Structural properties of docosahexaenoyl phospholipid bilayers investigated by solid-state 2H NMR spectroscopy

Article abstract of DOI:10.1021/ja011745n

Polyunsaturated lipids in cellular membranes are known to play key roles in such diverse biological processes as vision, neuronal signaling, and apoptosis. One hypothesis is that polyunsaturated lipids are involved in second messenger functions in biologi

Full text of DOI:10.1021/ja011745n

J. Am. Chem. Soc. 2001, 123, 12611-12622
12611
Structural Properties of Docosahexaenoyl Phospholipid Bilayers
2
Investigated by Solid-State H NMR Spectroscopy
Horia I. Petrache,†,§ Amir Salmon,‡ and Michael F. Brown*,‡
Contribution from the Department of Physiology, Johns Hopkins UniVersity School of Medicine,
Baltimore, Maryland 21205, and Department of Chemistry, UniVersity of Arizona, Tucson, Arizona 85721
ReceiVed July 17, 2001
Abstract: Polyunsaturated lipids in cellular membranes are known to play key roles in such diverse biological
processes as vision, neuronal signaling, and apoptosis. One hypothesis is that polyunsaturated lipids are involved
in second messenger functions in biological signaling. Another current hypothesis affirms that the functional
role of polyunsaturated lipids relies on their ability to modulate physical properties of the lipid bilayer. The
2
present research has employed solid-state H NMR spectroscopy to acquire knowledge of the molecular
organization and material properties of polyunsaturated lipid bilayers. We report measurements for a homologous
series of mixed-chain phosphatidylcholines containing a perdeuterated, saturated acyl chain (n:0) at the sn-1
position, adjacent to docosahexaenoic acid (DHA, 22:6ω3) at the sn-2 position. Measurements have been
performed on fluid (LR)-state multilamellar dispersions as a function of temperature for saturated acyl chain
lengths of n ) 12, 14, 16, and 18 carbons. The saturated sn-1 chains are therefore used as an intrinsic probe
with site-specific resolution of the polyunsaturated bilayer structure. The 2H NMR order parameters as a function
of acyl position (order profiles) have been analyzed using a mean-torque potential model for the chain segments,
and the results are discussed in comparison with the homologous series of disaturated lipid bilayers. At a
given absolute temperature, as the sn-1 acyl length adjacent to the sn-2 DHA chain is greater, the order of the
initial chain segments increases, whereas that of the end segments decreases, in marked contrast with the
corresponding disaturated series. For the latter, the order of the end segments is practically constant with acyl
length, thus revealing a universal chain packing profile. We find that the DHA-containing series, while more
complex, is still characterized by a universal chain packing profile, which is shifted relative to the homologous
saturated series. Moreover, we show how introduction of DHA chains translates the order profile along the
saturated chains, making more disordered states accessible within the bilayer central region. As a result, the
2
area per lipid headgroup is increased as compared to disaturated bilayers. The systematic analysis of the H
NMR data provides a basis for studies of lipid interactions with integral membrane proteins, for instance in
relation to characteristic biological functions of highly unsaturated lipid membranes.
Introduction
tigated through dietary manipulations of the ω3 essential fatty
acid (EFA) precursor, R-linolenic acid (18:3ω3), in rats5,6 and
monkeys.7 It has been observed that a dietary reduction in ω3
PUFA content involving DHA is compensated by a concomitant
increase in ω6 PUFAs such as DPA (docosapentaenoic acid,
22:5ω6), although significant alterations of function still occur.
In humans, the influence of DHA in nervous system develop-
ment has been studied by Uauy et al.,8 who showed reduced
learning ability and significant visual disfunction of human
infants who were fed formulas with reduced DHA content.
These studies linked the high levels of DHA found in human
placental blood and breast milk with the need for preformed
DHA in infant development. Consequently, there is increased
emphasis on supplementation of human infant formulas with
preformed DHA.9-11 DHA has also been shown to be involved
Docosahexaenoic acid (DHA, 22:6ω3) is the most unsaturated
fatty acid commonly present in biological membranes. Large
quantities of polyunsaturated lipids, especially DHA, are found
in the brain gray matter and synaptic membranes.1,2 In particular,
the outer segment membranes of the rod cells of the retina
contain close to 50 mol % DHA of the total fatty acid
components.3,4 Evidence for a functional role of DHA lipids
has been accumulating from various physiological studies. For
example, biological effects of polyunsaturated fatty acids
(PUFAs) such as DHA on neural processes have been inves-
* Corresponding author. Tel.: 520-621-2163. Fax: 520-621-8407. E-
mail: mfbrown@u.arizona.edu.
† Johns Hopkins University School of Medicine.
‡ University of Arizona.
§ Current address: Laboratory of Physical and Structural Biology,
NICHD, National Institutes of Health, Bethesda, MD 20892.
(1) Salem, N.; Kim, H. Y.; Yergey, J. A. In Health Effects of
Polyunsaturated Fatty Acids in Seafoods; Simopoulos, A. P., Kifer, R. R.,
Martin, R. E., Eds; Academic Press: New York, 1986; pp 263-317.
(2) Salem, N.; Wegher, B.; Mena, P.; Uauy, R. Proc. Natl. Acad. Sci.
U.S.A. 1996, 93, 49-54.
(5) Bouree, J. M.; Durand, G.; Pascal, G.; Youyou, A. J. Nutr. 1989,
119, 15-22.
(6) Greiner, R. S.; Moriguchi, T.; Hutton, A.; Slotnick, B. M.; Salem,
N. Lipids 1999, 34, S239-S243.
(7) Neuringer, M.; Connor, W. E.; Lin, D. S.; Barstad, L.; Luck, S. Proc.
Natl. Acad. Sci. U.S.A. 1986, 83, 4021-4025.
(3) Stone, W. L.; Fransworth, C. C.; Dratz, E. A. Exp. Eye. Res. 1979,
28, 387-397.
(8) Uauy, R.; Peirano, P.; Hoffman, D.; Mena, P.; Birch, D.; Birch, E.
Lipids 1996, 31, S167-S176.
(4) Brown, M. F. Curr. Top. Membr. 1997, 44, 285-356.
(9) Neuringer, M. Am. J. Clin. Nutr. 2000, 71, 256S-267S.
10.1021/ja011745n CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/17/2001

