The origins and regulation of tissue tension: Identification of

Experimental Cell Research 312 (2006) 423 – 433
www.elsevier.com/locate/yexcr
Research Article
The origins and regulation of tissue tension: Identification of collagen
tension-fixation process in vitro
Massimo Marenzana, Nick Wilson-Jones, Vivek Mudera, Robert A. Brown*
University College London, RFUCMS, Tissue Repair and Engineering Centre, Institute of Orthopaedics, RNOH, Stanmore Campus, London HA7 4LP, UK
Received 8 July 2005, revised version received 18 October 2005, accepted 5 November 2005
Available online 6 December 2005
Abstract
The absence of a controllable in vitro model of soft tissue remodeling is a major impediment, limiting our understanding of collagen
pathologies, tissue repair and engineering. Using 3D fibroblast-collagen lattice model, we have quantified changes in matrix tension and
material properties following remodeling by blockade of cell-generated tension with cytochalasin D. This demonstrated a time-dependent
shortening of the collagen network, progressively stabilized into a built-in tension within the matrix. This was differentially enhanced by
TGFB1 and mechanical loading to give subtle control of the new, remodeled matrix material properties. Through this model, we have been
able to identify the Ftension remodeling_ process, by which cells control material properties in response to environmental factors.
D 2005 Elsevier Inc. All rights reserved.
Keywords: 3D collagen gel; Tissue bioreactor; Matrix tension; Cytochalasin D; TGFh1; Mechanical loading; Collagen remodeling
Introduction
Understanding of the remodeling process of adult
collagenous tissues (tendon, cartilage, skin, etc.) is as
critical to progress in tissue regeneration and surgery as it
is to engineering of tissues. Perhaps the most enigmatic
element of the Fremodeling_ process is how 3D tissue spatial
organization is produced in the first instance, maintained
and then replaced after injury (reviewed in [1,2]). The
central factor is that supra-molecular organization of
extracellular matrix (ECM) polymers dictates the material
properties of that ECM (and so its function in adults). In
most instances, fibrillar collagen is the critical, load-bearing
ECM polymer element. Although this function must operate
throughout postembryonic life, its real importance only
Abbreviations: ECM, extra cellular matrix; FPCL, fibroblasts populated
collagen lattice; CFM, culture force monitor; TGFh1, transforming growth
factor-beta 1; HDF, human derma fibroblasts; RTF, rat tendon fibroblasts;
CD, cytochalasin D; Fao, apparent total force output; Fc, cell contraction
force; Fm, fixed tension in the matrix; RMT, Residual Matrix Tension (at
either 2 or 12 hours); U, pseudo viscosity.
* Corresponding author. Fax: +44 20 8954 8560.
E-mail address: [email protected] (R.A. Brown).
0014-4827/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.yexcr.2005.11.005
becomes obvious in pathologies such as scar contracture,
fibrotic disease or Dupuytren’s contracture [3].
We and others have proposed the hypothesis that cells
generate tensile forces as part of the mechanism for
remodeling matrix material properties [1,3 –7]. While it is
not certain how fibroblasts spatially (re)organize tissuecollagen networks during growth or repair, it clearly happens
without interruption to the load-carrying function. The
central mechanism which now needs to be understood is
not only how new material is added but how that new material
is accommodated within the existing structure and shape (i.e.,
the spatial architecture) of the surrounding matrix. This is the
special element of interstitial growth and remodeling, which
distinguishes the biological from the engineering concepts of
growth and repair [8]. The latter, in fact, relies mainly on
material apposition or on whole unit replacement process
[9,10]. A human scale analogy of this process might be to
Fgrow_ the Eiffel tower by interstitially inserting an additional stage (Fig. 1A). If this were possible for a steel
structure, almost every surrounding girder would need to be
moved/lengthened, not once, but many times (i.e., a dynamic
process) in order to expand the structure and maintain spatial
arrangement. This is also true for remodeling associated with
424
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
Fig. 1. Interstitial remodeling of collagenous tissues. (A) Photographic composition using images of the Eiffel tower to illustrate how soft tissues grow (or
shrink) by Finterstitial_ remodeling (i.e., biological growth, lower panel). This is clearly only possible as a photographic trick and contrasts with the more
familiar conventional engineering process (engineering growth, upper panel) where material is added (or removed) at the extremities of the structure. (B)
Typical human dermal fibroblasts (HDFs) contraction profile with hypothetical and actual force drops after actin cytoskeletal disruption with cytochalasin D
(CD, arrowed). The solid line shows the actual contraction profile, the dotted curve shows the superimposed response actually reported to CD [22,23] and the
dashed line gives the hypothetical curve, predicted if resident cells had remodeled the collagen lattice to produce a Fshortened material_. Force components,
revealed by eliminating cell-generated force by cytochalasin D, can be expressed as: F ao (apparent total force output), F c (cell-generated force) and F m (fixed
matrix tension), with F ao = F c + F m. If F ao = 0 upon CD addition, i.e., upon F c being zeroed, then F m (dotted line) must be zero. On the contrary, if F ao <> 0
upon CD addition, then the remaining force must be F m <> 0.
growing soft tissues in adults during pregnancy, slimming/
obesity cycles and repair (reviewed in [3]). The paradox of
biological (in contrast to engineering) growth is compounded
by the absence of any clear understanding of how cells can
spatially shuffle apparently covalently cross-linked, fibrillar
collagen networks, without mass degradation and re-synthesis or loss of load-carrying function. Interestingly, this
process occurs in the presence of a background tension
across the tissue, maintained by cellular contraction and
connective tissue shortening [3,7,9,11,12]. This process is
clearly seen in pathologies such as scar and adhesion
contractures and Dupuytren’s disease [3,11,12] but has few
parallels outside cell physiology. The lack of a controllable in
vitro model of this process is a major limitation to our
understanding not only of how to engineer tissues but of cell
mechanics generally [2].
Physical shortening of the collagen network [3,11,12] has
been demonstrated in vivo in an experimentally lax
ligament. Surgically de-tensioning was followed over 3
weeks by spontaneous ECM re-tensioning [9]. Such
remodeling of high strength materials, with sparse cell
content, cannot be due to cell contraction alone, but to
physical shortening of the collagen network itself [3].
Guidry and Grinnell [6,14] first concluded that collagen
gel reorganization involves a physical rearrangement of preexisting collagen fibrils which occurs in a time-dependant
manner as cell-generated forces increase the proximity of
adjacent fibrils [13].
