Dynamics of the interaction between a fibronectin molecule and a living bacterium under mechanical force.

Abstract

Fibronectin (Fn) is an important mediator of bacterial invasions and of persistent infections like that of Staphylococcus epidermis. Similar to many other types of cell-protein adhesion, the binding between Fn and S. epidermidis takes place under physiological shear rates. We investigated the dynamics of the interaction between individual living S. epidermidis cells and single Fn molecules under mechanical force by using the scanning force microscope. The mechanical strength of this interaction and the binding site in the Fn molecule were determined. The energy landscape of the binding/unbinding process was mapped, and the force spectrum and the association and dissociation rate constants of the binding pair were measured. The interaction between S. epidermidis cells and Fn molecules is compared with those of two other protein/ligand pairs known to mediate different dynamic states of adhesion of cells under a hydrodynamic flow: the firm adhesion mediated by biotin/avidin interactions, and the rolling adhesion, mediated by L-selectin/P-selectin glycoprotein ligand-1 interactions. The inner barrier in the energy landscape of the Fn case characterizes a high-energy binding mode that can sustain larger deformations and for significantly longer times than the correspondent high-strength L-selectin/P-selectin glycoprotein ligand-1 binding mode. The association kinetics of the former interaction is much slower to settle than the latter. On this basis, the observations made at the macroscopic scale by other authors of a strong lability of the bacterial adhesions mediated by Fn under high turbulent flow are rationalized at the molecular level.

Full text

Page 1

Dynamics of the interaction between a fibronectin
molecule and a living bacterium under
mechanical force
Yasser Bustanji*†‡, Carla Renata Arciola§¶, Matteo Conti*‡, Enrico Mandello*‡, Lucio Montanaro§¶?, and Bruno Samorı ´*‡?
*Dipartimento di Biochimica, Universita ` degli Studi di Bologna, Via Irnerio 48, 40126 Bologna, Italy;‡Istituto Nazionale per la Fisica della Materia, Bologna,
Italy;§Laboratorio di Biocompatibilita ` dei Materiali da Impianto, Istituti Ortopedici Rizzoli, Bologna, Via di Barbiano 1?10, 40136 Bologna, Italy; and
¶Dipartimento di Patologia Sperimentale, Universita ` degli Studi di Bologna, Via San Giacomo 14, 40126 Bologna, Italy
Communicated by Carlos J. Bustamante, University of California, Berkeley, CA, August 20, 2003 (received for review November 25, 2002)
Fibronectin(Fn)isanimportantmediatorofbacterialinvasionsand
of persistent infections like that of Staphylococcus epidermis.
Similar to many other types of cell–protein adhesion, the binding
between Fn and S. epidermidis takes place under physiological
shear rates. We investigated the dynamics of the interaction
between individual living S. epidermidis cells and single Fn mole-
cules under mechanical force by using the scanning force micro-
scope. The mechanical strength of this interaction and the binding
site in the Fn molecule were determined. The energy landscape of
the binding?unbinding process was mapped, and the force spec-
trum and the association and dissociation rate constants of the
binding pair were measured. The interaction between S. epider-
midis cells and Fn molecules is compared with those of two other
protein?ligand pairs known to mediate different dynamic states of
adhesion of cells under a hydrodynamic flow: the firm adhesion
mediated by biotin?avidin interactions, and the rolling adhesion,
mediated by L-selectin?P-selectin glycoprotein ligand-1 interac-
tions. The inner barrier in the energy landscape of the Fn case
characterizes a high-energy binding mode that can sustain larger
deformations and for significantly longer times than the corre-
spondent high-strength L-selectin?P-selectin glycoprotein ligand-1
binding mode. The association kinetics of the former interaction is
much slower to settle than the latter. On this basis, the observa-
tions made at the macroscopic scale by other authors of a strong
lability of the bacterial adhesions mediated by Fn under high
turbulent flow are rationalized at the molecular level.
C
phology, motility, gene expression, and survival of cells (1).
Fibronectin (Fn) is an extracellular matrix protein that is widely
distributed in the tissues of all vertebrates and is a potential
