Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
i
Prediction of Bone Cell
Probability Distribution in
Weak Electromagnetic Fields
Song Chen, B.Sc, M.Eng (Hons)
This thesis is submitted to fulfil the requirements of The Australian National
University for the degree of
Doctor of Philosophy
September 2018
© Copyright by Song Chen 2018
All Rights Reserved
ii
Statement
The work presented within this thesis holds no material or information that has been
accepted for the award in any university for any degree. To the best of my knowledge,
this thesis does not contain any material written by another person except for the places
denoted by specific references. The content of this thesis is the product of research
work carried out at The Australian National University, since the starting of this
research program.
Supervisory Panel:
Professor Qinghua Qin, Research School of Engineering, The Australian National
University.
A/Professor Rachel W. Li, John Curtin School of Medical Research/ The Medical
School, The Australian National University.
Professor Paul N. Smith, The Medical School, The Australian National University.
Song Chen
September 2018
iii
Acknowledgements
My deepest appreciation and sincere gratitude for the guidance and support from
Professor Qinghua Qin. You have been great mentors throughout my journey in
completing this thesis. The enthusiasm that you have for the research is admirable, and
I am thankful for the knowledge you have imparted to me throughout the years.
I am extremely grateful to A/Professor Rachel W Li for the opportunity to undertake
such an exciting project. Your positivity, compassion and understanding have
provided me with the motivation to go through challenging times. You are like the
compass that has navigated me to the right direction of research.
I wish to thank Professor Paul N Smith, for your support and guidance throughout the
years, which has enabled me to successfully get through scientific research. Your
knowledge in the orthopaedic surgery and the bone field often inspired me to think
from a different perspective.
I am truly blessed to have the support and assistance from my colleagues. To Bobin
Xin, Cheuk Yu Lee, Haiyang Zhou and Shuang Zhang, thank you for your endless
encouragement, concern, support, assistance and above of all, for the fun times.
The completion of this thesis would not have been possible without the love, faith and
support from my family. I am grateful to my parents for the unconditional support and
sacrifices that you both have made to give me the best in life. To my dearest Haoning
Feng, you filled my heart with so much love and I feel blessed to have you in my life.
iv
Abstract
Electromagnetic field (EMF) effects on the cell membrane level, general and specific
gene expression and signal pathways of bone cells have been examined in numerous
studies. These studies were conducted on bioprocesses such as cell proliferation, cell
cycle regulation, cell differentiation, and metabolism. Genotoxic effects and apoptosis
were observed during in vitro experiments. However, several observations after EMF
exposure have been irreproducible and contradictory with other studies. Especially,
statistic insignificance of EMF effects in bioprocesses occurred when comparing the
exposure group with the control. Corresponding to the inconsistent observations, the
types of EMF apparatus in vitro are various in parameters. In this thesis, the biological
effects of EMFs on osteoblasts and osteoclasts were examined in the exposure of the
static magnetic field (SMF) and pulsed electromagnetic field (PEMF). At the
preliminary study, the nonlinear dose-response relationship was observed between the
intensity of EMF exposure and cell proliferation of osteoblasts. A hypothesis was
proposed for seeking the interpretation of nonlinearity by the principle of interference.
Several frameworks were formulated for building a theoretical structure of the
hypothesis. The verification of the hypothesis was rooted in the experimental design.
Two gradients of SMFs were applied to osteoblastic Saos-2 cells, and the biological
data represented the interference of cell probability. In the PEMF experiment, the
interference was also found in the cell proliferation of osteoblasts when they were
affected by the PEMFs. The similar results were reproduced at the co-culture of
osteoblasts and osteoclasts in SMFs and PEMFs, which resulted in a unique
entanglement of cells.
v
Abbreviations
ALP Alkaline phosphatase
BMU Bone multicellular unit
Cat K Cathepsin K
DMSO Dimethyl Sulphoxide
ELF Extremely low frequency
EMF Electromagnetic field
FBS Foetal bovine serum
FDM Finite difference method
FDTD Finite-difference-time-domain
FEM Finite element method
FSC Forward scatter
IGFs Insulin-like growth factors
M-CSF Macrophage colony-stimulating factor
mL Millilitre
mM Millimolar
mT Milli-Tesla
PBS Phosphate buffered saline
vi
PEMF Pulsed electromagnetic field
PTH Parathyroid hormone
RANK Receptor activator of NF-κB
RANKL RANK ligand
RT-PCR Reverse transcription polymerase chain reaction
SEM Scanning electron microscope
SEMs Standard errors of the mean
SMADs Sma and Mad-related proteins
SMF static magnetic field
SSC Side scatter
TRAP Tartrate-resistant acid phosphatase
v/v Volume/volume
w/v Weight/volume
Wnts Wingless-type protein
α-MEM Α-Modification of Eagle’s Medium
µL Microlitre
µm Micrometre
µM Micromolar
vii
Table of Contents
Acknowledgements ..................................................................................................... iii Abstract ....................................................................................................................... iv Abbreviations ............................................................................................................... v Table of Contents ....................................................................................................... vii
List of Figures .............................................................................................................. x List of Tables.............................................................................................................. xii Chapter 1 Biological Effects of Electromagnetic Fields .............................................. 1
1.1 Characteristics of electromagnetic fields ..................................................... 2 1.1.1 Frequency ................................................................................................. 2
1.1.2 Amplitude ................................................................................................. 3
1.1.2.1 Amplitude of electric field ................................................................ 3 1.1.2.2 Amplitude of magnetic field ............................................................. 3
1.1.2.3 Maxwell’s equations......................................................................... 4 1.2 Types of EMF applications in vitro and in vivo ........................................... 5
1.2.1 SMF .......................................................................................................... 5 1.2.1.1 Weak SMF ........................................................................................ 6
1.2.1.2 Moderate SMF ................................................................................. 6 1.2.1.3 Strong and ultra-strong SMF ........................................................... 8
1.2.2 PEMF ....................................................................................................... 9 1.2.3 Animal studies ........................................................................................ 11
1.3 Why effects of EMF in biology are irreproducible and contradictory ....... 11
1.3.1 Mechanisms of EMF .............................................................................. 12
Chapter 2 Bone Remodelling and Applied Mathematical Modelling ........................ 16 2.1 Bone modelling and remodelling ............................................................... 17
2.1.1 Osteoblast ............................................................................................... 18 2.1.2 Osteoclast ............................................................................................... 19 2.1.3 Gene network between osteoblasts and osteoclasts ............................... 20
2.2 Mathematical models for simulating cell growth and movement .............. 22 Chapter 3 Research Question and Hypothesis ........................................................... 29
3.1 Research question....................................................................................... 30 3.2 Hypothesis ........................................................................ 31
Chapter 4 Materials and Biological Methods ............................................................ 36 4.1 Materials ..................................................................................................... 37
4.1.1 Chemical reagents .................................................................................. 37 4.1.2 Commercial kits and molecular products ............................................... 38
4.1.3 Other products and consumables ........................................................... 39 4.1.4 Cytokines ............................................................................................... 39 4.1.5 Solutions ................................................................................................. 39 4.1.6 Equipment .............................................................................................. 41
4.1.7 Software ................................................................................................. 42
4.1.8 RT2 profiler PCR array........................................................................... 42 4.2 PEMF apparatus design ............................................................................. 46 4.3 Culture and in vitro studies of Sao-2 cell line ............................................ 46
4.3.1 Saos-2 cell line ....................................................................................... 47 4.3.2 Initiation of culture process .................................................................... 47
4.3.3 Subculture Saos-2 .................................................................................. 47 4.3.4 Count cell densities of Saos-2 cells ........................................................ 48
4.3.5 Cryopreservation and retrieval of cultured cells .................................... 48
viii
4.4 Isolation, culture and in vitro studies of primary cells ............................... 49 4.4.1 Isolation and culture of osteoblasts from trabecular bone ..................... 49 4.4.2 Harvest of monolayer cultures ............................................................... 49 4.4.3 In Vitro osteoblasts cell mineral formation assay .................................. 50
4.5 Osteoclasts induction and culture .............................................................. 51
4.5.1 Preparation of solution ........................................................................... 51 4.5.2 PBMC isolation ...................................................................................... 51 4.5.3 PBMC cell culture .................................................................................. 51 4.5.4 TRAP staining ........................................................................................ 52 4.5.5 Flow cytometer and analysis .................................................................. 53
4.6 Western blotting ......................................................................................... 53 4.6.1 Protein extract and assay ........................................................................ 53
4.6.2 Blotting and detection ............................................................................ 54
4.7 RNA extraction and quantification ............................................................ 55 Chapter 5 The Response of Cell Probability to External Stimulus ............................ 61
5.1 Introduction ................................................................................................ 62 5.2 Preliminaries ........................................................................ 62
5.2.1 Distribution of cell density ..................................................................... 62
5.2.2 Distribution of cell density under external stimulus .............................. 68
5.3 Cell probability ......................................................................................... 69
5.4 Joint cell probability ................................................................................... 77 5.5 Particular correlation between osteoblasts and osteoclasts ........................ 82
Chapter 6 Numerical Model: Predication of Osteoblast Cell Population under
Influence of Surface Grain Size ................................................................................. 90
6.1 Introduction ................................................................................................ 91 6.2 The relationship between osteoblast proliferation and substrate surface
roughness ............................................................................................................... 91 6.3 Governing equations for osteoblast adhesion on the material surface ....... 93
6.4 Numerical solution methods for the double-layered model ....................... 95 6.5 Numerical simulation on experimental data ............................................... 99 6.6 Comparison with experimental data and prediction ................................. 101
6.7 Parametric analysis................................................................................... 103 Chapter 7 Weak SMF Effects on Osteoblastic Cell Proliferation ............................ 118
7.1 Introduction .............................................................................................. 119 7.2 Experimental design ................................................................................. 119
7.3 Experimental results ................................................................................. 120 7.3.1 Effect of the SMF on the orientation of cultured Saos-2 cells ............. 120
7.3.2 Effect of SMF intensity on the proliferation of cultured Saos-2 cells . 121 7.3.3 Effect of SMF intensity on the cell viability of cultured Saos-2 cells . 121 7.3.4 The sensitivity of cultured Saos-2 cell proliferation to SMF intensity 121
7.4 A numerical model of the dose-dependent effect of SMF on osteoblastic
cell proliferation ...................................................................... 122
7.4.1 A stochastic model for cell probability in SMF ................................... 122 7.4.2 Numerical results ................................................................................. 124 7.4.3 Interference in SMF ............................................................................. 125
7.5 Discussion ................................................................................................ 125 7.5.1 Signal-to-noise ..................................................................................... 126
7.5.2 Physical mechanisms ........................................................................... 127 Chapter 8 PEMF Effects on Cell Proliferation of Osteoblasts ................................. 143
8.1 Introduction .............................................................................................. 144
ix
8.2 Experimental designs ............................................................................... 144 8.3 Experimental results ................................................................................. 145
8.3.1 Effect of PEMF intensity on the proliferation of cultured Saos-2 cells145 8.3.2 Effect of PEMF frequency on the proliferation of cultured Saos-2 cells ...
.............................................................................................................. 145
8.3.3 The sensitivity of cultured Saos-2 cell proliferation to PEMF ............ 146 8.3.4 Effect of PEMF on the orientation and mineralisation of human
osteoblasts ........................................................................................................ 147 8.3.5 Real-time RT-PCR of osteogenesis-related genes ............................... 147
8.4 Numerical solution of PEMF effects on osteoblastic cell proliferation .. 148
8.4.1 Model derivation for PEMF ................................................................. 148 8.4.2 Interference in PEMF ........................................................................... 149
8.5 Discussion ................................................................................................ 150
Chapter 9 PEMF and SMF Effects on Co-culture of Osteoblasts and Osteoclasts .. 167 9.1 Introduction .............................................................................................. 168 9.2 Prediction ................................................................................................. 169
9.2.1 The numerical interaction between osteoblasts and osteoclasts .......... 169 9.2.2 Numerical results ................................................................................. 172
9.3 Validation in experimental images........................................................... 172
9.3.1 TRAP staining ...................................................................................... 173 9.3.2 Fluorescence-based staining and flow cytometry analysis .................. 173 9.3.3 Scanning electron microscope (SEM) images ..................................... 174
Chapter 10 Conclusions ........................................................................................... 184 References ................................................................................................................ 187
Appendix .................................................................................................................. 202
x
List of Figures
Figure 2.1 BMP-SMAD and TGF-β/Activin pathway.. ............................................ 26 Figure 2.2 Map of osteo-gene network ..................................................................... 28 Figure 3.1 Double-slit model for osteoblasts ............................................................ 35 Figure 4.1 Self-made EMF device in controlling the frequency and signal type ...... 57
Figure 4.2 Coil tube ................................................................................................... 58 Figure 4.3 Picture of main body ................................................................................ 59 Figure 4.4 Support for each tube ............................................................................... 60 Figure 5.1 The structure factor .................................................................................. 89 Figure 6.1 Sketch of the double-layered model for osteoblast adhesion on the
material surface ................................................................................................ 107
Figure 6.2 Two examples of model fitting by experiment data .............................. 108 Figure 6.3 Function determination for molecular and cellular sensitivity in grain size
.......................................................................................................................... 110 Figure 6.4 Numerical models for Alumina and Titania .......................................... 111 Figure 6.5 Comparison of numerical results and experimental data.. ..................... 112 Figure 6.6 Simulation of molecular sensitivity ....................................................... 114
Figure 6.7 Simulation of cellular sensitivity ........................................................... 116 Figure 6.8 The impact of size factor ....................................................................... 117
Figure 7.1 Representation of the device used to generate the SMFs ...................... 130 Figure 7.2 Cell morphology of osteoblastic Saos-2 in SMF and control ................ 131
Figure 7.3 Effect of the SMFs on the proliferation of osteoblastic Saos-2 ............. 134
Figure 7.4 The effect of SMF exposure on cell viability of osteoblastic Saos-2 .... 135
Figure 7.5 Effect of the SMFs on the cell proliferation of osteoblastic Saos-2 in
combined G1+G2 ............................................................................................. 136
Figure 7.6 The sketch of the mathematical model for the movement of osteoblastic
cells in vitro under SMFs ................................................................................. 138 Figure 7.7 The predicted pattern of osteoblastic Saos-2 cell proliferation in weak
SMFs ................................................................................................................ 140 Figure 7.8 Parametric analysis for SMF coefficient ............................................... 142
Figure 8.1 Representation of the device used to generate the PEMFs .................... 153 Figure 8.2 Effect of the PEMF intensity on the proliferation of cultured Saos-2 cells
.......................................................................................................................... 154 Figure 8.3 Effect of the PEMF frequency on the proliferation of cultured Saos-2
cells .................................................................................................................. 155 Figure 8.4 Data distribution of PEMF groups along the control ............................. 156
Figure 8.5 A comparison between cultured Saos-2 cell proliferation to PEMF
generated by a solenoid and that generated by Helmholtz coils ...................... 157 Figure 8.6 PEMF effect on the orientation of human osteoblasts ........................... 158 Figure 8.7 Effect of PEMF and SMF on osteogenesis of human osteoblasts assessed
by Alizarin red-stained calcified nodules ......................................................... 159
Figure 8.8 Western blot analysis of IGF-1 protein at a loading of 10 and 30 µg ... 160 Figure 8.9 Waveform of frequency ......................................................................... 161 Figure 8.10 The interference of cell probability in PEMF ...................................... 162 Figure 8.11 Implicit functions of coefficient 1 and 2 .............................................. 164 Figure 9.1 Imaginary grid of osteoblasts and osteoclasts........................................ 175
Figure 9.2 Prediction of the relative position between osteoblasts ......................... 176 Figure 9.3 TRAP staining on osteoclasts performed in co-culture of human
osteoblasts and osteoclast in control, SMF and PEMF exposure .................... 179
xi
Figure 9.4 The fluorescence-based staining in control, SMF and PEMF exposure.
Compared with the control, the SMF and PEMF exposure prevented the
forming of mature osteoclasts .......................................................................... 181 Figure 9.5 Scanning electron microscope (SEM) images of osteoblasts and
osteoclasts co-culture in control, SMF and PEMF exposure ........................... 183
xii
List of Tables
Table 1-1 Animal studies .......................................................................................... 15 Table 8-1 The effects of SMF exposure on osteogenesis-related genes of human
osteoblasts ........................................................................................................ 165
Table 8-2 The effects of PEMF exposure on osteogenesis-related genes of human
osteoblasts ........................................................................................................ 165 Table 8-3 The frequency functions.......................................................................... 166
1
Chapter 1
Biological Effects of Electromagnetic Fields
2
1.1 Characteristics of electromagnetic fields
Bioelectrical phenomena play a vital role in bioprocesses by the separation, transport
and storage of electrical charge. The participation of electron transfer is essential for
electrochemical communication between molecules [1]. Electromagnetic fields
(EMFs) can influence the mobility of electron by induced forces in the structure of
matter, which might be responsible for biological effects ranging from increased
enzyme reaction rates to transcript levels of specific genes [2]. The electromagnetic
spectrums of EMFs interact with biological systems by various wavelengths in term
of frequency and amplitude.
1.1.1 Frequency
In unit of Hertz (Hz), frequency in an electromagnetic spectrum is categorized from
extremely low frequency (ELF) (0-103 Hz), radiofrequency (RF) (103-108 Hz),
microwaves (~109 Hz), infrared (~1012 Hz), visible light, ultraviolet, X-rays to gamma
rays [3]. Electromagnetic spectrum with a high order of frequencies is classified as
ionising radiation, while the radiation insufficient to break molecular bonds is
classified as non-ionizing radiation. Gamma rays, X- rays are examples of ionising
radiation. Radiations from microwaves, RF and ELF-EMFs are examples of non-
ionizing radiation. Therefore, non-ionizing radiation ranges from 0 to approximately
1011 Hz, while ionising radiation is considered above 1011 Hz [4]. The energy of
electromagnetic radiation G , is associated with frequency v by
G hv= , (1.1.1)
where h represents Planck’s constant.
3
1.1.2 Amplitude
The amplitude of an EMF wave comprises of electric and magnetic components. The
electric and magnetic components occur an EMF spontaneously by motions of charged
objects. The EMF weakens with increasing distance from the charged objects [5].
Generally, an EMF produces an electric field by the potential difference. The magnetic
field is created due to the electric current flowing in a conductor. Higher potential
difference correlates to the stronger electric field. Higher electric current value
correlates to the stronger magnetic field.
1.1.2.1 Amplitude of electric field
An electric field can be represented mathematically as a vector field E. The vector
field has a value defined at each point of space and time and is in a function of space
and time coordinates. The magnitude of electric field E is correlated with the vector of
electrical potential Φ ,
= −E Φ , (1.1.2)
with a solution for a uniform field in which the electric field is constant at each point,
Ed
= − , (1.1.3)
where d is the distance between potential difference, and the unit of the electric field
is measured in V/m.
1.1.2.2 Amplitude of magnetic field
A magnetic field can also be represented as a vector field B or H, which describes the
magnetic influence of electric currents and magnetised materials. The magnetic flux
density B is measured in units of Gauss (G) or Tesla (T) (1T= 10,000G), and the field
intensity H is measured in amperes per meter. In a vacuum, B and H are the same. In
4
a magnetised material, magnetic flux density B is proportional to field intensity H and
related by
=B H , (1.1.4)
where is the permeability of the magnetised material. In EMF, the magnetic field
generated by a steady current I is described by
0
2
ˆd
4
IB
r
=
l r , (1.1.5)
where dl is the vector line element with the same direction of the current I, 0 is the
magnetic constant and r is the distance between the location of dl and the location of
the measured magnetic field in the direction r .
1.1.2.3 Maxwell’s equations
If the electric field E is non-zero and constant in time, it is an electrostatic field.
Similarly, if the magnetic field B is non-zero and constant in time, it is a magneto-
static field. When either the electric or magnetic field is time-dependent, both fields
are considered as a coupled EMF governed by Maxwell’s equations. Coupled EMF
wave consists of electrical field E and magnetic field B vibrating in phase and
perpendicular to the direction of propagation. The Maxwell’s equations introduce the
electric and magnetic field with a time and location dependence,
0
02
0
0
1t
c t
+ =
− =
=•
•=
BE E
EB
B J
(1.1.6)
5
in which sources are represented by electric charge density and current density J.
The universal constants are the permittivity of free space 0 and the permeability of
free space 0 .
1.2 Types of EMF applications in vitro and in vivo
Numerous studies have examined EMF effects on the cell membrane level, general
and specific gene expression, and signal transduction pathways. These studies have
been conducted on bioprocesses such as cell proliferation, cell cycle regulation, cell
differentiation, and metabolism. However, several observations after EMF exposure
have been irreproducible and contradictory in other studies. Especially, statistic
insignificance of EMF effects in bioprocesses occurred in the comparison between the
exposure group and the control. Besides, the types of EMF apparatus in vitro and in
vivo are utilised in various parameters. In the following paragraphs, the biological
effects of EMFs are examined respectively in the term of the static magnetic field
(SMF) and pulsed electromagnetic field (PEMF).
1.2.1 SMF
SMF generated from the geomagnetic field is closely related to living and evolution
for organisms on the earth [6]. SMF can be produced by either a permanent magnet or
direct current (DC) electricity in EMF which has constant magnetic flux density over
the time interval. Man-made SMF is not associated with induced electric currents
except during activation and deactivation [7]. According to the magnetic flux density,
SMF is classified as weak (<1 mT), moderate (1mT to 1T), strong (1T to 5T), and
ultra-strong (>5T) [8]. The biological effects of SMFs have been extensively examined
in several biological systems although with inconsistent results.
6
1.2.1.1 Weak SMF
A weak SMF exposure of 20 μT for 30 minutes altered Ca2+ transport compared to
zero magnetic field exposure in cell-free conditions, measured by 660 nm absorbance
for calmodulin-dependent cyclic nucleotide phosphodiesterase activity [9]. Weak SMF
of 120 μT increased the proliferation of human umbilical vein cells by 40% throughout
2 days. Buemi, et al. [10] examined the effects of 0.5 mT SMF on the balance between
cell proliferation and death in renal cells and cortical astrocyte cultures from rats. After
2, 4 and 6 days of exposure to the SMF, they observed a gradual decrease in apoptosis
and proliferation, while a gradual increase in cells with a necrotic morphology
compared to the control group. Sonnier, et al. [11] measured transmembrane Na+ and
K+ currents of the action potential in SH-SY5Y neuroblastoma cells exposed to SMFs
of 0.1 and 0.5 mT. Application of the magnetic fields did not result in detectable
changes in any of the parameters of the action potential, suggesting that the studied
SMFs did not affect the cellular mechanisms responsible for generating the action
potential.
1.2.1.2 Moderate SMF
Exposure to a 120 mT SMF resulted in a reduction in the peak calcium current
amplitude and a shift in the current-voltage relationship in cultured GH3 cells [12].
When exposed to a 125 mT SMF, a less than 5% reduction in peak current in voltage-
activated Na+ channels was measured in GH3 [13]. Human osteosarcoma cell line
MG-63 became stellar shapes and formed multiple layers after exposed to SMF of 400
mT for 24 hours. The differentiation of MG-63 cells at 1-3 days was promoted with
increased expression of Alkaline phosphatase (ALP) [14]. Chiu, et al. [15] reported
that the proliferation of MG-63 was inhibited by SMF with average magnetic flux 100,
250 and 400 mT after 24 hours. The significant difference in cell population was only
7
observed between 400 mT SMF exposed group and control, with the reduction of cell
membrane fluidity. Cunha, et al. [16] analysed MG-63 after exposure to a 320 mT
SMF either continuously or by 1 hour at any 24 hours interval, demonstrating that SMF
significantly reduced the cell proliferation in both treatment manners after 7 days by
MTT assay. They also showed the absence of SMF effects on cell membrane integrity,
morphology and cytoskeleton organisation after 7 days. Feng, et al. [17] observed the
morphology change of MG-63 on a poly-L-lactide substrate when exposed to a 400
mT SMF for 24 hours. The effects of SMF 500 mT and 1 T on cells from human blood
were investigated by examining their influence on the frequency of gross lesions, sister
chromatid exchanges and on the proportion of amodal cells. Neither treatment had a
significant effect on any of the parameters measured [18]. Pacini, et al. [19] examined
morphological changes caused by exposure to a 200 mT SMF on human neuronal cell
culture (FNC-B4). The results showed dramatic changes in morphology in which
vortexes of cells were formed and exposed branched neurites featuring synaptic
buttons. Endothelin-1 release from FNC-B4 cells was also dramatically reduced after
5 minutes of exposure. They also reported that human skin fibroblast cell morphology
was modified with a concomitant decrease in the expression of some sugar residues of
glycoconjugates after 1-hour exposure to a 200 mT SMF [20]. Teodori, et al. [21]
investigated the exposure of HL-60 cells to SMF of 6 mT with or without DNA
topoisomerase I inhibitor, camptothecin for 5 hours. The SMF alone did not produce
any apoptogenic or necrogenic effect in HL-60. In combination with camptothecin,
SMF did not affect overall cell viability but accelerated the rate of cell transition from
apoptosis to secondary necrosis after induction of apoptosis by camptothecin.
8
1.2.1.3 Strong and ultra-strong SMF
Strong and ultra-strong SMFs are of sufficient intensity to alter the preferred
orientation of a variety of diamagnetic anisotropic organic molecules. Matrix proteins
such as fibrin fibres are orientated to the SMF of 8 T [22]. Kotani, et al. [23] reported
that cultured mouse osteoblastic MC3T3-E1 cells were transformed to rod-like shapes
and oriented in the direction parallel to the magnetic field 8 T after 60 hours exposure.
The SMF exposure did not affect cell proliferation of MC3T3-E1, but up-regulated
cell differentiation and matrix synthesis which was determined by ALP and alizarin
red staining. Wiskirchen, et al. [24] reported that population doublings and cumulative
population doublings of human fetal lung fibroblasts had an insignificant difference in
statistics between exposed and control cell groups after 9 hours exposure to an SMF
of 1.5 T in three weeks. Clonogenic activity, DNA synthesis, cell cycle, and
proliferation kinetics were not altered by an SMF exposure of 1.5 T. The SMF
exposure of 7 T after 64 hours produced a reduction in viable cell numbers in
melanoma, ovarian carcinoma and lymphoma cell lines. Prolonged exposure to the
SMF of 7 T slowed the growth of human cancer cells in vitro. Alterations in the cell
growth cycle and gross fragmentation of DNA were excluded as possible contributory
factors [25]. Aldinucci, et al. [26] investigated whether SMF at a flux density of 4.75
T, generated by an NMR apparatus (NMRF), could promote movements of Ca2+, cell
proliferation and the production of proinflammatory cytokines in human peripheral
blood mononuclear cells (PBMC) after exposure to the field for 1 hour. The same study
was also performed after the activation of cells with 5 mg/mL phytohemagglutinin
(PHA). The results demonstrated that the SMF had neither proliferative nor
proinflammatory effects on normal and PHA activated PBMC. The concentration of
interleukin‐1β (IL-1β), interleukin‐2 (IL-2), interleukin‐6 (IL-6), interferon and
9
tumour necrosis factor α (TNFα) remained unvaried in the exposed cells. Exposure to
SMF with various magnetic flux densities of less than 1.6 T had no significant effect
on either active or passive Rb+ influxes, the morphology of HeLa cells [27]. Gradient
magnetic fields of 6 T affected the convection of floating cell aggregations in the cell
culture flask, and reversibly changed the direction of conventional flow [28]. Hirose,
et al. [29] reported that human glioblastoma A172 cells embedded in collagen gels
were oriented perpendicular to the direction of the static magnetic field at 10 T. A172
cells cultured in the absence of collagen did not exhibit any specific orientation pattern
after 7 days exposure to the static magnetic field. Eguchi, et al. [30] observed that
cultured Schwann cells from dissected sciatic nerves of neonatal rats were oriented
parallel to the magnetic field at 8 T after the exposure for 60 hours.
1.2.2 PEMF
There is ongoing interest in the application of PEMF radiation as an alternative non-
invasive therapy for curing bone disease. PEMF is referred to as time-varying EMF
consisting of specific or arbitrary waveforms with pulse modulated frequency [31].
Significant numbers of peer-reviewed publications have demonstrated that PEMF
radiation facilities the process of wound repairs. For instance, low-frequency
sinusoidal waveforms from 10 Hz up to 500 Hz have been shown to enhance healing
when used as adjunctive therapy [32]. In PEMF, magnetic flux density changes at one
or more frequencies. A PEMF, produced by alternating current (AC) electricity, is
significantly more dynamic than SMF and can induce an electric charge in tissues
which creates a cascade of physiologic effects [33]. In past decades, efforts have been
made to elucidate the exact SMF’s effect on osteoblasts, but this topic is challenged
by the fact that the in vitro effects of SMF highly depends on cell type [34], magnetic
field intensity and modes of application [8]. For example, PEMF exposure induced
10
cell proliferation of MG-63 with 2.3 mT and 75 Hz [35]. PEMFs determine signal
transduction using the intracellular release of Ca2+ leading to an increase in cytosolic
Ca2+ and an increase in activated cytoskeletal calmodulin [36]. PEMFs induce a dose-
dependent increase in cartilage differentiation [37], and upregulation of mRNA
expression of extracellular matrix molecules, proteoglycan, and Type II collagen
[38]. The acceleration of chondrogenic differentiation is associated with the
increased expression of TGF-β1 mRNA and protein, suggesting the stimulation of
TGF-β1 may be a mechanism through which PEMFs affect complex tissue
behaviours such as cell differentiation, and through which the effects of PEMFs may
be amplified [39]. PEMFs also are postulated to affect membrane level by
influencing signal transduction of several hormones or growth factors such as
parathyroid hormone, IGF-2, producing the amplification of their transmembrane
receptors [40].
Extremely low-frequency electromagnetic fields (ELF-EMFs), with tissue gradients in
the range from 10-7 to 10-1 V/cm, are involved in essential physiological functions in
mammals. Numerous studies have addressed the effects of ELF-EMFs involving a
wide spectrum of calcium-dependent processes, such as cell membrane functions in
bone growth [41] and regulation of intercellular communication [42]. Noriega-Luna,
et al. [43] investigated the effect of a magnetic flux density of 0.65 mT and frequency
of 4 Hz on the proliferation of MG-63, observing a slight increase in cell number and
subtle difference in osteoblast morphology after 48 hours treatment. Zhou, et al. [44]
demonstrated that the effects of 50 Hz sinusoidal PEMF of different magnetic flux
density, from 0.9 to 4.8 mT with an interval of 0.3 mT, inhabited osteoblast
proliferation and promoted their differentiation and mineralisation with peak activities
at 1.8 and 3.6 mT. De Mattei, et al. [45] introduced an experiment showing a
11
correlation between PEMF exposure time and cell proliferation increase in human
osteosarcoma cell line MG-63 and human normal osteoblast cells. The results
indicated that a short PEMF exposure of 30 minutes could stimulate the cell
proliferation in MG-63 and the normal human osteoblasts in 6 to 9 hours. However,
an increase in the length of exposure time resulted in an insignificant difference in cell
proliferation.
1.2.3 Animal studies
Few clinical studies concern EMF therapy, but animal studies have been carried out to
determine the effectiveness of EMF on bone growth. These studies indicate that EMF
can contribute to bone formation and healing process in various manners of magnetic
flux density, frequency and exposure duration. Representative animal studies in recent
years are summarised in Table 1 and generally show some positive effects.
1.3 Why effects of EMF in biology are irreproducible and contradictory
Cells and tissues are affected by the magnetic field of EMF, and not all EMF exposures
lead to an altercation at the cellular level. Some effects are only noted at discrete
frequencies and amplitude of the magnetic field. Others depend on the strength,
orientation and duration of the exposed field. The effects of EMFs on biological media
have been studied by many researchers using a variety of in vitro exposure systems
[46, 47]. Magnetic field therapy conducted by EMF is considered beneficial for
different diseases, especially those involving bones [48]. EMF stimulation has been
investigated as a therapy for wound healing following results that EMFs can promote
healing by potentially increasing collagen synthesis, angiogenesis, and bacteriostatic
[49]. Commercial EMF stimulators have been used to promote bone healing, with the
setting of EMF parameters varying significantly. Laboratory-based exposure systems
for studying EMF effects on biological samples utilise Helmholtz coils to generate
12
uniform electromagnetic exposures, but there are discrepancies between the set of
magnetic flux density and frequency. Hence, researchers argued that the controversial
effects of EMF on biological objects were due to experiments not carried out in well-
defined conditions [50]. Extensive studies have been done to categorise the parameters
of EMF for optimal laboratory settings [51]. EMF signal induces electric and magnetic
signal to initiate a cascade of biochemical reactions [52]. The EMF stimulation is
characterised by factors such as magnetic field intensity, waveform, dose-response
pattern [53], exposure duration [54], localisation of stimulation and spatial orientation
of the exposure system [55]. Other factors can influence the response to magnetic field
exposure by frequency and modulation, field uniformity, a combination of coil system
and precise placement, field intensity and polarisation, noise, vibration and
temperature of conducting materials, voltage carrying wires and metal equipment,
construction materials and the experiment schedule [56]. Therefore, EMF parameters
are proposed to be considered prior to evaluating the effect of magnetic field on a
biological sample in aspect of (i) type of magnetic fields, (ii) magnetic flux density,
(iii) frequency, (iv) exposure duration, (v) pulse shape, (vi) spatial gradient (dB/dx)
and temporary gradient (dB/dt) [57].
1.3.1 Mechanisms of EMF on biological systems
The mechanisms by which EMF affects biological systems are not fully elucidated. A
quantitative approach was presented in understanding the interaction between electric
fields and biological systems [58]. Biological samples react to external electrical
stimulation through a complex series of specific and non-specific responses [59]. The
specific response is determined by the physical nature of the stimulation, while the
non-specific response depends upon the intrinsic features of the organism system.
