Cyclic Straining of Cell-Seeded Synthetic Ligament Scaffolds:

Development of Apparatus and Methodology

EL MOSTAFA RAIF, BAHAA B. SEEDHOM, MICHAEL J. PULLAN,and TAKASHI TOYODA

ABSTRACT

Cyclic tensile strains acting along a ligament implant are known to stimulate cells that colonize it to pro-liferate and to synthesize an extracellular matrix (ECM), which will then remodel and form a new ligamentstructure. However, this process of tissue induction is poorly understood. As a first step toward elucidatingthis process, we aimed to investigate the effect of cyclic tensile strain on the proliferation of, and possible ECMsynthesis by, cells colonizing ligament scaffolds. Because there was no commercially available apparatus to un-dertake such investigation the objectives of this study were to develop an apparatus for the application of cyclictensile strains on cell-seeded synthetic ligament scaffolds and to develop and validate (through preliminarydata obtained using the apparatus) methodology for studying the effect of cyclic strain on cell proliferation.

We designed a multi-station test apparatus that operated inside an incubator. It allowed the application oftensile cyclic strains of between 0.5% and 5% at a frequency of 1 Hz on cell-seeded polyester ligamentscaffolds immersed in culture medium. Test stations with windows in their bases could be easily de-coupledfrom the apparatus. This allowed monitoring of cell proliferation and morphology, with inverted lightmicroscopy, through the transparent glass bases of the culture wells.

Preliminary experiments lasting for 1 day or 9 weeks examined the effect of selected aspects of the cyclicstrain on proliferation of cells seeded onto ligament scaffolds. Tests lasting for 1 day showed that theapplication of cyclic tensile strain of 5% for 4 h increased cell proliferation 24% above that observed inunstrained controls ( p< .05).

Scanning electron microscopy data from tests lasting for 9 weeks demonstrated further the dependency ofcell proliferation and possible ECM synthesis on the magnitude of the strain. The larger the amplitude, thegreater was the coverage of the scaffold with cells and ECM. Transmission electron microscopy of the ECMobserved at 9 weeks showed evidence of collagen fibrils aligned in the direction of load in strained scaffolds,whereas the tissue on the control scaffolds was random.

INTRODUCTION

RECONSTRUCTION OF RUPTURED LIGAMENTS, in particular

that of the anterior cruciate ligament (ACL), has become

widespread because it is now well recognized that, if left

untreated, ACL-deficient knees of active athletes become

predisposed to further injury. They can frequently develop

degenerative changes that could lead to premature arthritis in

active, young individuals.1,2 Several studies have demonstra-

ted that the health and soundness of ligaments are dependent

on the cyclic tensile strains that act on these tissues during

various activities.3–5 These strains have also been implicated

in the outcome of the reconstructive procedures of knee lig-

aments such as the ACL, particularly with synthetic scaffolds.

In vivo studies in the canine and human models have

highlighted an association between the presence (or absence)

Division of Bioengineering, Academic Unit of Musculoskeletal Disease, Faculty of Medicine and Health, University of Leeds, Leeds,

United Kingdom.

TISSUE ENGINEERINGVolume 13, Number 3, 2007# Mary Ann Liebert, Inc.DOI: 10.1089/ten.2006.0065

629

of tensile strains and the induction and remodeling (or lack of

it) of tissue around and within ligament scaffolds.6,7 How-

ever, apart from this association, no deeper insights into said

process of tissue induction can be derived from in vivo stud-

ies. This is because their nature precludes vital control

of the experimental variables involved (e.g., the amount of

strain applied on a ligament implant) and limits the number of

time points at which the outcome measures can be evaluated.

Furthermore, over the years, surgeons have used grafts that

have progressively increased in bulk and strength and hence

in stiffness. Once implanted in a joint, these stiffer grafts

would experience lower strains under the same loads than

those experienced by the more-compliant ones previously

used. This raises the question of whether there is a threshold

for the strain amplitude below which the proliferation of cells

seeding the implant and their subsequent extracellular matrix

(ECM) synthesis do not occur.Were such a threshold to exist,

it would have to be considered when selecting a graft or

designing a scaffold. The variation in the postoperative ex-

ercise regimens (in terms of periods of exercise and rest)

prescribed by various surgeons does not help in identifying

which of those are most beneficial to the process of the tissue

induction. Finally, in in vivo studies, it is impossible to as-

certain at what time points any effect or benefit of exercise

becomes apparent.

Although in vitro studies have considerable limitations,

they afford a measure of control on the variables and con-

ditions involved in the process of tissue induction. Therefore

they might help address some of the above questions. Pre-

vious in vitro studies have investigated the activities of cells

seeded onto various structures and substrates that have been

subjected to stretch,8–18 pressure,19 and shear.20–22 Stretch-

ing is the main deformation (straining) mode experienced by

the cells in ligament and tendon tissues, and the most rele-

vant of the above studies are those by Altman et al.15 and

Cacou et al.18 The former designed an elaborate 24-test-

station bioreactor that enabled the application of controlled

multidimensional strain on substantial scaffolds. The latter

designed an apparatus for straining cell-seeded collagen

gels. Neither system was commercially available. There-

fore, our objectives were to develop and commission a ro-

bust apparatus for the application of cyclic tensile strains in

an appropriate range on cell-seeded synthetic ligament

scaffolds and to develop and validate, through preliminary

data, methodology for assessing the effect of cyclic strain on

cell proliferation.

In a recent publication by 2 of the current authors,23 a

large amount of data, which were obtained using the above-

mentioned apparatus and methodology, have been reported.

These are summarized and discussed later in this article.

The goal of this article is to describe in detail the design,

development, and various commissioning tests of the ap-

paratus that we have designed and manufactured in house.

It also describes the various methodologies that we have

developed for harvesting and seeding cells onto synthetic

scaffolds for subsequent investigations of the effect of cy-

clic tensile strain on cell activities. It also presents the data

obtained as an outcome of commissioning the apparatus to

validate the methodology.

