Thermodynamic and structural characterization of amino acid-linked dialkyl lipids

Article abstract of DOI:10.1016/j.chemphyslip.2004.11.001

Using differential scanning calorimetry (DSC), X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR), we determined some thermodynamic and structural parameters for a series of amino acid-linked dialkyl lipids containing a glutamic ac

Full text of DOI:10.1016/j.chemphyslip.2004.11.001

Chemistry and Physics of Lipids 134 (2005) 29–39
Thermodynamic and structural characterization of amino
acid-linked dialkyl lipids
a,∗
b
c
Stephanie Tristram-Nagle , Ruthven N.A.H. Lewis , Joseph W. Blickenstaff ,
c
d
b
Michael DiPrima , Bruno F. Marques , Ronald N. McElhaney ,
c
d
John F. Nagle , James W. Schneider
a
Department of Biological Sciences, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
b
Department of Biochemistry, University of Alberta, Edmonton, AB., Canada T6G 2H7
Department of Physics, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
c
d
Department Chemical Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
Received 27 September 2004; received in revised form 3 November 2004; accepted 3 November 2004
Available online 28 November 2004
Abstract
Using differential scanning calorimetry (DSC), X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR),
we determined some thermodynamic and structural parameters for a series of amino acid-linked dialkyl lipids containing a
glutamic acid-succinate headgroup and di-alkyl chains: C12, C14, C16 and C18 in CHES buffer, pH 10. Upon heating, DSC
shows that the C12, C14 and annealed C16 lipids undergo a single transition which XRD shows is from a lamellar, chain ordered
subgel phase to a fluid phase. This single transition splits into two transitions for C18, and FTIR shows that the upper main
transition is predominantly the melting of the hydrocarbon chains whereas the lower transition involves changes in the headgroup
ordering as well as changes in the lateral packing of the chains. For short incubation times at low temperature, the C16 lipid
appears to behave like the C18 lipid, but appropriate annealing at low temperatures indicates that its true equilibrium behavior is
like the shorter chain lipids. XRD shows that the C12 lipid readily converts into a highly ordered subgel phase upon cooling and
2
˚
suggests a model with untilted, interdigitated chains and an area of 77.2 A /4 chains, with a distorted orthorhombic unit subcell,
◦
˚
˚
a = 9.0 A, b = 4.3 A and β = 92.7 . As the chain length n increases, subgel formation is slowed, but untilted, interdigitated chains
prevail.
©
2004 Elsevier Ireland Ltd. All rights reserved.
Keywords: DSC; XRD; FTIR; Peptidic amphiphiles; Lipid bilayers
1. Introduction
∗
Corresponding author. Tel.: +1 412 268 3174;
fax: +1 412 681 0648.
Synthetic amino acid/lipid conjugates (sometimes
E-mail address: stn@cmu.edu (S. Tristram-Nagle).
referred to as peptidic amphiphiles when there are two
0
009-3084/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.chemphyslip.2004.11.001

3
0
S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
or more amino acids) of various types have been in-
vestigated in recent years to help better understand
membrane biophysics (Kunitake, 1992; Ringsdorf et
al., 1988), as a platform for biomaterials develop-
ment (Neumann and Ringsdorf, 1986; Winger et al.,
1
996; Yu et al., 1996; Dori et al., 2000; Hartgerink
et al., 2001), and also as a means to anchor func-
tional units to surfaces for intermolecular force mea-
surements (Israelachvili, 1992; Berndt et al., 1995b;
Pincet et al., 1994). An important feature of many of
these lipids is the competition between hydrophobic
and hydrophilic interactions. Tsonchev et al. (2003)
have pointed out that orientation-dependent interac-
tions between headgroups must be considered when
predicting the morphology of self-assembled structures
of peptidic amphiphiles.
Fig. 1. Sketch of the n = 14 member of the (Cn)2-Glu-C2-COOH
family.
The solid lines divide the molecule into the hydrocar-
bon chains, and a glutamic acid and a succinate residue.
The bold solid line emphasizes the comparison to a
more conventional lipid description, with hydrocarbon
tails, abridgingorconnectinggroup, andasuccinamide
headgroup. Following Berndt et al. (1995), these lipids
have been named the (Cn)2-Glu-C2-COOH lipids. The
molecule shown in Fig. 1 is in the neutral form, which
occurs for pH ≤ 4. At pH 7 and higher, the carboxyl
group is deprotonated. The series that we study consists
of disaturated C12, C14, C16 and C18 hydrocarbon
chains. DSC, XRD and FTIR, in contrast to SFA, mea-
sure the properties of bulk lipids, which self-assemble
to their energy-minimized structure in aqueous solu-
tion. The self-aggregated structure of these lipids de-
pends strongly on the hydrophobic force, as well as
on the electrostatic, hydration and hydrogen bonding
forces due to the headgroup. The structure is also mod-
ified by the van der Waals interactions between the
hydrocarbon chains, and the relative strength of this
interaction increases as the chain length increases.
