Bi et al.

Stem Cell Research & Therapy (2019) 10:302

https://doi.org/10.1186/s13287-019-1415-6

RESEARCH Open Access

Stromal vascular fraction promotes

migration of fibroblasts and angiogenesis
through regulation of extracellular matrix in
the skin wound healing process
Hongsen Bi1*†, Hui Li2†, Chen Zhang1, Yiqing Mao2, Fangfei Nie1, Ying Xing2, Wuga Sha2, Xi Wang2,
David M. Irwin3* and Huanran Tan2*

Abstract
Background: A refractory wound is a typical complication of diabetes and is a common outcome after surgery.
Current approaches have difficulty in improving wound healing. Recently, non-expanded stromal vascular fraction
(SVF), which is derived from mature fat, has opened up new directions for the treatment of refractory wound
healing. The aim of the current study is to systematically investigate the impact of SVF on wound healing, including
the rate and characteristics of wound healing, ability of fibroblasts to migrate, and blood transport reconstruction,
with a special emphasis on their precise molecular mechanisms.
Methods: SVF was isolated by digestion, followed by filtration and centrifugation, and then validated by
immunocytochemistry, a MTS proliferation assay and multilineage potential analysis. A wound model was
generated by creating 6-mm-diameter wounds, which include a full skin defect, on the backs of streptozocin-
induced hyperglycemic mice. SVF or human adipose-derived stem cell (hADSC) suspensions were subcutaneously
injected, and the wounds were characterized over a 9-day period by photography and measurements. A scratch
test was used to determine whether changes in the migratory ability of fibroblasts occurred after co-culture with
hADSCs. Angiogenesis was observed with human umbilical vein endothelial cells. mRNA from fibroblasts,
endotheliocyte, and skin tissue were sequenced by high-throughput RNAseq, and differentially expressed genes,
and pathways, potentially regulated by SVF or hADSCs were bioinformatically analyzed.
Results: Our data show that hADSCs have multiple characteristics of MSC. SVF and hADSCs significantly improved
wound healing in hyperglycemic mice. hADSCs improve the migratory ability of fibroblasts and capillary structure
formation in HUVECs. SVF promotes wound healing by focusing on angiogenesis and matrix remodeling.
Conclusions: Both SVF and hADSCs improve the function of fibroblast and endothelial cells, regulate gene
expression, and promote skin healing. Various mechanisms likely are involved, including migration of fibroblasts,
tubulogenesis of endothelial cells through regulation of cell adhesion, and cytokine pathways.
Keywords: Stromal vascular fraction, Human adipose-derived stem cells, Wound healing

* Correspondence: [email protected]; [email protected];

[email protected]
†
Hongsen Bi and Hui Li are co-first authors (These authors contributed
equally to this study).
1
Department of Plastic Surgery, Peking University Third Hospital, Beijing
100191, China
3
Department of Laboratory Medicine and Pathobiology, University of
Toronto, Toronto, Ontario M5S1A8, Canada
2
Department of Pharmacology, Peking University, Health Science Center,
Beijing 100191, China

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 2 of 21

Background used by SVF to promote wound healing are still unclear,

Diabetes is a serious metabolic disease that threatens hu- which impedes the progress in the application of SVF
man health and results in a huge economic burden for cells. The aim of our study is to determine whether ther-
China [1]. Vascular lesions are the pathological basis for apy with non-expanded cell SVF is as efficient as ex-
many of the chronic complications of diabetes, including panded ADSCs in the treatment of wounds and to
major vascular lesions and microvascular lesions [2]. In evaluate the therapeutic potential of autologous SVF.
diabetic patients, large vascular lesions can cause ischemic We systematically investigated the effects of SVF and
changes in the skin, leading to ulcers, infections, and hADSCs on wound healing, including the rate of wound
wounds that do not heal [3]. The refractory wound is a healing, characteristics of the newly generated epithe-
typical complication of diabetes and is also a common oc- lium, effect on the migration ability of fibroblasts, and
currence after surgery [4]. These wounds are characterized blood transport system reconstruction, with a special
by having many complications that yield a large influence emphasis on the precise mechanisms. The results pro-
on their appearance [5]. Serious cases can lead to amputa- vide a new experimental basis for further understanding
tion. Effective control and treatment of diabetic wound mechanisms used by SVF and support promoting the
healing can significantly reduce the risk of amputation, re- clinical application of these cells.
duce medical expenses, and improve the quality of the life
of patients [6, 7]. However, to date, there is no ideal treat- Materials and methods
ment; therefore, improving the healing of refractory Ethical approvals
wounds is of great clinical value and significance. All experiments were performed in accordance with the
Skin wound healing is a complex biological integration guidelines and study protocols of the Peking University
process and includes the aggregation of extracellular Third Hospital Medical Science Research Ethics Com-
matrix molecules, soluble media and cytokines, prolifera- mittee (Approval No. IRB00006761-M2017391).
tion and migration of multiple cells, extracellular matrix
deposition, and reconstruction of blood transport sys- Harvesting of fat and SVF cell preparation
tems [8]. However, in diabetics, the orderly integration Human adipose tissue was obtained from male and female
of these processes is disrupted [9]. At present, conven- diabetic patients by a casing needle (fat suction) after in-
tional clinical treatment methods include removal of the formed consent and approval of the Peking University
trauma, local wound medication, and some newer treat- Third Hospital Medical Science Research Ethics Commit-
ment methods such as electrical stimulation, treatment tee. After fat suction, adipose tissue was digested with 0.1%
with growth factor, and negative pressure treatment collagenase I (Gibco, USA, 17100017) and then centrifuged
[10]. These methods have had only moderate clinical ef- at 1500 rpm for 5 min to obtain a cell pellet. After filtration
ficacy in improving chronic wound healing [11, 12]. Re- through a 425-μm mesh (Solarbio, CHN), the SVF cells
cently, stem cells, especially adipose-derived stem cells were counted. Some of the SVFs were immediately used in
(ADSCs) that have multidirectional differentiation po- animal experiments, with the remaining cells cultured.
tential and an ability to secrete various growth factors After 2 days of culture, media was changed to remove non-
and inflammation factors [13, 14], have become import- adherent cells while the attached cells were continued to be
ant tools in transplantation therapy to promote the heal- cultured in DMEM (Gibco, USA, 12800017) with 10% FBS
ing of refractory wounds [15]. Several research teams (Bioind, Israel, 04-001-1ACS). After the cell colonies
have shown that ADSCs can effectively promote the reached 80% confluence, cells were digested with trypsin.
healing of refractory wounds [16, 17]. However, isolation These cells were designated the first passage of human
of ADSCs requires in vitro cell culture and expansion adipose-derived stem cells (hADSCs) [21]. In this study, we
and can take from days to weeks. These steps increase used cells from the third to the fifth passage.
the possibility of infection by microorganisms, such as
bacteria, and thus, can be difficult for the clinical appli- Immunocytochemical characterization of SVF and hADSCs
cation of ADSC-based cytotherapy [18]. At the same Unamplified SVF cells and pure adherent homogenous
time, studies have found that non-expanded stromal vas- populations of cells obtained after culturing the SVF
cular fraction (SVF) that is derived from digestion, filtra- cells were characterized for the presence or absence of
tion, and centrifugation of mature fat cells contains a MSC marker proteins. After fixing with 4% polyformal-
variety of fat cells and mesenchymal stem cells (MSCs) dehyde and incubation with 5% BSA (Amersco, USA,
that opens up new directions for the treatment of diffi- A0332), cells were then incubated with mouse antibodies
cult wounds [19]. to CD105, CD44, and CD90 (each 1:100) and with rabbit
Although SVFs are helpful for the treatment of many antibodies to CD29 (1:100) and CD45 (1:200) (Abcam,
diseases, currently it is not commonly used in the clinic USA, Mesenchymal Stromal Cell Marker Antibody Panel
[20]. Despite extensive research, the precise mechanisms (CD44, CD45, CD90, CD29, CD105) - Human, ab93758)
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 3 of 21

