Assessing the cluster glycoside effect during the binding of concanavalin A to mannosylated artificial lipid rafts

Article abstract of DOI:10.1039/b910976e

Mannosyl glycolipids with perfluoroalkyl membrane anchors have been synthesised. When inserted into vesicles, these mannosyl lipids either dispersed evenly over the surface or, in the presence of cholesterol, phase-separated into artificial lipid rafts. At 1% mol/mol, the affinity of dispersed mannosyl lipids for Con A was 3-fold weaker than in solution, perhaps reflecting steric blocking by the surface. However increasing membrane loading 5-fold increased Con A affinity by up to 75% and indicated weak intramembrane chelation of Con A. Despite this observation, concentrating the mannosyl lipids into artificial lipid rafts did not significantly improve affinity for Con A. This lack of a cluster glycoside effect was ascribed to lipid congestion inhibiting intra-raft chelation of Con A, and implies that glycolipids located in lipid rafts may not necessarily be preorganised for multivalent binding.

Full text of DOI:10.1039/b910976e

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Assessing the cluster glycoside effect during the binding of concanavalin A
to mannosylated artificial lipid rafts†
Gavin T. Noble,a Sabine L. Flitsch,*a Kwan Ping Liema and Simon J. Webb*b
Received 4th June 2009, Accepted 23rd September 2009
First published as an Advance Article on the web 26th October 2009
DOI: 10.1039/b910976e
Mannosyl glycolipids with perfluoroalkyl membrane anchors have been synthesised. When inserted into
vesicles, these mannosyl lipids either dispersed evenly over the surface or, in the presence of cholesterol,
phase-separated into artificial lipid rafts. At 1% mol/mol, the affinity of dispersed mannosyl lipids for
Con A was 3-fold weaker than in solution, perhaps reflecting steric blocking by the surface. However
increasing membrane loading 5-fold increased Con A affinity by up to 75% and indicated weak
intramembrane chelation of Con A. Despite this observation, concentrating the mannosyl lipids into
artificial lipid rafts did not significantly improve affinity for Con A. This lack of a cluster glycoside
effect was ascribed to lipid congestion inhibiting intra-raft chelation of Con A, and implies that
glycolipids located in lipid rafts may not necessarily be preorganised for multivalent binding.
Biomimetic studies using artificial lipid rafts may reveal if, as
might be expected, concentrating mannosylated lipids in lipid rafts
Lipid rafts are phase-separated regions of liquid-ordered lipids
accentuates the cluster glycoside effect when binding to lectins.
1. Introduction
found within the fluid cell membrane.1 They contain high con-
However, rather than mannosylating the sphingolipids that are
often used to create rafts,1 mannose can instead be linked to a
fluorescent pyrene-perfluoroalkyl group. This membrane anchor
can phase-separate from fluid cholesterol-containing phospho-
lipid bilayers, particularly when the membrane is in the liquid
ordered (lo) phase.11,12 This phase separation mimics the dynamic
formation of lipid rafts, in this case providing mannoside clusters
on the membrane surface (Fig. 1(a)).
centrations of cholesterol, sphingolipids and membrane-bound
proteins, with the latter often linked to glycosylphosphatidylinos-
itol (GPI) membrane anchors.2 In addition to several lipid tails,
these GPI anchors are comprised of a heavily mannosylated glycan
linker that is connected to the protein through the nonreducing end
of the terminal mannose. The roles of GPI anchors are unclear;
the lipid tails are believed to be responsible for co-location within
lipid rafts, while the main role of the glycan section may be as a
spacer.3 However, IgG recognition of the glycan section of GPIs
from Plasmodium falciparum has been shown to be important for
the antimalarial immune response4 and the recognition of glycan
saccharides is also required by certain bacterial toxins.5 Potentially,
concentrating GPI anchors in lipid rafts may facilitate such
multivalent recognition events with bacterial toxins or antibodies
via the “cluster glycoside effect”, i.e. the enhancement in the per
binding site (valence corrected) affinity of the multivalent ligand.6
These lectin–sugar binding enhancements have been attributed
to either the formation of multiple links between binding part-
ners (chelation), statistical effects,7 “bind and slide”8 or ligand
oligomerisation.9 For membrane-bound glycolipids binding to
lectins, the enhancement depends upon ligand and lipid geometry,
since chelation requires the membrane-bound glycolipids to bridge
binding sites on the lectin. However recent studies have suggested
the “cluster glycoside effect” can also occur when glycolipids
cannot bridge binding sites.10
Fig. 1 (a) Binding of multivalent lectins to mannosylated lipids that are in
artificial “lipid rafts” or dispersed over the bilayer surface could change the
lateral distribution of these lipids. (b) Increasing the distance between the
surface and the mannosyl group may facilitate chelation of the multivalent
lectin.
The multivalent mannose-binding lectin concanavalin
A
(ConA) is extensively used tomodellectin/sacchariderecognition.
This tetrameric mannose-binding lectin, which agglutinates red
blood cells, is extracted from the common jack bean (Canavalia
ensiformis) where it is believed to be an insect antifeedant. It is also
a good quencher of fluorescence, which should allow binding to
fluorescent lipids to be monitored. Like many lectins, the affinity
of Con A for its target monosaccharides is relatively weak, with
an affinity for methyl mannoside of 8000 M-1 at 25 ◦C in water.6b,c
Nonetheless the cluster glycoside effect enhances Con A binding
to polymannosyl ligands in solution, even if the mannose units
aManchester Interdisciplinary Biocentre, The University of Manchester and
School of Chemistry, 131 Princess Street, Manchester, M1 7DN, UK.
E-mail: sabine.flitsch@manchester.ac.uk
bManchester Interdisciplinary Biocentre and the School of Chemistry,
University of Manchester, 131 Princess St, Manchester, M1 7DN, UK.
E-mail: S.Webb@manchester.ac.uk; Fax: +44-161-306-5201; Tel: +44-161-
306-4524
† Electronic supplementary information (ESI) available: Syntheses of lit-
erature compounds, incorporation studies (GPC), additional fluorescence
studies, and ITC data. See DOI: 10.1039/b910976e
˚
are too close together to bridge the 65 A distance between binding
sites. However, it is unclear whether the cluster glycoside effect
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 5245–5254 | 5245
©

View Article Online
also occurs when Con A binds to mannosyl lipid clusters in
phospholipid bilayers. To this end, Sasaki and co-workers studied
Con A recognition of clustered mannose-capped pyrenyl lipids
embedded in phospholipid vesicles.13 Binding to Con A drove the
mannosylated lipids apart, but this process was slow, taking up to
3 days at 25 ◦C to equilibriate. The affinity of these mannosylated
lipid clusters for Con A was reported as 3 ¥ 106 M-1, but this value
was calculated after a 30 hour incubation period, by which time
the protein had penetrated and disrupted the bilayers.
To determine if the cluster glycoside effect also operates during
Con A binding to artificial lipid rafts, mannosylated perfluoroalkyl
pyrene lipids 2 and 3 were synthesised (Scheme 1). Vesicle-based
studies similar to those of Sasaki and co-workers were used
as both intramembrane and intermembrane binding could be
assessed; the former should be enhanced by increasing the spacing
between the mannosyl moiety and the membrane surface (lipid 3
in Scheme 1, Fig. 1(b)). In view of previous observations,13 binding
analyses were carried out after a short incubation window to avoid
compromising vesicle integrity. These studies may also reveal if the
addition of multivalent Con A to mannosylated phase-separated
lipids leads to reorganisation of the lipids, such as partitioning out
of the artificial lipid rafts (Fig. 1(a)).
in the amount of excimer formed, which is reported by an
increase in fluorescence emission at 460 nm. If the addition of
a multivalent ligand changes the ratio of excimer to monomer
emission (375 nm) in the fluorescence emission spectrum (the
E/M ratio), this indicates the extent of phase separation of the
mannosylated lipids in the membrane has changed.
