Synthesis and characterization of carbohydrate-based phospholipids

Article abstract of DOI:10.1021/ja025542q

Novel carbohydrate-based phospholipids containing two saturated C₁₂ (dilauroyl ribo-phosphocholine) (DLRPC), C₁₄ (dimyristoyl ribo-phosphocholine) (DMRPC), and C₂₀ (diarachadonyl ribo-phosphocholine) (DARPC) carboxylic aci

Full text of DOI:10.1021/ja025542q

Published on Web 05/01/2002
Synthesis and Characterization of Carbohydrate-Based
Phospholipids
Geoffrey S. Hird, Thomas J. McIntosh,† Anthony A. Ribeiro,‡ and Mark W. Grinstaff*
Contribution from the Departments of Chemistry, Ophthalmology, and Biomedical Engineering,
Paul M. Gross Chemical Laboratory, Duke UniVersity, and Department of Cell Biology,
Duke NMR Spectroscopy Center, and Department of Radiology, Duke UniVersity Medical
Center, Durham, North Carolina 27708
Received January 8, 2002
Abstract: Novel carbohydrate-based phospholipids containing two saturated C12 (dilauroyl ribo-phospho-
choline) (DLRPC), C14 (dimyristoyl ribo-phosphocholine) (DMRPC), and C20 (diarachadonyl ribo-phospho-
choline) (DARPC) carboxylic acid chains were synthesized. The physical properties of the supramolecular
structures formed by these compounds were compared to those formed by their direct glycerol analogues
dilauroyl phosphocholine (DLPC), dimyristoyl phosphocholine (DMPC), and diarachadonyl phosphocholine
(DAPC). Modulated differential scanning calorimetry (MDSC) and X-ray diffraction data indicated that with
chain lengths e14 carbons, the carbohydrate backbone increased the thermal stability of the bilayer below
the phase-transition temperature (Tm) as compared to the glycerol-based lipids. With longer chains (C20),
the bilayer structure was destabilized as compared to glycerol-based lipids. NMR studies of a DMRPC
vesicle dispersion reveal split choline headgroup signals and distinct magnetization transfer effects arising
from the “inner” and “outer” surfaces of the bilayer vesicle. Modulated differential scanning calorimetry
also demonstrated that glycerol- and carbohydrate-based lipids mix, as evidenced by a single intermediate
Tm. In addition, carbohydrate-based lipid/cholesterol mixtures exhibited a decrease in enthalpy with an
increase in cholesterol concentration. Unlike glycerol phospholipids, carbohydrate lipids were resistant to
enzymatic degradation by phospholipase A2 (PLA2).
Introduction
phase contrast agents, dye/fragrance carriers, and as models for
biological membranes.9-19 To optimize for a specific structural
Phospholipids are a major component of all prokaryotic and
eukaryotic membranes.1-3 The chemical structure of the phos-
pholipid dictates the self-assembled bilayer structure formed,
as well as the membrane physical and mechanical properties.4-8
For example, phospholipids possessing a conical shape such as
a lysolipid, with a relatively large headgroup and small mono-
chained tail, tend to pack in spherical micelles containing hy-
drophobic tails on the inside and hydrophilic headgroups on
the outside. Alternatively, if the phospholipid possesses a cylin-
drical geometry, as in the case of a diacyl phosphocholine, then
bilayer structures spontaneously rearrange to form spherically
self-closed liposomes in solution. Liposomes are of interest as
gene transfection agents, drug delivery vehicles, ultrasound
or mechanical property of the bilayer, a number of research
groups are altering the structure of the phospholipid.20-27 Current
modifications of phospholipid structure are primarily limited
(9) Harvie, P.; Wong, F. M.; Bally, M. B. J. Pharm. Sci. 2000, 89, 652-663.
(10) Lasic, D. D.; Barenholz, Y., Eds. Handbook of Nonmedical Applications
of Liposomes; CRC Press: New York, 1996; Vol. 4.
(11) Goldberg, B. B.; Liu, J.; Forsberg, F. Ultrasound: Its Appl. Med. Biol.
1994, 20, 319-333.
(12) Lasic, D. D. In Vesicles; Rosoff, M., Ed.; Marcel Dekker: New York, 1996;
Vol. 62, pp 447-476.
(13) Lasic, D. D. Liposomes in Gene DeliVery; CRC Press: New York, 1997.
(14) Felgner, J. H.; Kumar, R.; Sridhar, C. N.; Wheeler, C. J.; Tsai, Y. J.; Border,
R.; Ramsey, P.; Martin, M.; Felgner, P. L. J. Biol. Chem. 1994, 269, 2550-
2561.
(15) Guillaume-Gable, C.; Floch, V.; Mercier, B.; Audrezet, M.; Gobin, E.; Le
Bolc’h, G.; Yaouanc, J.; Clement, J.; Abbayes, H. D.; Leroy, J.; Morin,
V.; Ferec, C. Hum. Gene Ther. 1998, 9, 2309-2319.
(16) Meager, A., Ed. Gene Therapy Technologies, Applications, and Regulations;
John Wiley & Sons: New York, 1999.
* To whom correspondence should be addressed. E-mail: mwg@
chem.duke.edu; http://www.chem.duke.edu/∼mwg/.
† Department of Cell Biology.
(17) Banerjee, R.; Das, P. K.; Srilakshmi, G. V.; Claudhuri, A. J. Med. Chem.
1999, 42, 4292-4299.
(18) Litzenger, D.; Huang, L. Biochim. Biophys. Acta 1992, 1113, 201-227.
(19) Needham, D.; Lasic, D. D. Chem. ReV. 1995, 95, 2601.
(20) Rui, Y.; Wang, S.; Low, P. S.; Thompson, D. H. J. Am. Chem. Soc. 1998,
120, 11213-11218.
‡ Duke NMR Spectroscopy Center.
(1) Rosoff, M. Vesicles; Marcel Dekker Inc.: New York, 1996.
(2) Hanahan, D. J. A Guide to Phospholipid Chemistry; Oxford University
Press: New York, 1997.
(21) Anderson, V. C.; Thompson, D. H. Biochim. Biophys. Acta 1992, 1109,
33-42.
(3) Cevc, G., Ed. Phospholipids Handbook; Marcel Dekker: New York, 1993.
(4) Israelachvili, J. N. Intermolecular and Surface Forces; Academic Press
Inc.: San Diego, 1992.
(22) Svenson, S.; Thompson, D. H. J. Org. Chem. 1998, 63, 7180-7182.
(23) Zalipsky, S.; Mullah, N.; Hading, J. A.; Gittelman, J.; Guo, L.; Defrees, S.
A. Bioconjugate Chem. 1997, 8, 111-118.
(5) Fuhrhop, J.; Koning, J. Membranes and Molecular Assemblies: The
Synkinetic Approach; The Royal Society of Chemistry: Cambridge, 1994.
(6) Evans, E.; Skalak, R. CRC Crit. ReV. Bioeng. 1980, 3, 181-418.
(7) Tanford, C. The Hydrophobic Effect, 2nd ed.; John Wiley and Sons: New
York, 1980.
(24) Needham, D.; Anyarambhatla, G.; Kong, G.; Dewherst, M. Cancer Res.
2000, 60, 1197-1201.
(25) Menger, F. M.; Wong, Y. J. Org. Chem. 1996, 61, 7382-7390.
(26) Geiger, H. C.; Perlstein, J.; Lachicotte, R. J.; Wyrozebski, K.; Whitten, D.
G. Langmuir 1999, 15, 5606-5616.
(8) Seddon, J. M. Biochim. Biophys. Acta 1990, 1031, 1-69.
9
10.1021/ja025542q CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 5983-5992
5983

A R T I C L E S
Hird et al.
