2012 Outstanding Paper: Runner-up
Implications of adipose-derived stromal cells in a 3D culture system for
osteogenic differentiation: an in vitro and in vivo investigation
Francis H. Shen, MDa,*, Brian C. Werner, MDa, Haixiang Liang, MDa, Hulan Shang, MSb,
Ning Yang, PhDb, Xudong Li, MD, PhDa, Adam L. Shimer, MDa, Gary Balian, PhDa,c,
Adam J. Katz, MD, FACSb,d,e
aDepartment of Orthopaedic Surgery, University of Virginia, Box 800159, Charlottesville, VA 22908, USA
bDepartment of Plastic Surgery, University of Virginia, Box 800376, Charlottesville, VA 22908, USA
cDepartment of Biochemistry and Molecular Genetics, University of Virginia, Box 800376, Charlottesville, VA 22908, USA
dDepartment of Biomedical Engineering, University of Virginia, Box 800759, Charlottesville, VA 22908, USA
eDivision of Plastic and Reconstructive Surgery, Department of Surgery, University of Florida College of Medicine, PO Box 100138,
Gainesville, FL 32610-0138, USA
Received 2 February 2012; revised 20 December 2012; accepted 8 January 2013
AbstractBACKGROUND CONTEXT: Healthy mammalian cells in normal tissues are organized in com-
plex three-dimensional (3D) networks that display nutrient and signaling gradients. Conventional
techniques that grow cells in a 2D monolayer fail to reproduce the environment that is observed
in vivo. In recent years, 3D culture systems have been used to mimic tumor microenvironments
in cancer research and to emulate embryogenesis in stem cell cultures. However, there have been
no studies exploring the ability for adipose-derived stromal (ADS) cells in a 3D culture system to
undergo osteogenic differentiation.
PURPOSE: To characterize and investigate the in vitro and in vivo potential for human ADS cells
in a novel 3D culture system to undergo osteogenic differentiation.
STUDY DESIGN: Basic science and laboratory study.
METHODS: Human ADS cells were isolated and prepared as either a 2D monolayer or 3D mul-
ticellular aggregates (MAs). Multicellular aggregates were formed using the hanging droplet tech-
nique. Cells were treated in osteogenic medium in vitro, and cellular differentiation was
investigated using gene expression, histology, and microCT at 1-, 2-, and 4-week time points.
In vivo investigation involved creating a muscle pouch by developing the avascular muscular inter-
val in the vastus lateralis of male athymic rats. Specimens were then pretreated with osteogenic me-
dium and surgically implanted as (1) carrier (Matrigel) alone (control), (2) carrier with human ADS
cells in monolayer, or (3) human ADS cells as MAs. In vivo evidence of osteogenic differentiation
was evaluated with micro computed tomography and histologic sectioning at a 2-week time point.
RESULTS: Human ADS cells cultured by the hanging droplet technique successfully formed MAs
at the air-fluid interface. Adipose-derived stromal cells cultured in monolayer or as 3D MAs retain
FDA device/drug status: Investigational (stem cells).
Author disclosures: FHS: Royalties: Elsevier Publishing (B); Consul-
ting: Synthes Spine (B), DePuy Spine (B); Speaking/Teaching Arrange-
ments: DePuy Spine (B); Board of Directors: MTF (B, Paid directly to
institution); Fellowship Support: AO (D, Paid directly to institution); Other
Office/Editorial Advisory board: The Spine Journal (none), SPINE (none),
European Spine Journal (none), SpineLine (none). BCW: Grants: OREF
(C, Paid directly to institution), OTA (B, Paid directly to institution).
HL: Nothing to disclose. HS: Nothing to disclose. NY: Nothing to dis-
close. XL: Nothing to disclose. ALS: Nothing to disclose. GB: Grants:
NIH (F, Paid directly to institution), Department of Defense (Amount
not disclosed, Paid directly to institution); Support for travel to meetings
for the study: NIH (Grants disclosed), DOD (Grants disclosed); Speak-
ing/Teaching Arrangements: Seminar (Amount not disclosed); Trips/
Travel: Seminar (Amount not disclosed); Scientific Advisory Board/Other
office: Journal of Orthopedic Research (none), Connective Tissue Research
(none). AJK: Grant: NIH (F, Paid directly to institution); Royalties: Uni-
versity of Pittsburgh (E); Stock Ownership: Cytori Therapeutics (B); Pri-
vate Investments: The GID Group (G); Consulting: LifeNet Health (D),
MicroAire (unknown); Board of Directors: The GID Group (equity), Plur-
oGen Therapeutics (equity).
The disclosure key can be found on the Table of Contents and at www.
This work was supported by an NIH T32 training grant, an Orthopae-
dic Trauma Association (OTA) resident research grant, and an Orthopaedic
Research and Education Fund (OREF) resident research grant.
* Corresponding author. Department of Orthopaedic Surgery, Univer-
sity of Virginia, PO Box 800159, Charlottesville, VA 22908-0159, USA.
Tel.: (434) 243-0291; fax: (434) 243-0242.
E-mail address: firstname.lastname@example.org (F.H. Shen)
1529-9430/$ - see front matter ? 2013 Elsevier Inc. All rights reserved.
The Spine Journal 13 (2013) 32–43
their ability to self-replicate and undergo multilineage differentiation as confirmed by increased
runx2/Cbfa2, ALP, and OCN and increased matrix mineralization on histologic sectioning. Multi-
cellular aggregate cells expressed increased differentiation potential and extracellular matrix pro-
duction over the same human ADS cells cultured in monolayer. Furthermore, MAs reseeded
onto monolayer retained their stem cell capabilities. When implanted in vivo, significantly greater
bone volume and extracellular matrix were present in the implanted specimens of MAs confirmed
on both microCT and histological sectioning.
