Introduction

The growth and progression of solid tumors are dependent on newly-formed microvessels, where niches for tumor cells as well as cancer stem cells exist [1–4]. Both angiogenesis and vasculogenesis contribute to tumor neovascularization, which provides oxygen and nutrients for tumor cell survival and proliferation [5–7]. Angiogenesis is a process of new capillary formation from pre-existing host blood vessels, while vasculogenesis is conducted by circulating endothelial progenitor cells (EPCs). However, it remains unclear whether and how vasculogenesis by EPCs incorporates into angiogenesis by sprouting pre-existing endothelial cells (ECs) in the process of neovascularization.

Tumor microvessls are structurally diverse, which forms the basis for tumor microvascular architecture phenotype heterogeneity (T-MAPH) [8] and differential potential targets for anti-cancer therapy. Unfortunately, the existing antiangiogenic agents have not shown consistently effective tumor suppression in clinical trials [9–11]. This raises the possibility that the complexity of T-MAPH needs to be taken into consideration because T-MAPH may consist of ECs derived from different sources such as from existing host vessels and circulating precursors in the blood.

In this study, we examined the possible incorporation of EPCs in host circulation into glioma neovascularization. We found EPCs from human umbilical cord blood (HUCB) were present in newly-formed microvessels in gliomas subcutaneously implanted into SCID mice, in addition to their capacity of forming capillary-like tubules in Matrigel in vitro. Moreover, both ECs differentiated from implanted human EPCs and host ECs participating in neovascularization contributed to the formation of T-MAPH.

Materials and methods

Isolation and culture of EPCs

HUCB was obtained from the informed and consenting donors at Southwest Hospital, Third Military Medical University, Chongqing, China. The umbilical cords were clamped and cord blood collection was performed ex utero according to standard hospital procedures.

EPCs were isolated and cultured according to previous description [12, 13]. Briefly, 70–80 ml HUCB from each donor was collected into a sterile bag containing 28 ml citrate phosphate dextrose solution. Mononuclear cells (MNCs) were isolated by density gradient centrifugation with Histopaque-1077 (Sigma, St. Louis, MO), seeded in culture flasks or on glass cover slips pre-coated with 10 μg/ml human fibronectin (Sigma) and maintained in EGM-2 BulletKit system (Clonetics, San Diego, CA) consisting of endothelial basal medium (EBM-2), 5% fetal bovine serum, human epidermal growth factor (hEGF), vascular endothelial growth factor (VEGF), human basic fibroblast growth factor (hbFGF), insulin-like growth factor-1 (IGF-1), hydrocortisone, GA-1000 (gentamicin sulfate and amphotericin), ascorbic acid and heparin. On day 2, non-adherent cells were removed, followed by addition of new medium. Thereafter, media were changed every 3 days. When adherent cells formed colonies and continued to grow into cobblestone-like monolayer, the cells were reseeded in 6-well plates pre-coated with 10 μg/ml human fibronectin at a density of 400 cells/well to acquire secondary colonies.

Characterization of EPCs

Flow cytometry

Freshly-isolated MNCs and those collected after 4 days of culture were incubated with fluorescein isothiocyanate (FITC)-conjugated mouse monoclonal anti-human CD34 antibody (Miltenyi Biotech, Bergisch Gladbach, Germany) and phycoerythrin (PE)-conjugated mouse monoclonal anti-human CD133/1 antibody (Miltenyi Biotech) for 10 min at 4°C. FITC-conjugated anti-mouse IgG2a (Miltenyi Biotech) and PE-conjugated anti-mouse IgG1 antibody (R&D Systems, Wiesbaden, Germany) were used as isotype controls. The cells were then washed with PBS containing 0.5% bovine serum albumin (BSA) and 2 mM ethylenediamine tetraacetic acid (EDTA). Quantitative analyses were performed using a FACS SCAN flow cytometer and CellQuest software (Becton Dickinson, Heidelberg, Germany). The EPCs were calculated as percentage of all MNCs. Each analysis included 100,000 events.

