Induction of Fracture Repair by Mesenchymal Cells Derived from
Human Embryonic Stem Cells or Bone Marrow
Anita Undale,1 Daniel Fraser,1 Theresa Hefferan,2 Ross A. Kopher,3 James Herrick,2 Glenda L. Evans,2 Xiaodong Li,2
Sanjeev Kakar,2 Meredith Hayes,2 Elizabeth Atkinson,4 Michael J. Yaszemski,2 Dan S. Kaufman,3 Jennifer J. Westendorf,2
Sundeep Khosla1
1Endocrine Research Unit, College of Medicine, Mayo Clinic, Rochester, Minnesota, 2Orthopedic Research, College of Medicine, Mayo Clinic,
Rochester, Minnesota, 3Department of Medicine and Stem Cell Institute, University of Minnesota, Minneapolis, Minnesota 55455, 4Biostatistics,
College of Medicine, Mayo Clinic, Rochester, Minnesota
Received 5 January 2011; accepted 23 May 2011
Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.21480
ABSTRACT: Development of novel therapeutic approaches to repair fracture non-unions remains a critical clinical necessity. We evalu-
ated the capacity of human embryonic stem cell (hESC)-derived mesenchymal stem/stromal cells (MSCs) to induce healing in a fracture
non-union model in rats. In addition, we placed these findings in the context of parallel studies using human bone marrow MSCs
(hBM-MSCs) or a no cell control group (n ¼ 10–12 per group). Preliminary studies demonstrated that both for hESC-derived MSCs and
hBM-MSCs, optimal induction of fracture healing required in vitro osteogenic differentiation of these cells. Based on biomechanical
testing of fractured femurs, maximum torque, and stiffness were significantly greater in the hBM-MSC as compared to the control
group that received no cells; values for these parameters in the hESC-derived MSC group were intermediate between the hBM-MSC
and control groups, and not significantly different from the control group. However, some evidence of fracture healing was evident by
X-ray in the hESC-derived MSC group. Our results thus indicate that while hESC-derived MSCs may have potential to induce fracture
healing in non-unions, hBM-MSCs function more efficiently in this process. Additional studies are needed to further modify hESCs to
achieve optimal fracture healing by these cells.
2011 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop
Res
Keywords: embryonic stem cells; mesenchymal stem/stromal cells; osteogenic; bone repair; fracture non-union
Five to ten percent of all fractures are complicated by
delayed union or non-union.1 Thus, development of
novel therapeutic approaches to treat and repair non-
union fractures remains a critical clinical necessity.
Human bone marrow mesenchymal stem/stromal cells
(hBM-MSCs) are considered a promising candidate for
clinical applications. The availability of autologous
hBM-MSCs from patients is advantageous for thera-
peutic applications. hBM-MSCs are multipotent and
possess the ability to regenerate cell types specific for
different tissues, including adipose tissue, bone, and
cartilage. Implantation of autologous hBM-MSCs
using different scaffolds resulted in bone regeneration
in several animal models.2–5 However, long-term
cultures of hBM-MSCs have limitations because, over
time, the cells exhibit a reduced proliferation rate
attributable to telomere shortening and senescence.6
In recent years, significant advances have been
made in examining the potential of human embryonic
stem cells (hESCs) in regenerative medicine. ESCs
are undifferentiated, pluripotent cells derived from
mammalian pre-implantation blastocysts. Unlike
hBM-MSCs, hESCs can be maintained in culture in-
definitely in an undifferentiated state, thus offering
potential advantages over the use of adult hBM-MSCs.
Since the initial report describing the derivation of
hESCs in 1998,7 these cells have been widely used for
studies of human developmental biology and to define
methods that support the differentiation of hESCS
into specific lineages.8–10
Recently, several groups reported the development
of MSCs from hESCs.11–14 These MSCs have the sur-
face phenotype typical of hBM-MSCs isolated from
hBM, including expression of CD73, CD105, Stro-1,
and CD44. Molecular studies show expression of
transcription factors and matrix proteins also typically
produced by hBM-MSCs. A stromal cell co-culture
system demonstrated that hESC-derived CD73þ cells
function as MSCs with robust formation of adipocytes,
chondrocytes, and osteoblasts.13,14
The goal of our study was to define the optimal con-
ditions for using hESC-derived populations to repair
fractures in a fracture non-union model.15,16 We ini-
tially identified optimal culture conditions for CD73þ
hESC-derived MSCs to induce fracture healing. Next,
we evaluated the capacity of these CD73þ hESC-
derived MSCs to induce healing in a fracture non-
union model in rats. In addition, we placed these
findings in the context of parallel studies using hBM-
MSCs or a control group that received no cells.
