This is the author’s version of a work that was submitted/accepted for pub-
lication in the following source:
Fan, Wei, Crawford, Ross, & Xiao, Yin (2011) The ratio of VEGF/PEDF
expression in bone marrow mesenchymal stem cells regulates neovascu-
larization. Differentiation, 81(3), pp. 181-191.
This file was downloaded from: http://eprints.qut.edu.au/41403/
c ? Crown copyright 2010 Published by Elsevier Inc.
Notice: Changes introduced as a result of publishing processes such as
copy-editing and formatting may not be reflected in this document. For a
definitive version of this work, please refer to the published source:
The ratio of VEGF/PEDF expression in bone marrow mesenchymal
stem cells regulates neovascularization
Wei Fan1, Ross Crawford1, Yin Xiao1
1Institute of Health and Biomedical Innovation, Queensland University of Technology,
Running title: BMSCs in neovascularization
Key words: Bone marrow, stem cells, angiogenesis, VEGF, PEDF
Dr. Yin Xiao
Institute of Health and Biomedical Innovation, Queensland University of Technology,
Kelvin Grove Campus, Brisbane, Qld 4059 Australia
Tel: +61 7 3138 6240
Fax: +61 7 3138 6030
Angiogenesis, or neovascularization, is a finely balanced process controlled by pro-
and anti-angiogenic factors. Vascular endothelial growth factor (VEGF) is a major
pro-angiogenic factor, whereas pigment epithelial-derived factor (PEDF) is the most
potent natural angiogenesis inhibitor. In this study, the regulatory role of bone
marrow stromal cells (BMSCs) during angiogenesis was assessed by the endothelial
differentiation potential, VEGF/PEDF production and responses to pro-angiogenic
and hypoxic conditions. The in vivo regulation of blood vessel formation by BMSCs
was also explored in a SCID mouse model. Results showed that PEDF was
expressed more prominently in BMSCs compared to VEGF. This contrasted with
human umbilical vein endothelial cells (HUVECs) where the expression of VEGF
was higher than that of PEDF. The ratio of VEGF/PEDF gene expression in BMSCs
increased when VEGF concentration reached 40 ng/ml in the culture medium, but
decreased at 80 ng/ml. Under CoCl2-induced hypoxic conditions, the VEGF/PEDF
ratio of BMSCs increased significantly in both normal and angiogenic culture media.
There was no expression of endothelial cell markers in BMSCs cultured in either
pro-angiogenic or hypoxia culture conditions when compared with HUVECs. The in
vivo study showed that VEGF/PEDF expression closely correlated with the degree of
neovascularization, and that hypoxia significantly induced pro-angiogenic activity in
BMSCs. These results indicate that, rather than being progenitors of endothelial
cells, BMSCs play an important role in regulating the neovascularization process, and
that the ratio of VEGF and PEDF may, in effect, be an indicator of the pro or
anti-angiogenic activity of BMSCs.
Angiogenesis or neovascularization is thought to be a regulated process between pro-
and anti-angiogenic factors. Vascular endothelial growth factor (VEGF) is one of the
most important pro-angiogenic factor and can initiate angiogenic differentiation of
endothelial progenitor cells (EPCs) (Koch et al., 2006; Asahara and Kawamoto,
2004). VEGF is expressed by a wide range of cells including bone marrow derived
mesenchymal stem cells (BMSCs) (Farhadi et al., 2005). Pigment epithelial-derived
factor (PEDF) was originally purified from the conditioned media of human fetal
retinal pigment epithelial (RPE) cells, and is the most potent natural angiogenesis
inhibitor (Ohno-Matsui et al., 2003). PEDF is thought to be a key factor associated
with avascularity of the cornea (Ohno-Matsui et al., 2003) and the balance between
VEGF and PEDF expression plays a crucial role in retinal vascularization (Zhang et
al., 2006). It is therefore clear that the balance between VEGF and PEDF is vital in
the processes of angiogenesis and neovascularization. There is, however, little
information to be found in the scientific literature concerning VEGF and PEDF
expression in BMSCs with respect to angiogenesis and neovascularization, but it is
known that BMSCs are involved in a number of pro-angiogenic and anti-angiogenic
activities (Chen et al., 2003). It is therefore of interest to explore how VEGF and
PEDF expression patterns affect the fate of BMSCs in the regulation of angiogenesis
The bone marrow comprises two dominant stem cell populations, namely
hematopoietic stem cells and mesenchymal stem cells (Bianco et al., 2001; Bianco et
al., 2006). When cultured in vitro, the latter rapidly adhere to the tissue culture
substrate and is thus easily separated from the non-adherent hematopoietic cells
through repeated media changes. Hematopoietic stem cells (HSCs) serve as the
reservoir for the various blood cells, such as erythrocytes, leukocytes, macrophages or
platelets. They also contain endothelial progenitor cells (EPCs), which are capable of
differentiating into mature endothelial cells (Loomans et al., 2006; Schatteman et al.,
2007). BMSCs are well characterized and are negative for CD45, CD14, CD31 and
CD34, but positive for CD105, CD 44, CD73 and CD90. They are a heterogeneous
mix of multipotent progenitor cells, capable of differentiating into mesodermal cell
lineages, such as osteoblasts, chondrocytes, fibroblasts and adipocytes (Reyes et al.,
2001). However, whether BMSCs are capable of differentiating into endothelial cell
lineages has yet to be definitively proven and therefore remains an open question
(Oswald et al., 2004; Zhang et al., 2007).