We report here the first example of a tissue-engineered
model of collagen tension-driven remodeling under tension,
based on the three-dimensional (3D), isometrically tethered,
fibroblast-populated collagen lattice (FPCL) model [14 – 17].
The FPCL model has been used extensively to study cell
force generation, responsive to growth factors and lattice
cell – matrix contraction and to study fibroblast behavior in
remodeling in 3D [3,18]. Over a short term (24 h), resident
fibroblasts generate a tensile force within the collagen,
physically shortening the FPCL between its anchor points
[19 – 24]. Disruption of the cell cytoskeletal motor with
cytochalasin [22,23] results in a complete loss of tension in
this model, indicating that there is no significant permanent
spatial remodeling of the collagen or physical shortening of
the ECM. Rather, gel contraction was entirely due to
temporary deformation of the collagen fibrillar network
operated by adhering and spreading cells.
The hypothesis under test here (proposed in [3]) is that the
tension generated across isometrically constrained FPCLs, by
the resident cells, would, in time, be stabilized as new matrix
architecture to give a physically shorter structure which is
also able to resist significant tensional force (i.e., a sort of
Ffixed tension_ within the remodeled matrix). Clearly,
repetition of this shortening would alter material properties
(elastic modulus, strength etc.) as well as geometry by the
previously postulated Fslip and ratchet mechanism_ [3]. To
test this hypothesis, we used the actin cytoskeleton disrupting
agent cytochalasin D to abolish cell-mediated contraction
[25], with real-time monitoring of the tension using a culture
force monitor (CFM). This force recording in the CFM was
linearly dependent on the displacement of one end of the
tethered FPCL from the other, thus the FPCL length (in which
the FPCL can sustain a given tension) and so its shortening
was monitored. The real-time force output of the CFM gave a
quantitative analysis of changing architecture (i.e., a measure
of the stabilized tension in the FPCL after cell force abolition)
under the effect of environmental factors. The in vitro
contracture model described here has been used to quantify
time-dependent physical shortening of the FPCL network, as
enhanced by TGFBI and mechanical loading. A critical
additional finding is the subtlety by which the cell –
biomechanical environment controls these new material
properties and the implications of this new understanding
for engineering of connective tissues.
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
Methods
Fibroblast culture and reagents
Human dermal fibroblasts (HDF) were prepared as
described previously [23]. Rat tendon fibroblasts (RTF) were
isolated from adult Achilles’ tendons by collagenase digestion or from explants. Briefly, tendon cubes (¨1 mm3) for
explants were stuck to petri dishes in Dulbecco’s Modified
Eagle’s Medium (DMEM) containing Penicillin/Streptomycin (100 U/ml and 100 Ag/ml, Gibco BRL, Paisley, Scotland)
and l-glutamine (2 mM, ICN, Biochemicals Ltd, Thyne, UK)
with 10% fetal calf serum (FCS: First Link, West Midlands,
UK) until cellular outgrowths had formed. Alternatively,
cubes were digested for 30 min in plain DMEM with 2 mg/ml
collagenase and the cell suspension re-plated in complete
(10% FCS) medium. Fibroblasts were used up to passage 8.
Human recombinant TGFh1 (PeprotechEC, UK and Sigma
Chemicals, Dorset, UK) was made up to concentrated stock
solutions in 0.1% bovine serum albumin (BSA: Sigma
Chemicals, UK) dissolved in 4 mM HCl and added to
cultures to give a final concentration of 15 ng/ml, known to
optimally stimulate fibroblast contraction [4].
Culture force monitor, TGFb1, mechanical loading and cell
force blockade
The culture force monitor (CFM), constructed and
calibrated as previously described [23,26,27], is an instrument capable of quantitatively measuring forces generated
by cells seeded into a rectangular 3D collagen gel, tethered
at its short edges. This force recording in the CFM is
linearly dependent on the displacement of one end edge of
the tethered gel from the other thus is linearly dependent on
the FPCL length at which the FPCL can sustain a given
tension (i.e., measure of shortening under tension). In brief,
5 ml of 2.28 mg/ml native acid soluble type I rat tail
collagen (First Link, West Midlands, UK) was mixed with
0.625 ml of 10 DMEM (Gibco BRL, Paisley, Scotland),
neutralizing with 1 M NaOH before addition of 1 ml
fibroblast suspension, to give 106 cells/ml [or 1 ml DMEM
only for controls]. [Note: the collagen used here was intact
native tropocollagen (acid soluble), rather than telopeptidefree, pepsin extracted collagen. The fibrilogenesis and
potential for telopeptide cross-linking sites were therefore
not impaired.] Where appropriate, stock TGFh1 or vehicle
was added to a final gel concentration of 15 ng/ml and the
mixture poured into T-shaped purpose-constructed Derlin
rectangular wells (Intertech Ltd., UK), with free floatation
bars at either end (comprising 5 layers of plastic mesh,
HeeBee Designs, UK) and allowed to set for 15 min, at
37-C, 5% CO2. Once set, gels were floated in 15 ml DMEM
with 10% FCS and, where appropriate, stock TGFh1 at a
final concentration of 15 ng/ml. The floating gel was
tethered through its floatation bars to a force transducer
(made of a cantilevered beam whose deflection is measured
425
by a strain gauge—Measurement Group, UK) at one end
and an anchor point at the other. Tensional forces through
the long axis of the gel were collected from the force
transducer to a PC (at a rate of one data points per second)
and were averaged into 10 min data points (i.e., 600
readings per point) for plotting the continuous force – time
output. Real-time graphical output, data storage and data
analysis used LabVIEW software (v. 6.01 National Instruments, USA). Mechanical loading of cultures [16,28] used a
tensioning CFM (tCFM), essentially the same as a CFM in
which the fix point was displaced, along the (long) axis of
gel tethering, by a computer-controlled stepper motor
(Parker, Germany) controlled through the PC. Cyclical
loading involved allowing the gel to contract for 8 h
followed by application of one loading cycle each hour,
comprising 15 min of loading (to 1% strain), 15 min no
movement, 15 min removal of the same load and 15 min no
movement (i.e., 16 cycles between 8 and 24 h).
At the end of a contraction/remodeling period (4, 18, 24,
61 h), cell-mediated force generation was abolished by
addition of a saturating dose of cytochalasin D (to disrupt
fibrillar actin) directly to the culture chamber [22,23,25].
