Original Full Length Article
MMP9 regulates the cellular response to inflammation after skeletal injury
Xiaodong Wanga, Yan Yiu Yua, Shirley Lieua, Frank Yanga, Jeffrey Langa, Chuanyong Lua, Zena Werbb,
Diane Hua, Theodore Miclaua, Ralph Marcucioa, Céline Colnota,c,⁎
aDepartment of Orthopaedic Surgery, University of California at San Francisco, San Francisco, USA
bDepartment of Anatomy, University of California at San Francisco, San Francisco, USA
cINSERM U781, Université Paris Descartes-Sorbonne Paris Cité, Institut Imagine, Hôpital Necker Enfants Malades, Paris, France
a b s t r a c ta r t i c l e i n f o
Received 8 June 2012
Revised 14 September 2012
Accepted 15 September 2012
Available online 23 September 2012
Edited by: Thomas Einhorn
cellular mechanisms are not well understood. To assess the role of inflammation on skeletal cell differentiation,
the first model, fractures are rigidly stabilized leading to direct bone formation, while in the second model, frac-
ture instability causes cartilage and bone formation. We compared the inflammatory response in these two me-
chanical environments and found changes in the expression patterns of inflammatory genes and in the
recruitment of inflammatory cells and osteoclasts. These results suggested that the inflammatory response
could influence skeletal cell differentiation after fracture. We then exploited matrix metalloproteinase 9
(MMP9) that is expressed in inflammatory cells and osteoclasts, and which we previously showed is a potential
regulator of cell fate decisions during fracture repair. Mmp9−/−mice heal stabilized fractures via endochondral
ossification, while wild type mice heal via intramembranous ossification. In parallel, we observed increases in
macrophages and T cells in the callus of Mmp9−/−compared to wild type mice. To assess the link between the
profile of inflammatory cells and skeletal cell fate functionally, we transplanted Mmp9−/−mice with wild type
bone marrow, to reconstitute a wild type hematopoietic lineage in interaction with the Mmp9−/−stroma and
tion and exhibited a normal profile of inflammatory cells. Moreover, Mmp9−/−periosteal grafts healed via
intramembranous ossification in wild type hosts, but healed via endochondral ossification in Mmp9−/−hosts.
Weobserved that macrophages accumulatedattheperiostealsurfaceinMmp9−/−mice,suggesting thatcelldif-
ferentiation in the periosteum is influenced by factors such as BMP2 that are produced locally by inflammatory
cells. Taken together, these results show that MMP9 mediates indirect effects on skeletal cell differentiation by
regulating the inflammatory response and the distribution of inflammatory cells, leading to the local regulation
of periosteal cell differentiation.
© 2012 Elsevier Inc. All rights reserved.
The recruitment of skeletal stem/progenitor cells and their differen-
tiation into osteoblasts and chondrocytes is key to the success of bone
repair. Many factors can influence skeletal progenitors during the
early stages of repair, including mechanical stimuli and inflammatory
factors. The mechanical environment is crucial in determining healing
via endochondral versus intramembranous ossification [1,2]. A stabi-
lized environment favors osteogenic differentiation, whereas the loss
of stabilization favors chondrogenic differentiation at the fracture site.
These cell fate decisions occur during the inflammatory phase of
fracture repair [3,4], however the role of inflammatory signals in skele-
tal cell fate is not well characterized.
The inflammatory phase of bone repair is marked by the infiltration
of inflammatory cells that contribute to the formation of the hematoma
and removal of damaged tissue. While a controlled inflammatory re-
sponse is necessary for stimulating tissue regeneration, prolonged in-
flammation can hinder the completion of the repair process [5–7]. The
multiple inflammatory cell types and factors involved make it difficult
to define the specific impacts of inflammation on bone healing. Thus
far, several studies have shown negative effects of the adaptive immune
er, these studies mostly revealed inflammatory functions during the re-
modeling phase of repair. Less is known about the role of inflammation
on skeletal progenitors that are recruited at the beginning of the inflam-
matory phase during the first few days after fracture. Inflammatory me-
diators such as tumor necrosis factor-α (TNFα) are required for bone
Bone 52 (2013) 111–119
⁎ Corresponding author at: INSERM U781, Université Paris Descartes-Sorbonne Paris
Cité, Institut Imagine, Tour Lavoisier 2ème étage, Hôpital Necker-Enfants Malades, 149
rue de Sèvres-75015 Paris, France. Fax: +33 1 44 49 51 50.
E-mail address: firstname.lastname@example.org (C. Colnot).
8756-3282/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/bone
formation but can also impair later stages of repair by stimulating carti-
lage degradation [12–14]. Likewise, apparent opposite results have
been reported on the role of nonsteroidal anti-inflammatory drugs
(NSAIDs), which target various cell types at different stages of repair in-
cluding osteoblasts, chondrocytes and osteoclasts [15–18]. In orthopedic
to which they affect repair is not well defined. The lack of Cox2, a target
of NSAIDs, inhibits osteoblast differentiation in the periosteum [19,20].
triene inhibitors can stimulate cartilage and bone formation in the early
phase of repair through direct actions on chondrocytes . Interesting-
ly, transplanted mesenchymal stem cells (MSCs), which can improve
healing, have systemic anti-inflammatory effects on the cytokines re-
leased after fracture including TNFα and interleukin-1β (IL1β) .
fluence each other.
Matrix metalloproteinases(MMPs) playimportant roles in bone de-
velopment and repair and these enzymes may participate in the inter-
action between inflammatory cells and skeletal progenitors [3,23–27].
MMP9, along with other MMPs, is expressed in inflammatory cells
and regulates inflammation in other tissues and diseases [28–33]. Our
previous work showed that MMP9 operates both during the inflamma-
tory and remodeling phases of repair . During the inflammatory
phase, MMP9 may regulate skeletal cell fate, as MMP9−/−but not wild
type skeletal progenitors differentiate into chondrocytes in both
non-stabilized and stabilized fractures. We hypothesized that MMP9
stimuli during bone repair. In this study, we assessed the relationship
between inflammation and the differentiation of skeletal progenitors
under different mechanical stimuli at the fracture site, and the role of
matrix metalloproteinase 9 (MMP9) in coordinating these events. We
used bone marrow transplantation and bone grafting to functionally
test the role of the mechanical environment and inflammation on cell
differentiation within the periosteum, a key source of skeletal progeni-
tors for bone repair .