Immunofluorescent staining for CD34, CD133 and KDR

To determine the expression of CD34, CD133 and KDR, EPCs reseeded on coverslips for 8 or 15 days were fixed in 2% paraformaldehyde for 15 min. For direct immunofluorescent staining, the cells were incubated with FcR Blocking Reagent (Miltenyi Biotech) and then in a specific antibody-cocktail diluted in PBS containing 0.5% BSA as well as 2 mM EDTA at 37°C for 30 min. The specific antibody-cocktail consisted of either (1) FITC-conjugated mouse monoclonal anti-human CD34 (1:11), and PE-conjugated mouse monoclonal anti-human CD133/1 (1:11), or (2) FITC-conjugated mouse monoclonal anti-human CD34 (1:11), and allophycocyanin (APC)-conjugated mouse monoclonal anti-human KDR (1:3.5) (R&D Systems) antibodies. FITC-conjugated mouse IgG2a and PE-conjugated mouse IgG1 or APC-conjugated mouse IgG1 were used as isotype controls. Samples were mounted with 50% glycerol in carbonate buffer and examined by laser confocal scanning microscopy.

Assays for ac-LDL uptake and lectin binding

To evaluate the ability of acetylated low-density lipoprotein (ac-LDL) uptake and binding of lectin on EPCs, cells cultured for 11 days were washed with EBM-2 (serum-free medium) and incubated with 1,1′-dioctadecyl 3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI)-ac-LDL (Molecular Probes, Leiden, Netherlands) at a concentration of 10 μg/ml in EBM-2 for 3 h at 37°C. Then, the cells were fixed with 2% paraformaldehyde for 10 min and incubated with FITC-conjugated lectin from Ulex europeus (UEA-1) (Sigma, Deisenhofen, Germany) at a concentration of 10 μg/ml in EBM-2 for additional 1 h at 37°C. After washing, double-positive (Dil-ac-LDL and FITC-UEA-1) cells were observed under laser confocal scanning microscopy.

Immunocytochemistry for CD31 and vWF

For detection of differentiation potential of EPCs, cells after 40 days of culture were incubated with normal goat serum for 30 min at 37°C in a humidified chamber, and then were incubated overnight with monoclonal mouse anti-human CD31 antibody (1:50 dilution; BD Pharmingen) or an anti-human von Willebrand factor (vWF) (1:100 dilution; Zhongshan Jingqiao Tech, Beijing, China) at 4°C. FITC-conjugated goat anti-mouse IgG (1:100 dilution; Zhongshan Jingqiao Tech) was used to detect CD31 and biotinylated goat anti-rabbit IgG, peroxidase-labeled ultra-streptavidin as well as DAB (Zhongshan Jingqiao Tech) were used to detect vWF. The stained cells were observed under laser confocal scanning microscopy as well as light microscopy.

Assays for EPC vasculogenesis

EPCs cultured in EGM-2 BulletKit system for 10 days were transferred to EBM-2 for 24 h and harvested using 1 mmol/l EDTA for assays of proliferation, migration, tubulogenesis and in vivo formation of vasculature.

Proliferation assay

EPC proliferation was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays. 1,000 EPCs in 200 μl/well were cultured in EBM-2 with or without 50 ng/ml VEGF (Peprotech, UK) in 96-well culture plates pre-coated with 10 μg/ml human fibronectin. After being cultured for 24, 48, or 72 h, EPCs were supplemented with 20 μl MTT (5 mg/ml, Sigma) and incubated for another 4 h. The supernatant was discarded and the EPC preparations were shacked in 200 μl dimethyl sulfoxide (DMSO, Sigma) for 10 min, followed by measurement of optical density value at 490 nm. Results from three independent experiments in triplicates were presented.

Migration assay

EPC migration was evaluated using a modified Boyden chemotaxis chamber assay. Polyvinylpyrrolidone-free polycarbonate filter transwell inserts (8 μm pores; Costar, Cambridge, MA) were incubated with 10 μg/ml human fibronectin and air dried. Inserts were placed in a 24-well plate containing 500 μl/well EBM-2 medium with or without 50 ng/ml VEGF. EPCs (10,000 cells/well) in EBM-2 were added to the upper chamber of the inserts. After 24 h at 37°C, non-migratory cells were removed from the upper chamber by wiping the upper surface of the filter and the migrated cells on the lower side of the filter were washed with PBS and fixed for 10 min with 2% paraformaldehyde. For quantification, the migrated cells were stained with Diff-Quik (American Scientific Products, Edison, NJ, USA) and the number of migrated cells was counted in five random high power (200×) fields per insert.

Tubulogenesis in Matrigel

To examine the ability of EPCs to participate in the formation of vascular network in vitro, growth factor reduced Matrigel (Becton Dickinson Labware) was placed in a 24-well plate at 37°C for 1 h for solidification, EPCs were then replated (30,000 cells/well) on top of the solidified Matrigel. Cells were grown with EBM-2 containing 50 ng/ml of VEGF or EBM-2 alone, and incubated at 37°C for 24 h. The tubule structure formed by EPCs was photographed at 200× magnification under phase-contrast microscopy.

