Model system for cell adhesion mediated by weak carbohydrate-carbohydrate interactions

Article abstract of DOI:10.1021/ja210304j

The multivalent carbohydrate-carbohydrate interaction between membrane-anchored epitopes derived from the marine sponge Microciona prolifera has been explored by colloidal probe microscopy. An in situ coupling of sulfated and non-sulfated disaccharides to

Full text of DOI:10.1021/ja210304j

Communication
pubs.acs.org/JACS
Model System for Cell Adhesion Mediated by Weak Carbohydrate−
Carbohydrate Interactions
†
¶
†
‡
‡
́
Bar
Milena Stephan, Eva Sunnick, Deniz Yu
̈
bel Lorenz, Luis Alvarez de Cienfuegos, Marieelen Oelkers, Ella Kriemen, Christian Brand,
†
†
§
§
§
,†̈
ksel, Venkateshwarlu Kalsani, Krishna Kumar,
‡
Daniel B. Werz, and Andreas Janshoff*
†
‡
Institute of Physical Chemistry and Institute of Organic and Biomolecular Chemistry, University of Go
7077 Gottingen, Germany
Department of Chemistry, Tufts University, 62 Talbot Avenue, Medford, Massachusetts 02155, United States
¶Department of Organic Chemistry, Universidad de Granada, Granada, Spain
̈
ttingen, Tammannstrasse 6,
3
§
̈
*
S Supporting Information
challenge. Recently, Anselmetti and co-workers were able to
2
+
ABSTRACT: The multivalent carbohydrate−carbohydrate
interaction between membrane-anchored epitopes derived
from the marine sponge Microciona prolifera has been ex-
plored by colloidal probe microscopy. An in situ coupling of
sulfated and non-sulfated disaccharides to membrane-coated
surfaces was employed to mimic native cell−cell contacts.
The dynamic strength of the homomeric self-association was
measured as a function of calcium ions and loading rate.
A deterministic model was used to estimate the basic
energy landscape and number of participating bonds in
the contact zone.
measure the self-recognition of g-200 as a function of Ca in
5
solution using single-molecule force spectroscopy. Synthetic
derivatives of the sulfated disaccharide coupled to bovine serum
8
9,10
11
albumin, gold nanoparticles,
and gold-coated AFM tips,
unambiguously showed that the presence of Ca2+ is necessary
to induce the required conformational changes to stabilize
dimers formed by the sulfated disaccharide β-D-GlcNAc(3S)-
11
(
1→3)-α-L-Fuc.
One of the major experimental problems in studying multi-
valent binding of otherwise weak cell surface based non-
covalent bonds is the interference by non-specific forces due4
to the absence of the natural matrix, the plasma membrane.
The fluidity of the bilayer paired with an abundance of choline
headgroups disfavors non-specific interactions to a large extent.
Moreover, Helfrich repulsion due to thermal undulation and
steric repulsion of the glycocalyx prevents biomembranes from
eak interactions between glycoconjugates displayed on
cell surfaces play a pivotal role in cell adhesion, and
W
further in cell development, fertilization, metastasis, myelin
12
1
adhering easily. Apart from these technical issues, the impact
compaction, and cell aggregation in sponges. In marine sponges,
of lateral organization on affinity and vice versa is of great
dissociated cells can reaggregate through calcium-mediated asso-
ciation of cell-surface proteoglycans (aggregation factors) in a
13
interest in binding cooperativity.
2
−4
Here, we present a versatile platform based on colloidal probe
microscopy employing lipid bilayer-coated surfaces that allows
measurement of weak multivalent carbohydrate−carbohydrate
interactions in a native-like environment. Scheme 1a shows the
chemical structure of the two disaccharide epitopes derived from
M. prolifera used in this study (see SI for their synthesis). The
disaccharides 1 and 2 were covalently coupled to preformed lipid
bilayers through conjugation of the free thiol group to a
maleimide moiety on a fraction of phospholipids to control the
species-specific manner.