CONCLUSIONS: This is the first study to investigate the capability of human ADS cells in a 3D
culture system to undergo osteogenic differentiation. The results confirm that MAs maintain their
stem cell characteristics. Compared with analogous cells in monolayer culture, the human ADS
cells as MAs exhibit elevated levels of osteogenic differentiation and increased matrix mineraliza-
tion. Furthermore, the creation of uniform spheroids allows for improved handling and manipula-
tion during transplantation. These findings strongly support the concept that 3D culture systems
remain not only a viable option for stem cell culture but also possibly a more attractive alternative
to traditional culture techniques to improve the osteogenic potential of human adipose stem
cells. ? 2013 Elsevier Inc. All rights reserved.
Keywords: Human adipose stromal cell; Osteogenic differentiation; Multicellular aggregate; Hanging droplet; Three-
Traditionally, bone marrow has been the major source
for multipotential stromal cells [1,2]; however, bone mar-
row harvest is an invasive, painful procedure with reported
complication rates of up to 30% [3,4]. Recently, an analo-
gous subpopulation of cells have also been found to exist
within adipose tissue. Studies suggest that adipose-
derived stromal (ADS) cells are more than space fillers
and under appropriate conditions can differentiate down
the osteoblastic lineage [5,6]. Studies have demonstrated
that ADS cells are an abundant source of mesenchymal
stem cells, are capable of self-renewal, and undergo multi-
lineage differentiation [7,8].
The high multidifferentiative potential of the ADS cells
combined with their reduced immunogenicity, when col-
lected autologously, makes them excellent candidates for
use in regenerative medicine [9,10]. Furthermore, adipose
stem cells(ASCs)can be easily obtained from adiposetissue
bone marrow . However, despite their relative abun-
dance, typically ADS cells still require their numbers to be
expanded in vitro before in vivo implantation.
Currently, conventional expansion techniques focus on
culturing cells in monolayer. The microenvironment estab-
lished by the extrinsic and intrinsic cellular signals from the
stem cells has a profound influence on their biologic func-
tion [12,13]. Studies suggest that the mere process of isolat-
ing multipotential mesenchymal stem cells from their
native three-dimensional (3D) environment and culturing
them in a 2D adherent monolayer can alter normal physio-
logical behavior resulting in loss of replicative ability,
colony-forming efficiency, and differentiation capabilities
over time [14,15].
Healthy mammalian cells naturally exist within a 3D mi-
croenvironment, which is affected by both complex cell-to-
cell and cell-to-extracellular matrix (ECM) interactions. In
recent years, 3D culture systems have been used to mimic
tumor microenvironments in cancer research [16,17] and
to emulate embryogenesis in stem cell cultures [18,19];
however, to the best of our knowledge, there have been
few studies that have used this technique for osteogenesis.
More specifically, there have been no studies exploring the
use of ADS cells in this manner. The purpose of this study
is to characterize and investigate the potential for human
ADS cells in a novel 3D culture system to undergo osteo-
Materials and methods
The series of in vitro and in vivo experiments, as well as
culture conditions and assays, is depicted in Fig. 1.
Cell isolation and preparation
All adipose tissues were obtained from intraoperative
suction lipectomy of a single patient. Subcutaneous adipose
tissue was obtained from this nondiabetic patient undergo-
ing elective surgical procedures via a protocol approved by
the Human Investigation Committee. ASCs were isolated
using previously described methods . Briefly, harvested
tissue was washed, centrifuged, decanted to remove oil, en-
zymatically dissociated, and centrifuged. Pelleted stromal
cells were then recovered, washed, filtered twice (250 mm
mesh followed by 105 mm mesh), centrifuged, and dec-
anted. Contaminating erythrocytes were lysed with an os-
motic buffer, and the stromal cells were plated onto
culture plastic (Thermo Fisher Scientific, Rochester, NY,
USA). Cultures were washed with buffer 24 to 48 hours af-
ter plating to remove unattached cells and then refed with
a fresh medium. Cells were grown to confluence after the
initial plating (p50), typically within 10 to 14 days, lifted
33 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
using TrypLE (Invitrogen, Carlsbad, CA, USA), and
Multicellular aggregate preparation
Passage zero (p50) cells were labeled with anti-CD31
and anti-CD45 antibodies (BD Biosciences, San Jose,
CA, USA), and magnetic sorting (MACS Miltenyi Biotec,
Auburn, CA, USA) was used to remove mature endothelial
cells and leukocytes. Mature endothelial cells are CD31þ,
and mature leukocytes are CD45þ. Both of these distinct
cell subpopulations are present in adipose tissue and do
not possess the replicative capability of CD31?/CD45?
stem cells. The resulting CD31?/CD45?cells were col-
lected and replated as passage 1 (p51) cells. After 2 weeks
of culture, the cells were released, counted, and resus-
pended at 625,000 cells/mL in a low-serum medium.
Forty-microliter droplets were placed on the inside of
lular aggregate (MA) formation using the hanging drop
method [21–23]. Cells were allowed to form 3D aggregates
comprised of 50,000 cells per droplet. After 24 hours of cul-
ture, MAs were transferred into suspension culture in ultra–
low attachment culture plates (Corning, Lowell, MA, USA).
Separate aliquots of the same CD31?/CD45?ASCs were
cultured as adherent monolayers, using standard culture
dishes, and the same low-serum (1% serum) medium was
used to culture MAs. Both MAs and monolayer cells were
cultured for 1 week in the low-serum medium before use in
in vitro and in vivo studies.