Stock cytochalasin D (CD: Sigma Chemicals UK) in DMSO
was added directly to culture wells to a final concentration
of 20 Ag/ml. [Note: human fibroblast contraction was
completely blocked, as previously reported, by 2 Ag/ml
CD, but control studies here established that a 10-fold
higher dose was needed to block RTF contraction.] An
alternative was to block contraction by hypo-osmotic cell
lysis by removal of most of the culture medium from the
chamber and replacement with an equal volume of sterile,
distilled water and the process repeated.
Residual matrix tension (RMT) and pseudo-viscosity
measures
Residual matrix tension (RMT) was determined from the
CFM output following addition of cytochalasin D (CD, 20
Ag/ml). In a typical HDF contraction profile (Fig. 1B),
apparent total force output ( F ao) at 24 h was postulated to
comprise the force due to cell contraction ( F c) plus any
contribution due to force of Ffixed tension in the matrix_
( F m) due to collagen remodeling by the fibroblasts, giving
the formula:
Fao ¼ Fc þ Fm :
Since the culture tension for HDFs was returned to
baseline levels by CD, remodeling could not have produced
any significant stable fixed tension (i.e., F ao = F c, F m = 0),
and this was used as the null control (Fig. 1B). However,
where cultures produced a clear F m, there remained a degree
of uncertainty as to its exact value as the fall off in force, postCD, was close to exponential due to diffusion-time (minute
timescale) and the stress – relaxation dynamics displayed by
the material (hour timescale). As a result, a standard cut-off
point of 2 h post-CD was taken as the RMT value and termed
426
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
Collagen gels were fixed under tension in 2.5%
glutaraldehyde in 0.1 M phosphate buffer, pH 7.5 for 1 h
at 4-C, washed twice in buffer only, with secondary fixation
in 1% osmium tetroxide (Agar Scientific Ltd., UK) in 0.1 M
sodium cacodylate buffer, pH 7.4 (1 h, room temperature)
followed by washing in 0.1 M sodium cacodylate buffer.
Full thickness specimens of collagen lattice (approximately
10 mm 5 mm) were snap-frozen in liquid nitrogen for 2
min, fractured and placed onto a carbon adhesive disc (Agar,
UK). Degassed specimens (Joel JJM 5500LV scanning
electron microscope in low vacuum mode) were sputter
coated with gold/palladium (60/40) for 2 min (Emitech
K550 coater) and examined in the same scanning electron
microscope (high vacuum mode).
which cells remodel a collagen matrix [3,19 –23,25]. In our
previous studies, force generation by human dermal
fibroblasts in tethered FPCLs was monitored over a 24 h
period (Fig. 1B). Two phases of force generation were
identified: (I) cell traction and (II) cell contraction. It is
notable that after 24 h this cell type (human dermal
fibroblasts—HDFs) produced minimal remodeling or
Ffixing_ of the collagen into a physically shorter material
(i.e., the force due to fixed tension in the matrix F m = 0).
This is shown by the near complete loss of tension
following addition of CD, which blocks cell contraction
by disrupting F-actin formation (Fig. 1B).
The presence of progressive fixing of matrix tension (due
to collagen network shortening) would add a 3rd phase
(where it occurred) and would result in only partial loss of
tension after addition of CD (dashed prediction plot and FF c_
in Fig. 1B). Hypothesizing that force components are
linearly additive, it is also predicted that, when this 3rd
phase, i.e., stable spatial remodeling, did occur, there would
be an increase in the apparent total force output ( F ao: i.e.,
prior to CD addition) as the new fixed matrix tension F m
would be added to the force due to cell contraction ( F c).
Under these conditions, if F c is assumed to be constant, the
change in apparent force (DF ao) will equal the DF m.
A qualitative measure of the degree of HDF-mediated
spatial reconfiguration of the collagen architecture over 24
h is shown in Fig. 2. Untethered (multi-vector) loading gave
a random fibril alignment in contrast to parallel-aligned
fibrils in uniaxially tethered gels. Yet, the quantitative
response to CD (Fig. 1B) (measured as tension across the
tethered FPCLs) shows that this was not a stable remodeling
as the inbuilt tension by the cells dropped to basal precontraction level.
Results
Identification of collagen network shortening with rat
tendon fibroblast: time dependency
RMT2 (>85% of the total force fall by 2 h). Further falls in
force after 2 h (RMT12) were regarded as pure material
stress – relaxation. Since the material relaxation between 2
and 12 h was near linear, a parameter representing the
dynamic resistance of the material to sudden load was
developed. This parameter, defined as pseudo-viscosity
(q 2 – 12 h), was obtained from the ratio of total force fall
between 2 and 12 h (DF 2 – 12 h) and the average velocity of
dropping force in the same timeframe (v df(2 – 12 h)). Hence, the
expression:
q212 h ¼ DF212 h=vdf ð212
hÞ:
Despite non-linearity in the 0 – 2 h timeframe, the
pseudo-viscosity could also be calculated in that timeframe
(defined as q 0 – 2h) to test its consistency as a timeindependent material parameter.
Scanning electron microscopy
Contracture model formulation using human dermal
fibroblasts FPCLs
Contraction of the fibroblast-populated collagen lattice
(FPLC) is widely regarded as a model of the process by
In contrast to HDFs, the same 24 h contraction profile for
rat tendon fibroblasts (RTFs) resulted in a substantial
Fresidual tension_ after CD treatment (Fig. 3A). Identification of this remodeling effect allowed us to study the
process of physical collagen network shortening in vitro.
Fig. 2. (A) Scanning electron micrograph of 24 h FPCLs contracted under uniaxial isometric tension (tensional axis is across the page). Arrowheads point
collagen fibrils, arrows indicate show fibroblasts. (B) As panel A but contracted without tethering (no tension).
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
427
Fig. 3. Identification of the residual matrix tension. (A) Force – time plot showing the contraction profile for rat tendon fibroblasts (RTF) with its plateau period
(4 – 6 h) and linear second rise in force from 9 to 24 h. At 24 h, cytochalasin D (CD) was added, producing the fall off in matrix tension (dotted bars show the
fall in force due to loss of cell contraction and the non-cell-based residual matrix tension—RMT). Force was monitored for a further 12 h, allowing
measurement of RMT at 2 and 12 h post-CD treatment. (B) Force – time plot showing the addition of the first dose of cytochalasin D – as in panel A – (after 22
h) followed by a second identical dose (to confirm CD dose saturation) and subsequent cell lysis with distilled water. RMT remained stable over the 12
h following these treatments. The remodeled lattice was then loaded with an additional 1% strain, equivalent to 0.5 mN. Similar RTM levels were seen using
hypo-osmotic lysis alone. (C) To follow functional matrix remodeling over a time course (4 to 61 h), RTM was measured 2 h (RMT2) after addition of CD.