Attractive interactions between peptidic headgroups
might be expected to stabilize self-assembled struc-
tures. Shimizu and Hato (1993) observed significant
increases in the hydrocarbon chain-melting transition
temperature (TM) for four-and five-mer peptidic am-
phiphiles containing glycine as a headgroup compared
to other amino acids. They ascribed the stabilizing
effect to a lower degree of hydration between the
glycine moieties, facilitating inter-glycine hydrogen
bonding. Those peptidic amphiphiles that contained
glycine exhibited poor dispersability in water, even
◦
when sonicated at 70–80 C for several minutes, fur-
ther implicating the stabilizing effect of interbilayer
hydrogen bonds. Strong headgroup–headgroup inter-
actions with attendant strong adhesion between glycine
amphiphiles were manifested in opposed bilayers after
contact using the surface-force apparatus (Schneider
et al., 2002). This effect was removed at higher pH
levels, where the glycine amphiphiles acquired a nega-
tive charge, presumably increasing the hydration state
and attenuating interbilayer hydrogen bonds. Such pH-
induced control of the inter-headgroup interactions
may serve as a general mechanism for developing
controlled-release liposomes, or other inducible mi-
crostructures using peptidic amphiphiles.
In this study we use differential scanning calorime-
try (DSC), X-ray diffraction (XRD), and Fourier trans-
form infrared spectroscopy (FTIR) to investigate the
physical properties of a series of novel dialkyl amino
acid lipids that contain a glutamic acid residue that is
a component used in the synthesis. Fig. 1 shows the
chemical structure which can be described in two ways.
2. Materials and methods
2.1. Synthesis of di-O-alkyl glutamic acid,
N-succinamide ((Cn)2-Glu-C2-COOH)
All reagents and solvents were of at least analyt-
ical grade quality and were obtained through Fisher
Scientific (Pittsburgh, PA). In the first step, di-O-alkyl
glutamicacidp-toluenesulfonate(dialkyl-glu-pTs)was
synthesized by the acid-catalyzed Fischer esterification

S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
31
◦
used by Berndt et al. (1995a). Briefly, 7.36 grams of
l-glutamic acid, an appropriate long chain alcohol
quired at a scan rate of ∼10 C/h. For the MC-2 scans,
4.5 mg was added to 3 ml CHES buffer or water in nal-
gene vials. These samples were cycled above the TM
(i.e. dodecanol, tetradecanol, hexadecanol or octade-
◦
canol) and p-toluenesulfonic acid monohydrate were
dissolved in ∼500 mL of toluene in a molar ratio
and then to −20 C three times with vortexing at each
◦
temperature. These samples were stored at 0–5 C for
1
:2:1.5 and heated under reflux until a stoichiomet-
up to two weeks prior to the scans. 1.20 ml was added
to the sample cell of the MC-2 and CHES buffer was
placed into the reference cell. The MC-2 was run at
ric amount of water was recovered in a Dean–Stark
trap (∼6 h). The product was then concentrated by ro-
tary evaporation and recrystallized three times from
acetone. The purity and identity of the crystalline ma-
terial was determined by a combination of NMR, IR,
and electrospray mass spectroscopy. In the second step,
◦
∼60 C/h.
2.4. X-ray diffraction
7
.27 g of dialkyl-glu-pTs and 1.25 equiv. of diisopropy-
2.4.1. Capillary samples
lethylamine (DIPEA) were dissolved in 200 mL of
THF/chloroform (1:1) and 1.15 equivalents of succinic
anhydride were added with stirring. The reaction mix-
Twenty-five milligrams lipid was added to 125 mg
CHES buffer at pH 10 in 100 ␮l nalgene vials. The
◦
samples were heated once or twice briefly to 60 C,
◦
◦
ture was stirred for 2 h at 40 C after which time the re-
vortexed and cooled to 20 C. About 10 mg of lipid
action was essentially complete. The product was then
concentrated by rotary evaporation and recrystallized
three times from ethanol. The identity and purity of the
recrystallized material were also determined by NMR,
IR, and electrospray mass spectroscopy.
dispersion was loaded into 1.0 mm glass X-ray capil-
laries (Charles Supper Co.) as in Tristram-Nagle et al.
(1993). Various temperature protocols as described in
Section 3 were carried out on individual capillaries in
an effort to form highly crystalline, low temperature
phases. Samples were studied using a Rigaku RUH3R
rotating copper anode X-ray source operated at 6 kW.
In order to avoid degrading the sample at one posi-
tion, capillaries were translated vertically after each
40 min exposure. A graphite monochromator selected
2
.2. Chemicals
For the preparation of the buffer solution, 250 mM
CHES (Fisher Scientific, Pittsburgh, PA) was used.
The pH was adjusted to 10.0 using concentrated
NaOH. Phosphate buffer was 100 mM sodium phos-
phate buffer, pH 7.4.
˚
Cu K␣ radiation (λ = 1.5418 A). The beam size was
1 mm × 1 mm, collimated using two X–Y slits before
the sample. Data were collected in duplicate scans of
20 or 40 min/sample using a Rigaku Mercury CCD
2
.3. DSC
(Woodlands, TX) with 1024 × 1024 pixels and a pixel
size of 70 ␮m. The S-distance (sample-to-CCD) for this
experiment was 9.34 cm.
High-sensitivity DSC measurements were per-
formed using a Calorimetry Sciences Multi-Cell instru-
ment (Calorimetry Sciences Corporation, American
Fork, UT) at UAlberta and a Microcal MC-2 calorime-
ter (Microcal LLC, Northampton, MA) at CMU. For
most of the experiments, 4–6 mg lipid was weighed
and 500 ␮l CHES buffer was added to the hastelloy
ampoules of the multi-cell calorimeter. After sealing
the ampoule, the lipid was hydrated by heating above
the main phase transition temperature (TM) and then
2.5. Thin layer chromatography
The lipids were checked for purity and degradation
both before and after DSC and X-ray diffraction.