The hanging droplet method is based on the natural pre-
disposition of cells to form into a 3D cellular aggregate
without the need for polymer scaffolds. Although modified
[21–23] since its original description , practically speak-
ing, the hanging droplet technique places a cellular suspen-
sion within an inverted well, which is held in place by
surface tension. Although cellular ‘‘sheets’’ have also been
describedwith othermethods,inthecase ofthehanging
droplet technique, the 3D MAs adopt a uniform spheroid
shape that accumulates at the liquid-air interface (Fig. 2).
Monolayer and 3D MA cultures
For monolayer culture, human ADS cells, at passage 3,
were plated onto 24-well tissue culture plates at a density of
5?104cells/cm2. For MA cultures, human ADS cells
(5?104cells per MA) prepared as detailed earlier were
placed in individual wells of 96-well plates for suspension
Otherwise, both cultures were then maintained at 37?C in
a humidified atmosphere of 95% air and 5% CO2. The
monolayer cells were grown up to 75% confluence in Dul-
becco’s Modified Eagle Medium (DMEM)/F12. Control me-
dium was high-glucose DMEM. Osteogenic differentiation
media consisted of high-glucose DMEM with 0.01 mM
1,25-dihydroxyvitamin D3 (R&D System, Inc., Indianapolis,
MN, USA), 50 mM L-ascorbate-2-phosphate, 50 mM dexa-
methasone, and 10 mM b-glycerophosphate. Medium in
all groups was changed every 3 days. Samples were col-
lected for gene studies at 1, 2, and 4 weeks. Samples were
fixed for histological studies at the 4-week time point.
Replating MAs as monolayer culture
Multicellular aggregates of human ADS cells (5?104
cells per MA) prepared as detailed earlier were cultured
for 2 weeks in suspension culture with standard medium
(DMEM). After this time period, each MA (approximately
5 mm in diameter) was placed in a single well of a 24-well
plate. These cultures were maintained at 37?C in a humidi-
fied atmosphere of 95% air and 5% CO2. Over a 2-week pe-
riod, the MA was allowed to slowly dissociate and reach
75% confluence in monolayer culture. Once 75% conflu-
ence was achieved, both control and osteogenic medium
(OM) experimental groups were grown in parallel as men-
tioned previously. Additionally, confirmation of multiline-
age capacity was performed at this point in separate
experiments. Multicellular aggregates replated in mono-
layer culture were differentiated in standard adipogenic
and chondrogenic media. Samples were collected for gene
expression and histological studies at the 4-week postosteo-
genic induction time point.
RNA extraction and purification
Briefly, the cultured cells from monolayer were rinsed
twice with phosphate-buffered saline. The cells were lysed
The use of osteogenic human stem cells holds promise for
spinal fusion as well as general orthopedic applications.
In this paper, the authors compared osteogenic activity of
adipose-derived cells cultured using a three-dimensional
technique versus monolayer culturing versus an acellular
control. The three-dimensional cultured and implanted
cells performed best in the rat model.
The behavior of cells cultured in different ways can vary
widely. This phenomenon has implications for basic sci-
ence studies from which conclusions drawn regarding
the biological activities from individually cultured
monolayer cells may not predict those seen in humans,
which are the intended host. It is also important prag-
matically (as in this study) where the aim was to have
the cells exhibit a therapeutic advantage.
34 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
with RLT buffer (Qiagen, Chatsworth, CA, USA) and
stored in a ?70?C freezer. Alternatively, MAs from 3D cul-
ture were rinsed twice with phosphate-buffered saline. The
MAs were lysed with RLT buffer (Qiagen) and significant
mechanical agitation and stored in a ?70?C freezer.
Gene expressions of runx2/core-binding factor alpha-1
(runx2/Cbfa1), alkaline phosphatase (ALP), and osteocalcin
(OCN) were determined at 1, 2, and 4 weeks. Total RNA
was extracted and purified using an RNeasy kit (Qiagen).
The purified RNA was then stored at ?20?C. The yield of
RNAwas determined by measuring absorbance at 260 nm.
Real-time polymerase chain reaction
Briefly, reverse transcriptase reactions were annealed at
37?C for 10 minutes followed by first-strand complementary
DNA (cDNA) synthesis at 42?C for 30 minutes and heat in-
activation at 95?C for 5 minutes. The resulting cDNA was
stored frozen at ?20?C until assay by real-time polymerase
chain reaction (PCR). Using the QuantiTect SYBR Green
PCR kit (Qiagen), the real-time PCR reactions were per-
formed with 25 mL of the SYBR Green kit master mix
and forward and reverse primers (300 nmol/L each). The
3-mL cDNA sample was added to the final reaction mixture
undiluted from the reverse transcriptase reaction. The 96-
well real-time PCR format included six 10-fold dilutions
in duplicate of the plasmid DNA standards. Each sample
was analyzed at least in duplicate with the iCycler instru-
ment (BioRad Laboratories, Hercules, CA, USA).
The PCR protocols used involved the activation of Am-
pliTaq Gold DNA Polymerase followed by 40 cycles of
denaturation at 94?C for 30 seconds, annealing at 60?C
for 30 seconds, and extension at 72?C for 30 seconds.
The PCR threshold cycle number (Ct) for each sample
was calculated at the point where the fluorescence exceeded
the threshold limit. The threshold limit was fixed along
the linear logarithmic phase of the fluorescence curves at
10 to 20 standard deviations above the average background
fluorescence. Relative expression of the target gene was
normalized to the 18S expression by the delta-delta cT
method with the 18S primer (QuantumRNA 18S PCR prod-
ucts; Ambion, Austin, TX, USA).