12612 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Petrache et al.
in human diseases as diverse as cancer, cystic fibrosis, and
multiple sclerosis (see Stillwell12 and Jenski13 for recent
reviews).
cholines, with n ) 12, 14, 16, and 18 carbons, in comparison
with the homologous series of the corresponding symmetric
lipids, di(m:0-d2m-1) phosphatidylcholines. By contrasting the
features of the two homologous series, rather than individual
lipids, one has a qualitative gain in understanding the effect of
polyunsaturation on the bilayer dynamic structure. An alternative
route to investigate the effects of polyunsaturation has been
undertaken by Holte et al.,25,26 who compared lipids with
different levels of unsaturation in the sn-2 position, for a fixed
sn-1 chain (18:0-d35) labeled for 2H NMR spectroscopy. From
this perspective, using a mixed-chain, unsaturated series of
stearoyl phosphatidylcholines, Holte et al.25,26 showed that the
2H NMR spectra of the saturated perdeuterated (18:0-d35) chain
changed markedly with the nature of the sn-2 chain, and
consequently that the order parameter profile changed shape,
with significant perturbation in the midchain region.
One interesting aspect is that the mammalian organism goes
to a long extent to produce and maintain high concentrations
of DHA.14 In particular, mammals must obtain the associated
essential fatty acid precursor 18:3ω3 from their diet in order to
form DHA through a series of enzymatic reactions involving
elongases and desaturases.2 In addition, the susceptibility to
autooxidative damage necessitates large amounts of vitamin E
(R-tocopherol) to prevent degradation of DHA,15 again consis-
tent with an essential biological role of PUFAs.14 However, the
exact mechanisms of action of PUFAs are still largely unknown.
One possibility is that PUFAs act as ligands for receptors
involved in transcription16 or biological signaling.1,17-20 Yet
the large concentration of PUFA-lipids found in the brain and
retina is difficult to explain by such a mechanism. An alterna-
tive hypothesis is that PUFAs yield characteristic membrane
properties associated with lipid-protein interactions.21-23 In
agreement with the second hypothesis, measurements using flash
photolysis have suggested that the bulk properties of DHA-
containing phospholipids are required for proper photoactivity
of the G protein-coupled receptor, rhodopsin.23,24 These proper-
ties may include the elastic curvature stress of the membrane
coupled to the activation of the receptor protein, and subsequent
signal transduction.23 The above observations have led to a
renewed interest in the biophysical properties (structure in
particular) of polyunsaturated bilayers.14,23,25-29 Knowledge of
structural properties enhances our understanding of molecular
level interactions between lipid molecules themselves, as well
as with protein components, because these interactions are
reflected in the intimate molecular packing within the cell
membrane.30-33
In this paper, we follow the approach of Barry et al.34 and
compare bilayers of the polyunsaturated series (n:0-d2n-1)-
(22:6)PC with the disaturated series of di(m:0-d2m-1) phosphati-
dylcholines. Here we focus on the detailed structural properties
of polyunsaturated lipid bilayers in the liquid-crystalline (LR)
state. We have previously investigated35 the homologous series
of disaturated phosphatidylcholines, di(m:0)PC in the LR phase,
and shown how the 2H NMR measurements can be interpreted
using a statistical model for the acyl chain configurational
disorder. The present study represents a natural extension to
the case of mixed-chain, saturated-polyunsaturated bilayers,
in which influences of polyunsaturation on the bilayer physical
2
properties have been systematically explored using H NMR
spectroscopy. By doing so, we discovered a number of interest-
ing new aspects. First, as shown by Salmon et al.30 for the case
of (16:0)(22:6)PC in the fluid state, polyunsaturation at the
glycerol sn-2 position increases the sn-1 chain disorder for all
mixed-chain lipids in the series. This leads to an increased cross-
sectional area (lower surface density) and reduced hydrocarbon
thickness when compared to their disaturated counterparts at a
given absolute temperature. A second important aspect is that,
at fixed absolute temperature, longer sn-1 chains show decreased
order at the chain end (bilayer center) but increased order in
the vicinity of the lipid headgroups. Despite this complex
behavior, we find a universal chain packing curve for the mixed-
chain saturated-polyunsaturated bilayers, just as previously
obtained for the disaturated series.35 By comparing the packing
profiles, we are able to show how the DHA chain affects the
overall chain packing within the bilayer. Finally, the detailed
structural results that emerge from this systematic analysis of
2H NMR data provide a firm conceptual framework for studies
The structural and dynamical properties of polyunsaturated
chains are revealed by comparison with saturated bilayers, as
2
demonstrated by Salmon et al.,30 who contrasted the H NMR
spectrum of (16:0-d31)(22:6)PC with that of di(16:0-d31)PC lipid
bilayers in the fluid (LR) state. The differences in the 2H NMR
spectra were interpreted in terms of an increase in the
configurational freedom of the palmitoyl (16:0) chains due to
2
the adjacent DHA chain. A more extensive H NMR analysis
was performed by Barry et al.34 for the homologous series of
mixed-chain, polyunsaturated (n:0-d2n-1)(22:6) phosphatidyl-
(10) Uauy, R.; Mena, P.; Rojas, C. Proc. Nutr. Soc. 2000, 59, 3-15.
(11) Jorgensen, M. H.; Hernell, O.; Hughes, E. L.; Michaelsen, K. F. J.
Ped. Gastro. Nutr. 2001, 32, 293-296.
(12) Stillwell, W. Curr. Org. Chem. 2000, 4, 1169-1183.
(13) Jenski, L. J. Curr. Org. Chem. 2000, 4, 1185-1200.
(14) Bloom, M.; Linseisen, F.; Lloyd-Smith, J.; Crawford, M. In Magnetic
Resonance and Brain Function: Approaches from Physics; Maraviglia, B.,
Ed.; IOS Press: Amsterdam, 1999; pp 527-553.
(25) Holte, L. L.; Peter, S. A.; Sinnwell, T. M.; Gawrisch, K. Biophys.
J. 1995, 68, 2396-2403.
(26) Holte, L. L.; Separovic, F.; Gawrisch, K. Lipids 1996, 31, S199-
S203.
(27) Litman, B. J.; Mitchell, D. C. Lipids 1996, 31, S193-S197.
(28) Huster, D.; Jin, A. J.; Arnold, K.; Gawrisch, K. Biophys. J. 1997,
73, 855-864.
(29) Huster, D.; Arnold, K.; Gawrisch, K. Biochemistry 1998, 37, 17299-
17308.
(30) Salmon, A.; Dodd, S. W.; Williams, G. D.; Beach, J. M.; Brown,
M. F. J. Am. Chem. Soc. 1987, 109, 2600-2609.
(31) Bloom, M. J.; Evans, E.; Mouritsen, O. G. Q. ReV. Biophys. 1991,
24, 293-397.
(15) Kelley, E. E.; Buettner, G. R.; Burns, C. P. Arch. Biochem. Biophys.
1995, 319, 102-109.
(16) de Urquiza, A. M.; Liu, S. Y.; Sjoberg, M.; Zetterstrom, R. H.;
Griffiths, W.; Sjovall, J.; Perlmann, T. Science 2000, 290, 2140-2144.
(17) Albino, A. P.; Juan, G.; Traganos, F.; Reinhart, L.; Connolli, J.;
Rose, D. P.; Darzynkiewicz, Z. Cancer Res. 2000, 60, 4139-4145.
(18) Chen, Z. Y.; Istfan, N. W. Prostaglandins, Leukotrienes Essent. Fatty
Acids 2000, 63, 301-308.
(19) Siddiqui, R. A.; Jenski, L. J.; Neff, K.; Harvey, K.; Kovacs, R. J.;
Stillwell, W. Biochim. Biophys. Acta 2001, 1499, 265-275.
(20) Kim, H. Y.; Akbar, M.; Lau, A.; Edsall, L. J. Biol. Chem. 2000,
275, 35215-35223.
(32) Petrache, H. I.; Grossfield, A.; MacKenzie, K. R.; Engelman, D.
M.; Woolf, T. B. J. Mol. Biol. 2000, 302, 727-746.
(33) Petrache, H. I.; Zuckerman, D. M.; Sachs, J. N.; Killian, J. A.;
Koeppe, R. E.; Woolf, T. B. Langmuir, in press.
(34) Barry, J. A.; Trouard, T. P.; Salmon, A.; Brown, M. F. Biochemistry
1991, 30, 8386-8394.
(21) Gibson, N. J.; Brown, M. F. Biochem. Biophys. Res. Commun. 1990,
169, 1028-1034.
(22) Mitchell, D. C.; Straume, M.; Litman, B. J. Biochemistry 1992, 31,
662-670.
(23) Brown, M. F. Chem. Phys. Lipids 1994, 74, 159-180.
(24) Liebman, P. A.; Botelho, A. V.; Brown, M. F.; Parkes, J. H. Biophys.
J. 2001, 80, 244.
(35) Petrache, H. I.; Dodd, S. W.; Brown, M. F. Biophys. J. 2000, 79,
3172-3192.

2H NMR of Polyunsaturated Bilayers
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12613
length, requires a model for the segmental motions to calculate the
average cos â(i) from the value of cos2 â(i) . For this, one needs to
consider the distribution function f (cos â(i)) that gives the relative
populations of all possible tilt angles â(i).35,40,41 The conventional
approach is based on a diamond lattice representation of chain
conformations,42 which gives the segmental projection along the bilayer
normal as43,44
of lipid interactions with integral membrane proteins, including
G-protein coupled receptors such as rhodopsin.
Methods
Experimental Procedures. Polyunsaturated phospholipids (n:0-
d
2n-1)(22:6)PC, with n ) 12, 14, 16, and 18 carbons, were synthesized
as described.30 Perdeuteration of fatty acids (Sigma, St. Louis, MO)
2
was performed with H gas using a 10% Pd-charcoal catalyst. The
cos â(i) ) 1/2 - S(i)
(2)
docosahexaenoic acid was purchased from Nu Chek Prep. Inc. (Elysian,
MN) and used without further purification. Disaturated phospholipids
were synthesized from the perdeuterated fatty acid anhydrides and the
cadmium chloride adduct of sn-glycero-3-phosphocholine,36 and then
purified on silica gel columns. Next, the sn-2 chains of the disaturated
lipids were hydrolyzed with snake venom phospholipase A2 from
Crotalus adamanteus (Sigma, St. Louis, MO). The resulting lysophos-
pholipid was reacylated with the anhydride of docosahexaenoic acid
to obtain the mixed-chain phospholipids, followed by silica gel
chromatography. The docosahexaenoic acid-containing samples were
all handled under argon with use of solvents containing butylated
hydroxytoluene (BHT) as an antioxidant (in a molar proportion of 0.1%
BHT to phospholipid). Treatment of the samples with phospholipase
A2, followed by transesterification of the released sn-2 fatty acids and
gas-liquid chromatography, detected less than 2% acyl chain migration.
This indicated a high isomeric purity of the product phospholipid. NMR
samples were prepared as 50 wt % multilamellar dispersions of 200-
250 mg of lipid in 67 mM sodium phosphate buffer, pH 7.0, using
2H-depleted water (Aldrich). Samples were stored under argon at -80
°C while not in use. Solid-state 2H NMR spectra were acquired at 55.43
MHz as described,34,35 and measurements were made for each lipid as
a function of temperature. The temperature was controlled within (0.5
°C. All samples were subsequently checked by thin-layer chromatog-
raphy, and revealed no contamination or degradation.
CD
A more recent result35,41 uses a continuum description of segmental
(i)
orientations and, for |S | > 1/8 gives
CD
(i)
CD
-8S - 1
1
2
cos â(i)
)
1 +
(3)
(
)
x
3
(i)
CD
(For |S | < 1/8, numerical calculations are needed as discussed
previously.35,41) This expression has been shown to be superior to the
diamond lattice model based on molecular dynamics simulations.41,45
In the framework of this continuum model, segmental orientations
are described using a mean-field potential U(â), which is the generator
of the segmental distribution function f (â) according to
U(â)
1
Z
f (â) ) exp -
(4)
(
)
kBT
with the partition function Z being given by
π
U(â)
Z )
exp -
sin â dâ
(5)
∫
2
(
)
0
kBT
The H NMR spectral powder patterns were numerically deconvo-
luted (de-Paked) to yield the θ ) 0° oriented spectra, as described by
Bloom et al.37 Peak assignments were made with the assumptions that
the order parameters decrease in magnitude from C2 (R carbon) to the
methyl end, and that the area under each peak is proportional to the
Here the carbon index i has been suppressed for clarity. The mean-
field potential U(â) (called the potential of mean torque) is determined
empirically from the experimentally measured order parameters SCD
.
number of deuterons generating the peak.36 The C-2H bond order
In a first-order approximation, it can be written as
(i)
parameters, S of the de-Paked 2H NMR spectra were calculated
CD
from the quadrupolar splittings using the relation38
U(â) ≈ -kBTꢀ1 cos â
(6)
3
2
(i)
CD
|∆νQ| ) øQ|S ||P2(cos θ)|
(1)
where ꢀ1 is a dimensionless model parameter which quantifies the
orientational confinement of the acyl segment in kBT energy units.35
Note that the potential of mean torque is an odd function of the segment
tilt because it changes sign upon inversion. The advantage of the first-
where øQ ≡ e2qQ/h ) 167 kHz,39 θ represents the angle between
the bilayer director axis and the static magnetic field direction, and
P2(cos θ) ) (3cos2 θ - 1)/2, which for θ ) 0° is equal to unity. The
quadrupolar splitting is designated as (∆νQ)| for the bilayer normal
parallel to the main magnetic field (θ ) 0°), and (∆νQ) for the
order mean-torque model is that it allows conversion of the order
(i)
CD
parameter profiles S
into an equivalent mean-torque parameter
(mean-torque strength) profile, given by the ꢀ(1i) values as a function of
acyl chain segment index i. In what follows, we apply the first-order
perpendicular orientation (θ ) 90°). Based on geometrical consider-
(i)
CD
2
ations, the values of S are assumed to be negative.
mean-torque model to the H NMR order parameters of the saturated
Mean-Torque Model of Segmental Conformations. Before pre-
senting the structural results, we briefly summarize the theoretical
considerations of the acyl chain configurational properties. The SCD
chains of the mixed-chain lipids, viz. (n:0-d2n-1)(22:6) phosphatidyl-
cholines, to calculate average structural parameters.
Calculation of Structural Parameters. For planar bilayers, the main
structural parameters of interest are the average cross-sectional area of
the acyl chain AC and the average acyl chain length LC, which are
defined and calculated as explained.35 In the present case we are
interested in comparing the properties of the mixed-chain, saturated-
polyunsaturated PC series to the corresponding disaturated series, based
on measurements on the saturated sn-1 chain. For the plateau region,
assuming that the saturated segmental cross-sectional area and the
projected length are inversely correlated,38,40,41 we have that
2
observables from H NMR spectroscopy represent a measure of the
carbon-deuterium (CD) bond orientations with respect to the static
(i)
external magnetic field axis.38 The order parameter S of carbon
CD
segment (i) is essentially the ensemble average 3cos2 â(i) - 1 /2, where
â
(i) denotes the angle between the C-2H bond direction (principal axis
frame) and the bilayer normal (director axis), and the brackets indicate
a time or ensemble average. As discussed elsewhere,35,40 the conversion
of the order parameters into structural quantities, such as acyl chain
(36) Williams, G. D.; Beach, J. M.; Dodd, S. W.; Brown, M. F. J. Am.
Chem. Soc. 1985, 24, 6868-6873.
(41) Petrache, H. I.; Tu, K.; Nagle, J. F. Biophys. J. 1999, 76, 2479-
2487.
(37) Bloom, M.; Davis, J. H.; MacKay, A. L. Chem. Phys. Lett. 1981,
80, 198-202.
(42) Seelig, A.; Seelig, J. Biochemistry 1974, 13, 4839-4845.
(43) Thurmond, R. L.; Dodd, S. W.; Brown, M. F. Biophys. J. 1991, 59,
108-113.
(38) Brown, M. F. In Biological Membranes; Merz, K. M., Roux, B.,
Eds.; Birkha¨user: Boston, 1996; pp 175-252.
(39) Davis, J. H. Biochim. Biophys. Acta 1983, 737, 117-171.
(40) Jansson, M.; Thurmond, R. L.; Barry, J. A.; Brown, M. F. J. Phys.
Chem. 1992, 96, 9532-9544.
(44) Douliez, J.-P.; Le´onard, A.; Dufourc, E. J. Biophys. J. 1995, 68,
1727-1739.
(45) Smondyrev, A. M.; Berkowitz, M. L. J. Chem. Phys. 1999, 110,
3981-3985.