MATERIALS AND METHODS

Ligament scaffold materials

The ligament scaffolds used were made from polyester

(polyethylene terephthalate) and had identical weave to that

currently used in ligament and tendon implants (The Leeds

Keio Ligament, Xiros PLC, Leeds, UK). They had a nom-

inal tensile strength of 300 N and an average stiffness of

15N/mm. The ligament scaffold had an open weave (known

as mock leno) and comprised 9 yarns, each consisting of 96

monofilaments, 20 mm in diameter each. The scaffolds,

which are commercially available (Xiros PLC) were treated

with electronic discharge plasma, which was shown to

render the surfaces of the graft hydrophilic and improve cell

attachment to the scaffold.24 The scaffolds used in this study

were sterilized with gamma irradiation, as are scaffolds used

for clinical use.

Cells

We used cells derived from the synovium of the meta-

tarsophalangeal joints of skeletally mature bovines. The

tissue was harvested within 1 h of slaughter.

Apparatus

The multi-test-station apparatus, which was designed and

manufactured in house, initially comprised 6 test stations;

the current version has 8. These were all mounted onto a

common base plate (Fig. 1). Each test station housed

1 scaffold, and the simultaneous loading of all the scaffolds

was achieved using a common camshaft driven by an ex-

ternal motor and reduction gearbox via a flexible drive.

This latter passed through a sealed tunnel in the back of the

incubator, wherein the apparatus was operated.

Figure 2A is a schematic illustration of 1 complete test

station and its loading mechanism. The scaffold is gripped

between 2 clamps; one of these is stationary and attached to

the base plate, and the other is free to move under the effect

of tension applied via a spring and cam arrangement. The

free clamp is attached to 1 end of a loading rod that passes

through 2 linear bearing blocks. At the other end of the rod,

a cam follower is mounted. The loading spring surrounds

the rod and is constrained between 1 of the bearing blocks

and a ring that is firmly attached to the rod. By adjusting the

position of this ring on the loading rod, the compression of

the spring can be varied. The spring is thus pre-compressed,

and so, because it tends to expand, the scaffold, which

becomes continually tensed, constrains it. For a proportion

of each complete rotation, the cam unloads the scaffold as it

pushes the follower back toward the scaffold, causing it

630 RAIF ET AL.

‘‘buckle’’ and so loose the tensile load acting along it. The

profile of the cam determined this proportion of the cycle

during which the ligament is unloaded. At all times, the

spring remains in compression. The ratio of the load to no-

load periods thus remains constant regardless of the mag-

nitude of the load applied to the ligament. By adjusting the

extent of compression of the spring, the load, and hence

the strain it produces along the scaffold, could be varied.

The range of tensile loads that can be applied with this

system is between 15 N and 150 N, and the corresponding

range of strains these loads produced is between 0.5% and

5%, respectively.

Because the scaffold was in series with the loading

spring, it was necessary to ensure that any creep in the

scaffold did not appreciably change the load that the spring

exerts on the scaffold. We therefore followed a common

practice in the design of such loading systems as this by

selecting a spring of a much lower stiffness than that of the

scaffold. Thus the spring stiffness was 3N/mm, which was

1/50th that of the scaffold (*150N/mm). The creep likely

to occur in the scaffold was expected to be small enough so

as to produce a negligible change in the scaffold tension.

Subjecting one of these scaffolds to cyclic tensile load

(cycling between 10 N and 150 N) in an Instron materials

testing machine later corroborated this. On reaching steady

state (which occurred after some 100 cycles), the maximum

creep measured was less than 2mm. This meant that, after

100 s from commencing the experiment, the load acting on

the scaffold reduced from 150 N to 145 N (i.e., 3%) and

remained at that value for the remainder of the test. It was

possible to adjust the position of the carriage to take this

creep into account. The apparatus thus operates almost in a

load-control mode.

Figure 2B and C shows one of the test station assem-

blies and its constituent components. The assembly shown

in Figure 2B and C comprised 2 clamps (made from 316

stainless steel) for firmly gripping a scaffold so as to pre-

vent it from slipping under the influence of the cyclic load.

The lower portions of the clamps (where the scaffold was

gripped) were submerged in culture medium (see Appen-

dix for specification) inside a rectangular culture well with

a glass slide base for examination of cell morphology with

inverted microscopy. The well had a plastic cover in which

2 rectangular holes had to be made to accommodate the

protruding portions of the clamps at either end of the well.

The holes left rectangular annular spaces around the

clamps, which were sealed with wide rectangular stainless

steel rings so that the culture medium was not exposed to

the hazard of contamination. These rings were not con-

strained and so could slide on the cover of the culture well.

Figure 2D is a schematic illustration of a partial section

through the apparatus incorporating the detail of a test

station, with all the components labeled and described in

the legend. This section illustrates one important feature of

the apparatus; any test station could be easily de-coupled as

an undisturbed assembly and removed from the machine.

Because the metallic carriage in which the culture well was

placed had a rectangular opening in its base, this opening

revealed almost the whole of the glass transparent base of

the culture well. Cell proliferation and morphology could

thus be examined using inverted light microscopy.

EXPERIMENTAL WORK

Commissioning of the apparatus

Efficiency of the clamps. The clamps were first tested in

isolation in an Instron materials testing machine while

gripping a ligament scaffold in the configuration shown in

FIG. 1. The apparatus has 8 test stations mounted on a base plate onto which is mounted a common camshaft that is connected to an

external motor with a flexible drive. The motor rotates the camshaft that controls the loading and unloading of the scaffolds gripped

within their individual culture chambers. Color images available online at www.liebertpub.com/ten.

DEVELOPMENT OF APPARATUS AND METHODOLOGY 631

Figure 2B. The clamped scaffold was subjected to cyclic

tensile loading (between 10 N and 150 N), which produced

the maximum strain intended for use in the investigation.

The frequency of load application was 20Hz, and the total

number of load cycles applied was 300,000, which was

larger than the number of load cycles envisaged in any

planned tests. This test was intended to examine the effi-

cacy of the clamps by assessing whether the scaffold slip-

ped during cyclic load application.

