The bacterial fimbrial tip acts as a mechanical force sensor.

Abstract

There is increasing evidence that the catch bond mechanism, where binding becomes stronger under tensile force, is a common property among non-covalent interactions between biological molecules that are exposed to mechanical force in vivo. Here, by using the multi-protein tip complex of the mannose-binding type 1 fimbriae of Escherichia coli, we show how the entire quaternary structure of the adhesive organella is adapted to facilitate binding under mechanically dynamic conditions induced by flow. The fimbrial tip mediates shear-dependent adhesion of bacteria to uroepithelial cells and demonstrates force-enhanced interaction with mannose in single molecule force spectroscopy experiments. The mannose-binding, lectin domain of the apex-positioned adhesive protein FimH is docked to the anchoring pilin domain in a distinct hooked manner. The hooked conformation is highly stable in molecular dynamics simulations under no force conditions but permits an easy separation of the domains upon application of an external tensile force, allowing the lectin domain to switch from a low- to a high-affinity state. The conformation between the FimH pilin domain and the following FimG subunit of the tip is open and stable even when tensile force is applied, providing an extended lever arm for the hook unhinging under shear. Finally, the conformation between FimG and FimF subunits is highly flexible even in the absence of tensile force, conferring to the FimH adhesin an exploratory function and high binding rates. The fimbrial tip of type 1 Escherichia coli is optimized to have a dual functionality: flexible exploration and force sensing. Comparison to other structures suggests that this property is common in unrelated bacterial and eukaryotic adhesive complexes that must function in dynamic conditions.

Full text

Page 1

The Bacterial Fimbrial Tip Acts as a Mechanical Force
Sensor
Pavel Aprikian1., Gianluca Interlandi2., Brian A. Kidd2, Isolde Le Trong3, Veronika Tchesnokova1, Olga
Yakovenko2, Matt J. Whitfield2, Esther Bullitt4, Ronald E. Stenkamp3, Wendy E. Thomas2*, Evgeni V.
Sokurenko1*
1Department of Microbiology, University of Washington, Seattle, Washington, United States of America, 2Department of Bioengineering, University of Washington,
Seattle, Washington, United States of America, 3Departments of Biological Structure and Biochemistry, University of Washington, Seattle, Washington, United States of
America, 4Department of Physiology & Biophysics, Boston University School of Medicine, Boston, Massachusetts, United States of America
Abstract
There is increasing evidence that the catch bond mechanism, where binding becomes stronger under tensile force, is a
common property among non-covalent interactions between biological molecules that are exposed to mechanical force in
vivo. Here, by using the multi-protein tip complex of the mannose-binding type 1 fimbriae of Escherichia coli, we show how
the entire quaternary structure of the adhesive organella is adapted to facilitate binding under mechanically dynamic
conditions induced by flow. The fimbrial tip mediates shear-dependent adhesion of bacteria to uroepithelial cells and
demonstrates force-enhanced interaction with mannose in single molecule force spectroscopy experiments. The mannose-
binding, lectin domain of the apex-positioned adhesive protein FimH is docked to the anchoring pilin domain in a distinct
hooked manner. The hooked conformation is highly stable in molecular dynamics simulations under no force conditions but
permits an easy separation of the domains upon application of an external tensile force, allowing the lectin domain to
switch from a low- to a high-affinity state. The conformation between the FimH pilin domain and the following FimG
subunit of the tip is open and stable even when tensile force is applied, providing an extended lever arm for the hook
unhinging under shear. Finally, the conformation between FimG and FimF subunits is highly flexible even in the absence of
tensile force, conferring to the FimH adhesin an exploratory function and high binding rates. The fimbrial tip of type 1
Escherichia coli is optimized to have a dual functionality: flexible exploration and force sensing. Comparison to other
structures suggests that this property is common in unrelated bacterial and eukaryotic adhesive complexes that must
function in dynamic conditions.
Citation: Aprikian P, Interlandi G, Kidd BA, Le Trong I, Tchesnokova V, et al. (2011) The Bacterial Fimbrial Tip Acts as a Mechanical Force Sensor. PLoS Biol 9(5):
e1000617. doi:10.1371/journal.pbio.1000617
Academic Editor: Katrina T. Forest, University of Wisconsin-Madison, United States of America
Received August 17, 2010; Accepted March 30, 2011; Published May 10, 2011
Copyright: ? 2011 Aprikian et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was financially supported by a postdoctoral fellowship from the American Heart Association number 0820107Z to GI, a NIH R01 number
1R01 AI50940 to EVS and WT, and a T32 training grant GM008268 to BK. The funders had no role in study design, data collection and analysis, decision to publish,
or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
Abbreviations: AFM, atomic force microscope; BSA, bovine-serum-albumin; Ld, lectin domain; MD, molecular dynamics; Pd, pilin domain; RMSD, root mean
square deviation; SASA, solvent accessible surface area.
* E-mail: evs@u.washington.edu (EVS); wendyt@u.washington.edu (WET)
. These authors contributed equally to this work.
Introduction
Most adhesive biological processes are exposed to mechanical
stress resulting from fluid flow-induced shear. Thus, the molecular
structures that mediate adhesive interactions are adapted to
function in mechanically dynamic conditions. In the case of gram-
negative bacterial cells, the interaction with the host tissue is
known to be mediated by adhesive proteins (adhesins) that are, in
many cases, positioned at the tip of multimeric hair-like
appendages called fimbriae (or pili) and bind to receptor molecules
on the target cells or tissues [1,2]. The 30 kDa FimH protein is the
most common, mannose-specific adhesin of Escherichia coli located
on the tip of type 1 fimbriae [3,4].
Bacterial adhesion mediated by type 1 fimbriae is enhanced by
shear stress [5,6], and single molecule force spectroscopy
experiments have shown that a tensile force extends the lifetime
of the bond between FimH and the mannose receptor [7]. The
force-enhanced, so-called catch bond mechanism of FimH
binding involves allosteric activation of the mannose-binding
lectin domain (Ld), which switches from a low- to a high-affinity
conformation upon separation from the anchoring pilin domain
(Pd).
The type 1 fimbria consists of a 1–2 mm long fimbrial rod,
which is built by thousands of copies of the non-adhesive major
subunit FimA, and the fimbrial tip, which comprises three minor
subunits, i.e., FimF, FimG, and the FimH adhesion [2,3,8].
Several crystallographic and nuclear magnetic resonance studies
have investigated the structure of the monomeric or dimerized
minor subunits [9–14]. Most of these studies were performed with
FimH, either with isolated Ld [14] or with the entire protein in
complex with the chaperone FimC wedged between Pd and Ld
[9,11]. Recently, the X-ray structure of a native form of FimH
PLoS Biology | www.plosbiology.org1May 2011 | Volume 9 | Issue 5 | e1000617

Page 2

was obtained, where the FimH adhesin is incorporated into the
fimbrial tip complex comprised also of one FimG and two FimF
subunits, with FimC wedged between the two FimF copies
(Figure 1a) [8]. Importantly, when comparing the tertiary
structure of Ld in the tip-incorporated FimH with that in the
isolated Ld or FimH/FimC complex, some remarkable confor-
mational differences were observed [8]. In the tip complex, Pd is
docked onto Ld causing compression of the b sandwich fold of Ld
by twisting two b sheets relatively to one another. The mannose-
binding pocket is located on the opposite side of the binding
domain relative to the Pd/Ld inter-domain region (Figure 1b),
but the twisting in the inter-domain region leads to opening of the
mannose pocket because the rigidity of the b sheet transmits
structural perturbations over long distances [8]. In contrast, when
Pd is separated, Ld assumes an elongated, less twisted
conformation with a tight conformation of the mannose-binding
pocket that has a more than 200-fold higher affinity to mannose
than the low-affinity, compressed conformation of Ld [5]. It has
been suggested that FimH Ld functions like a molecular finger-
trap that switches from a low- to a high-affinity conformation
upon separation from Pd, which is caused by tensile mechanical
force originated by shear stress [8]. Such force-induced activation
of the FimH adhesin is the basis of the shear-enhanced catch
bond mechanism of the type 1 fimbriae-mediated bacterial
adhesion under flow [5].
While the tertiary structure of Ld has been the primary focus of
a recent study [8], the dynamic mechanical properties of the entire
tip complex have not been studied and it is not clear how the
quaternary structure of the tip is adapted to function under shear
conditions. Using a combination of flow chamber experiments,
single molecule force spectroscopy, and molecular dynamics (MD)
simulations, we investigated here how the structural properties of
all fimbrial tip components are optimized to facilitate the initial
interaction of FimH with the surface receptor, its switch to the
activated form, and then its sustained binding under dynamic flow
conditions. In summary, the fimbrial tip acts as a mechanical force
sensor.
Results
Functional Properties of the Fimbrial Tip
The protein FimH is generally highly conserved in different E.
coli strains, though naturally occurring point mutations are known
to make binding less dependent on shear by increasing the
adhesion under low or no flow conditions [15]. While such mutant
variants are found among uropathogenic strains, the bulk of FimH
variants among all E. coli pathotypes, including uropathogenic
ones, clearly demonstrate shear-dependent binding. In particular,
the FimH variant crystallized in the fimbrial tip represents the
most common protein variant found in E. coli causing extra-
intestinal infections and belonging to the so-called B2 clonal
group.
The shear-dependent properties of the tip-incorporated FimH
have been demonstrated previously using either yeast mannan,
mannose coupled to bovine-serum-albumin (BSA), or guinea pig
red blood cells [5,7,16,17], all of which are surrogate receptors for
the type 1 fimbriae [15]. We tested whether the type 1 fimbriae
mediate shear-dependent adhesion to a natural target like bladder
epithelial cells. Bacteria expressing type 1 fimbriae, with FimH,
FimG, and FimF structurally identical to the ones in the
crystallized tip complex, were used in parallel plate flow chamber
experiments over the monolayer of bladder cell line T24. Bacterial
adhesion to the cells increased more than 20-fold when shear was
switched from 0.01 Pa to 0.1 Pa (Figure 2a,b). The pattern of E.
coli adhesion to uroepthelial cells under different shears was similar
to the bacterial binding to mannose-BSA coated on a surface
(Figure 2b), indicating the monomannose specific mechanism of
the shear-dependent E. coli adhesion to the bladder cells.
Moreover, purified fimbrial tips that were used for the X-ray
studies, when coupled to plastic beads, also mediated shear-
enhanced binding to a mannose-BSA coated surface (Figure 2c).
We then used single molecule force spectroscopy to establish
how the bond between the fimbrial tips and mannose responds to
various amounts of mechanical force. Previous single molecule
force spectroscopy experiments with fimbrial tips never resulted in
a measure of the dissociation rate as a function of force. This is
because the bond lifetime was too long (many minutes) to be
measured one molecule at a time in constant force experiments [8]
and because the alternative approach, a constant loading rate, had
not been analyzed in a way that calculated the dissociation rate
and was also performed on different variants of FimH [7]. Here, a
new method was used to analyze constant loading rate
experiments to estimate dissociation rates at various levels of force
[18]. This method also shows the catch bond behavior in a more
direct, intuitive fashion than either previous method. The force
was increased at a constant loading rate on bonds between fimbrial
tips and mannose-BSA to obtain histograms of rupture forces
(Figure 2d). The instantaneous dissociation (off-) rate was then
estimated from the number of bonds that break relative to the
number remaining for each bin. This method provides a
measurement of the force dependence of the effective off-rate
using a single constant loading rate. A loading rate of 300 pN/s
was previously shown to provide an even distribution between the
low and high force peaks in the histogram [7], and thus was used
here to provide adequate statistics in both regions. This force-
dependent effective dissociation rate (red line in Figure 2d) shows
that the off-rate decreases upon the force increasing between 30
and 80 pN (before beginning to increase above 90 pN). The
existence of a regime in which increased force decreases the
dissociation rate is the modern definition of catch bonds [19,20].
Thus, taken together, these results indicate that the fimbrial tip
complex used for the X-ray analysis [8] exhibits shear-enhanced
Author Summary
Noncovalent biological interactions are commonly sub-
jected to mechanical force, particularly when they are
involved in adhesion or cytoskeletal movements. While
one might expect mechanical force to break these
interactions, some of them form so-called catch bonds
that lock on harder under force, like a nanoscale finger-
trap. In this study, we show that the catch-bond forming
adhesive protein FimH, which is located at the tip of E. coli
fimbriae, allows bacteria to bind to urinary epithelial cells
in a shear-dependent manner; that is, they bind at high
but not at low flow. We show that isolated fimbrial tips,
consisting of elongated protein complexes with FimH at
the apex, reproduce this behavior in vitro. Our molecular
dynamics simulations of the fimbrial tip structure show
that FimH is shaped like a hook that is normally rigid but
opens under force, causing structural changes that lead to
firm anchoring of the bacteria on the surface. In contrast,
the more distal adaptor proteins of the fimbrial tip create a
flexible connection of FimH to the rigid fimbria, enhancing
the ability of the adhesin to move into position and form
bonds with mannose on the surface. We suggest that the
entire tip complex forms a hook-chain, ideal for rapid and
stable anchoring in flow.
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org2May 2011 | Volume 9 | Issue 5 | e1000617

