Phosphonate lipid tubules II

Article abstract of DOI:10.1021/ja012137a

We describe a new chiral tubule-forming lipid in which the C-O-P phosphoryl linkage of the archetypal tubule-forming molecule, 1,2-bis(10,12-tricosadiynoyl)-sn-glycero-3-phosphocholine, "DC(8,9)PC", is replaced by a C-P linkage. Tubule formation with this phosphonate analogue proceeds under the same mild conditions as with DC(8,9)PC and produces similar yields, but synchrotron small-angle X-ray scattering, atomic force microscopy, and optical microscopy show the new tubules to have diameters 1.94 times as great, to be significantly shorter, and to be thinner-walled. A significant portion of the enantiomerically pure chiral phosphonate precipitate is in the form of stable open helices, and these helices are divided almost evenly between left- and right-handed members.

Full text of DOI:10.1021/ja012137a

Published on Web 01/25/2002
Phosphonate Lipid Tubules II
Britt N. Thomas,*,† Christopher M. Lindemann, Robert C. Corcoran,
‡
§
§
§
|
Casey L. Cotant, Janet E. Kirsch, and Phillip J. Persichini
Contribution from the Department of Chemistry, Louisiana State UniVersity,
Baton Rouge, Louisiana 70803, Merck & Company, P.O. Box 2000, Rahway, New Jersey 07065,
Department of Chemistry, The UniVersity of Wyoming, Laramie, Wyoming 82071, and
Department of Chemistry, Allegheny College, MeadVille, PennsylVania 16335
Received September 4, 2001
Abstract: We describe a new chiral tubule-forming lipid in which the C-O-P phosphoryl linkage of the
archetypal tubule-forming molecule, 1,2-bis(10,12-tricosadiynoyl)-sn-glycero-3-phosphocholine, “DC(8,9)PC”,
is replaced by a C-P linkage. Tubule formation with this phosphonate analogue proceeds under the same
mild conditions as with DC(8,9)PC and produces similar yields, but synchrotron small-angle X-ray scattering,
atomic force microscopy, and optical microscopy show the new tubules to have diameters 1.94 times as
great, to be significantly shorter, and to be thinner-walled. A significant portion of the enantiomerically pure
chiral phosphonate precipitate is in the form of stable open helices, and these helices are divided almost
evenly between left- and right-handed members.
I. Introduction
Dimyristoylphosphatidylcholine and dipalmitoylphosphati-
dylcholine analogues in which diacetylene groups have been
inserted midway in the long hydrocarbon tails self-assemble to
yield tubules - stable, hollow cylindrical crystalline vesicles,
as first reported by Yager and co-workers1 in 1984. The S-
enantiomer of the prototypical tubule-forming molecule, 1,2-
bis(10,12-tricosadiynoyl)-sn-glycero-3-phosphocholine, 1, here-
after referred to as “DC(8,9)PC”, is shown in Figure 1.
-4
Tubules made from DC(8,9)PC and analogues in which the
diacetylene moiety is repositioned along the hydrocarbon tails
display a single helical trace upon their exteriors, suggesting
these cylindrical structures are the product of the helical winding
of a phospholipid bilayer ribbon. A remarkable correspondence
between the sense of handedness of this helical trace and
DC(8,9)PC monomer chirality is known: the R-DC(8,9)PC
monomer always produces tubules that have a right-handed
exterior helical trace, while the S- enantiomer always yields
tubules that have a left-handed exterior helical trace. This widely
known correspondence has served as the impetus for a number
of tubule structure theories that link helix handedness to
molecular chirality.5
-12
Another intriguing aspect of tubule
*
To whom correspondence should be addressed.
Louisiana State University.
Merck & Co.
The University of Wyoming.
Allegheny College.
†
Figure 1. (top) DC(8,9)PC. (middle) Compound 2, the “C4” phosphonate
analogue of DC(8,9)PC. (bottom) Compound 3, the “C3” phosphonate
analogue of DC(8,9)PC.
‡
§
|
(
(
(
1) Schoen, P. E.; Yager, P.; Priest, R. G. NATO Sci. Ser., Ser. E 1985, 102-
Polydiacetylenes), 223-232.
2) Yager, P.; Schoen, P. E.; Davies, C.; Price, R.; Singh, A. Biophys. J. 1985,
8, 899-906.
3) Schoen, P. E.; Yager, P. J. Polym. Sci., Polym. Phys. Ed. 1985, 23, 2203-
216.
(
structure is seen in tubules formed in ethanolic solvents; these
tubules are multilamellar, that is, they are made of several evenly
4
2
(
(
(
(
(
4) Yager, P.; Schoen, P. E. Mol. Cryst. Liq. Cryst. 1984, 106, 371-381.
5) de Gennes, P. G. C. R. Acad. Sci. Paris 1987, 304, 259.
6) Helfrich, W.; Prost, J. Phys. ReV. A: Gen. Phys. 1988, 38, 3065-3068.
7) Helfrich, W. J. Chem. Phys. 1986, 85, 1085-1087.
(9) Ou-Yang, Z.; Liu, J. Phys. ReV. Lett. 1990, 65, 1679-1682.
(10) Selinger, J. V.; Schnur, J. M. Phys. ReV. Lett. 1993, 71, 4091-4094.
(11) Selinger, J. V.; MacKintosh, F. C.; Schnur, J. M. Phys. ReV. E: Stat. Phys.,
Plasmas, Fluids, Relat. Interdiscip. Top. 1996, 53, 3804-3818.