After the test described above was completed, the ulti-

mate tensile strength of the scaffold itself was determined

using the Instron machine and compared with that of un-

strained scaffolds (controls). This test was undertaken to

determine whether the clamping process has damaged the

scaffold and thus affected its strength. There were 6 scaf-

fold samples in the test and control groups.

Direct measurement of load and strain acting on the

scaffold. A test station of the apparatus was instrumented

with a load cell and displacement transducer (Fig. 3) to si-

multaneously record the actual load applied to the scaffold

and the resulting extension and strain. The measurements

were taken approximately 2min after commencement of the

experiment, which was the period after which the loading

cycle experienced by the scaffold reached a steady state. The

measurements were taken at 5 load values between 15 N and

130 N to examine the repeatability of the load cycle and the

relationship between the amplitudes of load and resulting

extensions and strains.

Biocompatibility of the test stations. The biocompatibil-

ity of the environmental chamber was investigated to de-

termine whether the changes made to the cover of the culture

well, to accommodate the steel clamps, and the presence of

the clamps affect cell behavior. Therefore, the biocompat-

ibility of the test station (clamps and adapted culture well)

was investigated by comparing the cell viability, prolifera-

tion, and morphology in 2 configurations. In 1, a monolayer

culture of bovine synovial cells (104 cell/cm2) was seeded

in 1 well, with modified cover, in which 2 clamps were as-

sembled in the same way intended in a regular experiment.

In the other, cells were seeded in the same conditions in a

standard well, in which 2 agarose blocks were placed. The

agarose blocks had the same surface area as that of the

portions of the clamps that would be immersed in the wells

during a regular experiment. Cell viability and proliferation

were examined at 2 and 7 days of incubation using neutral

red assay25 and total deoxyribonucleic acid (DNA) content

assay, respectively. Cell morphology was investigated after

4 days of incubation. The samples were next prepared for

examination with light microscopy. The method of prepa-

ration for microscopic observations was the most com-

monly used by microscopists, in which the specimen was

fixed with 3.7% formaldehyde for 10min at room tempera-

ture then stained with 0.1% w/v toluidine blue in phosphate

FIG. 2. (A) Schematic illustration of 1 complete test station and

its loading mechanism [1] scaffold, [2] stationary clamp, [3] free

clamp, [4] spring, [5] connecting rod, [6] bearing block, [7] cam

follower, [8] ring for adjusting spring compression, [9] cam, [10]

frame/base plate of apparatus. (B) Assembly of a test station. (C)

Constituent components of a test station. (D) Central section

through a test station schematically illustrating the ligament

scaffold [1] attached within two clamps [2]. The scaffold is

sandwiched between two plates [3] within each clamp and gripped

by tightening the screws [4] within the clamps. These latter are

partially immersed in a rectangular culture well with a glass slide

bottom [6] to allow microscopic investigation without disturbing

the assembly. The culture well has a cover with 2 rectangular

holes through which the clamps protrude. The annular spaces

around these are covered with 2 rectangular rings [8] that are

allowed to slide over the cover as the clamps move under the

effect of load. The culture well is filled with the medium [9] to

above the scaffold. One of the clamps is kept stationary within its

holder [10], which is constrained by a steel rod [11] fixed to the

upper base plate of the apparatus [12]. The other clamp is attached

within a similar arrangement to end of the rod [13] where the

tensile load is applied via the spring/cam and cam follower ar-

rangement (not shown in this detail).

632 RAIF ET AL.

buffered salkine. Tests in the above configurations were

replicated 4 times at both time periods.

Validation of methodology—using the apparatus

in preliminary tests

The apparatus was used for short- and medium-term

investigations to validate the methodology. The short-term

tests were performed for just over 1 day, and the medium-

term tests spanned 9 weeks. The main aim of these tests,

particularly the latter, was to be assured of cell viability and

absence of contamination when performing tests over long

periods. The tests also aimed to obtain preliminary data on

the effect of cyclic strain application on the proliferation of,

and ECM synthesis by, synovial cells that were seeded onto

plasma treated ligament scaffolds. (See Appendix for pro-

tocols of cell harvesting, culture, and seeding onto scaffolds

and for assays referred to in this section.)

For the short-term tests, cells were harvested from the

synovium of metatarsophalangeal joints from freshly

slaughteredmature bovines. Cells from each animal were ex-

panded and then seeded onto 6 scaffolds. The cell-seeding

process was performed while the scaffolds were already

gripped in their respective clamps. Of these, 3 scaffolds

(test group) were subjected to a cyclic tensile strain of 5%

at a frequency of 1Hz for 4 h. The other 3 scaffolds were

used as controls. The scaffolds in the control group were

not subjected to cyclic strain but were placed in 3 other cell

culture chambers in the apparatus in which medium was

stirred to the same degree as the medium surrounding the

scaffolds subjected to cyclic strain. This was necessary be-

cause the application of mechanical strain engenders cul-

ture medium displacement and as a consequence enhances

the mixing of nutrients, which is likely to increase cell pro-

liferation. This test was thus repeated in triplicate using

cells derived from 3 animals.

The medium-term tests examined the ECM and tissue

using scanning electron microscopy (SEM) and transmis-

sion electron microscopy (TEM). The cell-seeded ligament

scaffolds were incubated in culture media with 1% fetal

bovine serum. The scaffolds were then subjected to cyclic

strain of 4 different amplitudes between 0% and 5% at a

frequency of 1Hz for 1 h per day (5 days per week) over a

9-week period. The same culture medium composition was

used throughout, and the medium was changed twice a

week. Because this experiment took 9 weeks, we deemed it

appropriate in this methodology article to use 2 scaffolds

for each of the strain amplitudes and 2 further scaffolds as

controls.

We have used standard procedures for specimen prepa-

rations for the 2 microscopic investigations.

Data processing and statistical analyses

Wherever only two groups were involved, the means and

standard deviations (SDs) of the data related to these two

groups were calculated, and comparison of the data was

performed with simple t- tests. Critical significance levels

were set at p< 0.05.

The data arising from the short-term tests in which the

DNA synthesis was assessed according to thymidine uptake

by cells (see Appendix). This was measured in 3 repli-

cate tests and 3 controls from each of 3 different animals.