overnight at 4 °C. Antibody binding was detected using 7 days after treatment with a Roche blood glucose monitor
goat anti-mouse antibodies conjugated to FITC (Zsbio, (Roche, Germany).
CHN, ZF-0312) or goat anti-rabbit antibodies conju- Hyperglycemic mice were raised for an additional 21
gated to TRITC (Zsbio, CHN, ZF-0316) by incubating in days, and then full-thickness wounds were created.
the dark for 1 h. The nuclei of cells were stained with Under general anesthesia, hair in the surgical area was
300 nM DAPI (Solarbio, CHN, C0065), and fluorescence removed and a full-thickness excisional wound of 6 mm
was observed through a confocal microscope (Leica, diameter was produced on the back of each mouse. To
Germany). prevent the wound from spontaneously contracting, a
plastic ring was placed in the wound. Hyperglycemic
mice (n = 19) were divided into three groups: SVF
Assays for adipogenic and osteogenic differentiation
treated (SVF, injection of SVF suspension with FBS into
potential
the four quadrants, n = 5), hADSC treated (hADSC, in-
Assays for adipogenic differentiation potential was per-
jection of hADSC suspension with FBS into the four
formed according to the manufacturer’s instructions (Cya-
quadrants, n = 6), and control (CON, injection of FBS,
gen, CHN, HUXMD-90031). Briefly, hADSCs were seeded
n = 8). Local injections of cell suspensions contained
into six-well plates, with human fibroblast (hF) cells used as
1.5 × 105 cells. On the day of the operation and every
control. When the cells reached 80% confluence, the media
day thereafter, the state of the wound was recorded by
was replaced with adipogenic differentiation medium A
photography. Nine days after the operation, a 2-mm-
(Cyagen, CHN, HUXMD-90031). After using medium A
wide slice of skin from around the wound surface was
for 3 days, medium B (Cyagen, CHN, HUXMD-90031) was
harvested, with half of each sample paraffin-embedded
replaced for 24 h. After using medium A and B alternately
for HE staining and the remainder frozen in liquid nitro-
for five times, medium B was used continuously for 7 days
gen and stored at − 80 °C for RNA analysis. Figure 1 de-
until the fat drops become large and round enough. During
scribes the experimental design used in this study.
the culture period of medium B, fresh medium B was re-
The long and short diameters of the wounds were
placed every 3 days. After about 27 days of culture, cells
measured with a cursor caliper, and the area was calcu-
were stained with Oil Red O according to the following
lated using the following formula:
method. Cells were fixed with 50% ethanol for 3 min and
then dyed with 5% Oil Red O for 30 min at room
Area ¼ π  long diameters  short diameters:
temperature. Induced fat cells contain orange-red oil
droplets.
For assays for osteogenic differentiation potential, as
Histology and immunohistochemistry
with the adipogenic differentiation experiments, media
For HE and Masson staining, 4-μm-thick paraffin-
was replaced with osteogenic differentiation medium
embedded sample slices were prepared. These experi-
(Cyagen, CHN, HUXMD-90021) when confluence
ments were performed by Servicebio (CHN). Images
reached 70% in the six-well plates. Osteogenic media
were taken using an Olympus1X71 microscope (Olym-
was changed every 3 days. After about 21 days of culture,
pus, JPN). Image-Pro Plus 6 (Media Cybernetics, USA)
Alizarin red S staining was performed. Cells were fixed
was used to measure the thickness of the epidermis and
with 50% ethanol for 3 min, and then stained with 1%
evaluate the integrity of the new skin in the HE staining.
Alizarin red S for 10 min at room temperature. Induced
osteoblast cells contain red crystal particles.

Healing of full-thickness wounds in diabetic mice

Male C57 mice (4–6 weeks old) were purchased from
the Animal Center of the Peking University Health Sci-
ence Center (Beijing, CHN). Animals were housed under
controlled environmental conditions, including temperature,
humidity, and lighting. Standard laboratory chow and water
were provided ad libitum. A protocol for these experiments,
following the “Guidelines for Animal Experiments,” was ap-
proved by the Peking University Health Science Center. C57
mice were randomly assigned to the following two groups: Fig. 1 Timeline for the experimental design. Initiation of treatment
blank (normal saline, intraperitoneal injection) and STZ with SVF or hADSCs was considered to be day 0 (indicated in the
blue box), with wounds photographed and measured each day until
(streptozocin 200 mg/kg, intraperitoneal injection) (Coola-
day 9
ber, CHN, CS10491). Fasting serum glucose was measured
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 4 of 21

Collagen was quantified by calculating the ratio of the is the scratch width at the nth hour after scratch
positive area (blue–green collagen) to the total area in formation.
the skin tissue Masson staining.
In order to detect the presence of CD31, immunohis- Tube formation assay
tochemistry was carried out. Rehydrated paraffin sec- Human umbilical vein endothelial cells (HUVECs) (iCell
tions were incubated in a microwave oven for 8 min. All Bioscience, CHN) were cultured in PriMed-iCELL-002
sections were blocked with 3% BSA at 37 °C for 30 min, (iCell Bioscience, CHN) supplemented with 5% FBS.
followed by incubation with anti-CD31 antibody (Servi- Subsequent experiments were conducted using cells at
cebio, GB11063-3, 1:800) at 4 °C overnight. After wash- passage 2 to 4. Culture plates were pre-cooled to − 20 °C
ing with PBS, a goat anti-rabbit secondary antibody before the experiments. Matrigel (Becton, Dickinson &
(Servicebio, GB23303, 1:200) coupled with horseradish Co., USA, 356230) was laid on the bottom of the lower
peroxidase was employed. Samples were incubated at transwell chamber, with the chamber then placed at
37 °C for 1 h. Sections were visualized with diaminoben- 37 °C for 2 h for solidification of the matrigel. HUVECs
zidine tetrahydrochloride, and staining was detected with (1 × 105/well) were seeded into the lower chamber.
a light microscope (Olympus, JPN). CD31 levels were hADSCs or HUVECs (as a control) (1 × 104/well) were
quantified by calculating the ratio of the positive area plated into the upper chamber of the transwell, followed
(yellowish-brown) to the total area using Image-Pro Plus by incubation for 24 h before the experiments. Angio-
6 (Media Cybernetics, USA). genesis (tube formation) was observed and photo-
graphed after 0, 2, 4, 6, 8, and 10 h. The number of tube
MTS proliferation assay structures was counted using an Image-Pro Plus 6
hADSCs (2 × 103 per well) were seeded into 96-well (Media Cybernetics, USA).
plates, and then formazan production was determined
using the CellTiter 96 AQueous Non-Radioactive Cell mRNA preparation for RNAseq
Proliferation Assay (Promega, USA, G3582) according to We used the transwell assay as a non-contacting co-
the manufacturer’s instructions. Dehydrogenase pro- culture system to evaluate the impact of hADSCs on
duced by living cells can convert MTS to formazan, with fibroblast (3T3 cells) and endotheliocyte (MS1 cells).
formazan having an absorption peak at 490 nm. OD Fibroblast and endotheliocyte cells were seeded into the
value at 490 nm reflects the ability of cells to proliferate. lower chamber of the transwell with hADSCs in the
upper chamber. Cells were co-cultured for 72 h at 37 °C.
Migration ability assay Cells in the lower chamber were washed by PBS, and
A scratch test was used to determine changes in the mi- total RNA was isolated using Trizol LS reagent (Invitro-
gratory ability of fibroblast cells in co-culture with gen, USA, 10296-010) following the instructions of the
hADSCs. Primary human fibroblasts (hFs) were cultured manufacturer.
from normal human skin. hADSC or hF cell suspensions Skin around the wounds was harvested at the end of
that had been continuously cultured for 4–8 passages the wound healing experiment. One half of each sample
were seeded (1 × 104 cells) into the upper chamber of 8- was snap-frozen, and total RNA was isolated using Tri-
mm pore size 24-well transwell plates (Corning, USA), zol LS reagent (Invitrogen, USA, 10296-010) following
with 5 × 104 hFs cultured in the lower chamber. Trans- the instructions of the manufacturer.
well plates were cultured for 24 h with DMEM full cul- mRNA was purified from the 3 μg of total RNA per
ture medium, with cells attaining 90% confluence. A sample (cells or skin tissue) using poly-T oligo-attached
scratch model was prepared by drawing a straight line magnetic beads, and then RNAseq was performed by
on the surface of the lower chamber with a 200-μl pip- Majorbio (Shanghai, CHN). Clean high-quality data ob-
ette tip, with PBS used to wash off any exfoliated cells. tained after removing reads containing ploy-N, reads con-
Adherent cells were continued to be co-cultured with taining adapters, and low-quality reads from the raw data
hADSCs or hFs for 0, 6, 12, 24, 36, 48, 60, and 72 h, and was used for the subsequent analysis. DEGSeq R package
the scratch width was observed and analyzed with (1.12.0) was used for the differential expression analysis.
Image-Pro Plus 6 (Media Cybernetics, USA). Change in In this study, log2 (fold change) of > 1.00 and a corrected
scratch width (ΔScratch) was calculated using the fol- P value < 0.05 (adjusted using the Benjamini and Hoch-
lowing formula: berg method) were necessary to classify a differentially
expressed gene. Data were analyzed on the free online
Majorbio I-Sanger Cloud Platform (www.i-sanger.com).
ΔScratch widthn ¼ SWi −SWn
Significantly enriched KEGG and GO terms for differ-
entially expressed genes were those with corrected P
where SWi represents the initial scratch width and SWn values less than 0.05. A protein-protein interaction (PPI)
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 5 of 21

network of the differential genes was constructed using

the STRING database (http://string-db.org/) and was vi-
sualized using Cytoscape for further analysis.

Gene expression analysis using quantitative reverse

transcription polymerase chain reaction (qRT-PCR)
Cell or skin tissue samples were homogenized, and total
RNA was isolated using Trizol LS reagent (Invitrogen,
USA, 10296-010) following the instructions. Amplifica-
tions were performed with the StepOne™ System with
PowerUpTM SYBR® Green MasterMix (ABI, USA,
4367659) with the primers listed in Table 1. The amplifi-
cation steps are as follows: initial denaturation at 95 °C
for 10 min, followed by 40 cycles at 95 °C for 15 s and
60 °C for 1 min. GAPDH was used as an internal
reference.

Statistical analysis
Results are shown as means ± SD. Differences between
groups were analyzed by one-way ANOVA (SPSS 13.0
for Windows, SPSS Inc., USA). P values of less than 0.05
were considered to be statistically significant.