(12) Schnur, J. M. Science 1993, 262, 1669-1676.
8) Lubensky, T. C.; Prost, J. J. Phys. II 1992, 2, 371-382.
10.1021/ja012137a CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002 1227

A R T I C L E S
Thomas et al.
spaced coaxially nested cylinders. Interestingly, it has recently
been demonstrated that the tubule interior lamellae may have
either a left- or a right-handed helical sub-structure.1
synthesis and study of DC(8,9)PC phosphonate analogue 3, in
which the polar headgroup has been moved closer to the
stereogenic center contained within the glycerol backbone. This
translation is accomplished through the remoVal of the linking
phosphate oxygen atom, rather than the (-CH2-) replacement
that led to phosphonate analogue 2. The (-CH2-)-for-oxygen
substitution leading to phosphonate analogue 2 resulted in the
lengthening of the DC(8,9)PC three-carbon glycerol backbone
to four contiguous carbon atoms, which led us to conveniently
refer to 2 as the “C4” phosphonate; in the same manner we
refer to the molecule of this study, phosphonate analogue 3, as
the “C3” phosphonate.
3
Tubule dimensions and morphology, particularly their hol-
lowness, have suggested a host of potential technological
applications, ranging from nanofabrication and purification to
medical encapsulation.1
2,14-16
Realization of the full techno-
logical potential of tubules requires the ability to optimize their
morphology for a given application. While tubule length is
moderately controllable over a range of a few microns to
17
hundreds of microns through solvent composition or kinetics
1
8
control, tubule diameter has proven to be insensitive to
regulation.
II. Experimental Section
In previous work,19 we reported a 2-fold increase in tubule
diameter attained through a subtle modification to DC(8,9)PC,
the replacement of the phosphoryl oxygen linking the glycerol
backbone to the phosphocholine headgroup with a methylene
II.a. Synthesis. II.a.1. General Procedures. Methylene chloride
and pyridine were distilled from CaH
Triethyl phosphite was distilled under reduced pressure prior to use.
,2:5,6-Di-O-isopropylidene-D-mannitol, bromotrimethylsilane, 4-(di-
2
and solid KOH, respectively.
1
(
2
-CH2-) group, resulting in DC(8,9)PC phosphonate analogue
. Our synthesis of 2 was motivated by the predicted dependency
methylamino)pyridine, carbon tetrabromide, and choline chloride were
purchased from Aldrich. 10,12-Tricosadiynoic acid was purchased from
Lancaster. All of these materials were used without any further
purification. Silica gel 60 (230-400 mesh) was used for column
of tubule diameter upon chiral packing of the tubule-forming
molecules within the lamellae.1
0-12
We speculated that small
1
13
chromatography. H and C NMR spectra were obtained in CDCl
3
changes in bond lengths and angles near the stereogenic center
might be sufficiently subtle to conserve tubule-forming capacity,
yet large enough by virtue of their placement to affect product
morphology.
solutions at 270 and 67.5 MHz, respectively. The high-resolution mass
spectrum of compound 3 was obtained by the Washington University
Resource for Biomedical and Bio-organic Mass Spectrometry at
Washington University, St. Louis, Missouri.
Concurrent with the 2-fold diameter increase found with
tubules derived from 2 was a halving of the tubule wall
thickness. That is, the number of coaxially nested cylinders
composing the phosphonate tubule wall was found to be half
that found for DC(8,9)PC tubules. Despite the large changes to
tubule diameter and wall thickness, the space between the
coaxial cylindrical lamellae was virtually unchanged. Tubule
length was also unaffected. Comparatively large numbers of
open helices were found in the DC(8,9)PC phosphonate
analogue preparations, that is, helically wound ribbons whose
widths are too narrow to enable fusion of the ribbon edges to
form closed cylinders. Very surprisingly, the open helices and
the fully formed tubules in this system were found to violate
the widely known helical handedness/molecular chirality cor-
respondence; indeed, roughly equal numbers of left-handed and
right-handed helices and tubules were formed from enantio-
merically pure preparations. Helical structures of the unexpected
sense of handedness have subsequently been described in the
study of an enantiomerically pure, multicomponent (water, bile
salt, phosphatidylcholine, and cholesterol) system,2 although
one sense of handedness is found to be dominant in this model
bile system.
II.a.2. (S)-4-Hydroxymethyl-2,2-dimethyl-1,3-dioxolane (4). A
solution of 1,2:5,6-di-O-isopropylidene-D-mannitol (10.5 g, 40 mmol)
in ethyl acetate (160 mL) was cooled to 0 °C (during which time the
mannitol crystallized). The ice bath was removed, and lead tetraacetate
(17.8 g, 40.2 mmol) was added in portions over a 10 min time period.
After an additional 35 min the reaction was filtered. The precipitate
was washed with ethyl acetate, and the combined organics were washed
with brine/saturated sodium bicarbonate (1:1, 2 × 20 mL) and brine
2 4
(20 mL) and were dried (Na SO ). To this solution was added ethanol
(20 mL) followed by portionwise addition of sodium borohydride (5.2
g). After the last addition, acetone (10 mL) was added, and the solution
was diluted with diethyl ether (250 mL) and carefully washed with
1:1 brine/5% aqueous HCl (2 × 50 mL), brine (50%, 50 mL), and
2 4
brine/saturated bicarbonate (50 mL) and was dried (Na SO ) and
concentrated under reduced pressure. The residue was brought up in
methanol (175 mL) which had been saturated with ammonia gas, left
to stand 12 h, and again concentrated under reduced pressure. The crude
material was purified by flash chromatography on silica gel using a
gradient elution from hexane/ethyl acetate (3:1) to ethyl acetate (100%)
to give (S)-4-hydroxymethyl-2,2-dimethyl-1,3-dioxolane, 4 (4.56 g,
1
13
21
8
6.4%), having H and C NMR identical to published spectral data.