The synthesis of both glycolipids 2 and 3 was achieved
in
a convergent manner (Scheme 1). First, the perfluo-
roalkyl pyrene membrane anchor 1 was synthesised using pub-
lished methods.11b Schmidt glycosylation of N-CBz protected
ethanolamine or monoazido(triethylene glycol) with tetra-O-
acetylmannose trichloroacetimidate afforded the linked sugars
in good yield (62 and 72% respectively).14,15 Palladium-catalysed
hydrogenation was then employed to remove the CBz or reduce
the azide.16 Coupling of the aminosugars with the perfluoroalkyl
pyrene membrane anchor 1 was then performed using N,N¢-
dicyclohexylcarbodiimide (DCC) and N-hydroxysuccinimide
(NHS) in CH2Cl2. Finally, the acetyl protecting groups were
cleaved using standard Ze´mplen conditions to yield lipids 2 and 3
(in 66 and 44% yields respectively).17
2.2 Assessing the binding of Con A to mannose-capped lipids 2
and 3 in solution
Dynamic light scattering (DLS) experiments revealed that when
mannose-capped lipids 2 and 3 were dispersed and sonicated in
buffer at low concentration (2 mM), aggregates with a distinct
particle size were observed. DLS revealed the presence of lipid
aggregates with average diameters of (67 15) nm for 2 and (63
14) nm for 3.† These aggregates gave strong excimer emission
(E/M = 3.7 0.6 for 2, E/M = 65 7 for 3).18 Dilution of these
solutions until sharp decreases in E/M were observed gave critical
aggregation concentrations (CAC) values for 2 and 3 of 18.5 and
158 nM respectively.† The 10-fold diminution in CAC for 3 com-
pared to 2 could be due to the doubling in length of the hydrophilic
portion in 3, which should improve water solubility. These CAC
values are comparable to critical micelle concentrations of pyrene–
lecithin lipids (CMC = 150 nM) and suggested both compounds
should incorporate into phospholipid bilayers.
To measure binding of Con A to aqueous dispersions of lipids 2
and 3, a competition assay with the fluorescence probe mannose–
umbelliferone (MUM) was developed. Con A can strongly quench
fluorescence,19 which forms the basis of assays involving MUM.20
For MUM (but unlike pyrene) quenching is largely static, i.e.
through formation of a distinct Con A/mannose–umbelliferone
complex.21 Although the emission spectra of both the pyrenyl
lipids and MUM overlapped after excitation at 317 nm (the
excitation wavelength for MUM), it was possible to deconvolute
the MUM emission band at 375 nm.† In aqueous buffer the
mannose lipids had only weak monomer emission at 379 nm and
strong excimer emission at 460 nm; the weak pyrene monomer
emission could be subtracted from the strong emission at 375 nm
by calibration, providing the MUM emission.
First we parameterised our assay, which used Con A dissolved
in MOPS buffer in the absence of excess manganese(II) and
calcium(II).22 The Con A solution was titrated into MUM,
providing K = (3.37 0.09) ¥ 104 M-1, a similar value as reported
in the literature (4.3 ¥ 104 M-1 at pH 7.2).23 Then a competition
experiment was used to determine K (per binding site) for 2 and
Scheme 1 Synthesis of mannose-capped lipids 2 and 3.
2. Results and discussion
2.1 Synthesis of lipids 2 and 3
The key component of these phase-separating lipids is the perflu-
oroalkyl group, which inserts into the membrane of phospholipid
vesicles. The perfluoroalkyl group causes these lipids to become
excluded from regions that enter ordered phases (due to increased
intermolecular contacts), leading to the formation of phase-
separated patches. The extent of this phase separation is reported
by the pyrene fluorophore, as the increased local concentration of
pyrene within the phase-separated regions results in an increase
5246 | Org. Biomol. Chem., 2009, 7, 5245–5254
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
3. The diminution of the apparent affinity of MUM for Con A
caused by competition with lipid 2 revealed that the mannose
lipid bound to each Con A binding site with an affinity of (5.4
0.5) ¥ 103 M-1, a figure that is similar to that observed for methyl
mannoside (8000 M-1 at 25 ◦C in water).6b,c Similarly for lipid 3 the
valence-corrected value of K was (6.9 1.1) ¥ 103 M-1, the longer
TEG spacer perhaps facilitating binding. Given these titrations
were carried out above the CACs of 2 and 3, the observed binding
constants will be diminished relative to methyl mannoside, as 2
and 3 must be removed from higher assemblies prior to binding.24
The value of K for 3 binding to Con A in solution was supported
using ITC, which gave K ~ (9 3) ¥ 103 M-1 although accurate
determination of the affinity was hampered by low heat flows
during complexation. Indeed, ITC showed that unlike methyl
mannose,25 there was a significant entropic component in the
binding of lipid 3 to Con A.†
Fig. 2 (a). Change in turbidity of 50% mol/mol DMPC/chol vesicles
bearing 1% mol/mol lipid 2 (᭺) and lipid 3 (᭹) with increasing Con A
(each 2 mM total lipid) at 25 ◦C. (b), (c) Representative fluorescence mi-
crographs of vesicle aggregates 2 hours after the addition of FITC-labelled
Con A (4 eq. of binding sites) to 50% mol/mol DMPC/chol vesicles
bearing 1% mol/mol lipid 2 (b) and lipid 3 (c). Scale bar represents 10 mm.
emission with the vesicle aggregates showed that cross-linking was
Con A-dependent. The corresponding EC50 values for vesicle
aggregation by 2 and 3 are 80 mM and 40 mM respectively,
with the lower value for 3 consistent with the longer spacer
diminishing inter-vesicle repulsion and facilitating aggregation.
Both these values are lower than the dissociation constant of
Con A for methyl mannose (Kd of 125 mM) and could be mistaken
for evidence of multivalency. However, given sub-stoichiometric
amounts of lectin are needed to crosslink vesicles and turbidity
changes are not proportional to aggregate size,28 turbidity studies
can only be indicative. Although the strength of intervesicular
links also depends on bilayer hydration and vesicle surface charge,
the extensive aggregation of these vesicles reflects the effectiveness
of Con A in forming intermembrane links, perhaps at the expense
of intramembrane links.
It was hoped that binding of Con A would induce changes in
the lateral distribution of lipids 2 and 3, which would be reflected
in the E/M ratio. The addition of Con A (4 equivalents of binding
sites) gave the expected initial rapid drop of both monomer and
excimer emission from 2 and 3 due to fluorescence quenching. The
fluorescence emission remained constant for ten minutes before
slowly dropping for a period of two hours; nonetheless the E/M
ratio showed little change over this time. Turbidity measurements
and fluorescence microscopy revealed that these fluorescence
changes reflected changes in the sizes of the vesicle aggregates.
After 20 hours the aggregates were large and clearly visible by
the naked eye, flocculating in the sample solution despite rapid
stirring. This effectively changed the concentration of pyrene lipid
in the cuvette and gave the observed decrease in emission. Despite
these large changes in aggregate size, the lack of change in E/M
suggests that although intermembrane binding of Con A appears
to increase, this does not cause corresponding lipid reorganisation
(either flip-flop or lateral) of 2 or 3.29
2.3 Aggregation of vesicles bearing lipids 2 and 3
Vesicles were prepared by extrusion of suspensions of the ap-
propriate lipid mixture through polycarbonate membranes with
800 nm diameter pores. Incorporation of the mannose-capped
lipids before extrusion ensured that both leaflets were populated
evenly with the pyrene lipids. The unilamellar vesicle suspensions
produced (20 mM total lipid concentration) were then incubated
overnight at 25 ◦C to ensure the systems were at equilibrium before
further studies. To quantify the extent of pyrene lipid incorpo-
ration, the vesicle suspensions were analysed by gel permeation
chromatography (GPC). Vesicle suspensions containing 1, 2 or
5% mol/mol loading of either lipid 2 or 3 (20 mM of 2 or 3) in
dimyristoyl phosphatidylcholine (DMPC) were passed through a
PD10 Sephadex GPC column. Fractions (1 mL) were collected,
which showed that all the fluorescent lipid eluted with the vesicles
in the first 6 mL. No unincorporated lipid was collected and no
pyrene fluorescence remained on the PD10 column. Fluorescence
spectroscopy on the collected fractions confirmed incorporation
was ~99%. In contrast, GPC treatment of aqueous dispersions
of the lipids (also 20 mM) showed that unincorporated lipids
remained on the columns, which retained the bright fluorescence
of the pyrene group.†
The addition of ethanolic solutions of 2 and 3 (40 mL, 1 mM)
to DMPC vesicles (2 mL, 2 mM) revealed the dynamics of these
mannosylated lipids in bilayers. The addition of these lipids to
DMPC vesicle suspensions resulted in an immediate drop in
excimer emission (from E/M = 3.7 to 0.11 and 65 to 0.071 for
lipid 2 and 3 respectively), showing the lipids rapidly inserted
into the outer leaflet of the vesicles. This was followed by slower
rearrangement of the lipids in the membrane, with monomer
emission steadily increasing to give final E/M values of 0.065
for 2 and 0.057 for 3. This slower second step corresponds to
the flip-flop of the mannose-capped lipids to the interior leaflet
of the DMPC vesicles, and the half-lives for both lipids (~3.5
hours) correspond well to literature values for the flip-flop of
phospholipids in vesicles.26
2.4 Assessing intramembrane binding of Con A to dispersed
mannosylated lipids 2 and 3 in vesicles
The next step was to determine the affinity of Con A for 2 and
3 when they are not clustered and are evenly dispersed over the
vesicle surface. Vesicles composed of DMPC were used as these
are in the liquid disordered phase (ld, also known as La),30,31 which
does not force the clustering of lipids 2 and 3 into artificial rafts
(E/M ratios 0.06 and 0.09 respectively).