Electrical signals are the basis of information transportation in the nervous system
13
[60]. These electrical signals in the form of minute electric currents flow around and
within the cells and are of critical importance for their normal functioning and can
accelerate normal cellular function such as endocytosis [61]. EMF perturb these
currents and charges and positively influence the process of cellular functioning. It has
been suggested that EMFs may trigger specific, measurable cellular responses such as
DNA synthesis, transcription, and protein synthesis by altering or augmenting pre-
existing endogenous electrical fields [62]. This mechanism adds further evidence to
the fact that external non-invasive electric stimulation is a potent tool in augmentation
of cells, tissue and organs. A well-established phenomenon in physics, ion cyclotron
resonance (ICR) and ion parametric resonance (IPR) models were proposed, which
state that ions resonate when exposed to a specific combination of alternating and static
magnetic fields [63]. Hall effect may provide the electrical basis of EMF in the bone
lacuna canalicular system, which attaches the positive ions at the negative interface of
the bone matrix [64]. When EMF is applied, the moving charged particles encounter a
Lorentz fore perpendicular to their direction. Cations accumulate at the downward
surface, and anions go upward to forming a hall voltage [65].
The magnetic control operates in biology by strong and weak interactions. The strong
magnetic interaction, such as magnetic resonance imaging, requires a strong and
durable magnetic field. It can be manifested in processes involving particles and
membranes. If these particles and membranes have anisotropy of magnetic
susceptibility, the energy of the strong magnetic interaction will result in re-orientation
of particles and deformation of membranes. Consequently, the properties and the
chemical reactivity of particles and membranes change with the re-orientation and
deformation [66]. Evidence suggested that cell membrane is in response to transducing
EMF [67] where EMF can interact with moving charges by Lorentz forces [68].
14
However, an argument exists against Lorentz forces in cells since dielectric media and
plasma lack moving charges [69]. Buchachenko and Kuznetsov [70] emphasised the
molecular radical pair paradigm as a reliable basis for understanding and deliberately
using biochemical magnetic effects in medicine. Biochemical processes are
accompanied by generation or participation of ion-radicals in pairs. Radical pair
mechanism implies that two radicals are produced simultaneously with paralleled or
anti-paralleled electron spins. Chemical reactions are spin selective which are allowed
only for those spin states of reactants with identical total spins. The spin states of the
pair, singlet and triplet are different in chemical reactivity but identical in structure. In
triplet state, the reactions are forbidden. Weak magnetic interaction might provide a
manner to overcome the spin prohibition of processes in biochemistry. Magnetic
interactions induce singlet-triplet spin conversion and switch over the reaction
between spin-allowed and spin-prohibited channels, controlling the reaction pathways
and chemical reactivity. Radical pair mechanism has been shown functionally on the
molecular level in biochemical reactions [71]. Three types of magnetic interactions are
considered for catalysing chemical and biochemical reactions, namely Zeeman
interaction, Fermi interaction and microwaves. Magnetic catalysis is controllable and
switchable by using magnetic isotopes or paramagnetic ions [72]. The magnetic
interaction effects on the enzymatic ATP synthesis were detected for the creatine
kinase in vitro [73]. The magnetic control of enzymatic DNA synthesis was observed
in time-varying magnetic fields [74].
15
Table 1-1 Animal studies
Animal Aim Magnetic flux
density/Frequency
Exposure
duration
Results Reference
10-week-old
Wistar
female rats
Recovery of
osteoporosis
Gradient SMF
180 mT
3 weeks Significantly increased the BMD values of the
osteoporotic lumbar vertebrae in ovariectomized rats
without significantly influencing the E2 levels.
[75]
12-week-old
Sprague Dawley
male rats
Type 1 diabetes
mellitus (T1DM)
Helmholtz coils
4 mT
16 weeks Significantly prevented the deterioration of bone
architectural deterioration and strength reduction,
promotion of bone formation and weak modulation of
bone resorption.
[76]
12-week-old
Wistar albino
male rats
Chronic exposure
to bone
MRI
1.5 T
8 weeks Induced low-frequency fields within the tissues which
could exceed the exposure limits necessary to
deteriorate bone microstructure and vitamin D
metabolism.
[77]
20-week‐old
Sprague Dawley
male rats
Hyperthyroidism‐
induced
osteoporosis
Helmholtz coils
15 Hz/1 mT
12 weeks Significantly inhibited bone loss and microarchitecture
deterioration in hyperthyroidism rats, which might
occur due to reduced THR expression.
[78]
4-week-old Wistar
female rats
Effects of
exposure duration
Solenoid
50 Hz/1.8 mT
8 weeks Comparing the effects of 0.5, 1.0, 1.5, 2.0, 2.5, and
3.0 hours/day, 1.5 hours/day was the optimal exposure
duration to increase the peak bone mass of young rats.
[79]
12-week-old
Wistar albino
male rats
Bone fracture
healing
Helmholtz coils
50 Hz/1.5 ± 0.2 mT
4 weeks The EMF had a positive but modest effect on bone
fracture healing.
[80]
12-week-old
Sprague Dawley
male/female rats
Bone loss Helmholtz coils
8 Hz/3.8 mT
12 weeks The EMF prevented the diabetes-induced bone loss
and reversed the deterioration of bone
microarchitecture by restoring Runx2 expression
through regulation of Wnt/β-catenin signalling.
[81]
16
Chapter 2
Bone Remodelling and Applied
Mathematical Modelling
17
2.1 Bone modelling and remodelling
Bone is a continuously updated tissue and constituted mainly by BMSCs, osteoblast,
osteocytes and osteoclasts. The dynamic balance between bone formation and
resorption, such as bone modelling and remodelling, has a pivotal role to play in the
normal bone metabolism, bone integrity and appropriate bone strength [82]. Bone
modelling and remodelling are both involved in osteogenesis and skeletal growth.
Bone modelling is characterised by the process of bone formation and resorption with
a net increase in bone mass. The mechanism involves activation-formation and
activation-resorption, which primarily occurs in childhood [83]. Activation is signalled
by local tissue strain and involves the recruitment of progenitor cells that differentiate
into mature osteoblasts or osteoclasts. Once the appropriate cell population is
activated, the processes of formation and resorption happen until sufficient bone mass
is altered for normalising local strains. In contrast, bone remodelling is defined as
renewing and maintaining bone in which a coupled process occurs between the
catabolic effects of bone-resorbing osteoclasts and anabolic effects of bone-forming
osteoblasts [84]. During a bone remodelling process, bone mass is removed at sites
where the mechanical loads are low, while bone is formed where mechanical stimuli
are transmitted repeatedly.
Bone, therefore, can maintain itself, depending on the external mechanical and
physiological stimuli from the systemic environment [85]. The process of bone
remodelling is executed by a temporary anatomic structure incorporating a cohort of
cells known as the bone multicellular unit (BMU). The BMU mainly consists of
osteoblasts, osteoclasts and osteocytes. Bone-lining cells, osteomacs and vascular
endothelial cells have also been reported to be associated with the BMU. Bone is a
highly vascularized organ, and networks of blood capillaries have been observed at the
18
site of the BMU [86]. Vascularisation and angiogenesis are prerequisites for bone
formation and remodelling. They serve multiple purposed bone cell progenitors via
blood capillaries to establish the modelling and remodelling site. The blood supply
also allows the BMU to become accessible to immune cell infiltration and interaction
with bone cells.
Bone remodelling involves the balance between osteoclast and osteoblast activity,
which is regulated by numerous signalling pathways [87]. Deregulation of signalling
pathways in bone cells may lead to bone diseases such as osteoporosis and
osteoarthritis. In the big picture, osteoblasts facilitate bone formation by laying down
a matrix which subsequently is mineralised, and produce receptor activator of nuclear
factor - B ligand (RANKL) initiating osteoclastogenesis [88]. Osteoclasts, activated
by proinflammatory cytokines and RANKL/macrophage colony stimulating factor
(M-CSF), initiate bone resorption by releasing catalytic enzymes like Cathepsin K (Cat
K) and matrix metalloproteases (MMPs) in a resorptive pit formed on the bone surface
[89]. Protection against bone damage can be achieved by (i) inhibiting RANKL
production by increasing the production of osteoprotegerin (OPG), (ii) suppressing the
proinflammatory cytokines, (iii) inhibiting production of Cat K and MMPs, and (iv)
inhibiting osteoclast formation. The cellular coupling between osteoblasts and
osteoclasts is complex and is massively coordinated by several regulatory systems to
keep both remodelling and resorption processes synchronised. The accentuation of one
or the other process eventually leads to bone fragility and clinical diseases of the
skeleton, such as osteoporosis, arthritis and osteolysis [90].
2.1.1 Osteoblast
Osteoblasts are the primary cells responsible for bone formation. They originate from
mesenchymal stem cells, which have the potential to differentiate into other
19
musculoskeletal tissues such as cartilage, fat, muscle, ligament and tendon [91]. The
commitment of these stem cells to the osteoblast lineage is highly driven by the growth
factors wingless-type proteins (Wnts) and bone morphogenetic proteins (BMPs). Once
skeletal stem cells are committed to the osteoblast lineage, the proliferation of
osteoblast precursor cells begins. These cells produce type I collagen for the basic
building block of bone, osteocalcin and alkaline phosphatase for mineral deposition.
Osteoblast precursor cells mature into cuboidal osteoblasts. Besides their role in bone
formation, the osteoblasts are involved in the recruitment and maintenance of
osteoclasts by expressing M-CSF, RANKL and OPG [92]. Mature osteoblasts may
undergo apoptosis or coalesce into the heterogeneous population bone-lining cells, or
eventually become encased and trapped within the matrix to become osteocytes. The
proportion of osteoblasts following each possibility varies in all mammals and is not
conserved among different types of bone. The age of the mammal may also influence
the number of osteoblasts that transform into osteocytes [93].
2.1.2 Osteoclast
Osteoclasts are motile macrophage-like and multinucleated cells, which are formed by
the fusion of myeloid hematopoietic precursors. Osteoclast precursors are formed in
the bone marrow and circulate in the blood. The differentiation of osteoclast precursors
into mature osteoclasts requires factors like M-CSF and RANKL [94]. The coupling
of receptor-ligand complexes between osteoclast precursors and mature osteoblasts are
indispensable for osteoclastogenesis [95]. A variety of cytokines also regulate the
mechanisms of osteoclast differentiation and fusion in normal and pathological states.
Parathyroid hormone (PTH) stimulates bone marrow stromal cells and mature
osteoblasts to produce RANKL and enhances osteoclast formation [96]. Tumour
necrosis factor (TNF) activates the autocrine and paracrine mechanism in osteoclast
20
precursors, controlling their formation and the activity of bone resorption [97]. The
osteoclasts serve as bone-resorbing cells by eroding bone, enabling the tissue to be
remodelled during growth and in response to stresses. Bone resorption is an essential
event during bone growth, tooth eruption, fracture healing and the maintenance of
blood calcium level. Osteoclasts are derived from hematopoietic stem cells [98] and
are created by the differentiation of monocyte/macrophage precursor cells at or near
the bone surface [99]. Hence, osteoclasts share a common pathway with that of
macrophage and dendritic cells [100].
2.1.3 Gene network between osteoblasts and osteoclasts
Critical to maintaining the strength of bone, the process of bone remodelling must be
tightly regulated [83]. Thus, bone mass and structure are ultimately the consequence
of interactions of multiple pathways in the network that are modulated by hormonal,
cytokine and immune systems. This control is exquisitely sensitive to external stimuli
such as EMF [101, 102]. In vitro studies clearly show that a variety of molecules in
bone metabolism are affected by EMF application, including BMP-2, TGF-β and IGF-
II [103, 104] (Figure 2.1). EMFs resulted in the activation of the extracellular signal-
regulated kinase (ERK), mitogen-activated protein kinase (MAPK) and prostaglandin
synthesis, which may also lead to stimulatory effects on bone [105, 106].
TFG-ß/Nodal/Activin signals are transduced through type I and type II receptors for
each member to R-SMAD proteins, such as SMAD 2 and SMAD 3, while BMP/GDF
signals are transduced through type I and type II receptors for each member to R-
SMAD proteins. Phosphorylated R-SMADs associated with SMAD 4 are translocated
to the nucleus to activate transcription of target genes [107]. BMP/GDF family genes
within the human genome have been extensively studied [108]. However, the
transcriptional regulation of BMP/GDF family members by the canonical Wnt
21
signalling pathway remains unclear. Wnts constitute a family of proteins important in
cell differentiation, notably playing a critical role in OB cell differentiation and bone
formation [109]. Upon binding of Wnt to Frizzled receptors and the low-density
lipoprotein receptor-related protein (LRP) coreceptors-5 and -6, the activity of GSK-
3β is inhibited, leading to the stabilisation of β-catenin and its translocation to the
nucleus [110]. There, it associates with T cell factor (TCF) 4 or lymphoid enhancer
binding factor (LEF) 1 to regulate gene transcription [111]. Sclerostin (SOST)
produced by bone cells, has recently emerged as an essential modulator of anabolic
signalling pathways in bone, particularly PTH stimulation and mechanical loading
[112, 113]. These facts make Wnt a suitable target to derive a bone anabolic response
[114]. On the other hand, Wnt pathway components, including Wnts, Fzds, Lrps, and
Tcf family members, are also expressed in osteoclast lineage cells [115]. Thus, Wnt/β-
catenin signalling appears to reduce bone resorption. This area requires further
investigation to resolve the scope of Wnt influences on bone metabolism. The
landmark discoveries of the three molecules RANK, RANKL and OPG have moved
bone research into a new era [116, 117]. In bone, the expression of RANKL allows the
maturation, differentiation and activation of OCs by binding to its receptor, RANK,
present on the surface of pre-osteoclasts. On the other side of the coin, OPG exerts a
protective effect on bone acting as a decoy receptor for RANK–RANKL binding. The
balance between RANKL and OPG determines bone resorption [118, 119]. Activated
T cells express RANKL and support osteoclast formation and activation by cytokines,
including IL-1, IL-6, TNF-α and IL-17. In contrast, several other T cell-derived
interleukins and cytokines, IFN-γ, IFN-β, IL-4, IL-10, IL-13, GM-CSF, osteoclast
inhibitory lectin and secreted Frizzled-related proteins (sFRPs) potently inhibit
osteoclast formation [120, 121]. Complex interactions have been visualised by
22
conceptual block diagram using data and text mining in Figure 2.2. Established
knowledge from feedback control of dynamic systems in engineering, allows us to
define four primary functions for known regulations: induce, inhibit, relate, and
complex. These primary functions enable construction of a network with systematic
loops.
2.2 Mathematical models for simulating cell growth and movement
Efforts for modelling mechanism of EMF on bone remodelling have focused on
mechanical functions or biological functions based on signalling pathways [122, 123].
Mathematical models of bone remodelling have been established on the
RANK/RANKL/OPG pathway under the influence of PTH, mechanical force, and
EMF at the cellular level [124-126]. These models can be extended to different
waveforms of EMF and other external perturbation [127, 128].
Bone cell differentiation and proliferation are important factors during bone
remodelling, and clinical PEMF devices have been shown to affect differentiation and
proliferation of bone cells in vitro [129, 130]. It has been proposed that gap junctions
which are specialised intercellular junctions are considered as mediators of the PEMF-
related cellular responses [129, 131-133]. Nevertheless, the underlying mechanism at
the cellular level that regulates bone remodelling under PEMF remains poorly
understood because of the inconsistent or even contradictory results from experiments.
For example, cell proliferation, as assayed by cell number and H-thymidine
incorporation, has been reported to increase [134], decrease [135], and remain
unaffected [136] by PEMF exposure. Similarly, the production of alkaline phosphatase
has been reported to either increase [137] or decrease [132] following PEMF
exposure.
23
In order to remove the limitations to generalisation concerning causes and effects of
bone remodelling under PEMF, mathematical models provide a dynamic, quantitative
and systematic description of the relationships among interacting components of the
biological system [138]. Kroll [139] and Rattanakul, et al. [140] each proposed a
mathematical model accounting for the differential activity of PTH administration on
bone accumulation. Komarova, et al. [141] presented a theoretical model of autocrine
and paracrine interactions among osteoblasts and osteoclasts. Komarova [142] also
developed a mathematical model that describes the actions of PTH at a single site of
bone remodelling, where osteoblasts and osteoclasts are regulated by local autocrine
and paracrine factors. Potter, et al. [143] proposed a mathematical model for the PTH
receptor (PTH1R) kinetics, focusing on the receptor’s response to PTH dosing to
discern bone formation responses from bone resorption. Lemaire, et al. [144]
incorporated detailed biological information and a RANK-RANKL-OPG pathway into
the remodelling cycle of a model that included the catabolic effect of PTH on the bone,
but the anabolic effect of PTH was not described. Based on the Lemaire’s model,
Wang, et al. [145] developed a mathematical model that could simulate the anabolic
behaviour of bone affected by intermittent administration of PTH, as well as a
theoretical model and its parametric study of the control mechanisms of bone
remodelling under the mechanical stimulus. Pivonka, et al. [146], [147] extended the
bone-cell population model based on the Lemaire’s model to explore the model
structure of cell-cell interactions theoretically and then investigated the role of the
RANK-RANKL-OPG system in bone remodelling.
The numerical models of bone remodelling are built under external loads and
examined by bone density. Finite element method (FEM) [148-154] is employed by
such models to simulate the relationship between bone density change and mechanical
24
stimulus [155]. FEM is successful in analysing the macroscopic level of bone structure,
especially in studying microcracks [156] and reconstruction of bone models from CT
images [157, 158]. Such problems can refer to similar studies from material research
[159-165]. Although the FEM may put forward the quantitative prediction of cell
behaviours, it is not clear exactly how mechanical loading affect the activities of
osteoblasts and osteoclasts in each cell cycle. Besides, the parametric simulation of
cell signalling introduces unknowns into the model and raises a question of how
signalling pathways interact with each other instead of the answer for the coupling
mechanism between activities of osteoblasts and osteoclasts at the cellular level. It
might be feasible to describe each component mathematically on a small scale of
signalling (Figure 2.1), while the mission is impossible for a signalling network
(Figure 2.2). Further, Spatial and temporal heterogeneity of molecules or proteins
require a comprehensive mathematical description instead of connecting signalling
pathways with simple positive and negative circuit feedbacks.
The multi-scale methodology of bone remodelling simulation is created to find
solutions for unknowns from experiments. One method of organising the multi-scale
bone remodelling simulation is to integrate signalling-based bone cell population
model into a micromechanical representation of cortical bone [166]. Theoretically, the
bone cell population can affect the mechanical properties of bone and the unknowns
for the model of bone cell population could be found when the mechanical properties
of bone are calculated in the experiment. Bone cell development proceeds in three
main periods: proliferation, extracellular matrix maturation, and mineralisation [167].
Unfortunately, according to the model built by Owen, et al. [168], the bone cell
population cannot fully link with mechanical properties in these periods. Another
method of multi-scale bone remodelling simulation is based on the fact that electric
25
and electromagnetic fields gain significance in the therapy of bone fracture healing
and bone disease [169].
Mathematical modelling is a powerful tool for testing and analysing various
hypotheses in complex systems, yet it is hard to simulate the biological system
accurately. The challenges are to elucidate the interactions of a biological system
mathematically in both cellular and molecular level, and to verify whether the
prediction derived from the formulated biological system match the experimental
results.
26
Figure 2.1 BMP-SMAD and TGF-β/Activin pathway. The intracellular signalling
networks follow a specific pathway from one gene to another. Models for intracellular
signalling networks are written from the original physiological map and woven into a
web in which single elements in models can receive information from multiple inputs.
However, despite impressive accomplishments at small scale, current models fall apart
at the genome scale. The failures occurring at the large scale are attributed to the reason
that the pre-assumption of relatively homogeneous and static components hardly
reflects the heterogeneity of biological signalling network.
27
28
Figure 2.2 Map of osteo-gene network. The remodelling units of OCs and OBs and
related genes are divided into three levels: extracellular, intracellular and nuclear
levels. Known regulations are applied in the diagram to establish interactions among
genes, and unknown regulations are highlighted with a question mark. Osteo gene
network represents the importance of a gene or a protein in multiple and interacted
pathways. This network provides a possible frame for a systematic and quantitative
measure of organising the gene and protein expression of various samples in one
picture. It is possible to simulate and study how modulating specific molecules and
cellular functions affect critical physiological and pathological outcomes within an
integrated system context.
29
Chapter 3
Research Question and Hypothesis
30
3.1 Research question
Researchers have been confused about the effect of EMF on a biological system in
past decades. Chapter 1 has shown the efforts to elucidate one or several factors for
explaining the underlying mechanism of the EMF effect on a biological system.
Nevertheless, we realise that different dataset yield contradictions when articulating
them into one picture. In classical physics, data obtained by using various instruments
supplement each other and can be combined into a consistent picture where the
influence of the measuring procedure can easily be subtracted from its outcome. On
the contrary, the effect of EMF on cells illustrates that changing the experimental
arrangement is equivalent to changing the holistic phenomena. Consequently, we
could not predict the effect of EMF on the experiment results even though numerous
similar experiments have been carried out.
Such contradictions might be overcome by loosely articulating the information derived
from various experimental arrangements in complementarity. Bohr introduced
complementarity in 1927, “Evidence obtained under different experimental conditions
cannot be comprehended within a single picture but must be regarded as
complementary in the sense that only the totality of the phenomena exhausts the
possible information about the objects” [170]. The principle of complementarity was
applied (i) between incompatible variables, (ii) between causation and spatiotemporal
location, and (iii) between the continuous and discontinuous pictures. Following the
principle of complementarity, we could describe cell behaviours in a rough quantity
without giving precise laws, or we could give the precise laws in their abstract form
without the quantitative presentation. Interestingly, experimental results in Chapter 1
described the biological effects of EMF in a rough quantity, and cell signalling and
mathematical models in Chapter 2 illustrated the precise laws.
31
Different pieces of information are taken into being mutually exclusive, yet jointly
indispensable to characterise a specific object. We write research questions here about
the consistency of bone cells’ bio-data in weak EMF.
• In the same experimental setting except for magnetic intensity, whether the
relationship between osteoblasts population and magnetic intensity follows the
same pattern throughout the time.
• In the different types of EMFs but with a similar scale of magnetic intensity,
whether the relationship between osteoblasts population and magnetic intensity
obeys the same principle.
• Whether the principle derived from osteoblasts behaviour in weak EMF is
consistent in the co-culture of osteoblasts and osteoclast.
3.2 Hypothesis
The primary hypothesis assumes that EMFs can affect the movement of osteoblasts.
The mechanisms of EMFs on biological systems in Chapter 1 provide this hypothesis
with substantial evidence. However, it is an open question of how to describe the
movement of osteoblasts mathematically. We prefer the concept of “cell probability”
which means the chance that cells appear at a specified location by counting the
number of cells at a specified location in a particular time and then taking the ratio of
this number to the total number during that time.
The cell population is usually presented as a mean ± standard deviation when repeating
the same experiment of cell proliferation several times. Here we consider the cell
population as a normal distribution with the central of mean. This normal distribution
might not be established since the sample size is limited and each sample has the same
weight in occurrence. Instead, cell probability might be described. Cell probability is
an imaginary picture that indicates the tendency of cell movement.
32
We imagine the effects of EMF on cell probability as a slit, which is a derivative model
of the double-slit experiment (Figure 3.1). The culture environment consists of at least
two slits, with and without EMF. When osteoblasts are cultured in EMFs, the cell
probability has two possible patterns after passing through slits. Therefore, the
superposition of probability amplitudes could occur. We write the probability
amplitude as
Cell final at Cell initial at x s = . (3.2.1)
Following the expression in physics, the right part to the vertical line gives the initial
condition, and the one on the left indicates the final condition. The abbreviation of
(3.2.1) is
x s = . (3.2.2)
Such an amplitude might be a complex number. Each replicate in cell culture has the
same probability amplitude rather than the cell population. If a factor is amplified in
the culture environment such as hormones, this will create a new slit. We mark the two
slits as 1 and 2. The resultant amplitude is calculated by multiplying in succession the
amplitude for each of the successive events,
1 1 2 2x s x s x s= + , (3.2.3)
which equals to
1 2 = + . (3.2.4)
The cell probability P is given by
2
P = . (3.2.5)
Hence, cell probability could happen with or without interference. If there are mutually
exclusive, indistinguishable alternatives in an experiment, the interference happens,
33
2 2 2
1 2 1 2 1 22 cosP = + = + + , (3.2.6)
where is the phase difference between 1 and 2 . If an experiment is performed
with the capability of determining whether one or another alternative is taken, the
interference is lost,
2 2 2
1 2 1 2P = + = + . (3.2.7)
For instance, screening tests in biological experiments are taken without interference.
When a PEMF is specified in magnetic intensity and frequency, we calculate a much
more complicated problem. In physics, when an event can occur in several alternative
ways, the probability amplitude for the event is the sum of the probability amplitudes
for each way considered separately. We build two filters for PEMF, where the first
filter represents the magnetic intensity and the second filter represents the frequency.
If the first filter has i slits and the second filter have j slits, the complete amplitude
is written by
j i
x s x j j i i s= . (3.2.8)
When osteoclasts appear in the experiment, the disturbance happens between
osteoclasts and the slits. The amplitude of osteoblasts is described in the equation
(3.2.3). The amplitude 1 s occurs when osteoblasts go from s to slit 1 and multiply
the amplitude a when osteoblasts contact with osteoclasts at slit 1. Then the
amplitude that osteoblast go from s to x via slit 1 and contact osteoclasts is
1 1 1a x a s = . (3.2.9)
Similarly, the amplitude that osteoblasts go via slit 2 and contact osteoclasts is
2 2 2b x b s = . (3.2.10)
34
The probability of both amplitudes is
2
1 2P a b = + , (3.2.11)
where the time is involved in the amplitude and the corresponding probability.
Different results are detected if the measurement occurs at different places and times.
The function should satisfy a differential equation which could be a wave equation
analogous to the equation for electromagnetic waves. However, it must be noticed that
this function is not a real wave in space even when it satisfies the wave equation. Since
cells take a long time in movement compared with the particles with a given energy,
we may consider the cells equivalent to particles with a very long wavelength in the
interference. Therefore, we might observe the alteration of cell possibility and the
occurrence of interference when osteoblasts are cultured in EMFs.
35
Figure 3.1 Double-slit model for osteoblasts. Thinking that osteoblasts start to move
at S and stop at X, osteoblasts pass through the silts created by culture environment.
Cell probability has a pattern along the culture plate. Each replicate in the experiment
has the same cell probability rather than the cell population. The number of slits
depends on the variety of external stimulus such as EMFs. When osteoblasts pass slits,
the superposition of probability amplitudes could occur with or without interference.
36
Chapter 4
Materials and Biological Methods
37
4.1 Materials
4.1.1 Chemical reagents
Chemical reagents involved in experiments conducted in this thesis were of analytical
grade unless indicated otherwise and were purchased from the following
manufacturers:
Description Manufacturer
1α,25-dihydroxyvitamin D3 Sigma-Aldrich, USA
4-(2-hydroxyethyl)-1-
piperazineethanesulfonic acid (HEPES)
ThermoFisher, Australia
α-MEM (α-modification of Eagle’s medium) Life Technologies, Australia
β-mercaptoethanol Sigma-Aldrich, USA
Alizarin red S Biochemicals Inc, USA
Bovine serum albumin Sigma-Aldrich, USA
Bromophenol blue Bio-Rad, USA
Chloroform ThermoFisher, Australia
Dexamethasone Sigma-Aldrich, USA
Dimethyl Sulphoxide (DMSO) ThermoFisher, Australia
D-MEM (Dulbecco’s-modification of Eagle’s
medium)
Life Technologies, Australia
Ethanol (100%) EMD Millipore Corporation, USA
FBS (Foetal Bovine Serum) Life Technologies, Australia
Hanks’ balanced salt solution Life Technologies, Australia
HEPES (4-(2-hydroxyethyl)-1-
piperazineethanesulfonic acid)
Sigma-Aldrich, USA
38
L-ascorbic acid phosphate Sigma-Aldrich, USA
L-Glutamine Sigma-Aldrich, USA
Liquid nitrogen (N2) Air Liquid, Australia
Penicillin Life Technologies, Australia
Skim Milk Powder Standard supermarket brand
Sodium chloride (NaCl) Bio-Rad, USA
Sodium dodecyl sulphate (SDS) Bio-Rad, USA
Sodium hydroxide (NaOH) Sigma-Aldrich, USA
Sodium phosphate dibasic (Na2HPO4) Sigma-Aldrich, USA
Sodium phosphate monobasic (NaH2PO4) Sigma-Aldrich, USA
Tween-20 Bio-Rad, USA
4.1.2 Commercial kits and molecular products
Description Manufacturer
10 x Reaction Buffer, 1 mL of 200 mM Tris-
HCl, pH 8.3, 20 mM MgCl2
Sigma-Aldrich, USA
5 x First Strand Buffer Invitrogen, USA
Amplification Grade DNAse I ThermoFisher, Australia
BD Pharm LyseTM red cell lysis buffer (10 x) BD Bioscience, USA
DAPI (4’,6-diamidino-2-phenylindole),
dilactate
Biotium, USA
Fastlane Cell cDNA Kit QIAGEN Gmbh, Australia
RNEasy Mini kit QIAGEN Gmbh, Australia
TRIzol® Invitrogen, USA
39
4.1.3 Other products and consumables
Description Manufacturer
0.2 mL PCR Tubes ThermoFisher, Australia
50 mL Centrifuge Tubes ThermoFisher, Australia
BD 1 mL Syringe BD Bioscience, USA
BD PrecisionGlideTM Needles 25 G BD Bioscience, USA
Coverslips-round 5 mM, 22×22 mm2 and
22×40 mm2
Knittle Glaser, Germany
Filter Paper Whatman International, UK
Glass Slides Knittle Glaser, Germany
Parafilm Laboratory Film Merck Millipore, Australia
Transfer pipettes Sigma-Aldrich, USA
4.1.4 Cytokines
Description Manufacturer
Recombinant human macrophage-colony
stimulating Factor (M-CSF)
EMD Millipore Corporation, USA
Recombinant human soluble RANK ligand
(sRANKL)
EMD Millipore Corporation, USA
4.1.5 Solutions
Solution and buffers were prepared using Mili-Q distilled water (ddH2O). HCl and
NaOH were used to adjust the pH unless stated otherwise.
40
Solution Composition and preparation
Flow cytometry wash buffer PBS containing 1% FBS and 0.1% (w/v) sodium
azide, stores at 40C.
Phosphate buffered saline
(PBS)
10 x PBS stock solution in ddH2O: 70 mM
Na2HPO4, 30 mM NaH2PO4 and 1.3 M NaCl. 1 x
PBS:1 in 10 dilutions of 10 x PBS stock solution
and calibrated to pH 7.4.
SDS-PAGE running buffer 10x stock solution in ddH2O: 25 Trizma base, 1.92
M glycine and 1% (v/v) SDS. 1 x working solution:
1in 10 dilutions of the 10x stock solution in ddH2O.
SDS-PAGE sample loading
buffer
4x stock solution dissolved in ddH2O:240 mM Tris-
HCl (pH 6.8), 8% (w/v) SDS, 40% (v/v) glycerol,
0.04% (w/v) bromophenol blue and 5% (v/v) β-
mercaptoethanol. Stored at 4 0C.
SDS-PAGE separating gel
buffer
1.5 M Tris-HCl (pH 8.8) and 0.4% (w/v) SDS
prepared in ddH2O.
SDS-PAGE stacking gel
buffer
1 M Tris-HCl (pH 6.8) and 0.4% (w/v) SDS
dissolved in ddH2O.
Sodium deoxycholate
solution
10% (w/v) solution dissolved in ddH2O.
Tartrate-resistant acid
phosphatase (TRAP) stain
5 mg Naphthol AS-MX dissolved in 250 µL 2-
ethoxyethanol with the addition of 40 mg Fast Red-
Violet LB salt dissolved in TRAP stain solution A.
Aliquots are stored at -20 0C, thawed at 37 0C and
filtered before use.
41
TRAP stain solution A 100 mM sodium acetate trihydrate, 50 mM sodium
tartrate dihydrate and 0.22% (v/v) glacial acetic
acid. Adjusted to pH of 5.0.
Western blot transfer buffer 25 mM Trizma base, 192 mM glycine and 10%
(v/v) methanol in ddH2O.
4.1.6 Equipment
Description Manufacturer
4300 SE/N Schottky Field Emission Electron
Microscopy
Hitachi, Japan
96-well Thermal Cycler Applied Biosystems, USA
Automatic CO2 Incubator Thermo Scientific, USA
Digital Teslameter with the Probe 3B Scientific, Germany
Eclipse TE2000s Fluorescent Microscope Nikon Instruments Inc., USA
Function Generator 3B Scientific, Germany
ImageQuant LAS 4000 GE Healthcare, USA
MiniSpin® microcentrifuge Eppendorf AG, Germany
MilliQ Water System Millipore Corporation, USA
Nanodrop® ND-1000 UV-Vis
Spectrophotometer V3.3
Thermo Scientific, USA
Nikon CoolPix S4 Digitial Camera Nikon Corp., Japan
Olympus CK30 Microscope Olympus Optical Co Ltd, Japan
Real-Time qPCR System Life Technologies, USA
42
4.1.7 Software
Description Manufacturer
ANSYS Ansys, Inc. USA
FACSDiva BD Biosciences, USA
Microsoft Research Image Composite Editor Microsoft, USA
MATLAB MathWorks, USA
FlowJo FlowJo LLC, USA
Prism GraphPad Software, USA
Origin OriginLab, USA
4.1.8 RT2 profiler PCR array
The PCR array catalogue is PAHS-026z.