II.a.3. (R)-4-Bromomethyl-2,2-dimethyl-1,3-dioxolane (5). A solu-
tion of triphenylphosphine (8.33 g, 31.8 mmol) in CH Cl (15 mL)
0
2
2
was added dropwise over a 9 h period (via syringe pump) to a well-
The success obtained by substituting the DC(8,9)PC linking
phosphoryl oxygen with a (-CH2-) group has motivated the
stirred solution of 4 (4.22 g, 31.9 mmol) and carbon tetrabromide (13.98
g, 42.2 mmol) in CH
was diluted with hexane (200 mL) and washed with aqueous sodium
bicarbonate (5%, 40 mL) and brine (40 mL) and was dried (Na SO
2 2
Cl (8 mL) at room temperature. The reaction
2
4
)
(
13) Thomas, B. N.; Lindemann, C. M.; Clark, N. A. Phys. ReV. E 1999, 59,
3
040-3047.
and concentrated. The crude material was purified by flash chroma-
(
(
(
14) Archibald, D. D.; Mann, S. Chem. Phys. Lipids 1994, 69, 51-64.
tography on silica gel using hexane/ethyl acetate (5:1) to give (R)-4-
15) Baum, R. Chem. Eng. News 1993, August 9, 19-20.
1
bromomethyl-2,2-dimethyl-1,3-dioxolane, 5 (2.15 g, 34.5%).
H
16) Johnson, D. L.; Polikandritou-Lambros, M.; Martonen, T. B. Drug DeliVery
1
996, 3, 9-15.
NMR: δ 4.35 (m, 1H), 4.12 (dd, 1H, J ) 12 Hz, 9 Hz), 3.87 (dd, 1H,
J ) 9 Hz, 6 Hz), 3.45 (dd, 1H, J ) 9 Hz, 6 Hz), 3.33 (dd, 1H, J ) 12
Hz, 9 Hz), 1.45 (s, 3H), 1.35 (s, 3H).
II.a.4. Diethyl [(R)-O-3,4-Isopropylidene-3,4-dihydroxypropyl]
Phosphonate (6). A solution of 5 (2.05 g, 10.5 mmol) in triethyl
(
(
(
(
17) Ratna, B. R.; Baral-Tosh, S.; Kahn, B.; Schnur, J. M.; Rudolph, A. S. Chem.
Phys. Lipids 1992, 63, 47-53.
18) Thomas, B. N.; Safinya, C. R.; Plano, R. J.; Clark, N. A. Science 1995,
2
67, 1635-1638.
19) Thomas, B. N.; Corcoran, R. C.; Cotant, C. L.; Lindemann, C. M.; Kirsch,
J. E.; Persichini, P. J. J. Am. Chem. Soc. 1998, 120, 12178-12186.
20) Zastavker, Y. V.; Asherie, N.; Lomakin, A.; Pande, J.; Donovan, J. M.;
Schnur, J. M.; Benedek, G. B. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 7883-
(21) Pouchert, C. J.; Behnke, J. The Aldrich Library of 13C and 1H FT-NMR
Spectra; Aldrich Chemical Co.: Milwaukee, Wisconsin, 1983.
7
887.
1228 J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002

Phosphonate Lipid Tubules II
A R T I C L E S
Scheme 1. Synthesis of DC(8,9)PC Phosphonate Analog 3a
A suspension of the phosphonic acid, choline chloride (2.27 g, 16.3
mmol), and dry pyridine (125 mL) was heated to 60 °C under a N
2
atmosphere. Trichloroacetonitrile (15.23 mL) was added, and the
reaction was stirred at 60 °C for 44 h. The solvent was removed under
reduced pressure, and the residue was dissolved in CH Cl . Activated
2 2
carbon was added to the solution, and the mixture was filtered through
Celite. The filtrate was concentrated under reduced pressure, and the
crude product was purified by flash chromatography on silica gel using
3 3 2
CHCl /CH OH/H O (2:0.8:0.15) to give 1-[(S)-3,4-bis(10,12-tri-
cosadiynoyloxy)propyl] phosphonic acid, 2-(N,N,N-trimethylammo-
nium) ethyl ester 7 (580 mg). This material was further purified by
1
trituration with acetone to afford 347.8 mg (14.0%). H NMR: δ 5.24
(m, 1H), 4.57 (m, 1H), 4.34 (m, 2H), 4.07 (m, 1H), 3.81 (m, 2H), 3.35
(
bs, 9H), 2.24 (m, 12H), 1.90 (m, 2H), 1.26-1.55 (m, 56H), 0.88 (t,
6
7
5
H, J ) 6.5 Hz). 13C NMR (67.5 MHz): δ 173.3, 173.2, 77.60, 77.58,
7.39, 76.93, 71.80 (d, Jcp ) 15.3 Hz), 66.86, 65.28, 65.19, 64.85,
7.66, 54.54, 34.43 (d, Jcp ) 31.4 Hz), 31.86, 30.89, 29.61, 29.59,
a
29.57, 29.54, 29.51, 29.48, 29.45, 29.43, 29.41, 29.39, 29.28, 29.12,
(
a) Pb(OAc)4, NaBH4; (b) P(Ph)3, CBr4; (c) P(OEt)3, ∆; (d) p-TsOH,
EtOH; (e) 10,12-tricosadiynoic acid, DMAP, DCC, CH2Cl2; (f) (CH3)3SiBr,
CH2Cl2, THF, H2O; (g) choline chloride, Cl3CCN, pyridine.
