influence of flow rate and scaffold pore size on cell behavior
TRANSCRIPT
Royal College of Surgeons in Irelande-publications@RCSI
Anatomy Articles Department of Anatomy
4-1-2012
Influence of flow rate and scaffold pore size on cellbehavior during mechanical stimulation in a flowperfusion bioreactor.Ryan J. McCoyRoyal College of Surgeons in Ireland
C JungreuthmayerAustrian Institute of Technology
Fergal O'BrienRoyal College of Surgeons in Ireland, [email protected]
This Article is brought to you for free and open access by the Departmentof Anatomy at e-publications@RCSI. It has been accepted for inclusion inAnatomy Articles by an authorized administrator of [email protected] more information, please contact [email protected].
CitationMcCoy RJ, Jungreuthymayer C, O'Brien FJ. Influence of flow rate and scaffold pore size on cell behavior during mechanicalstimulation in a flow perfusion bioreactor. Biotechnology and Bioengineering. 2012;109(6):1583-1594.
— Use Licence —
Attribution-Non-Commercial-ShareAlike 1.0You are free:• to copy, distribute, display, and perform the work.• to make derivative works.Under the following conditions:• Attribution — You must give the original author credit.• Non-Commercial — You may not use this work for commercial purposes.• Share Alike — If you alter, transform, or build upon this work, you may distribute the resulting work onlyunder a licence identical to this one.For any reuse or distribution, you must make clear to others the licence terms of this work. Any of theseconditions can be waived if you get permission from the author.Your fair use and other rights are in no way affected by the above.This work is licenced under the Creative Commons Attribution-Non-Commercial-ShareAlike License. Toview a copy of this licence, visit:URL (human-readable summary):• http://creativecommons.org/licenses/by-nc-sa/1.0/URL (legal code):• http://creativecommons.org/worldwide/uk/translated-license
This article is available at e-publications@RCSI: http://epubs.rcsi.ie/anatart/51
Influence of Flow-rate and Scaffold Pore Size on Cell Behaviour During
Mechanical Stimulation in a Flow Perfusion Bioreactor
McCoy, R.J. Eng.D1,2
Tel.: +353 1 402 8508;
E-mail address: [email protected]
Jungreuthmayer, C. Ph.D3,4
Phone: +43 50550 6413
Fax: +43 50550 6595
Mail: [email protected]
O’Brien, F.J. Ph.D1,2
(Corresponding Author)
Tel.: +353 1 402 2149;
fax: +353 1 402 2355.
E-mail address: [email protected]
1. Department of Anatomy
Royal College of Surgeons in Ireland
123 St. Stephen’s Green
Dublin 2
Ireland
2. Trinity Centre for Bioengineering,
Trinity College Dublin,
Dublin,
Ireland
3. Austrian Institute of Technology (AIT),
Mobility Department,
Giefinggasse 2,
1210 Vienna,
Austria
4. Austrian Centre of Industrial Biotechnology (ACIB),
Muthgasse 11/DG,
1190 Vienna,
Austria
Condensed Running Title: Cell Detachment in Highly Porous Scaffolds
Abstract
Mechanically stimulating cell-seeded scaffolds by flow-perfusion is one approach utilised for
developing clinically applicable bone graft substitutes. A key challenge is determining the
magnitude of stimuli to apply that enhances cell differentiation but minimises cell detachment
from the scaffold. In this study we employed a combined computational modelling and
experimental approach to examine how the scaffold mean pore size influences cell
attachment morphology and subsequently impacts upon cell deformation and detachment
when subjected to fluid-flow. Cell detachment from osteoblast-seeded collagen-GAG
scaffolds was evaluated experimentally across a range of scaffold pore sizes subjected to
different flow-rates and exposure times in a perfusion bioreactor. Cell detachment was found
to be proportional to flow-rate and inversely proportional to pore size. Using this data, a
theoretical model was derived that accurately predicted cell detachment as a function of mean
shear stress, mean pore size and time. Computational modelling of cell deformation in
response to fluid flow showed the percentage of cells exceeding a critical threshold of
deformation correlated with cell detachment experimentally and the majority of these cells
were of a bridging morphology (cells stretched across pores). These findings will help
researchers optimise the mean pore size of scaffolds and perfusion bioreactor operating
conditions to manage cell detachment when mechanically simulating cells via flow perfusion.
Keywords
Perfusion Bioreactor, Computational Fluid Dynamics, Finite Element Analysis, Shear Stress,
Pore Size, Cell Detachment
Introduction
Increasing limitations with respect to traditional autograft and allograft bone replacement
procedures, including the restricted amount of bone available, requirement for additional
invasive surgery, risk of infectious disease transmission and lack of available donors, is
driving research towards the development of clinically applicable bone graft substitutes. Bone
tissue-engineering approaches aim to address this unmet clinical need by using one or more
arms of the tissue-engineering triad; scaffolds, cell signalling factors and cells.
To develop bone graft substitutes the simplest methodology is to statically culture cells on
highly porous scaffolds in the presence of differentiation medium (Farrell, E et al., 2006;
Farrell et al., 2007; Keogh et al., 2010; Kim et al., 2005; St-Pierre et al., 2005; Tierney et al.,
2009). However, static culturing regimes neglect the important role of biomechanical stimuli
in the normal formation, development and homeostasis of bone tissue: the effects of which
are clearly observed when normal, physiologically relevant levels of mechanical stimulation
are removed (Morey and Baylink, 1978; Zerwekh et al., 1998). Therefore, in-vitro bone tissue
cultivation strategies have sought to enhance the osteogenic differentiation of cell-seeded
constructs by employing mechanical stimulation via flow-perfusion (Bancroft et al., 2002;
Cartmell, S. et al., 2003; Keogh et al., 2011; Sikavitsas et al., 2003; Sikavitsas et al., 2005;
Vance et al., 2005). We have successfully employed an in-house designed perfusion
bioreactor (Jaasma et al., 2008) to examine the effects of mechanical stimulation on gene
expression levels of key osteogenic markers for MC3T3 (pre-osteoblasts) cell-seeded
collagen-GAG (CG) scaffolds subjected to a range of fluid flow regimes and for exposure
times of 1h to 14 days (Jaasma and O’Brien, 2008; Partap et al., 2009; Plunkett et al., 2010).
One key challenge of such a strategy is determining the magnitude of stimuli to apply that
maximises the induction of osteogenic differentiation but minimises cell detachment.
