Conceived and designed the experiments: NCN PJP PM. Performed the experiments: NCN PM. Analyzed the data: NCN PJP PM. Contributed reagents/materials/analysis tools: NCN PJP PM. Wrote the paper: NCN PJP PM.
The authors have declared that no competing interests exist.
Although it is known that mechanical forces are needed for normal bone development, the current understanding of how biophysical stimuli are interpreted by and integrated with genetic regulatory mechanisms is limited. Mechanical forces are thought to be mediated in cells by “mechanosensitive” genes, but it is a challenge to demonstrate that the genetic regulation of the biological system is dependant on particular mechanical forces in vivo. We propose a new means of selecting candidate mechanosensitive genes by comparing in vivo gene expression patterns with patterns of biophysical stimuli, computed using finite element analysis. In this study, finite element analyses of the avian embryonic limb were performed using anatomically realistic rudiment and muscle morphologies, and patterns of biophysical stimuli were compared with the expression patterns of four candidate mechanosensitive genes integral to bone development. The expression patterns of two genes, Collagen X (ColX) and Indian hedgehog (Ihh), were shown to colocalise with biophysical stimuli induced by embryonic muscle contractions, identifying them as potentially being involved in the mechanoregulation of bone formation. An altered mechanical environment was induced in the embryonic chick, where a neuromuscular blocking agent was administered in ovo to modify skeletal muscle contractions. Finite element analyses predicted dramatic changes in levels and patterns of biophysical stimuli, and a number of immobilised specimens exhibited differences in ColX and Ihh expression. The results obtained indicate that computationally derived patterns of biophysical stimuli can be used to inform a directed search for genes that may play a mechanoregulatory role in particular in vivo events or processes. Furthermore, the experimental data demonstrate that ColX and Ihh are involved in mechanoregulatory pathways and may be key mediators in translating information from the mechanical environment to the molecular regulation of bone formation in the embryo.
While mechanical forces are known to be critical to adult bone maintenance and repair, the importance of mechanobiology in embryonic bone formation is less widely accepted. The influence of mechanical forces on cells is thought to be mediated by “mechanosensitive genes,” genes which respond to mechanical stimulation. In this research, we examined the situation in the developing embryo. Using finite element analysis, we simulated the biophysical stimuli in the developing bone resulting from spontaneous muscle contractions, incorporating detailed morphology of the developing chick limb. We compared patterns of stimuli with expression patterns of a number of genes involved in bone formation and demonstrated a clear colocalisation in the case of two genes (Ihh and ColX). We then altered the mechanical environment of the growing chick embryo by blocking muscle contractions and demonstrated changes in the magnitudes and patterns of biophysical stimuli and in the expression patterns of both Ihh and ColX. We have demonstrated the value of combining computational techniques with in vivo gene expression analysis to identify genes that may play a mechanoregulatory role and have identified genes that respond to mechanical stimulation during bone formation in vivo.
It is widely accepted that there is a relationship between the morphology of skeletal
structures and the mechanical forces acting upon them. Such a relationship begins in
the embryo where the importance of muscle for normal bone formation has been clearly
demonstrated
It is, however, more challenging to examine candidate genes in an in vivo context. To
date, the study of Kavanagh et al.
Considering embryonic bone formation specifically, a number of genes involved in key
steps have been identified as mechanosensitive in in vitro cell culture assays
In this paper, we hypothesise that mechanical forces influence embryonic bone
formation by regulating expression of mechanosensitive genes. To test this
hypothesis, the involvement of four genes in transducing mechanical information from
spontaneous muscle contractions during ossification was assessed; these are ColX,
FGFr2, Ihh and PTHrP. The genes were selected for this study based on their
importance for bone formation and evidence of their mechanosensitivity in vitro.
Using a novel approach, the potential in vivo mechanosensitivity of these genes is
initially assessed using computationally derived data on the biophysical
environment. The candidate genes were first examined by correlating their expression
patterns with patterns of biophysical stimuli across stages of development when
ossification begins. We carried out a detailed analysis of expression of the 4
candidate genes and, by using the results of finite element analyses based on 3-D
rudiment morphologies and realistic muscle loading schemes described in a previous
paper
Morphological and gene expression analyses were carried out on the tibiotarsal
rudiment in the hindlimb of the embryonic chick. Dissected embryos were staged
according to the Hamburger and Hamilton (HH) system
The BBSRC (Biotechnology and Biological Sciences Research Council, U.K.) ChickEST
Database (
Each EST clone was sequenced to verify identity. Plasmid DNA carrying the EST of interest was linearized with appropriate restriction enzymes (EcoR1 or Not1). Antisense and sense digoxigenin-labelled RNA probes were transcribed in vitro from 1 µg of linearized plasmid using T7 and T3 promoter sites (according to insert orientation) in the pBluescript II KS+ vector (all components for in vitro transcription from Roche, Germany). DNA template was degraded by incubation of probes with RNase free DNase (Roche). The probes were then purified on G25 columns (Amersham Biosciences, USA) according to the manufacturer's instructions. Probe concentrations were determined by spectophotometry and probes were stored at −20°C.
