Membrane-Grafted Hyaluronan Films: a Well-Defined
Model System of Glycoconjugate Cell Coats
Ralf P. Richter, Kai K. Hock, Jeffrey Burkhartsmeyer, Heike Boehm, Pit Bingen, Guoliang Wang,
Nicole F. Steinmetz, David J. Evans and Joachim P. Spatz
Materials and Methods
Materials: Lyophilized dioleoylphosphatidylcholine (DOPC) and dioleoylphospatidylethanolamine-
CAP-biotin (DOPE-CAP-Biotin) were purchased from Avanti Polar Lipids (Alabaster, AL, USA).
Lyophilized end-biotinylated hyaluronan with polydispersity indices of less than 1.04 and molecular
weights, MW, of approximately 58 kDa (HA50) and 1083 kDa (HA1000), respectively, were purchased
from Hyalose (Oklahoma City, OK, USA). Lyophilized streptavidin, biotinylated bovine serum albumin
(BSA) and other chemicals were purchased from Sigma. Ultrapure water with a resistivity of 18.2 MΩ
(Millipore, Schwalbach, Germany) was used.
A buffer solution made of 150 mM NaCl, 3 mM NaN3 and 10 mM HEPES, pH 7.4, was prepared in
ultrapure water. 2 mM CaCl2 was added for the incubation step leading to the formation of a supported
lipid bilayer. 2 mM EDTA was added for all other incubation steps.
Lipid mixtures were prepared in chloroform, dried, resuspended in EDTA-containing buffer and
homogenized as described earlier1. Small unilamellar vesicles (EUVs) were obtained by successive
extrusion through polycarbonate membranes with pore sizes of 100 nm and 30 nm (Avanti Polar
Lipids). Concentrations and mixing ratios were estimated from the dry masses of employed lipid
material. Before use vesicle suspensions were diluted to approximately 50 μg/mL.
Lyophilized streptavidin and BSA were reconstituted in ultrapure water to a concentration of
approximately 1 mg/mL as described by the manufacturer. Before use the solutions were diluted to
about 10 μg/mL. The concentration of streptavidin was determined from optical density measurements
at a wavelength of 280 nm using ε280 = 2.2 g-1 L cm-1. Hyaluronan was reconstituted in EDTA-
containing buffer and used at final concentrations of 5 to 40 μg/mL.
Biotinylated Cowpea mosaic virus (CPMV) particles were prepared as follows. The propagation and
purification of CPMV was performed by standard procedures2. CPMV particles were functionalized at
solvent-exposed amines with a succinimide ester reactive biotin (biotin-LCLC-NHS; Perbio,
Cramlington, UK). CPMV particles in 10 mM sodium phosphate buffer at pH 7.0 were exposed to a
3000-fold molar excess of biotin-LCLC-NHS dissolved in dimethyl sulfoxide (DMSO; Aldrich, Poole,
UK). The final DMSO concentration was adjusted to 20% by volume. After overnight incubation at 4°C
the reaction mixture was purified via a 100 K centrifugal device (Pall, Portsmouth, UK). The
biotinylated CPMV particles were still intact after chemical modification, as verified by transmission
electron microscopy (TEM). Successful biotinylation was confirmed by a dot blot technique as
described earlier3. Before use, the solution of biotinylated CPMV particles was diluted to approximately
Ferritin from equine spleen (Aldrich, Poole, UK) was biotinylated at solvent exposed amines. After a
clearing spin at 13000 rpm for 10 min in a table centrifuge and buffer exchange to 10 mM sodium
phosphate buffer at pH 7.0 using a 100 K centrifugal device, biotin-LCLC-NHS, dissolved in DMSO,
was added in a 240-fold molar excess. The final DMSO concentration was adjusted to 5% by volume.
After overnight incubation, the protein was purified from the reaction mixture via a 100 K centrifugal
device. The integrity of the ferritin particles after modification was confirmed by TEM. Western blot
analysis was used to confirm successful biotinylation. After electrophoresis in a 15%
polyacrylamide/SDS gel, the proteins were transferred to nitrocellulose membranes by electroblotting.
The blots were probed with streptavidin-alkaline phosphate (Aldrich, Poole, UK) and detection was
carried out using 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium tablets (Aldrich, Poole,
UK). The ability of streptavidin to bind specifically to a band of the correct molecular weight
demonstrated successful linking of biotin to ferritin. Before use, the ferritin solution was diluted to
approximately 20 μg/mL.