METHODS
Animals
Male athymic nude rats (Charles River Labs, Wilmington,
MA), age 8–12 weeks, weighing 270–370 g were used. Follow-
ing pilot studies, the rats were divided into three groups for
the main study: control, atelocollagen matrix infused with
saline (n ¼ 10); atelocollagen infused with CD73þ hESC-de-
rived MSCs differentiated under osteogenic conditions (see
below) for 7 days (n ¼ 10); and atelocollagen infused with
hBM-MSCs differentiated under osteogenic conditions (see
below) for 7 days (n ¼ 12). All rats were housed, treated, and
handled in accordance with the guidelines set forth by the
Mayo Clinic Institutional Animal Care and Use Committee.
Correspondence to: Sundeep Khosla (T: þ1-507-255-6663; F: þ1-
507-293-3853 E-mail: khosla.sundeep@mayo.edu)

2011 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
JOURNAL OF ORTHOPAEDIC RESEARCH 1

Cell Culture
The hESC line H9 (obtained from WiCell) was maintained
as undifferentiated cells as previously described17,18 by
coculture with irradiated mouse embryonic fibroblast (MEF)
cells in DMEM/F12 supplemented with 15% Knockout
Serum Replacer (KOSR) (Invitrogen, Carlsbad, CA), 1%
MEM-nonessential amino acids (Invitrogen), 0.5% penicillin–
streptomycin (P/S), 2 mM L-glutamine, 0.1 mM b-mercaptoe-
thanol (Sigma-Aldrich, St. Louis, MO), and 4 ng/ml human
bFGF (Invitrogen). hBM-MSCs (Lonza, Allendale, NJ) were
cultured in phenol red-free aMEM (Invitrogen) supple-
mented with 10% fetal bovine serum (FBS) and 1% P/S.
Media were changed every 2–3 days, and cells were passaged
upon reaching 80–90% confluency. Sorted CD73þ hESC-
derived MSCs and hBM-MSCs were differentiated on plastic
for 7 days in the presence of osteogenic differentiation medi-
um containing phenol red-free aMEM, 10% FBS, 50 mM
ascorbic acid, and 10
8 dexamethasone.
Mesenchymal Differentiation by Stromal Co-Culture
The mouse bone marrow stromal cell line M2-10B4 (ATCC)
was grown in DMEM (Invitrogen) containing 10% FBS
(Hyclone, Rockford, IL), 1% P/S, 1% MEM-nonessential ami-
no acids, and 0.1 mM b-mercaptoethanol. M2-10B4 cells
were inactivated with 10 mg/ml mitomycin C in M2-10B4-
conditioned medium for 3 h at 378C with 5% CO2 prior to
culture on gelatin-coated plates (Sigma). Mesenchymal differ-
entiation of hESCs occurred during co-culture on stromal cell
layers. hESCs were passaged onto M2-10B4 mouse stromal
cells with differentiation medium consisting of RPMI supple-
mented with 15% defined FBS (Hyclone), 2 mM L-glutamine,
0.1 mM b-mercaptoethanol, 1% MEM-nonessential amino
acids, and 1% P/S. Media were changed every 2–3 days. After
17–21 days, differentiated hESCs were made into a single
cell suspension by treatment with collagenase IV (1 mg/ml)
(Invitrogen), followed by 0.05% trypsin/EDTA (Gibco/Invitro-
gen, Carlsbad, CA) supplemented with 2% chick serum (Sig-
ma). Cells were treated with an antibody against human
CD73 conjugated with PE (Pharmingen, Vancouver, BC,
Canada). Antibodies for PE conjugated with magnetic beads
were used to select CD73þ cells from single cell suspension
using the EasySep PE selection kit (Stemcell Technologies,
San Diego, CA). Post-sorted CD73þ hESCs were plated onto
gelatin-coated plates and grown in mesenchymal stem cell
medium consisting of aMEM (Invitrogen) supplemented with
10% FBS, 1% P/S, 1% MEM-nonessential amino acids, 2 mM
L-glutamine, and 0.1 mM b-mercaptoethanol. Media were
changed on sorted cells every 2–3 days. For flow cytometric
analysis, post-sorted cell populations were washed with
Mg2þ and Ca2þ-free phosphate buffered saline (PBS)
(Hyclone) and removed from culture plates using 0.25% tryp-
sin/EDTA. After washing with PBS supplemented with 2%
FBS and 0.1% sodium azide, the cells were incubated with
antibodies for CD34-APC, CD73-PE, CD90-PE, CD105-APC,
and CD146-FITC (all IgG1, all from Pharmingen) and their
corresponding IgG1 controls. Live cell populations identified
using 7AAD exclusion were analyzed on FACS Calibur (Bec-
ton Dickenson, Franklin Lakes, NJ) for different surface an-
tigen expression using Flow-Jo Software (TreeStar).