Results
Characterization of SVF surface markers
Adipose-derived SVF (Fig. 2a) are a heterogeneous
population of cells. According to the International Soci-
ety for Cellular Therapy position statement, MSCs are
positive for CD105, CD44, CD90, and CD29 and have
low expression of the negative marker CD45 [22]. SVF
cells were characterized for these markers by
Fig. 2 SVF surface marker characteristics. a After digestion and
centrifugation of fat suction, obtained SVF (red arrow). b
Table 1 Primers for quantitative reverse transcription Immunofluorescence staining of CD105 (green), CD44 (green), CD90
polymerase chain reaction (qRT-PCR) (green), CD29 (red), and CD45 (red) in SVF, counterstained with
Gene Primer sequence (5′–3′) DAPI (blue)
name
Forward Reverse
edn1 CGTCGTACCGTATGGACTGG GCCTGGTCTGTGGCCTTATT
immunocytochemistry, with our results confirming that
lmod1 ATGGTTGGACAGTGCCTGAG TGGCATCAACCCTTCAGGTC
the main cell type in the SVF population is hADSC
Osm AGAATCAGGCGAACCTCACG GTGTGTCCTCACTGGGGAAG (Fig. 2b).
Ccl4 AGTCATCATTGAACAGGCTTTCT CGAGTGCTGTGCTGATCTGA
Ccr1 ACTCTGGAAACACAGACTCACT TGATGTGCTGCTGCGAGATT HADSCs share multiple characteristics with MSCs
Col1a1 ACAGTCGCTTCACCTACAGC GGGTGGAGGGAGTTTACACG The hallmarks of ADSCs are their characteristic surface
Col4a2 CTGGTGAAGCACAGCCAAAC GTCACCCGGATTGCAGTACA
protein expression, multipotency, and proliferative activ-
ity in vitro [23]. Freshly obtained SVF were immediately
Itgb4 CACGGACACCGGGATTATGT AGACTCCTGTCCGTTTCATCG
sorted and put into culture to observe their expansion,
Sdc1 CCTAACGCAGAGGAAGGGAC TTCGTGGATCCGAGTTGCC with adherent colonies forming after 24 h. Third passage
Thbs1 GCTGCCAATCATAACCAGCG TTCGTTAAAGGCCGAGTGCT hADSCs exhibited a fibroblast-like morphology. In order
Mmp9 CCAGCCGACTTTTGTGGTCT TGGCCTTTAGTGTCTGGCTG to evaluate the differentiation potential of hADSCs into
Vipr1 GCCTCCACACAAGGCAAATG GTGTTTCCAGGTAGGGCACA adipocytes and osteoblasts, hADSCs were cultured in
Angptl7 TCACGGGGAAGGAAGAAAACT GCCTGTTCCCCTTAGAGTCTG
adipogenic or osteogenic inducing medium in vitro.
Under lipid-forming conditions, hADSCs differentiate
GAPDH ACAAAGTGGACATTGTTGCC AAACATGGTGGTGAAGACGC
into adipocytes with Oil Red O (ORO)-positive lipid
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 6 of 21

droplets. Osteogenic inducing medium promoted CD90, and CD29 and have low expression of the nega-
hADSCs to produce Alizarin red-positive particles. tive marker CD45 (Fig. 3b). There was no significant dif-
There changes were significantly different from those ference between cells obtained from diabetic or healthy
seen with hF cells (Fig. 3a). individuals regarding cell morphology or MSC marker
The expression of surface proteins by hADSCs at pas- expression.
sage 3 was examined using immunocytochemistry. Our To validate the proliferative capability of hADSCs, we
cells fulfilled the defining criteria for MSCs according to tested the growth rate of hADSCs by MTS. Our results
the International Society for Cellular Therapy position show that hADSCs have a good proliferative activity.
statement. Our hADSCs are positive for CD105, CD44, The proliferative activity of hADSCs from diabetic

Fig. 3 hADSCs have multiple characters of MSCs. a Osteogenic differentiation (upper panel) of hADSCs (left) and fibroblasts (hFs, right) stained
with Alizarin red S and adipogenic differentiation (lower panel) of hADSCs (left) and fibroblasts (hFs, right) stained with Oil Red O. b
Immunofluorescence staining of hADSCs from diabetic and nondiabetic donors for CD105 (green), CD44 (green), CD90 (green), CD29 (red), and
CD45 (red) counterstained with DAPI (blue). c Cell proliferation of hADSCs measured by the MTS assay, with cell proliferation calculated as a
percentage. The proliferative activity of hADSCs from diabetic patients (SVF-D) was not significantly different compared to those from healthy
patients (SVF-ND)
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 7 of 21

patients (SVF-D) was not significantly different com- To examine the effect of hADSCs and SVF on microvessel
pared to those from healthy patients (SVF-ND). This re- density, immunohistochemistry was performed to determine
sult shows that hADSCs from diabetic patients did the levels of CD31 (Fig. 5e). Treatment with hADSCs or
not exhibit an impaired proliferative activity (Fig. 3c). SVF significantly increased the number of CD31-positive
Taken together, these data suggest that hADSCs ex- microvessels in the dermis, numbers that were significantly
panded from SVF of diabetic donors maintain MSC higher than in the control group (Fig. 5f, g).
characteristics.
Influence of SVF and hADSCs on differentially expressed
Influence of SVF and hADSCs on wound healing genes in the skin of mice
processes in hyperglycemic mice To elucidate mechanisms involved in the improved wound
Hyperglycemic mice were induced by a single intraperi- healing due to hADSCs and SVF, we applied high-
toneal injection of STZ. Serum glucose levels in these throughput RNAseq to investigate the gene expression pro-
mice were significantly increased at 7 days after injection files of skin after 9 days of treatment with hADSCs or SVF.
(Fig. 4a) and remained stable and high at 28 days. In this Fold change was defined as the ratio of the mRNA level in
hyperglycemic mouse model, wound healing in the hADSC or SVF group to the control group using FPKM. In
hADSC and SVF groups was enhanced compared with this study, log2 (fold change) of > 1.00 and a P value < 0.05
the control groups. As shown in Fig. 4c, the appearance were necessary to classify a gene as differentially expressed.
of the wounds at different times post-wounding were A total of 6177 significantly differentially expressed genes
quite different. Wound area at the different time inter- were identified in the hADSC-treated mouse skin and 2244
vals was determined. On days 6, 7, and 9 after wounding, in the SVF-treated mice, of which 1409 genes were signifi-
wound area was significantly decreased after hADSC and cant in both the hADSC and the SVF groups, as shown in
SVF injection compared to untreated wounds at the cor- the Venn diagram in Fig. 6a.
responding times (Fig. 4d). Hence, hADSC and SVF cells The potential functions of genes that are differen-
significantly accelerated wound healing compared with tially expressed in both hADSC and SVF groups were
the control group, especially at days 6, 7, and 9 post- characterized based on their annotations in the KEGG
surgery. database. Among the 325 enriched pathways, the 20
pathways with the higher enrichment indices are
Influence of SVF and hADSCs on mRNA expression in mainly associated with cytokine-cytokine receptor
mouse skin interaction, the IL-17 signaling pathway, the Toll-like
Expression of Endothelin1 and Leiomodin1 mRNAs in receptor signaling pathway, and the Jak-STAT signal-
mouse skin were detected by qRT-PCR. Endothelin1 and ing pathway. A total of 54 genes matched to the
Leiomodin1 mRNA levels were significantly decreased “cytokine-cytokine receptor interaction” pathway, sig-
upon hADSC or SVF treatment (Fig. 4e, f). nificantly more than for any other pathway. The lar-
gest spot (shown in red) in the KEGG enrichment
Influence of SVF and hADSCs on changes in skin diagram shown in Fig. 6b represents the “cytokine-
histology cytokine receptor interaction.”
We analyzed the histology of skin around the wounds 9 To analyze the functional significance of the changes in
days after injury. When skin sections from the control abundance of proteins due to the hADSC or SVF treatment,
(CON) were compared to the hADSC or SVF groups, we we used the STRING program to look for functional net-
noted that in addition to the increased healing rate, the works involved with the enriched genes in the KEGG gene
epidermis of the hADSC- or SVF-treated mice was set “cytokine-cytokine receptor interaction” pathway, which
thicker than that in controls (Fig. 5a, b). More collagen was visualized with Cytoscape. Figure 6c provides a graph-
fibers were observed in the hADSC and SVF groups than ical overview of the identified protein-protein interactions
in the control group on postoperative day 9, as detected (PPI) and implicated changed behaviors in response to
by Masson staining (collagen appears blue). In the con- hADSC or SVF treatment due to these enriched proteins.
trol group, disordered collagen and reduced levels of col- The Cytoscape network analysis found that chemokines (in-
lagen in the skin around the wound were found cluding ccl2, ccl4, ccl6, ccl7, and ccl9), chemokine receptors
(Fig. 5c). The content of collagen in the skin tissue was (including cxcr2, cxcr3, cxcr4, cxcr6, and ccr1), interleukins
measured and compared with control mice; the collagen (including il1b, il12a, and il18), and oncostatin M (osm)
content of hADSC- or SVF-treated mice was signifi- have high “betweenness centrality” and “degree” and play a
cantly increased. These results indicate that hADSC or key role in the PPI network (Fig. 6c).
SVF treatment improves the collagen content of the Through a hierarchical clustering analysis (expression
skin, but no significant difference was seen between indicator: TPM; linkage rules: complete linkage; similar-
these two treatment groups (Fig. 5d). ity measure: Euclidean), the 26 differentially expressed
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 8 of 21

