the buffer pH values. In each case clear, overlapping pKa
curves were generated. The data included in the graphs and
tables represent the signals that showed the largest shifts.
When the influence of counterions other than chloride was
investigated, NaCl and HCl were replaced by the correspond-
ing sodium salt and acid of the counterion of interest.
Su p p or tin g In for m a tion Ava ila ble: Copies of NMR
spectra of compounds 12a and 12b (2 pages). This material
is contained in libraries on microfiche, immediately follows this
article in the microfilm version of the journal, and can be
ordered from the ACS; see any current masthead page for
ordering information.
U-Tu be Exp er im en ts. In a 150 mL U-shaped glass tube,
equipped with a magnetic stirrer, the compound under study
J O960367F