Symbol Description
ACVR1 Activin A receptor, type I
AHSG Alpha-2-HS-glycoprotein
ALPL Alkaline phosphatase, liver/bone/kidney
ANXA5 Annexin A5
BGLAP Bone gamma-carboxyglutamate (gla) protein
BGN Biglycan
BMP1 Bone morphogenetic protein 1
BMP2 Bone morphogenetic protein 2
BMP3 Bone morphogenetic protein 3
BMP4 Bone morphogenetic protein 4
BMP5 Bone morphogenetic protein 5
BMP6 Bone morphogenetic protein 6
BMP7 Bone morphogenetic protein 7
43
BMPR1A Bone morphogenetic protein receptor, type IA
BMPR1B Bone morphogenetic protein receptor, type IB
BMPR2 Bone morphogenetic protein receptor, type II
CALCR Calcitonin receptor
CD36 CD36 molecule (thrombospondin receptor)
CDH11 Cadherin 11, type 2, OB-cadherin (osteoblast)
CHRD Chordin
COL10A1 Collagen, type X, alpha 1
COL14A1 Collagen, type XIV, alpha 1
COL15A1 Collagen, type XV, alpha 1
COL1A1 Collagen, type I, alpha 1
COL1A2 Collagen, type I, alpha 2
COL2A1 Collagen, type II, alpha 1
COL3A1 Collagen, type III, alpha 1
COL5A1 Collagen, type V, alpha 1
COMP Cartilage oligomeric matrix protein
CSF1 Colony stimulating factor 1 (macrophage)
CSF2 Colony stimulating factor 2 (granulocyte-macrophage)
CSF3 Colony stimulating factor 3 (granulocyte)
CTSK Cathepsin K
DLX5 Distal-less homeobox 5
EGF Epidermal growth factor
EGFR Epidermal growth factor receptor
FGF1 Fibroblast growth factor 1 (acidic)
44
FGF2 Fibroblast growth factor 2 (basic)
FGFR1 Fibroblast growth factor receptor 1
FGFR2 Fibroblast growth factor receptor 2
FLT1
Fms-related tyrosine kinase 1 (vascular endothelial growth
factor/vascular permeability factor receptor)
FN1 Fibronectin 1
GDF10 Growth differentiation factor 10
GLI1 GLI family zinc finger 1
ICAM1 Intercellular adhesion molecule 1
IGF1 Insulin-like growth factor 1 (somatomedin C)
IGF1R Insulin-like growth factor 1 receptor
IGF2 Insulin-like growth factor 2 (somatomedin A)
IHH Indian hedgehog
ITGA1 Integrin, alpha 1
ITGA2 Integrin, alpha 2 (CD49B, alpha 2 subunits of VLA-2 receptor)
ITGA3
Integrin, alpha 3 (antigen CD49C, alpha 3 subunits of VLA-3
receptor)
ITGAM Integrin, alpha M (complement component 3 receptor 3 subunit)
ITGB1
Integrin, beta 1 (fibronectin receptor, beta polypeptide, antigen
CD29 includes MDF2, MSK12)
MMP10 Matrix metallopeptidase 10 (stromelysin 2)
MMP2
Matrix metallopeptidase 2 (gelatinase A, 72kDa gelatinase, 72kDa
type IV collagenase)
MMP8 Matrix metallopeptidase 8 (neutrophil collagenase)
45
MMP9
Matrix metallopeptidase 9 (gelatinase B, 92kDa gelatinase, 92kDa
type IV collagenase)
NFKB1 Nuclear factor of kappa light polypeptide gene enhancer in B-cells 1
NOG Noggin
PDGFA Platelet-derived growth factor alpha polypeptide
PHEX Phosphate regulating endopeptidase homolog, X-linked
RUNX2 Runt-related transcription factor 2
SERPINH1
Serpin peptidase inhibitor, clade H (heat shock protein 47), member
1, (collagen binding protein 1)
SMAD1 SMAD family member 1
SMAD2 SMAD family member 2
SMAD3 SMAD family member 3
SMAD4 SMAD family member 4
SMAD5 SMAD family member 5
SOX9 SRY (sex determining region Y)-box 9
SP7 Sp7 transcription factor
SPP1 Secreted phosphoprotein 1
TGFB1 Transforming growth factor, beta 1
TGFB2 Transforming growth factor, beta 2
TGFB3 Transforming growth factor, beta 3
TGFBR1 Transforming growth factor, beta receptor 1
TGFBR2 Transforming growth factor, beta receptor II (70/80kDa)
TNF Tumour necrosis factor
TNFSF11 Tumour necrosis factor (ligand) superfamily, member 11
46
TWIST1 Twist homolog 1 (Drosophila)
VCAM1 Vascular cell adhesion molecule 1
VDR Vitamin D (1,25- dihydroxy vitamin D3) receptor
VEGFA Vascular endothelial growth factor A
VEGFB Vascular endothelial growth factor B
ACTB Actin, beta
B2M Beta-2-microglobulin
GAPDH Glyceraldehyde-3-phosphate dehydrogenase
HPRT1 Hypoxanthine phosphoribosyltransferase 1
RPLP0 Ribosomal protein, large, P0
HGDC Human Genomic DNA Contamination
RTC Reverse Transcription Control
PPC Positive PCR Control
4.2 PEMF apparatus design
Self-made EMF device was used in controlling the frequency and signal type (
Figure 4.1). The EMF Apparatus consists of two tubes (Figure 4.2), one protective
shield, one platform (Figure 4.3) and two supports (Figure 4.4), generating PEMF
signal and magnetic field strength by Helmholtz coils or solenoid of around 300 turns
copper wires each. This apparatus also includes a digital oscilloscope and teslameter
for measurement.
4.3 Culture and in vitro studies of Sao-2 cell line
Cells were cultured in plates or flasks inside an incubator with 5% CO2, 95% air at 37
0C. Cultured cells were handled using aseptic techniques in UV sterilised biological
safety cabinets (Class II).
47
4.3.1 Saos-2 cell line
Saos-2 (sarcoma osteogenic) was purchased from Australian agent (Sigma Aldrich,
catalogue number: 89050205), which was a non-transformed cell line derived from
human primary osteogenic sarcoma of an 11-year old female Caucasian. The Saos-2
cell line was from the ECACC collection, used as a permanent line of human
osteoblast-like cells and a source of bone-related molecules.
4.3.2 Initiation of the culture process
The medium of 5 mL was pre-equilibrated within a 25 cm2 culture flask for 2 hours in
an incubator. Saos-2 cells were removed from frozen storage and thawed in a 37 0C
water bath with agitation. The cryovial was removed when the ice melted and then
rinsed with 70% ethanol. The cells were resuspended and transferred to a 25 cm2 flask
with 5 mL equilibrated medium, then incubated 30 minutes for the settlement. The
cells were rinsed with 5ml of warmed fresh medium to remove the cryoprotectant. The
cells were cultured in fresh media and monitored daily for use. Saos-2 cells could grow
in a stationary flask using medium supplemented with 10% FBS, 2 mM L-glutamine
and 0.01% kanamycin. The cells were kept in a 2 mm medium layer in a flask with the
caps opening a quarter turn to balance the aeration and nutrition.
4.3.3 Subculture Saos-2
For a 25 cm2 culture flask, cells were subcultured when reaching 70-80% confluent.
The cells were washed with 10 mL warm and sterile PBS after removing the culture
medium. Trypsin solution of 2 mL was added to cells. The reaction lasted for 6 minutes
in the incubator. Growth medium of 5 mL was added to stop the reaction, and detached
cells were transferred to a sterile 15 mL centrifuge tube. The cell suspension was
centrifuged for 6 minutes at 300 x g. The cell pellet was resuspended in complete α-
48
MEM of 2 mL supplemented with 10 ng/mL MCSF and cultured in culture flasks or
plates.
4.3.4 Count cell densities of Saos-2 cells
The cells of 100 µL were transferred to a well in a 96-well plate with 10 µL sterile
0.4% trypan blue. The mixture was ensured homogenous and transferred 10 µL to the
haemocytometer. The dead cell was counted in the colour of blue and divided by the
total number of cells for the proportion of dead or lysed cells. The viability was one
minus the proportion of dead cells.
4.3.5 Cryopreservation and retrieval of cultured cells
The cells were frozen when exceeding 80% in confluence. The cells were gently
dislodged and centrifuged for 6 minutes at 300 x g. The cell pellet was resuspended to
5-6 x106 cells/mL and added 10% DMSO. Plastic cryovials were loaded with a cell
suspension of 1 mL. The vials were placed on ice tub for 15-30 minutes to allow
equilibration of the cells with the DMSO-containing medium. The vials were
transferred to the tray of the cell freezer and adjusted the height of the ray according
to the manufacturer’s instructions. The freezing unit was inserted in the neck of the
liquid nitrogen refrigerator for approximately 3 hours. Transfer frozen vials were
transferred to a storage canister in liquid nitrogen.
In order to survive freezing and thawing, the cells were treated with a cryoprotective
agent. DMSO or glycerol was added to resuspension medium 10% acted to
permeabilize the plasma membrane. Cryoprotectants depressed the freezing point so
that ice crystals began to form at about -5 ºC. If at -5 ºC to -15 ºC, the cell was
sufficiently dehydrated accompanied by osmotic shrinkage and concentration of ice
crystals formed in the surrounding medium but not in the cell interior. In practice, the
warming rate was more challenging to control than the cooling rate. General practice
49
was to immerse frozen vials in a warm water bath (37-40 0C) so that the cells thawed
in about 1.5 minutes.
4.4 Isolation, culture and in vitro studies of human osteoblasts
4.4.1 Isolation and culture of osteoblasts from human trabecular bone
Pieces of excised human bone specimens were placed in sterile saline or PBS at time
of collection, and the samples were transferred in a foam box without ice. Completed
α-MEM (α-MEM supplemented with 10% FBS, 2 mM L-glutamine, 100 U/mL
Penicillin and 100 µg/mL Streptomycin) and filtered PBS were warmed in 37 0C water
bath. The centrifuge was set at 300 x g, 4 0C and 6 minutes. Aseptic technique was
used to cut the bone specimen into small pieces about 1-2 mm using a scalpel blade or
scissors. The bone chips were washed five times with sterile PBS and transferred into
a sterile Falcon tube. Contaminating red blood cells were removed using a vortex
mixer. 8-10 pieces of bone were placed into a T75 flask along with 20 mL completed
α-MEM. The flask was placed in an incubator with 5% CO2 at 37 0C. After 1 week,
the medium was removed, and fresh culture medium was added. The bone pieces were
cultured for another week without disturbing. After that, the culture medium was
changed every 2 days. Between 3-4 weeks the cells could be confluent about 1-2 x 106
cells/flask.
4.4.2 Harvest of monolayer cultures
Collagenase (1 mg/mL) of 1 mL was added to the flask for 2 hours at 37 0C. Cell
culture medium was removed using a vacuum, and the cells were immediately washed
with 10 mL warm and sterile PBS. 3-5 mL of TryExpress solution was added to cover
the cell surface for 6 minutes reaction in the incubator. α-MEM of 5 mL was added to
stop the reaction, and the cells were transferred to a 50 mL centrifuge tube. Cell
50
solution was centrifuged at 300 x g, 4 °C for 6 minutes, and the supernatant was
disposed. α-MEM of 3 mL was added to the flask for the cell suspension. A pipette
was used to break cell clusters by pressing the tip of the pipette against the bottom of
the flask. The cell suspension was passed through a needle of 10 mL syringe 5-6 times
with the needle bevel pressing against the tube wall to further separate the cells.
4.4.3 In Vitro osteoblasts cell mineral formation assay
Alizarin red-S (AR-S) is a dye-binding selectively to calcium salts to quantify calcium
mineral content by measuring the amount of AR-S bound to mineralised nodules in
the cultures. Cells were briefly washed with PBS 1 mL/well. Ice-cold ethanol (70%)
was added to the cultures and reacted at 4 ºC for a minimum of 1 hour. The cells were
washed with ddH2O and added 1 mL/35-mm well 40 mM AR-S, pH 4.2. The plate
was placed on the shaker at room temperature for 10 minutes. The cells were washed
with ddH2O for 5 times and with PBS once. The plate was then shaken at room
temperature for 15 minutes. AR-S solution of 40 mM was diluted in 1: 400 with CPC
buffer to make 100 µM AR-S solution, which was diluted in a series dilution of every
3-fold with CPC buffer to make AR-S standards. Eppendorf tubes were labelled, and
CPC buffer was added to each tube. AR-S solution (100 µM) of 200 µL was added to
the tube. CPC buffer was added to each well to extract AR-S for 15 minutes at room
temperature. CPC buffer of 90 µL was added to each well in a 96-well plate. 10 µL
the AR-S was added to each well (1:10 dilution) and mixed. The dilution factor can be
changed depending on estimating of mineral content such as colour. AR-S standard
solution was added to the wells in replicates. AR-S concentration was determined by
absorbance measurement at 562 nm. The AR-S concentration was calculated using
AR-S standard curve.
51
4.5 Induction and culture of human osteoclasts
4.5.1 Preparation of solution
Osteoclast cells can be differentiated from human peripheral blood mononuclear cells
(PBMCs) using differentiating culture medium. HANK’s solution was prewarmed in
a water bath at 37 0C and filtered using a filter cup (0.2 µM) before use. Completed
osteoclast culture medium was made by α-MEM supplemented with 1% L-Glutamine,
10% FCS, 1% PSN, 10-8 M Dexamethasone, 10-8 M Vitamin D3 and 25 ng/mL M-
CSF. DMSO was used to dissolve Vitamin D3. Human recombinant RNAKL was
prepared at 50 ng/mL.
4.5.2 PBMC isolation
The Buffy coat 15 mL in a falcon tube and 20 mL pre-warmed HANKs was added to
dilute the buffy coat. Ficoll-Paque of 7.5 mL was added to an empty 50 mL Falcon
tube, and diluted buffy coat of 10 mL was loaded over Ficoll solution using a syringe.
The ratio of Ficoll and diluted buffy coat was 3:4. After loading, all of the tubes were
placed on a centrifuge, spun at 500 x g, 25 C for 30 minutes. The mononuclear cells
(PBMCs) were collected at the interface using a sterile pipette directly inserted into
the PBMCs layer. This layer was like a cake in between orange colour and white
colour. The PBMCs were transferred to a fresh 50 mL falcon tube. Cells were washed
with 20 mL HANKs and centrifuged at 500 x g, 25 C for 5 minutes. The supernatant
was discarded and completed osteoclast cell culture medium was added to make 5-10
mL cell suspension, depending on how many cells isolated.
4.5.3 PBMC cell culture
The cell viability was checked using 20 µL, 0.2% trypan blue working solution plus
20 µL cell suspension. The cells suspension was diluted if the concentration was high.
52
The cells were counted under bright field microscopy. For example, seeding to 24-well
plate and the cell suspension was 0.5 mL/well, the volume of cell seeding suspension
was 12 mL. The cell seeding suspension was made by adding the above volume into
14 mL completed osteoclast cell culture medium. Cell seeding suspension of 0.5 mL
was added to each well of a 24-well plate. The cells were incubated at 37 0C, 5% CO2
and 95% moisture incubator. In day 4, the culture medium was changed. In day 7, the
culture medium was changed with adding human recombinant RANKL into the cell
culture to induce osteoclasts. The final concentration of RANKL in the culture media
was 50 ng/mL. In day 10, the osteoclasts could be observed under bright field
microscopy.
4.5.4 TRAP staining
When multinucleated osteoclasts were formed, the cells were gently washed once with
1 x PBS and fixed with 4% paraformaldehyde for 15 minutes at room temperature,
followed by three washes with 1 x PBS. The cells were subsequently stained with
filtered Tartrate-resistant acid phosphatase (TRAP) solutions. The TRAP solutions
were warmed up to 37 0C and filtered before staining. The slides were dewaxed and
hydrated. Excess liquid was removed from the slide without letting the section dry out.
Filtered pre-warmed TRAP solution was applied, and the slides were incubated at 37
0C. The slides were checked every 30 minutes to monitor the development of the stain.
After approximately 2 hours, with the observation of magenta-stained osteoclasts, the
slides were rinsed in 1 x PBS three times and stored at 4 0C until ready for imaging.
The osteoclasts were visualised using the Olympus microscope. Serial images were
taken at desired magnification covering the entire well.
53
4.5.5 Flow cytometer and analysis
The FACSCalibaur, FACSCanto II and LSR Fortessa TM SORP were used for flow
cytometric studies. FACSDiva software was used for application setup, data
acquisition and data compensation. The live and single mononuclear cells were gated
using Forward Scatter (FSC) and Side Scatter (SSC) plots. Events were recorded.
Acquired and compensated data was imported into FlowJo for analysis.
4.6 Western blotting
Western blotting allows the specific detection of proteins on membranes after
separation by SDS-PAGE. It combines the resolution of gel electrophoresis with the
specificity of immunochemical detection. This technique is beneficial for the
identification and semi-quantification of specific proteins in complex mixtures of
proteins.
4.6.1 Protein extract and assay
At the different time point of the treatment, cells were extracted by using RIPA buffer.
The cells were washed with PBS to remove residual media. RIPA buffer (1 x) of 400
μL was added for 10 cm2 (6-well-plate: 9cm2 needs 360 μL; 25cm2 needs 1 mL). The
cells were incubated on ice for 5 minutes. The cells were sonicated briefly and
centrifuged for 10 minutes at 14000 x g, 4 0C. The supernatant was transferred to a
new tube for use. Protein concentration was measured by using the Bio-Rad Protein
Assay. Dye reagent was prepared by diluting 1 Dye Reagent Concentrate with 4 parts
DDH2O. This diluted reagent may be used up to 2 weeks when kept at room
temperature. Five dilutions of a protein standard were prepared. The linear range of
this microtiter plate assay is 0.05 mg/mL to approximately 0.5 mg/mL, for instance,
0.05 mg/mL, 0.1 mg/mL, 0.2 mg/mL, 0.4 mg/mL and 0.5 mg/mL. Protein solutions
54
were assayed in duplicate or triplicate. The standard and sample solution of 10 μL each
were put into separate microliter plate wells. Diluted dye reagent of 200 μL was added
to each well. The sample and reagent were mixed thoroughly using a microplate mixer.
The mixture was incubated at room temperature for at least 5 minutes. Since
absorbance could increase over time, samples should incubate at room temperature for
no more than 1 hour. Measure absorbance was at 595 nm.
4.6.2 Blotting and detection
Proteins were loaded to each well of SDS-PAGE gels. The sample was heated for
denaturing electrophoresis for optimal results at 95 0C for 5 minutes. NuSep Tris-
Glycine SDS Running Buffer was used for the electrophoresis. Appropriate protein
molecular weight marker (10 µL Novex Sharp Pre-Stained Protein Standard) was
loaded. Electrophoresis was operated at room temperature based on gel type. PVDF
protein side was marked with a pencil and NuSep Transfer Buffer (1 x) of 1000 mL
was prepared. A PVDF membrane was soaked in methanol for 1 minute and placed in
distilled water. A PVDF membrane, four blotting pads, and two filter papers were
placed in a shallow tray filled with NuSep Transfer Buffer (1X) for 10 minutes. Gel
apparatus was disassembled. Stacking gel was cut off with a clean razor blade and
soaked in NuSep Transfer Buffer (1 x) for 10 minutes. Transfer apparatus gel cassettes
were open with the black panel lying flat on the bottom of the tray filled with NuSep
Transfer Buffer (1 x). The transfer sandwich was prepared on the black panel in the
tray filled with NuSep Transfer Buffer (1 x). The sandwich was covered with the
transparent panel, fasten with the latch, and inserted into the electrode module with the
black panel facing the cathode electrode panel. The gel from sandwich was removed
and rinsed with NuSep Transfer Buffer (1 x). PVDF membranes were blocked with
Blocking Buffer on a shaker for 2 hours at 37 0C or overnight at 4 0C. The membrane
55
was washed three times with Basic Buffer on the shaker for 10 minutes each time.
Primary antibody was diluted with Primary Antibody Dilution Buffer. The membrane
was incubated with the diluted primary antibody on a shaker for 1 hour at room
temperature. The membrane was then washed three times with Washing Buffer on the
shaker for 10 minutes each time. The secondary antibody was diluted with Primary
Antibody Buffer containing 5% dry milk. The membrane was incubated with the
diluted secondary antibody on a shaker for overnight at 4 0C. The membrane was then
washed three times with Washing Buffer on the shaker for 10 minutes each time.
The detection reagents were taken from the storage at 4 0C and equilibrated to room
temperature before opening. Detection solutions were mixed to the final volume of 0.1
mL/cm2. Protein side of the washed membrane was placed upon a transparency film.
The mixed detection reagent was added on the membrane for a reaction of 3 minutes
at room temperature. The reagents covered the entire surface of the membrane, held
by surface tension on to the surface of the membrane. Excess detection agent was
drained off by holding the membrane gently in forceps and touching the edge against
a tissue. The membrane ‘protein-side’ was placed facing the camera. Blots were
exposed to GE Healthcare ImageQuant LAS 4000 at least three different exposures.
4.7 RNA extraction and quantification
Total RNA was extracted from the cultured cells using TRIzol reagent. All media were
aspirated off, and the flask was washed twice with PBS. TRIzol was added directly to
cells (2 mL/25cm2, 4 mL/75cm2), and the homogenised cells were incubated in TRIzol
for 10 minutes at room temperature. The cell TRIzol extract was aspirated 10 times
through a 23 G needle using 5 mL syringe. Chloroform of 0.2 mL was added into 1
mL cell TRIzol extract. The mixture was vigorously shaken by hand for 15 seconds
and kept at room temperature for 3 minutes. The mixture was centrifuged at 12000 x
56
g, 4 °C for 15 minutes. Ethanol with DEPC H2O (70%) was prepared. The upper
aqueous phase after the centrifuge was transferred, containing the RNA, into a
microcentrifuge tube. An equal volume of 70% ethanol was added to mix thoroughly.
Visible precipitates could form after the addition of ethanol, resuspend precipitates by
vigorously shaking. Aliquots of the sample of 700 µL were transferred to an RNAeasy
mini spin column placed in a 2 mL collection tube. The sample was centrifuged at
≥8000 x g at room temperature for 15 seconds. The flow was discarded through after
each centrifugation. Buffer RW1 (commercial kit) of 350 µL was added to the column
with centrifuging at >8,000 x g for 15 seconds. DNaseI stock solution of 10 µL was
added to Buffer RDD of 70 µL. The DNase I solution of 80 µL was added directly
onto the RNeasy silica gel membrane and incubated at room temperature for 15
minutes. Buffer RW1 of 350 µL was pipetted into the column and centrifuged at
>8,000 x g for 15 seconds. The column was placed into a new collection tube, and
Buffer RPE of 500 µL was added to the column followed by centrifuging at >8,000 x
g for 15 seconds. Another 500 µL Buffer RPE was added to the column and centrifuged
at >8,000 x g for 2 minutes to dry the membrane. A new collection tube was placed in
a microfuge tube, and the old collection tube was discarded. To elute, the column was
placed into a new 1.5 mL RNase-free collection tube, and 50 µL RNase-free water was
added to the spin column membrane. The tube stood for 1 minute at room temperature
and was centrifuged at >8,000 x g for 1 minute. Another 50 µL RNase-free water was
added to the spin column membrane, and the tube was centrifuged as above if the
expected RNA yield is >30 µg. The RNA quality was measured by Nanodrop.
57
Figure 4.1 Self-made EMF device in controlling the frequency and signal type. The
EMF Apparatus consists of two tubes, one protective shield, one platform and two
supports, generating PEMF signal and magnetic field strength by Helmholtz coils or
solenoid of around 300 turns copper wires. This apparatus also includes a digital
oscilloscope and tesla-meter for measurement.
58
(A)
(B)
Figure 4.2 Coil tube. There are two tubes for this apparatus. Coils spool the cylinder
surface in a single layer. Longitude dimensions of the tube: 1 cm for each flange depth;
18cm for main tube depth. Thus, the total length of one coil tube is 20 cm. Transverse
dimensions of coil tube: D1 (internal) = 22.5 cm, D2 (external) = 25 cm; Flange
diameter: D3 = 28 cm.
59
(A)
(B)
Figure 4.3 Picture of the main body. The main body contains two parts: a protective
shield and a platform. Two tubes will be connected outside the shield. Protective shield
length is 40 cm as well as Platform length. Dimensions of transverse protective shield
are: D22 = 20 cm, D23 = 22.5 cm; H25 = 3.5 cm; V21 = 2.5 cm, V20 = V24 = 3 cm.
Dimensions of transverse platform are: H3 = 18 cm; V1 = 2 cm.
60
(A)
(B)
Figure 4.4 Support for each tube. Dimensions following X axis are: tube fix = 18 cm;
flange fix = 1 cm; edge fix = 1 cm. Thus, the total depth is 22 cm. Transverse
dimensions for support are: H5 = 29 cm; V2 = 20 cm; D3 = 26.5 cm. Diameter for the
groove is D3 = 28 cm.
61
Chapter 5
The Response of Cell Probability to
External Stimulus
62
5.1 Introduction
The hypothesis in Chapter 3 describes the qualitative response of bone cells to EMFs.
The EMFs as an external stimulus alter the original environment of cell culture. In this
chapter, we attempt to establish several frameworks for quantitatively interpreting the
evolution of cell probability under a general situation of external stimulus. The
frameworks involve two major components, (i) an external stimulus and (ii) the
cellular response to the stimulus. The response is in two steps, (i) the detection of the
stimulus and (ii) the transduction of the stimulus into the controls of cellular interaction.
Ideally, each cell is distinguished from their direction of motion, reorientation
frequency and step length between reorientations in response to the stimulus. However,
the coupling between cell density and the local stimulus intensity might prevent the
process of distinguishing. For example, when cells aggregate or diffuse in response to
a stimulus at one time-step, the distributions of local cell density and stimuli are likely
to be altered at the following time-step.
Therefore, the spatiotemporal distribution of cell density is studied in preparation for
derivations of cell probability. The assumptions are made to correlate the local cellular
mobility with cell density. The cellular mobility is partially dependent on (i) available
cellular space, (ii) cellular adhesion, involving the attachments between neighbouring
cells via extracellular receptors, mediated cellular recognition and positions, and (iii)
the external stimulus which interact with cells by mediating a series of processes such
as cell migration and cellular signalling.
5.2 Preliminaries
5.2.1 Distribution of cell density
A cellular system is necessarily a balance sheet, based on components in the system
and the processes by which the components are determined. The components include
63
contents of proteins, nucleic acid, carbohydrate and lipid. The processes include such
as individual metabolites, related signalling pathways and cell differentiation. These
descriptions are considered equivalent to the exchanges of matter and energy between
the cells and their environment. If such exchanges are quantitatively represented by
chemical equations for metabolism and biosynthesis, they construct a conceptual
input-output matrix [171].
Following the concept of the input-output matrix in biology, we consider a cellular
system S with n components 1 2, nC C C , and each component has an
independent value id , then the value vector is
1 2{ , , }nd d d=d . (5.2.1)
When every component is presented in the same system S linked with a relationship
1 1 2 2i i i in n iC a C a C a C d= + + + + , (5.2.2)
where the coefficient ija , 1,2...j n= forms a n n matrix,
11 1
1
n
n nn
a a
a a
=
A . (5.2.3)
If the components have values under a specific condition
1 2{ , , }nc c c=c , (5.2.4)
these values satisfy the relation
= +c A c d . (5.2.5)
The equation (5.2.5) has a solution
1( )−= −c I A d , (5.2.6)
64
with the identity matrix I , in which 0, ; 1, ij ijI i j I i j= = = . Consider a set of cell
densities extracted from the same system S ,
1 2{ , , }mp p p=p , (5.2.7)
with m data series, m N , for example, time series 1 2, , mt t t . Suppose each cell
density is formed by n biochemical components 1 2, nC C C , then the same
coefficient matrix A exists in a relation which is analogous to (5.2.6),
1( )−=C I - A D , (5.2.8)
with
11 1 11 1
1 1
,
m m
n nm n nm
c c d d
c c d d
= =
C D . (5.2.9)
Hence C is a matrix for cell densities p , which could represent the history of cell
proliferation. Equations from (5.2.7) to (5.2.9) show that a solution of m cell
proliferation needs 1( )−I - A of dimension n n and D of dimension n m . The
equations (5.2.8) and (5.2.9) also indicate a linear relationship between matrix C and
matrix D , which is contrary to the nonlinearity of cell proliferation. To incorporate a
nonlinearity in the equations (5.2.8) and (5.2.9), at least one variable nmd in the
matrix D is a polynomial of degree which is not one. The similar methods of
modelling nonlinear biological systems have been described in the literature [172,
173].
However, this method has inherent defects. Taking a series of biosynthetic reactions
as an example, it is hard to write the degrees of components unless every component
is well-defined in related to enzymes. It is relatively more difficult to tell the difference
65
between normal and abnormal conditions of macromolecules. Even if a matrix D is
well-established after performing m independent experiments in the same system, the
degrees of components might be different when incorporating the 1m+ th data.
An invariant relationship 1( )−I - A is proposed to avoid the method of defects for the
same system S , which requires a unified approach to constructing the matrix D . This
approach should be able to build a sufficiently large and relevant matrix involving all
components during cell proliferation. If such a matrix was directly composed of
components in biosynthetic reactions, it could be lack of relevance since these
components have their chemical structures and biological properties. The cell
proliferation is highly dependent on cell migration that involves molecular
components such as adhesion receptors, cytoskeletal linking proteins, and extracellular
matrix ligands. At a specific time-step, cell migration is relevant to the distribution of
cell density at the position ix . Therefore, we expand (5.2.7) with n positions
1 1 1 2 1
2 1 2 2 2
1 2
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
n
n
m m m n
p x p x p x
p x p x p x
p x p x p x
=
p , (5.2.10)
which is a matrix in the dimension of m n . The equation (5.2.8) multiples with
(5.2.10) forming a new n n matrix in (5.2.11) and this new matrix *
p represents the
distribution of the cell population after the interactions of components in the cellular
system
66
* *
11 11 1 1 1 1 1
*
* *
1 1 1
11 1 1 1 1
1
1 1
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( )
( )
( ) ( )
mn n
n n n n nm m m n
m n
n nm m m n
c cp x p x p x p x
p x p x c c p x p x
d d p x p x
I - A
d d p x p x
−
= =
=
p
(5.2.11)
with a relationship similar to (5.2.2),
*
1 1 2 2( ) ( ) ( ) ( )n n n n n n nm m np x c p x c p x c p x= + + + . (5.2.12)
Since cell population cannot be infinite, the limit of * ( )n np x must exist. Define the
requirement condition
lim 1nmm
m
c→
= , (5.2.13)
which indicates that the accuracy of * ( )n np x is determined by the amount of
experimental data. Based on this assumption, we can determine whether the accuracy
of the prediction results can be further improved and the affected components in the
following sections.
When m n= , (5.2.12) has a particular solution like formula (5.2.2),
0
1 1 2 2( ) ( ) ( ) ( )n n n n n n nn n n nnp x c p x c p x c p x p= + + + + . (5.2.14)
Hence,
1( )−= 0p I -C p , (5.2.15)
with
0 0
11 1
0 0
1
n
n nn
p p
p p
=
0p . (5.2.16)
67
This particular solution (5.2.15) shows that the cell density distribution p is linear
with the initial cell density distribution 0
p .
We further consider the conditions for the existence of this solution. Consider a time
series 1 2, , nt t t and corresponding cell density 1 2, , np p p . If it satisfies the
relation (5.2.15), the cell growth during time series is linear, which is different from
the log growth of cell proliferation. A linear growth can be approximately realised in
a short time interval between t and t t+ . When the relation 1( )−I -C is invariant in
this time interval, the distribution of cell density should be the same regardless of how
many times experiments are performed. When this time interval becomes longer, the
matrix D is affected by the experimental observation. Even if the relation 1( )−I - A
is invariant, the distribution of cell density p is altered with the number and duration
of experiments. Consequently, if there is an initial distribution of cell density 0
p ,
there must be a corresponding interval ct , satisfying
1
1
( ) ,
( ) ,
c
c
t t
t t
−
−
=
=
0
0
p I - C p
p I - A Dp (5.2.17)
where t indicates the time duration of the experiment. Equation (5.2.17) introduces
the matrix of cellular function to the distribution of cell density. When ct t , the
initial distribution of cell density p directly determines the final distribution of cell
density 0
p . When ct t , the final distribution of cell density p is determined by
the initial distribution of cell density 0
p and the matrix of cellular function D
together.
68
5.2.2 Distribution of cell density under external stimulus
According to the equation (5.2.17), both the matrix D and the initial distribution of
cell density 0
p can be affected by an arbitrary external stimulus. In this section, we
firstly assume a coupling effect of the external stimulus on the matrix D and initial
distribution 0
p . The coupling effect results in an evolution of cell distribution
changed with time under the external stimulus, which is different from the traditional
distribution of particles under a force. The difference might be observed when
inputting the data of in vitro experiment into a particle distribution model under an
external stimulus. The model from Keller and Segel [174] for the chemotactic
movement of particles is employed to develop a continuum-based model of cellular
movement,
2
1
2
2
{ ( , ) } ( , )
( , )
p
F
D ft
D ft
= − +
= +
pp p p F p p F
FF p F
(5.2.18)
where p represents the distribution of cell density ( , )p x t and F represents the
distribution of an arbitrary external stimulus ( , )F x t . PD and FD are referred to as
diffusion coefficients of cell and stimulus distribution respectively. The stimulus
affects the cell density in a negative-cross term which is modified by the sensitivity
( , ) p F . Cellular and external kinetics are given by 1f and 2f respectively.
This model has been shown to demonstrate pattern forming properties with cells
accumulating into high-density aggregations. Therefore, this model results in blow-up
after finite time steps without specific limitation.