In this study we investigated, both in vitro and in vivo, the differentiation potential of
BMSCs into endothelial cells and the regulatory role in angiogenesis in response to
pro-angiogenic and hypoxic conditions.
Materials and Methods
Cells and cell culture
Human bone marrow was sourced from patients undergoing elective surgery at the
orthopaedics department at the Prince Charles Hospital (PCH) in Brisbane,
Queensland, Australia. Informed consent was given by all participants and the
project had approval from the ethics committees of the PCH and the Queensland
University of Technology. Mononuclear cells (MNCs) were isolated from the bone
marrow by density gradient centrifugation over Lymphoprep (Axis-Shield PoC AS,
Oslo, Norway) according to the manufacturer’s protocol and plated out in tissue
culture flasks in low glucose Dulbecco’s Modified Eagle Medium (DMEM;
Invitrogen Australia Pty Ltd., Mt Waverley, VIC, Australia) containing 10% (v/v) fetal
calf serum (FCS; InVitro Technology, Noble Park, VIC, Australia) and 1% (v/v)
penicillin/streptomycin (Invitrogen). Unattached hematopoietic cells were removed
by subsequent media changes. When reaching the 70-80% confluence, the attached
mesenchymal cells were subcultured at a seeding density of 3×103 / cm2 after
treatment with 0.25% Trypsin/EDTA (Invitrogen). Only early passage cells (P2-P5)
were used in this study. Human umbilical vein endothelial cells (HUVECs;
Clonetics, San diego, CA, USA) were used as positive controls when required.
HUVECs were cultured in a defined endothelial cell growth medium (ECGM)
containing VEGF, FGF-2, IGF-1, EGF, ascorbic acid and hydrocortisone (EGM-2;
Lonza Australia Pty Ltd., Mt Waverley, VIC, Australia) supplemented with 2% FCS.
The medium was changed for both BMSCs and HUVECs every three days until the
cells had reached confluence.
Endothelial cell differentiation of BMSCs under the stimulation of VEGF
BMSCs sourced from five patients (one female and four males ranging from 40 to 78
years of age, average of 61 years) were included in this experiment. At
approximately 80% confluence, angiogenic differentiation media, low glucose
DMEM supplemented with 5% FCS and 20, 40 or 80 ng VEGF (R&D Systems Inc.,
Minneapolis, MN, USA) per ml(Oswald et al., 2004; Zhang et al., 2007; Xu et al.,
2009), were applied to the cells. The media were replenished halfway through a 6-day
culture period, after which the cells were harvested and subjected to real time
quantitative PCR (RT-qPCR), western blot, enzyme-linked immunosorbent assay
(ELISA), and immunohistochemistry. The ability of these BMSCs to form in vitro
vessel-like structures was tested on a Matrigel substrate (Oswald et al., 2004; Xu et
al., 2009)(BD biosciences, North Ryde, NSW, Australia).
Endothelial cell differentiation of BMSCs under cobalt chloride-induced hypoxia
BMSCs sourced from six patients (two female and four male patients ranging from 47
to 76 years of age, average 57 years) were used in this experiment. The cells were
grown to 80% confluence and then subcultured into the following three groups: (i)
BMSCs cultured in normal DMEM containing 5% FCS; (ii) BMSCs cultured in
DMEM containing 5% FCS and 100 µM cobalt chloride; and (iii) BMSCs cultured in
DMEM containing 5% FCS, 100 µM cobalt chloride and 20 ng VEGF/ml. The
media were changed every three days and on day 6, RT-qPCR, western blot, ELISA
and immunohistochemistry were conducted. The ability of these BMSCs to form in
vitro vessel-like structures was tested on a Matrigel substrate.
Real time quantitative polymerase chain reaction
Total RNA was extracted using Trizol reagent (Invitrogen) and cDNA synthesized
from 1 µg of RNA using SuperScript III reverse transcriptase (Invitrogen).