12614 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
2VCH
Petrache et al.
2
1
A(Cn:0)
)
(7)
DM cos â
Here VCH represents the volume of a methylene group, and DM ) 2.54
2
Å is the length of a virtual bond between carbons i - 1 and i + 1. As
discussed previously,35 the area factor q ≡ 1/cos â can be ap-
proximated by
q ≈ 3 - 3 cos â + cos2 â
(8)
which involves the first two moments of the orientational distribution
function f (cos â), having odd and even parity, respectively. Both
moments can be calculated from the measured order parameter SCD
.
For each of the segments, the first moment cos â is given by eq 3
above, while the second moment is given by35
1 - 4SCD
cos2 â )
(9)
3
Figure 1. Schematic representation of (n:0-d2n-1)(22:6)PC chemical
structures, for n ) 12, 14, 16, and 18 carbons. As n is varied, the length
mismatch between the (perdeuterated) saturated sn-1 acyl chain and
the docosahexaenoic acid (DHA) sn-2 chain is altered. The 2H-labeled
saturated chain is used to monitor the average properties of the highly
polyunsaturated bilayers in the fluid, liquid-crystalline (LR) state. The
chemical structures are drawn for heuristic purposes only, and do not
show the energy minimized DHA conformations.
With the area factor q obtained from eq 8, we then calculate the average
cross-sectional area per saturated acyl chain as
2VCH
2
A(Cn:0) ) q
(10)
DM
For comparison with the disaturated lipids, we also calculate the
associated Volumetric thickness, defined in relation with the saturated
chain volume V (Cn:0) as
where the sum is now over all carbons from i ) 2 to i ) nC - 1. The
V (Cn:0)
A(Cn:0)
length L(Cn:0) is typically larger than L/C by 1-2 Å, but still less than the
D(Cn:0)
≡
(11)
volumetric thickness DC(n:0)
.
35
We emphasize that the structural parameters D(Cn:0) and AC(n:0) are
based on measurements of the saturated sn-1 chain only. The average
molecular area per chain A(Cn:0) , as well as the directly related
volumetric thickness D(Cn:0), reflects the packing of the saturated (n:0)
chain within the bilayer and accounts only for its contribution to the
hydrocarbon region. We have argued that for disaturated lipid bilayers,
the volumetric thickness D(Cn:0), defined as in eq 11, properly reflects
the hydrophobic span of one monolayer.35 However, even for symmetric
disaturated lipids, D(Cn:0) is not necessarily equal to the average chain
projection onto the bilayer normal, denoted by L(Cn:0), as discussed in
refs 35, 41, and 46. The exact relationship between D(Cn:0) and LC(n:0)
depends on the inter-monolayer packing at the bilayer center and, in
the case of mixed-chain polyunsaturated lipids, on possible length
mismatch between the saturated sn-1 chain and the polyunsaturated
sn-2 chain. Thus, for completeness, we report the results for the
Results
Deuterium NMR Spectroscopy of Docosahexaenoyl Phos-
pholipids in the Liquid-Crystalline State. Solid-state 2H NMR
spectra were measured for the homologous series of mixed-
chain lipids (n:0-d2n-1)(22:6)PC, with n ) 12, 14, 16, and 18,
at temperatures between -10 and 50 °C, as described below.
Based on the chemical structures (see Figure 1), fully extended
conformations (all-trans) of the sn-1 chains show a variable
degree of length mismatch between the saturated sn-1 chain
and the extended polyunsaturated sn-2 chain. The question is
whether such a mismatch persists in fluid bilayers where, due
to thermal disorder, the acyl chains assume a wide range of
conformations. Consequently, one is interested in how the chain
packing is affected by the intrinsic chain mismatch, and,
ultimately, the effects on the bilayer properties. As we show
conventional chain projection, L(Cn:0), together with an additional
quantity, denoted by LC/(n:0)
.
The latter gives the chain extent
35,41
2
next, such questions can be investigated in depth by H NMR
spectroscopy.
Representative solid-state H NMR spectra acquired for the
between the second carbon (i ) 2) and the ω-carbon (i ) nC), calculated
as a sum of segmental projections, Di , along the bilayer normal,
2
saturated sn-1 chain of the mixed-chain, polyunsaturated series
are presented in Figure 2. The powder-type H NMR spectra
nC-1
2
L/C(n:0) ≡ z(2) - z(n )
)
Di
(12)
C
∑
of the randomly oriented, multilamellar dispersions are shown
at the left, and the de-Paked subspectra, corresponding to the θ
) 0° orientation of the bilayer normal with respect to the
magnetic field, are shown at the right. The quality of the spectra
indicates a uniform spherical distribution of bilayer orientations,
characteristic of homogeneous multilamellar lipid dispersions
(so-called powder-type samples). In addition, the 2H NMR
spectra in Figure 2 indicate that the phospholipids undergo
axially symmetric motions about the bilayer normal, with a
distribution of quadrupolar splittings due to the various in-
equivalent methylene and methyl groups of the perdeuterated
saturated acyl chains. Based on studies of specifically deuterated
phospholipids in the LR state,42 it is known that the largest
i)3,5,...
where z(i) represents the average coordinate of carbon i. Note that the
summation in eq 12 is performed over odd values of i only, from i )
3 to nC - 1. The chain extent L/C(n:0) is a well-defined structural
parameter, but it does not account for chain contributions beyond the
end carbons C2 and Cn . To include these contributions, the conventional
C
approach30,35,41 defines the chain length (projection) LC(n:0) as
nC-1
1
2
L(Cn:0)
)
Di + Dn -1
(13)
∑
C
i)2,3,...
(46) Nagle, J. F. Biophys. J. 1993, 64, 1476-1481.