The data were processed as follows. For each animal, the

thymidine uptake from each strained scaffold was com-

pared with the mean uptake from the 3 control scaffolds. To

FIG. 3. Instrumentation of a pair of clamps in a workstation for measuring the load and extension experienced by the scaffold. Color

images available online at www.liebertpub.com/ten.

DEVELOPMENT OF APPARATUS AND METHODOLOGY 633

make comparisons between the data from the 3 animals, the

data were normalized as follows. For each animal, each of

the 3 test values for thymidine uptake was expressed as a

ratio of the mean control uptake value. The average and SD

of the 3 values of this ratio were then plotted as a histo-

gram. The data were subject to univariate analysis of var-

iance. Tukey’s post hoc comparisons were derived to test

the significance of differences between individual test re-

sults and their respective un-stretched controls. The values

of these ratios for each animal were then compared with

those obtained for the other animals. Critical significance

levels were set at p< 0.05.

RESULTS

Commissioning the apparatus

Efficiency of the clamps and their effect on scaffold

strength. There was no evidence of scaffold slippage from

between the clamps, nor was there any sign of yarn rupture

of any of the scaffolds that were subjected to the cyclic

tensile loading. The ultimate tensile strength of the same

scaffold group was determined, and its mean value� SD

was 290.3� 22.5 N. The control scaffold group, which had

not been subjected to cyclic tensile strain, had a mean

strength of 295� 13.4 N. Statistical analysis using the t-test

showed that there was no significant difference in ultimate

strength between the 2 groups.

Load and strain cycles applied on the scaffold. Cyclic

tensile loads of amplitudes of 15 N, 25 N, 60 N, 75 N, and

130 N were applied in turn to the scaffold, and the resulting

extension at each of the 5 loads was measured. The am-

plitudes of corresponding strains produced by these loads

were approximately 0.5%, 0.9%, 2%, 3%, and 4.7%. Figure

4A and B shows the load-extension data obtained for the

maximum and minimum values of the load applied in this

test. The strain rate applied was also calculated from these

data and was approximately 0.35/s.

Biocompatibility of the test stations. After 24 h of seed-

ing, more than 95% of the total number of cells were at-

tached on the base of the culture well in both configurations

of the test described earlier in the Methodology section.

The DNA content monitored at day 2 and day 7 showed that

synovial cells were proliferating at the same rate in both

culture configurations. Furthermore, the number of cells

observed in the presence of the clamps was approximately

96% and 92%, respectively, of that observed in the control

group. However, the differences between the 2 groups and

the control group were not statistically significant. In addi-

tion, the neutral red assay confirmed that the clamps within

the modified culture well did not adversely affect cell via-

bility. In addition, the number of cell relative to the number

of cells in the control group, determined using neutral red,

showed that cell viability in the clamp group was 96% and

94% of that in the control groups after 2 days and 7 days,

respectively, in culture. However, the difference between

the 2 groups was not statistically significant.

Morphological analysis using optical microscopy showed

that, after 72 h in culture, the presence of the clamps and

the agarose cubes did not appreciably affect the morphol-

ogy of synovial fibroblasts. In both culture conditions, the

synovial cells retained their characteristic shape, which

compared well with that of the cells in standard monolayer

culture of synovial cells (Fig. 5).

Validation of methodology—preliminary data

Short-term tests. The application of strain of an ampli-

tude of 4.5% for 4 h at a frequency of 1Hz on cell-seeded

scaffolds induced greater thymidine uptake than in the

unstrained control group ( p< 0.05). Figure 6 shows data

from 3 separate animals. Each histogram bar represents the

average of 3 replicates from an animal normalized as de-

scribed in the section on data processing and statistical

analysis. The bars represent the SD. The asterisks indicate

that, for each animal, thymidine uptake in the test replicates

was significantly higher than in the control replicates

( p¼ 0.05). There were no significant differences between

0

0 0.5 1 1.5

Time [sec]

2 2.5 3

0

0.2

0.4

0.6

0.8

1

Dis

pla

cem

ent [m

m]

1.2

1.4

1.6

1.8

2

25

50

75

Load [N

]

100

125

150

Load

Displ

0

25

50

75

Load [N

]

100

125

150

0.5 1 1.5

Time [sec]

2 2.5 30

Load

Displ

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Dis

pla

cem

ent [m

m]

FIG. 4. Load and extension data from the instrumented clamp,

at the 2 extreme values in the range of loading used in the study.

634 RAIF ET AL.

the normalized data of the 3 animals. The normalized

values (expressed as ratios) of thymidine uptake for the 3

animals were 1.29� 0.09; 1.19� 0.11, and 1.23� 0.14,

respectively.

Medium-term tests. Cyclic tensile strain of 4 different

amplitudes was applied to 4 groups of cell-seeded scaffolds

for 1 h per day (5 days per week) for 9 weeks. The induced

tissue was examined using SEM and TEM. Figure 7 shows

SEM micrographs of the cell-seeded ligament scaffold after

it was subjected to cyclic tensile strain for 9 weeks at dif-

ferent tensile strain amplitudes. The larger the amplitude,

the greater was the coverage of the scaffold with cells and

ECM. TEM was used on the thin layer of tissue covering

scaffold specimens after 9 weeks of strain application. The

preliminary investigation was encouraging in that it showed

fibers (more likely immature collagen fibers) aligned along

the direction of the strain application on the scaffold when

the strain applied was 2.5% or more, whereas in the control

and 1% strain groups the fibers were randomly oriented

(Fig. 8).

DISCUSSION

The process of tissue induction and remodeling that

follow ligament reconstruction, whether with autogenous

tissue or synthetic scaffolds, is still far from thoroughly

understood. It was argued earlier that it was appropriate to

elucidate this process by conducting in vitro studies. Al-

though these studies have considerable limitations, they

afford a measure of control over the variables affecting the

tissue induction process. This is an advantage that in vivo

studies do not offer, although in the long run, in vivo studies

will provide the most useful data. However, in vivo studies

should be undertaken later and their designs be based on

insights gained from in vitro studies.