29.07, 28.93, 28.84, 28.80, 28.33, 25.60, 24.87 (d, Jcp ) 17.3 Hz),
22.65, 19.17, 14.10. 31P NMR (132 MHz, 85% H
PO
3
4
external
reference): δ 18.59. HRMS calcd for C54H93NO P: 898.6689. Found:
7
phosphite (12 mL, 70.0 mmol) was refluxed for 14 h. The residual
triethyl phosphite was removed under reduced pressure, and the crude
product was purified by flash chromatography on silica gel using ethyl
acetate to give diethyl [(R)-O-3,4-isopropylidene-3,4-dihydroxypropyl]
898.6685.
II.b. Physical Characterization. II.b.1. Tubule Preparation.
DC(8,9)PC and “C3” phosphonate tubules were prepared by adding 1
mg of lipid to 1 mL of EtOH/H O solvent, heating to 50 °C with
2
1
phosphonate 6 (2.37 g, 89.4%). H NMR: δ 4.38 (m, 1H), 4.13 (m,
vigorous stirring, and cooling the clear solutions to room temperature
to obtain flocculent off-white precipitates. Solvent compositions that
maximized precipitate yields for DC(8,9)PC and the “C3” phosphonate
3 were 0.7:0.3 and 0.6:0.4 (v:v), respectively; these solvent compositions
were used throughout the work presented in this report. It has been
shown that DC(8,9)PC tubule lengths depend not only upon the solvent
composition, but upon the rate at which the spherical vesicles from
which they form are cooled18 as well. Accordingly, the DC(8,9)PC and
the “C3” phosphonate preparations were driven through their tubule
formation temperatures (38 and 37 °C for DC(8,9)PC and the “C3”
phosphonate, respectively) at the same rate, 12.8 °C/h. Tubule yields
were determined by centrifuging and desiccating the precipitate to
constant mass.
5
H), 3.68 (dd, 1H, J ) 11 Hz, 7.5 Hz), 2.33-1.91 (m, 2H), 1.43-1.11
(m, 12H).
II.a.5. Diethyl [(R)-3,4-Dihydroxypropyl] Phosphonate (7). A
solution of 6 (1.91 g, 7.6 mmol) and p-toluenesulfonic acid (211 mg,
.1 mmol) in ethanol (70 mL) was stirred for 14 h at room temperature
under a N atmosphere. The mixture was neutralized with solid sodium
bicarbonate, filtered, and concentrated under reduced pressure. The
residue was brought up in CHCl (50 mL) and filtered through Celite.
This material was concentrated under reduced pressure and purified
by flash chromatography on silica gel using CHCl /CH OH (7:1) to
1
17
2
3
3
3
give diethyl [(R)-3,4-dihydroxypropyl] phosphonate 7 (1.06 g, 66.3%).
1
H NMR: δ 4.12 (m, 5H), 3.70 (m, 1H), 3.54 (dd, 1H, J ) 12 Hz, 6
Hz), 2.42 (bs, 2H), 2.12-1.87 (m, 2H), 1.35 (dt, 6H, J ) 7.5 Hz, 3
Hz).
II.b.2. Optical Microscopy. Differential interference contrast mi-
croscopy (DIC) was conducted with a Nikon Diaphot 300 inverted
microscope equipped with 60× and 100× Nomarski objectives, and
images were digitized using a Dage VE1000 CCD video camera
coupled to a high-resolution video capture board or were photographed
with a Nikon N6006 35 mm camera. Tubule lengths were tabulated on
a Macintosh computer using the public domain NIH Image program,
developed at the U.S. National Institutes of Health, and are available
on the Internet at http://rsb.info.nih.gov/nih-image/. All specimens were
handled as gently as possible to minimize tubule breakage.
II.a.6. Diethyl [(S)-3,4-Bis(10,12-tricosadiynoyloxy)propyl] Phos-
phonate (8). A solution of 1,3-dicyclohexylcarbodiimide (2.27 g, 11.0
mmol) in dry CH
phosphonate 7 (1.06 g, 5.0 mmol), 10,12-tricosadiynoic acid (3.64 g,
0.5 mmol), and (dimethylamino)pyridine (128.3 mg, 1.05 mmol) in
dry CH Cl (35 mL) and was stirred under a N atmosphere. After
2 2
Cl (7 mL) was added to a stirred solution of diethyl
1
2
2
2
stirring for 48 h at room temperature, the reaction mixture was filtered
through Celite, and the filtrate was evaporated to furnish a light-yellow
solid. The solid was purified by flash chromatography on silica gel
using hexane/ethyl acetate (2:1) to give diethyl [(R)-3,4-bis(10,12)-
II.b.3. Atomic Force Microscopy. Contact- and tapping-mode
atomic force microscopy probes were conducted on a vibrationally and
thermally isolated Digital Instruments BioScope, equipped with silicon
nitride NanoProbe SPM tips. Tubule external diameters were measured
by determining the perimeter of the cross-sectional AFM trace of the
tubule and dividing by π. Finite tip-size convolution error has been
found to be negligible.19 Local membrane elasticity probes were
conducted on a Digital Instruments Nanoscope IIIa Multi-Mode
1
tricosadiynoyloxy)propyl] phosphonate 8 (3.06 g, 70.4%). H NMR:
δ 5.36 (m, 1H), 4.40 (dd, 1H, J ) 12 Hz, 3 Hz), 4.12 (m, 5H), 2.37-
2
.11 (m, 14H), 1.70-1.20 (m, 62H), 0.93 (t, 6H, J ) 7.5 Hz).