Incorporation of EPCs into tumor vasculature

We utilized a glioblastoma transplantation model to study the ability of EPCs forming endothelial capillaries in vivo. Briefly, human U87 glioma xenografts were grown subcutaneously after injection of 6 × 106 cells (ATCC, Manassas, VA, USA) into the flanks of sublethally irradiated (350 cGy) 6-week-old SCID mice (Laboratory Animal Center, Third Military Medical University). After the appearance of tumors (day 7 after implantation), EPCs (5 × 105/100 μl) were injected into tumor-bearing mice via tail vein (n = 6). Tumor-bearing mice without EPCs were used as control group (n = 5). The animals were raised under specific pathogen-free (SPF) condition. The tumor volumes were measured by a vernier caliper and calculated by the formula: Volume = l (length) × w 2 (width)/2. Twenty-eight days later, the animals were euthanized and tumors were removed and frozen in liquid nitrogen for inclusion into OCT (optimal cutting temperature) compound. Cryosections of 10 μm were prepared. In order to quantitate EPCs incorporation, tumor cryosections were double labeled with a monoclonal mouse anti-human CD31 antibody (1:50 dilution; BD Pharmingen, San Diego, CA) and a monoclonal rat anti-mouse CD31 antibody (1:50 dilution; BD Pharmingen). Cryosections were subsequently washed and labeled simultaneously with tetraethyl rhodamine isothiocyanate (TRITC)-conjugated goat anti-mouse IgG and FITC-conjugated goat anti-rat IgG (1:100 dilution; Zhongshan JingqiaoTech, Beijing, China). Finally, cellular nuclei were counterstained by 4′,6-diamidino-2-phenylindole (DAPI, Sigma). The sections were examined under laser confocal scanning microscopy. At low power field (40×), six most intensely vascularized fields (so called “hot spots”) were randomly selected under laser confocal scanning microscopy. Microvessel counting of these areas was performed at high power field (100×) by Image Pro-plus 6.0 software and the mean of microvessel number expressing human CD31 in the six fields was obtained. Then, the mean of total number of microvessels expressing both human CD31 and mouse CD31 in the same areas was calculated. Finally, the proportion of anti-human CD31 in total vessels was calculated. The relative contribution of engrafted EPCs to tumor vasculature was estimated by dividing total number of anti-human CD31-positive vessels by combined number of anti-human and mouse CD31-positive vessels.

In order to confirm the incorporation of transplanted EPCs into growing tumor vasculature, EPCs (5 × 105) were labeled in vitro with cell tracker carboxyfluorescein diacetate succinimidyl ester (CFSE, Molecular Probes) according to the manufacturer’s instructions and suspended in 100 μl PBS, then injected via tail vein into sublethally irradiated (350 cGy) 6-week-old SCID mice (n = 5). U87 cells (6 × 106) were injected on the same day in the right flanks of the mice. When tumors were palpable, the tumors, spleens and livers were harvested and cryosections of 10–60 μm were prepared. The cryosections were examined under laser confocal scanning microscopy to visualize the distribution and localization of engrafted CFSE-labeled EPCs. To test if the engrafted EPCs possessed EC phenotype and contributed to microvascular heterogeneity in morphology and 3D spatial distribution, tumor cryosections containing green fluorescent EPCs were also immunostained with a monoclonal mouse anti-human CD31 antibody or a monoclonal rat anti-mouse CD31 antibody overnight at 4°C, followed by rinsing with PBS and incubating for 1 h with TRITC-conjugated goat anti-mouse IgG, Cy5-conjugated goat anti-mouse IgG or TRITC-conjugated goat anti-rat IgG (1:100 dilution; Zhongshan Jingqiao Tech). After washing with PBS and counterstained with DAPI for revealing nuclei, the sections were examined with laser confocal scanning microscopy. Cryosections of 60 μm were scanned at 0.5 μm intervals to reconstruct three-dimension of the tumor vessels by an image analysis software (3D projection).

Additionally, fresh tumor tissues were minced and digested with protease (0.05% w/v) (Roche, Mannheim, Germany) at 37°C for 1 h with gentle shaking, washed twice with PBS, and digested again with collagenase/diapase (0.035% w/v) (Roche) at 37°C for 1 h. Cells were filtered through 75 μm sterile steel mesh, spun down at 1,500 rpm, and resuspended in PBS. These cells were labeled with anti-human CD31 or anti-mouse CD31 antibody and assessed by FACS.