Materials and methods
Non-stabilized and stabilized fractures
All protocols were approved by the Institutional Animal Care and
Use Committee of the University of California at San Francisco.
Mmp9−/−mice (12- to 16-week-old males) and their wild-type (WT)
littermates were anesthetized with an intraperitoneal injection of
Ketamine/Medetomidine. Non-stabilized or stabilized tibia fractures
were generated in the mid-diaphysis via three point-bending [1,3]. Mice
received analgesics and were monitored for signs of pain as described
neal injection of 2% Avertin 2, 5, 7, 10, and 14 days after fracture.
One centimeter of the wild type fractured hind limb between knee
and ankle was collected free of skin followed by RNA extraction at day
2 (stabilized: n=4; non-stabilized: n=3) and day 7 (non-stabilized:
n=4; stabilized: n=4) and from uninjured limbs (n=4) using
Trizol (Invitrogen, Carlsbad, CA). Microarrays were performed using
Agilent Mouse single-color 4×44 K arrays. Image analysis was
performed using Agilent's Extraction 9.1. Comparisons of stabilized
vs. no fracture, non-stabilized vs. no fracture, and stabilized vs.
non-stabilized at each time point were assessed with ANOVA and
t-statistics were assigned using R/Bioconductor. Genes with differen-
tial expression were analyzed for functional enrichment using DAVID
(NIAID/NIH). K-means clustering was performed in MeV4.1.02.
Fluorescence-activated cell sorting
Wild type and Mmp9−/−mice were euthanized following anesthe-
sia at days 0 (unfractured), 2 and 5 post-fracture (n=5 or 6 per
group). The fractured hind limb between the knee and the ankle was
collected free of skin. The bone marrow and soft tissues (periosteum,
muscle and hematoma) were separated. Bone marrow cells were
flushed and harvested in PBS. The minced soft tissue was digested
with collagenase/dispase (Roche, Palo Alto, CA) for 1 h at 37 °C and
the cells were filtered through a cell strainer (BD Falcon, Bedford,
MA). The cells were incubated with a panel of monoclonal antibodies
to identify various inflammatory cell types including anti-F4/80, anti-
CD4 and anti-CD8 (Serotec, Raleigh, NC), anti-FcεRIα, anti-CD11b,
anti-CD19, anti-Ly6G, and anti-IgM (eBiosciences, San Diego, CA), and
anti-CD117 (BD Pharmingen, San Diego, CA). Monoclonal antibodies
were conjugated with fluorescein isothiocyanate (FITC) or phycoery-
thrin (PE). Single or double staining was performed to identify the fol-
lowing inflammatory cell populations: macrophages (F4/80+ or
CD11b+/Ly6G-), neutrophils (CD11b+/Ly6G+), T-helper lympho-
cytes and cytotoxic T cells (CD4+ and CD8+, respectively), B cells
(CD19+/IgM+), and mast cells (CD117+/FcεRIα+). Cells were
washed and analyzed by fluorescence-activated cell sorting (FACS)
using a Facscaliber (Becton Dickinson, San Jose, CA). The percentage of
inflammatory cells was determined after gating on the entire viable
population. Cells were analyzed by flow cytometry using the Cell
Quest software (Becton Dickinson, San Jose, CA). The results were ana-
lyzed using both ANOVA and unpaired Student's t test to assess the ef-
fects of genotype, fracture stability, tissue type (bone marrow versus
soft tissue) and time post-fracture. Significance was determined at
Bone marrow transplantation
Ten-week-old Mmp9−/−or wild type males were lethally irradiated
with two 6 Gy doses of γ-irradiation 3–4 h apart. Bone marrow cells
from wild type and Mmp9−/−mice were transplanted into irradiated
wild type (WT-WT and KO-WT) and Mmp9−/−(WT-KO and KO-KO)
hosts from the same FVB/N background as previously described .
Following a 6-week recovery period, stabilized or non-stabilized frac-
tures were produced. Mice were sacrificed 2 days after fracture to col-
lect cells from bone marrow and soft tissue for FACS analysis of
inflammatorycell populations. At10 and 14 days post-injury,callustis-
sues from stabilized and non-stabilized fractures respectively were col-
lected and processed for histological and histomorphometric analyses
(n=5 or 6 per group).
Histological and histomorphometric analyses
Fractured tibiae were harvested free of skin, fixed overnight in 4%
paraformaldehyde at 4 °C, decalcified in 19% EDTA (pH 7.4) for
14 days, then dehydrated and embedded in paraffin. Sections
(10 μm) through the entire callus were collected. Histomorphometry
was performed as previously described [3,6]. To determine the volume
of cartilage within each callus, every thirtieth section (300 μm) was
stained with Safranin-O/Fast Green (SO/FG) and photographed using a
Leica DM 5000 B light microscope (Leica Microsystems GmbH,Wetzler,
Germany) that was equipped with a camera (Diagnostic Instruments,
Inc., Sterling Heights, MI). The captured images were analyzed with
Adobe Photoshop to determine the area of cartilage in each section.
Immunohistochemistry was performed every tenth slide through-
out the callus using rat anti-F4/80 and rat anti-Ly6G antibodies
(eBiosciences, San Diego, CA) to detect macrophages and neutrophils
X. Wang et al. / Bone 52 (2013) 111–119
respectively. Sections were treated with 20 μg/mL proteinase K (37 °C,
20 min), incubated with primary antibodies at 4 °C overnight. Endoge-
nous peroxidase activity was blocked by incubating the sections with
0.3% H2O2in PBS at room temperature for 10 min. Sections were then
incubated with Biotin Goat Anti-Rat IgG Polyclonal (BD Pharmingen,
San Diego, CA) at room temperature for 1 h followed by streptavidin–
horseradish peroxidase conjugate (Amersham Biosciences, Arlington.
Heights,IL). Stainingwasdetected usingdiaminobenzideneand the tis-
sue was counterstained with 0.1% fast green. On adjacent sections,
tartrate-resistant acid phosphatase (TRAP) staining was performed to
detect osteoclasts using a leukocyte acid phosphatase kit (Sigma, St.
Louis, MO) .