II.a.7. 1-[(S)-3,4-Bis(10,12-tricosadiynoyloxy)propyl] Phosphonic
Acid, 2-(N,N,N-Trimethylammonium) Ethyl Ester (3). Bromotri-
methylsilane (1.21 mL, 9.91 mmol) was added over a period of 1 h to
a stirred solution of diethyl [(S)-3,4-bis(10,12-tricosadiynoyloxy)propyl]
19
scanning probe microscope using oxide-sharpened silicon nitride tips.
phosphonate 8 (3.06 g, 3.52 mmol) in dry CH
2
Cl
2
(7 mL) under a N
2
II.b.4. X-ray Diffraction. High-resolution small-angle X-ray scat-
tering was done at Beamline X20A of the National Synchrotron Light
Source at Brookhaven National Laboratory. X-rays (1.6122 Å) were
atmosphere. Six hours after completion of the addition, the reaction
mixture was concentrated in vacuo, the residue was brought up in
aqueous THF (10%, 15 mL), and the solution was heated at reflux for
-
1
used with an in-plane resolution of 0.0007 Å , commensurate with
the Ge[1,1,1] analyzer crystal, while out-of-plane resolution was relaxed
9
0 min. The solvent was removed under reduced pressure, and the
residue was dissolved in CHCl (30 mL). The CHCl solution was dried
Na SO ), filtered, and concentrated under reduced pressure to give
.80 g (98%) of phosphonic acid which was utilized without any further
purification (Scheme 1).
-
1
3
3
to 0.008 Å to increase signal intensity. Tubule samples of c ≈ 100
mg lipid/mL were prepared by centrifuging at 10 000g at 20 °C for 30
min; optical microscopy of the resuspended pellet revealed no discern-
ible change in tubule morphology induced by centrifugation. This
(
2
2
4
J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002 1229

A R T I C L E S
Thomas et al.
concentrated paste was placed in standard 1.5 mm diameter quartz
diffraction capillaries, yielding unoriented (powder) samples.
Tubules polymerize upon exposure to ionizing radiation and acquire
a deep purple color in time intervals well below those required for
X-ray data acquisition. Radiation damage was minimized by translating
yields. Deduction of the small percentage of granular precipitate
from the overall “C3” phosphonate yield leads us to conclude
that tubule-forming capacity has not been substantially altered
as a result of the changes made to the DC(8,9)PC backbone.
III.b. Tubule Gross Morphology. The most striking mor-
phological difference between tubules made from the “C3”
phosphonate analogue 3 and DC(8,9)PC is a 1.94-fold increase
in external diameter. Contact-mode atomic force microscopy
yielded a mean “C3” phosphonate tubule external diameter of
1.076 ( 0.090 µm; the same protocol yielded a mean diameter
determination of 0.554 ( 0.053 µm for DC(8,9)PC tubules.
Tubules made from the previously studied “C4” phosphonate
5
0 mm of the 80 mm capillary length continuously through the ∼0.5
mm tall collimated X-ray beam during data acquisition, which had the
effect of dispersing radiation damage throughout an approximately 100-
fold larger powder sample than a stationary sample would present.
Accordingly, one should expect a diminution of radiation damage as
great as 2 orders of magnitude. Complete data sets were acquired
with markedly less sample discoloration and presumably less radiation-
induced damage. Each point in reciprocal space was acquired with an
integral number of sample translations to ensure that any chance sample
inhomogeneities and domain alignments were distributed evenly
throughout the reciprocal space scan.
2
2
2
were found to have a mean diameter of 1.182 ( 0.135 µm.
Because the AFM probe used to determine these micron-scale
tubule diameters has nanometer-scale resolution, the uncertain-
ties we report are essentially the standard deviations of the
respective distributions about the mean values. It is evident from
the narrow (approximately (10%) standard deviations that the
“C3” and “C4” phosphonate mean tubule diameters, while close,
are nevertheless distinct.
The instrument resolution was deconvolved from the tubule inter-
lamellar (00l) peak and fit to the powder average of the scattering from
infinite length multilayer tubule structures having outer diameter D,
wall thickness T, and layer spacing d. DC(8,9)PC and “C4” phosphonate
tubules formed in ethanol/water solvent systems are known to be
composed of coaxially nested hollow cylindrical lamellae, and this is
also the case with tubules formed from the “C3” phosphonate 2. These
nested layers were assumed to be perfect cylinders, corresponding to
membranes with no bending fluctuations, that is, rigid membranes.2
From this model and the measured half-width at half-maximum, we
obtained a correlation length ê which we take to be the mean tubule
wall thickness. Because both phosphonate and DC(8,9)PC tubule outer
diameters D are rather narrowly distributed, determination of the
correlation length ê and the lamellar spacing d permits estimation of
the number of lamellae comprising the tubule wall.
Tubule lengths were measured microscopically, and once
again comparison of the instrument resolution to the magnitude
of measured quantity leads to uncertainties that are essentially
standard deviations about the mean. It was found that length
deviations for all three systems are large, being about (50%
of the mean value. As a result, the DC(8,9)PC and the previously
studied “C4” phosphonate tubule lengths, 25.8 ( 12.7 µm and
23.5 ( 11.1 µm respectively, may be considered to be essentially
indistinguishable. In contrast, the 16.6 ( 8.6 µm mean length
of the “C3” phosphonate tubules of this study is significantly
smaller. Because all three preparations were handled identically
and with great care to avoid tubule breakage, we interpret shorter
“C3” phosphonate tubule mean length to be an intrinsic property
of the “C3” phosphonate self-assembly product.