RT-qPCR was performed with SYBR Green: 12.5 µl 2X SYBR Green QPCR master
mix (Roche, Castle Hill, NSW, Australia) was mixed with 5 µl water, 2.5 µl reverse
and forward primers (Sigma-Aldrich Pty. Ltd, Castle Hill, NSW, Australia) and 2.5 µl
cDNA template for a 25 µl final volume in a 96-well PCR plate. The mRNA
expressions of PEDF, VEGF, CD31, VEGF receptor 2 (VEGFR2), von Willebrand
factor (vWF) and endothelium nitric oxide synthases III (eNOSIII) were assayed and
normalized against the 18s house keeping gene. Forward and reverse primers of each
assayed gene are detailed in Table 1. Each sample was performed in triplicate and
the reactions were run on ABI Prism 7300 sequence detection system (AB Applied
Biosystems, Melbourne, Australia). The mean cycle threshold (Ct) value of each
target gene was normalized against Ct value of 18s; the relative expression calculated
using the following formula: 2 -(normalized average Cts) × 104.
Total protein was harvested in a cell lysis buffer supplemented with a proteinase
inhibitor cocktail (Roche Products Pty. Ltd., Dee Why, NSW., Australia). The
protein concentration was determined with a BCA protein assay kit (Sigma) and 10 µg
proteins from each sample were separated on SDS-PAGE gels. The proteins were
transferred onto a nitric cellulous membrane (Pall corporation, East Hills, NY, USA)
by semi-dry transfer method. The membranes were probed with hypoxia inducible
factor 1 alpha (HIF-1α) (1:1000, rabbit anti-human; Santa Cruz Biotechnology Inc.,
Santa Cruz, CA. U.S.A.), VEGF (1:2,000, rabbit anti-human; Thermo Fisher
Scientific, Fremont, CA, USA), PEDF(1:2,000, mouse anti-human; Millipore, North
Ryde, NSW, Australia), and α-tubulin (1:5,000, rabbit anti-human; Abcam Inc.,
Cambridge, MA, USA) primary antibodies. These were bound to HRP-conjugated
anti-mouse or rabbit secondary antibodies (1:10,000; Thermo Fisher Scientific,
Fremont, CA, USA). The bands were visualized using Super-Signal substrate
(Thermo Fisher Scientific, Fremont, CA, USA) and images captured on X-ray films.
Enzyme-linked immunosorbent assay (ELISA)
The protein expression of VEGF and PEDF by BMSCs under different treatment
conditions were tested using ELISA assay kits for VEGF (R&D Systems Inc.) and
PEDF (CHEMICON, Millipore) . Briefly, BMSCs cultured in flasks were thoroughly
washed with phosphate buffer saline (PBS) for three times and cell lysate was
produced by adding the lysis buffer supplemented with a proteinase inhibitor cocktail
(Roche) and shaking at 4°C for 1h. Then the lysate was transferred into eppendorf
tubes and spinned at 10,000 g for 10mins to remove cell debris. The supernatant was
collected for the ELISA tests. The test was performed in triplicates and results were
expressed as the amount (pg) of VEGF or PEDF in per µg total cell protein, which
was measured by the BCA protein assay kit (Sigma).
BMSCs and HUVECs were cultured in chamber slices and fixed in 4%
paraformaldehyde and washed at least three times with PBS. The cells were
permeablized with 0.1% Triton X solution for 6 min and endogenous peroxidase
activity quenched by incubating the sample slices with 3% H2O2 for 15 min, then
blocked with 10% swine serum for 1 h. The cells were incubated with the vWF
(Rabbit anti-human, 1:300, Millipore) and VEGFR2 (goat anti-human, 1:100, R&D
Systems, Inc., Minneapolis, MN, USA), VEGF (rabbit anti-human, 1:50, Thermo
Fisher Scientific) and PEDF (mouse anti-human, 1:100, Millipore) primary antibodies
overnight at 4°C, followed by incubation with the biotinylated swine-anti-mouse,
rabbit, goat universal secondary antibody (DAKO Multilink, CA, USA) for 15 min,
and then with horseradish peroxidase-conjugated avidin-biotin complex (DAKO
Multilink, CA, USA) for another 15 min. The slides were counter stained with
Mayer’s haematoxylin (HD Scientific Pty Ltd., Kings Park, NSW, Australia) and the
antibody complexes were visualized by the addition of a buffered diaminobenzidine
(DAB) substrate for 4 min.
Aliquots (50 µl) of growth factor-reduced phenol red free Matrigel matrix (BD
biosciences, North Ryde, NSW, Australia) were added into the wells of 96-well cell
culture plates and allowed to polymerize at 37°C for a minimum of 30 min. 3×103
BMSCs (randomly sourced from one patient described above), HUVECs or BMSCs
cultured under different conditions for 6 days, i.e. with different VEGF concentrations
or under cobalt chloride-induced hypoxia, were added onto the matrigel matrix and
original culture conditions were maintained. The new vessel-like networks formed on
the Matrigel was observed using a microscope at 40× magnification after 6 h
incubation at 37°C and 5% CO2 condition. To study the interactions between the
HUVECs and BMSCs, 3 ×103 HUVECs were mixed with either CoCl2-treated
BMSCs or untreated BMSCs in serum-free DMEM at the ratio of 1:1, and seeded
onto the Matrigel. Unmixed HUVECs and CoCl2-treated BMSCs were used as
controls. Vessel networks formed on the Matrigel by different cell combinations were
photographed after 24 and 72 h incubation. The total number of vessels in five
randomly selected areas from each well was recorded using the Axion software (Carl
Zeiss Microimaging GmbH, Göttingen, Germany). The experiment was done in
triplicates and the average was taken for statistical analysis.