Blow-up implies the formation of cell aggregates into infinite density, which is
unrealistic from a biological standpoint. The additional conditions can be set for the
69
sensitivity ( , ) p F with (i) (0, ) 0 F and (ii) existence of a 0ip for
( , ) 0i ip F = . The simplest solution is
0( , ) (1 )p F p = − , (5.2.19)
which was rigorously proved that the above condition could lead to repulsion at high
densities and a limit upon the aggregation size [175]. Equation (5.2.19) indicates the
density-dependence of cell motility, which may arise via a series of processes, such as
(i) the dependence between the cell migration and the space availability within cellular
environment, (ii) cellular adhesion involving the contact and attachment between
neighbouring cells via extracellular receptors, and (iii) the secretion of external
diffusible substances which allows cells to intercommunicate.
5.3 Cell probability
In the absence of any condition concerning the external stimulus, it is assumed that all
positions of the cells in the asymmetric cluster are equally probable. On this basis, it
is possible to determine the prior probability distribution of a structure factor. If an
experiment is performed and a certain magnetic intensity is observed, we may obtain
a posterior probability distribution of the structure factor. Our fundamental postulate
is that the posterior distribution is primarily determined by the values of observed
intensities when the number of observed intensities is sufficiently large, and this
distribution is relatively insensitive to the assumed prior distribution form. For this
purpose, we need to assume even less; the structure factor depends only on the values
of a sufficiently large number of observed magnetic intensities. This assumption is
plausible since the observed experimental results are not highly arbitrary. It is,
therefore, reasonable to suppose that, provided the number of available intensities is
greater than the minimum number required to determine the structure in the strict
70
algebraic sense, any structure factor of posterior distribution will agree with its real
value. The higher the number of observed intensities, the higher will be the statistical
significance, and the more reliable the final answer.
The probability distribution P(A) of a structure factor is an even function of A given
that the cells in the asymmetric cluster are assumed to occupy all positions with equal
probability. Thus, the structure factor can be positive as well as negative. However,
once a set of magnetic intensities is known, the cells in a cluster no longer occupy all
positions with equal probability. If the cells in the asymmetric unit are assumed to
move randomly throughout the unit cluster subject to the constraints imposed by the
magnetic intensities, the resulting probability distribution of a structure factor is no
longer an even function. The probability that the structure factors have a particular
sign is different from one-half. The purpose is to derive these probabilities on the basis
that specific magnetic intensities are specified and to derive a procedure for phase
determination. The probability distribution of a structure factor is derived for a cluster
of cells (Figure 5.1). The structure factor for the cluster is given by
1
( , , , )N
j j jh jhj
F f x y z h=
= , (5.3.1)
where jh
f is the scattering factor, N is the number of cells in the unit and
( , , , )j j jx y z h is a trigonometric function of h ( , , )h k l= . The probability that
( , , , )j j j jx y z h = lies between c and c+dc is denoted by ( )p c dc , where
( )p c is an even function within c. By this means, the coordinates of the jth cell cannot
exceed the region within c. Therefore, a cluster existing at c is given by
1 1( ) ( ) ( )N NR
Q c p p d d = , (5.3.2)
71
where the region R in the -space of N dimensions consists of points 1( , , )N
and the corresponding points ( , , )j j jx y z , 1,2, ,j N= yield to the equation
(5.3.1). Due to the region R within c, we introduce a discontinuous function
1( , , )NT to simplify the region with
1 1 1( ) ( ) ( ) ( , , )N N NQ c p p T d d
− −= , (5.3.3)
where
1
1
( , , ) 0
( , , ) 1
N k
N k
T if F c
T if F c
=
= . (5.3.4)
Moreover, kF inherits from (5.3.1),
-1
1
, , 1,2, , .k
k j j k k k k
j
F f F F f k N =
= = + = (5.3.5)
We establish the function 1( , , )NT to satisfy (5.3.4),
10
sin(( ) )1 1( , , )
2
kN
F c xT dx
x
−= − . (5.3.6)
Then
1 10
1 10
sin(( ) )1 1( ) ( ) ( ) ( )
2
1 1( ) ( ) sin(( ) )
2
kj N N
N k N
F c xQ p p dx d d
x
dxp p F c x d d
x
− −
− −
−= −
= − −
(5.3.7)
and
72
-1
-1 -1
( )sin(( ) )
( )sin(( ) )
( )sin(( ) )cos( ) ( )cos(( ) )sin( ) .
k k k
k k k k k
k k k k k k k k k k
p F c x d
p F f c x d
p F c x f x d p F c x f x d
−
−
− −
−
= + −
= − + −
(5.3.8)
When 1k = , we define
0 0F = , (5.3.9)
moreover, rewrite(5.3.8) at 1k = ,
1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
( )sin(( ) )
( )sin( )cos( ) ( )cos( )sin( )
sin( ) ( )cos( ) cos( ) ( )sin( ) .
p F c x d
p cx f x d p cx f x d
cx p f x d cx p f x d
−
− −
− −
−
= − +
= − +
(5.3.10)
Since ( )p a is an even function, we wish that its integral is also an even function.
Then we approximate (5.3.10) at c by
1 1 1 1( )sin(( ) ) sin( ) ( )cos( )p F c x d cx p c f xc dc
− −− = − . (5.3.11)
Then
-1( )sin(( ) ) sin(( ) ) ( )k k k k kp F c x d F c x q f x
−− = − , (5.3.12)
where
( ) ( )cos( )k kq f x p c f xc dc
−= . (5.3.13)
The (5.3.12) and (5.3.13) are an approximation to (5.3.8), thereby simplifying
(5.3.7) by
73
0
1
1 1 sin( )( )
2
N
k
k
cxQ q f x dx
x
=
= + . (5.3.14)
According to (5.3.14), we find the probability that a cluster lies between arbitrary
area A and A+dA,
0
1
( ) 1( ) cos( ) ( )
N
k
k
Q AP A Ax q f x dx
dA
=
= = . (5.3.15)
To solve 1
( )N
k
k
q f x dx=
, we use
11
log ( ) log( ( ))N N
k k
kk
q f x dx q f x dx==
= , (5.3.16)
and
11
( ) exp( log( ( )))N N
k k
kk
q f x q f x==
= , (5.3.17)
and find the Maclaurin series of ( )kq f x from (5.3.13)
2
2
1
( 1) ( )( ) 1
(2 )!
n n
kk n
n
f xq f x m
n
=
−= + , (5.3.18)
where
2
2 ( )n
nm c p c dc
−= . (5.3.19)
n is a positive integer. Therefore, we rewrite (5.3.16) in Maclaurin series
74
11
2
2
1 1
2 2 22
2 2 2
1 1 1 1
2
2
1
( ) exp( log( ( )))
( 1) ( )exp( log(1 ))
(2 )!
( 1) ( ) ( 1) ( ) ( 1) ( )1 1exp( (1 ( ) ))
(2 )! 2 (2 )! 3 (2 )!
( 1)exp(
(2 )!
N N
k k
kk
n nNk
n
k n
n n n n n nNk k k
n n n
k n n n
n n
n
n
q f x q f x
f xm
n
f x f x f xm m m
n n n
x
n
==
= =
= = = =
=
=
−= +
− − −= − + −
−=
2 222 2
2 2 2
1 1
( 1) ( 1)1 1(1 ( ) )) ,
2 (2 )! 3 (2 )!
n n n n
n nn n n
n n
x xm m m
n n
= =
− −− + −
(5.3.20)
where
2
2
1
Nn
n k
k
f=
= . (5.3.21)
When 1n = the first term of (5.3.20) is
2 2 2 422 2 2
2 2 2
2 2 4 3 62 32 2 2
2 2 2
(12 4 12
.2 8 24
x x xm m m
x x xm m m
− + + +
= − − − −
(5.3.22)
When 2n = the second term is
4 2 8 3 122 34 4 4
4 4 4 .4! 2 4! 3 4!
x x xm m m
− + −
(5.3.23)
Substituting (5.3.22) and (5.3.23) into (5.3.20), and expanding Taylor series except
for the term 2x
22 2 42 2
2 2 4 4
1
22 2 42 2
2 2 4 4
1( ) exp( )exp( (3 ) )
2 4!
1exp( ) (1 (3 ) ) .
2 4!
N
k
k
m xq f x m m x
m xm m x
=
= − − −
= − − − +
(5.3.24)
75
Substituting (5.3.24) into (5.3.15), we find the cluster probability P(A),
2
2 2 42 22 2 4 4
0
2 2 2 4 42 2 42 2
2 2 4 40
1 1( ) cos( )exp( )(1 (3 ) )
2 4!
1 1exp( )(1 )(1 (3 ) ) .
2 2! 4! 4!
m xP A Ax m m x dx
m x A x A xm m x dx
= − − − +
= − − + − − − +
(5.3.25)
Using the form of Gaussian integral,
2 2
20exp( )
2
x adx
a
− = , (5.3.26)
2 2 1 2 22
2 10
(2 -1)!exp( ) ( ) (2 -1)!
2 2 2
nn n
n
x a n a ax dx n
a
+
+− = = , (5.3.27)
with 2
2 2
2a
m= to solve (5.3.25)
4 2 22
2 2 4 4
2 2
2 2 2 2 2 2
31( ) (1 )
2 2 4
A m mAP A
m m m
− += − + − . (5.3.28)
The form of (5.3.28) implies
2 22 2 4
2 2 4 4
2 2 2 2
2 2 2 2 2 2 2 2 2 2
31( ) exp( )(1 (1 ) ).
2 2 8 2 8
m mA A AP A
m m m m m
−= − − + − + −
(5.3.29)
Under the condition,
2 2( )n
n = (5.3.30)
the approximation of (5.3.29) is given by
76
22 2
2 4
2
2 2 2 2 2 2 2
22
2 4
2
2 2 2 2 2 2 2
31( ) exp( )(1 exp( ))
2 2 8 2
31 1exp( ) .
2 2 8 2
m mA AP A
m m m m
m mA
m m m m
−= − −
−= − −
(5.3.31)
From (5.3.31), we notice that P(A) has a very similar form to the probability density
of the normal distribution with a variance 2 2m which is independent of the
coordinate system. The boundary conditions of 2 2m as well as (5.3.31) depend on
the particular position of clusters (5.3.19) and scattering coefficient of cells (5.3.21),
2 2
2 2
1
( ) , N
k
k
m c p c dc f
−=
= = . (5.3.32)
The prior distribution may be found regardless that the cluster contains cells in
particular positions or general positions.
The phase problem is to determine the phases of structure factors h
F in (5.3.1) and it
may get a practical solution from a joint probability distribution of the structure factors.
Although the magnitude of h
F is independent of the choice of origin, the cluster
structure alone is not sufficient to determine all phases because the definition (5.3.1)
implies an initial origin. Both the functional form j and the values of the cell
coordinates ( , , )j j jx y z depend on which centre is chosen as the origin. However, for a
fixed functional form of structure factor, it is possible to find certain linear
combinations of the phases whose values depend on the structure alone and are
independent of the choice of origin within a range permitted by the chosen form of
structure factor. This property leads to structure semi-invariants which depend only on
the choice of the unit cluster, and their values depend on the chosen form of structure
factor. If a sufficient number of magnetic intensities can determine the cluster
77
structure, these intensities determine the values of structure semi-invariants rather than
the phases of cells. Therefore the phase problem can be started with the problem of
determining the values of structure semin-variants for each fixed form of structure
factor, once a sufficient number of intensities has been tested. The phases may then be
obtained from the values of the structure semi-invariants by specifying origin. As
introduced in the previous section, the probability theory provides a solution for the
identification of the structure sem-invariants. On the other hand, it should be
emphasised that the understanding of the form of structure semi-invariants is an
invaluable aid in interpreting the results from the probability theory. In the following
section, we devote to discuss this subject for a cluster with cells in general positions.
In general the function in (5.3.1) depends on the choice of origin which ideally is a
centre of symmetry. Two origins are equivalent if the functional forms of are
identical. In other words, two origins are equivalent if they are geometrically related
in the same way to all the symmetry elements. It is noticed that the value of referred
to two similar origins may be equal or contrary to each other. The concept of equivalent
origins leads to equivalent structure factors. If two structure factors are equivalent,
their structure semi-invariants have the same value. As previously mentioned, the
values of the structure invariants are only determined by the structure while the values
of the phased also depend on the choice of origin. It will be seen that the origin may
be chosen by selecting the form of the structure factor and then specifying the signs of
structure factors.
5.4 Joint cell probability
For the probability distribution of a structure factor when certain magnitudes or phases
are specified is readily derivable from the joint distribution. We define the joint
78
probability 1 1( , , )j jm j jmp d d for the interval between jn and jn jnd + , where
n 1,2, ,m= and m is a positive integer. Hence, a similar definition to (5.3.1) is
written
( , , , )njn j j jx y z h = . (5.4.1)
Let 1 1 1( , , )m mP A A dA dA be the joint probability for the interval nA and
n nA dA+ ,
1 m
1 1 m
1 m
Q(A , ,A )P (A , ,A )=
A A
m
, (5.4.2)
the probability 1( , , )mQ A A is estimated following the procedure from (5.3.3) to
(5.3.6) ,
1 1 1 1
1 1
( , , ) ( ( , , ) ) ( , , )N m
m j jm j jm n mn
j n
Q A A p d d T
− −= =
= ,
(5.4.3)
where
10
0
sin(( ) )1 1( , , )
2
exp( ( )) exp( ( ))1 1 ,
2 2
kn n nn mn n
n
n kn n n kn nn
n
F AT d
i F A i F Ad
i
−= −
− − − −= −
(5.4.4)
and
1
1
( , , ) 0
( , , ) 1 .
n mn kn n
n mn kn n
T if F A
T if F A
=
= (5.4.5)
79
We continue the procedure from (5.3.7) to (5.3.14), and find that 1( , , )mQ A A
can be written in two parts
1 1 2 1( , , ) ( , , )m mQ Q A A Q A A= + . (5.4.6)
It indicates
11 1 1 1 1 1
0 01
1
exp( )1 1
( , , ) ( , , ),2 (2 )
m
n n Nn
m m j jn mmm mj
n
n
i A
Q A A d d q f f
i
=
=
=
−
= −
(5.4.7)
where
1 1 1 1 1
1
( , , ) ( , , ) exp( )m
j jm m j jm jn jn n j jm
n
q f f p i f d d
− −=
= ,
(5.4.8)
12 1 1 2 1 1
0 01
1
exp( )1 1
( , , ) ( , , ),2 (2 )
m
n n Nn
m m j jn mmm mj
n
n
i A
Q A A d d q f f
i
=
=
=
−
= −
(5.4.9)
where
2 1 1 1 1
1
( , , ) ( , , ) exp( )m
j jm m j jm jn jn n j jm
n
q f f p i f d d
− −=
= − .
(5.4.10)
Therefore,
1 1 1 1 1 10 0
1 1
1( , , ) (exp( ) ( , , ) ,
(2 )
Nm
m n n j jm m mmn j
P A A i A q f f d d
= =
= −
80
(5.4.11)
and
2 1 2 1 1 10 0
1 1
1( , , ) (exp( ) ( , , ) .
(2 )
Nm
m n n j jm m mmn j
P A A i A q f f d d
= =
=
.
(5.4.12)
Equations (5.4.11) and (5.4.12) are the starting point for deriving the probability
distribution by the structure factors. As in the derivation of (5.4.11) and (5.4.12) the
cells in the asymmetric unit are assumed to move randomly throughout the cluster
except that they are restricted by known magnetic magnitudes or a specified set of
structure factors. Then we follow the procedures described from (5.3.18) to (5.3.19)
and obtain
1
1 1 1 1 1
0
( )
( , , ) ( , , )!
mk
jn jn n
nj jm m j jm j jm
k
i f
q f f p d dk
=
− −=
=
,
(5.4.13)
12 1 1 1 1
0
( )
( , , ) ( , , )!
mk
jn jn n
nj jm m j jm j jm
k
i f
q f f p d dk
=
− −=
−
=
.
(5.4.14)
By defining a mixed moment
1
1 1 1 1( , , ) m
m j jm j jm j jmm p d d
− −= , (5.4.15)
we obtain a simple form of (5.4.13) and (5.4.14)
1
11 1 1
0
( )
( , , )!m
mk
jn n
nj jm m
k
i f
q f f mk
=
=
=
, (5.4.16)
81
1
12 1 1
0
( )
( , , )!m
mk
jn n
nj jm m
k
i f
q f f mk
=
=
−
=
, (5.4.17)
where m N and the interdependence of the vectors is related to the value of
moments. For k is even, the joint probability distribution is likely to vanish. For k is
odd, the joint probability distribution is enhanced. We consider a joint probability of
two interdependent events ( 2m= ) and (5.4.16) expresses as
1 1 1 2 2
2 2 2 2
1 1 10 2 2 01 1 1 20 2 2 02 1 2 1 2 11
( , )
11 ( ) ( 2 )
2
j j
j j j j j j
q f f
i f m f m f m f m f f m
=
+ + − + + −
(5.4.18)
Similarly, (5.4.17) expresses as
2 1 1 2 2
2 2 2 2
1 1 10 2 2 01 1 1 20 2 2 02 1 2 1 2 11
( , )
11 ( ) ( 2 )
2
j j
j j j j j j
q f f
i f m f m f m f m f f m
= − + + + + +
(5.4.19)
Then, we can solve (5.4.11) and (5.4.12) in the scenario ( 2m= ) by following the
procedure from (5.3.20) to (5.3.24)
2 2 2 2
1 1 1 2 2 20 1 1 02 2 2
1 11
10 1 1 01 2 2 11 1 2 1 2
1 1 1 1
1 1( , ) exp( )
2 2
(1 ),
N N N
j j j j
j jj
N N N N
j j j j
j j j j
q f f m f m f
im f im f m f f
= ==
= = = =
= − −
+ + − −
(5.4.20)
and
2 2 2 2
2 1 1 2 2 20 1 1 02 2 2
1 11
10 1 1 01 2 2 11 1 2 1 2
1 1 1 1
1 1( , ) exp( )
2 2
(1 ).
N N N
j j j j
j jj
N N N N
j j j j
j j j j
q f f m f m f
im f im f m f f
= ==
= = = =
= +
− − + +
(5.4.21)
82
Substituting (5.4.20) and (5.4.21) into (5.4.11) and (5.4.12), we obtain
2 2
1 21 1 2
2 2 2 2
20 02 1 2 20 1 02 2
1 1 1 1
1 1( , ) exp( ) (1 )
22 2
N N N N
j j j j
j j j j
A AP A A
m m f f m f m f
= = = =
= − − +
(5.4.22)
and
2 2
1 22 1 2
2 2 2 2
20 02 1 2 20 1 02 2
1 1 1 1
1( , ) exp( ) (1 )
22 2
N N N N
j j j j
j j j j
A AiP A A
m m f f m f m f
= = = =
= + −
.
(5.4.23)
Therefore, the probability distribution 1 2( , )P A A contains the distribution of the
real part (5.4.22) and an imaginary part (5.4.23).
5.5 Correlation between osteoblasts and osteoclasts
The section specifically addresses the correlation between osteoblasts and osteoclasts.
These two types of cells are relevant and interacted in BMU which has been mentioned
in Chapter 2. Equations (5.4.22) and (5.4.23) propose a structure for two relevant types
of cells in a joint probability distribution. The structure illustrates the orthogonal
relationship between the joint probability distribution. We assume that the interaction
between osteoblasts and osteoclasts also satisfy an orthogonal relationship.
Set cell probability of osteoblasts with a vector ( , )r tOB and osteoclasts with ( , )r tOC
, where r represents space and t represents time. ( , )r tOB and ( , )r tOC satisfy the
orthogonal form in the following,
( )OBt
− = −
OCOB P , (5.5.1)
83
2
2
1( )OC
t t t
= +
OC OB OB , (5.5.2)
where , ,OB OC indicate transition constants for osteoblasts and osteoclasts
respectively. P denotes the factor of apoptosis. Submitting (5.5.1) into (5.5.2),
2
2
1[ ( )] 0OC OB
OB t t
− + + =
P OB OBOB , (5.5.3)
with the boundary conditions
0n =OB , (5.5.4)
0n =OB . (5.5.5)
For an isotropic condition, apoptosis is proportional to cell population
OB OB =P OB . (5.5.6)
Substituting (5.5.6) into (5.5.3),
2
2
1[ (1 ) ] 0OC OB OB
t t
− + + =
OB OBOB . (5.5.7)
The solution is obtained by the extreme value of the functional, which is a one-field
principle in term of OB ,
( ) ( )2
2
1 2( ) (1 ) 2 d ,
2OC OB OB
VV
t t
= − • + • + •
OB OBOB OB OB OB OB
(5.5.8)
and
( ) ( ) ( ) ( ) 1
(1 ) (1 ) d2
1 1 d .
OC OB OB OC OB OBV
V
V
V
= − • + − •
+ • + • + • + •
OB OB OB OB
OB OB E OB OB OB OB OB
(5.5.9)
84
The first variation on the time difference is set to zero for an arbitrary instance t ,
0, 0 = =OB OB . (5.5.10)
It can be simplified as
( ) ( ) ( ) ( ) 1
(1 ) (1 ) d2
1 d .
OC OB OB OC OB OBV
V
V
V
= − • + − •
+ • + •
OB OB OB OB
OB OB OB OB
(5.5.11)
Consider the first term on the right-hand side, which can be rewritten as
( ) ( ) ( ) ( )
( ) ( )
1(1 ) (1 ) d
2
(1 ) d .
OC OB OB OC OB OBV
OC OB OBV
V
V
− • + − •
= − •
OB OB OB OB
OB OB
(5.5.12)
To transfer the curl operator away fromOB , we invoke the differential vector
calculus identity
( )• = • − •a b b a a b , (5.5.13)
( )• = • −• a b b a a b . (5.5.14)
Let
( )(1 ) , OC OB OB − → →OB a OB b , (5.5.15)
Hence,
( ) ( )
( ) ( )
(1 )
(1 ) (1 ) ,
OC OB OB
OC OB OB OC OB OB
− •
= • − −• −
OB OB
OB OB OB OB
(5.5.16)
85
and
( ) ( )
( ) ( )
(1 ) d
(1 ) d (1 ) d .
OC OB OBV
OC OB OB OC OB OBV V
V
V V
− •
= • − − • −
OB OB
OB OB OB OB
(5.5.17)
Applying the divergence theorem
d dV S
V S• = • f n f , (5.5.18)
we obtain
( ) ( )
( ) ( )
(1 ) d
(1 ) d (1 ) d .
OC OB OBV
OC OB OBV
OC OB OBS
V
V S
− •
= • − − • −
OB OB
OB OB n OB OB
(5.5.19)
Using the vector identity
( ) ( ) ( )• = • = • a b c b c a c a b , (5.5.20)
we further have
( ) ( )
( ) ( )
(1 ) d
(1 ) d (1 ) d .
OC OB OBV
OC OB OBV
OC OB OBS
V
V S
− •
= • − − • −
OB OB
OB OB OB n OB
(5.5.21)
Consequently, the first variation equation of the function is rewritten as
( )
( )
(1 ) d
1d(1 ) d ,
OC OB OBV
VOC OB OB
S
V
VS
= • −
− • + •
• − +
OB OB
OB OB OB OBOB n OB
(5.5.22)
86
Rearranging,
( )
( )
( )
( )
( )
1(1 ) d
(1 ) d
1(1 ) d
(1 ) d
(1 ) d
OB
I
OC OB OBV
OC OB OBS
OC OB OBV
OC OB OBS
OC OB OBS
OC O
V
S
V
S
S
= • − + +
− • −
= • − + +
− • −
− • −
− •
OB OB OB OB
OB n OB
OB OB OB OB
E n OB
E n OB
E n ( )(1 ) d .OC
B OBS
S− OB
(5.5.23)
The first integral on the right-hand side enforce the governing equation (5.5.23), the
second integral should vanish due to the prescribed osteoblasts field constraint over
the boundary OBS , the third one enforces the continuity conditions on the inter-
element edges IS when co-considered with those from the neighbourhood elements,
and the last integral enforces the boundary condition on osteoclasts field OCS . When
the stationary requirement 0 = is imposed for arbitrary variation of OB . For the
convenience of the derivation of the stiffness equation, the functional is rewritten in
the form of a vector operation,
( ) ( )2
T T T
2
1 2( ) (1 ) 2 d .
2OC OB OB
VV
t t
= − + +
OB OBOB OB OB OB OB
(5.5.24)
In the finite element formulation, after a robust continuum is discretised into a finite
number of elements, the osteoblasts field OB at an arbitrary point is interpolated
regarding nodal electric fields d
( , ) ( ) ( )t t=OB x N x d , (5.5.25)
87
where N denote the shape function matrix.
1
1
1
2 2
1 1
3
3
1
1
( ) ( )
( , ) ( ) ( ) ( ) ( )
( ) ( )
( ) ( )
n
i i
i
n n
i i i i
i i
n
i i
i
n
j ij i
i
N d t
E
t E d t N d t
E
N d t
OB N d t
=
= =
=
=
= = =
=
x
OB x N x x
x
x
, (5.5.26)
where n is the total number of unknowns
1
2
3
( )
( ) ( )
( )
i
i i
i
N
N
N
=
x
N x x
x
. (5.5.27)
Subsequently, the curl definition of the osteoblasts field OB is given by
( ) ( )1
( ) ( )n
i i
i
d t=
= = OB N x N d , (5.5.28)
with
1 2( ) ( ) ( )n = N N x N x N x . (5.5.29)
We obtain
( ) ( )
( )
TT T T
T T
1 2( ) (1 ) d d
2
d
OC OB OBV V
V
V V
V
= − +
+
E d N N d d N N d
d N N d
,
(5.5.30)
which can be written in matrix form as
T T T1
2 = + +d Kd d Cd d Md , (5.5.31)
88
with
( ) ( )
( )
T
T
T
(1 ) d
2d
d .
OC OB OBV
V
V
V
V
V
= −
=
=
K N N
C N N
M N N
(5.5.32)
The stationary condition of the functional regarding unknown d yields
T
1
2
= + + =
Kd Cd Md 0
d . (5.5.33)
The solution of the stationary condition (5.5.33) should have the similar form of
(5.4.22) and (5.4.23), which includes real part and imaginary part. If the real part is
assigned to reflect the possibility distribution of osteoblasts, the imaginary part will
indicate the possibility distribution of osteoclasts.
89
Figure 5.1 The structure factor. The structure factor can be positive as well as negative.
However, once a set of magnetic intensities is known, the cells in a cluster no longer
occupy all positions with equal probability. If the cells in the asymmetric unit are
assumed to move randomly throughout the unit cluster subject to the constraints
imposed by the magnetic intensities, the resulting probability distribution of a structure
factor is no longer an even function. The probability that the structure factors have a
particular sign is different from one-half. The purpose is to derive these probabilities
on the basis that specific magnetic intensities are specified and to derive a procedure
for phase determination.
90
Chapter 6
Numerical Model:
Predication of Osteoblast Cell Population
under Influence of Surface Grain Size
91
6.1 Introduction
The frameworks developed in Chapter 5 are utilised to investigate cell population of
osteoblasts on material surfaces of nano-grain size. In this case study, a numerical
model is proposed for simulating osteoblast behaviours on two different surface-
engineered biomaterials and investigating the functional importance of specific grain
sizes at the nanophase. The proposed model consists of molecular and cellular motions
based on theoretical and experimental evidence and creates predictive simulations
from sparse experimental data. The comparison of numerical solutions and
experimental data reveals that the proposed model can explain the nonlinearity of the
cell population on material surfaces concerning nanophase grain size (0-100 nm). The
numerical results further provide insight into the optimisation of nanophase grain size
on the surface of the biomaterial. Establishing a mathematical relationship in the
process between material surface and osteoblast allows deducing the coupling effects
of surface roughness and osteoblast adhesion.
6.2 The relationship between osteoblast proliferation and substrate surface
roughness
Consequent to the demographic changes in society with an ageing population and the
corresponding increase in the incidence of the musculoskeletal diseases of ageing such
as arthritis and osteoporosis, there is a pressing clinical need to develop new
biomaterials for orthopaedic implants. In an ideal situation, these materials would be
biocompatible, mechanically robust and actively direct osteogenic progenitor cells,
such as osteoblasts. The interaction between a material substrate and the biological
system is complex [176]. The interplay between the substrate and the extracellular
matrix (ECM) determines the cell response.
The development of a bone-implant biomaterial depends on the interactions of
osteoblasts and material surfaces. The quality of cell and material interactions
92
influence cellular proliferation and differentiation [177]. In surface engineering of
biomaterials, the primary strategy for enhancing the interactions between the cell and
the material surface is to modulate the properties of the material surface, such as
surface charge and roughness, to a state in which ECM adhesion proteins can
appropriately function [178]. Experiments highlight the importance of optimising the
roughness of the material surface in vitro and in vivo. [179] investigated the effect of
surface roughness of hydroxyapatite (HA) on human bone marrow cells and found that
cell adhesion, proliferation and detachment strength were sensitive to the surface
roughness of HA. Intriguingly, osteoblast adhesion and proliferation appear to vary
nonlinearly with the roughness of the material surface. Webster, et al. [180] reported
that osteoblast proliferation was significantly higher on nanophase Alumina and
titanium. Their experiment [181] also showed that the ability of osteoblasts to adhere
to these materials was not proportional to the reduction in grain size. Also, Huang, et
al. [182] observed that significant enhancements of osteoblast adhesion, proliferation,
maturation and mineralisation were exhibited on the nano-grained surface (below 100
nm), but little improvement was found on the ultrafine-grained (100 nm-1000 nm)
surface compared to the conventional coarse-grained surface.
The nonlinearity between material surface roughness and osteoblast adhesion might
be caused by multiple factors, such as cell population motility, proliferation and
synthesis of ECM proteins of anchorage-dependent cells [183]. Biological recognition
mediates the interaction between cells and implantable materials by binding specific
receptors on the cell surface to ligands on the material surface. Such ligands could be
proteins spontaneously absorbed upon the material surface when materials contact
with body fluid in vivo or cell culture medium in vitro [184]. Uncertainties exist as to
not only which aspects of cell behaviours are the most important in the interaction
93
between materials with a surface roughness but also what constitutes the best surface.
For instance, osteoblasts can adhere and maintain active when attached to some
surfaces (e.g., the cell culture plate or surface modified alloys), but almost inactive
when attached to others (e.g., many artificial surfaces).
In vitro and in vivo experimentation about materials and osteoblast cellular responses
to the substrate are expensive and time-consuming. Thus far because of the inherent
complexity of the biological milieu, there is no alternative to these experiments and
strategies in attempting to predict the response to a biomaterial at the cellular level.
Therefore, creating mathematical models of osteoblast responses to implantable
biomaterials is becoming increasingly popular as a strategy in biomaterial and drug
development. As an alternative to experimental approaches, a robust mathematical
model will facilitate more efficient optimisation of the molecular interface between
material surface and osteoblasts.
6.3 Governing equations for osteoblast adhesion on the material surface
We propose a double-layered model for investigating osteoblast adhesion on a material
surface (Figure 6.1). This model concerns the biological process of osteoblast
adhesion on the corresponding material surface with two relatively independent steps.
The first step describes the movement of molecules on the material surface due to the
diffusion-controlled transportation of proteins. This process includes biological
recognition [184]. During the transportation, the molecules have random motions in a
potential field formed by the interactions between the material surface and the cells
attached to the surface. The motions of molecules form a “molecular layer” in the
proposed model, which consists of cell signalling, ligand binding and adsorbed
proteins on the material surface [185]. The second step describes a “cellular layer”
including cell adhesion and migration at the cellular level. Since osteoblasts are
94
anchorage-dependent cells and the process of cell migration is inextricably linked with
the process of cell adhesion [186], we speculate that the process of osteoblast migration
is highly dependent on the process of osteoblast adhesion at the “cellular layer”. We
also assume that the osteoblast adhesion is related to the “molecular layer” because the
extracellular matrix (ECM) and specific transmembrane receptor (e.g., integrins)
contribute critically to the cell adhesion [187].
We introduce a local coordinate to describe the behaviours of the model on a material
surface. In the horizontal direction (x-axis) of the molecular layer, the model describes
the molecular motion by its probability distribution. Similarly, the model describes the
process of cell migration by a cellular probability distribution. Then we separate the
cell adhesion into two variables in the vertical direction (y-axis). One variable
represents the interaction between the material surface and the molecular layer.
Another variable represents the interaction between the molecular layer and the
cellular layer. Finally, the model links the variables in the vertical direction with the
probability distribution of cells and molecules in the horizontal direction. The double-
layered model consequently studies the osteoblast adhesion on a material surface by
identifying the probability distribution of molecules and cells. Based on the
assumptions above, we apply continuum mathematical descriptions to find the
probability distribution of molecules and cells. The descriptions assume the motions
of molecules and cells in stochastic processes, which lead to the Fokker-Planck
equations of stochastic processes for the probability density of molecules mp and cells
cp .
95
( )
( )
m mm m m
c cc c c
p pD p
t x x
p pD p
t x x
= −
= −
, (6.3.1)
where m and c refer to the drift coefficients among molecules and cells, which
determine their motions under external perturbations. mD and cD refer to the
corresponding diffusion coefficients indicating the random walk. We then consider the
influence of the cell adhesion and adsorbed proteins on the horizontal motions of
molecules and cells,
( ( ) )
( )
m mm m m
c cc c c
p pD p
t x x x
p pD p
t x x
= −
= −
, (6.3.2)
where path modulator expresses external perturbation from the surface roughness
R . When culture cells on a substrate, the surface roughness prevents the dispersion
of molecules through the friction between the molecular layer and material surface and
subsequently affects the cellular motion. Therefore, the probability density of
molecules mp and cells cp are bounded by conditions such that
( , , ( ), )
( , , )
m m m
c m c c
p f R D
p f p D
. (6.3.3)
The functions of drift coefficients m and c in Equation (6.3.2) are unknown
without experimental data.