RMT2 increased in a near linear fashion with culture time (4 to 61 h, filled symbols). The corresponding peak forces achieved (just before CD addition, open
symbols) had a different trend over time with a reduced rate of increase. The dashed line represents the basal contraction force of cell-free collagen lattice over
time. Error bars represent T SE (n = 3 separate constructs). The 4 h point RMT2 was negligible and not different to cell-free lattices, but there was a ¨2-fold
increase between 24 and 61 h ( P < 0.005).
Many features of the RTF contraction are similar to HDFs
including a correlation of phase I, cell traction force with
cell spreading (not shown). However, there were key
differences, including a more rapid attainment of the force
plateau (after 4 h for RTF and 8 h for HDF). In addition, the
RTF plateau was sustained for a shorter period, giving way
to a linear increase in force generation from 8 to 24
h (increase of 54.9%; to a mean peak force of 1.32 mN, n =
7) rather than the typical sustained force plateau for HDFs.
Addition of CD to rat cells at 24 h (Fig. 3A) identified a
fixed tension (shortening), measured here as a Fresidual
matrix tension_ (RMT2: i.e., 2 h post-CD treatment) of 0.47 T
0.02 mN (n = 7). This represented 35.6% of total apparent
force, pre-CD ( F ao), clear evidence of mechanically
functional cell-mediated collagen remodeling (shortening),
not seen in cell-free collagen gels.
The method for determining RMT and its stability was
validated through extended tests (Fig. 3B). Addition of a
second dose of CD, 2 h after the first, demonstrated that
blockade of cell force generation was saturated, and no
further reduction in RMT ( F m) could be produced.
Furthermore, lysis of the resident cells with hypo-osmotic
medium (Fig. 3B) also had no additional effect on RMT.
Hypo-osmotic cell lysis alone was used to confirm the
presence of RMT2 without use of CD (data not shown). At
this stage, the remodeled matrix was mechanically stable
as judged by (i) the minimal fall in tension over 13 h and
(ii) its ability to withstand an additional 0.5 mN of external
loading (generated by 1% external strain, Fig. 3B). The
effect of culture period on the development of fixed
tension was tested by progressively increasing the duration
of incubation prior to CD addition (Fig. 3C). RMT2
increased in a near linear manner from the basal level at 4
h stage to 24 h and 61 h. Differences between time points
were highly significant (n = 4, P < 0.005) with almost a 2fold increase in RMT2 between 24 and 61 h, while total
force generated ( F ao) increased by less than 20% over the
same period.
TGFb1 or mechanical loading and residual matrix tension
RMT2
Treatment of RTF cultures with TGFh1 (at a concentration known to stimulate contraction) also altered matrix
428
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
remodeling, as RMT2 (Fig. 4A). In the presence of 10%
serum, TGFh1 treatment did not affect contraction up to
4 h but then shortened the plateau stage (from 4 to 2 h)
and increased the rate of force generation almost 4-fold
thereafter (from 0.05 to 0.19 mN/h) up to 3.12 T 0.24
mN by 24 h. The mean RMT2 of 0.72 T 0.10 mN was
54.1% higher than untreated cultures ( P < 0.05, n = 7),
though this represented a much smaller proportion
(23.2%) of the total force prior to CD ( F ao) than for
cultures without TGFh1. Clearly, then, the pattern and
magnitude of tension fixation (remodeling) were altered
by this mechano-active growth factor. A second treatment
mode was to apply external uniaxial cyclical loading (Fig.
4B). This slow rate, low strain loading (1% strain at 1
cycle/h) applied between 8 and 24 h increased the RMT2
1.58-fold or 58.2% above the untreated RMT2 ( P < 5 104, n = 3). Interestingly, mechanical loading appeared
to produce a more stable remodeling as the RMT2 was a
larger fraction (54.2%) of the total force output than for
control (35.6%) or TGFh1 (23.2%) treatments. While
both treatments increased the RMT2 significantly over
controls, the difference between treatments was not
significant (Fig. 4C).
Viscoelastic and material properties of remodeled FPCL
When the cell force element ( F c) was abolished with
CD, the recoil of the CFM sprung force transducer beam
applied an equivalent reaction force, equal to F c, onto the
collagen gel. The greater the final cell-generated force, the
greater would be the sudden load transfer to the newly
remodeled collagen matrix. It is inevitable that such overloading would result in some rupture of new interfibril
bonds in a manner proportional to the loading ( F c) and the
mechanical properties of the remodeled collagen. The vast
majority (>85%) of the total force drop occurred over the 0
to 2 h period (RMT2). However, the continuing fall in
tension over 2 to 12 h was an important indicator of stress –
relaxation behavior in the remodeled material (Fig. 4D).
Average stress – relaxation curves for the untreated and
treated lattices (TGFh1 or mechanically loaded) illustrated
how the initial force ( F ao) affected the total force drop and
the rate of fall (Fig. 4D). Matrix tension fell non-linearly
over the first 2 h post-CD (mean linear rate of fall ranging
from 0.36 mN/h and 0.43 mN/h, for loaded and untreated
respectively, to 1.18 mN/h for TGFh1). This was followed
(2 – 12 h stage) by a near linear slow relaxation (R 2 > 0.80
Fig. 4. Effects of culture treatments on tension remodeling. Culture treatments (TGFh1, cyclic loading) influenced the level of RMT seen after CD treatment.
Panel A shows the rat tendon fibroblast contraction with/without 15 ng/ml TGFh1 (time zero). CD was given at 24 h and the drop in force monitored over the
following 2 h. The difference in RTM2 T TGFh1 is indicated as Fd_. Error bars represent the SEM (n = 7 separate contractions) displayed at 2 h intervals, for
clarity. (B) Cyclic loading (1% strain at 1 cycle/h) was applied to cultures from 8 to 24 h. CD was then given at 24 h and the drop in force monitored over the
following 2 h. The difference in RTM2 T loading is indicated as Fd_. Error bars represent the SEM (n = 3 – 7). (C) Summary chart comparing mean RMT2 in
TGFB1-treated, loaded and untreated (Control) cultures. Both treatments yielded a significant increase in RMT2 (#P < 0.05 and *P < 5 104). (D) Typical
force fall off and stress – relaxation profiles started immediately after CD addition, comparing control, TGFh1 and loaded responses, showing the exponential
fall over the first 2 h and the slow extension between 2 and 12 h. Note the more rapid 2 – 12 h fall with TGFh1 treatment.