In vivo vascularization assessment
Three female 6 week-old severe combined immunodeficient (SCID) mice (Animal
Resources Centre, Canning Vale, WA, Australia) were used to assess the in vivo
regulatory role of BMSCs in neovascularization. Ethics approval for this experiment
was granted from the QUT Animal Ethics Committee. The animals were
anesthetized with 10 µl/g bodyweight of a mixture of ketamine (100 mg/ml) and
xylazine (20 mg/ml), injected intraperitoneally. A total of 5 × 104 cells of one of five
conditions: (i) BMSCs, (ii) HUVECs, (iii) BMSCs treated with 20 ng/ml VEGF, (iv)
BMSCs treated with 100 µM cobalt chloride or (v) BMSCs treated with both 20
ng/ml VEGF and 100 µM cobalt chloride, were mixed with Matrigel at a ratio of 1:2
to reach a final volume of 300 µl. Gel without cells was used as a negative control.
Each mouse was injected with six gels (one from each group) with three on either side
of the dorsal area, approximately 1 cm apart. All animals had recovered by the
following day, and were sacrificed after 10 days and the implants were retrieved,
photographed and fixed in 4% paraformaldehyde. After paraffin embedding, the
implants were sectioned and three serial sagittal slices, close to the centre of each
implant, used for immunohistochemical staining. A vWF antibody (rabbit
anti-human, 1:300, Millipore) was used to detect the endothelial cells, and all vWF+
positive cells, capillaries or blood vessels were counted on each slice and normalized
to the slice area (mm2). The average from each group was used for statistical
analysis. To determine the relationship of VEGF vs. PEDF expression and the
degree of neovascularization, the VEGF (1:50) and PEDF (1:100) antibodies were
used to stain the slices.
Analysis was performed using SPSS software (SPSS Inc., Chicago, Il., USA). All the
data was analyzed using Student-t, one-way ANOVA or Friedman test. The
significance level was set at p ≤0.05.
VEGF and PEDF expressions in BMSCs and HUVECs
RT-qPCR revealed that gene expression of both VEGF and PEDF in BMSCs was
significantly higher than that in HUVECs (Student-t test, p<0.01) (Fig. 1A), and also
that the ratio of VEGF to PEDF differed between the two cell types. The PEDF
expression in BMSCs was significantly higher compared to VEGF (Student-t test,
p<0.05) (Fig. 1A), whereas in HUVECs the PEDF expression was significantly less
than the VEGF expression (Student-t test, p<0.05) (Fig. 1A). The average
VEGF/PEDF expression ratio in HUVECs was around 8.0, which was 16 times
greater than that (around 0.5) of BMSCs (Student-t test, p<0.01) (Fig. 1B). Western
blot analysis showed stronger expression of PEDF and VEGF in BMSCs compared
with HUVECs (Fig. 1C). Immunohistochemical staining of VEGF and PEDF
confirmed the high expressions in BMSCs and weak VEGF and PEDF expressions
were noted in HUVECs (Fig. 1D).
The VEGF/PEDF expression pattern and endothelial cell differentiation of
BMSCs under different extracellular VEGF concentrations
When cells were cultured in angiogenic media with increasing VEGF concentrations,
the VEGF/PEDF gene expression in BMSCs showed an interesting pattern (Fig 2).
The VEGF expression increased to around 1.5 times the original expression at 20
ng/ml extracellular VEGF, returning to its original expression when extracellular
VEGF increased from 40 ng/ml to 80 ng/ml (Friedman test, p<0.05) (Fig.2 A).
When the extracellular VEGF increased from 0 to 40 ng/ml, PEDF expression
decreased by nearly half the original expression level. At 80 ng/ml VEGF, however,
the PEDF expression returned to approximately 75% of its original expression
(Friedman test, p<0.05) (Fig.2 B). The VEGF/PEDF gene expression ratio reached a
peak average value of 0.8 at 40 ng/ml extracellular VEGF (Friedman test, p<0.05)
(Fig.2 D). Western blot image showed a similar pattern in VEGF/PEDF expression
ratio (Fig.2 C). ELISA results showed the general increasing trend for VEGF and
decreasing trend for PEDF despite the absence of statistical differences (Fig.4).
Endothelial differentiation of BMSCs at different extracellular VEGF concentrations
was also tested with RT-qPCR of the endothelial cell markers CD31, VEGFR2, vWF
and eNOSIII, with HUVECs serving as control. The gene expression of these
markers were all significantly higher in HUVECs than in BMSCs cultured in either
normal cell culture media or VEGF supplemented media (one-way ANOVA, p<0.01)
(Fig.2 E). There was no significant difference in the expression of those endothelial
cell markers between different BMSCs groups (one-way ANOVA, p>0.05) (Fig.2 E).