Page 3

binding to uroepithelial cells under flow conditions and catch-
bond behavior under tensile mechanical force.
Quaternary Structure of the Fimbrial Tip
To predict how the components of the tip complex might
behave under dynamic force conditions, we first evaluated the
overall quaternary structure of the tip based on the X-ray data. In
the fimbrial tip, Pd is connected to FimG, FimG to FimF, and
FimF to another copy of FimF via a donor-strand complemen-
tation mechanism [21], where a missing b strand in each b
sandwich shaped subunit is complemented by an N-terminal
strand of the following subunit. The linkage via the complement-
ing strand mechanism is among the strongest of non-covalent
bonds and, thus, to ease the presentation we will refer here to the
FimG and FimF subunits (with the complementing strands) as
domains, similarly to the Ld and Pd in FimH that are covalently
linked to each other via a linker chain, and Pd includes also the
FimG donor strand (see also Figure 1a and ‘‘Materials and
Methods’’).
A striking feature of the quaternary conformation of the fimbrial
tip crystal structure is the end-to-end position of all domains
relative to one another that results in an extended structure of the
tip complex, with a total length of approximately 230 A˚
(Figure 1a). However, the extent of quaternary interaction in the
various inter-domain interfaces is drastically different.
While the wedged FimC chaperone made it difficult to evaluate
the FimF-FimF interface, the buried surface between FimF and
FimG (501 A˚2) was smaller than between FimG and Pd (760 A˚2)
and even smaller than the interface buried between Ld and Pd
(1,040 A˚2; see ‘‘Materials and Methods’’ for a description of how
the surface buried between domains was calculated). There were
only three side chain interactions and no hydrogen bonds or salt
bridges between the FimF and FimG domains (Figure 3c), while
between FimG and Pd there were six side chain contacts, one
hydrogen bond, and one salt bridge (Figure 3b). The most
extensive interactions were between Pd and Ld, with 14 side chain
contacts and three hydrogen bonds, with one of them involving
backbone atoms, i.e., Cys161 NH … O Ser114, and two involving
the side chain of Arg166 and the carboxyl oxygen of Ala115
(Figure 3a and Table 1).
Besides the largest buried inter-domain interface, there was
another notable feature of the Ld-Pd quaternary conformation,
namely it had a ‘‘hooked’’ conformation. The domains Ld and Pd
were hinged at an angle of 128u (Figure 1b). In contrast, the angle
between the other extensively interacting domains, Pd and FimG,
was almost completely open at an angle of ca. 169u (Figure 1a and
Figure S1a). In summary, the proximal (i.e., closer to the fimbrial
rod) part of the tip has significantly fewer interdomain contacts
relatively to the distal portion of the tip, which is also characterized
by a distinctly hooked conformation of Ld and Pd.
Tip Flexibility in the Absence of Tensile Force
In order to test the flexibility of the different interdomain
interfaces, MD simulations were performed with three pairs of
neighboring fimbrial tip domains: Ld-Pd, Pd-FimG, and FimG-
FimF (Table 2). The quaternary conformation of the complexes
Ld-Pd and Pd-FimG was mostly stable in the simulations, with the
Ca root mean square deviation (RMSD) from the initial
conformation remaining below 3.5 A˚
(Figure 4a and Figure S2a). In contrast, the Ca RMSD for the
FimG-FimF complex exceeded 5.5 A˚
simulation (Figure 4a and Figure S2a), indicating that the
for both complexes
in the course of the
Figure 1. Crystallographic structures. (a) Entire fimbrial tip (stereoview). Subunits are distinguished using different colors. A subunit consists of a
polypeptide chain, whereas a domain is defined as the globular part of a subunit and (with the exception of the FimH lectin domain) a donated
strand of the neighboring subunit (see ‘‘Materials and Methods’’ for the exact definition of start and end residues of domains). The FimH subunit
consists of two domains, which are also distinguished with different colors: cyan for the lectin and purple for the pilin domain, respectively, except for
the donated strand of FimG, which is colored orange as the subunit it belongs to. The two FimF subunits are colored magenta and green,
respectively. The chaperone protein FimC is in yellow. Residues that are located between domains define linker chains and their backbone is colored
in silver. The arrows in yellow represent the principal axes (PA1, PA2, and PA3) of each domain (see ‘‘Materials and Methods’’ for the determination of
principal axes). They are used to calculate hinge and twist angles between adjacent domains (Figures S3, S4, S5d,e). (b) FimH subunit with donated
FimG strand (stereoview). The principal axes of the FimH subunit are labeled in brown.
doi:10.1371/journal.pbio.1000617.g001
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org3 May 2011 | Volume 9 | Issue 5 | e1000617

Page 4

quaternary structure of these domains is relatively flexible. The
observed flexibility is mainly due to rigid body movements of the
domains relative to each other, since the Ca RMSD for all
domains individually was mostly below 2 A˚
Similarly, the distance between centers of mass of neighboring
domains and the amount of buried surface area in the inter-
domain interface remained constant for the Ld-Pd and Pd-FimG
(Figure S2b).
complexes but fluctuated notably in the FimG-FimF complex
(Figure 4b,c and Figure S2c,d).
In order to further describe the movement of the domains
relative to each other, the angles between the principal axes of the
domains were monitored. The angle between the first principal
axes (PA1 in Figure 1b) subtracted from 180u is called the ‘‘hinge
angle,’’ whereas the angle between the second principal axes (PA2
Figure 2. Shear-enhanced adhesion and catch bond behavior. (a) Binding of E. coli to uroepithelial cells at low (0.01 Pa) and high (0.1 Pa)
shear stress in a flow chamber. (b) Level of E. coli binding under low (0.01 Pa) and high (0.1 Pa) shear stress to uroepithelial cells and mannose-BSA
coated surface. (c) Binding of fimbrial tip-coated beads to mannose-BSA coated surface. (d) Binding of fimbrial tips to mannose-BSA in single
molecule force spectroscopy experiments. The histograms in black (ordinate on the left, abscissa at the bottom) show the fraction of total pulls
rupturing within a bin of a force range. The red line (ordinate on the right, abscissa at the top) displays the calculated unbinding rate (k_off) as a
function of the force.
doi:10.1371/journal.pbio.1000617.g002
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org4 May 2011 | Volume 9 | Issue 5 | e1000617