ligand for most cell types (2). Fn is also an important mediator
of bacterial invasions and of persistent infections like that of
Staphylococcus epidermidis (3). Until now, S. epidermidis was
considered a mere saprophyte, usually harboring on the skin and
mucosae. Recently, this organism has been shown to also act as
a pathogen, mainly in association with surgical applications of
biomaterials (4).
Like many other cell–protein adhesions, the binding between
Fn and staphylococci takes place under physiological shear rates
(5). It has been reported that the high shear stress characteristic
of the turbulent flow at blood vessel entrances and bifurcations
can prevent Fn-mediated adhesion of Staphylococcus aureus to
the endothelial cells lining the vessel, reducing, in this way, the
chances of bacterial invasion (6). Thus, a thorough character-
ization of the binding?unbinding dynamics of these interactions
is necessary to better understand the molecular mechanisms of
Fn-mediated pathogenesis. Moreover, specific cell–protein in-
teractions have been selected by nature to mediate different
dynamic processes of adhesion of a cell under hydrodynamic
flow, as in the case of the rolling adhesion of leukocytes whose
velocity in the blood vessels is controlled by transient binding
ell–protein adhesion regulates essential processes in multi-
cellular organisms and provides signals that affect the mor-
events to the inner walls. It is of interest to understand how
different cell?protein adhesions are optimized in their functions
through the details of the interactions at the molecular level. In
this article, we address this issue by comparing the interaction
dynamics of three different adhesion processes: firm adhesion
mediated by the biotin?avidin pair, rolling adhesion, which is
mediated by the L-selectin?P-selectin glycoprotein ligand 1
(PSGL-1) system, and cell?bacterium adhesion mediated by
Fn?adhesin binding.
Traditionally, adhesion interactions have been studied in
reversible equilibrium conditions. However, in vivo, these inter-
actions take place under significant shear forces with pulling
rates that are much faster than the relaxation rates of the binding
pair, and thus, the binding?unbinding process occurs under
nonequilibrium, irreversible conditions. The development of
single-molecule manipulation methods, such as scanning force
microscopy (SFM) and optical tweezers, has made it possible to
investigate the dynamics of these processes under nonequilib-
rium conditions, and to measure their force-dependent dissoci-
ation kinetics. Here, we investigate the dynamics of the inter-
actions between individual living S. epidermidis cells and single
Fn molecules, map the energy landscape of the binding?
unbinding process, and measure the association and dissociation
rate constants of the binding pair by using SFM. In particular, we
show that two parameters, the bond length extension required to
reach the transition state from the minimum energy equilibrium
position of the bound state, and the bond survival time extrap-
olated at zero force, play a crucial role in determining the
dynamics of the interaction.
Materials and Methods
Bacterial Strains. Single colonies of each bacterial strain were
seeded in 8 ml of triptycase soy broth. After incubation for 24 h
at37°C,thebrothculturewasfractionatedin1-mlaliquotsstored
at ?80°C. Their genus and species were identified by the
Api-Staph test (BioMerieux, Charbonnier les Bains, France) and
by PCR methods (7) (see Supporting Materials and Methods,
which is published as supporting information on the PNAS
web site).
Preparation of Bacterial Confluent Lawns. Slime-producing strains.
Polystyrene discs cut from Petri dishes (Bibby Sterilin, Stone,
Staffordshire, U.K.) were incubated in 3.5 ml of triptycase soy
broth (TSB) with the addition of 350 ?l of stored strain aliquot
at 37°C. After 24 h, 40% of the culture medium was substituted
with fresh TSB and the incubation was continued for another
Abbreviations: SFM, scanning force microscopy; Fn, fibronectin; PSGL-1, P-selectin glyco-
protein ligand 1.
†Present address: Faculty of Pharmacy, University of Jordan, Amman 19328, Jordan.
?To whom correspondence should be addressed. E-mail: lucio.montanaro@alma.unibo.it.
or samori@alma.unibo.it.
© 2003 by The National Academy of Sciences of the USA
13292–13297 ?
PNAS ?
November 11, 2003 ?
vol. 100 ?
no. 23www.pnas.org?cgi?doi?10.1073?pnas.1735343100