Human ADS cells, at passage 3, were plated onto 24-
well tissue culture plates at a density of 5?104cells/cm2,
and each corresponding treatment group was subsequently
cultured in its respective media for 4 weeks. Cell phenotype
and mineralization were evaluated by microscopy. The
presence of calcium deposition was determined by Alizarin
Fig. 1. Experimental design.
35 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
Briefly, the cells from each group were first washed
twice with phosphate-buffered saline and then were fixed
with 10% (vol/vol) formaldehyde at room temperature for
15 minutes. The cells were rinsed with distilled water and
then stained by 1 mL of Alizarin Red (pH 4.1; American
MasterTech, Lodi, CA, USA) at room temperature for 30
minutes. The cells were then washed again several times
with distilled water. After being visualized by microscopy,
the plate was stored at ?20?C for quantification.
For quantification, 800 mL of 10% (vol/vol) acetic acid
was added to each well. This was then incubated in room
temperature for 30 minutes with shaking. The cells were
then scraped from the plate and transferred to a 1.5-mL mi-
crocentrifuge tube. After vortexing for 30 seconds, the
tubes were sealed, heated to 85?C for 10 minutes, cooled
in ice, and centrifuged for 15 minutes. Five hundred micro-
liters of the supernatant was transferred to a new tube and
neutralized with 10% (vol/vol) ammonium hydroxide. One
hundred fifty microliters of supernatant was triplicate read
with standard blocks at 405 nm in a 96-well plate.
In regard to the MAs, at 4 weeks, MAs from both the
control and the experimental group were fixed in 10%
(vol/vol) formaldehyde at room temperature for 15 minutes,
suspended in OCT compound (Tissue Tek, Sakura, Tor-
rance, CA, USA), and finally placed in liquid nitrogen until
uniformly frozen to allow for frozen sectioning. Frozen sec-
tions were stained with both hematoxylin and eosin (H&E)
staining to examine basic cellular structure and Alizarin
Red to determine the presence of calcification.
In brief, section MAs on slides were rinsed in tap water
and then Alizarin Red stain was applied for 15 to 30 sec-
onds. Excess stain was removed, and the slides were
dehydrated in acetone, followed by acetone/xylene solu-
tion, and finally cleared with xylene. After mounting, the
slides were examined under light microscopy.
In vivo cell preparation
Human ADS cells were cultured with DMEM/F12 me-
dium containing 10% fetal bovine serum in 5% CO2at
37?C. Cells were maintained at subconfluence before pas-
sage 3. Each hanging droplet MA was made with 5?104
cells. Both MAs and monolayer-cultured cells were cul-
tured in OM (high-glucose DMEM, 10% fetal bovine
serum, 1% Antibiotic-Antimycotic, 200 mM L-ascorbate-
2-phosphate, 10 nM 1,25-dihydroxyvitamin D3, 10 nM
dexamethasone, and 10 mM b-glycerophosphate) for 2
weeks before implantation. Each MA was cultured with
200 mL medium in a 96-well, cell nonadherence culture
plate. Media were changed every 3 days.
Recombinant human bone morphogenetic protein 2
tration of 0.25 mg/mL. Before implantation, monolayer-
cultured cells were digested by 2% tyrosine and suspended
the supernatant was carefully removed and the cells were re-
tration of 1?106cells/mL. Twenty microliters Matrigel
aliquots containing 2?106cells were transferred to 1 well
in a 96-well, cell nonadherence plate. The plate was kept in
the incubator at 37?C for 10 minutes to allow the Matrigel
to solidify. Hanging droplet MAs were washed by
phosphate-buffered saline twice and then mixed with
Fig. 2. Hanging droplet technique. Multicellular aggregate within hanging drop (left inset). Multicellular aggregate at 50? (right insert).
36F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
Matrigel containing rhBMP2 on ice. Twenty microliters of
Matrigel was mixed with 20 MAs and put in 1 well in a 96-
in the incubator at 37?C for 10 minutes before implantation.
Muscle pouch surgical technique
The animal protocol was approved by the Institutional
Animal Care and Use Committee. Four-week-old male
athymic rats (NIH-RNU, National Cancer Institution, Fred-
erick, MA, USA) were used for the study. All the animals
received general anesthesia. The low back and two hind
legs were prepared with disinfectant techniques as per pro-
tocol. Then a 1-cm incision was made on the skin of lateral
aspect of the thigh parallel with the middle third of the fe-
mur. The fascia of the vastus lateralis muscle was separated
with blunt dissection, and a muscle pouch was created.
Twenty microliters of Matrigel with or without cells was
implanted in the muscle pouch. Four surgical sites received
hanging droplet MA implantation, and four sites were im-
planted with monolayer-cultured cells. The muscle fascia
was closed with absorbable sutures. Skin was closed with
nonabsorbable sutures. Animals were allowed to move
freely in the cage after surgery and served normal food
MicroCT scans were performed after implantation of
cells at 3-day and 2-week time points. Animals were anes-
thetized with general anesthesia. A 3D volume rendering of
each animal was acquired from 360 projection CT images
with a Feldkamp back-projection algorithm (Exxim Corp.,
Pleasanton, CA, USA). The results were then analyzed with
ImageJ software (National Institution of Health). Bone vol-
ume was measured by using a fixed threshold setting for all
Animals were sacrificed 2 weeks after operation. The lo-
cation of the muscle pouch was identified by sutures used
for stitching the fascia. At least 1 cm2with whole thickness
of vastus lateralis muscle tissue surrounding the suture area
was harvested. Samples were fixed with 10% neutral for-
malin for 24 hours and then embedded in OCT compound
for frozen section. Cryosections of 7 mm thickness slices
were made with a cryomicrotome. The slices were stained
with H&E staining and Masson’s trichrome staining.