These findings were further confirmed by immunohistochemical staining against two
typical endothelial cell markers VEGFR2 and vWF (Fig. 5).
The VEGF/PEDF expression pattern and endothelial cell differentiation of
BMSCs under cobalt chloride induced hypoxia
The hypoxia induced by CoCl2 was firstly confirmed by the accumulated HIF-1α
protein within the CoCl2-treated BMSCs revealed by the western blot test (Fig.3 C).
When BMSCs were cultured under hypoxic condition induced by 100 µM CoCl2, the
VEGF mRNA expression increased, on average, nearly 5 fold (Fig.3 A) and reduced
PEDF mRNA expression by nearly 50% (Friedman test, p<0.05) (Fig.3 B). The
addition of 20 ng/ml VEGF did not significantly influenced the effect of CoCl2 on the
VEGF/PEDF gene expression (Friedman test, p>0.05) (Fig.3 A&B). The
VEGF/PEDF gene expression ratio increased to about 4.0 under CoCl2 culture
condition (Fig.3 F). PEDF and VEGF protein expression detected by western blot and
ELISA confirmed the gene expression pattern under the CoCl2-induced hypoxia
condition (Fig.3 D & Fig.4). Evidence of endothelial cell differentiation of BMSCs
under hypoxic conditions was tested with RT-qPCR of CD31, VEGFR2, vWF and
eNOSIII endothelial gene expression. There was no significant difference in the
expression of these markers between BMSC controls and BMSCs treated with either
100 µM CoCl2 or 100 µM CoCl2 plus 20 ng/ml VEGF (one-way ANOVA, p>0.05)
(Fig.3 E), and also confirmed by immunohistochemical staining against two typical
endothelial cell markers VEGFR2 and vWF (Fig. 5).
Enhanced vessel formation of HUVECs on Matrigel by CoCl2-treated BMSCs
The Matrigel assays showed that HUVECs formed new vessel-like structures more
readily within 6 h (Fig.6G), compared to undifferentiated BMSCs (Fig.6A) or
endothelial cell-differentiated BMSCs in 20 ng/ml VEGF (Fig.6B), 40 ng/ml VEGF
(Fig.6C), and 80 ng/ml VEGF (Fig.6D), as well as in CoCl2 (Fig.6E) or CoCl2 plus 20
ng/ml VEGF (Fig.6F). The number of vessels formed on Matrigel was significantly
higher in HUVECs (one-way ANOVA, p<0.01) (Fig.6P). There was no difference in
the ability to form vessel-like structures between BMSC groups (one-way ANOVA,
HUVECs, when mixed with CoCl2-treated BMSCs (Fig.6 H&L) in serum-free
DMEM, formed more stable vessel-like structures at both 24 (Fig.6 H-K) and 72 h
(Fig.6L-O) incubation on the Matrigel compared to HUVECs with untreated BMSCs
(Fig.6 I&M), CoCl2 treated BMSCs alone (Fig.6 J&N) or HUVECs alone (Fig.6
K&O). The number of vessel-like structures formed by HUVECs with CoCl2-treated
BMSCs was more than any other group after both 24 and 72 h incubation (one-way
ANOVA, p<0.001) (Fig.6Q).
In vivo neovascularization of BMSCs
The Matrigel implants were harvested from the SCID mice and images captured with
a stereomicroscope (Fig.7 a-f). There was no obvious neovascularization visible in
the Matrigel control (Fig. 7a), or in Matrigels containing untreated BMSCs (Fig. 7b)
or BMSCs treated with VEGF (Fig. 7c). Matrigels with CoCl2 treated BMSCs (Fig.
7d) or CoCl2+VEGF (Fig.7e), or HUVECs (Fig. 7f) showed an obvious blood vessel
formation inside implants, and this was confirmed by vWF staining of the samples.
Almost no vWF expression was detected in Matrigel without cells (Fig.7g), with
untreated BMSCs (Fig.7h) or with BMSCs treated with 20 ng/ml VEGF (Fig.7i).
vWF positive cells were found in Matrigel with BMSCs treated with CoCl2 (Fig.7j),
Matrigel with BMSCs treated with CoCl2 plus VEGF (Fig.7k) and Matrigel with
HUVECs (Fig.7l). Higher magnification (400x) of vWF staining of the Matrigel with
BMSCs treated with CoCl2 showed blood vessels and capillaries (Fig.7m) and a small
artery growing into the gel (Fig.7n). Matrigel containg HUVECs also showed positive
capillaries in the higher magnification (Fig.7o). The number of vWF positive cells and
blood vessels on each slice were counted and normalized against the area of each
slice. These results showed that Matrigels containing HUVECs, BMSCs treated
with CoCl2 alone or together with VEGF had significantly more vWF+ cells and
blood vessels in comparison with Matrigels containing untreated BMSCs or BMSCs
treated with only VEGF (Friedman test, p<0.01) (Fig.7p).