6.4 Numerical solution methods for the double-layered model
Using the Ornstein-Uhlenbeck process as a reference for defining the motion of the
molecular layer, we attempt to obtain a numerical solution of the model. The Ornstein-
96
Uhlenbeck process describes the velocity of diffusing particles under the influence of
friction, in which there is a random walk tendency to move towards a central location.
Here we assume that the molecular motions follow a standard process, thereby creating
a probability distribution of molecules. The probability density of molecules mp of
the Ornstein-Uhlenbeck process satisfies
2
0 2( ) [( ) ]m
m m
px x p p
t x x
= − +
, (6.4.1)
where ( ) is environmental sensitivity of the molecular layer and is the
diffusion coefficient of the molecular layer. Taking 0 0x = for simplicity, the initial
condition of the equation (6.4.1) is
2
2 ( ) 2 ( )
( ) ( )exp[ ]
2 (1 ) 2 (1 )m t t
xp
e e
− −
−=
− − . (6.4.2)
Alternatively, the molecular sensitivity can be expressed as
( )
= . (6.4.3)
We substitute equation (6.4.3) into the equation (6.4.2) and solve the stationary
solution,
2
( ) exp( )2 2
m
xp x
= − , (6.4.4)
and the molecular sensitivity is dependent on the material surface property R ,
( )mf R = . (6.4.5)
Based on the equation (6.4.5), the surface property may hinder molecular motions and
alter the molecular probability density. For the continuity, the initial cellular
97
distribution should inherit the characteristics of the final molecular distribution. Thus,
we assume the initial probability density of osteoblasts to be approximately equal to
the stationary probability density of molecules.
0( , ) ( )c mp x t p x= , (6.4.6)
where 0( )cp t is the initial solution for cellular probability density. Considering the
evolution of the cell probability density cp in equations (6.3.2) and (6.3.3), we find
that the complexity of osteoblast adhesion leads to inhomogeneous cell distribution in
the cellular layer. Using the approach of Stevens and Othmer [188], we aim to
generalise an inhomogeneous distribution for a continuous-time discrete-space
random walk along the horizontal direction. This method restricts the time evolution
to one-step time jumps and postulates the discrete quantity ( )ip t as an
approximation to cellular probability density cp at the position ix ih= , iZ as
well as time t beginning at 0i = , 0t = . Thus, ( )ip t evolves in a manner such that
1 1 1 1
( )[ ( ) ( )] [ ( ) ( )]i
i i i i i i i i
dp tp t p t p t p t
dt
+ − + −
− − + += − − − . (6.4.7)
Here i
− denotes the transitional probability that cells enter from 1i − to i , and i
+
denotes the transitional probability that cells leave from i to 1i+ (Figure 6.1). The
equation (6.4.7) is highly dependent on the form of transitional probabilities i
that
reflect the physical parameters of the cell type under investigation. In the context of
the movement of osteoblasts, the transitional probabilities i
are determined by the
ambient cellular probability density 1ip , the cellular sensitivity . The cellular
98
sensitivity works to prevent cells from leaving the present position. Then, a simple
form of i
may be constructed as
2
1 1(1 )(1 ) /i i ip p h
= − − . (6.4.8)
The cellular sensitivity inherits the correlation with surface property R as
( )cf R = . (6.4.9)
Substituting equation (6.4.8) into equation (6.4.7) gives
1 22
( ) 1[ ( ) ( ) (1 ( )) ( )] 0i
i i
dp tp t f t p t f t
dt h+ − − = , (6.4.10)
where
1 1 1 1 1( ) (1 ( ))(1 ( )) (1 ( ))(1 ( ))i i i if t p t p t p t p t + − − += − − + − − , (6.4.11)
and
2 1 2 1 2( ) ( )(1 ( )) ( )(1 ( ))i i i if t p t p t p t p t − − + += − + − . (6.4.12)
Equations (6.4.11) are unconditionally stable for 0 ( ) 1ip t . For matching the
experiment data, the probability density of cells is converted to the cell density per unit
area. We estimate the probability density of cells at the discrete position i and
integrate the probability density by a circular domain on the assumption that the
cellular probability density cp is equal at each position with the same radius,
00
( ) 2r
k cP t P xp dx= , (6.4.13)
where 0P is the initial cell density at 0i = , 0t = . x and t denote the spatial and
temporal variables respectively. The scale of the studied cell determines the number
of points. The cell density kP is the integration of cellular probability density cp .
99
6.5 Numerical simulation on experimental data
In this section, we used our model to simulate osteoblast adhesion on two different
material surfaces of nanophase grain size by inputting the published experimental
results [181]. Webster et al. provided evidence that osteoblast adhesion enhanced on
nanophase Alumina ( 2 3Al O ) and Titania ( 2TiO ) in vitro. The method of evaluating cell
adhesion was to lift osteoblasts enzymatically using trypsin and to count the number
of adherent cells in five random fields per substrate. The authors observed that the
number of adherent osteoblasts increased with time, and the cell density of adherent
osteoblasts varied on the surfaces of Alumina and Titania with different grain sizes.
From their observation, they concluded that an increase in osteoblast adhesion could
be accompanied by a decrease in Alumina and Titania grain size.
Since the osteoblast size was around 20 µm [189], 250 osteoblasts could be arrayed
along a radius of 0.5 cm. In 1 cm2, the size was set at 500 points. The model was
initialised with molecular sensitivity and cellular sensitivity by fitting the
experimental data of adherent osteoblasts on Alumina and Titania. As an example, the
model fitted the experimental results with 2.2 = and 0.26 = on Alumina of
grain size 77 nm (Figure 6.2 A), and with 0.6 = and 0.22 = on Titania of
grain size 32 nm (Figure 6.2 B). We then fitted the data for each grain size of Alumina
and Titania in the experiment. All and were plotted along grain sizes to
determine the functions of molecular and cellular sensitivities. Trendlines were added
to find the modified functions of molecular and cellular sensitivities by curve fitting.
For Alumina, the functions were cellular sensitivity (Figure 6.3 A),
0.24 0.11exp( 0.018 )Al R = + − , (6.5.1)
and molecular sensitivity (Figure 6.3 B),
100
2.27 0.90exp( 0.027 )Al R = − − . (6.5.2)
The functions for Titania were found for cellular sensitivity and molecular sensitivity.
For both Alumina and Titania, the molecular sensitivity increased with grain size
whereas the cellular sensitivity decreased with grain size. We checked the correlation
of exponential fitting for both materials and found that the sensitivity functions for
Titania had better consistency with the experimental data. The experimental data
varied around the sensitivity functions of Alumina, in particular for the cellular layer.
These results may imply that the cell adhesion on Titania might be stable whereas the
cellular motions of osteoblasts on Alumina are likely to be affected by the external
perturbation.
Based on the modified functions, we recalculated the model to show the cell adhesion
in the profile of adherent cell density with both time and grain size. We further draw
the profile with different colours indicating the various cell density. As can been seen,
although both the cell density of adherent osteoblasts on Alumina (Figure 6.4 A) and
Titania (Figure 6.4 B) increased with time, their paths were predicted to be in different
ways. The predicted adherent cell density on Alumina was observed that the adherent
cell density keeps at the same number in the beginning, but the duration varied from 1
to 2 hours in the range of grain size from 0 to 100 nm. The duration approximately
reached an average of 1.5 hours when the grain size was 35 nm. Then the predicted
adherent cell density increased quickly and reached 3200 cells/cm2 after 4 hours. This
process was fast when the grain size of Alumina under 40 nm.
In contrast to the Alumina, the predicted adherent cell density on Titania increased in
a lag phase but varied on the grain size. When the grain size of Titania was beyond 40
nm, the predicted adherent cell density increased in a linear approximation of time.
Between 10 nm to 40 nm, the predicted path showed a similar trend with that of
101
Alumina. Under 10 nm, the predicted adherent cell density appeared no longer to
increase. The predicted results inspired us to think about the existence of critical grain
size on Alumina and Titania in the following section.
6.6 Comparison with experimental data and prediction
In this section, we compare the numerical results from the proposed model with the
original experimental results [181]. The numerical results reflect the historical profile
of adherent cell density concerning time and grain size. Here we focus on discussing
the influence of grain size on the adherent cell density. The numerical result of
adherent cell density on Alumina matches Webster’s experimental data (Figure 6.5
A). The adherent cell density is significantly higher on Alumina with grain sizes in the
range 0–40 nm than on Alumina with grain sizes in the range 60–100 nm. There is an
exponential decay on Alumina with grain sizes 40–60 nm. The numerical result of
Titania also agrees with the experimental data (Figure 6.5 B). The adherent cell
density is significantly higher on Titania with grain sizes in the range 0–20 nm and
followed by a linear decay from 20 to 100 nm. Webster et al. observed in their
experiments that variation existed in the adherent cell density of adherent osteoblasts
per grain size and concluded that there might be a critical grain size of Alumina and
Titania in mediating osteoblast adhesion. A critical grain size here can be defined as a
point on the axis of grain size in which the slope of adherent cell density and grain size
dependence significantly changes. On Alumina, our model identifies two critical grain
sizes at 36 nm and 60 nm, an observation which agrees on the with the conclusion of
Webster et al. that there may be a critical grain size between 49 nm and 67 nm. Our
model predicts a critical grain size at 36 nm with cell density 3042 cells/cm2. This
potential interaction between grain size and adherent cell density has not previously
been identified.
102
In contrast to the findings for Alumina, our model predicts a different interaction
between the osteoblast and Titania. The Webster’s experimental data implied a range
of the critical grain size for Titania from 32 nm to 56 nm, whereas the model predicts
a continuous decline of adherent cell density from 30 nm to 100 nm. Further
experiments, particularly examining grain sizes in the range of 40-70 nm, may well
clarify this issue further. The model surprisingly indicates that adherent cell density on
Titania drops significantly with the grain size smaller than 10 nm. There is limited
published experimental data regarding culturing osteoblasts on a substrate surface
made of Titania with grain size under 10 nm. Park, et al. [190] demonstrated that the
one-dimensional surface nano-topography of 15 nm Titania nanotubes promoted
osteoblast formation. They evaluated the adhesion, spreading and growth of
osteoblasts on the surface of Titania nanotubes from 15 nm to 100 nm diameters. Cells
were found to adhere and proliferate best on 15 nm tubes, and the adhesion decreased
with an increase in the nanotube diameter. Although the topography of nanophase
ceramic is different from the nanotube, it has been speculated that an optimal spacing
between cell and substrate exists for osteoblast adhesion on Titania. The putative
optimal spacing may derive from focal contacts or adhesion plaques, which are
junction locations of about 10-15 nm between adherent cells and material surfaces
[191]. On the other hand, theoretically, smaller grain sizes provide a more available
surface area of higher energy to promote protein interactions such as vitronectin and
fibronectin for adhesion [192].
The optimal spacing may well represent a combination of the coupling of focal
contacts and topography between osteoblasts and material surfaces. The topography is
a function not only of the material surface itself but also of the adherent protein layer
intimate with that surface. The focal contacts at the protein layer can create an optimal
103
spacing where the osteoblast adhesion is the highest, and adhesion is reduced if the
spacing becomes either smaller or larger. The spacing is altered by factors such as the
thickness or the composition of adsorbed proteins, both of which are highly dependent
on the grain size of the material surface. Smaller grain sizes promote protein
interactions so that they form a thicker layer. This potential mechanism may explain
the trends seen in the model’s predicted curve of Alumina with optimal grain sizes in
the range 36–60 nm and the curve of Titania with grain sizes in the range 10–100 nm.
It is notable that these predicted curves of both Alumina and Titania show stable trends
with increasing grain sizes beyond the above predicted critical values. This finding
agrees with the previous experimental observation [182] as well as with the
considerable clinical experience of the practical use of both these biomaterials where
grain size is not homogeneous in the end-use product. Other factors apart from
nanophase grain size may also play a role in cell-substrate adhesion. The surface
topographies of Alumina and Titania differ, with the arrangement of Alumina grains
being more linear than that on the Titania surface. As cell-substrate adhesion includes
integrins that have nanoscale features, cells will respond to surfaces with nanoscale
characteristics of the pores, ridges and fibres of the basement membranes [193].
6.7 Parametric analysis
We used the model to simulate the adhesion of osteoblasts on engineered surfaces in
this case. However, it is wondering whether this model is capable of simulating the
growth of osteoblasts in the long term. The growth curves of osteoblasts exhibit three
phases: lag phase, exponential phase and stationary phase. The lag phase of osteoblast
proliferation in vitro suggests the cells are adapted to the culture conditions and some
of them probably died. After that, cell growth is exponential, and a tendency for a
stationary phase is observed after a certain culture time. When osteoblasts are cultured
104
under different conditions, these three phases reflects the temporal characteristics of
osteoblast growth. The functions of molecular and cellular sensitivity can modify the
temporal properties of osteoblast proliferation in addition to influencing both the
maximal and minimal threshold of the cell proliferation.
We begin by considering the molecular sensitvity on cell probability of osteoblasts
(Figure 6.6). The numerical simulation represents the temporal characteristics in
which cell probability of osteoblasts with various molecular sensitvities changes from
different lag phase to one stationary phase. The increase in the molecular sensitvity
leads to long-term of lag phase with low cell density, suggesting that cells are hard to
settle down and likely to die under a high diffusion environment. If we consider that
both the morphological aspects of osteoblasts on materials and molecular sensitvities
are affected exclusively by the surface roughness of materials, the above statement
implies that the spreading of the osteoblast cell and the formation of continuous cell
layer are better on smooth surfaces as compared to the rough ones. The period of
exponential phase is also extended under high molecular sensitvity. The period of
exponential phase continues 100-time steps when the molecular sensitvity equals to
0.5, and it is extended to 276-time steps when the molecular sensitvity equals 10.
Moreover, the amplified area shows that the stationary phase is made by periodical
oscillation. This periodical oscillation indicates the fact that a dynamic balance of cell
proliferation and apoptosis exists even osteoblasts reach the stationary phase.
We further examine the influence of cellular sensitivity on cell probability (Figure
6.7). Numerical results show that the cellular sensitivity [0,1] determines the
qualitative behaviour of solutions to this simulation. For each value of , the
simulation is identified by analysis of dynamic changes in cell density in response to
a gradual increase in time step. The simulation exhibits the opposite behaviour of mode
105
as plotted by a function of . The cell probability of lag phase remains the same
regardless of various values. The increase in values can shorten the duration of
the lag phase and consequently reduce the time required for the stationary phase.
Compared with 0.05 = , the duration of lag phase at 0.5 = is shortened dramatically
by time step difference 590t = and the time requirement is shortened by 930t = . A
higher value of results in higher cell density at the stationary phase. The simulation
exhibits that osteoblast proliferation is promoted by greater cellular sensitivity, which
is observed in the study of cell-substrate adhesion [194]. When [0,1] , curves of cell
probability have a similar pattern. When 1 , unstable solutions appear and an
unpredictable pattern in the section [1.5,2] is shown in Figure 6.7. Cell probability
distributes symmetrically on both the positive and negative sides. This result is not
realistic for cell distribution since the value of cell probability cannot be negative.
Nevertheless, it may work for the binding of proteins if the positive values express the
unbinding of proteins while the negative values indicate the binding of proteins. In
other words, this model could be extended to study objectives at different size level by
changing the constants.
The size factor n affects the path-dependence integration and thereby justifying the
dynamic model behaviour (Figure 6.8). To illustrate the effect of the increase in size
factor, simulations of size 1000n = and 10000n = performance to compare. The results
show that the path of exponential growth is modified with the alteration for size factor.
When the size factor increases, time steps for reaching steady state increase as well as
the cell density. The higher accumulation of cell density at steady state can be
explained from the definition of size factor. A more substantial size factor indicates
that more cells existing at the target area and thereby results in higher cell density
while the size factor is highly dependent on the size of the target cell.
106
(A)
(B)
1i− i 1i +
1 1( ) ( )i i i ip t p t+ −
− − −1 1( ) ( )i i i ip t p t+ −
+ + −
( ) /ip t t
(C)
107
Figure 6.1 Sketch of the double-layered model for osteoblast adhesion on the material
surface. (A) The graph represents two osteoblast cells (OBs) on an undefined material
surface. The biological processes are assigned to the model. It is noted that osteoblast
cells are in monolayer in vitro after settlement. Small circles (pink) represent
molecules involved in cell signalling and ligand binding. The material surface
properties affect the molecular motions at the molecular layer and further affect
cellular motions at the cellular layer. Osteoblast adhesion works as the transmission
during this progress, deciding how strongly the double-layered system is disturbed.
(B) Mathematical description of cellular probability density ( )ip t during t . It
indicates the difference between cells that stay at the transitional site from 1i − to i
and cells that stay at the transitional site from i to 1i+ . This mechanism achieves a
continuous-time discrete-space random walk at i for equation. (C) Discrete
distribution of cellular probability density on the interval [0, ]r . This interval is
separated into n subintervals of length rhn
= , with endpoints
, 0,1,...,ix ih i n= = . The probability density at a point ix is denoted by ip .
108
(A)
(B)
Figure 6.2 Two examples of model fitting by experimental data. (A) Model fitting of
experimental data for Alumina on 77 nm grain size. (B) Model fitting of experimental
data for Titania on 32 nm grain size. The time scale used 400 time-step to match 4
hours of experimental duration.
109
(A)
(B)
110
Figure 6.3 Function determination for molecular and cellular sensitivity in grain size.
(A) Cellular sensitivity to Alumina and Titania; (B) molecular sensitivity to Alumina
and Titania. The functions on Alumina were calculated for sensitivity (A) and
molecular sensitivity. (C) The functions on Titania were calculated for cellular
sensitivity (B) and molecular sensitivity (D). It was noted that these functions had
similar trends but very different value scales because each point came from
independent experimental data. The functions linked to every independent experiment
together by defining the molecular and cellular sensitivity. Hence the errors between
the experimental data and the model result rooted from these functions.
111
(A)
(B)
Figure 6.4 Numerical models for Alumina and Titania. (A) Profile of adherent cell
density on Alumina with the scale of time and nanophase grain sizes. (B) Profile of
adherent cell density on Titania with the scale of time and nanophase grain sizes. The
colour bar indicates the number of cell density (cells/cm2).
112
(A)
(B)
Figure 6.5 Comparison of numerical results and experimental data. (A) Critical points
for Alumina are marked at grain sizes 36 nm and 60 nm. (B) The critical point for
Titania is marked at grain size 10 nm and 5 nm. The enlarged part shows the decline
of adherent cell density on Titania below grain size 10nm.
113
(A)
(B)
114
Figure 6.6 (A) Simulation of molecular sensitivity. For 0.5 = (red line), the cell
probability density N moves through the lag phase 1 1.12N = at a time step 1t = ,
the exponential phase at 145t = and the stationary phase at 245t = ,245 1.61N = . For
1 = (blue line), the lag phase begins at 1t = ,1 0.99N = , the exponential phase
begins at 305t = , and the stationary phase is reached at 509t = ,509 1.60N = . For
10 = (purple line), the lag phase begins at 1t = ,1 0.39N = , the exponential phase
begins at 755t = , and the stationary phase is reached at 1031t = ,1031 1.59N = . (B)
Oscillation occurs at 0.016=b for 0.5 = , while 0.011=b for 1 = and 10 = .
Calculations are performed using the following set of parameters: 0.1 = , 100n = .
115
(A)
(B)
116
Figure 6.7 (A) Simulation of cellular sensitivity. For 0.05 = (red line), the cell
probability density starts at the lag phase 1 0.99N = , the exponential phase at 645t =
and reaches the stationary phase at 1049t = ,975 1.58N = . For 0.1 = (blue line),
dynamic behaviours repeat the blue line in Figure 6.6. For 0.5 = (purple line), the
lag phase begins at 1t = ,1 0.99N = , exponential phase begins at 55t = , and the
stationary phase is reached at 119t = ,119 1.65N = . Note that the oscillatory change
0.03=b for 0.05 = , 0.011=b for 0.1 = and 0.008=b for 0.5 = . (B) When 1.5 = ,
the simulation appears random walk. Calculations are performed using the following
set of parameters: 1 = , 100n = .
117
(A)
(B)
Figure 6.8 The impact of size factor. The comparison is made between the size 1000n =
and 10000n = . The results show that the path of exponential growth is modified with
the alteration for size factor. When the size factor increases, reaching steady state
needs more time steps, at the same time, cell density can be matched directly since the
value of y-axis is dramatically amplified. A high size factor indicates that substantial
cells existing at the target area and thereby results in a high cell density.
118
Chapter 7
Weak SMF Effects on Osteoblastic Saos-2
Cell Proliferation
119
7.1 Introduction
According to the research findings in Chapter 1 and the hypothesis in Chapter 3, the
“controversial results” for the EMF effect on osteoblasts may originate from the
alteration of ambient factors. Here we attempted to restrain the parametric study of the
EMF effect on osteoblasts. An in vitro experiment was designed for observing
osteoblastic Saos-2 cell proliferation in two gradients SMFs. In this chapter,
mathematical modelling was highly involved in the investigation of SMF effect on the
proliferation of osteoblasts. Therefore, we must guarantee the statistical accuracy and
reproducibility in each experiment. Compared with osteoblasts isolated from human
bone, osteoblastic Saos-2 cell line provided more stable statistical results for
quantifying cell proliferation.
We examined whether the osteoblastic cell proliferation was sensitive to the SMF
intensity on the order of milli-tesla (mT). The experimental data were analysed by the
numerical model developed in Chapter 6, exploring the relevance between the SMF
intensity and osteoblastic cell proliferation. The gradient field of SMF can generate six
different magnetic flux densities at the same time, which provide us with an
opportunity for observing the interference of cell probability.
7.2 Experimental design
This study used a solenoid coil to generate SMFs. The coil (125 mm in diameter, 200
turns of copper wire with a width of 33 mm) was mounted horizontally on a shelf
inside the incubator. A function generator drove the coil with a signal generator (3B
Scientific, Germany). The cells were placed in 6-well plastic culture plates mounted
on a platform across the centre of the solenoid. The culture plates with 85 mm x125
mm surface area were used throughout the experiment. The test area was 85 mm x 250
mm where two 6-well plates were placed horizontally. A magnetic field sensor (3B
120
Scientific, Germany) was used to measure the magnetic intensity at the centre of each
well (Figure 7.1 A).
In this study, multiple magnetic flux densities were generated at an identical
circumstance to reducing the statistical deviation from performing the experiments
separately. The device in Figure 7.1 was calibrated for generating weak SMF (~1 mT).
Limited by the accuracy of the device (~0.05 mT) and background noise (~0.01 mT),
the minimum difference of two magnetic flux densities was set > 0.06 mT.
Consequently, weak SMFs were established with proximal magnetic flux density at 2
and 2.5 mT, and with two gradient fields spontaneously. The first gradient field G1
was measured at the centre of the culture well in 2, 1.1, 0.6, 0.48, 0.37 and 0.3 mT.
The second gradient field G2 was measured at the centre of the culture well in 2.5, 1.6,
1, 0.75, 0.55, and 0.42 mT (Figure 7.1 B). Both fields satisfied the minimum
difference of magnetic intensity.
The control was placed in a different incubator. The background was placed in the
same incubator but different layer which was perpendicular to the coil at
approximately 50 cm. The exposure was applied for 2 hours at any 24-hour interval,
envisaging a possible clinical setting application. 6000 cells per square centimetre
were plated into a 6-well and cultured for two and four days, in the absence or presence
of SMFs.
7.3 Experimental results
7.3.1 Effect of the SMF on the orientation of cultured Saos-2 cells
The Saos-2 cells appeared fibroblasts-like cells in morphology after three days of
seeding. After SMF exposure for 4 hours, part of the cells formed a whirlpool-like
cluster, in contrast to the control which did not have such formation of the cluster
(Figure 7.2). The formed cluster indicated the alteration of cell distribution in SMF.
121
7.3.2 Effect of SMF intensity on the proliferation of cultured Saos-2 cells
Two test groups for SMF exposure, G1 and G2, were investigated on the Saos-2
proliferation (Figure 7.3). G1 was exposed in the gradient magnetic field from 2 mT,
and G2 was exposed in the field from 2.5 mT. Two groups were set in the same
condition except for the SMF intensity. Compared with the control, cell density in G1
only showed a significant increase after SMF exposure of 0.6 mT for 4 hours while
cell density in G2 was not changed significantly. The background of 0.01 mT in G1
and G2 was indifferent than control.
7.3.3 Effect of SMF intensity on the cell viability of cultured Saos-2 cells
The SMF exposure resulted in the lower cell viability compared with the background
and the control (Figure 7.4) The effect was pronounced in the early stage.
7.3.4 The sensitivity of cultured Saos-2 cell proliferation to SMF intensity
The sensitivity of cell proliferation to SMF intensity was focused on the difference
between adjacent data. The Saos-2 cell proliferation was not sensitive to the SMF
intensity of background in comparison with the control (Figure 7.3). The statistical
analysis in G1 and G2 illustrated that the Saos-2 cell proliferation was more sensitive
to SMF intensities after accumulated exposure for 4 hours than that after the exposure
for 8 hours. In G1, the difference of cell density was observed among adjacent SMF
intensities after 4 hours (Figure 7.3 A) but the difference was reduced to three intervals
with the apparent distinction (P<0.01) between the SMF exposure of 0.6 and 1.1 mT
(Figure 7.3 B). The similar phenomenon is also found in G2 that the sensitive intervals
were reduced from three to two (Figure 7.3 C, D). After the accumulated exposure of
8 hours, the distinction was found between the exposure of 0.75 and 1 mT in G2
(Figure 7.3 D). If combining G1 and G2 into one picture, the picture of G1+G2
showed sensitive to the SMF intensities after the accumulated exposure of 4 hours.
122
The distinction only appeared between 0.55 and 0.6 mT after 8 hours. It was
recognised for the feasibility of combined G1+G2. The distinction between the results
after the accumulated exposure of 4 hours and 8 hours implies that the SMF intensity
has a stronger influence on the Saos-2 proliferation at an early stage.
7.4 A numerical model of the dose-dependent effect of SMF on osteoblastic
cell proliferation
7.4.1 A stochastic model for cell probability in SMF
The first assumption is that a gradient SMF can affect the cell probability of
osteoblasts. Critical parametric functions are expected to be extracted by fitting cell
data in vitro experiment with a theoretical cell probability in silico. These parametric
functions facilitated the understanding of how osteoblastic cell perceived SMF. Our
model comprised of the mathematical analysis of random walks [188] and
Chemotactic movement [195], describing the spatial-temporal evolution of cell
probability density function under the influence of random forces as described in
Chapter 5 [196]. The fitting created an S-shaped curve for cell proliferation. In this
study, SMF can affect osteoblastic cell proliferation through the alteration of cell
formation. We write the Fokker-Planck equation of this stochastic process in term of
probability density function p and magnetic flux B ,
( , ) [ ( ) ( , ) ( ) ( , )]B Bp x t D p x t p x tt x x
= −
, (7.4.1)
where the evolution of the probability density function is affected by diffusion D and
drift . To simulate the dynamics of cell probability in the equation (7.4.1), we
suppose that the conditional probability density ip is implemented via an approximate
discretisation of the culture area ( i N ),
123
1 1 1 1( ) ( )i i i i i i i i ip W p W p W p W p+ − + −
− − + + = − − − (7.4.2)
where
1 1
1(1 )(1 )i i i iW p p
h
= − − (7.4.3)
Where the diffusion D is incorporated into the conditional probability density ip ,
and the drift indicates the correlation for neighbour coupling. Therefore, we
establish a three-term recurrence relation in a culture area where its radius r is divided
into points 0 ... i with a step h . Each point i corresponds to a conditional cell
probability density ip . After this step, the algorithm guarantees the nearest neighbour
coupling, and the whole system of cell probability density can be arranged in a finite
chain. In an ideal circumstance, we can fit any cell population N with the initial cell
population 0N by integrating the cell probability density p with a spatial component
i and temporal component j ,
0 ( )ji
j i
N N f p= (7.4.4)
where N represented the observed cell population and 0N represented the initial cell
population, for example, cell number cultured at day 0. The equation (7.4.4) was
calculated by a circular integration. Hence, the cell population tN matched the cell
number in a culture well after culturing certain days. When designing the experiment
for this study, we selected three temporal points to check the cell number at day 0, 2
(4 hours) and 4 (8 hours)respectively. The algorithm was solved by MATLAB® with
three predicted function. These functions were interpolated into the equation (7.4.4),
124
0 3 2 1( ) ( ) ( ( ))B B ji B
j i
N N f f p f= . (7.4.5)
The exact formation of these functions was related to the experimental data.
7.4.2 Numerical results
The numerical results of the stochastic model for G1 and G2 represented that Saos-2
cell proliferation was not linear with SMF intensities, but an approximation of the sine
curve in term of SMF intensities (Figure 7.7). After 8-hour SMF exposure in four
days, a period was found by the peak-to-peak amplitude as 1 mT in G1 and 0.8 mT in
G2. The system was meta-stable when examining the results after 4-hour SMF
exposure in two days. The oscillation was evident in the intensities from 2 to 3 mT in
G1, and from 0 to 0.5 mT in G2. The oscillation was reduced by combining G1 and
G2 into G1+G2. However, the oscillation of G1+G2 remained the same as that of the
G1 from 2 to 3 mT. The oscillation is an essential part of the stochastic model referring
to the interference of cell probability. A sine function is written based on the
relationship between osteoblastic cell proliferation and SMF intensity,
( ) ( )sin( )cN t N A t kB= + , (7.4.6)
where ( )N t is the cell population at the time t , and cN is a constant that could be a
population of control. The amplitude ( )A t of the sine function changes with time and
might satisfy a wave function. Its property illustrates a harmonic wave function in
which each point on the axis of the wave has a constant amplitude at a defined time.
The wave oscillates in time but has a spatial dependence that is stationary. Substitute
time t with a spatial component i and temporal component j ,
( ) sin( )c ji i
j i
N t N a p kB= + . (7.4.7)
125
We can write a square matrix to satisfy the equation (5.2.17) by
11 1 1
1 1
sin( B ) sin( B )
=
sin( B ) sin( B )
n n
n nn n
p k p k
p k p k
p . (7.4.8)
7.4.3 SMF Interference
According to the equation (7.4.5), the description of the interference is related to the
parametric functions f . Three coefficients are analysed for the linearity (Figure 7.8
A). The coefficient 1 represents the spatial modification on cell probability density,
and the coefficient 2 represents the temporal modification. The coefficient 3
contributes to the interference if applicable. The linearity reduces from coefficient 1
to coefficient 3. The correlation between coefficient 1 and the SMF intensity is linear,
and the coefficient 2 is approximately linear with the SMF intensity. The coefficient 3
is nonlinear with the SMF intensity. The results indicate that the interference occurs
without the dependence on the cell probability density,
1: 1.25sin[ ( 0.47)],0.49
2 : 0.68sin[ ( 0.36)],0.41
1 2 : 0.79sin[ ( 0.47)].0.51
B
B
B
G f
G f
G G f
= −
= +
+ = −
(7.4.9)
The interference results in various cell probability distribution along the cell plate
(Figure 7.8 B). The experimental results only showed several possibilities. The
possible variety of the cell population is enormous.
7.5 Discussion
Data presented in this chapter demonstrate that the osteoblastic Saos-2 cell
proliferation is sensitive to the weak-intensity SMF. The numerical results of the dose-
126
dependent effect of SMF on osteoblastic cell proliferation satisfy a wave equation.
This wave equation changes with the SMF circumstance. The experiment of 2.5 mT
has an enhanced frequency and reduced amplitude compared with the experiment of 2
mT. Hence the difference of the SMFs intensities and the arrangement of the
experiment can both modify osteoblastic cell proliferation. The wave equation is also
the representation of the interference, which could be the reason for the observation of
various experimental results at the same SMF intensity.
7.5.1 Signal-to-noise
Although weak-intensity SMFs may stimulate the unique responsiveness of organisms
in vitro, experimental data for the biological effects of weak-intensity SMFs on cell
cultures are limited compared with the effects of moderate-intensity SMFs [7].
Notably, the weak-intensity SMFs rarely expose on osteoblastic cells since the
underlying signal-to-noise problem between identification of weak-intensity SMFs
bio-effects and standard population deviation of cell proliferation [197]. However, if
the signal-to-noise problem can be solved, the study of weak-intensity SMFs on
osteoblastic cells will serve two purposes. The first one is to understand the dose
pattern of osteoblastic growth in SMFs. Compared with moderate-intensity SMFs,
weak SMFs have fewer combinations of physical parameters and more consistent
biological reactions at the cellular level, just because of the narrow range of SMFs
intensities. The second purpose is to verify the role of weak-intensity SMFs in clinical
applications by answering whether the weak-intensity SMFs can enhance cell
proliferation. We realise that the signal-to-noise problem in weak-intensity SMFs bio-
effect is part of the reasons for the controversial results of the SMFs influence on
osteoblastic cells mentioned above, which is highly conditional on their physical
mechanisms. Therefore, we propose a computational model to simulate the physical
127
process of osteoblastic cell in weak SMFs as the framework for experimental design.
Different from previous models focusing on Ca2+ and ion channels [198], our model
emphasises the cellular movement and general interaction between cells and
molecules. Inspired by the window effect, we made the underlying assumption of this
article that any magnetic signal can be understood by cells through finite physiological
processes. This assumption led to two derivations for designing in vitro and in silico
model. (i) Optimal cell response can be observed in any range of magnetic flux density.