Macrophages, neutrophils and osteoclasts were quantified using
stereology in all tissues of the hind limb between knee and ankle, ex-
cluding the bone marrow compartment. This area corresponds the
soft tissue compartment analyzed via FACS. Immunoreactive- or
TRAP-positive cells were counted using an Olympus CAST system
(Olympus, Center Valley, PA) and software byVisiopharm (Visiopharm,
Hørsholm, Denmark). The total area was outlined at low magnification
(2×) and the cells were counted at high magnification (20×) in
counting frames: ten to thirty fields covering approximately 3% of the
tissue section were randomly acquired by unbiased uniform random
sampling. Four counting frames covering 50% of the area within a field
were overlaid on each field. The number of positive cells and the points
of area within these fields were determined. Values were expressed as
number of cells per mm2tissue. The results were analyzed using
Student's t test. Significance was determined at Pb0.05.
Detection of BMP-2 expressing cells on tissue sections
Immunohistochemistry was performed using a goat anti-BMP-2 an-
tibody (Santa Cruz Biotechnology, Santa Cruz, CA) to detect BMP-2 ex-
pressing cells. After deparaffinization and rehydration, sections were
treated with 0.3% Triton X-100 in PBS for 10 min to permeabilize the
cells and then treated with 0.05% Trypsin in PBS for recovery of antige-
nicity (37 °C, 20 min). Endogenous peroxidase activity was blocked by
incubating the sections with 0.3% H2O2in PBS at room temperature for
20 min. Potential nonspecific binding sites were blocked with 3%
bovine serum albumin in PBS for 1 h. Sections were incubated with
the primary antibody in humidity chambers at 4 °C overnight. After
washes in PBS, sections were incubated with biotin-conjugated donkey
anti-goat polyclonal IgG-B (Santa Cruz Biotechnology, Santa Cruz, CA)
for 1 h and then incubated with avidin-biotinylated horseradish perox-
idase conjugate (Vector Laboratories, Burlingame, CA) for 30 min.
Staining was detected using diaminobenzidene and the tissue was
counterstained with 0.1% Fast Green.
Bone graft transplantation and histological analyses
Live bone graft transplantation was performed as previously de-
scribed . Briefly, bone grafts were isolated from adult wild type
mice that express GFP and Mmp9−/−
5-month-old). A fragment of cortical bone was cut in the anterior–
proximal area of each tibia. The periosteum was kept intact while
the endosteum and bone marrow were removed from the graft
using a razor blade. Fresh bone grafts were transplanted into FVBN
wild type mice and Mmp9−/−hosts (males, 10 week–5-month-old)
to obtain four experimental groups: wild type grafts into wild type
donors (WT–WT; n=5), Mmp9−/−grafts into Mmp9−/−hosts
(Mmp9−/−–Mmp9−/−; n=6), wild type grafts into Mmp9−/−hosts
(WT–Mmp9−/−; n=5) and Mmp9−/−grafts into wild type hosts
(Mmp9−/−–WT; n=5). Tibias were harvested and fixed for 24 h at
4 °C in 4% paraformaldehyde (PFA) at 4 °C, then samples were
decalcified and cryo-embedded . Safranin-O/Fast Green and
trichrome staining were performed to visualize cartilage and bone,
mice (males, 10 week–
respectively. GFP immunostaining was performed on adjacent sections
to confirm the origins of donor versus host cells (data not shown) .
Inflammatory genes are differentially expressed in stabilized and
To begin understanding how fracture stability may regulate inflam-
response in stabilized and non-stabilized mechanical environments
using microarray analysis. mRNA transcripts of inflammatory genes
appeared at days 2 and 7 after injury in stabilized and non-stabilized
fracture environments (Table 1 and Supplemental Fig. 1), indicating
that inflammatory genes are up-regulated after fracture regardless of
discernible in stabilized and non-stabilized fractures at day 2 after injury
(data not shown), inflammatory genes were significantly up-regulated
at day 7 after injury in the stabilized fractures as compared to the
non-stabilized fractures (Table 1 and Supplemental Fig. 1), indicating
that the inflammatory response lasts longer in stabilized fractures.
We identified distinct sets of genes differentially expressed in stabi-
lized and non-stabilized fractures. Although inflammatory genes were
up-regulated in stabilized fractures, genes encoding extracellular ma-
trix proteins were up-regulated in non-stabilized fractures (Supple-
mental Fig. 1), including several members of the Mmp family, such as
Mmp2, Mmp9, Mmp10 and Mmp13, that are involved in extracellular
matrix remodeling and inflammation. These results suggested that
co-regulation of members of clusters that include important candidates
for regulating osteogenesis and chondrogenesis occurs during the in-
flammatory phase of fracture healing.
Genes related to inflammation are significantly up-regulated after fracture. Microarray
analysis of fracture calluses at day 2 and day 7 post stabilized or non-stabilized fracture
identified genes with significant changes in expression (⁎⁎⁎pb0.001;⁎⁎pb0.01;
⁎pb0.05 and NS = Not Significant) compared with no-fracture limbs. Changes were
also found between stabilized and non-stabilized fractures at day 7 after fracture.
Note: M>2, B≥0 for all comparisons between fractured and non-fractured limbs;
M>1, FDRb.05 for all comparisons between stabilized and non-stabilized limbs; all
p-values are Bonferroni-corrected for multiple comparisons.
ComparisonDay 2 vs. No Fx Day 7 vs. No FxS vs. NS
Gene sets (Gene hits)Up in S Up in NSUp in S Up in NSUp in S
Chemokine receptor binding
Small chemokine, C–C
Immune system process
Acute inflammatory response
Response to stress
Response to wounding
X. Wang et al. / Bone 52 (2013) 111–119
Inflammatory cell recruitment differs in stabilized and non-stabilized
To determine whether molecular changes correlate with changes
in the recruitment of inflammatory cells in wild type stabilized
and non-stabilized fractures, we quantified the inflammatory cell
populations within bone marrow and soft tissues around the fracture
site using FACS analysis. In unfractured tibias, immune/inflammatory
cells were detected in the bone marrow, but not in the soft tissues
surrounding the bone (Fig. 1 and data not shown). After fracture, in-
flammatory cells were detected in bone marrow as well as in soft tis-
sues surrounding the fracture site. In stabilized and non-stabilized
fractures, all inflammatory cell types were detected at the fracture
site and surrounding soft tissues by day 2, except for CD8+ T cells
(Fig. 1 and Supplemental Fig. 2). In both mechanical environments,
most inflammatory cells were found around the fracture site from
days 2 to 5 (Fig. 1). However, we observed increased proportions of
macrophages and CD4+ T lymphocytes at days 2 and 5 in stabilized
fractures compared to non-stabilized fractures (Figs. 1B, H; pb0.05),
whereas the proportions of neutrophils, mast cells and B cells were
unchanged in the two groups (Fig. 1E and Supplemental Fig. 2).