The binding sites of Con A are at the vertices of a tetrahedron
and are 65–70 A apart;9,27 given Con A crosslinks and agglomerates
˚
cells, vesicle aggregation would be expected. Indeed, FITC-
labelled Con A aggregated vesicles containing 1% mol/mol of
mannose lipids 2 and 3 (Fig. 2), and the localisation of FITC
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 5245–5254 | 5247
©

View Article Online
In vesicles, the MUM competition assay was ineffective, as
strong pyrene monomer emission significantly interfered with
MUM emission. Even so, the affinity is available directly from
pyrene fluorescence quenching by Con A, although complicated by
the potential for dynamic quenching. Significant dynamic quench-
ing of pyrene was anticipated given the long excited-state lifetime
of pyrene lipids compared to MUM (30–140 ns vs. 0.4 ns).20,32
Indeed, when Con A was titrated into suspensions of vesicles
containing only the perfluoroalkyl-pyrene membrane anchor 1 (no
mannose binding group, Scheme 1) at 1% mol/mol, significant
quenching of pyrene monomer fluorescence was observed, with
weaker quenching of excimer fluorescence. Quenching constants
were calculated from the quenching of monomer emission, as
excimer quenching depends upon both the concentration of
excited pyrene monomer33 and the relative ability of Con A to
dynamically quench excimer vs. monomer fluorescence.34 Fur-
thermore, fluorescence measurements were taken within 60 s of
mixing to minimise disruption of the bilayer by Con A.35 Repeated
titrations enabled the determination of the dynamic quenching
constant (Kdyn, representing a diffusive encounter between Con A
and 2 or 3) as (3.1 0.1) ¥ 103 M-1 in DMPC vesicles and (3.8
0.4) ¥ 103 M-1 in chol/DMPC vesicles (50% mol/mol), whilst a
non-binding galactose analogue of 2† gave Kdyn = (3.8 0.3) ¥
103 M-1 in DMPC vesicles and Kdyn = (3.1 0.4) ¥ 103 M-1 in
chol/DMPC vesicles.† The similarity between these Kdyn values
suggests that neither the structure of the perfluoroalkyl pyrene
lipid nor the composition of the membrane affects the magnitude
of dynamic quenching of pyrene by Con A.
of Con A for 2 and 3 in solution and weaker than Con A affinity for
methyl mannoside (8000 M-1). The low affinity of Con A for 2 and
3 at 1% mol/mol in vesicles is consistent with monovalent binding
of Con A to the mannose-capped lipid, with an affinity that is
diminished by steric repulsion from the vesicle surface.38 Other
workers have used the Langmuir isotherm to quantify Con A
binding to mannosylated surfaces; K varies from 6.5 ¥ 103 M-1 to
5.6 ¥ 106 M-1 at coverages between 10 and 100% of mannose.10,39,40
However high surface coverages can provide misleading results due
to the very large number of binding sites available. Keissling and
co-workers found a lower loading of 10% mol/mol mannosylated
lipid in phospholipid bilayers gave a reliable value of K = 2.7 ¥
104 M-1.41
Wilczewski et al. suggested that chelation of Con A by
surface-immobilized mannose only occurred at surface loadings
of mannose above 10%,10 so higher loadings of 2 and 3 were
needed to reveal intravesicular chelation of Con A. Using a
simple model42,43 for the binding of adhesive lipids to multivalent
ligands (eqn 2) shows that increasing membrane loading (c) whilst
maintaining the same total concentration of mannosylated lipid
(i.e. phospholipid concentration decreases instead) should only
increase the affinity of membrane-embedded lipids 2 and 3 for
Con A if intramembrane binding occurs.
4
(2)
K stat
=
K 1 K inter +k c K inter +k c K inter +k c
)( )(
(
)
2
2
3
3
4
4
Despite our assays being unable to provide sufficient indepen-
dent data to elucidate all the intermembrane (Ki2nter, K3inter and Ki4nter
)
Quenching of the mannose-capped lipids 2 and 3 by Con A
combines this dynamic quenching with static quenching resulting
from the direct complexation of the mannose headgroup in
the binding pocket. If Kdyn is assumed to be the same as that
for the perfluoroalkyl-pyrene membrane anchor, then the static
quenching constant (Kstat, which is equivalent to the binding
constant of 2 or 3 to each mannose binding site on Con A) can
be calculated from the initial slope of the Stern–Volmer curves
(Fig. 3), since as [Con A] → 0 the slope approximates Kdyn + Kstat
(eqn 1).36
and intramembrane (k2c, k3c and k4c) contributions, the overall
intramembrane contribution to Kstat should increase as membrane
loading (c) increases; a strong increase in the per binding site
affinity (Kstat) with increasing c would suggest one of k2, k3 or k4
is significant.
The membrane loading of mannose-capped lipids 2 and 3 in
the DMPC vesicles was increased from 1% to 2%, and then
5%. At each loading, complete incorporation of lipid in the
membrane of the vesicles was checked by GPC. Con A was
titrated into these suspensions and Kstat obtained from the slope
of the Stern–Volmer plots as previously. Interestingly, despite our
previously hypothesised monovalent intervesicular binding model,
the increase in Kstat with increasing membrane loading suggests
there is an intramembrane component to the binding of Con A to
2 and 3, i.e. these membrane-bound lipids can chelate Con A. In
addition, the assumption that the longer triethylene glycol linker
in 3 would facilitate multiple intramembrane links was supported
by the higher sensitivity of Kstat to c for 3 (Fig. 4).
F0
2
(1)
= 1+ K + K dyn Con A + K stat K dyn Con A
]
[
)
]
[
)
(
(
stat
F
Fig. 3 Stern–Volmer plots for the addition of Con A to vesicle suspen-
sions containing 1% mol/mol of pyrenyl lipids in DMPC: (a) ᭹ = lipid
anchor 1, ᭺ = mannose lipid 2; (b) ᭹ = lipid anchor 1, ᭺ = mannose
lipid 3.
In vesicles with 1% mol/mol of 2 or 3, the values of Kstat obtained
were between 2000 and 3000 M-1, although subtracting Kdyn from
the slope resulted in significant errors (up to 450 M-1).37 Each
individual binding event is around 3-fold weaker than the affinity
Fig. 4 Dependence of Kstat on membrane loading of mannose-capped
lipids 2 (ꢀ) and 3 (᭹).
5248 | Org. Biomol. Chem., 2009, 7, 5245–5254
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Table 1 Affinity of Con A binding sites for lipids 2 and 3 in vesicles at
Nonetheless, even a 5-fold increase in the membrane loading
of 3 only resulted in a 75% increase in affinity for Con A binding
sites. This compares poorly to other membrane-bound multivalent
binding events; an increase from 1% to 5% in the membrane
loading of an ethylenediamine-capped lipid resulted in an increase
in affinity for copper(II) of 150%,43 while a 5-fold increase in the
membrane loading of a cyclodextrin-capped lipid resulted in an
increase in affinity for a bis(adamantane) ligand of 300%.44
different membrane concentrations and compositions
K
stat/103 M-1
Vesicle composition Loading/% mol/mol
Lipid 2
Lipid 3
In solution
DMPC
DMPC
DMPC
DMPC/chol
—
1
2
5
5.4 0.5
2.3 0.3
2.6 0.2
3.0 0.3
2.4 0.1
6.9 1.1
2.0 0.4
2.8 0.2
3.5 0.3
2.7 0.4
1 (clustered)
2.5 Assessing the binding of Con A to clustered mannosylated
lipids 2 and 3 in vesicles
This observation of weak intramembrane binding even when lipids
2 and 3 are dispersed over the surface of the vesicles suggested that
the formation of mannosylated lipid rafts should further enhance
avidity towards Con A. Furthermore, given the observation of
the cluster glycoside effect in mannosylated dendrimers and lipids
where the individual mannose groups are too close together to
bridge between binding sites on the Con A, there was also the
potential that higher “rebinding” probabilities would contribute
to the cluster glycoside effect.41
As expected, at a membrane loading of 1% mol/mol in
vesicles containing 50% mol/mol cholesterol, both fluorinated
lipids 2 and 3 phase-separated into mannosylated lipid rafts. At
these concentrations of cholesterol the bilayers enter the liquid
ordered phase and phase-separate the fluorinated lipids, which
was confirmed by E/M ratios of (0.6 0.3) and (1.4 0.4) for
lipids 2 and 3 respectively. The phase separation of lipid 2 was
confirmed by studies of supported DMPC and DMPC/cholesterol
bilayers on glass. Small domains up to 0.3 mm in diameter
were observed by confocal microscopy for 1% mol/mol 2 in
DMPC/cholesterol bilayers, whilst no domains were observed for
2 in DMPC bilayers.†45 Domains ~0.1 mm2 in area should be able
to accommodate several bound Con A proteins, although smaller
domains would not be visible optically and domain size in 800 nm
diameter vesicles (~8 mm2 surface area) may be different.
