Animal Procedures
All procedures were approved by the Stanford Committee on Animal Research.
Skeletally mature (ten to twelve weeks of age) male mice were used for all
studies. Following anesthesia and analgesia, the right cheek was shaved and
the skin cleansed. An incision was made along the mandibular ramus, and the
buccinator muscle was divided until the lateral surface of the mandible was
exposed; the periosteum was preserved. With use of a 1.0-mm drill-bit, a hole
was created that penetrated one cortex. The region was irrigated, and the skin
was closed with use of a nonabsorbable suture. Mice were killed at multiple
time points that represented the inflammatory, hard callus, and remodeling
phases of healing.
Histology
Tissues were harvested and fixed in 4% paraformaldehyde overnight, followed
by standard processing into paraffin blocks and 12-µm-thick sections.
Histological staining was performed with use of a modified Movats pentachrome
staining
method4.
Molecular and Cellular Analyses
In Situ Hybridization
The relevant digoxigenin-labeled mRNA antisense probes were prepared from
cDNA templates for Wnt2b, Wnt3a, and Wnt7a. In situ hybridization was
performed as previously
reported5.
Beta-Galactosidase Staining
Cells responsive to Wnt signaling express ß-galactosidase, which can
be detected by Xgal staining. For Xgal staining, tissues were fixed with 0.25%
glutaraldehyde for fifteen minutes and stained with Xgal overnight at
37°C.
Adenovirus-Mediated Inhibition of Wnt Signaling
All adenoviral constructs were generated
previously6. Wnt
inhibition was achieved by tail-vein injection of Ad-Dkk1 and the control
Ad-Fc, and injuries were generated immediately.
Wnt Signaling Is Activated During the Early Phase of Bone
Regeneration
In order to investigate the role of ß-catenin-dependent Wnt signaling
in craniofacial bone repair, we modified our monocortical skeletal defect
model7. We chose the
mandibular ramus as the defect site due to its anatomical properties (i.e.,
small bone-marrow cavity, two cortices, and thickest diameter). TOPgal Wnt
reporter mice8 were
employed for these studies to gain a temporal and spatial understanding of the
role of ß-catenin-dependent Wnt signaling during skeletal repair.
Seven days after injury, the defect site had filled with a cell-rich matrix
and there were no signs of inflammation. The first signs of bone matrix
deposition were evident in the periosteum, close to the cortical cut edge and
in the endosteum (Fig. 1,
A). Xgal staining of an adjacent slide revealed abundant
Wnt reporter activity in the entire skeletal defect
(Fig. 1, B). Seven
days later, the periosteal callus dramatically increased in size, and about
50% of the defect was filled with osseous matrix
(Fig. 1, C). Compared
with the earlier time point, the number of Xgal-positive cells was greatly
decreased (Fig. 1, D).
At the twenty-one-day time point, the entire injury site was bridged by a
mature osseous callus, which was surrounded by a neoperiosteum
(Fig. 1, E). In
contrast to the earlier time points, Wnt-responsive cells were only detectable
in the outermost layer of the callus, in direct contact with the neoperiosteum
(Fig. 1, F). Finally,
after twenty-eight days of bone repair, the callus was reduced in size through
osteoclast remodeling (Fig. 1,
G), and Wnt-responsive cells were absent from the
regenerate (Fig. 1,
H).
From these data, we conclude that ß-catenin-dependent Wnt signaling
participates in the early phases of skeletal wound repair when cells are
proliferating and forming the callus. Wnt signaling diminishes in the later
stages of repair, when osteoclastic remodeling of the callus predominates.
Multiple Different ß-Catenin-Dependent Wnt Ligands Are Expressed
During Early Mandibular Repair
To gain a better appreciation for which Wnt ligands were involved in this
process of craniofacial skeletal regeneration, we employed an in situ
hybridization-based screening for components of the ß-catenin-dependent
Wnt pathway. There are a number of variables that needed to be considered for
this screening process. There are at least forty-eight members of the Wnt
pathway that include ligands, receptors, inhibitors, and cofactors—all
of which are potentially relevant to understand Wnt action during skeletal
regeneration. In addition, Xgal data indicate that Wnt responsiveness begins
within twenty-four hours of
injury9 and persists
for at least two weeks (Fig. 1,
F). To add to the complexity, Wnts are likely to be
involved in early (e.g., survival, proliferation) and late (differentiation)
events in the healing process, so multiple time points need to be examined.
Here, we chose to analyze the participation of Wnt ligands in the early stages
of bone repair. We focused on the day-7 time point, and we show the expression
of three Wnt ligands that signal through the ß-catenin-dependent pathway.
Wnt2b, Wnt3a, and Wnt7a were broadly expressed throughout the early callus
(Fig. 2, A through
C). Higher magnification of the cut edge of the cortical
bone revealed expression of all three ligands in the periosteum and in the
endosteal surfaces of the injury (Fig. 2,
D through F). In addition, all three ligands
were strongly expressed in the osteoblasts that lined the newly formed bone
(Fig. 2, G through
I).
Inhibition of ß-Catenin-Dependent Wnt Signaling Results in a
Delay in Bone Matrix Deposition
Thus far, our data indicate that at least three Wnt ligands are expressed
in the callus by day 7 after injury, and that cells responding to this
endogenous Wnt signaling are in greatest numbers in the early stages of callus
formation. Taken together, these data suggest that Wnt signaling participates
in the process of cranial bone repair. To test this hypothesis directly, we
inhibited Wnt signaling in vivo with use of a gene-transfer strategy. Dkk1 is
a soluble antagonist of Wnt
signaling10, and we
employed an adenovirus expressing Dkk1 (Ad-Dkk1) to treat the mandibular
injury site. An adenovirus expressing the Fc portion of the mouse
immunoglobulin (Ad-Fc) was used as a control. The adenoviral constructs were
injected intravenously into TOPgal mice, a technique that has been shown to
produce a conditional, reversible inhibition in Wnt
signaling6. Viral
infection was assessed by Western blot analyses, which confirmed high levels
of protein expression of both Dkk1 and the Fc fragment within forty-eight
hours of injection (data not shown).