Physical detachment of cells from a scaffold occurs when the fluid flow generated forces
acting upon the cell exceed the adhesive forces generated through integrin-matrix
interactions. It is a particularly undesirable artefact of flow perfusion regimes (Alvarez-
Barreto and Sikavitsas, 2007; Alvarez-Barreto et al., 2007; Cartmell, S. et al., 2003; Plunkett
et al., 2010) having multiple implications from scientific (cell to cell communication),
economic (cell culture costs) and clinical efficacy (minimum cell number) perspectives.
Initial cell attachment levels within highly porous collagen-based scaffolds have been shown
to be a function of the mean pore size (Murphy and O’Brien, 2010; Murphy et al., 2010;
O’Brien et al., 2005; O’Brien et al., 2007) with cells attaching in either a flat (attached to a
single strut, akin to monolayer culture (2D)) or bridged (attached to multiple struts, stretching
across void spaces within the scaffold) morphology (Annaz et al., 2004; McMahon, 2007).
The proportion of cells adopting a bridged morphology is hypothesised to increase with
decreasing mean pore size because of increased numbers of strut junctions (Harley et al.,
2008), which are believed to act as nucleation points for cell bridging. Computational
modelling has shown that bridged cells can experience up to 500 fold greater levels of
cytoskeletal deformation than flat cells under the same flow conditions (Jungreuthmayer et
al., 2009b); their alignment to the direction of flow creates a greater level of resistance and at
the extreme end of the scale one can consider them to behave like a sail in the wind. We
hypothesise it is this increased cytoskeletal deformation of bridging cells that promotes
osteogenesis at lower mean shear stress levels () (~1-100x10-3
Pa) in highly porous scaffolds
compared to 2D (0.1–1.5Pa) (McCoy and O’Brien, 2010), or scaffolds with very large pore
sizes where a flat morphology type dominates (Vance et al., 2005). Furthermore, assuming
constant cell properties (e.g integrin density), bridged cells would be expected to have fewer
integrin-matrix interactions as they have smaller surface areas of attachment, resulting in a
reduced adhesive force. This reduced adhesive force coupled with increased cell deformation
(displacement caused by the mechanical stresses trying to pull the cell away from the
scaffold) is expected to make bridging cells more susceptible to detachment. We hypothesise
this is why cell detachment is observed in highly porous scaffolds subjected to flow perfusion
at a (~1-100x10-3
Pa) well below the critical (the required to remove 50% of attached
cells) shown to detach cells in 2D (~1-100Pa). Thus we believe a critical threshold (Tc) of
cell deformation exists for bridging cells, below which cells are stimulated and above which
they are detached, and this occurs at much lower than for flat cells.
Cell detachment has been extensively studied in 2D, with studies typically conducted in one
of two ways; parallel plate flow chambers (Bouafsoun et al., 2006; Kapur and Rudolph, 1998;
van Kooten et al., 1994; Truskey and Proulx, 1993; Xiao and Truskey, 1996) or rotating
disc/flow devices (Deligianni et al., 2001; Engler et al., 2009; Garcia et al., 1997; García et
al., 1997; Goldstein and DiMilla, 2003). To date, the authors are unaware of any studies
directed at evaluating the detachment of cells cultured on highly porous scaffolds in response
to fluid-flow. Highly porous scaffolds contain exceedingly irregular geometries as a
consequence of the freeze drying (Mandal and Kundu, 2008; O’Brien et al., 2004) or salt
leaching (Kim et al., 2005) fabrication processes, resulting in non-uniform flow profiles that
make shear stress characterisation very challenging. Furthermore, technical limitations in
visualising cells internally within the construct, means correlating cell detachment to the
magnitude of shear stress at the point of detachment is exceptionally difficult.
The overall goal of this paper was to characterise cell detachment in response to fluid flow
from highly porous CG scaffolds with different mean pore sizes and evaluate the relationship
between cell loss, cell morphology and cell deformation. We employed a novel approach
(Figure 1) combining two previously developed computational models (Jungreuthmayer et
al., 2009b; Jungreuthmayer et al., 2009a) with a series of experimental investigations. A finite
volume method (FVM) approach to characterising fluid flow rates in terms of shear stress,
was used in conjunction with CT imaging of the scaffolds to accurately characterise the
shear stress distribution within each scaffold under particular bioreactor operating conditions.
This data was utilised in conjunction with experimental studies to derive a predictive model
capable of characterising cell detachment based on mean scaffold pore size, and exposure
time to flow, which was subsequently validated against previously published literature. Based
on our hypothesis that bridging cells are predominantly detached from the scaffold, we used
this experimental data to set the number of bridging cells for the second computational
model. A finite element (FE) elastostatic simulation that allowed cell deformation
(displacement) to be calculated for individual cells based on their attachment morphology and
in response to the flow perfusion conditions they experienced within the scaffold. From this a
Tc for cell deformation, beyond which cells would detach, was determined. This yielded good
correlation with experimentally observed levels of detachment with the majority of the cells
exceeding Tc having a bridged morphology.
Materials and Methods
Experimental Methodology
Scaffold Fabrication
Scaffolds were fabricated in accordance with a previously developed protocol (O’Brien et al,
2004). Briefly, a CG suspension was created by blending micro-fibrillar bovine tendon
collagen (Integra Life Sciences, Plainsboro, NJ) with chondroitin-6-sulphate sodium salt,
isolated form shark cartilage (Sigma-Aldrich, St. Louis, MO, USA) in 0.05M acetic acid
(Fisher Scientific Loughborough, Leicestershire, UK); the CG suspension was degassed
under vacuum and freeze-dried (VirTis Co., Gardiner, NY). Final freezing temperatures of -
10oC, -40
oC and -60
oC yielded scaffold sheets with mean pore sizes of 325, 120 and 85m
respectively (Haugh et al., 2010). Post-lyophilisation scaffold sheets were dehydrothermally
cross-linked at 105oC for 24h in a vacuum oven at 50 mTorr (VacuCell, MMM, Germany).
Individual scaffold discs (diameter=12.7mm; depth=3-4 mm) were punched out of the sheets
and strengthened through chemical cross-linking; scaffolds were submersed in an aqueous
solution of 14mM N-(3-dimethylaminopropyl)-N0-ethylcarbodiimide hydrochloride (EDAC)
and 5.5mM N-Hydroxysuccinimide (NHS) for 2h before being washed and stored in PBS.