Thin layer chromatography (TLC) using chloro-
form:methanol:7N NH4OH (46:18:3, v/v) on samples
before experiments revealed <1% lysolipid when
stained with vapor from iodine crystals. Negligible
degradation was observed under the gentle conditions
of hydration that were used with the multi-cell
calorimeter and with preparation of the samples for
X-ray diffraction. After X-ray diffraction, the lipids
were degraded 5–10%. Care was taken not to heat to
◦
◦
cooling to −7 C, both at a rate of 90 C/h. For the
greaterchainlengthsamples, anadditionalheatingscan
was performed to just below the phase transition tem-
perature to further anneal the sample before cooling
◦
to −7 C. Heating and cooling thermograms were ac-

3
2
S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
◦
0 C, since at pH 10 this caused considerable degra-
9
dation (∼50% loss of the succinic acid headgroup),
while DPPC (the control) was only 5% degraded under
similar conditions.
2
.6. FTIR spectroscopy
Samples were prepared for FTIR spectroscopy by
dispersing 1–2 mg of lipid in 75–100 ␮l of a D2O-
based CHES buffer of a composition similar to that
used for the DSC studies. The dispersion was vig-
◦
orously agitated at temperatures ∼5–10 C above the
TM of the lipid and then squeezed between the CaF2
windows of a heatable, demountable liquid cell (NSG
Precision Cells, Farmingdale, NY) equipped with a
Fig. 2. Calorimetric transition temperature as a function of pH and
chain length. Circles are from the data in Fig. 3 at pH 10; triangles
from pH 7.4; squares from pH 4.
2
5 ␮m Teflon spacer. Once mounted in the sample
holder of the spectrometer, the sample was heated
◦
and cooled between 0 and 60 C by an external,
sition temperature for each lipid. Only the C16 and
C18 chain lengths had an additional lower temperature
transition, and the C18 lipid had a second, smaller tran-
computer-controlled water bath. Infrared spectra were
acquired as a function of temperature with a Digilab
FTS-40 Fourier transform spectrometer (Biorad, Dig-
ilab Division, Cambridge, MA) using data acquisition
and data processing protocols following Lewis et al.
◦
sition at 42 C. During cooling, a downward (exother-
mic), sharp transition was observed only for the higher
chain lengths, C16 and C18. In addition to this sharp
(2001).
◦
transition, an additional broader transition at ∼22 C
was also observed for C16. For the lower chain lengths,
C12 and C14, multiple smaller and broader transitions
were observed. Although the downward exotherms are
3
. Results
3
.1. Differential scanning calorimetry
When aqueous dispersions of the (Cn)2-Glu-C2-
COOH lipids were prepared as for the MC-2 samples,
the pH was measured to be 4, due to dissociation of the
proton from the lipid headgroup (see Fig. 1). Prelim-
inary DSC measurements indicated rather high tran-
sition temperatures (see Fig. 2) and non-reproducible
enthalpies, due to the insolubility and aggregation
of the lipids. At pH 7.4 (phosphate buffer) and pH
1
0 (CHES buffer), the lipids exhibited similar pat-
terns of thermotropic phase behavior, aside from small
changes in the transition temperatures (see Fig. 2).
However, due to considerable clumping of the lipids
at pH <8, most of the DSC, X-ray and FTIR experi-
ments were performed at pH 10 where handling was
easier.
The initial heating and cooling scans are shown in
Fig. 3. The initial heating scans in Fig. 3 exhibit the
main transition which is identified as the highest tran-
Fig. 3. DSC heating scans (upward peaks) after brief incubation at
◦
−
7 C and cooling scans (downward peaks) of (Cn)2-Glu-C2-COOH
lipids in CHES buffer at pH 10 as a function of chain length. The
◦
data shown were acquired with the Multicell DSC at 9–10 C/h and
are displaced on the ordinate for clarity.

S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
33
3.2. X-ray diffraction
Fig. 5A shows diffraction from (C12)2-Glu-C2-
◦
COOH at 10 C in the wide-angle region. All of the
lipids displayed similar wide-angle patterns when first
◦
observed at 10 C. When the C12 lipid was heated to
4
◦
8 C, the pattern in Fig. 5B emerged. The wide-angle
pattern is much more diffuse than in the gel phase,
which is typical of fluid phase lipids. Fig. 5C shows a
very ordered pattern that resulted when the C12 chain
◦
length was first heated to 48 C, then cooled to and
◦
held at 10 C for one day. This pattern is typical of
a highly ordered subgel phase. However, the identical
spacings of the strongest wide-angle reflections due to
chain packing in Fig. 5A and C suggest that these sam-
ples are in the same phase and that it is only the addi-
tional annealing that results in a more perfectly ordered
sample that clearly exhibits the reflections better.
The lamellar nature of the low-angle scattering was
especially evident in the C18 sample which displayed
eight orders of lamellar diffraction (see Fig. 6A). The
other chain lengths displayed a strong first Bragg order,
but the higher orders were weaker than those for the
C18 lipid (data not shown). Nevertheless, low-angle
D-spacings were calculated and are summarized in
Table 2. Fig. 6B shows the wide-angle radial average of
diffracted intensity data of (Cn)2-Glu-C2-COOH lipids
corresponding to the type of data shown in Fig. 5. The
averagedintensitiesshownherearefromourmostwell-
ordered samples. We attempted to order each lipid as
well as possible by following a T-jump thermal pro-
tocol (Tristram-Nagle et al., 1987) that was designed
to successfully nucleate and grow the subgel phase of
DPPC (Tristram-Nagle et al., 1994). Each sample was
Fig. 4. Main transition enthalpies (open circles) and also the sum of
the enthalpies of both the lower and main transitions (solid squares)
for the pH 10 data of Fig. 3 (initial scans). Lines are to guide the eye.
shifted in temperature only slightly from the main en-
dotherms for C16 and C18, there is a marked hysteresis
upon cooling for the lower chain lengths. Fig. 2 shows
that the main melting transition temperatures (TM) in-
crease systematically as a function of chain length n
for each pH.