Immediately after systemic infection, skeletal defects were generated. The
Ad-Fc and Ad-Dkk1-treated mandibles were collected on day 7 after injury and
then subjected to whole-mount Xgal staining. Uninjured, untreated control
mandibles generated in TOPgal mice showed no Xgal staining, but the anterior
portion of the osseous mandible and the tongue showed strong Xgal staining
(Fig. 3, A). The
injured, Ad-Fc-treated mandible exhibited high reporter activity throughout
the buccinator muscle seven days after injury
(Fig. 3, B), which
indicated that injury triggers upregulation of the endogenous Wnt pathway.
This Wnt responsiveness was greatly reduced when animals were treated with
Ad-Dkk1 (Fig. 3, C).
Xgal staining was not only reduced at the injury site, but was also absent in
the tongue and other sites of endogenous Wnt responsiveness. These data showed
that mandibular injuries triggered activation of Wnt signaling and that this
effect was ameliorated by the soluble Wnt antagonist Dkk.
Next, we used histology to evaluate the consequences of Wnt inhibition on
bone repair. While Ad-Fc-treated injuries healed following the normal time
course of repair (n = 6; Fig. 3,
D), Ad-Dkk1 treatment appeared to inhibit osteoblast
differentiation, which resulted in a lack of new bone matrix (n = 6;
Fig. 3, E). Normally
the mandibular periosteum thickens to almost eight times its normal width in
response to trauma (Fig. 1, C and
E), but in Ad-Dkk1 samples, this periosteal response was
abolished (Fig. 3, E).
These data demonstrate that inhibition of ß-catenin-dependent Wnt
signaling decelerates bone matrix deposition both at the injury site and in
the periosteum. Both of these effects result in a delay in cranial skeletal
repair.
Constitutively Active Wnt Signaling Results in Increased Bone Mass
and Decelerated Bone Repair
Our data show that Wnt inhibition is detrimental to bone regeneration
(Fig. 3). We next tested
whether activation of the Wnt pathway had a beneficial effect on bone healing.
A gain-of-function mutation in the Wnt co-receptor LRP5 is associated with a
high bone-mass
phenotype11,12;
thus, we evaluated mandibular bone regeneration in transgenic mice harboring
the same mutation (i.e., LRP5-G171V mice).
We first examined the intact LRP5-G171V skeleton and confirmed the
increased bone-mass phenotype. In the cranial skeleton, this manifested as a
reduced periodontal space around the molar roots. For example, in wild-type
mice, the periodontal ligament surrounds each molar root and its fibers bridge
a gap of about 50 to 100 µm (Fig. 4,
A). In the LRP5-G171V mice, this gap was greatly reduced
in width, in part due to calcification of the fibers that serve in attaching
the root (Fig. 4, B).
In addition to this cranial skeletal defect, the bone of the mandible was
thicker (data not shown).
Next, we evaluated how the healing of skeletal defects was affected by the
LRP5-G171V mutation. Seven days after creating a monocortical defect in
LRP5-G171V mice, we noticed some intramembranous ossification in the
periosteal compartment (Fig. 4,
C). This ossification was not accompanied by the usual
signs of matrix deposition at the injury site, however
(Fig. 4, C). These
data are consistent with the results we obtained from experiments involving a
tibial defect in this mouse
strain9. In this
mandibular skeletal defect, bone repair was delayed also, apparently due to
exuberant cell proliferation in the early repair process, which delayed
osteoblast differentiation.
This study addresses the role of ß-catenin-dependent Wnt signaling
during craniofacial bone regeneration. First, we showed that Wnt-responsive
cells populate the injury site until the defect is completely bridged by a
hard callus. Second, we used an in situ hybridization screening method to
detect the Wnt ligands that were expressed at the injury site at the time of
osteoblast differentiation. These two findings together implied that the
ß-catenin-dependent Wnt-signaling pathway participates in the early
regenerative response. In a gene-transfer approach, we inhibited this
Wnt-signaling pathway by overexpressing the soluble Wnt antagonist Dkk1. This
induced a downregulation of Wnt responsiveness, shown by Xgal staining in
reporter mice, and a delay in bone matrix deposition at the injury site.
Finally, we used a mouse mutant model, in which Wnt signaling is
constitutively active. Here, we observed calcified periodontal ligaments, in
keeping with the published high bone-mass phenotype caused by activating
mutations in LRP5. When we generated cranial skeletal injuries in LRP5-G171V
mice, we observed a delay in bone repair, similar to the postponed bone
healing we have observed in the
tibia9.
Is Wnt Signaling Equally Involved in Craniofacial and Appendicular
Bone Healing?
When considering the different origins of the craniofacial and appendicular
skeleton, one might wonder if each repair process involves distinct and unique
signaling pathways. Previously published data provide evidence that
ß-catenin-dependent Wnt signaling in appendicular skeletal regeneration
is essential for proliferation and differentiation of osteoprogenitor cells at
the injury site9.
Our data show that ß-catenin-dependent Wnt signaling is also essential
for craniofacial bone repair. Therefore, both skeletal compartments employ a
similar signaling pathway despite their different embryologic origin.
?