2H NMR of Polyunsaturated Bilayers
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12615
series and varies with the sn-1 acyl length. This is the result of
increased acyl chain disorder (as compared to the disaturated
series), which reduces the number of equivalent carbons with
overlapping resonances (plateau region).
Order Parameter Profiles of Polyunsaturated Phospho-
2
lipids. Further analysis of the H NMR spectral data in terms
of Cs2H bond order parameters allows for a more insightful
comparison of polyunsaturated lipids among themselves and in
relation to the corresponding disaturated lipid bilayers. Figure
(i)
CD
4 shows the order parameters |S | as a function of carbon
index (i) for the perdeuterated, saturated sn-1 chain adjacent to
the sn-2 DHA chain for the homologous series of mixed-chain
phosphatidylcholines in the liquid-crystalline (LR) state. Each
panel shows the data for a single lipid at temperatures ranging
between -5 and 50 °C. The order profiles are typical for
saturated acyl chains: there is an initial plateau region,
corresponding to the acyl chain segments next to the glycerol
and headgroup regions, followed by a monotonic decrease of
the order parameters moving toward the chain ends closer to
the bilayer center. As observed previously for disaturated
lipids,35 the global chain order decreases with increasing
temperature, indicating increased thermal motion within the fluid
lipid bilayers. A large effect of temperature is evident for the
plateau region, which becomes more disordered as the temper-
ature increases. In addition, one can identify a distinct subplateau
region which seems to be the most sensitive to temperature
variations. Especially for the longer (18:0-d35)(22:6)PC lipid,
this corresponds to a change in curvature of the order profiles
near the midchain region with increasing temperature.
2
Figure 2. (a-d) Representative H NMR spectra of the homologous
series of mixed-chain, saturated-polyunsaturated phosphatidylcholines,
(n:0-d2n-1)(22:6)PC, for n ) 12, 14, 16, and 18 carbons. The
phospholipids have a perdeuterated sn-1 acyl chain at the glycerol sn-1
position and a protiated DHA chain at the sn-2 position. Samples were
fully hydrated (67 mM Na phosphate buffer, pH 7.0; 50 wt % H2O)
2
and were in the liquid-crystalline (LR) state at 50 °C. Powder-type H
NMR spectra are shown at the left and the corresponding de-Paked 2H
NMR spectra at the right. As the sn-1 acyl length is increased, i.e.,
additional acyl mass is added to the hydrocarbon region, further
quadrupolar splittings appear, in marked contrast to the behavior of
disaturated phosphatidylcholines (see text).
Further comparison among the homologous phospholipids can
be made at fixed absolute temperature. As shown previously,35
and also seen in Figure 3, for the disaturated lipids the
nonplateau order parameters depend only weakly on the acyl
chain length, while those in the plateau region are higher for
longer chains. Figure 5 shows that the mixed-chain, saturated-
polyunsaturated lipids behave rather differently: at any given
temperature, the nonplateau order parameters decrease as the
acyl length increases, indicating increased disorder at the bilayer
center as the saturated chain becomes longer (cf. also Figure
3). This suggests a broadening of the chain termination
distribution along the bilayer normal, which could in principle
be detectable by small-angle X-ray scattering.47 A second
observation, most noticeable at T ) 50 °C, is that the order
profiles for the different lipids tend to approach each other in
the upper part of the acyl chain, indicating that the DHA chain
imposes a stronger perturbation in this chain region. Similar
quadrupolar splittings correspond to the acyl chain segments
closest to the lipid-water interface (called the plateau region),
followed by a progressive decrease along the chains toward the
terminal methyl groups. We have assumed that the same trend
holds for the mixed-chain, polyunsaturated lipid bilayers in the
fluid, liquid-crystalline (LR) state. The spectral assignments and
the numerical quadrupolar splittings, (∆νQ)|, corresponding to
the θ ) 0° orientation, are summarized in Tables 1-4.
As shown previously by Salmon et al.,30 the 2H NMR spectra
of multilamellar dispersions of (n:0-d2n-1)(22:6)PC lipids differ
both qualitatively and quantitatively from those of disaturated
phospholipids. Specifically, the former appear roughly triangular
in shape (apart from the central methyl component), as opposed
to the rectangular shape characteristic of disaturated phospho-
lipids (not shown).35 Moreover, an inspection of the line shapes
in Figure 2 reveals an interesting behavior with increasing acyl
chain length at fixed absolute temperature. Namely, the largest
quadrupolar splittings increase, while the smaller splittings
including the methyl group splitting decrease with increasing
acyl chain length. The contrasting features of the 2H NMR line
shapes between the disaturated and the mixed-chain, polyun-
saturated phosphatidylcholines are clearly visible in Figure 3,
where we show the variation of the quadrupolar splittings (θ )
0°) with acyl chain length for each series. For the disaturated
series,35 part a, the quadrupolar splittings (∆νQ)| are roughly
constant or increase slightly (positive slope), indicating equal
or higher orientational order of the corresponding segments as
the number of carbons (m) is increased. By contrast, for the
polyunsaturated series, part b, the values of (∆νQ)| for the sn-1
chain are seen to decrease (negative slope), indicating a
reduction in the orientational order as the number of carbons
(n) is increased. Note also that the number of resolved
quadrupolar splittings is larger in the case of the polyunsaturated
2
results on the shape of the H NMR order profile have been
obtained by Holte et al.25 using a steroyl (18:0) chain as a probe
of the mixed-chain lipid systems. As we will show, further data
reduction allows us to better resolve the differences between
the two series. This will be discussed on the basis of a uniVersal
chain packing profile35 that allows for a unified description of
lipid bilayers.
Before discussing the chain packing profiles, however, let
us first examine the global effect of the DHA chain on the
bilayer structure. In this regard, there are two natural questions
to address: (i) How is the lateral density (area per lipid) affected
by DHA for the different sn-1 chain lengths? (ii) What is the
variation of the bilayer thickness? We shall address these
questions using a statistical model35 to reduce the C-2H bond
order parameters to structural parameters, including average
cross-sectional areas and average chain lengths. Specifically,
(47) Petrache, H. I.; Tristram-Nagle, S.; Nagle, J. F. Chem. Phys. Lipids
1998, 95, 83-94.

12616 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Petrache et al.
2
Table 1. De-Paked H NMR Spectral Assignments and (∆νQ)| Splittings for (12:0-d23)(22:6)PC in the Liquid-Crystalline (LR) Phase
(∆νQ)| /kHza
carbon
peak
i
-9 °C
-5 °C
0 °C
5 °C
10 °C
15 °C
20 °C
30 °C
50 °C
A
B
C
D
E
F
2-6
7
8
64.1
64.1
58.6
50.9
44.1
33.2
11.1
57.6
53.3
50.6
43.4
37.3
26.9
9.5
55.9
51.5
48.2
41.0
35.4
25.5
8.6
53.1
48.8
45.4
38.4
32.9
23.6
8.0
51.5
46.8
43.5
36.7
31.2
22.3
7.5
49.9
45.0
41.7
35.1
29.7
21.0
6.8
47.8
43.3
40.2
33.4
28.2
20.0
6.6
44.7
39.5
36.5
30.2
25.5
17.8
5.7
39.6
34.7
31.7
25.9
21.5
15.0
4.6
9
10
11
12
G
a Quadrupolar splitting corresponding to θ ) 0° orientation of bilayer normal with respect to magnetic field.
2
Table 2. De-Paked H NMR Spectral Assignments and (∆νQ)| Splittings for (14:0-d27)(22:6)PC in the Liquid-Crystalline (LR) Phase
(∆νQ)| /kHza
carbon
peak
i
0 °C
15 °C
22.5 °C
25 °C
30 °C
35 °C
40 °C
45 °C
50 °C
A
B
C
D
E
F
G
H
I
2-4
5-7
8
59.7
54.8
54.8
48.6
44.4
37.2
31.7
23.0
7.4
53.6
49.4
47.8
41.9
37.7
31.4
26.2
18.7
6.0
51.4
47.6
46.3
39.8
35.6
29.0
24.5
17.0
5.4
50.4
46.6
44.4
38.5
34.6
28.2
23.3
16.5
5.4
48.5
44.5
42.3
36.8
32.8
27.0
22.2
15.7
4.9
47.6
44.1
41.4
35.6
31.6
25.8
21.1
14.7
4.5
46.4
42.6
39.6
34.3
30.7
24.8
20.3
14.3
4.3
45.0
41.3
37.8
32.8
29.2
23.6
19.4
13.4
4.1
43.5
39.3
36.3
31.5
27.5
22.3
18.1
12.8
3.8
9
10
11
12
13
14
a Quadrupolar splitting corresponding to θ ) 0° bilayer orientation.
2
2
Table 3. De-Paked H NMR Spectral Assignments and (∆νQ)|
Table 4. De-Paked H NMR Spectral Assignments and (∆νQ)|
Splittings for (18:0-d35)(22:6)PC in the Liquid-Crystalline (LR)
Phase
Splittings for (16:0-d31)(22:6)PC in the Liquid-Crystalline (LR)
Phase
(∆νQ)| /kHza
(∆νQ)| /kHza
carbon
carbon
peak
i
-5 °C 0 °C 5 °C 10 °C 15 °C 30 °C 50 °C
peak
i
-5 °C 0 °C 5 °C 10 °C 15 °C 20 °C 30 °C 50 °C
A
B
C
D
E
F
G
H
I
2-5
6-7
8
67.1 61.5 58.7 57.7
67.1 61.5 58.7 54.5
67.1 61.5 58.7 54.5
61.8 54.8 52.2 49.3
58.8 53.0 48.4 45.8
52.2 44.9 42.3 40.0
47.4 40.9 37.7 35.6
39.8 34.0 31.5 29.5
33.7 28.6 26.5 24.6
24.7 21.1 19.3 18.0
55.8
52.2
52.2
47.1
43.8
37.9
33.7
27.8
23.1
16.8
5.2
51.1
46.2
46.2
40.9
37.7
32.2
28.2
22.9
18.9
13.7
4.1
45.2
41.9
39.4
35.2
31.8
26.5
23.0
18.5
15.1
10.8
3.0
A
B
C
D
E
F
G
H
I
2-7
8
9
65.2 62.4 59.1 56.7 54.7 53.4 51.2 46.4
65.2 62.4 59.1 56.7 54.7 53.4 51.2 42.6
65.2 57.8 53.4 51.5 49.5 47.5 43.9 37.5
59.4 55.3 51.2 49.0 45.5 44.3 40.8 34.5
54.5 49.8 45.9 43.5 40.5 39.0 35.6 29.8
50.9 45.8 41.8 39.8 37.1 35.4 32.0 26.3
44.8 40.2 36.5 34.5 32.1 30.2 26.9 22.1
41.2 36.4 32.4 30.9 28.3 26.8 23.6 18.7
34.7 30.7 27.3 25.9 23.6 22.2 19.7 15.7
29.7 26.0 23.2 21.8 19.9 18.7 16.4 12.9
9
10
11
12
13
14
15
16
17
18
10
11
12
13
14
15
16
J
K
J
K
L
7.8
6.7
6.1
5.6
22.6 19.6 17.4 16.3 14.9 14.1 12.3
7.3 6.3 5.4 5.2 4.6 4.3 3.7
9.8
2.8
a Quadrupolar splitting corresponding to θ ) 0° bilayer orientation.
a Quadrupolar splitting corresponding to θ ) 0° bilayer orientation.
the analysis of the order parameters is done in terms of acyl
chain orientational distributions, within the framework of a
Here nn:0 denotes the number of saturated (n:0) chains and n22:6
the number of polyunsaturated (22:6) chains, where V (Cn:0) and
V (C22:6) are the associated partial molecular volumes. Integra-
tion of eq 15 holding the component ratios constant leads to
the following sum rule for the mixed-chain bilayer:
2
mean-torque model35 previously used for disaturated lipid H
NMR data.
Structural Parameters of Mixed-Chain Polyunsaturated
Phosphatidylcholines. For each of the lipids in the (n:0-d2n-1)-
2
(22:6)PC series, we have used the H NMR order parameters
VC ) V (Cn:0)nn:0 + V C(22:6)n22:6
(16)
of the perdeuterated, saturated sn-1 chain (see Tables 1-4 for
the measured quadrupolar splittings) to calculate its molecular
cross-sectional area A(Cn:0) and molecular volumetric thickness
D(Cn:0) as a function of temperature and acyl chain length n (see
Methods). In analogy with the concept of partial molar quanti-
ties, one can also introduce the partial molecular volumes of
the saturated and polyunsaturated acyl components as follows,
where the bilayer is treated as a continuum medium. The total
differential of the hydrocarbon chain volume is
where nn:0 ) n22:6. The volumetric thickness of the bilayer can
then be written in terms of the partial volumes and the total
interfacial area, A ) A(Cn:0) + AC(22:6) , as
VC
A
V (Cn:0) + V (C22:6)
A(Cn:0) + A(C22:6)
DC ≡
)
(17)
Note that this definition for the hydrocarbon thickness (per
monolayer) DC is consistent with the bilayer structure determi-
nation from small-angle X-ray scattering data.35,47
∂VC
∂VC
dVC )
dnn:0
+
dn22:6
(14)
(15)
(
)
(
)
∂nn:0
∂n22:6
T,P
T,P
Using the 2H NMR data for the saturated sn-1 chain, we can
calculate the partial cross-sectional area A(Cn:0) of this chain,
≡ V (Cn:0) dnn:0 + V C(22:6) dn22:6