Further analyses of non-stabilized fractures by stereology showed
that following this initial phase of inflammation, the proportion of
neutrophils decreased in the callus by day 7 compared to days 2
and 5 (Fig. 1F, pb0.05), whereas macrophages and osteoclasts were
still found within the callus at day 7 post-fracture and their propor-
tions were increased compared to days 2 and 7 (Figs. 1 C, I, pb0.05).
Loss of MMP-9 affects inflammatory cell populations in fracture calluses
MMP9 isa known mediator of inflammation and plays a role in frac-
ture repair [3,29]. The profile of inflammatory cell recruitment differed
between Mmp9−/−and wild type mice and changes were mostly
observed for macrophages and CD4 T cells between genotypes
(pb0.001). In Mmp9−/−stabilized fractures, we observed an increase
in the proportions of macrophages at days 2 and 5, and neutrophils at
day 2 compared to wild type stabilized fractures (Figs. 1A–B, D–E). Al-
though the proportion of CD4+ T cells was increased in the bone
% CD4 T CELLS
FACS-BONE MARROW FACS -SOFT TISSUE
NS S NS S
NS S NS S
% CD4 T CELLS
Fig. 1. Mechanical environment and loss of MMP9 alters the inflammatory cell populations in tibial fractures. Quantification of (A–C) macrophages, (D–F) neutrophils, (G–H) CD4 T
cells and (I) osteoclasts in the tibia. FACS analyses of the bone marrow (left) and soft tissue (middle) of Mmp9−/−and wild type (WT) mouse tibia at day 0 (D0, uninjured), and at
days 2 (D2) and 5 (D5) following non-stabilized (NS) and stabilized (S) fracture. The percentage of macrophages (Ly6G-negative and CD11b-positive), neutrophils (Ly6G-positive
and CD11b-positive) and CD4-positive T cells relative to the total number of cells is reported. (Right) Quantification by stereology of macrophages (C), neutrophils (F) and osteo-
clasts (I) in the soft tissue of Mmp9−/−and wild type non-stabilized (NS) fractures at days 2 (D2), 5 (D5) and 7 (D7) post-fracture. Bars represent mean±standard deviation,
Student's t-test:⁎Pb0.05,⁎⁎Pb0.01 (n=5 or 6 per group).
X. Wang et al. / Bone 52 (2013) 111–119
marrow of Mmp9−/−compared to wild type stabilized fractures, this dif-
proportions of other inflammatory cells were unchanged in Mmp9−/−
and wild type stabilized fractures (Supplemental Fig. 2).
In non-stabilized fractures, the proportion of macrophages and
CD4+ T cells increased in Mmp9−/−compared to wild type mice
(Figs. 1 A–C, G–H). In contrast, osteoclast recruitment was markedly
decreased in the Mmp9−/−calluses compared to wild type (Fig. 1I).
No difference was observed for other inflammatory cell types includ-
ing neutrophils, mast cells and B cells via flow cytometry (Figs. 1 D–E
and Supplemental Fig. 2). Thus the absence of MMP9 affected the re-
cruitment of inflammatory cells in the callus of stabilized and non-
Bone marrow transplantation rescues the fracture healing phenotype in
cells in Mmp9−/−mice impact skeletal cell differentiation in stabilized
and non-stabilized fractures. MMP9 is expressed both by bone
marrow-derived myeloid cells and non-myeloid cells . To deter-
mine if the production of MMP9 by myeloid cells was responsible for
the changes in inflammatory cell recruitment and skeletal cell differen-
tiation during fracture repair, we transplanted Mmp9−/−bone marrow
to provide Mmp9−/−hematopoietic cells in a wild type host environ-
ment [25,35]. When we created stabilized fractures in these wild type
mice that received Mmp9−/−bone marrow transplants, healing oc-
curred via direct bone formation as observed in wild type mice or in
control wild type hosts transplanted with wild type bone marrow
(Fig. 2B). Therefore the presence of Mmp9−/−inflammatory cells
alone was not sufficient to induce chondrogenic cell differentiation at
the fracture site. However, transplantation of wild type bone marrow
served in Mmp9−/−calluses as well as in Mmp9−/−controls
transplanted with Mmp9−/−bone marrow (Fig. 2B). Indeed, 0 of 5
Mmp9−/−mice transplanted with wild type bone marrow exhibited
cartilage in the callus, while 3 of 5 Mmp9−/−mice transplanted with
Mmp9−/−bone marrow exhibited cartilage at day 10 (Fig. 2B). There-
fore, wild type bone marrow could rescue the cartilage phenotype in
Bone marrow transplantation also rescued the cartilage remodeling
defect in Mmp9−/−non-stabilized fractures. By histomorphometric
analyses, the cartilage volume in the Mmp9−/−mice transplanted
with wild type bone marrow cells was significantly smaller than the
Mmp9−/−controls transplanted with Mmp9−/−bone marrow at
14 days post non-stabilized fracture (Fig. 2C). We verified that changes
in inflammatory cell profiles in Mmp9−/−mice paralleled the cartilage
phenotype in stabilized and non-stabilized fractures. The proportions
of macrophages decreased in Mmp9−/−mice transplanted with wild
type compared to Mmp9−/−bone marrow (Fig. 2D). The proportions
of neutrophils remained unchanged and the proportions of CD4+ T
cells decreased although not significantly (Fig. 2D). Osteoclast recruit-
ment was also restored as osteoclast number was significantly higher
mice transplanted with wild type compared to
Mmp9−/−bone marrow (Fig. 2E).