When stained with VEGF and PEDF it was revealed in non-vascularized areas of the
Matrigels containing untreated BMSCs that stromal cells showed stronger expression
of PEDF (Fig.8 A&B) compared to the highly vascularized areas of Matrigels
carrying CoCl2-treated BMSCs (Fig.8 C&D). In Matrigels containing CoCl2-treated
BMSCs, robust VEGF expression was seen in both newly formed blood vessel walls
(Fig.8 E) and in the surrounding stromal cells (Fig.8 F&G). The positive staining of
vWF was also noted in the endothelial cell of newly form blood vessel (Fig.8 H).
BMSCs are cells isolated from the bone marrow mononuclear cell and which attach to
cell culture plastic surfaces. It has been argued that in vitro cultured BMSCs might
be capable of differentiating into endothelial cells lineages, or in other words, that in
vitro cultured BMSCs contained endothelial progenitor cells (Oswald et al., 2004;
Zhang et al., 2007). To obtain BMSCs, the standard method for BMSCs isolation
and expansion was used in this study, and only cells from passage 2 to 5 were used in
an effort to eliminate HSCs contamination. Following a previously described
angiogenic differentiation protocol, which prescribes a 6 day angiogenic induction
period (Cipriani et al., 2007; Oswald et al., 2004; Zhang et al., 2007), BMSCs were
not capable of trans-differentiation into endothelial cells under either VEGF
stimulation or CoCl2-induced hypoxic conditions, based on criteria of endothelial cell
surface marker expression and Matrigel tube structure formation, when compared
with the endothelial cell line of HUVECs. An extended endothelial differentiation
period (2 and 3 weeks) was also investigated, including the application of
HUVECs-specific media (EGM-2), with similar negative results (data not shown).
BMSCs are reported to be involved in a number of tissue development and
regeneration processes, such as osteogenesis, chondrogenesis and angiogenesis (Xia
et al., 2008; Connelly et al., 2008; Lamagna and Bergers, 2006; Bexell et al., 2009;
Ozerdem et al., 2005). The role of BMSCs during angiogenesis is far from clear,
although some studies have shown that BMSCs may support and stabilize newly
formed blood vessels as pericytes (Lamagna and Bergers, 2006; Bexell et al., 2009;
Ozerdem et al., 2005). Within the angiogenic environment, VEGF is one of the major
pro-angiogenic growth factor that affect endothelial cell differentiation, migration and
blood vessel formation (Shibuya, 2008; Otrock et al., 2007; Yamazaki and Morita,
2006), and VEGF concentration tends to be elevated in the angiogenic areas
(Harlozinska et al., 2004). PEDF, on the other hand, is probably the most potent
angiogenesis inhibitor capable of preventing vascularisation and inducing apoptosis in
endothelial cells (Notari et al., 2006; Ohno-Matsui et al., 2003). The ratio or balance
between these two antagonistic factors (VEGF/PEDF) in a local environment
therefore has a great effect on angiogenesis and vascularization (Zhang et al., 2006).
VEGF expression in BMSCs has been investigated in the past (Sena et al., 2007;
Wrobel et al., 2003; Cai et al., 2002), but studies concerning the VEGF feedback loop
and interactions between VEGF and PEDF in BMSCs has not been forthcoming.
The findings reported here reveal for the first time that PEDF is much more strongly
expressed in BMSCs than is VEGF. Interestingly, VEGF expression in HUVECs,
although quite low when compared with BMSCs, is much higher than PEDF, the
VEGF/PEDF ratio being more or less diametrically opposite that of BMSCs. Together
these findings suggest that BMSCs may not be an angiogenesis-promoting cell
population, when in a normoxia and neutral environment. This view is supported by a
recent report which suggests that MSCs have a negatively regulatory role and are
capable of inhibiting capillary growth at high cell numbers (Otsu et al., 2009).
In a pro-angiogenic environment, extracellular VEGF concentration increases during
the angiogenic process (Harlozinska et al., 2004). In this study, it was found that
depending on the concentration, extracellular VEGF set up a general positive
feedback with the endogenous VEGF expression of BMSCs, although a certain high
level of extracellular VEGF concentration (80 ng/ml in this study) dropped the
endogenous VEGF production of the BMSCs; On the other hand, the increasing
extracellular VEGF concentration suppressed the PEDF expression of BMSCs.
These changes in the VEGF/PEDF expression pattern indicates that when
environmental VEGF is at an appropriate concentration, BMSCs may play a more
active role in the angiogenic process.