Page 2

48 h. With the replacement of part of the culture medium, the
periodoftheexponentialgrowthofthemicrobialpopulationwas
prolonged,achievingacompleteanduniformcolonizationofthe
discs. At the end of the incubation, the discs were washed five
times with physiologic solution (saline) to remove nonadherent
bacteria.
Non-slime-producing strains.Instead of polystyrene discs, Fn-coated
discs were used. The coating was achieved by covering a stainless
steel disk (?1 cm in diameter) with a 5-nm layer of 99.99%
titanium (Sigma), followed by a 30-nm layer of 99.99% gold
(Sigma), by resistive evaporation in high vacuum (10?6torr; 1
torr ? 133 Pa). These discs were immediately immersed into a
100 ?g?ml solution of human Fn (Sigma) in PBS for 20–30 min.
The growth of the bacterial lawn was then obtained on these
Fn-coated discs as above. To obtain the maximal expression of
adhesins on bacterial surface, exponential growth conditions
were achieved. After 24 h, 40% of the culture medium was
substituted with fresh triptycase soy broth (previously brought to
37°C) and incubation was continued for another 48 h. With the
replacement of part of the culture medium, the period of
exponential growth of the microbial population is prolonged,
achieving a complete and uniform colonization of the disk by
staphylococcal cells, which in this growth phase exhibited,
besides, their maximal expression of adhesins (8). Although the
number of adhesin molecules per cell has not been described for
S. epidermidis, in the case of S. aureus, in exponential growth
15,000 adhesins per cell have been estimated by flow cytometry
(9). The Fn adhesins are known to be uniformly distributed on
the cell surface (10, 11), with different concentrations in the
different strains.
Dynamic Force Spectroscopy (DFS) Experiments. Silicon nitride
(Si3N4) probes (Microlevers, ThermoMicroscopes; Sunnyvale,
CA) were gold-coated, following the same procedure as with the
metallic discs (see above). A few Fn molecules were covalently
bound to the probe by sulfur–gold bonds. The probes after that
functionalization with Fn were made to approach a bacterial
confluent lawn, kept in contact for a desired time (see below),
and then retracted. This cycle is normally repeated multiple
times in sequence. Whenever a molecular bridge is formed
because of a bond settled between at least one Fn molecules on
the tip and Fn adhesins on the cell surface, the force acting on
this bridge is reported vs. the tip displacement, as in Fig. 1. All
of the force curves thus obtained were recorded in PBS (150 mM
NaCl?150mMsodiumphosphate,pH7.4),byusingaNanoscope
IIIa (Digital Instruments, Santa Barbara, CA). The spring
constant(k)ofthecantileverswas?50?10pN?nm,determined
by the thermal oscillations method (12, 13). The loading rate
values were obtained by multiplying the experimental spring
constant of the cantilever by the pulling velocity. All of the probe
approaches were made at a constant speed of 1 ?m?s, and the
loadingforceatthecontactpointwaskeptatavalueof?400pN.
The saturation of the bacterium adhesins with Fn was achieved
directly into the Nanoscope III fluid cell by injecting a 10 ?g?ml
Fn solution in PBS, and, after ?20 min, the unbound Fn was
washed out with PBS before starting recording the force curves.
Results and Discussion
Force Spectroscopy. Fn exists in the plasma and other body fluids
as a soluble dimeric molecule, whereas in the extracellular
matrix, it is present in insoluble multimeric forms. The dimeric
molecule consists of two 220 to 250-kDa monomers joined by
two disulfide bonds at the C terminus of each subunit (2). Fn is
rich in sulfur atoms. Besides these two disulfide bridges, each
monomer contains 28 additional disulfide bridges and two free
cysteine groups (14). We could take advantage of these sulfur
atoms to immobilize the Fn molecules on a previously gold-
coated tip through covalent sulfur–gold bonds. These bonds
do not lead to a loss of the binding ability of Fn molecules
to the Fn-adhesins on the surface of S. aureus (11) or of S.
epidermidis (15).
DFS measurements on the wall of a single cell were performed
on a confluent bacterial lawn to increase the chances that the tip
will contact one bacterium in each approach of the SFM tip to
the surface, thus circumventing the need for prior imaging and
localization of the cells. This strategy also avoided potential tip
contamination, which often occurs during imaging, and that can
affectthesubsequentforcemeasurements,andresultinartifacts.
The S. epidermidis strains used in the present study were
clinically isolated and selected on their ability to bind Fn and to
produce slime.
Preliminary DSF experiments were carried out with slime-
producing strains, but the force-curves profiles obtained with
them were poorly reproducible and acquired the characteristic
shapes shown in Fig. 1 only after many approach-retraction
cycles. Thus, we decided to carry out the experiments with a
non-slime-producing strain. The choice of using bacteria without
slime for the experiments obliged us to develop another method
ofdeposition,becausenon-slime-producingbacteriadonotform
confluent lawns on polystyrene surfaces. To avoid the use of
glutaraldehyde as a cross-linker (16), we took advantage of the
dual capability of Fn to bind both to a gold-coated substrate
(through the thiol–gold bonds), and to the surfaces of the
bacteria. A gold-coated disk was functionalized with Fn and then
used as a substrate to immobilize the bacteria by Fn?Fn-adhesin
interactions. Confluent bacterial lawns were obtained, as con-
firmed by SFM imaging.
The surface of the bacteria not adhering to the substrate
surface was exposing Fn adhesins. The DFS experiments were
carried out on that surface by approaching and retracting the
functionalized tip. Force curves were recorded with force load-
ing rates ranging from 5 ? 103to 105pN?s, which brackets the
loading rate range (3 ? 104to 6 ? 105pN?s) acting between a
cell moving in a shear flow and a substrate, as described and
related to physiological conditions by Chen and Springer (17).
Fig. 1.
obtained when an SFM tip functionalized with Fn molecules is made to
approach (dotted trace) to a living bacterium and then retracted (solid trace).
The curves with a single peak (as in A) monitor the rupture of interactions
between the tip and the bacterium sustained by single molecular bridges; the
peaks in curves with multiple peaks (as in C) may be ascribed either to
unfolding and?or to unbinding events. The curves as in B might be due only
tononspecificinteractions,andthosecurvesasinDmaybeduetoexperiments
in which no binding interactions were settled.
Representative profiles of the force (F) vs. extension (E) curves
Bustanji et al.
PNAS ?
November 11, 2003 ?
vol. 100 ?
no. 23 ?
13293
BIOCHEMISTRY
BIOPHYSICS