Page 5

in Figure 1b) is labeled ‘‘twist angle.’’ (The angle between the third
principal axes, PA3 in Figure 1b, was observed to essentially
correlate with the twist angle and thus was not monitored.) The
FimH domains remained in the hooked conformation throughout
all runs and the hinge and twist angles measured 133u64u and
41u65u, respectively. The Pd-FimG complex also retained its
conformation, with a similar amount of variation in the hinge
angle (168u65u) but a slightly more pronounced variability in the
twist angle (2109u610u; the negative values are due to the
location of the donor strand complementation grooves in Pd and
FimG on opposite sides). The highest variability was between
FimG and FimF, where the hinge angle measured 252u611u and
the twist angle was 104u and fluctuated with a standard deviation
of 16u (Figure 4d,e and Figure S2e,f), supporting the hypothesis
that the FimG-FimF interface is very flexible and explores a
relatively large space compared to the other domains.
The differences in flexibility between the three complexes, with
Ld-Pd being the most rigid and FimG-FimF the most flexible, are
Figure 3. Inter-domain side chain and electrostatic contacts in MD simulations. (a, b, and c) Inter-domain interfaces in the X-ray structure.
(d, e, and f) Conformation after 14 ns in a 300 K simulation with two neighboring domains (the conformation in (d) is taken from the 40 ns run). (g, h,
and i) Conformation after 10 ns in a pulling simulation. Side chains involved in contacts or salt bridges or hydrogen bonds are displayed in the stick
and ball representation (see ‘‘Materials and Methods’’ for the definition of side chain and electrostatic contacts). Backbone atoms forming hydrogen
bonds are also represented. In the central and right column, those contacts are displayed that are observed to be persistent in the 300 K simulations
(i.e., they occur in at least 66% of the simulation frames, see ‘‘Materials and Methods’’). Side chains involved in persistent contacts are labeled only in
the central column. The carbon atoms of the displayed side chains are colored differently depending upon which domain they belong to: light grey
for the lectin domain, dark grey for the pilin domain, black for FimG, and tan for FimF. The domains are colored as in Figure 1. The figure was
prepared with VMD [60].
doi:10.1371/journal.pbio.1000617.g003
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org5 May 2011 | Volume 9 | Issue 5 | e1000617

Page 6

consistent with the number of native inter-domain interactions
(Figure 3d–f, Table 1, and Figure S2g). The interface between Ld
and Pd (Figure 3d) presents the largest number of native side chain
contacts and hydrogen bonds (Table 1), the latter being buried
throughout the MD simulations.
Conformational Changes under Tensile Force
In order to investigate the conformational changes induced by a
tensile force onto the fimbrial tip, pulling simulations were
performed with the complex containing Ld, Pd, FimG, and the
first copy of FimF (Figure 1). A constant force of 200 pN was
applied for 10 ns in three separate simulations between mannose-
binding site residues on the apex of Ld and the C-terminus of the
donor strand connecting the two FimF subunits (Table 2).
Most conformational changes were observed in the first 2 ns
during the pulling runs (Figures S3, S4, S5) with the tip extending
in total 37 A˚ (22% of the native structure). The most stable
interdomain interface during the pull was between Pd and FimG,
which remained almost unchanged, keeping a similar geometry as
the native state (Figure 4 and Figures S3, S4, S5). In contrast, the
quaternary structure of FimG-FimF underwent substantial con-
formational changes, with an increase of the Ca RMSD from the
native conformation (Figure 4a), an increase in the distance
between the centers of mass of the domains (Figure 4b), and a
virtual elimination of the buried surface area (Figure 4c) and native
side chain contacts (Figure 4f and Figures S3, S4, S5). Also, after
the straightening under tensile force, the quaternary structure of
the FimG-FimF complex was observed to be much more flexible
than in the absence of force, as indicated by the large fluctuations
in the twist angle (Figure 4e).
However, the most drastic changes occurred in the Ld-Pd
structure, where the inter-domain hook straightened to an almost
flat angle (Figure 4d and Figure S1e), with a large increase in the
distance between the centers of mass of the domains, a significant
decrease in buried surface area, and elimination of most native
contacts (Figure 4 and Figures S3, S4, S5, S6, S7, S8, S9). In
order to improve the statistical sampling of the events occurring
during separation of the lectin from the pilin domain, three
additional pulling simulations were performed with just the FimH
protein. In both sets of pulling runs the rupture of inter-domain
contacts between Ld and Pd happened in a sequential manner
(Figure S10). Importantly, the combined results analysis indicates
that the order of the contact breakage was inversely correlated
with the contact’s distance from the hinge axis, i.e., bonds further
away from the hinge ruptured earlier in the simulations than
contacts located closer to it (Figure 5). The Pearson’s linear
correlation coefficient was 0.78 with a p-value,0.01. This
suggests that Pd and Ld unzip under tensile force, where only
one or a few inter-domain contacts break at a time instead of
most or all of the contacts breaking simultaneously. The fact that
the sequence of breaking events is statistically correlated with the
distance of the contacts from the hinge axis is indicative that the
hooked shape of FimH allows a sequential unzipping of the
stabilizing contacts as the hinge opens.
In conclusion, these simulations demonstrate that mechanical
force can relatively easy and in an unzipping manner open the
hinge angle between the two domains and eliminate the native
interdomain contacts. These native interdomain contacts were
implicated previously in allosterically maintaining the low-affinity
conformation of Ld, so that loss of these contacts should lead to
activation [8]. Allosteric conformational changes are not expected
to be observed in molecular dynamics simulations due to the short
time scale of the simulations (10 ns) relative to that of typical
allosteric changes (microsecond to millisecond), so this cannot be
directly confirmed in the simulations. Nevertheless, these results
strongly suggest that FimH would allosterically switch its
conformation from low to high affinity once the hinge opens.
Table 1. Number of inter-domain contacts.
Interface X-Raya
300 K
Nativeb
300 K
Averagec
Pullingd
Side chain
interactions
Ld-Pd14 9 (9)8.5460.730.9860.14
Pd-FimG6 7 (6)6.1960.8 5.6260.88
FimG-FimF3 2 (2) 1.4460.560.160.3
Hydrogen bonds
Ld-Pd3 3 (3)2.660.56 0.5860.5
Pd-FimG1000
FimG-FimF0000
Salt bridges
Ld-Pd0000
Pd-FimG1000
FimG-FimF0000
See ‘‘Materials and Methods’’ for a definition of side chain contacts, hydrogen
bonds, and salt bridges.
aContacts observed in the crystallographic structure.
bContacts present in at least 66% of the simulation frames of the 300 K runs
(referred to here as native contacts), excluding the first 10 ns, which are
considered equilibration (the number in parentheses refers to the subset of
native contacts observed also in the X-ray structure).
cAverage and standard deviation of the number of native contacts during the
300 K runs, excluding the first 10 ns (see also Figure 4f).
dAverage and standard deviation of the number of native contacts during the
last 2 ns of the pulling simulations (see also Figure 4f).
doi:10.1371/journal.pbio.1000617.t001
Table 2. Simulation systems.
Simulated
Domainsa
Forceb
Duration Description
FimH Ld+FimH Pd
FimH Ld+FimH Pd
FimH Ld+FimH Pd
FimH Pd+FimG
FimG+FimF
40 ns+27 ns+25 ns Three 300 K runsc
27 ns 300 K run
25 ns300 K run
28 ns 300 K run
28 ns 300 K run
FimH Ld to FimF 200 pN 10 nsPull_1d
FimH Ld to FimF 200 pN10 ns Pull_2d
FimH Ld to FimF200 pN10 ns Pull_3d
FimH Ld+FimH Pd
FimH Ld+FimH Pd
FimH Ld+FimH Pd
200 pN10 ns Pull_FimH_1
200 pN 10 nsPull_FimH_2
200 pN10 ns Pull_FimH_3
aDomain structures used to start the simulations; includes also the linker chain
covalently connecting the domains.
bThe simulations where no force was applied were used to study the flexibility
of the domains at room temperature and determine persistent inter-domain
contacts.
cAll three simulations with the FimH domains were started from the X-ray
structure, but different velocities were assigned to the atoms at the beginning
of each run.
dThe pulling runs with the fimbrial tip were started from the conformation
obtained after equilibrating the structure at 300 K during 3 ns.
doi:10.1371/journal.pbio.1000617.t002
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org6May 2011 | Volume 9 | Issue 5 | e1000617

Page 7

Mechanical Force Sensor
PLoS Biology | www.plosbiology.org7 May 2011 | Volume 9 | Issue 5 | e1000617

Page 8

Mechanics of the Fimbrial Hook
Several independent measurements showed previously that
most FimH bonds dissociated within a second if little or no force
was applied across the complex, but most bonds were long-lived
under a sufficient tensile force load [8]. Thus, mannose unbinding
from the low-affinity state is the dominant event at no and low
force, but stable binding to the high-affinity (activated) state
appears to dominate at high force. While the pulling simulations
show that tensile force can open the hinge angle enough to lead to
FimH activation, it remains to be addressed whether opening of
the hinge angle will occur before mannose is pulled out from the
pocket.
According to kinetic rate theory [22], the rate at which a
physical event occurs—in our case, dissociation of mannose or,
alternatively, opening of the hinge angle between the FimH
domains—is exponentially related to the size of the energy barrier
DE of the reaction transition state relative to the original
equilibrium state of the system: k0=A exp [2DE/kBT], where
A is the Arhenius prefactor and kBT is thermal energy. However,
mechanical force can speed up a reaction by pulling the protein
into the transition state if it is elongated relative to the native state.
We define here Dx as the increase in length between the transition
state and the native state, thermally averaged and projected onto
the direction of force. Then, a constant force field contributes an
amount of energy F Dx, to help overcome the transition state
energy barrier, effectively decreasing its size as illustrated in
Figure 6a. This exponentially increases the reaction rate: k(F)=k0
exp [F Dx/kBT] [23]. Thus, the larger the elongation distance, the
greater will be the effect of the same force onto the reaction rate.
To determine how force affects FimH, we thus need to know the
elongation distances for the two transitions in question: mannose-
unbinding versus hinge opening and activation.
Transition state elongation distances can be estimated from MD
simulations [24], which provide structural details explaining force
dependence in force spectroscopy data [25]. To measure
elongation distances from the pulling trajectories, we estimate
the location of the transition state and then measure the total
elongation necessary to reach this state from the native state. The
transition state for mannose to be pulled out of FimH (i.e.,
breaking of the bond) consists of the mannose molecule being
Figure 4. Quantitative data from MD simulations. Bars filled with dots indicate averages of quantities measured during the 300 K simulations,
while bars with angular hatching indicate averages from the last 2 ns of pulling runs. Error bars show standard deviations (SD). To highlight quantities
with large SD, the error bar is thicker if the SD is larger than the average of all SDs of a given quantity. (a) Ca RMSD of pairwise adjacent domains. (b)
Distance between the centers of mass of two adjacent domains. (c) Surface area buried between two adjacent domains. (d) Hinge and (e) twist angle
between two adjacent domains, respectively. (f) Number of native side-chain contacts between two adjacent domains.
doi:10.1371/journal.pbio.1000617.g004
Figure 5. Unhinging pathway of FimH under tensile force in MD simulations. (a) Location of the hinge axis (stereoview) determined with
the program DynDom [61] by comparing the conformation of FimH at the end of the run pull 1 with its native conformation. (b) Sequence of rupture
events of contacts between Ld and Pd versus distance from the hinge axis. A contact was defined as broken at the first time point when it ruptured
and was not seen to reform within 300 ps. The Ca RMSD was calculated at the time of rupture (time averaged over 200 ps, Figures S3a, S4, S5a) in all
six pulling runs and ranked according to a fractional ranking algorithm (where equal values receive the same raking as their respective ordinal
rankings). The average and standard error of the mean of the ranked RMSD values are plotted against the distance of the respective contact from the
hinge axis. The Ca RMSD from the native structure is a better measure of progress than the time of rupture itself, because rupture events are
observed to occur in approximately the same order in every simulation but the time point when they occur varies across the simulations. The
distance of a contact from the hinge axis is calculated as the distance of the geometric center of the involved side chains from the axis (in the case of
hydrogen bonds, the geometric center of the D–H … A atoms is used). Significant inverse linear correlation (Pearson’s r=0.78; p value=0.004) shows
that the ranking number decreases with increasing distance, suggesting that the larger the distance from the hinge axis, the earlier the rupture of a
contact.
doi:10.1371/journal.pbio.1000617.g005
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org8 May 2011 | Volume 9 | Issue 5 | e1000617