Confirmation of osteogenic potential in monolayer and
Osteogenic potential for human ADS cells cultured in
monolayer and as MAs in 3D culture was confirmed with
gene expression, histology, and CT imaging. Gene expres-
sions of runx2/Cbfa1, ALP, and OCN for cells cultured in
monolayer and treated in OM demonstrated increased
values over the controls (control medium, DMEM, which
is not osteogenic) in every group and at each time period
confirming their ability to undergo osteogenic differentia-
tion (Fig. 3). Similarly, MAs in the 3D culture system also
Fig. 3. Gene expression of human adipose-derived stromal cells cultured
in monolayer and treated either with control medium (DMEM) or with os-
teogenic medium (OM) for markers of osteogenic differentiation: (Top)
runx2, (Middle) ALP, and (Bottom) OCN. Cells treated with OM demon-
strated increased expression of all osteogenic markers at all time points
compared with controls.
37 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
demonstrated significantly greater values over controls at
all time periods for both ALP and OCN (p!.05) (Fig. 4).
There was no significant difference in runX2 expression be-
tween the MAs and monolayer-cultured cells.
Using Alizarin Red staining to evaluate the presence of
mineralization, ADS cells in monolayer culture and MA
treated with OM demonstrated significantly greater calcium
deposition than the corresponding untreated control group
(p!.05) (data not shown). CT images demonstrated that
the MAs had significantly greater bone volume than controls
when treated with OM. The in vitro ADS cells in monolayer
are 2D with no significant depth; therefore, no bone volume
was noted when the 24-well plate with monolayer cells was
scanned with the microCT. This confirmed the osteogenic
potential of ADS cells in monolayer and 3D culture
(Fig. 5). Our study confirmed that the reorganization of
ADS cells into a 3D MA body did not disrupt its ability
to undergo osteogenic differentiation (Fig. 6).
Potentiation of gene expression for osteogenic
differentiation of MA over time
When directly comparing gene expression of 50,000 hu-
man ADS cells cultured in monolayer versus 50,000 of the
same cells as an MA in 3D culture, in regard to ALP, initial
values at 1 week demonstrated greater expression in mono-
layer; however, by 2 and 4 weeks, this trend had reversed
and there was significantly greater expression of ALP in
the MA group in comparison to monolayer. Similarly,
OCN expression was initially greater in the monolayer
group; however, this had also reversed by 4 weeks with
nearly a twofold greater expression in the MA group
(Fig. 7). No significant difference in runX2 expression
Maintenance of stem cell properties of renewal and
Histological images clearly demonstrate the MA ability
to readhere to a monolayer scaffold when reseeded onto
a tissue culture dish. Sequential imaging captures the MA
serving as a ‘‘depot’’ or reservoir of cells, which retain their
capability to divide and form confluent cells in monolayer.
Over the course of several weeks, cells can be seen expand-
ing in culture and extending pseudopods from the MA as
they reattach to the monolayer. This confirmed the ability
for the cells in the MA to expand in culture once recultured
When cultured in adiopogenic medium, the MAs al-
lowed to expand in monolayer culture differentiate into
Fig. 4. Gene expression of human adipose-derived stromal cells cultured as three-dimensional multicellular aggregates (MAs) and treated with either control
medium (DMEM) or osteogenic medium (OM) for markers of osteogenic differentiation: (Left) ALP and (Right) OCN. Multicellular aggregates treated with
OM demonstrated increased expression of all osteogenic markers at all time points compared with controls.
Fig. 5. Bone volume of control specimens and multicellular aggregates
(MAs) in vitro using microCT at preset threshold value. There was no de-
tectable bone formation in the control specimens, whereas significant bone
formation was detectable in the MA specimens cultured in osteogenic me-
38 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
adipocytes demonstrating abundant lipid droplets (Fig. 8A).
Furthermore, when cultured in OM, the cells underwent os-
teogenic differentiation with complete mineralization of the
ECM (Fig. 8B). The ability to expand in culture combined
with the potential to undergo multilineage differentiation
confirmed that the cells preserved their multipotent proper-
ties even after 3D culture.
CT scanning of in vivo muscle pouch specimens
No specimens demonstrate any evidence of mineraliza-
tion in the muscles at 3 days after implantation. By 2 weeks,
100% of specimens with implanted MAs demonstrated mi-
croCT evidence osteogenesis. However, at 2 weeks, none
of the specimens with implanted monolayer-cultured cells
demonstrated any bone formation. Quantitative evaluation
of the difference in bone volume between the two groups
was statistically significant (p!.05) (Fig. 9). Compared with
both control specimens treated with Matrigel alone, or with
Matrigel with cells cultured in monolayer, there was a statis-
tically greater volume of bone present in specimens treated
by MA at all time points (Fig. 10).
Histological sections of in vivo muscle pouch specimens
Two weeks after surgical implantation, samples were
harvested and underwent H&E staining. The 2-week time
point was chosen based on microCT results demonstrating
bone formation in all MA specimens. In all specimens, the
implanted cells, the unresorbed Matrigel, and a cuff of
normal tissue were identified confirming complete harvest
of the surgical site. Histological staining from both the
monolayer- and the MA-implanted specimens identified
the surrounding muscle from the implanted cells, collagen,
and mineralized matrix. The overall size of collagen and
implanted cells was notably increased in the MA group
compared with the monolayer group. Staining confirmed
an increased number of cellular elements and greater vol-
ume of collagen in the MA- over monolayer-implanted
group (Fig. 11).