Scheme 1. Synthetic Scheme for
Bis-(2,3-lauroyl)-1-methoxy-5-(phosphocholine)-ribose (DLRPC),
Bis-(2,3-myristoyl)-1-methoxy-5-(phosphocholine)-ribose (DMRPC),
and Bis-(2,3-arachadonyl)-1-methoxy-5-(phosphocholine)-ribose
(DARPC)a
Figure 1. Chemical structures of dilauroyl phosphocholine (DLPC),
dimyristoyl phosphocholine (DMPC), diarachadonyl phosphocholine (DAPC),
dilauroyl ribo-phosphocholine (DLRPC), dimyristoyl ribo-phosphocholine
(DMRPC), and diarachadonyl ribo-phosphocholine (DARPC).
to the hydrophilic head and hydrophobic tail groups.25,28-30
While nonglycerol-based phospholipids such as sphingolipids2
occur in nature, synthetic structural variations or complete
substitutions of the backbone are less common with many of
the chemical modifications retaining the basic glycerol three-
carbon unit (e.g., cyclopentane-based phospholipids)31 and the
phosphonolipids.32,33 These modified phospholipids exhibit
different physical properties from their glycerol-based analogues,
indicating the importance of backbone on bilayer structure. We
are interested in the effect on vesicle structure and mechanical
property when ribose is substituted for the conventional glycerol
backbone.
We previously showed that the carbohydrate-based phospho-
lipid, dilauroyl ribo-phosphocholine (DLRPC), possesses a
greater phase-transition temperature and enthalpy, as compared
to dilauroyl phosphocholine (DLPC).34 In this study, we selected
three chain lengths spanning from C12 to C20 to characterize
and develop the structure-property relationships for this class
of phospholipids. Herein we report the synthesis and physical
properties of dimyristoyl ribo-phosphocholine (DMRPC) and
diarachadonyl ribo-phosphocholine (DARPC), and discuss the
physical property relationships between these new carbohydrate-
based C12, C14, and C20 phospholipids and the corresponding
glycerol analogues (Figure 1).
a Key: (a)TrCl, C5H5N, 3 h, 120 °C; (b) DCC, DMAP, lauric/myristic/
arachadonic acid, DMF, 48 h, 60 °C; (c) acetic acid, H2O, 12 h, 50 °C; (d)
2-cyanoethyl-N,N′-diisopropylchlorophosphoramidite, DiPEA, CH2Cl2, 2 h,
22 °C; (e) (i) tetrazole, choline chloride, 22 °C, 7 h, (ii) I2, ACN, 3 h, 22
°C; (f) TEA (aq), 3 h, 22 °C.
or 9, respectively. These intermediates were purified on a silica
gel column and then immediately dissolved in aqueous acetic
acid to remove the trityl protecting group. Compounds 4, 7,
and 10 were subsequently purified by silica gel column
chromatography with an overall yield of 88, 49, and 36% from
2, respectively. The synthesis of compounds 5, 8, and 11 was
accomplished by first reacting 4, 7, or 10 with 2-cyanoethyl-
N,N′-diisopropylchlorophosphoramidite followed by addition of
choline chloride in the presence of tetrazole. The phosphorus-
(III) compound was subsequently oxidized to phosphorus(V)
by I2. Finally, the cyanoethyl protecting group was removed
by dissolving the mixture in 0.14 M (aq) TEA and stirring for
3 h at room temperature. Compound 5 or 8 was isolated after
alumina (72/28 chloroform/methanol - 69/27/4 chloroform/
methanol/H2O), Sephadex G-10 size exclusion (50/50 chloroform/
methanol), and reverse-phase C-18 Sep Pak chromatography
(10/90 methanol/chloroform). The overall yield for these three
steps (d-f) was 33 and 22%, respectively. Because 11 possesses
longer chains, its purification was slightly different; 11 was
dissolved in benzene, and the impurities were removed by
filtration. The benzene solution was concentrated to a minimum
before layering with MeOH and refrigerating overnight. The
resulting precipitate was filtered from solution and purified using
a G-10 size exclusion column in 50/50 CHCl3/MeOH. Finally,
a neutral alumina column (72/28 CHCl3/MeOH - 69/27/4
CHCl3/MeOH/H2O) was used to isolate 11 in 5% yield.
Supramolecular Structures. DLRPC, DMRPC, and DARPC
self-assemble into liposome-like structures in aqueous solution.
For example, a light micrograph of DMRPC hydrating is shown
in Figure 2 (bar ) 50 microns). To obtain this micrograph,
DMRPC is first deposited on a slide, and water is added. After
incubation at 40 °C, vesicle structures bud off the solid, and
these myelin figures form rapidly as the lipid rehydrates.
Vesicles, most likely multilamellar, are released from the solid
and are the spherical structures present in solution. Such struc-
tures are termed carbohydro-liposomes or carbohydrosomes.34
The possible formation of micelles cannot be ruled out, since
such structures are not observable at this resolution.
Results and Discussion
Synthesis of Carbohydrate-Based Phospholipids (DLRPC,
DMRPC, and DARPC). Methyl-2,3-di-O-lauroyl-â-D-ribo-5-
phosphocholine, 5 (DLRPC), methyl-2,3-di-O-myristoyl-â-D-
ribo-5-phosphocholine, 8 (DMRPC), and methyl-2,3-di-O-
arachadonyl-â-D-ribo-5-phosphocholine, 11 (DARPC), were
synthesized as shown in Scheme 1. The first step involved
protecting the primary hydroxide of 1 using trityl chloride.
Purification of 2 was accomplished by silica gel column
chromatography in 87% yield with an eluent of 3% methanol/
chloroform. Next, a DCC coupling with DMAP and lauric/
myristic/arachadonic acid in DMF afforded compounds 3, 6,
(27) Guillod, F.; Greiner, J.; Riess, J. G. Chem. Phys. Lipids 1995, 78, 149-
162.
(28) Menger, F. M.; Chen, X. Y. Tetrahedron Lett. 1996, 37, 323-326.
(29) Wang, G.; Hollingsworth, R. I. J. Org. Chem. 1999, 64, 4140-4147.
(30) Sommerdijk, N. A. J. M.; Hoeks, T. H. L.; Synak, M.; Feiters, M. C.;
Nolte, R. J. M.; Zwanenburg, B. J. Am. Chem. Soc. 1997, 119, 4338-
4344.
(31) Hancock, A. J. Methods Enzymol. 1981, 72, 640-672.
(32) Engel, R. Chem. ReV. 1977, 77, 349-367.
Alternatively, vesicles can be prepared using a high-pressure
extrusion technique. This technique affords vesicle sizes from
0.05 to 15 µm. Extruding DMRPC through a 200 nm polycar-
(33) Bittman, R. In Lipid Synthesis and Manufacture; Gunstone, F. D., Ed.;
CRC Press: Boca Raton, FL, 1999; pp 185-207.
(34) Hird, G. S.; McIntosh, T. J.; Grinstaff, M. W. J. Am. Chem. Soc. 2000,
122, 8097-8098.
9
5984 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Synthesis of Carbohydrate-Based Phospholipids
A R T I C L E S
Figure 4. Phase-transition temperature (Tm, °C) vs lipid chain length.
Figure 5. Phase-transition enthalpy (kcal/mol) vs lipid chain length.
Figure 2. Light micrograph of rehydrating DMRPC at 40 °C (bar ) 50
µm). The arrow points to a myelin figure.
Figure 3. Stability of DMRPC carbohydro-liposomes over time at 22 °C.
bonate filter at 21 °C rendered vesicles with an average particle
size of 197 nm (fwhm ) 69 nm). The vesicles are stable at 22
°C with no significant change in vesicle size for more than 30
days as shown in Figure 3.
Thermal Analysis. The phase-transition temperatures (Tm)
of the bilayers formed by DLRPC, DMRPC, and DARPC were
determined by modulated differential scanning calorimetry
(MDSC). In a typical experiment, 1 mg of lipid and 10 µL of
water were hermetically sealed in an aluminum pan. Phase-
transition temperatures of 15.75 ( 0.05, 29.6 ( 0.3, and 63.5
( 0.6 °C were observed for DLRPC, DMRPC, and DARPC,
respectively (Figure 4). The corresponding transition enthalpies
for DLRPC, DMRPC, and DARPC are 3.3, 6.1, and 5.1 kcal/
mol, respectively (Figure 5). DSC traces for DMPC and
DMRPC are shown in Figure 6. The phase transitions for the
carbohydrate-based lipid systems were all broader (∼10 °C;
fwhm ) 4.1 °C) than the phase transitions observed for the
corresponding glycerol-based lipids (∼2 °C; fwhm ) 0.4 °C).
The pretransition for DMPC is observed in the heat flow trace.