Substrate preparation: Silica-coated QCM-D sensors (Q-Sense, Gothenburg, Sweden) were cleaned
by immersion in a 2% sodium dodecyl sulfate solution for 30 min, rinsing with ultrapure water, blow-
drying with nitrogen, and exposure to oxygen plasma (0.4 mbar, 150 W) (100-E Plasma System, TePla,
Feldkirchen, Germany) for 30 min. Cleaned substrates were stored in air and again exposed to oxygen
plasma (5 min) prior to use.
Glass cover slips (Roth, Karlsruhe, Germany) were cleaned by rubbing with a soft lint-free tissue,
followed by immersion in a solution of H2O2 and H2SO4 (1:4 (v/v), 60 min), rinsing with ultrapure
water, blow-drying with nitrogen, and exposure to oxygen plasma (20 min), and stored in air. Prior to
use, cleaned cover slips were again exposed to oxygen plasma (5 min) and attached with vacuum grease
(GE Bayer Baysilone Paste, Roth) on a custom-made liquid cell.
Determination of the radius of gyration: The radius of gyration, Rg, of HA molecules was
021 . 0
as reported by Takahashi et al.4 for HA in 200 mM NaCl.
6 . 0
Quartz Crystal Microbalance with Dissipation Monitoring (QCM-D)
QCM-D measurements, described in detail elsewhere5, were performed with a Q-Sense E4 system
(Q-Sense). Briefly, upon interaction of (soft) matter with the surface of a sensor crystal, changes in its
resonance frequency, Δf, related to attached mass (including coupled water), and in its dissipation, ΔD,
related to frictional (viscous) losses in the adlayer are measured with a time resolution of better than 1 s.
The system was operated in flow mode, with a flow rate ranging from 5 to 100 μL/min. Sample
solution was continuously delivered to the measurement chamber by the aid of a peristaltic pump
(ISM597D, Ismatec, Zürich, Switzerland). In order to switch between sample liquids, the flow was
interrupted for a few seconds without disturbing the QCM-D signal. With this setup, adsorption and
interfacial processes can be followed in situ while successively exposing different solutions to the
surface. The working temperature was 24°C.
Resonance frequency and dissipation were measured at 6 harmonics (15, 25 … 65 MHz),
simultaneously. If not stated otherwise, changes in dissipation and normalized frequency (Δf = Δfn/n,
with n being the overtone number) of the third overtone (n = 3, i.e., 15 MHz) are presented.
Viscoelastic Modelling of QCM-D data: Fitting to a viscoelastic model12,13 was performed with the
software QTools (Q-Sense). The hyaluronan film was represented as a viscoelastic layer with viscosity
η and shear modulus μ. The viscoelastic properties were allowed to be linearly frequency dependent
(“extended model”). The layer’s thickness and density were kept fixed at the value determined by 3W-
RICM and at 1 g/cm3, respectively. The semi-infinite bulk solution was assumed to be Newtonian with
a viscosity of 0.9 mPa·s. The composite streptavidin-SLB layer exhibited very low dissipation and was
therefore treated as rigid (“Sauerbrey layer”). All measured overtones were included in the fitting
For all hyaluronan films, a unique fit of good quality was obtained (figure S1). The viscosities were
well-determined, with changes of less than ±2% leading to more than twofold increases in χ2. Shear
moduli were small but not negligible, and were determined within an error of less than ±70% within the
same limit of χ2. Both viscosity and shear modulus exhibited no or little frequency dependence. Slight
variations in the chosen film thickness only weakly affected the viscoelastic properties. This can be
rationalized by the fact that the penetration depth of the acoustic shear wave is similar to (for HA50) or
smaller (for HA1000) than the film thickness.
3579 11 13
3579 11 13
Figure S1. Experimental and fitted QCM-D responses for films made of HA50 (A) and HA1000 (B)
after 2 hours of incubation. Frequencies are normalized to the overtone number n.
Triple-Wavelength Reflection Interference Contrast Microscopy (3W-RICM)
RICM, a microinterferometric technique, described in detail elsewhere6,7, measures the distance
between a transparent planar substrate (here a glass cover slip) and the surface of a colloidal probe, that
contacts or hovers over the substrate. Polystyrene beads (Polysciences, Eppelheim, Germany) with
around 20 μm diameter were used as colloidal probes. Layer thicknesses were determined with a simple
model (parallel plate approximation with incident light parallel to the surface normal)8 from the position
of the first extremum in the radially averaged interferometric intensity profile. The diameter of the bead
was determined from bright field images.
