Force spectroscopy of collagen fibers to investigate their mechanical properties and structural organization.
ABSTRACT Tendons are composed of collagen and other molecules in a highly organized hierarchical assembly, leading to extraordinary mechanical properties. To probe the cross-links on the lower level of organization, we used a cantilever to pull substructures out of the assembly. Advanced force probe technology, using small cantilevers (length <20 microm), improved the force resolution into the sub-10 pN range. In the force versus extension curves, we found an exponential increase in force and two different periodic rupture events, one with strong bonds (jumps in force of several hundred pN) with a periodicity of 78 nm and one with weak bonds (jumps in force of <7 pN) with a periodicity of 22 nm. We demonstrate a good correlation between the measured mechanical behavior of collagen fibers and their appearance in the micrographs taken with the atomic force microscope.
- SourceAvailable from: kuleuven.be[show abstract] [hide abstract]
ABSTRACT: Spider capture silk is a natural material that outperforms almost any synthetic material in its combination of strength and elasticity. The structure of this remarkable material is still largely unknown, because spider-silk proteins have not been crystallized. Capture silk is the sticky spiral in the webs of orb-weaving spiders. Here we are investigating specifically the capture spiral threads from Araneus, an ecribellate orb-weaving spider. The major protein of these threads is flagelliform protein, a variety of silk fibroin. We present models for molecular and supramolecular structures of flagelliform protein, derived from amino acid sequences, force spectroscopy (molecular pulling) and stretching of bulk capture web. Pulling on molecules in capture-silk fibres from Araneus has revealed rupture peaks due to sacrificial bonds, characteristic of other self-healing biomaterials. The overall force changes are exponential for both capture-silk molecules and intact strands of capture silk.Nature Material 05/2003; 2(4):278-83. · 35.75 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Extracellular matrix proteins are thought to provide a rigid mechanical anchor that supports and guides migrating and rolling cells. Here we examine the mechanical properties of the extracellular matrix protein tenascin by using atomic-force-microscopy techniques. Our results indicate that tenascin is an elastic protein. Single molecules of tenascin could be stretched to several times their resting length. Force-extension curves showed a saw-tooth pattern, with peaks of force at 137pN. These peaks were approximately 25 nm apart. Similar results have been obtained by study of titin. We also found similar results by studying recombinant tenascin fragments encompassing the 15 fibronectin type III domains of tenascin. This indicates that the extensibility of tenascin may be due to the stretch-induced unfolding of its fibronectin type III domains. Refolding of tenascin after stretching, observed when the force was reduced to near zero, showed a double-exponential recovery with time constants of 42 domains refolded per second and 0.5 domains per second. The former speed of refolding is more than twice as fast as any previously reported speed of refolding of a fibronectin type III domain. We suggest that the extensibility of the modular fibronectin type III region may be important in allowing tenascin-ligand bonds to persist over long extensions. These properties of fibronectin type III modules may be of widespread use in extracellular proteins containing such domain.Nature 06/1998; 393(6681):181-5. · 38.60 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Here we demonstrate the implementation of a single-molecule force clamp adapted for use with an atomic force microscope. We show that under force-clamp conditions, an engineered titin protein elongates in steps because of the unfolding of its modules and that the waiting times to unfold are exponentially distributed. Force-clamp measurements directly measure the force dependence of the unfolding probability and readily captures the different mechanical stability of the I27 and I28 modules of human cardiac titin. Force-clamp spectroscopy promises to be a direct way to probe the mechanical stability of elastic proteins such as those found in muscle, the extracellular matrix, and cell adhesion.Proceedings of the National Academy of Sciences 02/2001; 98(2):468-72. · 9.74 Impact Factor
Force Spectroscopy of Collagen Fibers to Investigate Their
Mechanical Properties and Structural Organization
Thomas Gutsmann, Georg E. Fantner, Johannes H. Kindt, Manuela Venturoni, Signe Danielsen,
and Paul K. Hansma
Department of Physics, University of California, Santa Barbara, California 93101
extraordinary mechanical properties. To probe the cross-links on the lower level of organization, we used a cantilever to pull
substructures out of the assembly. Advanced force probe technology, using small cantilevers (length \20 mm), improved the
force resolution into the sub-10 pN range. In the force versus extension curves, we found an exponential increase in force and
two different periodic rupture events, one with strong bonds (jumps in force of several hundred pN) with a periodicity of 78 nm
and one with weak bonds (jumps in force of\7 pN) with a periodicity of 22 nm. We demonstrate a good correlation between the
measured mechanical behavior of collagen fibers and their appearance in the micrographs taken with the atomic force
Tendons are composed of collagen and other molecules in a highly organized hierarchical assembly, leading to
Tendons are among the most important stress-carrying
structures in mammals. They have a key role in musculo-
skeletal biomechanics. The knowledge of the mechanical
properties of tendons is important for the development of
new materials based on biological polymers and also for an
understanding of collagen-related diseases. Furthermore, the
properties of collagen and associated polymers are very
important to understand the structure and functional
mechanisms of biocomposites such as bone and cartilage
on the microscopic level.