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
for the trend lines). Mean rates of fall were 1.13 102,
2.17 102 and 4.26 102 mN/h for untreated, loaded
and TGFh1-treated respectively (that is, TGFh1 produced
2- to 3-fold greater relaxation). There was a highly
significant linear correlation between F ao (total force, preCD) and RMT2 (R 2 = 0.98). However, this correlation
largely disappeared over the following 2 –12 h period, i.e.,
for the RMT12 (R 2 = 0.53). This highlights the underestimation of fixed tension and its dependence on total force,
pre-CD (that is, RMT2 is the fixed tension which survives
sudden stress application). This analysis emphasizes that
material properties (elastic modulus, shear modulus, break
strength) of the cell-remodeled collagen will depend on the
local cell environment, including growth factors and
mechanical configuration of collagen fibrillar network
(e.g., tension field intensity and direction, fibers density
and alignment). Since RMT2 appeared to underestimate the
full extent of the remodeling, a further level of analysis was
needed.
Information on the material properties of the remodeled
collagen matrix could be derived from the RMT12 stress –
relaxation behavior. The mean RMT12 showed that the
429
newly remodeled matrices following cyclical loading were
able to withstand a significantly higher tension (¨50%)
compared to untreated and TGFh1-treated (TGFh1 RMT12
was actually lower than controls: Fig. 5A). The pseudoviscosity for 2 –12 h stage, q 2 – 12 (Fig. 5B), identified the
same differences between culture treatment modes as
RMT12, showing pseudo-viscosity significantly increased
(2-fold) by cyclic loading, relative to TGFh1 treatment. The
reduction in pseudo-viscosity with TGFh1 compared to
control cultures was clearest over the 2 –12 h period, when
control and loaded values were most similar (Fig. 5B). The
similar trend for pseudo-viscosity q 0 – 2, over 0 –2 h (Fig.
5C) as q 2 – 12, indicated that this material parameter was
consistent across the timeframe observed (while RMT
parameter represented a snapshot of the state of the
remodeled matrix). These parameters (RMT and q) were
far more sensitive to collagen material remodeling changes
than the more conventional measure of elastic modulus (Fig.
5D) which did not identify any significant differences
between TGFh1, loaded cultures and untreated controls at
12 h post-CD treatment. However, the elastic modulus was
significantly higher in all culture conditions (i.e., cell-
Fig. 5. RMT dynamics and material properties of the remodeled collagen lattices. (A) Changes in RMT12 for TGFh1, loaded and control cultures. Bars indicate
the SEM (n = 4). *Indicates significant difference (*P < 0.05) between loading treatment and both TGFh1-treated and untreated (control). (B) Variation of
pseudo-viscosity q in the 2 – 12 h timeframe (q 2 – 12) for the 3 treatments. Bars indicate the SEM (n = 4). Mechanical loading produced a significantly greater
q 2 – 12 than TGFh1 treatment (^P < 0.01) and untreated (Control, #P < 0.05). q 2 – 12 for TGFh1-treated was even lower than control (*P < 0.05). (C) Variation
of pseudo-viscosity q in the 2 – 12 h timeframe (q 0 – 2) for the 3 treatments. Bars indicate the SEM (n = 4). q 0 – 2 of loaded gels was significantly greater than
q 0 – 2 for TGFh1-treated and untreated controls (#P < 0.05). Again, q 0 – 2 of TGFh1-treated FPCLs was significantly lower than q 0 – 2 for controls (*P < 0.05).
Panel D shows the Young modulus for 3 treatments and cell-free lattices. Cell-seeded gels in all culture condition, treated with cytochalasin D for 12 h,
presented a significantly higher modulus than cell-free gels. The modulus fell after CD elimination of cell force in all culture conditions, but the difference was
statistically significant only for TGFh1-treated gels. There was no significant difference between the 3 treatments, after CD (12 h post-CD addition). Bars
indicate the SEM (n = 4). *Indicates significant differences in respect to the ‘‘Gel-Only’’ group (*P < 0.05) and # between TGFh1-treated (pre-CD) and all
other culture conditions (#P < 0.005).
430
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
remodeled gels) compared to cell-free gels (non-remodeled
gels).
Discussion
It is difficult to overstate the importance of dynamic
spatial control of tissue architecture to normal and repair
function in almost all mammalian tissues. This can be
illustrated through the difference between normal, functionally adaptive tissues and dysfunctional non-adaptive scar
tissues. Repaired tissues (scars) typically do not regain the
native 3D architecture of the original tissue [29]. [Note: all
tissues can scar where repair, as opposed to regeneration,
occurs.]
Although regulation of dynamic 3D collagen architecture
is far less well understood than compositional biochemistry,
there is a longstanding thread of investigation into the
problem and the concepts of interfibrillar slip [6,10,12
13,30,31]. Recent studies have considered the possibility of
molecular mediators, such as decorin [32], types V and XIV
collagen [33,34], proteoglycan [35] and collagen molecular
orientation [1,18]. In a pivotal work, Glimcher and Peabody
[12] argued the key importance of the process by which
cells of a tissue are able to remodel the dimensions and
material properties of a collagen network. In particular, they
focused on the mechanisms by which fibroblasts are able to
produce the stable progressive shortening of collagenous
networks seen in Dupuytren’s disease and scar contracture.
The use of cytochalasin [13,20] to measure the matrix
shortening process of cell remodeling has now made it
possible to produce a quantifiable cell culture model of the
process.
The current findings together with our past reports
[4,15,23,27] suggest that the different phases of force
generation are responsible for different phases of remodeling. The initial phase, following cell-seeding, correlates with
cell spreading and force generation by traction [3,5,19,26],
and it may be important that this was much shorter for RTFs
(4 h) than for HDFs (8 h) [26]. The subsequent phase, in
which cell force output flattens to a relatively constant
plateau level, has been considered a contraction phase,
representing a form of tensional homeostasis [15]. Again,
this phase seemed to be shortened in RTFs where the
equilibrium force was maintained for only 4 h. This pattern
suggests that HDFs and RTFs differ in their rates of collagen
remodeling (i.e., RTFs > HDFs) but that the process is
basically the same (it would be surprising if human
fibroblasts were more active than young rat). Indeed, under
suitable conditions, production of RMT has recently been
identified with HDF and bovine tendon fibroblasts (Beckett,
Mudera, Marenzana and Brown, in preparation), indicating
the process is neither rat- nor tendon-specific. Proof that this
seemingly fundamental process is not cell-specific is beyond
the scope of the study, but knowledge of tissue specific
variants would be of great importance. For example, it
would be plausible that cells from different tissues/states
(e.g., blood vessel wall, nerve sheath, normal, scleroderma
and Dupuytren’s dermis or fascia) do have substantially
different responses to TGFh1 or mechanical signals in terms
of generating RMT. Distinct responses to combinations
would seem particularly likely, resulting either in typical
adaptive differences between normal tissues or completely
fresh mechanisms to understand connective tissue pathologies. It is clear though that the remodeling process in RTFs
operates at a convenient rate for the experimental model
used here.