3,24
III. Results
III.a. Tubule Yields. The straightforward measure of how
well tubule-forming capacity has been conserved is to compare
DC(8,9)PC and “C3” phosphonate tubule yields. This proved
difficult, however, for small amounts of irregular granules,
approximately 1 µm in diameter, were present in the “C3”
phosphonate preparations. These granules are absent in the
corresponding dilute DC(8,9)PC and “C4” phosphonate prepara-
tions, but are identical in appearance to the bulk lamellar
coprecipitate that forms in overly concentrated DC(8,9)PC
solutions. We were unable to separate the two morphologies.
In view of the apparent similarity to granules formed in overly
concentrated DC(8,9)PC solutions, we decreased the “C3”
phosphonate concentrations used for tubule formation, but the
granules continued to appear even when the lipid concentration
was reduced 20-fold. The possibility that the granules are merely
undissolved bulk lipid was investigated by vigorously stirring
dilute phosphonate solutions at 75 °C for extended periods prior
to tubule formation, but the same results were obtained. On the
basis of extensive microscopic observations indicating that a
very small percentage of the microscope field-of-view is
occupied by these granules, and taking into account that tubules
are hollow and the granules appear solid, we can estimate
roughly that less than 5% of the “C3” phosphonate precipitate
by mass is in the form of granules. The oVerall “C3” phospho-
nate precipitate yield (comprising tubules and granules) is 90%,
the value found for DC(8,9)PC and “C4” phosphonate tubule
III.c. Phosphonate Tubule Membrane Deformability.
Tubules derived from phosphonates 2 and 3 appear to be more
flexible than those made of DC(8,9)PC and are often partially
unwound at the tubule end, as seen in the Nomarski differential
interference contrast micrograph of Figure 2. Phosphonate
tubules also deform through flattening upon drying. This
deformability was discovered in the course of AFM probes of
tubule structure; while tubules sink in solution and lie atop the
substrate, they remain mobile and are pushed about by the
probing AFM tip, necessitating air-drying to fix them. The extent
of the resultant flattening in the “C3” and “C4” phosphonate
tubules is so great, however, that the height difference between
alternating helix segments was found to be of the same
magnitude, or less, than that of the flattened ribbon roughness.
The membrane elasticity suggested by this deformation was
characterized through local force-modulation AFM. In this
technique, the AFM tip is forced down upon the sample, and
the amplitude of the cantilever tip deflection versus height above
the sample is measured. The magnitude of the oscillating
cantilever’s deflection as it is pushed onto the sample is inversely
related to the extent of indentation induced in the sample. A
“
soft” sample will deform with little resistance as the AFM tip
approaches, causing only a small change in the amplitude of
the oscillating AFM cantilever deflection. We found the AFM
cantilever tip had to be driven with 1.25× greater amplitude
with “C3” phosphonate tubules to obtain the same deflection
(
(
(
22) Thomas, B.; Safinya, C. R.; Plano, R. J.; Clark, N. A.; Ratna, B. R.;
Shashidar, R. Mater. Res. Soc. Symp. Proc. 1992, 248, 83-88.
23) Safinya, C. R.; Roux, D.; Smith, G. S.; Sinha, S. K.; Dimon, P.; Clark, N.
A.; Nellocq, A. M. Phys. ReV. Lett. 1986, 57, 2718-2721.
24) Roux, D.; Safinya, C. R. J. Phys. (Paris) 1988, 49, 307-0318.
1230 J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002

Phosphonate Lipid Tubules II
A R T I C L E S
Figure 2. Nomarski differential interference contrast micrograph of “C3”
phosphonate tubules. Notice the gentle bending of the tubule axes and the
unwinding occurring at the tubule ends. The vertical tubule found at the
right of the figure and the diagonal tubule it touches appear to have opposite
senses of handedness. The scalebar is 5 µm long.
obtained with DC(8,9)PC tubules, indicating significantly greater
“C3” phosphonate tubule wall elasticity. The identical technique
applied to the previously studied “C4” phosphonate tubules
indicated an even greater elasticity; the cantilever tip had to be
driven with 1.77× greater amplitude over that required for
DC(8,9)PC tubules.
Figure 3. Small-angle X-ray scattering showing the interlamellar funda-
mental peaks from “C4” phosphonate tubules (top), “C3” phosphonate
tubules (middle), and DC(8,9)PC tubules (bottom). The data are drawn to
the same scale but offset vertically for clarity. The solid trace is the 0.0007
-
1
III.d. Tubule and Helix Handedness. The desiccation-
induced flattening unexpectedly reduced the height differences
between alternating helix segments to that of membrane surface
roughness, frustrating straightforward attempts to determine
helix handedness through acquiring AFM height maps. It was
found, however, that water immersion of air-dried, fixed
specimens caused the tubules and helices to resume their
unflattened shape, yet remain fixed to the substrate, enabling
underwater contact-mode AFM.19 It was found that, as is the
case with the “C4” phosphonate 2, open helices formed from
the enantiomerically pure “C3” phosphonate 3 had roughly equal
probabilities of being right- or left-handed.
Å
fwhm instrument resolution, drawn to a different vertical scale.