Tendons have a hierarchical structure consisting of
smaller entities called fascicles. These fascicles consist of
fibrils, which again consist of sub- and/or microfibrils. The
fibrils consist mainly of two components, collagen and
proteoglycan (Scott, 1991). Type I collagen, the most
abundant form of collagen (;90%), is present in many
tissues, such as bone, tendon, and skin. The type I collagen
molecule is a triple-stranded coil consisting of two a1-chains
and one a2-chain. Three of the left-handed coiled collagen
molecules form a right-handed coiled triple helix called
tropocollagen (Ramachandra and Karthan, 1955). For the
structural organization of tropocollagen, there are different
models discussed in literature. Most of the research on the
structure and the mechanical properties of collagen has been
done on type I collagen, which has the most well-organized
structure of all collagen types. Type I collagen fibrils in their
native form typically display a banding pattern with
a periodicity of 67 nm, called D period, when visualized
with transmission electron microscopy (Schmitt et al., 1942;
Chapman and Hulmes, 1984) or AFM (Baselt et al., 1993;
Revenko et al., 1994). The structural organization of the 67-
nm banding is still a source of discussion. Petruska and
Hodge developed a widely accepted model of the organiza-
tion of collagen molecules into fibrils that was based on the
results they obtained from transmission electron microscopy
micrographs of negatively stained samples. The 67-nm
periodicity was explained by a repetition of overlap and gap
regions (Petruska and Hodge, 1964). In addition to the 67-
nm banding, smaller and larger periodicities were also
observed (Venturoni et al., 2003).
Collagen is described as an exceptional design for elastic-
energy storage and as a modest design for strength and
toughness (Gosline et al., 2002). Therefore, the knowledge
of the function of collagen fibrils is of general interest for
material researchers. The stress versus strain curve of
tendons (Fig. 1) can be divided into five distinct regions
(for review see Fratzl et al., 1998):
1. Toe region: A small strain leads to the removal of
macroscopic crimps in the collagen fibrils, which is
visible in the light microscope (Diamant et al., 1972).
2. Heel region: There is a straightening of kinks in the
collagen structure at strains beyond 3%. This occurs first
at the fibrillar and then at the molecular level by reducing
the disorder in the lateral molecular packing.
3. Linear region: Higher strains lead to a stretching of
collagen triple helices or of the cross-links between
helices. This stretching increases the length of the gap
region with respect to the length of the overlap region.
Moreover, it was proposed that molecular gliding within
the fibrils also plays a role because not all of the
elongation could be explained by stretching of the fibrils.
4. Plateau region: This region is pronounced in cross-link
deficient collagen and indicates creep behavior caused by
slippage within fibrils (Puxkandl et al., 2002).
Submitted September 2, 2003, and accepted for publication December 4,
Address reprint requests to Thomas Gutsmann, Tel.: 149-45-37-188-291;
Fax: 149-45-37-188-632; E-mail: firstname.lastname@example.org.
Thomas Gutsmann’s present address is Research Center Borstel, Leibniz
Center for Medicine and Biosciences, Dept. of Immunochemistry and
Biochemical Microbiology, Div. of Biophysics, Parkallee 10, D-23845
? 2004 by the Biophysical Society
3186Biophysical Journal Volume 86May 2004 3186–3193
5. Rupture: Ultimately, a higher strain leads to a disruption
of the structure of the fibril.
Collagen fibrils are capable of reversible deformation.
This property makes collagen an elastic protein (Gosline
et al., 2002) with a resilience of ;90% (Shadwick, 1990).
Besides collagen, other molecules, like proteoglycans (Cribb
and Scott, 1995), also play a role in the mechanical
properties of tendon fibrils. It was proposed that these other
molecules are important for the gliding of fibrils in the matrix
(Puxkandl et al., 2002). All common models are based on the
assumption of laterally homogeneous close packing of
fibril is an inhomogeneous structure composed of a relatively
hard shell and a less dense core (Gutsmann et al., 2003).
Thus, the mechanical properties of the whole fibril depend on
the properties of the shell and the core.