Therefore the objectives of the present study were to

design an apparatus and develop methodology, using the

apparatus, for investigating in the first instance the effect of

cyclic tensile strain on the proliferation and ECM synthesis

of synovial cells, which have been seeded onto synthetic

ligament scaffolds. Because there were no commercially

available apparatus to conduct such a study, we have de-

signed and manufactured in house a multi-test-station ap-

paratus in which it was possible to control the variables

investigated, in particular the amplitude and frequency of

the cyclic strain applied, as well as the period of strain

application. We have also developed methodology and

various protocols to commission and validate the apparatus

and then to obtain preliminary data on the effect of cyclic

FIG. 5. Morphology of synovial cells in monolayer culture; cells were fixed with 3.7% formaldehyde in phosphate buffered saline

then stained with Toluidine Blue O. (A) Cells cultured in the presence of agarose cubes of similar dimensions to the immersed sections

of the clamps. (B) Cells cultured in the presence of the clamps. Color images available online at www.liebertpub.com/ten.

FIG. 6. Effect of cyclic tensile strain on cell proliferation. Cell-

seeded plasma-treated polyester scaffolds were subjected to cyclic

strain of 4.5% at 1Hz for 4 h. Cells were labeled with 3H-

thymidine during the 0- to 24-h window (0 is the starting point of

the strain application). Each histogram bar is the average of

triplicate readings from one animal. Data were normalized as

described in the data processing section. *Statistically significant

differences in thymidine uptake of test specimens compared with

their respective controls ( p< 0.05; n¼ 3).

DEVELOPMENT OF APPARATUS AND METHODOLOGY 635

tensile strain on the proliferation and ECM synthesis of

cells seeded onto synthetic ligament scaffolds.

Scaffolds

Two comments are appropriate with regard to the scaf-

folds used in this study. There were no suitable commer-

cially available scaffolds in any other material than the one

cited earlier. The choice was then narrowed down to the

strength the scaffold should have. To use a scaffold of the

regular strength used for ACL reconstruction (*2000–

2400 N) would have required extremely large clamps

to grip the scaffold firmly, and this in turn would have

reduced the number of test stations greatly, to the disad-

vantage of the study. It would also have rendered the nec-

essarily large apparatus unsuitable to operate within an

incubator. Therefore, we used smaller scaffolds that had a

nominal strength of 300 N, but these allowed the obser-

FIG. 7. SEM micrographs of the synovial cell–seeded ligament scaffold after being subjected to a regime of cyclic tensile strain for 9

weeks, at different tensile strain amplitudes; the larger the amplitude, the greater was the coverage of the scaffold with cells and ECM.

FIG. 8. (A) Transmission electron microscopyTEM showing (arrow) fibers aligned along the direction of strain application on

scaffolds seeded with synovial cells after being subjected to a 2.5% cyclic tensile strain for 9 weeks. (B) Random fibers were observed

on control scaffolds, which were not subjected to cyclic strain.

636 RAIF ET AL.

vations and various planned measurements to be per-

formed.

In due course, when scaffolds of different materials be-

come commercially available, these can be used for further

studies.

Cells

Cells harvested from the synovium, rather than fibro-

blasts derived from ligaments, have been appropriately

used to seed the scaffolds in this study. One reason for this

choice is that the authors envisage that, in the clinical sit-

uation, un-seeded scaffolds will be used. When implanted

(e.g., as in the case of ACL reconstruction), the scaffold is

usually passed into bone tunnels made where the original

ligament is attached to the femur and tibia. Making the

tunnels involves the removal of the ligament’s remnants,

and in doing so, fibroblasts from the remnants would not

be available to colonize the scaffold. The synovial mem-

brane lining the capsule and surrounding the posterior

cruciate ligament would be a likely source of cells to col-

onize the scaffold. Synovial cells are also suitable, because

they can be induced consistently into multi-lineage differ-

entiation pathways and can conserve this multi-lineage

in vitro.26

Mesenchymal stem cells (MSCs) from human bone

marrow would be equally appropriate, because they would

likely migrate from the bone tunnels and colonize the scaf-

fold, where they could proliferate, differentiate, and syn-

thesize ECM that would remodel under the effect of

physiological tensile strains.

The apparatus

Once the apparatus was commissioned, it was found

satisfactory for the intended use. The apparatus is compact

so that it can be operated within the confines of an incu-

bator and so requires no special environmental controls to

replicate the operating conditions of the incubator. Because

it comprised 8 test stations, it could simultaneously ac-

commodate a group of up to 4 test specimens and a similar

group of control specimens. It was thus possible to conduct

simultaneous test and control replicate measurements on

cells from the same animal.

The loading mechanism using low-stiffness springs in

series with the high-stiffness scaffold overcame the potential

problem of creep of the scaffold, which could have altered

the tensile load applied on the scaffold. The tests conducted

showed that load-extension curve obtained reached a steady

state after approximately 100 load cycles.

The application of the maximum load on the scaffold for

approximately 300,000 cycles, using the specially designed

clamps, did not result in damage to the scaffold.

It was possible to apply cyclic tensile strains on the

scaffold in the range of 0.5% to 5% at a frequency of 1Hz.

A useful feature of this apparatus was that it allowed

visual monitoring of the morphology of cells and their

spread over the scaffold. This was possible by virtue of the

detachable test station design and the use of a culture well

with a glass slide base. It was thus possible to observe,

using inverted light microscopy, the cell morphology and

the extent to which the cells and ECM filled the intra-yarn

space on the open-weave structure of the scaffold.

Strain values used in the planned tests

This subject has been controversial (and will probably

continue to be so for some time), because ACL strains in

the human have been directly measured only in patients

during various regimens of rehabilitation. Such direct

measurements will have been justified on clinical grounds

and after informed consent of the patients involved. These

data are summarized in the paper by Fleming et al.,27 which

addressed strains in the ACL during stair climbing. Strains

in the ACL during the various exercises listed ranged be-

tween 0.1� 0.9% in passive flexion-extension of the knee

and 4.4� 0.6% in isometric contraction of the quadriceps

muscles at 158 under the influence of 30Nm of exten-

sion torque. During stair climbing, the ACL strain had an

average of some 2.8%, reaching a maximum of 7.3% in 1

of 5 patients tested. In another study, Fleming et al.28

measured ACL strains in patients during bicycling ranging

from –3.4% to 5.1%. Thus the literature contains much

useful data on ACL strains during a number of rehabilita-

tion exercises.