Page 3

The retraction (pulling) traces exhibited either one or multiple
peaks. Among the retraction traces with one peak, only those
displaying a sign change in curvature after the contact point (Fig.
1A) were considered for further analysis because they were
interpreted as resulting from specific interactions (18, 19). All
other traces (see, for example, Fig. 1B) were assumed to result
from nonspecific interactions and were discarded. Among the
traces with more than one peak (Fig. 1C), only the last peak was
included in the analysis (20), because this is the only one that can
be unambiguously identified with a single Fn–cell detachment
event. In contrast, the other peaks can either result from the
breaking of single ligand–receptor interaction, or from the
mechanically induced unfolding of the weakest domains in Fn
(21, 22).
The possibilities that the Fn–cell detachment events resulted
either from the detachment of Fn from the gold coated tip or
from the uprooting of the Fn-adhesin from the surface of the
bacteria, were excluded because of the following two reasons:
First, Fn was covalently attached to the tip and covalent bonds
can withstand forces of ?2nN (23). Second, the experiments
reported in Figs. 2 and 4 were carried out with the same tip: at
its end, there were only a few Fn molecules and uprooting events
should have resulted in a loss of its functionality. The ability to
discriminate between receptor-ligand dissociation and receptor
uprooting constitutes one of the advantages of single-molecule
mechanical experiments over classical kinetics studies of cellular
adhesion, where such distinction is usually very difficult to
establish (24).
The force curves we first recorded with a tip-pulling velocity
of 1 ?m?s led to the distribution of the rupture forces shown in
Fig. 2a (n ? 1,000). The zero force bar corresponds to the tip
approaches to the surface (65%) that did not result in an
adhesion interaction (Fig. 1D), i.e., to the unsuccessful binding
events. This distribution of rupture forces shows a single peak
centered at ?85 ? 9 pN (Fig. 2A). This monomodal distribution
indicates a well defined and discrete quantum of mechanical
force and is consistent with the breaking of interactions between
individual molecules of Fn and adhesins on the surface of S.
epidermidis.
When the adhesins on the surface of the bacteria were
passivated by flooding a concentrated solution of Fn, most
(78%) of the curves did not show any adhesion event. The other
curves led to a very broad and flat distribution of the rupture
forces (Fig. 2b). Moreover, a monomodal distribution, like that
in Fig. 2a, reappeared after the passivating molecules of Fn were
washed out with a diluted solution of acetic acid (25). When a
clinically isolated mutant strain (7) that did not express adhesins
was used, a distribution of the rupture forces was obtained (Fig.
2c), which was very similar to that in Fig. 2b. These two
independent control experiments confirmed that the 85 pN
deadhesionforcewasassociatedtoaspecificinteractionbetween
a Fn molecule and S. epidermidis bacterium.
To determine the location of the specific interaction between
a Fn molecule and S. epidermidis bacterium, we used specific
antibodies against different domains of Fn. We found that the
specific peaks in the force curves disappeared when the Fn-
functionalized tip was preincubated for 2–4 h at 37°C with a 100
?g?ml solution of a monoclonal antibody against the 12 type III
domain (heparin-binding site) of Fn located near the C terminus
(Takara Biomedicals, Tokyo; clone no. FNH3-8). This conclu-
sion was confirmed at a macroscopic level by incubating a
Fn-functionalized gold-coated disk with the same antibody; no
bacteriaadheredtothediskundertheseconditions.Thelocation
of the interaction at the heparin-binding site near the C terminus
by these passivating experiments is consistent with surface
plasmon resonance (SPR) results that shows the higher affinity
of S. epidermidis for the C-terminal fragment of Fn over its
N-terminal fragment (26).
MappingtheDeadhesionEnergyLandscapeUnderaMechanicalForce.
The value of 85 pN corresponding to the most probable value of
the deadhesion force is not a fundamental property of this
ligand-receptor pair, but it depends on the loading rate used in
these experiments. The loading rate is a measure of the rate at
which the force is applied to the sample. An external mechanical
force, properly directed, tilts and deforms the energy landscape
of the unbinding path (Fig. 3) so that the lifetime of the bond is
shortened significantly below its natural lifetime (? ? 1?koff).
Under external force, the dissociation rate is modified as de-
scribed in refs. 27 and 28:
koff(F) ? koff(0)e?Fx??kBT?
[1]
Fig. 2.
c) The histograms are relative to the two control experiments: one with a passivation of the Fn adhesins by flooding Fn (b), and one with a mutant strain not
expressing the Fn adhesins (c). The bars at zero force corresponds to the unsuccessful binding approaches.
(a) Distribution of the binding probability vs. the rupture forces (measured with a 1 ?m?s pulling velocity) of the Fn?Fn adhesins interactions. (b and
Fig.3.
deforms the energy landscape of an unbinding path, and how, on increasing
the force, one barrier along the reaction coordinate (rc) is reduced and
ultimately suppressed. If the highest barrier is located sufficiently out, it is
possible for an inner, smaller barrier to emerge as the rate-determining
barrier in a mechanical experiment. In principle, by varying the force loading
rate on the molecule, one can use direct mechanical manipulation to make
different barriers emerge as the rate-determining barriers, and thus map the
dissociation landscape.
Aschematicviewillustratinghowanexternalmechanicalforce(F1,F2)
13294 ?
www.pnas.org?cgi?doi?10.1073?pnas.1735343100Bustanji et al.