Besides the macroscopic stretching of whole tendons, Sun
et al. (2002) used the optical tweezer technique to determine
mechanical properties of single collagen monomers. The
persistence length of collagen I monomers was determined to
be 14.5 nm and the contour length was 309 nm (Sun et al.,
2002). However, there is a big gap in determining
mechanical properties between the stretching of single
molecules and the stretching of whole tendons. The atomic
force microscope has proven to be a powerful tool for the
characterization of biological macromolecules through high-
resolution topography imaging (Bustamante et al., 1997;
Hansma et al., 1997) and single molecule force spectroscopy
(Clausen-Schaumann et al., 2000; Zlatanova et al., 2000;
Carrion-Vazquez et al., 2000). Force spectroscopy experi-
ments have been performed on single molecules like titin
(Oberhauser et al., 2001), DNA (Strunz et al., 1999), and
polysaccharides (Marszalek et al., 2002). Becker et al.
recently published data on pulling spider capture-silk
molecules out of original fibers (Becker et al., 2003). We
previously used an atomic force microscope to perform force
spectroscopy experiments on collagen molecules that were
not structured in fibrils (Thompson et al., 2001).
Here we present force spectra (pulling curves) of
molecules pulled out of native fibrils from rat tail tendons.
To get a high-force resolution we used a small-cantilever
prototype AFM. The force curves show various character-
istics such as two distinct periodicities in the rupture events,
which we correlated with topography images of collagen
MATERIAL AND METHODS
Tails from rats, sacrificed for other experiments, were frozen at ?208C
typically for weeks, before our experiments. Rat tail tendons were removed
from thawed rat tails and stored in a buffer containing 150 mM NaCl and 2
mM Tris adjusted to a pH of 7.4 for several hours before sample preparation.
Fibrils from rat tail tendons were prepared wet on a glass disc, and dried
with a stream of filtered air. They were then imaged with an AFM (Veeco/
Digital Instruments, Santa Barbara, CA) in contact mode using commercial
cantilevers (CSC12, MikroMasch, Portland, OR).
The dried tendons were cut into pieces (;7 mm long) and the two ends
were glued to the glass disc with epoxy (2-Ton Clear Epoxy, Devcon,
Danvers, MA). The fibrils were rehydrated in a buffer containing 150 mM
NaCl, 2 mM Tris adjusted to a pH of 7.4. The pulling experiments were
conducted using different types of conventional and small cantilevers
(length \20 mm; width \8 mm) and different types of AFM: a prototype
small-cantilever AFM, a Veeco/Digital Instruments multimode, and
a molecular force puller MFP1 (Asylum Research, Santa Barbara, CA).
We used different controls for the z-position, such as strain gauges and the
sensored z axis of the MFP1. All data shown in this article was conducted
using the prototype small-cantilever AFM. The cantilevers that were
typically used had spring constants of ;25 mN/m, and resonance
frequencies of ;40 kHz, which were determined by the measurement of
the thermal fluctuations of the cantilevers. On the cantilevers, a 0.5-mm-long
tip was deposited by electron beam deposition. The pulling curves shown
were taken with loading rates of 7 or 14 mm/s.
For the histogram in Fig. 3 C we used data from 25 randomly chosen
pulling curves. The distances between consecutive significant rupture events
(rupture events with a change in force of [5 pN) were collected. No
distances below 16 nm were included because the evaluation of some of
these short distances was not precise enough due to noise, bending of the
cantilever, and other influences. No events were collected in the very
complex initial phase (200–300 nm), where several uncontrollable factors
make an interpretation of the data unreliable.
Before performing the pulling experiments, we imaged the
dry samples and found two different banding periodicities of
the collagen fibrils: the 65-nm D-spacing (of dried collagen
fibrils) that is typical for collagen fibrils and a small 23-nm
spacing (Fig. 2), which occurred in \3% of the collagen
fibrils. For the pulling experiments the collagen fibrils were
Most of the force (F) versus extension (z) curves showed
very complex structures containing different features. Fig.
3 A shows a force-extension curve recorded after pressing
the cantilever tip into the collagen fibril for[10 s. In many
cases, the force increased up to several nN and showed
a small hysteresis effect (see Fig. 3 A). Due to the
relatively high forces, we propose that these force-
curve is divided into five regions: (A) Toe region, (B) heel region, (C) linear
region, (D) plateau region, and (E) the rupture of the tendon.
Stress versus strain curve of a rat tail tendon. The stress-strain
Force Spectroscopy of Collagen Fibers 3187
Biophysical Journal 86(5) 3186–3193
extension curves resulted from pulling on structural
elements that were higher in the hierarchy of the collagen
fibril than a microfibril.