The periosteal response is influenced by MMP9 and the inflammatory
The periosteum is a major contributor of chondrocytes and osteo-
blasts to bone repair, and cell fate decisions in the periosteum influence
healing via endochondral versus intramembranous ossification
[34,38–40]. To determine the role of inflammation and MMP9 on cell
differentiation within the periosteum, we used a bone graft approach.
via endochondral ossification in the absence of mechanical stimuli,
compared to Mmp9−/−grafts in wild type hosts, that healed via
intramembranous ossification (Figs. 3B–C, pb0.05). This result indicat-
ed that a wild type inflammatory environment could prevent cartilage
differentiation in the Mmp9−/−periosteum. Conversely, wild type
grafts healed via intramembranous ossification in both wild type and
Mmp9−/−hosts (Figs. 3D–E). Similar to the bone marrow transplanta-
tion results, the Mmp9−/−environment did not induce cartilage differ-
entiation in the wild type periosteum. Thus, the lack of MMP9 in both
osteal surface and within the callus more closely. We observed that
macrophages remained within the callus as healing progressed, while
other inflammatory cell types were found at the periphery of the callus
by day 7 (Fig. 1 and data not shown). By immunohistochemistry, mac-
rophages accumulated as early as day 3 near the periosteal surface of
the Mmp9−/−compared to wild type stabilized fracture calluses
(Figs. 4A–B, arrowheads). The altered distribution of macrophages in
Mmp9−/−stabilized fractures affected BMP2 expression. We observed
a stronger BMP2 immunostainingin the periosteum of Mmp9−/−stabi-
lized fractures compared to wild type (Figs. 4C–D, arrowheads).
The mechanical environment affects the inflammatory response and
skeletal cell differentiation during bone repair
Bone repair is initiated by an inflammatory response, which coin-
cides with the recruitment and differentiation of skeletal progenitors.
To elucidate therole of inflammationand mechanical signals in skeletal
cell differentiation, we compared the inflammatory response between
stabilized and non-stabilized tibial fractures. Inflammatory genes were
up-regulated by day 2 in both stabilized and non-stabilized fractures,
with no major differences in gene expression profiles in the two me-
chanical environments. This is likely due to the fact that inflammation
is first triggered by the same injury in these two fracture models.
These results correlate with the timing of inflammatory response that
was previously described in several animal models of fracture repair,
with a peak of inflammatory genes and cytokines expression reported
at day 2 post fracture [41–44]. Simultaneously, the major inflammatory
cell types are recruited during this early stage of repair, a crucial stage
for skeletal cell fate decisions [3,45]. Although we detected no differ-
ences at the transcriptional level, we observed that stabilization of the
fractured bones increased the proportions of macrophages and CD4+
T cells compared to non-stabilized fractures. These results suggested
that macrophages and/or T cells play a predominant role in influencing
the mode of repair during the inflammatory phase.
In most tissue repair processes, inflammation must resolve to
support healing . Following the initial inflammatory response, the
expression of inflammatory genes remained high in stabilized fractures
indicating a prolonged inflammatory response. In contrast, in non-
stabilized factures, the decrease in inflammatory gene expression
paralleled cartilage and bone production within the fracture callus.
Thus, differences in the inflammatory environment clearly influence
the formation of cartilage and bone in the callus, as chondrogenesis
and osteogenesis are more robust when inflammation resolves faster.
Indeed, systemic inflammation associated with polytrauma has nega-
tive effects onfracture repairand has been associatedwiththe accumu-
lation of inflammatory factors [46,47]. Robust cartilage formationin the
non-stabilized fracture calluses could also play a role in suppressing in-
flammation . Moreover, non-stabilized fractures exhibit an en-
hanced angiogenic response compared to stabilized fractures, which
may facilitate the removal of inflammatory cells . Careful analyses
of histological sections showed that inflammatory cells were present
flammatory cells were then slowly excluded from the newly forming
X. Wang et al. / Bone 52 (2013) 111–119
fracture callus, where chondrocytes and osteoblasts are differentiating.
portions increased until day 7. These data provided further evidence
that macrophages are a key inflammatory cell type that may interact
with chondrocytes and osteoblasts during callus formation.
MMP9 regulates macrophage recruitment and periosteal cell fate
MMP9 is expressed bothbybonemarrow-derivedmyeloid cells and
osteoclasts that are involved in the inflammatory response and extra-
cellular matrix remodeling during bone repair [3,37]. In the absence of
% SAMPLES WITH
% CD4 T CELLS
WT or Mmp9-/- host
Bone Marrow Transplant from
WT or Mmp9-/- donor
Stabilized tibial fracture
Non-stabilized tibial fracture
X. Wang et al. / Bone 52 (2013) 111–119
MMP9, the inflammatory cell profile at the fracture site was altered
comparedtowildtypeanimals,and thiswasfunctionally linkedtophe-
notypic changes in Mmp9−/−mutants. Osteoclast recruitment was im-
paired in Mmp9−/−fracture calluses and bone marrow transplants
rescued the cartilage remodeling defects in non-stabilized fractures
via providing MMP9-expressing osteoclasts. This result was in concor-
dance with the rescue of the growth plate phenotype in Mmp9−/−de-
veloping long bones . Similarly, the recruitment of T cells and
macrophages wasaffectedin Mmp9−/−fracture calluses andtransplan-
tation of wild type bone marrow rescued the cartilage phenotype in
Mmp9−/−stabilized fractures. Since bone marrow transplantation
mostly provides cells from the hematopoietic lineage with long-term
engraftmentcapacities [22,25,35,51–55], we concludethatthemodula-
tions of the inflammatory response can indirectly impact osteogenesis
and chondrogenesis in the fracture callus. However, transplantation of
phenotype in stabilized fractures, indicating that cartilage induction
was not due to a cell autonomous defect in inflammatory cells. Using
periosteal graft transplantation, we uncovered the fact that MMP9
acts indirectly at the level of the periosteum, a target of many signals
that regulate skeletal cell fate during the early stages of repair
This indirect effect of MMP9 on periosteal cell fate was via the regu-
lation of macrophage localization. Only macrophages were located
cell differentiation. CD4+ T cells were found away from the perioste-
um, and so are unlikely to have an effect on chondrogenesis and osteo-
genesis. Previous studies reported that changes in the infiltration of
macrophages within the fracturecallus were related to the levelof frac-
ture stability, but did not provide a functional link with changes in the
osteogenic or chondrogenic response . Another study illustrated
the important role of resident macrophages, so-called osteomacs,
which are closely associated with bone lining osteoblasts. These
osteomacs are also associated with osteoclasts, and their depletion im-
pairs bone repair via intramembranous ossification . Surprisingly,
the complete absence of macrophages in CCR2 KO mice does not have
consequences on the early stages of repair, but mostly impairs the re-
modeling phase of repair .
