VCAM-1+ placenta chorionic villi-derived mesenchymal stem cells display potent pro-angiogenic activity

Introduction Mesenchymal stem cells (MSCs) represent a heterogeneous cell population that is promising for regenerative medicine. The present study was designed to assess whether VCAM-1 can be used as a marker of MSC subpopulation with superior angiogenic potential. Methods MSCs were isolated from placenta chorionic villi (CV). The VCAM-1+/− CV-MSCs population were separated by Flow Cytometry and subjected to a comparative analysis for their angiogenic properties including angiogenic genes expression, vasculo-angiogenic abilities on Matrigel in vitro and in vivo, angiogenic paracrine activities, cytokine array, and therapeutic angiogenesis in vascular ischemic diseases. Results Angiogenic genes, including HGF, ANG, IL8, IL6, VEGF-A, TGFβ, MMP2 and bFGF, were up-regulated in VCAM-1+CV-MSCs. Consistently, angiogenic cytokines especially HGF, IL8, angiogenin, angiopoitin-2, μPAR, CXCL1, IL-1β, IL-1α, CSF2, CSF3, MCP-3, CTACK, and OPG were found to be significantly increased in VCAM-1+ CV-MSCs. Moreover, VCAM-1+CV-MSCs showed remarkable vasculo-angiogenic abilities by angiogenesis analysis with Matrigel in vitro and in vivo and the conditioned medium of VCAM-1+ CV-MSCs exerted markedly pro-proliferative and pro-migratory effects on endothelial cells compared to VCAM-1−CV-MSCs. Finally, transplantation of VCAM-1+CV-MSCs into the ischemic hind limb of BALB/c nude mice resulted in a significantly functional improvement in comparison with VCAM-1−CV-MSCs transplantation. Conclusions VCAM-1+CV-MSCs possessed a favorable angiogenic paracrine activity and displayed therapeutic efficacy on hindlimb ischemia. Our results suggested that VCAM-1+CV-MSCs may represent an important subpopulation of MSC for efficient therapeutic angiogenesis. Electronic supplementary material The online version of this article (doi:10.1186/s13287-016-0297-0) contains supplementary material, which is available to authorized users.


Introduction
Peripheral arterial disease (PAD), characterized by the critical limb ischemia (CLI) with high morbidity and mortality risks, is gradually becoming an urgent lifethreatening disease in our aging society. To date, the main treatments for PAD are bypass grafting and endarterectomy. However, surgery has not always been allowed [1]. Numerous studies have demonstrated that mesenchymal stem cells (MSCs) derived from different tissue sources exert therapeutic efficacy on ischemia [2][3][4][5]. Varieties of reports have highlighted the therapeutic angiogenesis of MSCs by focusing on differentiation and paracrine mechanisms [6]. Several angiogenic cytokines and enzymes secreted by MSCs, including vascular endothelial cell growth factor (VEGF)-A [7], hepatocyte growth factor (HGF) [8], interleukin (IL)-8 [9], transforming growth factor beta (TGFβ) [10], matrix metalloproteinases (MMPs) [11], and so forth, have been widely reported to initiate angiogenesis. Based on their angiogenic properties, MSCs are attractive in various clinical trials [12]. However, MSCs have been known to be heterogeneous [13,14] and it remains to be determined whether some MSC subpopulations exert superior angiogenic activities and are more suitable for therapeutic angiogenesis.
Vascular cell adhesion molecule 1 (VCAM-1), also known as CD106, is extensively expressed on endothelial cells [15], and is also constitutively expressed on some stromal cells, existing in a particular vascular niche [16]. VCAM-1 plays a critical role in early embryonic development since VCAM-1-deficient mice often die early or show multiple severe defects in placental development [17]. In addition, soluble VCAM-1 (sVCAM-1) has shown evidence of mediating angiogenesis in rat cornea [18] and the sVCAM-1/α4 integrin pathway plays an important role in inflammatory stimuli-induced angiogenesis [19]. Recent studies demonstrated that VCAM-1 overexpression was associated with tumor angiogenesis, such as gastric carcinoma [20], breast cancer [21], and renal cancer [22]. These studies suggest VCAM-1 may be involved in angiogenesis.
We have previously isolated a VCAM-1 + MSC subpopulation of placenta chorionic villi (CV) that displayed unique immunomodulation capacity. VCAM-1 + CV-MSCs secreted not only inflammatory factors but also angiogenic cytokines [23]. The aim of this work was to assess the angiogenic potential of the VCAM-1 + CV-MSC subpopulation, and to explore its therapeutic application in an animal model of vascular ischemic disease.