(ii) Cell data in SMF can be understood by parameters of physiological processes and
magnetic flux densities. The first assumption can be partially proved by the evidence
that the Ca2+ and ion transport system can be affected by weak-intensity SMFs (< 1mT)
[199]. The second assumption is made to guarantee that the biological system can
reach a steady state after a finite time.
7.5.2 Physical mechanisms
If the evidence mentioned above are considered reasonable, then the osteoblastic cell
must perceive SMF by more than one local mechanism. There were three types of
physical mechanisms of SMF interacting with biological processes, including
electrodynamic interactions, magneto-mechanical interactions and effects on
electronic spin states [200]. Insight into these mechanisms was provided by
examination of the interaction of SMF with moving charges, enzyme reaction rates
and signal-transduction pathways involved in response to SMF [1]. Although these
mechanisms might target different subtle cell functions, their influence (individually
or collectively) was suspected of being observed from cell distribution and
morphology at the end of cell development in vitro. This provides a chance to answer
the question of how osteoblastic cell perceives SMF in vitro by a quantitative
description of osteoblastic cell distribution and magnetic flux intensity. A quantitative
128
description was conceptually built by a hypothesis “window effect”, which indicated
that specific magnetic flux intensities could make biological systems achieve optimal
response [57]. An analogy of the window hypothesis is a quantitative resonant
interaction of magnetic field with ions in biological systems [201]. Different from the
resonant interaction, the window effect implied the optimal response of a biological
system caused not only by the resonance of ions but also by other physiological
processes, and thereby allows multiple optimal responses to a range of magnetic flux
densities. The existence of probability cell density causes the uncertain of position and
velocity. Therefore, we may infer that a strong SMF would generate a fluctuation with
high frequency and small amplitude in cell growth which means that a strong SMF
would not affect cell proliferation but reduce the deviation of cell number between the
exposed group and control group. A study investigated the effects of a strong SMF (8
T) exposure on cultured mouse osteoblastic MC3T3-E1 cells and found that cells were
orientated in the direction parallel to the magnetic field. The interesting fact was that
this strong SMF exposure did not affect cell proliferation regardless of 24-hour or 60-
hour exposure. Also, there was no significant difference between the growth curve for
the exposed and non-exposed groups in the following three days and six days [23].
The underlying mechanisms for the existence of fluctuations in reconstructed dose-
dependent patterns may root in the physical mechanisms of SMFs interacting with the
biological process of bone cells. We examined two primary proposed physical
mechanisms, electrodynamic interactions and magneto-mechanical interactions [202].
The electrodynamic interactions may result in Hall Effect which leads to redistribute
surface charges on cell membrane [203] and may affect bone cell activity by altering
streaming potential. It has been shown that streaming potential can affect bone cell
129
activity in vitro, but how altered streaming potential is perceived by bone cells remains
poorly understood [204].
On the other hand, magneto-mechanical effects are related to the uniformity of SMFs
and inherent magnetic properties of materials. In an inhomogeneous SMF, bone cells
may experience a force (Lorentz force perhaps) and orientate along the direction of the
field gradient. Mouse osteoblastic MC3T3-E1 cells were orientated in the direction
parallel to a strong SMF (8 T) after 60 hours exposure in vitro as well as the orientation
of bone formation in and around the BMP-2/collagen pellets implanted subcutaneously
in vivo [23]. Interestingly, the migration of osteoblastic cells under strong SMF follow
the direction of magnetic flux rather than encounter a Lorentz force in the case that the
motion direction is not parallel to the applied magnetic field.
The diffusion function in osteoblastic cells after the treatment of SMFs indicates that
SMFs can reduce the diffusion activity with increasing magnetic flux. This conclusion
explains the phenomenon that we observed in the experiment. Naturally, the magneto-
mechanical effects are related to the uniformity of SMFs and inherent magnetic
properties of materials. In inhomogeneous SMFs, materials experience a force and
orientate along the direction of the field gradient. If the magnetic gradient is large
enough, diamagnetic levitation can be achieved. In a uniform magnetic field, there is
no force, but magnetic torque, which makes materials with anisotropy of magnetic
susceptibility tend to rotate until they reach a stable orientation. It has been approved
that osteoblasts can be oriented along the magnetic direction under high SMFs [23].
Another possibility is that SMFs can influence the rates of specific chemical reactions
in chemistry and biology because on radical ion pair of the reaction intermediates. The
radical ion pair can be altered by SMFs of modest and weak intensity [205, 206].
130
(A)
(B)
Figure 7.1 Representation of the device used to generate the SMFs. (A) The device
consisted of a signal generator and platform. The platform included a solenoid and two
six-well plates in line. (B) The strength of SMF was measured by a tesla-meter in the
centre of each well, labelled by A1 to A6. Two SMF groups G1 and G2 were treated
on osteoblasts. Two replicates were tested in one experiment, each data point with n=4.
The platform was placed in an incubator of 5% CO2, 37 0C.
131
(A)
(B)
Figure 7.2 Cell morphology of osteoblastic Saos-2 in SMF and control. After SMF
exposure of 0.6 mT, part of the cells formed a whirlpool-like cluster, in contrast to the
control which did not have such formation of the cluster.
132
(A)
(B)
133
(C)
(D)
134
Figure 7.3 Effect of the SMFs on the proliferation of osteoblastic Saos-2. Cells of
G1(A, B) and G2(C, D) were harvested after SMF exposure for accumulated 4 hours
and 8 hours respectively. The number of cells per well in 6-well plate was counted.
Two-tailed t-tests are performed on each intensity versus control and means of them
are not significantly different. Statistical significance is indicated among different
intensities. Data are expressed as the mean (symbols) ± SEMs (error bars), *P<0.05
**P<0.01 ***P<0.001.
135
Figure 7.4 The effect of SMF exposure on cell viability of osteoblastic Saos-2. The
SMF exposure resulted in the lower cell viability compared with the background and
the control. The effect was pronounced in the early stage. Data are expressed as the
mean (symbols) ± SEMs (error bars), *P<0.05 **P<0.01 ***P<0.001.
(A)
(B)
136
(A)
(B)
Figure 7.5 Effect of the SMFs on the cell proliferation of osteoblastic Saos-2 in
combined G1+G2. The picture of G1+G2 was sensitive to the SMF intensities after
the accumulated exposure of 4 hours. The obvious distinction was only shown between
0.55 and 0.6 mT after 8 hours. It was recognised for the feasibility of combined
G1+G2. Data are expressed as the mean (symbols) ± SEMs (error bars), *P<0.05
**P<0.01 ***P<0.001.
137
138
Figure 7.6 The sketch of the mathematical model for the movement of osteoblastic
cells in vitro under SMFs. (A) The movement of the osteoblastic cell is separated into
cellular movement and molecular movement under SMFs for modelling. (B) A logic
concept is built for initialising the mathematical description of modelling, established
in the relationship that the magnetic density can affect the molecular motions and
subsequently affect cellular motions. Cellular motions may reversely influence the
property of SMFs. (C) A realistic scenario is established for enriching the details of
modelling. A logic description can only provide a null hypothesis to the influence of
SMFs on osteoblastic cells, and we further set a detailed scenario to calculate the
cellular probability density in a culture well, for instance, one well in a 6-well plate.
(D) The algorithm of cellular probability density is written in both spatial and temporal
scale. (E) The mathematical model can be solved by fitting the experimental data into
the algorithm.
139
(A)
(B)
140
(C)
(D)
Figure 7.7 The predicted pattern of osteoblastic Saos-2 cell proliferation in weak
SMFs. The prediction is consistent with the experimental data. The oscillation was
evident in the intensities from 2 to 3 mT in G1 (A), and from 0 to 0.5 mT in G2 (B).
The oscillation was reduced by combining G1 and G2 into G1+G2 (C). However, the
oscillation of G1+G2 remained the same as the G1 from 2 to 3 mT. The symmetric
interval was observed with the central of 0.7 mT (D).
141
(A)
(B)
142
Figure 7.8 Parametric analysis. (A) Three coefficients are analysed for the linearity.
The coefficient 1 represents the spatial modification on cell probability density, and
the coefficient 2 represents the temporal modification. The coefficient 3 contributes to
the interference if applicable. The linearity reduces from coefficient 1 to coefficient 3.
The correlation between coefficient 1 and the SMF intensity is linear, and the
coefficient 2 is approximately linear with the SMF intensity. The coefficient 3 is
nonlinear with the SMF intensity which mainly contributes to the interference. (B) The
interference results in various cell probability distribution along the cell plate.
143
Chapter 8
PEMF Effects on Cell Proliferation of
Human Osteoblasts
144
8.1 Introduction
Despite a long history of clinical use of PEMFs and some mechanistic studies, the
optimal parameters for usage remain poorly understood. PEMFs can modulate the
proliferation and differentiation of osteoblasts and alter the morphology and
intracellular calcium levels of osteoblasts. Optimal therapeutic effects of PEMFs
require a series of optimisation in the methodology of treatment. PEMFs obtain various
results at different frequencies and intensities, which also diverse on signal waveform,
signal duration, cell type and cell stage. The study in Chapter 7 indicated that weak
SMFs resulted in the interference of cell probability in osteoblastic Saos-2 cells. A
periodic function was written for the relationship between SMF intensities and cell
population at the time. In this chapter, we investigated the effects of PEMF on cell
proliferation of osteoblastic Saos-2 cells, and further on the formation, mineralisation
and gene expression of primary human osteoblasts (passage 2). The interference of
cell probability was found in term of PEMF frequency and intensity.
8.2 Experimental designs
This study used two cell types, osteoblastic Saos-2 cells and primary human
osteoblasts (passage 2). Saos-2 cells were tested for the effects of PEMF on cell
proliferation. Primary human osteoblasts (passage 2) were used for the observation of
morphology, mineralisation and gene expression. Two types of apparatus were set to
generate PEMFs, a solenoid and Helmholtz coils. The coils were mounted horizontally
on a shelf inside the incubator and driven by a signal generator with alternative current
(AC) (3B Scientific, Germany). The signal function was sinusoidal. The cells were
placed in 6-well plastic culture plates mounted on a platform across the centre of the
coils. The culture plates of 85 mm125 mm were used throughout the experiment. A
self-made magnetic field sensor with an oscilloscope was used to measure the
145
magnetic intensity at the centre of each well (Figure 8.1 A). The effects of PEMF
intensity and frequency were analysed on cultured Saos-2 cells. The exposure method
utilised the decline of magnetic density at the proximal to the solenoid. Six magnetic
flux densities were investigated at the centre of each well: 0 (control), 0.1, 0.15, 0.25,
0.53 and 0.69 mT. The range of frequency was investigated in 40, 80, 160, 320, 640
and 800 Hz (Figure 8.1 B). The exposure was turned on for 2 hours at any 24-hour
interval, envisaging a possible clinical setting application. The control was set in a
separate incubator. 6000 cells per square centimetre were plated into a 6-well and
cultured for five days, in the absence or presence of a PEMF. Cells were settled in day
0, followed by PEMF exposure in day 1-4 and harvested on day 5.
8.3 Experimental results
8.3.1 Effect of PEMF intensity on the proliferation of cultured Saos-2 cells
Compared to the control group, PEMF exposure significantly promoted the cell
proliferation in the combination treatment of 0.15 mT with 40 and 800 Hz, 0.25 mT
with 320 and 800 Hz, and 0.69 mT with 160 Hz (Figure 8.2). No significant difference
was found between the control and PEMF exposure with 80 and 640 Hz. At the same
frequency, the perturbation of the cell population was observed under different
intensities of PEMF which was like the effect of SMF exposure.
8.3.2 Effect of PEMF frequency on the proliferation of cultured Saos-2 cells
At the same intensity of PEMF, the various frequency induced fluctuations around the
control (Figure 8.3). Compared with the range of control, 80, 160 and 640 Hz showed
adverse effects on cell population at 0.1 mT. 40, 320 and 800 Hz indicated positive
impacts on cell population at 0.15 mT. At 0.25 mT, 40 and 800 Hz showed positive
impacts on cell population while 160 Hz had a negative impact. 160 and 640 Hz
146
indicated adverse effects on cell population at 0.53 mT. At 0.69 mT, 80 Hz presented
negative impact on cell population while 160 Hz presented positive effect.
8.3.3 The sensitivity of cultured Saos-2 cell proliferation to PEMF
The sensitivity of Saos-2 cell proliferation to PEMF was illustrated by Figure 8.4 in
the format that the variety of PEMF intensities were correlated with cell proliferation
along the PEMF frequencies. The data on cell proliferation in various PEMF intensities
at the same frequency were collected at the same time. The sensitivity of cell
proliferation to PEMF was evaluated by the continuity of plotted data at each
frequency.
The Saos-2 cell proliferation was not sensitive to PEMF at 40, 80 and 320 Hz.
Compared with the control, the cell proliferation was sensitive to the points of 0.69
mT combined with 160 and 640 Hz, and 0.15 mT combined with 640 Hz. The cell
proliferation was also sensitive to the intervals of 0.25 ~ 0.53 mT at 160 and 640 Hz,
and 0.15 ~ 0.25 mT at 800 Hz. The combination of 0.69 mT and 160 Hz might be
unique since it was a discrete point. The significant perturbation of cell proliferation
was observed at 640 Hz, while the consistency of cell proliferation was observed at
320 Hz. It indicated that the Saos-2 cell proliferation might react to specific PEMF
frequencies by strengthening or weakening the perturbation.
A comparison was made between the sensitivity of cultured Saos-2 cell proliferation
to PEMF generated by a solenoid and that generated by Helmholtz coils (Figure 8.5).
The solenoid created a field gradience of 0.25 mT when the central magnetic flux
density was 0.69 mT. The Helmholtz coils created a consistent field of 0.69 mT. Two
types of PEMFs were carried out in the same protocol and platform. The results
showed a good continuity in data of cell proliferation at 40, 80, 320 and 800 Hz. The
solenoid indicated a better promotion of cell proliferation at 160 and 640 Hz. Weak
147
PEMFs from solenoid and Helmholtz coils presented a similar effect on Saos-2 cell
proliferation. The solenoid could strengthen the perturbation of cell proliferation in
comparison with Helmholtz coils.
8.3.4 Effect of PEMF on the orientation and mineralisation of human osteoblasts
The human osteoblasts presented the rod-like shape in morphology (Figure 8.6).
PEMF exposure of 0.69 mT with 160 Hz for 8 hours induced a variety of the
orientation of osteoblasts. At the same scale, the control illustrated consistent
orientation. After 12-day culture, cells stained by Alizarin red-aligned along a
whirlpool-like circle in PEMF, while the control did not form this unique shape
(Figure 8.7). Interestingly, the SMF exposure also showed a similar orientation pattern
with the group in PEMF. Effect of PEMF and SMF on osteogenesis of human
osteoblasts were assessed by Alizarin red-stained calcified nodules formed. The red
nodule was observed in PEMF exposure of 0.69 mT with 160 Hz, while the control
and the SMF exposure showed the insignificant difference.
8.3.5 Real-time RT-PCR of osteogenesis-related genes
The effects of PEMF and SMF exposure on osteogenesis-related genes of human
osteoblasts were assessed by quantitative reverse transcription polymerase chain
reaction (RT-PCR) with RNA isolation from osteoblasts after 8 hours treatment of
SMF (Table 8-1) and PEMF (Table 8-2). Compared with the control, both PEMF and
SMF exposure at 0.69 mT significantly increased IGF-1 and decreased BMP-1
expression after 8 hours of treatment. However, PEMF significantly increased PHEX
while SMF decreased PHEX expression in comparison with the control. The protein
level of IGF-1 indicated a slight increase by SMF and PEMF compared with the
control, which is not compatible with the significance of RT-PCR result.
148
8.4 Numerical solution of PEMF effects on osteoblastic cell proliferation
8.4.1 Model derivation for PEMF
The parametric functions were extracted by fitting experimental data in both scenarios
of intensity and frequency, which indicated how osteoblastic cell perceived PEMF.
The derivation described the spatial-temporal evolution of cell probability distribution
under the PEMF effect. The fitting created an S-shaped curve for cell proliferation.
PEMF affected osteoblastic cell proliferation by the alteration of cell distribution, and
a Fokker-Planck equation of this stochastic process was written in term of the
probability density function p , magnetic flux B and frequency v ,
( , ) [ ( , ) ( , ) ( , ) ( , )]B Bp x t D p x t p x tt x x
= −
, (8.4.1)
where the evolution of the probability density function was affected by diffusion and
drift. The equation (8.4.1) was incorporated with transitions to simulate the effect of
PEMF on the dynamics of cell distribution. Suppose that the conditional probability
density ip ( i N ) is implemented via an approximate discretisation of the culture
area, and attributed to the magnetic intensity B
ip and frequency v
ip respectively,
1 1 1 1( ) ( )B B B B B
i i i i i i i i ip W p W p W p W p+ − + −
− − + + = − − − , (8.4.2)
and
1 1 1 1( ) ( )i i i i i i i i ip W p W p W p W p + − + −
− − + + = − − − , (8.4.3)
where the diffusion is incorporated into the conditional probability density, and the
drift shows the correlation for neighbour coupling as in an equation (7.4.3). Therefore,
we establish a three-term recurrence relation in a culture area where its radius is
divided into points 0 ... i with a step h . Each point i corresponds to a conditional cell
probability density. After this step, the algorithm guarantees the nearest neighbour
149
coupling, and the whole system of cell probability density can be arranged in a finite
chain. Any cell population N is built with the initial cell population 0N by
integrating the cell probability density p with a spatial component i and temporal
component j ,
0 3 2 1( , ) ( , ) ( ( , ))B B ji B
j i
N N f f p f = , (8.4.4)
where N represented the observed cell population and 0N represented the initial cell
population. Using the matrix for the expression of functions, column entries show
inputs to a process sector while row entries represent outputs from a given sector as
the equation (5.2.17),
1
1
( ) ,
( ) ,
c
c
t t
t t
−
−
=
=
0
0
N I - C N
N I - A DN.
This format shows how dependent each sector is on every other sector. The coefficient
matrix 1( )−I -C and 1( )−I - A are redefined by n square matrix. The actual construction
of the coefficient matrix may require addressing the affected processes in PEMF.
8.4.2 PEMF Interference
The numerical results showed that Saos-2 cell proliferation was not dose-dependent
on PEMF frequency but fitted a sine function in term of PEMF frequency (Figure 8.9).
After 8-hour PEMF exposure, the fitting on PEMF frequency was created in five
equations (Table 8-3). The fitting was stable by comparing the results from PEMF
exposure of different intensities. The data were combined into one picture,
[0.37cos( ) 0.15sin( )] [0.18cos( ) 0.45sin( )]141.3 141.3 0.13 0.13
B BN
= − − + .
(8.4.5)
150
The equation articulated the information derived from five experimental arrangements
in complementarity as the hypothesis in Chapter 3. The equation (8.4.5) established a
pattern of interference which met the whole data set with an adjusted R-square of 0.5
(Figure 8.10 A). The interference inherited the similar properties of the mathematical
description in Chapter 7 (Figure 8.10 B). The metastability was found in the cell
probability in PEMF.
Set
1 2, 141.3 0.13
B
= = , (8.4.6)
the equation (8.4.5) had a general form as
2 1 2 1 2 11 22 sin( )[sin( ) cos( )]
2 2 2N C C
+ − −= − + , (8.4.7)
where C1 and C2 referred to Coefficient 1 and 2. The difference between each dataset
and the standard equation (8.4.5) was expressed in implicit functions (Figure 8.11).
Coefficient 1 indicated numerical distinction near the combination of 0.2 mT and 800
Hz. Coefficient 2 showed a numerical distinction near the combination of 0.3 mT and
300 Hz. The incorporation of two implicit functions into the standard equation built a
general model for analysing the PEMF effect on osteoblastic cell proliferation.
8.5 Discussion
The PEMF interference (8.4.5) represented a consistent description of cell probability
with SMF interference (7.4.9). In comparison with the SMF interference, the PEMF
interference employed two more components for the frequency influence and thereby
using more dataset to create a coefficient matrix. Twelve datasets were used in SMF
while thirty datasets were used in PEMF. Interestingly, the number of datasets was
proportional to the number of gene expression alteration in Table 8-1 and Table 8-2.
151
Interaction with signalling pathways is a potential mechanism by which deficient
energy PEMF might produce metabolic responses. Hormone and neurotransmitter
receptors are specialised protein molecules involved in a variety of biochemical
processes to passing chemical signals from the outside of a cell across the plasma
membrane to the interior. Since weak PEMFs have too little energy to traverse the
membrane directly, it is possible that they may modify the existing signal transduction
process in cell membranes, providing both transduction and biochemical amplification
[207].
We postulate that the observed effects of PEMF on bone cell metabolism correlate to
PEMF coupling with signal transduction which contributes to the hormonal regulation
of osteoblast proliferation and differentiation. Compared with the SMF, the PEMF
coupling covered a wide range of signal pathways in Table 8-2. For example, PHEX
was upregulated after the PEMF exposure, which was reported to enhance osteogenic
differentiation, extracellular matrix deposition and mineralisation [208]. We also
observed the enhanced mineralisation in the PEMF group, yet the insignificant
difference between the SMF group and the control (Figure 8.7). Hence, the PEMF
coupling exhibited stronger influence on calcium-dependent transduction than the
SMF coupling.
The phenomenon of signal transduction is central to a wide range of cellular activities
triggered by ligand-gated binding of hormones, antigen molecules, growth factor and
other cell-surface agonists. Calcium ions appear to be essential in the first steps of
transduction coupling of exogenous physical signals at the cell membrane and the
ensuing steps of calcium-dependent signalling in intracellular enzyme systems. The
modulation of calcium signalling by PEMF is thereby suggested to be a plausible
candidate for the activation of biochemical reactions. PEMF coupling with cellular
152
targets may occur via highly cooperative steps. For example, calcium-dependent steps
in the target pathway may include, (i) initial detection of PEMF at specific binding
sites with resultant electrochemical changes, (ii) membrane-bound proteins signalling
to the cell interior, and (iii) PEMF coupling with the cytoskeleton and other subcellular
constituents. Consequently, the PEMF coupling builds a hierarchical system from top
to bottom, starting with the calcium signalling. The depth of the hierarchical system
may depend on the magnetic intensity and frequency. At the same level of magnetic
intensity, the PEMF coupling illustrates no difference than the SMF coupling with
signal transductions (Figure 8.8), which implies that frequency may not affect the
activation of biochemical reactions. The frequency of PEMF may provide an
opportunity to couple with a series of pathways such as IGFs (Table 8-2). This
mechanism gives the PEMF an ability to influence the immune system (Figure 2.2).
An impact on the immune system leads to more evident results in vivo than in vitro,
which explains that the PEMF has significant therapeutic results on bone healing for
animals (Table 1-1) rather than the increase on cell proliferation of cultured
osteoblasts.
153
(A)
(B)
Figure 8.1 Representation of the device used to generate the PEMFs. (A) The device
consisted of a signal generator and platform. The platform contained a solenoid or
Helmholtz coils. (B) The PEMF intensity was measured by a tesla-meter in the centre
of each well, labelled by A1 to A6. Two replicates were tested in one experiment and
each data point with n=4. The platform was placed in an incubator of 5% CO2, 37 0C.
154
Figure 8.2 Effect of the PEMF intensity on the proliferation of cultured Saos-2 cells.
In each experiment, osteoblasts were exposed to an EMF with a fixed frequency.
Exposures were applied on osteoblasts for 2 hours per day and continued for four days.
The data are represented by mean ± SEMs (n=4). PEMF groups were compared with
the control, *P<0.05 or ** P<0.01.
155
Figure 8.3 Effect of the PEMF frequency on the proliferation of cultured Saos-2 cells.
Perturbations were induced by different frequencies of PEMF exposures for 8 hours.
The control was shown by an interval of mean ± SEMs (n=24). Six frequencies 40, 80,
160, 320, 640 and 800 Hz were compared at each PEMF intensity respectively. The
curves indicated the frequency-dependent tendency. Results are presented as mean ±
SEMs (n=4).
156
Figure 8.4 Data distribution of PEMF groups along the control. The data on cell
proliferation in various PEMF intensities at the same frequency were collected at the
same time. The sensitivity of cell proliferation to PEMF was evaluated by the
continuity of plotted data at each frequency. The data are represented by mean ±SEMs.
Each point in the EMF group includes n=4, and the interval of control includes n=24.
157
Figure 8.5 A comparison between cultured Saos-2 cell proliferation to PEMF
generated by a solenoid and that generated by Helmholtz coils. The solenoid created a
gradient field with the central magnetic flux density of 0.69 mT. The Helmholtz coils
created a consistent field of 0.69 mT. Two types of PEMFs were carried out in the
same protocol and platform. The results showed a good continuity in data of cell
proliferation at 40, 80, 320 and 800 Hz. The data are represented by mean ± SEMs.
Each point in the EMF group includes n=4, and the interval of control includes n=24.
158
(A)
(B)
Figure 8.6 PEMF effect on the orientation of human osteoblasts. (A) The human
osteoblasts presented the rod-like shape in morphology. (B) PEMF exposure of 0.69
mT with 160 Hz for 8 hours induced a variety of the orientation of osteoblasts.
159
(A) Control (1 mm)
(B) Control (500 µm)
(C) PEMF 0.69 mT,160 Hz (1 mm)
(D) PEMF 0.69 mT,160 Hz (500 µm)
(E) SMF 0.69 mT (1 mm)
(F) SMF 0.69 mT (500 µm)
Figure 8.7 Effect of PEMF and SMF on osteogenesis of human osteoblasts assessed
by Alizarin red-stained calcified nodules. The red nodule was observed in PEMF
exposure of 0.69 mT with 160 Hz, while the control and the SMF exposure showed
the insignificant difference.
160
(A) IGF-1 10 µg
(B) IGF-1 30 µg
Figure 8.8 Western blot analysis of IGF-1 protein at a loading of 10 and 30 µg. The
protein level of IGF-1 indicated a slight increase by SMF and PEMF compared with
the control, which is not compatible with the significance of RT-PCR result.
161
Figure 8.9 Waveform of frequency. The numerical results showed that Saos-2 cell
proliferation was not dose-dependent on PEMF frequency but fitted a sine curve in
term of PEMF frequency. After 8-hour PEMF exposure, the fitting on PEMF
frequency was created in five wave equations. The fitting was stable by comparing the
results from PEMF exposure of different intensities.
162
(A)
(B)
Figure 8.10 The interference of cell probability in PEMF. (A) The graph articulated
the data derived from five experimental arrangements, which established a normalised
pattern of interference (B) on the whole dataset with an adjusted R-square of 0.5. The
interference inherited the mathematical description.
163
(A) Coefficient 1
(B) Coefficient 2
164
Figure 8.11 Implicit functions of coefficient 1 and 2. The metastability was found in
the cell probability in PEMF. The implicit functions of Coefficient 1 and 2 were
expressed in graphs. Coefficient 1 (A) indicated numerical distinction near the
combination of 0.2 mT and 800 Hz. Coefficient 2 (B) showed a numerical distinction
near the combination of 0.3 mT and 300 Hz. The incorporation of two implicit
functions into a normalised pattern created a general model for analysing the PEMF
effect on osteoblastic cell proliferation.
165
Table 8-1 The effects of SMF exposure on osteogenesis-related genes of human
osteoblasts. SMF treated cells with the setting of a magnetic flux density of 0.69 mT
for 8 hours in 5 days. Fold difference is 1.5.
Genes Under-Expressed in
Genes Over-Expressed in
Gene Symbol Fold Gene Symbol Fold
ALPL -2.0 BMP4 1.9
BMP1 -1.8 BMP6 1.6
CSF3 -2.0 IGF1 2.6
CTSK -2.1
MMP9 -2.0
PHEX -1.5
VCAM1 -1.7
Table 8-2 The effects of PEMF exposure on osteogenesis-related genes of human
osteoblasts. PEMF treated cells with the setting of the magnetic flux density of 0.69
mT and a frequency of 160 Hz for 8 hours in 5 days. Fold difference is 1.5.
Genes Under-Expressed in
Genes Over-Expressed in
Gene Symbol Fold Gene Symbol Fold
AHSG -1.7 BMPR1B 1.5
BMP1 -1.6 CSF2 2.7
COL2A1 -2.6 FLT1 2.2
COL5A1 -3.0 IGF1 1.9
DLX5 -1.6 MMP8 1.5
IGF2 -1.5 PHEX 1.5
ITGA3 -1.9 TGFB3 1.6
166
Table 8-3 The frequency functions.
Function: sin( )cc
xN y A f
w w
= + −
yc A xc w Adj.R-square
0.1 mT 2.45 0.43 -88.78 142.27 0.84
0.15 mT 2.90 -0.34 473.23 184.92 0.79
0.25 mT 2.77 0.69 215.49 135.61 0.98
0.53 mT 2.47 0.24 153.83 0.40 0.98
0.69 mT 2.78 0.55 179.55 215.99 0.0035
167
Chapter 9
PEMF and SMF Effects on Co-culture of
Human Osteoblasts and Osteoclasts
168
9.1 Introduction
Bone remodelling plays an essential role in the maintenance of the integrity of skeletal
structures spatially and temporally, which is a coordinated process involving bone cells
in BMU. Osteoblasts and osteoclasts work in a coupling following an activation-
resorption-formation sequence [138]. This coupling has been simulated quantitatively
in signalling-based models [144, 209]. Signalling-based models are developed to
interpret bone remodelling from physiological aspects [142]. Parathyroid hormone
(PTH) is considered to mediate the underlying mechanisms of bone remodelling, and
the temporal effect of PTH is investigated as the key to establish a nonlinear
remodelling process. Signalling interactions among osteoblasts and osteoclasts, such
as RANK, RANKL and OPG signalling pathways, are also incorporated into the
mathematical model for bone remodelling [146]. However, signalling-based models
inevitably introduce unknowns for data weights of signalling factors and raise a
question of how to balance signalling weights in a network.
In Chapter 3 and 5, we proposed a hypothesis and two frameworks for solving joint
cell possibility in EMFs. The effects of the hypothesis and frameworks aim to describe
the coupling between osteoblasts and osteoclasts mathematically by path integration
instead of signalling weight. The path integration deals with cell movement by
physical principles, while the signalling weight must face biochemical reactions.
Therefore, path integration could satisfy the simplicity of formulation. In this chapter,
we solved the framework proposed in Chapter 5 under a simplified circumstance. Even
so, a prediction was presented for showing the change of relative positions between
osteoblasts and osteoclasts in EMFs. Further, we examined the prediction by the
experiment of osteoblasts and osteoclasts co-culture in EMFs.
169
9.2 Prediction
The exploration of the physical structure of bone cells discussed here is not limited to
single cell type but expanded to the interactions of two cell types. To simplify the
analysis without loss of generality, we omit the influence of cell size and membrane
and study the relative positions between osteoblasts and osteoclasts. Under this
assumption, we consider the interaction between osteoblasts and osteoclasts in a two-
dimensional picture in which osteoblasts and osteoclasts are put separately into two
different layers (Figure 9.1). In each layer, an imaginary grid is built with black or
grey nodes. After osteoblasts and osteoclasts occupy the nodes of grids, their
generalised mechanical properties are automatically defined. Consequently, we denote
cell probability of osteoblasts by a vector ( , )r tOB and cell probability of osteoclasts
by ( , )r tOC , where r represents spatial parameter and t represents temporal
parameter. Ideally, ( , )r tOB and ( , )r tOC attach to each node of black and grey grid
respectively, which can be solved by finite difference methods (FDM).
9.2.1 The numerical interaction between osteoblasts and osteoclasts
Although cell probability ( , )r tOB and ( , )r tOC are expected to be inhomogeneous,
the cell probabilities have the same temporal coordinate. In other words, time might
be the only consistent coordinate for different cell types. The temporal derivative of
( , )r tOB is related to the spatial derivative of ( , )r tOC when they have interactions
with each other, and vice-versa. This property is expressed as follows:
( , ) ( , )
( , ) ( , )
OC
OB
r t r tt x
r t r tt x
+
−
OB OC
OC OB
. (9.2.1)
We assume that the relations of (9.2.1) have a general form of equations,
170
( , ) ( ) ( , ) ( , )a
r t r r t r tt r
= +
OB OC , (9.2.2)
and
( , ) ( ) ( , ) ( , )b
r t r r t r tt r
− = +
OC OB . (9.2.3)
Where and are coefficients only containing a spatial variable, ( , )r t and
( , )r t contain both spatial and temporal variables. The solution relies on the
transformation in the spatial coordinate. On a finite domain, when , (1,2]a b ,
equations in (9.2.3) are fractional diffusion equation. When 1a b= = , we can solve
equations with finite-difference-time-domain (FDTD) under the simplest assumption,
1, ( , ) 0, ( , ) 0r t r t = = = = . (9.2.4)
Then,
( , ) ( ) ( , )r t a r r tt r
=
OB OC , (9.2.5)
( , ) ( ) ( , )r t b r r tt x
− =
OC OB . (9.2.6)
Following the construction of FDTD, the temporal derivative of ( , )r tOB with second-
order accuracy is approximated,
1
( , / 2) ( ( , ) ( , ))r t t r t t r tt t
+ = + −
OB OB OB . (9.2.7)
At the same temporal point, the spatial derivative of ( , )r tOC is evaluated with the
same method as:
1
( , / 2) [ ( / 2, / 2) ( / 2, / 2)]r t t r r t t r r t tr r
+ = + + − − +
OC OC OC
(9.2.8)
171
( , ) ( , ) a( ) [ ( / 2, / 2) ( / 2, / 2)]t
r t t r t r r r t t r r t tx
+ = + + + − − +
OB OB OC OC
(9.2.9)
Note that the values of r and t exist on the discrete points in space and time. We
continue to update the second equation from time to time / 2t t+ at the spatial point
/ 2r r− with a similar method, and obtain:
( / 2, / 2) ( / 2, / 2) b( ) [ ( , ) ( , )]t
r r t t r r t t r r r t r tx
− + = − − + − −
OC OC OB OB
(9.2.10)
/r t is defined as the diffusion velocity, which is proportional to cell proliferation
during a finite time domain. Then a( ) /r t r is inversely proportional to cell
proliferation of osteoclasts during a finite time domain and b( ) /r t r is inversely
proportional to cell proliferation of osteoblasts during a finite time domain. In the
program of simulation, the finite time domain is realised by several time steps.