Cell Culture
The pre-osteoblastic cell line, MC3T3-E1 (<passage 29), was cultured in -MEM (Sigma-
Aldrich, Ireland, Dublin) supplemented with 10% FBS (Biosera, East Sussex, UK), 1% L-
glutamine (Sigma–Aldrich) and 2% penicillin/streptomycin (Sigma–Aldrich). Cells were
maintained as sub-confluent monolayers in T175 flasks under standard conditions (37oC, 5%
CO2). Cells were detached with Trypsin–EDTA (Sigma–Aldrich) and re-suspended at a
concentration of 1x107 cells/mL. Scaffolds (85, 120 and 325m mean pore size) were seeded
with a total of 2 x 106 cells. In 6 well plates, 100 L of cell suspension was added drop-wise
onto the top surface of each scaffold and the plates placed into the incubator for 15 min to
allow for cell attachment. Scaffolds were then turned over and the procedure repeated. After
the second incubation period, 5 mL of growth media was added to each well and the scaffolds
were pre-cultured statically for 6 days (media change on day 3) (Partap et al, 2009).
Bioreactor Experiments
Constructs were then cultured in our in-house designed flow perfusion bioreactor (Jaasma et
al., 2008). During bioreactor culture, six constructs (n=2 per mean pore size) were cultured
simultaneously using a steady flow regime for 6, 24 or 48h, at either 0.05, 1 or 5mL/min.
These flow-rates were chosen as they equate to between 9x10-4
and 8.79x10-1
Pa, values
below that shown to stimulate or cause cell detachment of flatly attached cells in 2D (McCoy
and O’Brien, 2010). Thus cell detachment would be expected to be predominantly related to
bridged cells. Furthermore these have been previously shown to enhance osteogenesis of
MC3T3 cell-seeded scaffolds by our group, but resulted in cell detachment of up to 40% for
scaffolds with mean pore sizes of 120m (Jaasma and O’Brien, 2008). Static controls were
transferred to new 6-well plates and cultured in 5mL of fresh medium. Post-culture constructs
were flash frozen in liquid nitrogen and stored at -80oC until analysis.
DNA Extraction and Quantification
DNA was isolated from the constructs by homogenisation in RLT lysis buffer (Qiagen, USA)
using a rotor-stator homogeniser (Omni International). Cell lysates were centrifuged using QI
Shredder columns (Qiagen) and the supernatant stored at -800C until quantification. DNA
quantification was performed using a Quant-iT™ PicoGreen dsDNA kit (Invitrogen) in
accordance with manufacturer’s instructions. Sample fluorescence was measured (excitation
480nm, emission 538nm) and DNA concentration deduced using a standard curve.
High Resolution Scanning Electron Microscopy (HRSEM)
Constructs were prepared for HRSEM by fixation (2% gluteraldehyde/2% formaldehyde for
2h), washing (PBS) and secondary fixation (1% OsO4) for 1h, before being dehydrated in
10min stages through a series of ethanols (30%, 50%, 70%, 90%) followed by 3x20min in
100% ethanol. Constructs were chemically dried using a series of Hexamethyldisilazane in
ethanol steps (30%, 60%, 100%), sputter coated with Palladium to a nominal thickness of
7nm using a Cressington sputter coater and imaged using a Carl Zeiss Ultra scanning electron
microscope at 5kV. Images were falsely coloured (Adobe Photoshop) to highlight cells.
Computational Methodology
Computational models developed in house (Jungreuthmayer et al., 2009a; Jungreuthmayer et
al., 2009b) were used in this study. Computational fluid dynamic (CFD) simulations of fluid
flow through empty CG scaffolds were used to calculate the under different flow
conditions. These values were used for deriving a predictive equation of cell detachment.
CFD simulations of fluid flow through cell-seeded CG scaffolds were used to determine the
shear stresses and hydrostatic pressures acting on individual cells numerically seeded onto the
scaffold. These values were fed into elastostatic deformation simulations to determine the cell
deformation experienced by each cell in the cell-seeded scaffold. Mesh generation and
simulations were conducted using the opensource CFD toolbox OpenFOAM (OpenFOAM
version 1.3, OpenCFD Ltd, Reading, Berkshire, UK). Simulations and associated pre- and
post processing were conducted on a 32 bit, dual processor, 4Gb RAM system. In brief, a
three step (i, iii and iv) or five step procedure was required to successfully execute the empty
and cell-seeded models respectively; (i) scaffold reconstruction, (ii) numerical cell seeding,
(iii) mesh creation, (iv) CFD simulation and (v) elastostatic simulation. Steps ii and v were
not required for empty scaffold simulations as they involve the numerical seeding of cells
onto the scaffold and the deformation simulations of individual cells.
Scaffold Reconstruction
MicroCTCT) scans (SCANCO Medical AG, Bassersdorf, Switzerland) conducted using a
special filter technology to increase the contrast of the CG scaffolds (10,240m x 10,240m
x 520m), with a resolution of 6m x 6m x 6m, were obtained for scaffolds having mean
pore sizes of 85, 120 and 325m. Three smaller randomly chosen sub-volumes (768m x
768m x 384m) were extracted from each raw data file to lessen the associated
computational costs. The grey-scale data for each sub-volume was Gaussian rank filtered (to
reduce background noise) and subjected to a thresholding procedure that resulted in a black
(scaffold) and white (void/interstice) image of the geometry. Scaffold porosity after
thresholding was approximately 90% in all cases.
Numerical Cell Seeding
For the part of this study examining cell deformation in response to flow, scaffolds were
numerically seeded with cells using a cell seeding program developed in house
(Jungreuthmayer et al., 2009b). It uses two similar algorithms to seed cells with either
bridged or flat morphologies having minimum and maximum cell lengths of 50 and 80m
respectively; attachment lengths observed in-vitro for MC3T3 cells at high confluencies in
2D (unpublished data). This procedure was repeated until 255 cells per sub-volume were
successfully seeded onto the scaffold (equal to a cell seeding density of 5x105 cells/scaffold).
Mesh Creation
The mesh generation utility blockMesh was used to decompose the domain geometry of the
scaffold interstice into hexahedral elements. Scaffold meshes (seeded or non-seeded)
contained approximately 1.1x106 elements. Additionally, in the case of cell-seeded scaffolds
a mesh of each cell was generated (~250 elements) for use in the elastostatic simulations.
CFD Simulation
CFD simulations were performed using the laminar solver icoFoam (Transient solver for
incompressible, laminar flow of Newtonian fluids) with the settings: laminar fluid flow,
incompressible Newtonian fluid with a kinematic viscosity of 8.1x10-7
m2/s, no-slip boundary
conditions on walls, constant velocity inlet (1175, 235 and 12m/s, which correspond to the
experimental peak flow rates of 5, 1, 0.05mL/min), zero-pressure outlet, and impermeable
cell and scaffold walls. Cells were considered rigid for the CFD simulation.