Page 9

shifted from a horizontal into a perpendicular position within the
binding pocket, elongating the Ld-mannose complex approxi-
mately 3 A˚[26]. The minimal length change for the alternative
transition state, where the domains unhinge and Ld activates, was
estimated from the pulling simulations above. Screening of the
trajectories revealed that the rupture of inter-domain contacts
involving the side chain of Arg166 (a side chain contact with
Val155 and the hydrogen bonds with the carbonyl oxygen of
Ala115) was the first rupture event observed in all pulling
simulations (Figures S6, S7, S8, S9). Since the protein elongated
essentially monotonically, any later events would involve even
greater elongation distances. For this reason, the Arg166 rupture
event provides a lower limit for the location of the dominant
transition state. The importance of this event is further
strengthened by the fact that mutations of Arg166 cause strong
activation of FimH [24]. The elongation distance associated with
this event was calculated to be 9.161.2 A˚as measured between
the N-terminus of FimH and the C-terminus of FimG (Table S1
and ‘‘Materials and Methods’’ for details on the calculation of this
distance).
It follows that the activation pathway involves a longer
elongation than does the mannose unbinding pathway, since the
lower limit of 9.161.2 A˚of the former is greater than 3 A˚of the
latter. Thus, increased force will favor activation. To provide an
intuitive idea of how important this difference in length is,
Figure 6b depicts the effect of force, given the transition state
elongation distances described above, on the rate constants for the
two transitions. For this estimation, the rates in the absence of
force were taken from previous experimental results to be 6 s21for
mannose unbinding [16] but only 0.00125 s21for opening of the
hinge angle [27]. At an elongation distance of 9.1 A˚, the rate of
hinge opening surpasses the rate of unbinding at a critical force of
58 pN (Figure 6b). If the rate limiting step occurs after the Arg166
separation, the larger elongation distance will mean a steeper slope
in Figure 6b and a lower critical force. Thus, unbinding dominates
below a critical force of 58 pN, while hinge opening and activation
dominates above the critical force.
The critical force determined by the calculations presented here
is consistent with single molecule force experiments (Figure 2d)
where the bond between the fimbrial tip and mannose is observed
to become high affinity at similar force magnitudes. This supports
the model that the hook opening leads to the switch from the low-
to the high-affinity state, significantly slowing the mannose
unbinding rate [8]. The hook shape of FimH thus ensures that
there is a much larger distance, and thus a greater responsiveness
to force, for hinge opening relative to mannose unbinding. This
physical property allows a tensile force to activate FimH rather
than pulling mannose out of the pocket.
Discussion
A long-standing assumption about the microbial adhesive
organellae is that their quaternary conformation should be flexible
to not impede the ability to explore a target surface in order to
quickly find the corresponding receptor molecule and thereby
maximize the effectiveness of the adhesins’ function. For so-called
class 1 fimbriae, where the adhesin is located on the organella’s tip
(like in the type 1 fimbriae or di-galactose-specific P fimbriae), the
quaternary structure of the multi-protein fimbrial tip complex was
also proposed to be highly flexible to optimize the binding rate
[28]. Indeed, high levels of mobility have been observed previously
in NMR studies of FimG-FimF and FimF-FimF dimers in the type
1 fimbrial tip [10]. Furthermore, in the original X-ray structure of
FimH that was obtained in complex with FimC [9], the Ld and Pd
were separated from one another and, because it was then believed
that the separated domains conformation is native, this was also
Figure 6. Effect of force on dissociation kinetics versus unhinging and activation kinetics. (a) Schematic representation of the energy
landscape of FimH bound to mannose in the absence of force (blue) and in the presence of an external tensile force (brown). The energy barrier for
the unhinging conversion from the low- to the high-affinity state is represented as solid lines. The barrier of the mannose unbinding process from the
low-affinity state is shown in the insert as dashed lines. The energy added to the system as a result of the applied tensile force is indicated by a red
line. The energy landscape in the presence of force is a result of the superposition of the energy landscape in the absence of force and the product of
force times elongation. The elongation between the native and transition states is indicated. (b) Representation of the slopes for the rate constants
for unhinging and unbinding. Since the elongation until conversion to the high-affinity state is not known, the slopes are calculated for the lower
limit, 8.5 A˚, and the upper limit, 18 A˚, of the range. The intersection between the slopes for the unhinging rate and the slope for the mannose
unbinding rate are indicated by blue circles. Where the slopes intersect the rate of unbinding and the rate of unhinging is equal.
doi:10.1371/journal.pbio.1000617.g006
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org9 May 2011 | Volume 9 | Issue 5 | e1000617

Page 10

attributed to the need of high mobility of the mannose-binding Ld.
Mobility is likely to be important in the fimbrial tips because the
main shaft of the type 1 fimbriae is rigid [3] and thus would diffuse
slowly within a very restricted exploration space between the
bacteria and the target cell surface. However, considering that
type 1 fimbriae and other bacterial adhesins are able to mediate
shear-enhanced adhesion via the formation of catch bonds [7,29],
the adhesive organellas not only need to be adapted to ensure high
mobility of the adhesin, they must also have an effective
mechanism to permit the adhesins to be activated by tensile force.
As shown in this article, this results in more complex than
previously thought mechanical properties of the quaternary
structure of the adhesion apparatus.
The primary structure of the 300 amino acid-long FimH is 99%
conserved across E. coli strains. The vast majority of the naturally
occurring FimH variants from fecal and pathogenic E. coli
mediates well-manifested shear-dependent adhesion. Though
FimH in uropathogenic E. coli, especially in the strains causing
infection of kidneys, is under positive selection to acquire
mutations that increase binding at static or low-shear conditions,
most of the FimH variants in uropathogenic strains still manifest
shear-dependent phenotype to at least some extent [30–32]. For
example, the FimH variant crystallized in previous studies [9,11],
which was obtained from a model uropathogenic strain J96 and is
a common natural variant, exhibits a distinct shear-dependent
binding [29,32]. The FimH protein crystallized in the fimbrial tip
complex [8] and analyzed here differs from the former variant in
three amino acids (A27V, S70N, N78S) but is an even more
common (and, actually, evolutionary primary) variant among E.
coli that causes extra-intestinal infections in humans, including
urinary tract infections [15]. FimH variants with completely shear-
independent properties are relatively rare and appear to be
selected out very fast outside the urinary tract [31].
Using E. coli expressing the tip-associated FimH variant, we
showed here that bacterial adhesion to the bladder uroepithelial
cells increases 20-fold when shear increases from 0.01 to 0.1 Pa.
Physiological shear stress along the bladder surface is difficult to
evaluate and can result from the flow dynamics of urine as well as
from the bladder stretching/contraction. The flow-derived level of
wall shear stress equals 4Vm/pR3, where V is volumetric flow rate,
m is fluid viscosity, and R is the radius of the tube. In the human
urinary tract, one can estimate that shear stress is quite low in the
ureter (0.001 Pa, based on a 3 mm diameter and 0.01 ml/s flow)
but can reach 0.3 to 0.5 Pa in the urethra in the course of
urination (based on the average urethra diameter of 5–6 mm and
urine flow of 20–30 ml/s). Thus, though the shear stress along the
bladder surface is expected to vary dramatically depending on the
specific compartment or contraction state of the bladder, the shear
stress levels tested here are likely to be within the physiological
range. These levels are also within the range observed in other
compartments. For example, shear along vascular endothelial cells
ranges from 0.1 to 0.2 Pa on the venous side and 1 to 2 Pa (up to
5 Pa) on the arterial side of the circulation [6,7]. Shear stress
generated at the tooth surface by salivary flow is approximately
0.08 Pa [8]. Shear stress in the intestines is estimated at 1 to 2 Pa
(or higher with viscous lumen) [33] due to peristalsis [34].
The force-enhanced, catch bond fimbrial properties are
ultimately based on the ability of the binding domain of FimH,
i.e., Ld, to assume two conformational states, a twisted compressed
form with a low affinity and an elongated form with a high affinity
toward mannose. In turn, these states are intimately linked to
whether Ld is docked to or separated from the anchoring domain
of FimH, i.e., Pd. Thus, the effectiveness of FimH in mediating
bacterial adhesion depends not only on the ability to quickly find
the mannose receptor but also on the property of the two domains
to then separate, allowing the conformational switch that
strengthens binding.
The entire 95 kDa type 1 fimbrial tip complex of E. coli, which
includes three other structural proteins in addition to FimH, was
first tested for the ability to support the shear dependent properties
of FimH. Not only was the purified tip complex able to reproduce
the shear-dependent binding to a mannose-coated surface
demonstrated by fimbriated bacteria, but the force-enhanced
interaction between the tip-associated FimH and mannose could
be shown in the single molecule force spectroscopy experiments in
which force is increased at a constant loading rate with an atomic
force microscope. Thus, the shear-dependent and force-enhanced
binding properties are intrinsic to the fimbrial tip complex
crystallized recently [8].
Currently, it is very difficult to evaluate the quaternary structural
properties of such a large protein complex like the fimbrial tipunder
dynamic conditions by using direct experimental approaches like
NMRorotherformsofproteinspectroscopyanalysis.Thoughsome
predictions were made from the analysis of the static crystallo-
graphicstructure,here weused moleculardynamicssimulationsasa
tool to study in silico the dynamic properties of complex
biomolecular structures in solution. It would have been more
realistic to perform the simulations with mannose in the binding
pocket and apply the force to mannose instead of the 13 mannose-
interacting residues. However, the crystallographic structure of the
fimbrialtipwith FimHinthelow-affinitystate (whereLdand Pdare
interacting with each other) does not contain mannose. Further-
more, a conformational switchof Ld to the high-affinity state (where
the domains are separated) is not expected to occur in the time scale
of the simulations. Thus, applying the force onto mannose would
have had little influence onto the outcome of the simulations, where
the goal was mainly the description of how Ld and Pd separate. The
key findings of the simulations with and without force are
summarized by a cartoon presented in Figure 7a–d.
Before tensile force is applied, Ld and Pd are docked onto each
other forming a hook shaped structure (Figure 7a) stabilized by
inter-domain side chain interactions and hydrogen bonds. In
contrast, the interface between FimG and FimF presents very few
stabilizing contacts and thus is flexible. This interface, and most
likely FimF-FimF as well, acts like a hinge allowing for exploration
and fast on rates. Moreover, upon application of tensile force,
Figure 7. Cartoon illustrating force-induced conformational
changes in fimbrial tips. The mannose substrate is represented by
black arrows attached to a surface, whereas tensile force is indicated by
arrows attached to the second copy of FimF. (a) Fimbrial tip initially
binds to mannose. (b) Tensile force due to flow straightens the flexible
FimF interfaces. (c) Transition state where FimH Ld and Pd are separated
but Ld is still in the low-affinity state. (d) Final conformation under flow
with FimH Ld in the high-affinity state [8].
doi:10.1371/journal.pbio.1000617.g007
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org10 May 2011 | Volume 9 | Issue 5 | e1000617