Therefore, more time steps are divided in a finite time domain, less cell proliferation
of bone cells would be at each time step, which directly affects the magnitude of
a( ) /r t r and b( ) /r t r . It is noted that their magnitude may not be constant for
each time step since the cell proliferation is nonlinear.
The equations can exist only if the initial ( , )r tOB and ( , )r tOC keep stable.
Considering the cell cycle of osteoblasts and osteoclasts, they can only survive during
a limited time. Hence the time step is limited. After cell death, the system needs to be
re-initialised. However, history affects the re-initialisation and leads to the
accumulation in a new time step.
172
9.2.2 Numerical results
The boundary of ( , )r tOB in the numerical simulation was set as 21/( t) (0,1] , and
positions of cells were expressed as lg( )r . The magnitude of lg( / )r t determined the
relative position of OB and OC . When lg( / )r t increased from 1 to 9, the spatial
interval between OB and OC was minimised and the interference of cell probability
occurred (Figure 9.2). The function lg( / )r t was evaluated as a factor of diffusion.
The factor of diffusion profoundly affected the interference between cell probability
of osteoblasts and osteoclasts. High diffusion resulted in substantial entanglement
among osteoblasts and osteoclasts. In contrast, low diffusion led to spatial separation
among osteoblasts and osteoclasts. When osteoblasts and osteoclasts are cultured in
EMFs, the diffusion may vary from the situations of on/off EMFs, which forms a co-
existence of high and low diffusion. This circumstance generates interference in a
wave pattern over the cell culture. Therefore, entanglement and separation between
cells are expected to exist at the same time.
9.3 Validation in experimental images
Co-culture of human osteoblasts and osteoclasts began with separately culturing
osteoblasts and osteoclasts with the protocols in Chapter 4. Co-culture was initiated on
Day 8 by seeding osteoblasts on the plate of RANKL induced PBMC. The medium of
co-culture used completed osteoclast cell culture medium with RANKL. The medium
was changed every three days. The EMFs exposure started from day 9, and the
experiment finished on day 14. The co-culture was exposed under SMF of 0.69 mT
and PEMF of 0.69 mT with 160 Hz.
173
9.3.1 TRAP staining
TRAP staining on osteoclasts was performed in co-culture of human osteoblasts
(passage 2) and human osteoclasts in control, SMF and PEMF exposure (Figure 9.3).
The intensity of SMF exposure was 0.69 mT, and the PEMF exposure was at 0.69 mT
with 160 Hz. The substantial entanglement of cells was observed in the group under
the SMF and PEMF exposure (Figure 9.3 E-L). The entanglement illustrated that
osteoblasts were more likely to contact with osteoclasts under the SMF (Figure 9.3 J)
and PEMF (Figure 9.3 F) in comparison with the control (Figure 9.3 B). The
entanglement with the PEMF exposure promoted the forming of mature osteoclasts in
Figure 9.3 G, while inhabited the forming of mature osteoclasts in Figure 9.3 E. The
entanglement with the SMF exposure showed similar results. Moreover, the SMF
exposure limited the size of mature osteoclasts in contrast to the control and PEMF.
9.3.2 Fluorescence-based staining and flow cytometry analysis
The fluorescence-based staining illustrated similar results with TRAP staining.
Compared with the control (Figure 9.4 A), the SMF (Figure 9.4 E) and PEMF
exposure (Figure 9.4 C) limited the size of mature osteoclasts. Additionally, the SMF
exposure inhabited the forming of mature osteoclasts in contrast to PEMF exposure.
The results of flow cytometry presented the promotion on cell proliferation of
osteoblasts under the PEMF (Figure 9.4 D) and SMF exposure (Figure 9.4 F). The
SMF exposure resulted in higher cell proliferation of osteoblasts in contrast to PEMF
exposure. The observation indicated that the reduction of forming mature osteoclasts
companied with the increase in cell proliferation of osteoblasts in co-culture under
EMFs, which is consistent with the theoretical interaction between osteoblasts and
osteoclasts in the equation (9.2.1). The interference of cell probability in EMFs could
contribute to the separation of cell groups in Figure 9.4 D, F.
174
9.3.3 Scanning electron microscope (SEM) images
The SEM images were taken to elucidating the possible cellular interaction in SMF
and PEMF exposure (Figure 9.5). The co-culture of osteoblasts and osteoclasts in
control distributed evenly (Figure 9.5 A, B). After the PEMF and SMF exposure, the
overlapping cells were observed (Figure 9.5 C-F). The overlap was shown densely
after the SMF exposure (Figure 9.5 E, F) which could be the reason for preventing the
forming of mature osteoclasts. The dense overlap of cells after the SMF exposure also
showed that magnetic intensity worked in a major role for the cellular contact among
bone cells. The interference of cell probability in EMFs may originate from the overlap
of cells which is equivalent to the superposition of cell probability in Chapter 3.
175
Figure 9.1 Imaginary grid of osteoblasts (OB) and osteoclasts (OC). Two types of
cells are assigned to black and grey nodes respectively. Any given node of the black
grid is surrounded by neighbours that belong to the grey grid. The spatial variables
change with time.
176
(A) Initial stage
(B) lg( / ) 3r t =
(C) lg( / ) 6r t =
(D) lg( / ) 9r t =
Figure 9.2 Prediction of the relative position between osteoblasts ( OB ) and
osteoclasts ( OC ). When lg( / )r t increases from 1 to 9, the intervals between
osteoblasts and osteoclasts tend to be minimised, which leads to the interference of
cell probability. The function lg( / )r t contains the factor of diffusion. The numerical
results illustrated that high diffusion resulted in a substantial interference of osteoblasts
and osteoclasts.
177
(A) Control 1; (100 µm)
(B) Control 1; (50 µm)
(C) Control 2; (100 µm)
(D) Control 3; (100 µm)
(E) PEMF 1;
0.69 mT, 160 Hz (100 µm)
(F) PEMF 1;
0.69 mT, 160 Hz (50 µm)
178
(G) PEMF 2;
0.69 mT, 160 Hz (100 µm)
(H) PEMF 3;
0.69 mT, 160 Hz (100 µm)
(I) SMF 1; 0.69 mT (100 µm)
(J) SMF 1; 0.69 mT (50 µm)
(K) SMF 2; 0.69 mT (100 µm)
(L) SMF 3; 0.69 mT (100 µm)
179
Figure 9.3 TRAP staining on osteoclasts performed in co-culture of human osteoblasts
and osteoclast in control, SMF and PEMF exposure. The SMF exposure was at the
intensity of 0.69 mT, and the PEMF exposure was at 0.69 mT with 160 Hz. The active
interference was observed in the group under SMF and PEMF exposure. Compared
with PEMF exposure, SMF exposure showed substantial entanglement.
180
(A) Control
(B) Control
(C) PEMF 0.69 mT, 160 Hz
(D) PEMF 0.69 mT, 160 Hz
(E) SMF 0.69 mT
(F) SMF 0.69 mT
181
Figure 9.4 The fluorescence-based staining in control, SMF and PEMF exposure.
Compared with the control, the SMF and PEMF exposure prevented the forming of
mature osteoclasts. SMF and PEMF exposure inhabited the forming of mature
osteoclasts while promoted cell proliferation of osteoblasts. This phenomenon is
consistent with the prediction of spatial interaction between osteoblasts and osteoclasts
182
(A) Control (2k x)
(B) Control (7k x)
(C) PEMF 0.69 mT, 160 Hz (2k x)
(D) PEMF 0.69 mT, 160 Hz (7k x)
(E) SMF 0.69 mT (2k x)
(F) SMF 0.69 mT (10k x)
183
Figure 9.5 Scanning electron microscope (SEM) images of osteoblasts and osteoclasts
co-culture in control, SMF and PEMF exposure. The co-culture of osteoblasts and
osteoclasts in control distributed evenly. In the PEMF and SMF exposure, the
overlapping cells were observed. The overlap was shown densely after the SMF
exposure which could be the reason for preventing the forming of mature osteoclasts.
184
Chapter 10
Conclusions
185
In conclusion, this thesis achieved the following objectives.
1. A weak PEMF and SMF system is designed and constructed. The exposure
system for studying the effects of weak PEMF and SMF on bone cells was
constructed with self-design parts and commercial devices. The exposure
chamber is switchable between the solenoid and Helmholtz coil systems. The
magnetic flux can be either a gradient field or a uniform field. The structural
design of the exposure system allows experiments to be conducted at a proper
temperature on a maximum of four 6-well plates at the same time.
2. A mathematical modelling is built for articulating different dataset into one
picture. A general model was established for analysing the experimental data
in PEMF and SMF exposure. This model can be incorporated with the various
framework and able to calculate the numerical results in the various scenario
of the experimental environment.
3. Experimental effects of weak PEMF and SMF on osteoblasts and osteoclasts
are evaluated at various parameters, including (i) type of magnetic fields, (ii)
magnetic flux density, (iii) frequency, (iv) exposure duration, and (v) spatial
gradient (dB/dx). For SMF, two gradient fields of SMFs were applied to
cultured Saos-2 cells. The effects of cell proliferation were examined after the
exposure of 4 hours and 8 hours respectively. The first gradient field contained
intensities at the centre of the culture well with 2, 1.1, 0.6, 0.48, 0.37 and 0.3
mT. The second gradient field contained intensities at the centre of the culture
well with 2.5, 1.6, 1, 0.75, 0.55, and 0.42 mT. For PEMF, the effects of
sinusoidal PEMF intensity and frequency were analysed on cultured Saos-2
cells. Six magnetic flux densities were investigated at the centre of each well:
186
control, 0.1, 0.15, 0.25, 0.53 and 0.69 mT. The range of frequency was
investigated in 40, 80, 160, 320, 640 and 800 Hz.
4. The interference term is derived from the mathematical description of cell
behaviour in EMF. Cell probability was proposed to be evolving with spatial
and temporal factors. Different results could be obtained when the experiment
was carried out in different places and measured at a different time. The
interference should satisfy a differential equation which could be a wave
equation analogous to the equation for electromagnetic waves. However, it
must be emphasised that this function is not a real wave in space even when it
satisfies the wave equation. The existence of interference assumed that the cells
were equivalent to particles with a very long wavelength. Hence the
interference can be observed in the experiment if the various experimental
arrangement were set up at the same time.
5. The cell proliferation of osteoblasts and co-culture of osteoblasts and
osteoclasts are affected by the mutual interference generated in the
environment of SMFs and PEMFs. The proof included the numerical analysis
of cultured Saos-2 cell proliferation, morphology of human osteoblasts and
imaging evidence in co-culture of osteoblasts and osteoclasts. Based on the
observation of interference term in cell proliferation, the optimised EMF
exposure for promoting the growth of bone cells in vitro might be obtained by
the combination of various PEMFs or adoption of gradient SMFs.
187
References
188
[1] P. Kovacic, R. Somanathan, Electromagnetic fields: mechanism, cell signaling,
other bioprocesses, toxicity, radicals, antioxidants and beneficial effects,
Journal of Receptors and Signal Transduction, 30(4) (2010) 214-226.
[2] R. Goodman, M. Blank, Insights into electromagnetic interaction mechanisms,
Journal of cellular physiology, 192(1) (2002) 16-22.
[3] K. Hug, M. Röösli, Therapeutic effects of whole‐body devices applying pulsed
electromagnetic fields (PEMF): A systematic literature review,
Bioelectromagnetics, 33(2) (2012) 95-105.
[4] R.H. Funk, T. Monsees, N. Ozkucur, Electromagnetic effects - From cell biology
to medicine, Progress in histochemistry and cytochemistry, 43(4) (2009) 177-
264.
[5] R.G. Martin, Electromagnetic field theory for physicists and
engineers:Fundamentals and Applications, 2006.
[6] K.J. Lohmann, Q&A: Animal behaviour: Magnetic-field perception, Nature,
464(7292) (2012) 1140.
[7] L. Dini, L. Abbro, Bioeffects of moderate-intensity static magnetic fields on cell
cultures, Micron, 36(3) (2005) 195-217.
[8] A.D. Rosen, Mechanism of action of moderate-intensity static magnetic fields on
biological systems, Cell Biochemistry and Biophysics, 39(2) (2003) 163-173.
[9] A. Liboff, S. Cherng, K. Jenrow, A. Bull, Calmodulin‐dependent cyclic
nucleotide phosphodiesterase activity is altered by 20 μT magnetostatic
fields, Bioelectromagnetics, 24(1) (2003) 32-38.
[10] M. Buemi, D. Marino, G. Di Pasquale, F. Floccari, M. Senatore, C. Aloisi, F.
Grasso, G. Mondio, P. Perillo, N. Frisina, Cell proliferation/cell death balance
in renal cell cultures after exposure to a static magnetic field, Nephron, 87(3)
(2001) 269-273.
[11] H. Sonnier, O. Kolomytkin, A. Marino, Action potentials from human
neuroblastoma cells in magnetic fields, Neuroscience Letters, 337(3) (2003)
163-166.
[12] A.D. Rosen, Inhibition of calcium channel activation in GH3 cells by static
magnetic fields, Biochimica et Biophysica Acta (BBA)-Biomembranes,
1282(1) (1996) 149-155.
[13] A.D. Rosen, Effect of a 125 mT static magnetic field on the kinetics of voltage
activated Na+ channels in GH3 cells, Bioelectromagnetics, 24(7) (2003) 517-
523.
[14] H.M. Huang, S.Y. Lee, W.C. Yao, C.T. Lin, C.Y. Yeh, Static magnetic fields
up-regulate osteoblast maturity by affecting local differentiation factors,
Clinical Orthopaedics and Related Research, 447 (2006) 201-208.
[15] K.H. Chiu, K.L. Ou, S.Y. Lee, C.T. Lin, W.J. Chang, C.C. Chen, H.M. Huang,
Static magnetic fields promote osteoblast-like cells differentiation via
increasing the membrane rigidity, Annals of Biomedical Engineering, 35(11)
(2007) 1932-1939.
[16] C. Cunha, S. Panseri, M. Marcacci, A. Tampieri, Evaluation of the effects of a
moderate intensity static magnetic field application on human osteoblast-like
cells, American Journal of Biomedical Engineering, 2(6) (2012) 263-268.
[17] S.W. Feng, Y.J. Lo, W.J. Chang, C.T. Lin, S.Y. Lee, Y. Abiko, H.-M. Huang,
Static magnetic field exposure promotes differentiation of osteoblastic cells
grown on the surface of a poly-L-lactide substrate, Medical & Biological
Engineering & Computing, 48(8) (2010) 793-798.
189
[18] P. Cooke, P. Morris, The effects of NMR exposure on living organisms. II. A
genetic study of human lymphocytes, The British Journal of Radiology,
54(643) (1981) 622-625.
[19] S. Pacini, G.B. Vannelli, T. Barni, M. Ruggiero, I. Sardi, P. Pacini, M. Gulisano,
Effect of 0.2 T static magnetic field on human neurons: remodeling and
inhibition of signal transduction without genome instability, Neuroscience
Letters, 267(3) (1999) 185-188.
[20] S. Pacini, M. Gulisano, B. Peruzzi, E. Sgambati, G. Gheri, S.G. Bryk, S.
Vannucchi, G. Polli, M. Ruggiero, Effects of 0.2 T static magnetic field on
human skin fibroblasts, Cancer Detection and Prevention, 27(5) (2003) 327-
332.
[21] L. Teodori, J. Grabarek, P. Smolewski, L. Ghibelli, A. Bergamaschi, M. De
Nicola, Z. Darzynkiewicz, Exposure of cells to static magnetic field
accelerates loss of integrity of plasma membrane during apoptosis,
Cytometry, 49(3) (2002) 113-118.
[22] M. Iwasaka, S. Ueno, H. Tsuda, Effects of magnetic fields on fibrinolysis,
Journal of Applied Physics, 75(10) (1994) 7162-7164.
[23] H. Kotani, H. Kawaguchi, T. Shimoaka, M. Iwasaka, S. Ueno, H. Ozawa, K.
Nakamura, K. Hoshi, Strong static magnetic field stimulates bone formation
to a definite orientation in vitro and in vivo, Journal of Bone and Mineral
Research, 17(10) (2002) 1814-1821.
[24] J. Wiskirchen, E. Groenewaeller, R. Kehlbach, F. Heinzelmann, M. Wittau, H.
Rodemann, C. Claussen, S. Duda, Long‐term effects of repetitive exposure
to a static magnetic field (1.5 T) on proliferation of human fetal lung
fibroblasts, Magnetic Resonance in Medicine: An Official Journal of the
International Society for Magnetic Resonance in Medicine, 41(3) (1999) 464-
468.
[25] R.R. Raylman, A.C. Clavo, R.L. Wahl, Exposure to strong static magnetic field
slows the growth of human cancer cells in vitro, (1996).
[26] C. Aldinucci, J.B. Garcia, M. Palmi, G. Sgaragli, A. Benocci, A. Meini, F.
Pessina, C. Rossi, C. Bonechi, G.P. Pessina, The effect of strong static
magnetic field on lymphocytes, Bioelectromagnetics: Journal of the
Bioelectromagnetics Society, The Society for Physical Regulation in Biology
and Medicine, The European Bioelectromagnetics Association, 24(2) (2003)
109-117.
[27] H. Miyamoto, H. Yamaguchi, T. Ikehara, Y. Kinouchi, Effects of
electromagnetic fields on K+(Rb+) uptake by HeLa cells, in: Biological
effects of magnetic and electromagnetic fields, Springer, 1996, pp. 101-119.
[28] M. Iwasaka, K. Yamamoto, J. Ando, S. Ueno, Verification of magnetic field
gradient effects on medium convection and cell adhesion, Journal of Applied
Physics, 93(10) (2003) 6715-6717.
[29] H. Hirose, T. Nakahara, J. Miyakoshi, Orientation of human glioblastoma cells
embedded in type I collagen, caused by exposure to a 10 T static magnetic
field, Neuroscience Letters, 338(1) (2003) 88-90.
[30] Y. Eguchi, M. Ogiue-Ikeda, S. Ueno, Control of orientation of rat Schwann cells
using an 8-T static magnetic field, Neuroscience Letters, 351(2) (2003) 130-
132.
[31] Z. Zhao, J. Tao, J. Su, C. Ai, Y. Liu, J. Wang, The Design and Measurement of
Pulsed Magnetic Field, in: Measuring Technology and Mechatronics
190
Automation (ICMTMA), 2011 Third International Conference on, 2011, pp.
744-747.
[32] W. Latham, J.T.C. Lau, Bone Stimulation: A Review of Its Use as an Adjunct,
Techniques in Orthopaedics, 26(1) (2011) 14-21.
[33] L.Y. Sun, D.K. Hsieh, P.C. Lin, H.T. Chiu, T.W. Chiou, Pulsed electromagnetic
fields accelerate proliferation and osteogenic gene expression in human bone
marrow mesenchymal stem cells during osteogenic differentiation,
Bioelectromagnetics, 31(3) (2010) 209-219.
[34] B. Tenuzzo, A. Chionna, E. Panzarini, R. Lanubile, P. Tarantino, B.D. Jeso, M.
Dwikat, L. Dini, Biological effects of 6 mT static magnetic fields: a
comparative study in different cell types, Bioelectromagnetics, 27(7) (2006)
560-577.
[35] V. Sollazzo, G.C. Traina, M. DeMattei, A. Pellati, F. Pezzetti, A. Caruso,
Responses of human MG‐63 osteosarcoma cell line and human osteoblast‐like cells to pulsed electromagnetic fields, Bioelectromagnetics: Journal of
the Bioelectromagnetics Society, The Society for Physical Regulation in
Biology and Medicine, The European Bioelectromagnetics Association, 18(8)
(1997) 541-547.
[36] C.T. Brighton, W. Wang, R. Seldes, G. Zhang, S.R. Pollack, Signal transduction
in electrically stimulated bone cells, Journal of Bone and Joint Surgery-
American Volume, 83(10) (2001) 1514-1523.
[37] L. Massari, F. Benazzo, M. De Mattei, S. Setti, M. Fini, Effects of electrical
physical stimuli on articular cartilage, Journal of Bone and Joint Surgery-
American Volume, 89(suppl_3) (2007) 152-161.
[38] R.K. Aaron, D.M. Ciombor, G. Jolly, Stimulation of experimental endochondral
ossification by low‐energy pulsing electromagnetic fields, Journal of Bone
and Mineral Research, 4(2) (1989) 227-233.
[39] R.K. Aaron, D.M. Ciombor, H. Keeping, S. Wang, A. Capuano, C. Polk, Power
frequency fields promote cell differentiation coincident with an increase in
transforming growth factor‐β1 expression, Bioelectromagnetics, 20(7)
(1999) 453-458.
[40] R.J. Fitzsimmons, J.T. Ryaby, F.P. Magee, D.J. Baylink, IGF‐II receptor
number is increased in TE‐85 osteosarcoma cells by combined magnetic
fields, Journal of Bone and Mineral Research, 10(5) (1995) 812-819.
[41] R. Selvam, K. Ganesan, K.N. Raju, A.C. Gangadharan, B.M. Manohar, R.
Puvanakrishnan, Low frequency and low intensity pulsed electromagnetic
field exerts its anti-inflammatory effect through restoration of plasma
membrane calcium ATPase activity, Life sciences, 80(26) (2007) 2403-2410.
[42] M.L. Pall, Electromagnetic fields act via activation of voltage‐gated calcium
channels to produce beneficial or adverse effects, Journal of Cellular and
Molecular Medicine, 17(8) (2013) 958-965.
[43] B. Noriega-Luna, M. Sabanero, M. Sosa, M. Avila-Rodriguez, Influence of
pulsed magnetic fields on the morphology of bone cells in early stages of
growth, Micron, 42(6) (2011) 600-607.
[44] J. Zhou, L.G. Ming, B.F. Ge, J.Q. Wang, R.Q. Zhu, Z. Wei, H.P. Ma, C.J. Xian,
K.M. Chen, Effects of 50 Hz sinusoidal electromagnetic fields of different
intensities on proliferation, differentiation and mineralization potentials of rat
osteoblasts, Bone, 49(4) (2011) 753-761.
191
[45] M. De Mattei, A. Caruso, G.C. Traina, F. Pezzetti, T. Baroni, V. Sollazzo,
Correlation between pulsed electromagnetic fields exposure time and cell
proliferation increase in human osteosarcoma cell lines and human normal
osteoblast cells in vitro, Bioelectromagnetics: Journal of the
Bioelectromagnetics Society, The Society for Physical Regulation in Biology
and Medicine, The European Bioelectromagnetics Association, 20(3) (1999)
177-182.
[46] S. Ahmadian, S.R. Zarchi, B. Bolouri, Effects of extremely‐low‐frequency
pulsed electromagnetic fields on collagen synthesis in rat skin, Biotechnology
and Applied Biochemistry, 43(2) (2006) 71-75.
[47] C. Bauréus Koch, M. Sommarin, B. Persson, L. Salford, J. Eberhardt,
Interaction between weak low frequency magnetic fields and cell membranes,
Bioelectromagnetics, 24(6) (2003) 395-402.
[48] V. Akpolat, M.S. Celik, Y. Celik, N. Akdeniz, M.S. Ozerdem, Treatment of
osteoporosis by long-term magnetic field with extremely low frequency in
rats, Gynecological Endocrinology, 25(8) (2009) 524-529.
[49] N.M. Shupak, F.S. Prato, A.W. Thomas, Therapeutic uses of pulsed magnetic-
field exposure: a review, URSI Radio Science Bulletin, 76(4) (2003) 9-32.
[50] B. Strauch, C. Herman, R. Dabb, L.J. Ignarro, A.A. Pilla, Evidence-Based Use
of Pulsed Electromagnetic Field Therapy in Clinical Plastic Surgery,
Aesthetic Surgery Journal, 29(2) (2009) 135-143.
[51] A.R. Sul, S.N. Park, H. Suh, Effects of sinusoidal electromagnetic field on
structure and function of different kinds of cell lines, Yonsei Medical Journal,
47(6) (2006) 852-861.
[52] C.F. Martino, D. Belchenko, V. Ferguson, S. Nielsen Preiss, H.J. Qi, The effects
of pulsed electromagnetic fields on the cellular activity of SaOS‐2 cells,
Bioelectromagnetics, 29(2) (2008) 125-132.
[53] V. Grote, H. Lackner, C. Kelz, M. Trapp, F. Aichinger, H. Puff, M. Moser,
Short-term effects of pulsed electromagnetic fields after physical exercise are
dependent on autonomic tone before exposure, European Journal of Applied
Physiology, 101(4) (2007) 495-502.
[54] K. Chang, W.H.S. Chang, M.L. Wu, C. Shih, Effects of different intensities of
extremely low frequency pulsed electromagnetic fields on formation of
osteoclast‐like cells, Bioelectromagnetics, 24(6) (2003) 431-439.
[55] J.L. Kirschvink, Uniform magnetic fields and double‐wrapped coil systems:
improved techniques for the design of bioelectromagnetic experiments,
Bioelectromagnetics, 13(5) (1992) 401-411.
[56] W.R. Adey, Biological effects of electromagnetic fields, Journal of Cellular
Biochemistry, 51(4) (1993) 410-416.
[57] M.S. Markov, How living systems recognize applied electromagnetic fields, The
Environmentalist, 31(2) (2011) 89-96.
[58] K. R. Foster, Mechanisms of interaction of extremely low frequency electric
fields and biological systems, Radiation Protection Dosimetry, 106(4) (2003)
301-310.
[59] N. Salansky, A. Fedotchev, A. Bondar, Responses of the nervous system to low
frequency stimulation and EEG rhythms: clinical implications, Neuroscience
& Biobehavioral Reviews, 22(3) (1998) 395-409.
[60] A. De Loof, The electrical dimension of cells: the cell as a miniature
electrophoresis chamber, in: International review of cytology, Elsevier, 1986,
pp. 251-352.
192
[61] Y. Antov, A. Barbul, R. Korenstein, Electroendocytosis: stimulation of
adsorptive and fluid-phase uptake by pulsed low electric fields, Experimental
Cell Research, 297(2) (2004) 348-362.
[62] R. Goodman, A.S. Henderson, Some biological effects of electromagnetic
fields, Bioelectrochemistry and Bioenergetics, 15(1) (1986) 39-55.
[63] P. Volpe, Interactions of zero-frequency and oscillating magnetic fields with
biostructures and biosystems, Photochemical & Photobiological Sciences,
2(6) (2003) 637-648.
[64] D.H. Trock, Electromagnetic fields and magnets: investigational treatment for
musculoskeletal disorders, Rheumatic Disease Clinics of North America,
26(1) (2000) 51-62.
[65] R.C. Riddle, H.J. Donahue, From streaming‐potentials to shear stress: 25 years
of bone cell mechanotransduction, Journal of Orthopaedic Research, 27(2)
(2009) 143-149.
[66] C. Zhu, G. Bao, N. Wang, Cell mechanics: mechanical response, cell adhesion,
and molecular deformation, Annual Review of Biomedical Engineering, 2(1)
(2000) 189-226.
[67] J. Kirschvink, J. KDiaz‐Ricci, S. Kirschvink, Magnetite in human tissues: a
mechanism for the biological effects of weak ELF magnetic fields,
Bioelectromagnetics, 13(S1) (1992) 101-113.
[68] C. Polk, Physical mechanisms by which low-frequency magnetic fields can
affect the distribution of counterions on cylindrical biological cell surfaces,
Journal of Biological Physics, 14(1) (1986) 3-8.
[69] A. Buchachenko, Why magnetic and electromagnetic effects in biology are
irreproducible and contradictory?, Bioelectromagnetics, 37(1) (2016) 1-13.
[70] A. Buchachenko, D. Kuznetsov, Magnetic control of enzymatic
phosphorylation, Journal of Physical Chemistry and Biophysics, 4(2) (2014)
9.
[71] P. Hore, Are biochemical reactions affected by weak magnetic fields?,
Proceedings of the National Academy of Sciences, 109(5) (2012) 1357-1358.
[72] I.A. Shovkovy, Magnetic catalysis: a review, in: Strongly Interacting Matter in
Magnetic Fields, Springer, 2013, pp. 13-49.
[73] A.L. Buchachenko, D.A. Kuznetsov, Magnetic field affects enzymatic ATP
synthesis, Journal of the American Chemical Society, 130(39) (2008) 12868-
12869.
[74] A. Liboff, j.T. Williams, D. Strong, j.R. Wistar, Time-varying magnetic fields:
effect on DNA synthesis, Science, 223(4638) (1984) 818-820.
[75] S. Xu, H. Okano, N. Tomita, Y. Ikada, Recovery effects of a 180 mT static
magnetic field on bone mineral density of osteoporotic lumbar vertebrae in
ovariectomized rats, Evidence-Based Complementary and Alternative
Medicine, 2011 (2011).
[76] H. Zhang, L. Gan, X. Zhu, J. Wang, L. Han, P. Cheng, D. Jing, X. Zhang, Q.
Shan, Moderate-intensity 4 mT static magnetic fields prevent bone
architectural deterioration and strength reduction by stimulating bone
formation in streptozotocin-treated diabetic rats, Bone, 107 (2018) 36-44.
[77] H.R. Gungor, S. Akkaya, N. Ok, A. Yorukoglu, C. Yorukoglu, E. Kiter, E.O.
Oguz, N. Keskin, G.A. Mete, Chronic Exposure to Static Magnetic Fields
from Magnetic Resonance Imaging Devices Deserves Screening for
Osteoporosis and Vitamin D Levels: A Rat Model, International Journal of
Environmental Research and Public Health, 12(8) (2015) 8919-8932.
193
[78] C. Liu, Y. Zhang, T. Fu, Y. Liu, S. Wei, Y. Yang, D. Zhao, W. Zhao, M. Song,
X. Tang, H. Wu, Effects of electromagnetic fields on bone loss in
hyperthyroidism rat model, Bioelectromagnetics, 38(2) (2017) 137-150.
[79] B.Y. Zhu, Z.D. Yang, X.R. Chen, J. Zhou, Y.H. Gao, C.J. Xian, K.M. Chen,
Exposure Duration Is a Determinant of the Effect of Sinusoidal
Electromagnetic Fields on Peak Bone Mass of Young Rats, Calcified Tissue
International, 103(1) (2018) 95-106.
[80] Y. Atalay, N. Gunes, M.D. Guner, V. Akpolat, M.S. Celik, R. Guner,
Pentoxifylline and electromagnetic field improved bone fracture healing in
rats, Drug Design, Development and Therapy, 9 (2015) 5195.
[81] J. Zhou, X. Li, Y. Liao, W. Feng, C. Fu, X. Guo, Pulsed electromagnetic fields
inhibit bone loss in streptozotocin-induced diabetic rats, Endocrine, 49(1)
(2015) 258-266.
[82] J.C. Crockett, M.J. Rogers, F.P. Coxon, L.J. Hocking, M.H. Helfrich, Bone
remodelling at a glance, Journal of Cell Science, 124(7) (2011) 991-998.
[83] H.K. Datta, W.F. Ng, J.A. Walker, S.P. Tuck, S.S. Varanasi, The cell biology of
bone metabolism, Journal of Clinical Pathology, 61(5) (2008) 577-587.
[84] N.A. Sims, J.H. Gooi, Bone remodeling: Multiple cellular interactions required
for coupling of bone formation and resorption, Seminars in Cell &
Developmental Biology, 19(5) (2008) 444-451.
[85] E. Eriksen, Cellular mechanisms of bone remodeling, Reviews in Endocrine &
Metabolic Disorders, 11(4) (2010) 219-227.
[86] S. Khosla, The bone and beyond: a shift in calcium, Nature medicine, 17(4)
(2011) 430-431.
[87] K. Miyazono, S. Maeda, T. Imamura, BMP receptor signaling: Transcriptional
targets, regulation of signals, and signaling cross-talk, Cytokine &
Growth Factor Reviews, 16(3) (2005) 251-263.
[88] S.i. Harada, G.A. Rodan, Control of osteoblast function and regulation of bone
mass, Nature, 423(6937) (2003) 349.
[89] J. Crockett, D. Mellis, D. Scott, M. Helfrich, New knowledge on critical
osteoclast formation and activation pathways from study of rare genetic
diseases of osteoclasts: focus on the RANK/RANKL axis, Osteoporosis
International, 22(1) (2011) 1-20.
[90] T. Phan, J. Xu, M. Zheng, Interaction between osteoblast and osteoclast: impact
in bone disease, Histology and histopathology, 19(4) (2004) 1325-1344.
[91] P. Ducy, T. Schinke, G. Karsenty, The osteoblast: a sophisticated fibroblast
under central surveillance, Science, 289(5484) (2000) 1501-1504.
[92] P. Chatakun, R. Núñez-Toldrà, E.D. López, C. Gil-Recio, E. Martínez-Sarrà, F.
Hernández-Alfaro, E. Ferrés-Padró, L. Giner-Tarrida, M. Atari, The effect of
five proteins on stem cells used for osteoblast differentiation and
proliferation: a current review of the literature, Cellular and Molecular Life
Sciences, 71(1) (2014) 113-142.
[93] T. Nakashima, M. Hayashi, T. Fukunaga, K. Kurata, M. Oh-hora, J.Q. Feng,
L.F. Bonewald, T. Kodama, A. Wutz, E.F. Wagner, J.M. Penninger, H.
Takayanagi, Evidence for osteocyte regulation of bone homeostasis through
RANKL expression, Nature medicine, 17(10) (2011) 1231-1234.