In few force-extension curves, we observed relatively
steep spikes (Fig. 3 B). The force increased over a distance of
10 nm by a force of ;400 pN. Because of the complex initial
phase of each of these spikes, it is not possible to perform
a reliable worm-like chain (WLC) or other fit.
Most curves showed a superposition of different features
making it impossible to describe all of them in detail.
However, a histogram of the distances between rupture
events (Fig. 3 C) shows two distinct distances, one around 23
and one around 77 nm. In most cases the two distinct
distances appear in repeating patterns that are described
below. It is important to notice that the observed ruptures are
irreversible; once a bond breaks it will not reform again. A
histogram of the respective peak forces did not show specific
levels of forces (data not shown).
In Fig. 4 the shape of the force-extension curve resem-
bles an exponential increase. In numerous previous force
spectroscopy studies, the WLC model was used to char-
acterize mechanical unfolding of proteins (Oberhauser et al.,
1998; Rief et al., 1999; Fisher et al., 1999; Oroudjev et al.,
2002). The WLC describes the relationship between the
protein extension and the entropic force generated as a result
of such extension. However, fitting our data with this model
led to values of persistence lengths of\10 pm, which is be-
low the length of a single atom, thus making this model not
suitable. This is also true for the (extended) Langevin
fiber showing the 67-nm D-spacing; (B) fiber showing a 23-nm banding
pattern, which is observable in \3% of the fibrils. These images show the
AFM deflection signals.
AFM images of rat tail tendon collagen.(A) A typical collagen
of curves can be found in the force versus extension curves. (A) A reversible
increase to high force values shows hysteresis; (B) steep increase in force
that may be due to stretching individual molecules. (C) The histogram of the
distances between rupture events shows two peaks, one around 23 (arrow a)
and one around 77 nm (arrow b). The inset shows the histogram of the same
data, but with twice the bin size to demonstrate the significance of the two
peaks. Most of these two distinct distances were observed in repeating
Force spectroscopy of rat tail tendon fibrils. Different shapes
3188 Gutsmann et al.
Biophysical Journal 86(5) 3186–3193
function (Rief et al., 1997). The force curve F(z) is well
described by an exponential function:
FðzÞ ¼ F01ðFP? F0Þexp
z ? zP
with F0being the initial force starting the stretching on one
individual part of the structure, FPbeing the rupture force, zP
being the peak position, and D being a characteristic length.
We introduced F0because when pulling on assemblies of
molecules, it is sometimes possible that parallel substruc-
tures of the whole assembly will only begin to stretch at
forces starting higher than zero. For the pulling curve in Fig.
4 B we determined the characteristic length D to be 80 6 3
nm. The jumps down to zero force were not reversible, thus,
they were ruptures and not a reversible unfolding of proteins
like titin. However, we do not know whether this rupture
originates from the breaking of bonds between the tip and the
molecule or from the breaking of bonds between two or more
molecules. It is unlikely that covalent bonds in the molecules
Interestingly, we found two different patterns of rupture
events showing a distinct periodicity that clearly show up in
the histogram (Fig. 3 C). The first pattern shows a periodicity
of 78 nm and is easily visible in the histogram, an example is
shown in Fig. 5. After an increase in force up to [500 pN,
repeating rupture events occurred. They resulted in a stepped
decrease in the maximum force reached before the breaking
of each bond. The average distance between the rupture
events was 78 6 3 nm (Fig. 5 B). In several hundred pulling
curves, we never observed more then five of these repeating
events in one pulling curve.
The second periodicity is barely detectable because the
jumps in force were \7 pN. These measurements became
possible with the use of low spring-constant, high-resonance
frequency, small cantilevers (Viani et al., 1999). These
rupture events occurred in a quasilinear region of the force-
extension curve (Fig. 6 A). The average distance between
these rupture events was 22 6 2 nm (Fig. 6 B) and the
average increase in force between two consecutive events
was 53 6 4 pN. The overall linear increase does not allow
the direct conclusion that it resulted from a linear stretching
of the molecules. The shape of the curves between two
rupture events appears to be more linear than exponential.
However, the distance between the ruptures is too short to
perform a precise analysis, even with the high-force
resolution achieved by using a small cantilever.
Collagen fibrils are important for the function of musculo-
skeletal biomechanics in various species. The mechanical
the observed force versus extension curves can be fitted by an exponential
function with a characteristic length D of ;80 nm. (A) Overview of one
pulling experiment on lin/log scale with the same exponential function fitted
to different events showing that most of the events have the same char-
acteristics. (B) Detail of the same experiment on lin/lin scale with an
pulling curves, a regular pattern with an average distance between the
rupture events of 78 6 3 nm can be observed. This pattern is statistically
significant and is clearly visible in the distance histogram (Fig. 3 C). The
jumps in force are not reversible and they always lead to a stepwise decrease
of the force versus distance curve.