Healthy mammalian cells in normal tissues are orga-
nized in complex 3D networks that display nutrient and sig-
naling gradients . Therefore, conventional techniques
that culture cells in a 2D monolayer fail to reproduce the
environment that is observed in vivo. This may be reflected
as a reduced cellular replicative ability, decreased colony-
forming efficiency, and loss of differentiation capabilities
of multipotential cells cultured in monolayer over time
[12,27]. In an attempt to address these issues, organotypic
modeling aims to replicate a cellular culture system that
Fig. 6. Three-dimensional multicellular aggregate containing 50,000 adipose-derived stromal cells cultured in (Left) control medium alone or (Right) 4
weeks in osteogenic medium. Notice the abundant Alizarin Red staining confirming calcium deposition in the osteogenic medium group.
Fig. 7. Comparison of (Left) ALP and (Right) OCN expressions between human adipose-derived stromal cells cultured as monolayer and multicellular ag-
gregates (MAs). Multicellular aggregates demonstrated significantly increased ALP expression at 2 and 4 weeks and increased OCN expression at 4 weeks.
39 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
is based on what occurs in native tissues and organs
Because one of the important properties of stem cells for
tissue engineering is the ability to undergo multipotential
differentiation, it was important to confirm that the har-
vested ADS cells can undergo osteogenic differentiation
in both monolayer culture and as a 3D MA. To confirm
whether osteogenic differentiation had occurred, we se-
lected several well-described early and late markers of os-
teogenesis [7,8,30]. In addition, histological sections
verified the presence of calcified matrix from cells in both
monolayer culture and as a 3D MA.
Our study confirmed that the reorganization of ADS
cells into a 3D MA body did not disrupt its ability to
undergo osteogenic differentiation. Furthermore, when re-
seeded back onto tissue culture dishes, they formed conflu-
ent cells in monolayer, and on stimulation with specific
differentiation medium, they were able to once again un-
dergo multilineage differentiation. Perhaps, one of the most
fascinating aspects of these in vitro experiments is best cap-
tured in the image of the MA as it disperses back into
a monolayer. Thousands of cells can be seen extending
pseudopods from the MA as they reattach to the monolayer
(Fig. 12). Interestingly, this process occurs naturally with-
out the need for trypsin or any other enzymatic agents.
By maintaining the ability to renew and continue to un-
dergo multipotential differentiation, this series of in vitro
experiments clearly demonstrates that ADS cells preserve
not only their stem cell capabilities in 3D culture but also
through multiple transitions from monolayer to 3D culture
and back again. Over time, the cells in 3D culture had a rel-
atively greater increase in their gene expression for markers
of osteogenic differentiation and matrix mineralization
compared with those cultured in monolayer. The results
were normalized to the total number of cells; thus, these
findings suggest that cells cultured in a 3D system demon-
strate increased differentiation potential and ECM produc-
tion than those in 2D monolayer. The increased gene
expression and elevated protein production could be ob-
served on both the histological sections and the microCT
images as well.
Although mechanistically unclear, these results suggest
that 3D culture systems do not themselves necessarily influ-
ence stromal cells to undergo differentiation but that the in-
creased levels of osteogenic expression were a direct result
of the cells in the 3D environment. If this were not the case,
then we would expect the cells to lose their multipotential
Fig. 8. Multicellular aggregates cultured in osteogenic medium (A), adipogenic medium (B), or chondrogenic medium (C and D) differentiated into oste-
oblasts, adipocytes, and chondrocytes, respectively, indicating multilineage differentiation.
Fig. 9. Quantification of bone volume in the muscle pouch specimens of
the monolayer and multicellular aggregate group. Although significant var-
iance was noted, there was no bone formation in any of the monolayer
specimens, making the difference between the monolayer and multicellular
aggregate culture groups statistically significant (p!.05).
40 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
properties and remain differentiated and continue to express
elevated levels even after reseeding into monolayer. This
supports the concept that increased osteogenic differentia-
tion and matrix mineralization may be intrinsic to the 3D
culture system and not because of permanent alterations
to the ADS cells themselves. It is likely that the elevated
levels are because of the fact that 3D culture systems
closely resemble a normal physiological microenvironment
in several ways.
teractions than it can occur in a 2D monolayer. This is be-
cause cells in a spheroid are in closer association with one
another and in contact with a greater number of cells than
those in monolayer. This allows for improved intercellular
signaling than cells in monolayer, where secreted molecules
must be present in high concentrations to ensure effective
communication . Second, the cell-to-ECM interactions
are maintained more effectively than in conventional mono-
layer culture techniques. It is well known that the ECM
servesas a scaffoldandmodulatorforcellulargrowth,prolif-
eration, and differentiation . Last, the change in cellular
phenotype likely alters the form and the function of the cells.
The ADS cells in the MA ‘‘ball up’’ and become more
rounded in comparison to the flatter cells seen in monolayer,
and cell shape alone has been demonstrated to regulate the
gene expression and protein production in cells .
To better characterize the MA ability to undergo osteo-
genic differentiation, we undertook a series of in vivo in-
vestigations as well. The muscle pouch model was
specifically selected as the experiment of choice because
of its highly challenging environment for osteogenesis
. By placing the MA into a relatively avascular inter-
muscular plane, this model eliminates the potential contri-
bution of cells from the vasculature, more specifically in
the form of pericytes. Furthermore, because the MAs are
implanted in a site remote from any osseous structure
and not directly onto a decorticated bony surface, it elimi-
nates contributions from osteocytes, osteoblasts, and
Fig. 10. Two-week microCT images of 2?106cells surgically implanted
in muscle pouch as (Top) monolayer culture or (Bottom) multicellular ag-
gregates (MAs). There was a considerable amount of bone volume in the
muscle pouch at all time periods in the specimens that received the MA
(Bottom), whereas animals that received carrier alone or carrier with cells
from monolayer culture demonstrated no bone on microCT imaging (Top).