Fig. 6 Binding of Con A to 2 or 3 is inhibited by steric repulsion from the
surface and does not cause lateral reorganisation of the lipids. (a) Dispersed
lipids allow (i) intermembrane binding and (ii) weak intramembrane
chelation and (iii) the approach of more Con A. (b) Clustering of mannosyl
lipids causes steric congestion, inhibiting both intramembrane chelation
and the approach of more Con A.
dynamic quenching of other phase-separated pyrenyl lipids, with
the increase in E/M attributed to the existence of two monomer
subpopulations in the bilayer, only one of which is able to form
excimer.32
These affinities for Con A binding to mannosylated lipids
clustered in artificial lipid rafts are only slightly higher than those
observed for the dispersed lipids 2 and 3 in DMPC vesicles; Kstat
=
(2.3 0.2) ¥ 103 M-1 with an E/M ratio of 0.06 for 2 and Kstat
=
(2.0 0.4) ¥ 103 M-1 with an E/M ratio of 0.09 for 3 (Table 1). This
small enhancement shows that the high effective concentration of
the lipid in these mannosyl lipid clusters, directly reflected in the
E/M value, does not give a significant cluster glycoside effect. It
suggests that concanavalin A binds mannosylated lipids in these
“artificial lipid rafts” in a monovalent manner, with the remaining
binding sites participating in intermembrane binding to generate
large vesicle agglomerates. The lack of any cluster glycoside effect
shows the mannose groups are less available, which could be due to
steric crowding of the mannose groups in the lipid rafts. A similar
explanation has been invoked to rationalise decreases in avidity
of other multivalent ligands for phase-separated adhesive lipids,
such as cholera toxin B to GM1 lipid, and avidin to biotinylated
lipids.12,46
Con
A was titrated into suspensions of 2 and 3 in
DMPC/cholesterol vesicles and the quenching of the pyrene
monomer emission from lipids 2 and 3 recorded after each aliquot.
Then given Kdyn = (3.1 0.1) ¥ 103 M-1, the Stern–Volmer plots
(Fig. 5) gave Kstat = (2.4 0.1) ¥ 103 M-1 for lipid 2 and Kstat = (2.7
0.4) ¥ 103 M-1 for lipid 3. After four binding site equivalents of
Con A had been added, the E/M ratios had increased by 20% and
5% for 2 and 3 respectively, but rather than lipid reorganisation this
reflected weaker quenching of excimer fluorescence compared to
monomer fluorescence. Similar observations have been made for
Conclusions
Two novel glycolipids (2 and 3), which possess mannose head-
groups and perfluoroalkyl pyrene membrane anchors, have been
synthesised and shown to insert into phospholipid vesicles at
membrane loadings between 1 and 5% mol/mol. The fluorinated
membrane anchors caused these lipids to form mannosylated
“artificial lipid rafts” in bilayers containing cholesterol, allowing
investigation of the cluster glycoside effect during lectin binding
Fig. 5 Stern–Volmer plots for the addition of Con A to vesicle suspen-
sions containing 1% mol/mol of (a) lipid 2 and (b) lipid 3 in DMPC (᭺)
and 50% mol/mol cholesterol in DMPC (ꢁ). Dynamic quenching of lipid
anchor 1 (᭹) is also shown.
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 5245–5254 | 5249
©

View Article Online
to phase-separated glycolipids. A longer triethylene glycol spacer
was incorporated into lipid 3 to increase the distance between
the mannosyl group and the vesicle surface, which was hoped to
enhance the cluster glycoside effect by facilitating intramembrane
chelation of mannose lipids.
The addition of concanavalin A efficiently aggregated vesicles
with 1% mol/mol of 2 or 3, mimicking the well-characterised
agglutination of cells by lectins. These aggregates increased in size
over time, reaching sizes greater than 100 mm after 2 hours, but
the increase in size was not accompanied by an increase in the
E/M ratio. This suggests that unlike the system developed by
Sasaki and co-workers,13 there is no reorganisation of lipids 2 or
3 upon binding of Con A, and is consistent with the monovalent
intervesicular binding mode suggested by Burke et al.,9 rather than
chelation of 2 or 3 by Con A.
geometry between binding sites caused clustered biotin lipids
(in lo membranes) to disperse and dispersed biotin lipids (in ld
membranes) to cluster. Nonetheless, steric crowding in “artificial
lipid rafts” of biotinylated lipids also caused avidin to bind
clustered lipids with half the strength compared to dispersed biotin
lipids.12 This steric crowding effect is mirrored in the work of
Cremer et al., who observed that GM1 binding to cholera toxin
was impeded by phase separation of the GM1 receptor in a planar
bilayer.46
Both cholera toxin and avidin are proteins that present adjacent
and directionally-aligned binding sites that are able to facilitate
multivalent binding. Therefore it is reasonable to assume that
intramembrane glycolipid binding and the subsequent glycolipid
reorganisation can only be truly observed with lectins that
present binding sites with the correct spatial configuration. Studies
continue with such lectins, like wheat germ agglutinin, and their
corresponding carbohydrate binding partners.
At a membrane loading of 1% mol/mol, fluorescence quenching
experiments showed Con A binds to 2 or 3 dispersed over the sur-
face of DMPC vesicles (Kstat ª 2200 M-1) with lower affinity than
to the same lipids dispersed in aqueous buffer (K ª 6000 M-1). This
weakened affinity may be due to steric repulsion from the bilayer
surface and reinforces the suggested monovalent intermembrane
binding model. However, increasing the membrane loading whilst
keeping the global concentration of 2 or 3 constant caused an
increase in the affinity of the membrane-bound mannosyl lipids
for the binding sites on Con A. Such an increase suggests there is
a contribution to affinity from intramembrane chelation of Con A
by the membrane-bound lipids (Fig. 6(a)), although changes in the
local environment at the surface could also play a role. The affinity
increased more strongly for 3 compared to 2 when the membrane
loading was increased, which suggests the longer triethylene glycol
spacer between the membrane anchor and the mannose group
facilitated intramembrane chelation of Con A. Nonetheless, even
for 3 the increase in affinity caused by a 5-fold increase in
membrane loading is lower than that for other vesicle-bound
adhesive lipids binding to their conjugate multivalent ligands, and
probably reflects the tetrahedral geometric relationship between
adjacent binding sites on Con A disfavouring intramembrane
chelation.
Changing the phase state of the membrane by adding cholesterol
at 50% mol/mol drove the phase separation of these lipids (at 1%
mol/mol) to form mannosylated artificial lipid rafts. Despite the
high concentration of mannose residues in these phase-separated
regions, no cluster glycoside effect was observed and the measured
avidities were only marginally greater (ª 2500 M-1 at 1% mol/mol)
than those measured at when the lipids dispersed evenly over the
surface of the vesicles (ª 2200 M-1 at 1% mol/mol). The lack of a
cluster glycoside effect may reflect a change in the availability of the
mannose group upon phase separation into these “artificial lipid
rafts”, such as deeper embedding of the lipids in the membrane
or steric congestion (Fig. 6(b)). This counterbalances the higher
binding expected through more efficient intramembrane chelation
and statistically more probable Con A rebinding to mannosylated
clusters.
Experimental section
General procedures
Unless otherwise specified, emission spectra were recorded from
360–600 nm, exciting at 346 nm using the slit settings 15 nm
excitation, 2.5 nm emission, on a Perkin Elmer LS55 Lumines-
cence Spectrometer with an attached Julabo F25 waterbath at
25 ◦C. UV-visible absorption measurements were taken using
a Jasco V-660 Spectrophotometer. Fluorescence micrographs
were produced using a Zeiss Axio Imager A1 fluorescence
microscope fitted with a Canon Powershot G6 digital cam-
era. Images of representative vesicles aggregates were taken at
intervals of 0, 10, 30, 60 and 120 minutes. Dynamic light
scattering experiments were performed on a Wyatt DynaPro
99. Unless otherwise stated, all reagents were purchased from
Sigma-Aldrich and used as received. Monotosylated triethylene
glycol,47 monoazido triethylene glycol,48 2,3,4,6-tetra-O-acetyl-
D-mannopyranosyl trichloroacetimidate,14 2,3,4,6-tetra-O-acetyl-
1-(2-aminoethoxy)-mannose,49 2,3,4,6-tetra-O-acetyl-1-(9-azido-
triethylene glycol)mannose,15 2,3,4,6-tetra-O-acetyl-1-(9-amino-
triethylene glycol)mannose16 were all synthesised by modifications
of literature procedures.†
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-10-(pyren-1-eth-
oxy)decyloxy-N-(2-a-D-(2,3,4,6-tetra-O-acetyl)mannopyranosy-
lethyl)acetamide.