There is growing evidence that
carbohydrate recognition in marine sponges such as Microciona
prolifera could be the allogeneic determinant of sponge self-/
non-self-recognition. This may have provided the basis for early
metazoans ability to resist the negative effects of chimerism, for
5
instance, the insertion of fitness-reducing genes. For cell−cell
adhesion two polyvalent domains of the proteoglycan M. prolifera
aggregation factor (MAF) are important. MAF is composed of
two N-glycosylated proteins, MAFp3 and MAFp4; the former
carries copies of a 200 kDa acidic glycan (g-200), whereas the
14−16
number of carbohydrate groups displayed.
Since thermal
1
undulation of lipid bilayers renders analysis of binding affinity
cumbersome, we used solid-supported membranes (SSMs)
which preserve fluidity but suppress spatial bilayer fluctuations.
Fluidity is crucial considering that lateral organization of ligand-
receptors into clusters as a result of competition between binding
enthalpy and mixing entropy is only possible in laterally mobile
latter possesses about 50 copies of a 6 kDa glycan (g-6). The
MAFp4 arms of the sunburst-like proteoglycans are linked cell-
surface binding receptors, while the exposed g-200 glycans of
the MAFp3 ring participate in calcium-dependent self-association.
Two oligosaccharide epitopes, the pyruvated trisaccharide β-D-
Gal(4,6-(R)-Pyr)-(1→4)-β-D-GlcNAc(1→3)-α-L-Fuc and the sul-
fated disaccharide β-D-GlcNAc(3S)-(1→3)-α-L-Fuc, have been
17,18
matrices.
6
,7
The general setup of our experiment involves a membrane-
coated borosilicate sphere with a diameter of 15 μm attached
shown to be essential for cell−cell interactions in M. prolifera.
The measurement of forces governing self-association of
biologically significant carbohydrates at the single molecule
level with minimal influence from non-specific interactions
in a native-like environment poses a significant experimental
Received: November 2, 2011
Published: January 31, 2012
©
2012 American Chemical Society
3326
dx.doi.org/10.1021/ja210304j | J. Am. Chem. Soc. 2012, 134, 3326−3329

Journal of the American Chemical Society
Communication
a
Scheme 1. Experimental Setup
Figure 1. Typical force retraction curves of membrane-coated colloidal
probes with different functionalizations (a−c) and corresponding
rupture force histograms (d). (a) No disaccharide epitopes, (b)
disaccharide 1, and (c) disaccharide 2 attached to maleimide-bearing
phospholipids (10 mol %) in a DPPC (substrate) and POPC (probe)
matrix. Interactions were probed in calcium buffer before (gray) and
after functionalization of membranes with disaccharide 1 (blue) and
a
Structure of disaccharides 1 and 2 (a) probed by colloidal force
microscopy employing membrane-coated surfaces (b). The disacchar-
ides are covalently coupled to the lipid bilayer through maleimide
chemistry allowing the probing of interaction forces as a function of
calcium ion concentration (c).
2
(green), respectively. Force curves obtained in EDTA containing
to a tip-less cantilever (Scheme 1b). The silicate sphere is
coated with a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocho-
line (POPC) bilayer doped with 10 mol% of 1,2-dioleoyl-sn-
glycero-3-phosphoethanolamine-N-[4-(p-maleimidomethyl)-
cyclohexanecarboxamide] (MCC-DOPE) while the silica substrates
are coated with gel-phase 1,2-dipalmitoyl-sn-glycero-3-phospho-
line (DPPC) doped with 10 mol% 1,2-dipalmitoyl-sn-glycero-3-
phosphoethanolamine-N-[4-(p-maleimidomethyl)cyclohexane-
carboxamide] (MCC-DPPE). The silicate sphere is coated with
a POPC bilayer doped with 10 mol% of MCC-DOPE while the
silica substrates are coated with DPPC doped with 10 mol%
MCC-DPPE. Maleimide functionalized lipids (10 mol%) were
chosen to mimic the high density of sugars displayed by g-200
proteoglycans. DPPC was used to provide stability in the
contact zone and to minimize detachment of the bilayer from
the surface. Formation of SSMs was achieved by vesicle
spreading and monitored by fluorescence microscopy. Fluid
buffer are shown in red. Measurements were performed at a loading
rate of 10 nN/s, 1 s dwell time, and a load force of 200 pN prior to
retraction of the cantilever.