Elastostatic Simulation
The cells were assumed to be an isotropic, homogeneous material with a constant Young’s
modulus of 1kPa, simplifying the complex mechanical nature of the cells to a linear elastic
behaviour (Karcher et al., 2003; Theret et al., 1988). Cell deformation was simulated as a
function of the wall shear stress and the hydrostatic wall pressure computed during the CFD
simulation; applied to the nodes of cell faces exposed to fluid flow. The nodes of cell faces
touching the scaffold were constrained: ux=uy=uz=0, where ux, uy, and uz are the
displacements of a node in x-, y-, and z-direction. This constraint yielded immobile nodes at
the cell-scaffold contact area (cells are stuck to the scaffold). Each cell was individually
deformed using the FE (finite element) solver stressFemFoam. Only cells residing within an
internal sub-volume (660m x 660m x 330m, ~130 cells) of each extracted sub-volume
were analysed to avoid potential boundary artefacts.
Results
Experimental Evaluation of Cell Detachment
Using our in-house designed flow-perfusion bioreactor system we experimentally
characterised changes in osteoblast (MC3T3) cell number, as measured by change in total
DNA quantity, for different inlet flow-rates (static, 0.05, 1, 5mL/min), across a complete
range of CG scaffolds with mean pore sizes (85, 120 and 325m) and over culture time
durations of 0, 6, 24 and 48h (Figure 2). The maximal levels of cell detachment were
proportional to flow-rate and inversely proportional to mean pore size. The most favourable
conditions, a mean pore size of 325m and a flow-rate of 0.05mL/min yielded cell
detachment of ~3% after 48h, whilst the most unforgiving set-up, a mean pore size of 85m
and a flow-rate of 5mL/min, caused ~40% of cells to detach after 48h. For scaffolds with the
larger mean pore sizes of 120 and 325m, decreases in cell number for the 48h time point
were minimal upon increasing the flow-rate from 1 to 5mL/min, suggesting that the
remaining proportion of cells are significantly less susceptible to the fluid flow forces than
the previously detached proportion; a bi-modal distribution of cell detachment, whereby we
propose the bridging cells have detached under the flow conditions studied, leaving just the
flatly attached cells bound to the scaffold.
Characterisation of Shear Stress
In order to create a predictive model of cell detachment that can be utilised across a broad
range of system types and scaffold geometries/porosities, it was firstly necessary to express
flow-rate as a more relevant engineering parameter, mean shear stress (). Computational
modelling was used to determine the shear stress distribution and within the complex
internal architecture of CG scaffolds as a function of the different mean pore sizes and flow-
rates studied experimentally. CT images of 85, 120 and 325m mean pore size scaffolds
were used to provide accurate 3D information on the scaffold architectures. Due to the highly
porous nature of the scaffolds (~90%), no variation in maximum shear stress or the shear
stress distribution were observed as a function of pore size. Example data for a flow-rate of
1mL/min is shown in Figure 3. A linear relationship was observed (=c*V; where V is the
velocity of the fluid entering the scaffold and c is a constant value of 7.5e-5 Pa.s/m) between
flow-rate and (Table 1). For all flow-rates, the maximum shear stress was approximately
one order of magnitude greater than the , however, only a very small percentage of the
scaffold/cell surface areas were exposed to these conditions as typically 95% of the
scaffold/cell surface area was subjected to shear stresses less than 3-4 times the (Figure 3).
Cell Detachment Characterisation and Validation
Mathematical functions describing the shape of the data were fitted to the experimental data
points (Supplementary material). These different functions were then condensed to form the
following equation;
t
P
P
o
Pe
Pa
CC 0766.0907.5
0124.015
exp12312.25651
1824.10003.0
0036.01
Where C = fraction of cells in the scaffold at time t, Co = the number of cell initially, is the
mean shear stress in the scaffold, P = the mean pore size and a = maximum fold change in
cell number under static conditions (amount of cell growth in the absence of flow).
We validated the equations predictive ability against previously published literature for
MC3T3 cells cultured on CG scaffolds (Jaasma and O’Brien, 2008). In the first scenario, the
derived function accurately predicted cell detachment for a complex flow regime consisting
of 1h at 1mL/min followed by 7h at 0.05mL/min, repeated over a 49h period (Figure 4). Such
cycles are typically used in our laboratory to minimise cellular desensitization to mechanical
stimulation. In the second study a continuous low flow-rate regime of 0.05mL/min was
examined over a 49h period (Figure 4). The predicted levels of cell detachment for this
regime were well within the standard deviation of the experimental study.
Cell Deformation Correlates to Cell Detachment
As shown in Figure 3, pore size was found to have no effect on the average shear stress,
maximum shear stress or the shear stress distribution for scaffolds having a constant porosity
when subjected to the same flow-rate (Figure 3). This suggests that observed differences in
cell detachment can’t be attributed to shear dependent events, such as localised high shear
zones or variations in distribution profiles that may occur within scaffolds of different mean
pore size. One characteristic that we hypothesise to vary as a function of pore size is cell
attachment morphology, thus we examined computationally whether there was any
correlation between cell morphology, deformation and detachment. Firstly, we confirmed the
capability of MC3T3 cells to form both flat and bridged morphology types in CG scaffolds
by HRSEM (Figure 5). One commonly observed bridging cell morphology was limited to
two points of attachment, whilst flat cells typically had long thin morphologies as they ran
along the edge of struts (Figure 6), validating the choice of cell shape used in our
computational modelling scheme. After 48h changes in cell numbers appear to stabilise,
suggesting that cell detachment/re-attachment dynamics have reached a plateau, thus based
on our hypothesis that bridging cells are preferentially detached, the maximal detachment
levels observed experimentally (Table 2) were assumed to equal the percentage of bridging
cells present in the scaffold at 0h. Therefore cells were numerically seeded onto CT
reconstructed scaffolds at ratios of 32% bridged to 68% flat and 23% bridged to 77% flat for
120m and 325m mean pore size scaffolds respectively. Simulations were then run using
experimentally studied flow-rates and the level of cell deformation assessed. From
computational simulations conducted for 5mL/min flow-rates for both 120 and 325m
scaffolds, the critical threshold of deformation, Tc (the level of deformation above which a
cell would be considered to detach), was determined so that the same detachment levels were
observed as seen experimentally; for both pore sizes this was approximately 30nm. The
simulation were then re-run for the other flow rates tested and the percentage of cells that
could be considered to have detached (deformation levels > 30nm) under each set of flow
conditions were calculated. These values correlated well with that of the actual cell
detachment levels based on the mathematical curve fitting of the experimental data (Figure
7). In all cases greater than 95% of the detached cells from the computational simulations
were of the bridged morphology type.