(2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadeca-
fluoro-10-(pyren-1-ylmethoxy)decyloxy)acetic acid (50 mg,
0.068 mmol) was dissolved in dry CH2Cl2 (2 mL). The solution
was stirred under nitrogen and dicyclohexylcarbodiimide
(DCC) (18 mg, 0.087 mmol, 1.29 eq) was added followed by
N-hydroxysuccinimide (10 mg, 0.0879 mmol, 1.27 eq.). The
solution was left to stir for 3 h at room temperature, after which
time a milky white precipitate had formed. The reaction mixture
was filtered through cotton wool to remove the urea precipitate
and 2,3,4,6-tetra-O-acetyl-1-(2-aminoethoxy)mannose (37 mg,
0.095 mmol, 1.40 eq.) was added in dry CH2Cl2 (0.25 mL). The
mixture was stirred under dry nitrogen overnight. The crude
mixture was washed with NaHCO3 and water, dried over MgSO4
and the CH2Cl2 evaporated under reduced pressure. The crude
mixture was then purified using flash column chromatography
(3:2:1 ethyl acetate/chloroform/cyclohexane on silica) to yield
The biomimetic system presented herein suggests that the
binding protein geometry is the overriding factor for determining
whether multivalent binding at a membrane surface is possible.
In a concurrent study, reorganisation of a biotin lipid was
observed upon introduction of avidin; the stronger affinity of
avidin for biotin (>109 M-1 in this case) and a more favourable
5250 | Org. Biomol. Chem., 2009, 7, 5245–5254
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
a light-yellow solid (47 mg, 71%). TLC Rf 0.27 (3:2:1 ethyl
acid (1) (40 mg, 0.055 mmol) was dissolved in dry CH2Cl2
(2 mL). DCC (14.4 mg, 0.068 mmol, 1.29 eq) was added
followed by N-hydroxysuccinimide (8 mg, 0.070 mmol, 1.27 eq.)
and the solution was stirred under nitrogen for 3 h at room
temperature, after which a milky white precipitate had formed.
The reaction mixture was filtered through cotton wool to remove
the precipitate and 2,3,4,6-tetra-O-acetyl-1-(9-amino-triethylene
glycol)-a-D-mannose (52 mg, 0.109 mmol, 2 eq.) was added
in dry CH2Cl2 (0.25 mL). The reaction was left stirring under
N2 overnight. The crude mixture was washed with NaHCO3
and water, dried over MgSO4 and the CH2Cl2 evaporated under
reduced pressure. The crude mixture was then purified using flash
column chromatography (silica gel/19:1 chloroform/methanol) to
yield a milky-yellow oily solid (32.1 mg, 49%). TLC Rf 0.28 (19:1
CHCl3/MeOH), brown spot with ninhydrin followed by heat); 1H
1
acetate/chloroform/cyclohexane); H NMR (500 MHz, CDCl3,
25 ◦C): dH 1.98 (3 H, s, Ac CH3), 2.02 (3 H, s, Ac CH3), 2.09
(3 H, s, Ac CH3), 2.14 (3 H, s, Ac CH3), 3.55–3.59 (3 H, m,
CH2NHCO, OCHaHbCH2), 3.78 (1 H, ddd, 3J = 4.0, 4.7, 9.9 Hz,
OCHaHbCH2), 3.95 (1 H, ddd, 3J = 2.4, 5.7, 8.4 Hz, sugar
3
H-5), 4.02 (2 H, t, J (H,F) = 14.1 Hz, CH2CF2), 4.05 (2 H,
3
3
t, J (H,F) = 13.8 Hz, CH2CF2), 4.11 (1 H, dd, J = 2.2, 12.4,
3
sugar H-6a), 4.17 (2 H, s, OCH2CONH), 4.25 (1 H, dd, J =
5.9, 12.2 Hz, sugar H-6b), 4.83 (1 H, d, 3J = 1.3 Hz, sugar H-1),
3
3
5.24 (1 H, dd, J = 1.7, 3.2 Hz, sugar H-2), 5.27 (1 H, dd, J =
9.7, 10.1 Hz, sugar H-4), 5.31 (1 H, dd, 3J = 3.2, 10.1 Hz, sugar
H-3), 5.41 (2 H, s, CH2Ar), 6.81 (1 H, t, 3J = 5.6, NH), 7.99–8.36
◦
(9 H, m, 9 ¥ pyrene CH); 13C NMR (125 MHz, CDCl3, 25 C):
dC 19.5–19.8 (4 ¥ CH3), 37.4 (CH2), 61.5 (CH2), 65.1 (CH), 65.4
◦
3
3
(t, J = 25.9 Hz, CH2), 66.0 (CH), 67.1 (t, J = 25.4 Hz, CH2),
67.7 (CH), 67.8 (CH), 69.6 (CH2), 70.8 (CH2), 72.0 (CH2), 96.7
(CH), 120.0 (CH), 123.5 (CH), 123.6 (C), 123.9 (C), 124.5 (CH),
125.1 (CH), 126.3 (2 ¥ CH), 126.8 (CH), 127.1 (CH), 127.5 (CH),
NMR (400 MHz, CDCl3, 25 C): dH 1.98 (3 H, s, Ac CH3), 2.03
(3 H, s, Ac CH3), 2.09 (3 H, s, Ac CH3), 2.15 (3 H, s, Ac CH3),
3
3.52 (2H, t, J = 5.6 Hz, CH2NH), 3.57–3.64 (9 H, m, 4 ¥ TEG
CH2 + TEG C(1)Ha), 3.79 (1 H, dd, 3J = 4.2, 8.9 Hz, TEG C(1)Hb),
3.98-4.06 (5 H, m, sugar H-5, 2 ¥ CH2CF2), 4.10 (1 H, dd, 3J = 1.9,
11.5 Hz, sugar H-6a), 4.14 (2H, s, OCH2CONH), 4.28 (1H, dd,
3J = 4.8, 12.1 Hz, sugar H-6b), 4.86 (1 H, d, 3J = 1.6 Hz, sugar H-1),
5.26 (1 H, dd, 3J = 1.7, 3.3 Hz sugar H-2), 5.30 (1 H, dd, 3J = 9.9,
10.2 Hz, sugar H-4), 5.35 (1 H, dd, 3J = 3.3, 10.0 Hz, sugar H-3),
5.41 (2H, s, pyrene-CH2O), 6.78 (1H, t, 3J = 5.3 Hz, NH), 8.00–
8.37 (9 H, m, 9 ¥ pyrene CH); 13C NMR (125 MHz, CDCl3, 25 ◦C):
dC 20.6–20.9 (4 ¥ CH3), 38.7 (CH2), 62.4 (CH2). 66.2 (CH), 66.5 (t,
3J = 26.1 Hz, CH2), 67.4 (CH2), 68.1 (t, 3J = 25.4 Hz, CH2), 68.5
(CH), 69.1 (CH), 69.6 (CH), 69.7 (CH2), 70.0 (CH2), 70.4 (CH2),
70.6 (CH2), 71.8 (CH2), 73.1 (CH2), 97.7 (CH), 123.0 (CH), 124.5
(CH), 124.6 (C), 125.0 (C), 125.5 (CH). 126.1 (CH), 127.3 (CH),
127.4 (CH), 127.9 (CH), 128.2 (CH), 129.1 (CH), 129.6 (C), 130.8
=
128.1 (C), 128.5 (C), 129.7 (C), 130.2 (C), 130.8 (C), 167.1 (C O),
+
+
=
168.7–169.6 (4 ¥ C O); MS (ES ) m/z 1130.1 [M + Na] (9%),
1108.4 [M + H]+ (4%), 449.5 (30%), 420.2 (76%), 392.2 (41%),
215.2 [pyreneCH2]+ (100%); HRMS (ES+) m/z calculated for
C45H41F16NO13Na+ 1130.2215, found 1130.2247.