The hydrocarbon chain length dependence of the
phase transition enthalpies for these lipids at pH 10
is shown in Fig. 4. Justification for showing the com-
bined enthalpies of the lower and main transitions will
be given in the FTIR results section. With careful an-
nealing of C16 at lower temperatures, the lower tran-
sition disappears and the main transition increases in
enthalpy. The uncertainties for C16 in Fig. 4 indicate
that these variations in enthalpy arise from different
thermal histories. For C18, there is a larger contribu-
tion from the lower transition that did not change sig-
nificantly with low temperature annealing. A numeri-
cal summary of the calorimetry data after annealing is
given in Table 1. FWHM stands for full-width at half-
maximal height and is the measure of the breadth of
the transition.
◦
first heated to just above its TM, then cooled at 10 C/h
◦
to 1 C, where it was held for 1–2 h in order to initiate
nucleation of some subgel domains. Then the temper-
◦
ature was raised to 10 C below the TM in the hope
that no additional subgel domains would be nucleated.
The temperature was held there for one day so that the
Table 1
Summary of calorimetric results for annealed lipids at pH 10
◦
◦
Lipid
TM ( C)
FWHM ( C) (main)
ꢀH (kcal/mol) (main)
ꢀH (kcal/mol) (2 lower)
(C12)2-Glu-C2-COOH
(C14)2-Glu-C2-COOH
(C16)2-Glu-C2-COOH
(C18)2-Glu-C2-COOH
19.9
33.2
44.9
56.7
0.70
0.60
0.63
0.40
16.6
19.2
19.4
12.3
–
–
–
10.8

3
4
S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
Fig. 6. Radially averaged X-ray intensities. (A) Low-angle data with
eight orders of lamellar diffraction for (C18)2-Glu-C2-COOH lipid
◦
at 10 C obtained from a capillary sample in pH 10 CHES buffer. (B)
◦
Wide-angle data collected at 10 C in pH 10 CHES buffer, except
that (C14)2-Glu-C2-COOH data were collected at 25 C.
◦
nucleated domains might grow into a complete subgel
phase. After this T-jump protocol, all capillaries were
◦
loaded into a cassette and stored at 5 C for one month
◦
prior to X-ray study at 10 C. This T-jump method was
only partially successful for these di-alkyl lipids, and
additional growth of the subgel occurred with temper-
ature changes during the X-ray experiment. The C12
sample, with results shown in Fig. 5C, was the best or-
dered phase. The C14 sample increased its order when
◦
◦
the temperature was raised from 10 C to 25 C, evi-
denced by the increasing sharpness of its wide-angle
−
1
˚
peaks (1.3 < q < 1.8 A ). Fig. 6 shows that the wide-
angle peaks are sharper for C12 and C14; the broadness
oftheC16andC18peakssuggeststhatthesubgelphase
is still not fully formed.
Fig. 5. X-ray diffraction patterns of fully hydrated lipids in pH 10
◦
CHES buffer in capillaries. (A) (C12)2-Glu-C2-COOH at 10 C. (B)
3
.3. Fourier transform infrared spectroscopy
◦
(
C12)2-Glu-C2-COOH at 48 C. (C) (C12)2-Glu-C2-COOH held at
◦
◦
1
0 C for one day after heating to 48 C.
Fig. 7 shows the FTIR spectra obtained in both the
◦
subgel phase (0 C) and in the melted, fluid phase for
C12 and C18. The broad absorption band between 1550
Table 2
◦
Summary of X-ray results for (Cn)2-Glu-C2-COOH lipids in CHES buffer at 10 C
Lipid
D-Spacings ( A˚ )
Low-angle
Medium-angle
Wide-angle
(C12)2-Glu-C2-COOH
(C14)2-Glu-C2-COOH
(C16)2-Glu-C2-COOH
(C18)2-Glu-C2-COOH
44.8
48.5
52
13.5, 10.8, 8.9, 8.3, 7.6, 6.1, 5.9, 5.4, 5.1
14.7, 9.0, 6.0, 4.9
8.4, 7.3
4.49, 4.12, 3.94, 3.81
4.52, 4.28, 4.00, 3.81
4.53, 4.19, 3.98, 3.80
4.53, 4.21, 3.98, 3.80
55.7
8.4, 7.4

S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
35
◦
Fig. 7. FTIR spectra of (C12)2-Glu-C2-COOH and (C18)2-Glu-C2-COOH in the subgel phase (0 C) and in the fluid phase.
and 1600 cm 1 arises from the carboxylate group of the
−
higher temperature in Fig. 7 the sharp band is replaced
for both the C12 and C18 lipids by a broader band
succinic moiety. In the subgel phase of C12 this band
occurs near 1580 cm and shifts to near 1567 cm
in the fluid phase. This indicates that the interaction of
the carboxylate with its counterion, most likely Na ,
changes during the phase transition; this could be due
to an increase in hydration of the headgroup in the
fluid phase. This shift did not occur for C18, sug-
gesting that the headgroup in the subgel phase may
be more hydrated in this sample. Fig. 7 also shows a
−1
−
1
−1
involving components centered near 1737 cm and
−
1
1
719 cm . The latter component is in a range consis-
+
tent with hydrated or hydrogen-bonded ester carbonyl
groups.