Fig. 4 SVF and hADSCs improve the wound healing processes in hyperglycemic mice. a Fasted basal serum glucose levels treated with hADSCs
and SVF (*P < 0.05, compared to blank). b Full-thickness excisional wounds of the same 6-mm-diameter size were created on the backs of each
mouse. c Photographs of wounds 0–9 days after skin injury. d Quantification of the wound area, which was measured using the long and short
diameters of the wounds, at different days after wounding. (*P < 0.05, hADSCs compared to control (CON); #P < 0.05, SVF compared to CON).
Influence of SVF and hADSCs on the expression of Endothelin1 and Leiomodin1 mRNAs. Gene expression of Endothelin1 (e) and Leiomodin1 (f) was
evaluated by qRT-PCR from skin at the edges of the wounds treated or untreated with hADSCs or SVF. Results were normalized to GAPDH
expression. (Statistical significance of the differences is indicated by ***P < 0.001, **P < 0.005, and *P < 0.05)

genes in the PPI network could be divided into 9 distinct hADSCs, compared to the untreated group (Fig. 6d).
clusters according to their expression profiles: 7 of Among these genes, Osm, Ccl4, and Ccr1 have the
which exhibited upregulated genes and 2 clusters exhi- largest changes.
biting downregulated genes (Fig. 6d). This result is con- qRT-PCR was performed to verify the RNAseq results.
sistent with the observation that chemokines When qRT-PCR was conducted, its results were consist-
(including Ccl2, Ccl4, Ccl6, Ccl7, and Ccl9), chemo- ent with the gene expression levels identified by RNA-
kine receptors (including Cxcr2, Cxcr3, Cxcr4, Cxcr6, seq. In particular, qRT-PCR results showed that the
and Ccr1), interleukins (including Il1b and Il18), and gene expression levels of Osm, Ccr1, and Ccl4 were sig-
oncostatin M (Osm) are upregulated and that Il12a is nificantly increased in skin from mice treated with
downregulated in mouse skin treated with SVF or hADSCs or SVF (Fig. 7).
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 9 of 21

Fig. 5 Histological changes induced in skin by hADSCs and SVF. a Hematoxylin and eosin (H&E) images of wound edge skin at days 0 and 9
post-injury. Dotted lines outline the boundary between the epidermis and the dermis. b Quantification of the epidermal thickness was performed
using Image-Pro Plus 6. (***P < 0.001, compared to control (CON)). c Masson trichrome staining of sections of the wound edges from control and
treated mice at 0 and 9 days after wounding. d Quantification of the Masson-positive area was performed using Image-Pro Plus 6. (*P < 0.05,
**P < 0.005 compared to CON). e Immunohistological analysis of wound sections from control and treated mice 9 days after injury using
antibodies directed against CD31. Arrows mark the CD31+ cells. f Quantification of the CD31-positive cells was performed using Image-Pro Plus 6
analysis software. (*P < 0.05, **P < 0.005 compared to CON). g Quantification of the microvessel density in sections as determined by CD31
expression (*P < 0.05, **P < 0.005 compared to CON)

Influence of hADSCs on migratory ability of fibroblasts migration signals enhancing the migratory ability of fibro-
The scratch test and the transwell system (Fig. 8a) were blasts. (Fig. 8b, c).
used to investigate the influence of hADSCs on the migra-
tory ability of fibroblasts. Results of the scratch test Influence of hADSCs on the differential expression profile
showed that the migratory ability of fibroblasts was signifi- of fibroblasts
cantly increased with hADSC co-culture at 24, 36, 48, 60, To clarify mechanisms involved in the migration promo-
and 72 h compared to untreated controls (B-hFs and hFs- tion of hADSCs, we used high-throughput RNAseq and
hFs). These data suggest that hADSCs functioned as pro- calculated the relative expression levels of genes with
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 10 of 21

Fig. 6 Analysis of differential gene expression in the skin from mice treated with hADSCs (M-hADSC) or SVF (M-SVF) compared to untreated skin
(M-CON). a Venn diagram displaying the numbers and the overlap of the identified differentially expressed genes in M-hASC vs M-CON (red) and
M-SVF vs M-CON (green). b Significantly enriched KEGG pathways. As shown in the bubble diagram, the X-axis represents the enrichment score
of the gene sets, the Y-axis shows gene set names, and the size of the bubble indicates the number of genes in that gene set. KEGG pathways
with P < 0.05 are significantly enriched. The top 20 significantly different KEGG terms are shown. c Expanded view of the network imported from
Cytoscape, where nodes represent core differentially expressed genes in the gene set KEGG “cytokine-cytokine receptor interaction” pathway and
edges the interaction. Node size represents “betweenness centrality,” node color represents “degree.” d Heat map of core differentially expressed
genes in the gene set KEGG “cytokine-cytokine receptor interaction” pathway with Cytoscape network analysis. The colors in the heat map
represent gene expression level (log10(TPM+1)). Red represents higher expression; blue represents lower expression. The brackets represent
cluster analysis

FPKM to study the effect of 72 h of hADSC treatment of 1720 unique differentially expressed genes in fibro-
on gene expression in fibroblasts. In this study, log2 blast cells treated with hADSCs were identified.
(fold change) of > 1.00 and a P value < 0.05 were re- The annotations of genes that were differentially
quired to classify differentially expressed genes. A total expressed between the hADSC-treated and control cells
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 11 of 21

Fig. 7 Comparison of the results obtained from RNAseq and qRT-PCR methodologies for three genes. Influence of hADSCs or SVF on Osm (a),
Ccl4 (b), and Ccr1 (c) mRNA abundance. Relative gene expression evaluated by qRT-PCR is represented by the columns and TPM evaluated by
RNAseq represented by the scatter diagram in skin treated or untreated with hASCs or SVF. qRT-PCR results were normalized to GAPDH
expression. (*P < 0.05 compared to M-CON)

were obtained from the KEGG database. Significantly promotion of endothelial cell tubulegenes by hADSCs
enriched pathways (corrected P value < 0.05) among the (Fig. 9a). We found that hADSCs promoted the forma-
326 successfully mapped pathways are mainly associated tion of tubules after 2 to 10 h of treatment compared to
with the cell cycle–yeast, DNA replication, cell cycle, controls (B-EC and EC-EC) (Fig. 9b). Counting of the
extracellular matrix (ECM)-receptor interaction, hom- number of branches in the tubes generated in the co-
ologous recombination, mismatch repair, TNF signaling culture with hADSCs revealed that the hADSC co-
pathway, cytokine-cytokine receptor interaction, and culture significantly increased the number of new tube-
MAPK signaling pathway (Fig. 8d). like structures compared to those generated by un-
A requirement that a KEGG pathway be regulated by treated HUVECs (Fig. 9c).
all of treatment groups, including mice wound skin with
hADSCs and SVF, and fibroblasts treated with hADSCs, Influence of hADSCs on the differential expression profile
increases the likelihood that the identified KEGG path- of endotheliocyte
way is consistently affected in the fibroblasts both the To elucidate the pro-angiogenic mechanisms of
in vivo and the in vitro models. Using this strategy, 68 hADSCs, we applied high-throughput RNAseq to inves-
KEGG pathways were found to be significantly enriched tigate the effect of hADSC treatment for 72 h on EC
in the hADSC-treated mice skin, 33 pathways in the gene expression. In this study, log2 (fold change) of >
SVF-treated mice skin, and 13 pathways in the hADSC- 1.00 and a P value < 0.05 were necessary to classify dif-
treated fibroblasts cells, with 2 of these pathways over- ferentially expressed genes. A total of 1577 genes were
lapping in all of these groups. The two pathways were identified in EC treated with hADSCs, which were then
ECM-receptor interaction (Pathway ID: map 04512) and used for GO, KEGG, and PPI analyses.
cytokine-cytokine receptor interaction (Pathway ID: map Genes that were differentially expressed between the
04060). We generated Venn diagrams to show the extent hADSC and control treatments were annotated using
of similar response in the in vivo and in vitro models the KEGG database. Significantly enriched pathways
(Fig. 8e). There are 16 differentially expressed genes (corrected P value < 0.05) among the 327 successfully
matched to the “ECM-receptor interaction” pathway and mapped pathways are mainly associated with the ECM-
29 genes matched to the “cytokine-cytokine receptor receptor interaction, protein digestion and absorption,
interaction” pathway in the fibroblast cells. selenocompound metabolism, focal adhesion, folate bio-
synthesis, relaxin signaling pathway, and TNF signaling
Influence of hADSCs on HUVEC capillary structure pathway (Fig. 9d).
formation Only one KEGG pathway affected in EC overlapped
During wound healing, new blood vessels are needed to with both in the in vivo and in vitro models, including
ensure sufficient blood flow to supply oxygen and nutri- mice wound skin with hADSCs and SVF and EC
ents for the metabolism of the various types of cells treated with hADSCs. The overlapping KEGG path-
around the wound. An important process in angiogen- way was ECM-receptor interaction (Pathway ID: map
esis is the formation of tubule structures by endothelial 04512). We drew Venn diagrams to show the extent
cells accompanied by the migration of fibroblasts [24]. of similarities of the responses of the in vivo and
To characterize the angiogenesis process, we studied the in vitro models (Fig. 9e). A total of 20 differentially
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 12 of 21