Page 4

where koff(F) is the dissociation rate constant of the binding pair
under the applied force, (F), x?is the reaction length over which
the force must be applied to reach the transition state from the
equilibrium position of the bound state, kB is the Boltzmann
constant, and T is the absolute temperature. In this expression,
koff(0) is the dissociation rate constant extrapolated to zero
force, whose linear correlation with the unbinding force was first
demonstrated by Schwesinger et al. (29). For an energy land-
scape of only one barrier, this force-activated rate, koff(0),
becomes equal to the ‘‘natural’’ koffthat can be determined by
bulk methods like SPR (30). In the case of rugged energy
landscapes with more than one barrier, koff(0) and koffwill not
necessarily coincide because their values could be related to
different barriers.
The bond strength is defined as the most probable value for
the rupture force Fmp, (31, 32), i.e., to the maximum of the
distribution in Fig. 2 and depends on the loading rate dF?dt on
the ligand (28) according to:
Fmp? (kBT?x?)ln[dF?dt x??koff(0)kBT].
[2]
The most probable rupture forces obtained at loading rates
ranging from 5 ? 103to 105pN?s are plotted against the natural
logarithm of the loading rate in Fig. 4. A linear plot is obtained
with a slope (kBT?x?) that corresponds to a barrier located at x?
? 3.3 Å along the reaction coordinate. This same plot yields a
valueof?(0)?1?koff(0)?0.21sforthesurvivaltimeofthebond
extrapolated to zero force.
Force Spectra of Cell-Adhesion Proteins. Fig. 4 also shows the force
spectra for two representative cases of two other protein?ligand
pairs that mediate different dynamic states of adhesion of a cell
under a hydrodynamic flow: L-selectin or P-selectin?PSGL-1
that mediate the so-called rolling adhesion (33), and avidin?
biotin, which mediates firm cell adhesion, as reported by Evans
et al. (34) and by Merkel et al. (35), respectively. Unlike the plot
obtained here for the Fn?Fn–adhesin interaction, which shows a
single linear regime, these two plots reveal two distinct linear
regimes. As shown by Evans and Ritchie (28), each linear regime
in a force spectrum corresponds to the overcoming of a single
energy barrier along the unbinding pathway of the system.
Detachment from a bound state by a ligand confined by more
than one sharp barrier should lead to multiple linear regimes in
the force spectrum plot, as observed for selectin and avidin
interactions. These linear regimes make it possible to locate the
barriers along the unbinding pathways, and to yield the koff(0)
andthereforethetimesurvival,?(0),ofthebondwithineachwell
determined by each barrier. On the basis of these two param-
eters, potential energy profiles can be temptatively sketched as
in Fig. 5.
Thermally induced dissociation is always controlled by the
highest barrier of the landscape. In contrast, mechanically
induced dissociation is controlled not only by the height of the
barrier but also by the position of the barrier along the reaction
coordinate. In particular, the reduction and ultimate suppression
of the barrier height by the external force is greater the further
Fig. 4.
Gaussian fit of distributions like that in Fig. 2a and reported with their SD) at
different loading rates (pN?s) identify for the Fn?Fn-adhesin interaction one
linear regime that maps, in an energy landscape, a barrier at x?? 3.3 Å and a
timesurvival(extrapolatedatzeroforce)?(0)?0.21s.Thecorrespondingplots
of L-selectin?PSGL-1 (B), and avidin?biotin (C) are reproduced here from refs.
34 and 35, respectively. The two linear regimes of the avidin?biotin and of
L-selectin?PSGL-1 pairs locate two barriers along the unbinding pathways of
thetwosystems.Thehigh-strengthregimeslocatetheinnerbarriersat1.2and
0.6Å,respectively.Thetwolower-strengthregimesmapouterbarriersatx??
3and4Å,respectively.Extrapolationofeachlinearregimetozeroforceyields
the koff(0) and therefore the time survival, ?(0), of the bond under those
experimental conditions. In biotin?avidin, ?(0) was found to be equal to 0.03
and1.4s(35),andintheL-selectin?PLSG-1case,thevalueswere0.01and0.33s
(41), respectively, for the higher- and lower-strength regimes.
(A) The most probable unbinding forces (pN) (determined by a
Fig. 5.
of barriers along the reaction coordinate (rc) can determine the dynamic
characteristics of adhesion interaction under load. The potential energy
curves have been sketched for the L-selectin?PSGL-1, the avidin?biotin, and
the Fn?Fn-adhesin interactions on the basis of the plots in Fig. 4. The binding
lifetimeextrapolatedtozeroforce,?(0),isreportedforeachpotentialwell.(A)
Rollingadhesiontakesplacewhenthecellvelocitydropsbecauseoftransient
binding events: barriers characterized by short x?can sustain high forces but
small deformations (brittle character), and large x?can sustain lower forces
but larger deformations (elastic or compliant character). In the case of the
L-selectin?PSGL-1 curve, the inner barrier provides the high-strength attach-
ment needed to initiate the leukocyte tethering, but only for very short
loading times [?(0) ? 0.01 s]; the additional interaction ensures longer resi-
dence at vessels [?(0) ? 0.33 s]. (B) Transition from a transient to a firm
adhesionregimecanbeachievedbycarefullytailoringthevaluesofx?and?(0)
(33). The values of x? and ?(0) for the inner barrier of the avidin?biotin
interactionareabouttwiceasmuchasthoseoftheinnerbarrierforL-selectin?
PSGL-1. These parameters are consistent with the function of avidin?biotin
known to mediate firm adhesion under flow. (C) The Fn?Fn-adhesin values of
x?and ?(0) are much larger than those found for the inner barrier of the
biotin?avidin and L-selectin?PSGL-1 and are comparable to those of their
outer barriers. This set of values typifies an elastic interaction selected to
sustain relatively large deformations at low forces and low loading rates, but
for significantly longer times.
A schematic view illustrating how the position (x?) and the number
Bustanji et al.
PNAS ?
November 11, 2003 ?
vol. 100 ?
no. 23 ?
13295
BIOCHEMISTRY
BIOPHYSICS