These FTIR data provide valuable information re-
garding the phase transitions seen by calorimetry in
Fig. 3. Fig. 8 shows the temperature dependencies
−
1
broad absorption band between 1600 and 1650 cm
which arises from the amide bond between the suc-
cinic and glutamic acid moieties. With both C12 and
−
1
C18 this amide I band occurs near 1625 cm in the
−
1
subgel phases and shifts to 1640 cm in the fluid
phases. This behavior is consistent with either a de-
crease in the polarity of the local environment of the
amide carbonyl group, or a decrease in the strength of
the hydrogen-bonding interactions in the fluid phase,
or both. The weakening of the hydrogen-bonding inter-
actions is the more likely reason; this could be caused
by increased interlamellar water in the fluid phase. Fi-
−
1
nally, there is also an absorption band near 1735 cm
that arises from the ester carbonyl groups between the
glutamic acid and the alcohol moieties. This band is
sharp in the low temperature phases of both lipids at
◦
0
C, suggesting the existence of highly immobilized
ester carbonyl groups (Lewis and McElhaney, 2002).
The frequency of this band is outside the range that
is normally attributed to hydrogen-bonded ester car-
bonyl groups, suggesting that these groups are in rela-
tively nonpolar environments in the subgel phase and
are not involved in hydrogen-bonding interactions with
water or other groups in the headgroup region. At the
Fig. 8. FTIR data (top) showing ester C=O bandwidth ꢀν (triangles)
and CH2 stretching ν (squares) during heating of (C12)2-Glu-C2-
COOH (open symbols) and (C18)2-Glu-C2-COOH (solid symbols).
On the bottom are two heating scans from the DSC data of Fig. 3.

3
6
S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
for C12 and C18 of the width ꢀν of the ester C O
−
1
stretching band near 1735 cm and the frequencies
of the CH2 symmetric stretching band. Transitions in
these two quantities occurred for C12, C14 and C16 at
the same temperature as the DSC melting transitions,
as shown for C12 in Fig. 8. Also, transitions in the
−
1
CH2 scissoring frequencies near 1470 cm (data not
shown) occurred at these same temperatures. However,
for C18 the transitions in the C O stretching bandwidth
and in the CH2 scissoring frequency both occurred at
◦
Fig. 9. Model of (C12)2-Glu-C2-COOH lipid at 10 C.
the same temperature as the lower DSC transition near
◦
5
2 C, and the transition in the CH2 stretching fre-
quency occurred at the same temperature as the sharp,
main melting transition near 57 C. This indicates that
4. Discussion
◦
major changes in the lateral packing interactions be-
tween hydrocarbon chains and in the hydrogen bond-
ing interactions in the headgroup interfacial region of
C18 both coincide with the broad lower temperature
transition, whereas changes in hydrocarbon chain con-
formational order, i.e., hydrocarbon chain melting, oc-
cur at the main transition TM. We therefore assign the
broad lower temperature transition in C18 to a subgel-
to-gel phase transition and the main transition to a
gel-to-fluid phase transition. In C12 and C14 all the
spectroscopic changes described above coincide with
the single calorimetric transition which we assign to a
subgel-to-fluid phase transition. The behavior of C16
is more complex. Of the two transitions observed by
DSC in Fig. 3, all of the structural changes observed
by FTIR occur at the higher melting transition. In addi-
tion, the lower transition can be thermally annealed out,
suggesting that the lower temperature peak is probably
attributable to domain inhomogeneities in the subgel
phase formed when initially cooled to low tempera-
tures. Thus, on full equilibration C16 is more similar
to C12 and C14 where the phase transition is from an
Lc phase to an L␣ phase.
The preceding justifies the way we compared the en-
thalpies of transition in Fig. 4. In order to compare the
energetics of melting all the way from the subgel phase
to the fluid phase for all the chain lengths, the lower
and main transition enthalpies are combined for C16
and C18. As shown in Fig. 4, there was the expected
increase in the combined enthalpies with chain length.
Deviations from a smooth increase are most likely as-
sociated with incomplete formation of the subgel phase
in longer chains due to their slower kinetics of subgel
formation.
We suggest the structural model for the ordered,
lamellar subgel phase of (C12)2-Glu-C2-COOH whose
main features are shown in the two-dimensional sketch
in Fig. 9. The sharp (resolution limited) reflections seen
in Fig. 6B in the wide-angle region of the (C12)2-Glu-
C2-COOH subgel phase indicate that the hydrocarbon
chains are all-trans and not tilted. The X-ray reflec-
tions can be indexed to a distorted orthorhombic sub-
˚
˚
˚
cell with d20 = 4.49 A, d1–1 = 3.94 A, and d11 = 3.81 A.
˚
The unit cell parameters for this subcell, a = 9.0 A,
◦
˚
b = 4.30 A and β = 92.7 , can be used to calculate
the area AChain = 19.3 A per hydrocarbon chain. The
2
˚
distorted orthorhombic subcell is similar to that ob-
served in another charged lipid with a small head-
group, DM-TAP (Lewis et al., 2001), with a nearly
2
˚
identical AChain = 19.1 A . It is also not much larger
2
˚
than AChain = 19.0 A obtained for DPPC subgel phase
(Ruocco and Shipley, 1982).