2H NMR of Polyunsaturated Bilayers
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12617
(i)
Figure 4. Order parameters, |S |, of acyl chain C-2H bonds as a
CD
function of segment index i, for the (n:0-d2n-1)(22:6)PC series in the
liquid-crystalline (LR) state. Each panel shows the data for one member
of the series at various temperatures. Note that the overall order
decreases with increasing temperature, and that a larger effect is
observed on the initial carbon segments (plateau region). Consequently,
the order profiles do not scale uniformly with temperature, implying a
differential thermal expansion along the acyl chains.
Figure 3. Dependence of 2H NMR quadrupolar splittings (θ ) 0°) on
perdeuterated sn-1 acyl chain length nC (≡ m or n) for (a) the disaturated
phosphatidylcholine series, di(m:0-d2m-1)PC, and (b) the mixed-chain
(n:0-d2n-1)(22:6)PC series in the LR state. Data in part a are from ref
35. Note that for the disaturated PC series, part a, the quadrupolar
splittings (∆νQ)| remain approximately constant or increase slightly with
increasing sn-1 acyl chain length at the same absolute temperature.
The new splittings are near the largest (plateau) values and increase
with the acyl length. By contrast, for the mixed-chain polyunsaturated
series, part b, the quadrupolar splittings decrease with increasing sn-1
acyl carbons at the same absolute temperature, so that the new splittings
at larger values are accompanied by additional intermediate splittings.
Consequently, the packing of the (saturated) sn-1 acyl chains must differ
substantially in the two cases.
as explained in Methods. One needs additional information, viz.
A(C22:6) , to determine the total lipid cross-sectional area A
and the hydrocarbon half-thickness DC, as we show later. Note
that for the disaturated series, because of the chain symmetry,
Figure 5. Same data as in Figure 4, but now each panel compares the
different lipids at the same absolute temperature. As the number of
carbons in the sn-1 acyl chain increases, the order parameters of the
initial segments increase, while those of the end segments decrease,
causing a noticeable change of shape of the order profile.
one has VC ) 2V (n:0) and A ) 2 AC(n:0) , and consequently DC
C
) D(Cn:0). The situation is more complicated for the mixed-
chain bilayers, where the hydrocarbon thickness per monolayer,
DC, may differ from D(n:0), depending on the structural param-
C
with previous results for disaturated lipids, where the total area
per molecule A ) 2 A(Cn:0) corresponds to both acyl chains. A
summary of the structural results is provided in Table 5.
From part a of Figure 6 we note that the partial volumetric
thickness D(Cn:0) decreases with temperature, as for the disatu-
rated lipids, while at fixed absolute temperature, D(Cn:0) in-
creases with increasing number of carbons in the acyl chain
(larger n). Moreover, the overall variation with temperature is
similar for the two homologous series, with the polyunsaturated
eters of the sn-2 chain, viz. V (C22:6) and A(C22:6) , as indicated by
eq 17. However, it is still useful to consider the (partial)
volumetric thickness due to the sn-1 chain, D(Cn:0), as defined by
eq 11, for comparison with previous results for the disaturated
series.35 The values of DC(n:0) and A(Cn:0) for the mixed-chain
polyunsaturated bilayers, obtained using eq 11 and eq 10
respectively, are shown in Figure 6. The saturated chain volumes
V (Cn:0) were calculated using the temperature dependence given
by VCH ≈ V 0CH + RCH (T - 273.15 K), with empirical pa-
2
2
2
rameters V 0CH ) 26.5 Å3 and RCH ) 0.0325 Å3 K-1 35
In
.
(48) Nagle, J. F.; Wilkinson, D. A. Biophys. J. 1978, 23, 159-175.
2
2
(49) Petrache, H. I.; Feller, S. E.; Nagle, J. F. Biophys. J. 1997, 72, 2237-
addition, it was assumed that the terminal methyl volume VCH
3
2242.
48-50
is roughly twice the methylene volume VCH
.
The area
2
(50) Armen, R. S.; Uitto, O. D.; Feller, S. E. Biophys. J. 1998, 75, 734-
744.
results in Figure 6 are plotted as 2 A(Cn:0) to enable comparison