Page 5

the location of the barrier along the reaction coordinate. Thus,
if the highest barrier is located sufficiently out, it is possible for
an inner, smaller barrier to emerge as the rate-determining one
in a mechanical experiment (Fig. 3). In principle, by varying the
force-loading rate on the molecule, one can use direct mechan-
ical manipulation to make different barriers emerge as the
rate-determining ones, and thus map the dissociation landscape
(Fig. 4).
Consequently, the observation of a single linear regime in the
force spectrum of the Fn?Fn–adhesin pair (Fig. 4) cannot be
interpreted as an unequivocal evidence of a single barrier along
the unbinding pathway. If this were the case, both the thermal
and the mechanically induced dissociation process would be
controlled by this barrier, and the time survival of the bond
extrapolated to zero force ?(0) obtained by force spectroscopy
experiments would coincide with the bond lifetime, (?), mea-
sured for thermally activated dissociation. In this case, the value
of?(0)?0.21sobtainedbyDSFforFnissixordersofmagnitude
smaller than that estimated by SPR (26). The uncertainties in the
loading-rate evaluation in the DSF experiments, and rebinding
and avidity effects that might have affected the SPR ?(0) value
(as suggested by one referee), cannot account for such an
enormous difference. This difference proves that the energy
landscape of Fn?Fn–adhesin interaction is controlled by more
than one potential barrier and that the regime identified by our
measurement corresponds to an inner barrier (Fig. 5).
Barriers characterized by short x?can sustain high forces but
small deformations and thus display a ‘‘brittle’’character; con-
versely, barriers possessing large x?are ‘‘elastic’’ or compliant,
being capable of sustaining low forces but larger deformations
(36). The small distance to the first barrier from the equilibrium
bound position observed in rolling-adhesion mediated by L-
selectin?PSGL-1 (x?? 0.6 Å) results in a binding mode that can
sustain high stress, but only for very short loading times (?(0) ?
0.01 s). Such a binding mode can provide the high strength
attachment needed to initiate cell tethering and to interrupt cell
translocation in flow, but it requires additional interactions to
provide longer residence of leukocytes at vessels after being first
arrested by the inner barrier. These additional binding interac-
tionsareprovidedbyanouterbarrier(x??4Å),which,although
capable of sustaining smaller forces (typically ?65 pN), have
significantlylargersurvivaltimes(?(0)?0.33s).Firmadhesions,
such as those mediated by the biotin?streptavidin interaction, in
contrast, are characterized by an inner barrier whose values for
x?and ?(0) are about twice as much as those of the inner barrier
for L-selectin?PSGL-1 (Fig. 5). These parameters are consistent
with the function of avidin-biotin known to mediate firm adhe-
sion under flow (33).
In contrast, the Fn?Fn–adhesin values of x?and ?(0) reported
here are much larger than those found for the inner barrier of
the biotin–avidin and L-selectin–PSGL-1 and comparable to
those of their outer barriers (Fig. 5). The larger value of x?
determined for the Fn?Fn–adhesin thus typifies an elastic in-
teraction selected to sustain relatively large deformations at low
forces and low loading rates, but for significantly longer times.
This observation rationalizes the strong liability of bacterial
adhesion to high, turbulent flow.
TheAdhesionProbabilityandtheKineticRateConstants.Simulations
of adhesion dynamics by Chang et al. (33) have shown that for
fixed koff(0) and x?, increasing the association rate, kon, decreases
the velocity of cell rolling so that a rolling adhesion can become
afirmadhesion.Whatistheroleplayedbythekineticsofbinding
in Fn?Fn–adhesin interactions and how do these compare with
the binding kinetics of the L-selectin?PSGL-1 pair? Binding
kinetics in single-molecule studies can be characterized in terms
of adhesion probability (P), that is, the ratio between the number
of force curves showing a specific adhesion event (Fig. 1 A and
C) to the total number of curves. In these experiments, the
Fn-functionalized tip is kept in contact with the surface of the
bacteriafordifferenttimesbeforeitispulledawaywithavelocity
of 1 ?m?s. More than 4,000 force curves were recorded in this
way.AsshowninFig.6,theadhesionprobabilityvanishedatzero
contact time and increased exponentially with increasing contact
time up to 200 ms. Beyond this threshold, the value of the
adhesion probability reached a plateau. The adhesive interaction
of an SFM tip to a surface due to a specific binding reaction can
be described by:
dP?dt ? ?kruptureP ? kbind(1 ? P)
[3]
where kruptureand kbindare the rupture and binding rates of the
interaction (37). In particular, kbindis assumed to be a pseudo
first-orderrateconstant.BeingatzerocontacttimeP(0)?0,this
equation can be integrated to give:
P?t? ? kbind(1 ? e??kbind? krupture]t)?(kbind? krupture).
[4]
Eq. 4 accounts for the behavior of the adhesion probability on
the contact time observed experimentally for Fn (see Fig. 6). At
contact times shorter than 200 ms, the adhesion probability is
determined by the on-rate constant, kbind, alone. When the
contact times are increased up to values of the same order of
magnitude of the lifetime of the interaction in the experimental
conditions (?rupture ? 1?krupture), the complex starts breaking
spontaneously before being stretched significantly, leading to a
decrease of observable adhesion events, and in this second
regime, as predicted by Eq. 4, a plateau is reached. Using Eq. 4,
the rates kruptureand kbindcan thus be determined by a least-
square fit of the experimental data of adhesion probability vs.
contact time reported in Fig. 6. This analysis yields krupture?
1.38 ? 0.34 s?1and a kbind? 11.76 ? 0.79 s?1.
Whereas the kbindcan be converted into an on-rate only if we
know the effective concentration of binding partners in the
volume of the interaction between the tip and the bacterium
surface, krupturedirectly corresponds to a koff(37). The difference
between the value of koff(0) ? 4.7 s?1determined by the binding
analysis and estimated from the plot in Fig. 4 with the krupture
determined from Fig. 6 is likely due to the fact that the latter is
really an apparent rate constant. Indeed, whereas in the first set
of experiments the data were selected among those pulling
curves in which a rupture event could be unequivocally dis-
cerned, in the second set the data included both those curves
Fig. 6.
(t) in the case of the binding between Fn and S. epidermidis (curve A), and in
that of P-selectin and PSGL-1 (curve B, reproduced from ref. 37). Eq. 4 was
fitted (A, solid curve) to the experimental points. Both the approach and the
retractionwereperformedat1?m?svelocity.Foreachcontacttimethreesets
of experiments were carried out with at least 400 approach-retraction cycles
each. The average values and the SD are reported in the plot.
The dependence of the adhesion probability (P) on the contact time
13296 ?
www.pnas.org?cgi?doi?10.1073?pnas.1735343100Bustanji et al.