Our model in Fig. 9 emphasizes interdigitation of
the hydrocarbon chains. One interesting feature illus-
trated by the figure is that the two chains on a single
lipid molecule are not nearest neighbors, whereas the
pair of chains of more conventional lipids that form
interdigitated bilayers probably are nearest neighbors.
This difference in chain packing in our model in Fig. 9
is possible because the backbone linkage on the COOH
lipids is longer than the usual glycerol linkage. How-
ever, this kind of detail of the chain packing cannot
be determined from our data, any more than it can be
for conventional interdigitated phases. Furthermore, it
must be emphasized that Fig. 9 is a crude sketch re-
garding the chain packing.
The main purpose of Fig. 9 is to illustrate the bi-
layer thickness, which should not depend much upon

S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
37
˚
chain packing, and which is about 38 A in our sug-
the DSC cooling scans (Fig. 3) the presence of broader
transitions at lower temperatures and lack of a strong,
sharp transition near TM suggest that the shorter chain
lengths do not cool into a gel phase, but rather nucleate
and slowly grow directly into a subgel phase. A sep-
arate phase that could be a gel phase occurs between
the subgel and fluid phases for C18, and transiently
for C16. The equilibrium melting behavior of the C12,
C14 and C16 lipids is similar to that of other lipids
that have thermal transitions directly from the subgel
phase to the fluid phase, such as DMPE (Wilkinson
and Nagle, 1984; Lewis and McElhaney, 1993) and
DM-TAP (Lewis et al., 2001). They also share with
MPPC the property of having no gel phase (Serrallach
et al., 1984), although MPPC does have a ripple phase,
which these lipids do not have. Mini-interdigitation
was proposed to be an important factor for MPPC
subgel formation (Tristram-Nagle et al., 1999); the
chains of the (Cn)2-Glu-C2-COOH lipids appear to
be fully interdigitated in the subgel phase so this pre-
disposition is already present. As the chain length in-
creases, the total van der Waals interactions between
chains increases. If strong headgroup interactions are
also required for subgel formation, then the longer
chains could inhibit this formation since their van der
Waals interactions are relatively more dominant. The
same behavior was observed for the PCs (Lewis et al.,
1987).
Further confirmation that the low temperature phase
in the (Cn)2-Glu-C2-COOH lipids is a subgel, not a gel
phase, is that the combined enthalpies of melting are
more than double the main melting transition of the
zwitterionic PCs. What is the reason for these larger
enthalpies? The reason is probably not due to melt-
ing of interdigitated chains alone, since Laggner et al.
(1987) showed that interdigitated chains in dihexade-
cyl PC have nearly the same enthalpies of melting as
DPPC. Our explanation for the larger enthalpies is that
the sharp transitions that appear in Fig. 3 for the C12
and C14 chain lengths are combined sub- and main
transitions. The melting of the subgel phase directly to
the liquid crystalline phase also had a high enthalpy of
16.2 kcal/mol in DMPE (Wilkinson and Nagle, 1984;
Aokietal., 2001)and15.9 kcal/molinDM-TAP(Lewis
et al., 2001).
gested model. Together with the low-angle D-spacing
of 44.8 A, this indicates a steric water spacing D W of
A between adjacent bilayers in an MLV. This water
spacing is smaller than that of the non-interdigitated
subgel phase of DPPC (10.4 A) (Nagle and Wiener,
988). It is also slightly smaller than the water spacing
8 A) of the interdigitated subgel phase of dihexade-
cylphosphatidylcholine (DHPC) reported by Lohner et
al. (1987). In contrast, the cationic lipid DM-TAP had
virtually no water between bilayers in its subgel phase
ꢀ
˚
˚
7
˚
1
(
˚
(Lewis et al., 2001).
Let us also consider a model in which the chains
are not interdigitated. Untilted chains would require a
steric bilayer thickness of 55 A which is larger than
the D-spacing. To reduce the bilayer thickness to the
measured D-spacing would require tilting the chains
by more than 30 . Such a large tilt would broaden the
wide-anglepeaksoftheunorientedMLVsamplesmuch
more than is shown in Fig. 6b. Furthermore, the in-
crement in D-spacing of 3.7 A as a function of chain
length (see Table 2) is consistent with interdigitated
chains for all the chain lengths. A calculated increase
˚
◦
˚
˚
˚
of 2 × 1.27 A = 2.54 A would result from the increase
in two interdigitated carbons/chain, plus an increase
˚
˚
in water spacing of 1.2 A to yield 3.7 A. If the chains
were not interdigitated, either the water spacing would
have to decrease by (5.1–3.7) A upon increasing n by
˚
◦
2
, or the tilt angle would have to exceed 43 to ac-
˚
commodatethemeasuredincrement3.7 AinD-spacing
versus chain length. The latter possibility contradicts
the wide-angle data and the former possibility appears
less likely than our preferred model. Unfortunately, the
lack of an electron dense headgroup precludes obtain-
ing electron density profiles that would confirm this
conclusion as has been done for the interdigitated gel
phase of DPPC in the presence of surface active small
molecules (McIntosh et al., 1983).