[94] W.J. Boyle, W.S. Simonet, D.L. Lacey, Osteoclast differentiation and
activation, Nature, 423(6937) (2003) 337-342.
[95] D.J. Mellis, C. Itzstein, M.H. Helfrich, J.C. Crockett, The skeleton: a multi-
functional complex organ: the role of key signalling pathways in osteoclast
194
differentiation and in bone resorption, Journal of Endocrinology, 211(2)
(2011) 131-143.
[96] T. Wada, T. Nakashima, N. Hiroshi, J.M. Penninger, RANKL–RANK signaling
in osteoclastogenesis and bone disease, Trends in Molecular Medicine, 12(1)
(2006) 17-25.
[97] F. Shen, M.J. Ruddy, P. Plamondon, S.L. Gaffen, Cytokines link osteoblasts and
inflammation: microarray analysis of interleukin‐17‐and TNF‐α‐induced genes in bone cells, Journal of Leukocyte Biology, 77(3) (2005) 388-
399.
[98] Y. Tanio, H. Yamazaki, T. Kunisada, K. Miyake, S.-I. Hayashi, CD9 molecule
expressed on stromal cells is involved in osteoclastogenesis, Experimental
Hematology, 27(5) (1999) 853-859.
[99] W.J. Boyle, W.S. Simonet, D.L. Lacey, Osteoclast differentiation and
activation, Nature, 423(6937) (2003) 337.
[100] T. Nakashima, H. Takayanagi, The dynamic interplay between osteoclasts and
the immune system, Archives of Biochemistry and Biophysics, 473(2) (2008)
166-171.
[101] R.K. Aaron, S. Wang, D.M. Ciombor, Upregulation of basal TGFβ1 levels by
EMF coincident with chondrogenesis – implications for skeletal repair and
tissue engineering, Journal of Orthopaedic Research, 20(2) (2002) 233-240.
[102] T. Bodamyali, B. Bhatt, F.J. Hughes, V.R. Winrow, J.M. Kanczler, B. Simon,
J. Abbott, D.R. Blake, C.R. Stevens, Pulsed Electromagnetic Fields
Simultaneously Induce Osteogenesis and Upregulate Transcription of Bone
Morphogenetic Proteins 2 and 4 in Rat Osteoblastsin Vitro, Biochemical and
Biophysical Research Communications, 250(2) (1998) 458-461.
[103] H.H. Guerkov, C.H. Lohmann, Y. Liu, D.D. Dean, B.J. Simon, J.D. Heckman,
Z. Schwartz, B.D. Boyan, Pulsed Electromagnetic Fields Increase Growth
Factor Release by Nonunion Cells, Clinical Orthopaedics and Related
Research, 384 (2001) 265-279.
[104] J.H. Jansen, O.P. van der Jagt, B.J. Punt, J.A. Verhaar, J.P. van Leeuwen, H.
Weinans, H. Jahr, Stimulation of osteogenic differentiation in human
osteoprogenitor cells by pulsed electromagnetic fields: an in vitro study,
BMC musculoskeletal disorders, 11 (2010) 188.
[105] K. Nie, A. Henderson, MAP kinase activation in cells exposed to a 60 Hz
electromagnetic field, Journal of Cellular Biochemistry, 90(6) (2003) 1197-
1206.
[106] M. Schnoke, R.J. Midura, Pulsed electromagnetic fields rapidly modulate
intracellular signaling events in osteoblastic cells: comparison to parathyroid
hormone and insulin, Journal of Orthopaedic Research, 25(7) (2007) 933-
940.
[107] C.H. Heldin, K. Miyazono, P. ten Dijke, TGF-beta signalling from cell
membrane to nucleus through SMAD proteins, Nature, 390(6659) (1997)
465-471.
[108] A. von Bubnoff, K.W.Y. Cho, Intracellular BMP Signaling Regulation in
Vertebrates: Pathway or Network?, Developmental Biology, 239(1) (2001) 1-
14.
[109] J.J. Westendorf, R.A. Kahler, T.M. Schroeder, Wnt signaling in osteoblasts
and bone diseases, Gene, 341 (2004) 19-39.
[110] E. Canalis, Update in new anabolic therapies for osteoporosis, The Journal of
Clinical Endocrinology & Metabolism, 95(4) (2010) 1496-1504.
195
[111] T.L. McCarthy, M. Centrella, Novel links among Wnt and TGF-beta signaling
and Runx2, Molecular Endocrinology, 24(3) (2010) 587-597.
[112] D.G. Monroe, M.E. McGee-Lawrence, M.J. Oursler, J.J. Westendorf, Update
on Wnt signaling in bone cell biology and bone disease, Gene, 492(1) (2012)
1-18.
[113] M.K. Sutherland, J.C. Geoghegan, C. Yu, E. Turcott, J.E. Skonier, D.G.
Winkler, J.A. Latham, Sclerostin promotes the apoptosis of human
osteoblastic cells: a novel regulation of bone formation, Bone, 35(4) (2004)
828-835.
[114] S.L. Holmen, S.A. Robertson, C.R. Zylstra, B.O. Williams, Wnt-independent
activation of beta-catenin mediated by a Dkk1-Fz5 fusion protein,
Biochemical and Biophysical Research Communications, 328(2) (2005) 533-
539.
[115] Y.W. Qiang, Y. Chen, N. Brown, B. Hu, J. Epstein, B. Barlogie, J.D.
Shaughnessy, Jr., Characterization of Wnt/beta-catenin signalling in
osteoclasts in multiple myeloma, British Journal of Haematology, 148(5)
(2010) 726-738.
[116] B.R. Wong, R. Josien, S.Y. Lee, B. Sauter, H.L. Li, R.M. Steinman, Y. Choi,
TRANCE (tumor necrosis factor [TNF]-related activation-induced cytokine),
a new TNF family member predominantly expressed in T cells, is a dendritic
cell-specific survival factor, Journal of Experimental Medicine, 186(12)
(1997) 2075-2080.
[117] J. Xu, J.W. Tan, L. Huang, X.H. Gao, R. Laird, D. Liu, S. Wysocki, M.H.
Zheng, Cloning, sequencing, and functional characterization of the rat
homologue of receptor activator of NF-kappaB ligand, Journal of Bone and
Mineral Research, 15(11) (2000) 2178-2186.
[118] J. Caetano Lopes, H. Canhao, J.E. Fonseca, Osteoimmunology--the hidden
immune regulation of bone, Autoimmunity Reviews, 8(3) (2009) 250-255.
[119] E. Ang, Q. Liu, M. Qi, H.G. Liu, X. Yang, H. Chen, M.H. Zheng, J. Xu,
Mangiferin attenuates osteoclastogenesis, bone resorption, and RANKL-
induced activation of NF-kappaB and ERK, Journal of Cellular Biochemistry,
112(1) (2011) 89-97.
[120] M.T. Gillespie, Impact of cytokines and T lymphocytes upon osteoclast
differentiation and function, Arthritis Research & Therapy, 9(2) (2007) 103.
[121] J. Xu, H.F. Wu, E.S. Ang, K. Yip, M. Woloszyn, M.H. Zheng, R.X. Tan, NF-
kappaB modulators in osteolytic bone diseases, Cytokine & Growth Factor
Reviews, 20(1) (2009) 7-17.
[122] B.P. Ayati, C.M. Edwards, G.F. Webb, J.P. Wikswo, A mathematical model of
bone remodeling dynamics for normal bone cell populations and myeloma
bone disease, Biology Direct, 5 (2010) 28.
[123] E.F. Eriksen, Cellular mechanisms of bone remodeling, Rev Endocr Metab
Disord, 11(4) (2010) 219-227.
[124] Y. Wang, Q.H. Qin, A theoretical study of bone remodelling under PEMF at
cellular level, Computer methods in biomechanics and biomedical
engineering, 15(8) (2012) 885-897.
[125] X.Q. He, C. Qu, Q.H. Qin, A theoretical model for surface bone remodeling
under electromagnetic loads, Arch Appl Mech, 78(3) (2008) 163-175.
[126] C.Y. Qu, Q.H. Qin, Y.L. Kang, A hypothetical mechanism of bone remodeling
and modeling under electromagnetic loads, Biomaterials, 27(21) (2006)
4050-4057.
196
[127] J.C. Crockett, D.J. Mellis, D.I. Scott, M.H. Helfrich, New knowledge on
critical osteoclast formation and activation pathways from study of rare
genetic diseases of osteoclasts: focus on the RANK/RANKL axis,
Osteoporosis International, 22(1) (2011) 1-20.
[128] N.J. Pavlos, T.S. Cheng, A. Qin, P.Y. Ng, H.T. Feng, E.S. Ang, A. Carrello,
C.H. Sung, R. Jahn, M.H. Zheng, J. Xu, Tctex-1, a novel interaction partner
of Rab3D, is required for osteoclastic bone resorption, Molecular and
Cellular Biology, 31(7) (2011) 1551-1564.
[129] C.H. Lohmann, Z. Schwartz, Y. Liu, Z. Li, B.J. Simon, V.L. Sylvia, D.D.
Dean, L.F. Bonewald, H.J. Donahue, B.D. Boyan, Pulsed electromagnetic
fields affect phenotype and connexin 43 protein expression in MLO-Y4
osteocyte-like cells and ROS 17/2.8 osteoblast-like cells, Journal of
Orthopaedic Research, 21(2) (2003) 326-334.
[130] K. Chang, W.H.S. Chang, M.L. Wu, C. Shih, Effects of different intensities of
extremely low frequency pulsed electromagnetic fields on formation of
osteoclast-like cells, Bioelectromagnetics, 24(6) (2003) 431-439.
[131] K.J. McLeod, C.T. Rubin, The effect of low-frequency electrical fields on
osteogenesis, Journal of Bone and Joint Surgery-American Volume, 74A(6)
(1992) 920-929.
[132] M.A. Vander Molen, H.J. Donahue, C.T. Rubin, K.J. McLeod, Osteoblastic
networks with deficient coupling: Differential effects of magnetic and electric
field exposure, Bone, 27(2) (2000) 227-231.
[133] F. Tabrah, M. Hoffmeier, F. Gilbert, S. Batkin, C.A.L. Bassett, Bone-density
changes in osteoporosis-prone women exposed to pulsed electromagnetic-
fields (pemfs), Journal of Bone and Mineral Research, 5(5) (1990) 437-442.
[134] M. De Mattei, A. Caruso, G.C. Traina, F. Pezzetti, T. Baroni, V. Sollazzo,
Correlation between pulsed electromagnetic fields exposure time and cell
proliferation increase in human osteosarcoma cell lines and human normal
osteoblast cells in vitro, Bioelectromagnetics, 20(3) (1999) 177-182.
[135] D.C. Fredericks, J.V. Nepola, J.T. Baker, J. Abbott, B. Simon, Effects of
pulsed electromagnetic fields on bone healing in a rabbit tibial osteotomy
model, Journal of Orthopaedic Trauma, 14(2) (2000) 93-100.
[136] P. Diniz, K. Shomura, K. Soejima, G. Ito, Effects of pulsed electromagnetic
field (PEMF) stimulation on bone tissue like formation are dependent on the
maturation stages of the osteoblasts, Bioelectromagnetics, 23(5) (2002) 398-
405.
[137] K.J. McLeod, L. Collazo, Suppression of a differentiation response in MC-
3T3-E1 osteoblast-like cells by sustained, low-level, 30 Hz magnetic-field
exposure, Radiation Research, 153(5) (2000) 706-714.
[138] Q.-H. Qin, Mechanics of Cellular Bone Remodeling: Coupled Thermal,
Electrical, and Mechanical Field Effects, CRC Press, 2013.
[139] M. Kroll, Parathyroid hormone temporal effects on bone formation and
resorption, Bulletin of Mathematical Biology, 62(1) (2000) 163-188.
[140] C. Rattanakul, Y. Lenbury, N. Krishnamara, D.J. Wolwnd, Modeling of bone
formation and resorption mediated by parathyroid hormone: response to
estrogen/PTH therapy, Biosystems, 70(1) (2003) 55-72.
[141] S.V. Komarova, R.J. Smith, S.J. Dixon, S.M. Sims, L.M. Wahl, Mathematical
model predicts a critical role for osteoclast autocrine regulation in the control
of bone remodeling, Bone, 33(2) (2003) 206-215.
197
[142] S.V. Komarova, Mathematical model of paracrine interactions between
osteoclasts and osteoblasts predicts anabolic action of parathyroid hormone
on bone, Endocrinology, 146(8) (2005) 3589-3595.
[143] L.K. Potter, L.D. Greller, C.R. Cho, M.E. Nuttall, G.B. Stroup, L.J. Suva, F.L.
Tobin, Response to continuous and pulsatile PTH dosing: A mathematical
model for parathyroid hormone receptor kinetics, Bone, 37(2) (2005) 159-
169.
[144] V. Lemaire, F.L. Tobin, L.D. Greller, C.R. Cho, L.J. Suva, Modeling the
interactions between osteoblast and osteoclast activities in bone remodeling,
Journal of Theoretical Biology, 229(3) (2004) 293-309.
[145] Y.N. Wang, Q.H. Qin, S. Kalyanasundaram, A theoretical model for
simulating effect of Parathyroid Hormone on bone metabolism at cellular
level, Journal of Molecular and Cellular Biomechanics., 6(2) (2009) 101-112.
[146] P. Pivonka, J. Zimak, D.W. Smith, B.S. Gardiner, C.R. Dunstan, N.A. Sims, T.
John Martin, G.R. Mundy, Model structure and control of bone remodeling:
A theoretical study, Bone, 43(2) (2008) 249-263.
[147] P. Pivonka, J. Zimak, D.W. Smith, B.S. Gardiner, C.R. Dunstan, N.A. Sims, T.
John Martin, G.R. Mundy, Theoretical investigation of the role of the RANK-
RANKL-OPG system in bone remodeling, Journal of Theoretical Biology,
262(2) (2010) 306-316.
[148] H. Wang, Q.H. Qin, FE approach with Green’s function as internal trial
function for simulating bioheat transfer in the human eye, Archives of
Mechanics, 62(6) (2010) 493-510.
[149] H. Wang, Q.H. Qin, Hybrid FEM with fundamental solutions as trial functions
for heat conduction simulation, Acta Mechanica Solida Sinica, 22(5) (2009)
487-498.
[150] Q.H. Qin, H. Wang, Matlab and C programming for Trefftz finite element
methods, CRC Press, 2008.
[151] Q.H. Qin, Trefftz finite element method and its applications, Applied
Mechanics Reviews, 58(5) (2005) 316-337.
[152] Q.H. Qin, Variational formulations for TFEM of piezoelectricity, International
Journal of Solids and Structures, 40(23) (2003) 6335-6346.
[153] Q. Qin, Hybrid-Trefftz finite element method for Reissner plates on an elastic
foundation, Computer Methods in Applied Mechanics and Engineering,
122(3-4) (1995) 379-392.
[154] Q. Qin, Hybrid Trefftz finite-element approach for plate bending on an elastic
foundation, Applied Mathematical Modelling, 18(6) (1994) 334-339.
[155] H. Weinans, R. Huiskes, H.J. Grootenboer, The Behavior of Adaptive Bone-
Remodeling Simulation-Models, Jounral of Biomechanics, 25(12) (1992)
1425-1441.
[156] D. Kardas, U. Nackenhorst, Studies on Bone Remodeling Theory Based on
Microcracks Using Finite Element Computations, PAMM, 9(1) (2009) 147-
148.
[157] C.L. Lin, Y.H. Lin, S.H. Chang, Multi-factorial analysis of variables
influencing the bone loss of an implant placed in the maxilla: Prediction
using FEA and SED bone remodeling algorithm, Jounral of Biomechanics,
43(4) (2010) 644-651.
[158] R. Hambli, H. Katerchi, C.-L. Benhamou, Multiscale methodology for bone
remodelling simulation using coupled finite element and neural network
198
computation, Biomechanics and Modeling in Mechanobiology, 10(1) (2011)
133-145.
[159] Q.H. Qin, S.W. Yu, An arbitrarily-oriented plane crack terminating at the
interface between dissimilar piezoelectric materials, International Journal of
Solids and Structures, 34(5) (1997) 581-590.
[160] Q.H. Qin, Y.W. Mai, S.W. Yu, Some problems in plane thermopiezoelectric
materials with holes, International Journal of Solids and Structures, 36(3)
(1999) 427-439.
[161] Q.H. Qin, Y.W. Mai, S.W. Yu, Effective moduli for thermopiezoelectric
materials with microcracks, International Journal of Fracture, 91(4) (1998)
359-371.
[162] Q.H. Qin, Y.W. Mai, A closed crack tip model for interface cracks
inthermopiezoelectric materials, International Journal of Solids and
Structures, 36(16) (1999) 2463-2479.
[163] Q.H. Qin, Y.W. Mai, Thermoelectroelastic Green's function and its application
for bimaterial of piezoelectric materials, Archive of Applied Mechanics,
68(6) (1998) 433-444.
[164] Q.H. Qin, Thermoelectroelastic Green's function for a piezoelectric plate
containing an elliptic hole, Mechanics of Materials, 30(1) (1998) 21-29.
[165] Q. Qin, Y. Mai, Crack growth prediction of an inclined crack in a half-plane
thermopiezoelectric solid, Theoretical and Applied Fracture Mechanics, 26(3)
(1997) 185-191.
[166] S. Scheiner, P. Pivonka, C. Hellmich, Coupling systems biology with
multiscale mechanics, for computer simulations of bone remodeling,
Computer Methods in Applied Mechanics and Engineering, 254(0) (2013)
181-196.
[167] J.B. Lian, G.S. Stein, Concepts of osteoblast growth and differentiation: basis
for modulation of bone cell development and tissue formation, Critical
Reviews in Oral Biology & Medicine, 3(3) (1992) 269-305.
[168] T.A. Owen, M. Aronow, V. Shalhoub, L.M. Barone, L. Wilming, M.S.
Tassinari, M.B. Kennedy, S. Pockwinse, J.B. Lian, G.S. Stein, Progressive
development of the rat osteoblast phenotype in vitro: Reciprocal relationships
in expression of genes associated with osteoblast proliferation and
differentiation during formation of the bone extracellular matrix, Journal of
Cellular Physiology, 143(3) (1990) 420-430.
[169] M. Hartig, U. Joos, H.P. Wiesmann, Capacitively coupled electric fields
accelerate proliferation of osteoblast-like primary cells and increase bone
extracellular matrix formation in vitro, Eur Biophys J, 29(7) (2000) 499-506.
[170] N. Bohr, Atomic physics and human knowledge, Courier Dover Publications,
2010.
[171] H.M. Shapiro, Input-output models of biological systems: formulation and
applicability, Computers and Biomedical Research, 2(5) (1969) 430-445.
[172] V.Z. Marmarelis, Identification of nonlinear biological systems using Laguerre
expansions of kernels, Annals of Biomedical Engineering, 21(6) (1993) 573-
589.
[173] M.J. Korenberg, I.W. Hunter, The identification of nonlinear biological
systems: Volterra kernel approaches, Annals of Biomedical Engineering,
24(2) (1996) 250-268.
[174] E.F. Keller, L.A. Segel, Model for chemotaxis, Journal of Theoretical Biology,
30(2) (1971) 225-234.
199
[175] T. Hillen, K. Painter, Global existence for a parabolic chemotaxis model with
prevention of overcrowding, Advances in Applied Mathematics, 26(4) (2001)
280-301.
[176] R.W. Li, N.T. Kirkland, J. Truong, J. Wang, P.N. Smith, N. Birbilis, D.R.
Nisbet, The influence of biodegradable magnesium alloys on the osteogenic
differentiation of human mesenchymal stem cells, Journal of Biomedical
Materials Research Part A 102(12) (2014) 4346-4357.
[177] A. Arifin, A.B. Sulong, N. Muhamad, J. Syarif, M.I. Ramli, Material
processing of hydroxyapatite and titanium alloy (HA/Ti) composite as
implant materials using powder metallurgy: a review, Materials & Design, 55
(2014) 165-175.
[178] Z. Ma, Z. Mao, C. Gao, Surface modification and property analysis of
biomedical polymers used for tissue engineering, Colloids and Surfaces B:
Biointerfaces, 60(2) (2007) 137-157.
[179] D.D. Deligianni, N.D. Katsala, P.G. Koutsoukos, Y.F. Missirlis, Effect of
surface roughness of hydroxyapatite on human bone marrow cell adhesion,
proliferation, differentiation and detachment strength, Biomaterials, 22(1)
(2000) 87-96.
[180] T.J. Webster, C. Ergun, R.H. Doremus, R.W. Siegel, R. Bizios, Enhanced
functions of osteoblasts on nanophase ceramics, Biomaterials, 21(17) (2000)
1803-1810.
[181] T.J. Webster, R.W. Siegel, R. Bizios, Osteoblast adhesion on nanophase
ceramics, Biomaterials, 20(13) (1999) 1221-1227.
[182] R. Huang, S. Lu, Y. Han, Role of grain size in the regulation of osteoblast
response to Ti–25Nb–3Mo–3Zr–2Sn alloy, Colloids and Surfaces B:
Biointerfaces, 111 (2013) 232-241.
[183] M. Vandrovcová, L. Bacakova, Adhesion, growth and differentiation of
osteoblasts on surface-modified materials developed for bone implants,
Physiological Research, 60(3) (2011) 403-417.
[184] D.L. Elbert, J.A. Hubbell, Surface Treatments of Polymers for
Biocompatibility, Annual Review of Materials Science, 26(1) (1996) 365-
394.
[185] K. Anselme, Osteoblast adhesion on biomaterials, Biomaterials, 21(7) (2000)
667-681.
[186] K.C. Dee, T.T. Andersen, R. Bizios, Osteoblast population migration
characteristics on substrates modified with immobilized adhesive peptides,
Biomaterials, 20(3) (1999) 221-227.
[187] K. Anselme, L. Ploux, A. Ponche, Cell/Material Interfaces: Influence of
Surface Chemistry and Surface Topography on Cell Adhesion, Journal of
Adhesion Science and Technology, 24(5) (2010) 831-852.
[188] A. Stevens, H.G. Othmer, Aggregation, blowup, and collapse: the ABC's of
taxis in reinforced random walks, SIAM Journal on Applied Mathematics,
57(4) (1997) 1044-1081.
[189] Wheeless' Textbook of Orthopaedics, Duke University Medical Center's
Division of Orthopedic Surgery, Data Trace Internet Publishing, 1996.
[190] J. Park, S. Bauer, K.A. Schlegel, F.W. Neukam, K. von der Mark, P. Schmuki,
TiO2 Nanotube Surfaces: 15 nm—An Optimal Length Scale of Surface
Topography for Cell Adhesion and Differentiation, Small, 5(6) (2009) 666-
671.
200
[191] S. Minagar, J. Wang, C.C. Berndt, E.P. Ivanova, C. Wen, Cell response of
anodized nanotubes on titanium and titanium alloys, Journal of Biomedical
Materials Research Part A, 101A(9) (2013) 2726-2739.
[192] M. Sato, A. Aslani, M.A. Sambito, N.M. Kalkhoran, E.B. Slamovich, T.J.
Webster, Nanocrystalline hydroxyapatite/titania coatings on titanium
improves osteoblast adhesion, Journal of Biomedical Materials Research Part
A, 84A(1) (2008) 265-272.
[193] E.K.F. Yim, K.W. Leong, Significance of synthetic nanostructures in dictating
cellular response, Nanomedicine: Nanotechnology, Biology and Medicine,
1(1) (2005) 10-21.
[194] B. Stevens, Y. Yang, A. Mohandas, B. Stucker, K.T. Nguyen, A review of
materials, fabrication methods, and strategies used to enhance bone
regeneration in engineered bone tissues, Journal of Biomedical Materials
Research Part B: Applied Biomaterials: An Official Journal of The Society
for Biomaterials, The Japanese Society for Biomaterials, and The Australian
Society for Biomaterials and the Korean Society for Biomaterials, 85(2)
(2008) 573-582.
[195] K.J. Painter, T. Hillen, Volume-filling and quorum-sensing in models for
chemosensitive movement, The Quarterly Journal of Mechanics and Applied
Mathematics, 10(4) (2002) 501-543.
[196] S. Chen, C.Y. Lee, R.W. Li, P.N. Smith, Q.H. Qin, Modelling osteoblast
adhesion on surface-engineered biomaterials: optimisation of nanophase grain
size, Computer Methods in Biomechanics and Biomedical Engineering, 20(8)
(2017) 905-914.
[197] D.J. Muehsam, A.A. Pilla, A Lorentz model for weak magnetic field
bioeffects: Part I—Thermal noise is an essential component of AC/DC effects
on bound ion trajectory, Bioelectromagnetics, 30(6) (2009) 462-475.
[198] C.L.M. Bauréus Koch, M. Sommarin, B.R.R. Persson, L.G. Salford, J.L.
Eberhardt, Interaction between weak low frequency magnetic fields and cell
membranes, Bioelectromagnetics, 24(6) (2003) 395-402.
[199] J. Miyakoshi, Effects of static magnetic fields at the cellular level, Progress in
Biophysics and Molecular Biology, 87(2) (2005) 213-223.
[200] S. Yamaguchi-Sekino, M. Sekino, S. Ueno, Biological effects of
electromagnetic fields and recently updated safety guidelines for strong static
magnetic fields, Magnetic Resonance in Medical Sciences, 10(1) (2011) 1-10.
[201] V. Lednev, Possible mechanism for the influence of weak magnetic fields on
biological systems, Bioelectromagnetics, 12(2) (1991) 71-75.
[202] J. Zhang, C. Ding, L. Ren, Y. Zhou, P. Shang, The effects of static magnetic
fields on bone, Progress in Biophysics and Molecular Biology, 114(3) (2014)
146-152.
[203] T. Lemaire, E. Capiez-Lernout, J. Kaiser, S. Naili, V. Sansalone, What is the
importance of multiphysical phenomena in bone remodelling signals
expression? A multiscale perspective, Journal of the Mechanical Behavior of
Biomedical Materials, 4(6) (2011) 909-920.
[204] R.C. Riddle, H.J. Donahue, From streaming-potentials to shear stress: 25 years
of bone cell mechanotransduction, Journal of Orthopaedic Research, 27(2)
(2009) 143-149.
[205] S. Ghodbane, A. Lahbib, M. Sakly, H. Abdelmelek, Bioeffects of static
magnetic fields: oxidative stress, genotoxic effects, and cancer studies,
BioMed Research International, 2013 (2013).
201
[206] P.J. Hore, Are biochemical reactions affected by weak magnetic fields?,
Proceedings of the National Academy of Sciences, 109(5) (2012) 1357-1358.
[207] H.B. Murray, B.A. Pethica, A follow-up study of the in-practice results of
pulsed electromagnetic field therapy in the management of nonunion
fractures, Orthopedic Research and Reviews, 55 (2016) 67-72.
[208] X. Bai, D. Miao, D. Panda, S. Grady, M.D. McKee, D. Goltzman, A.C.
Karaplis, Partial rescue of the Hyp phenotype by osteoblast-targeted PHEX
(phosphate-regulating gene with homologies to endopeptidases on the X
chromosome) expression, Molecular Endocrinology, 16(12) (2002) 2913-
2925.
[209] Q.H. Qin, Y.N. Wang, A mathematical model of cortical bone remodeling at
cellular level under mechanical stimulus, Acta Mechanica Sinica, 28(6)
(2012) 1678-1692.
202
Appendix
203
#An example of programming SMF data in MATLAB
clear all close all
% raw data raw_x=[20 26 32 56 97]'; raw_y=[2600 2500 2800 1900 2000]'; raw_e=[200 100 450 350 200]';
%N N=500; %R R=(0:200)';
%del_t del_t=60; %final_t final_t=60000;%1000 steps %t vector t=(0:del_t:final_t)'; %h h=1/N; %xi i=(0:N)'; x=i*h; %cellular sensitivity of ti a=0.17+0.11*exp(-0.027*R); %molecular sensitivity of ti d=1.37-1.29*exp(-0.02*R); %initial cell density A=1800;
%rescaling: e.g. final projected time = n*(del_t*coe)-
>myscale=del_t*coe final_projected_time=8; myscale=final_projected_time/length(t);
%preallocation p=zeros(size(t,1),size(x,1)); g=zeros(size(t,1),size(x,1)); f=zeros(size(t,1),size(x,1));
%t=0 initialisation %pi(0) var=1./d;
%% clear p g f p{1}=zeros(length(var),length(x)); g{1}=zeros(length(var),length(x)); f{1}=zeros(length(var),length(x)); for i=1:length(var) p{1}(i,:)=sqrt(var(i)/2./pi).*exp(-0.5.* var(i)*x.^2); end %g and f for i=1:length(x) switch 1 case i==1 j=3;
204
g{1}(:,i)=p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-
1).*(1-a.*p{1}(:,j-2)); f{1}(:,i)=(1-p{1}(:,j+1)).*(1-a.*p{1}(:,j-
1))+p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-1).*(1-a.*p{1}(:,j-
2))+(1-p{1}(:,j-1)).*(1-a.*p{1}(:,j+1)); case i==2 j=3; g{1}(:,i)=p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-
1).*(1-a.*p{1}(:,j-2)); f{1}(:,i)=(1-p{1}(:,j+1)).*(1-a.*p{1}(:,j-
1))+p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-1).*(1-a.*p{1}(:,j-
2))+(1-p{1}(:,j-1)).*(1-a.*p{1}(:,j+1)); case i==length(x)-1 j=length(x)-2; g{1}(:,i)=p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-
1).*(1-a.*p{1}(:,j-2)); f{1}(:,i)=(1-p{1}(:,j+1)).*(1-a.*p{1}(:,j-
1))+p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-1).*(1-a.*p{1}(:,j-
2))+(1-p{1}(:,j-1)).*(1-a.*p{1}(:,j+1)); case i==length(x) j=length(x)-2; g{1}(:,i)=p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-
1).*(1-a.*p{1}(:,j-2)); f{1}(:,i)=(1-p{1}(:,j+1)).*(1-a.*p{1}(:,j-
1))+p{1}(:,j+1).*(1-a.*p{1}(:,j+2))+p{1}(:,j-1).*(1-a.*p{1}(:,j-
2))+(1-p{1}(:,j-1)).*(1-a.*p{1}(:,j+1)); case i>2 && i<length(x)-1 g{1}(:,i)=p{1}(:,i+1).*(1-a.*p{1}(:,i+2))+p{1}(:,i-
1).*(1-a.*p{1}(:,i-2)); f{1}(:,i)=(1-p{1}(:,i+1)).*(1-a.*p{1}(:,i-
1))+p{1}(:,i+1).*(1-a.*p{1}(:,i+2))+p{1}(:,i-1).*(1-a.*p{1}(:,i-
2))+(1-p{1}(:,i-1)).*(1-a.*p{1}(:,i+1)); end end %g and f along with time index k for k=1:size(t,1)-1 for i=1:length(x) switch 1 case i==1 j=3;
g{k}(:,i)=p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-
1).*(1-a.*p{k}(:,j-2)); f{k}(:,i)=(1-p{k}(:,j+1)).*(1-a.*p{k}(:,j-
1))+p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-1).*(1-a.*p{k}(:,j-
2))+(1-p{k}(:,j-1)).*(1-a.*p{k}(:,j+1)); case i==2 j=3; g{k}(:,i)=p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-
1).*(1-a.*p{k}(:,j-2)); f{k}(:,i)=(1-p{k}(:,j+1)).*(1-a.*p{k}(:,j-
1))+p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-1).*(1-a.*p{k}(:,j-
2))+(1-p{k}(:,j-1)).*(1-a.*p{k}(:,j+1)); case i==length(x)-1 j=length(x)-2; g{k}(:,i)=p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-
1).*(1-a.*p{k}(:,j-2)); f{k}(:,i)=(1-p{k}(:,j+1)).*(1-a.*p{k}(:,j-
1))+p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-1).*(1-a.*p{k}(:,j-
2))+(1-p{k}(:,j-1)).*(1-a.*p{k}(:,j+1)); case i==length(x)
205
j=length(x)-2; g{k}(:,i)=p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-
1).*(1-a.*p{k}(:,j-2)); f{k}(:,i)=(1-p{k}(:,j+1)).*(1-a.*p{k}(:,j-
1))+p{k}(:,j+1).*(1-a.*p{k}(:,j+2))+p{k}(:,j-1).*(1-a.*p{k}(:,j-
2))+(1-p{k}(:,j-1)).*(1-a.*p{k}(:,j+1)); case i>2 && i<length(x)-1 g{k}(:,i)=p{k}(:,i+1).*(1-a.*p{k}(:,i+2))+p{k}(:,i-
1).*(1-a.*p{k}(:,i-2)); f{k}(:,i)=(1-p{k}(:,i+1)).*(1-a.*p{k}(:,i-
1))+p{k}(:,i+1).*(1-a.*p{k}(:,i+2))+p{k}(:,i-1).*(1-a.*p{k}(:,i-
2))+(1-p{k}(:,i-1)).*(1-a.*p{k}(:,i+1)); end
p{k+1}(:,i)=(h.^2.*p{k}(:,i)+del_t.*g{k}(:,i))./(h.^2+del_t.*f{k}(:,
i)); end
end
clear N N=zeros(length(t),length(d)); for i=1:length(t)
%Trapezoidal numerical integration b1=repmat(transpose(A.*2.*pi.*x(1:end-
1)),length(d),1).*p{i}(:,1:end-1);
b2=repmat(transpose(A.*2.*pi.*x(2:end)),length(d),1).*p{i}(:,2:end); N(i,:)=sum(h.*(b2+b1)/2,2);
end
myh=surf(N); set(myh,'EdgeColor','none') figure plot(N(300,:),'k-') hold on errorbar(raw_x,raw_y,raw_e,'b.','MarkerSize',20);