Discussion
In 2D, osteogenic differentiation is enhanced at in the range of 0.1–1.5Pa, typically well
below the lower limit associated with cell detachment (1-100Pa) (McCoy and O’Brien,
2010). In highly porous scaffolds, osteogenesis can be promoted at much lower (~1-
100x10-3
Pa), yet cell loss is a prevalent issue (McCoy and O’Brien, 2010). The overall goal
of this paper was to firstly characterise cell detachment in response to fluid flow from highly
porous CG scaffolds with different mean pore sizes, and secondly, to evaluate if the cell loss
observed could be related to cell morphology and cell deformation. The study successfully
characterised cell detachment experimentally and in combination with computational
modelling developed a model capable of accurately predicting cell detachment, the
methodologies of which may be applied to perfusion-scaffold bioreactor systems in general.
Furthermore, evidence to support the hypothesis that cell detachment can be attributed to a
bridging morphology type absent in 2D was presented.
Differences in geometrical configurations of perfusion bioreactor systems and the complex
internal architecture of scaffolds (Bancroft et al., 2002; Bonvin et al., 2010; Jaasma et al.,
2008; Wendt et al., 2003; Zhao and Ma, 2005) makes correlating observed biological
outcomes on the basis of inlet fluid flow-rate very difficult. Calculation of the within
systems allows comparisons to be made on a single representative scale. In this study we
successfully used established FVM techniques (Cioffi et al., 2006; Jungreuthmayer et al.,
2009a; Maes et al., 2009; Raimondi et al., 2006; Sandino et al., 2008), to characterise the
shear stress distribution and values within CT imaged CG scaffolds of differing mean pore
sizes under different flow-rates. Pore size was found to have no effect on the average shear
stress, maximum shear stress or the shear stress distribution for scaffolds having a constant
porosity when subjected to the same flow-rate (Figure 3). This suggests that observed
differences in cell detachment cannot be attributed to shear dependent events, such as
localised high shear zones or variations in distribution profiles that may occur within
scaffolds of different mean pore size. However, one cellular characteristic that is constant in
2D, but is expected to vary as a function of mean pore size in highly porous constructs, is cell
attachment morphology. We hypothesised that smaller pore sizes would have a higher ratio of
bridging to flat morphology types and the cell detachment observed could be attributed to
bridging cells, which are expected to be more susceptible to detachment when subjected to
flow. We confirmed the capability of the cells to form both flat and bridged morphology
types in the CG scaffolds by HRSEM (Figure 5 and Figure 6). Assuming cells cannot
spontaneously reach across a void space in order to bridge, and have a finite maximum
attachment length, then to initiate bridging cells need to be localised at junction apexes where
they can “migrate” in multiple directions along one or more struts, away from the apex, to
form a bridged morphology. For larger mean pore sizes (>150m), where the cell is
significantly smaller than the length of a scaffold strut, then the majority of cells will not be
located within the vicinity of strut junctions and hence are more likely to attach flatly.
Limited cell bridging will still occur, but will be constrained to the regions surrounding strut
junctions. As pore size decreases the density of strut junctions increases per unit volume
(Harley et al., 2008) (junctions per pore remain constant) and strut length decreases,
increasing the proportion of cells in the locality of strut junctions. Thus as pore size
decreases, a greater proportion of cells are predicted to have a bridged morphology type.
Therefore, if bridging cells are more susceptible to the shear stresses than flat cells, greater
levels of cell detachment would be expected to occur in smaller pore sized scaffolds under
equal shear stresses. This trend was noted in our data (Figure 2) supporting the hypothesis of
a role for cell morphology with respect to cell detachment. Whilst different cell morphologies
could be visualised by HRSEM (Figure 5 and 6), a limitation of this study is that a systematic
analysis quantifying the total number of cells having either bridging or flat morphologies in
different pore sized scaffolds was not conducted. This would be a very challenging
undertaking technically and thus a more simplified approach was used in this study based on
the relationship between and trends observed in experimental cell detachment levels. In
general, increasing the induced greater levels of detachment (Figure 2). However,
increasing from 0.0176 to 0.0879Pa in both 120 and 325m mean pore size scaffolds
caused minimal changes in the total cell detachment, suggesting a susceptible cell population
(bridged cells) had been detached and the remaining cells (flat cells) had significantly greater
adhesion strengths; indicative of a bi-modal distribution of cell adhesion strength. This may
be explained by surface area of attachment, where flat cells are expected to have larger
surface areas of attachment than bridged cells, so assuming uniform scaffold material
properties (ligand density) and cell characteristics (integrin expression), bridged cells would
be expected to have reduced adhesion strengths and be more susceptible to detachment. We
therefore estimated the size of the bridging population, as a function of pore size, to be equal
to cell detachment levels observed experimentally at the highest after 48h, when cell
reattachment/detachment dynamics were considered to have stabilised; 32% and 23% for
120m and 325m mean pore size scaffolds respectively.
Cell attachment morphology does not solely affect cell adhesion, it also influences cell
deformation; the degree of displacement the cell experiences in response to physical forces
caused by fluid flow. Cells have non-uniform attachment lengths or orientations to the
direction of flow, with larger cells and orientations more normal to the direction of flow
resulting in greater levels of cell deformation. Computational modelling has shown that
bridged cells experience up to 500 fold greater levels of cytoskeletal deformation than flat
cells (Jungreuthmayer et al., 2009b) under flow conditions that have been shown to promote
osteogenic differentiation, but cause cell detachment from the scaffold (Alvarez-Barreto and
Sikavitsas, 2007; Alvarez-Barreto et al., 2007; Cartmell, S. et al., 2003; Jaasma and O’Brien,
2008; Partap et al., 2009; Plunkett et al., 2010). We hypothesised the greater levels of cell
deformation experienced by bridging cells is the primary contributing factor to cell
detachment observed in cell-seeded scaffolds undergoing flow perfusion. To examine the
relationship between cell deformation, morphology and detachment, as a function of flow, we
used the information gained experimentally to set the number of bridged cells for our
computational simulations. Thus cells were numerically seeded in the computational model at
a ratio of 32% to 68% and 23% to 77% (bridged to flat) for 120m and 325m mean pore
size scaffolds respectively. Based on our experimental data that cells susceptible to
detachment should all be detached at of 0.0879Pa, the minimum deformation required to
detach 32% and 23% of cells for the 120m and 325m mean pore size scaffolds
respectively was calculated; a value equating to 30nm, the critical threshold of detachment,
Tc. Assuming that cell morphology profiles and hence adhesion strength profiles are the same
for any scaffold of a given pore size, then changes in the deformation profiles with respect to
the Tc reflect changes in cell detachment levels (Figure 8). Computational simulations were
conducted for the other used experimentally and the percentage of cells >Tc correlated well
with cell detachment (Figure7); in all cases >95% of the cells experiencing deformation
levels >30nm(detached cells) were of a bridging morphology type, supporting the hypothesis
that bridging cells are the predominant morphology type detached from highly porous
scaffolds.