Figure S2. Representative 3W-RICM data. (A-C) Interferometric images of a colloidal bead hovering
over a HA50-film. The three images were acquired simultaneously at wavelengths of 490 nm (A),
546 nm (B) and 630 nm (C). (D) Typical trace of the in-plane movements of a bead’s center over a
period of 25 s: directed movement (most likely due to convective movement of surrounding fluid)
occurs in addition to random thermal movements, and indicates that the bead remains mobile on the
hyaluronan film. (E) Expected normalized intensities in the centre of the bead as a function of bead-to-
surface distance (thin lines in blue (490 nm), green (546 nm) and red (630 nm)). For each individual
wavelength, a multitude of distances would be in agreement with the interferometric data obtained on a
HA1000-film (bold curve segments). By combining all three wavelengths, a solution can be found
(black vertical line), which is unique within an interval of at least 900 nm of bead-to-surface distance.
For the unambiguous determination of distances in the range of several hundred nanometers,
simultaneous measurements at several wavelengths are required. Details of the technical
implementation of triple-wavelength RICM, inspired by the approach of Schilling et al.6, will be
described in a forthcoming publication.
Representative data from our 3W-RICM measurements are assembled in figure S2. The distances
determined from all three wavelengths (490, 546 and 630 nm) were usually found to overlap within an
interval of less than 10 nm. Tracking of the in-plane movement of the bead’s center revealed random
thermal movements, sometimes coexisting with directed movements (figure S2D). This indicates that
the beads remain mobile on the hyaluronan films, with little or no attractive interaction with the
hyaluronan chains. Moreover, any electrostatic interactions that do occur are rather short ranged under
the applied conditions: the Debye length is below 1 nm. Hence we expect the measured bead-surface
distances to be a good measure of the HA-films’ thickness.
We employed reflectometry9 to determine the mass of adsorbed hyaluronan. A prototype setup that
combines reflectometry and QCM-D on the same support was kindly provided by Q-Sense. Details of
the technical implementation will be described in a forthcoming publication. Briefly, the ratio S = Ip/Is
of the intensities of the p- and s-polarized light is measured, as the light is reflected from the support.
Assuming the adsorbed film to be homogeneous with a given thickness, d, and refractive index, n, the
adsorbed mass can be determined according to:
with S0 being the intensity ratio before film deposition. The sensitivity factor, As, depends on the optical
setup (incident angle and wavelength of the polarized light, properties of the support and the
surrounding solvent) and was calculated with custom-made software. Refractive index increments,
dn/dc, of 0.180 cm3/g and 0.151 cm3/g were used for streptavidin10 and hyaluronan4, respectively.
For thin layers (d < 25 nm) and refractive indices characteristic for our streptavidin layers
(n-nsolvent < 0.2), we found the sensitivity factor to be independent of d and n (As = 0.055±0.005 nm-1).
We determined a mass of 220 ng/cm2 for a full monolayer of streptavidin. Within less than 10% error
this is in accordance with measured data available in the literature10 as well as estimates for a close-
packed layer of streptavidin11, and thus validates our quantitative approach.
For refractive index changes pertinent to the hyaluronan layers (n-nsolvent < 0.01), the dependence of
As on the refractive index remains small even for thicker layers (figure S3B). The sensitivity factor,
however, becomes dependent on d. In order to choose the appropriate sensitivity factor, we therefore
used the thicknesses determined by 3W-RICM.
Figure S3. (A) Representative reflectometric data for the formation of a HA50-film. (B) Sensitivity
factor as a function of layer thickness for different refractive index differences, Δn = n-nsolvent, that are
relevant for the HA-films. The dependency on the refractive index is weak.
The assumption of homogeneity for these layers was chosen for simplicity. As many polymer brushes
exhibit a pronounced density profile, this approximation may be rather rough for the hyaluronan films.
In effect, the layer thickness that is effectively seen by reflectometry may be smaller than the thickness
determined by RICM, due to differences in the perception of contrast by these techniques12. Given the
dependence of As on the layer thickness, the calculated masses therefore represent an upper bound for
the adsorbed amounts of hyaluronan.
The average distances, s, between neighboring sites of hyaluronan anchorage were calculated
, with NA being Avogadro’s number and by assuming hexagonal
packing of the sites. The hydration was calculated by comparing the mass of adsorbed hyaluronan with
the mass that corresponds to a water film with a thickness as determined by 3W-RICM.