Fig. 8 hADSCs promote migration of fibroblasts. a Diagram of the transwell setup. hF cells were seeded into the lower chamber of the 24-well
transwell (pore size: 0.8 μm); hFs or hADSCs were plated in the upper chamber. b Representative images showing fibroblast scratches. Images
were taken at 0 to 72 h. Arrow shows the width of the scratch wound. c Quantification of the scratch width over time measured using Image-Pro
Plus 6 software. (*P < 0.05, compared to B-hFs; #P < 0.05, compared to hFs-hFs). d Significantly enriched KEGG pathways are shown as a bubble
diagram. The X-axis represents the enrichment score of the gene sets, the Y-axis shows gene set names, and the size of the bubble indicates the
number of genes in that gene set. The KEGG pathways with P < 0.05 are significantly enriched. e Venn diagram displaying the number and
overlap of the enriched KEGG pathway in the Fs-CON vs Fs-hADSC group (blue), M-CON vs M-hADSC group (green), and M-CON vs M-SVF
group (yellow)

expressed genes matched to the “ECM-receptor inter- (Lamb3 and Lamc3), and collagen (Col6a3, Col1a1,
action” pathway in endotheliocyte. Col9a3, and Col4a2), had expression levels that were sig-
nificantly upregulated and that thrombospondins (Thbs1
and Thbs3) were downregulated in fibroblasts and
Influence of hADSCs on the “ECM-receptor interaction” endotheliocyte treated with hADSCs. The location and
pathway role of these different genes of fibroblasts and endothe-
Our KEGG analysis showed that the, in vivo and liocyte in the KEGG pathways were analyzed (Fig. 10),
in vitro, hADSC-treated fibroblasts and endotheliocyte and it was found that these genes play an important role
all had similar genetic changes in the KEGG pathway in extracellular matrix function.
“ECM-receptor interaction.” In the current study, we To explore the functional consequences of the changes
verified that the ECM-receptor interaction-related genes, in gene expression associated with hADSCs, we analyzed
including integrins (Itga7, Itga2b, and Itgb4), laminin the enrichment genes in the gene set in the KEGG
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 13 of 21

Fig. 9 hADSCs promote tube formation. a Diagram of the transwell setup. HUVECs were seeded into the lower chamber of the transwell (pore
size: 0.8 μm) with HUVECs or hADSCs plated in the upper chamber. b Representative images showing endothelial cell tube formation. Images
were taken at 0 to 10 h. c Numbers of tube structure over time were counted. (*P < 0.05, compared to B-EC; #P < 0.05, compared to EC-EC). d
Significantly enriched KEGG pathways are shown as a bubble diagram. The X-axis represents the enrichment score of the gene sets, the Y-axis
shows gene set names, and the size of the bubble indicates the number of genes in that gene set. The enrichment score was calculated using
the total number of genes in each functionality group and the total number of genes that could be functionally annotated. P < 0.05 was used as
the thresholds in selecting significant KEGG pathways. The significantly different KEGG terms are shown. e Venn diagram displaying the number
and overlap of the enriched KEGG pathway in EC-CON vs EC-hADSC group (red), M-CON vs M-hADSC group (green), and M-CON vs M-SVF
group (yellow)

pathway “ECM-receptor interaction” (Fig. 11a). We used Influence of hADSCs on gene expression in fibroblasts
the STRING programs to look for functional networks and endotheliocytes
of these enriched genes, which was visualized with qRT-PCR was performed to verify the RNAseq results
Cytoscape. Figure 11b provides a graphical overview of and was found to be consistent with the RNAseq results.
the STRING results and indicates that the changed be- Genes to be confirmed were selected based on their im-
havior in response to hADSC treatment in proteins. The portance and likely functions for wound healing. qRT-
Cytoscape network analysis found that genes, including PCR results show that the expression levels of genes as-
Itgb4, Itga7, Itga6, Col4a2, Thbs1, Tnc, Col4a1, Col1a1, sociated with extracellular matrix, including Col1a1,
Thbs3, and Itgb7, have a high “betweenness centrality” in Col4a2, Itgb4, Sdc1, and Mmp9, were significantly in-
the PPI networks of “ECM-receptor interaction.” creased in both fibroblasts and endotheliocyte treated
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 14 of 21

Fig. 10 Differentially expressed genes in the ECM-receptor interaction pathway are shown. The ECM-receptor interaction pathway is adapted
from the KEGG database. Genes upregulated in endotheliocytes and fibroblast by hADSCs are in (red), only upregulated in fibroblast are in red
with right half filled with yellow, only upregulated in endotheliocytes are in red with left half filled with yellow, and downregulated are in green.
Gene with log2 (fold change) > 1.00 is considered to be a significantly differentially expressed gene

with hADSCs. Meanwhile, Thbs1 was significantly de- Discussion

creased in fibroblasts and endotheliocyte treated with The stromal vascular fraction (SVF) obtained by collage-
hADSCs. In addition to this, Lmod1 (Leiomodin1) and nase digestion and centrifugal separation contains a var-
Edn1 (Endothelin1) were significantly decreased in fibro- iety of biologically active cells [25]. A subgroup of SVF
blasts and endotheliocyte treated with hADSCs. In con- cells have a large proportion of characteristics similar to
trast, gene expression levels of Vipr1 (Vasoactive those of mesenchymal stem cells from bone marrow
intestinal polypeptide receptor 1) and Angptl7 (Angio- sources, and are named “adipose-derived stem cells
poietin-like 7) were significantly increased in the fibro- (ADSCs)” [26]. They are an ideal source of stem cells
blasts and endotheliocyte treated with hADSCs. The [25]. SVFs collected according to a normal operation
results of qRT-PCR are shown in Fig. 11c–l, and the ex- method can promote the survival of transplanted fat and
pression patterns of these 10 genes are consistent with assist in the treatment of refractory wound. In animal
the RNAseq results. tests, SVF was found to promote angiogenesis, increase
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 15 of 21

Fig. 11 hADSC-related gene-regulatory network in “ECM-receptor interaction.” a Heat map of core differentially expressed genes in the gene set
KEGG “ECM-receptor interaction” pathway. The colors in the heat map represent gene expression level (log10(TPM+1)). Red represents higher
expression, blue represents lower expression. b Expanded view of the network imported from Cytoscape, where nodes represent core difference
gene in the gene set KEGG “ECM-receptor interaction” pathway and edges the interaction. Node size represents “betweenness centrality,” node
color represents “degree.” c–l Comparison of the results of 10 genes obtained using RNAseq and qRT-PCR methodology. Relative gene expression
was evaluated by qPCR in Fs and EC treated or untreated with hADSCs. Results were normalized to GAPDH expression (***P < 0.001,
**P < 0.005, *P < 0.05)
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 16 of 21

vascular density, and improve blood supply to ischemic role in wound healing [29]. The thickness of skin is
myocardium. It is well known that the purity and num- likely associated with the proliferation of fibroblasts and
ber of stem cells obtained by the direct separation other skin cells stimulated by SVF. Angiogenesis and the
method are relatively low, and it takes a long time to migration of fibroblasts are essential in wound healing
purify and cultivate them in vitro. During this process, it by allowing an increase in the blood supply near the
is possible that these cells become infected by microor- wound to improve cell proliferation and metabolism.
ganisms such as bacteria. These factors weaken the po- Therefore, improving the function of fibroblasts and
tential of ADSC-based cell therapy. In contrast to endotheliocytes and promoting angiogenesis are of
ADSCs, SVF can be isolated in real time and contains a great significance to wound healing [30]. Our Masson
sufficient number of cells to eliminate the need for stains and CD31 immunohistochemistry experiments
in vitro amplification. Therefore, SVF could be an ideal showed that hADSCs and SVF significantly induced
cell source for locally applied treatment of refractory collagen levels and microvascular formation in skin
wounds [27]. wounds, which should strongly promote wound heal-
ing. Taken together, hADSCs and SVF promote
hADSCs have multiple characteristics of MSCs wound healing by inducing the angiogenic process
Our results show that hADSCs expanded to the third and collagen production.
passage have characteristics similar to those of the Leiomodin1 is an important factor in the contractile
MSCs. Immunocytochemical tests show that hADSCs activity of smooth muscle cells and maintains vascular
express high levels of CD105, CD44, CD90, and CD29 homeostasis [31]. Leiomodin has three isoforms, with
and are CD45 negative. Additionally, hADSCs expanded Leiomodin1 being mainly expressed in the smooth
from SVF in vitro had adipocytic and osteoblastic differ- muscle cells found in vascular tissue. Although the
entiation potential. In vitro, amplified hADSCs also have structure, function, and distribution of the different
a vigorous proliferation potential. The proliferative activ- leiomodin isoforms vary greatly between tissues, experi-
ity of hADSCs from different patients was significantly ments have shown that actin polymerization is strongly
different; however, hADSCs from a diabetic patient did promoted by Leiomodin in vitro [32]. Actin
not exhibit an impaired proliferative activity. This pro- polymerization can regulate smooth muscle tension and
vides an experimental basis for patients with diabetes to induce vasoconstriction [33]. Endothelin1 is a strong
carry out autologous SVF transplantation for wound vasoconstrictor that is secreted by vascular endothelial
healing. In addition to multipotency, an expanded ADSC cells, which is a powerful vasoconstrictor in mammals. A
exerts paracrine effects to regulate the function and high level of endothelin1 results in vasoconstriction and
mRNA expression of fibroblasts and endothelial cells. decreased compliance of the blood vessel walls and is as-
sociated with inflammation and thrombosis, which can
hADSCs and SVF significantly improve the wound healing lead to heart failure, hypertension, and atherosclerosis
processes by regulating the “cytokine-cytokine receptor [34]. Our data revealed that treatment with hADSCs or
interaction” pathway SVF decreased the expression of Endothelin1 and Leio-
In this study, using a mouse model of skin wound heal- modin1 mRNAs, which potentially improves blood sup-
ing, we investigated the role of hADSCs and SVF in ply to the wound by inhibiting the contraction of local
regulating the wound healing processes and its mecha- blood vessels.
nisms. SVF or hADSCs were subcutaneously injected to RNAseq and network analysis of gene expression pro-
determine their effects on skin wound healing, allowing files of skin tissue around the wound found that the ex-
observation of changes to skin structure after healing. pression of 54 genes in the “cytokine-cytokine receptor
Our results show that wound healing in SVF-treated interaction” pathway has changed after SVF or hADSC
mice was significantly more rapid than in control mice, treatment for 9 days. Chemokines play an important role
but with no significant difference from the hADSC- in regulating multiple processes in wound healing [35],
treated group. Similar to the literature [28], our results which is consistent with the observation that chemo-
show that SVF contains a percentage of hADSCs. Con- kines, chemokines receptor, interleukins, and oncostatin
sidering that SVF and hADSC groups use the same M (Osm) were upregulated in mouse skin treated with
number of cells, therefore, it can be inferred that SVF or hADSCs. Among these genes, expression levels
hADSCs in SVF have better activity than expanded of Osm, Ccl4, and Ccr1, which have the largest changes,
hADSCs. An increase in epidermal thickness, compared were analyzed by qRT-PCR, which was consistent with
to control, was found with SVF or hADSC treatment. the RNAseq results.
Skin contains many types of cells, including dermal fi- Various types of inflammatory cells and factors are in-
broblasts that participate in the synthesis of extracellular volved in inflammatory disorders, which contribute to
matrices, secrete growth factors, and play an important delayed wound healing. OSM, an Il-6 family protein, is
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 17 of 21