12618 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Petrache et al.
comes from the polar headgroups rather than the acyl length.
The fact that the area per lipid decreases with increasing acyl
chain length might appear counterintuitive at first, considering
the repulsion generated by chain confinement. The explanation,
however, resides in the fact that longer chains (more mass
added) have increased van der Waals attraction that overcomes
repulsive entropic effects. In effect, the longer chains keep the
headgroups closer together on average, in agreement with the
higher melting temperatures.35
Material Properties of DHA Lipids Revealed by the
Saturated Chain Behavior. To better quantify the variation
of the structural parameters with both the thermodynamic
variable T and the carbon number n, let us consider the following
total differential:35
1
V
1 ∂V
V ∂T n,P
1 ∂V
V ∂n T,P
dV )
dT +
dn
(
)
(
)
Figure 6. Structural parameters of the saturated sn-1 chain, as a
function of temperature (a,b) and acyl chain length (c,d). Parts a and b
show results for the partial volumetric thickness, D(Cn:0), and twice the
partial area per saturated chain, 2 A(Cn:0) , respectively, for the mixed-
chain, (n:0-d2n-1)(22:6)PC, lipids. The thick lines show corresponding
results for the disaturated series, di(m:0-d2m-1)PC, taken from ref 35.
At a given absolute temperature, acyl polyunsaturation at the sn-2
position generates a smaller hydrocarbon thickness and a larger cross-
sectional area. Parts c and d show semilogarithmic plots of the relative
variation of D(Cn:0) and 2 A(Cn:0) for the mixed-chain polyunsaturated
series. The reference values DM and AM represent the maximum
segmental projection onto the bilayer normal and the corresponding
≡ R dT + ν dn
(18)
The above total differential defines the total volumetric expan-
sion coefficients R and ν under conditions of constant bulk
pressure P. The coefficient R represents the thermal expansion,
while ν describes the relative volume change due to additional
carbons, and reduces to ν ) 1/n. Using the relationship V C(n:0)
) DC(n:0) AC(n:0) (and suppressing the superscript (n:0) for
clarity), we can write
dV
dDC d AC
C ) d ln VC )
+
minimum area (AM ≡ 4VCH /DM), respectively. The effect of increasing
VC
DC
2
AC
the saturated acyl length n (mass added to the system) is reflected more
on the partial volumetric thickness, D(Cn:0), than on the cross-sectional
) d ln DC + d ln AC
(19)
area, A(Cn:0)
.
This expression allows for a decomposition of the two bulk
coefficients R and ν into contributions from the relative thickness
variation (parallel component) and the relative area variation
(perpendicular component):
curves being slightly shifted to lower D(Cn:0) values (by roughly
0.5 Å). Alternatively, the two sets of curves can be made to
overlap by a relative shift along the temperature axis of about
15-20 °C. This shows that at the same absolute temperature,
the saturated sn-1 chains of the DHA lipids are more disordered
than for the corresponding disaturated series, which results in
∂ln VC
∂ln DC
∂ln AC
∂T
R ≡
)
+
(
)
(
)
(
)
∂T n,P
∂T n,P
n,P
smaller values of DC(n:0) and larger cross-sectional areas A(Cn:0)
.
≡ R| + R
(20)
One should also note that the area differences among the
polyunsaturated lipids are significantly smaller than the spread
of cross-sectional areas of their disaturated counterparts, as
shown in part b of Figure 6. With the exception of the (12:0)-
(22:6)PC bilayer, the polyunsaturated lipids have roughly the
same area per saturated chain, especially in the temperature
range below 30 °C. This important result suggests that the short
(12:0) chain is a special case, namely there is a qualitative
difference between the way the DHA chain packs with the short
(12:0) sn-1 chain and with the other longer saturated chains.
This aspect will be further considered in the Discussion section.
Let us next turn to the acyl chain length dependence of the
structural parameters for the mixed-chain, polyunsaturated
bilayers. The relative variations of D(Cn:0) and AC(n:0) with the
saturated acyl chain length n are shown in parts c and d of Figure
6. The advantage of plotting the relative (dimensionless)
variations in semilogarithmic fashion is that the effects of the
chain length on the thickness and area per chain can now be
directly compared. It can be seen that as mass is added to the
saturated sn-1 chain, the thickness D(Cn:0) increases, while the
area per chain A(Cn:0) undergoes a slight contraction. Clearly,
there is a more pronounced effect of the additional acyl mass
on the bilayer thickness, rather than on the area. Consequently,
for the polyunsaturated series, as found previously for the
disaturated series,35 the major influence on the area per lipid
and
∂ln VC
∂ln DC
∂ln AC
∂n
ν ≡
)
+
(
)
(
)
(
)
∂n T,P
∂n T,P
T,P
≡ ν| + ν
(21)
The above expansion coefficients, calculated by interpolation
of the structural parameters from Figure 6, are given in Table
5. Note that the isobaric thermal expansion of the saturated sn-1
chain, quantified by R| and R , is similar for both the disaturated
and the mixed-chain polyunsaturated series. The thickness
D(Cn:0) decreases with temperature by a fractional change R| on
the order of -2 × 10-3 to -4 × 10-3 K-1, while the partial
cross-sectional area A(Cn:0) increases with R between 3 ×
10-3 and 5 × 10-3 K-1. The reason the thermal expansion
coefficients for the disaturated and polyunsaturated series are
similar is evident from Figure 6. While the magnitude of the
structural parameters varies between the two series, the thermal
variation is the same. In particular, a relative temperature shift
of 15-20 °C would make the two families of curves overlap.
Discussion
Effect of Docosahexaenoyl Acyl Chains on Bilayer Prop-
erties. Polyunsaturated lipids are found as constituents of cellular

2H NMR of Polyunsaturated Bilayers
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12619
Table 5. Summary of Structural Results for Saturated sn-1 Chain of (n:0)(22:6)PC Lipids in the Liquid-Crystalline (LR) State
D(Cn:0)/Åb
A(Cn:0) /Å2 c
R| /10-3K-1 d
R /10-3K-1 e
ν| /10-2 f
ν /10-2 g
L/C(n:0)/Åh
L(Cn:0)/Åi
a
n
T/°C
Sp
12
0
30
50
0
30
50
0
30
50
0
30
50
0.223
0.179
0.158
0.238
0.194
0.174
0.246
0.204
0.181
0.249
0.205
0.185
11.2
10.2
9.6
28.3
32.4
35.3
27.6
31.2
33.6
27.2
30.6
33.0
27.1
30.5
32.6
-3.1
-3.2
-3.4
-3.6
-2.6
-2.7
-3.9
-2.6
-2.7
-4.1
-2.1
-1.9
4.3
4.4
4.6
4.8
3.8
3.9
5.1
3.8
3.9
5.4
3.3
3.1
11.3
13.5
13.4
9.3
10.8
10.7
7.9
9.0
8.9
6.8
7.7
-3.0
-5.2
-5.0
-2.1
-3.7
-3.5
-1.6
-2.7
-2.6
-1.3
-2.2
-2.1
8.8
7.9
7.5
10.6
9.6
9.0
12.5
11.2
10.5
14.3
12.6
11.9
10.4
9.3
8.8
14
16
18
13.5
12.3
11.7
15.6
14.4
13.6
17.6
16.2
15.5
12.2
10.9
10.2
14.1
12.5
11.6
15.8
13.9
13.0
7.6
a Plateau order parameter values used to calculate the area factor q (eq 8). b Volumetric thickness (eq 11). c Cross-sectional area (eq 10). d Thermal
expansion coefficient along the bilayer normal (eq 20). e Thermal expansion coefficient in the membrane plane (eq 20). f Volumetric expansion
coefficient along the bilayer normal due to additional carbons (eq 21). g Volumetric expansion coefficient in the membrane plane due to additional
carbons (eq 21). h Average chain projection calculated using eq 12. i Average chain projection calculated using eq 13.
membranes, and thus knowledge of their structural properties
is needed to understand their roles in biological membrane
function.12,13 One hypothesis regarding the role of polyunsatu-
rated lipids in biological functions (such as vision) is based on
the distinct physical properties of lipid bilayers conferred by
the presence of the polyunsaturated chain.14,23,53 For instance,
a native-like lipid environment involving the presence of
phospholipid DHA chains is sufficient, although not necessary
for native-like visual function of rhodopsin in the membrane.53-56
Rather, biophysical properties of the membrane bilayer appear
to be involved in the meta I-meta II transition of rhodopsin,
the signal transducing event of visual excitation. As a rule, the
properties of lipid bilayers are governed by a balance of forces
acting on both the headgroup (hydrophilic) and hydrocarbon
(hydrophobic) regions. We have observed that introduction of
a polyunsaturated DHA chain at the sn-2 position increases the
cross-sectional area per lipid headgroup in liquid-crystalline
bilayers.30 Consequently, the force balance is shifted according
to the modified acyl chain interactions, inducing a change in
the curvature stress involving the aqueous interface. This aspect
has been addressed in the literature with regard to the “spon-
taneous curvature” induced by the effective lipid shape.23,57,58
The specific structural and dynamical properties of polyun-
saturated lipids are best revealed by comparison with saturated
bilayers.25,30 In this regard, a deeper understanding can be
acquired by comparison of the members of a homologous series,
rather than individual lipids.22,25,34 Therefore, we have analyzed
the properties of the polyunsaturated (n:0)(22:6)PC bilayers in
relation to the disaturated di(m:0)PC series.34 Based on chemical
structures (Figure 1), the mixed-chain (n:0)(22:6)PC series is
characterized by a systematic variation of the acyl chain
mismatch as the sn-1 acyl length is varied. While this intrinsic
mismatch might not be fully preserved in bilayers formed at
finite temperatures, it is expected to influence the bilayer
properties.
Length Mismatch and Area Differences between the
Saturated and Polyunsaturated Acyl Chains in Fluid Bilay-
ers. Let us first note that the polyunsaturated DHA chain has a
larger molecular volume than any of the saturated chains
considered in this study. At 30 °C, for example, one estimates
a DHA molecular volume of about 526.4 Å3, using VCH
)
2
27.475 Å3 (see above) and assuming a ratio VCH /VCH of
2
≈1.31.59 With V C(22:6) ) 526.4 Å3, the ratio between the sn-1
and the sn-2 chain volumes in the mixed-chain, saturated-
polyunsaturated series is found to vary from 0.63 for n ) 12 to
0.94 for n ) 18. Given that the DHA chain has a large molecular
volume, the question is whether the DHA chain contributes to
enlarging the cross-sectional area, increasing the bilayer thick-
ness, or both. There are only a few reported measurements on
the interfacial area of DHA-containing lipid bilayers. For fully
hydrated (18:0)(22:6)PC at 30 °C, Koenig et al.51 give a value
of A ) 69.2 Å2 from a combined 2H NMR and X-ray analysis.
Subtracting our results for the sn-1 partial area A(C18:0) ) 30.5
Å2 (cf. Table 5), we obtain A(C22:6) ) 38.7 Å2 for the DHA
chain. This suggests that the DHA molecular cross-sectional
area is larger than that of the adjacent (18:0) chain by 8-9 Å2.
We can therefore conclude that the larger cross-sectional area
for the mixed-chain (n:0)(22:6)PC bilayer is due to both
chains: not only is the sn-1 chain more disordered, but also
the DHA chain is larger.25,52 Note that this decomposition into
partial areas associated with each chain does not make any
particular assumption about the shape of the two chains. The
results for the partial areas give a ratio A(C18:0) / AC(22:6) ≈ 0.79.
Since this is not equal to the volumetric ratio calculated above,
it implies that a mismatch of chain length projections along the
bilayer normal is also present. To explore this possibility, we
consider the DHA volumetric thickness defined as DC(22:6)
≡
V (C22:6)/ AC(22:6) . With the volume and area values from above,
we obtain D(C22:6) ≈13.6 Å, with an error estimate of (0.3 Å.
Compared to D(C18:0) ) 16.2 Å for the saturated (18:0) chain
(51) Koenig, B. W.; Strey, H. H.; Gawrisch, K. Biophys. J. 1997, 73,
1954-1966.
(Table 5), it shows that the DHA volumetric thickness, DC(22:6)
,
(52) Applegate, K. R.; Glomset, J. A. J. Lipid Res. 1986, 27, 658-680.
(53) Wiedmann, T. S.; Pates, R. D.; Beach, J. M.; Salmon, A.; Brown,
M. F. Biochemistry 1988, 29, 6469-6474.
is lower by 2-3 Å. As mentioned above, the hydrophobic
thickness (per monolayer) DC is calculated from eq 17 using
the lipid total hydrocarbon volume and total cross-sectional area.
For (18:0)(22:6)PC at 30 °C we obtain DC ) 14.8 Å, which is
(54) Brown, M. F.; Gibson, N. J. In Essential Fatty Acids and
Eicosanoids; Sinclair, A., Gibson, R., Eds.; American Oil Chemist’s Society
Press: Champaign, IL, 1992; pp 134-138.
an intermediate value between D(C22:6) ) 13.6 Å and D(C18:0)
)
(55) Gibson, N. J.; Brown, M. F. Biochemistry 1993, 32, 2438-2454.
(56) Botelho, A. V.; Gibson, N. J.; Wang, Y.; Thurmond, R. L.; Brown,
M. F. Biophys. J. 2001, 80, 2466.
16.2 Å, as expected from eq 17. The relationship between DC
and the single-chain volumetric thicknesses, D(Cn:0) and DC(22:6)
,
(57) Gruner, S. M.; Shyamsunder, E. Ann. N. Y. Acad. Sci. 1991, 625,
685-697.
(58) Gawrisch, K.; Holte, L. L. Chem. Phys. Lipids 1996, 81, 105-116.
(59) Marsh, D. CRC Handbook of Lipid Bilayers; CRC Press: Boca
Raton, FL, 1992.