Cobalt chloride is a hypoxia mimicking agent commonly used to activate
hypoxia-related responses in cells (Lee et al., 2007). The biochemical and molecular
mechanisms of CoCl2 induced hypoxia have been shown to be similar to those in low
oxygen tension (Lee et al., 2007). Changes to gene expression in BMSCs as a result
of hypoxia has been reported (Ohnishi et al., 2007), however, little is known about
hypoxia effect on the endothelial cell differentiation and VEGF/PEDF expression
patterns in BMSCs. In this study we showed that CoCl2-induced hypoxia
significantly increases VEGF expression and suppresses PEDF expression, suggesting
that hypoxia drives BMSCs to favor angiogenesis and neovascularisation. We also
demonstrated that CoCl2-treated BMSCs appeared to enhance and stabilize the
vessel-like structure formed by HUVECs in Matrigel, further demonstrating a
supporting role for BMSCs in neovascularization when BMSCs were in a hypoxia
Our in vivo study showed that the VEGF/PEDF expression ratio in BMSCs was
closely correlated with neovascularization. The VEGF/PEDF ratio in undifferentiated
or VEGF-treated BMSCs was low (less then “1” at both gene and protein levels), and
no in vivo blood vessel formation was found after in vivo transplantation. However, in
CoCl2-induced hypoxia culture condition the VEGF/PEDF expression ratio in BMSCs
reversed up to around “4” at both gene and protein levels, and once transplanted a
higher degree of neovasculariztion was observed. The distribution of VEGF and
PEDF in the neovascularized areas further confirmed that more VEGF was expressed
in blood vessel forming areas, whereas more PEDF was expressed in areas of
relatively low vessel formation.
The findings in this study suggest that BMSCs are important regulators in the
neovascularization, rather than as a source of endothelial progenitor cell, under both
normoxia and hypoxia conditions. The ratio of VEGF and PEDF may therefore be
an indicator of the pro- and anti-angiogenic activity of BMSCs.
The authors wish to thanks Mr Thor Friis for his contribution in preparing this
manuscript. The work is supported by Queensland University of Technology.
Asahara, T., and Kawamoto, A. (2004) Endothelial progenitor cells for postnatal vasculogenesis. Am J
Physiol Cell Physiol 287:C572-579.
Bexell, D., Gunnarsson, S., Tormin, A., Darabi, A., Gisselsson, D., Roybon, L., Scheding, S., and
Bengzon, J. (2009) Bone marrow multipotent mesenchymal stroma cells act as pericyte-like
migratory vehicles in experimental gliomas. Mol Ther 17:183-190.
Bianco, P., Kuznetsov, S.A., Riminucci, M., and Gehron Robey, P. (2006) Postnatal skeletal stem cells.
Methods Enzymol 419:117-148.
Bianco, P., Riminucci, M., Gronthos, S., and Robey, P.G. (2001) Bone marrow stromal stem cells:
nature, biology, and potential applications. Stem Cells 19:180-192.
Cai, S., Ma, Q., Yu, X., Dang, G., and Ma, D. (2002) Expression of human VEGF(121) cDNA in mouse
bone marrow stromal cells. Chin Med J (Engl) 115:914-918.
Chen, J., Zhang, Z.G., Li, Y., Wang, L., Xu, Y.X., Gautam, S.C., Lu, M., Zhu, Z., and Chopp, M. (2003)
Intravenous administration of human bone marrow stromal cells induces angiogenesis in the
ischemic boundary zone after stroke in rats. Circ Res 92:692-699.
Cipriani, P., Guiducci, S., Miniati, I., Cinelli, M., Urbani, S., Marrelli, A., Dolo, V., Pavan, A., Saccardi,
R., Tyndall, A., Giacomelli, R., and Cerinic, M.M. (2007) Impairment of endothelial cell
differentiation from bone marrow-derived mesenchymal stem cells: new insight into the
pathogenesis of systemic sclerosis. Arthritis Rheum 56:1994-2004.
Connelly, J.T., Garcia, A.J., and Levenston, M.E. (2008) Interactions between integrin ligand density
and cytoskeletal integrity regulate BMSC chondrogenesis. J Cell Physiol 217:145-154.
Farhadi, J., Jaquiery, C., Barbero, A., Jakob, M., Schaeren, S., Pierer, G., Heberer, M., and Martin, I.
(2005) Differentiation-dependent up-regulation of BMP-2, TGF-beta1, and VEGF expression
by FGF-2 in human bone marrow stromal cells. Plast Reconstr Surg 116:1379-1386.
Harlozinska, A., Sedlaczek, P., Kulpa, J., Grybos, M., Wojcik, E., Van Dalen, A., and Einarsson, R.
(2004) Vascular endothelial growth factor (VEGF) concentration in sera and tumor effusions
from patients with ovarian carcinoma. Anticancer Res 24:1149-1157.
Koch, S., Yao, C., Grieb, G., Prevel, P., Noah, E.M., and Steffens, G.C. (2006) Enhancing angiogenesis
in collagen matrices by covalent incorporation of VEGF. J Mater Sci Mater Med 17:735-741.
Lamagna, C., and Bergers, G. (2006) The bone marrow constitutes a reservoir of pericyte progenitors. J
Leukoc Biol 80:677-681.