Page 6

which displayed and those curves that did not display ruptures.
In some cases, when a rupture was not seen, we concluded that
there had been no binding. In reality, a binding event might have
takenplace,butthecorrespondingruptureeventoccurredbelow
the force resolution of the SFM (20 pN), leading to an under-
estimation of krupture.
Forcomparison,theassociationkineticsofP-selectin?PSGL-1
reported by Anselmetti and colleagues (37) is reproduced in Fig.
6. It can be seen that these data have a mirror image relationship
with that obtained here for Fn?Fn–adhesin. In the former case,
the binding probability starts from 100% at zero contact time,
then exponentially decreases until a plateau at ?20% is reached.
This behavior indicates that the binding is very fast. As suggested
byAnselmettiandcolleagues(37),thishigh-adhesionprobability
in short contact time, could be a general feature of selectin–
ligand interactions, and the underlying molecular property that
makes the leukocyte tethering process more effective with
increased shear flow. On the contrary, Fn?Fn–adhesins binding
probability starts from 0% at zero contact time and increases
exponentially until it reaches a plateau. This result explains why
and how the bacterial infections mediated by Fn are modulated
by the blood velocity as observed at the macroscopic scale by
ReddyandRoss(6).Theseauthorsfoundthatwhentheflowrate
increases, the Fn?Fn–adhesin binding is greatly decreased. At
high shear flow, i.e., when the contact time is very short, the
elastic binding mode of Fn?Fn–adhesins resulting from the inner
barrier described here is not efficient and fast enough to capture
bacterial cells, contrary to the leukocyte rolling adhesion case.
We thank G. Zuccheri (University of Bologna, Bologna, Italy) for his
crucial and helpful advice and F. Grandi (University of Bologna) for
technical assistance. This work was supported by Italian Ministry of
Health Grant SVE 225?2001; Programmi Biotechnologie legge 95?95
Ministero dell’Universita ´ e della Ricerca Scientifica e Tecnologica 5%;
Ministero dell’Universita ´ e della Ricerca Scientifica e Tecnologica
Progetti di Ricerca di Interesse Nazionale 1999 and 2001; Progetti
Pluriennali (2001) Universita ` di Bologna; and Progetto Fondo Integra-
tivo Speciale per la Ricerca (Ministero Tesoro, 2003).
1. Gumbiner, B. M. (1996) Cell 84, 345–357.
2. Ruoslahti, E. (1988) Annu. Rev. Biochem. 57, 375–413.
3. Sinha, B., Francois, P. P., Nusse, O., Foti, M., Hartford, O. M., Vaudaux, P.,
Foster, T. J., Lew, D. P., Herrmann, M. & Krause, K. H. (1999) Cell Microbiol.
1, 101–117.
4. Frere, J., Dubus, A. & Fonze, E. (1999) Nat. Biotechnol. 17, BV17–BV18.
5. Goldsmith, H. L. & Turitto, V. T. (1986) Thromb. Haemostasis 55, 415–435.
6. Reddy, K. & Ross, J. M. (2001) Infect. Immun. 69, 3472–3475.
7. Montanaro, L., Arciola, C. R., Borsetti, E., Collamati, S. & Baldassarri, L.
(1999) New Microbiol. 22, 331–336.
8. Projan, S. J. & Novick, R. P. (1997) in The Staphylococci in Human Diseases,
eds. Crossley, K. B. & Archer, G. L. (Churchill Livingstone, New York), pp.
55–82.