12620 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Petrache et al.
can be better seen by rewriting eq 17 as
A(Cn:0)
A
A(C22:6)
DC )
D(Cn:0)
+
D(C22:6)
(22)
A
This expression shows that the hydrophobic thickness, DC, can
be regarded as an average of single-chain volumetric thicknesses,
weighted by the partial areas. In this sense, the single-chain
volumetric thicknesses play the role of partial quantities with
respect to the hydrocarbon thickness DC.
Note that the partial thicknesses D(Cn:0) and DC(22:6) may, in
general, differ from the geometric chain length projections,
denoted by LC(n:0) and LC(22:6), calculated as a sum over segmental
projections (see Methods). As discussed in detail else-
where,35,41,46 due to the highly disordered acyl chains in the
fluid state, the hydrophobic thickness DC depends on the chain
packing at the bilayer center (inter-monolayer packing) and
differs from the average chain length L(Cn:0). This is true for
both the like-chain, disaturated bilayers and for the mixed-chain,
saturated-polyunsaturated bilayers. For example, as shown in
Table 5, for the (18:0)(22:6)PC at 30 °C, we estimate for the
saturated chain length a value of L(C18:0) ) 13.9 Å, to be
compared to D(C18:0) ) 16.2 Å. A discussion of chain length
mismatch would more naturally involve the two chain length
projections L(Cn:0) and L(C22:6), but the latter cannot be estimated
from the current data. As an alternative, we have considered
the partial volumetric thicknesses D(Cn:0) and DC(22:6). These can
both be determined and conveniently used to characterize the
structure of mixed-chain polyunsaturated bilayers in comparison
with the disaturated series and X-ray data.35,47 In particular, for
like-chain disaturated bilayers, the volumetric thicknesses of the
two acyl chains are expected to be the same. This is no longer
true for the mixed-chain lipids, as we discuss next.
To discuss how the chain mismatch varies with the number
of sn-1 carbons (n), let us assume a constant DHA area at fixed
temperature. This is reasonable since only a slight area contrac-
tion is expected as hydrocarbon mass is added in the hydro-
carbon region (see above). This means that the value of
A(C22:6) ) 38.7 Å2 obtained for n ) 18 is an underestimate for
shorter chains. However, the underestimate is more likely to
be on the order of 1 Å2, even for the extreme case of n ) 12
studied here. Constant A(C22:6) would then imply constant
volumetric thickness DC(22:6) across the series at a given abso-
lute temperature. In Figure 7, the estimated D(C22:6) value
(assumed constant) for the (n:0)(22:6)PC series at 30 °C is
compared with the saturated chain volumetric thickness DC(n:0)
and with the hydrophobic thickness per monolayer, DC. The
results suggest that the volumetric thickness mismatch between
the sn-1 and the sn-2 chains is minimal for n ) 14 and n ) 16,
and crosses over in between. Interestingly, the value of DC )
13.0 Å for (14:0)(22:6)PC from Figure 7 is practically the same
as the hydrocarbon thickness per monolayer for di(14:0)PC,
obtained previously.35
Figure 7. Volumetric thickness results for the (n:0-d2n-1)(22:6)PC
series at 30 °C, showing the partial volumetric thicknesses D(Cn:0) and
D(C22:6) due to the individual chains, and the total hydrocarbon half-
thickness (i.e., thickness per monolayer) DC (eq 17). The asymmetric
distribution of mass along the lipid long axis generates an acyl length
mismatch between the saturated sn-1 and the polyunsaturated sn-2
chains, which crosses over between n ) 14 and n ) 16. It follows that
the chain packing at the bilayer center is qualitatively different for the
members of the homologous series.
tions are dominated by two preferred configurations, denoted
as angle-iron and helical. These conformations have been
considered by Applegate and Glomset52 in the context of energy-
minimized (optimal) DHA chain packing. Only recently have
renewed computational efforts been directed toward polyun-
saturated phospholipids in the fluid state, in an attempt to
determine the accessible DHA states at biologically relevant
temperatures. Huber et al.60 have shown that, while DHA has a
tendency to adopt extended conformations (helical, angle-iron)
involving triene sequences, a large number of other disordered
conformations are significantly populated, including back bend-
ing (chain upturns), all of which loosen the chain packing and
increase the lipid cross-sectional area. Even at low temperatures,
Applegate and Glomset52 found a wide range of DHA cross-
sectional areas, between 25.8 and 32.9 Å2, depending on the
chain arrangement. Angle-iron DHA shapes tended to give lower
values of the cross-sectional area than helical conformations.
In addition, somewhat lower cross-sectional areas were found
for DHA chains packed together with saturated chains in all-
trans configurations,52 with an optimum DHA area between 24.3
and 27.0 Å2. Noting that optimal packing of all-trans saturated
chains occurs at roughly 20 Å2 per chain,52,61 this finding
suggests that, in its energy-minimized state, the DHA chain
occupies a lateral area which is larger by about 4-7 Å2. It is
therefore plausible that, as the temperature is increased, the area
difference increases to 8-9 Å2 at 30 °C, as we find in this study.
In addition to altering the bilayer structure, substitution of
the polyunsaturated DHA chain at position sn-2 also leads to a
decreased cooperativity of the main phase transition, as shown
2
by both H NMR34 and Raman spectroscopy.62 Furthermore,
the magnitude of the effect is found to differ for the various
sn-1 saturated acyl lengths. Specifically, the case of 12 carbon
acyl chains (nC ) 12) is close to an intersection point between
the melting curve of the disaturated and mixed-chain DHA lipids
and shows the largest hysteresis between cooling and heating
melting temperatures. By contrast, the nC ) 18 lipids show the
largest reduction of melting temperature upon unsaturation and
At this juncture, the DHA chain conformations deserve some
attention. We have found that at 30 °C, the DHA chain occupies
a partial cross-sectional area that is larger by 8-9 Å2 than the
saturated chain area. How can one interpret this result in terms
of DHA conformations? Unfortunately, the specific DHA
conformations and acyl packing are harder to determine.
Typically, detailed molecular dynamics simulations are needed
to explore the complex configurational space generated by the
polyallylic sequence of alternating cis double bonds separated
by single methylene carbons.60 In principle, the DHA conforma-
(60) Huber, T.; Rajamoorthi, K.; Kurze, V. F.; Beyer, K.; Brown, M. F.
J. Am. Chem. Soc., in press.
(61) Sun, W.-J.; Tristram-Nagle, S.; Suter, R. M.; Nagle, J. F. Biophys.
J. 1996, 71, 885-891.
(62) Litman, B. J.; Lewis, E. N.; Levine, I. W. Biochemistry 1991, 30,
313-319.