Three factors have been identified here as regulators of
collagen remodeling: time in culture, TGFh1 treatment and
low magnitude, uniaxial cyclic loading. The strong time
dependency suggests that the process is rate-limited,
potentially at a number of points (consistent with the
differences between cell types). The time dependency may
also be a function of a need for synthesis of new collagen
which can be slow in culture [3]. This is based on the
theoretical requirement for small amounts of new collagen
to link existing fibrils, predicted for dynamic spatial
reorganization [3].
The cyclical loading, applied here, was selected as an
example of a carefully characterized and previously defined
pattern rather than as a means to simulate any particular
natural tissue loading. The system has been previously
established as a means of stimulating resident cell mechanoresponses within the highly compliant collagen gel
[15,16,27]. Although the forces applied and the rate of
cycling (rate of strain) are unusually low for natural tissue
loading, these must be appropriate to the scaffold in which
the cells sit. The collagen gel scaffold used here provides
minimal stress shielding, compared to that of native tissue,
such that cells undergo substantial deformation (actual
strain). These have been shown previously to elicit
biochemical and motility responses [16,27], potentially
modeling supra-physiological cell loading in native connective tissues.
Both TGFh1 and cyclic loading were found to increase
the RMT. Each of these treatments was selected as
representative examples of environmental factors known
to modify the collagen network in vivo and in vitro. TGFh1
promotes synthesis and accumulation of fibrillar collagen
(and other matrix elements) while decreasing overall MMP
synthesis [36 – 38] but also increases force generation and
modifies fibroblast attachment in early collagen lattice
culture [4,25,39]. Cyclical tensile loading promotes collagen
synthesis and matrix alignment and alters MMP and TIMP
expression [16,27,28,40]. Importantly, the increase in stable
remodeling/RMT in response to cyclical loading did not
seem to substantially increase force generated by cells, in
contrast to TGFh1. This suggests the possibility of more
than one mechanism, with fibroblasts responding to multielement local cues. The hypothesis that forces are simply
additive (i.e., an increase in F m necessarily implies an
increase in F ao, hypothesizing F c constant) is complicated
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
by the finding that mechanical loading of gels increased the
F m but not (significantly) the F ao. On the other hand, the
fact that F m did not proportionally follow the large increase
in F ao with TGFh1 may be partly due to underestimation of
RMT2 (as mentioned above) and partly a function of the
different times taken for the growth factor to affect cell
motor only, F c, and matrix remodeling. Identification of this
additional complexity was an unexpected benefit of using
TGFh1 as an exemplar environmental variable in that it
highlights potential mechanisms (in principle) for subtle
local cellular control of matrix mechanics. Such a complex
control system would help to explain the local variations in
collagen matrix properties seen in vivo as well as in this
simple FPCL model. Clearly, local cell-mechanical cues,
simplified by the simple collagen-only-based substrate, will
be further complicated by other (tissue type) connective
tissue components, such as elastin or proteoglycans.
The stress – relaxation phenomenon is known and
expected in viscoelastic materials such the collagen lattice
following sudden load steps. However, the relaxation seen
here is clearly distinct from the standard stress –relaxation
response (as shown by studies on cell-free lattices in our
laboratory; data not shown). It took 2 h to reach even a nearsteady level (continuing beyond 12 h) after 24 h remodeling,
as opposed a few minutes for standard stress relaxation. This
is consistent with the idea that sudden loss of cell force after
addition of CD leads to a complex pattern of rupture of
heterogeneous (newly formed) interfibrillar bonds, collapsing gradually at different times depending on their strength
and position.
Cell proliferation is unlikely to play a major role here in
the changes in matrix remodeling, not least because
proliferation in collagen lattice cultures is widely regarded
as low [41,42], particularly over the 24 h time course used
here [17,19 – 23,25]. Certainly the rate, scale and time of
onset of TGFh1 and cyclic loading responses, seen here and
in previous studies (i.e., after only a few minutes or hours),
do not correlate with long-term changes in cell number
characteristic of FPCLs. In addition, total FPCL force
generation (known to be cell density dependent) did not
correlate with RMT in this study.
Wakatsuki et al. (2000) identified changes in collagen
matrix elastic modulus following FPCL contraction and
treatment with cytochalasin. Consistent with their work, all
cell-seeded gels (in all culture conditions) post-CD treatment showed a significantly higher modulus than cell-free
gels (Fig. 5D). Thus, both F m and elastic modulus of the
matrix alone (post-CD treatment) were consistently increased by the cell remodeling activity. This was in addition
to changes in stiffness due to the contribution of cell
cytoskeleton, lost with cytochalasin treatment. Reinterpretation of their findings in the light of the present model
suggests that at least some of material changes described in
their work were a result of the tension-driven collagen
remodeling, i.e., RMT. To our knowledge, this is the first
description of the mechanisms underlying this effect,
431
although the cell-independent contraction (in the shapechange FPCL model) reported by Grinnell and Ho [25] was
almost certainly a result of the same process. Clear
identification of the nature and characteristics of the effect,
or its consequences for new material properties, was not
possible without the real-time quantitative measure of force
generation provided by the CFM. In turn, this has led to the
identification of a surprising level of subtlety in the ability
of fibroblasts to produce different matrix material properties
in response to combinations of environmental signals, such
signals in this case signals were mechanical loading and
TGFh1.