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-10-(pyren-1-etho-
xy)decyloxy-N-(2-a-D-mannopyranosylethyl)acetamide (2). 2,2,
3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-10-(pyren-1-ethoxy)-
decyloxy-N-(2-a-D-(2,3,4,6-tetra-O-acetyl)mannopyranosyleth-
yl)acetamide (47 mg, 0.043 mmol) was dissolved in dry methanol
(2 mL). A pH 10, sodium methoxide (NaOMe) solution was
made by dissolving sodium metal in dry methanol until the
required pH was reached. The NaOMe solution (pH 10, 2 mL)
was then added slowly to the stirring lipid solution, and the
mixture stirred for 30 min. Ion-exchange resin (Amberlite IR-120)
was then added until the solution was pH 7. The resin was
then filtered off through cotton wool, the ethanol removed under
reduced pressure and a milky-yellow solid was obtained (37 mg,
=
(C), 131.2 (C), 131.8 (C), 132.1 (C), 167.9 (C O), 169.7–170.6
+
+
=
(4 ¥ C O); MS (ES ) m/z 1218.0 [M + Na] (10%), 1196.5 [M +
H]+ (16%), 1012.5 (8%), 624.0 (34%), 508.1 (23%), 480.4 (12%),
331.1 (75%), 215.1 [pyrene-CH2]+ (12%); HRMS (MALDI) m/z
calculated for C49H49F16NO15Na+ 1218.2739, found 1218.2720.
1
0.0394 mmol, 93%). TLC Rf 0.2 (9:1 chloroform/methanol); H
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-10-(pyren-1-met-
hoxy)decyloxy-N-(2-(2-(2-a-D-mannopyranosylethoxy)ethoxy)et-
hyl))amide (3). 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-
10-(pyren-1-methoxy)decyloxy-N-(2-(2-(2-a-D-(2,3,4,6-tetra-O-
acetyl)mannopyranosylethoxy)ethoxy)ethyl))amide (32.1 mg,
0.0269 mmol) was dissolved in dry methanol (1 mL). 25 wt%
NaOMe/MeOH solution (0.1 mL) was then added slowly to
the stirring lipid solution. The mixture was stirred for 30 min.
Ion-exchange resin (Amberlite IR-120) was then added until the
solution reached pH 7. The resin was then filtered off through
cotton wool, the methanol removed under reduced pressure and
a milky-yellow oily solid was obtained (24.7 mg, 0.0241 mmol,
90%). TLC Rf 0.05 (9:1 chloroform/methanol/1 drop Et3N,
brown spot with ninhydrin followed by heat); 1H NMR (400 MHz,
acetone-d6, 25 ◦C): dH 3.46 (2H, m, TEG CH2), 3.53 (1H, t,
3J = 6.2 Hz, TEG CHaHb), 3.59–3.72 (13H, m, 4 ¥ TEG
CH2, TEG CHaHb, sugar OH-2,3,4,6), 3.80–4.10 (6H, m, sugar
NMR (400 MHz, acetone-d6, 25 ◦C): dH 3.49–3.86 (14 H, m,
CH2CH2NH, CH2CH2NH, sugar CH-2,3,4,5,6a,6b, sugar OH-
2,3,4,6), 4.22 (2H, s, CH2CONH), 4.35 (2 H, t, 3J (H,F) = 14.5 Hz,
3
CH2CF2), 4.37 (2 H, t, J (H,F) = 15.0 Hz, CH2CF2), 4.81 (1H,
d, 3J = 1.3 Hz, sugar H-1), 5.55 (2 H, s, CH2Ar), 7.40 (1H, t, 3J =
5.7 Hz, NH), 8◦.11-8.51 (9 H, m, pyrene CH); 13C NMR (100 MHz,
acetone-d6, 25 C): dC 40.0 (CH2), 63.1 (CH2), 67.4 (t, 3J = 24.8 Hz,
3
CH2), 69.1 (t, J = 24.7 Hz, CH2), 69.8 (CH), 70.8 (CH2), 72.4
(CH), 73.0 (CH2), 73.3 (CH), 74.1 (CH2), 74.9 (CH), 101.8 (CH),
125.0 (CH), 126.3 (CH), 126.4 (C), 126.5 (C), 127.1 (2 ¥ CH), 127.9
(CH), 129.0 (CH). 129.1 (CH), 129.4 (CH), 129.5 (CH), 131.8
=
(C), 131.9 (C), 132.3 (C), 132.5 (C), 132.9 (C), 168.7 (C O); MS
(ES+) m/z 962.7 [M + Na]+ (44%), 940.0 [M + H]+ (3%), 747.7 [2-
pyreneCH2+Na]+ (16%), 453.9 (50%), 412.9 (100%), 215.0 [pyrene-
CH2]+ (72%); HRMS (ES+) m/z calculated for C37H33F16NO9Na+
expected 962.1798, found 962.1788, 963.1846, 964.1879.
3
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9-Hexadecafluoro-10-(pyren-1-met-
hoxy)decyloxy-N-(2-(2-(2-a-D-(2,3,4,6-tetra-O-acetyl)mannopy-
ranosylethoxy)ethoxy)ethyl))-amide. (2,2,3,3,4,4,5,5,6,6,7,7,8,
8,9,9-Hexadecafluoro-10-(pyren-1-ylmethoxy)decyloxy)acetic
CH-2,3,4,5,6a,6b), 4.22 (2 H, s, OCH2CONH), 4.37 (2 H, t, J
(H,F) = 14.3 Hz, CH2CF2), 4.40 (2H, t, 3J = 14.2 Hz, CH2CF2),
4.84 (1 H, d, 3J = 1.4 Hz, sugar H-1), 5.55 (2 H, s, CH2Ar), 7.31
3
(1H, t, J = 5.4 Hz, NH), 8.11–8.51 (9 H, m, pyrene CH); 13C
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 5245–5254 | 5251
©

View Article Online
◦
NMR (100 MHz, acetone-d6, 25 C): dC 34.6 (CH2), 64.3 (CH2),
fluorescence emission at 379 and 460 nm was plotted against
elution volume. Very little free 2 or 3 eluted from the GPC columns
within the first 10 mL, whereas almost all (~99%) vesicle-bound 2
or 3 in DMPC was recovered irrespective of membrane loading of
mannosyl lipid.
3
3
67.5 (CH2), 67.8 (t, J = 25.1, CH2), 68.6 (t, J = 25.0, CH2),
69.0 (CH), 69.3 (CH), 69.5 (CH2), 70.0 (CH2), 70.1 (CH2), 70.7
(CH2), 71.0 (CH2), 71.6 (CH), 71.8 (CH), 74.2 (CH2), 99.3 (CH),
124.5 (CH), 125.5 (C), 125.6 (CH), 125.9 (C), 126.8 (2 ¥ CH),
127.2 (CH), 128.3 (2 ¥ CH), 128.7 (CH), 128.9 (CH), 130.3 (C),
130.4 (C), 131.2 (C), 131.8 (C), 132.2 (C), 162.9 (C O); MS (ES )
+
=
Determination of K for 2 and 3 binding to Con A in solution
(competition experiment with MUM)
m/z 1050.6 [M + Na]+ (8%), 318.2 (45%), 268.2 (11%), 267.2
(64%), 215.0 [pyrene-CH2]+ (62%); (MALDI) m/z calculated for
C41H41F16NO11Na+ expected 1050.2316, found 1049.9567 (100%)
[M + Na]+; HRMS (ES+) calculated for C41H41F16NO11Na+
1050.2322, found 1050.2297.
15 ¥ 10 mL titres of Con A/MUM solution with 2 or 3 (1.57 mM
Con A, 20 mM 2 or 3, 2 mM MUM) were added to a solution
of MUM with 2 or 3 (2 mM MUM, 20 mM 2 or 3, 2 mL,
20 mM MOPS, 100 mM NaCl, pH 7.4). Fluorescence spectra were
recorded (ex. at 317 nm) 60 s after each addition. The emission of
MUM (375 nm) and pyrene monomer (379 and 395 nm) overlaps
when excited at 317 nm, while that of the pyrene excimer was
clear. Therefore analogous titrations in the absence of MUM
were carried to give E/M ratios that allowed the determination
of the contribution of pyrene monomer and thus MUM to the
fluorescence at 375 nm. In a modification of the procedure of
Landschoot et al.,50 a Scatchard plot was employed to determine
K from the resulting data. Eqns 3 and 4 were used to resolve
values for the response (amount of bound ligand, r) and amount
of unbound pyrene lipid [P]f ([2]f or [3]f) at each titre point.