The broad phase transition of DMRPC collected on heating as
observed in the heat flow trace has a maximum at approximately
30 °C with a shoulder at approximately 4 °C lower in
temperature. These two broad peaks are slightly more separated
(7 °C) on the cooling trace. Three possible explanations for the
broad phase transition are the existence of extra phases, a lack
Figure 6. DSC trace of DMPC and DMRPC.
in cooperativity between the lipid molecules, or impurities.
HPLC analysis showed only a single peak for the carbohydrate
lipid, suggesting the first two possibilities are more likely.
DLRPC exhibits a Tm of 15.75 °C, 16 °C higher than the Tm
of the glycerol analogue DLPC (-1.1 ( 0.8 °C) (see Figure
4). A melting temperature of 29.6 °C is observed for DMRPC.
This Tm is approximately 6 °C higher than the phase transition
of DMPC (23.5 °C).42 The increase in Tm for both DLRPC and
(35) McIntosh, T. J.; Simon, S. A.; MacDonald, R. C. Biochim. Biophys. Acta
1980, 597, 445-463.
(36) McIntosh, T. J. Biophys. J. 1980, 29, 237-246.
(37) Weinstein, J. N.; Ralston, E.; Leserman, L. D.; Klausner, R. D.; Dragsten,
P.; Henkart, P.; Blumenthal, R. In Liposome Technology; Gregoriadis, G.,
Ed.; CRC Press: Boca Raton, 1984; Vol. 3, pp 183-204.
(38) Yuan, W.; Berman, R. J.; Gelb, M. H. J. Am. Chem. Soc. 1987, 109, 8071-
8081.
(39) Deems, R. A.; Dennis, E. A. Methods Enzymol. 1981, 71, 703-710.
(40) Kawana, M.; Kuzuhara, H.; Emoto, S. Bull. Chem. Soc. Jpn. 1981, 54,
1492-1504.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 5985

A R T I C L E S
Hird et al.
DMRPC indicates a more stable state below the Tm as compared
to the glycerol analogues. Increased bilayer stability from
replacing glycerol with ribose is a likely consequence of
enhanced carbohydrate-carbohydrate and chain-chain interac-
tions within the bilayer. The decrease in Tm difference between
the carbohydrate and glycerol systems as the chain length
increases from C12 to C14 is consistent with increased hydro-
carbon chain packing reducing the contribution from the ribose
in the bilayer. Interestingly, DARPC possesses a melting
temperature of 63.5 °C which is slightly below the phase-
transition temperature of DAPC (66 °C).42 For this C20 analogue,
DARPC, the similarity in Tm to DAPC is not entirely unexpected
since tail-tail interactions are a major contributor to Tm when
the tail length is long.
Figure 7. Phase-transition temperature (Tm, °C) vs DMPC/DMRPC mol
% ratios.
other can be investigated with MDSC. If two lipids mix without
forming separate domains, then a single intermediate peak is
observed between their respective phase-transition temperatures.
If mixing is incomplete, two separate phase-transition temper-
atures corresponding to the two fractions are observed. Although
DMPC and DMRPC as well as DLPC and DLRPC possess
chemically and structurally different backbones, a single mixed
bilayer is formed. For example, lipid mixtures with a single
transition temperature between 29 (100% DMRPC) and 23.5
°C (100% DMPC) are formed by simply varying the DMPC/
DMRPC ratios, illustrating that DMRPC mixes with DMPC to
form mixed lipid bilayers (Figure 7).
At the phase-transition temperature, the bilayer transforms
from the gel to disordered liquid crystalline phase in glycerol-
based phospholipids such as DLPC, DMPC, and DAPC. This
endothermic process is observed as two discrete events. A small
pretransition peak, corresponding to the conversion from a
lamellar gel (Lâ′) phase to a rippled gel (Pâ′) phase, occurs below
a larger sharp main transition peak corresponding to conversion
of the Pâ′ phase to the liquid crystalline LR phase.43 The
temperature at which the pretransition exists as well as the
enthalpy of the transition both change as a function of tail length.
This presence of two well-defined phase transitions is not
observed for the carbohydrate-based lipids, as discussed earlier.
X-ray Diffraction Studies. To further characterize the bilayer
structures formed by DLRPC, DMRPC, and DARPC, X-ray
diffraction data were collected. DLRPC and DMRPC exhibited
similar phases below and above the Tm. For DLRPC at 10 °C,
patterns with wide-angle reflections at 7.7, 6.1, 4.9, 4.6, and
3.9 Å were observed. For DMRPC at 22 °C, X-ray patterns
consisted of several sharp wide-angle reflections at 7.74, 6.2,
4.1, and 3.9 Å.
The relationship between the enthalpy of phase transition and
the hydrocarbon chain length for the glycerol-based phospho-
lipids is shown in Figure 5. The literature values for the glycerol-
based phospholipids (b) as well as our experimentally deter-
mined data for the glycerol-based phospholipids (×), and for
the novel carbohydrate-based (]) phospholipids, are also shown
in Figure 5. The enthalpy of DLRPC is 3.3 ( 0.3 kcal/mol and
is approximately double the enthalpy observed for DLPC (1.5
kcal/mol). The enthalpy of transition for DMRPC is 6.1 ( 1.0
kcal/mol and is similar to the enthalpy for DMPC (measured
5.6 ( 1.1 kcal/mol; lit. 5.9 ( 0.6 kcal/mol).42 As the hydro-
carbon chain length increases from C12 to C14, the phase-
transition temperatures and enthalpies of the carbohydrate-based
lipids approach those of the glycerol-based lipids. This is
consistent with chain-chain interactions becoming more im-
portant as chain length increases, reducing the influence of
backbone modification on enthalpy. The Tm’s of DAPC and
DARPC are similar, but the enthalpy of the carbohydrate-based
DARPC is lower than that of DAPC. The enthalpy of DAPC is
12.6 ( 0.7 kcal/mol (lit. ) 11.9 kcal/mol),42 while DARPC,
the carbohydrate analogue, possesses an enthalpy of 5.1 ( 1.5
kcal/mol. The lower enthalpy for DARPC indicates the carbo-
hydrate backbone is destabilizing the bilayer membrane, and
this likely reflects a decrease in the collective order of the
bilayer. The interaction of cholesterol with a bilayer affords a
similar consequence, where the Tm remains constant but the
enthalpy decreases with increased mol % cholesterol.44
These X-ray data are consistent with hydrocarbon chains
crystallized in the plane of the bilayer;45 this is commonly called
an Lc phase. Such crystalline hydrocarbon chain packing is not
typically observed with glycerol-based phospholipids below their
phase-transition temperature.42 This crystalline chain packing
indicates a stable bilayer arrangement within the carbohydrate-
based vesicle below the phase-transition temperature. The
increase in Tm and enthalpy observed for DLRPC and DMRPC
as compared to DLPC and DMPC also supports this conclusion.
X-ray diffraction patterns of DARPC at 22 °C reveal the effect
of longer chains on bilayer packing. Wide angle reflections are
observed at 4.1 and 3.8 Å. These data indicate a gel phase
similar to that of DAPC instead of an Lc phase as seen in
DLRPC and DMRPC. The bilayer formed by DARPC is similar
to the bilayers formed by glycerol-based phospholipids. Because
the phase transition is from a gel-liquid crystalline phase, the
enthalpy is expected to be lower as compared to the change
from a Lc-liquid crystalline phase.42 The enthalpy of transition
for DARPC is approximately one-half that exhibited for the
glycerol-based DAPC (see Figure 8). This suggests that the
carbohydrate backbone is likely destabilizing the gel state
relative to the glycerol analogues by disrupting the bilayer or
affording a decrease in cooperativity between the long-chain
phospholipids in the bilayer.
In nature, cell membranes contain a mixture of different
phospholipids. The ability of phospholipids to mix with each
(41) Browning, J. L. In Liposomes: From Physical Structure to their Applica-
tions; Knight, C. G., Ed.; Elsevier/North-Holland Biomedical Press: New
York, 1981; Vol. 47, pp 189-242.
(42) Marsh, D. CRC Handbook of Lipid Bilayers; CRC Press: Boca Raton,
FL, 1990.
(44) McMullen, T. P. W.; Lewis, R. N. A. H.; McElhaney, R. N. Biochemistry
1993, 32, 516-522.
(45) Blaurock, A. E.; McIntosh, T. J. Biochemistry 1986, 25, 299-305.
(43) Silvius, J. R. Chem. Phys. Lipids 1991, 57, 241-252.