secreted by activated macrophages and inhibits the ex- associated with angiogenesis as well as the chemotaxis
pression of inflammatory cytokines TNFα and Il-1β at and activation of macrophages. Our study found that 9
wound sites. In the late stages of inflammation, OSM days after injury, the expression of multiple chemokines
changes the behavior of the inflammatory factors and in- and the density of new capillaries in the wound locally
duces an anti-inflammatory reaction, thereby promoting increased. Vascular and non-vascular cells coordinated
wound healing [36]. Small molecular weight chemotaxic to promote physiological healing of wounds. Interaction
cytokines are called chemokines. In injury, a variety of between capillary endothelial cells and peripheral cells
cells, including endothelial cells and fibroblasts, secrete are required to determine the remodeling process of
cytokines with molecular weights between 8 and 12 kDa microvessels [38]. Evidence suggests that microvascular
[30]. Chemokines can promote angiogenesis and immu- dysfunction is responsible for the pathogenesis of dia-
noregulation in all stages of wound healing, with several betic wound healing [3]. The function of endothelial
contributing to multiple stages of wound healing. CCL4, cells, the basic component of blood vessels, is an import-
also referred to as MIP-1β (macrophage inflammatory ant factor affecting angiogenesis. We found that
protein-1β), is a major chemokinetic factor associated with HUVECs treated with hADSCs increased in tubulogen-
the recruitment of monocytes. Chemokine receptor 1 esis ability.
(CCR1), expressed on the surface of mononuclear cells, The connective tissue contains a large number of fi-
interacts with its ligands CCL3 and CCL5 [37]. Once broblasts, accounting for a majority of their cellular
monocytes enter a wound, they differentiate into macro- components. By synthesizing ECM in the dermis, fibro-
phages, secrete growth factors, and function as antigen- blasts maintain the normal structure of ECM in normal
presenting cells and scavenger cells. A study by Kaesler and injured tissues, thereby promoting wound healing
et al. showed that chemokine CCL-6 is a powerful chemo- [40]. Fibroblasts accelerate the damage repair process by
kinetic factor secreted by macrophages. CCL6 expression increasing matrix synthesis, secreting growth factors,
increases during skin injury and through chemotaxis at- and promoting wound contraction. We found that fibro-
tracts a large number of macrophages, which increases the blasts treated with hADSCs have increased migratory
number of macrophages in the healing of skin wounds ability. In the process of angiogenesis, fibroblasts and
[37]. A variety of CXCs promote the regeneration of blood vascular endothelial cells bond, and chemovascular sig-
vessels around the wound with studies showing that local nals, including growth factors, activate the migration of
levels of CXCL1, CXCL2, CXCL5, CXCL7, CXCL8, and vascular endothelial cells, which eventually forms a com-
CXCL12 increase during wound healing [30]. prehensive vascular network [41]. ECM plays an import-
In our results, CC-chemokines were upregulated in ant role in providing a place for cell adhesion and
wound healing, including CCL2, CCL4, CCL6, CCL7, participates in the formation of new blood vessels [42].
and CCL9, which are all able to chemoattract macro- We used RNAseq and network analysis tools to study
phages; simultaneously, OSM, produced by activated pathways affected by hADSCs. KEGG analysis found that
macrophages, is also upregulated. This suggests a high the hADSC-treated fibroblasts and endotheliocytes had
content of macrophages in the wound. These data similar genetic changes in the “ECM-receptor inter-
demonstrate that macrophage chemotaxis by SVF or action” pathway. By directly interacting with cell surface
hADSCs can offer a potential therapeutic benefit for receptors, ECM provides a substrate for cell adhesion
wound healing. and migration and participates in the repair or regener-
ation of damaged skin tissue [43]. Recent studies have
hADSCs improve angiogenesis by regulating the “ECM- found that some ECM proteins regulate the distribution
receptor interaction” pathway and bioavailability of growth factors, promoting wound
New blood vessels help transport nutrients and oxygen healing through their involvement in controlling the
to the wound area. In the wound healing process, core behaviors of cells [44]. In our study, we verified that
growth of microvessels of the size of capillaries is re- ECM-receptor interaction-related genes, including colla-
quired [38]. The proliferation and subsequent apoptosis gen, laminin, integrins, and syndecans, have significantly
of endotheliocyte, the proliferation and migration of fi- upregulated expression levels and that thrombospondins
broblasts, and the synthesis of collagen are all required were downregulated. Itgb4, Itga7, Itga6, Col4a2, Thbs1,
to simultaneously occur in a coordinated fashion to pro- Tnc, Col4a1, Col1a1, and Thbs3 play key roles in the
mote wound closure. The timing of the changes in the PPI networks identified in fibroblasts and endothelio-
levels of a variety of soluble factors, such as VEGF and cytes after treatment with hADSCs. These regulated
chemical chemokines (CXCL10, CXCR3 ligands), are genes participate in cellular pathways closely associated
carefully regulated to induce the growth of new blood with wound healing.
vessels in the early stages of injury and their degradation Collagen consists of type I collagen (about 80%) and
in the late stages of healing [39]. These factors are also type III collagen (about 10%), which forms a crosslinked
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 18 of 21

network. These two types of collagen are the main fi- the current study, we showed that five members of in-
brins making up healthy dermal ECM [45]. In this study, tegrin family, including a2b, a6, a7, b4, and b7, were up-
type I collagen alpha 1 (Col1a1) was significantly upreg- regulated by hADSCs or SVF in mice skin, fibroblasts,
ulated not only in mice with skin wounds, but also in fi- and endotheliocytes.
broblasts and endotheliocytes treated with hADSCs and Syndecans are transmembrane proteins on the plasma
SVF. Col4a2 and 3, Col6a2 and 3, and Col9a3 were also membrane that participate in cell signal transmission
upregulated by hADSCs in fibroblasts and endothelio- through multiple pathways [47]. Syndecan 1 is the core
cytes. The fibrin molecule surrounds the periphery of protein, with syndecan 4 found in the cytoplasm. Syn-
the elastic fiber bundle composed of highly crosslinked decan interacts with integrins through syndecans 1 and
elastic protein cores and constitutes a bundle of elastic 4 and promotes integrin-mediated angiogenesis and cell
fibers. These elastic fibers are mixed with collagen net- attachment through an increase in Rho-mediated FAK
works to form fibers that are necessary for skin exten- phosphorylation [48]. In this study, we showed that
sion and compliance. After skin injury, the extracellular syndecans 4, 1, and 2 are expressed at higher levels in
matrix in dermis is destroyed with the formation of new fibroblasts and endotheliocytes than syndecan 3, where
fibrin matrix [45]. Importantly, fibrin and other proteins syndecans 1 (fold change: 2.49 in EC, 2.42 in Fs cells)
in the collagen-elastic protein networks interact with and 3 (fold change: 2.42 in EC, 2.72 in Fs cells) expres-
soluble signaling molecules (such as growth factors) and sion significantly increased after treatment with
present cell adhesion sites [45]. The most abundant hADSCs. Syndecan 2 and 4, on the other hand, in-
glycoprotein in the basal membrane ECM is laminins. creased to a lesser extent. As syndecan 1 not only has
Laminins are cruciform trimers whose surface is a three- higher expression than the other syndecans, but also
stranded spiral structure that is widespread in a variety has a large increase in expression, suggests that synde-
of organizations. The laminin structure consists of three can 1 plays an important role in promoting wound
chains, α, β, and γ, and is accordingly named by its αβγ healing.
chain. Cellular receptors for laminins include the integ- Various metzincin enzymes participate in the activa-
rins and syndecans. Laminins are involved in regulating tion of syndecan, and hydrolyzed fragments have bio-
core cellular activity, such as cell adhesion and migra- logical activity. Studies have found that a variety of
tion, cell apoptosis, proliferation, and differentiation, and MMPs can digest syndecan to release active fragments
provide a structural scaffold for cells. As such, in skin in vitro and in vivo [49]. For example, activation of syn-
wound healing, laminins play a critical role in re- decans 1, 2, and 4 requires cleavage by the gelatinases
epithelialization and angiogenesis. Studies have shown MMPs 2 and 9 [50]. In this study, we found that mmp9
that the rate of protease hydrolysis of LAMA3 is related mRNA expression increased, as measured by RNAseq
to wound healing. Laminin has a high affinity with a var- and qRT-PCR, in fibroblasts and endotheliocytes after
iety of growth factors, and the α chain laminin type G treatment by hADSCs.
domain is the binding domain with heparin. These re- Thrombospondins are a family of large multi-domain
gions promote the attachment of endothelial and fibro- glycoproteins that regulate a range of cellular functions,
blasts cells by binding to the cell surface receptor including intracellular signals, migration, and prolifera-
syndecan. The application of laminin for wound healing tion, and they interact with a wide range of receptors,
in type 2 diabetic mice can significantly increase the per- growth factors, proteases, and matrix molecules [43].
ipheral growth factor concentration and promote wound Thrombospondin 1 (thbs1) inhibits the migration and
healing [44]. Here, we showed that Lama3, Lamb3, and proliferation of endothelial cells and promotes apoptosis
Lamc3 were significantly upregulated in fibroblasts, of endothelial cells by directly regulating CD36 and
endotheliocytes, and mice wound skin after treatment with CD47 receptors, thus inhibiting angiogenesis [43, 51]. In
hADSCs or SVF. This upregulation of laminin may be clin- this study, we found that thbs1 and thbs3 mRNA expres-
ically useful and lead to effective tissue regeneration. sion decreased in endotheliocytes and fibroblasts treated
Integrin is the main type of transmembrane cell sur- by hADSCs. This downregulation of thrombospondins
face receptor involved in the interaction between cells may promote angiogenesis.
and the extracellular matrix. Collagen and glycoprotein Taken together, these data suggest that hADSCs pro-
in the ECM have multiple cell-binding sites, including mote endotheliocyte and fibroblast adhesion to ECM
those for integrin receptors, and provide platforms for proteins, which is mediated by ECM-receptor inter-
migrating cells [30]. Integrin can regulate cell adhesion action. At the same time, hADSCs promoted migration
and migration as well as other cellular processes, includ- of fibroblasts and tubulogenesis properties of endothelio-
ing, for example, proliferation and differentiation [46], cytes and inhibited the contraction of local blood vessels,
by identifying short sequences present in the extracellu- all of which cooperate to improve blood supply and
lar matrix, including collagen, histone, and fibrin [30]. In angiogenesis during wound healing.
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 19 of 21