Fig. 11. In vivo histological sections with hematoxylin and eosin (H&E) or Masson’s trichrome of implanted cells in (Top) monolayer and (Botom) mul-
ticellular aggregates (MAs). Note the increased number of cellular elements and greater volume of collagen in the MA- over monolayer-implanted group.
41 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
pro-osteoblasts from the periosteum and surrounding corti-
cal or cancellous bone as well.
In this study, the MAs used contained 50,000 cells with
an average diameter of 300 to 500 mm placed within 20
mL. The advantage of the creation of uniform spheroids
allows for improved handling and manipulation during
transplantation. The MAs are cultured in a scaffold-free
environment; thus, there is no need to lift the cells from
an adherent monolayer, keeping the surrounding ECM in-
tact and maintaining the osteogenic micromilieu. Whereas
apparent during the in vitro studies, the clinical signifi-
cance of this becomes more important in the in vivo
Both groups received the same number of cells at the
time of initial in vivo implantation. Therefore, the only im-
mediate difference between the two was not the number of
cells delivered but the form in which they were implanted.
Cells from the monolayer were placed into suspension after
chemical and mechanical disruption of the ECM. This is
completely different from the 3D culture system where
the ECM and cells were implanted together as the MAs.
Because MAs are cultured scaffold free, there was no need
to physically or chemically disrupt the ECM to obtain the
cells. As a result, even though both had the same number
of cells at initial implantation, with the ECM intact, the
MAs were already in a more osteogenic micromilieu before
When this was evaluated using microCT imaging, there
was a considerable amount of bone volume in the muscle
pouch at all time periods in the specimens that received
the MAs, whereas animals that received carrier alone or
carrier with cells from monolayer culture demonstrated
no bone on microCT imaging. Compared with both control
specimens treated with Matrigel alone, or with Matrigel
with cells cultured in monolayer, there was a statistically
greater volume of bone present in specimens treated by
MA at all time points.
Furthermore, histological sections of the MA both
in vitro and in vivo confirmed that the MAs are solid aggre-
gates and not an outer rim of cells and ECM with a necrotic,
apoptotic center. In vitro histology demonstrated robust cal-
cification and matrix mineralization at 4 weeks. Interest-
ingly, calculations of the bone volume on microCT do not
match 100% of the expected volume of the implanted spec-
imen, which most likely reflects a portion of the healing
that is cartilaginous. This would be expected for bone that
forms through endochondral ossification and was confirmed
on 2-week in vivo histology where a portion of the MAs
served as cartilage anlages. Regardless, based on microCT
and histological sectioning, there was statistically greater
number of cells, increased ECM, and evidence of osteogen-
esis in the MA group compared with control and cells in
monolayer (Fig. 7).
The clinical significance of these findings is several fold.
First, the MAs are much easier to handle and manipulate
in vivo compared with monolayer-cultured cells and bone
marrow aspirate and allow direct implantation on a biologic
surface. In addition to improved handling and manipula-
tion, the MAs demonstrate greater, more efficient, and more
rapid osteogenic differentiation, which could lead to quick-
er spine fusion in a clinical setting. The burden and po-
tential complications for patients are significantly less
compared with bone marrow stromal cells, as the harvest
for these cells involves simple liposuction.
Bonemorphogenetic protein2 itselfinduces bone formation;
therefore, it would have been beneficial to evaluate the oste-
ogenic potential of the implanted MAs in vivo without addi-
tion of rhBMP2. The time point of 2 weeks for the in vivo
studies was chosen because the experiments were designed
to evaluate early bone formation; however, the authors ac-
knowledge that typically most in vivo experiments use 4-
to 8-week time points. Bone formation at the 4- or 8-week
time points may have demonstrated different results. Finally,
the authors chose to evaluate a human cell line from a single
donor to eliminate between-donor differences. The authors
acknowledge that different donors may have different osteo-
genic capabilities that could lead to different results when
experiments are repeated with different cell lines.
study to investigate the use of human ADS cells in a 3D cul-
ture system to undergo osteogenic differentiation. The re-
sults of this study confirm that human ADS cells as MAs
maintain their stem cell characteristics and can undergo
self-renewal and multipotential differentiation. Further-
more, compared with analogous cells treated in monolayer,
those grown in the 3D culture system exhibit markedly ele-
vated levels of osteogenic differentiation and increased
Fig. 12. Multicellular aggregate (MA) 2 weeks after reseeding onto ad-
herent monolayer culture. Pseudopods can be seen extending out from
MA as they reattach to the monolayer. The MA acts as an adipose-
derived stromal ‘‘seed’’ of cells for the culture plate.
42F.H. Shen et al. / The Spine Journal 13 (2013) 32–43
matrix mineralization both invitro and invivo. More impor- Download full-text
tantly, the creation of uniform spheroid allows for improved
handling and manipulation during transplantation, where
they readily undergo osteogenesis. These findings strongly
support the concept that 3D culture systems remain not only
aviable option butalso possibly a more attractivealternative
to more traditional culture techniques and should be investi-
 Mauney JR, Volloch V, Kaplan DL. Role of adult mesenchymal stem
cells in bone tissue engineering applications: current status and future
prospects. Tissue Eng 2005;11:787–802.
 Benayahu D. The hematopoietic microenvironment: the osteogenic
compartment of bone marrow: cell biology and clinical application.
 Heary RF, Schlenk RP, Sacchieri TA, et al. Persistent iliac crest donor
site pain: independent outcome assessment. Neurosurgery 2002;50:
 Banwart JC, Asher MA, Hassanein RS. Iliac crest bone graft harvest
donor site morbidity. A statistical evaluation. Spine 1995;20:
 Parker AM, Katz AJ. Adipose-derived stem cells for the regeneration
of damaged tissues. Expert Opin Biol Ther 2006;6:567–78.