9
5986 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Synthesis of Carbohydrate-Based Phospholipids
A R T I C L E S
Figure 8. 600 MHz proton NMR spectra of DMRPC vesicles in D2O at
39 °C. The expanded spectra in the insets show the splitting for the N+-
(CH3)3, OMe, and ribose H-3 resonances. X indicates residual HOD or
impurity resonances.
NMR Studies of the Bilayer. Figure 8 shows the NMR
resonances at 600 MHz of the choline, ribose, and acyl protons
of a freshly sonicated vesicle bilayer dispersion of DMRPC in
D2O. The NMR assignments for this sugar-based phosphocho-
line were derived from 2-D COSY from the assignments for
methyl-2,3-di-O-lauroyl-â-D-ribo-5-phoshocholine in CD3CN,34
and from the literature for glycerol-based phosphocholine lipids
such as egg phospatidylcholine.46-51 The terminal choline
N+(CH3)3 signal for the DMRPC vesicles is seen to be split
into two resonances about 11 Hz apart, which is identical to
the splitting observed for the terminal choline N+(CH3)3 signal
of unilamellar, sonicated egg phosphocholine vesicles at 600
MHz field.48 This splitting of the N+(CH3)3 headgroup signal
of phosphocholine (and other glycerol-based phospholipids)
arises from the headgroup signals in both the interior and the
exterior monolayer of the phosphocholine liposome bilayer.46-49
In addition to the splitting of the terminal N+(CH3)3 signal,
the ribose OMe and three individually resolved ribose reso-
nances of the DMRPC vesicles are also split into two signals.
The line shapes of the ribose signals are very similar to that of
the N-terminal headgroup. In each case, the lowfield signal
attributed to the “outer” layer is more intense and narrow, while
the upfield signal associated with the “inner” layer is less intense
and broad. Analysis of the integrated intensities gives a ratio
of 1.45:1.00 for the “outer” to “inner” DMRPC signals. The
chemical shift splittings for the ribose H-1, H-2, H-3, and OMe
signals are ∼32, 24, 21, and 20 Hz, respectively, or ap-
proximately a factor of 2 more pronounced than that of the
terminal choline N+(CH3)3 signal. The choline CH2OP and
ribose H-4 protons resonate coincidentally as a broad, over-
lapped signal. The choline CH2N+, the R, â and main methylene
moieties, and the ω methyl group of the hydrophobic myristoyl
chains appear as single, broad resonances for our ribose-based
vesicle dispersion. Two previous papers report split choline
CH2N+, CH2OP, acyl CH2, and ω methyl groups for dipalmi-
toyl- and dimyristoyl-phosphatidyl choline vesicles,52,53 but, in
Figure 9. Expanded two-dimensional 1H-1H COSY spectra at 600 MHz
of DMRPC vesicles in D2O at 39 °C. X indicates residual HOD or impurity
resonances. The cross-peaks from DMRPC molecules in the “inner” and
“outer” layer are indicated with an i and ï.
general, most of the glycerol-based liposomes show broad
unresolved choline CH2N+, choline CH2OP, and acyl chain
resonances.52,54
On the basis of the 1H NMR spectrum of DLRPC in
CD3CN, the ribose H-5 protons are expected to resonate near
3.8 ppm. The NMR spectrum of the DMRPC vesicles shows
no detectable resonances in the 3.8-3.9 ppm region, indicating
that the H-5 signals are too broad to be observed and the
backbone region near H-5 is relatively immobile.
The ribose-based liposomes were sufficiently long-lived for
a 2-D NMR experiment. The expanded 2-D COSY spectrum
shown in Figure 9 manifests the coherence pathways of
magnetization transfer in the lipid resonances. The terminal
N+(CH3)3 and OMe signals are resonances with singlet character
and show no off-diagonal COSY cross-peaks. They are split
due to the contributions from DMRPC molecules in the “outer”
and “inner” layers of the carbohydro-liposome. The ribose
anomeric H-1 resonance is split into the tall, narrow lowfield
signal at 4.97 ppm arising from DMRPC molecules in the
“outer” layer of the vesicle, and the short, broad upfield signal
at 4.91 ppm arising from DMRPC molecules in the “inner” layer
of the vesicle. A strong and a weak off-diagonal coherence
cross-peak are, respectively, observed from the 4.97 and 4.91
ppm H-1 resonances to ribose H-2 signals at 5.23 and 5.19 ppm,
which correspond to the tall, narrow lowfield and short, broad
upfield H-2 resonances. The 5.23 and 5.19 ppm H-2 signals, in
turn, respectively show a strong and a weak coherence cross-
peak to the tall and short ribose H-3 signals at 5.35 and 5.31
ppm. The two H-3 resonances then exhibit strong and weak
coherence cross-peaks to the ribose H-4, which is not resolved
in the one-dimensional NMR spectrum due to the coincidental
overlap with the choline CH2OP near 4.31 ppm.
(46) Jendrasiak, G. L. Chem. Phys. Lipids 1973, 9, 133-146.
(47) Jendrasiak, G. L.; Smith, R.; Ribeiro, A. A. Physiol. Chem. Phys. Med.
NMR 1991, 23, 1-11.
(48) Jendrasiak, G. L.; Smith, R.; Ribeiro, A. A. Biochim. Biophys. Acta 1993,
1145, 25-32.
The remaining 2-D coherence connectivity is from the choline
CH2OP to the choline CH2N+ signal near 3.69 ppm. Note that
(49) Kostelnik, R. J.; Castellano, S. M. J. Magn. Reson. 1973, 9, 291-295.
(50) McLaughlin, A. C. Magn. Reson. 1982, 49, 246-256.
(51) Bystrov, V. F.; Dubrovina, N. I.; Barsukov, L. I.; Bergelson, L. D. Chem.
Phys. Lipids 1971, 6, 343-347.
(53) Neumann, J. M.; Zachowski, A.; Tran-Dinh, S.; Devaux, P. F. Eur. Biophys.
J. 1985, 11, 219-223.
(54) Sheetz, M. P.; Chan, S. I. Biochemistry 1972, 11, 4573-4581.
(52) Schuh, J. R.; Banerjee, U.; Muller, L.; Chan, S. I. Biochim. Biophys. Acta
1982, 687, 219-225.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 5987

A R T I C L E S
Hird et al.
this connectivity is highly skewed and asymmetric, manifesting
the presence of two overlapped and unresolved “outer” and
“inner” cross-peaks. The 1-D CH2N+ signal of egg phospho-
choline vesicles has previously been resolved into “inner” and
“outer” signals with the addition of anion probes.46-48 In
summary, the COSY data show that the contributions of
coherence transfer are localized to DMRPC molecules within
each bilayer; that is, coherence transfer effects occur in DMRPC
molecules in the inner layer of the vesicle independent of effects
in DMRPC molecules in the outer layer, and vice versa.
Loss of the ribose H-5 proton signal is consistent with a
decrease in the backbone flexibility arising from replacement
of the glycerol backbone of a phospholipid with a constrained
cyclic sugar ring such as ribose. In addition, the spacing between
the choline headgroup and the acyl tails is increased due to the
larger ribose backbone. This, in turn, could lead to larger
differences in the radii of curvature of the inner and outer
monolayers for the carbohydro-liposomes as compared to
liposomes. If this is indeed the case, then the packing effects
that give rise to the difference in the chemical shifts of the
“inner” and “outer” surfaces of the bilayer vesicle could become
more pronounced at the sugar backbone in line with our current
observations. All together, the NMR data suggest that the outer
layer of a DMRPC vesicle is similar to the outer layer of a
DMPC vesicle, while the inner layer is structurally different.
One interpretation of these data is that the ribose backbone and
its inherently restricted conformation have greater difficulty
accommodating the imposed structure of the inner layer, which
has a smaller radius of curvature, as compared to a glycerol
analogue.
Figure 10. Phase-transition enthalpy (kcal/mol) vs mol % cholesterol in
DMRPC vesicles.
is zero at 50 mol % cholesterol for DMRPC. For the glycerol-
based phospholipid, cholesterol packs in the bilayer membrane
with its hydroxyl group at the same level as the carbonyl
groups.59 In this structural arrangement, cholesterol primarily
interacts with the tails and not the headgroup or backbone.