Page 11

FimG and FimF separate from each other (Figure 7b) while their
rotational freedom is still preserved (Figure 4e). This flexibility
despite the presence of force is likely to reduce the torsional forces
on mannose in the binding pocket, stabilizing the adhesive
interaction under dynamic flow conditions.
Importantly, upon further application of tensile force, most of
the interactions between Ld and Pd also break and the hook
shaped structure of FimH straightens with the domains separating
(Figures 3g and 7c). As discussed above, separation of the domains
leads to a switch of Ld from a low- to a high-affinity state
(Figure 7d) [16,35]. Thus, while the hinge opening is the primary
event in the native fimbrial tip-incorporated FimH that is required
for the catch bond to be formed, it is the ensuing conformational
change within the binding domain itself that is a key event in
strengthening the bond with mannose, observed in single molecule
force spectroscopy here and in previous studies [8,17]. Notably,
despite the fact that, in the absence of force, the Ld-Pd interface is
stabilized by more bonds and is more stable than the interface
between Pd and FimG, Ld separates from Pd when force is
applied, whereas the Pd-FimG interface remains almost unaltered.
Thus, we propose here that the hooked conformation of FimH is
of key significance because it allows for fast sequential unzipping of
the contacts stabilizing the Ld-Pd interface. At the same time, the
rigid open Pd-FimG conformation provides a lever arm that
facilitates the FimH hook opening under force (Figure 7c).
In summary, every interface in the fimbrial tip appears to play a
specific functional role. The FimG-FimF (and likely also FimF-
FimF) interface presents lateral and rotational flexibility to
efficiently explore the target surface and prevent the stress applied
through the fimbrial rod from disrupting the interaction between
FimH and the receptor. The hook shaped structure of FimH acts
like a force sensor that is activated if a force threshold is reached.
Finally, the relatively stiff interface between Pd and FimG extends
the lever arm of the pilin domain. This may explain why so many
different subunits are found in the fimbrial tip, instead of FimH
being directly attached to the major subunit, FimA. The function
of the fimbrial tip can be compared to that of an anchor connected
to a flexible chain. The flexible connection to the chain allows the
relatively stiffer hook to efficiently explore the surrounding tissue
until it engages in bond formation. After the bond is formed, the
mechanical stress onto the bacterial cell is transmitted to the hook,
which acts as a force sensor switching to a high-affinity state.
The importance of mechanical forces in modulating biological
adhesion is highlighted by the structure and function of the
adhesive fimbrial tip described here. In particular, the quaternary
conformation of this adhesive complex is optimized for initializing
the binding (fast on rates), fast switching to the strong bond, and
sustaining the binding under dynamic shear conditions (slow off
rates). It needs to be noted that the rest of the fimbrial rod is also
likely to play an important role in shear conditions. Namely, it has
been shown that the fimbrial rod can uncoil and prevent breakage
of the high-affinity mannose bond when the force load becomes
too high [36]. The rod can recoil when external force drops,
sustaining internal mechanical tension on FimH to keep it in a
high-affinity conformation by preventing re-docking of the two
domains into the hook conformation.
The type 1 fimbrial tip is the only complex of a native fimbrial
tip described at atomic resolution so far. It is possible that the
quaternary conformation of adhesive tips in other fimbriae will
have similar mechanical properties. For example, di-galactose-
specific P fimbriae of E. coli also exhibit shear-enhanced binding
[28] and possess a complex filamentous adhesive tip. The multi-
domain structures of some eukaryotic adhesins have similarities to
the FimH fimbrial tip as well.
It has been shown that many other receptor-like interactions in
eukaryotes are also governed by the catch-bond mechanism.
Examples include the binding mechanism of selectins [20,37,38],
integrins [39], and also the interaction between von Willebrand
Factor and the platelet surface receptor glycoprotein Iba [40,41],
between actin and myosin [42], and between kinetochores and
single microtubules [43]. The distinctive hinge or hook shape
between the adhesive and anchoring domains of the P-/L-selectins
[37,44], integrins [45,46], and myosin [47], as well as the curled
shape of microtubule protofilaments in the weak-binding disas-
sembly mode [48], suggest a similar mechanism by which
mechanical force can apply enough energy to activate these
proteins and complexes [19,46,49]. Indeed, a common character-
istic of protein complexes displaying catch bond behavior is the
presence of multiple domains or subunits. One role for the many
nonadhesive domains in integrins, and the two to nine consensus
repeats in selectins, for example, might be to confer additional
molecular flexibility to enhance binding rates and minimize
torsional forces. The similarities in quaternary structural mechan-
ics between type 1 fimbrial tips in bacteria and these eukaryotic
structures suggest convergent evolution, in which a wide range of
proteins that must mediate adhesion in dynamic conditions have
evolved multiple domains or subunits creating both rigid hooks to
capture energy from mechanical force to initiate conformational
changes and flexible regions to initiate and stabilize adhesion in
dynamic conditions. This raises the question as to whether the
quaternary structures of other multidomain adhesive complexes,
including matrix proteins like fibronectin, blood proteins like
fibrinogen and von Willebrand Factor, cell-cell adhesion proteins
like cadherins, and receptors like selectins and platelet GPIba, are
also optimized for mechanical functions. As novel native structures
of adhesive complexes are elucidated in the future, the paradigm
of mechanically regulated cell adhesion will provide new insights
into the molecular details and physiological conditions of cell-cell
interactions.
It is possible that quaternary structural changes are also
involved in other biological interactions that are exposed to
mechanical force, even if they do not involve catch bonds. Force
regulation and, thus, the catch-bond mechanism might not
provide advantages in interactions that provide irreversible
adhesion, like between the holdfast of bacteria and colonizing
surface that involves one of the strongest adhesive biological
interactions [50]. Catch bonds are also unlikely to be involved in
the interactions where the binding strength is regulated by the
chemical modification of the receptor or ligand, like covalent
chemistry performed by adhering mussels [51]. At the same time,
other adhesive interactions that provide reversible adhesion, like
the foot of the gecko where spatula use pre-tension to control the
adhesion strength that drops off at a critical angle, [52,53] could
benefit from some version of force regulation of the quaternary
structures of the involved proteins. It remains to be determined,
however, to what extent the catch-bond mechanism can be
generalized to any reversible (non-covalent) biological interactions
regulated by mechanical force.
Materials and Methods
Flow Chamber Experiments
Bacterial binding to uroepithelial cells.
chamber experiments were in general performed as described
previously [5]. A confluent monolayer of T24 human urinary
bladder carcinoma cells was grown in Corning brand polystyrene
tissue culture dish in McCoy medium supplemented with 10%
FBS and penicillin/streptomycin. Flow chamber with 1 cm wide
Parallel plate flow
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org11 May 2011 | Volume 9 | Issue 5 | e1000617