9. Mohamed, N., Visai, L., Speziale, P. & Ross, J. M. (2000) Microb. Pathog. 29,
357–361.
10. Proctor, R. A. Mosher, D. F. & Olbrantz, P. J. (1982) J. Biol. Chem. 257,
14788–14794.
11. Vann, J. M., Hamill, R. J., Albrecht, R. M., Mosher, D. F. & Proctor, R. A.
(1989) J. Infect. Dis. 160, 538–542.
12. Florin, E. L., Rief, M., Lehmann, H., Ludwig, C., Dornmair, V. T., Moy, V. T.
& Gaub, H. E. (1995) Biosens. Bioelectron. 10, 895–901.
13. Hutter, J. L. & Beckhhoefer, J. (1993) Rev. Sci. Instrum. 64, 1868–1873.
14. Potts, J. R. & Campbell, I. D. (1996) Matrix Biol. 15, 313–320.
15. Pavey, K. D., Barnes, L. M., Hanlon, G. W., Olliff, C. J., Ali, Z. & Paul, F.
(2001) Lett. Appl. Microbiol. 33, 344–348.
16. Razatos, A., Ong, Y. L., Sharma, M. M. & Georgiou, G. (1998) Proc. Natl.
Acad. Sci. USA 95, 11059–11064.
17. Chen, S. & Springer, T. A. (2001) Proc. Natl. Acad. Sci. USA 98, 950–955.
18. Conti, M., Bustanji, Y., Falini, G., Ferruti, P., Stefoni, S. & Samori, B. (2001)
ChemPhysChem 2, 610–613.
19. Hinterdorfer, P., Schilcher, K., Baumgartner, W., Gruber, H. J. & Schindler,
H. (1998) Nanobiology 4, 177–188.
20. Baumgartner, W., Hinterdorfer, P. & Schindler, H. (2000) Ultramicroscopy 82,
85–95.
21. Rief, M., Gautel, M., Schemmel, A. & Gaub, H. E. (1998) Biophys. J. 75,
3008–3014.
22. Oberhauser, A. F., Badilla-Fernandez, C., Carrion-Vazquez, M. & Fernandez,
J. M. (2002) J. Mol. Biol. 319, 433–447.
23. Grandbois, M., Beyer, M., Rief, M., Clausen-Schaumann, H. & Gaub, H. E.
(1999) Science 283, 1727–1730.
24. Alon, R., Chen, S., Fuhlbrigge, R., Puri, K. D. & Springer, T. A. (1998) Proc.
Natl. Acad. Sci. USA 95, 11631–11636.
25. Flock, J. I., Froman, G., Jonsson, K., Guss, B., Signas, C., Nilsson, B., Raucci,
G., Hook, M., Wadstrom, T. & Lindberg, M. (1987) EMBO J. 6, 2351–2357.
26. Holmes, S., May, K., Johansson, V., Markey, F. & Critchley, I. (1997) J.
Microbiol. Methods 28, 77–84.
27. Bell, G. I. (1978) Science 200, 618–627.
28. Evans, E. & Ritchie, K. (1997) Biophys. J. 72, 1541–1555.
29. Schwesinger, F., Ros, R., Strunz, T., Anselmetti, D., Guntherodt, H. J.,
Honegger, A., Jermutus, L., Tiefenauer, L. & Pluckthun, A. (2000) Proc. Natl.
Acad. Sci. USA. 97, 9972–9977.
30. Dettmann, W., Grandbois, M., Andre, S., Benoit, M., Wehle, A. K., Kaltner,
H., Gabius, H. J. & Gaub, H. E. (2000) Arch. Biochem. Biophys. 383, 157–170.
31. Izrailev, S., Stepaniants, S., Balsera, M., Oono, Y. & Schulten, K. (1997)
Biophys. J. 72, 1568–1581.
32. Evans, E. & Ritchie, K. (1999) Biophys. J. 76, 2439–2447.
33. Chang, K. C., Tees, D. F. & Hammer, D. A. (2000) Proc. Natl. Acad. Sci. USA
97, 11262–11267.
34. Evans, E., Leung, A., Hammer, D. & Simon, S. (2001) Proc. Natl. Acad. Sci.
USA 98, 3784–3789.
35. Merkel, R., Nassoy, P., Leung, A., Ritchie, K. & Evans, E. (1999) Nature 397,
50–53.
36. Liphardt, J., Onoa, B., Smith, S. B., Tinoco, I. J. & Bustamante, C. (2001)
Science 292, 733–737.
37. Fritz, J., Katopodis, A. G., Kolbinger, F. & Anselmetti, D. (1998) Proc. Natl.
Acad. Sci. USA 95, 12283–12288.
Bustanji et al.
PNAS ?
November 11, 2003 ?
vol. 100 ?
no. 23 ?
13297
BIOCHEMISTRY
BIOPHYSICS