 Strem BM, Hicok KC, Zhu M, et al. Multipotential differentiation of
adipose tissue-derived stem cells. Keio J Med 2005;54:132–41.
 Shen FH, Zeng Q, Lv Q, et al. Osteogenic differentiation of adipose-
derived stromal cells treated with GDF-5 cultured on a novel three-
dimensional sintered microsphere matrix. Spine J 2006;6:615–23.
 Zeng Q, Li X, Choi L, et al. Recombinant growth/differentiation
factor-5 stimulates osteogenic differentiation of fat-derived stromal
cells in vitro. Connect Tissue Res 2006;47:264–70.
 Yoshimura K, Eto H, Kato H, et al. In vivo manipulation of stem cells
for adipose tissue repair/reconstruction. Regen Med 2011;6:33–41.
 Levi B, Hyun JS, Nelson ER, et al. Nonintegrating knockdown and
customized scaffold design enhances human adipose-derived stem
cells in skeletal repair. Stem Cells 2011;29:2018–29.
 Fraser JK, Zhu M, Wulur I, et al. Adipose-derived stem cells.
Methods Mol Biol 2008;449:59–67.
 Kabiri M, Kul B, Lott WB, et al. 3D mesenchymal stem/stromal cell
osteogenesis and autocrine signalling. Biochem Biophys Res Com-
 Baraniak PR, McDevitt TC. Scaffold-free culture of mesenchymal
stem cell spheroids in suspension preserves multilineage potential.
Cell Tissue Res 2012;347:701–11.
 Baer PC, Griesche N, Luttmann W, et al. Human adipose-derived
mesenchymal stem cells in vitro: evaluation of an optimal expansion
medium preserving stemness. Cytotherapy 2010;12:96–106.
 Park E, Patel AN. Changes in the expression pattern of mesenchymal
and pluripotent markers in human adipose-derived stem cells. Cell
Biol Int 2010;34:979–84.
 Huang CP, Lu J, Seon H, et al. Engineering microscale cellular niches
for three-dimensional multicellular co-cultures. Lab Chip 2009;9:
 Marushima H, Shibata S, Asakura T, et al. Three-dimensional culture
promotes reconstitution of the tumor-specific hypoxic microenviron-
ment under TGFbeta stimulation. Int J Oncol 2011;39:1327–36.
 Ismail S, Samadikuchaksaraei A, Bishop A, et al. Development of
a novel three-dimensional, automatable and integrated bioprocess
for the differentiation of embryonic stem cells into pulmonary alve-
olar cells in a rotating vessel bioreactor system. Tissue Eng Part C
 Ou DB, He Y, Chen R, et al. Three-dimensional co-culture facilitates
the differentiation of embryonic stem cells into mature cardiomyo-
cytes. J Cell Biochem 2011;112:3555–62.
 Katz AJ, Tholpady A, Tholpady SS, et al. Cell surface and transcrip-
tional characterization of human adipose-derived adherent stromal
(hADAS) cells. Stem Cells 2005;23:412–23.
 Banerjee M, Bhonde RR. Application of hanging drop technique for
stem cell differentiation and cytotoxicity studies. Cytotechnology
 Amos PJ, Kapur SK, Stapor PC, et al. Human adipose-derived stro-
mal cells accelerate diabetic wound healing: impact of cell formula-
tion and delivery. Tissue Eng Part A 2010;16:1595–606.
 Kelm JM, Timmins NE, Brown CJ, et al. Method for generation of
homogeneous multicellular tumor spheroids applicable to a wide va-
riety of cell types. Biotechnol Bioeng 2003;83:173–80.
 Medvinskii LA, Kolesnichenko TS. [Modification of the ‘‘hanging
drop’’ method and its application to cultivation of reaggregates of dis-
sociated cells and explants of embryonic lungs]. Biull Eksp Biol Med
 Cheng NC, Wang S, Young TH. The influence of spheroid formation
of human adipose-derived stem cells on chitosan films on stemness
and differentiation capabilities. Biomaterials 2012;33:1748–58.
 Chandrasekaran S, Giang UB, King MR, et al. Microenvironment
induced spheroid to sheeting transition of immortalized human kera-
tinocytes (HaCaT) cultured in microbubbles formed in polydimethyl-
siloxane. Biomaterials 2011;32:7159–68.
 Gruber HE, Hanley EN Jr. Human disc cells in monolayer vs 3D cul-
ture: cell shape, division and matrix formation. BMC Musculoskelet
 Papadimitropoulos A, Scherberich A, Guven S, et al. A 3D in vitro
bone organ model using human progenitor cells. Eur Cell Mater
 Gurkan UA, Kishore V, Condon KW, et al. A scaffold-free multicel-
lular three-dimensional in vitro model of osteogenesis. Calcif Tissue
 Zeng Q, Li X, Beck G, et al. Growth and differentiation factor-5
(GDF-5) stimulates osteogenic differentiation and increases vascular
endothelial growth factor (VEGF) levels in fat-derived stromal cells
in vitro. Bone 2007;40:374–81.
 Bartosh TJ, Ylostalo JH, Mohammadipoor A, et al. Aggregation of
human mesenchymal stromal cells (MSCs) into 3D spheroids en-
hances their antiinflammatory properties. Proc Natl Acad Sci USA
 Wolf MT, Daly KA, Reing JE, et al. Biologic scaffold composed
of skeletal muscle extracellular matrix. Biomaterials 2012;33:
 Scherberich A, Muller AM, Schafer DJ, et al. Adipose tissue-derived
progenitors for engineering osteogenic and vasculogenic grafts. J Cell
43 F.H. Shen et al. / The Spine Journal 13 (2013) 32–43