Through a curve fitting approach we derived an equation capable of predicting cell
detachment based on , mean pore size and exposure time for MC3T3 cell-seeded CG
scaffolds. This predictive model was validated against previously published literature for
MC3T3 cells cultured on CG scaffolds (Jaasma and O’Brien, 2008). The predictive model
not only predicted cell loss under constant with respect to time, but could also be applied to
more complex flow regimes where the is changed with respect to time. Thus this predictive
model can be used with confidence to help understand how changes to pore size or bioreactor
operating conditions will influence detachment in MC3T3 cell-seeded CG scaffolds.
Translating this work to other systems having different scaffold properties or cell types,
would currently require the repetition of the experimental and computational work conducted
herein. However, if the scaffold properties could be incorporated into the predictive model
then repetition of the experimental work may be reduced and the power of equation derived
enhanced further. To date, what we have shown is that the combined use of such modelling
and experimental studies can be successfully applied to the development of predictive models
of cell detachment in scaffold-perfusion systems.
Finally, it is important to consider the trade-off between cell detachment and the induction of
osteogenic differentiation. Further studies need to be conducted to determine the most osteo-
inductive flow-rates. This will be a complex undertaking and is not within the scope of this
study, but we can hypothesise that if cell differentiation is triggered by cytoskeleton
deformation, then bridged cells are likely to have enhanced levels of activation in comparison
to flat cells. However, if all bridged cells were detached, then enhanced differentiation of flat
cells may still be achievable if is increased to 2D levels. Thus a bi-modal distribution for
induction of cell differentiation may also exist in response to mechanical stimulation. Either
way, researchers developing perfusion-scaffold systems can hopefully use understanding
gained from studies such as this to design scaffolds and control bioreactor operating
conditions to better control population wide responses.
Conclusion
Characterisation of cell detachment in highly porous scaffolds following mechanical
simulation in a flow perfusion bioreactors is a challenging undertaking due to the internal
complexity of scaffold architecture and the inability to visualise cells in real-time. We
employed a novel approach combining computational modelling and experimental techniques
to characterise cell detachment and demonstrated that detachment in porous scaffolds occurs
at well below those shown to cause cell detachment in 2D and attributed this to a bridging
cell morphology type not present in 2D. Furthermore, this study illustrated the potential use
of computational models in combination with experimental techniques for the development
of tissue engineered constructs.
Acknowledgements
This study has received funding from the European Research Council under the European
Community's Seventh Framework Programme (FP7/2007-2013) / ERC grant agreement n°
239685. This work was enabled by the CRANN Advanced Microscopy Laboratory [AML],
Trinity College Dublin. This work has also been supported by the Austrian BMWFJ, BMVIT,
SFG, Standortagentur Tirol, and ZIT through the Austrian FFG-COMET- Funding Program.
The authors would like to thank Prof. Dimitri Scholz and Dr. Cormac O’Connell (UCD
Conway Institute of Biomolecular and Biomedical Research) for technical assistance with
sample preparation for SEM and Colm Faulkner (Centre for Research on Adaptive
Nanostructures and Nanodevices (CRANN), Trinity College Dublin) for technical assistance
and training with respect to SEM operation.
References
Alvarez-Barreto JF, Sikavitsas VI. 2007. Improved mesenchymal stem cell seeding on RGD-modified
poly(L-lactic acid) scaffolds using flow perfusion. Macromolecular Bioscience 7:579-588.
Alvarez-Barreto J, Linehan S, Shambaugh R, Sikavitsas V. 2007. Flow perfusion improves seeding of
tissue engineering scaffolds with different architectures. Annals of Biomedical Engineering
35:429-442.
Annaz B, Hing KA, Kayser M, Buckland T, Silvio LD. 2004. Porosity variation in hydroxyapatite and
osteoblast morphology: a scanning electron microscopy study. Journal of Microscopy
215:100-110.
Bancroft GN, Sikavitsas VI, van den Dolder J, Sheffield TL, Ambrose CG, Jansen JA, Mikos AG.
2002. Fluid flow increases mineralized matrix deposition in 3D perfusion culture of marrow
stromal osteoblasts in a dose-dependent manner. Proceedings of the National Academy of
Sciences of the United States of America 99:12600-12605.
Bonvin C, Overney J, Shieh AC, Dixon JB, Swartz MA. 2010. A multichamber fluidic device for 3D
cultures under interstitial flow with live imaging: Development, characterization, and
applications. Biotechnology and Bioengineering 105:982-991.
Bouafsoun A, Othmane A, Kerkeni A, Jaffrézic N, Ponsonnet L. 2006. Evaluation of endothelial cell
adherence onto collagen and fibronectin: A comparison between jet impingement and flow
chamber techniques. Materials Science and Engineering: C 26:260-266.
Cartmell, S., Porter, B.D, García, A.J., Guldberg, R.E. 2003. Effects of medium perfusion rate on cell-
seeded three-dimensional bone constructs in vitro. Tissue Engineering 9:1197.
Cioffi M, Boschetti F, Raimondi MT, Dubini G. 2006. Modeling evaluation of the fluid-dynamic
microenvironment in tissue-engineered constructs: A micro-CT based model. Biotechnology
and Bioengineering 93:500-510.
Deligianni DD, Katsala ND, Koutsoukos PG, Missirlis YF. 2001. Effect of surface roughness of
hydroxyapatite on human bone marrow cell adhesion, proliferation, differentiation and
detachment strength. Biomaterials 22:87-96.
Engler AJ, Chan M, Boettiger D, Schwarzbauer JE. 2009. A novel mode of cell detachment from
fibrillar fibronectin matrix under shear. J Cell Sci 122:1647.
Farrell, E, O’Brien, FJ, Doyle, P, Fischer J, Yannas, I, Harley, BA, O’Connell Brian, Prendergast, PJ,
Campbell, VA. 2006. A collagen-glycosaminoglycan scaffold supports adult rat mesenchymal
stem cell differentiation along osteogenic and chondrogenic routes. Tissue Engineering
12:459.