Consequently, substitution of the glycerol by ribose is not
expected to dramatically alter the location of cholesterol within
the bilayer.
Phospholipids possessing tails of 17 carbons or less have a
phase-transition temperature that increases as cholesterol content
increases.44 The phase-transition temperature increases as a
function of cholesterol concentration and affords up to a 17 °C
increase at 40 mol % cholesterol for DMPC. A different effect
of cholesterol on bilayer structure is observed with long-chain
glycerol derivatives. As the chain length increases to greater
than 17, the phase-transition temperature decreases. A mismatch
between the hydrophobic length of the cholesterol molecule and
the length of the hydrophobic core of the lipid molecule
determines if an increase or decrease in Tm is observed.44
Therefore, cholesterol can either stabilize or destabilize a gel
state.
For DMRPC, a slight decrease in phase-transition temperature
is observed with increasing mol % cholesterol. The phase-
transition temperature decreases from ∼30 °C at 0 mol %
cholesterol to ∼27 °C at 35 mol % cholesterol. This result is
unexpected since the tail length is less than 17 carbons. The
decrease in phase-transition temperature indicates a slight
destabilization of the bilayer membrane with incorporation of
cholesterol. This suggests a mismatch between the hydrophobic
length of DMRPC and cholesterol, disrupting the interactions
between DMRPC in the bilayer. The carbohydrate backbone
of our molecule may be altering the effective hydrophobic length
of the bilayer area as compared to DMPC; thus DMRPC
interacts with cholesterol in a manner similar to longer tail chain
glycerol analogues. X-ray data on 20 mol % cholesterol/DMRPC
confirm the presence of the Lc state. A repeat period of 54 Å is
observed with weak reflections at crystal spacings of 3.7 and
4.8 Å.
Cholesterol Interactions with the Bilayer. Cholesterol is
present at varying concentrations in biological membranes and
plays a role in modifying bilayer physical properties.55 The
consequences of adding cholesterol to a bilayer were reviewed
by McMullen et al.,44 and include a broadening and eventual
elimination of the cooperative gel-liquid crystal transition of
the bilayer, a decrease/increase in the orientational order of the
hydrocarbon chains below/above the phase transition, the
expansion/condensation of the gel/liquid-crystalline bilayer, a
decreased gel phase chain tilt angle, an increase/decrease in the
passive permeability of the gel/liquid-crystalline phases, and a
loss of a pretransition at low cholesterol concentrations. Ad-
ditional studies on lipid/cholesterol bilayer mechanical properties
such as critical area strain, tensile strength, and elastic area
compressibility modulus demonstrated that mechanical proper-
ties are also affected by cholesterol. For example, the area
expansion modulus increases with cholesterol content until
reaching a plateau at 55 mol % cholesterol.56
Both the pre- and the main-phase transitions decrease for
glycerol-based phospholipids with the addition of cholesterol.
The pretransition peak disappears when the cholesterol con-
centration is 5-10 mol %. Higher cholesterol concentrations
are necessary to completely abolish the main transition. The
enthalpy of transition for saturated diacyl glycerol-based phos-
pholipids reaches zero regardless of chain length at about 50
mol % cholesterol.44,56-58 A similar trend is observed with
carbohydrate-based lipids. As shown in Figure 10, the enthalpy
The interaction of cholesterol with phospholipids containing
modified backbones has previously been studied.60 Bittman and
Blau investigated the interaction of cholesterol with two different
(57) Ladbrooke, K. D.; Williams, R. M.; Chapman, D. Biochim. Biophys. Acta
1968, 150, 333-340.
(58) Lecuyer, H.; Bervichian, D. G. J. Mol. Biol. 1969, 45, 39-57.
(59) New, R. R. C., Ed. Liposomes: A Practical Approach; Oxford University
Press: New York, 1990.
(55) Yeagle, P. L., Ed. The Biology of Cholesterol; CRC Press Inc.: Boca Raton,
FL, 1988.
(56) Needham, D.; Nunn, R. S. Biophys. J. 1990, 58, 997-1009.
(60) Bittman, R.; Blau, L. Biochemistry 1972, 11, 4831-4839.
9
5988 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Synthesis of Carbohydrate-Based Phospholipids
A R T I C L E S
Conclusions
The carbohydrate-based analogues of DLPC, dilauroyl ribo-
phosphocholine (DLRPC), DMPC, dimyristoyl ribo-phospho-
choline (DMRPC), and DAPC, diarachadonyl ribo-phospho-
choline (DARPC), possess unique chemical and physical
properties. These compounds self-assemble into multibilayer
structures in aqueous solution. Differential scanning calorimetry
reveals larger phase-transition temperatures for DMRPC and
DARPC as compared to DLRPC. This is consistent with an
increase in tail chain length. The shorter tailed carbohydrate-
based lipids such as DLRPC and DMRPC possess both an
increased Tm and an increased enthalpy as compared to those
of their glycerol-based analogue. DARPC, however, exhibits a
similar Tm to DAPC but a lower enthalpy. MDSC studies
indicate that the glycerol- and carbohydrate-based lipid mix as
evidenced by a single intermediate phase-transition temperature.
X-ray diffraction studies on DLRPC and DMRPC indicate a
more ordered Lc phase below the phase-transition temperature.
However, DARPC and DAPC exhibit a gel phase below the
Tm. For carbohydrate-based phospholipids with C20 tails, the
bilayer structure is less ordered as evidenced by the presence
of a gel phase. A larger enthalpy per CH2 group is observed
for DLRPC and DMRPC (which have crystalline packing below
Tm) as compared to DARPC (which has a gel phase below Tm).
As in glycerol-based phospholipids, the length of the tail groups
greatly influence bilayer physical structure and properties. NMR
studies of DMRPC vesicle dispersions reveal split headgroup
and backbone resonances arising from protons either in the outer
or in the inner half of the bilayer, similar to vesicles from
glycerol-based phopholipids. 2-D COSY data show that coher-
ence transfer due to coupling effects occurs in DMRPC
molecules in the inner layer of the vesicle independently of the
coupling effects in DMRPC molecules in the outer layer.
Cholesterol doping studies show that the enthalpy of transition
decreases to zero at 50 mol %, comparable to the effect of
cholesterol with glycerol-based phospholipids. DMRPC, unlike
DMPC, is not a substrate of PLA2. Carbohydrate-based phos-
pholipids, such as those described here, possess several favorable
attributes for evaluating lipid-lipid interactions and supramo-
lecular structure formation. These results may also provide new
insight for tailoring vesicle properties for pharmaceutical,
cosmetic, and other industrial applications.
Figure 11. PLA2 hydrolysis of DMPC and DMRPC.
types of phosphonolipids. They found that if the oxygen
connecting phosphate to the glycerol backbone is replaced with
a carbon, as in DL-2-hexadecoxy-3-octadecoxypropylphospho-
nylcholine, then the lipid is able to interact with cholesterol. If
the phosphonate was attached directly to the sn-3 position,
however, as in DL-3,4-di-octadecoxybutylphosphonylcholine,
then the interaction of cholesterol within the bilayer is sterically
inhibited. With the phosphonate attached directly to the sn-3
position, the headgroup phosphate is forced to face toward the
aqueous face of the bilayer, therefore inhibiting hydrogen
bonding with the terminal hydroxide of cholesterol. When the
molecule is isosteric with conventional glycerol-based phos-
pholipids as in the case of the phosphonate DL-2-hexadecoxy-
3-octadecoxypropylphosphonylcholine, this hydrogen-bonding
interaction remains.
Phospholipase Assays on DMRPC. Enzymes play an
important role in phospholipid metabolism.2,61 In the biological
milieu, phospholipase A2 (PLA2) catalyzes the hydrolysis of
the fatty acid from the sn-2 position of the glycerol backbone
yielding a free fatty acid and a lyso-phospholipid. Glycerol-
based phospholipids such as phosphoethanolamines and phos-
phocholines are known PLA2 substrates.2,62
PLA2 enzymatic activity was recorded using a pH-stat, which
maintains a specific pH by adding base during the reaction
which liberates fatty acid.63 At 40 °C, DMPC was hydrolyzed
with an initial average rate of 6.19 × 10-8 mol/min (std dev
6.72 × 10-10). However, DMRPC was not hydrolyzed as shown
in Figure 11.