2H NMR of Polyunsaturated Bilayers
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12621
Figure 8. Comparison of order profiles of mixed-chain, poly-
unsaturated (n:0-d2n-1)(22:6)PC lipids (symbols) with disaturated
di(m:0-d2m-1)PC lipids (solid lines) at T ) 50 °C for chain lengths nC
(≡ m or n) between 12 and 18 carbons. The C-2H bond order
parameters are plotted as a function of reverse carbon number, ending
with the ω carbon (terminal methyl). Note that for the disaturated lipids,
the nonplateau segments essentially trace a common curve. In addition,
the (12:0-d27)(22:6)PC data fall on this common saturated curve.
However, the order profiles for the other mixed-chain, polyunsaturated
lipids show a dramatic change of shape, viz. curvature, as the length
of the sn-1 chain increases. (Similar behavior is observed at other
temperatures.)
minimal hysteresis. How can we interpret these systematic
variations in the context of the new findings reported here?
2
In Figure 8, we plot the H NMR order parameter profiles
for both the disaturated and the mixed-chain polyunsaturated
series at 50 °C in reverse order; viz., all curves are shifted to
coincide at the ω-carbon. (Note that di(18:0)PC is not in the
fluid state at this temperature, and it is therefore not included.)
Plotted this way, the experimental data facilitate a direct, model-
free comparison of the two homologous series. As can be seen,
the order profiles for the end acyl segments of the disaturated
series trace a common curve, from which the polyunsaturated
series deviates progressively for longer sn-1 chains. One can
conclude that the different sn-1 saturated acyl chains pack
qualitatively differently with the adjacent sn-2 DHA chain. The
n ) 12 profile nearly coincides with the disaturated profile,
while the n ) 18 case shows the largest deviation. In agreement
with this result, Holte et al.25 found that addition of three or
more double bonds into the sn-2 chain significantly changed
the order profile of the 18:0-d35 sn-1 chain, and that the change
was not homogeneous along the chain. Here we show that the
conclusion is valid for the whole (n:0)(22:6)PC series, and that
the effect is progressively reduced for shorter sn-1 chains. Quite
interestingly, one can still find a common characteristic feature
for this series, as we show next.
Figure 9. Mapping of 2H NMR order profiles into chain packing
curves, z(i) , obtained as a sum over acyl chain segmental projections
onto the bilayer normal. (a) Saturated sn-1 chains of mixed-chain,
polyunsaturated (n:0)(22:6)PC series at 50 °C. (b) Disaturated
di(m:0)PC series at 65 °C. The curves z(i) are plotted as a function of
reverse carbon index i′ ≡ nC - i + c, where nC denotes the chain
length (m or n) and c is a relative shift. A value of c ) 2 was chosen
in parts a and b for all the lipids shown. In part c, the packing curves
for the mixed-chain (n:0)(22:6)PC lipids are adjusted by shifting the
individual curves along both the horizontal and the vertical axes. The
resulting curve represents a universal packing profile for the saturated
sn-1 acyl chain adjacent to a DHA chain at position sn-2. This clearly
shows there is a midchain region of similar disorder (neutral volume)
for all four mixed-chain, polyunsaturated lipids considered, with the
longer saturated sn-1 chains having increased order toward the
headgroup region and decreased order near the bilayer center.
shift value. For example, choosing c ) 2 reverses the indexing,
giving i′ ) 2 for the ω carbon and i′ ) nC for the C2 carbon.
The chain packing profiles for the polyunsaturated series at
50 °C and the disaturated series at 65 °C are shown in Figure
9, parts a and b, respectively. (The value of T ) 65 °C was
selected for the disaturated series in order to have all four
bilayers in the fluid state; similar behavior is seen at other
temperatures.) From part a of Figure 9, it can be observed that
the polyunsaturated series exhibits distinct packing profiles for
each lipid, in contrast to the behavior of the disaturated lipids,
which follow a common curve as shown in part b. For the
saturated sn-1 chains, adjacent to a sn-2 DHA chain, part a,
proceeding along the chain from the ω-carbon toward the lipid
headgroup, the order increases less for longer chains. In parts
a and b of Figure 9, a common value of c ) 2 for the index
shift was chosen for all lipids, such that all curves are coincident
at the ω-carbon. With a closer inspection, one can find that all
packing curves for the mixed-chain lipids in part a are actually
sampled from a common profile, as shown in part c of Figure
9. Here, the curves from part a have been shifted along both
axes, taking advantage of the fact that both the carbon average
Universal Chain Packing Curve for Mixed-Chain Poly-
unsaturated Phosphatidylcholines in the Fluid State. We shall
now compare the disaturated and the polyunsaturated series by
considering their acyl chain packing profiles, as explained below.
The mean-torque model35 used for calculation of the partial
cross-sectional area per chain A(Cn:0) can be used to determine
average projections for each carbon segment (cf. Methods). The
resulting segmental projections permit the calculation of average
carbon positions, z(i) , relative to an arbitrary point chosen as
reference. The resulting chain packing curves yield insight into
the acyl chain geometry and facilitate direct comparison between
different bilayer structures. We have previously shown35 that,
for the case of disaturated lipids, it is convenient to choose the
methyl (ω) carbon as a reference and proceed along the chain
toward the headgroup region. For convenience, the carbon atoms
i are reindexed as i′ ) nC - i + c, where c denotes an arbitrary

12622 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Petrache et al.
the polyunsaturated series. For the like-chain disaturated phos-
phatidylcholines, part a, the acyl chain packing near the bilayer
center changes very little with the addition of carbon segments,
whereas the carbon segments next to the headgroups become
more ordered. By contrast, for the mixed-chain, saturated-
polyunsaturated phosphatidylcholines, part b, the end segments
of the saturated chain become more disordered for longer acyl
chains. An order increment of the carbon segments next to the
headgroups is still present, albeit less pronounced than for the
disaturated series. For both series, the variation with the acyl
chain length follows a universal profile, as quantified by the
chain packing profiles plotted in Figure 9.
Conclusions
We have shown that polyunsaturation at the glycerol sn-2
position affects the bilayer structure by comparing the homolo-
gous series of mixed-chain, saturated-polyunsaturated phos-
phatidylcholines with the corresponding disaturated series. A
2
systematic analysis of the H NMR data has revealed that the
highly unsaturated DHA chains have significant disordering
effects on the chain packing within the lipid bilayer, and that
the magnitude of the effect increases progressively with increas-
ing saturated acyl chain length. The variations across the series
can be understood in terms of a universal, continuous chain
packing curve characteristic of saturated acyl chains, which is
sampled in discrete fashion by individual acyl chains. In all
instances, DHA induces a loosening of the acyl chain packing,
increasing the cross-sectional area per lipid headgroup as
compared to disaturated bilayers. This disordering effect of DHA
on the bilayer structure is expected to influence the interaction
of the bilayer environment with proteins and membrane-bound
peptides and, possibly, to modulate their biological functions.
Figure 10. Schematic representation of saturated chain packing
behavior of (a) disaturated di(m:0)PC series and (b) mixed-chain,
polyunsaturated (n:0)(22:6)PC series as a function of sn-1 acyl length
(m or n, respectively). The figure emphasizes the common midchain
region (neutral volume), with increased order toward the lipid headgroup
and increased disorder toward the chain ends near the bilayer center,
as quantified by the chain packing curves shown in Figure 9. The
universal chain packing curve, here depicted schematically, can be
regarded as a continuous profile sampled discretely by individual
saturated acyl chains. For the disaturated series (part a), more ordered
states are sampled by longer acyl chains, while keeping the order at
the bilayer center invariant. By contrast, the presence of DHA (part b)
makes more disordered states accessible for the end segments, while
reducing the accessibility to ordered states for the initial chain segments
next to the headgroup.
Abbreviations
(12:0-d23)(22:6)PC, 1-perdeuteriolauroyl-2-docosahexaenoyl-
sn-glycero-3-phosphocholine; (14:0-d27)(22:6)PC, 1-perdeute-
riomyristoyl-2-docosahexaenoyl-sn-glycero-3-phosphocho-
line; (16:0-d31)(22:6)PC, 1-perdeuteriopalmitoyl-2-docosahexa-
enoyl-sn-glycero-3-phosphocholine; (18:0-d35)(22:6)PC, 1-per-
deuteriostearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocho-
line; di(12:0-d23)PC, 1,2-diperdeuteriolauroyl-sn-glycero-3-phos-
phocholine; di(14:0-d27)PC, 1,2-diperdeuteriomyristoyl-sn-
glycero-3-phosphocholine; di(16:0-d31)PC, 1,2-diperdeuteriopalmi-
toyl-sn-glycero-3-phosphocholine; di(18:0-d35)PC, 1,2-diperdeu-
teriostearoyl-sn-glycero-3-phosphocholine; DHA, docosahexa-
enoic acid; DPA, docosapentaenoic acid; EFA, essential fatty
acid; NMR, nuclear magnetic resonance; PC, phosphatidylcho-
line; PUFA, polyunsaturated fatty acid.
position, z(i) , and the index number i′ are relative quantities
which depend linearly on the choice of a reference point. This
result shows that, for the whole series, there is a region having
a common degree of segmental order as for the disaturated
series, but this region is shifted up along the chain, toward the
headgroup, as the acyl length increases. In particular, at 50 °C,
we find a shift of roughly one carbon for each two additional
methylenes. The emerging chain packing curve can be viewed
as a continuous profile that is sampled discretely by individual
saturated acyl chains having different lengths.
These results suggest a picture of the saturated acyl chain
packing within planar bilayers, as schematically diagrammed
in Figure 10. As the length of the saturated sn-1 chain varies,
there is a central (midchain) section, as delimited by the dashed
horizontal lines in Figure 10, parts a and b. In this region, the
acyl chain ordering and packing properties are largely conserved.
This section of the bilayer can be regarded as a “neutral
volume”; i.e., it is left invariant upon addition of methylene
carbons to the saturated acyl chain. Outside of this region, the
segmental order parameters change as a function of the acyl
chain length, with a different variation for the disaturated versus
Acknowledgment. We thank Thomas Huber for many
valuable discussions and critical reading of the manuscript.
H.I.P. gratefully acknowledges Thomas B. Woolf for support
and discussions related to this work. This research was supported
by the U.S. National Institutes of Health and by the Ro¨ntgen-
Professorship of Physics at the University of Wu¨rzburg.
JA011745N