Page 12

gasket was assembled on top of the monolayer. Recombinant
GFP-expressing E. coli were grown overnight in LB with shaking,
washed, and resuspended in McCoy medium to OD 1.0. Then E.
coli were flown into the chamber at low (0.01 Pa) or high (0.1 Pa)
shear stress conditions for 30 min at 37uC. Then the flow was
switched to McCoy medium without bacteria, and the flow
chamber was mounted on a Nikon TE200 inverted microscope
with a 10-fold phase-contrast objective and fluorescent lamp. The
T24 cells monolayer was examined under transmitted light, and
images of several fields of view per each plate with intact
monolayer were acquired both in transmitted and fluorescent
green light. The number of green dots, which corresponds to
bacteria sticking on the surface, was counted and averaged over 10
pictures at each shear. Examples of these pictures for both low and
high shear are presented in Figure 2a.
Bacterial binding to mannose.
mannose coated dishes was performed as described previously
[32]. The dishes coated with monomannosylated bovine serum
albumin were inserted into a parallel plate flow chamber. Then E.
coli were washed into the chamber at either 0.01 Pa or 0.1 Pa for
5 min at 37uC and the number of bacteria sticking onto the
surface was determined as described in ‘‘Bacterial Binding to
Uroepithelial Cells.’’ A comparison of E. coli binding to T24 cells
or mannose at both shears is presented in Figure 2b.
Fimbrial tip binding to mannose.
3 mm diameter were incubated with 100 mg/ml mannosylated
bovine serum albumin (V-labs Inc., Covington, LA) for 75 min
and washed with 0.2% bovine serum albumin in phosphate
buffered saline (PBS-BSA) to reduce nonspecific binding. Fimbrial
tips were immobilized on a Corning brand polystyrene tissue
culture dish at 0.1 mg/ml total protein for 1.25 h at 37uC, then
blocked with PBS-BSA overnight. To measure the rate of
accumulated binding of beads to the surface, the suspended
beads were washed over the surface at the indicated shear stress
levels for 5 min and the number of adherent beads at the end of
5 min was counted and plotted in Figure 2c.
Binding of bacteria to
Polystyrene beads with a
Constant Velocity Single Molecule Force Spectroscopy
Single molecule force spectroscopy experiments were performed
as described previously [7]. Fimbrial tips were immobilized on
plates as described above except at a much higher concentration
on account of the small size of the atomic force microscope (AFM)
cantilever tip. Olympus Biolever cantilevers were incubated with
100 mg/ml man-BSA at 37uC for 1.25 h and blocked overnight in
PBS-BSA. An Asylum MFP-3D AFM was used to probe the forces
on single bonds between the cantilever and surface in PBS-BSA.
The tip was pressed to the surface and then the tip withdrawn at a
constant velocity calculated to create 300 pN/s, given the spring
constant of the cantilever tips (calculated with the thermal method
and found to vary between 4.38 and 5.69 pN/nm for different
tips). The force at rupture was calculated as the difference between
the peak of tensile force and the average baseline force following
rupture, using an automated script. Nonspecific interactions
between the tip and surface were measured by adding 4%
alpha-methyl mannose to the PBS-BSA solution to prevent specific
bonds from forming. The resulting histogram of rupture forces was
plotted as a difference between the number of total events and
non-specific rupture events at a given force range (Figure 2d, bars).
The off rates (Figure 2d, circles) were then calculated as a function
of force according to the method of Evans et al. [18]. The
probability of rupture was estimated for the kthforce bin by
dividing the number of interactions that break in the force range
represented by the kthbin (DNk) by the number remaining at the
start of the bin (Nk=DNk+DNk+1+…). The rate of rupture was
calculated as this probability divided by Dtk, the time spent in that
bin, calculated as the width of force bin divided by loading rate.
Initial Structures and Definition of Domains
The conformations used in the molecular dynamics simulations
were derived from the crystallographic structure of the fimbrial tip
(PDB code 3JWN). The quaternary structure of the crystallized
fimbrial tip consists of four subunits and a chaperone protein
FimC (Figure 1a). Each subunit donates a b strand to the N-
terminal neighboring subunit, thus allowing for polymerization
(Figure 1b). In order to study the flexibility and the function of the
fimbrial tip, the entire structure was subdivided into domains.
Each domain consisted of the globular part of a subunit and
included the strand donated by the neighboring subunit. The
donated strand is covalently linked to the neighboring domain
through a short unstructured chain, termed a linker chain. Starting
from the N-terminus, the names and in parenthesis the exact
amino acid sequences of each domain are as follows: FimH lectin
(1–158), FimH pilin (161–279 and 1–12 of donated strand by
FimG subunit), FimG (16–144 and 1–12 of donated strand by
FimF subunit), and FimF (16–154 and 1–12 of donated strand by
the second FimF). In this article, if not specified, FimG and FimF
always refer to the domains and FimH to the lectin domain (Ld)
and pilin domain (Pd) including the linker chain. Otherwise, the
word ‘‘subunit’’ will accompany the name, e.g., FimG sub-unit.
The constant force simulations (see section ‘‘Simulations’’) were
run with the first four domains of the fimbrial tip, i.e., FimH Ld,
FimH Pd, FimG, and FimF, after having been equilibrated for
3 ns. In this construct, the strand donated by the second copy of
FimF was truncated after residue 14. In the simulations run with
just FimH, the protein was cleaved after residue 13 of the FimG
donated strand. In the runs with FimH Pd-FimG or FimG-FimF,
the protein was cleaved after residue 14 of the respective donated
strand.
Simulations
The MD simulations were performed with the program NAMD
[54] using the AMBER03 force field [55] and the TIP3P model of
water [56]. In the simulations with only two domains, the protein
was inserted into a cubic water box with a side length of 115 A˚
such that the distance between the protein and the edge of the box
measured at least 12 A˚. In the simulations with the four domains
(FimH Ld to FimF), a rectangular water box was used with side
lengths of 70 A˚670 A˚6200 A˚to allow for extension of the protein
when a pulling force was applied. The water molecules
overlapping with the protein or the ions were removed if the
distance between the oxygen atom of a water molecule and any
atom of the protein or any ion was smaller than 3.1 A˚. This value
is equivalent to the distance between water molecules at room
temperature and 1 atmosphere of pressure. To avoid finite size
effects, periodic boundary conditions were applied. Different initial
random velocities were assigned whenever more than one
simulation was performed with the same molecule. Electrostatic
interactions were calculated within a cutoff of 10 A˚, while long-
range electrostatic effects were taken into account by the Particle
Mesh Ewald summation method [57]. Van der Waals interactions
were truncated with the use of a switch function starting at 8 A˚
and turning off at 10 A˚.
Before production, the starting conformation and the solvent
were minimized by performing 6,000 steps of the conjugate
gradient method. Following minimization, the system was heated
by increasing the temperature stepwise in increments of 10 K each
every 1 ps during a total time of 30 ps until the target temperature
of 300 K was reached. During production, the temperature was
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org 12May 2011 | Volume 9 | Issue 5 | e1000617

Page 13

kept constant by using the Berendsen thermostat [58] with a
relaxation time of 0.1 ps, while the pressure was held constant at 1
atm by applying a pressure piston [59]. For the 300 K runs where
no external force was applied, the first 10 ns of unconstrained
simulation time were also considered part of the equilibration and
were not used for the analysis. The dynamics were integrated with
a time step of 2 fs. The covalent bonds involving hydrogens were
rigidly constrained by means of the SHAKE algorithm with a
tolerance of 1028. Snapshots were saved every 10 ps for trajectory
analysis.
The simulations where no external force was applied were
checked for convergence by calculating averages and standard
deviations of the magnitudes presented in Figure 4 on time
windows of 10 ns. In the case of the simulations with either the
entire FimH protein or the complex between Pd and FimG, the
averages over the 10 ns time window differed on values smaller
than the standard deviations, indicating that the runs had
converged. On the other hand, the quaternary structure of the
FimG-FimF complex was observed to be highly flexible because of
the near absence of inter-protein contacts; thus, the averages over
10 ns time windows had relatively larger fluctuations.
Constant force pulling.
A force of 200 pN was applied in
opposite directions to simulate the extension of the four domains
(FimH Ld to FimF). The force was applied to the Ca of residue 14
of the FimF donated strand located at the C-terminus and to the
center of mass of the Ca atoms of the following 13 residues located
near the mannose binding pocket at the N-terminus: F1, I13, N46,
D47, Y48, I52, D54, Q133, N135, Y137, N138, D140, and D141.
These are the same residues used to pull the protein in a previous
study by us [5]. Though the fimbrial tip crystal structure was
obtained in the absence of mannose, an assumption was made that
these residues interact with mannose based on the previously
obtained crystal structure of the FimH-FimC complex with
separated FimH domains [9,11]. The force was applied during
in total 10 ns and three runs were performed. Prior to pulling, the
four-domain construct was equilibrated at 300 K for 3 ns. In
addition, three 10 ns pulling simulations were performed also with
the isolated FimH protein, where the initial structure was taken
after 10 ns equilibration of the 40 ns run at 300 K (Table 2).
Determination of Native Contacts
The conformations sampled at room temperature were used to
determine native hydrogen bonds and salt bridges. To define a
hydrogen bond, a H … O distance cutoff of 2.7 A˚and a D–H …
O angle cutoff of 120u was used, where a donor D could either be
an oxygen or a nitrogen. Side chains were defined to form a
contact when the distance between their geometric centers was not
larger than 6 A. An interaction was defined as salt bridge if the
atoms Nf of Lys or Cf of Arg were closer than 4 A˚or 5 A˚,
respectively, from either the Cc of Asp or Cd of Glu. All histidines
were assumed neutral. Those hydrogen bonds and side chain
contacts present in at least 66% of the frames of the 300 K
simulations were selected as native contacts. Native inter-domain
contacts were used to monitor the separation of the domains from
each other in the pulling simulations.
Inter-Domain Buried Surface and Angles
Inter-domainburied
surface area (SASA) buried at the interface between two
domains was calculated by subtracting the SASA of the two
domains without the linker chain from the sum of the SASA of the
two domains independently.
Inter-domain hinge and twist angles.
for each domain were calculated by diagonalizing the moment of
surface.
Thesolvent accessible
The principal axes
inertia tensor using the program VMD [60]. The hinge angle
between two neighboring domains was defined as the angle
between their longest principal axes subtracted from 180u. The
twist angle was defined as the angle between the second longest
principal axes.
Calculation of the Elongation of the Protein
Elongation of the fimbrial tip under pulling was calculated
between the N-terminus of FimH (residue 1) and the C-terminus of
FimG (residue Asp12 of the donated FimF strand) because the
interface between FimG and FimF is rather flexible whereas the
interface between Pd and FimG is relatively stiff (Figure 4). This
distance for the native state was determined by averaging over the
last 2 ns of a 3 ns simulation with the entire fimbrial tip, where no
force was applied, and measured 12161 A˚. Because of the high
force used in the pulling simulations, the single domains were likely
to experience overstretching. Thus stretching of the single domains
was subtracted by calculating the distance between the termini of
the Ld, Pd, and FimG domains (see section ‘‘Initial Structures and
Definition of Domains’’ above for the exact definition of domains)
and comparing it to their average lengths during the last 2 ns of
the 3 ns run with the fimbrial tip.
Supporting Information
Figure S1
fimbrial tip domains and principal axes. (a) Crystallographic
structure of the fimbrial tip containing the four domains used in
the simulations. The angle between the first principal components
of neighboring domains (hinge angle) is indicated. (b) Conforma-
tion of FimH after 14 ns of simulation at 300 K from the 40 ns
run. (c) Conformation of the complex between FimH pilin domain
and FimG obtained after 14 ns of simulation at 300 K. (d)
Conformation of the complex between FimG and FimF obtained
after 14 ns of simulation at 300 K. Note that after 14 ns FimF is in
a different configuration as in the X-ray structure. (e) Conforma-
tion of the fimbrial tip after pulling at constant force for 10 ns
(pull 1).
Found at: doi:10.1371/journal.pbio.1000617.s001 (10.04 MB
EPS)
Comparison between different conformations of
Figure S2
domains during room temperature simulations (of the three 300 K
simulations with FimH, the longest, 40 ns, is shown here). (a) Ca
RMSD of adjacent domains from the native structure. (b) Ca
RMSD of single domains from the native structure. (c) Distance
between the centers of mass of adjacent domains. (d) Surface
buried at the interface between adjacent domains. (e) Hinge and (f)
twist angle between adjacent domains (see also Figure 1 and
‘‘Materials and Methods’’). The twist angle between FimH Pd and
FimG is negative and the corresponding ordinate is indicated on
the right. (g) Number of native side-chain contacts between
adjacent domains. The native side-chain contacts were determined
from 300 K runs performed with pairwise domains (see ‘‘Materials
and Methods’’).
Found at: doi:10.1371/journal.pbio.1000617.s002 (1.16 MB EPS)
Time series of quantities describing the flexibility of
Figure S3
changes at the interface between domains during the pull 1
simulation. Prior to pulling, 3 ns of equilibration were performed
and are shown in the left part of the plot. All plotted quantities are
time averages over a window of 200 ps. (a) Pairwise Ca RMSD of
adjacent domains from the native structure. (b) Distance between
the centers of mass of adjacent domains. (c) Surface buried at the
interface between adjacent domains. (d) Hinge and (e) twist angle
Time series of quantities describing conformational
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org13 May 2011 | Volume 9 | Issue 5 | e1000617