The (Cn)2-Glu-C2-COOH lipids become more or-
dered than typical gel phases at low temperature and
this suggests that the low temperature phase should
be considered to be a subgel phase. This was espe-
cially apparent for C12 and C14 in the X-ray data in
Fig. 5C and Table 2 that showed several wide-angle and
many medium-angle reflections, which are not present
in typical gel phases, indicating periodicities associ-
ated with well-packed hydrocarbon chains and head-
group ordering (Raghunathan and Katsaras, 1995). In
The proposed phase behavior is supported by the
FTIR data in Fig. 8 that reveal that there are two sep-
arable events. The first of these is manifested by the

3
8
S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
changes in the C O stretching and the CH2 bending
regions that are the result of degradation of lateral
close contact interactions between the lipid molecules;
this is consistent with what occurs in the transition
from subgel to gel phases (see Lewis and McEl-
haney, 1990, 1992). The second event is manifested
by the change in the CH2 stretching frequency that re-
flects the trans-gauche isomerizations during melting
of the chains (Lewis and McElhaney, 2002). Both of
these events occur simultaneously at the single tran-
sition in C12, C14 and (with annealing) C16, but
for C18 the first event occurs at a lower tempera-
ture than the CH2-stretching transition. This confirms
that the lower transition in C18 is a subgel transition
and that the upper transition is a chain melting main
transition.
To return to an initial goal of this work, to investigate
the relative importance of the van der Waals, hydration
and electrostatic forces to the structure and phase be-
havior of these lipids, we discuss the effect of using
pH 10 CHES buffer. When CHES buffer is not used,
these lipids are quite acidic (pH 4.0) in pure water. This
results in higher TM’s for all the chain lengths, indicat-
ing a more stable low temperature structure at low pH
search. BFM and JWS were supported by the National
ScienceFoundationBES-0093538, theArnoldandMa-
bel Beckman Foundation, and a joint grant from the Air
Force Office of Scientific Research and the DARPA
SIMBIOSYS program.
References
Aoki, H., Koto, T., Kodama, M., 2001. Calorimetric inves-
tigation of conversion to the most stable subgel phase
of phosphatidylethanolamine-water system. J. Thermal Anal.
Calorim. 64, 299–306.
Berndt, P., Fields, G., Tirrell, M., 1995a. Synthetic lipidation of pep-
tides and amino acids—monolayer structure and properties. J.
Am. Chem. Soc. 117, 9515–9522.
Berndt, P., Kurihara, K., Kunitake, T., 1995b. Measurement of forces
between surfaces composed of two-dimensionally oriented, com-
plementary and noncomplementary nucleobases. Langmuir 11,
3083–3091.
Boggs, J.M., 1980. Intermolecular hydrogen bonding between lipids:
influence on organization and function of lipids in membranes.
Can. J. Biochem. 58, 755–770.
Dori, Y., Bianco-Peled, H., Satija, S., Fields, G.B., McCarthy, J.B.,
Tirrell, M., 2000. Ligand accessibility as a means to control cell
response to bioactive bilayer membranes. J. Biomed. Mater. Res.
50, 75–81.
Hartgerink, J.D., Beniash, E., Stupp, S.I., 2001. Self-assembly and
mineralization of peptide-amphiphile nanofibers. Science 294,
(see Fig. 2). At pH 4.0, the carboxyl group may be
partially protonated compared to pH 10 with buffer,
thereby reducing electrostatic repulsive interactions,
allowing tighter headgroup hydrogen bonding. Simi-
larly, the PEs have higher TM’s than the corresponding
PCs due to weak and transient hydrogen bonding in
the headgroup region (Nagle, 1976; Boggs, 1980). In
addition, osmotic pressure measurements have shown
that adhesion between adjacent PE bilayers is stronger
than between the corresponding PC lipids (McIntosh
and Simon, 1996). It is therefore consistent that the
COOH lipids tended to clump for lower pH and it was
not until pH 10 that these lipids could be hydrated into
the usual opaque suspensions that are easier to study
by the methods employed in this paper.
1
684–1688.
Israelachvili, J.N., 1992. Intermolecular and Surface Forces, Aca-
demic Press, NY.
Kunitake, T., 1992. Synthetic bilayer membranes: molecular design,
self-organization, and application. Angew. Chem. Int. Ed. Engl.
31, 709–726.
Laggner, P., Lohner, K., Degovics, G., Muller, K., Schuster,
A., 1987. Structure and thermodynamics of the dihexadecyl-
phosphatidylcholine-water system. Chem. Phys. Lipids 44,
31–60.
Lewis, R.N.A.H., Mak, N., McElhaney, R.N., 1987. A differential
scanning calorimetric study of the thermotropic phase behavior
of model membranes composed of phosphatidylcholines con-
taining linear saturated fatty acyl chains. Biochemistry 26, 6118–
6126.
Lewis, R.N.A.H., McElhaney, R.N., 1990. The subgel phases
of n-saturated diacyl phosphatidylcholines. A Fourier trans-
form infrared spectroscopic study. Biochemistry 29, 7946–
7
953.
Acknowledgments
Lewis, R.N.A.H., McElhaney, R.N., 1992. Structures of the subgel
phases of n-saturated diacyl phosphatidylcholine bilayers: FTIR
1
3
2
spectroscopic studies of
J. 61, 63–77.
Lewis, R.N.A.H., McElhaney, R.N., 1993. Calorimetric and spectro-
scopic studies of the polymorphic phase behavior of a homolo-
gous series of n-saturated 1,2-diacyl phosphatidylethanolamines.
Biophys. J. 64, 1081–1096.
C O and H labeled lipids. Biophys.