In contrast, coupling of sulfated disaccharides 2 to maleimide
groups of the outer membrane leaflets resulted in a shift to larger
2
+
force that was entirely dependent on the presence of Ca . In
the presence of 2, membrane tethers were frequently formed.
Addition of Ca2+ chelators such as EDTA (5 mM after rinsing
with buffer) abolished any attractive interaction in all cases (red
semitransparent histograms). The histogram of forces found for
the homomeric interaction between 2 in the presence of 10 mM
Ca displays several resolvable peaks that we attribute to attrac-
tive background interaction between the bilayers at low forces
and the disruption of nanoclusters of various sizes at higher
forces. These data establish the necessity of the sulfato group for
self-recognition, and the essential role of Ca2+ ions in mediating
the interactions. Along these lines, Vliegenthart and co-workers
found that the interaction between carbohydrates derived from
M. prolifera is completely abolished upon substitution with other
2+
2
bilayers display diffusion coefficients of 1.4 ± 0.1 μm /s as
determined by fluorescence recovery after photobleaching
only slightly reduced compared to giant liposomes or free-
2+
2+
2+
15
divalent cations such as Mg or Mn , while Cd preserves the
interaction.
standing bilayers. Coupling of the epitopes was quantified by
reflectometric interference spectroscopy (RIfS), revealing a
coupling efficiency of 50% for 2. Details of the experimental
procedures can be found in the SI.
8,11
Figure 2 shows rupture force histograms recorded at differ-
ent loading rates illustrating that the contact zone between
membrane-coated colloidal probe and the SSM comprises a
number of parallel bonds. The estimation of the exact number
of participating bonds requires some general thermodynamic
and kinetic considerations.
The formation of nanoclusters composed of more than a single
dimeric bond is driven by an intricate competition between
enthalpic and entropic contributions on the one hand and kinetic
assembly schemes on the other hand.
argued that cluster size is governed by binding enthalpy favoring
growing clusters and dispersion driven by gain in entropy.
essence, weak bonds with a free energy of only few k T do not
Figure 1 shows a compilation of typical force retraction
curves obtained from colloidal probe microscopy. Membrane-
coated silicate spheres were brought in contact for a dwell time
of 1 s and subsequently retracted from the surface (Figure 1a−c).
Interaction forces in the absence of disaccharides attached to the
SSMs exhibit a rupture force of 25−50 pN that is entirely abol-
ished in the presence of EDTA-containing buffer (Figure 1a).
We attribute the background force to calcium-mediated inter-
action between the choline headgroups. Attachment of disaccha-
ride 1 lacking the sulfato group to the lipid bilayers did only lead
to a small shift in interaction force compared to that in the
absence of sugar epitopes (Figure 1d).
13,17−19
Smith and Seifert
19,20
In
B
lead to large clusters. Those bonds are only formed by
stochastic matching upon contact between substrate and probe,
3
327
dx.doi.org/10.1021/ja210304j | J. Am. Chem. Soc. 2012, 134, 3326−3329

Journal of the American Chemical Society
Communication
barrier and bound state distance, x . Therefore, these two parame-
u
ters contained in eq 1 are fitted to the mean rupture force of a
given cluster size as a function of loading rate (Figure 2).