Cell isolation and culture
This study was approved by the Ethical Committee and the Institutional Review Board of the Chinese Academy of Medical Science and Peking Union Medical College, Tianjin, China. All volunteers provided informed consent. CV-MSCs were harvested and cultured as described previously [23]. The regular culture medium for CV-MSCs was DF12 medium (Gibco, Grand Island, NY, USA), 10 % fetal bovine serum (FBS), 10 ng/ml epidermal growth factor (EGF; Peprotech, Rocky Hill, NJ, USA), 2 mM glutamine (Sigma, St.Louis, MO, USA), 1 % nonessential amino acids (Gibco), and 100 U/ml penicillin-streptomycin (Invitrogen, Carlsbad, CA, USA). Human umbilical vein endothelial cells (HUVECs) were harvested by digesting umbilical vein with 0.25 % trypsin (Gibco) for 15 minutes at 37°C. The HUVECs were then cultured in EGM2-MV (Lonza, Walkersville, MD, USA).

RNA extraction, reverse transcription, and real-time PCR
Total RNA was extracted using the E.Z.N.A. Total RNA Kit I (OMEGA, Norcross, GA, USA), and cDNA synthesis was performed using the MLV RT kit (Invitrogen). All of the procedures followed the manufacturer's instructions. Real-time PCR was performed on an Applied Bio system 7900 Real-Time PCR System (Foster City, CA, USA), using a SYBR Green-based real-time detection method. Primers used are shown in Additional file 1: Table S1. Each sample was performed in triplicate.

Tubular network formation assay in vitro
Pairs of VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs were seeded at 2 × 10 4 cells/well gently on a Matrigelcoated (BD Biosciences, Bedford, MA, USA) 96-well plate. Photographs were taken by Microscope (Olympus, Melville, NY, USA) 12 hours later. Tube numbers in each well were counted. Three pairs of VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs were used, and each sample was performed in triplicate.

Matrigel plug angiogenesis assay in vivo
Six-week-old nude male mice were purchased from the Institute of Experimental Animal (Beijing, China). All of the animal experiments followed the Peking Union Medical College Animal Care and Use Committee guidelines. VCAM-1 +/− CV-MSCs or nonseparated (NS) CV-MSCs (10 6 cells) were suspended in 400 μl Matrigel and injected subcutaneously into the dorsal area of nude mice. Matrigel supplement with phosphate-buffered saline (PBS) served as the negative control. Each group contained three to six mice. Three weeks later, Matrigel implants were harvested, photographed, fixed, sliced, and stained with hematoxylin and eosin (H & E; Sigma). Vessel numbers were counted under the microscope. Frozen slices stained with alpha-smooth muscle actin (α-SMA; Invitrogen) and von Willebrand factor (vWF; Abcam, Cambridge, MA, USA) were employed to detect the neovascular structures in the Matrigel plug. Photographs were taken at × 20 and × 60 objectives by confocal microscopy (UltraView; Perkin-Elmer, Waltham, Massachusetts, USA).

Conditioned medium preparation and proliferation assay
Pairs of 10 6 VCAM-1 +/− CV-MSCs were incubated in EBM2 medium (Lonza) for 48 hours. Then their conditioned mediums (CMs) were collected, centrifuged at 1800 rpm for 10 minutes to remove cell debris, filtered through 0.2 μm filters (Pall Corporation, Ann Arbor, MI, USA), and frozen at -80°C. To determine the proproliferative effect, VCAM-1 +/− CV-MSC CM supplemented with 2 % FBS were used to culture HUVECs for 72 hours. EBM2 supplemented with 2 % FBS, and EGM2-MV (endothelial cells commercial culture medium; Lonza) served as the negative and positive control, respectively. The Cell Counting Kit 8 (Dojindo, Rockville, MD, USA) method was used to measure HUVEC proliferation at 24, 48, and 72 hours. ΔOD450 indicated the final data after subtracting the background. Each sample was performed in quadruplicate.