We utilized an orthotopic model to determine if EPCs contributed to tumor growth in the microenvironment of brain glioma. Intracranial gliomas were obtained by injecting 1 × 105 U87 cells in the brain of SCID mice (n = 5) according to previous description [14]. Tumors were harvested when animals showed signs of neurological impairment. The animal experiments were approved by the Ethics Committee of our institution.

Statistical analysis

All data are presented as mean ± s.d. Differences between group means were assessed by an unpaired Student’s t-test for single comparisons or by ANOVA for multiple comparisons using SPSS 13.0. P-values less than 0.05 were considered as statistically significant.

Results

EPCs formed colonies and functionally differentiate to ECs

During 3–7 days of culture, MNCs attached to the flask bottom became elongated and showed linear cord-like structures (Fig. 1A). On day 9 through 17, the cells formed colonies (Fig. 1B). The colony-forming cells were reseeded, and their offspring cells produced secondary colonies (Fig. 1C). Cells from the secondary colones gradually evolved into cobblestone-like monolayer reaching confluence (Fig. 1D).

Fig. 1
figure 1

Morphology of EPCs in culture. A Adherent and elongated cells on the culture days 3–7. B Increased EPC colonies with sprouting cells on the days 9–17. C Secondary colonies formed by reseeded EPCs. D Cobblestone-like monolayer. Phase-contrast light microscopy, 200×

Full size image

In order to define the nature of potential EPCs, flow cytometry was performed with isolated MNCs cultured on day 0 through day 4. In freshly isolated MNCs, 1.20% expressed CD133, 1.85% CD34, and 1.06% were CD34/CD133 positive (Fig. 2A). After culture for 4 days, the expression of CD133 decreased to 0.78% which implied MNCs began to differentiate, while CD34 increased to 3.85%, and CD34/CD133 decreased to 0.26% (Fig. 2B). Coexpression of CD34/CD133 or CD34/KDR on EPCs was examined on day 8 and 15. During this period, EPCs continued to positively express CD34 and KDR, while CD133 became less positive (Fig. 3A). Most adherent cells (>95%) after 11 days in culture showed uptake of DiI-ac-LDL and binding of FITC-UEA-1 (Fig. 3B). After 40 days of culture, EPCs were found to express markers of mature ECs, CD31 and vWF (Fig. 3C, D).

Fig. 2
figure 2

Phenotype of EPCs detected by flow cytometry. A The expression of CD133, CD34, and CD133/CD34 by freshly-isolated MNCs. B The expression of CD133, CD34, and CD133/CD34 by MNCs after 4-day culture

Full size image
Fig. 3
figure 3

The expression of endothelial markers by EPCs. A The expression of CD34 and KDR on the days 8–15. B Uptaking of DiI-ac-LDL and binding FITC-UEA-1 by MNCs cultured for 11 days. C, D The expression of CD31 (C) and vWF (D) by EPCs on the day 40. A–C Immunofluorescence under laser confocal scanning microscopy. D Immunoperoxydase labeling, under light microscopy (200×)

Full size image

VEGF promoted EPC proliferation, migration and tubulogenesis

When stimulated with VEGF (50 ng/ml), EPCs showed increased cell number as compared to control group at 24, 48, and 72 h (P < 0.05) (Fig. 4). In addition, when stimulated with VEGF (50 ng/ml) for 24 h, the number of migrated EPCs significantly increased (P < 0.05) (Fig. 5A, B). Furthermore, in Matrigel tubule assays, VEGF stimulated EPCs formed increased number of tubules as compared with cells in the absence of VEGF (P < 0.05) (Fig. 6A, B).

Fig. 4
figure 4

Growth of EPCs with and without VEGF. * Significant increase of VEGF-stimulated EPCs proliferation as compared to those without VEGF stimulation (P < 0.05)

Full size image
Fig. 5
figure 5

The effect of VEGF on EPCs migration. A, B Significantly increased migration of EPCs in response to 50 ng/ml VEGF. * P < 0.05 as compared to cell migration in the absence of VEGF

Full size image
Fig. 6
figure 6

The effect of VEGF on tubulogenesis of EPCs in Matrigel. Increased number of tubules formed by EPCs with stimulation by VEGF (A), * P < 0.05 as compared to cells in the absence of VEGF (B)