Direct measurement of ACL strains in vivo during nor-

mal locomotion of normal healthy humans is not ethically

justified, and therefore no such data exist. These must there-

fore be estimated from calculated ACL tensions during

locomotion and ACL tensile properties determined from

cadaveric measurements. Morrison29–31 analyzed the knee

forces during 5 activities: level walking, stair ascent and

descent, and walking up and down a ramp. The ACL forces

ranged between 169 N during level walking and 447 N

during stair descent. The corresponding strains to these

forces, as determined from the load-extension graph for a

young human,32 were approximately 3.1% and 6.9%.

From the foregoing, it appears that a strain of 5% might

reasonably be considered ‘‘physiological’’ and appropri-

ate to use in this study, although the strains occurring dur-

ing locomotion are estimates and might be considered to be

somewhat ‘‘soft’’ data.

The lower values of strain were also appropriate to use

in the planned investigation because the implants used

by surgeons (whether autogenous tissue or synthetic scaf-

folds) are generally stiffer than the natural ACL and would

therefore experience lower strains during use. It was there-

fore useful to undertake tests at decreasing values of strain

to determine the threshold for the strain at and below

which cell proliferation drastically reduced or ceased al-

together.

DEVELOPMENT OF APPARATUS AND METHODOLOGY 637

Strain rate

The strain rate used in this preliminary study was 0.35/s,

compared with a physiological value of 0.4/s, estimated on

the basis of 0.05 strain applied in the tests and the rise time

(approximately 0.12 s) of the ACL force from 0 to peak

value during locomotion, in the paper by Collins and

O’Connor.33 Strain rates could be greater in faster sporting

activities, but locomotion is the most prevalent of humans

activities. Nevertheless, should the effect of strain rate be

necessary to investigate, this could be undertaken by chang-

ing the profile of the cam in the present apparatus.

Preliminary data

The methodology developed and protocols adopted for bi-

ological commissioning of the apparatus demonstrated that:

1. the environment of the apparatus is biocompatible;

cell viability was evident

2. according to the preliminary data, in the short and me-

dium term, cyclic tensile strain stimulates cell prolif-

eration and that the results obtained in the short- term

tests from 3 different animals were consistent

3. the SEMobservations in themedium-term experiments

showed that cell proliferation and ECM synthesis are

dependent on the amplitude of the cyclic strain

Finally, the methodology described in the Appendix aims

at studying, in the first instance, the effect of cyclic tensile

strain on the proliferation of cells and tissue induction, at a

phenomenological level. However, the apparatus described

here can still be used in any future study in which other

biochemical assays are adopted to investigate cell activities

at the molecular level.

Additional data

It is appropriate to mention here other data that we have

obtained using the apparatus and methodology described

in this article but have been published23 before this one. We

briefly summarize the results of this study in which we used

206 ligament scaffolds. The experiments undertaken ex-

plored the effect of the following variables on cell prolifera-

tion, which was investigated using the uptake of thymidine

with which cells were labeled after 24 h according to pre-

scribed protocols.

(a) The period of strain application: Tests were carried

out at a strain of 4.5% and a frequency of 1Hz, applying

strain for 0.5 h, 1 h, and 4 h. From this test, it was estab-

lished that cell proliferation had a maximum value when

the period of strain application was 1 h.

(b) The amplitude of strain: Scaffolds were subjected to

cyclic tensile strain for periods of 1 h at strain amplitudes of

4.5%, 2.5%, 1%, and 0% (controls) at a frequency of 1Hz.

This test showed that cell proliferation was related to the

amplitude of strain and showed further that there is a threshold

for the amplitude of the strain (1%) at and below which cell

proliferation was not significantly different from that ob-

served in control specimens.

The tests described in (a) and (b) lasted for one day each.

(c) Short-term cumulative effect of strain application: In

a series of tests lasting for 5 weeks, cyclic tensile strain of

different amplitudes was applied to cell-seeded scaffolds

for 1 h per day (5 days per week) for 5 weeks. Cell (and

possible ECM) growth into the inter-yarn spaces formed in

the weave of the scaffold was monitored using light mi-

croscopy. The area occupied by the proliferating cells (or

degree of fill of these rectangular spaces in the scaffold

weave) was observed and expressed as percentage of the

total area of the respective rectangular space examined. The

results showed that, in the control group, cells covered an

average of 8% (range 6–14%) of the total intra-yarn space

area. Non-significant differences were observed when the

cells were subjected to cyclic tensile strain of 1%; the av-

erage of the area occupied by the proliferating cells and

tissue was 10% (6–18%) of the total area of the intra-yarn

space. The corresponding percentages of the areas covered

by cells and tissue of the inter-yarn spaces were 20% (range

16–30%) for a strain amplitude of 2.5% and 44% (range

24–70%) for a strain amplitude of 4.5%. These were sig-

nificantly different from each other and individually from

the control and the 1% strain groups ( p< .05).

CONCLUSION

This article describes design, development, and com-

missioning of an apparatus for the application of cyclic

tensile loads onto cell-seeded ligament scaffolds. A meth-

odology has been also developed for using the above ap-

paratus to investigate the effect of cyclic tensile strain on

the proliferation of bovine synovial cells seeded onto syn-

thetic ligament scaffolds. The experimental work under-

taken in this study has validated the function of the

apparatus in so far as its capacity to perform the functions

for which it has been intended. The preliminary data pre-

sented have demonstrated that the methodology developed

is appropriate. The apparatus can be used with other meth-

odologies and appropriate assays for investigating the ef-

fect of cyclic tensile strain on other metabolic activities from

the synovium or other origins such as bone marrow.