Induction of Fracture Non-Union
We modified a fracture non-union protocol described by
Matsumoto and co-workers.15,16 Briefly, a lateral parapatel-
lar knee incision was made on the right limb to expose the
distal femoral condyle. To avoid displacement of the fracture,
a K-wire was inserted from the trochlear groove into the fem-
oral canal in a retrograde manner. The tendons were at-
tached over the patellofemoral joint with absorbable sutures.
To avoid complex fractures, we modified the previously pub-
lished protocol,15,16 and did not cut the femur with a thin
saw cut but still used 3-point bending directly to fracture the
femur. Thus, a transverse femoral shaft fracture was pro-
duced in the right femur of each rat using a C-shaped instru-
ment applying 3-point bending. To achieve non-union, the
periosteum was cauterized circumferentially at a distance of
2 mm on each side of the fracture. The rats immediately
received a local implantation of saline or cells mixed with
atelocollagen (105 cells of each type in 100 ml of the solution)
at the fracture site.16 The wound was closed, and post-
operative pain was managed by administration of subcutane-
ous injection of buprenorphine hydrochloride after surgery.
Normal weight bearing activities were allowed following the
operation.
Implantation of Cells
Immediately after the creation of the non-union fracture, the
rats received the appropriate local implantation of saline
or cells mixed with atelocollagen (105 cells of each type in
100 ml of the solution) at the fracture site. We used atelocol-
lagen as a scaffold based on an established model of fracture
repair.16
Radiographical Assessment of Fracture Healing
Radiographs of the fractured femurs were taken under anes-
thesia immediately after surgery and every 2 weeks with the
animal in the supine position and both limbs fully extended.
Radiographs were evaluated by two blinded readers and
scored for fracture healing using a modification of a previous-
ly describe scale19: 0—no change from post-fracture appear-
ance; 1—trace of radiodense material in the fracture defect;
2—flocculent radiodensity with flecks of calcification; 3—
defect bridged at least at one point with material of non-uni-
form density; 4—at least one of four cortices obscured by new
bone, or the fracture was bridged on dorsal and ventral sides
with uniform callus; and 5—defect bridged by uniform new
bone, cut ends of cortex no longer distinguishable. For each
animal and time point, we used the mean of the scores of the
two readers.
Tissue Harvesting
Eight weeks after fracture, rats were euthanized by CO2
inhalation. The fractured femurs used for biomechanical
testing were dissected free from the surrounding muscle, and
testing was performed as described below. As part of the pilot
studies, femurs were also used in histological analyses; bones
were harvested and embedded in methyl methacrylate.
Serial sections were mounted on silane-coated glass slides
and stained immediately as described below.
Biomechanical Assessment of Fracture Union
Without removal of the intramedullary K-wires, standard-
ized torsional testing was performed on the fractured site.
The torsional load was applied at a speed of 58/s for a maxi-
mum of 36 s. Maximum torque (N-cm) and stiffness (N-cm/
degree) were used as the primary endpoints.20 Maximum
torque was the highest load that the bone sustained before
fracture, and stiffness was calculated from the linear portion
of the loading curve (higher values for both are indicative of
2 UNDALE ET AL.
JOURNAL OF ORTHOPAEDIC RESEARCH