Differences in peak intensities are attributable to differences in tubule
suspension concentrations in the three samples.
It should be noted that while Nomarski DIC is capable of
resolving the sense of handedness of an open helix, it is
generally incapable of doing so with fully formed tubules. DIC
optics are designed to enhance the index of refraction discon-
tinuity found at object-solution interfaces, such as at the ribbon
edges, but no such discontinuity exists along the fused helical
trace of fully formed, closed tubules. AFM examination of the
helical ridges found upon “C3” phosphonate tubules shows them
to be far less pronounced than those found upon DC(8,9)PC
tubule exteriors. This comparative smoothness, in conjunction
with the propensity of phosphonate tubules to flatten, made
AFM handedness determination of fully formed phosphonate
tubules considerably more difficult than with DC(8,9)PC tubules.
Nevertheless, AFM probes reveal roughly equal numbers of
right- and left-handed exterior helical traces on fully formed
“C3” phosphonate tubules.
Corroborating evidence supporting our “mixed” handedness
findings comes from Nomarski Differential Interference Contrast
(DIC) microscopy probes of undesiccated phosphonate tubules.
In the inverted microscope configuration utilized, the tubules
sink to the bottom of the ethanolic solution and lie atop the
glass coverslip. The ability to consistently and unambiguously
focus upon the side of unflattened helices in contact with the
19
coverslip has been demonstrated, and because optically probed
helices need not be fixed and may remain in solvent, flattening
does not occur. The intersection of the narrow microscope focal
plane with the near side of an open helically wound ribbon yields
a two-dimensional alternating pattern of ribbon and interstitial
void. As long as the microscope focal plane is unequivocally
placed at the helix side nearest the observer, the helical
handedness generating this pattern is resolvable in the same way
that the sense of handedness of a machine screw may be
ascertained from its two-dimensional photograph. We detected
left- and right-handed helices in nearly equal numbers, a finding
in agreement with the AFM probes and similar to our earlier
III.e. Phosphonate Tubule Internal Structure. Small-angle
X-ray scattering (SAXS) data from DC(8,9)PC, “C4”, and “C3”
phosphonate tubule suspensions are given in Figure 3. The
DC(8,9)PC, “C4”, and “C3” phosphonate tubule (00l) peaks
are centered at |k0| ) 4π sin(2θ/2)/λ ) 0.0998, 0.09624, and
-
1
0.100346 Å , respectively. Despite the differences in tubule
diameters for the three compounds, the interlamellar spacings
d are quite close: 62.97 Å for DC(8,9)PC and 65.29 and 62.62
Å for the “C4” and “C3” phosphonates, respectively. It is worth
noting that the DC(8,9)PC interlamellar spacing lies between
that of the two phosphonates. We interpret this remarkable
conservation of interlamellar spacing throughout a 2-fold
increase in tubule diameter as an indication of the relative
“
C4” phosphonate tubule results.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002 1231

A R T I C L E S
Thomas et al.
importance of interlamellar forces in comparison to those that
determine the cylindrical radii of curvature for the three systems.
Peak-shape analyses for completely rigid membranes leads
to a correlation length for the “C3” phosphonate êC3 ) 197.9
Å, whereas for DC(8,9)PC and the “C4” phosphonate tubules
the correlation lengths were êDC(8,9)PC ) 431.08 Å and êC4 )
and it has been speculated that localized impurities, rapid
formation kinetics, or thermal fluctuations that nucleate vortex-
antivortex pairs are the origins of these chiral defects.27
At the same time, the striking uniformity of the helical trace
found on enantiopure DC(8,9)PC tubule exteriors cannot be
ignored. If a statistical mechanism in determining helical
handedness is operative, molecular chirality must, at least in
the case of DC(8,9)PC, overwhelmingly bias the otherwise
random ensemble tilt. The corresponding lack of uniformity in
the equilibrium phosphonate tubule preparations indicates that,
whatever mechanism is in effect, the modifications we have
installed near the DC(8,9)PC stereogenic center have substan-
tially diminished the influence that molecular chirality exerts
upon self-assembly product morphology.
2
07.0 Å, respectively. Because of the narrow dispersion of
tubule diameters for each molecule, we interpret these values
to be equivalent to the tubule wall thickness T. The ratio of
tubule wall thickness to interlamellar spacing is the number n
of bilayers composing the tubule; we find n to be 6.9 for
DC(8,9)PC and 3.2 for both phosphonates, a finding intuitively
consistent with the observed phosphonate tubule flattening and
deformability. Thus, we find that the two phosphonates produce
tubules having nearly identical wall thicknesses (about half that
of tubules made from DC(8,9)PC) and with similarly close
cylindrical diameters (about twice that of DC(8,9)PC tubules).
While the number of coaxially nested cylinders composing
tubules is the same for both phosphonates, as is the interlamellar
distance between these cylinders, the AFM-determined surface
normal elasticities are not equal. Because the AFM probe was
applied to “C3” and “C4” phosphonate tubules having the same
number of lamellae as well as interlamellar separation, we
surmise that the elasticity variation reflects a structural difference
intrinsic to the lamellae. That is, the elasticity variation is not
a consequence of different tubule wall thicknesses offering more
or less resistance to the AFM probe tip. We note that the more
elastic “C4” phosphonate membrane produces tubules with the
largest diameter. If we set aside wall thickness considerations
momentarily and extend our comparison to include DC(8,9)PC
tubules, we find this trend continues; the least elastic membrane
is associated with the smallest diameter tubule. Because our
phosphonate study is currently limited to two molecules, and
we include DC(8,9)PC in this comparison as well, it is premature
to conclude that a membrane elasticity/tubule diameter correla-
tion exists, although it is in accord with the limited amount of
data.