Fig. 12 Proposed molecular mechanisms for improved wound healing by SVF. SVF promotes wound healing by inducing angiogenesis, matrix
remodeling, and cell migration. Most processes are initiated through increased expression of chemokines and cytokines. Red arrows indicate the
increase initiated by SVF or ADSCs. Blue arrows indicate the decrease

In summary, both SVF and hADSCs improve the func- Acknowledgements

tion of endotheliocytes and fibroblasts, regulate gene ex- Not applicable.

pression, and jointly promote skin healing (Fig. 12).

There is no significant difference in the effect, or mech- Authors’ contributions
anism, between SVF and hADSCs. Considering the con- HB, HT, and DMI are the guarantors of this work, and as such, they had full
access to all the data in this study and take full responsibility for the integrity
venience of SVF applications, SVF can replace hADSCs
of the data and accuracy of the data analysis. HB and HL researched the
for wound healing. data and wrote the manuscript; DMI reviewed and edited the manuscript;
CZ, YM, FN, YX, WS, and XW researched the data. All authors read and
approved the final manuscript.

Conclusions
Stem cell therapy is a new research direction in medi- Funding
cine, and SVF-assisted treatment of wounds is a promis- This study was sponsored by the interdisciplinary medicine Seed Fund of
Peking University and the Fundamental Research Funds for the Central
ing treatment. Various mechanisms are likely involved,
Universities (Grant Numbers BMU2018MX001), the National Natural Science
including migration of fibroblasts, and tubulogenesis of Foundation of China (NSFC) (Grant Numbers 81874321 and 81673459),
endotheliocytes through the regulation of cell adhesion National Key Technologies R&D Program (Grant Numbers 2015BAK45B01,
2012BAK25B01), and a grant from the National Science Foundation of China
and the cytokine pathway. Since SVF affects multiple as-
– Canadian Institutes of Health Research (NSFC-CIHR) China-Canada Joint
pects of wound healing, we predict that interactions with Health Research Initiative (Grant Numbers 81061120525 and CCI-109605).
surrounding cells and intricate paracrine activity play
important roles in vivo. Further study of the interaction
Availability of data and materials
network of SVF and wound microenvironment is neces- The datasets used and/or analyzed during the current study are available
sary and should provide a theoretical basis for the ad- from the corresponding author on reasonable request.
vancement of SVF-related cell therapy.
Ethics approval and consent to participate
Abbreviations All experiments were performed in accordance with the guidelines and
Angptl7: Angiopoietin-like 7; CCL: Chemokine (C-C motif) ligand; study protocols of the Peking University Third Hospital Medical Science
CCR: Chemokine receptor; Col: Collagen; EC: Endothelial cells; Research Ethics Committee (Approval No. IRB00006761-M2017391).
ECM: Extracellular matrix; Edn1: Endothelin1; FPKM: Fragments per kilobase of
exon model per million reads mapped; Fs: Fibroblasts; hADSCs: Human
adipose-derived stem cells; HE: Hematoxylin and eosin; hFs: Primary human Consent for publication
fibroblasts; HUVEC: Human umbilical vein endothelial cell; Il: Interleukin; Not applicable.
Itg: Integrins; Lam: Laminin; Lmod1: Leiomodin1; MSC: Mesenchymal stem
cells; Osm: Oncostatin M; qRT-PCR: Quantitative reverse transcription
polymerase chain reaction; STZ: Streptozocin; SVF: Stromal vascular fraction; Competing interests
Thbs: Thrombospondins; Vipr1: Vasoactive intestinal polypeptide receptor The authors declare that they have no competing interests.
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 20 of 21