STN and JFN were supported by US National Insti-
tutes of Health grant GM44976-12. RNAHL and RNM
were supported by Operating and Major Equipment
grants from the Canadian Institutes of Health Research
and the Alberta Heritage Foundation for Medical Re-

S. Tristram-Nagle et al. / Chemistry and Physics of Lipids 134 (2005) 29–39
39
Lewis, R.N.A.H., Tristram-Nagle, S., Nagle, J.F., McElhaney, R.N.,
001. The thermotropic phase behavior of cationic lipids: calori-
Ruocco, M.J., Shipley, G.G., 1982. Characterization of the sub-
transition of hydrated dipalmitoylphosphatidylcholine bilayers.
Biochim. Biophys. Acta 684, 309–320.
Schneider, J., Dori, Y., Haverstick, K., Tirrell, M., 2002. Force titra-
tion of Langmuir-Blodgett bilayers of glycine amphiphiles: JKR-
type measurements using the surface-force apparatus. Langmuir
18, 2702–2709.
Serrallach, E.N., de Haas, G.A., Shipley, G.G., 1984. Structure and
thermotropic properties of mixed-chain phosphatidylcholine bi-
layer membranes. Biochemistry 23, 713–720.
2
metric, infrared spectroscopic and X-ray diffraction studies of
lipid bilayer membranes composed of 1,2-di-O-myristoyl-3-
N,N,N trimethylaminopropane (DM-TAP). Biochim. Biophys.
Acta 1510, 70–82.
Lewis, R.N.A.H., McElhaney, R.N., 2002. Vibrational spectroscopy
of lipids. In: Chalmers, J.M., Griffiths, P.R. (Eds.), The Handbook
of Vibrational Spectroscopy, vol. 5. John Wiley & Sons Ltd., pp.
3
447–3464.
Lohner, K., Schuster, A., Degovics, G., Muller, K., Laggner, P., 1987.
Thermal phase behaviour and structure of hydrated mixtures be-
tweendipalmitoyl-anddihexadecyl-phosphatidylcholine. Chem.
Phys. Lipids 44, 61–70.
Tristram-Nagle, S., Zhang, R., Suter, R.M., Worthington, C.R., Sun,
W.-J., Nagle, J.-F., 1993. Measurement of chain tilt angle in
fully hydrated bilayers of gel phase lecithins. Biophys. J. 64,
1097–1109.
McIntosh, T.J., Simon, S.A., 1996. Adhesion between phos-
phatidylethanolamine bilayers. Langmuir 12, 1622–1630.
McIntosh, T.J., McDaniel, R.V., Simon, S.A., 1983. Induction of
an interdigitated gel phase in fully hydrated phosphatidylcholine
bilayers. Biochim. Biophys. Acta 731, 109–114.
Tristram-Nagle, S., Wiener, M., Yang, C.-P., Nagle, J.F., 1987. Kinet-
ics of the subtransition in dipalmitoylphosphatidylcholine. Bio-
chemistry 26, 4288–4294.
Tristram-Nagle, S., Suter, R.M., Sun, W.-J., Nagle, J.F., 1994. Ki-
netics of subgel formation in DPPC: X-ray diffraction study
proves nucleation-growth hypothesis. Biophys. Biochim. Acta
1191, 14–20.
Nagle, J.F., 1976. Theory of lipid monolayer and bilayer phase tran-
sitions: effect of headgroup interactions. J. Membr. Biol. 27,
2
33–250.
Tristram-Nagle, S., Isaacson, Y., Lyatskaya, Y., Liu, Y., Brummond,
K., Katsaras, J., Nagle, J.F., 1999. Polymorphism in myris-
toylpalmitoylphosphatidylcholine. Chem. Phys. Lipids 100,
101–113.
Tsonchev, S., Schatz, G.C., Ratner, M.A., 2003. Hydrophobically-
driven self-assembly: a geometric packing analysis. Nano Lett.,
623–626.
Wilkinson, D.A., Nagle, J.F., 1984. Metastability in the phase behav-
ior of dimyristoylphosphatidylethanolamine bilayers. Biochem
23, 1538–1541.
Nagle, J.F., Wiener, M.C., 1988. Structure of fully hydrated bilayer
dispersions. Biochim. Biophys. Acta 942, 1–10.
Neumann, R., Ringsdorf, H., 1986. Peptide liposomes from am-
phiphilic amino acids. J. Am. Chem. Soc. 108, 487–490.
Pincet, F., Perez, E., Bryant, G., Lebeau, L., Mioskowski, C., 1994.
Long-range attraction between nucleosides with short-range
specificity:directmeasurements. Phys. Rev. Lett. 73, 2780–2783.
Raghunathan, V.A., Katsaras, J., 1995. Structure of the LC’ phase
in a hydrated lipid multilamellar system. Phys. Rev. Lett. 74,
4
456–4459.
Winger, T.M., Ludovice, P.J., Chaikof, E.L., 1996. Lipopeptide con-
jugates: biomolecular building blocks for receptor activating
membrane-mimetic structures. Biomaterials 17, 437–441.
Yu, Y.-C., Berndt, P., Tirrell, M., Fields, G.B., 1996. Self-assembling
amphiphiles for construction of protein molecular architecture.
J. Am. Chem. Soc. 118, 12515–12520.
Ringsdorf, H., Schlarb, B., Venzmer, J., 1988. Molecular archi-
tecture and function of polymeric oriented systems: models
for the study of organization, surface recognition, and dynam-
ics of biomembranes. Angew. Chem. Int. Ed. Engl. 27, 113–
1
58.