Previously, the tendency has been to assume that local or
anatomical differences in material properties (e.g., skin,
tendon, fascia) were due to local fibroblast sub-populations
or phenotypic shifts. However, these findings suggest a new
approach to understanding heterogeneities in tissue collagen
properties. In this simple model, one pattern of loading and
a single, exemplar concentration of one TGFh isoform
elicited changes in remodeling and quality of the remodeled
collagen matrices. It seems reasonable, then, that complex
local combinations of such micro-environmental factors
could largely explain local heterogeneity of tissue properties
in vivo. The cell mechanisms by which these differential
controls operate are presently uncertain, but it is clear that
the two factors used here operate at least partially
independently. Indeed, preliminary results (not shown)
suggest that combinations of TGFh1 plus cyclic loading
result in material properties dominated by the TGFh1 rather
than an average of the two (in preparation). It seems
inevitable from previous work that both growth factor and
mechanics will alter cell enzyme activity or collagen
secretion to affect matrix composition [36,37,40]. It is
equally plausible that they influence the processes of cell
motility and force generation, necessary for supra-molecular
collagen fibril assembly and so material properties
[4,16,38]. One such indirect route might be through
increasing cell force generation (seen here with TGFh1)
and so selection of particular interfibrillar bonding by
rupture of weaker bonds at an early stage. A possible role
for other structurally important ECM components, such as
glycosaminoglycans, within the collagen might be predicted
from a study using a comparable monitoring system [43].
However, the collagen substrate described in that paper was
dense, macroporous and insoluble, making direct comparison of cyto-mechanical responses difficult without further
analysis. The stabilization mechanism reported here is
consistent with the pioneering concept that stabilization of
collagen fibrils by cell-mediated remodeling is based only
on the force intensity and force application time [13]. It is
now possible to extend the concept to a more complex,
multivariate mechanism for mechano-regulation of matrix
material properties.
This culture model of 3D collagen matrix remodeling
opens a new window on how fibroblasts may produce
subtly distinct local matrix properties. It also presents a
432
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
novel approach to understanding and engineering of
connective tissues [1] with far-reaching implications for
rebuilding connective tissues. Limited understanding of
collagen adaptive remodeling has been a key obstacle to a
range of problems from micro-gravity changes and growth
defects to bioreactor tissue engineering [1,2]. The model
here provides the means to tackle such questions. Notably,
as a 3D in vitro model, it is easily reproducible, defined
and quantifiable, simplifying the overwhelming complexity
of in vivo models where both tissues and systemic
processes are interconnected.
Acknowledgments
Work was funded by the European Commission (framework 5 FBITES_ programme) and Biotechnology and
Biological Sciences Research Council (UK). We are grateful
to Mike Kayser for his expert assistance with electron
microscopy.
Contract grant sponsors: European Commission and
Biotechnology and Biological Sciences Research Council
(UK). Contract grant number: EU framework 5, FBITES_
program.
References
[1] R.A. Brown, Tissue engineering: clinical applications and mechanical
control, in: J.M. Polak, L.L. Hench, P. Kemp (Eds.), Future Strategies
for Tissue and Organ Replacement, Imperial College Press, London,
2002.
[2] R.A. Brown, Bioartificial implants: design and tissue engineering, in:
M. Elices (Ed.), Structural Biological Material. Design and Structure –
Property Relationships, Pergamon, Amsterdam, 2000.
[3] J.J. Tomasek, G. Gabbiani, B. Hinz, C. Chaponnier, R.A. Brown,
Myofibroblasts and mechano-regulation of connective tissue remodelling, Nat. Rev., Mol. Cell Biol. 3 (2002) 349 – 363.
[4] R.A. Brown, K.K. Sethi, I. Gwanmesia, D. Raemdonck, M. Eastwood,
V. Mudera, Enhanced fibroblast contraction of 3D collagen lattices
and integrin expression by TGF-beta1 and -beta3: mechanoregulatory
growth factors? Exp. Cell Res. 274 (2002) 310 – 322.
[5] R.T. Tranquillo, Self-organization of tissue-equivalents: the nature and
role of contact guidance, Biochem. Soc. Symp. 65 (1999) 27 – 42.
[6] C. Guidry, F. Grinnell, Studies on the mechanism of hydrated collagen
gel reorganization by human skin fibroblasts, J. Cell Sci. 79 (1985)
67 – 81.
[7] I.B. Bischofs, U.S. Schwarz, Cell organization in soft media due to
active mechanosensing, Proc. Natl. Acad. Sci. U. S. A. 100 (2003)
9274 – 9279.
[8] P. Muller, L.E. Dahners, A study of ligamentous growth, Clin. Orthop.
Relat. Res. 229 (1988) 274 – 277.
[9] A.L. Wallace, R.M. Hollinshead, C.B. Frank, Creep behavior of a
rabbit model of ligament laxity after electrothermal shrinkage in vivo,
Am. J. Sports Med. 30 (2002) 98 – 102.
[10] R.A. Brown, P.D. Byers, Swelling of cartilage and expansion of the
collagen network, Calcif. Tissue Int. 45 (1989) 260 – 261.
[11] M.J. Glimcher, The biochemistry, structure and macromolecular
organisation of collagen in Dupuytren’s disease, in: A.H. Kang, M.E.
Nimni (Eds.), Collagen; volume V, Pathobiology, CRC Press, Boca
Ratton, 1992.
[12] M.J. Glimcher, H.M. Peabody, Collagen organisation, in: R. McFarlane, D.A. McGrouther, M.H. Flint (Eds.), Dupuytrens disease,
Churchill Livingstone, Edinburgh, 1990.
[13] C. Guidry, F. Grinnell, Contraction of hydrated collagen gels by
fibroblasts: evidence for two mechanisms by which collagen fibrils are
stabilised, Collagen Relat. Res. 6 (1986) 515 – 529.
[14] F. Grinnell, C.R. Lamke, Reorganization of hydrated collagen lattices
by human skin fibroblasts, J. Cell Sci. 66 (1984) 51 – 63.
[15] R.A. Brown, R. Prajapati, D.A. McGrouther, I.V. Yannas, M.
Eastwood, Tensional homeostasis in dermal fibroblasts: mechanical
responses to mechanical loading in three-dimensional substrates,
J. Cell. Physiol. 175 (1998) 323 – 332.
[16] V.C. Mudera, R. Pleass, M. Eastwood, R. Tarnuzzer, G. Schultz, P.
Khaw, D.A. McGrouther, R.A. Brown, Molecular responses of human
dermal fibroblasts to dual cues: contact guidance and mechanical load,
Cell Motil. Cytoskeleton 45 (2000) 1 – 9.
[17] T. Wakatsuki, M.S. Kolodney, G.I. Zahalak, E.L. Elson, Cell
mechanics studied by a reconstituted model tissue, Biophys. J. 79
(2000) 2353 – 2368.
[18] E. Tamariz, F. Grinnell, Modulation of fibroblast morphology and
adhesion during collagen matrix remodeling, Mol. Biol. Cell 13
(2002) 3915 – 3929.
[19] A.K. Harris, D. Stopak, P. Wild, Fibroblast traction as a mechanism for
collagen morphogenesis, Nature 290 (1981) 249 – 251.