While macrophages are not required for the inductionof endochon-
dral ossification within the callus, they are required for bone induction
in pathological conditions such as heterotopic ossification (HO) .
This bone induction may be due to the production of BMPs that are
expressed by macrophages and other inflammatory cells, but that are
not normally active in the muscle environment [60,38]. During fracture
repair, BMPs are produced by many cell types and are up-regulated in
the periosteum of non-stabilized fractures, which heal via endochon-
dral ossification [38,56,61]. Ectopic localization of macrophages in the
periosteum of Mmp9−/−stabilized fractures may be responsible for
the induction of endochondral ossification as we found an increase in
BMP2 expression. There are other examples of indirect mechanisms of
action of MMP9. MMP9 could regulate the mechanical properties of
the extracellular matrix, which may directly impact periosteal stem
cells [57,62,63]. During bone development and during the replacement
giogenesis by regulating the bioavailability of VEGF [3,37,50]. Thus,
MMP9 can modulate the function of important growth factors by
Proportions of samples with donor-derived cartilage
5/8 0/5 0/50/5
WT or Mmp9-/- Periosteal Bone Graft
WT or Mmp9-/- Host Tibia
Fig. 3. Analysis of cartilage formation during healing of Mmp9−/−and wild type periosteal bone grafts. (A) Experimental design. (B–E) Longitudinal sections through the tibia and
histological evaluation using Safranin-O staining at day 10 post-surgery of Mmp9−/−periosteal bone grafts transplanted in Mmp9−/−(B, Mmp9−/−–Mmp9−/−) or wild type hosts
(C, Mmp9−/−–WT) or wild type periosteal bone grafts transplanted in wild type (D, WT–WT) or Mmp9−/−hosts (D, WT–Mmp9−/−). Cartilage (red) is only observed in Mmp9−/−
mice transplanted with Mmp9−/−bone grafts. Student's t-test:⁎Pb0.05 (Mmp9−/−–Mmp9−/−compared to WT–Mmp9−/−, WT–WT and Mmp9−/−–WT; n=5 or 6 per group).
Fig. 2. Transplant of wild type bone marrow prevents cartilage formation in Mmp9−/−stabilized fractures and alters inflammatory cell profile. (A) Experimental design. (B, left)
Safranin-O Fast Green staining of day 10 stabilized fracture calluses of Mmp9−/−mice transplanted with Mmp9−/−bone marrow (Mmp9−/−-Mmp9−/−) or with wild type bone marrow
of total cartilage volume at day 10 post stabilized fracture and percentage of samples with cartilage observed at the fracture site. (C, left) Safranin-O Fast Green staining of day 14
non-stabilized fracture calluses of Mmp9−/−mice transplanted with Mmp9−/−bone marrow (Mmp9−/−–Mmp9−/−) or with wild type bone marrow (WT–Mmp9−/−). (C, right)
Histomorphometric measurements of total cartilage volume at day 14 post non-stabilized fracture. (D) FACS analyses at day 2 post non-stabilized fracture of macrophages, neutrophils
and CD4-positive T cells in bone marrow (BM) and soft tissue (ST) of Mmp9−/−mice transplanted with Mmp9−/−(Mmp9−/−–Mmp9−/−) or wild type bone marrow (WT–Mmp9−/−).
(E) Quantification by stereology at day 5 post-stabilized fracture of osteoclasts in Mmp9−/−mice transplanted with Mmp9−/−(Mmp9−/−–Mmp9−/−) or wild-type bone marrow
(WT–Mmp9−/−). Bars represent mean±standard deviation, Student's t-test: * Pb0.05; NS=Not Significant (n=5 or 6 per group).
X. Wang et al. / Bone 52 (2013) 111–119
controlling their release from the extracellular matrix or by controlling
thespatialdistributionof cellsthat secretethese growthfactors [39,56].
In conclusion, by elucidating the indirect effects of MMP9 on peri-
osteal cell fate, we have unraveled one of the multiple mechanisms by
which inflammatory cells and mechanical stimuli may influence skel-
etal cell differentiation during bone repair. It is clear that specific in-
flammatory cell types such as macrophages can be targeted to
stimulate bone repair, and that other cell types including T cells
may also be targeted. To improve the management of inflammation
following bone injury clinically, more studies will be required to dis-
sect the roles of numerous inflammatory and mechanical factors, and
their impact on skeletal stem cells [64–66]. These factors act in a com-
plex healing environment where major signaling pathways including
BMPs, Wnt, Hedgehogs and Parathyroid Hormone also regulate skel-
etal stem cells, and may be modulated by mechanical stimuli and in-
flammatory cytokines [67–69].
Supplementary data to this article can be found online at http://
This work was funded by NIH-NIAMS R01 AR053645 to TM, R01
AR057344 to CC and TM, INSERM ATIP-AVENIR and Marie Curie Inter-
national Reintegration grant to CC, R01 AR046238 and R01 CA057621
to ZW. Microarray analysis was performed at the UCSF Lung Biology
Microarray Core Facility. We thank Jesse Shantz for help with statisti-
 Thompson Z, Miclau T, Hu D, et al. A model for intramembranous ossification dur-
ing fracture healing. J Orthop Res 2002;20(5):1091-8.
 Le AX, Miclau T, Hu D, et al. Molecular aspects of healing in stabilized and
non-stabilized fractures. J Orthop Res 2001;19(1):78-84.
 Colnot C, Thompson Z, Miclau T, et al. Altered fracture repair in the absence of
MMP9. Development 2003;130(17):4123-33.
 Miclau T, Lu C, Thompson Z, et al. Effects of delayed stabilization on fracture
healing. J Orthop Res 2007;25(12):1552-8.
 Pape HC, Marcucio R, Humphrey C, et al. Trauma-induced inflammation and frac-
ture healing. J Orthop Trauma 2010;24(9):522-5.
 Lu C, Miclau T, Hu D, et al. Cellular basis for age-related changes in fracture repair.
J Orthop Res 2005;23(6):1300-7.
 Claes L, Recknagel S, Ignatius A. Fracture healing under healthy and inflammatory
conditions. Nat Rev Rheumatol 2012;8(3):133-43.