Farrell E, Byrne EM, Fischer J, O’Brien FJ, O’Connell BC, Prendergast PJ, Campbell VA. 2007. A
comparison of the osteogenic potential of adult rat mesenchymal stem cells cultured in 2-D
and on 3-D collagen glycosaminoglycan scaffolds. Technology and Health Care 15:19-31.
Garcia AJ, Ducheyne, P, Boettiger, D. 1997. Cell adhesion strength increases linearly with adsorbed
fibronectin surface density. Tissue Engineering 3:197.
García AJ, Ducheyne P, Boettiger D. 1997. Quantification of cell adhesion using a spinning disc
device and application to surface-reactive materials. Biomaterials 18:1091-1098.
Goldstein AS, DiMilla PA. 2003. Examination of membrane rupture as a mechanism for mammalian
cell detachment from fibronectin-coated biomaterials. Journal of Biomedical Materials
Research Part A 67A:658-666.
Harley BAC, Kim H-D, Zaman MH, Yannas IV, Lauffenburger DA, Gibson LJ. 2008.
Microarchitecture of three-dimensional scaffolds influences cell migration behavior via
junction interactions. Biophysical Journal 95:4013-4024.
Haugh MG, Murphy CM, O’Brien FJ. 2010. Novel Freeze-Drying Methods to Produce a Range of
Collagen–Glycosaminoglycan Scaffolds with Tailored Mean Pore Sizes. Tissue Engineering
Part C: Methods 16:887-894.
Jaasma MJ, O’Brien FJ. 2008. Mechanical stimulation of osteoblasts using steady and dynamic fluid
flow. Tissue Eng Part A 14:1213-1223.
Jaasma MJ, Plunkett NA, O’Brien FJ. 2008. Design and validation of a dynamic flow perfusion
bioreactor for use with compliant tissue engineering scaffolds. Journal of Biotechnology
133:490-496.
Jungreuthmayer C, Jaasma MJ, Al-Munajjed AA, Zanghellini J, Kelly DJ, O’Brien FJ. 2009b.
Deformation simulation of cells seeded on a collagen-GAG scaffold in a flow perfusion
bioreactor using a sequential 3D CFD-elastostatics model. Medical Engineering & Physics
31:420-427.
Jungreuthmayer C, Donahue SW, Jaasma MJ, Al-Munajjed AA, Zanghellini J, Kelly DJ, O’Brien FJ.
2009a. A comparative study of shear stresses in collagen-glycosaminoglycan and calcium
phosphate scaffolds in bone tissue-engineering bioreactors. Tissue Eng Part A 15:1141-1149.
Kapur R, Rudolph AS. 1998. Cellular and cytoskeleton morphology and strength of adhesion of cells
on self-assembled monolayers of organosilanes. Experimental Cell Research 244:275-285.
Karcher H, Lammerding J, Huang H, Lee RT, Kamm RD, Kaazempur-Mofrad MR. 2003. A Three-
Dimensional Viscoelastic Model for Cell Deformation with Experimental Verification.
Biophysical Journal 85:3336-3349.
Keogh MB, Partap S, Daly JS, O’Brien FJ. 2011. Three hours of perfusion culture prior to 28 days of
static culture, enhances osteogenesis by human cells in a collagen GAG scaffold. Biotechnol.
Bioeng 108:1203-1210.
Keogh MB, O’Brien FJ, Daly JS. 2010. Substrate stiffness and contractile behaviour modulate the
functional maturation of osteoblasts on a collagen-GAG scaffold. Acta Biomaterialia 6:4305-
4313.
Kim HJ, Kim U-J, Vunjak-Novakovic G, Min B-H, Kaplan DL. 2005. Influence of macroporous
protein scaffolds on bone tissue engineering from bone marrow stem cells. Biomaterials
26:4442-4452.
van Kooten TG, Schakenraad JM, van der Mei HC, Dekker A, Kirkpatrick CJ, Busscher HJ. 1994.
Fluid shear induced endothelial cell detachment from glass - influence of adhesion time and
shear stress. Medical Engineering & Physics 16:506-512.
Maes F, Ransbeeck PV, Oosterwyck HV, Verdonck P. 2009. Modeling fluid flow through irregular
scaffolds for perfusion bioreactors. Biotechnology and Bioengineering 103:621-630.
Mandal BB, Kundu SC. 2008. Non-bioengineered silk fibroin protein 3D scaffolds for potential
biotechnological and tissue engineering applications. Macromolecular Bioscience 8:807-818.
McCoy RJ, O’Brien FJ. 2010. Influence of Shear Stress in Perfusion Bioreactor Cultures for the
Development of 3D Bone Tissue Constructs: a Review. Tissue Engineering Part B: Reviews
16:587-601.
McMahon L. 2007. The effect of cylic tensile loading and growth factors on the chondrogenic
differentiation of bone-marrow derived mesenchymal stem cells in a collagen-
glycosaminoglycan scaffold.
Morey E, Baylink D. 1978. Inhibition of bone formation during space flight. Science 201:1138-1141.
Murphy CM, O’Brien FJ. 2010. Understanding the effect of mean pore size on cell activity in
collagen-glycosaminoglycan scaffolds. Cell Adh Migr 4:377-81.
Murphy CM, Haugh MG, O’Brien FJ. 2010. The effect of mean pore size on cell attachment,
proliferation and migration in collagen-glycosaminoglycan scaffolds for bone tissue
engineering. Biomaterials 31:461-466.
O’Brien FJ, Harley BA, Yannas IV, Gibson LJ. 2005. The effect of pore size on cell adhesion in
collagen-GAG scaffolds. Biomaterials 26:433-441.
O’Brien FJ, Harley BA, Waller MA, Yannas IV, Gibson LJ, Prendergast PJ. 2007. The effect of pore
size on permeability and cell attachment in collagen scaffolds for tissue engineering.
Technology and Health Care 15:3-17.
O’Brien FJ, Harley BA, Yannas IV, Gibson L. 2004. Influence of freezing rate on pore structure in
freeze-dried collagen-GAG scaffolds. Biomaterials 25:1077-1086.
Partap S, Plunkett N, Kelly D, O’Brien F. 2009. Stimulation of osteoblasts using rest periods during
bioreactor culture on collagen-glycosaminoglycan scaffolds. J Mater Sci: Mater Med
21:2325-2330.
Plunkett NA, Partap S, O’Brien FJ. 2010. Osteoblast response to rest periods during bioreactor culture
of collagen-glycosaminoglycan scaffolds. Tissue Eng Part A 16:943-951.
Raimondi MT, Moretti M, Cioffi M, Giordano C, Boschetti F, Laganà K, Pietrabissa R. 2006. The
effect of hydrodynamic shear on 3D engineered chondrocyte systems subject to direct
perfusion. Biorheology 43:215-222.