Preparation of vesicles
800 nm unilamellar vesicles were prepared by dissolving the
required molar ratio of DMPC, cholesterol and synthetic lipid
in ethanol-free spectroscopic grade chloroform (total lipid =
20 mmol) followed by removal of the solvent to give a thin
film of lipid. The appropriate buffer (20 mM MOPS, 100 mM
NaCl, pH 7.4 at room temperature, 1 mL) was added to the
flask, and the thin film detached by vortex mixing. Extrusion
(19¥) through a single 800 nm polycarbonate membrane using
an Avestin Liposofast extrusion apparatus above the lipid Tm gave
unilamellar vesicles. The final concentration of phospholipid was
20 mM, that of the synthetic lipid 1 mM. For vesicles containing
2 or 5% mol/mol synthetic lipid, the synthetic lipid concentration
was constant but the amount of phospholipid was reduced to
10 mmol and 4 mmol respectively. All vesicle suspensions were
freshly extruded prior to fluorescence or UV studies.
⎛
⎞
⎟
⎟
⎟
⎟
⎠
B
1
⎜
r = 1−
U +
]
⎜
(3)
(4)
[
⎜
⎜
C
[ ]
K cu −BK cu
⎝
⎛
⎜
⎝
⎞
⎟
⎟
⎟
⎟
⎠
1
P
= P − C −B U −
⎜
[ ] [ ] [ ]
[
]
f
⎜
⎜
K cu −BK cu
where [C], [U] and [P] are the total concentrations of Con A,
MUM and pyrene lipid respectively. Kcu is the association constant
of MUM with Con A and B = 1 - Fu(0)/Fu (Fu(0) is the initial
MUM fluorescence at 375 nm). Plotting r/[P]f vs. r gave a linear
plot with a gradient of -K, which gave K = 5.4 0.5 ¥ 103 M-1 for
2 and K = 6.9 1.1 ¥ 103 M-1 for 3.
Dynamic light scattering (DLS)
Dispersions of lipid 2 or 3 (2 mM, in 20 mM MOPS, 100 mM NaCl,
pH 7.4) were sonicated for 5 min and 0.2 mL was transferred to a
DLS cuvette. 10 ¥ 10 s acquisitions at 25 C using a laser power
of 30% were performed for each lipid and were repeated 3 times.
Results were analysed using Wyatt Dynamics software (version
6.7.7.9).
◦
Determination of K for 4-methylumbelliferyl-a-D-
mannopyranoside (MUM) binding to Con A
Critical aggregation concentration (CAC) determination
Concanavalin A/MUM solution (1.57 mM Con A, 2 mM MUM,
15 ¥ 10 mL titres) was added to a buffered solution of MUM
(2 mL, 2 mM MUM, 20 mM MOPS, 100 mM NaCl, pH 7.4).
Fluorescence spectra were recorded (ex. at 317 nm) 60 s after each
addition. The emission at 375 nm was fitted to a Stern–Volmer
plot, which gave K = 3.37 0.09 ¥ 104 M-1 (Kstat = K).
Buffer (1 mL 20 mM MOPS, 100 mM NaCl, pH 7.4) was added
to a thin film of 2 or 3, the lipid solubilised using vortex mixing
and sonication for 5 min to give a lipid dispersion (200 mM 2 or 3).
The fluorescence spectrum was recorded, then the solution diluted
10-fold using buffer and fluorescence spectrum re-recorded. This
procedure was repeated until the fluorescence was too weak to
measure (2 nM). The excimer (460 nm) to monomer (395 nm)
fluorescence ratio was plotted vs. log[lipid]; a sharp increase in
excimer emission occurred at the CAC, 0.0185 mM for 2 and
0.157 mM for 3.
Determination of quenching of vesicle-bound pyrenyl lipids 1, 2 or
3 by Con A
(A) Titrations. Freshly extruded vesicle suspensions (200 mL)
containing the desired synthetic lipid (1, 2 or 3) were diluted to
2 mL with buffer (2 mM total lipid, 20 mM MOPS, 100 mM
NaCl, pH 7.4) and the fluorescence spectrum measured. Using a
constant analyte concentration method, 20 ¥ 10 mL titres of Con A
solution (62.4 mg mL-1, 2.4 mM (binding sites), 2 mM vesicles)
were added and the fluorescence spectrum recorded 60 s after each
addition. These fluorescence data were corrected for dilution and
fitted to a Stern–Volmer plot. Control experiments using synthetic
Determination of 2 and 3 insertion into vesicles
Dispersions of mannosyl lipids (2.5 mL, 20 mM 2 or 3) in buffer
(20 mM MOPS, 100 mM NaCl, pH 7.4) or suspensions of DMPC
vesicles with 1, 2 or 5% mol/mol 2 or 3 were loaded onto GPC
columns. The samples were eluted with buffer and fluorescence
spectra recorded for each 1 mL of eluent. For each sample the
5252 | Org. Biomol. Chem., 2009, 7, 5245–5254
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
lipid 1 possessing no binding groups gave the dynamic quenching
constants. The association constants for 2 or 3 to Con A were
determined by subtracting the observed Kdyn values for titrations
with 1 from the slopes of the lines in the Stern–Volmer plots for 2
or 3.
13 B. Bondurant, J. A. Last, T. A. Waggoner, A. Slade and D. Y. Sasaki,
Langmuir, 2003, 19, 1829.
14 (a) R. R. Schmidt and J. Michel, Angew. Chem., Int. Ed. Engl., 1980, 19,
731; (b) R. Karamanska, J. Clarke, O. Blixt, J. I. MacRae, J. Q. Zhang,
P. R. Crocker, N. Laurent, A. Wright, S. L. Flitsch, D. A. Russell and
R. A. Field, Glycoconjugate J., 2008, 25, 69.
15 N. Nagahori and S.-I. Nishimura, Biomacromolecules, 2001, 2, 22.
16 (a) Y. Koshi, E. Nakata, M. Miyagawa, S. Tsukiji, T. Ogawa and
I. Hamachi, J. Am. Chem. Soc., 2008, 130, 245; (b) O. Otman, P.
Boullanger, D. Lafont and T. Hamaide, Macromol. Chem. Phys., 2008,
209, 2410.
17 G. Zemple´n and E. Pacsu, Chem. Ber., 1929, 62, 1613.
18 Although there are significant differences in E/M between 2 and 3, it
is difficult to compare E/M values for different pyrenyl compounds if
E/M is already large, i.e. monomer emission is weak. E/M is sensitive to
several factors including excited-state lifetime and the rate of collision
between pyrene moieties. See ref. 33.
(B) Timecourse. Freshly extruded vesicle suspensions (200 mL)
containing the desired synthetic lipid (1, 2 or 3) were diluted to
2 mL with buffer (2 mM total lipid, 20 mM MOPS, 100 mM NaCl,
pH 7.4) and the fluorescence spectrum measured. Con A (2.08 mg,
4 eq. binding sites) was added and fluorescence spectra recorded
every 60 s for 10 min, then every 600 s for 2 h and finally once
every 2 h until 24 h had passed.
19 (a) R. L. Phillips, I.-B. Kim, L. M. Tolbert and U. H. F. Bunz, J. Am.
Chem. Soc., 2008, 130, 6952; (b) C. Xue, S. P. Jog, P. Murthy and H.
Liu, Biomacromolecules, 2006, 7, 2470.
20 R. B. Thompson and J. R. Lakowicz, Biochim. Biophys. Acta, 1984,
790, 87.
Determination of vesicle aggregation by Con A
(A) Titrations. Freshly extruded vesicle suspensions (200 mL)
containing the desired synthetic lipid (2 or 3) were diluted to 2 mL
with buffer (2 mM total lipid, 20 mM MOPS, 100 mM NaCl,
pH 7.4) and the absorption spectrum measured. Using a constant
analyte concentration method, 20 ¥ 10 mL titres of a concanavalin
A solution (40.8 mg mL-1, 1.57 mM Con A, 2 mM vesicles) were
added and the absorption spectrum recorded 1 min after each
addition. The absorption at 700 nm was used as the UV indicator
of turbidity.
21 (a) F. G. Loontiens, R. M. Clegg and T. M. Jovin, Biochemistry, 1977,
16, 159; (b) R. M. Clegg, F. G. Loontiens and T. M. Jovin, Biochemistry,
1977, 16, 167; (c) D. J. Christie, G. M. Alter and J. A. Magnuson,
Biochemistry, 1978, 17, 4425; (d) P. C. Harrington and R. G. Wilkins,
Biochemistry, 1978, 17, 4245.