APPENDIX: PROTOCOLS FOR CELLHARVESTING, EXPANSION, SEEDING

ONTO SCAFFOLDS, AND MEASUREMENTOF CELL PROLIFERATION

Cell culture

The synovium specimens, aseptically harvested from the

synovium of metatarsophalangeal joints within 1 h after

slaughter of bovines younger than 18 months old were

638 RAIF ET AL.

digested at 378C for 3 h using 0.25 % (w/v) collagenase type

IA (Sigma) inDulbecco’smodifiedEagle’smedium (DMEM)

(Sigma). The digest was then centrifuged at 500 g for 10min

and the pellet suspended in DMEM before passing through a

70-mm nylon filter to remove undigested residue. Cells were

isolated by centrifugation at 500 g for 10min and re-sus-

pended in DMEM. Primary cells were seeded at 104 cells/cm2

in DMEM, supplemented with 10% fetal bovine serum,

penicillin, streptomycin, and amphotericin. The culture me-

dium was changed 3 times per week, and cells were passaged

after 7 to 9 days. Only passages 1 to 3 were used in the study.

Cell seeding the ligament scaffold

It was essential that the ligament scaffolds were gripped

in the clamps before they were seeded. This was dictated by

practical reasons: the relatively long time required for

precise assembling of the clamps and the requirement for

sterility and ease of handling. When placed in the culture

wells, the scaffolds were approximately 3mm above the

base of the well, and because cells tended to gravitate to the

bottom of the wells, their access to the scaffold would have

been decreased. The base of the well was therefore raised

by coating it with a layer of agarose gel of the same height

to overcome this difficulty. To further maximize the access

of cells to the scaffold, the space occupied by the cell

culture medium was greatly reduced by filling the majority

of the well with agarose gel such that the space occupied

with the culture medium was a narrow central channel

surrounding the scaffold. To achieve this, an H-shaped

mold made of stainless steel was used, around which the

agarose gel was cast. When the mold was removed, it left

voids that accommodated the 2 clamp sections gripping the

scaffold, a central channel also accommodating the latter,

and the culture medium. A volume of 1mL of culture

medium containing 105 cells was added to each clamped

scaffold placed on the channel-shaped agarose-coated

wells. The chambers were covered with the adapted cover

and the annular space then sealed with the silicone rubber

and incubated at 378C for 24 h. The scaffolds were trans-

ferred to new culture wells containing 3mL of the culture

medium with 10% fetal bovine serum.

Cell proliferation

Seventy-two h after seeding, the cells were rendered

quiescent by incubation for 20 h in DMEM with 0.5% fetal

bovine serum. Then after adding 1 mCi/mL [Methyl-3H]-

thymidine to each sample, test group of scaffolds were

subjected to a cyclic tensile strain of 4.5% at a frequency of

1Hz for 4 h. In the control group, the clamps were sub-

jected to the same movement as the test group without

applying any strain to the scaffold but engendering a stir-

ring of the culture medium to the same degree as in the test

group. Labeling was then terminated after 24 h by removal

of culture medium, at which point the cell-seeded scaffolds

were rinsed 3 times with phosphate buffered saline. The

cells were digested for 24 h at 658C with 1mL of 0.1 M

sodium phosphate, pH 6.0, containing 5mM disodium

ethylene diaminetetraacetate, 5mM cysteine hydrochloric

acid (HCl), and 1.25 mg/mL papain (Sigma). The digest was

then centrifuged at 18000 g for 5min. Thymidine uptake

(deoxyribonucleic acid (DNA) synthesis) was determined

and used as a measure of cell proliferation. All tests were

repeated in triplicate, and the experiments were repeated

using synovial cells from 3 different animals.

Assay of DNA synthesis

DNA was precipitated by mixing 250 mL of the papain-

digested cells with 250 mL 10% ice-cold trichloroacetic

acid-tannic acid followed by incubation for 30min at 08C.The supernatant was discarded after centrifugation, and the

precipitate was suspended in 0.5 N NaOH followed by

neutralisation with 0.5 N HCl. An aliquot was mixed with a

scintillation cocktail, and the radioactivity was measured

using a liquid scintillation counter. The results, expressed

as disintegrations per minute (dpm) per 1 mg of DNA con-

tent and measured as described below, were taken as an in-

dication of DNA synthesis.

Assay of total DNA content

Fifty mL of papain-digested cells was applied to an op-

tical methacrylate cuvette (Kartell Plastics UK Ltd). Three

mL of 0.1 mg/mL Hoechst 33258 (Sigma) solution in

10mM Tris buffer was then added to each sample. Fluo-

rescence was measured in a fluorescence spectrometer

(Perkin-Elmer LS-3, Perkin-Elmer Ltd., Beaconsfield, UK)

with the excitation and emission wavelengths set at 348 and

458 nm, respectively. DNA content was determined from

the standard curve obtained with increasing amounts of calf

thymus DNA (Sigma).

ACKNOWLEDGMENTS

This work was supported by a ROPA Grant GR/N23608/

01 from The Engineering and Physical Sciences Research

Council. The authors wish to thank Mr. Brian Whitham for

helping with the manufacture of the apparatus.

REFERENCES

1. Pelletier J.P., Martel-Pelletier J, Howell D.S. Etiopathogen-

esis of osteoarthritis. In: Koopman WJ, ed. Arthritis and Al-

lied Conditions. Baltimore: Williams and Wilkins. 1997, pp.

1969–84.

2. Noyes F.R., Mooar PA, Matthews DS, Bulter DL. The

symptomatic anterior cruciate-deficient knee. J Bone Joint

Surg Am. 65, 154, 1983.

3. Vailas A.C., Tipton C.M., Matthes R.D. et al. Physical ac-

tivity and its influence on the repair process of medial col-

lateral ligaments. Connect Tissue Res. 9, 25, 1981.

DEVELOPMENT OF APPARATUS AND METHODOLOGY 639

4. Woo SLY, Gomez M.A., Woo Y.K. et al. Mechanical prop-

erties of tendons and ligaments. II. The relationships of im-

mobilization and exercise on tissue remodelling. Biorheology.

19, 397, 1982.

5. Kamps B.S., Linder L.H., DeCamp C.E. et al. The influence

of immobilization versus exercise on scar formation in the

rabbit patellar tendon after excision of the central third. Am J

Sports Med. 22, 803, 1994.