After dissection, limbs selected for in situ hybridisation were fixed in 4% paraformaldehyde (PFA) in PBS over night, and dehydrated through a series of methanol/PBT (PBT = 0.1% Triton X-100 in PBS; 25, 50, 75%; 1×10 minute) washes, followed by 2×10 minutes in 100% methanol and stored at −20°C in 30 or 50 ml tubes until needed. On the morning of sectioning, limbs were re-hydrated through a series of methanol/PBT (75, 50, 25%; 1×10 minute) washes at 4°C. After 2×10 minutes washes in PBT, excess tissue surrounding the skeletal rudiments was removed in order to give optimal sectioning performance. The specimens were embedded in 4% Low Melting Agarose/PBS (Invitrogen, UK). 80 or 100 µm sections were cut in the longitudinal direction with a vibrating microtome (VT1000S, Leica) and stored in PBS in 12-well plates.
After 2×10 minute washes in PBT, free-floating sections were treated with proteinase K (20 µg/ml in PBT) for 5 minutes at room temperature. Sections were then washed twice in PBT and fixed for 20 minutes in 0.2% glutaraldehyde/4% paraformaldehyde (PFA). Fixation was followed by washes (3×5 minutes) in PBT at room temperature, and a further 30 minute PBT wash at 55°C. The sections were then prehybridised at 55°C overnight in a hybridization solution containing 2% blocking reagent (Roche), 50% formamide, 5× SSC (Saline-sodium citrate buffer), 0.5% 3-[(3-Cholamidopropyl-[(3-Cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), 500 µg/ml Heparin, 1 µg/ml Yeast RNA, 0.1% Tween 20 and 5 mM EDTA (ethylenediamine tetraacetic acid) (all components from Sigma, UK, unless otherwise stated). Antisense and sense probes were denatured at 80°C for 3 minutes and sections were then incubated at 55°C over 2–3 nights in hybridization solution containing either antisense or sense probe at minimum concentrations of 2 ng/µl.
Post-hybridization washes were carried out at 60°C as follows: 2×10 minutes in 2× SSC; 3×20 minutes in 2× SSC/0.1% CHAPS; 3×20 minutes in 0.2× SSC/0.1% CHAPS. The sections were then washed for 2×10 minutes in TNT (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Tween 20) at room temperature and blocked in blocking buffer (0.1 M maleic acid, 0.15 M NaCl, 3% blocking reagent (Roche)) plus 10% goat serum overnight at 4°C. Sections were incubated overnight in fresh blocking buffer (plus 10% serum) containing a 1∶1000 dilution of anti-digoxigenin Fab fragments conjugated with alkaline phosphatase (Roche) at 4°C, with rocking. The sections were then washed (5×1 hour) at room temperature in TNT and left rocking in TNT over 2 nights at 4°C. On the day the signal was developed, sections were washed in 3 changes of NMT (100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl2) for 15 minutes each. The chromogenic reaction was carried out in NMT containing 17.5 µg/ml 4-nitro blue tetrazolium chloride (NBT; Roche) and 6.25 µg/ml 5-bromo-4-chloro-3-indolyl-phophate (BCIP; Roche). Sections were developed in the dark at room temperature with rocking for 6–8 hours and then fixed in 4% PFA/PBS for 1 hour before mounting on slides with Aquapolymount (Polysciences, Inc).
Two sets of immobilisation experiments were performed at different timepoints;
named Set A and Set B. In Set A, 120 eggs were assigned as experimental embryos,
and 80 as controls, while in Set B, 100 eggs were assigned as experimental
embryos, and 80 as controls. The eggs were incubated for 3 days, after which 4
ml of albumen was removed with a syringe so that the embryo would sink lower in
the egg and a window could be cut in the shell without rupturing the
chorioallantoic membrane. Administration of the neuromuscular blocking agent
Decamethonium Bromide (DMB)
All harvested embryos were stained to reveal cartilage and bone using Alcian Blue
(cartilage) and Alizarin Red (bone) using a modification of the protocol of
Hogan et al.