 Toben D, Schroeder I, El Khassawna T, et al. Fracture healing is accelerated in the
absence of the adaptive immune system. J Bone Miner Res 2011;26(1):113-24.
 Xing Z, Lu C, Hu D, et al. Multiple roles for CCR2 during fracture healing. Dis Model
 Alexander KA, Chang MK, Maylin ER, et al. Osteal macrophages promote in vivo
intramembranous bone healing in a mouse tibial injury model. J Bone Miner Res
 Colburn NT, Zaal KJ, Wang F, et al. A role for gamma/delta T cells in a mouse model
of fracture healing. Arthritis Rheum 2009;60(6):1694-703.
 Gerstenfeld LC, Cho TJ, Kon T, et al. Impaired intramembranous bone formation
during bone repair in the absence of tumor necrosis factor-alpha signaling. Cells
Tissues Organs 2001;169(3):285-94.
 Gerstenfeld LC, Cho TJ, Kon T, et al. Impaired fracture healing in the absence of
TNF-alpha signaling: the role of TNF-alpha in endochondral cartilage resorption.
J Bone Miner Res 2003;18(9):1584-92.
 Alblowi J, Kayal RA, Siqueira M, et al. High levels of tumor necrosis factor-alpha
contribute to accelerated loss of cartilage in diabetic fracture healing. Am J Pathol
 Radi ZA. Pathophysiology of cyclooxygenase inhibition in animal models. Toxicol
 Abdul-Hadi O, Parvizi J, Austin MS, et al. Nonsteroidal anti-inflammatory drugs in
orthopaedics. J Bone Joint Surg Am 2009;91(8):2020-7.
 Chang JK, Wu SC, Wang GJ, et al. Effects of non-steroidal anti-inflammatory drugs
on cell proliferation and death in cultured epiphyseal-articular chondrocytes of
fetal rats. Toxicology 2006;228(2–3):111-23.
 Simon AM, O'Connor JP. Dose and time-dependent effects of cyclooxygenase-2 in-
hibition on fracture-healing. J Bone Joint Surg Am 2007;89(3):500-11.
 Xie C, Ming X, Wang Q, et al. COX-2 from the injury milieu is critical for the initi-
ation of periosteal progenitor cell mediated bone healing. Bone 2008;43(6):
 Zhang X, Schwarz EM, Young DA, et al. Cyclooxygenase-2 regulates mesenchymal
cell differentiation into the osteoblast lineage and is critically involved in bone re-
pair. J Clin Invest 2002;109(11):1405-15.
 Wixted JJ, Fanning PJ, Gaur T, et al. Enhanced fracture repair by leukotriene antago-
a novel role of the cysteinyl LT-1 receptor. J Cell Physiol 2009;221(1):31-9.
 Granero-Molto F, Weis JA, Miga MI, et al. Regenerative effects of transplanted
mesenchymal stem cells in fracture healing. Stem Cells 2009;27(8):1887-98.
 Vu TH, Werb Z. Matrix metalloproteinases: effectors of development and normal
physiology. Genes Dev 2000;14(17):2123-33.
 Stickens D, Behonick DJ, Ortega N, et al. Altered endochondral bone development in
matrix metalloproteinase 13-deficient mice. Development 2004;131(23):5883-95.
 Behonick DJ, Xing Z, Lieu S, et al. Role of matrix metalloproteinase 13 in both en-
dochondral and intramembranous ossification during skeletal regeneration. PLoS
 Lieu S, Hansen E, Dedini R, et al. Impaired remodeling phase of fracture repair in
the absence of matrix metalloproteinase-2. Dis Model Mech 2011;4(2):203-11.
 Dan H, Simsa-Maziel S, Hisdai A, et al. Expression of matrix metalloproteinases dur-
ing impairment and recovery of the avian growth plate. J Anim Sci 2009;87(11):
Fig. 4. Distribution of macrophages near the periosteal surface during fracture repair. Immunodetection of (A–B) macrophages using F4/80 antibody and (C–D) BMP2-expressing
cells at the periosteal surface of wild type and Mmp9−/−stabilized tibial fractures at day 3 post-injury. (A–B) Macrophages are found throughout the callus and are concentrated
near the periosteum in Mmp9−/−stabilized fractures (B, arrowheads) compared to wild type fractures. (C–D) BMP2 immunostaining is detected in inflammatory cells at the frac-
ture site and near the periosteum with stronger staining found in the periosteum of Mmp9−/−stabilized fractures (D, arrowheads) compared to wild type.
X. Wang et al. / Bone 52 (2013) 111–119
 Esparza J, Kruse M, Lee J, et al. MMP-2 null mice exhibit an early onset and severe
experimental autoimmune encephalomyelitis due to an increase in MMP-9 ex-
pression and activity. FASEB J 2004;18(14):1682-91.
 Corry DB, Kiss A, Song LZ, et al. Overlapping and independent contributions of
MMP2 and MMP9 to lung allergic inflammatory cell egression through decreased
CC chemokines. FASEB J 2004;18(9):995-7.
 Greenlee KJ, Corry DB, Engler DA, et al. Proteomic identification of in vivo sub-
strates for matrix metalloproteinases 2 and 9 reveals a mechanism for resolution
of inflammation. J Immunol 2006;177(10):7312-21.
 Hsu JY, McKeon R, Goussev S, et al. Matrix metalloproteinase-2 facilitates wound
healing events that promote functional recovery after spinal cord injury.
J Neurosci 2006;26(39):9841-50.
 Gutierrez-Fernandez A, Inada M, Balbin M, et al. Increased inflammation delays
wound healing in mice deficient in collagenase-2 (MMP-8). FASEB J 2007;21(10):
 Reif S, Somech R, Brazovski E, et al. Matrix metalloproteinases 2 and 9 are markers
of inflammation but not of the degree of fibrosis in chronic hepatitis C. Digestion
 Colnot C. Skeletal cell fate decisions within periosteum and bone marrow during
bone regeneration. J Bone Miner Res 2009;24(2):274-82.
 Colnot C, Huang S, Helms J. Analyzing the cellular contribution of bone marrow to
fracture healing using bone marrow transplantation in mice. Biochem Biophys
Res Commun 2006;350(3):557-61.
 Colnot CI, Helms JA. A molecular analysis of matrix remodeling and angiogenesis
during long bone development. Mech Dev 2001;100(2):245-50.