22 The addition of 1 mM Mn(II) and 1 mM Ca(II) made no difference
to the values of K obtained either for MUM or lipids 2 and 3. The
addition of an excess of these ions is not required for Con A-mannose
binding as the Mn(II) in S1 is tightly bound (Kd = 1.6 mM at 25 ◦C and
pH 6.5; A. Sadhu and J. A. Magnuson, Biochemistry, 1989, 28, 3197)
and the Ca(II) in S2 is not required; (C. F. Brewer, D. M. Marcus, A. P.
Grollman and H. Sternlicht, J. Biol. Chem, 1974, 249, 4614). Excluding
1 mM Mn(II) improved the signal-to-noise ratio by 31%.
23 A. Van Landschoot, F. G. Loontiens and C. K. De Bruyne,
Eur. J. Biochem., 1978, 83, 277.
24 (a) R. Holm, W. Shi, R. A. Hartvig, S. Askjær, J. C. Madsend and P.
Westh, Phys. Chem. Chem. Phys., 2009, 11, 5070; (b) E. Junquera, L.
Pen˜a and E. Aicart, Langmuir, 1997, 13, 219.
25 T. K. Dam, R. Roy, S. K. Das, S. Oscarson and C. F. Brewer, J. Biol.
Chem., 2000, 275, 14223.
26 E. Middlekoop, B. H. Lubin, J. A. F. Op den Kamp and B. Roelfsen,
Biochim. Biophys. Acta, 1986, 855, 421.
(B) Timecourse. Freshly extruded vesicle suspensions (200 mL)
containing the desired synthetic lipid (1, 2 or 3) were diluted to
2 mL with buffer (2 mM total lipid, 20 mM MOPS, 100 mM NaCl,
pH 7.4) and the absorption spectrum measured. Con A (2.08 mg,
4 eq. binding sites) was added and absorption spectra recorded
every 60 s for 10 min, then every 600 s for 2 h and finally once
every 2 h until 24 h had passed.
Acknowledgements
27 H. U. Ahmed, M. P. Blakeley, M. Cianci, D. W. J. Cruickshank,
J. A. Hubbard and J. R. Helliwell, Acta Crystallogr., Sect. D: Biol.
Crystallogr., 2007, 63, 906.
We thank the BBSRC for providing studentship funding for KPL
and GTN, and the Leverhulme Trust for additional funding.
28 E. V. Pozharski, L. McWilliams and R. C. MacDonald, Anal. Biochem.,
2001, 291, 158.
References and notes
29 Instantaneous changes in distribution (<60 s) cannot be excluded due
to the obscuring effect of pyrene quenching by Con A, but such rapid
changes would be unlikely given the viscous nature of the membrane.
30 The terms La and ld both represent lamellar liquid crystalline phases of
phospholipid bilayers, with the term ld often used when a co-existing
liquid ordered lo phase is present in the membrane. See ref. 2.
31 The addition of lipids bearing the pyrenyl-perfluoroalkyl membrane
anchor in 2 and 3 has been found to depress the Tm of phospholipid
bilayers by 1–2 ◦C. See R. J. Mart, K. P. Liem and S. J. Webb, Chem.
Commun., 2009, 2287.
32 Y. Barenholtz, T. Cohen, E. Haas and M. Ottolenghi, J. Biol. Chem.,
1996, 271, 3085.
33 H. J. Galla and E. Sackmann, J. Am. Chem. Soc., 1975, 97, 4114.
34 This effect means changes in E/M ratio will not necessarily reflect
changes in lateral distribution if the multivalent ligand causes signifi-
cant dynamic quenching. See ref. 32.
1 W. H. Binder, V. Barragan and F. M. Menger, Angew. Chem., Int. Ed.,
2003, 42, 5802.
2 M. Luckey, in Membrane Structural Biology, Cambridge University
Press, New York, 1st edn, 2008, ch. 4, pp. 73–76.
3 M. G. Paulick and C. R. Bertozzi, Biochemistry, 2008, 47, 6991.
4 F. Kamena, M. Tamborrini, X. Liu, Y.-U. Kwon, F. Thompson, G.
Pluske and P. H. Seeberger, Nat. Chem. Biol., 2008, 4, 238.
5 D. B. Diep, K. L. Nelson, S. M. Raja, E. N. Pleshak and J. T. Buckley,
J. Biol. Chem., 1998, 273, 2355.
6 (a) J. J. Lundquist and E. J. Toone, Chem. Rev., 2002, 102, 555;
(b) D. K. Mandal, N. Kishore and C. F. Brewer, Biochemistry, 1994,
33, 1149; (c) J. B. Corbell, J. J. Lundquist and E. J. Toone, Tetrahedron:
Asymmetry, 2000, 11, 95.
7 M. Kanai, K. H. Mortell and L. L. Kiessling, J. Am. Chem. Soc., 1997,
119, 9931.
35 Replicating the procedure of Bondurant et al. (ref. 13) resulted in a
broad and featureless pyrene emission spectrum. We believe the large
excess of Con A compared to phospholipid and mannose lipid (a 1:1
ratio of Con A to total lipid) may cause rapid bilayer disruption, similar
to that observed by Bondurant et al. over several hours.
36 B. Valeur, in Molecular Fluorescence: Principles and Applications,
Wiley-VCH, Weinheim, 1st edn, 2002, ch. 4, p. 86.
37 The quenching constants obtained from these Stern–Volmer plots were
verified by iterative fitting of the data to a 1+1 binding isotherm in
Dynafit. See P. Kuzmic, Anal. Biochem., 1996, 237, 260.
8 T. K. Dam and C. F. Brewer, Biochemistry, 2008, 47, 8470.
9 S. D. Burke, Q. Zhao, M. C. Schuster and L. L. Kiessling, J. Am. Chem.
Soc., 2000, 122, 4518.
10 M. Wilczewski, A. Van der Heyden, O. Renaudet, P. Dumy, L. Coche-
Gue´rente and P. Labbe´, Org. Biomol. Chem., 2008, 6, 1114.
11 (a) R. J. Mart, X. Wang, K. P. Liem and S. J. Webb, J. Am. Chem.
Soc., 2006, 128, 14462; (b) S. J. Webb, K. Greenaway, M. Bayati and L.
Trembleau, Org. Biomol. Chem., 2006, 4, 2399.
12 K. P. Liem, G. T. Noble, S. L. Flitsch and S. J. Webb, Faraday Discuss.,
2010, DOI: 10.1039/b907114h.
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 5245–5254 | 5253
©

View Article Online
38 (a) H. Jung, T. Yang, M. D. Lasagna, J. Shi, G. D. Reinhart and P. S.
Cremer, Biophys. J., 2008, 94, 3094; (b) S. Zhao, D. S. Walker and W. M.
Reichert, Langmuir, 1993, 9, 3166.
39 E. A. Smith, W. D. Thomas, L. L. Kiessling and R. M. Corn, J. Am.
Chem. Soc., 2003, 125, 6140.
45 These domains have a comparable size to those observed for other
phase-separating pyrenyl lipids. See C. Hayden, J. S. Hwang, E. A.
Abate, M. S. Kent and D. Y. Sasaki, J. Am. Chem. Soc., 2009, 131,
8728.
46 J. Shi, T. Yang, S. Kataoka, Y. Zhang, A. J. Diaz and P. S. Cremer,
J. Am. Chem. Soc., 2007, 129, 5954.
47 A. Bouzide and G. Sauve, Org. Lett., 2002, 4, 2329.
48 S. Ciampi, T. Bo¨cking, K. A. Kilian, J. B. Harper and J. J. Gooding,
Langmuir, 2008, 24, 5888.
40 These are usually the macroscopic binding constants for Con A; the
valence corrected values are 4-fold weaker.
41 D. A. Mann, M. Kanai, D. J. Maly and L. L. Kiessling, J. Am. Chem.
Soc., 1998, 120, 10575.
42 X. Wang, R. J. Mart and S. J. Webb, Org. Biomol. Chem., 2007, 5, 2498.
43 E. L. Doyle, C. A. Hunter, H. C. Phillips, S. J. Webb and N. H. Williams,
J. Am. Chem. Soc., 2003, 125, 4593.
44 C. W. Lim, B. J. Ravoo and D. N. Reinhoudt, Chem. Commun., 2005,
5627.
49 T. Hasegawa, T. Fujisawa, M. Numata, T. Matsumoto, M. Umeda, R.
Karinaga, M. Mizu, K. Koumoto, T. Kimura, S. Okumura, K. Sakurai
and S. Shinkai, Org. Biomol. Chem., 2004, 2, 3091.
50 A. Van Landschoot, F. G. Loontiens and C. K. De Bruyne,
Eur. J. Biochem., 1980, 103, 307.
5254 | Org. Biomol. Chem., 2009, 7, 5245–5254
This journal is
The Royal Society of Chemistry 2009
©