Control Embryos | Experimental Embryos | Total | |
Day 8 | 13 | 23 | 36 |
Day 9 | 5 | 10 | 15 |
Day 11 | 30 | 25 | 55 |
As described in detail in Nowlan et al.
In the normal analysis, muscles on the ventral aspect are active during the flexion contraction and muscles on the dorsal are active in the extension contraction. In the experimental situation, both sets of muscles are activated at the same time, at 75% of the load magnitudes of the normal situation.
Collagen X (ColX) expression was found in the region of hypertrophic
chondrocytes in the internal cartilage of the tibiotarsal rudiments, and
also in the perichondrium and periosteum (where present) at all three stages
examined (
(A) Avian hindlimb at HH34 stained with Alcian Blue to highlight cartilage, (B–D) ColX expression patterns in sections of the avian tibiotarsus at stages HH30, HH32 and HH34. ColX is expressed in the hypertrophic chondrocytes (white arrowheads) and in the perichondrium/periosteum (black arrowheads). (D) The approximate location of the bone collar is indicated with a green line. (E) ColX expression in HH34 metatarsal rudiments showing bands of expression in the perichondrium (arrows). Up is distal, down is proximal. Scale bars 1 mm.
FGFr2 was found to be expressed in the perichondrium and in the periarticular
cartilage of the tibiotarsus at all stages, and also in the periosteum at
HH34, as shown in
Up is distal, down is proximal. tt: tibiotarsus, mt: metatarsals. The approximate location of the bone collar is indicated with a green line. FGFr2 is expressed in the perichondrium and periosteum (black arrowheads), and in the periarticular cartilage (white arrowheads). Scale bars 1 mm.
Ihh was found to be expressed in bands of pre-hypertrophic chondrocytes
within the cartilage core of the tibiotarsus for the three stages examined
(
Up is distal, down is proximal. tt: tibiotarsus, mt: metatarsals. The approximate location of the bone collar is indicated with a green line. Ihh is expressed in the hypertrophic (HH30) and pre-hypertrophic (HH30–32) zones (black arrowheads), elevated expression at the periphery highlighted with white arrowheads. Scale bars 1 mm.
PTHrP expression was evident in the periarticular regions of the rudiments of
the hindlimb at stages HH30 and HH32, as highlighted with arrows in
Up is distal, down is proximal. tt: tibiotarsus, mt: metatarsals. PTHrP is expressed in the peri-articular cartilage (arrows). Scale bars 1 mm.
The expression patterns of ColX, FGFr2, Ihh and PTHrP illustrated in
PTHrP (green) expressed in peri-articular cartilage. Ihh (red) expressed in pre-hypertrophic chondrocytes. ColX (black) expressed in hypertrophic chondrocytes (diffuse colouring), and in perichondrium/periosteum (dashed black line). (Right) Comparison of fluid velocity and maximum principal strain for normal embryos, mid-flexion (red solid line) and immobilised embryos in rigid paralysis (dashed blue line) at HH30, 32 and 34, along ventral and dorsal paths. Section shown is mid-line longitudinal section through the rudiment.
In order to verify that the conditions used to alter the mechanical
environment had an effect on skeletal development, the skeletons of controls
and embryos treated with the neuromuscular blocking agent DMB were compared
with particular focus on the tibiotarsus. The immobilisation treatment was
found to have a dramatic effect on overall skeletal morphology. Treated
embryos were smaller, with abnormal rib formation, joint contractures and
spinal curvature (results not shown); effects which had previously been
reported in other immobilisation studies
When the state of rigid paralysis induced by treatment with DMB was simulated
in Finite Element Analysis, dramatic differences in pattern and magnitude
were observed when compared with the results from the normal skeletal
rudiments, as described in Nowlan et al.
Due to the correlation with patterns of biophysical stimuli, ColX and Ihh were chosen for comparison between control and immobilized specimens at the mid timepoint of the experiment (day 9, roughly HH33). Each gene was examined in seven treated specimens and four control specimens; (of the ten experimental and five control specimens at day 9, three were damaged in the sectioning process). The analysis focussed on day 9 because, as it coincides with an early stage in the ossification process, it maximised the chances of seeing an effect on the candidate mechanosensitive genes.
Scale bars 1 mm. The extension of expression in the hypertrophic region is demarcated by white arrowheads; expression in the periosteum and perichondrium is highlighted by black arrow heads. B–C: ColX staining in the perichondrium is more restricted proximo-distally and does not extend beyond the hypertrophic zone.
Scale bar 1 mm. Expression in pre-hypertrophic chondrocytes is demarcated by arrowheads.