Scratch wound healing assay
When endothelial cells reached confluence, a scratch wound was generated across each well using a pipette tip. After washing with PBS, pairs of CM supplemented with 2 % FBS, EGM2-MV, or EBM2 + 2 % FBS were used to culture endothelial cells for 18 hours. The cleared area of each well was photographed under × 40 magnification at 0 and 18 hours, and measured by ImageJ software (NIH, USA). The percentage of area repopulation was calculated by the following formula:

Human cytokine antibody array
The human cytokine antibody array (AAH-CYT-G1000) was performed following the manufacturer's instructions (RayBiotech, Norcross, GA, USA) to detect 120 cytokine expressions in supernatants (SN) of VCAM-1 +/− CV-MSCs. Cytokine signals above 200 were further studied, and the cytokine signal ratio in VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs was calculated. This was statistically significant if the cytokine signal ratio was >1.3 or <0.75. Two pairs of VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs were used. Each sample was performed in duplicate. The targeted names of all cytokines involved are presented in Additional file 1: Table S2.

Transplantation of VCAM-1 +/− CV-MSCs in the hind limb ischemia model
Nude mice (male, 7-8 weeks old, 18-22 g) were intraperitoneally anesthetized with 100 mg/kg sodium pentobarbital (Sigma). Unilateral femoral artery ligation and excision were performed as described previously [24]. Nude mice were randomly divided into three groups (PBS, VCAM-1 + CV-MSCs, and VCAM-1 − CV-MSCs groups) after arterial ligations, and then 100 μl of a 10 6 cell suspension or PBS was intramuscularly injected into ischemic hind limbs within 6 hours post surgery. Blood perfusion in ischemia and nonischemia limbs was measured by the PeriCam PSI System (PERIMED AB Company, Järfälla, stockholm, Sweden) on day 0, day 7, and day 20. Ischemia damage and functional assessment of ischemic hind limbs in each treatment group were assessed on day 20 according to the semiquantitative scores that had been described previously [24].

Angiography
On day 20, after blood perfusion detection, mice were sacrificed for angiography to evaluate the vessel density in ischemic limbs. Angiographic images of hind limbs in three treatments were acquired by the Kodak In-Vivo FX ProImaging System (Kodak, New Haven, Connecticut, USA), and the angiography score was employed [24] to quantitatively analyze the collateral vessel formation at the ischemia site.

Histological analysis
On day 20, after angiography, the ischemia adductor muscle of nude mice in each group was collected, fixed in 10 % formaldehyde (Sigma) overnight, and embedded in paraffin. To detect capillary densities in ischemic sites, H & E staining was performed and images were taken under × 200 magnification. Vessels containing barium sulfate or erythrocytes were counted, and the vessel density in each group was calculated and compared.

Statistical analysis
Statistical analysis was performed using Graph Pad Prism 6.0 (Graph Pad Software, Inc., San Diego, CA, USA). All data are presented as mean ± standard error of the mean. The Mann-Whitney test and one-way analysis of variance (ANOVA) were performed to determine the significance. Fisher's exact test (Freeman-Halton) was employed to assess the outcome of transplantation via a 3 × 3 contingency table. The difference was considered to be significant if p <0.05.