Received: 10 July 2019 Revised: 7 September 2019 24. Demidova-Rice TN, Durham JT, Herman IM. Wound healing angiogenesis:
Accepted: 11 September 2019 innovations and challenges in acute and chronic wound healing. Adv
Wound Care (New Rochelle). 2012;1(1):17–22.
25. Morris ME, Beare JE, Reed RM, Dale JR, LeBlanc AJ, Kaufman CL, Zheng H,
References Ng CK, Williams SK, Hoying JB. Systemically delivered adipose stromal
1. Wang L, Gao P, Zhang M, Huang Z, Zhang D, Deng Q, Li Y, Zhao Z, Qin X, vascular fraction cells disseminate to peripheral artery walls and reduce
Jin D, Zhou M, Tang X, Hu Y, Wang L. Prevalence and ethnic pattern of vasomotor tone through a CD11b + cell-dependent mechanism. Stem Cells
diabetes and prediabetes in China in 2013. JAMA. 2017;317(24):2515–23. Transl Med. 2015;4(4):369–80.
2. Verhulst MJL, Loos BG, Gerdes VEA, Teeuw WJ. Evaluating all potential oral 26. Granel B, Daumas A, Jouve E, Harle JR, Nguyen PS, Chabannon C, Colavolpe
complications of diabetes mellitus. Front Endocrinol. 2019;10:56. N, Reynier JC, Truillet R, Mallet S, Baiada A, Casanova D, Giraudo L, Arnaud L,
3. Okonkwo UA, DiPietro LA. Diabetes and wound angiogenesis. Int J Mol Sci. Veran J, Sabatier F, Magalon G. Safety, tolerability and potential efficacy of
2017;18(7):1419. injection of autologous adipose-derived stromal vascular fraction in the
4. Cheng B, Fu X. The focus and target: angiogenesis in refractory wound fingers of patients with systemic sclerosis: an open-label phase I trial. Ann
healing. Int J Lower Extrem Wounds. 2018;17(4):301–3. Rheum Dis. 2015;74(12):2175–82.
5. Virador GM, de Marcos L, Virador VM. Skin wound healing: refractory 27. Zhou L, Xu L, Shen J, Song Q, Wu R, Ge Y, Xin H, Zhu J, Wu J, Jia R.
wounds and novel solutions. Methods Mol Biol. 2019;1879:221–41. Preischemic administration of nonexpanded adipose stromal vascular
6. Elraiyah T, Tsapas A, Prutsky G, Domecq JP, Hasan R, Firwana B, Nabhan M, fraction attenuates acute renal ischemia/reperfusion injury and fibrosis.
Prokop L, Hingorani A, Claus PL, Steinkraus LW, Murad MH. A systematic Stem Cells Transl Med. 2016;5(9):1277–88.
review and meta-analysis of adjunctive therapies in diabetic foot ulcers. J 28. El-Habta R, Sloniecka M, Kingham PJ, Backman LJ. The adipose tissue
Vasc Surg. 2016;63(2 Suppl):46S–58S e1–2. stromal vascular fraction secretome enhances the proliferation but inhibits
7. Wang Z, Hasan R, Firwana B, Elraiyah T, Tsapas A, Prokop L, Mills JL Sr, the differentiation of myoblasts. Stem Cell Res Ther. 2018;9(1):352.
Murad MH. A systematic review and meta-analysis of tests to predict 29. Qiu S, Jia Y, Sun Y, Han P, Xu J, Wen G, Chai Y. Von Hippel-Lindau (VHL)
wound healing in diabetic foot. J Vasc Surg. 2016;63(2 Suppl):29S–36S e1–2. protein antagonist VH298 improves wound healing in streptozotocin-
8. Eming SA, Martin P, Tomic-Canic M. Wound repair and regeneration: induced hyperglycaemic rats by activating hypoxia-inducible factor- (HIF-) 1
mechanisms, signaling, and translation. Sci Transl Med. 2014;6(265):265sr6. signalling. J Diabetes Res. 2019;2019:1897174.
9. Li X, Li D, Wikstrom JD, Pivarcsi A, Sonkoly E, Stahle M, Landen NX. 30. Ridiandries A, Tan JTM, Bursill CA. The role of chemokines in wound healing.
MicroRNA-132 promotes fibroblast migration via regulating RAS p21 protein Int J Mol Sci. 2018;19(10):3217.
activator 1 in skin wound healing. Sci Rep. 2017;7(1):7797. 31. Nanda V, Miano JM. Leiomodin 1, a new serum response factor-dependent
10. Sun M, He Y, Zhou T, Zhang P, Gao J, Lu F. Adipose extracellular matrix/ target gene expressed preferentially in differentiated smooth muscle cells. J
stromal vascular fraction gel secretes angiogenic factors and enhances skin Biol Chem. 2012;287(4):2459–67.
wound healing in a murine model. Biomed Res Int. 2017;2017:3105780. 32. Boczkowska M, Rebowski G, Kremneva E, Lappalainen P, Dominguez R. How
11. Nicholas MN, Yeung J. Current status and future of skin substitutes for Leiomodin and Tropomodulin use a common fold for different actin
chronic wound healing. J Cutan Med Surg. 2017;21(1):23–30. assembly functions. Nature Communications. 2015;6:8314.
12. Han G, Ceilley R. Chronic wound healing: a review of current management 33. Weise-Cross L, Sands MA, Sheak JR, Broughton BRS, Snow JB, Gonzalez Bosc
and treatments. Adv Ther. 2017;34(3):599–610. LV, Jernigan NL, Walker BR, Resta TC. Actin polymerization contributes to
13. Duscher D, Barrera J, Wong VW, Maan ZN, Whittam AJ, Januszyk M, Gurtner enhanced pulmonary vasoconstrictor reactivity after chronic hypoxia. Am J
GC. Stem cells in wound healing: the future of regenerative medicine? A Physiol Heart Circ Physiol. 2018;314(5):H1011–H21.
mini-review. Gerontology. 2016;62(2):216–25. 34. Davenport AP, Hyndman KA, Dhaun N, Southan C, Kohan DE, Pollock JS,
14. Kirby GT, Mills SJ, Cowin AJ, Smith LE. Stem cells for cutaneous wound Pollock DM, Webb DJ, Maguire JJ. Endothelin. Pharmacol Rev. 2016;68(2):
healing. Biomed Res Int. 2015;2015:285869. 357–418.
15. Nourian Dehkordi A, Mirahmadi Babaheydari F, Chehelgerdi M, Raeisi 35. Falanga V. Wound healing and its impairment in the diabetic foot. Lancet.
Dehkordi S. Skin tissue engineering: wound healing based on stem-cell- 2005;366(9498):1736–43.
based therapeutic strategies. Stem Cell Res Ther. 2019;10(1):111. 36. Ganesh K, Das A, Dickerson R, Khanna S, Parinandi NL, Gordillo GM, Sen CK,
16. Chen YW, Scutaru TT, Ghetu N, Carasevici E, Lupascu CD, Ferariu D, Pieptu Roy S. Prostaglandin E(2) induces oncostatin M expression in human
D, Coman CG, Danciu M. The effects of adipose-derived stem cell- chronic wound macrophages through Axl receptor tyrosine kinase pathway.
fifferentiated adipocytes on skin burn wound healing in rats. J Burn Care J Immunol. 2012;189(5):2563–73.
Res. 2017;38(1):1–10. 37. Balaji S, Watson CL, Ranjan R, King A, Bollyky PL, Keswani SG. Chemokine
17. Seo E, Lim JS, Jun JB, Choi W, Hong IS, Jun HS. Exendin-4 in combination involvement in fetal and adult wound healing. Adv Wound Care (New
with adipose-derived stem cells promotes angiogenesis and improves Rochelle). 2015;4(11):660–72.
diabetic wound healing. J Transl Med. 2017;15(1):35. 38. Nagaraja S, Chen L, DiPietro LA, Reifman J, Mitrophanov AY. Predictive
18. van Dongen JA, Harmsen MC, van der Lei B, Stevens HP. Augmentation of approach identifies molecular targets and interventions to restore
dermal wound healing by adipose tissue-derived stromal cells (ASC). angiogenesis in wounds with delayed healing. Front Physiol. 2019;10:636.
Bioengineering. 2018;5(4):91. 39. DiPietro LA. Angiogenesis and wound repair: when enough is enough. J
19. Deng C, Wang L, Feng J, Lu F. Treatment of human chronic wounds with Leukoc Biol. 2016;100(5):979–84.
autologous extracellular matrix/stromal vascular fraction gel: A STROBE- 40. Dulmovits BM, Herman IM. Microvascular remodeling and wound healing: a
compliant study. Medicine. 2018;97(32):e11667. role for pericytes. Int J Biochem Cell Biol. 2012;44(11):1800–12.
20. Sabol RA, Bowles AC, Cote A, Wise R, Pashos N, Bunnell BA. Therapeutic 41. Genova T, Grolez GP, Camillo C, Bernardini M, Bokhobza A, Richard E,
potential of adipose stem cells. Adv Exp Med Biol. 2018. https://doi.org/10. Scianna M, Lemonnier L, Valdembri D, Munaron L, Philips MR, Mattot V,
1007/5584_2018_248. Serini G, Prevarskaya N, Gkika D, Pla AF. TRPM8 inhibits endothelial cell
21. Abo-Aziza FAM, A AZ. The impact of confluence on bone marrow migration via a non-channel function by trapping the small GTPase Rap1. J
mesenchymal stem (BMMSC) proliferation and osteogenic differentiation. Cell Biol. 2017;216(7):2107–30.
Int J Hematol-Oncol Stem Cell Res. 2017;11(2):121–32. 42. Moniri M, Boroumand Moghaddam A, Azizi S, Abdul Rahim R, Zuhainis Saad
22. Bourin P, Bunnell BA, Casteilla L, Dominici M, Katz AJ, March KL, Redl H, W, Navaderi M, Arulselvan P, Mohamad R. Molecular study of wound
Rubin JP, Yoshimura K, Gimble JM. Stromal cells from the adipose tissue- healing after using biosynthesized BNC/Fe3O4 nanocomposites assisted
derived stromal vascular fraction and culture expanded adipose tissue- with a bioinformatics approach. Int J Nanomedicine. 2018;13:2955–71.
derived stromal/stem cells: a joint statement of the International Federation 43. Masli S, Sheibani N, Cursiefen C, Zieske J. Matricellular protein
for Adipose Therapeutics and Science (IFATS) and the International Society thrombospondins: influence on ocular angiogenesis, wound healing and
for Cellular Therapy (ISCT). Cytotherapy. 2013;15(6):641–8. immuneregulation. Curr Eye Res. 2014;39(8):759–74.
23. Madrigal A, Tan L, Zhao Y. Expression regulation and functional analysis of 44. Ishihara J, Ishihara A, Fukunaga K, Sasaki K, White MJV, Briquez PS, Hubbell
RGS2 and RGS4 in adipogenic and osteogenic differentiation of human JA. Laminin heparin-binding peptides bind to several growth factors and
mesenchymal stem cells. Biol Res. 2017;50(1):43. enhance diabetic wound healing. Nature communications. 2018;9(1):2163.
Bi et al. Stem Cell Research & Therapy (2019) 10:302 Page 21 of 21

45. Briquez PS, Hubbell JA, Martino MM. Extracellular matrix-inspired growth
factor delivery systems for skin wound healing. Adv Wound Care (New
Rochelle). 2015;4(8):479–89.
46. LaFoya B, Munroe JA, Miyamoto A, Detweiler MA, Crow JJ, Gazdik T, Albig
AR. Beyond the matrix: the many non-ECM ligands for integrins. Int J Mol
Sci. 2018;19(2):449.
47. Park PW. Isolation and functional analysis of syndecans. Methods Cell Biol.
2018;143:317–33.
48. Vuong TT, Reine TM, Sudworth A, Jenssen TG, Kolset SO. Syndecan-4 is a
major syndecan in primary human endothelial cells in vitro, modulated by
inflammatory stimuli and involved in wound healing. J Histochem
Cytochem. 2015;63(4):280–92.
49. Jang B, Jung H, Choi S, Lee YH, Lee ST, Oh ES. Syndecan-2 cytoplasmic
domain up-regulates matrix metalloproteinase-7 expression via the protein
kinase Cgamma-mediated FAK/ERK signaling pathway in colon cancer. J
Biol Chem. 2017;292(39):16321–32.
50. Manon-Jensen T, Multhaupt HA, Couchman JR. Mapping of matrix
metalloproteinase cleavage sites on syndecan-1 and syndecan-4
ectodomains. FEBS J. 2013;280(10):2320–31.
51. Gunda V, Sudhakar YA. Regulation of tumor angiogenesis and choroidal
neovascularization by endogenous angioinhibitors. J Cancer Sci Therapy.
2013;5(12):417–26.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.