Sandino C, Planell JA, Lacroix D. 2008. A finite element study of mechanical stimuli in scaffolds for
bone tissue engineering. Journal of Biomechanics 41:1005-1014.
Sikavitsas VI, Bancroft GN, Holtorf HL, Jansen JA, Mikos AG. 2003. Mineralized matrix deposition
by marrow stromal osteoblasts in 3D perfusion culture increases with increasing fluid shear
forces. Proceedings of the National Academy of Sciences of the United States of America
100:14683-14688.
Sikavitsas V, Bancroft G, Lemoine J, Liebschner M, Dauner M, Mikos A. 2005. Flow perfusion
enhances the calcified matrix deposition of marrow stromal cells in biodegradable nonwoven
fiber mesh scaffolds. Annals of Biomedical Engineering 33:63-70.
St-Pierre J-P, Gauthier M, Lefebvre L-P, Tabrizian M. 2005. Three-dimensional growth of
differentiating MC3T3-E1 pre-osteoblasts on porous titanium scaffolds. Biomaterials
26:7319-7328.
Theret DP, Levesque MJ, Sato M, Nerem RM, Wheeler LT. 1988. The application of a homogeneous
half-space model in the analysis of endothelial cell micropipette measurements. J Biomech
Eng 110:190-199.
Tierney CM, Jaasma MJ, O’Brien FJ. 2009. Osteoblast activity on collagen-GAG scaffolds is affected
by collagen and GAG concentrations. Journal of Biomedical Materials Research Part A
91A:92-101.
Truskey GA, Proulx TL. 1993. Relationship between 3T3 cell spreading and the strength of adhesion
on glass and silane surfaces. Biomaterials 14:243-254.
Vance J, Galley S, Liu DF, Donahue SW. 2005. Mechanical stimulation of MC3T3 osteoblastic cells
in a bone tissue-engineering bioreactor enhances prostaglandin E2 release. Tissue Eng
11:1832-1839.
Wendt D, Marsano A, Jakob M, Heberer M, Martin I. 2003. Oscillating perfusion of cell suspensions
through three-dimensional scaffolds enhances cell seeding efficiency and uniformity.
Biotechnology and Bioengineering 84:205-214.
Xiao Y, Truskey GA. 1996. Effect of receptor-ligand affinity on the strength of endothelial cell
adhesion. Biophysical Journal 71:2869-2884.
Zerwekh JE, Ruml LA, Gottschalk F, Pak CYC. 1998. The effects of twelve weeks of bed rest on
bone histology, biochemical markers of bone turnover, and calcium homeostasis in eleven
normal subjects. Journal of Bone and Mineral Research 13:1594-1601.
Zhao F, Ma T. 2005. Perfusion bioreactor system for human mesenchymal stem cell tissue
engineering: Dynamic cell seeding and construct development. Biotechnology and
Bioengineering 91:482-493.
Figure Captions
Figure 1: Overview of the work undertaken in this publication. Computational and
experimental techniques were combined to produce a predictive mathematical model of cell
detachment in highly porous scaffolds that was validated against published literature and used
to provide insight into the role of cell morphology with respect to cell detachment.
Figure 2: Fold change in cell number as a function of scaffold pore size and flow-rate.
Cell detachment levels as a function of scaffold pore size (85 m, top; 120 m, middle; 325
m, bottom), flow-rate (, Static control; , 0.05mL/min; , 1mL/min and 5mL/min)
and exposure time were determined experimentally. A curve fitting approach was employed
to mathematically describe the data (see Supplementary Material) giving the illustrated
curves; static control (solid line); 0.05mL/min (dotted line); 1mL/min (medium dash) and
5mL/min (large dash). Error bars represent the standard deviation (n=2).
Figure 3: Pore size has no effect on mean shear stress, maximum shear stress or shear
stress distribution when porosity and flow-rate are kept constant. Scaffolds of different
mean pore sizes were reconstructed from CT images and 3 smaller sub-volumes were
extracted to reduce computational costs. All sub-volumes had a constant porosity (~90%).
CFD simulations showed no difference in (a) mean shear stress (b) maximum shear stress or
(c) shear stress distribution.
Figure 4: Comparison of predicted levels of cell detachment to published data. Predicted
levels () of cell detachment calculated from the derived predictive model (Equation 1) were
compared to published data, (Jaasma et al, 2008; ) for MC3T3s in 120m pore size
scaffolds for either a standard flow regime (top) consisting of cyclic bouts of 1h at 1mL/min
followed by 7h at 0.05mL/min, or low a flow regime (bottom), which consisted of a constant
flow-rate of 0.05mL/min. Error bars represent the standard deviation of the experimental
data.
Figure 5: High resolution scanning electron microscopy of cell morphology attachment
types in highly porous scaffolds. MC3T3 cells seeded on collagen-GAG scaffolds were
assessed by high resolution scanning electron microscopy to confirm the presence of both
bridged (c, d, e, f) and flat morphology (a, b, c) types prior to flow. Cells have been falsely
coloured to enhance their appearance. Bridging cells are shown in purple or orange and flatly
attached cells in green.
Figure 6: Comparison of experimentally observed cell morphology types and
computationally seeded cells. High resolution scanning electron microscopy images (A-C)
of bridging (purple) and flatly (green) attached cells and comparative images of numerically
cell-seed scaffolds (D,E) used for deformation modelling. Numerically cell-seeded scaffolds
were cropped three dimensionally to aid visualisation of the cells within the construct. Images
have been falsely coloured to aid cell visualisation.
Figure 7: Comparison between experimentally observed cell detachment and
computationally predicted levels as a function of deformation. Cell detachment observed
experimentally for 120m () and 325m () pore size scaffolds was compared to
computationally predicted levels as function of the percentage of cells exceeding a critical
threshold (30nm) of cell deformation for scaffold with 120m () and 325m () mean
pore sizes and for all flow-rates studied experimentally (0.05mL/min, 1mL/min and
5mL/min). Error bars show standard deviations of computational modelling studies (n=3).
Figure 8: Relationship between flow-rate, cell morphology and cell deformation.
Assuming cell morphology and adhesion profiles are equal in two scaffolds of the same pore
size, then if the minimum level of deformation (critical threshold, Tc) for detaching an entire
population (or sub-population) of cells is known (i.e all bridged cells) then reducing the flow-
rate will result in decreased shear stresses and reduced levels of cell deformation. Shifts in the
deformation profile with respect to the critical threshold will therefore reflect changes in the
proportion of the population that are detached.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8