Full size image

EPCs incorporated into newly-formed microvessels in vivo

In order to evaluate the capacity of cultured EPCs to participating in neovascularization in transplanted tumors, we utilized a subcutaneous glioblastoma xenograft model. To exclude possible competition by autologous EPCs with engrafted EPCs for incorporation into tumor vasculature [15], SCID mice used in the study were sublethally irradiated (350 cGy) before implantation with U87 glioma cells. Subcutaneous tumors were formed 7 days after transplantation, and then human EPCs were injected via tail vein. All the SCID mice that received sublethal irradiation survived during period of experiment. On day 21 after U87 implantation, the average volume of subcutaneous tumors in SCID mice with EPCs was 244 mm3 ± 38 mm3, significantly larger than that in SCID mice without EPCs (122 mm3 ± 20 mm3). Similarly, the average volume (24.21 mm3 ± 4.64 mm3) of intracranial tumors with EPCs was significantly larger than that of intracranial tumors without EPCs (16.24 mm3 ± 3.98 mm3). This result suggested that EPCs might promote more tumor growth. We harvested subcutaneous tumors on day 28. Immunofluorescence revealed numerous vessels in subcutaneous tumors in which human CD31-expressing cells were found in linear or tubular arrangement in tumor tissues (Fig. 7A). Therefore, injected human EPCs were integrated into the endothelial lining of the tumor microvasculature. For quantification, we calculated the percentage of vessels of human origin in tumor-associated vessels. Vessels expressing human CD31 accounted for 18.68% ± 1.32% of the total vessels, suggesting significant contribution of EPCs to tumor neovascularization. The incorporation of EPCs was also demonstrated by the presence of CFSE-labeled EPCs along human CD31-expressing vascular channels in subcutaneous tumors (Fig. 7B). To evaluate the distribution of transplanted EPCs in non-tumoral organs in tumor-bearing mice, we examined CFSE-labeled cells in cryosections of the spleen and liver. No CFSE-labeled EPCs were detected in the spleen and liver suggesting marked tropism of EPCs for tumors. FACS revealed that ECs from the host blood vessels represented 0.55% among cells from implanted tumor tissues and tumor ECs from implanted human EPCs represented 0.07%. Thus ECs derived from donor EPCs represented a distinct population (11.3% of all ECs in tumor vessels). We examined the role of engrafted EPCs in microvascular architecture by immunofluorescence and three-dimensional reconstruction technology. We found that the vasculature partly incorporated by implanted EPCs were aberrant in morphology showing strip-like structure, arborization, or serpentine sinusoid structure. ECs derived from human EPCs mostly localized at the branching points of the microvessels emanating from mother vessels (Fig. 8A). Three-dimensional chimeric vascular network containing CFSE+ ECs was chaos with different branches, lengths and angles, CFSE+ ECs linked mouse ECs in branches to form thickened vessels of multicellular clusters and or a blind loop (Fig. 8B). These results suggested the contribution of ECs to the microvascular architecture heterogeneity.

Fig. 7
figure 7

EPC incorporation into newly-formed microvessels in tumor xenografts. Human EPCs were injected via tail vein into SCID mice bearing subcutaneously implanted gliomas. Tumors were sectioned on day 28. A Double staining with anti-human CD31 and anti-mouse CD31 antibodies revealing ECs derived from human EPCs (human CD31) that formed new microvessels (arrow) or lined the lumen of murine vessels (arrowhead). B ECs differentiated from CFSE-labeled EPCs in microvessel walls positive for human CD31 in tumor. ECs were revealed in yellow color (arrows)

Full size image
Fig. 8
figure 8

Contribution of EPC to T-MAPH. A Double staining with anti-human CD31 and anti-mouse CD31 antibodies revealing CFSE+ cells expressed human CD31 that form strip structure (arrow), branching pattern (arrowhead), and sinusoid structure (triangle). B Three-dimensional reconstruction of the vessels revealing chaos chimeric vascular network contained of CFSE+ ECs (arrow) and mouse-derived ECs (arrowhead) with different branches, lengths and angles

Full size image

Discussion

The study by Asahara et al. [16] demonstrated for the first time that purified CD34+ progenitor cells from adult peripheral blood could differentiate ex vivo to exhibit an endothelial phenotype. These cells were thereafter called EPCs. Besides from peripheral blood, EPCs have also been collected from adult bone marrow, HUCB, and human fetal liver [17–19]. More recently, HUCB became a common source of EPCs. Accumulating evidence indicates that EPCs are actively recruited to sites of tumor, trauma, inflammation and ischemia, where the cells differentiate in situ into mature ECs and integrate into new blood vessels [20–22].