Next, we investigated if DMRPC inhibited PLA2 hydrolysis
of DMPC. Using DMPC (5 mM) as our substrate, DMRPC was
added to the reaction mixture at varying concentrations from
0.1 to 1 mM. DMRPC did not inhibit PLA2-catalyzed hydrolysis
of the sn-2 acyl of DMPC. Additional studies using concentra-
tions of DMRPC up to 4 mM DMRPC did not yield a significant
change in reaction rate (6.32 × 10-8 mol/min (std dev 1.36 ×
10-8) vs 6.19 × 10-8 mol/min (std dev 6.72 × 10-10)). These
data indicate that DMRPC is neither a PLA2 substrate nor a
potent inhibitor.
Experimental Section
1
(a) Materials. Instruments. One-dimensional (1-D) H, 13C, and
31P NMR spectra characterizing synthetic intermediates and ribo-
phospholipids were obtained on a Varian Inova 400 MHz spectrometer.
Two-dimensional (2-D) COSY and HMQC spectra leading to full
assignments were obtained on a Varian Unity 500 MHz spectrometer.
Thin-layer chromatography (TLC) was performed on Merck KGA Silica
Gel 60 (230-400 mesh; ASTM; phosphomolybdic acid). Low- and
high-resolution FAB mass spectra were obtained using a JEOL JMS-
SX102A mass spectrometer with 4-nitrobenzyl alcohol as the matrix
and Xe as the ionizing gas. EI and CI(NH3) mass spectra were collected
on a Hewlett-Packard 5988A mass spectrometer. Modulated differential
scanning calorimetry was performed on a TA Instruments 2920
modulated differential scanning calorimeter. Polycarbonate filters were
purchased from Poretics Products. A Lipex extruder was used for the
high-pressure extrusion experiments, and a Brookhaven Instruments
Synthetic lipids such as the cyclopentanoid-based phospho-
lipids have been tested as substrates for PLA2. While a few
cyclopentanoid analogues were not observed to be substrates
for PLA2, the majority of the lipids were hydrolyzed as they
maintain the basic glycerol backbone.64
(61) Cao, Y.; Tam, S. W.; Arthur, G.; Chem, H.; Choy, P. C. J. Biol. Chem.
1987, 262, 16927-16935.
(62) Menashe, M.; Romero, G.; Bitonen, R. L.; Lichtenberg, D. J. Biol. Chem.
1986, 261, 5328-5333.
(63) Reynolds, L. J.; Washburn, W. N.; Deems, R. A.; Dennis, E. A. Methods
Enzymol. 1991, 197, 3-23.
(64) Lister, M. D.; Hancock, A. J. J. Lipid Res. 1988, 29, 1297-1308
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 5989

A R T I C L E S
Hird et al.
Corp. Zeta Plus Potential Analyzer was used for dynamic light scattering
particle sizing. A Shimadzu RF-1501 spectrofluorimeter was used for
leakage assays. Osmometry was performed on an Advanced Wide
Range osmometer, Model 3W2 Advanced Instruments, Inc. (Needham
Heights Mass.). C-18 Sep Pak columns were purchased from Waters
Corp.
Chemicals. Chemicals were purchased from Acros, Sigma/Aldrich,
Kodak, and Alfa Aesar. The oxidizing solution, 0.2 M I2 in THF/Pyr/
H2O, was purchased from Glen Research. Sephadex G-50 was
purchased from Sigma/Aldrich chemicals. All solvents indicated as dry
were distilled in the laboratory over either Na/benzophenone or CaH
with the exception of DMF, which was purchased from Aldrich as
99.5% pure in a Sureseal bottle.
(b) Methods. Modulated Differential Scanning Calorimetry. One
to two milligrams of lipid was hermetically sealed with 9-11 µL of
water in an aluminum pan. The modulation was set to (1.00 °C every
40 s, and the pan was equilibrated at -15 °C. The temperature was
increased at 0.5 °C/min to 70 °C where it was held for 2 min. The
temperature was then reduced to -10 °C and held at this temperature
for 2 min. This heating-cooling cycle was repeated two more times
before the sample was held isothermal at -10 °C for 20 min. The data
collected on the third cycle were analyzed. Integrating under the heat
flow curve of the main peak afforded the enthalpy of the transition.
TA Instruments Universal Analysis computer program was used to
analyze the data.
X-ray Diffraction. Multilamellar suspensions of lipid were formed
by hydrating 3-10 mg of compound in 1-2 mL of water at a
temperature above the phase-transition temperature of the lipid. This
mixture was cycled below the phase-transition temperature three times
with vortexing between each cycle.35 The mixture was centrifuged to
provide a hydrated pellet of multilamellar bilayers. Pellets of DMRPC/
cholesterol mixtures do not easily form usable pellets for X-ray studies.
The pellet was transferred to a sealed quartz-glass X-ray capillary which
was mounted in a temperature controllable chamber on a point-focus
collimator. A stationary anode Jerrel-Ash generator (Jerrel-Ash Div.,
Fisher Scientific Co., Waltham, MA) was used to produce Cu KR
X-radiation.36 Diffraction patterns were obtained using a flat plate film
cassette loaded with Kodak DEF X-ray film. The specimen to film
distance was ∼10 cm with exposure times of 2-6 h. The low angle
reflections were in accordance with Bragg’s law, 2d sin θ ) hλ, where
λ is the wavelength (1.54 Å), d is the repeat period, h is the number of
the diffraction order, and θ is the Bragg angle.
NMR Studies of Bilayer Dispersion. Biophysical NMR studies
exploring the bilayer of a freshly sonicated DMRPC dispersion were
performed using a Varian Inova 600 spectrometer equipped with a Sun
Ultra 5 computer and 5 mm Varian probe. 1-D 1H NMR spectra were
recorded with a spectral window of 5.1 kHz, 8.2 s acquisition time,
and digitized using 84 K points to yield a digital resolution of 0.12
Hz/pt. 2-D 1H-1H COSY spectra were recorded in the absolute value
mode with a 5.1 kHz spectral width, 2 K points, 1 s relaxation delay,
and 64 scans/increment. A total of 512 time increments was collected,
and the data were processed with sine-bell weighting in the two
dimensions before Fourier transformation and symmetrization to give
a final 2-D matrix of 2 K by 2 K points. The Varian COSY sequence
was modified to allow a long, low power transmitter pulse for solvent
signal elimination at a frequency different from the nonselective high
power pulses at the carrier frequency. The methyl-2,3-di-O-myristoyl-
â-D-ribo-5-phosphocholine (DMRPC) lipid bilayers have an ordered
gel to liquid crystalline phase transition near 30 °C. NMR experiments
were carried out at 39 °C on a freshly sonicated vesicle bilayer
dispersion of DMRPC at a concentration of 2 mg/0.6 mL D2O.
Phospholipase A2 Studies. The phospholipase assays were per-
formed on a Metrohm 718 STAT Titrino pH-stat following a procedure
described by Yuan et al.38,39 Phospholipase A2 (PLA2) has a molecular
weight of 30 000 Da and is activated by calcium. Before the assays
were performed, the pH-stat was calibrated at pH 7 and pH 4. The
NaOH (0.01 N) titrant was standardized by titration to pH 8.65 with a
weighed amount of potassium hydrogen-phthalate in 40 mL of DI water.
The lipid in methylene chloride/methanol (98/2) was transferred to a 5
mL round-bottom flask, and the solvent was removed by rotoevapo-
ration. High vacuum then removed trace amounts of the remaining
solvent. The lipid film was rehydrated with 2 mL of an aqueous solution
containing Triton X-100 (40 mM) and CaCl2 (10 mM) and then
sonicated briefly to solubilize the lipids (DMPC; [5 mM]). Next, the
solution was immersed in a temperature-controlled bath at 40 °C and
stirred with a magnetic stir bar. A stream of argon was passed over the
reaction mixture to prevent CO2 absorption. A microelectrode (Metrohm
6.0224.100 pH 1-11/0-60 °C Idrolyte) was inserted through the top
of the reaction vessel along with the dosing tube from the pH-stat.