Lee, H., Bien, C.M., Hughes, A.L., Espenshade, P.J., Kwon-Chung, K.J., and Chang, Y.C. (2007)
Cobalt chloride, a hypoxia-mimicking agent, targets sterol synthesis in the pathogenic fungus
Cryptococcus neoformans. Mol Microbiol 65:1018-1033.
Loomans, C.J., Wan, H., de Crom, R., van Haperen, R., de Boer, H.C., Leenen, P.J., Drexhage, H.A.,
Rabelink, T.J., van Zonneveld, A.J., and Staal, F.J. (2006) Angiogenic murine endothelial
progenitor cells are derived from a myeloid bone marrow fraction and can be identified by
endothelial NO synthase expression. Arterioscler Thromb Vasc Biol 26:1760-1767.
Notari, L., Baladron, V., Aroca-Aguilar, J.D., Balko, N., Heredia, R., Meyer, C., Notario, P.M.,
Saravanamuthu, S., Nueda, M.L., Sanchez-Sanchez, F., Escribano, J., Laborda, J., and
Becerra, S.P. (2006) Identification of a lipase-linked cell membrane receptor for pigment
epithelium-derived factor. J Biol Chem 281:38022-38037.
Ohnishi, S., Yasuda, T., Kitamura, S., and Nagaya, N. (2007) Effect of hypoxia on gene expression of
bone marrow-derived mesenchymal stem cells and mononuclear cells. Stem Cells Download full-text
Ohno-Matsui, K., Yoshida, T., Uetama, T., Mochizuki, M., and Morita, I. (2003) Vascular endothelial
growth factor upregulates pigment epithelium-derived factor expression via VEGFR-1 in
human retinal pigment epithelial cells. Biochem Biophys Res Commun 303:962-967.
Oswald, J., Boxberger, S., Jorgensen, B., Feldmann, S., Ehninger, G., Bornhauser, M., and Werner, C.
(2004) Mesenchymal stem cells can be differentiated into endothelial cells in vitro. Stem Cells
Otrock, Z.K., Makarem, J.A., and Shamseddine, A.I. (2007) Vascular endothelial growth factor family
of ligands and receptors: review. Blood Cells Mol Dis 38:258-268.
Otsu, K., Das, S., Houser, S.D., Quadri, S.K., Bhattacharya, S., and Bhattacharya, J. (2009)
Concentration-dependent inhibition of angiogenesis by mesenchymal stem cells. Blood
Ozerdem, U., Alitalo, K., Salven, P., and Li, A. (2005) Contribution of bone marrow-derived pericyte
precursor cells to corneal vasculogenesis. Invest Ophthalmol Vis Sci 46:3502-3506.
Reyes, M., Lund, T., Lenvik, T., Aguiar, D., Koodie, L., and Verfaillie, C.M. (2001) Purification and ex
vivo expansion of postnatal human marrow mesodermal progenitor cells. Blood
Schatteman, G.C., Dunnwald, M., and Jiao, C. (2007) Biology of bone marrow-derived endothelial cell
precursors. Am J Physiol Heart Circ Physiol 292:H1-18.
Sena, K., Sumner, D.R., and Virdi, A.S. (2007) Modulation of VEGF expression in rat bone marrow
stromal cells by GDF-5. Connect Tissue Res 48:324-331.
Shibuya, M. (2008) Vascular endothelial growth factor-dependent and -independent regulation of
angiogenesis. BMB Rep 41:278-286.
Wrobel, T., Mazur, G., Surowiak, P., Wolowiec, D., Jelen, M., and Kuliczkowsky, K. (2003) Increased
expression of vascular endothelial growth factor (VEGF) in bone marrow of patients with
myeloproliferative disorders (MPD). Pathol Oncol Res 9:170-173.
Xia, Z., Locklin, R.M., and Triffitt, J.T. (2008) Fates and osteogenic differentiation potential of human
mesenchymal stem cells in immunocompromised mice. Eur J Cell Biol 87:353-364.
Xu, J., Liu, X., Chen, J., Zacharek, A., Cui, X., Savant-Bhonsale, S., Liu, Z., and Chopp, M. (2009)
Simvastatin enhances bone marrow stromal cell differentiation into endothelial cells via notch
signaling pathway. Am J Physiol Cell Physiol 296:C535-543.
Yamazaki, Y., and Morita, T. (2006) Molecular and functional diversity of vascular endothelial growth
factors. Mol Divers 10:515-527.
Zhang, S.J., Zhang, H., Hou, M., Zheng, Z., Zhou, J., Su, W., Wei, Y., and Hu, S. (2007) Is it possible
to obtain "true endothelial progenitor cells" by in vitro culture of bone marrow mononuclear
cells? Stem Cells Dev 16:683-690.
Zhang, S.X., Wang, J.J., Gao, G., Parke, K., and Ma, J.X. (2006) Pigment epithelium-derived factor
downregulates vascular endothelial growth factor (VEGF) expression and inhibits
VEGF-VEGF receptor 2 binding in diabetic retinopathy. J Mol Endocrinol 37:1-12.