VCAM-1 + CV-MSCs displayed angiogenic potential on Matrigel assay in vitro and in vivo
To determine the angiogenic potential of VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs, a tubular network assay was performed in vitro. To our surprise, without exogenous VEGF, VCAM-1 + CV-MSCs spontaneously formed about 4.14-fold intact tubular structures on Matrigel compared with VCAM-1 − CV-MSCs (n = 3, p <0.01; Fig. 3a). Matrigel plug angiogenesis assays in vivo [25] were then performed to explore the angiogenic differences. Interestingly, plenty of macroscopic blood vessels were observed in the Matrigel plugs of the VCAM-1 + CV-MSCs and NS CV-MSCs groups rather than the VCAM-1 − CV-MSCs and PBS groups (Fig. 3b-i). H & E staining revealed that the new outgrowth contained erythrocytes and the smooth muscle layer (Fig. 3b ii, iii). Moreover, vessel densities in the VCAM-1 + CV-MSCs and NS CV-MSCs groups were significantly higher than in the VCAM-1 − CV-MSCs and PBS groups (10.66 ± 0.67 and 11.84 ± 1.23 per mm 2 vs. 0.36 ± 0.24 and 0.27 ± 0.19 per mm 2, n = 3, p <0.0001; Fig. 3c). However, the vessel density in the VCAM-1 + CV-MSCs and NS CV-MSCs groups was similar (p >0.05). Besides that, a larger vessel lumen was observed in the VCAM-1 + CV-MSCs group rather than in the NS CV-MSCs group, which could be related to a higher VCAM-1 + CV-MSC proportion in the transplanted cells. Moreover, immunostaining of vWF and α-SMA revealed that the fresh blood vessels contained endothelial cells (labeled with anti-vWF antibodies) and smooth muscle cells (labeled with anti-α-SMA antibodies; Fig. 3d), indicating that the vessel structures were intact and mature.

VCAM-1 + CV-MSC CM effectively promoted endothelial cell proliferation and migration
To explore the paracrine activities of VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs, we collected their CMs and performed endothelial cell proliferation and scratch wound healing assay. Our data revealed that compared with the VCAM-1 − CV-MSC CM , VCAM-1 + CV-MSC CM significantly promoted endothelial cell proliferation during 48 hours (n = 3, p <0.01), with the most significant point at 24 hours (n = 3, p <0.001). But this pro-proliferative effect was not significant after 72 hours (n = 3, p >0.05; Fig. 4a). The reason for this might be the exhaustion of angiogenic cytokines. In addition, scratch assay that mimicked the wound healing process in vitro was used to evaluate the promigratory effects. After incubation for 18 hours, we surprisingly found that endothelial cells cultured in VCAM-1 + CV-MSC CM reached confluence again. Representative photographs were taken under × 40 magnification and the percentage of area repopulation was calculated by Image J software (NIH, USA) (Fig. 4b). VCAM-1 + CV-MSC CM significantly increased the cleared area recovery compared with VCAM-1 − CV-MSC CM (80.58 ± 6.88 vs. 56.36 ± 4.23, n = 3, p <0.01; Fig. 4c), indicating that VCAM-1 + CV-MSC CM was richer in pro-migratory cytokines than VCAM-1 − CV-MSC CM . To figure out the paracrine mechanism of VCAM-1 + CV-MSCs, we performed VEGF and  Fig. 4d), while the sVCAM-1 concentration was <20 pg/ ml (Additional file 1: Figure. S1). The fact that VEGF can induce endothelial cell proliferation and migration [7] may partially explain the pro-proliferative and promigratory differences between VCAM-1 + CV-MSC CM and VCAM-1 − CV-MSC CM on endothelial cells.

VCAM-1 + CV-MSCs exerted therapeutic efficacy on hind limb ischemia
To investigate the therapeutic neovascularization of VCAM-1 + CV-MSCs, we constructed a vascular ischemia animal model and intramuscularly injected 10 6 VCAM-1 +/− CV-MSCs into the ischemic limbs within 6 hours post surgery. PBS served as a negative control. To estimate the therapeutic effect, we classified mice into three outcomes: limb salvage, foot necrosis, and limb loss. Different percentage distributions of outcomes among three groups were calculated and Fisher's exact test (Freeman-Halton) was used to analyze this result (p = 0.10, n = 11; Fig. 6a). From the data, mice in the PBS group suffered the maximal amputation rate (54.5 %) and foot necrosis rate (27 %). The amputation rate in the VCAM-1 − CV-MSCs group was much higher than in the VCAM-1 + CV-MSCs group (36.4 % vs. 9 %), while the foot necrosis rate in both of them was 18.2 %. Semiquantitative scores of ischemia damage and ambulatory impairment were used to assess ischemic states and physiological function of ischemic limbs. Results indicated that VCAM-1 + CV-MSCs significantly alleviated the ischemia damage and ambulatory impairment (0.77 ± 0.37 and 0.59 ± 0.24), much better than the PBS group (2.77 ± 0.52 and 1.82 ± 0.33, n = 11, p <0.05), while VCAM-1 − CV-MSCs showed a slight improvement compared with PBS treatment (1.86 ± 0.57 and 1.18 ± 0.36, n = 11, p >0.05; Fig. 6b, c).