In this study, we set out to test the hypothesis that mechanical forces influence
embryonic bone formation by regulating certain mechanosensitive genes. In a first
analysis, the expression patterns of four genes; ColX, FGFr2, Ihh and PTHrP, were
characterised and compared with patterns of biophysical stimuli. ColX and Ihh
expression patterns correlated with stage-matched patterns of biophysical stimuli,
whereas FGFr2 and PTHrP expression patterns did not. This identified ColX and Ihh as
potential mechanosensitive genes regulating ossification in the embryo. ColX and Ihh
expression patterns followed the same dynamic sequence of events as the patterns of
biophysical stimuli, with one peak of expression at the mid-diaphysis at the
youngest stage (HH30), and two peaks progressively more proximal and distal to the
mid-diaphysis at HH32 and HH34. The ColX expression at the surface (on the
perichondrium) correlates with the locations of peak biophysical stimuli also at the
surface, while Ihh expression in the pre-hypertrophic cartilage is at the same
longitudinal position in the rudiment as, and adjacent to, the peak levels of
biophysical stimuli. In order to corroborate the hypothesis that ColX and Ihh may
act as mechanosensitive genes for bone formation in the chick limb, an
immobilisation assay was established, where rigid paralysis was induced with the
prevention of skeletal muscle contractions. The morphological analysis of the
immobilised embryos clearly demonstrated the effect of an altered mechanical
environment on skeletal development, with immobilisation leading to shorter
tibiotarsi and decreased bone collar formation. Finite Element Analyses of skeletal
elements under rigid paralysis indicated a dramatic alteration in patterns of
biophysical stimuli both in terms of stage-dependant patterns of biophysical stimuli
and magnitudes of stimuli in comparison with the normal case. Aspects of the
expression of ColX and Ihh indeed showed altered expression patterns following
immobilisation in a proportion of specimens; (see
The identification of Ihh as mechanosensitive in vivo is of particular interest since
this gene has been shown to be a key regulator of bone formation in the mouse, and
in particular formation of the bone collar,
In this study, there was a certain amount of variability in the effect of the
neuromuscular blocking agent, and the change in expression patterns of candidate
mechanosensitive genes were not seen in all immobilised (drug-treated) specimens.
This variability is not unexpected since the alteration to muscle contractions is
effected by exposure to a pharmaceutical agent where the response to a set dose can
vary across individual specimens. A variable response was also evident when movement
in the experimental embryos was quantified; while movement was clearly reduced, it
was not completely removed in all specimens. However, detectable changes in gene
expression were seen for two different genes in multiple specimens, showing a
repeatable effect, and the statistically significant decrease in rudiment length and
bone formation serves as confirmation of the immobilisation treatment as a means of
altering the mechanical environment. The magnitudes of the muscle loads applied for
the embryos subjected to rigid paralysis may be an overestimation, because while we
have assumed the same volume of muscle in our simulations, it has been widely
reported that muscle mass is reduced in immobilised embryos
The study presented here has revealed the alteration of gene expression as a result
of mechanical stimulation. Even though we have identified the in vivo
mechanosensitivity of two genes in the developing limb, we do not know what
signalling cascades prompted the change in ColX and Ihh expression patterns. For
example, focussing on Ihh in particular, while it has been suggested that Ihh
regulates proliferation of chondrocytes through the activation of stretch activated
channels by mechanical stimulation
Many researchers have recognized the importance of the interaction between mechanical
and biological factors for bone development. A range of biophysical stimuli
parameters have been hypothesised to promote ossification, such as low levels of
hydrostatic stress and principal strain
The work presented here has provided a new insight into mechanoregulation of embryonic long bone ossification. This is the first study where finite element analyses of the embryonic limb using anatomically accurate rudiment and muscle morphologies have enabled comparison of predicted biophysical stimuli patterns with gene expression patterns, and the characterisation of the biophysical environment in the growing rudiment when skeletal muscle contractions are prevented. A means of corroborating candidate mechanosensitive genes was proposed and tested, revealing ColX and Ihh as mechanosensitive in vivo during embryonic bone formation, and also identifying them as potential key mediators in translating information from the mechanical environment to the molecular regulation of bone formation in the embryo.
The authors are grateful to Ms. Suzanne Miller and Dr. Kevin Mitchell, Genetics Department, Trinity College Dublin, and Ms. Kristen Summerhurst from the Zoology Department, Trinity College Dublin, for their advice on and assistance with this research. We thank Dr. Cheryl Tickle, University of Dundee, for providing the Ihh in situ probe.