Determination of Mean Diffusion Constants in Thin Hyaluronan Films
Diffusion through hyaluronan films was studied by QCM-D detection of biotinylated probes as they
adsorbed from a bulk reservoir through the film onto streptavidin-covered SLBs. Under conditions of
laminar shear flow in a slit, a steady state could be achieved in the initial phase of adsorption, as
confirmed by the constant adsorption rates observed:
Here, c0 is the probe concentration in the bulk. The overall kinetic constant, k, is determined by the
transport (diffusion and convection) of the probe in the bulk solution and the probe’s diffusion through
The interplay of these two contributions can be adequately described by a theoretical framework as
outlined by Déjardin et al.14. Here, we associate kinetic constants, kb and kf, to the transport in the bulk
and to the diffusion through the film, respectively. The inverse of the kinetic constants can be
interpreted as resistances which, in a simple approximation, are additive14:
kb can be determined from a reference measurement of the adsorption rate, (
streptavidin-coated SLB without immobilized hyaluronan under otherwise identical conditions. With
the experimentally accessible parameter
( ) (
If the adsorption is predominantly controlled by the diffusion through the film (kb > kf, i.e., α < 0.5), the
simple approximation overestimates kf by less than 15% 14. For α < 2/3, the error is still smaller than
Alternatively, the total transport process can be viewed as diffusion through two layers, one layer
(index f) being the hyaluronan film itself, the other (index b) being an adjacent layer that becomes
depleted in probe molecules14. In this case15:
with l and D being the layers’ thickness and diffusion constant, respectively. From equations 2 to 5, one
, on a
The kinetic constants, kb, can in principle be determined independently for all probes of interest. This
requires, however, an accurate determination of the bulk concentration, c0, and the degree of hydration,
as sensed by QCM-D, of each probe. It appears simpler to determine kb for only one (reference) probe at
i The interfacial reaction of the biotinylated probe with streptavidin is neglected here, as adsorption of
streptavidin to a biotin-covered surface was previously found to be transport controlled under similar
conditions (Reimhult, E.; Larsson, C.; Kasemo, B.; Höök, F. Anal. Chem. 2004, 76, 7211-7220).
a given flow rate. The kinetic constants for other probes and other flow speeds can than be easily
calculated according to the following scheme. For the adsorption under laminar flow conditions along a
slit of given geometry, the kinetic constant kb is given by16
flow of fluid through the slit. A is a constant that is determined by the geometry of the slit. The probes’
bulk diffusion constant can be estimated from the Stokes-Einstein equation
where RT/NA and ηb are the thermal energy and the bulk viscosity, respectively, and d is the effective
diameter of the probeii. Thus
, and the kinetic constant for one probe (index 2) with a given
diameter and at a given flow rate can be determined from the kinetic constant obtained for another
probe (index 1) by:
( ) 1
We chose the adsorption of streptavidin to a biotinylated SLB as our reference system, and assumed
an average hydration of around 80% as reported by Reimhult et al.10 for low coverage.
(1) Richter, R. P.; Mukhopadhyay, A.; Brisson, A. Biophys. J. 2003, 85, 3035-47.
(2) Wellink, J. Meth. Mol. Biol. 1998, 81, 205-9.
(3) Steinmetz, N. F.; Calder, G.; Lomonossoff, G. P.; Evans, D. J. Langmuir 2006, 22, 10032-7.
(4) Takahashi, R.; Kubota, K.; Kawada, M.; Okamoto, A. Biopolymers 1999, 50, 87-98.
(5) Rodahl, M.; Höök, F.; Krozer, A.; Brzezinski, P.; Kasemo, B. Rev. Sci. Instrum. 1995, 66, 3924-
(6) Schilling, J.; Sengupta, K.; Goennenwein, S.; Bausch, A. R.; Sackmann, E. Phys. Rev. E 2004, 69,
(7) Rädler, J.; Sackmann, E. Langmuir 1992, 8, 848-53; Rädler, J.; Sackmann, E. J. Phys. II France
1993, 3, 727-48.
(8) Kühner, M.; Sackmann, E. Langmuir 1996, 12, 4866-76.
(9) Dijt, J. C.; Cohen Stuart, M. A.; Fleer, G. J. Adv. Colloid Interface Sci. 1994, 50, 79-101.
(10) Reimhult, E.; Larsson, C.; Kasemo, B.; Höök, F. Anal. Chem. 2004, 76, 7211-20.
(11) Brisson, A.; Bergsma-Schutter, A.; Oling, F.; Lambert, O.; Reviakine, I. J. Cryst. Growth 1999,
(12) Domack, A.; Prucker, O.; Rühe, J.; Johannsmann, D. Phys. Rev. E 1997, 56, 680-9.
(13) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scripta 1999, 59, 391-6.
(14) Déjardin, P.; Vasina, E. N. Colloids and Surfaces B: Biointerfaces 2004, 33, 121-7.
(15) Crank, J. The Mathematics of Diffusion; 2nd edition ed.; Clarendon Press: Oxford, 1975.
(16) Hermens, W. T.; Benes, M.; Richter, R. P.; Speijer, H. Biotechn. Appl. Biochem. 2004, 39, 277-
Q ADk =
, with Q being the volume
ii For the sake of simplicity, we neglect the effect of particle shape.