EPCs are immunologically identified by cell surface markers and functionally by tubulogenesis. Peichev et al. [17] identified a subset of endothelial precursors from circulating CD34+ cells, which express VEGFR-2 and AC133 (CD133). More recent studies define primitive EPCs derived from bone marrow as CD133+/CD34+/KDR+. CD133+ EPCs incubated with medium containing angiogenic growth factors progressively lost CD133 expression with increased cell surface mature EC makers such as vWF, CD31, CD105 and vascular endothelial (VE) cadherin (VE-cadherin) [18]. In addition, Yoder et al. [23] redefined EPCs by their capacity to form colonies with robust proliferation potential in vitro and perfused vessel formation in vivo. Therefore, EPCs are believed to possess the characteristic as being non-EC that can give rise to ECs with clonal expansion and stem cell properties [24].

We have shown that potential EPCs isolated from HUCB incorporated into tumor neovasculature after systemic injection. EPCs expressed progenitor cell markers CD34 and CD133 and possessed the capacity of generating colonies in short term culture. These EPCs exhibited cobblestone-like morphology with phenotypic and functional features similar to mature ECs. These results suggest that EPCs thus obtained may serve as valuable tools for studies of mechanisms of neovascularization in vivo.

It has been reported that bone marrow-derived circulating EPCs contribute to neovascularization in adult. This is in contrast to the hypothesis that vasculogenesis occurs only during the embryonic period [25]. EPCs can be mobilized from bone marrow into the blood stream by a variety of angiogenic growth factors such as VEGF and chemokines such as stromal derived factor 1 (SDF-1). They subsequently are recruited and incorporated into vessel walls at the sites of active physiological or pathological vasculogenesis such as endometrial neovascularization, tissue ischemia, vascular trauma as well as cancer [21, 26, 27]. EPCs have been detected at increased frequency in the peripheral blood circulation of cancer patients [28, 29] and the levels of EPCs may serve as indicators of tumor progression and patient prognosis. EPCs may also indirectly promote vasculogenesis by releasing growth factors, such as VEGF, hepatocyte growth factor (HGF), and granulocyte (macrophage) colony stimulating factor (GM-CSF) [30], etc.

Anti-angiogenic therapies have become one of the most promising approaches in the anti-cancer drug development. Antiangiogenic drugs exert their effects on cancer progression through several mechanisms: (a) by interfering with the activity of angiogenic factors, their receptors or downstream signaling pathways; (b) by upregulating or delivering endogenous inhibitors, and (c) by directly targeting tumor vasculature [31, 32]. Although experimental models offered promising therapeutic results, many anti-angiogenesis agents fail to show consistent tumor suppressing results in cancer patients. These discrepancies in results obtained in preclinical versus clinical studies call for better understanding of the mechanisms of tumor neovascularization.

Tumor vessels are structurally and functionally abnormal, with tumor blood flow in chaotic and variable conditions [33]. Tumor vasculature is highly disorganized, tortuous and dilated, with uneven diameter, excessive branching and shunts. Such abnormality may cause difficulties in achieving effective anti-angiogenesis cancer therapy [34]. We proposed the concept of tumor microvascular architecture phenotype heterogeneity (T-MAPH) based on the observations that tumor blood vessels are diverse and heterogeneous in organization, morphology, density and 3D spatial distribution [8]. Our present study indicates that the chimeric vasculature incorporated by implanted EPCs showed aberrantly morphologic diversity such as arborization, strip-like structure, and serpentine sinusoid structure and chaos three-spatial distribution with different branches, lengths and angles. These results indicate the contribution of EPCs to tumor neovascularization that may be involved in the formation of T-MAPH.

Malignant gliomas are characterized by the presence of abundant newly-formed microvessels. In previous studies, we have shown that tumor cells especially those surrounding microvessels in very aggressive glioblastomas, are highly positive for VEGF staining thus may actively recruit ECs, support their proliferation, and promote the formation of new vasculature. The exposure of ECs to excessive VEGF may also result in the formation of T-MAPH [8]. Circulating EPCs show similar responses to VEGF within tumor tissues. In the present study, EPCs responded to VEGF by proliferation, migration and tubule formation in vitro. HUCB-derived EPCs when injected systemically incorporated into neovessels of tumors formed by VEGF-producing glioblastoma cells. These findings provide a basis for better understanding the mechanism of T-MAPH formation and the potential for the development of more effective anti-cancer vasculature agents.