[20] T. Elsdale, J. Bard, Collagen substrata for studies on cell behavior,
J. Cell Biol. 54 (1972) 626 – 637.
[21] P. Delvoye, P. Wiliquet, J.L. Leveque, B.V. Nusgens, C.M. Lapiere,
Measurement of mechanical forces generated by skin fibroblasts
embedded in a three-dimensional collagen gel, J. Invest. Dermatol.
97 (1991) 898 – 902.
[22] M.S. Kolodney, R.B. Wysolmerski, Isometric contraction by fibroblasts and endothelial cells in tissue culture: a quantitative study, J. Cell
Biol. 117 (1992) 73 – 82.
[23] M. Eastwood, D.A. McGrouther, R.A. Brown, A culture force monitor
for measurement of contraction forces generated in human dermal
fibroblast cultures: evidence for cell – matrix mechanical signalling,
Biochim. Biophys. Acta 1201 (1994) 186 – 192.
[24] P. Roy, W.M. Petroll, H.D. Cavanagh, C.J. Chuong, J.V. Jester, An in
vitro force measurement assay to study the early mechanical
interaction between corneal fibroblasts and collagen matrix, Exp. Cell
Res. 232 (1997) 106 – 117.
[25] F. Grinnell, C.H. Ho, Transforming growth factor beta stimulates
fibroblast-collagen matrix contraction by different mechanisms in
mechanically loaded and unloaded matrices, Exp. Cell Res. 273
(2002) 248 – 255.
[26] G. Talas, T.S. Adams, M. Eastwood, G. Rubio, R.A. Brown,
Phenytoin reduces the contraction of recessive dystrophic epidermolysis bullosa fibroblast populated collagen gels, Int. J. Biochem. Cell
Biol. 29 (1997) 261 – 270.
[27] M. Eastwood, V.C. Mudera, D.A. McGrouther, R.A. Brown, Effect of
precise mechanical loading on fibroblast populated collagen lattices:
morphological changes, Cell Motil. Cytoskeleton 40 (1998) 13 – 21.
[28] D. Kessler, S. Dethlefsen, I. Haase, M. Plomann, F. Hirche, T. Krieg,
B. Eckes, Fibroblasts in mechanically stressed collagen lattices assume
a ‘‘synthetic’’ phenotype, J. Biol. Chem. 276 (2001) 36575 – 36585.
[29] D.S. Jackson, The dermal scar, in: M.I.V. Jayson, J.B. Weiss (Eds.),
Collagen in Health and Disease, Churchill Livingston, Edinburgh,
1982.
[30] T. Nemetschek, H. Riedl, R. Jonak, H. Nemetschek-Gansler, J.
Bordas, M.H. Koch, V. Schilling, Functional properties of parallel
fibred connective tissue with special regard to viscoelasticity (author’s
transl.), Virchows Arch. A Pathol. Anat. Histol. 386 (1980) 125 – 151.
[31] J.F. Woessner, in: M.I.V. Jayson, J.B. Weiss (Eds.), Collagen in Health
and Disease, Churchill Livingston, Edinburgh, 1982.
[32] K.G. Danielson, H. Baribault, D.F. Holmes, H. Graham, K.E. Kadler,
R.V. Iozzo, Targeted disruption of decorin leads to abnormal collagen
fibril morphology and skin fragility, J. Cell Biol. 136 (1997) 729 – 743.
M. Marenzana et al. / Experimental Cell Research 312 (2006) 423 – 433
[33] K.E. Kypreos, D. Birk, V. Trinkaus-Randall, D.J. Hartmann, G.E.
Sonenshein, Type V collagen regulates the assembly of collagen fibrils
in cultures of bovine vascular smooth muscle cells, J. Cell. Biochem.
80 (2000) 146 – 155.
[34] B.B. Young, M.K. Gordon, D.E. Birk, Expression of type XIV
collagen in developing chicken tendons: association with assembly
and growth of collagen fibrils, Dev. Dyn. 217 (2000) 430 – 439.
[35] H.K. Graham, D.F. Holmes, R.B. Watson, K.E. Kadler, Identification of collagen fibril fusion during vertebrate tendon morphogenesis. The process relies on unipolar fibrils and is regulated
by collagen – proteoglycan interaction, J. Mol. Biol. 295 (2000)
891 – 902.
[36] A.B. Roberts, M.B. Sporn, R.K. Assoian, J.M. Smith, N.S. Roche,
L.M. Wakefield, U.I. Heine, L.A. Liotta, V. Falanga, J.H. Kehrl, et al.,
Transforming growth factor type beta: rapid induction of fibrosis and
angiogenesis in vivo and stimulation of collagen formation in vitro,
Proc. Natl. Acad. Sci. U. S. A. 83 (1986) 4167 – 4171.
[37] O. Eickelberg, A. Pansky, R. Mussmann, M. Bihl, M. Tamm, P.
Hildebrand, A.P. Perruchoud, M. Roth, Transforming growth factorbeta1 induces interleukin-6 expression via activating protein-1 con-
[38]
[39]
[40]
[41]
[42]
[43]
433
sisting of JunD homodimers in primary human lung fibroblasts, J. Biol.
Chem. 274 (1999) 12933 – 12938.
R.A. Ignotz, J. Massague, Cell adhesion protein receptors as targets
for transforming growth factor-beta action, Cell 51 (1987) 189 – 197.
M.J. Reed, R.B. Vernon, I.B. Abrass, E.H. Sage, TGF-beta 1 induces
the expression of type I collagen and SPARC, and enhances
contraction of collagen gels, by fibroblasts from young and aged
donors, J. Cell. Physiol. 158 (1994) 169 – 179.
M. Parsons, E. Kessler, G.J. Laurent, R.A. Brown, J.E. Bishop,
Mechanical load enhances procollagen processing in dermal fibroblasts by regulating levels of procollagen C-proteinase, Exp. Cell Res.
252 (1999) 319 – 331.
F. Grinnell, Fibroblasts, myofibroblasts, and wound contraction, J. Cell
Biol. 124 (1994) 401 – 404.
J. Fringer, F. Grinnell, Fibroblast quiescence in floating or released
collagen matrices: contribution of the ERK signaling pathway and actin
cytoskeletal organization, J. Biol. Chem. 276 (2001) 31047 – 31052.
T.M. Freymann, I.V. Yannas, R. Yokoo, L.J. Gibson, Fibroblast
contractile force is independent of the stiffness which resists the
contraction, Exp. Cell Res. 272 (2002) 152 – 162.