Stepwise rupture of bonds in the collagen fiber. In some
Force Spectroscopy of Collagen Fibers3189
Biophysical Journal 86(5) 3186–3193
properties of whole tendons, as well as the chemical
composition and the topography of collagen fibrils, were
studied in detail. However, very little is known about the
mechanical properties of the fibrils on the molecular level.
Knowledge of this will lead to a better understanding of the
binding between tropocollagen molecules, how they assem-
ble to fibrils, and with that an understanding of the
topography as well as the macroscopic mechanical behavior.
Later, we will discuss our results along with the topography
of the fibrils and the macroscopic stress versus strain curves
Exponential increase in force
The extension of collagen fibrils inside the tendon is
considerably less than the total extension of the tendon.
Puxkandl et al. proposed that the deformation must occur
outside the collagen fibrils, presumably in the proteoglycan-
rich matrix (Puxkandl et al., 2002). Thus, the tendon is
considered as a composite material with collagen fibrils
embedded in a proteoglycan-rich matrix.
There are relatively few theoretical models describing the
rupture of multiple parallel bonds under dynamic loading
(Seifert, 2000) and most of them simplify the model by using
Hookean springs with zero rest length. Puxkandl et al.
introduced a Voight-Kelvin mechanical model consisting of
parallel elastic (spring) and viscous (dashpot) portions for the
collagen fibrils, as well as for the proteoglycan matrix
(Puxkandl et al., 2002). This model leads to an exponential
stress versus strain function. This is in agreement with the
exponential function we observed in the pulling curves. The
most likely explanation on the molecular level is a stretching
of whole assemblies of collagen and other components of
the fibrils, e.g., proteoglycan. The stretching of individ-
ual molecules, molecular cross-links and matrix shearing all
probably contribute to the pulling forces. However, without
a suitable model, it is difficult to make an unambiguous
connection between the characteristic length D and the aver-
age distance between the repeating rupture pattern of ;80
nm. We suggest that the described mechanical behavior on
the molecular level corresponds mainly to the heel and/or
linear region on the macroscopic level (Fig. 1). We exclude
the other regions because the toe region corresponds to the
removal of macroscopic crimps in the collagen fibrils and the
plateau region corresponds to the slippage within fibrils,
neither of which lead to an exponential increase in force.
Repeating rupture pattern of 78 nm
The periodicity of 78 6 3 nm between rupture events in the
force spectra (Fig. 5) shows that there are bonds between
molecules in the fibrils with a distance of ;78 nm. We
believe that this pattern results from pulling an assembly of
molecules out of fibrils, followed by a sequential rupture of
individual subunits. Most likely the molecules are tropocol-
lagen, however, it is possible that other molecules, like
proteoglycans, participate. Therefore, the decrease of the
peak forces with each successive rupture event can be
explained by the fact that when a bond ruptures, the force on
the remaining bonds increases. Thus, the overall force that
needs to be applied to a bundle of molecules to break the
weakest bond in the bundle decreases with the number of
force-carrying subunits in the bundle. For the interpretation
of our data, the precise knowledge of the involved molecules
is not so important because we focused on the structure of
The measured periodicity of 78 nm is ;16% longer
as compared to the 67-nm D-banding obtained from the
topography of wet collagen fibrils. The 67 nm is the
projection of the wavy fibril structure onto the x-y plane and,
with a height corrugation of at least 3–7 nm. The surface
distance is therefore longer than 67 nm by at least 6 nm
(Venturoni et al., 2003; Lin and Goh, 2002). Furthermore, it
was proposed that the collagen molecules are tilted by ;168
to the fibril’s axis leading to a length of 70 nm. These two
effects lead to a length of at least 76 nm. Katsura and Ono
proposed that the precoiled length of D in the a-chain is ;81
nm and the length of D in the tropocollagen triple helix is
;75 nm (Katsura and Ono, 1998). This is in agreement with
our data suggesting a periodicity of bonds between the
experiments, the linear-like increase in force is due to very small jumps in
force. These jumps are ;7 pN every 22 nm corresponding to every 53 pN.
This 22-nm periodicity is also statistically significant and appears as a spike
in the distance histogram (Fig. 3 C).
Small rupture events in the linear-like region. In most of the
3190Gutsmann et al.
Biophysical Journal 86(5) 3186–3193