The reaction mixture was adjusted to a pH of 8.2 by addition of NaOH
(0.1 N). Next, 0.2 units of PLA2 (naja naja venom, Sigma) (0.1 units/
mL) were added, and the reaction was started. The pH was allowed to
drop to 8.0 and held at this value by addition of NaOH. At pH 8, the
fatty acid chains hydrolyzed in the presence of PLA2 are fully ionized,
and therefore the amount of base added as a function of time provides
kinetic data for the rate of the reaction. For the mixed DMRPC and
DMPC experiments, the concentration of DMPC was held constant (5
mM), and the concentration of DMRPC was 0.25, 0.5, 1.0, 2.0, and
4.0 mM.
(c) Synthesis. 5-Trityl-1-â-methoxy-ribose 2. â-Methoxy ribose 1
(0.82 g, 5.01 mmol) and trityl chloride (1.63 g, 5.83 mmol) were
dissolved in 100 mL of dry pyridine and heated to 120 °C for 3 h. The
solvent was evaporated under high vacuum, and the resultant oil was
dissolved in chloroform. The chloroform was washed with water, 0.5
N HCl, and then again with water before being dried with Na2SO4.
Silica gel column chromatography was performed (0-3% methanol in
chloroform) yielding 2 in an 87% yield (1.77 g, 4.40 mmol).40 1H
NMR (CDCl3, 400 MHz, ppm): phenyl protons, 7.19-7.6 (15H,
m); H(1), 4.860 (1H, s); H(3), 4.232 (1H, dd, J ) 6.4 Hz, 4.8 Hz);
H(4), 4.092 (1H, m); H(2), 4.0005 (1H, d, J ) 4.4 Hz); OMe(1), 3.321
(3H, s); H(5), 3.246 (2H, m). 1H NMR data agreed with literature
values.40
Bis-(2,3-lauroyl)-1-methoxy-5-hydroxy-ribose, 4. 2 (1.12 g, 6.80
mmol), lauric acid (4.36 g, 21.75 mmol), and DMAP (2.66 g, 21.75
mmol) were dissolved in 175 mL of dry DMF. Next, DCC (4.49 g,
21.75 mmol) was added, and the reaction was stirred at 60 °C for 48
h to yield 3. Acetic acid (5 mL of 5%) was added to the reaction
vessel, and the white precipitate was then filtered. After removing the
solvent via high vacuum, the resultant oil was dissolved in chloro-
form and washed with 0.5 N HCl, 5% NaHCO3, and water, and then
dried with Na2SO4 before evaporation via high vacuum. Silica gel
column chromatography (10% ethyl acetate in hexane) yielded
crude 3 which was dissolved in 150 mL of aqueous acetic acid and
heated at 50 °C for 12 h. The solvent was removed via high vacuum,
and the oil was azeotroped with toluene. Silica gel chromatography
(10-30% ethyl acetate in hexane) afforded 4 in 41% yield (1.47 g,
2.78 mmol). 1H NMR (CDCl3, 400 MHz, ppm): H(3), 5.337 (1H, t, J
) 5.6 Hz); H(2), 5.231 (1H, d, J ) 4.8 Hz); H(1), 4.882 (1H, s); H(4),
4.195 (1H, m); H(5), 3.859 (2H, m); OMe, 3.410 (3H, s); H(R), 2.329
(m); H(â), 1.607 (m); (CH2)m, 1.256 (m); ω CH3, 0.859 (t, J ) 7.2
Hz). FAB mass spectrometry ((M - H)+ theoretical ) 528.4, observed
) 527.4).
Bis-(2,3-lauroyl)-1-methoxy-5-(phosphocholine)-ribose, 5. 2-Cya-
noethyl-N,N′-diisopropylchlorophosphoramidite (0.75 g, 3.35 mmol)
was added dropwise to a 0 °C solution of 4 (1.47 g, 2.79 mmol) and
DIPEA (0.75 mL, 3.90 mmol) in 50 mL of dry CH2Cl2. After stirring
at 22 °C for 2 h, 1 mL of methanol was added. The solvent was removed
via high vacuum after an additional hour of stirring. The resultant oil
was dissolved in CH2Cl2 and washed with 5% NaHCO3 and water
before being dried with Na2SO4. Choline chloride (0.47 g, 3.35 mmol)
was added to the flask containing the oil and dried under high vacuum
overnight. Tetrazole (0.25 g, 3.62 mmol) was next added with 100 mL
9
5990 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Synthesis of Carbohydrate-Based Phospholipids
A R T I C L E S
of dry acetonitrile, and the reaction mixture was stirred for 7 h. Next,
an oxidizing solution of 0.1 M I2 in THF/Pyr/H2O (approximately 20
mL) was added until I2 remained. The solvent was removed via high
vacuum yielding the intermediate (4.50 g). To remove the cyanoethyl
protecting group, 0.34 g of the intermediate was dissolved in 43 mL
of 0.14 M (aq) TEA and stirred for 3 h. The TEA was removed via
high vacuum, and the remaining water was removed by lyophilization.
Purification of 5 was achieved using alumina chromatography (72/28
chloroform/methanol - 69/27/4 chloroform/methanol/H2O), Sephadex
G-10 size exclusion chromatography (50/50 chloroform/methanol), and
finally C-18 Sep Pak chromatography (10/90 methanol/chloroform).
A white solid 5 was obtained in 33.5% yield for the last three steps
(45.4 mg). 2-D COSY and HMQC spectrum were performed to assign
the resonances observed in the H NMR spectra. H NMR (CD3CN,
500 MHz, ppm): H(3), 5.205 (1H, t, J ) 5.40 Hz); H(2), 5.122 (1H,
dd, J ) 4.80 Hz); H(1), 4.868 (1H, d, J ) 1.6 Hz); H(4), 4.173 (1H,
m); CH2OP, 4.173 (1H, m); H(5), 3.848 (1H, m); H(5), 3.773 (1H, m);
CH2N+, 3.498 (2H, m); OMe, 3.345 (3H, s); N+(CH3)3, 3.144 (9H, s);
H(R), 2.284 (4H, m); H(â), 1.548 (4H, m); (CH2)n, 1.266 (m); ω CH3,
0.872 (t, 6.40 Hz).41 13C NMR (CD3CN, 125 MHz, ppm): CdO,
173.54; CdO, 173.50; C(3), 75.15; C(2), 72.53; C(1), 107.13; C(4),
81.91; CH2OP, 59.97; C(5), 59.53; CH2N+, 67.48; OMe, 55.44; N+-
(CH3)3, 54.80; C(R), 34.32; C(â), 25.401; ω CH3, 12.21; (CH2)n, 30.2
(31P NMR 0.067 ppm). High-resolution FAB mass spectrometry ((MH)+
theoretical ) 694.4659, observed ) 694.4653).
4.190 (1H, m); CH2OP, 4.190 (1H, m); H(5), 3.880 (1H, m); H(5),
3.801 (1H, m); CH2N+, 3.513 (2H, m); OMe, 3.344 (3H, s); N+(CH3)3,
3.141 (9H, s); H(R), 2.282 (4H, m); H(â), 1.542 (4H, m); (CH2)n, 1.261
(m); ω CH3, 0.871 (t, 7.2 Hz). 13C NMR (CD3CN, 100 MHz, ppm):
CdO, 173.62; CdO, 173.35; C(3), 75.12; C(2), 72.17; C(1), 107.15;
C(4), 80.97; CH2OP, 60.40; C(5), 59.20; CH2N+, 67.10; OMe, 55.75;
N+(CH3)3, 54.89; C(R), 34.52; C(â), 25.51; ω CH3, 14.39; (CH2)n, 30.65
(31P NMR 0.595 ppm). High-resolution FAB mass spectrometry ((MH+)
theoretical ) 750.5285, observed ) 750.5255).
Bis-(2,3-arachadonyl)-1-methoxy-5-hydroxy-ribose, 10. 2 (1.02 g,
6.20 mmol), arachadonic acid (6.19 g, 19.81 mmol), and DMAP (2.42
g, 6.20 mmol) were dissolved in 200 mL of dry DMF. DCC (4.09 g,
6.20 mmol) was added dropwise, and the reaction was stirred at 40 °C
for 48 h. Next, 5 mL of 5% acetic acid was added to the reaction vessel,
and the white precipitate was filtered. The resultant solution was
evaporated under high vacuum before being dissolved in CHCl3 and
washed with 0.5 N HCl, 5% NaHCO3, and water. After drying with
Na2SO4, the solution was rotoevaporated to dryness. Silica gel column
purification (9/1 Hex/EtAC) was performed. The product was dissolved
in aqueous acetic acid and stirred at 50 °C for 12 h to remove the trityl
protecting group. The deprotected product was purified by silica gel
column chromatography (9/1 Hex/EtAC - 7/3 Hex/EtAc) to afford
10 in a 36% yield (1.60 g). 1H NMR (CDCl3, 400 MHz, ppm): H(3),
5.338 (1H, t); H(2), 5.235 (1H, d); H(1), 4.887 (1H, s); H(4), 4.197
(1H, m); H(5), 3.867 (2H, m); OMe, 3.415 (3H, s); H(R), 2.332 (m);
H(â), 1.613 (m); (CH2)µ, 1.258 (m); ω CH3, 0.862 (t).