Figure 2a shows the impact of loading rate on the rupture
force. As expected, the rupture forces shift to larger values with
higher pulling velocity. Importantly, the deterministic model
predicts that rupture forces do not linearly increase with cluster
size as expected from a simple superposition of parallel springs
carrying a shared load. In fact, softer molecular springs with
lower k -values produce less dynamic strength (F ) for a given
s
rup
cluster size (Figure 2b). This behavior can be rationalized by
considering two limiting cases in current force experiments
in which the distance and not the force is controlled. Seifert
distinguishes shared load in which k ≫ k and the force per
s
c
bond depends on the N(t) existing bonds from the non-
cooperative case where N k ≪ k and the force per bond is
b
s
c
2
3,24
independent of N(t).
It is therefore mandatory to estimate
the molecular stiffness prior to a detailed analysis of the dyna-
mic strength. We computed ⟨N k ⟩ from >400 force curves
b
s
and by assuming a Poisson distribution of k used the identity
Figure 2. (a) Rupture force histograms for different loading rates
subject to multipeak fitting of five Gaussian functions to capture the
various cluster sizes. (b) Rupture force as a function of cluster size
obtained by solving the ordinary differential equation of time depen-
dent dissolution of multiple bonds subject to a linear force ramp as a
s
2
k = σ /⟨N k ⟩ to determine k = 3.4 mN/m and N ≈ 4.
s
b
s
s
b
According to the rupture force histograms that were best des-
cribed by five Gaussians (Figure 2a), we assigned cluster sizes
comprising three to five sugar−sugar bonds. This is in good
function of bond stiffness: k = 0.1 (□), 0.025 (○), 0.01 (△), and
s
accordance to estimates from Poisson statistics of k . A third way
s
−1
−1
0
0
5
.005 N m (▽). Parameters for modeling: koff = 0.14 s , x =
u
to guess the initial cluster size very coarsely is to divide the
measured rupture force of the cluster by the rupture force of a
single bond. De Souza et al. measured an average rupture force
of 30 ± 6 pN for the individual bond between two sulfated
disaccharides in the presence of calcium, which corresponds
to approximately seven to eight bonds participating in the
.36 nm, μ = 1 nN/s. (c) Rupture force of clusters consisting of N =
b
bonds as a function of loading rate (▲) subject to fitting the
parameters of eq 1 represented by the solid line. Assumed parameters:
−1
kon = 0, koff = 0.0015 s , k = 0.01 N/m, k = 3.4 mN/m, x = 0.25 nm.
c
s
u
while in contrast strong bonds accumulate in the contact zone.
Assuming that we deal with weak bonds and only one mobile
membrane, accumulation of bonds in the contact zone can be
largely excluded.
11
cluster. In the following analysis we assume N = 5 as derived
b
from the rupture force histograms. Figure 2c shows a fit of eq 1
to the dynamic strength of the largest clusters comprising five
parallel bonds identified from the force histograms (figure 2a,
arrows).
Rupture of multiple bonds under dynamic loading can be
described by a simple kinetic model introduced by Bell and
21−23
The experimental data could be best described by an equi-
further investigated by Seifert.
The rate equation for the
−
1
ibrium off-rate of k ≈ 0.0015 s , and a potential width of x =
time-dependent decrease in the number of bonds N(t) from
off
u
0
.25 nm that are in good accordance with force spectroscopy
initially N reads
b
data obtained from Anselmetti and co-workers measuring the
5
∂tN = −N(t)koff exp(F(t)x /k T) + (N − N(t))k
on
dynamic strength of a single g-200−g-200 interaction. The
u
B
b
potential width is slightly closer to the barrier than predicted
(1)
(
xu = 0.35 nm). The loss in strength due to a closer distance to
with the dissociation rate k , the association rate kon in
the barrier might be rationalized by the more complex oligosac-
charides involved in the homomeric g-200 interaction rendering
the deviation within a reasonable range.