Discussion
Previous studies have reported that MSCs displayed remarkable therapeutic properties on vascular ischemic diseases such as myocardial infarction, stroke, and perivascular ischemic diseases [12]. However, the mechanisms of therapeutic angiogenesis induced by MSCs have not yet been well defined. Several investigators have proposed that paracrine factors secreted from MSCs, including a core of angiogenic cytokines (i.e., VEGF, HGF, IL-8, TGFβ), exosomes [26], and microvesicles [27], might be the major contributors [28]. Gnecchi et al. [29] reported that injection with the CM of Akt-modified MSCs abundant with VEGF, bFGF, HGF, and TB4 significantly improved cardiac performance after induced myocardial infarction. Recent studies using cell labeling [30] and single cell technology [31] also supported the major status of paracrine action in MSC-mediated angiogenesis.
In this study, we have firstly demonstrated that the VCAM-1 + CV-MSC subpopulation displayed a potent angiogenic property and exerted enhanced therapeutic efficacy on regeneration after ischemia in comparison with the VCAM-1 − CV-MSC subpopulation.
We then wanted to know why VCAM-1 + CV-MSCs possessed superior pro-angiogenic activities than VCAM-1 − CV-MSCs. We were interested to note a superior angiogenic secretome from VCAM-1 + CV-MSCs, including HGF, IL-8, ANG, ANGPT2, CXCL1/GRO-α, μPAR, IL-1β, IL-1α, CSF2/GM-CSF, CSF3/G-CSF, MCP-3, CTACK/ CCL27, and OPG. Previous studies have shown that HGF potently stimulated endothelial cell motility and growth [32]. IL-8 promoted angiogenesis via directly enhancing endothelial cell proliferation, survival, and MMP production [33]. ANG potently induced new blood vessel formation [34]. ANGPT-2 potentiated the effects of other angiogenic cytokines in vivo and initiated neovascularization [35]. CXCL1 enhanced microvascular endothelial cell  migration and tube formation [36]. μPAR induced endothelial cell invasion and proliferation in the initial period of angiogenesis [37]. IL-1β [38], IL-1α [39], GM-CSF [40], and G-CSF [41] were reported to initiate angiogenesis by stimulating VEGF production or activating the angiogenesis-related pathway. MCP-3 stimulated the migration of circulating angiogenic cells and angiogenesis partially via the chemokine (C-X-C motif ) receptor 1 (CCR1) [42]. CTACK/CCL27 was reported to accumulate the CD34 + bone marrow cells (expressing CCR10) to participate in skin wound healing and repair [43]. OPG was a positive regulator of microvessel formation in vivo and could activate endothelial colonyforming cells [44]. In addition, we have performed the endothelial cell differentiation assay in vitro and have not found significant differences between VCAM-1 + CV-MSCs and VCAM-1 − CV-MSCs (seen by immunostaining of vWF) under a confocal microscope (Additional file 1: Figure. S2). Based on these studies, we believed that paracrine action rather than differentiation was the principal mechanism of the therapeutic angiogenesis induced by MSCs. Besides, the superior angiogenic effect of VCAM-1 + CV-MSCs could be a result of a synergic effect of multiple angiogenic factors secreted by cells.
To date, the identification of MSC still relies on the minimal criteria specified in 2006 (plastic adhesion, expressing a set of membrane antigens and tridifferentiation capacities) [45]. Besides these properties, the trait of MSCs varies among different origins and individuals; that is, the paracrine actions [46], and immunomodulatory [23] and hematopoietic support capacities [47]. In addition, MSCs isolated from the same tissue also comprised a heterogeneous population. A variety of markers (i.e., Stro-1, SSEA-4, CD271, CD146) have hence been adopted to investigate the potential of particular MSC subpopulations [14]. Psaltis et al. [48] reported that stro-1 + bone marrow-derived MSCs possessed unique cardiovascular paracrine activities. Interestingly, Gronthos et al. [49] employed VCAM-1 as a coexpressed maker to enrich stro-1 + MSCs. Our data agree with Psaltis et al.'s study, which verified the consistent angiogenic potentials of VCAM-1 + MSCs. Most recently, Wang et al. [50] reported that MSCs pretreated with IL-1β and tumor necrosis factor alpha could enhance the therapeutic efficacy on cardiovascular ischemia via upregulating VCAM-1 expression. Consistently, our study demonstrated the presence of a natural VCAM-1 + MSC subpopulation in vivo in placenta CV that exerted excellent paracrine action. Additionally, it has been shown that placenta CV and bone marrow abundant with capillaries contained many more VCAM-1 + MSCs (68 % and 13 %) than adipose tissue and umbilical cord (0.24 % and 4 %) [23], suggesting that VCAM-1 + MSCs might play important roles in the physiological vasculogenesis and angiogenesis.