 Ortega N, Wang K, Ferrara N, et al. Complementary interplay between matrix
metalloproteinase-9, vascular endothelial growth factor and osteoclast function
drives endochondral bone formation. Dis Model Mech 2010;3(3–4):224-35.
 Yu YY, Lieu S, Lu C, et al. Immunolocalization of BMPs, BMP antagonists, receptors,
and effectors during fracture repair. Bone 2010;46(3):841-51.
 Yu YY, Lieu S, Lu C, et al. Bone morphogenetic protein 2 stimulates endochondral os-
sification by regulating periosteal cell fate during bone repair. Bone 2010;47(1):
 Wang Q, Huang C, Zeng F, et al. Activation of the Hh pathway in periosteum-derived
mesenchymal stem cells induces bone formation in vivo: implication for postnatal
bone repair. Am J Pathol 2010;177(6):3100-11.
 Einhorn TA, Majeska RJ, Rush EB, et al. The expression of cytokine activity by frac-
ture callus. J Bone Miner Res 1995;10(8):1272-81.
 Cho TJ, Gerstenfeld LC, Einhorn TA. Differential temporal expression of members
of the transforming growth factor beta superfamily during murine fracture
healing. J Bone Miner Res 2002;17(3):513-20.
 Einhorn TA. The cell and molecular biology of fracture healing. Clin Orthop Relat
Res 1998;46(Suppl. 355):S7–S21.
 Rundle CH, Wang H, Yu H, et al. Microarray analysis of gene expression during the
inflammation and endochondral bone formation stages of rat femur fracture re-
pair. Bone 2006;38(4):521-9.
 Heiner DE, Meyer MH, Frick SL, et al. Gene expression during fracture healing in
rats comparing intramedullary fixation to plate fixation by DNA microarray.
J Orthop Trauma 2006;20(1):27-38.
 Claes L, Ignatius A, Lechner R, et al. The effect of both a thoracic trauma and a
 Recknagel S, Bindl R, Kurz J, et al. Experimental blunt chest trauma impairs frac-
ture healing in rats. J Orthop Res 2011;29(5):734-9.
 Ulivi V, Lenti M, Gentili C, et al. Anti-inflammatory activity of mono-
galactosyldiacylglycerol in human articular cartilage in vitro: activation of
an anti-inflammatory cyclooxygenase-2 (COX-2) pathway. Arthritis Res
 Lu C, Saless N, Hu D, et al. Mechanical stability affects angiogenesis during early
fracture healing. J Orthop Trauma 2011;25(8):494-9.
 Vu TH, Shipley JM, Bergers G, et al. MMP-9/gelatinase B is a key regulator of
growth plate angiogenesis and apoptosis of hypertrophic chondrocytes. Cell
 Giordano A, Galderisi U, Marino IR. From the laboratory bench to the patient's
bedside: an update on clinical trials with mesenchymal stem cells. J Cell Physiol
 Satija NK, Singh VK, Verma YK, et al. Mesenchymal stem cell-based therapy: a new
paradigm in regenerative medicine. J Cell Mol Med 2009;13(11–12):4385-402.
 Xing Z, Lu C, Hu D, et al. Rejuvenation of the inflammatory system stimulates frac-
ture repair in aged mice. J Orthop Res 2010;28(8):1000-6.
 Simmons PJ, Przepiorka D, Thomas ED, et al. Host origin of marrow stromal cells fol-
lowing allogeneic bone marrow transplantation. Nature 1987;328(6129):429-32.
 Rieger K, Marinets O, Fietz T, et al. Mesenchymal stem cells remain of host origin
even a long time after allogeneic peripheral blood stem cell or bone marrow
transplantation. Exp Hematol 2005;33(5):605-11.
 Wang Q, Huang C, Xue M, et al. Expression of endogenous BMP-2 in periosteal
progenitor cells is essential for bone healing. Bone 2011;48(3):524-32.
 Colnot C, Zhang X, Tate ML. Current insights on the regenerative potential of the
periosteum: Molecular, cellular, and endogenous engineering approaches.
J Orthop Res 2012, http://dx.doi.org/10.1002/jor.22181 [Epub ahead of print].
 Huang Y, Zhang X, Du K, et al. Inhibition of beta-catenin signaling in chondrocytes
induces delayed fracture healing in mice. J Orthop Res 2012;30(2):304-10.
 Hankemeier S, Grassel S, Plenz G, et al. Alteration of fracture stability influences
chondrogenesis, osteogenesis and immigration of macrophages. J Orthop Res
 Kan L, Liu Y, McGuire TL, et al. Dysregulation of local stem/progenitor cells as a com-
mon cellular mechanism for heterotopic ossification. Stem Cells 2009;27(1):150-6.
 Tsuji K, Bandyopadhyay A, Harfe BD, et al. BMP2 activity, although dispensable for
bone formation, is required for the initiation of fracture healing. Nat Genet
 Karamichos D, Skinner J, Brown R, et al. Matrix stiffness and serum concentration
effects matrix remodelling and ECM regulatory genes of human bone marrow
stem cells. J Tissue Eng Regen Med 2008;2(2–3):97–105.
 McBride SH, Dolejs S, Brianza S, et al. Net change in periosteal strain during stance
shift loading after surgery correlates to rapid de novo bone generation in critically
sized defects. Ann Biomed Eng 2011;39(5):1570-81.
 Mountziaris PM, Spicer PP, Kasper FK, et al. Harnessing and modulating inflam-
mation in strategies for bone regeneration. Tissue Eng B Rev 2011;17(6):393-402.
 Huang C, Ogawa R. Mechanotransduction in bone repair and regeneration. FASEB
 Palomares KT, Gleason RE, Mason ZD, et al. Mechanical stimulation alters tissue
differentiation and molecular expression during bone healing. J Orthop Res
 Rosen V. Harnessing the parathyroid hormone, Wnt, and bone morphogenetic
protein signaling cascades for successful bone tissue engineering. Tissue Eng B
 Sibai T, Morgan EF, Einhorn TA. Anabolic agents and bone quality. Clin Orthop
Relat Res 2011;469(8):2215-24.
 Chen Y, Whetstone HC, Lin AC, et al. Beta-catenin signaling plays a disparate role
in different phases of fracture repair: implications for therapy to improve bone
healing. PLoS Med 2007;4(7):e249.
X. Wang et al. / Bone 52 (2013) 111–119