Page 14

between adjacent domains (see also Figure 1 and ‘‘Materials and
Methods’’). The twist angle between FimH Pd and FimG is
negative and the corresponding ordinate is indicated on the right.
(f) Number of native side-chain contacts between adjacent
domains. The native side-chain contacts were determined from
300 K runs performed with pairwise domains (see ‘‘Materials and
Methods’’). (g) Solvent accessible surface area of the carboxyl
oxygen of S114 and the amid group of C161, which are involved
in a native inter-domain hydrogen bond.
Found at: doi:10.1371/journal.pbio.1000617.s003 (0.39 MB EPS)
Figure S4
changes at the interface between domains during the pull 2
simulation. Prior to pulling, 3 ns of equilibration were performed
and are shown in the left part of the plot. All plotted quantities are
time averages over a window of 200 ps. (a) Pairwise Ca RMSD of
adjacent domains from the native structure. (b) Distance between
the centers of mass of adjacent domains. (c) Surface buried at the
interface between adjacent domains. (d) Hinge and (e) twist angle
between adjacent domains (see also Figure 1 and ‘‘Materials and
Methods’’). The twist angle between FimH Pd and FimG is
negative and the corresponding ordinate is indicated on the right.
(f) Number of native side-chain contacts between adjacent
domains. The native side-chain contacts were determined from
300-K runs performed with pairwise domains (see ‘‘Materials and
methods’’). (g) Solvent accessible surface area of the carboxyl
oxygen of S114 and the amid group of C161 which are involved in
a native inter-domain hydrogen bond.
Found at: doi:10.1371/journal.pbio.1000617.s004 (0.39 MB EPS)
Time series of quantities describing conformational
Figure S5
changes at the interface between domains during the pull 3
simulation. Prior to pulling, 3 ns of equilibration were performed
and are shown in the left part of the plot. All plotted quantities are
time averages over a window of 200 ps. (a) Pairwise Ca RMSD of
adjacent domains from the native structure. (b) Distance between
the centers of mass of adjacent domains. (c) Surface buried at the
interface between adjacent domains. (d) Hinge and (e) twist angle
between adjacent domains (see also Figure 1 and ‘‘Materials and
Methods’’). The twist angle between FimH Pd and FimG is
negative and the corresponding ordinate is indicated on the right.
(f) Number of native side-chain contacts between adjacent
domains. The native side-chain contacts were determined from
300 K runs performed with pairwise domains (see ‘‘Materials and
Methods’’). (g) Solvent accessible surface area of the carboxyl
oxygen of S114 and the amid group of C161, which are involved
in a native inter-domain hydrogen bond.
Found at: doi:10.1371/journal.pbio.1000617.s005 (0.39 MB EPS)
Time series of quantities describing conformational
Figure S6
contacts during the pull 1 simulation. The time series during the
3 ns equilibration run prior to start pulling are displayed on the
left and separated from the pulling plots by a vertical dashed line.
Side chain contacts between FimH Ld and FimH Pd are colored
in cyan, those between FimH Pd and FimG in orange, and those
between FimG and FimF in magenta. Native inter-domains
hydrogen bonds, observed only between FimH Ld and FimH Pd,
are colored in blue. The ordinate on the left lists the amino acids
involved in contacts. The residue on the left of the ‘‘-’’ is contained
in the domain closer to the N-terminus. In most cases, residues
involved in inter-domain contacts are also contained within the
subunit with the same name. The only exception is residue R12 in
the FimF subunit, which belongs to the FimG domain and
contacts A59 in FimF (see ‘‘Materials and Methods’’ for the
definition of sub-domains and subunits).
Found at: doi:10.1371/journal.pbio.1000617.s006 (1.07 MB EPS)
Time series of the formation of native inter-domains
Figure S7
contacts during the pull 2 simulation. The time series during the
3 ns equilibration run prior to start pulling are displayed on the
left and separated from the pulling plots by a vertical dashed line.
Side chain contacts between FimH Ld and FimH Pd are colored
in cyan, those between FimH Pd and FimG in orange, and those
between FimG and FimF in magenta. Native inter-domains
hydrogen bonds, observed only between FimH Ld and FimH Pd,
are colored in blue. The ordinate on the left lists the amino acids
involved in contacts. The residue on the left of the ‘‘-’’ is contained
in the domain closer to the N-terminus. In most cases, residues
involved in inter-domain contacts are also contained within the
subunit with the same name. The only exception is residue R12 in
the FimF subunit, which belongs to the FimG domain and
contacts A59 in FimF (see ‘‘Materials and Methods’’ for the
definition of sub-domains and subunits).
Found at: doi:10.1371/journal.pbio.1000617.s007 (1.05 MB EPS)
Time series of the formation of native inter-domains
Figure S8
contacts during the pull 3 simulation. The time series during the
3 ns equilibration run prior to start pulling are displayed on the
left and separated from the pulling plots by a vertical dashed line.
Side chain contacts between FimH Ld and FimH Pd are colored
in cyan, those between FimH Pd and FimG in orange, and those
between FimG and FimF in magenta. Native inter-domain
hydrogen bonds, observed only between FimH Ld and FimH
Pd, are colored in blue. The ordinate on the left lists the amino
acids involved in contacts. The residue on the left of the ‘‘-’’ is
contained in the domain closer to the N-terminus. In most cases,
residues involved in inter-domain contacts are also contained
within the subunit with the same name. The only exception is
residue R12 in the FimF subunit, which belongs to the FimG
domain and contacts A59 in FimF (see ‘‘Materials and Methods’’
for the definition of sub-domains and subunits).
Found at: doi:10.1371/journal.pbio.1000617.s008 (1.10 MB EPS)
Time series of the formation of native inter-domains
Figure S9
contacts during all three pulling simulations with FimH. Side
chain contacts between FimH Ld and FimH Pd are colored in
cyan. Native inter-domain hydrogen bonds between FimH Ld and
FimH Pd are colored in blue. The ordinate on the left lists the
amino acids involved in contacts. The residue on the left of the ‘‘-’’
is contained in the domain closer to the N-terminus (see ‘‘Materials
and Methods’’ for the definition of sub-domains and subunits and
native contacts).
Found at: doi:10.1371/journal.pbio.1000617.s009 (2.44 MB EPS)
Time series of the formation of native inter-domain
Figure S10
hinge axis of residues involved in inter-domain contacts between
Ld and Pd. The values were averaged over three pulling
simulations with the fimbrial tip (left) and three pulling runs with
the isolated FimH protein (right). A straight line was fitted to show
that the contacts break in a sequential manner with contacts
located farther away from the hinge axis (Figure 5a in the article)
breaking earlier in the simulations. The Pearson’s linear
correlation coefficient is 0.70 and 0.88 for the runs with the
fimbrial tip and the runs with isolated FimH, respectively, while
the p values are 0.02 and 8e24, respectively. Figure 5 in the article
contains a ranking analysis of the Ca RMSDs versus distance from
the axis averaged over all six pulling simulations.
Found at: doi:10.1371/journal.pbio.1000617.s010 (0.03 MB EPS)
Ca RMSD at time of rupture versus distance to the
Table S1
FimH-FimG complex has to undergo until the transition state is
reached (see ‘‘Materials and Methods’’ for details). The time point
along the three pulling simulations is determined where the side
Estimate of the minimal amount of elongation that the
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org14 May 2011 | Volume 9 | Issue 5 | e1000617