off
equilibrium, and the force per bond F(t) = k νt = μt, with
sys
2
3
v, the pulling velocity and μ, the loading rate. The spring
constant, ksys = k k /(N(t)k + k ), depends on the spring
Considering the rather large membrane-coated probe used in
our study with a radius of 7.5 μm, the dynamic strength of only
very few bonds are assessed after a dwell time of 1s. We attri-
bute this to a combination of various contributions. The effec-
tive contact area is very small considering that only those bonds
carry the load efficiently that reside at the bottom of the poten-
tial which is in the center of the adhesion zone. This effect is
enhanced by the finite surface roughness of the probe (∼0.7
c
s
s
c
constant of the cantilever k and that produced by the involved
c
molecules N(t)k , with k the stiffness of a single-bonded
s
s
molecule. The rupture force of this deterministic model is
defined by Frup = μτ, with τ the lifetime of the cluster defined by
N(τ) = 1. Equation 1 is solved numerically with initial
condition N(t=0) = N (see SI for details). The strategy to
b
obtain as many intrinsic parameters of the system from force
spectroscopy works as follows. First, the number of bonds
involved in the colloidal probe measurement is estimated from
measuring the mean stiffness of the clusters and its variance.
Alternatively, counting the number of maxima occurring in a
rupture force histogram also serves this purpose (vide infra).
25
nm rms). Due to the inherent curvature of the probe, bonds
are only formed at a distance of 70 nm from the center of the
sphere. Taking into account that koff increases exponentially
with distance from the center of the spherical probe only few
bonds effectively participate in a cluster. Enrichment of bonds
in the contact zone is largely prohibited by entropy costs due to
the low binding energy of the homodimers. Besides, the non-
bound sugar moieties act as repeller molecules that produce
repulsion upon close contact. Due to the finite softness of the
The former method also provides the molecular stiffness k of a
s
single-bonded molecule. The second step comprises rupture
force measurements as a function of loading rate to determine
the equilibrium off-rate k at zero load and the distance between
off
3
328
dx.doi.org/10.1021/ja210304j | J. Am. Chem. Soc. 2012, 134, 3326−3329

Journal of the American Chemical Society
Communication
bound molecules, the rupture force becomes less strongly in-
fluenced by the number of bonds constituting the cluster, which
(6) Spillmann, D.; Thomasoates, J. E.; Vankuik, J. A.; Vliegenthart,
J. F. G.; Misevic, G.; Burger, M. M.; Finne, J. J. Biol. Chem. 1995, 270,
5
(
089.
7) Santos, J. I.; de Souza, A. C.; Canada, F. J.; Martin-Santamaria, S.;
Kamerling, J. P.; Jimenez-Barbero, J. ChemBioChem 2009, 10, 511.
8) Haseley, S. R.; Vermeer, H. J.; Kamerling, J. P.; Vliegenthart,
J. F. G. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 9419.
9) de Souza, A. C.; Halkes, K. M.; Meeldijk, J. D.; Verkleij, A. J.;
eventually leads to a systematic underestimation of N . In the
b
limiting case of a very soft linker it is conceivable that rupture
forces of clusters are independent of the number of participating
bonds.
(
In conclusion, we investigated the formation and dynamic
strength defined as the rupture force of carbohydrate nano-
clusters between two opposing membranes equipped with dis-
accharide epitopes derived from the marine sponge M. prolifera.
The weak binding affinity prevents clustering beyond statistical
matching and therefore allows observing the bond breakage
from individual bonds to very small ensembles comprising only
few dimers. The setup based on colloidal force probe allows
monitoring the rupture force as a function of bilayer fluidity,
loading rate, and ionic strength in the absence of thermal
undulations and minimal non-specific interaction forces. The
technique permits measuring the cooperativity of molecular
recognition in a chemically and physically defined way.
Compared to single-molecule mechanics, small clusters allow
the measurement of the effect of rebinding that enhances the
measured strength at low loading rates. The system described
here will permit the measurement of kinetics of dimer forma-
tion and dissolution, with both partners displayed on lipid
bilayers. Many biological systems are constructed in this
manner and studies of vesicle docking as an initial step in fusion
and cellular adhesion will benefit from this general scheme if
the epitopes are synthetically accessible.
(
Vliegenthart, J. F. G.; Kamerling, J. P. Eur. J. Org. Chem. 2004, 4323.