Conclusion
Our comparative studies at multiple levels on the angiogenic properties of VCAM-1 + CV-MSCs and VCAM-1 -CV-MSCs showed that VCAM-1 could be used as a surface marker to select a MSC subpopulation with superior pro-angiogenic activity. Moreover, the exciting therapeutic efficacy of VCAM-1 + CV-MSCs on ischemic nude mice not only provided a novel strategy for cellbased therapy of ischemic diseases, but also a hint for banking appropriate MSCs for clinical usage.

Additional file
Additional file 1: is Table S1 presenting primers for real-time reverse transcription PCR, Table S2 presenting the description of the human cytokine antibody array (AAH-CYT-G1000), Table S3 presenting the phenotype of three donor-derived CV-MSCs, Table S4 presenting differential angiogenesis cytokines of VCAM-1 -CV-MSCs and VCAM-1 + CV-MSCs, Fig. S1 showing sVCAM-1 concentration in 48-hour CM of VCAM-1 + CV-MSCs and VCAM-1 -CV-MSCs measured by ELISA, and (See figure on previous page.) Fig. 6 Transplantation of VCAM-1 + CV-MSCs significantly enhanced the blood perfusion and the generation of collateral vessels in the ischemic sites. a VCAM-1 +/− CV-MSCs or PBS were injected into the ischemic site of nude mice. Percentage distributions of limb salvage, foot necrosis, and limb loss in the three groups are shown and analyzed by the Fisher's exact test (n = 11, p = 0.102). Ischemia damage and physiological function of ischemic limbs were semiquantified by ischemia scores b and ambulatory impairment scores c (n = 11, *p <0.05). d Blood perfusion in ischemic/ healthy limb was detected by the Kodak In-Vivo FX Pro Imaging System on days (D) 0, 7, and 20. Different colors indicate different blood perfusion. Blood flow increased from dark blue to red. e Blood perfusion ratio in ischemic to healthy limbs was used to quantitatively analyze the blood flow restoration in ischemic limbs (n = 11, *p <0.05, **p <0.01). f Angiography was performed to assess the collateral vessel generation in the ischemic site. Red curves indicated the site of the femoral arteries incision; black arrows showed the collateral vessels in the ischemic hind limb. g Angiography score indicated that VCAM-1 + CV-MSCs were superior to VCAM-1 − CV-MSCs in augmenting collateral vessels (n = 3-5, *p <0.05). h H & E staining was performed to study the vessel density in ischemia limbs. Pictures showed that the blood vessels were full of barium sulfate (silver, scale bar = 100 μm). i Vessel density in VCAM-1 + CV-MSC or VCAM-1 − CV-MSCs group was significantly greater than the PBS group (n = 7, ***p <0.001, ****p <0.0001). Furthermore, VCAM-1 + CV-MSC transplantation apparently promoted the vessel generation compared with the VCAM-1 − CV-MSCs group (**p <0.05). CV chorionic villi, MSC mesenchymal stem cell, PBS phosphate-buffered saline, VCAM-1 vascular cell adhesion molecule 1 (Color figure online)