Page 15

chain of Arg166 loses its inter-domain contacts with Ld (one side
chain contact with Val155 and two hydrogen bonds with the
carbonyl oxygen of Ala115; Figures S6, S7, S8, S9). This event is
observed to always be the first rupture event in all three
simulations. Thus it is assumed that the location of the transition
state will either be at exactly this event or later. It is worth
mentioning that in three pulling simulations with just FimH,
rupture of contacts involving Arg166 was also observed to be the
first rupture event (Figures S9 and S10), providing further
statistical evidence.
Found at: doi:10.1371/journal.pbio.1000617.s011 (0.03 MB
DOC)
Acknowledgments
We thank the anonymous reviewers for providing excellent points for
manuscript improvement, Steve Moseley for critical reading of the
manuscript and very helpful suggestions, Goragot Wisedchaisri for sharing
outstanding expertise, and Tamir Gonen for collegial discussions. The
computations were performed on the Abe supercomputer at the National
Center for Supercomputing Applications.
Author Contributions
The author(s) have made the following declarations about their
contributions: Conceived and designed the experiments: WET EVS.
Performed the experiments: PA BAK VT OY. Analyzed the data: GI
MJW. Contributed reagents/materials/analysis tools: ILT RES. Wrote the
paper: GI WET EVS. Contributed to data collection, analysis, andinter-
pretation: EB.
References
1. Beachey E (1981) Bacterial adherence–adhesion-receptor interactions mediating
the attachment of bacteria to mucosal surfaces. J Infect Dis 143: 325–345.
2. Klemm P (1997) Fimbriae: adhesion, genetics, biogenesis, and vaccines CRC
Press.
3. Hahn E, Wild P, Hermanns U, Sebbel P, Glockshuber R, et al. (2002) Exploring
the 3D molecular architecture of Escherichia coli type 1 pili. J Mol Biol 323:
845–857.
4. Jones C, Pinkner J, Roth R, Heuser J, Nicholes A, et al. (1995) FimH adhesin of
type-1 pili is assembled into a fibrillar tip structure in the enterobacteriaceae.
Proc Natl Acad Sci U S A 92: 2081–2085.
5. Thomas WE, Trintchina E, Forero M, Vogel V, Sokurenko EV (2002) Bacterial
adhesion to target cells enhanced by shear force. Cell 109: 913–923.
6. Anderson BN, Ding AM, Nilsson LM, Kusuma K, Tchesnokova V, et al. (2007)
Weak rolling adhesion enhances bacterial surface colonization. J Bacteriol 189:
1794–1802.
7. Yakovenko O, Sharma S, Forero M, Tchesnokova V, Aprikian P, et al. (2008)
FimH forms catch bonds that are enhanced by mechanical force due to allosteric
regulation. J Biol Chem 283: 11596–11605.
8. Le Trong SN, Aprikian P, Kidd BA, Forero-Shelton M, Tchesnokova V, et al.
(2010) Structural basis for mechanical force regulation of the adhesin FimH via
finger trap-like b sheet twisting. Cell 141: 645–655.
9. Choudhury D, Thompson A, Stojanoff V, Langermann S, Pinkner J, et al.
(1999) X-ray structure of the FimC-FimH chaperone-adhesin complex from
uropathogenic Escherichia coli. Science 285: 1061–1066.
10. Gossert AD, Bettendorff P, Puorger C, Vetsch M, Herrmann T, et al. (2008)
NMR structure of the Escherichia coli type 1 pilus subunit FimF and its
interactions with other pilus subunits. J Mol Biol 375: 752–763.
11. Hung C, Bouckaert J, Hung D, Pinkner J, Widberg C, et al. (2002) Structural
basis of tropism of Escherichia coli to the bladder during urinary tract infection.
Mol Microbiol 44: 903–915.
12. Eidam O, Dworkowski FSN, Glockshuber R, Gruetter MG, Capitani G (2008)
Crystal structure of the ternary FimC-FimF(t)-FimD(N) complex indicates
conserved pilus chaperone-subunit complex recognition by the usher FimD.
FEBS Lett 582: 651–655.
13. Puorger C, Eidam O, Capitani G, Erilov D, Gruetter MG, et al. (2008) Infinite
kinetic stability against dissociation of supramolecular protein complexes
through donor strand complementation. Structure 16: 631–642.
14. Bouckaert J, Berglund J, Schembri M, De Genst E, Cools L, et al. (2005)
Receptor binding studies disclose a novel class of high-affinity inhibitors of the
Escherichia coli FimH adhesin. Mol Microbiol 55: 441–455.
15. Weissman S, Chattopadhyay S, Aprikian P, Obata-Yasuoka M, Yarova-
Yarovaya Y, et al. (2006) Clonal analysis reveals high rate of structural mutations
in fimbrial adhesins of extraintestinal pathogenic Escherichia coli. Mol
Microbiol 59: 975–988.
16. Thomas W, Forero M, Yakovenko O, Nilsson L, Vicini P, et al. (2006) Catch-
bond model derived from allostery explains force-activated bacterial adhesion.
Biophys J 90: 753–764.
17. Tchesnokova V, Aprikian P, Yakovenko O, Larock C, Kidd B, et al. (2008)
Integrin-like allosteric properties of the catch bond-forming FimH adhesin of
Escherichia coli. J Biol Chem 283: 7823–7833.
18. Kinoshita K, Leung A, Simon S, Evans E (2010) Long-lived, high-strength states
of ICAM-1 bonds to beta(2) integrin, II: lifetimes of LFA-1 bonds under force in
leukocyte signaling. Biophys J 98: 1467–1475.
19. Thomas WE, Vogel V, Sokurenko E (2008) Biophysics of catch bonds. Annu
Rev Biophys 37: 399–416.
20. Marshall B, Long M, Piper J, Yago T, McEver R, et al. (2003) Direct
observation of catch bonds involving cell-adhesion molecules. Nature 423:
190–193.
21. Le Trong I, Aprikian P, Kidd BA, Thomas WE, Sokurenko EV, et al. (2010)
Donor strand exchange and conformational changes during E. coli fimbrial
formation. J Struct Biol 172: 380–388.
22. Evans E (2001) Probing the relation between force - lifetime - and chemistry in
single molecular bonds. Ann Rev Biophys Biomol Struct 30: 105–128.
23. Kramers H (1940) Brownian motion in a field of force and the diffusion model of
chemical reactions. Physica 7: 284–304.
24. Izrailev S, Stepaniants S, Balsera M, Oono Y, Schulten K (1997) Molecular
dynamics study of unbinding of the avidin-biotin complex. Biophys J 72:
1568–1581.
25. Isralewitz B, Baudry J, Gullingsrud J, Kosztin D, Schulten K (2001) Steered
molecular dynamics investigations of protein function. J Mol Graph Model 19:
13–25.
26. Nilsson LM, Thomas WE, Sokurenko EV, Vogel V (2008) Beyond induced-fit
receptor-ligand interactions: structural changes that can significantly extend
bond lifetimes. Structure 16: 1047–1058.
27. Whitfield M, Ghose T, Thomas W (2010) Shear-stabilized rolling behavior of E.
coli examined with simulations. Biophys J 99: 2470–2478.
28. Nilsson L, Thornas W, Sokurenko E, Vogel V (2006) Elevated shear stress
protects Escherichia coli cells adhering to surfaces via catch bonds from
detachment by soluble inhibitors. Appl Environ Microbiol 72: 3005–3010.
29. Aprikian P, Tchesnokova V, Kidd B, Yakovenko O, Yarov-Yarovoy V, et al.
(2007) Interdomain interaction in the FimH adhesin of Escherichia coli regulates
the affinity to mannose. J Biol Chem 282: 23437–23446.
30. Sokurenko E, Chesnokova V, Doyle R, Hasty D (1997) Diversity of the
Escherichia coli type 1 fimbrial lectin—differential binding to mannosides and
uroepithelial cells. J Biol Chem 272: 17880–17886.
31. Weissman SJ, Beskhlebnaya V, Chesnokova V, Chattopadhyay S, Stamm WE,
et al. (2009) Differential stability and trade-off effects of pathoadaptive mutations
in the Escherichia coli FimH adhesin (vol. 75, pg 3548, 2007). Infection and
Immunity 77: 1720.
32. Thomas WE, Nilsson LM, Forero M, Sokurenko EV, Vogel V (2004) Shear-
dependent ‘stick-and-roll’ adhesion of type 1 fimbriated Escherichia coli. Mol
Microbiol 53: 1545–1557.
33. Jeffrey B, Udaykumar H, Schulze K (2003) Flow fields generated by peristaltic
reflex in isolated guinea pig ileum: impact of contraction depth and shoulders.
Am J Physiol Gastrointest Liver Physiol 285: G907–G918.
34. Lentle RG, Janssen PWM (2008) Physical characteristics of digesta and their
influence on flow and mixing in the mammalian intestine: a review. J Comp
Physiol B Biochem Syst Environ Physiol 178: 673–690.
35. Thomas W (2008) Catch bonds in adhesion. Annu Rev Biomed Eng 10: 39–57.
36. Forero M, Yakovenko O, Sokurenko EV, Thomas WE, Vogel V (2006)
Uncoiling mechanics of Escherichia coli type I fimbriae are optimized for catch
bonds. PLoS Biol 4: 1509–1516. doi:10.1371/journal.pbio.0040298.
37. Phan UT, Waldron T, Springer TA (2006) Remodeling of the lectin-EGF-like
domain interface in P- and L-selectin increases adhesiveness and shear resistance
under hydrodynamic force. Nat Immunol 7: 883–889.
38. Sarangapani K, Yago T, Klopocki A, Lawrence M, Fieger C, et al. (2004) Low
force decelerates L-selectin dissociation from P-selectin glycoprotein ligand-1
and endoglycan. J Biol Chem 279: 2291–2298.
39. Kong F, Garcia AJ, Mould AP, Humphries MJ, Zhu C (2009) Demonstration of
catch bonds between an integrin and its ligand. J Cell Biol 185: 1275–1284.
40. Yago T, Lou J, Wu T, Yang J, Miner JJ, et al. (2008) Platelet glycoprotein Iba
forms catch bonds with human WT vWF but not with type 2B von Willebrand
disease vWF. J Clin Invest 118: 3195–3207.
41. Interlandi G, Thomas WE (2010) The catch bond mechanism between von
Willebrand Factor and platelets investigated by molecular dynamics simulations.
Proteins: Struct, Funct, Bionf 78: 2506–2522.
42. Guo B, Guilford WH (2006) Mechanics of actomyosin bonds in different
nucleotide states are tuned to muscle contraction. Proc Natl Acad Sci U S A 103:
9844–9849.
43. Uezumi A, Fukada S-I, Yamamoto N, Takeda Si, Tsuchida K (2010)
Mesenchymal progenitors distinct from satellite cells contribute to ectopic fat
cell formation in skeletal muscle. Nat Cell Biol 12: 143–152.
Mechanical Force Sensor
PLoS Biology | www.plosbiology.org15 May 2011 | Volume 9 | Issue 5 | e1000617