(10) de Souza, A. C.; Halkes, K. M.; Meeldijk, J. D.; Verkleij, A. J.;
Vliegenthart, J. F. G.; Kamerling, J. P. ChemBioChem 2005, 6, 828.
(11) de Souza, A. C.; Ganchev, D. N.; Snel, M. M. E.; van der Eerden,
J. P. J. M.; Vliegenthart, J. F. G.; Kamerling, J. P. Glycoconjugate J. 2009,
2
(
2
(
6, 457.
12) Boulbitch, A.; Guttenberg, Z.; Sackmann, E. Biophys. J. 2001, 81,
743.
13) Weikl, T. R.; Asfaw, M.; Krobath, H.; Rozycki, B.; Lipowsky, R.
Soft Matter 2009, 5, 3213.
(14) Abdulreda, M. H.; Moy, V. T. Biophys. J. 2007, 92, 4369.
(15) Lorenz, B.; Keller, R.; Sunnick, E.; Geil, B.; Janshoff, A. Biophys.
Chem. 2010, 150, 54.
(16) Schuy, S.; Treutlein, B.; Pietuch, A.; Janshoff, A. Small 2008, 4,
970.
(17) Smith, A.-S.; Sackmann, E. ChemPhysChem 2009, 10, 66.
(18) Fenz, S. F.; Smith, A.-S.; Merkel, R.; Sengupta, K. Soft Matter
2
011, 7, 952.
(19) Smith, A.-S.; Seifert, U. Soft Matter 2007, 3, 275.
(20) Fenz, S. F.; Bihr, T.; Merkel, R.; Seifert, U.; Sengupta, K.; Smith,
A.-S. Adv. Mater. 2011, 23, 2622.
(
(
(
21) Bell, G. I. Science 1978, 200, 618.
22) Bell, G. I.; Dembo, M.; Bongrand, P. Biophys. J. 1984, 45, 1051.
23) Seifert, U. Phys. Rev. Lett. 2000, 84, 2750.
ASSOCIATED CONTENT
Supporting Information
Detailed experimental procedures and analytical data for
disaccharides 1 and 2 as well as their precursors, H and
(24) Erdmann, T. Ph.D. thesis, University of Potsdam, 2005.
25) van Zwol, P. J.; Palasantzas, G.; van de Schootbrugge, M.;
■
(
*
S
de Hosson, J. T. M.; Craig, V. S. J. Langmuir 2008, 24, 7528.
1
13
C
NMR spectra, RIfS data, and numerical computation of eq 1.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
ajansho@gwdg.de
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
D.B.W. thanks the Deutsche Forschungsgemeinschaft (Emmy
Noether Fellowship as well as project A5 in SFB 803 “Func-
tionality controlled by organization in and between membranes”)
and the Fonds der Chemischen Industrie. B.L. gratefully acknow-
ledges financial support from the IMPRS “Physics of Biological
and Complex Systems”. This work was partially funded by
the National Institutes of Health (GM065500 and CA125033
to K.K.).
REFERENCES
■
(
(
1) Bucior, I.; Burger, M. M. Curr. Opin. Struct. Biol. 2004, 14, 631.
2) Misevic, G. N.; Finne, J.; Burger, M. M. J. Biol. Chem. 1987, 262,
5
(
2
(
870.
3) Bucior, I.; Scheuring, S.; Engel, A.; Burger, M. M. J. Cell Biol.
004, 165, 529.
4) Garcia-Manyes, S.; Bucior, I.; Ros, R.; Anselmetti, D.; Sanz, F.;
Burger, M. M.; Fernandez-Busquets, X. J. Biol. Chem. 2006, 281, 5992.
5) Fernandez-Busquets, X.; Koernig, A.; Bucior, I.; Burger, M. M.;
Anselmetti, D. Mol. Biol. Evol. 2009, 26, 2551.
(
3
329
dx.doi.org/10.1021/ja210304j | J. Am. Chem. Soc. 2012, 134, 3326−3329