Bis-(2,3-arachadonyl)-1-methoxy-5-(phosphocholine)-ribose, 11.
10 (1.18 g, 2.46 mmol) was dissolved in 45 mL of dry CH2Cl2, and
DIPEA (0.73 mL, 4.18 mmol) was added. The reaction vessel was then
chilled to 0 °C before adding 2-cyanoethyl-N,N ′-diisopropylchloro-
phosphoramidite (0.02 mL, 3.70 mmol). The reaction was stirred for 4
h at 22 °C and subsequently evaporated under high vacuum for 1.5 h.
The crude material was dissolved in dry CH2Cl2 and then washed with
5% NaHCO3 and H2O before being dried with Na2SO4. Choline chloride
(0.49 g, 3.20 mmol) was azeotroped twice with toluene, dried under
high vacuum overnight at 45 °C, and then added to the flask containing
the intermediate. The reaction mixture was then dissolved in 94/6
CH3CN/CH2Cl2, and tetrazole (0.22 g, 3.20 mmol) was added. The
reaction was stirred for 12 h. Oxidation of the P(III) was accomplished
using I2 in THF/Pyr/H2O, and the reaction was stirred for 3 h and then
evaporated under high vacuum. The resultant mixture was dissolved
in 113 mL of H2O containing 2 mL of TEA and stirred for 3 h at 22
°C. The solvent was subsequently removed by lyophilization. To isolate
the product, the reaction mixture was first dissolved in benzene and
filtered. The benzene solution was then concentrated to a minimum
before layering with MeOH and refrigerating overnight. Next, the
resulting product was filtered from solution and purified using a G-10
size exclusion column in 50/50 CHCl3/MeOH followed by a neutral
alumina (70-230 mesh) column (72/28 CHCl3/MeOH - 69/27/4
CHCl3MeOH/H2O) to afford 0.12 g (5.2% yield) of product. 1H NMR
(CD3OD, 400 MHz, ppm): H(3), 5.353 (1H, t, J ) 5.6 Hz); H(2),
5.214 (1H, m); H(1), 4.910 (1H, m); H(4), 4.250 (1H, m); CH2OP,
4.280 (1H, m); H(5), 3.951 (1H, m); H(5), 3.951 (1H, m); CH2N+,
3.624 (2H, m); OMe, 3.390 (3H, s); N+(CH3)3, 3.212 (9H, s); H(R),
2.341 (4H, m); H(â), 1.605 (4H, m); (CH2)m, 1.295 (m); ω CH3, 0.860
(t, 7.2 Hz). 13C NMR (CD3OD, 100 MHz, ppm): CdO, 173.43; Cd
O, 173.18; C(3), 74.83; C(2), 71.81; C(1), 106.64; C(4), 80.60; CH2-
OP, 62.73; C(5), 59.30; CH2N+, 67.11; OMe, 55.507; N+(CH3)3, 54.72;
C(R), 34.32; C(â), 25.12; (CH2)m, 30.41; ω CH3, 14.18 (31P NMR 2.22
ppm). High-resolution FAB MS (MH+) (theoretical ) 918.7099,
observed ) 918.7169).
1
1
Bis-(2,3-myristoyl)-1-methoxy-5-hydroxy-ribose, 7. 2 (1.22 g, 7.46
mmol), myristic acid (5.45 g, 23.86 mmol), DMAP (2.92 g, 23.86
mmol), and DCC (4.93 g, 23.86 mmol) were stirred in 150 mL of DMF
for 48 h at 60 °C to afford 6. The solution was filtered, and the resultant
mixture was purified on a silica gel column with the eluent of 9/1 Hex/
EtAc. Next, the solid was immediately dissolved in aqueous acetic acid
and stirred at 50 °C for 12 h. The product, 7, was subsequently purified
by silica gel column chromotography (eluent 7/3 Hex /EtAc) with an
1
overall yield of 49% for the two steps (2.12 g, 3.63 mmol). H NMR
(CDCl3, 400 MHz, ppm): H(3), 5.339 (1H, t); H(2), 5.233 (1H, d);
H(1), 4.888 (1H, s); H(4), 4.196 (1H, m); H(5), 3.863 (2H, m); OMe,
3.413 (3H, s); H(R), 2.331 (m); H(â), 1.610 (m); (CH2)m, 1.260 (m);
ω CH3, 0.863 (t).
Bis-(2,3-myristoyl)-1-methoxy-5-(phosphocholine)-ribose, 8. Com-
pound 8 was prepared by first reacting 7 (0.64 g, 1.09 mmol) with
2-cyanoethyl-N,N ′-diisopropylchlorophosphoramidite (0.37 mL, 1.64
mmol) and DIPEA (0.33 mL, 1.86 mmol) in 60 mL of dry CH2Cl2 at
0 °C. The reaction was stirred for 2 h at 22 °C under N2. This reaction
mixture was then quenched with 2 mL of MeOH, and then the solvent
was removed by high vacuum. The mixture was dissolved in distilled
CH2Cl2, and then washed with 5% NaHCO3 and water before being
dried with Na2SO4. Dried choline chloride (0.20 g, 1.42 mmol) was
added to the reaction vessel. Next, the resultant mixture was dissolved
in 100 mL of dry CH3CN containing tetrazole (0.10 g, 1.42 mmol),
and then stirred for 7 h. An oxidizing solution of 0.2 M I2 in THF/
Pyr/H2O was added with stirring until the solution remained yellow
indicating that the phosphorus was fully oxidized, yielding the
cyanoethyl protected intermediate. This mixture was again evaporated
and azeotroped with benzene. To remove the cyanoethyl protecting
group, the contents of the reaction vessel were dissolved in 0.134 M
(aq) TEA and stirred for 3 h at room temperature. The TEA was
removed with high vacuum, and the water was removed by lyophiliza-
tion. The crude product was purified by alumina column chromatog-
raphy (72/28 CHCl3/MeOH - 69/27/4 CHCl3/MeOH/H2O), by Sepha-
dex G-10 size exclusion chromatography (50/50 CHCl3/MeOH),
followed by C-18 Sep Pak chromatography (10/90 MeOH/CHCl3).
Compound 8 (0.18 g, 0.25 mmol) was obtained in 22.4% yield for the
last three steps d-f. 2-D COSY and HMQC spectrum were performed
Acknowledgment. This work was supported by the North
Carolina Biotechnology Center (M.W.G.), the American Heart
Association (M.W.G.), the Pharmaceutical Sciences NIH Train-
ing Grant (G.S.H.), NIH grant GM27278 (T.J.M.), and Duke
1
1
to assign the resonances observed in the H NMR spectra. H NMR
(CD3CN, 400 MHz, ppm): H(3), 5.209 (1H, t, J ) 31.2 Hz); H(2),
5.120 (1H, dd, J ) 2.2 Hz); H(1), 4.870 (1H, d, J ) 0.8 Hz); H(4),
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 5991

A R T I C L E S
Hird et al.
University. The Duke NMR Center is supported in part by NIH
NCI P30-CA-14236 (A.A.R.). NMR instrumentation was funded
by NSF, NIH, the North Carolina Biotechnology Center, and
Duke University. The authors would like to thank Jeff Mills,
Dr. David Needham, Dr. Jason Keiper, Dr. Steve Aubuchon,
and Dr. Stephen Lee. M.W.G. also thanks the Pew Foundation
for a Pew Scholars Award in the Biomedical Sciences, the
Dreyfus Foundation for a Camille Dreyfus Teacher-Scholar, and
the Alfred P. Sloan Foundation for a Research Fellowship.
Supporting Information Available: NMR spectra of DLRPC,
DMRPC, and DARPC (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
JA025542Q
9
5992 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002