6. Fujikawa K, Seedhom B.B., Matsumoto H et al. The Leeds-

Keio ligament. In: Strover A. ed. Intra-Articular Reconstruc-

tion of the Anterior Cruciate Ligament. London: Butterworth

Heinemann. 1993, p. 173–207.

7. Fujikawa K, Iseki F and Seedhom B.B. Arthroscopy after

anterior cruciate reconstruction with the Leeds-Keio ligament.

J Bone Joint Surg. 71, 566, 1989.

8. De Witt M.T., Handley C.J., Oakes BW et al. In vitro re-

sponse of chondrocytes to mechanical loading: the effect of

short-term mechanical tension. Connect Tissue Res. 12, 97,

1984.

9. Hasegawa S, Sato S, Saito S et al. Mechanical stretching

increases the number of cultured bone cells synthesizing DNA

and alters their pattern of protein synthesis. Calcif Tissue Int.

431, 1985.

10. Ngan PW, Crock B, Varghese J et al. Immunohistochemical

assessment of the effect of chemical and mechanical stimuli

on cAMP and prostaglandin E levels in human gingival fi-

broblasts in vitro. Archs Oral Biol. 33, 163, 1988.

11. Pender N, McCulloch CAG. Quantitation of actin polymeri-

zation in two human fibroblast sub-types responding to me-

chanical stretching. J Cell Sci. 100, 187, 1991.

12. Sutker B.D., Lester G.E., Banes A.J. et al. Cyclic strain

stimulates DNA and collagen synthesis in fibroblast cultured

from rat medial collateral ligaments. Trans Orthop Res Soc.

15, 130, 1990.

13. Neidlinger-Wilke C, Wilke H.J., Claes L. Cyclic stretching of

human osteoblasts affects proliferation and metabolism: a

new experimental method and its application. J Orthop Res.

12, 70, 1994.

14. Akai Y, Homma T, Burns K.D. et al. Mechanical stretch/

relaxation of cultured rat mesangial cells induces proto-

oncogenes and cyclooxygenase. Am J Physiol. 267, C482,

1994.

15. Altman G.H., Horan R.L., Martin I et al. Cell differentiation

by mechanical stress. FASEB J. 16, 270, 2001.

16. Toyoda T, Matsumoto H, Fujikawa K, Saito S, Inoue K.

Tensile load and metabolism of anterior cruciate ligament

cells. Clin Orthop Relat Res. 353, 247, 1998.

17. Kim SG, Akaike T, Sasagawa T, Atomi Y, Kurosawa H. Gene

expression of type I and type III collagen by mechanical

stretch in anterior cruciate ligament cells. Cell Struct Funct.

27, 139, 2002.

18. Klein-Nulend J, Veldhuijzen J.P., Burger E.H. Increased

calcification of growth plate cartilage as result of compressive

force in vitro. Arthritis Rheum. 29, 1002, 1986.

19. LaPlaca M.C., Thibault L.E. An in vitro traumatic in-

jury model to examine the response of neurons to a hydro-

dynamically-induced deformation. Ann Biomed Eng. 25, 665,

1997.

20. Carver S.E., Heath C.A. Semi-continuous perfusion system

for delivering intermittent physiological pressure to re-

generating cartilage. Tissue Eng. 5, 1, 1999.

21. Hung C.T., Allen F.D., Pollack S.R., Attia E.T., Hannafin

J.A., Torzilli P.A. Intracellular calcium response of ACL and

MCL ligament fibroblasts to fluid-induced shear stress. Cell

Signal. 9, 587, 1997.

22. Cacou C, Palmer D, Lee D.A., Bader D.L., Shelton J.C. A

system for monitoring the response of uniaxial strain on cell

seeded collagen gels. Med Eng Phys. 22, 327, 2000.

23. Raıf E.M., Seedhom B.B. Effect of cyclic tensile strain on

proliferation of synovial cells seeded onto synthetic ligament

scaffolds – an in vitro simulation Bone 36, 418, 2005.

24. Rowland J.R.J., Tsukazaki S, Kikuchi T et al. Radiofre-

quency-generated glow discharge treatment: potential benefits

for polyester ligaments. J Orthop Sci. 38, 198, 2000.

25. Babich H, Borenfreund E. Neutral Red assay for toxicology

in vitro. In: Watson RR, ed. In Vitro Methods of Toxicology.

1992, pp. 237–251.

26. De Bari C, Dell0Accio F, Tylzanowski P, Luyten F.P. Mul-

tipotent mesenchymal stem cells from adult human synovial

membrane, Arthritis Rheum. 44, 1928, 2001.

27. Fleming B.C., Beynnon B.D., Renstrom P.A. et al. The strain

behavior of the anterior cruciate ligament during stair

climbing: an in vivo study. Arthroscopy. 15, 185, 1999.

28. Fleming B.C., Beynnon B.D., Renstrom P.A., Peura G.D.,

Nichols C.E., Johnson R.J. The strain behavior of the anterior

cruciate ligament during bicycling. An in vivo study. Am J

Sports Med. 26, 109, 1998.

29. Morrison J.B.. Bioengineering analysis of force actions

transmitted by the knee joint. Biomed Eng. 3, 164, 1968.

30. Morrison JB. Function of the knee joint in various activities.

Biomed Eng. 4, 573, 1969.

31. Morrison J.B. The mechanics of the knee joint in relation to

normal walking. J Biomech. 3, 51, 1970.

32. Grood E.S., Noyes F R. Cruciate ligament prosthesis:

strength, creep and fatigue properties. J Bone Joint Surg. 58,

1083, 1976.

33. Collins J.J., O’Connor J.J. Muscle-ligament interaction at the

knee during walking. Proceedings of the Institute of Me-

chanical Engineering, Part H. J Eng Med 205, 11, 1991.

Address reprint requests to:

Dr. Bahaa B. Seedhom

Division of Bioengineering

Academic Unit of Musculoskeletal Disease

Faculty of Medicine and Health

University of Leeds

Leeds

United Kingdom

E-mail: b.b.seedhom@leeds.ac.uk

640 RAIF ET AL.