A key prediction of chiral-packing theories is that tubule
diameter is a consequence of chiral intralamellar packing
efficiency. In racemic mixtures, one would expect chiral packing
effects to average to zero, and the theory predicts tubule diameter
will diverge to infinity; that is, flat lamellar sheets are the
predicted self-assembly product. DC(8,9)PC R-, S- enantiomer-
mixing experiments have failed to show this; indeed the
racemate produces left- and right-handed tubules of unchanged
28
diameter. This outcome has been explained by positing a
spontaneous DC(8,9)PC enantiomer separation that occurs in
the course of tubule formation; in this way, tubule handedness
remains traceable down to the dimensions of chiral molecular
packing. This enantiomeric separation is well precedented;
indeed Pasteur’s resolution of D- and L-tartaric acids through
the painstaking microscopic examination and subsequent separa-
tion of crystal habit is the basis of modern experimental
29
stereochemistry.
The presence of two crystal habits (left- and right-handed
helices) in our enantiopure phosphonate preparations, however,
is inconsistent with a chiral-packing mechanism. While the
evolution of left- and right-handed helices in mixed-enantiomer
DC(8,9)PC systems has been reasonably interpreted in terms
of chiral-packing theory, the experimental outcome does not
exclude the possibility that a chiral symmetry-breaking mech-
anism is operative. Specifically, it is possible that the intended
zero net-chirality state of the DC(8,9)PC enantiomer-mixing
experiment was in fact attained and that the subsequent
formation of left- and right-handed helices is reflective of the
random nature of the chiral symmetry-breaking mechanism. This
interpretation does not link tubule diameter to enantiomeric
excess, which is consistent with the enantiomer-mixing outcome,
and it also obviates the postulated enantiomeric separation.
Nevertheless, the enantiomer separation/chiral-packing models
remain a valid possibility.
IV. Discussion
The self-assembly behaviors exhibited by phosphonates 2 and
3
raise a number of questions regarding the role of molecular
structure, particularly chirality, in determining tubule handedness
and dimensions. Specifically, the surprising observation of stable
right- and left-handed helices in enantiopure phosphonate
preparations suggests helix formation can be statistical in nature,
rather than being based in chiral intermolecular interactions.
For example, a tilt-driven spontaneous chiral symmetry breaking
of the membrane that would produce both senses of helical
handedness has been proposed.25 Within this theory’s frame-
work, chiral membranes may result even from achiral molecules,
and the random tilt direction chosen by the ensemble should
produce our experimentally observed ∼50:50 ratio of left- to
right-handed helical structures. The helix may be otherwise
viewed as a macroscopic breaking of chiral symmetry, analo-
gous to the spiral defects seen in freely suspended hexatic films
of chiral molecules.26 No dependence of the spiral’s sense of
handedness upon molecular chirality is found in these systems,
The profound morphological differences between DC(8,9)PC
and tubules made from either phosphonate, as well as the
smaller, but distinct changes between “C3” and “C4” phospho-
nate tubules, are our first steps in delineating the dependence
of tubule formation upon the so-called “backbone” of the tubule-
forming molecule. The dramatic effects evoked by minor
modifications about the chiral center within this backbone
(
27) Kamien, R. D.; Selinger, J. V. J. Phys.: Condens. Matter 2001, 13, R1-
R22.
(
25) Seifert, U.; Shillcock, J.; Nelson, P. Phys. ReV. Lett. 1996, 77, 5237-
240.
26) Dierker, S. B.; Pindak, R.; Meyer, R. B. Phys. ReV. Lett. 1986, 56, 1819-
822.
(28) Spector, M. S.; Selinger, J. V.; Schnur, J. M. J. Am. Chem. Soc. 1997,
119, 8533-8539.
5
(
(29) Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of Organic
Compounds; John Wiley & Sons: New York, 1994.
1
1232 J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002

Phosphonate Lipid Tubules II
A R T I C L E S
suggest that additional bond length and angle modifications may
produce tubules of new, technologically useful and theoretically
interesting dimensions. That no significant loss of tubule-
forming capacity has occurred thus far supplies additional
motivation to continue the synthetic strategy we have under-
taken. While the results of this report are in general supportive
of chiral symmetry-breaking models, the striking uniformity of
handedness found in enantiopure DC(8,9)PC preparations
demands that molecular chirality, absent in chiral symmetry-
breaking models, be included in tubule formation theory.
research. J.E.K. was supported in part by a Research and
Training Program award of the Exxon Education Foundation.
C.L.C. was supported by a University of Wyoming Grant-In-
Aid. We thank Professor David L. Johnson of the University
of Oklahoma Health Sciences Center Department of Oc-
cupational & Environmental Health for discussions regarding
fibrous aerosols. High-resolution mass spectra were obtained
by the Washington University Resource for Biomedical and Bio-
organic Mass Spectrometry at Washington University, St. Louis,
Missouri. Portions of this work were executed at Beamline
X20A of the National Synchrotron Light Source at Brookhaven
National Laboratory, administered by the United States Depart-
ment of Energy.
Acknowledgment. B.N.T. was supported by NSF CAREER
Grant CHE-9734266. Acknowledgment is made by B.N.T. to
the donors of the Petroleum Research Fund, administered by
the American Chemical Society, for partial support of this
JA012137A
J. AM. CHEM. SOC.
9
VOL. 124, NO. 7, 2002 1233